Sunscreens can preserve human skin microbiome upon erythemal UV exposure

Ultraviolet radiation (UVR) is a known environmental key factor for premature skin ageing. Only few scientific evidence is available to support the effects of UVR on the skin microbiome. This in vivo pilot study aimed to evaluate the impact on the skin microbiome upon erythemal UV exposure and the protection of UV‐exposed skin microbiome by UV filters.


INTRODUCTION
At the boundary between an individual and the environment, the skin harbours a multitude of microorganisms such as bacteria, fungi, viruses, archaea, and mites that compose the skin microbiome.Bacteria were found to be the most abundant skin microorganisms, with more than 1000 distinct species.The most common phyla on skin are Actinobacteria, Firmicutes, Proteobacteria, and Bacteroidetes, identified by sequencing the bacterial 16S small-subunit ribosomal RNA gene [1].While the composition of the skin microbiome has been found to significantly vary between individuals and across body sites, it can be characterized primarily by the physiology of the skin site, whether they are sebaceous (oily), moist, or dry [2,3].For example, lipophilic bacteria and fungi of the genus Cutibacterium and Malassezia, respectively, prefer sebaceous sites such as the face, chest, and back, whereas Staphylococcus and Corynebacterium species more abundantly colonize less exposed, moist areas, including the groin, axilla, and the feet.
The skin microbiome stimulates the breakdown of natural products on the skin surface.The resulting building blocks and metabolites can be shared with other microorganisms and host cells.The skin surface constitutes a hostile environment to the microbiota as it lacks many nutrients beyond basic proteins and lipids derived from stratum corneum, sweat, and sebum.In order to survive, skin microorganisms employ various strategies that involve interactions among themselves as well as mutualistic or commensal interactions with mammalian host cells.These symbiotic interactions affect microbial fitness, population dynamics, and functional capacities within the microbiome.The skin microbiome also promotes defence and immune responses.In the interplay between commensal residents and the cutaneous immune system, it protects against colonization and infection by opportunistic or pathogenic organisms and promotes tissue repair and barrier functions.Skin commensals such as Staphylococcus epidermidis, one of the most abundant bacteria on healthy skin, can produce biofilms, antimicrobial compounds, and proteases to limit colonization of pathogens and thereby modulate the innate immune function [2,3].
It has been generally recognized that solar ultraviolet radiation (UVR) is the key environmental factor causing premature skin ageing [4].High UVR causes cutaneous alterations, starting with acute effects such as erythema (sunburn), pigmentation (tanning), immune modulation, and vitamin D synthesis.In the long term, chronic UVR exposure may result in the well-known consequences of pigmentary disorders, photoaging, photo-immunosuppression, photo-carcinogenesis, and photo-dermatoses [5].However, limited evidence exists in the literature regarding the effects of UV radiation on skin microbiota [6][7][8].High-energy UV radiation exerts a general germicidal effect on microorganisms.Short-wavelength UVC irradiation (100-280 nm) is a common disinfection method for air, water, and food to kill or inactivate microorganisms.The effectiveness of germicidal UV irradiation depends on the exposure time, intensity, and wavelength of the UV radiation.While solar UVC radiation is absorbed by the atmospheric ozone layer, only UVB (280-315 nm) and UVA (315-400 nm) are environmentally relevant for the skin (microbiome).Burns et al. demonstrated that erythemal UVR doses had a profound qualitative and quantitative impact on the composition of the skin microbiome when a small number of participants were exposed to doses of UVA (22-47 J/cm 2 ) or UVB (100-350 mJ/cm 2 ) [6].Clear increases were observed in some bacterial groups, such as those in the phylum Cyanobacteria, and decreases in others, such as those in the family Lactobacillaceae, in subjects exposed to UV radiation.Previous studies have demonstrated that the presence of the microbiome on the skin of mice plays a crucial role in the epidermal and immune responses to UV radiation [7][8][9].The skin microbiome diminishes immunosuppressive responses to UV radiation by modulating gene expression and cellular microenvironment of the skin [8].It has been suggested that protecting the nature of a healthy skin microbiome against UV-induced microbiome shifts, for example, by prophylactic pro-or prebiotics, might be beneficial in the treatment of cutaneous skin disorders related to UV exposure [6,7].
Over the past few decades, effective sun protection strategies have been proposed, focusing on personalized approaches such as wearing sun-protective clothing and applying sunscreen.These strategies aim to shield individuals from the detrimental effects of the sun while still allowing them to benefit from its positive aspects.Sunscreens were originally been developed to minimize erythema and are specified by the sun protection factor (SPF) which is a quantitative measure of its ability to inhibit erythema.Modern sun protection products include photostable, broad-spectrum UVA/B filters to protect against sunburn, photocarcinogenesis, photodermatoses, and photoaging, and may be combined with substantiated active ingredients that protect against hyperpigmentation and premature skin ageing beyond the UV range.This clinical exploratory trial specifically assessed the impact of erythemal UV irradiation (2 MED, minimal erythema dose) on human skin with regards to the skin microbiome composition and the medium protection of the UVexposed skin microbiome by UV filters in a representative SPF 20 sunscreen versus a placebo formulation.A medium protection level of SPF 20 is clearly sufficient for a realistic exposure scenario of low erythemal UV dose (2 MED) and is applicable for facial care as the face is the most exposed skin region on a daily basis.

