Skin aging‐related microbial types separated by Cutibacterium and α‐diversity

Studies on the skin microbiome have been conducted to uncover the relationship between skin microbes and the host. However, most of these studies have primarily focused on analyzing individual microbial compositions, which has resulted in a limited understanding of the overall relationship.


| INTRODUC TI ON
The relationship between skin aging and microorganisms has been well established, with numerous studies indicating that microorganisms play a role in the aging process of the skin. 1,24][5][6][7] Among these dominant genera, Cutibacterium, which includes Cutibacterium acnes, is the most abundant in many individuals and is known to have age-related characteristics, as it tends to be found in higher concentrations in younger populations. 5ditionally, α-diversity, measured by the Shannon index, has also been shown to be related to age, with low diversity in young individuals and high diversity in the elderly. 5Aside from the relationship with age, these two factors have significant relationships with the skin characteristics, so they are widely used in research on the relationship between skin and microorganisms.
In recent times, advanced analyses including microbial typing have been conducted, which have the potential to yield more comprehensive results and enhanced utility.For example, in a study of the gut microbiome, enterotypes were defined and their association with long-term diet was discovered. 8Also, in the case of skin microbiome, two studies identified the microbial subtypes of the female skin using principal coordinate analysis and functional analysis. 6,7ese studies highlighted the importance of functional studies and provided clues about microbial types on our body.However, the type classifiers in these studies were counter-intuitive and complex to use for actual type classifications.
In this study, we focused on the two most representative indicators, Cutibacterium and the Shannon index, to classify facial skin microbiome types.Rather than comparing the skin microbiome simply by age, we sought to find a meaningful way to classify microbial types using skin aging features.To prevent misunderstandings due to chronological age, age-adjusted analyses were performed.Our results showed that, when compared in the same age group, healthy skin has higher microbial diversity, consistent with the general trend.The samples were then divided into four skin types based on the Shannon index and the proportion of Cutibacterium.Finally, we found significant differences in skin aging characteristics between each type.The type classification system developed in this study improves our ability to interpret and explain the results in the field of microbial research, as well as facilitating the explanation of skin conditions to consumers.

| Study participants and sampling method
A total of 101 healthy Korean women aged 18-70 years old, without any skin diseases such as acne vulgaris and psoriasis, were recruited in Seoul, and their skin microbiomes were sampled from November 18 to November 22, 2019.Basic characteristics of the participants are summarized in Table 1.They were fully informed of the study and provided written consent after reading the research protocol, which was approved by the institutional review board.All research was conducted in accordance with relevant guidelines and regulations.
On the day before sampling, participants were instructed to clean their facial skin before 10:00 p.m. On the following day, their facial skin microbiome was collected from the right cheek using the MySkin® kit (sampling tape) according to the manufacturer's protocol.The sampling was performed at the Research Center of LG H&H without any makeup or cleansing on the skin.Participants were then instructed to cleanse their face and wait in a room with a constant temperature (20-24°C) and humidity (40%-60%) for 30 min.After skin stabilization, the facial skin characteristics were measured using the Janus-III measurement system.

| Measurement of skin characteristics by Janus-III
Using the Janus-III measurement system, three images of the participants' faces were captured using different light sources (normal, polarized, and ultraviolet).The images were then segmented into the forehead, beside left/right eyes, under left/right eyes, nose, and left/right cheeks.In cases of poor segmentation, manual adjustments were made.Pores, wrinkles, and pigmented spots were calculated as the weighted sum of areas separated by color differences with adjacent pixels, with values ranging from 0 to 100.Facial skin color was extracted from the left and right cheeks using the CIE-LAB space.Sebum and porphyrin were evaluated by counting the fluorescent dots above a certain size and dividing them into sebum or porphyrin based on color criteria.The values of sebum and porphyrin were then log-transformed before analysis.

