FEMS Immunology and Medical Microbiology 24 (1999) 43^47

Bacillus thuringiensis serotype H34 isolated from human and

insecticidal strains serotypes 3a3b and H14 can lead to death of
immunocompetent mice after pulmonary infection
Eric Hernandez a , Franc°oise Ramisse b; *, Thierry Cruel c , Robert le Vagueresse a ,
Jean-Didier Cavallo d
a
Laboratoire de Biologie, HIA Percy, 92141 Clamart, France
b
Centre d'Etudes du Bouchet Laboratoire de Microbiologie, BP 3, 91710 Vert-le-Petit, France
c
Laboratoire d'Anatomo Pathologie, H.I.A du Val de Graêce, 75014 Paris, France
d
Laboratoire de Biologie, HIA Begin, Saint-Mandeè, France

Received 24 September 1998 ; received in revised form 8 January 1999; accepted 12 January 1999

Abstract

In 1995, we isolated a strain of Bacillus thuringiensis serotype H34 from severe human tissue necrosis. This bacterium was
able to induce myonecrosis in immunosuppressed mice after cutaneous infection. Its potential pathogenicity for
immunocompetent hosts was investigated in a mouse model of pulmonary infection. Mice infected intranasally by a
suspension containing 108 spores died within 8 h in a clinical toxic-shock syndrome. In the same conditions, infection with a
mutant without crystalline toxin, with the supernatant from a culture containing 108 bacteria ml31 and by the insecticidal strain
serotypes 3a3b or H14 led to identical results. Lower inocula simply induced a local inflammatory reaction with bacterial
persistence observed during the course of 10 days. z 1999 Federation of European Microbiological Societies. Published by
Elsevier Science B.V. All rights reserved.

Keywords : Bacillus thuringiensis; Pulmonary infection ; Hemolysin

1. Introduction Infection in humans is unusual, and apart from

The ubiquitous soil bacterium Bacillus thuringien-
sis encodes a diverse array of pesticidal proteins
widely used around the world for insect pest control.
Recently, DNA sequences encoding the delta-endo-
toxin have been included in plants to promote resist-
ance to insects.
* Corresponding author. Tel.: +33 (1) 6990-8381;
Fax: +33 (1) 6493-5266; E-mail: f.ramisse@wanadoo.fr
gastrointestinal tract infections or those following
laboratory contamination, there are only two clinical
reports of B. thuringiensis infection [1,2]. In 1995, we
isolated a strain of B. thuringiensis var. konkukian
(serotype H34) from soft tissue necrosis following
severe war wounds caused by a land mine explosion.
The ability of this strain to induce myonecrosis in
immunosuppressed mice after cutaneous infection
has been previously described [1]. Since B. thuringien-
sis spores and its parasporal body are commonly

0928-8244 / 99 / $20.00 ß 1999 Federation of European Microbiological Societies. Published by Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 8 - 8 2 4 4 ( 9 9 ) 0 0 0 0 5 - X

