A Cellular Deficiency of Gangliosides Causes Hypersensitivity to
Clostridium perfringens Phospholipase C*
Received for publication, January 10, 2005, and in revised form, May 26, 2005
Published, JBC Papers in Press, May 26, 2005, DOI 10.1074/jbc.M500278200
Marietta Flores-Dı´az‡§, Alberto Alape-Giro´n‡§¶
, Graeme Clark**, Bruno Catimel‡‡,
Yoshio Hirabayashi§§, Ed Nice‡‡, Jose´-Marı´a Gutie´rrez§, Richard Titball**,
and Monica Thelestam‡
From the ‡Microbiology and Tumor Biology Center, Karolinska Institutet, Stockholm S-17177, Sweden, §Instituto
Clodomiro Picado, Facultad de Microbiologı´a, and ¶Departamento de Bioquı´mica, Facultad de Medicina, Universidad de
Costa Rica, San Jose´, 2060 Costa Rica, **Defence Science and Technology Laboratory, Porton Down, Salisbury, Wilts SP4
0JQ, United Kingdom,‡‡Ludwig Institute for Cancer Research, P.O. Royal Melbourne Hospital, Parkville, Melbourne,
Victoria 3050, Australia, and §§Institute of Physical and Chemical Research (RIKEN) 2-1 Hirosawa,
Wako-shi, Saitama 351-01, Japan
Clostridium perfringens phospholipase C (Cp-PLC),
also called 
-toxin, is the major virulence factor in the
pathogenesis of gas gangrene. Previously, a cellular
UDP-Glc deficiency was related with a hypersensitivity
to the cytotoxic effect of Cp-PLC. Because UDP-Glc is
required in the synthesis of proteoglycans,N-linked gly-
coproteins, and glycosphingolipids, the role of these gly-
coconjugates in the cellular sensitivity to Cp-PLC was
studied. The cellular sensitivity to Cp-PLC was signifi-
cantly enhanced by glycosphingolipid synthesis inhibi-
tors, and a mutant cell line deficient in gangliosides was
found to be hypersensitive to Cp-PLC. Gangliosides pro-
tected hypersensitive cells from the cytotoxic effect of
Cp-PLC and prevented its membrane-disrupting effect
on artificial membranes. Removal of sialic acids by
C. perfringens sialidase increases the sensitivity of cul-
tured cells to Cp-PLC and intramuscular co-injection of
C. perfringens sialidase, and Cp-PLC in mice potentiates
the myotoxic effect of the latter. This work demon-
strated that a reduction in gangliosides renders cells
more susceptible to the membrane damage caused by
Cp-PLC and revealed a previously unrecognized syner-
gism between Cp-PLC and C. perfringens sialidase, pro-
viding new insights toward understanding the patho-
genesis of clostridial myonecrosis.
Bacterial phospholipases C (PLCs)1 (EC 3.1.4.3) are secreted
proteins that display a variable preference for different phos-
pholipids (1). Bacterial PLCs that hydrolyze sphingomyelin
(SM) and phosphatidylcholine (PC) are important virulence
factors in the pathogenesis of diseases caused by Listeria mono-
cytogenes, Pseudomonas aeruginosa, and several Clostridium
sp. (2–4). Clostridium perfringens, the pathogenic bacterium
most widely distributed in nature (5), produces one phospho-
lipase C (referred to here as Cp-PLC), which is highly toxic; it
decreases cardiac contractility, increases capillary permeabil-
ity, and displays platelet-aggregating, hemolytic, cytotoxic, and
myotoxic activities (6). Cp-PLC is also called 
-toxin and has
been associated with enteritis in domestic animals, Crohn dis-
ease, and gas gangrene in humans (6).
Gas gangrene is an acute and life-threatening infection most
frequently caused by C. perfringens and characterized by mas-
sive local edema, severe myonecrosis, and the accumulation of
gas at the site of infection (7). Gas gangrene without external
injury is associated with diabetes mellitus, peripheral vascular
disease, or an underlying gastrointestinal or hematologic ma-
lignancy (8). More often, gas gangrene is associated either to
trauma or surgery, occurring after the introduction of bacterial
spores in a deep lesion or in a surgical wound (7). Despite the
use of antibiotics and intensive care regimes, in many cases of
gas gangrene radical amputation is the treatment of choice to
avoid shock, multiorgan failure, and death (7). The plc gene,
encoded in the C. perfringens chromosome, shows minimal
inter-strain sequence variations and is highly expressed in all
strains associated with gas gangrene (6, 9). Homologous genes
are found in other Clostridium species such as Clostridium
sordellii, Clostridium absonum, and Clostridium novyi, which
occasionally cause gas gangrene (10, 11).
Several lines of evidence indicate that Cp-PLC is the major
virulence factor in C. perfringens-induced gas gangrene. First,
when injected intramuscularly in mice, recombinant Cp-PLC
causes myonecrosis and reproduces the histopathological fea-
tures of gas gangrene (12). Second, immunization with the
recombinant Cp-PLC C-terminal domain protects mice from
the intramuscular challenge with a lethal dose of C. perfringens
vegetative cells (13). Third, a C. perfringens mutant strain, in
which the plc gene has been inactivated by homologous recom-
bination, is unable to produce gas gangrene (14). Although
* This work was supported by Costa Rica-United States Foundation
Grant CT13-02, Swedish Cancer Society Grant 3826-B96-01XAB, Con-
sejo Nacional de Investigaciones Cientificas y Tecnologicas, Vicerrecto-
ria de Investigacio´n, Universidad de Costa Rica Grants 741-98-287,
741-A0-510, and 741-A3-503, Swedish Research Council Grant 16X-
05969, and Karolinska Institutet Research Funds. The costs of publi-
cation of this article were defrayed in part by the payment of page
charges. This article must therefore be hereby marked “advertisement”
in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

 To whom correspondence should be addressed: Instituto Clodomiro
Picado, Facultad de Microbiologı´a, Universidad de Costa Rica, 2060 San
Jose´, Costa Rica. Tel.: 506-2290344 or 506-2293135; Fax: 506-2920485;
E-mail: aalape@cariari.ucr.ac.cr.
1 The abbreviations used are: PLCs, phospholipases C; SM, sphingo-
myelin; PC, phosphatidylcholine; Cp-PLC, C. perfringens PLC; UDPG:
PP, UDP-Glc pyrophosphorylase; ER, endoplasmic reticulum; GSL(s),
glycosphingolipid(s); GlcCer, glucosylceramide; GalCer, galactosylcer-
amide; SiaAc, sialic acid; C-PLC, C. perfringens PLC C-terminal do-
main; B. asper MT II, B. asper myotoxin II; mAb(s), monoclonal anti-
bodies; L-cycloserine, L- 4-amino-3-isoxazolidinone; PPMP, DL-threo-1-
phenyl-2-palmitoylamino-3-morpholino-1-propanol; GlcUAT-I, UDP-
GlcUA:Gal1,3Gal-R glucuronosyltransferase I; UDPG:CGT, UDP-Glc:
ceramide glucosyltransferase; FBS, fetal bovine serum; PBS,
phosphate-buffered saline; HPTLC, high performance thin layer chro-
matography; LDH, lactate dehydrogenase; CK, creatine kinase; PKC(s),
protein kinase(s) C; wt, wild type; CHO, Chinese hamster ovary; RU,
resonance units.
THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 280, No. 29, Issue of July 22, pp. 26680–26689, 2005
© 2005 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.
This paper is available on line at http://www.jbc.org26680
considerable progress has been made in recent years in the
knowledge of the structure of Cp-PLC, our understanding of
the cellular and molecular basis of its toxic effects is still
incomplete (15).
A mutant cell line hypersensitive to Cp-PLC, referred to here
as Don Q, was isolated previously (16). This cell line has a
permanent low UDP-glucose (UDP-Glc) level due to a recessive
point mutation in the UDP-Glc pyrophosphorylase (UDPG:PP)
gene (17). We showed previously (18) that upon one allele
reversion of the mutation or transfection with a wild type
UDPG:PP, the mutant cell compensates the UDP-Glc defi-
ciency and regains the same relative resistance to Cp-PLC as
the one displayed by the parental cell. However, the molecular
basis relating the cellular UDP-Glc deficiency to the sensitiza-
tion to Cp-PLC has not been established.
UDP-Glc is required for the synthesis of the carbohydrate
moiety of cell surface glycoconjugates, such as proteoglycans,
N-glycosylated glycoproteins, and glycosphingolipids (GSLs) (17).
Proteoglycans consist of variable polysaccharide chains, the gly-
cosaminoglycans, linked to a core polypeptide. The mature gly-
cosaminoglycans contain glucosamine or galactosamine alternat-
ing in glycosidic linkages with glucuronic acid, iduronic acid, or
galactose and are sulfated or acetylated to varying degrees (19).
