Phosphate-catalyzed degradation of d-glucosone in aqueous solution is accompanied by C1-C2 transposition

Article abstract of DOI:10.1021/ja3020296

Pathways in the degradation of the C₆ 1,2-dicarbonyl sugar (osone) d-glucosone 2 (d-arabino-hexos-2-ulose) in aqueous phosphate buffer at pH 7.5 and 37 °C have been investigated by ¹³C and ¹H NMR spectroscopy with the use of singly and doubly ¹³C-labeled isotopomers of 2. Unlike its 3-deoxy analogue, 3-deoxy-d-glucosone (3-deoxy-d-erythro-hexos-2-ulose) (1), 2 does not degrade via a 1,2-hydrogen shift mechanism but instead initially undergoes C1-C2 bond cleavage to yield d-ribulose 3 and formate. The latter bond cleavage occurs via a 1,3-dicarbonyl intermediate initially produced by enolization at C3 of 2. However, a careful monitoring of the fates of the sketetal carbons of 2 during its conversion to 3 revealed unexpectedly that C1-C2 bond cleavage is accompanied by C1-C2 transposition in about 1 out of every 10 transformations. Furthermore, the degradation of 2 is catalyzed by inorganic phosphate (P_i), and by the P_i-surrogate, arsenate. C1-C2 transposition was also observed during the degradation of the C₅ osone, d-xylosone (d-threo-pentose-2- ulose), showing that this transposition may be a common feature in the breakdown of 1,2-dicarbonyl sugars bearing an hydroxyl group at C3. Mechanisms involving the reversible formation of phosphate adducts to 2 are proposed to explain the mode of P_i catalysis and the C1-C2 transposition. These findings suggest that the breakdown of 2 in vivo is probably catalyzed by P_i and likely involves C1-C2 transposition.

Full text of DOI:10.1021/ja3020296

Article
pubs.acs.org/JACS
Phosphate-Catalyzed Degradation of D-Glucosone in Aqueous
Solution Is Accompanied by C1−C2 Transposition
Wenhui Zhang and Anthony S. Serianni*
Department of Chemistry and Biochemistry, University of Notre Dame, Notre Dame, Indiana 46556, United States
S
* Supporting Information
ABSTRACT: Pathways in the degradation of the C6 1,2-
dicarbonyl sugar (osone) D-glucosone 2 (D-arabino-hexos-2-
ulose) in aqueous phosphate buffer at pH 7.5 and 37 °C have
1
been investigated by 13C and H NMR spectroscopy with the
use of singly and doubly 13C-labeled isotopomers of 2. Unlike
its 3-deoxy analogue, 3-deoxy-D-glucosone (3-deoxy-D-erythro-
hexos-2-ulose) (1), 2 does not degrade via a 1,2-hydrogen shift
mechanism but instead initially undergoes C1−C2 bond cleavage to yield D-ribulose 3 and formate. The latter bond cleavage
occurs via a 1,3-dicarbonyl intermediate initially produced by enolization at C3 of 2. However, a careful monitoring of the fates of
the sketetal carbons of 2 during its conversion to 3 revealed unexpectedly that C1−C2 bond cleavage is accompanied by C1−C2
transposition in about 1 out of every 10 transformations. Furthermore, the degradation of 2 is catalyzed by inorganic phosphate
(Pi), and by the Pi-surrogate, arsenate. C1−C2 transposition was also observed during the degradation of the C5 osone, D-
xylosone (D-threo-pentose-2-ulose), showing that this transposition may be a common feature in the breakdown of 1,2-dicarbonyl
sugars bearing an hydroxyl group at C3. Mechanisms involving the reversible formation of phosphate adducts to 2 are proposed
to explain the mode of Pi catalysis and the C1−C2 transposition. These findings suggest that the breakdown of 2 in vivo is
probably catalyzed by Pi and likely involves C1−C2 transposition.
INTRODUCTION
■
The current global diabetes epidemic1 has spurred renewed
interest in combating this debilitating disease that afflicts both
young and old. Current therapeutic approaches range from
better monitoring and control of blood glucose concentrations
to gene replacement and stem cell therapies to restore aberrant
insulin production and/or glucose transport.2−4 The elevated
detrimental effects of glycation in diabetic patients,9 although
glucose concentrations in the blood and tissues of diabetic
patients cause covalent modification of proteins through the
side effects from this treatment have been reported.10
process of glycation, the spontaneous, non-enzyme-catalyzed
reaction of the acyclic (aldehydo) form of glucose with lysine
side chains to initially form Schiff bases.5 The latter undergo
spontaneous rearrangement to form protein-bound Amadori
adducts, a process that is considered the committed (i.e.,
irreversible) step in protein glycation.6
The biochemical fates of Amadori adducts are still a matter of
debate, but it is commonly assumed that C−C bond
fragmentation in these adducts, or in their oxidized derivatives,
produces reactive oxygen species (ROS) which inflict oxidative
damage to proteins and other structures in their vicinity.7 This
damage is largely responsible for the multiple negative
biological consequences of glycation.
Glycation reduces the biological activities of proteins and
enzymes, especially those with long lifetimes.8 Thus, in addition
to the treatment strategies mentioned above, efforts have been
made to develop therapies that reduce glycation in vivo and/or
prevent the degradation of Amadori adducts to ROS. For
example, the B6 vitamer, pyridoxamine, appears to reduce the
In addition to the primary reaction between glucose and
proteins in vivo, secondary reactions also occur. Glucose
spontaneously rearranges and/or degrades in buffered
(phosphate) aqueous solution at pH 7.5, and the byproduct
reactive carbon species (RCS; e.g., methyl glyoxal and glyoxal)
from this degradation (autoxidation) in vivo may inflict greater
tissue damage than glucose itself.11 Among the potential
rearrangement products are the 1,2-dicarbonyl sugars (osones),
3-deoxy-D-glucosone (3-deoxy-D-erythro-hexos-2-ulose; 3DG;
1) and D-glucosone (D-arabino-hexos-2-ulose; GlcOS; 2).
3DG is produced in vivo by spontaneous (nonenzymic)
reaction of D-glucose with protein (Lys side chains) to give an
initial aldimine (Schiff base), which rearranges in several steps
to liberate free 3DG (Scheme 1).12 The same initial Schiff base
also partitions into the competing Amadori pathway (Scheme
S1, Supporting Information). Other pathways for the formation
Received: February 29, 2012
Published: May 31, 2012
© 2012 American Chemical Society
11511
dx.doi.org/10.1021/ja3020296 | J. Am. Chem. Soc. 2012, 134, 11511−11524

