The skin microbiome, a complex ecosystem covering the largest organ of the body, is home to a myriad microorganisms including bacteria, viruses and fungi. Among these, the fungi population, or the mycobiome, albeit less abundant than its bacterial counterparts, plays a crucial yet often underappreciated role in maintaining skin health and disease.
Log in to view the full article
The skin microbiome, a complex ecosystem covering the largest organ of the body, is home to a myriad microorganisms including bacteria, viruses and fungi. Among these, the fungi population, or the mycobiome, albeit less abundant than its bacterial counterparts, plays a crucial yet often underappreciated role in maintaining skin health and disease.
The mycobiome represents less than 5% of the total skin microbiota.1 It maintains a low diversity that is predominantly comprised of the genus Malassezia.2 Found in healthy adults, this community exhibits stability,3 suggesting its critical role in the skin ecosystem.
Research has uncovered a variety of fungi on healthy skin, including Candida along with multiple Malassezia species such as Malassezia globosa, Malassezia restricta, Malassezia sympodialis and Malassezia furfur,4 each playing distinct roles across different skin environments. Beyond mere occupants, these fungal communities actively engage in skin metabolism.
Malassezia species, for instance, contribute significantly to the skin lipidome through their metabolism of lipids, breaking them down into smaller fatty acids.5 This process is vital in driving skin homeostasis and modulating immune responses, including eliciting a type 17 immune response that fosters antifungal immunity under homeostatic conditions.6 Such interactions elucidate the fungi's pivotal role in skin health, balancing microbial colonization and facilitating immune defense mechanisms.
While forming an integral part of the skin microbiome, fungi are also implicated in various skin disorders. Malassezia, despite being predominant in healthy skin, is associated with conditions such as tinea versicolor, atopic dermatitis and seborrheic dermatitis.7 These associations highlight a complex relationship where shifts in fungal abundance, diversity and species composition lead to or exacerbate skin issues.8-11
For example, an imbalanced fungal community can trigger exaggerated immune responses in disrupted skin barriers, leading to increased inflammation.6 Moreover, the involvement of other fungal species such as Candida spp. and Cryptococcus spp. in skin disorders underscores the importance of a broader investigation into the mycobiome’s role in dermatological health.12
Despite their significance, fungi have been relatively overlooked in microbiome studies, which have predominantly focused on bacteria. Shifts in fungal diversity reveal a distinct correlation with the severity of skin conditions, contrasting the patterns observed in bacterial diversity.13-17
In earlier work, the authors elucidated the critical role of bacteria in defining skin health – and understanding the full profile and interaction between commensal and pathogenic microbiota could hold the potential to treat skin disorders and other conditions. To uncover such insights objectively, a digital platform was built to support at-home microbiome data collection via kit activation, metadata collection and depersonalized sequencing.
Through this tool, the aggregated data from study participants is clustered into more precise skin profiles to allow for efficient formulation development and personalized regimens through artificial intelligence (AI) classification.18 The previous work used hierarchical clustering based on species-level microbiome data combined with self-reported factors such as skin health and lifestyle choices. This identified Cutibacterium acnes as a pivotal determinant in the formation of facial skin profiles, particularly in clusters indicative of an oily skin phenotype, skin longevity and health.
Building upon these previous findings, the authors next sought to unravel the taxonomic structure and composition of the fungal communities within these same clusters. Interestingly, as will be shown, incorporating fungal microbiome data alongside bacterial data provided enhanced resolution, enabling the further segmentation of larger C. acnes-dominated clusters into four smaller, distinct groups. Adopting a holistic view that integrates the complex interactions between bacterial and fungal communities within the skin microbiome could therefore be a key to unlocking advanced dermatological insights for the development of innovative skin care products.
Methods
Participant criteria: The present study included participants based on their initial completion of a commercial skin microbiome sampling kit (n = 323). Participant information was processed in accordance with the privacy policy and de-identified results were used for the analysis.
Microbiome sample collection: Participants self collected microbiome samples from their forehead area using cotton swabs. They were instructed to avoid using facial products or washing their face for 24 hours prior to sampling. When collecting the samples, participants swabbed their foreheads with the supplied swab for 80 seconds. The swabs were then placed into a sample tube containing liquid buffer and immediately stored in the freezer until shipping. Swab samples were shipped to the lab for sequencing within four days of sampling.