The glycosaminoglycans are assembled in the lumen of the Golgi
apparatus by sequential addition of monosaccharides to the
linker tetrasaccharide GlcUA1,3Gal1,3Gal1,4Xyl-, which is
bound to a Ser or a Thr residue of the core polypeptide (20).
UDP-Glc is essential for the synthesis of UDP-glucuronic acid
and UDP-galactose, two monosaccharide donors used in glyco-
saminoglycans synthesis.
N-Glycosylated glycoproteins have covalently bound to an
Asn residue, oligosaccharides of variable composition, all of
which arise from the common core precursor Glc3Man9GlcNAc2-
PP-dolichol (21). The oligosaccharide moiety of this precursor is
synthesized by the stepwise addition of monosaccharide units
onto dolichol pyrophosphate on the endoplasmic reticulum (ER)
membrane and is then transferred en bloc onto the nascent
polypeptide chains (21). UDP-Glc is actively transported to the
lumen of the ER, where it serves as donor of the three terminal
glucoses of the core precursor (21). Within the ER, UDP-Glc is
also used by the UDP-Glc:glycoprotein glucosyltransferase,
which recognizes and glucosylates misfolded glycoproteins (21).
This glucosylation prolongs the contact of N-linked glycopro-
teins with the lectin-type chaperones calnexin and calreticulin
and thus leads to their retention within the ER until they are
correctly folded (21).
GSLs are composed of a variable carbohydrate moiety linked
to ceramide and derive either from glucosylceramide (GlcCer)
or galactosylceramide (GalCer) (22). GalCer is synthesized
from ceramide and UDP-galactose on the luminal face of the
ER membrane, whereas GlcCer is synthesized from ceramide
and UDP-Glc on a pre-Golgi compartment or on the cytosolic
face of the Golgi apparatus (23). Ceramide, a common precur-
sor of GSLs and SM, is either synthesized de novo from L-serine
and palmitoyl-CoA (22) or by a salvage pathway from sphingo-
sine (23) (Fig. 1). The de novo synthesis of ceramide involves
serine palmitoyltransferase, 3-ketosphinganine reductase,
sphinganine/sphingosine N-acyltransferase, and dihydrocer-
amide desaturase (22), whereas the salvage pathway requires
only the N-acyltransferase (23) (Fig. 1). In addition, GSLs from
the extracellular space are incorporated in the plasma mem-
brane and internalized along the lysosomal pathway (23). In
mammalian cells, the most abundant derivatives of GlcCer are
gangliosides, for which lactosylceramide serves as a precursor
(Fig. 1). Gangliosides may have either one (a series), two (b
series), or three (c series) sialic acid (SiaAc) residues linked to
the 3-position of the inner galactose moiety or may lack SiaAc
at this position (0 series) (23). Complex gangliosides have an-
other SiaAc moiety linked to the 3-position of the terminal
galactose (23).
Because a UDP-Glc deficiency leads to a cellular hypersen-
sitivity to Cp-PLC (18) and this metabolite is essential in the
synthesis of proteoglycans, glycoproteins, and GSLs, the aim of
this work was to clarify the role of these glycoconjugates in the
cellular sensitivity to this toxin.
EXPERIMENTAL PROCEDURES
Toxins, Antibodies, and Chemicals—Wild type recombinant Cp-PLC
from the strain NCTC 8237 or its C-terminal fragment (C-PLC) were
expressed in Escherichia coli and purified as described (24). Bacillus
cereus PLC, 1-deoxynojirimycin, castanospermine, and the ganglioside
GM3 were from Calbiochem. Bothrops asper myotoxin II (B. asper MT
II) was purified as described (25). L-4-Amino-3-isoxazolidinone (L-cyclo-
serine) and DL-threo-1-phenyl-2-palmitoylamino-3-morpholino-1-propa-
nol (PPMP) were from Biomol. The ganglioside-specific monoclonal
antibodies (mAbs) DH2, KM871, 101, and 102 were produced and
purified as described (26–28). The anti-glucosylceramide rabbit anti-
serum was from Glycobiotech GmbH (Ku¨kels, Germany); B. cereus
sphingomyelinase, C. perfringens, and Vibrio cholerae sialidases, re-
combinant endoglycosidase H from Streptomyces plicatus, endoglycosi-
dase F1 from Chryseobacterium meningosepticum, and all other chem-
icals were from Sigma.
Cells and Cell Culture—Chinese hamster lung fibroblasts, referred
to here as Don wt, mouse melanoma cells (B16), Chinese hamster ovary
cells (CHO-K1), and human larynx carcinoma cells (HEp-2) were from
ATCC. The UDP-Glc-deficient mutant of Don wt cells, Don Q, which has
a defective UDPG:PP, the spontaneous revertant of this mutant, Don
QR, and the UDPG:PP transfectant clone. B9 was described previously
(18). A proteoglycan-deficient mutant of CHO cells, referred to here as
pgsG-110, which has a defective UDP-GlcUA:Gal1,3Gal-R glucurono-
syltransferase I (GlcUAT-I), and the GlcUAT-I transfectant clone pgs-T
FIG. 1. Metabolic pathway of glycosphingolipid synthesis and
sites susceptible to pharmacologic inhibition. The inhibitors of
the pathway used in this study are indicated in italics.
Ganglioside Reduction Causes 
-Toxin Hypersensitivity 26681
were described previously (20). A ganglioside-deficient mutant of B16
cells, referred to here as GM95, which has a defective UDP-Glc ceram-
ide glucosyltransferase (UDPG:CGT, EC 2.4.1.80), and the UDPG:CGT
transfectant clone CG-1 were described previously (29). Don wt, Q, QR,
B9, B16, GM95, and CG-1 cells were cultivated at 37 °C in Eagle’s
minimum essential medium (Invitrogen) supplemented with 10% fetal
bovine serum (FBS), 5 mM L-glutamine, penicillin (100 units/ml), and
streptomycin (100 g/ml) in a humid atmosphere containing 5% CO2.
CHO, pgs-110, and pgs-T cells were grown in Ham’s F-12 medium
(Invitrogen), supplemented as above.
Cytotoxicity Assays—Cells in 96-well plates grown to 80–90% of con-
fluence were exposed to Cp-PLC or to wheat germ lectin in 200 l of
medium/well, and cell viability was assessed 24 h later using a neutral red
assay (18). Briefly, the cells were incubated for 2 h with 200 l/well of
neutral red (50 g/ml) dissolved in supplemented culture medium, and
the incorporated dye was extracted with 100 l of acetic acid/ethanol/
water (1:50:49), before recording the absorbances at 540 nm. Cell survival
was expressed as a percentage, considering as 100% the value of parallel
cultures incubated with only medium or with medium and drugs at the
indicated concentrations. Assays were performed with 3–6 replicate sam-
ples. When not shown, the standard deviation was 
10%.
Pharmacological or Enzymatic Treatments and Evaluation of Their
Effects—Cells were preincubated for 18–24 h with 1-deoxynojirimycin
(1–90 g/ml), castanospermine (1–90 g/ml), swainsonine (1–90 g/ml),
brefeldin A (0.1–9 g/ml), monensin (1–90 g/ml), tunicamycin (0.1–9
g/ml), -chloroalanine (1–90 g/ml), L-cycloserine (3–300 g/ml), myri-
ocin (3–300 g/ml), fumonisin B1 (0.3–30 M), PPMP (0.3–30 g/ml),
C. perfringens large sialidase (0.3–100 milliunits/ml) or V. cholerae
sialidase (0.3–100 milliunits/ml) or for 2 h with galactocerebrosides
(90–900 g/ml), gangliosides (90–900 g/ml), in supplemented medium
containing 10% FBS, or for 18–24 h in Opti-MEM (Invitrogen) with
S. plicatus endoglycosidase H (1–9 milliunits/ml) or C. meningosepti-
cum endoglycosidase F1 (1–9 milliunits/ml).
Proteins in lysates of untreated and treated cells were submitted to
SDS-PAGE and electroblotted onto 0.45-m nitrocellulose membranes
as described (17). Carbohydrate moieties in glycoproteins were oxidized
with periodate and detected using biotin-hydrazide and streptavidin-
alkaline phosphatase with the GlycoTrack carbohydrate detection kit
(Oxford Glycosystems, Abingdon, UK).
GSLs were isolated from trypsinized cells and pelleted by centrifu-
gation at 4 °C at 300  g for 10 min. Cell pellets were lyophilized and
extracted twice with chloroform/methanol 1:1 (v/v). The combined ex-
tracts were reduced to 25% of the original volume and kept at 4 °C
overnight, and then the insoluble material was removed by centrifuga-
tion at 4 °C at 100  g for 10 min. The supernatant was dried under
nitrogen and lyophilized, and the lipids were partitioned twice in 6 ml
of diisopropyl ether/1-butanol, 60:40 (v/v), and 3 ml of 0.1% aqueous
NaCl. The aqueous phase was lyophilized and redissolved in 300 l of
distilled water. Salts were removed by gel filtration in Sephadex G-50.
