Content » Vol 98, Issue 5

Clinical Report

Propionibacterium acnes Abundance Correlates Inversely with Staphylococcus aureus: Data from Atopic Dermatitis Skin Microbiome

Wojciech Francuzik1, Kristin Franke1, Ralf R. Schumann2, Guido Heine1 and Margitta Worm1

1Department of Dermatology, Venereology and Allergology, and 2Institute for Microbiology and Hygiene, Charité Universitätsmedizin Berlin, Berlin, Germany

ABSTRACT

The microbiome may influence disease severity in atopic dermatitis. The skin of atopic dermatitis patients and healthy individuals was sampled in a standardized manner and the microbial composition analysed using next-generation sequencing. Optical density measurements were used to investigate bacterial growth under defined conditions in vitro. Lesional skin from patients with atopic dermatitis had a higher abundance of Staphylococcus aureus and reduced quantities of Propionibacterium acnes and Lawsonella clevelandensis compared with non-lesional skin. The abundance of P. acnes correlated negatively with that of S. aureus (ρ= –0.6501, p < 0.0001). Fermentation products of P. acnes inhibited the growth of S. aureus and S. epidermidis. Serum from patients with atopic dermatitis inhibited the growth of S. aureus to a greater extent than did serum from healthy individuals. These results suggest that selective modification of the skin microbiome could potentially be used as a therapeutic strategy in atopic dermatitis.

Key words: P. acnes; atopic dermatitis; S. aureus; microbiota, high-throughput nucleotide sequencing; skin; eczema.

Accepted Jan 29, 2018; Epub ahead of print Jan 30, 2018

Acta Derm Venereol 2018; XX: XX–XX.

Corr: Margitta Worm, Department of Dermatology, Venereology and Allergology, Charité Universitätsmedizin Berlin, Charitéplatz 1, DE-10117, Berlin, Germany. E-mail: margitta.worm@charite.de

INTRODUCTION

Atopic dermatitis (AD) is a chronic, pruritic inflammatory skin disease with an estimated prevalence of 10–20% among children in Western countries and 2–3% among adults (1). A genetically determined defect in epidermal function is thought to be the initial mechanism in the pathogenesis of AD (2). This barrier defect results in increased water loss, xerosis and reactivity of immune cells to exogenous antigens, such as pollen and microbial flora. The ensuing immune reaction in the skin leads to pruritus and scratching, which further promote inflammation. Lesional overgrowth with Staphylococcus aureus has been described in several reports. S. aureus is a well-known pathogen that can increase disease severity (3) and has been found to be more frequent in patients with AD than in healthy individuals (4).

The Human Microbiome Project described the composition of the healthy skin microflora (5), and skin microbiome dysregulation has been reported previously in patients with AD (6, 7). Mechanisms that either promote or depend on changes to the microbiome are currently of high scientific interest. It has been shown that the commensal skin microbiome impacts on local and systemic immunity (8). The lesional overgrowth of S. aureus correlates with AD flare-ups (9). S. epidermidis, which is considered a harmless skin commensal, promotes an innate immune response (10, 11) and, moreover, inhibits the growth of S. aureus and Propionibacterium acnes (12).

The skin microbiome in healthy individuals and patients with AD with lesions in various anatomical sites was studied using an unbiased approach in order to determine whether a common form of microbial dysregulation is present in patients with AD. By analysing how skin bacteria interact with each other, the aim of this study is to elucidate the underlying mechanisms of AD and provide a basis for the development of novel therapeutic strategies for this disease.

METHODS
Patient recruitment and sampling

Ethical approval for this single-centre research study was provided by the Charité Ethical Committee (EA1/194/14). After obtaining written informed consent, patients with AD and healthy individuals (7 in each group) were included in the study. All procedures were performed in accordance with the Declaration of Helsinki (13). The participants were recruited from the outpatient clinic of the Department of Dermatology Venereology and Allergology. The inclusion criteria for the AD group (modified from (9)) comprised a diagnosis of child-onset AD according to the criteria of Hanifin & Rajka (14); abstaining from the use of topical corticosteroids, skin disinfectants or antibiotics for at least 4 weeks; and avoiding skin washing for 12 h prior to sampling. All participants were male, between 18 and 60 years of age, and were included after a thorough examination at the Comprehensive Allergy Center Charité.

