Unique dermal bacterial signature differentiates atopic dermatitis skin from healthy

Abstract

Gaining a deeper understanding of the variation in skin microbiota across habitats and layers provides critical insights into the complex host-microbial interactions that drive inflammatory skin diseases. Our study investigated dermal versus epidermal microbiota in lesional and non-lesional skin of 37 adult atopic dermatitis (AD) patients and 37 healthy controls. Skin biopsies were partitioned into epidermal and dermal compartments, while serial tape strips collected the superficial epidermis. Bacterial communities were analyzed by cultivation, matrix-assisted laser desorption ionization time-of-flight mass spectrometry, confocal laser scanning microscopy, and metagenomic sequencing. We found that the effects of AD were evident across skin layers. The natural variation between skin layers and habitats diminishes in AD-affected skin, intensifying the impact of the microenvironment and host factors. A remarkably distinct dermal bacterial community was discovered among AD patients, being more conserved and providing a clearer difference between skin habitats, while the epidermis varied substantially. Importantly, comparisons between AD patients and controls revealed more genera differed when studying the dermal samples than the epidermal ones. Staphylococcus, Corynebacterium, and Cutibacterium genera differed with AD status across all samples, but Prevotella and Mitsuokella only differed in the dermis. In the dry and moist dermis, this translated into 14 and 61 gene pathways significantly varying with AD status, including many related to the biosynthesis of menaquinones (vitamin K2). These results suggest dermal sampling would allow for the role of the skin microbiome within AD pathogenesis to be better resolved since these communities are simpler and less prone to environmental contamination.

Importance: This study sheds light on the profound impact of skin microbiota's complex composition and distribution in atopic dermatitis (AD). The distinctive bacterial profile and activity, especially within the dermal skin compartment, vividly mirrored the cutaneous conditions in this inflamed microenvironment. The striking similarity in bacterial communities across different skin habitats in atopic skin underscores the high influence of atopic dermatitis-the genetic predisposition to an amplified immune response. This finding suggests that the dermal bacterial profile could be a valuable tool for longitudinally monitoring changes during the disease's relapsing phases, allowing for a precise categorization of patients into specific AD endotypes. Broadening the focus throughout the entire eczema-affected skin paves the way for treatments capable of modulating dermal biological factors, offering more effective management of AD. By further centering the interest in host-microbial interactions, we can refine personalized treatments, ultimately improving the lives of millions suffering from atopic dermatitis.

Keywords: atopic dermatitis; core microbiome; cutaneous community; dermal microbiota; microbial functionality; skin microbiota.

Fig 1

Schematic of participants, locations, samples, and analysis and metadata of participants in dot plots. Samples (a) were collected at three different habitats—dry (d), moist (m), and sebaceous (s) from the cheek, elbow pit, and back of hand, respectively. Fifteen consecutive tape strips (b) were collected for confocal laser scanning microscopy (CLSM) (tapes 1, 6, and 15), cultivation, and MALDI-TOF (pools 2–5, 7–10, and 11–14). Additionally (c), four biopsies, two from each elbow pit and back of hand, were collected for CLSM and metagenomics, respectively. The latter were divided into epidermal and dermal compartments before DNA extraction. (d) The distribution of age and sex is shown for the two groups of participants, atopic dermatitis (AD) and healthy (HC). Each dot represents one participant. The healthy participants were sex- and age-matched to the included AD patients. The average age of the participants was 34 years. (e) The total severity score (TSS) of eczema based on lichenification, excoriation, erythema, crusting/oozing, dryness/scaling, and edema (score of 0–3) is shown. Each dot represents the total TSS of the sampled area. Eczema categories: non-lesional = 0–1; mild = 2–5; moderate = 6–9; and severe = 10–18. TSS is shown with dotted lines.

Fig 2

Visualization of bacterial aggregates and scattered single cells. Large bacterial aggregates (~1,500 µm in diameter) were detected on tape strips attached to the collected corneocytes in (a) AD moist skin habitat and (b) healthy sebaceous epidermal skin. Bacteria were stained blue by DAPI (DNA stain) and visualized using CLSM. Bacterial aggregates were detected in the stratum corneum of (c) AD and (d) healthy moist skin habitats and in (e) moderate lesional dry skin habitat. The aggregates detected in AD and healthy skin (white circles) sections were mostly <10 µm in diameter no matter the AD status. In the lesional skin, (e) bacterial aggregates were occasionally scattered deeper in the epidermis due to barrier abnormalities and structural changes. The paraffin-embedded and sectioned skin biopsies were stained by a universal bacterial Texas Red-conjugated PNA-FISH probe (mRNA stain) and DAPI. Human nuclei were stained blue by DAPI, while erythrocytes appeared yellow, and the surrounding tissue was green due to autofluorescence. The dual staining of the skin sections visualizes the bacterial metabolic activity (red = active, blue = inactive), showing more active bacterial aggregates in (e) lesional skin.

Fig 3

Specific species and genera composition by cultivation and 16S amplicon sequencing. (a) A bar chart shows the most common cultivated species in all skin groups for AD lesional, (black) non-lesional (gray), and healthy skin (white). The statistical difference (χ-square) is shown above the bars. (b) Box plots visualizing significant differences in relative abundance/ASV richness of abundant genera between AD patients and healthy controls by 16S rRNA amplicon sequencing. Sample types: De = dry epidermal, Dd = dry dermal, Me = moist epidermal, and Md = moist dermal compartment. Asterisks denote significant differences. Pseudomonas showed significant differences in ASV richness within the Me samples but was removed for ease of reading.

Fig 4

Bacterial specificity in skin compartments in AD and healthy controls. Heat trees of ASVs differing between the bacterial community of AD (orange) and healthy (blue) within the four skin compartments: (a) the dry epidermis, (b) dry dermis, (c) moist epidermis, and (d) moist dermis. Gray nodes are equally represented in both comparing groups. After correction for multiple comparisons, no ASVs statistically varied in relative abundance between AD and healthy, while many ASVs remained significantly different within the dermal compartment.

Fig 5

Significant variation in gene functions of the bacterial communities in AD and healthy controls. (a) A total of 27 pathways significantly differed in two or more skin groups. These 27 pathways were plotted, with their change in relative abundance associated with AD mapped (orange is AD; blue is controls). Pathways consistently differed between skin groups; however, the dermal and/or moist environments best differentiated AD skin from healthy controls. (b) Ten pathways significantly differed in all three skin groups (none varied in the dry epidermis). Nine of the 10 pathways were consistently higher in AD patients than healthy controls.