HPTLC separation and immunostaining were performed as described
(30) using mAb DH2 (26) or a rabbit anti-glucosylceramide antiserum
(31). Densitometric scanning analysis was performed with a Shimadzu
CS-930 HPTLC scanner.
To evaluate ganglioside incorporation into the membrane, cells were
trypsinized, centrifuged, and washed once with PBS, and the cell pellet
was resuspended and fixed with 4% formaldehyde for 30 min at 37 °C.
After three washes with PBS, cells were incubated with 20 g/ml of the
ganglioside-specific mAbs for 1 h on ice. After three washes with PBS,
cells were incubated with fluorescein-labeled anti-mouse immunoglobu-
lins for 30 min on ice. Flow cytometry analysis was performed on a
FACSort flow cytometer (BD Biosciences). Data were collected and
analyzed using CellQuest software (BD Biosciences).
To evaluate the effects of the treatments in sensitivity to toxins, cells
were grown to 60–80% of confluence in 96-well plates and exposed to
the treatments for the mentioned periods of time before and during
exposure to serial dilutions of the toxins. Cell viability was measured
24 h later using the neutral red assay as described. At the concentra-
tions used, none of the substances by itself caused significant cytotox-
icity (viability was higher than 95% in comparison with untreated
cells). In control experiments after ganglioside addition, mild
trypsinization was carried out with 0.1% trypsin in phosphate-buffered
saline (PBS), pH 7.8, for 15 min at 37 °C.
Hydrolysis of Choline-containing Phospholipids from Prelabeled
Cells—Cells were incubated with 5 Ci/ml [methyl-3H]choline (Amer-
sham Biosciences) for 18 h, washed three times with Hanks’ balanced
salt solution, and exposed to Cp-PLC (20 ng/ml) in supplemented me-
dium for 0.5, 1.5, 2, 4, and 6 h. The amount of radioactive choline
incorporated into membranes and the choline remaining after Cp-PLC
treatment were determined by counting the radioactivity in the lipid
phase after extraction with chloroform/methanol (18). The released
counts/min were measured and related either to the amount of cellular
protein determined in parallel cultures using the Bio-Rad DC protein
assay or to the total counts/min incorporated into the membranes.
Release of Lactate Dehydrogenase—The lactate dehydrogenase
(LDH, EC 1.1.1.27) released to the medium after exposure of cells to
toxins for 2 h was determined by using a kit from Roche Applied
Science. LDH release was expressed as a percentage, considering as 100
and 0% the values of parallel cultures exposed to 0.1% Triton X-100 or
medium only. Assays were performed with 3–4 replicate samples.
Myotoxicity Studies—Myotoxicity was estimated by measurement of
the creatine kinase (CK) activity in the plasma after intramuscular
injection of the toxin in the right gastrocnemius in groups of five CD-1
mice, as described (24). Animals were housed, fed, and handled accord-
ing to the principles and practices approved by the Institutional Com-
mittee for Care and Handling of Experimental Animals of Universidad
de Costa Rica. The CK activity in plasma was determined using a
kinetic assay (Sigma, CK-10) 3 h after injection of 0.7 g of Cp-PLC
alone or together with different amounts of C. perfringens sialidase.
Membrane-disrupting Effect in Liposomes—Unilamellar liposomes
composed of cholesterol and either PC or SM (in 1:1 molar ratio) encap-
sulating carboxyfluorescein were prepared as described (32). In some
cases purified gangliosides were incorporated into the liposomes at
concentrations of 100, 33, and 11 g/ml (which would correspond to
PC/ganglioside or SM/ganglioside molar ratios of about 30:1; 90:1,
and 270:1). After a 20-min exposure to Cp-PLC (20 g/ml), fluores-
cence at 520 nm (excitation at 485 nm) was measured using a Fluo-
roskan Ascent fluorimeter (PerkinElmer Life Sciences). Samples
were assayed in triplicate. Carboxyfluorescein release was expressed
as a percentage, considering as 100 and 0% the values of liposomes
exposed for 1 h to 1% Triton X-100 at 37 °C or to borate saline buffer,
pH 7.6, respectively.
Biosensor Studies—Analysis was performed using a BIAcore 2000
biosensor and the carboxymethyl-dextran-coated sensor chip, CM5, or
the lipophilic pioneer L1 sensor chip (Pharmacia Biosensor, Uppsala,
Sweden). GT1b ganglioside was immobilized onto a CM5 surface as
described previously (27). Briefly, ganglioside GT1b was dissolved in
50% methanol/ethanol at a final concentration of 1 mg/ml, and 50 l of
this solution was diluted (1:3 v/v) in running buffer (10 mM Hepes, pH
7.4, 150 mM NaCl, 2 mM CaCl2) and injected at a flow rate of 5 l/min
over the unmodified surface. The surface was then washed with 10 mM
NaOH until a stable base line was obtained (27). Approximately 0.12
ng/mm2 (120 RU) was immobilized onto the sensor surface. For immo-
bilization onto the L1 chip, GT1b was incorporated in PC vesicles (100
g of GT1b/1 mg of PC), and the surface was then washed with 10 mM
NaOH to remove unattached vesicles. Approximately 3.0 ng/mm2 (3028
RU) was immobilized onto the sensor surface, whereas PC vesicles
(3.2 ng/mm2, 3254 RU) were immobilized onto the control surface.
mAbs, Cp-PLC, or C-PLC were buffer-exchanged into the biosensor
running buffer using a Fast desalt column (PC 3.2/10) connected to a
SMART system (Amersham Biosciences). Anti-GT1b ganglioside IgG
mAbs 101 and 102 (28) and the anti-GD3 ganglioside IgG mAb KM871
(27) were injected (30 l, 165 nM) at a flow rate of 10 l/min over
immobilized GT1b ganglioside to confirm surface specificity. Cp-PLC or
C-PLC were also injected (30 l and 2.35 and 3.5 mM, respectively) at a
flow rate of 10 l/min over immobilized GT1b ganglioside. The sensor-
grams shown have been subtracted with the corresponding signal ob-
tained when the sample was passed over a nonderivatized CM5 surface
or over the PC control surface (L1 chip).
RESULTS
Inhibition of Proteoglycan Synthesis Does Not Affect Cellular
Sensitivity to Cp-PLC—To evaluate the role of proteoglycans in
the cell sensitivity to Cp-PLC, the following two cell lines
having extreme differences in their proteoglycan content due to
a genetic reason were used: the proteoglycan-deficient mutant
cell line, pgsG-110, which has an impaired activity of the en-
zyme GlcUAT-I, that catalyzes the first step in the synthesis of
the tetrasaccharide primer of glycosaminoglycans (20), and the
clone pgs-T, which is a stable transfectant having a wild type
GlcUAT-I gene. Remarkably, these two cell lines exhibited the
same sensitivity to Cp-PLC as the wild type CHO cell (Fig. 2),
which conclusively demonstrates that a deficiency in proteogly-
cans does not sensitize cells to Cp-PLC.
Ganglioside Reduction Causes 
-Toxin Hypersensitivity26682
Inhibition of N-Linked Glycoprotein Processing Does Not In-
crease the Cellular Sensitivity to Cp-PLC—We assessed the role
of N-linked glycoproteins in the cellular sensitivity to Cp-PLC
by treating the cells prior to and during the toxin exposure with
substances that interfere with their processing or transport to
the surface. These substances lead to the production of glyco-
proteins with under-processed or incomplete oligosaccharide
chains and thus alter the cellular glycoprotein pattern (33, 34).
Deoxynojirimycin, an inhibitor of glucosidase I, and castano-
spermine, an inhibitor of glucosidases I and II, prevent removal
of the glucose residues from the oligosaccharide core thereby
inhibiting further processing of the oligosaccharide chain (33).
Swainsonine affects a later step of glycoprotein processing by
inhibiting the Golgi 
-mannosidase II and thus preventing
further glycoprotein processing at the Golgi apparatus (33).
Brefeldin A and monensin interfere with the organization of
the Golgi apparatus and thereby interrupt the trafficking of
glycoproteins through this compartment (34). As reported for
other cell lines, growth of Don wt cells in the presence of
deoxynojirimycin, castanospermine, swainsonine, brefeldin A,
or monensin altered the glycoprotein pattern (not shown).
