Model Systems for Flavoenzyme Activity. Control of Flavin Recognition via Specific Electrostatic Interactions

Article abstract of DOI:10.1021/ol015838l

matrix presented A model system has been used to study the interactions of dipole-containing aromatic systems with oxidized and reduced flavin. Ab initio computational and experimental studies show that dipole orientation within the host is a critical determinant for recognition and redox behavior of the flavin guest.

Full text of DOI:10.1021/ol015838l

ORGANIC
LETTERS
2001
Vol. 3, No. 10
1531-1534
Model Systems for Flavoenzyme
Activity. Control of Flavin Recognition
via Specific Electrostatic Interactions
Allan J. Goodman, Eric C. Breinlinger, Catherine M. McIntosh,
Lisa N. Grimaldi, and Vincent M. Rotello*
Department of Chemistry, UniVersity of Massachusetts, Amherst, Massachusetts 01003
rotello@chem.umass.edu
Received March 13, 2001
ABSTRACT
A model system has been used to study the interactions of dipole-containing aromatic systems with oxidized and reduced flavin. Ab initio
computational and experimental studies show that dipole orientation within the host is a critical determinant for recognition and redox behavior
of the flavin guest.
The interaction of aromatic rings with aromatic1 and non-
aromatic moieties2 is an important motif in molecular
recognition.3 These complex interactions are composed of
dipoles, quadrupoles, and higher order influences4 that can
be either attractive or repulsive in nature, differing dramati-
cally in both their distance dependence and their direction-
ality.5 The numerous variables involved results in a com-
plicated additivity of forces, making direct interpretation and
prediction of these interactions difficult.
Side chain-flavin and substrate-flavin interactions are a
common motif in flavoenzyme structures, with aromatic
stacking6 and aromatic-dipole interactions both prominent
features. One example of the importance of specific flavin-
protein aromatic stacking is found in the DesulfoVibrio
Vulgaris flavodoxin (Figure 1a), where Swenson et al. have
shown that the Y98F active site mutation results in a
substantial change in the recognition and redox activity of
the adjacent FMN cofactor.7 In this system, as well as other
flavoenzymes, both dipolar and quadrupolar interactions are
possible between the electron-poor flavin and adjacent
(1) Hunter, C. A. Chem. Soc. ReV. 1994, 23, 101-109. Coates, G. W.;
Dunn, A. R.; Henling, L. M.; Dougherty, D. A.; Grubbs, R. H. Angew.
Chem., Int. Ed. Engl. 1997, 36, 248-251. Greaves, M. D.; Galow, T. H.;
Rotello, V. M. Chem. Commun. 1999, 169-170. Gellman, S. H. Chem.
ReV. 1997, 97, 1231-1232. Martin, C. B.; Mulla, H. R.; Willis, P. G.;
Cammers-Goodwin, R. J. Org. Chem. 1999, 64, 7802-7806.
(2) Scrutton, N. S.; Raine, A. R. C. Biochem. J. 1996, 319, 1-8.
Dougherty, D. A. Science 1996, 271, 163-168. Madhavi, N. N. L.; Katz,
A. K.; Carrell, H. L.; Nangia, A.; Desiraju, G. R. Chem. Commun. 1997,
1953-1954. Adams, H.; Carver, F. J.; Hunter, C. A.; Osborne, N. J. Chem.
Commun. 1996, 2529-2530. Adams, H.; Harris, K. D. M.; Hembury, G.
A.; Hunter, C. A.; Livingstone, D.; McCabe, J. F. Chem. Commun. 1996,
2531-2532. Rotello, V. M. Heteroatom Chem. 1998, 9, 605-606.
(3) Breslow, R. Acc. Chem. Res. 1995, 28, 146-153. Gellman, S. H.
Chem. ReV. 1997, 97, 1231-1232.
(5) Glusker, J. P. Directional aspects of intermolecular interactions; 1998;
Vol. 198, pp 1-56.
(6) Zhou, Z. M.; Swenson, R. P. Biochemistry 1996, 35, 15980-15988.
(7) Swenson, R. P.; Krey, G. D. Biochemistry 1994, 33, 8505-8514.
Chang, F. C.; Swenson, R. P. Biochemistry 1997, 36, 9013-9021. Tam-
Chang, S. W.; Mason, J.; Iverson, I.; Hwang, K. O.; Leonard, C. Chem.
Commun. 1999, 65-66.
(4) Ma, J. C.; Dougherty, D. A. Chem. ReV. 1997, 97, 1303-1324. Cozzi,
F.; Ponzini, F.; Annunziata, R.; Cinquini, M.; Siegel, J. S. Angew. Chem.,
Int. Ed. Engl. 1995, 34, 1019-1020. Chen, C. T.; Siegel, J. S. J. Am. Chem.
Soc. 1994, 116, 5959-5960.
10.1021/ol015838l CCC: $20.00 © 2001 American Chemical Society
Published on Web 04/20/2001

