lntemational

EISEVIER

Journal

of FoodMicrobiology

International Journal of
Food Microbiology 28 (1995) 169-185

Bacteriocins: modes of action and potentials in food

preservation and control of food poisoning
Tjakko Abee

al* , Lothar

Krockel

b, Colin Hill

aFood Chemistry and -Microbiology Section, Department of Food Science, Wageningen Agricultural
University, Bomenweg 2, 6703 HD Wageningen, The Netherlands
b Bundesanstalt fiir Fleischforschung, Kulmbach, Germany
Department of Microbiology, University College, Cork, Ireland

1. Introduction

Lactic acid bacteria (LAB) play an essential role in the majority of food
fermentations, and a wide variety of strains are routinely employed as starter
cultures in the manufacture of dairy, meat, vegetable and bakery products. One of
the most important contributions of these microorganisms is the extended shelf life
of the fermented product by comparison to that of the raw substrate. Growth of
spoilage and pathogenic bacteria in these foods is inhibited due to competition for
nutrients and the presence of starter-derived inhibitors such as lactic acid, hydrogen peroxide and bacteriocins (Ray and Daeschel, 1992). Bacteriocins, are a
heterogenous group of anti-bacterial proteins that vary in spectrum of activity,
mode of action, molecular weight, genetic origin and biochemical properties.
Currently, artificial chemical preservatives are employed to limit the number of
microorganisms capable of growing within foods, but increasing consumer awareness of potential health risks associated with some of these substances has led
researchers to examine the possibility of using bacteriocins produced by LAB as
biopreservatives.
The major classes of bacteriocins produced by LAB include: (I) lantibiotics, (II)
small heat stable peptides, (III) large heat labile proteins, and (IV) complex
proteins whose activity requires the association of carbohydrate or lipid moieties
(Klaenhammer, 1993). Significantly however, the inhibitory activity of these substances is confined to Gram-positive bacteria and inhibition of Gram-negatives by
these bacteriocins has not been demonstrated,
an observation which can be

* Corresponding
author. Tel. +31 (- 317) 484981;
Tjakko.Abee@ALGEMEEN.LENM.WAV.NL

Fax

0168-1605/95/$09.50 0 1995 Elsevier Science B.V. All rights reserved

SSDZO168-1605(95)00055-0

+31

(- 317) 484893;

e-mail

170

T. Abee et al. /ht.

J. Food Microbiology 28 (1995) 169-185

cell wall

outer membrane
surface layer
teichoic acid

peptidoglycan

cytoplasmic membrane
GRAM-POSITIVE
Fig. 1. Schematic presentation
lipopolysaccharide.

GRAM-NEGATIVE
of the cell envelope of Gram-positive and Gram-negative bacteria. LPS,

explained by a detailed analysis and comparison of the composition of Gram-positive and Gram-negative bacterial cell walls (Fig. 1). In both types the cytoplasmic
membrane which forms the border between the cytoplasm and the external
environment, is surrounded by a layer of peptidoglycan which is significantly
thinner in Gram-negative bacteria than in Gram-positive bacteria. Gram-negative
bacteria possess an additional layer, the so-called outer membrane which is
composed of phospholipids, proteins and lipopolysaccharides
(LPS), and this
membrane is impermeable to most molecules. Nevertheless, the presence of porins
in this layer will allow the free diffusion of molecules with a molecular mass below
600 Da. The smallest bacteriocins produced by lactic acid bacteria are approximately 3 kDa and are thus too large to reach their target, the cytoplasmic
membrane (Klaenhammer, 1993; Stiles and Hastings, 1991). However, Stevens et
al. (1991) and Ray (1993) have demonstrated that SuZmonellu species and other
Gram-negative bacteria become sensitive to nisin after exposure to treatments that
change the permeability barrier properties of the outer membrane (see below).
This review will focus on the mode of action of lantibiotics (class I) and class II
LAB bacteriocins and their potentials in food preservation and control of food
poisoning.

2. Lantibiotics produced by lactic acid bacteria (LAB)

The best studied member of this group is Nisin A, a 34-residue antibacterial
peptide that is produced by several strains of Lacfococc~ Zuctis and strongly
inhibits the growth of a wide range of Gram-positive bacteria (Hurst, 1981; Jung
and Sahl, 1991). The mature peptide displays several unusual features, such as the
dehydrated residues dehydroalanine and dehydrobutyrine, which are derived from
serine and threonine residues, respectively, and lanthionine and p-methyllanthionine residues which form five intramolecular thioether bridges (Jung and
Sahl, 1991). Molecular structures similar to that in mature nisin are found in three
other bacteriocins produced by L. luctis (lacticin 4811, Luctobucilfus sake (lactocin

T. Abee et al. /ht.

J. Food Microbiology 28 (199.5) 169-185

171

S) and Cumobactetium piscicofu (carnocin UI49) together forming the Class I LAB
bacteriocins (Piard et al., 1992; Mortvedt et al., 1991; Stoffels et al., 1992;
Klaenhammer, 1993).
In Gram-positive bacteria nisin has been shown to act on energized membrane
vesicles to disrupt the proton motive force (PMF), inhibit uptake of amino acids,
and cause release of accumulated amino acids (Jung and Sahl, 1991). Nisin Z, a
natural nisin variant, was isolated from L. 1ucti.s subsp. luctis strain NIZO 22186.
The gene for this lantibiotic, designated n&Z, has been cloned and its nucleotide
sequence was found to be identical to that of the precursor nisin gene apart from a
single mutation resulting in a substitution of His2 for Asn2 in the mature
polypeptide (Mulder et al., 1991). Exposure of the food pathogen List&a monocytogenes to nisin Z resulted in immediate loss of cellular potassium ions, depolarization of the cytoplasmic membrane, hydrolysis and partial efflux of cellular ATP
(Abee et al., 1994) demonstrating that in this species, the primary target for nisin Z
is the cytoplasmic membrane.
Recently, Mortvedt-Abildgaard et al. (1995) showed that production and bactericidal activity of lactocin S are highest at acid pH values. Mode of action studies
indicated that bactericidal (bacteriolytic) activity was confined to pH values of six
and slightly lower. This is possibly due to the influence of two positively-charged
(lysine) and two negatively-charged (glutamate and aspartate) amino acids and two
histidine residues with a positive charge at pH 6 or lower (pK, = 6 for His) and
having a major role in determining the effective charge of the peptide which is
crucial for activity.
The lantibiotic carnocin U149 produced by C. piscicolu U149 was also shown to
act at the cytoplasmic membrane of L. luctis (Stoffels et al., 1994). Here studies
with L. fuctis derivatives, harbouring different segments of the nisin gene cluster,
indicated that membrane-located
proteins encoded by specific genetic determinants within this gene cluster may function as a receptor for carnocin prior to its
bacteriolytic activity on the cytoplasmic membrane.
2.1. Pore formation by kin

