Enhancement of the immune response to poorly immunogenic gangliosides after incorporation into very small size proteoliposomes (VSSP)
Certain gangliosides are tumor-associated antigens that constitute potential targets for cancer immunotherapy. A major drawback in the design of ganglioside-based cancer vaccines, however, is the poor immunogenicity of these glycolipids. Here we report the immunological and physicochemical properties of very small size proteoliposomes (VSSP) obtained by using anionic detergents to incorporate gangliosides into the outer membrane protein complex (OMPC) of N. meningitidis. VSSP of three different gangliosides, GM3, NGcGM3 and GD3, were tested. These gangliosides differ in level of expression in normal tissues and in immunogenicity in different animal species. We show that the immunization with VSSP in an oil adjuvant consistently induced both IgM and IgG anti-ganglioside antibodies. In the mouse, the anti-ganglioside IgG fraction was not restricted to the typical T-independent isotype IgG3. Unexpectedly, significant levels of the T-dependent IgG1, IgG2a and particularly IgG2b were also found. VSSP-mediated enhancement of the immunogenicity was not restricted to the relatively immunogenic ganglioside GD3, satisfactory immune responses against highly tolerated GM3 and NGcGM3 were also obtained. Similar results were achieved in chickens and monkeys. No reactogenicity was observed even when self-gangliosides were used for immunization. VSSP overcame natural tolerance to gangliosides in an adjuvant dependent fashion.
Enhancement of the immune response to poorly immunogenic
gangliosides after incorporation into very small size
*, Adriana Carr
, Leobaldo Solorzano
, Oscar Valiente
, Osquel Barroso
, Gustavo Victoriano Sierra
, Luis Enrique Fernandez
Finlay Institute, Ave. 27 No. 19805, La Lisa, P.O. Box 16 017, C. de La Habana, Cuba
Center of Molecular Immunology, Calle 216 esq. a 15, Atabey, Playa, P.O. Box 16 040, C. de La Habana, Cuba
Received 25 November 1998; received in revised form 19 April 1999; accepted 26 April 1999
Certain gangliosides are tumor-associated antigens that constitute potential targets for cancer immunotherapy. A major
drawback in the design of ganglioside-based cancer vaccines, however, is the poor immunogenicity of these glycolipids. Here we
report the immunological and physicochemical properties of very small size proteoliposomes (VSSP) obtained by using anionic
detergents to incorporate gangliosides into the outer membrane protein complex (OMPC) of N. meningitidis. VSSP of three
dierent gangliosides, GM3, NGcGM3 and GD3, were tested. These gangliosides dier in level of expression in normal tissues
and in immunogenicity in dierent animal species. We show that the immunization with VSSP in an oil adjuvant consistently
induced both IgM and IgG anti-ganglioside antibodies. In the mouse, the anti-ganglioside IgG fraction was not restricted to the
typical T-independent isotype IgG3. Unexpectedly, signi®cant levels of the T-dependent IgG1, IgG2a and particularly IgG2b
were also found. VSSP-mediated enhancement of the immunogenicity was not restricted to the relatively immunogenic
ganglioside GD3, satisfactory immune responses against highly tolerated GM3 and NGcGM3 were also obtained. Similar results
were achieved in chickens and monkeys. No reactogenicity was observed even when self-gangliosides were used for
immunization. VSSP overcame natural tolerance to gangliosides in an adjuvant dependent fashion. # 1999 Published by
Elsevier Science Ltd. All rights reserved.
Keywords: Cancer vaccines; Proteosomes; Gangliosides; Proteoliposomes
Gangliosides are sialylated glycosphingolipids pre-
sent on the plasma membrane of mammalian cells. In
tumors, however, they are overexpressed or display
abnormal glycosylation patterns [1,2]. Sometimes, free
gangliosides are also shed into the blood stream of
patients suering certain malignancies [3,4].
It has been observed that cancer patients who
develop antibodies against certain gangliosides have
better prognosis and prolonged survival compared to
those that are antibody -negative . Similarly, anti-
bodies agains t dierent gangliosides have been shown
to mediate protection in animal models against chal-
lenge with ganglioside-expressing tumors [6,7]. These
®ndings, among others, point to gangliosides as prom-
ising targets for the active speci®c immunotherapy of
cancer . However, a major drawback of gangli oside
vaccines is the poor immunogenicity of these glyco-
lipids. Gangliosides are generally autoantigens toler-
ated immunologically to a greater or lesser degree.
Furthermore, they are categorized as T-cell -indepen-
dent antigens, i.e. they are not recognized by T lym-
phocytes, which therefore do not provide help
Vaccine 18 (2000) 190±197
0264-410X/99/$ - see front matter # 1999 Published by Elsevier Science Ltd. All rights reserved.
