The skin microbiome: current perspectives and future challenges

Abstract

Complex communities of bacteria, fungi, and viruses thrive on our skin. The composition of these communities depends on skin characteristics, such as sebaceous gland concentration, moisture content, and temperature, as well as on host genetics and exogenous environmental factors. Recent metagenomic studies have uncovered a surprising diversity within these ecosystems and have fostered a new view of commensal organisms as playing a much larger role in immune modulation and epithelial health than previously expected. Understanding microbe-host interactions and discovering the factors that drive microbial colonization will help us understand the pathogenesis of skin diseases and develop new promicrobial and antimicrobial therapeutics.

Introduction

Beginning with van Leeuwenhoek’s invention of the microscope in the 17^th century, studies have linked microbes to human disease by uncovering direct, one-to-one relationships between pathogens and skin pathologies. Seminal discoveries include human papillomavirus (HPV) as a cause of squamous cell cancer and Treponema pallidum as the cause of syphilis. More recently, metagenomic advances have allowed us to examine not just one pathogen at a time but thousands of different microbes simultaneously. With these techniques, scientists have uncovered surprisingly diverse and complex microbial communities thriving on the epithelial surfaces of every individual. These communities influence human physiology, immunity, and disease in ways that we are now just beginning to appreciate.

An estimated 1 million bacteria, with hundreds of distinct species, inhabit each square centimeter of skin¹. Many studies have suggested that microbes may contribute even to noninfectious pathologies, such as atopic dermatitis, psoriasis, rosacea, and acne though recent molecular studies are beginning to explain the complex relationship between host and microorganism^2–6. These studies have established a new paradigm for how microbes cause disease, where not just pathogens but also imbalances in the commensal ecosystem cause skin pathology. Whether this imbalance is primary or secondarily caused by changes in host skin and immunity and how this imbalance potentiates epithelial dysfunction, immune dysregulation, or overgrowth of pathogenic microbes are new questions on the research frontier that will impact how we understand and treat skin diseases.

Recent reviews have comprehensively summarized the work to date on the skin microbiome^7–10. This review will briefly describe representative studies of the skin microbiome but will focus primarily on the current gaps in research, relevant clinical questions, and potential methods for addressing these questions.

What is metagenomics?

Historically, characterizing cutaneous microbes involved culturing skin swabs or biopsies. However, less than 1% of bacterial species can be cultivated with standard lab conditions, and many that do grow are competed out by faster-growing organisms¹¹. Consequently, easily cultivated bacteria or fungi, such as Staphylococcus or Malassezia species, were overrepresented in early microbial surveys. Recent advances in DNA amplification and sequencing technology can now bypass the culture steps and allow for more complete, unbiased views of skin microbiota and their genetic content, collectively called the “microbiome” (for glossary, see Table 1).

Table 1.

Glossary

Term	Definition
metagenomics	culture-free, genomic analysis of microbes by direct extraction and cloning of DNA from a particular ecosystem, such as the skin1–3
microbiome	original definition: the "ecological community of commensal, symbiotic, and pathogenic microorganisms that literally share our body space"4
	common usage: aggregate gene content within a microbial ecosystem5
16S rRNA gene	ribosomal gene conserved across bacteria with conserved regions used for PCR amplification and variable regions used for taxonomic classification6
MINE	maximal information-based nonparametric exploration
	a group of statistical methods to find and characterize associations in large datasets with many variables7
metatranscriptomics	culture-free analysis of total RNA (both mRNA and rRNA) isolated from a microbial ecosystem8

^1.References Handelsman J. Metagenomics: application of genomics to uncultured microorganisms. Microbiol Mol Biol Rev 2004;68:669-85.

^2.Riesenfeld CS, Schloss PD , Handelsman J. Metagenomics: genomic analysis of microbial communities. Annu Rev Genet 2004;38:525-52.

^3.Handelsman J, Rondon MR, Brady SF, Clardy J , Goodman RM. Molecular biological access to the chemistry of unknown soil microbes: a new frontier for natural products. Chem Biol 1998;5:R245-9.

^4.Lederberg J MA. 'Ome Sweet 'Omics - a genealogical treasury of words. Scientist 2001;15.

^5.A framework for human microbiome research. Nature 2012;486:215-21.

^6.Turnbaugh PJ, Ley RE, Hamady M, Fraser-Liggett CM, Knight R , Gordon JI. The human microbiome project. Nature 2007;449:804-10.

^7.Reshef DN, Reshef YA, Finucane HK, Grossman SR, McVean G, Turnbaugh PJ et al. Detecting novel associations in large data sets. Science 2011;334:1518-24.

^8.Urich T, Lanzen A, Qi J, Huson DH, Schleper C , Schuster SC. Simultaneous assessment of soil microbial community structure and function through analysis of the meta-transcriptome. PLoS One 2008;3:e2527.

