Syntheses of TN building blocks Nα-(9- fluorenylmethoxycarbonyl)-O-(3,4,6-tri-O-acetyl-2-azido-2-deoxy-α-D- galactopyranosyl)-L-serine/L-threonine pentafluorophenyl esters: Comparison of protocols and elucidation of side reactions

Article abstract of DOI:10.1016/j.carres.2005.02.029

T_N antigen building blocks N^α-(9- fluorenylmethoxycarbonyl)-O-(3,4,6-tri-O-acetyl-2-azido-2-deoxy-α-d- galactopyranosyl)-l-serine/l-threonine pentafluorophenyl ester [Fmoc-l-Ser/l-Thr(Ac₃-α-d-GalN₃)-OPfp, 13/14]

Full text of DOI:10.1016/j.carres.2005.02.029

Carbohydrate
RESEARCH
Carbohydrate Research 340 (2005) 1273–1285
Syntheses of TN building blocks Na-(9-fluorenylmethoxycarbonyl)-O-
(3,4,6-tri-O-acetyl-2-azido-2-deoxy-a-D-galactopyranosyl)-L-
serine/L-threonine pentafluorophenyl esters: comparison of
protocols and elucidation of side reactionsq
Mian Liu,a,b Victor G. Young, Jr.,b, Sachin Lohani,c, David Livea,* and George Baranyb,*
aDepartment of Biochemistry, Molecular Biology and Biophysics, University of Minnesota, Minneapolis, MN 55455, USA
bDepartment of Chemistry, University of Minnesota, Minneapolis, MN 55455, USA
cDepartment of Pharmaceutics, University of Minnesota, Minneapolis, MN 55455, USA
Received 15 December 2004; accepted 22 February 2005
Available online 11 April 2005
Abstract—TN antigen building blocks Na-(9-fluorenylmethoxycarbonyl)-O-(3,4,6-tri-O-acetyl-2-azido-2-deoxy-a-D-galactopyranos-
yl)-L-serine/L-threonine pentafluorophenyl ester [Fmoc-L-Ser/L-Thr(Ac3-a-D-GalN3)-OPfp, 13/14] have been synthesized by two dif-
ferent routes, which have been compared. Overall isolated yields [three or four chemical steps, and minimal intermediary
purification steps] of enantiopure 13 and 14 were 5–18% and 6–10%, respectively, based on 3,4,6-tri-O-acetyl-D-galactal (1). A
byproduct of the initial azidonitration reaction of the synthetic sequence, that is, N-acetyl-3,4,6-tri-O-acetyl-2-azido-2-deoxy-a-D-
galactopyranosylamine (5), has been characterized by X-ray crystallography, and shown by 1H NMR spectroscopy to form
complexes with lithium bromide, lithium iodide, or sodium iodide in acetonitrile-d3. Intermediates 3,4,6-tri-O-acetyl-2-azido-
2-deoxy-a-D-galactopyranosyl bromide (6) and 3,4,6-tri-O-acetyl-2-azido-2-deoxy-b-D-galactopyranosyl chloride (7) were used to
glycosylate Na-(9-fluorenylmethoxycarbonyl)-L-serine/L-threonine pentafluorophenyl esters [Fmoc-L-Ser/L-Thr-OPfp, 11/12]. Previ-
ously undescribed low-level dehydration side reactions were observed at this stage; the unwanted byproducts were easily removed by
column chromatography.
ꢀ 2005 Elsevier Ltd. All rights reserved.
Keywords: Glycopeptide synthetic building blocks; Glycosylation; Side reactions; Metal ion–carbohydrate complexes; Dehydroalanine;
Z-Dehydrothreonine
1. Introduction
for several reasons, including the microheterogeneity
found naturally in their glycosylation.1 Chemical syn-
thesis has provided an alternative way to access well-de-
fined glycopeptide structures that can be used to probe
mucin glycoprotein properties.2 This has been carried
out either in solution,3,4 or (preferably for larger frag-
ments) on solid supports.5 At present, the most efficient
and versatile method for preparation of mucin-related
O-linked glycopeptides, in particular those bearing the
TN antigens, uses protected glycosylated amino acids
as building blocks in stepwise assembly.6 Facile and effi-
cient procedures to prepare appropriately protected
amino acid building blocks are therefore important to
the practice of solid-phase glycopeptide synthesis. In
In recent years, there has been an increased appreciation
for the biological roles of mucin glycoproteins and gly-
coprotein domains. Quantitative studies of these glyco-
conjugates from natural sources have been challenging
q A preliminary report of portions of this work was presented at the
3rd International/28th European Peptide Symposium, Prague, Czech
Republic, September 5–10, 2004.
*
Corresponding authors. Tel.: +1 612 625 1028; fax: +1 612 626 7541
(G.B.); tel.: +1 612 626 3805; fax: +1 612 624 5121 (D.L.); e-mail
addresses: david@nmr1.biochem.umn.edu; barany@umn.edu
These authors are responsible for the X-ray crystallographic analysis
of compound 5.
0008-6215/$ - see front matter ꢀ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.carres.2005.02.029

1274
M. Liu et al. / Carbohydrate Research 340 (2005) 1273–1285
general, such building blocks are obtained by direct
reaction of an Na-9-fluorenylmethoxycarbonyl (Fmoc)-
protected amino acid pentafluorophenyl (Pfp) ester with
a glycosyl donor. Fmoc is preferred as an Na-amino pro-
tecting group in peptide and glycopeptide synthesis since
it is stable to glycosylation conditions. It can be re-
moved during stepwise assembly under mild conditions
with bases such as morpholine or piperidine, and it is
stable to acidic conditions used in cleavage of glycopep-
tides from the support. Likewise, the Pfp ester serves a
ÔdualÕ function, insofar as it is stable and protects the
Ca-carboxyl under conditions of glycosylation, while it
provides activation of that same building block carboxyl
group when needed for subsequent coupling. Prior syn-
theses of serine and threonine building blocks carrying
galactosamine have been described and summarized.7–12
Here, we report some new findings on the syntheses
of building blocks Na-Fmoc-O-(3,4,6-tri-O-acetyl-2-
azido-2-deoxy-a-D-galactopyranosyl)-L-serine/L-threonine
pentafluorophenyl esters (13/14). By comparing proto-
cols from the literature, those that are efficient and ro-
bust can be identified and recommended. The synthetic
efforts reported here are part of a larger research pro-
gram to probe by NMR spectroscopy the conforma-
tional properties of mucin glycopeptides, and are
motivated in part by the need for optimal procedures
that allow incorporation of isotopes as NMR probes.
2. Results and discussion
2.1. Syntheses of Na-Fmoc-O-(3,4,6-tri-O-acetyl-2-azido-
2-deoxy-a-D-galactopyranosyl)-L-serine/L-threonine pen-
tafluorophenyl ester (13/14)
Two routes were used to synthesize the title building
blocks (Scheme 1). Azidonitration of 3,4,6-tri-O-acetyl-
D-galactal (1) under previously described conditions13
gave 3,4,6-tri-O-acetyl-2-azido-2-deoxy-b-D-galactopyr-
anosyl nitrate (2), as well as the corresponding a-anomer
3 and the corresponding talo-isomer 4 [2:3:4, 1.0:1.25
( 0.20):0.23( 0.03)]. The azidonitration reaction was
monitored by reversed-phase high-performance liquid
chromatography (RP-HPLC), and shown to be complete
within 10 h at ꢀ20 ꢁC, as judged by disappearance of
galactal 1. Also produced under the conditions was
N-acetyl-3,4,6-tri-O-acetyl-2-azido-2-deoxy-a-D-galacto-
pyranosylamine (5) (similar level as 2), as discussed in
Section 2.2. The reaction mixture was transformed
further, without purification of intermediates to
3,4,6- tri-O-acetyl-2-azido-2-deoxy-a-D-galactopyrano-
syl bromide (6) (Route A) or to 3,4,6-tri-O-acetyl-2-
azido-2-deoxy-b-D-galactopyranosyl chloride (7) via
a-iodide 8 (Route B). Either compound 6 or compound
7 was then coupled directly with Na-(9-fluorenylmethoxy-
carbonyl)-L-serine/L-threonine pentafluorophenyl esters
OAc
AcO
AcO
OAc
OAc
AcO
OAc
AcO
OAc
AcO
OAc
N3
OAc
OAc
OAc
O
a
O
O
O
O
+
+
+
ONO2
AcO
N3
N3
N3
ONO2
1
2
3
4
ONO2
5
NHAc
Azidonitration Reaction Products
b
c
OAc
AcO
OAc
OAc
AcO
B
A
OAc
O
OAc
OAc
AcO
OAc
O
OAc
f
O
d
O
Cl
6
8
N3
AcO
N3
O
7
N3
N3
Br
e
g
g
13
Fmoc-Ser-OPfp
I
f
13
14
tBu
OAc
h
OAc
OAc
AcO
OAc
Fmoc-Ser-OPfp
11
O
Fmoc-Ser-OPfp
O
AcO
9
Cl
tBu
N3
N3
O
h
Fmoc-Thr-OPfp
12
Fmoc-Thr-OPfp
10
7
Fmoc-Thr-OPfp
14
Scheme 1. Synthetic routes to building blocks Fmoc-L-Ser/L-Thr(Ac3-a-D-GalN3)-OPfp (13/14). Reagents and conditions: (a) CAN, NaN3, CH3CN,
ꢀ20 ꢁC, 10 h, 42% [following Ref. 13]; (b) LiBr, CH3CN, 8 h, 82% [following Ref. 13]; (c) LiI, CH3CN, 20 min, 48% [following Ref. 13]; (d) TEAC,
CH3CN, 25 ꢁC, 2.5 min, b:a = 5.5–6.8:1, 92% [following Ref. 13]; (e) TEAC, CH3CN, 28 ꢁC for varying times, that is, b:a = 2.7:1 after 0.5 h and
˚
b:a = 1:2 after 1.5 h [following Ref. 17]; (f) Ag2CO3/AgClO4, 1:1 toluene–CH2Cl2, molecular sieves (3 A), N2, in dark, reaction of 11 (48 h) gives 13
˚
(50%) and reaction of 12 (24 h) gives 14 (28%) [following Ref. 11]; (g) Ag2CO3/AgClO4, toluene–CH2Cl2 (1:1), molecular sieves (3 A), N2, in dark,
reaction of 11 (4 days) gives 13 (23%) and reaction of 12 gives 14 (27%) [following Ref. 9]; (h) TFA, 92% for 11 (0.5 h reaction) and 95% for 12 (1 h
reaction) [following Ref. 14].

