Spectroscopic and Theoretical Study on Electronically Modified Chromophores in LOV Domains: 8-Bromo- and 8-Trifluoromethyl-Substituted Flavins

Article abstract of DOI:10.1002/cbic.201200670

Two chemically synthesized flavin derivatives, 8-trifluoromethyl- and 8-bromoriboflavin (8-CF₃RF and 8-BrRF), were photochemically characterized in H₂O and studied spectroscopically after incorporation into the LOV domain of the blue light photoreceptor YtvA from Bacillus subtilis. The spectroscopic studies were paralleled by high-level quantum chemical calculations. In solution, 8-BrRF showed a remarkably high triplet quantum yield (0.97, parent compound riboflavin, RF: 0.6) and a small fluorescence quantum yield (0.07, RF: 0.27). For 8-CF₃RF, the triplet yield was 0.12, and the fluorescence quantum yield was 0.7. The high triplet yield of 8-BrRF is due to the bromine heavy atom effect causing a stronger spin-orbit coupling. Theoretical calculations reveal that the decreased triplet yield of 8-CF₃RF is due to a smaller charge transfer and a less favorable energetic position of T₂, required for intersystem crossing from S₁ to T₁, as an effect of the electron-withdrawing CF₃ group. The reconstitution of the LOV domain with the new flavins resulted in the typical LOV photochemistry, consisting of triplet state formation and covalent binding of the chromophore, followed by a thermal recovery of the parent state, albeit with different kinetics and photophysical properties. Vision in blue: Flavin derivatives, substituted at position 8 (R=bromo-, trifluoromethyl-) were synthesized and incorporated as chromophores into the blue-light-sensing LOV domain of YtvA from Bacillus subtilis. The resultant strong change in electronegativity significantly modifies the electronic and functional properties of the reconstituted LOV domains.

Full text of DOI:10.1002/cbic.201200670

CHEMBIOCHEM
FULL PAPERS
DOI: 10.1002/cbic.201200670
Spectroscopic and Theoretical Study on Electronically
Modified Chromophores in LOV Domains: 8-Bromo- and
8-Trifluoromethyl-Substituted Flavins
Madina Mansurova,*[a] Julian Simon,[a] Susanne Salzmann,[b] Christel M. Marian,[c] and
Wolfgang Gꢀrtner*[a]
Two chemically synthesized flavin derivatives, 8-trifluorometh-
yl- and 8-bromoriboflavin (8-CF3RF and 8-BrRF), were photo-
chemically characterized in H2O and studied spectroscopically
after incorporation into the LOV domain of the blue light pho-
toreceptor YtvA from Bacillus subtilis. The spectroscopic studies
were paralleled by high-level quantum chemical calculations.
In solution, 8-BrRF showed a remarkably high triplet quantum
yield (0.97, parent compound riboflavin, RF: 0.6) and a small
fluorescence quantum yield (0.07, RF: 0.27). For 8-CF3RF, the
triplet yield was 0.12, and the fluorescence quantum yield was
0.7. The high triplet yield of 8-BrRF is due to the bromine
heavy atom effect causing a stronger spin–orbit coupling. The-
oretical calculations reveal that the decreased triplet yield of 8-
CF3RF is due to a smaller charge transfer and a less favorable
energetic position of T2, required for intersystem crossing from
S1 to T1, as an effect of the electron-withdrawing CF3 group.
The reconstitution of the LOV domain with the new flavins re-
sulted in the typical LOV photochemistry, consisting of triplet
state formation and covalent binding of the chromophore, fol-
lowed by a thermal recovery of the parent state, albeit with
different kinetics and photophysical properties.
Introduction
LOV (light, oxygen, voltage) domain proteins, such as YtvA of
Bacillus subtilis, are members of a large blue-light-sensitive
photoreceptor family. Their plant counterparts are called pho-
totropins.[1,2] The response towards light in a LOV domain de-
pends strongly on the photophysical properties of the incorpo-
rated chromophore, flavin mononucleotide (FMN). Light excita-
tion generates the triplet state, which, during its decay over
a few ms, forms a covalent bond to the protein through the
side chain of a closely positioned cysteine residue (Cys62 in
YtvA-LOV), concomitantly with a rearrangement of the hydro-
gen bonding network in the chromophore pocket.[3–5] This
photoadduct (LOV390) is considered the signaling state of YtvA.
In the dark, the covalent bond in LOV390 is slowly reopened
over several hours, yielding the initial (parent) state of the pho-
toreceptor (LOV447).[6]
cal/photochemical properties of these biological photorecep-
tors are now being intensely investigated.[10] The main strat-
egies for controlling the signaling mechanism are through
changes in the protein by site-directed mutations or through
replacement of the native chromophore with chemically syn-
thesized or modified flavins. The objective of the latter ap-
proach is based on the assumption that a new functional
group at a modified chromophore might change its electronic
properties and also might interfere with the hydrogen bonding
network between the chromophore and the protein. As well as
the natural chromophore (FMN), riboflavin (RF) and (in small
amounts) flavin adenine nucleotide (FAD) have also been iden-
tified in LOV domains.[11] In the dark state, the isoalloxazine
ring of flavin is involved in many interactions between the
chromophore and protein that are important for regulation of
the kinetics of the LOV domain photocycle, whereas the phos-
phate group at the ribityl moiety establishes electrostatic inter-
actions with the side chains of two arginine residues on the
surface of the protein.[12]
LOV domains have recently attracted attention because of
their very versatile properties, which make them interesting
candidates for potential biotechnological applications.[7–9] Ac-
cordingly, effects of structural modifications of the photophysi-
Here we report the synthesis and photophysical properties
of two electronically modified flavins (Scheme 1)—8-(trifluoro-
methyl)riboflavin (8-CF3RF) and 8-bromoriboflavin (8-BrRF)—in
H2O and after incorporation into the LOV domain of YtvA from
B. subtilis, in combination with high-level quantum chemical
calculations. There are still very few reports on the use of flavin
derivatives as chromophores in LOV domains. One interesting
case was the in vivo incorporation of roseoflavin (8-dimethyla-
mino FMN), which showed a strong bathochromic shift of its
absorbance band (lmax =520 nm).[13] However, those authors
demonstrated that roseoflavin did not form a covalent bond
[a] Dr. M. Mansurova, J. Simon, Prof. Dr. W. Gꢀrtner
Max Planck Institute for Chemical Energy Conversion
Postfach 10135, 45410 Mꢁlheim an der Ruhr (Germany)
E-mail: wolfgang.gaertner@cec.mpg.de
[b] Dr. S. Salzmann
Department of Chemistry, Aarhus University
Langelandsgade 140, 8000 Aarhus C (Denmark)
[c] Prof. Dr. C. M. Marian
Institute of Theoretical and Computational Chemistry
Heinrich-Heine University Dꢁsseldorf
Universitꢀtsstrasse 1, 40225 Dꢁsseldorf (Germany)
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 645 – 654 645

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
quired positions[22,23] or on multistep synthesis from appropri-
ately substituted aromatic building blocks as starting materi-
als.[24] However, the synthesis of 8-halogenated flavins is not
straightforward, because halogen substituents reduce the nu-
cleophilicities of the amine functions of the meta-halogenated
aryl systems from which the synthesis starts. On the other
hand, selective introduction of halogens into the complete iso-
alloxazine ring is rather troublesome due to the presence of
many competing functional groups in this complex molecule.
Here, the synthesis of the both 8-substituted chromophores
was performed in similar manner to published synthetic routes
(Scheme 2, for details see the Experimental Section).[24,25] The
previous authors reported the protection group chemistry ap-
plied here in great detail. In brief, the free OH groups in the ri-
bitylated anilines 2[26] were protected by treatment with 2,2-di-
methoxypropane, and the resulting compounds 3 were al-
lowed to react with 2,4,6-trichloropyrimidine, yielding mixtures
of isomers with compounds 4 as major components. Substitu-
tion on the pyrimidine ring with benzyloxy functions, followed
by hydrogenation (Pd/C), resulted in intermediates 6 in good
yield. The completion of the isoalloxazine ring (ring closure re-
action) was performed by treatment with sodium nitrite[27] in
the presence of acetic acid, yielding the N-oxides 7. The de-
sired compounds 9 were then obtained by reduction with
sodium thiosulfite, followed by deprotection of the ribityl OH
groups. The overall yields were 3% and 4.3% for 8-CF3RF and
8-BrRF, respectively. These apparently moderate yields should
be seen in the context of the reaction sequence of eight reac-
tion steps. In addition, 3!4 involves the formation of mixtures
of isomers of which only one (with the correct steric configura-
tion) could be used in each case for further synthesis; further-
more, the substituent change (chloride to benzyl, 4!5) and
the deprotection chemistry had to be processed for both posi-
tions. For the good yields of the individual steps see the Exper-
imental Section.
