Microbiota on the human skin plays a major role in body odour production. The human microbial and chemical signature displays a qualitative and quantitative correlation. Genes may influence the chemical signature by shaping the composition of the microbiota. Recent studies on human skin microbiota, using 16S rRNA gene sequencing, found a high inter- and intrapersonal variation in bacterial species on the human skin, which is relatively stable over time. Human body odours mediate the attraction of mosquitoes to their blood hosts. Odours produced by skin microbiota are attractive to mosquitoes as shown by in vitro studies, and variation in bacterial species on the human skin may explain the variation in mosquito attraction between humans. Detailed knowledge of the ecology and genetics of human skin microbiota is needed in order to unravel the evolutionary mechanisms that underlie the interactions between mosquitoes and their hosts.
Introduction
Human-associated microorganisms on the skin have long been studied because of their role as pathogens. For example, Staphylococcus aureus and Streptococcus pyogenes, which cause skin infections and a variety of other diseases (Bisno & Stevens, 1996; Kluytmans et al., 1997), are probably two of the best-studied microbial pathogens on the human skin. The role of human skin microorganisms as pathogens and in the defence against diseases has been reviewed extensively by Cogen et al. (2008). Skin-inhabiting bacteria also act as producers of odours, which may play a role in the interactions between disease vectors (e.g. mosquitoes, biting midges, biting flies, triatomine bugs, mites, ticks) and their human hosts. Olfaction is the principal sense with which many of these vectors locate their blood hosts, and hence chemical cues affect disease transmission (Takken & Knols, 1999; Logan & Birkett, 2007). The present paper provides an overview of how skin microorganisms shape the human odour profile and the way in which this knowledge may be exploited for novel strategies of vector control and a reduction of vector-borne disease risk. As an example, we discuss how skin microbiota affects the interaction between malaria vector mosquitoes and humans, and suggest how this knowledge can be used to identify the microbial volatiles that mediate this behaviour.
Human microbiota
Since molecular tools became available, there has been an enormous increase in knowledge on the ecology of the human microbiota (Gao et al., 2007; Costello et al., 2009; Grice et al., 2009; Vrieze et al., 2010). The human intestinal microbiota proves to be a complex community of many different species and plays an important role in human health and disease (Egert et al., 2006; Vrieze et al., 2010). Upon disturbance, probiotics can help to restore microbial populations in the intestines and thereby restore the function of the intestinal tract (Ouwehand et al., 2002; Rastall et al., 2005).
In contrast to the intestinal microbiota, until recently, only little was known about the microbial community on the human skin. Most of the knowledge gathered was based on research using selective growth media. The application of molecular techniques shows that only a small proportion of the existing microbial community is detected using these media (Gao et al., 2007; Grice et al., 2008). Recent studies, using molecular techniques, have shown the complexity of the microbial community on the skin and that humans differ strongly in their skin microbiota profiles (Gao et al., 2007; Fierer et al., 2008; Costello et al., 2009; Grice et al., 2009). This knowledge can help to unravel some of the basic functions of microorganisms on the human skin.
Volatile production by microbiota on the human skin
Human sweat is odourless only gets its characteristic smell after incubation with bacteria (Shelley et al., 1953). The microbiota of the skin plays an important role in human odour production and the number of certain microorganisms is strongly correlated with the intensity of the odour emitted (Leyden et al., 1981; Jackman & Noble, 1983; Rennie et al., 1990, 1991; Taylor et al., 2003; Ara et al., 2006). The human skin microorganisms are most abundant in the vicinity of skin glands, where they metabolize the skin gland secretions (Fig. 1).
Distribution of sebaceous (blue), eccrine (green) and apocrine (brown) glands on the human skin surface. Darker colours indicate higher densities (modified after Wilson, 2008).
Human skin glands can be divided into sebaceous and sweat glands (Fig. 2). Sebaceous glands produce sebum, which consists of lipids and dead cells (Figs 1 and 2) (Stoddart, 1990). Sweat glands comprise eccrine and apocrine glands, and produce mainly water (Fig. 1 and 2). Eccrine glands are located all over the body (Fig. 1) and produce sweat to cool the body by evaporative heat loss when its temperature increases. Apocrine glands are mainly found in the axillary region, and are hypothesized to play a role in human pheromone production (Stoddart, 1990). When fresh apocrine secretions are sterilized, they are odourless (Shelley et al., 1953).
Human skin glands and their most important excretions. ( ), no convincing evidence.
