TOPICAL REVIEW: Fluorescence lifetime imaging microscopy in life sciences

Abstract

Fluorescence lifetime imaging microscopy (FLIM) and fluorescence anisotropy imaging microscopy (FAIM) are versatile tools for the investigation of the molecular environment of fluorophores in living cells. Owing to nanometre-scale interactions via Förster resonance energy transfer (FRET), FLIM and FAIM are powerful microscopy methods for the detection of conformational changes and protein-protein interactions reflecting the biochemical status of live cells. This review provides an overview of recent advances in photonics techniques, quantitative data analysis methods and applications in the life sciences.

Full text

Fluorescence lifetime imaging microscopy in life sciences
This article has been downloaded from IOPscience. Please scroll down to see the full text article.
2010 Meas. Sci. Technol. 21 102002
(http://iopscience.iop.org/0957-0233/21/10/102002)
Download details:
IP Address: 137.224.252.10
The article was downloaded on 02/03/2012 at 16:34
Please note that terms and conditions apply.
View the table of contents for this issue, or go to the journal homepage for more
Home Search Collections Journals About Contact us My IOPscience

IOP PUBLISHING MEASUREMENT SCIENCE AND TECHNOLOGY
Meas. Sci. Technol. 21 (2010) 102002 (21pp) doi:10.1088/0957-0233/21/10/102002
TOPICAL REVIEW
Fluorescence lifetime imaging microscopy
in life sciences
Jan Willem Borst and Antonie J W G Visser
Laboratory of Biochemistry, Microspectroscopy Centre, Wageningen University, PO Box 8128,
6700 ET Wageningen, The Netherlands
E-mail: janwillem.borst@wur.nl and antoniejvisser@gmail.com
Received 4 June 2010, in final form 2 August 2010
Published 14 September 2010
Online at stacks.iop.org/MST/21/102002
Abstract
Fluorescence lifetime imaging microscopy (FLIM) and fluorescence anisotropy imaging
microscopy (FAIM) are versatile tools for the investigation of the molecular environment of
fluorophores in living cells. Owing to nanometre-scale interactions via F
¨
orster resonance
energy transfer (FRET), FLIM and FAIM are powerful microscopy methods for the detection
of conformational changes and protein–protein interactions reflecting the biochemical status of
live cells. This review provides an overview of recent advances in photonics techniques,
quantitative data analysis methods and applications in the life sciences.
Keywords: FLIM, FRET, fluorescence, anisotropy, GFP, biosensors
(Some figures in this article are in colour only in the electronic version)
1. Introduction
All living organisms are embedded in a complex environment
and have developed sophisticated sensing mechanisms.
Consequently, living cells consist of a complicated network
of biological molecules to cope with internal and external
cues. Visualization of these molecular processes in vivo like
signal transduction reactions, post-translational modifications,
apoptosis or ion fluxes is currently possible by the use
of fluorescence imaging methodologies. One approach to
detect molecules in living cells is by development of optical
biosensors, which are ideal candidates for industrial use
because of the high specificity, selectivity and adaptability. A
typical example in healthcare is the application of disposable
blood glucose biosensors, in which blood sugar levels of
diabetic patients are monitored in real time [1]. Biosensors
used in cell biology generally imply real-time monitoring
of molecular activities and/or interactions in living cells [2].
New developments of optical biosensors in combination with
fluorescence microscopy give insight into the responses of
molecular components in the cell such as ligands, receptors
and other signalling proteins, which are activated in a
dynamic network. Live cell imaging provides the time as
an additional dimension and has evolved as an important
approach in cell biology to monitor intracellular dynamics
[3–5]. Nowadays localization and dynamics can be connected
with system properties like active-protein state, mobility of
cytosolic and membrane-bound proteins, electric field effects
and oxygen, calcium or proton fluxes to address different
biological questions [6]. To visualize and quantify cellular
proteins of interest, advanced spectroscopic techniques are
combined with microscopy and biosensors, so that specific
molecular information on cells is obtained. In this respect the
development and application of the green-fluorescent protein
(GFP) technology has been of crucial importance for studying
fluorescently labelled proteins in their natural cellular habitat
[7].
The introduction of high spatial resolution confocal
microscopy gave the opportunity to investigate the co-
expression of different proteins in living cells. The spatial
resolution of a microscope allows detection at sub-cellular
level, but physical molecular interactions between proteins
and receptors on nanometre scale cannot be visualized. The
spatial resolution of a typical confocal or wide-field image
is diffraction limited meaning that the use of excitation
light of 488 nm will result in an optical resolution of
0957-0233/10/102002+21$30.00 1 © 2010 IOP Publishing Ltd Printed in the UK & the USA

Meas. Sci. Technol. 21 (2010) 102002 Topical Review
around 220 nm by applying the empirical relation of
0.61λ/NA with λ the excitation wavelength (488 nm) and
NA the numerical aperture of the microscope objective
(1.4). One possibility to go beyond the optical limitations
is to apply super-resolution microscopy methods like
stimulated emission depletion (STED) microscopy, saturated
structural illumination microscopy (SSIM), photoactivatable
localization microscopy (PALM) or stochastic optical
reconstruction microscopy (STORM) as reviewed by Hell [8].
Another possibility is to combine F
¨
orster resonance energy
transfer (FRET) methodology with fluorescence microscopy.
FRET is a fluorescence technique based on dipolar interactions
between different coloured donor and acceptor molecules
making this technique very sensitive to intermolecular
distances in the range between 1 and 10 nm. Many fluorescent
biosensors are based on the FRET methodology [9, 10].
The outline of this review is as follows. First we give an
introduction to fluorescence in general with some emphasis
on time-resolved fluorescence spectroscopy. Furthermore, the
use of time-resolved fluorescence spectroscopy in combination
with fluorescence microscopy as enhanced contrast mode in
fluorescence microscopy, which is also known as fluorescence
lifetime imaging microscopy (FLIM), will be highlighted. A
survey of the different techniques for the measurement of
fluorescence lifetimes under a microscope is given. We also
give a survey of classical molecular biological /biochemical
approaches to measure protein interactions in cells. Recent
FLIM applications in living systems including FRET detected
by FLIM as well as applications of time-resolved fluorescence
anisotropy imaging are discussed. A few remarks on advanced
data analysis methods are given.
2. Basic concepts of fluorescence spectroscopy
2.1. History
Fluorescence spectroscopy has a long history; a description of
the phenomenon dates from previous centuries. The quantity
known as the Stokes shift, which is the shift to lower energy of
the fluorescence spectrum relative to the absorption spectrum
of a compound, was already observed and published in 1852
by Stokes, who also coined the term fluorescence [11]. In
the first half of the previous century fluorescence turned
out to be already a settled method as illustrated by famous
phenomena like, among others, quenching of the fluorescence
[12], polarization of fluorescence [13, 14], the construction of
a phase fluorometer for measuring fluorescence lifetimes of
dye molecules [15], the Jablonski diagram [16], radiationless
energy transfer [17] (its history excellently reviewed by
Clegg [18]) and so forth. Applications to biochemical and
biophysical systems have been pioneered by Weber [19]. The
very first FLIM system has been based on a frequency-domain
microscope setup [20].
During the last few decades, fluorescence has been
established as a sensitive technique at the interface
of chemistry, physics and biology. Its popularity is
greatly enhanced by newly developed techniques and
methods such as femtosecond and picosecond mode-
locked lasers, rapid proximity-focused microchannel
plate (MCP) photomultipliers (PMTs) and single-photon
avalanche photodiodes (SPADs), modern electronics, electron-
multiplied charge-coupled device (ECCD) cameras, new
microscope setups, powerful computers and data analysis
software, rapid multidimensional data acquisition and so
forth. The popularity stems also from the fact that the
method is extremely sensitive allowing the investigation
of even single fluorescent molecules. Fluorescence is an
established technique in cell biology, because proteins in
cells can be made fluorescent by fusing them with GFP
variants and different cellular compartments can be stained
with fluorescent materials. Subsequently intracellular details
can be visualized by using sensitive fluorescence imaging
microscopy. Many excellent textbooks and edited books on
fluorescence principles and fluorescence microscopy and its
applications have been published (see, for instance, [21–26]).
Why is it that fluorescence has emerged as one of the most
frequently used spectroscopic techniques in biology, chemistry
and physics? The answer is quite simple: the time scale of
molecular interactions falls exactly within the expected range
relevant for physiological conditions. This aspect can be
illustrated using the Einstein relation:
x
2
=2Dt (1)
stating that the mean-square molecular displacement is related
to the translational diffusion coefficient D and the travelling
time t. If we substitute a time of 10 ns (order of magnitude
of a fluorescence lifetime) and a diffusion constant D =
10
−6
cm
2
s
−1
(typically for a protein of mass 10 kDa), the
displacement x
2

1/2
is 1 nm, which is of similar magnitude
to the diameter of a large protein segment and equivalent to
a large protein conformational change. This relatively simple
example is taken to illustrate that fluorescence spectroscopy
is a sensitive method to register dynamic protein events in
biological systems.
2.2. Jablonski diagram
Basic principles of fluorescence can be clarified with the help
of the Jablonski diagram (figure 1). This diagram illustrates the
electronic states of a fluorescent molecule and the transitions
between them. The electronic states are arranged vertically
by the energy level. They are grouped horizontally by spin
multiplicity. In the left part of the diagram three singlet states
with antiparallel spins are shown: the singlet ground state
(S
0
) and two higher singlet excited states (S
1
and S
2
). Singlet
states are diamagnetic, as they do not interact with an external
magnetic field. The triplet state (T
1
) is the electronic state with
parallel spins. A molecule in the triplet state interacts with
an external magnetic field. Transitions between electronic
states of the same spin multiplicity are allowed. Transitions
between states with different spin multiplicities are formally
forbidden, but may occur owing to a process called spin–
orbit coupling. This transition is called intersystem crossing.
Superimposed on these electronic states are the vibrational
states, which are of much smaller energy. When a molecule
absorbs a photon with an appropriate energy, a valence electron
is promoted from the ground state to a particular vibrational
level in the excited singlet manifold (figure 1). The process
2

Meas. Sci. Technol. 21 (2010) 102002 Topical Review
Figure 1. Jablonski diagram illustrating the electronic states of an aromatic molecule and the transitions between them. The left panel
represents the light absorption process, the middle panel the fluorescence process and the right panel intersystem crossing to the triplet state.
Solid lines denote radiative transitions, dashed lines non-radiative transitions and the dotted line intersystem crossing. Note that two
low-energy photons can be simultaneously absorbed in a single quantum event to excite the molecule.
of light absorption is extremely rapid, in the order of 1 fs.
This means that the nuclei of the molecule are fixed during
the transition, because of their much larger mass, and that
the Born–Oppenheimer approximation is valid. It should
be noted that the point of departure is always the lowest
vibrational level of S
0
designated S
0,0
. After light absorption
the excited molecule ends up at the lowest vibrational level
of S
1
(S
1,0
) via vibrational relaxation and internal conversion.
This radiationless process takes place in about 1 ps. The lowest
vibrational level of S
1
is the starting point for fluorescence
emission to the ground state S
0
, non-radiative decay to S
0
(internal conversion) and transition to the lowest triplet state
(intersystem crossing) (figure 1). Fluorescence takes place on
the nanosecond timescale and, depending on the molecular
species and its duration, amounts to 1–100 ns. It is clear from
the Jablonski diagram that fluorescence always originates from
the same level, irrespective of which electronic energy level is
excited. The emitting state is the lowest vibrational level of the
first excited state S
1,0
. It is for this reason that the fluorescence
spectrum is shifted to lower energy (higher wavelength) than
the corresponding absorption spectrum (Stokes shift). In
figure 1, the triplet state is also drawn. Once the molecule has
reached this state, it will reside there for a very long time (from
microseconds to seconds) before i t will decay to the ground
state. This is due to the spin-forbidden transitions involved
in the excited singlet–triplet and triplet–singlet ground state
transitions. In rigid solutions or in deoxygenated solutions,
long-lived phosphorescence (milliseconds to seconds) from
this state can be observed. Because of its long lifetime, the
triplet state of an aromatic molecule is the starting point for
photochemical reactions. One reaction in particular is very
prominent, namely the production of very reactive singlet
oxygen. The oxygen molecule (O
2
) possesses a triplet ground
state. In solution, frequent collisions between an aromatic
molecule in the triplet state and oxygen result in energy transfer
and generation of singlet oxygen that can oxidize and destroy
the aromatic molecule. These photochemical reactions are
important to understand photobleaching effects that occur in
fluorescence microscopy leading to a decrease in fluorescence
intensities.
2.3. Fluorescence parameters
The quantum yield, , is defined as the ratio of the number of
fluorescence photons emitted by the sample n
E
to the number
of photons absorbed n
A
. Furthermore it can be shown that 
is the ratio of the rate of the radiative transition ( k
r
)tothesum
of rates of all transitions (k
r
+ k
nr
), in which the excited state
is involved. Therefore, any molecular mechanism leading to
a non-radiative depopulation of the excited state reduces the
quantum yield:
 =
n
E
n
A
=
k
r
k
r
+ k
nr
. (2)
The other important characteristic feature of fluorescence is
its time response, namely the decay of fluorescence intensity
following infinitesimally short or δ-type excitation. The
scheme presented in figure 2 is a simplified Jablonski diagram
that can be used to explain the basic kinetics of fluorescence.
The population of the excited molecules [M
∗
] generated at the
moment of excitation, t = 0, starts to decrease exponentially
through the radiative ( k
r
) and non-radiative (k
nr
) transitions to
the ground state. The characteristic time of this process, τ ,is
called the fluorescence lifetime:
τ =
1
k
r
+ k
nr
. (3)
The intensity of fluorescence, I(t), emitted at any moment of
this process is proportional to [M
∗
](t), thus
I(t) = I
0
exp(−t/τ). (4)
The fluorescence lifetime τ has the physical meaning of the
time needed for the fluorescence intensity to decrease to 1/e
(= 1/2.71) of its initial value I
0
.
Any molecule, which interacts with the fluorophore and
reduces its quantum yield or lifetime, is called a quencher.
In the case of dynamic quenching the quenching molecule
Q collides with the excited fluorophore M
∗
(see figure 2).
The excited state kinetics is affected by forming an additional
way of depopulating the excited state. Consequently, the
fluorescence decay is modified. Both steady-state and time-
resolved fluorescence yield the famous Stern–Volmer constant
3

