Activation and tuning of green fluorescent protein chromophore emission by alkyl substituent-mediated crystal packing

Article abstract of DOI:10.1021/ja806962e

OAlkyl synthetic analogues of the green fluorescent protein chromophore are nonfluorescent in solution but demonstrate a bright emission in the crystalline state. Three-dimensional steady-state and time-resolved emission spectroscopies in the solid state,

Full text of DOI:10.1021/ja806962e

Published on Web 12/23/2008
Activation and Tuning of Green Fluorescent Protein
Chromophore Emission by Alkyl Substituent-Mediated Crystal
Packing
Jian Dong, Kyril M. Solntsev,* and Laren M. Tolbert*
School of Chemistry and Biochemistry, Georgia Institute of Technology,
901 Atlantic DriVe, Atlanta, Georgia 30332-0400
Received September 3, 2008; E-mail: solntsev@gatech.edu; tolbert@chemistry.gatech.edu
Abstract: O-Alkyl synthetic analogues of the green fluorescent protein chromophore are nonfluorescent in
solution but demonstrate a bright emission in the crystalline state. Three-dimensional steady-state and
time-resolved emission spectroscopies in the solid state, as well as single-crystal X-ray diffraction, reveal
the nature of complex emission in the crystals. A hypsochromic emission shift with an increase of the alkyl
size is determined by the monomer-aggregate emission ratio.
Introduction
providing steric hindrance to the cis/trans isomerization pathway
in the excited state and allowing the superb fluorescence
The widespread use of green fluorescent protein (GFP) as a
biological marker in molecular biology, medicine, and cell
biology1-3 culminated in the 2008 Nobel Prize in chemistry.
The chromophore of this protein, a p-hydroxybenzylideneimi-
dazolinone (p-HOBDI), is covalently bound to the middle of
an internal R-helix directed along a rigid ꢀ-barrel axis and
further constrained via an extended hydrogen-bond network
involving the phenolic hydroxyl and carbonyl groups, with some
degree of rotational freedom.4 Upon photoexcitation, facile
excited-state proton transfer from the neutral excited chro-
mophore occurs to form the fluorescent anionic state. However,
due to the free rotation around the exo-methylene double bond
in the excited state,5 the fluorescence of either denatured protein
chromophores or synthetic chromophore derivatives is tremen-
dously quenched in fluid solution at room temperature.6-8
Escaping constraints of the ꢀ-barrel, the cis/trans isomerization
rotation exhibits a weak viscosity dependence9-11 and can only
be hindered by cooling of the solution to 77 K, resulting in
fluorescence recovery. The ꢀ-barrel thus plays a critical role in
properties of GFP and its mutants. However, the nature of the
interplay among various intermolecular forces involved in such
proteins is very complex, and even as simple a question as the
degree of constraint necessary to achieve high fluorescence
quantum yields is still unresolved.
Previous studies have shown that the protein cavity surround-
ing the chromophore in GFP or GFP-like proteins in the ground
state is not ideal for a planar chromophore.5 The chromophore
itself is planar because of its highly conjugated π system. The
protein matrix, conversely, imposes some distortions away from
planarity. This distorting force should have an effect on the
fluorescence of the chromophore, for example, serving as the
initial accelerant for the light-induced photocycle;12 however,
the photophysical mechanism remains unclear.
Unfortunately, most in vitro characterizations of GFP chro-
mophores have been performed only in solution, which is
different from the environment of native GFP chromophore,
since ꢀ-barrel prevents the chromophore fluorescence quenching
from external solvents.1 Some gas-phase studies were reported
recently.13 Again, the deficiency of the restriction from inter-
(1) Zimmer, M. Chem. ReV. 2002, 102, 759–781.
(2) Tsien, R. Y. Annu. ReV. Biochem. 1998, 67, 509–544.
(3) Green Fluorescent Protein: Properties, Applications, and Protocols,
2nd ed.; Chalfie, M., Kain, S. R., Eds.; Wiley-Intersciences; Hoboken,
NJ, 2005.
(12) Yamada, A.; Ishikura, T.; Yamato, T. Proteins: Struct., Funct.
Bioinform. 2004, 55, 1063–1069.
(13) (a) Andersen, L. H.; Bluhme, H.; Boye´, S.; Jørgensen, T. J. D.; Krogh,
H.; Nielsen, I. B.; Nielsen, S. B.; Svendsen, A. Phys. Chem. Chem.
Phys. 2004, 6, 2617–2627. (b) Lammich, L.; Petersen, M. Å.; Nielsen,
M. B.; Andersen, L. H. Biophys. J. 2007, 92, 201–207.
(4) Chen, M. C.; Lambert, C. R.; Urgitis, J. D.; Zimmer, M. Chem. Phys.
2001, 270, 157–164.
(5) Maddalo, S. L.; Zimmer, M. Photochem. Photobiol. 2006, 82, 367–
372.
(14) (a) Deans, R.; Kim, J.; Machacek, M. R.; Swager, T. M. J. Am. Chem.
Soc. 2000, 122, 8565–8566. (b) An, B.-K.; Kwon, S.-K.; Jung, S.-D.;
Park, S. Y. J. Am. Chem. Soc. 2002, 124, 14410–14415. (c) Luo, J.;
Xie, Z.; Lam, J. W. Y.; Cheng, L.; Chen, H.; Qiu, C.; Kwok, H. S.;
Zhan, X.; Liu, Y.; Zhu, D.; Tang, B. Z. Chem. Commun. 2001, 1740–
1741. (d) Chen, J.; Xu, B.; Ouyang, X.; Tang, B. Z.; Cao, Y. J. Phys.
Chem. A 2004, 108, 7522–7526. (e) Li, Z. H.; Wong, M. S.; Tao, Y.;
Lu, J. Chem. Eur. J. 2005, 11, 3285–3293.
