Synthesis and immunological properties of N-modified GM3 antigens as therapeutic cancer vaccines

Article abstract of DOI:10.1021/jm0494422

The problem of immunotolerance to GM3, an important tumor-associated trisaccharide antigen, seriously hinders its usage in cancer vaccine development. To solve this problem, the keyhole limpet hemocyanin (KLH) conjugates of a series of GM3 derivatives wer

Full text of DOI:10.1021/jm0494422

J. Med. Chem. 2005, 48, 875-883
875
Synthesis and Immunological Properties of N-Modified GM3 Antigens as
Therapeutic Cancer Vaccines
Yanbin Pan,† Peter Chefalo,‡ Nancy Nagy,‡ Clifford Harding,*,‡ and Zhongwu Guo*,†
Departments of Chemistry and Pathology, Case Western Reserve University, Cleveland, Ohio 44106-7078
Received July 14, 2004
The problem of immunotolerance to GM3, an important tumor-associated trisaccharide antigen,
seriously hinders its usage in cancer vaccine development. To solve this problem, the keyhole
limpet hemocyanin (KLH) conjugates of a series of GM3 derivatives were synthesized and
screened as therapeutic cancer vaccines. First, the â-linked anomeric azides of differently
N-acylated GM3 analogues were prepared by a highly convergent procedure. Next, a pentenoyl
group was linked to the reducing end of the carbohydrate antigens following selective reduction
of the azido group. The linker was thereafter ozonolyzed to give an aldehyde functionality
permitting the conjugation of the antigens to KLH via reductive amination. Finally, the
immunological properties of the resultant glycoconjugates were studied in C57BL/6 mice by
assessing the titers of specific antibodies induced by the GM3 analogues. While KLH-GM3
elicited low levels of immune response, the KLH conjugates of N-propionyl, N-butanoyl, N-iso-
butanoyl, and N-phenylacetyl GM3s induced robust immune reactions with antibodies of
multiple isotypes, indicating significantly improved and T-cell dependent immune responses
that lead to isotype switching, affinity maturation, and the induction of immunological
“memory”. It was suggested that GM3PhAc-KLH is a promising vaccine candidate for
glycoengineered immunotherapy of cancer with GM3 as the primary target.
Introduction
ingly modified mannosamine, which serves as a biosyn-
thetic precursor of modified sialic acid, to initiate the
expression of the artificial antigen in place of the
natural TACA on tumors. The activated immune system
will thereby eliminate the marked tumor cells. The
concept is further supported by recent results from
Jennings’ group.21
In principle, this strategy should be applicable to
various TACAs, provided that two conditions are ful-
filled. First, there must be an effective synthetic vaccine
that can induce a specific immune response in cancer
patients. Second, there must be a proper protocol that
can be used to engineer cancer cells to express the same
artificial TACA. This work is focused on the first issue,
namely, the identification of an artificial antigen for
vaccine design, using GM3 (Figure 1), a sialylated
TACA, as the target molecule.1,2
GM3 is one of the most abundant TACAs on several
types of tumors, such as malignant melanoma and
neuroectodermal tumors.22 Consequently, an effective
vaccine based on GM3 can be widely useful. Unfortu-
nately, GM3 is poorly immunogenic, and attempts to
break its immunotolerance by traditional techniques,
such as linkage to KLH covalently, only achieved limited
success.22,23 Even though the incorporation of GM3 into
“very small size proteoliposomes” made of lipoproteic ex-
tracts from bacteria could improve the immunogenicity
of GM3,24 the system is complex and difficult to control.
We anticipate that our strategy might circumvent the
immunotolerance problem of GM3 and result in effective
cancer immunotherapy based on GM3. Therefore, we
have synthesized the natural GM3 and its artificial
N-acyl derivatives, linked them to KLH, and studied the
immunological properties of the resultant glycoconju-
gates.
It is well-established that oncological transformations
are accompanied by the change of cell surface glycosy-
lation patterns.1,2 The abnormal glycans expressed on
tumor cells are known as tumor-associated carbohydrate
antigens (TACAs). Since the early 1970s, many TACAs
have been identified2-6 and become useful molecular
templates and targets in the design and development
of therapeutic cancer vaccines.4-10 Traditionally, cancer
vaccines are created by coupling TACAs to a carrier
protein, such as keyhole limpet hemocyanin (KLH).
Resultant glycoconjugates can be employed to immunize
cancer patients to fight tumors that bear the same
antigens. This method has witnessed success with some
TACAs, and a few cancer vaccines thus developed are
now in clinical trials.11 However, a majority of TACAs
are not functional due to the problem of immunotoler-
ance, namely, the indifference of patients’ immune
system to TACAs and their protein conjugates.
To overcome this problem, we have recently explored
a new strategy12 that is based upon metabolic engineer-
ing of cell surface N-acetylneuraminic acid (Neu5Ac),
which was pioneered by Bertozzi13-17 and Reutter.18,19
Neu5Ac is a natural sialic acid and is significantly
overexpressed by many tumors.20 The underlining con-
cept of our strategy is to first immunize cancer patients
with a synthetic vaccine consisting of an artificial
derivative or analogue of a natural TACA that has its
sialic acid residues chemically modified. After an im-
mune response specific to the artificial antigen is
established, the patients are treated with correspond-
*
Author to whom correspondence should be addressed. Phone:
1-216-368-5059 (C.H.); 1-216-368-3736 (Z.G.). Fax: 1-216-368-0494
(C.H.); 1-216-368-3006 (Z.G.). E-mail: cvh3@case.edu; zxg5@case.edu.
† Department of Chemistry.
‡ Department of Pathology.
10.1021/jm0494422 CCC: $30.25 © 2005 American Chemical Society
Published on Web 01/13/2005

876 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 3
Pan et al.
sialic acid itself showed that hydrophobic substitutes
could often promote immunogenicity.30
Our plan was to couple the N-acyl derivatives of GM3
to a carrier protein to form glycoconjugate vaccines,
which may provide T-cell “help” for B-cell responses.31
B-cells that express membrane-bound antibodies specific
for derivatized GM3 should exhibit antibody-enhanced
uptake of the glycoconjugate, allowing enhanced B-cell
presentation of the carrier protein to T-cells. This
mechanism drives T-cell responses that promote carbo-
hydrate-specific B-cells to proliferate, undergo isotype
class switching and affinity maturation, and differenti-
ate into long-lived memory B-cells. Thus, conjugation
of carbohydrate antigens to carrier protein enhances not
only the magnitude but also the quality of antibody
responses.31
Figure 1. The structure of GM3 antigen.
KLH, a well-established carrier protein for experi-
mental cancer vaccines,6 was employed to form glyco-
conjugate vaccines 2a-e (Figure 2), because KLH con-
jugates are often highly immunogenic and induce IgG
antibody responses that are desirable for cancer immu-
notherapy. Alternatively, human serum albumin (HSA)
conjugates 3a-e were prepared as capture reagents for
ELISA assays in immunological studies, which would
avoid detection of antibodies specific for KLH.
A pentenoyl group was utilized as a linker to conju-
gate GM3 antigens to proteins. We have recently
demonstrated that as a linker a pentenoyl group has
some advantages.32 For instance, the linker and the
antigens are coupled by a relatively stable amide bond
through a selective reaction between pentenoyl anhy-
dride and glycosylamines. Moreover, because the cou-
pling reaction can be achieved at the final stage of
synthesis, protection tactics required for oligosaccharide
preparation can be significantly simplified. Ozonolysis
of the double bond of the pentenoyl group eventually
gives an aldehyde functionality that facilitates the
conjugation of GM3 antigens to proteins via reductive
amination.
Figure 2. Structures of N-modified GM3 and conjugate
vaccines.