Scheme 1. Structural formulas of riboflavin and its halogenated derivatives.
The corresponding lumiflavin compounds each carry a methyl group at N10.
(i.e., a photoadduct) with the protein. We were recently able
to show that the naturally occurring flavin chromophores RF,
FMN, and FAD, bearing differently substituted ribityl chains,
differ only slightly[14] in their photophysical characteristics, with
the largest variations in the thermally driven dark recovery ki-
netics; accordingly, the RF forms of the modified flavins were
used in the experiments reported here. Both substituents
should disturb the steric interactions with the protein cavity
only marginally, because their van der Waals radii are only
slightly larger than that of the substituted methyl group. How-
ever, the introduction of a bromine atom should increase elec-
tron density in the flavin and modify the triplet formation yield
strongly, whereas the trifluoromethyl group should have
a strongly electron-withdrawing effect. Selection of these sub-
stituents and their position was motivated by previous findings
that indicated that the transition dipole moment extends from
position 8 of the isoalloxazine ring system to position 3.[15,16]
Both in vivo[13,17] and in vitro[12,14,18] chromophore exchange
procedures are available; here we followed our recently report-
ed protocol for chromophore exchange, based on unfolding of
the wild-type (WT) protein, removal of the native chromo-
phore, and introduction of a modified flavin upon refolding of
the protein into its native conformation.[14] The quantum
chemical calculations performed here explain the observed ex-
cited state behavior (these calculations were performed for the
lumiflavin forms, i.e., the sugar-free forms of the modified fla-
vins).
Spectroscopic properties of 8-CF3RF and 8-BrRF in buffered
solution (pH 7)
Both flavins had typical flavin double-peaked absorption spec-
tra (Figure 1). The spectrum of 8-BrRF was essentially identical
to that of riboflavin, also with respect to its extinction coeffi-
cient (lmax =450 nm, emax =12460mꢀ1 cmꢀ1). This finding is in
excellent agreement with the theoretical calculations, which
yield vertical absorption energies of 8-BrLF that closely match
those of LF. Interestingly, the S2 transition of 8-CF3RF showed
a hypsochromic shift (330 nm, compared to 360 nm for the
parent compound, RF, or for 8-BrRF, vide infra), and this com-
Results and Discussion
The two flavins reported here were designed especially be-
cause of the strong electronic changes induced by the halo-
genated substituents at position 8, because calculations reveal
the direction of both transition dipole moments from the
phenyl ring to the uracil component. In particular, the
S2 transition dipole moment is oriented between positions 8
and 3. In addition, the bromine should exert a strong heavy-
atom effect on the fluorescence/triplet reaction channels such
that the triplet yield should increase at the expense of the
fluorescence.
pound also exhibited a lower extinction coefficient (lmax
=
450 nm, emax =8450mꢀ1 cmꢀ1, Table 1). The theoretically calcu-
lated vertical absorption energies of 8-CF3LF show convincing
agreement with the measured absorbance band maxima and
reveal the origin of the hypsochromic shift mentioned above.
For the parent compound, the energetic position of the
second absorption band is known to be solvent-dependent,
ranging between 332 nm (3.73 eV) in dioxane and 367 nm
(3.38 eV) in water.[28] The underlying electronic transition is ac-
Chemical synthesis
Several methods for the synthesis of flavin derivatives have
been reported;[19–21] they are based either on displacement re-
actions on flavins containing good leaving groups at the re-
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 645 – 654 646

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
Scheme 2. Synthesis of 8-substituted flavins. a) d-Ribose, NaCNBH3; b) 2,2-dimethoxypropane/acetone, TsOH; c) 2,4,6-trichloropyrimidine, K2CO3, Bu4NBr;
d) PhCH2ONa; e) H2/Pd; f) NaNO2, AcOH; g) Na2S2O3; h) TFA.
state is preferentially stabilized by polar solvents. The more
polar and protic the surrounding medium, the more this band
shifts to lower energy. Our calculations reveal that for the tri-
fluoromethyl derivative even the electronic transition in the
isolated molecule requires higher excitation energy. In the
parent compound all heteroatoms are concentrated in one
half of the molecular scaffold, whereas in the trifluoromethyl
compound, polar heteroatoms with increased electronegativity
are added on the aliphatic half of the molecule. A transfer of
charge density from the benzene to the pteridine moiety, as
takes place in the excited state, is thus less favorable. In addi-
tion, the impact of the solvent on the energetic position of the
absorbance band in question is less than it is in the parent
Figure 1. UV/VIS spectra of 8-BrRF (dotted line) and 8-CF3RF (solid line) in
compound. The origin of this variation can be explained by
the significantly decreased dipole moment of the 8-CF3LF.
H2O, pH 7. The absorption spectrum of 8-BrRF fully coincides with that of ri-
boflavin.
Fluorescence and triplet state
properties of halogenated fla-
vins
Table 1. Spectroscopic properties of 8-substituted riboflavins and of the parent compound (RF) in water, pH 7.
In addition, calculated vertical excitation wavelengths are listed.
labs (max) [nm] labs (calcd) [nm] emax [mꢀ1 cmꢀ1
]
lem (max) [nm] tF [ns]
FF
The fluorescence maxima for
both synthetic flavins were simi-
lar to that for RF (520 nm, H2O).
However, both 8-CF3RF and 8-
BrRF significantly differed from
RF
8-CF3RF
8-BrRF
360, 450
330, 450
360, 450
357, 432
333, 435
359, 433
12460
8450
12460
520
520
520
4.8
4.9
0.27
0.70
3.5 (85%), 7.0 (15%) 0.07
companied by a transfer of charge density from the benzene
to the pteridine moiety of the isoalloxazine ring, thus increas-
ing the molecular dipole moment with respect to the ground
state.[29] It is thus not surprising that the electronically excited
RF in their fluorescence quantum yields (0.7 and 0.07, respec-
tively, in comparison with 0.27 for RF[30]). Time-resolved fluores-
cence measurements yielded a biexponential decay for 8-BrRF
[tF1 =3.5 ns (85%), tF2 =7.0 ns (15%)], whereas the decay of 8-
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 645 – 654 647

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
CF3RF was monoexponential (tF =4.9 ns) and in the same
range as that of the parent compound (Table 1).
Table 2. Transient absorption properties of modified flavins (triplet
decay).
Study of the triplet parameters revealed significant variations
with respect to those of the parent compound: both modified
riboflavins showed broad transient absorptions with maxima at
520 and at approximately 700 nm, similarly to RF (Figure 2,
Table 2),[31] but the quantum yield determined for 8-CF3RF
(FT =0.12, with a decay time tT of 10 ms) was much smaller
lT (max) [nm]
tT [ms]
FT
RF
8-CF3RF
8-BrRF
520, 700
520,[a] 650
520, 700
29
10
5
0.60
0.12
0.97
[a] Very weak signal.
the chromophore exchange protocol with naturally occurring
flavins (RF, FMN, FAD), and also with the alkyl-modified flavins
in the former work;[14] the halogenated flavin chromophores
were lost in relatively short times, causing the apoprotein to
aggregate and precipitate. Because the riboflavin forms were
used here, one might speculate that interaction was less
stable, due to the lack of the phosphate group. Previous ex-
periments,[14] however, revealed no destabilizing effect from
the removal of the phosphate group: a riboflavin reconstituted
LOV domain showed no change in its spectral properties, even
when kept for several weeks at 48C.
Figure 2. Transient spectra of 8-CF3RF (solid line) and 8-BrRF (dotted line) in
H2O, pH 7. Compound 8-CF3RF: interval 1.5–3.5 ms; 8-BrRF: 0.5–1.5 ms.
Nevertheless, both 8-CF3LOV and 8-BrLOV showed the func-
tional properties of the LOV photocycle: an absorption maxi-
mum around 450 nm, the photoproduct formation after illumi-
nation with blue light, and the thermally driven dark recovery
(Figure 3). The hypsochromically shifted S2 absorption band of
8-CF3 seen in aqueous solution was identically present in the
protein-incorporated form, allowing in this case the identifica-
tion of three isosbestic points during the generation of the
than that for RF (FT =0.6, tT =29 ms).[31;32] In earlier work on RF,
3
we found that the second pp* state (T2) acts in aqueous solu-
tion as a doorway state for intersystem crossing from the S1 to
the T1 state.[29] Like its singlet counterpart that gives rise to the
second band in the absorption spectrum, the T2 state of 8-
CF3RF experiences a smaller solvent shift than that of the
parent compound. It may there-
fore be assumed that the cross-
ing point between the S1 and T2
potential energy surfaces of 8-
CF3RF occurs at higher energies
with respect to the S1 minimum
than in RF, thus explaining the
smaller triplet quantum yield. In
contrast, the quantum yield of 8-
BrRF was remarkably high: FT =
0.97 (decay time=5 ms), which is
clearly attributable to the heavy-
atom effect of the introduced
bromine.
