Bacteriocin‑Producing Enterococci Modulate Cheese Microbial Diversity

Vol.:(0123456789) Microbial Ecology (2024) 87:175 https://doi.org/10.1007/s00248-025-02491-7 RESEARCH Bacteriocin‑Producing Enterococci Modulate Cheese Microbial Diversity ClaudiaTeso‑Pérez1· AreliLópez‑Gazcón1· JuanManuelPeralta‑Sánchez2· ManuelMartínez‑Bueno1· EvaValdivia1,3· MaríaEstherFárez‑Vidal4,5· AntonioM.Martín‑Platero1 Received: 16 October 2024 / Accepted: 6 January 2025 © The Author(s) 2025 Abstract Cheese production involves various lactic acid bacteria (LAB) that break down lactose, milk proteins, and fats, producing key nutrients and influencing the cheese’s flavor. They form communities that play a crucial role in determining the cheese’s organoleptic properties. The composition of cheeses’ microbial communities is shaped by physicochemical factors (e.g., temperature, pH, and salinity) and biological factors (i.e. microbial interactions). While starter cultures are introduced to control these communities, non-starter LAB represent a significant portion of the final microbial assemblage, but their interactions remain unclear. LAB often produce bacteriocins, antimicrobial peptides that antagonize other bacteria, but their role within LAB communities is not fully understood. This study aimed to assess the impact of bacteriocin production on LAB diversity in cheese, using Enterococcus as a model organism, a common bacteriocin producer. We analyzed enterocin production of enterococcal isolates by antimicrobial assays and microbial diversity differences in raw milk cheeses by two approaches: 16S RNA gene amplicon metagenomic sequencing for the whole microbial community and multi-locus sequence analysis (MLSA) for the enterococcal diversity. Our results revealed that LAB communities were dominated by lactococci, lactobacilli, and streptococci, with enterococci present in lower numbers. However, cheeses containing bacteriocin-producing enterococci exhibited higher microbial diversity. Interestingly, the highest diversity occurred at low levels of bacteriocin producers, but this effect was not observed within enterococcal populations. These findings suggest that bacteriocin production plays a key role in shaping LAB communities during cheese ripening, although further research is needed to understand its broader implications in other microbial ecosystems. Keywords Enterococcus· Cheese microbiota· Microbial interactions· Microbial diversity· Bacteriocins· Enterocins Introduction Cheese is a dynamic ecosystem that is constantly influenced by external and internal factors, including milk composition, cheese-making methods, ripening conditions, and interactions among microbial communities. Multiple biochemical reactions and microbial interactions during the ripening process are involved in the development of desirable product sensory and organoleptic characteristics, including a pleasing taste and aroma and protection against spoilage, foodborne pathogens, and negative health effects [1, 2]. The microbial communities in cheese are highly complex, with lactic acid bacteria (LAB) being the most abundant microorganisms. LAB are a heterogeneous group of bacteria with the ability to ferment carbohydrates into lactic acid via homo fermentation, or into different end products (CO2, lactate and acetate, or ethanol) via heterofermentative * María Esther Fárez-Vidal [email protected] * Antonio M. Martín-Platero [email protected] 1 Departamento de Microbiología, Universidad de Granada, Avda. Fuentenueva, S/N, 18071Granada, Spain 2 Departamento de Zoología, Facultad de Biología, Universidad de Sevilla, 41012Seville, Spain 3 Instituto de Biotecnología, Universidad de Granada, C/ Ramón y Cajal, 4, 18071Granada, Spain 4 Departamento de Bioquímica y Biología Molecular III e Inmunología, Facultad de Medicina, Universidad de Granada, 18016Granada, Spain 5 Instituto de Investigación Biomédica IBS. Granada. Complejo Hospitalario Universitario de Granada, Universidad de Granada, 18071Granada, Spain C.Teso-Pérez et