International Journal of Food Microbiology 174 (2014) 98–109

Contents lists available at ScienceDirect

International Journal of Food Microbiology journal homepage: www.elsevier.com/locate/ijfoodmicro

Microbial biodiversity in cheese consortia and comparative Listeria growth on surfaces of uncooked pressed cheeses Cécile Callon ⁎, Emilie Retureau, Robert Didienne, Marie-Christine Montel INRA, URF 545 Fromagères, 20 Côte de Reyne, 15000 Aurillac, France

a r t i c l e

i n f o

Article history: Received 19 April 2013 Received in revised form 11 December 2013 Accepted 3 January 2014 Available online 9 January 2014 Keywords: Natural antilisterial consortium Reconstituted consortia Biodiversity Inhibitory activities Uncooked pressed cheese rinds Organic acids

a b s t r a c t The study set out to determine how changes in the microbial diversity of a complex antilisterial consortium from the surface of St-Nectaire cheese modify its antilisterial activities. On the basis of the microbial composition of a natural complex consortium named TR15 (Truefood consortium 15), three new consortia of different species and strain compositions were defined: TR15-SC (58 isolates from TR15 collection), TR15-M (pools of isolates from selective counting media) and TR15-BHI (pools of isolates from BHI medium). Their antilisterial activities on the surfaces of uncooked pressed cheese made with pasteurised milk were compared with the activity of complex consortium TR15 and a control cheese inoculated only with starter culture (Streptococcus thermophilus, Lactobacillus delbrueckii). The natural consortium TR15 was the most inhibitory, followed by reconstituted consortium TR15-BHI. The dynamics of the cheese rind microbial flora were monitored by counting on media and by isolate identification using 16S rDNA sequencing and direct 16S rDNA Single Strand Conformation Polymorphism analysis. The combination of these methods showed that rind with natural consortium TR15 had greater microbial diversity and different microbial dynamics than cheese rinds with reconstituted consortia. Cheese rind with the natural consortium showed higher citrate consumption and the highest concentrations of lactic and acetic acids, connected with high levels of lactic acid bacteria such as Carnobacterium maltaromaticum, Vagococcus fluvialis, Enterococcus gilvus, Leuconostoc mesenteroides, Brochothrix thermosphacta and Lactococcus lactis, ripening bacteria such as Arthrobacter nicotianae/arilaitensis, and Gram negative bacteria (Pseudomonas psychrophila and Enterobacter spp.). The highest L. monocytogenes count was on rind with TR15-M and was positively associated with the highest pH value, high succinic and citric acid contents, and the highest levels of Marinilactibacillus psychrotolerans and Gram positive catalase positive bacteria represented by Staphylococcus vitulinus, Brevibacterium linens, Microbacterium gubbeenense and Brachybacterium tyrofermentans. The results show that the species composition of consortium is more important than the number of species. It is likely that inhibition mechanisms differ from one consortium to another; investigating gene expression will be an effective way to elucidate microbial interactions in cheese. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Raw milk cheeses are often criticised on safety grounds although they are known to be safe products. During the past decade the prevalence of Listeria monocytogenes in raw milk cheese has been relatively low (EFSA, 2013; Jakobsen et al., 2011; Little et al., 2008) owing to good health and hygiene procedures. Dairy products are involved in fewer than 3.45% of European outbreaks (EFSA, 2011). European food safety alerts (EFSA, 2011; Koch et al., 2010) suggest that Listeria contamination is more common in soft cheeses made from pasteurised or heattreated milk than in those made from raw milk. Complex microbial communities can contribute to their self-protection. The importance of the role played by a cheese system's microbial biodiversity in limiting pathogen growth has often been mentioned. Several multispecies ⁎ Corresponding author at: INRA, URF 545, 20, Côte de Reyne, 15000 Aurillac, France. Tel.: +33 04 71 45 64 12; fax: +33 04 71 45 64 13. E-mail address: [email protected] (C. Callon). 0168-1605/$ – see front matter © 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.ijfoodmicro.2014.01.003

microbial consortia with antilisteria activity have been identified on the surfaces of complex red-smear cheeses (Bleicher et al., 2010; Eppert et al., 1997; Imran et al., 2010; Maoz et al., 2003; Roth et al., 2010) and on the surfaces or in the cores of Saint-Nectaire cheeses (Millet et al., 2006; Retureau et al., 2010; Saubusse et al., 2007). All these studies suggest that no single strain was responsible for all the inhibitory activities of a consortium and that quite different microbial consortia in terms of species composition, level of diversity and structure were able to inhibit L. monocytogenes. Maoz et al. (2003) described a red-smear cheese consortium whose inhibitory activity was associated with Gram positive non-lactic acid bacteria (Corynebacterium, Brevibacterium etc.). Marine lactic acid bacteria (Vagococcus, Facklamia, Alkalibacterium, Marinilactibacillus) may be involved in inhibition on the surfaces of red smear cheeses (Bleicher et al., 2010; Roth et al., 2010, 2011). Adding a microbial species to a consortium, or omitting one, can increase or decrease its inhibitory properties against Listeria (Callon et al., 2011; Imran et al., 2010). On a cheese agar, the association of

C. Callon et al. / International Journal of Food Microbiology 174 (2014) 98–109

five individual strains of yeast, Gram positive and Gram negative bacteria had inhibiting effects similar to the initial complex consortium (Imran et al., 2010, 2013). In uncooked cheese cores, Listeria was inhibited by an association of four species of lactic and non-lactic acid bacteria (Callon et al., 2011). By successive propagation on cheese surfaces, Monnet et al. (2010) selected a consortium with strong antilisterial activity, composed of Yarrowia lipolytica and the Vagococcus–Carnobacterium– Enterococcus group. The microbial interactions governing inhibition are still unknown. The production of antagonistic substances in supernatant (lactate, acetate (Callon et al., 2011), bacteriocins and heat stable nonproteinaceous molecules (Bleicher et al., 2010)) was found to be associated with inhibition, but no causal effect was demonstrated. Maoz et al. (2003) found Listeria to be in a stressed state when in contact with antilisteria consortia. Elsewhere, the transcriptome of Listeria, monitored by microarray analysis, changed with the induction of genes involved in energy supply, stress response and cell wall synthesis (Hain et al., 2007). Most of the literature deals with consortia from the surfaces of red smear cheeses; studies of the surfaces of other cheese types are scarce. The purpose of this study was to determine how changes in the microbial composition and diversity of a complex antilisterial consortium from the surface of St-Nectaire, an uncooked pressed cheese, modify its antilisterial activities. On the basis of the microbial composition of a previously studied consortium named TR15 (Retureau et al., 2010), several new consortia, differing in their species and strain compositions, were defined. The growth of Listeria on the surfaces of uncooked pressed cheeses made with these consortia was compared with growth on cheeses made with the initial complex consortium, TR15. The microbial population dynamics were compared by culture-dependent analysis and Single Strand Conformation Polymorphism analysis. Production of lactic, acetic, succinic, citric and formic acids was measured by HPLC. 2. Materials and methods 2.1. Preparation of cheeses 2.1.1. Cheesemaking Cheeses (600 g) were manufactured from pasteurised milk collected at an agricultural school farm (ENILV, Aurillac, France), using an uncooked pressed cheese technology (Callon et al., 2011). Milk was pasteurised at 72 °C for 20 s, cooled to 33 °C, then inoculated with 0.6% of a commercial starter culture (MY800, Streptococcus thermophilus; Lactobacillus delbrueckii spp. bulgaricus, Danisco, Paris La Défense, France) and with a commercial mould culture of Penicillium commune (2.5 ml/200 l, Laboratoire Interprofessionnel de Production — LIP, Aurillac, France). Forty-five minutes after adding 40 ml/100 l of calf rennet (Beaugel 520—Ets Coquard, Villefranche sur Saône, France), the curd was cut and gently stirred to eliminate whey. The curds were moulded and pressed for 24 h at 2.1 bar and 22 °C. After pressing, the one-day-old cheeses were vacuum-packed and frozen at −20 °C until surface inoculation. 2.1.2. Preparation of the consortia The complete experimental design of the study is shown in Fig. 1. Different consortia named TR15, TR15-M, TR15-SC and TR15-BHI were prepared for inoculating the cheese surface (Section 2.1.3). The natural microbial consortium (TR15), taken from 1/4 of the rind (3 mm thick) of a raw milk St-Nectaire cheese after 28 days of ripening, was diluted 1/10 in phosphate buffer at pH 7.5 and blended for 4 min using a Stomacher blender (Retureau et al., 2010). The suspension was aliquoted into sterile tubes and frozen at − 20 °C until use. Before inoculation, the suspension was thawed at 25 °C. Three different consortia were reconstituted from this natural microbial consortium TR15. Consortium TR15-SC consisted of 58 strains isolated after inoculating TR15 onto selective media (Fig. 2). For this purpose, 200 microbial