The cytoplasmic membrane of the bacterial cell is the primary target for nisin
activity. This lantibiotic has been shown to associate with non-energised liposomes
with the greatest interaction being observed with negatively charged phospholipids.
This indicated that the initial association of these positively charged peptides with
the membrane may also be, in part, charge dependent (Garcia-Garcera et al., 1993;
Driessen et al., 1995). A truns-membrane orientation is not adopted prior to the
application of a membrane potential (negative inside) of approximately - 80 to
- 100 mV (Jung and Sahl, 1991). The threshold potential might be influenced by
various parameters such as the pH and the phospholipid composition of the
membrane. Nisin A and Z displayed increased activity at acidic pH values and
could permeabilize membranes at membrane potentials which were very low and
even completely absent (Gao et al., 1991; Garcia-Garcera et al., 1993; Abee et al.,
1994). Nisin A can form transient multistate pores with diameters ranging from 0.2

172

CM

T. Abee et al. /ht.

.I, Food Microbiology 28 (1995) 169-185

B
CM

--b

Fig. 2. Models for pore formation by lantibiotics (A) and non-Iantibiotic Class II LAB bacteriocins (B).
(AI The mature nisin molecule is schematically presented with the N-terminal (N) 1-19 amino acid
residue part containing one Lys+, connected via a flexible hinge region to the 21-34 C-terminal (C)
amino acid residue part, which contains two Lys+. The barrel-stave mechanism involves three discrete
steps: 1. binding of nisin molecules to the membrane; 2. AY (inside negative&dependent insertion into
the membrane; and 3. aggregation of monomers resulting in the formation of a water-filled pore. (B)
Model for pore formation by Class II LAB bacteriocins. I. the proteinaceous receptor is involved in
bacteriocin binding; 2. PMF-independent
insertion of the bacteriocin into the membrane; and 3.
aggregation of monomers in the membrane results in pore formation. The light and dark shaded halves
represent the hydrophilic and hydrophobic regions of the amphiphilic peptides, respectively. See text
for details.

to 1.2 nm in black lipid membranes when tram-negative potentials are applied.

Such pores would allow the passage of hydrophilic solutes with molecular masses
up to 0.5 kDa. Indeed nisin A and Z have been shown to induce leakage of ATP
from target cells (Jung and Sahl, 1991; Abee et al., 1994). It has been proposed
that lantibiotics belong to the group of cytolytic pore-forming proteins, which
function through a so-called barrel-stave mechanism (Fig. 2; Ojcius and Young,
1991). Recently, Driessen et al. (1995) proposed a model where the mode of action
of nisin was determined by the phospholipid composition of the membrane. In
liposomes composed of phosphatidylcholine (PC), nisin was suggested to act (even
in the absence of a PMF) as an anion-selective carrier. The action of nisin against
membranes containing anionic phospholipids such as phosphatidylglycerol (PG),
involves the local perturbation
of the bilayer structure, and a A+ (inside
negative)-dependent
insertion of nisin into the membrane. Furthermore,
this
model suggested that electrostatic interactions between the nisin molecules and
the phospholipids could bring the lipid head-groups into the pore lining (Driessen
et al., 1995).
Nisin has also been shown to act on Clustridium and Bacillus spores, but in

T Abee et al. /ht.

J. Food Microbiology 28 (1995) 169-185

173

these cases the exact mechanism of action has not been elucidated (Hurst, 1981).
Sublethal heat treatment of spores causes sufficient injury to induce sensitivity to
nisin (Jung and Sahl, 1991, Ray and Daeschel, 1992).

3. Class II LAB bacteriocins

Class II LAB bacteriocins are small heat-stable, non-lanthionine containing
membrane-active peptides. The mature bacteriocins are predicted to form amphiphatic helices with varying amounts of hydrophobicity and p-sheet structure.
Klaenhammer (1993) defined three subgroups within this class of bacteriocins that
are: ZZa, Listeria-active peptides with a consensus sequence in the N-terminal of
-Tyr-Gly-Asn-Gly-Val-. Zlb, poration complexes formed by oligomers of two different proteinaceous peptides and, ZZc, Thiol-activated peptides requiring a reduced
cysteine residues for activity.
3.1. Mode of action of pediocin and other Class ZJ Listeria-active

peptides

Pediococci are widely applied in the fermentation of meat and vegetables. The
best studied bacteriocin produced by this genus is pediocin PA-l from Pediococcus
acidifactici and recently this was shown to be identical to pediocin AcH (Klaenhammer, 1993; Ray and Daeschel, 1992). This bacteriocin shares sequence similarities with various other important anti-listerial bacteriocins (Sakacin A and P,
Leucocin A, and Carnobacteriocin BMl and B2) produced by LAB associated with
meats. These (pediocin-like) peptides are active against a broad range of Grampositive bacteria including L. monocytogenes.
The function of the consensus
sequence in the N-terminal region of these mature peptides (see above) is
unknown. Mature pediocin PA-l is a highly hydrophobic, positively charged
peptide consisting of 44 amino acids. Pediocin PA-l acts on the cytoplasmic
membrane thereby dissipating ion gradients and inhibiting transport of amino
acids in sensitive cells: The same activity was observed in membrane vesicles
derived from these cells, whereas liposomes made from the membrane lipids were
not affected (Chikindas et al., 1993).
Pediocin PA-l contains two disulfide bonds and it was shown that the bond
between the cysteine residues at positions 24 and 44 is essential for activity.
Preliminary studies on the membrane permeabilizing effects of a number of these
pediocin-like bacteriocins indicated that sakacin A and P (Chikindas et al., 19931,
leucocin and carnobacteriocin B2 and BMl (van Belkum and Abee, unpublished
data) may function like pediocin PA-l. It was concluded that pediocin PA-1 forms
hydrophilic pores in the cytoplasmic membrane of target cells in a protein
receptor-mediated,
voltage-independent
manner (Chikindas et al., 19931, analogous to action of Lactococcin A (LcnA), a bacteriocin produced by L. Zactis (van
Belkum et al., 1991). LcnA is small 54-amino-acid hydrophobic peptide that
specifically inhibits the growth of other L. lactis subspecies. The effect of purified
Len A on whole lactococcal cells and membrane vesicles indicated that the

174

T Abee et al. /ht.

