Stereospecific molybdic acid-catalyzed isomerization of 2-hexuloses to branched-chain aldoses

Article abstract of DOI:10.1016/S0008-6215(99)00112-3

On treatment with a catalytic amount of molybdic acid in aqueous solution, the 2-ketohexoses D-fructose, L-sorbose and D-tagatose undergo a stereospecific intramolecular rearrangement to give the corresponding 2-C-(hydroxymethyl)aldoses, 2-C-(hydroxymethyl)-D-ribose (D-hamamelose), 2-C-(hydroxymethyl)-L-lyxose, and 2-C-(hydroxymethyl)-D-xylose, respectively. At equilibrium, the ratio of 2-ketose to 2-C-(hydroxymethyl)aldose ranged from 14:1 (fructose) to 32:1 (sorbose). A similar treatment of D-psicose failed to yield a significant amount of the corresponding branched-chain aldose. The equilibria can be shifted with the addition of boric acid to the reaction mixture; under these conditions, ratios of 3:1 and 7:1 were obserwed for D-fructose and L-sorbose, respectively. A mechanistic study with D-(3-¹³C)fructose afforded D-(1-¹³C)hamamelose, thus confirming C-3 - C-4 bond cleavage with concomitant C-2 - C-3 transposition suggested from recent studies with D-(2-¹³C)fructose. Copyright (C) 1999 Elsevier Science Ltd.

Full text of DOI:10.1016/S0008-6215(99)00112-3

www.elsevier.nl/locate/carres
Carbohydrate Research 319 (1999) 38–46
Stereospecific molybdic acid-catalyzed isomerization of
2-hexuloses to branched-chain aldoses
a
a
Zuzana Hricov´ıniova´-B´ılikova´ , Milos
'
Hricov´ıni a, Ma´ria Petrus
ova´ ,
'
Anthony S. Serianni b, Ladislav Petrus a,
'
*
a Institute of Chemistry, Slo6ak Academy of Science, SK-842 38 Bratisla6a, Slo6ak Republic
b Department of Chemistry and Biochemistry, Uni6ersity of Notre Dame, Notre Dame, IN 46556-5670, USA
Received 22 February 1999; accepted 27 April 1999
Abstract
On treatment with a catalytic amount of molybdic acid in aqueous solution, the 2-ketohexoses
-sorbose and -tagatose undergo a stereospecific intramolecular rearrangement to give the corresponding 2-C-(hy-
droxymethyl)aldoses, 2-C-(hydroxymethyl)-D-ribose ( -hamamelose), 2-C-(hydroxymethyl)- -lyxose, and 2-C-(hy-
droxymethyl)- -xylose, respectively. At equilibrium, the ratio of 2-ketose to 2-C-(hydroxymethyl)aldose ranged from
14:1 (fructose) to 32:1 (sorbose). A similar treatment of -psicose failed to yield a significant amount of the
corresponding branched-chain aldose. The equilibria can be shifted with the addition of boric acid to the reaction
mixture; under these conditions, ratios of 3:1 and 7:1 were obserwed for -fructose and -sorbose, respectively. A
mechanistic study with
-(3-13C)fructose afforded -(1-13C)hamamelose, thus confirming C-3ꢀC-4 bond cleavage
D-fructose,
L
D
D
L
D
D
D
L
D
D
with concomitant C-2ꢀC-3 transposition suggested from recent studies with
Science Ltd. All rights reserved.
D
-(2-13C)fructose. © 1999 Elsevier
Keywords: Branched-chain aldoses; 2-C-(Hydroxymethyl)aldoses; Molybdic acid-catalysis; Carbon skeleton rearrangement; D-(3–
13C)Fructose; Mechanism study
the starting aldose becomes C-2 of the product
aldose and vice versa [8–10]. It has been
proposed that this exchange of carbon atoms
with their substituents is mediated by a tetra-
dentate dimolybdate complex involving four
neighboring hydroxyl groups, including one of
the aldehydrol hydroxyls. This complex pro-
motes the formation of a new C-1ꢀC-3 bond
while simultaneously breaking the C-2ꢀC-3
bond, thereby causing epimerization with con-
comitant C-1ꢀC-2 transposition. The process
as applied to isotopically uniform aldoses for-
mally results in C-2 epimerization only.
1. Introduction
The mutual interconversion of C-2-epimeric
aldoses catalyzed by molybdic acid is known
as the B´ılik reaction [1–7]. Due to its simplic-
ity, this reaction has been used to prepare
several rare aldoses on a commercial scale.
