JOURNAL OF BACTERIOLOGY, Nov. 2007, p. 7531–7538
Copyright © 2007, American Society for Microbiology. All Rights Reserved.
Vol. 189, No. 21
Epsilon-Toxin Plasmids of Clostridium perfringens Type D Are Conjugative?†
Meredith L. Hughes,1‡ Rachael Poon,1‡ Vicki Adams,1Sameera Sayeed,2Juliann Saputo,3
Francisco A. Uzal,3Bruce A. McClane,1,2and Julian I. Rood1*
Australian Research Council Centre of Excellence in Structural and Functional Microbial Genomics, Department of Microbiology,
Monash University, Victoria 3800, Australia1; Department of Molecular Genetics and Biochemistry, University of Pittsburgh
School of Medicine, Pittsburgh, Pennsylvania 152612; and California Animal Health and Food Safety Laboratory,
San Bernardino Branch, School of Veterinary Medicine, University of California,
Davis, San Bernardino, California 924083
Received 16 May 2007/Accepted 3 August 2007
Isolates of Clostridium perfringens type D produce the potent epsilon-toxin (a CDC/U.S. Department of
Agriculture overlap class B select agent) and are responsible for several economically significant enterotox-
emias of domestic livestock. It is well established that the epsilon-toxin structural gene, etx, occurs on large
plasmids. We show here that at least two of these plasmids are conjugative. The etx gene on these plasmids was
insertionally inactivated using a chloramphenicol resistance cassette to phenotypically tag the plasmid. High-
frequency conjugative transfer of the tagged plasmids into the C. perfringens type A strain JIR325 was
demonstrated, and the resultant transconjugants were shown to act as donors in subsequent mating experi-
ments. We also demonstrated the transfer of “unmarked” native ?-toxin plasmids into strain JIR325 by
exploiting the high transfer frequency. The transconjugants isolated in these experiments expressed functional
?-toxin since their supernatants had cytopathic effects on MDCK cells and were toxic in mice. Using the widely
accepted multiplex PCR approach for toxin genotyping, these type A-derived transconjugants were genotypi-
cally type D. These findings have significant implications for the C. perfringens typing system since it is based
on the toxin profile of each strain. Our study demonstrated the fluid nature of the toxinotypes and their
dependence upon the presence or absence of toxin plasmids, some of which have for the first time been shown
to be conjugative.
The gram-positive anaerobe Clostridium perfringens is the
causative agent of severe gastrointestinal disease (including
enterotoxemia and enteritis) in animals and gas gangrene and
food poisoning in humans (13, 23, 34). C. perfringens isolates
produce many different toxins but are divided into toxin types
A to E based solely on the production of four major toxins:
?-toxin (encoded by the plc gene), ?-toxin (cpb), ε-toxin (etx),
and ?-toxin (iap) (14, 25).
C. perfringens strains of all five toxin types produce the chro-
mosomally encoded ?-toxin, and type C, D, and E isolates also
produce plasmid-encoded ?-toxin, ε-toxin, and ?-toxin, respec-
tively. Type B isolates produce ?-toxin, ?-toxin, and ε-toxin.
The demonstration that the genes encoding three of these
typing toxins are found on large virulence plasmids led to the
hypothesis that C. perfringens types B to E may represent type
A strains that have received toxin-encoding plasmids by hori-
zontal gene transfer (7, 21).
In C. perfringens conjugative plasmids are associated primar-
ily with resistance to antibiotics, particularly tetracycline (1, 2).
The paradigm C. perfringens tetracycline resistance plasmid,
pCW3, has been sequenced, and its conjugation region, the tcp
locus, has been shown to be essential for conjugative plasmid
transfer (4). All known conjugative C. perfringens plasmids
have the tcp region (4). Until recently, there was little or no
evidence that the toxin plasmids from C. perfringens may be
conjugative, with the exception of the enterotoxin (CPE) plas-
mid (5). Subsequent studies have shown that the tcp locus is
present in several CPE plasmids, as well as ?-, ε-, and ?-toxin
plasmids from C. perfringens type C, D, and E strains, respec-
tively (4, 16, 30). Recently, it was shown that type D isolates
may harbor several large plasmids (48 to 110 kb), which could
carry up to three different toxin genes (30). These plasmids
have not yet been shown to be conjugative, but some were
shown to carry the tcp region (4, 30).
ε-Toxin is a pore-forming cytotoxin and is one of the most
potent clostridial toxins (33). It is classified as an overlap class
B select agent by the CDC and U.S. Department of Agricul-
ture due to its potential for misuse as a bioterrorism agent.
ε-Toxin-producing type D isolates are the etiological agents of
highly lethal enterotoxemias, particularly in sheep and goats
(13, 35). The exact role of ε-toxin in C. perfringens type D
disease is still unknown; however, this potent neurotoxin is
thought to mediate the rapid decline of diseased animals by
entering the bloodstream and exerting both systemic and neu-
rological effects (13, 35, 37). In addition, in goats, ε-toxin can
produce severe damage to the intestinal tract without having to
be absorbed into the systemic circulation (37).
The potent nature of ε-toxin and type D-mediated disease
highlights the importance of understanding the mechanisms by
which plasmids carrying the etx gene and other lethal toxin
genes are disseminated. To this end, we tagged two type D
ε-toxin plasmids by insertionally inactivating the etx gene with
* Corresponding author. Mailing address: Department of Microbi-
ology, Monash University, Victoria 3800, Australia. Phone: (613) 9905
4825. Fax: (613) 9905 4811. E-mail: email@example.com.
† Supplemental material for this article may be found at http://jb
‡ M.L.H. and R.P. contributed equally to this paper and are joint
?Published ahead of print on 24 August 2007.
a chloramphenicol resistance cassette and used these marked
plasmids to determine if the plasmids were conjugative. Here
we report the first demonstration of the conjugative transfer of
both marked and unmarked wild-type ε-toxin plasmids from
type D to type A strains of C. perfringens and characterization
of the resultant transconjugants.
