lable at ScienceDirect

Food Microbiology 27 (2010) 1095e1103

Contents lists avai

Food Microbiology

journal homepage: www.elsevier .com/locate/ fm

From undefined red smear cheese consortia to minimal model communitiesboth exhibiting similar anti-listerial activity on a cheese-like matrix

M. Imran, N. Desmasures*, J.-P. VernouxUnité des Micro-organismes d’Intérêt Laitier et Alimentaire, EA 3213, IFR146 ICORE, Université de Caen Basse-Normandie, Esplanade de la paix, 14032 CAEN cedex, France

a r t i c l e i n f o

Article history:Received 1 February 2010Received in revised form13 July 2010Accepted 20 July 2010Available online 27 July 2010

Keywords:Microbial communitiesYeastsBacteriaAnti-listerial activitySmear cheese

* Corresponding author. Tel.: þ33 231565621; fax:E-mail address: nathalie.demasures@unicaen.fr (N

0740-0020/$ e see front matter � 2010 Elsevier Ltd.doi:10.1016/j.fm.2010.07.016

a b s t r a c t

Starting from one undefined cheese smear consortium exhibiting anti-listerial activity (signal) at 15 �C,50 yeasts and 39 bacteria were identified by partial rDNA sequencing. Construction of microbialcommunities was done either by addition or by erosion approach with the aim to obtain minimalcommunities having similar signal to that of the initial smear. The signal of these microbial communitieswas monitored in cheese microcosm for 14 days under ripening conditions. In the addition scheme,strains having significant signals were mixed step by step. Five-member communities, obtained byaddition of a Gram negative bacterium to two yeasts and two Gram positive bacteria, enhanced the signaldramatically contrary to six-member communities including two Gram negative bacteria. In the erosionapproach, a progressive reduction of 89 initial strains was performed. While intermediate communities(89, 44 and 22 members) exhibited a lower signal than initial smear consortium, eleven- and six-member communities gave a signal almost as efficient. It was noteworthy that the final minimal modelcommunities obtained by erosion and addition approaches both had anti-listerial activity while con-sisting of different strains. In conclusion, some minimal model communities can have higher anti-listerialeffectiveness than individual strains or the initial 89 micro-organisms from smear. Thus, microbialinteractions are involved in the production and modulation of anti-listerial signals in cheese surfacecommunities.

� 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Red smear cheeses are examples of food ecosystems involvingcomplex microbial communities which consist of diverse genera ofyeast and bacteria. Ripening starts with the growth of yeasts withdominant species like Debaryomyces hansenii, Geotrichum candidum,Kluyveromyces lactis or Kluyveromyces marxianus and Yarrowia lip-olytica. Then, a salt-tolerant but acid sensitive usually very complexbacterial consortium begins to develop on the cheese surface,including coryneformbacteria e.g.Arthrobacter sp.,Brevibacterium sp.and also sometimes Gram negative bacteria e.g. Psychrobacter sp.,Halomonas sp. and Pseudomonas putida (Larpin-Laborde, 2006;Maozet al., 2003). Analysing the microbial diversity at surface of Livarotcheese, Larpin-Laborde (2006) found that Gram negative and Grampositive bacteria were co-dominant (about 109e1010 cfu/cm2 at theend of ripening), consisting of about forty different species while for

þ33 231566179.. Desmasures).

All rights reserved.

yeast there was ten species, G. candidum being the dominating one(Larpin et al., 2006).

The balance of this microbiota on the cheese surface is crucial foravoiding the growth of contaminant pathogenic micro-organisms(Bockelmann and Hoppe-Seyler, 2001). One well-known pathogeniccontaminant having potential for growth under cheese ripeningconditions is Listeria monocytogenes. A survey of its incidence inEuropean soft and semi-soft red smear cheeses by Rudolf andScherer (2001) indicated a higher incidence of L. monocytogenes incheeses manufactured from pasteurized milk (8.0%) than in cheesesmanufactured from raw milk (4.8%). These data have been recentlyconfirmed for European cheeses by the European Food SafetyAgency (EFSA, 2009) with incidences of 4.2e5.2% and 0.3e0.4%,respectively. Low incidence of L. monocytogenes in raw milk cheesesmay be due to the presence of some natural factors in raw milk andthereafter in raw milk cheeses, which might be destroyed duringpasteurization (Marielle and Albert, 2005). L. monocytogenes hasa ubiquitous nature, is able to survive and grow at refrigerationtemperatures (2e4 �C) (Rocourt and Cossart, 1997) and is tolerant tolow pHs (below pH 5.0) and high (up to 10%) sodium chloride levels(Peterson et al., 1993). Mechanisms of survival under adverse

M. Imran et al. / Food Microbiology 27 (2010) 1095e11031096

environmental conditions make it difficult to control in a food pro-cessing environment due to its wide range of survival temperatureand the formation of biofilms with phenomena of quorum sensing(Gandhi and Chikindas, 2007).

One option for the control of L. monocytogenes in cheese ecosys-tems could be microbial antagonism through competition for nutri-ents or space and/or by the production of one or more antimicrobialmetabolites such as organic acids, hydrogen peroxide, antimicrobialenzymes or bacteriocins (Holzapfel et al., 1995). Bio-preservationstrategies based on addition of bacteriocin or bacteriocin-producingmicrobial strains have been widely studied. However, effectiveapplication is scarce and the long-term effectiveness of inhibitorystrains has been questioned because of their instability and reducedefficacy in the complex food matrix (Chen and Hoover, 2003; Gálvezet al., 2007). Moreover, bacteriocin-resistant strains have beenobserved recently (Nilsson et al., 2005; Peschel and Sahl, 2006;Vadyvaloo et al., 2004). The metabolites, produced by a variety ofmicro-organisms, which proved to be effective against L. mono-cytogenes in vitro, often showed reduced or lost effectiveness in situ.Possible reasons could be higher concentration requirements,degradation or cross reaction with other food ingredients(Dieuleveux and Gueguen,1998). Moreover, the use of single hurdlescan stimulate the resistance in L.monocytogenes, and sub-lethal stressdeveloped more resistance in advance phases of growth in specificcircumstances (Gravesen et al., 2002).

