Molecular basis of human body odour formation: insights deduced from corynebacterial genome sequences


  • H. Barzantny,

    1. Institut für Genomforschung und Systembiologie, Centrum für Biotechnologie, Universität Bielefeld, Universitätsstraße 27, D-33615 Bielefeld, Germany
    Search for more papers by this author
  • I. Brune,

    1. Institut für Genomforschung und Systembiologie, Centrum für Biotechnologie, Universität Bielefeld, Universitätsstraße 27, D-33615 Bielefeld, Germany
    Search for more papers by this author
  • A. Tauch

    1. Institut für Genomforschung und Systembiologie, Centrum für Biotechnologie, Universität Bielefeld, Universitätsstraße 27, D-33615 Bielefeld, Germany
    Search for more papers by this author

Helena Barzantny, Institut für Genomforschung und Systembiologie, Centrum für Biotechnologie, Universität Bielefeld, Universitätsstraße 27, D-33615 Bielefeld, Germany. Tel.: +49 521 106 12253; fax: +49 521 106 89041; e-mail:


During the past few decades, there has been an increased interest in the essential role of commensal skin bacteria in human body odour formation. It is now generally accepted that skin bacteria cause body odour by biotransformation of sweat components secreted in the human axillae. Especially, aerobic corynebacteria have been shown to contribute strongly to axillary malodour, whereas other human skin residents seem to have little influence. Analysis of odoriferous sweat components has shown that the major odour-causing substances in human sweat include steroid derivatives, short volatile branched-chain fatty acids and sulphanylalkanols. In this mini-review, we describe the molecular basis of the four most extensively studied routes of human body odour formation, while focusing on the underlying enzymatic processes. Considering the previously reported role of β-oxidation in odour formation, we analysed the genetic repertoire of eight Corynebacterium species concerning fatty acid metabolism. We particularly focused on the metabolic abilities of the lipophilic axillary isolate Corynebacterium jeikeium K411.


Pendant les décennies passées, il ya eu un intérêt croissant pour le rôle essentiel des bactéries commensales de la peau dans la formation des odeurs du corps humain. Il est maintenant généralement admis que les bactéries de la peau sont à l’origine de ces odeurs corporelles par biotransformation de composants de la sueur axillaire. On a, en particulier, montré que les corynebacteria aérobiques contribuent fortement à l’odeur axillaire, tandis que d’autres résidents de la peau semblent avoir une influence moindre. L’analyse de composants odoriférants de la sueur a montré que les substances majoritairement responsables de l’odeur incluent des dérivées de stéroïdes, des acides gras volatils à courtes chaines et des sulphanylalcanols. Dans cette mini revue, nous décrivons les bases moléculaires des quatre voies les plus largement étudiées dans la formation des odeurs corporelles, en nous concentrant sur les processus enzymatiques sous-jacents. En considérant le rôle précédemment décrit de la β oxydation dans la formation d’odeur, nous avons analysé le répertoire génétique de huit espèces de Corynebacterium concernant le métabolisme des acide gras. Nous nous sommes particulièrement concentrés sur les capacités métaboliques de Corynebacterium jeikeium K411 isolé de la composante lipophile axillaire.


The human skin is a multifunctional organ that protects the human body from dehydration, regulates body temperature and forms the first line of defence against bacterial pathogens and viruses [1]. The latter function is achieved by secretion of antimicrobial substances, the establishment of a low pH (5.5) and the maintenance of a commensal bacterial community on the skin surface [2]. It is currently believed that the benefits of healthy skin microflora are comparable with those of healthy gut flora, as demonstrated by several studies under the human microbiome project [3–5]. The human skin can be colonized by many different bacterial species, because it offers physiologically diverse niches characterized by a sebaceous, dry or moist environment. Sebaceous areas like the external auditory canal are predominantly colonized by propionibacteria and staphylococci species, whereas dry skin largely harbours β-proteobacteria and flavobacteriales [4]. The moist environment of the human axilla is characterized by the presence of oily and odourless fluids containing proteins, cholesterol, steroid derivatives, squalene and a wide range of lipids [6–8]. These substances are secreted by eccrine, sebaceous and apocrine glands specifically located in the underarm region of the human body and some of which are stimulated by emotional stress [9, 10]. Emotional sweating does not occur until puberty and is often associated with development of strong body odour [11].

In the past few decades, several studies have focused on the relationship between these odourless sweat secretions and the formation of a strong body odour. It has been known since the 1950s that the metabolic activity of microorganisms that inhabit the skin surface is essential for the development of body odour [12], but the exact processes could not be fully elucidated until now. The composition of the bacterial community colonizing the axilla is quite diverse, when different individuals are compared and even slight differences can be observed between the right and left axilla of just one individual [13]. Nevertheless, closer investigation of the bacterial community structure in the axillary microenvironment showed that it is usually colonized by four major groups of bacteria, namely staphylococci, aerobic coryneforms, propionibacteria and micrococci [4]. High levels of strong body odour were observed in individuals with a microflora dominated by aerobic coryneforms, whereas staphylococci-dominated axillae only revealed low levels of odour [14].

