Effect of the skincare product on facial skin microbial structure and biophysical parameters: A pilot study

Abstract Daily use of cosmetics is known to affect the skin microbiome. This study aimed to determine the bacterial community structure and skin biophysical parameters following the daily application of a skincare product on the face. Twenty‐five Korean women, who used the same skincare product for four weeks participated in the study. During this period, skin hydration, texture, sebum content, and pH were measured, and skin swab samples were collected on the cheeks. The microbiota was analyzed using the MiSeq system. Through these experiments, bacterial diversity in facial skin increased and the microbial community changed after four weeks of skincare product application. The relative abundance of Cutibacterium and Staphylococcus increased, significant changes in specific bacterial modules of the skin microbial network were observed, and skin hydration and texture improved. It was suggested that daily use of skincare products could affect the microbial structure of facial skin as well as the biophysical properties of the facial skin. These findings expand our understanding of the role of skincare products on the skin environment.


| INTRODUC TI ON
The skin is an ecosystem comprising various host structures and colonizing microorganisms, including bacteria, fungi, and viruses. Its composition is unique to each person and part of the body (Byrd et al., 2018;Perez et al., 2016). Through evolution, skin microorganisms have adapted to individual host environments and cells. On the one hand, the skin provides nutrients and abiotic factors (e.g., temperature and humidity) that let skin microorganisms grow (Findley et al., 2013;Kong, 2011); on the other hand, these microorganisms prevent the colonization of pathogens, directly and indirectly benefiting the host (Schommer & Gallo, 2013).
The development of next-generation sequencing techniques has facilitated the study of the human microbiome, first in the gut and later on also on the skin. As a result, inter-personal or intra-personal skin microbiome diversity has been revealed. Specifically, the structure of the skin microbiome has been seen to vary depending on the environment Leung et al., 2020), gender (Ross et al., 2017;Ying et al., 2015), race , and age Shibagaki et al., 2017), as well as over time. Several studies have shown that long-term stability reflects the initial status of the microbiome and the host's specific lifestyle (Flores et al., 2014;Grice et al., 2009;Hillebrand et al., 2021;Oh et al., 2016). Furthermore, a balanced skin microbiome is known to play an important role in skin health, as any alterations lead to the overgrowth of pathogenic strains linked to various skin diseases Kong et al., 2012;Williams & Gallo, 2015). The daily use of cosmetics might also affect the skin microbiome, and may be determined by product type, duration of use, and participant characteristics (Ciardiello et al., 2020;Lee et al., 2018;Two et al., 2016;Wallen-Russell, 2019).
While most research in this sense has focused on microbial diversity or changes to individual bacterial strains, little is known about the impact of cosmetics on the overall microbial structure.
In the present study, the whole microbial community structure was analyzed, and biophysical parameters of the skin were measured following the use of a skincare product in Korean women.
The co-occurrence network between bacterial community and skin biophysical parameters offers a broad understanding of the role of cosmetics on the skin ecosystem.

| Participant recruitment and study design
Twenty-five healthy Korean women between 30 and 58 years of age, and residing in Daejeon were recruited in this study (average age: 43 years). Participants who met the following conditions were excluded: (1) were pregnant or lactating; (2) had a lesion like spots, acne, erythema, or atopic dermatitis at the test site; (3) had infectious skin disease; (4) were sensitive to cosmetics, pharmaceuticals, or daily exposure to light; and (5) had undergone skin treatment (scaling, fillers, botox, laser treatment, etc.) within 3 months. The essence type of a moisturizing skincare product (su:m37° Secret Essence, LG Household & Health Care Ltd) was provided to all volunteers and the ingredients were listed in Table A1. Participants were asked to apply the skincare product on their face twice a day (morning and evening) after facial washing with their cleanser for four weeks. They were allowed to maintain their own skincare routines except for prohibiting the use of antibiotics, steroids, and cosmetics with similar formulations or ingredients to the target product. Swab sampling and measurements of skin biophysical parameters were performed on the cheek (previously unwashed for at least 8 h) three times during the experiment: before the use of skincare product (T0), and two (T2) and four (T4) weeks after. Before measurements, the participants relaxed under constant temperature and humidity conditions (indoor temperature 20-25°C, humidity 40-60%) for at least 30 min while the weather was dry and cold during this study ( Figure A1).

| Measurements of skin biophysical parameters
Skin biophysical parameters, including hydration, texture, sebum content, and pH, were assessed. Skin hydration levels were measured using a Corneometer ® (Courage+Khazaka electronic GmbH, Köln, Germany) and expressed as arbitrary units (A.U.). Facial skin texture was analyzed with a Visioscan® camera (Courage+Khazaka electronic GmbH) and expressed as SEr (roughness) values. Sebum content was measured using the Visioscan® camera and Sebufix® F 16 foil (Courage+Khazaka electronic GmbH) and was expressed as Area %. Lastly, facial pH was measured with a skin pH meter (Courage+Khazaka electronic GmbH).

