Density of pathogens
In 1964, Bendy et al.  first proposed that the number of bacteria plays a critical role for the development of infection. Thereafter, some authors suggested that a bacterial density equal to or higher than 104 or 105 g−1 (or cm−3) of tissue may be required to cause wound infection . Finally, the concept of critical colonization emerged as a stage in which bacterial density is not sufficient to trigger infection but may suffice to surpass host defence without overt generalized immunological reaction. Critical colonization thus appears to be a precursor state of infection and has been implicated in delayed healing. In agreement with this concept, it has been recently shown by Xu et al.  that in neuropathic DFU, a high bacterial load was associated with a worse rate of wound healing and there was a significant inverse relationship between colony-forming unit (CFU) count from exudates and wound healing rate. However, how relevant is this concept in clinical practice? Ten years ago, Dow et al.  had already observed that β-haemolytic streptococci at 102 CFU g−1 of tissue were able to induce tissue damage, whereas a count greater than 105 CFU of less pathogenic organisms was of little significance. In the same way, Robson and Heggers  documented that wound healing may occur despite a high number of bacteria. Moreover, this concept, relying mostly on studies in acute wounds, is difficult to define in either clinical or microbiological terms and does not take into account the presence of specific pathogens. From a practical point of view, diagnosis of critical colonization should require to systematically perform a tissue biopsy for every wound, whose interpretation may be questionable because of heterogeneity of bacteria distribution within the wound tissue. Finally, it remains questionable that use of bacterial colony count may help the clinician to decide antibiotic therapy. Therefore, critical colonization appears as a debatable and unpractical concept.
Testing the virulence power of bacteria seems to be a more promising way to characterize infection and to differentiate it from colonization. Our studies focused on S. aureus, the most prevalent pathogen isolated from DFU, using several techniques, including DNA microarray-based genotyping, multiplex polymerase chain reaction (PCR) and Caenorhabditis elegans and zebrafish killing assays.
DNA microarray is a technique from molecular biology able to detect genes related to virulence and antibiotic resistance . Using this method, we were able to show that virulence and resistance genes were far more often present in clinically infected DFU than in uninfected wounds. Hence, 20 of 22 S. aureus isolates from uninfected ulcers were free of virulence genes, whereas these genes were detected in all but one of the 85 strains from infected ulcers . Multiplex PCR technique enabled us to screen a larger number of virulence genes. With this technique, we showed that the prevalence rate of ten genes was significantly higher in strains from infected ulcers than in those from noninfected ulcers . Moreover, by logistic regression analysis, a combination of five genes (sea, sei, hlgv, cap5 and lukDE) allowed to differentiate infected from noninfected DFU with a good sensitivity (0.98) and specificity (0.87). In a more recent study, we used a new generation of DNA microarray that covered 334 different genes for genotyping 195 S. aureus isolated as the sole pathogen in DFU. We found that a single gene (lukDE) may suffice to differentiate uninfected from infected ulcers (personal data). Moreover, lukDE was able to predict the outcome of grade 1 DFU. Finally, this new DNA microarray generation allowed to compare the strains, determine their clonality and thus predict more specifically the colonizing or infecting profile of S. aureus.
The importance of the bacterial virulence to induce infection was confirmed by killing assays using C. elegans and zebrafish as host models. C. elegans is a transparent and mostly hermaphrodite nematode, about 1 mm in length with a limited number of cells, which feeds mainly on environmental bacteria in soil and E. coli OP50 in the laboratory. This nematode has been proposed as a useful animal model to study host–pathogen interactions and assess the virulence potential of microorganisms , and the C. elegans killing assay has been validated to study the virulence of S. aureus. The test is based on ingestion of S. aureus by the nematodes, leading ultimately to the death of the worms. Because many of the virulence factors used by S. aureus to induce disease in humans are also required for full pathogenicity in nematodes, the percentage of killed worms along the time (worm survival curve) when exposed to different S. aureus strains is an indirect marker of virulence capacity. We found that for all but one S. aureus strain isolated from infected DFU, the time required to kill 50% of the worms (LT50) was shorter than 2 days, whereas for all but one S. aureus strain isolated from uninfected (grade 1) ulcers, LT50 was longer than 3 days . These encouraging results are currently being validated using another animal model, the zebrafish (personal data). In the latter, innate immune system is close to that of humans  and a model of infection by S. aureus has already been established . The transparency of the embryos and use of fluorescent bacteria make it possible to follow infections in real time.
