Consumer Safety Considerations of Skin and Oral Microbiome Perturbation

The Human Microbiome in Health and Well-BeingMicrobiotas associated with the oral mucosa and the skin help program the human immune system to recognize pathogens (52, 53), reduce the risk of invasion by undesired organisms (54), and produce vitamins and other metabolites, such as short-chain fatty acids (55). In skin, phenol-soluble modulins (PSMs) and bacteriocins (56) contribute to the ecological and structural maintenance of the niche (54). Commensal skin organisms, such as Staphylococcus epidermidis and C. acnes, use distinct mechanisms to inhibit pathogens and maintain a healthy skin barrier. S. epidermidis produces antimicrobial peptides which can reportedly control the growth of S. aureus (16), as well as serine proteases to inhibit biofilm formation (16); fermentation products, such as succinic acid, that may inhibit the overgrowth of the opportunistic pathogen C. acnes (46); and a unique form of lipoteichoic acid that can inhibit skin inflammation during skin injury (57). C. acnes also has a protective role as a commensal by converting sebum to free fatty acids, which in consequence inhibit colonization of opportunistic pathogens and contribute to the maintenance of an acidic skin pH (46).

In the oral cavity, some streptococci generate hydrogen peroxide, which can inhibit the caries-associated bacterium S. mutans (58). The oral microbiome also has nonantimicrobial functions of importance to health and disease, where nitrate-reducing oral bacteria can convert dietary nitrate into nitrites, which can influence cardiovascular health and blood pressure (59). Nutritional functions of the oral microbiota are delivered by complex communities via cross feeding and syntrophy. For example, streptococci have both glycosidic and endopeptidase activity, while species of Prevotella and Porphyromonas have endopeptidase activity and species of Fusobacterium and Peptostreptococcus have aminopeptidase activity (60). Bearing in mind the roles of the skin and oral microbiome that are currently understood and the fact that other activities remain unknown, the maintenance and protection of the healthy functionality of the microbiome are important considerations when assessing the effect of personal care products.

Microbiome Composition versus FunctionInitiatives such as the Human Microbiome Project (HMP) (13, 53, 55) and other studies (14, 52, 61) have enhanced our understanding of the baseline skin and oral microbial composition, but the search for attributes that define a healthy human microbiome continues. As part of the HMP, where 200 healthy individuals were examined, the core microbiome of different body sites, including saliva, plaque, tongue, and other oral tissues, ranged from 0 to 8 operational taxonomic units (OTUs) when analyzed for percent prevalence of 100%, compared to a higher range of 19 to 75 OTUs when the percentage was lowered to 50% (62). Interpretation of the core microbiome to measure the similarity of samples depends on the taxonomic resolution employed since samples may decrease in apparent similarity when analyzed to the genus or OTU level compared to the phylum level (8, 61, 63). While a specific group of microorganisms may be shared between individuals, interindividual variation may still be considerable at the species level and for the presence of rare microorganisms (8, 14, 61). Care is therefore required when classifying microbiome composition as healthy or otherwise, especially in the absence of species-level classification. This is of particular importance in the oral cavity, where different species within the same genera can have contrasting associations between health and disease.

The functions provided by compositionally different microbiomes can be relatively similar between individuals (55). Exploring which of these general functions are associated with health represents an alternative to the concept of “healthy composition” (64). A proposed functionality-based definition of a “healthy microbiome” involves three functions: those associated with health-related housekeeping functions, human functions, and specialized functions (53). Housekeeping functions involve energy production and the generation of metabolites and other requirements to maintain the microbial community itself; human-associated functions comprise interactions with the host, such as developing and influencing the activity of the immune system; and specialized functions include regulation of the pH in a specific body site. A functional core has been described for metabolic pathways detected in more than 75% of individuals (55). Pathway cores were identified for either multiple or single body sites, reflecting the fact that some core functions are broadly distributed and general to the human host, while others are an adaptation to a specific body site. It should be noted that core functions are not necessarily beneficial to the host. Among site-enriched pathways, nitrate reduction has been identified to be important in the oral cavity (55). These core pathways are generally associated with microbial consortia. Such functional observations may provide further insights when studied across populations and during longer temporal studies with a controlled microbial change. If functional characterization of the human microbiome can be achieved, measuring or predicting the loss of a beneficial function or the introduction of an undesired function could be used as a functional index during consumer safety assurance.