Mechanistic studies with regiospecifically hy-
drogen- and carbon-isotopically labelled and
substituted aldoses have revealed a remark-
able carbon skeleton rearrangement that ac-
companies C-2-epimerization in which C-1 of
Recently it has been shown that treatment
of 2-C-(hydroxymethyl)-
nose [11,12] and
acid results in dramatic structural changes.
D-glucose and -man-
* Corresponding author. Tel.: +421-7-5941-0272; fax: +
421-7-5941-0222.
E-mail address: chemlpet@savba.sk (L. Petrus' )
D-fructose [13] with molybdic
0008-6215/99/$ - see front matter © 1999 Elsevier Science Ltd. All rights reserved.
PII: S0008-6215(99)00112-3

Z. Hrico6´ınio6a´ -B´ıliko6a´ et al. / Carbohydrate Research 319 (1999) 38–46
39
The two branched-chain aldoheptoses rear-
NMR spectra were recorded with spectral
widths of 1500 Hz and typically 16 transients
were accumulated to obtain sufficient signal/
noise ratios. Presaturation of the HDO reso-
nance was achieved by low-power irradiation
during part of the relaxation delay. A QNP
probe (5 mm) was used for the measurement
of 1D 13C NMR spectra. Standard 2D tech-
niques with pulse field gradients were used to
obtain COSY, HMBC and HSQC data; the
latter experiment was performed in the phase
sensitive-enhanced pure-absorption mode [14].
A 60 ms delay time was set for the measure-
ment of long-range proton–carbon coupling
constants in the HMBC experiment. Spectral
widths in heteronuclear experiments were 1200
Hz (1H) and 5000 Hz (13C), and spectra were
zero-filled before Fourier transformation to
give final digital resolutions of 1.2 and 5.9
Hz/pt, respectively.
ranged stereospecifically and reversibly to
D-
manno-heptulose
and
-fructose gave 2-C-(hy-
-hamamelose). The
D
-gluco-heptulose,
respectively, while
D
droxymethyl)-
D
-ribose (
D
equilibria of these reactions strongly favored
the 2-ketoses. The carbon skeleton rearrange-
ment which was expected to accompany these
13
1
interconversions was confirmed by C and H
NMR studies of the conversion of -(2-
13C)fructose to -(2-13C)hamamelose [13].
In this report an improved strategy for the
preparation of -hamamelose and studies of
the molybdic acid-catalyzed isomerization of
three additional 2-ketohexoses ( -sorbose,
tagatose and -psicose) to their corresponding
D
D
D
L
D-
D
2-C-branched aldoses are described. Further-
more, in order to examine the mechanism of
this isomerization in more detail, the behavior
of
D
-(3-13C)fructose upon treatment with a
catalytic amount of molybdic acid was
investigated.
Isomerization of hexuloses.—A solution of
L-sorbose,
D
-tagatose or
D-psicose (100 mg,
5.6 mmol) in 0.2% aqueous molybdic acid (5
mL) was heated at 80 °C. Samples (0.5 mL)
were taken at selected intervals (1, 2, 3, 5, 7, 10
2. Experimental
h
for
L-sorbose and for
D
-tagatose;
5, 10, 20, 30, 40, 50 h for
D
-psicose) and
General methods.—Melting points were de-
termined on an electrothermal Kofler appara-
tus. Specific optical rotations were measured
at 20 °C with an automatic polarimeter
(Perkin–Elmer, model 141) using a 10-cm 1-
mL cell. The composition of mixtures was
examined by descending PC on Whatman No.
1 sheets using the solvent system S1: 5:1:4
butan-1-ol–ethanol–water, or S2: 8:2:1 ethyl
acetate–pyridine–water as mobile phases, fol-
lowed by visualization with alkaline silver ni-
trate. Preparative chromatography was
performed on a column (denoted C1) (95
cm×1.6 cm) of Dowex 50W X8 (200–400
mesh) in the Ba2+ form, using water as the
eluant, usually at a flow rate of ꢀ5 mL/h.
Evaporations were conducted under reduced
pressure at 40 °C.
treated with Amberlite IRA 400 (HCO3−, 3
mL), filtered, and the filtrates were evapo-
rated, exchanged with D2O and analyzed by
1H NMR. Selected resonances were integrated
to determine the ratios of the sugars present.