The control group was composed of non-atopic, healthy volunteers with no personal or family history of allergic rhinitis, allergic asthma, or atopic dermatitis, with negative standard prick test results, and normal serum IgE concentrations. To allow for naïve microbial skin colonization, volunteers were excluded if they had used skin disinfectants, topical corticosteroids or antibiotics in the 4 weeks prior to sampling.

Sampling was performed according to the protocol by Oh (15). However, we enlarged the sampling area to increase DNA concentrations in the final samples. Catch-All swabs (Epicentre, Madison, WI, USA) were moisturized with DNAse-free water and used in 3 strictly defined locations (antecubital fossa 40 cm2, interscapular region 40 cm2 and retroauricular region 10 cm2). DNA was extracted using the PowerSoil DNA Isolation Kit (MO BIO Laboratories, Inc., Carlsbad, CA, USA), quantified using a BioPhotometer (Eppendorf AG, Hamburg, Germany) and confirmed by conventional PCR (using 341F: CCTACgggAggCAgCAg and 785R: TgACTACAAgggTATCTAATCC primers, TIB MOLBIOL GmbH, Berlin, Germany).

Next-generation sequencing (NGS)

The bacterial 16S ribosomal RNA genes were amplified by PCR using the 341F-785R primers covering the 3rd to 4th hypervariable regions (16). Sequencing was performed on a miSeq Illumina platform (LCG Berlin Lab, Berlin, Germany). The Bioconductor Workflow (17) was used to analyse the microbiome data. Briefly, raw reads were taken of the forward and reverse strands and those that did not meet the quality threshold were filtered out (more than 3 low-quality reads in the forward strands and more than 6 low-quality reads in the reverse strands). Low-quality reads were determined based on the quality scores incorporated into the FASTA files from the Illumina sequencer. The reads were subsequently de-replicated and nearly all sequencing errors were removed using the Divisive Amplicon Denoising Algorithm (dada2) (18). Next, after merging both strands and eliminating chimeras, we obtained a high-quality database and assigned taxonomies to it.

Functional microbiology studies

S. aureus and S. epidermidis were cultured in 96-well plates under standard conditions (200 µl lysogeny broth medium (LB medium, Sigma-Aldrich, Saint Louis, MO, USA) at 37°C in a shaker-incubator, 150 rotations per minute). Microbial growth was measured as the optical density over time (OD600nm MRX II, Dynex, Chantilly, VA, USA). The influence of P. acnes on S. aureus was determined using supernatants from 5 clones of P. acnes (provided by the Institute for Microbiology and Hygiene, Charité Universitätsmedizin Berlin, Germany) cultured for 5 days in the following conditions: LB medium, with or without the addition of 0.01% glycerol, at 37°C in anaerobic conditions (BD Gaspack, New Jersey, USA) in a shaker-incubator at 150 rpm. Colonies were centrifuged at 5,000 rpm for 30 min. S. aureus was then cultured in fresh LB medium with or without the addition of P. acnes supernatant at increasing concentrations (0%, 6.25%, 12.5%, 25%, 50%, and 100%). The impact of propionic acid on the growth of S. aureus was analysed in cultures treated with increasing concentrations of propionic acid (Sigma-Aldrich, Saint Louis, MO, USA), using OD measurements.

Statistical analysis

Statistical analysis was performed using the “phyloseq” (19) and “microbiome” (20) packages for the R Statistical Platform (21). Differences in the relative abundances of specific bacteria were analysed using a Mann–Whitney U test or, if 3 groups were compared, a Kruskal–Wallis test with a Nemenyi post-hoc test and Holm’s correction. Pairwise correlations of microbial abundances were estimated using Spearman’s method. Growth curves were evaluated using the “growthcurver” (22) package by comparing their mean area under the curve (AUC) values using pairwise Mann–Whitney U tests with Holm’s correction to account for multiple comparisons. To measure the diversity of the microbial communities the Shannon Diversity Index was applied.