al. 175 Page 2 of 15 metabolism [3, 4]. This rapid acidification of fermented milk inhibits the growth of the majority of undesirable microorganisms [5], thereby extending the shelf life of the fermented food. LAB also contribute to the flavor, texture, and nutritional value of the cheese by producing diacetyl, acetoin, acetaldehyde, acetic acid, and other organic acids, and a wide range of volatile compounds [6, 7]. Although the LAB naturally present in milk have traditionally been used in cheese-making, it is now more common to select fully characterized LAB strains as starter cultures to improve the consistency of fermentation [8]. Growth of the cheese microbiota is also controlled by regulating temperature, salt, and humidity levels [9]. Nevertheless, a secondary or adventitious LAB microbiota, known as non-starter lactic acid bacteria (NSLAB), spontaneously develops in cheese made with either pasteurized or raw milk, especially during ripening [4]. This microbiota does not usually contribute to acid production [10] but plays a major role in cheese ripening, influencing the final flavor and texture [11]. In particular, cheese made with raw milk usually has a complex microbiota characterized by the succession of different microorganisms throughout the cheesemaking process, from starter LAB to NSLAB [12]. This succession of communities may be influenced by the ripening conditions and by dynamic interactions among the microorganisms [1], which can be beneficial, neutral, or harmful, and affect the outcome of cheese ripening. Antagonism is a frequent type of interaction in dairy fermentation, either through the production of acids via metabolism of carbohydrates, the presence of bacteriophages, or the production of antimicrobial peptides such as bacteriocins [1]. Bacteriocins are ribosomal synthesized antimicrobial peptides that kill or inhibit the growth of closely related bacteria [13, 14]. They are considered to be safe and natural preservatives with immense potential for utilization alone or alongside other techniques in food preservation [15]. Bacteriocin-producing strains commonly have self-protec- tion mechanisms against their own bacteriocins, allowing their possible utilization as “protective cultures” in cheese to inhibit the growth of pathogens and spoilage microorganisms [16]. The fact that a large majority of LAB produce bacteriocins [15, 17] contrasts with the high diversity of the microbial communities, raising questions about their ecological function. In studies on intertidal marine predators, Paine [18] was the first to propose that negative interactions favor greater biodiversity by preventing one species from outcompeting others. A similar effect has been proposed for antagonistic interactions mediated by bacteriocins in bacterial communities, where bacteriocins may also play an important role in promoting biodiversity by producing ecological units of producers and resistant strains [19]. These ecological units could arise from the dynamics between producing, resistant, and sensitive strains in a similar manner to the game of “rock-paper-scissors.” In this way, the toxin producer can kill the toxin-sensitive strain, the toxin-sensitive strain can outgrow the toxin-resistant strain, and the toxin-resistant strain can outgrow the toxin-producing strain [20]. This model can be extended to multiple populations if they form units of mutually immune strains that act as a single entity in this rock-paper-scissors game against each other [21]. Given this background, the objective of this study was to examine the effect of bacteriocin producer populations on community diversity. To this end, we used Enterococcus as a model organism because there are frequent bacteriocin producers in this genus, and its population levels in cheese are variable, ranging from 103 to 108CFU/g in ripened cheese [22, 23]. This allows to study the effect of enterocin producers under different diversity scenarios on cheese, which is an excellent model system for studying the mechanistic