Two-sided fluorescence resonance energy transfer for assessing molecular interactions of up to three...

Two-Sided Fluorescence Resonance Energy Transfer

for Assessing Molecular Interactions of up to Three

Distinct Species in Confocal Microscopy

Zsolt Fazekas,1,y Miklos Petras,2,y Akos Fabian,2 Zsuzsanna Palyi-Krekk,2 Peter Nagy,2

Sandor Damjanovich,1,2 Gyorgy Vereb,2 Janos Szollo†si1,2*

� AbstractThe role of the expression patterns of proteins involved in oncogenesis can be under-stood after characterizing their multimolecular interactions. Conventional FRETmeth-ods permit the analysis of interaction between two molecular species at the most, whichnecessitates the introduction of new approaches for studying multicomponent signalingcomplexes. Flow cytometric as well as microscopic donor (dbFRET) and acceptor(abFRET) photobleaching FRETmeasurements were performed to determine the asso-ciation states of ErbB2, b1-integrin, and CD44 receptors. Based on consecutivelyapplied abFRET and dbFRETmethods (two-sided FRET), the relationship of b1-integ-rin–ErbB2 heteroassociation to ErbB2 homoassociation and of b1-integrin–ErbB2 het-eroassociation to ErbB2–CD44 heteroassociation was studied by correlating pixel-by-pixel FRET values of the corresponding abFRET and dbFRET images in contour plots.Anticorrelation was observed between b1-integrin–ErbB2 heteroassociation and ErbB2homoassociation on trastuzumab sensitive N87 and SK-BR-3 cells, while modest posi-tive correlation was found between b1-integrin–ErbB2 and ErbB2–CD44 heteroassocia-tion on trastuzumab resistant MKN-7 cells. The FRET efficiency values of b1-integrin–ErbB2 heteroassociation were markedly higher at the focal adhesion regions on attachedcells than those measured by flow cytometry on detached cells. In conclusion, we imple-mented an experimental set-up termed two-sided FRET for correlating two pairwiseinteractions of three arbitrarily chosen molecular species. On the basis of our results,we assume that the homoassociation state of ErbB2 is dynamically modulated by itsinteraction with b1-integrins. ' 2007 International Society for Analytical Cytology

� Key termsreceptor tyrosine kinase; ErbB2, b1-integrin; CD44; trastuzumab resistance; laser scan-ning confocal microscopy; fluorescence resonance energy transfer; tsFRET; dbFRET;abFRET

UNDERSTANDING the molecular mechanisms of signal transduction processes

requires the elucidation of assembling and/or disassembling of protein complexes

including the conformational changes accompanying these processes. One of the best

approaches for studying these molecular rearrangements is fluorescence resonance

energy transfer (FRET), which is a sensitive method for measuring intra- or intermo-

lecular distances. FRET-based methods are capable of resolving molecular associa-

tions and conformational changes in the 1–10 nm range far exceeding the diffraction

limit of the conventional microscope. FRET is a nonradiative process in which energy

is transferred from the first excited electronic state of a donor molecule (D) to a

nearby acceptor molecule (A) in a dipole–dipole interaction under favorable condi-

tions, resulting in the quenching of donor fluorescence accompanied by the increase

(also known as sensitization) of acceptor fluorescence (1–3). For mapping the physi-

cal associations of various membrane molecules, flow cytometric and microscopic

1Cell Biology and Signaling ResearchGroup of the Hungarian Academy ofSciences, Research Center for MolecularMedicine, Medical and Health ScienceCenter, University of Debrecen, Hungary2Department of Biophysics and CellBiology, Research Center for MolecularMedicine, Medical and Health ScienceCenter, University of Debrecen, Hungary

Received 20 July 2007; Accepted 8October 2007

This article contains supplementalmaterial which is available atwww.interscience.wiley.com/jpages/1552-4922/suppmatyZsolt Fazekas and Mikl�as Petr�ascontributed equally to this paper.

*Correspondence to: J�anos Sz€ollo†si,Department of Biophysics and CellBiology, Faculty of Medicine, Medicaland Health Science Center, University ofDebrecen, P.O. Box 39, Nagyerdei krt.98, H-4012 Debrecen, Hungary

Email: [email protected]

Published online 28 November 2007 inWiley InterScience(www.interscience.wiley.com)

DOI: 10.1002/cyto.a.20489

Original Article

Cytometry Part A � 73A: 209�219, 2008

applications of the FRET phenomenon have been established

(4). The flow cytometric FRET (FCET) technique can provide

statistically accurate information on the distribution of energy

transfer efficiency on thousands of individual cells (5,6). In

contrast, microscopic FRETmeasurements are statistically less

accurate, nevertheless, they can reveal the heterogeneity in cell

surface distributions and interactions of membrane receptors

within a single cell (7–9).

Disregulation of signal transduction processes starting

from assemblies of cell surface molecules contributes to the

pathogenesis of various cancers. In this respect, members of

the ErbB family of transmembrane receptor tyrosine kinases

have been studied extensively (10–13). It has been revealed

that certain tumor cells express multiple ErbB members,

which can homo- and heteroassociate during signal transduc-

tion, leading to receptor crossactivation, and induction of

downstream signaling (14). One of the ErbB proteins, ErbB2,

was found to be the preferred heteroassociation partner of all

other ErbB proteins, playing a mediator role in lateral signal

transmission and transmembrane signaling by its intrinsic ty-

rosine kinase activity (15,16). In addition to the association

pattern of the ErbB molecules, tumor development, the prolif-

eration, and survival of cells also depend on the interactions

between cells and their environment, including the extracellu-

lar matrix (ECM). Many of these interactions are mediated by

the integrin family of adhesion receptors, playing a pivotal

role in migration, attachment, cell cycle progression, pro-

grammed cell death, and regulation of other signal transduc-

tion pathways (17,18). Colocalization of and FRET efficiency

between ErbB2 and b1-integrin molecules revealed that these

molecules are in close vicinity (19,20). This finding provides

additional evidence for the crosstalk between certain integrins

and receptor tyrosine kinases suggesting that they play an im-

portant role in tumor progression (21). In addition, higher

order clusters of the ErbB2 molecule—containing hundreds of

proteins—have also been identified (22,23). Although these

ErbB2 clusters colocalize with lipid rafts (24), the degree of

homoassociation of ErbB2 is inversely proportional to both

the local density of the raft marker GM1 gangliozide and that

of ErbB3 (25). Thus, it appears that the local density/presence

of molecular species can correlate with molecular interactions

that eventually define the outcome of various signaling path-

ways, but not in a straightforward and easily predictable man-

ner. In simple FRETmeasurements, molecular interactions are

investigated in isolation, i.e., without regard to the influence

exerted by other proteins, which, considering the above, is an

oversimplification. In contrast, a direct demonstration at mi-

croscopic resolution of the (anti)correlation of two distinct

interactions sharing a common molecular species would vastly

enhance our understanding of signaling systems.

