Review
TRENDS in Cell Biology Vol.11 No.5 May 2001
203
Imaging biochemistry inside cells
Fred S. Wouters, Peter J. Verveer and Philippe I.H. Bastiaens
Proteins provide the building blocks for multicomponent molecular units,
or pathways, from which higher cellular functions emerge. These units consist of
either assemblies of physically interacting proteins or dispersed biochemical
activities connected by rapidly diffusing second messengers, metabolic
intermediates, ions or other proteins. It will probably remain within the realm of
genetics to identify the ensemble of proteins that constitute these functional units
and to establish the first-order connectivity. The dynamics of interactions within
these protein machines can be assessed in living cells by the application of
fluorescence spectroscopy on a microscopic level, using fluorescent proteins that
are introduced within these functional units. Fluorescence is sensitive, specific
and non-invasive, and the spectroscopic properties of a fluorescent probe can be
analysed to obtain information on its molecular environment. The development
and use of sensors based on the genetically encoded variants of green-fluorescent
proteins has facilitated the observation of ‘live’ biochemistry on a microscopic
level, with the advantage of preserving the cellular context of biochemical
connectivity, compartmentalization and spatial organization. Protein activities
and interactions can be imaged and localized within a single cell, allowing
correlation with phenomena such as the cell cycle, migration and morphogenesis.
GFP
applications
the use of probes emitting in the near-infrared1,
and the localization of Zn2+ release was imaged with
anatomical resolution in organotypical hippocampal
cultures after electrical stimulation2. Another payoff
of spatial resolution is the possibility to integrate
data from different biosensors or other cell-state
parameters to gain additional information – for
example, on causal connections. The biological
machinery inside cells can be investigated by
various microscopic techniques and biosensors.
Here, we divide these approaches into three sections:
first, the redistribution of biosensors in cells detected
by their fluorescence intensity; second, fluorescent
indicators that change their intrinsic fluorescence
properties as a function of a physiological parameter
or chemical reaction state; and finally, sensors for
protein interactions.
Imaging fluorescence patterns
Fred S. Wouters
Peter J. Verveer
Philippe I.H. Bastiaens*
Cell Biology and Cell
Biophysics Program,
European Molecular
Biology Laboratory,
Meyerhofstraße 1,
D-69117 Heidelberg,
Germany.
*e-mail:
Philippe.Bastiaens@
embl-heidelberg.de
Fluorescence spectroscopy approaches, previously
confined to cuvette-based measurements, are
progressively making their way into the field of cell
biology. This novel development adds an aspect other
than spatial resolution to microscopy – detection of
protein activity in the cell. Mainly because of the
availability of an ever-increasing range of
intrinsically fluorescent proteins that can be
genetically fused to virtually any protein of interest,
the area of their application as fluorescent biosensors
has reached the inner workings of the living cell.
The most basic use of fluorescent biosensors is the
collection of photons from a cell or tissue to detect
the occurrence of a process with temporal resolution.
Spatial information expands the usability of
biosensors by adding subcellular and supracellular
information. On the subcellular level, the read-out of
the biosensor can be sampled with spatio-temporal
resolution, enabling the morphological dissection
of the studied process: for example, what are the
organelles or compartments participating and is
the process polarized? On the supracellular level,
fluorescence imaging of biosensors in a collection of
cells allows the determination of cell-to-cell variation.
Many studies show biosensor images of multiple cells
to demonstrate the repeatability or homogeneity of
the process. However, heterogeneity of a process in
a cell population can be of considerable biological
interest, and the simultaneous imaging of multiple
cells has been used to assess the extent of variation
in cellular responses. Apart from cell cultures,
biosensors can also be used to map processes in
multicellular tissues. For example, cathepsin D
proteolytic activity was imaged in tumors in situ by
A simple but useful design of a biosensor consists of a
minimal protein domain fused to green-fluorescent
protein (GFP) that interacts specifically with
molecules that are transiently generated at specific
sites in cells. This allows monitoring of secondmessenger generation by imaging translocation of the
fluorescent protein molecule. Examples of this type of
sensor are those based on lipid-binding domains such
as the C1 domains of protein kinase C (PKC)3, the PA
domain of Raf4 and the pleckstrin-homology (PH)
domain of several proteins5–7. After cell stimulation,
these three types of sensors translocate to the plasma
membrane upon an increase in diacylglycerol (DAG),
phosphatidic acid or 3′ phosphoinositide content,
respectively. Lipid metabolite production can be
quantified by determining the ratio of cytosolic to
membrane fluorescence intensity. This approach was
pioneered in an elegant study by Oancea and Meyer8
that showed how signal-transduction processes
use the same signaling molecules to achieve different
cellular responses. The kinetics of translocation of the
PKC DAG-binding C1 domain and Ca2+-binding C2
domain were compared with those of full-length PKC.
It was shown that signal specificity could be achieved
by temporal coordination of Ca2+ and DAG signals in
PKC activation. Sustained Ca2+ oscillations were
found to be necessary for the release of the
inaccessible ‘clamped’ C1 binding site at the plasma
membrane to allow a prolonged activation by DAG
binding. A further example where unique insight
was obtained with a fluorescent biosensor approach
is a study where a minimal PH domain was used to
image the generation and distribution of
3′-phosphoinositides in the plasma membrane by
http://tcb.trends.com 0962-8924/01/$ – see front matter © 2001 Elsevier Science Ltd. All rights reserved. PII: S0962-8924(01)01982-1
Review
204
TRENDS in Cell Biology Vol.11 No.5 May 2001
Box 1. Fluorescence intensity measurements of FRET in a microscope
Interactions between two proteins can be imaged by detecting
fluorescence resonance energy transfer (FRET) between donor
Protein
excitation, F DA(i), to obtain the sensitized emission. Apparent
FRET efficiencies, EA(i), can then be calculatedi,j, analogous
to Eqn II:
Acceptor-labeled
protein
E A(i) =
Sensitized
emission
FR
ET
GFP
Donor
emission
Donor
emission
TRENDS in Cell Biology
Fig. I.
and acceptor fluorophores (Fig. I). FRET is a photophysical
phenomenona–c where energy is transferred non-radiatively from
a donor fluorophore to an acceptor fluorophore with an efficiency
defined by:
E=
R 60
R 60
+r6
[I]
where r is the distance between the two fluorophores, and R0 is the
distance at which 50% energy transfer takes place (typically 2–6 nm).
