Team:Paris Bettencourt/Project/Old People Smell



2-nonenal is what "old people smell" is: an odor compound unique to the sweat of old individuals. The smell of 2-nonenal is described as a combination of cucumber, orris and fat. Most of the 2-nonenal on the skin surface comes from the breakdown of sweat-secreted omega-7 fatty acids.


The goal of this project is to isolate bacterial strains that are able to digest, scavenge and eliminate 2-nonenal from the skin. These strains could be packaged into a bacterial cosmetic that attenuates old people smell.


  • Isolated 4 bacterial species capable of growing on 2-nonenal as a carbon source.
  • Determined the critical smell detection limit of 2-nonenal.
  • Analyzed 2-nonenal tolerance and degradation in natural isolates.

Introduction Motivation Aims Results

Figure 1. Isolation of 2-nonenal metabolizing strains. Samples were taken from 3 different body sites with an inoculating loop and struck on M9 2-nonenal plates. After 1 week of incubation at 37C, we identified 3 growing colonies: Micrococcus luteus, Curtobacterium flaccumfaciens and Escherichia fergusonii.

Figure 2. 2-nonenal accumulates in elder skin. 2-nonenal is believed to come from the oxidation of palmitoleic acid accelerated by lipid peroxides (Haze et al., 2001)

Figure 3. A word cloud of odor descriptors for 2-nonenal. Words were freely assigned by 17 smellers to 2-nonenal at a concentration of 100 ppm.


Body odor is affected by age. The existence of a characteristic "old person smell" has been long discussed anecdotally and recently confirmed experimentally. Even untrained smellers can classify age by scent alone (Mitro et al., 2012) The presence of 2-nonenal in body odor is correlated with age and detectable only in people over 40 years old (Haze et al., 2001).

2-nonenal is an unsaturated aldehyde smelling of orris, cucumber and fat (Flavornet database,). It is detectable to the human nose at concentrations as low as 3 ppm. The exact source of 2-nonenal in sweat is currently not known, but it is believed to derive from skin-secreted lipids, oxidized by bacterial lipid peroxidases (Mitro et al., 2012).

The appearance of 2-nonenal in sweat with age has no negative health consequences. The perception of 2-nonenal is subjective, and may be pleasant or unpleasant depending on the smeller and the context. However, "old person smell" can carry a social stigma and some people may prefer to remove it from their body odor for cosmetic reasons. Also, 2-nonenal can be a case study in targeted microbiome modulation. If we can selectively remove this molecule, we may learn to alter other properties of the complex skin surface.

Figure 4. Diffusion model of 2-nonenal through air. This animation shows the concentration of 2-nonenal diffusing through air to equilibrium. 2-nonenal rapidly diffuses to reach the detection threshold of the human nose, even at low skin surface concentrations (2.6 +/- 3.5 ppm).

Figure 5. M. luteus grows on M9-nonenal media. Colonies of M. luteus were struck out and grown for 5 days at 37C. E. coli, the control strain, grows on LB media but not M9-nonenal.

Box 1. Selected information about strains isolated on 2-nonenal.

Figure 6. Humans can detect 2-nonenal at 100 ppm. 12 human smellers were asked to identify pure 2-nonenal in water at the following fold dilutions: 102 104 106 and 108. All samples were solubilized with tween 80 at a concentration of 0.05%. All smell tests were double-blind. Samples are scored by the percentage of smellers who were able to positively distinguish them from tween-only controls.

Figure 7. Smell test results. 15 people smelled the samples of M.luteus or C. flaccumfaciens grown in LB with 625ppm 2-nonenal, or a control sample witn 2-nonenal but no bacteria. The 2-nonenal smell was the least intense in the culture containing M.luteus (p-value: 1.04e-09, HSD turkey test). C. flaccumfaciens was also decreased 2-nonenal smell (p-value: 9.72e-05 HSD turkey test). Error bars represent 95% confidence intervals.


Diffusion Model of 2-nonenal through air

A simple model was created for the diffusion of 2-nonenal into the air using COMSOL Multiphysics (a physics-based interface to solve partial differential equations). The model is shown in Fig.4. From literature, it was found that the odor threshold of 2-nonenal in air is approximately 10-4 mg/kg air (Grosch, 2009). This odor threshold was reached approximately 7 hours in the model, at a distance of 5 cm from the skin surface.

Isolation of strains on 2-nonenal

We sought to isolate bacterial strains adapted to use 2-nonenal as a carbon source. Such strains could hypothetically be used to actively scavenge 2-nonenal from the skin, neutralizing the smell. We chose to look for these strains within the natural human skin flora. Human sweat is rich in fatty acids (Callewaert et al., 2014), and lipophilic phenotypes are common among skin isolates. Natural skin bacteria are pre-adapted to the skin environment, so are more likely to be viable and metabolically active.

We prepared minimal 2-nonenal plates and inoculated them with human skin samples. M9 agar plates were prepared with 0.2% 2-nonenal solubilized with 0.05% tween 80. No other carbon sources were present. An inoculating loop was used to streak plates with samples collected from human forehead, hands, and outer nose. In total we sampled 3 body sites from 8 individuals.

