Dialogue between Staphylococcus aureus SA15 and Lactococcus garvieae strains experiencing oxidative stress

Background Staphylococcus aureus is an important foodborne pathogen. Lactococcus garvieae is a lactic acid bacterium found in dairy products; some of its strains are able to inhibit S. aureus growth by producing H2O2. Three strains of L. garvieae from different origins were tested for their ability to inhibit S. aureus SA15 growth. Two conditions were tested, one in which H2O2 was produced (high aeration) and another one in which it was not detected (low aeration). Several S. aureus genes related to stress, H2O2-response and virulence were examined in order to compare their level of expression depending on the inoculated L. garvieae strain. Simultaneous L. garvieae H2O2 metabolism gene expression was followed. Results The results showed that under high aeration condition, L. garvieae strains producing H2O2 (N201 and CL-1183) inhibited S. aureus SA15 growth and impaired its ability to deal with hydrogen peroxide by repressing H2O2-degrading genes. L. garvieae strains induced overexpression of S. aureus stress-response genes while cell division genes and virulence genes were repressed. A catalase treatment partially or completely restored the SA15 growth. In addition, the H2O2 non-producing L. garvieae strain (Lg2) did not cause any growth inhibition. The SA15 stress-response genes were down-regulated and cell division genes expression was not affected. Under low aeration condition, while none of the strains tested exhibited H2O2-production, the 3 L. garvieae strains inhibited S. aureus SA15 growth, but to a lesser extent than under high aeration condition. Conclusion Taken together, these results suggest a L. garvieae strain-specific anti-staphylococcal mechanism and an H2O2 involvement in at least two of the tested L. garvieae strains. Electronic supplementary material The online version of this article (10.1186/s12866-018-1340-3) contains supplementary material, which is available to authorized users.


Background
Staphylococcus aureus is an opportunistic human pathogen that can be responsible for food poisoning [1]. Its pathogenic activity is due to the production of various enzymes and toxins. It can be found in different environments including milk and dairy products [2]. In cheese, its level should not exceed 10 5 CFU.g − 1 (European Community Regulation No. 852-853/2004).
Lactococcus garvieae is an ubiquitous LAB (Lactic Acid Bacteria) that can be found in various fermented foods including dairy products [3,4], in fish, ruminant or human microbiota [5] and can be associated with pathologies such as fish lactococcoses [6,7]. LABs such as Lactococcus lactis or Lactococcus garvieae are able to inhibit the proliferation of pathogens in cheese by the production of hydrogen peroxide [8], bacteriocins [9], by competition for nutrients [10,11] or by acidification of the medium [12][13][14]. Depending on its concentration, hydrogen peroxide has a bactericide or a bacteriostatic effect on S. aureus [15]. L. garvieae raises our interest because a specific dairy strain of this LAB, N201 strain, can inhibit S. aureus growth by the production of H 2 O 2 [2,16,17]. Moreover, it has a strong technological potential as a ferment for cheese production [18] and has almost no effect on acidification of the medium, compared to other LABs [2]. The inhibitive properties of L. garvieae N201 were confirmed on 2 strains of S. aureus: a human pathogenic strain, MW2 [19] and a non-pathogenic dairy strain, SA15, isolated from Saint-Nectaire cheese [17]. Delpech et al. [20] showed that S. aureus had no effect on L. garvieae growth and H 2 O 2 -related gene expression. On the contrary, L. garvieae N201 impaired the capacity of both strains of S. aureus to deal with the presence of H 2 O 2 which led to growth deficiency [2,16,17].
In order to investigate the interaction between S. aureus and L. garvieae in oxidative stress-inducing culture conditions, genes involved in H 2 O 2 metabolism were first choice targets, whose expression has been monitored by Delpech et al. [17,20] in co-cultures of L. garvieae N201 and S. aureus SA15 or MW2.