(6) Niwa, H.; Ionuye, S.; Horano, T.; Matsuno, T.; Kojima, S.; Kubota,
M.; Ohashi, M.; Tsuji, F. I. Proc. Natl. Acad. Sci. U.S.A. 1996, 93,
13617–13622.
(7) Kojima, S.; Ohkawa, H.; Hirano, T.; Maki, S.; Niwa, H.; Ohashi, M.;
Inouye, S.; Tsuji, F. I. Tetrahedron Lett. 1998, 39, 5239–5242.
(8) Litvinenko, K. L.; Webber, N. M.; Meech, S. R. J. Phys. Chem. A
2003, 107, 2616–2623.
(9) Mandal, D.; Tahara, T.; Webber, N. M.; Meech, S. R. Chem. Phys.
Lett. 2002, 358, 495–501.
(15) Dreuw, A.; Plo¨ner, J.; Lorenz, L.; Wachtveitl, J.; Djanhan, J. E.;
Bru¨ning, J.; Metz, T.; Bolte, M.; Schmidt, M. U. Angew. Chem., Int.
Ed. 2005, 44, 7783–7786.
(10) Mandal, D.; Tahara, T.; Meech, S. R. J. Phys. Chem. B 2004, 108,
1102–1108.
(16) Ooyama, Y.; Yoshida, K. New J. Chem. 2005, 29, 1204–1212.
(17) Mizukami, S.; Houjou, H.; Sugaya, K.; Koyama, E.; Tokuhisa, H.;
Sasaki, T.; Kanesato, M. Chem. Mater. 2005, 17, 50–56.
(11) Litvinenko, K. L.; Webber, N. M.; Meech, S. R. J. Phys. Chem. A
2003, 107, 2616–2623.
9
662 J. AM. CHEM. SOC. 2009, 131, 662–670
10.1021/ja806962e CCC: $40.75
2009 American Chemical Society

Tuning GFP Chromophore Emission
A R T I C L E S
Figure 1. Real-color photograph of ROBDI crystals under daylight and UV illumination. The structure of ROBDI molecules is presented in the middle.
molecular hydrogen bonding and π-π interaction limits further
understanding of the chromophore in GFP and its mutants.
Compared to that in solution, some fluorophore molecules can
be restricted in their motions and adopt planar conformation as
aggregates in the solid state, which results in more effective
conjugation between the electron donor and acceptor groups of
the molecules. This effective conjugation may further induce
the red-shift of the emission maximum and enhance the
fluorescence quantum yields.14 However, organic fluorophores
with high solid-state fluorescence are less common. Many
organic chromophores are highly emissive in dilute solutions
but become weakly luminescent or nonluminescent in their solid
states. The formation of molecular aggregates by strong
intermolecular π-π interactions15-17 or intermolecular hydro-
gen bonding18 in the solid state generally results in reduction
of luminescence efficiency. In view of the fact that most GFP
chromophore derivatives are nonfluorescent in room-tempera-
ture solutions, it is illuminating to examine their intermolecular
interactions in bulk materials that lead to the formation of
aggregates, significantly altering the photophysical and photo-
chemical properties of these molecules, resulting in large
changes in their spectral characteristics and luminescence
efficiency. Moreover, the optical and photophysical properties
of organic π-conjugated molecules in the solid state, defined
largely by their molecular packing, which in turn is controlled
by various noncovalent forces, have been used as key elements
in optoelectronic devices.19,20 Understanding the nature of these
forces and controlling them for the design of functional materials
is of significant importance, especially with reference to tuning
their application in devices such as light-emitting diodes,21
photovoltaics,22 and field effect transistors.23,24 In order to
provide further insight into the photophysics of this important
chromophore, therefore, we embarked on a study of such
chromophores in the solid state, where solid-state crystal packing
might be expected to play a significant role.
Our initial results were obtained with the well-known
hydroxybenzylideneimidazolinone derivative in which the pro-
tein side chains were replaced with methyl groups (HOBDI).
Taking advantage of the well-known propensity of long-chain
alkyl groups to affect aggregation states in solution and packing
in the solid, we also prepared and obtained the fluorescence
properties of several isoelectronic O-alkyl benzylideneimida-
zolinone (ROBDI) GFP chromophore derivatives in solution
and in the solid state. The role of crystal packing in controlling
the intensity and wavelength of the solid-state fluorescence of
these derivatives is elucidated, and its relevance to GFP itself
is described.
Results
Photophysical Properties in Solution. The synthesis and
purification of alkoxybenzylideneimidazolinones (ROBDI) de-
rivatives (R ) C6H13 (“Hex”), C12H25 (“Ddec”)) were analogous
to those of the MeOBDI reported by us earlier25 and are
described in the Supporting Information. The absorption spectra
of all neutral ROBDI derivatives were measured in various
solvents. As in previous measurements of MeOBDI,25 the
absorption maxima of these derivatives (ca. 370 nm) were nearly
identical and exhibited the weak solvent dependence reported
earlier. Similar to HOBDI, all ROBDI derivatives exhibited
an extremely weak steady-state emission in various solvents at
room temperature,26 which was obscured by Raman scattering.
The λmax values observed for these derivatives (∼450 nm) were
also consistent with those previously reported.26,27
Solid-State Luminescence. In contrast to solutions, the crystals
of HexOBDI and DdecOBDI exhibited strong greenish-yellow
and blue fluorescence emission, respectively, upon irradiation
with a hand-held UV lamp, whereas MeOBDI exhibited weaker
dark-red fluorescence (see Figure 1). The crystals of HOBDI,
however, still remained nonfluorescent.
(18) Yoshida, K.; Uwada, K.; Kumaoka, H.; Bu, L.; Watanabe, S. Chem.
Lett. 2001, 8, 808–809.
(19) Kumar, N. S. S.; Varghese, S.; Rath, N. P.; Das, S. J. Phys. Chem. C.
2008, 112, 8429–8437.