Metadata collection: Participants also completed an online survey paired with the microbiome sampling kit. Metadata included demographic information (gender, location and year of birth), as well as skin information (current condition, routine, issues, goals, etc.) and diet.
Sequencing and clean-up: Microbial DNA were extracted from the swab samples and sequenced at 16S ribosomal RNA gene (16S rRNA) V1-V3 region for bacteria and Internal Transcribed Spacer 2 (ITS2) region of the ITS region for fungi. Sequencing results were then analyzed for contaminants and cleaned with a proprietary algorithm. All processed samples contained high quality bacterial and fungal sequence data.
Microbiome analysis: Raw bacterial and fungal microbiome data from all samples was transformed to relative abundances. Bray-Curtis dissimilarity was computed using combined bacterial and fungal abundances. RStudio/Posit software was used as the local analytics environment, and preprocessing of sequencing data was conducted with the phyloseq package. Relative abundance was calculated with the mia package, transforming abundance counts to relative abundance using the relabundance method.
Hierarchical clustering based on Bray-Curtis distance was conducted using the hclust function with the vegan package. Silhouette profiles were obtained by the NbClust package to find the appropriate number of clustering. Each cluster was then characterized using the grouped averages of their respective metadata and microbiome features. The three control samples from the total sample number were excluded from analyses after hierarchical clustering. Correlation analysis of the metadata and top microbial species was performed using the microViz package in R.
Privacy and cybersecurity: As mentioned, all personally identifiable information (PII) was removed prior to analysis. Analysis was performed on local FileVault (Apple, 2003) and BitLocker (Microsoft, 2007) encrypted hardware. Data downloaded and uploaded from/to the company’s servers relied on HTTPS connections with up-to-date SHA-256 and RSA encryption in the browsers and software. Two-factor Authentication was required for essential team resources, including GitHub (GitHub, 2008), AWS (Amazon, 2006), and OneDrive (Microsoft, 2007).
Results
As noted, the previous study leveraged hierarchical clustering on combined bacterial and fungal microbiome data from 323 forehead skin samples.18 Quantitative microbiome metrics included Shannon diversity, microbial load and taxonomy, alongside self-reported survey data about skin health, age, gender and lifestyle. Here, 14 distinct clusters were identified based on their microbial composition (see Figure 1).
Comprehensive analysis aligning microbiome clusters with self-reported information highlighted two major clusters, 6 and 8, that together, harbored the majority of the samples collected (n = 205). These clusters were characterized by a dominant presence of C. acnes, accounting for more than 80% of the species composition.
Survey responses in these clusters were mainly from individuals in their 20s to 40s, with more than 50% having combination skin and nearly 25% having oily skin. A majority of these also reported a prevalence of acne (see Figure 2 and Table 1).
Further clustering was achieved in the present study through the application of fungal microbiome data, which enabled the segmentation of each of the two large clusters into four subclusters: 6A-D and 8A-D (see Figure 3). Here, more than 70% of participants for all subclusters were women. Also, subcluster 6D contained an even distribution of age between 20 to 50 years of age, while all other subclusters contained participants primarily within their 30s.
In subcluster 8D, an interesting group of characteristics was observed. This cluster contained the oldest set of participants, had no reports of oily skin and was comparatively lower in terms of reported acne issues – instead, wrinkles were the most concerning skin issue.
Furthermore, combination skin, followed by dry skin, dominated most clusters except for subclusters 6A and 8D. Acne was most prevalent in all subclusters except for subcluster 8D. In addition, subcluster 6C contained the least number of participants with skin sensitivity (see Table 1).
Bacterial vs. Fungal Load Dynamics
Exploring the differences in microbial load between the identified subclusters revealed an interesting pattern within Cluster 6 (6A-D). In terms of bacterial load, this remained stable; but the fungal load fluctuated across subclusters (see Figures 3c and 3a, respectively). Also, subclusters 6B and 6D had significantly lower fungal loads than subclusters 6A and 6C (p < 0.01). Furthermore, subcluster 6A presented a notably higher fungal load relative to 6D, demonstrating distinct variations in fungal bioburden among the subclusters.
