The epimerase activity of anthocyanidin reductase from Vitis vinifera and its regiospecific hydride transfers

Article abstract of DOI:10.1515/BC.2010.015

Anthocyanidin reductase (ANR) from Vitis vinifera catalyzes an NADPH-dependent double reduction of anthocyanidins producing a mixture of (2S,3R)- and (2S,3S)-flavan-3-ols. At pH 7.5 and 30°C, the first hydride transfer to anthocyanidin is irreversible, and no intermediate is released during catalysis. ANR reverse activity was assessed in the presence of excess NADPq. Analysis of products by reverse phase and chiral phase HPLC demonstrates that ANR acts as a flavan-3-ol C₃-epimerase under such conditions, but this is only observed with 2R-flavan-3-ols, not with 2S-flavan-3-ols produced by the enzyme in the forward reaction. In the presence of deuterated coenzyme 4S-NADPD, ANR transforms anthocyanidins into dideuterated flavan-3-ols. The regiospecificity of deuterium incorporation into catechin and afzelechin - derived from cyanidin and pelargonidin, respectively - was analyzed by liquid chromatography coupled with electrospray ionization-tandem mass spectrometry (LC/ESI-MS/MS), and it was found that deuterium was always incorporated at C₂ and C₄. We conclude that 3-epimerization should be achieved by tautomerization between the two hydride transfers and that this produces a quinone methide intermediate which serves as C₄ target of the second hydride transfer, thereby avoiding any stereospecific modification of carbon 3. The inversion of C₂ stereochemistry required for reverse epimerization suggests that the 2S configuration induces an irreversible product dissociation.

Full text of DOI:10.1515/BC.2010.015

Article in press - uncorrected proof
Biol. Chem., Vol. 391, pp. 219–227, February/March 2010 • Copyright ꢀ by Walter de Gruyter • Berlin • New York. DOI 10.1515/BC.2010.015
The epimerase activity of anthocyanidin reductase from
Vitis vinifera and its regiospecific hydride transfers
Mahmoud Gargouri1,2, Jean Chaudie`re1,*,
Claude Manigand1, Chloe´ Mauge´1, Katell Bathany1,
Jean-Marie Schmitter1 and Bernard Gallois1
Introduction
For the past decade, the expression and partial characteri-
zation of several enzymes (see Scheme 1) have greatly
improved our understanding of the biosynthetic pathway(s)
of flavan-3-ols of distinct stereochemistries from leucoantho-
cyanidins produced by dihydroflavonol reductase (DFR). In
the first pathway, a reductive dehydration step performed by
leucoanthocyanidin reductase (LAR) converts leucoantho-
cyanidins to (2R,3S)-trans-flavan-3-ols such as (q)-catechin
(Tanner et al., 2003; Pfeiffer et al., 2006). In parallel, antho-
cyanidin synthase (a dioxygenase) transforms leucoantho-
cyanidins into chromophoric anthocyanidins (Saito et al.,
1999; Turnbull et al., 2001; Wilmouth et al., 2002; Abrahams
et al., 2003). The latter are unstable compounds which could
be either glycosylated by UDP-glucose:flavonoid 3-O-glu-
cosyltransferase into anthocyanins (Springob et al., 2003;
Veitch and Grayer, 2008) or reduced by anthocyanidin reduc-
tase (ANR, EC 1.3.1.77) into flavan-3-ols at the expense of
two NADPH equivalents (Xie et al., 2003).
There are very few enzymes that use two equivalents of
a nicotinamide coenzyme to reduce a single cosubstrate. In
addition to ANR, the best documented examples are proba-
bly the NADPH-dependent pteridine reductase PTR1 from
Leishmania tarentolae and Leishmania major (Wang et al.,
1997; Gourley et al., 2001), and the recently described
NAD(P)H-dependent diketoreductase from Acinetobacter
baylyi (Wu et al., 2009). A mechanistic study of the regio-
and stereospecificity of ANRs would therefore help to under-
stand the unusual potentialities of this enzyme family. In a
comparative study of ANR from Arabidopsis thaliana and
Medicago truncatula (At- and Mt-ANR in Scheme 1), Xie et
al. (2004) reported that (-)-epicatechin w(2R,3R) cis structurex
was the major reduction product of cyanidin in either case,
and they assumed that the minor amount of (-)-catechin
w(2S,3R) trans structurex which was additionally produced
might be as a result of non-enzymatic epimerization. They
proposed a speculative but plausible reductive mechanism in
which two hydride transfers from NADPH to carbons 2 and
3 of cyanidin would lead to a 2,3-cis product of unique ster-
eochemistry, i.e., (2R,3R). However, in a recent report (Gar-
gouri et al., 2009a), we showed that ANR from Vitis vinifera
(Vv-ANR) exclusively used pro(S) hydrogen from NADPH
and we provided non-equivocal evidence for the enzyme-
mediated production of a mixture of C3-epimers of 2S con-
figuration, such as (-)-catechin and (q)-epicatechin derived
from cyanidin. Thus, Vv-ANR does not only behave as a
reductase but also as a C3-epimerase, which suggests that its
1 Chimie et Biologie des Membranes et des Nanoobjets
UMR CNRS 5248, Baˆtiment B8, Avenue des Faculte´s,
Universite´ Bordeaux 1, F-33405 Talence Cedex, France