The culture-free, sequence-based method of analyzing any collection of microorganisms, such as skin microbiota, can be referred to as "metagenomics"¹². In analyzing bacterial microbiomes, this method most often involves amplifying the 16S ribosomal RNA (16S rRNA) gene by PCR directly from skin samples (Fig. 1A)^{13, 14}. The 16S rRNA gene exists in all bacteria and archaea but not in eukaryotes. It contains both conserved regions that serve as binding sites for PCR primers and variable regions for taxonomic classification after high-throughput sequencing of the PCR products (Fig. 1B)^{15, 16}. Sequences that are more than 97% identical can often be classified within one species. Within one species, sequence variations are assumed to be due to intra-species strain variations. Also, the number of sequences counted within one species represents the relative abundance of that species in the original skin sample. Thus, this metagenomic approach gives a comprehensive picture of the bacterial community by providing both identification and relative abundances of all present species (Fig. 2).

Figure 1. Metagenomics is a culture-free method to assess skin microbiota.

(A) DNA is purified directly from a skin swab or biopsy. This DNA contains a mixture of genomic DNA from skin and microbial cells. PCR is used to amplify all bacterial DNA with primers that anneal to the conserved region of the 16S rRNA gene. Then, these PCR amplicons are sequenced. Finally, sequences can phylogenetically classified to give the species identities within the microbiome and sequences can be counted to give relative abundances of each species. (B) An alignment of the 16S rRNA gene between S. aureus and S. epidermidis downloaded from NCBI and aligned via Geneious (http://www.geneious.com/). Blue lines show nucleotides that differ between the two species. Inset shows an example of the specific sequence differences.

Figure 2. Composition of a single metagenome.

(A) Phylogenetic tree of an example metagenome downloaded from MG-RAST (data from Fierer et al²⁶). The number of sequences in the metagenome that correspond to each phylogenetic category is listed. For example, 3790 sequences making up 62% of the metagenome's sequences were found to be Actinobacteria by similarity to references sequences. (B) Pie chart showing microbial composition within the same example metagenome. Chart was generated using Krona on the MG-RAST website (http://metagenomics.anl.gov/).

The normal microbiome on human skin

In 2007, the National Institutes of Health (NIH) launched the Human Microbiome Project to survey microbial content across 242 healthy adults, develop a reference catalog of microbial genome sequences, and understand how specific habitats in the gut, genitourinary system, and skin contribute to health and disease states^{14, 17–19}. Recently, results from the Human Microbiome Project were published that describe their metagenomic methods and the publicly available databases of whole genome and 16S rRNA gene sequences¹⁸. This work and other studies in the past decade have characterized the skin microbiome of healthy volunteers and its variation across different spatial niches, individuals, and time (Table 2).

Table 2.

