Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2024 May 28.
Published in final edited form as: Ann Rheum Dis. 2022 Dec 12;82(4):507–514. doi: 10.1136/ard-2022-223389

Alterations in the cutaneous microbiome of patients with psoriasis and psoriatic arthritis reveal similarities between non-lesional and lesional skin

Alba Boix-Amorós 1, Michelle H Badri 2, Julia Manasson 3, Rebecca B Blank 3, Rebecca H Haberman 3, Andrea L Neimann 4, Parvathy V Girija 3, Anthony Jimenez Hernandez 3, Adriana Heguy 5,6, Sergei B Koralov 6, Richard Bonneau 2,7, Jose C Clemente 1, Jose U Scher 3
PMCID: PMC11131958  NIHMSID: NIHMS1915798  PMID: 36600182

Abstract

Objectives

To investigate the cutaneous microbiome spanning the entire psoriatic disease spectrum, and to evaluate distinguishing features of psoriasis (PsO) and psoriatic arthritis (PsA).

Methods

Skin swabs were collected from upper and lower extremities of healthy individuals and patients with PsO and PsA. Psoriatic patients contributed both lesional (L) and contralateral non-lesional (NL) samples. Microbiota were analysed using 16S rRNA sequencing.

Results

Compared with healthy skin, alpha diversity in psoriatic NL and L skin was significantly reduced (p<0.05) and samples clustered separately in plots of beta diversity (p<0.05). Kocuria and Cutibacterium were enriched in healthy subjects, while Staphylococcus was enriched in psoriatic disease. Microbe–microbe association networks revealed a higher degree of similarity between psoriatic NL and L skin compared with healthy skin despite the absence of clinically evident inflammation. Moreover, the relative abundance of Corynebacterium was higher in NL PsA samples compared with NL PsO samples (p<0.05), potentially serving as a biomarker for disease progression.

Conclusions

These findings show differences in diversity, bacterial composition and microbe–microbe interactions between healthy and psoriatic skin, both L and NL. We further identified bacterial biomarkers that differentiate disease phenotypes, which could potentially aid in predicting the transition from PsO to PsA.

INTRODUCTION

Psoriasis (PsO) is a chronic, immune-mediated, inflammatory skin disorder affecting 125 million people worldwide1 and more than 7.5 million adults in the USA.2 Environmental triggers, such as infection, have been associated with disease onset and exacerbation.3 4 There is mounting evidence that microbes contribute to PsO pathogenesis, both in humans5–9 and in animal models.10 This notion is supported by recent studies, which describe the cutaneous microbiome in psoriatic patients using culture-independent, high-throughput sequencing techniques. They have reported shifts in microbial diversity and taxonomic abundance when comparing healthy, psoriatic non-lesional (NL) (unaffected; NL) and psoriatic lesional (L) skin.11–16

Although a few studies have specifically looked at the microbial features of psoriatic NL skin, findings conflict with one another.17 Furthermore, the cutaneous microbial analysis of psoriatic arthritis (PsA) as a separate entity is entirely lacking, even though close to a third of PsO patients develop synovio-entheseal inflammation and transition to PsA.18 The passage from PsO to PsA may, in part, be driven by microbial triggers, which deserves further investigation. We set out to characterise the cutaneous microbiota of healthy, PsO and PsA individuals, comparing healthy, psoriatic NL and psoriatic L skin to identify differences in microbial composition and microbe–microbe interactions across the entire psoriatic disease spectrum.

METHODS

Study participants

Psoriatic patients were recruited from dermatology and rheumatology clinics at the NYU Grossman School of Medicine/ NYU Langone Health (NYU). In addition to biosampling, skin and musculoskeletal exams were performed and/or reviewed from medical records. Skin was evaluated for the presence of PsO, the severity of which was quantified with percentage (%) of body surface area (BSA) affected, and categorised as: mild (<3%), moderate (3%–10%) or severe (>10%). When applicable, musculoskeletal evaluations focused on the classification of PsA, tender joint counts (TJC) and swollen joint counts (SJC). In addition to psoriatic patients, non-arthritic healthy individ uals were identified within NYU and enrolled as controls.