To identify which bacteria are responsible for the production of human odour, Leyden et al. (1981) compared the microbiology of the axilla of 229 human subjects with the intensity of their axillary odour. Corynebacteria were found to be responsible for the typical apocrine odour (Leyden et al., 1981). This was confirmed by other studies (Jackman & Noble, 1983; Rennie et al., 1991; Taylor et al., 2003), although an association with the total number of aerobic bacteria and the number of micrococci on the skin was also suggested (Taylor et al., 2003).
Most studies have investigated the relation between axillary odour and microbiota composition and have therefore focused on apocrine sweat. Feet are a major source of human body odour and contain high numbers of eccrine glands (Fig. 1). Increased foot odour is associated with higher population densities of microorganisms with lipase and proteinase activity (Marshall et al., 1987) and/or higher numbers of Bacillus species (Ara et al., 2006). The human scalp has a high density of sebaceous glands (Sastry et al., 1980). Sebaceous glands contain numerous propionibacteria that hydrolyse triglycerides into volatile fatty acids (Kearney et al., 1984).
After a correlation between odour production and bacterial composition was recorded, in vitro studies further revealed the underlying mechanisms of odour production by specific bacterial species. Gower et al. (1994) showed that 5α-androstenone, which is suggested to contribute to axillary odour in men, is produced by two coryneform bacterial strains when supplied in vitro with four nonvolatile 16-androstenes. Corynebacteria also play a pivotal role in the generation of volatile fatty acids, which are associated with malodour (James et al., 2004). When skin lipids are catabolized into long-chain fatty acids, it seems that only corynebacteria are capable of transforming these long-chain fatty acids into short- and medium-chain fatty acids (C2–C11), causing malodour (James et al., 2004). Brevibacteria and micrococci metabolize these short- and medium-chain fatty acids even further (James et al., 2004). Another example of the specificity of biotransformations by certain bacterial species is the conversion of branched-chain amino acids. Staphylococcus species, but not Corynebacterium species, can convert these amino acids to highly odorous short-chain amino acids (C4–C5) (James et al., 2004). The above examples illustrate that different species of skin bacteria each have their own specific metabolism and, therefore, generate a characteristic odour profile.
The complete genomes of the skin bacterial species Bacillus subtilis (Kunst et al., 1997), Corynebacterium jeikeium (Tauch et al., 2005), Propionibacterium acnes (Bruggemann et al., 2004) and Staphylococcus aureus (Kuroda et al., 2001) have been sequenced. A next step will be to link these sequences to the pathways and mechanisms of odour production by these bacteria on the human skin.
Bacterial enzymes involved in odour production, by converting nonvolatile compounds into volatile compounds, have been identified and their genes brought to expression in Escherichia coli (Natsch et al., 2003, 2004). Specific odour-producing enzymes can be blocked by inhibitors, which have a higher affinity for the enzyme than the original substrate (Ara et al., 2006).
Another group of components involved in odour production on the human skin involves odour-binding proteins. In the apocrine glands, they transport odour molecules to the skin surface (Spielman et al., 1995; Zeng et al., 1996). When the amount of these Apocrine Secretion Odour-Binding (ASOB) proteins was measured on the skin of individuals of Chinese and non-Chinese descent, a lower amount of protein was detected on the panellists of Chinese ancestry (Jacoby et al., 2004). This may result in lower body odour production.
Research on human body odour is often focused on a specific group of odours or pathways, often related to malodour. Studies that aimed at analysing all components of human body odour have produced long lists of chemicals, often dependent on the methods used for the collection and analysis of the body odours. Bernier et al. (1999, 2000) analysed human skin emanations collected on glass beads and reported 346 compounds including carboxylic acids, alcohols, esters, aldehydes, aliphatics, aromatics and ketones. The majority of the intense peaks in the chromatograms were fatty acids and the pattern of these peaks appeared to be similar to that observed from bacteria that convert triglycerides on the skin into these fatty acids (Bernier et al., 2000). The collection and analysis of skin emanations through direct skin contact result in many compounds, often nonvolatile (Bernier et al., 2000; Penn et al., 2006). The use of solid phase microextraction (SPME) or Dynamic Headspace Sampling (DHS) to analyse compounds released from human skin limits the results to volatile compounds, and in these studies, often fewer, but also different compounds are detected than in studies analysing skin emanations through direct skin contact (Curran et al., 2005; Gallagher et al., 2008).
