Novel substrates for efficient enzymatic transglycosylation by Bacillus circulans

Article abstract of DOI:10.1139/v02-156

To develop transglycosylation for efficient preparation of N-acetyllactosamine (Galβ 1→4)GlcNAc, LacNAc), β-galactosidase mediated transglycosylation using novel substrates Bacillus Circulans was explored. To make transglycosylation entropically favorable

Full text of DOI:10.1139/v02-156

1174
Novel substrates for efficient enzymatic
transglycosylation by Bacillus circulans
Shiro Komba and Yukishige Ito
Abstract: To develop transglycosylation for efficient preparation of N-acetyllactosamine (Galꢀ(1J4)GlcNAc,
LacNAc), ꢀ-galactosidase mediated transglycosylation using novel substrates Bacillus Circulans was explored. To
make transglycosylation entropically favorable over hydrolysis, donor (lactose or galactose) and acceptor (N-acetyl-
glucosamine, GlcNAc) components were connected to a single molecule. For that purpose, 1,2- and 1,3-benzene-
dimethanol and 2-hydroxy-5-nitro- and 5-hydroxy-2-nitro-benzyl alcohol were screened as linkers and enzymatic
transglycosylation was examined under several conditions. In the case of 2-hydroxy-5-nitro-benzyl connected substrate
40, an indication of the occurrence of intramolecular transglycosylation was observed, and the desired product (58) was
obtained in 26% isolated yield. The same reaction in the presence of CMP sialic acid and ꢁ-(2J6)-sialyltransferase
gave sialyl LacNAc 87 in one pot in 39% isolated yield. Additionally, the effect of the C-2 substituent of the acceptor
component was briefly examined using substrates containing NHAlloc (72), NHTroc (73), and N3 (74) groups. Al-
though the occurrence of intramolecular transglycosylation was not clear in these cases, disaccharides 81–83 were
obtained in reasonable yields.
Key words: galactosidase, intramolecular transglycosylation, N-acetyllactosamine, sialyltransferase.
Résumé : Faisant appel à de nouveaux substrats Bacillus circulan, on a étudié la réaction de transglycolisation
catalysée par une ꢀ-galactosidase dans le but de développer une méthode de transglycosylation permettant de préparer
la N-acétyllactosamine (Galꢀ(1J4)GlcNAc, LacNAc) d’une façon efficace. Afin de rendre la réaction de
transglycosylation entropiquement favorable et de la rendre plus favorable que l’hydrolyse, on a relié les blocs donneur
(lactose ou galactose) et accepteur (N-acétylglucosamine, GlcNAc) dans une seule molécule. À cette fin, on a utilisé les
benzène-1,2- et 1,3-diméthanols ainsi que les alcools 2-hydroxy-5-nitro et 5-hydroxy-2-nitrobenzyliques comme
connecteurs et on a étudié les réactions de transglycosylation sous plusieurs conditions. Dans le cas du substrat 40,
faisant appel à l’alcool 2-hydroxy-5-nitrobenzylique comme connecteur, on a observé des indications à l’effet de
l’existence d’une réaction de transglycosylation intramoléculaire et on a obtenu le produit désiré, 58, avec un
rendement de 26% de produit isolé. La même réaction effectuée en présence de CMP de l’acide sialique et d’ꢁ-(2J6)-
sialyltransférase dans une réaction monotope conduit à la formation de la LacNAc sialyée, 87, avec un rendement de
39% en produit isolé. De plus, on a fait une brève étude de l’effet du substituant en C-2 du composant accepteur en
utilisant des substrats portant des groupes NHAlloc (72), NHTroc (73) et N3 (74). Même si les possibilités de réactions
de transglycosylation intramoléculaire ne sont pas claires dans ces cas, on a obtenu les disaccharides 81–83 avec des
rendements raisonnables.
Mots clés : galactosidase, transglycosylation intramoléculaire, N-acétyllactosamine, sialyltransférase.
[Traduit par la Rédaction] Komba and Ito 1185
Introduction
and CD22 (5). It was first isolated from porcine gastric
mucin as a growth factor for Lactobacillus bifidus var.
pennysylvanicus by Tomarelli et al. (6). A number of chemi-
cal approaches toward LacNAc have been reported (7–11).
Most typically, it can be prepared from lactose (Lac) via
lactal using azidonitration, developed by Wong and
Lemieux (12) and Lemieux and Ratcliff (13), as the key trans-
formation.
On the other hand, Wong et al. (14) developed a highly ef-
ficient LacNAc producing system, by using ꢀ-(1J4)-gala-
ctosyltransferase (EC 2.4.1.90), in combination with
multiple enzymes for UDP–galactose recycling. This land-
mark achievement opened the way for glycosyltransferase
catalyzed preparative scale oligosaccharide synthesis (15).
Limitations in terms of the cost, availability, and stability of
the enzymes and sugar nucleotides are still not negligible,
however.
N-Acetyllactosamine (LacNAc) is a widespread compo-
nent of glycoproteins (1), glycolipids (2), and glycosamino-
glycans (3). It is particularly important as a core structure of
ligands for cell adhesion molecules, for instance selectin (4)
Received 3 December 2001. Published on the NRC Research
Press Web site at http://canjchem.nrc.ca on 4 September 2002.
Dedicated to the memory of late Professor Ray Lemieux in
recognition of his greatest contribution to carbohydrate
chemistry.
S. Komba and Y. Ito.1 The Institute of Physical and
Chemical Research (RIKEN), 2–1 Hirosawa, Wako-shi,
Saitama 351–0198, Japan.
1Corresponding author (e-mail: yukito@postman.riken.go.jp).
Can. J. Chem. 80: 1174–1185 (2002)
DOI: 10.1139/V02-156
© 2002 NRC Canada