Results and Discussion
Design of Modified GM3 Antigens and Glyco-
conjugate Vaccines. The major objectives of this work
are to understand how chemical modifications of GM3
would affect its immunogenicity and to establish effec-
tive vaccines for cancer immunotherapy using GM3. In
this context, we are especially interested in modifying
the sialic acid residue of GM3, because previous studies
suggest that chemically modified sialic acids and sialo
carbohydrates are significantly more immunogenic than
their natural counterparts.25-28 More importantly,
Bertozzi,13-17 Reutter,18,19 and recently Varki29 have
demonstrated that a variety of cells, including tumor
cells,12,21 can uptake unnatural N-acyl mannosamines
to biosynthesize artificial sialic acids and sialoglycocon-
jugates and present them onto cell surfaces. Thus,
metabolic engineering of GM3 on tumor cells may be
conveniently achieved by incorporating unnatural sialic
acids, which will meet the second requirement of our
new immunotherapy. A caveat is that in the design of
modified sialo antigens we need to bear in mind that
the correspondingly N-modified mannosamines have to
be available to the enzymes involved in the biosynthesis
of sialic acid.
We have designed and studied four unnatural N-acyl
derivatives of GM3 (Figure 2), including N-propionyl
GM3 (GM3NPr, b), N-butanoyl GM3 (GM3NBu, c),
N-iso-butanoyl GM3 (GM3NiBu, d), and N-phenylacetyl
GM3 (GM3NPhAc, e), which comprise a range of
structural variations. GM3NPr and natural GM3 have
a small structural difference, as the propionyl group is
only one carbon longer than an acetyl group. Neverthe-
less, a similar structural variation of polysialic acid
could substantially improve the immunogenicity.25,26
Therefore, it is interesting to examine how a propionyl
group would affect the immunological properties of
GM3. Butanoyl, iso-butanoyl, and phenylacetyl groups
are structurally more different from an acetyl group,
and we expect that their derivatives would be even more
immunogenic. Additionally, all acyl groups in GM3NPr,
GM3NBu, GM3NiBu, and GM3NPhAc are more hydro-
phobic than the acetyl group, which we hope would
further improve the immunogenicity. Our studies on
Consequently, glycoconjugates 2a-e and 3a-e were
synthesized for studying the immunological properties
of N-modified GM3 antigens.
Synthesis of N-Acyl GM3 Antigens and Their
Protein Conjugates. Despite the significant progress
in carbohydrate chemistry in recent years, the chemical
synthesis of complex oligosaccharides remains a signifi-
cant undertaking. One of the major problems in the
synthesis of GM3 and its derivatives is that the sialy-
lation reactions employed to introduce the sialic acid
residue usually give very poor yields and stereoselec-
tivity.33-39 To address this problem, we planned to
examine a new strategy developed by Boons and co-
workers,40,41 in which an N-trifluoroacetylneuraminic
acid derivative was utilized as the glycosyl donor to
afford significantly improved results. The use of a
trifluoroacetyl group as a protecting group for the amino
group also has other advantages in our synthesis. For
instance, a trifluoroacetamido group can be easily
deprotected under mild conditions, which can simplify
the design of protection tactics and the final stage
modification of GM3.
Another important issue in the preparation of the
protein conjugates of GM3 and its derivatives is that a
pentenoyl group as the appropriate linker has to be

Synthesis and Immunology of N-Acyl GM3
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 3 877
Scheme 1a
Scheme 2a
a Reagents and conditions: (a) O3, MeOH, -70 °C, 0.5 h; then
Me2S, to room temperature, 2 h, 85-90%; (b) KLH or HSA,
NaBH3CN, 0.1 N NaHCO3, 37 °C, 3 days.
The glycosyl acceptor 5 and donor 8 were prepared
from lactose (4) and Neu5Ac (6) according to reported
procedures.32,42 The azido group was attached to the
glycosyl acceptor segment at an early stage. The thiogly-
coside of sialic acid (7) was smoothly converted into a
trifluoroacetyl derivative 8 in two steps. The sialylation
of 5 by 8 under the conventional conditions40,41 afforded
trisaccharide 9 in a good yield (ca. 63%) and excellent
stereoselectivity, but the purification process was very
tedious due to the partial overlapping of 9 with the
excessive substrate 5 on the column. Because we found
that after introduction of the linker the product was
more easily purified, we directly applied the partially
purified 9 to the next step. The selective reduction of
the azido group of 9 and acylation of the resultant amine
were performed in one pot to afford the N-trifluoroacetyl
1
GM3 conjugate 10 in an excellent yield (90%). The H
NMR data of 10 (H-3e: δ 2.64) showed that the sialic
acid residue was R-linked33-38,40,41 and the glycosyl-
amino bond had a â-configuration (H-1: δ 4.90, J ) 9.0
Hz). The R-configuration of sialic acid was further con-
firmed in the deprotected product 11 that showed a
downfield shift of H-3e NMR signal (δ 2.63) compared
to that of the â-isomer (δ < 2.40). It is worth pointing
out that this is the first synthesis of a glycosylazido
derivative of GM3, although a number of procedures
for GM3 and other derivatives have been described
previously.33-39
After 10 was obtained, its trifluoroacetyl group was
removed with 0.5 N NaOH solution at room tempera-
ture. Under this condition, the hydroxyl and carboxyl
groups were also deprotected, but the linker was unaf-
fected. The resultant free amino group was then selec-
tively acylated in methanol by acetic, propionic, butyric,
iso-butyric, and phenylacetic anhydrides, respectively,
to afford the N-acyl derivatives of GM3 11a-e. The
products were purified on a Biogel P2 column, and their
identities and purities were characterized with MS and
high-resolution NMR.
GM3 derivatives were conjugated to the carrier
proteins through a two-step procedure (Scheme 2). First,
each GM3 derivative was treated with ozone in metha-
nol to selectively break the carbon-carbon double bond
of the linker and give an aldehyde, as shown in 12a-e.
The products were purified on a Biogel P2 column, and
their NMR spectra clearly showed a signal of the
hydrated aldehyde at about 5.0 ppm and the disappear-
ance of the double bond signals. The coupling of 12 to
carrier proteins, including KLH and HSA, was ac-
a Reagents and conditions: (a) MsOH (cat.), MeOH, reflux, 24
h; then CF3COOMe, Et3N, MeOH; (b) Ac2O, pyridine, 84% (two
steps); (c) NIS, TfOH, MS 4 Å, MeCN, ca. 63%; (d) Lindlar catalyst,
H2, MeOH/EtOAc (1:1), 4-pentenoic anhydride, 6 h, 90%; (e) 0.5
N NaOH, 10 h; then various anhydride, rt, 83-94%.
attached to the GM3 antigens. For this purpose, we
planned to have the reducing end of derivatized GM3s
capped by an azido group that can serve as a latent
amino group. Thus, after the azido group is selectively
reduced under mild conditions to give a primary amine,
the pentenoyl group can be introduced to the GM3
reducing end even in the presence of free hydroxyl
groups. The azido group can also serve as a protecting
group, because an azido group is stable to most reactions
involved in the oligosaccharide synthesis.
Meanwhile, because a trifluoroacetyl group can be
selectively removed in the presence of a pentenoyl
group, it will facilitate a procedure of introducing the
linker at an early stage before the deprotection and
acylation of the amino group on C-5. Deprotecting and
modifying the sialic acid residue at the latest stage can
lead to a highly convergent synthetic design as shown
in Scheme 1.

878 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 3
Pan et al.
Table 1. Carbohydrate Loading of Glycoconjugates
KLH conjugates
HSA conjugates
2a
(Ac)
2b
(Pr)
2c
(Bu)
2d
(iBu)
2e
(PhAc)
3a
(Ac)
3b
(Pr)
3c
(Bu)
3d
(iBu)
3e
(PhAc)
sample
loading (%)
12.4
13.5
12.8
13.5
10.8
15.0
15.4
17.5
13.2
11.9
Figure 3. Antigen-specific total antibody contents in sera analyzed by ELISA. Each line represents the antibody level in serum
pooled from six mice. Anti-GM3 and anti-derivatized GM3 specific antibody levels were obtained from mice immunized with
various GM3-KLH glycoconjugates. Anti-KLH specific antibody level, which was used as a positive control, was obtained from
mice immunized with GM3NPr, and anti-GM3 specific antibody level of preimmune serum was obtained from the same group as
the negative control (equivalent results were obtained from the other groups). For ELISA assays, the corresponding GM3-HSA
glycoconjugates were used as the capture antigens. Goat anti-mouse κ antibodies were used to detect antibodies bond to the
capture antigens. Error bars are smaller than the symbol width.
complished by reductive amination carried out in a 0.1
M NaHCO3 buffer in the presence of NaBH3CN. The
GM3 derivatives were employed in excess, and the
glycoconjugates were readily separated from the unre-
acted GM3 derivatives on a Biogel A0.5 column. The
product that gave positive results in assays for both
sugar and protein was naturally the first component
eluted out of the column and was collected, dialyzed
against distilled water, and finally freeze-dried to afford
the glycoconjugates as white fluffy solids.
