Syntheses of highly fluorescent GFP-chromophore analogues

Article abstract of DOI:10.1021/ja710388h

Eight B-containing compounds, i.e., 1a-h, were prepared as mimics of the green fluorescent protein (GFP) fluorophore. The underlying concept was that synthetic GFP chromophore analogues are not fluorescent primarily because of free rotation about an aryl-alkene bond (Figure 1b). This rotation is not possible in the β-barrel of GFP; hence, the molecule is strongly fluorescent. In compounds 1a-h, radiationless decay via this mechanism is prevented by complexation of the BF₂ entity. The target materials were prepared via two methods; most were obtained according to the novel route shown in Scheme 1b, but compound 1f was made via the procedure described in Scheme 2. Both syntheses involved formation of undesired compounds E-4a-h that formed simultaneously with the desired isomeric intermediates Z-4a-h. Both compounds form BF₂ adducts, i.e., 1a-h and 5a-h, respectively. Methods used for spectroscopic characterization and differentiation of compounds in the series 1 and 5 are discussed, and these are supported by single-crystal X-ray diffraction analyses for compounds 1c, 5c, 1f, and 5f. Electronic spectra of compounds 1a-h and 5a-h were studied in detail. Those in the 5 series were shown to be only weakly fluorescent, but the 1 series were strongly fluorescent compounds (comparable to the boraindacene, BODIPY, dyes). Compounds 1g and 1h are water soluble, and 1h has particularly significant potential as a probe, since it also has a carboxylic acid group for attachment to biomolecules.

Full text of DOI:10.1021/ja710388h

Published on Web 03/06/2008
Syntheses of Highly Fluorescent GFP-Chromophore
Analogues
Liangxing Wu and Kevin Burgess*
Department of Chemistry, Texas A & M UniVersity, Box 30012, College Station, Texas 77841
Received December 4, 2007; E-mail: burgess@tamu.edu
Abstract: Eight B-containing compounds, i.e., 1a-h, were prepared as mimics of the green fluorescent
protein (GFP) fluorophore. The underlying concept was that synthetic GFP chromophore analogues are
not fluorescent primarily because of free rotation about an aryl-alkene bond (Figure 1b). This rotation is
not possible in the â-barrel of GFP; hence, the molecule is strongly fluorescent. In compounds 1a-h,
radiationless decay via this mechanism is prevented by complexation of the BF2 entity. The target materials
were prepared via two methods; most were obtained according to the novel route shown in Scheme 1b,
but compound 1f was made via the procedure described in Scheme 2. Both syntheses involved formation
of undesired compounds E-4a-h that formed simultaneously with the desired isomeric intermediates Z-4a-
h. Both compounds form BF2 adducts, i.e., 1a-h and 5a-h, respectively. Methods used for spectroscopic
characterization and differentiation of compounds in the series 1 and 5 are discussed, and these are
supported by single-crystal X-ray diffraction analyses for compounds 1c, 5c, 1f, and 5f. Electronic spectra
of compounds 1a-h and 5a-h were studied in detail. Those in the 5 series were shown to be only weakly
fluorescent, but the 1 series were strongly fluorescent compounds (comparable to the boraindacene,
BODIPY, dyes). Compounds 1g and 1h are water soluble, and 1h has particularly significant potential as
a probe, since it also has a carboxylic acid group for attachment to biomolecules.
solution.11,12 They fluoresce at somewhat shorter wavelengths
Introduction
than their parent proteins and with extremely low quantum yields.