2 Laboratoire de Physiologie Mole´culaire des Plantes,
Centre de Biotechnologie de Borj-Cedria, B.P. 901, 2050
Hamman-Lif, Tunisia
*Corresponding author
e-mail: j.chaudiere@cbmn.u-bordeaux.fr
Abstract
Anthocyanidin reductase (ANR) from Vitis vinifera catalyzes
an NADPH-dependent double reduction of anthocyanidins
producing a mixture of (2S,3R)- and (2S,3S)-flavan-3-ols. At
pH 7.5 and 308C, the first hydride transfer to anthocyanidin
is irreversible, and no intermediate is released during catal-
ysis. ANR reverse activity was assessed in the presence of
excess NADPq. Analysis of products by reverse phase and
chiral phase HPLC demonstrates that ANR acts as a flavan-
3-ol C3-epimerase under such conditions, but this is only
observed with 2R-flavan-3-ols, not with 2S-flavan-3-ols pro-
duced by the enzyme in the forward reaction. In the presence
of deuterated coenzyme 4S-NADPD, ANR transforms antho-
cyanidins into dideuterated flavan-3-ols. The regiospecificity
of deuterium incorporation into catechin and afzelechin –
derived from cyanidin and pelargonidin, respectively – was
analyzed by liquid chromatography coupled with electro-
spray ionization-tandem mass spectrometry (LC/ESI-MS/MS),
and it was found that deuterium was always incorporated
at C2 and C4. We conclude that C3-epimerization should be
achieved by tautomerization between the two hydride trans-
fers and that this produces a quinone methide intermediate
which serves as C4 target of the second hydride transfer,
thereby avoiding any stereospecific modification of carbon
3. The inversion of C2 stereochemistry required for ‘reverse
epimerization’ suggests that the 2S configuration induces an
irreversible product dissociation.
Keywords: anthocyanidin reductase; deuterium labeling;
double reduction; epimerase; hydride transfer; NADPH;
quinone methide; regiospecificity.
2010/284
Brought to you by | Purdue University Libraries
Authenticated
Download Date | 5/22/15 5:02 PM

Article in press - uncorrected proof
220 M. Gargouri et al.
Scheme 1 Biosynthetic pathways of flavan-3-ols.
catalytic mechanism could substantially differ from that of
previously described ANRs. Our subsequent studies of
steady-state kinetics and binding at equilibrium (Gargouri et
al., 2009b) led to the conclusion that the kinetic mechanism
of Vv-ANR was rapid-equilibrium ordered (Bi Uni Uni Bi),
with NADPH binding first and NADPq released last. In the
present study, we investigated the regiospecificity of hydride
transfers to anthocyanidins by means of deuterium-labeling
and LC/ESI-MS/MS experiments, and we demonstrated that
the enzyme performs as a pure C3-epimerase of 2R-flavan-
3-ols in the presence of excess NADPq. Our results provide
mechanistic clues to the epimerase activity which is associ-
ated with anthocyanidin reduction or flavan-3-ol oxidation.
2009a) that these two flavan-3-ols are (-)-catechin and (q)-
epicatechin.
To assess the possibility that ANR exhibits significant cat-
alytic activity in the backward direction of the reaction, the
enzyme was then incubated at pH 7.5 and 308C with each
of the four (epi)catechin stereoisomers in the presence of a
large concentration of NADPq. No trace of 578-nm chro-
mophore could be detected in any of these four experiments.
Moreover, (-)-catechin and (H)-epicatechin were both recov-
ered unchanged from the reaction medium, even upon 16 h
of incubation in the presence of 50 m
M ANR.
However, as shown in Figure 2, we found that (H)-cate-
chin and (-)-epicatechin were independently transformed into
their C3-epimer. In either case, the yield of epimerization
exceeded 30% upon a 30-min incubation of 50 m
with 50 m ANR and 500 m
NADPq. There was no spon-
taneous isomerization in the absence of ANR, and with only
5 m ANR the yield of epimerization was approximately
M flavanol
M
M
Results
Cyanidin strongly absorbs visible light with a maximal
absorbance wavelength of 578 nm (´;16 533
M
-1
M
cm-1) at
twice smaller by 30 min.
pH 7.5, and this wavelength was used to monitor the time
course of cyanidin disappearance with or without ANR.
Methanolic samples of products derived from either cyanidin
or pelargonidin upon completion of the ANR-mediated reac-
tion were resolved by reverse phase and chiral phase HPLC.
As shown in the reverse phase chromatogram of Figure 1A,
cyanidin is transformed into equivalent amounts of flavan-
3-ols having the retention times of catechin (peak 1) and
epicatechin (peak 2). Based on the chiral phase chromato-
gram shown in Figure 1B, we confirm (Gargouri et al.,
It was also approximately twice smaller by 16 h when the
reaction medium contained 100 m
M NADPH in addition to
500 mM
NADPq. Thus, ANR has the unexpected property
to slowly but selectively epimerize 2R epimers of its own
reduction products when the ratio wNADPqx/wNADPHx is
high. This ‘reverse’ C3-epimerase activity is not associated
with irreversible consumption of NADPq, for there is no
measurable increase in 340-nm absorbance.