The sialic acid contents of the resultant glycoconju-
gates were analyzed by the Svennerholm43 method and
then converted into the levels of carbohydrate loading
according to the equation presented in the Experimental
Section. In general, the coupling reactions were very
efficient and gave glycoconjugates containing 11-18%
(w/w) of carbohydrates (Table 1). These levels of antigen
loading are at the ideal range for glycoconjugates being
used as vaccines in immunological studies26 and as
capture reagents in ELISA assays.
Immunological Studies of Protein Conjugates of
N-Acyl GM3 Antigens. The immunogenicity of the
KLH conjugates of N-acyl GM3 derivatives was inves-
tigated in C57BL/6 mice, a well-characterized inbred
strain that is appropriate for further studies. As men-
tioned earlier, conjugating GM3 derivatives to a carrier
protein, namely, KLH herein, allows B-cells to present
GM3 derivative specific immunogenic peptides, gener-
ated by proteolytic cleavage of the KLH molecule, to
T-cells, which should lead to a significantly improved
immune response.31
At day 27 and day 35 following intraperitoneal im-
munizations (Experimental Section), animal sera were
withdrawn and pooled with respect to the individual
group and date. We assessed the total antigen-specific
antibody titers of the pooled sera by means of ELISA.
Then, the titers of specific antibody isotypes, including
IgM, IgG1, IgG2a, and IgG3, were assessed. The iso-
types of the antibodies have different capacities for
inducing the effector function important to antitumor
response (e.g., antibody-dependent cell-mediated cyto-
toxicity and the ability to fix complement on antibody
bound membranes).6,45,46 The sera obtained at day 27
and day 35 gave similar results in terms of different
GM3 antigens, but only the results of day 35 are shown
and discussed here.
Figure 3 shows the total antibody titers with the
readings of optical density (OD) plotted against pooled
sera dilutions in an ELISA assay in which plate wells
were coated with a derivatized GM3-HSA conjugate or
KLH protein alone as the capture antigen. Higher OD
values indicate higher concentrations of serum antibod-
ies specific for the capture antigen. These wells were
bound with half-log dilutions of pooled sera from day
-1 and day 35 bleeds of six immunized mice per
derivatized GM3-KLH conjugate. Pooling the serum of
the six immunized mice gives a mean serum antibody
concentration of the six immunized mice. Serum anti-
bodies bound to the capture antigen were detected with
goat anti-mouse κ light chain specific and horseradish
peroxidase (HRP) conjugated antibody. In mice, κ light
chain antibodies constitute about 95% of all antibodies,
so this assay provides a good approximation of total
antibody response.46 Using the GM3-HSA conjugates
The immunizations made use of Ribi adjuvant, an oil
emulsion system that is approved for use in humans.44

Synthesis and Immunology of N-Acyl GM3
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 3 879
Figure 4. Titer analysis of antigen-specific antibodies determined by ELISA assay (Experimental Section). Each represents the
titer in pooled serum obtained on day 35 after primary and booster immunizations (Experimental Section) from six replicate
animals.
as capture antigens allows us to detect those antibodies
that are specific for the modified GM3 component of the
conjugates because the mice have never been exposed
to HSA and the serum therefore will not contain
antibodies to this protein. In fact, none of the pooled
antisera showed binding to HSA protein (data not
shown). However, the experiment using KLH as the
capture antigen allows the determination of a positive
control, because KLH is well-established as a strong
immunogen and will elicit a high titer of antibodies
specific for KLH. Figure 3 indicates that although all
of the tested glycoconjugates (2a-e) were immunogenic,
GM3NPhAc showed the highest level of antigen-specific
antibodies. The natural GM3 antigen gave the lowest
concentration of GM3-specific antibodies. There was
essentially no binding of preimmune serum to various
GM3-HSA conjugates.
Although the raw data allows us to roughly estimate
the relative immunogenicities, a titer analysis was
performed to normalize the data and quantify more
accurately the immunogenicities of the derivatized
glycoconjugates (Figure 4). Panel A of Figure 4 shows
titers of total glycoconjugate-specific antibodies, and as
above, the assays measured the κ light chain. A hier-

880 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 3
Pan et al.
Figure 5 can also help to conclude that the linker could
not induce significant antibody responses or a strong
cross reactivity should be observed even for GM3NPhAc.
These results are understandable. At the molecular
level, GM3PhAc is the most dissimilar to the natural
GM3. Therefore, GM3PhAc should be the most im-
munogenic, while antibodies against this structure
should be least likely to bind to the natural form of GM3.
However, GM3NPr is very similar to the natural GM3,
so it was less immunogenic, and its antibodies had the
highest level of cross reactivity with the natural GM3.
In summary, an expedient method was established
to synthesize N-modified derivatives of GM3 and their
protein conjugates. The properties of these glycoconju-
gates were studied to explore the impact of structural
changes of GM3 on its immunogenicity. It was revealed
that derivatized GM3 antigens are generally more
immunogenic than GM3 itself. Moreover, whereas GM3
induces mainly IgM responses, derivatized GM3 anti-
gens induce both IgM and IgG responses. From the
viewpoint of cancer therapy, IgG responses are more
interesting because of their desirable properties such
as immunological memory, affinity maturation, and
improved antibody-dependent cell-mediated cytotoxic-
ity.45,46
GM3NPhAc proved to be the most immunogenic
among the GM3 derivatives investigated, and its anti-
sera were the least cross reactive with the natural GM3.
These results indicate that GM3NPhAc is an excellent
candidate as an antitumor vaccine. Meanwhile, our
studies47 have also shown that the N-phenylacetyl
derivative of mannoseamine is one of the best substrates
for N-acetyl sialic acid adolase, an enzyme that may be
involved in bypassing the biosynthetic bottleneck in the
glycoengineering of sialic acid and sialoglycoconjugates
on cells.48 In view of these results, we anticipate that
the combination of a GM3PhAc-based vaccine with the
glycoengineering of cancer cells by N-phenylacetyl-
mannosamine holds great promises as an effective
immunotherapy of cancer. In pursuit of this goal, we
are presently focusing on the in vivo studies of the
treatment of melanoma by using N-phenylacetylman-
noseamine as the glycoengioneering precursor and
GM3PhAc conjugates as the vaccines.
Figure 5. Titer analysis of antibodies cross reactive with the
natural GM3 determined by ELISA (Experimental Section).
ELISA plates were coated with GM3-HSA conjugate. Each
bar represents total anti-GM3 reactivity in pooled sera from
six replicate animals immunized with the indicated derivatized
GM3.
archy of immunogenicity was evident. GM3NPhAc was
the most immunogenic, followed by GM3NiBu, GM3NBu,
and then GM3NPr. The natural GM3 antigen induced
the lowest concentration of serum antibodies.
Panels B-E of Figure 4 show the titers of different
antibody isotypes. It was observed that different con-
jugates produced relatively equivalent concentrations
of IgM antibody (panel B). IgM antibody has important
complement fixing functions. However, because it is the
first antibody produced by antigen-specific B-cells in a
T-cell independent manner and it is not subject to
affinity maturation, IgM immune responses are limited
by lack of memory.31 Major titer differences were
observed with IgG responses to derivitized GM3 anti-
gens. The IgG1 antibody titers seen in panel C of Figure
4 were substantially higher than that of IgM, and the
glycoconjugates displayed the same hierarchy of immu-
nogenicity seen in panel A (i.e., GM3NPhAc being the
most immunogenic, followed by GM3NiBu, GM3NBu,
and GM3NPr). The natural GM3 induced very low levels
of IgG1 antibody. Panel D of Figure 4 shows that all of
the tested glycoconjugates gave very low IgG2a antibody
levels with only GM3PhAc, GM3NiBu, and GM3NBu
having measurable IgG2a titers. IgG3 titers (Figure 4,
panel E) showed the same hierarchy as seen in panels
A and C with a more pronounced difference in the
magnitude of the GM3NPhAc titer relative to the other
gycoconjugates.