Journal of the American Chemical Society
Article
Scheme 1
of 1 in vivo are also possible.13,14 The production of 2 in vivo
may not be protein-mediated but rather initiates from D-
glucose, which rearranges to the 1,2-enediol, with subsequent
oxidation giving 2 (Scheme 2).12 Recent studies suggest that 2
may also form in vivo from the cleavage of Amadori products
mediated by peroxynitrite.15
(compressed air, ∼20 mL/min), and the solution pH was maintained
at 7.0 with periodic additions of 0.01 N NaOH until the pH stopped
dropping (∼3 h). The reaction mixture was then filtered though a 0.2
μm membrane filter, and the filtrate was concentrated at 30 °C in
vacuo. The solution was applied to a chromatographic column (2.5 ×
110 cm) containing Dowex 50 × 8 (200−400 mesh) ion-exchange
resin in the Ca2+ form.20 The column was eluted with distilled,
decarbonated water at ∼1.5 mL/min, and fractions (12 mL) were
collected and assayed by TLC (silica gel; spots detected by charring
after spraying with 1% (w/v) CeSO4−2.5% (w/v) (NH4)6Mo7O24−
10% aq H2SO4 reagent).21 Fractions nos. 22−29 containing osone 2
were pooled and concentrated at 30 °C in vacuo. The product 2 (168
mg, 0.94 mmol, 85% yield) was identified by its characteristic 13C
Scheme 2
We showed recently that 1 rearranges in aqueous phosphate
buffer at pH 7.5 and 37 °C to give two major products, 3-
deoxy-D-ribo-hexonic acid and 3-deoxy-D-arabino-hexonic
acid.16 Deuterium and 13C-labeling experiments revealed that
this rearrangement occurs through an intramolecular 1,2-
hydrogen transfer mechanism.17 Competing with this skeletal
rearrangement at a lower level is C1−C2 bond cleavage, which
yields formate and 2-deoxy-D-ribonate.16
Degradation studies of 1 led to the question posed in the
present work: Does the 3DG analogue, D-glucosone 2, degrade
in aqueous phosphate buffer in a manner similar to that found
for 1? We show in this report, through the use of 13C-labeled
isotopomers of GlcOS (21, 22, 2,1,2 and 21,3; superscripts denote
labeled carbons) and NMR that replacement of the CH2 group
at C3 of 1 by a HCOH group affects the degradation route
significantly. Furthermore, the degradation of 2 is accompanied
by an unexpected transposition of the C1−C2 fragment via a
mechanism mediated by inorganic phosphate.
chemical shifts,22 and purity was >95% based on 13C NMR assay.
Aqueous solutions of 2 were stored at 4 °C prior to use.
D-[1-13C]Xylosone 51 (D-[1-13C]threo-pentos-2-ulose) was prepared
from D-[1-13C]xylose as described previously by Vuorinen and
Serianni.23
C. General Procedure Used in Degradation Studies of 2 by
NMR. The 13C-labeled D-glucosone (21, 22, 21,2, or 21,3) was dissolved
in 100 mM sodium phosphate buffer at pH 7.5 to give a 10−20 mM
solution in 2 containing sodium azide (about 0.2 mg/mL solution)
EXPERIMENTAL METHODS
■
2
A. Reagents. D-[1-13C]Glucose, D-[2-13C]glucose, D-[1,2-13C2]-
glucose, D-[1,3-13C2]glucose, D-[1-13C]xylose, D-[1-13C]ribulose 31,
and DL-[1-13C]glyceraldehyde 41 were obtained from Omicron
Biochemicals, Inc. (South Bend, IN).
and H2O (10% v/v), and the resulting solution was incubated at 37
°C in the dark. NMR spectra of the reaction mixture were obtained
periodically by withdrawing an aliquot from the reaction mixture and
transferring it into a 5-mm NMR tube. NMR spectra were collected at
22 °C on a 600 MHz FT-NMR spectrometer as described below.
D. NMR Spectroscopy. High-resolution 1D 1H and 13C{1H}
NMR spectra of reaction mixtures were obtained at 22 °C. Samples
were analyzed in 5-mm NMR tubes on a 600-MHz FT-NMR
spectrometer equipped with a 5 mm 1H−19F/15N−31P dual broadband
B. Synthesis of 13C-Labeled D-Glucosones 2 and D-Xylosone
5. Four 13C isotopomers of GlcOS 2 were prepared: D-[1-13C]-
glucosone (21), D-[2-13C]glucosone (22), D-[1,2-13C2]glucosone (21,2),
and D-[1,3-13C2]glucosone (21,3). The synthetic route for their
preparation (Scheme S2, Supporting Information) involved treatment
of 13C-labeled D-glucoses with pyranose 2-oxidase (glucose 2-oxidase,
PROD, EC 1.1.3.10; Sigma).16,18,19 Labeled D-glucose (200 mg, 1.11
mmol) was dissolved in deionized water (60 mL) in a 250-mL three-
neck flask, and PROD (5.8 mg) and catalase (2.0 mg; Sigma) were
added. The solution was stirred gently at 25 °C and aerated
1
probe. H NMR spectra (600 MHz) were typically collected with a
∼7000 Hz spectral window and a ∼3.5 s recycle time. 13C{1H} NMR
spectra (150 MHz) were collected with ∼37 000 Hz spectral windows
and ∼3.0 s recycle times. FIDs were processed to optimize spectral S/
1
N, and final spectra had digital resolutions of ∼0.05 Hz/pt for H
11512
dx.doi.org/10.1021/ja3020296 | J. Am. Chem. Soc. 2012, 134, 11511−11524

Journal of the American Chemical Society
Article
NMR and ∼0.14 Hz/pt for 13C NMR. Chemical shifts were referenced
[1-13C]ribose 81 (103.58, 98.91, 96.45, and 96.16 ppm) and D-
[1-13C]arabinose 91 (99.44 and 95.24 ppm) were detected,26
with the former in greater abundance. The upfield region
contained signals from D-[1-13C]ribulose 31, D-[1-13C]xylulose
101 (68.63 and 65.13 ppm), and [2-13C]glycolate 72 (63.89
ppm). After 50 days of reaction (Figure S1, Supporting
Information), only signals from [1-13C]glycolate 71, [13C]-
formate, and [1-13C]arabinonate 61 were detected downfield,
and the anomeric carbon region showed new weak signals26
from D-[1-13C]lyxose 111 (96.73 ppm) in addition to those
arising from D-[1-13C]ribose 81 and D-[1-13C]arabinose 91. The
strong upfield signal at 63.88 ppm arises from [1-13C]glycolate
71.
externally to sodium 4,4-dimethyl-4-silapentane-1-sulfonate (DSS).
RESULTS AND DISCUSSION
■
A. Degradation of D-[2-13C]Glucosone 22. Testing for
the Formation of D-[1-13C]Ribulose 31. Prior studies
reported by Baynes and co-workers24 have shown that 2
degrades in aqueous phosphate buffer (200 mM, pH 7.4, 37
°C) to give D-ribulose 3 in 20% yield and other unidentified
products. A reasonable mechanism for the formation of 31 from
22 is shown in Scheme 3; C1−C2 bond cleavage in the
presumed 1,3-dicarbonyl intermediate is promoted by OH−
attack on C1 to give unlabeled formate as the second product.
Studies of 22 confirm the production of D-[1-13C]ribulose 31,
a result consistent with the mechanism shown in Scheme 3. In
addition, D-[1-13C]arabinonate 61 is detected as a degradation
product, with C1 of 22 presumably lost as unlabeled formate
(Scheme 4). Unexpectedly, H13COO− is detected early in the
degradation of 22, and glycolate 7 is detected with 13C-labeling
at either C1 or C2. D-[1-13C]Ribose 81, D-[1-13C]arabinose 91,
D-[1-13C]xylulose 101, and D-[1-13C]lyxose 111 presumably
arise from the D-[1-13C]ribulose 31 intermediate via isomer-
ization and epimerization.
Scheme 3
To test the mechanism in Scheme 3, 22 was incubated in 100
mM Na-phosphate buffer, pH 7.5, at 37 °C, and aliquots were
withdrawn from the reaction vessel for assay by 13C{1H} NMR.
The 13C{1H} NMR spectrum obtained after 1 day is shown in
Figure 1. The downfield region contained three weak signals
arising from D-[1-13C]arabinonate 61 (181.87 ppm), [13C]-
formate (173.62 ppm), and [13C]bicarbonate (162.89 ppm).
The anomeric region contained signals from unreacted 22
(103.41, 99.99, 96.29, and 95.72 ppm), and the upfield region
contained only signals25 arising from D-[1-13C]ribulose 31
(69.10, 65.45, and 65.19 ppm). After 15 days of reaction
(Figure 2), an additional downfield signal was observed from
[1-13C]glycolate 71 (182.48 ppm) (this signal first appeared
after ∼2 days of reaction; data not shown). The anomeric
region was virtually devoid of signals from unreacted 22, but D-
Additional insight into the degradation of 22 was obtained by
investigating the behavior of authentic D-[1-13C]ribulose 31
when incubated under reaction conditions similar to those used
in the degradation of 22. 13C{1H}NMR spectra of the reaction
solution after 1 day at room temperature, followed by 2 days
and 10 days at 37 °C, are shown in Figure S2 (Supporting
Information). These data show that the unidentified
intermediates (very weak signals at ∼62 ppm) in Figures 2D
and S1D apparently arise from the degradation of the D-
Figure 1. The 13C{1H} NMR spectrum of the reaction mixture with 22 (100 mM NaPi; pH 7.5; 37 °C) after 1 day. (A) Full spectrum. (B)
−
Downfield region showing signals from [1-13C]arabinonate 61 (a), H13COO− (b), and H13CO3 (c). (C) The anomeric carbon region showing
signals (a) from unreacted 22. (D) Upfield region showing signals (a) from D-[1-13C]ribulose 31 (α- and β-furanose and keto forms).25
11513
dx.doi.org/10.1021/ja3020296 | J. Am. Chem. Soc. 2012, 134, 11511−11524

