Cis-Trans Photoisomerization of Fluorescent-Protein Chromophores
Valerio Voliani,*,†Ranieri Bizzarri,†,‡Riccardo Nifosı `,†,‡Stefania Abbruzzetti,§Elena Grandi,§
Cristiano Viappiani,§and Fabio Beltram†,‡
Scuola Normale Superiore, Italian Institute of Technology, Piazza dei CaValieri 7, I-56126 Pisa, Italy, NEST,
Scuola Normale Superiore and CNR-INFM, Via della Faggiola 19, I-56126 Pisa, Italy, and NEST CNR-INFM,
Dipartimento di Fisica, UniVersita ` di Parma, Viale G. P. Usberti 7A, I-43100 Parma, Italy
ReceiVed: March 19, 2008; ReVised Manuscript ReceiVed: June 5, 2008
Photochromic variants of fluorescent proteins are opening the way to a number of opportunities for high-
sensitivity regioselective studies in the cellular environment and may even lead to applications in information
and communication technology. Yet, the detailed photophysical processes at the basis of photoswitching
have not been fully clarified. In this paper, we used synthetic FP chromophores to clarify the photophysical
processes associated with the photochromic behavior. In particular, we investigated the spectral modification
of synthetic chromophore analogues of wild-type green fluorescent protein (GFP), Y66F GFP (BFPF), and
Y66W GFP (CFP) upon irradiation in solutions of different polarities. We found that the cis-trans
photoisomerization mechanism can be induced in all the chromophores. The structural assignments were
carried out both by NMR measurements and DFT calculations. Remarkably, we determined for the first time
the spectra of neutral trans isomers in different solvents. Finally, we calculated the photoconversion quantum
yields by absorption measurements under continuous illumination at different times and by a nanosecond
laser-flash photolysis method. Our results indicate that cis-trans photoisomerization is a general mechanism
of FP chromophores whose efficiency is modulated by the detailed mutant-specific protein environment.
Fluorescent proteins (FPs) are arguably the most important
fluorescent probes in molecular biology.1,2This largely stems
from the fact that their primary sequence can provide all the
information needed for fluorescence expression. As a conse-
quence fluorescent fusion constructs consisting of a given protein
of interest and an FP can be rather straightforwardly expressed
in living cells.1The green fluorescent protein (GFP) from
jellyfish Aquorea Victoria was the first such protein to be
discovered and cloned in eukariotes. GFP is the ancestor of the
most widespread FP family for in ViVo detection.2
The chromophore of GFP is a p-hydroxybenzylidene imida-
zolinone that is generated by the subsequential cyclyzation/
oxidation/dehydration of the Ser65-Tyr66-Gly67motif, on the
aftermath of the primary sequence folding into a ?-barrel tertiary
structure.3Successful engineering of both absorption and
emission spectra of GFPs was achieved by substituting selected
aminoacids in the primary sequence.4Blue-shifted variants of
parent GFP were obtained by replacing Tyr66with triptophan
(cyan fluorescent protein, CFP), histidine (blue fluorescent
protein, BFP), or phenylalanine (Y66F GFP, hereafter denoted
as BFPF).5The reduced electron conjugation in the chromophore
when an indolyl, imidazolyl or phenyl ring replaces the phenol
of GFP accounts for the blue-shifted fluorescence of these
mutants. Despite lower quantum yield and photostability with
respect to GFP, these mutants found widespread use in cellular
multitracking studies,8colocalization analyses,9and as donors
in fluorescence resonance energy transfer (FRET) in ViVo
measurements with green10or yellow FPs as acceptors.6
Recently, the use of photoswitchable proteins for in ViVo
studies has attracted wide interest.12–14Irreversibly photoswit-
chable variants from dark to bright states7or from one color to
another8expanded noticeably the range of applications of
conventional FPs.17–19In 2003 we suggested that the dark state
of photoswitching GFP mutants, including one developed by
us,9carried a trans chromophore configuration (in contrast to
the normal cis configuration) achieved through cis-trans
photoisomerization around the exocyclic double bond.10Our
argument was based on theoretical calculations and indirect
spectral evidence, and referred also to photophysical studies
carried out on models of the GFP chromophore.22–26Such a
photoinduced cis-trans isomerization was recently demon-
strated by X-ray analysis for a FP variant from Anemonia
Sulcata,11for a mutant derived from ClaVularia,12and for the
popular Dronpa variant from Pectidinaee.13In spite of a poor
homology in primary sequence, these proteins display ?-barrel
tertiary structures and highly conjugated imidazolinone chro-
mophores resembling those of GFP. Cis-trans photoisomer-
ization is presumably a general feature of this large class of
chromophores. Indeed, photoinduced cis-trans isomerization
was verified experimentally at least for HBDI, a synthetic
analogue of the wtGFP chromophore.10The optical and pho-
tochromic properties of the trans state, however, were not
It is worth noting that the FP photochromism is a complex
molecular process for which a loss of structural rigidity of the
chromophore in the protein barrel is the determining factor.12,14,15
Additionally, for some proteins a proton transfer step seems to
be related to the structural change associated with photo-
chromism. The photophysical properties of the cis-trans
isomerization must be interpreted in such a perspective. Indeed,
the most relevant structural feature to allow the cis-trans
isomerization is the possibility of a concerted torsion around
* To whom correspondence should be addressed. E-mail: email@example.com.
†Scuola Normale Superiore, Italian Institute of Technology.
‡NEST, Scuola Normale Superiore and CNR-INFM.
§NEST CNR-INFM, Dipartimento di Fisica, Universita ` di Parma.