RESULTS
Sample characteristics

Both groups of patients were comparable regarding age and sex (mean ± standard deviation (SD) age: AD 35 ± 9.64 and control 29 ± 6.08 years). Topical treatment with corticosteroids was stopped 4 weeks prior to sampling in the AD patient group. The median (interquartile range [IQR]) Eczema Area and Severity Index (EASI) of the patients was 29 (12.7), the median (IQR) Dermatology Life Quality Index (DLQI) was 13 (24), and the mean ± SD body surface area (BSA) was 31.9 ± 16.7%.

P. acnes colonization is reduced in patients with atopic dermatitis

The 5 most abundant species (among a total of 56) identified from the 3 anatomical sites were P. acnes, S. epidermidis, S. aureus, P. pseudogenitalium, and a recently described representative of Corynebacterineae: Lawsonella clevelandensis. The bacterial composition was clearly different in AD lesions vs. healthy skin. Much more S. aureus was detected in lesional skin, whereas P. acnes was reduced compared with non-lesional skin (Fig. 1 and Fig. S1). In locations with an overgrowth of S. aureus, an abundance of P. acnes was decreased. However, the microbiome of non-lesional AD skin resembled the microbiome of healthy skin controls (Table I). The microbial diversity of lesional skin did not differ significantly from non-lesional counterpart skin (p > 0.2, Fig. S2a). The overall abundance of bacterial organisms (based on the raw number of 16S gene reads) was highest in the retroauricular region and lowest in the antecubital fossa.


Fig. 1. (a) Ordination plot of the 10 most abundant species sampled from 3 skin locations of 7 patients with atopic dermatitis (AD) and 7 healthy controls. The samples were taken from lesional skin or non-lesional skin of either AD patients or healthy controls. A multidimensional scaling algorithm with Jensen-Shannon Divergence for distance estimation was used. (b) The mean relative abundances of the 10 most abundant species. Less abundant species were grouped together into “others” and included in the graph for completeness.


Table I. Differences in the relative abundances of the 10 most abundant bacterial species

Subsequently, the correlation of bacteria abundance relative to one another was analysed (Fig. 2). The results show that the relative abundances of P. acnes and S. aureus were negatively correlated (ρ = –0.6501, p < 0.0001), which suggested that a lack of P. acnes may promote the growth of S. aureus. Abundance of P. acnes was decreased in the lesional samples irrespective of the sampling area (Fig. S2b). Two positive correlations in abundance were also identified: between M. luteus and S. infantis and between F. magna and C. pseudogenitalium.


Fig. 2. Comparison of the relative abundance correlation of the 10 most prevalent species. Blue (negative) and red (positive) colours correspond to Spearman’s ρ values.

P. acnes inhibits S. aureus growth by producing propionic acid

Based on the above findings, it was assumed that propionic acid, a fermentation product of P. acnes, impacts the growth of S. aureus. Therefore, P. acnes was cultured in the presence of glycerol to generate propionic acid. The supernatants from these cultures were subsequently used to culture S. aureus. Growth inhibition was observed in the supernatant-treated S. aureus cultures (99.96% mean reduction in AUC compared with control, p = 0.0079; Fig. 3 and Fig. S3a). It was hypothesized that this inhibition was caused by an increased concentration of propionic acid, which is the main fermentation product of P. acnes when cultured in the presence of glycerol. To test this hypothesis, S. aureus was cultured in medium containing increasing concentrations of propionic acid (Fig. S3b, right-hand panel). The growth of S. aureus was completely abolished by propionic acid at concentrations as low as 0.002 mol/dm3 (99.98% AUC reduction compared with control). Finally, it was investigated whether the inhibitory effect of propionic acid was also relevant for the growth of S. epidermidis. In this experiment, the results were similar to those observed for S. aureus (Fig. S3b, left-hand panel).


Fig. 3. Fermentation products of P. acnes inhibit the growth of S. aureus. Growth curves of S. aureus incubated with the 6.25% of supernatants from P. acnes cultures (Sup and controls (Sup-). P. acnes was cultured in LB medium in the absence (GLY–, n = 5) or in the presence (GLY+, n = 5) of 0.01% glycerol for 5 days. Subsequently, the cultures were centrifuged at 5,000 rpm for 30 min, and the supernatant was used as the medium to culture S. aureus. Error bars represent 95% confidence intervals.