Journal of the American Chemical Society
Article
Figure 2. The 13C{1H} NMR spectrum of the reaction mixture with 22 (100 mM NaPi; pH 7.5; 37 °C) after 15 days. (A) Full spectrum. (B)
Downfield region showing signals from D-[1-13C]arabinonate 61 (a), H13COO− (b), H13CO3− (c), and [1-13C]glycolate 71 (d). (C) The anomeric
carbon region showing signals from D-[1-13C]ribose 81 (a) and D-[1-13C]arabinose 91 (b). (D) Upfield region showing signals from D-[1-13C]ribulose
31 (α- and β-furanose and keto forms) (a),25 D-[1-13C]xylulose 101 (b), and [2-13C]glycolate 72 (c).
Scheme 4
Scheme 5
ribulose intermediate. Furthermore, after an extended reaction
period, only [13C]bicarbonate and [2-13C]glycolate 72 were
observed; signals from the intermediates at ∼62 ppm eventually
disappear.
The formation of labeled products other than D-[1-13C]-
ribulose 31 generated from 22 during degradation leads to
questions about the overall course of the reaction. How does
the [13C]formate produced early in the reaction arise?
[2-13C]Glycolate 72 arises presumably from the degradation
of the D-[1-13C]ribulose 31 intermediate, while [1-13C]glycolate
71 may arise from direct C2−C3 bond cleavage of 22 (Scheme
5). Degradation studies of D-[1-13C]glucosone 21 were
therefore undertaken to investigate these proposed trans-
formations.
still observed (Figure 4A and 4B), but no signals from
unreacted 21 were detected (Figure 4C). Anomeric carbon
signals, however, were observed for D-[1-13C]ribose 81 and D-
[1-13C]arabinose 91. The upfield region showed the presence of
D-[1-13C]ribulose 31 (69.09, 65.46, and 65.19 ppm) and
[2-13C]glycolate 72, although the latter contributed the
dominant signal in this region, in contrast to the results
obtained after 1 day (Figure 3D). This enhancement appears to
come at the expense of the unidentified intermediates, whose
signals are dominant in Figure 3D but relatively weak in Figure
4D.
B. Degradation of D-[1-13C]Glucosone 21. The 13C{1H}
NMR spectrum of the reaction mixture after 1 day of
degradation of 21 is shown in Figure 3. The strong downfield
signal at 173.62 ppm arises from [13C]formate (H13COO−),
presumably generated by C1−C2 bond cleavage in 21 via routes
shown in Schemes 3 and 4. The anomeric carbon region
(Figure 3C) contained four intense signals arising from
unreacted 21 (97.70, 97.27, 92.70, and 92.03 ppm) and several
weaker unassigned signals. Importantly, several upfield signals
between 62 and 66 ppm were observed, and two of these were
assigned to the α- and β-furanose forms of D-[1-13C]ribulose 31
(65.45 and 65.19 ppm) and a third to [2-13C]glycolate 72
(63.89 ppm). The closely spaced signals at ∼62 ppm (62.35,
62.25, and 62.24 ppm) are similar to the very weak signals
observed during the degradation of 22 (Figures 2D and S1D),
which were shown to arise during the degradation of 31 (Figure
S2). After 15 days, the strong signal arising from H13COO− was
The above NMR results with 21 as reactant show that, unlike
the degradation of 3-deoxy-D-glucosone 1,17 degradation
involving an intramolecular 1,2-hydrogen shift (benzylic acid
rearrangement) is not favored for 2, based on the absence of
carboxylate (C1) carbon signals attributable to D-[1-13C]-
gluconate 121 and D-[1-13C]mannonate 131 in Figures 3 and 4
(Scheme 6). The detection of H13COO− supports the
occurrence of C1−C2 bond cleavage in 21, yielding unlabeled
D-arabinonate as shown in Scheme 4.
During the degradation of 21, intermediate D-ribulose 3 is
produced with 13C-labeling at C1, an unexpected result
inconsistent with the route shown in Scheme 3 (i.e., when 21
is the reactant, 3 should be unlabeled). The detection of 31
11514
dx.doi.org/10.1021/ja3020296 | J. Am. Chem. Soc. 2012, 134, 11511−11524

Journal of the American Chemical Society
Article
Figure 3. The 13C{1H} NMR spectrum of the reaction mixture with 21 (100 mM NaPi; pH 7.5; 37 °C) after 1 day. (A) Full spectrum. (B)
Downfield region showing the intense H13COO− signal. (C) The anomeric carbon region showing signals (a) from unreacted 21. (D) Upfield region
showing signals from D-[1-13C]ribulose 31 (α- and β-furanose forms) (a), [2-13C]glycolate (b), and unidentified intermediates (c).
Figure 4. The 13C{1H} NMR spectrum of the reaction mixture with 21 (100 mM NaPi; pH 7.5; 37 °C) after 15 days. (A) Full spectrum. (B)
Downfield region showing the intense signal from H13COO−. (C) The anomeric carbon region showing signals from D-[1-13C]ribose 81 (a), D-
[1-13C]arabinose 91 (b), and an unidentified species (x) . (D) Upfield region showing signals from D-[1-13C]ribulose 31 (α- and β-furanose and keto
forms) (a), [2-13C]glycolate 72 (b), and unidentified intermediates (c).
suggests that C1−C2 bond cleavage in 2 may be preceded by
C1−C2 transposition. If this transposition had occurred during
the degradation of 22 (see above), the unlabeled D-ribulose 3
that formed (in addition to 31) would not be observed easily in
the spectrum, because the 13C NMR assay detects only those
carbons that are 13C-labeled (i.e., the concentration of
unlabeled 3 is too low to be observed).
Scheme 6
The detection of [2-13C]glycolate 72 during the degradation
of 21 is consistent with a route involving C2−C3 cleavage, as
shown in Scheme 5. If C1−C2 transposition occurred prior to
cleavage and degradation, then a portion of the [2-13C]glycolate
11515
dx.doi.org/10.1021/ja3020296 | J. Am. Chem. Soc. 2012, 134, 11511−11524