Therefore, we developed, and demonstrated the applic-

ability of, a method, two-sided FRET (tsFRET), to measure

the relationship between the association states of two mole-

cule-pairs of three arbitrarily chosen molecular species. Two-

sided FRET involves acceptor depletion (abFRET) and donor

photobleaching (dbFRET) measurements on the same sample

consecutively, using three appropriately chosen fluorophores.

The feasibility of tsFRETwas explored using three noncompet-

ing monoclonal antibodies binding to the same ErbB2 mole-

cule. Two-sided FRET was then used to reveal that b1-integ-rin–ErbB2 heteroassociation interfered with ErbB2–ErbB2

homoassociation, while ErbB2–b1-integrin heteroassociation

showed a modest positive correlation with ErbB2–CD44

heteroassociation. These examples show that tsFRET can be

used to reveal crossassociation and mutual interference of

molecular interactions on a pixel-by-pixel basis.

MATERIALS AND METHODS

Cell Culture

The gastric adenocarcinoma cell lines MKN-7 (Immuno-

Biological Laboratories Cell Bank, Gunma, Japan) and NCI-

N87 (#CRL-5822, American Type Culture Collection (ATCC),

Manassas, VA), and the breast cancer cell lines SK-BR-3

(#HTB-30, ATCC) and JIMT-1 (from Jorma Isola, University

of Tampere, Tampere, Finland) were grown according to their

specifications to subconfluency. For microscopic experiments,

cells were cultured on 12 mm diameter coverslips or on Lab-

Tek II chambered coverglass (Nalge Nunc International, Roch-

ester, NY). For flow cytometric measurements cells were tryp-

sinized before antibody labeling.

Antibodies

The following monoclonal antibodies were used: 2C4,

7C2 (gifts from Genentech, South San Francisco, CA) and

trastuzumab (Hoffman-La Roche AG, Germany) against

ErbB2. Monoclonal antibodies 528 (Mab 528) against ErbB1,

TS2 against b1-integrin, and Hermes3 against CD44 were pur-

ified from supernatants produced by the hybridoma cell lines

528 (IgG2a, #HB-8509), TS2/16.2.1 (IgG1, #HB-243), and

Hermes3 (IgG2a, #HB-9480) (ATCC, Manassas, VA), respec-

tively, using Sepharose 4B Fast Flow Protein G beads in the

case of IgG2a antibodies or Protein A beads in the case of

IgG1 antibodies (Sigma-Aldrich, St. Louis, MO). Labeling of

antibodies with Cy3 and Cy5 monoreactive sulfoindocyanine

N-hydroxysuccinimidyl esters (Amersham Biosciences Europe,

Freiburg, Germany), Alexa Fluor1 546, Alexa Fluor1 555,

Alexa Fluor1 647 monosuccinimidyl-esters and X-fluores-

cein-isothiocyanate (XFITC) (Molecular Probes, Invitrogen

Corp., CA) was carried out according to the manufacturers’

Grant sponsor: European Community; Grant numbers: EU FP6 LSHB-CT-2004-503467, EU FP6 LSHC-CT-2005-018914, EU FP6 MCRTN-CT-035946-2 and EU FP6 MRTN-CT-2005-019481; Grant sponsor: HungarianMinistry of Health; Grant number: ETT 066/2006; Grant sponsor:Hungarian National Research Fund; Grant numbers: OTKA K68763,

K62648 and F049025; Grant sponsors: Hungarian National Office forResearch and Technology, €Oveges grant, and Hungarian Ministry ofEducation and Culture.

© 2007 International Society for Analytical Cytology

ORIGINAL ARTICLE

210 Two-Sided FRET in CLSM

specifications. The dye to protein labeling ratio determined by

spectrophotometry was �3:1.

Labeling of Cells With Antibodies

For flow cytometric measurements, cells were trypsinized

followed by resuspending 1 3 106 cells in 50 ll PBS supple-

mented with 0.1% bovine serum albumin (BSA) and labeling

with a saturating concentration (10–20 lg/ml) of Alexa546- or

Alexa647-conjugated antibodies on ice (48C) for 30 min. After

labeling, they were washed three times in PBS and fixed in 500

ll of 1% formaldehyde-PBS.

For microscopic measurements, adherent cells were

washed three times in PBS, labeled with saturating concentra-

tions of XFITC-, Cy3-, Alexa555, or Cy5-conjugated antibo-

dies on ice for 30 min, then washed three times in PBS, and

fixed in 500 ll of 4% formaldehyde-PBS for 20 min. Cover-

slips were mounted in glycerin onto precleaned microscopic

slides, while samples on chambered coverglasses were stored

and measured under 500 ll of 1% formaldehyde-PBS.

Flow Cytometric Determination of Cell Surface

Receptor Expressions

Quantitation of receptor expression was carried out on a

FACSVantage SE with DiVa option (Becton Dickinson,

Franklin Lakes, NJ) using the FL-6 channel (excitation at 633

nm, detection through a 650 nm long pass filter) in linear

mode. The background corrected mean of histograms from

20,000 events was converted to antigen numbers using the

known expression (800,000 per cell) of ErbB2 on SK-BR-3

cells and the respective labeling ratios of the antibodies used.

Expression of ErbB2 on SK-BR-3 cells was quantified using

QIFIKIT (DakoCytomation, Glostrup, Denmark) according to

the manufacturer’s instructions.

Flow Cytometric Fluorescence Resonance Energy

Transfer Measurements

To determine the homo- and heteroassociation states of

the labeled receptors, flow cytometric fluorescence reso-

nance energy transfer measurements (FCET) were per-

formed either on the FACSVantage SE instrument with DiVa

option equipped with 488, 532, and 633 nm lasers or on a

FACSArray bioanalyzer (Becton Dickinson) equipped with

532 and 635 nm lasers. In experiments on the FACSVantage

SE instrument, the FRET efficiencies were calculated with

cell-by-cell correction for autofluorescence (5). The cellular

autofluorescence was measured in the FL1 channel (excited

by the 488 nm laser) through a 530/30 nm band pass filter.