Summary of microbiome studies

Author	Disease studied	Skin site	Number patients
				Method	Results
Frank et al 20031	normal	outer ear canal	24	RFLP; Sanger sequencing	45 spp; microbial community complexity greater in males and older individuals; species composition correlated with consanguinity but not with household association
Dekio et al 20052	normal	forehead	5	16S rRNA gene sequencing	32 OTUs; 62% sequences P. acnes; 9 species previously not known to live on skin
Gao et al 20073	normal	volar forearm	6 (4 resampled 8–10 mos later)	16S rRNA gene sequencing	182 OTUs; 19% sequences Proteobacteria, 51% Actinobacteria, 24% Firmicutes; 63% of sequences common to all subjects; 54% common to both time points; 50–77% identity between R and L arms
Grice et al 20084	normal	antecubital fossa	6 (4 resampled 8–10 mos later)	skin swab, scrape, or punch biopsy; 16S rRNA gene sequencing	113 OTUs; swab, scrape, and punch gave similar results; >90% Proteobacteria (Pseudomonas spp predominant)
Fierer et al 20085	normal	palm	51	16S rRNA gene sequencing	4,742 OTUs; 32% sequences Propionibacterium, 17% Streptococcus, 8% Staphylococcus; 17% identity between R and L hands; 13% identity between individuals; greater diversity in female hands
Grice et al 20096	normal	20 skin sites	10 (5 resampled 4–6 mos later)	16S rRNA gene sequencing	Similar habitats (moist, sebaceous, or dry) had similar microbial compositions; Outer ear and nares were most stable over time; Popliteal fossae, arms, buttocks least stable over time
Costello et al 20097	normal	18 skin sites	9 (resampled 4 times in 3 mos)	16S rRNA gene sequencing	Same habitats are similar between individuals; Inoculation of habitat with bacteria from another site did not significantly change microbiota over time
Dominguez-Bello et al 20108	normal newborn	forearm, forehead	10 newborns, 9 mothers	16S rRNA gene sequencing	Newborns are homogenously colonized; Microbial composition dependent on delivery mode (C-section versus vaginal delivery)
Capone et al 20119	normal infant	volar forearm, buttock, forehead	31 infants age 3 to 52 wks; 5 mothers	16S rRNA gene sequencing	Diversity "evenness" increases with age; Staph and Strep spp decrease in relative abundance with age. Bacilli, Clostridia, Actinobacteria are most frequent classes in infant skin.
Oh et al 201210	normal child and adult	nares, volar forearm, antecubital fossa, popliteal fossa,	28 individuals, age 2–40 yrs	16S rRNA gene sequencing	Microbial diversity of nares increases with sexual maturity. Dominant phyla across all skin sites differed globally between children and adults. S. aureus was overrepresented in nares of younger children and correlated to presence at other skin sites.
Bek-Thomsen et al 200811	acne	facial folicle, cheek	5 pts (not on therapy), 3 normal	16S rRNA gene sequencing	30 OTUs; Healthy follicles contained only P. acnes; Acne follicles had 53–92% P. acnes but also had other bacteria
Price et al 200912	chronic wounds	wound	24 pts	16S rRNA gene sequencing from wound cultures	Recent antibiotics changes wound microbiota, increases Pseudomonadaceae; Diabetic wounds have increased Streptococcaceae
Gontcharova et al 201013	diabetic ulcers	wound and contralateral normal skin	23 wound, 28 normal	16S rRNA gene sequencing	Wounds less diverse than intact skin; Wounds have more Corynebacteriaceae, Streptococcaceae, and anaerobes; Intact skin and wound on same individual do not have more related microbiomes
Paulino et al 200614	psoriasis	plaque, forearm	3 pts, 5 normal	18s rRNA gene sequencing	No significant difference in Malassezia populations in healthy versus psoriatic skin
Gao et al 200815	psoriasis	plaque	6 pts	16S rRNA gene sequencing	Psoriatic plaques have more Firmicutes (39% vs 24% in normal skin)
Fahlen et al 201216	psoriasis	plaque biopsy	10 pts, 12 normal	16S rRNA gene sequencing	652 OTUs; Psoriatic plaques have more Streptococcus, similar levels of Firmicutes
Sugita et al 200117	AD	scalp, back, nares	32 pts (on topical steroids), 18 healthy	IGS gene sequencing	AD pts have different composition of Malasseiza species with more M. globosa, M. restricta, and M. furfur
Sugita et al 200418	AD		13 AD, 12 normal	IGS gene sequencing	Different strains of M. restricta in AD versus normal skin
Dekio et al 200719	AD	face	13 AD, 10 normal	16S rRNA RFLP analysis	18 species; Stenotophomonas maltophilia most common colonizer of AD skin; S. aureus not seen in any samples
Kong et al 201220	AD	antecubital fossa, popliteal fossa, forearm, nare	12 AD, 11 healthy	16S rRNA gene sequencing	AD flares characterized by 65% increase in Staphylococcus spp, predominantly S. aureus and S. epidermidis; AD flares while on intermittent treatment only 15% slight increase in Staphylococcus colonization; Complete treatment of flares re-diversified microbiome

Abbreviations: AD = atopic dermatitis, OTU = operational taxonomic unit, pts = patients, mos = months, spp = species

^1.References Frank DN, Spiegelman GB, Davis W, Wagner E, Lyons E, Pace NR. Culture-independent molecular analysis of microbial constituents of the healthy human outer ear. Journal of clinical microbiology 2003;41:295-303.

^2.Dekio I, Hayashi H, Sakamoto M, Kitahara M, Nishikawa T, Suematsu M et al. Detection of potentially novel bacterial components of the human skin microbiota using culture-independent molecular profiling. Journal of medical microbiology 2005;54:1231-8.

^3.Gao Z, Tseng CH, Pei Z , Blaser MJ. Molecular analysis of human forearm superficial skin bacterial biota. Proceedings of the National Academy of Sciences of the United States of America 2007;104:2927-32.

^4.Grice EA, Kong HH, Renaud G, Young AC, Bouffard GG, Blakesley RW et al. A diversity profile of the human skin microbiota. Genome research 2008;18:1043-50.

^5.Fierer N, Hamady M, Lauber CL , Knight R. The influence of sex, handedness, and washing on the diversity of hand surface bacteria. Proceedings of the National Academy of Sciences of the United States of America 2008;105:17994-9.

^6.Grice EA, Kong HH, Conlan S, Deming CB, Davis J, Young AC et al. Topographical and temporal diversity of the human skin microbiome. Science 2009;324:1190-2.

