Using this approach, all 13C signals of FMN bound to LOV1 and LOV2 domains of Avena sativa and to the LOV2 domain of the fern, Adiantum capillus-veneris, could be unequivocally assigned
Trang 1Wolfgang Eisenreich1, Monika Joshi1, Boris Illarionov1, Gerald Richter2, Werner Ro¨misch-Margl1, Franz Mu¨ller3, Adelbert Bacher1and Markus Fischer4
1 Lehrstuhl fu¨r Organische Chemie und Biochemie, Technische Universita¨t Mu¨nchen, Garching, Germany
2 School of Chemistry, Cardiff University, UK
3 Wylstrasse 13, Hergiswil, Switzerland
4 Institut fu¨r Biochemie und Lebensmittelchemie, Abteilung Lebensmittelchemie, Universita¨t Hamburg, Germany
13C-Labeled flavocoenzymes have played an important
role for the spectroscopic analysis of flavoenzymes [1],
but their use was limited by the costs and effort for
the preparation of the13C-labeled cofactors More
spe-cifically, 13C can be introduced with relative ease into
the pyrimidine moiety of the isoalloxazine system [2],
and the xylene moiety of flavin cofactors can be
labeled by enzyme-assisted strategies [3], whereas
ribi-tyl carbon atoms have been rarely included in labeling
studies due to technical hurdles We have shown earlier that mixtures of 13C isotopologues of the riboflavin precursor, 6,7-dimethyl-8-ribityllumazine, can be pre-pared by biotransformation of 13C-labeled glucose
in vivo[4] In the present study, we report the transfor-mation of these isotopologue libraries into random libraries of 13C isotopologues of FMN and their utili-zation for NMR studies of the plant blue light recep-tor, phototropin [5,6]
Keywords
blue light receptor; isotopologue libraries;
LOV domain; NMR spectroscopy;
phototropin
Correspondence
W Eisenreich, Lehrstuhl fu¨r Organische
Chemie und Biochemie, Technische
Universita¨t Mu¨nchen, Lichtenbergstrasse 4,
D-85747 Garching, Germany
Fax: +49 89 289 13363
Tel: +49 89 289 13336
E-mail: wolfgang.eisenreich@ch.tum.de
M Fischer, Institut fu¨r Biochemie und
Lebensmittelchemie, Abteilung
Lebensmittelchemie, Universita¨t Hamburg,
Grindelallee 117, 20146 Hamburg, Germany
Fax: +49 40 4283 84342
Tel: +49 40 4283 84357
E-mail:
markus.fischer@chemie.uni-hamburg.de
(Received 6 June 2007, revised 12 August
2007, accepted 19 September 2007)
doi:10.1111/j.1742-4658.2007.06111.x
The plant blue light receptor phototropin comprises a protein kinase domain and two FMN-binding LOV domains (LOV1 and LOV2) Blue light irradiation of recombinant LOV domains is conducive to the addition
of a cysteinyl thiolate group to carbon 4a of the FMN chromophore, and spontaneous cleavage of that photoadduct completes the photocycle of the receptor The present study is based on 13C NMR signal modulation observed after reconstitution of LOV domains of different origins with ran-dom libraries of13C-labeled FMN isotopologues Using this approach, all
13C signals of FMN bound to LOV1 and LOV2 domains of Avena sativa and to the LOV2 domain of the fern, Adiantum capillus-veneris, could be unequivocally assigned under dark and under blue light irradiation condi-tions.13C Chemical shifts of FMN are shown to be differently modulated
by complexation with the LOV domains under study, indicating slight differences in the binding interactions of FMN and the apoproteins
Abbreviations
C(4a), carbon 4a; TARF, tetraacetylriboflavin.