Experimental Section
N-{O-[Methyl 4,7,8,9-tetra-O-acetyl-3,5-dideoxy-5-tri-
fluoroacetamido-D-glycero-r-D-galacto-non-2-ulopyrano-
sylonate]-(2f3)-O-(2,6-di-O-acetyl-â-D-galactopyranosyl)-
(1f4)-2,3,6-tri-O-acetyl-â-D-glucopyranosyl} 4-Penten-
amide (10). After a mixture of 8 (500 mg, 0.85 mmol), 5 (982
mg, 1.7 mmol), and activated molecular sieves (4 Å, 2.0 g) in
anhydrous acetonitrile (5.0 mL) was stirred at room temper-
ature for 20 h under N2, the mixture was cooled to -35 °C,
and NIS (1.7 mmol) and TfOH (0.2 mmol) were added. It was
kept at -35 °C for 1 h and then diluted with DCM (20 mL).
The solid material was filtered off and washed with DCM (5
mL). The combined filtrates were washed with aqueous
Na2S2O3 (20%) and water. The organic phase was dried over
Na2SO4 and concentrated under reduced pressure. Silica gel
column chromatography of the residue afforded the desired
trisaccharide 9 (585 mg, 0.53 mmol, 63%) containing a small
amount of impurities. A solution of 9 (550 mg, 0.5 mmol) and
4-pentenoic anhydride (20 mg) in MeOH and EtOAc (1:1, v/v,
5 mL) was stirred vigorously with Lindlar catalyst (60 mg)
and H2 at room temperature for 3 h. The solid material was
filtered off, and the filtrate was condensed under reduced
We next examined the reactivity of sera from mice
immunized with derivatized GM3 conjugates to the HSA
conjugate of natural GM3 (Figure 5). This allowed us
to assess the potential for autoimmune reactions in
response to immunization with the derivatized glyco-
conjugates. In general, sera from these animals showed
much lower reactivity with GM3-HSA than that with
the correspondingly modified GM3-HSA shown in
Figure 4A. This was especially true for GM3NPhAc. In
fact, the cross reactivity of GM3NPhAc sera to the
natural GM3 was almost negligible. Moreover, the
hierarchy of cross-reactivity titers is the inverse of the
titers of immunogenicity, with GM3NPhAc sera the
least cross-reactive, followed by GM3NBu and GM3NiBu.
However, the reactivity of GM3NPr sera to GM3NPr
and the natural GM3 were rather similar, suggesting
that the immune response induced by GM3NPr had
little selectivity. The result of GM3NPhAc shown in

Synthesis and Immunology of N-Acyl GM3
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 3 881
pressure. The residue was purified by column chromatography
to afford 10 (515 mg, 90%) as a white solid: Rf 0.30 (acetone
and toluene 1:1); [R]D +2.4 (c 1.0, CHCl3). 1H NMR (CDCl3,
600 MHz): δ 7.13 (d, 1 H, J 7.8 Hz, NH), 6.25 (d, 1 H, J 9.2
Hz, NH), 5.70 (m, 1 H, J 6.0, 10.8, 16.8 Hz, CH2dCH), 5.39
(m, 1 H, J 4.2 Hz, H-8′′), 5.34 (d, 1 H, J 8.4 Hz, H-7′′), 5.19
(dd, 1H, J 9.0, 10.2 Hz, H-2), 5.18 (t, 1 H, J 10.2 Hz, H-3),
4.98 (d, 1 H, J 16.8 Hz, CH2dCH), 4.92 (d, 1 H, J 10.2 Hz,
CH2dCH), 4.90 (d, 1 H, J 9.0 Hz, H-1), 4.88 (dd, 1 H, J 5.0,
10.8 Hz, H-4′′), 4.76 (dd, 1H, J 9.0, 9.6 Hz, H-2′), 4.46 (d, 1 H,
J 7.8 Hz, H-1′), 4.38 (d, 1 H, J 11.4 Hz, H-9′′a), 4.33 (d, 1 H,
J 10.8 Hz, H-6′′), 4.22 (d, 1 H, J 10.2 Hz, H-3′), 4.7 (m, 2 H),
4.13 (dd, 1 H, J 12.0, 4.2 Hz, H-6), 4.04-4.00 (m, 2 H), 3.94
(q, 1 H, J 10.2 Hz, H-5′′), 3.74 (m, 1 H, J 9.0 Hz, H-4), 3.72 (s,
3 H), 3.67 (d, 1 H, J 8.4 Hz, H-5), 3.58 (t, 1 H, J 6.0 Hz, H-5′),
3.35 (bs, 1 H, H-4′), 2.82 (d, 1 H, J 3.6 Hz, OH), 2.64 (m, H-3′′e),
2.26 (m, 2 H), 2.18 (m, 2 H), 2.06, 2.03, 2.02, 2.01, 2.00, 1.98,
1.97, 1.96, 1.94 (9s, 9 × 3H, OAc), 1.70 (dd, 1 H, J 12.6, 12.0
Hz, H-3′′a). 19F NMR (CDCl3): δ -76.7 (s). 13C NMR (CDCl3,
50 MHz): δ 172.7, 171.2, 170.9, 170.7, 170.6, 170.6, 170.5,
169.8, 169.7, 169.4, 168.2, 158.0 (q, J 37.6 Hz), 136.4, 115.8,
115.0 (q, J 285 Hz), 100.8, 96.9, 77.9, 75.1, 74.5, 72.5, 71.7,
71.0, 69.6, 68.6, 68.5, 68.4, 66.9, 66.8, 62.7, 62.1, 61.9, 53.2,
48.7, 37.8, 35.6, 28.9, 21.4, 20.9, 20.8, 20.7, 20.7, 20.6, 20.5,
20.4, 20.3. FABMS Calcd for C47H64F3N2O28 [M + H]+: 1161.4.
Found: 1162.0. Calcd for C47H63F3N2NaO28 [M + Na]+: 1183.4.
Found: 1184.0.
N-{O-[5-Acylamido-3,5-dideoxy-D-glycero-r-D-galacto-
non-2-ulopyranosylonic acid]-(2f3)-O-(â-D-galactopyra-
nosyl)-(1f4)-â-D-glucopyranosyl} 4-Pentenamide (11a-
e). Compound 10 (500 mg, 0.45 mmol) was dissolved in 0.5 N
NaOH (30 mL), and the mixture was stirred at room temper-
ature for 10 h. After neutralization and condensation under
reduced pressure, the crude product was directly used for the
acylation. To a solution of the resultant amine (20 mg, 0.03
mmol) in 2.5 mL of MeOH and 0.5 mL of NaOH (0.5 N) was
added 0.1 mL of an acyl anhydride dropwise in an ice-water
bath. After the reaction finished within 6 h as indicated by
TLC, the mixture was condensed under reduced pressure. The
residue was purified on a Biogel P-2 column with H2O as the
eluent. Fractions containing the expected product were com-
bined and freeze-dried to afford various N-acyl GM3 deriva-
tives 11a-e.
+ H]+: 834.4. Found: 834.7. HR-FABMS Calcd for C29H48N2-
NaO19 [M + Na]+: 751.2749. Found: 751.2532.
