576 JOURNAL OF COSMETIC SCIENCE
suggested that Corynebacteria must have mutualist trophic relationships with the main skin
colonizers, like Staphylococci. The literature mentions that Corynebacteria colonize all body
sites (dry, sebaceous, moist), even if in low proportions on the skin (10%)1, and Dimitri
et al. (2019) also suggested that some species like C kroppenstedtii are associated with skin
aging and wrinkles. Overall, these data suggest that Corynebacteria are ubiquitous members
of the skin microbiota.
Micrococcus sp. and Streptococci are low abundant species (5%),36,9 yet several studies of the
face microbiota report their presence and describe potentially beneficial roles for the skin.
For instance, Streptococci have been associated to the recovery of the barrier function in vitro
through the secretion of spermidine while Micrococci have been positively associated with
protection against signs of skin aging.37,38
Together, these data indicate that not only C acnes, S epidermidis, Micrococci, Streptococci, and
Corynebacteria are representative of the skin microbiota, but they also play key roles in the
skin homeostasis maintenance. One of the challenges of the present study was to obtain
a consortium with these five species. Despite their different nutritional needs and growth
requirements, an adapted protocol was elaborated that allowed the growth of wild isolates
for these five bacteria.
REPRODUCIBILITY OF THE MODEL
Overall, the species proportions (%)in the coculture after 8 hours of incubation are close
to those that are described in the literature (Figure 4).9 Nevertheless, the total bacterial
concentration in the consortium after 8 h of incubation is of more than 109 CFU/mL.
Thus, percentages (per 102) are not relevant indicators to study the model reproducibility.
Consequently, the concentrations’ variability were analyzed rather than the proportions.
Coefficients of variation are inferior to 4%, indicating a very good reproducibility of the
coculture preparation protocol.
IMPACT OF COSMETIC PRODUCTS ON THE SKIN MICROBIOTA
In order to analyze the effects and compare tested products, the LR on the five species
were calculated. The calculation of LR is common to measure the bactericidal efficacy
of disinfectants or preservatives that indicates the quantity of microorganisms
eliminated by a substance in a given contact time.39,40 When studying biocidal
substances, LR at least equal to five log are requested for a treatment to be effective.
In a challenge test for cosmetic formulations, an LR of three is required on bacteria
to validate the preservative efficacy.41 On the contrary, LRs less than or equal to 1
(LR ≤ 1) are not considered significant. In the present study, a score (sum of the 5
LR) was calculated, in order to aggregate the data on the five species. Consequently,
scores between 0 and –5 (LR ≤ 1) are obtained with neutral ingredients for the
microbiota (microbiota-friendly). Such scores were observed with proline, glycine,
glycerol, glucose, and inulin. All these substances are potential substrates for bacteria
and could have stimulated bacterial growth. However, the contact time of 8 hours
was not long enough to observe a significant increase of bacteria. We concluded that
the model was stable enough to be well adapted to discriminate between neutral and
harmful substances.

577 COSMETIC INGREDIENTS THAT RESPECT SKIN MICROBIOTA
The maximum LR that can be obtained for a species in the present model is –8.5 (maximal
concentration =8.5 log CFU/mL). A score of approximately –42.5 is therefore the maximum
score that can be obtained. Between –5 and –42.5, ingredients have deleterious effects on
the microbiota. Indeed, scores for phenoxyethanol (1%) and parabens (1%) were –28.4 and
–41.9, respectively, which is congruent with their antimicrobial properties. It is interesting
to note that a concentration of 0.5% of phenoxyethanol has no inhibitory effect on the
consortium after 8 hours of contact. A concentration inferior to 1% of phenoxyethanol is
recommended in finished formulas,42 yet inferior concentrations still exhibit antimicrobial
activity. More specifically, Wang et al. showed that S epidermidis and M luteus monocultures
were inhibited by phenoxyethanol at 0.5% in an in vitro minimum inhibitory concentration
(MIC) assay.14 On the contrary, no effect was observed on these bacteria in the present
assay (data not shown). The reduced contact time and the presence of several species in
the consortium, combined with specific growth conditions (lower pH, low temperature)
could explain the low efficacy of phenoxyethanol at a concentration of 0.5%. In general,
antimicrobial substances lead to expected reductions of bacteria in the model, which
indicates that the designed protocol is adapted to test ingredients with unknown effect.
Finally, among the common substances used in cosmetics tested here, most of them
had little effect on the consortium. Pentylene glycol (2.5%), a solvent not listed as an
antimicrobial, reached the score of –3.6. While no significant impact of pentylene glycol
was observed at this concentration, pentylene glycol is known to enhance the efficacy of
preservatives in cosmetic formulas by lowering the water activity.42 Puschman et al. (2018)
also demonstrated that the antimicrobial activity of a preservative (phenoxyethanol) in
emulsion gels is dependent on the lipophilicity of the oil phase and the concentration of
emulsifiers.43 Future studies should focus on testing several ingredients simultaneously, and
possibly complete cosmetic formulas, to have a more precise understanding of the effect of
substance combinations on the skin microbiota, and identify synergistic or antagonistic
effects.
Evolutions of the protocol should also be considered. For instance, increasing the incubation
time could be useful to analyze potential beneficial ingredients. The introduction of
pathogen species or the disruption of the proportions of the consortium would also bring
interesting data on known dysbiosis such as atopic dermatitis or acne vulgaris.
CONCLUSION
As proposed by van Belkum et al., cosmetic products should have the least possible impact
on the skin’s microorganisms.13 To do so, the authors evoke the need of first intention
in vitro prescreening methodologies, and in vivo testing of cosmetic products to assess the
friendliness toward the microbiota. Since in vivo assays are more complex to implement,44
i.e., require extensive toxicity assessment, ethic validation, specific sampling strategies
and complex analysis, one would recommend reserving in vivo assays for finished products
or formulas. Instead, in vitro assays such as the one presented here, are useful to analyze
the precise impact of ingredients composing formulas. To date, there is no consensus
or standard methodology for such evaluations, nor regulation to claim the “microbiota
friendliness” of cosmetic products. The present article is a proposition of an in vitro protocol
to measure the impact of ingredients and, potentially, combinations of ingredients on the
skin’s main bacterial commensals, which brings additional data, complementary to the
safety and efficacy data.