Trang 2The gene specifying phototropin, the first in the
emerging family of blue light receptors in plants, was
initially cloned from Arabidopsis thaliana and was
shown to specify a cytoplasmic protein comprising
a serine⁄ threonine protein kinase domain and two
FMN-binding LOV domains (designated LOV1 and
LOV2), which are members of the PAS domain
super-family [6–8] (Fig 1) Blue light irradiation of
recombi-nant LOV domains results in substantial modulation
of the visible absorption spectrum [9], which was
inter-preted to result from the formation of an adduct
between the thiol group of a cystein residue and
car-bon 4a [C(4a)] of the FMN chromophore by Vincent
Massey (a contribution to the discussion at the 13th
International Congress on Flavins and Flavoproteins;
29 August to 4 September 1999, Konstanz) (Fig 2)
This interpretation was confirmed by site-directed
mutagenesis, NMR spectroscopy and X-ray
crystallo-graphy [9–13] The photocycle is best described as an
addition⁄ elimination sequence
The structure of recombinant LOV2 domain of the fern Adiantum capillus-veneris has been determined by X-ray crystallography in the dark as well as the light state [12,13] (Fig 3) The protein is characterized by five antiparallel b-sheets and four a-helices that form a central pocket harboring the FMN chromophore A light-induced change in the position of the side chain
of cysteine 966 is well in line with the adduct forma-tion [12–14] In addiforma-tion to the cysteine 966 residue,
11 amino acid residues were shown to contact the FMN chromophore via hydrogen bonds or van der Waals contacts Notably, these residues are highly con-served in all LOV domains, indicating a canonical FMN binding motif (Fig 2)
NMR studies with the LOV2 domain of Avena
sati-va showed that the photoadduct formation involves a conformational change in the ribityl side chain of the flavin cofactor as indicated by 31P and13C NMR data [10] It was also shown that the adduct formation trig-gers the unfolding of the helical domain Ja, which
Fig 1 Organization of phototropins used in the preent study A, A sativa NPH1-1 (accession no O49003); B, A capillus-veneris phy3 (accession no Q9ZWQ6).
Fig 2 Amino acid sequence alignment of domains used in the present study Amino acid residues derived from the vector are italicized Asterisks indicate amino acid residues of protein from A capillus-veneris in direct contact with the FMN chromophore [13] Identical amino acid residues are shown in black, and similar amino acid residues are in grey shadow typeface The formation of a cysteinyl-flavin-C(4a) cova-lent adduct after irradiation of the LOV–FMN complex with blue light is shown and the cystein residue involved in adduct formation is marked by an arrow.
Trang 3serves as a linker between the LOV2 domain and the
kinase domain in the LOV2 domain of A sativa [11]
That unfolding is believed to modulate the activity of
the kinase domain, which is conducive to its
autophos-phorylation
The exact role of the LOV1 domain in regulating
phototropin activity is not fully understood, but the
overall architecture of LOV1 from Chlamydomonas was
found to be almost identical with that of LOV2 [15]
Recent studies indicate that LOV2 acts as the principal
light sensing domain, which is coupled via the Ja helix
with the kinase activity [11], whereas LOV1 may play a
crucial role in receptor dimerization [16,17]
The present study was initiated in order to monitor more closely the light-induced chemical shift modula-tion of the FMN cofactors complexed to LOV domains of different origins For the unequivocal assignment of the 13C NMR signals of all carbon atoms in the FMN chromophore, the apoproteins were reconstituted with random and ordered 13C isotopo-logue libraries of FMN The signal intensity modula-tion reflecting the different isotopologue composimodula-tions
in samples with random isotopologue libraries of FMN served as the basis for isotope abundance editing
of the 13C NMR signals The method can be adapted for NMR signal assignment in a variety of other pro-tein⁄ ligand systems
Results
We have reported earlier on the preparation of isoto-pologue mixtures of 6,7-dimethyl-8-ribityllumazine (3; Fig 4) by in vivo biotransformation of13C-labeled glu-cose using a recombinant Escherichia coli strain [4] These isotopologue mixtures were used in the present study as a starting material for the preparation of iso-topologue mixtures of FMN by enzyme-assisted syn-thesis