FEMSIM 996 21-4-99

44 E. Hernandez et al. / FEMS Immunology and Medical Microbiology 24 (1999) 43^47
used as aerosolised biopesticides [3], the aim of this
study was to evaluate the potential pathogenicity of
this strain and of the insecticidal serotypes 3a3b and
H14 for immunocompetent mice after pulmonary ex-
perimental infection.
2. Materials and methods
2.1. Mice
Mice used were 5-week-old female BALB/c
(Charles River, Saint-Aubin-les-Elbeuf, France)
kept in a biosafety containment facility in groups
of ¢ve, with sterile water and food.
2.2. Bacteria
B. thuringiensis serotype H34-konkukian was iso-
lated from a severe war wound infection [1]. B. thur-
ingiensis H34 without crystalline toxin was a sponta-
neous mutant isolated from the same patient. Both
strains were identi¢ed with biochemical tests and H-
serotyping performed by the WHO collaborating
centre for Entomopathogens (Dr. Lecadet, Uniteè
des Bacteèries Entomopathogeénes, Institut Pasteur,
Paris, France). B. thuringiensis serotype H12 was ob-
tained from a clinical specimen and was considered
as clinically irrelevant. B. thuringiensis serotypes
3a3b and H14 were obtained from Abbot laborato-
ries.
2.3. Cultures
Spore suspensions were prepared from a 10-day-
old culture on poor agar medium (yeast extract, 10 g
l31 ; NaCl, 5 g l31 ; Agar, 20 g l31 ) suspended in
sterile water and incubated for 1 h at 65³C in order
to kill vegetative forms. Dilutions were made in ster-
ile water to obtain inocula of 105 ^108 spores per
mouse. Supernatants were obtained from a 24-h sta-
tionary-phase cultures containing 108 CFU ml31
centrifuged 15 min at 7000Ug and ¢ltered on 0.22
Wm.
2.4. Infection
For each strain, four groups of ¢ve mice were
FEMSIM 996 21-4-99
infected intranasally, under light ether anaesthesia,
by suspensions containing from 105 ^108 spores in
50 Wl. Five mice were infected with the supernatant.
Each experiment was repeated three times.
2.5. Bacterial counts
Lungs were dissected from the main bronchi, di-
luted in sterile PBS (Sigma), and plated on trypti-
case-soy agar medium. Bacterial counts from lung
homogenates were expressed in log10 CFU ml31 as
mean þ S.E.
2.6. Histological examinations
Histological examinations were performed imme-
diately after death or 48 h after infection. Mice were
killed by injection of a pentothal overdose. Lungs
were perfused with 500 Wl of 10% formaldehyde by
intratracheal injection, taken out and ¢xed in 10%
formalin. The tissues were embedded in para¤n
blocks and sectioned at 4^5 Wm thickness. The sec-
tions were mounted onto regular slides for staining
with haematoxylin-eosin-sa¡ron or Gram stain.
2.7. Haemolytic activity
Supernatants were centrifuged on Centriprep 30
(Amicon, Epernon, France) in order to concentrate
and separate the proteins by molecular weight. Hae-
molytic activities were quanti¢ed by incubating 50 Wl
of serial 2-fold dilutions of supernatants with 50 Wl
of 0.5% packed and washed rabbit erythrocytes. The
mixtures were incubated 60 min at 37³C and left to
sediment for 60 min at room temperature [4]. The
titre was given by the last dilution giving 100% hae-
molysis. Erythrocytes with sterile broth were used as
negative control and erythrocytes with distilled water
as positive control.
3. Results
3.1. Infections with spore suspensions
All the mice instilled with 108 spores of B. thur-
ingiensis var. konkukian died within 8 h with a clin-
ical toxic-shock syndrome. Autopsy showed large
 E. Hernandez et al. / FEMS Immunology and Medical Microbiology 24 (1999) 43^47 45

Table 1
Pulmonary bacterial counts of Bacillus thuringiensis H34
Time 108 spores per mouse 107 spores per mouse 105 spores per mouse 103 spores per mouse
postinfection (log10 8) (log10 7) (log10 5) (log10 3)
4 h 7.30 þ 0.12 ND ND ND
8 h dead ND ND ND
24 h 6.87 þ 0.17 4.33 þ 0.14 2.81 þ 0.26
48 h 7.09 þ 0.25 4.79 þ 0.24 3.17 þ 0.02
10 days 5.20 þ 0.16 3.16 þ 0.13 1.10 þ 0.08
Values are mean þ S.E. of mice per point. ND, not determined.