^7.Costello EK, Lauber CL, Hamady M, Fierer N, Gordon JI , Knight R. Bacterial community variation in human body habitats across space and time. Science 2009;326:1694-7.

^8.Dominguez-Bello MG, Costello EK, Contreras M, Magris M, Hidalgo G, Fierer N et al. Delivery mode shapes the acquisition and structure of the initial microbiota across multiple body habitats in newborns. Proceedings of the National Academy of Sciences of the United States of America 2010;107:11971-5.

^9.Capone KA, Dowd SE, Stamatas GN, Nikolovski J. Diversity of the human skin microbiome early in life. The Journal of investigative dermatology 2011;131:2026-32.

^10.Oh J, Conlan S, Polley E, Segre JA, Kong HH. Shifts in human skin and nares microbiota of healthy children and adults. Genome medicine 2012;4:77.

^11.Bek-Thomsen M, Lomholt HB, Kilian M. Acne is not associated with yet-uncultured bacteria. Journal of clinical microbiology 2008;46:3355-60.

^12.Price LB, Liu CM, Melendez JH, Frankel YM, Engelthaler D, Aziz M et al. Community analysis of chronic wound bacteria using 16S rRNA gene-based pyrosequencing: impact of diabetes and antibiotics on chronic wound microbiota. PloS one 2009;4:e6462.

^13.Gontcharova V, Youn E, Sun Y, Wolcott RD, Dowd SE. A comparison of bacterial composition in diabetic ulcers and contralateral intact skin. The open microbiology journal 2010;4:8-19.

^14.Paulino LC, Tseng CH, Strober BE, Blaser MJ. Molecular analysis of fungal microbiota in samples from healthy human skin and psoriatic lesions. Journal of clinical microbiology 2006;44:2933-41.

^15.Gao Z, Tseng CH, Strober BE, Pei Z, Blaser MJ. Substantial alterations of the cutaneous bacterial biota in psoriatic lesions. PloS one 2008;3:e2719.

^16.Fahlen A, Engstrand L, Baker BS, Powles A, Fry L. Comparison of bacterial microbiota in skin biopsies from normal and psoriatic skin. Archives of dermatological research 2012;304:15-22.

^17.Sugita T, Suto H, Unno T, Tsuboi R, Ogawa H, Shinoda T et al. Molecular analysis of Malassezia microflora on the skin of atopic dermatitis patients and healthy subjects. Journal of clinical microbiology 2001;39:3486-90.

^18.Sugita T, Tajima M, Amaya M, Tsuboi R, Nishikawa A. Genotype analysis of Malassezia restricta as the major cutaneous flora in patients with atopic dermatitis and healthy subjects. Microbiology and immunology 2004;48:755-9.

^19.Dekio I, Sakamoto M, Hayashi H, Amagai M, Suematsu M, Benno Y. Characterization of skin microbiota in patients with atopic dermatitis and in normal subjects using 16S rRNA gene-based comprehensive analysis. Journal of medical microbiology 2007;56:1675-83.

^20.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. Genome research 2012;22:850-9.

In utero, fetal skin is sterile, but minutes after birth, colonization begins to occur^20–22. Newborns are first homogenously colonized with a similar, low-diversity microbiome over all skin sites^{20, 22}. As infants contact environmental microbiota and as different areas of the skin develop distinct moisture, temperature, and glandular characteristics, individual skin habitats arise with divergent, increasingly diverse microbiota²². These habitats then continue to transform with puberty, aging, and environmental exposures^23–27. Metagenomic studies using 16S rRNA sequencing in adults show that the vast majority of skin bacteria as well as gut flora fall into four phyla: Actinobacteria, Firmicutes, Bacteroidetes, and Proteobacteria, but within these phyla exist thousands of distinct species^{1, 26, 28–37}. A survey of the palm microbiome, for instance, found 4,742 distinct species in 51 healthy subjects, with an average of 158 species coexisting on a single palm²⁶.

Surveys of microbiomes over 20 different skin sites show that similar habitats, such as the axillae and the popliteal fossae, have similar microbial compositions (Fig. 3)^{31, 38}. For instance, in all individuals, Propionibacterium species dominate sebaceous areas like the forehead, retroauricular crease, and back, while Staphylococcus and Corynebacterium species dominate moist areas, such as the axillae (Fig. 3). Surprisingly, abundant Gram-negative organisms, previously thought to colonize the skin rarely as gastrointestinal contaminants, were found in the microbiomes of dry skin habitats, such as the forearm or leg.

Figure 3. Microbiome composition on normal human skin.

