Context and background: One of the most common complications of diabetes is the development of ulcers in the foot. The causes of these diabetic foot ulcers (DFU) can be numerous and even result in lower extremity amputations.
One of the major contributors to healing impairment of any wounds is microbial infection. DFU infection and biofilm formation contributes to the risk of amputation, with almost 90% DFU related amputations having a wound infection. Previous studies have identified polymicrobial infections with a mix of aerobic and anaerobic bacteria in the infected DFUs (see  for a review on diabetic foot infections). Antibiotics are used for treatment of DFUs, but if these have an effect on the different species of the polymicrobial infections or if they confer an advantage to a certain species, limiting the efficiency of the antibiotic, is unknown.
Though previous studies have identified specific bacterial species (like Staphylococcus aureus, Streptococcus species and Pseudomonas aeruginosa) in DFU infections, these methods are largely culture based or amplicon-based and have their own limitations.
This study takes the approach of shotgun metagenomics to understand bacterial infections in DFUs at a strain-specific level. They also study the biofilm lifestyle, the antibiotic resistance, effects of therapeutics and the host response based on different clinical isolates identified in a mouse model.
Experimental setup: Samples from DFUs of 100 subjects were analysed at initial presentation and followed up every 2 weeks until the DFU healed, was amputated or up to 12 weeks if neither occurred. All wounds were subjected to debridement, to clear necrotic tissue and callus formed, at every follow up. Out of the total cohort, 30% received antibiotics (due to other complications). Host response was tested in vitro using keratinocytes grown in cell-free spent media from planktonic cultures or 72 hour-biofilm cultures of the clinical isolates identified in this study. A murine model was used to test response in vivo to these isolates.
Strain-level identification of bacteria in DFU infections: For the cohort in this study, the most abundant genera identified (Staphylococcus (18.95%), Corynebacterium (14.64%), Pseudomonas (9.37%), and Streptococcus (7.32%)) were consistent using with 16S sequencing data from another study. However, a genera-level identification is not satisfactory as can be seen by the fact that Staphylococcus accounts for species associated with healthy skin (S. epidermidis) as well as pathogens (S. aureus). Using shotgun metagenomics, the study was able to achieve strain-level identification of Staphylococcus aureus, Pseudomonas aeruginosa, Corynebacterium striatum, and Alcaligenes faecalis as the main constituents in all DFU infections. A single strain (S. aureus 7372) was identified as the major strain of the Staphylococcus species present.
Healing vs non-healing wound microbiomes respond differently to debridement: Wounds that did not heal till the 12-week mark of this study were termed as non-healing. No differences were obtained with respect to diversity in DFU microbiomes of antibiotic receiving and not receiving subjects, thus it did not relate to the healing or non-healing phenotypes. A therapeutic intervention that all subjects received was wound debridement. All healing wounds showed a change in the microbiome composition immediately post debridement when compared to pre-debridement. Anaerobic bacteria such as Anaeroccocus, Porphyromonas, Prevotella, and Veillonella spp. were found to be reduced in these wounds post debridement. No such change was seen in non-healing wounds, indicating that the microbiome of these doesn’t respond to the initial debridement. Furthermore, deep and poorly oxygenated wounds correlated with biofilm-like metabolism, indicating that non-healing wounds are associated with biofilm formation.
S. aureus strain level differences detected in healing vs non-healing wounds: S. aureus was found in 94% samples with varying wound outcomes, suggesting that strain-level differences may play a role. Using a phylogenetics approach, particular strains were identified. S. aureus 7372 (SA7372) was found across all wound types and S. aureus 10757 (SA10757) was found only in non-healing wounds. Whole genome sequencing of these 2 strains revealed genetic differences which were also phage-dependent.
Host response to the different strains: To study if these strains elicited different host response, an in vitro approach was taken. Corynebacterium striatum (considered opportunistic pathogen but not regularly identified in the clinical laboratory), Alcaligenes faecalis (considered non-pathogenic environmental contaminant) along with the two S. aureus strains were used. Wounded primary keratinocytes were grown in cell free spent media from either planktonic or biofilm cultures. After 8 hours, cytokine profiling of A. faecalis infected keratinocytes revealed an increased production of IL-8, G-CSF, GM-CSF, IL-6, TGF-α, TNF-α, IP-10 and platelet growth factor. Thus A. faecalis is found to elicit a strong immune response and thus could play a beneficial role in wound repair.
To understand what happens in vivo, the same strains were used to form biofilms and transferred onto wounds created on the mouse dorsa. Wounds were followed and photographed at various intervals upto day 28. The A. faecalis and C. striatum infected wounds displayed slight delays in healing, but then resumed similar healing as the control. Both S. aureus strains displayed a persistent delayed healing, with SA10757 being worse than SA7372 wounds. These findings indicate an effect of strain-level variation on the healing process in hosts.
Interesting aspects of the study: This study takes a unique approach to understand microbial infections in DFUs and identify the bacterial strains involved. They further correlate their observations of healing vs non-healing wounds to debridement and the differences of the wound conditions like depth and oxygenation to the microbiome. The in vitro study of cytokine profiles reveals a plausible beneficial role for A. faecalis in wound repair by eliciting an immune response. The correlation of the healing dynamics in vivo to the observed strain differences from the metagenomics data of healing vs non-healing wounds supports the hypothesis that strain-level differences give rise to different wound outcomes. The study provides a method to study DFU microbiomes in more details, and could be applied to other wound types too for a clearer understanding of the wound microbiome.
Questions for the authors: The SA10757 strain was found in ~18% of non-healing wounds. Do the other strains identified also have similar genetic differences? It would be interesting to compare the genome sequences of the other strains found specific to non-healing wounds and see how similar or different they are.
Since most wound infections are polymicrobial, it would be interesting to see the healing dynamics in vivo using mixed biofilms. Given that A. faecalis elicits a strong immune response, would we expect a mixed biofilm of A. faecalis and S. aureus to show faster healing rates as compared to just S. aureus biofilms? I think given the interspecies interactions, there are various interesting questions that can be answered here.
- Alexiadou, Kleopatra, and John Doupis. “Management of diabetic foot ulcers.” Diabetes Therapy 3.1 (2012): 4.
- Hobizal, Kimberlee B., and Dane K. Wukich. “Diabetic foot infections: current concept review.” Diabetic foot & ankle 3.1 (2012): 18409.
Posted on: 24th October 2018