Model Systems for Flavoenzyme Activity: Relationships between Cofactor Structure, Binding and Redox Properties

Article abstract of DOI:10.1021/ja036940b

A series of flavins were synthesized bearing electron-withdrawing and -donating substituents. The electrochemical properties of these flavins in a nonpolar solvent were determined. The recognition of these flavins by a diamidopyridine (DAP) receptor and the effect this receptor has on flavin redox potential was also quantified. It was found that the DAP-flavin binding affinity and the reduction potentials (E_1/2) for both the DAP-bound and unbound flavins correlated well with functions derived from linear free energy relationships (LFERs). These results provide insight and predictive capability for the interplay of electronics and redox state-specific interactions for both abiotic and enzymatic systems.

Full text of DOI:10.1021/ja036940b

Published on Web 11/27/2003
Model Systems for Flavoenzyme Activity: Relationships
between Cofactor Structure, Binding and Redox Properties
Yves-Marie Legrand,†,‡ Mark Gray,† Graeme Cooke,*,‡ and Vincent M. Rotello*,†
Contribution from Department of Chemistry, UniVersity of Massachusetts,
Amherst, Massachusetts 01003, and Centre for Biomimetic Design and Synthesis,
Chemistry,William H. Perkin Building, School of Engineering & Physical Sciences,
Heriot-Watt UniVersity, Riccarton, Edinburgh, UK EH14 4AS
Received June 27, 2003; E-mail: g.cooke@hw.ac.uk; rotello@chem.umass.edu
Abstract: A series of flavins were synthesized bearing electron-withdrawing and -donating substituents.
The electrochemical properties of these flavins in a nonpolar solvent were determined. The recognition of
these flavins by a diamidopyridine (DAP) receptor and the effect this receptor has on flavin redox potential
was also quantified. It was found that the DAP-flavin binding affinity and the reduction potentials (E1/2) for
both the DAP-bound and unbound flavins correlated well with functions derived from linear free energy
relationships (LFERs). These results provide insight and predictive capability for the interplay of electronics
and redox state-specific interactions for both abiotic and enzymatic systems.
Introduction
adenine dinucleotide (FAD), both derived from riboflavin
(vitamin B2).
Flavoenzymes are a ubiquitous class of proteins that catalyze
a variety of redox transformations including the oxidation of
amines to imines, thiols to disulfides and the hydroxylation
of aromatic species.1-4 Flavoproteins also mediate between
single-electron redox processes involving iron-heme and iron-
sulfur clusters and the obligate two-electron redox processes
of NADH. Flavoenzymes consist of two essential components,
the apoprotein and a flavin-based redox cofactor. The latter is
usually present as either flavin mononucleotide (FMN) or flavin
The apoprotein in flavoenzymes serves both to form a binding
pocket for the cofactor and to regulate cofactor redox properties.
The distinctive differences between flavin microenvironment
tune the redox properties of the cofactor to meet the function
of the given flavoenzyme. X-ray crystallography has provided
a great deal of information about the identities and relative
positions of the components of the flavoprotein microenviron-
ment, but does not yield direct insight into the mechanism or
redox/recognition properties of the enzymes.5
Biochemical researchers have employed artificial flavins
featuring electron-donating and -withdrawing substituents as
structural probes to explore the environment of flavin-binding
sites and as mechanistic probes that have led into new insights
into flavoenzyme function. In the former category artificial
flavins have been widely used for their spectral properties, where
the sensitivity of the flavin chromophore to changes in its local
dielectric has yielded valuable information about active-site
polarity.6 Moreover, synthetic flavins that exist in two or more
tautomeric forms have confirmed the presence or absence of
particular patterns of hydrogen bonds, permanent charges, and
strong dipoles through perturbation of the natural equilibria
between these states.7 In other work Massey and co-workers
have studied free energy changes associated with flavodoxins
reconstituted with a variety of artificial flavins. This work
examined the role microenvironmental effects of the binding
pocket have upon the redox couple of the cofactor, and the
related effects on cofactor protonation state and stability within
the active site.8
† University of Massachusetts.
‡ Heriot-Watt University.
(1) Walsh, C. Acc. Chem. Res. 1980, 13, 148-155.
(2) Ghisla, S.; Massey, V. Eur. J. Biochem. 1989, 181, 1-17.
(3) Bruice, T. C. Isr. J. Chem. 1984, 24, 54-61.
(4) Mueller, F. In Topics in Current Chemistry; Boschke, F. L., Ed.; Springer-
Verlag: Berlin, 1983; Vol. 108, pp 71-108.
(5) Ghisla, S.; Massey, V. Biochem. J. 1986, 239, 1-12.