Until now, several routes of odour formation have been detected including the transformation of glycerol and lactic acid, and the conversion of aliphatic amino acids into volatile fatty acids [15]. Of all, the four routes most extensively studied and the generally accepted mechanisms are as follows: (i) the biotransformation of steroids, (ii) the release of short branched-chain fatty acids from glutamine-conjugates, (iii) the release of short sulfanylalkanols from glycine-cysteine- or cysteine-(S)-conjugates and (iv) the biotransformation of long-chain fatty acids into volatile short branched-chain fatty acids.

In this mini-review, we summarize the current knowledge of the molecular processes contributing to human axillary malodour formation. We present the four most extensively studied routes of odour formation and their underlying molecular mechanisms and enzymatic processes. Considering the previously reported role of β-oxidation in axillary odour formation, we analyse and compare the repertoire of fatty acid-degrading enzymes found in corynebacterial genomes [16, 17]. As a model organism, we use the skin isolate Corynebacterium jeikeium K411, for which the complete genome sequence is available and transcriptomic as well as proteomic methods have been established, facilitating future research approaches in the field of human body odour formation [18–20].

Major routes and mechanisms of human body odour formation

Biotransformation of steroids

There is only a limited understanding of the transformation of steroids into volatile odorous products until now, even though substances such as 5α-androst-16-en-3-one, androst-16-en-3α-ol and its β-epimer have been identified as malodour components with low olfactory thresholds in male body odour [21]. The routes of generating these components are still not clear, but can be traced back to the enzymatic abilities of lipophilic aerobic corynebacteria, because enrichment of the culture medium with glucose and oxygen limitation led to decreased biotransformation in vitro [22]. Furthermore, it has been shown that bacterial communities collectively operate to generate odorous steroids and that the interplay between different bacterial species in the human axilla is necessary for efficient biotransformation [23]. It is hypothesized that these bacteria convert odourless precursor steroids into 16-androstenes and thereby contribute to human malodour. In agreement, none of the 16-androstenes has been detected so far in fresh sweat [21]. Decréau et al. were able to show that steroids of the androsta-5,16-diene series are a major source of androsta-4,16-dien-3-one and, therefore, likely precursors of human malodour components [22]. Furthermore, the authors demonstrated that the stereochemistry of the precursor molecules determines the extent of malodour. For instance, the α-epimers of different 3-sterols were more readily transformed into odorous 3-oxo-steroids by aerobic corynebacteria than the corresponding β-epimers [21, 22, 24]. The analysis of biotransformation and the detection of intermediates allowed the prediction of enzyme functions involved in the process of body odour development. It is proposed that the interplay between 3α(β)-sterol hydrogenase, 4, 5- or 5α-reductase, steroid-4,5-isomerase and hydroxylases leads to the conversion of non-odouriferous precursor molecules into odouriferous steroid derivatives [22]. No 5α-reductase activity could be verified in vitro until now, but the requirement of a reductase system in the process of steroid biotransformation has been demonstrated [22, 25]. Still, further research is necessary to clarify the actual contribution of steroids to human body odour and the enzymatic steps involved in the generation of axillary malodour. Detailed analysis of the bacterial community structure of the human axilla can be performed by using metagenomics, independently of the culturability of the skin microbes [26, 27]. Additionally, the application of metatranscriptomics will give an overview of enzymatic activities present in the human axillae. This will help to identify new enzymes contributing to human body odour and to get new insights into the interplay between bacterial species that coexist in the human axilla [28].

Cleavage of glutamine-conjugates by the action of aminoacylases

In the early stages of human malodour research, it was suggested that axillary odour mainly originates from steroid derivatives metabolized on the human skin. At the same time, little attention was paid to the role of short and volatile branched-chain fatty acids. However, a comprehensive study by Zeng et al. changed this view and led to a more accurate understanding of the complexity of human malodour. A wide range of straight-chain, branched and unsaturated C6-C11 acids as well as C5-C10γ-lactones were identified along with (E)-3-methyl-2-hexenoic acid (3M2H), the most dominant and characteristic component of human axillary odour [29]. 3M2H can be detected in two stereo isomeric forms and is released upon bacteriolysis from a non-odorous precursor molecule present in axillary secretions. Natsch et al. demonstrated that 3M2H, similarly to its odorous derivative 3-hydroxy-3-methyl hexanoic acid (HMHA), is covalently bound to a glutamine residue and released by the action of a Nα-acylglutamine aminoacylase (AgaA) cloned from the skin microbe Corynebacterium striatum Ax20 (Fig. 1) [30]. Experimental data demonstrated that AgaA is located in the cytoplasm of C. striatum and that its activity is Zn2+ dependent. Enzymes homologous to AgaA are found in archaea, eubacteria, plants and the skin microbe C. jeikeium K411 (Fig. 2). The latter encodes three aminoacylases with 30-40% sequence identity to AgaA. Although no experimental data are available, it is likely that these enzymes are responsible for the cleavage of Gln-conjugates in C. jeikeium. Yet, further research is necessary to investigate their contribution to body odour formation. Although most of the enzymes shown in Fig. 2 are Zn2+-dependent, no common zinc-binding domain, such as the HEXXH motif, has been identified by sequence comparison [31, 32]. Only two highly conserved motifs are present containing several amino acids potentially involved in zinc binding. The conserved cysteine and histidine residues are of particular interest, because they are found in other metal-binding domains as well [33].

