Acidic Skin Care Promotes Cutaneous Microbiome Recovery and Skin Physiology in an Acute Stratum Corneum Stress Model

Cutaneous microbiome and skin physiology are important indicators of the epidermal homeostasis status. The microbe biodiversity in the upper epidermis is influenced by the interaction with exogenous factors. This interaction includes environmental aspects encompassing exposure to a specific professional or leisure activity or skin care habit [1-5], e.g., microbiota from cohabiting family members are shared and also with their dogs [6]. Exogenous stressors like mechanical abrasion, chemical irritants, and ultraviolet irradiation have an impact on functional homeostasis and the microbiome [3, 7, 8].

The complex interaction and variability of exogenous factors influencing the homeostasis of an organism are summarized by the term “exposome” [9, 10]. Lifestyle factors like food with a high content of antioxidative ingredients, living with a partner or a dog, as well as smoking, hygiene procedures, prescription drugs (e.g., antibiotics and immuno-suppressants) have been shown to influence both skin physiology [11, 12] and the microbiome [4, 13, 14]. Studies with an integrated look at skin physiology and the microbiome might reveal the relation between the homeostasis of functional skin parameters and adaptation of the microbiome.

The importance of the epidermal acid mantle has been described [15-18]. The understanding of the origin of the stratum corneum (SC) acid buffer system is mainly based on phospholipid/free fatty acid, histidase, and Na+–H+ exchanger-1 (NHE1) pathways [19, 20]. Acidic skin pH resides in the upper SC and shows a gradient toward neutral values in the deeper part of the epidermis in relation to active, energy-requiring mechanisms involved in keratinocyte differentiation processes [21]. The consequences of altered epidermal functions associated with an elevated skin pH are seen in inflammatory skin diseases, under hyperhydrated as well as very dry SC and skin infections [22, 23].

An acidic pH is important in leave-on and rinse-off skin care products [24]. Acidic skin pH care exerts positive effects on skin physiology and specific microbial colonization patterns, especially in atopic dermatitis [22, 25, 26].

The skin microbiome has been characterized by specific patterns in highly hydrated areas and sebum-rich areas [27-29]. The relation of sebum and hydration levels with microbiome diversity was confirmed specifically in the facial region [30-33]. The use of an emollient in infants at risk for atopic dermatitis decreased the surface pH and subsequently increased the proportion of Streptococcus salivarius [34].

Stress models are used in assessing barrier-related parameters to reveal pathological mechanisms and modulating effects during the recovery phase [35-37]. This so-called “epidermal treadmill stress model” reveals underlying pathophysiological mechanisms of skin diseases with altered epidermal functions [38]. The tape stripping (TS) stress model was adapted to microbiome research: microbiome changes in the epidermis upon an acute barrier disruption by sequential TS have been characterized in humans [39]. Regeneration of the diversity after acute removal from superficial parts of the SC induces active repair processes. Bacterial deoxyribonucleic acid (DNA) was evidenced using independent detection techniques down to the dermis and the dermal adipose tissue of normal human skin [40].

In the present study, we studied the short-term and mid-term recovery of skin physiology in relation to the microbiome changes after TS. Then modulation of recovery was studied in untreated areas compared to two cosmetic leave-on lotions with different pH (pH 5.5 vs. pH 9.3). We hypothesized that (i) barrier disruption with a mild acute stress model would induce significant changes in skin physiology, (ii) the stress model would lead to an immediate decrease in the diversity of the microbiome, (iii) the recovery of both skin physiology and microbiome alteration could be accelerated with leave-on skin care, (iv) an acidic pH of the leave-on lotion would be favorable for the recovery of skin physiology and the microbiome.

The study was approved by the Ethical Committee of the Euroderma Clinic in Sofia on July 9, 2020, with protocol number 2/July 9, 2020. The 25 healthy volunteers gave their written informed consent prior to the start of the study. The statistical analysis was performed with Prism 6. The male-female ratio was 10:15; mean age 30.7 years (standard error of the mean 2.1), body mass index 21.7 (standard error of the mean 0.8), smoker: nonsmoker ratio 12:13. The mean room temperature was 19.7–20.2°C and relative humidity 44.1–46.7% as recommended [41, 42]. All treatments, measurements, and microbiome assessments were performed on the volar forearm during the late morning and the early afternoon hours.