(6) (a) Ghisla, S.; Massey, V.; Lhoste, J. M.; Mayhew, S. G. Biochemistry
1974, 13, 589-597. (b) Muller, F.; Mayhew, S. G.; Massey, V. Biochem-
istry 1973, 12, 4654-4662. (c) Massey, V.; Ganther, H. Biochemistry 1965,
4, 1161-1173. (d) Ghisla, S.; Massey, V. J. Biol. Chem. 1975, 250, 577-
584. (e) Biemann, M.; Claiborne, A.; Ghisla, V.; Massey, V.; Hemmerich,
P. J. Biol. Chem. 1983, 258, 5440-5448.
(7) See for example: (a) Moore, E. G.; Ghisla, S.; Massey, V. J. Biol. Chem.
1979, 254, 8173-8178. (b) Massey, V.; Ghisla, S.; Moore, E. G. J. Biol.
Chem. 1979, 254, 9640-9650.
(8) Ludwig, M. L.; Schopfer, L. M.; Metzger, A. L.; Pattridge, K. A.; Massey,
V. Biochemistry 1990, 29, 10364-10375.
(9) Blankenhorn, G. Eur. J. Biochem. 1976, 67, 67-80.
(10) Massey, V.; Nishino, T. In FlaVins and FlaVoproteins; Yagi, K., Yamano,
T., Eds.; University Park Press: Baltimore, 1980.
(11) Murthy, Y. V. S. N.; Meah, Y.; Massey, V. J. Am. Chem. Soc. 1999, 121,
5344-5345.
(12) Hasford, J. J.; Rizzo, C. J. J. Am. Chem. Soc. 1998, 120, 2251-2255.
(13) Edmondson, D. E.; Ghisla, S. In FlaVins and FlaVoproteins 1999;
Proceedings of the 13th International Symposium, Konstanz, Germany,
August 29-September 4, 1999; Ghisla, S., Kroneck, P., Macheroux, P.,
Sund, H., Eds.; Rudolf Weber: Agency for Scientific Publications, 1999;
pp 71-76.
In experiments designed to reveal mechanistic information,
it has been shown that flavin substitution can completely
eliminate reactivity.9 In other cases more subtle substitutions
can be made to change the redox potential of the cofactor, where
(14) For flavin electrochemical processes under aqueous conditions see: (a)
Muller, F. In Chemistry and Biochemistry of FlaVoenzymes; Muller, F.,
Ed.; CRC Press: Boca Raton, FL, 1991; Vol. 1, pp 1-71. (b) Stankovich,
M. T. In Chemistry and Biochemistry of FlaVoenzymes; Muller, F., Ed.;
CRC Press: Boca Raton, FL, 1991; Vol. 1, pp 401-425.
9
10.1021/ja036940b CCC: $25.00 © 2003 American Chemical Society
J. AM. CHEM. SOC. 2003, 125, 15789-15795
15789

A R T I C L E S
Legrand et al.
Scheme 1. Flavin Numbering and Common Oxidation/Protonation
States Observed for the Flavin Cofactor
Figure 1. (a) Lipoamide dehydrogenase active site.15 (b) DAP-flavin
complex.
simulate the overall dielectric of flavin binding compared with
aqueous environments. As in the natural analogues, flavin
binding by DAP causes a stabilization of the radical anion form
of the cofactor, manifested by a substantial cathodic shift in
the reduction potential. Here we report experiments using
artificial flavins bearing systematic functional group variations
at the 7 and 8 positions, allowing the simplicity of the DAP-
flavin model system to act as a benchmark for biochemical
experiments using intact flavoproteins. In these studies, we
demonstrate that the reduction potentials of the flavins free in
solution, the binding of DAP with flavin in the oxidized state,
and the reduction potentials of the resulting complexes can be
standardized using LFERs. To our knowledge this is the first
time the effects of subtle electronic modifications on redox-
modulated molecular recognition has been explored in a
systematic manner.
in one notable example, 7,8-dichloro-FAD was employed to
show that product release is the rate-determining step for
D-amino acid oxidase.10 Recently Massey showed that not only
could the reactivity of a flavoprotein be altered kinetically
through cofactor replacement, it could also be altered mecha-
nistically. In these studies, the NADPH-dependent reductase Old
Yellow Enzyme (OYE), functioned as an oxygen-dependent
desaturase when reconstructed with 8-cyano-FMN.11 Flavopro-
teins are, however, incredibly complex pieces of biochemical
machinery, and as such it is often difficult to determine precise
cause and effect relationships using enzymatic systems. One
way to address this problem is through the use of appropriate
model systems, where interactions between receptor and sub-
strate can be added and tuned systematically to address a given
problem.