J. Phys. Chem. B 2008, 112, 10714–10722
10.1021/jp802419h CCC: $40.75
2008 American Chemical Society
Published on Web 07/31/2008
the bonds connecting the two rings, which results in a space-
saving isomerization mechanism termed hula-twist.16We believe
that detailed understanding of the FP photochromism requires
a comparison between the photoisomerization efficiency of the
chromophore alone and when embedded in the protein barrel.
On account of our long-standing interest in the photophysical
properties of FPs,4,9,10,17we set off to investigate how the
chromophore structure influences the cis-trans photoisomer-
ization and the optical differences between the two states. In
this paper we report on our studies regarding the photoisomer-
ization of synthetic analogues of the GFP, CFP, and BFPF
chromophores in solutions of different polarities. We shall show
by optical and NMR measurements and by computer modeling,
that the same photoinduced, reversible cis-trans mechanism
is shared by all these chromophores. We shall argue that this
result supports the idea that reversible cis-trans isomerization
is a general photoprocess that does or does not take place in a
given folded mutant on account of the detailed local chro-
mophore environment. In addition, we shall report for the first
time the absorption spectra of the trans chromophores and
quantify the efficiency of the photoisomerization process. These
data will be compared with the reported photoconversion
efficiencies of some FPs, in view of estimating the role of protein
matrix rigidity on the emergence of the photochromic protein
Materials and Methods
Chromophore Synthesis. The chromophores were synthe-
sized according to the established procedure, involving the
preliminary synthesis of azalactone compounds followed by their
aminolysis with methylamine.18The details of this procedure
are reported as Supporting Information.
Absorption Spectroscopy. Ultraviolet-visible (UV-vis)
spectra were measured on a Jasco 550 spectrophotometer (Jasco,
Tokyo, Japan) supplied with a Jasco ETC-505T thermostat.
NMR Spectrometry. NMR spectra were collected by Varian
Unity 300 spectrometer (Varian, Palo Alto, CA, USA) at room
temperature on 5-10% solutions in DMSO-d6, Dioxane-d8, or
CDCl3.1H NMR spectra were collected at 300 MHz with the
follow experimental conditions: 32768 points, 3 kHz of spectral
width, 30° impulse, 2 s acquisition time, and 1024 transients.
13C NMR spectra were collected at 75 MHz with the following
experimental conditions: 65536 points, 15 kHz of spectral width,
70° impulse, 0.8 s acquisition time, and (1-2) × 105transients.
Photoconversion Experiments. Solutions of chromophore
or protein in water at pH 6.5-7 or organic solvents were placed
in 1500 µL quartz cells with a 1 cm path length for absorption
spectroscopy experiments; solutions with concentration 1 to 10
mM were placed into 700 µL quartz cells with a 0.5 cm path
length for NMR spectrometry experiments. Photoconversion was
obtained by irradiating at 360, 406, or 458 nm by using either
an Ar-Kr laser (Beamlok 2060, Spectra Physics, Mountain
View, CA).Temperature was controlled by a Peltier thermostat
(Varian, Palo Alto, CA) and continuous stirring was ensured.
Nanosecond Laser Flash Photolysis. Laser flash photolysis
experiments were performed with a previously described
setup.19,20Photoexcitation was achieved by the third harmonic
(355 nm) of a nanosecond Nd:YAG laser, and detection of the
absorbance change was performed with the cw output of a Xe
arc lamp and a Si photodiode as a detector (see Supporting
Information for further details).
Theoretical Calculations. We evaluated the excitation ener-
gies using the time-dependent extension of DFT in the linear
perturbation theory,17with B3LYP exchange and correlation
functional21and 6-31+G* basis set, on structures optimized with
B3LYP/6-31G*. We included implicit solvent effects through
the polarizable continuum method (PCM).10The performance
of these techniques for fluorescent protein chromophores were
calculated by the GIAO method22(B3LYP/6-311G**), on
structures optimized at the same level of theory. Since the two
pairs of opposite carbons on the benzene ring are equivalent
on the time scale of the NMR experiment, the average chemical-
shift value was taken for comparison with experiment. Gaussian
0323was used for the calculations.
13C NMR chemical shifts were
Synthesis of GFP Chromophores. The synthesis of model
chromophores was accomplished by means of an established
synthetic procedure involving two steps (Scheme 1, reactions
i-ii). Note that, unless sterically hindered reactants are used in
the first step, this procedure is known to yield (Z)-isomers.24,25
The higher thermodynamic stability of the (Z)-isomers of our
chromophores was indeed predicted by theoretical calculations
(see below). For clarity of discussion, we will refer to the (Z)-
and (E)-isomers as cis and trans, respectively, in agreement with
the practical nomenclature commonly adopted for the chro-
mophores of FPs. The prefixes c and t will be added to the
SCHEME 1: Chemical Synthesis (i-ii) and Reversible Cis-Trans Photoisomerization of Chromophores (iii-iv)a
aKey: (i) CH3COONa, (CH3COO)2, 110-130 °C; (ii) CH3NH2(aq), ethanol, Na2CO3, 120-130 °C; (iii) hν; (iv) hν and/or ∆. The stereochemically-
relevant dihedral angles ? and τ are shown: cis and trans configurations correspond to τ ) 0° and τ ) 180°, respectively.
