The bacterial strains studied are listed in Table 1.

A collection of 23 clinical isolates of H. kunzii collected from DFU in a multicentre study performed between February 2008 and August 2013 was used (Vergne et al., 2015). Moreover, to assess the co-infection between H. kunzii and S. aureus, in addition to the reference S. aureus strain Newman, two clinical S. aureus strains, both isolated and characterized in our hospital, were used: NSA1385 (a colonizing strain collected from uninfected ulcer) and NSA739 (an infecting strain collected from deep DFI; Sotto et al., 2008; Messad et al., 2015). Escherichia coli OP50 was used as control for nematodes. This bacterium is the standard feeding strain for Fer-15 nematodes. It harbors no known uropathogenic virulence factors.

The different bacteria were grown in Mueller-Hinton (MH) and Luria Bertani (LB) broth or agar at 37°C except H. kunzii strains which were grown on Columbia agar supplemented with 5% fresh sheep blood (bioMérieux, France) under a 5% CO₂ atmosphere at 37°C during 48 h.

S. aureus, H. kunzii and E. coli grew identically alone or in association on the nematode growth medium (NGM) used to worms experiments at 37°C (Figure S1).

PFGE analysis of genomic DNA fragments of the 23 clinical isolates of H. kunzii was carried out after digestion with the restriction endonuclease SmaI, as previously published for enterococci (Bourdon et al., 2011). The electrophoresis was performed using a CHEF-DRIII apparatus (BioRad, France) and PFGE patterns were interpreted according to well-established criteria (Tenover et al., 1995).

The nematode infection assay was carried out as previously described using the Fer-15 mutant line (a temperature sensitive fertility defect; Lavigne et al., 2008). Overnight cultures of the studied bacterial strains in the NGM were harvested, centrifuged and suspended in phosphate buffered saline solution (PBS) at a concentration of 10⁵ CFU/mL. Ten microliters of these bacterial suspensions were inoculated on NGM agar plates. A ratio 1:1 (10⁵/10⁵) was prepared during co-infection assays. To validate this ratio, we evaluated the CFU of both bacteria prior to the seeding onto agar to confirm that S. aureus and H. kunzii were present in an equal amount. The plates were incubated at 37°C for 8−10 h. Around 30 L4 stage nematodes per plate were thus seeded and then incubated at 25°C. An independent reader (blind to the culture) scored each day the number of live nematode under a stereomicroscope (Leica, France).

C. elegans were infected with two representative H. kunzii strains (H10 with no virulence and H13 with a low virulence). After 12 h, 30 nematodes were transferred to NGM medium containing the different S. aureus strains. In the same way, to evaluate the bacterial persistence, nematodes were coinfected with H. kunzii and S. aureus. After 12 h, 30 nematodes were transferred to NGM medium containing OP50 strain. The conditions of nematodes preparation were strictly similar to previous assays described before. Final analysis established the Lethal time 50% (LT50), which corresponds to time (in days) required to kill 50% of the worms.

Three replicates repeated five times were performed for each studied strain.

Firstly, all the studied bacterial strains were grown in LB broth media (with or without anaerobes conditions) at 25°C during 16 h. The cultures were then spotted on NGM plates. Around 30 L4 stage nematodes were deposed in the center of the bacteria lawn. To establish the occupancy assays, the number of nematodes inside or outside each lawn was counted after overnight incubation as previously published (Lavigne et al., 2013). The results were presented in percent occupancy (number of worms in the bacterial lawn on the total number of C. elegans). The experiments were performed in triplicate.

Secondly, we determined the number of bacteria within the nematode gut (Garsin et al., 2001; Lavigne et al., 2008). Briefly, nematodes were picked at 72 h, and the surface bacteria were removed by washing the nematodes twice in M9 medium containing 25 μg/ml gentamicin. The C. elegans were then mechanically disrupted in M9 medium containing 1% Triton X-100. Finally, after serial dilutions, 100 μl of the mixture were plated on LB agar medium and on Columbia agar supplemented with 5% fresh sheep blood (when the co-infection experiments were performed). The colonies [aspect and haemolytic activity (α haemolytic for H. kunzii and β haemolytic for S. aureus)] were counted after 24 h and the identification of each species was confirmed by MALDI-TOF (Vitek MS, BioMérieux). Three replicate assays were performed for each strain.