MATERIALS AND METHODS
Bacterial strains and culture conditions. All strains and plasmids used in this
study are shown in Table 1. Strains were grown overnight in FTG medium
(Oxoid) and then on TSC agar, C. perfringens agar base (TSC and SFP; Oxoid)
containing 0.04% D-cycloserine (Sigma). When required, antimicrobial agents
were added to solid media at the following concentrations: chloramphenicol, 30
?g ml?1; thiamphenicol, 30 ?g ml?1; rifampin, 20 ?g ml?1; nalidixic acid, 20 ?g
ml?1; streptomycin, 1 mg ml?1; tetracycline, 10 ?g ml?1; and saturated potas-
sium chlorate, 1% (vol/vol). Agar cultures were grown in an atmosphere con-
taining 10% CO2and 10% H2in N2at 37°C overnight.
Conjugation experiments. Mixed plate mating was carried out as previously
described (24, 26), except that selection was on TSC agar supplemented with the
appropriate antibiotics. The recipients were the strain 13 derivatives JIR325
(rifampin and nalidixic acid resistant) and JIR4394 (streptomycin and potassium
chlorate resistant). The donor strains were the type D ?etx::catP mutants
JIR4981 and JIR4982. Transconjugants were selected on TSC agar containing
thiamphenicol, rifampin, and nalidixic acid when JIR325 was the recipient and
on TSC agar containing thiamphenicol, streptomycin, and potassium chlorate
when JIR4394 was the recipient. The conjugative transfer efficiency was defined
as the number of transconjugants per donor cell. Strains carrying pCW3 were
used as positive controls in all mating experiments.
In the unmarked plasmid conjugations, the type D strain CN1020 was used as
the donor and JIR325 was used as the recipient. After mating as described above,
derivatives of JIR325 were selected on agar medium containing rifampin and
nalidixic acid. Transconjugants were identified by colony hybridization carried
out according to the manufacturer’s instructions (Roche), using a digoxigenin
(DIG)-labeled etx probe.
Preparation of culture supernatants. Culture supernatants from 4-h (late-log-
phase) TPG broth cultures (10) were centrifuged at 6,700 ? g for 10 min, and this
was followed by filter sterilization using a 0.45-?m-pore-size filter (Millipore). To
activate ε-toxin activity, 0.05% trypsin (1:250; Sigma) was added, and the culture
supernatants were incubated at 37°C for 30 min prior to use in Madin-Darby
canine kidney (MDCK) cell assays for cytotoxicity. When required, supernatants
were concentrated 10-fold using Amicon ultracentrifugation columns (10,000-Da
cutoff). For each strain supernatants from cultures grown on at least three
separate occasions were assayed.
Molecular methods. C. perfringens genomic DNA for use as a template in
PCRs was prepared as described previously (4). Plasmid DNA from Escherichia
coli was isolated by alkaline lysis (QIAGEN). PCR amplification using Pfo
polymerase and 0.5 ?M of each primer was performed using 30 cycles of dena-
turation (95°C), annealing (55°C), and extension (72°C). Extension times were
varied from 1 to 4 min depending on the expected PCR product length.
QIAquick PCR purification kits were used to purify PCR products before clon-
TABLE 1. Bacterial strains and plasmids
Strain or plasmid Descriptiona
Source or reference(s)
C. perfringens strains
CN1020Type D wild-type strain, carries 48-kb ε-toxin plasmid pJIR3118 Burroughs-Wellcome Collection via
R. G. Wilkinson
Burroughs-Wellcome Collection via
R. G. Wilkinson
P. Johanesen, D. Lyras, and
J. Rood, unpublished
J. Parsons, T. Bannam, and
J. Rood, unpublished
Transconjugant, JIR4981 ? JIR325
Transconjugant, JIR4982 ? JIR325
Transconjugant, CN1020 ? JIR325
Transconjugant, CN1020 ? JIR325
CN3718Type D wild-type strain, carries 48-kb ε-toxin plasmid pJIR3119
Strain 13 RifrNalr
Strain 13 StrrChlr
CN1020(pJIR3120) ?etx::catP Cmr/Tmr
CN3718(pJIR3121) ?etx::catP Cmr/Tmr
47 kb, conjugative, Tcr
E. coli-C. difficile shuttle vector, 7.1 kb, C. difficile repA Emr
Clostridial suicide plasmid, 6.1 kb, pT7Blue-3 backbone ?bla?ermB,
?kan?oriT, EcoRI?catP, EmrCmr
8.4-kb plasmid with Asp718/PstI-digested 2.3-kb 5? etx PCR product
cloned into KpnI/PstI-digested pJIR2783, EmrCmr
9.9-kb plasmid, pJIR2906 with 1.5-kb 3? etx PCR product cloned into
XbaI (T4 filled), EmrCmr
6.9-kb plasmid, pJIR2907 digested with XmnI and religated to
remove oriT and ermB, Cmr
14-kb ?etx::catP recombination vector, 7.1-kb FspI-digested
pMTL9301 vector ligated into KpnI/NotI-digested pJIR3076.ColE1,
C. difficile repA oriT ermB ?etx::catP EmrCmr/Thr
Wild-type 48-kb ε-toxin plasmid from CN1020
Wild-type 48-kb ε-toxin plasmid from CN3718
pJIR3118 ?etx::catP, Cmr/Tmr
pJIR3120Transformation with pJIR3077
Transformation with pJIR3077
pJIR3121pJIR3119 ?etx::catP, Cmr/Tmr
aRifr, Nalr, Tcr, Strr, Chlr, Emr, and Cmr/Tmr, resistance to rifampin, resistance to nalidixic acid, resistance to tetracycline, resistance to streptomycin, resistance to
potassium chlorate, resistance to erythromycin, and resistance to chloramphenicol and thiamphenicol, respectively.