It has been well documented that interactions between micro-organisms have impacts on different functions like flavour, colourand taste (Molimard et al., 1994). In their consideration of ripeningactivity, Bockelmann and Hoppe-Seyler (2001) used a five-strainsmicrobial group (1 yeast and 4 bacteria) with equivalent ripeningactivity of initial smear cheese in model cheese. In this minimalecosystem, interactions between micro-organisms can easily bedetermined and the role of each strain toward the generation ofa function can be proposed. Bonaiti et al. (2005) used an alternativemethod to study the characteristics of complex microbialcommunities by simplification based on the maintaining of biodi-versity and characteristic function. Based on the progressivesimplification of Livarot ecosystem, odour was used as functionalbase. A model ecosystem consisting of 10 strains was prepared,which made it possible to study the role of each microorganismduring the ripening process leading toward specific odour. Thecolour development in the red smear cheese results from thecomplex interaction between components of the cheese surfacemicrobiota (Galaup et al., 2007). Two cheeses with similar bacterialdistribution and population had different surface colour, implyingthat species-specific pigmentation of the bacteria also differsdepending on the yeasts present in the complex consortium(Mounier et al., 2008).

It has also been shown that some ripening consortia can inhibitfood pathogens (Eppert et al., 1997; Maoz et al., 2003; Mareikeet al., 2006). Eppert et al. (1997) have identified smears (complexmicrobial communities) which were very inhibitory to L. mono-cytogenes reducing its growth by 4 log cycles while a pure culture ofbacteriocin-producing bacteria isolated from smear cheese onlyreduced its growth by 2 log cycles. It was shown that listerialgrowth was clearly dependent on the microbial composition of theundefined surface flora but the molecular basis of the effect wasunknown.

The background to the present study is a body of research aimingat an understanding of how micro-organisms from Livarot cheesesurface consortia (Livarot is a French traditional red smear cheese)interact in communities to express anti-listerial activity. The objec-tive of the work was to identify representative members of Livarotcheese microbial surface consortia and to use them for preparingminimal microbial communities with similar anti-listerial activity

than that of the initial smear. Minimal communities had to be quitesimple so that interaction could easily be tested further. Theyalsohadto reflect part of the Livarotmicrobial diversity. So itwas decided thatthey consist of two yeasts and four bacteria including two Gramnegative strains. To construct model communities, two approacheswere used. One was based on successive addition of individual anti-listerial strains and the other was based on erosion starting froman anti-listerial defined smear. The anti-listerial activity of thesemodel microbial communities was tested in vitro and on a cheesemicrocosm.

2. Materials and methods

2.1. Smear samples, strains, media and culture conditions

Six Livarot cheeses had been obtained, in a previous survey(Larpin et al., 2006) from three different dairies (1, 2 and 3) bysampling at different stages during ripening (Early stage ofripening: E ¼ 2e3 days, Middle stage: M ¼ 17e24 days, Late stage:L ¼ 58e61 days). After scraping off part of the smear and theisolation of surface microbial flora, cheese samples were packed insterile bags and frozen at �76 �C.

From the smear sample 3M, fifty yeasts and 39 bacteria that hadpreviously been isolated were used. Bacterial isolates had beenpreviously characterized by a phenotypic approach using Bio-mérieux Biotype 100 strips, clustered and dereplicated by Rep-PCRusing Jacquard similarity coefficient andWard’s clusteringmethods(Larpin-Laborde, 2006). Yeasts had been dereplicated and charac-terized by FTIR spectroscopy, and identification had been achievedby phenotypic tests and genotyping (Larpin-Laborde, 2006). For thepresent study, identity of all bacterial and yeast isolates wasconfirmed by partial 16S rRNA and 26S rRNA gene sequencing,respectively. All these isolates were stored at �76 �C in glycerol.

For testing inhibitory activity, L. monocytogenes WSLC 1685(Scott A) and WSLC 1211, Staphylococcus aureus UCMA 7735,Escherichia coli C267 (O157:H7, non-shigatoxic) and SalmonellaTyphimurium ATCC 13311 were used as indicator strains, depend-ing on experiments.

Pre-culturing of microbial isolates was done by transferring onecolony to 5 ml of Trypticase Soya Broth (Merck, Darmstadt,Germany) supplemented with 6 g/L yeast extract (Oxoid, Basing-stoke, England) and 18 g/L D (þ) Glucose (AES Laboratoires, Com-bourg, France) (TSB-YEG). This medium was proven to be welladapted to the growth of both bacteria (including pathogens) andyeast strains. Incubation was done at 37 �C for 24 h (L. mono-cytogenes, S. aureus, E. coli and Salmonella Typhimurium) or at 25 �C(smear isolates) under 120 rpm shaking using a Novotron shaker(VWR, Fontenay sous bois, France) for 24e48 h, depending onisolates. Culturing was done by transferring 50 ml of pre-culture to5 ml of TSB-YEG and incubating under the same conditions.

2.2. Preparation of suspensions of smears, of individual strainsand of model microbial communities

Smears of frozen cheeses (about 1 cm2) were scraped off fromcheese, suspended in 1 ml trypton-salt solution and kept at roomtemperature for 1 h before experiments.

Enumeration of all 89 yeast and bacterial cultures (see cultureconditions) was done on TSA-YE (þglucose) after 48 h incubation at25 �C. Aliquots of each strain culture were taken in microtube andcentrifuged at 6000 g for 10 min to collect microbial cells. Twowashings were given with 0.9% sterile NaCl solution. Using enumer-ation data, concentration of each bacteria and yeast suspension wasadjusted with trypton-salt diluent to 109 cfu/ml for bacteria and107 cfu/ml for yeasts. Microbial communities suspensions were

M. Imran et al. / Food Microbiology 27 (2010) 1095e1103 1097

prepared by mixing equal volumes of individual microbialsuspensions.

2.3. Determination of anti-listerial activity of cheese smears,individual strains and model microbial communities

Anti-listerial activity of six available cheese smears and 89individual strains from inhibitory smear 3M was first tested in vitroagainst L. monocytogenes WSLC 1685 and, only for individualstrains, against L. monocytogenes WSLC 1211. Inhibition was deter-mined by using an agar membrane assay according to a methodadapted from Goerges et al. (2006) where the soft agar used wassupplemented with glucose to a 20 g/L final concentration. Thescore scale used was based on visual observation, according toGoerges et al. (2006).