haemorrhagic lique¢ed lesions of the lungs and the
liver. Histological examinations of the lung revealed
lesions of acute bronchitis associated with ulceration,
injuries of the mucociliary apparatus, oedema and
alveolar damage. These lesions were associated with
a neutrophilic in¢ltrate. Blood cultures obtained
after intracardiac puncture were positive and re-
vealed a pure culture of B. thuringiensis. In the
same conditions, inocula from 105 to 107 spores
per mouse led only to a local in£ammatory reaction,
with a bacterial persistence observed for at least 10
days (Table 1). Bacterial counts gave the same results
when the lung homogenates were heated or not at
65³C in order to kill vegetative forms.
Infection by the mutant without parasporal body
inclusion led to the same results, including for the
pulmonary bacterial counts.
Instillation of 108 spores of B. thuringiensis sero-
type 3a3b led to a lethality of 80%. The lethality was
only 4O% with B. thuringiensis serotype H14 when
using the same inoculum (Table 2). Histological ex-
aminations revealed identical lesions to those ob-
served with B. thuringiensis serotype H34.
Infection by 108 spores of B. thuringiensis H12,
simply induced a lung in£ammatory reaction.
3.2. Nasal instillation of supernatants
Since the obvious lesions were lung haemorrhagic
su¡usions, we supposed, as previously described in
the mice cutaneous infectious model [1], that the se-
creted haemolysins of B. thuringiensis were the toxins
involved in the pathogenicity. Two di¡erent haemo-
lysins, closely related to those of Bacillus cereus, have
been described [5^7]. While one, thiol-dependent, is
inhibited by cholesterol, the other is not. Instillation
with the supernatant of a stationary phase culture
(24 h at 37³C) containing 108 CFU of B. thuringien-
sis H34 led to an 80% lethality rate. This ¢gure
dropped to 10% when cholesterol was added to the
supernatant at the concentration of 60 Wg ml31 . In
contrast, the supernatant of B. thuringiensis H12 was
not able to kill the mice.
After ¢ltration of the supernatants on Centriprep
30, the haemolytic activity corresponding to the frac-
tion higher than 30 kDa was: 1/256 for B. thurin-
giensis H34 crystal‡ ; 1/128 for B. thuringiensis sero-
type 3a3b; 1/32 for serotype H14; and 1/2 for B.
thuringiensis serotype H12. The concentrated frac-
tion was able to kill mice in 30 min for B. thurin-
giensis H34 and in less than 5 h for B. thuringiensis
3a3b.
The fractions corresponding to the proteins of a

Table 2
Titration of the haemolytic activity in the supernatants (24 h after ¢ltration on Centriprep 30), and lethality rates induced by 108 spores
ml31
Strain Haemolysin titre Lethality at 24 h postinfection due to 108 spores
B. thuringiensis H34 cry‡ 1/256 100%
B. thuringiensis 3a3b 1/128 80%
B. thuringiensis H14 1/32 40%
B. thuringiensis H12 1/2 0%
Titres are the last dilution giving 100% haemolysis.

