Using this cheese mimicking matrix we screened the

Using this cheese-mimicking matrix, we screened the antifungal activity of 44 LAB fermented milk-based products and 23 LAB isolates used as protective cultures against 4 fungal targets. The LAB were obtained from the culture collections of CIRM-BIA (Centre International de Ressources Microbienne-Bactéries d'Intérêt Alimentaire, Rennes, France) and LUBEM (Laboratoire Universitaire de Biodiversité et Ecologie Microbienne, Plouzané, France). For MMP-2/MMP-9 Inhibitor I product preparation, a 10%-reconstituted low-heat milk supplemented with 45% anhydrous milk fat and 0.5% litmus (LH medium) was sterilized for 30 min at 110°C. Then, it was individually inoculated with a suspension of the LAB to be tested (1% vol/vol), obtained after 2 subcultures of 24 h at 30°C in MRS broth (Difco, Le Pont de Claix, France). The LH medium inoculated with LAB was then incubated for 20 h at 30°C; 100 µL of the resulting fermentation product was then deposited on the surface of each mini-cheese. After drying for 2 h at room temperature under laminar air flow, plates were surface inoculated with 1 of 4 fungal targets: Mucor racemosus UBOCC-A-116002, Penicillium commune UBOCC-A-116003, Galactomyces geotrichum UBOCC-A-216001, or Yarrowia lipolytica UBOCC-A-216006, which were previously isolated from dairy products (Garnier et al., 2017) and obtained from the Université de Bretagne Occidentale Culture Collection (UBOCC, Plouzané, France). Except for yeasts that were first cultured in potato dextrose broth, 10 µL of suspension containing 5 × 103 spores or cells/mL, obtained as previously described (Delavenne et al., 2012), was spotted at the center of each mini-cheese (1 tested fungus/plate). Plates were then incubated at 12°C for 5, 6, and 8 d for M. racemosus, G. geotrichum, and P. commune and Y. lipolytica, respectively. Antifungal activity was then determined by visually evaluating fungal growth compared with a negative control without any fermentation product (Figure 1). To test the antifungal activity of LAB isolates for potential use as protective cultures, we applied the same methodology except that LAB isolates to be tested were individually suspended in sterilized milk to reach a final concentration of 107 cfu/mL of retentate and inoculated concomitantly with the commercial starter MA016, pH indicator, and rennet before distribution in 24-well plates. In this context, 23 LAB cultures were tested in duplicate against the same fungal targets as those described above, after 2 pre-cultures for 24 h at 30°C in MRS broth. Incubation to obtain the mini-cheeses was performed as described above. Plates were then surface-inoculated with 1 of the 4 fungal targets and antifungal activity was evaluated as described above. All tested fungal species grew well in the cheese-mimicking model (Supplemental Figure S1; https://doi.org/10.3168/jds.2017-13518) and the results obtained for both replicates were similar, confirming the suitability of this matrix to sustain fungal growth and the reproducibility of the screening method. Among the 46 tested LAB fermentation products, 25 showed antifungal activity against at least one fungal target, and 22 out of the 23 tested protective cultures showed antifungal activity against at least one target. More precisely, intermediate antifungal activities against at least one fungal target were observed for 14 (30%) fermentation products and 12 (55%) isolates, whereas complete inhibition against at least one fungal target was only observed for LAB isolates (n = 10; 45%; Table 1). The most active (++ and +++) isolates or fermentation products corresponded to Lactobacillus casei, Lactobacillus plantarum, Lactobacillus brevis, Lactobacillus paracasei, Leuconostoc mesenteroides, and, to a lesser extent, Lb. rhamnosus. Antifungal activity varied depending on the utilization mode (protective culture or fermentation product). For example, for Lb. plantarum strains tested both for their fermentation products and as protective culture, fermentation products showed no activity against M. racemosus but could inhibit the latter when used as protective cultures (Table 1). In contrast, Lb. brevis CIRM-BIA 608 showed activity only against M. racemosus and P. commune when used as a fermentation product. None of the tested isolates of Lactobacillus pentosus, Lactobacillus sakei, or Leuconostoc citreum showed any activity (Table 1). As expected, within a species, not all tested isolates showed the same antifungal activity. For example, each tested Lb. casei isolate showed antifungal activity but the inhibited fungal target and the obtained antifungal scores were different among them, confirming that antifungal activity is a strain-dependent trait, as suggested by Cortés-Zavaleta et al. (2014). Finally, under the tested conditions, P. commune was the most frequently inhibited target, followed by M. racemosus and G. geotrichum. None of the strains used to produce fermentation products was able to inhibit growth of Y. lipolytica, whereas only 3 strains used as protective cultures (Lb. brevis CIRM-BIA 128, Lb. plantarum L244, and Leu. mesenteroides CIRM-BIA 1187) showed intermediate antifungal activity against this target (Table 1). Overall, antifungal activity of the tested strains was higher when they were used as protective cultures; indeed, all isolates did show antifungal activity. The reproducibility of antifungal activities obtained with 2 batches of mini-cheeses for 2 protective cultures and fermentates was evaluated (Supplemental Figure S2; https://doi.org/10.3168/jds.2017-13518). The obtained antifungal activities showed a good reproducibility despite slight variation.