The transformation of 3 into riboflavin catalyzed by the enzyme riboflavin synthase proceeds as a dismuta-tion whereby two equivalents of 3 are transformed into one equivalent each of riboflavin (5; Fig 4) and 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione (4; Fig 4) In order to avoid the inherent loss of isotope-labeled precursor, the second product 4 resulting from the dismutation can be reconverted into 3 by treatment with lumazine synthase using 3,4-dihydroxy-2-buta-none 4-phosphate as cosubstrate The cosubstrate can
be prepared in appropriately 13C-labeled form by enzymatic conversion of 13C-labeled glucose By that approach, the yield of riboflavin based on isotope-labeled 3 can be optimized (for details, see Experimen-tal procedures)
The riboflavin arising by the in vitro biotransforma-tion can be converted into FMN by treatment with riboflavin kinase in situ; ATP required as kinase sub-strate can be conveniently recycled using phosphoenol pyruvate as phosphate donor One-pot reaction mix-tures catalyzing the formation of FMN from randomly labeled 3 and specifically 13C-labeled glucose comprise nine enzyme catalysts and afford the product at a yield
of over 90% based on the13C-labeled 3 [3]
Synthetic genes specifying the LOV1 (Fig 5) and LOV2 domains of phototropin NPH1-1 from A sativa and the LOV2 domain of phototropin phy3 from the fern A capillus-veneris were constructed as described
A
B
Fig 3 Adiantum phy3 LOV2 structures (A) dark and (B) light state
(protein databank ID code 1G28, respectively, 1JNU) [12,13] Atoms
in amino acid residues are colored by elements: carbon, white;
oxy-gen, red; nitrooxy-gen, blue; sulfur, yellow.
Trang 4in the Experimental procedures All genes were
opti-mized for hyperexpression in E coli host strains
Gen-erally, the assembled DNA fragments were cloned into
an expression vector specifying fusion proteins
com-prising hisactophilin from Dictyostelium discoideum
and a thrombin cleavage site All sequences have been
deposited in GenBank The cognate fusion proteins
were expressed efficiently in recombinant E coli strains
and could be bound to nickel-chelating Sepharose due
to the large number of histidine residues present in the
hisactophilin domain The column was washed with
buffer containing 8 m urea to release the
protein-bound FMN, and the resulting apoprotein was
recon-stituted on the column with isotope-labeled FMN The
protein was then eluted with imidazole The solution
was treated with thrombin and passed again through a
nickel-chelating column in order to remove the cleaved
hisactophilin domain that was bound, whereas the
LOV2 domains were not retained
Figure 6 shows 13C NMR signals of the
recombi-nant LOV2 domain from A capillus-veneris (LOV2fern)
reconstituted with [U-13C17]FMN and with two isoto-pologue mixtures of FMN obtained by biotransforma-tion of [2-13C1]- or [3-13C1]glucose, respectively The left panel shows spectra that were acquired under dark conditions In the spectrum of protein reconstituted with universally13C-labeled FMN (Fig 6A), all signals with the exception of C(2) appear as broadened multi-plets due to 13C13C coupling of directly adjacent carbon atoms In the samples reconstituted with the isotopologue mixtures, the carbon signals of the bound FMN appear as singlets, and their apparent intensities vary over a wide range (Fig 6B,C) This intensity vari-ation is due to the presence of the singly 13C-labeled isotopologues at different abundances in the FMN iso-topologue mixtures The relative intensities of the indi-vidual carbon signals observed in the protein sample reflect the relative abundances of the different FMN isotopologues (cf.13C enrichments of the FMN speci-mens from the different 13C-labeled glucoses indicated
by filled circles in Fig 6) and constitute the basis for unequivocal signal assignment
Fig 4 Synthesis of isotopologue libraries of FMN from [2-13C 1 ]glucose 1, Ribulose 5-phosphate; 2, 3,4-dihydroxy-2-butanone 4-phosphate;
3, 6,7-dimethyl-8-ribityllumazine; 4, 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimdinedione; 5, riboflavin.