Inclusion and exclusion criteria

All participants were 18 years of age or older. Participants in the PsA group met Classification Criteria for Psoriatic Arthritis19 and had active plaque PsO of the skin. Participants in the PsO group were diagnosed with plaque PsO and had no evidence of arthritis, enthesitis or dactylitis attributable to PsA. Healthy controls were age, sex and race/ethnicity-matched individuals with no personal history of PsO or PsA. Participants in the PsO and PsA groups were naïve to DMARDs and biological agents or had an extensive wash-out period (see table 1 and online supplemental material 1) prior to biosampling.

Table 1.

Baseline demographic and clinical characteristics of the study population

Healthy (n=20) PsO (n=23) PsA (n=31) P value
Age, mean±SD (range) 41.0±9.3 (24–57) 47.3±12.6 (27–69) 40.9±13.1 (21–66) 0.113
Sex 0.465
 Female (%) 9 (45.0) 12 (52.2) 11 (35.5)
 Male (%) 11 (55.0) 11 (47.8) 20 (64.5)
Race 0.541*
 Caucasian 15 (75.0) 19 (82.6) 27 (87.1)
 Non-Caucasian 5 (25.0) 4 (17.4) 4 (12.9)
Ethnicity 0.563*
 Hispanic 4 (20.0) 4 (17.4) 9 (29.0)
 Non-Hispanic 16 (80.0) 19 (82.6) 22 (71.0)
BSA severity 0.678*
 Mild (<3% BSA) (%) N/A 4 (17.4) 8 (25.8)
 Moderate (3%–10% BSA) (%) N/A 10 (43.5) 9 (29.0)
 Severe (>10% BSA) (%) N/A 3 (13.0) 6 (19.4)
 Not recorded (%) N/A 6 (26.1) 8 (25.8)
TJC, median (range) N/A N/A 5.5 (0–20)† N/A
SJC, median (range) N/A N/A 2 (0–17)† N/A
Dactilytis N/A N/A 11 (35.5) N/A
Enthesitis N/A N/A 17 (54.9) N/A
Treatment for lesions 0.431
 Topical steroids N/A 20 25
 UV N/A 7 6
 Intralesional steroids N/A 1 2
Open in a new tab

Statistical significance calculated using the χ2 test for dichotomous variables and the one-way ANOVA for continuous variables.

*

One or more comparison groups has<5 entries, an assumption that is used in the χ2 test.

†

One subject did not have joint counts recorded.

ANOVA, analysis of variance; BSA, body surface area; N/A, not available; PsA, psoriatic arthritis; PsO, psoriasis; SJC, swollen joint count; TJC, tender joint count; UV, ultraviolet.

Participants were excluded from the study if they met any of the following exclusion criteria: antibiotic use in the last 3 months, active malignancy or gastrointestinal tract surgery leaving permanent residua (eg, gastrectomy, bariatric surgery and colectomy).

Sample collection

Skin swabs were obtained from the upper and/or lower extremities as previously described.20 In participants with active skin PsO (PsO and PsA groups), samples were collected from a site of L skin and a matching contralateral site of NL unaffected skin. In healthy controls, samples were collected from the elbow/forearm (upper extremity) and the knee/shin (lower extremity).

DNA extraction and 16S rRNA gene sequencing

DNA was extracted from skin swabs using the MoBio Power-Soil DNA Isolation kit (MoBio Laboratories, Carlsbad, California, USA), following a previously described protocol.21 16S rRNA bacterial gene sequencing of the V4 hypervariable region was performed with the 515 F-806R forward-barcoded reverse complement primers22–24 on the Illumina MiSeq platform (150 bp read length, paired-end protocol), as previously described.25 Sequencing was performed at the NYU Langone Health Genome Technology Centre.

