Bacterial landscape of human skin: seeing the forest for the trees

Authors

  • Oleg A. Alexeyev

    Corresponding author
    1. Department of Medical Biosciences/Pathology, Umeå University, Umeå, Sweden
    • Correspondence: Oleg A. Alexeyev, Department of Pathology, Umeå University Hospital, S-90185 Umeå, Sweden, Tel.: +46-90-7852841, Fax: +46-90-139853, e-mail: oleg.alexeyev@medbio.umu.se

    Search for more papers by this author

Abstract

Skin harbours large communities of colonizing bacteria. The same bacterial species can exist in different physiological states: viable, dormant, non-viable. Each physiological state can have a different impact on skin health and disease. Various analytical methodologies target different physiological states of bacteria, and this must be borne in mind while interpreting microbiological tests and drawing conclusions about possible cause–effect relationships.

Introduction

Skin serves as a protective barrier and harbours large communities of colonizing bacteria. Recent advances in DNA sequencing methodology have shed new light on human skin microbiome [1-3]. Sequence-based approaches have identified a greater diversity of cutaneous bacteria than studies using traditional cultivation techniques; 205 bacterial genera have been reported [1]. While much attention has been paid to emerging new information about the microbiome diversity, the most relevant clinical question that remains to be answered is how many viable, metabolically active bacteria eventually colonize the skin? Like any ecological system, the skin bacterial community will likely be represented by live, dormant and dead cells, with all components contributing to the stability of the system but having different impacts on disease and health. In this viewpoint, I will discuss major physiological states of bacteria, their possible contribution to the skin well-being as well as current methodologies used to study the skin bacterial landscape.

Bacteria exist in different physiological states

A simplified two-valued logic system is often useful to pronounce a biological organism, including humans, either live or dead. In order to apply a similar logic to microorganisms, we need to define the vital signs for bacteria [4]. What makes it possible to pronounce the microorganism alive (viable) or dead (non-viable)? This simple question is of paramount importance in skin microbiology as it governs antibacterial therapy and links bacteria to various dermatological disorders. Classical microbiology equates viability with culturability [4]. However, it is well recognized that a substantial number of bacteria are not culturable despite being viable. The bacterial cells that have reversibly lost the ability to proliferate have been referred to as being ‘dormant’. Dormancy is commonplace within both plant and animal kingdoms, and it is therefore not surprising to observe it in bacteria; arguably, the most diverse and frequent life form. The term ‘reversibility’ is a core feature of dormancy as opposed to truly dead (non-viable) bacteria (Table 1). Most ecosystems in nature are unstable, exposing bacteria to a variety of stresses that can be unfavourable for their growth [5]. To survive in stressful conditions, bacteria may enter a dormant or resting state and resume growth once conditions become more favourable [6]. An extracellular bacterial DNA left after bacterial membrane degradation reflects the final step in an individual bacterium life cycle.

Table 1. Physiological states of bacteria
Physiological statePhenotype
Viable (high metabolic state)Capable to proliferate and form colonies on culture media (if available)
Dormant (low metabolic state)Temporarily lost or have a markedly diminished capability to proliferate and form colonies on culture media
Non-viable (dead)Permanently lost the capability to proliferate and grow on culture media.

Pathogenic potentials of viable, dormant and non-viable bacteria are likely to be different

The pathogenic potential of a given bacterial species is determined by genotype and manifested by phenotype and is beyond the scope of this essay. Instead, I will focus on the relationship between the physiological state of a bacterial cell and virulence.

Viable bacteria

Arguably, the best-studied bacterial form in any clinical discipline, including dermatology, since Robert Koch devised reliable and reproducible techniques for bacterial culture, is the viable cell. It is only through the isolation of individual bacterial species in pure culture that a comprehensive characterization of physiological properties and a full assessment of virulence potential may be undertaken [7]. Detection of bacterial mRNA is not a reliable marker of viability [8]. It is through culturing that we implicate viable bacteria in the course of several skin diseases, including Propionibacterium acnes in acne vulgaris and Staphylococcus aureus in atopic dermatitis. Moreover, viable bacteria can be beneficial to humans by providing defense against pathogenic bacteria [9]. The skin microbiota have an autonomous role in controlling the local inflammatory milieu [10, 11] and tuning resident T lymphocyte function [12], in stimulating the adaptive immune response [13]. Bacteria produce antimicrobial peptides themselves [14, 15] as well as promote expression of antimicrobial peptides by keratinocytes [16-18].