However, none of these treatments increase the sensitivity of
Don wt cells to Cp-PLC (Fig. 3A, bars 1–5), in comparison with
untreated cells. Consistently, the sensitivity of Don wt cells to
the cytotoxic effect of Cp-PLC was not affected when the cells
were pretreated neither with endoglycosidase H, which cleaves
high mannose and hybrid N-linked oligosaccharides from gly-
coproteins, nor with endoglycosidase F, which cleaves high
mannose, hybrid, and biantennary complex N-linked oligosac-
charides (Fig. 3A, bars 6 and 7). In contrast, the exposure to the
same inhibitors of glycoprotein processing/transport or en-
doglycosidases changed significantly the cellular sensitivity to
the cytotoxic effect of wheat germ lectin (Fig. 3B), which re-
quires binding to cell surface glycoproteins in order to cause
cytotoxicity (35). These results corroborate that the treatments
with the inhibitors or the enzymes had the predicted effect and
indicate that a reduction in cell surface N-linked glycoproteins
does not increase the cellular sensitivity to Cp-PLC. However,
pretreatment of Don wt with tunicamycin, an inhibitor of the
N-acetylglucosamine phosphotransferase, which blocks the
synthesis of all N-linked glycoproteins (21), was found to in-
crease the cellular sensitivity to both Cp-PLC and wheat germ
lectin, in comparison with control cells (Fig. 3, A and B, bars 8).
Because tunicamycin also blocks the synthesis of GSLs (36–
38), we were prompted to study the effects of specific inhibitors
of GSLs synthesis in the cellular sensitivity to Cp-PLC.
A Specific Reduction in Cellular Gangliosides Increases the
Sensitivity to the Cytotoxic Effect of Cp-PLC—We initially as-
sessed the role of GSLs in the cellular sensitivity to Cp-PLC by
treating the cells prior to and during the toxin exposure with
various inhibitors of the enzymes involved in GSLs synthesis
(see Fig. 1). Myriocin, L-cycloserine, and -chloroalanine inhibit
the serine palmitoyltransferase, blocking the pathway of cer-
amide synthesis de novo (22). Fumonisin B1 inhibits the sphin-
ganine/sphingosine fatty-acyl-CoA N-acyltransferase, blocking
both ceramide synthesis de novo and the salvage pathway (22),
whereas PPMP inhibits UDPG:CGT, the enzyme that catalyzes
glucosylceramide synthesis (39). The treatment of Don cells
with these substances reduces the synthesis of glucosylcera-
mide and GM3 to 20–35% in comparison with untreated cells
(not shown) and all increased significantly their sensitivity to
Cp-PLC (Fig. 4A). In contrast, none of these GSLs synthesis
inhibitors affected the cellular sensitivity to the cytotoxic effect
of the B. cereus PLC (not shown). Similar results were obtained
with the same treatments in the cell lines CHO and HEp-2 (not
shown). These results suggest that the reduction of glucosyl-
ceramide-derived GSLs specifically sensitizes different cell
lines to Cp-PLC. To substantiate this conclusion further, we
compared the sensitivity to Cp-PLC of two cell lines having
extreme differences in their GSLs content due to a genetic
reason. We used the ganglioside-deficient mouse melanoma cell
line GM95, which has an impaired activity of UDPG:CGT (29),
and the clone CG-1, which is a stable transfectant having a
wild type UDPG:CGT and has the same ganglioside content as
the parental cell (29). Remarkably, the ganglioside-deficient
cell GM95 was found to be 105 times more sensitive to Cp-PLC
than the transfectant clone CG-1, which is as resistant to the
cytotoxic effect of the Cp-PLC as the parental B16 cell (Fig. 4B).
In contrast, GM95, CG-1, and B16 exhibited the same relative
resistance to the B. cereus PLC (not shown). These results
conclusively demonstrated that a cellular deficiency of glu-
cosylceramide-derived GSLs specifically causes hypersensitiv-
ity to the cytotoxic effect of Cp-PLC. Indeed, the content of the
ganglioside GM3, evaluated by HPTLC and immunostaining
with mAb DH2, was found to be similar in Don Q and GM95
cells (2.4  0.3 and 2.0  0.4 nmol/108 cells respectively) and is
significantly reduced (about 4 and 19 times, respectively) in
FIG. 2. A proteoglycan deficiency due to a genetic defect does
not sensitize cells to Cp-PLC. The proteoglycan-deficient mutant
CHO cell, pgsG-110, which has a defective GlcUAT-I (filled triangles),
the GlcUAT-I transfectant clone pgs-T (open circles), and the wild type
CHO cells () were exposed to serial 10-fold dilutions of Cp-PLC (214
g/ml). Cell viability was determined 24 h later using the neutral red
assay. The results are expressed as the percentages of neutral red
uptake in controls incubated without Cp-PLC.
FIG. 3. Specific inhibition of N-glycoprotein processing or
transport does not affect the cellular sensitivity to Cp-PLC but
changes significantly the sensitivity to wheat germ lectin. A,
Don wt cells were incubated with only medium () or pretreated with 5
mg/ml deoxynojirimycin (bar 1), 15 mg/ml castanospermine (bar 2), 3
mg/ml swainsonine (bar 3), 3 mg/ml brefeldin A (bar 4), 6 mg/ml mon-
ensin (bar 5), 3 milliunits/ml S. plicatus endoglycosidase H (bar 6), 6
milliunits/ml C. meningosepticum endoglycosidase F1 (bar 7), or 7
mg/ml tunicamycin (bar 8) before exposure to Cp-PLC (21.4 ng/ml) (A)
or to wheat germ lectin (100 mg/ml) as described under “Experimental
Procedures.” B, cell viability was determined 24 h later by using the
neutral red assay, and the results are expressed as the percentages of
neutral red uptake in controls incubated with the same treatments but
without Cp-PLC or wheat germ lectin.
Ganglioside Reduction Causes 
-Toxin Hypersensitivity 26683
comparison with that of the corresponding wild type cells Don
wt and B16.
Gangliosides Protect Hypersensitive Cells from the Cytotoxic
Effect of Cp-PLC—It has been shown that gangliosides added
to culture media are incorporated in the plasma membrane by
a wide range of cells (23). Most interestingly, Don Q or GM95
cells preincubated with a mixture of gangliosides extracted
from bovine brain were protected in a dose-dependent manner
from the cytotoxic effect of Cp-PLC (not shown). In contrast,
pretreatment of Don Q or GM95 cells with galactocerebrosides
extracted from bovine brain did not have any effect on their
sensitivity to Cp-PLC (not shown). Because exogenously added
gangliosides could attach to cells as micelles removable by
proteases (40), control experiments were performed in which
cells were treated with trypsin after ganglioside addition. How-
ever, trypsinization did not affect the protective effect of the
gangliosides (not shown). In addition, the incorporation of gan-
gliosides was verified by flow cytometry using the mAbs 101
and 102 in cells exposed to the ganglioside mixture and then
treated with trypsin (not shown). Taken together, these results
indicated that the protection from the cytotoxic effect of Cp-
PLC was exerted by ganglioside molecules inserted into the
plasma membrane.
To evaluate whether the sensitization to Cp-PLC caused by
tunicamycin was preventable by gangliosides, cells were
treated with tunicamycin in the presence of different concen-
trations of bovine brain gangliosides. This mixture fully pro-
tected Don wt cells (Fig. 5A) or CHO cells (not shown) from the
sensitization caused by tunicamycin, thus demonstrating that
the sensitization caused by tunicamycin was indeed due to the
inhibition of ganglioside synthesis rather than to the inhibition
of N-linked glycoprotein synthesis.
To determine whether protection from the cytotoxic effect of
Cp-PLC is conferred by specific type(s) of gangliosides, experi-
ments were performed adding different gangliosides to the
cells. Pretreatment of Don Q cells with the ganglioside asialo-
GM1a did not protect from the cytotoxic effect of Cp-PLC (Fig.
5B). However, pretreatment with gangliosides of the a series
(GM3, GM1a, or GD1a) or the b series (GD1b or GT1b) (Fig. 1)
increased cell resistance to Cp-PLC in a dose-dependent man-
ner (Fig. 5B). The protective effect increased with the content of
SiaAc of the ganglioside oligosaccharide chain, indicating that
multiple SiaAc units of complex gangliosides play a protective
role. Remarkably, the protective effect of GM3 and GT1b could
be reduced or even abolished if they are added in the presence
of the “C. perfringens large sialidase” (Fig. 5C), which catalyzes
the hydrolysis of 
2,3 SiaAc residues in gangliosides (41).
FIG. 4. Inhibition of glycosphingolipid synthesis sensitizes
cells to Cp-PLC. A, Don wt cells were incubated without any pretreat-
ment (bar 1), pretreated with 60 mg/ml -chloroalanine (bar 2), 60
mg/ml L-cycloserine (bar 3), 60 mg/ml myriocin (bar 4), 0.01 mM fumo-
nisin B1 (bar 5), and 6 mg/ml PPMP (bar 6), as described under “Ex-
perimental Procedures,” before exposure to Cp-PLC (21.4 ng/ml). Cell
viability was determined 24 h later using the neutral red assay. The
results are expressed as the percentages of neutral red uptake in
controls incubated with the same concentration of the inhibitors but
without Cp-PLC. B, a ganglioside deficiency due to a genetic defect
sensitizes cells to Cp-PLC. The mouse melanoma ganglioside-deficient
mutant cell line GM95, which has a deficient UDPG:CGT (filled circles),
the UDPG:CGT transfectant clone CG-1 (), and the wild type B16 cells
(open circles) were exposed to serial 10-fold dilutions of Cp-PLC (214
g/ml). Cell viability was determined 24 h later using the neutral red
assay. The results are expressed as the percentages of neutral red
uptake in controls incubated without Cp-PLC.