Meas. Sci. Technol. 21 (2010) 102002 Topical Review
Figure 2. Kinetic scheme to explain unperturbed fluorescence (left), dynamic quenching (middle) and FRET (right). δ(0) is excitation with
a δ-pulse, k
r
is the rate constant of radiative transition, k
nr
is the rate constant of non-radiative transition, k
q
is the rate constant of quenching,
Q is an external quenching molecule and k
T
is the rate constant of FRET from donor molecule D to acceptor molecule A.
K
SV
= τ
0
k
q
, in which τ
0
is the fluorescence lifetime without
quencher and k
q
is the diffusion controlled rate constant of
quenching. K
SV
has the dimension M
−1
.
2.4. F
¨
orster resonance energy transfer
F
¨
orster developed the quantitative theory for resonance energy
transfer in the late 1940s [17]. Therefore, we call this
process F
¨
orster resonance energy transfer or FRET. FRET
is a photophysical process where the excited-state energy
from a donor molecule is transferred non-radiatively to an
acceptor molecule (in the ground state) at close distance via
weak dipole–dipole coupling (figure 2). F
¨
orster derived the
following expression for the rate constant of transfer k
T
:
k
T
(R) =
1
τ
D

R
0
R

6
(5)
where R is the distance between donor and acceptor molecules
and τ
D
is the fluorescence lifetime of the donor (without
acceptor). Since the transfer rate is proportional to the inverse
6th power of the distance R, the transfer rate is an extremely
sensitive parameter for obtaining distances between 1 and
10 nm. The distance at which the excitation energy of the
donor is transferred to the acceptor with probability 0.5 is
called the F
¨
orster or critical distance R
0
and can be calculated
using the relevant spectroscopic properties of the participating
molecules:
R
6
0
=
9000 ln(10)κ
2

0
128π
5
N
A
n
4

∞
0
F
D
(σ )ε
A
(σ )
σ
4
dσ, (6)
where κ
2
is the orientation factor, 
0
is the quantum yield
of donor fluorescence (without acceptor), N
A
is Avogadro’s
number and n is the refractive index of the intervening medium.
The integral (J,M
−1
cm
3
) is the degree of spectral overlap
between donor fluorescence spectrum (F
D
) and acceptor
absorption spectrum (ε
A
), given by either wave number (σ )
or wavelength (λ) scale:
J =

∞
0
F
D
(σ )ε
A
(σ )
σ
4
dσ =

∞
0
F
D
(λ)ε
A
(λ)λ
4
dλ. (7)
The orientation factor κ
2
is given by
κ
2
= [cos θ
T
− 3 cos θ
D
cos θ
A
]
2
= [sin θ
D
sin θ
A
cos ϕ − 2 cos θ
D
cos θ
A
]
2
, (8)
where θ
T
is the angle between transition moments of donor and
acceptor, θ
D
and θ
A
are the angles of the transition moments
of donor and acceptor with the separation vector and ϕ is
the angle between the two planes formed by the transition
moments of donor and acceptor and the s eparation vector. For
systems without any three-dimensional, spatial information
the orientation factor is the indeterminate parameter in the
evaluation of R
0
having any value 0 <κ
2
< 4. All other
parameters can be measured or evaluated. F
¨
orster also
introduced the transfer efficiency E, which is only a function
of actual (R) and critical (R
0
) distances:
E =
1
1+

R
R
0

6
. (9)
There are several methods available for quantification of
FRET, of which the one based on donor fluorescence lifetimes
is the most straightforward, because the fluorescence lifetime
is a concentration-independent property, while fluorescence
intensity is not. Donor fluorescence lifetimes in the absence
(τ
D
) and presence (τ
DA
) of acceptor molecules are often
measured for the observation of FRET and a decreased
fluorescence lifetime of the donor is then an indication of
molecular interactions. From this reduction in lifetimes we
immediately obtain the experimental FRET efficiency:
E = 1 −
τ
DA
τ
D
. (10)
In figure 3 we have summarized the main concepts of
FRET, which are illustrated with the widely used FRET
couple eCFP (enhanced cyan-fluorescent protein, donor)
and eYFP (enhanced yellow-fluorescent protein, acceptor).
FRET is used extensively for monitoring interactions
and conformational changes between or within biological
macromolecules conjugated with suitable donor–acceptor
pairs. Because of its sensitivity FRET also forms the
basis for ‘sensing’ important biological molecules in many
applications.
2.5. Time-resolved fluorescence
Fluorescence decay measurements are a very useful tool for
investigating the molecular dynamics of excited states. The
two main methods for obtaining time-resolved fluorescence
data are single-photon timing (also called time-correlated
single photon counting (TCSPC)) and multi-frequency phase-
modulation fluorometry. Both techniques yield essentially the
same information and differ mainly in how the time-resolved
fluorescence data are obtained, for instance in the time domain
or in the frequency domain. The principle of t he single-
photon timing method is illustrated in figure 4 and that of
4

Meas. Sci. Technol. 21 (2010) 102002 Topical Review
(A)(B)
(C)(D)
Figure 3. Main concepts of FRET. The occurrence of FRET depends on distance (A) and the spectral overlap (B). The reduction of a
lifetime from 2.5 to 2.0 ns (D) corresponds to a transfer efficiency of 0.2 (20%) (C) and to a distance between CFP and YFP of 6.3 nm given
a critical distance of 5.0 nm.
(A)
(B)
Figure 4. (A) Schematics of time-correlated single photon counting. The arrival time of the first photon after an excitation pulse is
measured and stored in memory. The histogram of many arrival times of photons represents the ‘fluorescence intensity versus time’ curve.
(B) Total fluorescence decay analysis of eGFP obtained with discrete lifetime analysis. Excitation and emission wavelengths were 487 and
539 nm. Experimental (noisy curve) and fitted (black solid line) time-resolved fluorescence curves are shown together with the reference
compound erythrosine B (reference lifetime 85 ps). The optimal fit resulted in lifetimes of 1.47 ns (normalized amplitude 10%) and 2.70 ns
(normalized amplitude 90%).
5

Meas. Sci. Technol. 21 (2010) 102002 Topical Review
Figure 5. Determination of the phase difference (φ
ω
)and
modulation ratio (m
ω
) in the phase modulation method of
fluorescence lifetime measurements.
frequency domain measurements in figure 5. For more detailed
information on the two methods we refer to several textbooks,
edited books and reviews [21, 22, 27–31].
For a fluorophore, which decays mono-exponentially with
lifetime τ , the fluorescence δ-response function I(t) is
I(t) = α exp(−t/τ), (11)
where α is the pre-exponential factor or amplitude. In single-
photon timing experiments where the fluorophore is excited
by an excitation pulse p(t), the observed fluorescence decay,
I
obs
(t), is a convolution of I(t) and p(t):
I
obs
(t) =

t
0
I(t

)p(t − t

) dt

=

t
0
I(t − t

)p(t

) dt

= I ⊗ p, (12)
where ⊗ denotes the convolution operator, t and t

represent
time. The instrument response function p(t) is usually
obtained by measuring the scattered excitation light. The
use of reference compounds with well-known fluorescence
lifetimes is much better to avoid wavelength-dependent time
response (the colour effect) of photomultiplier tubes [32]
and other parts of the detection s cheme. When I
obs
(t) and
p(t) are experimentally known, the parameters α and τ of
I(t) can be estimated by a variety of least-squares fitting
techniques. The example shown in figure 4 is the experimental
and fitted fluorescence decay of eGFP in aqueous buffer
with reference compound erythrosin B in water [33]. The
time resolution that can be achieved using a TCSPC setup
depends much on the light source and detector properties.
Using high repetition rate mode-locked lasers in combination
with MCP PMTs, instrument response function widths of
30 ps are possible. Finally, deconvolution algorithms will
result in fluorescence lifetimes in the range of 5–20 ps.
Diode lasers and light-emitting diodes (LEDs) which are
much cheaper can also be used but have the disadvantage
giving longer excitation pulses (width in the range 50–
200 ps). However, fluorescence lifetimes of several hundred
picoseconds can still be resolved.
In the frequency-domain method (figure 5), the values
of the phase shift, φ, and the relative modulation, m,fora
fluorophore with lifetime τ ,aregivenby
φ = tan
−1
(ωτ ) (13)
m =
1
√
1+ω
2
τ
2
, (14)
where ω is the angular frequency of the sinusoidally modulated
excitation light. The fluorescence lifetime can be determined
independently from the phase shift (equation (13)) and the
relative modulation (equation (14)). A good test for a single-
exponential decay is the identity of the τ values determined by
phase and modulation, whatever the modulation frequency
ω = 2πν is ( ν the generator-set frequency). For precise
determination of fluorescence lifetimes, the phase difference
(φ) and the modulation ratio (m) are measured as function
of different frequencies ω. The curves can be analysed
by least-squares fitting using theoretical expressions of the
sine and cosine Fourier transforms of the δ-pulse response.
The resolution of the phase-modulation method depends
on the modulation frequency. Fluorescence lifetimes in
the picosecond range can be measured using modulation
frequencies between 2 and 10 GHz or using the harmonics of
a mode-locked laser pulse train. Fluorescence lifetimes in the
1–10 ns range can be measured using modulation frequencies
between 2 and 200 MHz.
2.6. Fluorescence anisotropy
Photons can be absorbed only when their energy fits to the
energy gap between the ground and excited energy levels of
a particular molecule. Another condition for light absorption
is that the electric component or vector of the electromagnetic
wave must be parallel or close to parallel, to the transition
moment of the molecule. In a solution the orientations of
the transition moments are completely random. Therefore,
if we excite such a system with linearly polarized light,
the excitation will be efficient only for those molecules
whose transition moments are at the moment of excitation
oriented similarly to the direction of polarization. The
initial distribution of orientations of the excited molecules
will then be highly anisotropic. This ordering effect is
called photoselection. After excitation, the molecules start
to fluoresce with their characteristic fluorescence lifetime
τ and, simultaneously, Brownian rotational motion causes
the initial orientational order of the excited molecules to
vanish. Polarization of fluorescence is determined by the
orientation of the fluorophores’ transition moment at the
instant of fluorescence emission. This gives an opportunity to
determine rotational diffusion of fluorophores by detecting the
anisotropy of their fluorescence (see figure 6). Fluorophores in
the sample are excited with short pulses of vertically polarized
light causing photoselection. A polarizer in the fluorescence
channel (x) can be rotated from t he vertical to the horizontal
position. In the first part of the experiment, the decay of
intensity of vertically polarized fluorescence I

(t) is measured.
Then the polarizer is moved to horizontal orientation and I
⊥
(t)
6

Meas. Sci. Technol. 21 (2010) 102002 Topical Review
(A)
(B)
Figure 6. (A) Geometry of fluorescence anisotropy experiment.
(B) Definition of anisotropy r(t) and illustration how the individual
polarized intensity components are globally analysed to a model
function of r(t) with known total fluorescence decay function I(t).
is detected. Both decays are combined to form the fluorescence
anisotropy function r(t), defined as
r(t) =
I
||
(t) − I
⊥
(t)
I
||
(t) +2I
⊥
(t)
. (15)
For spherical and freely rotating fluorophores the anisotropy
decays exponentially:
r(t) = r
0
exp(−t/τ
c
) (16)
with the characteristic decay constant τ
c
, called rotational
correlation time, and r
0
= 2/5 for randomly distributed
molecular systems where the transition moments of absorption
and emission coincide.
According to the theory of rotational diffusion, τ
c
is
related to the hydrated radius r
h
of a rotating sphere by the
equation:
τ
c
=
4πr
3
h
η
3k
B
T
. (17)
Here k
B
T is the Boltzmann factor and η is the viscosity of the
solvent.
As an example of rotational motion the fluorescence
anisotropy decay of eGFP in aqueous buffer has been presented
in figure 7. The fluorophore is tightly bound in the protein
matrix and rotates together with the whole protein with a
correlation time of 14 ns, in agreement with equation (17).
It is also clear that the initial fluorescence anisotropy (at
t = 0) is close to the limiting value of 0.4, implying nearly
parallel absorption and transition moments. The steady-state
anisotropy ris obtained from integration of the time-resolved
anisotropy, weighted by the total fluorescence intensity decay
I(t), which is assumed to be an exponential with fluorescence
lifetime τ :
r=

∞
0
r(t)I(t)dt

∞
0
I(t)dt
. (18)
(B)
(A)
Figure 7. (A) Molecular model of eGFP and rotation around a
cylindrical axis (dashed arrow). (B) Experimental (noisy curve) and
fitted (black solid line) fluorescence anisotropy decay curve of eGFP
in aqueous buffer (pH 7.8). The initial anisotropy r
0
= 0.38 and the
rotational correlation time τ
c
= 14 ns.
When rotations form the main depolarization process the
steady-state fluorescence anisotropy can be related to the
famous Perrin equation (14):
r=
r
0
1+
τ
τ
c
, (19)
where τ is the fluorescence lifetime and τ
c
is the r otational
correlation time.
Another source to detect in anisotropy decay is when
energy transfer takes place. Actually, this technique is
in principle sufficient to detect homo-FRET between two
identical fluorescent proteins. Shown in figure 8 is the
fluorescence anisotropy decay of two eGFP molecules linked
by a peptide giving a bi-exponential anisotropy decay with
a long component due to rotation of dimeric eGFP and a
short correlation time τ
T
, which arises from reversible energy
transfer and is equal to the reciprocal transfer rate constant:
τ
T
=
1
2k
T
, (20)
where the factor of 2 indicates the reversibility of the FRET
process. It should be noted that the critical distance R
0
(with
7