Cluster 8 exhibited different dynamics in microbial loads. The bacterial load was stable between 8A, 8B and 8C but was significantly lower in 8D than 8C (p < 0.05) – a shift not observed in Cluster 6 (see Figure 3d). For fungal load, subclusters 8A, 8B and 8D were significantly lower than 8C (see Figure 3b). This observation suggests a single divergent pattern of fungal distribution within Cluster 8, distinguishing it from the distribution profiles of Cluster 6.
Fungal Microbiome x Bacterial Differentiation: Cluster 6
Further investigation of microbiome diversity in Cluster 6 revealed distinct differences in bacterial and fungal communities between subclusters.
Bacteria: Bacterial communities demonstrated low Shannon diversity across all subclusters (see Figure 4c), where Cluster 6D displayed significantly higher diversity compared with Clusters 6A through 6C – albeit still a low level of diversity (p < 0.001). Cluster 6 bacterial composition was predominantly characterized by C. acnes (see Figure 4d), although Subcluster 6D demonstrated a significantly lower presence of C. acnes compared with the other subclusters (p < 0.001).
In contrast, Staphylococcus epidermidis was significantly higher in Subcluster 6D than 6A (p < 0.001) and 6B-C (p < 0.01). Also, Subcluster 6D contained additional bacterial species in significantly higher levels, compared with subclusters 6A and 6B, including Corynebacterium kroppenstedtii, Corynebacterium bovis and Staphylococcus hominis. In addition, Corynebacterium sp were significantly higher in subcluster 6D than subcluster 6B (p < 0.001). Bacterial composition in subclusters 6A through 6C were highly similar, except for one variation: Streptococcus mitis was significantly higher in subcluster 6A compared to subcluster 6B (p < 0.05), adding differentiation between the subclusters.
Fungi: Compared to the bacterial compositions, fungal communities showcased wide variability between subclusters. Overall, Cluster 6 fungal communities demonstrated high alpha diversity, particularly in subclusters 6B and 6D (see Figure 4b). Alpha diversity was significantly elevated in subcluster 6B compared to subclusters 6A and 6C (p < 0.001), and subcluster 6D compared to subclusters 6A (p < 0.05) and 6C (p < 0.001).
Fungal communities were primarily dominated by the yeast M. restricta and other fungal species (see Figure 4d). A wide array of additional taxa is also noted within Cluster 6, with elucidation between subclusters by M. restricta and other fungal species. There was a significantly higher presence of M. restricta in subcluster 6A compared to subclusters 6B (p < 0.001) and 6D (p < 0.01), in addition to subcluster 6C compared to subclusters 6B and 6D (p < 0.001). In contrast, other fungal species were significantly higher in subcluster 6B compared to subclusters 6A and 6C (p < 0.001), as well as subcluster 6D compared to subclusters 6A (p < 0.01) and 6C (p < 0.001).
Fungal Microbiome x Bacterial Differentiation: Cluster 8
Bacteria: In analyzing bacterial microbiome data for Cluster 8, distinct patterns of Shannon diversity emerged across the groups (see Figure 5c), illustrating the variation within this cluster. Specifically, the diversity within subcluster 8D was markedly higher than that of subclusters 8A and 8C (see Figure 5c). In contrast, subcluster 8B exhibited significantly greater diversity than 8C. Further examination revealed significant differences in the taxonomic composition and relative abundance of key bacterial species across the subclusters (see Figure 5d).
A significant variation in the abundance of C. acnes was observed, with subcluster 8A displaying a higher presence compared to 8B (p < 0.01) and 8D (p < 0.001). Conversely, subcluster 8C showed a significantly higher abundance of C. acnes than both 8B and 8D (p < 0.001). Additionally, the levels of S. epidermidis varied significantly across the subclusters, with 8B having elevated levels compared to both 8A (p < 0.01) and 8C (p < 0.001). Other significant bacterial species that varied between subclusters included C. bovis, Finegoldia magna, Staphylococcus aureus and S. mitis, each contributing to the unique microbial composition characterizing each subcluster.
Fungi: Similarly, the fungal microbiome data revealed a distinct pattern (see Figure 5a); the Shannon diversity index of subcluster 8C was significantly lower than that observed in subclusters 8A, 8B and 8D. This low fungal diversity in subcluster 8C is attributed to its high prevalence of M. restricta, which constituted more than 90% of the fungal community – significantly higher compared to the other three subclusters (p < 0.001; see Figure 5b). The increased fungal diversity in subclusters 8A, 8B and 8D results from lower levels of M. restricta and a higher incidence of other fungal species, which were significantly lower in 8C, compared to the rest.
