To our knowledge, this work describes the first dynamic study of naturally developing anti-listerial cheese surface consortia. The monitoring of two complex consortia obtained from industrial productions was carried out with TTGE, a culture independent fingerprinting technique which enabled species-level detection of high-GC and low-GC bacteria in separate runs.
Previous studies reported a broad range of biodiversity in smear consortia, with 2 to 15 bacterial species detected [2, 5, 22, 23]. High bacterial diversity was observed in consortium F, with 13 species detected at dominant level by culture independent analysis. The cultivation approach detected only 9 of the 13 species present at dominant level in consortium F, but enabled detection of 6 additional species present at subdominant level. TTGE is a semiquantitative approach with limited sensitivity compared to the cultivation approach. However, as fingerprinting technique, TTGE enabled to overcome the arbitrary selection exercised on the flora by the cultivation step, giving a more complete view of biodiversity at dominant level. The combined use of both approaches led to a detailed knowledge of biodiversity in cheese smear flora, as already observed by Feurer et al. and Mounier et al. [5, 24]. The identification strategy used in the present study for the cultivation approach, i.e. all cultivable isolates grouped by TTGE profiles and subsequent sequencing, enabled the detection of intraspecies diversity differentiation in 3 dominant species. This strategy greatly simplified the identification of bands in the TTGE fingerprints of complex consortia corresponding to intraspecies variability. Consortium M displayed slightly less diversity than F with 10 species detected at the dominant level by culture independent analysis.
A total of 20 species were detected in consortia F and M, including eight coryneform bacteria. C. variabile, C. casei, B. linens and Mc. gubbeenense are common ripening microorganisms of smear cheeses detected on soft cheeses [5, 9] and semi-hard cheeses [2, 8, 23]. Br. tyrofermentans was first isolated from Gruyère cheese  and was recently shown to colonize the surface of soft cheeses [5, 9]. To our knowledge, this is the first time that Br. paraconglomeratum has been detected in cheese although this species has been previously isolated from milk . Agrococcus casei was first isolated from Gubbeen, an Irish semi-hard cheese . Three Staphylococcus species were isolated in addition to coryneforms. St. equorum is common on smear cheeses [6, 8, 27–29] while St. vitulinus was only isolated by Irlinger et al.  from French cheeses. St. epidermidis, a human skin inhabitant, was detected on various Irish semi-hard cheeses [2, 8]. Two Gram-positive marine lactic acid bacteria (LAB) and an uncultured bacterium from marine sediment were also part of the dominant flora. M. psychrotolerans has been detected in the smear of soft cheeses from Germany and France [5, 9]. Alkalibacterium sp. was found to be present in many European cheeses including Tilsiter, a semi-hard smear cheese . We also identified potentially undesirable species of enterococci in the subdominant flora of consortium F. Enterococci have a controversial status in the dairy industry. They are considered naturally occurring ripening organisms for artisan Mediterranean cheese , but also appear as emerging pathogens due to the virulence factors they tend to harbor . To our knowledge, this study is the first report of the presence of Facklamia sp. in cheese. F. tabacinasalis was first isolated from powdered tobacco by Collins et al.  and has recently been detected in raw milk by Delbès et al.  in a French farm producing Saint-Nectaire cheese and by Hantsis-Zacharov and Halpern  in a farm from northern Israel equipped with modern automated milking facilities. The presence of F. tabacinasalis on the surface of smear cheese may constitute a health hazard, as this species was shown to be α-haemolytic on horse blood . Moreover, from six Facklamia species described to date, four were isolated from human clinical specimen .
We observed highly similar microbial community structures of consortia F and M, with 9 species being common to both consortia at dominant level, despite different ripening procedures. High interbatch diversity was described by Rea et al.  in a single cheese ripening facility of Gubbeen, an Irish semi-hard smear cheese, over 8 years production, which may be related to a lack of humidity and temperature control during ripening of Gubbeen cheese. In the present study, the production of Swiss Raclette type cheese with defined production and ripening parameters led to the development of a similar flora in two distinct dairies. The source of this highly diverse flora remains unidentified but possible sources could be the brine bath, skin of the workers or wooden shelves, as shown by Mounier et al.  for Gubbeen cheese.