Secondly, anti-listerial activity was tested in conditions near toreal cheese ecosystem. A recently developed quantitative methodwas used (Imran et al., 2008) involving a cheese microcosm con-sisting of cheese agar (composition near to Livarot cheese) as biotopeand microbial communities/isolates as biocoenosis. More preciselycheese agar was prepared according to Guichard and Bonnarme(2005) from unsalted Livarot curd, and distributed in 96-wellmicroplates. To reach smear-like bacterial and yeast concentration(108 and 106 cfu/cm2 respectively (Larpin-Laborde, 2006)), fiftymicrolitres of smears or microbial communities suspension (asexplained in previous paragraph) were used to inoculate the cheeseagar surface (0.5 cm2 per well). The anti-listerial activity was testedby monitoring growth kinetics of L. monocytogenes WSLC 1685,initially co-inoculated at level of 100 cfu/cm2 of cheese agar, withsmears or microbial communities or microbial isolates. The incuba-tion of microplates was done at 15 �C � 1 �C (cheese ripeningtemperature) and 98% humidity in a climatic incubator (VWR) for 14days. Cheese agar after 3, 5, 8 and 14 days of incubationwas removedwith help of a sterile toothpick and transferred into a microtube. Thehomogenizationwas done with help of the same toothpick followedbyavigorous vortex. Serial dilutionsweredone in trypton salt and thelisterial count was determined on Oxford agar (AES Laboratoires)after incubating at 37 �C for 48 h. Cell counts were calculated per cm2

of cheese agar. Growth kinetic was determined in three independentexperiments. In selected experiments, cheese agar wells were dedi-cated to surface pH measurement using Reflectometer (Merck). Theanti-listerial activity was calculated by determination of the differ-ence between concentrations of L. monocytogenes WSLC 1685 onassay and on control cheese microcosms composed of L. mono-cytogenesWSLC1685 and ofG. candidumUCMA523, a non-inhibitorystrain.

2.4. Construction of model microbial communities byaddition and erosion approaches

In order to finally select minimal communities having anti-lis-terial activity and consisting of two yeasts and four bacteria withtwo Gram negative strains, two approaches were used.

For the addition approach, a selection of isolates was made onthe basis of their individual anti-listerial activity in vitro. Commu-nities were prepared by adding, step by step, two yeasts, two Grampositives and then two Gram negative bacteria to reach six-straincommunities (Table 1).

For the erosion approach, the method used by Bonaiti et al.(2005) was adapted. All the 89 yeast and bacterial strains wereinitially pooled together and named mother microbial communityE. In the next step, community E was simplified into four daughtercommunities E1, E2, E3 and E4 (Table 1) each made of 44 yeast andbacterial strains. For this step and all further erosion steps, two ofthe four daughter communities were selected and simplified into

four new communities each composed of half of the strainsinvolved in the previous stage. Selection of individual strains wasbased on representation of each bacterial Rep-PCR and yeast FTIRcluster. In the final erosion step, four model communities with sixstrains each (two yeasts, two Gram positives and then two Gramnegative bacteria) were constructed.

At each step for both experimental approaches, the selection ofthe microbial communities was based primarily on expression ofhigher anti-listerial activity but when difference was not significantbetween communities, selection was based on microbial biodiver-sity and the two most different communities were selected tomaintain maximum representation of the three following groups:yeasts, Gram negative, and Gram positive bacteria.

3. Results

3.1. Bio-preservation effectiveness of undefined read smearcheese flora

The six undefined Livarot smears tested in vitro (Table 2)exhibited variable inhibition depending on target pathogens.Salmonella Typhimurium ATCC 13311 was the most sensitive strain.Four smear samples 1M, 2M, 3M and 3L had higher inhibitoryactivity toward the four pathogenic strains tested.

Further experiments were dedicated to the study of anti-listerialactivity on cheese microcosm. The initial pH of cheese agar was5.2� 0.1 and it remained constant in case of L. monocytogenesWLSC1685 as single culturing microorganism (Fig. 1). However thiscondition did not mimic pH evolution of Livarot cheese surfaceduring ripening (from 5 to 7e7.5 at the end of ripening). Thisevolution is due to yeasts such as G. candidum or D. hansenii, thetwo major species used as ripening agent for Livarot cheese prep-aration. Using some strains of these yeast species in co-culture withL. monocytogenes, increase in pH was observed and was correlatedwith an enhanced growth of the pathogen (data not shown).Negative control strains were selected as being non-inhibitorytoward L. monocytogenes WSLC 1685. Among them, the non-inhibitory strain G. candidum UCMA 523 was selected (Fig. 1) forpreparing a control cheese microcosm.

Under these conditions, anti-listerial activity of smears wasobserved. It rose from the 3rd to the 5th day and tended to decreaseor to remain stable up to the 14th day (Table 3). There was nosignificant difference between most smears. The consortium 3Mwas selected for having high biodiversity traits and for being effi-cient in reducing L. monocytogenes on the cheese microcosm.

3.2. Idenfication of 89 yeasts and bacterial strains isolatedfrom Livarot cheese smear 3M

All identified strains are given in Table 1. Twenty-two differenttaxawere identified. G. candidumwas the dominant species in yeastisolates. Among bacterial isolates, 23 were Gram negatives and 13were Gram positives and three bacterial isolates could not beidentified. Arthrobacter arilaitensis and Ps. putida were the domi-nant species in Gram positives and Gram negatives, respectively.

3.3. Anti-listerial activity of the model microbial communitiesobtained during erosion process

Compared to the smear 3M, the mother microbial community(E) consisting of the 89 strains exhibited lower anti-listerial activity(1e2 log cycles) (Fig. 2). The anti-listerial activity of all fourdaughter microbial communities E1, E2, E3 and E4 (44 memberseach, Table 1) was not significantly different from each other andfrom that of mother community E (Fig. 2). Community E3 showed

Table 1Compositions of model communities and anti-listerial activity in vitro of individual strains isolated from Livarot smear 3M. *St.: Staphylococcus; Ps.: Pseudomonas; R.: Raoultella; A.:Arthrobacter; Ac.: Acinetobacter; S.: Serratia; G.: Geotrichum; Yar.: Yarrowia; Ca.: Candida; T.: Torulaspora, E: Ewingella, H: Hafnia, n.d: not determined. **n.g.: no growth of the testedstrain; n.a.: not assessable; x/na.: variable from x to n.a. ***: REP- (bacteria) or FT-IR (yeast) cluster to which the isolate belongs. y Identified as Brachybacterium alimentarium bybiotype 100 profile (Larpin-Laborde, 2006).

M. Imran et al. / Food Microbiology 27 (2010) 1095e11031098

Table 2In vitro anti-listerial activity (inhibition score) of six Livarot smears against fourbacteria.