(20) (a) Hudson, A. J.; Weaver, M. S. In Functional Organic and Polymeric
Materials; Richardson, T. H., Ed.; John Wiley & Sons: New York,
2000. (b) Kaneko, M. In Handbook of Organic Molecules and
Polymers; Nalwa, H. S., Ed.; Wiley: New York, 1997.
(23) (a) Dimitrakopoulos, C. D.; Malenfant, P. R. L. AdV. Mater. 2002,
14, 99–117. (b) Garnier, F.; Hajlaoui, R.; Yassar, A.; Srivastava, P.
Science 1994, 265, 1684–1686. (c) Li, X.-C.; Sirringhaus, H.; Garnier,
F.; Holmes, A. B.; Moratti, S. C.; Feeder, N.; Clegg, W.; Teat, S. J.;
Friend, R. H. J. Am. Chem. Soc. 1998, 120, 2206–2207. (d) Kelly,
S. M.; O’Neill, M. In Handbook of AdVanced Electronic and Photonic
Materials; Nalwa, H. S., Ed.; Academic Press: San Diego, CA, 2000;
Vol. 1,Chapter 1.
(21) (a) Burroughes, J. H.; Bradley, D. D. C.; Brown, A. R.; Marks, R. N.;
Mackay, K.; Friend, R. H.; Holmes, P. L.; Holmes, A. B. Nature 1990,
347, 539–541. (b) Kraft, A.; Reichert, A. Tetrahedron 1999, 55, 3923–
3930. (c) Organic Light Emitting DeVices: Synthesis, Properties and
Applications; Mu¨llen, K., Scherf, U., Eds.; Wiley-VCH: Weinheim,
2006. (d) Yamaguchi, Y.; Ochi, T.; Miyamura, S.; Tanaka, T.;
Kobayashi, S.; Wakamiya, T.; Matsubara, Y.; Yoshida, Z.-I. J. Am.
Chem. Soc. 2006, 128, 4504–4505. (e) Armaroli, N.; Accorsi, G.;
Holler, M.; Moudam, O.; Nierengarten, J.-F.; Zhou, Z.; Wegh, R. T.;
Welter, R. AdV. Mater. 2006, 18, 1313–1316.
(24) Davis, R.; Kummar, N. S. S.; Abraham, S.; Suresh, C. H.; Rath, N. P.;
Tamaoki, N.; Das, S. J. Phys. Chem. C 2008, 112, 2137–2146.
(25) Dong, J.; Solntsev, K. M.; Tolbert, L. M. J. Am. Chem. Soc. 2006,
128, 12038–12039.
(26) Solntsev, K. M.; Poizat, O.; Dong, J.; Rehault, J.; Lou, Y.; Burda, C.;
Tolbert, L. M. J. Phys. Chem. B 2008, 112, 2700–2711.
(27) Kummer, A. D.; Kompa, C.; Niwa, H.; Hirano, T.; Kojima, S.; Michel-
Beyerle, M. E. J. Phys. Chem. B 2002, 106, 7554–7559.
(22) (a) O’Neill, M.; Kelly, S. M. AdV. Mater. 2003, 15, 1135–1146. (b)
Shirota, Y. J. Mater. Chem. 2005, 15, 75–93. (c) Strohriegl, P.;
Grazulevicius, J. V. AdV. Mater. 2002, 14, 1439–1452.
9
J. AM. CHEM. SOC. VOL. 131, NO. 2, 2009 663

A R T I C L E S
Dong et al.
Figure 2. (Top) Solid-state fluorescence spectra of HOBDI (crystal) and
MeO-DdecOBDI (thin films). (Bottom) Normalized solid-state fluorescence
spectra of MeOBDI; λex ) 370 nm.
In order to investigate the difference in the solid-state photo-
physical properties among these ROBDI derivatives, we measured
the fluorescence excitation and emission spectra of the crystals and
thin films. With the exception of MeOBDI, the emission spectra
in their powdered crystalline state and in thin films formed from
their melts were almost identical. Considering the lack of reproduc-
ibility of the fluorescence intensity in loading powdered crystalline
samples,28,29 thin films of these ROBDI derivatives were used in
the measurements. A thin-film form was not available for HOBDI,
because it decomposed above 200 °C. The fluorescence spectra of
these ROBDI derivatives were much broader, and their emission
intensities were significantly higher, compared to those in solution.
As shown in Figure 2, the fluorescence intensity of HOBDI was
still low and showed a very weak emission peak at about 465 nm.
Compared to HOBDI, more than a 10-fold fluorescence enhance-
ment was observed for its alkyl ethers. The emission maximum of
a DdecOBDI thin film sample was also at ∼465 nm, with a peak
position similar to that of the solid-state HOBDI sample; however,
the spectrum of MeOBDI was red-shifted to λmax ) 560 nm. The
HexOBDI sample showed the broadest and highest fluorescence
intensity and appeared as a superposition of the MeOBDI and
DdecOBDI emission bands.
The optical properties of MeOBDI samples depended signifi-
cantly on the nature of their preparation (Figure 2, bottom). The
emission spectra of its crystals showed two structured bands, with
one sharp peak at 450 nm and the other broad band centered at
∼550 nm. The emission spectra of thin films formed by heating
above its melting point and rapidly cooling to room temperature
showed a band centered at ∼600 nm. However, this kind of thin
film was less stable thermally, and a blue-shift in its emission
maximum was observed after 2 days at room temperature. The
Figure 3. Excitation-emission contour plots of solid ROBDI samples.
The difference between the neighboring lines corresponds to the constant
intensity difference on a linear scale. Deeper colors correspond to higher
intensities.
difference in fluorescence of two different solid states could be
attributed to the formation of a thermodynamically unstable
aggregation form upon melting and rapid cooling. A similar
polymorphic phenomenon was also reported in ref 19. In contrast,
slowly cooled thin films were very stable at ambient conditions.