Expression of proteins attached to green fluorescent protein
(GFP; Figure 1a) is one of the most widely used strategies for
labeling inside live cells.1-5 Most of the GFP molecule is not
directly useful for the fluoresence of this material; in fact, the
chromophore is relatively small (Figure 1b).6,7 The GFP
chromophore is formed via autocatalytic dehydration of a Ser-
Tyr-Gly tripeptide motif to give an imidazolinone that is then
air-oxidized (Figure 1c).8 Chromophores in GFP analogues are
formed via similar condensation/oxidation routes from other
tripeptides (e.g., blue fluorescent protein, BFP, from Ser-His-
Gly; cyan fluorescent protein, CFP, from Ser-Trp-Gly).1,8
Molecules representing the chromophores of GFP (A9 and
B6) and point mutants (C for Y66F mutant of GFP; D for cyan
fluorescent protein CFP representing the Y66W mutant of GFP;
E for blue fluorescent protein BFP representing the Y66H
mutant of GFP)9 or naturally occurring analogues (F for red
fluorescent protein RFP)10 have been prepared and studied in
This striking difference in quantum yields has been much
discussed in the literature, but the consensus opinion is relatively
simple.13 Loss of fluorescence energy from the synthetic
chromophores in solution is mainly the result of radiationless
transfer arising from free rotation about the aryl-alkene bond
(blue arrows, Figure 2). Isomerization of the alkene (red arrows)
and the polar nature of aqueous media relative to the apolar
environment within the â-barrel protein structures may also be
contributing factors, but the main one is that free rotation
parameter. Steric and electronic factors prevent free rotation of
the aryl substituents when the chromophores are encapsulated
in the proteins. The environment also disfavors E/Z-isomeriza-
tion and provides an apolar medium that is often conducive to
fluorescence.
The research discussed here explores a hypothesis that highly
fluorescent analogues of the GFP chromophore could be made
by including boron. The conceptual genesis of these analogues
1 and the particular compounds that in fact have now been made
are represented in Figure 3. It was supposed that inclusion of
the boron atom would preclude free rotation of the aryl-alkene
bond, so these chromophore analogues would be highly
fluorescent in solution. Further, the structures of the target
(1) Tsien, R. Y. Annu. ReV. Biochem. 1998, 67, 509-544.
(2) Tsien, R. Y. Annu. ReV. Neurosci. 1989, 12, 227-253.
(3) Miller, L. W.; Cai, Y.; Sheetz, M. P.; Cornish, V. W. Nat. Methods 2005,
2, 255-257.
(4) Wang, S.; Hazelrigg, T. Nature 1994, 369, 400-403.
(5) Chalfie, M.; Tu, Y.; Euskirchen, G.; Ward, W. W.; Prasher, D. C. Science
1994, 263, 802-805.
(6) Niwa, H.; Inouye, S.; Hirano, T.; Matsuno, T.; Kojima, S.; Kubota, M.;
Ohashi, M.; Tsuji, F. I. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 13617-
13622.
(11) Dong, J.; Solntsev, K. M.; Poizat, O.; Tolbert, L. M. J. Am. Chem. Soc.
2007, 129, 10084-10085.
(12) Chen, K.-Y.; Cheng, Y.-M.; Lai, C.-H.; Hsu, C.-C.; Ho, M.-L.; Lee, G.-
H.; Chou, P.-T. J. Am. Chem. Soc. 2007, 129, 4510-4511.
(13) Follenius-Wund, A.; Bourotte, M.; Schmitt, M.; Iyice, F.; Lami, H.;
Bourguignon, J.-J.; Haiech, J.; Pigault, C. Biophys. J. 2003, 85, 1839-
1850.
(7) Shimomura, O. FEBS Lett. 1979, 104, 220-222.
(8) Zimmer, M. Chem. ReV. 2002, 102, 759-781.
(9) Kojima, S.; Ohkawa, H.; Hirano, T.; Maki, S.; Niwa, H.; Ohashi, M.;
Inouye, S.; Tsuji, F. I. Tetrahedron Lett. 1998, 39, 5239-5242.
(10) He, X.; Bell, A. F.; Tonge, P. J. Org. Lett. 2002, 4, 1523-1526.
