549

Journal of Food Protection, Vol. 63, No. 4, 2000, Pages 549–552

Copyright Q, International Association for Food Protection

Research Note

Ability of Lactobacillus and Propionibacterium Strains to

Remove Aflatoxin B1 from the Chicken Duodenum
HANI EL-NEZAMI,1† HANNU MYKKÄNEN,2 PASI KANKAANPÄÄ,1‡ SEPPO SALMINEN,1,3 AND JORMA AHOKAS1*

1Key Centre for Applied and Nutritional Toxicology, RMIT-University, GPO Box 2476V, Melbourne, Vic 3001, Australia; 2Department of Clinical
Nutrition, University of Kuopio, P.O. Box 1627, 70211 Kuopio, Finland; and 3Department of Biochemistry and Food Chemistry, University of Turku,
20014 Turku, Finland

MS 98-324: Received 11 December 1998/Accepted 7 July 1999

ABSTRACT
The ability of Lactobacillus rhamnosus strains GG and LC-705 to remove AFB1 from the intestinal luminal liquid medium
has been tested in vivo using a chicken intestinal loop technique. In this study, the GG strain of L. rhamnosus decreased AFB1
concentration by 54% in the soluble fraction of the luminal fluid within 1 min. This strain was more efficient in binding AFB1
compared with L. rhamnosus strain LC-705 (P , 0.05) that removed 44% of AFB1 under similar conditions. Accumulation
of AFB1 into the intestinal tissue was also determined. There was a 74% reduction in the uptake of AFB1 by the intestinal
tissue, in the presence of L. rhamnosus strain GG compared with 63% and 37% in the case of Propionibacterium freudenreichii
ssp. shermanii JS and L. rhamnosus strain LC-705, respectively. The complexes formed in vitro between either L. rhamnosus
strain GG or L. rhamnosus strain LC-705 and AFB1 were stable under the luminal conditions for a period of 1 h.

The occurrence of aflatoxin in feedstuff is common all
over the world (12). Aflatoxins have been detected as con-
taminants of crops before harvest, during harvesting and
drying, in storage, as well as after processing and manu-
facturing (3). Following ingestion of aflatoxin-contaminat-
ed feeds, accumulation of AFB1 and aflatoxin M1 into eggs
and milk has been reported (1, 10). Even low levels of
aflatoxins in feeds may result in decreased milk yield and
egg production by farm animals. In experimental animals
low levels of AFB1 have been shown to be immunosup-
pressive and to possess mutagenic and teratogenic activity
causing cancer and liver damage (2, 9). Thus, chronic ex-
posure to aflatoxins may not only significantly alter food
production and farm animal productivity, but consumption
of aflatoxin-contaminated food commodities may also rep-
resent a health risk to the consumer. It is therefore important
to reduce and/or prevent human exposure by developing
practical and effective methods to detoxify aflatoxin-con-
taminated foods and feedstuffs.
AFB1 appears to be absorbed rapidly from the small
intestine into mesenteric venous blood. Kumagai (8) in-
jected [3H]AFB1 directly into the stomach and various sites
of the small intestine of rats and measured the radioactivity
in the bile 30 min later. He demonstrated that the site of
absorption was the small intestine; and the absorption was
most efficient from the duodenum.
* Author for correspondence. Tel: 1 61 3 992 52650; Fax: 1 61 3 9663
6087; E-mail: Toxicology@rmit.edu.au.
† Present address: Department of Clinical Nutrition, University of Kuopio,
P.O. Box 1627, 70211 Kuopio, Finland.
‡ Present address: Department of Biochemistry and Food Chemistry, Uni-
versity of Turku, 20014 Turku, Finland.
In an effort to develop a practical method for aflatoxin
detoxification, we have investigated the ability of selected
dairy strains of lactic acid bacteria to remove AFB1 from
liquid media (4–6). Lactobacillus rhamnosus strains GG
and LC-705 removed a significantly greater amount of
AFB1, compared to other strains of either gram-positive or
gram-negative bacteria. The AFB1 removal by these two
strains was a rapid process. The aims of the current study
were to investigate: (i) the ability of the selected strains to
bind AFB1 under the conditions inside the duodenum; (ii)
the ability of the selected strains to influence AFB1 uptake
by the duodenal tissue; and (iii) the stability of the complex
formed in vitro between these strains and AFB1 under the
conditions inside the duodenum.
A strain of Propionibacterium freudenreichii ssp. sher-
manii JS was also included in this study.
MATERIALS AND METHODS
Bacterial strains and determination of bacterial concen-
tration. Bacteria used were L. rhamnosus strains GG (ATCC
53103) and LC-705 and P. freudenreichii ssp. shermanii JS. The
bacterial strains were obtained from Valio Ltd. (Helsinki, Finland)
as a freeze-dried powder. The concentration of lyophilized bacteria
(number of bacteria/g of powder) was measured by flowcytometry
using a Coulter Electronics EPICS XL cytometer equipped with
an air-cooled 488 nm argon-ion laser operating at 15 mW. Via-
bility was assessed using the method described by Virta et al. (11).
In vivo binding assay. The University of Kuopio Animal
Ethics Committee approved experiments. One-week-old broiler
chicks (Vilppulan Hybrid Ltd., Vilppula, Finland) were used in
these experiments. The chicks were anesthetized by an intraperi-
toneal injection with 0.1 ml/100 g body weight of Mebunat so-
550 EL-NEZAMI ET AL. J. Food Prot., Vol. 63, No. 4