In addition, the fluorescence of crystalline MeOBDI, as well as
HexOBDI and DdecOBDI, vanished above its melting point
(Figure S1, Supporting Information), but the fluorescence reap-
peared after cooling. This process was reversible. In contrast to
the crystals of the Photoactive Yellow Protein synthetic chro-
mophore30 no fluorescence turn-off or turn-on was observed upon
prolonged irradiation of the ROBDI crystals, demonstrating an
absence of photoinduced cis-trans isomerization.
For a more detailed study of the photophysical properties of
these ROBDI derivatives in the solid phase, we investigated
their excitation-emission matrix (EEM) spectroscopy. EEM
spectroscopy has recently emerged as an effective method for
spectroscopic characterization of multifluorophoric systems.31-34
We collected the emission spectra of the fluorophores at room
temperature, scanning the excitation wavelength from 250 to
(30) Usman, A.; Masuhara, H.; Asahi, T. J. Phys. Chem. B 2006, 110,
20085–20088.
(31) Weber, G. Nature 1961, 190, 27–29.
(32) Clower, C.; Solntsev, K. M.; Kowalik, J.; Tolbert, L. M.; Huppert, D.
J. Phys. Chem. A 2002, 106, 3114–3122.
(28) Ramachander, R.; Jiang, Y.; Li, C.; Eris, T.; Young, M.; Dimitrova,
M.; Narhi, L. Anal. Biochem. 2008, 376, 173–182.
(33) Gill, D. M.; Wright, J. C.; McCaughan, L. Appl. Phys. Lett. 1994, 64,
2483–2485.
(29) Gardecki, J. A.; Maroncelli, M. Appl. Spectrosc. 1998, 52, 1179–1189.
(34) Divya, O.; Mishra, A. K. Anal. Chim. Acta 2007, 592, 82–90.
9
664 J. AM. CHEM. SOC. VOL. 131, NO. 2, 2009

Tuning GFP Chromophore Emission
A R T I C L E S
Figure 4. Emission spectra of solid ROBDI samples at various excitation wavelengths. Color coding is the same for all graphs.
550 nm at 5 nm increments. In Figure 3, the emission spectra
of the ROBDI samples are displayed as contour plots, which
allowed the observation of every emission band and its
correlation to the appropriate excitation wavelength. The contour
plot of MeOBDI was the simplest, showing mainly only one
symmetric peak at Ex/Em of 470/550 nm. The contour plots of
HexOBDI and DdecOBDI were highly asymmetrical, demon-
strating two major emissive species: a long-wavelength one
similar to that of MeOBDI with Ex/Em peaks of 470/550 nm
for HexOBDI and 470/515 nm for DdecOBDI, and another at
shorter wavelengths with Ex/Em peaks of 420/460 nm for
HexOBDI and 440/460 nm for DdecOBDI. Some emission was
observed from HOBDI only upon long-wavelength visible
excitation, but not with UV light (Figures 1 and 3). Manual
deletion of the strong excitation light scattering signals resulted
in the empty zones in the bottom right corner of each graph. It
should be noted that the light scattering, as well as refraction
from crystal or film morphology and geometrical effects
observed on microspectrophotometer devices, may slightly alter
the fluorescence spectra. Some selected emission spectra from
each sample demonstrating the spectral evolution upon excitation
wavelength shifts are presented in Figure 4.
Lifetime measurements used in fluorescence spectroscopy, for
instance, time-correlated single-photon counting (TCSPC), are
expected to be less dependent on concentration and method of
preparation than steady state, thus providing an easier comparison
of crystalline samples. A combination of time-resolved and steady-
state emission techniques may allow the deconvolution of contribu-
tions from multichromophoric systems. Fluorescence lifetimes
obtained from measurements of the solid-state of materials are often
complex, and multiexponential decay profiles are observed, which
could be attributed to the presence of molecules in different states
of aggregation in the solid state.35 In addition, due to the complexity
of the excitation spectra in the solid-state ROBDI derivatives, it
is feasible to use pulsed diode lasers with two different wavelengths
(372 and 467 nm) as excitation sources in the TCSPC system. To
reveal the complex nature of the emitting species in the solid-state
samples, we collected intensity-wavelength-time 3D surfaces
similarly to the steady-state results.
Figure 5 demonstrates the decay surfaces of alkoxyBDIs. At
372 nm excitation, the surfaces of all compounds clearly
demonstrate an existence of two major decaying species: a short-
decaying one with a maximum around 450 nm, and another
significantly longer, decaying with the maximum around 570
nm therefore resembling steady-state results. The relative
amplitude of the short-lived component increased dramatically
in the order of C1 > C6 > C12. The HOBDI emission at 372
nm excitation was too weak and short-lived to be detected with
our TCSPC instrument. The time-resolved emission surfaces
of all compounds at 467 nm excitation demonstrated only one
component with the emission maximum at 570 nm observed in
the previous case. The decays of this component were uniform
through the whole spectrum for each compound. Some selected
2D fluorescence decay curves are presented in Figure 6. One
can see that the nonexponential decay at 570 nm for each sample
was practically independent of the excitation wavelength. The
average lifetime of the former was higher for the long-chain
derivatives. The decay at 450 nm was ultrashort for MeOBDI
and much longer for HexOBDI and DdecOBDI. The data
measured on the 10 or 20 ns time scales were fitted to a sum of
two or three (n ) 2 or 3) exponentials,
n
I(t)
I(0)
)
A exp(-t ⁄ τ ) X IRF(t)
(1)
∑
i
i
i)1
and convoluted with the instrument response function (IRF). It
was normalized so that ∑in) 1 Ai ) 1. The results of fitting are
presented in Table 1.
(35) Li, H.; He, L.; Zhong, B.; Li, Y.; Wu, S.; Liu, J.; Yang, G.
ChemPhysChem. 2004, 5, 124–127.