Figure 1.

 Schematic overview of two important mechanisms of human body odour formation. The precursor molecules (bright green) are transported intracellularly into the vacuoles of secretory cells by the transport protein ABCC11 (orange) [70]. It is hypothesized that precursor molecule-filled vacuoles release their content into the lumen of the apocrine sweat glands. These glands then secrete, among others, the malodour precursors Gln-3-hydroxy-3-methylhexanoic acid (bright green) and Gly-Cys-(S)-3-methyl-3-sulphanylpentan-1-ol (orange/red) onto the skin surface. The precursors are cleaved locally by bacterial enzymes resulting in the release of volatile short fatty acids and sulphanylalkanols, which contribute strongly to human malodour.

Figure 2.

 Partial sequence comparison between the putative aminoacylases from Corynebacterium jeikeium K411 and enzymes of known function. The amino acid sequences were aligned using the ClustalX algorithm. Two amidohydrolases from Pyrococcus horkoshii OT3 (NP_142952) and Geobacillus stearothermophilus (CA70000), a carboxypeptidase from Sulfolobus solfataricus (CAA88397), an amino acid hydrolase from Arabidopsis thaliana (P54968), the aminoacylase AgaA from Corynebacterium. striatum (AAN77164), a hippuricase from Campylobacter jejuni (CAA85396) and the putative aminoacylases AmiA, AmiB and AmiC from C. jeikeium K411 (jk1673, jk0802 and jk0500, respectively) were used for the alignment. Highly conserved sequences are marked with red boxes [34]. Residues potentially involved in enzyme activity are marked with asterisks [33].

Interestingly, the activity of AgaA has been shown to be highly specific for the Gln-residue, because replacement of Gln by various other amino acids resulted in abolishment of substrate cleavage [34]. Variation in the acyl chain of the Gln-precursor, in contrast, seems to have no negative influence on enzyme activity [34]. Considering its broad substrate specificity, AgaA has been used in the in vitro testing of substances that control human malodour, such as deodorant additives. Experiments with AgaA and precursor molecules, composed of a Gln-residue and different fragrance alcohols, clearly showed an odour-masking effect. Moreover, the affinity for malodour precursors was reduced, because the fragrance precursor competed for the active site of AgaA [30]. Overall, this strategy holds promise in the development of new deodorant additives, which specifically target an enzyme, responsible for malodour formation, rather than the entire microbial axillary community. The identification of the bacterial enzyme responsible for malodour formation boosted the search for further malodour components released upon AgaA activity. In a comprehensive GC/MS analysis, Natsch et al. identified 26 odorous acids in AgaA-treated sweat and two of the corresponding Gln-conjugates in fresh sweat samples (Table I) [35].

Table I.   List of acids identified in human sweat samples according to Natsch et al. [34]
CategoryChemical compound
  1. *Acids previously unreported in nature.

  2. Acids with Gln-precursor molecules found in fresh sweat samples.

Hydroxy acids3-Hydroxy-3-methylhexanoic acid
3-Hydroxy-4-methylhexanoic acid*
3-Hydroxy-3-methylheptanoic acid*
3-Hydroxy-4-methylheptanoic acid*
3-Hydroxyoctanoic acid
3-Hydroxy-3-methyloctanoic acid*
3-Hydroxy-4-methyloctanoic acid*
3-Hydroxy-4-methylnonanoic acid*
3-Hydroxydecanoic acid
Unsaturated branched acids4-Methyloct-4-enoic-/4-methylideneoctanoic acid
(Z)-4-Methyloct-3-enoic acid*
(E)-4-Methyloct-3-enoic acid*
(Z)-4-Methylnon-3-enoic acid*
(E)-4-Methylnon-3-enoic acid
Amino acid degradation products/others2-Hydroxypropanoic acid (lactic acid)
3-Methyl-2-oxopentanoic acid
4-Methyl-2-oxopentanoic acid
4-Ethylheptanoic acid
Phenylacetic acid
4-Ethyloctanoic acid (‘goat acid’)
8-Hydroxyoctanoic acid
Octanedioic acid (suberic acid)
9-Hydroxynonanoic acid
(4-Hydroxyphenyl)acetic acid
Non-anedioic acid (azelaic acid)