PII: S 026 4 - 4 1 0 X ( 9 9 ) 00 21 9 - 4
* Corresponding author. Present address: Center of Molecular
Immunology, Calle 216 esq. a 15, Atabey, Playa, A. Postal 16040, C.
de La Habana, Cuba. Tel.: +53-7-335049; fax: +53-7-333509.
E-mail address: firstname.lastname@example.org (F. Estevez)
(cytokines) to drive the maturation of the immune re-
To augment ganglioside immunogenicity, both a
protein carrier and a strong adjuvant must be used.
Dierent approaches have been studied [9,10], includ-
ing the use of carrier proteins capable to non-covalent
bind gangliosides. As no chemical modi®cation is
needed to bring carrier protein and ganglioside
together, the full antigenic structure of the latter
remains unchanged, making achievement of success in
immunization more likely.
In a previous work, Portoukalian et al.  immu-
nizing with 9-0-acetyl-GD3 ganglioside adsorbed onto
VLDL (very low density lipoproteins) could induce
both IgM and IgG antibodies in mice. Another work
by Livingston et al.  showed that the addition of
GD3 to puri®ed preparations of meningococcal outer
membrane proteins (proteosomes) generated conju-
gates that consistently induced anti-GD3 IgM an ti-
bodies in mice .
These results, however, have been obtained only for
very low-expressed gangliosides, which are known to
be fairly good immunogens . Previous work in our
laboratory has shown that VLDL-based preparations
do not induce antibodies against highly expressed
gangliosides (GM3 in chicken, GM3 and NGcGM3 in
mice, etc.) even when very strong adjuvants (complete
Freund's adjuvant, CFA; incomplete Freund's adju-
vant, IFA; or Montanide ISA 51) are used (unpub-
lished results). On the other hand, the immune
response obtained with proteosome-based ganglioside
preparations has been, until now, mostly restricted to
the IgM isotype.
We have observed, however, that proteosome-based
preparations can be optimized in such a way that they
can render immunogenic even highly tolerated ganglio-
sides. Here we report the physicochemical properties
and the immunogenicity of very small size proteolipo-
somes (VSSP) obtained by the incorporation of
gangliosides into the outer membrane protein complex
(OMPC) of N. meningitidis using anionic detergents.
VSSP of three dierent gangliosides, GM3, NGcGM3
and GD3, were tested. These gangliosides are present
at dierent levels in the normal tissues of the three ani-
mal species studied. The immune response induced by
these preparations was evaluated in terms of a time
course pro®le of antibody production, isotype/subclass
distribution and antibody speci®city. Some results are
compared among the dierent animal species.
2. Materials and methods
GM3 and N-glycolylneuraminic acid-containing
GM3 (NGcGM3) were extracted from dog and horse
erythrocytes respectively by a modi®cation of the
method described by Folch . The solvent partition
step was replaced by a mild base treatment followe d
by a solvent extraction with hexane. Further puri®-
cation was achieved by Silica Gel 60 (Merck
Darmstadt, Germany) chromatography in chloroform±
methanol±ammonia 2.5 M (65:25:4).
GD3 ganglioside was obtained from bovine butter-
milk powder and puri®ed as previously described .
GM1, GD1a and GT1b were isolated from bovine
brain  and NGcGM2 from BALB/c mice livers
Purity was monitored by HPLC  or HPTLC as
described below and referred to standard gangliosides.
2.2. Animals and immunization schedule
Three dierent animal models were used: 10±12
week-old outbred chickens, line B4 (LABIOFAM,
Cuba); BALB/c female mice, 7±8 weeks of age, pur-
chased from CENPA LAB (Havana, Cuba); and young
non-human primates (Macacas fascicularis ), which
were maintained in the animal house facility of
CENPALAB (Havana, Cuba).
Immunization was carried out by the intramuscular
route as speci®ed for each experiment. Immunogens
were prepared either by mixing eq ual volumes of vac-
cine and a 2 mg/ml solution of alum (Alhydrogel;
Superfos Biosector, Denmark) or by emulsifying equal
volumes of vaccine and an oil adjuvant. Two dierent
oil adjuvants were used: CFA (Sigma, St. Louis) and
incomplete Freund adjuvant (Montanide ISA 51)
(Seppic, France). The latter is a highly puri®ed form
of IFA in which the irritant emulsify ing agent Alarcel
A has been replaced with mannide monooleate
(Montanide 80). This form was preferred because it is
less toxic than IFA and has been extensively used in
human trials of contraceptive, cancer and AIDS vac-
cines . Each dose contained 120 mg of ganglioside
in a total volume of 0.1 ml. Animals were bled before
each immunization and 14 d after the last one, unless
otherwise stated. Sera were stored at ÿ208C.