In S. aureus, Catalase (KatA) and Alkyl hydroxyperoxidase (Ahp) play a role in H 2 O 2 degradation [29,30]. Ahp leads to a dual function in oxidative-stress resistance, environmental persistence and host-pathogen interaction [29]. Amongst the targeted genes, dnaK is known to be involved in H 2 O 2 -resistance [31,32], clpC has an important role in oxidative stress regulation, and ctsR is a transcriptional repressor of stress-genes [33]. Moreover, the 2 latter genes, belonging to the dcw cluster involved in cellular division, might be modulated by H 2 O 2 -stress, leading to S. aureus growth impairment [34]. So, if LAB can modify this genes cluster expression, it could have an inhibiting effect on cellular division.
In addition to S. aureus growth modulation, L. garvieae may also have an effect on its virulence. Enterotoxins are the main toxins responsible for S. aureus food poisoning. Ninety-four per cent of S. aureus isolated from cow milk have at least one enterotoxin-encoding gene [35]. Amongst the different enterotoxins, enterotoxin C, encoded by sec4 gene, is the most frequently involved in food poisoning [36][37][38][39]. Cretenet et al. [40] showed that this virulence-related gene expression can be modified by L. lactis in cheese. S. aureus virulence is under control of the agr system which is involved in regulating many stress response and virulence genes [41][42][43]. agrA is a response regulator [44] and is able to induce hld, a δ-lysin gene [43], and enterotoxin C encoding gene sec4 [45]. The agr system itself is controlled by several molecular intermediates such as SaeRS and SrrAB. SaeRS is a two-component system involved in response to environmental stress which could inhibit agr system [46] and also control virulence genes [47][48][49][50]. SrrAB is also a two-component system activated in an anaerobic environment [51]. It would be involved in virulence gene control [52] in response to H 2 O 2 . Indeed, in the presence of H 2 O 2 , srrA is repressed in S. aureus [53]. Under anaerobic conditions, SrrA represses agrA and hld expression [54,55]. CodY, a regulatory protein involved in repressing virulence gene expression in S. aureus, is also involved in controlling agr system and virulence genes [56].
The level of aeration can change the level of H 2 O 2 -production according to the LAB strain. Indeed, it has been demonstrated that the transcriptomes of L. lactis and L. garvieae are significantly modified by the aeration level [20,57]. Lactobacillus crispatus can produce H 2 O 2 in high aeration conditions but not in static conditions [58]. Contrariwise, Lactobacillus delbrueckii subsp. bulgaricus can also produce H 2 O 2 in static conditions, even though the amount produced is lower than in high aeration conditions [59].
The aim of the present study was to compare the transcriptional response involved in the antagonistic interaction between S. aureus and different strains of L. garvieae. In addition to N201 strain isolated from raw milk Saint-Nectaire cheese and already well described [2,16], two other L. garvieae strains were selected for their different capacities to produce H 2 O 2 : a H 2 O 2 -producing strain, CL-1183 (VIVASET, Veterinary Faculty, Complutense University from Madrid) isolated in Brazil from the milk from buffalo cows affected by subclinical mastitis [5] as well as a H 2 O 2 -non producing strain, Lg2, a fish pathogenic strain isolated in Japan [60]. The expression of S. aureus genes related to oxygen metabolism, response to stress, cell division and virulence was measured as well as the expression of L. garvieae genes related to oxygen metabolism.

Results
In order to investigate the interaction between S. aureus and L. garvieae in oxidative stress-inducing culture conditions, we followed the growth of the strains and measured the amount of H 2 O 2 in two culture conditions i.e. under high and low aeration levels. The genes whose expression has been monitored by Delpech et al. [17,20] were chosen to compare the transcriptional response involved in the antagonistic interaction between S. aureus and different strains of L. garvieae. The expression of S. aureus genes related to oxygen metabolism, response to stress, cell division and virulence was measured as well as the expression of L. garvieae genes related to oxygen metabolism ( Table 1).