Baseline values on SC hydration (SCH), surface pH, skin color, barrier function, and microbiome analysis were assessed. TS was performed with 5 sequential strippings with Corneofix® applied with standardized pressure (with the help of stamp with a built-in spring exerting pressure of 225 g/cm²) for 5 s [43, 44]. Three different areas on the volar forearms received TS and one remained unstripped. Skin color, barrier function, pH, SCH, and microbiome swabs were obtained before, immediately after TS, 2 h after TS, after 2 days and 7 days.

Epidermal barrier function was measured with a Tewameter TM300 (g/m²h). SCH was assessed with Corneometer CM 825 (arbitrary units). Surface pH was quantified with the pH-Meter (pH-units). Skin color was measured with the Mexameter MX 18 (erythema and melanin index; both arbitrary units). All instruments are manufactured by Courage&Khazaka electronics (Cologne, Germany).

Lotion A with a pH of 5.5 and lotion B of the same composition with a pH of 9.3 (adjusted and buffered by the addition of bicarbonate and sodium hydrogencarbonate) were compared. The pH 9.3 was selected for formulation reasons and to have a clear difference in applied pH. The International Nomenclature Cosmetic Ingredients list of lotion ingredients was as follows: aqua, glycerin, cetearyl alcohol, sorbitol, hexyldecanol, hexyldecyl laurate, Chamomilla recutita flower extract, allantoin, sodium cetearyl sulfate, citric acid, sodium hydroxide, sodium acetate, sodium ascorbate, xanthan gum, dimethicone, parfum, alcohol, phenoxyethanol, sodium benzoate, benzyl alcohol. Each volunteer had all four treatment areas (lotion A, lotion B, TS untreated and nonstripped and untreated) and served as his own control on both volar forearms. The assignment to the different treatment modalities was selected according to a randomization list for each volunteer. A fingertip unit of both lotions was applied twice daily for 6 consecutive days.

DNA was extracted and purified into 50 µL elution buffer using the DNeasy PowerSoil Pro-Kit (Qiagen). Empty cremes were analyzed as controls. 2.6 pg of DNA from Salinibacter ruber (DSM 13855) was added as an internal spike-in control. Amplification of the V3-V4 region of the 16S-rRNA gene was done with the ultra-clean production multiplex polymerase chain reaction (PCR) master mix (Qiagen) from 5 µL DNA using the primers CCTACGGGNGGCWGCAG (forward) and GACTACHVGGGTATCTAATCC (reverse) following the Illumina 16S library preparation protocol (https://support.illumina.com/documents/documentation/chemistry_documentation/16s/16s-metagenomic-library-prep-guide-15044223-b.pdf). Sequencing was done on an Illumina MiSeq (v2 reagents) with 2 × 250 bp paired-end reads.

After sequencing, the standard UPARSE 16S protocol was used to merge the paired reads by their 3′-ends, filter reads to a minimum length of 100 bp, followed by clustering unique reads into operational taxonomic units (OTUs) using a 97% identity threshold and quantifying OTU abundances by mapping reads to the OTU sequences [45]. OTUs with a relative abundance of less than 0.01% were discarded. Each OTU sequence was searched in the NCBI Targeted Loci 16S database (NCBI accession PRJNA33175) using NCBI BLAST [46] with an identity cut-off of 97% and E-value cut-off of 0.01, and assigned to the annotated species. If no match was found, the sequence was also searched in the NCBI Nucleotide database using the same cut-offs but also excluding unclassified species or environmental sequences (based on the taxon names in the NCBI taxonomy). If no match was found in either database, the OTU was labeled as unclassified. OTUs with multiple best matches were assigned to the least common ancestor of all alignments with the same BLAST bit score. OTUs assigned to the Salinibacter ruber spike-in were discarded.

Alpha diversity is measured as the number of OTUs in each sample. Alpha diversity is a parameter in microbiome research that represents the diversity/heterogeneity in a single microbiome sample of a specific environment. Here we define alpha diversity as the number of OTUs in each sample. Relative abundances for species were calculated from classified OTUs and summarized on taxonomic ranks from species to phylum. Visualization of microbiome data was done using ggplot2 [47].

