New Insights into the Intrinsic and Extrinsic Factors That Shape the Human Skin Microbiome

Abstract

Despite recognition that biogeography and individuality shape the function and composition of the human skin microbiome, we know little about how extrinsic and intrinsic host factors influence its composition. To explore the contributions of these factors to skin microbiome variation, we profiled the bacterial microbiomes of 495 North American subjects (ages, 9 to 78 years) at four skin surfaces plus the oral epithelium using 16S rRNA gene amplicon sequencing. We collected subject metadata, including host physiological parameters, through standardized questionnaires and noninvasive biophysical methods. Using a combination of statistical modeling tools, we found that demographic, lifestyle, and physiological factors collectively explained 12 to 20% of the variability in microbiome composition. The influence of health factors was strongest on the oral microbiome. Associations between host factors and the skin microbiome were generally dominated by operational taxonomic units (OTUs) affiliated with the Clostridiales and Prevotella A subset of the correlations between microbial features and host attributes were site specific. To further explore the relationship between age and the skin microbiome of the forehead, we trained a Random Forest regression model to predict chronological age from microbial features. Age was associated mostly with two mutually coexcluding Corynebacterium OTUs. Furthermore, skin aging variables (wrinkles and hyperpigmented spots) were independently correlated to these taxa.IMPORTANCE Many studies have highlighted the importance of body site and individuality in shaping the composition of the human skin microbiome, but we still have a poor understanding of how extrinsic (e.g., lifestyle) and intrinsic (e.g., age) factors influence its composition. We characterized the bacterial microbiomes of North American volunteers at four skin sites and the mouth. We also collected extensive subject metadata and measured several host physiological parameters. Integration of host and microbial features showed that the skin microbiome was predominantly associated with demographic, lifestyle, and physiological factors. Furthermore, we uncovered reproducible associations between chronological age, skin aging, and members of the genus Corynebacterium Our work provides new understanding of the role of host selection and lifestyle in shaping skin microbiome composition. It also contributes to a more comprehensive appreciation of the factors that drive interindividual skin microbiome variation.

Keywords: Corynebacterium; age; demographic; forehead; host lifestyle; host physiology; metadata; scalp; skin microbiome.

FIG 1

Microbiome diversity. (A) Bacterial alpha diversity at each site. (B) NMDS ordination displaying bacterial composition similarity (Bray-Curtis dissimilarities) among samples, color-coded according to site. (C) Stacked bar plot showing the relative abundance of the top ten most abundant bacterial families in each of the five sampled body sites. The “Others” category represents less-abundant taxa.

FIG 2

Effect sizes of variables on microbiome composition. Variables found to be significantly correlated with overall forehead (A) and oral (B) microbiome variation, sorted by their relative importance (% of R²) within predefined categories. R² values represent the fractions of microbial composition variation explained by the variables in each category. Variables are additionally described in Table S1B.

FIG 3

(A) Mean relative importance of variable categories. (B) Aggregate relative importance of categories stratified by site.

FIG 4

Associations between bacterial taxa and subject variable categories. Networks displaying the top associations between OTUs and variable categories on the forehead (A) and in the mouth (B). Edges are color-coded according to the log-transformed permutation test q-values (see HAllA methods for additional details); yellow hues indicate stronger associations. Node size is proportional to the number of associations between OTUs and variables. Variable codes are explained in Table S1B and the main text.

FIG 5

Forehead microbiome and age. (A) Relative importance of OTUs explaining age as a function of their contribution to the Random Forest model mean square error (MSE). (B) Subject age in relation to the relative abundance of Corynebacterium OTUs deemed most important in the Random Forest age model. (C) Relative abundance-based coexclusion of the age-predictive corynebacterial OTUs. (D) Spearman correlations between forehead OTUs (those present in at least 25% of the samples) and the skin aging variables Wrinkles (VAWAF) and PigmentedSpots (VASAF). The color bar represents correlation values, and the crosses represent significant associations (FDR < 0.05).

FIG 6

Oligotyping analysis. (A) Distribution of Corynebacterium oligotypes, sorted by overall median relative abundance. (B) Relative abundance among age classes of oligotypes tentatively assigned to corynebacterial OTU3 and OTU10.