9
J. AM. CHEM. SOC. VOL. 131, NO. 2, 2009 665

A R T I C L E S
Dong et al.
similar to HOBDI, but the angle between two mean planes of
alternate molecules increases from 31.9° to 50.4° (see Figure
8a). Since the hydroxyl proton is replaced by a methyl group
in MeOBDI, the possibility of an inter- or intramolecular
hydrogen bond is eliminated. Moreover, this herringbone
configuration is stabilized in the AB couple by aromatic
C-H· · ·O (2.651 Å) and aromatic C-H· · ·N (2.845 Å) interac-
tions,24 as shown in Figure 8b. In the case of AA′ (not shown),
aromatic π-π interactions from the phenyl rings of both A and
A′ are seen, and the overlapping of the parallel layers is partially
reduced compared to that in HOBDI. The dipole moments of
the molecules in the AA′ pair are still antiparallel as for HOBDI.
The structure of HexOBDI is significantly different from
those of HOBDI and MeOBDI. It has only two molecules per
unit cell (Figure S4, Supporting Information). The relative
alignment of A and B is the most altered, as they are aligned
antiparallel to each other (Figure 9a), while in both HOBDI
and MeOBDI they are arranged in an angular fashion with
respect to each other. Two strong intermolecular aromatic
C-H· · ·O (2.677 Å) interactions (electronically) are responsible
for the planar AB alignment, whereas the AA′ unit is character-
ized by the presence of both aliphatic C-H· · ·O and aliphatic
C-H· · ·π interactions (Figure 9b). Such C-H· · ·O hydrogen-
bonding-stabilized planarization of crystal structures has been
reported for various molecules.24,38 The only aggregation
interaction that can be observed consists of aromatic C-H and
imidazoline moiety interactions with a distance of 3.298 Å
between neighboring molecules. However, since A and A′ are
translated along both the long and short axes relative to each
other with parallel dipole moments, the π-overlap between them
is also minimized (Figure 9c).
Figure 5. Time-resolved emission spectra of solid ROBDI samples at 372
and 467 nm laser excitation.
The structure of DdecOBDI is very similar to that of
HexOBDI, with AB- and AA′-type dimer units with aromatic
C-H· · ·O, aliphatic C-H· · ·O, and aliphatic C-H· · ·π interac-
tions (Figures S5,Supporting Information; Figure 10). Compared
to HexOBDI, the A and B units of DdecOBDI make a stronger
interaction for the coplanar alignment while the aromatic planes
of the shifted parallel units A and A′ in DdecOBDI are located
at a larger distance compared to HexOBDI (see Figures 9b,c
and 10b,c).
Crystal Structure. Single crystals of all ROBDI compounds
were grown at room temperature in dichloromethane or ethyl
acetate solvents. Their crystal data are summarized in Table 2.
HOBDI has eight molecules per unit cell (Figure S2,
Supporting Information), with the molecules arranged into two
nonequivalent stacks with their dipoles pointing in opposite
directions in a herringbone fashion.36 A part of the X-ray
structure of HOBDI containing four monomer molecules is
shown in Figure 7a. The figure also includes two major
alignments of the molecules, namely, (1) AB (alternate stack)
and (2) AA′ (neighboring stack), which are separately shown
in Figure 7b,c. Most significantly, an intermolecular hydrogen
bond between the hydroxyl proton and the carbonyl oxygen
(O-H · · · O distance and angle are 2.673 Å and 173.7°, re-
spectively) is observed in AB couples of HOBDI. The
hydrogen-bond formation is accompanied by the CdO bond
lengthening in HOBDI as compared to all other ROBDIs (see
Table 2). In addition, the imidazolinone and phenyl moieties in
AA′ and BB′ couples overlap, with interplane distances of 3.590
and 3.546 Å. Intermolecular hydrogen bonding18 and π-π
interactions in the solid state15,16,37 have been reported as general
factors in fluorescence quenching.
Discussion
The GFP synthetic chromophore HOBDI and its derivatives
are nonfluorescent in solutions at room temperature, and their
fluorescence quenching is caused presumably by a nonadiabatic
transition induced by the free rotation around the exo-methylene
double bond in the excited state. In their solid states, the
formation of aggregates from intermolecular interaction sig-
nificantly alters the photophysical and photochemical properties
of the molecules, resulting in large changes in their spectral
characteristics and luminescence efficiency.39-41
The most striking difference among the solid-state fluores-
cence data is the quenching of the HOBDI fluorescence. The
crystal packing of HOBDI indicates the electronic interaction
is maximum for AA′, BB′ couples; in addition, the effect of an
One unit cell of MeOBDI contains four molecules (Figure
S3, Supporting Information). The structure of MeOBDI is
(38) (a) Desiraju, G. R. Acc. Chem. Res. 1996, 29, 441–449. (b) Desiraju,
G. R. Acc. Chem. Res. 2002, 35, 565–573. (c) Vargas, R.; Garza, J.;
Dixon, D. A.; Hay, B. P. J. Am. Chem. Soc. 2000, 122, 4750–4755.
(d) Jeffrey, G. A. An Introdution to Hydrogen Bonding; University
Press: Oxford, U.K., 1997.
(36) (a) Koren, A. B.; Curtis, M. D.; Francis, A. H.; Kampf, J. W. J. Am.
Chem. Soc. 2003, 125, 5040–5050. (b) Gierschner, J.; Ehni, M.;
Egelhaaf, H.-J.; Medina, B. M.; Beljonne, D.; Benmansour, H.; Bazan,
G. C. J. Chem. Phys. 2005, 123, 144914. (c) Spano, F. C. J. Chem.
Phys. 2003, 118, 981. (d) Nishio, M. CrystEngComm 2004, 6, 130–
158.
(39) Cornil, J.; Beljonne, D.; Calbert, J.-P.; Bre´das, J.-L. AdV. Mater. 2001,
13, 1053–1067.