| Bacterial community analysis
The library construction and sequencing were conducted by TAK Inc. (Japan) using the V1-V2 region of the 16S ribosomal RNA (rRNA) gene.During the extraction process, one sample was excluded due to low quality control as determined by the MySkin® protocol.The V1-V2 region of the 16S rRNA gene was amplified using the primers 28F (5′-GRGTT TGA TYM TGG CTCAG-3′) and 338R (5′-TGCTG CCT CCC GTA GGAGT-3′).
On average, 46,987 reads were obtained from 100 samples, and community analysis was performed with an average of 42,763 reads after quality control using DADA2 9 and the QIIME2 10 pipeline.
The analysis was performed at the genus level as assigned by the Greengenes 11 reference database.For evaluating microbial diversity, the Shannon index was used to quantify the number of discovered taxa and their evenness.The Jensen-Shannon divergence was used as a distance metric to measure the β-diversity of the microbiome, considering both the presence/absence of taxonomy and differences in relative abundances, particularly in taxa that are present in multiple samples with drastic differences in abundance.PICRUSt2 12 was used to infer the gene contents of the microbial community and map them to KEGG pathways. 13

| Statistical analysis
All statistical analyses and visualization were carried out using R (version 3.6.1).Linear regressions were used for association tests and residuals calculation with age adjustments because age was strongly associated with both skin characteristics and microbial features such as Cutibacterium and Shannon index.The Kruskal-Wallis test was used to compare skin characteristics (after adjusting for age) between microbial types, while the Jonckheere-Terpstra test was used for ordered trend tests.False-discovery rates were calculated using the Benjamini-Hochberg procedure to adjust for multiple testing.Hierarchical clustering, partitioning around method, and density-based clustering were used to identify distinct microbial clusters and defined types.The functions of skin microbes that were most likely to explain differences among the facial skin microbial types were predicted using linear discriminant analysis (LDA) effect size analysis, 14 with a minimum LDA score of 2.0 and a "one-againstone" testing option.

| Association between skin microbiome and skin characteristics
Linear regression tests were performed to assess the relationship between skin characteristics and representative microbial factors, the Shannon index and Cutibacterium (Table 2).The results showed that facial pigmented spots-polarized light (FPS-PL), sagging pores, pores, LAB-a*, sebum, and porphyrin had negative associations with the Shannon index and positive associations with the proportion of Cutibacterium.LAB-L* showed a negative association only with Cutibacterium.The beta coefficient signs of the linear regression results for the Shannon index and Cutibacterium proportion were opposite directions, but the association significance (p-value) was similar for each skin trait.
In summary, more age-related skin features were observed when the abundance of Cutibacterium was relatively higher and the Shannon index was relatively lower in the subjects.The skin features such as pigment spots, pores, skin redness, and skin darkening were more prominent in Cutibacterium-rich subjects (Table 3).

| Differences in skin characteristics among microbial types
To verify the relevance of type classification in skin aging research, we compared and analyzed differences in skin characteristics based on microbial type (Figure 3) and briefly summarized them in

| Functional prediction of microbial type
We used PICRUSt2 to predict the functional content of the microorganisms and to determine whether the differences in skin aging indices by microbial type are related to the functions of the microorganisms (Figure 4 and Table 3(