M. Liu et al. / Carbohydrate Research 340 (2005) 1273–1285
1275
(11/12)14 to give the corresponding a-glycosylated build-
ing blocks 13/14.9,11 According to the literature,15 the
b-glycosyl chloride 7 was expected to induce higher a-ste-
reoselectivity than the a-glycosyl bromide 6, a conclusion
confirmed experimentally in the present study.
that is, prior to workup. Thus, while the pathway for
generation of 5 must somehow involve water,13 this en-
ters the nominally anhydrous reaction milieu at ꢀ20 ꢁC.
As a practical matter, even if compound 5 is formed—
despite literature-described precautions to minimize its
formation—it is readily separated by column chroma-
tography. Indeed, we found that the fraction enriched
in the mixture of 2, 3, and 4 [of which 2 and 3 are needed
to continue on to the a-halo glycosylating agents 6 or 8]
did not include any 5, as judged again by the ESIMS
assay. That same chromatographic procedure did
provide 5 in sufficient purity to allow its crystallization;
such crystals were suitable for X-ray structural analysis
(Figs. 1 and 2; Tables 1 and 2).
The molecules of 5 are found in the standard chair
conformation in space group P21. The stereochemistry
at C1 is clearly a. Bond lengths and angles are as ex-
pected (Table 2); in particular, the three nitrogens of
the azide moiety are linear. After we solved the X-ray
crystallographic structure of 5 in an anhydrous form
at 173 K (spring 2004), another laboratory reported in
this journal the structure of 5 in a monohydrate form,
solved at 293 K.19 Both structures are monoclinic with
the same space group, and have similar but not identical
unit cell constants. Bond lengths and bond angles (Ref.
19 vs our Table 2) are the same, within experimental
error. However, there are significant differences in the
hydrogen-bonding pattern and in the packing, primarily
due to inclusion of water. Indeed, the unit cell constants
of the monohydrate form show an expansion parallel to
the a axis. In our case, we find only one hydrogen bond:
N1–H1Bꢂ ꢂ ꢂO10 (Table 3), while in the monohydrate
form, this hydrogen bond is broken with the concomi-
tant insertion of water. The water oxygen in the litera-
ture structure accepts the acetylamide proton, while
Route A (Scheme 1), involving three synthetic steps
and two separations by silica gel chromatography, gave
13 and 14 in overall yields of 18% and 10%, respectively,
based on 3,4,6-tri-O-acetyl-D-galactal (1). Route B
(Scheme 1), involving four synthetic steps and two sep-
arations by silica gel chromatography, gave 13 and 14
in overall yields of 5% and 6%, respectively, also from
1. For the serine derivative 13, the a:b ratio—after gly-
cosylation but before the final chromatography that re-
moved residual b—was 6:1 and 4:1 by routes A and B,
respectively. The poorer ratio in route B is attributable
to the b:a ratio of 5.5:1 in 7, as discussed below; thus,
route A is preferred. In either case, after regular silica
gel chromatography, the pure building block 13 (>99%
enantiopure) needed for glycopeptide synthesis was ob-
tained. For the threonine derivative 14, route A via a-
bromide 6 was preferred. About 10% of the b-anomer
of 14 remained after standard silica gel chromatogra-
phy, but further purification by normal-phase HPLC
gave pure 14 (>99.9% enantiopure). In practice, 14 with
ꢁ10% of unwanted b was used for solid-phase glycopep-
tide synthesis. This did not create an insurmountable
problem (Liu et al., submitted; see also Ref. 16), since
the desired a-glycopeptides (even those containing three
GalNAc-Thr units) were eventually readily separated,
using RP-HPLC, from glycopeptides contaminated with
one or more GalNAc-Thr b-anomers.
Note that there are two ways to access b-chloride 7.
The method due to Lemieux and Ratcliffe13 (route B)
uses a-iodide 8 as an intermediate and gave a ratio of
b:a in the range of 5.5–6.8:1. Alternatively, compound
7 was obtained by treating pure 6 with tetraethylammo-
nium chloride (TEAC) (route A). Under those reaction
conditions, the a-stereoisomer was also generated. Ra-
tios of b to a species were under kinetic control, that
is, 2.7:1 after 0.5 h (the recommendation of Paulsen
et al.17) 1:2 after 1.5 h, and 1:10 after 14 days at 25 ꢁC.
Thus, with time, the less stable b-chloride 7 isomerizes
to its a-anomer; since the goal was to maximize forma-
tion of the b form, reactions were carried out as rapidly
as possible, even without full conversion of 6.
2.2. X-ray crystallographic structure of N-acetyl-3,4,6-
tri-O-acetyl-2-azido-2-deoxy-a-D-galactopyranosylamine
(5)
Supporting what has been reported previously,13,18 we
confirm that compound 5 is a byproduct formed, along
with intended products 2, 3, and 4, during the azidoni-
tration reaction (Scheme 1). Through use of ESIMS,
we show here that compound 5 is already present in situ,
Figure 1. Thermal ellipsoid drawing,
a numbering scheme for
compound 5 as determined by X-ray crystallography. Both the azide
(N2, N3, and N4) and one acetyl (O8, C9, C10) groups are disordered.
The major occupancies of each were used in this drawing.