Photocycle and photophysical
properties of 8-CF3RF and 8-
BrRF in LOV
Both 8-substituted chromo-
phores were successfully incor-
porated into the blue light pho-
toreceptor YtvA by the chromo-
phore exchange protocol report-
ed previously.[14] The newly re-
constituted
proteins
were
Figure 3. Photoproduct formation in 8-CF3LOV (top left) and in 8-BrLOV (bottom left) upon blue light exposure.
The insets show the difference spectra between the lit and the dark states. Dark recovery kinetics are shown on
relatively unstable relative to
LOV domains reconstituted by the right for 8-CF3LOV (top) and for 8-BrLOV (bottom) with the exponential fit curves given in white.
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 645 – 654 648

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
photoproduct (lmax =370 nm, Figure 3). Quantitatively, the for-
mation of the photoproduct was calculated for the first 30 s of
irradiation and then after saturating blue light irradiation, by
using YtvA WT as reference. Dark recoveries were monitored at
lmax =445 nm for both reconstituted proteins. The instability of
reconstituted 8-CF3LOV mentioned above, however, impeded
complete dark recovery. The summarized data for the protein
photochemistry of 8-CF3LOV and 8-BrLOV are presented in
Table 3.
Table 3. Photocycle parameters of LOV-reconstituted flavins.
Photoproduct formation after
[b]
tph[a] [min] trec [min] 30 s [%]
sat. irradiation [%]
YtvA WT
8-CF3RF LOV 6.5
8-BrRF LOV
6.5[d]
7
129
116[c]
86
27
45
8
100
100
15
[a] Time of photoproduct formation. [b] Dark-state recovery kinetics.
[c] The recovery was not complete, probably due to partial denaturation
of the sample during the experiment. [d] There were no further changes
after the indicated time and the complete lit state was not achieved.
A very interesting behavior was observed for 8-BrLOV.
Whereas the S2 absorption band for this compound was very
pronounced when measured in H2O (lmax =360 nm, Figure 1),
in the protein-bound state only the S1 absorption band was
clearly detected (Figure 3). In addition, this S1 absorption band
showed a peculiarity in that the high-energy shoulder of the
450 nm band is of very low intensity, in contrast to the
“normal” flavin chromophores that exhibit a very characteristic
“three-peaked” 450 nm absorbance band. Interestingly, blue
light irradiation of 8-BrLOV yielded a very limited bleaching of
the absorption at 450 nm, yet the increase at 390 nm is clearly
present (Figure 3). Estimation of the vertical excitation energies
of 8-BrLOV shed light on this behavior. As elucidated in earlier
work,[33] two distinct conformations (confA and confB) at the
sulfur atom of Cys62 have been detected in the X-ray structure
of YtvA-LOV, and these give rise to slightly different vertical ab-
sorption spectra. The calculated difference in the vertical exci-
tation energies is most pronounced for the excitation that
yields the S2 band and results from the conformation-depen-
dent interaction of the sulfur lone-pair orbital of Cys62
(Figure 4). In both conformers, remarkable red shifts of this
band relative to the water-solvated cofactor are observed. In
confA, in which the sulfur atom of Cys62 is located above the
benzene moiety of the isoalloxazine ring, the stabilization of
the upper electronic state is particularly strong. It can therefore
be assumed that the unusual shape of the measured absorp-
tion spectrum is due to merging of the S1 and S2 bands.
Figure 4. Comparison between experimentally measured absorption spectra
of 8-BrRF/8-BrLOV and DFT/MRCI vertical excitation energies and intensities
of the lumiflavin derivatives. The top panel shows the results for aqueous
solution and the bottom panel for the protein environment. The calculated
vertical excitation energies (bottom) correspond to the two different con-
formers of the 8-BrLOV complex (confA: gray lines; confB: light gray lines).
For further explanation see text.
Table 4. Spectroscopic properties and triplet decays of 8-CF3RF and 8-
BrRF in the LOV domain.
tF [ns]
FF
tT [ms]
YtvA WT
8-CF3LOV
8-BrLOV
2.2 (85%), 4.6 (15%)
3.3 (70%), 0.7 (30%)
0.5 (81%), 3.4 (19%)
0.22
1.7
1.7
3.6
0.10[a]
0.07
[a] The protein was not stable over longer times (i.e., a decrease in pro-
tein amount by 40% was found upon repetition of the measurement
after 24 h).
(which would be indicative of chromophore release) was de-
tected upon blue light irradiation. Moreover, the observed dis-
tribution (3.3 ns (70%) and 0.6 ns (30%); Table 4) was the op-
posite of the fluorescence decay of 8-CF3LOV; this could lead
to the conclusion that the 19% component in 8-BrLOV (3.4 ns)
might be the active 8-BrLOV species. If it is assumed that this
fraction of the chromophore is the active one, the typically ob-
served acceleration of the fluorescence decay in the protein-
bound form is not detected. Still, both lifetimes of 0.5 and
3.4 ns are within the reported range of LOV proteins.[34–37]
Another possible explanation for the low level of conversion
to the photoproduct could be steric hindrance introduced by
The reasons for the reduced yield of photoconversion are
not clear. Careful observation of the fluorescence lifetime iden-
tified a biexponential decay, with one species (ꢁ19%) show-
ing tF =3.4 ns (Table 4), identical to that of the major species
in the fluorescence lifetime decay of the unbound 8-BrRF (in
H2O, pH 7). However, the absence of the S2 band is clear evi-
dence for the bound chromophore, and no signal at 360 nm
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 645 – 654 649

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
the bulkier bromine atom or, alternatively, its electronic influ-
ence on the surrounding amino acid network in the chromo-
phore pocket, possibly resulting in an increased distance be-
tween the C-4a position of the isoalloxazine ring and the reac-
tive Cys62 residue. These steric or electronic restrictions in the
chromophore pocket might also contribute to the low stability
of the protein–chromophore complex.
effects of the introduced halogen substituents on the electron-
ic properties of the flavin chromophore and reveal different
types of interactions between the chromophore and its sur-
rounding amino acid network.
Experimental Section
The data in Tables 3 and 4 do not allow for a discussion in
quantitative terms with comparison to time-resolved triplet
state determination. However, the extent of photoadduct for-
mation after 30 s is higher for the trifluoromethyl derivative
than for the WT protein. This is even more surprising if its
lower extinction coefficient and, in addition, the very low trip-
let yield in water are taken into account. This interpretation
would point to an additional reaction channel involving the
singlet state, as already proposed for the LOV2 domain of the
alga Chlamydomonas reinhardtii.[38]
Synthesis of 8-substituted flavins
General: All chemicals were obtained from Sigma–Aldrich, except
for 3-isopropyl-4-methylaniline, which was purchased from Daxian
Chemical Institute, Ltd. All moisture- and oxygen-sensitive reac-
tions were carried out under inert atmosphere (Ar). All reported
yields are for isolated pure products. Flash chromatography (FC)
purifications were performed on Merck silica gel 60 (230–
400 mesh). HPLC purifications were carried out on Kromasil C18
ODS-5-100 (RP-C18, 5 mm, 250 mmꢂ21 mm, flow rate 0.8 mLminꢀ1),
with elution with systems A (H2O+2% CH3CN+0.05% TFA) and B
(30:70 H2O+CH3CN+0.05% TFA), if not indicated otherwise. The
NMR spectra were recorded with a Bruker AVANCE DRX 400 spec-
trometer. Chemical shifts are given as d values (ppm) downfield
relative to tetramethylsilane as an internal standard. MS characteri-
zation was carried out with a Finnigan MAT 8200 (EI) or Bruker Es-
quire 3000 (ESI) instrument; exact mass determinations (HRMS)
were performed with a Finnigan MAT 95 instrument.
The bromo derivative has different properties, if the finding
of a large unreactive fraction is taken into account. Only 15%
of the sample could converted even under saturating irradia-
tion conditions (Table 3). If this 15% fraction is assumed to be
the functionally active species, we would already have convert-
ed more than 50% (8% of 15%) after 30 s of irradiation. This
calculation would indeed—especially for the bromo deriva-
tive—demonstrate the correlation of an increased triplet yield
and, in parallel, increased photoadduct formation.