J. Phys. Chem. B, Vol. 112, No. 34, 2008 10715
chromophore name to identify its stereoisomeric form (e.g.,
cGFP: cis-GFP chromophore). Furthemore, we shall adopt the
protein data bank (PDB) nomenclature of FPs to denote specific
atoms in the chromophores.26
Thermodynamic Stability As a Function of Chromophore
Stereochemistry. Given the planarity of the imidazolinone and
aromatic rings and the symmetry of the methyl groups, the
stereochemical configuration of the chromophores is mainly
defined by the ? and τ dihedral angles, which refer to the torsion
of the exocyclic C-C or the CdC bonds respectively (Scheme
1). Cis and trans isomers correspond to τ ≈ 0° and τ ≈ 180°,
respectively. To assess the relative thermodynamic stability of
distinguishable stereochemical configurations, we carried out
DFT-B3LYP energy calculations on each chromophore by
assuming either gas phase as medium or an implicit solvent of
high polarity (methanol or water). Note that solvation free-
energy terms (PCM method) are always taken into account when
the implicit solvent is added and the vibrational entropy was
found to be nearly unaffected by stereochemical changes (∼0.1
kcal/mol). Thus, the obtained energy values can be identified
with stereochemical free-energies and are denoted by G(?,τ)
cGFP, tGFP, cBFPF, and tBFPF are all characterized by a
symmetrical phenol /phenyl ring, leading to a 2-fold symmetry
for rotation around ?. We found that in all four structures the
minimum-energy configuration is nearly planarsi.e. (? ≈ 0°,
τ ≈ 0°) for cis isomers and (? ≈ 0°, τ ≈ 180°) for trans
isomerssproviding extended electronic conjugation while avoid-
ing steric hindrance. Note that the finite tolerance of the
optimization procedure is likely at basis of the slight, i.e. <2°,
deviations from the perfect all-planar configuration (? ) 0°, τ
) 0°) found in these cases.
The calculated values indicate that cGFP f tGFP isomer-
ization is associated with ∆Gi ) Gtrans - Gcis ) 2.41-2.47
kcal/mol (Table 1), in good agreement with data reported
previously (∆Gi ) 2.10 kcal/mol).27cBFPFftBFPF isomer-
ization is associated with an even larger free-energy change
(Table 1). Hence, about 98-99% of molecules of these
chromophores must be in the cis state at room temperature.
Pleasantly, the1H NMR spectra of cGFP and cBFPF showed
resonances attributable to a single stereoisomer (data not show).
The asymmetric indolyl ring of CFP chromophore disrupts
the 2-fold symmetry for rotation around ?. Accordingly, the
theoretical calculation showed that cCFP and tCFP are both
characterized by two energy minima, each one corresponding
to a nearly all-planar configuration, which were denoted as
c1CFP/c2CFP and t1CFP/t2CFP, respectively (Scheme 2). We
found that t1CFP is more stable than c2CFP in all media (Table
1). On the other hand, t2CFP has a very high relative G(?,τ),
likely owing to the unfavorable steric relationship between the
indolyl and the imidazolinone rings (Scheme 2), which leads
to the highest stereochemical distorsions from the all-planar
configuration. From the free-energy values in methanol we
calculated that at room temperature 73.5%, 12.5%, 14%, and
<10-3% of the chromophore are in the c1CFP, c2CFP, t1CFP,
and t2CFP configurations, respectively. The1H NMR spectrum
of cCFP in deuterated methanol at room temperature strongly
substantiated these results, as it evidenced that around 12% of
the chromophore is in a different stereoisomeric form, lately
identified with the trans form (see below).
Photoconversion of Chromophores. The absorption spectra
of the neutral cis chromophores are characterized by two main
bands: one, more intense, peaked in the 340-400 nm range
(low-energy band), and one, less intense, peaked between 260
and 300 nm (high energy band) (Figure 1a, black trace). On
the basis of the TD-DFT B3LYP methods, we assigned the two
bands to the HOMO->LUMO and to the HOMO-3->LUMO
transitions, respectively. Owing to the ionization of the phenolic
group and the formation of a highly conjugated anionic
imidazolinone,28the low-energy band of cGFP in alkaline water
(pH 12) is red-shifted of about 56 nm (not shown).
In the typical photoconversion experiment, a cis chromophore
was irradiated close to the low-energy band maximum for times
ranging from few seconds to 10 min by means of 1-5 mW
continuous laser light. The irradiation experiments were carried
out in solvents of different characteristics, namely methanol
TABLE 1: Calculated Stereochemical Free-Energy G(?,τ) and Dipole Moments µ and Measured Photoisomerization Quantum
µ (D) G(?,τ)ametc(kcal/mol)
0.0 (180°) 0.03.650e
cGFP00 0.21 ( 0.01g
0.21 ( 0.05h
0.18 ( 0.02g
0.10 ( 0.0 + 2g
0.08 ( 0.05h
0.2 ( 0.1g
0.24 ( 0.13h
0.9 ( 0.3g
aG(0,0) was set to 0 as reference for each chromophore.
fPhotoisomerization quantum yield.gfrom continuous irradiation experiments.hFrom flash photolysis experiments.
bgp: gas phase.
enc: not calculated.
SCHEME 2: Stereochemical Configurations of CFP
Chromophore Corresponding to the Energy Minima
Calculated by the DFT-B3LYP Method
J. Phys. Chem. B, Vol. 112, No. 34, 2008
Voliani et al.
(protic solvent, ?r) 33), acetonitrile (aprotic solvent, ?r) 36.6),
dioxane (aprotic solvent, ?r) 2.2), and, for cGFP, also in neutral
or alkaline water (protic solvent, ?r) 80.1). Indeed, the internal
environment of the proteins is typically different from pure water
Figure 1. Photochromic characteristics of chromophores. They are exemplified by the optical behavior of cBFPF upon irradiation. (a, left panel)
Continuous irradiation at 360 nm leads to the time-dependent decrease and red-shift of the absorption spectrum due to the formation of a photoproduct
(identified with tBFPF by NMR, see text); (a, right panel) constant irradiation of photoconverted cBFPF at 406 nm, leads to the reversion of the
absorption spectrum. b) The photoconversion rate of cBFPF at 360 nm is dependent on irradiation light power, but the same photostationary state
is always achieved; photoconversion curves are fitted to eq S6 (Supporting Information) to yield the photoisomerization quantum yields (see text).