Previously to immunization and blood collection,
monkeys were anesthetized with 1 mg of Ketamina
(Lab. Reig Jofre, S.A., Barcelona) per kg of weight by
the intramuscular route. All the animals were treated
according to the Cuban National Laboratory Animal
2.3. Preparation of vaccines
The outer membrane proteins complex (OMPC) was
puri®ed at the Finlay Institute (Havana, Cuba) as pre-
viously describ ed . Group B N. meningitidis strain
385 (B4 P1.15) was cultivated and the culture centri-
F. Estevez et al. / Vaccine 18 (2000) 190±197 191
fuged to obtain 500 g (wet weigh t) of biomass. This
biomass was resuspended in 25 l of buer containing
0.09% sodium deoxycholate (DOC), 50 mM Tris±HCl,
pH 8.5, and 2 mM EDTA. The extraction process was
carried out at 48C for 2.5 h. During this period, 10
treatments of 30 s each were given in an ultrasonic
bath (Ultraware, Radleys, UK), alternating with mag-
netic stirring at 250 rpm.
Cell debris was separated by centrifugation at
10,000g and the supernatant treated with DNase and
RNase (5 mg/l each) at 378C. The extra ct was centri -
fuged at 100,000g for 2 h and the pellet resuspended in
200 ml 5% sodium deoxycholate/(50 mM) Tris±EDTA
buer, pH 9. Gel ®ltration on Sephacryl S-300 HR
(Pharmacia Biotech, Sweden) was carried out using
1% sodium deoxycholate as eluant. The ®rst eluted
peak was the basic material containing the vesicles as
revealed by electron microscopy following negative
staining with uranyl acetate. The OMPC was precipi-
tated with ethanol, was hed with the same solvent, and
stored at ÿ808C in 40 mg portio ns.
To incorporate gangliosides into the OMPC (VSSP
preparation procedure), 40 mg aliquots of OMPC were
dissolved to a ®nal concentration of 1 mg/ml in 0.01
M Tris±HCl buer, pH 8.5, containing 12 mM DOC
and 1 mM sodium dodecylsulfate (SDS). Forty mg
ganglioside was then added and the mixture stirred.
The resultant solution was dialyzed for 14 d against
0.01 M Tris±HCl buer, pH 8.5, using a 3.5 KDa cut-
o membrane (Spectra/Por, Spectrum Medical
Industries, Houston, Texas) to remove detergents. The
dialyzed preparation was centrifuged for 1 h at
100,000g, pellets were discarded and supernatants
assayed for protein and ganglioside contents. Finally,
the VSSP solution obtained was ®lter-sterilized (0,2 mm
cellulose acetate membranes; Sartorius, Germany) and
stored at 48C.
Vaccine lots NGcGM3/VSSP, GM3/VSSP and
GD3/VSSP were prepared with the respective ganglio-
side and the OMPC. The lot designa ted OMPC was
prepared in the same way but without the addition of
any ganglioside, and the lot designated GM3 was pre-
pared with the ganglioside alone.
2.4. Physical and chemical characterization of vaccines
Protein assay was performed by a modi®cation of
the method of Lowry . Gangliosides were identi®ed
by HPTLC as described below and their concentration
determined by the resorcinol method .
For the electron microscopy study, samples were
adsorbed 5 min onto carbon-coated Formvar grids
and neg atively stained with 1% phosphotungstic acid,
pH 7.0, or 2% uranyl acetate. Grids were examined at
20±80,000-fold magni®cation at 60±80 kV.
Gel ®ltration chromatography on Sepharose CL-4B
(Pharmacia Biotech, Sweden) was carried out in a
C10/40 column equilibrated with 0.01 M Tris±HCl
buer, pH 8.5, containing 0.15 M NaCl. Fracti ons of
1 ml were colle cted and assayed for protein and
ganglioside concentration. The partition coecient K
was done as descri bed by Sofer and Nystrom .
Standard proteins were used as molecular weight mar-
kers in the same conditions as the samples. A cali-
bration curve, K
vs. MW, was established.
CsCl equilibrium density gradient analysis of the
vaccine was performed with an initial one-step gradient
with a density of 1.525 g/ml in the lower half and
1.005 g/ml in the upper half. Tris±HCl buer (0.01 M,
pH 8.5) was present throughout the gradient, and the
vaccine was initially uniformly distributed in the upper
solution. Centrifugation was performed in a Kontron
TFT 70.13 angle head rotor at 55,000 rpm (211,000g )
and 108 C for 48 h. Fractions were collected, and den-
sity, sialic acid content and protein concentration
determined for each fraction.