Microbiome Stability as an Indicator of HealthThe variability of the microbiome over time in healthy individuals has been assessed and was found to be low (65). This "temporal stability" has been interpreted as a sign that the community is in equilibrium, regardless of the fact that some microbes may at the same time be changing as a response to disturbances (66). Stability in the classical sense, that is, the ability of the microbiome to remain at equilibrium when exposed to a perturbation (resistance) or to return to it afterwards (resilience), has also been proposed to be a key feature of a healthy microbiome (53). When referring to a microbiota as stable, we include both of these two concepts of stability. Despite their importance for understanding microbial community dynamics and responses to perturbations, long-term longitudinal studies are still rare. However, based on the available evidence, the composition of the human microbiome is relatively stable over time, with the main variation within an individual being between body sites (13), and considerable temporal stability has also been reported for the microbiome in healthy skin. Oh and colleagues (14) generated metagenomic sequence data from longitudinal samples collected over 2 years and reported that bacterial, fungal, and viral communities were largely stable over that time, despite exposure to the external environment. This stability was observed to be site specific, with body sites harboring high microbiome diversity being more variable than low-diversity (sebaceous) body sites. Observations of temporal stability in the skin microbiome have been interpreted as evidence for colonization resistance and used as the basis for clinical studies exploring the skin microbiome in disease states, where compositional changes in the microbiome have been reported. Costello et al. (10) assessed the resilience of the skin microbiota by disinfecting plots on the forehead and left volar (i.e., underside of) forearms of volunteers and then inoculating them with foreign microbiotas (i.e., microbiotas taken from the tongue and skin of other individuals). The microbiotas of forearm plots (n = 16) that had been inoculated with tongue scrapings were more similar to tongue communities than to those normally associated with the forearm in relative abundance between 2 and 8 h after inoculation. However, communities more similar to those normally associated with the forehead developed on forehead plots that had been similarly inoculated with tongue material. It can be inferred, therefore, that for some reason (potentially the presence of sebaceous lipids), the forehead environment exerted a stronger selection pressure than the forearm. Furthermore, following interpersonal and intergender reciprocal swaps of forehead and forearm microbiotas, developing communities resembled those of the recipient rather than those of the donor, demonstrating the importance of the environment and, possibly, the action of endogenous mechanisms for individualization and microbiota perpetuation. The authors hypothesized that the stronger selection at forehead sites was due to sebaceous secretions, which, in contrast to dry sites like the volar forearm, (i) may have been more strongly selective and/or (ii) could have supported the more rapid recolonization from appendageal structures, which is in agreement with the hypothesis outlined above.

The oral microbiota may also remain stable over time in healthy individuals (6), although it is also sufficiently malleable to be beneficially manipulated through hygienic intervention (39). It is, however, important to consider what stability means when referring to a host-associated microbiota. Belstrom and colleagues collected saliva from five volunteers without oral disease every 4 h for 24 h, repeated this 7 days later (67), and profiled the salivary microbiome. While caution is necessary, given the small sample size, the author’s tentative conclusion was that “little or no variation” within salivary microbiomes was observed over time. The oral cavity is a complex environment with various distinct areas, and saliva, often purported to contain microorganisms originating from multiple sites on the mouth, may vary less in terms of microbiome composition than, for example, a tooth surface, where, in individuals following the recommended oral health regime of twice-daily brushing, microbial abundance is very low immediately after cleaning but can exceed 10⁷ bacteria per cm² following regrowth.

Maintaining microbiome stability in healthy individuals ensures that the beneficial microbial functions are maintained (68), so the measurement of microbiome stability and its recovery following disturbance is important in understanding potential risks. While the human microbiome is relatively stable, its composition can be altered both by pathologies, such as gingivitis and dandruff, and by treatment.