99

isolates were taken from media for lactic acid bacteria (FH, MSE and SB), ripening bacteria (RPF, CRBM), Gram negative bacteria (PCAI) and yeasts (OGA), in the conditions described in Section 2.3.1. The isolates were purified before being identified at species level by DNA extraction (using Easy DNA kit with phenol/chloroform Invitrogen, Cergy Pontoise, France), ribosomal 16S rRNA gene amplification (1450 bp) and sequencing, as described by Retureau et al. (2010). The Blast programme was used to compare the sequences with those available in the GenBank database. In this study, a 99% similarity was taken as the criterion to assign an isolate to a species or group of species. Yeasts were identified by a combination of phenotype tests and sequencing of the D1/D2 domain of the 26S rRNA gene, as described by Callon et al. (2006). Then 58 strains (Fig. 2) were chosen according to their frequency and establishment on the rind, as described by Retureau et al. (2010). Each strain was cultured in 10 ml of broth medium (MRS medium for lactobacilli and leuconostocs, M17 medium for enterococci, BHI for other lactic acid bacteria, Gram-positive catalase-positive bacteria and Gram-negative bacteria, and YPG medium for yeasts). Cultures were centrifuged for 15 min at 5000 rpm and 4 °C. Pellets were then resuspended in sterile milk with 15% glycerol, with 0.5% ascorbate added for Gram positive bacteria but not for Gram negative bacteria or yeasts. They were then frozen at − 20 °C. Each tube used for inoculation was thawed for 5 min in a 30 °C bath, enumerated, and diluted in a pH 7.5 phosphate buffer in order to inoculate the cheese surfaces at the appropriate concentrations (Fig. 2). In order to take better account of strain diversity, a pooled consortium TR15-M was composed as follows: all colonies growing on each of the selective media inoculated with consortium TR15 (FH dilution plate −5, MSE dilution −4, SB dilution −4, RPF dilution −2, CRBM dilution −3, PCAI dilution −3 and OGA dilution −4 (see Section 2.3.1)) were scraped off, making one pool per medium. Each pool was then resuspended in 2 ml of phosphate buffer. Suspensions were washed twice by centrifuging for 15′ at 5000 rpm and pellets were then frozen at − 20 °C. Before use, the pellets were thawed and the suspensions enumerated on their respective isolation media, then diluted in a pH 7.5 phosphate buffer to obtain the appropriate concentrations on the cheese surfaces. The diluted suspensions were pooled together to form consortium TR15-M. A further consortium, TR15-BHI, was composed by pooling all colonies growing on non-selective BHI medium inoculated with TR15 (see Section 2.3.1). BHI medium was selected because numerous bacterial species with antilisterial properties can grow on it, as shown by Monnet et al. (2010). The suspensions were prepared and processed as described above. 2.1.3. Inoculation of experimental cheese surfaces and ripening The one-day-old experimental cheeses made from pasteurised milk were thawed for 5 h at room temperature under a laminar hood and turned over each hour to dry them. The surfaces of all cheeses were inoculated by depositing and spreading 1 ml of the culture of L. monocytogenes (strain 167) with a sterile tooth-brush (Millet et al., 2006). The strain was precultured for 18 h at 37 °C in casein soya broth supplemented with 0.6% yeast extract and then processed as described in Section 2.1.2 for the TR15-SC strains. It was inoculated at 2 to 5 CFU/cm2. The cheese surfaces were then inoculated with either the natural microbial consortium (TR15) or one of the reconstituted consortia (TR15BHI, TR15-SC and TR15-M). Each cheese surface (area = 132 cm2) was inoculated with 1 ml of reconstituted consortium prepared as described in Section 2.1.2. The inoculation levels of microorganism strains or pools were based on mean counts obtained by different microbial enumerations of the natural microbial consortium TR15 on the corresponding medium (Fig. 2). The experiment included a control cheese whose the rind was inoculated only with 1 ml L. monocytogenes strain (167), used as control for Listeria growth.

100

C. Callon et al. / International Journal of Food Microbiology 174 (2014) 98–109

Rinds of 1 day old pasteurized cheese Inoculation with L. monocytogenes strain N167 (2 to 5 cfu/cm²)

Assays

Reference

Inoculation with 4 different consortia

No inoculation

3 reconstituted consortia

1 complex consortium

TR15 Natural consortium from 28d rind of raw milk St-Nectaire cheese inhibiting Listeria (Retureau et al., 2010)

TR15SC

TR15M

TR15BHI

58 strains from the isolate collection from the TR15 inoculated on different media

Pool of bulked bacteria from TR15 growing on FH, MSE, SB, CRBM, RPF, PCAI and OGA media a

Pool of bulked bacteria from TR15 growing on BHI medium

Cheese ripening

Listeria analysis Establishment and dynamics of microflora Culture dependent method Counting on FH, MSE, SB, CRBM, RPF, PCA, BHII and OGA media a

Culture independent method Direct 16S rDNA V2 region SSCP b Analysis of cheese rind total DNA

Identification of isolates on CRBM, BHI and PCAI media

Fig. 1. Experimental design of the study. aAll counting media are described in Section 2.3.1. bSingle strand conformation polymorphism.

Inoculated cheeses were ripened for 28 days in sterile stainless steel boxes in INRA's cellars at 8–9 °C and 98% relative humidity. After 8 and 18 days of ripening, the cheeses were washed with sterile salt water (20% NaCl) using sterile wipes. 2.2. Sampling Rind samples (3 mm thick, 10 g) were taken from inoculated cheeses after 1, 8, 14, 18 and 28 days of ripening in order to perform biochemical analysis and Single Strand Conformation Polymorphism (SSCP) on microbial 16S rDNA (V2 region). 2.3. Microbial analysis 2.3.1. Counting on media The rind samples were blended in phosphate buffer (20 mmol/l KH 2 PO 4 , 0.01 mol/l K 2 HPO 4 ) pH 7.5 using a Stomacher blender (Interscience, St. Nom la Bretèche, France). Each suspension was diluted with Ringer's solution and appropriate dilutions were spread on agar plate media using a spiral system (DS +, Interscience, St Nom la Bretèche, France). L. monocytogenes was counted as described in ISO 11290-2, by an accredited laboratory (LIAL, Aurillac, France). Microbial analysis was performed as described by Retureau et al. (2010). Then the microbial populations were enumerated on the following agar plate media: coagulase positive and negative staphylococci on

Rabbit Plasma Fribrinogen agar (RPF); facultatively heterofermentative Lactobacilli on Facultatively Heterofermentative agar (FH); dextraneproducing leuconostocs on Mayeux Sandine Elliker agar medium (MSE); enterococci on Slanetz and Bartley agar (SB); Pseudomonas on Cephalosporin Fucidin Cetrimide agar (CFC); Gram-positive catalasepositive bacteria on Cheese Ripening Bacteria Medium (CRBM); Gram negative bacteria on Plate Count Agar (PCA) supplemented with milk, 5 mg/l vancomycin and 5 g/l crystal violet (PCAI), and yeasts and moulds on Oxytetracycline Glucose Agar (OGA) medium. In addition to the media used by Retureau et al. (2010), the microbial flora were also counted on Brain Heart Infusion (BHI) medium. All media were purchased from Biokar. On OGA medium, white colonies corresponding to D. hansenii or C. sake were distinguished from cream-coloured colonies corresponding to Y. lipolytica. Colony counts were expressed as log of Colony Forming Units (CFU) per cm2 of cheese rind. Counts below 10 CFU/cm2 were recorded as b 1 log/cm2. As it was difficult to quantify at genus or species level on some media, 647 isolates were taken from CRBM, BHI and PCAI media after culturing 8- and 28-day-old cheese rinds and were identified by 16S rDNA sequencing as described in Section 2.1.2. In order to easily compare the L. monocytogenes populations on the surfaces of the different cheeses during ripening, the area of inhibition (AI) (Wenzel and Marth, 1990) between two ripening days (t1 and t2) was calculated using the following formula: AI = (t2 − t1) / 2 ∗ [(Ct1 + Ct2) − ( Tt1 + Tt2)] (C = count of L. monocytogenes in a

C. Callon et al. / International Journal of Food Microbiology 174 (2014) 98–109

101

TR15 Complex consortium

Counting on media FH, MSE, SB, CRBM a RPF, PCAI and OGA

TR15-M Pools of bulked bacteria from Media = undefined

FH (3.26) b

BHI a

TR15-SC

TR15-BHI

Selection of 58 strains among the collection of 200 identified isolates = defined c

Pools ofbulked bacteria = undefined

Lb. casei/paracasei (13) d Lb. curvatus (1)

MSE (2.22)

Ln. pseudomesenteroides (5)

SB (1.54 )

E. faecalis (3)

Cb. mobile Cb. maltaromaticum

Cb. mobile (1) Marini. psychrotolerans (5)

