Existence of a Colonizing Staphylococcus aureus Strain Isolated in Diabetic Foot Ulcers

Staphylococcus aureus is by far the most common and virulent pathogen in diabetic foot infection (1,2). However, this causative pathogen is a common inhabitant of the skin microflora and colonizes the nares and other human mucosa. It may be considered as an opportunistic colonizing organism. Recently, we demonstrated the coexistence of two populations of S. aureus strains isolated from diabetic foot ulcer (DFU): strains isolated from uninfected ulcers with a low virulence potential as opposed to strains isolated from infected ulcers with a high virulence potential (3,4). Moreover the strains belonged to two clonal complexes (CC8/CC5) that appeared to be linked to uninfected ulcers, enabling us to distinguish uninfected from infected wounds (5). In this study, we describe for the first time an insertion of a phage in the CC8 lineage of methicillin-sensitive S. aureus (MSSA), which is associated with the colonizing S. aureus strains, and we report the impact of this phage on biofilm formation and bacterial virulence.

All bacterial strains used in this study are listed in Table 1. Bacteria were grown at 37°C in Luria Bertani broth or brain-heart infusion broth.

Fer-15 worms were maintained and infected as previously described (6). All experiments were conducted in triplicate and repeated at least five times for each strain. S. aureus virulence was assessed using the nematode survival curve and calculating the LT50 and LT100 (median lethal time 50% and 100%, respectively).

The presence of S. aureus in the Caenorhabditis elegans digestive tract was determined at 72 h as described by Garsin et al. (7). Three replicates were performed for each strain. Infection of zebrafish embryos was carried out as previously described (8). More information regarding the two models can be found in the Supplementary Data.

Twenty-two strains from our panel of colonizing and infecting S. aureus strains isolated from DFUs (5) were chosen for optical mapping. Optical maps were provided by OpGen (Gaithersburg, MD), prepared on the Argus Optical Mapping System as described previously (9), and analyzed with the support of Phylogene (Bernis, France). The optical maps of the studied strains were then compared with the in silico restriction maps of 19 sequenced S. aureus isolates whose sequence genomes were available in GenBank and transformed by using the MapSolver v.2.1.1 software (OpGen SA).

Genomic DNA of S. aureus NSA1385 was sequenced using a 454 Life Sciences-Roche platform by Lifesequencing S.L. (Valencia, Spain). The combination of scaffolds and contigs resulted in an estimated genome size of 3.2 Mb.

The PCR protocol is presented in the Supplementary Data. After purification, PCR products were sequenced using a PerkinElmer ABI 377 sequencer and compared with sequences in GenBank by BLAST (http://www.ncbi.nlm.nih.gov/blast).

To evaluate the biofilm formation, we used the BioFilm Ring Test (BioFilm Control, Saint Beauzire, France) according to the manufacturer’s recommendations (10). Three experiments with three repeats each (three wells per slide) were performed per strain and incubation time.

To detect and evaluate spontaneous excision of the ROSA-like phage from the hotspot region ΦNM2, we used different procedures already described: the TMS medium with or without FeCl₃ (50 μmol/L) to create iron-repleted and iron-restricted growth conditions, mitomycin C, and UV treatment (11–13). We evaluated the occurrence frequency of NSA1385 mutants that lost the phage by counting the number of NSA1385 without phage colony-forming unit (CFU) and compared with the NSA1385 CFU number. Phage excision was confirmed using a PCR assay and genome sequencing.

The Mann-Whitney test was used to compare the in vivo bacterial growth of the different strains. To compare overall survival curves in nematode and zebrafish killing assays, a Cox regression was used. For pairwise comparison of two survival curves in nematode and zebrafish killing assays, we used a log-rank test. Statistical analysis was performed using the S-PLUS 2000 software package (Insightful Corporation, Seattle, WA), and results were considered significant at P < 0.05.