Consumer Products Can Alter Microbiome Composition or FunctionThe hypothesis that the skin microbiota, once established, is perpetuated by continuous endogenous inoculation is supported by an investigation by Grice et al. (12) in which the skin microbiota was sampled using swabs and biopsies and profiled by high-throughput sequencing. An attractive explanation is that secretions from sweat glands and the outward migration of differentiating skin cells could transport bacterial cells from within appendageal structures continuously onto the skin surface (as proposed by Kong [69]). Daily hygiene regimens may, however, affect the microbiome, and some routines, such as tooth brushing and hand washing, do this intentionally to control or reduce the risk of oral disease and to reduce the transmission of pathogens, respectively (54, 70). Exposure to antimicrobials through the use of household and personal care products has shown minimal long-term effects on the microbiome. In this respect, two human studies monitored how the use of toothpaste, liquid and bar soap, and dishwashing liquid with and without triclosan perturbed the microbiome. The first study, a crossover control study involving healthy individuals, showed no significant impact on the human oral or gut microbiome composition during 4 months of exposure to the antibacterial compound triclosan (71). A longitudinal survey of the gut microbiota in infants and mothers during the first year following birth also did not show major compositional changes or a loss of microbial diversity (72). It is highly likely that environmental modulation of the skin microbiota has been occurring since the ancient origins of the microbiome for the skin through UV irradiation, friction, and washing and for the oral cavity through diet, friction, and cleaning. In personal care, antiperspirants are used by approximately 50% of the global population and have been shown to reduce the bacterial load in the axilla. Individuals that do not use antiperspirants have been observed to harbor a greater axillary microbiome diversity than individuals that use antiperspirants do (73). For antiperspirant and deodorant users who ceased use of product, an increase in Staphylococcaceae was observed; in comparison, Corynebacterium species dominated in nonusers. In contrast, microbiome diversity was reported to be greater in antiperspirant users than in deodorant users or nonusers. In a separate study of nine cohorts, axillary diversity was similarly found to be greater in antiperspirant (and deodorant) users than in nonusers (74). A recent study on the effect of cosmetic products on the microbiome of facial skin of high- and low-hydration groups indicated that the baseline bacterial diversity was greater in the low-hydration group than in the high-hydration group and that the use of cosmetic products decreased the differences between the two groups (38).

Microbiome IndividualizationEvidence suggests that both environment and host genetics play important roles in determining the composition of individual microbiomes. Salivary microbiome studies in twins indicate that overall microbial abundance and some aspects of the microbial population structure are influenced by heritability (75). With respect to the skin microbiome, Blekhman and colleagues (76) analyzed shotgun metagenomic data from the HMP, collecting data on host genetic variation for 93 individuals. They reported significant associations between host genetics and the microbiome composition for 10 of the 15 sites that they assessed, including the oral cavity and the skin. Thus, as well as extrinsic environmental factors, host genetics appears to play a role in the composition of the oral and skin microbiotas, probably through immunological and other mechanisms. These examples partly explain the variability between individuals observed in microbiome research (8) and highlight the need to separate a significant change from individual variation when assessing specific perturbations.

Extrinsic factors also influence the stability of the microbiome, since activities, such as smoking tobacco, have been shown to influence the composition of oral biofilms (77), suggesting that smoking promotes the acquisition and colonization of pathogenic bacteria. The development of gingivitis and its progression from gingivitis to periodontitis and the promotion of dental plaque biofilm colonization partly depend on the host immune response (78). Gomez and colleagues (79) illustrated the impact of host genetics through a human volunteer study involving a large cohort of monozygotic and dizygotic twin children with and without active caries, with the aim of elucidating the contributions of host genotype and shared environment on the oral microbiomes (supragingival plaque) of children. They observed that the similarity in oral microbiomes was higher between monozygotic twins, regardless of caries state, with certain taxa being identified as highly heritable, but that most of the variation was determined by the specific growth microenvironment. The caries state, however, was not associated with the more highly heritable bacteria, suggesting that lifestyle, diet, and oral hygiene practices might outweigh parental heritability in establishment of a caries-associated microbiome. The more heritable species were detected at a lower abundance with increasing age and sugar consumption.

