Factor Analysis of Spectroelectrochemical Reduction of FAD Reveals The P K of The Reduced State and The Reduction Pathway
Factor Analysis of Spectroelectrochemical Reduction of FAD Reveals The P K of The Reduced State and The Reduction Pathway
Factor Analysis of Spectroelectrochemical Reduction of FAD Reveals The P K of The Reduced State and The Reduction Pathway
The free flavin adenine dinucleotide (FAD) cofactor is known to exhibit a pH-dependent midpoint
potential involving a simultaneous two-electron transfer step (n 2). Uv-vis spectroelectrochemical
reductions of FAD at constant pH, ranging from 5 to 9, were recorded and analyzed by factor analysis.
Principal factor analysis was used to determine the number of species present at each pH. The results
indicate that only two composite forms of FAD are present: the oxidized and the reduced forms.
Window factor analysis was used to extract the concentration profiles of the controlling species. The
oxidized form was found to be a single pH-independent species, whereas the reduced form consists
of two species. The pH-dependent spectroscopic changes of reduced FAD were best modeled by a
single proton transfer step involving two different ionization states with an apparent pKa 6.3. This
value compares favorably with those obtained from NMR and from midpoint potential measurements. At pH 6, the reduction of FAD was found to be first order, whereas at pH 9 the reduction is zero
order; these observations are explained in terms of the reaction pathway involving xanthine oxidase,
its substrate, and the pH. Copyright # 2007 John Wiley & Sons, Ltd.
KEYWORDS: factor analysis; window factor analysis; Uv-vis spectroscopy; flavin adenine dinucleotide; FAD reduction
1. INTRODUCTION
Flavin adenine dinucleotide (FAD) is a well-characterized
coenzyme that is derived from riboflavin [1]. Flavoproteins
containing FAD carry out a broad range of biological redox
reactions, some involving the transfer of two electrons in
individual successive steps (i.e. two n 1 steps) or in a single
simultaneous step (i.e. an n 2 process). The FAD cofactor in
flavoenzymes is known to exist in three oxidation states (the
quinone, semiquinone, and hydroquinone), and each of these
oxidation states can exist in different ionization states. Over
the past half century, there have been numerous spectroscopic and electrochemical studies to identify different
ionization and oxidation states of free FAD, as well as
attempts to establish the reduction potentials and acid
dissociation constants associated with these species [16].
Notably, several spectroscopic studies of FAD [1,6], from pH
5 to 9, revealed no ionizable oxidized species and only a
single ionizable reduced species.
Factor analysis of FAD reduction reveals pKa of reduced state and reduction pathway 521
3. EXPERIMENTAL
3.1. Sample preparation
2. THEORY
Factor analysis encompasses a number of mathematical and
statistical methods. It is used to analyze large data sets
without relying upon any preconceived chemical model. In
many applications, PFA is first used to determine the number
of controlling factors involved in a set of data. Details of the
chemical system must be known in order to exclude
experimental artifacts that may appear as unique factors.
There are a number of complicating issues such as detector
noise, baseline drift, signal nonlinearity, data pretreatment
effects, sampling error, etc. Such undesirable effects often
confuse the analysis because they produce ghost factors that
have no chemical meaning. Noise is a component of all
experimental data sets, and therefore care must be taken to
exclude this artifact as a chemical species. Also, it is
important to note that the number of chemical species
present is not necessarily the same as the number of factors.
For example, if the concentrations of two or more species
change but remain in the same ratio due to chemical
equilibrium only a single factor will be observed for these
multiple components.
A variety of factor analytical techniques have been
developed to determine the number of significant factors
responsible for a data matrix, methods such as the residual
standard deviation, the factor indicator function, F-tests for
establishing significance levels, distribution of misfit,
chi-squared tests, cross validation, etc. [13] The residual
standard deviation (RSD), also known as the real error (RE) is
calculated from the eigenvalues extracted from the data
matrix. The primary factors are associated with the largest
eigenvalues, accounting for the maximum variation in the
data. The secondary factors are associated with the smallest
set of eigenvalues, accounting for the noise as well as other
undesirable instrumental and experimental artifacts. If a
reasonable estimate of the noise in the data (i.e. the standard
deviation) can be determined prior to the factor analysis the
number of primary factors can be determined. The number of
factors corresponds to the point where the standard
Copyright # 2007 John Wiley & Sons, Ltd.
