Anja Dietrich2002

7/31/2019 Anja Dietrich2002

1/21

.Reviews in Molecular Biotechnology 82 2002 211231

/Fluorescence resonance energy transfer FRET andcompeting processes in donoracceptor substitutedDNA strands: a comparative study of ensemble and

single-molecule data

Anja Dietrich, Volker Buschmann, Christian Muller, Markus SauerU

Physikalisch-Chemisches Institut, Uni ersitat Heidelberg, Im Neuenheimer Feld 253, 69120 Heidelberg, Germany

Abstract

.We studied the fluorescence resonance energy transfer FRET efficiency of different donoracceptor labeledmodel DNA systems in aqueous solution from ensemble measurements and at the single molecule level. The donor

. . .dyes: tetramethylrhodamine TMR ; rhodamine 6G R6G ; and a carbocyanine dye Cy3 were covalently attached to .the 5-end of a 40-mer model oligonucleotide. The acceptor dyes, a carbocyanine dye Cy5 , and a rhodamine

.derivative JA133 were attached at modified thymidine bases in the complementary DNA strand with donoracceptor distances of 5, 15, 25 and 35 DNA-bases, respectively. Anisotropy measurements demonstrate that none ofthe dyes can be observed as a free rotor; especially in the 5-bp constructs the dyes exhibit relatively high anisotropy

values. Nevertheless, the dyes change their conformation with respect to the oligonucleotide on a slower time scale in .the millisecond range. This results in a dynamic inhomogeneous distribution of donorracceptor DrA distances and

orientations. FRET efficiencies have been calculated from donor and acceptor fluorescence intensity as well as fromtime-resolved fluorescence measurements of the donor fluorescence decay. Dependent on the DrA pair anddistance, additional strong fluorescence quenching of the donor is observed, which simulates lower FRET efficiencies

at short distances and higher efficiencies at longer distances. On the other hand, spFRET measurements revealedsubpopulations that exhibit the expected FRET efficiency, even at short D rA distances. In addition, the measuredacceptor fluorescence intensities and lifetimes also partly show fluorescence quenching effects independent of theexcitation wavelength, i.e. either directly excited or via FRET. These effects strongly depend on the DrA distanceand the dyes used, respectively. The obtained data demonstrate that besides dimerization at short DrA distances, anelectron transfer process between the acceptor Cy5 and rhodamine donors has to be taken into account. To explaindeviations from FRET theory even at larger DrA distances, we suggest that the -stack of the DNA double helix

UCorresponding author. Tel.: q49-6221-548460; fax: q49-6221-544255.

.E-mail address: [email protected] M. Sauer .

1389-0352r02r$ - see front matter 2002 Elsevier Science B.V. All rights reserved. .PII: S 1 3 8 9 - 0 3 5 2 0 1 0 0 0 3 9 - 3


2/21

( )A. Dietrich et al. rRe iews in Molecular Biotechnology 82 2002 211231212

mediates electron transfer from the donor to the acceptor, even over distances as long as 35 base pairs. Our datashow that FRET experiments at the single molecule level are rather suited to resolve fluorescent subpopulations inheterogeneous mixture, information about strongly quenched subpopulations gets lost. 2002 Elsevier Science B.V.

All rights reserved.

.Keywords: Fluorescence resonance energy transfer FRET ; Single-molecule spectroscopy; DNA mediated electron transfer

1. Introduction

Electronic energy transfer between ground andexcited states of chromophores play a key role in

chemistry, biology and physics Mataga andKubota, 1970; Turro, 1978; Agranovich and

.Galanin, 1982 . Generally these photophysicalprocesses involve non-radiative transfer of elec-tronic excitation from an excited donor, DU to aground state acceptor molecule A, and occur ontime scales from femtoseconds to milliseconds atdistances ranging from a few to approximately

100 A. For donoracceptor distances within theweak coupling limit, i.e. donoracceptor distances

.) 2 nm, Forster 1948, 1968 derived an expres-sion for the rate constant k for dipoledipole-ET

w .xinduced energy transfer Eq. 1 .

2 . .9000ln10 F dD D A .k s 1HET 5 4 6 4128 n N R 0A D

.Eq. 1 expresses the rate constant for energytransfer in measurable spectroscopic quantitiessuch as: the refractive index of the medium, n;the orientation factor 2 which is generally as-

sumed to be 2r3 for random orientations Dale.et al., 1979 ; the fluorescence quantum yield of

the donor, ; its fluorescence lifetime, ; Avo-D Dgadros number, N ; the normalized fluorescence

A .spectrum of the donor, F ; the absorptionDspectrum of the acceptor, expressed by its extinc-

.tion coefficient, ; and the average transitionAy1 .frequency in cm . Eq. 1 can be written in

terms of the Forster critical transfer radius R , 0the distance at which the transfer efficiency equals

w .x50% Eq. 2 .

6R1 0 .k s 2ET / RD

The efficiency of FRET, E is then defined tobe equal to:

1 .Es 36 .1 q RrR0

Forster type resonant energy transfer occursfor allowed singletsinglet transitions if the emis-sion of DU and the absorption of A overlapsignificantly. For such transitions the critical

transfer radii range from 10 to 100 A Berlman,.1973 . In combination with the strong distance-

dependence, fluorescence resonance energy .transfer FRET is ideally suited to obtain infor-

mation about structure and structural changes ofbiologically important molecules Stryer and

Haugland, 1967; Veatch and Stryer, 1977; Stryer,

1978; Stuhmeier et al., 1997; Tuschl et al., 1994;.Parkhurst et al., 1996; Szollosi et al., 1998 . Dur-

ing the last few years, improvements in sensitivityand spatial resolution of conventional fluores-cence microscopy have led to an enforced practi-

cal application of FRET for review see Weiss,.1999, 2000; Selvin, 2000 . In combination with

genetically encoded dyes, such as green fluores- . .cent protein GFP and its relatives Tsien, 1998 ,

FRET established the principal ability to monitorinteractions and distances between molecules

even in living cells.However, one should be aware of the fact that

the distance range that can be efficiently probedby FRET is limited. Due to the 1rR6 distancedependence, distances in the range 0.51.5 R ,0i.e. FRET efficiencies, E in the range 0.980.10,are suitable for FRET measurements. At higherdistances, the FRET efficiency drops to zero, atshorter distances, the FRET efficiency is close tounity, and distance changes result in very smallchanges in FRET efficiency. Furthermore, the


3/21

( )A. Dietrich et al. rRe iews in Molecular Biotechnology 82 2002 211231 213

FRET methodology can be applied only for dis-tances within the weak coupling limit. In a moregeneral description, electronic energy transfer in-

volves non-radiative transfer of electronic excita-tion energy from an excited donor DU to anacceptor molecule, A, independent of the dis-tance. Only when long range coulombic interac-

tions contribute weak coupling between donor.and acceptor , the energy transfer process can be

formulated in terms of dipoledipole interactions . .via Eq. 1 Speiser, 1996 .

For intermediate or strong coupling betweenthe donor and acceptor, short range exchange

.interaction, as formulated by Dexter 1953 can

dominate the electronic energy transfer process,even when the relevant electronic transitions areforbidden. In addition, the uncertainty in theorientation factor 2 renders the application ofFRET for the determination of absolute distancesmore difficult. Therefore, FRET is rather suitedfor the detection of dynamic distance changes.On the other hand, conformational changes suchas folding or unfolding of a protein are difficult toreveal from ensemble measurements due to thelack of synchronization. Furthermore, subpopula-tions with slightly changed DrA distances or ori-

entations are averaged out in ensemble experi-ments. Therefore, the spectroscopic observationof individual DrA pairs seems to be the methodof choice to overcome the described problems.

.Ha et al. 1996a first demonstrated single pair .fluorescence resonance energy transfer spFRET

on double-labeled DNA strands adsorbed on adry surface. Fluctuations in FRET efficiencieshave also been used to study conformational dy-namics of: single immobilized SNase protein

.molecules during catalysis Ha et al., 1999a ; lig-

and-induced conformational changes in single .RNA molecules Ha et al., 1999b ; folding dy-namics of individual GCN4 peptides Jia et al.,

.1999 ; and the effect of salt on the dissociation ofthe coiled coil dipeptide -tropomyosin Ishii et

.al., 1999 . However, care must be taken to ensureminimal perturbation from the immobilization ofDNA, RNA, or proteins on modified glass sur-

.faces Osborne et al., 2001 . Therefore, spFRETmeasurements on freely diffusing molecules seemto be a valuable approach. On the one hand,

fluorescence bursts from single molecules travers-ing the laser beam in solution are small, i.e. onlya limited number of photon counts in the order of

tens to hundreds can be detected from an individ-ual molecule. On the other hand, detailed analy-sis of the photon bursts can provide invaluableinformation about the distributions of molecular

properties, undisturbed by surface effects Deniz.et al., 1999, 2000; Dahan et al., 1999 . Due to the

availability of hundreds to thousands of events inonly a few minutes, single molecule studies insolution can uncover easily subpopulations of an-alyte molecules in heterogeneous ensemblesSauer et al., 1998; Eggeling et al., 1998; Deniz et

.al., 1999 .Independent of the applied experimental con-ditions, great care must be taken in attributingthe change of spFRET efficiency to a distancechange between donor and acceptor. In singlemolecule experiments there are several factors

which influence the measured FRET efficiencies . including a digital photobleaching Ha et al.,