There are a number of ways to measure FRET in a microscope that
can be broadly divided into intensity-based methods and
fluorescence decay-kinetics-based methods (see Table 1).
Excitation of a donor fluorophore in a FRET pair leads to
quenching of donor emission and in an increased, sensitized,
acceptor emission. Intensity-based FRET detection techniques
make use of these effects. Because it is possible to measure the
donor fluorescence emission specifically without the contaminating
leak-through of the acceptor emissiond, a good way to detect FRET
is to compare the quenched with the unquenched donor emission
after specific photobleaching of the acceptor fluorophoree–h. An
apparent energy transfer efficiency ED(i) can be defined in each
position i of an imagee, which is equal to the true FRET efficiency in
the complex E multiplied by the fraction of donor-tagged molecules,
αD(i ), that are in a complex at position i :
ED(i) = 1 −
F D (i)
D
F pb
(i)
= E ⋅ α D(i)
[II]
D (i) are the donor emission images before and
where F D(i) and Fpb
after photobleaching. The detection of donor quenching is
experimentally straightforward because of the intrinsic specificity of
the donor fluorescence detection and acceptor photo-destruction.
However, this method has limited applicability in living cells
because photobleaching requires prolonged illumination during
which relocation of the donor-tagged molecules can occur.
Detection of FRET through the sensitized acceptor emission
presents a more complex situation than donor-only detection
because the measured image must be corrected for leak-through
of the donor emission and for direct excitation of the acceptor.
The acceptor emission upon specific acceptor excitation, F A(i),
and the fluorescence through the donor filter upon donor
excitation, F D(i), are measured at each position i of an image.
These are used to correct the acceptor emission upon donor
http://tcb.trends.com
F DA(i) − F D(i) ⋅ RD − F A(i) ⋅ RE
F A(i)
= C ⋅ E ⋅ αA (i)
[III]
where the constant C is the ratio of donor and acceptor
brightness, and αA(i) is the fraction of acceptor-tagged
molecules that are in a complex at position i. RD is the ratio of
the detection efficiencies of the donor emission intensities
through the acceptor and donor filter sets. It can be obtained
from a sample that has only donor molecules by exciting at a
single wavelength and dividing the total intensities detected
through the acceptor and donor filters. RE is the ratio of the
extinction coefficients of the acceptor when excited at the
acceptor and donor wavelengths. It can be obtained by exciting
a sample with only acceptor molecules at the donor and
acceptor wavelengths, and dividing the total intensities
detected through the acceptor filter. These two ratios are
constant factors that are assumed to be spatially invariant in
the image. A possible pitfall with this approach is that the
quantum yields of donor and acceptor might vary in the sample
owing to environmental factors such as pH, compromising the
scalar correction factors RD and RE. In the case where relative
concentrations of donor and acceptor are constant at each
measurable position in a sample, a ratio measurement of F DA(i)
and F D(i) can be used to detect FRET qualitatively. This
approach should be avoided unless a constant stoichiometry
of the donor–acceptor pair can be ensured, as is the case in
chimeric constructs where donor and acceptor are attached to
the same molecule and changes in FRET occur owing to
changes in conformation (e.g. cameleonsk).
References
a Förster, T. (1948) Zwischenmolekulare Energiewanderung und Fluoreszenz.
Ann. Phys. 2, 57–75
b Clegg, R.M. (1996) Fluorescence resonance energy transfer. In Fluorescence
Imaging Spectroscopy and Microscopy (Wang, X.F. and Herman, B., eds),
pp. 179–252, Wiley
c Herman, B. (1989) Resonance energy transfer microscopy. Methods Cell Biol.
30, 219–243
d Bastiaens, P.I.H. and Pepperkok, R. (2000) Observing proteins in their
natural habitat: the living cell. Trends Biochem. Sci. 25, 631–637
e Bastiaens, P.I.H. and Jovin, T.M. (1998) FRET microscopy. In Cell Biology: A
Laboratory Handbook (Vol. 3) (Celis, J.E., ed.), pp. 136–146, Academic Press
f Wouters, F.S. et al. (1998) FRET microscopy demonstrates molecular
association of non-specific lipid transfer protein (nsL-TP) with fatty acid
oxidation enzymes in peroxisomes. EMBO J. 17, 7179–7189
g Bastiaens, P.I.H. et al. (1996) Imaging the intracellular trafficking and
state of the AB5 quaternary structure of cholera toxin. EMBO J. 15,
4246–4253
h Bastiaens, P.I.H. and Jovin, T.M. (1996) Microspectroscopic imaging
tracks the intracellular processing of a signal transduction protein:
fluorescent-labeled protein kinase C βI. Proc. Natl. Acad. Sci. U. S. A.
93, 8407–8412
i Gordon, G.W. et al. (1998) Quantitative fluorescence energy transfer
measurements using fluorescence microscopy. Biophys. J. 74, 2702–2713
j Nagy, P. et al. (1998) Intensity-based energy transfer measurements in
digital imaging microscopy. Eur. Biophys. J. 27, 377–389
k Miyawaki, A. et al. (1999) Dynamic and quantitative Ca2+ measurements
using improved cameleons. Proc. Natl. Acad. Sci. U. S. A. 96, 2135–2140
Review
TRENDS in Cell Biology Vol.11 No.5 May 2001
205
Table 1. Different approaches for measuring FRETa
Intensity based
Refs
Fluorescence-decay-kinetics based
Refs
Donor intensity with
acceptor photobleaching
20,28,34,39,45,48–50,52,54
Donor photobleaching
36,40,48–50,54,55,57,58
Donor intensity
1,56
Donor fluorescence lifetime
26–28,40,50,54,59,60
Sensitized emission
22,29,35-37,41–45,51
Donor–acceptor
intensity ratios
16,17,19,24,25,38
Acceptor lifetime in-growth
23
aThe
listed references refer to research that used the approaches given in each row.
Abbreviation: FRET, fluorescence resonance energy transfer.
using GFP fused to PH domains derived from the
ADP-ribosylation factor nucleotide-binding-site
opener (ARNO) and the general receptor for
3′-phosphoinositides (GRP1)7. Spatial gradients
of membrane 3′-phosphoinositides correlated with the
direction of platelet-derived growth factor (PDGF)induced chemotaxis of cells6. This type of sensor
design will certainly be applied to other protein
modules – for example to detect protein
phosphorylation by using Src-homology 2 (SH2) or
phosphotyrosine-binding (PTB) domains fused to
GFP variants. It will be interesting to see whether
proteins containing multiple phosphoamino-acidbinding domains are also able to integrate or decode
temporal patterns of protein phosphorylation in
signalling pathways.