General procedure for riboaniline formation: Compound 1
(1 mmol) was dissolved in MeOH (10 mL), and NaCNBH3 (2 equiv)
was added with stirring, followed by d-ribose (2 equiv). The result-
ing mixture was stirred under reflux for 43–48 h, the solvent was
then removed under reduced pressure, and the remaining material
was added to HCl (1m, 2 mL) and swirled until gas generation fin-
ished. The solution was then carefully neutralized with saturated
NaHCO3, extracted into EtOAc (3ꢂ25 mL), dried over Na2SO4, and
concentrated. FC on silica gel, with elution with CH2Cl2/MeOH
(90:10), provided the product 2.
Conclusions
Both, 8-CF3RF and 8-BrRF were synthesized in good overall
yields and could be incorporated into the LOV domain of
B. subtilis photoreceptor YtvA. Interestingly, the positions and
shapes of the absorption bands, the fluorescence properties,
and the triplet lifetimes and yields of the flavin derivatives in
water are significantly different from those of RF: the intro-
duced bromine substituent produces a remarkably high triplet
quantum yield, whereas the properties of the trifluoromethyl
compound are similar to those of the parent compound. Also
in the protein-incorporated form, both flavin derivatives
showed significant changes in photochemical behavior. Espe-
cially strong effects were observed for 8-BrRF, for which the S2
state absorption was practically absent and the high-energy
shoulder of the S0 absorption band was significantly less pro-
nounced. No such strong effects were found in the case of the
CF3 substituent; this chromophore showed properties very sim-
ilar to those of the RF-assembled protein, except that the
S2 absorption appears at higher energy. The large changes in
triplet yields (much higher for the bromo derivative and lower
for the trifluoromethyl derivative) observed in water could not
be identified similarly clearly in the protein-bound forms, due
to the relatively unstable derivative–LOV domains. However,
the apparently higher yield of photoproduct formation of the
trifluoromethyl flavin might be indicative of (with all reluc-
tance) an alternative pathway for photoadduct formation to
that of the WT-LOV under identical conditions. Site-directed
mutagenesis might help in clarifying this preliminary observa-
tion. All these findings are evidence for the strong modifying
3-Bromo-4-methylriboaniline (2a): Yield: 78%; 1H NMR (400 MHz,
CD3OD): d=7.20 (d, 1H; Ar), 6.62 (s,1H; Ar), 6.43 (d, 1H; Ar), 4.81
(brs; 4OH), 3.91 (m, 1H; 5’-CHH), 3.76 (m, 2H; 5’-CHH, 2’-CH), 3.65
(m, 2H; 3’-CH, 4’-CH), 3.30 (dd, 1H; 1’-CHH), 3.13 (m, 1H; 1’-CHH),
2.25 ppm (s, 3H; CH3); 13C NMR (100 MHz, CD3OD): d=149.8, 138.9,
133.4 (Ar), 116.5, 113.6, 112.0 (Ar), 74.7 (C-3’), 74.6 (C-2’), 72.2 (C-4’),
64.6 (C-5’), 47.6 (C-1’), 23.1 ppm (CH3); MS (EI): m/z: 319 [M], 198
[MꢀCH2OH(CHOH)3].
3-Trifluoromethyl-4-methylriboaniline (2b): Yield: 85%; 1H NMR
(400 MHz, CDCl3): d=7.05 (d,1H; Ar), 6.93 (d, 1H; Ar), 6.77 (dd, 1H;
Ar), 4.80 (brs; 4OH), 3.91 (m, 1H; 5’-CHH), 3.80–3.72 (m, 2H; 5’-
CHH, 4’-CH) 3.66–3.61(m, 2H; 2’-CH, 3’-CH), 3.44 (dd, 1H; 1’-CHH),
3.15 (dd, 1H; 1’-CHH), 2.29 ppm (s, 3H; CH3); 13C NMR (100 MHz,
CDCl3): d=148.5, 133.7 (Ar), 130.2, 129.9 (CF3), 127.7, 124.6, 117.0,
111.3 (Ar), 74.6 (C-3’), 74.4 (C-2’), 72.2 (C-4’), 64.4 (C-5’), 47.3 (C-1’),
18.3 ppm (CH3); MS (EI): m/z: 309 [M], 188 [MꢀCH2OH(CHOH)3].
General procedure for OH protection: A riboaniline 2 (1 mmol)
was dissolved in a mixture of dimethoxypropane/dry acetone (1:2,
12 mL), and TsOH·H2O (1.0 equiv) was added. The reaction mixture
was stirred at RT for 1.5 h, and the solvents were then removed in
vacuo. CH2Cl2 (20 mL) was added, and the resulting mixture was
washed with saturated NaHCO3 (5 mL), followed by H2O (5 mL). FC
on silica gel with elution with CH2Cl2!CH2Cl2/Et2O (99:1) resulted
in the desired 2,3:4,5-bis(methylethylidene)riboaniline and in
2,5:3,4-bis(methylethylidene)riboaniline as a side product.
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 645 – 654 650

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
3-Bromo-4-methyl [2,3:4,5-bis(methylethyliden)ribo]aniline (3a): Yield:
45%; 1H NMR (400 MHz, CDCl3): d=7.00 (d, J=8.5 Hz, 1H; H-5),
6.89 (m, 1H; H-2), 6.57 (m, 1H; H-6), 4.37 (m, 1H; 2’-CH), 4.11 (m,
2H; 3’-CH, 4’-CH), 4.02 (m, 1H; 5’-CHH), 3.92 (m, 1H; 5’-CHH), 3.40
(m, 1H; 1’-CHH), 3.32 (m, 1H; 1’-CHH), 2.26 (s, 3H; 4-CH3), 1.43,
1.40, 1.35, 1.32 ppm (4s, 12H; 4CH3); 13C NMR (100 MHz, CDCl3):
d=146.3 (C-3), 131.1 (C-5), 131.0 (C-1), 125.4 (C-4), 117.2 (C-6),
113.2 (C-2), 110.0, 108.9 (2C; 2C(CH3)2), 78.2 (C-3’), 75.4 (C-2’), 73.2
(C-4’), 68.1 (C-5’), 44.1 (C-1’), 28.0, 26.9, 25.45, 25.39 (4C; 4CH3),
21.7 ppm (CH3); MS (EI): m/z (%): 401, 399 (32) [M], 386, 384 (20)
[MꢀCH3], 200 (95), 198 (100) [MꢀCH2(OH)(CHOH)3]. Yield of isolat-
ed 3-bromo-4-methyl [1,5:3,4-bis(methylethyliden)ribo]aniline: 6%.
(100 MHz, CDCl3, signals that could be interpreted): d=164.4 (C-
10a), 163.0 (C-4), 139.7 (C-6), 133.9 (C-2), 131.4 (C-5a), 131.3 (C-9),
109.8, 109.1 (2C; 2C(CH3)2), 101.9 (C-4a), 77.9 (C-3’), 74.5 (C-2’), 73.2
(C-4’), 68.1 (C-5’), 50.9 (C-1’), 27.7, 26.8, 25.5, 25.1 (4C; 4CH3),
19.1 ppm (CH3-7a); MS (EI): m/z (%): 535 (3) [M], 520 (20) [MꢀCH3],
376 (15) [MꢀCF3(C6H3)CH3], 334 (35) [MꢀCH2(OH)(CHOH)3], 321 (90)
[Mꢀribose].
1
Yield of the side product: 20%; H NMR (400 MHz, CDCl3, only sig-
nals that differ from 4b are listed): d=7.30 ppm (1H; pyrimidine-
H); 13C NMR (100 MHz, CDCl3, signal that differs from 4b): d=
109.9 ppm (pyrimidine-CH); MS (EI): m/z (%): 535 (6) [M], 520 (25)
[MꢀCH3], 376 (15) [MꢀCF3(C6H3)CH3], 334 (40) [MꢀCH2(OH)-
(CHOH)3], 321 (100) [Mꢀribose].