We isolated three bacteria strains capable of growing on 2-nonenal (Fig. 5). Colonies appeared after 1 week. The strains were identified by 16S sequencing with universal primers (Box 1, information on each strain). We chose to focus further experiments on Micrococcus luteus as it performed the best in the M9 the 2-nonenal medium.

M. luteus is a natural human skin bacterium and a well-described oligotroph, or nutrient scavenger. Their genome sequence indicates a complete fatty acid degradation pathway, meaning they can plausibly degrade 2-nonenal.

We next sought to characterize the degradation of 2-nonenal by M. luteus. We first identified the smell detection threshold of 2-nonenal. Human smellers could consistently detect 2-nonenal at a concentration of 1000ppm (Fig. 6).

While this work was in progress, we learned that 2-nonenal is bactericidal for many species at concentrations as low as 1000 ppm (Cho et al., 2004) We therefore determined a threshold of toxicity of 2-nonenal for M. luteus. LB agar plates were prepared with varying concentrations of 2-nonenal. Growth of M. luteus was significantly inhibited at 2-nonenal concentrations above 1250ppm.

2-nonenal smell attenuation

M.luteus in medium containing 625 ppm 2-nonenal lowers the intensity of the ‘old people smell’ significantly (p-value: 1.04e-09, HSD turkey test), while C. flaccumfaciens lowers the 2-nonenal smell mildly (p-value: 9.72e-05, HSD turkey test) in comparison to the medium with 625ppm 2-nonenal medium only (Fig. 7).


Diffusion model of 2-nonenal through air

The model is based on a 2D axisymmetric geometry, with the air modeled as the space 5 cm around the skin surface. Here, the skin surface is kept at a constant concentration, C0 = 2.6 +/- 3.5 ppm (in terms of concentration of skin surface lipids) (Haze et al., 2001). The diffusivity of 2-nonenal through the air was calculated using the Chapman-Enskog theory of gas diffusivity given by the following equation:

D = (1.858E-3 * T(3/2) * (1/M1 + 1/M2)(1/2)) / (p * s122 * w)

where D is the diffusivity of the 2-nonenal
T is the room temperature, 298 K
p is the air pressure, 1 atm
M1 is the mass of the 2-nonenal, 140.22 g/mol
M2 is the mass of the air, 29 g/mol
s12 is the average collision diameter, which is around 340 Angstroms for gas molecules in air, and w is the dimensionless temperature-dependent collision integral, usually on the order of 1.

Thus, the diffusion coefficient of 2-nonenal was tabulated to be 1.687E-9 m2/s.

The odor threshold was a value found in literature for the diffusion of (E)-2-nonenal (Grosch, 2009).

Isolation of strains on 2-nonenal
M9 minimal agar with 0.2% 2-nonenal was used for selection of 2-nonenal metabolizing strains. 0.05% tween 80 was used to solubilize the 2-nonenal. No other carbon sources were present. An inoculating loop was used to streak plates with samples collected from human forehead, hands, outer nose and armpit. In total we sampled 3 body sites from 8 individuals. The plates were incubated for one week at 37 °C and then checked for the appearance of colonies.

Colony PCR and 16S sequencing
Colony PCR was performed with universal 16S rRNA specific primers 8F (5'-AGA GTT TGA TCC TGG CTC AG) and 1492R (5'-CGG TTA CCT TGT TAC GAC TT). Sanger sequencing was performed commercially by GATC Biotech (Germany). Strains were identified by aligning the sequenced DNA fragments to the Greengenes database. Hits with the highest similarity were used to name isolates.

2-nonenal detection threshold
Serial dilutions of pure 2-nonenal were prepared in water at dilution factors of 102 104 106 and 108. All samples were solubilized with tween 80 at a concentration of 0.05%. All smell tests were double-blind. Samples were scored by the percentage of smellers who were able to positively distinguish them from tween-only controls. Samples were evaluated by 12 volunteers 20-30 years old, without serious smell disability.

Toxicity threshold
Toxicity of 2-nonenal was evaluated at a range of concentrations (300 to 10000 ppm) on LB agar plates. M. luteus liquid cultures were grown to saturation in LB, serially diluted, and plated on LB - 2-nonenal. CFU counts were determined following overnight incubation.

Attenuation of 2-nonenal odor
2-nonenal was dissolved in LB at 625 ppm, near the sensitivity threshold we determined for the human nose. Samples were solubilized with tween 80 at 0.05%. LB-nonenal was innoculated with M. luteus, C. flaccumfaciens, or left as an untreated control. Samples were allowed to incubate overnight at 37 °C. 20 individuals scored all the three samples according to the scale 0 - no smell ; 1 - light smell ; 2 - medium smell, 3 - strong smell. All smell tests used randomized tube labels and opaque tubes to hide the appearance of the media. One-way anova was performed to compare mean scores of the three samples followed by HSD Tukey multiple comparisons of means with 95% family-wise confidence levels.

Centre for Research and Interdisciplinarity (CRI)
Faculty of Medicine Cochin Port-Royal, South wing, 2nd floor
Paris Descartes University
24, rue du Faubourg Saint Jacques
75014 Paris, France
+33 1 44 41 25 22/25
Copyright (c) 2014 All rights reserved.