Preparation
of
2-C-(hydroxymethyl)-
L-
lyxose.—(A) A solution of
L
-sorbose (1 g, 5.6
mmol) and boric acid (1.4 g, 22.6 mmol) in 2%
aqueous molybdic acid (5 mL) was heated at
80 °C for 15 h. The reaction mixture was
evaporated to dryness and then again four
times from methanol (3×10 mL). The residue
was dissolved in water (10 mL) and the mix-
ture was treated with Amberlite IRA 400 in
the HCO3− form (50 mL). The resin was re-
moved by filtration and washed with water
(3×5 mL). The deionized solution was con-
centrated and crystallized from MeOH to af-
NMR spectroscopy.—1H (300.13 MHz) and
13C (75.45 MHz) NMR spectra were recorded
in D2O solutions at 40 °C on a Bruker DPX
300 spectrometer equipped with a 5 mm in-
verse broadband probe with a shielded z-gra-
dient. Proton and carbon chemical shifts were
ford a part of the starting -sorbose (0.7 g, 3.9
L
mmol). The mother liquor was fractionated on
column C1. Fraction 1 (eluting between 110
and 130 mL) gave a second portion of recov-
ered
L
-sorbose (110 mg, 0.6 mmol) after evap-
1
expressed relative to internal TSP. 1D H
oration. Fraction 2 (eluting between 130 and

40
Z. Hrico6´ınio6a´ -B´ıliko6a´ et al. / Carbohydrate Research 319 (1999) 38–46
Subsequent rechromatography on C1 afforded
a 1:1 mixture of the two sugars (0.12 g; eluting
between 160 and 165 mL). Final preparative
PC (Whatman No. 3, solvent S2, 12 h; Rf 1.27
140 mL) was a 2:1 mixture of
L
-sorbose and
2-C-(hydroxymethyl)-
L
-lyxose (75 mg). Frac-
tion 3 (eluting between 140 and 160 mL)
contained chromatographically pure 2-C-(hy-
for
D-tagatose, 1.69 for 2-C-(hydroxymethyl)-
droxymethyl)- -lyxose (92 mg, 0.5 mmol, 9%),
L
D
-xylose) afforded pure 2-C-(hydroxymethyl)-
which after evaporation was obtained as a
D
-xylose (40 mg, 0.6 mmol, 0.5%): mp
colorless syrup: Rf 1.21 (S2), [h]D (c 1, water)
1
108–110 °C (EtOH), [h]D (c 1, water)
−3.1“ −2.6° (24 h); H NMR (D2O, 300.13
1
MHz): l 5.01 (H-1 ap), 4.77 (H-1 bp), 3.90
(H-5e bp), 3.82–3.87 (H-4 a, bp), 3.63–3.82
(H-5a, H-5e 3ap), 3.72 (CH2 (C-2) ap), 3.55–
3.68 (CH2 (C-2) bp), 3.61 (H-3 ap), 3.59 (H-3
bp), 3.21 (H-5a bp); 13C NMR (D2O, 75.45
MHz): l 103.49 (C-1 af ), 99.63 (C-1 bf ),
97.54 (C-1 bp), 97.19 (C-1 ap), 83.50 (C-2 af ),
82.95 (C-4 bf ), 82.17 (C-2 bf ), 82.04 (C-4 af ),
78.18 (C-2 ap), 77.64 (C-2 bp), 74.93 (C-3 bp),
74.93 (C-3 bf ), 74.21 (C-3 ap), 73.66 (C-3 af ),
69.97 (C-4 ap), 69.97 (C-4 bp), 67.71 (C-5 bp),
66.32 (CH2(C-2) ap), 65.93 (CH2(C-2) bf ),
65.36 (C-5 bf ), 64.11 (CH2(C-2) af ), 65.93
(CH2(C-2) bf ), 65.36 (C-5 bf ), 64.11 (CH2(C-
2) af ), 64.11 (C-5 ap), 63.95 (C-5 af ), 63.32
(CH2(C-2) bp).
+29.6“ +16.5 (24 h) [16]. H l 5.05 (H-1
ap), 4.70 (H-1 bp), 4.07 (H-5e bp), 3.86 (CH2
(C-2) bp), 3.83 (H-3 ap), 3.67–3.78 (CH2 (C-
2) ap), 3.73 (H-5a, H-5e ap), 3.69–3.74 (H-4
a, bp), 3.57 (H-3 bp), 3.37 (H-5a bp). 13C
NMR (D2O, 75.45 MHz): l 100.61 (C-1 bp),
95.18 (C-1 ap), 78.02 (C-3 bp), 77.56 (C-2 ap),
76.01 (C-2 bp), 76.01 (C-3 ap), 70.75 (C-4 ap),
70.75 (C-4 bp), 66.67 (C-5 bp), 64.86 (C-5 ap),
64.13 (CH2(C-2) ap), 63.60 (CH2(C-2) bp).