| D ISCUSS I ON AND CON CLUS I ON
The skin microbiome has a significant impact on skin aging due to its close interaction with the skin.There is evidence that the skin microbiome can predict age of host better than the gut microbiome, 15 highlighting the strong link between aging and the skin microbiome.
Although many studies suggested age-related microbes, it is not necessarily the case that these microorganisms are directly related to skin health.For example, C. acnes was frequently found on the skin of young individuals, but an abundance of C. acnes does not necessarily indicate young and healthy skin.However, after adjusting for age, we found that the relative abundance of Cutibacterium showed a positive association with skin aging characteristics.This means that when comparing people of the same age, those with more Cutibacterium exhibited more aging-related characteristics such as wrinkles and pigmentation.This phenomenon also applies to comparisons of skin properties between microbial types.
Type C, which is composed of subjects with a high proportion of Cutibacterium, included many young people.If we compare the skin features of type C without age correction, it showed relatively low values of wrinkles and pigmentation.However, this pattern arises because both Cutibacterium and skin aging traits are each affected by actual age.Since the two factors are related to age, the analysis reveals an indirect negative correlation between them, even though in reality, there is a positive correlation.Therefore, to avoid age-induced misunderstandings, age-adjusted residuals were used for comparison when assessing associations between aging traits and microbial types.As seen in the results section, after age adjustment, wrinkles and pigmented spots were relatively higher in type C.
According to the results section, type C exhibited more severe signs of aging, while type B showed less severe aging.These findings suggest a negative association between skin microbial diversity and skin aging, which is in line with the general trend of the association between microbial diversity and health indicators. 3However, it is Predictive functions significantly present in each microbial type.
| 1073 not enough to explain skin aging based solely on microbial diversity.
In response, we examined functional potentials of skin microbiome and compared them to identify valuable explanations for the association between microbial types and skin aging characteristics.
Type C exhibited many distinct metabolic pathways, including carbohydrate and lipid metabolism.Variations in nutrient utilization by the skin microbiota could be attributed to differences in metabolism-associated genes observed between types.Types C and O shared the common features of high sebum and porphyrin.
However, type C had a higher potential for microbial utilization of sebum compared to other skin types, as it is enriched in genes related to lipid metabolism.Cutibacterium, which is abundant in type C skin, secretes lipase that can alter sebum composition and harm hair follicle walls, leading to reactive oxygen species production and inflammation. 16Moreover, free fatty acids produced by lipase can have several negative effects on the skin, such as inducing inflammation or promoting bacterial cell adhesion. 17,18 the contrary, type O had a higher genetic potential related to cellular processes, such as cell growth, death, and motility, than metabolic-related genes.Since these characteristics differed from those of type C, they can be explained the differences in skin features between types C and O.However, since type O was a group of all subjects dominated by microorganisms other than Cutibacterium, it is challenging to conclude that high cellular activity and low metabolic potential are microbial traits common to all type O subjects.As the interpretation of functional potential is limited to possibilities, further studies are necessary to clarify the functional relationship between microbial types and skin phenotypes.
In summary, we developed a classification model for skin microbial types based on two major indicators Shannon index and the relative abundance of Cutibacterium.Our analysis showed significant associations between skin characteristics and the classification of microbial types, particularly Cutibacterium, which has the potential to impact skin aging.Notably, by adjusting for age, we were able to examine the associations between microorganisms and skin health deterioration more clearly.Our classification system offers a simple and intuitive way to explain skin phenomena, making it useful not only for dermatology but also for the cosmetic industry.
While our proposed classification method is user-friendly, it has some limitations.The criteria for classification can vary depending on the study population, and the types of classification need to be further subdivided based on the study's objectives.As mentioned earlier, the associations between microbial types and skin aging patterns do not imply causality.Moreover, the taxonomic resolution achieved by targeting the 16S rRNA gene with short-read sequencing was limited to the genus level, as compared to the higher resolution provided by entire gene sequencing. 19 our study, we specifically focused on the microbial community of the cheek, known for its relatively high sebum content compared to other skin sites.While the facial skin microbiota includes not only bacteria but also fungi and viruses, bacteria were found to be more that the specific concentrations of sweat and sebum exerted a notable influence on the composition of the microbial community, primarily due to variations in nutrient availability for each microorganism. 21Cutibacterium, in particular, metabolizes sebum components to support its own growth 22 and regulate the growth of other microbes. 23Consequently, modulating sebum production has the potential to induce changes in the defined microbial types within our study.