Community Composition by PCoA
The distribution of the microbial community composition across different subclusters was analyzed using the Bray-Curtis dissimilarity Principal Coordinates Analysis (PCoA) plot (see Figure 6). This revealed a significant difference among the subclusters [permutational multivariate analysis of variance (PERMANOVA), Adonis p-value = 1e-04, R² = 0.12, F = 3.81] (see Figure 6a). The bacterial community showed a more pronounced clustering effect (PERMANOVA, Adonis p-value = 1e-04, R² = 0.41, F = 19.51), with clear separations between some subclusters.
This suggests strong differential variations between the groups (see Figure 4b), particularly evident in the bacterial clusters, where each subcluster within Cluster 6 was more similar; this pattern was also similar to that of Cluster 8. Thus, using the fungal microbiome data enabled further segmentation within the two C. acnes-dominated clusters, demonstrating the importance of aligning both bacterial and fungal microbiome data in delineating microbial community structures with greater precision.
Correlation of Abundant Taxa and Metadata
Visualization of the Pearson correlation coefficients with heatmaps also provided several insights into the microbial-metadata relationships (see Figure 7). Within fungal communities, non-Malassezia species were found to increase overall with age and wrinkles. Specifically, a significant positive correlation (p < 0.05) of Cladosporium sp. and Cladosporium lignicola, and significant negative correlation (p < 0.05) with various Malassezia species (M. globosa, M. arunalokei and M. sympodialis) were observed with age.
In contrast, Malassezia sp. were increased in the presence of acne, rosacea and sensitivity, whereas Alternaria sp had a significant decrease (p < 0.05) with acne and sensitivity (see Figure 7a). In addition, most Malassezia species, except for M. slooffiae, were clustered based on their correlations with various survey variables, correlating negatively with age and wrinkles and positively with acne.
Thus, the relationships between bacterial communities and participant metadata can be better understood at a more granular level. With age, overall Corynebacterium species, particularly C. bovis (p < 0.05), showed a positive correlation, while C. acnes decreased significantly (p < 0.05). C. acnes was also weakly negatively correlated with psoriasis and eczema, but positively correlated with oily skin (p < 0.05), acne (p < 0.01) and rosacea.
In addition, C. kroppenstedtii, C. pseudogenitalium and other low abundant species had a significantly negative correlation with acne. On the other hand, C. kroppenstedtii had a significant positive correlation with rosacea (p < 0.001). Also noted was a significant positive correlation between wrinkles and one bacterium, S. mitis (p < 0.05) (see Figure 7b).
Discussion
The described findings highlight how leveraging fungal microbiome data can further delineate a substantial population of individuals sharing similar bacterial profiles. This demonstrates the critical role of incorporating both data types, bacteria and fungi, in discriminating among closely related subclusters, offering a finer resolution into the complex relationship between fungal and bacterial communities on the skin. The study also demonstrates the dynamic nature of fungal communities in response to environmental skin changes, contrasting with the more stable bacterial loads, and underscores the significance of a balanced mycobiome for skin health.
Incorporating these findings, subcluster 8D emerged with distinctive features: the highest Shannon diversity, the lowest presence of C. acnes and oldest mean age among the subclusters. Participants in this subcluster reported no oily skin and the lowest incidence of acne, thereby suggesting a potential link between age, microbial diversity and skin health. This observation aligns with current research indicating that the skin microbiome undergoes notable changes of increasing diversity due to the decrease of C. acnes.19-21
Furthermore, subcluster 8D was distinguished by having the highest concern of wrinkles, alongside a significantly lower microbial load. Yet intriguingly, this analysis found no significant correlation between microbial load and age, unlike the observed relationship between diversity and age (data not shown). This differentiation implies that while increased diversity within the microbiome correlates with aging, the overall microbial load does not follow a parallel trend.