Meas. Sci. Technol. 21 (2010) 102002 Topical Review
(B)
(A)
Figure 8. (A) Schematic model of dimeric eGFP and non-radiative,
reversible energy transfer between the chromophores (dashed
two-sided arrow). (B) Experimental (noisy curve) and fitted (black
solid line) fluorescence anisotropy decay curve of dimeric eGFP in
aqueous buffer (pH 7.8). The initial anisotropy r
0
= 0.36 and the
(long) rotational correlation time τ
c
= 28 ns and (short) transfer
correlation time τ
T
= 1.4 ns.
κ
2
= 1) between two eGFP molecules is 4.9 nm and identical
to that between eCFP and eYFP [34].
When the sole depolarization process is homo-FRET in
a dimeric system with transition moments fixed in space
(no motion), the fluorescence anisotropy function obeys an
exponential decay function plus a constant term [35, 36]:
r(t) = β
1
exp(−2k
T
t) + β
2
= (r
0
− r
∞
) exp(−2k
T
t) + r
∞
.
(21)
By summation and subtraction of the amplitudes β
1
and
β
2
, information on the intra- and intermolecular transition
moments can be obtained:
β
2
+ β
1
=
3
5
cos
2
δ −
1
5
β
2
− β
1
=
3
5
cos
2
θ
T
−
1
5
,
(22)
where δ is the intramolecular angle between the absorption
and emission transition moments in one molecule and θ
T
is
the intermolecular angle between the absorption transition
moment of molecule 1 and the emission transition moment of
molecule 2. The brackets in equation (22) denote an average of
two values (1 → 2 and 2 → 1). The steady-state fluorescence
anisotropy r becomes for this case:
r=
β
1
1+2k
T
τ
+ β
2
=
r
0
− r
∞
1+2k
T
τ
+ r
∞
. (23)
3. Fluorescence lifetime measurements under a
microscope
The integration of fluorescence spectroscopy in light
microscopy adds a new dimension to microscopy, since spatial
information about the molecular behaviour of fluorescent
molecules can be obtained. The microscopic contrast is
provided by fluorescence parameters such as quantum yield
(intensity), spectrum, lifetime, anisotropy (polarization) or
combinations. The popularity of fluorescence microscopy
arises from the fact that fluorescence is both extremely
sensitive and minimally invasive, which can be applied
to living cells, tissues or whole bodies. Let us take
the measurements of fluorescence lifetimes under the
microscope as an example. The acronym FLIM can cover
identical meanings, namely Fluorescence Lifetime I maging
Microscopy, Fluorescence Lifetime IMaging or Fluorescence
LIfetime Microscopy. Fluorescence lifetimes are absolute
quantities. The most important advantage of FLIM over
fluorescence intensity imaging is that fluorescence lifetimes
are independent of fluorophore concentration and laser
excitation intensity. Since the fluorescence lifetime of a
fluorophore is sensitive to the local environment (pH, charge,
presence of quenchers, refractive index, temperature and
so forth), their measurements under a microscope offer
the important advantage of contrast by spatial variations of
lifetimes. Methods to visualize molecular interactions can be
based on FRET, for example measured by FLIM. Detection of
FRET in a microscopic i mage using the quenching of donor
fluorescence quantum yield is not straightforward, since one
should have knowledge on the fraction of donor molecules
having an acceptor in the vicinity as compared to those
having no acceptor. Detection of FRET under a microscope
using the shortening of the donor fluorescence lifetime is a
much more direct method than that of fluorescence intensities.
Using FLIM the donor fluorescence lifetime as a function
of space can be measured. Suppose that the cross-section
of a microscopic object such as a cell containing fluorescent
donor and acceptor molecules, which are spatially divided
into 100 × 100 pixels. All pixels contain 10
4
decay curves.
In each microscopic pixel a donor fluorescence lifetime can
be calculated from the decay curve. FRET imaging by donor
fluorescence lifetime reduction under the microscope allows
quantitative detection of interacting proteins in live cells.
Various review papers and edited books on FLIM and FRET
combinations have appeared in the recent literature with many
back [37–42]. The book edited by Periasamy and Clegg [41],
in particular, provides a comprehensive treatment of theory,
instrumentation, data analysis and biological applications of
FLIM.
8

Meas. Sci. Technol. 21 (2010) 102002 Topical Review
4. FLIM instrumentation
4.1. Widefield FLIM
For widefield FLIM detection, gated or modulated image
intensifiers are used in combination with CCD readout. FLIM
can be performed in either time- or frequency domain. For
time-domain measurements short pulses of excitation light are
used and the fluorescence decay is directly measured with a
gated detection system or TCSPC. Frequency-domain FLIM
measurements are performed by periodic intensity modulation
of the excitation source and (homodyne or heterodyne) phase-
sensitive detection. Both time- and frequency-domain FLIM
systems require high repetition rate pulsed lasers or rapidly
modulated continuous wave (CW) lasers, which traditionally
are the most expensive part of the setup. The introduction
of inexpensive light-emitting diodes (LEDs) has made the
technique more popular by greatly reducing the cost of FLIM
technology. Experimental FLIM measurements using LEDs
have been described for both time-domain [43] and frequency-
domain [44] setups. The advantage of using widefield FLIM
is the rapid acquisition speed, as lifetime information is
simultaneously collected in all the pixels of the image. This
permits imaging of fluorescence lifetimes in real time with
millisecond speed. A disadvantage is that the signal is sampled
only for a fraction of the emission time, resulting in a loss
of photons and limited lifetime accuracy. Widefield FLIM
systems have the highest accuracy per second at the cost of
higher excitation powers.
4.2. Time- and space-correlated single photon counting
The concept of time- and space-correlated single photon
counting (TSCSPC) method is used to obtain picosecond
FLIM and anisotropy data of microscopic samples [45–47].
TSCSPC is the imaging variant of the classical TCSPC
method. The TSCSPC quadrant-anode (QA) detector is
a non-scanning time-resolved imaging detector that avoids
excessively high average-peak-powers that are encountered in
single-point time-resolved scanning systems. By making use
of novel ‘ coded-anode’ detectors, simultaneous acquisition
of fluorescence lifetime information in up to 500 × 500
space channels is possible, at a time resolution of about
15 ps (1/10 of the instrument response function of 150 ps
full width at half maximum (FWHM)) and a dynamic range
of up to 10
7
. The unique feature of the setup is that for
each detected photon it is known (i) from which position it
was emitted, (ii) at what time it arrived after the excitation
pulse, (iii) at what absolute time it arrived and (iv) what its
colour and polarization were; in other words the full set of
system parameters describing the physical properties of each
individual photon can be obtained. With this microscope
setup it is possible to study the dynamics of intracellular
protein–protein interactions by monitoring changes in images
of (multi-exponential) fluorescence dynamics, time-resolved
fluorescence anisotropy or emission spectra, all at picosecond
time resolution. Because the QA-detector can cope with
minimal-invasive excitation levels, long-period observation of
these phenomena can be achieved. Several more recent FRET–
FLIM examples have been reported by using this detector
[48–50].
4.3. Scanning FLIM
For FLIM with beam scanning, time-correlated single-photon
counting (TCSPC) is widely employed [30]. Typically,
FLIM measurements are performed using either pulsed diode
lasers (one-photon excitation) or Ti:sapphire lasers (one or
two-photon excitation). The recent availability of compact
broadband supercontinuum laser sources allows one-photon
excitation across visible wavelengths of the spectrum [51].
The fluorescence photons are detected either with SPADs in
the descanned mode or using PMTs in a non-descanned setup.
A disadvantage of the TCSPC approach is that the photon
signal count rate must be kept significantly below the excitation
pulse rate (typically by a factor of 100) to prevent pulse ‘pile-
up’. TCSPC signals are accumulated until a sufficient number
of photons are collected. The low signal rate combined
with the sequential nature of data acquisition requires long
measurement times, in the order of 1 min or more. On the
other hand, TCSPC-based FLIM with scanning systems has
an excellent signal-to-noise ratio with the highest accuracy
per photon. The high signal-to-noise ratio enables optical
sectioning of fluorescence images with lifetime as read-out
parameter. Another advantage is that lifetimes can easily be
combined with spectral detection using multi-channel photon
counting detectors. A two-photon excitation scanning FLIM
system used in our laboratory for investigations in plants has
been described in [52, 53]. A schematic view of this setup
and associated FLIM data analysis (see section 9)aregivenin
figure 9.
4.4. FLIM with single-molecule detection sensitivity
An advanced confocal microscope setup using pulsed
excitation and TCSPC detection has been described that
sequentially registers the emitted photons from single
molecules in a solution allowing parallel recording
of fluorescence correlation/cross correlation spectroscopy
(FCS/FCCS), fluorescence lifetimes and pixel/image
information over time periods of hours [54]. Using this
four-channel (containing four SPADs) confocal microscope
one is able to simultaneously detect, in real time,
spectra, fluorescence intensity, lifetime and anisotropy [55].
The analysis of the pixel fluorescence information in
higher dimensional histograms maximizes the selectivity of
fluorescence microscopic methods for detection of specific
events. Multi-parameter fluorescence detection (MFD) in
confocal fluorescence microscopy has been recently reviewed
and its reliability illustrated for various molecular interaction
studies in different complex environments [56].
Another example to put forward in this review i s
fluorescence lifetime correlation spectroscopy (FLCS), which
is a hybrid method of TCSPC and FCS. A typical experiment
is performed in a single-channel confocal microscope using
a high-repetition pulsed picosecond laser whereby the
emitted photons are stored in a time-tagged time-resolved
9

Meas. Sci. Technol. 21 (2010) 102002 Topical Review
(A)
(B )(C)
(E )
(D)
Figure 9. Schematic overview of FLIM experiment and associated
analysis. (A) Diagram of setup. (B) Fluorescence intensity image of
a plant protoplast transfected with eCFP after excitation with 860 nm
light; red contour is the region of analysis. (C) FLIM image colour
coding of globally analysed fluorescence lifetime in each pixel.
(D) E xperimental and fitted fluorescence decays in 1 pixel. (E)
Frequency histogram of recovered lifetimes in all pixels.
mode. The principles are lucidly explained by Kapusta and
colleagues [57]. The method allows the separation of the
autocorrelation f unction of individual components of a mixture
of fluorophores, given that they have different fluorescence
lifetimes. The separation is performed by applying
mathematical lifetime filters calculated from individual and
overall fluorescence decay patterns. FLCS offers the unique
advantage when using a single dye to monitor its diffusion
in two distinct environments as done successfully for a
fluorescent lipid coexisting in a supported lipid bilayer and
single unilamellar vesicles [58] and for a dye intercalated in
unfolded and folded DNA domains [59].
4.5. Recent advances in FLIM
More rapid acquisition in multiphoton FLIM can be
accomplished by the use of multiple scanning beams and
spinning (Nipkow) discs. Kumar and co-workers [60]
used multi-beam (64 scanned beams) and multiphoton
excitation in combination with a 16-channel TCSPC detector
to detect optical sectioning of labelled pollen grains and
NAD(P)H autofluorescence in living cells. Multiphoton
excitation minimizes out-of-focus photobleaching and
multifocal excitation reduces in-plane photobleaching
effects.
Padilla-Parra and co-workers [61] employed both
picosecond time-domain FRET–FLIM with TSCSPC and
fast FRET–FLIM acquisition in live cells using multifocal
excitation and gated detection. They demonstrated the
presence of protein–protein interactions in a quantitative
manner by determining the minimal fraction of interacting
donor molecules. This fraction is directly obtained from
FLIM data without any fitting procedures. The authors have
applied the methodology to detect the interaction between
the bromodomain of a transcriptional factor and acetylated
histones (H4) in nuclei of living cells. The method is
independent of the type of donor such as eCFP having multiple
lifetimes or eGFP with a single lifetime. In a follow-
up study Padilla-Parra and colleagues have tested different
FRET couples such as eGFP peptide-linked to different red-
fluorescent protein acceptors and the new donor mTMP1 (teal-
fluorescent protein) linked via a polypeptide chain to eYFP
[62]. On the basis of the highest minimal fraction of interacting
donor molecules (0.70) the mTMP1-YFP FRET couple was
by far the best couple.
Van Munster and co-workers [63] integrated a spinning-
disc unit into a frequency-domain FLIM instrument and
observed a considerable reduction of measuring artefacts,
while maintaining the advantages of wide field. Grant and
colleagues [64] designed a time-domain optically sectioned
FLIM microscope developed for high-speed (10 frames per
second) live cell imaging, combining wide-field parallel-pixel
detection with confocal sectioning utilizing spinning Nipkow
disc microscopy.
With spectrally resolved FLIM, the lifetime and spectrum
of the fluorescence are recorded in each pixel of an image.
Different measuring schemes were designed. Becker and co-
workers [65] projected the fluorescence image on the back
aperture of the objective lens on a circular fibre bundle, which
output is matched to the input slit of a spectrograph. The
spectrum at the output of the spectrograph is projected on a
16-anode PMT. Fluorescence photons are counted by a TCSPC
acquisition card and distributed over four dimensions covering
photon arrival times, wavelengths and the coordinates of the
scan area.
Emission-wavelength-resolved FLIM (hyperspectral
FLIM) has been used in a slit-scanning microscope to provide
optical contrast by using the autofluorescence of unstained
tissue [66]. Combination with a tunable continuum laser
source allows the registration of the complete excitation–
emission–lifetime matrix, for instance to diagnose diseased
tissue [67].
Streak-camera-based systems for simultaneous
acquisition of fluorescence decay and spectrum under a
confocal microscope have also been utilized [68–71].
To improve lateral and axial spatial resolutions a
stimulated emission depletion (STED) FLIM system has
been described consisting of excitation light derived from
a supercontinuum source in a micro-structured optical fibre,
beam shaping and aberration-correction optics, a (commercial)
laser scanning confocal microscope, detection by TCSPC and
200 nm beads as test samples [72].
Very recently a fluorescence lifetime endomicroscope
employing a fibre bundle probe and TCSPC has been described
showing FLIM images of stained pollen grains, FRET of live
cells and tissue autofluorescence [73].
10