| Amplification of the V3-4 region of the 16S rRNA gene
To analyze the microbiome community, the variable V3-4 region (approximately 400-500 bp) of the 16S ribosomal RNA (rRNA) gene was amplified. The primer sets used for PCR amplification of target genes in the V3-4 region are listed in Table 1. The PCR reaction mixture (25 μl total volume) contained 10 ng of DNA template, 2.5 μl of 16S-v34 Fs (5.0 μM) primers, 2.5 μl of 16S-v34-R (5.0 μM) primers, and 12.5 μl of 2X KAPA HiFi HotStart ReadyMix (Roche, Midrand, South Africa). The PCR thermal profile consisted of an initial denaturation step at 96°C for 3 min, followed by 25 cycles at 96°C for 30 s, 55°C for 30 s, and 72°C for 30 s, as well as a final step at 72°C for 7 min. The PCR products were cleaned with AMPure XP beads (Beckman Coulter) using a 1.4× ratio, quantified using Picogreen fluorescence (Life Technologies), and volume-adjusted prior to the second round of PCR. The PCR products were used to construct 16S rDNA gene libraries according to guidelines for sequencing on the MiSeq System (Illumina Inc.).
Each OTU was assigned a taxonomy ID based on the Silva and NCBI databases using RDP classifiers (Soergel et al., 2012;Wang et al., 2007). The alpha_diversity.py program of QIIME was applied to analyze alpha diversity (Gotelli & Colwell, 2001). Differences between samples were evaluated using the Mann-Whitney U-test and Kruskal-Wallis test in R. Statistical significance was set at p < 0.05.
Canonical correspondence analysis (CCA) was performed to assess the correlation between skin biophysical parameters and microbial composition using the Bray-Curtis distance matrix in R. Significance was evaluated using the permutation test in R.
Spearman's correlation coefficient was applied to evaluate the correlation among bacteria or between the skin biophysical parameters and bacteria. OTUs with less than 25% prevalence in each group were trimmed. Microbial networks were constructed using the Gephi program (Bastian et al., 2009), with the following criteria: threshold = 0.6 and adjusted p < 0.05. The modularity of the bacterial network was visualized by a heatmap using the pheatmap package in R. The correlation between bacteria and skin biophysical parameters (Spearman's correlation threshold = 0.4, adjusted p < 0.05) was visualized in Cytoscape (Shannon et al., 2003). Bacterial profiles in each sampling group were compared by linear discriminant analysis effect size (LEfSe).
Microbiota composition was compared at the order and genus level (Figure 1c,d). The dominant orders differed with sampling time (Figure 1c), particularly between T0 samples and those collected after use of the skincare product (T2 and T4). Before treatment, Rickettsiales was the main order (average 31.57%) but was rarely found after treatment (< 1% in T2 and T4). In contrast, Bacillales, a common order in human skin, was more abundant in T2 samples (20.30%) than T0 (6.92%) and T4 ( Figure 2b).

| Effect of the skincare product on skin biophysical parameters
Every two weeks, biophysical parameters, including facial skin hydration, texture, sebum amount, and pH were measured (Figure 3).

| Correlation between skin microbial community and skin biophysical parameters
CCA was used to investigate the correlation between the skin microbiome and skin biophysical parameters (Figure 4a). The total inertia of the CCA plot was 3.17; whereas the constrained inertia was 0.81, of which 18.8% was explained by the CCA1 axis and 2.1% by CCA2. In the CCA To investigate the exact relationship between skin microbial composition and its biophysical parameters, Spearman's correlation matrix was applied ( Figure 5). In the correlation network, the occurrence of Propionibacteriales, Corynebacteriales, and Bacillaceae orders was associated with improved skin texture (r = 0.49, 0.47, and 0.76, respectively, p < 0.05, Figure 5a). In contrast, Clostridiales (r = −0.63) and Verrucomicrobiales (r = −0.53), both of which were members of module 1 in Figure 2b, correlated negatively with skin texture. Pseudomonadales and Actinomyces were related to skin hydration (r = 0.52, 0.43, respectively, p < 0.05). As shown in Figure 5b, the correlations between skin biophysical parameters and genera were mostly consistent with those at the order level. While Snodgrassella, which was not detected at the order level, was associated with sebum content (r = 0.46).