principles that drive microbial diversity [24, 25]. Results obtained indicate a positive effect of enterocin production by enterococcal populations on the overall LAB community biodiversity when bacteriocin-producer populations are small. These findings shed light on the impact of antimicrobial peptide production as a driver of LAB biodiversity. Material andMethods Sampling andMicrobial Enumeration Given the higher diversity of raw milk cheeses [2], the study included 15 different commercial ripened raw milk cheeses (Supplementary TableS1): five from cow’s milk, five from sheep’s milk, and five from goat’s milk. Cheese samples (5g) were homogenized for 2min in 45mL of a prewarmed (37°C), sterile, 2% sodium-citrate solution in sterile plastic bags with lateral filters using a masticator lab blender (IUL Instruments, Barcelona, Spain). Tenfold serial dilutions of this homogenate were then prepared in sterile 2% sodiumcitrate solution up to a dilution of 10–6. A 100 μL aliquot was spread in triplicate on agar plates for bacterial enumeration, using Brain Heart Infusion agar (BHI, VWR Chemicals) for total LAB, and Kenner Fecal agar (KF, VWR Chemicals) for enterococci. Viable counts were obtained after incubation for 3days at 28°C for total LAB enumeration and 37°C for enterococci enumeration. Bacterial counts were calculated as the mean values of three determinations after log transformation. Antimicrobial Assays andMicrobial Isolation The abundance of enterocin producers was estimated by testing randomly 45–52 enterococcal colonies per sample (KF medium) for their antimicrobial capacities, employing Bacteriocin-Producing Enterococci Modulate Cheese Microbial Diversity Page 3 of 15 175 the double-layer plate method of Gratia and Fredericq [26]. Enterococcal colonies were transferred by toothpick onto BHI agar buffered (BHA-B) in 0.1M sodium phosphate buffer at pH7 (in triplicates) and incubated at 37°C overnight. Next day, plates were overlaid with 5mL of BHA-B soft agar (1.8% BHI, 0.8% agar) inoculated at 2% with an overnight culture of the indicator strains Listeria innocua (CECT4030) or Enterococcus faecalis S-47 (lab collection) and left to solidify at room temperature. Plates were incubated again overnight at 37°C for the growth of the indicator strain. Colonies with an evident inhibition zone around the strain spot were considered to be enterocin producers. Subsequently, a representative set of 67 isolates were collected (5–7 from each sample) for further analysis, selected according to their zone of antibacterial activity. Pure cultures were stored in BHI medium containing 20% glycerol at − 80°C. To evaluate cross-immunity between selected isolates, all strains were tested against each other using the aforementioned double layer plate technique. Plates were incubated again overnight at 37°C for growth of the indicator strain, and colonies with an evident inhibition zone around the strain spot were considered to be enterocin producers. DNA Extraction Enterococcal populations and microbial communities were characterized by extracting total DNA from each strain and each cheese sample, following the modification of the salting-out procedure (MSOP) of Martin-Platero etal. [27]. Briefly, cells from 1mL of an overnight culture of each isolate or cells from 2.5g of cheese were resuspended in 100 µL of lysozyme buffer and incubated for 30min at 37°C to hydrolyze the cell wall, followed by the addition of 600 µL of lysis buffer and a further incubation step at 80°C to facilitate cell lysis. Cell debris and proteins were then removed by adding 200 µL of protein precipitation solution, mixing, chilling on ice for 10min, and centrifuging for 10min at 20,000 × g. DNA was then precipitated with an equal volume of isopropanol and washed with 70% ethanol. The DNA was finally dried to remove any ethanol trace and dissolved in 200 µL of 0.5 × Tris–EDTA buffer (Tris-ClH 10mM pH 8, EDTA 1mM). DNA concentrations were measured using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific). Multi‑Locus