J. Food Microbiology 28 (1995) 169-185

bacteriocin increases the permeability of the cytoplasmic membrane of sensitive

lactococci in a voltage-independent
way. The specificity of Len A for lactococci
seems to stem from the fact that the bacteriocin recognises a Lactococcus-specific
membrane receptor protein. The receptor protein may well be involved in pore
formation since Len A could not permeabilize liposomes composed only of
phospholipids obtained from sensitive lactococcal cells (van Belkum et al., 1991).
Furthermore,
protease K treatment was shown to render membrane vesicles
insensitive to LcnA, which is apparently due to proteolytic digestion of the
bacteriocin receptor (Venema et al., 1994). A model for the mode of action of
these class II LAB bacteriocins is shown in Fig. 2 (van Belkum et al., 1991;
Chikindas et al., 1993; Venema et al., 1994).

4. Bacteriocins: applications in food preservation

This section will deal with recent developments in this area and describe novel
strategies with potential application in the food industry.
4.1. Application of bacteriocins in the preservation of dairy products
A problem often encountered in cheese production is the outgrowth of butyric
acid bacteria such as Clostridium tyrobutyricum (Hurst, 1981). In dairy practice,
nitrate is commonly added to cheesemilk to prevent outgrowth of clostridia spores.
This chemical preservative can be very efficiently replaced by nisin A. Outgrowth
of C. tyrobutyricum spores in nitrate-free Gouda cheese was completely prevented
when a nisin A producing strain was added to the starter culture (10% nisin A
producers) (Hugenholtz and de Veer, 1991). Nisin A is also an effective inhibitor
of L. monocytogenes, and growth of this pathogen was effectively inhibited by
Nisin A in camembert (Maisnier-Patin et al., 1992) and in cottage cheese at 4C as
well as at 37C (Benkerroum and Sandine, 1988). These results strongly suggest a
potentially wider role for nisin A in the future preservation of a variety of dairy
products.
Recently, the relevant physicochemical and biological properties of nisin A and
nisin Z were analysed (de Vos et al., 1993). Identical MICs of nisin A and nisin Z
were found with all tested indicator strains of six different species of Gram-positive bacteria. However, at concentrations
above the MICs, with nisin Z the
inhibition zones obtained in agar diffusion assays with all tested indicator strains
were larger than those obtained with nisin A. These results suggested that nisin Z
has better diffusion properties than nisin A in agar. Whether nisin Z will perform
better as a biopreservative in certain foods than nisin A remains to be investigated.
The application of nisin in dairy foods which require lactic acid starter bacteria
presents a problem because the wide spectrum of inhibition associated with nisin
includes LAB themselves. An alternative approach which could be used to control
specific pathogens or spoilage organisms in dairy foods is to employ bacteriocins
with a highly specific activity range. The pediocin-like, heat-stable bacteriocin

T. Abee et al. /ht.

J. Food Microbiology 28 (1995) 169-185

175

enterocin 1146, which is produced by Enterococcus faecium DPC1146, is extremely

active against L. monocytogenes at levels which have no effect on lactococcal
starters (Parente and Hill, 1992a, b). E. f aecium DPC1146 was used to ferment
milk, which was subsequently pasteurized. The bacteriocin is produced in milk and
is unaffected by the heat treatment. This milk was mixed with fresh milk and used
for cheese making. The lactococcal starters were shown to grow and produce acid
normally in this mix, whereas L. monocytogenes introduced in at the same time
was rapidly killed. This inhibitory effect was not observed when a variant of
DPC1146 was used which no longer produced the bacteriocin.
4.2. Biopreservation of meat products
Over the past three decades there has been an increasing research interest in
the development of nitrite-free meat curing systems. The principle concern with
the use of nitrite for curing of meat is the eventual formation of carcinogenic
N-nitrosamines. Recently, attempts have been made to use nisin A as an altemative to nitrite. While the use of this bacteriocin alone was not successful, promising
results were obtained when it was combined with reduced levels of nitrite: loo-250
ppm nisin A combined with 120 ppm nitrite was more effective than the conventional 156 ppm nitrite (Shahidi, 1991). Nisin A is apparently not the bacteriocin of
choice for meat preservation in contrast to its effectiveness in dairy products (see
above). Bacteriocins produced by LAB associated with meat and meat fermentations such as Pediococcus, Leuconostoc, Camobacterium and Lactobacilh
spp.
are likely to have much greater potential as meat preservatives (Stiles and Hastings, 1991; Shahidi, 1991; Yousef et al., 1991).
L. monocytogenes is a food-borne pathogen which is ubiquitous in the environment and can be isolated from foods of different origin, including meats and meat
products. In meat processing plants it may be present in slicing rooms and
eventually contaminate pasteurized products during slicing and packaging. Recently, some biopreservation techniques have been applied to meat products and
these involved the introduction of a competitive microflora of LAB as protective
cultures for chill-stored ready-to-eat meat products, including bacteriocin producing LAB; and the use of purified anti-listerial bacteriocins added directly as
natural food additives.
Lactobacilh
sake Lb674, a mildly acidifying lactic acid bacterium originally
isolated from meat, produces the bacteriocin sakacin 674, which is identical to
sakacin P and very similar to pediocin PA-1 (Krockel, 1992; Tichaczek et al., 1994,
Holck et al., 1994). On vacuum-packed sliced Bologna-type sausage stored at
+ 7C L. sake Lb674 produces detectable amounts of bacteriocin and delays or
completely inhibits the growth of L. monocytogenes when inoculated at levels of at
least lo-lo6 LAB/g, while bacteriocin negative LAB had no inhibitory effect on
growth of this organism (Fig. 3). As a purified additive, sakacin 674 exhibits a
marked initial effect against L. monocytogenes and reduces listerial growth during
storage of this fermented meat product (data not shown).
Yousef et al. (1991) investigated the growth of L. monocytogenes in packed