The origin of the Gln-conjugates found in axillary secretions has not been hitherto elucidated. However, it has been suggested that the Gln-conjugates are produced in other organs (e.g. the liver) and thereafter specifically transported to the axillae [35]. On the other hand, it is possible that the synthesis of Gln-linked malodour precursors occurs uniquely in the apocrine glands located in the axillary vault. The development and number of apocrine glands has been reported to vary between different ethnic groups. Therefore, it is plausible that the development of body malodour is to some extent determined by genetic predisposition [36]. Studies comparing gender-specific differences concerning malodour components further support the importance of genetic factors in malodour formation [24]. More recently, Martin et al. have further proven the influence of genetic predisposition [37]. It was hypothesized that the ATP-binding cassette transporter subfamily C member 11 (ABCC11), found in apocrine sweat glands, is involved in the secretion of odour precursors in the axillae. In fact, ABCC11 has been reported to be able to transport different lipophilic compounds [38]. A single nucleotide polymorphism (SNP; 538G>A) in the ABCC11 gene, expressed in ceruminous glands, alters the human earwax type from wet and yellow to dry and white [39]. Because ceruminous glands and apocrine sweat glands are closely related, Martin et al. hypothesized that the ABCC11 SNP in the apocrine sweat glands may also alter axillary odour formation [40]. Therefore, the ABCC11 genotype of 25 volunteers was determined, and their odour profiles were closely examined. The results indicated that ABCC11 is indeed responsible for the secretion of Gln-conjugates as well as other malodour components and that the reported SNP in the ABCC11 gene leads to a loss of transport activity [37]. Consequently, in certain ethnic groups (e.g. East Asians) and individuals carrying the SNP in ABCC11, malodour formation is reduced, because the precursor molecules are deficiently secreted onto the skin surface and are therefore not accessible for bacteriolysis.

Cleavage of glycine-cysteine-(S)-conjugates by the action of metallopeptidases and C-S lyases

The third group of molecules contributing to human body odour formation, besides odoriferous steroid derivatives and short volatile fatty acids, comprises several volatile thiols. Sulphur-containing substances have previously been reported to contribute to a wide range of different odours, from cat urine to more pleasant fragrances such as the aroma of passion fruit juice and the bouquet of white wine or aged champagne [41–44]. Therefore, several investigators concentrated on the identification of sulphuric components in human axillary odour. In 2004, Hasegawa et al. detected 3-methyl-3-sulphanylhexan-1-ol (3M3SH) as a malodour component in incubated human sweat along with 2-methyl-3-sulphanylbutan-1-ol (2M3SB), 3-sulphanylpentan-1-ol and 3-sulphanylhexan-1-ol [45]. Meanwhile, another study by Natsch et al. detected the same sulphuric components in NaOH-treated axilla secretions, supporting the previous findings by Hasegawa et al. [46]. These sulphanylalkanols have very low olfactory threshold values (1–8 pg L−1), indicating the high impact of sulphuric compounds on axillary malodour, even when only trace amounts are present [46]. Because none of these substances could be detected in fresh sweat samples, the question arose whether the malodour components are released by bacterial enzymes from non-odorous precursor molecules, as described previously for (E)-3-methyl-2-hexenoic acid [30].

A detailed analysis of flavour-active volatile thiols in Sauvignon blanc must extract identified Cys-(S)-conjugates as non-odorous precursor molecules. Incubation with a cell-free bacterial lysate resulted in cleavage of Cys-precursor molecules and release of volatile thiols. The responsible enzyme was found to be a β-lyase [47]. These findings prompted Natsch et al. to synthesize specific Cys-(S)-conjugates for 3M3SH to investigate the release of the malodour component upon bacterial incubation. Skin isolates of Staphylococcus and Corynebacterium were incubated with synthetic Cys-3M3SH, (S)-S-benzyl-cysteine and Cys-2M3SB, whereas significant malodour development was only detected with corynebacteria. Thereupon, the corynebacterial C-S lyase AecD (MetC) of C. striatum Ax20 was cloned and shown to cleave the synthetic Cys-(S)-conjugates and, furthermore, to release a sulphuric odour from fresh odourless sweat samples [46]. Additionally, it was demonstrated that the substrate specificity of AecD is especially dependent on the cysteine residue of the conjugate, as shown before for the aminoacylase AgaA and Gln-conjugates [34]. The hypothesis of Cys-(S)-conjugates as non-odorous precursor molecules of volatile thiols was further supported by the findings of Starkenmann et al., who detected Gly-Cys-(S)-conjugates of 3M3SH, 3-sulphanylhexan-1-ol and 2-methyl-3-sulphanyl-pentan-1-ol in sterile human sweat samples [48].

To investigate whether 3M3SH can be directly released from the Gly-Cys-precursor by an enzymatic reaction, the dipeptide conjugate was incubated with total cell extracts of several bacterial skin isolates or a purified C-S lyase [49]. 3M3SH was released upon incubation with the cell extract of C. striatum Ax20, but not with the C-S lyase alone, clearly indicating the presence of an additional enzyme involved in precursor cleavage. To verify this, the metal-chelating agent o-phenanthroline was added to the cell extract, which resulted in abolishment of precursor cleavage. Moreover, the incubation of the Cys-(S)-conjugates with the cell lysate and o-phenanthroline still resulted in 3M3SH release, and the activity of the C-S lyase on Cys-(S)-conjugates was not influenced by the metal-chelating agent. All in all, these findings suggested the presence of a metal-dependent enzyme, which was subsequently identified as the Zn2+-dependent dipeptidase TpdA and cloned from C. striatum Ax20 [49]. The release of 3M3SH from its Gly-Cys-(S)-precursor was verified in vitro by combining the two enzymes (TpdA and AecD) cloned from the skin isolate C. striatum Ax20 (Fig. 1). Further, it was demonstrated that C. jeikeium K411 is also able to release 3M3SH from the dipeptide precursor and that genes homologous of tpdA (jk0266) and aecD (jk0592) are encoded in its genome. The release of the thiol was only detected after a prolonged incubation with C. jeikeium K411 and enzyme assays revealed a 25-fold decreased vmax compared to TpdA from C. striatum [49].