Meas. Sci. Technol. 21 (2010) 102002 Topical Review
Figure 10. Three possible approaches to design FRET biosensors.
5. Survey of non-FLIM approaches in cell biology to
determine molecular interaction
5.1. GFP-based biosensors
Visualization of biochemical processes in a living cell has
been made possible by the application of visible fluorescent
protein technology. The GFP from the jellyfish Aequorea
victoria was discovered in the early 1960s and the application
of this technology has been of crucial importance for imaging
of intracellular proteins [7]. Typically, cDNA of GFP is
coupled to cDNA of a protein of interest via genetic methods.
Transformation of animal- or plant cells using the fusion DNA
construct allows its fluorescence to be observed within several
hours. Currently, fluorescent proteins, also from coral species,
emitting from violet to red are available, enabling differently
labelled proteins to be tracked simultaneously in the cell
[74–78]. Different combinations of the fluorescent proteins
can be used as donor–acceptor pairs in FRET applications.
So far, the enhanced forms of cyan-fluorescent protein (eCFP)
and yellow-fluorescent protein (eYFP) are the most commonly
used FRET pair in cell biology. Many improvements of
the eCFP spectral properties have been made resulting in an
optimization of the eCFP–eYFP FRET pair. Kremers and
co-workers [79] have developed a series of monomeric super
CFP (sCFP) molecules. Campbell and co-workers [80]have
developed the first monomeric version of the tetrameric CFP
(cyan FP) cFP484 from Clavularia coral, which finally resulted
in a red-shifted CFP mTFP. A key condition for FRET is the
strong spectral overlap between donor emission- and acceptor
absorption spectra. Mutations of eCFP and eYFP variants
have resulted in a better FRET pair yielding a F
¨
orster distance
of about 5.0 nm. However, red-shifted donor–acceptor pairs
would increase the F
¨
orster radius due to the fourth power
contribution of the wavelength to the spectral overlap integral
(see equation (7)). The red-shifted FRET pair eGFP–mRFP
has a F
¨
orster radius of about 5.5 nm and studies of proteins
tagged with these fluorescent moieties have already shown that
this is a good pair for FRET–FLIM applications [81].
5.2. Split GFP-based interactions
An alternative fluorescence method for monitoring protein
interactions is based on the bimolecular fluorescence
complementation (BiFC) technology [82–84]. The BiFC is
a method where an intact GFP fluorophore is r estored from
its non-fluorescent N- and C-terminal domains. When two
potentially interacting proteins, fused to two complementing
GFP fragments, are in close proximity, the proteins can bring
the N- and C-terminal GFP domains into a favourable position
and orientation and thereby facilitate their association into
a functional fluorophore [84, 85]. Since only a standard
fluorescence microscope is required for this technique, this
approach is an attractive, alternative method to analyse protein
interactions [86]. However, the method has also some
drawbacks. First, the complementation is irreversible thereby
limiting to observe dynamic interactions. The second concern
is to find proper controls. As soon as a fluorescence signal
is detected, protein interaction is confirmed. However, a lack
of fluorescence signal does not necessarily implicate that the
proteins are far apart, because a wrong orientation may prevent
complementation.
5.3. FRET-based biosensors
Frommer and Wang and co-workers have reviewed in detail
the design and applications of genetically encoded FRET
biosensors [87, 88]. FRET-based biosensors have become
essential t ools to detect biological activities in living cells with
high spatial and temporal resolutions. They operate because
activity is accompanied by a change in the FRET signal via
ratiometric imaging of fluorescence intensities in donor and
acceptor channels (abbreviated as ratiometric imaging). Their
modes of action of FRET biosensors are outlined in figure 10
and further discussed.
11

Meas. Sci. Technol. 21 (2010) 102002 Topical Review
There are two approaches to design a FRET biosensor.
An intermolecular FRET probe monitors a FRET signal,
when two proteins both labelled with two different fluorescent
dyes come in close proximity. The second configuration,
an intramolecular FRET sensor, consists of a protein tagged
with two fluorescent molecules. Upon stimulation or ligand
binding a conformational change will result in an altered
FRET signal. The first FRET sensors were indicators for
in vivo measurements of calcium concentrations and were
nicknamed ‘cameleons’ because of the change in ratiometric
imaging [89–91]. Nagai and co-workers [91] have developed
one of the most sensitive cameleons (YC3.60), which
shows a fivefold difference in ratiometric imaging between
calcium-free and calcium-bound forms. Like the cameleons,
genetically encoded FRET indicators have been developed
for the intracellular measurement of protons (pH) [92] and
chloride [49], but also sensors for the detection of sugars
(maltose and glucose) have been reported [93]. Also the
onset of apoptosis in living cells can be monitored with a
FRET sensor. Proteolytic enzymes called caspases recognize
a specific peptide bond sequence. When this specific peptide
substrate flanked by two FRET-active fluorescent proteins is
genetically encoded in the cell, specific caspases will split
the two fluorescent domains and FRET disappears. In this
way, programmed cell death can be monitored by the action
of specific caspases [94]. Recently, dual FRET sensors have
been developed to detect simultaneously the levels of ions and
second messengers and caspase-3 activity [95, 96]. Niino and
co-workers [96] report on four-colour imaging using single-
wavelength excitation light and subsequent linear unmixing
of the fluorescence intensities in order to distinguish the
fluorescent proteins. In these experiments FRET sensors were
composed of either sapphire/RFP or eCFP/eYFP enabling
simultaneous imaging of cAMP and cGMP levels in single
cells.
5.4. Co-localization or molecular interaction of proteins
The question, whether a protein is monomeric, dimeric or
multimeric, can be addressed by several approaches. One
approach is immunoprecipitation of intact protein complexes,
also known as co-immunoprecipitation (Co-IP). Co-IP works
by selecting an antibody that recognizes a known protein that
is believed to exist in a larger protein complex. By targeting
this known member with an antibody, the entire protein
complex may be pulled out of solution. Mass spectrometry
analysis enables the identification of unknown members of
the complex. This concept of pulling protein complexes
out of solution is also referred to as a ‘pull-down’ and is
used regularly in cell biology to analyse protein interactions.
Another method is two-hybrid screening, also known as the
yeast two-hybrid system, for identification of protein–protein
or protein–DNA interactions. The surmise behind this method
is the activation of downstream reporter gene(s) by binding
of a transcription factor onto an upstream activating sequence
(UAS). For two-hybrid screening, the transcription factor is
split into two separate fragments. The binding domain
is necessary for binding to the UAS and the activating domain
is responsible for transcriptional activation.
Although the above-described methods are good reporters
for proteins interacting in a complex, most acquired data
are from a homogenized population of cells, which can be
considered as a non-natural habitat. The optical resolution
of a microscope allows detection of fluorescently labelled
proteins at sub-cellular level. At most, co-localization of
two proteins equipped with two different fluorophores can be
established, but molecular interactions between these proteins
on nanometre scale cannot be determined. The spatial
resolution of light microscopy is orders of magnitude larger
than the average size of a protein molecule (diameter about
3 nm for a globular protein of 30 kDa). Therefore, it is unclear
whether molecules observed in the same region under a light
microscope interact or not. One possibility to go beyond that
optical limitation is to apply FRET microscopy.
5.5. Intensity-based FRET imaging
Several methods to quantify FRET have been developed and
the three most common will be discussed briefly in this and
next paragraphs and in the next section. In general, the
combination of optical microscopy with FRET spectroscopy
provides quantitative, temporal and spatial information about
binding and interaction of proteins in vivo [25, 97]. FRET can
be quantified using steady-state or time-resolved fluorescence
techniques. In the steady-state approach the fluorescence
intensities of fluorescent donor and fluorescent acceptor are
monitored under a fluorescence microscope. This steady-
state approach relies on the observation that the fluorescence
intensity of the donor is reduced and that of the acceptor is
enhanced when energy transfer takes place. The disadvantage
of this approach is that the signals are highly dependent on
the local concentrations of donor and acceptor molecules.
Photobleaching needs to be avoided, because it alters the
effective donor and acceptor concentrations. Artefacts like
bleed-through of the donor fluorescence in the acceptor
detection channel and direct excitation of the acceptor also
need to be taken into account for quantitative analysis. In
several publications corrected FRET imaging methods have
been demonstrated and further improved [98–100]. Elder
and co-workers [101] have recently reviewed the protocols
required to obtain FRET efficiencies both from quenching
of donor fluorescence intensity and for sensitized acceptor
emission intensity.
5.6. FRET measured by acceptor photobleaching
Acceptor photobleaching (APB) is another method, in which
FRET can be used to examine intracellular molecular
interactions between proteins. The method has been applied
in several s tudies reviewed in [4]. APB is not only restricted
to intensity-based fluorescence microscopy, but can also be
applied to FLIM [102]. APB is often employed to prove the
occurrence of FRET. APB measurements under a confocal
microscope have been critically assessed by Karpova and
colleagues [103]. They describe APB experiments using a
confocal laser scanning microscope, which is not only able to
selectively excite both donor and acceptor fluorophores, but
can also selectively photobleach the fluorescent acceptor. The
12

Meas. Sci. Technol. 21 (2010) 102002 Topical Review
fluorescence intensities of donor and acceptor are determined
before and after applying a highly intense laser pulse that
irreversibly photobleaches the fluorescent acceptor. When
donor and acceptor molecules interact, the destruction of the
previously fluorescent acceptor will result in increased donor
fluorescence intensity, since the energy cannot be transferred
to the acceptor molecule any more. The measurement of FRET
efficiencies by the APB approach requires several checkpoints.
First, selective bleaching of the acceptor is required, because
bleaching of the donor will result in underestimation of donor
de-quenching [4, 104]. Second, APB is still an intensity-
based approach and therefore it is sensitive for both donor and
acceptor concentrations. By using a FRET indicator where
only one eCFP and one eYFP are connected by a linker, this
problem can be avoided. Third, the method is very often
confined to fixed cells because of rapid diffusion of fresh
acceptor molecules into the bleached area and rapid recovery
of acceptor fluorescence.
6. FLIM applications
6.1. FRET–FLIM
In the previous paragraphs intensity-based FRET methods for
the quantification of protein–protein interactions are described.
However, these methods show severe disadvantages because
the results are very much dependent on differences in
local probe concentrations and removal of experimental
artefacts. An alternative method is FLIM, which measures the
fluorescence lifetime pixel by pixel in a fluorescence image.
FLIM overcomes problems of intensity-based methods, since
only the fluorescence lifetime of the donor molecule is
determined. Molecular interaction between donor and
acceptor will result in the quenching of donor fluorescence.
This donor fluorescence quenching will result in a decrease of
the fluorescence lifetime, since energy transfer will introduce
an additional relaxation path from the excited state to the
ground state. The amount of lifetime reduction is directly
correlated with the experimental FRET efficiency E via
equation (10), which is a function of the distance between
donor and acceptor via equation (9).
Currently many FRET–FLIM applications have been
performed in a variety of different cell systems. In our
own laboratory we focus on plant research. Molecular
interactions of receptor kinases have been demonstrated in
plant protoplasts, which are plant cells without cell wall [105].
Specific transcription factor interactions in the nucleus of
plant protoplasts could be demonstrated as well [106 ]. The
formation of aquaporins, which are membrane-bound water-
transporting protein complexes, has been visualized in maize
protoplasts [107]. In all these experiments donor fluorescence
lifetimes were displayed as false-colour coded images. FLIM
images of cells expressed with proteins fused with only a donor
fluorescent protein are compared with those where proteins
of interest are separately fused with both donor and acceptor
fluorescent proteins. Differences in the colour coding of FLIM
images indicate where protein interactions are taking place.
Russinova and colleagues [105] have shown the presence
of receptor complexes only in specific areas of the plasma
membrane, whereas aquaporin proteins nicely displayed a
reduced donor lifetime throughout the complete membrane
[107]. The ability to dissect where specific molecular
interactions take place provides us with better insight into
particular processes occurring in plant systems.
In cancer research, gene expression profiling has provided
information in the prediction of cancer. However, microscopic
techniques are indispensable to understand the physiological
changes and to identify tumours having specific changes in
molecular pathways. Advanced in vivo imaging methods with
high spatial resolution and optimal contrast could assess post-
translational modifications and protein–protein interactions
involved in this disease [108]. FLIM measurements could
be performed in cell line models of cancer, fresh human
tissues and formalin-fixed paraffin-embedded tissue. In animal
models, dynamic deep-tissue FRET–FLIM imaging of cancer
cells is now also feasible allowing rapid screening of target
modifiers [108].
6.2. Multi-parameter FRET–FLIM applications
To date, most FRET imaging experiments have utilized only
single donor–acceptor pairs. Imaging of multiple FRET pairs
within a single cell—and thus to correlate multiple signalling
events—has been limited by the significant spectral overlap of
the commonly used genetically expressed fluorescent proteins.
Grant and co-workers [109] combined FRET–FLIM of the
novel red-shifted TagRFP/mPlum FRET pair with spectral
ratiometric imaging of an eCFP/Venus pair. The fluorophore
emission was separated maximally, while at the same time the
low quantum yield of the far-red-emitting acceptor mPlum was
avoided.
6.3. A novel, quantitative FRET–FLIM method: rise times of
acceptor fluorescence
Shortening of the donor fluorescence lifetime in FRET–FLIM
measurements is an indicator of FRET, and the difference be-
tween the donor lifetime with and without acceptor allows
quantification of the FRET process. A complicating issue
is the fact that FRET systems are often heterogeneous, since
they can also contain a population of donor molecules that
for several reasons cannot transfer their excitations. This has
been observed even for purified visible-protein FRET pairs
with time-resolved fluorescence spectroscopy [33, 110]. The
average donor lifetime then originates from interacting and
non-interacting molecules. Therefore, the average donor fluo-
rescence lifetime does not reflect the true FRET efficiency, and
distances between molecules calculated based on these life-
times are overestimated (they are simply too long). Another
method of measuring FRET uses the detection of the rise time
of acceptor fluorescence following donor excitation. The rise
time of the acceptor fluorescence is equal to the fluorescence
lifetime of the donor in the presence of acceptor. The main
advantage of this approach is that only those molecules that
are involved in energy transfer are monitored and thus pre-
selected. As compared to the more commonly used donor-
fluorescence-quenching method, the transfer rate/efficiency
can be accurately determined and thus yielding more accurate
13