Clinical test products
The investigational products consisted of two test formulations; an SPF 20 sunscreen formulation including broad-spectrum UV protection by four UV filters (INCI names: Ethylhexyl Salicylate, Octocrylene, Bis-Ethylhexyloxyphenol Methoxyphenyl Triazine, Butyl Methoxydibenzoylmethane; DSM Nutritional Products, Switzerland) and a vehicle formulation replacing the UV filters with a higher oil content ("placebo") to assess the potential of protecting the UV-exposed skin microbiome (Table 1).

Human study design
A single-blind clinical trial was conducted by a clinical research institute (Eurofins Dermscan, Gdansk, Poland) according to Helsinki Declaration (1964) and its successive updates, as well as the guidance on Good Clinical Practice CPMP/ICH /135/95 (R2).The study was submitted to the Bioethics Committee of the Regional Chamber of Physicians (Gdansk, Poland) and approved in March 2021.Subjects received information about the study conditions, and a written informed consent was collected.Ten healthy female subjects between 20 and 45 years old with Fitzpatrick skin type II-III that complied with the inclusion/exclusion criteria were enrolled in the study.A commercial gentle rinse-off product (Perfume-free Cleansing Bar, Gallinée, France) was distributed to the subjects for a once-daily body wash in the conditioning phase for 7 days before measurement.Subjects did not use shampoo or body lotions on their backs at least 36 h before the treatment.A final cleansing of the back area with the provided rinse-off product was performed at least 4 h before the MED dosing (i.e., at least 24 h before treatment).Individual minimal erythema dosing of the subjects was initiated the day before the treatment.The next day (D0), MED was determined.Four zones of 2.5 cm × 10 cm were assigned on their upper middle backs, that is, untreated/ non-exposed (baseline); untreated/exposed; placebo and sunscreen treatment on the UV-exposed skin (see Supporting Information, Figure A body map).Skin swabbing samples were collected on the untreated, unexposed baseline zone for microbiome analysis.Then, approximately 2 mg of the two test formulations per 1 cm 2 of body surface was applied to two test areas on the back of 10 subjects prior to UV irradiation.Three zones (i.e., untreated, placebo, and sunscreen treated) on the back of approximately 25 cm 2 each were exposed with 2 MED of UVA/ UVB radiation using a monoport solar simulator.Photos were taken of the whole back using a Nikon D90 digital camera after UV irradiation to visualize skin redness.Skin surface swabs were collected on the three irradiated zones 2 h after UV exposure for the analysis of skin microbiome.Spectroscopical measurements of the dedicated non-treated zones were performed using a MINOLTA CM700-d Spectrophotometer before and 2 h after irradiation and were expressed as L* a* values of the colour space defined by the International Commission on Illumination (CIE) in 1976.L* stands for perceptual lightness and a* for the two unique colours of human vision: red, green.Occurrent adverse events were recorded.The skin swabbing samples were sequenced (BaseClear B.V., Leiden, NL) and analysed by the different conditions.