Even though the molecular basis of Cys-Gly-(S)-conjugate-cleavage in C. striatum has been unveiled, no experimental data are available with respect to the regulation of gene expression in this organism. Nonetheless, the transcriptional regulation of aecD has been demonstrated recently for C. jeikeium K411 [50]. Real-time reverse transcriptase polymerase chain reaction (RT-PCR) experiments revealed a decreased expression of aecD mediated by the TetR-like repressor McbR (jk0101), in cells cultivated with 5 mmol L−1 methionine. A transcription factor-binding site was identified within the aecD promoter region, indicating that this organism tightly regulates methionine biosynthesis. Whether, however, the mentioned differences in thiol release between the C-S lyase of C. jeikeium and C. striatum are because of the ability to regulate its expression remains to be elucidated.

Biotransformation of long-chain fatty acids (LCFAs)

The fourth major route of human body odour formation proposed in the literature is the conversion of LCFAs into volatile short-chain fatty acids (VFAs) [15, 16]. The surface of the human skin is characterized by the presence of a wide range of lipids that mainly originate from sebaceous and apocrine glands. Their oily secretions are composed of squalene, cholesterol, wax esters, triacylglycerols, unesterified fatty acids and glyco- or phospholipids, which differ in complexity from internal tissue lipids [51]. More odd-numbered, branched-chain and unsaturated fatty acids are found on the skin surface; chain lengths of C14-C30 are commonly detected [52]. Commensal bacteria have adapted to these easily accessible nutrients on the human skin, because they developed special metabolic abilities to survive and optimize proliferation in this unique ecological niche.

The generation of free fatty acids from skin lipids, catalyzed by secreted lipases, is the first step in the biotransformation of LCFAs. Propionibacteria and aerobic coryneforms have been particularly shown to exhibit strong lipase activity [16]. It has been hypothesized that the subsequent degradation of LCFAs is only partial, so that VFAs, which contribute to human malodour, are generated [16]. Therefore, James et al. tested several skin isolates for their ability to degrade various fatty acids representing the different types found on the human skin surface: saturated, mono-unsaturated and methyl-branched fatty acids. The study demonstrated that species of a certain genus, such as Micrococcus or Brevibacterium, were able to fully catabolize most of the tested substrates. On the other hand, several corynebacteria generated only intermediates or non-utilizable end products of β-oxidation [16]. Because strong malodour was shown to be associated with a dense population of aerobic corynebacteria in the axillae [14], the partial degradation of fatty acids might be a prominent cause of body odour development in humans. So far, no enzymes responsible for partial fatty acid degradation in the human axillae have been identified. Nevertheless, sequencing of the skin isolate C. jeikeium K411 in 2005 and the analysis of its proteome provided insights into the genetic repertoire of aerobic corynebacteria and their β-oxidation-related features [19, 20].

According to its genome annotation, C. jeikeium K411 encodes about 31 proteins potentially involved in fatty acid metabolism: an acyl-CoA synthetase (FadD [11 genes]), an acyl-CoA dehydrogenase (FadE [8 genes]), an enoyl-CoA hydratase (EchA [5 genes]), a hydroxyacyl-CoA dehydrogenase (FadB [2 genes]), a ketoacyl-CoA thiolase (FadA [3 genes]) and a 2,4-dienoyl-CoA reductase (FadH [1 gene]) (Figs 3 and 4). The latter is needed for the metabolism of unsaturated fatty acids with double bonds extending from an even-numbered carbon atom [53]. Additionally, the C. jeikeium K411 genome encodes one acx gene coding for an acyl-CoA oxidase, which catalyzes the second step of β-oxidation. Acx is different from FadE in the reoxidation of FADH2 with O2, which leads to the formation of H2O2 [54].

Figure 3.

 Candidate corynebacterial genes encoding enzymes involved in fatty acid metabolism. Colour code: dark blue, Corynebacterium jeikeium; orange, Corynebacterium urealyticum; rose, Corynebacterium kroppenstedtii; red, Corynebacterium glutamicum; green, Corynebacterium efficiens; purple, Corynebacterium diphtheriae; light blue, Corynebacterium striatum; grey, Corynebacterium aurimucosum. Genes and products: fadD, acyl-CoA synthetase; fadE, acyl-CoA dehydrogenase; echA, enoyl-CoA hydratase; fadB, hydroxyacyl-CoA dehydrogenase; fadA, ketoacyl-CoA thiolase; fadH, 2,4-dienoyl-CoA reductase; acx, acyl-CoA oxidase.

Figure 4.