aromatics in relation to the redox active flavin nucleus
(Figure 1b). This model utilizes hydrogen bonding to align
the flavin nucleus over aromatic groups featuring oriented
dipolar functionality, without significantly changing the
overall electron density of the aromatic stacking moiety. This
allows the direct quantification of dipole/quadrupole-flavin
energetics in reference to unsubstituted receptor 1. Com-
parison of the binding of oxidized and reduced species then
affords insight into the stabilizing and destabilizing effects
of dipolar interactions on flavin redox properties.
Preliminary insight into the general role of dipole orienta-
tion on flavin recognition was obtained through computa-
tional analysis of the fluoro-substituted compounds 2a-c.
Following initial PM3 optimization of the entire 1‚3ox
complex, the imide proton of 3 and the amide proton of the
receptor were constrained to the PM3-predicted distance and
the resulting fragment geometry optimized at the HF 3-21G*
level. The structures of the receptor-flavin complexes 2a-
c‚3ox were then optimized in a similar fashion, with multiple
starting points used to determine the minimum energy
binding modes. To investigate the trend in the reduced
species, the corresponding radical anions were likewise
optimized. After ab initio geometry optimization, DFT
B3LYP 6-31G* single-point calculations were used for each
of the systems to obtain energies of complexation and
electrostatic potential surfaces12 (Figure 2).
In the oxidized series we see a predicted trend of increasing
stability as the fluoro substituent moves from ortho to para.
The structure and calculated energy difference of the receptor
2a‚3ox fragment indicates that a dipole-dipole repulsion
between the fluoro substituent and the carbonyl oxygen
(O(4)) forces the aromatic moiety out of plane with the flavin
nucleus. The interaction of the meta-substituted receptor 2b
with flavin is energetically more favorable than that of
receptor 2a, due in part to the interaction of the fluorine atom
with the weakly positive N(10) position of the flavin ring
system.
The strongest favorable computationally predicted interac-
tion is seen for the para-substituted receptor 2c. In this
receptor, a strong donor atom-π interaction can be seen
between the fluoro substituent and the highly electropositive
portion of the flavin.13
A similar trend is seen in the reduced series of host-guest
species. A pronounced repulsion is seen in the ortho-
substituted dyad 2a‚3red as compared to the oxidized system,
evidenced by a “swinging out” of the phenyl ring from
underneath the flavin ring system. This loss of aromatic
stacking is also observed for the meta- and para-substituted
scaffolds, indicating similar repulsion of the phenyl ring from
the flavin radical anion. Some stability is gained in the para-
substituted dyad through two edge-to-face hydrogen bond
interactions between the phenyl hydrogens and flavin car-
bonyl O(2), thus generating the most stable complex.
Figure 1. (a) Tyrosine takes part in the aromatic-dipole interaction
in the FMN binding site of the D. Vulgaris flavodoxin.7 (b) The
analogous host-guest complexes of receptors 1 and 2 with flavin
3ox.
aromatic residues.8 The orientation and nature of these
intermolecular contacts determines their resultant effect on
the microenvironment of the active site, providing specific
recognition based on oxidation state and concomitant regula-
tion of the redox behavior of the cofactor.
The selectivity of the apoprotein for a specific cofactor
oxidation state provides the enzyme with the ability to control
cofactor redox potentials. This selectivity can be partially
deciphered via mutagenesis studies of intact biological
systems.9 These results, however, are difficult to attribute
to a single interaction due to multiple secondary effects that
arise when residues are altered.10 Model systems that are able
to focus on a single interaction thus serve as complements
to biological studies.11
To address the issue of the roles of specific electrostatic
factors on flavin-aromatic interactions, we have developed
a model system that directly controls the orientation of polar
(8) Chemistry and Biochemistry of FlaVoenzymes; Mu¨ller, F., Ed.;
CRC: Boca Raton, 1991; Vols. 1-3.
(12) Previous studies have shown that this methodology effectively
replicates key experimental attributes of host-guest complexes. See: Cuello,
A. O.; McIntosh, C. M.; Rotello, V. M. J. Am. Chem. Soc. 2000, 122, 3517-
3521.
(13) Breinlinger, E. C.; Keenan, C. J.; Rotello, V. M. J. Am. Chem. Soc.
1998, 120, 8606-8609.
(9) Stockman, B. J.; Richardson, T. E.; Swenson, R. P. Biochemistry
1994, 33, 15298-15308.
(10) Bradley, L. H.; Swenson, R. P. Biochemistry 1999, 38, 12377-
12386.
(11) Rotello, V. M. Curr. Opin. Chem. Biol. 1999, 3, 747-751.
1532
Org. Lett., Vol. 3, No. 10, 2001