Upstream sequence processing

To obtain better coverage, only forward sequences were analysed. Upstream sequence processing was performed with Quantitative Insights into Microbial Ecology 2 (QIIME2) V.2020.11.26 Sequences underwent demultiplexing, denoising with DADA2,27 and quality filtering of amplicon sequence variants (ASVs) against the SILVA reference database.28–30 A phylogenetic tree was constructed, and taxonomy assigned using a naïve Bayes classifier against the SILVA reference database.28–30 Finally, ASVs assigned to chloroplast, mitochondrial, or eukaryotic features, and those that could not be classified below the kingdom level were removed.

Downstream bioinformatic analysis

Microbiome bioinformatic and statistical analyses were performed with QIIME2 V.2020.1126 and R V.4.0.4,31 using the following R libraries: phyloseq,32 NetCoMi,33 igraph,34 DESeq2,35 qiime2R,36 tidyverse,37 dplyr,38 reshpae2,39 FSA,40 ggplot2,41 ggpubr42 and ggrepel.43 P values <0.05 were considered significant. Benjamini-Hochberg false discovery rate (FDR) adjustments44 were applied to pairwise comparisons that involved multiple hypothesis testing. Adjusted p values <0.05 or <0.10 were considered significant depending on the analysis.

We estimated Shannon alpha-diversity, which measures the diversity within a sample; the number of observed features, which measures the richness within a sample; and unweighted UniFrac beta diversity,45 which measures the dissimilarity between sample pairs. Based on the UniFrac distance matrix, we also performed principal coordinate analysis (PCoA), a dimensionality reduction technique to visualise similarity of samples. Kruskal-Wallis and permutational multivariate analysis of variance (PERMANOVA) with 999 permutations were used to calculate the respective statistics. The average relative abundance of features between comparison groups of interest was visualised at the order rank. In addition, the potential influence of demographic and clinical data on microbiome diversity were studied. Continuous variables were analysed using Spearman correlations, and categorical variables were analysed using either Wilcoxon test or Kruskal-Wallis, depending on the number of groups being compared.

Microbe-microbe networks were computed to learn associations among features in healthy, NL (PsO and PsA) and L (PsO and PsA) skin using the SPRING method46 within NetCoMi.47 Networks were constructed from an identical set of 222 features at the genus rank, and visualised with the igraph library34 in R. In each network, nodes represent unique features, hubs represent features with a high number of associations, and edges (connections) between nodes represent positive or negative associations between features. Edge lengths are inversely proportional to the association strength. Pairs of networks were compared using the Jaccard similarity index with different measures (degree, betweenness centrality, closeness centrality, eigenvector centrality) among sets of the most central nodes. The Jaccard index ranges from 0 to 1, with higher scores indicating larger similarity.

DESeq235 differential expression analysis was applied to identify differentiating features between PsO and PsA NL skin and visualised in a volcano plot. Statistically significant features (p<0.05) were also compared using average relative abundance and significance of the difference tested with the Mann-Whitney test.

Additional details are described in online supplemental material 1.

RESULTS

Baseline characteristics

Baseline characteristics of the healthy, PsO and PsA cohorts are shown in table 1. A total of 74 participants (20 healthy, 23 PsO, and 31 PsA) and 148 samples were analysed. Most participants with disease were naïve to biological (94.4%) or synthetic disease-modifying antirheumatic drugs (DMARDs; 81.5%); those who previously took systemic agents had wash-out periods of>8 months (range 8–240 months) prior to biosampling. There were no significant differences in age, sex, race and ethnicity across the three groups, although the PsA cohort was predominantly male. Phenotypically, there were no significant differences in PsO %BSA between the PsO and PsA cohorts. In the PsA group, the median TJC was 5.5 (range 0–20), and the median SJC was 2 (range 0–17).