Fluorescence indicators of physiological state or for
biochemical reactions
Chemical sensors for Ca2+ or other ions were probably
the first to use a change of fluorescence property, such
as quantum yield or spectral profile, upon chelation of
the ion. These probes allow the quantitative imaging
of the ion concentration inside cells after calibration
of their fluorescence response in a separate in vitro
experiment. Nowadays, a plethora of molecular
physiological indicators, often based on GFP, are
available for measuring H+ concentration9–13, Zn2+
concentration2, Cl– concentration14 and Ca2+
concentration15–18. In addition, biologically active
sequences can be incorporated into the fluorescent
probe. For example, GFP-based physiological
indicators can be targeted to specific cellular
compartments by incorporating a specific localization
signal in the fusion construct16,17.
Fusion constructs containing two GFP variants
allow the determination of pH inside cells and
organelles9. The GFP variants lose their emission
at low pH because of chromophore protonation. Their
different pKa values, ranging from 6.1 to 7.2, lead to
a difference of the relative intensity of the two GFP
variants, which can be quantified by ratio imaging.
By judicious choice of mutations, a pH-sensing probe
consisting of a single GFP mutant has been created
whose spectral properties, unlike native GFP, are
strongly coupled to the bulk pH so that it can be used
as a ratiometric probe10,12. One class of compound
sensors is based on a change in the efficiency of
http://tcb.trends.com
fluorescence resonance energy transfer (FRET)
between a donor and acceptor GFP variant (Box 1)
fused to protein domains or subunits that change
their interaction upon ligand binding. Ca2+-sensing
‘cameleons’16,17, consisting of blue/cyan GFP donor
molecules separated from green/yellow GFP acceptor
molecules by calmodulin and the M13 calmodulinbinding domain units, are probably the best-known
examples of this type of sensor. Ca2+ causes the
connecting units to interact, resulting in compression
of the construct and therefore an increase in the
efficiency of FRET. Since the donor–acceptor
stoichiometry of these kinds of constructs is constant,
this change can be reliably detected by the change in
fluorescence emission ratio of the two GFPs (see
Box 1). Similar probes have been designed to detect
small organic compounds such as ATP, GTP or
adenosine 3′,5′-cyclic monophosphate (cAMP). Back
in 1991, an indicator for the signaling molecule cAMP
was reported19. This sensor was based on a
reconstituted fluorescent enzyme, consisting of donorlabeled regulatory and acceptor-labeled catalytic
subunits of protein kinase A, that was microinjected
into cells. cAMP production leads to dissociation of
these subunits, which is detected by the
corresponding reduction in FRET efficiency between
the donor and acceptor fluorophores. Nowadays, a
genetically encoded sensor for cAMP is available were
the fluorophores are introduced by fusing the cyan
and yellow variants of GFP20. Recently, generic
fluorescent reporters were developed that are based
on the insertion of foreign peptide sequences into an
internal GFP loop15,21. Alternatively, these peptides
can be fused to the new C- and N-termini after
forming a circular permutation of GFP15,18. These
foreign grafts can be either two protein domains that
interact or sequences that, upon ligand binding,
induce a conformational change. Thus, a structural
change in GFP is effected that decreases fluorescence
emission from the chromophore, possibly by
decreasing its pKa. So far, this has only been applied
to the detection of Ca2+ ions15,18, but it will not take
long for other single GFP sensors to be developed that
report on the concentration of small molecules such
as cAMP or detect changes in phosphorylation or
proteolytic processing.
Complementary to the detection of small
compounds, fluorescent sensors have been developed
206
Review
TRENDS in Cell Biology Vol.11 No.5 May 2001
Box 2. Fluorescence decay kinetic measurements of FRET in a microscope
In kinetic-based approaches of fluorescence resonance
energy transfer (FRET) determination, the excited-state
decay kinetics of the donor or the acceptor
fluorescence are measured. The most widely used
method is fluorescence lifetime imaging microscopy
(FLIM)a,b, which makes use of the decrease in the
fluorescence lifetime of the donor owing to the
depopulation of its excited state by FRET. In these
approaches, a single lifetime value is obtained at each
position in an image. Assuming that this lifetime value
is a sum of the true lifetimes of the donor in bound
and unbound state, weighted by their respective
populations, and assuming single component lifetime
decays, an apparent energy transfer efficiency EL(i)
can be defined that is the same as for the acceptor
photobleaching method (see Box 1, Eqn I):
EL (i) = 1 −
τ (i)
= E ⋅ α D (i)
τD
[I]
where αD(i) is the fraction of donor-tagged molecules
that is in a complex, τ(i) is the average lifetime
detected at each position i, and τD is the true lifetime
of free donorb that can be obtained from a sample
without acceptor or after photobleaching the
acceptor. In practice, the lifetimes obtained by FLIM
are not a weighted sum of the true lifetimes, but this
is often a good approximation. For a description of
an exact quantitative approach with FLIM see Box 3.
Because the fluorescence lifetime is independent
of probe concentration and light path, τD does not
necessarily need to be calculated at each point in a
cell but can also be derived as an average from a
lifetime measurement after photobleachingc or from
a different sample without acceptor. This is especially
useful in live-cell imaging, where a reference cannot
be obtained at each point in a cell by acceptor
photobleaching. In fact, the contrast in measured
lifetimes is already a reliable indicator for spatial
variations in donor–acceptor association. A possible
to report on cellular protein activity or conformational
or covalent state. There is a great deal of interest in
monitoring proteolytic activity in cells, probably
originating from the interest in the cascade of
caspase activities that regulate apoptosis in cells.
A genetically encoded fluorescent probe for caspase
activity22–24 is based on a sequence of two variants
of GFP linked by a sequence that is recognized and
cleaved by a caspase, thereby abolishing FRET
between the molecules. Protein phosphorylation is
another important biological activity that has been
studied by FRET measurements. To monitor PKA
activity in cells, a conformational phosphorylation
sensor was made by fusing two GFP variants with the
kinase-inducible domain of the transcription factor
cAMP-responsive-element binding protein (CREB).
Phosphorylation of the sensor by PKA resulted in a
http://tcb.trends.com
pitfall for FLIM is that the fluorescence lifetime can
vary in an image owing to environmental influences.