Brocho. thermosphacta

A. nicotianae/arilaitensis (3) BHI (6.71)

A. bergerei/ardleyensis(4) CRBM (3.92)

Brevi. linens (2)

A. bergerei/ardleyensis Brevi. linens

Brevi. antiquum (1)

Microb. gubbeenense

Brachy. tyrofermentans (1) RPF (1.2)

Leucob. aridicollis/iarus

Staph. vitulinus (2) Staph. xylosus/saprophyticus (1)

Ser. proteamaculans

Prot. vulgaris (1) PCAi (1.54)

Ser. proteamaculans (1) Ps. psychrophila (2) C. sake (7)

OGA (1.68)

Y. lipolytica (2) D. hansenii (3)

Fig. 2. Microbial composition of reconstituted consortia TR15-M, TR15-SC and TR15-BHI and levels of inoculation (log cfu/cm2). aAll media are described in Section 2.3.1. bThe level of inoculation (log cfu/cm2) was defined according to the results of the enumeration of TR15 species on each medium. cThe 58 isolates were chosen from the species able to grow on cheese surfaces. dNumber of inoculated isolates of the species. TR15 = complex consortium; TR15-SC = reconstituted consortium with selected strains from TR15 (58 isolates); TR15-M and TR15-BHI = reconstituted consortia with pools of colonies cultured on FH, MSE, SB, CRBM, RPF, PCAI, CFC and OGA media or BHI medium of the TR15 culture.

control cheese whose surface was not inoculated, and T = count of L. monocytogenes in cheeses whose surfaces were inoculated with microbial consortia).

determined as Pi = ai × 100/Σai, where ai was the peak area and Σai was the sum of the peak areas from all the SSCP patterns.

2.3.2. Single strand conformation polymorphism (SSCP) analysis of the cheese rinds Total genomic DNA was extracted from 1 g of cheese rind sample (control, TR15, TR15-M, TR15-SC and TR15-BHI) after 1, 8, 14, 18, 21 and 28 days of ripening, using the phenol–chloroform method described by Delbès et al. (2007). Partial 16S rRNA gene sequences were amplified using Gram1F-Gram2R primers with Gram1F labelled with a hexachloro derivative of fluorescein (HEX) and Gram2R labelled with fluorescein phosphoramidite (NED) (Saubusse et al., 2007). The PCR-SSCP products were analysed by SSCP capillary electrophoresis on an ABI Prism 310 genetic analyser. The fluorescence signal was analysed using the GeneScan analysis software. The SSCP peak patterns were aligned using the internal standard GeneScan-400 Rox. The dominant peaks in the SSCP patterns were assigned by comparing their migration with that of the bacterial strains from the TR15 bacterial collection. To analyse the different profiles, relative peak proportion Pi was

2.4. Chemical analysis 2.4.1. pH measurements Cheese surface pH was determined after 1, 8, 14, 21 and 28 days of ripening at 3 locations on each cheese using a 926 VTV pH-meter with Ingold electrode 406 MX (Mettler-Toledo S.A., Viroflay, France). The results are the means of the three measurements.

2.4.2. Analysis of organic acid contents of cheese surfaces by high pressure liquid chromatography Organic acids were measured in the cheese rinds after 8, 14 and 28 days of ripening by using High Pressure Liquid Chromatography (HPLC) on an Aminex HPX-87H column (Biorad, Marnes-la-Coquette, France), as described by Leclercq-Perlat et al. (1999). The results were expressed in mg/g of dry rind content.

102

C. Callon et al. / International Journal of Food Microbiology 174 (2014) 98–109

2.4.3. Ability of microbial strains to produce acids on cheese medium The 20 main strains identified from the TR15 complex consortium were tested for their ability to produce lactic, acetic and formic acids after being cultured on a cheese medium (sodium caseinate 12 g/l, casein peptone 3 g/l, yeast extract 1 g/l, CaCl 0.1 g/l, MgSO4 0.5 g/l, NaCl 20 g/l, pH adjusted to 7.0 with chlorhydric acid), supplemented or not with 2 g/l of glucose, galactose, lactose or citrate. Before inoculation, all the strains were pre-cultured in casein-soy broth incubated at 25 °C until the end of the exponential phase. The Petri dishes containing the cheese medium were inoculated with 104 CFU/g of strain and incubated at 25 °C. The inoculated agar was centrifuged at 41,415 g for 15 min at 20 °C. The supernatant was immediately analysed by HPLC as described in Section 2.4.2.

the inhibition area on rinds with TR15, TR15-SC and TR15-BHI were similar (AI (8–1) around 6). From 14 days and until the end of ripening, AI values with reconstituted consortium TR15-BHI were significantly lower than with the natural consortium TR15 (Table 1). After that, overall, TR15-BHI (AI = 41.96) was significantly the most inhibitory of the reconstituted consortia. As shown in Table 1, the pH values were between 5.35 and 5.61 at 1 day and increased sharply in all cheese rinds through ripening, reaching values between 7.39 and 7.80 at the end of ripening. Until 21 days of ripening, the most inhibitory cheese rind, TR15, had the significantly lowest pH values, similar to those of TR15-SC (at 1 and 21 d). At 21 and 28 days, the pH of rind with T15-BHI was higher than that with TR15-M and TR15-SC, but TR15-BHI was nevertheless the most inhibitory.

2.5. Data analysis 3.2. Microbial composition of inoculated consortia For AI, pH and microbial counts at different ripening times, standard analysis of variance (main-effects ANOVA) was performed to compare the effect of the consortia inoculation. Concerning the acid production by strains, the 20 strains tested were classified under an Ascending Hierarchical Classification (AHC) performed using Euclidean distances and Ward's method. The classification incorporated the lactic, acetic and formic acid outputs from the different substrates. The dendrogram obtained made it possible to significantly display four groups at an aggregation distance of 50. This analysis was followed by K-means clustering to establish the characteristics of the four groups. Principal composant analysis with microbial levels, listeria, pH and acid contents of different cheese rinds at 8 and 28 d has been realized. All statistical analyses were performed using Statistica software (Statsoft, version 6.1, Tulsa, USA). 3. Results 3.1. Inhibition of L. monocytogenes and pH evolution on cheese surfaces inoculated with the different consortia L. monocytogenes increased throughout ripening to reach 7.7 log cfu/cm2 at 28 d on the surface of the control cheese inoculated only with starter culture, whereas it did not exceed 4.5 log for any other consortium and was only 3.3 log at 28 d with the natural consortium (TR15) (data not shown). Throughout ripening, the inhibition area was significantly greatest on cheese rind inoculated with TR15 (ΣAI = 70.44) and least after inoculation with TR15-M (ΣAI = 21.56) (Table 1). At 8 days,

The composition of the natural consortium TR15 was evaluated by identifying colonies on selective media (FH, MSE, SB for LAB, RPF and CRBM for Gram-positive catalase-positive bacteria, PCAI for Gramnegative bacteria and OGA for yeasts) and on the non-selective medium BHI. Twenty different bacterial species and 3 yeast species (Fig. 2, Table 2) were identified. On selective media, the lactic acid bacteria identified were Lactobacillus casei/paracasei, Lactobacillus curvatus, Leuconostoc pseudomesenteroides, Enterococcus faecalis, Marinilactibacillus psychrotolerans and Carnobacterium mobile. The Gram positive catalase positive bacteria were Arthrobacter nicotianae/arilaitensis, Arthrobacter bergerei/ardleyensis, Brevibacterium linens, Brevibacterium antiquum, Brachybacterium tyrofermentans, Staphylococcus vitulinus, Staphylococcus xylosus/saprophyticus. The Gram-negative isolates were identified as Pseudomonas psychrophila, Serratia proteomaculans and Proteus vulgaris. Yeast isolates were Candida sake, Yarrowia lipolytica and Debaryomyces hansenii. Besides the species C. mobile, A. bergerei/ardleyensis, B. linens and S. Proteamaculans, already identified on selective media, 16S rDNA sequencing of 23 isolates growing on BHI medium identified 4 other species: Carnobacterium maltaromaticum, Brochothrix thermosphacta, Leucobacter aridicollis/iarus and Microbacterium gubbeenense. Consortium TR15-BHI from the pool of all colonies growing on this medium contained at least these 8 species. TR15-SC was composed of one or more strains of each of the 16 bacterial species and 3 yeast species isolated from selective media (58 strains in all) (Fig. 2) whereas TR15-M resulted from pooling all the colonies growing on all these media. TR15-M contained at least these species but may be more diverse at species or strain level.