Sebaceous (blue text), moist (orange text), and dry (green text) habitats are labeled anatomically. Microbial composition differs among the habitats (pie charts at right). The four major phyla are shown: Actinobacteria, Firmicutes, Proteobacteri, and Bacteroidetes. Within these phyla, the three most abundant genera are also shown: Propionibacterium, Corynebacterium, and Staphylococcus. Figure is compiled with data pooled from many metagenomes, from Grice et al³¹. Figure is adapted from Figure 3 in Grice et al⁹ with permission from Nature Publishing Group.

In addition to differing species compositions, each habitat also has its own characteristic level of microbial diversity and temporal fluctuation. For example, antecubital fossae had the highest variance in species composition between subjects, called beta diversity, but each single antecubital fossa had less alpha diversity, or fewer unique species within one habitat when compared to other sites¹⁹. Different skin sites also have different levels of temporal variability. Partially occluded sites, such as the inguinal crease, had more stable bacterial communities over time³¹, while dryer and more exposed skin sites, such as the palm, had higher diversity and more temporal fluctuation³⁸. Characterization of skin habitats by indices such as alpha diversity, beta diversity, and temporal volatility provides information about community structure and can be a quantitative method to follow changes in the skin microbiome after antibiotics, pathogen arrival, and other perturbations.

Consistent with the idea of ecological niches, transplanting microbes from one habitat to another, such as from the tongue to the forehead, caused only a transient presence of tongue microbiota on the forehead with eventual return to a forehead microbiome³⁸. Individual genetics and environmental exposures also contribute to microbiome composition, as contralateral habitats within an individual are more similar than the same habitat across different individuals^{1, 31, 38}. Additionally, within one species of bacteria, strain-level genotypic differences exist in subsets of the populations, potentially correlating to the genetic or immune characteristics of host individuals¹⁹.

Although metagenomics studies using 16S rRNA gene sequencing have revolutionized our understanding of the healthy skin microbiome, many questions need to be addressed. A recent study showed that the nares, antecubital fossa, volar forearm, and popliteal fossa of children differ globally from the same sites in adults in terms of bacterial composition³⁹ (Table 2). For example, S. aureus was more abundant in the nares of children, and this was significantly correlated to S. aureus colonization at other skin sites³⁹. Continued investigation of skin microbiome composition in a variety of age and ethnic groups may help elucidate why certain populations are more susceptible to certain pathologies and the host or environmental factors that determine the composition of skin ecosystems.

In addition to the abundant Propionibacterium, Staphyloccocus, and Corynebacterium species, most species in the skin microbiome each make up less than 1% of the total flora in any particular habitat. These minority species are not well studied and many were not previously known to colonize the skin, but low abundance species could nonetheless be linchpins of the skin ecosystem. Metagenomic studies of soil ecosystems have shown that several low abundance fungal species are actually highly active in essential decomposition processes⁴⁰. Therefore, it is possible that low abundance skin microbes also exert large influences over abundant species, such as S. epidermidis, or pathogenic species, such as S. aureus. One way to detect relationships between pairs of species in a microbiome is to use maximal information-based nonparametric exploration (MINE) statistics. This statistical tool was recently developed and has been applied to a variety of large datasets, including the gut microbiome⁴¹. Application of MINE to skin microbiome data could hint at which pairs of bacterial species are functionally symbiotic or antagonistic and how disruptions in a few species could change the ecosystem as a whole.

Another method to study low abundance species in the skin microbiome is metatranscriptomics, which has been used to study soil microbiomes^{40, 42}. All published studies surveying skin microbiomes have used a DNA-centered, genomic approach. By contrast, in metatranscriptomics, RNA, not DNA, is purified from a skin sample before sequencing. Since the cell itself has already amplified the RNA, this approach can better detect low abundance organisms. Additionally, transcriptome data capture metabolic activity and can reveal whether a low abundance species contributes proportionally more to the ecosystem. Furthermore, since RNA is much less stable than DNA, the meta-transcriptome would only identify microorganisms that are alive, providing a more accurate snapshot in time than metagenomics. However, one technical challenge to this approach is the limitation on skin biopsy size compared to a soil sample, which makes isolating enough RNA more difficult. Therefore, this metatranscriptomics approach may be more applicable when single-molecule DNA sequencing can be performed in a more inexpensive, high-throughput manner.

The microbiome in atopic dermatitis

One frequently studied disease using metagenomics is atopic dermatitis (AD). Although AD is noninfectious, flares may relate to changes in cutaneous microbes. AD is a chronic, relapsing disorder that affects approximately 15% of children in the United States. Many hypotheses have been invoked for the pathogenesis of AD, including a deficiency in the epithelial barrier protein filaggrin, colonization by S. aureus, and immune hypersensitivity^43–47. Empirically effective treatments for AD include antibiotics, steroids, and dilute bleach baths⁴⁸. These are thought to work by decreasing bacterial load and inhibiting a dysfunctional, exuberant immune response to skin flora.