Meas. Sci. Technol. 21 (2010) 102002 Topical Review
distances. Borst and co-workers [110] have demonstrated this
novel approach by picosecond-resolved FRET experiments on
the calcium-sensor yellow cameleon YC3.60. In the absence
and presence of calcium, the FRET efficiency changed from
approximately 50% to almost 100%. Because the transfer
efficiency in the calcium-bound form of YC3.60 approaches
100% (corresponding to highly quenched donor fluorescence),
the distance between the two fluorescent proteins must be so
short that they must be nearly adjacent to each other. From
analysis of the FRET experiments, geometric parameters
have been derived allowing the generation of cartoon repre-
sentations for calcium-free and calcium-bound conformations
[110].
Laptenok and colleagues [111] have applied this approach
in the global analysis of FLIM images. A fusion of the
FRET pair eCFP–eYFP (linker of 13 amino acids) was
transfected in plant protoplasts. Donor and acceptor images
after donor excitation were recorded sequentially and all FLIM
images were analysed simultaneously. Global analysis has
the advantage that donor fluorescence lifetime and acceptor
rise time can be linked in multiple FLIM images in order
to increase the precision of transfer rate determinations.
Acceptor rise times led to realistic distances between the
two fluorescent proteins in the construct, in contrast to the
unrealistically long distances derived from the average donor
fluorescence lifetime.
6.4. GFP and other dyes as FLIM-reporter molecules
GFP has gained widespread use as a reporter to visualize
physiological processes within cells. Other examples with
the use of FLIM are the ability to measure local refractive
indexes of proteins in living cells, and to determine the pH
in different cell organelles. The average fluorescence lifetime
of eGFP has been used to report on the refractive index of
its environment [112]. The inverse fluorescence lifetime of
GFP is proportional to the refractive index squared and the
effect has a long range in the order of micrometres. It was
found that GFP-tagged proteins in cells suspended in different
refractive-index media resulted in a decrease of the average
fluorescence lifetime of GFP on addition of glycerol or sucrose
to the media in which the fixed cells are suspended [113 ].
Van Manen and co-workers [114] showed local refractive
index of intracellular GFP by expressing GFP fusions of
Rac2 and gp91phox. These proteins are both subunits of the
phagocyte NADPH oxidase enzyme and confocal microscopy
showed that both GFP-Rac2 and GFP-gp91phox are targeted
to the cytosol and to membranes in human myeloid PLB-985
cells. FLIM experiments of the two eGFP-labelled proteins
resulted in average fluorescence lifetimes of 2.70 ns for
cytosolic eGFP-Rac2 and 2.31 ns for membrane-bound eGFP-
gp91phox. A comparison of these lifetimes with a calibration
curve obtained by measuring eGFP lifetimes in phosphate-
buffered saline (PBS)/glycerol mixtures of known refractive
index, yielded local refractive indices of cytosolic eGFP-Rac2
and membrane-targeted eGFP-gp91phox of 1.38 and 1.46,
respectively, which is in good correspondence with reported
values for the cytosol and plasma membrane measured by other
techniques [115].
Intracellular pH can be influenced by cellular metabolism,
ion channel conductivity, contractility, ion transport and
cell cycle control. It was found that the absorbance and
fluorescence properties of eGFP are pH dependent in aqueous
solutions and in intracellular compartments in living cells.
eGFP has two absorption bands arising from neutral form and
anionic form at around 400 nm and 490 nm, respectively.
Kneen and Haupts and co-workers already demonstrated a
tenfold reversible change in absorbance and fluorescence of
purified recombinant GFP mutants [115, 116]. Time-resolved
fluorescence studies indicate that the fluorescence lifetime
of the anionic form is in the 2–3 ns range and that of the
neutral form (having a blue-shifted emission) is in the sub-
nanosecond range in a buffer s olution [117]. The eGFP
mutant (F64L/S65T) has also been used to probe cytoplasmic
and organellar pH levels [115]. Mitochondrial pH was
found to be >7.5, but could not be determined accurately
because of the much lower pK
a
(6.0) of the GFP mutant
used. Llopis and co-workers [118] employed a number of
fluorescent protein mutants with pK
a
between 6.15 and 7.1
to measure the pH in cytosolic, mitochondrial and Golgi
regions of mammalian cells. The enhanced YFP variant
(S65G/S72A/T203Y/H231L) was found to be suitable for
measuring the pH in all three sub-cellular domains, while
eGFP was only suitable f or pH measurements in cytosolic
and Golgi regions [118]. Nakabayashi and co-workers
[119] applied FLIM of eGFP in human cervical carcinoma
cells. The fluorescence lifetime of eGFP in HeLa cells was
found to decrease with decreasing intracellular pH, which
probably arises from the pH-dependent ionic equilibrium of
the chromophore in the ground state [119]. Chemical dyes
can also be used for pH imaging but these probes are difficult
to introduce into cells and no organelle specificity is obtained.
Hille and co-workers [120] checked the suitability of several
fluorescent dyes for pH imaging in living cells using FLIM.
BCECF is the most suitable fluorescent dye out of four dyes
tried and displays the widest range of the average decay time,
both in vitro (3.0–3.9 ns) and in vivo (3.3–3.6 ns) [120].
Viscosity can directly be linked to diffusion of bio-
macromolecules. Changes in viscosity have been linked
to diseases and malfunction at cellular level because of
perturbation of mobility of chemicals influencing signalling
and transport. Kuimova and co-workers [121] developed a
new approach to image local intracellular microviscosity by
measuring the fluorescence lifetime of molecular motors with
the spatial resolution of a confocal microscope.
7. Time-resolved fluorescence anisotropy imaging
7.1. Fluorescence anisotropy imaging microscopy
Fluorescence anisotropy imaging microscopy (FAIM) has
been comprehensively reviewed by Tramier and Coppey-
Moisan providing a very useful introduction to this
topic including principles, technical requirements and back
references [122]. Rather than measuring rotational diffusion
in a living cell it has been pointed out that FRET between
identical fluorophores (for instance eGFP) will lead to a
14

Meas. Sci. Technol. 21 (2010) 102002 Topical Review
distinct depolarization process, which directly yields the
transfer rate constant (see equation (20)). This process is
called homo-FRET and its detection is an elegant way to s tudy
protein–protein interactions in cells. One distinct advantage
over hetero-FRET as in FRET–FLIM applications is the fact
that only a single fluorescent protein linked to the protein of
interest is needed.
Time-resolved fluorescence anisotropy imaging can be
performed in either the time-domain [122] or the frequency
domain [123]. Multifocal two-photon excitation [122, 124],
widefield [125] and confocal [126] methods can be used.
In the time domain, a pulsed laser is focused into a high-
numerical-aperture objective and the fluorescence collected
by the objective is detected by time gating or TCSPC.
The optical design of the microscope requires in the
detection light path either a rotatable polarizer with a single
detector or polarizing beamsplitter with two photo-detectors
for acquiring sequential or simultaneous polarization-resolved
measurements of the same sample spot, respectively. The
advantage of simultaneous acquisition of the orthogonal
polarization components is the faster acquisition minimizing
unwanted photobleaching and other disturbing effects like
sample movement and laser intensity fluctuations. Four
geometric intensity components of fluorescence polarization
in the microscope must be taken into account: I
vh
, I
hv
, I
vv
and I
hh
with the first index denoting excitation polarization
direction and the second index emission polarization direction.
I
vh
and I
hv
pertain to the parallel direction relative to the
propagation direction of the laser excitation and I
vv
and
I
hh
to the perpendicular direction. Since each geometric
intensity component has a different transmission and detection
efficiency and depolarization in the microscope objective must
be taken into account, the experimental anisotropy must be
corrected for by a G-factor, which is determined by steady-
state intensity measurements of the four geometric intensity
components [122]:
G =

I
vv
· I
vh
I
hh
· I
hv
. (24)
When this correction is not applied, the initial anisotropy r
0
is significantly reduced. However, it should be noted that the
depolarization arising from the use of high numerical aperture
objectives always leads to lower r
0
values. I n two-photon
excitation the theoretical r
0
value is higher (r
0
= 0.571) than
in one-photon excitation (r
0
=0.4) offering a higher dynamical
range of anisotropy measurements [127].
Since time-resolved fluorescence anisotropy imaging
requires sophisticated microscopic instrumentation and
sufficient time resolution to recover fast transfer correlation
times, it is technically less demanding to carry out steady-
state fluorescence anisotropy imaging. Since most FAIM
publications pertain to imaging steady-state fluorescence
anisotropy, it is therefore useful to investigate some limiting
situations. By using the Perrin equation (19) and substituting
τ = 2.6 ns (fluorescence lifetime of eGFP) and τ
c
= 24 ns
(rotational correlation time of eGFP in the cell cytoplasma)
we obtain r=0.90r
0
.Forτ
c
= 200 ns (rotational
correlation time of eGFP fused to large protein aggregate
or immobilized protein) r will not deviate from r
0
(r=
0.99r
0
). For a dimeric eGFP system with homo-FRET as
the only depolarization process (with k
T
 1/τ ) we can
distinguish two cases. When the transition moments are
aligned parallel (or antiparallel, θ
T
= 0
◦
), we arrive at the
result that r≈r
0
= r
∞
(for r
0
= 0.4). In other words, energy
transfer is invisible in this steady-state anisotropy experiment.
When both transition moments are perpendicular (θ
T
= 90
◦
),
it can be shown that r≈r
∞
= 0.1 (for r
0
= 0.4). Therefore
it can be concluded from these limiting situations that steady-
state fluorescence anisotropy for GFP-tagged proteins in cells
is much more sensitive to FRET than to rotational motion,
provided that the transition moments involved are not parallel.
7.2. Homo-FRET
The first application of homo-FRET between identical
GFP chromophores detected by time-resolved fluorescence
anisotropy imaging has been reported by Gautier and co-
workers [128]. GFP was tagged to monomers of Herpes
simplex virus (HSV-1) thymidine kinase (TK) yielding a
chimeric protein TK366GFP. In cells, this protein can be
present either as monomers or dimers. The rotational
correlation time of the monomeric GFP-fusion protein was
estimated to be close to 100 ns and that of the dimeric protein
twice as long. These correlation times cannot be recovered
from the anisotropy decay because the fluorescence lifetime
of eGFP is much shorter (2.6 ns) yielding a time range of about
15 ns to extract information from time-resolved fluorescence
anisotropy (see figure 7 and [129] to illustrate this point). The
fast transfer correlation time of 2.4 ns revealed from the fit
of the fluorescence anisotropy decay arises from homo-FRET
between two GFP chromophores of interacting TK366GFP
monomers. It should be noted that the transfer time is in the
same order of magnitude as the fluorescence lifetime and is
easily detected as a rapid relaxation in the anisotropy decay
curve.
Bader and colleagues [130] described the use
of fluorescence-anisotropy-based homo-FRET detection
methods to study clustering of identical proteins in cells. The
methods are evaluated on cells expressing GFP constructs that
contain one or two FK506-binding proteins. This system
can form dimers and oligomers in a controlled fashion in
order to establish an experimental relation between anisotropy
and cluster size. Quantitative cluster size measurements
of other membrane-bound GFP-fusion proteins have been
performed using anisotropy imaging. Experiments on
glycosylphosphatidylinisotol (GPI)-anchored proteins reveal
that GPI forms clusters with an average size of more than two
subunits, whereas for the epidermal growth factor receptor
(EGFR), approximately 40% of unstimulated receptors are
present in the plasma membrane as pre-existing dimers.
The methods are based on measurements of steady-state
fluorescence anisotropies r (see equation (19)) and on the
difference r between the initial anisotropy r
0
and the limiting
anisotropy r
∞
(see equation (21)) for both one-photon and
two-photon excitation. Runnels and Scarlata [131]have
previously provided the theoretical background to interpret
15