FIG. 5. Gangliosides protect hypersensitive cells from the cy-
totoxic effect of Cp-PLC. A, bovine brain gangliosides protect from
the sensitization to Cp-PLC caused by tunicamycin. Don wt cells were
untreated (open circles) or pretreated with tunicamycin (7 g/ml) for
24 h, incubated for 2 h without gangliosides (black circles) or with
gangliosides at 79 mg/ml (filled squares), 238 mg/ml (filled triangles), or
714 mg/ml (), and exposed to serial 10-fold dilutions of Cp-PLC (214
g/ml). Cell viability was determined 24 h later using the neutral red
assay. The results are expressed as the percentages of neutral red
uptake in controls incubated with the same concentrations of tunica-
mycin and gangliosides but without Cp-PLC. B, gangliosides protect
Don Q cells from the cytotoxic effect of Cp-PLC in a dose-dependent
manner. Don Q cells were preincubated for 2 h with the indicated
gangliosides at 18.5 g/ml (black bars), 37.5 g/ml (hatched bars), or
87.5 g/ml (white bars) and then exposed to Cp-PLC (21.4 ng/ml). Cell
viability was determined 24 h later using the neutral red assay. The
results are expressed as the percentages of neutral red uptake in
controls incubated with the same concentration of gangliosides but
without Cp-PLC. C, sialidase abolishes the capacity of gangliosides to
protect GM95 cells from the cytotoxic effect of Cp-PLC. GM95 cells were
preincubated for 24 h in Opti-MEM only (white bars) or with 87.5 g/ml
of the ganglioside GM3 (hatched bars) or GT1b (black bars) in the
absence or the presence of the indicated amounts of C. perfringens large
sialidase and then exposed to Cp-PLC (3.5 ng/ml). Cell viability was
determined 24 h later using the neutral red assay. The results are
expressed as the percentages of neutral red uptake in controls incu-
bated with the same concentration of gangliosides and sialidase but
without Cp-PLC.
Ganglioside Reduction Causes 
-Toxin Hypersensitivity26684
These results further substantiate the conclusion that the
SiaAc of these GSLs play a role in their cytoprotective effect
from Cp-PLC.
Exposure to Sialidases Potentiates the Toxicity of Cp-PLC—
The effect of a pretreatment of the target cells with sialidases
in the sensitivity to Cp-PLC was evaluated. The exposure of
Don wt or CG-1cells to the C. perfringens large sialidase did not
cause cell death by itself but increased significantly their sen-
sitivity to the cytotoxic effect of Cp-PLC in a dose-dependent
manner (Fig. 6A). In contrast, the exposure of Don wt cells to
C. perfringens large sialidase does not affect the sensitivity of
cultured cells to other cytotoxic phospholipases such as
B. cereus PLC or B. asper myotoxin II (not shown). Remark-
ably, pretreatment of the ganglioside-deficient cells Don Q with
the C. perfringens large sialidase increased their sensitivity to
Cp-PLC to a lower extent (Fig. 6B) than in their parental cells
(Fig. 6A). Similarly, pretreatment of the ganglioside-deficient
cells GM95 with the C. perfringens large sialidase sensitizes
them to Cp-PLC to a lower extent (Fig. 6B) than in their
transfected counterparts, the CG1 cells (Fig. 6A). These results
support the conclusion that the C. perfringens large sialidase
synergizes with Cp-PLC by cleaving the sialic acids of the
cellular gangliosides.
To evaluate whether a potentiation of Cp-PLC toxicity by the
C. perfringens large sialidase also would occur in vivo, both
proteins were injected separately and together intramuscularly
in mice, and myotoxicity was evaluated as the increase in
plasma CK activity. The sialidase was not myotoxic by itself at
the doses used, but increased the muscle damage caused by
Cp-PLC in a dose-dependent manner (Fig. 6C).
Cp-PLC Causes Disruption of the Cellular Membrane in Gan-
glioside-deficient Cells—We have shown previously that there
are no significant differences in the incorporation of [3H]cho-
line among cells with different UDP-Glc levels (18). We now
compared the degradation of choline-containing phospholipids
and the disruption of the cellular membrane caused by Cp-PLC
in the cell lines Don wt, Q, and QR, which differ in their cellular
UDP-Glc concentration (17). The degradation of choline-con-
taining phospholipids induced by Cp-PLC, measured as the
release of radioactive choline from prelabeled cells, occurred in
a dose- and time-dependent manner (Fig. 6, A and B). Remark-
ably, Cp-PLC induced a higher release of labeled choline from
the membrane of Don Q than from Don wt or Don QR cells,
which have partially compensated the UDP-Glc deficiency (17)
(Fig. 7, A and B). Furthermore, it also induced a higher release
of LDH from Don Q than from the UDPG:PP transfectant clone
B9 (Fig. 8A). Similarly, the release of LDH induced by Cp-PLC
from GM95 cells was higher than that induced from the UDPG:
CGT transfectant clone CG-1 (Fig. 8A). In contrast, B. asper
myotoxin II, a membrane-disrupting toxin with broad cytolytic
specificity (25), induced a similar LDH release from Don Q and
GM95 as from their transfectant counterparts (Fig. 8B). There-
fore, these data show that Cp-PLC causes extensive membrane
damage in Don Q and GM95 cells and suggest that their hy-
persensitivity to this toxin depends, at least in part, on an
increased disruption of the cellular membrane.
Gangliosides Protect PC- or SM-containing Liposomes from
the Membrane-disrupting Effect of Cp-PLC—Although Cp-PLC
has a broad spectrum of substrates (1), it only disrupts artifi-
cial membranes made of cholesterol and PC or SM, but not
those made of cholesterol and phosphatidylserine, phosphati-
FIG. 6. C. perfringens large sialidase potentiates the toxicity of
Cp-PLC. A, Don wt (black bars) or CG-1 (white bars) cells were prein-
cubated with or without C. perfringens large sialidase overnight before
exposure to Cp-PLC (21.4 ng/ml). Cell viability was determined 24 h
later using the neutral red assay. The results are expressed as the
percentages of neutral red uptake in controls incubated with the same
concentration of sialidase but without Cp-PLC. B, Don Q cells (black
bars) or GM95 cells (white bars) were preincubated with C. perfringens
large sialidase overnight before exposure to Cp-PLC (2.1 ng/ml). Cell
viability was determined 24 h later using the neutral red assay. The
results are expressed as the percentages of neutral red uptake in
controls incubated with the same concentration of sialidase but without
Cp-PLC. C, groups of five CD-1 mice were injected in the right gastro-
cnemius muscle with 35 milliunits of C. perfringens large sialidase
alone, 0.7 g of Cp-PLC alone, or with a combination of Cp-PLC and
17.5 or 35 milliunits of the sialidase, and 3 h later plasma CK activity
was measured as described under “Experimental Procedures.”
FIG. 7. Degradation of choline-containing phospholipids by
Cp-PLC occurs in a time- and dose-dependent manner. Don wt
(filled triangles), Q (empty squares), and QR cells (filled squares) were
prelabeled with [methyl-3H]choline as described under “Experimental
Procedures” and then exposed to Cp-PLC (21.4 ng/ml) for 0.5, 1.5, 2, 4,
and 6 h (A) or to serial 10-fold dilutions of Cp-PLC for 4 h (B). The
supernatant was collected; the counts/min were measured using a
-counter and related to the amount of cellular protein (A) or to the total
amount of counts/min incorporated into membranes (B).
Ganglioside Reduction Causes 
-Toxin Hypersensitivity 26685
dylethanolamine, or phosphatidylglycerol (42). The influence of
the addition of gangliosides on the susceptibility of artificial
membranes to disruption by Cp-PLC was studied. Incorpora-
tion of asialo-GM1a to PC- or SM-containing liposome assays
did not protect them from the membrane-disrupting effect of
Cp-PLC (Fig. 9, A and B). However, incorporation of ganglio-
sides of the a series (GM1a or GD1a) or b series (GD1b or GT1b)
readily prevented the membrane disruption caused by Cp-PLC
in a dose-dependent manner (Fig. 9, A and B).