The regiospecificity of hydride transfers from NADPH to
either cyanidin (phenolic ring B) or pelargonidin (catecholic
Brought to you by | Purdue University Libraries
Authenticated
Download Date | 5/22/15 5:02 PM

Article in press - uncorrected proof
Epimerase activity of anthocyanidin reductase and hydride transfers 221
Figure 1 Chromatographic analysis of products derived from cyanidin in the presence of ANR and excess NADPH.
Initial concentrations are 200 m NADPH and 50 m cyanidin. (A) Reverse-phase HPLC; peak 1 is eluted at the retention time of the
M
M
two catechin standards, whereas peak 2 is eluted at the retention time of the single epicatechin standard; peak 3 contains a dimer of cyanidin.
(B) Chiral phase HPLC; external standards indicated by the dotted line: (-)-catechin wIx, (-)-epicatechin wIIx, and (H)-catechin wIIIx.
Figure 2 Chromatographic analysis of products derived from 2R-flavan-3-ols in the presence of ANR and excess NADPq.
(A, A9) Reverse phase HPLC; peaks 1 and 2 contain catechins and epicatechins, respectively, as in Figure 1. (B, B9) Chiral phase HPLC.
Samples are methanolic solutions of polyphenols obtained upon 30-min incubation of 50 mM ANR with 500 mM M flavan-
NADPq and 50 m
3-ol at pH 7.5 and 308C; flavan-3-ols initially introduced are (H)-catechin in (A) and (B), and (-)-epicatechin in (A9) and (B9); external
standards shown by the dotted line: (-)-catechin wIx, (-)-epicatechin wIIx, and (H)-catechin wIIIx.
ring B) was then investigated by means of deuterium label-
ing. For this purpose, 4S-NADPD was substituted for
NADPH, and the deuterated products were analyzed by LC-
MS and LC-MS/MS. Positive ion mass spectra of enzymat-
ically deuterated catechin and afzelechin are reproduced in
Figures 3 and 4, respectively. In Figure 3A, the major ion at
m/z 293 can be assigned to an (MqHq) ion derived from
dideuterated catechin, in agreement with the expected double
reduction of cyanidin. The relative abundance of the peak at
m/z 294 is in agreement with the expected value for catechin
(isotopic contribution of 13C). Similar conclusions can be
drawn from Figure 4A: the spectrum attributed to dideute-
rated afzelechin is virtually the same if one takes into
account the expected 16-Da shift in m/z values.
Brought to you by | Purdue University Libraries
Authenticated
Download Date | 5/22/15 5:02 PM

Article in press - uncorrected proof
222 M. Gargouri et al.
MS/MS spectra of precursor ions at m/z 293 and 277 are
shown in Figures 3B and 4B, respectively. For dideuterated
catechin (Figure 3B), major fragments have m/z values of
124, 140, and 167. The corresponding fragments occur at
m/z 108, 140, and 151 in the MS/MS spectrum of dideute-
rated afzelechin (Figure 4B). Our interpretation of the MS/
MS spectra of dideuterated catechin and afzelechin is
summarized in Tables 1 and 2 (peaks assigned to 13C back-
ground result from the isolation width of the precursor ion
that was isolated with its isotopic cluster). This interpretation
is based on the m/z values of MS/MS fragments which we
observed with non-deuterated catechin and afzelechin. All
such values are reported in Scheme 2 which confirms most
of the fragmentation steps reported by Li and Deinzer (2007)
for catechin (m/z values outside parentheses) and uses our
own data to extend them to afzelechin (m/z values inside
parentheses). Clearly, the conclusion which can be drawn
from Tables 1 and 2 is that the two deuterium transfers have
been targeted to C2 and C4, respectively. The lack of m/z 170
monodeuterated fragment in the MS/MS spectrum of cate-
chin is not surprising, because the abundance of the corre-
sponding (m/z 169) fragment reported by Li and Deinzer
(2007) for non-deuterated catechin was extremely low and
actually undetectable in our own spectra (see Scheme 2). The
abundance of the m/z 149 fragment is low but significant in
the MS/MS spectrum of deuterated catechin, and the absence
of the corresponding fragment (m/z 133) in that of deuterated
afzelechin is likely to reflect the influence of changing phe-
nol for catechol in ring B. Finally, the m/z 251 fragment ion
of deuterated catechin does not have a corresponding m/z
249 or 250 fragment in the MS/MS spectrum of non-deuter-
ated catechin reported by Li and Deinzer (2007), as well as
in our own data (see Scheme 2). A similar fragmentation is
observed for deuterated afzelechin with a 16-Da mass shift,
leading to the m/z 235 ion. Although it is tempting to think
that it results from a loss of ethynol (loss of 42 Da), we
cannot use it in an analysis of deuteration regiospecificity
because we do not have a realistic fragmentation pathway to
suggest.