Indicator strains Smears

1M 2E 2M 2L 3M 3L

L. monocytogenes WSLC 1685 3 1 4 2 3 3Escherichia coli O157:H7 C267 4 0 4 2 4 4Staphylococcus aureus UCMA 7735 4 0 4 2 5 4Salmonella Typhimurium ATCC 13311 3 1 5 5 4 4

M. Imran et al. / Food Microbiology 27 (2010) 1095e1103 1099

a more stable anti-listerial activity and was selected for furthersimplification along with E1, which had a more different microbialcomposition than E3. Communities E1 and E3 were furthersimplified into four daughter communities E1.1, E1.2, E3.1, and E3.2(22 members each, Table 1). The pattern of their anti-listerialactivity remained very low and variable and all communities endedup with a lowest anti-listerial activity at 14th day (Fig. 2). Followingthe same strategy as in the previous step, E3.1 and E1.1 wereselected and simplified in turn into four communities (11 memberseach, Table 1). Surprisingly, communities E1.1.1, E1.1.2 and E3.1.2exhibited anti-listerial activity similar to that of smear 3M and evenmore at 3rd day (Fig. 2). E1.1.1 and E3.1.2 were selected. Fourmicrobial communities E1.1.1.1, E1.1.1.2, E3.1.2.1 and E3.1.2.2 wereobtained (six members each, Table 1).

Mostly, anti-listerial activity was not significantly differentbetween microbial communities at different incubation times,probably due to variation between experiments. Mainly twomicrobial communities (E31 and E311) were significantly (P< 0.05)different from the others at 3rd day. E1.1.1.1 was the only commu-nity significantly (P < 0.05) different from all others at 14th day.This community was retained for future studies.

3.4. Anti-listerial activity during addition process

3.4.1. Anti-listerial activity of the 89 single strainsAnti-listerial activity was tested in vitro at 15 �C (Table 1). Most

yeasts strains had weak inhibition (inhibition score < 3) againstL. monocytogenes WSLC 1685 and moderately high against L. mon-ocytogenesWSLC 1211. Bacterial strains showed higher anti-listerialactivity against L. monocytogenes WSLC 1685 with inhibition scoreup to 4. Most inhibitory strains belonged to Serratia sp., Ps. putida,and A. arilaitensis. In some cases, a white staining of soft agar belowculture spots grown on membrane occurred due to unknowncauses and was therefore considered not to be assessable. For someof the tested bacterial strains no growth (n.g, Table 1) was observed

1.0E+00

1.0E+01

1.0E+02

1.0E+03

1.0E+04

1.0E+05

1.0E+06

1.0E+07

1.0E+08

1.0E+09

1.0E+10

0 3 5 8 14

incubation time (days)

L.m

on

oc

yto

gen

es

W

LS

C 1

68

5 (G

ro

wth

K

in

etic

s

CF

U/c

m²)

2.5

3

3.5

4

4.5

5

5.5

6

6.5

7

7.5

Su

rfa

ce

p

H

Growth kinetics (L.monocytogenes WSLC 1685)

Growth Kinetics (L.monocytogenes WSLC 1685+

G.candidum UCMA 523)

pH (L.monocytogenes WSLC 1685+ G.candidum UCMA

523)

pH (L.monocytogenes WSLC 1685)

Fig. 1. Monitoring of surface pH and growth kinetics of L. monocytogenes WSLC 1685(mean of three independent experiments) on cheese agar in absence or presence of thenon-inhibitory strain Geotrichum candidum UCMA 523.

in a reproducible way in presence of L. monocytogenes which couldact as a reverse inhibitor. For the addition process, the selection oftwo G. candidum UCMA 3724 and 3765 was made from the twomain FTIR clusters and based on their anti-listerial activity. Candidanatalensis UCMA 3722 and Torulaspora delbrueckii UCMA 3745 werealso chosen. A. arilaitensis UCMA 3890 and UCMA 3913, Paeniba-cillus sp. UCMA 3887, Brachybacterium alimentarium UCMA 3920,Serratia liquefaciens UCMA 3879, Serratia grimesii UCMA 3895,Marinomonas UCMA 3912 and Pseudomonas stutzeri UCMA 3883were selected for both having anti-listerial activity andmaintainingbiodiversity.

3.4.2. Addition approachThe two strains of G. candidum had moderate to weak anti-

listerial activity (Fig. 3), alone or in combinationwith another yeast(A1, A2, A3, A4). For the next step, A1 (G. candidumUCMA 3724withCa. natalensis UCMA 3722) and A4 (G. candidum UCMA 3765 withT. delbrueckii UCMA 3745) were retained (Table 1). Addition ofA. arilaitensis UCMA 3890 or 3913 resulted in A1.1, A1.2, A4.1, andA4.2 with little increase in anti-listerial activity. Following the samestrategy as in the previous step A1.1 and A4.2 were selected. In thenext step Paenibacillus sp. UCMA 3887 and B. alimentarium UCMA3920 were added (A1.1.1, A1.1.2, A4.2.1, and A4.2.2) resulting invariable changes in anti-listerial activity. A1.1.1 and A1.1.2 were notsignificantly different from each other, it was also the case forA4.2.1, A4.2.2 but these two groups were highly different (P < 0.01)from each other. Communities A1.1.1 and A4.2.1 were retained.Gram negative bacteria S. liquefaciens UCMA 3879 or S. grimesiiUCMA 3895 were then added. Anti-listerial activity was dramati-cally increased for all communities (up to 4.77 logs for A4.2.1.2 at5th day) and was significantly higher (P < 0.05) than smear activityat 3rd and 5th day. Communities A1.1.1.1 and A4.2.1.2 were retainedfor addition of a second Gram negative strain Marinomonas sp.UCMA 3912 or Ps. stutzeri UCMA 3883 which resulted in a signifi-cant decrease (P < 0.05) of anti-listerial activity at 3rd day(A1.1.1.1.2, A4.2.1.2.1) or at most of ripening times for A4.2.1.2.Association A1.1.1.1.1 was the only one which showed stable anti-listerial activity with slight variations from the 3rd day to the 14thday (Figs. 3 and 4) and it was selected as one of the most repre-sentative minimal model community for future studies.