. .1996a and b so called blinking due to: rotatio-nal jumps Ha et al., 1996b; Ruiter et al., 1997;

Bartko and Dickson, 1999; Weston and Goldner,.2001 ; intersystem crossing into long lived triplet

states Veerman et al., 1999; Weston et al., 1999;. English et al., 2000a,b ; spectral diffusion Lu and

Xie, 1997; Yip et al., 1998; Weston and Buratto,.1998 ; cistrans isomerization as in case of the

indocarbocyanine dye Cy5 Widengren and.Schwille, 2000 ; and fluctuations in the excited

.state kinetics Tinnefeld et al., 2000, 2001 .In addition, intermolecular quenching of the

donor due to dynamic interactions with the bio-molecule, e.g. DNA, has to considered. As Seidel

.et al. 1996 already reported, coumarin dyes are

more or less efficiently quenched by all four DNAnucleotides. Simultaneously, we found that mostrhodamine and oxazine dyes are efficiently

quenched by the DNA base guanosine Sauer etal., 1995; Nord et al., 1997; Lieberwirth et al.,

.1998 . Covalent linking of these dyes to oligonu-cleotides containing guanosine residues results ina diminished fluorescence quantum yield and de-cay time dependent on the distance between theguanosine residue and the fluorescent dye. Inaddition, we demonstrated that the fluorescence


4/21


kinetics of the dye are influenced mainly whenthe guanosine residue is located in close vicinity

.to the dye Sauer et al., 1998 . This quenching

effect has been supported by several other groups .Vamosi et al., 1996; Widengren et al., 1997 and`has been used to study conformational fluctua-tions in DNA oligonucleotides at the singlemolecule level by time resolved fluorescence

spectroscopy Edman et al., 1996; Jia et al., 1997;.Eggeling et al., 1998; Sauer et al., 1998 . As in the

case of stilbene labeled hairpin oligonucleotideswith dCdG stems, a photo-induced electrontransfer reaction from the guanine ground stateto the excited rhodamine or oxazine singlet state

provides a plausible mechanism for fluorescence .quenching Lewis et al., 1997 . The difference in

behavior of neighboring dG compared with dA,dT, or dC bases can be attributed to the loweroxidation potential of dG vs. dA or the pyrimidine

bases dT and dC Sauer et al., 1995; Seidel et al.,.1996; Steenken and Jovanovic, 1997 . Besides the

monitoring of the dynamical behavior of DNAoligonucleotides, the quenching influence ofguanosine residues on the attached reporter dyecan also be used as a powerful tool to probe the

local DNA sequence in double- or single-stranded .DNA Knemeyer et al., 2000 .Furthermore, it should be noted that each of

these processes influences the measured FRETefficiency to a different degree, and even moreimportant, the dye structure of the donor andacceptor itself might control the contributionsfrom non-distance change processes.

Therefore, it is essential to choose suitablecontrol samples and examine both single moleculeand bulk measurements to fully understand the

influence of non-distance change processes. Inaddition, most problems might be circumventedby a careful spectroscopic study of the only donorand only acceptor labeled molecule and by chang-ing the donor and acceptor dye and the couplingposition.

Motivated by these considerations we investi-gated and compared the FRET efficiency of DNAmolecules labeled with different donor and accep-tor molecules at different positions in ensemblemeasurements as well as at the single molecule

level. As donor molecules we used two rhodamine .derivatives: rhodamine 6G R6G ; and tetrameth-

.ylrhodamine TMR , and a carbocyanine deriva-

tive, Cy3 coupled to the 5-end of a 40-meroligonucleotide. With a persistence length of ; 50

.nm Bustamante et al., 1994 , 40mer double-stranded DNA should represent an ideal FRETsystem with relatively fixed DrA distance, influ-enced only by the conformational flexibility of theused linkers. The acceptor dyes, a carbocyanine

.derivative Cy5 and a rhodamine derivative .JA133 , were coupled to the complementarystrand at different distances of 5, 15, 25 and 35

.base pairs Fig. 1 . Since TMR and R6G are

efficiently quenched by guanosine residues in aclose neighborhood, we used a specific oligonu-cleotide sequence containing no guanosineresidues at the 5-end. The FRET efficiencies

were calculated via fluorescence intensity and timeresolved data. Independent of the used method,the observed spectroscopic data indicate the pres-ence of an additional quenching path. Our dataevidence that other fluorescence quenchingprocesses like electron transfer reaction, either

.through space in case of the 5 base pair distanceor DNA-mediated via the -stack of the DNA

bases Meggers et al., 1998; Kelley and Barton,.1999; Ye and Jiang, 2000; Lewis et al., 2001 , have

to be taken into account.

2. Results and discussion

2.1. Design of FRET constructs

A set of differently labeled FRET constructs

.with varying DrA DrA base pair separationwas synthesized to investigate and compare thedistance dependence and the influence of the dyestructure on the measured spectroscopic charac-teristics. To keep parameters such as diffusionrate and thermal stability constant, the differentDNA constructs had the same sequence and aconstant length of 40 base pairs. From inter-molecular quenching experiments it is known thatthe two rhodamine derivatives R6G and TMR are

efficiently quenched by guanosine residues Sauer


5/21


.et al., 1995, 1998; Nord et al., 1997 . On the otherhand, the rhodamine derivative JA133 and mostindocarbocyanine dyes such as Cy5 are not

quenched by DNA nucleotides Lieberwirth et al.,.1998 . Due to these observations, we decided to

use FRET constructs where guanosine residuesare so far apart from the donor dyes that anyquenching influence of the DNA base guanineshould be minimized. The selected 40meroligonucleotide consists of six consecutive adeno-sinethymidine base pairs at the donor side of

.the construct Fig. 2 .The Forster radii, R of the four investigated 0

.DrA pairs of: 63.5 A R6GrCy5 ; 64.5 A . .

TMRrCy5 ; 55.8 A Cy3rCy5 ; and 59.0 A .TMRrJA133 were calculated from the spectraloverlap of the separate absorption and emissionspectra of the donor and acceptor only double-stranded oligonucleotides, and the fluorescencequantum yields of the donor only constructs, re-spectively. Extinction coefficients of 1.1 = 105 and

5 y1 y1 2.5= 10 l mol cm for JA133 Sauer et al.,

.1995 and Cy5, respectively, have been used. Alldyes were assumed as free rotors.

As can be seen from the data in Table 1, the

donor dyes R6G and Cy3 exhibit monoexponen-tial fluorescence decay times attached at thedoubled-stranded DNA. The TMR labeledoligonucleotide shows a second shorter fluores-cence decay time of 1.25 ns with a relatively smallamplitude. Due to the flexibility of the used C -6aminolinkers, the donor dyes can adopt differentconformations with respect to the oligonu-cleotide, thereby preventing or promoting rota-tional mobility of some side chains such as theamino groups in case of TMR. Hence, the differ-

ent rotational mobility is directly reflected in themeasured fluorescence decay time and quantum .yield Drexhage, 1977; Vogel et al., 1988 . How-

ever, together with the relatively high fluores-cence quantum yields, the data imply that thedonor dyes are not quenched by DNA nu-cleotides. In other words, the donor only con-structs exhibit relatively homogeneous spectros-

.Fig. 1. Molecular structures of the used dyes. The donor dyes 5-carboxyrhodamine 6G R6G , 5-carboxytetramethylrhodamine .TMR , and the indocarbocyanine dye Cy3 as well as the acceptor dye Cy5 were obtained as functionalized N-hydroxysuccinimidylesters. As a second acceptor we used a rhodamine derivative, JA133.


6/21


.Fig. 2. a Schematic diagram of the optical setup. For efficient excitation of the donor dyes TMR, R6G and Cy3 we used afrequency-doubled Nd:YAG laser emitting at 532 nm. The collimated laser beam was directed into an inverted microscope and

.coupled into the microscope objective with high numerical apertures oil immersion, 100 = , NA 1.4 via a dichroic beam splitter.Within the microscope objective, the beam was focused into the sample to detect freely diffusing FRET constructs. Thefluorescence light was collected through the same objective and imaged onto a 100-m pinhole to reject out-of-focus light. Thetransmitted fluorescence light is then split by a dichroic mirror and focused onto the active areas of two avalanche photodiodes,

.APDs SPCM AQR-14 . To further isolate the donor and acceptor signal we used additional band pass filters in front of the APDs .570DF60 and 675RDF50 . The signals of both APDs were coupled to a counting board and a personal computer. Sample solutions y11 . y610 M were prepared from 10 M stock solutions by several dilution steps. For diffusion measurements, the average excitation

. .power at the sample was adjusted to be 325 W. b Model of the DrA DNA constructs with varying distance: D- N -A;5 . . . . UD- N -A; D- N -A; and D- N -A. The 40mer complementary oligonucleotides: i 5- ATA TAA GCT ATG CAA TGC TAT15 25 35

. U U U UGGT AAC GTA TCG AAT CGT A-3; and ii 5-T ACG AT T CGA TAC GTT ACC ATA GCA T TG CAT AGC TT A .TAT-3 were custom synthesized. All donor dyes were coupled to the 5 end of oligonucleotide i via 5-aminomodifier C . Acceptor6

.dyes were also coupled to C amino-modified thymidine bases at four different positions in the complementary oligonucleotide ii6resulting in DrA distances of 5, 15, 25 and 35 base pairs, respectively. Since 10 base pairs make one turn in B form double helix, therelative positions of both dyes are similar in all FRET constructs. To ensure a defined and comparable environmental influence

U .independent of the DrA distance, we used similar sequences nearby the acceptor positions T . Coupling reactions were carriedout in 250 mM carbonate buffer, pH 9.3, at room temperature for 2 h. The labeled oligonucleotides were purified by reversed phase .RP18-column HPLC using a gradient of 075% acetonitrile in 0.1 M aqueous triethylammonium acetate. As confirmed by

.absorption spectroscopy, this method yields 100% labeled DNA. The constructs are referred to as Donor- N -Acceptor, where nn . .indicates the base pair separation between the dyes, e.g. TMR- N -Cy5 for tetramethylrhodamine labeled oligonucleotide i25

.hybridized to Cy5 labeled oligonucleotide ii with a DrA distance of 25 base pairs.