Skin microbiome sampling
Skin surface sampling was performed by trained and consistent personnel by swabbing the untreated, unexposed areas at baseline (D0) as well as at three selected irradiated zones after UV exposure that were previously treated with the placebo, the active formulation, or left untreated.Skin swabs were collected from the back of the study participants using Hydraflock® flocked swabs (Puritan, Guilford, ME, USA), which were previously soaked in a sterile 0.9% NaCl + 0.1% Tween 20 solution.The swab was firmly rubbed 10 times over the sampling site (surface of about 25 cm 2 ) for 10 s.The swab was then placed in a 2 mL collection tube containing 1 mL of DNA stabilization buffer (Zymo Research Corp., Irvine, CA, USA).The samples were stored at −80°C until further processing.

DNA extraction
DNA was extracted using the ZymoBIOMICS DNA Miniprep kit (ZYMO Research, Freiburg, Germany) and the associated protocol.In brief, samples were put into a lysis tube and then lysis solution was added.Lysis tubes were placed in a bead beater and then spun down in a microcentrifuge.
The supernatant was then filtered and washed until a final elution step to produce DNA suitable for sequencing.

16S rRNA amplicon sequencing
DNA was extracted from 36 swabbing samples of 9 subjects (V1-V9) as well as from one blank sample with the swabbing buffer (as contamination control) and sequenced on an Illumina MiSeq in a 2 × 300 paired configuration.Primers targeted the V3-V4 region of the bacterial 16S rRNA gene.Sequences were then processed into amplicon sequence variants (ASV) using DADA2 [10].ASVs represent distinct variants that can vary by as little as one base pair, differing from operational taxonomic units that rely on selecting medoid sequences from similar sequences.In brief, input sequences were trimmed for 10 sequences on the left, truncated to 250 base pairs, and then merged with a maximum number of mismatches set to three.Taxonomy was assigned using the RDP Taxonomy 16.

Illumina shotgun sequencing and data processing
Illumina demultiplexing paired-end sequence reads were generated using the Illumina NovaSeq 6000.The sequences generated with the NovaSeq 6000 were performed under accreditation according to the scope of BaseClear B.V. (L457; NEN-EN-ISO/IEC 17025).FASTQ read sequence files were generated using bcl2fastq version 2.20 (Illumina).Initial quality assessment was based on data passing the Illumina Chastity filtering.Subsequently, reads containing PhiX control signal were removed using an inhouse filtering protocol.In addition, reads containing (partial) adapters were clipped (up to a minimum read length of 50 bp).
The second quality assessment was based on the remaining reads using the FASTQC quality control tool version 0.11.8.

Alignment-based filtering (Illumina)
The analysis was performed by aligning the Illumina reads against the reference (human) sequence using BBmap v38.79 to filter out host contamination.Forward and reverse reads were trimmed using CutAdapt (v3.7) [11] to remove poor-quality sequences, and only pairs that passed were used downstream.Trimmed reads were then mapped against the human genome, with hits being removed, and only non-human reads were utilized for taxonomic and pathway classification.Taxonomy was identified using MetaPhlan2 with database mpa_v30_ CHOCOPhlAn_201901 and pathways were characterized using Humann2 [12].

Canonical correspondence analysis (CCA)
To evaluate the relationships between biological assemblages of species and their environment, CCA was applied using Vegan v2.5.7 [13,14].CCA evaluated participants, the treatments, and the irradiation blocks and significance was determined through an ANOVA-like permutation test.