7532HUGHES ET AL. J. BACTERIOL.
ing, sequencing, or DIG labeling. C. perfringens typing was performed using a
multiplex PCR assay as previously described (8). PCR products were separated
on a 1% agarose gel by electrophoresis. Restriction endonuclease digestion,
agarose gel electrophoresis, and DNA ligation were performed using standard
laboratory methods (28). All oligonucleotide sequences are shown in Table S1 in
the supplemental material.
Construction of C. perfringens mutants by allelic exchange. Based on the
sequence of the ε-toxin plasmid pJGS1721 (G. Myers, I. Paulsen, J. G. Songer,
B. McClane, R. Titball, J. Rood, and S. Melville, unpublished data) primers
JRP2240 and JRP2241 were designed to amplify an upstream 2.3-kb fragment
which incorporated the first 248 nucleotides of the etx gene and 2 kb of upstream
sequence. A 1.5-kb downstream fragment that included the last 265 nucleotides
of etx and 1.2 kb of downstream sequence was amplified with primers JRP2112
and JRP2242. These fragments were cloned into pJIR2783 on either side of the
catP gene, resulting in the 9.9-kb plasmid pJIR2907. Digestion of pJIR2907 with
XmnI and religation resulted in deletion of the oriT site and ermB gene, yielding
the 6.9-kb plasmid pJIR3076. The 7.1-kb FspI fragment of pMTL9301 (22) was
cloned into NotI/KpnI-digested pJIR3076 to construct the 14-kb recombination
vector pJIR3077, which then was introduced into the wild-type type D strains
CN1020 and CN3718 by electroporation (31). Thiamphenicol-resistant transfor-
mants were isolated and passaged in nonselective FTG broth and then plated
onto TSC agar containing thiamphenicol. Thiamphenicol-resistant colonies were
patched onto medium containing erythromycin (to ensure that free pJIR3077 did
not persist), and erythromycin-sensitive colonies were chosen for further analysis.
Southern hybridization. Genomic and plasmid DNA was digested with HindIII,
and fragments were separated by electrophoresis on 0.8% agarose gels. DNA was
visualized by ethidium bromide staining prior to transfer to nitrocellulose mem-
branes. DIG labeling, hybridization, chemiluminescent detection, stripping, and
reprobing were carried out as described in the DIG user’s manual (Roche).
DIG-labeled probes were generated by PCR using the following primers: for etx,
JRP2074 and JRP2495; for catP, JRP2142 and JRP2143; for ermB, JRP1924 and
JRP1925; for repA, JRP2457 and JRP2465; and for tcpH, JRP1758 and JRP1661.
PFGE. C. perfringens genomic DNA was embedded in agarose plugs, and the
uncut genomic DNA was subjected to pulsed-field gel electrophoresis (PFGE)
prior to transfer to nylon membranes and Southern hybridization using a DIG-
labeled etx probe, as described previously (30).
Cell culture and cytotoxicity assays. MDCK cells were grown in Dulbecco’s
modified Eagle’s medium supplemented with 2 mM glutamate, 10% fetal calf
serum, and 100 U/ml of streptomycin and penicillin (Invitrogen). For cytotoxicity
assays, six-well tissue culture trays were seeded with 2 ? 105cells/well and
incubated overnight in 5% CO2to obtain semiconfluence. Cell monolayers were
washed three times with phosphate-buffered saline and immersed in 1 ml of
Dulbecco’s modified Eagle’s medium prior to addition of 50 ?l of trypsin-treated
or non-trypsin-treated culture supernatant. Cytopathic effects were observed
using an inverted confocal microscope (Olympus 1X71) with a 37°C heated stage
over a 24-h period. For neutralization of ε-toxin activity, monoclonal antibody
(MAb) 5B7 (kindly provided by Paul Hauer, Center for Veterinary Biologics,
Ames, IA) (9) was added to trypsin-treated culture supernatants and incubated
at room temperature for 1 h as described previously (29) before addition of the
treated supernatant to MDCK cell monolayers. As a negative control, culture
supernatants were treated with a polyclonal antibody reactive against the ErmB
Quantification of ?-toxin, perfringolysin O, and ?-toxin levels in culture
supernatants. ?-Toxin was assayed (32) using nutrient agar supplemented with
4% egg yolk by reference to a C. perfringens phospholipase C standard (Sigma).
The total protein content was determined using a bicinchoninic acid kit (Pierce),
and the specific activity was expressed in phospholipase C units/mg of total
protein. Perfringolysin O activity on horse red cells was determined by a doubling
dilution assay; the perfringolysin O titer was defined as the reciprocal of the last
dilution showing complete hemolysis (36). ε-Toxin levels were determined by
comparing the chemiluminescent signal intensities for known ε-toxin standards
with the test samples in Western blots, using ε-toxin MAb 5B7, as described
Mouse intravenous injection model. The toxicity of the strains was analyzed
using a mouse intravenous assay to calculate the 50% lethal dose (LD50) of
culture supernatant, as previously described (29). Briefly, sterile culture super-
natants were incubated with 0.05% trypsin (1:250; Difco) for 30 min at 37°C and
then serially diluted twofold (between 1:100 and 1:1,600) in 1% peptone water.
Then 0.5 ml of each relevant dilution was injected intravenously into pairs of
mice. As a negative control, mice were inoculated with trypsinized 1% peptone
water. The double reciprocal of the highest dilution inducing lethality (within
48 h) in at least one of the paired mice was defined as the LD50. Two indepen-
dent batches of filtered culture supernatants were assayed, and the results were
expressed as the average of the LD50values for each batch. Experimental pro-
cedures involving animals were approved by the Animal Care and Use Commit-
tee of the California Animal Health and Food Safety Laboratory, University of
California, Davis (permit 04-11593).