9
10.1021/ja710388h CCC: $40.75 © 2008 American Chemical Society
J. AM. CHEM. SOC. 2008, 130, 4089-4096
4089

A R T I C L E S
Wu and Burgess
Figure 3. Thought process that led to the design of analogues 1 and the
derivatives discussed in this paper.
concerns the syntheses and photophysical properties of com-
pounds 1a-h.
Figure 1. GFP: (a) X-ray structure; (b) chromophore; (c) biogenesis of
the chromophore.
Results and Discussion
Syntheses of the Highly Fluorescent GFP-Chromophore
Analogues. To the best of our knowledge, nearly6 all the
syntheses of GFP-chromophore analogues are based on con-
densation of hippuric or aceturic acid derivatives with aldehydes
(the “Erlenmeyer azlactone synthesis”15).9,12,13,16-18 However,
at least one report indicates these conditions do not work well
for pyrrole-2-carbaldehyde,19 and in fact, attempts to use them
for this substrate were unsuccessful (Scheme 1a). Consequently,
a new strategy was developed that began with the synthesis of
an imidazolinone and then condensation of this with the
pyrrole-aldehyde. That route was successful for compounds
1a-e.
We initially believed the main challenge in the synthesis
described in Scheme 1b was that compounds 4 tended to form
as mixtures of E/Z isomers, but later it was determined that
these materials isomerized under the reaction conditions to form
1 and 5. This conclusion was reached in the following way.
Compound 1c and the corresponding intermediates 4c were
chosen as models for the series. First, evidence was collected
for assignment of E- or Z-stereochemistry for 4c and related
Figure 2. Some synthetic analogues of chromophores in strongly fluores-
cent proteins.
(14) Loudet, A.; Burgess, K. Chem. ReV. 2007, 107, 4891-4932.
(15) Erlenmeyer, E. Justus Liebigs Ann. Chem. 1893, 275, 1-8.
(16) He, X.; Bell, A. F.; Tonge, P. J. FEBS Lett. 2003, 549, 35-38.
(17) Bourotte, M.; Schmitt, M.; Follenius-Wund, A.; Pigault, C.; Haiech, J.;
Bourguignon, J.-J. Tetrahedron Lett. 2004, 45, 6343-6348.
(18) Yampolsky, I. V.; Remington, S. J.; Martynov, V. I.; Potapov, V. K.;
Lukyanov, S.; Lukyanov, K. A. Biochemistry 2005, 44, 5788-5793.
(19) Herz, W. J. Am. Chem. Soc. 1949, 71, 3982-3984.
molecules 1 are reminiscent of the highly fluorescent dyes in
the 4,4-difluoro-4-bora-3a,4a-diaza-s-indacene, or BODIPY
(hereafter abbreviated to BODIPY) class, G.14 Thus, this paper
9
4090 J. AM. CHEM. SOC. VOL. 130, NO. 12, 2008

Highly Fluorescent GFP-Chromophore Analogues
A R T I C L E S
Table 1. Spectroscopic Properties of Z- and E-4c Isomers
Scheme 1. Routes to Target Compounds 1a
3JH6,C5
(Hz)
νC
δH6
(ppm)
δC5
(ppm)
δH1
(ppm)
δH4
′
(ppm)
d
O
′
(cm-
)
1
Z-4c
E-4c
3.8
8.9
1686
1663
7.17
7.33
169.8
168.6
10.62
13.01
5.87
5.99
to compounds like H9 and I18 could have led to isomeric
mixtures, but apparently that was not a problem for those
structures. Thus, the Z-isomer of compound H was shown to
be more stable than the E-form via experimental measure-
ments.16 Similarly, B3LYP calculations for molecule I indicated
that the Z-isomer was thermodynamically preferred.21 However,
the same B3LYP method was applied to compound 4c in the
current work, and this led to the wrong conclusion. Specifically,
the E-isomer was calculated to be more stable both in Vacuo
and in a medium of the same dielectric constant as DMSO (by
1.33 and 0.61 kcal‚mol-1 at 25 °C and by 0.82 and 0.60
kcal‚mol-1 at 100 °C, respectively), whereas NMR experiments
in DMSO-d6 showed that the Z-isomer was in fact the
thermodynamically preferred one. Details of the calculations
are given in the Supporting Information, and spectra for the
thermal isomerization studies are shown in Figure 4a. In fact,
the thermal isomerization in DMSO-d6 at 100 °C was quite slow
(the reaction approaches equilibrium after 51 h, indicative of a
very high-energy barrier to E,Z isomerization). The E,Z ratio
at equilibrium was about 0.6:1.0, indicating that the Z-isomer
was more stable by 0.41 kcal‚mol-1 at 100 °C. Photoisomer-
ization of two different E/Z mixtures of 4c in CDCl3 irradiated
at 360 nm was also studied; the data (Figure 4b) indicate that
the E-isomer is dominant in the photostationary state.