Songbird
Songbird is a tool that compares ratios of taxa, or any compositional set of features, to remove bias of the unknown microbial load using relative abundance data [15].Songbird then models the differential rankings to provide a relationship between each feature and the categorical or quantitative measurement in question.Songbird was applied to the categorical variables of irradiation, treatment, and sunscreen to investigate the differential rankings of genes, pathways, and taxonomy both from a metagenomic and 16S rDNA perspective.These different features then contain a coefficient to describe their relationship to the modelled variable.
[16] was applied to 16S rDNA count data and MetaPhlanderived taxonomic relative abundance data.The generated networks were then trimmed using igraph to remove any nodes with a degree less than one.Exploration of cluster interactions via metabolism utilized taxonomic references to a close neighbour species as the taxonomic resolution of 16S data cannot get down to strain-level classification.
Organisms were selected from organisms already annotated in the KEGG reference database.

In vitro UV model
The microbial strains Lactobacillus crispatus CIP 103604 and S. epidermidis ATCC 12228, and Cutibacterium acnes ATCC 11827 were utilized from commercial strain collections for this study.The cultures were reconstituted according to the suppliers' manuals.That is, from CRBIP (Institut Pasteur Collection, France) or ATCC (American Type Culture Collection, USA).Briefly, culture media (140 μL) and the calibrated inoculum (20 μL) of each bacterial strain were added to a 96-well plate to reach 1 × 10 6 colony forming units per millilitre (CFU/ mL) in the final concentration.After inoculation, the test sample was added (40 μL) in triplicate and was thoroughly mixed to ensure homogeneous distribution.Subsequently, the plate was submitted to UV exposure by an UV irradiator system (BioSun, Vilber Lourmat, France) with an 1:19 energy ratio between UVA and UVB.An aliquot of each culture was taken in triplicate 4 h after UV stress and incubated for 48 h at the optimal growing conditions before population counting.The specific UV dose for each strain was determined before by an UV dose range-finding tests that led to a survival rate of the microbial population between 20% and 49% compared to the initial population 4 h after exposure (i.e., L. crispatus: 500 mJ/cm 2 , S. epidermidis: 150 mJ/cm 2 , C. acnes: 30 mJ/ cm 2 ).For the microbial population counts, aliquots of the contaminated samples of L. crispatus and S. epidermidis were serially diluted (4 decimal dilutions followed by 8 binary dilutions, in triplicate) in culture media on 96well plates with the dye triphenyltetrazolium chloride.The plates were incubated at 37 ± 2.5°C for 48 h considering the respiratory type of each strain.After 48 h at the optimal growing conditions, the last dilution showing a pink colouration (triphenyltetrazolium) or turbidity allowed the determination of the decrease rate of the microbial population by the Most Probable Number MPN method.The counting of C. acnes population was performed by plating aliquots on agar plates.As controls the unirradiated and irradiated bacteria cultures in culture media, as well as the vehicle control, were used.The microbial population counts were expressed in CFU/ mL and performed in triplicate.The use level of the single UV filter and combinations in the test formulations was calculated to obtain a predicted in vitro SPF value of 5.75 ± 0.15 and 23.45 ± 0.65, respectively.The UV filters were mixed in two oil solvents, Tegosoft XC (55%) and Miglyol 829 Eco (ad100%), plus the emulsifier Tween 80 (5%).The effect of selected UV dose conditions on the survival of the microbial population in the presence of the tested samples 1-10 at 20% v/v final concentration was compared with the corresponding strain in the absence of samples 1-10 without exposure to UV irradiation.Finally, the survival rate of each microbial population and the relative protection by the UV filter(s) were calculated.The CFU counts 4 h after UV exposure were compared to the initial population of bacteria and expressed in percentage of bacterial survival (%).The relative percentage of protection (%) was calculated from the difference between the survival percentages of samples and irradiated vehicle controls divided to the survival percentage of the irradiated vehicle controls.