176

T. Abee et al. /ht.

0

J. Food Microbiology 28 (1995) 169-185

bat-

LAB

0 bat+

LAB
7.0

o-o-o
0

- 6.5

14

21

28

Days

of

5.0

Storage

Fig. 3. Growth of L. monocytogenes on vacuum-packaged Bologna-type sausage at 7C in the presence

of non-bacteriocin producing lactic acid bacteria (bat- LAB) and bacteriocin producing LAB (bat+
LAB), and pH progression during storage. The sausage was artificially contaminated during slicing with
a pool of four different serovars of L. monocytogenes. (bat- LAB): a pool of 20 bacteriocin negative
LAB strains from meat samples. (bat+ LAJ3): sakacin 674 producing strain of Lactobacih sake
Lb674.

wiener sausage, a fully-cooked, cured meat product which is susceptible to contamination by L. monocytogenes before packaging. These researchers provided evidence that Pediococcus inoculants or purified pediocin can function as biopreservatives to eliminate Gram-positive pathogenic bacteria in cooked meats during
extended refrigerated storage.
4.3. Biopreservation of fkh
The application of nisin A in the preservation of fish products has been studied
by Taylor et al. (1990) who showed that nisin treatment of cod, herring, and
smoked mackerel fillets inoculated with Clostridium botulinum spores brought
about a delay in toxin production of 5 days at 10C but only by half a day at 26C.
Nisin treatment did not interfere with growth of non-pathogenic bacteria and in all
samples botulinurn toxin was formed before spoilage was evident.
The effects of nisin Z, carnocin U149 and bavaricin A on bacterial growth and
shelf life of brined shrimp was recently evaluated and compared with those of a
benzoate-sorbate
solution and a control with no added preservatives (Einarsson
and Lauzon, 1995). Typically this product contains 3 to 6% NaCl and sorbic and
benzoic acids in concentrations from 0.05 to l.O%, with pH ranging from 5 to 6,
and is stored at temperatures from 0 to 6C. The benzoate-sorbate
solution
preserves the brined shrimp for the whole storage period (59 days). The shelf life

T. Abee et al. / ht. J. Food Microbiology 28 (199.5) 169-185

117

of the shrimp in the absence of preservatives was found to be 10 days. Carnocin

U149 had no influence on shelf life, while crude bavaricin A (a cell-free supernatant of Lactobacilh
bauaricus MI 401) extended the shelf life to 16 days.
Significantly, when crude or purified nisin Z was applied to the same material the
shelf life was extended to 31 days. Such results offer clear perspectives for the
biopreservation of certain fish products with nisin Z.

5. Factors affecting LAB bacteriocin

action

5.1. Intrinsic food factors

The action of bacteriocins against sensitive microorganisms is influenced to a

large degree by factors such as pH, cell concentration, lipid content, proteolytic
enzymes, and liquid vs. solid system (Ray and Daeschel, 1992). The effiency of
nisin Z against cells of L. monocytogenes was recently shown to be significantly
reduced in the presence of di- and trivalent cations such as Mg2+, Ca2+ or Gd3+
(Abee et al., 1994) which may interact with the negatively charged phospholipid
headgroups of PG and cardiolipin present in the cytoplasmic membrane (Harwood
and Russel, 1984; OLeary and Wilkinson, 1988). This can result in inhibition of
electrostatic interactions between positive charges on the bacteriocins and the
negatively charged headgroups of the phospholipid molecules, and/or neutralization of the negative charges of the headgroups inducing a condensation of these
lipids resulting in a more rigid membrane (Abee et al., 1994). The presence of
these di- and trivalent ions in foods could potentially reduce the efficiency of nisin
action against Gram-positive spoilage bacteria and pathogens.
5.2. Effect of temperature

The action of nisin Z is also dependent on the temperature. The rate of Nisin
Z-induced K+ efflux from cells of L. monocytogenes grown at 30C was shown to
be severely reduced at decreased temperatures. The ordering of the lipid hydrocarbon chains which occurs at lower temperatures resulting in a decrease in membrane fluidity are probably responsible for the reduced nisin Z efficiency observed
(Abee et al., 1994). L. monocyfogenes adapts to low temperature growth by
increasing the proportion of short and/or branched fatty acyl chains of the lipids
thereby maintaining an optimum fluidity (Gounot, 19911, an adaptation which may
well be responsible for the remaining detectable activity of nisin Z against cells
grown at 4C (Abee et al., 1994). This is in line with the observation that similar
MIC values for nisin Z against food pathogens and food spoilage bacteria are
found when cells are grown in BHI or in low-fat milk and at high or low
temperatures (Table 1). The necessary adaptations at the level of the cytoplasmic
membrane for growth at low temperature allows nisin Z to act efficiently against a
broad range of sensitive bacteria over a wide range of temperatures.

178

T. Abee et al. / Int. J. Food Microbiology 28 (1995) 169-185

Table 1
Minimal inhibitory concentration (ME, fig/l) of nisin Z for food spoilage micro-organisms
food-borne pathogens grown in BHI or in low-fat milk at various temperatures

and

Temperature
7C

21C

400
400
400
10
5

400
400
400

800
400
200

800
400

10
10
800
400

NG
25
400
800

800

1200

400

400

1200

30C

Growth in BHI a
Bacillus cereus
Lactobacillus brevis
Lactobacillus plantarum
Brochotti
thermospacta
Pediococcus acidilactici
Listeria innocw
Listeria monocytogenes Scott A

Growth in milk
Bacillus cereus
Lkteria monocytogenes

Scott A

a The initial inoculum was approx. lo4 to 10 cells per ml. Mic values in BHI were determined using
OD measurements and the MIC values in low-fat milk were determined using plate counting. NG, no
growth possible at this temperature; -, not determined.