Serum of patients with atopic dermatitis inhibits growth of S. aureus

During the course of a flare-up, scratching induced by pruritus may lead to the formation of wounds on the skin and to the extravasation of blood. Therefore, we hypothesized that the serum of patients with AD might impact the growth of skin bacteria. In our analysis we showed that serum from patients with AD and from healthy controls inhibited the growth of S. aureus and S. epidermidis in vitro when added to LB medium (Fig. 4). Serum from patients with AD inhibited bacterial growth to a greater extent than did serum from healthy controls (AUC reduction 31.28%, p = 0.0063).


Fig. 4. Serum inhibits the growth of S. aureus in a culture. The growth curves of S. aureus in the presence of serum (2.5% in LB medium) were evaluated by measuring the OD600nm. Vertical bars represent 95% confidence intervals. Asterisk indicates statistical significance. A serum concentration of 2.5% was selected based on dose-response experiments, indicating that this concentration had a negligible impact on the pH of the culture (data not shown).

rRNA gene sequence variants (RSV) indicate multiple entities of the core species

Based on the single-nucleotide variants in amplicon sequences, we identified 5 ribosomal gene sequence variants (RSVs) for S. aureus, 7 for S. epidermidis, 2 for L. clevelandensis. P. acnes and C. pseudogenitalium did not show variants in rRNA gene sequences. The same RSVs were found in samples deriving from the same patients. RSVs of the same species were often positively correlated in their prevalence in samples. However, S. aureus variant 4 did not follow this pattern and correlated positively with the abundance of S. epidermidis variants and C. pseudogenitalium (Fig. S4).

Two RSVs were identified that were nearly identical (406/407 and 405/407 bases) to the bacterium L. clevelandensis strain X1698 (GenBank accession number KM401816). L. clevelandensis was present in 17 (39.53%) samples and in 8 (57.14%) patients. A total of 16 samples (94.12%) were isolated from non-lesional skin.

DISCUSSION

The results of this study confirm the composition patterns of bacterial species in lesional and non-lesional skin from patients with AD, despite the use of different sequencing technologies (9). However, using the novel “dada2” algorithm it was possible to investigate the skin microbial communities beyond the species level and to identify genetic variants of the core bacterial species. Moreover, it was shown that the recently described L. clevelandensis is commonly present on healthy skin. The most prominent difference between lesional and non-lesional skin in AD is the negative correlation between S. aureus and P. acnes. This may be related due to an uncontrolled outgrowth of S. aureus or a decrease in P. acnes abundance on the skin.

Skin infections are significantly more frequent in patients with AD than in patients with other skin diseases (i.e. psoriasis) (23), suggesting that the abundance of bacteria depends on factors that modify the micro-environment. For example, a disrupted epidermal barrier (24) due to dysfunctional filaggrin leads to increased: (i) transepidermal water loss (TEWL), (ii) bacterial colonization, and (iii) contact with allergens. The outside-in penetration of external antigens (25) activates keratinocytes to produce thymic stromal lymphopoietin (TSLP) leading to a Th2-type milieu in the skin of patients with AD (26). As pruritus is one of the central symptoms of AD, a possible explanation for the overgrowth of S. aureus during a flare-up might be intensive scratching. The skin barrier is mechanically altered by scratching, and adhesive matrix molecules (AMMs, e.g. fibronectin, fibrinogen or collagen) are enriched at the lesional skin sites (27, 28). These AMMs bind to the microbial surface components recognizing adhesive matrix molecules (MSCRAMMs) of S. aureus. Although the barrier disruption in AD promotes expression and secretion of antimicrobial peptides (AMPs) after superficial injury (29), the diminished acidity of the skin alters the performance of these AMPs (which reach optimum performance in acidic pH), promoting bacterial growth (30). S. aureus boosts the underlying inflammation through its Toll-like receptor (TLR) 2 ligands. Kaesler and colleagues described that TLR2 ligands can promote AD by interleukin 4 (IL-4)-mediated suppression of IL-10 (31). As a result, despite the antimicrobial activity mediated by the immune system (e.g. through the increased release of IL-22 (32), IL-17A and IL-17F), S. aureus is more prone to adhere to the injured skin regions than other commensal bacteria.