 Schematic overview of β-oxidation and mycolic acid biosynthesis in corynebacteria. (A) Fatty acids are activated by an acyl-CoA synthetase to form acyl-CoA in an ATP-dependent reaction. The acyl-CoA is oxidized to trans-Δ2-enoyl-CoA by an acyl-CoA dehydrogenase in a reaction that uses the cofactor FAD. Enoyl-CoA is subsequently hydrated by an enoyl-CoA hydratase. A further oxidation step, mediated by a hydroxyacyl-CoA dehydrogenase, converts hydroxyacyl-CoA into ketoacyl-CoA. The last step of β-oxidation is catalyzed by a ketoacyl-CoA thiolase leading to an acyl-CoA with a shortened acyl chain (n-2) and acetyl-CoA. The acetyl-CoA can then be used in the tricarboxylic acid cycle (TCA) to generate energy, whereas the shortened acyl-CoA is used in a new round of β-oxidation. (B) In corynebacteria, two C16 fatty acids are first condensed and afterwards reduced to form a C32 mycolic acid, which can either associate with the outer lipid bilayer (inner leaflet) or with trehalose to form non-covalently bound glycolipids (outer leaflet). Adapted from Portevin et al. [71] and Lea-Smith et al. [72].

Although this comprehensive set of 31 enzymes encoded in the C. jeikeium K411 genome and potentially contributing to human body odour formation has been described, the pathways of fatty acid degradation and their regulation have not been elucidated until now. To analyse in detail the characteristics and function of this comprehensive set of genes and to evaluate the potential role of fatty acid-degrading enzymes in odour formation, we screened the previously sequenced genomes of eight corynebacteria species for genes associated with fatty acid metabolism. Genes homologous between C. jeikeium, Corynebacterium urealyticum, Corynebacterium aurimucosum, Corynebacterium kroppenstedtii, C. striatum, Corynebacterium diphtheriae, Corynebacterium efficiens and Corynebacterium glutamicum were identified using the edgar software [55] and the ‘Gene Ortholog Neighborhoods’ tool from the DOE Joint Genome Institute [56].

In total, we identified 51 genes in the corynebacterial genomes coding for an acyl-CoA-synthetase including the fadD1 genes, which are known to be necessary for mycolic acid biosynthesis (Fig. 3) [57]. As an exception, C. kroppenstedtii lacks a fadD1 gene, which is consistent with the absence of mycolic acids in its cell wall. This species is, similarly to C. jeikeium and C. urealyticum, dependent on lipids, because it lacks a gene coding for a fatty acid synthase. It can be speculated that C. kroppenstedtii might use exogenously available fatty acids mainly for cell membrane assembly and derive its energy from sugars, for which it encodes several sugar uptake phosphotransferase systems [58]. In C. jeikeium and C. urealyticum, these uptake mechanisms are missing, and energy supply is mainly ensured by degradation of fatty acids. The amino acid sequence of FadD enzymes of C. jeikeium is variable, but the characteristic AMP-binding motif of the acyl-CoA synthetase is well conserved among the ten paralogues (FadD2-FadD11) (Fig. 5A). It can be speculated that the overall diversity might be because of different substrate specificities, which would enable C. jeikeium to utilize a broader range of fatty acid substrates in its natural habitat leading to a growth advantage for such well-adapted species.

Figure 5.

 Signature motifs of C. jeikeium K411 fatty acid-metabolizing enzymes. The amino acid sequences of enzymes potentially involved in the fatty acid metabolism of C. jeikeium K411 were aligned using the ClustalX algorithm. Characteristic sequence motifs were obtained from the PROSITE database and are marked with a red box. Enzymes: (A) FadD2-11, acyl-CoA synthetase (PDOC00427, PS00455); (B) FadA1-3, 3-ketoacyl-CoA thiolase (PDOC00092, PS00098); (C) EchA1-5, enoyl-CoA hydratase (PDOC00150, PS00166); (D) FadB1-2, 3-hydroxyacyl-CoA dehydrogenase (PDOC00065, PS00067); and (E) FadE1-8, acyl-CoA dehydrogenase (PDOC00070, PS00072/3).

For the second step of β-oxidation, 22 fadE genes were found in the analysed corynebacterial genomes (Fig. 3). No fadE gene was identified in C. glutamicum, C. striatum and C. diphtheriae. Thus, these species are unable to degrade fatty acids, because the oxidation pathway is interrupted. In C. jeikeium, eight paralogues of fadE are found, all containing the two characteristic FAD-binding motifs (Fig. 5E). Similar to FadD, the amino acid sequence of these paralogues varies substantially, again indicating divers substrate specificities for the eight acyl-CoA dehydrogenases.