Risks of Pathogen ColonizationOne of the beneficial activities of the microbiotas of the skin and oral cavity is the protection of the host tissue from pathogens (as summarized in Fig. 1). Perturbation of commensal communities may therefore be a factor contributing to the pathogenesis of certain inflammatory conditions. In some circumstances, overgrowth of commensal microorganisms with pathogenic potential (pathobionts) or colonization by external pathogenic organisms (transients) can cause disease. The ability of transient organisms to colonize is likely to depend on the interactions with the commensals residing at each specific body site. In this respect, microbial communities with more competitive interactions than cooperative interactions are assumed to be more resilient, in the sense that cooperation causes coupling between species involving several species to change at the same time and destabilize the system (80). In the mouth, loss of colonization resistance through antibiotic use can lead to infections by opportunistic pathogens, such as Candida species and S. aureus (as reviewed in references 60 and 81). In this regard, microbial changes that do not increase the opportunity for pathogens to colonize are unlikely to adversely affect the well-being of the host.

• Open in new tab
• Download powerpoint

FIG 1

Assessing the safety of perturbations of the skin and oral microbiome.

The Human Body as a Microbial NicheThe skin and oral cavity present distinct environments, and ecological conditions in situ have a large influence on the compositional differences in microbiota between body sites. Oily, moist, and dry skin sites regulate nutrients and harbor specific microbial taxa (46, 52). The mouth can be broadly divided into different habitats, the gingiva and hard palate, the tongue and throat, and dental plaque, with each one being colonized by a microbiome characteristic of the specific site (60). The microbiota present in the oral cavity forms biofilms by attaching to the different surfaces, which confer spatial structure and provide the conditions required for different organisms to survive within the community (9). The availability of oxygen is one of the drivers of microbiota composition, and in this context, a succession during the formation of dental plaque has been proposed whereby teeth are initially colonized by facultative genera, such as Streptococcus, with a shift to a microbial community better adapted to anaerobic conditions occurring as the biofilm matures. Bacterial succession on the tooth surface can also be strongly influenced by nutrient availability, mechanical stress, and saliva flow (6, 61, 82) and by binding of bacteria to proteins in the salivary pellicle coating the tooth surface (83).

Interactions with the external environment can also drive selection. For example, an increase in sugar intake or a reduction in saliva flow may induce a reduction in pH that allows the expansion of aciduric organisms (82). Loss of moisture, changes in temperature, and exposure to UV radiation can also result in microbiota alteration in the skin (84). Similarly, changes in the spatial structure may also influence the microbial community within a given body site (9, 84).

Microbial Diversity in HealthSeveral indices have been employed to differentiate microbiomes associated with health and disease. Among these, microbial (ecological) diversity is frequently measured. Ecological diversity can be measured as richness (the number of taxa present) and evenness (the abundance of microbial constituents). Although not universally applicable, higher diversity has been associated with health in specific contexts, when considering that more diverse microbes may supply the host with increased functional traits. However, microbial diversity on its own is not an accurate measure for determining disease etiology or health. While reduced microbial diversity has frequently been observed in conditions such as atopic dermatitis and psoriasis (85, 86), this is not always the case; for example, in both psoriatic and unaffected elbows (79), richness has been reported to be the same, while an increase in bacterial diversity due to the rise of species of minor abundance has been observed in gingivitis and periodontitis (63, 87). The measurement of diversity also does not account for interactions among species, and two microbiomes with the same level of diversity may be different. It may therefore be more pertinent to observe the entire community of microbes present and, by extension, how they function, rather than relying on richness alone as a predictor of disease (88).