PolySorp 96-well plates (Nunc, Denmark) were
coated with 0.16 nmol/well of ganglioside dissolved in
methanol and dried at 378C for 1.5 h. To reduce back-
ground, plates were washed with PBS containing
Tween 20 (0.05%, v/v) plus 0.2 M extra of NaCl and
then blocked with skimmed milk (2%, w/v) in PBS.
Serum samples, biotinylated goat anti-IgM or anti-IgG
antibodies speci®c for each animal species and strepta-
vidin-conjugated alkaline phosphatase (Jackson, West
Grove, PA) were diluted in blocking buer and
sequentially incubated on the plates for 2 h at room
temperature with extensive was hing between each
incubation. The enzymatic reaction was visualized with
p-nitrophenyl phosphate (PNPP), 1 mg/ml, dissolved
in 1M diethanolamine buer, pH 9.6, plus 1 mM
To eliminate the eect of non-speci®c recognition,
sera were also tested on wells to which no gangliosides
had been added. The absorbance at each serum di-
lution obtained on these wells was subtracted from
that of the ganglioside-coated wells. Serological titer
was de®ned as the inverse of the highest dilution yield-
ing a ®nal ab sorbance value higher than 0.1.
Determination of IgG isotype pro®le was performed
using secondary isotype-speci®c biotinylated rat anti-
mouse IgG1, IgG2a, IgG2b and IgG3 antibodies
(PharMingen, San Diego, CA). Optimal secondary
antibody dilutions were established by ELISA with
the MAbs 14F7 (IgG1), T3 (IgG2a), T4 (IgG2b)
(Center of Molecular Immunology, Havana, Cuba)
and R24 (IgG3) (kindly provided by Dr. Philip O.
F. Estevez et al. / Vaccine 18 (2000) 190±197192
Livingston, Memorial Sloan Kettering Cancer Center,
2.6. HPTLC and enzyme immunos taining
High-performance thin-layer chromatography
(HPTLC) was carried out on Silica Gel 60 plates (E.
Merck AG, Germany) and developed with chloro-
form±methanol±ammonia 2.5 M/KCl 0.25% (50:40:10,
v/v). Ganglioside spots were visualized with resorcinol
reagent (0.2% resorcinol plus 0.25 mM CuSO
Immunostaining on HPTLC plates was performed
according to the method of Magnani et al.  with
slight modi®cations. Brie¯y, gangliosides were spotted
on plates elimate this as described above. Plates were
then dried in a warm air current, soaked in a 0.1%
solution of polyisobutylmethacrylat e (Aldrich
Chemical Co., Ltd., Gillingham, Dorset) in hexane for
1 min and then allowed to air dry. To eliminate back-
ground, plates were blocked in PBS buer containing
2% (w/v) skimmed milk at room temperature for 30
min. After overnight incubation with sera diluted
1:100 in blocking buer, plates were extensively
washed with PBS and incubated with biotinylated goat
anti-mouse IgG antibodies. Following another washing
step, Streptavidin-conjugated Alkaline Phosphatase
(Jackson, WestGrove, PA) diluted in blocking buer
was added and incubated for 2 h at room temperature.
Spots were visualized with a 0.1 mg/ml solution of 5-
bromo-4-chloro-3-indolyl phosphate in 0.1 M glycine
buer, pH 10.4. The reaction was stopped by washing
Fig. 1. Elution pro®le of vaccine lots GM3 and OMPC (A) and
GM3/VSSP (B) on Sepharose CL-4B. Continuous lines represent the
absorbance at 224 nm and broken lines the analytical results.
Arrows signaled Vo and Vf indicate the void volume and the end of
the run, respectively.
Fig. 2. Isopycnic CsCl density gradient centrifugation of vaccine lots
GM3/VSSP (A), GM3 (B) and OMPC (C). Fractions were assayed
for density by a gravimetric procedure (right axis) and for protein
and sialic acid contents by spectrophotometric methods (left axis).
F. Estevez et al. / Vaccine 18 (2000) 190±197 193
3.1. Physicochemical characteristics of the conjugates
VSSP preparations are quite transparent solutions
that can be easily ®lter-sterilized using a 0.2 mm ®lter.
Although, by electron microscopy, the OMPC dis-
played vesicular structure prior to its incorporation
into VSSP (data not shown), no VSSP preparation
did. Contrary to other proteosome-based preparations
that completely precipitate, the VSSP insoluble frac-
tion sedim enting by centrifugation for 1 h at 100,000g
(step included in the producti on procedure) constituted
less than 15% of the vaccine, indicating that VSSP
possess increased solubility.