In other bacteria such as Escherichia coli or Salmonella, the last three steps of fatty acid degradation are carried out by a multi-enzyme complex encoded by the fadAB operon [59]. FadB from E. coli is characterized by two protein domains conferring enoyl-CoA hydratase and hydroxyacyl-CoA dehydrogenase activity to the enzyme, whereas FadA possesses a thiolase domain only active in combination with FadB [60]. In corynebacteria, the process of β-oxidation has not been investigated in detail, as it is not clear whether FadA and FadB form a multi-enzyme complex. Moreover, corynebacteria encode EchA, an additional enzyme containing an enoyl-CoA hydratase domain (Fig. 5C) [20]. In the corynebacterial genomes, 16 genes coding for EchA were identified. The only species for which an echA gene has not been found is C. aurimucosum (Fig. 3). Several species lack a fadB gene, and, altogether, nine homologues are found in the genomes of C. jeikeium, C. urealyticum, C. kroppenstedtii and C. efficiens. In C. jeikeium, the two FadB enzymes not only differ in size, but also seem to exhibit different enzymatic activities. Whereas FadB1 has a molecular weight of 82.5 kDa and an enoyl-CoA hydratase as well as a hydroxyacyl-CoA dehydrogenase domain, FadB2 is much smaller (30.7 kDa) and contains only the dehydrogenase domain (Fig. 5D). It can be speculated that the interplay between FadB2 and EchA might exhibit the same enzymatic function as FadB1 alone.

For FadA, 15 homologues were detected in the analysed genomes, and only C. glutamicum and C. diphtheriae lack genes coding for the ketoacyl-CoA thiolase. In C. jeikeium, all of the three FadA enzymes possess the characteristic thiolase signature motif, yet they differ in their primary protein structure (Fig. 5B). Interestingly, only the skin isolates C. jeikeium and C. urealyticum encode the 2,4-dienoyl-CoA reductase, enabling the species to utilize mono-unsaturated fatty acids commonly found on the skin surface (Fig. 3) [16].

When examining the whole process of β-oxidation, it becomes clear that several genes are missing in the analysed species. C. glutamicum and C. diphtheriae do not encode homologous genes of fadE, fadB and fadA; C. aurimucosum is lacking echA and fadB, and C. striatum is missing a fadB homologue, which results in the disruption of β-oxidation in these organisms (Fig. 3). The other species, however, have the genetic repertoire for fatty acid oxidation, including the lipophilic species C. jeikeium, C. urealyticum and C. kroppenstedtii, where the dependency on exogenous fatty acids for survival has been demonstrated in vitro [61–63]. For C. efficiens, in contrast, no experimental data are available regarding the ability to utilize fatty acids as sole carbon source. However, it is known that this organism, like its close relative C. glutamicum, is able to synthesize fatty acids de novo from acetyl-CoA [64]. This is also true for the species C. diphtheriae, C. aurimucosum and C. striatum, which is consistent with the non-lipophilic lifestyle of these five species.

Strikingly, close examination of the distribution of the comprehensive set of fatty acid-metabolizing genes within the C. jeikeium genome revealed that some of the genes are grouped in clusters (Fig. 6A). The first cluster is the mycolic acid biosynthesis cluster, which has been described in detail in C. glutamicum (Fig. 6B) [65]. It includes genes coding for an acyl-CoA carboxylase (accD3), a polyketide synthase (pks13), an acyl-CoA synthetase (fadD1), an envelope lipids regulation factor (elrF), a trehalose corynomycol transferase subunit A and B (cmtB and cmtC) and an additional hypothetical protein (jk0135). In particular, pks and accD3 have been shown to be essential for mycolic acid biosynthesis in C. glutamicum, because deletion of these two loci leads to loss of extractable and cell wall-bound mycolic acids [65]. The second cluster includes genes coding for an acetyl-CoA carboxylase (accD1 and accBC1), an acyl-CoA dehydrogenase (fadE8), a hypothetical enoyl-CoA hydratase (jk1548), a citrate lyase β-subunit (citE), an acyl-CoA synthetase (fadD5), an aycl-CoA:3-ketoacid-CoA transferase (scoA and scoB), a ketoacyl-CoA thiolase (fadA3) and two hypothetical proteins with transmembrane domains (jk1541 and jk1542) (Fig. 6C). This cluster and gene composition has not been previously described for other bacteria, yet some genes are found in a similar arrangement in Mycobacterium tuberculosis H37Rv [66]. An enzymatic function in fatty acid biosynthesis has been proposed for the CitE protein of M. tuberculosis. In C. jeikeium, it might play a different role, considering its lipid-dependent lifestyle. Most of the genes encoded in this cluster seem to be linked to fatty acid degradation rather than biosynthesis, which leads to the assumption that this arrangement of genes might be a special fatty acid degradation cluster. Interestingly, the gene preceding the cluster is predicted to encode a TetR-like repressor, which might be involved in repression of the downstream genes. The role of this particular cluster in fatty acid degradation needs to be elucidated in the future to understand the complex mechanisms of lipid utilization in C. jeikeium.

Figure 6.