Ability of L. garvieae strains to produce H 2 O 2 and to inhibit S. aureus growth The effect of aeration level on microbial growth was tested by determining cellular concentrations of both microorganisms. In pure culture, S. aureus growth was almost identical under both high and low aeration conditions. However, the population count reached at 24 h was 1.2 log CFU/ml lower under low aeration than under high aeration (Fig. 1a). The 3 different strains of L. garvieae grew as well under high aeration level as under low aeration level independent of the presence of S. aureus (Additional file 1). pH values remained stable (7.0 ± 0.2) in all cultures over the whole experiment (data not shown).
Detectable amounts of H 2 O 2 were produced exclusively under high aeration level by N201 and CL-1183  Table 2). Under high aeration, when co-cultivated with N201, SA15 growth was 2.1 log [CFU.ml − 1 ] lower than in pure culture as early as 6 h with a maximal growth inhibition of 5.8 log [CFU.ml − 1 ] at 24 h. SA15 growth co-cultivated with CL-1183 was 1.7 log [CFU.ml − 1 ] lower compared to pure culture as early as 6 h. The maximal growth inhibition was observed after 24 h of co-culture (4.5 log [CFU.ml − 1 ] lower). SA15 co-cultivated with Lg2 did not show any significant impairment in growth throughout the experiment compared to pure culture (Fig. 1a).
To determine the involvement of H 2 O 2 produced by N201 and CL-1183 in growth inhibition of SA15, co-cultures of SA15 with these 2 strains were performed under high aeration level in presence of catalase ( Fig. 1). No detectable amount of H 2 O 2 was observed at any time. S. aureus cell concentration was not significantly different in presence of catalase compared to control ( Fig. 1) during both exponential and stationary phases. Statistical analysis, using ANOVA, showed that in presence of catalase there was no significant difference of SA15 count after 24 h in co-culture with CL-1183 compared to pure culture. In contrary, SA15 cell concentration in co-culture SA15/N201 was still 0.8 log [CFU.ml − 1 ] lower than in pure culture. These results showed that the inhibition of SA15 growth by N201 and CL-1183 was partially or completely suppressed by catalase.

Effect of L. garvieae strains on S. aureus genes expression under high aeration level (Additional file 2)
Whatever the L. garvieae strain cultivated with SA15, none of the H 2 O 2 -response genes tested was up-regulated. When SA15 was cultivated with L. garvieae N201 or CL-1183, katA was down-regulated, at 9 h and 24 h with N201 (3.7 and 3.0-times, respectively), or at 6 and 9 h with CL-1183 (3.9 and 3.2-times, respectively). When cultivated with N201, ahpF expression in SA15 was not modified in comparison with pure culture, whereas in co-cultures SA15/CL-1183 and SA15/Lg2, it was down-regulated (2.1-times at 6 h and 2.9-times at 24 h, 6.9-times at 24 h, respectively). sodA was down-regulated at 6, 9 and 24 h in co-cultures with N201 (2.2, 4.4 and 3.9-times, respectively) or at 24 h (2.2-times) with CL-1183. No change in expression of katA and sodA was observed with Lg2.
L. garvieae N201 induced an up-regulation of the three SA15 stress-response genes tested at 9 h; clpC, ctsR and dnaK were 2.0-times, 3.7-times and 2.5-times more expressed than in pure culture, respectively. When cultivated with CL-1183, only dnaK was 2.8-times up-regulated at 6 h, while clpC and ctsR expressions were not affected. Conversely, in co-culture SA15/Lg2, clpC and dnaK were 3.9-times and 19.8-times down-regulated respectively, while ctsR did not show any significant difference in its expression.
In co-culture, the five SA15 virulence-related regulator genes tested displayed contrasted patterns of expression. At 9 h, in both SA15/N201 and SA15/CL-1183 co-cultures, agrA was up-regulated (4.0-times with N201 and 3.4-times with CL-1183), while srrA was down-regulated (7.1-times or 6.4-times with N201 or CL-1183 respectively). codY expression was down-regulated 4.1-times at 24 h by N201 and 3.6-times at 6 h by CL-1183. saeS expression was 2.4-times lower at 24 h in co-culture with N201 than in pure culture, whereas it was not affected by CL-1183. hld expression remained stable until 24 h when SA15 was in co-culture with N201 or CL-1183. In co-culture SA15/Lg2, at 24 h, agrA, codY and hld expressions were all down-regulated (3.0-times, 2.6-times and 3.9-times, respectively), while saeS and srrA expressions remained stable.