Current implementations in a widefield microscope
have the added disadvantage compared with
intensity-based FRET measurements that the
spatial resolution is somewhat reduced. Confocal
fluorescence lifetime microscopy does not suffer
from that disadvantage. A recent promising approach
to FRET imaging with FLIM is the simultaneous
detection of the acceptor and donor fluorescence
decay kinetics, which enables the use of spectrally
similar donor–acceptor pairsd.
Rather than determining the fluorescence lifetime,
it is also possible to make use of the photobleaching
kinetics of the donore,f. The donor steady-state
intensity decays with a rate proportional to its
fluorescence lifetime and therefore the energy
efficiency calculation is the same as Eqn I, with the
photobleaching rates substituted for the lifetimes.
Obviously, this measurement is time-consuming and
can only be applied to fixed samples.
References
a Gadella, T.W.J., Jr and Jovin, T.M. (1995) Oligomerization
of epidermal growth-factor receptors on A431 cells studied
by time-resolved fluorescence imaging microscopy –
a stereochemical model for tyrosine kinase receptor
activation. J. Cell Biol. 129, 1543–1558
b Bastiaens, P.I.H. and Squire, A. (1999) Fluorescence lifetime
imaging microscopy: spatial resolution of biochemical
processes in the cell. Trends Cell Biol. 9, 48–52
c Wouters, F.S. and Bastiaens, P.I.H. (1999) Fluorescence
lifetime imaging of receptor tyrosine kinase activity in
cells. Curr. Biol. 9, 1127–1130
d Harpur, G.H. et al. (2001) Imaging FRET between spectrally
similar GFPmolecules in single cells. Nat. Biotechnol. 19, 167–169
e Jovin, T.M. and Arndt-Jovin, D.J. (1989) FRET microscopy:
digital imaging of fluorescence resonance energy transfer. In Cell
Structure and Function by Microspectrofluorometry (Kohen, E.
and Hirschberg, J.G., eds.), pp. 99–115, Academic Press
f Jovin, T.M. and Arndt-Jovin, D.J. (1989) Luminescence digital
imaging microscopy. Annu. Rev. Biophys. Biophys. Chem.
18, 271–308
reduction of FRET between the GFPs that could be
followed by ratiometric imaging in cells25 (Box 1).
Because these activity probes are based on genetic
fusion of the donor–acceptor pair, this design allows
straightforward ratio measurements to be used.
To examine situations where activity is detected by,
or arises from, interactions of initially dispersed
compounds, a different approach must be followed.
Examples are the measurements of the activity of
GFP-tagged PKC26 or ErbB127,28, as judged by their
phosphorylation state, by detecting FRET upon
binding of Cy3-labeled phosphoamino-acid-specific
antibodies. In cases such as these, ratio
measurements cannot be applied because a fixed
donor–acceptor stoichiometry cannot be assumed.
However, alternative fluorescence-microscopic
methods are available to detect quantitatively the
Review
GFP
Cy3
TRENDS in Cell Biology Vol.11 No.5 May 2001
Populations
0.9
0
0.8
0
Fig. 1. Spreading of ErbB1–GFP phosphorylation (green: ErbB1–GFP, red: EGF beads) upon a one-minute
stimulation with a bead covalently labeled with epidermal growth factor (EGF) and Cy5 (left panels).
ErbB1 phosphorylation was detected by fluorescence resonance energy transfer (FRET) between
ErbB1–GFP and Cy3-labeled antibodies. Shown are antibody staining patterns (middle panels) and maps
of the fraction of phosphorylated receptor (right panels) for detection with the generic PY72 antibody
against phosphotyrosine (upper panels) and the 9H2 antibody that specifically recognizes
phosphotyrosine 1173 in the activated ErbB1 (Alexis Biochemicals, San Diego, USA). Experimental details
and further examples can be found elsewhere27. Note that the extent and distribution of ErbB1 activation
are indistinguishable for both antibodies, demonstrating the specificity of the FRET measurement.
occurrence of FRET (Boxes 1 and 2). In these
phosphorylation studies, the activity of the donortagged proteins is detected, and thus the acceptortagged antibodies are present in excess, which is
necessary to detect all active proteins. FRET is
detected by the decrease in GFP fluorescence lifetime
using fluorescence lifetime imaging microscopy
(FLIM, see Box 2). Because the acceptor fluorescence
is excluded from the measurement, saturating
amounts of the acceptor-labeled antibody can be used
and absolute specificity of the antibody for the
phosphorylated epitope is not required. Furthermore,
the exclusive GFP readout in this approach enables
the detection of activated ErbB1 receptor against a
background of phosphotyrosine labeling on other
proteins. An example of this situation is presented in
Fig. 1, where FRET between GFP-tagged ErbB1 and
Cy3-labeled generic antibody against phosphotyrosine
detects the same receptor activation event as revealed
by use of Cy3-labeled specific antibody against the
phosphorylated ErbB1 receptor. This approach also
allows the quantitative determination of the fraction
of phosphorylated protein at microscopic resolution by
exploiting prior knowledge about the biological system
in a global analysis of the fluorescence decay data
(Box 3). By imaging the evolution of ErbB1
phosphorylation after stimulation with EGF
covalently attached to sub-micron beads, it was shown
in these studies that extensive lateral spreading of
receptor activation takes place after local stimulation.
http://tcb.trends.com
207
Another FRET-based activity measurement using
the diagnostic binding of a reporter molecule to detect
the active state of a protein was used for the GTPase
Rac29. In this study, an acceptor-labeled minimal
binding domain derived from the PAK1 kinase
(PBD) – that is known to bind to Rac exclusively when
Rac is in the GTP-bound (active) state – was used in
combination with a Rac–GFP chimera donor. In these
experiments, high levels of activated Rac were
observed in membrane ruffles, and a gradient of
activated protein was seen inside motile cells that
coincided with the direction of movement. This implies
that a graded, rather than localized, distribution of Rac
activation is responsible for directional persistence of
cellular motion. This experimental set-up represents
an elegant fluorescent analog of a classical biochemical
binding study. Here, binding was detected by FRETinduced sensitized acceptor-emission (see Box 1). In
this study, an excess of acceptor-PBD is needed to
detect all activated Rac proteins. To compare the
concentration of active Rac relative to that of total Rac,
the sensitized emission would need to be compared
with the total amount of fluorescence originating from
GFP–Rac in the absence of FRET, which is difficult to
achieve with this type of measurement. Furthermore,
it would have been possible to use a donor-based
method, such as acceptor-photobleaching, or FLIM,
to overcome the problems associated with excess
acceptor-tagged probe.