3-Trifluoromethyl-4-methyl
[2,3:4,5-bis(methylethyliden)ribo]aniline
1
(3b): Yield: 41%; H NMR (400 MHz, CDCl3): d=7.04 (m, 1H; H-5),
6.88 (m, 1H; H-2), 6.68 (m, 1H; H-6), 4.38 (m, 1H; 2’-CH), 4.14 (m,
2H; 3’-CH, 4’-CH), 4.02 (m, 1H; 5’-CHH), 3.93 (m, 1H; 5’-CHH), 3.46
(m, 1H; 1’-CHH), 3.34 (m, 1H; 1’-CHH), 2.33 (s, 3H; 4-CH3), 1.43,
1.41, 1.35, 1.33 ppm (4s, 12H; 4CH3); 13C NMR (100 MHz, CDCl3):
d=145.7 (C-3), 132.7 (C-5), 132.6 (C-1), 129.5 (d, CF3), 124.8 (C-4),
115.9 (C-6), 110.7 (C-2), 110.0, 108.9 (2C; 2C(CH3)2), 78.2 (C-3’), 75.7
(C-2’), 73.2 (C-4’), 68.1 (C-5’), 43.6 (C-1’), 28.0, 26.8, 25.4, 25.3 (4C;
4CH3), 18.2 ppm (CH3); MS (EI): m/z (%): 389 (23) [M], 188 (100)
[MꢀCH2(OH)(CHOH)3].
General procedure for substitution of 4 with sodium benzoate:
An intermediate 4 (1 mmol) was added to a solution of sodium
benzoate (5m in benzyl alcohol, 5 equiv) in toluene (10 mL), and
resulting mixture was heated at reflux with stirring for 20–24 h. Af-
terwards, the reaction mixture was allowed to cool down, Celite
(1 g) was added, and the crude product was filtered out, washed
with toluene (2ꢂ10 mL), and concentrated in vacuum. FC on silica
gel (CH2Cl2/toluene 98:2) afforded the desired product as a colorless
solid.
8-Bromo intermediate 5a: Yield: 71%; 1H NMR (400 MHz, CDCl3):
d=7.55–7.28 (m, 13H; aromatic), 5.41, 5.28 (2s, 4H; 2CH2ꢀBz),
5.22 (s, 1H; H-4a), 4.54 (m, 2H; 2’-CH, 3’-CH), 4.07 (m, 2H; 4’-CH, 5’-
CHH), 3.93 (m, 1H; 5’-CHH), 3.85 (m, 1H; 1’-CHH), 3.67 (m, 1H; 1’-
CHH), 2.40 (s, 3H; CH3-7a), 1.32, 1.27, 1.25 ppm (3s, 12H; 4CH3);
13C NMR (100 MHz, CDCl3, DEPT): d=171.3 (C-4), 165.1 (C-2), 164.0
(C-10a), 142.6 (C-9a), 140.9, 137.2, 137.0, 136.7 (Bz), 132.7 (C-6),
131.4 (C-9), 128.5, 128.4, 128.3, 128.1, 128.0, 127.9 (Bz), 127.8 (C-5a),
127.7, 127.6 (Bz), 127.1 (C-7), 125.0 (C-8), 109.8, 108.5 (2C;
2C(CH3)2), 82.1 (C-4a), 78.1 (C-3’), 75.4 (C-2’), 73.0 (C-4’), 69.5, 68.1
(2C; 2CH2ꢀBz), 67.9 (C-5’), 50.4 (C-1’), 27.9, 26.7, 25.5, 25.3 (4C;
4CH3), 22.5 ppm (CH3-7a); MS (EI): m/z (%): 689 (3) [M], 674 (12)
[MꢀCH3], 488 (15) [MꢀCH2(OH)(CHOH)3], 475 (45) [Mꢀribose].
8-Trifluoromethyl intermediate 5b: Yield: 49%; 1H NMR (400 MHz,
CDCl3, HH-COSY): d=7.66–7.16 (m, 13H; aromatic), 5.40, 5.28 (2s,
4H; 2CH2ꢀBz), 5.22 (s, 1H; H-4a), 4.69 (m, 1H; 1’-CHH), 4.55 (m,
1H; 2’-CH), 4.06 (m, 2H; 4’-CH, 5’-CHH), 3.94 (m, 1H; 3’-CH), 3.85
(m, 1H; 5’-CHH), 3.58 (m, 1H; 1’-CHH), 2.48 (s, 3H; CH3-7a), 1.32,
1.25, 1.24 ppm (3s, 12H; 4CH3); 13C NMR (100 MHz, CDCl3, HMBC,
HMQC): d=171.3 (C-4), 165.0 (C-2), 163.7 (C-10a), 141.8 (C-9a),
137.2, 136.7, 133.1, 132.1, 129.0, 128.5 (2C), 128.4 (2C), 128.2, 128.1
(2C), 128.02 (2C), 128.98, 127.9, 125.3 (17C; Bz, C-5a, C-6, C-7, C-8,
C-9), 115.0 (d, J=81 Hz, CF3), 109.8, 108.9 (2C; 2C(CH3)2), 82.1 (C-
4a), 78.2 (C-3’), 75.5 (C-2’), 73.2 (C-4’), 68.6, 68.0 (2C; 2CH2ꢀBz),
68.2 (C-5’), 50.6 (C-1’), 27.8, 26.7, 25.4, 25.3 (4C; 4CH3), 17.8 ppm
(CH3-7a); MS (EI): m/z (%): 679 (5) [M], 664 (10) [MꢀCH3], 520 (8)
[Mꢀ(CH3)(CF3)C6H3], 478 (18) [MꢀCH2(OH)(CHOH)3], 465 (85) [Mꢀri-
bose].
3-Trifluoromethyl-4-methyl
[2,5:3,4-bis(methylethyliden)ribo]aniline:
1
Yield: 38%; H NMR (400 MHz, CDCl3): d=7.03 (d, J=8.0 Hz, 1H; H-
5), 6.92 (s, 1H; H-2), 6.71 (m, 1H; H-6), 4.13 (m, 1H; 2’-CH), 3.94 (m,
2H; 3’-CH, 4’-CH), 3.89 (m, 1H; 5’-CHH), 3.81 (m, 1H; 5’-CHH), 3.52
(m, 1H; 1’-CHH), 3.06 (m, 1H; 1’-CHH), 2.32 (s, 3H; 4-CH3), 1.51,
1.35 (2C), 1.24 ppm (3s, 12H; 4CH3); 13C NMR (100 MHz, CDCl3):
d=145.3 (C-3), 132.7 (C-5, C-1), 129.5 (d, CF3), 123.3 (C-4), 116.5 (C-
6), 110.8 (C-2), 108.6 (dioxolane-C(CH3)2), 102.0 (dioxepane-C(CH3)2),
76.7 (C-3’, C-2’, hidden under CDCl3 signal), 68.6 (C-4’), 58.4 (C-5’),
45.9 (C-1’), 28.4, 25.6, 25.2, 23.7 (4C; 4CH3), 18.2 ppm (CH3); MS
(EI): m/z (%): 389 (28) [M], 188 (100) [MꢀCH2(OH)(CHOH)3].
General procedure for condensation with 2,4,6-trichloropyrimi-
dine: A protected riboaniline 3 (1 mmol), KHCO3 (4 equiv), and tet-
rabutylammonium bromide (cat. amount) were suspended in tolu-
ene (4 mL). A solution of 2,4,6-trichloropyrimidine (2 equiv) in tolu-
ene (1 mL) was added, and the resulting mixture was stirred under
argon at 1008C for 22–26 h. After filtration from solid material, the
mixture was directly loaded on top of a FC column and eluted
with CH2Cl2/Et2O (95:5). Three compounds were isolated in the fol-
lowing order: undesired side-product (20–27%), starting com-
pound, and desired product (35–42%).
8-Bromo intermediate 4a: Yield: 42%; 1H NMR (400 MHz, CDCl3):
d=7.33 (m, 1H; H-6), 7.22 (s, 1H; H-9), 7.14 (m, 1H; H-5a), 6.00 (m,
1H; H-4a), 4.47 (m, 1H; 2’-CH), 4.35 (m, 1H; 3’-CH), 4.09 (m, 2H; 4’-
CH, 5’-CHH), 3.95 (m, 3H; 5’-CHH, 1’-CH2), 2.34 (s, 3H; CH3-7a), 1.40,
1.34, 1.29, 1.28 ppm (4s, 12H; 4CH3); 13C NMR (100 MHz, CDCl3):
d=164.2 (C-10a), 159.7 (C-4), 140.3 (C-2), 138.7 (C-9a), 132.1 (C-6),
129.0 (C-9), 128.2 (C-5a), 127.1 (C-7), 125.3 (C-8), 109.9, 109.3 (2C;
2C(CH3)2), 102.0 (C-4a), 77.8 (C-3’), 74.4 (C-2’), 73.2 (C-4’), 68.1 (C-5’),
50.6 (C-1’), 27.8, 26.9, 25.6, 25.3 (4C; 4CH3), 22.6 ppm (CH3-7a); MS
(EI): m/z (%): 545 (3) [M], 530 (12) [MꢀCH3], 344 (15) [MꢀCH2(OH)-
(CHOH)3], 331 (60) [Mꢀribose]; HRMS: m/z calcd: 568.037608
[M+Na]+; found: 568.037204.