Tagging the ?-toxin plasmids with catP. The objective of
these studies was to determine if the ε-toxin plasmids are
conjugative. From our previous study of type D strains of C.
perfringens (29) we chose two ε-toxin-producing isolates,
CN1020 and CN3718, which were transformable using stan-
dard laboratory methods and had been determined to have the
type D genotype by multiplex PCR. CN1020 harbors a single,
approximately 48-kb plasmid, now designated pJIR3118, which
carries IS1151, the etx gene, and the tcp locus (30). CN3718
also has a single 48-kb etx plasmid (data not shown), desig-
The ε-toxin plasmids in these strains were marked with the
catP gene, which confers chloramphenicol and thiamphenicol
resistance, so that resistance could be used as a selective
marker for plasmid transfer. We utilized a novel insertional
inactivation approach based on the successful mutagenesis of
several genes in Clostridium difficile in a previous study (18). In
that study, an E. coli-C. perfringens recombination vector that
was relatively unstable in C. difficile was used to introduce the
insertionally inactivated gene and allow time for recombina-
tion events to occur. Growth in the absence of the plasmid-
determined resistance marker followed by screening for resis-
tance to the gene used for insertional inactivation enabled
selection for chromosomal mutants. It was reasoned that it
might be possible to use the reverse approach in C. perfringens,
and subsequent stability assays showed that the E. coli-C. dif-
ficile shuttle vector pMTL9301 (18), which confers erythromy-
cin resistance, was unstable in C. perfringens in the absence of
selective pressure (data not shown).
A pMTL9301-based recombination vector, pJIR3077, that
contained an etx gene that had been insertionally inactivated
with catP was constructed and used to transform the type D
strains CN1020 and CN3718. After initial growth in the pres-
ence of thiamphenicol and subsequent growth in absence of
antibiotics to provide time for the loss of pJIR3077, thiam-
phenicol-resistant, erythromycin-sensitive colonies were se-
lected as potential ?etx::catP mutants.
Genomic DNA from these putative mutants was analyzed
initially by PCR. An 805-bp etx fragment was amplified from
CN1020 and CN3718 (Fig. 1A, lanes 2 and 4), and a 1.5-kb
fragment was amplified from the mutants and the suicide vec-
tor (Fig. 1A, lanes 3, 5, and 6). As expected, catP-specific PCR
products were amplified from the mutants but not from the
wild-type strains (Fig. 1B). The absence of ermB- and repA-
specific PCR products in the mutants (data not shown) indi-
cated that the vector did not persist in these strains. These
results are as predicted for insertional inactivation of the etx
genes by allelic exchange.
Southern hybridization of HindIII-digested DNA from the
wild-type and mutant strains confirmed the insertional inactiva-
tion of the etx gene. Hybridization with an etx-specific probe
showed that the 2.1-kb etx band in the wild-type strains had been
replaced by a 1.8-kb band in the mutants (Fig. 1C), as predicted
VOL. 189, 2007CONJUGATIVE TRANSFER OF EPSILON-TOXIN PLASMIDS 7533
(Fig. 1E). In addition, only the mutants had the 1.1-kb band that
hybridized to the catP probe. Probes specific for ermB and repA
hybridized only to the recombination vector pJIR3077 (data not
shown). These data provided convincing evidence that etx mu-
tants had been successfully constructed by double-crossover
events. The mutants were designated JIR4981 and JIR4982, and
their mutated plasmids were designated pJIR3118?etx::catP or
pJIR3120 and pJIR3119?etx::catP or pJIR3121, respectively.
Southern blotting with the plasmid-specific tcpH gene, which is
plasmid was still present in these strains (data not shown).
Marked ?-toxin plasmids are conjugative. To determine if
the plasmids from these type D strains were conjugative, we
used the tagged ?etx::catP mutants as donors in separate mixed
plate matings, using the type A strain JIR325 as the recipient.
Transconjugants were selected on medium containing thiam-
phenicol, selecting for the marked plasmids, and on medium
containing rifampin and nalidixic acid, selecting for the recip-
ient. No colonies were observed in control matings that in-
volved only the donor or recipient strains. Conjugative transfer
of thiamphenicol resistance (i.e., the marked plasmids) was
observed at very high frequencies comparable to those ob-
served for the paradigm tetracycline resistance plasmid, pCW3
(Table 2). PCR analysis confirmed that the plasmids had in-
deed been transferred into strain JIR325. In particular, a
?etx::catP-specific fragment was amplified from the transcon-
jugants but not from the recipient strain, and the cpb2 gene,
which was specific for the type A recipient, was amplified from
the transconjugants but not from the donor (Fig. 2A). In ad-
dition, two of the resultant transconjugants, JIR4983 and
JIR4985, one from each donor, could be used as donors in
subsequent matings and yielded transfer frequencies similar to
those of the original type D-to-type A matings (Table 2).
Conjugative transfer of an unmarked type D ?-toxin plas-
mid. The high transfer frequencies suggested that we might be
able to detect transfer of the native ε-toxin plasmid from a type
D strain to a type A strain without marking the plasmid. To this
end, plate matings were performed using CN1020 as the donor
and JIR325 as the recipient. In these unmarked plasmid con-
jugation experiments, transconjugants carrying the etx gene
were detected, by colony hybridization using an etx-specific
probe, at a frequency of 0.8%, and the results were confirmed
by PCR analysis (data not shown). Two of these independently
derived JIR325(pJIR3118) transconjugants, JIR12012 and
JIR12013, were chosen for further study.
Multiplex PCR analysis revealed that the conjugative trans-
fer of the ?-toxin plasmid to a type A strain converted this
strain to a genotypically type D isolate. A multiplex PCR assay
(8) that detects six C. perfringens lethal toxin genes, etx (ε-
toxin), plc (?-toxin), cpb (?-toxin), cpb2 (?2-toxin), cpe (CPE),
and iap (enzymatic component of ?-toxin), was used to analyze
the unmarked transconjugants. By definition, type A strains
carry the plc gene and may have other accessory toxin genes,
such as cpb2 or cpe, but do not have the etx, cpb, and iap genes.