We used two infection models to confirm our previous observations that clinical S. aureus strains isolated from grade 1 DFUs are less virulent than strains isolated from grade 2–4 DFUs (4). We analyzed the behavior of five clinical strains: the two uninfecting strains (NSA1322 and NSA1385 from grade 1 DFU), the two infecting strains (NSA739 and NSA18026 from grade 2–4 DFU), and the reference strain Newman. In the C. elegans model, the five studied strains killed the nematodes more rapidly than the avirulent Escherichia coli OP50 strain used as nutrient for the nematodes (P < 0.001) (Table 1). The LT50 were similar for the two colonizing strains and the strain Newman but significantly longer (P < 0.001) than the LT50 of the infecting strains (4.0–4.6 ± 0.3 vs. 1.6–1.7 ± 0.2 days, respectively) (Table 1). The differences in virulence were not due to differences in the survival and proliferation of strains within the nematode intestine, since the intestine colonization by the different strains was not significantly different (Table 1).

Based on the results obtained with C. elegans, we investigated whether the two uninfecting strains (NSA1322 and NSA1385) also exhibit lower virulence in zebrafish embryos. We compared mortality rates over a 92-h period. The two colonizing strains caused less embryo deaths than the two infecting strains (P < 0.001) (Fig. 1A and B).

To investigate the difference between colonizing and infecting strains isolated from DFUs (4), we used optical mapping to analyze their genome organization (Supplementary Fig. 1). The five colonizing strains were clonal (>99% similarity) and clustered closely to two reference strains (Newman and NCTC8325) with ∼98% similarity. The majority of the infecting strains belonged to different clonal groups with <90% similarity to the colonizing strains. Interestingly, five infecting strains (NSA739, NSA6759, NSA11260, NSA18026, and NSA56348) exhibited 96.5% similarity with the colonizing strains. All these strains belonged to the CC8-MSSA clonal complex.

The major difference between the colonizing and the infecting CC8-MSSA strains was the presence of a large insertion located exclusively in all of the colonizing strains (CC8- and CC5-MSSA). The insertion was located at the previously described ΦNM2 integration hotspot (Supplementary Fig. 2), a known hotspot for genetic insertions, with phage insertions identified in the published genomes of the two reference strains (Newman and NCTC8325) (14).

To characterize the genetic insertion present in the colonizing strains, a draft genome sequence of NSA1385 was determined to ∼29× coverage. The integration site is in the intergenic region between rpmF (encoding the 50S ribosomal protein L32) and isdB (encoding the staphylococcal hemoglobin receptor required for heme-iron uptake). The insertion has a high G+C content (35.4%, compared with 32% for the chromosome in total) and appears to be an integrated bacteriophage. This phage has a genome of 44,031 bp with 73 open reading frames and is allocated within the unclassified dsDNA phages group. The comparison of the phage sequence with sequences in GenBank highlighted that the phage integrates at the same sequence as the phage ROSA, a previously described phage (15) with no known function. The sequences were comparable but we noted an inversion of a part of the phage sequence (Supplementary Fig. 2). We named this genetic island ROSA-like. More information can be found in the Supplementary Data.

To evaluate whether the ROSA-like phage insertion is a common feature from uninfecting strains, we tested a collection of S. aureus strains isolated in different clinical situations in France (Table 2). The insertion was identified in 40 of 392 strains (10.2%) of our collection, with 39 of 75 (52%) from colonizing DFUs and 1 of 131 (0.8%) from nose. The ROSA-like phage was almost exclusively found in grade 1 ulcers (39 of 44 uninfected wounds, 88.6%), and all the strains harboring the insertion belonged to the CC8/CC5-MSSA lineages.

To determine the impact of the genetic insertion, we studied biofilm formation using the BioFilm Ring Test. In brain-heart infusion medium, we observed that the colonizing NSA1385 strain formed biofilms more rapidly than the infecting NSA739 strain (150 vs. 180 min) (Fig. 2). These results confirmed the colonizing role of the investigated strains.

An important issue in clinical practice is to know if the ROSA-like phage is stable or not. Using different DNA-damaging stimuli, we observed that the frequency of excision was <5 × 10⁻⁹, suggesting a strong selective pressure for its maintenance. One strain with a confirmed loss of the ROSA-like phage (detected by PCR, optical map, and DNA sequencing) was selected for further experiments and named NSA1385(P−).