General procedure for the reduction of the pyrimidine ring: A
flask containing a dibenzyl compound 5 (0.5 mmol) and Pd/C
(25 mg, 10%) in methanol (2.5 mL) was flushed with argon, and
the reaction mixture was stirred under hydrogen for 2 h and then
flushed again with argon to remove the hydrogen. After removal
of solid material by filtration and the concentration of the solvent,
the corresponding product 6 was purified by FC on silica gel with
elution with CH2Cl2/methanol/toluene 98:1:1.
8-Trifluoromethyl intermediate 4b: Yield of desired 4b: 35%;
1H NMR (400 MHz, CDCl3): d=7.60 (m, 1H; H-6), 7.40-7.24 (m, 2H;
H-9, H-5a), 6.00 (m, 1H; H-4a), 4.49 (m, 1H; 2’-CH), 4.12 (m, 2H; 3’-
CH, 4’-CH), 4.02 (m, 1H; 5’-CHH), 3.85 (m, 3H; 5’-CHH, 1’-CH2), 2.53
(s, 3H; CH3-7a), 1.41, 1.31, 1.29, 1.27 ppm (4s, 12H; 4CH3); 13C NMR
8-Bromo intermediate 6a: Yield: 79%; 1H NMR (400 MHz, CDCl3):
d=9.32 (brs, 1H; NH), 8.58 (brs, 1H; NH), 7.28 (d, 1H; H-6), 7.20
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 645 – 654 651

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
(m, 2H; H-9, H-5a), 4.70 (s, 1H; H-4a), 4.39 (m, 1H; 2’-CH), 4.04 (m,
3H; 3’-CH, 4’-CH, 5’-CHH), 3.94 (m, 1H; 5’-CHH), 3.89 (m, 1H; 1’-
CHH), 3.75 (m, 1H; 1’-CHH), 2.40 (s, 3H; CH3-7a), 1.42, 1.36,
1.31 ppm (3s, 12H; 4CH3); 13C NMR (100 MHz, CDCl3, HMBC,
HMQC): d=164.6 (2C; C-4, C-2), 154.0 (C-10a), 139.9 (C-9a), 138.7
(C-6), 132.0 (C-9), 126.9 (C-5a), 125.7 (C-7), 125.2 (C-8), 110.2, 109.7
(2C; 2C(CH3)2), 78.9 (C-4a), 77.8 (C-3’), 74.8 (C-2’), 73.0 (C-4’), 67.9
(C-5’), 52.5 (C-1’), 27.6, 26.7, 25.3, 25.1 (4C; 4CH3), 22.5 ppm (CH3-
7a); MS (EI): m/z (%): 509 (6) [M], 494 (5) [MꢀCH3], 308 (10)
[MꢀCH2(OH)(CHOH)3], 295 (45) [Mꢀribose].
8-Trifluoromethyl intermediate 6b: Yield: 77%; 1H NMR (400 MHz,
CDCl3, HH COSY): d=8.55 (brs, 1H; NH), 8.47 (brs, 1H; NH), 7.57
(m, 1H; H-6), 7.36 (m, 1H; H-5a), 7.24 (s, 1H; H-9), 4.64 (s, 1H; H-
4a), 4.39 (m, 1H; 2’-CH), 4.10 (m, 2H; 1’-CHH, 5’-CHH), 4.00 (m, 2H;
3’-CH, 4’-CH), 3.90 (m, 1H; 5’-CHH), 3.74 (m, 1H; 1’-CHH), 2.50 (s,
3H; CH3-7a), 1.43, 1.35, 1.32, 1.31 ppm (4s, 12H; 4CH3); 13C NMR
(100 MHz, CDCl3, HMBC, HMQC): d=164.2 (C-4), 154.1 (C-2), 150.4
(C-10a), 139.3 (C-9a), 137.4 (C-7), 134.0 (C-6), 131.2 (C-9), 128.5 (d,
J=81 Hz, CF3), 125.8 (C-5a), 122.3 (C-8), 110.3, 110.0 (2C; 2C(CH3)2),
79.0 (C-4a), 77.8 (C-3’), 75.0 (C-2’), 73.1 (C-4’), 68.0 (C-5’), 52.8 (C-1’),
27.6, 26.7, 25.3, 25.1 (4C; 4CH3), 19.0 ppm (CH3-7a); MS (EI): m/z
(%): 499 (15) [M], 484 (10) [MꢀCH3], 298 (15) [MꢀCH2(OH)(CHOH)3],
285 (100) [Mꢀribose].
8-Bromo-[2,3:4,5-bis(methylethyliden)ribo]flavin (8a): Yield: 89%;
1H NMR (400 MHz, CDCl3, HH COSY): d=8.61 (brs, 1H; NH-3), 8.10
(2s, 2H; H-6, H-9), 5.09 (m, 1H; 1’-CHH), 4.91 (m, 1H; 1’-CHH), 4.76
(m, 1H; 2’-CH), 4.22 (m, 3H; 3’-CH, 4’-CH, 5’-CHH), 4.02 (m, 1H; 5’-
CHH), 2.55 (s, 3H; CH3-7a), 1.50, 1.39, 1.21 ppm (3s, 12H; 4CH3);
13C NMR (100 MHz, CDCl3, HMQC, HMBC): d=159.0, 154.5 (2C; C-4,
C-2), 150.6 (C-10a), 137.5, 135.0, 134.2, 133.2 (C-6 or C-9), 132.1 (C-
9), 129.0 (C-5a), 128.2 (C-7), 120.5 (C-6 or C-9), 110.4 (2C(CH3)2),
78.0 (C-3’), 74.2 (C-2’), 73.6 (C-4’), 68.2 (C-5’), 46.1 (C-1’), 27.6, 26.9,
25.5, 25.3 (4C; 4CH3), 22.6 ppm (CH3-7a); MS (EI): m/z (%): 520 (3)
[M], 504 (10) [MꢀO], 462 (8) [MꢀC(CH3)2O], 320 (5) [MꢀCH2(OH)-
(CHOH)3], 304 (10) [Mꢀribose]; HRMS: m/z calcd: 520.095762
[M+Na]+; found: 520.095525.
8-Trifluoromethyl-[2,3:4,5-bis(methylethyliden)ribo]flavin (8b): Yield:
1
95%; H NMR (400 MHz, CDCl3, HH COSY): d=9.10 (brs, 1H; NH),
8.18, 8.17 (m, 2H; H-6, H-9), 5.22 (m, 1H; 1’-CHH), 4.86 (m, 1H; 1’-
CHH), 4.76 (m, 1H; 2’-CH), 4.24 (m, 3H; 3’-CH, 4’-CH, 5’-CHH), 4.00
(m, 1H; 5’-CHH), 2.62 (s, 3H; CH3-7a), 1.46, 1.44, 1.37, 1.18 ppm (4s,
12H; 4CH3); 13C NMR (100 MHz, CDCl3, HMBC, HMQC): d=158.7 (C-
4), 154.7 (C-2), 150.8 (C-10a), 139.7 (C-4a), 136.6 (C-6 or C-9), 135.2
(C-7), 134.6, 131.2, 124.6, 121.9, 115.5 (C-9 or C-6), 110.5 (2C;
2C(CH3)2), 78.0 (C-3’), 74.3 (C-2’), 73.6 (C-4’), 68.2 (C-5’), 46.2 (C-1’),
27.4, 26.7, 25.5, 25.2 (4C; 4CH3), 18.9 ppm (CH3-7a); MS (EI): m/z
(%): 452 (23) [MꢀOC(CH3)2], 495 (30) [MꢀCH3], 510 (5) [M].
General procedure for ring closure: Sodium nitrite (5 equiv) was
added to a solution of a compound 6 (0.5 mmol) in glacial acetic
acid (2.5 mL) in a dark environment. The resulting mixture was
stirred at room temperature for 3 h, and then water (2 mL) was
added. The resulting suspension was stirred for an additional 3 h,
after which the solvents were evaporated in vacuum. The crude
product was purified on silica gel with elution with CH2Cl2/metha-
nol/toluene (96:3:1) to afford the corresponding compound 7 as
an orange solid.
General procedure for deprotection of OH groups: A compound
8 (0.04 mmol) was stirred in TFA/H2O solution (10:1, 0.5 mL) at RT
for 4 h. The solvents were then evaporated, and ether (2 mL) was
added to the crude product. The formed precipitate was centri-
fuged (108C, 3000 rpm, 10 min), and washed with ether (1 mL).