(c) Spontaneous recovery to the native state of the photoproducts of cGFP (green line, 360 nm-absorbance) and cCFP (cyan line, 406 nm-absorbance)
in methanol at 20 °C. (d) Single high-energy pulse irradiation at 355 nm of cGFP in methanol (OD ) 0.6) leads to a fast and irreversible absorption
change (<5 ns) detected at 420 nm (green line); methylene blue in water (OD ) 0.67) irradiated at the same wavelength shows a fast conversion
to triplet state followed by slower (a few microseconds) decay to the ground singlet state because of quenching by molecular oxygen (red line).
(inset) By using methylene blue as an actinometer (see text), the comparative method32yields ?cas a function of laser-pulse energy: extrapolation
to zero energy gives the actual ?cvalue.
J. Phys. Chem. B, Vol. 112, No. 34, 2008 10717
and in some cases can be strongly hydrophobic. We selected
these solvents to explore a large interval of polarity, as expressed
by their dielectric constants, and the capability to establish or
not hydrogen bonds with the chromophore molecules. Note that
methanol and acetonitrile have approximately the same polarity,
although only the former solvent is able to partecipate to
H-bonding with the chromophores.
In all cases, steady-state photoexcitation led to the decrease
in intensity and red-shift of the low-energy band (Figure 1a,
left panel); conversely, the high energy band was enhanced upon
irradiation. Remarkably, the presence of two sharp isosbestic
points indicated the direct conversion of the native chromophore
in a photoproduct with different spectral properties. As expected,
the rate of photoconversion increased with the excitation power
(Figure 1b). For each chromophore the same constant spectrum
was asymptotically reached regardless of the irradiation power,
thus suggesting that a common photostationary state, character-
ized by equal forward and reverse light-induced photoconversion
rates, had been achieved (Figure 1b).
An optical state close to the native one was restored upon
irradiation of the photoproduct at longer wavelengths (>40 nm
from the maximum of the cis chromophore, Figure 1a, right
panel). Again, the process rate was found to increase with the
excitation source power. Hence, we concluded that these
compounds displayed reversible photochromism. The nature of
the solvent did not change the observed photochromic pattern
(data not shown).
Interestingly, the photoproducts of cGFP and cCFP were
found to decay back to the native state when kept at room
temperature and in the dark (Figure 1c). This process was found
to be much faster (minutes/few hours) in protic solvents than
in aprotic ones (some hours/days). In contrast, the photoproduct
of cBFPF remained stable for weeks in all solvents. In methanol
at 20 °C, the photoproduct of cGFP relaxed back thermally to
their native states following a nonmonoexponential kinetics,
whereas the data of cCFP should be fitted to a monoexponential
law, with τ2 ) 195 min. To unveil any pH effect on the
spontaneous recovery, we also separately addressed the neutral
and anionic forms of cGFP. In both cases we found nonmo-
noexponential decay kinetics much faster than recorded in
methanol (not shown), but pH did not change significantly the
average time-constant when data were fitted to a biexponential
law (pH 6.1, 〈τ〉 ) 8.1 min; pH 10.2, 〈τ〉 ) 8.3 min).
1H NMR Analysis of the Photoproduct. We compared the
1H NMR resonances of each photoproduct with those of the
parent cis isomer, in view of assessing the structural changes
associated with the photoconversion. Deuterated dioxane was
chosen as solvent owing to the slow thermal recovery of the
photoproducts of cGFP and cCFP to the native state.
The most interesting resonance pattern changes were detected
in the 6.5-9.0 ppm aromatic region. The photoproduct of cGFP
was found to display a downfield shift of the HD1, HD2, and
HB2 proton resonances, whereas HE1 and HE2 were only
slightly affected (Figure 2a). Remarkably, similar spectral
changes were described for irradiated cGFP in deuterium
oxide,27deuterated DMSO,29and deuterated methanol.29Con-
sistently with these authors, we can identify the photoproduct
with the trans isomer tGFP upon analogy with the1H NMR
spectra of structurally related azlactone moieties.30Similar
resonance shifts were observed for the photoproducts of cBFPF
and cCFP (Figure 2a), which were accordingly identified with
tBFPF and tCFP. From peak integration, we calculated that
tGFP, tBFPF, and tCFP represent the 57%, 71%, and 76% of
the moles in the photoirradiated mixtures, respectively.
13C NMR Structural Identification of the Photoproduct
of cBFPF. To further confirm the identification of the photo-
products with the trans isomers, we carried out a thorough13C
NMR analysis on cBFPF before and after irradiation. This
chromophore was selected for its simpler structure and because
it did not show thermal recovery to the native state. Deuterated
dioxane was chosen as solvent owing to its aprotic and apolar
nature that prevents large modifications of the NMR pattern
provoked by strong solvent-solute interactions.
The fully decoupled13C NMR spectrum of the photoproduct
resembled closely that of cBFPF, although it was on average
shifted 6-8 ppm downfield. Under the assumption of cis-trans
isomerization, we carried out GIAO/DFT calculations of the
chemical shifts of both isomers and we compared them with
those retrieved experimentally. The13C NMR pattern of cBFPF
is in the expected antiparallel correlation (slope ) 1.018) with
the DFT chemical shifts of the cis isomer (Figure 2b, blue open
circles). Remarkably, such a correlation was verified also
between the signals attributable to the photoproduct and those
calculated for the trans isomer (Figure 2b, red open circles).