Table I. Residual standard deviations, RSD(n), obtained from PFA (showing only the first three residuals) compared to the
standard deviations, STD, estimated from the isosbestics point at 340 nm for the buffered solutions FADpHx, and from the baseline
at 550 nm for FADox and FADred matrices. The number of chemical factors, n, corresponds to the RSD level that is closest to the
estimated STD
Data Matrix
STD (103)
RSD(1) (103)
RSD(2) (103)
RSD(3) (103)
FADpH 5.0
FADpH 5.5
FADpH 6.0
FADpH 6.2
FADpH 6.5
FADpH 6.7
FADpH 7.0
FADpH 7.5
FADpH 8.0
FADpH 9.0
FADox
FADred
8.56
4.22
4.35
4.75
2.65
1.60
3.47
3.00
1.94
2.22
18.79
15.73
93.41
113.48
93.68
116.45
128.50
125.29
135.97
140.88
116.71
108.35
15.47
31.12
3.06
3.54
2.63
2.56
2.12
2.62
2.58
2.48
2.15
2.34
10.12
12.71
1.97
0.69
0.85
1.15
0.88
0.92
0.74
0.75
0.88
0.62
8.05
5.55
2
2
2
2
2
2
2
2
2
2
1
2
absorbances were not corrected for baseline nor for variations in the initial concentrations of FAD.
4. RESULTS
4.1. Principal factor analysisnumber
of factors
The reduction of FAD at pH 7 as a function of time is
illustrated in the spectra of Figure 1. Similar data were
obtained at other pHs. The noise in the data was quantified
by determining the standard deviation (STD) about the
isosbestic point located at 340 nm. Values obtained as a
function of pH are provided in Table I. The absorption data
were then analyzed by PFA yielding RSD(n), at each pH as a
function of factor level, n. Table I displays only the first three
residuals, all other residuals are superfluous because they
systematically decrease as the factor level increases. At each
pH, the standard deviation estimated from the isosbestic
point is closest to RSD(2) corresponding to the two-factor
level, giving evidence that two factors account for each
FADpH data matrix. It is interesting to note that all other
common criterion [13] for determining the number of
relevant factors, such as F-tests, cross validation, the factor
indicator function, etc., yielded an excessive number of
Copyright # 2007 John Wiley & Sons, Ltd.
Figure 1. Reduction of 122 mM free FAD in 200 mM phosphate at pH 7. Individual spectra were collected every 5 min
intervals but only selected spectra are shown. The isosbestics
point at 340 nm was used to determine the standard deviation
in the absorbance measurements.
J. Chemometrics 2007; 21: 520528
DOI: 10.1002/cem
Factor analysis of FAD reduction reveals pKa of reduced state and reduction pathway 523
(1)
(2)
5. DISCUSSION
5.1. Factor analysis
The FADpHx data matrices, described in the Experimental
Procedures Section, conducted at 10 different pH, were
subjected to principal factor analysis (PFA). PFA revealed
Copyright # 2007 John Wiley & Sons, Ltd.
Factor analysis of FAD reduction reveals pKa of reduced state and reduction pathway 525
2
(3)
pH
FADred mj
FADred bj
FADox mj
FADox bj
5.0
5.5
6.0
6.2
6.5
6.7
7.0
7.5
8.0
9.0
0.968
1.041
1.037
1.001
0.991
0.928
0.949
0.929
1.038
1.055
0.0208
0.0327
0.0100
0.0142
0.0060
0.0138
0.0135
0.0101
0.0229
0.0181
1.016
1.035
1.036
1.008
0.981
0.909
0.990
0.921
1.039
1.089
0.0347
0.0226
0.0066
0.0152
0.0214
0.0023
0.0189
0.0203
0.0164
0.0339
(5a)
FAD2 H FADH
(5b)
(6a)
Path 2:
FADH H FADH2
(4)
(6b)
The first steps in Paths 1 and 2, (5a) and (6a), represent the
ionization which involves reduction of FAD by xanthine
(Xan) catalyzed by xanthine oxidase (XanOx(cat)). The
second steps, (5b) and (6b), represent single proton transfers.
Overall, Path 1 involves a single proton transfer, whereas
Path 2 involves two proton transfers, the reduction step
accounting for one of the transfers. Both paths are
hypothetically possible. Both mechanisms assign the dissociation constant, 106.3, to the second step, which differs
stoichiometrically for each model. We will, however, present
evidence to show that Path 2 represents the correct
mechanism.