The next step in this project was to make the final products
1 from the intermediates 4. Two isomeric products could be
formed, 1 and 5, and the spectroscopic strategies used to
differentiate them were similar but not identical to those applied
to compound 4c. Coupling between the C6-H proton and the
C5dO carbonyl carbon is complicated by additional spin
pairings involving the B and 19F nuclei. Consequently, that
particular spectroscopic probe is not so conveniently measured;
however, the E-isomer is easily identified by the presence of
coupling between the 19F to C5dO atoms. Further, the carbonyl
stretch for compounds 5 is less than that for 1; in this case, this
is because the boron atom is directly coordinated with the
a (a) Conventional condensation approaches were unsuccessful for
pyrrole-2-carboxaldehyde; (b) the strategy developed here involving
condensation of imidazolinones.
materials.16,20 Table 1 shows selected spectroscopic parameters
for the two isomers of 4c. Coupling between the C6-H proton
and the C5dO carbonyl carbon was probably the most definitive
stereochemical probe; a larger coupling constant was observed
for one isomer that was therefore assigned an E-configuration.
This assignment was supported by observation of a reduced
wavenumber for stretching of the carbonyl group in the putative
E-isomer, as would be expected from internal H-bonding. With
this assignment, chemical shift differences of the protons at H6,
H1′, H4′ and the carbon at C5 were noted as possible indicators
of E- or Z-stereochemistry.
Work by others on syntheses of GFP-chromophore analogues
(via approaches analogous to that shown in Scheme 1a) leading
(20) de Dios, A.; Luz de la Puente, M.; Rivera-Sagredo, A.; Espinosa, J. F.
(21) Nemukhin, A. V.; Topol, I. A.; Burt, S. K. J. Chem. Theory Comput. 2006,
2, 292-299.
Can. J. Chem 2002, 80, 1302-1307.
9
J. AM. CHEM. SOC. VOL. 130, NO. 12, 2008 4091

A R T I C L E S
Wu and Burgess
Figure 4. Isomerization of E,Z-mixtures of compound 4c: (a) thermally in DMSO-d6 at 100 °C; (b) photochemically in CDCl3 at 25 °C and 360 nm
radiation.
Table 2. Spectroscopic Properties of 1c and 5c
νC
δC5
(ppm)
δH6
(ppm)
δH4
′
(ppm)
δF
(ppm)
d
(cm-
O
1
)
Figure 5. Trends in chemical shift differences for compounds 5 relative
to compounds 1 illustrate that these can be used to differentiate between
them. Here, the zero line is the chemical shift values for compounds 1, and
the relative chemical shift difference for the corresponding compound 5 is
indicated. The chemical shift differences for the 19F signals shown here
are 0.1× their actual values.
1c
5c
1705
1610
160.7
158.4
7.41
7.64
6.01
6.28
37.02
34.20
(t, JCF
)
6.1 Hz)
carbonyl CdO only for compounds 5. Single-crystal X-ray
analyses were also obtained for compound 1c and 5c, confirming
the assignment in this case (see below). Table 2 shows the key
spectroscopic parameters for compounds 1c and 5c. In retro-
spect, the magnitude and direction of small chemical shift
differences for these two series of compounds show consistent
trends (Figure 5) that could be used to make tentative assign-
ments for new compounds in the series.