Sequence Analysis (MLSA) Representative bacterial isolates (67 strains; 5–7 from each sample) were characterized by MLSA to estimate the population variability of the genus Enterococcus in each cheese sample. To this end, we constructed a phylogenetic tree based on the concatenated sequences of five housekeeping genes (Table1): adk (adenylate kinase), atpA (ATP synthase, alpha subunit), gyd (glyceraldehyde-3-phosphate dehydrogenase), gdh (glucose-6-phosphate dehydrogenase) [28], and groEL (chaperone GroEL). Primers for groEL gene were designed using Hyden software [29], allowing for two mismatches and two degenerate positions and targeting an amplicon size between 400 and 600bp. Each gene for each bacterial isolate was PCR-amplified using the following conditions: an initial denaturing step of 94°C for 3min followed by an amplification step of 35 cycles of 30s at 94°C, 30s at 50°C, and 30s at 72°C, with a final extension of 5min at 72°C. Reactions were performed in a final volume of 50 µL with Taq polymerase (1x) and buffers from IBIAN Technologies. Primers were used at a concentration of 0.5µM, and 100ng of DNA was added. PCR products were purified with the NucleoFast 96 PCR clean-up kit (Macherey–Nagel) and sequenced with PCR forward primers by Sanger sequencing [30], using the Stab Vida (Universidad de Nova de Lisboa) sequencing service. Next, each sequence was taxonomically assigned according to the best blast hit on the NCBI blast tool (https:// blast. ncbi. nlm. nih. gov/ Blast.cgi) [31]. Finally, a phylogenetic tree was constructed from the concatenated sequences. First, each gene was aligned and Table 1 Primers used for MLSA analysis Primer pair Gene Primer sequence (5′- > 3′) Amplicon size (bp) Reference adk adk2 Adenylate kinase TAT GAA CCT CAT TTT AAT GGG GTT GAC TGC CAA ACG ATT TT´ 437 [28] atpA1 atpA2 ATP synthase, alpha subunit CGG TTC ATA CGG AAT GGC ACA AAG TTC ACG ATA AGC CAC GG 556 gyd1 gyd2 Glyceraldehyde-3-phosphate dehydrogenase CAA ACT GCT TAG CTC CAA TGGC CAT TTC GTT GTC ATA CCA AGC 395 gdh1 gdh2 Glucose-6-phosphate dehydrogenase GGC GCA CTA AAA GAT ATG GT CCA AGA TTG GGC AAC TTC GTC CCA 530 GroELF GroELR Chaperone GroEL GYG AAA AAT TWC AAG AAC G ACG ACW GCT TCA GTY GTT AA 480 This study C.Teso-Pérez et al. 175 Page 4 of 15 trimmed with MEGA-X software (version 10.1.8) [32] using the clustalW algorithm [33]. Then, the concatenation of the five genes was used to construct a maximum likelihood phylogenetic tree in MEGA-X with 1000 bootstrap replications following the Tamura-Nei evolution model (ML heuristic method: nearest-neighbor-interchange, number of threads: 3) [34]. Data Analysis 16S rRNA Metagenomic Sequencing The microbial composition of cheese samples was determined by constructing and sequencing 16S rRNA metagenomic libraries for each sample. We constructed 16S rRNA libraries corresponding to the V4 variable region by a two-step amplification approach. The first primer pair (Mi_ U515F, 5′- TCGTC GGCAG CGTCA GATGT GTATA AGAGA CAGGT GCCAG CMGCC GCGGT AA −3′ and Mi_E786R, 5′- GTCTC GTGGG CTCGG AGATG TGTAT AAGAG ACAGG GACTA CHVGG GTWTC TAAT −3′) contained the primer sequences of U515F and E786R targeting the V4 region of the 16S rRNA gene with partial overlap of Illumina primers, as previously described [35]. PCR was carried out in a final volume of 25 µL containing 12.5 µL of Phusion Flash High-Fidelity PCR Master Mix (Thermo Scientific™), 0.3µM of each primer, and 5 µL of template DNA. The amplification program comprised an initial denaturing step of 98°C for 10s followed by an amplification step of 20 cycles of 1s at 98°C, 5s at 52°C, and 5s at 72°C, with a final extension of 1min at 72°C. PCR was purified using the MEGAquick-spin™ Plus Total Fragment DNA Purification Kit (iNtRON Biotechnology) and was re-amplified in a second PCR to introduce a unique combination of two barcodes for each sample. This PCR was performed in a final volume of 25 µL containing 12.5 µL of Phusion Flash High-Fidelity PCR Master Mix (Thermo