Trang 5For example, the position 8a methyl group, but not
the position 7a methyl group, is significantly labeled in
the sample of 3 obtained by biotransformation of
[2-13C1]glucose, and the signal detected at 23.2 p.p.m
in the spectrum with the isotopologue library from [2-13C1]glucose can be clearly assigned to C(8a) (Fig 6B) The C(7a) atom is not 13C-enriched from either [2-13C1]- or [3-13C1]glucose; therefore, no signal
Fig 5 Construction of a synthetic gene for
A sativa LOV1 domain Alignment of the wild-type DNA sequence (ASNPH1), and the synthetic DNA sequence (ASLOV1-syn) with 5¢ and 3¢ overhangs including the synthetic BglII and HindIII sites Changed codons are shaded in black New single restriction sites are shaded in grey Oligonucleotides used
as forward primers are drawn above, and reverse primers below, the aligned DNA sequences.
Fig 6 13 C NMR signals of 13 C-labeled FMN complexed to LOV2 domain from A capillus-veneris under dark conditions or under blue light irradiation A, [U-13C 17 ]FMN; B, FMN obtained from [2-13C 1 ]glucose; C, FMN obtained from [3-13C 1 ]glucose Asterisks indicate impurities.
Trang 6can be detected in the 13C NMR spectra of the
corre-sponding protein samples On the other hand, a second
methyl signal (doublet with a coupling constant of
43 Hz) is observed at 21.9 p.p.m in the spectrum with
[U-13C17]FMN as a cofactor It is immediately obvious
that this signal has to be assigned to C(7a)
Due to the specific 13C enrichments in the ribityl
moiety of the FMN samples, the signals for C(1¢),
C(2¢) and C(4¢) are observed in the isotopologue
mix-ture from [2-13C1]glucose, whereas only the signals for
C(2¢) and C(3¢) are detected in the spectrum of the
iso-topologue mixture from [3-13C1]glucose with higher
intensity of the C(2¢) signal On this basis, all ribityl
signals can be unequivocally assigned (Table 1 and
Fig 6)
Using the same isotopologue editing approach,
unequivocal signal assignments can be obtained for the
carbon atoms of the isoalloxazine ring Thus, label
from [2-13C1]glucose is diverted to the ring carbon
atoms 4a, 5a, 6 and 8 with 13C enrichments of
6 > 4a > 5a 8 (cf filled circles in Fig 6) The
car-bon atoms 4, 5a, 7, 8, 9a and 10a acquire 13C label
from [3-13C1]glucose with enrichments in the order of
5a 8 > 10a > 4 > 9a 7 Indeed, in the signal
region for aromatic carbon atoms (115–165 p.p.m), four signals were observed in the protein samples with FMN from [2-13C1]glucose (Fig 6B), and six signals were detected with FMN from [3-13C1]glucose (Fig 6C) The signal intensities were found to vary in the same pattern as determined for the free FMN iso-topologue mixture, and thus provided the basis for the assignments Additional validation is provided by the simultaneous detection of the signals for C(8), and C(5a) in both samples because both molecular posi-tions acquire 13C enrichment from [2-13C1]glucose, as well as from [3-13C1]glucose Due to low 13C enrich-ments of C(9) in the used FMN libraries, no signal should be detectable for that atom However, a signal for C(9) has to be present in the sample with [U-13C17]FMN and was indeed observed at 118.9 p.p.m (Fig 6A) In summary, the observed sig-nal intensities in the spectra with universally 13 C-labeled FMN and two isotopologue libraries of FMN (i.e obtained from biotransformation of [2-13C1]- and [3-13C1]glucose) allowed the assignments of all 17 car-bon atoms of FMN The results are summarized in Table 2 The validity of the experimental approach was confirmed by signal assignments using an ordered library of 13C-labeled FMN isotopologues More spe-cifically, we measured the 13C NMR chemical shifts
of seven selectively 13C-labeled FMN isotopologues
Table 1 13 C abundance of FMN obtained from [2- 13 C1]glucose and
FMN obtained from [3-13C 1 ]glucose bound to the LOV2 domain
from A capillus-veneris under dark and light conditions The
corre-sponding values with the isotopologue mixtures of free FMN are
given for comparison On the basis of the low signal-to-noise ratios
of the NMR spectra, the errors can be estimated as ± 30% of a
given 13 C abundance value.