Using culture methods, S. aureus colonization and infection have been commonly associated with AD⁴⁹. Consistent with this, a metagenomic study showed that Staphylococcus species increased from 35% to 90% of the microbiome during flares, but surprisingly, both S. aureus and S. epidermidis increased⁵⁰. Thus, microbiome data suggest that understanding how S. aureus affects AD will require understanding S. aureus fluctuations as part of a larger, complex ecosystem. S. epidermidis can produce molecules that selectively inhibit S. aureus⁵¹, arguing that S. epidermidis may be antagonistic to S. aureus. However, in the gut, pathogenic species can more easily colonize when closely related commensal species are also abundant⁵², suggesting that Staphylococcus species may be mutualistic. Given the differing data above, in the case of AD skin, it is still unclear whether S. aureus and S. epidermidis mutually enhance each other's colonization or whether S. epidermidis increases as an antagonistic response to an increasing S. aureus population. In addition to the obvious changes in S. aureus and S. epidermidis abundance, many unrelated, non-staphylococcal species also appear to change in abundance during an AD flare⁵⁰. Future research should examine whether a change in host skin first triggers changes in species composition, thus allowing for Staphylococcus overgrowth, or if Staphylococcus overgrowth is a primary event that then forces other species to change in abundance.

These questions might be further investigated in mouse models of AD, such as the NC/Nga mouse, which develops disease that is clinically and histologically similar to AD after exposure to environmental aeroallergens^{53, 54}. Importantly, understanding how S. aureus relates to microbiome fluctuations as a whole may reveal novel treatments of AD flares such as rebalancing and re-diversifying the skin microbiome rather than eliminating S. aureus or bacterial burden on the skin. Lessons learned from AD might also inform our understanding of other skin pathologies, such as psoriasis, acne, and chronic wounds, which may also be related to microbiome imbalances.

Microbiome studies similar to those in AD have been performed in patients with psoriasis^{5, 55, 56}, chronic wounds^57–59, or acne⁶⁰ (Table 2). In chronic wounds, the microbiome was found to be less diverse than that of healthy skin but no consensus microbiome was found, even among wounds of the same etiology^{57–59, 61}. In contrast, the follicular microbiome in acne was more diverse than that of healthy follicles, which are colonized almost exclusively by P. acnes⁶⁰. And in psoriasis, there is a lack of consensus in how and if the microbiome of psoriatic plaques differs from that of normal skin^{5, 55, 56}. Metagenomic studies with more detailed stratification based on patients’ clinical status and treatment regimens may help elucidate the clinical significance of these findings.

Antibiotics and the microbiome

A major gap in our current understanding is how current therapies affect the microbiome. Many dermatological treatments are bactericidal or immunosuppressive and may have unexpected effects on the microbiome. In the gut, antibiotics were found to cause not only a transient loss in bacterial diversity but also a long-term loss of microbiome members beyond the direct antibiotic targets^62–64. Even though vancomycin targets only Gram-positive bacteria, Gram-negative populations were depleted after vancomycin treatment⁶⁴. This effect on off-target microbes likely occurs due to indirect relationships between bacterial species that are forged through ecosystem-wide processes, such as metabolite exchange and waste product removal⁶⁵.

Furthermore, after cessation of antibiotic treatment and even after restoration of bacterial density in the gut, the long-term changes in microbial community composition facilitate colonization by pathogens, such as vancomycin-resistant Enterococcus, which then potentiates bloodstream invasion⁶⁶. Therefore, using bactericidal treatments like antibiotics in AD or UV light in psoriasis may have wide-reaching, unknown effects on the microbiome and disease recurrence. Currently, the data on probiotic treatments for skin diseases, such as atopic dermatitis remain controversial. A meta-analysis of seven Cochrane and non-Cochrane reviews showed no clear evidence that interventions such as probiotics, maternal antigen avoidance, and different antigen-avoidance diets reduced the incidence of atopic dermatitis⁶⁷. Although pooled data showed a reduction in eczema incidence with exclusive breastfeeding for at least six months and with maternal probiotic supplementation, these data were based on small trials⁶⁷. Additionally, these trials focused on the modulating the gut microbiome to affect skin health. Future investigation into treatments for microbe-related skin pathologies could be directed toward probiotic regimens that directly modulate the skin microbiome.