N-{O-[5-Butanoylamido-3,5-dideoxy-D-glycero-r-D-ga-
lacto-non-2-ulopyranosylonic acid]-(2f3)-O-(â-D-galac-
topyranosyl)-(1f4)-â-D-glucopyranosyl} 4-Pentenamide
(11c, 19 mg, 87%): [R]D 3.1 (c 0.7, MeOH). 1H NMR (D2O, 600
MHz): δ 5.74 (m, 1 H, J 6.6, 10.2, 16.8 Hz, CH2dCH), 4.96
(dd, 1 H, J 1.2, 17.4 Hz, CH2dCH), 4.90 (d, 1 H, J 10.2 Hz,
CH2dCH), 4.83 (d, 1 H, J 9.6 Hz, H-1), 4.39 (d, 1 H, J 7.8 Hz,
H-1′), 3.97 (dd, 1 H, J 3.0, 9.6 Hz), 3.42 (dd, 1 H, J 1.8, 9.0
Hz), 3.28 (dd, 1 H, J 9.0, 9.6 Hz), 2.62 (dd, 1 H, J 12.0, 4.2 Hz,
H-3′′e), 2.29 (m, 2 H), 2.24 (m, 2 H), 2.13 (t, 2 H, J 7.2 Hz,
COCH2CH2CH3), 1.66 (t, 1 H, J 12.0 Hz, H-3′′a), 1.47 (m, 2 H,
COCH2CH2CH3), 0.77 (t, 3 H, J 7.8 Hz, COCH2CH2CH3).
FABMS Calcd for C30H50N2NaO19 [M + Na]+: 765.3. Found:
765.4. Calcd for C30H49N2Na2O19 [M - H + 2Na]+: 787.3.
Found: 787.3. Calcd for C34H62N3O21 [M + DEA + H]+: 848.4.
Found: 848.8. HR-FABMS Calcd for C34H62N3O21 [M + DEA
+ H]+: 848.3876. Found: 848.3874.
N-{O-[5-iso-Butanoylamido-3,5-dideoxy-D-glycero-r-D-
galacto-non-2-ulopyranosylonic acid]-(2f3)-O-(â-D-ga-
lactopyranosyl)-(1f4)-â-D-glucopyranosyl} 4-Pentena-
mide (11d, 21 mg, 94%): [R]D 2.8 (c 1.0, MeOH). 1H NMR
(D2O, 600 MHz): δ 5.74 (m, 1 H, J 6.0, 10.2, 16.8 Hz, CH2d
CH), 4.95 (d, 1 H, J 16.2 Hz, CH2dCH), 4.90 (d, 1 H, J 10.2
Hz, CH2dCH), 4.83 (d, 1 H, J 9.0 Hz, H-1), 4.39 (d, 1 H, J
7.2 Hz, H-1′), 3.97 (dd, 1 H, J 3.0, 9.6 Hz), 3.28 (t, 1 H, J 9.0
Hz), 2.62 (dd, 1 H, J 12.0, 4.2 Hz, H-3′′e), 2.38 (m, 1 H,
COCHMe2), 2.29 (m, 2 H), 2.20 (m, 2 H), 1.66 (t, 1 H, J 12.0
Hz, H-3′′a), 0.96, 0.95 (2 d, 2 × 3 H, J 6.6 Hz, COCHMe2).
HR-FABMS Calcd for C30H50N2NaO19 [M + Na]+: 765.2905.
Found: 765.2936. Calcd for C30H49N2Na2O19 [M - H + 2Na]+:
787.2725. Found: 787.2728.
N-{O-[5-Phenylacetamido-3,5-dideoxy-D-glycero-r-D-
galacto-non-2-ulopyranosylonic Acid]-(2f3)-O-(â-D-ga-
lactopyranosyl)-(1f4)-â-D-glucopyranosyl} 4-Pentena-
mide (11e, 20 mg, 83%): [R]D 1.1 (c 0.7, MeOH). 1H NMR
(D2O, 600 MHz): δ 7.25-7.15 (m, 5 H), 5.73 (m, 1 H, J 7.0,
10.2, 16.8 Hz, CH2dCH), 4.96 (d, 1 H, J 1.2, 17.4 Hz, CH2d
CH), 4.89 (d, 1 H, J 10.2 Hz, CH2dCH), 4.83 (d, 1 H, J 9.0 Hz,
H-1), 4.37 (d, 1 H, J 7.8 Hz, H-1′), 3.99 (dd, 1 H, J 2.0,
9.6 Hz), 3.42 (s, 2 H, PhCH2), 3.28 (t, 1 H, J 7.2 Hz), 3.22
(dd, 1 H, J 9.0 Hz), 2.62 (dd, 1 H, J 4.0, 12.2 Hz, H-3′′e), 2.29
(m, 2 H), 2.21 (m, 2 H), 1.77 (t, 1 H, J 12.2 Hz, H-3′′a). HR-
FABMS Calcd for C34H50N2NaO19 [M + Na]+: 813.2905.
Found: 813.2907. Calcd for C34H49N2Na2O19 [M - H + 2Na]+:
835.2725. Found: 835.2750.
N-{O-[5-Acylamido-3,5-dideoxy-D-glycero-r-D-galacto-
non-2-ulopyranosylonic acid]-(2f3)-O-(â-D-galactopy-
ranosyl)-(1f4)-â-D-glucopyranosyl} 4-Oxo-butanamide
(12a-e). Ozone was bubbled into individual solutions of
11a-e (0.02 mmol) in MeOH (5 mL) at -78 °C until a blue
color appeared and remained (ca. 30 min). The solutions were
kept at -78 °C for another 10 min, and then nitrogen was
introduced to remove the remaining ozone. After Me2S (0.2
mL) was added at -78 °C, the resulting solution was allowed
to warm to room temperature over a period of 1 h and stand
for another 1 h before it was condensed in a vacuum. The crude
product was purified on a Sephadex G10 column using distilled
water as the eluent to give the aldehyde products 12a-e as
white solids, which were directly used in the conjugation
reactions without further purification.
N-{O-[5-Acetamido-3,5-dideoxy-D-glycero-r-D-galacto-
non-2-ulopyranosylonic acid]-(2f3)-O-(â-D-galactopyra-
nosyl)-(1f4)-â-D-glucopyranosyl} 4-Pentenamide (11a, 20
1
mg, 92%): [R]D 2.4 (c 0.7, MeOH). H NMR (D2O, 600 MHz):
δ 5.73 (m, 1 H, J 6.6, 10.2, 17.2 Hz, CH2dCH), 4.96 (dd, 1 H,
J 1.8, 17.2 Hz, CH2dCH), 4.90 (dd, 1 H, J 1.8, 10.2 Hz, CH2d
CH), 4.83 (d, 1 H, J 9.0 Hz, H-1), 4.39 (d, 1 H, J 7.8 Hz, H-1′),
3.96 (dd, 1 H, J 3.0, 9.6 Hz), 3.80 (dd, 1 H, J 1.8, 12.6 Hz),
3.66-3.45 (dd, 1 H, J 1.8, 10.2 Hz), 3.44 (dd, 1 H, J 9.6, 10.2
Hz), 3.28 (dd, 1 H, J 9.0, 9.6 Hz), 2.61 (dd, 1 H, J 12.6, 4.8 Hz,
H-3′′e), 2.28 (m, 2 H), 2.24 (m, 2 H), 1.88 (s, 3 H, COCH3),
1.65 (t, 1 H, J 12.0 Hz, H-3′′a). FABMS Calcd for C28H46N2-
NaO19 [M + Na]+: 737.2. Found: 737.3. Calcd for C28H45N2-
Na2O19 [M - H + 2Na]+: 759.2. Found: 759.3. Calcd for
C32H58N3O21 [M + DEA + H]+: 820.3. Found: 820.4. HR-
FABMS Calcd for C28H46N2NaO19 [M + Na]+: 737.2592.
Found: 737.2422.