Table 1 Statistical analysis (main-effects ANOVA) with inhibition areas of L. monocytogenes and pH values of cheese rinds inoculated with complex consortium TR15 or reconstituted consortia after 1, 8, 14, 21 and 28 days of ripening. Consortia

AI (8–1)

TR15 TR15-BHI TR15-SC TR15-M

5.54 5.89 6.36 5.00

+/− +/− +/− +/−

AI (14–8) 1.99 NS 1.1 NS 2.7 NS 0.3 NS

Consortia

pH 1 d

Reference TR15 TR15-BHI TR15-SC TR15-M

5.35 5.37 5.52 5.36 5.61

+/− +/− +/− +/− +/−

10.3 5.78 5.53 2.45

+/− +/− +/− +/−

0.9 c 0.8 b 2.3 b 0.6 a

AI (18–14)

AI (21–18)

AI (28–21)

10.28 +/− 1.2 b 3.23 +/− 0.5 a 2.33 +/− 0.3 a 2.4 +/− 0.7 a

11.94 +/− 0.6 d 5.55 +/− 1.0 c 4.19 +/− 0.7 b 2.47 +/− 0.5 a

32.52 21.49 17.37 11.93

pH 8 d 0.04 a 0.02 a 0.01 b 0.09 a 0.06 b

6.51 5.99 6.18 6.79 6.45

+/− +/− +/− +/− +/−

pH 14 d 0.15 b 0.10 a 0.07 a 0.12 c 0.05 b

7.76 6.97 7.57 7.51 7.07

+/− +/− +/− +/− +/−

0.04 d 0.05 a 0.06 c 0.03 c 0.07 b

+/− +/− +/− +/−

Σ AI 0.6 c 1.7 b 0.7 b 1.8 a

70.44 +/− 2.29 d 41.96 +/− 5.2 c 35.8 +/− 6.6 b 21.56 +/− 3.8 a

pH 21 d

pH 28 d

7.7 +/− 0.08 a 7.71 +/− 0.06 a 7.93 +/− 0.01 b 7.73 +/− 0.05 a 7.88 +/− 0.03 b

7.43 +/− 0.15 b 7.8 +/− 0.07 d 7.62 +/− 0.04 c 7.39 +/− 0.04 a 7.53 +/− 0.07 b

AI = inhibition areas between two different times = (t2 − t1) / 2 ∗ [(Ct2 + Ct1) − (Tt2 + Tt1)] where C = counts of L. monocytogenes in reference cheese and T = counts of L. monocytogenes in trial cheeses inoculated with different consortia, t2, t1 = ripening times (days) (1 d, 8 d, 14 d, 18 d, 21 d and 28 d). Σ AI = total of AI at all times of ripening. The results of AI are the means of 3 numerations of L. monocytogenes. The results of cheese rind pH are the means of the measurements in 3 places on each cheese. Means within column with different letters are significantly different (P b 0.05) with a b b b c according to statistical test of Newman–Keuls. NS = non-significant. Reference = commercial starter only; TR15 = complex consortium; TR15-SC = reconstituted consortium with strains selected from TR15 (58 isolates); TR15-M, TR15-BHI = reconstituted consortia with pools of colonies cultured on selective media (TR15-M) or non-selective media BHI (TR15-BHI) inoculated with TR15.

C. Callon et al. / International Journal of Food Microbiology 174 (2014) 98–109

3.3.1.1. Lactic acid bacteria. The TR15 cheese rind showed the highest levels of Lactobacillus, Leuconostoc and Enterococcus (Table 3). It also displayed the greatest diversity in lactic acid bacteria (Table 4). At 28 d, L. mesenteroides, Enterococcus gilvus, C. mobile, L. lactis, C. maltaromaticum, Vagococcus fluvialis and M. psychrotolerans were detected on BHI or CRBM media at over 7.5 log cfu/cm2. Enterococcus faecalis (5.7 log cfu/cm2) decreasing during ripening, was at lower level at 28 days. The TR15-BHI cheese rind showed the lowest total counts of Lactobacillus and Leuconostoc at 8 and 28 d (Table 3). However, high levels of L. mesenteroides were detected: 7.5 log cfu/cm2 at 28 d, similar to the level with TR15 (Table 4). As with TR15, the species detected on TR15-BHI rind at 28 d were E. gilvus, C, maltaromaticum, V. fluvialis (all at more than 9.3 log cfu/cm2) and M. psychrotolerans and C. mobile. Also as with TR15, E. faecalis decreased during ripening. The TR15-SC and TR15-M rinds had less diversity in lactic acid bacteria on BHI and CRBM media than the TR15 and TR15-BHI rinds (Table 4). Their dominant populations were M. psychrotolerans and C. mobile, at levels above 7.5 cfu/cm2 at 28 d. TR15-M rind also had the highest level (8.1 log cfu/cm2) of L. mesenteroides at 28 days and of E. faecalis at 8 d. The growth of most of these species was delayed in TR15-SC but L. curvatus was only detected at 8 d in this rind, into which it had been inoculated. E. gilvus, C. maltaromaticum, V. fluvialis and L. lactis were not detected in either of these rinds.

Table 2 Closest relative accession number and % identity of identified species by 16S rDNA sequencing. Species

Closest relative AN

% identity

C. maltaromaticum E. gilvus E. faecalis V. fluvialis B. thermosphacta L. mesenteroides S. vitulinus M. psychrotolerans C. mobile A. nicotianae/arilaitensis B. tyrofermentans B. linens L. aridicollis/iarus S. proteamaculans L. curvatus A. bergerei/ardleyensis S. saprophyticus/xylosus M. gubbeenense Enterobacter spp. L. casei/paracasei L. lactis P. vulgaris B. antiquum P. psychophila

KC213911.1 AB742448.1 KF303455.1 JF690756.1 KC346293.1 KC510002.1 NR024670.1 JQ680459.1 NR040926.1 KF055023.1/NR074608.1 NR026272.1 DQ361016.1 KC764981.1 HQ219939.1 KC787548.1 AJ609633.2/AJ551162.2 AM237352.1/KF192271.1 JQ680448.1 EU430753.1 HM058898.1/CP002391.1 HQ293075.1 NR025336.1 EF459545.1 JQ968688.1

100 100 100 99 100 100 100 100 99 100 98 99 100 100 98 100 100 99 98 100 100 100 97 99

103

3.3.1.2. GRAM-positive catalase-positive bacteria. The TR15 and TR15-BHI rinds had the lowest counts of Gram-positive catalase-positive bacteria (Table 3). The composition of this bacterial group was quite different from one consortium to another and varied greatly during ripening (Table 4). B. thermosphacta and B. tyrofermentans were found in both TR15 and TR15-BHI rinds at 8 days. At 28 days, A. arilaitensis/nicotianae and B. linens dominated on TR15 rind (7.5 log cfu/cm2). Surprisingly, L. aridicollis/iarus, A. bergereri/ardleyensis and B. thermosphacta were all present on the TR15-BHI rind at 28 d although they were not found on TR15 rind (probably under the detection limit). At 8 days, the TR15-M rind displayed the highest levels (between 7.8 log and 9.2 cfu/cm2) and the greatest diversity of Gram-positive catalase-positive bacteria; S. vitulinus, S. saprophyticus/xylosus, M. gubbeenense, B. linens and B. tyrofermentans were all present. At 28 d, the latter two species were no longer detected. The TR15-SC rind was only dominated by S. vitulinus and B. linens at 8 days but was more diverse at 28 days, with B. tyrofermentans, S. saprophyticus/ xylosus, and M. gubbeenense at levels above 8 log cfu/cm2. The inoculated species A. nicotianae/arilaitensis was not detected.

3.3. Dynamics of flora on inoculated cheeses during ripening The dynamics of microbial flora on cheese rinds inoculated with the different consortia were monitored by a culture-dependent method (counting on media and identification of dominant populations growing on media after 0, 8 and 28 days of ripening) and culture-independent methods (direct SSCP analysis of the V2 region of 16S rDNA on cheese rinds at different times during ripening: 1, 8, 14, 18 and 28 days). The dynamics on TR15, TR15-SC, TR15-M and TR15-BHI rinds were compared by combining the results of the different methods (Tables 4 and 5).

3.3.1. Culture-dependent approach Levels of the dominant bacterial populations in the different consortia (TR15, TR15-BHI, TR15-SC and TR15-M) were evaluated at 0, 8 and 28 d, by counting on selective media (Table 3) and by species-level identification of colonies using 16S rDNA sequencing (Table 4).