Journal of the American Chemical Society
Article
pool could be produced from the intermediate D-[1-13C]-
ribulose 31 (Scheme 5); the coexistence of both pathways is
only implied from these data based on the detection of 31 in the
reaction mixture.
C. Degradation of D-[1,2-13C2]Glucosone 21,2. Studies
with 21,2 were undertaken to determine the chemical fate of the
C1−C2 fragment during degradation, specifically to determine
the presence or absence of C1−C2 connectivity in the
degradation products as detected through signal splitting
from one-bond 13C−13C spin-coupling (1JCC).
The generation of D-[1-13C]ribose 81 and D-[1-13C]arabinose
91 (Figure 4C) presumably occurs by isomerization of 31 to the
corresponding C2-epimeric aldopentoses via a 1,2-enediol
intermediate. The detection of 81 and 91 thus provides
additional indirect evidence for the production of D-[1-13C]-
ribulose 31 during the degradation of 21.
After 1 day of reaction, H13COO− was detected (Figure 6).
Downfield signals were observed for [1,2-13C2]glycolate 71,2
(doublet from the carboxyl carbon; 182.67 ppm/182.30 ppm),
D-[1-13C]arabinonate 61 (181.88 ppm), and H13CO3 . The
−
The production of D-[1-13C]ribulose 31 and [13C]formate
during the degradation of both 21 and 22 must be accompanied
by the formation of unlabeled 3 and unlabeled formate if C1−
C2 transposition occurred during degradation. The unlabeled
fraction of these pools would not be detected by 13C NMR
given their low concentrations in solution. Because detection of
these unlabeled components would provide confirmatory
evidence for the proposed C1−C2 transposition, 21 and 22
were separately incubated in 100 mM phosphate buffer at pH
7.5 and 37 °C for 3 days, and the formate pools were analyzed
in the resulting solutions by 1H NMR (Figure 5). The formate
anomeric region was dominated by signals from the C1 and C2
carbons of unreacted 21,2, with each of the eight major signals
1
split into doublets due to JC1,C2. The upfield region contained
signals from D-[1-13C]ribulose 31 (69.08, 65.45, and 65.18
ppm), [1,2-13C2]glycolate 71,2 (from the hydroxymethyl carbon;
64.05 ppm/63.69 ppm)), and from unidentified intermediates
(∼62 ppm; ∼62.44 ppm/∼62.03 ppm). The latter signals
appeared as doublets, indicating that the C1−C2 fragment from
21,2 remained intact in these species.
After 6 days of reaction, the carboxyl signal from 71,2 became
more pronounced (182.67 ppm/182.30 ppm), signals from 21,2
disappeared, and the anomeric region contained signals from D-
[1-13C]ribose 81 and D-[1-13C]arabinose 91 (Figure 7). Signals
from 31 were still detected, in addition to those arising from the
C2 carbon of 7 and from the unidentified intermediates.
However, signals arising from the latter two species changed in
character; in both cases, a new line appeared in the center of
each doublet, suggesting the production of a second source of
glycolate during degradation, and revealing a pool of
unidentified intermediates that lacks an intact 13C−13C
fragment. These findings show that, early in the reaction, 7
forms directly from 21,2 via C2−C3 bond cleavage, giving 71,2,
whereas later in the reaction, when the solution contains little
or no 21,2, a second source of glycolate derives from the
degradation of intermediate D-[1-13C]ribulose 31 via C2−C3
bond cleavage, giving 72 (i.e., these findings confirm the two
routes proposed for glycolate production in Scheme 5).
After 22 days of degradation, little changed in the downfield
region of the 13C NMR spectrum, but the anomeric region
contained new signals from D-[1-13C]xylose 141 and D-
[1-13C]lyxose 111 (Figure S3 in Supporting Information).
The upfield region also contained new signals from D-
[1-13C]xylulose 101. The C2 signal of 7 appeared as a triplet
due to the presence of both 71,2 and 72 in solution (Figure S4,
Supporting Information). Similarly, signals from the unidenti-
fied intermediates at ∼62 ppm revealed a mixture of singly and
doubly 13C-labeled isotopomers. After 61 days (Figure S5,
Supporting Information), signals from all four D-[1-13C]-
aldopentoses were observed in the anomeric region, and the
upfield region was dominated by signals from [2-13C]glycolate
72 (major isotopomer) and [1,2-13C2]glycolate 71,2 (minor
isotopomer).
1
Figure 5. Downfield regions of H NMR spectra of reaction mixtures
after 3 days of reaction for 21 (A) and 22 (B) showing only signals
from formate at δ = 8.385 ppm. In spectrum A, [H13COO−]/
[H12COO−] ≅ 9/1; in spectrum B, this ratio is ∼1/9. In spectra A and
1
B, JCH in H13COO− was 194.8 Hz. The 13C isotope effect on δH in
formate (δ13C − δ12C) is −1.8 Hz at 600 MHz.
D. Degradation of D-[1,3-13C2]Glucosone 21,3. Degrada-
tion studies with 21,3 were conducted to test for the formation
of D-[1,2-13C2]ribulose 31,2, which is expected if C1−C2
transposition occurred during degradation. Detection of C1
signals in 3 split by the one-bond 13C−13C spin-coupling would
provide direct evidence for this process.
After 20 days of reaction, downfield signals due to D-
[2-13C]ribulose 32 (keto form; 215.14 ppm) and H13COO−
were detected (Figure 8). The dominant signals in the
anomeric carbon region arose from D-[2-13C]ribulose 32 (cyclic
forms; 108.26 and 105.15 ppm) and unreacted 21,3 (97.66,
pool produced from 21 contained ∼90% H13COO− and ∼10%
H12COO−, whereas that produced from 22 contained ∼10%
H13COO− and ∼90% H12COO−. Because these pools were
generated early in the reaction before substantial degradation of
2 had taken place, the detection of substantial unlabeled
formate from 21, and substantial labeled formate from 22,
provides indirect support for the hypothesis that C1 and C2 are
transposed during degradation. Experiments with doubly 13C-
labeled 2 were conducted to obtain more direct support for this
hypothesis.
11516
dx.doi.org/10.1021/ja3020296 | J. Am. Chem. Soc. 2012, 134, 11511−11524

Journal of the American Chemical Society
Article
Figure 6. The 13C{1H} NMR spectrum of the reaction mixture with 21,2 (100 mM NaPi; pH 7.5; 37 °C) after 1 day. (A) Full spectrum. (B)
−
Downfield region showing weak signals from [1,2-13C2]glycolate 71,2 (a), D-[1-13C]arabinonate 61 (b), H13COO− (c), and H13CO3 (d). Signal x
arises from an unidentified intermediate that disappears over time. (C) The anomeric carbon region showing signals (doublets) from C1 (a1) and C2
(a2) of unreacted 21,2. (D) Upfield region showing signals from D-[1-13C]ribulose 31 (a), [1,2-13C2]glycolate 71,2 (b), and unidentified intermediates
(c). Signal x was also unidentified.
Figure 7. The 13C{1H} NMR spectrum of the reaction mixture with 21,2 (100 mM NaPi; pH 7.5; 37 °C) after 6 days. (A) Full spectrum. (B)
−
Downfield region showing weak signals from [1,2-13C2]glycolate 71,2 (a), D-[1-13C]arabinonate 61 (b), H13COO− (c), and H13CO3 (d). Signal x
arises from an unidentified intermediate that disappears over time. (C) The anomeric carbon region showing signals from D-[1-13C]ribose 81 (a) and
D-[1-13C]arabinose 91 (b). (D) Upfield region showing signals from D-[1-13C]ribulose 31 (a), [1,2-13C2] 71,2 and [2-13C]glycolate 72 (b), and
unidentified intermediates (c). Signal x was also unidentified.
97.25, 92.68, and 92.01 ppm). Closer inspection of the former
signals revealed the presence of weak satellites on each signal
(108.43 ppm/108.08 ppm; 105.32 ppm/104.98 ppm)
indicative of a minor population of D-[1,2-13C2]ribulose 31,2.
The upfield region contained signals from D-[1,2-13C2]ribulose
31,2, with the C1 signals of the α- and β-furanose and keto
1
forms split by JC1,C2 values consistent with those reported
1
previously in authentic 3 (1JC1,C2 (αf) = 51.8 Hz; JC1,C2 (βf) =
51.3 Hz; JC1,C2 (keto) = 41.5 Hz).25 Signals from [2-13C]-
1
glycolate 72 (63.75 ppm) and unidentified intermediates (∼62
11517
dx.doi.org/10.1021/ja3020296 | J. Am. Chem. Soc. 2012, 134, 11511−11524