Chromatography of GM3/VSSP vaccine on
Sepharose CL-4B (Fig. 1B) showed that both protein
and ganglioside elute in a peak of Kav 0.62 whose
equivalent molecular weight is about 100±300 KDa
when referred to globular proteins analyzed in the
same conditions. Independent runs of the GM3 and
lot OMPC (Fig. 1A) showed that GM3 alone eluted
with a K
of about 0.8, with OMPC eluting in the
void volume. No signi®cant deviations from this
elution pro®le were observed among dierent ganglio-
side/VSSP preparat ions (data not shown).
More convincing data on the association between
proteins and ganglioside was obtained by isopycnic
gradient centrifugation of the vaccine in CsCl (see Fig.
2). Ganglioside alone (lot GM3) produced a single
band at a density of 1.236 g/ml, whereas the ganglio-
side in the vaccine (lot GM3/VSSP) was found at a
higher density (1.265 g/ml). Proteins in the vaccine
were also found in a narrow band at 1.265 g/ml,
whereas non-ganglioside-containing proteins (lot
OMPC) were found dispersed throughout the gradient,
with a peak at 1.282 g/ml. These results demonstrate
an intimat e association between ganglioside and the
OMPC, probably hydrophobic in nature as indicated
by the high resistance of the complex to very high con-
centrations of CsCl.
3.2. Immunogenicity of NGcGM3/VSSP and GM3/
VSSP in chickens
The very low-expressed NGcGM3 and the highly
expressed GM3 gangliosides [25,26] were selected to
study the immunogenicity of VSSP in chicken. Four
groups of four chickens were each immunized intra-
muscularly on days 0, 14, 28 and 42 with NGcGM3/
VSSP or GM3/VSSP vaccines administered in either
alum or emuls i®ed complete Freund's adjuvant (CFA).
The serological study revealed that no animal devel-
oped antibodies against GM3 and only one of four
against NGcGM3 (titer 40,000) when vaccines were
administered in alum. On the contrary, when CFA
was used (see Fig. 3) all chickens developed high titer
IgG antibodies against both NGcGM3 and GM3.
3.3. Immunogenicity of NGcGM3/VSSP, GM3/VSSP
and GD3/VSSP in mice
Groups of ten mice each were immunized with
NGcGM3/VSSP, GM3/VSSP and GD3/VSSP vaccines
either alone or in M ontanide ISA 51 adjuvant on days
0, 14, 28 and 42. Fourteen days after the last immuniz-
ation (day 56), mice were bled and sera tested by
ELISA against the speci®c ganglioside.
No animal developed antibodies against the studied
gangliosides when vaccines were administered alone.
On the contrary, immunization with the three vaccine s
in Montanide ISA 51 adjuvant was very eective (see
Table 1). At least 80% of animals developed immuno-
globulin titers equal to or higher than 80 (IgM and
IgG). This response was not only observed in mice
immunized with the more heterologous (very low-
Fig. 3. IgG response against GM3 (closed symbols) and NGcGM3
(open symbols) in chicken immunized with GM3/VSSP and
NGcGM3/VSSP, respectively, in CFA. Each symbol represent an in-
dividual chick; arrows, immunization days.
Serological response induced by ganglioside/VSSP preparations in mice
Immunogen Adjuvant n Anti-ganglioside IgG ELISA titers (n ) Anti-ganglioside IgM ELISA titers (n )
GM3/VSSP Montanide ISA 51 10 10240 (3), 5120 (2), 640 (2), 320 (3) 640 (1), 320 (2), 160 (3), 80 (1), 0 (3)
NGcGM3/VSSP Montanide ISA 51 9 1280 (2), 640 (5), 320 (2) 1280 (3), 640 (2), 160 (3), 80 (1)
GD3/VSSP Montanide ISA 51 10 5120 (1), 1280 (1), 160 (2), 80 (4), 0 (2) 5120 (1), 1280 (1), 640 (4), 320 (3), 80 (1)
F. Estevez et al. / Vaccine 18 (2000) 190±197194
expressed) GD3 but also with autologous GM3 and
IgG antibodies induced by VSSP preparations
against gangliosides included the whole panel of iso-
types with slightly higher titers of IgG3 and IgG2b
(Fig. 4) (cross-reactivity between isotyp e-speci®c sec-
ondary antibodies was insi gni®cant). IgG antibodies
against the OMPC, however, were restricted to the T-
cell-dependent IgG isotypes IgG1, IgG2a and IgG2b.
No IgG3 titers were detected in this case.
The speci®city of IgG antibodies detected in mouse
sera was studied by HPTLC immunostaining using as
standard gangliosides GM1, GD1a, GT1b, GM3,
GD3, NGcGM3 and NGcGM2. The results of a
representative serum of each group are shown in Figs.