1276
M. Liu et al. / Carbohydrate Research 340 (2005) 1273–1285
Figure 2. Molecular packing diagram of 5, looking down the b axis.
Table 1. Selected X-ray crystallographic parameters for compound 5
making two additional hydrogen bonds with acetyl oxy-
gens on neighboring molecules [the authors did not
explicitly locate the water protons, but the reported
Empirical formula
Molecular weight
Space group
C14H20N4O8
372.34
P21
˚
Oꢂ ꢂ ꢂO contacts of 2.500(6) and 2.608(8) A are strongly
˚
consistent with such an interpretation].19 In comparing
the two structures, it might appear that the monohy-
drate form should be more stable, since no disorder
was found, while the anhydrous form exhibited disorder
in one of the acetyl groups and in the azide.
Unit cell dimensions
a = 8.2146(8) A; a = 90ꢁ
˚
b = 12.5045(13) A; b = 95.458(2)ꢁ
˚
c = 8.7710(9) A; c = 90ꢁ
896.87(16)
3
˚
V, A
Z
2
Density (calculated)
Absorption coefficient
F(000)
1.379 mg/mL
0.114 mmꢀ1
392
As indicated in Section 2.3, knowledge of the struc-
ture of 5 helps to provide a basis for why the next step
in the overall preparative process (Scheme 1) may be
simplified by eliminating separation of 5 from the mix-
ture before proceeding with further synthetic steps.
Crystal color, morphology
Crystal size
Colorless, prisms
0.20 · 0.15 · 0.10 mm3
2.33–27.53ꢁ
h Range for data collection
Index ranges
ꢀ10 6 h 6 10, 0 6 k 6 16,
0 6 l 6 11
9120
2146[ R(int) = 0.0282]
Reflections collected
Independent reflections
2.3. Discovery of lithiumand sodiumion binding proper-
ties of N-acetyl-3,4,6-tri-O-acetyl-2-azido-2-deoxy-a-D-
galactopyranosylamine (5)
Observed reflections [I P 2r(I)] 2007
Completeness to h = 27.53ꢁ
Data/restraints/parameters
Goodness-of-fit on F2,a
Final R indices [I > 2r(I)]
R Indices (all data)
99.3%
2146/7/257
1.061
bR1 = 0.0396, bwR2 = 0.0977
cR1 = 0.0428, bwR2 = ꢀ0.31001
To determine if the presence of byproduct 5 would have
an adverse effect on conversion of the azidonitration
product mixture to a-halo glycosylating agents, we
decided, with pure 5 in hand, to investigate particularly
its interaction with the salts, that is, lithium bromide and
lithium iodide, used in this conversion (Scheme 1), as
well as with sodium iodide as a control. These investiga-
˚
Largest diff. peak and hole
h
0.340 and ꢀ0.278 e A
i1=2
2
a GooF ¼ S ¼ R½wðF o2 ꢀ F 2cÞ Š=ðn ꢀ pÞ
:
b R1 = RkF0jꢀjFCk/RjF0j.
h
i1=2
:
2
2
c wR2 ¼ R½wðF 02 ꢀ F c2Þ Š=R½wðF 20Þ Š
.
i
Â
1
2
w ¼ q r2ðF 20Þ þ ðaÃPÞ þ bÃP þ d þ eà sinðhÞ :
tions used H NMR spectroscopy (Fig. 3; Table 4).

M. Liu et al. / Carbohydrate Research 340 (2005) 1273–1285
1277
˚
Table 2. Selected bond lengths (A) and bond angles (ꢁ) for compound
Table 3. Hydrogen bonding in compound 5
5
D–Hꢂ ꢂ ꢂA
d(D–H) d(Hꢂ ꢂ ꢂA) d(Dꢂ ꢂ ꢂA) <(DHA)
˚
(A)
˚
(A)
˚
(A)
C1–C2
O5–C1
C3–C2
1.528(3)
1.427(3)
1.528(3)
1.479(3)
1.228(3)
1.122(4)
1.121(5)
1.526(3)
1.444(3)
1.528(3)
1.446(3)
O5–C1–C2
C3–C2–C1
C3–C2–N2
C1–C2–N2
N2–N3–N4
C4–C3–C2
O3–C3–C2
C3–C4–C5
O4–C4–C3
O5–C5–C4
O5–C5–C6
108.95(18)
111.60(18)
110.7(2)
(ꢁ)
N1–H1Bꢂ ꢂ ꢂO10a 0.88
2.18
3.009(3)
157.5
a Symmetry transformations used to generate equivalent atoms: x ꢀ 1,
y, z. D is donor, A is acceptor.
C2–N2
N2–N3
N3–N4
N30–N40
C3–C4
108.55(18)
172.7(5)
110.68(19)
106.01(18)
108.21(18)
108.69(18)
110.74(19)
103.60(19)
Compound 5 was subjected to the conditions used in
the preparation of a-bromide 6 or a-iodide 8 (Scheme 1),
and then reisolated by an extractive workup. From the
superimposability of spectra (Fig. 3), recorded in aceto-
nitrile-d3 before or after exposure to lithium bromide,
lithium iodide, or sodium iodide, it was concluded that
no significant reactions occur between ions and sugar
5. The a configuration of the sugar is retained upon
interactions with metal ions. As a practical point, then,
it was possible to take the entire azidonitration product
mixture, which includes about 30 mol % of 5, and carry
out the subsequent reaction. Nevertheless, the amount
of salts needed for full conversion of 2 and 3 had to
O3–C3
C4–C5
O4–C4
C5–C61.514(3)
O5–C5
C6–O6
–O6C51–0C8.60(2)
1.432(3)
1.440(3)
1.313(5)
1.382(9)
1.195(5)
1.206(10)
1.493(7)
1.491(14)
1.440(3)
1.359(3)
N1–C1–C2
N1–C7–O7
N1–C7–C8
C9–O3–C90
C90–O3–C3
C9–O3–C3
O8–C9–O3
O80–C90–O3
O8–C9–C10
O80–C90–C100
112.75(19)
122.0(2)
115.4(2)
35.3(4)
O3–C9
O3–C90
O8–C9
121.1 (4)
118.4(2)
122.7(4)
121.5(7)
125.3(7)
125.8(14)
O80–C90
C9–C10
C90–C100
N1–C1
N1–C7
Figure 3. 1H NMR spectra of compound 5 in acetonitrile-d3. Chemical shift assignments were determined based on 2D COSY spectra. Panel A:
Compound 5, after purification from among the products of the azidonitration reaction. Note that when pure 5 was subjected to reaction conditions
involving LiBr, LiI, or NaI (details in text), following which, the reaction was quenched, organic product was extracted into CH2Cl2, while aqueous
washes removed the salts, and the organic phase was concentrated, and the residue was dissolved in acetonitrile-d3, the NMR spectrum was
superimposable on the one now shown in Panel A; Panel B: Compound 5 (40 mM) and LiBr (5 equiv) were mixed; Panel C: the same, but with
7 equiv of LiI; Panel D: the same, but with 5 equiv of NaI.

1278
M. Liu et al. / Carbohydrate Research 340 (2005) 1273–1285
a,b
Table 4. Chemical shift increments (d: ppm) of compound 5 in the presence of lithium bromide, lithium iodide, or sodium iodide in acetonitrile-d3
Compound 5
N–H
H-3
H-5
H-1
H-2
H-4
H-6
In CD3CN
7.30
1.75
1.22
0.71
5.30
0.59
0.38
0.30
4.065.85
0.43
0.36
4.25
0.04
5.39
4.06
0.01
Increment (+ LiBr)
Increment (+ LiI)
Increment (+ NaI)
ꢀ0.03
0.01
ꢀ0.01
ꢀ0.03
0
ꢀ0.02
ꢀ0.01
ꢀ0.03
ꢀ0.01
0.26
ꢀ0.02
a Significant increments are shown in boldface for emphasis.
b The effects of sodium bromide, potassium bromide, and potassium iodide could not be evaluated, because of the limited solubility of these salts in
acetonitrile-d3.
be increased [e.g., in the case of LiBr, from 1.2 to
1.8 mass equiv] when 5 was also present. After the reac-
tion was complete, superfluous 5 was readily removed
by column chromatography.
NMR spectra of 5 in acetonitrile-d3 were then re-
corded in situ, in the presence of salts [LiBr, LiI, or
NaI; Fig. 3, Table 4]. In all cases, significant changes
in chemical shifts of 5 were induced at the N–H, H-3,
and H-5 sites, while the shift effects on H-1, H-2, H-4,
and H-6, among others, were negligible. These NMR
data are consistent with binding of the alkali ion by
sugar 5, and make it possible to pinpoint the molecular
contacts.
The NMR-suggested formation of complexes of lith-
ium or sodium ions with 5 was supported further by
ESIMS data (Table 5). When 5 was mixed with lithium
or sodium salts, mass spectra show as the base peaks the
expected 1:1 adducts. Interpretation of the results are
complicated somewhat by the control spectra of 5 alone,
which showed significant levels of metal:cation adducts,
for cation plus sodium or potassium adsorbed from the
external medium; even ions consisting of 5 plus cation in
a 2:1 ratio were observed. Nevertheless, the data are
clearly consistent with a strong affinity of sugar 5 to al-
kali cations.
With the X-ray structure (Figs. 1 and 2) and NMR re-
sults (Table 4) of 5 in hand, we were in a position to de-
velop a more specific understanding of how the alkali
metal ions interact with the protected carbohydrate.
While other examples of metal–carbohydrate complexes
have been reviewed,20–24 these all differ from 5 insofar as
free hydroxyl groups are available for coordination with
metal ions.
OAc
OAc
H6'
H6
H2
O
6
4
H4
5
2
H1
AcO
1
3
N−
N+
H5
-
M+
X
−
N1δ
H3
H
Ac
N
X = Br, M = Li;
X = I, M = Li, Na
Figure 4. Proposed model for complex of compound 5 with LiBr, LiI,
or NaI. Significant chemical shifts changes observed for N–H, H-3,
and H-5 are consistent with their proximity to the metal cation. The
resonance structure of the azide group shown in the above is arbitrary.
binding by azide-containing sugar 5 in a polar aprotic
organic solvent. Both the lone pair of nitrogen N1 and
the azide group are involved, but none of the oxygens
of the O-acetyl protected hydroxyl groups participate
in metal binding. Also, H-3 and H-5 are in close proxi-
mity to the cation, consistent with the experimentally
observed downfield chemical shifts diagnostic of signifi-
cant deshielding. H-5 is not as close to the cation as H-3,
so the deshielding of the former is less than that of the
latter. It was also determined, from NMR data, that
the maximal cation interaction was with lithium, and
the maximal anion interaction was with bromide.
Thus, the chemical shift changes of N–H (1.22 vs 0.71),
H-3 (0.38 vs 0.30), and H-5 (0.36vs 0.26) are greater with
lithium (iodide) than with sodium (iodide); and compar-
ing N–H (1.75 vs 1.22), H-3 (0.59 vs 0.38), and H-5 (0.43
vs 0.36) shows how the effects of (lithium) bromide are
greater than those of (lithium) iodide (Table 4). While al-
kali metals and 5 do not undergo chemical reac-
tions, the fact that they are coordinated explains why
The X-ray structural data on the disposition of func-
tional groups was taken into account to develop a model
(Fig. 4) for the likely contact points of alkali cation
Table 5. Mass spectrometric characterization of complexes of 5 with lithium bromide, lithium iodide, and sodium iodidea
Compound 5
Compound 5 + LiBr
Compound 5 + LiI
Compound 5 + NaI
373.1 (6) [M+H]+
379.1 (100) [M+Li]+
751.3 (8) [2M+Li]+
379.1 (100) [M+Li]+
395.1 (100) [M+Na]+
395.1 (100) [M+Na]+
411.1 (51) [M+K]+
767.2 (20) [2M+Na]+
783.2 (6) [2M+K]+
a Compound 5 plus salt were dissolved in methanol and subjected to ESIMS. The Table reports m/z in the positive mode of the major monoisotopic
peaks in each spectrum, followed by relative intensity (within parentheses) and assignment [within brackets].