The X-ray structures of molecules 1c and 5c (Figure 6) have
at least two interesting features; one is relevant to the spectro-
scopic assignments made above, and the other relates more to
their fluorescence properties. First, the CdO bond length for
compound 1c is 1.226(3) Å, whereas the corresponding distance
in compound 5c is 1.311(5) Å. The first of these carbonyls is
not complexed, so the bond length is shorter, and the bond is
presumably stronger than in 5c where its bond order is reduced
by interaction with the BF2 entity. Second, the pyrrole-derived
ring in compound 1c is almost exactly coplanar with the one
originating from an imidazolinone; in fact, the dihedral angle
between those two rings is only 0.51°. However, for compound
5c the corresponding dihedral angle is 9.94°. Electronic spectra
of compounds 1 and 5 are described in the next section. Briefly,
compounds 1 tend to absorb at longer wavelengths than their
isomers 5, with higher extinction coefficients, and they also
fluoresce at longer wavelengths. All these properties are
consistent with 1 being the more planar, conjugated structure.
Moreover, fluorescence quantum yields for 5 are, without
9
4092 J. AM. CHEM. SOC. VOL. 130, NO. 12, 2008

Highly Fluorescent GFP-Chromophore Analogues
A R T I C L E S
Figure 6. Top and side views of structures based on single molecule X-ray diffraction analyses of (a) compound 1c and (b) compound 5c.
exception, 10-100 times less than for the corresponding
structures 1. It may be that one contributing factor to this is
rapid equilibration between the two enantiomeric twisted
conformations serving as a pathway for radiationless decay.
In the event, E,Z-isomerization of compounds 4 apparently
does not significantly impact the syntheses of products 1 because
the isomers seem to interconvert under the conditions used in
the final step, as supported by the following observations.
Synthesis of 4c via condensation mediated by HBr/TFA
(condition ii in Scheme 1b) in fact gave E-4c almost exclusively.
However, when this material was converted to the boron-
containing final products, compound 5c and 1c were formed in
a ratio of approximately 2:1. Similarly, when a 1:1 mixture of
E-4c and Z-4c was converted to the boron-containing products
via the same reaction conditions, 5c and 1c were formed in a
ratio of approximately 2:1. It appears that formation of
compound 1c from 4c is a stereorandom process; i.e., little or
no stereochemical information in the starting material is retained.
The pathway outlined in Scheme 1b worked well for
syntheses of compounds 1a-e, but there was a problem when
the same protocol was applied to the 4-nitrobenzene-substituted
product 1f. Specifically, the intramolecular “aza-Wittig” closure
did not afford the desired azlactone 3f; instead, a complex
mixture of products formed. Consequently, the route described
in Scheme 2 was devised. Here, the azlactone was formed via
an intramolecular condensation of the hippuric acid derivative;15
this intermediate was not isolated but instead was condensed
with 3,5-dimethyl-2-pyrrolaldehyde to give the azlactone deriva-
tive 6. Conversion of 6 to the corresponding imidazolinone 4f
was achieved via reaction with methylamine.9 Finally, the BF2
entity was introduced using the same conditions as outlined for
the other compounds in the series in Scheme 1b.
Compounds 1f and 5f were crystallized and studied by X-ray
diffraction. They show the same features as outlined for the
structures of 1c and 5c. Thus, the carbonyl of 1f is shorter than
that in 5f, and the molecular shape is almost perfectly flat
whereas 5f is twisted. Further details are given in the Supporting
Information.