Another previously reported factor influencing the pathogenesis of AD and dysbiosis of the skin microbiota is the decrease in the number and volume of sebaceous glands (33, 34). Sebaceous gland insufficiency is associated with reduced production of glycerol, ceramides and free fatty acids (35). This might: (i) explain the predilection of a given microbial species to a specific skin site; (ii) restrict the growth of P. acnes, and therefore promote the outgrowth of pathogenic bacteria, i.e. S. aureus.; (iii) explain the low water content in the skin and the associated deterioration of the skin barrier seen in patients with AD (36). Kim et al. (37) found a significant negative correlation between sebum production and skin’s pH, both in healthy individuals and patients with acne vulgaris. Decreased colonization with P. acnes may also be responsible for the observed lower prevalence of acne vulgaris in patients with AD (38). The impact of sebaceous glands on the modulation of the skin microbiome equilibrium has been studied previously (39).

Propionic acid released by P. acnes reduces the pH of the skin and therefore increases the activity of skin’s antimicrobial proteins (AMPs), i.e. dermcidin, modifying the microbial composition. This effect is not specific for AD. Shu et al. (40) have proven that the main product of P. acnes glycerol fermentation is indeed propionic acid, which inhibits the growth of S. aureus even when buffered with the appropriate compound to eliminate the influence of a lower pH on the culture (41). Both the decrease in intact filaggrin expression and the lower sebum production promote increased pH levels in the skin of patients with AD, who exhibit higher pH values in healthy and lesional skin compared with healthy controls (42).

Frequent exposition to S. aureus in colonized lesions promotes the emergence of specific IgE and IgG2 antibodies (43). Serum from patients with AD inhibited the growth of S. aureus to a greater extent compared with serum from healthy individuals. This might be due to increased activation of the complement cascade or enhanced antimicrobial activity via S. aureus recognizing immunoglobulins.

We also identified the recently isolated bacterium L. clevelandensis (44) as a rather common inhabitant of healthy skin. Given the fact that there are several reports on this bacterium causing abscesses (45) it might be regarded as an opportunistic pathogen. It was not discovered previously, probably due to demanding growth requirements and long incubation time. However, NGS seems to be an ideal alternative for the identification of bacteria, which are otherwise difficult to culture in vitro.

NGS data analysis using an appropriate bioinformatic analysis can identify single nucleotide variations in the rRNA gene sequences. Although they better represent the true microbial community, the clinical significance of RSVs is unknown, and it might be due to: (i) the exact same genome can have different sequences, because of multiple copies of the target gene; (ii) 2 strains from the same species can have the same sequence in the amplified region; (iii) 2 strains from the same species can have different sequences in the amplified region; (iv) 2 different species can have the same sequence in the amplified region.

In conclusion, the restoration of the balance in the skin microbiome, by topical treatment to specifically reduce S. aureus (46), promoting P. acnes (40), or both, may partially impact the course of AD. Whether a direct microbial supplementation with commensal Propionibacteria, for example, may contribute to the stabilization of the skin microbiome and prevent flares is an interesting target for future therapeutic approaches.

Study limitations

A high variance in bacterial communities was found within all individuals, demonstrating the high sensitivity of the NGS technique used. Because the skin carries relatively few microbes compared with the gut, we sought to reduce bias by standardized sampling. Nevertheless, we derived our conclusions by comparing relative abundances and not absolute values.

No differences were observed in the alpha diversity of bacterial communities from the AD and healthy skin. This might be dependent on the fact that in each case (in healthy, non-lesional or lesional samples) the communities were dominated by a single bacterial species, whereby the others were less abundant. Our results were based on the Divisive Amplicon Denoising Algorithm, whereby singletons are not present due to better quality control of the reads. Therefore, the resulting microbiome is less diverse, but resembles the true composition more adequately compared with clustering based methods (18).