The Importance of Bacterial AbundanceCompositional studies of the skin and oral microbiomes have suggested that the load or abundance of organisms can be more significant than their presence in the progression of disease. A 65% increase in the proportion of S. aureus in atopic dermatitis sufferers at flare sites and a partial correlation between S. aureus abundance and disease severity have been reported (106). Similarly, S. epidermidis was significantly more abundant during flares than after flares and in controls, although the underlying reasoning for the increase in S. epidermidis was not determined. Several studies have reported an increased C. acnes abundance in individuals with acne than in unaffected volunteers (90). While differences between the absolute numbers of bacteria between individuals with inflammatory acne, papules, and pustules have been reported, there appears to be progressively higher bacterial loads vis-à-vis the severity of the disease (91). The use of quantification methods, such as quantitative PCR, has revealed higher levels of S. mutans and Streptococcus sobrinus in children with caries than in caries-free children (92). In other oral diseases, such as gingivitis, severity is better correlated with the plaque load and maturity than with some specific bacteria (60). It should, however, be borne in mind that NGS is not well-suited to determining differences in bacterial absolute abundance (the quantified genetic or microbial load within a sample), such that two samples with an identical relative abundance (the genetic representation of microbes within a sample ranked against all taxa in the sample) could differ markedly in absolute abundance (93).

Host-Microbiota InteractionsSkin functions as a two-way barrier, which helps to preserve hydration levels and prevent entry of noxious substances into the body. Skin function may be shaped by the commensal organisms. and in this respect, Naik et al. (94) demonstrated that germfree mice had a weakened immune response to the parasite Leishmania major compared to mice raised under specific-pathogen-free conditions. The impaired response in the germfree mice could be rescued by colonization with S. epidermidis (94), implying a role for the microbiota in promoting host immunity. More recent evidence suggests that the microbiota is fundamental to skin structure. Conventionally reared mice showed altered gene expression compared with germfree mice. Meisel et al. (95) reported that 2,820 genes were differentially regulated by microbial colonization; these included genes associated not only with the host immune response but also with epidermal differentiation. Crucially, the expression of 9 genes involved in the epidermal differentiation complex (EDC), a collection of genes involved in terminal differentiation of keratinocytes (reviewed in refrence 89), was regulated by the microbiota. When the skin of conventionally raised mice was compared to that of germfree mice, differences in the balance of proliferation and differentiation were observed. These data support the view that the microbiome may be associated with the development of the skin architecture since the EDC has been implicated in dermatological diseases, such as psoriasis (reviewed in reference 96). Various studies have shown that the microbiota is associated with the outcome of the healing response when wounding breaches the skin barrier. In broken skin, the commensal microorganisms can behave as pathogens and colonization of wound sites can result in the release of microbial metabolites that can further damage host tissues (reviewed in reference 97). It is therefore unsurprising that accelerated wound healing has been observed in the absence of microbiota (98, 99), but it is also the case that the commensal microbiota can produce antimicrobial peptides (AMPs) that can inhibit the invasion of wound sites by pathogens (100). There is also evidence that S. epidermidis can inhibit the uncontrolled inflammation sometimes associated with wounding. Part of the mechanism for this may involve the inhibition of cytokine release by keratinocytes (57).

With respect to beneficial effects, S. epidermidis has been reported to augment tight junction function in keratinocytes (101), where the interaction of keratinocyte monolayers with S. epidermidis increased the transepithelial electrical resistance (a measure of tight junction function) within a short time of exposure to this bacterium. Furthermore, Toll-like receptor (TLR) ligands, such as lipoteichoic acid or peptidoglycan, may augment tight junction function in keratinocyte monolayers (102). These data suggest that skin commensals, like those of the gut, are probably involved in many aspects of epithelial barrier homeostasis.

Metagenomic ProfilingStudies employing both ribosomal profiling and metagenomics have sought to identify microbes linked to either oral or cutaneous disease, whether it is at the community level or that of individual taxa. Several studies have reported changes in the proportion of bacteria on the skin in psoriasis (25, 26, 85). Gao et al. (26) and others (25), for example, reported that Firmicutes were significantly overrepresented in psoriasis lesions compared to uninvolved skin, while the Actinobacteria and Propionibacterium species were reportedly present at a significantly lower relative abundance in psoriatic lesions. Apart from bacteria, the fungal genus Malassezia has also been associated with psoriasis (25, 26, 85, 103, 104). Altered microbial community profiles have also been reported in atopic dermatitis, where an increased proportion of Staphylococcus species, particularly S. aureus and S. epidermidis, was observed during disease flares in comparison to that at baseline or posttreatment and correlated with increased disease severity (19, 105, 106).