By contrast, type D strains have both the plc and etx genes. In
this assay the type D donors CN1020 and CN3718 were both
etx?plc?and cpb2 negative, and the type A recipient JIR325
was plc?cpb2?and etx negative (Fig. 2A and 2B). The multi-
plex PCR profiles of the unmarked etx?transconjugants,
JIR12012 and JIR12013, identified these strains as C. perfrin-
gens type D strains since they had PCR products of the ex-
pected sizes when both the plc and etx primers were used,
despite the fact that the recipient in these matings was the type
A strain JIR325 (Fig. 2B). Both strains were clearly derived
from JIR325 since they carried the cpb2 gene. Finally, the use
FIG. 1. Confirmation of ?etx::catP mutants. (A and B) PCR anal-
ysis using etx- and catP-specific primers, respectively. Lanes 2 to 6
contained strains CN1020, JIR4891, CN3718, and JIR4892 and recom-
bination plasmid pJIR3077, respectively. Lanes 1 and 7 contained size
markers (Promega PCR marker and HindIII-digested ?cI857, respec-
tively). (C and D) Southern blots of HindIII-digested genomic DNA
obtained using etx- and catP-specific probes, respectively. The contents
of lanes 2 to 6 were the same as those described above. Lane 1
contained HindIII-digested ?cI857 molecular size markers. (E) Rep-
resentation of the etx region (based on previous studies ) in the
wild-type ε-toxin plasmids pJIR3118 and pJIR3119, compared to the
same region after insertional inactivation of the etx gene by allelic
exchange with the recombination vector pJIR3077. The positions of
HindIII sites (H) are indicated, as are the sizes of the predicted
HindIII fragments (in kb). The etx and catP probes used for Southern
blotting are indicated by the solid lines under the genes.
7534HUGHES ET AL. J. BACTERIOL.
of tcpH primers provided evidence that a conjugative plasmid
carrying the tcp region had been transferred (data not shown).
PFGE confirms the movement of large ?-toxin plasmids. To
confirm that ε-toxin plasmids had been transferred in these
experiments, PFGE followed by Southern blotting was carried
out using uncut genomic DNA (30). Using an etx-specific
probe, a single hybridizing band at approximately 48 kb was
detected for the parent type D strain CN1020 and for the
JIR325-derived transconjugants containing the native ε-toxin
plasmid pJIR3118 (data not shown). Similar results were ob-
tained with the marked transconjugants derived from the type
D strain CN3718, but no hybridization was observed with
JIR325. These results are consistent with the conjugative trans-
fer of the etx plasmids and their maintenance in the transcon-
?-Toxin plasmids have a tcp conjugation region that is very
similar to that of pCW3. To determine if the tcp conjugation
region was also present in the ε-toxin plasmids, we analyzed
genomic DNA from the marked plasmids pJIR3120 and
pJIR3121. PCR analysis was performed with 27 primer com-
binations that together encompassed the entire tcp locus, from
the intP gene to the dcm gene (4). The tcp conjugation region
was present in both plasmids, and most primer pairs generated
products that were the same size as those for pCW3. However,
there was some variation in the tcpF and tcpH regions, as
previously shown with another ε-toxin plasmid, pJGS1721 (4).
Type A transconjugants express ?-toxin, ?-toxin, and per-
fringolysin O. To determine if the etx gene was expressed in
the type A background and if the ε-toxin plasmid had any effect
on the production of other toxins, Western blotting was done
with culture supernatants using ε-toxin MAb 5B7. The results
showed that the type A-derived transconjugants that originated
from transfer of the native ε-toxin plasmid pJIR3118 produced
levels of ε-toxin (Fig. 3) that were 2.5-fold lower than the
expression levels in the type D donor. However, the transcon-
jugants produced ?-toxin and perfringolysin O at levels com-
parable to those produced by the type A recipient strain (Fig.
3C and D). These results showed that the lower level of pro-
duction of ?-toxin and perfringolysin O in the type D strain was
not the result of the presence of a repressor gene in the ε-toxin
?-Toxin produced in the type A-derived transconjugants is
active on MDCK monolayers and is lethal in the mouse intra-
venous injection model. ε-Toxin is expressed as a relatively
inactive protoxin that is activated after cleavage of both N-
terminal and C-terminal domains by either trypsin or extracel-
lular proteases produced by C. perfringens (15). Since purified,
trypsin-treated ε-toxin is known to be cytotoxic to MDCK cells
(19, 20), we treated culture supernatants from several strains
with trypsin to activate any ε-toxin that was present and exam-
ined the cytopathic effects on MDCK cell monolayers. In initial
experiments cell rounding and blebbing were observed as soon
as 30 min after inoculation with culture supernatants that con-
tained activated ε-toxin, and severe loss of adherence and
obvious cytotoxicity was apparent after 24 h. After 24 h, tryp-
sin-treated supernatants from CN1020 or CN3718, diluted at
least 10-fold, still caused clear cytopathic effects (Fig. 4). Cul-
ture supernatants of JIR325 derivatives carrying the native
ε-toxin plasmid also were highly toxic to the MDCK cells (Fig.
4 and data not shown). By comparison, MDCK cells incubated
for 24 h with trypsin-treated undiluted culture supernatants
from JIR325 or with either JIR4983 or JIR4985 (JIR325 de-
rivatives carrying the mutated ?etx::catP plasmids) showed no
Pretreatment with an MAb that was specific for ε-toxin com-
pletely abrogated the destructive effects of trypsin-activated
supernatants on the MDCK cells. However, a control poly-
clonal antibody specific for the unrelated ErmB RNA methyl-
TABLE 2. Transfer of marked etx plasmids
DonorRecipient PlasmidPlasmid propertiesTransfer frequencya
pJIR3118 ?etx::catP (Tmr)
pJIR3119 ?etx::catP (Tmr)
pJIR3118 ?etx::catP (Tmr)
pJIR3119 ?etx::catP (Tmr)
(2.9 ? 4.0) ? 10?1
(3.8 ? 6.6) ? 10?2
(2.0 ? 2.0) ? 10?1
(4.8 ? 3.9) ? 10?1
(1.0 ? 6.3) ? 10?1
(5.7 ? 2.3) ? 10?2
aThe transfer frequencies are expressed as the number of transconjugants per donor cell and are the averages of at least three separate biological replicates.