In 1998, Rizzo and co-workers studied differences in the
reduction potentials within a series of flavins substituted at the
7 and/or 8 positions.12 They showed that a linear free energy
relationship (LFER) provided good correlation between the
flavin reduction potential and the electronic nature of the
substituents. In a later reappraisal of this work by Edmonson
and Ghisla, a refined model of flavin reduction was presented
that explicitly took into account electronic differences between
the 7 and 8 positions.13 Rizzo’s studies were carried out in
aqueous environments which differ from the flavin microenvi-
ronment provided by apoproteins in both the overall dielectric
and the absence of the controlled interactions. Additionally,
reduction of flavin in water initiates a series of processes that
lead inevitably to production of the flavohydroquinone anion
(Scheme 1).14 Therefore, this series of experiments is valuable
in the context of the many two-electron processes implicated
in flavin biochemistry; however, the rich single-electron chem-
istry of the flavoenzymes remains relatively unexplored by a
systematic LFER study. Additionally, these studies in aqueous
media do not allow the effect of specific interactions on flavin
redox properties to be determined.
Results and Discussion
(1) DAP-Flavin Binding. The association constants for all
10 flavins were determined with DAP using 1H NMR titration
(Table 1).17 In these studies, the downfield shift of the imide
(N(3)) proton resonance was fitted to a 1:1 binding isotherm
using iterative nonlinear curve-fitting methods, giving values
accurate to within 5%.
Analysis of the data shows that in general the measured
binding constant increases with increasing electron-donating
capacity of the substituents at the 7 and 8 positions of the flavin.
This can be rationalized in a qualitative fashion from a
hydrogen-bond donor-acceptor viewpoint.18 The nature of the
interaction between DAP and flavin is a three-point hydrogen-
bonded network.19 Of the three hydrogen bonds, flavin provides
two sites that act as hydrogen-bond acceptors, the O(2) and O(4)
carbonyl oxygens, but only one site that can operate as a
hydrogen-bond donor, the N(3) imide proton. Electron-donating
groups will have the effect of increasing the electron density
throughout the flavin system. As a direct consequence, this effect
will increase the electron density at the imide proton, making
this contact a poor hydrogen-bond donor. However the greater
electron density at the two carbonyl oxygen atoms will make
(16) For studies featuring multipoint interactions between flavin analogues and
artificial receptors, see: Kajiki, T.; Moriya, H.; Hoshino, K.; Kuroi, T.;
Kondo, S.; Nabeshima, T.; Yano, Y. J. Org. Chem. 1999, 64, 9679-9689.
(17) Ka and E1/2 values were determined in noncompetitive solvents with similar
dielectric constants (CDCl3 and CH2Cl2) due to their suitability for NMR
and electrochemical work, respectively. Previously, we have shown that
neither the interchange between these solvents nor the presence of the
electrolyte necessary for conducting the electrochemical experiments has
any measurable effect on the results obtained. See: Deans, R.; Niemz, A.;
Breinlinger, E. C.; Rotello, V. M. J. Am. Chem. Soc. 1997, 119, 10863-
10864.
(18) (a) Stone, A. J. The Theory of Intermolecular Forces; Clarendon Press:
Oxford, 1996. (b) Scheiner, S. Hydrogen Bonding: A Theoretical Perspec-
tiVe; Oxford University Press: New York, 1997.
(19) Breinlinger, E.; Niemz, A.; Rotello, V. M. J. Am. Chem. Soc. 1995, 117,
5379-5380.
Previously, we have shown that diaminopyridine derivatives
(e.g., DAP) may be used as a simple, yet informative, mimic
of apoproteins.15 These receptors recognize the flavin through
three-point hydrogen bonding to the imide moiety of the cofactor
in a manner reminiscent of most natural flavoproteins such as
lipoamide dehydrogenase (Figure 1).16 Furthermore nonpolar
aprotic solvents (e.g., methylene chloride and chloroform) better
(15) Niemz, A.; Rotello, V. M. Acc. Chem. Res. 1999, 32, 44-52.
9
15790 J. AM. CHEM. SOC. VOL. 125, NO. 51, 2003

Model Systems for Flavoenzyme Activity
A R T I C L E S
Table 1. Summary of Electrochemical and Binding Data for Substituted Flavins
flavin
number
C7
(meta)
C8
(para)
flavinox + DAP
K (M-1 b
flavin
flavin + DAP
1/2 (mV)d
flavinred+DAP
1/2 (mV)c
E
K (M-1 e
)
a
C7 σm
C8 σpa
)
E
a
a
1a
1b
1c
1d
1e
1f
Br
Br
0.37
0.37
-0.06
0.00
-0.06
0.37
0.37
-0.06
0.10
0.26
0.24
365
-1039
-1048
-1163
-1206
-1235
-1231
-1258
-1286
-1358
-1378
-944
-949
14 800
17 100
14 500
18 900
15 400
9 500
15 900
18 600
13 700
17 100
Cl
Cl
362
404
432
480
456
792
517
606
960
Me
H
Me
Cl
Cl
Me
MeO
Me
Cl
H
H
MeO
NMe2
Me
MeO
NMe2
0.24
-1071
-1109
-1146
-1153
-1181
-1194
-1278
-1304
0.00
0.00
-0.28
-0.83
-0.14
-0.28
-0.83
1g
1h
1i
1j
-0.06
a Values obtained from Hansch, Leo and co-workers.21 b Values (5%, CDCl3, 23 °C. c [Flavin] ) 5 mM; E1/2 ( 3 mV vs ferrocene, CH2Cl2, 23 °C, 0.1
M TBAP used as electrolyte. d [Flavin] ) 5 mM, [DAP] ) 50 mM; E1/2 ( 3 mV vs ferrocene, CH2Cl2, 23 °C, 0.1 M TBAP used as electrolyte. e Obtained
1/2
1/2
from the relationship Ka(red)/Ka(ox) ) exp[(nF/RT)(E (bound) - E (unbound))].