One of the methods commonly used to determine microbial profiles is denaturing gradient gel electrophoresis analysis, which allows for DNA fingerprinting of complex microbial communities. Comparison of this microbial profile of the skin with its chemical profile shows a correlation, but only when the subjects strictly followed some basic rules of behaviour, such as no deodorant use for 48 h and wearing a t-shirt provided by the researchers (Xu et al., 2007). The importance of behavioural rules for the volunteers indicates that, in addition to the microbial composition, environmental factors such as grooming habits and diet also influence the formation of human body odour (Xu et al., 2007; Havlicek & Lenochova, 2008).
Genetic origin of human odour
The human odour profile is, at least partly, genetically based. By smell, humans can match monozygotic twins, but not dizygotic twins, based on their body odours even when the twins are living apart (Wallace, 1977; Roberts et al., 2005). Humans are able to discriminate their relatives from nonrelatives, based on their odour (Porter et al., 1985; Weisfeld et al., 2003). This kin recognition may indicate a genetic basis for the human odour profile, although the exact mechanism remains unknown (Lenochova & Havlicek, 2008).
The volatile compounds released by monozygotic twins can also be matched based on qualitative and quantitative similarities (Sommerville, 1994). Recently, Kuhn & Natsch (2009) compared the body odour of monozygotic twins using two-dimensional GC, focusing on volatile carboxylic acids known to be the principal components of body odour, and released by skin bacteria. First, a fresh sweat sample was taken from the axilla of the individuals. Next, to induce the production of volatile odours from the fresh sweat sample, a recombinant enzyme from axilla bacteria was used instead of skin bacteria themselves. In this way, the influence of personal grooming habits on the bacterial population of the skin was ruled out. The study showed a clear and strong contribution of genetic factors to the relative pattern of volatile fatty acids released and the importance of bacterial enzymes in the production of volatile odorants from fresh sweat samples (Kuhn & Natsch, 2009).
In mice, genes of the major histocompatibility complex (MHC) have been shown to influence body odour (Yamazaki et al., 1976; Penn & Potts, 1998a), and several studies have attempted to link MHC genes to body odour in humans. The genes of the MHC complex encode antigens, which are involved in the immune system. When humans had to judge the body odours coming from worn T-shirts, they preferred body odours from persons with another MHC gene profile (Wedekind & Furi, 1997; Wedekind & Penn, 2000). This leads to the conclusion that individuals have a distinct body-odour type, which is determined, at least partly, by their inherited MHC alleles.
In 1998, Penn & Potts (1998b) reviewed the available data on the connection between MHC genes and body odour and summarized the possible explanations on how these genes might influence body odour. They combined the hypothesis that MHC genes influence odour by shaping the commensal microbiota (Howard, 1977) and the hypothesis that converted MHC molecules transport aromatic molecules (Pearse-Pratt et al., 1992). It was hypothesized that MHC molecules bind allele-specific subsets of peptides, from which metabolites are volatilized by the activity of the commensal microorganisms (Penn & Potts, 1998b). In an experiment to test this hypothesis, the GC-MS analysis of the axillary odour of 18 volunteers revealed a number of compounds, including 3-methylbutanal. This compound was observed to be a potential link to the human leukocyte antigen (HLA) genes (Savelev et al., 2008), which are the genes located in the MHC region of humans, encoding for specific antigens. An in vitro experiment was conducted with axillary bacteria from individuals with a high or a low abundance of 3-methylbutanal. Media with two different HLA peptides were incubated with the bacteria of individuals with a high or a low production of 3-methylbutanal in their armpit. The results showed that different HLA peptides can alter the production of 3-methylbutanal by skin bacteria and that the microbial populations themselves influence the production of 3-methylbutanal (Savelev et al., 2008).
Inter- and intrapersonal diversity in skin microbiota
As discussed, odour production on the human skin is in part a function of the composition of the skin microbiota. Because the human odour profile is at least partly genetically based, it can be expected that intrapersonal variation in the microbiota composition over time is lower than interpersonal variation. This has indeed been shown for the composition of faecal microbiota (Holdeman et al., 1976; Eckburg et al., 2005), but studies investigating intra- and interpersonal variations of skin microorganisms have been rare.
Recently, the human microbial community was studied in more detail, using 16S rRNA gene sequencing. This technique allows for the identification of more bacterial species than is possible with culture-based methods alone (Gao et al., 2007; Grice et al., 2008), and it allows for a better overview of the composition of the human microbial community. On the surface of the average hand palm, more than 150 bacterial species were found (Fierer et al., 2008) and 88 on the inner elbow (Grice et al., 2008). The variation in the composition of bacterial communities on the human skin is often high and these communities are more diverse than bacterial communities found in throat, stomach and faecal environments (Fierer et al., 2008; Costello et al., 2009).