FIG. 2. Multiplex PCR analysis of the wild type and transconju-
gants. (A) Lanes 2 to 8 contained type D strain CN1020, CN1020 etx
mutant JIR4981, JIR4981 ? JIR325 transconjugant JIR4983, type D
strain CN3718, CN3178 etx mutant JIR4982, JIR4982 ? JIR325
transconjugant JIR4985, and JIR325, respectively. Lane 1 contained
Promega PCR molecular size markers. (B) Lanes 2 to 5 contained
strains JIR325, JIR12012, JIR12013, and CN1020, respectively. Lane 1
contained Promega PCR molecular size markers.
VOL. 189, 2007 CONJUGATIVE TRANSFER OF EPSILON-TOXIN PLASMIDS7535
ase protein had no effect (Fig. 4). In addition, non-trypsin-
treated supernatants of JIR325 and the transconjugants caused
no cytopathic effects. These data demonstrated that active
ε-toxin in the culture supernatants, not other toxins such as
?-toxin or perfringolysin O, was responsible for the cytopathic
effects on the MDCK cells.
Finally, trypsin-treated supernatants were tested for the ability
to cause disease in the mouse intravenous injection model (29).
The results showed that the type A-derived JIR325(pJIR3118)
transconjugants had toxicity (141 to 400 minimal lethal doses)
comparable to that of their type D parent strain, CN1020 (141 to
200 minimal lethal doses). Supernatants from JIR325 were not
lethal under these conditions. These data confirmed that conju-
gative transfer of a native ε-toxin plasmid from a type D strain to
a type A strain can convert the latter to an ε-toxin-producing,
phenotypically type D strain whose products are lethal for mice.
Previous studies have revealed conjugative transfer of nu-
merous pCW3-related tetracycline resistance plasmids (1, 2,
11) and a cpe plasmid (6) between type A strains of C. perfrin-
gens. The potential for conjugative transfer of the ε-toxin plas-
mids was suggested previously since these plasmids have the
tcp conjugation genes (4, 16, 30), which are present in all
known conjugative plasmids from C. perfringens. In this study
we showed that two ε-toxin plasmids can transfer by conjuga-
tion. Conjugation was initially demonstrated by using a novel
genetic method to construct genetically marked derivatives of
these plasmids and then using the marker to select for
transconjugants. The efficiency of transfer was so high that in
subsequent matings it was possible to isolate transconjugants
without any selection for the donor, which has not been
achieved before in C. perfringens. In addition, this is the first
report of intraspecies conjugative transfer between type D and
type A strains of C. perfringens. Since the resultant transcon-
jugants from these matings could act as donors in subsequent
mating experiments, these 48-kb ε-toxin plasmids must encode
all of the functions required for conjugation. C. perfringens has
many different toxin-encoding plasmids, and many of these
plasmids carry the tcp conjugation locus (4, 16, 30). Based on
the current results and the highly conserved nature of the tcp
regions, it is highly likely that many, if not all, of these toxin
plasmids will prove to be conjugative.
The movement of an unmarked wild-type ε-toxin plasmid
into a type A recipient enabled us to analyze etx gene expres-
sion in a type A background. In this background, functional
ε-toxin was produced and the transconjugants were both ge-
notypically and phenotypically type D, as determined by mul-
tiplex PCR analysis and toxin assays. These results provide
experimental evidence, at least for type D strains, that the
different toxin types of C. perfringens result from the horizontal
transfer of toxin-carrying plasmids (7, 21); that is, as other
workers have suggested, type B, C, D, and E isolates of C.
perfringens may represent type A strains that have acquired a
conjugative toxin-encoding plasmid (7, 21). However, due to
the extensive genetic diversity found even in type A strains, a
view of strain typing based only on toxin carriage may be
simplistic and inadequate. Recently, analysis of the toxin plas-
mids of 23 type D isolates revealed considerable variation in
both the size and the number of large plasmids. Isolates car-
rying the etx, cpe, and cpb2 genes on a single plasmid or on
three distinct plasmids were identified (30). Clearly, a typing
system that can discriminate between the chromosomal back-
ground and the plasmid content of an isolate would be more
useful for strain typing and would provide a more accurate
picture of which genetic elements C. perfringens requires for
host specificity and virulence.
It was recently demonstrated that ε-toxin is responsible for
the lethality of type D culture supernatants in the intravenous
mouse lethality assay (29). In this study we demonstrated that
activated culture supernatants from type A-derived etx?
FIG. 3. Toxin production by pJIR3118-derived transconjugants. (A) Western blotting of culture supernatants with anti-ε-toxin MAb 5B7. Lanes
1 to 4 contained JIR325, CN1020, JIR12012, and JIR12013, respectively. (B) Quantification of ε-toxin in culture supernatants of transconjugants
JIR12012 and JIR12013 (lanes 1 and 2, respectively) and CN1020 (lane 3). The other lanes contained different concentrations of the purified
ε-toxin standard (50 to 500 ng). (C and D) Phospholipase C (Plc) and perfringolysin O (PfoA) activities of the strains indicated. The data are the
averages for three independent biological samples.
7536 HUGHES ET AL.J. BACTERIOL.
transconjugants caused ε-toxin-mediated MDCK cell cytotox-
icity and were lethal in the intravenous mouse model, even
though they had less ε-toxin than the parent type D strain. If
ε-toxin is the only type D-specific requirement for type D-
mediated enterotoxemia, then our type A-derived transconju-
gants should be virulent in a large animal model of disease.