TABLE 1. AFB1 recovered from the intestinal tissue and luminal fluid of 1-week-old chicks after intraduodenal injection of AFB1 with
or without bacteriaa
The concentration of AFB1 6 SD extracted from:

Soluble fraction
Duodenal tissueb of luminal fluidc Bacterial pelletc

Group 1 min 60 min 1 min 60 min 1 min 60 min

LBGGd only ND (5) NDg (5) ND (5) ND (5) ND (5) ND (5)

AFB1 only 0.27 6 0.09 (5) ND (5) 1.04 6 0.36 (5) 0.05 6 0.01 (5) ND (5) ND (5)
LBGGd 1 AFB1 0.07 6 0.05 (3) ND (5) 0.48 6 0.15 (3) ND (5) 0.76 6 0.04 (3) 1.38 6 0.16 (5)
LC705e 1 AFB1 0.17 6 0.11 (2) ND (5) 0.58 6 0.10 (2) 0.08 6 0.06 (3) 0.54 6 0.10 (2) 1.07 6 0.12 (3)
PJSf 1 AFB1 0.10 6 0.05 (5) ND (5) 0.67 6 0.13 (5) 0.13 6 0.02 (2) 0.55 6 0.11 (5) 1.24 6 0.06 (2)
a Intraduodenal dose; 1010 CFU 1 1.5 mg of AFB1 in 0.5 ml of PBS. The original number of chickens per group was 5 but the number
of successful experiments used to obtain these data is indicated in brackets.
b Concentration expressed as mg AFB /g tissue.
1
c Concentration expressed as mg AFB /ml.
1
d L. rhamnosus strain GG.
e L. rhamnosus strain LC-705.
f P. freudenreichii ssp. shermanii JS.
g ND, below the detection limit of the HPLC method used (200 pg AFB /ml methanol).
1