Probes for detecting protein interactions
As entire sequences of genomes from several
organisms, most notably Homo sapiens30,31, are
becoming increasingly available, uncovering the
functional connectivity of the proteome has become a
main effort of modern biology. Interactions are being
mapped by large-scale approaches32 to identify clusters
of proteins that perform specific functions. Optical
approaches in live cells will be instrumental in this
effort, but, mainly due to technological difficulties,
they have been applied primarily to detect functional
interactions between selected proteins of interest.
FRET measurements provide a very useful tool to
detect molecular associations of fluorescently tagged
proteins33. Some of the available approaches allow the
researcher to obtain quantitative information on
protein interactions as they occur in the living cell. This
sets the use of FRET in cells apart from the numerous
analytical biochemical and genetic approaches that are
currently in use. Accordingly, a clear increase in the use
of FRET-based interaction detection is apparent in the
recent body of scientific literature. An overview of
recent publications, classified by the approach taken,
can be found in Table 1. Here, we will discuss the
application of FRET approaches in these studies.
Detection of FRET could potentially be used in a
proteomics strategy to identify novel binding partners
or to establish an interaction matrix between sets of
(related) proteins. The obvious attraction of live cellbased interaction screening is that the measurements
208
Review
TRENDS in Cell Biology Vol.11 No.5 May 2001
Box 3. Quantitative detection of protein interactions by FLIM
Fluorescence resonance energy transfer (FRET) can be
used to image protein interactions (see Boxes 1 and 2).
In each resolvable volume element, a mixture of states
(bound or unbound) exists (Fig. Ia). The total intensity
state, and the influence of the environment in the
specimen, must be taken into account in the analysis.
In the simplest case, the fluorescence of each state can
be assumed to be a single exponential, and the
(a)
(b)
α( ) =
2
α(
5
)=
7
α(
7
α( ) =
11
α(
2
)=
α( )
Global
analysis
0
1
τ( ), τ(
FLIM data
at each point of an image is a sum of the intensities of
both states, weighted by the relative amounts α
(populations) of bound and unbound molecules. In
fluorescence lifetime imaging microscopy (FLIM), the
measured lifetime is a nonlinear weighted function of
the true lifetimes and the populations of each state. A
quantitative approach is desirable, where the
populations of each species are calculated, along with
the true energy-transfer efficiency in the complex. This
is feasible with FLIM as the donor fluorophore has
different fluorescence kinetics in the bound or
unbound states. Recently, instruments have become
available that enable the measurement of the
individual decay kinetics of bound and unbound
species along with their populationsa,b. Until now,
these have not been used to determine populations of
bound and unbound molecules by FRET, because
signal-to-noise issues complicate the measurements.
This can be solved using advanced data analysis
methods such as global analysisc that use a priori
knowledge of the biochemical system. The
fluorescence decay kinetics of each state must be
correctly modeled – that is, the correct decay model
(single exponential or multi-exponential) for each
are made in the context of relevant physiochemical
conditions and that spatial restrictions to possible
interactions, as imposed by compartmentalization,
are maintained. The screening for binding partners
constitutes an optical analog of a genome-wide yeasttwo-hybrid or other biochemical screening strategy.
To date, technological difficulties have prevented such
an approach on a large scale. Owing to the extremely
short working range of FRET and possible
unfavorable orientations of the donor–acceptor pair,
the chance of false-negative detection of interactions is
high. Intensity-based approaches to measure FRET in
)
0
Populations
Fig. I.
http://tcb.trends.com
1
13
13
)
TRENDS in Cell Biology
influences of the environment are negligibled. Global
analysis then enables estimation of the lifetime of
each state along with the relative concentrations in
each resolvable volume element (Fig. 1b), without
further reference measurements. This allows
calculation of the true FRET efficiency in the complex,
using the following expression:
E = 1−
τF
τD
[I]
where τD is the lifetime of the donor without FRET
(unbound state) and τF is the lifetime of the donor
when FRET occurs (bound state).
References
a Scully, A.D. et al. (1996) Development of a laser-based
fluorescence microscope with subnanosecond time resolution.
J. Fluoresc. 6, 119–125
b Squire, A. et al. (2000) Multiple frequency fluorescence
lifetime imaging microscopy. J. Microsc. 197, 136–149
c Verveer, P.J. et al. (2000) Global analysis of fluorescence
lifetime imaging microscopy data. Biophys. J. 78,
2127–2137
d Verveer, P.J. et al. (2000) Quantitative imaging of lateral
ErbB1 receptor signal propagation in the plasma membrane.
Science 290, 1567–1570
addition suffer from a high level of false-positive
detections when not properly disentangled from direct
fluorescence contributions. This is further complicated
when spatial variations in the quantum yield – for
example, owing to pH effects – affect the fluorescence
emission intensity (Box 1). Another problem is that
the noise introduced by the corrections can be so
large that the estimated sensitized emission is not
statistically significant. Excellent papers are
available that provide the guidelines for proper FRET
determination34–37, and a brief overview is provided
in Box 1.
Review
TRENDS in Cell Biology Vol.11 No.5 May 2001
A smaller-scale approach to study the function of a
collection of proteins that constitute a given pathway
has been successful. This approach has been applied
to map the pathways of nuclear transport factors
through the nuclear pore complex38. In this study, an
interaction matrix between a CFP-tagged importin
or exportin receptor with a panel of YFP-tagged
nucleoporins was established in living yeast cells.
Powerful genetic manipulation possibilities in yeast,
including chromosomal gene insertion to prevent
overexpression artifacts and the use of a genetic
background where function of the fusion protein
is required for viability, make this organism an
attractive vehicle for screening purposes. Indeed, two
previously reported interactions were detected in the
FRET screening, and a novel interaction was verified
by co-immunoprecipitation. The connectivity
uncovered in this study showed the existence of a
substantial overlap, with distinct contact points for
the two receptor translocation pathways. In these
experiments, FRET was quantified by a variant of
the donor/acceptor intensity ratio method (see Box 1),
where sensitized acceptor emission and total donor
emission are used. This method is only appropriate
under conditions where the stoichiometry of the
interacting proteins is fixed and known. The validity of
protein interactions that were identified in this FRET
screen therefore needs to be verified by rigorous
correction for non-FRET fluorescence contributions
(Box 1) or by using an alternative concentrationindependent FRET detection technique.