Table 3 Statistical analysis (main-effects ANOVA) with microbial levels on all cheese rinds counted on selective media at 8 and 28 d. Media

FH

MSE

SB

CRBM

RPF

PCAI

OGA

Lactobacillus

bacteria

Staphylococcus

Gram-bacteria yeasts

Leuconostoc

Enterococcus

Ripening

Total counts (log cfu/cm2) at 8 d TR15 6.6 +/− 0.08 c TR15-BHI 1.9 +/− 0.1 a TR15-SC 5.1 +/− 0.04 b TR15-M 5 +/− 0.1 b

8.7 4.1 6.4 7.4

+/− +/− +/− +/−

0.04 c 0.3 a 0.02 b 0.4 b

6.0 5.8 3.4 4.0

+/− +/− +/− +/−

0.1 c 0.02 c 0.05 a 0.2 b

6.8 7.3 7.1 9.4

+/− +/− +/− +/−

0.03 a 0.5 a 0.2 a 0.3 b

6.5 5.1 7.2 9.5

+/− +/− +/− +/−

0.2 b 0.03 a 0.1 c 0.3 d

7.8 6.7 6.8 5.5

+/− +/− +/− +/−

1.0 b 0.1 c 0.02 b 0.1 a

8.4 7.0 8.0 8.6

+/− +/− +/− +/−

0.3 NS 0.2 NS 0.2 NS 0.5 NS

Total counts (log cfu/cm2) at 28 d TR15 7.2 +/− 0.02 b TR15-BHI 3.7 +/− 0.1 a TR15-SC 7.1 +/− 0.3 b TR15-M 7.7 +/− 0.2 b

7.4 3.7 6.5 7.5

+/− +/− +/− +/−

0.5 b 0.2 a 0.1 b 0.0 b

7.8 7.2 7.9 7.0

+/− +/− +/− +/−

0.7 NS 0.07 NS 0.05 NS 0.3 NS

7.8 7.6 9.5 9.2

+/− +/− +/− +/−

0.4 a 0.04 a 0.2 b 0.05 b

5.1 3.9 7.2 7.7

+/− +/− +/− +/−

0.3 a 0.1 a 0.4 b 0.4 b

8.7 9.5 8.6 9.3

+/− +/− +/− +/−

0.1 a 0.01 b 0.1 a 0.07 b

7.1 7.2 8.2 7.5

+/− +/− +/− +/−

0.1 a 0.04 a 0.1 b 0.08 a

All counting media FH, MSE, SB, CRBM, RPF, PCAI and OGA are described in Section 2.3.1. The results of counts at 8 and 28 d are the means of two analyses. Means within column with different letters are significantly different (P b 0.05) with a b b b c according to statistical test of Newman–Keuls. NS = non-significant level. TR15 = complex consortium; TR15-SC = reconstituted consortium with selected strains from TR15 (58 isolates); TR15-M and TR15-BHI = reconstituted consortia with pools of colonies cultured on FH, MSE, SB, CRBM, RPF, PCAI, CFC and OGA media or BHI medium of the TR15 culture.

104

C. Callon et al. / International Journal of Food Microbiology 174 (2014) 98–109

Table 4 Establishment and dynamics of flora in cheeses inoculated with different consortia, monitored by culture-dependant method: total counts of cheese rind on media and identification of isolates at 8 and 28 d. Bold data in table 4 highlight the dominant counts for each sample at different times. Media

TR15 ia

TR15-BHI 8

28

i

7.5 – – 6.5 – 5.7 b6.5 – – b5 6.5 5.7 7.6

7.5 7.5 7.5 8.3 8.4 – – 7.7 b7.5 6.9 8.4 7.8 8.8

– 6.6 6.5 – 5.7 b6.5 – – – 6.4 b6.5 – –

7.5 – – – – – – 7.5 b7.5 6.9 b7.5 – –

i i

7 6.5 7.5 5.8 6.5 –

7.5 7.9 8 7.7 – –

i

8.4 8.4 nd

6.5 6.5 5.9

TR15-SC

TR15-M

8

28

i

8

28

5.5 7.4 6.5 – 6.5 5.6 b5.5 5.6 b5.5 – – 5.9 7.4

7.5 9.6 9.3 – 9.3 – – 7.4 b7.5 6.3 b7.5 7.3 9.9

2.22 ni ni ni nid 2.54

–

–

– – – – – – – 5.2

9.2 7.8 8.7 8.2 b7.5 7.5 9.3 9.1

– 7.1 5.8 – 5.6 b5.5 – – – – – – –

6.3 6.3 b7.5 7.8 – – 7.8 – – – – – –

3.7 ni

–

–

– – 5.8 – 6.3 7.2 5.5 – –

b7 7.5 8.7 – 8.2 8.9 7.8 8.4 8

6.5 6.2 – – – –

8.2 9.2 – – – –

0.5 ni 0.2

b6 4.9 b6 3.5

b8 8.3 – –

b4.0

7 – 6.8

7.2 – –

2.3 2 1.9

7.8 7.1 6.1

i

8

28

– – – – – – 8.6 8.8 8.5 8.5 8.8 9.0

8.1 – – – – – 7.5 b8 8.4 – 8 b8 8.5

– – – – – 8 – 8.2 7.8 9.2 8.9 8.2 8.6

– – – – – b7 – b7 8.3 9.2 9.5 – 8.3

b7.0

6.3 4.8 3.5 3.5 – –

8.7 8.8 – – – –

8 – 6.3

7.9 nd 6.8

7 7.1 7

b

a) Lactic acid bacteria Leuconostoc mesenteroides Carnobacterium maltaromaticum Vagococcus fluvialis Lactococcus lactis Enterococcus gilvus Enterococcus faecalis Marinilactibacillus psychrotolerans Carnobacterium mobile Total lactic acid bacteria b) Gram + catalase + bacteriab Arthrobacter arilaitensis/nicotianae Brochothrix thermosphacta Leucobacter aridicollis/iarius Brachybacterium tyrofermentans Arthrobacter bergereri/ardleyensis Brevibacterium linens Staphylococcus vitulinus Staphylococcus saprophyticus/xylosus Microbacterium gubbeenense c) Gram negative bacteriab Serratia proteamaculans

BHI BHI BHI BHI BHI CRBM BHI CRBM BHI CRBM BHI CRBM BHI

CRBM CRBM BHI BHI CRBM BHI BHI CRBM BHI CRBM BHI CRBM BHI

BHI PCAI BHI PCAI PCAI PCAI

Pseudomonas psychrophila Enterobacter Proteus vulgaris

i

i

i i i i

i i i i i i i i i

i i i i i

i

i

i

i i

3.7 3.7

ni 3.7 3.7 3.7 1.72 1.72 ni

0.2

i

c

d) Yeasts Candida sake Debaryomyces hansenii Yarrowia lipolytica

OGA OGA OGA

Number of isolates taken from each medium at 8 and 28 d Consortium

TR15

Medium CRBM BHI PCAI

8d 14 34 10

TR15-BHI 28 d 24 41 44

8d 33 39 15

TR15-M 28 d 14 56 21

8d 18 45 20

TR15-SC 28 d 14 56 21

8d 9 31 27

28 d 12 58 23

TR15 = complex consortium; TR15-SC = reconstituted consortium with selected strains from TR15 (58 isolates). TR15-M = reconstituted consortium with pools of colonies from FH, MSE, SB, CRBM, RPF, PCAI and OGA media inoculated with TR15 consortium. TR15-BHI = reconstituted consortium with pools of colonies from BHI medium inoculated with TR15 consortium. All media are described in Section 2.3.1. Results highlighted in grey indicate similarities in cheese rind characteristics. The results are not highlighted in grey as in the file which we had sent. a i = inoculation expressed as log cfu/cm2. The level of inoculation was defined according to the results of the enumeration of TR15 on each medium. b Counts of species by identification by 16S rDNA sequencing of isolates growing on CRBM, BHI and PCAI media (647 isolates described in the table below). All media are described in Section 2.3.1. c Counts of species by the morphology on OGA medium. d ni = not inoculated.

3.3.1.3. Gram negative bacteria. At the beginning of ripening, levels of Gram negative bacteria were lower on rinds with TR15-SC and TR15-M than on those with TR15 and TR15-BHI, but at the end of ripening they were similar (Table 3). S. proteamaculans was the dominant Gram negative population in all rinds (Table 4) except with TR15 and increased sharply between 8 days and 28 days, especially with TR15-SC and TR15-M (N8 log cfu/cm2 at 28 d). At 8 days, the TR15 consortium included P. psychrophila and Enterobacter spp., which were not detected in other consortia. Of these two species, only P. psychrophila was still

detected after 28 days. P. vulgaris, inoculated in the TR15-SC consortium, was no longer detected. 3.3.1.4. Yeasts. At 8 d, levels of yeasts, mainly represented by C. sake on all rinds and D. hansenii on the TR15 rind, were highest on the TR15 and TR15-M rinds (8.4 and 8.7 log cfu/cm2) (Tables 3 and 4). Then they decreased, especially on the TR15 rind (6.5 log cfu/cm2 at 28 d). At 28 d they were highest on TR15-SC rind (8.2 log cfu/cm2). Y. lipolytica was present at lower levels (6 to 7 log cfu/cm2). The

25 2 7

5 6

35

8

48

36

8 5

L. monocytogenes

Leucob. aridicollis/iarius

Brevi. linens

A. bergerei/ardleyensis

Staph. vitulinus

5 Cb. mobile

Lb. delbrueckii b

Lb. curvatus

Lb. casei/ paracasei

Brocho. thermosphacta

36

10 4

64

TR15 = complex consortium; TR15-SC = reconstituted consortium with selected strains from TR15 (58 isolates); TR15-M and TR15-BHI = reconstituted consortia with pools of colonies from FH, MSE, SB, CRBM, RPF, PCAI, CFC and OGA media or BHI medium inoculated with TR15. Results highlighted in grey indicate similarities in cheese rind characteristics. a Proportion Pi of each peak on SSCP profiles by analysis of V2 region (brin Hex) of cheese rind total DNA at 1, 8, 14, 18 and 28 days with Pi = ai ∗ 100/Σai where ai = peak area and Σai = sum of peak area on profile. Only dominant species at different times of ripening are represented. Each cheese rind was co-inoculated with L. monocytogenes and one of the different consortia. b S. thermophilus and L. delbrueckii are the two species inoculated as starter culture in milk. Bold data in Table 5 highlight dominant proportions for each sample at different times.