Scientific™), 0.4µM of each primer, and 5 µL of the purified PCR product from the first PCR. The amplification conditions were an initial denaturation step of 98°C for 10s followed by an amplification step of 9 cycles of 1s at 98°C, 5s at 55°C, and 5s at 72°C, with a final extension of 1min at 72°C. This second PCR was purified again using the aforementioned kit. Next, the DNA concentration was measured using a Qubit® 3.0 Fluorometer (Invitrogen, Carlsbad, CA, USA) and normalized to the same concentration. High-throughput sequencing was carried out on an Illumina MiSeq platform at the Scientific Instrumental Center of the University of Granada (CIC-UGR, Spain). Sequences were uploaded to the Sequence Read Archive (SRA) on the Genbank—NCBI webpage (https:// www. ncbi. nlm. nih. gov/ sra) under the PRJNA988471BioProject (BioSamples- SAMN36020620 to SAMN36020683). Sequences andData Analysis QIIME2 2021.11 was used to process the 16S Illumina reads [36]. First, primers were trimmed with the cut adapt plugin [37], discarding untrimmed sequences. Then, amplicon sequence variants (ASVs) were inferred from paired reads by the dada2 plugin [38]. Given the expected insertion size of 253 bases for the construction, ASVs shorter than 245 bases were filtered out. Next, all ASVs were aligned by the mafft method [39] and used to construct a phylogeny. The representative sequences were taxonomically classified using the Silva-138 database clustered at 99% similarity [40] by the classify-sklearn method [41]. Differences in bacterial diversity were explored by estimating alpha diversity from the ASV richness, the Shannon diversity index [42], and Faith’s phylogenetic diversity (PD) [43] in QIIME2 using a rarefied table at a sequencing depth of 17,000 sequences per sample. We constructed general linear models (GLM) with the diversity indices as dependent variables and the presence of Enterococcus-bacteriocin producers as fixed factor. Parametric tests were applied, because all variables followed a Gaussian distribution (Shapiro–Wilk’s test of normality, p > 0.05). In addition, the effect of antimicrobial strains on enterococci population diversity was determined by using Faith’s PD to estimate the alpha diversity [43], quantifying this index by using a phylogenetic tree inferred from MLSA results. We also calculated the population richness in each cheese sample by assigning each strain with a sequence type (ST), which was determined by clustering unique sequences of each gene using the USEARCH program (version 11.0.66; identity = 100%) [44]. PHYLOViZ software (version 2.0) was used to identify strains with the same allelic profile [45]. Strains grouped in the same cluster in four out of the five genes were considered the same population. The non-parametric Kruskal–Wallis test was applied to explore the relationship between the diversity indices and the presence of Enterococcus-bacteriocin producers, because none of the indices followed a Gaussian distribution (Shapiro–Wilk’s test of normality, p > 0.05). Pearson’s correlation test was also applied to evaluate the relationship between the proportion of enterococci producers and the corresponding alpha diversity. Finally, the Student’s t-test for dependent variables was used to establish the significance of differences in sensitivity between strains from the same cheese and those from different cheeses. Statistica 10.0 [46] was used for all statistical analyses. Bacteriocin-Producing Enterococci Modulate Cheese Microbial Diversity Page 5 of 15 175 Results Differential Diversity Among Samples Microbial counts and compositions varied widely among the studied LAB communities (n = 15). The total LAB microbial count ranged from 2.8 × 105 to 2.6 × 108CFU/g (Supplementary TableS1; Fig.1), and the enterococcal count ranged from 6.6 × 102 to 1.1 × 106CFU/g (Supplementary TableS1; Fig.1) except in two samples that showed no growth (not depicted in Fig.1). Hence, enterococcal populations were on average threefold lower than total LAB populations. According