Analysis of the mRNA levels of spa, hla, and agr was performed following the method previously published (Doumith et al., 2009; Kriegeskorte et al., 2014). These 3 genes are essential in the S. aureus pathogenicity notably in worms model (Sifri et al., 2003, 2006). S. aureus isolates and H. kunzii H13 were grown alone or in association in MH broth to an OD₆₀₀ of ~0.7. The total RNA extraction was performed using the RNeasy Mini kit (Qiagen, France) during exponential stages. Purity and concentration were determined by the NanoDrop 2000 spectrophotometer (Fisher Scientific, USA). The iScript^TM Select cDNA Synthesis Kit (Biorad, Hurcules, CA) was used to the synthesis of cDNA from 1 μg of total RNA for each sample. Real-time PCR were done in a LightCycler® 480 (Roche, France) using the LightCycler FastStart DNA Master^PLUS SYBRGreen I kit, 100 ng of cDNA and 10 pmol of target primers (Table 4). Amplifications were analyzed in triplicate from three different RNA preparations. Cycle threshold (Ct) values of the different target genes were compared with the Ct-values of the house-keeping gene (gyrB) (Sihto et al., 2014). The Newman strain was used as control. The normalized relative expressions of the studied genes were obtained for each strain following the equation: 2^−ΔΔCt (ΔΔCt = (Ct_gene–Ct_gyrB)_{studied strain}—(Ct_gene–Ct_gyrB)_{control strain}) (Livak and Schmittgen, 2001). Results obtained for each gene were log-transformed to obtain a fold change difference between strains.

For selected genes involved in C. elegans response against infection [hlh-30, lys-5, lgg-1, clec-7 (Visvikis et al., 2014), cyp-37B1 (Irazoqui et al., 2010), sodh-1 (JebaMercy and Balamurugan, 2012)], transcript level analysis was performed by qRT-PCR following the same protocol described before. C. elegans were infected between 12 h with studied bacterial strains. The nematodes were then washed twice in water. Total RNA from animals was extracted by using TRIzol® RNA Isolation Reagents (ThermoFisher, France). The target primers used were presented in Table 4. The 2^−ΔΔCT method was used to analyze transcriptional changes in target genes using snb-1 as the housekeeping control gene (Livak and Schmittgen, 2001). Data analysis was performed with the Pfaffl method (Pfaffl, 2001). Error bars indicate the standard deviation (SD) of three independent experiments.

Statistics and graphs were performed using GraphPad Prism 6.0 software.

For the nematode killing experiments, a log-rank (Mantel-cox) test was used to evaluate differences in survival rates between the different strains.

Log-transformed data were used for RT-PCR. The effects of bacterial infections on expression of selected genes involved in S. aureus virulence and in host response were performed using one-way ANOVA followed by Dunnett's multiple comparisons test. A statistically significant difference was retained for P < 0.01.

The virulence of a clinical panel of 23 H. kunzii isolates was evaluated in a nematode model. The genetic comparison of the 23 strains showed that the isolates were not clonally related (Figure S2) eliminating a clonal impact of the virulence behavior. Out of 23, 17 (74%) H. kunzii strains were non-virulent with a behavior similar to the non-pathogenic E. coli OP50 (LT50s: 6.1 days ± 0.4 vs. 7.1 days ± 0.5, respectively), a laboratory reference strain used to feed nematodes [P-value, non significant (NS)]. The other six H. kunzii strains (H7, H10, H17a, H20, H21, H22b) were significantly more virulent than OP50 (P < 0.001; Table 1, Table S1).