lution (Orion, Finland) containing 60 mg/ml of sodium pentobar-
bital. This dose was sufficient to keep the animal under anesthesia
for 60 min. A 2–3-cm cross-sectional cut in the lower abdominal
region was made and a 5–7-cm-long segment of the duodenum
around the pancreas was separated by two ligatures. Before tight-
ening of the second ligature the test solutions were injected into
the loop.
Chickens were divided into three groups per bacterial strain
studied (two tests and one control). For the first test group (n 5
5), the dose was 0.5 ml of phosphate-buffered saline (PBS, pH
7.4) containing 1.5 mg of AFB1. The second test group was as-
signed to examine the ability of the bacterial strains (n 5 5 for
each strain and time point) to bind AFB1 in vivo; 0.1 g of lyoph-
ilized bacteria (1010 CFU) was suspended in 0.25 ml of PBS and
injected into the duodenal loop immediately followed by an in-
jection of 0.25 ml of PBS containing 1.5 mg AFB1 (final concen-
tration of AFB1 was 1.5 mg/0.5 ml). The chicks were killed either
within 1 min or 60 min after injection. As controls, one group of
chicks (n 5 5) was injected with 0.5 ml of bacterial suspension
and killed after 60 min.
In vivo stability of the AFB1–bacterial complex. The sta-
bility of the AFB1–bacterial complex formed in vitro was exam-
ined under the conditions inside the duodenum. For this purpose,
lyophilized bacteria (1010 CFU) were suspended in 1.5 ml of PBS
containing 1.5 mg AFB1. The mixture was incubated at 378C for
60 min after which the suspension was centrifuged. The amount
of AFB1 in the supernatant was quantified to verify that AFB1
was bound to the bacteria. The bacterial pellets with the bound
AFB1 were then resuspended in 0.5 ml of PBS and injected into
the duodenal loop of chicks (n 5 5 per strain) that were killed 1
or 60 min after the injection. The amount of AFB1 in both intes-
tinal tissue and the supernatant fluid of the luminal content were
quantitated as in the method of El-Nezami et al. (4), and the
amount of the AFB1 released was calculated.
In vitro stability of AFB1–bacterial complex. The in vivo
stability was compared with the in vitro stability of the AFB1–
bacterial complex. In this experiment after incubation of AFB1
with bacteria as above, the supernatant was removed and analyzed
for AFB1 residues. The bacterial pellet was suspended in water
and adjusted to pH 5 with 0.1 M lactic acid to simulate the pH
of the loop content. The mixture was then incubated for a period
of 60 min, after which the suspension was centrifuged and the
amount of AFB1 in the supernatant was quantitated as in El-Ne-
zami et al. (4), and the amount of AFB1 released was calculated.
Extraction of aflatoxin. The contents of the intestinal loop
were rinsed out with 5 ml of PBS after the chicks were killed,
the intestinal contents were centrifuged (at 2,000 3 g and 48C for
15 min), and the supernatant was extracted twice with 2 ml chlo-
roform. The two extracts were then pooled and evaporated to dry-
ness and the residue was reconstituted in 0.5 ml methanol. In order
to confirm that the removal of AFB1 from the intestinal content
was a result of binding of the toxin to the bacteria, the pellet was
suspended in 2 ml chloroform, shaken vigorously, and centrifuged
at 2,000 3 g. The chloroform layer was evaporated to dryness
and the residue was then suspended in 0.5 ml methanol.
The ability of the bacterial strains to influence the uptake of
AFB1 by the intestinal tissue was also investigated. The duodenal
tissue was homogenized with 4 ml chloroform. The homogenate
was filtered, the chloroform evaporated, and the residue reconsti-
tuted in 0.5 ml methanol.
All samples were analyzed by high-performance liquid chro-
matography using the conditions described in El-Nezami et al. (4).
Statistical analysis. The Mann-Whitney test was used to test
for statistical differences for AFB1 binding and release between
different bacterial strains.
RESULTS AND DISCUSSION
The amount of AFB1 taken up by the intestinal tissue
in the presence of L. rhamnosus strain GG at 1 min (0.07
mg/g duodenal tissue) was less than that taken up in the
case of the positive control (0.27 mg/g duodenal tissue) (Ta-
ble 1). This indicates over 70% reduction in the uptake of
AFB1 by the duodenal tissue when AFB1 was injected in
the presence of L. rhamnosus strain GG. The AFB1 uptake