A number of studies have been designed to prove
the existence of expected protein interactions or
confirm them in the setting of the living cell – for
example, to exclude false-positive or indirect
interactions from biochemical approaches. This is
essentially a low-throughput proteomics approach.
Of course, detection of an unconditional – that is,
constitutive – interaction merely establishes the
architecture of a multi-molecular particle, even
though it provides mechanistic insight.
Constitutively oligomerized states were observed for
the Fas receptor39, where a qualitative sensitized
emission-based cell-sorting assay was complemented
by microscopic quantitation, and for the EGF
receptor40 as judged by quantitative donor bleaching
kinetics and FLIM (Box 2). Constitutive homooligomerization was also found for SNARE
complexes41, showing that significant amounts of
stable SNARE complexes exist before synaptic vesicle
fusion and that these complexes are assembled
during the preceding vesicle docking and priming
phase. Interactions between different proteins were
detected between the Four-and-a-half LIM-only
FHL2 and FHL3 proteins42 and the apoptosisregulating proteins Bcl2 and Bax in mitochondria43
and between Pit-1 and Ets-1 transcription factors in
the nucleus44. These studies rely on the detection of
sensitized emission, but only in the Bcl2/Bax and
SNARE studies was the sensitized emission corrected
http://tcb.trends.com
209
for direct fluorescence contributions (Box 1).
Physiological, biochemical and FRET imaging
evidence for the functional light-induced interaction
between phyB and cry2 proteins has been presented45.
Notably, in this study, FRET between the green- and
red-emitting spectral variants of GFP was used.
Sensitized emission measurements were reported
without correction for direct fluorescence
contributions, but, because these measurements
are susceptible to misinterpretation, the study was
complemented by acceptor photobleaching
measurements. The use of living cells in this study
implies that rapid acceptor-photobleaching was used
as, otherwise, labeled proteins would be able to relocate
before acquisition of the unquenched donor
fluorescence image. In general, care should be taken
when using DsRed as an acceptor because it suffers
from two properties that limit its use. First, the
native form of DsRed is a tetramer46. Even though the
stoichiometry in fusion constructs is possibly different,
specific residues have been identified that mediate
homo-association47. Second, DsRed matures through
a green-emitting intermediate46. This emission is
generally not observed owing to efficient FRET to
mature molecules in the multimeric complex. All
DsRed-based FRET observations – intensity- and
lifetime-based – will be complicated by this occurrence
of intrinsic FRET. The recent publication of the
crystal structure of DsRed47 will hopefully lead to
the availability of monomeric DsRed variants.
The association of the nsL-TP lipid carrier with
a number of peroxisomal fatty acid β-oxidation
enzymes has been monitored by FRET48, suggesting
that there is an enzyme complex in which nsL-TP
performs a substrate-presentation role. FRET
imaging between antibodies against the cholera
toxin A-subunit and directly labeled pentameric
B-subunits49 has implied that the Golgi serves as the
intracellular target and site of toxin disassembly.
Separate detection of the holo-toxin and its subunits
by FRET indicated that the A-subunit is redirected
to the plasma membrane, whereas the B-subunit
persists in the Golgi after disassembly.
The detection of an inducible interaction can offer
a higher level of information, giving more insight into
function. The proteolysis of PKC β1 induced by
phorbol myristate acetate (PMA) in the nucleus50
represents a signaling event that is part of the PKCmediated signaling pathway. Recruitment of Grb2 to
activated EGF receptors51 can be used as a measure of
EGF-induced receptor activation, analogous to the
determination of its phosphorylation status27,28. In
this case, the sensitized emission should be compared
with the unquenched donor fluorescence, which, with
the labeling strategy used, would be more easily
obtained by acceptor photobleaching.
Monitoring the change of a conditional interaction
can be used as a diagnostic read-out for (part of) the
activity of a pathway. Many papers have used
knowledge derived from analytical biochemistry
210
Review
TRENDS in Cell Biology Vol.11 No.5 May 2001
approaches to develop a cell-based FRET approach.
The Rac activity sensor29 described earlier is a good
example of such a measure of a conditional interaction
as a read-out. Additionally, the activation of nuclear
hormone-mediated transcription was detected by the
interaction of the nuclear hormone receptor with
co-activator proteins52. Similarly, the interaction of the
YFP-tagged co-activators peroxisome proliferatoractivated receptor binding protein (PBP) and steroid
receptor coactivator-1 (SCP-1) with CFP-tagged
retinoic acid receptor (RAR) and estrogen receptor
(ER) was used to demonstrate ligand-dependent
binding. Agonist and antagonist-dependent FRET
effects can be discerned, validating the use of this
assay to measure the regulation of the transcription
process and to screen for new regulators or ligands.
Outlook
Considering the central biological question of the
physiological relevance of observed chemical
reactions and the increasing collection of ‘traditional’
analytical assays being translated into cell-based
biosensor approaches, it is to be expected that the
functional imaging field has yet to reach its final
potential. The possibilities for the quantitative and
selective detection of biochemical states and reactions
of proteins by the various FRET techniques should
provide a wealth of data. FRET allows the detailed
workings of the protein machines underlying the
various cellular processes to be investigated, which
would otherwise be compromised by the preparative
steps unavoidable in biochemical approaches.
Furthermore, visualization of intracellular gradients
in enzymatic activities, such as phosphorylation and
GTPase activity, can now be related to morphogenetic
References
1 Tung, C.H. et al. (2000) In vivo imaging of
proteolytic enzyme activity using a
novel molecular reporter. Cancer Res.
60, 4953–4958
2 Thompson, R.B. et al. (2000) Fluorescence
microscopy of stimulated Zn(II) release from
organotypic cultures of mammalian hippocampus
using a carbonic anhydrase-base biosensor
system. J. Neurosci. Methods 96, 35–45
3 Oancea, E. et al. (1998) Green fluorescent protein
(GFP)-tagged cysteine-rich domains from protein
kinase C as fluorescent indicators for
diacylglycerol signaling in living cells. J. Cell Biol.
140, 485–498
4 Rizzo, M.A. et al. (2000) The recruitment of Raf-1
to membranes is mediated by direct interaction
with phosphatidic acid and is independent of
association with Ras. J. Biol. Chem. 275,
23911–23918
5 Teruel, M.N. and Meyer, T. (2000) Translocation
and reversible localization of signalling proteins:
a dynamic future for signal transduction. Cell
103, 181–184
6 Haugh, J.M. et al. (2000) Spatial sensing in
fibroblasts mediated by 3′ phosphoinositides.