5 and 6. Only speci®c gangliosides were recognized by
serum IgG. Similar results were obtained with sera of
the other two animal models used in the study.
3.4. Immunogenicity in monkeys
To evaluate the immunogenicity of autologous
GM3/VSSP vaccine in an animal model closer to
human beings, two monkeys were immunized intra-
muscularly with GM3/VSSP vaccine in Montanide
ISA 51. Injections were given on days 0, 14, 28 and
42. Boosters were thereafter administered once a
month and sera were collected before each immuniz-
ation. Signi®cant levels of both IgM and IgG anti-
bodies were detectable as early as day 14 after the ®rst
immunization (see Fig. 7). IgM peaked on day 14 and
IgG on days 42±56. After a short period of decline,
levels of both immunoglobulins remained at a plateau
for the next four months of the study.
The term proteosome or proteoliposome was for-
merly used to designate liposome-like puri®ed prep-
Fig. 4. IgG isotypes of serum antibodies. Groups of BALB/c mice
were immunized four times with either NGcGM3/VSSP, GM3/VSSP
or GD3/VSSP vaccines in Montanide ISA 51 adjuvant at 14 d inter-
vals. Five day-56 sera of each immunized group were assayed by
ELISA against the speci®c antigen. The geometric mean (n=5) of
the ELISA titers were calculated for each particular immunoglobulin
isotype. Data are plotted as the percent represented by each isotype
2SE within the total pool of IgG immunoglobulin.
Fig. 5. HPTLC immune staining of sera of mice immunized with
GM3/VSSP vaccine (B) and NGcGM3/VSSP vaccine (C) in
Montanide ISA 51. Plate A was stained with resorcinol.
Fig. 6. HPTLC immune staining of sera of mice immunized with
GD3/VSSP vaccine (B) in Montanide ISA 51. Plate A was stained
Fig. 7. Immunogenicity of vaccine GM3/VSSP/ Montanide ISA 51
in monkeys. Open symbols represent anti-GM3 IgM and closed sym-
bols anti-GM3 IgG. each symbol represents an individual animal;
arrows, immunization days.
F. Estevez et al. / Vaccine 18 (2000) 190±197 195
arations of the outer membrane proteins of N. menin-
gitidis [27,28]. The physicochemical properties of
VSSP, however, dier from other proteosome-based
preparations described before . Ganglioside incor-
poration into the OMPC leads to a dramatic redu ction
in the size of the complex that, unlike other proteo-
some-based preparations, is not visible by electron mi-
croscopy. On the other hand, the relatively high
density of VSSP (1.265 g/ml) reveals that VSSP are
non-vesicular proteo-lipidic bodies from which most
water has been excluded.
The basis for this compact structure might be
explained if we consider the molecular properties of
gangliosides. These glycosphingolipids have very large
oligosaccharide moieties relative to their hy drocarbon
portions. For reasons of molecular geometry, these
structures are not thermodynamically compatible with
the formation of bilayer vesicles but with micelles .
The presence of high concentrations of gangliosides in
a solution of membrane proteins, high ly dissociated by
the presence of anionic detergents, leads to the for-
mation of mixed micelles from which most water is
excluded. After detergent removal, the system should
reach a new equilibrium in which the ganglioside prop-
Our study on the immunogenicity of VSSP con®rms
and extends a previous repo rt , and shows that not
only IgM but also high-level IgG anti-ganglioside anti-
bodies can be induced in the mice. However, in ad-
dition to GD3 ganglioside, which is considered the
most immunogenic in this animal model , highly
expressed gangliosides such as GM3 and NGcGM3
were also highly immunogenic when used as VSSP.
Similar results were obtained in the other two animal
species studied. Unexpectedly, the anti-ganglioside IgG
fraction in mice was not restricted to the typical T-
independent isotype IgG3. High, levels of the Th1-re-
lated IgG istotypes IgG1, IgG2a and particularly
IgG2b were also found. These phenomena, however,
were strongly dependent on the use of an oil-in-water
adjuvant. VSSP overcame natural tolerance to ganglio-
side in an adjuvant- dependent fashion.
The mechanisms involved in VSSP immuno-enhan-
cing properties are not yet understood. The observed
eect is not inherent to the adjuvant itself since in pre-
vious experiences GM3/VLDL and NGcGM3/VLDL
conjugates administered either in CFA or in
Montanide ISA 51 were not immunogenic in mice. It
is known that serotype proteins, which are the main
components of the OMPC, induce proliferation and
activation of lymphocytes and lead to secretion of IL-2
by a mechanism dierent to that of polyc lonal activa-
tors such as LPS . On the other hand, the adjuvant
activity of oil-in-water emulsions has been suggested to
be linked to their interaction with dendritic cells 
which, at the same time, have been implicated in the
presentation of non-classical antigens that are very clo-
sely related to gangliosides . The xenogenic en-
vironment created by the bacterial proteins and some
residual lipids that, like LPS, are natural activators of
the immune system, together with the very small size
of the complexes, might facilitate ganglioside uptake
by this subset of antigen-presenting cells that triggers
the maturation of an incipient immune response.