M. Liu et al. / Carbohydrate Research 340 (2005) 1273–1285
1279
additional salt must be present when 5 is carried through
to the next step.
mechanism. Note that hydrogen chloride catalyzes the
dehydration reaction more efficiently than hydrogen
bromide.
2.4. Dehydration reaction during glycosylation of Na-(9-
fluorenylmethoxycarbonyl)-L-serine/L-threonine pentafluo-
rophenyl esters (11/12)
As a control, compounds 9 and 10 were exposed to
trifluoroacetic acid (TFA), which resulted in effective re-
moval of tert-butyl ether side-chain protection without
any kind of dehydration reaction, as evidenced by
ESIMS (spectra not shown) and RP-HPLC (Fig. 5).
The geometry of the double bond of Thr-derived
dehydration product 16 was investigated by a 1D nucle-
ar Overhauser enhancement (1D NOE) experiment (Fig.
6). Selective excitation of the methyl group at 1.91 ppm
provided enhanced signals on the NH proton at
6.16 ppm, Fmoc–CH2 protons at 4.51 ppm, Fmoc–CH
proton at 4.27 ppm, and Ar–H proton at 7.60 ppm
(Panel B). However, saturation of the NH proton at
6.16 ppm did not yield observable NOE enhancement,
perhaps due to difficulty in saturating the broad peak.
These data revealed that a single stereoisomer of dehy-
drothreonine ester was formed, that is, compound 16,
which has a Z geometry. The existence of the Z isomer
was further demonstrated when saturation of the ole-
finic quartet at 7.22 ppm in the NOE experiment did
not yield any enhancement except for the methyl group
at 1.91 ppm (Panel C). The exquisite stereospecificity of
the Thr dehydration was observed previously in another
context, and established by similar 1D NOE experimen-
tal analysis.25
During the reactions of Na-(9-fluorenylmethoxycarbo-
nyl)-L-serine/L-threonine pentafluorophenyl esters (11/
12) with 3,4,6-tri-O-acetyl-2-azido-2-deoxy-a-D-galacto-
pyranosyl bromide (6) or 3,4,6-tri-O-acetyl-2-azido-2-
deoxy-b-D-galactopyranosyl chloride (7), dehydration
products Na-(9-fluorenylmethoxycarbonyl)-dehydroala-
nine/dehydrothreonine pentafluorophenyl esters (15/16)
formed at low yields [Scheme 2: 2% and 5% of 15 in
Ser family; 3% and 4% of 16 in Thr family with glycos-
ylated donors 6 and 7, respectively]. Mechanistically, we
hypothesize that the dehydration reactions are catalyzed
by the acids, hydrogen bromide, and hydrogen chloride,
which are generated during glycosylation. Thus, bro-
mide and chloride ions are leaving groups from 6 and
7, with assistance of Ag+ ions. After combining with
protons released from 11 and 12, the haloacids could
recombine with hydroxyl groups on 11 and 12, and then
form dehydrated byproducts 15 and 16 by E1 or E2
procedures. This competes with formation of building
blocks 13 and 14, which are products of Ko¨nigs–
Knorr a-glycosylation reactions that follow an SN1
OAc
OAc
O
OAc
AcO
OAc
AcO
6
Ag+
Ag+
O
Cl
+
N3
N3
OAc
O
OAc
AcO
7
OAc
O
OAc
OAc
AcO
OAc
AcO
+
N3
O
Br
N3
N3
O
R
O
H+
CH
SN1
Fmoc N
H
C
H
C OPfp
R
O
O
OH2
R
13 or 14
H
CH
CH
O
H+X-
Fmoc
N
H
C
H
C OPfp
Fmoc
N
H
C
C OPfp
X —
H
11 or 12
(
)
E1 or E2
—H2O
R = H (Ser) or CH3 (Thr)
X = Br or Cl
H
R
C
C
15 or 16
Fmoc
OPfp
N
H
C
O
Scheme 2. Proposed mechanism for the dehydration of Fmoc-L-Ser/L-Thr-OPfp (11/12) as a side reaction during the intended reaction of 6 or 7 with
11/12 to provide 13/14.