to the corresponding 16S rRNA metagenomes, LAB communities were dominated by lactococci and streptococci, followed by lactobacilli, at proportions that varied among samples (Table2 and Fig.2). In all samples, populations of enterococci were four to five orders of magnitude lower than those of lactococci or streptococci, including samples where enterococci did not grow in their selective media (KF). The total number of ASVs ranged from 12.83 to 53.53, with the ranking for enterococci ranging from 8 to 40th position (average of 24th position) according to their abundance in the samples. Alpha diversity scores ranged from 0.1593 to 3.4817 for the Shannon index and from 1.190 to 3.360 for Faith’s PD. MLSA identified four species of enterococci present in our cheese samples: E. faecalis, E. faecium, E. durans, and E. hirae (Fig.3). E. faecalis and E. faecium were the most frequently isolated species. The phylogeny inferred from MLSA results showed no homogenous clusters in relation to the sample or antimicrobial activity, indicating that the isolated enterococci were not part of a clonal population. In the case of E. faecalis, Fig. 1 Cheese microbial counts. Total LAB counts and their corresponding enterococci counts are shown in logarithmic scale. Total LAB counts ranged from 5.4 to 8.4 log10 (cfu/g), while enterococci counts were on average threefold lower than total LAB counts; no growth of enterococci was observed in two samples. Colors indicate milk source Table 2 The ten most abundant bacterial genera per cheese according to 16S rRNA metagenomic analysis. All values are expressed as percentages Cheese sample Lactobacillus Lactococcus Streptococcus Leuconostoc Enterococcus Acinetobacter Staphylococcus Ochrobactrum Pseudomonas Chryseobacterium AM1901QU01 50.42 30.17 9.88 7.67 0.56 0.45 0.25 0.14 0.13 0.11 AM1901QU02 4.71 22.28 72.36 0.43 0.05 0.00 0.00 0.01 0.00 0.00 AM1902QU03 31.80 41.77 12.74 0.00 0.08 0.00 12.05 0.51 0.03 0.00 AM1902QU04 28.61 67.70 0.75 1.91 0.05 0.00 0.61 0.05 0.01 0.20 AM1902QU05 4.24 1.25 93.98 0.00 0.07 0.09 0.00 0.04 0.00 0.00 AM1902QU07 1.32 96.43 1.95 0.00 0.04 0.11 0.03 0.05 0.00 0.02 AM1902QU08 12.06 8.60 79.12 0.05 0.00 0.01 0.00 0.02 0.01 0.00 AM1902QU09 15.96 30.94 51.49 0.00 0.01 0.03 1.23 0.05 0.00 0.00 AM1902QU10 1.37 54.03 44.51 0.01 0.00 0.01 0.03 0.02 0.00 0.00 AM1902QU11 41.12 0.04 58.17 0.00 0.01 0.00 0.00 0.01 0.00 0.00 AM1902QU12 19.17 14.96 57.17 8.27 0.01 0.00 0.03 0.08 0.00 0.00 AM1903QU14 8.20 21.06 67.96 0.72 0.01 0.91 0.09 0.03 0.05 0.01 AM1903QU15 16.00 80.59 2.74 0.34 0.12 0.09 0.01 0.04 0.00 0.00 AM1903QU16 29.64 68.91 0.37 0.04 0.03 0.01 0.81 0.12 0.01 0.00 AM1903QU17 0.50 98.87 0.09 0.00 0.00 0.12 0.00 0.02 0.06 0.11 C.Teso-Pérez et al. 175 Page 6 of 15 Fig. 2 Bacterial genera present in raw milk cheeses. In all samples, the most abundant genera were Lactococcus, Streptococcus, Lactobacillus, and Leuconostoc. Color legend shows the ten most abundant genera overall ranked from most to least abundant Bacteriocin-Producing Enterococci Modulate Cheese Microbial Diversity Page 7 of 15 175 Fig. 3 Phylogenetic tree of Enterococcus strains isolated from cheese made from raw milk. The phylogenetic tree is based on multi-locus sequence analyses (MLSA). From left to right, the first annotation bar represents the species: E. faecalis (green), E. faecium (orange), E. durans (blue), and E. hirae (red); the second bar represents the cheese sample and the third and fourth bars represent activity against L. innocua and E. faecalis S47, respectively C.Teso-Pérez et al. 175 Page 8 of 15 however, the genetic distance between strains was very short, with low bootstrap values. Nevertheless, determination of enterococcal population richness by the assignation of type sequences yielded 57 different sequence types, observing 2–7 different type sequences in each cheese sample. Differential Antagonistic Producer Frequency Among Enterococci Between forty-four and fifty-two bacterial strains from each cheese were screened for