Cutaneous bacterial dysbiosis is present in PsO and PsA

We compared the cutaneous microbiota of upper and lower extremity in NL and L skin from PsO and PsA patients as well as healthy skin from control individuals. We found no significant differences in the alpha diversity of upper and lower extremities (p=0.57, Mann-Whitney), and the unweighted UniFrac distances between samples from the same extremity and from different extremities were not significantly different (p=0.82, Mann-Whitney). We therefore performed additional analyses on data combining both sites. The Shannon alpha diversity index was significantly higher in healthy skin compared with psoriatic (PsO and PsA) L skin (figure 1A, p<0.05, Kruskal-Wallis with Benjamini-Hochberg correction; online supplemental table S1). Shannon diversity was also significantly higher in healthy skin compared with psoriatic NL skin (figure 1A, p<0.05, Kruskal-Wallis with Benjamini-Hochberg correction; online supple mental table S1). Similarly, bacterial richness, measured as the number of observed features was higher in healthy skin compared with psoriatic (PsO and PsA) L skin (figure 1B, p<0.05; online supplemental table S1) and compared with psoriatic NL skin, although differences were only significant between healthy and PsA skin (figure 1B, p<0.05; online supplemental table S1). No significant differences were observed in alpha diversity estimates between PsO and PsA skin, both L and NL (figure 1A,B). Furthermore, PCoA showed that healthy skin clustered separately from L and NL psoriatic skin (figure 1B, p<0.05, PERMANOVA; online supplemental table S2). Average differences were also observed in the relative abundance of specific features at the order rank. For instance, there was a lower overall abundance of Staphylococcales in healthy samples compared with psoriatic samples, a lower abundance of Corynebacteriales in PsO samples compared with healthy and PsA samples, and a lower abundance of pseudomondales in PsA samples compared with healthy and PsO samples (figure 1C).

Figure 1.

Figure 1

Open in a new tab

Bacterial composition in healthy, psoriasis (PSO) and psoriatic arthritis (PSA) individuals. (A) Healthy samples demonstrate significantly higher Shannon alpha diversity compared with PSO and PSA non-lesional (NL) and lesional (L) samples (p<0.05, Kruskal-Wallis with Benjamini-Hochberg correction for pairwise comparisons). There are no significant differences between PSO and PSA samples. Only statistically significant results are shown. (B) Healthy samples demonstrate significantly higher richness, as measured by the number of observed features, compared with PSO and PSA NL and lesional (L) samples (p<0.05). There are no significant differences between PSO and PSA samples. (C) Principal coordinate analysis plot of unweighted UniFrac beta diversity shows healthy samples clustering separately from nl and L PSO and PSA samples (p<0.05, PERMANOVA with correction for pairwise comparisons). There are no significant differences between PSO and PSA samples. (C) Barplot of feature relative abundance at the order RANK demonstrates average differences between healthy and psoriatic samples. Relative abundance is shown on a scale of 0%–100%. Legend Lists only the top taxa. PERMANOVA, permutational multivariate analysis of variance.

Given the overall differences between healthy and psoriatic microbial communities, we next performed analysis to identify bacterial commensals associated with each of the groups in NL skin (online supplemental figure S1). We found significantly higher levels of Kocuria (p=0.011) and Cutibacterium (p=0.016) in healthy subjects, while Corynebacterium and Staphylococcus were higher in psoriatic NL skin, although not significantly so. Comparing healthy and psoriatic L skin (online supplemental figure S2), we again found higher levels of Kocuria (p=0.021), Cutibacterium (p=0.0017) in healthy, while Staphylococcus was numerically higher in psoriatic L skin (p=ns).

In addition, we tested the relationship between demographic and clinical data on both Shannon diversity and the observed number of features (online supplemental table S3). Among all variables, we only observed significantly higher Shannon diversity in L skin from PsA participants with clinically evident enthesitis compared with that from PsA patients without enthesitis (Wilcoxon test, p=0.033).

Bacterial associations in psoriatic and healthy skin

Next, we constructed bacterial networks to identify associations in psoriatic NL and L skin compared with healthy skin (figure 2; online supplemental tables S4 and S5). Each of the three networks was learnt from features at the genus rank. The network representing healthy skin was denser, had a higher clustering coefficient and a higher proportion of negative associations (lower positive edge percentage), as well as a lower average path length compared with networks representing psoriatic NL and psoriatic L skin (online supplemental table S4). Moreover, hubs in the NL and L networks (ie, nodes with a higher proportion of associations) were predominantly from the Firmicutes phylum, whereas hubs found in the healthy network were predominantly from the Proteobacteria phylum.