Discussion
The two hydrides which are sequentially transferred to
anthocyanidin come from pro(S) hydrogens of NADPH, and
the three-dimensional structure of the apoenzyme exhibits a
single Rossman fold. Moreover, binding equilibrium data
support a single NADPH-binding site per enzyme molecule
(Gargouri et al., 2009a,b). This means that the orientation of
these two transfers should be the same, either from above or
from below the average plane of the benzopyran ring. Our
data suggest that the first reduction step is virtually irrevers-
ible because no trace of anthocyanidin could be observed at
the 578 nm wavelength in the backward reaction, even with
2R-flavanols for which the last step is apparently reversible.
Because no significant amount of polyphenolic structures
other than anthocyanidin and two flavan-3-ol epimers could
be visualized in any of our chromatograms, it is likely that
the expected dihydrocyanidin intermediate is tightly bound
to the enzyme and never released in the aqueous solvent, and
this is in agreement with the rapid-equilibrium ordered mech-
anism deduced from steady-state kinetics and equilibrium-
binding studies (Gargouri et al., 2009b). The second hydride
transfer is irreversible with natural products of 2S configur-
ation. However, we did observe C3-epimerization of 2R-cat-
echin or -epicatechin in the presence of a large excess of
NADPq, which means that hydride transfer to 2R-dihydro-
cyanidin(s) is reversible, but markedly displaced in the direc-
tion of NADPq production. Altogether, these observations
suggest that an ANR-dihydrocyanidin transient complex
undergoes reversible epimerization between the two steps of
Figure 3 Positive ion mass spectra of enzymatically deuterated
catechin.
(A) ESI-MS spectrum; (B) MS/MS spectrum of the major ESI-MS
ion (m/z 293).
Figure 4 Positive ion mass spectra of enzymatically deuterated
afzelechin.
(A) ESI-MS spectrum; (B) MS/MS spectrum of the major ESI-MS
ion (m/z 277).
Brought to you by | Purdue University Libraries
Authenticated
Download Date | 5/22/15 5:02 PM

Article in press - uncorrected proof
Epimerase activity of anthocyanidin reductase and hydride transfers 223
Table 1 Carbon targets of deuterium labeling in the major MS/MS fragments
of dideuterated catechin.
m/za
Equivalenceb
Deuteration
Deuterium position(s)
123
124
125
139
140
141
148
149
151
152
166
167
251
274
275
123
Non-deuterated
Monodeuterated
Dideuterated
Non-deuterated
Monodeuterated
Dideuterated
Monodeuterated
Dideuterated
Non-deuterated
Monodeuterated
Monodeuterated
Dideuterated
?
–
C2
123q1
123q2
139
139q1
139q2
147q1
147q2
151
151q1
165q1
165q2
?
273q1
273q2
C13 background
–
C4
C13 background
C2 or C3 or C4
(C2, C4) or (C2, C3) or (C3, C4)
–
C13 background
C2 or C3 or C4
(C2, C4) or (C2, C3) or (C3, C4)
?
C2 or C3 or C4
(C2, C4) or (C2, C3) or (C3, C4)
Monodeuterated
Dideuterated
am/z values of the most abundant fragments are shown in bold font.
bm/z reference values are those of MS/MS fragments of non-deuterated catechin.
Table 2 Carbon targets of deuterium labeling in the major MS/MS fragments
of dideuterated afzelechin.
m/za
Equivalenceb
Deuteration
Deuterium position(s)
108
109
139
140
141
150
151
169
208
235
258
259
107q1
107q2
139
139q1
139q2
149q1
149q2
169
Monodeuterated
Dideuterated
Non-deuterated
Monodeuterated
Dideuterated
Monodeuterated
Dideuteratedc
Non-deuterated
?
C2
C13 background
–
C4
C13 background
C2 or C3 or C4
(C2, C4) or (C2, C3) or (C3, C4)
–
?
?
?
?
?
257q1
257q2
Monodeuterated
Dideuterated
C2 or C3 or C4
(C2, C4) or (C2, C3) or (C3, C4)
am/z values of the most abundant fragments are shown in bold font.
bm/z reference values are those of MS/MS fragments of non-deuterated
afzelechin.
cPlus non-deuterated m/z 151.
hydride transfer. Although we do not know which of the
for a mechanism in which the observed epimerization at C3
is actually as a result of non-stereospecific proton addition.
Scheme 3 is compatible with such constraints, as well as with
the rapid-equilibrium ordered Bi Uni Uni Bi mechanism. In
the first step of Scheme 3, a stereospecific reduction at C2
leads to the enolic structure II which instead of being trans-
formed into its 3-keto equivalent by spontaneous tautome-
rization undergoes an enzyme-catalyzed tautomerization
which leads to quinone methide epimers (structures III and
IV). The 2,3-stereochemistry of the final products is already
in place at this stage, and the quinone methide structure then
provides a good target for a second hydride transfer, this time
at C4. The formation of the two quinone methide epimers
requires at least a basic catalytic side chain on the enzyme
to facilitate deprotonation of one of the phenolic hydroxyl
groups at C5 or C7. In principle, the overall proton exchange
anthocyanidin flavylium cation or quinoidal conjugate bases
preferentially binds to ANR as a true substrate, in all cases,
the electrophilic target of the first hydride transfer should be
either carbon 2 or carbon 4, i.e., those which are in ortho or
para position with regard to oxygen 1.