One-way ANOVA analysis within time groups was calculated followed by Sidak’s multiple comparisons test with a significance level of p < 0.05.

Comparison of diversity between treatment groups was performed with the Durbin-Conover test and Holm multiple testing correction using ggstatsplot [48], also using a significance level of p < 0.05.

Transepidermal water loss (TEWL) and erythema index were assessed immediately after and 120 min post-TS procedure to evaluate the validity of epidermal barrier disruption (Fig. 1a). A significant increase immediately after (p < 0.001) and 120 min (p < 0.01) after TS was observed, confirming the validity of the stress model of skin barrier disruption. No variations were noted for melanin and erythema index (online suppl. Fig. S4; see www.karger.com/doi/10.1159/000526228 for all online suppl. material), indicating that the stress model was of mild intensity.

A significant increase in SCH was observed on day 2 and day 7 in comparison to untreated skin sites for both lotions (p < 0.001). The change in SCH upon product application is shown in Figure 1c. Lotion A showed slightly higher values compared to lotion B without reaching statistical significance.

A clear difference in skin surface pH was observed at day 2 and day 7 (Fig. 1d). Lotion B application resulted in a significant increase in pH (p < 0.001) both at day 2 and day 7 in comparison to lotion A and untreated TS area. Skin surface pH did not differ between lotion A and the untreated TS site. However, the values of all groups increased slightly (nonsignificant) compared to baseline, which is within normal variations with a 1-week study.

No significant difference between sites in TEWL values (Fig. 1b), erythema and melanin index (online suppl. Fig. S1) could be revealed at day 2 and 7, indicating no relevant induction of irritation (increased erythema index) or desquamation (decrease of melanin index).

Data from 24 out of the 25 individuals were analyzed (one individual was excluded because of missing samples). The number of sequencing reads per sample ranged from 1,117 to 168,807 (mean 207,764, median 11,356) and was on average higher in the treated samples and lowest in untreated samples after TS (online suppl. Fig. S2). A comparison between the fraction of numbers of OTUs that are classified to a taxon and the fraction of reads belonging to classified OTUs shows that on average 83% of OTUs are classified, but they comprise on average 95% of the reads, i.e., most unclassified OTUs are of low abundance (online suppl. Fig. S3). The rarefaction analysis (online suppl. Fig. S4) shows the number of OTUs for the 11 samples per individual after subsampling to half read depth, one-fourth read depth, and further dilutions down to 1/1,024. Rarefaction curves reached a plateau in all samples, with at least 95% of OTUs present at half read depth.

The alpha diversity was unchanged during the 7 days of the study in the nonstripped and nontreated control areas (Fig. 2a), with mean values from 183 to 204. TS significantly reduced the alpha diversity to 125. The values increased in the untreated areas to 177 after 2 days and 228 after 7 days (Fig. 2b). Both lotions significantly accelerated the re-population regarding the alpha diversity already at 2 days compared with the untreated areas to a mean of 281 for lotion A and 272 for lotion B (Fig. 2c). After 7 days, no significant differences between the groups were found, but we observe a decrease of alpha diversity compared with the measurements from 2 days after TS in both lotions to levels compared to untreated areas (Fig. 2d).

The distribution of phyla (Fig. 3a) was unchanged in the control area over the 7 days of the study. After TS, the relative abundance of Proteobacteria was increased, whereas Actinobacteria abundances decreased. This shift from Actinobacteria toward Proteobacteria recovered partially after 2 days and normalized almost completely at day 7. Treatment with lotion A and lotion B accelerated the normalization: already after 2 days, the distribution was comparable to the baseline nonstripped area and remained similar on day 7. Examination of the most abundant orders revealed an increase of Burkholderiales after TS as well as a reduction, in particular, of the lower abundant orders like Corynebacteriales, Micrococcales, Propionibacteriales, and Lactobacillales, with the exception of Xanthomonadales, where the relative abundance increased (Fig. 3b). Abundances of Bacillales and Pseudomonadales remained largely unchanged.