The low number of samples studied could limit the ability to discover differences in abundance of the less prevalent representatives of the skin microbiota and prevented us from drawing conclusions concerning RSVs of the core species.

Reference sequences of the 16S rRNA gene in regions V3–V4 of S. aureus and S. simiae are identical (accession numbers: NZ_AEUN01000485.1, NZ_CP009828.1), which prevented us from differentiating between these 2 bacteria.

ACKNOWLEDGEMENTS

The authors thank Dennis Ernst and Inga Wyro?lak for outstanding technical assistance, and Benjamin J. Callahan and Paul J. McMurdie for fruitful discussion and constant development of essential open source software.

Funding. WF received the Charité Scholarship for the duration of this project.

The authors have no conflicts of interest to declare.

REFERENCES
  1. Weidinger S, Novak N. Atopic dermatitis. Lancet 2016; 387: 1109–1122.
    View article    Google Scholar
  2. Palmer CNA, Irvine AD, Terron-Kwiatkowski A, Zhao Y, Liao H, Lee SP, et al. Common loss-of-function variants of the epidermal barrier protein filaggrin are a major predisposing factor for atopic dermatitis. Nat Genet 2006; 38: 441–446.
    View article    Google Scholar
  3. Zollner, Wichelhaus, Hartung, Mallinckrodt V, Wagner, Brade, et al. Colonization with superantigen-producing staphylococcus aureus is associated with increased severity of atopic dermatitis. Clin Exp Allergy 2000; 30: 994–1000.
    View article    Google Scholar
  4. Gong J, Lin L, Lin T, Hao F, Zeng F, Bi Z, et al. Skin colonization by staphylococcus aureus in patients with eczema and atopic dermatitis and relevant combined topical therapy: a double-blind multicentre randomized controlled trial. Br J Dermatol 2006; 155: 680–687.
    View article    Google Scholar
  5. Huttenhower C, Gevers D, Knight R, Abubucker S, Badger JH, Chinwalla AT, et al. Structure, function and diversity of the healthy human microbiome. Nature 2012; 486: 207–214.
    View article    Google Scholar
  6. Cole C, Kroboth K, Schurch NJ, Sandilands A, Sherstnev A, O’Regan GM, et al. Filaggrin-stratified transcriptomic analysis of pediatric skin identifies mechanistic pathways in patients with atopic dermatitis. J Allergy Clin Immunol 2014; 134: 82–91.
    View article    Google Scholar
  7. Shi B, Bangayan NJ, Curd E, Taylor PA, Gallo RL, Leung DY, et al. The skin microbiome is different in pediatric versus adult atopic dermatitis. J Allergy Clin Immunol 2016; 138: 1233–1236.
    View article    Google Scholar
  8. Naik S, Bouladoux N, Linehan JL, Han S-J, Harrison OJ, Wilhelm C, et al. Commensaldendritic-cell interaction specifies a unique protective skin immune signature. Nature 2015 520: 104–108.
    View article    Google Scholar
  9. Kong HH, Oh J, Deming C, Conlan S, Grice EA, Beatson MA, et al. Temporal shifts in the skin microbiome associated with disease flares and treatment in children with atopic dermatitis. Gen Res 2012; 22: 850–859.
    View article    Google Scholar
  10. Lai Y, Cogen AL, Radek KA, Park HJ, MacLeod DT, Leichtle A, et al. Activation of TLR2 by a small molecule produced by staphylococcus epidermidis increases antimicrobial defense against bacterial skin infections. J Invest Dermatol 2010; 130: 2211–2221.
    View article    Google Scholar
  11. Cogen AL, Yamasaki K, Sanchez KM, Dorschner RA, Lai Y, MacLeod DT, et al. Selective antimicrobial action is provided by phenol-soluble modulins derived from staphylococcus epidermidis, a normal resident of the skin. J Invest Dermatol 2010; 130: 192–200.
    View article    Google Scholar