Donor intensities were recorded in the FL4 channel through

a 585/42 nm bandpass filter, while acceptor and sensitized

acceptor intensities were detected in FL6 and FL5 channels,

respectively, through 650 nm long pass filters. The donor

(Alexa546) and acceptor (Alexa647) molecules were excited

by the 532 and 633 nm laser lines, respectively. In measure-

ments on the FACSArray bioanalyzer the donor intensities

(excited by the 532 nm laser) were recorded in the Yellow

channel through a 585/42 nm bandpass filter, while the

acceptor and sensitized acceptor intensities were excited by

the 633 nm and the 532 nm lasers, respectively, and were

detected in the Red and Far-Red channels through a 661/16

nm bandpass and a 685 nm long pass filter, respectively. List

mode data were stored in FCS 3.0 file format. Calculation of

cell-by-cell FRET efficiency was carried out using the soft-

ware REFLEX (26). FRET efficiencies are presented as mean

values—indicating the standard error of the mean—of

approximately normally distributed, unimodal FRET histo-

grams of 20,000 cells.

Confocal Microscopy

A Zeiss LSM 510 confocal laser-scanning microscope

(CLSM, Carl Zeiss AG, Jena, Germany) with a Plan-Apo-

chromat 633/1.40NA oil immersion objective and an UV/

488/543/633 beam splitter was used to image cells. XFITC,

Cy3, and Cy5 were excited at 488, 543, and 633 nm and

detected through 505–530 nm, 560–605 nm bandpass, and

650 nm longpass filters, respectively. Horizontal, 1–3 lmthick, four-times averaged optical sections (512 3 512 pixels

at 12 bit) were taken of either the top or bottom flat layer of

the cell membrane using 2.51 ls pixel dwell time and 33zoom (100 nm per pixel, two times oversampling) in frame

mode. Before acceptor photobleaching FRETmeasurements,

the optimal bleaching time for reducing the Cy5 fluores-

cence to less than 20% of the prebleach value was deter-

mined. Images were acquired in multitrack mode to avoid

crosstalk between channels. The alignment of detection

channels was regularly checked with samples of FocalCheck

fluorescent microspheres (Molecular Probes, Invitrogen

Corp., CA).

Donor Photobleaching Method

In donor photobleaching FRET introduced in 1989 by

Jovin and Arndt-Jovin (27), the energy transfer is calculated

by comparing the photobleaching kinetics of the donor in the

presence and absence of the acceptor. This kinetics can be

described by a time constant (s) which is inversely related to

the donor quantum yield. Assuming that the donor photo-

bleaching occurs from the excited singlet state, it can be

derived, that:

E ¼ 1� sblesAble

ð1Þ

where E denotes the FRET efficiency, sble and sAble are the

photobleaching time constants in the absence and presence,

respectively, of the acceptor. In these experiments, the speci-

mens were bleached by acquiring 40–50 images as time series

using the 488 nm Ar-ion laser at high (20–40%) power.

Images were median or low-pass filtered when necessary (3 33 kernel, LSM v 3.2 software), then exported and fit after

thresholding by the dbFRET software (28), yielding images of

the photobleaching time constants.

Acceptor Photobleaching Method

This method is based on irreversibly destroying the

acceptor fluorophore by photobleaching it and thereby

dequenching the donor molecule (29,30). FRET efficiency (E)

ORIGINAL ARTICLE

Cytometry Part A � 73A: 209�219, 2008 211

in each pixel was measured using the acceptor photobleaching

protocol (31) and calculated as follows:

Eði;jÞ ¼ 1�FBDði;jÞ � Savr4 � FB

Aði;jÞcavr : FA

Dði;jÞð2Þ

where FBD(i,j) and FA

D(i,j) are the background subtracted donor

fluorescence values before and after photobleaching the

acceptor, respectively, and FBA(i,j) denotes the background sub-

tracted acceptor fluorescence value of pixel (i,j) before photo-

bleaching the acceptor. cavr is a factor correcting for the photo-bleaching of the donor during the whole protocol, calculated

as the median of

cði;jÞ ¼FrefD;BDði;jÞ

FrefD;ADði;jÞ

ð3Þ

where F refD,BD(i,j) and F refD,A

D(i,j) are the background subtracted donor

intensities in pixels (i,j) before (B) and after (A), respectively,

running an identical acceptor photobleaching protocol on a

reference sample labeled with donor only (refD) (31). S avr4

is also a correction factor compensating for the crosstalk of the

acceptor label into the donor channel. It is the median of

S4ði;jÞ ¼FrefADði;jÞFrefAAði;jÞ

ð4Þ

where F refAD(i,j) and F refA

A(i,j) are the background subtracted donor

and acceptor intensities in pixels (i,j) of a reference sample la-

beled with acceptor only (refA). Correction for incomplete

acceptor bleaching was not necessary, because samples with

less than 80% depletion of acceptor were not evaluated. Con-

trol experiments revealed that calculated FRET values showed

negligible dependence on the extent of acceptor bleaching, if

at least 80% of the acceptor was bleached (see Supplemental

Material).

The measurement protocol was implemented on an LSM

510 confocal laser scanning microscope described earlier. A

custom C algorithm was written and used in SCIL Image

(TNO, Institute of Applied Physics, Amsterdam, The Nether-

lands) (32) to evaluate image sequences and calculate average

FRET efficiency after background subtraction, correction for

shift, and Gauss filtering for pixels above threshold in each

frame, usually containing 2–5 cells. Output results were stored

in standard image cytometry data (ICS 1.0) format (33).

RESULTS

Relative Expression Levels of Cell Surface Receptors

A general assumption in oncology is that the expression

level and interaction partners of proteins determine the malig-

nant phenotype of cancer cells. We presently investigate the

molecular background of resistance against trastuzumab, a

monoclonal antibody used in the treatment of ErbB2-overex-

pressing breast cancer. Since previous results implied that the

expression levels of ErbB1, ErbB2, adhesion molecules (b1-integrin, CD44), and MHC-I are important in influencing

trastuzumab responsiveness, we characterized the expression

levels of these molecules on trastuzumab resistant (JIMT-1,

MKN-7) and sensitive (SK-BR-3, N87) cell lines by flow cyto-

metry. Trastuzumab resistant cell lines showed lower ErbB2,

but higher b1-integrin and CD44 expression levels than the

sensitive ones. The relative receptor numbers of ErbB1 and

MHC-I did not reveal characteristic differences between the

examined resistant and sensitive cell lines (Fig. 1).