In the most plausible double-reduction mechanisms that
could be envisaged, the second hydride transfer would be
targeted to carbon 3, i.e., to a ketone group resulting from
tautomerization of either a 4,3- or a 2,3-enol intermediate.
These two mechanisms are formally excluded by the deute-
rium incorporation experiments, because they cannot lead to
the observed (C2, C4) dideuteration. Moreover, they are
incompatible with the observed C3-epimerization as well as
with the production of a mixture of cis- and trans-products
of strict 2S configuration. This means that we have to look
Brought to you by | Purdue University Libraries
Authenticated
Download Date | 5/22/15 5:02 PM

Article in press - uncorrected proof
224 M. Gargouri et al.
Scheme 3 (C2, C4) sequence of hydride transfers in the forward
reaction.
is only observed with 2R-flavan-3-ols which are C2-epimers
of its ‘normal’ products in the forward reaction. This activity
requires NADPq, which suggests that hydride transfer from
flavan-3-ol to NADPq is the first step (Scheme 4). For such
a dehydrogenation step, proton extrusion is an absolute pre-
requisite of hydride delivery, and this leaves CH in positions
2, 3, and 4 as possible sources of hydride. In the forward
pathway, hydride delivery from NADPH to carbon 2 pro-
duces a 2S structure, which means that CH in position 2 of
Scheme 2 Fragmentation pathways of catechina and afzelechinb.
a(RsOH, m/z outside parentheses); b(RsH, m/z inside parentheses);
Nsnot observed.
RDA, Retro Diels-Alder; BFF, benzofuran-forming fission.
would be facilitated by simultaneous acidic catalysis in the
vicinity of carbon 3, but a single acidic side chain would
favor stereospecific protonation at C3 instead of the observed
epimerization. Therefore, one has to assume that the proton
captured by carbon 3 comes from the aqueous solvent.
The involvement of a quinone methide intermediate as a
target of the second hydride transfer is the most plausible
explanation of deuteration at C4. Moreover, this structure
bears striking resemblance with the quinone methide inter-
mediate which is the target of hydride transfer in eugenol
synthase, an NADPH-dependent reductase (Louie et al.,
2007). Interestingly, recent structural data obtained in our
laboratory strongly support the involvement of a similar qui-
none methide intermediate in the catalytic mechanism of
LAR (Mauge´ et al., in preparation). In LAR, the production
of a quinone methide should result from a dehydration step.
With ANR, none of the structures that could be envisaged
as a partial reduction intermediate of anthocyanidin would
enable such a dehydration mechanism, and this is why
Scheme 3 assumes that it results from proton exchange
between the phenolic hydroxyl at C5 or C7 and the enolic
carbon 3 of structure II.
If Scheme 3 is essentially correct, it should be compatible
with the reverse epimerase activity of ANR, even if the latter
Scheme 4 Reverse epimerization.
Brought to you by | Purdue University Libraries
Authenticated
Download Date | 5/22/15 5:02 PM

Article in press - uncorrected proof
Epimerase activity of anthocyanidin reductase and hydride transfers 225
2R-flavanols could not deliver a hydride ion to NADPq in
the reverse pathway, following deprotonation of a phenolic
or aliphatic hydroxyl. Moreover, the epimerization that is
observed at C3 is just as efficient with flavan-3-ols of
(2R,3R) configuration as with flavan-3-ols of (2R,3S) con-
figuration. This means that we can rule out hydrogenated
carbon 3 as a source of hydride, because the oxidized nico-
tinamide ring of NADPq must stack on its polyphenolic part-
ner in the same configuration as the reduced nicotinamide
ring of NADPH in the forward pathway, hydride transfer
from carbon 3 of (H)-catechin to NADPq can be ruled out,
and we conclude that the hydride ion transferred to NADPq
should come from the methylene group in position 4, which
strongly supports a (C2, C4) sequence of hydride transfers in
the forward/reductive pathway. In other words, our ‘reverse
epimerization’ data support the reductive (C2, C4) sequence
of Scheme 3, but not a (C4, C2) sequence.
The reverse epimerization pathway would then be
described by Scheme 4, which might not require any other
catalytic assistance than that which was assumed for Scheme
3. In the presence of a large excess of NADPq, the ANR/
NADPq/flavan-3-ol complex would be in equilibrium with
an ANRØNADPHØquinone methide complex, and the latter
would slowly epimerize through interconversion with its
enolic tautomer. The fact that this reverse epimerization only
works with 2R-flavan-3-ols is more puzzling. This could be
as a result of a much higher dissociation constant of 2S-
flavan-3-ols from the last ternary complex. A 2S-induced
geometrical constraint would indeed solve the problem of
product release, which is frequently the limiting step in
NAD(P)(H)-dependent enzymes.
crystallize a binary or ternary complex of ANR have been
unsuccessful. In the three-dimensional structure of the crys-
tallized apoenzyme (Gargouri et al., 2009a), the active site
is virtually locked in a configuration that would hardly leave
any space for binding of either NADPH or cyanidin, because
it is obstructed by a large peptidic loop.