Figure 2.

Figure 2

Open in a new tab

Networks representing bacterial-bacterial associations in healthy, psoriatic non-lesional (NL) and psoriatic lesional (L) skin. Networks were generated using the SPRING method within NetCoMi from an identical set of 222 features. Nodes represent features at the genus rank and are coloured by phylum. nodes are sized by the centred log-ratio of each taxon. The top 25 nodes in each network are labelled, with hubs indicated in blue and taxa of interest indicated in green. Edges represent bacterial associations, with the edge length inversely proportional to the strength of the association. Grey edges represent positive associations and red negative associations. Pairs of networks were compared using the Jaccard similarity of the computed betweenness centrality of the most central nodes, represented by dashed lines. The Jaccard index ranges from 0 to 1; the higher the score, the higher the similarity. The same force-directed layout was used for NL and L networks.

We next compared pairs of networks using the Jaccard similarity of betweenness centrality measure among sets of the most central nodes. The betweenness centrality measure represents the fraction of times a node lies on the shortest path between all other nodes. We observed that the psoriatic NL network more closely resembled the psoriatic L network than the healthy network (figure 2; online supplemental table S5). We noted a similar pattern when using measures of degree and closeness centrality to compare network nodes (online supplemental table S6). This finding suggests that the bacterial associations and community dynamics in psoriatic NL skin are closer to that of psoriatic L skin than to those of healthy skin.

Microbial differences in PsO and PsA

Having observed commonalities between psoriatic L and NL skin, we sought to identify differences between PsO and PsA samples, particularly in NL skin since structural changes and barrier disruption in L skin are expected to alter bacterial community composition. Using DESeq2 analysis,35 we identified several candidates at the genus rank, which were differentially abundant in PsO NL compared with PsA NL skin (figure 3A; online supplemental table S6). Features meeting statistical significance without correction (p<0.05) are shown in blue, while those meeting statistical significance with correction for multiple hypothesis testing (q<0.10) are shown in red. Corynebacterium, Cloacibacterium, Atopobium and Megasphaera demonstrated significant differences in relative abundance (figure 3B, p<0.05, Mann-Whitney; online supplemental table S7), with the first two found to be enriched in the PsA cohort, while Atopobium and Megasphaera were enriched in the PsO cohort.

Figure 3.

Figure 3

Open in a new tab

DESeq differential abundance analysis comparing psoriasis (PsO) and psoriatic arthritis (PsA) non-lesional (NL) skin. (A) Volcano plot of features at the genus rank that differentiate PsO NL skin from PsA NL skin. Features more abundant in PsA are shown on the right, while those more abundant in PsO are shown on the left. Blue points represent statistically significant features (p<0.05). Red points represent features that also meet significance with correction for multiple hypothesis testing (q<0.10). Point size is proportional to the mean relative abundance of each feature across all samples that were compared. One outlier point, Tumebacillus, is not shown. (B) Boxplots demonstrating significant differences in the average relative abundance between PsO NL and PsA NL skin. Statistical significance was calculated with the Mann-Whitney U test.

DISCUSSION

The cutaneous microbiome is a topic of scientific interest in psoriatic disease, and several prior studies have described microbial alterations in psoriatic compared with healthy skin. Intriguingly, despite the relatively small cohort size, these investigations converge on a handful of differentiating phyla—Firmicutes, Actinobacteria and Proteobacteria However, no previous studies looked at the cutaneous microbiome in PsA as a separate disease entity.48 Ours is the first study to do so, which simultaneously characterises the cutaneous microbiota across the entire psoriatic disease spectrum.