Dormant bacteria

Bacteria existing in a state in which they are alive but no longer replicating are referred to as dormant bacteria. Microbial species from all domains have the ability to enter the dormant state [19]. A range of environmental factors can cause growing bacteria to sense that their surrounding environment is incapable of sustaining continued growth. These include nutrient starvation or limitation, toxic chemical concentrations, changes in temperature, pressure or pH. Dormant bacteria have a notable reduction in metabolism, to the point of absolute dormancy in some cases [6]. The metabolism reduction can manifest in little or no cell wall synthesis [20]. Dormancy is believed to be an important mechanism for bacterial persistence and is exemplified by Mycobacterium tuberculosis infection [21]. The entrance of cells into a dormant, persistent state is largely responsible for the multidrug tolerance of infections [22]. This tolerance, referred to as physiological resistance, is dependent upon growth arrest resulting in inhibition of cell wall synthesis or translation, which are the common antibiotic targets. The tolerance is different from resistance where the acquisition of specific mutations renders the bacterium insensitive to these antibiotics arises [23]. As mentioned above, the dormancy state is fully reversible, and bacterial cells can exit dormancy when they detect permissive environmental conditions [6]. Growth-promoting conditions include, but are not limited to, adequate nutritional resources, lowering levels of natural (e.g. lysozyme, antibacterial peptides) and artificial (antibiotics) antimicrobials. Hypothetically, only a fraction of the cells in any ecological system might be active at any time, while the rest are in a dormant state [19]. As gene expression in bacteria is growth rate-dependent [24], dormant cells have low, if any, virulence but re-establish it after exiting the dormant status.

Non-viable bacteria

Bacteria possessing damaged membranes represent non-viable/dead cells. In natural environments, like marine sediments, dead cells represented the most common and important fraction (70–74%) of bacterial assemblages [25]. Pathogenic potential of non-viable cells in skin disorders has not been studied. Experimental infection with heat-treated P. acnes is reported to induce an inflammatory response in the skin [26]. Programmed bacterial cell death is, however, an important aspect of biofilm development contributing to intercellular adhesion [27].

Free bacterial DNA

Free DNA released from dead bacteria represents the final stage in the bacterial life cycle. Bacterial cell death often leads to autolysis and release of cytoplasmic contents, including DNA. Extracellular DNA molecules are present in most terrestrial and aquatic environments. Naturally, competent bacteria are able to take up exogenous DNA and undergo genetic transformation [28]. Through DNA transformation, bacteria can acquire potentially useful genetic information, such as novel metabolic functions, virulence traits or antibiotic resistance. Additionally, extracellular DNA can serve as templates for the repair of DNA damage as well as a nutritional source of carbon and nitrogen [29]. It must be stressed that amplifiable DNA is preserved long after bacterial cell death [30] and can persist in the environment. Short bacterial 16S rDNA fragments from ancestral sources dating back thousands of years have been successfully amplified [31, 32].

Biofilms are an emerging issue in skin microbiology

Biofilm is a microbial aggregate embedded in extracellular matrix with phenotypic characteristics different from individual bacterial species that comprise it. Biofilms may be the default evolutionary mode of microbial growth [33], and their formation protects cells from harmful conditions in the environment and facilitates escaping from host surveillances. Cell-to-cell communication in the biofilm is mediated via quorum sensing enabling adjustment of gene expression and highly orchestrated community behaviour [34]. Biofilms contain different bacterial populations, including actively growing cells, dormant cells, dead cells and extracellular DNA [35, 36]. An important feature of biofilms is their antibiotic resistance despite a high sensitivity of individual bacterial cells. The resistance was earlier attributed to a physical barrier in the form of extracellular matrix. Many studies have, however, shown that the penetration of antibiotics is not limited in bacterial biofilms [37]. The presence of dormant cells is now believed to account for biofilm tolerance to many antimicrobials [22, 38, 39]. Only direct visualization with electron microscopy, immunofluorescence or in situ hybridization may confirm the presence of biofilms in human skin. Chronic wounds [40], rosacea [41], acne vulgaris [42, 43] and hidradenitis suppurativa [44, 45] have all been associated with biofilm formation.