Metagenomics to investigate cutaneous infections

Metagenomic studies have provided insights into AD, psoriasis, acne, and chronic wounds. These diseases are noninfectious but can be influenced by shifts and imbalances in skin microbiota. Organisms that cause cutaneous infections can also be studied via metagenomics, which could be particularly useful in those infections associated with a wide range of clinical features and wide geographic and host variability. One such organism is Staphylococcus aureus, a major source of hospital- and community-acquired infections. Its manifestations range from asymptomatic nasal carriage to impetigo, enterotoxin-mediated desquamation, severe necrotizing pneumonia, and septicemia. In addition to a wide range of virulence and toxin-producing capabilities, S. aureus also exhibits variable antibiotic susceptibility, including methicillin and vancomycin resistance. Its widespread pathogenicity and increasing antibiotic resistance coupled with declining treatment options makes S. aureus an important pathogen to study from a patient safety and public health perspective^68–71.

Thus far, 14 strains of S. aureus have been fully sequenced, with many more partially sequenced^72–77. Additionally, DNA microarrays have been developed for genome comparisons between strains of S. aureus^78–82. Studies using whole genome sequencing and DNA microarrays show that virulence and antibiotic resistance are associated with both host-specific and lineage-specific factors^{72, 80, 83} and are encoded in many different ways, including point mutations or small inserts in certain genes, large mobile genetic elements composed of many virulence genes that travel together^{78, 84, 85}, and conjugative plasmids from unrelated species, such as vancomycin-resistant Enterococcus (VRE)⁸⁶.

The study of vancomycin-resistant S. aureus (VRSA) provides an example of how genomic studies can characterize the emergence and epidemiology of antibiotic-resistant strains to identify future therapeutic targets. Since its emergence, all VRSA isolates have been found to be strains within the lineage CC5 and resistance seems to arise from acquisition of the plasmid Tn1546 from VRE during the course of each infection rather than spread of VRSA between individuals^86–89. Recently, a comparative study of 12 whole VRSA genomes revealed that CC5 strains have several genetic features not present in other S. aureus lineages, which could promote acquisition of plasmids from other bacterial species while also impairing host immune function⁹⁰.

In addition to techniques using whole genomes, methods such as multi-locus sequence typing (MLST)⁹¹ and spa typing^{92, 93} have been developed to analyze S. aureus epidemiology across hundreds of samples that differ in clinical, geographic, or host characteristics. Similar to the 16S rRNA metagenomic method described earlier, MLST and spa typing rely on culture-free sample collection, then sequencing of specific regions that exist in all S. aureus strains, and finally classification of strains based on strain-specific alleles. Studies using these approaches have shown that although a large number of S. aureus lineages are present worldwide, only ten predominate and among these, three lineages are rarely associated with methicillin resistance⁹⁴. Further advances in and more widespread use of genomics to study pathogen epidemiology will continue to improve our understanding of how genetic information in pathogens encode pathology and host specificity. Similar studies outside of S. aureus have already shown strong associations between the genotypes of Helicobacter pylori strains and host ethnicity and migratory patterns^95–97.

The microbiome in immune development

As a first line of defense against infection, the skin is both a physical and immunological barrier. Along with the gut, the skin is one of the most heavily immune-surveyed sites in the body. The immune system must not only distinguish between self and other but also perform the more difficult task of distinguishing between beneficial and pathogenic microbes. Since all microbes share similar molecular patterns of lipopolysaccharides and peptidoglycans, it has been a challenge to understand what exactly alerts the immune system to pathogenicity. Evidence now suggests that both skin and gut microbiota play a crucial role in educating and assisting the immune system.

Experiments in germ-free laboratory mice offered the first insights into how crucial microbiota are to immune development. These mice exhibit defective development of gut-associated lymphoid tissue and mesenteric lymph nodes, reduced epithelial expression of immune molecules, and improper T cell differentiation^98–102. Studies have also shown that disturbances in gut microbiota contribute to diseases of immune dysregulation^103–107. Similarly, a recent study has shown that germ-free mice without commensal skin microbes have abnormal cytokine production and cutaneous T cell populations¹⁰⁸. These germ-free mice could not mount an appropriate immune response against intradermal Leishmania major infection; however, immunity could be rescued by allowing Staphyloccocus epidermidis colonization on the skin of germ-free mice¹⁰⁸. These results offer tantalizing evidence that, like the gut, the skin has well-developed immune functions at both the epithelial and associated immune tissue levels. Thus, many of the same principles and lines of investigation in the gut microbiome can be applied to the skin microbiome.

Healthy skin barrier consists of both immune surveillance and epidermal keratinocytes, which produce antimicrobial peptides (AMPs) that contribute to innate immunity (Fig. 4)^109–111. Expression of these AMPs are upregulated by the presence of Propionibacterium species and other Gram-positive bacteria^{112, 113}. In addition to AMPs, sebocytes can produce antimicrobial free fatty acids by hydrolyzing sebum triacylglycerides. This triacylglyceride hydrolysis is also performed by commensal bacterial flora such as P. acnes and S. epidermidis^{114, 115}.

Figure 4. The microbiome and skin immunology.