3.5. Comparison of complex and minimal model microbialcommunities and individual strains

Table 3 allows a comparison of the anti-listerial effect betweencomplex and minimal microbial communities and individualstrains on cheese microcosm with regard to the pH. There was nocorrelation (r ¼ 0.49) between pH and L. monocytogenes growthreduction (log) in our tested conditions. There was neither syner-gistic nor additive anti-listerial activity found in the final minimalmodel microbial communities in comparison to anti-listerialactivity of each of its component i.e. individual strains in pureculture (Table 3). However, the anti-listerial activity of the twoselected minimal communities was higher than the individualactivity of the most inhibitory member on cheese microcosm. Anti-listerial activity of individual members on cheese microcosm wasnot always consistent with in vitro data (Tables 1 and 3).

4. Discussion

Antipathogenic activity of smear samples collected from six redsmear cheeses (Livarot cheese) was observed in vitro towards fourpotentially pathogenic bacterial strains. While the various smearstested were all complex with different microbial composition(Larpin-Laborde, 2006) they were not very different in terms of

Table 3Reduction of listerial growth in log (mean of three independent experiments) in presence of natural smears, model microbial communities and some selected individual strainson cheese agar matrix, Surface pH is in brackets.

Ripening time (days)

Microbial community/strain tested (number of strains) 3 5 8 14

Smears 1M 2.51(5.2) 3.59(6.8) n.d. 3.21(6.7)2E 2.23(6.2) 2.75(6.7) n.d. 1.81(6.8)2M 2.51(6.7) 2.64(7.1) n.d. 2.97(6.6)2L 1.55(6.6) 2.58(7.0) n.d. 2.85(6.6)3L 1.05(6.5) 2.94(7.0) n.d. 2.74(6.1)3M 1.90(6.1) 3.54(6.1) 3.07(6.4) 3.28(6.5)

Erosional communities E (89) 0.86(5.5) 1.76(6.1) 1.64(6.0) 1.61(5.8)E1 (44) 0.44(5.7) 1.70(6.2) 1.35*(6.1) 1.43*(6.2)E2 (44) 1.32(5.8) 1.46(6.2) 1.78*(6.2) 1.44*(6.5)E3 (44) 1.75(5.6) 1.58(6.2) 1.56*(6.3) 1.61*(7.0)E4 (44) 0.45(5.6) 1.68(6.2) 1.80*(6.0) 1.64*(6.4)E.1.1 (22) 0.49* 1.59* 1.30* 0.76*E.1.2 (22) 0.45* 1.65* 1.32* 0.64*E.3.1 (22) 0.82* 1.52* 1.51* 0.69*E.3.2 (22) 0.49* 1.81* 1.28* 0.61E.1.1.1 (11) 3.35* 3.21 2.83 3.03E.1.1.2 (11) 3.12* 3.13 2.60 2.68E.3.1.1 (11) 1.28 1.16 1.66 1.89*E.3.1.2 (11) 3.70* 2.99 2.90 2.31E.1.1.1.1 (6) 3.15 3.32 2.70 2.60E.1.1.1.2 (6) 3.04* 3.06 2.24 1.69*E.3.1.2.1 (6) 3.43 2.99 1.21* 1.66*E.3.1.2.2 (6) 3.73 3.02 1.92* 1.17*

Additional communities A1 (2) 1.75(5.6) 1.15*(6.1) 1.43*(7.0) 0.58*(6.3)A2 (2) 0.93(5.7) 1.19*(6.1) 0.68*(6.6) 0.14*(6.5)A3 (2) 1.71(5.5) 1.18*(6.0) 0.76*(6.2) 0.59*(5.6)A4 (2) 1.39(5.4) 0.86*(5.7) 0.70*(5.8) 0.26*(5.6)A1.1 (3) 3.38* 1.51 2.00 1.81A1.2 (3) 1.79 1.57* 1.60* 1.85*A4.1 (3) 1.92 1.32 1.30* 1.55*A4.2 (3) 1.38 1.69* 1.78* 1.82A1.1.1 (4) 0.18* 1.32* 1.20* 1.32*A1.1.2 (4) 0.27* 1.54* 0.96* 1.36A4.2.1 (4) 2.06 1.90* 1.43 1.43A4.2.2 (4) 1.82 2.06* 0.99* 1.38*A1.1.1.1 (5) 3.70* 4.75* 2.98 2.69A1.1.1.2 (5) 3.24* 4.17* 2.02 2.62A4.2.1.1 (5) 3.90* 4.52* 2.09 2.20A4.2.1.2 (5) 3.95* 4.77* 2.05 2.25A1.1.1.1.1 (6) 2.22 2.44 2.31 2.54A1.1.1.1.2 (6) 2.99 2.98 1.05 1.47A4.2.1.2.1 (6) 1.62 2.32 1.56 1.44A4.2.1.2.2 (6) 2.11 2.49* 0.88* 1.35*

Individual G. candidum UCMA 3724 (A) 1.75(6.4) 1.15(6.3) 1.08(6.8) 0.47(5.7)

strains G. candidum UCMA 3765 0.41(6.3) 0.75(5.8) 0.47(5.7) 0.22(5.6)Ca. natalensis UCMA 3722 (A) 0.73(5.7) 1.88(6.4) n.d. 1.86(6.4)A. arilaitensis UCMA 3872 (A) 1.60(6.7) 1.52(6.7) n.d. 1,26(6.3)S. liquefaciens UCMA 3879 (A) 3.29(5.9) 1.72(6.7) n.d. 2.28(6.7)Paenibacillus sp. UCMA 3887 (A) 1.14(5.8) 2.03(6.6) n.d. 1.55(6.5)Marinomonas sp. UCMA 3912 (A) 1.61(5.6) 2.77(5.6) n.d. 1.18(6.5)D. hansenii UCMA 3723 (E) 2.14(6.4) 2.21(6.5) n.d. 1.65(6.2)G. candidum UCMA 3730 (E) 0.73(6.2) 2.02(6.5) n.d. 2.00(6.3)A. arilaitensis UCMA 3890 (E) 1.58(5.5) 2.02(6.3) n.d. 1.44(6.6)Marinomonas sp. UCMA 3892 (E) 1.22(5.7) 2.30(6.4) n.d. 2.17(6.4)Ps. putida UCMA 3916 (E) 1.99(5.9) 3.69(6.0) n.d. 2.04(6.2)St. equorum UCMA 3917 (E) 2.80(6.3) 1.54(6.7) n.d. 1.90(6.4)

*Significantly different (P < 0.05) from cheese smear 3M.