In this study, we found that the skin microbiota of patients with PsO and PsA is less diverse compared with that of healthy controls, even in the absence of clinically apparent lesions. Similar results were observed by Alekseyenko et al, where specimens from L skin had significantly lower Shannon diversity compared with NL and healthy specimens at the phylum, class, order, family and genus ranks.13 However, our results are in direct contrast to those of Chang et al, where alpha diversity was higher in psoriatic upper (but not lower) extremity L and NL skin compared with healthy skin.49 However, and unlike our study, the L and NL skin specimens were not all matched. Other studies showed no differences in alpha diversity between healthy, NL and L elbow skin, though measurements were made using either the Chao1 index14 or the Gini-Simpson index and using shotgun metagenomics.16 Thus, alpha diversity results appear to vary widely across studies, which may be related to differences in sample size, location of L and NL skin being sampled, the type of sequencing performed, and the diversity measurements used to calculate differences. We also observed increased Shannon diversity in L samples from PsA participants who presented with enthesitis compared with those who did not. Only two previous studies have found gut microbiome changes in enthesitis-related arthritis.50 51 The differences between cohorts (infant and juvenile patients vs our cohort of adult patients), type of samples (faeces vs skin) and the fact that we only observed an effect in Shannon diversity, makes it difficult to draw more robust conclusions. Future studies will be required to evaluate more comprehensively the potential bidirectional interaction between enthesitis and the skin microbiome composition in PsA patients.

We identified several features that differentiate healthy and psoriatic skin. Comparing healthy and psoriatic NL skin, as well as healthy and psoriatic L skin, we found that genera such as Kocuria and Cutibacterium were enriched in healthy skin, while Staphylococcus was enriched in psoriatic skin, both L and NL. Our data is in agreement with that of Alekseyenko et al, who observed that the relative abundance of Staphylococcus was higher in psoriatic lesions than in healthy skin.13 It is also supported by Langan, et al, who showed a higher abundance of Staphylococcus and Prevotella in L skin.14 Still, others have proposed alternate taxa that account for differences between healthy and NL or L psoriatic skin.12 49 Interestingly, and although Tett et al also observed a higher prevalence of Staphylococcus in L skin, compared with NL, they showed that there is no single bacterial feature associated with psoriatic disease at the species rank, but rather a strain heterogeneity colonisation and functional variability in psoriatic skin, suggesting niche-specific adaptations in PsO.16

On the other hand, Cutibacterium (formerly Propionibacterium) is one of the most abundant bacteria in human skin.52 Although P. acnes has been classically associated with acne vulgaris, recent studies have shown that it is also a prevalent commensal in healthy skin, and that only specific pathogenic strains of P. acnes could be involved in acne vulgaris pathogenesis.49 53 54 In agreement with our results, other studies comparing healthy and L and NL psoriatic skin have also reported higher abundances of Cutibacterium in healthy skin,11 14 49 suggesting a protective role in skin physiology.

Very few investigations have described cutaneous microbe–microbe interactions in psoriatic disease. Fyhrquiest demonstrated that psoriatic disease is not associated with a single dominating species, but rather co-occurring communities of microbes.55 Stehlikova found correlations between Kocuria, Lactobacillus and Streptococcus with Saccharomyces in psoriatic skin at the elbow, which were not observed in healthy skin.56 Our study also explored microbe–microbe associations in healthy, psoriatic NL and psoriatic L skin using bacterial association networks. We compared taxa across all three phenotypes and found that the healthy network had a higher proportion of negative associations and fewer positive interactions, indicating mutual exclusive connections between taxa. These positive interactions can destabilise microbial communities by generating positive feedback loops where one species decreasing in abundance can lower the abundance of others that benefit from cooperative interactions.57

Unexpectedly, we discovered that the associations in psoriatic NL skin have more similarity to psoriatic L skin than to healthy skin, suggesting that even in the absence of active inflammatory plaques, a state of microbial dysbiosis is already present in the skin of patients with psoriatic disease. This finding could provide a rationale for the Koebner phenomenon, which occurs when psoriatic lesions in healthy-appearing skin emerge following an injury or trauma in patients with PsO.58 Our data suggest a potential mechanism whereby disruptions in microbial interactions of healthy-appearing skin in psoriatic patients serve as a primer for the development of lesions in the presence of a stressor (ie, microtrauma).