  12. Wang Y, Kuo S, Shu M, Yu J, Huang S, Dai A, et al. Staphylococcus epidermidis in the human skin microbiome mediates fermentation to inhibit the growth of propionibacterium acnes: implications of probiotics in acne vulgaris. Appl Microbiol Biot 2014; 98: 411–424.
    View article    Google Scholar
  13. General Assembly of the World Medical Association. World Medical Association Declaration of Helsinki: ethical principles for medical research involving human subjects. J Am Coll Dent 2014; 81: 14.
    View article    Google Scholar
  14. Hanifin J, Rajka G. Diagnostic features of atopic eczema. Acta Derm Venereol 1980; Suppl 92: 44–47.
    View article    Google Scholar
  15. Oh J. Human, bacterial and fungal amplicon collection and processing for sequencing. Bio-protocol 2015; 5: DOI: 10.21769/BioProtoc.1477.
    View article    Google Scholar
  16. Klindworth A, Pruesse E, Schweer T, Peplies J, Quast C, Horn M, et al. Evaluation of general 16S ribosomal RNA gene PCR primers for classical and next-generation sequencing-based diversity studies. Nucleic Acids Res 2012; 41: e1–1.
    View article    Google Scholar
  17. Callahan BJ, Sankaran K, Fukuyama JA, McMurdie PJ, Holmes SP. Bioconductor workflow for microbiome data analysis: from raw reads to community analyses. F1000 Res 2016; 5: 1492.
    View article    Google Scholar
  18. Callahan BJ, McMurdie PJ, Rosen MJ, Han AW, Johnson AJA, Holmes SP. DADA2: High-resolution sample inference from illumina amplicon data. Nat Methods 2016; 13: 581–583.
    View article    Google Scholar
  19. McMurdie PJ, Holmes S. Phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data. PLoS ONE 2013;8: e61217.
    View article    Google Scholar
  20. Leo Lahti, Sudarshan Shetty et al. (2017). Tools for microbiome analysis in R. Version 1.1.2. http://microbiome.github.com/microbiome.
    View article    Google Scholar
  21. R Core Team. R: a language and environment for statistical computing. Vienna: R Foundation for Statistical Computing 2016.
    View article    Google Scholar
  22. Sprouffske K, Wagner A. Growthcurver: an R package for obtaining interpretable metrics from microbial growth curves. BMC Bioinformatics 2016; 17: 172.
    View article    Google Scholar
  23. Christophers E, Henseler T. Contrasting disease patterns in psoriasis and atopic dermatitis. Arch Dermatol Res 1987; 279: S48–51.
    View article    Google Scholar
  24. Cork MJ, Danby SG, Vasilopoulos Y, Hadgraft J, Lane ME, Moustafa M, et al. Epidermal barrier dysfunction in atopic dermatitis. J Allergy Clin Immunol 2009; 129: 1892–1908.
    View article    Google Scholar
  25. Irvine AD, McLean WI, Leung DY. Filaggrin mutations associated with skin and allergic diseases. N Engl J Med 2011; 365: 1315–1327.
    View article    Google Scholar
  26. Liu Y-J. Thymic stromal lymphopoietin: master switch for allergic inflammation. J Exp Med 2006; 203: 269–273.
    View article    Google Scholar
  27. Cho S-H, Strickland I, Boguniewicz M, Leung DY. Fibronectin and fibrinogen contribute to the enhanced binding of Staphylococcus aureus to atopic skin. J Allergy Clin Immunol 2001; 108: 269–274.
    View article    Google Scholar
  28. Novak N. Immune mechanisms leading to atopic dermatitis. J Allergy Clin Immunol 2003; 112: S128–139.
    View article    Google Scholar
  29. Harder J, Dressel S, Wittersheim M, Cordes J, Meyer-Hoffert U, Mrowietz U, et al. Enhanced expression and secretion of antimicrobial peptides in atopic dermatitis and after superficial skin injury. J Invest Dermatol 2010; 130: 1355–1364.
    View article    Google Scholar
  30. Ong PY, Ohtake T, Brandt C, Strickland I, Boguniewicz M, Ganz T, et al. Endogenous antimicrobial peptides and skin infections in atopic dermatitis. N Engl J Med 2002; 347: 1151–1160.