FCET Measurements

As we observed differences in the expression levels of

ErbB2, b1-integrin, and CD44 in the cell lines, we then

assessed the molecular associations of these proteins using

FCET (Table 1). The homoassociation of ErbB2 correlated

with its expression level giving rise to high ErbB2 homoasso-

ciation in trastuzumab sensitive cell lines in which the expres-

sion level of ErbB2 was high (19,20). The heteroassociations

were usually measured using the more abundant cell surface

protein as acceptor. Significant interaction between ErbB2 and

b1-integrin was detected on the resistant JIMT-1 cell line,

while MKN-7 and N87 cells displayed moderate and SK-BR-3

cells low (\5%) FRET in this respect. Heteroassociation of

CD44 with b1-integrin or ErbB2 was high on the resistant

JIMT-1 cell line but weak on the similarly resistant MKN-7

cells (Table 1).

Model System for Two-Sided FRET

As proof-of-concept for the two-sided FRET (tsFRET)

approach, we devised a triple labeled model system applicable

in CLSM, in which the first fluorophore (XFITC) is a donor

dye (D1), the second fluorophore (Cy3 or Alexa555) acts as an

acceptor (A1) for the first fluorophore and as a donor (D2)

for the third fluorophore (Cy5), serving as acceptor (A2). In

the system above, the energy transfer efficiency can be meas-

ured sequentially using the abFRETmethod (between D2:A1

and A2) followed by the dbFRET method (between D1 and

A1:D2) (see Fig. 2A). Using appropriately chosen fluoro-

Figure 1. Number of cell surface molecules on trastuzumab resist-

ant (JIMT-1, MKN-7) and sensitive (SK-BR-3, N87) cell lines. Num-

ber of ErbB1 (h), ErbB2 (n), b1-integrin ( ), MHC-I ( ) and CD44( ) receptors were determined by flow cytometry. The trastuzu-

mab resistant cell lines showed lower ErbB2, but higher b1-integ-rin and CD44 expression levels than the sensitive ones.

ORIGINAL ARTICLE


phores, the abFRET and dbFRET methods can be combined

consecutively, since the excitation wavelengths (543 and 633

nm corresponding to the absorption of A1:D2 and A2,

respectively) used in abFRET do not excite the first fluoro-

phore (XFITC). To demonstrate the feasibility of this

approach, we used the ErbB2 molecule labeled with three non-

competing monoclonal antibodies as a model system. On the

basis of the nearly full-length ErbB2 model (34), it was known

that the 7C2, 2C4, and 4D5 antibodies do not compete with

each other. We labeled N87 cells with XFITC-7C2 (D1), Cy3

or Alexa555-conjugated 2C4 (A1:D2) and Cy5-4D5 (A2).

Both dbFRET and abFRET efficiency values demonstrated the

proximity of the labeled antibodies as expected (Fig. 3). Then,

we generated the contour plot from the corresponding pixel-

by-pixel dbFRET (axis ‘‘x’’) and abFRET (axis ‘‘y’’) efficiencies.

The pixel-by-pixel 2D matrix of the contour plot was calcu-

lated on 5–10% FRET ranges by a macro written for the soft-

ware Sigmaplot. The resulted 2D matrix was then plotted by

using the contour plot function of the software Igor Pro. We

derived the dependence of the two FRET efficiencies on each

other by calculating a trendline. The dbFRET axis was divided

into several bins depending on the number of data points, and

the mean abFRET value corresponding to pixels in the bin was

calculated and plotted against the binned dbFRET values. The

contour plot and the trendline (slope 5 0.003% by linear

regression) show no correlation between the FRET values,

since these energy transfer efficiency values show a Gaussian

distribution. This finding is in accordance with expectations

since both dbFRET and abFRET values represent intramolecu-

lar distances, which are constant.

Application of Two-Sided FRET

Having demonstrated the feasibility of the tsFRET

approach, we set out to characterize the effect of the b1-integ-rin–ErbB2 heteroassociation (measured by dbFRET) on ErbB2

homoassociation (measured by abFRET) on trastuzumab sen-

sitive cell lines (SK-BR-3, N87). The schematic model of the

experiment can be seen in Fig. 2B. JIMT-1 and MKN-7 were

not examined in this respect, because the homoassociation

state of ErbB2 was low on both cell lines as determined by

flow cytometric measurements (Table 1) and confirmed also

with abFRET (35). Two different peaks can be identified at

13% (abFRET) – 20% (dbFRET) and at 7% (abFRET) – 30%

(dbFRET) on the contour plot for N87 cells. The trendline for

N87 cells shows a decrease with a slope of 20.042% (Fig. 4A).

This finding implies an anticorrelation between b1-integrin–ErbB2 heteroassociation (dbFRET) and ErbB2 homoassocia-

tion (abFRET), as the higher the dbFRET efficiency measured

in a pixel, the lower the abFRET efficiency was, and vice versa.

Similar anticorrelation can be observed on SK-BR-3 cells

based on the trendline showing also a decrease with a slope

20.080%, however, the contour plot displayed only a single

peak at about 12% (abFRET) – 26% (dbFRET) (Fig. 4B).

Two-sided FRET is suited for investigating the relation-

ship between the heteroassociation states of three different

molecular species on the cell surface. As surmounting evidence

supports the crosstalk between ErbB2 and certain cell surface

adhesion molecules, we decided to analyze the interaction of

ErbB2, b1-integrin, and CD44 on trastuzumab resistant breast

(JIMT-1) and gastric (MKN-7) cancer cell lines. The sche-

matic model of these experiments is displayed in Fig. 2C. A

modest positive correlation between ErbB2–b1-integrin and

b1-integrin–CD44 heteroassociations is observed on the con-

tour plot of MKN-7 cells (Fig. 4C), since higher dbFRET

values are paired with higher abFRET values; two discrete

peaks are distinguishable at 10% (abFRET) – 22% (dbFRET)

and at 20% (abFRET) – 32% (dbFRET) and the trendline

shows an increase with a slope 0.077%. On the contrary, there

is only one peak at 12% (abFRET) – 21% (dbFRET) on the

contour plot for JIMT-1 cells (Fig. 4D), the trendline runs

with a slope 20.002%, thus no indication of a clear positive

or negative correlation between the dbFRET and abFRET effi-

ciency values can be observed.

Next, we examined the effect of the ErbB1–ErbB2 hetero-

association on the homoassociation of ErbB2 in SK-BR-3 cells

and also characterized the interactions of ErbB1, b1-integrin,and MHC-I receptors in JIMT-1. Mean dbFRET and abFRET

efficiency values were determined from the corresponding

images (Table 2). In these cases, neither of the contour plots

or trendlines showed a clear correlation between dbFRET and

abFRET values (data not shown).