For synthetic convenience in preparing compounds 1, concur-
rent formation of compounds 5 is a nuisance. Conditions to form
compounds 1 with high selectivities were not identified in this
study, but it was shown that the less fluorescent material 5 could
be stripped of the BF2 group, giving intermediates 4 that could
thereby be “recycled” (Scheme 3). Thus, for our most studied
compounds, those in series c, treatment of 5c with sulfuric acid
gave removal of the BF2 group to regenerate intermediate 4c.
If this intermediate is not heated above room temperature or
radiated with relatively intense and energetic light (e.g., 360
nm), then 4c will be isolated as a pure E-isomer. However, as
outlined above, this has no real consequence on the formation
of compound 1c because 5c is invariably also formed.
Review of the literature on BODIPY dyes leads to the
conclusion that there are an abundance of lypophilic probes of
that general type but relatively few water-soluble ones.14
Consequently, there was a strong motivation to prepare the first
water-soluble variants of dyes 1. Compound 1c was chosen as
a starting point because this material was conveniently obtained
in gram amounts and because it contains an aryl iodide
functionality that can also be manipulated. Sulfonation of
compound 1c was achieved by treatment with excess chloro-
sulfonic acid at 0 °C (Scheme 4). Practically, isolation of such
sulfonic acids can be difficult. Here, the crude reaction mixture
was quenched with NaHCO3(aq) and the dye accumulated in the
aqueous phase. The dichloromethane layer was removed, and
9
J. AM. CHEM. SOC. VOL. 130, NO. 12, 2008 4093

A R T I C L E S
Wu and Burgess
Scheme 4. Syntheses of the Water-Soluble Probes 1g and 1h
Scheme 2. Revised Procedure for Synthesis of Compounds 1f
and 5f
Scheme 3. Method To Recycle the Undesired Compounds 5 into
the Target Materials 1
raphy on silica using 10% MeOH in CH2Cl2. Overall, the
extraction and chromatography procedures are relatively easy
because the products are so strongly colored.
Scheme 4 also shows how the sulfonated product 1g was
coupled with 5-hexynoic acid via a Sonogashira reaction.22
Again, the sulfonated product could be isolated via column
chromatography on silica (20% MeOH in CH2Cl2). Compounds
1g and 1h are soluble in aqueous media, and 1h contains a
carboxylic acid functionality that could be used as a point of
attachment to biomolecules. To illustrate this and to measure
the spectroscopic properties of this probe on a protein (see
below), 1h was coupled to streptavidin. Measurement of the
dye/protein ratio via UV23,24 indicated this was almost exactly
3.0:1.0.
UV and Fluorescence Properties of the GFP-Chro-
mophore Analogues. The central hypothesis of this work is
that the fluorescence quantum yields of “unconstrained” GFP
(22) Sonogashira, K.; Tohda, Y.; Hagihara, N. Tetrahedron Lett. 1975, 16,
4467-4470.
the sodium salt of the product was extracted back into an organic
medium (1:1 CH2Cl2/iPrOH). After removal of the solvents, the
crude residue contained over 90% of the desired product. This
material was then further purified by flash column chromatog-
(23) Molecular Probes; Invitrogen Corporation: Carlsbad, CA, 2006. http://
probes.invitrogen.com.
(24) Albarran, B.; To, R.; Stayton, P. S. Protein Eng. 2005, 18, 147-152.
(25) Weber, G.; Teale, F. W. J. Trans. Faraday Soc. 1958, 54, 640-648.
9
4094 J. AM. CHEM. SOC. VOL. 130, NO. 12, 2008

Highly Fluorescent GFP-Chromophore Analogues
A R T I C L E S
Table 3. Spectroscopic Properties of 4c, 1c, and 5c in MeOH
and the isomeric locked product 5c are also given in Table 3.