1280
M. Liu et al. / Carbohydrate Research 340 (2005) 1273–1285
from anhyd CaH2 before use, and MeOH was main-
a
t
˚
tained over 3 A molecular sieves. N -Fmoc-L-Ser( Bu)-
OPfp (9) and Na-Fmoc-L-Thr(tBu)-OPfp (10) were
obtained from BACHEM (San Carlos, CA), and 3,4,6-
tri-O-acetyl-D-galactal was obtained from Aldrich.
TLC was carried out on E. Merck Silica Gel 60 F254
on aluminum sheets, and visualized by charring with
10% sulfuric acid in ethanol and/or by UV light
(254 nm), as appropriate. Regular silica gel chromato-
graphy was performed on silica gel (70–230 mesh) using
hexanes (92% n-hexane, obtained from Mallinckrodt
Laboratory Chemicals, Phillipsburg, NJ) and EtOAc
for elution. Analytical RP-HPLC was performed using
a Vydac C18 reversed-phase column (4.6 · 250 mm) on
a Beckman instrument configured with two model 112
pumps and a model 165 variable wavelength detector,
eluted at a 1.0 mL/min flow rate by linear gradients of
0.1% aqueous TFA (Buffer A) and 0.1% TFA in CH3CN
(Buffer B) over 40 min. Preparative normal-phase HPLC
was performed using a silica gel column (21.4 · 250 mm)
with an integrated guard module on a Beckman instru-
ment configured with a model 125P solvent module
and a model 166P detector (set at 280 nm), eluted at a
15 mL/min flow rate by an isocratic mixture of 4:1 hex-
anes–EtOAc. NMR spectra were recorded, with CDCl3
as solvent, on Varian Inova instruments operating for
1H at 300, 500, or 600 MHz and for 13C at 125 MHz.
The reported chemical shifts were referenced to Me4Si
and are considered accurate to 0.02 (1H) and 0.5
(13C). Electrospray ionization-mass spectrometry
(ESIMS) was performed on a Bruker BioTOF II instru-
ment in the University of Minnesota Department of
Chemistry, with samples being introduced via direct
infusion with a syringe pump.
Figure 5. RP-HPLC profiles of 9, 10, 11, and 12. The four panels
show, respectively (top to bottom): Fmoc-Ser(tBu)-OPfp (9) (tR
14.6min); Fmoc-Thr( tBu)-OPfp (10) (tR 17.2 min); Fmoc-Ser-OPfp
(11) (tR 6.1 min); and Fmoc-Thr-OPfp (12) (tR 7.0 min). Compounds
11 and 12 are the products from TFA-mediated deprotections of 9 and
10, and were detected directly by RP-HPLC analysis. Retention times
of dehydration products 15 and 16 are 14.6and 11.7 min, respectively
(RP-HPLC profiles not shown). Analytical RP-HPLC column (C18
:
4.6 · 250 mm) with detection at 280 nm, linear gradients of 0.1%
aqueous TFA (Buffer A) and 0.1% TFA in CH3CN (Buffer B) were run
at a 1.0 mL/min flow rate from 60% to 100% B over 40 min.
3. Conclusions
This paper compares several approaches to the prepara-
tion of Fmoc-L-Ser/L-Thr(Ac3-a-D-GalN3)-OPfp (13/
14), building blocks for glycopeptides synthesis. The
desired pure a-anomer of serine is obtained readily, with
the final stage being regular silica gel chromatography.
However, in the threonine family, isolation of enantio-
pure a-anomer 14 requires normal-phase HPLC. The
initial azidonitration reaction gives a mixture of prod-
ucts, including the byproduct, N-acetyl-3,4,6-tri-O-acet-
yl-2-azido-2-deoxy-a-D-galactopyranosylamine (5), the
structure of which has been confirmed by X-ray crystal-
lography. 1H NMR studies reveal that 5 can bind to lith-
ium or sodium cations, and a model to explain these
results has been proposed. Structural and mechanistic
insights garnered throughout this work have led to
streamlined, practical syntheses of intermediates, and
final building blocks.
4.2. Azidonitration of 3,4,6-tri-O-acetyl-D-galactal
Following Lemieux and Ratcliffe13 a solution of 3,4,6-tri-
O-acetyl-D-galactal (1) (2.11 g, 7.7 mmol) in CH3CN
(42 mL) was added dropwise over 5 min to a vigorously
stirred mixture of solid sodium azide (0.75 g, 11.5 mmol)
and solid ceric ammonium nitrate (CAN, 12.65 g,
28.1 mmol) at ꢀ20 ꢁC. To achieve and maintain this tem-
perature, the reaction vessel was immersed in a brine
solution that was in turn surrounded by powdered dry
ice. Starting material 1 (tR = 26.7 min) disappeared with-
in 10 h, as monitored by RP-HPLC (0–40% Buffer B over
40 min, detection at 225 nm). Aliquots were also taken
for ESIMS evaluation. The reaction was quenched by
adding cold Et2O (50 mL) and water (50 mL). The org-
anic layer was separated and washed with ice-cold water
(3 · 50 mL), and dried (MgSO4). Evaporation of the sol-
vent left a yellow syrup (2.0 g), which will be referred to
as Ôcrude azidonitration reaction product mixtureÕ. For
some of the goals of this research, the Ôcrude mixtureÕ
could be used successfully as a starting point for further
4. Experimental
4.1. General methods
Solvents and reagents were of the highest commercially
available grade, and generally used directly as obtained
from Aldrich Chemical Co. (Milwaukee, WI), ICN Bio-
medicals (Aurora, OH), or Alfa Aesar (Wardhill, MA).
However, CH2Cl2 and toluene were freshly distilled

M. Liu et al. / Carbohydrate Research 340 (2005) 1273–1285
1281
Figure 6. NMR experiments to establish the stereochemistry of compound 16. Panel A, 1H NMR spectrum of 16. Panel B, after the methyl group has
been selectively excited. Panel C, after the olefinic proton has been selectively excited.
transformations. In other cases, the Ôcrude mixtureÕ
(2.0 g) was dissolved in 3:2 hexanes–EtOAc (10 mL)
and applied to a silica gel column (16 · 3 cm) that had
been pre-equilibrated with hexanes. The chromatogram
was eluted with 3:2 hexanes–EtOAc to provide two frac-
tions. The first fraction (1.2 g, 42%) proved to be a
mixture of azidonitration products: 3,4,6-tri-O-acetyl-2-
azido-2-deoxy-b-D-galactopyranosyl nitrate (2), a-ano-
mer 3, and talo-isomer 4; ratios of compounds 2, 3,
and 4 were typically 1:1.25( 0.20):0.23( 0.03), as deter-
peared within a few hours, and further crystal growth
was allowed to proceed for 2 days. ESIMS: m/z 783.2
[2M+K]+, 767.2 [2M+Na]+, 411.1 [M+K]+, 395.1
1
[M+Na]+, 373.1 [M+H]+. H NMR data of 2–5 were
in agreement with those previously reported.13,19
4.3. 3,4,6-Tri-O-acetyl-2-azido-2-deoxy-a-D-galactopyr-
anosyl bromide (6)
4.3.1. Initial method. Again following Lemieux and
Ratcliffe13 Ôchromatographed azidonitration reaction
product 2/3/4 mixtureÕ (1.2 g, 3 mmol) was treated with
a suspension of LiBr (1.3 g, 15 mmol) in CH3CN
(11 mL) at 25 ꢁC for 8 h, with reaction progress being
monitored by ESIMS. The reaction mixture was diluted
with CH2Cl2 (100 mL) and water (2 · 50 mL), and the
organic phase was concentrated and applied to a regular
silica gel column (12 · 4 cm). Elution with 1:0 ! 1:1
hexanes–EtOAc provided the title product as a syrup
(0.96g, 2.42 mmol, 82%). Rf 0.66 (1:1 hexanes–EtOAc).
1
mined by the relative intensities of the diagnostic H
NMR shifts assigned to anomeric protons. Rf 0.48 (3:2
hexanes–EtOAc). ESIMS: m/z 415.1 [M+K]+, 399.1
[M+Na]+. This first chromatography fraction will be re-
ferred to as Ôchromatographed azidonitration reaction
product 2/3/4 mixtureÕ.
The second chromatographic fraction (0.5 g, 17%)
was determined to be N-acetyl-3,4,6-tri-O-acetyl-2-azi-
do-2-deoxy-a-D-galactopyranosylamine (5). Rf 0.21
(2:1 hexanes–EtOAc). Compound 5 was precipitated
from anhyd Et2O, and then redissolved in hot Et2O.
After filtration, the filtrate was brought to a brief boil,
and then allowed to cool slowly to 25 ꢁC. Crystals ap-
1
The H NMR data of this syrup agreed with earlier re-
ports13 and did not change for over a month of storage
at 4 ꢁC. 13C NMR (125 MHz, CDCl3): 170.4