7/21


Table 1Ensemble spectroscopic characteristics of the different FRET constructs in aqueous buffer containing 1 M NaCl

D D A D D D D . E E E ab s em f,rel 1 2 av av . . . . .nm nm ns ra ns ra ns1 2

.TMR- N 557 581 1.00 3.55r0.89 1.25r0.11 3.30 .TMR- N -Cy5 557r650 584r663 0.78 0.22 0.09 3.50r0.86 1.27r0.14 3.19 0.0335 .TMR- N -Cy5 557r650 584r666 0.63 0.37 0.22 2.85r0.88 0.81r0.12 2.61 0.2125 .TMR- N -Cy5 557r650 584r667 0.31 0.69 0.53 3.01r0.55 0.79r0.45 2.01 0.3915 .TMR- N -Cy5 552r652 584r667 0.11 0.89 0.63 3.26r0.82 0.62r0.18 2.78 0.165

.R6G- N 534 557 1.00 4.28r1.00 4.28 .R6G- N -Cy5 534r650 558r663 0.91 0.09 0.04 4.14r0.96 0.57r0.04 4.00 0.0735 .R6G- N -Cy5 534r650 557r663 0.70 0.30 0.12 3.77r0.96 0.84r0.04 3.65 0.1525 .R6G- N -Cy5 534r650 556r665 0.34 0.66 0.53 3.77r0.74 1.09r0.26 3.07 0.2815 .R6G- N -Cy5 529r652 548r668 0.24 0.76 0.45 3.82r0.91 0.98r0.09 3.56 0.175

.Cy3- N 551 565 1.00 0.90r1.00 0.90 .Cy3- N -Cy5 551r650 565r664 0.87 0.13 0.06 1.42r0.52 0.25r0.48 0.86 0.0435 .Cy3- N -Cy5 551r650 565r664 0.75 0.25 0.14 1.33r0.48 0.22r0.52 0.75 0.1725 .Cy3- N -Cy5 551r650 565r664 0.49 0.51 0.29 1.34r0.25 0.23r0.75 0.51 0.4315 .Cy3- N -Cy5 549r651 564r667 0.29 0.71 0.51 1.43r0.20 0.19r0.80 0.44 0.515

.TMR- N 557 581 1.00 3.55r0.89 1.25r0.11 3.30 .TMR- N -JA133 558r620 581r 0.77 0.23 0.03 3.48r0.89 1.06r0.11 3.21 0.0335 .TMR- N -JA133 559r621 581r634 0.69 0.31 0.08 3.39r0.78 1.45r0.22 2.96 0.1025 .TMR- N -JA133 558r622 581r638 0.36 0.64 0.25 3.01r0.61 0.99r0.39 2.22 0.3315 .TMR- N -JA133 554r625 581r639 0.20 0.80 0.04 3.31r0.78 1.11r0.22 2.83 0.145

Absorption and emission maxima of the donor and acceptor: relative fluorescence quantum yield, D ; and fluorescencef,rel

lifetime, D

of the donor; and FRET efficiencies calculated from the donor decrease, ED

; donor decrease and acceptor increase,A D .E , and from the donor average fluorescence lifetime, E . Absolute fluorescence quantum yields of donor only labeledavD . . .oligonucleotides, s0.40 for TMR- N , 0.70 for R6G- N and 0.20 for Cy3- N , were calculated using rhodamine 6G in ethanolf

.as standard with a quantum yield of 0.90 Arden-Jacob, 1992 . The instrument response function required for deconvolution of thefluorescence decays was obtained from a scattering solution. The quality of the decay fits was assessed by means of the reduced

2 .chi-squared statistical parameter . In most cases a multiexponential fit was necessary to describe the measured decays . y1 .. .satisfactorily I t sa exp yt . Here a are pre-exponential factors that describe the ratio of the excited species a s1 ,i i i i

and denote their lifetimes, respectively. Average fluorescence lifetimes were calculated via sa ra . All fluorescencei av av i i i idecays were fitted with 2 values -1.2. Fluorescence anisotropies at the emission maxima, r for the dyes were calculated from the

polarization of the emission components I , I , I and I where the subscripts denote the orientation of the excitation andVV VH HV HH .I y GIVV VH.emission polarizers as rs where GsI rI . R values were calculated from the overlap of the donorVH HH 0 .I q2GIVV VH

conjugate emission spectrum and the acceptor conjugate absorption spectrum in 1 M NaCl assuming a refractive index, n of 1343.

copic characteristics, which is at least important,if not the prerequisite, for successful FRET ex-periments.

2.2. Distance dependence of FRET

The absorption spectra of the hybridized dou-ble labeled FRET oligonucleotides show the ex-pected three peaks of donor absorption, acceptorabsorption and the absorption of the oligonu-

cleotide at approximately 260 nm. Without anydirect interaction between the dyes in the groundstate, the absorption spectra of the FRET con-structs should equal the sum of the absorptionspectra of donor and acceptor only labeledoligonucleotides. As can be seen from Table 1,this behavior can be observed for FRET con-structs with larger DrA distances of 15, 25 and 35base pairs. In strong contrast, the absorption max-ima of the donor and acceptor in the five base


8/21


pair constructs are always slightly shifted to theblue and red, respectively, independent of the

.donor or acceptor dye used Table 1 .

The emission spectra of the FRET constructswith DrA distances of 15, 25 and 35 base pairsshow the expected decrease in donor fluorescence

.and increase in acceptor fluorescence Fig. 3ad .In addition, the emission curves intersect in onepoint, which demonstrates a direct correlationbetween the decrease of donor fluorescence and

increase of acceptor fluorescence. However, againthe constructs with five base pair separation ex-hibit striking behavior. Although the donor inten-

sity decreases, the energy is not completely trans-ferred to the acceptor, i.e. with exception of theCy3rCy5 pair, the acceptor fluorescence intensi-ties of all other pairs are lower than for theconstructs with 15 base pair separation. In addi-tion, the 5-bp constructs show slightly shiftedemission maxima implying weak dyedye interac-

y6 . .Fig. 3. Fluorescence emission spectra of the different FRET constructs in aqueous buffer containing 1 M NaCl 25 C, 10 M . a . . .TMRrCy5 excited at 520 nm; b R6GrCy5 excited at 480 nm; c Cy3rCy5 excited at 520 nm; and d TMRrJA133 excited at 520

. .nm. For comparison the emission spectra of the donor and acceptor only labeled constructs, donor- N and N -acceptor, . .respectively, are given. Complementary combinations of donor and acceptor labeled oligonucleotides i and ii were mixed 1:1 at

room temperature in 100 mM Tris borate, pH 8.3, containing 1 M NaCl. The completeness of hybridization was verified by additionof small amounts of the acceptor labeled oligonucleotide while monitoring the donor fluorescence intensity. For complete 1:1hybridization, the donor fluorescence intensity should be independent on any further addition of acceptor labeled oligonucleotide.

.To exclude polarization effects, the fluorescence was recorded under the magic angle 54.7 .


9/21


Fig. 4. Ensemble fluorescence decays of the TMRrCy5 labeled y6 .DNA constructs 10 M in aqueous buffer containing 1 M

NaCl. In addition, the fluorescence decay of the only donorlabeled double-stranded DNA is shown. Excitation at 495 nm,emission at 570 nm, 37 psrchannel. Ensemble fluorescencelifetimes were determined from 10y6 M solutions of theconjugates at the emission maxima using either a pulsed LED .center wavelength: 495 nm or a diode laser emitting at 635nm as excitation source and time correlated single photon

.counting TCSPC .

tions even in the excited states. Only the Cy3-

.N -Cy5 construct exhibits at least a decrease5and increase in donor and acceptor fluorescence,respectively, compared to the fluorescenceproperties of the 15-bp construct.

With exception of the 5-bp constructs, the flu- .orescence decay functions Fig. 4 show the same

trend, i.e. with decreasing DrA distance, thelifetime decreases. All donor fluorescence decaysmeasured from FRET constructs could only bedescribed satisfactorily with at least a biexponen-tial model. Even at the shortest DrA distance of

5 bp, the donor fluorescence decay still exhibits along fluorescence component with relatively highamplitude. Interestingly, the 5-bp Cy3rCy5 con-struct exhibits the expected shorter decay time.Table 1 summarizes the relative fluorescencequantum yields and fluorescence decay times ofthe donor dyes obtained from ensemble measure-ments. The data strongly support the idea thatanother effect has to be taken into account atshort DrA distances. In addition, the biexponen-tial fluorescence decays with the longer compo-

nent, always comparable to the lifetime of thedonor only labeled construct, imply a strong con-formational heterogeneity, e.g. at least two ex-

treme conformations with different orientationsof the dipole moments.