J. Cell Biol. 151, 1269–1280
7 Oatey, P.B. et al. (1999) Confocal imaging of the
subcellular distribution of phosphatidylinositol
http://tcb.trends.com
8
9
10
11
12
13
14
15
processes, where the distribution of activity shapes
the cellular response53. Multiple approaches are at
the disposal of the researcher that utilize the different
spectroscopic consequences of FRET and will prove to
be instrumental in the challenge to uncover the
functional connectivity of the proteome. It will be
important to analyze the individual cell states defined
by multiple biochemical activities as measured by
fluorescence imaging and to correlate this with the
cellular phenotype or response.
Potential interactions can be predicted from
genomic sequences by using computational methods
that apply evolutionary criteria to proteins or
proteomes32. Consequently, the tools for the study of
pathways and processes of ever-increasing
magnitude and complexity are now becoming
available. At the same time, computer simulation of
interactions within functional units of the proteome
will become important because the increasingly
complex physiochemical network will exhibit
properties that are not intuitively apparent.
Ultimately, cells can be studied as nonlinear systems
where variations in responsiveness to the same
stimulus are expected within a cell population
dependent on the ‘internal protein setting’ of the
individual cell. A way of coping with this biological
complexity is to improve and extend on the current
technology. For instance, simultaneous detection of
multiple protein interactions is in principle possible
using FLIM. Thus multicomponent analysis of
protein processes in cells by fluorescence
microspectroscopy will complement classical
genetics, bioinformatics and protein biochemistry
in order to obtain the full picture of how information
is processed in a cell.
3,4,5-trisphosphate in insulin- and PDGFstimulated 3T3-L1 adipocytes. Biochem. J. 344,
511–518
Oancea, E. and Meyer, T. (1998) Protein kinase C
as a molecular machine for decoding calcium and
diacylglycerol signals. Cell 95, 307–318
Llopis, J. et al. (1998) Measurement for cytosolic,
mitochondrial, and Golgi pH in single living cells
with green fluorescent proteins. Proc. Natl. Acad.
Sci. U. S. A. 95, 6803–6808
Miesenböck, G. et al. (1998) Visualizing secretion
and synaptic transmission with pH-sensitive
green fluorescent proteins. Nature 394, 192–195
Dansen, T.B. et al. (2000) Peroxisomes in human
fibroblasts have a basic pH. Nat. Cell Biol. 2,
51–53
Matsuyama, S. et al. (2000) Changes in
intramitochondrial and cytosolic pH: early events
that modulate caspase activation during
apoptosis. Nat. Cell Biol. 2, 318–325
Wu, M.M. et al. (2000) Organelle pH studies using
targeted avidin and fluorescein–biotin. Chem.
Biol. 7, 197–209
Jayaraman, S. et al. (2000) Mechanism and
cellular applications of a green fluorescent
protein-based halide sensor. J. Biol. Chem. 275,
6047–6050
Baird, G.S. et al. (1999) Circular permutations
and receptor insertion within green fluorescent
16
17
18
19
20
21
22
23
proteins. Proc. Natl. Acad. Sci. U. S. A. 96,
11241–11246
Miyawaki, A. et al. (1997) Fluorescent indicators
for Ca2+ based on green fluorescent proteins and
calmodulin. Nature 388, 882–887
Miyawaki, A. et al. (1999) Dynamic and
quantitative Ca2+ measurements using improved
cameleons. Proc. Natl. Acad. Sci. U. S. A. 96,
2135–2140
Nakai, J. et al. (2001) A high signal-to-noise Ca2+
probe composed of a single green fluorescent
protein. Nat. Biotechnol. 19, 137–141
Adams, S.R. et al. (1991) Fluorescence ratio
imaging of cyclic AMP in single cells. Nature
349, 694–697
Zaccolo, M. et al. (1999) A genetically encoded,
fluorescent indicator for cyclic AMP in living cells.
Nat. Cell Biol. 2, 25–29
Doi, N. and Yanagawa, H. (1999) Design of generic
biosensors based on green fluorescent proteins
with allosteric sites by directed evolution. FEBS
Lett. 453, 305–307
Mahajan, N. et al. (1999) Novel mutant green
fluorescent protein protease substrates reveal the
activation of specific caspases during apoptosis.
Chem. Biol. 6, 401–407
Harpur, G.H. et al. (2001) Imaging FRET between
spectrally similar GFP molecules in single cells.
Nat. Biotechnol. 19, 167–169
Review
24 Tyas, L. et al. (2000) Rapid caspase-3 activation
during apoptosis revealed using fluorescenceresonance energy transfer. EMBO Reports 1,
266–270
25 Nagai, Y. et al. (2000) A fluorescent indicator for
visualizing cAMP-induced phosphorylation
in vivo. Nat. Biotechnol. 18, 313–316
26 Ng, T. et al. (1999) Imaging protein kinase Cα
activation in cells. Science 283, 2085–2089
27 Verveer, P.J. et al. (2000) Quantitative imaging of
lateral ErbB1 receptor signal propagation in the
plasma membrane. Science 290, 1567–1570
28 Wouters, F.S. and Bastiaens, P.I.H. (1999)
Fluorescence lifetime imaging of receptor tyrosine
kinase activity in cells. Curr. Biol. 9, 1127–1130
29 Kraynov, V.S. et al. (2000) Localized Rac
activation dynamics visualized in living cells.
Science 290, 333–337
30 Venter, J.C. et al. (2001) The sequence of the
human genome. Science 291, 1304–1351
31 International Human Genome Sequencing
Consortium (2001) Initial sequencing and analysis
of the human genome. Nature 409, 860–921
32 Walhout, A.J.M. and Vidal, M. (2001) Protein
interaction maps for model organisms. Nat. Rev.