Cp-PLC or Its C-terminal Domain Does Not Interact with
GT1b—The possibility that Cp-PLC or C-PLC interacts directly
with GT1b was evaluated by biosensor analysis. GT1b was
immobilized onto a carboxymethyl-dextran-coated CM5 sensor
chip or onto a lipophilic L1 sensor chip. To evaluate the speci-
ficity of the GT1b-immobilized surfaces, anti-GT1b ganglioside
mAb 101, mAb 102, or anti-GD3 ganglioside mAb KM871 (165
nM) were injected on both the GT1b-coated and the control
surfaces. The anti-GT1b mAbs 101 and 102 were shown to bind
specifically to immobilized GT1b via hydrophobic interaction
onto the CM5 surface (140 and 260 RU, respectively, Fig. 9A)
compared with the CM5 control surface. The mAb KM871 (Fig.
10A) and bovine serum albumin (not shown) did not bind to
immobilized GT1b. Similar results were obtained by using the
L1 chip. The anti-GT1b mAbs 101 and 102 were shown to bind
specifically to immobilized GT1b/PC vesicles (170 and 130 RU,
respectively, see Fig. 10B) compared with a PC control surface,
whereas the mAb KM871 (Fig. 10B) or bovine serum albumin
(not shown) did not bind to the GT1b/PC vesicles. No binding
was observed upon injection of Cp-PLC or its C-terminal do-
main (C-PLC) (2.35 and 3.5 mM, respectively) onto GT1b-coated
surfaces (Fig 10, A and B). Similar results were obtained when
Cp-PLC or C-PLC was injected at concentrations of 7 and 10
mM, respectively.
DISCUSSION
Cp-PLC is the major virulence factor in the pathogenesis of
gas gangrene, a devastating disease associated with traumatic
injuries or surgical wounds, which has increasing significance
in diabetes. Despite the recent identification of the residues
critical for toxicity in Cp-PLC (24, 43, 44), the molecular mech-
anism of action of this toxin is still not completely understood.
Cp-PLC is hemolytic, necrotizing, lethal, and highly toxic to
various cell types, particularly those that are deficient in UDP-
Glc (18). However, the reason why a UDP-Glc deficiency in-
duces cellular hypersensitivity to Cp-PLC was unknown. UDP-
Glc is an important precursor in the synthesis of cell surface
glycoconjugates such as proteoglycans, glycoproteins, and
GSLs. This work demonstrates that a reduction of cellular
gangliosides increases the cellular sensitivity to Cp-PLC. Sev-
eral lines of evidence support this conclusion. (i) Pharmacolog-
ical inhibition of ganglioside synthesis sensitizes different cell
lines to the cytotoxic effect of Cp-PLC. (ii) Mutant cell lines
with a reduced content of gangliosides, due to a mutation either
in the UDPG:CGT or the UDPG:PP, are 105 times more sensi-
tive to Cp-PLC than the corresponding clones transfected with
the wild type enzymes. (iii) The addition of exogenous ganglio-
sides protects hypersensitive cells from the cytotoxic effect
of Cp-PLC.
Regarding the type of gangliosides involved in this phenom-
enon, it was found that gangliosides protected artificial and
cellular membranes from the disruption caused by Cp-PLC,
whereas asialo-GM1a did not protect them. The fact that GM3
and GM1a have a protective effect, whereas asialo-GM1a does
not, demonstrates that the SiaAc attached to the internal ga-
lactose plays a role in protection. Furthermore, because GM1a
and GD1b were less protective than GD1a and GT1b, respec-
tively, the results indicate that the SiaAc attached to the ter-
minal galactose also contributes to the protective effect of com-
plex gangliosides. Two lines of evidence support the importance
of the SiaAc units in cytoprotection. (i) C. perfringens sialidase
abolishes the capacity of GM3 and GT1b to protect GM95 cells
from the cytotoxic effect of Cp-PLC. (ii) Pretreatment of cul-
tured cells with C. perfringens sialidase increases significantly
their sensitivity to Cp-PLC.
Our observations shed light on general issues of muscle
pathology during clostridial myonecrosis. When injected intra-
muscularly, Cp-PLC causes rapid and extensive damage to
FIG. 8. LDH release induced by exposure to Cp-PLC orB. asper
MT II occurs in a dose-dependent manner. Don Q (black bars), B9
(hatched bars), GM95 (gray bars), or CG-1 (white bars) cells were
exposed for 2 h to serial 10-fold dilutions of Cp-PLC (A) or B. asper MT
II (B) in medium containing 1% FBS. The supernatant was collected,
and 2 h later the activity of LDH was measured and related to the total
amount of LDH determined after lysis with 1% Triton X-100, as de-
scribed under “Experimental Procedures.”
FIG. 9. Gangliosides protect PC- and SM-containing liposomes
from the membrane-disrupting effect of Cp-PLC. The indicated
gangliosides (11 g/ml, black bars; 33 g/ml, hatched bars; 100 g/ml,
white bars) were incorporated in cholesterol and PC- or SM-containing
liposomes (A and B, respectively) encapsulating carboxyfluorescein and
then exposed to Cp-PLC (20 g/ml). Fluorescence at 520 nm (excitation
at 485 nm) was measured 20 min later using a fluorometer and related
to the total fluorescence determined after lysis with 1% Triton X-100, as
described under “Experimental Procedures.”
Ganglioside Reduction Causes 
-Toxin Hypersensitivity26686
muscle fibers, inducing a significant release of CK to the
plasma (24). However, the reason for the high sensitivity of
muscle cells to this toxin is not clear. Muscle is known to have
the lowest concentration of complex gangliosides among all
mammalian tissues studied so far (45, 46). Therefore, our find-
ings offer a possible explanation for the high susceptibility of
muscle fibers to the cytotoxic effect of Cp-PLC. A UDP-Glc
deficiency has been reported to occur in the muscular tissue
during diabetes (47), probably as a consequence of the defective
glucose transport and/or phosphorylation (48). The effect of this
metabolic deficiency on ganglioside synthesis could be a pre-
disposing factor to the severity of the tissue damage caused by
C. perfringens infections in these patients.
Sequence analysis of the C. perfringens chromosome reveals
the presence of two genes encoding secreted sialidases that are
able to hydrolyze the SiaAc residues from glycoconjugates of
the mammalian cell membranes (5). The nanI gene (Locus CPE
0725) encodes a 694-amino acid protein known as the “large
sialidase enzyme” (49). The nanJ gene (Locus CPE 0553) en-
codes a 1173-amino acid protein not yet characterized, which
displays a 51% sequence similarity to the 111-kDa sialidase
from Clostridium septicum (50). Sialidase activity is detectable
in culture supernatants during exponential growth of C. per-
fringens, and in vivo it is detectable in the bloodstream of
experimental animals or patients with gas gangrene in early
stages of the disease (51, 52). In patients suffering clostridial
myonecrosis sialidase in amounts as high as 200 units/liter and
110 milliunits/g can be found in exudates and tissue homoge-
nates, respectively (51, 52). The results of the myotoxicity stud-
ies performed in this work clearly support the hypothesis that
C. perfringens sialidase has a synergistic effect with Cp-PLC in
the pathology of clostridial myonecrosis.
The damage inflicted by Cp-PLC to the muscle tissue during
C. perfringens infections could be also potentiated by its effects on
diverse signal transduction pathways in endothelial cells, plate-
lets, and neutrophils (7). In these cells, Cp-PLC leads to the
uncontrolled production of several intercellular mediators and
adhesion molecules, which alters the traffic of neutrophils to the
infected tissue and promotes thrombotic events leading to further
ischemia (7). Because UDPG:CGT expression decreases rapidly
in cells exposed to low oxygen tension (53), the perfusion deficit
caused by Cp-PLC in the infected tissue might lead to a reduction
in cellular gangliosides, further increasing the susceptibility of
muscle cells to its membrane damaging effect.
The capacity of Cp-PLC to disrupt artificial membranes de-
pends on its ability to bind and enzymatically degrade their
constituent phospholipids (54). We showed that the disruption
of PC- or SM-containing liposomes by Cp-PLC is prevented by
the incorporation of gangliosides into the liposomes membrane.
It has been reported that the incorporation of gangliosides into
artificial membranes does not affect the binding of Cp-PLC
(55). Therefore, the inhibition of the membrane-disrupting ef-
fect of Cp-PLC by gangliosides could result from a direct inter-
action of the Cp-PLC with the SiaAc units, which would pre-
vent it from reaching the acyl-ester bonds to be cleaved.
However, the results of the biosensor studies excluded that
possibility. Because gangliosides reduce membrane fluidity
and shift the cut-off pressure to lower values (56), their inhib-
itory effect on the Cp-PLC membrane-disrupting activity could
result from a tighter packing of the phospholipids in the target
membrane. Alternatively, because the bulky head groups of
gangliosides protrude from the interface and create a polar and
hydrated microenvironment (56), the inhibition could be due
either to an alteration of the availability of substrate due to a
steric effect or to the electrostatic changes induced at the in-
terface. The latter possibility is supported by the fact that
GD1a and GT1b, which carry higher number of charges, exert
the highest inhibition.