(37) Mizukami, S.; Houjou, H.; Sugaya, K.; Koyama, E.; Tokuhisa, H.;
Sasaki, T.; Kanesato, M. Chem. Mater. 2005, 17, 50–56.
(40) Vaday, S.; Geiger, H. C.; Cleary, B.; Perlstein, J.; Whitten, D. G. J.
Phys. Chem. B 1997, 101, 321–329.
9
666 J. AM. CHEM. SOC. VOL. 131, NO. 2, 2009

Tuning GFP Chromophore Emission
A R T I C L E S
Figure 6. Fluorescence decays curves of solid ROBDI samples at various excitation wavelengths detected at various emission wavelengths. Color coding
is the same for all graphs.
Table 1. Parameters Obtained from Multi-exponential Fits (Eq 1)
bonding involving both hydroxyl and carbonyl groups, leading
to quenching, may have interesting analogies with other proteins
of the ROBDI Time-Resolved Emission Dataa
compound
λ
obs, nm
τ1, ns
A1/Ai
τ2, ns
A2/Ai
τ3, ns
having hydrogen bonds of similar nature. We have also recently
demonstrated26 that in meta-HOBDI the excited-state proton
transfer (ESPT) is diabatic and induces the S1fS0 deactivation
when it leads to a zwitterion (protonation of the m-HOBDI
anion, deprotonation of the m-HOBDI cation), but it is adiabatic
when it leads to a single-charged form (deprotonation of
m-HOBDI and m-MeOBDI). This behavior agrees perfectly
with the previous assumption about the role of zwitterion as a
quenching product in the GFPs.43
MeOBDI
450
570
450
570
450
570
570
0.020b
0.04
0.12
0.09
0.11
0.11
0.03
0.99
0.52
0.49
0.34
0.76
0.59
0.65
0.72
0.54
0.34
1.2
0.36
0.88
0.38
0.01
0.30
0.49
0.51
0.23
0.25
0.22
-
1.6
1.2
2.6
3.7
3.4
0.98
HexOBDI
DdecOBDI
HOBDI
a C1-C12 samples were excited at 372 nm. HO sample was excited
at 467 nm. The quality of fit was evaluated by the residuals analysis and
ꢁ2 values, which were always smaller than 1.15. b Below the resolution
of our TCSPC system.
In contrast to HOBDI, the solid-state luminescence of alkyl
ROBDI derivatives is enhanced dramatically compared to that
observed in solution, which may be attributed to “freezing” of
planar conformation of the fluorophores, resulting in more
effective conjugation between the electron donor and acceptor
groups and avoiding the quenching present in hydrogen bonding
crystals. However, such enhancement is not uniform, and we
conclude that freezing of conformation is a necessary but not
sufficient condition for efficient emission. Various emission
colors were observed from ROBDI crystals upon UV irradiation,
which was a little surprising since the electronic structures of
the chromophores were essentially the same. We will discuss
now how the inert alkyl groups play a dramatic role in color
tuning of the ROBDI crystals emission.
intermolecular hydrogen bond may lead to the efficient quench-
ing of the HOBDI solid-state fluorescence. This is in contrast
to the fluorescent proteins, where the presence of a hydrogen-
bonding arginine (R96) is an obligatory component of matura-
tion.42 The spectroscopic role of R96 has not been successfully
explained, and we suspect that, in the absence of R96 (pKa )
12.48 in water), other hydrogen-bonding residues may lead to
formal proton transfer in the excited state, leading to fluores-
cence quenching just as in the HOBDI crystal. Strong hydrogen
(41) (a) Sheikh-Ali, B. M.; Rapta, M.; Jameson, G. B.; Cui, C.; Weiss,
R. G. J. Phys. Chem. 1994, 98, 10412–10418. (b) Sheikh-Ali, B. M.;
Weiss, R. G. J. Am. Chem. Soc. 1994, 116, 6111–6120. (c) Lewis,
F. D.; Yang, J.-S.; Stern, C. L. J. Am. Chem. Soc. 1996, 118, 2772–
2773. (d) Lewis, F. D.; Yang, J.-S.; Stern, C. L. J. Am. Chem. Soc.
1996, 118, 12029–12037.
Consistent with previous reports,24,44 the solid-state fluores-
cence spectrum of the derivative possessing a shorter alkoxy
(42) Wood, T. I.; Barondeay, D. P.; Hitomi, C.; Kassmann, C. J.; Tainer,
J. A.; Getzoff, E. D. Biochemistry 2005, 44, 16211–16220.
(43) Weber, W.; Helms, V.; McCammon, J. A.; Langhoff, P. W. Proc.
Natl. Acad. Sci. U.S.A. 1999, 96, 6177–6182.
9
J. AM. CHEM. SOC. VOL. 131, NO. 2, 2009 667

A R T I C L E S
Dong et al.