1282
M. Liu et al. / Carbohydrate Research 340 (2005) 1273–1285
(C6–O–C@O), 169.8 (C4–O–C@O), 169.6 (C3–O–
C@O), 89.0 (C1), 71.5 (C5), 69.9 (C3), 66.6 (C4), 60.8
(C6), 58.7 (C2), 20.6–20.7 (3CH3). ESIMS: m/z 432.0
[M+K]+, 416.0 [M+Na]+, 394.1 [M+H]+.
6.82 (d, 1H, J1,2 4.2 Hz, H-1), 5.46(d, 1H, J3,4 3.2 Hz,
H-4), 5.18 (q, 1H, J2,3 10.7 Hz, H-3), 3.40 (q, 1H, H-
2), 2.15, 2.05, 2.00 (3s, 9H, 3CH3). ESIMS: m/z 479.9
[M+K]+, 464.0 [M+Na]+. Treatment of this yellow foam
with TEAC by the procedure just alluded to provided a
mixture of both chloride anomers (0.82 g, 92%,
b:a = 5.5–6.8:1).
4.3.2. Preferred method. The Ôcrude azidonitration
reaction product mixtureÕ (0.51 g, 1.35 mmol), together
with LiBr (0.6g, 7.0 mmol), was suspended in CH 3CN
(5 mL). The mixture was stirred at 25 ꢁC for 8 h, follow-
ing which the product was isolated by techniques already
described in the immediately preceding section. A syrup
(125 mg) was obtained, which comprised product 6, to-
gether with unreacted starting materials 2, 3, and 4
(1:0.42:0.78:0.20), as detected and quantified by NMR
[note that 5, though present, was not detected by NMR
in CDCl3, as first remarked on in Ref. 13]. The syrup
was dissolved again in CH3CN (5 mL), LiBr (0.6g)
was added, and stirring proceeded for 4 h. The reaction
mixture was diluted with CH2Cl2 (10 mL), washed with
water (2 · 5 mL), and then the organic phase was dried
(NaSO4), filtered, and concentrated to provide the title
compound (124 mg, ꢁ60%). Alternatively, a solution of
the Ôcrude azidonitration reaction product mixtureÕ
(1.0 g, ꢁ2.7 mmol) in CH3CN (10 mL) with LiBr
(1.8 g, 21 mmol) was stirred at 25 ꢁC for 8 h. The product
was isolated as described in Section 4.3.1 to provide the
title compound (354 mg, 80%).
4.5. Na-(9-Fluorenylmethoxycarbonyl)-L-serine pentafluo-
rophenyl ester (11)
Following Bock and co-workers14 compound 9 (4.54 g,
8.26mmol) was treated with TFA (80 mL) for 30 min
at 25 ꢁC. The reaction solution was concentrated to dry-
ness in vacuo and coevaporated with toluene (2 · 30 mL)
to provide the title compound 11 (3.75 g, 7.6mmol,
92%). tR 6.1 min (60–100% Buffer B over 40 min, detec-
1
tion at 280 nm). H NMR (300 MHz, CDCl3):26 d 7.74,
a
7.57, 7.38, 7.28 (8H, Ar–H), 5.93 (d, 1H, JNH–CH
7.2 Hz, NH), 4.39–4.51 (m, 2H, CHb), 4.18–4.22 (m,
3H, CHa, Fmoc–CH2), 4.00 (q, 1H, J1 11.3 Hz, J2
2.7 Hz, Fmoc–CH). The proton signal due to N–H was
suppressed by addition of a drop of D2O. ESIMS: m/z
532.1 [M+K]+, 516.1 [M+Na]+, 494.1 [M+H]+.
4.6. Na-(9-Fluorenylmethoxycarbonyl)-L-threonine pen-
tafluorophenyl ester (12)
Following Bock and co-workers14 compound 10 (4.0 g,
6.96 mmol) was treated with TFA (35 mL) for 1 h at
25 ꢁC. The reaction solution was concentrated to dryness
in vacuo and coevaporated with toluene (2 · 30 mL) to
provide the title compound 12 (3.34 g, 6.59 mmol,
95%). tR = 7.0 min (60–100% Buffer B over 40 min at
4.4. 3,4,6-Tri-O-acetyl-2-azido-2-deoxy-b-D-galactopyr-
anosyl chloride (7)
4.4.1. Initial method. Following Paulsen et al.,17 a solu-
tion of compound 6 (20 mg, 50 lmol) in CH3CN
(1.5 mL) was added to a 0.12 M solution of TEAC in
CH3CN (1.5 mL) at 28 ꢁC. After 1.5 h, the reaction mix-
ture was diluted with toluene (14.5 mL), washed with
water (2 · 5 mL), and dried (Na2SO4). 1H NMR
(300 MHz, CDCl3) revealed that b:a = 1:2. For b-ano-
mer 7: d 5.11 (d, 1H, J1,2 9.0 Hz, H-1), 4.84 (q, 1H,
J2,3 10.5 Hz, J3,4 3.3 Hz, H-3), 3.87 (q, 1H, H-2), 2.20,
2.08, 2.06(3s, 9H, 3CH 3). For a-anomer: d 6.19 (d,
1H, J1,2 3.6Hz, H-1), 5.52 (q, 1H, J4,5 1.2 Hz, J3,4
3.3 Hz, H-4), 5.39 (q, 1H, J2,3 10.8 Hz, H-3), 2.18,
2.08, 2.07 (3s, 9H, 3CH3). ESIMS: m/z 372.1
[M+Na]+, 350.1 [M+H]+, 314.1 [MꢀCl]+. When the
same material (already worked up) was rechecked after
14 days at 25 ꢁC, the b:a ratio was 1:10. When the same
reaction was carried out for 0.5 h at 28 ꢁC, the b:a ratio
after workup was 2.7:1.
1
280 nm). H NMR (300 MHz, CDCl3):26 d 7.79, 7.63,
a
7.42, 7.33 (8H, Ar–H), 5.71 (d, 1H, JNH–CH 9.6Hz,
NH), 4.73 (q, 1H, JCHa–CHb 1.8 Hz, CHa), 4.59–4.61 (m,
1H, CHb), 4.50 (q, 2H, JFmocCH –FmocCH 2.4 Hz, Fmoc–
2
2
CH2), 4.27 (t, 1H, JFmocCH–FmocCH 6.9 Hz, Fmoc–CH),
2
2.04 (br, 1H, OH), 1.36(d, 3H, JCH –CH 6.3 Hz, CH3).
3
b
Proton signals of N–H and O–H were suppressed by
addition of a drop of D2O. ESIMS: m/z 546.1 [M+K]+,
530.1 [M+Na]+, 508.1 [M+H]+.
4.7. Na-(9-Fluorenylmethoxycarbonyl)-O-(3,4,6-tri-O-
acetyl-2-azido-2-deoxy-a-D-galactopyranosyl)-L-serine
pentafluorophenyl ester (13)
4.7.1. Method A. Following Levine and co-workers11 a
solution of 11 (80 mg, 0.16mmol) in dry 1:1 toluene–
CH2Cl2 (10 mL) was stirred with Ag2CO3 (100 mg,
0.36mmol), AgClO 4 (15 mg, 0.066 mmol), and molecu-
4.4.2. Preferred method. Again following Lemieux and
Ratcliffe13 3,4,6-tri-O-acetyl-2-azido-2-deoxy-a-D-galacto-
pyranosyl iodide (8) (1.13 g, 48%) in the form of a yel-
low foam was obtained from the Ôchromatographed
azidonitration reaction product 2/3/4 mixtureÕ and evalu-
˚
lar sieves (215 mg, 3 A, powdered), in the dark and
under N2, at 20 ꢁC for 12 h. Next, a solution of compound
6 (96mg, 0.25 mmol) in dry 1:1 toluene–CH 2Cl2 (2 mL)
was added dropwise over 1 h, and the stirring was
1
ated by H NMR spectroscopy (300 MHz, CDCl3): d