them able to form stronger hydrogen bonds with the amide
protons of the receptor. In other words, electron-donating groups
diminish the strength of one hydrogen bond while simulta-
neously promoting the other two, with the net result being that
substitution of this type enhances DAP-flavin complexation.20
(2) Electrochemical Studies. We next compared the redox
behavior of the flavins both alone and in the presence of DAP.
The experiments were carried out using cyclic voltammetry and
square-wave voltammetry methods, using the ferrocene-fer-
rocenium couple as an internal standard.17 The half-wave reduc-
tion potentials (E1/2) corresponding to the conversion of the
oxidized substrate to the corresponding radical anion for both
the receptor-bound and free flavins are recorded in Table 1; a
10-fold excess of receptor was used to ensure complete binding.
As expected, substitution at the 7 and 8 positions of the flavin
results in pronounced differences in the values of E1/2, with
electron-donating and -withdrawing substituents making the
reduction potentials more and less negative, respectively.
Another dramatic difference in the redox behavior of the
DAP-bound flavins compared to the free flavins lie in the shapes
of the cyclic voltammograms (Figure 2). For the free “natural”
flavin (7,8-dimethyl flavin 1h) one reduction wave and two
distinct reoxidation waves were observed. In an earlier study,22
we showed that this behavior arose from an electrochemical-
chemical-electrochemical (e-c-e) process where the radical
anion formed at the electrode surface rapidly deprotonates
oxidized flavin in the bulk medium. The protonated flavin
radical produced in this process undergoes a further one-electron
reduction at the electrode surface to form the relatively stable
fully reduced flavin anion. In the present study, we observed
that the tendency to undergo this process depends strongly upon
the reduction potential of each flavin: those with strongly
electron-donating substituents (Figure 2a) have the most negative
reduction potentials and also show the most pronounced e-c-e
behavior. In contrast, the low reduction potentials caused by
electron-withdrawing substituents give rise to fully electro-
chemically reversible formation of the radical anion. The reason
for this change in behavior lies in changes in basicity of the
N(5) position of the flavin radical anion, with electron-
withdrawing groups having the effect of removing much of the
charge from this nitrogen and thus lowering the pKa at this
position to below that of the imide of the oxidized form.
Figure 2. Cyclic voltammetry responses of selected 7,8-substituted flavins
vs ferrocene, CH2Cl2, 23 °C, 0.1 M TBAP used as electrolyte. The solid
lines indicate each flavin free in solution (5 mM), the dashed lines are the
same flavins in the presence of DAP (50 mM). (a) 1j, (b) 1h, (c) 1d, (d)
1b.
In contrast, the redox behavior for all of the flavins in the
presence of DAP are much simpler, displaying only a single
electrochemical wave in their voltammograms, indicative of a
fully electrochemically reversible conversion to and from the
radical anion. There are two reasons for this difference in
electrochemical behavior: (i) binding to DAP effectively blocks
the imide proton of the oxidized flavin, preventing proton
transfer between the flavin radical anion and the fully oxidized
flavin and (ii) binding to flavin in the radical anionic form
dramatically increases the extent of delocalization of the anion
into the pyrimidine moiety, significantly lowering the reactivity
of the N(5) position.
(20) For substituent effects and hydrogen bonding, see: Wilcox, C. S.; Kim,
E.; Romano, D.; Kuo, L. H.; Burt, A. L.; Curran, D. P. Tetrahedron 1995,
51, 621-634.
(21) Hansch, C.; Leo, A.; Unger, S. H.; Kim, K. H.; Nikaitani, D.; Lien, E. J.
J. Med. Chem. 1973, 16, 1207.
(3) Linear Free Energy Analysis and Biochemical Impli-
cations. As mentioned previously, Rizzo and co-workers linked
9
J. AM. CHEM. SOC. VOL. 125, NO. 51, 2003 15791

A R T I C L E S
Legrand et al.
Figure 3. Simplification of the flavin radical anion into a pseudo-phenyl
system.
the reduction potential of a series of flavins to a LFER.12 In
this study, the aromatic ring of the flavin was treated as a phenyl
system, with the N(5) position bearing the bulk of the negative
charge upon reduction to the radical anion form. In comparison
to this position, the C(7) and C(8) positions of a flavin are
analogous to the meta and para positions respectively of a
simpler phenyl system (Figure 3).