In terms of the oral microbiota, changes in microbial composition have long been associated with dental caries and periodontitis. For caries, sequence analysis has confirmed that bacteria other than S. mutans are correlated with active caries (Lactobacillus and Bifidobacterium), and likewise, several taxonomic groups of bacteria are associated with periodontitis (28, 107–110). It is also clear that the etiology of disease also involves a complex interplay between the host and the resident microbial communities that is yet to be fully explored. When applied to the study of psoriasis, such approaches indicate that strain-level features and associated functional variation may be pertinent to disease (111).

This exploration of host-microbe interactions has been hindered by the fact that virulence and pathogenic determinants could be partitioned at the subspecies or strain level. It is well established that intraspecies genomic features lead to phenotypic variability (106, 112–114). Ribosomal genus-based profiling approaches lack strain-level resolution. Several recent computational tools to taxonomically (115–117) and functionally (118, 119) characterize individual members of the microbiome at strain-level resolution in metagenomic data sets have become available.

Profiling of Functional PotentialWhile understanding the community structure of a microbiome and the relationship between specific taxa and health or disease can be informative, knowledge of community function will probably be most useful in understanding the effect of perturbing the microbiome. Shotgun metagenomics provides the potential to access strain-level taxonomic features and the potential functional characteristics of the community, which has, until recently, been computationally challenging. This approach can be used for the investigation of functional traits, although it can only reveal the functional potential of communities. It can also be used to profile viruses, which are not amenable to ribosome-based profiling. The oral microbiome has assessed disease states, such as caries or periodontal disease, compared to healthy controls. Shi et al. (120) and Wang et al. (121) reported that community function around bacterial chemotaxis and cell motility are increased in disease compared to periodontal health. It has also been shown that in periodontal disease there is an increase in metabolic pathway genes associated with fatty acid metabolism (122), as well as an increase in genes associated with the metabolic degradation of nutrients (120) and those required for growth under anaerobic conditions (122). Healthy communities have been shown to exhibit increased functions in the areas of fatty acid biosynthesis, aspartate and homoserine metabolism, membrane transport, and signal transduction. Metagenomic studies of the skin are more difficult due to the low bacterial density and small sample surfaces available (123). Mathieu et al. (124) consider the skin microbiota to be a complete organism, reporting a predominance of catabolic genes and the ability of the skin bacteria to use the sugars, lipids, and iron that are found on human skin. They also found genes related to antibiotic resistance, as well as some linked to acid resistance, clearly a mechanism for tolerance of the natural acidity of the skin. Oh et al. (17) have described a “functional core” of about 30% of the community that can vary depending on the diversity and biogeography of the different skin microenvironments, which drives the functional capacity that is required by that community. For example, dry sites were found to favor functional traits surrounding the citrate cycle, and sebaceous sites showed increased function around glycolysis and ATP/GTP/NADH dehydrogenase I. While these metagenomic approaches provide more than a simple inventory of taxa and provide information on function and health/disease interrelationships, making judgments of community functional traits by reference genome comparison should be undertaken with care. There is a large genomic diversity that is just starting to be understood, for example, the association of only some C. acnes strains with acne vulgaris (116, 123). Further complicating the search for a functional understanding of the microbiome is the identification of new genes from metagenomic analysis approaches that are associated with health or disease but that cannot be assigned to any functional pathway.

Metatranscriptomic AnalysesShotgun transcriptomics can be used to determine the active functions of a microbiome (125), especially as the community composition of a microbiome alone is not necessarily reflective of its active community members (126). This is an emerging research area with fewer data available, and challenges remain, for example, in sampling sufficient mRNA material to enable analysis. However, the transcriptomic profile of a community is dynamic and can easily change in the same biological sample at different times as the microbiome responds continually to changing environmental and host conditions. Metatranscriptomic studies applied to the human microbiome are more limited than metataxonomic/metagenomics surveys.