The high biodiversity is particularly surprising in the case of dairy F, where the smear brine is freshly prepared prior to each smearing and inoculated with a defined ripening culture of only 3 bacterial species. Moreover, the smear brine is applied by a cheese ripening robot that smears the young cheeses first. However, the microflora of the brine bath is not controlled and might be one of the major sources. In particular, the brine bath (18-22% (w/v) NaCl) could be suitable to maintain the two halophilic and alkaliphilic marine LAB detected in consortium F, as some strains of M. psychrotolerans and Al. kapii were shown to grow at salt concentration as high as 21% (w/v) by Ishikawa et al. [37, 38].
Dynamic studies of consortia F and M inoculated at same cell counts on cheese surface revealed a similar sequential development of nine bacterial species, i.e. Lc. lactis, St. equorum, Al. kapii, C. casei, B. linens, C. variabile, an uncultured bacterium from marine sediment, Mc. gubbeenense and Ag. casei. The development of this microbial community prevented growth of Listeria innocua, inoculated at 5 × 103 CFU ml-1 smear brine on cheeses at day 7 and 8, over 60 to 80 days ripening. Contamination at day 7 and 8, i.e. when yeasts reached their highest density, provided optimal growth conditions for Listeria, as shown by the rapid Listeria growth on control cheese. Strong antilisterial activities were shown in this unfavorable condition for consortia F and M. Antilisterial activities of complex undefined cheese surface consortia were already observed in previous studies [9, 15]. Maoz et al.  reported a total inhibition of L. monocytogenens during 40 days ripening of a soft smear cheese with an initial contamination level of 1.6 × 103 CFU ml-1 smear brine.
The surface of smear cheese contains a limited range of substrates supporting growth of microorganisms, mainly lactose and lactate. Lactose is mostly metabolized by LAB during curd acidification and initial ripening. The residual lactose can be metabolized on the cheese surface by yeasts during the first days of ripening, as shown for soft cheeses by Leclercq-Perlat et al. . Lactate metabolized by yeasts into CO2 and H2O leads to deacidification of the cheese surface . As a result, lactate continuously diffuses from the core to the surface of the cheese. Lactate can be totally consumed by surface microorganisms in soft cheeses . Several smear bacterial species, i.e. Brevibacterium aurantiacum, C. casei, C. variabile, Mc. gubbeenense and St. saprophyticus, were shown to use lactate and casaminoacids for growth . In contrast, Listeria sp. can only use a limited range of carbon sources for growth, including glucose, glycerol, fructose and mannose, while no growth occurs on lactate or casaminoacids [43–46]. Premaratne et al.  showed that Listeria monocytogenes may utilize alternative carbon sources, such as N-acetylglucosamine and N-acetylmuramic acid, which are major components of bacterial and fungal cell walls [44, 47]. In addition, the yeast cell wall contains a mannan glycopeptide with mannose , a sugar metabolized by Listeria sp. Listeria growth on smear cheese can therefore be limited by a low availability of carbon source and stimulated by components of smear microorganisms.
Marine LAB ferment glucose into lactate and assimilate mannose [37, 38]. Ishikawa et al.  reported that Al. kapii can utilize a fairly limited range of carbon sources. In the present study, M. psychrotolerans and/or Al. kapii established early on cheeses treated by complex consortia, i.e. between day 14 and day 20. We believe competition for nutrients between marine LAB and Listeria sp. may be involved in Listeria inhibition in the smear since the development of M. psychrotolerans and Al. kapii occurred simultaneously with the decrease of Listeria counts for both cheeses treated with consortium F (first trial and repetition) and for one cheese treated with consortium M (repetition). In addition, Listeria growth on control cheese stopped when M. psychrotolerans and Al. kapii were first detected in the smear, i.e. on day 37. Hain et al.  reported a microarray experiment conducted with the antilisterial complex smear consortium described by Maoz et al. . Genes involved in energy supply were mostly up-regulated after 4 hours of contact between Listeria monocytogenes and the consortium, suggesting that Listeria had entered a state of starvation. While Maoz et al.  detected M. psychrotolerans in the aforesaid smear consortium by cultivation methods, they may have overlooked the presence of Al. kapii or related species.