In conclusion, we have shown that Vv-ANR performs as
a pro(S) reductase/C3-epimerase in the forward direction of
the reaction with hydride transfers at C2 and C4, whereas it
performs as a pure C3-epimerase in the backward direction.
The ‘reverse’ epimerase activity requires the presence of
ANR and NADPq, and instead of being observed with the
products of the forward reaction it is exclusively observed
with the corresponding C2-epimers. The most plausible
mechanism that could be envisaged from our data relies on
the transient production of an enzyme-bound quinone
methide intermediate following a first hydride transfer at C2,
and it implies a strict C4-regiospecificity of the second
hydride transfer. The (C2, C4) regiospecificity of the two
hydride transfers would therefore provide a mechanistic
rationale for C3-epimerization in the forward and backward
directions of the reaction.
Materials and methods
Chemicals, enzymes and reagents
Cyanidin chloride, pelargonidin chloride, (q)-catechin, and (-)-epi-
catechin were purchased from Extrasynthe`se (Genay, France). (-)-
Catechin G98% pure, glucose-6-phosphate dehydrogenase from
Baker’s yeast type XV, dithiothreitol, NADPH tetrasodium salt
G97% pure, NADPq sodium salt,
D-Glucose-1-d, antibiotics, and
A critical feature of Scheme 3 is the C4-regiospecificity
of the second hydride transfer, i.e., that which is also
observed with the closest structural homologs of ANR. A
search for potential structural homologs led to the conclusion
that the enzyme that has the greatest three-dimensional
homology with ANR is DFR from Vitis vinifera, another
enzyme of the short-chain dehydrogenase/reductase super-
family which is known to transfer a hydride equivalent to
carbon 4 of dihydroflavonols (Petit et al., 2007). The root
mean square distance obtained by alignment of the two
all buffer components were purchased from Sigma-Aldrich Chem-
ical Co. (S)-NADPD, i.e., 4S-NADP2H, was synthesized enzymat-
ically, as previously described (Gargouri et al., 2009a).
Expression and purification of recombinant Vv-ANR
A cDNA library isolated from Vitis vinifera Cabernet Sauvignon
post-veraison berries was used for PCR amplification of the open
reading frame of the Vv-ANR gene. All steps, including PCR ampli-
fication, production of a His6-tag fused GB1-ANR construct,
expression of His6-tagged fused protein GB1-ANR in Escherichia
coli, Tev-mediated removal of the GB1 domain including the hexa-
histidine tag at its N-terminal end, and purification of Vv-ANR to
homogeneity, were performed as previously described (Gargouri et
al., 2009a). Vv-ANR was concentrated by ultrafiltration and stored
˚
matrices was 1.6 A, with 253 aligned Ca positions. The
structural homolog that ranked second to DFR in this search
was vestitone reductase, an NADPH-dependent enzyme
which reduces 2-hydroxy-isoflavanone-3R vestitone, here
again by means of hydride transfer to carbon 4 of the ben-
zopyran ring (Shao et al., 2007). As already mentioned, an
enzyme-assisted deprotonation of a phenolic hydroxyl group
at C5 or C7 is necessary to produce quinone methide inter-
mediates of Scheme 3. The catalytic triad of the short-chain
dehydrogenase/reductase superfamily includes a lysine and a
tyrosine residue which could fulfill this function, and which
are conserved in ANR (Tyr168 and Lys172) and DFR (Tyr163
and Lys167) from Vitis vinifera. In DFR, the side chain of
Tyr163 is indeed at hydrogen-bonding distance from the C5
hydroxyl group (Petit et al., 2007). With ANR, a three-
dimensional structural assessment of our mechanistic predic-
tions is not possible at this time, because all attempts to
at -208C upon dialysis against 10 m
M tricine, 50 mM NaCl, 10 mM
dithiothreitol, 5 m EDTA, pH 7.5, plus 2.5% glycerol (v/v). ANR
M
protein concentration was measured by monitoring the absorbance
at 280 nm, using a theoretical extinction coefficient of 28 795
-1
M
cm-1.
Enzyme assays
Anthocyanidin reductase assays were carried out at pH 7.5 and 308C
in the forward and backward directions of the reaction. Two distinct
incubation times (30 min and 16 h) were used for each assay. Reac-
tion products were then extracted with 2=500 ml ethylacetate,
supernatants were pooled and dried under argon and the products
were dissolved in 40 ml methanol, stored at -208C or directly used
Brought to you by | Purdue University Libraries
Authenticated
Download Date | 5/22/15 5:02 PM

Article in press - uncorrected proof
226 M. Gargouri et al.
for HPLC analysis. In the forward direction, the enzyme activity
was assayed in a buffer containing 50 m HEPES, pH 7.5, plus
5% methanol (v/v), 50 m anthocyanidin chloride, 200 m
NADPH, and 2 m enzyme in a total volume of 200 ml. Antho-
cyanidins were always introduced last in final reactional media as
aliquots of stock solutions prepared in methanol containing 4%
spray voltage was set to 4.5 kV, the capillary temperature was
2708C, and the capillary voltage was 32 V. MS/MS spectra of the
major (Mq1) parent ion derived from deuterated products, i.e.,
(epi)catechin or (epi)afzelechin, were obtained by means of colli-
sion-induced dissociation in the ion trap, using a normalized colli-
sion energy set at 35% of the instrument scale (Aime´ et al., 2008).