Komba and Ito
1175
Scheme 1.
The glycosidase-based approach is attractive due to the
following reasons: (i) an inexpensive (e.g., lactose) or easily
obtainable (e.g., aryl ꢀ-D-galactoside) donor can be used; (ii)
compared with glycosyltransferases, this group of enzymes
are much more stable, easier to handle, and generally
cheaper. Zilliken et al. (16) reported the first enzymatic syn-
thesis of LacNAc from lactose and N-acetylglucosamine
(GlcNAc) in the presence of extracts from Lactobacillus
bifidus, although the yield was low from the standpoint of
practical use. More recently, quite extensive studies on the
use of ꢀ-D-galactosidase (EC 3.2.1.23) from Bacillus
circulans have been reported by several groups (17–22).
Although these pioneering works clearly revealed the po-
tential of galactosidase as a valuable tool for the synthesis of
biologically relevant oligosaccharides, attainable yields have
not been satisfactory. In most cases, a large excess of either
glycosyl donor (e.g., lactose) or acceptor (N-acetylgluco-
samine) was used in order that the transglycosylation can
compete with hydrolysis. It would be extremely difficult to
suppress the hydrolysis to an extent that transglycosylation
using an equimolar amount of glycosyl donor–acceptor can
be achieved in a synthetically useful efficiency. One poten-
tial solution would be to perform the glycosyl transfer in an
intramolecular manner by connecting the donor and acceptor
components together into a single molecule (Scheme 1).
Transglycosylation may be favorable over hydrolysis on the
condition that the donor and acceptor portions of the sub-
strate fit simultaneously into the binding pocket of the en-
zyme. In this paper, the first success in finding the substrate
that, in part, fulfills these demands, is reported (23).
1,3-(meta)-benzenedimethanol, and substrates 9 and 10 were
designed. These substrates were synthesized in a straightfor-
ward manner as depicted in Scheme 2. Although rather forc-
ing conditions (75°C, 12 h) were required, the reactions of 2
and 3 with oxazoline 1 (24) selectively gave 4 and 5 in 59
and 74% yield, respectively, without noticeable contamina-
tion with diglycosylated products. Subsequent glycosylation
with lactosyl trichloroacetimidate 6 (25) was effected in the
presence of TMSOTf to afford 7 and 8 in 75 and 79% yield,
respectively. Finally, O-acyl protecting groups were re-
moved by sodium methoxide, and the designed substrates 9
and 10 were obtained in quantitative yield. These com-
pounds were then examined as substrates for enzymatic
transglycosylation using ꢀ-galactosidase from B. circulans,
and the courses of the reactions were monitored by analyti-
cal HPLC (Fig. 1). Based on MALDI-TOF-MS and NMR
analysis, peaks assignable as the desired products 15 and 16
(MALDI-MS m/z: ([M + Na]+, 688) were identified, and
their structures were rigorously confirmed by NMR analysis
of corresponding acetates 17 and 18. The yields of
transglycosylated products 15 and 16 were rather low
(~5%), however. While hydrolysis toward 11 and 12 was the
overwhelmingly major pathway, it was noteworthy that the
formation of bis-disaccharides 13 and 14 was dominant over
15 and 16 throughout the reaction.2 Since the formation of
13 and 14 is possible only by intermolecular pathway, it is
most probable that the intramolecular pathway made only a
minor contribution, if any, to the formation of 15 and 16
(Scheme 3). Their formation would most likely be a result of
sequential intermolecular transglycosylation and degly-
cosylation (i.e., (9, 10)J(13, 14)J(15, 16)). This conclu-
sion was supported by the extremely similar profiles of the
ortho (9) and meta (10) isomers.
Results and discussion
As a tether to connect donor and acceptor components,
our initial attention was focused on (2) 1,2-(ortho)-, and (3)
The above negative results obtained by 9 and 10 implied
that the topological relationship of Gal and GlcNAc residues
2 In addition, products that, based on MALDI-TOF mass, are assignable as multiply galactosylated, were detected. Their structures were not
rigorously confirmed, however (see Figs. 1–5).
© 2002 NRC Canada

