Unpublished data on facial skin microbiome profiles and skin biophysical measurements were extracted from a previously performed single-center study (Campiche and Pascucci, 2023), which investigated the impact of a combination of a Nannochloropsis oculata microalgae extract and high-performance polysaccharide (Pullulan) [PEPHA^®-TIGHT CB] on skin texture, firmness and elasticity of female volunteers. As described therein, female volunteers applied over the course of 7 days twice daily either an active [PEPHA^®-TIGHT CB] or a placebo formulation on their face. Briefly, this study showed that application of [PEPHA^®-TIGHT CB] provided an instant facial skin care benefit compared to usage of a placebo formulation. Compositions of used formulations can be retrieved from (Campiche and Pascucci, 2023).

Healthy, Caucasian female volunteers, aged 40–65, were grouped according to their pre- (n = 14, average age 48.1 ± 3 years) or postmenopausal (n = 30, average age 60.6 ± 4 years) status, defined by whether menstrual cycles were regular or whether the last menstrual period occurred at least 12 months ago, respectively. Perimenopausal and menopausal individuals were excluded from this study, except where explicitly mentioned. In order to minimize confounding factors that are known to impact skin microbiome compositions, strict exclusion criteria for selection of study participants included, but were not limited to, being in the course of immunosuppressive treatment, under long-term treatment, in particular with products containing aspirin and derivatives, anti-inflammatory, antibiotics, antihistamines, steroid hormones, such as those being integral components of hormonal replacement therapies (HRT), beta blockers and/or desensitization drugs. Furthermore, subjects having a skin disease, having facial skin recently exposed to sunlight (natural or artificial) within 2 weeks preceding the inclusion, premenopausal subjects being pregnant, and subjects having applied anti-wrinkle, anti-aging and firming products during the 4 weeks preceding the starting day of the study, were not allowed to participate.

Samples were obtained in accordance with the Declaration of Helsinki. All study volunteers had given their informed consent to participate before enrolment. Skin samples were collected in France (Lyon) from forehead and from randomly chosen either left or right cheek sites using DNA/RNA shield swabs (Zymo Research, Cat #R1106). Samples were taken on day 0 and day 7, totaling 4 samples from each subject. The volunteers were instructed not to wash their faces on the day of sampling. The swab was premoistened in sterile PBS prior to rubbing the skin area of 2 × 2 cm (4 cm² surface area) for 30 s in different directions. Sample material was placed into DNA/RNA shield storage tubes (Zymo Research, Cat #R1106) and stored at −80°C until processing.

DNA was extracted according to standard protocols, Illumina 16S rRNA gene amplicon libraries were generated and sequenced at BaseClear B.V. as described previously (Deyaert et al., 2023).

In short, barcoded amplicons from the V3-V4 region of 16S rRNA genes were generated using a 2-step PCR with the primers 16S‐341F and 16S‐785R. In order to decrease potential biases associated with PCR amplification using 16S rRNA gene V3-V4 primers (Meisel et al., 2016) during this skin microbiome study, Phusion high-fidelity polymerase was used to reduce errors in sequence amplification (Sze and Schloss, 2019). The libraries were barcoded, multiplexed, and sequenced on an Illumina MiSeq machine with a paired-end 300 cycles protocol and indexing. Illumina sequencing data were quality checked and demultiplexed by BaseClear standards, and FASTQ files were generated. For data processing, paired-end sequence reads were merged into pseudoreads through sequence overlap. Chimeric pseudoreads were removed and remaining reads were aligned against the RDP database for bacterial organisms (Cole et al., 2014). Based on the alignment scores of the pseudoreads, the taxonomic depth of the lineage is based on the identity threshold of the rank; Species 99%, Genus 97%, Family 95%, Order 90%, Class 85%, Phylum 80%.

To evaluate aging-associated changes in facial skin biophysical properties of study participants, Cutometer^® measurements were performed according to standardized guidelines. Aberrant values were excluded from statistical analyses reported here. During the baseline visit, the viscoelastic properties of the forehead and cheek skin were evaluated using a Cutometer^® device, which measures the deformation of a cutaneous area submitted to a mechanical suction and its recovering power.

Preprocessing of abundance data was performed using MicrobiomeAnalyst 2.0 (Lu et al., 2023), while calculations and visualizations were performed in a Jupyter notebook in a Python v3.10.11 environment with the following relevant packages installed: Pandas 2.0.0, Numpy v1.23.5, Scipy v1.11.3, scikit-bio v0.5.9, sklearn v1.1.2, Seaborn v0.12.2, Matplotlib v3.7.1.