Clinical study results
Ten women between 22 and 43 years old (mean age 32 ± 2 years) with skin type II (6 subjects) and III (4 subjects) were enrolled.Four zones were assigned on their upper middle backs that were attributed to the conditions used: untreated/unexposed, untreated/exposed, placebo, and sunscreen treatment on the UV-exposed skin (see Body map Supporting Information, Figure A).The investigational product consisted in an SPF 20 sunscreen formulation based on a standard oil-in-water emulsion.For the placebo formulation, the four chemical UV filters were replaced with oil.All participants successfully completed the study.No adverse events were observed during the study, except for the expected erythema formation on the unprotected, exposed skin.Two hours after UV exposure at 2 MED, the appearance of distinct erythema was observed on all subjects and spectroscopically quantified by a significant decrease of the L* parameter (−7%) and a significant increase of the a* parameter (+53%, Table 2) in the CIE colour space.Erythema was clearly visible in the unprotected zones, partially visible in the placebo-treated zones, but not visible in the SPF 20-treated zones, showing an efficient SPF photoprotection (see Supporting Information, Figure B).

16S sequencing results
The relative abundance graphs clustered by subject showed high variation in microbial composition between individuals and conditions.(Figure 1, Supporting Information, Figure C).The most abundant genus was Cutibacterium, with a wide range of relative abundances between 3% and 91% in different individuals.
This is further confirmed by the multivariate method CCA.Results of CCA suggested that interpersonal differences were significant (p-value: 0.001), whereas irradiation (p-value: 0.219), the placebo (vehicle control, p-value: 0.130), and sunscreen (sunscreen, p-value: 0.731) played less of a key role in the microbiome shape.

Diversity
Cutibacterium acnes is a major component of the skin microbiome, especially in sebum-rich skin regions, and confirmed as the most abundant bacteria in this study.As a result, the relative abundance of C. acnes can drastically impact ecological measures.When producing 16S rRNA gene read information, it is also unclear if multiple ASVs are present due to sequencing errors or unique strain information contained within the amplicon.Therefore, diversity measures (Shannon diversity index) were repeated with and without C. acnes (Figure 2).When including C. acnes, the median Shannon diversity index was highest in skin samples that received the placebo or the sunscreen formulation (Figure 2a).However, these samples also exhibited a significant degree of variability.A high variance was also observed in the unexposed site, and after UV radiation, it became apparent that the variance was reduced suggesting that differences in microbial diversity across individuals may be reduced.Conversely, when excluding the contribution of C. acnes to the diversity, the variance in the unexposed site was noticeably diminished, suggesting that the abundance of C. acnes may vary significantly among individuals (Figure 2b).
In addition, it became evident that treating the skin led to an increase in the diversity of specifically non-C.acnes microbiota.However, the use of sunscreen over placebo did not likely contribute to any significant effect on the global microbiota diversity.Also, irradiation did not seem to significantly alter the diversity of the skin microbiota.This could mean that, while UVR may impact individual organisms, they might be replaced by others at an equal ratio as the overall diversity did not change.Application of placebo increased diversity without considering C. acnes (Wilcoxon rank-sum, p = 0.014) versus non-treatment, but not otherwise (p = 0.139).The variability in C. acnes confounds much of our overall understanding of the effect of the treatment.It may be best practice to remove dominant organisms to control for such confounding effects in ecological measures.