1176
Can. J. Chem. Vol. 80, 2002
Scheme 2.
of these compounds does not fit the binding pocket of ꢀ-
galactosidase. We therefore decided to attenuate the distance
between these residues by changing the linker. To that end,
2-hydroxy-5-nitrobenzyl (ortho) (20) and 5-hydroxy-2-
nitrobenzyl (meta) alcohol (21) were chosen, which each
have one less methylene unit compared with 2 and 3.
Because they are unsymmetrical diols, regioisomers can be
produced simply by changing the order of glycosylation
(i.e., 36–37 and 38–39). Additional diversity of the substrate
can be generated by incorporating the galactose residue, in-
stead of lactose, directly on the phenolic oxygen (i.e., 40 and
41). In the latter case, the substrates can be viewed as deriv-
atives of p-nitrophenyl galactoside, a well explored reactive
donor for enzymatic transglycosylation.
group, trichloroacetimidates 27 (28), 30 (25), and 33 (25)
gave highly satisfactory results. To our surprise, addition of
molecular sieves (MS), a common practise in glycosylation
to ensure anhydrous conditions, deteriorated the efficiency
of these particular reactions. Namely, when 22 and 23 were
reacted with 27 in the presence of MS AW-300, the yields
of 28 and 29 were less than 40%, while the same products
were obtained in excellent yields in their absence. Reactions
of 25 and 26 with 30 and 31 were likewise performed to
give 31, 32, 34, and 35 in 85–97% yield. Somewhat unex-
pectedly, in the case of the preparation of 34, a significant
amount of the corresponding ꢁ-isomer was also synthesized
(ꢁ:ꢀ = 1:3.3). The obtained products were then subjected to
deprotection, uneventfully, to give 36–41.
These compounds were examined as substrates for
transglycosylation in a similar manner as that described for 15
and 16. Namely, each substrate (8 mM) was incubated with B.
circulans ꢀ-galactosidase in 50 mM acetate buffer (pH 5.0),
and the reactions were followed by analytical HPLC (Fig. 2).
In the case of substrates carrying lactose as a donor com-
ponent (36–39), severe peak overlaps retarded the interpretation
Preparation of these substrates was achieved in a concise
and divergent manner as depicted in Scheme 4. It com-
menced with a BF3·OEt2-mediated chemoselective glycosyl-
ation of the primary hydroxy groups of 20 and 21, with
either lactosyl (19) (26) or glucosaminyl (24) (27) acetate to
give 22 and 23 or 25 and 26, respectively, in 67–85% yield.
For the glycosylation of the unreactive phenolic hydroxy
© 2002 NRC Canada

Komba and Ito
1177
Fig. 1. Time course of ꢀ-galactosidase catalyzed reactions: (a) compound 9 (8 mM) was incubated with 0.5 U of ꢀ-galactosidase in
2.5 mL of acetate buffer (50 mM, pH 5.0) at 36°C; (b) compound 10 (8 mM) was incubated with 1.2 U of ꢀ-galactosidase under identical
conditions as (a). Aliquots of the mixture were analyzed by HPLC. Peaks were assigned as follows based on MALDI-TOF-MS analysis:
᭿ starting material (9 and 10); ᭺ GlcNAcGlc (11 and 12); ᭹ LacNAcGlc (15 and 16); ٗ LacNAcLac (13 and 14); ᭜ GalLacNAcLac.
Scheme 3.
© 2002 NRC Canada

1178
Can. J. Chem. Vol. 80, 2002
Scheme 4. Reagents and conditions: (a) 2 equiv BF3·OEt2 in CH2Cl2 at rt for 1 h; (b) 0.5 equiv BF3·OEt2 in CH2Cl2 at rt for 1 h;
(c) (i) 75 equiv NH2NH2 in EtOH at 90°C for 3 h, (ii) Ac2O in MeOH at rt for 8 h.
of the HPLC profiles. To enhance the peak assignments,
aliquots of the mixtures were treated with ꢀ-N-acetylglu-
cosaminidase from Jack beans (Sigma Co., Ltd.). After this
treatment, galactosylated products, including the desired
products (54–57) and bis-disaccharides (48–51), were
readily distinguished from substrates and hydrolyzed prod-
ucts (42–45) by their resistance. In any event, no indication
of intramolecular glycosyl transfer was obtained from these
substrates; that is, at the early stage of incubation, the
,
formation of 48, 49, 50, and 51 was dominant over 54, 55
© 2002 NRC Canada