Diversity and richness were calculated using Shannon’s entropy and Chao1, respectively. Bray-Curtis divergence was used to compute the inter-sample distance. Principal Component Analysis (PCA) was performed on the obtained distance matrix. PERMANOVA (99999 permutations) and PERMDISP (999 permutations) testing was used to assess clustering differences between groups. Unless otherwise stated, non-parametric independent testing using Mann-Whitney U test was performed when comparing different subjects for measures within pre- and post-menopausal groups. Non-parametric paired testing using Wilcoxon test was performed when comparing subject dependent features. Original datasets are available in a publicly accessible repository: Sequence Read Archive (SRA) portal of NCBI under accession number PRJNA1079725.

We designed and executed an observational study to investigate a potential attribution of menopausal status rather than differences in aging skin and the underlying skin biophysical properties to a shift in skin microbial community structures, allowing to minimize influences of other age-associated confounders. With average group ages of 48.1 ± 3 and 60.6 ± 4 years, respectively, representative pre- and postmenopausal study participants were specifically chosen to belong to an age range close to menopausal transition. The median age at menopause among Caucasian women from industrialized nations is 51 (Gold, 2011). Of note, the oldest participants in the premenopausal group were older than the youngest participants in the postmenopausal group (Supplementary Table S1).

Assuming that age-dependent skin physiological parameters are critical determinants for microbial niche selection and population, we would not expect any significant changes in skin microbiome compositions between “younger” pre- and “older” postmenopausal groups, should their measurable skin biophysical conditions, such as viscoelastic properties, be similar. As previously reported, facial sagging is inherently linked with facial skin firmness, and hence is the most accurate predictor of chronological age (Flament et al., 2021), consistent across different ethnical groups (Voegeli et al., 2023). Given that the average age difference between the two examined groups was 12.4 years only, comparison of Cutometer^® readings between pre- and postmenopausal groups (Supplementary Table S1) revealed no significant differences in aging-associated skin elasticity and tiring effects (Figure 1). With the main discriminants between the examined study groups being their respective pre- or postmenopausal status and associated skin physiological parameters, we next examined genomic structures of facial skin bacterial communities.

To investigate an association between skin microbiome composition and menopausal status, we performed 16S rRNA gene amplicon sequencing of bacterial DNA isolated during two timepoints over the course of a week, separately from cheek and forehead skin sites. A total of 2′885′744 reads were analyzed with an average read count of 14′723 per sample, and the sequences were assigned to 5,854 operational taxonomic units (OTUs) based on ≥ 97% sequence identity (Supplementary Table S2). Skin microbiome profiling of each study participant twice during 1 week, on different facial skin sites and in dependence of topical application of the investigational cosmetic product, allowed not only to gain statistical power for comparative data analysis, but also to investigate temporal and intra-personal variability of skin microbiome profiles, as well as to explore the impact of specific extrinsic factors, such as cosmetic product use, on skin bacterial compositions: Topical application of either active or placebo formulations over the course of 1 week had no impact on skin bacterial compositions, as α- and β-diversities (measured by Shannon’s index and Bray-Curtis dissimilarity, respectively) assessed at the taxonomic bacterial genus level were not significantly different between placebo and verum groups after 7 days of applications (Supplementary Figure S1), and as such microbiome-friendliness can therefore be attributed to the formulations. In accordance with previous publications (Oh et al., 2016), skin microbial community structures on forehead and cheek were characterized by longitudinal stability, as α- and β-diversities were not significantly different between baseline sampling on day 0, and day 7 (Supplementary Figure S2). Given the temporal stability and resistance of skin-resident microbial compositions to tested active and placebo formulations, we decided to include all collected microbiome data sets for following analyses.

According to taxonomical assignments, which were performed using the pipeline integrated within the MicrobiomeAnalyst 2.0 package (Lu et al., 2023), the 5 most abundant genera, comprising >79% of the microbiomes, were Cutibacterium, Corynebacterium, Staphylococcus, Snodgrassella, and Streptococcus (Figure 2A). Relative abundances of bacterial genera were differentially distributed on facial skin between pre- and postmenopausal volunteers. For example, the lipophilic Cutibacterium genus was not only the most proportionally represented genus on skin of all study participants, but also significantly more abundant in the premenopausal group (Figure 2B). On the other hand, relative abundances of OTUs affiliated to the Streptococcus genus, which belongs to the Bacillota phylum with preference for skin sites with low sebaceous gland activity, such as infant skin (Luna, 2020), was found to be significantly increased in the postmenopausal group (Figure 2C). Moreover, the Streptococcus genus was previously shown to negatively correlate with facial sebaceous gland area (Howard et al., 2022). This finding aligns well with previously published data, that skin sites enriched with sebaceous lipids attract lipophilic microorganisms (Byrd et al., 2018). Other bacterial genera, whose relative abundances were significantly affected by menopausal statuses, are listed in Supplementary Table S3.