Preparation of
D
-hamamelose.— -Fructose
D
(0.5 g, 2.8 mmol) and boric acid (0.7 g, 11.3
mmol) were dissolved in 2% aqueous molyb-
dic acid (2.5 mL) and the reaction conditions
and purification of products followed those
for the preparation of 2-C-(hydroxymethyl)- -
L
Work-up of fraction 4 (eluting between 170
lyxose. Three chromatographic fractions were
obtained. Fraction 1 (eluting between 115 and
and 200 mL) gave
L-fructose (12 mg, 0.07
mmol, 1%) as a syrup: [h]D (c1, water) +95°
(24 h). Obtained data were in accordance with
those published previously [15].
135 mL) contained -sorbose, which was crys-
D
tallized from MeOH to yield 45 mg (0.25
mmol, 9%) of product; mp 164 °C, [h]D (c 1.0,
water) + 42°; Rf 0.91 (S1); Rf 0.96 (S2). These
data and the 13C chemical shifts were in accor-
dance with those published previously [15,17].
Fraction 2 (eluting between 175 and 205 mL)
(B) The same procedure conducted without
the addition of boric acid to the reaction
mixture afforded 32 mg (0.2 mmol, 3%) of
2-C-(hydroxymethyl)-
mmol, 1%) of -fructose.
Preparation of 2-C-(hydroxymethyl)-
lose.—A solution of -tagatose (7 g, 38.9
L
-lyxose and 9 mg (0.05
L
contained recovered
D-fructose (280 mg, 1.6
D-xy-
mmol, 57%). Concentration and crystalliza-
tion of fraction 3 (eluting between 225 and
D
mmol) in 0.2% aqueous molybdic acid (35
mL) was heated at 80 °C for 15 h. After
cooling, the reaction mixture was treated
batchwise with Amberlite IRA 400 in the
HCO3− form (300 mL). The resin was re-
moved by filtration and washed with water
(3×30 mL). The deionized solution was con-
centrated, and the residue was crystallized
from MeOH to afford a part of the starting
300 mL) gave pure
D
-hamamelose (101 mg,
0.6 mmol, 20%), mp 110–111 °C (absolute
ethanol), [h]D (c 1, water) +7.5“ −7.1°; Rf
1.26 (S1); Rf 1.88 (S2). These data were in
accordance with data published previously
[18].
Isomerization of
tion of
D
-(3-13C)fructose.—A solu-
D
-(3-13C)fructose (170 mg, 0.9 mmol)
in 0.2% aqueous molybdic acid (1.8 mL) was
heated at 80 °C for 5 h. After cooling, the
reaction mixture was stirred with Amberlite
IRA-400 in the HCO3− form (2 mL) for 15
min, filtered, and the resin was washed with
water (3×3 mL). The filtrate was concen-
trated under diminished pressure to give a
D
-tagatose (5.5 g, 30.6 mmol). The mother
liquor was twice fractionated on column C1. A
3:1 mixture of starting -tagatose and 2-C-
(hydroxymethyl)- -xylose (0.63 g) eluted first
D
D
(eluting between 150 and 170 mL and contain-
ing a substantial fraction of the latter sugar).

Z. Hrico6´ınio6a´ -B´ıliko6a´ et al. / Carbohydrate Research 319 (1999) 38–46
41
1
13
light-yellow syrup (175 mg). H and C NMR
spectra revealed the presence of three compo-
lyzed isomerization of
cal reaction conditions did not afford the
D
-psicose under identi-
nents,
namely,
and
D
-(3-13C)sorbose,
D-(3-
expected 2-C-(hydroxymethyl)-D-arabinose as
13C)fructose
D
-(1-13C)hamamelose.
1
determined by H NMR, even after 50 h of
Separation of this mixture on column C1 at a
flow rate of 5 mL/h afforded three fractions.
Fraction 1 (eluting between 90 and 105 mL)
reaction. The addition of boric acid to the
reaction mixture failed to shift the equilibria
with
D-tagatose and
D-psicose towards the
contained
D
-(3-13C)sorbose (4.5 mg, 0.03
corresponding 2-C-(hydroxymethyl)aldoses.
The different yields of 2-C-(hydrox-
ymethyl)aldoses produced by 2-ketose rear-
rangement under the conditions of the B´ılik
reaction may be caused by different amounts
of cyclic and acyclic molybdate complexes of
the 2-ketohexoses in the reaction mixtures.
mmol, 3%), fraction 2 (eluting between 125
and 155 mL) contained recovered
D-(3-
13C)fructose (150 mg, 0.8 mmol, 88%), and
fraction 3 (eluting between 185 and 235 mL)
contained
D
-(1-13C)hamamelose (12 mg, 0.07
mmol, 7%).