Carbon
position
13 C abundance (%)
[2-13C 1 ]glucose [3-13C 1 ]glucose
Free
FMN
LOV2-bound FMN
Free FMN
LOV2-bound FMN
a
Not determined due to signal overlapping.bSignal observed at high
(+) and very high intensity (+ +) c Reference value.
Table 2 NMR chemical shifts of TARF, free FMN and FMN bound
to LOV2 domain from A capillus-veneris in dark and light condi-tions.
FMN atom
NMR chemical shifts (p.p.m.)
TARF
Free FMN
LOV-2 bound FMN
Trang 7bound to the LOV2 domain of A capillus-veneris
(Fig 7) The chemical shifts observed with these
sam-ples were in agreement with the signal assignments
made on the basis of the random isotopologue libraries
(Fig 6)
The interpretation of the NMR chemical shifts of
protein-bound flavins is usually based on the
compari-son with the chemical shifts of free flavins [1]
There-fore, the published assignments for free FMN [10]
were checked by the isotopologue abundance editing
method using labeled FMN obtained from [1-13C1
]-[2-13C1]- and [3-13C1]glucose The previous assignments
[10] of the isoalloxazine ring carbons could be
com-pletely confirmed (Table 2 and Fig 6) The previous,
tentative assignments of C(3¢) and C(4¢) [10] had to be
interchanged All 13C NMR assignments of
tetra-acetylriboflavin (TARF) were assigned by 2D
13C inadequate experiments using [U-13C17]TARF In
this case, the previous assignment for C(2¢) and C(4¢)
[18,19] had to be interchanged (Table 2)
The isotopologue editing method was then used to
assign the 13C NMR signals of FMN bound to the
LOV domains under study under blue light irradiation
conditions In order to keep photodamage of the
pro-tein as low as possible, the acquisition times were
somewhat reduced under blue light as compared to
dark conditions As a consequence, the signal-to-noise ratios of the NMR spectra of the irradiated samples were lower than those of the corresponding spectra in the dark (Fig 6, right column) The signal assignments obtained from the random isotopologue libraries matched those from the selectively labeled FMN sam-ples (Fig 7 and supplementary Fig S1) The results are presented in Table 2 and confirm previous assign-ments for LOV2 from A sativa (LOV2oat) [10], except that the chemical shifts due to C(7) and C(8), C(6) and C(9), as well as those due to C(3¢) and C(4¢), have to
be interchanged
In analogy to the above procedure, the chemical shifts of the LOV1 domain from A sativa (LOV1oat) were also investigated The results are given in supple-mentary Tables S1 and S2 and supplesupple-mentary Figs S4 and S5 In summary, the FMN signals of (LOV2oat) and (LOV1oat) were detected at similar chemical shifts (± 0.2 p.p.m), with the exception of the signals for C(5¢) and C(7a), which were upfield shifted by 0.5 p.p.m and 0.7 p.p.m., respectively, and the signals for C(2) and C(4), which were downfield shifted by 1.6 p.p.m and 0.7 p.p.m., respectively, in the LOV1 domain A correlation diagram of the chemical shifts for all proteins investigated in this study is shown in Fig 8
Fig 7 13 C NMR signals of of 13 C-labeled FMN with LOV2 domain from A capillus-veneris under dark conditions A, [U-13C 17 ]FMN; B, [xylene-13C 8 ]FMN; C, [7a,9- 13 C2]FMN; D, [6,8a- 13 C2]FMN; E, [4,10a- 13 C 2 ]FMN; F, [7,9a- 13 C 2 ]FMN; G, [5a,8-13C 2 ]FMN; H, [4a-13C 1 ]FMN Asterisks indicate impurities.