Journal of the American Chemical Society
Article
Figure 8. The 13C{1H} NMR spectrum of the reaction mixture with 21,3 (100 mM NaPi; pH 7.5; 37 °C) after 20 days. (A) Full spectrum showing
signals from D-[2-13C]ribulose 32 (keto and furanose forms) (a) and H13COO− (b). (B) The anomeric carbon region showing the C2 signals (cyclic
forms) of 32 (a), D-[1,2-13C2]ribulose 31,2 (satellites on a), and unreacted 21,3 (b). (C) Upfield region showing signals from 31,2 (a), [2-13C]glycolate
72 (b), and unidentified intermediates (c).
2
ppm) were also observed. After 29 days, the latter glycolate
signal contained satellites attributed to the presence of a small
population of [1,2-13C2]glycolate 71,2 (Figure 9).
produced by hydrogen−deuterium exchange with the H2O
2
1
solvent (∼10% H2O in H2O, v/v) at sites adjacent to C2
(H1R/S and/or H3) after extended incubation, which would
cause small upfield shifts in the C2 signal. The upfield region is
similar to that in Figure 8, although the [2-13C]glycolate signal
now dominated those from 31,2, and weak satellites on the C2
signal of 72 were still observable, indicating the presence of
[1,2-13C2]glycolate 71,2. In summary, the key finding from the
degradation studies of 21,3 is that D-[1,2-13C2]ribulose 31,2 is
produced, thus providing key confirmatory evidence that C1−
C2 transposition occurred during the degradation (Scheme 7).
E. Degradation of D-[1-13C]Xylosone 51. The degrada-
tion of D-[1-13C]xylosone 51 (D-[1-13C]threo-pentos-2-ulose)
was investigated to determine whether this structurally related
C5 osone degrades in a manner similar to that of 2. The
13C{1H} NMR spectrum of the reaction mixture after 2 days
(Figure S6, Supporting Information) showed an intense signal
from H13COO− (173.62 ppm), and essentially no unreacted 51
was detected. The upfield region contained signals from
[2-13C]glycolate 72 (63.88 ppm) and from unidentified
intermediates having chemical shifts (∼62 ppm) similar to
those observed in reactions with 21. Importantly, signals from
C1 (68.48 ppm), C3 (78.51 ppm), and C4 (65.56 ppm) of the
predominant keto form of 151 were also observed,27 with that
from C1 considerably more intense than those from C3 and
C4, suggesting that D-[1-13C]erythrulose 151 formed during the
reaction. Glycolate 72 presumably arises from C2−C3 bond
cleavage of 151, analogous to the behavior of 31 (Scheme 5).
These findings demonstrate that C1−C2 transposition detected
during the degradation of 2 is not unique to this C6 osone but
instead may be a general characteristic of osone degradation
when C3 bears an hydroxyl substituent.
Figure 9. The C2 signals of glycolate 7 detected after 29 days of
degradation of 21,3. Signals “a” arise from 71,2 and are split by 1JC1,C2
=
54.7 Hz. Signal “b” at δ 63.87 ppm arises from 72. The 13C isotope
shift for C2 (δC1,2 − δC2) is −1.4 Hz at 150 MHz.
After 76 days of reaction, the C2 signals of D-[2-13C]ribulose
32 and D-[2-13C]xylulose 102 were observed in the anomeric
carbon region of the spectrum (Figure 10). Inspection of the
former signals showed them to be complex, with each
composed of singlets bracketed by weak satellites. Weak
singlets slightly upfield of the main singlet were also observed.
These results suggest the presence of a major population of D-
[2-13C]ribulose 32 and a minor population of D-[1,2-13C2]-
ribulose 31,2 in solution; the upfield weak singlets are probably
F. The Effect of Buffer on the Degradation Rates of 2.
A potential role of the buffer in the degradation of 2 was
investigated by comparing the reaction rate in 100 mM
phosphate buffer (pH 7.5) at 37 °C with those observed in no
buffer (pH 9.5, solution pH adjusted with NaOH), in 400 mM
phosphate buffer (pH 7.5), in 100 mM MOPS buffer (pH 7.5),
11518
dx.doi.org/10.1021/ja3020296 | J. Am. Chem. Soc. 2012, 134, 11511−11524

Journal of the American Chemical Society
Article
Figure 10. The 13C{1H} NMR spectrum of the reaction mixture with 21,3 (100 mM NaPi; pH 7.5; 37 °C) after 76 days. (A) Full spectrum. (B) The
anomeric carbon region showing the C2 signals of D-[2-13C]ribulose 32 (a; signal intensities are off-scale) and D-[1,2-13C2]ribulose 31,2 (satellites).
Note the weak singlets slightly upfield of the central signal, which are probably due to 3 containing 2H at H1 and/or H3. D-[2-13C]Xylulose 102 is
also detected (b). (C) Upfield region showing signals from 31,2 (a) and [2-13C]glycolate 72 (b). The weak satellites bracketing the latter signal
indicate the presence of [1,2-13C2]glycolate 71,2 in the solution.
Scheme 7
because arsenate often behaves as a phosphate surrogate in
chemical and biochemical structures and transformations.28
Under these solution conditions, the degradation reaction
proceeded rapidly and gave time-lapse NMR data similar to
those obtained with phosphate buffer (Figure S8, Supporting
Information).
G. Degradation of DL-[1-13C]Glyceraldehyde 41.
13C{1H} NMR results from the degradation of different 13C-
isotopomers of 2 confirm that an early and major intermediate
is D-ribulose 3. However, the generation of 3 from 2 does not
occur exclusively by direct C1−C2 bond cleavage as shown in
Scheme 3. In addition to this route, degradation is accompanied
by C1−C2 transposition ∼10% of the time based on NMR
signal integrations (Figure 5). Once 3 forms, it undergoes
further cleavage to give glycolate 7, with the C1 and C2 carbons
of 7 arising from C2 and C1, respectively, of 3 (Scheme 5).
This cleavage may give D-glyceraldehyde 4 as the C3 byproduct.
13C NMR evidence for the transient formation of 4 at low
concentrations was obtained (from experiments with 24; data
not shown); 4 never accumulates in the reaction mixture,
presumably due to its lability under the reaction conditions
(i.e., its steady-state concentration is very low). The potential
contribution of 4 to the degradation profile of 2 was
investigated by examining the degradation of authentic DL-
[1-13C]glyceraldehyde 41 in 100 mM phosphate buffer at pH
7.5 and 37 °C (Figure S9, Supporting Information). After one
day, most of 41 had isomerized to [1-13C]dihydroxyacetone 161
([1-13C]1,3-dihydroxypropanone); the aldose−ketose equili-
brium favors the keto form under these conditions (Scheme 8)
(Figure S9B).29 Ketose 161 undergoes subsequent degradation
to yield [13C]formate and [2-13C]glycolate 72, presumably via
C1−C2 cleavage similar to what occurs in 2-ketopentose 3
and in 100 mM sodium citrate buffer. In the absence of buffer
at pH 9.5, little if any degradation occurred after 36 days
(Figure S7, Supporting Information), even though the solution
pH was three units higher than that of the standard phosphate
buffer used in this investigation. Aqueous solutions of 2 appear
to be very stable at elevated temperatures and moderately
alkaline conditions. These findings suggested that phosphate
may play a role in catalyzing the degradation of 2.
Reactions conducted in the MOPS and citrate buffers gave
degradation rates similar to that found with no buffer. However,
when 400 mM phosphate buffer was employed, the degradation
rate increased significantly relative to that found in 100 mM
phosphate buffer.
Trace metals are common contaminants in commercial
sources of sodium phosphate and thus could be responsible for
the observed rate enhancements in phosphate buffer. However,
the reaction rate in 100 mM phosphate containing 5 mM
EDTA was the same as that in 100 mM phosphate, providing
further evidence that phosphate plays a specific role in
catalyzing the degradation of 2.
Given the findings in phosphate buffer, a reaction was
conducted in 100 mM sodium arsenate (pH 7.5) at 37 °C,
11519
dx.doi.org/10.1021/ja3020296 | J. Am. Chem. Soc. 2012, 134, 11511−11524