5.3. Resistance to bacteriocins

The complete nisin gene cluster in L. lactis consists of eleven genes with the
order nk4BTCZPZXFEG,
many of which have been implicated in nisin biosynthesis and nisi and the newly identified nisE, nisF and nisG having a putative role in
producer protection (Kuipers et al., 1993; Engelke et al., 1992; Siegers and Entian,
1995).
A number of other nisin resistance mechanisms have been described. Many
Gram-positive bacteria have been shown to be resistant to nisin due their ability to
synthesize an enzyme, nisinase, which could inactivate nisin. The enzyme was
isolated from several Bacillus spp. and was shown to be a dehydropeptide
reductase since it specifically reduced the C-terminal dehydroalanyllysine of nisin
to alanyllysine (Hurst, 1981). Another resistance mechanism involves adaptation of
cells to high concentrations of bacteriocins. Recently, Ming and Daeschel (1993)
evaluated the spontaneous nisin resistance frequencies in eight common foodbome
pathogenic and spoilage bacteria and characterized the phenotypic responses of a
derivative of L. monocytogenes Scott A resistant to high levels of nisin. In BHI
medium, spontaneous nisin resistance frequencies were in the range of 10e6 to
lo-* when cells were exposed to nisin at concentrations between 2 and 8 times the
MIC values. Detailed characterization of a resistant mutant of strain Scott A which
was obtained by a stepwise increase in exposure to nisin, revealed that changes had
occurred in the bacterial membrane structure i.e. the mutant had a higher phase
transition temperature, a higher percentage of straight chain fatty acids and a
lower percentage of branched chain fatty acids. As a result the fluidity of the

T. Abee et al. /ht.

J. Food Microbiology 28 (1995) 169-185

179

membrane was decreased which apparently resulted in a decreased efficiency of

nisin pore formation in this nisin resistant mutant (Ming and Daeschel, 1993).
Spontaneous sub-populations of variants (mutants) of L. monocytogenes resistant to pediocin AcH, mesenterocin 52, curvaticin 13 and plantaricin Cl9 were
reported recently (Ray and Daeschel, 1992; Rekhif et al., 1994). The occurrence of
spontaneous resistant mutants of L. rnonocytogenes to the latter three bacteriocins
was estimated to be in the range of 10e3 to 10e4, and strikingly, these mutants
showed cross resistance to the three bacteriocins. This resistance characteristic was
stable through many generations, even in the absence of the bacteriocins (Rekhif
et al., 1994). Interestingly, all the mutants appeared to be as sensitive to nisin as
the parental strains. Whether the high frequency of resistant mutants against these
pediocin-like bacteriocins is due to the loss of (proteinaceous)
receptor sites
remains to be elucidated.

6. Novel strategies in biopreservation

6.1. Paired starter cultures composed of nlsin producing and nisin resistant LAB

Dutch Gouda cheese is susceptible to the clostridium-associated

butyric acid
fermentation and nisin would appear to be an ideal inhibitor as an alternative to
nitrate which is currently used. However, direct addition of this bacteriocin to
cheese-milk is prohibited in the Netherlands. An attractive permissible approach
would be the use of nisin-producing starters during the fermentation. Unfortunately however, no nisin-producing LAB have the flavour-generating, eye-forming,
acidifying activities, and bacteriophage-resistance
necessary for Gouda cheese
manufacture. Recently, the genetic information for nisin production and immunity
was introduced into industrial strains selected from the complex mixed starter
cultures that are used in the Dutch dairy industry (Rauch and de Vos, 1992;
Hugenholtz and de Veer, 1991; Hugenholtz et al., 1995). Conjugative mobilization
was employed to transfer a transposon harbouring the genetic information for nisin
production and immunity to the citrate-utilizing component of the starter (L. lactis
subsp. lactis (biovar. diacetylactis)) and nisin immunity to the remaining component L. lactis subsp. cremoris which plays an essential role in proteolysis during
cheese production (Fig. 4). The cheese-making transconjugants harbouring the
nisin production/immunity
determinants were subsequently shown to have retained all previously mentioned traits essential in Gouda cheese manufacture and
thus concentrated starters were developed composed of the two strains. By varying
the ratios of the two strains, the level of nisin produced could be conveniently
manipulated. Resulting cheeses were thus protected from development of Clostridium tyrobutyricum and Staphylococcus aureus during the whole period of ripening
(Hugenholtz et al., 1995).
A similar approach was recently applied in industrial-scale cabbage fermentations in the USA (Harris et al., 1992). In the traditional process L. mesenteroides

180

T. Abee et al. /ht.

J. Food Microbiology 28 (1995) 169-185

L. lactis ssp. lactis

L. lactis ssp. cremoris

~~~

~~~

lac+ nis sue -

lac _nis sue +

1. factis ssp. cremoris

a
0

lac+

lac+
nis

nis

SW +

.suc+

Fig. 4. Conjugal transfer of nisin immunity and sucrose fermentation. Nisin immunity (nis) and genetic
information for growth on sucrose (Sue+) was introduced via conjugation in the lactose-negative
(Lac-) proteolytic Lactococcus lacks spp. cremoris strain. Growth on lactose (plasmid-encoded lac+,
indicated by closed circle), growth on sucrose (sue+ ) and immunity to nisin (nis) (transposon, indicated
by the closed bar) are the available food-grade selection markers, and lac-, sue- and nisS are the
respective counterparts.

dominates early in the fermentation and subsequently this genus is displaced by a

more acid tolerant homofermentative
Lactobaciflus plan&rum which completes
the fermentation. Adequate flavour development however, requires that the latter
strain does not dominate too soon in the process, therefore Harris et al. (1992)
developed a paired-starter
culture system which involved a nisin-resistant L.
mesenteroides and a nisin producing L. luctis strain. Here nisin production was
sufficient to retard proliferation of naturally present L. plunturum. In subsequent
studies Breidt et al. (1993) isolated L. mesenteroides strains which were resistant to
high levels of nisin (up to 25,000 IU/ml in broth) by a mechanism which did not
involve destruction of the bacteriocin. Use of these strains might allow the
production or application of nisin at sufficiently high levels in sauerkraut fermentations to prolong inhibition of L. plan&rum with a concomitant extension of the
heterofermentative
phase.
6.2. Application of hydrostatic pressure, electroporation and chelating agents in
combination with bacteriocins
In 1986, Kordel and Sahl reported that Escherichiu coli became sensitive to
nisin when the outer membrane was disrupted (Kordel and Sahl, 1986). Since then
several researchers have shown that the use of chelating agents, such as EDTA,