The analysis also revealed significant correlations between aging, C. acnes and skin conditions, notably identifying a positive association between S. mitis and wrinkle formation, echoing findings by Li, et al., regarding its link to aging-related wrinkles.22 Concurrently, an upsurge in C. kroppenstedtii, known to be prevalent in rosacea patients' skin microbiomes,23 was observed along with a positive correlation with C. humerusii and a notable decrease in S. epidermidis. This latter finding prompts further investigation, considering contrasting reports of elevated S. epidermidis in specific rosacea-affected areas.24, 25
Fungal patterns also emerged, with Malassezia species such as M. globosa, M. arunalokei and M. sympodialis showing a negative correlation with age — a trend consistent across wrinkles, dry skin and eczema but inversely related to rosacea, acne and sensitivity. This indicates the dominant presence of Malassezia on younger skin and its diminished abundance with age, as highlighted by Li, et al.,22 underscoring the complex interplay between microbial species and skin health across different ages and conditions.
Furthermore, in a longevity study of Korean women, the significant association between Malassezia species and the rise in sebum levels on the forehead was observed – a phenomenon that diminishes as one ages.26 Subclusters 6A and 6C, which had a smaller number of participants over 40, with counts of 1 and 3 respectively, had significantly elevated levels of Malassezia compared to 6B and 6D, which contained three to five times the number of participants over the age of 40. Additionally, although not statistically significant, subclusters 6B and 6D exhibited elevated levels of Cladosporium, a trend that showed a positive correlation with age.
Conclusion
In conclusion, this comprehensive investigation into the skin microbiome reveals the significant, yet often overlooked, contribution of the mycobiome to the skin, particularly as it relates to age. The results distinctly highlight the intricate interplay between fungal and bacterial communities and their collective impact on different skin profiles.
Also, the findings particularly highlight how diversity, along with the specific fungal and bacterial compositional shifts, correlate with longevity. These insights emphasize the necessity of incorporating the mycobiome into dermatological research, suggesting a shift toward more holistic approaches in skin care that consider the complex, age-related dynamics of the skin's microbial ecosystem.
References
1. Byrd, A.L., Belkaid, Y. and Segre, J.A. (2018). The human skin microbiome. Nature Reviews Microbiology. Available at https://doi.org/10.1038/nrmicro.2017.157
2. Ruchti, F. and LeibundGut-Landmann, S. (2023). New insights into immunity to skin fungi shape our understanding of health and disease. Parasite Immunology. Available at https://doi.org/10.1111/pim.12948
3. Findley, K., Oh, J., … Segre, J.A., et al. (2013). Topographic diversity of fungal and bacterial communities in human skin. Nature. Available at https://doi.org/10.1038/nature12171
4. Sugita, T., Suto, H., Unno, T., Tsuboi, R., Ogawa, H., Shinoda, T. and Nishikawa, A. (2001). Molecular analysis of Malassezia microflora on the skin of atopic dermatitis patients and healthy subjects. Journal of Clinical Microbiology. Available at https://doi.org/10.1128/JCM.39.10.3486-3490.2001
5. Tuor, M. and LeibundGut-Landmann, S. (2023). The skin mycobiome and intermicrobial interactions in the cutaneous niche. Current Opinion in Microbiology. Available at https://doi.org/10.1016/j.mib.2023.102381
6. Sparber, F., de Gregorio, C., … LeibundGut-Landmann, S., et al. (2019). The skin commensal yeast Malassezia triggers a Type 17 response that coordinates anti-fungal immunity and exacerbates skin inflammation. Cell Host & Microbe. Available at https://doi.org/10.1016/j.chom.2019.02.002
7. Gaitanis, G., Magiatis, P., Hantschke, M., Bassukas, I.D. and Velegraki, A. (2012). The Malassezia genus in skin and systemic diseases. Clinical Microbiology Reviews. Available at https://doi.org/10.1128/CMR.00021-11