In co-cultures SA15/N201 and SA15/CL-1183, amongst the 2 enterotoxins-encoding genes tested, only sec4 expression was modified at 24 h, with a 4.1-times and a 9.0-times down-regulation, respectively. sel2 expression was not modified whatever the SA15 culture conditions were. In co-culture SA15/Lg2, no enterotoxin gene expression was affected. No change in stress-response genes expression was observed when SA15 was cultivated with N201 or CL-1183. Concerning the cell division gene mraW, its expression was not modified, whatever the strain cultivated with SA15. Amongst virulence-related regulator genes, agrA was up-regulated in all three co-cultures at different stages of growth. It was the only virulence-related regulator gene differentially regulated by N201 (6.0-times up-regulated at 9 h). In the same way, this gene was 4.1-times up-regulated at 9 h by CL-1183, and 2.5-times up-regulated at 24 h by Lg2. codY was 2.0-times down-regulated at 6 h but 2.0-times up-regulated at 24 h with CL-1183, while codY and srrA were 2.1-and 2.7-times down-regulated at 9 and 24 h, respectively with Lg2.

Effect of aeration on L. garvieae strains (Additional file 4)
To evaluate the strain-specific effect of aeration on L. garvieae, expression of 5 H 2 O 2 -related genes of them (trxB1, ahpC, gpx, poxB and sodA) was monitored and compared between the 3 L. garvieae strains in co-culture with SA15. Overall, most differential expressions concerned H 2 O 2 -degradation genes. Differences between H 2 O 2 -producing and non-producing strains were essentially related to ahpC: both N201 and CL-1183 overexpressed this gene under low aeration (5.6-times at 24 h in N201, 4.0-times at 6 h and 6.5-times at 9 h in CL-1183). Conversely, Lg2 overexpressed ahpC (4.7-times) at 9 h under high level of aeration. Both CL-1183 and Lg2 overexpressed trxB1 under high level of aeration (2.9-times at 6 h in CL-1183 and 2.1-times 24 h in Lg2). Amongst H 2 O 2 -synthesis genes, only CL-1183 overexpressed poxB (4.1-times at 9 h under low aeration).

Discussion
The aim of the study was to compare the transcriptional response involved in the antagonistic interaction between S. aureus and different strains of L. garvieae. Three strains of L. garvieae from different origins were used: N201, a dairy-isolated strain known to be a S. aureus inhibiting strain; CL-1183, a strain isolated from milk from buffalo cows suffering from mastitis; and Lg2, a fish pathogenic strain. Under high aeration condition, two of them were shown to be able to produce detectable amount of H 2 O 2 (N201 and CL-1183), whereas the third one (Lg2) was not.  [29]. This could be correlated with the large amounts of H 2 O 2 detected in the co-cultures SA15/N201. The down-regulation of katA, at 9 h and 24 h, can be explained by the fact that srrA was also down regulated. Mashruwala and Boyd [61] found that srrAB mutant strain had decreased transcription of genes encoding for H 2 O 2 resistance factors as katA. They also reported that SrrAB positively influenced H 2 O 2 resistance during periods of high O 2 dependent respiratory activity, but not when cellular respiration was diminished as a result of lower O 2 availability. S. aureus SA15 genes involved in stress-response (clpC, ctsR, dnaK) were all up-regulated by N201 and only dnaK was up-regulated by CL-1183, whereas clpC and dnaK were down-regulated by Lg2. These results indicated that in presence N201 and CL-1183, SA15 detected an oxidative stress as a result of H 2 O 2 production by these two L. garvieae strains. However, the up-regulation of these genes did not allow SA15 to fight against this stress, although precedent studies reported that S. aureus clpC or dnaK mutants growth was impaired in the presence of H 2 O 2 stress [32,62,63] and that a basic level of expression of dnaK was sufficient in response to this stress [32]. All these data highlighted the complexity of the stress response machinery and the important role of clpC, ctsR and dnaK genes. Moreover, while mraW gene was not differentially expressed in presence of Lg2, it was repressed at 6 and 9 h by N201, confirming results obtained by Delpech et al. [17], and by CL-1183 in our study, when S. aureus growth was inhibited. The repression of mraW by N201 and CL-1183 was positively correlated with inhibition of SA15 growth. These results are supported by Cretenet et al. [40] showing that L. lactis was able to inhibit ftsH, ftsL and ftsZ genes also involved in cellular division. Interestingly, several studies reported that inhibition of cell division protein is a promising approach for anti-staphylococcal therapy [64,65].