Figure 4. Correlation between experimental and linear free energy derived
values for the reduction potentials of flavins in methylene chloride free in
solution. Slope ) 1.01 (except for methoxy-substituted flavins), r2 ) 0.9831.
This simplification of the flavin system allowed Rizzo to use
the corresponding Hammett σ constants to correlate his data
through the relationship:
Table 2. Summary of Experimental and LFER Predicted Free
Energy Changes for Reduction of Flavins 1a-h
flavin
a
number
C7 σmFm
C8 σpFpb
∆G c (kcal/mol)
∆G d (kcal/mol)
calc
exp
E1/2(X,Y) ) F{σm(X) + σp(Y)} + E1/2(H,H)
(1)
1a
1b
1c
1d
1e
1f
1g
1h
1i
-2.55
-2.55
0.41
0.00
0.41
-2.55
-2.55
0.41
-0.69
0.41
-1.20
-1.10
-1.10
0.00
0.00
1.29
3.82
0.64
1.29
3.82
24.1
24.2
27.1
27.8
28.2
26.6
29.1
28.9
28.4
32.0
24.0
24.2
26.8
27.8
28.5
28.4
29.0
29.7
31.3
31.8
While this treatment gave good qualitative agreement between
the experimental data and values predicted from this relationship,
Edmonson and Ghisla showed that better correlation could be
achieved for the prediction of flavin reduction potentials in an
aqueous environment through a simple separation of the
Hammett parameters, such that each position had an independent
sensitivity factor (F):13
1j
a Energetic contribution from the C7 substituent, values obtained using
Fm ) -6.9 kcal/mol/σ. b Energetic contribution from the C8 substituent,
values obtained using Fp ) -4.6 kcal/mol/σ. c Total LFER predicted energy
change for reduction, obtained using eq 3. d Experimental free energy change
for reduction, obtained from values listed in Table 1 using the relationship
E1/2(X,Y) ) Fmσm(X) + Fpσp(Y) + E1/2(H,H)
(2)
We applied a variation of the latter method to analyze the
data from our experiments, with replacement of electrochemical
half-wave potentials with ∆G values, allowing us to analyze
∆G ) -nFE1/2
.
1
of the medium and the difference in hydrogen bonding between
the aqueous and nonaqueous environments.
both the electrochemical and H NMR derived data sets using
one general equation:
The results obtained from 1H NMR titrations for the DAP-
flavin complex were analyzed using eq 3 after unit conversion.
It was found that the experimental data were fitted best when
the LFER was substituted with the values Fm ) 0.15 kcal/mol/σ
and Fp ) 0.45 kcal/mol/σ; thus, for binding to DAP the C(8)
position dominates. A graph showing the correlation between
the values obtained from these optimized F values and those
determined experimentally is presented in Figure 5 (see also
Table 3). Again an excellent correlation of r2 > 0.98 is observed
with the range of free energies produced, showing that effective
communication does take place between the imide portion of
the flavin and the distant C(7) and C(8) positions.
The final set of experimental data corresponding to flavin
reduction in the presence of the receptor was also fitted to the
standardized LFER (eq 3). Optimized values for the two
sensitivity parameters were found to be identical to those
obtained for the unbound flavins with Fm ) -6.9 kcal/mol/σ
and Fp ) -4.6 kcal/mol/σ (Figure 6, see also Table 4), with
the only difference being that the values are offset by 3 kcal/
mol. This suggests that the two events occurring within the
electrochemical cell, namely binding between flavin and receptor
and reduction of the cofactor to the radical anion, are indepen-
∆G(X,Y) ) Fmσm(X) + Fpσp(Y) + ∆G(H,H)
(3)
Using the generalized LFER of eq 3, the system-specific F
values were iteratively varied to give an optimized relationship
between the experimental and Hammett-derived values for the
reduction of the free flavins in an aprotic solvent (Table 2).