M. Imran et al. / Food Microbiology 27 (2010) 1095e11031100

antipathogenic activity. This suggest that their complexity was nota preponderant factor in explaining their inhibitory activity.

Anti-listerial activity was investigated more precisely by usingL. monocytogenes WSLC 1685 Scott A as an indicator strain. Usinga minimal microcosm, we reproduced some conditions (cheesematrix, ripening agent and physico-chemical conditions: pH,temperature, relative humidity) found in real Livarot cheeseecosystem. Growing L. monocytogenes on minimal microcosm it

was found that the physico-chemical conditions that occur duringripening, especially pH changes, can favour listerial growth asalready shown by Farber and Peterkin (1991).

Compared to minimal cheese microcosm, all six microcosmsmade from smear samples exhibited various degrees of anti-listerialactivity, resulting in aw3 logs inhibition after 14 days’ ripening. Anti-listerial activity has been previously reported in real cheese, likesmear cheeses (Carnio et al., 1999; Eppert et al., 1997; Maoz et al.,

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

4.00

4.50

5.00

M3

E 1E E

2 3E E

4 1.1.

E E1.

2. 1.3.

E

2.3.

EE

1.

1.1.

2.1.

1.E

1.1.

3.E E

3.

2.1.

1.1.

1.1.

E E.

2.1.

1.1

1.3.

E

1.2.

2.2.

1.3.

E

Microbial communities

)g

oL

(y

ti

vi

tc

al

ai

re

ts

ili

tn

a

3rd day 5th Day 8th day 14th day

44 Strains

22 Strains

11 Strains 6 Strains

89

Strains

Smear

Fig. 2. Anti-listerial activity (mean of three independent experiments) of model microbial communities prepared by erosion approach.

M. Imran et al. / Food Microbiology 27 (2010) 1095e1103 1101

2003) and hard cheeses (Saubusse et al., 2007). In our experiments,complete elimination of L. monocytogenes was not observed as insome of these previous studies (Eppert et al.,1997;Maoz et al., 2003).Our conditions could explain the difference: particularly, the use of

0.00

1.00

2.00

3.00

4.00

5.00

A1

A2

A3

A4

A

1.1

A 1

.2

A 4

.1

A

4.2

A

1.1

.1

A 1

.1.2

Micro

An

tilis

te

ria

l A

ctiv

ity

(L

og

)

3rd day 5th Day 8th day 14th

2 Strains

3 Strains

4

Fig. 3. Anti-listerial activity (mean of three independent experiments)

cheese microcosms, the microbial composition of smears, a higherinitial contamination level of L. monocytogenes (100 cfu/cm2

compared to 10 cfu/cm2) and the simultaneous inoculation ofL. monocytogenes with microbial communities.

A

4.2

.1

A

4.2

.2

A 1

.1.1

.1

A

1.1

.1.2

A

4.2

.1.1

A 4

.2.1

.2

A

1.1

.1.1

.1

A

1.1

.1.1

.2

A

4.2

.1.2

.1

A

4.2

.1.2

.2

bial Communities

day

Strains

5 Strains

6 Strains

of model microbial communities prepared by addition approach.

1.0E+02

1.0E+03

1.0E+04

1.0E+05

1.0E+06

1.0E+07

1.0E+08

1.0E+09

1.0E+10

0 2 4 6 8 10 12 14 16

Incubation time (days)

Lis

te

ria

l c

on

ce

ntra

tio

n (L

og

/c

m)

Livarot Smear (complex microbial population)

Smear like consortium (89 strains)

A 1.1.1.1.1(6 strains)

E 1.1.1.1 (6 strains)

Control negative(G.candidum UCMA 523)

A 1.1.1.1(5 strains)

Fig. 4. Growth kinetics of L. monocytogenes WSLC 1685 (mean of three independentexperiments) in cheese microcosms reconstituted starting from natural smears, andminimal model microbial communities.

M. Imran et al. / Food Microbiology 27 (2010) 1095e11031102

Using an erosion approach, anti-listerial activity of definedcommunities derived from an undefined smear sample wasmonitored at each erosion step. Successive erosion steps fromsmear to 22 strains generally led to progressive extinction of theactivity, suggesting first that there was little functional redundancyin the original community and that the loss of diversity resulted ina loss of activity. Surprisingly, the further erosion step led toa strong increase of anti-listerial activity which was maintainedduring the last step and became stronger than the activity ofmother community (89 strains) and comparable with originalsmear sample 3M (Fig. 4). It can be hypothesised that there is somefunctional redundancy between members of the initial communitybut that specific microbial interactions extensively modulateexpression of the activity. Microbial interactions seem to be themain key for explaining our results, as indicated in recent studies(Monnet et al., 2010; Retureau et al., 2010).

When using the additional approach, on the whole, little anti-listerial activity was expressed by communities with 2e4members.But adding the fifth member resulted in a dramatic increase of anti-listerial activity, while it decreased in communities with 6members. It seems that addition of one Gram negative bacteria(either S. liquefaciens or S. grimesii) enhanced anti-listerial activity(with an important boost of its own anti-listerial activity by thecommunity). This seems to be due to the high anti-listerial activityof these special strains, which would then be the main bacteriahaving anti-listerial activity in the communities. However additionof a second Gram negative bacterium changed the equilibrium toa less favourable state, resulting in a decreased anti-listerialactivity. It points out the important role of microbial interaction(probably competition between two Gram negative bacteria here).

To assess the anti-listerial activity of all communities, entireexperiments were repeated three times. While identical trendswere obtained, variability was observed (Figs. 2 and 3). This can beattributed to the complexity of cheese microcosm, as already indi-cated by Eppert et al. (1997) and Retureau et al. (2010) and makesunderstanding of microbial inhibition in situ a challenging task.