For
D-tagatose and especially
D-psicose, cata-
lytically inactive cyclic species are expected to
predominate in aqueous solution [19–21]. In
3. Results and discussion
contrast,
D
-fructose and
L
-sorbose, which
have the threo configuration at the C-3ꢀC-4
fragment, form mainly acyclic molybdate
complexes, which are expected to be catalyti-
cally active [13].
The treatment of
aqueous molybdic acid at 80 °C for 5 h gave a
mixture of -sorbose and 2-C-(hydrox-
ymethyl)- -lyxose in a 32:1 ratio. The ratio
L
-sorbose with 0.2%
L
L
was determined by integration of separated 1H
The mechanism of transformation of
D
-
signals between 3.42 and 3.53 ppm of
L
-sor-
-sor-
fructose to -hamamelose has been studied
D
1
recently by H and 13C NMR spectroscopy
bose (the H-1b and H-4 signals of a-
L
using 13C-labeled ketose [13].
Fructose was converted to
D
-(2-13C)-
-(2-13C)-
bopyranose, respectively) and the anomeric
signals of the branched chain -lyxose (4.75–
L
D
5.25 ppm). This ratio can be considered to be
a good estimate of the thermodynamic equi-
librium since it remained unchanged even af-
ter an additional 5 h of reaction. When the
reaction was performed in the presence of four
equivalents of boric acid (relative to the ketose
starting material), a new ratio of 7:1 was
achieved and the product 2-C-(hydroxy-
hamamelose stereospecifically, as determined
from 1D and 2D NMR spectra in which
1
additional H–1H splittings (1H spectra) and
13
the appearance of C resonances correspond-
ing to the quarternary C-2 were observed. In
the present study, further evidence of an in-
tramolecular C-1ꢀC-2ꢀC-3 rearrangement was
obtained using
ing ketose, affording
the product. The reaction of
D
-(3-13C)fructose as the start-
methyl)-
Similar co-application of boric acid in the
preparation of -hamamelose from -fructose
L-lyxose was obtained in 12% yield.
D
-(1-13C)hamamelose as
-(3-13C)fructose
D
D
D
with molybdic acid gave a 14:1 ratio of ketose
to branched-chain aldose which was identical
to ratios obtained from similar reactions with
gave a 20% yield of the branched-chain al-
dose. This modification makes the method
more useful from a preparative viewpoint,
since a yield of only ꢀ7% is obtained in the
absence of boric acid.
unlabeled
D
-fructose and
D
-(2-13C)fructose.
1
The 300 MHz H NMR spectrum of the -(1-
D
13C)hamamelose product is shown in Fig. 1.
Molybdic acid-catalyzed isomerization of -
D
1
The anomeric regions of the H spectra of
tagatose under similar reaction conditions
without boric acid led to a 19:1 ratio of the
natural
D
-hamamelose and of
D
-hamamelose
obtained from D-(3-13C)fructose differ consid-
erably. In the latter spectrum, two sets of four
signals due to anomeric protons correspond to
ketose to 2-C-(hydroxymethyl)- -xylose (ob-
D
tained by integration of H-6b signals of the a-
and b-pyranose forms of the former and the
anomeric signals of the latter) after 5 h of
reaction. In contrast, the molybdic acid-cata-
the four forms of
D
-(1-13C)hamamelose in
aqueous solution, namely, a-furanose (5.25

42
Z. Hrico6´ınio6a´ -B´ıliko6a´ et al. / Carbohydrate Research 319 (1999) 38–46
Fig. 1. The 300 MHz 1H NMR spectrum of
natural abundance compound. Four signals between 4.76 and 5.25 ppm are due to anomeric protons, namely, a-furanose (5.25
ppm, ꢀ), b-furanose (5.18 ppm, ), b-pyranose (5.09 ppm, ") and a-pyranose (4.76 ppm, +). (B) Spectrum of -hamamelose
D-hamamelose in aqueous solution at 40 °C. (A) Conventional spectrum of the
D
ꢀ
obtained from the reaction of
D
-(3-13C)fructose with a catalytic amount of molybdic acid. Signals of the anomeric protons are
split into doublets due to the presence of 13C at C-1 of the product branched-chain aldose. The magnitudes of the one-bond
proton–carbon coupling constants between H-1 and 13C-1 vary according to anomeric configuration and cyclic form; in furanose
1
forms, 171.3 Hz (a anomer) and 173.9 Hz (b anomer), and in pyranose forms, 170.0 Hz (b anomer) and 162.6 Hz (a anomer).