35

27 5 30 20

47 74 59

16 5

83

9 4

3 5 3

3 2 6 3

6

29

6

38

2

29 Ln. mesenteroides

E. faecalis

30

25 57 15

10

7

20 21

12 91 3 8 100 Str. thermophilus b

40

22

53 33

10 20

17 7

8

6

5 10 34

30

60

5

10

22

6

16

30

70

5

15

40

10

23

d28 d18 d14 38 63

TR15M

d8 d1 d18 d28 d14

TR15SC

d8 d1 d28 d18 d14

TR15BHI

d8 d1 d28 d14 d18 d8 d1

TR15 16S rDNA V2 SSCP peaks a

Table 5 Establishment and dynamics of microflora in cheeses inoculated with different consortia monitored by culture-independent method: V2 region 16S rDNA SSCP analysis of cheese rinds at 1, 8, 14, 18 and 28 days.

C. Callon et al. / International Journal of Food Microbiology 174 (2014) 98–109

105

detection of yeast at 7.0 log cfu/cm2 at 8 d on rind with TR15-BHI may have resulted from contamination, as no yeast was inoculated. 3.3.2. Culture-independent approach by SSCP The V2 region 16S rDNA SSCP profiles varied between cheese rinds inoculated with the different consortia (Table 5). At 1 day, all profiles were dominated by 2 peaks coeluting with S. thermophilus and to a lesser extent L. delbrueckii, inoculated as starter culture. Throughout ripening, the V2 SSCP profiles of the TR15 rind displayed two dominant peaks corresponding to L. mesenteroides and B. thermosphacta. The peak for L. mesenteroides increased proportionately until 14 d (64% of total peak area) and then decreased until 28 d (29% of total peak area). The peak for B. thermosphacta was greatest at 8 days (48% of total peak area), decreased at 14 d (36% of total peak area) and then remained stable until the end of ripening. The other minor peaks of the profiles (b10% of total area) indicated the presence of C. mobile and S. vitulinus throughout ripening and E. faecalis, L. casei/paracasei and L. curvatus from 18 d of ripening. The V2 SSCP profiles of TR15-BHI rind were largely dominated until day 18 by a peak coeluting with B. thermosphacta (83% of total peak area at 8 d, 74% at 18 d); between 14 and 28 d, the L. aridicollis/iarus peak increased proportionately (25 to 30% of total peak area). Peaks coeluting with L. mesenteroides (12% of total peak area at 18 d) and A. bergerei/ardleyensis (4 to 16% of total peak area from 8 d to 28 d) were detected during ripening. Peaks coeluting with S. vitulinus, C. mobile and L. curvatus were present in low proportions (3 to 5% of total peak area) at 8 d on the TR15-BHI profiles. Throughout ripening, the V2 SSCP profiles of the TR15-SC rind were dominated by two peaks corresponding to A. bergerei/ardleyensis and L. curvatus. However, the proportion of A. bergerei/ardleyensis decreased during ripening (57% of total peak area at 8 d, 30% at 28 d) whereas that of L. curvatus increased (20% of total peak area at 8 d, 60% at 28 d). Peaks coeluting with E. faecalis (6% of total peak area at 8 d), C. mobile (7% of total peak area at 8 d), S. vitulinus (10% of total peak area at 8 d), L. mesenteroides (17% of total peak area at 14 d) and L. casei (20% of total peak area at 14 d) and L. aridicollis/iarus (5% of total peak area at 1 d) were sporadically present on TR15-SC profiles. At 8 and 14 days, the V2 SSCP profiles of TR15-M rind were dominated by the peak coeluting with L. aridicollis/iarus (40% of total peak area at 8 d and 70% at 14 d) and to a lesser extent by L. mesenteroides (16% of total peak area at 8 d and 30% at 14 d). At 18 days, the SSCP profiles were more balanced, as besides these two peaks there were peaks coeluting with C. mobile, B. linens and representing 15 and 10% of the total peak area respectively. At 28 d, the peak coeluting with B. linens was predominant (35% of total peak area) whereas those coeluting with L. mesenteroides and L. aridicollis/iarus had decreased (to 23 and 27% of total peak area). 3.4. Acid concentrations The dominant acids found on the HPLC profiles of cheese rinds were lactic, acetic, succinic, citric and formic acids (Table 6). Concentrations of citric acid at 8 days were higher on rind with TR15-M than on those with TR15, TR15-BHI, TR15-SC or the control cheese. Thereafter the levels decreased on all rinds, especially the TR15 rind, on which it was not detected after 28 days of ripening (Table 4). Succinic acid was mainly detected at 28 d on cheese rind with TR15, TR15-BHI and TR15-M, reaching the highest level with TR15-M. At 8 days, lactic acid concentrations (Table 6) were lower on the TR15-M rind (13 mg/g dry cheese) than on those with the other consortia (all around 25 mg/g dry cheese). They increased between days 8 and 14 on cheese rinds inoculated with TR15 and TR15-M, then decreased until day 28. On rinds with the other consortia they decreased throughout ripening. At 14 days, concentrations were notably higher on rind with TR15 (30.2 mg/g dry cheese) than on those with TR15-BHI (14.6 mg/g dry cheese) or TR15-SC (6.20 mg/g dry cheese).

106

C. Callon et al. / International Journal of Food Microbiology 174 (2014) 98–109

Table 6 Concentrations of organic acids on cheese rinds during ripening, measured by HPLC after 8, 14 and 28 days of ripening. Citric acid

Control TR15 TR15-BHI TR15-SC TR15-M

Succinic acid

Lactic acid

Formic acid

Acetic acid

8d

14 d

28 d

8d

14 d

28 d

8d

14 d

28 d

8d

14 d

28 d

8d

14 d

28 d

6.74 6.04 6.71 5.59 7.35

5.27 5.74 4.64 5.16 5.75

5.14 0.00 4.86 3.67 3.91

0.00 0.00 0.00 0.00 1.32

0.00 0.00 0.31 0.00 0.00

0.00 3.12 3.45 0.41 5.17

25.65 26.00 25.64 24.23 13.24

16.55 30.2 14.6 11.89 23.4

9.08 6.77 6.09 4.77 7.8

0.00 0.00 0.00 0.00 0.00

0.00 0.00 0.00 0.00 0.00

0.00 0.95 0.00 0.29 0.00

0.00 0.65 0.09 0.05 1.29

0.00 1.00 1.53 2.06 1.18

0.00 5.29 3.98 2.53 3.28

The results are expressed in mg/g of dry cheese rind. Control = commercial starter only; TR15 = complex consortium; TR15-SC = reconstituted consortium with strains selected from TR15 (58 isolates); TR15-M, TR15BHI = reconstituted consortia with pools of colonies cultured on selective media (TR15-M) or BHI medium inoculated with TR15.

Acetic acid concentrations increased during ripening on all cheese rinds but no acetic acid was detected in the control cheese rind. Acetic acid concentration was highest at 14 d on rind with TR15-SC (2.06 mg/g dry cheese), but at 28 d with TR15 (5.29 mg/g dry cheese) and TR15-BHI (3.98 mg/g dry cheese). Formic acid was only detected at 28 days, at low levels, on rinds with TR15 (1 mg/g dry cheese) and TR15-SC (0.3 mg/g dry cheese). 3.5. Acid-producing ability of microbial flora The 20 main microbial strains identified in the different consortia can be divided into 4 clusters according to their ability to produce acids from glucose, galactose, lactose or citrate added to a cheese medium (Table 7). One group comprised L. casei and L. mesenteroides, characterized by high production of lactic acid from glucose (mean of 1584 mg/l),

galactose (761 mg/l), lactose (844 mg/l) and citrate. This group also produced acetic acid from citrate (92 mg/l), but no formate. The second group, comprising E. faecalis, L. curvatus and C. maltaromaticum, produced less lactic acid than group 1 but more acetic acid from glucose (118 mg/l) and galactose (231 mg/l). It was mainly distinguished from the other groups by its production of formic acid: 46 mg/l from glucose, 21 mg/l from galactose and 19 mg/l from lactose. V. fluvialis, E. gilvus, L. aridicollis/iarus and B. tyrofermentans made up the third group, distinguished by significant production of lactic acid from glucose, galactose and lactose and the highest level of acetic acid production from glucose (320 mg/l), galactose (546 mg/l), lactose (397 mg/l) and citrate (996 mg/l). The last and largest group, comprising B. thermosphacta, M. psychrotolerans, M. gubbeenense, S. thermophilus, S. vitulinus, P. psychrophila, P. vulgaris, A. nicotianae/arilaitensis, S. proteamaculans,