N-{O-[5-Propionylamido-3,5-dideoxy-D-glycero-r-D-ga-
lacto-non-2-ulopyranosylonic acid]-(2f3)-O-(â-D-galac-
topyranosyl)-(1f4)-â-D-glucopyranosyl} 4-Pentenamide
(11b, 20 mg, 91%): [R]D 2.1 (c 1.0, MeOH). 1H NMR (D2O, 600
MHz): δ 5.74 (m, 1 H, J 6.6, 10.2, 16.8 Hz, CH2dCH), 4.96
(dd, 1 H, J 1.8, 17.4 Hz, CH2dCH), 4.91 (dd, 1 H, J 1.8, 9.6
Hz, CH2dCH), 4.83 (d, 1 H, J 9.0 Hz, H-1), 4.39 (d, 1 H, J 8.4
Hz, H-1′), 3.80 (dd, 1 H, J 2.4, 12.6 Hz), 3.42 (t, 1 H, J 10.2
Hz), 3.28 (dd, 1 H, J 9.0, 9.6 Hz), 2.62 (dd, 1 H, J 12.6, 4.8 Hz,
H-3′′e), 2.28 (m, 2 H), 2.24 (m, 2 H), 2.15 (q, 2 H, J 7.8 Hz,
CH3CH2CO), 1.79 (t, 1 H, J 12.6 Hz, H-3′′a), 0.91 (t, 3 H, J 7.8
Hz, CH3CH2CO). FABMS Calcd for C29H48N2NaO19 [M + Na]+:
751.3. Found: 751.4. Calcd for C29H47N2Na2O19 [M - H +
2Na]+: 773.3. Found: 773.5. Calcd for C33H60N3O21 [M + DEA
N-{O-[5-Acetamido-3,5-dideoxy-D-glycero-r-D-galacto-
non-2-ulopyranosylonic acid]-(2f3)-O-(â-D-galactopyra-
nosyl)-(1f4)-â-D-glucopyranosyl} 4-Oxo-butanamide (12a).
1H NMR (D2O, 300 MHz): δ 5.05 (d, 1 H, J 9.0 Hz, H-1), 4.98
(t, 1 H, J 6.0 Hz, CHO), 4.54 (d, 1 H, J 7.7 Hz, H-1′), 3.96 (dd,
1 H, J 2.7, 9.7 Hz), 2.76 (dd, 1 H, J 12.6, 4.9 Hz, H-3′′e), 2.46
(m, 1 H), 2.38 (m, 2 H), 2.02 (s, 3 H, COCH3), 1.9 (m, 1 H),
1.79 (t, 1 H, J 12.5 Hz, H-3′′a). FABMS Calcd for C27H44N2-
NaO20 [M + Na]+: 739.3. Found 739.4. Calcd for C27H44N2-
KO20 [M + K]+: 755.2. Found 755.4. HR-FABMS Calcd for
C31H53N3NaO21 [M - H2O + DEA + Na]+: 826.3067. Found:

882 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 3
Pan et al.
826.3034. Calcd for C31H54N3O21 [M - H2O + DEA + H]+:
sample of the glycoconjugate (0.5 mg) in distilled water (1.0
mL) was mixed well with the resorcinol reagent (2.0 mL), the
mixture was heated in a boiling water bath for 30 min. It was
then cooled to room temperature, and to the mixture was
added an extraction solution (1-butanol acetate and 1-butanol,
85:15 v/v, 3.0 mL). The mixture was shaken vigorously before
it was allowed to stand still for ca. 10 min, allowing the organic
layer to separate well from the inorganic layer. The organic
layer was transferred to a 1.0 cm cuvette, and its absorbance
at 580 nm was determined by an UV-vis spectrometer, using
a blank organic solution as the control. The sialic acid content
of the glycoconjugate is determined against a calibration curve
created with the standard NeuNAcyl (Acyl ) Ac, Pr, Bu, iBu,
PhAc) solutions analyzed under the same conditions.43 The
carbohydrate loading of each glycoconjugate was calculated
according to the equation shown below.
804.3247. Found: 804.3247.
N-{O-[5-Propionylamido-3,5-dideoxy-D-glycero-r-D-ga-
lacto-non-2-ulopyranosylonic acid]-(2f3)-O-(â-D-galac-
topyranosyl)-(1f4)-â-D-glucopyranosyl} 4-Oxo-butana-
1
mide (12b). H NMR (D2O, 600 MHz): δ 4.88 (d, 1 H, J 9.0
Hz, H-1), 4.72 (t, 1 H, J 6.0 Hz, CHO), 4.40 (d, 1 H, J 7.8 Hz,
H-1′), 3.98 (dd, 1 H, J 2.4, 9.7 Hz), 2.62 (dd, 1 H, J 12.6, 4.8
Hz, H-3′′e), 2.50 (m, 1 H), 2.29 (m, 2 H), 2.16 (q, 2 H, J 7.8 Hz,
COCH2CH3), 1.80 (m, 1 H), 1.67 (t, 1 H, J 12.0 Hz, H-3′′a),
0.97 (t, 3 H, J 7.2 Hz, COCH2CH3). FABMS Calcd for C28H46N2-
NaO20 [M + Na]+: 753.3. Found: 753.4. Calcd for C28H46N2-
KO20 [M + K]+: 769.3. Found: 769.4. HR-FABMS Calcd for
C32H55N3NaO21 [M - H2O + DEA + Na]+: 840.3225. Found:
840.3286. Calcd for C32H56N3O21 [M - H2O + DEA + H]+:
818.3404. Found 818.3405.
N-{O-[5-Butanoylamido-3,5-dideoxy-D-glycero-r-D-ga-
lacto-non-2-ulopyranosylonic acid]-(2f3)-O-(â-D-galac-
topyranosyl)-(1f4)-â-D-glucopyranosyl} 4-Oxo-butana-
N-Acyl GM3 loading (%) )
amount of sialic acid (mg) in the sample
weight of the glycoconjugate sample (mg)
1
×
mide (12c). H NMR (D2O, 600 MHz): δ 4.85 (d, 1 H, J 9.0
Hz, H-1), 4.70 (t, 1 H, J 4.8 Hz, CHO), 4.40 (d, 1 H, J 7.8 Hz,
H-1′), 3.96 (dd, 1 H, J 2.7, 9.7 Hz), 2.60 (dd, 1 H, J 12.6, 4.8
Hz, H-3′′e), 2.40 (m, 1 H), 2.28 (m, 2 H), 2.11 (t, 2 H, J 7.2 Hz,
COCH2), 1.83 (m, 1 H), 1.65 (t, 1 H, J 12.0 Hz, H-3′′a), 1.47
(m, 2 H, COCH2CH2), 0.75 (t, 3 H, J 7.2 Hz, COCH2CH2CH3).
FABMS Calcd for C29H47N2Na2O20 [M - H + 2Na]+: 789.3.
Found 789.5. HR-FABMS Calcd for C33H57N3NaO21 [M - H2O
+ DEA + Na]+: 854.3380. Found: 854.3334. Calcd for
C33H58N3O21 [M - H2O + DEA + H]+: 832.3560. Found:
832.3556.
molecular weight of GM3
molecular weight of sialic acid
× 100%
Immunization of Mice. Six female C57BL/6 mice at the
age of 8 weeks (Jackson Laboratories, Bar Harbor, ME) were
immunized for each GM3-KLH glycoconjugate. Immuniza-
tions were intraperitoneal with glycoconjugate containing 2
µg of carbohydrate in 200 µL of saline mixed with 200 µL of
MPL/TDM Ribi adjuvant (Sigma Chemical, St. Louis, MO)
following the manufacturer’s protocol. The mice were boosted
with identical immunizations on days 14, 21, and 28 following
the initial immunization. The mice were bled by tail vein prior
to the initial immunization on day 0 and after immunization
on day 27 and day 35. Bleed was clotted to obtain sera, which
were stored at -80 °C.
Protocols for ELISA Analysis. ELISA plates were coated
with KLH or GM3 and GM3 derivatives conjugated to HSA.
These capture reagents allowed detection of antibodies specific
for KLH or the GM3 and derivatized GM3 components of the
glycoconjugates. Maxisorp ELISA plates (NuncNalgene, Roch-
ester, NY) were coated overnight at 4 °C with 100 µL of HSA
glycoconjugates or KLH (1 µg/mL 0.1 M bicarbonate buffer),
and then washed with PBS. Sera from the six mice per group
were pooled, diluted 1:300 to 1:72 900 in serial half-log
dilutions in PBS with 0.02% azide, and incubated overnight
in the coated ELISA plates (100 µL/well). The plates were then
washed and incubated with 1:1000 dilution of alkaline phos-
phatase linked anti-κ, anti-IgM, or anti-IgG2a antibodies or
with 1:2000 dilution of anti-IgG1 or anti-IgG3 antibodies
(Southern Biotechnology, Buckingham, AL) for 1 h at room
temperature. Plates were washed and developed with PNPP
substrate for colorimetric readout using a BioRad 550 plate
reader (BioRad, Hercules, CA) at 405 nM wavelength.