M. Liu et al. / Carbohydrate Research 340 (2005) 1273–1285
1283
continued for 48 h. Then, the mixture was diluted with
CH2Cl2 (40 mL) and filtered through dry Celite (ꢁ2 g)
and concentrated, and the filtrate was washed with 5%
aq NaHCO3 (3 · 10 mL) followed by water
(2 · 10 mL). The organic layer was dried over anhyd
Na2SO4 and concentrated in vacuo to dryness and sub-
jected to standard silica gel chromatography with 4:1
hexanes–EtOAc for elution, whereupon three fractions
were collected. Fraction A (1.6mg, 2%) was Fmoc–
dehydroanaline pentafluorophenyl ester (15). Rf 0.83
in the dark and under N2, at 20 ꢁC for 1 h. Next, Ag-
ClO4 (36mg, 0.18 mmol) was added and stirring was
continued under the same conditions for 20 min. Then,
6 (285 mg, 0.73 mmol) in dry 1:1 toluene–CH2Cl2
(10 mL) was added dropwise over 1 h and stirring was
continued for 24 h. Product isolation was the same as
described for the preparation of 13. Two fractions
(RfA = 0.73; RfB = 0.33. 1:1 hexanes–EtOAc) were ob-
tained by standard silica gel chromatography. Fraction
A 16 (6mg, 3%) was Fmoc–dehydrothreonine pentaflu-
orophenyl ester. 1H NMR (500 MHz, CDCl3): d 7.77 (d,
2H, J3,4 7.5 Hz, ArH-4), 7.60 (d, 2H, J1,2 7.2 Hz, ArH-
1), 7.41 (m, 2H, ArH-3), 7.34 (m, 2H, ArH-2), 7.22 (q,
1
(1:1 hexanes–EtOAc). H NMR (500 MHz, CDCl3): d
7.78 (d, 2H, J 7.5 Hz, Ar–H), 7.63 (d, 2H, J 7.5 Hz,
Ar–H), 7.45 (m, 2H, Ar–H), 7.36(m, 2H, Ar– H), 7.09
(s, 1H, CHa), 6.57 (br, 1H, NH), 6.17 (s, 1H, CHb),
4.52 (d, 2H, J 6.9 Hz, Fmoc–CH2), 4.27 (t, 1H, Fmoc–
CH). ESIMS: m/z 514.1 [M+K]+, 498.1 [M+Na]+. Frac-
tion B was title compound 13 (65 mg, 50%). Rf 0.51(1:1
hexanes–EtOAc). 1H NMR data were in agreement with
those previously reported.9 ESIMS: m/z 845.2 [M+K]+,
829.2 [M+Na]+, 807.2 (MH)+. Fraction C (11 mg,
8.5%, b-anomer) was Na-(9-fluorenylmethoxycarbonyl)-
O-(3,4,6-tri-O-acetyl-2-azido-2-deoxy-b-D-galactopyrano-
syl)-L-serine pentafluorophenyl ester. Rf 0.43 (1:1 hexa-
nes–EtOAc). 1H NMR, (300 MHz, CDCl3): d 7.77,
1H, JCH–CH 7.2 Hz, CH), 6.16 (br, 1H, NH), 4.51 (d,
3
2H, J 6.9 Hz, Fmoc–CH2), 4.27 (t, 1H, Fmoc–CH),
1.91 (t, 1H, CH3). ESIMS: m/z 512.1 [M+Na]+. Fraction
B was further purified by normal-phase preparative
HPLC and gave the enantiopure title product 14
(110 mg, 28%, tR 39.4 min), which was also purified by
medium-pressure silica gel chromatography.9 1H NMR
(500 MHz, CDCl3):9,11 d 7.77, 7.63, 7.40, 7.32 (8H, Ar-
a
H), 5.85 (d, 1H, JNH–CH 9.0 Hz, NH), 5.48 (d, 1H, J3,4
3.0 Hz, H-4), 5.31 (dd, 1H, J2,3 11.0 Hz, H-3), 5.18 (d,
1H, J1,2 3.6Hz, H-1, diagnostic of a), 4.78 (dd, 1H,
JCHa–CHb 2.0 Hz, CHa), 4.59 (m, 1H, CHb), 4.41–4.51
(m, 2H, Fmoc–CH2), 4.28–4.31 (m, 2H, H-5, Fmoc–
CH), 4.12 (dd, J5,6 6.0 Hz, J6a–6b 1.5 Hz, 2H, H-6),
3.77 (dd, 1H, H-2), 2.17, 2.08, 2.06(3s, 9H, COC H3),
a
7.61, 7.41, 7.31 (8H, Ar–H), 5.88 (d, 1H, JNH–CH
8.7 Hz, NH), 5.34 (dd, 1H, J4,5 0.9 Hz, H-4), 4.99 (m,
1H, CHa), 4.81 (dd, 1H, J3,4 3.3 Hz, H-3), 4.58 (dd,
1H, JCHa–CHb 3.0 Hz, CHba), 4.46(m, 2H, Fmoc–C H2),
4.42 (d, 1H, J1,2 8.1 Hz, H-1, diagnostic of b), 4.26(t,
1H, J1 6.9 Hz, J2 7.2 Hz, Fmoc–CH), 4.03–4.16(m,
2H, H-6), 3.99 (dd, 1H, JCHb–CHb 10.2 Hz, CHbb), 3.83
1.46(d, JCHb–CH 6.0 Hz, 3H, CH3). ESIMS: m/z 843.2
3
[M+Na]+, 821.2 [M+H]+. The b-anomer (15 mg, 3%,
tR 45.9 min) was also obtained by normal-phase pre-
a
b
1
(m, 1H, H-5), 3.73 (dd, 1H, J2,3 10.8 Hz, H-2), 2.17,
2.08, 2.03 (3s, 9H, COCH3). ESIMS: m/z 845.2
[M+K]+, 829.2 [M+Na]+, 807.2 [M+H]+.
parative HPLC. H NMR (500 MHz, CDCl3): d 4.53
(d, J1,2 7.8 Hz, H-1), diagnostic of b. ESIMS: m/z
843.2 [M+Na]+, 821.2 [M+H]+. (Additional NP-HPLC
spectra are shown in the Supplementary data.)
4.7.2. Method B. Following Paulsen et al.,9 a solution
of compound 11 (400 mg, 0.82 mmol) in abs dry 1:1 tolu-
ene–CH2Cl2 (4 mL), together with Ag2CO3 (500 mg,
1.81 mmol), AgClO4 (69 mg, 0.33 mmol) and molecular
4.8.2. Method B. Again following Paulsen et al.,9
solution of compound 12 (385 mg, 0.76mmol) in abs
dry 1:1 toluene–CH2Cl2 (20 mL), together with Ag2CO3
a
˚
(470 mg, 1.71 mmol) and molecular sieves (1.3 g, 3 A,
˚
sieves (1.1 g, 3 A, powdered), was stirred at 25 ꢁC for 1 h
powered) were stirred at 25 ꢁC for 1 h in the dark and
under a dry N2 atmosphere. Then AgClO4 (57 mg,
0.28 mmol) and compound 7 (398 mg, a:b = 1:5.5,
1.14 mmol) in dry 1:1 toluene–CH2Cl2 (10 mL) were
added. After 3 days, the reaction was quenched by add-
ing CH2Cl2 (40 mL) and filtered over dry Celite (ꢁ4 g).
The filtrate was washed with 5% aq NaHCO3
(3 · 10 mL) and water (2 · 10 mL), and then dried
(Na2SO4). Following solvent evaporation, the sample
was redissolved in EtOAc and applied to a regular silica
gel column (15 · 3 cm), which was eluted with 4:1 hex-
anes–EtOAc to provide two fractions. Fraction A was
16 (14 mg, 4%). Fraction B was further purified by nor-
mal-phase preparative HPLC and gave the enantiopure
title product 14 (165 mg, 27%) and the b-anomer (20 mg,
3%).
in dark and under dry N2 atmosphere. A solution of
compound 7 (420 mg, a:b = 1:5.5, 1.2 mmol) in dry 1:1
toluene–CH2Cl2 (30 mL) was added dropwise over
45 min. After 4 days, product isolation by regular silica
gel chromatography provided three fractions: fraction A
(15, 20 mg, 5%). fraction B (desired 13, 152 mg, 23%),
fraction C (37 mg, 5%, b-anomer).
4.8. Na-(9-Fluorenylmethoxycarbonyl)-O-(3,4,6-tri-O-
acetyl-2-azido-2-deoxy-a-D-galactopyranosyl)-L-threo-
nine pentafluorophenyl ester (14)
4.8.1. Method A. Again following Levine and co-work-
ers11 a solution of 12 (245 mg, 0.48 mmol) in dry 1:1 tolu-
ene–CH2Cl2 (30 mL) was stirred with Ag2CO3 (300 mg,
˚
1.1 mmol) and molecular sieves (0.85 g, 3 A, powdered)