by duodenal tissue was also reduced by either L. rhamnosus
strain LC-705 or P. freudenreichii ssp. shermanii JS. How-
ever, the reduction in the uptake was 37 and 63% for L.
J. Food Prot., Vol. 63, No. 4 IN VIVO BINDING OF AFLATOXIN B1 BY SELECTED BACTERIAL STRAINS
TABLE 2. Percentage of AFB1 bound when incubated with the
bacteria either in vitro or in vivo
% AFB1 bound 6 SD when AFB1
incubated with the bacteria either:
In vivoa
In vitro,b
1 min 60 min 60 min
Bacteria incubation incubation incubation
L. rhamnosus
strain GG 51 6 2c 92 6 1c 80 6 2
L. rhamnosus
strain LC-705 36 6 2c 71 6 2 77 6 4
P. freudenreichii ssp.
shermanii JS 37 6 1d 82 6 1e 22 6 2
a 1010 CFU/ml bacteria was incubated with 1.5 mg AFB1/ml for
1 or 60 min in the duodenum of 1-week-old chicks.
b 1010 CFU/ml bacteria was incubated with 5 mg AFB /ml for 60
1 be recovered in the case of L. rhamnosus strain LC-705
min at 378C and pH 7.3.
c Strains bind significantly less AFB in vivo than in vitro (P ,
1 1). These findings indicate that the complex formed be-
0.05).
d Strains binds significantly more AFB in vivo than in vitro (P
1
, 0.05).
rhamnosus strain LC-705 and P. freudenreichii ssp. sher-
manii JS, respectively. The reduction in AFB1 uptake by
the duodenal tissue suggests that the bacterial strains tested
in this study can reduce the amount of AFB1 absorbed into
the duodenal tissue and hence may affect aflatoxin bio-
availability.
Hsieh and Wong (7) demonstrated that over 50% of the
dose of 0.05 mg/kg [14C]AFB1 disappeared from the duo-
denal region within 60 min. In this study, 97% of the AFB1
dose disappeared from the duodenal region (luminal fluid
and duodenal tissue) within 60 min.
L. rhamnosus strain GG decreased AFB1 by 54% in
the soluble fraction of the luminal fluid in 1 min (Table 1).
This strain was more efficient in reducing AFB1 content in
the luminal fluid compared to the other two strains tested
(P , 0.05). Both L. rhamnosus strain LC-705 and P. freu-
denreichii ssp. shermanii JS were only able to reduce AFB1
by 44% and 36% in the soluble fractions, respectively. It
appears that this phenomenon is the result of binding of
AFB1 by bacteria, because the difference between the
amount of AFB1 added and that recovered from the duo-
denal tissue and luminal contents can be accounted for by
recovery from bacterial pellets as presented in Table 1. Af-
ter 60 min incubation in the duodenal loop approximately
92, 71, and 82% of the AFB1 dose (1.5 mg AFB1/ml) can
be recovered from the bacterial pellets of L. rhamnosus
strain GG, L. rhamnosus strain LC-705, and P. freudenrei-
chii ssp. shermanii JS, respectively (Table 1).
There is a significant difference between the amount
of AFB1 recovered from the bacterial pellets when the
strains tested were incubated in vivo either for 1 or 60 min
(Table 2). When the incubation was continued for 60 min,
the binding efficiency of L. rhamnosus strain GG and P.
freudenreichii ssp. shermanii JS were significantly in-
creased. This was not the case for L. rhamnosus strain LC-
705 where no significant difference was detected between
551
incubations for 1 min or 60 min. The difference in binding
efficiency was not noted when these strains were incubated
in vitro (4). A reduction was noted in the efficiency of the
two Lactobacillus strains to bind AFB1 in vivo when com-
pared to their binding efficiency in vitro (Table 2). As for
P. freudenreichii ssp. shermanii JS, the binding ability of
the bacteria was significantly enhanced when the incubation
was carried out in vivo either for 1 or 60 min when com-
pared to the in vitro binding. Such differences between in
vivo and in vitro binding are not surprising, because there
are more interfering factors at work in vivo.
In the absence of added bacteria, 3% of the AFB1 dose
was present in the luminal contents and no residue from
tissue extracts was found after 60 min, suggesting that
AFB1 was absorbed. No residue was present in the luminal
contents when AFB1 was injected in combination with L.
rhamnosus strain GG, but 5 and 8% of the AFB1 dose could
and P. freudenreichii ssp. shermanii JS, respectively (Table
tween AFB1 and L. rhamnosus strain GG is more stable
compared to that formed with L. rhamnosus strain LC-705
and P. freudenreichii ssp. shermanii JS. The ability of the
bacteria to bind AFB1 rapidly may result in a reduction of
AFB1 absorption via duodenal mucosa.