We calculated the FRET efficiency, ED fromthe decrease in donor fluorescence intensity via

Es 1 yI rI , where I , and I denote theDA D DA Dmeasured integrated fluorescence intensity of thedonor in the presence and absence of the accep-tor in the wavelength range 540600 nm, respec-

.tively Table 1 . FRET efficiencies calculated fromexcitation spectra of the acceptors show similar

.results data not shown . In addition, we calcu-

lated the FRET efficiency, E

A

from the quantumyield and cross-talk corrected acceptor intensity,I between 640 and 700 nm, and the donor inten-A

A sity, I between 540 and 600 nm via E sI r ID A D.qI . This method is generally applied to calcu-A

late the FRET efficiency from single moleculedata. Furthermore, Table 1 shows the FRET ef-

D .ficiencies, E obtained from the average fluor-aescence lifetimes measured for the donor dyes inthe presence and absence of the acceptor. Inde-pendent of the method used, the FRET efficien-

.cies measured at short DrA distances 5 bp

appear to be much to low. For long distances 35.bp , unexpectedly high FRET efficiencies were

observed for all four constructs if the efficiency iscalculated solely from the donor intensity. Com-parison of the differently calculated FRET effi-ciencies implies that an additional efficientquenching pathway has to be taken into accountto explain the observed strong deviations of theexperimental FRET data from theory. The overalllower FRET efficiencies calculated from the aver-age fluorescence lifetimes indicates the presence

of a short decay component which we can obvi-ously not resolve with our experimental timeresolved equipment. Hence, even the time

.resolved data although in an indirect way implythat a fraction of the donor dyes for example a

.conformational subpopulation is efficientlyquenched by another mechanism.

2.3. Anisotropy and distance distributions

Fluorescence anisotropy measurements show


10/21


that none of the investigated donor and acceptordyes can be regarded as a free rotor. For the 15,25 and 35 bp constructs, and donor or acceptor

only labeled oligonucleotides anisotropy, valuesvary between: 0.16 and 0.20 for R6G and TMR;0.10 and 0.15 for Cy3; 0.24 and 0.28 for Cy5; and0.26 and 0.29 for JA133 in 1 M NaCl. Closerexamination of the donor anisotropies shows asystematic increase with decreasing DrA distancedue to faster FRET. Interestingly, the anisotropyof all dyes increases considerably in the 5-bpdistance constructs, e.g. for Cy5 we measured an

.anisotropy of 0.38 in TMR- N -Cy5; 0.31 in5 . .R6G- N -Cy5; 0.35 in Cy3- N -Cy5; and for5 5

.JA133 we obtained rs 0.38 in TMR- N -JA133.5These data clearly indicate nearly fixed dipolemoments of the donor and acceptor dye.

As has been recently shown by several groupsSauer et al., 1995; Vamosi et al., 1996; Seidel et`al., 1996; Nord et al., 1997; Widengren et al.,

.1997; Lieberwirth et al., 1998 , rhodamine andoxazine dyes attached covalently to oligonu-cleotides have the tendency to interact with DNAbases. The degree of this aggregation is stronglycontrolled by the water solubility of the dye struc-ture and the flexibility of the used linker arm.

Therefore, relatively high anisotropy values mightappear. However, single molecule studies re-

vealed the existence of several different confor-mational states with respect to the oligonu-cleotide which interchange in the microsecond to

millisecond range Edman et al., 1996; Wen-nmalm et al., 1997; Eggeling et al., 1998; Sauer et

.al., 1998 . This conformational motion is slowcompared to the emission lifetime of the dye.Hence, the measured anisotropy values might in-dicate fixed rotors, which in fact change their

transition dipole on a slower time scale. In addi-tion, our fluorescence data demonstrate strongheterogeneity of FRET efficiency, which might becontrolled by different orientations of the transi-tion dipoles rather than by distance changes. Thisis supported by the observation that most fluor-escence decays contain more than 50% of a long

.fluorescence component Table 1 which is similarto the fluorescence lifetime measured for thedonor only labeled oligonucleotides, independentof the DrA distance.

2.4. spFRET measurements

To compare ensemble data with single molecule

measurements, 10y11 M solutions of the con-structs were excited at 532 nm and the donor and

acceptor emission were detected separately Fig.. 5 . As previously pointed out Dahan et al., 1999;

.Deniz et al., 1999 , the ability of single moleculedetection to measure distributions implies thatinhomogeneous populations of molecules can bestudied. However, the time resolution of the mea-surements controls whether the inhomogeneityappears as static or dynamic. For example, asalready mentioned above, the conformational dy-

namics of dyes attached to DNA occur on amicrosecond to millisecond time scale. Althoughthe dyes may adopt all kinds of possible confor-mations with different orientations of the transi-

tion dipole moment, the anisotropy originatingfrom the rotational mobility of the molecule dur-ing its excited state lifetime of a few nanosec-

.onds is high, which indicates strongly hinderedrotational mobility of the chromophores attachedto DNA.

The typical transition time of a ds 40meroligonucleotide in the detection volume ; 1 fem-

.toliter is approximately 1 ms. In other words, thisis within the time range of these conformationalchanges. Hence, by choosing an integration timeof 1 ms per bin, subpopulations exhibiting differ-

Fig. 5. Typical fluorescence trajectories monitored on the . . y1 1donor gray and acceptor black channel of a 10 M

. .solution of TMR- N -Cy5 in aqueous buffer 1 M NaCl .25


11/21


ent FRET efficiencies should be revealed. Fig. 5shows an example of a 1-ms integration timefluorescence burst trajectory for a 10y1 1 M solu-

.tion of TMR- N -Cy5 showing clearly correlated25photon bursts. Because direct excitation of theacceptor and leakage of the donor emission intothe acceptor channel is very small, fluorescencebursts on the acceptor channel were assigned toacceptor molecules excited via FRET. With anaverage excitation energy of 325 W, typicalbackground count rates of 0.62 kHz for the greenand 0.47 kHz for the red channel were measured.To discriminate dye fluorescence efficiently frombackground noise, only time bins containing in

.total green and red channel more than 30 counts .30 kHz were used to calculate the FRET effi-ciency. This criteria of thresholding selects onlythose fluorescent bursts where donor andror ac-

.ceptor emission are strong Ying et al., 2000 .

.The single pair FRET spFRET efficiencies,Esp were calculated from the background cor-rected fluorescence intensities of the donor Icorr,D

corr

w .xand acceptor I E q . 4 .A

Icorr y CAsp .E s 4corr corr .I yC qIA D

The cross-talk, C between the donor and ac-ceptor channel was calculated from ensembleemission spectra and the transmission of the filter

set 8.7% for TMR; 8.1% for Cy3; and 3.2% for.R6G .

Fig. 6 shows the spFRET efficiency histograms

that were generated from the single molecule .fluorescence intensity data of: TMR- N -Cy5;x . .R6G- N -Cy5; and Cy3- N -Cy5 constructs, andx x

of only donor labeled oligonucleotides measuredin aqueous buffer containing 1 M NaCl. In

Fig. 6. FRET histograms extracted from single molecule data of 10y11 M solutions of the differently labeled DrA constructs andcorresponding Gaussian fits.


12/21


absence of an acceptor, the donor only labeledoligonucleotides show only one peak with zeroFRET efficiency. For the 5-bp separation con-

structs, two peaks are evident, one centered atapproximately zero efficiency and a second at

.very high ) 0.95 efficiency. With increasing sep-aration length, the second peak clearly shifts tolower FRET efficiency, as expected for Forsterenergy transfer. Due to the overlap with the zeropeak, it was impossible to separate out the FRETpeaks for the 35-bp constructs which should ex-

hibit theoretically very low FRET efficiencies -.0.10 . It is generally assumed that the large peak

in zero energy transfer efficiency arises from

non-hybridized, single-stranded donor labeledoligonucleotides, and premature photobleachingof Cy5. As has been recently shown Grunwell et

.al., 2001 , the use of oxygen scavengers can drasti-cally decrease the photobleaching and oxidativedamage of Cy5, thereby decreasing the observedamplitude of the zero peak.

In addition, it should be pointed out that singlemolecule analysis is always subjected to a kind ofselection, in that single molecule experiments areonly performed on the subset of all moleculesthat are sufficiently bright to be detected and

investigated under the applied burst recognitionprocedure. For example, dim or dark acceptormolecules may also intrinsically exist, which con-tribute to the large zero peak. Especially in thecase of Cy5 as acceptor, the formation of a non-fluorescent cis state has to be taken into account.

It has been shown recently Widengren and.Schwille, 2000 , that irrespective of the excitation

rate, 50% of the Cy5 molecules are in an essen-tial non-fluorescent cis state. The rate of inter-change between the trans and cis state was found

to be proportional to the excitation rate. Very .recently Widengren et al., 2001 , this behaviorhas been used to extract the FRET efficiency by

.analyzing the acceptor fluorescence Cy5 usingfluorescence correlation spectroscopy. Further-more, the excitation spectra of the cis and transforms overlap significantly, i.e. the donor mighttransfer its energy to the acceptor in the cis state.Hence, the donor intensity is reduced but thereduction is not reflected as an increase of accep-tor fluorescence, as expected for FRET. Since,

the isomerization rates are fast compared to the .observation time in our experiments 1 msrbin

the cistrans fluctuations are smeared out. Hence,

dependent on how FRET efficiencies are calcu-lated, lower values might result. However, experi-ments with the rigid rhodamine derivative JA133as acceptor showed comparable FRET efficien-

.cies, and large zero peaks data not shown . Thisindicates that the cistrans isomerization of Cy5molecules is not responsible for the observeddeviations from theoretically predicted efficien-cies.