 Genomic location of genes involved in the fatty acid metabolism of Corynebacterium jeikeium K411. (A) Linear scheme of the C. jeikeium K411 genome showing the approximate location and distribution of candidate genes involved in fatty acid metabolism. (B) The mycolic acid biosynthesis cluster of C. jeikeium K411 (1) and Corynebacterium glutamicum ATCC 13032 (2). Genes and products: cmtB/cmytB, corynomycol transferase B; cmtC/cmytA, corynomycol transferase C/A; cg3185, transposase; jk0135/cg3181, putative secreted protein; fadD, acyl-CoA synthetase; pks, polyketide synthase; accD, acyl-CoA carboxylase β-subunit. (C) Gene cluster composed of several genes potentially involved in fatty acid degradation in C. jeikeium (1), including genes homologous to the citE operon region of Mycobacterium tuberculosis H37Rv (2) [66]. Genes and products: jk1541/jk1542, putative membrane protein; fadA, acyl-CoA thiolase; scoA/scoB, acyl-CoA:3-ketoacid-CoA tranferease; fadD, acyl-CoA synthetase; citE, citrate lyase β-chain; jk1548, hypothetical protein; fadE, enoyl-CoA dehydrogenase; accBC/accA, acyl-CoA carboxylase α-subunit; accD, acyl-CoA carboxylase β-subunit; jk1552/Rv2506, putative TetR-like transcriptional regulator; pdhA, pyruvate dehydrogenase α-subunit; and Rv2499c, possible oxidase regulatory-related protein.

Altogether, it becomes clear that the comprehensive repertoire of fatty acid-metabolizing enzymes seems to be a unique trait of lipophilic aerobic corynebacteria found on the human skin. These results are consistent with previous in vitro studies of James et al. [16], highlighting that corynebacteria can be differentiated on the basis of their ability to catabolize fatty acids (group A and B) as well as their lipophilicity. Various enzymes encoded in the genome of C. jeikeium represent potential targets for fighting human axillary odour with selective deodorant additives. Therefore, further research is necessary to elucidate not only the molecular processes taking place in human axillae, but also the intracellular metabolic reactions of bacteria residing on the human axillae.

Conclusions and prospects

During the past decades, the field of body odour research has become increasingly popular. Several novel routes of human malodour formation and molecular processes catalysed by enzymes of skin resident bacteria have been dissected. However, little is still known about the formation of odorous steroid derivatives. Further research is necessary to fully comprehend the molecular processes underlying steroid biotransformation. In contrast, detailed knowledge has been gained concerning the generation of volatile malodour compounds released after cleavage of Gln- or Gly-Cys-(S)-conjugates by bacterial enzymes. However, several questions remain to be answered. For instance, why are these compounds produced, and what is the origin of their precursor molecules? Nonetheless, the process of body odour formation depends on the interplay between the commensal bacteria with their catalytic enzymes and the unique sweat secretions of the human host.

Despite all of the recently gathered knowledge on the process of Gln-conjugate-biotransformation, much remains to be unveiled. The origin of the Gln-conjugates in the human body has not been elucidated yet. Moreover, it is still not clear whether these molecules are produced in the axillary apocrine glands or whether they are first synthesized in a different organ of the human body and later transported to the underarm region. The distribution of the mutated ABCC11 gene within certain ethnical groups suggests that odour formation in humans has evolved. To better understand the role of body odour in the evolutionary history of man, further research is warranted.

For the Gly-Cys-(S)-conjugates, it has been suggested that the molecules are generated via the glutathione detoxification pathway in human cells [48]. This glutathione S-transferase (GST)-dependent pathway has been studied in plants, mammals and bacteria. The mitochondrial glutathione S-transferase kappa 1 (GSTK1-1) has been suggested to participate in the catalytic activity, transport and folding of proteins involved in lipid metabolism [67, 68]. Interestingly, Wilke et al. noted that secretory cells of apocrine sweat glands are exceptionally rich in mitochondria and different granules, supporting a connection between that organelle and the formation of Cys-Gly-(S)-conjugates by the glutathione detoxification pathway [9]. Under emotional stress, strong body odour formation is often observed instantly, and it has been shown that hormones can trigger molecular processes in secretory cells of apocrine sweat glands [69]. These might also include GSTK1-1 activity, ultimately leading to the synthesis of malodour precursor. However, no experimental evidence has been obtained so far. In future, it would be of interest to elucidate the relationship between the glutathione detoxification pathway, lipid metabolism and body odour formation in humans.

On the bacterial side, further research is needed to identify the enzymes involved in steroid transformation and lipid degradation in the human axillae. More attention should be focused on understanding the production of VFAs by aerobic corynebacteria, because there is only limited information about the molecular and regulatory processes taking place at the cellular level. Elucidation of skin microbial fatty acid degradation pathways and application of next-generation sequencing methods, such as metatranscriptomics, will allow a better understanding of the pathways involved in human body odour formation. Furthermore, the complex structure of the axillary community and the enzymatic interactions responsible for the biotransformation of sweat components should be carefully examined to attain a comprehensive understanding of human body odour formation. This knowledge will ultimately be applied to the cosmetic industry and the development of new deodorant additives. These include substances that target specific enzymes or microbes rather than the entire microbial community, thereby preserving the benefits of the natural human axillary microflora.


HB acknowledges the scholarship granted by the CLIB-Graduate Cluster Industrial Biotechnology co-financed by the Ministry of Innovation, Science, and Research of North Rhine Westphalia.