Simultaneous skin biome and keratinocyte genomic capture reveals microbiome differences by depth of sampling

Capsule Summary

Novel skin tape strip method allows for simultaneous collection of the skin microbiome and underlying host DNA and RNA, and reveals that microbial ecology is dependent on the depth of sampling.

To the Editor,

Atopic dermatitis (AD) is a multifactorial, inflammatory skin condition affecting up to 20% of children, with approximately 60% of patients experiencing disease onset within the first year of life¹. AD varies in clinical presentation and phenotypes are difficult to define. Non-invasive methods to sample the skin are necessary to enable investigations of AD and improve disease phenotyping. Skin tape stripping is a common dermatologic procedure used to collect keratinocytes and other resident skin cells². Current non-invasive methods result in poor keratinocyte nucleic acid yields, requiring pooling of multiple tapes for downstream assays, such as proteomics ^3–5 and transcriptomics⁶. Pooling of tapes helps to increase the yield of material necessary for downstream assays, but it masks key differences that may be present as a function of the depth of sampling. The current study uniquely utilizes dissolvable tape strips, originally designed for hydrographic applications, for simultaneous collection of the skin microbiome, keratinocyte epigenetics, and keratinocyte gene expression in order to characterize the surface biome and the ensuing underlying host skin innate response. Our data demonstrate differences in gene expression and microbial ecology with each individual tape, an observation that would have been masked by tape pooling. Further, our findings highlight that the local environment is dependent on the depth of sampling. These attributes make this methodology a powerful clinical and research tool to enhance our understanding of normal skin and skin diseases.

We surveyed several adhesive tapes (Tegaderm, Transpore and Blenderm surgical tapes, Kroger brand Sheer bandage, D-Squame⁶, and a water-dissolvable tape designed for hydrographic applications, SmartSolve) for usability, tolerability, material cost (Supplementary Table I), and keratinocyte sampling by measuring filaggrin (FLG) by quantitative PCR (qPCR). This study was approved by the Institutional Review Board at Cincinnati Children’s Hospital Medical Center and all subjects provided informed consent. We collected eleven tapes per site (Fig 1A). Some subjects reported discomfort with Tegaderm beyond tape strip number 11, therefore, collections of all tapes were limited to 11 tapes per site. Keratinocytes collected using SmartSolve tape from three healthy adult controls showed the most consistent FLG expression from individual tape strips 8, 9, 10 or 11 compared to other tapes. FLG was detected in 1 out of 4 Bandage replicates, and 3 out 4 Tegaderm and Transpore replicates. FLG expression was not detected in the samples collected from D-Squame discs. (Fig 1B). While D-Squame discs have been used successfully for genome-wide transcriptomics studies, 20 consecutive tapes were pooled for the analyses⁶. We assayed each D-Squame from sequential discs 1–11 for FLG expression and no expression was detected (data not shown). Our data suggests that SmartSolve is the most suitable for a clinical setting, has better tolerability and lower cost, while it allows for single tape epidermal RNA sampling, therefore all subsequent analyses were performed with SmartSolve.

FIG 1. Epidermal marker expression in RNA using non-invasive skin tape stripping in children with atopic dermatitis (AD).

(A) Diagram illustration of the 11 tape strips collected and the downstream applications. (B) Relative FLG expression quantified by qPCR in individual tapes 8–11 collected from healthy control donors with SmartSolve, Tegaderm, Sheer Bandage, Transpore and D-Squame adhesive tapes. Expression of FLG was normalized to 18S. Representative data is shown (N=3). (C) Relative expression of filaggrin (FLG), loricrin (LOR), S100A9, keratin 1 (KRT1), transglutaminase 1 (TG1) and keratin 14 (KRT14) in each of the 11 sequential tape strips from healthy controls and a blank tape as negative control (Ø). Representative data is shown (N=3). (D) Host RNA isolated from lesional and non-lesional tapes 8 or 9 yielded host keratinocyte RNA suitable for quantitative PCR. Error bars denote technical replicates. Data from three subjects are shown (N=400).

In 3 adult healthy controls, we assayed mRNA isolated from each of the 11 sequential tapes for expression of the following markers by qPCR: FLG, loricrin (LOR), S100A9, keratin 1 (KRT1), transglutaminase 1 (TG1), and keratin 14 (KRT14) (Fig 1C) in order to assess the level(s) of the epidermis that may be sampled by each strip. The tapes consistently sampled the stratum corneum and, in some cases, tapes 8–11 may also sample deeper layers of the epidermis (Fig 1C). Although a recent transcriptome study reported difficulties in collecting skin tape samples from non-lesional compared to lesional skin⁶, we were able to consistently and reliably sample the normal skin of healthy adults with a success rate of at least 80%.

In order to confirm the scalability and potential clinical applicability of this new method, we implemented it in a clinical setting in 400 children with AD. We quantified epidermal gene expression (FLG, S100A8, and S100A9) in individual tape strips 8 and 9 taken from both non-lesional and lesional skin of children with AD. Expression of these genes was reliably detected in ≥90% of the subjects. There was variability in expression observed between AD subjects (Fig 1D) illustrating the potential for future studies to explore how expression may be related to disease endotypes.