Table 2. Summary of Crystallographic Data for ROBDI derivatives
HOBDI
MeOBDI
HexOBDI
DdecOBDI
empirical formula
formula weight
C12H12N2O2
216.24
C13H14N2O2
230.26
C18H24N2O2
300.39
C24H36N2O2
384.55
crystal system
space group
Z
monoclinic
P2(1)/n
8
monoclinic
P2(1)/n
4
triclinic
P1
2
triclinic
P1
2
j
j
a, Å
b, Å
c, Å
7.6165(3)
16.5619(7)
16.8917(5)
90
92.805(1)
90
2128.23(15)
1.350
0.767
11.4832(4)
7.7162(3)
13.3246(5)
90
102.553(2)
90
1152.43(7)
1.327
0.740
7.4623(5)
10.0972(6)
11.7537(8)
88.216(3)
79.002(2)
71.188(2)
822.48(9)
1.213
0.629
5807
2247
0.0173
7.4678(3)
10.2078(7)
15.4909(7)
85.971(3)
85.619(4)
71.602(4)
1115.91(10)
1.144
0.562
5685
3090
0.0244
R, deg
ꢀ, deg
γ, deg
V, A3
dcalc, Mg/cm3
µ, mm-1
total reflections
unique reflections
Rint
7768
3176
0.0130
7792
1863
0.0246
final R indices,
R1, wR2
0.0338, 0.0899
0.0361, 0.0997
0.0353, 0.1029
0.0449, 0.1261
R indices (all data),
R1, wR2
0.0349, 0.0908
0.0384, 0.1015
0.0396, 0.1077
0.0557, 0.1349
phenyl-imidazole interplanar angle, deg
CdO distance, Å
7.2
1.235
3.4
1.222
5.5
1.217
4.7
1.218
group (MeOBDI) is significantly broader and red-shifted
compared to that of the derivative with longer alkoxy chains
(DdecOBDI), while the solid-state intensive emission band of
HexOBDI is in between and apparently possesses peaks from
both solid-state derivatives. The red-shifted emission observed
in the solid state is clearly different from that in solution, because
the absorption and emission spectra are almost independent of
the length of the alkoxy substituent in the latter case. Solid-
state EEM spectra of MeOBDI clearly show a totally different
Ex/Em peak than that in solution. The significant red-shift in
the emission spectrum is presumably due to electronic coupling
resulting from aggregation of molecules, a widely reported
phenomenon.24,39,41,44 The studies on the dependence of the
emission spectra upon dilution with an inert material have shown
that the formation of red-shifted band is due to the formation
of aggregates and is not merely due to planarization. Moreover,
the fluorescence spectrum of the thin films of MeOBDI formed
by rapid cooling is more red-shifted, broader, and unstructured,
indicating strong intermolecular vibronic coupling. Similar
reports19 suggest the existence of the alternate molecular packing
changing from a herringbone-type arrangement in the powdered
crystalline state to a brickstone-type arrangement in thin-film
form. In the brickstone arrangement, the material would result
in a significant enhancement of the overlap of the π orbitals of
Figure 7. (a) Tetramer unit of HOBDI taken from the X-ray structure.
(b,c) Dimer units showing the AB and AA′ interaction. On this and the
following figures, all distances are in angstrom units.
Figure 8. (a) Tetramer unit of MeOBDI taken from the X-ray structure.
(b) Dimer unit showing the AB interaction.
9
668 J. AM. CHEM. SOC. VOL. 131, NO. 2, 2009

Tuning GFP Chromophore Emission
A R T I C L E S
Figure 9. (a) Tetramer unit of HexOBDI taken from the X-ray structure. (b,c) Dimer units showing the AB and AA′ interactions.
Figure 10. (a) Tetramer unit of DdecOBDI taken from the X-ray structure. (b,c) Dimer units showing the AB and AA′ interaction.
the molecules from alternate stacks, leading to significant
increase in electronic coupling between these molecules and
red-shifted emission. A significant number of publications
demonstrate a dramatic role for π-π stacking between the
chromophore and the neighboring aromatic amino acid,45
resulting in the important bathochromic emission shift of the
emission, as a consequence producing the whole family of
yellow fluorescent proteins.2,3,46
In the case of HexOBDI and DdecOBDI, an additional Ex/
Em peak at 420/460 nm, which is only ∼25 nm red-shifted from
the corresponding solution peak, is observed in their EEM
spectra in addition to aggregate peaks. We attribute this band
to the monomeric emission, and its small red-shift may be
related to planarization and rigidization of the molecules in the
solid state.24,47
Since fluorescence emission of the thin films of ROBDI
samples matches closely that of the corresponding crystals, we
conclude that the molecular packing responsible for the lumi-
nescent properties is the same in both cases. The X-ray data
demonstrate that the molecules are stacked in a cofacial manner,
and the dipole moments of the neighboring stacks (AA′ and
BB′) are antiparallel in HOBDI and MeOBDI and parallel in
HexOBDI and DdecOBDI. In contrast to HOBDI, within each
neighboring stack, the molecules of MeOBDI are slightly
9
J. AM. CHEM. SOC. VOL. 131, NO. 2, 2009 669

A R T I C L E S
Dong et al.
translated (slipped) along the long axis and partially minimize
the π-overlap between them, resulting in the highly red-shifted
intense emission band. Alternatively, since the chromophore in
GFP or GFP-like proteins is not consistent with a planar
chromophore in the ground state48 but freely rotating and highly
twisted based on calculation,5 it is intriguing to see if the
fluorescence quenching in solid-state HOBDI is induced by
twisting of the exo-methylene double bond in chromophore,
resulting in efficient IC decay at the conical intersection.43,49
Indeed, we have found that the angle between the phenyl and
imidazole planes is the highest in the HOBDI derivative (7.2°,
Table 2), which may correlate well with its retarded fluorescence.