Table 1. FCET measurements on the trastuzumab sensitive (SK-BR-3 and N87) and resistant (JIMT-1 and MKN-7) cell linesa

ASSOCIATION/CELL LINE SK-BR-3 JIMT-1 N87 MKN-7

ErbB2 homoassociation 16.1 (�1.2) 2.9 (�0.8) 11.5 (�0.1) 3.9 (�1.2)

ErbB2 intramolecular 56.2 (�2.4) 58.2 (�1.5) 65.4 (�0.3) 56.7 (�1.4)

ErbB2–b1-integrinb n.d. 13.1 (�1.1) n.d. 9.1 (�1.2)

b1-integrin–ErbB2b 3.3 (�2.0) n.d. 7.1 (�1.3) n.d.

b1-integrin homoassociation 9.8 (�1.9) 6.4 (�1.0) 5.2 (�1.9) 6.4 (�2.8)

ErbB2–CD44 n.d. 19.9 (�1.5) n.d. 4.2 (�0.7)

b1-integrin–CD44 n.d. 10.5 (�1.0) n.d. 3.7 (�0.5)

MHC-I–b1-integrin n.d. 14.9 (�1.3) n.d. 5.2 (�0.5)

a Mean energy transfer efficiency determined in flow cytometry. Mean � SEM (%) of 20,000 cells.b Heteroassociations were always measured using the more abundant cell surface protein as acceptor.

n.d. 5 not determined owed to low expression levels.

ORIGINAL ARTICLE


When comparing flow cytometric and microscopic data,

FCET values of b1-integrin–ErbB2 heteroassociation on SK-

BR-3, N87, and MKN-7 cells and of b1-integrin–CD44 inter-

action on MKN-7 cells are significantly lower (Table 1) than

the corresponding microscopic FRET values (Table 2).

Although trypsinization used before flow cytometry can alter

the distribution and association state of receptors, especially

those of adhesion receptors like b1-integrin or CD44, we

wanted to exclude the possibility that the microscopic FRET

procedure overestimates the association rate of these mole-

cules. When comparing dbFRET values for ErbB2–b1-integ-rin in SK-BR-3 and MKN-7 cells, after trypsinization and

cytocentrifugation of the cells, these values were significantly

lower (1.1% � 0.3%, 3.8% � 0.5%) than the corresponding

values measured on adherent, nontrypsinized cells (see Table

2). This substantiated our assumption that it is trypsiniza-

tion and not the flow cytometric FRET approach which low-

ers the FRETefficiency between b1-integrin and ErbB2.

During these experiments, we have observed that, in con-

trast to their inhomogeneous distribution on the surface of

non-trypsinized adherent cells, b1-integrin and CD44 were

homogeneously distributed on the surface of trypsinized, cyto-

centrifuged cells. In the former, these cell adhesion receptors

were found in a remarkably higher density at the adhering

(bottom) surface of cells than toward the top. ErbB2 displayed

a more homogeneous distribution all over the cell membrane

both in adherent and trypsinized cells. We have used both

dbFRET and abFRET on MKN-7 and N87 cells to explore the

dependence of FRET between ErbB2 and b1-integrin on the

distance of various optical slices from the plane of adherence.

The decrease in the expression level of b1-integrin away from

the plane of adherence toward the top of the cell was accom-

panied by a decrease in both abFRET (Fig. 5) and dbFRET

(data not shown) efficiency values. This implies that the asso-

ciation of b1-integrin and ErbB2 is significantly higher at the

adhering surface of the cells.

DISCUSSION

Dynamic interactions of membrane proteins play an im-

portant role in the regulation of signal transduction processes.

It is conceivable that heterodimerization tendency of given

molecular species can affect the association pattern of other

molecular species. To study the relationship between the asso-

ciations of two molecule pairs of three arbitrarily chosen mo-

lecular species, we established a fluorescence microscopic

approach termed two-sided FRET, in which an abFRET mea-

surement is followed by a dbFRET measurement in triple-

labeled cells. Using appropriate fluorophores, the abFRET and

the dbFRETmethods can be combined, because the excitation

wavelengths (543 and 633 nm) used in abFRET do not excite

the first fluorophore (XFITC) and the spectral overlaps

between the emission spectra of the second (Cy3, Alexa555) or

third (Cy5) fluorophores and the absorption spectrum of the

first fluorophore (XFITC) are negligible. Initially, we used Cy3

as the second fluorophore (A1:D2), but the Alexa555-Cy5

pair turned out to be even more suitable for abFRET experi-

Figure 2. Schematic models of two-sided FRET measurements on

CLSM. D1 denotes XFITC, the donor in dbFRET, A1:D2 refers Cy3or Alexa555 acting as an acceptor in dbFRET (A1) and a donor in

abFRET (D2), while A2 indicates Cy5, the acceptor in abFRET. (A)

The model system. ErbB2 was labeled with XFITC-7C2 (D1), Cy3-

2C4 or Alexa555-2C4 (A1:D2) and Cy5-4D5 (A2). (B) Influence ofb1-integrin-ErbB2 heteroassociation on the ErbB2 homoassocia-tion. Cells were labeled with XFITC-TS2 (D1), Cy3-2C4 or

Alexa555-2C4 (A1:D2) and Cy5-2C4 (A2). (C) Relationship of

ErbB2 � b1-integrin and b1-integrin � CD44 heteroassociation.

Cells were labeled with XFITC-2C4 (D1), Cy3-TS2 or Alexa555-TS2

(A1:D2), and Cy5-Hermes3 (A2).

ORIGINAL ARTICLE


ments than the conventional Cy3-Cy5 pair based on its spec-

tral properties, normalized fluorescence intensities, overlap in-

tegral, and normalized FRETefficiency (36).

To test the applicability of two-sided FRET, we used

ErbB2 labeled on three different epitopes with fluorescent

monoclonal antibodies. The combined application of dbFRET

Figure 3. Two-sided FRET on N87 cells triple-labeled with anti-ErbB2 antibodies. The raw, green color coded XFITC-7C2 (D1) image (A),

red color coded Cy3-2C4 (A1:D2) image (B), blue color coded Cy5-4D5 (A2) image (C), the false color dbFRET efficiency image (right top),the false color abFRET efficiency image (right middle) and the contour plot derived from pixel values of the corresponding dbFRET and

abFRET images (right bottom) are displayed.

ORIGINAL ARTICLE


Figure 4. Two-sided FRET measurements of two molecule-pairs of three arbitrarily chosen molecular species. N87 (A) and SK-BR-3 (B)

cells were labeled for b1-integrin with XFITC-TS2 (D1), and for ErbB2 with Cy3- or Alexa555-2C4 (A1:D2) and Cy5-2C4 (A2) antibodies.MKN-7 (C) and JIMT-1 (D) cells were labeled for ErbB2 with XFITC-2C4 (D1), b1-integrin with Cy3- or Alexa555-TS2 (A1:D2) and for CD44with Cy5-Hermes3 (A2) antibodies. Consecutive abFRET and dbFRET measurements were performed and the false color maps of dbFRET

(left column) and abFRET (middle column) efficiencies as well as the contour plot of the corresponding pixel-by-pixel dbFRET and abFRET

efficiency values (right column) are shown.