Mol. Cell. Biol. 2, 55–62
33 Bastiaens, P.I.H. and Pepperkok, R. (2000)
Observing proteins in their natural habitat: the
living cell. Trends Biochem. Sci. 25, 631–637
34 Bastiaens, P.I.H. and Jovin, T.M. (1998) FRET
microscopy. In Cell Biology: A Laboratory Handbook
(Vol. 3) (Celis, J.E., ed.), pp. 136–146, Academic Press
35 Gordon, G.W. et al. (1998) Quantitative
fluorescence energy transfer measurements using
fluorescence microscopy. Biophys. J. 74, 2702–2713
36 Jovin, T.M. and Arndt-Jovin, D.J. (1989) FRET
microscopy: digital imaging of fluorescence
resonance energy transfer. In Cell Structure and
Function by Microspectrofluorometry (Kohen, E. and
Hirschberg, J.G., eds), pp. 99–115, Academic Press
37 Nagy, P. et al. (1998) Intensity-based energy
transfer measurements in digital imaging
microscopy. Eur. Biophys. J. 27, 377–389
38 Damelin, M. and Silver, P.A. (2000) Mapping
interactions between nuclear transport factors in
living cells reveals pathways through nuclear
pore complex. Mol. Cell 5, 133–140
TRENDS in Cell Biology Vol.11 No.5 May 2001
39 Siegel, R.M. et al. (2000) Fas preassociation
required for apoptosis signaling and dominant
inhibition by pathogenic mutations. Science 288,
2354–2357
40 Gadella, T.W.J., Jr and Jovin, T.M. (1995)
Oligomerization of epidermal growth-factor
receptors on A431 cells studied by time-resolved
fluorescence imaging microscopy – a
stereochemical model for tyrosine kinase receptor
activation. J. Cell Biol. 129, 1543–1558
41 Xia, Z. et al. (2001) Stable SNARE complex prior
to evoked synaptic vesicle fusion revealed by
fluorescence resonance energy transfer. J. Biol.
Chem. 276, 1766–1771
42 Li, H.Y. et al. (2001) Protein–protein interaction of
FHL3 with FHL2 and visualization of their
interaction by green fluorescent proteins (GFP)
two-fusion fluorescence resonance energy
transfer (FRET). J. Cell. Biochem. 80, 293–303
43 Mahajan, N. et al. (1998) Bcl-2 and Bax
interactions in mitochondria probed with green
fluorescent protein and fluorescence resonance
energy transfer. Nat. Biotechnol. 16, 547–552
44 Day, R.N. (1998) Visualization of Pit-1
transcription factor interactions in the living
cell nucleus by fluorescence resonance energy
transfer microscopy. Mol. Endocrinol. 12,
1410–1419
45 Más, P. et al. (2000) Functional interaction of
phytochrome B and cryptochrome 2. Nature 408,
207–211
46 Baird, G.S. et al. (2000) Biochemistry,
mutagenesis, and oligomerization of DsRed, a red
fluorescent protein from coral. Proc. Natl. Acad.
Sci. U. S. A. 97, 11984–11989
47 Wall, M.A. et al. (2000) The structural basis for
red fluorescence in the tetrameric GFP homolog
DsRed. Nat. Struct. Biol. 7, 1133–1138
48 Wouters, F.S. et al. (1998) FRET microscopy
demonstrates molecular association of nonspecific lipid transfer protein (nsL-TP) with fatty
acid oxidation enzymes in peroxisomes. EMBO J.
17, 7179–7189
49 Bastiaens, P.I.H. et al. (1996) Imaging the
intracellular trafficking and state of the AB5
quaternary structure of cholera toxin. EMBO J.
15, 4246–4253
211
50 Bastiaens, P.I.H. and Jovin, T.M. (1996)
Microspectroscopic imaging tracks the
intracellular processing of a signal
transduction protein: fluorescent-labeled
protein kinase C βI. Proc. Natl. Acad. Sci.
U. S. A. 93, 8407–8412
51 Sorkin, A. et al. (2000) Interaction of EGF receptor
and Grb2 in living cells visualized by fluorescence
resonance energy transfer (FRET) microscopy.
Curr. Biol. 10, 1395–1398
52 Llopis, J. et al. (2000) Ligand-dependent
interactions of coactivators steroid receptor
coactivator-1 and peroxisome proliferatoractivated receptor binding protein with nuclear
hormone receptors can be imaged in live cells and
are required for transcription. Proc. Natl. Acad.
Sci. U. S. A. 97, 4363–4368
53 Carazo-Salas, R.E. et al. (2001) Ran-GTP
coordinates regulation of microtubule nucleation
and dynamics during mitotic-spindle assembly.
Nat. Cell Biol. 3, 228–234
54 Bastiaens, P.I.H. and Squire, A. (1999)
Fluorescence lifetime imaging microscopy: spatial
resolution of biochemical processes in the cell.
Trends Cell Biol. 9, 48–52
55 Jovin, T.M. and Arndt-Jovin, D.J. (1989)
Luminescence digital imaging microscopy.
Annu. Rev. Biophys. Chem. 18, 271–308
56 Sokol, D.L. et al. (1998) Real time detection of
DNA–RNA hybridization in living cells. Proc.
Natl. Acad. Sci. U. S. A. 95, 11538–11543
57 Gadella, T.W.J., Jr et al. (1994) Visualization of
lipid–receptor interactions on single cells by timeresolved imaging fluorescence microscopy.
J. Fluoresc. 4, 295–298
58 Schmid, J.A. et al. (2000) Oligomerization of the
human serotonin transporter and of the rat GABA
transporter 1 visualized by fluorescence energy
transfer microscopy in living cells. J. Biol. Chem.
276, 3805–3810
59 Gadella, T.W.J., Jr et al. (1999) GFP-based FRET
microscopy in living plant cells. Trends Plant Sci.
4, 287–291
60 Murata, S. et al. (2000) Fluorescence lifetime
imaging of nuclear DNA: effect of fluorescence
resonance energy transfer. Cytometry 41,
178–185
GFP applications
This Review by Fred Wouters, Peter Verveer and Philippe Bastiaens is the first in a series of articles on the uses
of GFP in a variety of applications and technologies. The next issue of Trends in Cell Biology features a review
by Andrew Belmont, ‘Visualizing chromosome dynamics with GFP’. Look out for other articles in the series to
follow in subsequent issues.
GFP in Motion 2
All Trends in Cell Biology subscribers will receive a free copy of the ‘GFP in Motion 2’ CD with this issue of the
journal. The CD contains a selection of movies of live-cell imaging using GFP-labeled proteins. It has been
compiled by Beat Ludin and Andrew Matus and is sponsored by BioImage A/S (with APBiotech), Chroma
Technology Corporation, Clontech, Life Imaging Services, Perkin Elmer and Universal Imaging Corporation. If
you are not a subscriber to Trends in Cell Biology and would like to purchase a copy of the CD, please contact:
[email protected]
http://tcb.trends.com