13C)fructose and the product branched-chain
ppm), b-furanose (5.18 ppm), b-pyranose (5.09
aldoses are shown in Fig. 2. From
13C)fructose (Fig. 2A), -(2-13C)hamamelose
was formed (Fig. 2C), whereas
-(3-13C)-
fructose (Fig. 2B) yielded
-(1-13C)hama-
melose (Fig. 2D). Thus, in the former rear-
rangement the carbonyl carbon of -fructose
was converted to the quarternary, branching
carbon in -hamamelose, whereas in the latter
the C-3 atom of
-(3-13C)fructose became C-1
D
-(2-
ppm) and a-pyranose (4.76 ppm). Each
D
anomeric proton is coupled by 160–175 Hz to
13
D
the enriched C nucleus at C-1 in the synthe-
D
sized
D
-hamamelose. The magnitudes of these
1JC1ꢀH1 values depend on both anomeric
configuration and ring form. In furanose
forms, 1JC1ꢀH1 values were 171.3 Hz (a anomer)
and 173.9 Hz (b anomer), whereas in the
pyranoses 170.0 Hz (b anomer) and 162.6 Hz
D
D
D
in
-(1-13C)hamamelose, as concluded from 1H
D
(a anomer) were observed1. Thus, the C-sub-
stituted C-3 in the postulated
13
NMR data. In addition, Fig. 2 shows 13C
NMR spectra (Fig. 2E,F) of a secondary
D
-(3-13C)-
fructose tetradentate dimolybdate complex 1
product generated during the reaction of
fructose with molybdic acid, namely, -sor-
bose. Although obtained in low yield,
-sorbose is also formed during the conversion
of -fructose to -hamamelose. However, as
observed in the 13C spectra, the formation of
-sorbose is not accompanied by carbon-skele-
ton rearrangement;
-(2-13C)fructose generates
-(2-13C)sorbose (Fig. 2A,E) and -(3-13C)-
fructose generates
-(3-13C)sorbose (Fig.
D
-
(Scheme 1) is transposed to C-1 in the
D
-(1-
13C)hamamelose product. We postulate the in-
volvement of transient state 2 during the
interconversion, where a new C-2ꢀC-4 bond
forms with concomitant cleavage of the origi-
nal 13C-3ꢀC-4 bond. This rearrangement was
also monitored by 13C NMR spectroscopy. 13C
D
D
D
D
D
D
NMR spectra of
D
-(2-13C)fructose and
D-(3-
D
D
D
1 The pyranose forms of
D
-hamamelose prefer the 1C4
2B,F). Isotope tracing clearly demonstrates the
differences of both processes upon molybdic
conformation, and thus the C-1ꢀH-1 bond is equatorial in the
b anomer [22].

Z. Hrico6´ınio6a´ -B´ıliko6a´ et al. / Carbohydrate Research 319 (1999) 38–46
43
Scheme 1.
acid treatment of isotopically substituted
fructoses (Scheme 2). 13C-3 of
13C)fructose (4) becomes 13C-1 in the primary
product,
-(1-13C)hamamelose (5), while the
secondary product of the transformation,
(3-13C)sorbose (6), remains labeled at C-3. If
-sorbose is formed from -fructose via a
D
-(3-
-
posed previously [8,9,24,25] during the molyb-
date-catalyzed epimerization of aldoses. In this
secondary process, the C-2ꢀC-3 threo-
diastereoisomeric aldoses mutually intercon-
vert without carbon-skeleton rearrangement.
This secondary process is probably accompa-
nied by a mutual exchange of the C-2ꢀH and
C-3ꢀH hydrogen atoms similar to that origi-
nally suggested for the primary process [10].
D
D
D-
D
D
process similar to the Lobry de Bruyn–Al-
berda van Ekenstein isomerization [23], other
isomerization products such as
D
-psicose,
D-
Interestingly, the treatment of
D
-(2-13C; 2-
tagatose and 3-ketoses might also be expected.
However, careful examination of the NMR
spectra of reaction mixtures generated from
2H)xylose with molybdic acid essentially elimi-
nated this secondary process, presumably due
to the deuterium isotope effect retarding pro-
D
-(2-13C)fructose and
D
-(3-13C)fructose did
ton abstraction; in this case, only
D
-(1-13C;
not show the presence of products other than
-(2-13C)hamamelose and -(2-13C)sorbose in
the former case [13] and
-(1-13C)hamamelose
and
-(3-13C)sorbose in the latter case. In
1-2H)lyxose was formed as the product of the
primary process [8].
