Fig 1.
Repeated exposure to oxazolone induces skin thickening and recruitment of immune cells.
OXA-induced inflammation was studied by measuring thickness of epidermis (A) and dermis (B) in H&E-stained skin sections. The number of total cells (C), lymphocytes (D), eosinophils (E) and neutrophils (F) were counted from H&E-stained skin tissue under a light microscope at 1000x magnification. P values for epidermis and dermis were calculated with Brown-Forsythe and Welch ANOVA and Dunnett’s T3 multiple comparison tests and P values for immune cells were calculated with Kruskal-Wallis and Dunn’s multiple comparison tests. Yellow significance stars indicate the test between naïve and positive or naïve and negative, and blue significance stars between negative and positive group. P values: * < 0.05. ** < 0.01, *** < 0.001, **** < 0.0001.
Fig 2.
Changes in mouse skin microbiota were induced by acetone and olive oil with and without oxazolone.
The most abundant bacterial families (n = 15) were chosen for visualisation using their relative abundances in the communities (A). Distance-based redundancy analysis (dbRDA) was performed for each timepoint to observe microbial community differences between the groups over time (B-E). Relative abundance data was used to calculate Bray-Curtis dissimilarity indices between samples with group and cage as categorical explanatory variables, while the ASVs were the response variables. In addition, DNA extraction batches and library sizes were included as conditioning variables and partialled out. Ellipses on the top of symbols mark the 85% confidence interval and the P values of ANOVA like permutation test for dbRDA (permutations = 9999) pairwise comparisons are included in each box.
Fig 3.
Microbial abundance in different groups and timepoints based on variance partition analysis.
Hierarchical Modelling of Species Communities (HMSC) model was utilised for the most abundant ASVs (n = 100) to examine changes of ASV abundance between groups and timepoints. ASV abundances were included as a response variable, while the interaction of group and timepoint in addition to library size were included as explanatory variables to model the communities. Each mouse and cage were included as random effects. The ASVs with the most variance originating from the interaction of positive group and timepoint (A) and of negative group and timepoint (B) are shown. The P values were calculated using differential abundance analysis using generalised linear models with zero-inflated log-normal and negative binomial distribution, including FDR correction. Yellow colour indicates the test between naïve and positive or naïve and negative, and blue between negative and positive group: * < 0.1, ** < 0.01, *** < 0.001.
Table 1.
Pairwise statistical test results of microbial abundance of within group timepoints based on HMSC.
Each ASV, its lowest rank, statistical test, comparison, and the result are given in each column. GLMM = generalised linear mixed model.
Fig 4.
Associations between skin thickness or immune cell recruitment and ASVs.
Positive group Spearman correlations were calculated between ASV abundance and epidermis/dermis thickness (A) as well as total immune cell counts (B). Cells were counted in 15 high-power fields at 1000 X magnification. Spearman Rho rank correlation test with FDR correction was used to calculate P values: < 0.1, * < 0.05, ** < 0.01, *** < 0.001.