The appearance of two real maxima in most ofthe FRET distribution histograms instead of

broad distributions implies that the conformatio-nal fluctuations between sub-states with differentorientations or DrA distances are on average

.slower than the measurement time 1 ms . If theconformational fluctuation rates were faster than

the measurement time which equals the diffu-.sion time , the high FRET peak would at least be

smeared out or gradually shifted to lower FRETefficiency.

Table 2 gives the calculated spFRET efficien-cies, Esp, and distribution widths, w, obtainedfrom Gaussian fits. There are several effects that

contribute to the observed peak broadening. Es-pecially the low signal intensity obtained fromsingle molecule measurement raises strong fluc-

Table 2spSingle-pair FRET efficiencies, E , and standard deviations,

w, revealed from Gaussian fits of the spFRET distributions

spE w

.TMR- N 0.01 0.1340 .TMR- N -Cy5 0.99 0.115 .TMR- N -Cy5 0.58 0.4915 .TMR- N -Cy5 0.21 0.3225

.R6G- N 0.01 0.0740 .R6G- N -Cy5 0.97 0.105 .R6G- N -Cy5 0.52 0.5415 .R6G- N -Cy5 0.19 0.1325

.Cy3- N 0.02 0.1740 .Cy3- N -Cy5 0.95 0.115 .Cy3- N -Cy5 0.48 0.3815 .Cy3- N -Cy5 0.24 0.3525


13/21


tuations in the FRET efficiency calculated fromintensity ratios. Another source of broadening isfluctuations in DrA distance and fluctuations in

2

, the orientation factor. Although a Gaussian fitis not optimal, at least for very low and highFRET efficiencies, it does not lead to major dis-crepancies. Surprisingly, spFRET efficiencies dif-fer substantially from the ensemble values. Inparticular, the 5-bp constructs exhibit a drasticallyincreased spFRET efficiency, i.e. as expected forsuch a short DrA distance. On the other hand,ensemble measurements show overall lower effi-ciencies at short distances which we ascribed toan additional efficient quenching of the donor.

These data clearly demonstrate the power ofspFRET measurements to reveal subpopulationsfrom heterogeneous ensembles.

2.5. Donorracceptor distance model

To compare the measured FRET efficiencieswith Forster theory, the bp separation of donorand acceptor was converted into distances usingthe model introduced for DNA-FRET efficiencies

.by Clegg et al. 1993 . According to this model, .the DrA distances R A for two dyes attachedDA

at a B form double helix in solution can beestimated from the structural properties of thedyes and the vector sum of two components: oneparallel; and one perpendicular to a cylindrical B

form DNA model with a typical diameter of 20 A,3.4 A rise and 36 per base pair:

2 2 2 . 'R s Kq 3.4N q L qL y 2L LDA D A D A . ..cos q 36N 5

Here, N represents the base pair separation;and K the distance between donor and acceptoralong the helical axis for Ns 0; L and L areD Athe normal distances of the donor and acceptorchromophore to the helical axis, respectively; and is the inter-dye angular separation for Ns 0.Using this helix model, i.e. describing thedouble-stranded DNA as a cylinder, the distancebetween the donor and acceptor is smaller whenboth chromophores are on the same side of thehelix, and longer when they are on opposite sides.

However, if only one of the dyes is close to thehelix, e.g. the dye adheres to the DNA indicatedby high anisotropy values, this is no longer the

case, and the modulation disappears. Based onthe structural properties of the dyes, the tethersand the linkers, we used two extreme start posi-

.tions: a fully stretched linkers with Ks 4.0 A, L s 22.0 A, L s22.0 A, L s29.2 A,TMR R6G Cy3

L s 29.2 A, L s 26.8 A and s 306; andCy5 JA133 .b Ks 4.0 A, L s 12.0 A, L s12.0 A,TMR R6G

L s 12.0 A, L s 12.0 A, L s 12.0 A andCy3 Cy5 JA133 s 306. This assumes that all chromophores re-fold and adhere to the DNA. Furthermore, weused three different R values for all DrA pairs0 .R as calculated from ensemble spectra

"

10%0to calculate the expected FRET efficiencies forthe series of constructs.

Fig. 7 shows the theoretically expected FRETefficiencies for differently labeled DNA con-structs assuming different R values and differ-0ent linker conformations. However, obviously nomodel can describe the measured ensemble val-ues accurately. If any, then the spFRET effi-ciencies can be described approximately by amodel assuming the dyes adhered to the DNA.The most striking difference from theory is the

relatively high FRET efficiencies measured forthe 25- and 35-bp construct. On the other hand,the FRET efficiencies retrieved from singlemolecule measurements with the 5-bp constructmatch nearly ideally the theoretical predicted val-ues.

2.6. Competing energy transfer mechanisms

Table 3 shows the ensemble fluorescenceproperties of the acceptor dyes in the FRET

.constructs excited at 635 nm direct excitation .Interestingly, the Cy5 fluorescence intensity de-creases upon binding of the donor, independentof the donor dye used. While for the 35-, 25- and15-bp constructs comparable quenching effi-ciencies are observed, quenching increases con-siderably in the 5-bp constructs. In contrast, theaverage fluorescence lifetimes increase slightly.On the other hand, the fluorescence of the accep-tor dye JA133 is not influenced by the donor dyeat longer distances. Once again at short distance


14/21


Fig. 7.


15/21


Table 3A A .Ensemble fluorescence properties relative fluorescence quantum yield, and fluorescence lifetime, of the acceptor dyesf,rel

upon direct excitation at 635 nm

A A A A f,rel 1 2 av . . .ns ra ns ra ns1 2

.N -Cy5 1.00 1.62r0.86 0.62r0.14 1.5635 .TMR- N -Cy5 0.83 1.65r0.59 0.81r0.41 1.4335 .TMR- N -Cy5 0.83 1.72r0.67 0.74r0.33 1.5425 .TMR- N -Cy5 0.86 1.67r0.62 0.74r0.38 1.4715 .TMR- N -Cy5 0.66 1.83r0.75 0.75r0.25 1.705

.R6G- N -Cy5 0.67 1.78r0.49 0.91r0.51 1.4835 .R6G- N -Cy5 0.77 1.82r0.59 0.82r0.41 1.5825 .R6G- N -Cy5 0.71 1.88r0.48 0.94r0.52 1.5515 .R6G- N -Cy5 0.50 1.80r0.78 0.62r0.22 1.695

.Cy3 N Cy5 0.84 1.65r0.73 0.67r0.27 1.5235 .Cy3 N Cy5 0.85 1.69r0.83 0.50r0.17 1.6225 .Cy3 N Cy5 0.91 1.72r0.73 0.76r0.27 1.5915 .Cy3 N Cy5 0.68 1.88r0.77 0.69r0.23 1.765

.N JA133 1.00 3.99r1.00 3.9935 .TMR N JA133 1.00 3.91r1.00 3.9135 .TMR N JA133 1.00 3.89r1.00 3.8925 .TMR N JA133 1.00 3.89r1.00 3.8915 .TMR N JA133 0.33 3.91r0.87 1.16r0.13 3.805

D .Absolute quantum yields of the acceptor only labeled oligonucleotides were determined to be s0.40 for N -Cy5 andf 35D . s0.80 for N -JA133.f 35

.5 bp the quantum yield is drastically reduced.Here a shorter fluorescence lifetime component

.appears in the time-resolved data Table 3 . How- .ever, the small amplitude 13% of the shorter

component of 1.16 ns can not explain the reduc-tion of the fluorescence quantum yield of 67%.Comparison of the measured intensity and

lifetime data of the acceptor fluorescence Table.3 suggest that, at least in the 5-bp construct, an

additional very efficient quenching process has to

be taken into account.Together with the observed shifts in the ab-

sorption and emission maxima the data strongly

imply direct ground and excited state interactionsbetween the donor and acceptor molecules, en-abled by the short separation distance and thelinker flexibilities. The idea of direct interactions,e.g. aggregation of donor and acceptor, is additio-nally supported by the measured high anisotropy

values for the dyes in the 5-bp constructs. It iswell known, that ionic dyes tend to aggregate,even in diluted aqueous solutions, to form non-

fluorescent dimers Drexhage, 1977; Kemnitz et

.al., 1986; Liang et al., 1997 . Comparison of themolecular structures of the donor and acceptor

.dyes Fig. 1 indicates that the indocarbocyanine

. . .Fig. 7. Theoretically expected FRET efficiencies for the FRET pairs: a TMRrCy5; b R6GrCy5; and c Cy3rCy5 attached toa double stranded DNA. Distances are calculated by the model shown above. Based on the structural properties of the dyes and the

. .linkers two extreme start positions were used: left side fully stretched linkers; and right side collapsed linkers. Furthermore, .three different R values have been used R as calculated from ensemble spectra "10% . Open circles represent the FRET0 0

efficiencies ED obtained from the decrease of donor intensity in ensemble measurements; open triangles represent EA , obtainedfrom ensemble donor and acceptor intensities; and the black squares are the FRET efficiencies calculated from single moleculedata. The error bars represent the standard deviation.