Figure 2. The position of the fluoro substituent leads to differing atom-flavin interactions: (a) ortho, (b) meta, and (c) para. Structures
represent the lowest energy conformations. Reported energies are relative to the unsubstituted receptor 1.
Overall, the energies derived from the calculations suggest
a trend of increased recognition for ortho-, to meta-, to para-
substituted complexes that should experimentally be observed
as increased binding affinities.
To test the hypothesis of substituent position dependency
and to verify ourcalculations, the receptors 1 and 2a-c were
synthesized and binding affinities of the dyads 1‚3ox and 2a-
c‚3ox were quantified by 1H NMR titrations in CDCl3.
Electrochemical studies on the redox active flavin were then
performed to measure the change in reduction potential for
the flavin. These changes were used to calculate free energies
of binding as well as the association constants of the reduced
cofactors 1‚3red and 2a-c‚3red (Table 1).
(2a) to meta (2b) to para (2c) for both fluoro and related
moieties.15 In the oxidized family of receptors, we see an
increase in free energy of 0.61 kcal/mol from ortho to para.
This is of a lower magnitude than the predicted values (3.45
kcal/mol), presumably due to entropic effects arising from
the flexibility of the ethyl chain and solvation effects not
accounted for in the calculations. Upon reduction of the flavin
cofactor, the geometry dependence trend is maintained.
Similar differences of calculated vs observed energy changes
as compared to the oxidized series are apparent (9.97 to 0.73
kcal/mol, respectively). An increase in binding for each of
the reduced scaffolds, highlighted by the positive ∆E1/2
Table 1. Binding Constants for Complexes of Oxidized and
Reduced Flavin 3 with Receptors 1 and 2
ring
Ka(ox)
∆G(ox)
∆E1/2
∆G(red)
Ka(red)
receptor substit (M-1
)
(kcal/mol) (mV)b (kcal/mol) (M-1
)
a
1
H
920
-4.04
-3.87
-4.23
-4.48
21
12
23
18
-4.45
-4.07
-4.69
-4.80
1839
2a
2b
2c
o-F
m-F
p-F
690
1270
1920
966
2761
3330
a CDCl3, 23 °C, all titrations are (5%, based on standard error of the
curve fits. b Concentrations used: 1, 2 ) 0.0025 M, 3 ) 0.0005 M in 0.1
M TBAP solution in CH2Cl2 at 23 °C. ∆E1/2 is for the bound vs unbound
flavin 3 referenced to ferrocene (E1/2(unbound) ) -1293 mV).14
In agreement with the ab initio predictions, there is a
dependence on the geometry of the fluoro substituent on the
binding affinities of the complexes (Figure 3). An increase
in binding is observed as the substituent moves from ortho
Figure 3. Binding constants for the oxidized and reduced species.
Org. Lett., Vol. 3, No. 10, 2001
1533

