Microbial community profiling shows dysbiosis in the lesional skin of Vitiligo subjects

Healthy human skin harbours a diverse array of microbes that comprise the skin microbiome. Commensal bacteria constitute an important component of resident microbiome and are intricately linked to skin health. Recent studies describe an association between altered skin microbial community and epidemiology of diseases, like psoriasis, atopic dermatitis etc. In this study, we compare the differences in bacterial community of lesional and non-lesional skin of vitiligo subjects. Our study reveals dysbiosis in the diversity of microbial community structure in lesional skin of vitiligo subjects. Although individual specific signature is dominant over the vitiligo-specific microbiota, a clear decrease in taxonomic richness and evenness can be noted in lesional patches. Investigation of community specific correlation networks reveals distinctive pattern of interactions between resident bacterial populations of the two sites (lesional and non-lesional). While Actinobacterial species constitute the central regulatory nodes (w.r.t. degree of interaction) in non-lesional skin, species belonging to Firmicutes dominate on lesional sites. We propose that the changes in taxonomic characteristics of vitiligo lesions, as revealed by our study, could play a crucial role in altering the maintenance and severity of disease. Future studies would elucidate mechanistic relevance of these microbial dynamics that can provide new avenues for therapeutic interventions.

Non-Lesional and Lesional samples. Identification of core taxa (genera/OTUs) was done using a bootstrapping approach, wherein a random subset (i.e. 75% of samples) were drawn from the whole set of samples, and core microbial taxa were deduced from the randomly drawn sample set. This approach was iterated 1000 times to arrive at 1000 sets of core microbial taxa for the whole sample set. A union of core taxa for all 1000 iterations was generated and a bootstrap score was assigned to each taxon based upon it's consistency in appearing as core taxon in all iterations. Bootstrap score was scaled between 0-100, wherein a score of 100 for a taxon indicated that the given taxon appeared as core in all 1000 iterations. In this study, only those taxa were considered as core which had minimum bootstrap score of 50. In this figure, each bar (blue, red) signifies the bootstrap score for the corresponding genus/OTU identified as core.    Intra-community network analysis: Non-lesional skin b).
Supplementary Figure S4 Intra-community network analysis: Lesional skin Figure S4b: Spearman correlation based micorbial communtiy interaction network generated for lesional samples. Nodes have been colored according to the phylum level affiliation of the network members, wherein each network members represents a microbial genus.   The assignment process proceeds in the following manner. If greater than two-thirds of reads in a cluster had an identical taxonomic assignment (i.e. classified by RDP to the same taxon), the OTU cluster is assigned to that taxon. In the event of any taxon in a cluster not attaining a cumulative percentage of 66%, all genus level assignments in that cluster are re-assigned to their immediately higher taxonomic level (i.e. corresponding family level), and the process of checking if any taxon exceeded 66% is repeated. This process is iterated until all clusters obtain a taxonomic assignment Schematic description of the algorithm used for Taxonomic assignment of individual OTU clusters Community composition of lesional and non-lesional skin of Vitiligo subjects. Samples were obtained from anatomically similar sites. Figure S7: Box-plots representing relative abundance analysis of the bacterial taxa discovered in samples obtained from anatomically similar Non-Lesional (Normal) and Lesional (Vitiligo) sites at genus (main) and phylum (inset) levels. Taxa with minimum median abundance of 1% were used for the comparison. The results depicted in the above figure indicate that the microbial abundances observed in the subset of samples (30) from anatomically similar skin sites are similar to that observed during analysis of the whole sample set (40 samples).

Supplementary Figure S5
Supplementary Figure S7 α-diversity trends in lesional (Vitiligo) and non-lesional (Normal) skin samples taken from anatomically similar sites Figure S8: a) Box-plots illustrating the comparison of diversity indices (Chao-1, Fisher, Simpson1-D and Shannon index) between Non-Lesional and Lesional samples from anatomically similar sites. b) Comparison of differences between successive relative contribution values (of the ordered genera) in Non-Lesional and Lesional samples. Only those genera were considered for calculating successive differences in relative contributions that had a minimum relative contribution of 1%. Consistent with the previously observed results for overall population, diversity analysis using the subset of samples from anatomically similar sites indicate similar pattern with respect to various diversity metrics.
Supplementary Figure S8 Core microbiota comparison between lesional and non-lesional skin of vitiligo subjects. The samples were taken from anatomically similar sites. Figure S9: Comparison of core-microbial (a) OTUs and (b) RDP genera between Non-Lesional (Normal) and Lesional (Vitiligo) samples taken from anatomically similar sites. Identification of core taxa (genera/OTUs) was done using the bootstrapping approach. Normal and Vitiligo samples pertaining to anatomically similar sites were observed to share a fairly common core taxa profile (18 OTUs and 17 RDP genera constituting the common core). OTUs pertaining to Enhydrobacter (OTU-116), Intrasporangiaceae (OTU-38), and Porphyrobacter (OTU-160) were observed to be the 'Core' set exclusive to Vitiligo samples. The RDP taxon corresponding to Enhydrobacter was observed to be a genus that is 'Core' exclusively in Vitiligo samples. The results with the samples corresponding to anatomically similar sites were observed to have good similarity with results obtained with the complete set of samples.
Supplementary Figure S9 Ordination Analysis of cutaneous microbiome of lesional and non-lesional skin. The samples were taken from anatomically similar sites   Supplementary Figure S11 Comparison between lesional and non-lesional bacterial composition using Intracommunity network analysis. The samples were taken from anatomically similar sites. Figure S12a: Spearman correlation based micorbial communtiy interaction network generated for non-lesional samples. Nodes have been colored according to the phylum level affiliation of the network members, wherein each network members represents a microbial genus. Also refer to supplementary table S3b.
Comparison between lesional and non-lesional bacterial composition using Intracommunity network analysis. The samples were taken from anatomically similar sites.  Intra-community network analysis of cutaneous microbiota. The samples were taken from anatomically similar sites.

a) b)
Supplementary Figure S13             Table S9: A list of differentially abundant OTUs between lesional (Vitiligo) and non-lesional (Normal) skin obtained using the Wilcoxon test coupled to a bootstrapping approach. In each iteration of the bootstrap method, taxa with significantly different abundance were initially identified using Benjamini-Hochberg p-value correction at an FDR of 0.0001. Subsequently, taxa which were observed as having a significantly different abundance (post BH correction) in at least 99.5% of iterations were retained and are shown below. Samples in this analysis were obtained from anatomically similar sites.