However, if there is more to the relationship between disease
and strain type than just the production of ε-toxin, then the
transconjugants may not colonize the sheep gastrointestinal
tract and cause disease.
C. perfringens type A and type D isolates can coexist as
commensals in the gastrointestinal tracts of ruminants, which
provides ample opportunity for the conjugative transfer of
toxin plasmids. Although it is generally accepted that ε-toxin
production is required for type D-mediated disease to occur,
the precise process by which disease develops is not fully un-
derstood. The overgrowth of commensal ε-toxin-producing
strains in the gastrointestinal tract due to changes in feeding,
environmental stress, or the entry of exogenous type D strains
has been suggested to be the potential trigger for disease (3,
17, 35). Our results suggest an alternative scenario for the
sudden onset of enterotoxemic disease, as first suggested for
FIG. 4. ε-Toxin-mediated cytopathic effects of culture supernatants. Monolayers of MDCK cells were incubated for 24 h in the presence of
trypsin-treated culture supernatants from different strains before examination with an Olympus 1X71 inverted microscope. ImageJ 1.37v software
(http://rsb.info.nih.gov/ij/index.html) was used to view the images. (A) JIR325. (B) CN1020 ?etx::catP. (C) Medium alone. (D) CN1020.
(E) CN1020 pretreated with MAb 5B7. (F) CN1020 pretreated with ErmB-specific antibody. (G) JIR12012 [JIR325(pJIR3118)]. (H) JIR12012
pretreated with MAb 5B7. (I) JIR12012 pretreated with ErmB-specific antibody. (J) JIR4983 [JIR325(pJIR3118?etx::catP)]. (K) JIR4983
pretreated with MAb 5B7. (L) JIR4983 pretreated with ErmB-specific antibody. The culture supernatant from CN1020 was diluted 50-fold prior
to use. Bar, 100 ?m.
VOL. 189, 2007 CONJUGATIVE TRANSFER OF EPSILON-TOXIN PLASMIDS7537
human non-food-borne disease mediated by plasmid-deter- Download full-text
mined CPE production (6). Under certain conditions, adher-
ent resident C. perfringens type A cells may be converted into
ε-toxin-producing type D strains by conjugative transfer of an
ε-toxin plasmid from either resident or exogenous type D cells.
This process would exponentially increase the number of ε-toxin-
producing bacteria in the gastrointestinal tract. In particular, if
transfer can occur in vivo, then it would no longer be necessary
for an exogenous type D strain to have the ability to colonize
the gastrointestinal tract; the strain would simply need to be
able to survive long enough to transfer its toxin plasmid to the
resident adherent flora.
We thank Trudi Bannam, Wee Lin Teng, and Jennifer Parsons for
helpful discussions. Paul Hauer is thanked for kindly supplying MAbs
This research was supported by grant AI056177-03 and training
grant T32 AI060525-01A1 from the National Institute of Allergy and
Infectious Diseases and by funds from the Australian Research Coun-
cil Centre of Excellence in Structural and Functional Microbial
1. Abraham, L. J., and J. I. Rood. 1985. Molecular analysis of transferable
tetracycline resistance plasmids from Clostridium perfringens. J. Bacteriol.
2. Abraham, L. J., A. J. Wales, and J. I. Rood. 1985. Worldwide distribution of
the conjugative Clostridium perfringens tetracycline resistance plasmid,
pCW3. Plasmid 14:37–46.
3. Adamson, R. H., J. C. Ly, M. Fernandez-Miyakawa, S. Ochi, J. Sakurai, F.
Uzal, and F. E. Curry. 2005. Clostridium perfringens epsilon-toxin increases
permeability of single perfused microvessels of rat mesentery. Infect. Im-
4. Bannam, T. L., W. L. Teng, D. Bulach, D. Lyras, and J. I. Rood. 2006.
Functional identification of conjugation and replication regions of the tet-
racycline resistance plasmid pCW3 from Clostridium perfringens. J. Bacteriol.
5. Brynestad, S., and P. E. Granum. 1999. Evidence that Tn5565, which in-
cludes the enterotoxin gene in Clostridium perfringens, can have a circular
form which may be a transposition intermediate. FEMS Microbiol. Lett.
6. Brynestad, S., M. R. Sarker, B. A. McClane, P. E. Granum, and J. I. Rood.
2001. Enterotoxin plasmid from Clostridium perfringens is conjugative. Infect.
7. Canard, B., B. Saint-Joanis, and S. T. Cole. 1992. Genomic diversity and
organization of virulence genes in the pathogenic anaerobe Clostridium per-
fringens. Mol. Microbiol. 6:1421–1429.
8. Garmory, H. S., N. Chanter, N. P. French, D. Bueschel, J. G. Songer, and
R. W. Titball. 2000. Occurrence of Clostridium perfringens beta2-toxin
amongst animals, determined using genotyping and subtyping PCR assays.
Epidemiol. Infect. 124:61–67.
9. Hauer, P. J., and N. E. Clough. 1999. Development of monoclonal antibodies
suitable for use in antigen quantification potency tests for clostridial veter-
inary vaccines. Dev. Biol. Stand. 101:85–94.
10. Johnston, J. L., J. Sloan, J. A. Fyfe, J. K. Davies, and J. I. Rood. 1997. The
recA gene from Clostridium perfringens is induced by methyl methanesulpho-
nate and contains an upstream Cheo box. Microbiology 143:885–890.
11. Lyras, D., and J. I. Rood. 1996. Genetic organization and distribution of
tetracycline resistance determinants in Clostridium perfringens. Antimicrob.
Agents Chemother. 40:2500–2504.