values, reflects the additional strength of the hydrogen bonds
formed in the reduced complex as compared to the oxidized.
These bonds are strong enough to maintain the bound state
and compensate for the loss of aromatic stacking and π-donor
interactions with the reduced cofactor, in agreement with
previous model systems.16
(as compared to the 1‚3 dyad) indicates an increased
difficulty in reducing the flavin cofactor and also correlates
with the substantial decrease in stability of the reduced
complex in vacuo. The binding affinities of the meta- and
para-substituted systems (2b‚3red and 2c‚3red, respectively)
follow the general trend observed for the oxidized series, in
which the increased hydrogen bonding typical of this
triazine-flavin pair in the reduced system increases the
overall binding affinity. This again supports our hypothesis
that the binding strength is affected by the position of the
fluoro moiety. The ∆E1/2 values for both the meta and para
positions are similar to those of unsubstituted ring, suggesting
specifically that the dipole-quadrapole interaction of the
phenyl ring with the flavin in these systems is lost.
In summary, we have predicted and experimentally
determined the effects of dipole orientation on aromatic
stacking in both the oxidized and radical anion states of
flavin. Since the electrostatic topology of the flavin surface
is not uniform, not all areas are expected to interact equally
with dipoles. Computational and experimental results confirm
this assertation, showing (experimentally) a modulation of
recognition of up to 0.61 kcal/mol in the oxidized systems
and 0.73 kcal/mol in the reduced complexes, indicating that
proper placement of dipoles is critical to recognition
processes. By quantifying the individual interactions present
in enzyme active sites, the requirement for particular residues
in the local environment of the cofactor can be more readily
interpreted and targeted in future experiments.
The reduction in binding for the oxidized ortho-substituted
host system when compared with that of the unsubstituted
phenethylamine complex 1‚3ox clearly supports the theory
that the unfavorable interaction caused by the proximity of
the fluoro substituent of 2a to the carbonyl group of flavin
3ox is of a higher magnitude than any favorable aromatic
stacking (present in 1‚3ox). In the case of the meta-substituted
system (2b‚3ox), a modest increase in binding affinity is seen
for the electron-deficient fluoro receptor as compared to the
unsubstituted system 1‚3ox. This slight increase in binding,
although still within the general trend of the fluoro series, is
unexpected on the basis of the results of our calculations.
This can possibly be attributed to the entropic benefits of
having several low-energy conformational states in the
experimental system. The para-substituted system, 2c‚3ox,
exhibits the expected increase in binding when compared
with the unsubstituted phenethylamine dyad 1‚3ox. This
shows favorable responses due to both aromatic stacking and
donor atom-π interactions, concomitant with previous
studies.17
In the reduced ortho-substituted system (2a‚3red), a greater
difference in free energy from the unsubstituted dyad for
the reduced as compared to the oxidized system is observed
experimentally. This is in good agreement with the calcula-
tions, which show a large unfavorable interaction for this
complex. The change in reduction potential to the negative
Acknowledgment. This research was supported by the
National Institutes of Health (GM 59249-0). V.R. acknowl-
edges support from the Alfred P. Sloan Foundation, Research
Corporation, and the Camille and Henry Dreyfus Foundation.
C.M. was supported by a NIH Chemistry-Biology Interface
Training Grant (GM 5-84524).
(14) Binding constants for the reduced species were calculated by single
electron-transfer processes using the following formula: Ka(red)/Ka(ox) )
1/2(bound)-E1/2
e(nF/RT)(E
(unbound)), where Ka(red) and Ka(ox) are the binding
Supporting Information Available: Synthesis, 1H NMR,
and elemental analyses for precursors and target molecules
and NMR titration and electrochemical conditions. This
material is available free of charge via the Internet at
http://pubs.acs.org.
constants for the reduced and oxidized species, respectively.
(15) A parallel, but less dramatic, substituent geometry dependency was
observed for the receptor with a methoxy substituent.
(16) Breinlinger, E. C.; Rotello, V. M. J. Am. Chem. Soc. 1997, 119,
1165-1166. Deans, R.; Niemz, A.; Breinlinger, E. C.; Rotello, V. M. J.
Am. Chem. Soc. 1997, 119, 10863-10864.
(17) Baba, A. S.; Tamura, N. Immunology 1977, 32, 251-256. Naito,
Y.; Sugiura, M.; Yamaura, Y.; Fukaya, C.; Yokayama, K.; Nakagawa, Y.;
Ikeda, T.; Senda, M.; Fujita, T. Chem. Pharm. Bull. 1991, 39, 1736-1745.
OL015838L
1534
Org. Lett., Vol. 3, No. 10, 2001