D
D
D
Of the 2 - ketohexose X 2 - C - (hydroxy-
methyl)aldose interconversions studied herein,
D
addition, total respective recoveries of all the
isotopically substituted sugars after purifica-
tion by chromatography were 98.0 and 97.8%,
respectively.
the most useful preparatively is the
D
-fructose
to -hamamelose interconversion. Since a cat-
D
alytic amount of molybdic acid is used (1:100
when expressed as the mole ratio of dimolyb-
date to sugar), the observed 14:1 ratio of
ketose to branched-chain aldose prob-
These results suggest the existence of a
fructose X -sorbose interconversion analo-
gous to the secondary reaction pathway pro-
D-
D

44
Z. Hrico6´ınio6a´ -B´ıliko6a´ et al. / Carbohydrate Research 319 (1999) 38–46
of interconverting sugars is shifted to those
species forming stronger sugarꢀborate com-
plexes. Preliminary results with calcium hy-
droxide, which is also known to catalyze a
similar carbon-skeleton rearrangement of al-
doses having the threo arrangement at their
C-2ꢀC-3 fragments [28–30], revealed that car-
bon atom transposition may occur when some
ketoses are treated with an equimolar amount
of the hydroxide, yielding 2-C-(hydrox-
ymethyl)aldose products.
4. Conclusions
Molybdic acid catalyzes two types of inter-
conversions among the sugars shown in
-Sorbose2 (7),
D
-fructose (9) and
Fig. 2. The 75.5 MHz 1H-decoupled 13C NMR spectra, in
aqueous solution at 40 °C, of the starting 13C-enriched com-
pounds (A,B) and the synthetic products obtained from iso-
Scheme 3.
D
D-tagatose (11) treated with this catalyst un-
merization reactions (C–F).
-(3-13C)fructose (B); -(2-13C)hamamelose (C) and
13C)sorbose (E) obtained from -(2-13C)fructose;
13C)hamamelose (D) and
-(3-13C)fructose.
D
-(2-13C)Fructose (A);
dergo highly stereospecific carbon-skeleton re-
arrangements (A, B, C) to produce equili-
brium mixtures containing 2-C-(hydroxy-
D
D
D
-(2-
-(1-
D
D
D
-(3-13C)sorbose (F) obtained from
methyl)-
D
-lyxose2 (8), 2-C-(hydroxymethyl)-
-
D
D
ribose (
D-hamamelose, 10) and 2-C-(hydroxy-
methyl)-
D
-xylose (12), respectively. These
ably reflects the thermodynamic equilibrium.
A recently described analogous interconver-
equilibria highly favor the hexuloses; ketose
and branched-chain aldose ratios of 32:1, 14:1
and 19:1, respectively, have been determined
by NMR. Due to the apparently extensive
production of catalytically inactive molybdate
sion of
D-fructose to
D
-hamamelose catalyzed
by Ni(II)ꢀdiamine complexes gave a 56% yield
of the latter sugar in an analytical-scale reac-
tion [26,27]. In this case, however, equimolar
complexes of
D
-psicose (13), interconversion
D starting from 13 was not observed; this
reaction remains to be investigated from
14. Importantly, reaction equilibria for
amounts of -fructose and catalyst were used,
D
and thus the Ni(II)ꢀdiamine-catalyzed process
is not expected to give a true thermodynamic
ratio of the interconverting free sugars but
rather that of their Ni(II) complexes. In this
respect the process is similar to the boric
acid-assisted molybdic acid-catalysis reported
herein, where the thermodynamic equilibrium
2 Although
(hydroxymethyl)-
L
-sorbose and its isomerization product, 2-C-
-lyxose, were studied in this paper, their
L
enantiomers are shown in Scheme 3 in order to indicate the
structural relationships among the interconverting sugars
within one enantiomeric series.
Scheme 2.

Z. Hrico6´ınio6a´ -B´ıliko6a´ et al. / Carbohydrate Research 319 (1999) 38–46
45
Scheme 3.
[3] V. B´ılik, W. Voelter, E. Bayer, Liebigs Ann. Chem., 759
(1972) 189–194.
[4] V. B´ılik, W. Voelter, E. Bayer, Liebigs Ann. Chem.,
(1974) 1162–1166.
[5] V. B´ılik, Chem. Z6esti, 29 (1975) 114–118.
[6] V. B´ılik, Chem. Listy, 77 (1983) 496–505.
[7] S.J. Angyal, G.S. Bethell, R.J. Beveridge, Carbohydr.
Res., 73 (1979) 9–18.