The analysis of H-C spin-spin coupling effect provided the
final confirmation to our structural identification. Indeed, it is
widely reported that the vicinal coupling
13C-CdC-H motif is stereospecific, as3JC,Htrans>3JC,Hcis.31
Such a difference is strongly pronounced when the carbons are
unaffected by γ-effect (such as quaternary carbons).31On
account of this property, we focused our attention on the
quaternary carbonylic C2 (signal at 170.8 ppm in cBFPF) and
we determined its vicinal coupling with the vinyl proton HB2
(3JC2,HB2) both in cBFPF and its photoproduct (Figure 2c). To
avoid interference from other protons at same bonding distance
we collected the13C NMR spectra by selectively decoupling
the methyl protons HA3. Our data showed that native cBFPF
is characterized by3JC2,HB2) 4.40 Hz while the photoconverted
mixture has3JC2,HB2) 8.80 Hz (Figure 2c). In their pioneering
work on azlactone stereoisomerism, Prokof’ev et al. found
3JC2,HB2cis) 5.5 Hz and3JC2,HB2trans) 12.5 Hz for the azlactone
precursor of cBFPF.31Taking into account both the stereospe-
cific dependence of3JC,Hin the13C-CdC-H structural motif
and the close structural homology between the chromophore
and its parent azlactone, these data provided an unequivocal
identification of the native and photoconverted forms with
cBFPF and tBFPF.
Absorption Spectra of Trans Chromophores and Photoi-
somerization Quantum Yields. Irradiated chromophore solu-
tions in dioxane-d6that had been previously analyzed by1H
NMR were used to prepare solutions of known cis-trans
composition in all the other solvents. From the absorption
spectra of these solutions and those of the pure cis forms we
were able to determine the absorption spectral details for pure
trans isomers in each solvent (Figure 3a-c, blue traces, and
Figures S1-S3 in the Supporting Information). Remarkably,
the low-energy band of the trans chromophores was red-shifted
of 5-15 nm compared to that of the cis chromophores,
depending on the molecular structure and the solvent nature.
Conversely, the high-energy band increased in intensity upon
isomerization. These effects seem explainable on account of
oscillator-strength transfer to the excitation involving the
molecular orbital localized on the phenolic ring (HOMO-3),
possibly due to decreased conjugation over the entire chro-
mophore in the trans configuration.
The theoretical description of photoisomerization (Supporting
Information) shows the direct relationship between the spectral
change upon irradiation, the optical characteristics of the pure
3JC,H in the
J. Phys. Chem. B, Vol. 112, No. 34, 2008
Voliani et al.
isomers, and the photoisomerization quantum yields. The latter
quantities, denoted here as Φc(cis f trans) and Φt(trans f cis),
represent the most relevant parameters of the photochromic
behavior, as they measure the efficiency of the nonradiative
decay channel(s) that lead to the stereochemical change upon
light absorption. For moderate optical densities eq S6 (Sup-
porting Information) shows that the solution absorbance follows
a first order kinetics upon irradiation, whose amplitude and time-
constant are connected to the light power, the solution volume,
the extinction coefficients of the two isomers, the thermal
recovery rate-constant, and (Φc, Φt). Accordingly, we fitted the
time decays of the low-energy band during the irradiation
experiments carried out in methanol to eq S6 (Figure 1b), and,
given the knowledge of the other parameters, we calculated the
photoisomerization quantum yields for each chromophore in
neutral form (Table 1). According to the determined Φ values,
the photoisomerization processes are very effective. Remarkably
about 100% of the absorbed photons by the trans isomers lead
to the stereochemical change in the BFPF and CFP chro-
mophores. Note that photoisomerization quantum yields of
reported photochromic proteins (GFP chromophore) range
between 10-6and 10-3.10,14
The Φcvalues were also determined by nanosecond laser flash
photolysis experiments. Here, the chromophores were exposed
to a single shot of a pulsed 355 nm laser light while monitoring
the absorbance changes at 420 nm (cGFP, Figure 1d). Different
detection wavelengths were used for cCFP (455 nm) and for
cBFPF (390 nm). All chromophores showed an irreversible (on
the time scale of the experiment) change in absorbance (∆A)
taking place below the time resolution of the apparatus (5 ns).
∆A was found to increase linearly with laser pulse energy at
low values and saturates above ≈ 20 mJ per pulse, as expected
Figure 2. (a)1H NMR spectra (aromatic region only) of photoconverted cGFP (green line), cBFPF (blue line), and cCFP (red line); molecular
structures and peak attribution according to PDB nomenclature is included; resonances of the cis and trans chromophores are denoted by (c)
and (t), respectively. (b) Comparison between DFT-calculated and experimentally measured13C NMR chemical shifts for cBFPF and tBFPF:
excellent correlation was found for all resonances. (c)13C NMR spectra of native and photoconverted cBFPF while adopting selective decoupling
between the methyl protons HA3 and C2: the detected3JC2,HB2values in the photoproduct correspond to those expected for the cis and trans
J. Phys. Chem. B, Vol. 112, No. 34, 2008 10719
for a reversible photoconversion (not shown). Methylene blue
was used as an actinometer (Supporting Information) and
allowed for the calculation of Φcvalues at different laser powers
by means of the comparative method (Supporting Information).32
Extrapolation to zero laser pulse energy (inset to Figure 1d)
yielded the actual Φcvalues (Table 1). Noticeably we found
Φcvalues in excellent agreement with those obtained by the
continuous irradiation experiment. Furthermore, the obtained
Φc for cGFP are in good agreement with the value recently
reported by Yang et al. for the same chromophore.29
In recent years, new scientific evidence pointed out that the
photophysics of some GFP mutants is not merely restricted to
fluorescence emission but comprises other remarkable properties
such as photoactivation and reversible photoswitching. The
essential role of the detailed folded structure of the protein in
determining such phenomena is generally acknowledged. Indeed,
the presence of the ?-barrel fold in GFP is necessary for
fluorescence, since it provides the structural rigidity and the
inaccessibility to external quenchers that are prerequisites for
the chromophore to emit. The complex photophysics of GFP
mutants, however, is ultimately linked to the peculiarities of
the chromophore that emerge also when the latter is buried
within the tridimensional fold. Notably, the (Z)-(E) diaster-
omerization of the chromophore, which is commonly referred
to as cis-trans isomerization, was proposed to be at the basis
of fluorescence photoswitching for some GFP mutants10and
was later demonstrated as an effective mechanism of fluores-
cence modulation in FPs from Anemonia Sulcata11and ClaVu-
laria.12From these studies, however, it is still unclear whether
the photoinduced cis-trans isomerization represents a general
feature of FP chromophores. The scope of this work was to
verify the validity of this hypothesis for three synthetic
analogues of GFP chromophores, namely wild type Y66 (GFP),
Y66F (BFPF), and Y66W (CFP).