Metatranscriptomic studies of the skin are more challenging than those of the oral microbiome due to the limitations of the microbial biomass in the sample material. Kang and colleagues (125) analyzed the metatranscriptomics of patients with acne vulgaris versus those of healthy controls. C. acnes was reportedly the most transcriptionally active organism and was predominant in both the healthy and diseased samples. Further analysis of the gene expression profile of C. acnes in the samples identified that the organism’s activity on acne-affected skin was distinct from its activity on healthy skin. Specifically, the vitamin B₁₂ biosynthesis pathway was observed to be significantly downregulated in acne. Additionally, a model of how vitamin B₁₂ modulates the transcriptional and metabolic activities of C. acnes in acne pathogenesis was suggested. The model underlined how shotgun metatranscriptomic approaches can enhance the understanding of disease pathogenesis. One of the limitations of metatranscriptome data is that the final metabolic products generated by a microbial community are not captured (126). In this respect, techniques such as proteomics, metabolomics, and lipidomics can help to have a deeper functional characterization of the microbiome.

Metatranscriptomics has been used in conjunction with metagenomics to investigate saliva from individuals with caries and periodontitis and to compare it with saliva from orally disease-free individuals. Belstrom et al. (67) identified 15 differentially expressed KEGG orthologs (KOs) between samples from patients with periodontitis or caries and samples from orally healthy controls. These included eight carbohydrate metabolism-associated KOs that were downregulated in periodontal disease and two KOs associated with glycan biosynthesis and carbohydrate metabolism that were upregulated in caries. In addition, the same study observed that lipid metabolism was increased in healthy samples compared with dental caries samples and concluded that longitudinal studies may reveal that screening for salivary metabolic gene expression can identify oral diseases preclinically. However, it is also clear that the development of such diagnostics is at a very early stage and that overcoming the very significant differences in complexity between the salivary and plaque microbiomes would be a substantial technical and clinical challenge.

Metabolomic AnalysesMicrobial metabolites can have a direct impact on oral or skin health (e.g., short-chain fatty acids and sulfides in periodontal diseases, organic acids in dental caries), or they can enter and modulate host metabolic processes. As such, metabolite exchange between the microbiome and host represents one mechanism through which these systems communicate. Variation in the bacterial species present can modulate the genetic library of the microbiome, changing its overall functional capacity, its metabolite production, and the downstream impact on host health. However, different species are known to possess similar or even the same metabolic traits. This functional redundancy means that studying composition alone may be insufficient to accurately determine the overall biotransformation capabilities of the microbiome and, therefore, its potential to modify host health. Metabolic profiling (metabolomics/metabonomics) has emerged as a powerful tool for studying the microbiota because it can ascertain the metabolic profile via low-molecular-weight compounds in a sample. These metabolic signatures contain thousands of molecular low-molecular-weight compounds reflecting biochemical events. These include host metabolic processes but also those performed by the resident microbes and products arising from interactions between the two. Studies using metabolomics to directly assess the functional status of the skin microbiota are limited. However, several studies have characterized the skin metabolome in a wider context. These have used a variety of sample types, including skin swabs, hydrogel micropatches (127), punch biopsy specimens, and sweat. In one study analyzing epidermal skin tissue, several bacterium-derived metabolites (128) and bacterial substrates were observed, including p-cresol, a product of bacterial tyrosine metabolism. This demonstrates that these tissue samples can be informative for studying the skin microbiome. Skin surface liquid extracts (sweat) represent another sample type of potential utility. These are complex mixtures of secretions derived from eccrine, apocrine, and/or sebaceous glands (depending on the body location) as well as from the microbiota inhabiting the skin (129). Attempts are being made to optimize and standardize the collection and analysis of sweat, and this may prove to be a useful resource for studying the skin microbiota.

Metabolic profiling of gingival crevicular fluid (GCF) has been used to study the importance of host-bacterium interactions in periodontal disease. Here, the depletion of antioxidants, degradation of host cellular components, and accumulation of bacterial products were seen in the disease state (130, 131). Attempts have been made to integrate salivary bacterial and metabolic data sets to identify metabolic products related to specific bacterial groups (132). Oral biofilms have also been studied by capillary electrophoresis-mass spectrometry (CE-MS)-based metabolomics. This has enabled the central carbon metabolic pathways to be investigated in the oral biofilm. One approach is to measure these pathways in supragingival plaque before and after a glucose rinse. Glucose can be degraded by bacteria to several metabolic products, including acetate, formate, lactate, and succinate. Assessing the metabolic content of this plaque after the rinse provides information on the functional capacity of the biofilm.