Trang 8The approach described in the present study opens a
new way to unambiguously assign all carbon atoms in
the 13C NMR spectra of protein-bound flavin
cofac-tors using no more than three FMN samples (i.e a
uniformly labeled and two partially labeled flavin
samples that can be biosynthetically obtained by in vivo biotransformation of [2-13C1]- and [3-13C1]glucose) Since, in the latter two cases, the degree of13C enrich-ment of a given carbon atom in the flavin samples dif-fers, the 13C NMR signal strength (amplitude) of a given carbon atom provides an additional constraint in the signal assignment procedure By contrast to previ-ous NMR work on flavoproteins [18], where variprevi-ous iosotopologues, selectively enriched in the xylene moiety of flavin, also were used [20], the isotopologues obtained by the new biosynthetic approach allow assignment of the carbon atoms of the ribityl side chain in the NMR spectra of flavin The 13C chemical shifts of these atoms can provide important informa-tion about the binding interacinforma-tion between the hydro-xyl groups of the side chain of flavin and the apoprotein, and could also report possible conforma-tional changes of the side chain (e.g due to reduction
of a flavoprotein)
In supplementary Table S2, the 13C chemical shifts
of LOV1 domain of A sativa in the two states are listed In the dark state, the chemical shifts of the iso-alloxazine moiety of flavin are very similar to those observed with LOV2 of the same species and of dif-ferent organisms (Fig 8) Most of the differences (± 0.3 p.p.m) between the two sets are within the accuracy limits of chemical shift determination, except for C(8) of LOV2fern, which is upfield shifted by 0.6 p.p.m., and C(8a) of LOV2fern, which is downfield shifted by 0.7 p.p.m., respectively, compared to LOV1oatand LOV2oat The chemical shifts of the side chain carbon atoms 1¢ and 3¢ of LOV2fernshow signifi-cant differences, which may be ascribed to variation in the strength of the hydrogen bond of the correspond-ing hydroxyl groups with the proteins and⁄ or to con-formational changes in the side chain (Fig 8) A similar effect is shown by the proteins in the blue light irradiated state The greatest difference in chemical shifts is observed for C(2) of LOV1oat, which is down-field shifted by 1.6 p.p.m compared to the LOV2 molecules Similarly, the C(4) and C(7a) of LOV1oat are downfield shifted by 0.7 p.p.m and upfield shifted
by 0.5 p.p.m., respectively, compared to LOV2oat and LOV2fern A significant difference is also observed for C(9) and C(1¢) of LOV2fern, which are upfield shifted
by 0.9 and 0.6 p.p.m., respectively, and C(2¢), which is downfield shifted by 0.7 p.p.m., compared to LOV1oat and LOV2oat
Based on extensive 13C and 15N NMR studies on free flavins in aprotic and protic media [21], which have shown that a direct correlation exists between the p-electron density and the 13C chemical shift of a particular atom of the flavin molecule, the observed
Fig 8 13 C Chemical shifts of 13 C-labeled FMN in complex with
LOV domains: black lines, dark conditions; blue lines, irradiated
with blue light.