Journal of the American Chemical Society
Article
Scheme 8
Scheme 9
(Figure S9C) (Scheme 5). These results confirm that 4 is
unstable under the solution conditions used for the degradation
of 2, thus explaining its lack of accumulation in the reaction
mixture.
CONCLUSIONS
■
The degradation of D-glucosone 2 in aqueous phosphate buffer
(pH 7.5, 37 °C) occurs by a series of reactions distinctly
different from that reported recently for 3-deoxy-D-glucosone
1.17 Unlike 1, osone 2 does not undergo carbon skeleton
rearrangement via intramolecular 1,2-hydrogen transfer to any
appreciable extent; neither D-gluconate 12 nor D-mannonate 13
could be detected in reaction mixtures. Instead, two backbone
fragmentation events are favored early in the degradation: C1−
C2 cleavage to give formate and D-ribulose 3, and C2−C3
cleavage to give glycolate and an unidentified C4 byproduct,
possibly D-erythrose or D-erythrulose. The D-ribulose inter-
mediate is susceptible to further degradation, which occurs
primarily by C2−C3 cleavage to give glycolate and a C3
byproduct, possibly D-glyceraldehyde or dihydroxyacetone.
Thus, in 2, degradation occurs primarily via backbone
fragmentation into 1- (formate) or 2- (glycolate) carbon
units; this behavior may have practical implications in the
conversion of sugar feedstocks to useful chemical precursors
and end-products. The identification of 2-ketopentose 3 as the
initially formed major degradation intermediate confirms prior
claims made by Baynes and co-workers,24 and the NMR studies
with 13C-labeled substrates lead to a mechanism for its
formation (Scheme 3). The degradation of the C3 and C4
byproducts generated from these C−C bond cleavage events
was not investigated here.
The production of D-ribulose 3 from D-glucosone 2, however,
does not occur exclusively by the mechanism shown in Scheme
3. If this mechanism were sufficient to account for all
conversions of 2 to 3, then 21 and 22 would yield only 3 and
31, respectively. The experimental data suggest otherwise. NMR
data obtained from the degradation of 21, 22, 21,2, and 21,3,
interpreted collectively, lead to the internally consistent
conclusion that the conversion can be accompanied by C1−
C2 transposition; for example, during the degradation of 21,3,
two 13C-isotopomers of 3 are produced, 32 and 31,2 (Scheme
9). Partitioning between these two 13C isotopomers is not
equal; ∼90% of 21,3 converts to 32, and ∼10% converts to 31,2.
C1−C2 transposition is a minor event but its occurrence is
nevertheless remarkable and unexpected. Furthermore, this
transposition does not appear to be unique to 2, because D-
xylosone 5 behaves similarly. It is thus likely that most, if not
all, 1,2-dicarbonyl sugars containing a hydroxyl group at C3
undergo the same rearrangement. 1,2-Dicarbonyl sugars
containing this consensus C1−C3 backbone substructure
possess a latent ability to undergo C1−C2 transposition during
degradation.
set of buffers was not examined, the present findings show that
inorganic phosphate (Pi), and the Pi surrogate, arsenate,
catalyze the degradation of 1,2-dicarbonyl sugars. Aqueous
solutions of 2 at moderately alkaline pH that are devoid of
buffer species are very stable, even at 37 °C. These findings
imply that 2 generated under in vivo conditions most likely
degrades via a Pi-catalyzed mechanism.
C1−C2 transposition during carbohydrate transformations,
although rare, has been documented previously, most notably
in the molybdate-catalyzed epimerization of aldoses (Scheme
10).30 In the latter process, dimolybdate is believed to bind
Scheme 10
reversibly to the acyclic hydrate (1,1-gem-diol) form of an
aldose substrate, and this complex promotes C2-epimerization
with concomitant C1−C2 transposition, doing so in a
stereospecific manner. Using this reaction as a model, possible
mechanistic explanations of the observed C1−C2 transposition
during the degradation of 2 were considered that involve
putative transient and reversible Pi adducts with the osone. In
doing so, an implicit assumption was made that not only does
Pi catalyze the overall rate of degradation, but it is also
associated with the concomitant C1−C2 transposition. Because
the rate of degradation of 2 is very low in the absence of Pi, it
remains unclear whether C1−C2 transposition occurs under
these solution conditions. Efforts were not made to conduct
degradation reactions over extended time periods to address
this question.
With the above caveats in mind, it is self-evident that the
significantly different pathways of degradation of 1 and 2 can be
attributed to differences in covalent structure at C3. The
presence of a C3 hydroxyl group α to the C2 carbonyl renders
H3 considerably more acidic in 2 than are either H3R or H3S
in 1. The reduced H3 acidities in 1 presumably render C1−C2
cleavage and/or 1,2-hydrogen transfer17 more favored than
enolization and subsequent rearrangement. Thus, a useful
starting point for considering how Pi-mediated C1−C2
The detection of C1−C2 transposition led to investigations
of the potential role of phosphate as a mediator of osone
degradation and/or C1−C2 transposition. While an exhaustive
11520
dx.doi.org/10.1021/ja3020296 | J. Am. Chem. Soc. 2012, 134, 11511−11524

Journal of the American Chemical Society
Article
Scheme 11
transposition might occur would be H3 abstraction and
subsequent enolization at C3.
Scheme 12
Potential ways by which Pi might form reversible adducts
with osone 2 were considered based on prior substrate
specificity studies of the enzyme, glycerol kinase (GK) (EC
2.7.1.30) (Scheme S3, Supporting Information).31 Both L-
glyceraldehyde and D-glyceraldehde bind to GK, but only the
former is phosphorylated to give L-glyceraldehyde 3P. When D-
glyceraldehde is the substrate, GK behaves like an ATPase,
presumably because phosphorylation occurs at one of the C1
hydroxyl groups of the putative bound 1,1-gem-diol form of the
substrate. Release of this phosphorylated product by GK is
followed by rapid hydrolysis in solution to give Pi and the
original starting triose. In this mechanism, Pi is transiently
bonded to a hemiacetal hydroxyl group, and this ester bond is
readily hydrolyzed in solution. Similar adducts thus appeared to
be reasonable intermediates in the Pi-mediated degradation of
2.
The above considerations led to a mechanism for the
conversion of osone 2 to ketose 3 involving Pi as a catalyst at
pH 7.5 (Scheme 11). Osone 2 reacts with Pi to give a 1-
phosphate adduct, which then cyclizes to give a 1,2-cyclic
phosphate intermediate (formation of the 2-phosphate could
also occur first). Adduct formation is favorable, because C1 and
C2 are electrophilic and thus susceptible to nucleophilic attack
by Pi. The intermediate 1,2-cyclic phosphate is now activated
for enolization, because abstraction of the relatively acidic H3 is
promoted by the proximal bound phosphate monoanion to give
the 2,3-enediol. Subsequent rearrangement of the latter gives a
1,3-cyclic phosphate, which can undergo hydrolysis to give an
unphosphorylated 1,3-dicarbonyl intermediate and Pi. The
phosphorylated 1,3-dicarbonyl intermediate (or its unphos-
phorylated equivalent) is postulated to be the key intermediate
in subsequent chemical transformations.
formation of glycolate from 2 (Scheme 13). Abstraction of the
acidic H2 generates a conjugated 1,2-enediol-3-keto inter-
mediate, which subsequently enolizes to a 2,3-dicarbonyl
intermediate. Cleavage of the C2−C3 bond in the latter,
promoted by OH−, gives glycolate 7 and D-erythrose. This
mechanism explains the formation of 71 from 22 (Scheme 5).
Scheme 13
The 1,3-cyclic phosphate serves as an intermediate in the
formation of 3 (Scheme 12). Attack by OH− at C1, with
concomitant cleavage of the C1−C2 bond, gives an
intermediate with the phosphate group tethered to the chemical
equivalents of formate and 3. Subsequent hydrolysis gives the
free ketose 3 after dehydration. Alternatively, OH− attack at C1
of the 1,3-cyclic phosphate and C1−C2 bond cleavage could
produce a untethered C1−C2 enediol intermediate (phosphate
serves as the leaving group) that undergoes isomerization to
give D-ribulose (also giving D-ribose and D-arabinose). The free
1,3-dicarbonyl sugar (see above) may also serve as a substrate
in the reaction sequence, in this case undergoing attack by OH−
to give 3 and formate directly.
The free 1,3-dicarbonyl intermediate, derived from hydrolysis
of the 1,3-cyclic phosphate in Scheme 11, is implicated in the
11521
dx.doi.org/10.1021/ja3020296 | J. Am. Chem. Soc. 2012, 134, 11511−11524