In several studies, using 16S rRNA gene sequencing, the variation between individuals and the temporal stability of the skin microbial communities have been investigated with various results. No personal bacterial profile was found when the bacterial variation between the left and the right inner elbow of an individual was compared with the left and the right inner elbow of different individuals (Grice et al., 2008) and also no longitudinal stability of the microbiota was found when samples from the forearm of four volunteers were taken 8–10 months apart (Gao et al., 2007). Other studies, however, showed that the variation between the left and the right body parts of the same individual was lower than the interpersonal variation (Gao et al., 2007; Fierer et al., 2008; Grice et al., 2009), and sampling multiple skin sites at multiple time points showed that the variation in skin microbiota between individuals was higher than within an individual (Costello et al., 2009; Grice et al., 2009). In addition to this, individuals can be identified by the bacterial traces they leave on objects such as a computer keyboard or mouse (Fierer et al., 2010).
These studies, using high-throughput molecular techniques, all showed a high bacterial diversity and indicate that the composition of bacterial communities on the human skin depends on skin site characteristics. Detailed studies with multiple sample sites and time points clearly indicate that humans have an individual microbial skin composition that is relatively stable over time (Costello et al., 2009; Grice et al., 2009).
Volatiles of human skin bacteria and vector–host interactions
Several insect families act as vectors of harmful diseases such as malaria, leishmaniasis, river blindness and West Nile virus (Takken & Knols, 1999; Resh et al., 2004; Reisen et al., 2005; Kamhawi, 2006). These vectors locate their blood hosts, including humans, from a distance using olfactory cues produced by their hosts (Takken & Knols, 1999; Logan & Birkett, 2007). Recent studies demonstrated the role of skin microbiota in interactions between disease vectors and their blood hosts (Braks & Takken, 1999; Verhulst et al., 2009; Ortiz & Molina, 2010), and the knowledge of human skin microbiota reviewed in this paper sheds further light on these interactions.
As an example, and because of recent discoveries on its olfactory behaviour, we focus on the malaria vector Anopheles gambiae Giles sensu stricto (hereafter referred to as A. gambiae). This mosquito species is anthropophilic and its host seeking is mainly accomplished by odour-mediated anemotaxis in which human odours provide essential cues (Takken & Knols, 1999). Many of the volatiles to which these mosquitoes respond are of bacterial origin. This became evident when washing human feet with a bactericidal soap proved to significantly alter the biting-site selection of A. gambiae females on a motionless naked volunteer (De Jong & Knols, 1995). In addition, human eccrine sweat is attractive to A. gambiae after incubation with skin bacteria for 1 or 2 days (Braks & Takken, 1999). Moreover, a recent study showed that the volatiles produced by human skin bacteria in vitro are attractive to A. gambiae when grown on agar plates (Verhulst et al., 2009). The headspace volatiles from cultures of these bacteria were identified and a synthetic blend was developed that attracted females of this mosquito species (Verhulst et al., 2009). However, it is not yet clear whether bacteria isolated from less attractive individuals produce these volatiles, whether other substrates can enhance the production of attractive volatiles and whether the production of these attractive volatiles is specific to bacterial species present only on human skin. These important issues deserve being addressed in future studies.
Further research on the role of skin microbiota in the host-seeking behaviour of mosquitoes can possibly reveal the evolutionary mechanisms of vector–host interactions in general. Rhodnius prolixus Stål (Hemiptera: Reduviidae: Triatominae), for example, is the vector of Chagas disease and its host-seeking behaviour is influenced by host odours (Guerenstein & Lazzari, 2009). Odours produced by bacteria also seem to play an important role for R. prolixus, because an antibacterial gel reduced the attractiveness of skin washings to this vector (Ortiz & Molina, 2010).
The discovery that volatiles produced by human skin microorganisms in vitro mediate A. gambiae host-seeking behaviour creates new opportunities for the control of this mosquito species. Identifying the volatiles in the headspace of bacterial cultures, or mass production of the bacteria themselves, may lead to the development of new odour-baited trapping systems. African malaria mosquitoes and men have coevolved for thousands of years, and this process may have led to the development of microorganisms on the human skin that repel mosquitoes. Identification of these repellent microorganisms and the volatiles that they produce can lead to the development of new repellent products. Blockage of odour-producing bacteria or specific odour-producing enzymes in bacteria (Ara et al., 2006) can possibly affect the attractiveness of a person to mosquitoes.