[8] M.L. Hayes, N.J. Pennings, A.S. Serianni, R. Barker, J.
Am. Chem. Soc., 104 (1982) 6764–6769.
[9] E.L. Clark Jr., M.L. Hayes, R. Barker, Carbohydr. Res.,
153 (1986) 263–270.
L
-sorbose and
D
-fructose can be shifted sig-
nificantly towards the 2-C-(hydroxymethyl)-
pentoses when boric acid is included in the
reaction mixture. Moreover,
L-sorbose and
D-
fructose also undergo a competitive secondary
isomerization process to a small extent, appar-
ently catalyzed by molybdic acid, affording
L
-fructose and
D
-sorbose, respectively; this
process (E) is shown in Scheme 3 for the
D
-enantiomers 7 and 9.
The proposed mechanism of the molybdic
[10] V. B´ılik, L. Petrus
690–696.
[11] L. Petrus
Chem. Papers, 50 (1996) 373–374.
[12] Z. Hricov´ıniova´, M. Hricov´ıni, M. Petrus
ulova´, L. Petrus, Chem. Papers, 52 (1998) 238–243.
[13] Z. Hricov´ıniova´, M. Hricov´ıni, L. Petrus, Chem. Papers,
52 (1998) 692–698.
' , V. Farkas' , Chem. Z6esti, 29 (1975)
'
, Z. Hricov´ıniova´, M. Petrus' ova´, M. Matulova´,
acid-catalyzed carbon-skeleton rearrangement
of ketoses to 2-C-(hydroxymethyl)aldoses,
'
ova´, M. Mat-
studied earlier with the use of
fructose, was tested further with the use of
-(3-13C)fructose. The results indicate that the
D
-(2-13C)-
'
'
D
[14] J. Schleucher, M. Sattler, C. Griesinger, Angew. Chem.,
Int. Ed. Engl., 32 (1993) 1489–1491.
[15] D.R. Lide, (Editor), Handbook of Chemistry and Physics,
CRC Press, London 1994, p. 3–464.
[16] R.J. Woods, A.C. Neish, Can. J. Chem., 32 (1954) 404–
414.
rearrangement proceeds in a fashion analo-
gous to the carbon-skeleton rearrangement of
aldoses catalyzed by molybdate ions.
[17] K. Bock, C. Pedersen, Ad6. Carbohydr. Chem. Biochem.,
41 (1983) 27–66.
[18] W.G. Overend, N.R. Williams, J. Chem. Soc., (1965)
3446–3448.
Acknowledgements
This investigation was supported in part by
VEGA Grant No. 2/4144/98 awarded by the
Slovak Academy of Sciences.
[19] M. Matulova´, V. B´ılik, Chem. Papers, 44 (1990) 97–103.
[20] J.P. Sauvage, S. Chapelle, J.F. Verchere, Carbohydr.
Res., 237 (1992) 23–32.
[21] J.P. Sauvage, J.F. Verchere, S. Chapelle, Carbohydr.
Res., 286 (1996) 67–76.
[22] G. Schilling, A. Keller, Liebigs Ann. Chem., (1977) 1475–
1479.
[23] C.A. Lobry de Bruyn, W. Alberda van Ekenstein, Recl.
Tra6. Chim. Pays-Bas, 14 (1895) 156, 203.
References
[24] V. B´ılik, L. Petrus
' , V. Farkas' , Coll. Czech. Chem. Com-
[1] V. B´ılik, Chem. Z6esti, 26 (1972) 183–186; 187–189;
372–375.
mun., 43 (1978) 1163–1166.
[25] V. B´ılik, I. Knezek, K. B´ılikova´, Chem. Papers, 42 (1988)
[2] V. B´ılik, W. Voelter, E. Bayer, Angew. Chem., Int. Ed.
Engl., (1971) 909.
401–405.

46
Z. Hrico6´ınio6a´ -B´ıliko6a´ et al. / Carbohydrate Research 319 (1999) 38–46
[26] R. Yanagihara, S. Osanai, S. Yoshikawa, Chem. Lett.,
(1992) 89–90.
[27] R. Yanagihara, J. Egashira, S. Yoshikawa, S. Osanai,
Bull. Chem. Soc. Jpn, 68 (1995) 237–242.
[28] R. Yanagihara, K. Soeda, S. Shiina, S. Osanai, S.
Yoshikawa, Bull. Chem. Soc. Jpn, 66 (1993) 2268–2272.
[29] S.J. Angyal, Carbohydr. Res., 300 (1997) 279–281.
[30] A. Kuzin, Ber., 69 (1936) 1041–1049.
.