Journal of the American Chemical Society
Article
Scheme 14
Scheme 15
In summary, the mechanism shown in Scheme 11 explains
the observed catalysis by Pi in the degradation of 2 and
implicates a 1,3-cyclic phosphate intermediate. The latter
phosphate, or the free 1,3-dicarbonyl sugar derived from it,
rearranges to D-ribulose (Scheme 12). The 1,3-dicarbonyl
intermediate is also implicated in the formation of 7 from 2
(Scheme 13). Approximately 90% of 2 degrades through these
routes based on the results of 13C-labeling studies. The
remaining 10% of 2 follows a different route responsible for
the observed C1−C2 transposition.
tethered C2 and C4 fragments (Scheme 14). After bond
rotation, the latter are rejoined at C1 and C3 (aldol-like
condensation). An internal redox reaction involving 1,2-
hydrogen transfer (analogous to what is observed during the
degradation of 1)17 gives the 1,3-dicarbonyl intermediate,
which undergoes OH−-mediated C1−C2 bond cleavage to give
3 and formate. The key transformation in this mechanism is the
1,2-hydrogen transfer that occurs after bond rotation and the
aldol-like condensation.
In an alternate mechanism, the tethered C1−C2 fragment
(glycolaldehyde) in Scheme 14 can lose water to give a tethered
enediol fragment, with subsequent transfer of phosphate to the
equivalent C2 hydroxyl. Addition of water gives the original two
A potential mechanism for C1−C2 transposition is initiated
with the 1,3-cyclic phosphate intermediate, which undergoes
C2−C3 bond cleavage (retro-aldol-like) to give phosphate-
11522
dx.doi.org/10.1021/ja3020296 | J. Am. Chem. Soc. 2012, 134, 11511−11524

Journal of the American Chemical Society
Article
carbon fragment in which C1 and C2 have been inverted.
Condensation to the original C3 leads to 3 containing the
original C1 of the 1,3-cyclic phosphate at C1 of the ketose. This
alternate mechanism for C1−C2 transposition eliminates the
need for a 1,2-hydrogen transfer step in Scheme 14.
tions of these findings for human health, especially for diabetes,
are at least 2-fold. The steady-state concentration of 2 in vivo
under diabetic conditions is not well established, but it is
probably lower than that of 1. The extent to which 2 reacts with
proteins in vivo is unclear, although its intrinsic reactivity
relative to 1 is probably higher (greater electrophilic character
at C2). Future in vivo metabolic studies involving the use of
isotopically labeled 2 must now include the possibility of C1−
C2 transposition when establishing the fates of specific labeled
carbons; a simple backbone degradation pathway cannot be
assumed. The present work was conducted in vitro under
phosphate-buffered and temperature conditions that only
crudely approximate those in vivo, and thus it would be
worthwhile conducting degradation studies in plasma or other
biological media to test whether the degradation route is
affected, especially with respect to the C1−C2 transposition
component. Second, the degradation of 2 is accompanied by
the formation of multiple degradation products, including
formate, glycolate, and possibly glyceraldehyde, which
themselves may affect physiological function. While some of
these byproducts are probably harmless when generated
sporadically, long-term chronic exposure to them in the
diabetic condition may lead to metabolic aberrations not fully
appreciated at the present time.
The proposed mechanisms shown in Schemes 11−14 involve
Pi in various saccharide adducts during the degradation of 2. In
these mechanisms, Pi is covalently attached to reaction
intermediates either via an ester bond to a hemiacetal/
hemiketal hydroxyl group or via a mixed anhydride bond to a
carboxylic acid. These attachments allow facile hydrolysis to
liberate Pi when the rearrangement is complete. Potential
mechanisms in which Pi is attached via ester linkage to a
primary or secondary hydroxyl group were excluded; these
bonds are relatively stable, would presumably hydrolyze slowly
under the reaction conditions, and would be expected to
accumulate during the reaction. NMR studies showed no
evidence of stable phosphorylated intermediates or end-
products; the latter would have been suggested from the
observation of 13C−31P spin-coupling when the 31P atom is
two- or three-bonds removed from the carbon under
observation.32
The competing degradation routes for 2 in aqueous
phosphate buffer are summarized in Scheme 15. Pathways A,
B, and C consume 2, with pathway A involving C1−C2 bond
cleavage with or without C1−C2 transposition to give D-
ribulose 3 and formate. A second C1−C2 bond cleavage mode
(pathway C) gives D-arabinonate 6 and formate. The third
pathway (B) involves C2−C3 bond cleavage to give glycolate 7
and a C4 byproduct. Two secondary degradation pathways
emanate from intermediate 3. Pathway D involves C2−C3
bond cleavage to give glycolate 7 and a C3 byproduct, and
pathway E involves isomerization to give the four D-
aldopentoses (8, 9, 11, 14) and D-xylulose 10. In the present
work, reaction mixtures were not incubated for more than ∼90
days, and thus additional degradation pathways may exist in
more prolonged incubations.
The observation that ∼10% of 2 undergoes C1−C2
transposition during degradation leads to the question of
whether this transposition could be made more favorable. If the
mechanisms in Scheme 12 and 14 are correct, how might the
C2−C3 bond cleavage (retro-aldol-like) reaction be made more
favorable than C1−C2 bond cleavage? Might, for example, the
presence of an electronegative substituent such as fluorine at
C4 of the starting osone (e.g., 4-fluoro-D-glucosone) promote
C2−C3 bond cleavage over C1−C2 bond cleavage? A equally
intriguing question is whether 1,2-dicarbonyl sugars undergo
C1−C2 transposition in vivo, presumably involving Pi catalysis.
Answers to these questions invite further study.
The mechanisms shown in Schemes 11−14 could be tested
in at least two ways. Preparation of the proposed 1,3-dicarbonyl
intermediate, 3-keto-D-glucose (by C3 oxidation and depro-
tection of 1,2;5,6-di-O-isopropylidene-α-D-glucofuranose33),
and investigations of its degradation in the presence and
absence of Pi might shed light on its putative central role in the
degradation of 2. In addition, NMR studies with 2H/13C doubly
labeled 2 could prove useful to test the mechanism shown in
Scheme 14 in which 1,2-hydrogen transfer is implicated. These,
and other experiments, will be the subjects of future reports.
The primary motivation for conducting this work was
mechanistic, namely, to examine the effect of C3 structure on
the degradation of 1,2-dicarbonyl sugars; the unusual backbone
rearrangement that occurred was unanticipated. The implica-
ASSOCIATED CONTENT
■
S
* Supporting Information
Scheme S1, general scheme of protein glycation via the
Amadori pathway; Scheme S2, synthesis of 13C-labeled D-
glucosones; Scheme S3, action of glycerol kinase on D- and L-
glyceraldehyde; Figure S1, 13C{1H} NMR spectrum of the
reaction mixture with 22 (100 mM NaPi; pH 7.5; 37 °C) after
50 days; Figure S2, 13C{1H} NMR spectra of reaction mixtures
with D-[1-13C]ribulose 31 (100 mM NaPi; pH 7.5; 37 °C);
Figure S3, 13C{1H} NMR spectrum of the reaction mixture
with 21,2 (100 mM NaPi; pH 7.5; 37 °C) after 22 days; Figure
S4, expansion of the 13C NMR spectrum shown in Figure S3;
Figure S5, 13C{1H} NMR spectrum of the reaction mixture
with 21,2 (100 mM NaPi; pH 7.5; 37 °C) after 61 days; Figure
S6, 13C{1H} NMR spectrum of the reaction mixture with 51
(100 mM NaPi; pH 7.5; 37 °C) after 2 days; Figure S7,
13C{1H} NMR spectrum of the reaction mixture with 21 after
36 days at pH 9.5 (adjusted with NaOH) and 37 °C; Figure S8,
13C{1H} NMR spectrum of the reaction mixture with 21 (100
mM sodium arsenate; pH 7.5; 37 °C) after 4 days; Figure S9,
degradation of D-[1-13C]glyceraldehyde 41 in 100 mM
phosphate buffer, pH 7.5, at 37 °C; Figure S10, standard
13C{1H} NMR spectrum of sodium D-arabinonate 6 in 100 mM
NaPi buffer at pH 7.5 and 22 °C; Figure S11, standard 13C{1H}
NMR spectrum of sodium glycolate 7 in 100 mM NaPi buffer
at pH 7.5 and 22 °C; Figure S12, standard 13C{1H} NMR
spectrum of D-erythrulose 15 in 100 mM NaPi buffer at pH 7.5
and 22 °C. This material is available free of charge via the
Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
aseriann@nd.edu
Notes
The authors declare no competing financial interest.
11523
dx.doi.org/10.1021/ja3020296 | J. Am. Chem. Soc. 2012, 134, 11511−11524