Overall the best fit for both the bound and unbound reduction
potentials was found for Fm ) -6.9 kcal/mol/σ and Fp ) -4.6
kcal/mol/σ, suggesting that flavin reduction is more sensitive
to electronic changes at the C(7) position than C(8). The only
data that did not correlate well using these values were those
where MeO was at C(8), where presumably the bulky nature
of this substituent interferes with resonance donation. If these
points are excluded from the analysis, r2 > 0.98 is found for
both the bound and unbound flavin reductions (Figure 4). These
results contrast strongly with those reported by Edmonson and
Ghisla for the reduction of flavins in an aqueous environment,
where changes at the C(8) position were shown to be the
dominant factor in controlling flavin electrochemistry (Fm
-3.2 kcal/mol/σ, Fp ) -4.7 kcal/mol/σ). We believe this change
in behavior arises from a combination of the change in dielectric
)
9
15792 J. AM. CHEM. SOC. VOL. 125, NO. 51, 2003

Model Systems for Flavoenzyme Activity
A R T I C L E S
Table 4. Experimental and LFER Predicted Free Energy Changes
for the Flavin Reduction in the Presence of DAP
flavin
number
a
C7 σmFm
C8 σpFpb
∆G c (kcal/mol)
∆G d (kcal/mol)
calc
exp
1a
1b
1c
1d
1e
1f
1g
1h
1i
-2.55
-2.55
0.41
0.00
0.41
-2.55
-2.55
0.41
-0.69
0.41
-1.20
-1.10
-1.10
0.00
0.00
1.29
3.82
0.64
1.29
3.82
21.8
21.9
24.9
25.6
26.0
24.3
26.8
26.6
26.2
29.8
21.8
21.9
24.7
25.6
26.4
26.6
27.2
27.5
29.5
30.1
1j
a Energetic contribution from the C7 substituent, values obtained using
Fm ) -6.9 kcal/mol/σ. b Energetic contribution from the C8 substituent,
values obtained using Fp ) -4.6 kcal/mol/σ. c Total LFER predicted energy
change for reduction, obtained using eq 3. d Experimental free energy change
for reduction, obtained from values listed in Table 1 using the relationship
Figure 5. Plots showing correlation between experimental and linear free
energy derived values for the change in free energy upon binding between
flavins and DAP in CDCl3. Slope ) 1.02, r2 ) 0.9841.
∆G ) -nFE1/2
.
Table 3. Experimental and LFER-predicted Free Energy of
DAP-Flavin Association in CDCl3
flavin
number
a
C7 σmFm
C8 σpFpb
∆G c (kcal/mol)
∆G d (kcal/mol)
calc
exp
1a
1b
1c
1d
1e
1f
1g
1h
1i
0.06
0.06
-0.01
0.00
-0.01
0.06
0.06
-0.01
0.02
0.12
0.11
0.11
0.00
0.00
-0.13
-0.37
-0.06
-0.13
-0.37
-3.42
-3.43
-3.49
-3.59
-3.60
-3.66
-3.91
-3.67
-3.70
-3.98
-3.49
-3.49
-3.55
-3.59
-3.66
-3.63
-3.95
-3.70
-3.79
-4.07
Figure 7. Thermodynamic cycle showing the relationship between the
flavin-DAP binding and reduction processes.
1j
-0.01
(gained upon reduction) back into areas of the molecule where
the HOMO has vacated, maintaining wave function orthogonal-
ity.23 This redistribution in turn will cause significant disruption
in the extent of communication between the phenyl and
pyrimidine subunits of the flavin.
a Energetic contribution from the C7 substituent, values obtained using
Fm ) 0.15 kcal/mol/σ. b Energetic contribution from the C8 substituent,
values obtained using Fp ) 0.45 kcal/mol/σ. c Total LFER predicted energy
change upon binding, obtained using eq 3. d Experimental free energy
change upon binding, obtained from values listed in Table 1 using the
relationship ∆G ) -RT ln Ka.
One way to test this hypothesis of independence and
noncommunication between the phenyl and pyrimidine subunits
of the flavin is by examination of the binding constants between
receptor and cofactor in the reduced form. If the hypothesis is
correct, this lack of communication would diminish the cor-
relation between experimental values and those based upon the
standard LFER. Although we were not able to quantify the
interaction between the flavin radical anion and DAP through
direct experimental observation, we were able to obtain values
for this process through use of a relationship that makes direct
use of the three experimental parameters:
Ka(red)/Ka(ox) ) exp[(nF/RT)(E1/2(bound) -
E1/2(unbound))] (4)
Using a thermodynamic cycle (Figure 7) it is possible to
derive a LFER that corresponds to the free energy change
associated with binding DAP to the flavin radical anion (∆G3-
(X,Y)) from the optimized parameters for (i) the binding to
flavin in the oxidized form (∆G1(X,Y)), (ii) the reduction
processes corresponding to the DAP-bound flavin (∆G2(X,Y)),
and (iii) the free flavin (∆G4(X,Y)). Inspection of this cycle
tells us that if the level of electronic communication between
Figure 6. Plot showing correlation between experimental and linear free
energy derived values for the reduction potentials of flavins in methylene
chloride in the presence of DAP. Slope ) 1.07 (methoxy-substituted flavins
not included), r2 ) 0.9884.
dent. This is plausible due to the nature of the hydrogen-bonding
system at the molecular orbital level. Interaction with the
receptor causes a significant polarization of the HOMO away
from N(5) to maximize the hydrogen-bond ability of the O(2)
and O(4) positions. Moreover, this causes a redistribution of
the spin density of the SOMO containing the unpaired electron
(22) Niemz, A.; Imbriglio, J.; Rotello, V. M. J. Am. Chem. Soc. 1997, 119,
887-892.
(23) Niemz, A.; Rotello, V. M. J. Am. Chem. Soc. 1997, 119, 6833-6836.
9
J. AM. CHEM. SOC. VOL. 125, NO. 51, 2003 15793

A R T I C L E S
Legrand et al.