The AFB1–bacterial complex that was incubated in
vivo (in the intestinal lumen) appeared to be more stable
than the complex incubated in vitro. For all the strains test-
ed, there was no release of AFB1 back to the duodenal
content when the AFB1–bacterial complex was incubated
in the intestinal loop for 60 min. When the complex was
incubated in vitro for the same period, a release of AFB1
back into the media was noticed with all strains tested. The
release from P. freudenreichii ssp. shermanii JS (64.4 6
0.9%) was significantly higher than that from L. rhamnosus
strain GG (13.9 6 4.2%) or strain LC-705 (12.0 6 4.4%).
This study is the first demonstration of the ability of
selected bacterial strains to bind AFB1 under in vivo con-
ditions. The findings from this study indicate that the con-
ditions inside the duodenum enhance the ability of both L.
rhamnosus strain GG and P. freudenreichii ssp. shermanii
JS to bind AFB1. These conditions also improve the sta-
bility of the complex formed between the strains tested and
AFB1. The results show that once the bacteria bind the
AFB1, it is not released back into duodenal fluid. We sug-
gest that binding of AFB1 by bacteria is a promising way
to reduce the bioavailability of AFB1 from contaminated
feed. Development of this approach has the potential to
reduce the accumulation of AFB1 and thus decrease its tox-
icity to both humans and animals. This approach would be
of benefit to human and animal health in developing coun-
tries where reduction of aflatoxin contamination is needed
but is limited by available resources and technology. Suc-
cessful application of this approach will not only benefit
human health but may result in increased productivity of
farm animals.
ACKNOWLEDGMENTS
The skillful assistance of Mrs. R. Hall and Mrs. D. Donohue in the
preparation and reviewing of this manuscript is gratefully acknowledged.
552 EL-NEZAMI ET AL.
We thank Food Science Australia and the Center for International Mobil-
ity, Finland for their financial support during the term of this study. As-
pects of this work are subject to a current International Patent Application
(application no. 5005/97).
REFERENCES
1. Allcroft, R., and R. B. A. Carnaghan. 1963. Groundnut toxicity: an
examination for toxin in human food products from animals fed toxic
groundnut meal. Vet. Res. 75:259–263.
2. Eaton, D. L., and E. P. Callagher. 1994. Mechanisms of aflatoxin car-
cinogenesis. Annu. Rev. Pharmacol. Toxicol. 34:135–172.
3. Ellis, W. O., J. P. Smith, and B. K. Simpson. 1991. Aflatoxins in food:
occurrence, biosynthesis, effects on organisms, detection, and meth-
ods of control. Crit. Rev. Food Sci. Nutr. 30:403–439.
4. El-Nezami, H., P. Kankaanpää, S. Salminen, and J. Ahokas. 1998.
Ability of dairy strains of lactic acid bacteria to bind food carcino-
gens. Food Chem. Toxicol. 36:321–326.
5. El-Nezami, H., P. Kankaanpää, S. Salminen, and J. Ahokas. 1998.
Physico-chemical alterations enhance the ability of dairy strains of
lactic acid bacteria to remove aflatoxins from contaminated media. J.
Food Prot. 61:466–468.
J. Food Prot., Vol. 63, No. 4
6. El-Nezami, H., S. Salminen, and J. Ahokas. 1996. Biologic control
of food carcinogen using Lactobacillus GG. Nutr. Today 31:41S–43S.
7. Hsieh, P. H., and J. J. Wong. 1994. Pharmacokinetics and excretion
of aflatoxins, p. 73–88. In D. L. Eaton and J. D. Groopman (ed.), The
toxicology of aflatoxins: human health, veterinary, and agricultural
significance. Academic Press, San Diego, Calif.
8. Kumagai, S. 1989. Intestinal absorption and excretion of aflatoxin in
rats. Toxicol. Appl. Pharmacol. 97:88–97.
9. Robens, J. F., and J. L. Richard. 1992. Aflatoxins in animal and human
health. Rev. Environ. Contam. Toxicol. 127:69–74.
10. Rodricks, J. V., and L. Stoloff. 1977. Aflatoxin residues from contam-
inated feed in edible tissues of food producing animals, p. 75–89. In
J. V. Rodricks, C. W. Hesseltine, and M. A. Mehlman (ed.), Myco-
toxins in human and animal health. Pathotox. Publ., Park Forest
South, Ill.
11. Virta, M., S. Lineri, P. Kankaanpää, M. Karp, K. Peltonen, J. Nuutila,
and E. M. Lilius. 1998. Determination of complement-mediated kill-
ing of bacteria by viability staining and bioluminescence. Appl. En-
viron. Microbiol. 64:515–516.
12. Wood, G. E. 1989. Aflatoxins in domestic and imported foods and
feeds. J. Assoc. Off. Anal. Chem. 72:543–548.