T. Abee et al. / ht. J. Food Microbiology 28 (1995) 169-185

181

which bind magnesium ions in the lipopolysaccharide layer, gives rise to Gramnegative bacteria with increased susceptibility to bacteriocins, antibiotics and
detergents (Stevens et al., 1991; Ray, 1993). In recent years ultrahigh hydrostatic
pressure (UHP) and pulsed electric field (PEF), because of their antimicrobial
effect have been investigated as possible nonthermal methods of food preservation
(Morris, 1993). Both UHP and PEF destroy microbial cells by destabilizing the
structural and functional integrity of the cytoplasmic membrane. Significantly,
Kalchayanand et al. (1994) reported that when the UHP- and PEF-related injury
was sublethal, surviving cells of Gram-negative bacteria had become sensitive to
pediocin AcH/PA-1 and nisin; members of two distinct bacteriocin groups. Schved
et al. (1994) showed recently, that permeabilization of the outer membrane of E.
coli and Salmonella typhimurium with EDTA without altering the cytoplasmic
membrane, rendered these cells sensitive to nisin but not pediocin SJ-1. This could
be attributed to a lack of a pediocin-receptor
(Chikindas et al., 1993) in the
cytoplasmic membrane of these Gram-negative bacteria. Apparently, bacteriocins
in combination with UHP and PEF have greater antibacterial effectiveness than
bacteriocins or UHP and PEF alone.
6.3. Heterologous expression of bacteriocins and construction of multiple-bacteriocinproducing LAB
The lactacin F complex, composed of LafA and LafX peptides, is produced by
Lactobacillus johnsonii VP111088 and is active against five other Lactobacillus
species and Enterococcus fuecalis (Klaenhammer, 1993). These peptides are pro-

cessed from prepeptides with similar N-terminal leader sequences characterized by

a Gly-Gly - * Xaa+ cleavage site, which resembles that of other bacteriocins. Of
all the class II bacteriocins, the N-terminal leader sequences of the lactacin F
peptides and that of the pediocin-like, listeria-active bacteriocins A, BMl, and B2
produced by C. pkcicolu LV17 (Quadri et al., 1994; Worobo et al., 19941, are
highly similar suggesting that systems for export and proteolytic processing may be
interchangeable.
Indeed, Allison et al. (1995) could show that lactacin F was
expressed in C. piscicolu and that lactacin F production (LafA and LafX) occurred
simultaneously with carnobacteriocin production via the carnobacteriocin excretion
and processing machinery. Transformants of C. pkcicola LV17 which produced
lactacin F in combination with the carnobacteriocins,
did not exhibit activity
beyond their normal range of sensitive indicator strains. In these experiments the
lactacin F operon, composed of the structural bacteriocin genes and putative
immunity protein were introduced into the heterologous host. Recently, other
reports of this type of heterologous expression have included the production of
lactacin F by Leuconostoc geZidum UAL187-22 (Klaenhammer, 1993) and helveticin J by L. johnsonii NCK64 (Fremaux and Klaenhammer, 1994). It was
suggested that the heterologous processing/excretion
mechanisms could provide a
basis for cloning broad-range bacteriocins into LAB which are used as starters in
fermentations, thereby increasing the safety and shelf life of these fermented food
products (Klaenhammer, 1993; Allison et al., 1995).

182

T. Abee et al. /ht.

J. Food Microbiology 28 (1995) 169-185

6.4. Protein engineering of bacteriocins

A protein engineering strategy was developed by Kuipers et al. (1992) which

enabled the production of mutant nisin Z species by site-directed mutagenesis.
Several mutants of nisin Z were produced, purified and characterized.
The
introduction or substitution of Ser or Thr residues was shown to lead to mutants
containing other or additional dehydrated residues and these variants showed
different antimicrobial activities compared to the wild-type nisin Z. Furthermore,
nisin Z variants (N27K and H31K nisin Z> with higher solubility at pH values
above 6 were produced, which retained the antimicrobial activity of nisin Z
(Kuipers et al., 1992). Protein engineering thus has significant potential in the
development of novel variants of existing bacteriocins which can be selected for
very specific applications.

7. Bacteriocins:

future prospects

The application of bacteriocins from lactic acid bacteria in combination with

traditional methods of preservation and proper, hygienic processing could be
effective in controlling spoilage and pathogenic bacteria, particularly human
pathogens such as L. monocytogenes, in a variety of food products. However, a
number of problems such as low production levels and instability in certain
environments/foods
need to be addressed. Recombinant DNA technology is
currently applied, to enhance production, to transfer of bacteriocin genes to other
species, and for mutation and selection of bacteriocin variants with increased
and/or broader activity spectra.
Some bacteriocin-producing
strains can be applied as protective cultures in a
variety of food products. For example, well characterized, homofermentative,
mildly acidifying, bacteriocinogenic LAB are ideal candidates for biopreservation
of meats where modification of the product is undesirable. However, relatively
high levels of these cultures may be required for protection against some pathogens.
In these cases bacteriocin producers should be selected which do not negatively
influence product taste and appearance when incorporated at high numbers. These
problems can be avoided if purified bacteriocins or inactivated cultures are used
directly as natural food additives, however additional hurdles may have to be
inchrded in order to prevent bacteriocin-resistant pathogens from growing.
Before bacteriocins can be applied in foods their cytolytic abilities should be
assessed in detail. This is a very important issue since recently a cytolysin produced
by E. fuecalis was described that possesses both hemolytic and bacteriocin activities (Gilmore et al., 1994).
Continued study of the physical and chemical properties, mode of action and
structure-function
relationships of bacteriocins is necessary if their potential in
food preservation is to be exploited. Further research into the synergistic reactions
of these compounds and other natural preservatives, in combination with advanced
technologies such as PEF and UHP could result in replacement of chemical

T. Abee et al. /ht.

J. Food Microbiology 28 (1995) 169-185

preservatives, or could allow less severe processing (e.g. heat) treatments,

still maintaining adequate microbiological safety and quality in foods.

183

while

Acknowledgements

The authors thank Edmund Kunji for the preparation of Fig. 2, Jeroen Hugenholtz for suggestions and for the supply of Fig. 4, Todd Klaenhammer for
unpublished results, and Aidan Coffey for advice and critical reading of the
manuscript. The AAIR Concerted Action PL920630 is thanked for support.