Differential ranking analysis
The differential ranking analysis performed with Songbird is a statistical method to facilitate compositional changes based on relative abundance data [15] and allowed identification of key taxa associated with the application of the sunscreen formulation (Figure 3).Particularly, two sequences belonging to the Lactobacillus genus were found to have a positive association exclusively with the UV filter containing formulation.In contrast, Lactobacillus showed a negative association with UV irradiation and with the placebo (Table 3).While the taxonomic classifiers were only able to resolve ASVs 53 and 57 to the genus Lactobacillus, using SILVA ACT (alignment, classification, and tree service) and adding neighbours suggested that ASV 53 and 57 were likely Lactobacillus crispatus (Figure 4).
The differential analysis performed in our study showed that L. crispatus was one of the taxa in the microbial composition that changed the most and that was positively associated with the application of the sunscreen formulation containing UV filters.On the other hand, Cutibacterium acnes positioned itself in the middle of the differential ranking graph, suggesting that the UV filter formulation did not particularly influence its relative abundance compared to other taxa (Figure 3a).Similarly, when performing a log-ratio analysis between Lactobacillus and Cutibacterium, it could be observed that UV exposure led to a reduction of the Lactobacillus/Cutibacterium log ratio compared to the unexposed skin.Placebo application further resulted in a log ratio reduction implying a higher abundance of Cutibacterium compared to Lactobacillus on those skin sites.The application of the UV filter containing formulation led to a re-establishment of the original ratio between the two taxa suggesting protecting effects.(Figure 3b).Lactobacillus crispatus was found to be the most abundant Lactobacillus species (of 31 spp.) in our skin study.The most prevalent taxa of Lactobacilli of the skin were also found among the dominant Lactobacillus taxa (L.crispatus, L. iners, L. gasseri, L. jensenii) of the vaginal community [17].L. crispatus is one of the predominant species found in the vaginal tract which has been shown to play a key role in women's health [18].Also, a close association with poultry gut health was reported [19].Due to its ability to produce lactic acid, hydrogen peroxide, and bacteriocin-like compounds with strong antimicrobial properties, the bacterium has emerged as a promising probiotic for promoting host health.Its capacity to combat pathogenic bacteria makes it valuable in maintaining a healthy microbial balance [18,20].It is hypothesized that Lactobacilli are more abundant after birth and their proportion decreases as we get older.Leoty-Okombi et al. recently explored the facial microbiota in 50 young and 50 elderly Caucasian subjects focusing specifically on the wrinkle areas and found that Lactobacillus crispatus was present in high abundance in the younger cohort while the bacterium was not detectable in the wrinkles of the older cohort.This finding may suggest that L. crispatus could potentially play a role in the maintenance of healthy and young skin [21].

Interaction network
The network generated using Spiec-Easi produced 3 clusters of organisms (Figure 5).Cluster 1 came with a negative association with both sunscreen and the placebo treatment  and a mixed association with irradiation (Table 4).These are organisms that are likely present on the skin but are removed, inhibited, or otherwise negatively impacted by treatment.Cluster 2 appeared to be Cutibacterium acnes, and these bacteria showed to be positively associated with radiation and negatively associated with any treatment, sunscreen, or placebo.This is also supported when we examined the diversity metrics with and without C. acnes.It is not clear why there is such an association of C. acnes with exposure to UV. Cluster 3 contained the most diverse set of organisms.Most of these organisms had a positive association with the placebo and sunscreen and a negative association with radiation.The sequence assigned to Micrococcus sits most central to this cluster with connection to Corynebacterium, Streptococcus, and Staphylococcus and more remotely to Kocuria.When organisms are more central to their own cluster, this is referred to as closeness centrality and describes the ability of this organism to impact its immediate neighbours more readily [22].
These organisms can influence each other through various mechanisms, including the sharing of metabolites or DNA, the removal of inhibitors, or the ability to fulfil auxotrophy requirements.These interactions may play a crucial role in shaping their mutual relationships and overall ecosystem dynamics.These organisms must also share the same ecological niche and therefore sufficiently diverse mechanisms for the uptake the same necessary nutrients must be present among all organisms.Recent work by Loomis et al., show Micrococcus luteus has an impact on the host tissue and may be interacting with other organisms via the host as well [23].Micrococcus is also known to be evolutionarily resistant to UV radiation and may be capable of sharing this ability to other nearby organisms [9].
Looking at the ABC transporters associated with these five organisms, all but Micrococcus appear to have a unique manganese transporter (Figure 6, Supporting Information, Figure D).Manganese plays a key role in combatting oxidative stress via its use in manganese superoxide dismutase that counters the generation of reactive-oxidative species (ROS) [24].Micrococcus may not have the same manganese requirement though it is expected to have superoxide dismutase for responding to the presence of ROS [25].Having the ability to uptake minerals in a nutrientpoor environment may give these organisms a competitive advantage over others [26].