The UV and fluorescence properties of E-4c are almost identical
to those of Z-4c. The locked compound 5c has a quantum yield
that is approximately 30-fold greater than E-4c, and its absor-
bance and emission wavelengths are also red-shifted. The
quantum yield of 5c is, however, 86-fold less than 1c, possibly
for the reason implied in the discussion of the X-ray structures
of 1c and 5c, i.e., rapid conformational equilibria between ring-
puckered forms leading to radiationless decay pathways. Other
consequences of deviation from planarity for 5c are that the
absorbance is blue-shifted and the extinction coefficient is less.
λmax
ꢀ
λmax
fwhma
(nm)
abs
emiss
1
(nm)
(M-1 cm-
)
(nm)
Φb
Z-4c
1c
E-4c
5c
458
492
458
463
40 200
57 300
48 000
42 700
515
532
515
530
69
54
73
54
0.0005
0.86 ( 0.02
0.0003
0.01
a Fluorescence full width at half-maximum peak height (fwhm). b Fluo-
rescein in 0.1 M NaOH as standard (φ ) 0.92).25
Table 4 gives the UV absorbance and fluorescence data for
all the compounds 1 and 5. The UV and fluorescence properties
of the lypophilic compounds 1a-f and 5a-f are quite similar;
Figure 8 shows illustrative spectra for compounds 1a-f and
5a-f.
The data in Table 4 show that the UV absorbance maxima
for all the lipophilic compounds 1a-f in MeOH are very similar
(λmax abs ) 494 ( 4 nm) except for compound 1a, which has a
methyl rather than an aryl substituent on the core. Measurements
of fluorescence full width at half-maximum (fwhm) peak heights
show that the methyl substituted compound 1a gives the sharpest
fluorescence in the series. Fluorescence emission maxima for
compounds 1b-e are in a 10 nm range (λmax emis ) 526 ( 5
nm). The outliers in terms of fluorescence are 1a (because it
does not have an aryl substituent) and the nitroaryl-substituted
compound 1f. Compound 1f, fluorescing at 598 nm, has a much
larger Stoke’s shift than the others in the series but a much
lower quantum yield. The anomalously low quantum yield is
most probably due to photoinduced electron transfer from the
excited state of the fluorescent core to the LUMO of the nitroaryl
group (i.e., d-PeT). When hexane was used as a medium in
place of methanol, then the quantum yield of the nitroaryl-
substituted compound 1f was only about 33% less than 1a-e
in MeOH. This implies that the d-PeT quenching mechanism
is not dominant in apolar media, reflecting changes of the
oxidation potentials in this molecule. The absorption and
fluorescence maxima of 1a-e were shifted by up to 20 nm to
the red when a less polar media (dichloromethane) was used
(see Supporting Information for an expanded version of Table
4).
Figure 7. UV absorbance and fluorescence of Z-4c and 1c in MeOH (about
10-6 M for UV and 10-6-10-7 M for fluorescence).
analogues like 4 will be greatly increased in compounds such
as 1 that are conformationally locked by a boron atom.
This was verified, and illustrative data are shown in Table 3.
The unconstrained molecule Z-4c has a very low quantum
yield (0.0005) compared to the locked analogue 1c (0.86); this
is a dramatic illustration of the hypothesized effect. Several other
differences in the electronic spectra of these materials were also
observed. The UV absorbance and fluorescence emissions of
compound 1c are both red-shifted relative to Z-4c, and the
fluorescent emission from 1c is sharper (Figure 7). For further
comparisons, data for the trans-unconstrained intermediate E-4c,
Table 4. Spectroscopic Properties of GFP-Chromophore Analogues
1
λmax abs (nm)
ꢀ
(M-1 cm-
)
λmax emiss (nm)
fwhm (nm)
Φa
solvent
1a
5a
1b
5b
1c
5c
1d
5d
1e
1f
1f
5f
5f
1g
1g
1h
1h
1s
477
426
490
463
492
463
493
455
495
492
498
473
486
488
481
488
482
482
57 300
37 000
58 700
50 500
57 300
42 700
51 300
38 600
37 900
44 000
n.d.