Komba and Ito
1179
Fig. 2. Time course of ꢀ-galactosidase catalyzed reactions using substrates derived from the nitro-containing linker (Scheme 4).
(a) Compound 36 (8 mM) was incubated with 0.34 U of ꢀ-galactosidase in 1.8 mL of acetate buffer (50 mM, pH 5.0) at 36°C;
(b) compound 37 (8 mM) was incubated with 0.34 U of ꢀ-galactosidase, under identical conditions as (a) aliquots of the mixture were
analyzed by HPLC. Peaks were assigned as follows based on MALDI-TOF-MS analysis: ᭿ starting material (36 and 37); ᭺ GlcNAcGlc
(42 and 43); ᭹ LacNAcGlc (54 and 55); ٗ LacNAcLac (48 and 49); × GlcNAcGalLac (regioisomer of 48 and 49, structure not con-
firmed); ᭜ GlcNAcGal2Lac (structure not confirmed); (c) compound 38 (8 mM) was incubated with 0.34 U of ꢀ-galactosidase in
1.8 mL of acetate buffer (50 mM, pH 5.0) at 36°C; (d) compound 39 (8 mM) was incubated with 0.10 U of ꢀ-galactosidase, under
identical conditions as (c). Aliquots of the mixture were analyzed by HPLC. Peaks were assigned as follows based on MALDI-TOF-
MS analysis: ᭿ starting material (38 and 39); ᭺ GlcNAcGlc (44 and 45); ᭹ LacNAcGlc (56 and 57); ٗ LacNAcLac (50 and 51);
× GlcNAcGalLac (regioisomer of 50 and 51, structure not confirmed); ᭜ GlcNAcGal2Lac (structure not confirmed).; (e) compound 40
(8 mM) was incubated with 0.13 U of ꢀ-galactosidase in 0.70 mL of acetate buffer (50 m M, pH 5.0) at 36°C; ( f ) compound 41
(8 mM) was incubated with 0.44 U of ꢀ-galactosidase under idential conditions as in (e). Aliquots of the mixture were analyzed by
HPLC. Peaks were assigned as follows based on MALDI-TOF-MS analysis: ᭿ starting material (40 and 41); ᭺ GlcNAc (46 and 47);
᭹ LacNAc (58 and 59); ٗ LacNAcGal (52 and 53); × Gal2GlcNAc (regioisomer of 52 and 53, structure not confirmed); ᭜
Gal3GlcNAc (structure not confirmed).
56, and 57.2 Nevertheless, it was observed that a significant
amount of the transglycosylated product (56) was formed
from ortho substrate 38.
On the other hand, the profile of substrate 40 was mark-
edly different. In this particular case, the formation of the
LacNAc derivative (closed circle) was dominant over that of
52 even at the early stage of incubation ([40] > [46]).2 Since
the rates of intermolecular glycosyl transfer are likely to be
similar between 40 and 46 (k1 W k2), it was inferred that the
intramolecular pathway is functional (k3 > k1, k2) (Scheme 5).
By contrast, when starting from the meta oriented substrate
41, it was found that the transglycosylation was far less efficient.
© 2002 NRC Canada

1180
Can. J. Chem. Vol. 80, 2002
Scheme 5.
performed at 20 mM concentration in acetate buffer (entry 3).
Thus, using 50 mg (94 mol) of compound 40, incubation
with ꢀ-galactosidase (430 mU) for 3 h afforded 26% yield
(13.1 mg) of 58, the structure of which was rigorously con-
firmed by NMR analysis of the corresponding acetate 70.
Subsequently, the effect of the glucosamine C-2 substituent
was examined. Considering the potential synthetic utility of
the expected products, three types of substrates were pre-
pared, namely N-allyloxycarbonyl (Alloc) protected 72,
The formation of intermolecularly glycosylated product 53
preceded that of 59, which means that the direct pathway
was insignificant.
Using the preferred substrate 40, the effects of substrate
concentration (i.e., Table 1, entries 1ꢂ4), organic solvents
(entries 6–12), and buffer pH (entry 5) were briefly examined.
Contrary to several precedents on enzymatic transglycosylation
(17, 20, 21), the beneficial effect of organic solvent was not
observed (entries 6–12). A preparative scale reaction was
© 2002 NRC Canada