Hypothesis: P. acnes biofilm dormancy model in acne vulgaris

Skin P. acnes may exist as mono-[43], P. acnes/fungal [46] and P. acnes/P. granulosum (Alexeyev, personal communication) biofilms. In this simplified P. acnes biofilm dormancy model, a P. acnes biofilm phenotype is determined by the ratio between metabolically active and dormant P. acnes. A high availability of sebum, a nutritional substrate for P. acnes, leads to an increased proportion of metabolically active bacteria and contributes to a pro-inflammatory phenotype of the P. acnes biofilm (Fig. 1). This may account for acne flares in the adolescent population characterized by increased hormone/sebum production. A decreasing production of sebum associated with either ageing or iatrogenic blockage by isotretinoin may favour a dormant phenotype of P. acnes biofilm. The latter will be mostly comprised of low active bacteria causing little or no inflammation. According to this model, antibiotic treatment will only target metabolically active but not dormant bacteria. In P. aeruginosa biofilms, the response to antibiotics is indeed dependent on the metabolic state of individual populations [47]. Moreover, exposure to antibiotics itself promotes switching to dormancy [48]. Once the biofilm is switched to a dormant phenotype, the continued administration of antibiotic only prevents emergence of metabolically active bacteria while dormant cells are unaffected. Moreover, the development of mutations associated with genuine antibiotic resistance is favoured. Once antibiotics are withdrawn, the dormant P. acnes cells re-establish the bacterial population. Interestingly, the presence of hydroxyl radicals promotes antibiotic-mediated killing of dormant cells [49]. Whether benzoyl peroxide has the same effect while combined with antibiotics for acne treatment is unknown. It must be stressed that the proposed model considers only follicular P. acnes biofilms which are inaccessible with swab, scrape and cyanoacrylate gel biopsy [50]. Moreover, specific P. acnes clonal lineages may be associated with acne [51, 52]. The presence of different P. acnes phylotypes in the same biofilm [43] may play an important role as well. Whether a pro-inflammatory P. acnes biofilm phenotype can be linked to acne vulgaris, while a dormant biofilm phylotype is more characteristic for asymptomatic colonization remains to be studied. The distribution of skin P. acnes biofilms is focal [43, 46], so the different phenotypes can be present in the same patient.

Figure 1.

Propionibacterium acnes biofilm dormancy model in acne vulgaris. (a) A dormant biofilm phenotype, (b) A pro-inflammatory biofilm phenotype. Metabolically active P. acnes are depicted in red and dormant in blue.

Identification of different physiological states of bacteria requires different methodologies

Culture

Cultivation has represented the gold standard for the identification of microorganisms, and we see a steady increase in the number of cultured bacteria. In 2003, 26 of the approximately 52 identifiable phyla within the domain Bacteria had cultivated representatives [53]. In 2009, there were estimated 61 bacterial phyla, of which 31 had no cultivable representatives [54]. It is still believed, however, that only 1% of excitable bacteria are culturable. The uncultured bacteria fall into two broad categories: (i) bacterial groups with no cultivated representatives (called yet-to-be-cultivated cells), and (ii) bacteria belonging to groups that have been previously cultivated in the laboratory but whose cells are in a state in which they are alive but no longer replicating (for example, dormant cells) [55].

16S rRNA detection

This analysis takes advantage of the universal presence of the 16S ribosomal RNA gene in all prokaryotes. The 16S rDNA contains highly conserved regions, which facilitate PCR, whereas hypervariable regions can be used for phylogenetic categorization [3, 56]. The 16S rDNA analysis offers high sensitivity and detects bacteria difficult or even impossible to culture. Sanger, next generation, and whole genome short-gun sequencing strategies can be used for 16S rDNA analysis [57]. The clinical significance of organisms identified solely by 16S rDNA gene sequencing is, however, unclear. Live, dormant and dead bacterial cells as well as extra-cellular DNA will be targeted by analysis. Another inherited bias is related to the inability to assess the metabolic activity of bacteria.