12. Lyristis, M., A. E. Bryant, J. Sloan, M. M. Awad, I. T. Nisbet, D. L. Stevens,
and J. I. Rood. 1994. Identification and molecular analysis of a locus that
regulates extracellular toxin production in Clostridium perfringens. Mol. Mi-
13. McClane, B. A., F. A. Uzal, M. F. Miyakawa, D. Lyerly, and T. Wilkins. 2007.
The enterotoxic clostridia, p. 698–752. In M. Dworkin, S. Falkow, E. Rosen-
burg, K. H. Schleifer, and S. E. (ed.), The prokaryotes: a handbook on the
biology of bacteria, vol. 4. Springer-Verlag, New York, NY.
14. McDonel, J. L. 1980. Clostridium perfringens toxins (type A, B, C, D, E).
Pharmacol. Ther. 10:617–635.
15. Minami, J., S. Katayama, O. Matsushita, C. Matsushita, and A. Okabe.
1997. Lambda-toxin of Clostridium perfringens activates the precursor of
epsilon-toxin by releasing its N- and C-terminal peptides. Microbiol. Immu-
16. Miyamoto, K., D. J. Fisher, J. Li, S. Sayeed, S. Akimoto, and B. A. McClane.
2006. Complete sequencing and diversity analysis of the enterotoxin-encod-
ing plasmids in Clostridium perfringens type A non-food-borne human gas-
trointestinal disease isolates. J. Bacteriol. 188:1585–1598.
17. Niilo, L. 1980. Clostridium perfringens in animal disease: a review of current
knowledge. Can. Vet. J. 21:141–148.
18. O’Connor, J. R., D. Lyras, K. A. Farrow, V. Adams, D. R. Powell, J. Hinds,
J. K. Cheung, and J. I. Rood. 2006. Construction and analysis of chromo-
somal Clostridium difficile mutants. Mol. Microbiol. 61:1335–1351.
19. Payne, D. W., E. D. Williamson, H. Havard, N. Modi, and J. Brown. 1994.
Evaluation of a new cytotoxicity assay for Clostridium perfringens type D
epsilon toxin. FEMS Microbiol. Lett. 116:161–167.
20. Petit, L., M. Gibert, D. Gillet, C. Laurent-Winter, P. Boquet, and M. R.
Popoff. 1997. Clostridium perfringens epsilon-toxin acts on MDCK cells by
forming a large membrane complex. J. Bacteriol. 179:6480–6487.
21. Petit, L., M. Gibert, and M. R. Popoff. 1999. Clostridium perfringens: toxino-
type and genotype. Trends Microbiol. 7:104–110.
22. Purdy, D., T. A. T. O’Keeffe, M. Elmore, M. Herbert, A. McLeod, M. Bokori-
Brown, A. Ostrowski, and N. P. Minton. 2002. Conjugative transfer of clos-
tridial shuttle vectors from Escherichia coli to Clostridium difficile through
circumvention of the restriction barrier. Mol. Microbiol. 46:439–452.
23. Rood, J. I. 2007. Clostridium perfringens and histotoxic disease, p. 753–770. In
M. Dworkin, S. Falkow, E. Rosenberg, K.-H. Schleifer, and E. Stackebrandt
(ed.), The prokaryotes: a handbook on the biology of bacteria, vol. 4. Springer-
Verlag, New York, NY.
24. Rood, J. I. 1983. Transferable tetracycline resistance in Clostridium perfrin-
gens strains of porcine origin. Can. J. Microbiol. 29:1241–1246.
25. Rood, J. I., and S. T. Cole. 1991. Molecular genetics and pathogenesis of
Clostridium perfringens. Microbiol. Rev. 55:621–648.
26. Rood, J. I., E. A. Maher, E. B. Somers, E. Campos, and C. L. Duncan. 1978.
Isolation and characterization of multiply antibiotic-resistant Clostridium
perfringens strains from porcine feces. Antimicrob. Agents Chemother. 13:
27. Rood, J. I., V. N. Scott, and C. L. Duncan. 1978. Identification of a trans-
ferable resistance plasmid (pCW3) from Clostridium perfringens. Plasmid
28. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: a
laboratory manual, 2nd ed. Cold Spring Harbor Laboratory, Cold Spring
29. Sayeed, S., M. E. Fernandez-Miyakawa, D. J. Fisher, V. Adams, R. Poon, J. I.
Rood, F. A. Uzal, and B. A. McClane. 2005. Epsilon-toxin is required for
most Clostridium perfringens type D vegetative culture supernatants to cause
lethality in the mouse intravenous injection model. Infect. Immun. 73:7413–
30. Sayeed, S., J. Li, and B. A. McClane. 2007. Virulence plasmid diversity In
Clostridium perfringens type D isolates. Infect. Immun. 75:2391–2398.
31. Scott, P. T., and J. I. Rood. 1989. Electroporation-mediated transformation
of lysostaphin-treated Clostridium perfringens. Gene 82:327–333.
32. Sloan, J., T. A. Warner, P. T. Scott, T. L. Bannam, D. I. Berryman, and J. I.
Rood. 1992. Construction of a sequenced Clostridium perfringens-Escherichia
coli shuttle plasmid. Plasmid 27:207–219.
33. Smedley, J. G., III, D. J. Fisher, S. Sayeed, G. Chakrabarti, and B. A.
McClane. 2004. The enteric toxins of Clostridium perfringens. Rev. Physiol.
Biochem. Pharmacol. 152:183–204.
34. Songer, J. G. 1997. Clostridial diseases of animals, p. 153–182. In J. I. Rood,
B. A. McClane, J. G. Songer, and R. W. Titball (ed.), The clostridia: mo-
lecular biology and pathogenesis. Academic Press, London, United King-
35. Songer, J. G. 1996. Clostridial enteric diseases of domestic animals. Clin.
Microbiol. Rev. 9:216–234.
36. Stevens, D. L., J. Mitten, and C. Henry. 1987. Effects of ? and ? toxins from
Clostridium perfringens on human polymorphonuclear leukocytes. J. Infect.
37. Uzal, F. A. 2004. Diagnosis of Clostridium perfringens intestinal infections in
sheep and goats. Anaerobe 10:135–143.
7538HUGHES ET AL.J. BACTERIOL.