To definitively understand the impact of the phage insertion in the colonizing strain, we evaluated the biofilm formation and the virulence of NSA1385(P−). We observed the restoration of biofilm formation (Fig. 2) and bacterial virulence using the two in vivo models (Table 1 and Fig. 1B). All these data suggest that the ROSA-like phage clearly influences the virulence of the colonizing S. aureus strain. The absence of the phage restores the bacterial virulence. However, the low level of excision indicates that the ROSA-like phage is very stable and suggests that this colonizing strain does not require antibiotic treatment.

S. aureus is one of the most frequent pathogens isolated from community-acquired and nosocomial infections and is the most prevalent in DFUs (2). Even though S. aureus strains can colonize different human mucosa and may be considered as a commensal organism, these bacteria are clearly pathogens. However, we recently described the existence of a colonizing S. aureus strain isolated from DFUs (4). Here, a comparative genomic strategy using a collection of clinical strains detected a genetic determinant (a ROSA-like phage) associated with the attenuation of the clonal group. To confirm that this element was responsible for the colonizing behavior, we assessed the distribution of the ROSA-like phage in disease-causing and asymptomatically carried S. aureus in a national, nonbiased population taken from national epidemiological studies. The results demonstrated, for the first time, that the prophage was associated with the ability of the bacteria to colonize chronic ulcers and was responsible for the noninvasive character.

The carriage of virulence determinants by phage is not an uncommon situation in bacterial pathogens (16–20). In S. aureus, many of the phages encode and disseminate potent staphylococcal virulence factors (e.g., Panton-Valentine leukocidin) or resistance determinants (SCCmec cassette) (21). However, more recently, authors have shown that some phages could affect bacterial virulence by preventing the production of toxins (22). In our study, we demonstrated that the phage insertion blocks the virulence potential. Moreover, in the colonizing S. aureus strain, the difficulties to induce phage excision (using DNA-damaging agents), the low phage excision frequency, and the different rearrangements noted (compared with the published genome sequence of phage ROSA) demonstrated a great stability of this insertion, suggesting a low ability to spread or transfer to other staphylococci.

The association with a high potential of biofilm formation and the avirulence of the S. aureus strains containing the ROSA-like phage may explain the ability of the bacteria to colonize chronic wound tissues and their ability to exist in the commensal state. Interestingly, these colonizing strains were exclusively found in uninfected ulcers that represent the niche of these strains. The role played by these bacteria would be directed toward the establishment of colonizing wounds and/or help other pathogens in the pathogenic process, the ROSA-like insertion bringing a selective advantage to this role. From a clinical point of view, the stability of phage and the very low potential of virulence of the colonizing strain suggest that it is not necessary to treat ulcers carrying this S. aureus type. Our findings may contribute to better diagnosis and improved treatment of diabetic ulcers.

Acknowledgments. The authors thank Lifesequencing S.L. (Valencia, Spain) for sequencing the strain. The authors acknowledge the editorial assistance of Mariella Lomma (University Hospital Carémeau). The authors thank René Courcol (Lille University Hospital) for his fruitful comments and Laure Vidal-Navarro and Christelle Ngba (Nîmes University Hospital) for their help in this study.

Funding. Work in INSERM U1047 is supported by INSERM and the University of Montpellier. This work was supported by grants from the French Society of Diabetes (ALFEDIAM grant 2008) and the Languedoc-Roussillon Region (Chercheur d’Avenir award 2009). N.M. received a PhD fellowship from the Infectiopôle Sud Foundation. T.K.P. was supported by a British Infection Association research fellowship. E.P.S. was supported by a grant from the National Institutes of Health (NIH; R01-AI-069233). Fer-15 nematodes were provided by the Caenorhabditis Genetics Center, a foundation of the NIH National Center for Research Resources.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. N.M. researched data and wrote the manuscript. T.K.P. and C.D.-R. researched data and reviewed the manuscript. G.L., D.O., S.J.F., S.A.R., and E.P.S. contributed to discussion and reviewed and edited the manuscript. M.B. researched data. F.V. contributed to discussion and reviewed the manuscript. A.S. designed the study and reviewed and edited the manuscript. J.-P.L. designed the study, researched data, and wrote the manuscript. J.-P.L. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.