In psoriatic NL skin, we found features that differentiate PsO from PsA. Of particular interest is Corynebacterium, a common cutaneous commensal, which was significantly more abundant in PsA compared with PsO. Importantly, Langan found that Corynebacterium in psoriatic L skin correlated with PASI severity,14 while Quan et al showed that Corynebacterium abundance correlated with severity of local lesions, as well as skin capacitance.15 This suggests that Corynebacterium might be linked to skin disease severity and potentially to progression from cutaneous disease to synovio-entheseal inflammation (ie, PsA). Additional prospective studies are needed to confirm this observation and test this hypothesis.

One important application of these data is the potential development of therapeutic options for the treatment of psoriatic disease and/or the prevention of PsA. For instance, in acne, topical probiotics have been shown to improve skin erythema, repair the skin barrier and reduce the size of lesions.59 In atopic dermatitis, Staphylococcus aureus dominates microbial communities characterised by a relative reduction of commensals.60 Topical probiotics have been shown to decrease the concentration of S. aureus, increase the level of skin ceramides, and improve erythema, scaling and pruritus.61–65 Expanding these studies to psoriatic disease might help identify microbes that can contribute to disease pathogenesis and/or protect from disease progression.

We note several limitations in our study. While our sample size is moderate, the number of participants is notably higher than in previous publications.11 12 14 16 49 We combined samples from upper and lower extremities given the relative similarity of their microenvironments and minimal differences in diversity measurements. We did not explore the cutaneous microbiome of the scalp, a common site for psoriatic lesions that has been associated with progression from PsO to PsA.66 Our study was limited to 16S rRNA gene sequencing, which lacks functional information that can be obtained with shotgun metagenomics. While initial studies using shotgun metagenomics were partially hindered by the low proportion of sequenced data that could be mapped to reference genomes, more recent work by the Skin Microbial Genome Collection has enabled the classification of the majority of skin metagenomic sequences, providing a more comprehensive view of skin microbial diversity.67 In addition, taxonomic sensitivity and specificity depends on the variable region of the 16S gene being targeted.68 Here, we sequenced the V4 region, which may present some limitations in annotating particular species in the skin, but is widely used in amplicon-based microbiome studies, thus allowing comparison of results to a broad literature of prior studies and across body sites,69 most importantly the gut. In addition, our analysis was focused primarily on taxonomic assignments at the genus level and above, thus avoiding biases that may occur in species-level identification. We lack data on fungi, which have also been shown to play a role in previous investigations.56 70 Further, we did not compare different protocols for sample collection, processing and DNA extraction, which can impact results in microbiome studies.71 Although our cohorts were recruited through multiple clinics in a large metropolitan academic centre, they were restricted to a single institution and using a unified protocol for collection, processing and analysis, in order to reduce potential biases. A major strength of our study is the fact that we excluded individuals with recent use of antibiotics and systemic medications (oral synthetic drugs and/or biological agents).

In summary, our study addresses an important gap in the research and literature of psoriatic disease by exploring the cutaneous microbiome across the entire psoriatic spectrum, including healthy subjects, PsO and PsA patients. We found differences in diversity, bacterial composition, and microbe–microbe interactions between healthy and psoriatic skin, both L and NL. We discovered that bacterial associations in NL psoriatic skin are more similar to psoriatic L skin than to healthy skin, which has potential clinical implications for the origin of the Koebner phenomenon. We identified Corynebacterium as a potential biomarker for distinguishing PsO and PsA, particularly in NL skin. Future directions for this research will include the validation of Corynebacterium as a potential biomarker using low-cost approaches (eg, qPCR); studying additional body sites commonly affected by PsO, (ie, scalp) as well as other microbiome niches (ie, gut); investigating fungal and functional data using 18S/ITS and shotgun sequencing; studying microbiome features involved in the transition from PsO to PsA as well as in those patients with PsA sine PsO; and characterising ways in which modulation of the cutaneous bacterial composition may affect inflammatory outcomes. These complementary approaches could provide new avenues for differentiating disease phenotypes with the ultimate goal of predicting (and possibly preventing) the progression from PsO to PsA through microbial-based strategies.