8. Schmid, B., Künstner, A., … Bosshard, P.P., et al. (2022). Dysbiosis of skin microbiota with increased fungal diversity is associated with severity of disease in atopic dermatitis. Journal of the European Academy of Dermatology and Venereology. Available at https://doi.org/10.1111/jdv.18347
9. Takemoto, A., Cho, O., Morohoshi, Y., Sugita, T. and Muto, M. (2015). Molecular characterization of the skin fungal microbiome in patients with psoriasis. The Journal of Dermatology. Available at https://doi.org/10.1111/1346-8138.12739
10. Tao, R., Li, R. and Wang, R. (2021). Skin microbiome alterations in seborrheic dermatitis and dandruff: A systematic review. Experimental Dermatology. Available at https://doi.org/10.1111/exd.14450
11. Han, S.H., Cheon, H.I., … Ahn, K.J., et al. (2018). Analysis of the skin mycobiome in adult patients with atopic dermatitis. Experimental Dermatology. Available at https://doi.org/10.1111/exd.13500
12. Jo, J.-H., Kennedy, E.A. and Kong, H.H. (2017). Topographical and physiological differences of the skin mycobiome in health and disease. Virulence. Available at https://doi.org/10.1080/21505594.2016.1249093
13. Kong, H.H., Oh, J., … Segre, J.A., et al. (2012). Temporal shifts in the skin microbiome associated with disease flares and treatment in children with atopic dermatitis. Genome Research. Available at https://doi.org/10.1101/gr.131029.111
14. Gao, Z., Tseng, C., Strober, B.E., Pei, Z. and Blaser, M.J. (2008). Substantial alterations of the cutaneous bacterial biota in psoriatic lesions. PloS One. Available at https://doi.org/10.1371/journal.pone.0002719
15. Alekseyenko, A.V., Perez-Perez, G.I., de Souza, A., et al. (2013). Community differentiation of the cutaneous microbiota in psoriasis. Microbiome. Available at https://doi.org/10.1186/2049-2618-1-31
16. Zhang, E., Tanaka, T., Tajima, M., Tsuboi, R., Nishikawa, A. and Sugita, T. (2011). Characterization of the skin fungal microbiota in patients with atopic dermatitis and in healthy subjects. Microbiology and Immunology. Available at https://doi.org/10.1111/j.1348-0421.2011.00364.x
17. Alcaïs, A., Abel, L. and Casanova, J.-L. (2009). Human genetics of infectious diseases: Between proof of principle and paradigm. The Journal of Clinical Investigation. Available at https://doi.org/10.1172/JCI38111
18. Park, H.G., Swaney, M.H., Nguyen, U.T., Lukianets, N. and Jungman, E. (2024). Redefining the skin type: Microbiome-based segmentation for efficient product development and personalization. IFSCC Magazine, 26(4).
19. Shibagaki, N., Suda, W., … Hattori, M., et al. (2017). Aging-related changes in the diversity of women’s skin microbiomes associated with oral bacteria. Scientific Reports. Available at https://doi.org/10.1038/s41598-017-10834-9
20. Howard, B., Bascom, C.C., … Isfort, R.J., et al. (2022). Aging-associated changes in the adult human skin microbiome and the host factors that affect skin microbiome composition. Journal of Investigative Dermatology. Available at https://doi.org/10.1016/j.jid.2021.11.029
21. Myers, T., Bouslimani, A., Song, S.J., et al. (2024). A multi-study analysis enables identification of potential microbial features associated with skin aging signs. Frontiers in Aging. Available at https://doi.org/10.3389/fragi.2023.1304705
22. Li, Z., Bai, X., … Hu, D., et al. (2020). New insights into the skin microbial communities and skin aging. Frontiers in Microbiology. Available at https://doi.org/10.3389/fmicb.2020.565549
23. Rainer, B.M., Thompson, K.G., … Chien, A.L., et al. (2020). Characterization and analysis of the skin microbiota in rosacea: A case-control study. American Journal of Clinical Dermatology. Available at https://doi.org/10.1007/s40257-019-00471-5
24. Whitfeld, M., Gunasingam, N., Leow, L.J., Shirato, K. and Preda, V. (2011). Staphylococcus epidermidis: A possible role in the pustules of rosacea. Journal of the American Academy of Dermatology. Available at https://doi.org/10.1016/j.jaad.2009.12.036
25. Woo, Y.R., Lee, S.H., Cho, S.H., Lee, J.D. and Kim, H.S. (2020). Characterization and analysis of the skin microbiota in rosacea: Impact of systemic antibiotics. Journal of Clinical Medicine. Available at https://doi.org/10.3390/jcm9010185
26. Kim, H.J., Oh, H.N., Park, T., et al. (2022). Aged related human skin microbiome and mycobiome in Korean women. Sci Rep 12, 2351; https://doi.org/10.1038/s41598-022-06189-5