Catalase treatment partially reduced SA15 growth inhibition by N201, confirming the result obtained by Delbes-Paus et al. [2], whereas the inhibition by CL-1183 was completely suppressed. Oogai et al. [66] observed that in presence of catalase, the S. aureus MW2 growth inhibition by Streptococcus sanguinis was completely suppressed. These observations confirmed the role of H 2 O 2 produced by LAB in growth inhibition of S. aureus. Our study demonstrated that under high level of aeration, the inhibition of S. aureus by L. garvieae involved strain specific mechanisms. Indeed, CL-1183 inhibited SA15 growth mainly by H 2 O 2 production whereas inhibition due to N201 may involve the combined action of H 2 O 2 and other antagonistic mechanism.
Under low aeration condition, S. aureus growth was almost identical to that measured under high aeration condition, except after 24 h, when we observed a 1,2 log reduction in the S. aureus population level that could be due to the depletion of oxygen. Ledala et al. [67] showed that the growth rate of S. aureus was independent of oxygen limitation over 12 h although its metabolome was significantly affected. S. aureus growth was slightly inhibited by the three L. garvieae strains, although none of them could produce detectable amounts of H 2 O 2 (Fig. 1). This suggests that there was another inhibitory mechanism involved. These results match those obtained with N201 in the previous studies [2,17,20]. Delbes-Paus et al. [2] have shown that in milk, even if H 2 O 2 was not detected, S. aureus growth was inhibited by L. garvieae N201. However, no clear hypothesis to explain the mechanism of the anti-staphylococcal activity under low aeration could be drawn from the gene expression data. However, the up-regulation of stress-response genes in the presence of Lg2 was observed. Indeed, clpC, ctsR and dnaK genes, were all up-regulated at 6 h in co-culture with Lg2, suggesting that Lg2 triggered a stress on SA15 which led to a growth inhibition.

L. garvieae strains and aeration level effect on S. aureus SA15 virulence gene expression
Under high level of aeration, L. garvieae N201 reduced S. aureus virulence-related genes expression confirming results obtained by Delpech et al. [17]. Cretenet et al. [40] and Queck et al. [68] have shown that the agr system, involved in the regulation of genes linked to S. aureus virulence, was repressed by L. lactis even if L. lactis does not produce H 2 O 2 in this condition. Molecular intermediaries SaeRS and SrrAB are involved in controlling the agr system [46]. We showed srrA expression was repressed in co-culture with N201 and CL-1183, when H 2 O 2 was produced as already observed by Chang et al. [53]. Moreover, Majerczyk et al. [56] have shown that a codY mutant could derepressed agr system. In our conditions, codY was repressed by CL-1183 at 6 h explaining agrA up-regulation at 9 h, whereas repression of codY at 24 h by Lg2 cannot explain agrA down-regulation suggesting that another regulator might be involved. Moreover, at 24 h, despite SA15 growth was not modified, Lg2 caused a down-regulation of three virulence-related regulator genes tested in this study (agrA, codY, hld). sec4 and sel2 are two enteroxin-encoding genes found in S. aureus SA15 [17]. Although enteroxin C-encoding gene sec4 is under the control of the agr system [45], agrA expression was only induced by N201 and CL-1183 and sec4 was repressed under high aeration level as shown by Delpech et al. [17]. Our data showed the repression of virulence associated genes and of enterotoxin encoding genes as well as simultaneous S. aureus growth inhibition in the presence of N201 or CL-1183, indicating that these 2 strains not only inhibited SA15 growth but also potentially attenuated its virulence.