Table 7 Characteristics of acid production on cheese medium by the 20 strains identified in the consortia, measured by HPLC. Clusters

Acid production (mg/l)

Substrates added to cheese mediumb

Lactate

Glucose Galactose Lactose Citrate Glucose Galactose Lactose Citrate Glucose Galactose Lactose Citrate

Acetate

Formate

1 (N = 2)

2 (N = 3)

3 (N = 4)

4 (N = 11)

L. casei L. mesenteroides

E. faecalis L. curvatus C. maltaromaticum

V. fluvialis E. gilvus L. aridicollis/iarus B. tyrofermentans

M. psychrotolerans S. thermophilusa C. mobile B. thermosphacta S. vitulinus A. nicotianae/ arilaitensis M. gubbeenense B. linens P. vulgaris P. psychrophila S. proteamaculans

1584c 761 844 62 85 128 92 174 0 0 0 5

397 143 265 47 118 231 29 120 46 21 19 12

447 214 355 54 320 546 397 996 0 0 0 0

35 15 57 16 91 41 30 62 2 4 4 2

Statistical description of the 4 clusters obtained by Ascending Hierarchical Classification (AHC) using Euclidian distances and Ward method (displayed at a threshold aggregation distance above 50) and followed by K-means of observations. In bold type: acid productions characterizing the most the different clusters. a S. thermophilus = starter strain used for manufacturing the cheeses. b Cheese medium supplemented with 2 g/l of glucose, galactose, lactose or citrate (Materials and methods, Section 2.4.3). c Means of the acid production (mg/l) by the strains composing the cluster.

C. Callon et al. / International Journal of Food Microbiology 174 (2014) 98–109

107

8d 6 5

Acetic acid

Cb. maltaromaticum V. fluvialis E. gilvus

Y. lipolytica Citric acid pH Listeria

4

TR15-BHI

3

CRBM

Fact. 2 : 34,51%

2 1

Brocho. thermosphacta

Succinic acid Microb. gubbeenense Marini. psychrotolerans Staph. saprophyticus Brevi. linens Brachy. tyrofermentans

TR15-M

0 -1

Lactic acid

TR15 SC

-2

Staphylococcus

-3

Carnob. mobile Staph. vitulinus

TR15

-4

Ln. mesenteroides -5 -6 -7

Ps. psychrophila Enterobacter spp. D. hansenii

Lc. lactis -6

-4

-2

0

C. Sake Lactobacillus

2

4

Yeasts

6

8

10

Fact. 1 : 50,41%

28d 8

Lactobacillus

6

Fact. 2 : 33,52%

4

Leuconostoc Carnob. mobile

Formic acid Ps. psychrophila Lc. lactis

TR15

Brevi. linens Y. Lipolytica Staph. vitulinus

2

Ln. mesenteroides Arthro. nicotianae/arilaitensis

Staphylococcus

TR15-M

Marini. psychrotolerans

0

E. gilvus Cb. maltaromaticum V. fluvialis

E. faecalis Microb. gubbeenense CRBM TR15-SC pH yeasts Staph. saprophyticus Brachy. tyrofermentans

-2

-4

Listeria

-6

TR15-BHI

Arthrob. Bergerei/ardleyensis

C. sake

Citric acid

Brocho. thermosphacta Leucob. Aridicollis/iarus

-8 -8

Acetic acid

-6

-4

-2

0

2

4

6

8

Fact. 1 : 51,12% Fig. 3. Principal Composant Analysis with microbial counts and diversity, listeria growth, pH and acid contents of cheese rinds at 8 and 28 d.

B. linens and C. mobile, was characterized by the lowest output of all acids in cheese medium. 4. Discussion Similarities and differences in the microbial composition of the natural consortium TR15 and reconstituted consortia and in their dynamics on cheese rinds were highlighted by culture-dependent methods (counting on different media and isolate identification by 16S rDNA sequencing) and by V2 region 16S rDNA SSCP analysis. BHI medium seems to be a useful medium for revealing bacterial species diversity

as several species of lactic acid bacteria (V. fluvialis, C. maltaromaticum, E. gilvus) and Gram-positive catalase-positive bacteria (M. gubbeenense, L. aridicollis/iarus and A. bergerei/ardleyensis) were only quantified on this medium. Moreover, E. faecalis and L. mesenteroides were counted at 2 to 3 log/cm2 higher on the non-selective BHI medium than on selective SB or MSE media. The 16S rDNA SSCP analyses were less accurate than culture-dependent methods for revealing species diversity, as among the 16 bacterial species identified on TR15 rind by culture dependent methods, 9 species (M. gubbeenense, M. psychrotolerans, A. nicotianae/ arilaiinsis, C. maltaromaticum, B. tyrofermentans, S. saprophyticus/xylosus, E. gilvus, L. lactis and V. fluvialis) were never detected by SSCP analysis.

108

C. Callon et al. / International Journal of Food Microbiology 174 (2014) 98–109

Gram negative bacteria were not detected because they are not amplified by the primers used. The antilisterial activity of the different consortia could be connected with their microbial composition and dynamics during ripening, their acid contents and their pH. They were all more inhibitory than the starter culture alone (Table 1). The highest count of L. monocytogenes at 8 days on rind with TR15-M was positively associated with the highest pH value and succinic and citric acid contents. It was associated with the highest levels of M. psychrotolerans and of Gram-positive catalasepositive bacteria, represented by S. saprophyticus/xylosus, B. linens, M. gubbeenense and B. tyrofermentans (Fig. 3). This is not surprising as most of these species are not able to produce acetic and lactic acids in large amounts (Table 7). The lower counts of L. monocytogenes with all other consortia was positively associated with citrate consumption, higher lactic and acetic acid concentrations, and high levels of and diversity in lactic acid bacteria (such as C. maltaromaticum, V. fluvialis, E. gilvus, L. mesenteroides, B. thermosphacta and L. lactis,) and Gram negative bacteria (P. psychrophila and Enterobacter spp.), especially with TR15 (Fig. 3). C. maltaromaticum, V. fluvialis, E. gilvus and L. mesenteroides can effectively produce large amounts of acetic acid (Table 7). At 28 days, the lower counts of L. monocytogenes in consortia TR15 and TR15-BHI was associated with high levels of C. maltaromaticum, V. fluvialis, E. gilvus and A. nicotianae/arilaiteinsis, which were not detected with the other consortia. Moreover, TR15, with markedly lower counts of Listeria, distinguished from TR15-BHI by high levels of acetic acid and formic acid, high citrate consumption during ripening, and high levels of P. psychrophila, L. lactis and L. mesenteroides. TR15-BHI was characterized by high levels of L. aridicollis/iarus and B. thermosphacta. Monnet et al. (2010), studying antilisterial consortia from red smear cheese, identified the group Vagococcus–Carnobacterium–Enterococcus as potentially having antilisterial properties. Carnobacterium was a component of antilisterial consortia identified on soft cheeses (CailliezGrimal et al., 2007). C. maltaromaticum could also contribute to inhibition due to its ability to produce L-lactic, acetic or formic acids (Afzal et al., 2010; Leisner et al. (2007); Zalán et al., 2010). The ability of Carnobacterium to prevent Listeria growth by bacteriocin production in fish and meat has been documented (Brillet et al., 2005; Laursen et al., 2005; Leisner et al., 2007). A. nicotianae was also a component of antilisterial red-smear cheese consortia identified by Maoz et al. (2003) and Mayr et al. (2004). B. thermosphacta has been isolated from the surfaces of Gorgonzola, Scimudin and Taleggio (Fontana et al., 2010). TR15-SC and TR15-M, with high levels of L. monocytogenes at 28 days, were very similar in their microbial composition, characterized by high levels of Staphylococcus, M. gubbeenense, B. linens and M. psychrotolerans, and in their low levels of acetic acid. M. psychrotolerans, S. vitulinus and M. gubbeenense are unable to produce this acid in large amounts. Nevertheless, these species have been found in complex antilisterial consortia on the surfaces of red smear cheeses (Bleicher et al., 2010; Cocolin et al., 2009; Eppert et al., 1997; Mayr et al., 2004; Roth et al., 2010). Roth et al. (2010) suggested that M. psychrotolerans may inhibit Listeria on Raclette cheese by competing for nutrients It may contribute to the inhibition, as L. monocytogenes growth was higher in the control (with only starter culture). Our results confirm the literature data (Bleicher et al. (2010) and Imran et al. (2010)), showing that the species composition of a consortium is more important than the number of species. In our consortia the bacteriostatic effect against L.monocytogenes may be due to their diversity and the predominance of acid-producing lactic acid bacteria. To date, no study has clearly demonstrated how microbial populations in multi species consortia interact and inhibit L. monocytogenes. It is likely that mechanisms differ from one consortium to another, and an understanding of them is still difficult to achieve. The development of new technology for investigating gene expression, including mutagenesis, will be a powerful tool, when coupled with sequencing, for elucidating microbial interactions in cheese.