Komba and Ito
1181
Fig. 3. Time course of ꢀ-galactosidase catalyzed reactions using C-2 modified substrates (Scheme 7); (a) compound 72 (20 mM) was
incubated with 86 mU of ꢀ-galactosidase in 86.5 L of acetate buffer (50 mM, pH 5.0) at 36°C; (b) compound 73 (20 mM) was incu-
bated with 86 mU of ꢀ-galactosidase under identical conditions as (a); (c) compound 74 (10 mM) was incubated with 172 mU of ꢀ-
galactosidase under identical conditions as (a). Aliquots of the mixture were analyzed by HPLC. Peaks were assigned as follows based
on MALDI-TOF-MS analysis: ᭿ starting material (72 and 73 and 74); ᭺ GlcNR (75 and 76 and 77); ᭹ LacNR (81 and 82 and 83);
ٗ LacNRGal (78 and 79 and 80); × Gal2GlcNR (regioisomer of 78 and 79 and 80, structure not confirmed); ᭜ Gal3GlcNR (structure
not confirmed).
molecularly transferred products (78–80);2 nonetheless,
Table 1. Effects of solvent and substrate concentration.
carbamate-protected lactosamine derivatives 81 and 82 were
40
(mM)
Enzyme
(mU)
Solvent
(mL)a
Time
(h)
Yield
(%)
formed with an efficiency comparable with 58. These prod-
ucts were quite resistant to galactosidase mediated hydroly-
sis. Preparative scale reactions using 72 (17.3 mol), 73 (15
mol), and 74 (9.6 mol) were performed, in the same
manner as described for 40, to give 81, 82, and 83 in 21, 21,
and 11%, respectively (Scheme 8), the structures of which
were rigorously confirmed by NMR analysis of the corre-
sponding acetates (84, 85, and 86). Since N-Alloc- (30) and
N-Troc-protected derivatives (31) and C-2 azide (13, 32) are
useful as glycosyl donors for the ꢀ- and ꢁ-selective gly-
cosylation of 2-amino sugars, respectively, this procedure
may be useful for the chemoenzymatic synthesis of complex
oligosaccharides.
Entry
1
2
3
4
5
6
7
8
9
0.8
8
20
20
20
8
8
8
8
8
43
86
430
430
86
130
130
130
130
43
4b
15b
26c
18b
15c
7b
14b
14b
2b
10b
13b
1b
A (2.3)
A (0.23)
A (4.7)
48
8
3
A (4.7)
4
B (0.09)
C (0.23)
C (0.23)
C (0.23)
D (0.23)
E (0.23)
E (0.23)
E (0.23)
4.5
24
72
96
24
8
10
11
12
8
8
43
43
24
72
We then turned our attention to the synthesis of sialyl
LacNAc by using galactosidase mediated transglycosylation,
in combination with sialyltransferase. Since it is obvious that
the LacNAc product 58 is susceptible to further digestion by
ꢀ-galactosidase (to provide 46), its yield under standard con-
ditions does not reflect the relative rates of intramolecular
transglycosylation and hydrolysis (i.e., k3 vs. ka). In fact, the
yield of 58 (26%) dropped to 18% after 4 h incubation (Ta-
ble 1, entries 3 and 4). To minimize the extent of post-trans-
glycosylation hydrolysis, and to gain a more accurate
estimation of the efficiency of transglycosylation, ꢁ-(2J6)-
sialyltransferase (rat, recombinant, Spodoptera frugiperda)
(Calbiochem Co., Ltd.) (33, 34) and CMP sialic acid were
added to the system to trap 58 as a galactosidase resistant
product 87 (Scheme 9, Figs. 4 and 5). As expected, smooth
formation of sialyl-ꢁ-(2J6)-LacNAc 87 (ٗ) was observed,
and its amount did not decrease after prolonged incubation.
In a preparative scale run, substrate 40 was used (25 mg, 47 mol),
incubated in 200 mM HEPES buffer (2.4 mL, pH 7.5) in the
presence of ꢀ-galactosidase (100 mU), ꢁ-(2J6)-
sialyltransferase (60 mU), and CMP-sialic acid (30 mg,
a(A) 50 mM acetate buffer, pH 5.0; (B) 200 mM HEPES, pH 7.5; (C)
50% acetonitrile in A; (D) 60% triethyl phosphate in A; (E) 20% 2-
ethoxyethyl ether in A.
bBased on HPLC relative peak integrations, quantified at 245 nm.
cIsolated yield.
N-trichloroethoxycarbonyl (Troc) protected 73, and C-2
azide 74. To synthesize these substrates, compound 34 was
used as the common intermediate. Thus, 34 was treated with
hydrazine hydrate in EtOH to remove N-phthalimido and O-
acyl groups. The resultant amine was treated with either
allyl chloroformate – NaHCO3, trichloroethyl chloroformate
– NaHCO3, or TfN3–Cu2SO4–K2CO3 (29), to afford 72, 73,
and 74 in 64, 80, and 53% yield, respectively (Scheme 6).
Enzymatic transglycosylations were examined under con-
ditions optimized for 40 (20 mM substrates in 50 mM ace-
tate buffer, pH 5.0, Fig. 3 (Scheme 7). In these cases, the
dominancy of intramolecular transglycosylation was not as
clear as 40, because the formation of the expected products
(81–83) was slightly behind the initial formation of inter-
© 2002 NRC Canada

1182
Can. J. Chem. Vol. 80, 2002
Scheme 6.
Scheme 7.
46 mol). After 17 h, successive purification by Sephadex
LH-20 and preparative HPLC afforded sialyl-ꢁ-(2J6)-
LacNAc 87 in 39% yield (15.4 mg).
In conclusion, a novel method for the preparation of
LacNAc and sialyl LacNAc was developed using ꢀ-
galactosidase mediated intramolecular transglycosylation,
© 2002 NRC Canada