    View article    Google Scholar
  31. Kaesler S, Volz T, Skabytska Y, Köberle M, Hein U, Chen K-M, et al. Toll-like receptor 2 ligands promote chronic atopic dermatitis through IL-4-mediated suppression of IL-10. J Allergy Clin Immunol 2014; 134: 92–99.e6.
    View article    Google Scholar
  32. Malhotra N, Yoon J, Leyva-Castillo JM, Galand C, Archer N, Miller LS, et al. IL-22 derived from t cells restricts staphylococcus aureus infection of mechanically injured skin. J Allergy Clin Immunol 2016; 138: 1098–1107.e3.
    View article    Google Scholar
  33. Prose P. Pathologic changes in eczema. J Pediatr 1965; 66: 170–178.
    View article    Google Scholar
  34. Wirth H, Gloor M, Stoika D. Sebaceous glands in uninvolved skin of patients suffering from atopic dermatitis. Arch Dermatol Res 1981; 270: 167–169.
    View article    Google Scholar
  35. Macheleidt O, Sandhoff K, Kaiser HW. Deficiency of epidermal protein-bound ?-hydroxyceramides in atopic dermatitis. J Invest Dermatol 2002; 119: 166–173.
    View article    Google Scholar
  36. Fluhr JW, Mao-Qiang M, Brown BE, Wertz PW, Crumrine D, Sundberg JP, et al. Glycerol regulates stratum corneum hydration in sebaceous gland deficient (asebia) mice. J Invest Dermatol 2003; 120: 728–737.
    View article    Google Scholar
  37. Kim MK, Choi SY, Byun HJ, Huh CH, Park KC, Patel RA, et al. Comparison of sebum secretion, skin type, pH in humans with and without acne. Arch Dermatol Res 2006; 298: 113–119.
    View article    Google Scholar
  38. Larsson, PA, Liden S. Prevalence of skin diseases among adolescents 12–16 years of age. Acta Derm Venereol 1980; 60: 415–423.
    View article    Google Scholar
  39. Nguyen MT, Hanzelmann D, Härtner T, Peschel A, Götz F. Skin-specific unsaturated fatty acids boost the staphylococcus aureus innate immune response. Infect Immun 2015; 84: 205–215.
    View article    Google Scholar
  40. Shu M, Wang Y, Yu J, Kuo S, Coda A, Jiang Y, et al. Fermentation of propionibacterium acnes, a commensal bacterium in the human skin microbiome, as skin probiotics against methicillin-resistant staphylococcus aureus. PLoS ONE 2013; 8: e55380.
    View article    Google Scholar
  41. Wang Y, Dai A, Huang S, Kuo S, Shu M, Tapia C, et al. Propionic acid and its esterified derivative suppress the growth of methicillin-resistant Staphylococcus aureus USA300. Benef Microb 2014; 5: 161–168.
    View article    Google Scholar
  42. Knor T, Meholji?-Fetahovi? A, Mehmedagi? A. Stratum corneum hydration and skin surface pH in patients with atopic dermatitis. Acta Dermatovenereol Croat 2010; 19: 242–247.
    View article    Google Scholar
  43. Baker BS. The role of microorganisms in atopic dermatitis. Clin Exp Immun 2006; 144: 1–9.
    View article    Google Scholar
  44. Bell ME, Patel NB, Metcalfe MG, Tucker T-A, Harrington SM, Bernard KA, et al. Lawsonella clevelandensis gen. nov., sp. nov., a new member of the suborder corynebacterineae isolated from human abscesses. Int J Syst Evol Micr 2016; 66: 2929–2935.
    View article    Google Scholar
  45. Harrington SM, Bell M, Bernard K, Lagacé-Wiens P, Schuetz AN, Hartman B, et al. Novel fastidious, partially acid-fast, anaerobic gram-positive bacillus associated with abscess formation and recovered from multiple medical centers. J Clin Microb 2013; 51: 3903–3907.
    View article    Google Scholar
  46. Tan WP, Suresh S, Tey HL, Chiam LY, Goon AT. A randomized double-blind controlled trial to compare a triclosan-containing emollient with vehicle for the treatment of atopic dermatitis. Clin Exp Dermatol 2010; 35: e109–112.
    View article    Google Scholar
Supplementary content
Figure S1
Figure S2
Figure S3
Figure S4