Under low aeration condition, S. aureus over-expressed agrA and enterotoxin-encoding genes sel2 and sec4 in presence of N201 and CL-1183, whereas Lg2 did not induce any modification. In presence of Lg2, agrA was up-regulated at 24 h. This result can be explained by the fact that SA15 srrA gene was down-regulated, at the same time. The same observation was reported by Yarwood et al. [55]. They showed that transcription of RNAIII from the agr locus was inversely dependent on expression of srrAB. Inherently, despite SA15 agrA up-regulation by Lg2, no modification in enterotoxin genes expression was observed. Yarwood et al. [55] also reported that the SA15 srrB mutant growth was significantly slower in anaerobic condition, which is in accordance with SA15 growth inhibition in presence of Lg2.
L. garvieae strain responses to low and high levels of aeration in co-culture with S. aureus SA15 It seemed that the three strains of L. garvieae differed essentially in their ability to degrade H 2 O 2 . Presence of H 2 O 2 would not depend only on H 2 O 2 -synthesis, but also on H 2 O 2 -degradation by L. garvieae strains, which is an original mechanism, and in accordance with the results obtained by Delpech et al. [20].
In high aeration level, Lg2 overexpressed two H 2 O 2 degradation genes ahpC and trxB at 9 and 24 h respectively. The non-detection of H 2 O 2 could be explained by the absence of production of H 2 O 2 or by its degradation by AhpC and TrxB. The latter mechanism could explain why, under low aeration, despite CL-1183 overexpressed an H 2 O 2 -synthesis gene, poxB, no detectable H 2 O 2 was observed. Indeed, at the same time, this strain overexpressed the H 2 O 2 degradation gene ahpC. Our data showed that the importance of the inhibition varies depending on the presence or absence of H 2 O 2 . Indeed, inhibition can occur while H 2 O 2 is not detected. As regards other potential explanations for the growth inhibition, the only putative bacteriocin identified in the L. garvieae N201 genome was homologous to garvieaecin Q (GarQ, data not shown), a class IId bacteriocin [69]. In view of the results of previous experiments on L. garvieae N201 [20], the inhibition was probably caused neither by garvieaecin Q nor by a protein, nor by a lipid, and nor by a polysaccharide. In the present study, we evidenced L. garvieae strain-specific response in S. aureus gene expression which suggests that molecular mechanisms involved in the inhibition could differ between L. garvieae strains. Further investigations are needed to elucidate these mechanisms. A detailed analysis of strains physiology is needed to evaluate the potential role of nutritional competition in the inhibition. A global transcriptomic approach of the interaction would shed some light on the metabolic dialogue between these strains, and more particularly the potential involvement of quorum-sensing.

Conclusions
Our data evidenced an impact of the aeration condition on the interaction between SA15 and L. garvieae strains.
Under high level of aeration, only N201 and CL-1183 produced detectable amounts of H 2 O 2 and consequently inhibited SA15 growth. Both L. garvieae strains activated SA15 stress-response genes and inhibited the expression of several genes involved in H 2 O 2 -degradation, virulence and cellular division. A catalase treatment partially or completely suppressed this inhibition and no H 2 O 2 was detected. Lg2 strain did not produce any detectable amount of H 2 O 2 and did not inhibit SA15 growth. However, as N201 and CL-1183, Lg2 inhibited SA15 virulence-related genes and stress-response related genes.
Under low aeration level, none of the L. garvieae strains produced any detectable amount of H 2 O 2 . N201 and CL-1183 were still able to inhibit SA15 growth, as well as, Lg2. While N201 and CL-1183 induced the up-regulation of the SA15 enterotoxin-encoding genes and related regulator agrA, Lg2 caused up-regulation of SA15 stress-related genes and down-regulated virulence-related regulators genes.