No reactogenicity was observed even when VSSP of
self-gangliosides such as GM3 were administered in
Montanide ISA in any of the three animal species
used. These gangliosides, as has been suggested before
, should remai n `cryptic' in normal tissues, out of
reach of circul ating antibodies.
Based on these observations and on our recent ®nd-
ing that threshold ino cula of syngeneic melanoma B16
tumor cells in mice immunized with GM3-containing
VSSP leads to a very high tumor rejection rate and
increased survival (submitted for publication), we have
started two clinical trials, following international
guidelines . Patients suering metastatic breast can-
cer have been immunized with GM3/VSSP and
NGcGM3/VSSP vaccines in Montanide ISA 51 to
evaluate the immunogenici ty and the therapeutic po-
tential of these preparations.
The authors thank Miss Marilyn Clavell and Mr.
Dariel Morales for expert technical assistance, Ms.
Amada Valdez for laboratory assistance, Dr. Ester M.
Fajardo for help with the ELISA techni que and Dr.
Philip O. Livingston for his kindness in providing us
with the R24 MAb and his words of encouragement.
 Hakomori SI. Tumor malignancy de®ned by aberrant glycosyla-
tion and sphingo(glyco)lipid metabolism. Cancer Res
 Morton DL, Ravindranath MH, Irie RF. Tumor gangliosides
as targets for active speci®c immunotherapy of melanoma in
man. In: Svennerholm L, Asbury AK, Reisfeld RA, Sandho
K, Suzuki K, Tettamanti G, Toano G, editors. Progress in
Brain Research. Amsterdam: Elsevier Science, 1994. p. 251.
 Portoukalian J, David MJ, Gain P, Richard M. Shedding of
GD2 ganglioside in patients with neuroblastoma. Int J Cancer
 Takahashi K, Ono K, Hirabayashi Y, Taniguchi M. Escape
mechanisms of melanoma from immune system by soluble mela-
noma antigens. J Immunol 1988;140:3244±8.
 Portoukalian J, Carrel S, Dore JF, Rumke P. Humoral immune
response in disease-free advanced melanoma patients after vacci-
nation with melanoma-associated gangliosides. Int J Cancer
F. Estevez et al. / Vaccine 18 (2000) 190±197196
 Zhang H, Zhang S, Cheung NV, Ragupathi G, Livingston PO.
Antibodies against GD2 ganglioside can eradicate syngeneic
cancer micrometastases. Cancer Res 1998;58:2844±9.
 Dohi T, Nores G, Hakomori S. An IgG3 monoclonal antibody
established after immunisation with GM3 Lactone: immuno-
chemical speci®city and inhibition of melanoma cell growth in
vitro and in vivo. Cancer Res 1988;48:5680±5.
 Livingston PO, Zhang S, Lloyd KO. Carbohydrate vaccines
that induce antibodies against cancer. 1. Rationale. Cancer
Immunol Immunother 1997;45:1±9.
 Ravindranath MH, Brazeau SM, Morton DL. Ecacy of
tumor cell vaccine after incorporating monophosphoryl lipid A
(MPL) in tumor cell membranes containing tumor-associated
ganglioside. Experientia 1994;50:648±53.
 Livingston PO. Approaches to augmenting the immunogenicity
of melanoma gangliosides: From whole cells to ganglioside-
KLH conjugate vaccines. Immunol Rev 1995;145:148±66.
 Dumontet C, Rebbaa A, Portoukalian J. Very low density lipo-
proteins and interleukin 2 enhance the immunogenicity of 9-O-
acetyl-GD3 ganglioside in BALB/c mice. J Immunol Methods
 Livingston PO, Calves MJ, Helling F, Zollinger WD, Blake
MS, Lowell GH. GD3/proteosome vaccines induce consistent
IgM antibodies against the ganglioside GD3. Vaccine
 Livingston PO, Ritter G, Calves MJ. Antibody response after
immunisation with the gangliosides GM1 GM2, GM3, GD2
and GD3 in the mouse. Cancer Immunol Immunother
 Folch PJ, Arsove S, Meath JA. Isolation of brain strandin, a
new type of large molecular tissue component. J Biol Chem
 Takamizawa K, Iwamori M, Mutai M, Nagai Y. Gangliosides
of bovine buttermilk. J Biol Chemistry 1986;261:5625±30.
