Is cutaneous microbiota a player in disease pathogenesis? Comparison of cutaneous microbiota in psoriasis and seborrheic dermatitis with scalp involvement

Patients

The diagnosis was established in each case by checking the patient’s file who were followed up and by confirming their disease with clinical examination. Patients with well-defined erythematous plaques covered by silvery-white scales on elbows, knees and other parts of the body were examined for scalp involvement by psoriasis vulgaris. Typical lesions on the scalp were taken as psoriasis vulgaris of the scalp. Poorly defined, salmon-coloured plaques with a greasy, yellowish scale on the seborrheic areas such as scalp and face were considered as seborrheic dermatitis. The diagnosis was confirmed in each case by a dermatologist and was rechecked by other dermatologists. If the dermatologists agreed on the diagnosis, the patient was recruited.

The controls were obtained from healthy volunteers who came to the outpatients’ clinic for cosmetic concerns such as hyperpigmentation, wrinkles, sun protection etc. and who had no lesions on the scalp.

Procedure

The study included consenting patients diagnosed with psoriasis vulgaris and seborrheic dermatitis and healthy volunteers (mean ages, 45.9, 48.1, and 47.7 years, respectively). The cases and controls were matched for age. Each group had five female and five male participants who had used no treatment (including topicals, medicated shampoos, antibiotics or any other medications) for the last two months, who had no other chronic diseases, and who had not taken a bath for at least 24 h before sample collection. Each of the disease groups and the control group had 10 subjects. So we had a total of 30 patients and 50 swabs. (20 swab samples were taken from the lesions, 20 swab samples were taken from the non-lesional parts of the scalp and 10 swab examples were taken from the healthy group).

The psoriasis patients had never received systemic psoriasis treatment. Psoriasis area and severity index (PASI) and seborrheic dermatitis area and severity index (SEDASI) were used to score the severity of the diseases. The average psoriasis area and severity index score was 9.75 ± 6.39 (range, 3.5-24.3) for psoriasis vulgaris patients, while the average seborrheic dermatitis area and severity index score was 17.3 ± 8.83 (range, 5-28) for patients with seborrheic dermatitis.

The study was planned as a preliminary study, and the sample size of the groups was the least required number for significant statistical analysis. The study approval was obtained from the Istanbul Medeniyet University ethics committee (15 August 2018/0310). The study protocol was registered at https://www.clinicaltrials.gov with the number NCT03807453.

Sample collection and preservation

Fifty swab samples were taken from both the lesions and lesion-free hairy scalp of the patients and the scalp of controls. Ten swab samples each were taken from the lesions of psoriasis, the lesions of seborrheic dermatitis, the non-lesional parts of psoriais, the non-lesional parts of seborrheic dermatitis and from the healthy group. A single investigator (M.A.K.) selected the sampling sites and collected the samples from all the cases and controls. Sterile swabs were placed inside DNA/ RNA Shield™ collection tube w/swab tubes. Samples were taken with different swabs from the lesions and lesion-free areas from the scalps of the same patients. The investigators usually tried to collect the samples from the controls from the same areas, usually the parietal area or vertex.The samples were taken by rubbing swabs soaked with sterile distilled water on the area for one minute. The swabs were immediately put in sterile solution, containing DNA/RNA Shield, inside closed tubes at -20°C until microbiota analysis was performed.

16S rRNA gene sequencing

In order to reveal the bacterial compositions of the samples, Amplicon-based sequencing targeting the 16S rRNA regions, which are used for bacterial identification and classification, was performed. For that purpose, ZymoBIOMICS targeted sequencing service for microbiome analysis (Zymo Research, USA) was used for DNA isolation and sequencing. The V3-V4 region of the 16S ribosomal RNA (rRNA) gene was targeted, and the prepared library was sequenced on the Illumina MiSeq sequencing platform in paired-end reading mode.

Bioinformatical analysis

Bioinformatical analysis was carried on the FASTQ reads obtained from the sequencing analysis. There were 2,131,964 reads obtained for 50 samples. These reads were imported into the quantitative insights into microbial ecology (Qiime2) software pipeline, version 2020.2, for analysis.²⁴ The sequences were demultiplexed using the “demux” plugin. DADA2 was chosen for quality filtering, trimming of adapters, denoising, error removal, and merging of the paired-end sequences. The sequences were rarefied to an equal depth of 8,350 sequences per sample, which retained 43.6% of features in all 50 samples. For taxonomy assignment, the 16S rRNA sequence database from bacteria and archaea was obtained from NCBI. The “q2-feature-classifier” plugin with the naïve Bayes classifier was trained on the V3-V4 region of the database. The sequences assigned to “eukaryote” and “mitochondria” and those that were “unassigned” were filtered out before proceeding to the downstream analysis.