485
485
521
517
532
530
521
528
531
598
587
527
545
531
518
538
526
529
19
25
46
64
54
54
38
69
48
120
53
73
97
53
50
58
56
56
0.89 ( 0.01
0.0007
0.87 ( 0.01
0.006
0.86 ( 0.02
0.01
0.85 ( 0.03
0.03
0.80 ( 0.01
0.0004
0.53 ( 0.01
0.005
MeOH
MeOH
MeOH
MeOH
MeOH
MeOH
MeOH
MeOH
MeOH
MeOH
hexanes
MeOH
hexanes
MeOH
phos 7.4b
MeOH
phos 7.4b
phos 7.4b
35 500
n.d.
0.05
48 100
46 500
35 000
34 800
n.d.
0.88 ( 0.01
0.87 ( 0.01
0.84 ( 0.01
0.82 ( 0.01
n.d.
a Fluorescein in 0.1 M NaOH as standard (φ ) 0.92). b 0.1 M lithium phosphate buffer (pH 7.4).
9
J. AM. CHEM. SOC. VOL. 130, NO. 12, 2008 4095

A R T I C L E S
Wu and Burgess
Figure 9. UV absorbance and fluorescence of 1h and the labeled
streptavidin 1s in 0.1 M lithium phosphate buffer, pH 7.4.
significantly different from that of the parent dye; in other words,
conjugation to this particular protein does not have significant
effect on the fluorescent properties.
Conclusions
Compounds 1 in this paper are conceptual hybrids of GFP-
chromophore analogues and BODIPY dyes. The data presented
prove that addition of the BF2 entities to the open-chain
intermediates 4 greatly increases the quantum yields across a
range of compounds. Overall, this provides more evidence for
the assertion that GFP chromophores within the protein have
enhanced quantum yields due to conformational locking within
the â-barrel cocoon.
The fluorescence properties of compounds 1 tend to be very
similar to those of BODIPY dyes. Both compounds are neutral,
uncharged, highly fluorescent molecules that absorb with high
extinction coefficients and emit at around 520-530 nm (unless
high conjugating substituents are added). BODIPY dyes have
many applications in chemistry and biology. Probes 1 could
presumably be used in the same ways, and there may be some
advantages to doing this, in terms of intellectual property at
the very least. Further, the data presented here show that dyes
1 can be sulfonated and modified to include a point of
attachment to biomolecules. The water-soluble modifications
to compounds 1 result in retention of their fluorescent properties
in aqueous media and when attached to a model protein
(streptavidin).
Acknowledgment. Financial support for this work was
provided by The National Institutes of Health (Grant GM72041)
and by The Robert A. Welch Foundation. TAMU/LBMS-
Applications Laboratory headed by Dr. Shane Tichy provided
mass spectrometric support, the Laboratory for Molecular
Simulation (Dr. Lisa Thompson) supported our molecular
simulations work, and the X-ray Diffraction Laboratory (Drs.
J. Reibenspies and N. Bhuvanesh) generated the crystallographic
data.
Figure 8. UV absorbance and fluorescence spectra for (a) GFP-chro-
mophore analogues 1 in MeOH and (b) GFP-chromophore analogues 5 in
MeOH. Concentrations throughout are 10-6 M for UV and 10-7 M for
fluorescence.
Table 4 also gives data for the water-soluble dyes 1g, 1h,
and 1s. The absorbance and emission maxima for 1g and 1h
were slightly blue-shifted in phosphate buffer relative to the
same dyes (and to 1b-e) in MeOH. Quantum yields for 1g
and 1h were also high in phosphate buffers, just as in MeOH.
The spectra of compound 1s, i.e., 1h on streptavidin, were not
Supporting Information Available: Procedures and charac-
terization data for all the new compounds and crystallographic
files in CIF format. This material is available free of charge
via the Internet at http://pubs.acs.org.
JA710388H
9
4096 J. AM. CHEM. SOC. VOL. 130, NO. 12, 2008