Acknowledgments This work received funding from TRUEFOOD, “Traditional United Europe Food”, an integrated project financed by the European Commission under the 6th Framework Programme for RTD (Contract no. FOOD-CT-2006-016264). The authors sincerely thank R. Lavigne and B. Desserre for their technical assistance in the cheesemaking and microbial analysis. We are also grateful to Daniel Picque and especially Armelle Delile (INRA, UMR782 Génie et Microbiologie des Procédés Alimentaires, 78850 Thiverval-Grignon, France) for the HLPC analysis. The English text was revised by Harriet Coleman. References Afzal, M.I., Jacquet, T., Delaunay, S., Borges, F., Millière, J.B., Revol-Junelles, A.M., CailliezGrimal, C., 2010. Carnobacterium maltaromaticum: identification, isolation tools, ecology and technological aspects in dairy products. Food Microbiol. 27, 573–579. Bleicher, A., Obermajer, T., Matijasic, B.B., Scherer, S., Neuhaus, K., 2010. High biodiversity and potent anti-listerial action of complex red smear cheese microbial ripening consortia. Ann. Microbiol. 60, 531–539. Brillet, A., Pilet, M.F., Prevost, H., Cardinal, M., Leroi, F., 2005. Effect of inoculation of Carnobacterium divergens V41, a biopreservative strain against Listeria monocytogenes risk, on the microbiological, chemical and sensory quality of cold-smoked salmon. Int. J. Food Microbiol. 104, 309–324. Cailliez-Grimal, C., Edima, H.C., Revol-Junelles, A.M., Millière, J.B., 2007. Carnobacterium maltaromaticum: the only Carnobacterium species in French ripened soft cheeses as revealed by polymerase chain reaction detection. J. Dairy Sci. 90, 1133–1138. Callon, C., Delbès, C., Duthoit, F., Montel, M.C., 2006. Application of SSCP-PCR fingerprinting to profile yeast community in raw milk Salers cheeses. Syst. Appl. Microbiol. 29, 172–180. Callon, C., Saubusse, M., Didienne, R., Buchin, S., Montel, M.C., 2011. Simplification of a complex microbial antilisterial consortium to evaluate the contribution of its flora in uncooked pressed cheeses. Int. J. Food Microbiol. 145, 379–389. Cocolin, L., Nucera, D., Alessandria, V., Rantsiou, K., Dolci, P., Grassi, M.A., et al., 2009. Microbial ecology of Gorgonzola rinds and occurrence of different biotypes of Listeria monocytogenes. Int. J. Food Microbiol. 133, 200–205. Delbès, C., Ali Mandjee, L., Montel, M.C., 2007. Monitoring bacterial communities in raw milk and cheese by culture-dependent and -independent 16S rRNA gene based analyses. Appl. Environ. Microbiol. 73, 1882–1891. EFSA, European Centre For Disease Prevention and Control, 2011. The European Union Summary Report on trends and Sources of Zoonoses, Zoonotic Agents and Foodborne Outbreaks in 2009. EFSA Journal 9, 2090–2468. EFSA, European Centre For Disease Prevention and Control, 2013. The European Union Summary Report on trends and Sources of Zoonoses, Zoonotic Agents and Foodborne Outbreaks in 2011. EFSA Journal 11, 3129–3378. Eppert, I., Valdes-Stauber, N., Gotz, H., Busse, M., Scherer, S., 1997. Growth reduction of Listeria spp. caused by undefined industrial red smear cheese cultures and bacteriocin-producing Brevibacterium linens as evaluated in situ on soft cheese. Appl. Environ. Microbiol. 63, 4812–4817. Fontana, C., Cappa, F., Rebecchi, A., Cocconcelli, P.S., 2010. Surface microbiota analysis of Taleggio, Gorgonzola, Casera, Scimudin and Formaggio di Fossa Italian cheeses. Int. J. Food Microbiol. 138, 205–211. Hain, T., Chatterjee, S.S., Ghai, R., Kuenne, C.T., Billion, A., Steinweg, C., et al., 2007. Pathogenomics of Listeria spp. Int. J. Med. Microbiol. 297, 541–557. Imran, M., Desmasures, N., Vernoux, J.P., 2010. From undefined red smear cheese consortia to minimal model communities both exhibiting similar anti-listerial activity on a cheese-like matrix. Food Microbiol. 27, 1095–1103. Imran, M., Bré, J.M., Guéguen, M., Vernoux, J.P., Desmasures, N., 2013. Reduced growth of Listeria monocytogenes in two model cheese microcosms is not associated with individual microbial strains. Food Microbiol. 33, 30–39. Jakobsen, R.A., Heggebo, R., Sunde, E.B., Skjervheim, M., 2011. Staphylococcus aureus and Listeria monocytogenes in Norwegian raw milk cheese production. Food Microbiol. 28, 492–496. Koch, J., Dworak, R., Prager, R., Becker, B., Brockmann, S., Wicke, et al., 2010. Large listeriosis outbreak linked to cheese made from pasteurized milk, Germany, 2006–2007. Foodborne Pathog. Dis. 7, 1581–1584. Laursen, B.G., Bay, L., Cleenwerck, I., Vancanneyt, M., Swings, J., Dalgaard, et al., 2005. Carnobacterium divergens and Carnobacterium maltaromaticum as spoilers or protective cultures in meat and seafood: phenotypic and genotypic characterization. Syst. Appl. Microbiol. 28, 151–164. Leclercq-Perlat, M.N., Oumer, A., Bergere, J.L., Spinnler, H.E., Corrieu, G., 1999. Growth of Debaryomyces hansenii on a bacterial surface-ripened soft cheese. J. Dairy Res. 66, 271–281. Leisner, J.J., Laursen, B.G., Prevost, H., Drider, D., Dalgaard, P., 2007. Carnobacterium: positive and negative effects in the environment and in foods. FEMS Microbiol. Rev. 31, 592–613. Little, C.L., Rhoades, J.R., Sagoo, S.K., Harris, J., Greenwood, M., Mithani, V., et al., 2008. Microbiological quality of retail cheeses made from raw, thermized or pasteurized milk in the UK. Food Microbiol. 25, 304–312. Maoz, A., Mayr, R., Scherer, S., 2003. Temporal stability and biodiversity of two complex antilisterial cheese-ripening microbial consortia. Appl. Environ. Microbiol. 69, 4012–4018.

C. Callon et al. / International Journal of Food Microbiology 174 (2014) 98–109 Mayr, R., Fricker, M., Maoz, A., Scherer, S., 2004. Anti-listerial activity and biodiversity of cheese surface cultures:influence of the ripening temperature regime. Eur. Food Res. Technol. 218, 242–247. Millet, L., Saubusse, M., Didienne, R., Tessier, L., Montel, M.C., 2006. Control of Listeria monocytogenes in raw-milk cheeses. Int. J. Food Microbiol. 108, 105–114. 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 the surface of smearripened cheeses. Food Microbiol. 27, 302–310. Retureau, E., Callon, C., Didienne, R., Montel, M.C., 2010. Is microbial diversity an asset for inhibiting Listeria monocytogenes in raw milk cheese? Dairy Sci. Technol. 90, 375–398. Roth, E., Miesher Schwenninger, S., Hasler, M., Eugster-Meier, E., Lacroix, C., 2010. Population dynamics of two antilisterial cheese surface consortia revealed by temporal temperature gradient gel electrophoresis. BMC Microbiol. 10, 74.

109

Roth, E., Schwenninger, S.M., Eugster-Meier, E., Lacroix, C., 2011. Facultative anaerobic halophilic and alkaliphilic bacteria isolated from a natural smear ecosystem inhibit Listeria growth in early ripening stages. Int. J. Food Microbiol. 147, 26–32. Saubusse, M., Millet, L., Delbès, C., Callon, C., Montel, M.C., 2007. Application of single strand conformation polymorphism — PCR method for distinguishing cheese bacterial communities that inhibit Listeria monocytogenes. Int. J. Food Microbiol. 116, 126–135. Wenzel, J.M., Marth, E.H., 1990. Changes in population of Listeria monocytogenes in a medium with internal pH control containing Streptococcus cremoris. Int. J. Dairy Sci. 73, 3357–3365. Zalán, Z., Hudáček, J., Štětina, J., Chumchalová, J., Halász, A., 2010. Production of organic acids by Lactobacillus strains in three different media. Eur. Food Res. Technol. 230, 395–404.