The mean relative abundances of the 20 most abundant genera in each group are shown in Figure 4. The first row shows control areas, the second row the TS/untreated area, and the third and fourth row show lotions A and B, respectively. The control group showed a stable genus composition over the whole period, with Staphylococcus as the dominant genus. TS induced a relative increase of Brevibacillus and a reduction of Staphylococci and Corynebacteria. After 7 days, the genus composition was similar to control areas. Treatment with both lotions resulted in normalization of the genus distribution already at day 2.

Noteworthy is a major reduction of the four typical skin-associated genera Corynebacterium, Cutibacterium, Staphylococcus, and Streptococcus after TS (Fig. 5a). Both lotions accelerated their recovery to normal and even elevated levels compared with control and untreated groups. When plotting the combined relative abundance of these four genera against the alpha diversity in each sample (Fig. 5b), we observe no change over time in control areas, but a shift toward low abundance and low alpha diversity in the TS areas. The groups of both lotions show a similar distribution as in the control group and no relevant shift of both variables between the two time points, indicating that lotion-treated skin areas returned to almost normal skin flora composition already after 2 days.

Figure 6 shows the mean relative abundance of the OTUs that are assigned to different Staphylococcus species (S. aureus/S. simiae, S. capitis/S. caprae, S. capitis/S. caprae/S. epidermidis, as well as summarized other Staphylococcus spp.) and the number of samples in which they were observed. TS induced a similar reduction in all Staphylococcus OTUs. This reduction returned to almost normal levels after 2 days but only after 7 days showed comparable abundances to untreated areas.

Exogenous environmental influences alter skin physiology and microbiome parameters in humans [2, 3, 7]. We hypothesized that an acute epidermal stress model would be suitable to study the effects of acidic skin care on recovery dynamics both of skin physiology and epidermal microbiome diversity and other microbiome parameters. The TS stress model is an established model to reveal subtle differences in pathological states and to study treatment responses [49, 50]. TS was also described to serve as a model in studying skin microbiome [39, 51]. We selected a mild TS model in healthy volunteers which showed to be correctly executed with mild barrier disruption and no measurable irritation.

The recovery of skin physiology was accelerated by skin care, however, not in a significant pH-dependent manner. Both lotions (only differing in the pH-related buffer substances) revealed a positive effect on barrier-related parameters. Only for skin surface pH, a significant difference was observed at day 2 and day 7 with an increased pH induced by basic/alkaline pH of lotion B. No significant differences between sites were detectable in barrier function and irritation at day 2 and day 7.

TS significantly reduced the alpha diversity, followed by a recovery over 7 days without treatment. Thus, our assumption of TS being a relevant model for studying the dynamics of microbial recovery after superficial removal of SC was confirmed. The re-population of the stripped skin section is coming both from lateral ingrowth and from preserved, deeper parts of the skin and hair follicles [39].

Both lotions accelerated the recovery of the microbiome, showing a significantly higher alpha diversity already after 2 days, with a slightly higher induction rate for lotion A (lower pH). The alpha diversity after 7 days of treatment is lower than after 2 days of treatment. There was significant increase of alpha diversity at day 2 induced by TS seen as an acceleration of the repopulation process. This process was still ongoing at day 7 but less pronounced comparing lotion A and B to untreated tape-stripped area. This suggests that the repopulation of stripped area started with an initial overcompensation caused by a replacement flora that rapidly fills the empty niche. After a few days, the normal flora replaces this transitional microbiome. Since the pH was identified as not significantly relevant for the accelerated re-population after TS, we assume that the ingredients of both formulations are the major drivers. In vivo studies are not designed to identify the role of specific ingredients. After TS, the abundance of Burkholderiales increased, whereas abundances of several other genera decreased. In particular, the abundance of the common skin-associated genera Staphylococcus, Corynebacterium, Streptococcus, and Cutibacterium was reduced after TS. Similar to the alpha diversity, the shift in the microbiome composition returned to a normal state over the course of 7 days in untreated areas. Topical treatment with lotions accelerated the normalization of the taxa distribution, and already after 2 days, the distribution was comparable to the control nonstripped areas and remained similar on day 7.

The inter-individual differences in species distribution between the volunteers (baseline levels) represent a limiting factor when interpreting microbiome data. The high variation makes it challenging to identify significant and relevant effects induced by pH changes since they might be masked by the inter-individual variation in the present cohort size.