ORIGINAL ARTICLE


and abFRET did not significantly alter the FRET values meas-

ured on double-labeled cells. This finding, the narrow gaus-

sian distribution of both intramolecular FRET values and the

lack of correlation between the dbFRET and abFRET values

implied that two molecular interactions sharing a common

species can be reliably studied by the two-sided FRET

method.

When trying to find a correlation between the trastuzu-

mab responsiveness of cell lines and their protein expression

levels and association states, we noticed that the trastuzumab

resistant cell lines showed lower ErbB2, but higher b1-integrinand CD44 expression levels than the sensitive ones. The b1-integrin–CD44, the ErbB2–CD44, and ErbB2–b1-integrin het-

eroassociations were modestly high on the resistant JIMT-1

cell line even by FCETmeasurements. This correlated with the

extremely high level of b1-integrin and CD44 expressions of

JIMT-1 cells, while the similarly resistant MKN-7 cells having

lower, but still high b1-integrin and CD44 expressions dis-

played weak ErbB2–CD44 and b1-integrin–CD44 heteroasso-

ciations. On the other hand, ErbB2 homoassociation seemed

to be high on N87 cells, where ErbB2 outnumbers other mole-

cules (e.g., b1-integrin and CD44) forming heteroassociations

with ErbB2. Therefore, it appears that molecules compete with

each other for finding association partners, i.e., association

states are not only influenced by their relative affinities, but

also by their relative expression levels. We applied the two-

sided FRET method to investigate the relationship between

association states of different molecules. Analyzing the con-

tour plots and the trendlines generated from dbFRET and

abFRET pixel values, we observed a clear anticorrelation

between b1-integrin–ErbB2 heteroassociation and ErbB2

homoassociation on N87 and SK-BR-3 cells. On the other

hand, we observed a modest positive correlation between

ErbB2–b1-integrin and b1-integrin–CD44 heteroassociations

on MKN-7 cells, whereas on JIMT-1 cells no correlation was

noticed. The later observation can be explained by the extre-

mely high b1-integrin and CD44 expression levels on this cell

line, and that ErbB2 expressed at a much lower level could

interfere only with a small portion of b1-integrin and CD44

molecules leaving the association state of the majority of them

undisturbed. According to these results, we may assume that

the homoassociation state of ErbB2 is dynamically modulated

by the interaction with b1-integrins, and it is tempting to

speculate that the heteroassociations and activation of cell ad-

hesion molecules can enhance the activity of survival pathways

during disease progression.

Flow cytometric FRET results were in agreement with mi-

croscopic ones for the homoassociation of ErbB2. When com-

paring intramolecular FRET values that are likely unaffected

by cytoskeletal organizing forces, abFRET gave lower values

Table 2. Two-sided FRET measurements among three distinct molecular species on trastuzumab sensitive (SK-BR-3 and N87) and

resistant (JIMT-1 and MKN-7) cellsa

CELL LINE D1 A1:D2 A2 dbFRET abFRET

JIMT-1 ErbB1 b1-integrin MHC-I 16.6 (�4.3) 7.2 (�0.4)

ErbB2 b1-integrin CD44 12.0 (�0.8) 9.2 (�1.5)

MKN-7 ErbB2 b1-integrin CD44 15.2 (�2.4) 21.2 (�3.4)

b1-integrin ErbB2 CD44 18.2 (�2.3) 6.0 (�4.0)

SK-BR-3 ErbB1 ErbB2 ErbB2 21.0 (�1.7) 11.2 (�1.5)

b1-integrin ErbB2 ErbB2 26.3 (�1.3) 16.0 (�2.8)

ErbB2b ErbB2b ErbB2b 37.0 (�3.4) 23.0 (�1.1)

N87 b1-integrin ErbB2 ErbB2 19.0 (�2.4) 9.3 (�1.8)

a Mean dbFRET and abFRET efficiency values (�SEM, %) determined from corresponding dbFRET and abFRET images.b ErbB2 receptors were triple-labeled with 7C2 (D1), 2C4 (A1:D2), and 4D5 (A2) antibodies tagged by the corresponding XFITC, Cy3 orAlexa555, and Cy5 fluorescent dye for consecutive abFRET and dbFRET measurements. All other ErbB2 indicated in the table denotes

labeling with 2C4 antibody.

Figure 5. Z-section images of ErbB2�b1-integrin heteroassocia-tion on adherent MKN-7 cells. MKN-7 cells were labeled with

Alexa555-2C4 and Cy5-TS2 in chambered coverglass. Z-section

image was taken as a series of line scans recorded with 1 lmdepth interval. The Alexa555-2C4 as donor (A), the Cy5-TS2 as

acceptor (B), and the false color energy transfer efficiency image

(C) are displayed. It can be seen that the association state of

ErbB2 and b1-integrin is decreasing from the plane of adherence

(bottom of the images) toward the top of the cell.

ORIGINAL ARTICLE


than those measured by dbFRET, which were, in turn, lower

than those provided by the FCET method. The discrepancy

between flow cytometric and dbFRET results can be explained

by the pronounced sensitivity of photobleaching microscopic

methods to environmental factors such as the local concentra-

tion of fluorochromes and oxygen (37). The overestimation by

dbFRET compared with abFRET can be related to the different

weighting of the methods (8), and to the fact that the dbFRET

method requires a reference donor photobleaching decay time

measured in the absence of the acceptor. The conditions under

which this reference parameter is measured (on a single donor

labeled sample or on a double-labeled sample in a region in

which the acceptor has been depleted) can strongly influence

the dbFRET values (7).

Interestingly, the dbFRET efficiency values of b1-integ-rin–ErbB2 heteroassociation were markedly higher on most

cells (except JIMT-1) than those from FCET measurements.

Similar differences could be observed when we examined the

b1-integrin–CD44 heteroassociation using abFRET, implying

that the overestimating effect of dbFRET (8) is not solely re-

sponsible for the divergence of the results. Furthermore, tryp-

sinized and cytocentrifuged cells in microscopy gave very low

FRET values. In addition, the distribution of FRET between

ErbB2 and b1-integrin along the z axis demonstrated that

higher FRET efficiency values were observed in the membrane

apposed to the glass surface than towards the top of the cell

(Fig. 5).