Metagenomics
Metagenomics still presents a challenge for low-biomass samples, as extracted nucleotides are frequently made up of host DNA rather than the sought-after microbial DNA.After QC PCR only 17 of 36 samples had sufficient microbial DNA to be eligible for shotgun metagenomic sequencing.The insufficient amount of bacterial biomass might be caused by the sampling procedure, even though we considered a higher swabbing surface area of 25 cm 2 as compared with the usual 4 cm 2 for the 16S rRNA gene sequencing.The 16S quantity from QC analysis was dependent on the specific subject than the applied conditions.We were not able to address the compliant versus non-compliant samples to specific subject characteristics such as age, skin appearance (i.e., dry, normal, and oily), or erythema intensity.However, there seemed to be a correlation between the samples with high abundances of C. acnes and compliant QC.Despite the incomplete data set the metagenomics data were further analysed on taxonomy and functional pathways by the conditions applied.Further data is represented in the Supporting Information.

In vitro UV model with isolated bacteria
Specific microorganisms can benefit from additional UV protection, as observed with L. crispatus in the clinical study.To confirm the in vivo results, a targeted in vitro liquid model was developed with individual bacteria and UV filters., Selective UV filters and combinations were screened for the survival of Lactobacillus crispatus, Cutibacterium acnes, and S. epidermidis (as reference strains) in 96-well microplate format upon UV stress.The UVA/B dose was adjusted to the different UV sensitivity of the used strains.The inhibition of bacterial growth as a consequence of UV irradiation was compared to that of the same strain without UV irradiation (Table 5) 4 h after UV exposure.
The UV filters exerted a specific level of protection on the tested bacteria upon UV exposure.In particular, the UV filters octocrylene, zinc oxide, ethylhexyl triazone, and butyl methoxydibenzoylmethane increased the survival of L. crispatus compared to the vehicle control.Also, the combinations of UV filters were effective: Sample 8 had the same UV filter combination as used in the in vivo study and confirmed the protecting effect on the L. crispatus population.On the other hand, selected UV filters such as butyl methoxydibenzoylmethane, ethylhexyl salicylate, and octocrylene, as well as combinations reduced the population of C. acnes within the skin microbiome upon UV irradiation.At the same time, the population of the commensal bacterium S. epidermidis was maintained or propagated by the different UV filters.As a result, a smart selection of UV filters in a sunscreen is supposed to control the C. acnes population upon UV exposure and ultimately, might mitigate the inflammatory acne breakouts.

CONCLUSIONS
Through this pilot study, we obtained valuable insights into the response of the skin microbiome to erythemal UV exposure as well as the potential protective effects of an SPF 20 sunscreen formulation on UV-irradiated skin.These findings contribute to our understanding of the complex dynamics between the skin microbiome and UV radiation, highlighting the importance of sun protection measures in maintaining a healthy skin ecosystem.The T A B L E 5 Test formulations of UV filter(s) and ingredients (* mixed with 55% Tegosoft XC, ad100% Miglyol 829 Eco, 5% Tween 80) in weight%.The use level and predicted in vitro SPF value was in the range of 5.6-5.9 for the single UV filter formulations and 23-24 for the filter combinations.Results of the survival 4 h after exposure of each microbial population was calculated in percentage [%] compared to initial population.In brackets: relative protection by the UV filter(s) in % versus irradiated vehicle control.relative abundances clustered by subject showed significant interindividual differences in microbial composition.UV irradiation and treatment had certain impacts on the relative abundances.Adding placebo as a treatment seemed to increase the microbial diversity, but no significant additional effect on diversity for sunscreen was detected.The placebo formulation appeared to show partial UV protection with an erythema intensity between unprotected and SPF 20-protected skin, which might reduce the differential effect compared to the sunscreen.As an exploratory study, we acknowledge that the recruitment of a limited number of volunteers represents a limitation, particularly in the context of microbiome analysis.While this preliminary investigation provides valuable insights, additional larger studies are needed to further substantiate our findings.Despite increasing the standard sampling area by a factor of 5, some of the samples did not provide enough biomass to allow the metagenomic sequencing, therefore our metagenomic data referred only to a subset of the samples.Thus, a higher surface area is not a guarantee for high-quality samples, whereas variability in the swabbing procedure, subject variability, body region, treatment, and removal of host DNA might also be critical, and methodological standardization should be considered in future studies.To partially overcome this limitation 16S rRNA sequencing was performed.All 36 samples had sufficient bacterial DNA to be sequenced by the 16S rRNA method to gain data on the full set of subjects and conditions.While diversity can be a simplified microbial metric, sunscreen application impacted the relative abundance of pathways.A species of Micrococcus was detected, most likely M. luteus, that is known to be relatively resistant to UV damage and can confer protection to other species.In the discovered interaction network associated with sunscreen, the closeness centrality of Micrococcus to Corynebacterium, Streptococcus, and Staphylococcus is an example of how Micrococcus might impact nearby organisms or even the host tissue and their abilities to UV resistance by sharing metabolites (e.g., siderophores, cofactors, and carotenoids).Increased supplementation with specific building blocks like the branched-chain amino acids that were associated with sunscreen might support the skin microbiome and host cells for better health.