References
Abee, T., Rombouts, F.M., Hugenholtz, J., Guihard, G. and Letellier, L. (1994) Mode of action of nisin
Z against Listeria monocytogenes Scott A grown at high and low temperatures. Appl. Environ.
Microbial. 60, 1962-1968.
Allison, G.E., Worobo, R.W., Stiles, M.E. and Klaenhammer, T.R. (1995) Heterologous expression of
the lactacin F peptides by Carrwbacterium pi&cola LV17. Appl. Environ. Microbial. 61, 1371-1377.
Benkerroum, R. and Sandine, W.E. (1988) Inhibitoty action of nisin against Listeria monocytogenes. J.
Dairy Sci. 71, 3237-3245.
Breidt, F., Crowley, K.A. and Flemming, H.P. (1993) Isolation and characterization of nisin-resistant
Leuconostoc mesenteroides for use in cabbage fermentations. Appl. Environ. Microbial. 59, 37783783.
Chikindas, M.L., Garcia-Garcera, M.J., Driessen, A.J.M., Ledeboer, A.M., Nissen-Meyer, J., Nes, I.F.,
Abee, T., Konings, W.N. and Venema, G. (1993) Pediccin PA-l, a bacteriocin from Pediococcus
acidizactici PACl.0, forms hydrophilic pores in the cytoplasmic membrane of target cells. Appl.
Environ. Microbial. 59, 3577-3584.
de Vos, W.M., Mulder, J.W.M., Siezen, R.J., Hugenholtz, J. and Kuipers, O.P. (1993) Properties of
nisin Z and distribution of its gene, nisZ, in Lactococcus lactis. Appl. Environ. Microbial. 59,
213-218.
Driessen, A.J.M., van den Hooven, H.W., Kuiper, W., van de Kamp., M., Sahl, H.-G., Konings, R.N.H.
and Konings, W.N. (1995) Mechanistic studies of Iantibiotic-induced permeabilization of phospholipid vesicles. Biochemistry 34, 1606-1614.
Einarsson, H. and Lauzon, H.L. (1995) Biopreservation of brined shrimp (Pandah borealis) by
bacteriocins from lactic acid bacteria. Appl. Environ. Microbial. 61, 669-676.
Engelke, G., Gutowski-Eckel, Z., Kiesau, P., Siegers, K., Hammelmann, M. and Entian, K-D. (19921
Regulation of nisin biosynthesis and immunity in Lactococcus lactis 6F3. Appl. Environ. Microbial.
60, 814-825.
Fremaux, C. and Klaenhammer, T.R. (1994) Helveticin J, a large heat-labile bacteriocin from Lnctobacillus helueticus. In: De Vuyst, L., and E.J. Vandamme (editors) Bacteriocins of Lactic Acid
Bacteria: Microbiology, Genetics and Applications, Elsevier Applied Science Publ., UK, pp. 397-418.
Gao, F.H., Abee, T. and Konings, W.N. (1991) Mechanism of action of the peptide antibiotic nisin in
liposomes and cytochrome c oxidase-containing proteoliposomes. Appl. Environ. Microbial. 57,
2164-2170.
Garcia-Garcera, M.J., Elferink, M.G.L., Driessen, A.J.M. and Konings, W.N. (1993) In vitro pore-forming activity of the lantibiotic nisin. Role of proton motive force and lipid composition. Eur. J.
B&hem. 212, 417-422.
Gilmore, M.S., Segarra, R.A., Booth, M.C., Bogie, C.P., Hall, L.R. and Clewell, D.B. (1994) Genetic
structure of the Enterococcus faecalis plasmid pADl-encoded
cytolytic toxin system and its
relationship to Iantibiotic determinants. J. Bacterial. 176, 7335-7344.

184

T. Abee et al. /ht.

J. Food Microbiology 28 (1995) 169-185

Gounot, A., 1991. Bacterial life at low temperature: physiological aspects and biotechnological implications. J. Appl. Bacterial. 71, 386-397.
Harris, L.J., Flemming, H.P., and Klaenhammer, T.R. (1992) Novel paired starter culture system for
sauerkraut, consisting of a nisin-resistant Leuconostoc mesenteroides strain and a nisin-producing
Lactococcus lactti strain. Appl. Environ. Microbial. 58, 1484-1489.
Harwood, J.L. and Russel, N.J. (1984) Lipids in Plants and Microbes, George Allen and Unwin,
London, UK.
Holck, A.L., Azelsson, L., Hiihne, K., Kr?_ickel,L. (1994) Purification and cloning of sakacin 674, a
bacteriocin from Lactobacillus sake Lb674. FEMS Microbial. Lett. 115, 143-150.
Hugenholtz, J. and de Veer, G.J.C.M. (1991) Application of nisin A and nisin Z in dairy technology. In:
G. Jung and H.-G. Sahl (editors) Nisin and Novel Lantibiotics, ESCOM, Leiden, pp. 440-448.
Hugenholtz, J., Twigt, M., Slomp, M. and Smith, M.R. (1995) Development of nisin-producing starteers
for Gouda cheese manufacture. International Dairy lactic acid bacteria conference, Palmerston
north, New Zealand, 19-23 February 1995. Book of Abstracts, S. 2.4
Hurst, A. (1981) Nisin. Adv. Appl. Microbial. 27, 85-123.
Jung, G. and Sahl, H.-G. (1991) Nisin and Novel Lantibiotics. ESCOM, Leiden.
Kalchayanand, N., Sikes, T., Dunne, C.P. and Ray, B. (1994) Hydrostatic pressure and electroporation
have increased bactericidal efficiency in combination with bacteriocins. Appl. Environ. Microbial.
60,4174-4177.
Klaenhammer, T.R. (1993) Genetics of bacteriocins produced by lactic acid bacteria. FEMS Microbial.
Rev. 12, 39-86.
Kordel, M. and Sahl, H.-G. (1986) Susceptibility of bacterial, eukaryotic and artificial membranes to the
disruptive action of the cationic peptides Pep5 and nisin. FEMS Microbial. L&t. 34, 139-144.
Krockel, L. (1992) Bacteriocine von Milchsiiurebakterien fiir Fleischerzeugnisse. Mittbl. Bundesanst.
Fleischforsch. Kulmbach 31, 207-215.
Kuipers, O.P., Rollema, H.S., Yap, W.M.G.J., Boot, H.J., Siezen, R.J. and de Vos, W.M. (1992)
Engineering dehydrated amino acid residues in the antimicrobial peptide nisin. J. Biol. Chem. 267,
24340-24346.
Kuipers, O.P., Beerthuyzen, M.M., Siezen, R.J. and de Vos, W.M. (1993) Characterization of the nisin
gene cluster nhABTCIPR of Lactococcus lactis: Requirement of expression of the n&-t and ni.rZ
genes for development of immunity. Eur. J. Biochem. 216, 281-291.
Maisnier-Patin, S., Deschamps, N., Tatini, S.R. and Richard, J. (1992) Inhibition of Lbteria monocytogenes in Camembert cheese made with a nisin-producing starter. Lait 72, 249-263.
Ming, X. and Daeschel, M.A. (1993) Nisin resistance of foodbome bacteria and the specific resistance
responses of Listeria monocytogenes Scott A. J. Food Protect. 26, 944-948.
Morris, C.E. (1993) High-pressure builds up. Food Eng. 65, 113-120.
Mortvedt, C.I., Nissen-Meyer, J. Sletten, K. and Nes, I.F. (1991) Purification and amino acid sequence
of lactocin S, a bacteriocin produced by Lactobacillus sake L45. Appl. Environ. Microbial. 57, 1892.
Mortvedt-Abildgaard, CL, Nissen-Meyer, J., Jelle, B., Grenov, B., Skaugen, M. and Nes, I.F. (1995)
Production and pH-dependent bactericidal activity of lactocin S, a lantibiotic from Lactobacillus
sake LA5. Appl. Environ. Microbial. 61, 175-179.
Mulder, J.W.M., Boerrigter, I.J., Rollema, H.S., Siezen, R.J. and de Vos, W.M. (1991) Identification
and characterization of the lantibiotic nisin Z, a natural nisin variant. Eur. J. Biochem. 201,
581-584.
Ojcius, D.M. and Young, J.D. (1991) Qtolytic pore-forming proteins and peptides: is there a common
structural motive? TIBS 16, 225-229.
OLeary, W.M. and S.G. Wilkinson, 1988. Gram-positive bacteria. In: R. Ratledge and S.G. Wilkinson
(editors), Microbial Lipids, Acad. Press, London, pp. 117-201.
Parente, E. and Hill, C. (1992a) Characterization of enterocin 1146, a bacteriocin from Enterococcus
faecium inhibitory to Listeria monocytogenes. J. Food Protect. 55, 497-502.
Parente, E. and Hill, C. (1992b) Inhibition of Listeria in buffer, broth, and milk by enterocin 1146, a
bacteriocin produced by Enterococcus faecium. J. Food Protect. 55, 503-508.
Piard, J.-C., Muriana, P.M., Desmazaud, M.J. and Klaenhamme, T.R. (1992) Purification and partial