Lifestyle factors such as BMI, dietary intake, smoking, and alcohol consumption have also showed associations with the skin microbiome. 24Additionally, the changes in microbial composition and diversity by therapeutic methods including phototherapy, 25 We applied clustering methods to determine the microbial types of facial skin and compared the results.Through hierarchical clustering of 100 samples based on the relative abundance of skin bacteria, three distinct clusters were identified, as shown by the thick black squares in Figure 1.The Shannon index decreased from left to right, and this decrease in diversity generally coincided with the increase in Cutibacterium, except for the rightmost group, where Neisseriaceae (classified only at the family level) were dominant.As results of applying clustering methods, groups were distinguished based on the proportion of Cutibacterium and the Shannon index, and the isolated group dominated by other genera was clearly separated from other groups (Figure 2A-H).Accordingly, we defined the microbial types using the two most prominent factors in the facial skin microbiome, the Shannon index and the abundance of Cutibacterium, as classification criteria (Figure 2I).A regression line of 4C + S = 4 was found between the Shannon index and the proportion TA B L E 1 Characteristics of study participants.
c)).A total of 34 pathways showed significant results, with approximately 10 pathways each appearing significantly for types C, B, and O, and no prominent pathways for type CB.Type C and type B are similar in that the prominent pathways belong to metabolism, but the detailed functional content is different between the two types.Type C had many pathways related to carbohydrate and lipid metabolism, while type B had pathways related to the biodegradation of xenobiotics and pathways related to DNA repair.In type O, pathways belonging to cellular processes were prominent, specifically, pathways related to bacterial cell growth and death, such as the cell cycle, meiosis, and apoptosis, as well as pathways related to cell motility, such as flagella assembly.These pathways were mainly predicted from the Neisseriaceae (F) family, which is abundant in type O.
abundant and influential in sebaceous regions compared to fungi or viruses in other skin sites.20However, considering the potential interactions between bacteria and fungi in relation to skin health, further investigations into the interplay between these microorganisms are required.Advanced research techniques, such as shotgun metagenomics or internal transcribed spacer sequencing for fungal DNA, could be utilized to explore the interactions and underlying mechanisms between bacteria and other microorganisms, providing a comprehensive examination of their associations.Despite its limitations, our study provides valuable insights into the relationships between skin microbial diversity, Cutibacterium abundance, and skin aging through a new classification method of skin microbial types.We found a significant correlation between sebum, one of the skin characteristics, and both microbial diversity and Cutibacterium.While environmental factors can impact the composition of the skin microbiome, the skin characteristics were measured and swab samples were collected within a week of individuals residing in Seoul, Korea, who shared similar environmental conditions in our analysis.Therefore, it is likely that the observed correlations between sebum and microbial features are maybe primarily influenced by host physiological factors, independent of other environmental influences.A previous research on the development of artificial sebum and sweat media revealed topical treatment for acne,26 and the use of basic cosmetic products27 have also been investigated.Subsequent studies on lifestyle factors and therapeutic methods, therefore, can serve as a valuable basis for suggesting personalized skincare guides based on distinct skin microbiome types.Such determinants will contribute to a deeper understanding of the skin microbiome and their potential role in promoting overall skin health.AUTH O R CO NTR I B UTI O N SYunkwan Kim, Joong-Gon Shin, and Nae Gyu Kang are involved in the conceptualization and project administration.Seo-Gyeong Lee, Hye Lim Keum, Ki-Nam Gu, Jung Yeon Seo, and Hae Jung Song were included in the analysis and visualization.Joong-Gon Shin, Seo-Gyeong Lee, and Jung Yeon Seo were involved in data collection and preparation.Seo-Gyeong Lee, Hye Lim Keum, Hae Jung Song, and Yunkwan Kim wrote the original drafts.Sangseob Leem, Hae Jung Song, Yunkwan Kim, Ki-Nam Gu, Seon Mi Lee, Woo Jun Sul, and Nae Gyu Kang were involved in writing the reviews and editing.