Meas. Sci. Technol. 21 (2010) 102002 Topical Review
the experimental data. They have shown that the steady-state
anisotropy r of oligomers consisting of N monomers will be
roughly N
−1
times the r value of the monomer. The time
resolution of the time-gated fluorescence anisotropy imaging
setup used by Bader and colleagues [126] was insufficient
to directly measure the transfer correlation time τ
T
(see
equation (20)).
7.3. Hetero-FRET
From the steady-state fluorescence anisotropy imaging papers,
reviewed by Levitt and colleagues [37], we wish to highlight
the contribution of Rizzo and Piston [132] that reports on
hetero-FRET between Cerulean (a bright monomeric version
of CFP that is the donor) peptide linked to Venus (a monomeric
YFP that is the acceptor) via depolarized sensitized emission.
In the absence of FRET, the fluorescence emission from
the donor fluorescent protein is highly polarized. Upon
donor excitation, depolarization of fluorescence emission is
observed only in the detection channel of the acceptor. This
approach is able to distinguish FRET between linked and
unlinked Cerulean and Venus fluorescent proteins in living
cells with a larger dynamic range than other approaches.
The observations are confirmed by steady-state fluorescence
anisotropy measurements obtained from aqueous solutions of
Cerulean alone and the Cerulean–Venus construct. There is
a distinct decrease in steady-state anisotropy in the sensitized
emission spectrum of the acceptor. We have performed time-
resolved fluorescence anisotropy measurements on a similar
system consisting of the purified calcium-sensing yellow
cameleon YC3.60 [110]. When the donor (eCFP in this case)
is excited and the time-dependent anisotropy of the sensitized
Venus is detected, the depolarization could be resolved in
time. The recovered relaxation time is the inverse transfer
rate constant (1/k
T
) and the end level of the anisotropy (r
∞
)is
determined by the mutual orientation of the transition moments
of donor and acceptor.
7.4. Imaging of intracellular NADH
Vishwasrao and colleagues [133] have presented an elegant
application of two-photon excited fluorescence lifetime and
time-resolved fluorescence anisotropy imaging experiments
of reduced nicotinamide adenine dinucleotide (NADH and/or
NADPH), which acts as electron donor in cellular oxidative
phosphorylation and glycolysis. The authors have employed
intrinsic NADH fluorescence as a natural, noninvasive probe
to study the metabolic transition from normoxia (tissue oxygen
level as in air) to hypoxia (lower oxygen level) in mitochondria
of neural tissue. The time-resolved anisotropy pattern is
peculiar, first a rapid decrease in anisotropy is observed
followed by a rise in anisotropy and, at longer time, decay
again. This pattern is characteristic for the presence of both
free and enzyme-bound states of intracellular NADH in neural
tissue. This pattern could be mimicked by measuring NADH
free in aqueous buffer (fast correlation time corresponding with
rapid rotational mobility) in equilibrium with NADH bound to
malate dehydrogenase (long correlation time corresponding to
slow rotation). Such a time-resolved fluorescence anisotropy
pattern can only be analysed by associated global analysis,
in which the fluorescence lifetime components of free NADH
are linked to the fast correlation time and those of bound
NADH to the long correlation time. Global analysis of
fluorescence and associated anisotropy decays of intrinsic
tissue fluorescence offers the possibility to determine the free
to bound NADH concentration ratio and draw conclusions
about neural energy metabolism. The response of NADH
to the metabolic transition from normoxia to hypoxia turned
out to be more complex than a simple increase in free NADH
concentration. The concentration of free NADH, and that of an
enzyme-bound form with a relatively short lifetime indicative
for a conformational change, increases preferentially over that
of other enzyme-bound NADH species. From the anisotropy
decay of free NADH (see equation (17 )), it is also observed
that the intracellular viscosity is reduced, likely due to osmotic
swelling of mitochondria.
Yu and Heikal [134] followed the same experimental
approach to image intracellular NADH in breast cancer
cells and normal breast cells for quantitative analysis of
NADH concentration and conformation. Two-photon FLIM
of intracellular NADH indicates sensitivity to both cell
pathology and inhibition of the respiratory chain activities
using potassium cyanide (KCN). The authors developed a non-
invasive assay to estimate the average NADH concentration,
which is 1.8 fold higher in cancer cells than in normal breast
cells. In addition, from time-resolved associated anisotropy
analysis the fractions of free and enzyme-bound conformations
of intracellular NADH could be established. These fractions
in normal breast cells are statistically different from those in
breast cancer cells.
Both studies demonstrate the potential of imaging
intracellular NADH dynamics for probing mitochondrial
anomalies associated with neurodegenerative diseases, cancer,
diabetes and aging. In general, the approach is applicable
to other metabolic and signalling pathways in living cells,
without the need for cell destruction as in conventional
biochemical assays.
8. FLIM data analysis
8.1. Global analysis
For accurate and quantitative analysis of FLIM data and of
FRET–FLIM data in particular, well-designed data analysis
protocols are required. Significant advantages and improved
accuracy in data analysis can be achieved by applying global
analysis, when data f rom different measurements are analysed
simultaneously. Using the fact that some parameters should
be invariant in different experiments significantly increases
the precision of the analysis. The dynamical features of
fluorescence decay are often well described by a small number
of kinetic processes, for which the fluorescence lifetimes in all
pixels have similar values, but the relative intensity values
may vary from pixel to pixel. Then the statistical challenge is
global analysis of the image: use of the fluorescence decay at
all locations to estimate the number of lifetimes associated with
the kinetic processes, as well as the amplitude of each kinetic
process at each location. The assumption that the fluorescence
16

Meas. Sci. Technol. 21 (2010) 102002 Topical Review
lifetime components are spatially invariant is a crucial one for
global analysis of a FLIM image, which has been justified
by Verveer and co-workers [135]. Given that typical FLIM
images represent on the order of 10
3
time points and 10
3
locations, meeting this challenge is computationally intensive
and requires large memory resources. Fortunately, it has been
demonstrated that global analysis of a bi-exponential kinetic
model function is for most cases sufficient, assuming the same
lifetimes but different amplitudes across all pixels measured.
Various robust and rapidly converging global fitting algorithms
have been tested on simulated and experimental FLIM data sets
[135–137].
We have also designed, tested and optimized a FLIM data
analysis protocol, which will be briefly summarized [138]. For
each pixel in a FLIM image, the fluorescence is measured as
a function of time. Each recording of the fluorescence decay
in 1 pixel can be considered as a separate experiment. The
resulting data of all pixels is stored as a matrix in which each
column represents the fluorescence decay associated with a
single pixel x, such that I(t
i
,x
j
) is the fluorescence intensity
at time t
i
in pixel x
j
. In practice, the fluorescence decay must
be convolved with the instrumental response function, which
is either obtained from a scattering sample or from a single-
fluorescence-lifetime standard. The matrix data set associated
with a single FLIM image
Ψ is modelled as a product of two
matrices C and A, where column i of matrix C represents the
time profile of component i of the fluorescence decay, and
column i of (transpose) matrix A represents the amplitude of
component i (α
i
) over all pixels. The free parameters of the
model are the lifetimes τ
i
and, for each pixel, the amplitudes
α
i
describing the fluorescence decay. In technical sense,
the parameter estimation problem associated with fitting the
model for
Ψ is an example of a separable nonlinear least-
squares optimization problem, which can be solved by a
variable projection algorithm. Full details are given in [ 138 –
140]. Software to reproduce the FLIM data analysis procedure
is provided in an open-source and freely available package
called ‘TIMP’ for ‘The R project for Statistical Computing’.
Very recently, estimation of parameters describing the rise
of acceptor fluorescence and the decay of donor fluorescence
via simultaneous global analysis of multiple FLIM images
has been implemented as well [ 111 ]. The example shown in
figure 9 (eCFP in plant protoplasts) has been analysed with
this software package.
8.2. Graphical analysis
Currently there is much interest in providing graphical
analysis of FLIM data that do not require nonlinear
fitting procedures. These analytical tools were originally
developed for frequency-domain FLIM applications [141,
142]. Graphical representation of FLIM data demonstrates
that a mixture of two components with different lifetimes
(such as FRET-active and FRET-inactive donor molecules)
can be resolved by frequency-domain FLIM measurements
at a single frequency [141]. Redford and Clegg [142]
have named this method a polar-plot representation in
analogy to a similar procedure used for analysis of dielectric
relaxation measurements. This polar-plot analysis provides a
diagnostic visualization of the participating fluorescent species
underlying a complex lifetime distribution. The method also
works for FLIM data obtained with TCSPC even when low
number of photon counts is collected in 1 pixel of an image.
Digman and colleagues [143] have transformed raw, pixelated
photon-counting histogram data into the phasor space. Like in
the polar-plot the phasor representation has a strong diagnostic
value. In the phasor plot of FLIM images, different fluorescent
species can be easily recognized, because the pixel values are
clustered in specific regions of the universal semi-circle. It
is also straightforward to identify a population of molecules
undergoing FRET and molecules giving rise to background
fluorescence. Trajectories between different clusters (for
instance, FRET-active and FRET-inactive populations) can be
investigated and quantitated by the combined use of cursor and
calculator. The large benefit of phasor representation is that
FLIM data analysis becomes more intuitive and accessible to
inexperienced users. The phasor approach has the potential
to analyse large FLIM data sets, because computationally
intensive, exponential decay analysis of large amount of pixels
is not required any more. Grecco and co-workers [144] applied
a Fourier filter to FRET–FLIM data obtained with TCSPC
and considerably enhances the robustness (and speed) of
global analysis using a bi-exponential decay model to recover
accurate, relative concentrations of FRET-active and FRET-
inactive molecular populations. As a benefit, the developed
method also yields the numerical values of the instrumental
response function, avoiding the use of extra measurements.
9. Conclusions and perspectives
We hope to have illustrated that the first decade of the
21st century has witnessed tremendous progress in FLIM
technology. This pertains not only to the improvement
of photonics hardware, thereby increasing the speed of
measurements and reducing data acquisition times, but also
to computational technology to cope with efficient data
processing and analysis. The progress goes hand in hand
with the GFP technology and other probe developments,
which enables us to look at nanoscale protein–protein
interactions in living cells. Label-free FLIM applications
like described for NADH as well as FLIM endoscopy also
promise great future because of clinical applications for tissue
monitoring of diseases. It is now timely to integrate advanced
fluorescence microscopy into systems biology, in particular
to study biochemical networks, imaging signal transduction
pathways and probing protein reaction states in living cells,
excellently reviewed in [145]. FLIM, and FRET–FLIM in
particular, has also a technological role in high-content and
high-throughput screening via specific assays. A general
high-throughput fluorescence microscopy approach has been
surveyed in [146], whereas an automatic FLIM setup has
been proposed to screen for ubiquination of α-synuclein, a
protein involved in Parkinson’s disease [147]. It has not been
emphasized in this review, but FLIM is also an important
tool in the study of microfluidic systems, in particular solvent
interactions and mixing [148] and DNA–dye interactions
17