1284
M. Liu et al. / Carbohydrate Research 340 (2005) 1273–1285
97.29 The space group P21 was determined based on sys-
tematic absences and intensity statistics. A direct-meth-
ods solution was calculated, which provided all
nonhydrogen atoms from the E-map. All nonhydrogen
atoms were refined with anisotropic displacement
parameters. Hydrogen atoms were placed in ideal geo-
metrical positions in a standard riding model, except
that methyl torsion angles were refined. All hydrogen
atoms were refined with relative isotropic displacement
parameters relative to the host atom. The final full ma-
trix least squares refinement converged to R1 = 0.0396
and wR2 = 0.0977 (F2 all data).
1
4.9. H NMR of N-acetyl-3,4,6-tri-O-acetyl-2-azido-2-
deoxy-a-D-galactopyranosylamine (5) in the presence of
LiBr, LiI, or NaI
LiBr (24 mg, 0.28 mmol) was suspended in a solution of
compound 5 (20 mg, 53.7 lmol) in CD3CN (1 g) with
stirring, at 25 ꢁC. Aliquots were taken for 1H NMR spec-
troscopic (300 MHz) evaluation in situ: (a) Before adding
1
LiBr. H NMR data was in accordance with that re-
ported before.13 (b) After 3 h: d (ppm) 9.05 (d, 1H, JNH,H1
9.9 Hz, NH), 5.86–5.93 (m, 2H, H-1, H-3), 5.40 (d, 1H,
J3,4 3.0 Hz, H-4), 5.49 (t, 1H, Ja 6.3 Hz, Jb 6.0 Hz, H-
5), 4.22 (q, J1,2 5.7 Hz, J2,3 11.4 Hz, 1H, H-2), 3.99–
4.11 (m, 2H, H-6), 2.18, 2.14, 2.04, 2.02 (4s, 12H,
4CH3). (c) The reaction was quenched by adding CH2Cl2,
following which the mixture was concentrated to provide
the organic product as a residue, which was then dis-
solved in CD3CN. 1H NMR data was in accordance with
that reported before.13 The procedure for treatment with
LiI (50 mg, 0.37 mmol) was the same as that of LiBr: (b)
After 20 min: d 8.52 (d, 1H, JNH,H1 9.6Hz, N H), 5.83 (q,
J1,2 5.7 Hz, 1H, H-1), 5.68 (dd, 1H, J2,3 11.3 Hz, H-3),
5.36(dd, 1H, J3,4 3.3 Hz, H-4), 4.42 (t, Ja 6.0 Hz, Jb
6.6 Hz, 1H, H-5), 4.26(q, 1H, H-2), 4.03 (m, 2H, H-6),
2.18, 2.13, 2.01, 2.00 (4s, 12H, 4CH3). The procedure
for treatment with NaI (40 mg, 0.27 mmol) was the same
Acknowledgements
We thank Professor Gary R. Gray for critical reading of
the manuscript and William W. Brennessel of the X-ray
Crystallographic Laboratory of the Department of
Chemistry at the University of Minnesota for assistance
in solving the structure of 5. An anonymous referee
made several helpful suggestions, which are appreciated
insofar as they guided revision of the manuscript. This
work was supported by grant GM 66148 from the Na-
tional Institute of Health.
1
as that of LiBr. (b) After 20 min: H NMR (300 MHz,
CD3CN): d 8.01 (d, 1H, JNH,H1 9.6Hz, N H), 5.84 (q,
J1,2 5.7 Hz, 1H, H-1), 5.60 (dd, 1H, J2,3 11.4 Hz, H-3),
5.38 (dd, 1H, J3,4 3.3 Hz, H-4), 4.32 (t, Ja 6.0 Hz, Jb
6.6 Hz, 1H, H-5), 4.23 (q, 1H, H-2), 4.06(m, 2H, H-6),
2.13, 2.09, 2.01, 2.00 (4s, 12H, 4CH3).
Supplementary data
The supplementary crystallographic data for this paper
is contained in CCDC 264154. These data can be ob-
tained free of charge via www.ccdc.cam.ac.uk/data_re-
quest/cif, or by emailing data_request@ccdc.cam.ac.uk,
or by contacting The Cambridge Crystallographic Data
Centre, 12, Union Road, Cambridge CB2 1EZ, UK; fax:
+44 1223 336033.
4.10. X-ray data collection, solution, and refinement of
compound 5
4.10.1. Data collection. A crystal (approximate dimen-
sions 0.20 · 0.15 · 0.10 mm3) was placed onto the tip of
a 0.1 mm diameter glass fiber and mounted on a Bruker
SMART Platform CCD diffractometer for a data collec-
tion at 173(2) K. A preliminary set of cell constants was
calculated from 66 reflections harvested from three sets
of 20 frames. The data collection was carried out using
Mo Ka radiation (graphite monochromator) with a
frame time of 30 s and a detector distance of 4.90 cm.
A randomly oriented region of reciprocal space was sur-
veyed to the extent of one sphere and to a resolution of
NMR spectra of compounds 2–8, and 11–16, as well
as some additional Tables and Figures related to the
X-ray crystallographic analysis of 5, have been compiled
(total 47 pages), and Supplementary data associated
with this article can be found, in the online version at
doi:10.1016/j.carres.2005.02.029.
References
1. Van den Steen, P.; Rudd, P. M.; Dwek, R. A.; Opdenak-
ker, G. Crit. Rev. Biochem. Mol. Biol. 1998, 33, 151–208.
2. Kunz, H. Angew. Chem., Int. Ed. Engl. 1987, 26, 294–308.
3. Sames, D.; Chen, X. T.; Danishefsky, S. J. Nature 1997,
389, 587–591.
4. Kuduk, S. D.; Schwarz, J. B.; Chen, X. T.; Glunz, P. W.;
Sames, D.; Ragupathi, G.; Livingston, P. O.; Danishefsky,
S. J. J. Am. Chem. Soc. 1998, 120, 12474–12485.
5. Hojo, H.; Nakahara, Y. Curr. Protein Pept. Sci. 2000, 1,
23–48.
˚
0.77 A. Final cell constants were calculated from 3813
strong reflections from the actual data collection after
integration (SAINT).27 The intensity data were cor-
rected for absorption and decay (SADABS).28 Addi-
tional crystal and refinement information has been
compiled (Table 1).
4.10.2. Structure solution and refinement. The structure
27
was solved using SHELXS-97 and refined using SHELXL-
6. Bielfeldt, T.; Peters, S.; Meldal, M.; Bock, K.; Paulsen, H.
Angew. Chem., Int. Ed. Engl. 1992, 31, 857–859.

M. Liu et al. / Carbohydrate Research 340 (2005) 1273–1285
1285
7. Luning, B.; Norberg, T.; Tejbrant, J. Glycoconjugate J.
¨
1989, 6, 5–19.
18. Broddefalk, J.; Nilsson, U.; Kihlberg, J. J. Carbohydr.
Chem. 1994, 13, 129–132.
8. Peters, S.; Bielfeldt, T.; Meldal, M.; Bock, K.; Paulsen, H.
J. Chem. Soc., Perkin Trans. 1 1992, 9, 1163–1171.
9. Paulsen, H.; Bielfeldt, T.; Peters, S.; Meldal, M.; Bock, K.
Liebigs Ann. Chem. 1994, 4, 369–379.
19. Srinivas, O.; Muktha, B.; Radhika, S.; Guru Row, T. N.;
Jayaraman, N. Carbohydr. Res. 2004, 339, 1447–1451.
20. Rendleman, J. A. Adv. Carbohydr. Chem. 1966, 21, 209–
270.
10. Yule, J. E.; Wong, T. C.; Gandhi, S. S.; Qiu, D.; Riopel,
M. A.; Koganty, R. R. Tetrahedron Lett. 1995, 36, 6839–
6842.
21. Angyal, S. J. Pure Appl. Chem. 1973, 35, 131–146.
22. Angyal, S. J. Adv. Carbohydr. Chem. Biochem. 1989, 47, 1–
43.
`
11. Gururaja, T. L.; Ramasubbu, N.; Levine, M. J. Lett. Pept.
Sci. 1996, 3, 79–88.
23. Verchere, J. F.; Chapelle, S.; Xin, F.; Crans, D. C. Prog.
Inorg. Chem. 1998, 47, 837–945.
12. Liu, M.; Live, D.; Barany, G. Chimica Oggi 2004, 22, 30–
34.
24. Yano, S.; Mikata, Y. Bull. Chem. Soc. Jpn. 2002, 75, 2097–
2113.
´
13. Lemieux, R. U.; Ratcliffe, R. M. Can. J. Chem. 1979, 57,
1244–1251.
14. Vargas-Berenguel, A.; Meldal, M.; Paulsen, H.; Bock, K.
J. Chem. Soc., Perkin Trans. 1 1994, 18, 2615–2619.
15. Kunz, H.; Birnbach, S. Angew. Chem., Int. Ed. Engl. 1986,
25, 360–362.
25. Somlai, C.; Lovas, S.; Forgo, P.; Murphy, R. F.; Penke, B.
Synth. Commun. 2001, 31, 3633–3640.
26. Jansson, A. M.; Meldal, M.; Bock, K. J. Chem. Soc.,
Perkin Trans. 1 1992, 13, 1699–1707.
27. SAINT V6.2, Bruker analytical X-ray systems, Madison,
WI, 2001.
16. Paulsen, H.; Peters, S.; Bielfedt, T.; Meldal, M.; Bock, K.
Carbohydr. Res. 1995, 268, 17–34.
28. Blessing, R. H. Acta Crystallogr., Sect. A: Found Crystal-
logr. 1995, 51, 33–38.
17. Paulsen, H.; Richter, A.; Sinnwell, V.; Stenzel, W.
Carbohydr. Res. 1978, 64, 339–362.
29. SHELXTL V6.10, Bruker Analytical X-ray Systems,
Madison, WI, 2000.