These models yield two different expressions for the
half-cell reactions as shown in Equations (7) and (8).
According to Nernst theory the midpoint reduction potential, Emid, is a function of the standard reduction potential, Eo,
the dissociation constant, Ka, and the hydrogen ion
concentration as shown in Equations (7) and (8).
Path 1:
Eo 59
Ka
log
2
2
Ka H
(7)
Eo 59
Ka
log
2
2
Ka H H 2
(8)
Emid
Path 2:
Emid
Factor analysis of FAD reduction reveals pKa of reduced state and reduction pathway 527
(12a)
a kKOH cato
(12b)
b KOH
(12c)
where
and
(9)
The hydrogen ion and the two electrons released from this
half-cell are used to reduce FAD (see Equation (6a)). The
overall reaction is initiated by xanthine oxidase catalyst,
XanOx(cat), which produces a steady-state intermediate
leading to the final products. The oxidation-reduction
process can be depicted as
FAD Xanthine OH XanOxcat
Intermediate ! FADH UricAcid
(10)
(11)
6. CONCLUSION
This investigation shows that factor analysis can yield
valuable information regarding the reduction of FAD. The
advantage of this method is that it can resolve a complex
system without imposing any prior assumptions about the
system. As clearly illustrated here, the method can be used to
extract information from experiments that are difficult to
control all of the variables, especially during sample
preparation.
The results obtained are consistent with those obtained by
investigators using quite different techniques. For example,
from 15-N chemical shifts Franken and coworkers [18] found
that the pKa of the reduced form of pure flavin mononucleotide (FMN) has a value of 6.8. This is in good agreement with
the pKa value of the reduced form of pure FAD determined to
be 6.3 in the present investigation. The difference in pKa
values may be due to the chemical difference between FMN
and FAD, or due to some artifact of the different techniques
used to obtain these values. Exact agreement is not to be
expected between the pKa determined in this study for FAD
and the pKa previously determined for FMN. At the present
time there is no known way to account for the effects of
different kinds of buffers on the spectra of the FAD species.
Their effects appear to be very small and have been ignored
in the present investigation as well those of Lowe and Clark
[2], Mayhew [5], and many others.
The proposed kinetic mechanism is also consistent with
the spectroscopic observations. It explains the switching
from first-order to zero-order reaction rate as a function of
pH. The nature of the intermediate cannot be determined
from these measurements because its signal is below the
detection limits of the instrument. It is tempting, however, to
speculate that the elusive semiquinone plays an important
role in the process.
J. Chemometrics 2007; 21: 520528
DOI: 10.1002/cem
REFERENCES
1. Muller F. Chemistry and Biochemistry Flavoenzymes, 1,
Chapter 1 CRC Press: Boca Raton, FL, 1991.
2. Lowe HJ, Clark WM. Studies on oxidation-reduction.
XXIV. Oxidationreduction of flavin adenine dinucleotide. J. Biol. Chem. 1956; 221: 983992.
3. Heelis PF. The photophysical and photochemical properties of flavins (Isoalloxazines). Chem. Soc. Rev. 1982; 11:
1538.
4. Draper RD, Ingraham LJ. A potentiometric study of the
flavin semiquinone equilibrium. Arch. Biochem. Biophys.
1968; 125: 802808.
5. Mayhew SG. The effects of pH and semiquinone formation on the oxidationreduction potentials of flavin
mononucleotide. Eur. J. Biochem. 1999; 265: 698702.
6. Dudley KH, Ehrenberg A, Hemmerich P, Muller F.
Spektren und strukturen der am flavin redox system
beteiligeten partikeln. Helv. Chim. Acta 1964; 47: 1354.
7. Kano K, Mori K, Uno B, Kubota T. Voltmmetric and
spectroscopic studies of pyrroloquinoline quinone under
neutral and basic conditions. Bioelect. Bioener. 1990; 23:
2272238.
8. Clark WM. Oxidation-Reduction Potentials of Organic Systems. RE Krieger Publishing Co.: Huntington, NY, 1972.
9. Smith ET, Davis CA, Barber MJ. Voltammetric simulations of multiple electron transfer/proton coupled
reactions: FAD as a model system. Anal. Biochem. 2003;
323: 114121.