16/21


dyes Cy3 and Cy5 exhibit the most pronouncedwater solubility. Due to the additional negativelycharged sulfonate groups, intermolecular interac-

tions in water such as formation of dimers arereduced. Hence, the observed additional quench-ing process should be reduced in the Cy3rCy5pair. This idea is strongly supported by the datashown in Table 1 and Table 3. The FRET effi-ciencies for the Cy3rCy5 constructs calculated

via the two different methods exhibit the smallestdifferences. In the other FRET pairs TMRrCy5,

.R6GrCy5, TMRrJA133 , intermolecular interac-tions between the donor and acceptor dye aremore likely. The absorption and emission maxima

shift slightly, and provide additional quenchingpathways such as dimer formation in the case ofthe 5-bp constructs. This reduces the observedquantum yield for both the donor and acceptor.Due to the fact that dimers are essentially non-fluorescent, the measured ensemble fluorescencelifetimes are not reduced. Hence, the donor en-ergy cannot be transferred completely to the ac-ceptor via FRET. Since the subpopulation ofnon-fluorescent or only weakly fluorescent dimersare not detected in single molecule experimentsdue to thresholding, higher FRET efficiencies are

calculated.The formation of non-fluorescent intermolecu-

lar ground state complexes cannot explain thereduced fluorescence quantum yield of the accep-tor Cy5 at longer separation distances. Several

reports have indicated Gasper and Schuster,.1997; Norman et al., 2000; Schuster, 2000 that in

5-labeled double-stranded DNA, the chro-mophores are associated with DNA by end cap-ping, i.e. the chromophore is stacked onto the endof the helix, in a manner similar to that of an

additional base pair. Therefore, efficient chargetransfer via the base pairs of DNA might bepromoted, even at longer distances. The possibil-ity that the -stacked base pairs of DNA mightmediate charge transfer was suggested over 30

.years ago Eley and Spivey, 1962 . Experimentalinvestigations and theoretical treatments ofphoto-induced charge transfer in DNA have re-

vealed the occurrence of at least two mechan-isms: a single step superexchange mechanism

which is strongly distance-dependent; and a

multi-step hole hopping mechanism which is onlyweakly distance-dependent Jortner et al., 1998;

.Bixon et al., 1999; Lewis et al., 2001 . In the

hopping mechanism, a hole which is generallyassumed to be a guanosine radical cation gener-ated photochemically via electron transfer to anexcited chromophore can reversibly hop from oneguanosine to another until it reaches a trap site .more than four subsequent ArT base pairs .

The energetics of photo-induced charge separa-tion and charge recombination processes can be

w .xestimated by using Wellers equation Eq. 6 ,

.G sE yE yE q C 6cs ox red 0,0

where E and E are the first one-electronox redoxidation potential of the donor and the firstone-electron reduction potential of the acceptorin the solvent under consideration. E is the0,0energy of the zerozero transition to the lowestexcited singlet state of the excited partner, and Cis the solvent-dependent Coulombic attraction

.energy Weller, 1982 . The value of C in highlypolar solvents like water is sufficiently small thatit can be neglected. Electrochemical measure-ments were made for R6G, JA133 and Cy5 in

acetonitrile solution at a glassy carbon electrode.Table 4 gives the redox potentials measured vs.

.the saturated calomel electrode SCE and thecorresponding transition energies of the dyes. The

Table 4First one-electron oxidation and reduction potentials, E andox

E , of R6G, Cy5 and JA133 in acetonitrile and correspond-reding zerozero transition energies, E0,0

E E Eox red 0,0 . . .VrSCE VrSCE eV

R6G 1.39 y0.95 2.27Cy5 0.82 y0.88 1.88JA133 1.20 y0.71 1.96

The redox potentials were determined by cyclic voltamme- .try CV in dry acetonitrile purchased from Aldrich. Tetra-n-

w x .butylammonium hexafluorophosphate TBA PF was used as6the supporting electrolyte. 0.1 M solutions of the electrolyte

were purified and dried using neutral aluminum oxide ICN.alumina N, super I . Measurements were performed in a three

electrode arrangement in a single cell at 20C. The scan speedwas 100 mVrs. The positive and negative voltage limits of thesystem are 2.4 V and y2.8 V vs. SCE in acetonitrile.


17/21


first one-electron oxidation potential for de-oxyguanosine E s 1.25 V vs. SCE in acetonitrileox

.was taken from Seidel et al. 1996 . The free

energy changes G for photo-induced electroncstransfer from a ground state guanosine residue tothe excited dye are: y0.07 eV for R6G; 0.00 eVfor JA133; and q0.25 eV for Cy5. These thermo-dynamic data support the experimentally observedbehavior, that among the dyes used in this study,only R6G and the relatively similar dye TMR are

quenched by guanosine residues Sauer et al.,1995; Nord et al., 1997; Eggeling et al., 1998;

.Lieberwirth et al., 1998 . Therefore, we used DNAconstructs in which the first GrC base pair is

located 6 bp away from the 5-terminus where thedonor dyes R6G, TMR and Cy3 were attached.The acceptor seems to be quenched due to the

presence of the donor at the DNA and vice versadue to another additional quenching process.

.From the measured redox potentials Table 4 ,Cy5 is the strongest electron donor among thedyes used. The free energy change G forcsphoto-induced electron transfer from Cy5 to R6Gis y0.50 eV if R6G is excited, and y0.11 eV ifthe Cy5 chromophore is excited. Therefore, effi-cient charge transfer might occur at least at short

separation distance like in the 5-bp constructs viadirect through space interactions Tierney et al.,

.2000 . If so, the fluorescence of the excited donorwould be quenched supplementary to the fluo-rescence resonant energy transfer to the acceptor.The fluorescence of the excited acceptor could bequenched in the same way, via charge transfer tothe ground state donor. The free energy changesfor charge transfer between a rhodamine donor . .TMR, R6G and a rhodamine acceptor JA133are only slightly exergonic, if at all. Nevertheless,

at short distances, strong fluorescence quenching .of the acceptor occurs Table 3 . This might beexplained by strong intermolecular interactionsbetween the two rhodamine dyes due to theirrelatively hydrophobic structure. At longer dis-

.tances 15, 25, 35 bp , the acceptor JA133 exhibitsnearly uninfluenced fluorescence characteristics.In strong contrast, the acceptor Cy5 exhibits areduced fluorescence quantum yield even at

.longer DrA distance Table 3 . Here we assumethat a subpopulation of DrA constructs adopts a

conformation in which both donor and acceptorinteract strongly with the -stack of the DNA,either by end capping or intercalation. Hence, the

DNA might facilitate photo-induced electrontransfer from Cy5 to the rhodamine. As already

.discussed by Kijima et al. 1998 , excited electronsof the donor might interact with the acceptor dyecovalently bound to the opposite end of a DNAby transfer across over 1000 base pairs. Depen-dent on the conformation of the dyes with respectto the -stack of the DNA, fluorescence of theexcited donor or acceptor might be efficientlyquenched. In single molecule experiments, thesesubpopulations cannot be detected due to the low

fluorescence quantum yields of these states andthe application of a threshold for data collection.Hence, the spFRET efficiencies obtained fromthe 5-bp construct are in accordance with the

theoretically expected FRET efficiencies Table.2; Fig. 7 .

In ensemble measurements, the only weaklyfluorescent subpopulation which undergoes effi-cient electron transfer and dimerization in combi-nation with the opening of new non-radiativedeactivation channels influences the calculated

FRET efficiencies in a way that the FRET effi-ciencies appear to be to low for short DrA dis-tances. On the other hand, at longer DrA dis-tances the different extent of the electron trans-fer efficiencies for the donor and acceptor alsofalsify the calculated ensemble FRET efficiencies.Since the free energy change for photo-inducedelectron transfer between rhodamines and Cy5 ismore negative for the excited donor than theexcited acceptor, the FRET efficiency calculatedfrom the ensemble fluorescence intensities, might

indicate higher FRET efficiencies. With the as-sumption that the charge transfer process occursonly in subpopulations with optimal conforma-tions of the dyes with respect to the -stack ofthe DNA base pairs, very efficient charge separa-tion can result. From the reduced fluorescence

.intensities of the acceptor Table 3 , on averageapproximately 1030% of all FRET constructsexhibit such a favorable conformation in whichfast electron transfer mediated by the DNA seemsto be possible. Therefore, short fluorescence


18/21


lifetimes might appear which, however, are onlypoorly resolved with standard spectroscopy. Dueto thresholding and low fluorescence intensities,

this subpopulation appears as dim or non-fluo-rescent in single molecule experiments and is notdetected. Therefore, single molecule experimentsreveal more accurate FRET efficiencies unper-turbed by additional quenching effects.

3. Conclusions

We studied the FRET efficiencies of four dif-ferent DrA pairs covalently attached to a

double-stranded 40 base pair oligonucleotide byensemble and single molecule spectroscopy inaqueous solution. All ensemble measurements re-

vealed that, especially at short DrA distances, anadditional fluorescence quenching pathway forboth the donor and acceptor has to be taken intoaccount. Our data demonstrate that due to theconformational flexibility of the linkers, the donorand acceptor dye can directly interact in the 5-bpconstructs to form partly non-fluorescent or only

weakly fluorescent complexes. Furthermore, wecould show that the extent of the process is

strongly controlled by the water solubility of thedyes. Anisotropy measurements demonstrate thatnone of the dyes can be observed as a free rotor;in particular the 5-bp constructs exhibit unusuallyhigh anisotropy values. Nevertheless, the dyeschange their conformation with respect to theoligonucleotide but on a slower time scale in themillisecond range. This results in a dynamic inho-mogeneous distribution of DrA distances andorientations.