Alpha and beta diversity analyses

The diversity within a sample is called alpha diversity.²⁵ Three common alpha diversity measures were used: the number of observed operational taxonomic units reflecting the richness (how many different bacteria are present in the sample) only, Shannon’s diversity index providing equal weight to richness and evenness (how evenly are these bacterial taxa distributed within the sample), and Faith’s phylogenetic diversity index accounting for relationships based on phylogenetics.²⁶

Beta diversity shows the variety of different bacteria among the samples. The principal component analysis with the unweighted or weighted UniFrac distance analysis was used. As the phylogenetic beta diversity measurement, the UniFrac measure with two types, weighted (qualitative) (considering abundance) and unweighted (quantitative) (considering presence or absence of bacterial taxa) analyses, were conducted.²⁷ By this way, clustering patterns between the different patient groups were examined to reveal if any inter-group differences exist.

Statistical analysis

Relative frequency data at the taxonomic level 2 (phylum) and 6 (genus), alpha and beta diversity metrics were exported from Qiime2. All statistical analyses and visualizations were carried using R (version 3.6.1). The Shapiro-Wilk test was used to test the distribution normality. For the alpha diversity analysis, one-way analysis of variance followed by the post hoc Tukey’s honest significant difference test, or the Kruskal-Wallis analysis of variance comparison followed by the posthoc Dunn’s test were applied for the normally and non-normally distributed data, respectively. The permutational multivariate analysis of variance test was performed using Qiime2 to determine the significance of differences in beta-diversity distances between the compared groups. The differences in the relative abundance of taxa between the groups were tested using the Mann-Whitney U test.

Psoriasis vulgaris

The alpha diversity analysis is shown in Figures 1a-1c. The skin microbiota values of the non-lesional psoriasis vulgaris samples lay between those of the healthy control and psoriasis lesional groups, but there was a statistical difference only between the lesional and control groups. The beta diversity analysis also revealed different clustering for the control and lesional psoriasis vulgaris samples [Figures 1d-1g]. On the other hand, the comparison of the lesional and non-lesional psoriasis vulgaris samples showed overlapping clustering. The pairwise comparison using the permutational multivariate analysis of variance test revealed a significant difference between the control and lesional group (unweighted: pseudo-F = 2.73, q = 0.003; weighted: pseudo-F = 5.29, q = 0.045), while no significant difference was detected between the lesional and non-lesional samples of psoriasis. (Unweighted: pseudo-F = 1.18, q = 0.27; weighted: pseudo-F = 0.9, q = 0.57).

Close

Figure 1:

Alpha diversity analysis using (a) observed operational taxonomic unit: (b) Shannon’s diversity index and (c) Faith’s phylogenetic diversity index, compared between the case groups. Differences in observed operational taxonomic unit across the case groups were compared using the one-way analysis of variance (top dashed line) and post hoc Tukey’s honest significant difference test (Different superscript letters indicate significant differences, P < 0.05) while Shannon’s diversity and Faith’s phylogenetic diversity indices were compared using the Kruskal-Wallis analysis of variance (top dashed line) and post hoc Dunn’s test (grey solid lines). Beta diversity analysis using principal component analysis plots based on unweighted (d and e) and weighted (f and g) Unifrac distances. Ellipses correspond to the 95% confidence intervals for each case group. OTU: Operational taxonomic unit, C: Control, PL: Psoriasis vulgaris lesional site, PLS: Psoriasis vulgaris non-lesional site, SDL: Seborrheic dermatitis lesional site, SDLS: Seborrheic dermatitis non-lesional site

Export to PPT

To reveal the differences in the bacterial composition of the control and psoriasis vulgaris groups, phylum and genus-based comparisons were carried out [Figure 2]. Actinobacteria, Firmicutes and Proteobacteria were the three phyla dominating the overall community in both healthy and psoriasis skin microbiota [Figure 3]. However, the balance between these three groups was determined to be disturbed in the psoriasis vulgaris microbiota. The relative abundance of Actinobacteria decreased in both lesional and non-lesional psoriasis groups compared to the control [Figure 3a]. The differences in the relative abundance of Proteobacteria were not significant [Figure 3b]. The pairwise comparisons of the psoriatic lesional and non-lesional microbiota compositions at the phylum level did not show any significant difference.

Close

Figure 2:

Most abundant phyla (top) and genera (bottom) across the case groups. The median relative abundance of the three most abundant phyla and ten most abundant genera are used for the illustration

Export to PPT

Close

Figure 3:

Comparison of the relative per cent abundances of the three dominant phyla, (a) Actinobacteria, (b) Proteobacteria and (c) Firmicutes, between the case groups. The differences between the two case subgroups were evaluated using the Mann-Whitney U test. The respective p values for the significant results are given in the figures. The differences observed in Proteobacteria were determined to be non-significant. C: Control, PL: Psoriasis vulgaris lesional site, PLS: Psoriasis vulgaris non-lesional site, SDL: Seborrheic dermatitis lesional site, SDLS: Seborrheic dermatitis non-lesional site

Export to PPT

However, there was a significant increase in the abundance of Firmicutes in the lesional and non-lesional psoriasis groups when compared to the controls [Figure 3c].