Journal of the American Chemical Society
Article
(hydrate), 76.10; C4 (keto), 65.56; C4 (hydrate), 64.28. Under these
solution conditions, the keto/hydrate ratio is ∼92/8.
(28) Tawfik, D. S.; Viola, R. E. Biochemistry 2011, 50, 1128−1134.
(29) Davis, L. Bioorg. Chem. 1973, 2, 197−201.
ACKNOWLEDGMENTS
■
This work was supported by the National Institute of Diabetes
and Digestive and Kidney Disease (DK065138).
(30) (a) Hayes, M. L.; Pennings, N. J.; Serianni, A. S.; Barker, R. J.
Am. Chem. Soc. 1982, 104, 6764−6769. (b) Zhao, S.; Petrus, L.;
Serianni, A. S. Org. Lett. 2001, 3, 3819−3822.
REFERENCES
■
(1) Wild, S.; Roglic, G.; Green, A.; Sicree, R.; King, H. Diabetes Care
2004, 27, 1047−1053.
(31) Hayashi, S.; Lin, E. C. C. J. Biol. Chem. 1967, 242, 1030−1035.
(32) Serianni, A. S.; Pierce, J.; Barker, R. Biochemistry 1979, 18,
1192−1199.
(33) Morris, P. E., Jr.; Hope, K. D.; Kiely, D. E. J. Carbohydr. Chem.
1989, 8, 515−530.
(2) Bolen, S.; Feldman, L.; Vassy, J.; Wilson, L.; Yeh, H.-C.;
Marinopoulos, S.; Wiley, C.; Selvin, E.; Wilson, R.; Bass, E. B.;
Brancati, F. L. Ann. Intern. Med. 2007, 147, 386−399.
(3) Kuroda, M.; Bujo, H.; Aso, M.; Saito, Y. J. Diabetes Invest. 2011, 2,
333−340.
(4) Santamaria, X.; Massasa, E. E.; Feng, Y.; Wolff, E.; Taylor, H. S.
Mol. Ther. 2011, 19, 2065−2071.
(5) Zhang, Q.; Ames, J. M.; Smith, R. D.; Baynes, J. W.; Metz, T. O. J.
Proteome Res. 2009, 8, 754−769.
(6) Voziyan, P. A.; Khalifah, R. G.; Thibaudeau, C.; Yildiz, A.; Jacob,
J.; Serianni, A. S.; Hudson, B. G. J. Biol. Chem. 2003, 278, 46616−
46624.
(7) Chetyrkin, S. V.; Mathis, M. E.; Ham, A.-J.; Hachley, D. L.;
Hudson, B. G.; Voziyan, P. A. Free Red. Biol. Med. 2008, 44, 1276−
1285.
(8) Ulrich, P.; Cerami, A. Recent Prog. Hormone Res. 2001, 56, 1−22.
(9) Degenhardt, T. P.; Alderson, N. L.; Arrington, D. D.; Beattie, R.
J.; Basgen, J. M.; Steffes, M. W.; Thorpe, S. R.; Baynes, J. W. Kidney
Int. 2002, 61, 939−950.
(10) House, A. A.; Eliasziw, M.; Cattran, D. C.; Churchill, D. N.;
Oliver, M. J.; Fine, A.; Dresser, G. K.; Spence, J. D. JAMA 2010, 303,
1603−1609.
(11) Chetyrkin, S.; Mathis, M.; Pedchenko, V.; Sanchez, O. A.;
McDonald, W. H.; Hachey, D. L.; Madu, H.; Stec, D.; Hudson, B.;
Voziyan, P. Biochemistry 2011, 50, 6102−6112.
(12) Thornally, P. J.; Langborg, A.; Minhas, H. S. Biochem. J. 1999,
344, 109−116.
(13) Zyzak, D. V.; Richardson, J. M.; Thorpe, S. R.; Baynes, J. W.
Arch. Biochem. Biophys. 1995, 316, 547−554.
(14) Szwergold, B. S.; Howell, S.; Beisswenger, P. J. Diabetes 2001,
50, 2139−2147.
(15) Nagai, R.; Unno, Y.; Hayashi, M. K.; Masuda, S.; Hayase, F.;
Kinae, N.; Horiuchi, S. Diabetes 2002, 51, 2833−2839.
(16) Chetyrkin, S. V.; Zhang, W.; Hudson, B. G.; Serianni, A. S.;
Voziyan, P. A. Biochemistry 2008, 47, 997−1006.
(17) Zhang, W.; Carmichael, I.; Serianni, A. S. J. Org. Chem. 2011, 76,
8151−8158.
(18) Giffhorn, F.; Kopper, S.; Huwig, A.; Freimund, S. Enzyme
̈
Microb. Technol. 2000, 27, 734−742.
(19) Freimund, S.; Huwig, A.; Giffhorn, F.; Kopper, S. Chem.Eur. J.
̈
1998, 4, 2442−2455.
(20) Angyal, S.; Bethell, G. S.; Beveridge, R. Carbohydr. Res. 1979, 73,
9−18.
(21) Tropper, F. D.; Andersson, F. O.; Grand-Maitre, C.; Roy, R.
Carbohydr. Res. 1992, 229, 149−154.
(22) Freimund, S.; Baldes, L.; Huwig, A.; Giffhorn, F. Carbohydr. Res.
2002, 337, 1585−1587.
(23) Vuorinen, T.; Serianni, A. S. Carbohydr. Res. 1990, 207, 185−
210.
(24) Wells-Knecht, K. J.; Zyzak, D. V.; Litchfield, J. E.; Thorpe, S. R.;
Baynes, J. W. Biochemistry 1995, 34, 3702−3709.
(25) Vuorinen, T.; Serianni, A. S. Carbohydr. Res. 1990, 209, 13−31.
(26) (a) Drew, K. N.; Zajicek, J.; Bondo, G; Bose, B.; Serianni, A. S.
Carbohydr. Res. 1998, 307, 199−209. (b) King-Morris, M. J.; Serianni,
A. S. J. Am. Chem. Soc. 1987, 109, 3501−3508.
(27) The 13C NMR signals for D-[1-13C]erythrulose 151 were
assigned based on spectral data obtained from a commercial (Sigma)
unlabeled sample of 15. The following 13C NMR signal assignments
(in ppm) were made in 100 mM NaPi buffer, pH 7.5, 9/1 v/v
1H2O/2H2O, 200 mM in 15, 22 °C: C1 (keto), 68.48; C1 (hydrate),
66.73; C2 (keto), 215.00; C2 (hydrate), 98.47; C3 (keto), 78.51; C3
11524
dx.doi.org/10.1021/ja3020296 | J. Am. Chem. Soc. 2012, 134, 11511−11524