Comparison of the FRET efficiencies obtained

from ensemble and single molecule experimentsdemonstrates that the technique of spFRET ex-periments in solution is a powerful technique touncover subpopulations. spFRET also facilitatescorrect interpretation of ensemble measurements.In single molecule experiments only those fluo-rescence signals that exhibit a fluorescence inten-sity above the signal threshold used contribute tothe measured FRET efficiency. Therefore,strongly quenched populations are not detected.The measured redox properties of the dyes imply

the possibility of a photo-induced electron trans-fer reaction between Cy5 and rhodamine chro-mophores. Hereby, the carbocyanine derivative

acts as a strong electron donor, independent ofwhether the donor or acceptor is excited. Thiseffect might seriously affect the FRET efficien-

.cies at a short DrA distance 5 bp . For largerDrA distances we assume that dependent on theconformations of the dyes with respect to theDNA, an efficient charge transfer via the -stackof the base pairs occurs. These subpopulations,

which appear strongly quenched and undetectedin single molecule experiments, control the mea-sured FRET efficiencies in ensemble measure-

ments. Our results and considerations demon-strate that not only direct intermolecular interac-tions between the donor and acceptor at shortseparation distance, but also long-range DNAmediated interactions have to be taken into ac-count. Furthermore, our data show that spFRETexperiments are only suited to resolve adequatelyfluorescent subpopulations in heterogeneous mix-ture. Information about strongly quenched sub-populations gets lost. Consideration of the de-scribed competing quenching processes is impor-tant for any application of the FRET technology.

Acknowledgements

The authors thank J. Wolfrum for fruitfulcooperation and stimulating discussion, and K.H.Drexhage and J. Arden-Jacob for the generousdisposal of the rhodamine derivative JA133. Fi-nancial support by the Volkswagen-Stiftung .Grant Ir74 443 and the Bundesministerium furBildung, Wissenschaft, Forschung und Tech-

.nologie Grant 11864 BFA082 is gratefully ac-knowledged.

References

Agranovich, V.M., Galanin, M.D., 1982. Electronic ExcitationEnergy Transfer in Condensed Matter. North Holland,Amsterdam.

Arden-Jacob, J., 1992. Synthese neuer langwelliger Xanthen-farbstoffe, Ph.D Thesis, Siegen.

Bartko, A.P., Dickson, R.M., 1999. Imaging three-dimensionalsingle molecule orientations. J. Phys. Chem. B 103,1123711241.


19/21


Berlman, I.B., 1973. Energy Transfer Parameters of AromaticCompounds. Academic Press, New York.

Bixon, M., Giese, B., Wessely, S., Langenbacher, T., Michel-Beyerle, M.E., Jortner, J., 1999. Long-range charge hopping

in DNA. Proc. Natl. Acad. Sci. USA 96, 11713 11716.Bustamante, C., Marko, J.F., Siggia, E.D., Smith, S., 1994.

Entropic elasticity of lambda-phage DNA. Science 265,15991600.

Clegg, R.M., Murchie, A.I.H., Zechel, A., Lilley, D.M., 1993.Observing the helical geometry of double-stranded DNA insolution by fluorescence resonance energy transfer. Proc.Natl. Acad. Sci. USA 90, 29942998.

Dahan, M., Deniz, A., Ha, T. et al., 1999. Ratiometric identi-fication and separation of single molecules diffusing insolution. Chem. Phys. 47, 85106.

Dale, R.E., Eisinger, J., Blumberg, W.E., 1979. The orientatio-nal freedom of molecular probes. Biophys. J. 26, 161 194.

Deniz, A.A., Dahan, M., Grunwell, J.R. et al., 1999. Single-pairfluorescence energy transfer on freely diffusing molecules:Observation of Forster distance dependence and subpopu-lations. Proc. Natl. Acad. Sci. USA 96, 36703675.

Deniz, A.A., Laurence, T.A., Beligere, G.S. et al., 2000.Single-molecule protein folding: diffusion fluorescence en-ergy transfer studies of the denaturation of chymotrypsininhibitor 2. Proc. Natl. Acad. Sci. USA 97, 5179 5184.

Dexter, D.L., 1953. A theory of sensitized luminescence insolids. J. Chem. Phys. 21, 836 850.

.Drexhage, K.H., 1977. In: Schafer, F.P. Ed. , Topics in Ap-plied Physics, 1. Springer Verlag, Berlin Heidelberg, NewYork, pp. 144179.

Edman, L., Mets, U., Rigler, R., 1996. Conformational transi-tions monitored for single molecules in solution. Proc. Natl.Acad. Sci. USA 93, 67106715.

Eggeling, C., Fries, J.R., Brand, L., Gunther, R., Seidel,C.A.M., 1998. Monitoring conformational dynamics of asingle molecule by selective fluorescence spectroscopy. Proc.Natl. Acad. Sci. USA 95, 15561561.

Eley, D.D., Spivey, D.I.T., 1962. Semiconductivity of organicsubstances. Trans. Faraday Soc. 58, 411415.

English, D.S., Furube, A., Barbara, P.F., 2000a. Single-mole-cule spectroscopy in oxygen-depleted polymer films. Chem.Phys. Lett. 324, 1519.

English, D.S., Harbron, E.J., Barbara, P.F., 2000b. Probingphotoinduced intersystem crossing by two-color, double

resonance single molecule spectroscopy. J. Phys. Chem. A104, 90579061.

Forster, Th., 1948. Zwischenmolekulare Energiewanderungund Fluoreszenz. Annalen der Physik 2, 55 75.

.Forster, Th., 1968. In: Sinanoglu, O. Ed. , Modern QuantumChemistry. Academic Press, New York, p. 93.

Gasper, S.M., Schuster, G.B., 1997. Intramolecular photoin-duced electron transfer to anthraquinones linked to duplexDNA: the effect of gaps and traps on long-range radicalmigration. J. Am. Chem. Soc. 119, 1276212771.

Grunwell, J.R., Glass, J.L., Lacoste, T.D., Deniz, A.A., Chemla,D.S., Schultz, P.G., 2001. Monitoring the conformationalfluctuations of DNA hairpins using single-pair fluorescence

resonance energy transfer. J. Am. Chem. Soc. 123,42954303.

Ha, T., Enderle, T., Ogletree, D.F., Chemla, D.S., Selvin, P.R.,Weiss, S., 1996a. Probing the interaction between two sin-

gle molecules: fluorescence resonance energy transferbetween a single donor and a single acceptor. Proc. Natl.Acad. Sci. USA 93, 62646268.

Ha, T., Enderle, Th., Chemla, D.S., Selvin, P.R., Weiss, S.,

1996b. Single molecule rotational and translational diffu-sion observed by near-field scanning optical microscopy.Phys. Rev. Lett. 77, 39793982.

Ha, T., Ting, A.Y., Liang, J. et al., 1999a. Single-molecule

fluorescence spectroscopy of enzyme conformational dy-

namics and cleavage mechanism. Proc. Natl. Acad. Sci.USA 96, 893898.

Ha, T., Zhuang, X., Kim, H.D., Orr, J.W., Williamson, J.R.,

Chu, S., 1999bb. Ligand-induced conformational changes

observed in single RNA molecules. Proc. Natl. Acad. Sci.USA 96, 90779082.

Ishii, Y., Yoshida, T., Funatsu, T., Wazawa, T., Yanagida, T.,

1999. Fluorescence resonance energy transfer between sin-

gle fluorophores attached to a coiled-coil protein in aque-ous solution. Chem. Phys. 247, 163 173.

Jia, Y., Sytnik, A., Li, L., Vladimirov, S., Cooperman, B.S.,

Hochstrasser, R.M., 1997. Nonexponential kinetics of asingle tRNAPhe molecules under physiological conditions.Proc. Natl. Acad. Sci. USA 94, 79327936.

Jia, Y., Talaga, D.S., Lau, W.L., Lui, H.S.M., DeGrado, W.F.,

Hochstrasser, R.M., 1999. Folding dynamics of single GCN4

peptides by fluorescence resonant energy transfer confocal

microscopy. Chem. Phys. 247, 6983.Jortner, J., Bixon, M., Langenbacher, T., Michel-Beyerle, M.E.,

1998. Charge transfer and transport in DNA. Proc. Natl.Acad. Sci. USA 95, 1275912765.

Kelley, S.A., Barton, J.K., 1999. Electron transfer betweenbases in double helical DNA. Science 283, 375 381.

Kemnitz, K., Tamai, N., Yamazaki, Y., Nakashima, N., Yoshi-

hara, K., 1986. Fluorescence decays and spectral properties

of rhodamine B in submono-, mono- and multilayer sys-tems. J. Phys. Chem. 90, 50945101.

Knemeyer, J.P., Marme, N., Sauer, M., 2000. Probes for detec-tion of specific DNA sequences at the single-moleculelevel. Anal. Chem. 72, 37173724.

Kijima, H., Spataru, N., Kawata, Y., Yano, S., Vartires, I.,

1998. Long-ranged electron interaction between carboxyte-

tramethylrhodamine and fluoresceinisothiocyanate boundcovalently to DNA, as evidenced by fluorescence quench-ing. J. Phys. Chem. B 102, 99819984.

Lewis, F.D., Wu, T., Zhang, Y., Letsinger, R.L., Greenfield,

S.R., Wasielewski, M.R., 1997. Distance-dependent elec-tron transfer in DNA hairpins. Science 277, 673676.