Viruses, bacteria, and fungi (purple, red, or green dots) cover human skin and its appendages. Keratinocytes produce antimicrobial peptides (AMPs). Sebocytes produce free fatty acids (FFAs). Some commensal microbes also produce AMPs and FFAs, as well as pheno-soluble modulins (PSMs). These molecules all inhibit pathogen colonization. Commensal microbes may additionally inhibit pathogen growth by competition and crowding on the skin surface. The microbiota also interact with immune cells to activate them or modulate their production of pro- and anti-inflammatory cytokines. Backbone skin diagram downloaded from Docstoc (www.docstoc.com).

A large number of Gram-positive commensals, including Lactococcus, Streptococcus, and Staphylococcus species, also produce bactericidal factors de novo¹¹⁶. Peptides called pheno-soluble modulins (PSMs) are produced by S. epidermidis and have selective activity against S. aureus, Group A Streptocooccus, and E. coli but not other S. epidermidis¹¹⁷. Interestingly, S. aureus also produce PSMs, but these have minimal antimicrobial activity and instead induce lysis of neutrophils while S. epidermidis PSMs have bacteria-killing activity but no effect on neutrophils^{118, 119}. Bacterially-produced AMPs do not just play a minor role in innate immunity but are abundant on skin and, in nanomolar amounts, can decrease the survival of pathogens on healthy human skin by 2–3 log fold^{119, 120}.

Microbiota not only activate and assist innate immunity but also influence adaptive immunity, although these interactions are more complex and less well understood. Studies in the gut show that the commensal Bacterioides fragilis elicits anti-inflammatory cytokines, primarily IL-10, and regulatory T cells¹²¹. Other studies on how gut microbiota might modulate the immune system are reviewed elsewhere^{122, 123}. How skin microbiota might influence the innate and adaptive immune system should now be an area of active investigation since so many autoimmune diseases—vitiligo, dermatomyositis, and lupus, to name a few—manifest on the skin even if they are also systemic.

Cancer immunology and the microbiome

Malignancy has been hypothesized to result from a breakdown in immune surveillance and from mutagenic and proliferative environments, such as chronic inflammation. Since the skin microbiome is important for developing a well-functioning immune system and for modulating inflammation, it may also protect against cancers. In support of this hypothesis, studies have shown that workers, such as farmers and waste incinerator workers, who were exposed heavily to environmental microbiota had lower cancer rates^124–126.

Cancer and inflammation are linked in multiple ways. Studies have shown that chronic inflammation and tissue injury increases the risk for cancer, as in the relationship between Helicobacter pylori infection and gastric cancer^{108, 127} or between burns and squamous cell carcinoma¹²⁸. On the other hand, acute inflammation can activate tumor necrosis factor and IL-12-induced antitumor activity, as in the case of Coley's toxin causing sarcoma regression^{129, 130}. Commensal skin bacteria have been shown to both reduce inflammation during wound healing¹³¹ and activate innate immunity and inflammatory cytokines¹³². This begs the question, how do commensal bacteria affect skin inflammation and does this contribute to or protect against malignancy?

Evidence has now been provided that certain microbial components actually do have antitumor activity against bladder and colon cancers, at least in part by heightening immunosurveillance^133–136. Thus far, there are no published studies on how the microbiome influences genesis and propagation of skin cancers. Global metagenomic assessments of microbiome differences between tumor sites and healthy skin may help explain the different propensities for cancer among individuals and skin habitats despite similar sun exposures and may open the door for new therapeutics.

Conclusions

Metagenomics has revolutionized our views about the skin microbiome and its interactions with the host epithelial and immune systems. Metagenomics has also yielded many new questions about what factors drive the composition and fluctuations in skin ecosystems, how changes in the microbiome contribute to disease, and how our medical interventions affect the microbiome. For a wide variety of diseases that relate to perturbations in the epidermis or the immune system, such as melanoma, graft-versus-host-disease, and autoimmune diseases, studying the microbiome may provide a new perspective to pathogenic factors and new therapeutic targets.

Capsule summary.

• -Recent metagenomic studies have revealed that diverse and complex microbial ecosystems inhabit the skin, collectively known as the skin microbiome.
• -This review summarizes recent studies characterizing the skin microbiome and highlights current gaps in research.
• -Understanding how the skin microbiome interacts with the host immune system and with pathogens could pave the way to new antimicrobial and promicrobial therapeutics for a wide array of diseases, including atopic dermatitis, psoriasis, chronic wounds, and cancer.

Abbreviations

16S rRNA

16S ribosomal RNA

OTU

operational taxonomic unit

NIH

National Institutes of Health

MINE

maximal information-based nonparametric exploration

AD

dermatitis

AMPs

antimicrobial peptides

PSMs

pheno-soluble modulins