Komba and Ito
1183
Scheme 8.
Scheme 9.
using substrates that have glycosyl donor and aglycon por-
tions connected by a tether.
benzyl alcohol. Optical rotations were measured with a DIP-
370 digital polarimeter (Japan spectroscopic Co., Ltd.) at
22°C. H NMR spectra were recorded at 400 MHz (JEOL
1
JNM-AL400), or at 500 MHz (JEOL JNM-ECP500), in
CDCl3 (internal Me4Si, ꢃ: 0.00), DMSO-d6, CD3OD, or D2O.
13C NMR spectra were recorded at 100 or 125 MHz, respec-
tively, on the same instruments in CDCl3, DMSO-d6,
CD3OD, or D2O (internal MeOH, ꢃ: 49.0). ꢀ-Galactosidase
from Bacillus circulans was purchased from Daiwa Kasei
K.K. (Japan). ꢁ-(2J6)-Sialyltransferase from rat, recombi-
nant, Spodoptera frugiperda was purchased from
Calbiochem Co., Ltd. ꢀ-N-Acetylglucosaminidase from Jack
beans was purchased from Sigma Co., Ltd. All solvents and
reagents were reagent grade and were used without further
purification.
Experimental3
General methods
Column chromatography was performed on Silica Gel
60N (spherical, neutral) (Kanto chemical Co., Inc.). Prepara-
tive reverse-phase HPLC was performed on a Waters HPLC
system (Millennium32 operation system), using a Mightysil
RP-18 GP 250–20 (5 m) (Kanto chemical Co., Inc.), with a
flow rate of 10 mL min–1, and detected at 256 nm with a
photodiode array detector (Waters 996). The solvent systems
were: (a) TFA–water (1:1000) and (b) TFA–acetonitrile–wa-
ter (1:900:100). Analytical reverse phase HPLC separations
were performed on a Waters HPLC system, using a
Mightysil RP-18 GP 250–3.0 (5 m) (Kanto chemical Co.,
Inc.), with a flow rate of 0.6 mL min–1, and detected at
256 nm with the same detector as above, using the same
solvent system as above. MALDI-TOF-MS was performed
in the positive ion mode on a Kompact MALDI IV tDE
(Shimadzu Co., Ltd.), using a matrix of 2,5-dihydroxy-
benzoic acid or ꢁ-cyano-4-hydroxycinnamic acid, and an an-
giotensin I (MW = 1296.5) as a standard. FAB-MS was
performed in the positive ion mode on a JMS-HX110
(JEOL), using a matrix mixture of glycerol and 3-nitro-
Enzymatic transglycosylation: typical procedure 1
A solution of substrate 9 (13.7 mg, 20.5 mol, 8 mM) in
50 mM sodium acetate buffer (2.5 mL, pH 5.0) was incu-
bated with ꢀ-D-galactosidase (0.5 U) at 36°C. After 30 min,
1 h, 4.5 h, and 8 h, aliquots of the reaction mixture were
taken (ca. 200 L) and terminated by heating at 95°C for
10 min. Each of them was separated by HPLC, first using
90% solvent A and 10% solvent B for 10 min, then the lin-
ear gradient 10ꢂ100% solvent B for 170 min. The eluted
peaks were detected 254 nm UV absorption and the
3 Experimental details for preparation of substrates were provided as Supplementary Material. Supplementary data may be purchased from the
Depository of Unpublished Data, Document Delivery, CISTI, National Research Council Canada, Ottawa, ON K1A 0S2, Canada
(http://www.nrc.ca/cisti/irm/unpub_e.shtml for information on ordering electronically).
© 2002 NRC Canada