The combined ethereal extracts were washed with water, and the
aqueous solution was lyophilized. The solid was dried in vacuo to
provide the corresponding pure compound 9 (85–95%).
1
8-Bromo-5-N-oxide-isoalloxazine (7a): Yield: 69%; H NMR (400 MHz,
8-Bromoriboflavin (9a): Yield: 85%; 1H NMR (400 MHz, [D6]DMSO,
HH COSY): d=11.42 (brs, 1H; NH-3), 8.35 (s, 1H; H-6), 8.08 (s, 1H;
H-9), 5.10 (m, 1H; 5’-CHH), 4.85 (m, 1H; 5’-CHH), 4.66 (m, 3H; 2’-
CH, 3’-CH, 4’-CH), 4.44 (m, 1H; 1’-CHH), 4.22 (m, 1H; 1’-CHH), 3.62
(m, 4H; 4OH), 2.48 ppm (s, 3H; CH3-7a); 13C NMR (100 MHz,
[D6]DMSO, HMQC): signals that could be identified: d=159.5, 155.1
(2C; C-4, C-2),135.4, 134.2, 131.2, 130.0, 117.0, 131.8 (C-9, from
HMQC), 121.0 (C-6, from HMQC), 73.3 (C-3’), 72.4 (C-2’), 68.9 (C-4’),
63.3 (C-5’), 49.2 (C-1’), 21.8 ppm (CH3-7a); MS (ESI): m/z (%): 385
(50) [MꢀBr+H+Na]+, 463 (100) [M+Na]+, 903 (5) [2M+Na]+;
HRMS: m/z calcd: 463.022930 [M+Na]+; found: 463.023083.
CDCl3): d=8.27 (brs, 1H; NH), 7.24 (m, 1H; H-6, signal partially cov-
ered by signal of CDCl3), 7.18 (m, 1H; H-9), 5.12 (m, 1H; 2’-CH),
4.74 (m, 2H; 3’-CH, 4’-CH), 4.23 (m, 3H; 5’-CH2, 1’-CHH), 4.00 (m,
1H; 1’-CHH), 2.56 (s, 3H; CH3-7a), 1.55, 1.46, 1.37, 1.23 ppm (3s,
12H; 4CH3); 13C NMR (100 MHz, CDCl3): d=155.8, 153.7 (2C, C-4, C-
2), 137.3 (C-10a), 134.1 (C-4a), 133.7 (C-9a), 133.3 (C-6), 129.0 (C-9),
128.2 (C-5a), 122.1 (C-7), 121.7 (C-8), 110.4, 110.3 (2C; 2C(CH3)2),
77.9 (C-3’), 74.7 (C-2’), 73.6 (C-4’), 68.2 (C-5’), 45.9 (C-1’), 27.6, 26.9,
25.6, 25.3 (4C; 4CH3), 22.8 ppm (CH3-7a); MS (EI): m/z (%): 507 (10)
[MꢀNO], 521 (3) [MꢀO].
8-Trifluoromethyl-5-N-oxide-isoalloxazine (7b): Yield: 73%; 1H NMR
(400 MHz, CDCl3): d=8.37 (brs, 1H; NH), 8.34 (m, 1H; H-6), 8.25 (m,
1H; H-9), 5.24 (m, 1H; 2’-CH), 4.75 (m, 2H; 1’-CHH, 5’-CHH), 4.23
(m, 3H; 3’-CH, 4’-CH, 5’-CHH), 4.00 (m, 1H; 1’-CHH), 2.63 (s, 3H; 7a-
CH3), 1.47, 1.43, 1.38, 1.21 ppm (4s, 12H; 4CH3); 13C NMR (100 MHz,
CDCl3): d=155.6, 153.7 (2C; C-4, C-2), 153.3 (C-10a), 137.9, 135.5,
135.1, 132.6, 126.2, 125.3, 124.3 (C-4a, C-5a, C-6, C-7, C-8, C-9, C-
9a), 110.5, 110.1 (2C; 2C(CH3)2), 78.0 (C-3’), 74.6 (C-2’), 73.6 (C-4’),
68.2 (C-5’), 46.1 (C-1’), 27.5, 26.8, 25.5, 25.2 (4C; 4CH3), 19.1 ppm
(CH3-7a); MS (EI): m/z (%): 497 (10) [MꢀNO], 511 (15) [MꢀO].
8-(Trifluoromethyl)riboflavin (9b): Yield: 95%; 1H NMR (400 MHz,
[D6]DMSO, HH COSY): d=11.52 (brs, 1H; NH), 8.38 (m, 1H; H-6),
8.17 (m, 1H; H-9), 4.80 (m, 2H; 1’-CH2, signal partially covered by
signal of H2O), 4.50 (m, 1H; 2’-H), 4.22 (m, 1H; 4’-CH), 4.00 (m, 1H;
3’-CH, 5’-CHH), 3.62 (m, 1H; 5’-CHH), 2.57 ppm (s, 3H; 7a-CH3);
13C NMR (100 MHz, [D6]DMSO, HMQC): d=159.4, 153.4 (2C; C-2, C-
4), 148.0 (C-10a), 140.9 (C-4a), 136.0 (C-9), 134.0, 132.4, 131.8, 133.3
(C-6), 129.0 (C-9), 73.4 (C-3’), 72.6 (C-2’), 68.7 (C-4’), 63.3 (C-5’), 47.4
(C-1’, from HMQC), 18.0 ppm (CH3-7a); MS (ESI): m/z (%): 453 (100)
[M+Na]+; HRMS: m/z calcd: 453.099242 [M+Na]+; found:
453.099723.
General procedure for reduction of N-oxides 7: Na2S2O4
(1.5 equiv) in water (3 mL) was added to a solution of a com-
pound 7 (0.085 mmol) in EtOH (3 mL) under argon. After the
system had been stirred at RT for 1.5 h, the solvents were evapo-
rated, and the crude yellow solid was washed with water (2ꢂ2 mL)
and dried under vacuum. FC on silica gel (elution with CH2Cl2/
MeOH 97:3) yielded the corresponding compound 8 (89–95%).
Isolation of YtvA protein and chromophore exchange: The isola-
tion of recombinant YtvA has been described elsewhere.[4,39,40] In-
corporation of both 8-CF3RF and 8-BrRF as chromophores in the
YtvA protein was achieved by a recently developed protocol based
on the unfolding/refolding of the protein on a His-Trap column by
treatment with a high concentration of urea and the exchange of
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 645 – 654 652

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
the native FMN chromophore with the artificial flavins during re-
folding.[14] The yield of chromophore incorporation was assessed
by determining the A280/A450 absorption ratios; this gave values of
7:1 for 8-CF3RF (WT YtvA: 4:1) and 5:1 for 8-BrRF (note that 8-
made use of the fact that the steric requirements of a bromine
substituent resemble those of a methyl group. Starting from the
optimized structures of the wild-type complexes,[33] we replaced
the methyl group in the 8-position by a bromine atom and relaxed
the C8–Br interatomic distance at the QM level while freezing the
geometric positions of all other centers and of the point charges
representing the MM environment.
CF3RF has
a lower extinction coefficient than FMN: emax =
8450 cmꢀ1 mꢀ1, in comparison with emax =12500 cmꢀ1 mꢀ1). Both
chromophores were used in their riboflavin forms, due to low
yields of the phosphorylation reactions. Our previous study
showed that presence of the 5’-phosphate group does not influ-
ence the photochemistry within the chromophore pocket in the
protein and causes only a minor change in the dark recovery kinet-
ics.[14]
Keywords: chromophore–protein interactions
·
flavins
·
photoreceptors · photophysics · QM/MM calculations
[1] J. M. Christie, Annu. Rev. Plant Biol. 2007, 58, 21–45.
[2] U. Krauss, B. Q. Minh, A. Losi, W. Gꢀrtner, T. Eggert, A. von Haeseler, K. E.
Jaeger, J. Bacteriol. 2009, 191, 7234–7242.
Analytical methods: Absorption spectra and kinetics traces were
recorded with
a UV2401-PC spectrophotometer (Shimadzu).
Steady-state fluorescence was measured on a Varian Eclipse fluo-
rimeter by using a 5 nm emission slit width and a scanning speed
of 300 nmminꢀ1. Fluorescence quantum yields were calculated
with RF (FF =0.27 in H2O)[30] and wild-type YtvA (FF =0.22 in phos-
phate buffer)[34] as references. Time-resolved fluorescence was mea-
sured with a single-photon-counting apparatus model FL900 (Edin-
burgh Analytical Instruments, Livingston, UK). Transient absorbance
changes after nanosecond-laser flash excitation were recorded
with a detection system (LFP111) from Luzchem, Ontario, Canada.