The molecular packing in the single crystal of longer alkoxy
chain derivatives HexOBDI and DdecOBDI, which unlike
MeOBDI exhibit greenish-yellow and blue fluorescence, re-
spectively, is very different. As in the other cases, the mole-
cules of alternate stacks (AB couples) arrange in a herringbone
fashion. The AB couples in HexOBDI and DdecOBDI are
antiparallel molecules which are close enough to interact
electronically (see Figures 9 and 10). From the crystal structure,
it becomes evident that in these arrangements, the overlap
between the HexOBDI and DdecOBDI chromophore parts of
these structures is minimal because of the interdigitated ar-
rangement. Instead, the main overlap occurs between the
neighboring stacks (AA′ and BB′ couples). The crystal packing
of HexOBDI and DdecOBDI shows that A and A′ (or B and
B′) are translated along both the long and short axes relative to
each other. The molecular number in each unit cell is reduced
in HexOBDI and DdecOBDI as well (see Table 2). As a result
of these factors, electronic interactions between the two
molecules are minimized and excitonic coupling is weak. The
interaction of the neighboring molecules in DdecOBDI is
relatively weak compared to that in HexOBDI, resulting in
dominating monomer-like blue fluorescence. Conversely, analy-
sis of the weak interactions in the crystal structure of HexOBDI
indicates the presence of both the monomer and red-shifted
aggregate emission. It is very interesting to note that, despite
the different crystal structure of all four molecules, the spectral
maxima of their monomer and aggregate emissions are very
close, demonstrating a weak dependence of the excitonic
coupling on the nature of packing. Weakening of the intermo-
lecular coupling in the order OH-C1-C6-C12 corresponds
well with the fluorescence lifetime increase of the aggregate
emission at 570 nm. Also, kinetic analysis of the monomer
emission at 450 nm (Figures 5 and 6, Table 1) reveals the nature
of the blue emission in DdecOBDI vs the yellow-green emission
in HexOBDI. While the monomer lifetimes were the same for
both species, the tailing of the aggregate emission in the low-
wavelength spectral area was much larger for DdecOBDI. This
phenomenon, and not the enhanced monomer emission, caused
the hypsochromic shift of the apparent emission from Dde-
cOBDI crystals. In summary, we have demonstrated that the
color tuning in ROBDI crystals by the alkyl size variation is
regulated by the monomer-aggregate emission ratio. One might
conclude that these existing molecular interactions in solid state
are different from those in the native ꢀ-barrel-isolated chro-
mophore. However, by extending the alkyl chain in ROBDI
derivatives, the emission spectrum of DdecOBDI becomes
similar to that of the isolated, rigid GFP chromophore. Thus,
isolation of the chromophore by effective control of packing
by the alkyl group returns its emission to that of the protein.
Conclusion
Fluorescence “turn-on” occurs in the solid state for synthetic
analogues of green fluorescence protein which are almost
completely nonfluorescent in solutions. 3D steady-state and
time-resolved emission spectroscopies, as well as X-ray mea-
surements, reveal the nature of complex emission in the crystals,
including the fluorescence of monomers and aggregates. The
size of O-alkyl substituents plays a dramatic role in color tuning
of crystalline luminescence. With the increase of alkyl group
from methyl to hexyl to dodecyl, the interaction between the
aromatic molecules in the lattice becomes weaker, resulting in
a hypsochromic shift in the apparent emission from the crystals.
No emission was observed from the hydroxy derivative,
probably due to efficient intermolecular proton transfer. There-
fore, the fluorescence of the GFP chromophores is not easily
reducible to simple questions of flexibility around the ben-
zylidene bond but rather is finely tuned by the subtle balance
among intermolecular forces. We would like to point out that
our observations are not limited only to the GFP, its mutants,
and synthetic chromophores. An application of the GFP fluo-
rophores as model compounds that are quenched due to free
intramolecular rotation in solutions and are bright in the solid
state could be extrapolated to a much wider class of organic
fluorophores. Our work demonstrates an easy way of tuning
their emission.
(44) Davis, R.; Rath, N. P.; Das, S. Chem. Commun. 2004, 74–75.
(45) (a) Kummer, A. D.; Wiehler, J.; Rehaber, H.; Kompa, C.; Steipe, B.;
Michel-Beyerle, M. E. J. Phys. Chem. B 2000, 104, 4791–4798. (b)
Suto, K.; Takenaka, H.; Takenaka, Y. Patent WO 2007086473. (c)
Kikuchi, A.; Fukumura, E.; Karasawa, S.; Mizuno, H.; Miyawaki, A.;
Shiro, Y. Biochemistry 2008, 47, 11573–11580. (d) Ai, H.-W.;
Hazelwood, K. L.; Davidson, M. W.; Campbell, R. E. Nat. Methods
2008, 5, 401–403.
Acknowledgment. We thank the U.S. National Science Foun-
dation (CHE-0456892 and 0809179 to L.M.T. and K.M.S.) for
generous financial support. The authors thank Dr. Veaceslav
Coropceanu for fruitful discussions.
(46) (a) Ormo¨, M.; Cubitt, A. B.; Kallio, K.; Gross, L. A.; Tsien, R. Y.;
Remington, S. J. Science 1996, 273, 1392–1395. (b) Wachter, R. M.;
Elsliger, M. A.; Kallio, K.; Hanson, G. T.; Remington, S. J. Structure
1998, 6, 1267–1277. (c) Griesbeck, O.; Baird, G. S.; Campbell, R. E.;
Zacharias, D. A.; Tsien, R. Y. J. Biol. Chem. 2001, 276, 29188–29194.
(d) Rekas, A.; Alattia, J.-R.; Nagai, T.; Miyawaki, A.; Ikura, M. J. Biol.
Chem. 2002, 277, 50573–50578. (e) Nagai, T.; Yamada, S.; Tominaga,
T.; Ichikawa, M.; Miyawaki, A. Proc. Natl. Acad. Sci. U.S.A. 2004,
101, 10554–10559.
Supporting Information Available: Crystallographic CIF files,
details of synthesis, and analytical and spectral characterization
data of compounds. This material is available free of charge
via the Internet at http://pubs.acs.org.
JA806962E
(47) (a) Da Como, E.; Loi, M. A.; Murgia, M.; Zamboni, R.; Muccini, M.
J. Am. Chem. Soc. 2006, 128, 4277–4281. (b) Nijegorodov, N.;
Zvolinsky, V.; Luhanga, P. V. C. J. Photochem. Photobiol. A 2008,
196, 219–226. (c) Chang, M.-H.; Hoffmann, M.; Anderson, H. L.;
Herz, L. M. J. Am. Chem. Soc. 2008, 130, 10171–10178.
(48) Yang, F.; Moss, L.; Phillips, G. Nat. Biotechnol. 1996, 14, 1246–
1251.
(49) Toniolo, A.; Olsen, S.; Manohar, L.; Martinez, T. J. Faraday Discuss.
2004, 127, 149–163.
9
670 J. AM. CHEM. SOC. VOL. 131, NO. 2, 2009