Our investigation started by carrying out DFT-B3LYP
theoretical calculations to evaluate the thermodynamic stability
of these molecules as function of the stereochemical configura-
tions (defined by the ? and τ dihedral angles, Scheme 1). In
this analysis we considered only the neutral state of the ionizable
GFP chromophore, in order to allow for a reliable comparison
with the other two derivatives. Both in gas phase and in polar
media we found that the symmetric phenol/phenyl ring of the
GFP or BFPF chromophores leads to just two ground-state
configurations, i.e. cis (τ ≈ 0°) and trans (τ ≈ 180°), the latter
being about 2.5-3.0 kcal/mol less stable than the former (Table
1). This result was confirmed by our experiments: no trace of
trans isomers was found in the1H NMR spectra of native cGFP
and cBFPF. Differently, calculations on the asymmetric indolyl
ring in the CFP chromophore yielded four ground-state con-
figurations, two cis (c1CFP, c2CFP) and two trans (t1CFP,
t2CFP) (Scheme 2). Notably, steric hindrance between the
indolyl and the imidazolinone ring leads to a strongly decreased
stability of the two ? ≈ 180° configurations (c2CFP, t2CFP,
see Table 1). As a result, t1CFP becomes more stable than
c2CFP, and this suggests its being detectably populated at room
temperature. The presence of significant trans population at
room temperature was indeed confirmed by our
measurements in methanol. It is worth noting that the calculated
free-energy minima were found to correspond to nearly all-
planar geometrical arrangements of the chromophores, owing
to the formation of an extended electronic conjugation along
the two aromatic rings.
Following the identification of the ground states, we dem-
onstrated that all chromophores were reversibly photoconvertible
between the cis and trans forms, regardless of the solvent
characteristics (Scheme 1, iii-iv). In our experiments, we always
made use of continuous wave light sources and moderate
illumination intensities to avoid triggering of multiphoton
processes. Indeed, it has been demonstrated that high photon
fluxes, such as those typical of fs- and ps-pulsed excitation, may
lead to unexpected photophysics. For example, some authors
found that the two-photon photobleaching of GFP is often
accompanied by higher-order effects, possibly owing to coherent
or sequential higher-order excitation of upper electronic states.33,34
Note that the cis-trans photoconversion of the Dronpa chro-
mophore (similar to cGFP/tGFP) has been reported to be a
single-photon process at illumination intensities similar to those
adopted by us.35
Cis-trans isomerization was found to be associated with an
intensity decrease and red-shift of the low-energy band in the
absorption spectrum (Figure 1a and 3). The red-shift of the trans
spectrum from the cis counterpart is not very large (5-10 nm),
but allowed us to address selectively each form and to show
the reversible photochromism of the chromophores. The struc-
tural identification of the native molecules and their photoprod-
ucts with the cis and trans isomers was made possible by1H
NMR/13C NMR analysis (Figure 2). To date only the1H NMR
spectral details of cGFP and tGFP were reported.27For cBFPF
and tBFPF,13C NMR data were compared with theoretical
calculations and a remarkable agreement between the two data
Figure 3. Molar absorption spectra of neutral GFP (a), BFPF (b), and CFP (c) chromophores in methanol. The cis and trans isomers are reported
in red and blue, respectively. The extinction coefficient scale is common for all chromophores in order to allow a comparison of molar absorption
J. Phys. Chem. B, Vol. 112, No. 34, 2008
Voliani et al.
sets was found (Figure 2b). Furthermore, selectively decoupled
13C NMR spectra allowed for the configurational assignment
to cBFPF and its photoproduct on a pure experimental basis
(Figure 2c). cBFPF and tBFPF were selected for this analysis
on account of their simpler molecular structures and thermal
stability in both forms (see below). Notably,1H NMR allowed
the determination the molar fractions of the cis and trans isomers
in the photoconverted mixtures. From these data and the
measured optical properties of the mixtures, we were able to
reconstruct the molar absorption spectra of pure trans chro-
mophores in solvents with different characteristics (Figure 3).
To our knowledge, this is the first time that the optical spectra
of pure trans FP chromophore forms are presented to the
tGFP and tCFP were shown not to be stable at room
temperature and in protic solvents. They spontaneously recon-
verted back into their cis counterparts (Scheme 1, part iv; Figure
1c), consistently with the higher free-energy of the trans forms.