Trang 9chemical shifts can be interpreted in terms of the
elec-tronic structure of the protein-bound flavin and its
perturbation by binding interaction and chemical
reac-tions Thus, the dark state interaction between FMN
and the LOV domains under study is characterized by
strong hydrogen bonding with the C(2)O group of
flavin The strength of the hydrogen bond corresponds
approximately to that of free FMN in water,
indi-cating polarization of the flavin along the axis
C(8)-C(6)-C(9a)-N(5)-C(10a)-C(2) This is manifested
by the observed downfield shifts of the corresponding
C atoms (Table 2 and supplementary Table S2)
Although there exists a hydrogen bond between the
protein and the flavin at C(4)O group, its strength is
considerably weaker than that observed in free FMN
in water These observations are in good agreement
with recent X-ray data showing a distance of 0.31 nm
between the Nd group of N998 and the oxygen atom
of the C(2)O group (Table 3) To the C(4)O group,
two hydrogen bonds (Negroup of Q1029, Ndgroup of
N1008) have been suggested by X-ray data, yet the
distance between the bond-forming atoms is larger
than that observed at C(2)O, supporting our
inter-pretation A strong hydrogen bond to C(4)O would
have influenced the chemical shift of the C(4a) atom
by a downfield shift compared to that of FMN [10]
The even slightly upfield shifted resonance of the
C(4a) atom in comparison to TARF indicates extra
p-electron density allocation to this position, released
from the N(10) atom, which is downfield shifted compared to TARF, as shown previously [10] The resonance position of C(4a) is thus in full agreement with the weak hydrogen bond observed at C(4)O The partial positive charge created on N(10) [21] by the release of electron density onto C(4a) is distributed mainly onto the C(5a) and the C(9) atom, and, to a lesser extent, onto the C(7) atom, in agreement with the fact that the latter atom experiences a smaller downfield shift than the other two atoms The data demonstrate that, with regard to the isoalloxazine moiety of flavin, there are only minor electronic differ-ences of the prosthetic group of the different proteins investigated in the present study Taking also into account the previously published 15N NMR data on LOV2 [10], it can be concluded that the chemical shift
of the N(5) atom of flavin indicates no hydrogen bond formation with this atom and a rather hydrophobic environment at this site This interpretation of the NMR data is fully supported by the X-ray data [22]
On the other hand, the chemical shift of the N(1) atom indicates the presence of a strong hydrogen bond at this site, although the X-ray data suggest the absence
of a hydrogen bond The opposite holds for the N(3) atom: the X-ray data propose a hydrogen bond with the protein whereas the 15N NMR data suggest the absence of such a bond
The chemical shifts of the C(3¢) and C(4¢) atoms
of the side chain of protein-bound flavin resemble
Table 3 Distances between FMN atoms (Ligand) and amino acid residues of LOV2 from Adiantum phy3 (chimeric fern photopreceptor) [12,13] For comparison, 13 C chemical shifts of bound FMN atoms are given.
Distance (A ˚ )
13 C NMR chemical shifts (p.p.m.)
Distance (A ˚ )
13 C NMR chemical shifts (p.p.m.)
Trang 10those of FMN in water, indicating stronger hydrogen
bonding interactions with the hydroxyl group at C(3¢)
and a somewhat weaker one with that at C(4¢)
compared to those observed in FMN This hydrogen
bonding pattern agrees with that observed by X-ray
crystallography [22] The resonance position due to
C(1¢) reflects the increased sp2 hybridization of N(10)
[21] Both C(2¢) and C(5¢) are upfield shifted by more
than 1 p.p.m compared to those of FMN Whereas
the X-ray data indicate no hydrogen bond between the
C(5¢)O group and the protein, but a strong one at
C(2¢)O, the NMR data do not indicate hydrogen
bonding at these atoms It is suggested that the
appar-ent absence of a hydrogen bond at C(2¢)O, as revealed
by 13C NMR, may be masked by a counteracting
factor, most probably a conformational change
Upon blue light irradiation of the proteins, rather
drastic changes are observed in the NMR spectra [10]
The most obvious one is the large upfield shift
()68.5 p.p.m) of the C(4a) resonance This proves the
conversion of the C(4a) atom from sp2 to sp3
hybrid-ization, in line with the formation of a covalent bond
between this atom of flavin, and the sulfur atom of