The comparison of the control and psoriasis vulgaris groups showed substantial alterations in some genera. In the lesional psoriasis samples, there was a statistically significant decrease in the Propionibacterium levels and a significant increase in the levels of Staphylococcus, Streptococcus, Neisseria, Mycobacterium, Finegoldia, Haemophilus, Azospirillum and Ezakiella [Figure 4]. The comparison between the non-lesional and lesional psoriasis vulgaris groups showed an increase in the Nocardioides and Anaerococcus abundance in the lesional group. In addition, when compared with the control group, Staphylococcus, Streptococcus and Haemophilus abundances were shown to be increased in the lesional samples.

Close

Figure 4:

Relative abundance of the genera of interest that were determined to significantly differ between the case groups. The pairwise comparisons were carried out using the Mann-Whitney U test for all case groups. The minimum (min), mean, median and maximum (max) per cent abundances in the respective case groups are presented in the charts under each genus plot

Export to PPT

Seborrheic dermatitis

The alpha diversity analysis revealed enhanced richness and evenness in seborrheic dermatitis, and the difference was more prominent and statistically significant in the lesional group. The diversity indices for the non-lesional seborrheic dermatitis samples lay between the values of the control and lesional seborrheic dermatitis groups, but the difference between the lesional and non-lesional samples was statistically significant only according to Faith’s phylogenetic diversity analysis [Figures 1a-1c]. The beta diversity analysis also confirmed the significant diverse clustering of the lesional seborrheic dermatitis group compared to the controls (unweighted: pseudo-F = 3.42, q = 0.003; weighted: pseudo-F = 6.43, q = 0.02) [Figures 1d-1g]. When the lesional and non-lesional groups of seborrheic dermatitis were compared with the beta diversity analysis, the unweighted UniFrac distance revealed a significant difference (unweighted: pseudo-F = 2.69, q = 0.003; weighted: pseudo-F = 1.3, q = 0.45).

The differences in the bacterial composition of the control and seborrheic dermatitis groups with phylum and genus-based comparisons are shown in Figure 2. Actinobacteria, Proteobacteria and Firmicutes were the three dominant phyla in the disease group [Figure 3]. The pairwise comparisons of the lesional seborrheic dermatitis and controls revealed a significant decrease in Actinobacteria [Figure 3a] and an abundance in Firmicutes [Figure 3c]. The increase in Firmicutes was also significant in the non-lesional group compared to the controls. The differences in Proteobacteria were not significant [Figure 3b]. There were also no statistically significant differences in the abundances between the lesional and non-lesional seborrheic dermatitis groups at phylum level.

The genus-based differences are illustrated in Figure 4. Compared to control, the only decrease in the lesional seborrheic dermatitis group was seen in Propionibacterium, and increased genera detected were Micromonospora, Staphylococcus, Streptococcus, Leptotrichia, Azospirillum, Aquabacterium, Neisseria and Enhydrobacter. There were no significant differences in the pairwise comparison of the control and non-lesional groups and that of the lesional and non-lesional seborrheic dermatitis groups at the genus level.

Psoriasis vulgaris and seborrheic dermatitis

The alpha diversity measurements showed similar richness in the lesional psoriasis vulgaris and seborrheic dermatitis groups, as well as the non-lesional groups [Figures 1a-1c]. Similarly, the UniFrac measures showed overlapping diversity shared by the lesional or non-lesional psoriasis vulgaris as well as seborrheic dermatitis microbiota [Figures 1d-1g], (lesional psoriasis vulgaris and lesional seborrheic dermatitis: unweighted: pseudo-F = 1.44, q = 0.097; weighted: pseudo-F = 0.37, q = 0.95, and non-lesional psoriasis vulgaris and non-lesional seborrheic dermatitis: unweighted: pseudo-F=1.44, q = 0.11; weighted: pseudo-F = 0.21, q=0.97). A significant difference was obtained from the comparison of the lesional psoriasis vulgaris and seborrheic dermatitis groups using the unweighted UniFrac analysis (unweighted: pseudo-F = 2.36, q = 0.008; weighted: pseudo-F = 1.27, q = 0.45).

Phylum-and genus-based comparisons of the psoriasis vulgaris and seborrheic dermatitis groups are shown in Figure 2. When the differences at the phylum level were compared between the two disease groups, no significant difference was found in lesional or non-lesional regions [Figure 3]. The comparison of the lesional disease groups at the genus level revealed that three distinct taxa, Mycobacterium, Ezakiella and Peptoniphilus were only present in psoriasis vulgaris [Figure 4]. An increased abundances of Nocardioides and Aquabacterium were observed in non-lesional seborrheic dermatitis when compared with non-lesional psoriasis vulgaris regions [Figure 4].

Limitations

The small sample size is a limitation of the study. Studies comparing the two diseases with larger sample sizes are necessary. Higher taxonomic resolution down to species level is necessary and can be achieved by third-generation sequencing technologies, which can produce longer reads, or shotgun metagenomics sequencing that would allow for the examination of not only bacteria but also yeasts such as Malassezia. Other omics analyses may also be beneficial to reveal the effect of dysbiosis on the host immune system.