The decays of tGFP and tCFP were found to be nonmonoex-
ponential, suggesting the presence of reaction intermediates. In
aprotic solvents, however, tGFP and tCFP displayed consider-
able thermal stability. Remarkably, tBFPF was found to be stable
for weeks at room temperature in any solvent. It is tempting to
attribute this complex phenomenology to proton displacement/
rearrangement mechanisms, which may lower the energy barrier
between the trans and cis forms and would be accessible only
in protic solvents. The overall kinetic scheme, however, is not
quite clear yet. The fast equilibrium between the anionic and
neutral GFP chromophore, for which different cis-trans energy
barriers were calculated,16cannot explain the presence of slow-
forming/slow-degrading intermediates on the route to the stable
cis isomer. Indeed, although the decay of tGFP was accelerated
in water compared to methanol, the recovery rate was rather
similar at neutral and alkaline pH.
The kinetic analysis of the photo/thermal isomerization
process predicts that, for moderate optical densities (<0.1) and
front-face illumination geometry, the chromophore absorbance
must follow a monoexponential law (eq S6, Supporting Infor-
mation). Accordingly, we fitted the experimental photoconver-
sion curves of neutral chromophores to eq S6 (Figure 1b) and,
from the knowledge of the irradiation power, the solution
volume, and the extinction coefficients of the pure isomers, we
determined the photoisomerization quantum yields Φcand Φt.
Φc values were also obtained by a flash-photolysis method
relying on single-shot irradiation of the chromophores at 355
nm and measurement of the absorption changes extrapolated at
zero power. Methanol was selected as solvent for these
experiments owing to its mixed protic/organic properties, in an
effort to mimic the chromophore environment in the protein.
In all cases we found that Φc and Φt are greater than 0.1,
suggesting that photoisomerization gives a major contribution
to the deactivation of the excited states of the chromophores.
The quantum yield for the photoconversion of cGFP to tGFP
is consistent with the value recently determined for the same
chromophore system in methanol.29Noticeably, tBFPF and tCFP
showed Φtvalues close to unity. This indicates that in these
chromophores the trans excited-state energy surface determines
de-excitation pathways almost completely leading to the cis
configuration. For comparison, the photoisomerization quantum
yield of the known photochromic proteins undergoing chro-
mophore cis-trans isomerization is in the 10-6-10-3range.10,14
This shows that the inclusion within the ?-barrel and the
establishment of the H-bonding network with the closest protein
environment can decrease by several orders of magnitude the
efficiency of chromophore photoisomerization. Such a difference
highlights how the stereochemical freedom of the protein
chromophore determines the accessibility of the photoisomer-
ization channel. Significantly, many authors have attributed the
determining factor of FP photochromism to a loss of structural
rigidity around the chromophore.12,14,15It is also likely that other
typical processes in FPs, such as proton transfer reactions, are
powerful modulators of the stereoisomerization, eventually
affecting the overall efficiency of photoconversion.36,37It is
worth noting the inhibiting effect of proton donor moieties on
the photoisomerization mechanism has been recently pointed
out for some model chromophores in protic solvents.29
In conclusion, we demonstrated that the cis-trans isomer-
ization is a common deactivation channel of the excited-state
of FP model chromophores. By means of a thorough NMR
analysis combined with DFT calculations, we were able to
attribute the stereochemistry of initial compounds and their
photoproducts. For the first time we reported the optical
absorption spectra of neutral trans chromophore analogues and
the values of photoisomerization quantum yields. The compari-
son with the data available for the reported photochromic
proteins suggests that the chromophore environment plays a
decisive role in determining the effectiveness of the photoin-
duced cis-trans isomerization.
Interestingly, the trans states of ionizable chromophores are
unstable in protic solvents and decay back to the correspondent
cis states in times ranging from few minutes to hours. This result
is consistent with the much higher thermodynamical stability
calculated for the cis isomers by our DFT analysis. Nonetheless,
our DFT analysis indicates that the CFP chromophore should
be characterized by a rather stable trans conformation, owing
to the steric hindrance of the indolyl lateral group. The presence
of detectable trans CFP chromophore at room temperature was
experimentally verified by1H NMR measurements. The char-
acteristics of the thermal relaxation phenomenon indicate that
mobile protons/hydrogen bonds play a relevant role in determin-
ing the activation energy barrier between the trans and cis states.
Acknowledgment. We thank Dr. Valentina Tozzini, Dr.
Stefano Luin and Dr. Guido Pintacuda for stimulating discus-
sions. Partial financial support by MUR within FIRB Project
RBLA03ER38 is gratefully acknowledged.
Supporting Information Available: Text giving details of
the procedure synthesis of the chromophores, the theoretical
description of the isomerization’s kinetics, and the description
of experimental setup for laser flash-photolysis and figures
showing the molar absorption spectra of pure cis-trans isomers
in all tested solvents. This material is available free of charge
via the Internet at http://pubs.acs.org.
References and Notes
(1) Tsien, R. Y. Annu. ReV. Biochem. 1998, 67, 509.
(2) Miyawaki, A. Neuron 2005, 48, 189.
(3) Wachter, R. M. Acc. Chem. Res. 2007, 40, 120.
(4) Tozzini, V.; Pellegrini, V.; Beltram, F. Green Fluorescent Proteins
and their applications to cell biology and bioelectronics. In CRC Handbook
of Organic Photochemistry and Photobiology; Horspool, W., Lenci, F., Eds.;
CRC Press: Boca Raton, FL, 2004; p 139.
(5) Heim, R.; Prasher, D. C.; Tsien, R. Y. Proc. Natl. Acad. Sci. U.S.A.
1994, 91, 12501.