The experiments described herein show that in the absence
of any additional specific interactions, in media consistent with
the dielectric of the flavin binding site of most flavoproteins,
structural variations at the C(7) and C(8) positions of the flavin
structure lead to a series of well defined and predictable changes
in the properties of the cofactor. First, binding between the imide
functionality of the cofactor and a model of the residues that
would bind to these positions in a native protein, is affected
strongly by the changes examined, meaning that a significant
degree of electronic delocalization exists through the full extent
of the neutral molecule. Moreover this suggests that binding
to the corresponding residues in the active site would inter-
play strongly with microenvironmental effects that would
perturb this delocalization in the form of secondary specific
interactions.
Figure 8. Plot showing the lack of correlation between experimental and
linear free energy-derived values for the DAP binding of flavins in the
reduced form.
Second, the single-electron reduction of both the free and
bound flavins to the corresponding radical anions also relates
strongly (but differently) to the identity of the substituents at
the C(7) and C(8) positions. Third, while the sensitivity factor
for substituents at C(8) remains essentially constant for the
reduction in both aprotic (Fp ) -4.6 kcal/mol/σ) and aqueous
environments (Fp ) -4.7 kcal/mol/σ),13 the contribution from
the C(7) position is much more variable. The sensitivity factor
for the latter position in a low dielectric medium (e.g., CH2Cl2,
Fm ) -6.9 kcal/mol/σ), which mimics the binding pocket of a
natural flavoenzyme, is twice that of measurements recorded
in water (Fm ) -3.2 kcal/mol/σ).13 As the C(7) position is out
of direct conjugation with N(5), the role differing substituents
(attached at C(7)) have on modulating the flavin’s properties
has been largely overlooked. However, the variability in
sensitivity on moving from pure association with the binding
site to electrochemical activity suggests that more importance
should be given to this site than at present when designing
experiments to modify and tune flavoprotein reactivity.
Table 5. Summary of Experimental and LFER Predicted Free
Energy Changes for the Free Energy Change upon Association of
Flavins in the Radical Anionic Form and DAP
flavin
number
a
C7 σmFm
C8 σpFpb
∆G c (kcal/mol)
∆G d (kcal/mol)
calc
exp
1a
1b
1c
1d
1e
1f
1g
1h
1i
0.06
0.06
-0.01
0.00
-0.01
0.06
0.06
-0.01
0.02
0.12
0.11
0.11
0.00
0.00
-0.13
-0.37
-0.06
-0.13
-0.37
-5.66
-5.67
-5.73
-5.83
-5.84
-5.90
-6.15
-5.90
-5.94
-6.21
-5.68
-5.77
-5.68
-5.83
-5.71
-5.42
-5.73
-5.82
-5.64
-5.77
1j
-0.01
a Energetic contribution from the C7 substituent, values obtained using
Fm ) 0.15 kcal/mol/σ. b Energetic contribution from the C8 substituent,
values obtained using Fp ) 0.45 kcal/mol/σ. c Total LFER predicted energy
change upon binding, obtained using eq 3. d Experimental free energy
change upon binding, obtained from values listed in Table 1. using the
relationship ∆G ) -RT ln Ka.
Conclusions
the phenyl and pyrimidine moieties of the flavin remains
constant, we then obtain eq 5. Substituting the optimized
parameters into this relationship gives eq 6, which simplifies
to give a LFER with the same Fm and Fp values as were
optimized for the binding to the oxidized flavins (eq 7).
Ten flavins bearing a variety of substitutions at the C(7) and
C(8) positions were synthesized. It was found that these
substitutions had marked effects on (i) the redox properties of
the flavin, (ii) the affinity for a synthetic receptor designed to
mimic the flavins enzymatic partner, and (iii) the redox
properties of the bound species. Subsequently, it was found that
the trends in all three physical properties affected could be
analyzed through use of linear free energy relationships. These
relationships showed that the redox behavior of the flavins had
greater sensitivity to substituents at the C(7) position than C(8).
For binding to the oxidized flavin the opposite effect was
observed with C(8) substituents modulating the affinity for the
receptor to a much greater extent than C(7) substituents. Further
analysis of the data showed that in the reduced form the
communication between the C(7) and C(8) positions of the flavin
and the imide moiety was effectively eliminated.
∆G1(X,Y) + ∆G2(X,Y) ) ∆G3(X,Y) + ∆G4(X,Y) (5)
(0.15σm(X) + 0.45σp(Y) + ∆G1(H,H)) + (∆G2(H,H) -
6.9σm(X) - 4.6σp(Y)) ) ∆G3(X,Y) + (∆G4(H,H) -
6.9σm(X) - 4.6σp(Y)) (6)
∆G3(X,Y) ) 0.15σm(X) + 0.45σp(Y) + ∆G1(H,H) +
∆G2(H,H) - ∆G4(H,H) (7)
Experimental values for the free energy change associated
with DAP-flavinred binding obtained using eq 4 are compared
to LFER-derived predicted values from eq 7 in Table 5 and
also graphically in Figure 8. As anticipated, there is little sim-
ilarity between the two data sets, with the experimental values
scattered over a very small energy range compared to values
predicted from eq 7. This shows that upon reduction of the flavin
to the radical anion, communication between the phenyl and
pyrimidine subsystems is effectively eliminated.