To compare the virulence between H. kunzii and S. aureus, we used well-characterized S. aureus strains: the S. aureus strain NSA739 (collected from DFI and harboring a high virulence potential), the S. aureus strain NSA1385 (collected from DFU and harboring a low virulence potential) and the S. aureus reference strain Newman (Sotto et al., 2008; Messad et al., 2015). We observed that the panel of H. kunzii presented significantly lower virulence than NSA739 (LT50: 2.8 days ± 0.4), NSA1385 (LT50: 4.7 ± 0.2) and Newman (LT50: 4.3 ± 0.3; P < 0.001), respectively. The difference of virulence between the infecting and the colonizing strains of S. aureus was previously demonstrated (Table 1, Table S1; Messad et al., 2015). These results confirmed that H. kunzii strains are low- or non-virulent bacteria.

When the different H. kunzii strains and NSA739 were used to co-infect C. elegans, an important attenuation of the S. aureus virulence was observed independently of H. kunzii virulence or non-virulence potential (Table 2 Table S2, Figure 1). The LT50 obtained with the strains coinfection varied between 4.1 and 5.7 days. They were significantly longer than the LT50 detected with S. aureus strain alone (LT50: 2.8 days; P < 0.001).

To confirm the role of H. kunzii in the diminution of S. aureus virulence, we tested the effect of H. kunzii on two others S. aureus strains: NSA1385 (colonizing strain) and reference strain Newman and on the E. coli OP50. If the majority of H. kunzii strains had statistically no impact on S. aureus virulence, 8 strains (H4, H6, H8, H13, H16, H17b, H22b, H23) reduced significantly the virulence of NSA1385 (LT50: 5.6–6.3 days; P < 0.001; Table 3, Table S3). On the other hand, we observed that 6 H. kunzii strains (H9, H13, H18, H22a, H22b, H23) reduced the virulence of the reference strain Newman (LT50: 5.8–7.1 days; P < 0.001; Table 3). Interestingly no difference could be noted when worms were fed with H. kunzii + OP50 (Table 3), suggesting a specific effect between S. aureus and H. kunzii.

All these findings strongly suggest the important role of particular H. kunzii strains in the attenuation of S. aureus virulence isolated from wounds notably for highly virulent S. aureus strains.

To exclude the possibility that the observed results in worms was due to a modification of their feeding behavior, an occupancy test was performed. None of the bacterial strains tested alone or in association presented strong avoidance behavior. No significant difference was noted in the fraction of nematodes on the bacterial lawn between the different associations studied (Tables 1–3, Tables S1–S3). We also measured the bacterial load in the intestine of nematodes at 72 h post infection (Lavigne et al., 2013). We found that all bacteria tested alone or in association can colonize and survive in the C. elegans intestine (Figure 2). The number of the H. kunzii and S. aureus CFU was around 4 × 10⁵ bacteria per nematode (IC95% = 2.8–7.9 × 10⁵) within the nematode intestine for each combination without statistical difference (P-value, NS).

These results confirm the low virulence of H. kunzii strains and suggest that the impact on the modulation of S. aureus virulence observed in C. elegans was not due to a modification of nematodes' feeding behavior, nor to a reduction of S. aureus infection rate or a hypothetical cytotoxicity effect of H. kunzii on C. elegans.

To estimate the host response during co-infection between H. kunzii and S. aureus, we carried out qRT-PCRs on six representative host defense genes of nematodes after infection by H. kunzii and S. aureus alone or in co-infection: hlh-30 (the key transcriptional factor-encoded gene for S. aureus host defense), lys-5 and clec-7 (antimicrobial-encoded genes), cyp-37B1 and sodh-1 (cytoprotective and detoxification-encoded genes) and lgg-1 (autophagy-encoded gene). Of the different co-infection combinations we choose to study one H. kunzii isolate non-virulent in nematode model and reducing the virulence of all S. aureus strains (H13) and one with low virulence in nematode model and that had effect exclusively on NSA739 virulence (H10).

We found that nematodes fed with the two H. kunzii strains did not show significant differences of expression of host defense genes compared to nematodes fed with the non-pathogenic strain OP50. This result confirms that H. kunzii strains do not stimulate the C. elegans immune response (Figure 3). On the other hand, when nematodes were fed with the three S. aureus strains, they significantly overexpressed hlh-30, lys-5, sodh-1, and cyp-37B1 compared to nematodes fed with OP50 (P < 0.01). Only the autophagy gene lgg-1 had no modification of expression whatever the strain and the condition tested (P-value, NS). Interestingly, no significant difference in the expression of host response genes could be observed between each S. aureus strains tested (colonizing or infecting; Figure 3). These results confirm a nematode host response when C. elegans were infected with S. aureus.