 Svennerholm L, Fredman P. A procedure for the quantitative
isolation of brain gangliosides. Biochim Biophys Acta
 Hashimoto Y, Otsuka H, Sudo K, Suzuki A, Yamakawa T.
Genetic regulation of GM2 expression in liver of the mouse. J
 Gazzotti G, Sonnino S, Ghidoni R. Normal-phase high-per-
formance liquid chromatographic separation of non-derivatized
ganglioside mixtures. J Chrom 1985;348:371±8.
 Ahlers JD, Dunlop N, Pendlenton CH, Newman M, Nara PL,
Berzofky JA. Candidate HIV type 1 multideterminant cluster
peptide-P18MN vaccine constructs elicit type 1 helper T cells,
cytotoxic T cells, and neutralising antibody, all using the same
adjuvant immunisation. AIDS Res Hum Retroviruses
 Campa C, Sierra GV, Gutierrez MM, Garcia L, Puentes GC,
Sampedro MC, Sotolongo F, Le Riverend EX, Galguera MA.
Vaccine against group B N. meningitidis, gammaglobulin and
transfer factor. European patent 0 301 992, US patent
 Markwell MA, Haas SM, Bieber LL, Tolbert NE. A modi®-
cation of the Lowry procedure to simplify protein determination
in membrane and lipoprotein samples. Anal Biochem
 Suennerholm L. Quantitative estimation of sialic acids. II A col-
orimetric resorcinol-hydrochloric acid method. Biochem
Biophys Acta 1957;24:604±11.
 Sofer G. Gel ®ltration. In: Sofer GK, Nystrom LE, editors.
Process Chromatography, A Practical Guide. New York:
Academic Press, Harcourt Brace Jovanovich Publishers, 1989.
 Magnani JF, Smith DF, Ginsburg V. Detection of gangliosides
that bind cholera toxin: direct binding of 125I-labeled toxin on
thin layer chromatograms. Anal Biochem 1980;109:399±402.
 Leeden RW, Yu RK. Chemistry and analysis of sialic acid. In:
Rosemberg A, Schengtrund C-L, editors. Biological Role of
Sialic Acid. New York: Plenum Press, 1976. p. 1.
 Kawai T, Kato A, Higashi S, Kato S, Naiki M. Quantitative
determination of N-glycolylneuraminic acid expression in
human cancerous tissues and avian lymphoma cell lines as a
tumorassociated sialic acid by gas chromatography-mass specto-
metry. Cancer Res 1991;51:1242±6.
 Lowell GH. Proteosomes, hydorohpobic anchors, iscoms and
liposomes for improved presentation of peptides and protein
vaccines. In: Woodrow GC, Levine MM, editors. New
Generation Vaccines. New York: Marcel Dekker, 1990. p. 141±
 Lowell GH, Ripley Ballou W, Smith LF, Wirtz RA, Zollinger
WD, Hockmeyer WT. Proteosome-lipopeptide vaccines:
Enhancement of immunogenicity for malaria CS peptides.
 Ruegg CL, Jae RI, Koster B, Sado JC, Zollinger WD,
Lowell GH, Strand M. Preparation of proteosome-based vac-
cines. Correlation of immunogenicity with physical character-
istics. J Immunol Methods 1990;135:101±9.
 Maggio B. The surface behavior of glycosphingolipids in bio-
membranes: a new frontier of molecular ecology. Prog Biophys
Molec Biol 1994;62:55±117.
 Liu MA, Friedman A, Oli AI, Tai J, Martinez D, Deck RR,
Shieh JT, Jenkins TD, Donnelly JJ, Hawe LA. A vaccine carrier
derived from N. meningitidis with mitogenic activity for lym-
phocytes. Proc Natl Acad Sci USA 1992;89:4633±7.
 Ott G. The adjuvant MF59: the 1998 perspective, clinical per-
formance and mechanism of action. Res Immunol 1998;149:25±
 Kawano T, Cui J, Koezuka Y, Toura I, Kaneko Y, Motoki K,
Ueno H, Nakagawa R, Sato H, Kondo E, Koseki H, Taniguchi
M. CD1d-restricted and TCR-mediated activation of Va 14
NKT cells by glycosylceramides. Science 1997;278:1626±9.
 Nores GA, Dohi T, Taniguchi M, Hakomori S. Density-depen-
dent recognition of cell surface GM3 by a certain anti-mela-
noma antibody, and GM3 lactone as a possible immunogen:
requirements for tumor-associated antigen and immunogen. J
 International Ethical Guidelines for Biomedical Research
Involving Human Subjects, 1993.
F. Estevez et al. / Vaccine 18 (2000) 190±197 197