M
M
M
M
(v/v) of methane sulfonate buffer (10 mM, pH 2), and diluted to the
appropriate concentrations by monitoring the maximal absorbance
in the visible region. The spontaneous degradation of such stock
solutions was undetectable for 24 h at 48C, and for at least 4 days
at -208C. In the backward direction, the enzyme activity was
Acknowledgments
assayed in a buffer containing 50 m
methanol, 50 m cis- or trans-flavan-3-ol of defined stereochem-
istry, 500 m enzyme in a total volume
NADPq and 5 or 50 m
M HEPES, pH 7.5, plus 5%
We thank Dr. Bernard Badet (CNRS, UPR 2301, Gif-sur-Yvette,
France) for helpful discussions.
M
M
M
of 200 ml. For the experiment which required (H)-epicatechin, the
latter was first prepared from the forward (reducing) reactional
medium, using a semi-preparative dc18 reverse phase column of
10-mm diameter and the same elution system as that described
below with the analytical column.
References
Abrahams, S., Lee, E., Walker, A.R., Tanner, G.J., Larkin, P.J., and
Ashton, A.R. (2003). The Arabidopsis TDS4 gene encodes leu-
coanthocyanidin dioxygenase (LDOX) and is essential for
proanthocyanidin synthesis and vacuole development. Plant J.
35, 624–636.
Routine HPLC assay of 2,3-cis- and 2,3-trans-flavan-
3-ols produced by ANR
2,3-cis-Flavan-3-ols and 2,3-trans-flavan-3-ols were routinely
resolved and quantified without stereospecific discrimination of the
corresponding diastereoisomers. This was achieved with a reverse
Aime´, C., Plet, B., Manet, S., Schmitter, J.M., Huc, I., Oda, R.,
Sauers, R.R., and Romsted, L.S. (2008). Competing gas-phase
substitution and elimination reactions of gemini surfactants with
anionic counterions by mass spectrometry. Density functional
theory correlations with their bolaform halide salt models. J.
Phys. Chem. B 112, 14435–14445.
Gargouri, M., Manigand, C., Mauge´, C., Granier, T., Langlois
d’Estaintot, B., Cala, O., Piannet, I., Batany, K., Chaudie`re, J.,
and Gallois, B. (2009a). Structural and functional characteriza-
tion of anthocyanidin reductase from Vitis vinifera. Acta Cryst.
D 65, 989–1000.
˚
phase column (Atlantis dc18, 4.6=250 mm, 300 A, 5 mm, from
Waters, Versailles, France) under a flow rate of 1 ml min-1. A linear
elution gradient was applied for 20 min, from 10% to 90% aceto-
nitrile, in the presence of 0.08% trifluoroacetic acid. The absorbance
wavelength of the detector (996 photodiode Array Detector, Waters)
was set at 214 nm to ensure optimal sensitivity. 2,3-cis- and 2,3-
trans-Flavan-3-ols were quantified by means of external standards.
Gargouri, M., Gallois, B., and Chaudie`re, J. (2009b). Binding-equi-
librium and kinetic studies of anthocyanidin reductase from Vitis
vinifera. Arch. Biochem. Biophys. 491, 61–68.
HPLC assay used for stereochemical characterization
of (epi)catechin products
Mixtures of polyphenols produced by ANR were extracted and
redissolved in methanol as described above. The chiral separation
of isomeric products contained in these samples was achieved on a
4.6=250 mm (5 mm) Chiralcel OJ-H column (Daicel, Chiral Tech-
nologies, Illkirch, France). An isocratic elution was performed with
hexane/ethanol (70/30, v/v), using a flow rate of 0.5 ml/min and a
detector absorbance wavelength of 214 nm. (q)-Catechin, (-)-cate-
chin, and (-)-epicatechin were used as external standards. The reten-
tion time of (q)-epicatechin which was the only missing external
standard was established from a previous study that included mass
spectrometry and NMR investigations (Gargouri et al., 2009a).
Gourley, D.G., Schu¨ttelkopf, A.W., Leonard, G.A., Luba, J., Hardy,
L.W., Beverley, S.M., and Hunter, W.N. (2001). Pteridine reduc-
tase mechanism correlates pterin metabolism with drug resis-
tance in trypanosomatid parasites. Nat. Struct. Biol. 8, 521–525.
Li, H.J. and Deinzer, M.L. (2007). Tandem mass spectrometry for
sequencing proanthocyanidins. Anal. Chem. 79, 1739–1748.