Finally, when we co-fed nematodes with H. kunzii and S. aureus strains, we observed that the majority of host defense genes (hlh-30, lys-5, cyp-37B1, and sodh-1) were overexpressed compared to OP50 or H. kunzii alone (P < 0.01). Gene expression levels were equivalent to those observed with nematodes fed with S. aureus alone (P-value, NS; Figure 3). Surprisingly the clec-7 gene has only significant variation of expression when C. elegans were fed with the colonizing S. aureus strain NSA1385 (P < 0.05).

These results suggest that the co-infection with H. kunzii and S. aureus induced an overexpression of some host defense genes. However, the variation of expression of clec-7 gene during the coinfection of H. kunzii and the virulent/non-virulent S. aureus strains could suggest some modulations of host defense. So, during the coinfection, the attenuation of S. aureus virulence in presence of H. kunzii seems to not be due to its capacity to trigger C. elegans host response that would help the fight against S. aureus infection. H. kunzii seems to directly act in the modulation of S. aureus virulence and to have no major role on the modulation of host immune defense.

To look into the possibility of direct attenuation of S. aureus virulence by H. kunzii, the expression levels of two representative virulence genes (hla and spa) and the main regulatory gene agr (that influences the expression of numerous S. aureus virulence genes) were measured for the different S. aureus strains associated with the H. kunzii isolate H13 and compared to those of S. aureus alone (Figure 4).

We observed that hla gene which encodes for the α-hemolysin (representing one of the most important virulence factors) was significantly derepressed in S. aureus NSA739 associated with H. kunzii H13 [Median −0.277; 95%CI (−0.41/−0.18); p < 0.01]. In the same way, agr was also significantly down regulated [Median −0.582; 95%CI (−0.36/−0.81); p < 0.001]. spa gene which encodes the protein A (representing one of the most important colonizing factor) showed a significant overexpression in S. aureus NSA739 associated with H. kunzii H13 [Median 0.197; 95%CI (0.15–0.27); p < 0.001]. The same results were noted for the two other S. aureus studied.

This data suggested that H. kunzii attenuated directly the virulence of S. aureus by a deregulation of virulence genes and the global regulator of this virulence.

To get a better understanding of the role of H. kunzii on S. aureus virulence, we evaluated the effect of sequential infections on nematodes. Firstly, we infected C. elegans with the different bacteria alone and in association during 12 h followed by a transfer of nematodes on OP50 strain. The results showed that S. aureus is clearly more virulent compared to H. kunzii (P < 0.001) and this virulence was not due to a constant reinfection since all the results were comparable to those obtained in the first experiments (Tables 2, 3).

Secondly, we sequentially inoculated the different associated bacteria. We infected C. elegans with the H. kunzii strains alone during 12 h followed by a transfer of nematodes on the different S. aureus strains tested. We observed that the LT50s for this protocol were significantly reduced (LT50: 2.5–4.1 vs. 4.1–5.7 days, respectively; P < 0.001). However, this impact was not clearly equivalent for the different combinations tested. Indeed, for the H. kunzii strain (H10) with different impact on the S. aureus virulence, the LT50 was similar to LT50 obtained for nematodes infected by NSA739 alone (LT50: 2.5 vs. 2.8 days, respectively; P-value, NS). For the H. kunzii strain (H13) with an impact on the virulence of all the S. aureus studied, LT50 remained significantly reduced compared to nematodes infected with NSA739 alone (LT50: 4.1 vs. 2.8 days, respectively; P < 0.001; Table 2). Thus, the direct association of H. kunzii and S. aureus has an impact on the attenuation of S. aureus virulence. This effect is significantly reduced or aborted when the infection is sequential suggesting the necessity to simultaneously co-infect with both H. kunzii and S. aureus to attenuate the virulence of S. aureus. H. kunzii seems to act directly on S. aureus reducing its virulence and thus the host response (showed by the reduction of cyp-37B1 and clec-7 expression previously).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.