A titer analysis was performed to normalize data and
calculate relative immunogenicities of derivatized glycocon-
jugates.30 Optical density (OD) values were plotted against
dilution values, and a best-fit line was obtained. The equation
of this line was used to calculate the dilution value at which
an OD of 0.5 was achieved, and antibody titer was calculated
at the inverse of this dilution value.
N-{O-[5-iso-Butanoylamido-3,5-dideoxy-D-glycero-r-D-
galacto-non-2-ulopyranosylonic acid]-(2f3)-O-(â-D-ga-
lactopyranosyl)-(1f4)-â-D-glucopyranosyl} 4-Oxo-butan-
1
amide (12d). H NMR (D2O, 600 MHz): 4.90 (d, 1 H, J 9.0
Hz, H-1), 4.82 (t, 1 H, J 6.0 Hz, CHO), 4.40 (d, 1 H, J 7.8 Hz,
H-1′), 3.85 (t, 1 H, J 9.6 Hz), 3.44 (dd, 1 H, J 8.0, 10.0 Hz),
3.40 (dd, 1 H, J 1.8, 9.6 Hz), 2.62 (dd, 1 H, J 12.6, 4.8 Hz,
H-3′′e), 2.57 (m, 1 H), 2.38 (m, 1 H, COCH), 2.32 (m, 1 H),
2.25 (m, 1 H), 1.82 (m, 1 H), 1.66 (t, 1 H, J 12.0 Hz, H-3′′a),
0.96, 0.95 (2d, 6 H, J 6.6 Hz, CHMe2). FABMS Calcd for
C29H48N2NaO20 [M + Na]+: 767.3. Found: 767.3. Calcd for
C29H48N2KO20 [M + K]+: 783.3. Found: 783.3. HR-FABMS
Calcd for C33H57N3NaO21 [M - H2O + DEA + Na]+: 854.3380.
Found: 854.3367. Calcd for C33H58N3O21 [M - H2O + DEA +
H]+: 832.3560. Found: 832.3576.
N-{O-[5-Phenylacetamido-3,5-dideoxy-D-glycero-r-D-
galacto-non-2-ulopyranosylonic acid]-(2f3)-O-(â-D-ga-
lactopyranosyl)-(1f4)-â-D-glucopyranosyl} 4-Oxo-butan-
amide (12e). 1H NMR (D2O, 600 MHz): δ 7.25-7.18 (m, 5
H), 4.90 (d, 1 H, J 9.6 Hz, H-1), 4.41 (t, 1 H, J 7.2 Hz, CHO),
4.39 (d, 1 H, J 7.8 Hz, H-1′), 3.95 (dd, 1 H, J 3.0, 9.6 Hz), 3.84
(t, 1 H, J 9.6 Hz), 3.48 (s, 2 H, PhCH2), 3.43 (dd, 1 H, J 9.6,
7.8 Hz), 3.22 (d, 1 H, J 9.0 Hz), 2.62 (dd, 1 H, J 4.8, 12.6 Hz,
H-3′′e), 2.55 (m, 1 H), 2.23-2.33 (m, 2 H), 1.82 (m, 1 H), 1.65
(t, 1 H, J 12.0 Hz, H-3′′a). FABMS Calcd for C33H48N2NaO20
[M + Na]+: 815.2698. Found: 815.3. Calcd for C33H48N2KO20
[M + K]+: 831.3. Found: 831.3. HR-FABMS Calcd for C37H57N3-
NaO21 [M - H2O + DEA + Na]+: 902.3380. Found: 902.3360.
Calcd for C37H58N3O21 [M - H2O + DEA + H]+: 880.3560.
Found: 880.3510.
General Procedure for the Coupling between 12a-e
and KLH or HSA. A solution of 12a-e (6 mg each), KLH or
HSA (5 mg), and NaBH3CN (5 mg) in 0.1 M NaHCO3 (0.4 mL,
pH 7.5-8.0) was allowed to stand at room temperature in the
dark for 3 days with occasional shaking. The reaction mixture
was then loaded onto a Biogel A0.5 column (1 cm × 15 cm)
and eluted with 0.1 M PBS buffer (I ) 0.1, pH ) 7.8). The
fractions containing the glycoconjugate, characterized by BCA
assay for proteins and by the Svennerholm 43 method for sialic
acid, were combined and dialyzed against distilled water for
2 days. It was then lyophilized to give a white powder of the
expected glycoconjugate.
Acknowledgment. This work was supported by a
research grant from NIH/NCI (1R01 CA95142).
Supporting Information Available: Experimental pro-
cedures for 5 and 8, and the 1H NMR and MS spectra of
compounds 10 and 11a-e. This material is available free of
charge via the Internet at http://pubs.acs.org.
References
(1) Hakomori, S. Possible functions of tumor-associated carbohy-
drate antigens. Curr. Opin. Immunol. 1991, 3, 646-653.
(2) Hakomori, S.; Zhang, Y. Glycosphingolipid antigens and cancer
therapy. Chem. Biol. 1997, 3, 97-104.
Analysis of the Carbohydrate Loading Levels of the
Glycoconjugates. After the solution of an exactly weighed

Synthesis and Immunology of N-Acyl GM3
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 3 883
(3) Reisfeld, R. A.; Cheresh, D. A. H. Human tumor antigens. Adv.
Immunol. 1987, 40, 323-377.
(4) Ben-Efraim, S. One hundred years of cancer immunotherapy:
A critical appraisal. Tumor Biol. 1999, 20, 1-24.
(5) Ragupathi, G. Carbohydrate antigens as targets for active
specific immunotherapy. Cancer Immunol. Immunother. 1998,
46, 82-87.
(6) Danishefsky, S. J.; Allen, J. R. From the laboratory to the
clinic: A retrospective on fully synthetic carbohydrate-based
anticancer vaccines. Angew. Chem., Int. Ed. 2000, 39, 837-863.
(7) Hellstrom, K. E.; Gladstone, P.; Hellstrom, I. Cancer vaccines:
Challenges and potential solutions. Mol. Med. Today 1997, 286-
290.
(8) Chen, C.-H.; Wu, T.-C. Experimental vaccine strategies for
cancer immunotherapy. J. Biomed. Sci. 1998, 5, 231-252.
(9) Livingston, P. O. Construction of cancer vaccines with carbohy-
drate and protein tumor antigens. Curr. Opin. Immunol. 1992,
4, 624-629.
(10) Toyokuni, T.; Singhal, A. K. Synthetic carbohydrate vaccines
based on tumor-associated antigens. Chem. Soc. Rev. 1995, 231-
242.
(11) Stipp, D. Closing in on the cancer vaccines. Fortune 1998, 138,
168-176.
(28) Ritter, G.; Boosfeld, E.; Asluri, S.; Calves, M. J.; Oettgen, H. F.;
Old, L. J.; Livingston, P. O. Antibody response after immuniza-
tion with ganglioside GD3, GD3 congeners (lactones, amide and
gangliosidol) in patients with malignant melanoma. Int. J.
Cancer 1991, 48, 379-385.
(29) Tangvoranuntakul, P.; Gagneux, P.; Diaz, S.; Bardor, M.; Varki,
N.; Varki, A.; Muchmore, E. Human uptake and incorporation
of an immunogenic nonhuman dietary sialic acid. Proc. Natl.
Acad. Sci. U.S.A. 2003, 100, 12045-12050.
(30) Chefalo, P.; Pan, Y. B.; Nagy, N.; Harding, C. V.; Guo, Z.
Preparation and immunological studies of protein conjugates of
N-acylneuraminic acids. Glycoconjugate J. 2004, 20, 407-414.
(31) Lanzavecchia, A. Antigen-specific interaction between T and B
cells. Nature 1985, 314, 537-539.