1184
Can. J. Chem. Vol. 80, 2002
Fig. 4. HPLC profiles of ꢀ-galactosidase – ꢁ-sialyltransferase re-
action of 40: (a) 1.5 h, (b) 6.5 h. Retention times of starting ma-
terial (40), Gal2GlcNAc (52), NeuAcGalGlcNAc (87), and
GlcNAc (46) are 17.87, 18.34, 22.25, and 22.53 min, respectively.
Enzymatic transglycosylation: typical procedure 2
A solution of substrate 36 (10 mg, 14.3 mol, 8 mM) in
50 mM sodium acetate buffer (1.8 mL, pH 5.0) was added to
ꢀ-D-galactosidase (0.34 U) and incubated at 36°C. After
30 min, 1.5 h, 7 h, and 24 h, aliquots of the reaction mixture
were taken and terminated by heating at 95°C for 10 min,
and part of them were analyzed by analytical HPLC, first us-
ing 100% solvent A for 5 min, then the linear gradient 0–
30% solvent B for 30 min.
Residual aliquots (180 L) were treated with ꢀ-N-
acetylglucosaminidase (62.5 mU) at 36°C for 2 days, then
the reaction mixtures were terminated by heating at 95°C for
10 min. Each of them was analyzed by HPLC, 90% solvent
A and 10% solvent B as isocratic mode. The eluted peaks
were detected at 254 nm using UV absorption and the com-
position was calculated by integral curves.
Combined transglycosylation–sialylation: preparation of
compound 87
To a solution of 40 (25 mg, 46.7 mol, 20 mM) and CMP-
Neu5Ac (30.8 mg, 46.7 mol) in 200 mM HEPES buffer
(2.3 mL, pH 7.5) containing 20 mM MgCl2, 5.3 mM MnCl2,
and 20 mM KCl, was added ꢁ-(2J6)-sialyltransferase (60
mU), followed by ꢀ-D-galactosidase (2.2 U). The mixture
was incubated at 36°C for 18 h, and terminated by heating at
95°C for 10 min. After concentration, the mixture was
roughly purified by Sephadex LH-20 gel filtration column
chromatography (H2O). Fractions containing 87 were con-
centrated and the mixture was purified by preparative HPLC,
first using 100% solvent A for 5 min, then the linear gradi-
ent 0–30% solvent B for 30 min to give sialylated compound
87 (15.4 mg, 39%), together with hydrolyzed compound 46
1
(10.8 mg, 61%). Compound 87: H NMR (500 MHz, D2O
composition was calculated by integral curves. The fractions
containing products were analyzed by MALDI-TOF-MS and
each product was acetylated for extensive 1H NMR analysis.
and MeOH) ꢂ: 8.06 (d, 1H, JPh4,Ph5 = 3.2 Hz, Ph6), 8.01 (dd,
1H, JPh3,Ph4 = 9.2 Hz, JPh4,Ph6 = 3.2, Ph4), 6.49 (d, 1H,
JPh3,Ph4 = 9.2 Hz, Ph3), 4.75 (d, 1H, Jgem = 12.8 Hz, PhCH),
Fig. 5. Time course of ꢀ-galactosidase – ꢁ-sialyltransferase reaction of 40. Compound 40 (20 mM) was incubated with 86 mU of
ꢀ-galactosidase and ꢁ-(2J6)-sialyltransferase (2.4 mU) in 95 L of HEPES (200 mM, pH 7.5), 20 mM MgCl2, 5.3 mM MnCl2,
20 mM KCl, and CMP-sialic acid (20 mM), at 36°C. Peaks were assigned as follows based on MALDI-TOF-MS analysis: ᭜ starting
material (40); × GlcNAc (46); ٗ NeuAcGalGlcNAc (87); ᭿ Gal2GlcNAc (52).
© 2002 NRC Canada