T. Abee et al. /ht.

J. Food Microbiology 28 (1995) 169-185

185

characterization of lacticin 481, a lanthionine-containing bacteriocin produced by Lactococcus lactis

subsp. Zacth CNRZ 481. Appl. Environ. Microbial. 58, 279-284.
Quadri, L.E.N., Sailer, M., Roy, K.L., Vederas, J.C. and Stiles, M.E. (1994) Chemical and genetic
characterization of bacteriocins produced by Camobacterium piscicola LV17B. J. Biol. Chem. 269,
12204-12211.
Rauch, P.J.G. and de Vos, W.M. (1992) Characterization of the novel nisin-sucrose conjugative
transposon Tn5276 and its insertion in Lactococcus lactis. J. Bacterial. 174, 1280-1287.
Ray, B. and Daeschel, M.A. (1992) Food Biopreservatives of Microbial Origin, CRC Press, Boca Raton,
FL.
Ray, B. (1993) Sublethal injury, bacteriocins, and food microbiology. ASM News 59, 285-291.
Rekhif, N., Atrih, A. and Lefebvre, G. (1994) Selection of spontaneous mutants of Listeria monocytogenes ATCC 15313 reistant to different bacteriocins produced by lactic acid bactreia strains. Curr.
Microbial. 28, 237-241.
Schved, F., Henis, Y. and Juven, B.J. (1994) Response of spheroplasts and chelator-permeabilized cells
of Gram-negative bacteria to the action of the bacteriocins pediocin SJ-1 and nisin. Int. J. Food
Microbial. 21, 305-314.
Shahidi, F. (1991) Developing alternative meat-curing systems. Trends Food Sci. Technol. September,
219-222.
Siegers, K. and Entian, K.-D. (1995) Genes involved in immunity to the lantibiotic nisin produced by
Lactococcus lacks 6F3. Appl. Environ. Microbial. 61, 1082-1089.
Stevens, K.A., Sheldon, B.W., Klapes, N.A. and Klaenhammer, T.R. (1991). Nisin treatment for the
inactivation of Salmonella species and other Gram-negative bacteria. Appl. Environ. Microbial. 57,
3613-3615.
Stiles, M.E. and Hastings, J.W. (1991) Bacteriocin production by lactic acid bacteria: potential for use
in meat preservation. Food Sci. Technol. 2, 235-263.
Stoffels, G., Nissen-Meyer, J. Sletten, K. and Nes, I.F. (1992) Purification and characterization of a new
bacteriocin isolated from a Camobacterium sp. Appl. Environ. Microbial. 58, 1417-1422.
Stoffels, G., Gudmundsdottir, A. and Abee, T. (1994) Membrane-associated proteins encoded by the
nisin gene cluster may function as a receptor for the lantibiotic camocin UI49. Microbiology 140,
1443-1450.
Taylor, L., Cann, D.D. and Welch, B.J. (1990) Antibotulin properties of nisin in fresh fish packaged in
an atmosphere of carbondioxide. J. Food Protect. 53, 953-959.
Tichaczek, P.S., Vogel, R.F., Hammes, W.P. (1994) Cloning and sequencing of sakP encoding sakacin P,
the bacteriocin produced by Lactobacillus sake LTH 673. Microbiology 140, 361-367.
van Belkum, M.J., Kok, J., Venema, G., Holo, H., Nes, I.F., Konings, W.N. and Abee, T. (1991) The
bacteriocin Lactococcin A specifically increases the permeability of lactococcal cytoplasmic membranes in a voltage-independent, protein-mediated manner. J. Bacterial. 173, 7934-7941.
Venema, K., Haverkort, R.E., Abee, T., Haandrikman, A.J., Leenhouts, K.J., de Leij, L., Venema, G.
and Kok, J. (1994) Mode of action of LciA, the lactococcin A immunity protein. Mol. Microbial. 14,
521-532.
Worobo, R.W., Henkel, T., Sailer, M., Roy, K.L., Vederas, J.C. and Stiles, M.E. (1994) Characterization and genetic determinant of a hydrophobic peptide bacteriocin, carnobacteriocin A, produced by
Camobactetium piscicola LV17A. Microbiology 140, 517-526.
Yousef, A.E., Luchansky, J.B., Degnan, A.J. and Doyle, M.P. (1991) Behavior of Listeria monocytogenes in Wiener exudates in the presence of Pediococcus acidilactici or Pediocin AcH during storage
at 4 or 25C. Appl. Environ. Microbial. 57, 1461-1467.