In most cases found in literature, inhibition mechanisms demon-strated in cheese are bacteriocinproduction by single strains (Valdès-Stauber and Scherer,1994) andpHor organic acid contents, especiallylactic acid (Millet et al., 2006).However, inour tested conditions therewas no correlation between pH and L. monocytogenes growthreduction(log).While testing individual strains, itwas found thatonlya minority of yeasts inhibit L. monocytogenes as previously reported(Goerges et al., 2006; Larpin-Laborde, 2006). In Livarot ecosystemG.candidum is thedominantyeast (Larpinetal., 2006). It could thenbe

assumed that it has a vital impact on the functionality of microbialcommunity (Decker and Nielsen, 2005).WhileG. candidum is knownfor inhibitory properties as it is a producer of phenyl-lactic acid andindollactic acid (anti-listerial compounds), it exhibited a very weakanti-listerial activity in situ in cheese (Dieuleveux et al., 1998). It maythus have a principal role in balancing the whole ecosystem. Anti-listerial activity has been demonstrated for several cheese-relatedGram positive bacterial strains due to the production of bacteriocins(Valdès-Stauber and Scherer, 1994). In our study the most inhibitoryspecies/strains were Staphylococcus equorum, A. arilaitensis UCMA3892, S. liquefaciens and Ps. putida, depending on ripening time. Theindividual anti-listerial activity of these strains was not sufficient toexplain the anti-listerial activity of the communities in which theywere present. This may be due to a synergistic effect of multipleinhibitory factors that produced a hurdle effect (Nilsson et al., 2000;Leistner and Gorris, 1995).

Although microbial competition is an important quality andsafety issue in food manufacturing, only limited work has beenconducted on mechanisms of microbial inhibition that do notinclude antimicrobial compounds. An interaction betweenSaccharomyces cerevisiae and non-Saccharomyces yeasts wasmediated by cell-to-cell contact, presumably as a result of compe-tition for space (Nissen and Arneborg, 2003). Contact-dependentinhibition of growth in E. coli was also reported (Aoki et al., 2005;Slechta and Mulvey, 2006). Studying anti-listerial activity ofcomplex cheese communities or strains, some authors demon-strated Jameson effect (Guillier et al., 2008). In another case the roleof microbial equilibrium in anti-listerial activity was shownthrough adaptation of cheese communities by propagation(Monnet et al., 2010). So different mechanisms could be involved ineffectiveness of inhibition in complex cheese communities.

On the whole, the two approaches (addition or erosion) devel-oped here both led to inhibitory minimal model communities withdifferent microbial compositions. At this stage it is difficult torecommend one or other approach for construction of minimalcommunities since results are very similar. Additional approachseems to be less time-consuming while erosion approach is moreinteresting from conceptual point of view. To understand microbialinteractions resulting in anti-listerial activity in cheese ecosystems,the way in which final communities are reached (includingconstruction of intermediate communities) and their study areequally important. Smearmicrobial communities fromLivarot cheesecan be a natural barrier against development of L. monocytogenes.Individual analysis of smear isolates and of different combinations ofisolates ledus toproduce some selectedminimalmodel communitieshaving anti-listerial features similar to smear. These can be moreeasily analysed at population and molecular levels to elucidate thebasis of inhibitory mechanisms (work in progress).

Acknowledgements

We thank to Prof. Christine Vernozy-Rozand from ENVL (Lyon,France) for providing the E. coli strain C267 and Prof. SiegfriedScherer from Abteilung Mikrobiologie, Zentralinstitut für Ernäh-rungs- und Lebensmittelforschung (Germany) for providingL. monocytogenes WSLC 1685 and WSLC 1211. Thanks to MarielleGueguen for her help regarding statistical analysis. This work wasmade possible by a PhD scholarship from Higher EducationCommission (HEC) of Pakistan.

References

Aoki, S.K., Pamma, R., Hernday, A.D., Bickham, J.E., Braaten, B.A., Low, D.A., 2005.Contact-dependent inhibition of growth in Escherichia coli. Science 309,1245e1248.

M. Imran et al. / Food Microbiology 27 (2010) 1095e1103 1103

Bockelmann, W., Hoppe-Seyler, T., 2001. The surface flora of bacterial smear-ripened cheeses from cow’s and goat’s milk. Int. Dairy J. 11, 307e314.

Bonaiti, C., Irlinger, F., Spinnler, H.E., Engel, E., 2005. An iterative sensory proce-dure to select odor-active associations in complex consortia of microorgan-isms: application to the construction of a cheese model. Int. Dairy J. 88,1671e1684.

Carnio, M.C., Eppert, I., Scherer, S., 1999. Analysis of the bacterial surface ripeningflora of German and French smeared cheeses with respect to their anti-listerialpotential. Int. J. Food Microbiol. 47, 89e97.

Chen, H., Hoover, D.G., 2003. Bacteriocins and their food applications. Compr. Rev.Food Sci. Food Saf. 2, 82e100.

Decker, M., Nielsen, P.V., 2005. The inhibitory effect of Penicillium camemberti andGeotrichum candidum on the associated funga of white mould cheese. Int. J.Food Microbiol. 104, 51e60.

Dieuleveux, V., Van Der Pyl, D., Chataud, J., Gueguen, M., 1998. Purification andcharacterization of anti-Listeria compounds produced by Geotrichum candidum.Appl. Environ. Microbiol. 64, 800e803.

Dieuleveux, V., Gueguen, M., 1998. Antimicrobial effects of D-3-phenyllactic acid onListeria monocytogenes in TSB-YE medium, milk, and cheese. J. Food Prot. 61,1281e1285.

EFSA, 2009. Trends and Sources of Zoonoses and Zoonotic Agents in the EuropeanUnion in 2007.

Eppert, I., Valdes-Stauber, N., Gotz, H., Busse, M., Scherer, S., 1997. Growth reductionof Listeria spp. caused by undefined industrial red smear cheese cultures andbacteriocin-producing Brevibacterium linens as evaluated in situ on soft cheese.Appl. Environ. Microbiol. 63, 4812e4817.

Farber, J.M., Peterkin, P.I., 1991. Listeria monocytogenes, a food-borne pathogen.Microbiol. Mol. Biol. Rev. 55, 476e511.

Galaup, P., Gautier, A., Piriou, Y., Villeblanche, A.d., Valla, A., Dufossé, L., 2007. Firstpigment fingerprints from the rind of French PDO red-smear ripened softcheeses Epoisses, Mont d’Or and Maroilles. Innov. Food Sci. Emerg. Technol. 8,373e378.

Gálvez, A., Abriouel, H., López, R.L., Omar, N.B., 2007. Bacteriocin-based strategiesfor food biopreservation. Int. J. Food Microbiol. 120, 51e70.

Gandhi, M., Chikindas, M.L., 2007. Listeria: a foodborne pathogen that knows how tosurvive. Int. J. Food Microbiol. 113, 1e15.

Goerges, S., Aigner, U., Silakowski, B., Scherer, S., 2006. Inhibition of Listeria mon-ocytogenes by food-borne yeasts. Appl. Environ. Microbiol. 72, 313e318.