Since the regular microscopic FRET measurements were

performed on adherent cells, and images were taken usually of

the attached bottom layer of the cells, the higher expression

level and probably increased activation state of b1-integrin at

these focal adhesion points could strongly affect the interac-

tion with ErbB2. These same interactions are likely disrupted

when cells are trypsinized for FCET.

In conclusion, we implemented a method termed two-

sided FRET for the investigation of the relationship of associa-

tions of two molecule pairs of three arbitrarily chosen molecu-

lar species. While at the single molecule level several new

approaches have been introduced recently for analyzing the

interactions of up to three single molecules in solution, such

as the pulsed interleaved excitation (PIE) method (38), the

three-color alternating-laser excitation (3c-ALEX) method

(39) simultaneous fluorescence decay time measurements

(40), multiparameter single-molecule fluorescence spectros-

copy (SMFS) (41), and confocal multicolor single-molecule

spectroscopy (42), our method provides a useful approach to

cellular systems where background from autofluorescence or

higher density of labels would preclude single molecule spec-

troscopy. It must also be kept in mind that the detection of

FRET between pairwise combinations of three molecules (e.g.,

between A and B, and between B and C) does not imply the

presence of a trimer, because dimers AB and BC can separately

exist in a single pixel of size limited by the resolving power of

the microscope. Measuring FRET between AB and BC sepa-

rately would actually give statistically relevant information

about the association state of these pairwise combinations;

however, the spatial relationship of the AB and BC dimers can

not be revealed in such an experiment. The pixel-by-pixel rela-

tionship of the two molecule pairs can be detected by doing

both dbFRET and abFRETmeasurements on the same sample

labeled by three noncompeting monoclonal antibodies yield-

ing information about how the association state of AB corre-

lates with that of BC in a single pixel on the same cell surface.

Therefore, the two-sided FRET method supplements, but not

supplants the triple-FRETmethod (43).

Using tsFRETwe found anticorrelation between b1-integ-rin–ErbB2 heteroassociation and ErbB2 homoassociation on

trastuzumab sensitive cells, implying that the homoassociation

state of the ErbB2 is dynamically modulated by the interaction

with b1-integrins. We have also observed a modest positive

correlation between ErbB2–b1-integrin and b1-integrin–CD44 heteroassociations on MKN-7 cells. These findings sup-

port the role of the heteroassociations and activation state of

cell adhesion molecules played in the signal transduction path-

ways of growth factor receptors.

ACKNOWLEDGMENTS

We wish to thank Ms. Gabriella O†ri, Ms. Tunde Terdik,

and Ms. Hajnalka Toldi for their technical assistance in protein

conjugation and wish to express sincere gratitude to Gabor

Horvath and Gergely Szentesi for their constructive criticism.

LITERATURE CITED

1. Szollosi J, Tron L, Damjanovich S, Helliwell SH, Arndt-Jovin D, Jovin TM. Fluores-cence energy transfer measurements on cell surfaces: A critical comparison of steady-state fluorimetric and flow cytometric methods. Cytometry 1984;5:210–216.

2. Tron L, Szollosi J, Damjanovich S, Helliwell SH, Arndt-Jovin DJ, Jovin TM. Flowcytometric measurement of fluorescence resonance energy transfer on cell surfaces.Quantitative evaluation of the transfer efficiency on a cell-by-cell basis. Biophys J1984;45:939–946.

3. Clegg RM. Fluorescence resonance energy transfer. Curr Opin Biotechnol 1995;6:103–110.

4. Szollosi J, Nagy P, Sebestyen Z, Damjanovich S, Park JW, Matyus L. Applications offluorescence resonance energy transfer for mapping biological membranes. J Biotech-nol 2002;82:251–266.

5. Sebestyen Z, Nagy P, Horvath G, Vamosi G, Debets R, Gratama JW, Alexander DR,Szollosi J. Long wavelength fluorophores and cell-by-cell correction for autofluores-cence significantly improves the accuracy of flow cytometric energy transfer measure-ments on a dual-laser benchtop flow cytometer. Cytometry 2002;48:124–35.

6. Szollosi J, Damjanovich S, Matyus L. Application of fluorescence resonance energytransfer in the clinical laboratory: Routine and research. Cytometry: Communica-tions in Clinical Cytometry 1998;34:159–179.

7. Jares-Erijman EA, Jovin TM. FRET imaging. Nat Biotechnol 2003;21:1387–1395.

8. Nagy P, Vamosi G, Bodnar A, Lockett SJ, Szollosi J. Intensity-based energy transfermeasurements in digital imaging microscopy. Eur Biophys J 1998;27:377–389.

9. Jares-Erijman EA, Jovin TM. Imaging molecular interactions in living cells by FRETmicroscopy. Curr Opin Chem Biol 2006;10:409–416.

10. Alroy I, Yarden Y. The ErbB signaling network in embryogenesis and oncogenesis: sig-nal diversification through combinatorial ligand-receptor interactions. FEBS Lett1997;410:83–86.

11. Slamon DJ, Clark GM, Wong SG, Levin WJ, Ullrich A, McGuire WL. Human breastcancer: Correlation of relapse and survival with amplification of the HER-2/neuoncogene. Science 1987;235:177–182.

12. Schechter A, Stern DF, Vaidyanathan L, Decker SJ, Drebin JA, Green MI, WeinbergRA. The neu oncogene: An erb-B-related gene encoding a 185,000-Mr tumour anti-gen. Nature 1984. pp 513–516.

13. Yarden Y, Sliwkowski MX. Untangling the ErbB signalling network. Nat Rev Mol CellBiol 2001;2:127–137.

14. Spencer KS, Graus-Porta D, Leng J, Hynes NE, Klemke RL. ErbB2 is necessary forinduction of carcinoma cell invasion by ErbB family receptor tyrosine kinases. J CellBiol 2000;148:385–97.

15. Graus-Porta D, Beerli RR, Daly JM, Hynes NE. ErbB-2, the preferred heterodimeriza-tion partner of all ErbB receptors, is a mediator of lateral signaling. EMBO J 1997;16:1647–1655.

16. Karunagaran D, Tzahar E, Beerli RR, Chen X, Graus-Porta D, Ratzkin BJ, Seger R,Hynes NE, Yarden Y. ErbB-2 is a common auxiliary subunit of NDF and EGF recep-tors: Implications for breast cancer. EMBO J 1996;15:254–264.

17. Albelda SM. Role of integrins and other cell adhesion molecules in tumor progressionand metastasis. Lab Invest 1993;68:4–17.