The antagonistic properties of L. garvieae against S. aureus were strain-and aeration level-dependent. Under high aeration, they involved H 2 O 2 production by two out of the three L. garvieae tested strains. This study provides new insights into microbial interactions mechanisms and shows the importance of investigating strain-specific effects.

Strains and culture conditions
Strains Lactococcus garvieae N201, CL-1183 and Lg2 and Staphylococcus aureus SA15 were cultivated in Brain-Heart Infusion buffered to pH 7 using Potassium Phosphate (BHI, Biokar Diagnostic, Pantin, France) at 30°C and 37°C, respectively, under static condition. After 20 h, cells were harvested at 3500 g during 15 min. Cell pellets were resuspended in 2 ml of BHI and syringed 5 times. Cell concentration was then determined using a Petroff-Hausser counting chamber. Fifty milliliters of buffered BHI, containing or not 4000 IU ml − 1 of catalase from bovine liver (ref. C100, Sigma), were then co-inoculated at~10 7 cells per ml for L. garvieae and~10 6 cells per ml for S. aureus, or just with L. garvieae or S. aureus at the same concentration in 250 ml Erlenmeyer flasks or in 50 ml Falcon tubes, as described previously [2,17].
Then, Erlenmeyer flasks were incubated for 6, 9 or 24 h at 30°C under shaking condition (150 rpm), corresponding to the high level of aeration condition. The low level of aeration condition corresponded to static fully filled 50 ml Falcon tubes. At 0, 6, 9 and 24 h, 1 ml of the cultures was removed to be syringed and serially diluted in Ringer's solution. After adequate mixing, 100 μl of each dilution were plated onto solid Baird Parker for S. aureus and BHI agar for L. garvieae for numeration. Colony-forming units were counted after overnight incubation at 37°C for S. aureus and 30°C for L. garvieae. Also, 1 ml was removed from the cultures for H 2 O 2 content analysis and 40 ml for RNA extraction. Cell pellets from 6, 9 and 24 h cultures were resuspended in 200 μl of cold Tris-EDTA buffer and then frozen at − 80°C. The experiment was carried out in triplicate within each of three independent biological replicates, for a total of 9 samples per time point. pH measurements were made with the rest of the cultures.

Quantitative analysis of H 2 O 2 in L. garvieae cultures supernatant
Hydrogen peroxide concentration was determined according to Batdorj et al. protocol [70] with slight modifications. Cells were harvested by centrifugation at 11500 g, 15°C for 10 min. One hundred microliter of the sample supernatant were mixed with 100 μl of 4-aminoantipyrine 4 mg.ml − 1 (Sigma-Aldrich, St. Louis, Missouri, USA), 20 μl of water-saturated phenol (Sigma-Aldrich, St. Louis, Missouri, USA), 750 μl of phosphate buffer Na 2 HPO 4 / NaH 2 PO 4 0.1 M (pH 7) and 30 μl of horseradish peroxidase type VI-A (500 U.ml − 1 in sodium phosphate buffer pH 6 (Sigma-Aldrich, St. Louis, Missouri, USA). Sample was mixed by inverting the tube and OD was measured at 505 nm. Blank was done by replacing sample by sterile medium. H 2 O 2 concentration was determined using a standard curve performed with concentrations ranging from 0 to 3 mM. The minimal concentration that could be detected was 0.5 mM.

RNA extraction and DNase treatments
Cells in Tris-EDTA buffer were thawed out and 25 μl of 20% SDS, 500 μl of phenol (pH4), 3.5μl of β-mercaptoethanol and 600 mg of Zirconium beads were added. The cells were broken twice in a tissue homogenizer (Precellys® 24, Bertin Technologies, Montigny-le-Bretonneux, France). Then, 200 μl of chloroform were added and mixed with the solution and sample was centrifuged at 11400 g for 20 min at 4°C. RNA extraction was performed using NucleoSpin RNA Midi kit (Macherey-Nagel, GmbH & Co. KG, Düren, Germany) following the supplier's instructions. The RNA extracts were treated twice with DNAse I using an Ambion DNA-free kit following the supplier's instructions (Ambion, Inc., Austin, Texas, USA). The RNA extracts were quantified using a Nanodrop™ 2000C (Thermo Fisher Scientific Inc., Waltham, Massachusetts, USA).

