Universal Dermal Microbiome in Human Skin

DISCUSSION

Different anatomic locations were expected to have different compositions and richness (4), but we demonstrated that neither bacterial community composition nor overall bacterial richness (as operational taxonomic units [OTU]) differed between the anatomic locations (Fig. 1a). This challenges previous findings, in which the cutaneous community composition differed greatly between skin locations, even within habitats (4). While local conditions, such as density of glands or hair follicles, as well as chemical factors, such as pH, moisture, and temperature, all influence bacterial composition (1, 4, 12, 15), the hip and knee are both dry skin habitats with similar topographical conditions, which may explain why no significant differences were observed. Additionally, while a strong and significant patient effect was expected based on many previous studies (1, 3–7, 17), this was limited to the epidermal communities, and no effect on the dermal communities was detected at all (for skin compartments: see Fig. S1a and b in the supplemental material).

Differences in skin compartments had the largest effect on determining the overall community composition. The dermal community consisted of a specific subset of the epidermal community, but the stark contrast between dermal and epidermal bacterial communities was surprising. Previous attempts to analyze divided skin biopsy specimens by 16S rRNA sequencing found similar microbial profiles in both compartments (9), whereas in our study, bacterial community composition and richness (as OTU richness) differed significantly between the dermal and epidermal compartments (Fig. 1). Additionally, when comparing community similarities (using Bray-Curtis similarity matrices), we found that the dermal compartment’s community was significantly less variable than that of the epidermal compartment (Fig. S2). The heat tree (Fig. 2) depicts the difference in OTU richness associated with each phylogenetic group and illustrates that the majority of OTUs were enriched in the epidermal compartment (blue), while relatively few groups demonstrated higher richness in the dermal compartment (yellow). The results here suggest that the bacterial community of the dermal compartment is less rich across nearly all taxonomic groups.

The distributions of the four major phyla were similar to those of previous investigations in human skin (1, 4, 9, 18). Other than the more abundant Pelomonas spp. within the phylum Proteobacteria, the most persistent genera showed abundance in the dermal community that was similar to or lower than the abundance in the epidermal community. Proteobacteria are found to colonize deep, cutaneous compartments and are presumably involved in regulating skin homeostasis between the host and the environment (19). Specifically, the dermal highly abundant Pelomonas spp. are one of the core commensals in cutaneous communities (19). Concurrently, Corynebacterium, a genus that is predominant in moist skin habitats but also prevalent in dry skin habitats (12), was mainly abundant and richer in the epidermal compartment than in the dermal compartment (Fig. S5). Together, these results suggest that the dermis, with its less complex and less variable bacterial community, is more specifically colonized (2).

To further characterize the differences between skin compartments, we analyzed OTUs that significantly differed in persistence between dermal compartments. An OTU may have a low abundance within the microbiome but nevertheless may be observed in many samples, suggesting a functional role. We define persistence as the percentage of samples in which an OTU is found (Fig. 4). A total of 75 OTUs differed significantly in persistence between the epidermal and dermal compartments. These OTUs were categorized as being present in both compartments or only the epidermal compartment and were classified into anaerobic or aerobic bacteria based on published literature findings. All OTUs, except Pelomonas saccharophila, were more persistent in the epidermal compartment, while there were no clear trends in the distributions of anaerobic and aerobic bacteria between the skin layers. Similarly, only a few individual OTUs increased in abundance within the dermis, which included a Methylobacterium species and a Brevundimonas species. These have been recorded as opportunistic pathogens in immunocompromised patients or in those suffering from an underlying disease (20, 21), and the dermis may serve as a reservoir for potential pathogens for patients at risk of skin and soft tissue infections. This is in agreement with a growing body of literature suggesting that a lack of regulation of the microbiome by the host (dysbiosis) can lead to previously commensal bacterial species becoming pathogenic (22). However, it should be noted that Methylobacterium spp. and other genera such as Xanthomonas spp. have previously been identified as common contaminants in DNA extraction kits (23). In our study, these genera were absent in our negative controls and were more abundant in the dermal communities, suggesting a role in the local microbiota rather than a systemic contamination, while other OTUs assigned to Atopobium, Cellulomonas, and Conchiformibius genera were likely contaminants.

Given that the dermal and epidermal bacterial communities differed in composition and richness, we examined the factors that regulate these differences. The effects of age, sex, smoking status, diabetic status, anatomic location, and interpersonal variation on the bacterial OTU richness of each skin compartment were modeled using generalized linear modeling, while envfit and multivariate general linear modeling (Table S1) were used to estimate their effects on the bacterial community composition. There was evidence that age, smoking status, diabetic status, and interpersonal variation affected the epidermal composition and bacterial OTU richness. There was also additional support that sex and anatomic location affected composition but not OTU richness (Table S1 and Fig. S3 and S4). The dermal microbiota was considerably less affected by these factors, with only interpersonal variation and age affecting bacterial composition, while no factor correlated with the dermal bacterial OTU richness. The results here therefore suggest that the epidermal bacterial composition was more affected by external factors, resulting in its higher variability. This is in accordance with the previous findings of many studies, where significant interpersonal variability was observed (6, 12, 16, 24–28). This variation is likely to be driven by greater exposure of the epidermis to the external environment. It is also quite possible that compositional differences among individuals’ dermal bacterial communities are at least partly driven by host genotypic variation as well as immunology (12).