Intriguingly, the α-diversity (Shannon’s index) and β-diversity (Bray-Curtis dissimilarity), assessed at the genus taxonomic level, were significantly different between pre- and postmenopausal groups at sampled facial skin sites (Figures 2D, E). Specifically, bacterial α-diversity was significantly increased in skin samples obtained from postmenopausal volunteers, a finding that is in line with previous reports stating higher α-diversity on ageing skin compared to significantly younger skin of females (Shibagaki et al., 2017; Jugé et al., 2018; Kim et al., 2022; Zhou et al., 2023). The statistical significance regarding differences in menopause-associated α- and β-diversities was maintained even after mathematical removal of the most abundant Cutibacterium genus from the equation (Supplementary Figure S3), suggesting that evenness deflation of the samples caused by dominance of Cutibacterium is not the only driver of bacterial diversification.

The measured skin parameters, such as elasticity and recovery rate, were not significantly different between pre- and postmenopausal groups in our study (Figure 1), whereas skin microbiome profiles were affected by the respective menopausal statuses (Figure 2). As expected, no significant associations between biophysical parameters and bacterial diversity on skin could be detected (Supplementary Figure S4), suggesting that the observed shift in skin microbial profiles can be attributed directly to menopausal status rather than to differences in aging-associated viscoelastic properties.

The topography of skin largely impacts the distribution of microbes, indicating the presence of selection pressure: Sebaceous skin, found, for example, on forehead, contains the lowest microbial diversity, while comparably drier sites, such as cheek skin, favors a broader diversity (Mukherjee et al., 2016). Accordingly, in our study the bacterial α- and β-diversities were significantly higher in cheek than in forehead samples and significantly dissimilar between the skin sites, respectively (Figures 3A, B). Aligning with previous findings (Grice et al., 2009; Findley et al., 2013), the relative abundances of lipophilic microbes, such as those belonging to the Cutibacterium genus, were significantly increased on forehead (Figure 3C), a skin site that is characterized by higher sebaceous gland density compared to cheek skin (Thody and Shuster, 1989).

Given these intra-personal variabilities between facial skin sites we wondered whether the observed menopause-regulated shift in facial skin microbiome profiles (Figure 2) was powered by forehead- and/or cheek-resident bacterial communities. Indeed, differences in bacterial genera distributions (not shown), as well as α- and β-diversities, on cheek skin between pre- and postmenopausal volunteers were reduced (Supplementary Figures S5A, B), compared to forehead skin of the same volunteers (Figures 3D–F). Nevertheless, Chao1 index was significantly increased in cheek skin samples of postmenopausal volunteers (Supplementary Figure S5C), suggesting that rare taxa richness, rather than their even distribution, was the main driver for the increased bacterial diversification. The difference in menopause-associated α-diversity remained statistically significant even after mathematical removal of the Cutibacterium genus from forehead samples (Supplementary Figure S6A), whereas bacterial communities became similar (Supplementary Figure S6B).

A small subject number (n = 4) reported irregular menstrual cycles and as such was classified as perimenopausal and excluded from the main study reported above. Nevertheless, inclusion of this group and subsequent direct comparison with pre- and postmenopausal volunteers revealed that respective bacterial genus-level α-diversities as measured by Shannon’s index in samples were significantly different (Supplementary Figure S7). In agreement with a previous report stating that microbial diversity was reduced on facial skin sites of female volunteers with self-evaluated irregular menstrual periods (Ma et al., 2023), Shannon’s index of perimenopausal skin samples was significantly lower than from pre- and postmenopausal subjects in our study (Supplementary Figure S7).

Worldwide life expectancy is steadily increasing, and the number of postmenopausal women is expected to reach 1.2 billion by 2030 (Afshari et al., 2020). This rapidly growing population now lives one-third of their lifetime in a state of estrogen deficiency, hence there is an unmet need to expand our knowledge on the role of skin microbiome in postmenopausal health. Several gut and vaginal microbiome-targeting therapies exist for treatment of menopause-associated burdens (Park et al., 2023). However, the potential of skin microbiome-modulating strategies to manage skin disorders associated with menopausal states is waiting to be exploited.

We hypothesize that an increased bacterial diversity and potentially associated dysbiosis in the skin microbiome may at least partially contribute to the development of skin disorders experienced by trans- and postmenopausal women. Thus, targeted microbiome-based interventions aiming at rebalancing the skin microbiome composition may provide alternative approaches to traditional symptom reduction or more drastic treatments, such as hormonal replacement therapy (HRT), and hence reduce associated side effects. HRT is often prescribed to women to alleviate adverse physiological changes experienced during menopause. While the effects of HRT on gut and vaginal microbial community structures have been investigated (Dothard et al., 2023), impacts on skin microflora are not known. Given that HRT stimulates sebum secretion (Sator et al., 2001), we would expect the skin of postmenopausal patients using HRT to be repopulated by lipophilic microorganisms, such as Cutibacterium acnes, followed by a decrease in microbiome diversity due to evenness deflation of competing microbial components.

Future studies into better understanding mechanisms of interactions between circulating sex hormonal levels and the skin microbiome are required for the development of non-hormonal, microbiome-targeting therapeutics for treatment of skin and hair menopausal symptoms.