Komba and Ito
1185
14. C.H. Wong, S.L. Haynie, and G.M. Whitesides. J. Org. Chem.
47, 5416 (1982).
15. C.H. Wong, R.L. Halcomb, Y. Ichikawa, and T. Kajimoto.
Angew. Chem. Int. Ed. Engl. 34, 412 (1995).
16. F. Zilliken, P.N. Smith, C.S. Rose, and P. György. J. Biol.
Chem. 217, 79 (1955).
17. (a) K. Sakai, R. Katsumi, H. Ohi, T. Usui, and Y. Ishido. J.
Carbohydr. Chem. 11, 553 (1992); (b) T. Usui, S. Morimoto,
Y. Hayakawa, M. Kawaguchi, T. Murata, Y. Matahira, and Y.
Nishida. Carbohydr. Res. 285, 29 (1996).
18. E. Farkas and J. Thiem. Eur. J. Org. Chem. 3073 (1999).
19. (a) S. Takayama, M. Shimazaki, L. Qiao, and C.H. Wong.
Bioorg. Med. Chem. Lett. 6, 1123 (1996); (b) T. Kitamura, S.
Takayama, H. Huang, and C.H. Wong. Angew. Chem. Int. Ed.
Engl. 35, 2348 (1996).
20. J.H. Yoon and J.S. Rhee. Carbohydr. Res. 327, 377 (2000).
21. K. Priya and D. Loganathan. Tetrahedron, 55, 1119 (1999).
22. J. Fang, W. Xie, J. Li, and P.G. Wang. Tetrahedron Lett. 39,
919 (1998).
4.63 (d, 1H, Jgem = 13.3 Hz, PhCH>), 4.57 (d, 1H, J1,2
8.2 Hz, H-1Gal), 4.40 (d, 1H, J1,2 = 7.8 Hz, H-1GlcN), 2.64
(dd, 1H, J3eq,4 = 4.1 Hz, Jgem = 11.9 Hz, H-3eq sialic acid),
=
2.00 and 1.95 (2s, 6H, 2NAc), 1.66 (dd, 1H, J3ax,4 = Jgem
=
11.9 Hz, H-3ax sialic acid). 13C NMR (150 MHz, D2O,
NaOD, and MeOH) ꢃ: 176.4 (CPh2), 175.1 (CNAc=O GlcN),
174.8 (CNAc=O sialic acid), 173.7 (C-1 sialic acid), 133.1
(CPh5), 127.6 (CPh4-H), 127.2 (CPh6-H), 126.5 (CPh1), 119.6
(CPh3-H), 103.8 (C-1 Gal), 100.5 (C-1 GlcN), 100.4 (C-2
sialic acid), 80.2 (C-4 GlcN), 74.8 (C-5 GlcN), 74.0 (C-5
Gal), 73.1 (C-3 Gal), 72.7 (C-3 GlcN), 72.7 (C-6 sialic
acid), 71.9 (C-8 sialic acid), 70.8 (C-2 Gal), 68.9 (C-4 Gal),
68.6 (C-7 sialic acid), 68.4 (C-4 sialic acid), 67.0 (CBn-H2),
63.6 (C-6 Gal), 62.8 (C-9 sialic acid), 60.7 (C-6 GlcN), 55.4
(C-2 GlcN), 52.2 (C-5 sialic acid), 40.5 (C-3 sialic acid),
22.3 and 22.2 (2CNAc-H3). MALDI-MS m/z: 848 ([M + Na]+).
Acknowledgments
23. A part of this work was published in a preliminary form: S.
Komba and Y. Ito. Tetrahedron Lett. 42, 8501 (2001).
24. C. Auge and A. Veyrierés. Carbohydr. Res. 46, 293 (1976).
25. R.R. Schmidt and J. Michel. Angew. Chem. Int. Ed. Engl. 19,
731 (1980).
This work was supported by the Special Postdoctoral Re-
searchers Program at RIKEN. We thank Dr. H. Koshino and
his staff for NMR measurements and Ms. A. Takahashi for
technical assistance.
26. C.S. Hudson and J.M. Johnson. J. Am. Chem. Soc. 37, 1270
(1915).
27. A. Medgyes, E. Farkas, A. Liptak, and V. Pozsgay. Tetrahe-
dron, 53, 4159 (1997).
28. G. Grundler and R.R. Schmidt. Carbohydr. Res. 135, 203 (1985).
29. (a) J.T. Lundquist, IV and J.C. Pelletier. Org. Lett. 3, 781
(2001); (b) P.B. Alper, S.H. Hung, and C.H. Wong. Tetrahe-
dron Lett. 37, 6029 (1996).
References
1. M. Fukuda. Biochim. Biophys. Acta, 780, 119 (1985).
2. T. Feize, H.C. Gooi, R.A. Childs, J.K. Pickard, K. Uemura,
L.M. Koomes, S.J. Thorpe, and E.F. Hounsell. Biochim. Soc.
Trans, 12, 591 (1984).
3. R.V. Iozzo and A.D. Murdoch. FASEB J. 10, 598 (1996).
4. R.K. Jain, B.G. Huang, E.V. Chandrasekaran, and K.L. Matta.
Chem. Commun. 23 (1997).
30. A. Vargas-Berenguel, M. Meldal, H. Paulsen, and K. Bock. J.
Chem. Soc. Perkin Trans. 1, 2615 (1994).
31. (a) J. Rademann, A. Geyer, and R.R. Schmidt. Angew. Chem.
Int. Ed. Engl. 37, 1241 (1998); (b) N. Sawada, M. Ito, H.
Ishida, and M. Kiso. Tetrahedron Lett. 42, 1745 (2001).
32. W. Kinzy and R.R. Schmidt. Justus Liebigs Ann. Chem. 1537
(1985).
33. For the previous work on the combined use of ꢀ-galactosidase
and sialyltransferase. G.F. Herrmann, Y. Ichikawa, C.
Wandrey, F.C.A. Gaeta, J.C. Paulson, and C.H. Wong. Tetrahe-
dron Lett. 34, 3091 (1993).
34. (a) J. Weinstein, U. de Sausa-e-Silva, and J.C. Paulson. J. Biol.
Chem. 257, 13 835 (1982); (b) M.A. Williams, H. Kitagawa,
A.K. Datta, J.C. Paulson, and J.C. Jamieson. Glycoconjugate J.
12, 755 (1995); (c) K.B. Wlasichuk, M.A. Kashem, P.V.
Nikrad, P. Bird, C. Jiang, and A.P. Venot. J. Biol. Chem. 268,
13 971 (1993).
5. T.M. Wrodnigg and A.E. Stutz. Angew. Chem. Int. Ed. 38, 827
(1999).
6. R.J. Tomarelli, J.B. Hassinen, E.R. Eckhardt, R.H. Clark, and
F.W. Bernhart. Arch. Biochem. Biophys. 48, 225 (1954).
7. A.A. Sherman, O.N. Yudina, A.S. Shashkov, V.M. Menshov,
and N.E. Nifant’ev. Carbohydr. Res. 330, 445 (2001).
8. S. Figueroa-Perez and V. Verez-Bencomo. Carbohydr. Res.
317, 29 (1999).
9. E. Lattova and L. Petrus. Carbohydr. Res. 235, 289 (1992).
10. T. Ogawa and S. Nakabayashi. Carbohydr. Res. 97, 81 (1981).
11. J. Alais and A. Veyrieres. Carbohydr. Res. 207, 11 (1990).
12. T.C. Wong and R.U. Lemieux. Can. J. Chem. 62, 1207 (1984).
13. R.U. Lemieux and R.M. Ratcliffe. Can. J. Chem. 57, 1244
(1979).
© 2002 NRC Canada