C966 in LOV2 of A capillus-veneris (Fig 2) The
reso-nance line of N(5) also undergoes a large upfield shift
()283 p.p.m) [10], indicating the change from an
aro-matic to an aliphatic nitrogen atom The other carbon
atoms of flavin most affected by conversion of the
pro-tein by light are: C(8) ()19.9 p.p.m), C(6) ()14.0 pm),
C(9a) ()6.7 p.p.m) and C(10a) (+ 5.9 p.p.m) All these
atoms are involved in the possible mesomeric
struc-tures of oxidized flavin [21] that are disturbed by the
C(4a) substitution The upfield shifts of the resonances
of these atoms, with the exception of C(10a), which
shows a downfield shifted signal, demonstrates the
allocation of the incoming electron density at these
positions The downfield shift of the resonance line
due to C(10a) is caused by the higher electron density
withdrawal from this atom by the further polarization
of the C(2)O group compared to that of the molecule
under dark conditions Since the sp2 hybridization of
the N(10) atom increases considerably upon formation
of the C(4a) adduct [10], the upfield shifts of the
reso-nances of C(5a) ()6.1 p.p.m) and C(7) ()2.9 p.p.m)
atoms can be ascribed to electron density release onto
these atoms from N(10) Overall, with regard to the
chemical shifts of the carbon atoms of the
isoalloxa-zine ring, the electronic structure of the different
pro-teins investigated in the present study is very similar, if
not almost identical Only the chemical shifts of the
C(2), C(4) and C(4a) atoms of the adduct of LOV1
and LOV2 proteins differ considerably from each
other The chemical shifts of the former protein are
downfield from those due to LOV2, indicating stronger hydrogen bonding in LOV1 than in LOV2 at these positions The hydrogen bond pattern as observed by NMR of the oxidized proteins is also observed in the adduct forms The hydrogen bond at N(3), as observed
by X-ray, is now also evident in the NMR data Whereas the chemical shifts of the C(4a) signals in LOV2oat and LOV1oat are very similar, the corre-sponding signal in LOV2fern is upfield shifted with respect to the former This observation suggests some structural difference(s) at the C(4a) position between the proteins from oat and that from fern
The hydrogen bonding pattern observed for the C(3¢) and the C(4¢) atoms of the ribityl side chain in the oxidized proteins is also observed in the corre-sponding adducts, but the strength of hydrogen bond-ing with the C(3¢)O group is considerably increased, especially that in LOV2fern With regard to C(5¢) atom, the LOV2 proteins exhibit similar chemical shifts for this atom, whereas that of LOV1 is upfield shifted by 0.5 p.p.m The chemical shift for the C(2¢) atom increases in the order: LOV2oat, LOV1oat and
LOV2-fern, possibly reflecting variations in the strength of hydrogen bonding interactions at this position
The13C NMR data show that there exist some subtle differences between the proteins investigated in the pres-ent study, as far as the isoalloxazine moiety of the flavin
in the resting and photo-adduct state is concerned The largest difference among the three proteins is observed for the resonance line of the C(2)O group (downfield shift) of LOV1 in the photo-adduct state However, there are some resonances of ribityl side chain carbon atoms, which differ to a greater extent among the three proteins (Fig 8), indicating variations in the interaction with the proteins and⁄ or conformational differences The LOV1 and LOV2 domains from A sativa have also been investigated by optical (light absorption, fluo-rescence, CD) and chemical techniques (kinetics) [9,23] The light absorption and 13C NMR data indicate a hydrophobic environment at the isoalloxazine moiety of the protein-bound flavin, whereas the fluorescence quan-tum yield of flavin bound to the two proteins reflect probably the sequence difference between the two pro-teins in the neighborhood of the flavin (residue 1010 is phenylalanine in LOV2 of A capillus-veneris and the positionally equivalent residue in LOV1 of A sativa is leucine 117) From these data, it can be concluded that the microenvironment of flavin in LOV2 is more hydro-phobic than that in LOV1 This property of the proteins appears to be correlated with the kinetic data, where it has been determined that LOV2 is more reactive to form the photo-adduct than LOV1 and its photo-adduct
is more stable than that of LOV1 [9] Moreover, the