For excitation, a Nd:YAG-driven tunable OPO laser was used
(Nd:YAG, Innolas, Garching, Germany; OPO, GWU Lasertechnik, Erft-
stadt, Germany). The experiments were performed in the linear
laser-energy dependence region of the transient absorbance
changes with lexc =450 nm. Triplet quantum yields (FT) of the
flavin derivatives were determined by comparison with the known
triplet quantum yield of RF. For this the literature value of FT =0.6
was used as reference.[32] Data analysis was performed with Origin
software. All measurements were performed at 208C by using
1 cm light-path quartz cuvettes. Photoequilibrium conditions, with
accumulation of the photoactivated state, were achieved by illumi-
nating the sample with a blue-light-emitting Led-Lenser V8 lamp
(Zweibrꢃder Optoelectronics, Solingen, Germany).
[3] M. Salomon, W. Eisenreich, H. Dꢃrr, E. Scheicher, E. Knieb, V. Massey, W.
Rꢃdiger, F. Mꢃller, A. Bacher, G. Richter, Proc. Natl. Acad. Sci. USA 2001,
98, 12357–12361.
[4] Y. Tang, Z. Cao, E. Livoti, U. Krauss, K.-E. Jaeger, W. Gꢀrtner, A. Losi, Pho-
tochem. Photobiol. Sci. 2010, 9, 47–56.
[5] S. Raffelberg, M. Mansurova, W. Gꢀrtner, A. Losi, J. Am. Chem. Soc. 2011,
133, 5346–5356.
[6] S. Kikuchi, M. Unno, K. Zikihara, S. Tokutomi, S. Yamauchi, J. Phys. Chem.
B 2009, 113, 2913–2921.
[7] T. Drepper, T. Eggert, F. Circolone, A. Heck, U. Krauss, J.-K. Guterl, M.
Wendorff, A. Losi, W. Gꢀrtner, K.-E. Jaeger, Nat. Biotechnol. 2007, 25,
443–445.
[8] S. Chapman, C. Faulkner, E. Kaiserli, C. Garcia-Mata, E. I. Savenkov, A. G.
Roberts, K. J. Oparka, J. M. Christie, Proc. Natl. Acad. Sci. USA 2008, 105,
20038–20043.
[9] A. Mçglich, K. Moffat, Photochem. Photobiol. Sci. 2010, 9, 1286–1300.
[10] A. Losi, W. Gꢀrtner, Annu. Rev. Plant Biol. 2012, 63, 49–72.
[11] B. D. Zoltowski, B. R. Crane, Biochemistry 2008, 47, 7012–7019.
[12] M. Salomon, J. M. Christie, E. Knieb, U. Lempert, W. R. Briggs, Biochem-
istry 2000, 39, 9401–9410.
[13] A. Tyagi, A. Penzkofer, T. Mathes, P. Hegemann, J. Photochem. Photobiol.
B 2010, 101, 76–88.
[14] M. Mansurova, P. Scheercousse, J. Simon, M. Kluth, W. Gꢀrtner, ChemBio-
Chem 2011, 12, 641–646.
[15] L. B. Johansson, A. Davidsson, G. Lindblom, K. R. Naqvi, Biochemistry
1979, 18, 4249–4253.
[16] T. Climent, R. Gonzalez-Luque, M. Merchan, L. Serrano-Andres, J. Phys.
Chem. A 2006, 110, 13584–13590.
[17] T. Mathes, C. Vogl, J. Stolz, P. Hegemann, J. Mol. Biol. 2009, 385, 1511–
1518.
[18] H. Dꢃrr, M. Salomon, W. Rꢃdiger, Biochemistry 2005, 44, 3050–3055.
[19] P. Hemmerich, B. Prijs, H. Erlenmeyer, Helv. Chim. Acta 1959, 42, 1604–
1611.
Computational methods: Calculations on 8-BrLF and 8-CF3LF
(without the sugar side chain) were carried out by using DFT for
the geometry optimizations and a combination of DFT and multi-
reference configuration interaction (MRCI) for the determination of
the excitation energies and transition moments. Throughout,
a basis set of valence triple zeta quality with polarization functions
(TZVP) was used in the quantum chemical treatments. Water solu-
tion was mimicked by adding four explicit water molecules form-
ing hydrogen bonds with the nitrogen and oxygen atoms of the
[20] G. Scola-Nagelschneider, P. Hemmerich, Eur. J. Biochem. 1976, 66, 567–
577.
chromophore combined with
a continuum solvation model
[21] V. M. Berezovskii, G. D. Glebova, Russ. J. Hetercycl. 1965, 1, 121–124.
[22] Y. V. S. N. Murthy, V. Massey, J. Biol. Chem. 1998, 273, 8975–8982.
[23] F. Yoneda, K. Shinozuka, K. Hiromatsu, R. Matsushita, Y. Sakuma, M.
Hamana, Chem. Pharm. Bull. 1980, 28, 3576–3583.
[24] H. Fenner, D. Hochhuth, Arch. Pharm. 1990, 323, 873–879.
[25] M. Mansurova, M. S. Koay, W. Gꢀrtner, Eur. J. Org. Chem. 2008, 5401–
5406.
[26] E. E. Carlson, L. L. Kiessling, J. Org. Chem. 2004, 69, 2614–2617.
[27] Y. M. Legrand, M. Gray, G. Cooke, V. M. Rotello, J. Am. Chem. Soc. 2003,
125, 15789–15795.
[28] E. Sikorska, I. V. Khmelinskii, W. Prukala, S. L. Williams, M. Patel, D. R.
Worrall, J. L. Bourdelande, J. Koput, M. Sikorski, J. Phys. Chem. A 2004,
108, 1501–1508.
[29] S. Salzmann, J. Tatchen, C. M. Marian, J. Photochem. Photobiol. A 2008,
198, 221–231.
(COSMO) for describing the solvent polarity. Further details on the
methods and procedures can be found in earlier work[29] (solvation
protocol IV), from which the theoretical results for the parent com-
pound were also taken. Spectroscopic properties of 8-BrLF in the
LOV domain of YtvA were studied by combined quantum-mechan-
ical/molecular-mechanical (QM/MM) methods. In addition to the
modified FMN cofactor, several important amino acid side chains,
including Gln66, Asn94, Asn104, and Gln123, as well as the reactive
residue Cys62, were treated quantum mechanically, whereas the
protein environment was represented by the CHARMM force field
(refer to Salzmann et al.[33] for a detailed description of the proto-
col). Because of the lack of force-field parameters suited for de-
scribing the nonbonded interactions between bromine and the
protein environment, the active regions of the two conformers
(confA and confB) of the modified protein–cofactor complex were
not fully geometry optimized at the QM/MM level. Instead, we
[30] P. Drçssler, W. Holzer, A. Penzkofer, P. Hegemann, Chem. Phys. 2003,
286, 409–420.
[31] S. Salzmann, V. Martinez-Junza, B. Zorn, S. E. Braslavsky, M. Mansurova,
C. M. Marian, W. Gꢀrtner, J. Phys. Chem. A 2009, 113, 9365–9375.
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 645 – 654 653

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
[32] L. Crovetto, V. Martinez-Junza, S. E. Braslavsky, Photochem. Photobiol.
2006, 82, 281–290.
[38] A. Penzkofer, L. Endres, T. Schiereis, P. Hegemann, Chem. Phys. 2005,
316, 185–194.
[33] S. Salzmann, M. R. Silva, W. Thiel, C. M. Marian, J. Phys. Chem. B 2009,
113, 15610–15618.
[39] A. Losi, E. Polverini, B. Quest, W. Gꢀrtner, Biophys. J. 2002, 82, 2627–
2634.
[34] A. Losi, Photochem. Photobiol. 2007, 83, 1283–1300.
[35] W. Holzer, A. Penzkofer, M. Fuhrmann, P. Hegemann, Photochem. Photo-
biol. 2002, 75, 479–487.
[40] A. Losi, B. Quest, W. Gꢀrtner, Photochem. Photobiol. Sci. 2003, 2, 759–
766.
[36] T. A. Schꢃttrigkeit, C. K. Kompa, M. Salomon, W. Rꢃdiger, M. E. Michel-
Beyerle, Chem. Phys. 2003, 294, 501–508.
Received: October 25, 2012
[37] T. E. Swartz, S. B. Corchnoy, J. M. Christie, J. W. Lewis, I. Szundi, W. R.
Briggs, R. A. Bogomolni, J. Biol. Chem. 2001, 276, 36493–36500.
Published online on March 1, 2013
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 645 – 654 654