For most vector-borne diseases, the primary factor influencing transmission is human exposure to bites of infected vectors (Téllez, 2005). Humans differ in their attractiveness to mosquitoes and these differences remain relatively stable over time (Schreck et al., 1990; Lindsay et al., 1993; Knols et al., 1995; Bernier et al., 2002; Mukabana et al., 2002; Qiu et al., 2004, 2006; Logan et al., 2008). Because the human skin microbiota plays an important role in the production of body odours and is attractive to A. gambiae when grown on agar plates, a correlation between skin microbiota composition and a person's attractiveness to mosquitoes can be expected. Within humans, this may lead to selection for skin bacteria that are less attractive or even repellent to mosquitoes.
Human attractiveness to mosquitoes may change when individuals are infected with Plasmodium parasites that cause malaria (Lacroix et al., 2005; Mukabana et al., 2007). Children harbouring Plasmodium gametocytes attracted about twice as many mosquitoes as children without infection or with parasites in the asexual stage (Lacroix et al., 2005). This increased attractiveness of infected children could be explained by an increase in body temperature, increased perspiration or by a change in breath composition. However, these factors are less likely to be involved in the observed change in attractiveness, as the infection was asymptomatic in all of the children involved (Lacroix et al., 2005). It remains to be investigated whether an infection with Plasmodium affects the composition and/or the number of skin bacteria. If so, it might be an additional explanation for the increase in attractiveness to malaria mosquitoes (Fig. 3).
Hypothetical model of mosquito–host interactions and the role of skin microbiota (modified after Knols, 1996). The most important factors involved are shown. Dashed lines indicate the effects when the host is infected with Plasmodium parasites. RH, relative humidity.
Concluding remarks and future directions
The microbiota on the human skin plays a major role in body odour production. Body odour is used by disease vectors to locate their blood hosts and the differential attractiveness of humans to mosquitoes may be explained by individual variation in the microbial composition on the skin. Differences in observed human attractiveness and a correlation with microbiota composition can be exploited to mediate vector behaviour as a tool for disease control. Bacteria such as corynebacteria and brevibacteria can have their own specific metabolism and, therefore, a characteristic odour profile. Recent technological advances revealed the pathways and mechanisms of odour production on human skin, and the contribution of specific microorganisms to this odour production. Next, parts of the genome sequences of skin bacteria need to be linked to these pathways. By screening for these specific sequences, the role and significance in odour production of the different bacteria on the human skin can be determined and will reveal their role in the attraction of humans to mosquitoes.
Several studies have indicated a link between HLA genes and the human odour profile (Wedekind & Furi, 1997; Wedekind & Penn, 2000; Savelev et al., 2008), although the evidence is not fully convincing as yet. These HLA studies focused on the axillae as these are suggested to play a role in human pheromone production from the apocrine glands. If HLA genes influence human body odour produced by eccrine glands and not apocrine glands, HLA-determined odours might be easier to detect from other body parts, because apocrine glands are mainly found in the axillae.
Molecular studies have mapped the human microbiota community in detail (Gao et al., 2007; Fierer et al., 2008; Grice et al., 2008, 2009; Costello et al., 2009), and the recently launched Human Microbiome Project (http://www.hmpdacc.org/resource.php) will further characterize the human microbiota. Several studies have examined the stability of the human microbial population over time (Gao et al., 2007; Costello et al., 2009; Grice et al., 2009). The results show that a large number of volunteers and many time points are essential in these studies. To be able to link the human odour profile with its skin microbiota, sequencing the 16S rRNA gene fragments may not yield the desired results because these genes are not involved in the metabolism of the bacteria. Sequencing the genome of the skin microbiota and blasting these sequences can indicate which genes involved in certain metabolic pathways are present. Whole transcriptome sequencing involves mRNA, which reflects genes that are actively expressed and codes for enzymes that regulate the metabolic activity of the skin microbiota at the time of sampling. At present, the known metabolic pathways and links to genes are limited and the methods are still costly (Zoetendal et al., 2006; Turnbaugh et al., 2007; Vrieze et al., 2010). To reveal a possible relationship between the human genome and the bacterial composition on the skin and the odour profile, blood samples for HLA analysis, detailed mapping of the human skin microbiota and odour samples for GC-MS analysis are required.
More detailed knowledge on how genes, human skin microbiota and human body odour are related can not only help to understand and develop treatment of skin diseases, but possibly also unravel the evolutionary mechanisms behind the host-seeking behaviour of mosquitoes and other blood-feeding insects and lead to novel means of vector-borne disease control.
Acknowledgements
This work was supported in part by a grant from the Foundation for the National Institutes of Health through the Grand Challenges in Global Health Initiative (GCGH#121).
References
Author notes
Editor: Ian Head