To our knowledge the work presented here provides the most
controlled examination of substitution effects on the electro-
chemical properties of a series of flavins yet reported. In addition
this paper is the first to explicitly examine a redox modulated
molecular recognition process in detail using linear free energy
relationships and, is thus a key step on the path towards the
rational design of electrochemically controllable switches and
9
15794 J. AM. CHEM. SOC. VOL. 125, NO. 51, 2003

Model Systems for Flavoenzyme Activity
A R T I C L E S
each curve gave excellent agreement with the 1:1 binding isotherm,
using the relationship
sensors. It is hoped that this work will likewise prove valuableto
those workers employing flavins and their analogues as mecha-
nistic probes and as therapeutic agents.
∂
obs ) ∂H +
1/2
2 - 4[Ht] + [Gt]
1
1
Ka
Currently we are continuing our investigations in this area
through extensive use of computational methods to achieve a
greater understanding of the root causes of the observed
behavior. Further experimental work will focus on the effects
these substituents have upon the affinity for more complex
guests that better mimic the flavoprotein active sites through
employment of specific π-π and donor atom-π interactions
with the flavin system. The results of these investigations will
be disclosed in due course.
(∂HG - ∂H) [Ht] + [Gt] +
-
[Ht] + [Gt] +
((
K ) ((
)
) )
a
2[Ht]
where [Gt] and [Ht], correspond respectively to the total concentrations
of guest and host; δobs, the observed chemical shift, δH represents the
chemical shift of the imide proton of the host in the absence of guest,
and δHG the chemical shift of the followed proton in the fully bound
complex, estimated from extrapolation from the limiting experimental
value.26
Electrochemistry. All electrochemical experiments were carried out
on a Cypress System potentiostat. A 1-mm platinum button and a gold-
plated electrode were utilized as working and counter electrodes,
respectively. A silver wire pseudoreference electrode was used, and
all potentials are referenced versus the ferrocene/ferrocenium couple.
The sweep rate was 200 mV/s, and the studies were run on an argon-
purged temperature-controlled cell. Solutions of the flavin being studied
alone and of the comparable complex with DAP were prepared,
maintaining a constant concentration of the flavin. Maximal shifts were
observed at ∼5 equiv of DAP; 10 equiv was used to ensure complete
binding. The solutions were degassed by bubbling argon through them
for at least 10 min, at which time cyclic and square wave voltammo-
grams were recorded (for representative voltammograms, see Supporting
Information).
Experimental Section
Materials and General Methods. Solutions utilized in electro-
chemical experiments were prepared using reagent grade CH2Cl2 dried
via distillation over CaH2. Tetrabutylammonium perchlorate (TBAP,
obtained from SACHEM, electrometric grade) was dissolved in CHCl3,
washed with distilled water, recrystallized twice from ethyl acetate,
and dried for several days under high vacuum. Other chemicals were
reagent grade, obtained from Aldrich, and used without further
purification.
NMR Titrations (General).24 These experiments were performed
in a noncompetitive solvent under constant concentration conditions
of the individual flavin at 298 K. Flavin stock solution (CDCl3 solvent;
2 mL, 5 mM) was used to prepare 20 mM solutions of the receptor
molecule (DAP ) 8.85 mg). Then 600 µL of the stock solution was
transferred to an NMR tube and the spectrum of this solution recorded
in a Bruker AC200 200 MHz NMR spectrometer. Aliquots of the
receptor solution were added, with spectra recorded for each addition,
and the incremental downfield change in chemical shift of the imide
proton noted.
Acknowledgment. We thank Justin Fermann of the Univer-
sity of Massachusetts Chemistry Department for helpful dis-
cussions. This research was supported by the NIH (GM59249-
04), the NSF (CHE-0213354) and Heriot-Watt University.
Supporting Information Available: Synthesis of flavins and
cyclic voltammetry for flavins and flavin-DAP complexes
(PDF). This material is available free of charge via the Internet
at http://pubs.acs.org.
The resulting data were analyzed to reveal the binding constant Ka
by application of nonlinear least-squares curve fitting.25 The data for
JA036940B
(24) NMR dilution experiments show that DAP and the flavins do not dimerize
or aggregate under the experimental conditions used in our study.
(25) Long, J. R.; Drago, R. S. J. Chem. Soc. 1982, 59, 1037-1039.
(26) For further information see: Conners, K. Binding Constants; Wiley and
Sons: New York, 1987.
9
J. AM. CHEM. SOC. VOL. 125, NO. 51, 2003 15795