Bacterial interactions: the role of biofilms
Bacterial interactions are currently considered important in many infectious diseases, including acute and chronic wounds . Bacteria may exist as planktonic or free-floating cells or sessile attached cells, contributing to form biofilms. According to Donlan and Costerton , biofilm is a (polymicrobial) sessile community characterized by cells that are irreversibly attached to a substratum or interface or to each other, and these cells are encased in a matrix of EPSs that they have produced. These cells exhibit an altered phenotype with respect to growth rate and gene transcription. Attachment of ‘pioneering’ bacteria is the first step towards biofilm development. Then, microbial cells begin to proliferate, forming small aggregates or microcolonies, and other bacterial colonizers coadhere within the biofilm under construction. This step generates chemical signals (auto-inducers), allowing cell-to-cell communication essential for biofilm formation, including EPS production, synergistic or antagonist interactions between cells and changes in bacterial phenotypic characteristics. This communication by auto-inducers is known as quorum sensing and gives information about cell density. So, when the number of cells has reached a critical threshold (the quorum), bacteria start to produce EPS and some genes get upregulated. In the mature biofilm, bacteria are embedded in EPS matrix and form three-dimensional mushroom-like structures containing water channels. Dispersal of bacterial cells from biofilm can be a passive phenomenon, such as erosion by fluid shear or an active one, known as ‘seeding dispersal’, and secondary to quorum-sensing-mediated death and lysis of cell subpopulation within the biofilm [52-54]. Biofilms are claimed to be involved in a number of infectious diseases such as endocarditis, prostatitis, periodontitis, otitis media as well as in infection from medical devices. Biofilms have been found in osteomyelitic bone , a common complication of DFU . More recently, James et al. , using light and scanning electron microscopy, reported that 60% of chronic wounds exhibited biofilms as opposed to 6% of acute wounds; of note, 13 of 50 chronic wounds that the authors analyzed were DFUs and 10 (77%) contained biofilm. According to Potera , 65% of human infections involve biofilms, but the presence of biofilms is not necessarily detrimental, occurring naturally in regions of the human body such as teeth, gastrointestinal tract or vaginal mucosa, and might be even protective against infection [52, 59].
Clinical consequences of biofilm are important. First, sessile bacteria show increased resistance to antibiotic agents. Thus, it was reported that the minimal inhibitory concentration and minimal bactericidal may be up to 100- to 1000-fold higher for the sessile form of the bacteria compared with their planktonic counterparts . Several mechanisms are put forward to account for this resistance [53, 54, 59]: EPS might constitute a chemical and/or physical barrier against penetration of some antibiotics. Modification of the micro-environment may also alter efficacy of antibiotic agents. Proximity of cells favours horizontal gene transfers and hence transfers of resistance gene. Sessile cells might also have upregulation of resistance gene. Because the biofilm is polymicrobial, antibiotic sensitivity is heterogeneous, with some species being susceptible to a given agent but others not. Finally, because biofilm-associated cells grow more slowly than planktonic cells, it has been proposed that they take up antibiotics more slowly. This phenomenon is more evident for the cells at the base of the biofilm, known as persister cells. As mentioned earlier, sessile bacteria are more virulent than their planktonic counterpart because of upregulation of genes encoding for virulence factors and horizontal gene transfer between bacteria. Another consequence of biofilm formation is resistance of bacteria to the immune system possibly due to decreased penetration of phagocytes through the EPS and/or defective opsonization impeding engulfment of bacteria by phagocytic cells [54, 61]. As a whole, those three characteristics (resistance to antibiotic agents, expression of virulence factors and resistance to the host immune system) may explain the frequent severity, chronicity and recidivism of DFI; moreover, the detachment of cells from the biofilm may facilitate the spread of infection notably to the bone.
The fundamental question is why are some biofilms ‘healthy’ and others able to induce infection. Two hypotheses are currently under discussion . The first, the ‘specific bacterial hypothesis’, suggests that only a few species of bacteria within the heterogeneous polymicrobial biofilm may be involved in the infectious process. Conversely, the ‘nonspecific bacterial hypothesis’ (or ‘community hypothesis’) considers the bacterial composition of biofilm as a whole such as ‘a functional unit’ and does not take account of individual pathogenic bacteria alone. Recently, the concept of FEPs was developed by Dowd et al. . According to this hypothesis, certain bacterial species considered as nonpathogenic when alone or species not capable of maintaining a chronic infection on their own may co-aggregate symbiotically in a pathogenic biofilm and act synergically to cause a chronic infection. Preliminary data supporting this concept have been published based on new biomolecular techniques that allow to identify various clusters in chronic DFU made up of particular species. A nonrandom distribution pattern of bacteria existed in the wounds. For example, a quantitative analysis of the distance of bacterial aggregates to the wound surface showed that the aggregates of Pseudomonas aeruginosa were located significantly deeper in the wound bed than those of S. aureus. This particular distribution of P. aeruginosa and S. aureus may explain the under-representation of P. aeruginosa and over-representation of S. aureus in chronic wounds by conventional culturing of wound swab samples. Anaerobes are present in most of FEPs, and distribution of the various pathogens in those multispecies biofilms is nonrandom; aerobic bacteria are localized in the upper surface where oxygen content is relatively high, whereas anaerobes are localized more deeply in niches created by oxygen consumption by aerobes. Biofilm and FEP concept may explain the delayed healing of chronic wounds, why cultures from chronically infected diabetic foot wounds are often polymicrobial and why anaerobes are frequently isolated when sampling, transport and culture methods are optimal. A complex sequence of events takes place during the formation of these biofilms, which includes random bacterial settlement of early colonizers, an increased competition among the present species and a niche differentiation resulting in very heterogeneous biofilms. This may also account for the resistance of some DFIs to a single antibiotic agent. In practice, the role of biofilms in DFI may encourage sharp debridement to remove biofilm bacteria and to obtain in-depth specimens for bacteriological culture. High prevalence of anaerobes in pathological biofilm requires using suitable culture media to isolate them. Because species traditionally considered as low virulent can be aggressive if they are part of an FEP, they should be viewed cautiously. Finally, techniques of molecular biology have to be developed to identify bacterial populations that are difficult to isolate by the usual culture techniques.
In conclusion, DFI is a complex pathology (Figure 2) involving to a varying extent the host and the bacteria. As a whole, this may explain why DFIs are frequent, often severe and difficult to treat. Finally, classic dogma such as Koch's postulates or the single-pathogen paradigm must be revisited in the light of new data from molecular biology.