antimicrobial activity, finding activity against E. faecalis S-47 in 23.64% of strains, activity against L. innocua in 18.5%, and activity against both in 21.25% (Fig.4 and Table3). Only three cheese samples had no strains with antimicrobial activity. Impact ofEnterocin Producers onPopulation andCommunity Diversity Enterococcus diversity was not significantly affected by the presence of enterocin producers in enterococcal populations. Given the lack of sharp clusters in the MLSA (Fig.3), we estimated the population richness and used the phylogenetic tree inferred from MLSA results to quantify Faith’s PD [43] in each sample. Neither the population richness estimation Fig. 4 Enterocin producers. Proportion of enterocin producers with antimicrobial activity against E. faecalis S-47 (yellow circles), L. innocua CECT4032 (blue circles), and both indicator strains (red circles) Table 3 Percentage of enterocin producers per sample Cheese sample % enterocin producers against E. faecalis S47 % enterocin producers against L. innocua AM1901QU01 8.16 0 AM1901QU02 0 0 AM1902QU03 29.54 20.45 AM1902QU04 11.54 1.92 AM1902QU07 0 0 AM1902QU08 36.54 32.69 AM1902QU09 23.08 26.92 AM1902QU10 46.15 50 AM1902QU11 30.43 30.43 AM1902QU12 28.85 28.85 AM1903QU14 39.2 39.2 AM1903QU15 0 0 AM1903QU16 53.8 11.5 Table 4 Effect of enterocin-producing populations on enterococcal diversity. Kruskal–Wallis non-parametric test to evaluate effects of the presence/absence of Enterococcus-bacteriocin producers on alpha diversity indices of the enterococci populations H N p Faith’s phylogenetic diversity index 1.02 13 0.31 Population richness 0.15 13 0.69 Fig. 5 Boxplot showing the difference in sensitivity between enterococci in the same cheese sample and those in different cheese samples Bacteriocin-Producing Enterococci Modulate Cheese Microbial Diversity Page 9 of 15 175 nor Faith’s PD significantly differed between producers and non-producers (Table4). Regarding the cross-immunity of each isolate (Supplementary TableS2), the frequency of inhibition was lower for enterococci that were present in the same cheese sample than for those that were not (Fig.5). This difference was not statistically significant, although it was close-to-significant (p = 0.068) (Table5). In contrast, the presence of enterocin producers increased the diversity of the whole community, being associated with a close-to-significant increase in ASV richness and Faith’s PD and with a statistically significant (p < 0.05) increase in Shannon index score (Table6 and Fig.6). This relationship between greater diversity and the presence of enterocin producers was explored by testing the correlation between these variables in the samples that contained enterocin producers. A negative relationship was observed between diversity and enterocin producers in these samples (Tables7 and 8), but statistical significance was only reached for the Shannon index with E. faecalis as indicator strain (Fig.7). Hence, maximum Table 5 Student’s t-test of dependent variables to compare sensitivity between enterococci from the same cheese and those from different cheeses d.f. degree of freedom d.f N p Same cheese * different cheeses 1.02 13 0.068 Table 6 General linear models on effects of the presence/absence of Enterococcus-bacteriocin producers on alpha diversity indices of the bacterial communities of cheeses d.f. degree of freedom The first number is the d.f. of the independent variable and the second is the d.f. of the error term. Significant p-value (p < 0.05) highlighted in bold d.f F p Shannon diversity index 1;11 6.12 0.031 Faith’s phylogenetic diversity index 1;11 1.90 0.196 AVS richness 1;11 3.74 0.079 Fig. 6 Boxplot showing differences in Shannon index (a), Faith’s phylogenetic diversity (b), and amplicon sequence variant richness (c) between raw milk cheeses with and without the presence of Enterococcusbacteriocin producers. Dots represent the average, boxes the standard error of the mean, and whiskers the confidence interval. *A significant difference