Gravesen, A., Ramnath, M., Rechinger, K.B., Andersen, N., Jansch, L., Hechard, Y.,Hastings, J.W., Knochel, S., 2002. High-level resistance to class IIa bacteriocins isassociated with one general mechanism in Listeria monocytogenes. Microbi-ology 148, 2361e2369.

Guichard, H., Bonnarme, P., 2005. Development and validation of a plate techniquefor screening of microorganisms that produce volatile sulfur compounds. Anal.Biochem. 338, 299e305.

Guillier, L., Stahl, V., Hezard, B., Notz, E., Briandet, R., 2008. Modelling thecompetitive growth between Listeria monocytogenes and biofilm microflora ofsmear cheese wooden shelves. Int. J. Food Microbiol. 128, 51e57.

Holzapfel, W.H., Geisen, R., Schillinger, U., 1995. Biological preservation of foodswith reference to protective cultures, bacteriocins and food-grade enzymes. Int.J. Food Microbiol. 24, 343e362.

Imran, M., Desmasures, N., Vernoux, J.-P., 2008. A rapid method for selection ofmicrobial communities exhibiting antilisterial activity on micro-cheese agarmedium, p. Ljubljana, Slovenia. First European Food Congress.

Larpin, S., Mondoloni, C., Goerges, S., Vernoux, J.P., Gueguen, M., Desmasures, N.,2006. Geotrichum candidum dominates in yeast population dynamics in Livarot,a French red-smear cheese. FEMS Yeast Res. 6, 1243e1253.

Larpin-Laborde, S., 2006. Geotrichum candidum et autres levures de surface duLivarot: inverntaire, dynamique et potentialites alleopathiques universite deCaen, France, PhD thesis.

Leistner, L., Gorris, L.G.M., 1995. Food preservation by hurdle technology. TrendsFood Sci. Technol. 6, 41e46.

Maoz, A., Mayr, R., Scherer, S., 2003. Temporal stability and biodiversity of twocomplex antilisterial cheese-ripening microbial consortia. Appl. Environ.Microbiol. 69, 4012e4018.

Mareike, W., Vera, T., Siegfried, S., 2006. Rapid analysis of two food-borne microbialcommunities at the species level by Fourier-transform infrared micro-spectroscopy. Environ. Microbiol. 8, 848e857.

Marielle, G., Albert, A., 2005. Factors moderating Listeria monocytogenes growth inraw milk and in soft cheese made from raw milk. Lait 85, 153e170.

Millet, L., Saubusse, M., Didienne, R., Tessier, L., Montel, M.C., 2006. Control of Lis-teria monocytogenes in raw-milk cheeses. Int. J. Food Microbiol. 108, 105e114.

Molimard, P., Lesschaeve, I., Bouvier, I., Vassal, L., Schlich, P., Issanchou, S.,Spinnler, H.E., 1994. Bitterness and nitrogen fractions of mold ripened cheese ofCamembert type: impact of the association of Penicillium camemberti withGeotrichum candidum. Lait 74, 361e374.

Monnet, C., Bleicher, A., Neuhaus, K., Sarthou, A.-S., Leclercq-Perlat, M.-N.,Irlinger, F., 2010. Assessment of the anti-listerial activity of microfloras from thesurface of smear-ripened cheeses. Food Microbiol. 27, 302e310.

Mounier, J., Monnet, C., Vallaeys, T., Arditi, R., Sarthou, A.S., Helias, A., Irlinger, F.,2008. Microbial interactions within a cheese microbial community. Appl.Environ. Microbiol. 74, 172e181.

Nilsson, L., Chen, Y., Chikindas, M.L., Huss, H.H., Gram, L., Montville, T.J., 2000.Carbon dioxide and nisin act synergistically on Listeria monocytogenes. Appl.Environ. Microbiol. 66, 769e774.

Nilsson, L., Hansen, T.B., Garrido, P., Buchrieser, C., Glaser, P., Knochel, S., Gram, L.,Gravesen, A., 2005. Growth inhibition of Listeria monocytogenes by a non-bacteriocinogenic Carnobacterium piscicola. J. Appl. Microbiol. 98, 172e183.

Nissen, P., Arneborg, N., 2003. Characterization of early deaths of non-Saccharo-myces yeasts in mixed cultures with Saccharomyces cerevisiae. Arch. Microbiol.180, 257e263.

Peschel, A., Sahl, H.G., 2006. The co-evolution of host cationic antimicrobialpeptides and microbial resistance. Nat. Rev. Microbiol. 4, 529e536.

Peterson, M.E., Pelroy, G.A., Paranjpye, R.N., Poytsky, F.T., Almond, J.S., Eklund, M.W.,1993. Parameters for control of Listeria monocytogenes in smoked fisheryproducts: sodium chloride and packaging method. J. Food Prot. 56, 938e943.

Retureau, Ã.m., Callon, C.c., Didienne, R., Montel, M.-C., 2010. Is microbial diversity anasset for inhibiting Listeria monocytogenes in rawmilk cheeses? Dairy Sci. Technol.

Rocourt, J., Cossart, P., 1997. Listeria monocytogenes. In: Doyle, M.P., Buechat, L.R.,Montville, T.J. (Eds.), Food Microbiology d Fundamentals and Frontiers.

Rudolf, M., Scherer, S., 2001. High incidence of Listeria monocytogenes in Europeanred smear cheese. Int. J. Food Microbiol. 63, 91e98.

Saubusse, M., Millet, L., Delbes, C., Callon, C., Montel, M.C., 2007. Application ofsingle strand conformation polymorphism e PCR method for distinguishingcheese bacterial communities that inhibit Listeria monocytogenes. Int. J. FoodMicrobiol. 116, 126e135.

Slechta, E.S., Mulvey, M.A., 2006. Contact-dependent inhibition: bacterial brakesand secret handshakes. Trends Microbiol. 14, 58e60.

Vadyvaloo, V., Arous, S., Gravesen, A., Hechard, Y., Chauhan-Haubrock, R.,Hastings, J.W., Rautenbach, M., 2004. Cell-surface alterations in class IIa bacte-riocin-resistant Listeria monocytogenes strains. Microbiology 150, 3025e3033.

Valdès-Stauber, N., Scherer, S., 1994. Isolation and characterization of Linocin M18,a bacteriocin produced by Brevibacterium linens. Appl. Environ. Microbiol. 60,3809e3814.