Sample
In conclusion, based on the data set obtained, we suggest that treating the skin microbiome with an SPF 20 sunscreen before UV exposure could potentially yield skin benefits.The concept could be extended to sunscreens with higher SPF values than SPF 20.We anticipate similar or even greater protective effects on the skin microbiome.However, it is important to empirically evaluate the microbiome compliance of each formulation, considering factors such as viability and microbial composition.This assessment is necessary due to variations in ingredient content among different formulations.
Certain microorganisms, such as Lactobacillus crispatus, appear to derive benefits from enhanced UV protection.It is worth noting that L. crispatus is not commonly considered a skin commensal strain.Similar to its crucial role in vaginal microbiota, L. crispatus may potentially form beneficial biofilms to safeguard host tissue and contribute to the innate immune response, ultimately promoting the maintenance of the skin's self-defence mechanisms.
The specific preservation of UV-sensitive Lactobacillus crispatus and reduction of Cutibacterium acnes in the presence of carefully selected UV filters hold the potential for novel cosmetic concepts and claims related to photoprotection.These developments could pave the way for effective strategies for safeguarding not just the skin but also its microbiome against the transient stress induced by solar UV radiation, thereby bolstering cutaneous resilience.

1
Relative abundances of subject V1 clustered by the four conditions unexposed, exposed, placebo, and sunscreen visualized by Genome Explorer (BaseClear B.V.).

F I G U R E 2 3 F I G U R E 4
Box-plots illustrating alpha diversity (Shannon Indices) clustered by conditions.(a) full data consideration; (b) data without Cutibacterium acnes abundance.F I G U R E 3 Visualization of the (a) coefficients from Songbird for two Lactobacillus crispatus species and the resulting (b) log-ratios between Lactobacillus and Cutibacterium acnes.Coefficients from Songbird of two Lactobacillus sequences.T:YES is the association of these features to sunscreen relative to T:NO, which is the control of that variable.RAxML results from the ARB-Silva phylogenetic tree of sequences classified as Lactobacillus, showing that their nearest neighbours are Lactobacillus crispatus.

F I G U R E 5
The network results of SPIEC-EASI showing the most resolved lineage and the sequence identifier associated with organism.T A B L E 4Integration of SPIEC-EASI network results, clusters, and associated Songbird coefficients.

F I G U R E 6
Metallic cation and iron-siderophore transporters for each of the 5 organisms associated with the treatment.Pathways that are present in more than one of the organisms are coloured orange.Corynebacterium (yellow), Streptococcus (blue), Micrococcus (green), Kocuria (purple), Staphylococcus (red).
Ingredient list of Placebo and SPF 20 sunscreen formulation.
T A B L E 1