| DISCUSS ION
The present study indicates that daily use of the skincare product could affect the skin microbiome. Shannon diversity increased after the use of the skincare product (Figure 1b), confirming an earlier finding by Ciardiello et al. (2020), who revealed that the use of cosmetics elevated skin microbial alpha diversity .
High alpha diversity is considered a hallmark of a healthy skin microbiome, as indicated by lower alpha diversity in damaged (Grice & Segre, 2011) and aged  skin. Therefore, according to our results, the skincare product might improve the microbial health of facial skin. immunity (Naik et al., 2012). In particular, C. acnes and S. epidermidis are known to prevent pathogenic bacterial colonization (Cogen et al., 2008;Fournière et al., 2020). More specifically, these two strains can directly inhibit pathogen growth by producing bacteriocins and competing for nutrients with other bacteria (Sanford & Gallo, 2013).
Moreover, C. acnes and S. epidermidis can stimulate human keratinocytes and sebocytes to produce antimicrobial peptides and maintain a balanced skin microbiome (Gallo & Nakatsuji, 2011). Consequently, we hypothesize that the skincare product upheld the microbial equilibrium by favoring the growth of Cutibacterium and Staphylococcus.
Notably, the negative correlation in network analysis did not represent direct competition among microbiota, and the relative abundance of these taxa varied among participants. To overcome this inherent limitation of network analysis, future studies should investigate direct interactions among microorganisms via co-culture experiments.
Generally, modularity in a co-occurrence network refers to phylogenetically close species or microbes inhabiting similar habitats (Olesen et al., 2007), and may disclose functional roles in bacterial ecosystems (Lurgi et al., 2019). Modules appearing in a specific F I G U R E 5 Correlation between skin microbiota and skin biophysical parameters. (a) Order level. (b) Genus level. Node colors in each network correspond to (a) phylum and (b) order level. Edge colors indicate Spearman's correlation coefficient environment are thought to exert specific functions, explaining why a shift in gut bacterial clusters might affect human health (Baldassano & Bassett, 2016;Liu et al., 2019). In the present study, the abundance of modules 2 and 3 gradually increased following the application of the skincare product ( Figure 2b). Therefore, we conclude that the skincare product created an environment optimized for the growth of these modules. Moreover, microorganisms belonging to modules 2 and 3 might exert specific functions in balancing the skin ecosystem. The shift in bacterial modules caused by the use of skincare products has not been investigated extensively. Further studies should be performed to clarify the function of microbes in such modules and, specifically, how they affect the human skin.
The skin environment, defined by its hydration level, smoothness, sebum secretion, and pH, tends to affect skin health (Mukherjee et al., 2016). Dry skin is known to cause skin irritation and aging (Flynn et al., 2001;Rawlings & Matts, 2005), and elevated pH can lead to skin disease (Youn et al., 2013). Among various skin biophysical parameters, hydration level and smoothness were gradually improved after the use of the skincare product, while pH and sebum content were maintained ( Figure 3). These results suggest that using skincare products might preserve skin health.
The correlation between skin microbial community and skin biophysical parameters revealed that hydration and texture were related to skin bacterial community composition (Figure 4a), with some orders and genera displaying significant correlations ( Figure 5). Chang et al. suggested that the order level was the most appropriate for selecting a microbial indicator representative of a specific environment (Chang et al., 2017). The Actinomycetales and Pseudomonadales orders belonging to module 3 correlated significantly with skin hydration levels (Figure 2b, Figure 5a). Accordingly, we speculate that microorganisms belonging to module 3 grew well on hydrated skin, even though the function and role of these bacteria on the face have not been characterized yet. The genus Delftia also exhibited a significant correlation with skin hydration (r = 0.43, Figure 5b), which is consistent with previous findings (Wallen-Russell, 2019). Delftia and Ralstonia, two genera of the Burkholderiales, have been reported as major bacterial taxa on the human face and forearm (Gao et al., 2007;Grice et al., 2008). Nevertheless, further studies are required to understand the relationship between these bacteria and skin hydration.
Various studies have investigated whether cosmetics could affect skin commensal microbes (Bouslimani et al., 2019;Wallen-Russell, 2019;Williams & Gallo, 2015;Xu et al., 2020). Lee et al. showed that certain bacteria grew by metabolizing cosmetic ingredients . Here, we found that the use of a skincare product could improve the skin environment, as well as change its microbiome structure. In particular, the appearance and rise of specific bacterial modules, and their relationship with skin biophysical parameters were observed. These microorganisms can be used as bio-indicators of facial skin conditions and a healthy skin microbial ecosystem. Furthermore, additional studies should be conducted to identify the main components or materials that are associated with changes in the skin microbiome and biophysical parameters.

ACK N OWLED G EM ENTS
The authors would like to thank 3BIGS Co., Ltd. (South Korea) for technical assistance during the analysis of sequencing data.

CO N FLI C T O F I NTE R E S T
All authors are employees of LG Household and Health Care.

E TH I C S S TATEM ENT
The study was conducted in compliance with the Declaration of