The validity of this model has been proven as shown in Figure 1. This model was used to evaluate the effect of skin care with different pH buffering of a marketed cosmetic product on the microbiome and skin physiology parameters. No significant change in TEWL at day 2 and day 7 was registered. A probable limitation is the spontaneous barrier recovery within 2 days after TS, reflected by the nonsignificant difference between untreated TS and untreated unstripped area already at day 2. Our mild stress model only induced a mild barrier damage with a relatively fast barrier recovery [38]. A more pronounced increase in TEWL or repetitive damage would have caused more severe disturbances, with a longer regeneration phase. The chosen mild barrier disturbance corresponds to the everyday situations.

SCH increased significantly under both treatments already at day 2 and was still higher at day 7 in comparison to untreated sites. Micro-morphological changes were detectable in a multiethnic study population treated for 14 days with an acidic lotion by optical coherence tomography and reflectance confocal microscopy [52]. Changes in hygiene routines have only limited effects on the microbiome and only detectable after usage of an antiperspirant and foot powder; yet no changes were observed after usage of lotions [14, 53].

Our findings provide evidence that a mild stress model (TS) in the assessment of skin surface microbiome homeostasis in relation to skin physiology is an applicable model. SCH increased significantly after treatment with both lotions already at day 2 and was still higher at day 7 compared to untreated sites. The process of normalization of the stressed microbiome parameters (including alpha diversity, mean relative taxa distribution on the phylum-level, abundance of selected genera) was started already after 2 h. and the parameters returned to normal distribution after 7 days.

Our hypothesis could be answered as follows: (i) that barrier disruption with a mild acute stress model induced significant changes in skin physiology, measured as TEWL without leading to a relevant inflammation, (ii) the stress model lead to an immediate decrease in the diversity of the microbiome, (iii) the recovery of both skin physiology and microbiome alteration was accelerated by leave-on skin care, (iv) a significant superiority of the acidic pH lotion compared to the basic/alkaline pH lotion could not be shown during the recovery of skin physiology and the microbiome parameters in this model, but indications for faster microbial rebalancing at pH 5.5 should be analyzed in further studies. In future studies, higher buffer capacities and eventually even using lotions with a pH lower than 5.5 might elucidate the modulating role of pH on microbiome parameters.

The potential mechanisms for the accelerated microbiome normalization could be (a) the composition of the lotion, (b) optimized hydration during the recovery phase, (c) the induced repair mechanism. The formulation might have a positive effect on the SCH and subsequently on cutaneous microbiome and skin physiology. Furthermore, this eventually has an implication in the modulation of exogenous stress-induced epidermal alterations. Further studies would elucidate the role of prolonged application of the acidic skin care in different stress.

We would like to thank the volunteers for their dedicated participation in the study. The patients in this article have given written informed consent for publication of their case details.

The study was approved by the Institutional Ethics Board of the Ethical Committee of the Euroderma Clinic in Sofia from July 9, 2020 with protocol number 2/July 9, 2020. All volunteers signed a written informed consent.

Joachim W. Fluhr and Razvigor Darlenski have received consulting fees from Sebapharma. Benjamin Kaestle, Michaela Arens-Corell, and Lina Praefke are employees of Sebapharma. All other authors have no competing interests regarding the content of this article.

This study was sponsored by Sebapharma, Boppard, Germany.

Conception of the study: Joachim W. Fluhr, Razvigor Darlenski, Nikolai K. Tsankov, Laurent Misery, Benjamin Kaestle, Michaela Arens-Corell, and Lina Praefke. Performing the study: Dessyslava G. Nikolaeva, Joachim W. Fluhr, and Razvigor Darlenski. Microbiome analysis: Peter Menzel and Rolf Schwarzer. Data analysis: Joachim W. Fluhr, Peter Menzel, Rolf Schwarzer, Dessyslava G. Nikolaeva, and Razvigor Darlenski. Interpretation of the data: all authors. Writing of the manuscript: all authors. All authors have read the manuscript, agreed on its submission, and consented to its publication.

All data generated or analyzed during this study are included in this published article and its online supplementary material. The sequencing data for all 264 microbiome samples are available at NCBI Genbank under Bioproject accession number PRJNA713953.