Bacterial Strain Diversity Within Wounds

Abstract

Significance: Rare bacterial taxa (taxa of low relative frequency) are numerous and ubiquitous in virtually any sample-including wound samples. In addition, even the high-frequency genera and species contain multiple strains. These strains, individually, are each only a small fraction of the total bacterial population. Against the view that wounds contain relatively few kinds of bacteria, this newly recognized diversity implies a relatively high rate of migration into the wound and the potential for diversification during infection. Understanding the biological and medical importance of these numerous taxa is an important new element of wound microbiology. Recent Advances: Only recently have these numerous strains been discovered; the technology to detect, identify, and characterize them is still in its infancy. Multiple strains of both gram-negative and gram-positive bacteria have been found in a single wound. In the few cases studied, the distribution of the bacteria suggests microhabitats and biological interactions. Critical Issues: The distribution of the strains, their phenotypic diversity, and their interactions are still largely uncharacterized. The technologies to investigate this level of genomic detail are still developing and have not been largely deployed to investigate wounds. Future Directions: As advanced metagenomics, single-cell genomics, and advanced microscopy develop, the study of wound microbiology will better address the complex interplay of numerous individually rare strains with both the host and each other.

Benjamin C. Kirkup, Jr., PhD

Figure 1.

Undersampling and undercharacterization. The upper left figure represents a population of bacteria, each with three characters. Through undersampling (upper right) the population is characterized as entirely uniform despite containing relatively high diversity. Despite exhaustive sampling, the same misconception is arrived at by limiting characterization to a relatively conserved character (lower left). The error is multiplied when both undersampling and undercharacterization occur, in whichever order (lower right). To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/wound

Figure 2.

Probability of detecting outbreaks with low-abundance strains. This graph represents the results of a simulation in which a variable number of isolates are characterized from five samples; the probability of detecting a strain that composes some fraction of the actual population at least once in each of the samples is charted. By characterizing one isolate per sample, even a strain that is 90% of the population is detected only half the time in all five of the samples. Thus, its ubiquity would be unacknowledged. As the number of strains characterized per sample increases, the probability of detecting strains of even lower abundance increases substantially. With 20 isolates per sample, even a strain that is only 10% of the population can be detected at least once in all five samples. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/wound

Figure 3.

Detecting low-frequency strains via shotgun metagenomics. In a normal wound sample, over 90% of the DNA recovered is human DNA. The human-DNA-depletion kits typically remove 90% of the human DNA. This creates a sample that is 91% bacterial DNA. In a sample dominated by a single genus, that genus may contribute 50% of the total bacterial DNA, and a species within that genus may thus represent ∼2% of the original DNA sample. Given that the sample is run alone on a MiSeq (an extremely conservative estimate), 6 M reads may be recovered from that one species. Using a marker gene present in all the members of the species to assess diversity, ∼2,400 reads will map to that gene. Given the length of an average gene and an average read, any given location along the length of the gene will be covered a median of 100-fold. This could be adequate for genotyping a single strain, but for detecting subpopulations, it is not. Most reads include errors, making detection of minority strains within the most-frequent species difficult at best. A potential solution is the use of even higher sequencing depths, but this carries a great expense, not only for sequencing but during data analysis and computation. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/wound

Figure 4.

Decomposition of taxa. As frequent higher taxa are decomposed into multiple less-frequent taxa, the number of rare taxa above the threshold of detection increases, but each taxa has a lower effective population size. This has population genetic implications. First, drift is a more significant factor in each of the populations, and this will tend to reduce the impact of selection. The addition of numerous taxa into a relatively short-lived habitat (the wound) suggests a relatively high rate of migration into the wound. The diversity may also be the result of relevant mutational processes. Finally, increased diversity implies greater opportunity for mating/recombination within the wound. The population genetic processes themselves have an impact on epidemiology, infection control, the rate of pathoadaptation, and bacterial response to antibiotic administration. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/wound

Figure 5.

Spatial variance. Community composition is shown on a spatial grid in the context of three environmental variables. Each variable has its own internal spatial autocorrelation as variables change continuously over space. The correlation between the collected environmental variables and the community composition is perceived as the impact of the environment on the community. The degree of autocorrelation within the environmental variables that overlaps with autocorrelation within the community explains some fraction of the autocorrelation within the community composition. The excess autocorrelation within the community composition is the degree to which migration limits the distribution of organisms in the wound. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/wound

Figure 6.

Sampling microhabitats. The process of sampling a wound may lead to the recovery of several different habitats in a sample, each represented in varying proportion. In the figure, the wound biofilm is simplified to four distinct communities. The two samples represent capturing all four communities in one sample but only one community in the other. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/wound