The three-dimensional structure of Cp-PLC shows two do-
mains joined by a short flexible linker region (57). The N-
terminal domain contains the active site, whereas the C-termi-
nal domain, which is needed to interact with aggregated
phospholipids, is analogous to C2 domains of intracellular eu-
karyotic proteins involved in signal transduction (57). Recent
crystallographic studies have revealed that Cp-PLC exists in
two conformations as follows: an open active form and a closed
inactive form. In the open form three Zn2 ions are bound, and
the active site is accessible (58). In the closed form there are
only two Zn2 ions bound, and the active site is hindered by the
loop encompassing residues 132–149 (58). It is suggested that
the binding of the Cp-PLC C-terminal domain to the target
membrane induces a conformational change in the N-terminal
domain, which uncovers the active site, allowing hydrolysis of
the aggregated phospholipid substrates (58). The changes in-
duced by the reduction of the gangliosides content of the target
membrane could favor that Cp-PLC acquires and/or maintains
its active conformation. Regardless of the mechanism, our re-
sults demonstrate that the reduction of gangliosides in the
target membrane increases the sensitivity to the Cp-PLC-in-
duced disruption.
Most interestingly, gangliosides have been reported to mod-
FIG. 10. Cp-PLC or its C-terminal domain do not interact with
GT1b. Ganglioside GT1b was immobilized (0.12 ng/mm2) onto the CM5
sensor surface via hydrophobic interaction, as described under “Experi-
mental Procedures” (A). Anti-GT1b specific mAbs (101 and 102) and the
anti-GD3 mAb KM871 were injected (30 l, 165 nM) at a flow rate of 10
l/min over immobilized GT1b ganglioside to confirm surface specificity.
Cp-PLC (PLC) and the C-terminal domain of Cp-PLC (C-PLC) were in-
jected (30 l and 2.35 and 3.5 mM, respectively) at a flow rate of 10 l/min
over the GT1b-coated surface. The sensograms shown have been sub-
tracted with the corresponding signal obtained when the sample was
passed over a nonderivatized CM5 surface. GT1b/phosphatidylcholine
lipids vesicles (100 g of GT1b/1 mg of PC) were immobilized (approxi-
mately 3.0 ng/mm2) onto an L1 sensor surface (B). PC lipids vesicles were
immobilized (3.2 ng/mm2) onto an L1 sensor surface as control. Anti-
GT1b-specific mAbs (101 and 102), the anti-GD3 mAb, KM871, Cp-PLC,
and C-PLC were injected as described in A. The sensograms shown have
been subtracted with the corresponding signal obtained when the sample
was passed over the PC control surface.
Ganglioside Reduction Causes 
-Toxin Hypersensitivity 26687
ulate the activities of some other membrane-interacting en-
zymes in vitro: classical protein kinases C (PKCs), secreted
class I phospholipase A2, and the B. cereus PLC (59–62). Gan-
gliosides inhibit the responsiveness of classical PKCs to diacyl-
glycerol (59). Although the mechanism of this inhibition has
not been clarified, it is worth noting that classical PKCs pos-
sess a C2 domain (63), analogous to that of Cp-PLC (57). Gan-
gliosides modulate the enzymatic activity of pancreatic phos-
pholipase A2 without acting directly on the active center or the
interfacial recognition region of the enzyme (55, 60). In addi-
tion, gangliosides also inhibit the fusion of large unilamellar
liposomes induced by the B. cereus PLC (61, 62), which lack the
C2-like domain present in Cp-PLC. This inhibition seems to
occur at the level of the lipid-water interface, affecting both the
maximum rate of catalysis of the enzyme and the availability of
the substrate (61, 62). It is not known whether Cp-PLC could
induce a similar effect in that type of liposomes, but it is clear
that its cytotoxic effect is not associated with fusion of the
cellular membranes of the target cells and syncytia formation
(15). Considering the substantial structural differences be-
tween B. cereus PLC and Cp-PLC, it is difficult to speculate
whether the mechanism by which gangliosides inhibit mem-
brane fusion induced by the first enzyme is somehow related to
the mechanism by which they inhibit the cytotoxic effect of
the latter.
Previous studies with genetically engineered Cp-PLC vari-
ants have shown that its catalytic capacity is required for
cytotoxicity (18), supporting a role for enzymatic phospholipids
degradation and membrane damage in cell death. The capacity
of cells to synthesize PC, required for an adaptive response of
cells to Cp-PLC, is needed for cell resistance to this toxin, and
therefore, cells with a lowered capacity to synthesize PC are
hypersensitive to Cp-PLC (64). However, we have found previ-
ously that a cellular UDP-Glc deficiency is not associated with
a defective synthesis of choline-containing phospholipids (18).
The present data showed that Cp-PLC causes a higher degra-
dation of choline-containing phospholipids in cells deficient in
UDP-Glc than in cells with normalized levels of this metabolite.
Therefore, the cytotoxicity of Cp-PLC in UDP-Glc deficient cells
correlates with the extent of hydrolysis of choline-containing
phospholipids of the plasma membrane and not with differ-
ences in the phospholipids-synthesizing metabolic pathways.
Thus, it seems likely that upon Cp-PLC treatment in ganglio-
side-deficient cells, the synthesis of choline-containing phos-
pholipids, albeit normal, cannot cope with the increased mem-
brane degradation, leading to structural membrane damage.
In conclusion, this work provides new insights toward un-
derstanding the factors that determine the susceptibility to the
toxic effect of Cp-PLC both in cultured cells and in the muscu-
lar tissue. The data provide compelling evidence that a de-
crease in the cellular content of gangliosides increases the
sensitivity to the cytotoxic effect of Cp-PLC and showed that
the disruption of the plasma membrane caused by this toxin is
potentiated by the “large C. perfringens sialidase.” This knowl-
edge might provide a rationale for the development of novel
therapeutic strategies to reduce tissue damage during clostrid-
ial myonecrosis.
Acknowledgments—We thank L. Monturiol, M. J. Pineda, and
M. Jayasekera for their valuable help in some experiments and
Prof. B. Lomonte (Universidad de Costa Rica) for critical reading of the
manuscript and for providing B. asper MT II. We also thank Dr. J. Esko
(University of California, San Diego) for providing the mutant CHO cell
pgsG-110 and the transfectant clone pgs-T; Dr. S. I. Hakomori (Univer-
sity of Washington, Seattle) and Dr. R. L. Schnaar (The Johns Hopkins
School of Medicine, Baltimore) for providing mAbs.
REFERENCES
1. Titball, R. W. (1999) in The Comprehensive Source Book of Bacterial Protein
Toxins (Alouf, J. E., and Freer, J. H., eds) 2nd Ed., pp. 310–329, Academic
Press, New York
2. Songer, J. G. (1997) Trends Microbiol. 5, 156–161
3. Vazquez-Boland, J. A., Jun, M., Berche, P., Chakraburry, T., Domı´nguez-
Bernal, G., Goebel, W., Gonza´lez-Zorn, B., Wehland, J., and Kreft, J. (2001)
Clin. Microbiol. Rev. 14, 584–640
4. Stonehouse, M. J., Cota-Gomez, A., Parker, S. K., Martin, W. E., Hankin, J. A.,
Murphy, R. C., Chen, W., Lim, K. B., Hackett, M., Vasil, A. I., and Vasil,
M. L. (2002) Mol. Microbiol. 46, 661–676
5. Shimizu, T., Ohtani, K., Hirakawa, H., Ohshima, K., Yamashita, A., Shiba, T.,
Ogasawara, N., Hattori, M., Kuhara, S., and Hayashi, H. (2002) Proc. Natl.
Acad. Sci. U. S. A. 99, 996–1001
6. Titball, R. W., Naylor, C. E., and Basak, A. K. (1999) Anaerobe 5, 51–64
7. Stevens, D. L. (2000) Int. J. Med. Microbiol. 290, 497–502
8. Burke, M. P., and Opeskin, K. (1999) Am. J. Forensic Med. Pathol. 20, 158–162
9. Rood, J. I. (1998) Annu. Rev. Microbiol. 52, 333–360
10. Karasawa, T., Wang, X., Maegawa, T., Michiwa, Y., Kita, H., Miwa, K., and
Nakamura, S. (2003) Infect. Immun. 71, 641–646
11. Clark, G. C., Briggs, D. C., Karasawa, T., Wang, X., Cole, A. R., Maegawa, T.,
Jayasekera, P. N., Naylor, C. E., Miller, J., Moss, D. S., Nakamura, S.,
Basak, A. J., and Titball, R. W. (2003) J. Mol. Biol. 333, 759–769
12. Bunting, M., Lorant, D. E., Bryant, A. E., Zimmerman, G. A., McIntyre, T. M.,
Stevens, D. L., and Prescott, S. M. (1997) J. Clin. Investig. 100, 565–574
13. Williamson, E. D., and Titball, R. W. (1993) Vaccine 11, 1253–1258
14. Awad, M. M., Bryant, A. E., Stevens, D. L., and Rood, J. I. (1995) Mol.
Microbiol. 15, 191–202
15. Flores-Dı´az, M., Thelestam, M., Clark, G. C., Titball, R. W., and Alape-Giro´n,
A. (2004) Anaerobe 10, 115–123
16. Florin, I. (1991) Microb. Pathog. 11, 337–346
17. Flores-Dı´az, M., Alape-Giro´n, A., Persson, B., Pollesello, P., Moos, M., von
Eichel-Striber, C., Thelestam, M., and Florin, I. (1997) J. Biol. Chem. 272,
23784–23791
18. Flores-Dı´az, M., Alape-Giro´n, A., Titball, R. W., Moos, M., Guillouard, I., Cole,
S., Howells, A. M., von Eichel-Streiber, C., Florin, I., and Thelestam, M.