FEMSIM 996 21-4-99

46 E. Hernandez et al. / FEMS Immunology and Medical Microbiology 24 (1999) 43^47
molecular weight under 30 kDa were not haemolytic
and were not able to kill mice after experimental
infection.
3.3. Inactivated supernatant
The lethal fraction of the supernatant of B. thur-
ingiensis H34 heated at 70³C for 10 min did not
express haemolytic activity. When instilled into
mice, the fraction was unable to kill them.
4. Discussion
Infection by B. thuringiensis is very uncommon in
humans, but this bacterium is not identi¢ed in rou-
tine culture. In this particular case, B. thuringiensis
has been identi¢ed in our medical laboratory by the
Vitek System (Vitek, St. Louis, MO) which was us-
ing, at the time, the industrial software. When using
the medical software, this bacterium was identi¢ed as
B. cereus. Compared to B. thuringiensis which is re-
sistant to penicillin G and ampicillin, B. cereus ex-
presses a higher resistance pattern and is resistant to
penicillin G, ampicillin, ticarcillin and cefalotin, but
remains sensitive to piperacillin [1^8].
Except for B. thuringiensis H12, commonly recov-
ered in soils and used as control, all the strains tested
in this protocol were, to di¡erent degrees, pathogenic
for mice. The commercial strains of B. thuringiensis
are commonly used in agriculture and in forestry
with no case of pathogenicity described for over 30
years, but infectious concentrations used for infec-
tion in this protocol are very high compared to those
used in agriculture. However, it has never been de-
scribed that any serotype of B. thuringiensis was able
to kill mice when applied at a high concentration by
the pulmonary route. B. thuringiensis H34 isolated
from our patient is not used as a biopesticide, and
is perhaps a rare and unusual isolate expressing a
high level of a B. cereus-like haemolytic toxin. How-
ever, this argument is not available for B. thuringien-
sis 3a3b and H14 which were obtained from a com-
mercial source.
In the case of B. thuringiensis H34, pathogenicity
in mice does not seem to be correlated with the pro-
duction of the delta-endotoxin. A decrease in lethal-
ity rate observed when cholesterol was added to the
FEMSIM 996 21-4-99
supernatant indicates that the thiol-dependent hae-
molysin is probably implicated in the pathogenicity
of B. thuringiensis serotypes H34, 3a3b, and H14.
Surprisingly, the concentration of 107 organisms
per mouse did not produce enough toxin to kill the
mice. This observation suggests that a high concen-
tration of haemolysin is required for lethality. This
hypothesis was con¢rmed by the instillation of con-
centrated supernatants which were able to kill mice
in less than 30 min for B. thuringiensis H34 and 5 h
for serotype 3a3b. However, haemolysin is probably
not involved alone in pathogenicity: a lower inocu-
lum induced pulmonary lesions with bacterial persis-
tence observed during a 10-day period. This persis-
tence was not associated with bacterial multiplication
(Table 1). The observation that bacterial counts in
lung homogenates, after heating at 65³C or not, were
the same, suggests that the inoculum is maintained
as sporulated forms. The exact localisation of spores
in lungs is unknown, but we suppose that the spores
are ingested, but not destroyed, by the alveolar phag-
ocytic cells.
Ongoing experiments will be performed to con¢rm
this hypothesis and to investigate the correlation be-
tween the serotypes, the amount of thiol-dependent
haemolysin, the phospholipase production, and the
pulmonary pathogenicity in mice.
Acknowledgments
We gratefully acknowledge Karine David, Agneés
Labarre and Rosy Smith for their technical assis-
tance.
References
[1] Hernandez, E., Ramisse, F., Cruel, T., Ducoureau, J.P., Alon-
so, J.M. and Cavallo, J.D. (1998) Bacillus thuringiensis serovar
H34-konkukian superinfection: report of one case and exper-
imental evidence of pathogenicity in immunosuppressed mice.
J. Clin. Microbiol. 36 (7), 2138^2139.
[2] Damgaard, P.H., Granum, P.E., Bresciani, J., Torregrossa,
M.V., Eilenberg, J. and Valentino, L. (1997) Characterization
of Bacillus thuringiensis isolated from infections in burn
wounds. FEMS Immunol. Med. Microbiol. 18, 47^53.
[3] Aronson, A.I., Beckman, W. and Dunn, P. (1986) Bacillus
thuringiensis and related insect pathogens. Microbiol. Rev.
50, 1^24.
 E. Hernandez et al. / FEMS Immunology and Medical Microbiology 24 (1999) 43^47 47

[4] Coolbaugh, J.C., Wende, R.D. and Williams, R.P. (1972) Mi-
crotitration of Bacillus cereus hemolysin. Appl. Microbiol. 24,
997^998.
[5] Coolbaugh, J.C. and Williams, R.P. (1978) Production and
characterization of two hemolysins of Bacillus cereus. Can.
J. Microbiol. 24, 1289^1295.
[6] Honda, T., Shiba, A., Seo, S., Yamamoto, J., Matsuyama, J.
and Miwatani, T. (1991) Identity of hemolysins produced by
Bacillus thuringiensis and Bacillus cereus. FEMS Microbiol.
Lett. 63, 205^209.
[7] Matsuyama, J., Yamamoto, K., Miwatani, T. and Honda, T.
(1995) Monoclonal antibody developed against a hemolysin of
Bacillus thuringiensis. Microbiol. Immunol. 39, 619^622.
[8] Cavallo, J.D., Hernandez, E. and Dubrous, J.P. (1997) Bacil-
lus cereus agent d'infection des plaies de guerre. Meèd. Armeèes
25 (5), 373^378.

FEMSIM 996 21-4-99