ORIGINAL ARTICLE


18. Fujita S, Watanabe M, Kubota T, Teramoto T, Kitajima M. Alteration of expression inintegrin b 1-subunit correlates with invasion and metastasis in colorectal cancer.Cancer Lett 1995;91:145–149.

19. Mocanu MM, Fazekas Z, Petras M, Nagy P, Sebestyen Z, Isola J, Timar J, Park JW,Vereb G, Szollosi J. Associations of ErbB2, b1-integrin and lipid rafts on Herceptin(Trastuzumab) resistant and sensitive tumor cell lines. Cancer Lett 2005;227:201–212.

20. Falcioni R, Antonini A, Nistico P, Di Stefano S, Crescenzi M, Natali PG, Sacchi A. a 6b 4 and a 6 b 1 integrins associate with ErbB-2 in human carcinoma cell lines. ExpCell Res 1997;236:76–85.

21. Guo W, Giancotti FG. Integrin signalling during tumour progression. Nat Rev MolCell Biol 2004;5:816–826.

22. Nagy P, Jenei A, Damjanovich S, Jovin TM, Szollosi J. Complexity of signal transduc-tion mediated by ErbB2: Clues to the potential of receptor-targeted cancer therapy.Pathol Oncol Res 1999;5:255–271.

23. Nagy P, Jenei A, Kirsch AK, Szollosi J, Damjanovich S, Jovin TM. Activation-depend-ent clustering of the erbB2 receptor tyrosine kinase detected by scanning near-fieldoptical microscopy. J Cell Sci 1999;112:1733–1741.

24. Brown DA, London E. Functions of lipid rafts in biological membranes. Annu RevCell Dev Biol 1998;14:111–136.

25. Nagy P, Vereb G, Sebestyen Z, Horvath G, Lockett SJ, Damjanovich, S, Park JW, JovinTM, Szollosi J. Lipid rafts and the local density of ErbB proteins influence the bio-logical role of homo- and heteroassociations of ErbB2. J Cell Sci 2002;115:4251–4262.

26. Szentesi G, Horvath G, Bori I, Vamosi G, Szollosi J, Gaspar R, Damjanovich S, JeneiA, Matyus L. Computer program for determining fluorescence resonance energytransfer efficiency from flow cytometric data on a cell-by-cell basis. Comput MethodsPrograms Biomed 2004;75:201–211.

27. Jovin TM, Arndt-Jovin DJ. Luminescence digital imaging microscopy. Annu Rev Bio-phys Biophys Chem 1989;18:271–308.

28. Szentesi G, Vereb G, Horvath G, Bodnar A, Fabian A, Matko J, Gaspar R, Damjano-vich S, Matyus L, Jenei A. Computer program for analyzing donor photobleachingFRET image series. Cytometry Part A 2005;67A:119–128.

29. Bastiaens PI, Majoul IV, Verveer PJ, Soling HD, Jovin TM. Imaging the intracellulartrafficking and state of the AB5 quaternary structure of cholera toxin. EMBO J1996;15:4246–4253.

30. Vereb G, Meyer CK, Jovin TM. Novel microscope-based approaches for the investiga-tion of protein-protein interactions in signal transduction. In: Heilmeyer LMG, edi-tor. Interacting Protein Domains. Their Role in Signal and Energy Transduction, Vol-

ume 102, NATO ASI Series, H: Cell Biology. Berlin, Heidelberg: Springer Verlag;1997. pp 49–52.

31. Vereb G, Matko J, Szollosi J. Cytometry of fluorescence resonance energy transfer.Methods Cell Biol 2004;75:105–152.

32. Van Balen R, Koelma D, Ten Kate TR, Mosterd B, Smeulders AWM. ScilImage: AMulti-layered Environment for Use and Development of Image Processing Software.In: H.I. Christiansen and J.L. Crowley, editors. Experimental Environments for Com-puter Vision and Image Processing. Singapore: World Scientific Publishing Co.; 1994;pp 107–126.

33. Dean P, Mascio L, Ow D, Sudar D, Mullikin J. Proposed standard for image cytome-try data files. Cytometry 1990;11:561–569.

34. Bagossi P, Horvath G, Vereb G, Szollosi J, Tozser J. Molecular modeling of nearly full-length ErbB2 receptor. Biophys J 2005;88:1354–1363.

35. Zsebik B, Citri A, Isola J, Yarden Y, Szollosi J, Vereb G. Hsp90 inhibitor 17-AAGreduces ErbB2 levels and inhibits proliferation of the trastuzumab resistant breast tu-mor cell line JIMT-1. Immunol Lett 2006;104:146–155.

36. Horvath G, Petras M, Szentesi G, Fabian A, Park JW, Vereb G, Szollosi J. Selecting theright fluorophores and flow cytometer for fluorescence resonance energy transfermeasurements. Cytometry Part A 2005;65A:148–157.

37. Song L, Hennink EJ, Young IT, Tanke HJ. Photobleaching kinetics of fluorescein inquantitative fluorescence microscopy. Biophys J 1995;68:2588–2600.

38. Muller BK, Zaychikov E, Brauchle C, Lamb DC. Pulsed interleaved excitation. Bio-phys J 2005;89:3508–3522.

39. Lee NK, Kapanidis AN, Koh HR, Korlann Y, Ho SO, Kim Y, Gassman N, Kim SK,Weiss S. Three-color alternating-laser excitation of single molecules: Monitoringmultiple interactions and distances. Biophys J 2007;92:303–312.

40. Rothwell PJ, Berger S, Kensch O, Felekyan S, Antonik M, Wohrl BM, Restle T, GoodyRS, Seidel CA. Multiparameter single-molecule fluorescence spectroscopy reveals het-erogeneity of HIV-1 reverse transcriptase:primer/template complexes. Proc Natl AcadSci USA 2003;100:1655–1660.

41. Heilemann M, Kasper R, Tinnefeld P, Sauer M. Dissecting and reducing the heteroge-neity of excited-state energy transport in DNA-based photonic wires. J Am Chem Soc2006;128:16864–16875.

42. Ross J, Buschkamp P, Fetting D, Donnermeyer A, Roth CM, Tinnefeld P. Multicolorsingle-molecule spectroscopy with alternating laser excitation for the investigation ofinteractions and dynamics. J Phys Chem B 2007;111:321–326.

43. Galperin E, Verkhusha VV, Sorkin A. Three-chromophore FRET microscopy to ana-lyze multiprotein interactions in living cells. Nat Methods 2004;1:209–217.

ORIGINAL ARTICLE


Two-sided fluorescence resonance energy transfer for assessing molecular interactions of up to three...

Documents

Transcript of Two-sided fluorescence resonance energy transfer for assessing molecular interactions of up to three...