(1998) J. Biol. Chem. 273, 24433–24438
19. Prydz, K., and Dalen, K. T. (2000) J. Cell Sci. 113, 193–205
20. Ge Wei, X. B., Sinha, A., and Esko, J. (1999) J. Biol. Chem. 274, 13017–13024
21. Helenius, H., and Aebi, M. (2001) Science 291, 2364–2369
22. Merrill, A. H. (2002) J. Biol. Chem. 277, 25843–25846
23. Tettamanti, G. (2004) Glycoconj. J. 20, 301–317
24. Alape-Giro´n, A., Flores-Dı´az, M., Guillouard, I., Naylor, C., Titball, R. W.,
Rucavado, A., Lomonte, B., Basak, A. K., Gutie´rrez, J. M., Cole, S., and
Thelestam, M. (2000) Eur. J. Biochem. 267, 5191–5197
25. Lomonte, B., Tarkowski, A., and Hanson, L. A. (1994) Toxicon 32, 1359–1369
26. Dohi, T., Nores, G., and Hakomori, S. (1988) Cancer Res. 48, 5680–5685
27. Catimel, B., Scott, A. M., Lee, F. T., Hanai, N., Ritter, G., Welt, S., Old., L. J.,
Burgess, A. W., and Nice, E. C. (1998) Glycobiology 8, 927–938
28. Lunn, M. P. T., Johnson, L. A., Fromholt, S. E., Itonori, S., Huang, J., Vyas,
A. A., Hildreth, J. E. K., Griffin, J. W., Schnaar, R. L., and Sheikh, K. A.
(2000) J. Neurochem. 75, 404–412
29. Kazuya, I. P., Hidari, J., Ichikawa, S., Fujita, T., Sakiyama, H., and Hiraba-
yashi, Y. (1996) J. Biol. Chem. 271, 14636–14641
30. Yu, R. K., and Ariga, T. (2000) Methods Enzymol. 312, 115–134
31. Brade, L., Vielhaber, G., Heinz, E., and Brade, H. (2000) Glycobiology 10,
629–636
32. Jepson, M., Howells, A., Bullifent, H. L., Bolgiano, B., Crane, D., Miller, J.,
Holley, J., Jayasekera, P., and Titball, R. W. (1999) Infect. Immun. 67,
3297–3301
33. Elbein, A. D. (1991) FASEB J. 5, 3055–3063
34. Dinter, A., and Berger, E. G. (1998) Histochem. Cell Biol. 109, 571–590
35. Liu, W. K., Sze, S. C. W., Ho, J. C. K., Liu, B. P. L., and Yu, M. C. (2004) J. Cell.
Biochem. 91, 1159–1173
36. Yusuf, H. K. M., Pohlentz, G., and Sandhoff, K. (1983) Proc. Natl. Acad. Sci.
U. S. A. 80, 7075–7079
37. Guarnaccia, S. P., Shaper, J. H., and Schnaar, R. L. (1983) Proc. Natl. Acad.
Sci. U. S. A. 80, 1551–1555
38. Tsai, B., Gilbert, J. M., Stehle, T., Lencer, W., Benjamin, T. L., and Rapoport,
T. A. (2003) EMBO J. 22, 4346–4355
39. Ichikawa, S., and Hirabayashi, Y. (1998) Trends Cell Biol. 8, 198–202
40. Schwarzmann, G. (2001) Semin. Cell Dev. Biol. 12, 163–171
41. Roggentin, P., Kleineidam, R. G., and Schauer, R. (1995) Biol. Chem. Hoppe-
Seyler 376, 569–575
42. Nagahama, M., Michiue, K., and Sakurai, J. (1996) Biochim. Biophys. Acta
1280, 120–126
43. Walker, N., Holley, J., Naylor, C. E., Flores-Dı´az, M., Alape-Giro´n, A., Carter,
G., Carr, F. J., Thelestam, M., Keyte, M., Moss, D. S., Basak, A. K., Miller,
J., and Titball, R. W. (2000) Arch. Biochem. Biophys. 384, 24–30
44. Jepson, M., Bullifent, H. L., Crane, D., Flores-Dı´az, M., Alape-Giro´n, A.,
Jayasekeera, P., Lingard, B., Moss, D., and Titball, R. W. (2001) FEBS Lett.
495, 172–177
45. Mu¨thing, J., and Cacic, M. (1997) Glycoconj. J. 14, 19–28
46. Mu¨thing, J., Maurer, U., Neumann, U., Kniep, B., and Weber-Schu¨rholz, S.
(1998) Carbohydr. Res. 307, 135–145
47. Basu, A., Basu, R., Shah, P., Vella, A., Johnson, M., Nair, K. S., Jensen, M. D.,
Schwenk, W. F., and Rizza, R. A. (2000) Diabetes 49, 272–283
48. Bonadonna, R. C., Del Prato, S., Bonora, E., Saccomani, M. P., Guilli, G.,
Natali, A., Frascerra, S., Pecori, N., Ferrannini, E., Bier, D., Cobelli, C., and
DeFronzo, R. A. (1996) Diabetes 45, 915–925
49. Traving, C., Schauer, R., and Roggentin, P. (1994) Glycoconj. J. 11, 141–151
50. Rothe, B., Rothe, B., Roggentin, P., and Schauer, R. (1991) Mol. Gen. Genet.
Ganglioside Reduction Causes 
-Toxin Hypersensitivity26688
226, 190–197
51. Schauer, R., Sander-Wever, M., Gutschker-Gdaniec, G. H., Roggentin, P.,
Randow, E. A., and Hobrecht, R. (1985) Clin. Chim. Acta 146, 119–127
52. Roggentin, T., Kleineidam, R. G., Majewski, D. M., Tirpitz, D., Roggentin, P.,
and Schauer, R. (1993) J. Immunol. Methods 157, 125–133
53. Zhao, H., Miller, M., Pfeiffer, K., Buras, J. A., and Stahl, G. L. (2003) FASEB
J. 17, 723–734
54. Nagahama, M., Michiue, K., Mukai, M., Ochi, S., and Sakurai, J. (1998)
Microbiol. Immunol. 42, 533–538
55. Bianco, I. D., Fidelio, G. D., and Maggio, B. (1990) Biochim. Biophys. Acta
1026, 179–185
56. Maggio, B. (1994) Prog. Biophys. Mol. Biol. 62, 55–117
57. Naylor, C. E., Eaton, J. T., Howells, A., Justin, N., Moss, D. S., Titball, R. W.,
and Basak, A. K. (1998) Nat. Struct. Biol. 5, 738–746
58. Eaton, J. T., Naylor, C. E., Howells, A. M., Moss, D. S., Titball, R. W., and
Basak, A. K. (2002) J. Mol. Biol. 319, 275–281
59. Kreutter, D., Kim, J. Y. H., Goldenring, J. R., Rasmunsen, H., Ukomandu, C.,
DeLorenzo, R. J., and Yu, R. K. (1987) J. Biol. Chem. 262, 1633–1637
60. Maggio, B., Bianco, I. D., Montich, G. G., Fidelio, G. D., and Yu, R. K. (1994)
Biochim. Biophys. Acta 199, 137–148
61. Daniele, J. J., Maggio, B., Bianco, I. D., Gon˜i, F. M., Alonso, A., and Fidelio,
G. D. (1996) Eur. J. Biochem. 239, 105–110
62. Basan˜ez, G., Fidelio, G. D., Gon˜i, F. M., Maggio, B., and Alonso, A. (1996)
Biochemistry 35, 7506–7513
63. Sutton, R. B., and Sprang, S. R. (1998) Structure (Lond.) 6, 1395–1405
64. Sleight, R., and Kent, C. (1983) J. Biol. Chem. 258, 824–830
Ganglioside Reduction Causes 
-Toxin Hypersensitivity 26689