Lewis, F.D., Letsinger, R.L., Wasielewski, M.R., 2001. Dy-

namics of photoinduced charge transfer and hole transportin synthetic DNA hairpins. Acc. Chem. Res. 34, 159170.

Liang, K., Farahat, M.S., Perlstein, J., Law, K.Y., Whitten,

D.G., 1997. Exciton interactions in nonconjugated squaraine


20/21


dimers. Mechanisms for coupling and consequences forphotophysics and photochemistry. J. Am. Chem. Soc. 119,830831.

Lieberwirth, U., Drexhage, K.H., Herten, D.P. et al., 1998.

Multiplex dye DNA sequencing in capillary gel elec- .trophoresis CGE with diode laser based time-resolved

fluorescence detection. Anal. Chem. 70, 47714779.

Lu, H.P., Xie, X.S., 1997. Single-molecule spectral fluctuations .at room temperature. Nature Lond. 385, 143146.

Mataga, N., Kubota, T., 1970. Molecular Interactions andElectronic Spectra. Dekker, New York.

Meggers, E., Michel-Beyerle, M.E., Giese, B., 1998. Sequencedependent long range hole transport in DNA. J. Am.Chem. Soc. 120, 1295012955.

Nord, S., Sauer, M., Arden-Jacob, J. et al., 1997. Ground andexcited state reactions of new red fluorescent dyes and theDNA base guanosine. J. Fluoresc. 7, 79S 81S.

Norman, D.G., Grainger, R.J., Uhrin, D., Lilley, D.M.J., 2000.Location of cyanine-3 on double-stranded DNA: impor-tance for fluorescence resonance energy transfer studies.Biochemistry 39, 63176324.

Osborne, M.A., Barnes, C.L., Balasubramanian, S., Klener-man, D., 2001. Probing DNA surface attachment and localenvironment using single molecule spectroscopy. J. Phys.Chem. B 105, 31203126.

Parkhurst, K.M., Brenowitz, M., Parkhurst, L.J., 1996. Simul-taneous binding and bending of promoter DNA by theTATA binding protein: real time kinetic measurements.Biochemistry 35, 74597465.

Ruiter, A.G.T., Veerman, J.A., Garcia-Parajo, M.F., van Hulst,N.F., 1997. Single molecule rotational and translational

diffusion observed by near-field scanning optical micros-copy. J. Phys. Chem. A 101, 7318 7323.

Sauer, M., Han, K.T., Muller, R. et al., 1995. New fluorescentdyes in the red region for biodiagnostics. J. Fluoresc. 5,247261.

Sauer, M., Drexhage, K.H., Lieberwirth, U., Muller, R., Nord,S., Zander, C., 1998. Dynamics of electron-transfer reac-tions between xanthene dyes and DNA nucleotides moni-tored on the single molecule level. Chem. Phys. Lett. 284,153163.

Schuster, G.B., 2000. Long-range charge transfer in DNA:transient structural distortions control the distance depen-dence. Acc. Chem. Res. 33, 253260.

Selvin, P., 2000. The renaissance of fluorescence resonanceenergy transfer. Nature Struct. Biol. 7, 730 734.

Seidel, C.A.M., Schulz, C., Sauer, M., 1996. Nucleobase-specificquenching of fluorescent dyes. 1. Nucleobase one-electronredox potentials and their correlation with static and dy-namic quenching efficiencies. J. Phys. Chem. A 100,55415553.

Speiser, S., 1996. Photophysics and mechanisms of intramolec-ular electronic energy transfer in bichromophoric molecu-lar systems: solution and supersonic jet studies. Chem. Rev.96, 19531976.

Steenken, S., Jovanovic, S.V., 1997. How easily oxidizable isDNA? One-electron reduction potentials of adenosine and

guanosine radicals in aqueous solution. J. Am. Chem. Soc.119, 617618.

Stryer, L., Haugland, R.P., 1967. Energy transfer: a spectros-copic ruler. Proc. Natl. Acad. Sci. USA 58, 719 730.

Stryer, L., 1978. Fluorescence energy transfer as a spectros-copic ruler. Ann. Rev. Biochem. 47, 819846.

Stuhmeier, F., Welch, J.B., Murchie, A.I.H., Lilley, D.M.J.,Clegg, R.M., 1997. Global structure of three-way DNA

junctions with and without additional unpaired bases: a

fluorescence resonance energy transfer analysis. Biochem-istry 36, 1353013538.

Szollosi, J., Damjanovich, S., Matyus, L., 1998. Application of fluorescence resonance energy transfer in the clinicallaboratory: routine and research. Cytometry 34, 159179.

Tierney, M.T., Sykora, M., Khan, S.I., Grinstaff, M.W., 2000.

Photoinduced electron transfer on an oligodeoxynucleotide

duplex: observation of the electron-transfer intermediate.

J. Phys. Chem. B 104, 75747576.Tinnefeld, P., Buschmann, V., Herten, D.P., Han, K.T., Sauer,

M., 2000. Confocal fluorescence lifetime imaging micros- .copy FLIM at the single molecule level. Single Mol. 3,

215223.

Tinnefeld, P., Herten, D.P., Sauer, M., 2001. Photophysical

dynamics of single dye molecules studied by spectrally-re-

solved fluorescence lifetime imaging microscopy. J. Phys.Chem. A. 105, 79898003.

Tsien, R.Y., 1998. The green fluorescent protein. Annu. Rev.Biochem. 67, 509544.

Turro, N.J., 1978. Modern Molecular Photochemistry. Ben-jamin, London.

Tuschl, T.C., Gohlke, T.M., Jovin, E., Westhofand, E., Eck-stein, F., 1994. A three-dimensional model for the hammer-

head ribozyme based on fluorescence measurements. Sci-ence 266, 766785.

Vamosi, G., Gohlke, C., Clegg, R.M., 1996. Fluorescence`characteristics of 5-carboxytetramethylrhodamine linked

covalently to the 5 end of oligonucleotide: multiple con-

formers of single-stranded and double-stranded dye DNAcomplexes. Biophys. J. 71, 972994.

Veatch, W., Stryer, L., 1977. The dimeric nature of gramicidin

A transmembrane channel: conductance and fluorescence

energy transfer studies of hybrid channels. J. Mol. Biol.113, 89102.

Veerman, J.A., Garcia-Parajo, M.F., Kuipers, L., van Hulst,

N.F., 1999. Time-varying triplet state lifetimes of singlemolecules. Phys. Rev. Lett. 83, 21552158.

Vogel, M., Rettig, W., Sens, R., Drexhage, K.H., 1988. Struc-

tural relaxation of rhodamine dyes with different n-sub-

stitution patterns: a study of fluorescence decay times andquantum yields. Chem. Phys. Lett. 147, 452460.

Weiss, S., 1999. Fluorescence spectroscopy of single bio-molecules. Science 283, 16761683.

Weiss, S., 2000. Measuring conformational dynamics of bio-

molecules by single molecule fluorescence spectroscopy.Nature Struct. Biol. 7, 724729.

Weller, A., 1982. Photoinduced electron transfer in solution:


21/21


exciplex and radical ion formation free enthalpies and theirsolvent dependence. Z. Phys. Chem. Neue Folg. 133, 93 98.

Wennmalm, S., Edman, L., Rigler, R., 1997. Conformationalfluctuations in single DNA molecules. Proc. Natl. Acad.

Sci. USA 94, 1064110646.Weston, K.D., Buratto, S.K., 1998. Millisecond intensity fluc-

tuations of single molecules at room temperature. J. Phys.Chem. A 102, 36353638.

Weston, K.D., Carson, P.J., DeAro, J.A., Buratto, S.K., 1999.Single-molecule detection fluorescence of surface-boundspecies in vacuum. Chem. Phys. Lett. 308, 5864.

Weston, K.D., Goldner, L.S., 2001. Orientation imaging andreorientation dynamics of single dye molecules. J. Phys.Chem. B 105, 34533462.

Widengren, J., Dapprich, J., Rigler, R., 1997. Charge transferand stacking reactions between Rh6G and dGTP in water a fluorescence correlation spectroscopy study. Chem.

Phys. 216, 417

426.Widengren, J., Schwille, P., 2000. Characterization of photoin-duced isomerization and back-isomerization of the cyanine

dye Cy5 by fluorescence correlation spectroscopy. J. Phys.Chem. A 104, 64166428.

Widengren, J., Schweinberger, E., Berger, S., Seidel, C.A.M.,

2001. Two new concepts to measure fluorescence reso-

nance energy transfer via fluorescence correlation spectros-

copy theory and experimental realizations, J. Phys.Chem., submitted for publication.

Ye, Y.-J., Jiang, Y., 2000. Electronic structures and long-range

electron transfer through DNA molecules. Int. J. Quant.Chem. 78, 112130.

Ying, L., Wallace, M.I., Balasubramanian, S., Klenerman, D.,

2000. Ratiometric analysis of single-molecule fluorescence

resonance energy transfer using logical combinations of

threshold criteria: a study of 12-mer DNA. J. Phys. Chem.B 104, 51715178.

Yip, W.T., Hu, D., Yu, J., Vanden Bout, D.A., Barbara, P.F.,

1998. Classifying the photophysical dynamics of single- and

multiple-chromophoric molecules by single molecule spec-troscopy. J. Phys. Chem. A 102, 75647575.

Documents

Anja Dietrich2002