Fluorescent in situ hybridization and immunofluorescence microscopy

Fluorescent in situ hybridization (FISH) combines the precision of oligonucleotide-binding with microscopy, thus enabling visualization of individual cells within diseased tissue [58]. Limitations associated with FISH are (i) the required prior sequence knowledge in order to design the oligonucleotide probe, (ii) the inability to detect slowly growing or metabolically inactive (dormant) cells, that is, cells with low ribosome content. Given the availability of monoclonal as well as polyclonal antibodies, the direct visualization of bacteria with immunofluorescence microscopy represents another methodology to detect bacteria. In addition, unspecific fluorescent stains are widely used for quantification of bacteria [59]. It is of paramount importance to take into account what type of a physiological state of bacteria is detected by a given analytical method before establishing any cause–effect link in dermatology.

Microbial sampling is a neglected bias in skin microbiology

Skin bacteria are collected for further testing with different sampling methods: swab, scrape, cyanoacrylate gel biopsy and needle biopsy. These techniques target different skin structures, and anatomical considerations must be taken into account when interpreting microbiological data [50] (Table 2). While the sampling of superficial and intrastratum corneum bacterial populations is quite straightforward, the sampling of hair follicle populations is more troublesome. Failure to take anatomical considerations into account may lead to the erroneous assumption that follicular bacterial populations are targeted as exemplified by the recent study of P. acnes in acne vulgaris [60]. Their usage of tape stripping for hair follicle sampling in acne vulgaris is misleading as various amounts of superficial and intrastratum corneum microbial populations will have been sampled. In addition, bacteria may reside in a deeper, anaerobic part of hair follicle [43, 46], which is inaccessible with swab, scrape, cyanoacrylate gel biopsy or tape stripping. Moreover, the usage of adhesive tapes will have sampled both acne-affected and normal follicles which coexist and therefore severely compromised their conclusions [61].

Table 2. Overview of skin populations targeted by different sampling methodologiesa
Bacterial populationsSampling method
SwabScrapeCyanoacrylateBiopsy
  1. a

    Adapted from [50].

  2. b

    Likely removed during pretreatment with antiseptics.

Superficial stratum corneum+++b
Intrastratum corneum+++
Follicle infundibulum++
Lower hair follicle+

Future clinical perspectives

All ecological systems in nature are believed to be in a stable equilibrium. At the same time, evidence points out that all living biological systems, including microbial communities, are non-linear, complex interactions and operate within Chaos theory [62]. Like any forest is comprised of blossoming trees, snags and coarse woody debris, the skin microbiota will likely be represented by continually changing populations of viable, dormant and dead bacteria surrounded by extracellular bacterial DNA. Skin topo-graphy, host and environmental factors will all have profound effects of the skin flora [2]. Skin surface, stratum corneum and hair follicles will likely have their own microbiota [50]. A pioneering work of Grice et al. [1] has unfolded an extreme diversity of skin microbiome. The next step is to separate living bacterial cells from ‘molecular ghosts’ predicted by comparative sequence analysis or fingerprinting [63]. Bacteria with high metabolic rates are more likely to play a pathogenic or beneficial role. Examining the ratio of 16S rRNA to rRNA genes (rDNA) of individual bacterial taxa has been successfully used to provide estimates of bacterial metabolic activities [64]. Assessing microbial metabolic and functional pathways in microbial communities will further advance our understanding of skin microbiota [65]. We are facing a formidable but, at the same time, an exciting challenge of learning what the army of microbes does on our skins. Keeping some bacteria happy and the others at bay could be the key to keeping our skin soft, supple and healthy [66].

Acknowledgements

The author thanks Anne Eady and Andrew McDowell for critical reading of the manuscript.

Conflict of interests

The author has declared no conflicting interests.