(32) Xue, J.; Pan, Y.; Guo, Z. Neoglycoprotein cancer vaccines:
Synthesis of an azido derivative of GM3 and its efficient coupling
to proteins through a new linker. Tetrahedron Lett. 2002, 43,
1599-1602.
(33) Earle, M. A.; Manku, S.; Hultin, P. G.; Li, H.; Palcic, M. M.
Chemoenzymatic synthesis of a trimeric ganglioside GM3 ana-
logue. Carbohydr. Res. 1997, 301, 1-4.
(12) Liu, T.; Guo, Z.; Yang, Q.; Sad, S.; Jennings, H. J. Biochemical
engineering of surface R(2-8)polysialic acid for immunotargeting
cancer cells. J. Biol. Chem. 2000, 275, 32832-32836.
(13) Bertozzi, C. R.; Kiessling, L. L. Chemical glycobiology. Science
2001, 291, 2357-2364.
(14) Saxon, E.; Bertozzi, C. R. Cell surface engineering by a modified
Staudinger reaction. Science 2000, 287, 2007-2010.
(15) Mahal, L. K.; Bertozzi, C. R. Engineering cell surfaces: Fertile
ground for molecular landscaping. Chem. Biol. 1997, 4, 415-
422.
(34) Nagao, Y.; Nekado, T.; Ikeda, K.; Achiwa, K. Synthesis of
ganglioside GM3 and GM4 analogs having mimics of ceramide
moieties and their binding activities with inluenza virus A.
Chem. Pharm. Bull. 1995, 43, 1536-1642.
(35) Eisele, T.; Schmidt, R. R. Synthesis of the thio-linked ganglioside
GM3 epitope. Liebigs Ann. 1997, 865-872.
(36) Lonn, H.; Stenvall, K. Exceptionally high yield in glycosylation
with sialic acid: Synthesis of a GM3 glycoside. Tetrahedron Lett.
1992, 33, 115-116.
(16) Yarema, K. J.; Mahal, L. K.; Bruehl, R. E.; Rodriguez, E. C.;
Bertozzi, C. R. Metabolic delivery of ketone groups to sialic acid
residues. Application to cell surface glycoform engineering. J.
Biol. Chem. 1998, 273, 31168-31179.
(37) Fujita, S.; Numata, M.; Sugimot, M.; Tomita, K.; Ogawa, T. Total
synthesis of the modified ganglioside de-N-acetyl-GM3 and some
analogues. Carbohydr. Res. 1992, 228, 347-370.
(38) Murase, T.; Ishida, H.; Kiso, M.; Hasegawa, A. A facile regio-
and stereoselective synthesis of ganglioside GM3. Carbohydr.
Res. 1989, 188, 71-80.
(39) Yan, F.; Mehta, S.; Eichler, E.; Wakarchuk, W. W.; Gilbert, M.;
Schur, M. J.; Whitfield, D. M. Simplifying oligosaccharide
synthesis: Efficient synthesis of lactosamine and siaylated
lactosamine oligosaccharide donors. J. Org. Chem. 2003, 68,
2426-2431.
(40) De Meo, C.; Demchenko, A. V.; Boons, G.-J. A stereoselective
approach for the synthesis of R-sialosides. J. Org. Chem. 2001,
5490-5497.
(41) De Meo, C.; Demchenko, A.; Boons, G. J. Triflouroacetamido
substituted sialyl donors for the preparation of sialyl galacto-
sides. Aus. J. Chem. 2002, 55, 131-134.
(42) Hasegawa, A.; Murase, T.; Ogawa, M.; Ishida, H.; Kiso, M.
Synthetic studies on sialoglycoconjugates 17: Synthesis of 4-O-,
9-O-, and 4,9-di-O-acetyl -N-acetylneuraminic acids and their
derivatives. J. Carbohydr. Chem. 1990, 9, 415-428.
(43) Svennerholm, L. Estimation of sialic acids. II. Colorimetric
resorcinol-hydrochloric acid method. Biochim. Biophys. Acta
1957, 24, 604-11.
(44) Evans, J. T.; Cluff, C. W.; Johnson, D. A.; Lacy, M. J.; Persing,
D. H.; Baldridge, J. R. Enhancement of antigen-specific im-
munity via the TLR4 ligands MPL adjuvant and Ribi.529. Expert
Rev. Vaccines 2003, 2, 219-229.
(45) Casadevall, A.; Pirofski, L. A. Antibody-mediated regulation of
cellular immunity and the inflammatory response. Trends
Immunol. 2003, 24, 474-478.
(17) Mahal, K. L.; Yarema, K. J.; Bertozzi, C. R. Engineering chemical
reactivity on cell surfaces through oligosaccharide biosynthesis.
Science 1997, 276, 1125-1128.
(18) Kayser, H.; Geile, C. C.; Paul, C.; Zeitler, R.; Reutter, W.
Incorporation of N-acyl-2-amino-2-deoxy-hexoses into glycosph-
ingolipids of pheochromocytoma cell line PC12. FEBS Lett. 1992,
301, 137-140.
(19) Schmidt, C.; Stehling, P.; Schnitzer, J.; Reutter, W.; Horskorte,
R. Biochemical engineering of neural cell surfaces by the
synthetic N-propanoyl-substituted neuraminic acid precursor.
J. Biol. Chem. 1998, 273, 19146-19152.
(20) Takano, R.; Muchmore, E.; Dennis, J. W. Sialylation and
malignant potential in tumor cell glycosylation mutants. Gly-
cobiology 1994, 4, 665-674.
(21) Zou, W.; Borrelli, S.; Gilbert, M.; Liu, T.; Pon, R. A.; Jennings,
H. J. Bioengineering of surface GD3 ganglioside for immuno-
targeting human melanoma cells. J. Biol. Chem. 2004, 279,
25390-25399.
(22) Bitton, R. J.; Guthmann, M. D.; Babri, M. R.; Carnero, A. J. L.;
Alonso, D. F.; Fainboim, L.; Gomez, D. E. Cancer vaccines: An
update with special focus on ganglioside antigens. Oncol. Rep.
2002, 9, 267-276.
(23) Livingston, P. O. Approaches to augmenting the immunogenicity
of melanoma gangliosides: from the whole melanoma cells to
ganglioside-KLH conjugate vaccines. Immunol. Rev. 1995, 145,
147-156.
(24) Estevez, F.; Carr, A.; Solorzano, L.; Valiente, O.; Mesa, C.;
Barroso, O.; Sierra, G. V.; Fernandez, L. E. Enhancement of the
immune response to poorly immunogenic gangliosides after
incorporation into very small size proteoliposomes (VSSP).
Vaccine 1999, 18, 190-197.
(25) Jennings, H. N-propionylated group B meningococcal polysac-
charide glycoconjugate vaccine against group B meningococcal
meningitis. Int. J. Infect. Dis. 1997, 1, 158-164.
(26) Jennings, H. J.; Sood, R. K. In Neoglycoconjugates: Preparation
and Applications; Lee, Y. C., Lee, R. T., Eds.; Academic Press:
San Diego, 1994; pp 325-371.
(27) Ritter, G.; Boosfeld, E.; Calves, M. J.; Oettgen, H. F.; Old, L. J.;
Livingston, P. O. Antibody response after immunization with
ganglioside GD3, GD3 lactones, GD3 amide and GD3 gangliosi-
dol in the mouse. GD3 lactone I induces antibodies reactive with
human melanoma. Immunobiology 1990, 182, 32.
(46) Goldsby, R. A.; Kindt, T. J.; Osborne, B. A.; Kuby, J. In
Fundamental Immunology; W. H. Freeman and Company: New
York, 2003; pp 76-105.
(47) Pan, Y.; Ayani, T.; Nadas, J.; Guo, Z. Accessibility of N-acyl
derivatives of D-mannosamine to N-acetylneuraminic acid aldo-
lase. Carbohydr. Res. 2004, 339, 2091-2100.
(48) Jacobs, C. L.; Goon, S.; Yarema, K. J.; Hinderlich, S.; Hang, H.
C.; Chai, D. H.; Bertozzi, C. R. Substrate specificity of the sialic
acid biosynthetic pathway. Biochemistry 2001, 40, 12864-12874.
JM0494422