Reverse transcription and quantitative PCR
Reverse Transcription (RT) was performed using 0.5 μg of RNA twice treated with DNase, 5 μl of 10X buffer, 2 μl of 25X dNTP, 5 μl of random primers and 2.5 μl of retrotranscriptase (Applied Biosystems®, Life Technologies, Foster City, California, USA). The RT was done in a thermocycler (Techne® Prime, Bibby Scientific, Stone, Staffordshire, UK) with the following parameters: 10 min at 25°C and 120 min at 37°C.
Genes Of Interest (GOI) Ct were determined in quantitative PCR (qPCR) assays using 2.5 μl of cDNA suspensions 10-fold diluted in RNAse-free water and 10 μl of qPCR mix containing 1.25 μl of each primer (10 mM, Tables 3 and 4), 6.25 μl of qPCR Rotor Gene SybrGreen mix (Qiagen, Hilden, Germany) and 1.25 μl of RNAse-free water. The qPCR was performed according to the protocol on Rotor Gene Q (Qiagen, Hilden, Germany) with the following parameters: 4 min at 94°C, then for 35 cycles, 30 s at 94°C, 30 s at 55°C, 60 s at 72°C. Each reaction was performed in triplicate within each of three independent biological replicates, for a total of 9 reactions per time point.
Primers efficiencies were determined according to a 10-fold template dilution standard and were all ranging from 1.90 (~95%) to 2.293 (~115%) according to the following equation: Efficiency = 10^(− 1/slope of calibration curve) [71]. The primers used were previously described by others [17]. One reference gene was used for L. garvieae data: tufB encoding the elongation factor Tu. Considering that the lag in growth phase reflects the inhibition of S. aureus by L. garvieae, we evaluated gene expression at identical time points corresponding to different population levels. Two reference genes whose expression was stable over recA a Recombinase A Housekeeping gene [74] agrA Accessory gene regulator A Virulence-related regulator [40] ahpF Alkyl hydroperoxidase F H 2 O 2 -response [17] clpC Clp proteinase C Stress-response [40] ctsR Transcriptional repressor of stress-genes Stress-response [40] codY Transcriptional repressor Virulence-related regulator [17] dnaK Chaperone protein DnaK Stress-response [40] hld Deltahemolysin Virulence-related regulator [40] katA Catalase H 2 O 2 -response [40] mraW S-adenosyl methyltransferase Cell division [17] saeS Histidine protein kinase Virulence-related regulator [40] sec4 Enterotoxin C Enterotoxin [40] sel2 Enterotoxin L Enterotoxin [40] sodA Superoxide dismutase H 2 O 2 -response [40] srrA Staphylococcal respiratory response Virulence-related regulator [40] a Reference genes time under the different conditions were used for S. aureus data: housekeeping genes recA, encoding the recombinase A, and hu, encoding a DNA-binding protein. The GOI expression was calculated according to the formula introduced by Hellemans et al. [72].

Comparison of gene expression
Differential gene expression was evaluated by comparing expression between two conditions. First, the influence of aeration on gene expression in pure culture was evaluated for S. aureus. Then, the influence of different strains of L. garvieae (N201, CL-1183 and Lg2) on S. aureus SA15 gene expression was observed under high or low aeration condition. Secondly, L. garvieae gene expression was compared between the different strains in co-culture with S. aureus SA15 under both aeration conditions. A gene was considered as differentially expressed when its expression was changed by at least a factor of 2.

Statistical analyses
Statistical analyses on microbial counts were performed using R software [73] by one-way analysis of variance (ANOVA) followed by Least Significant Difference (LSD) test. The gene expression statistical analyses were performed using R by ANOVA with Newmann-Keuls post-hoc test. Significance was declared at P < 0.05.