Meas. Sci. Technol. 21 (2010) 102002 Topical Review
[149]. Summarizing, FLIM is a unique and versatile tool to be
used by scientists working at the multi-disciplinary interface
of biology, chemistry, physics and engineering.
Acknowledgments
We would be like to thank Nina Visser for preparing figures
and Sergey Laptenok for supplying figure 9.
References
[1] Wilson G S and Gifford R 2005 Biosensors for real-time in
vivo measurements Biosens. Bioelectron. 20 2388–403
[2] B unt G and Wouters F S 2004 Visualization of molecular
activities inside living cells with fluorescent labels Int. Rev.
Cytol. 237 205–77
[3] B astiaens P I and Squire A 1999 Fluorescence lifetime
imaging microscopy: spatial resolution of biochemical
processes in the cell Trends Cell Biol. 9 48–52
[4] L ippincott-Schwartz J, Snapp E and Kenworthy A 2001
Studying protein dynamics in living cells Nat. Rev. Mol.
Cell Biol. 2 444–56
[5] Hink M A, Bisseling T and Visser A J 2002 Imaging
protein-protein interactions in living cells Plant Mol. Biol.
50 871–83
[6] Wouters F S, Verveer P J and Bastiaens P I 2001 Imaging
biochemistry inside cells Trends Cell Biol. 11 203–11
[7] Tsien R Y 1998 The green fluorescent protein Annu. Rev.
Biochem. 67 67509–44
[8] Hell S W 2007 Far-field optical nanoscopy Science
316 1153–8
[9] J ares-Erijman E A and Jovin T M 2003 FRET imaging Nat.
Biotechnol. 21 1387–95
[10] Jares-Erijman E A and Jovin T M 2006 Imaging molecular
interactions in living cells by FRET microscopy Curr.
Opin. Chem. Biol. 10 409–16
[11] Stokes G G 1852 On the change of refrangibility of light
Phil. Trans. R. Soc. 30 463–562
[12] Stern O and Volmer M 1919
¨
Uber die Abklingungszeit der
Fluoreszenz Z. Phys. 20 183–8
[13] Wawilow S J and Lewschin W L 1923 Beitr
¨
age zur Frage
uber polarisiertes Fluoreszenzlicht von Farbstoffl
¨
osungen
Z. Phys. 16 135–54
[14] Perrin F 1929 La fluorescence des solutions Ann. Phys. Paris
12 169–275
[15] Gaviola E 1926 Die Abklingzeiten der Fluoreszenz von
Farbstoffl
¨
osungen Z. Phys. 35 748–56
[16] Jablonski A 1933 Efficiency of anti-Stokes fluorescence in
dyes Nature 131 839–40
[17] F
¨
orster T 1948 Zwischenmolekulare Energiewanderung und
Fluoreszenz Ann. Phys. 2 55–75
[18] Clegg R M 2006 The history of FRET Reviews in
Fluorescence ed C D Geddes and J R Lakowicz (New
York: Springer) pp 1–45
[19] Weber G 1953 Rotational Brownian motion and polarization
of the fluorescence of solutions Adv. Protein Chem.
8 415–59
[20] Venetta B D 1959 Microscope phase fluorometer for
determining the fluorescence lifetimes of fluorochromes
Rev. Sci. Instrum. 30 450–7
[21] Valeur B 2002 Molecular Fluorescence (Weinheim:
Wiley-VCH)
[22] Lakowicz J R 2006 Principles of Fluorescence Spectroscopy
3rd edn (New York: Springer)
[23] Goldys E M (ed) 2009 Fluorescence Applications in
Biotechnology and Life Sciences (Hoboken, NJ: Wiley
Blackwell)
[24] Wang X F and Herman B (ed) 1996 Fluorescence Imaging
Spectroscopy and Microscopy (New York: Wiley)
[25] Herman B 1998 Fluorescence Microscopy (New York:
Springer)
[26] Pawley J B (ed) 2006 Handbook of Biological Confocal
Microscopy (New York: Springer)
[27] O’Connor D V and Phillips D 1984 Time-Correlated Single
Photon Counting (London: Academic)
[28] Jameson D M, Gratton E and Hall R D 1984 The
measurement and analysis of heterogeneous emissions by
multifrequency phase and modulation fluorometry Appl.
Spectrosc. Rev. 20 55–106
[29] Gratton E, Jameson D M and Hall R D 1984 Multifrequency
phase and modulation fluorometry Annu. Rev. Biophys.
Bioeng. 13 105–24
[30] Becker W 2005 Advanced Time-Correlated Single Photon
Counting Techniques (Berlin: Springer)
[31] Birch D J S and Imhof R E 1991 Time-domain fluorescence
spectroscopy using time-correlated single-photon counting
Topics in Fluorescence Spectroscopy vol 1 ed J R
Lakowicz (New York: Plenum) pp 1–95
[32] Boens N et al 2007 Fluorescence lifetime standards for time
and frequency domain fluorescence spectroscopy Anal.
Chem. 79 2137–49
[33] Visser A J, Laptenok S P, Visser N V, van Hoek A, Birch D J,
Brochon J C and Borst J W 2010 Time-resolved FRET
fluorescence spectroscopy of visible fluorescent protein
pairs Eur. Biophys. J. 39 241–53
[34] Hink M A, Visser N V, Borst J W, van Hoek A and
Visser A J W G 2003 Practical use of corrected
fluorescence excitation and emission spectra of fluorescent
proteins in F
¨
orster resonance energy transfer (FRET)
studies J. Fluoresc. 13 185–8
[35] Tanaka F and Mataga N 1979 Theory of time-dependent
photo-selection in interacting fixed systems Photochem.
Photobiol. 29 1091–7
[36] Bastiaens P I H, van Hoek A, Benen J A E, Brochon J C
and Visser A J W G 1992 Conformational dynamics and
intersubunit energy transfer in wild-type and mutant
lipoamide dehydrogenase from Azotobacter vinelandii.A
multidimensional time-resolved polarized fluorescence
study Biophys. J. 63 839–53
[37] Levitt J A, Matthews D R, Ameer-Beg S M and Suhling K
2009 Fluorescence lifetime and polarization-resolved
imaging in cell biology Curr. Opin. Biotechnol.
20 28–36
[38] Suhling K, French P M W and Phillips D 2005 Time-resolved
fluorescence microscopy
Photochem. Photobiol. Sci.
4 13–22
[39] Festy F, Ameer-Beg S M, Ng T and Suhling K 2007 Imaging
proteins in vivo using fluorescence lifetime microscopy
Mol. Biosyst. 3 381–91
[40] Wallrabe H and Periasamy A 2005 Imaging protein
molecules using FRET and FLIM microscopy Curr. Opin.
Biotechnol. 16 19–27
[41] Periasamy A and Clegg R M (ed) 2009 FLIM Microscopy in
Biology and Medicine (Boca Raton: Chapman & Hall
/CRC)
[42] Gadella T W J (ed) 2009 FRET and FLIM Techniques
(Amsterdam: Elsevier)
[43] Kennedy G T, Elson D S, Hares J D, Munro I, Poher V,
French P M W and Neil M A A 2008 Fluorescence
lifetime imaging using light emitting diodes J. Phys. D:
Appl. Phys. 41 094012
[44] Elder A D, Frank J H, Swartling J, Dai X and Kaminski C F
2006 Calibration of a wide-field frequency-domain
fluorescence lifetime microscopy system using light
emitting diodes as light sources J. Microsc.
224 166–80
18

Meas. Sci. Technol. 21 (2010) 102002 Topical Review
[45] Tramier M, Gautier I, Piolot T, Ravalet S, Kemnitz K,
Coppey J, Durieux C, Mignotte V and Coppey-Moisan M
2002 Picosecond-hetero-FRET microscopy to probe
protein-protein interactions in live cells Biophys. J.
83 3570–7
[46] Tramier M, Kemnitz K, Durieux C, Coppey J, Denjean P,
Pansu R B and Coppey-Moisan M 2000 Restrained
torsional dynamics of nuclear DNA in living proliferative
mammalian cells Biophys. J. 78 2614–27
[47] Tramier M, Kemnitz K, Durieux C and Coppey-Moisan M
2004 Picosecond time-resolved microspectrofluorometry
in live cells exemplified by complex fluorescence
dynamics of popular probes ethidium and cyan fluorescent
protein J. Microsc. 213 110–8
[48] Millington M, Grindlay G J, Altenbach K, Neely R K,
Kolch W, B encina M, Read N D, Jones A C, Dryden D T
and Magennis S W 2007 High-precision FLIM-FRET in
fixed and living cells reveals heterogeneity in a simple
CFP-YFP fusion protein Biophys. Chem.
127 155–64
[49] Jose M, Nair D K, Reissner C, Hartig R and Zuschratter W
2007 Photophysics of Clomeleon by FLIM: discriminating
excited state reactions along neuronal development
Biophys. J. 92 2237–54
[50] Jose M, Nair D K, Altrock W D, Dresbach T,
Gundelfinger E D and Zuschratter W 2008 Investigating
interactions mediated by the presynaptic protein bassoon
in living cells by Foerster’s resonance energy transfer and
fluorescence lifetime imaging microscopy Biophys. J.
94 1483–96
[51] Dunsby C et al 2004 An electronically tunable ultrafast laser
source applied to fluorescence imaging and fluorescence
lifetime imaging microscopy J. Phys. D: Appl. Phys.
37 3296–303
[52] Borst J W, Hink M A, van Hoek A and Visser A J W G 2003
Multiphoton microspectroscopy in living plant cells Proc.
SPIE 4963 231–8
[53] Borst J W, Nougalli-Tonaco I, Hink M A, van Hoek A,
Immink R G H and Visser A J W G 2006 Protein-protein
interactions in vivo: use of biosensors based on FRET
Reviews in Fluorescence 2006 ed C D Geddes and
J R Lakowicz (New York: Springer) pp 341–57
[54] Felekyan S, K
¨
uhnemuth R, Kudryatsev V, Sandhagen C,
Becker W and Seidel C A M 2005 Full correlation from
picoseconds to seconds by time-resolved and time-
correlated single photon detection Rev. Sci. Instrum.
76 083104
[55] Widengren J, Kudryavtsev V, Antonik M, Berger S,
Gerken M and Seidel C A 2006 Single-molecule detection
and identification of multiple species by multiparameter
fluorescence detection Anal. Chem. 78 2039–50
[56] Weidtkamp-Peters S, Felekyan S, Bleckmann A, Simon R,
Becker W, K
¨
uhnemuth R and Seidel C A 2009
Multiparameter fluorescence image spectroscopy to study
molecular interactions Photochem. Photobiol. Sci.
8 470–80
[57] Kapusta P, Wahl M, Benda A, Hof M and Enderlein J 2007
Fluorescence lifetime correlation spectroscopy J.
Fluoresc. 17 43–8
[58] Benda A, Fagul’ova V, Deyneka A, Enderlein J and Hof M
2006 Fluorescence lifetime correlation spectroscopy
combined with lifetime tuning: new perspectives in
supported phospholipid bilayer research Langmuir
22 9580–5
[59] Humpolickova J, Benda A, Sykora J, Machan R, Kral T,
Gasinska B, Enderlein J and Hof M 2008 Equilibrium
dynamics of spermine-induced plasmid DNA
condensation revealed by fluorescence lifetime correlation
spectroscopy Biophys. J. 94 L17–9
[60] Kumar S et al 2007 Multifocal multiphoton excitation and
time correlated single photon counting detection for 3-D
fluorescence lifetime imaging Opt. Express 15 12548–61
[61] Padilla-Parra S, Auduge N, Coppey-Moisan M and
Tramier M 2008 Quantitative FRET analysis by fast
acquisition time domain FLIM at high spatial resolution in
living cells Biophys. J. 95 2976–88
[62] Padilla-Parra S, Auduge N, Lalucque H, Mevel J C,
Coppey-Moisan M and Tramier M 2009 Quantitative
comparison of different fluorescent protein couples for fast
FRET-FLIM acquisition Biophys. J. 97
2368–76
[63] van Munster E B, Goedhart J, Kremers G J, Manders E M
and Gadella T W Jr 2007 Combination of a spinning disc
confocal unit with frequency-domain fluorescence lifetime
imaging microscopy Cytometry A 71 207–14
[64] Grant D M et al 2007 High speed optically sectioned
fluorescence lifetime imaging permits study of live cell
signaling events Opt. Express 15 15656–73
[65] Becker W, Bergmann A and Biskup C 2007 Multispectral
fluorescence lifetime imaging by TCSPC Microsc. Res.
Tech. 70 403–9
[66] De Beule P et al 2007 Rapid hyperspectral fluorescence
lifetime imaging Microsc. Res. Tech. 70 481–4
[67] Owen D M, Auksorius E, Manning H B, Talbot C B,
de Beule P A, Dunsby C, Neil M A and French P M 2007
Excitation-resolved hyperspectral fluorescence lifetime
imaging using a UV-extended supercontinuum source Opt.
Lett. 32 3408–10
[68] Krishnan R V, Masuda A, Centonze V E and Herman B 2003
Quantitative imaging of protein-protein interactions by
multiphoton fluorescence lifetime imaging microscopy
using a streak camera J. Biomed. Opt. 8 362–7
[69] Krishnan R V, Saitoh H, Terada H, Centonze V E
and Herman B 2003 Development of a multiphoton
fluorescence lifetime imaging microscope using a streak
camera Rev. Sci. Instrum. 74 2714–21
[70] Biskup C, Zimmer T and Benndorf K 2004 FRET between
cardiac Na+ channel subunits measured with a confocal
microscope and a streak camera Nat. Biotechnol. 22 220–4
[71] Biskup C, Zimmer T, Kelbauskas L, Hoffmann B, Klocker N,
Becker W, Bergmann A and Benndorf K 2007
Multi-dimensional fluorescence lifetime and FRET
measurements Microsc. Res. Tech. 70 442–51
[72] Auksorius E, B oruah B R, Dunsby C, Lanigan P M,
Kennedy G, Neil M A and French P M 2008 Stimulated
emission depletion microscopy with a supercontinuum
source and fluorescence lifetime imaging Opt. Lett.
33 113–5
[73] Kennedy G T, Manning H B, Elson D S, Neil M A,
Stamp G W, Viellerobe B, Lacombe F, Dunsby C
and French P M 2010 A fluorescence lifetime imaging
scanning confocal endomicroscope J. Biophotonics
3 103–7
[74] Shaner N C, Campbell R E, Steinbach P A, Giepmans B N,
Palmer A E and Tsien R Y 2004 Improved monomeric red,
orange and yellow fluorescent proteins derived from
Discosoma sp. red fluorescent protein Nat. Biotechnol.
22 1567–72
[75] Shaner N C, Steinbach P A and Tsien R Y 2005 A guide to
choosing fluorescent proteins Nat. Methods 2 905–9
[76] Shu X, Shaner N C, Yarbrough C A, Tsien R Y
and Remington S J 2006 Novel chromophores and buried
charges control color in mFruits Biochemistry 45 9639–47
[77] Verkhusha V V and Lukyanov K A 2004 The molecular
properties and applications of Anthozoa fluorescent
proteins and chromoproteins Nat. Biotechnol. 22 289–96
[78] Giepmans B N, Adams S R, Ellisman M H and Tsien R Y
2006 The fluorescent toolbox for assessing protein
location and function Science 312 217–24
19