The functional capabilities of the epidermal and dermal bacterial communities were computationally annotated to predict potential gene families. A PCA plot of the predicted genes demonstrates separate groupings of samples from the dermal and epidermal compartments (Fig. S6) based on predicted gene ortholog abundances. There was a significant difference (P < 1e−5) in relative gene abundances between the two compartments, suggesting a difference in overall functionality of the microbial communities. A heat map (Fig. 4a) and a simplified ridge plot (Fig. 4b) describing the 25 most significantly different pathways (all false discovery rates [q values] < 1e−5) clearly illustrate the variability in functionality throughout the stratified tissue. The difference in proportions of the pathways can lead to an inference of the functional relevance in the epidermal and dermal communities. The enriched pathways in the dermal community, such as nitrogen metabolism, bacterial chemotaxis, flagellar assembly, and lipopolysaccharide biosynthesis, indicate a possible dormant bacterial community that is initiating a biofilm mode of growth and survival by use of alternative energy pathways. Concurrently, the dermal depleted pathways include DNA replication, mismatch repair, homologous recombination, and pyrimidine/purine metabolism (Fig. 4), indicating bacteria with a decreased cell growth. These predictive, metagenomic analyses suggest a clear contrast in functional capability of the microbial communities between cutaneous compartments.

Various skin compartments exhibit specific milieus (2, 12). Hair follicles contain anoxic environments (12) that may allow anaerobic bacteria to thrive, cultivating their own unique localized microbiota (29). They stretch from the outermost layers of the epidermis deep into the dermis (14), contain a significant proportion of microorganisms, and should be considered pathways for the microbiota of the epidermis to enter the dermal layer of the skin and vice versa. Nevertheless, the dermal community is well protected against topical cutaneous antimicrobials, such as preoperational preparation antimicrobials that cannot penetrate the many cutaneous layers (30). The presence of a sebaceous plug in the infundibulum of hair follicles may furthermore hinder the penetration of drugs (31), and the deep regions of hair follicles thereby act as microbial reservoirs (32). This hidden dermal microbiota can easily reestablish the core of the epidermal microbiota or, during an operation, accidently be pushed further into the tissue and potentially initiate an infection.

In surgical site infections and chronic wounds, Staphylococcus spp. are ubiquitous (28, 33). Our findings of a medium abundance of Staphylococcus spp. in healthy skin (Fig. 2; Fig. S5b) suggest that this omnipresent genus possesses the potential to grow and outcompete other genera under favorable conditions. Coagulase-negative staphylococci (28), Cutibacterium acnes (34), and Pseudomonas aeruginosa (35) are also commonly found in surgical site infections and chronic wounds, as well as in healthy human skin (13). These potentially pathogenic species can cause complications when forced out of their natural habitat by surgical utensils or needles and possess the ability to develop chronic infections in opportunistic locations. These and other opportunistic pathogens sustain a low metabolism within the cutaneous compartments but thrive when favorable conditions exist.

Overall, our findings suggest an advanced adaptable epidermal microbiota and a less complex dermal core community in healthy human skin. Each of these communities can adapt to the compartment in which they thrive. Focusing on the dermal community may allow clinicians and researchers to simplify the relationship between the skin microbiome and skin disease, as it is less affected by environmental factors. These new findings may change the perception of the human skin microbiota as being entirely individual. However, distinctive details in the core microbiome may provide insight into health and genetics, as well as diet, lifestyle, and surroundings (16). While previous studies have attempted to map the entire epidermal skin microbiome (36), a greater understanding of the spatial scaling of microbial communities would also enhance our understanding of its function in chronic cutaneous diseases and infections. If the dermal communities serve as a reservoir for pathogenic bacteria that drive skin or soft tissue infections, strategies for preparing and managing patients undergoing surgery should be adjusted to reduce chances of complications. Speculatively, future preoperational procedures may include microbial analysis of the patient´s skin to predict potentially susceptible patients at risk for a postoperational infection. Correspondingly, the microbial composition of individuals suffering from chronic cutaneous conditions may predict upcoming flares of eczema and thereby improve prophylactic treatment. We therefore suggest that targeting different cutaneous compartments should be prioritized in future skin microbiome studies. A connection between autoimmune diseases or skin disorders and the microbiota may become clearer by investigating the dermal microbiota, as the dermal community is less affected by external factors and therefore possibly more stable. Further studies are needed to determine whether this core “universal microbiota” is present within dermal compartments across habitats. Nevertheless, future studies into chronic skin disorders would benefit from an increased focus on all skin compartments rather than just the epidermis, despite the invasive nature of biopsies.

The strength of this study is that the separation of cutaneous compartments allowed for the characterization of distinct communities at a level of detail not found in previous studies (2). The use of DNA metabarcoding allows for bacteria to be amplified even in very low concentrations and allows for noncultivable bacteria to also be described as part of the overall microbial composition (37). However, in future studies, a greater consideration of different skin habitats (including moist and sebaceous), ethnicities, and ages of participants should be included to extend the insight into the human microbiome. The addition of shotgun metagenomic and metatranscriptomic approaches would also considerably expand our understanding of the composition and functioning of the skin microbiome within different skin compartments (13).