The role of the commensal skin microbiome in dermal immunity and the pathogenesis of inflammatory skin conditions is increasingly evident⁷. We next determined the relative quantity of human versus microbial DNA present on each tape strip by metagenomic shotgun sequencing (MSS) to identify which tapes would be most optimal to conduct host and microbial analyses. The first three tapes collected yielded the highest levels of microbial DNA (11%, 9%, and 13%, respectively). In contrast, analysis of the DNA from the 11^th tape strip revealed that 98% of the reads mapped to the human host genome while only 2% of the reads mapped to microorganisms (Fig 2A). We next examined whether the microbial ecology was dependent on the relative depth of skin sampling as determined by the sequential tape strip number. Previous studies have shown that the distribution of microbial species is not uniform throughout the stratum corneum ². Further, the depth of S. aureus penetration into the dermis of AD lesional sites is associated with the presence of increased inflammatory cytokines in skin biopsies ^{2, 7}. We subjected genomic DNA from tape strips 1 and 7 taken from non-lesional and lesional sites of randomly selected AD subjects to MSS (Fig 2B). Clear differences in the skin microbial ecology were observed between tapes 1 and 7 in both non-lesional and lesional skin. Among the 200 most frequent microbial species identified, 71% (14/200) of these species had relative abundances that were markedly different between tapes 1 and 7 (median CV between paired tape samples greater than 10%). Cutibacterium acnes (11.1% vs 9.5%, p=0.03), Staphylococcus.sp.HMSC077B09 (0.4% vs 0.8%, p=0.03) are examples of subspecies that have significantly different relative abundances between tape 1 vs. tape 7 even with our limited sample size (Supplementary Table II). These data demonstrate substantial variability in microbial content between individual layers of tape.

FIG 2. Microbial and human DNA as a function of depth of tape strip sampling.

(A) Metagenomic shotgun sequencing results of each tape strip is displayed in a bubble plot where each individual microbial species is displayed in a different color and the size of each individual bubble corresponds to its relative abundance. Percentages of human DNA (light blue) and microbial DNA (multi-colored) for each tape strip are designated. Representative data from healthy subject is shown (N=15). (B) Most abundant bacterial species for tape strips 1 and 7 from non- lesional and lesional skin of atopic dermatitis children by MSS. Data from three randomly selected subjects are shown (N=21). (C) DNA methylation in the LDHC CG1, CG2 and CG3 promoter sites of DNA isolated from tape strip 4 from lesional atopic dermatitis children by pyrosequencing. Data from three randomly selected subjects are shown (N=10).

Studies of whole blood, T cells and B cells have revealed no differences in genome-wide epigenetic patterns between AD cases and controls⁸. However, epigenetic patterns differ in skin biopsies from individuals with AD compared to healthy controls⁸. Since skin biopsies are not feasible in large cohort studies and may not be feasible in clinical office settings, we next determined whether host DNA isolated from skin tapes could be utilized for epigenetic studies. We analyzed the DNA methylation status of three cytosine-phosphate-guanine (CpG) sites on the LDHC promoter, CG1, CG2 and CG3, which are known to be associated with steroid treatment response in asthmatic children⁹, on lesional tape strip 4. These data demonstrate that the individual tape strips yielded sufficient host keratinocyte DNA to perform epigenetic studies. There was significant variability in the methylation levels observed at the three CpG sites across three randomly selected AD subjects, with one having very high methylation levels at one site (subject 1), while others had high methylation levels at all 3 sites (subjects 2 and 3) (Fig 2C). These data demonstrate that there is considerable variation in skin methylation and that this methodology could be utilized to identify epigenetic biomarkers in the skin of children with AD.

We have developed a new non-invasive methodology to sample the skin. Our data demonstrate that this methodology simultaneously samples the surface skin biome as well as the underlying human DNA and RNA, which can be used to characterize host responses to environmental cutaneous exposures. This novel methodology is unique in that it utilizes dissolvable tapes and each single tape strip yields sufficient nucleic acids for metagenomics (biome characterization), keratinocyte epigenetic studies, and RNA expression studies. This is highly relevant as the biome varies by depth of sampling and this method enables characterization of each layer sampled by an individual tape. It is safe, well tolerated, efficient, low-cost, non-invasive. We successfully applied this methodology to a large cohort (N=400) of young children with AD aged 1–2, and therefore, has broad potential in clinical and research settings.

Abbreviations

AD

Atopic Dermatitis

BL+TG

Bacteria lysis (buffer) with 2% Thioglycerol

CpG

Cytosine-phosphate-guanine

LDHC

lactate dehydrogenase C

MPAACH

Mechanisms of Progression of Atopic Dermatitis to Asthma in Children

MSS

Metagenomic Shotgun Sequencing

PCR

Polymerase Chain Reaction