(6) Ting, A. Y.; Kain, K. H.; Klemke, R. L.; Tsien, R. Y. Proc. Natl.
Acad. Sci. U.S.A. 2001, 98, 15003.
(7) Patterson, G. H.; Lippincott-Schwartz, J. Science 2002, 297, 1873.
(8) Gurskaya, N. G.; Verkhusha, V. V.; Shcheglov, A. S.; Staroverov,
D. B.; Chepurnykh, T. V.; Fradkov, A. F.; Lukyanov, S.; Lukyanov, K. A.
Nat. Biotechnol. 2006, 24, 461.
J. Phys. Chem. B, Vol. 112, No. 34, 2008 10721
(9) Cinelli, R. A. G.; Pellegrini, V.; Ferrari, A.; Faraci, P.; Nifosi, R.; Download full-text
Tyagi, M.; Giacca, M.; Beltram, F. Appl. Phys. Lett. 2001, 79, 3353.
(10) Nifosi, R.; Ferrari, A.; Arcangeli, C.; Tozzini, V.; Pellegrini, V.;
Beltram, F. J. Phys. Chem. B 2003, 107, 1679.
(11) Andresen, M.; Wahl, M. C.; Stiel, A. C.; Grater, F.; Schafer, L. V.;
Trowitzsch, S.; Weber, G.; Eggeling, C.; Grubmuller, H.; Hell, S. W.;
Jakobs, S. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 13070.
(12) Henderson, J. N.; Ai, H. W.; Campbell, R. E.; Remington, S. J.
Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 6672.
(13) Andresen, M.; Stiel, A. C.; Trowitzsch, S.; Weber, G.; Eggeling,
C.; Wahl, M. C.; Hell, S. W.; Jakobs, S. Proc. Natl. Acad. Sci. U.S.A. 2007,
(14) Habuchi, S.; Ando, R.; Dedecker, P.; Verheijen, W.; Mizuno, H.;
Miyawaki, A.; Hofkens, J. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 9511.
(15) Loos, D. C.; Habuchi, S.; Flors, C.; Hotta, J.; Wiedenmann, J.;
Nienhaus, G. U.; Hofkens, J. J. Am. Chem. Soc. 2006, 128, 6270.
(16) Weber, W.; Helms, V.; McCammon, J. A.; Langhoff, P. W. Proc.
Natl. Acad. Sci. U.S.A. 1999, 96, 6177.
(17) Nifosı `, R.; Amat, P.; Tozzini, V. J. Comput. Chem. 2007, 28, 2366.
(18) Kojima, S.; Ohkawa, H.; Hirano, T.; Maki, S.; Niwa, H.; Ohashi,
M.; Inouye, S.; Tsuji, F. Tetrahedron Lett. 1998, 39, 5239.
(19) Banderini, A.; Sottini, S.; Viappiani, C. ReV. Sci. Instrum. 2004,
(20) VanBrederode, M. E.; Gensch, T.; Hoff, W. D.; Hellingwerf, K. J.;
Braslavsky, S. E. Biophys. J. 1995, 68, 1101.
(21) Cossi, M.; Barone, V. J. Chem. Phys. 2001, 115, 4708.
(22) Ruud, K.; Helgaker, T.; Bak, K. L.; Jørgensen, P.; Jensen, H. J. A.
J. Chem. Phys. 1993, 99, 3847.
(23) Frisch, M. et al. Gaussian Inc.: Pittsburgh, PA; revision B.03 ed.,
(24) Mukerjee, A.; Kumar, P. Heterocycles 1981, 16, 1995.
(25) Rao, Y.; Filler, R. Synthesis 1975, 12, 749.
(26) Berman, H.; Westbrook, J.; Feng, Z.; Gilliland, G.; Bhat, T.;
Weissig, H.; Shindyalov, I.; Bourne, P. Nucl. Acids. Res 2000, 28, 235.
(27) He, X.; Bell, A. F.; Tonge, P. J. FEBS Lett. 2003, 549, 35.
(28) Bell, A. F.; He, X.; Wachter, R. M.; Tonge, P. J. Biochemistry
2001, 40, 8619.
(29) Yang, J.; Huang, S.; Liu, Y.; Peng, S. Chem. Commun. 2008, 1344.
(30) Morgenstern, A.; Schutij, C.; Nauta, W. Chem. Commun. 1969,
(31) Prokof’ev, E.; Karpeiskaya, E. Tetrahedron Lett. 1979, 8, 737.
(32) Bensasson, R. V.; Land, E. J.; Truscott, T. G. Excited states and
free radicals in biology and medicine; Oxford University Press: Oxford,
(33) Mondal, P. P.; Diaspro, A. Phys. ReV. E: Stat. Nonlin. Soft Matter
Phys. 2007, 75, 061904.
(34) Patterson, G. H.; Piston, D. W. Biophys. J. 2000, 78, 2159.
(35) Habuchi, S.; Dedecker, P.; Hotta, J.; Flors, C.; Ando, R.; Mizuno,
H.; Miyawaki, A.; Hofkens, J. Photochem. Photobiol. Sci. 2006, 5, 567.
(36) Schafer, L. V.; Groenhof, G.; Boggio-Pasqua, M.; Robb, M. A.;
Grubmuller, H. PLoS Comput. Biol. 2008, 4, e1000034.
(37) Fron, E.; Flors, C.; Schweitzer, G.; Habuchi, S.; Mizuno, H.; Ando,
R.; Schryver, F. C.; Miyawaki, A.; Hofkens, J. J. Am. Chem. Soc. 2007,
J. Phys. Chem. B, Vol. 112, No. 34, 2008
Voliani et al.