Mathematical ModelingOral and skin microbial community dynamics are shaped by three broad factors: the host, the environment, and the community. The human host provides the microenvironment for the community and may alter this environment through hygiene and other behaviors. The genetic makeup of the host also influences the community's microenvironment. The surrounding environment offers a large species pool from which immigration into the local community may take place. Finally, community composition (richness, evenness, and interactions) as well as history (e.g., previous exposure to perturbations) may impact its dynamics.

A community model expresses in mathematical terms how selected factors influence community dynamics. Community models thus allow prediction of the response of the community to short-term (pulse) perturbations and altered conditions (press perturbations). Models can be coarse grained or detailed, describing populations or individuals. A general distinction can be made between phenomenological models that predict community behavior on the basis of immigration and mortality rates, interaction strengths, growth rates, and other parameters and metabolic models that take underlying molecular mechanisms of interactions into account. The generalized Lotka-Volterra equation and its variants (133–135), but also individual-based models, such as the neutral model (136) and its extensions, are examples of the former.

In the oral cavity, these models have to deal with the complication that most community members can exist both in a free-floating planktonic state and as part of a biofilm, which may have different growth rates and different access to nutrients and which may engage in different interactions. Previously, Schroeder and colleagues (137) proposed a discrete and continuous version of a model that describes the dynamics of both planktonic and sessile communities in drinking water pipes and that may be adapted to model community dynamics in the oral cavity. The programming language gro, which was designed for individual-based modeling of spatially structured microbial communities, may also be of interest in this respect (138). A range of factors, including growth rates, cell signaling, diffusing, and chemotaxis, can be factored in.

Metabolic models require the accurate reconstruction of each community member's metabolism (139), which is a major hurdle because of a lack of reliable and complete genome annotations and the large percentage of unknown gene functions. Metabolic reconstructions may be quickly generated automatically with tools such as ModelSEED or RAVEN (140, 141). This type of modeling presents some disadvantages, such as the requirement for a tedious manual curation to ensure an accurate reconstruction (142) and the assumption that community members are in a metabolic steady state. This assumption is relaxed by some dynamic metabolic models which require kinetic parameters, such as compound uptake rates (139). The dynamic individual-based metabolic modeling tools COMETS (143) and BacArena (144) additionally take spatial structure into account, which is important to model biofilms. Metabolic models can also exploit metaomics data as additional constraints on metabolic fluxes or as validation data (145). For example, gene expression data have been used to validate metabolic models (146). Despite their promise, to the best of our knowledge, metabolic models have been applied only to communities consisting of a small number of species. Metabolic models of species grown alone and in pairs can be exploited to predict ecological interactions (147). For instance, gut microbial interactions were predicted based on the semicurated reconstruction of 773 gut species (148). The extension of dynamic and spatial metabolic models to more complex microbial communities is a promising field for future research.

Community-level metabolic networks are a simpler form of metabolic models, where metabolites and reactions are represented as nodes and edges, respectively, but where stoichiometric coefficients are not taken into account (149). They offer a framework for the straightforward integration of metaomics data as node or edge weights (150). While metabolic networks can handle larger communities, they do not allow quantitative modeling (151).

Quantitative community models have parameters which need to be determined through measurements under well-controlled conditions. For instance, growth assays in mono- and coculture can provide growth rates and interaction strengths. Once a model is parameterized, it needs to be validated experimentally. Such a validation consists of comparing the outcomes of experimental perturbations with the outcomes predicted by the model. The model may undergo several rounds of adjustment and validation until it reaches sufficient accuracy, or it may fail to be predictive because important but unknown factors are not taken into account or the community dynamics are chaotic or predominantly stochastic. A model that predicts community dynamics to an acceptable level of accuracy can be applied to simulate the effects of yet untested perturbations on the community.