Louie, G.V., Baiga, T.J., Bowman, M.E., Koeduka, T., Taylor, J.H.,
Spassova, S.M., Pichersky, E., and Noel, J.P. (2007). Structure
and reaction mechanism of basil eugenol synthase. PLoS One 2,
e993.
Petit, P., Granier, T., d’Estaintot, B.L., Manigand, C., Bathany, K.,
Schmitter, J.M., Lauvergeat, V., Hamdi, S., and Gallois, B.
(2007). Crystal structure of grape dihydroflavonol 4-reductase,
a key enzyme in flavonoid biosynthesis. J. Mol. Biol. 368,
1345–1357.
Pfeiffer, J., Ku¨hnel, C., Brandt, J., Duy, D., Punyasiri, P.A., Fork-
mann, G., and Fischer, T.C. (2006). Biosynthesis of flavan 3-ols
by leucoanthocyanidin 4-reductases and anthocyanidin reduc-
tases in leaves of grape (Vitis vinifera L.), apple (Malus=
domestica Borkh.) and other crops. Plant Physiol. Biochem. 44,
323–334.
Saito, K., Kobayashi, M., Gong, Z., Tanaka, Y., and Yamazaki, M.
(1999). Direct evidence for anthocyanidin synthase as a 2-oxo-
glutarate-dependent oxygenase: molecular cloning and function-
al expression of cDNA from a red forma of Perilla frutescens.
Plant J. 17, 181–189.
LC/ESI-MS and -MS/MS of ANR reaction products
The regiospecificity of deuterium incorporation from pro(S)-
NADPD into the final products was assessed by LC-MS and -MS/
MS analysis. Non-deuterated catechins and epicatechins were used
as external standards. The samples (20 ml) were separated using the
Atlantis dc18 column (vide infra) connected to an electrospray ion-
ization-ion trap mass spectrometer (Surveyor LC system and LCQ
Advantage, Thermo Fisher, Waltham, MA, USA). The gradient was
the same as that used for the routine HPLC assay, except that 0.05%
formic acid was used instead of 0.08% trifluoroacetic acid. The 1-
ml/min flow rate was split post-column, with a flow rate of only
0.2 ml/min being sent to the electrospray ionization source. The
spectrometer was operated in the positive electrospray mode. The
Brought to you by | Purdue University Libraries
Authenticated
Download Date | 5/22/15 5:02 PM

Article in press - uncorrected proof
Epimerase activity of anthocyanidin reductase and hydride transfers 227
Shao, H., Dixon, R.A., and Wang, X. (2007). Crystal structure of
vestitone reductase from alfalfa (Medicago sativa L.). J. Mol.
Biol. 369, 265–276.
Springob, K., Nakajima, J., Yamazaki, M., and Saito, K. (2003).
Recent advances in the biosynthesis and accumulation of antho-
cyanins. Nat. Prod. Rep. 20, 288–303.
chain dehydrogenase/reductase PTR1. Arch. Biochem. Biophys.
342, 197–202.
Wilmouth, R.C., Turnbull, J.J., Welford, R.W., Clifton, I.J., Prescott,
A.G., and Schofield, C.J. (2002). Structure and mechanism of
anthocyanidin synthase from Arabidopsis thaliana. Structure 10,
93–103.
Tanner, G.J., Francki, K.T., Abrahams, S., Watson, J.M., Larkin, P.J.,
and Ashton, A.R. (2003). Proanthocyanidin biosynthesis in
plants. Purification of legume leucoanthocyanidin reductase and
molecular cloning of its cDNA. J. Biol. Chem. 278, 31647–
31656.
Turnbull, J.J., Prescott, A.G., Schofield, C.J., and Wilmouth, R.C.
(2001). Purification, crystallization and preliminary X-ray dif-
fraction of anthocyanidin synthase from Arabidopsis thaliana.
Acta Crystallogr. D Biol. Crystallogr. 57, 425–427.
Veitch, N.C. and Grayer, R.J. (2008). Flavonoids and their glyco-
sides, including anthocyanins. Nat. Prod. Rep. 25, 555–611.
Wang, J., Leblanc, E., Chang, C.F., Papadopoulou, B., Bray, T.,
Whiteley, J.M., Lin, S.X., and Ouellette, M. (1997). Pterin and
folate reduction by the Leishmania tarentolae H locus short-
Wu, X., Liu, N., He, Y., and Chen, Y. (2009). Cloning, expression,
and characterization of a novel diketoreductase from Acineto-
bacter baylyi. Acta Biochim. Biophys. Sin. (Shanghai) 41, 163–
170.
Xie, D.Y., Sharma, S.B., Paiva, N.L., Ferreira, D., and Dixon, R.A.
(2003). Role of anthocyanidin reductase, encoded by
BANYULS in plant flavonoid biosynthesis. Science 299, 396–
399.
Xie, D.Y., Sharma, S.B., and Dixon, R.A. (2004). Anthocyanidin
reductases from Medicago truncatula and Arabidopsis thaliana.
Arch. Biochem. Biophys. 422, 91–102.
Received September 24, 2009; accepted October 29, 2009
Brought to you by | Purdue University Libraries
Authenticated
Download Date | 5/22/15 5:02 PM