Meas. Sci. Technol. 21 (2010) 102002 Topical Review
[79] Kremers G J, Goedhart J, van Munster E B and
Gadella T W Jr 2006 Cyan and yellow super fluorescent
proteins with improved brightness, protein folding, and
FRET Forster radius Biochemistry 45 6570–80
[80] Ai H W, Henderson J N, Remington S J and Campbell R E
2006 Directed evolution of a monomeric, bright and
photostable version of Clavularia cyan fluorescent protein:
structural characterization and applications in fluorescence
imaging Biochem. J. 400 531–40
[81] Peter M, Ameer-Beg S M, Hughes M K, Keppler M D,
Prag S, Marsh M, Vojnovic B and Ng T 2005
Multiphoton-FLIM quantification of the EGFP-mRFP1
FRET pair for localization of membrane receptor-kinase
interactions Biophys. J. 88 1224–37
[82] Bracha-Drori K, Shichrur K, Katz A, Oliva M, Angelovici R,
Yalovsky S and Ohad N 2004 Detection of protein-protein
interactions in plants using bimolecular fluorescence
complementation Plant J. 40 419–27
[83] Walter M et al 2004 Visualization of protein interactions in
living plant cells using bimolecular fluorescence
complementation Plant J. 40 428–38
[84] Hu C D, Chinenov Y and Kerppola T K 2002 Visualization of
interactions among bZIP and Rel family proteins in living
cells using bimolecular fluorescence complementation
Mol. Cell 9 789–98
[85] Kerppola T K 2008 Bimolecular fluorescence
complementation (BiFC) analysis as a probe of protein
interactions in living cells Annu. Rev. Biophys. 37 465–87
[86] Weinthal D and Tzfira T 2009 Imaging protein-protein
interactions in plant cells by bimolecular fluorescence
complementation assay Trends Plant Sci. 14 59–63
[87] Frommer W B, Davidson M W and Campbell R E 2009
Genetically encoded biosensors based on engineered
fluorescent proteins Chem. Soc. Rev. 38 2833–41
[88] Wang H, Nakata E and Hamachi I 2009 Recent progress in
strategies for the creation of protein-based fluorescent
biosensors ChemBioChem 10 2560–77
[89] Miyawaki A, Griesbeck O, Heim R and Tsien R Y 1999
Dynamic and quantitative Ca
2+
measurements using
improved cameleons Proc. Natl Acad. Sci. USA
96 2135–40
[90] Miyawaki A, Llopis J, Heim R, McCaffery J M, Adams J A,
Ikura M and Tsien R Y 1997 Fluorescent indicators for
Ca
2+
based on green fluorescent proteins and calmodulin
Nature 388 882–7
[91] Nagai T, Yamada S, Tominaga T, Ichikawa M and
Miyawaki A 2004 Expanded dynamic range of fluorescent
indicators for Ca
2+
by circularly permuted yellow
fluorescent proteins Proc. Natl Acad. Sci. USA
101 10554–9
[92] Esposito A, Gralle M, Dani M A, Lange D and Wouters F S
2008 pHlameleons: a family of FRET-based protein
sensors for quantitative pH imaging Biochemistry
47 13115–26
[93] Medintz I L 2006 Recent progress in developing FRET-based
intracellular sensors for the detection of small molecule
nutrients and ligands Trends Biotechnol. 24 539–42
[94] Xu X, Gerard A L, Huang B C, Anderson D C, Payan D G
and Luo Y 1998 Detection of programmed cell death
using fluorescence energy transfer Nucleic Acids Res.
26 2034–5
[95] Ai H W, Hazelwood K L, Davidson M W and Campbell R E
2008 Fluorescent protein FRET pairs for ratiometric
imaging of dual biosensors Nat. Methods 5 401–3
[96] Niino Y, Hotta K and Oka K 2009 Simultaneous live cell
imaging using dual FRET sensors with a single excitation
light PLoS One 4 e6036
[97] Wallrabe H, Elangovan M, Burchard A, Periasamy A
and Barroso M 2003 Confocal FRET microscopy to
measure clustering of ligand-receptor complexes in
endocytic membranes Biophys. J. 85 559–71
[98] Gordon G W, Berry G, Liang X H, Levine B and Herman B
1998 Quantitative fluorescence resonance energy transfer
measurements using fluorescence microscopy Biophys. J.
74 2702–13
[99] Xia Z and Liu Y 2001 Reliable and global measurement of
fluorescence resonance energy transfer using fluorescence
microscopes Biophys. J. 81 2395–402
[100] Hoppe A, Christensen K and Swanson J A 2002 Fluorescence
resonance energy transfer-based stoichiometry in living
cells Biophys. J. 83 3652–64
[101] Elder A D, Domin A, Kaminski-Schierle G S, Lindon C,
Pines J, Esposito A and Kaminski C F 2009 A quantitative
protocol for dynamic measurements of protein interactions
by F
¨
orster resonance energy transfer-sensitized
fluorescence emission J. R. Soc. Interface
6 S59–81
[102] Wouters F S and Bastiaens P I 1999 Fluorescence lifetime
imaging of receptor tyrosine kinase activity in cells Curr.
Biol. 9 1127–30
[103] Karpova T S, Baumann C T, He L, Wu X, Grammer A,
Lipsky P, Hager G L and McNally J G 2003 Fluorescence
resonance energy transfer from cyan to yellow fluorescent
protein detected by acceptor photobleaching using
confocal microscopy and a single laser J. Microsc.
209 56–70
[104] Voss T C, Demarco I A and Day R N 2005 Quantitative
imaging of protein interactions in the cell nucleus
Biotechniques 38 413–24
[105] Russinova E, Borst J W, Kwaaitaal M, Cano-Delgado A,
Yin Y, Chory J and de Vries S C 2004 Heterodimerization
and endocytosis of Arabidopsis brassinosteroid receptors
BRI1 and AtSERK3 (BAK1) Plant Cell 16 3216–29
[106] Tonaco I A, Borst J W, de Vries S C, Angenent G C
and Immink R G 2006 In vivo imaging of MADS-box
transcription factor interactions J. Exp. Bot. 57 33–42
[107] Zelazny E, Borst J W, Muylaert M, Batoko H,
Hemminga M A and Chaumont F 2007 FRET imaging in
living maize cells reveals that plasma membrane
aquaporins interact to regulate their subcellular
localization Proc. Natl Acad. Sci. USA 104 12359–64
[108] Kelleher M T et al 2009 The potential of optical proteomic
technologies to individualize prognosis and guide rational
treatment for cancer patients Target Oncol. 4 235–52
[109] Grant D M et al 2008 Multiplexed FRET to image multiple
signaling events in live cells Biophys. J. 95 L69–71
[110] Borst J W et al 2008 Structural changes of yellow Cameleon
domains observed by quantitative FRET analysis and
polarized fluorescence correlation spectroscopy Biophys.
J. 95 5399–411
[111] Laptenok S P, Borst J W, Mullen K M, van Stokkum I H M,
Visser A J W G and van Amerongen H 2010 Global
analysis of F
¨
orster resonance energy transfer in live cells
measured by fluorescence lifetime imaging microscopy
exploiting the rise time of acceptor fluorescence Phys.
Chem. Chem. Phys. 12 7593–602
[112] Suhling K, Siegel J, Phillips D, French P M W,
L
´
ev
ˆ
eque-Fort S, Webb S E and Davis D M 2002 Imaging
the environment of green fluorescent protein Biophys. J.
83 3589–95
[113] Tregidgo C, Levitt J A and Suhling K 2008 Effect of
refractive index on the fluorescence lifetime of green
fluorescent protein J. Biomed. Opt. 13 031218
[114] van Manen H J, Verkuijlen P, Wittendorp P, Subramaniam V,
van den Berg T K, Roos D and Otto C 2008 Refractive
index sensing of green fluorescent proteins in living cells
using fluorescence lifetime imaging microscopy Biophys.
J. 94 L67–9
20

Meas. Sci. Technol. 21 (2010) 102002 Topical Review
[115] Kneen M, Farinas J, Li Y and Verkman A S 1998 Green
fluorescent protein as a noninvasive intracellular pH
indicator Biophys. J. 74 1591–9
[116] Haupts U, Maiti S, Schwille P and Webb W W 1998
Dynamics of fluorescence fluctuations in green fluorescent
protein observed by fluorescence correlation spectroscopy
Proc. Natl Acad. Sci. USA 95 13573–8
[117] Cotlet M et al 2001 Excited-state dynamics in the enhanced
green fluorescent protein mutant probed by picosecond
time-resolved single photon counting spectroscopy J.
Phys. Chem. B 105 4999–5006
[118] Llopis J, McCaffery J M, Miyawaki A, Farquhar M G
and Tsien R Y 1998 Measurement of cytosolic,
mitochondrial, and Golgi pH in single living cells with
green fluorescent proteins Proc. Natl Acad. Sci. USA
95 6803–8
[119] Nakabayashi T, Wang H P, Kinjo M and Ohta N 2008
Application of fluorescence lifetime imaging of enhanced
green fluorescent protein to intracellular pH measurements
Photochem. Photobiol. Sci. 7 668–70
[120] Hille C, Berg M, Bressel L, Munzke D, Primus P,
Lohmannsroben H G and Dosche C 2008 Time-domain
fluorescence lifetime imaging for intracellular pH sensing
in living tissues Anal. Bioanal. Chem. 391 1871–9
[121] Kuimova M K, Yahioglu G, Levitt J A and Suhling K 2008
Molecular rotor measures viscosity of live cells via
fluorescence lifetime imaging J. Am. Chem. Soc.
130 6672–3
[122] Tramier M and Coppey-Moisan M 2008 Fluorescence
anisotropy imaging microscopy for homo-FRET in living
cells Fluorescent Proteins ed K F Sullivan (London:
Academic) pp 395–414
[123] Clayton A H, Hanley Q S, Arndt-Jovin D J, Subramaniam V
and Jovin T M 2002 Dynamic fluorescence anisotropy
imaging microscopy in the frequency domain (rFLIM)
Biophys. J. 83 1631–49
[124] Benninger R K, Onfelt B, Neil M A, Davis D M and
French P M 2005 Fluorescence imaging of two-photon
linear dichroism: cholesterol depletion disrupts molecular
orientation in cell membranes Biophys. J. 88 609–22
[125] Suhling K, Siegel J, Lanigan P M, L
´
ev
ˆ
eque-Fort S,
Webb S E, Phillips D, Davis D M and French P M 2004
Time-resolved fluorescence anisotropy imaging applied to
live cells Opt. Lett. 29 584–6
[126] Bader A N, Hofman E G, van Bergen en Henegouwen P M
and Gerritsen H C 2007 Imaging of protein cluster sizes by
means of confocal time-gated fluorescence anisotropy
microscopy Opt. Express 15 6934–45
[127] Volkmer A, Subramaniam V, Birch D J and Jovin T M 2000
One- and two-photon excited fluorescence lifetimes and
anisotropy decays of green fluorescent proteins Biophys. J.
78 1589–98
[128] Gautier I, Tramier M, Durieux C, Coppey J, Pansu R B,
Nicolas J C, Kemnitz K and Coppey Moisan M 2001
Homo-FRET microscopy in living cells to measure
monomer-dimer transition of GFP-tagged proteins
Biophys. J. 80 3000–8
[129] Borst J W, Hink M A, van Hoek A and Visser A J W G 2005
Effects of refractive index and viscosity on fluorescence
and anisotropy decays of enhanced cyan and yellow
fluorescent proteins J. Fluoresc. 15 153–60
[130] Bader A N, Hofman E G, Voortman J, van Bergen en
Henegouwen P M and Gerritsen H C 2009 Homo-FRET
imaging enables quantification of protein cluster sizes with
subcellular resolution Biophys. J. 97 2613–22
[131] Runnels L W and Scarlata S F 1995 Theory and application
of fluorescence homotransfer to melittin oligomerization
Biophys. J. 69 1569–83
[132] Rizzo M A and Piston D W 2005 High-contrast imaging of
fluorescent protein FRET by fluorescence polarization
microscopy Biophys. J. 88 L14–6
[133] Vishwasrao H D, Heikal A A, Kasischke K A and
Webb W W 2005 Conformational dependence of
intracellular NADH on metabolic state revealed by
associated fluorescence anisotropy J. Biol. Chem.
280 25119–26
[134] Yu Q and Heikal A A 2009 Two-photon autofluorescence
dynamics imaging reveals sensitivity of intracellular
NADH concentration and conformation to cell physiology
at the single-cell level J.Photochem. Photobiol. B 95 46–57
[135] Verveer P J, Squire A and Bastiaens P I 2000 Global analysis
of fluorescence lifetime imaging microscopy data Biophys.
J. 78 2127–37
[136] Pelet S, Previte M J, Laiho L H and So P T 2004 A fast
global fitting algorithm for fluorescence lifetime imaging
microscopy based on image segmentation Biophys. J.
87 2807–17
[137] Barber P R , Ameer-Beg S M, Gilbey J, Carlin L M,
Keppler M, Ng T C and Vojnovic B 2009 Multiphoton
time-domain fluorescence lifetime imaging microscopy:
practical application to protein-protein interactions using
global analysis J. R. Soc. Interface 6 S93–105
[138] Laptenok S P, Mullen K M, Borst J W, van Stokkum I H M,
Apanasovich V V and Visser A J W G 2007 Fluorescence
Lifetime Imaging Microscopy (FLIM) data analysis with
TIMP J. Statist. Software 18 1–20
[139] van Stokkum I H, Larsen D S and van Grondelle R 2004
Global and target analysis of time-resolved spectra
Biochim. Biophys. Acta 1657 82–104
[140] Mullen K M, Vengris M and van Stokkum I H M 2007
Algorithms for separable nonlinear least squares with
application to modelling time-resolved spectra J. Global
Optim. 38 201–13
[141] Clayton A H, Hanley Q S and Verveer P J 2004 Graphical
representation and multicomponent analysis of
single-frequency fluorescence lifetime imaging
microscopy data J. Microsc. 213 1–5
[142] Redford G I and Clegg R M 2005 Polar plot representation
for frequency-domain analysis of fluorescence lifetimes J.
Fluoresc. 15 805–15
[143] Digman M A, Caiolfa V R, Zamai M and Gratton E 2008 The
phasor approach to fluorescence lifetime imaging analysis
Biophys. J. 94 L14–6
[144] Grecco H E, Roda-Navarro P and Verveer P J 2009 Global
analysis of time correlated single photon counting
FRET-FLIM data Opt. Express 17 6493–508
[145] Verveer P J and Bastiaens P I 2008 Quantitative microscopy
and systems biology: seeing the whole picture Histochem.
Cell. Biol. 130 833–43
[146] Pepperkok R and Ellenberg J 2006 High-throughput
fluorescence microscopy for systems biology Nat. Rev.
Mol. Cell. Biol. 7 690–6
[147] Esposito A, Dohm C P, Bahr M and Wouters F S 2007
Unsupervised fluorescence lifetime imaging microscopy
for high content and high throughput screening Mol. Cell
Proteomics 6 1446–54
[148] Benninger R, Hofmann O, McGinty J, Requejo-Isidro J,
Munro I, Neil M , de Mello A and French P 2005
Time-resolved fluorescence imaging of solvent
interactions in microfluidic devices Opt. Express
13 6275–85
[149] Benninger R K, Hofmann O, Onfelt B, Munro I, Dunsby C,
Davis D M, Neil M A, French P M and de Mello A J 2007
Fluorescence-lifetime imaging of DNA-dye interactions
within continuous-flow microfluidic systems Angew.
Chem. Int. Ed Engl. 46 2228–31
21