Mapping axillary microbiota responsible for body odours using a culture-independent approach

Abstract

Background: Human axillary odour is commonly attributed to the bacterial degradation of precursors in sweat secretions. To assess the role of bacterial communities in the formation of body odours, we used a culture-independent approach to study axillary skin microbiota and correlated these data with olfactory analysis.

Results: Twenty-four Caucasian male and female volunteers and four assessors showed that the underarms of non-antiperspirant (non-AP) users have significantly higher global sweat odour intensities and harboured on average about 50 times more bacteria than those of AP users. Global sweat odour and odour descriptors sulfury-cat urine and acid-spicy generally increased from the morning to the afternoon sessions. Among non-AP users, male underarm odours were judged higher in intensity with higher fatty and acid-spicy odours and higher bacterial loads. Although the content of odour precursors in underarm secretions varied widely among individuals, males had a higher acid: sulfur precursor ratio than females did. No direct correlations were found between measured precursor concentration and sweat odours. High-throughput sequencing targeting the 16S rRNA genes of underarm bacteria collected from 11 non-AP users (six females and five males) confirmed the strong dominance of the phyla Firmicutes and Actinobacteria, with 96% of sequences assigned to the genera Staphylococcus, Corynebacterium and Propionibacterium. The proportion of several bacterial taxa showed significant variation between males and females. The genera Anaerococcus and Peptoniphilus and the operational taxonomic units (OTUs) from Staphylococcus haemolyticus and the genus Corynebacterium were more represented in males than in females. The genera Corynebacterium and Propionibacterium were correlated and anti-correlated, respectively, with body odours. Within the genus Staphylococcus, different OTUs were either positively or negatively correlated with axillary odour. The relative abundance of five OTUs (three assigned to S. hominis and one each to Corynebacterium tuberculostearicum and Anaerococcus) were positively correlated with at least one underarm olfactory descriptor.

Conclusions: Positive and negative correlations between bacterial taxa found at the phylum, genus and OTU levels suggest the existence of mutualism and competition among skin bacteria. Such interactions, and the types and quantities of underarm bacteria, affect the formation of body odours. These findings open the possibility of developing new solutions for odour control.

Keywords: 16S rRNA gene sequencing; Bacterial communities; Skin.

Figure 1

Comparisons of underarm global sweat odour intensities. Data for the left and right axillae were averaged for the four assessors (circles), and then, the median values (thick, horizontal lines) were calculated for (A) each subject or (B) each session (T1–T4) with respect to the sex and antiperspirant user type. Odour intensity was evaluated on a scale from 0 to 10. T1: Morning session on day 1; T2: Afternoon session on day 1; T3: Morning session on day 2; T4: Afternoon session on day 2; AP: Antiperspirant user; Non-AP: Non-antiperspirant user; F: Female; M: Male.

Figure 2

Comparisons of underarm sweat odour intensities. Data for the left and right axillae were averaged for the four assessors (circles), and then, the median values (thick, horizontal lines) were calculated for each subject. Odour intensity was evaluated on a scale from 0 to 10. AP: Antiperspirant user; Non-AP: Non-antiperspirant user; F: Female; M: Male.

Figure 3

Bacterial load on armpit skin as determined by qPCR in DNA extracts. DNA quantity in purified extracts was calculated by using S. aureus MW2 genomic DNA as a reference. The S. aureus MW2 genome weighs approximately 2.9 fg and contains six 16S rDNA copies. The correlation between cycle threshold values and the S. aureus MW2 DNA amount in the qPCR mixture was linear in the 0.1 pg to 1 ng range. T1: Morning session on day 1; T2: Afternoon session on day 1; T3: Morning session on day 2; T4: Afternoon session on day 2; AP: Antiperspirant user; Non-AP: Non-antiperspirant user; F: Female; M: Male.

Figure 4

Relative abundance of predominant bacterial taxa across skin microbiomes. (A) Proportion of 16S sequences assigned to the four major phyla. For each subject, the four sessions (1–4) are presented in the left-to-right direction. (B) Heat map showing the relative abundance of genera across samples. Genera with a relative abundance >0.5% in at least one sample are presented. The four fields within a frame correspond to the four sessions (1–4) in the left-to-right direction. The proportion is indicated by the scale at the bottom of the plot. The analysed subjects were non-antiperspirant users.

Figure 5

Mean intra-individual and inter-individual similarity. The Bray-Curtis similarity was calculated by using the square-root-transformed relative abundance of OTUs. To calculate inter-individual similarity, we compared samples taken at the same session. Error bars correspond to standard errors. *P < 0.05; ***P < 0.001 by two-sided t test; F: Female; M: Male; F-M: Comparison between males and females; Intra: Intra-individual; Inter: Inter-individual. The analysed subjects were non-antiperspirant users.

Figure 6

Principal coordinate analysis for skin bacterial communities. The analyses were performed by using (A) the Bray-Curtis similarity matrix based on the square-root-transformed relative abundance of OTUs and (B) weighted UniFrac based on the OTU relative abundance. The percentage of variation explained is given in parentheses. The analysed subjects were non-antiperspirant users.

Figure 7

Acid-spicy odour intensity as a function of the ‘absolute’ abundance of OTU154509. Odour intensity was evaluated on a scale from 0 to 10. The ‘absolute’ abundance is expressed in arbitrary units obtained by multiplying the relative abundance of the OTUs (fraction of all OTUs determined by pyrosequencing) with DNA quantity in pg determined by qPCR in purified extracts using S. aureus MW2 genomic DNA as a reference. The S. aureus MW2 genome weighs approximately 2.9 fg and contains six 16S rDNA copies.

Scheme 1

Chemical structures of the most important odourless water soluble precursors secreted by apocrine glands and odorant compounds obtained by microbial activity. Precursor 1, N-α-3-hydroxy-3-methylhexanoyl-(L)-glutamine; precursor 2, N-α-3-methylhexenoyl-(L)-glutamine; precursor 3, S-[1-(2-hydroxyethyl)-1-methylbutyl]-(L)-cysteinylglycine; precursor 4, S-[1-(2-hydroxyethyl)-butyl]-(L)-cysteinylglycine.