Supplementary Material

supplemental 1
supplemental 3
supplemental 2

WHAT IS ALREADY KNOWN ON THIS TOPIC

  • ⇒ Prior studies support the contribution of the cutaneous microbiome to psoriatic disease pathogenesis. However, the skin microbial communities in patients with psoriasis (PsO) and psoriatic arthritis (PsA) as separate phenotypes have not been previously analysed.

WHAT THIS STUDY ADDS

  • ⇒ We demonstrate microbiome perturbations in the skin of predominantly systemic drug-naïve PsO and PsA patients, which exhibit lower diversity compared with healthy controls even in the absence of clinical lesions.

  • ⇒ Our study simultaneously characterised the cutaneous microbiota across the entire psoriatic disease spectrum, and identified features that differentiate PsO from PsA.

  • ⇒ Bacterial association networks in psoriatic non-lesional (NL) skin are more similar to psoriatic L than to healthy skin, suggesting an underlying dysbiotic process in the cutaneous surface of patients with psoriatic disease even in the absence of clinically evident lesions.

  • ⇒ The common cutaneous commensal Corynebacterium was enriched in NL PsA, compared with NL PsO, and could serve as a biomarker of disease progression.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

  • ⇒ These findings offer novel insights regarding microbiome variations in psoriatic disease, and show that although PsO and PsA skin microbiomes share common traits, they also exhibit differences in key taxa that might potentially be used as diagnostic biomarkers. Our results could inform clinical decisions for patients with PsO, particularly in those deemed at risk for disease progression to PsA.

Acknowledgements

We would like to thank Rhina Medina, Luz Alvarado, Rochelle Castillo, Soumya Reddy, Gary Solomon, Eileen Lydon, Pamela Rosenthal, as well as the attendings and staff at the NYU Langone Ambulatory Care 23rd Street Clinic, the Bellevue Arthritis Clinic, and the NYU Langone and Bellevue Dermatology Clinics for their efforts in the recruitment of participants for this study. We would also like to thank Leopoldo Segal, Benjamin Wu, and Yonghua Li for their assistance with the initial stages of this project, as well as James Morton for his guidance and advice throughout the data analysis process.

Funding

National Psoriasis Foundation Early Career Grant (PI: Manasson), NIH/NIAMS R01AR074500 (PI: Scher, Co-I: Manasson), NIH/NIAMS T32 AR069515 (PI: Buyon, Trainee: Manasson), Rheumatology Research Foundation Investigator Award (PI: Manasson), Psoriatic Arthritis Center (PI: Scher), Riley Family Foundation (PI: Scher), Beatrice Snyder Foundation (PI: Scher), Cancer Center Support Grant P30CA016087 at the Laura and Isaac Perlmutter Cancer Center (support for NYU Genome Technology Center), National Psoriasis Foundation (PI: Koralov and Scher), LEO Foundation (PI: Koralov), Colton Center for Autoimmunity Pilot Grant (PI: Koralov and Scher), National Psoriasis Foundation Early Career Grant (PI: Manasson), NIH/ NIAMS R01AR074500 (PI: Scher, Co-I: Manasson), NIH/NIAMS T32 AR069515 (PI: Buyon, Trainee: Manasson), Rheumatology Research Foundation Investigator Award (PI: Manasson), Psoriatic Arthritis Center (PI: Scher), Riley Family Foundation (PI: Scher), Beatrice Snyder Foundation (PI: Scher), Cancer Center Support Grant P30CA016087 at the Laura and Isaac Perlmutter Cancer Center (support for NYU Genome Technology Center), National Psoriasis Foundation (PI: Koralov and Scher), LEO Foundation (PI: Koralov), Colton Center for Autoimmunity Pilot Grant (PI: Koralov and Scher).

Footnotes

Competing interests RHH: received consulting fees from Janssen ALN: served as a consultant for Janssen, UCB, AbbVie and Bristol-Myers Squibb; has an immediate family member who owns shares of stock in J&J, Eli Lilly, AbbVie and Pfizer JUS: served as a consultant for Janssen, Abbvie, Novartis, Sanofi, UCB and BMS.

Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

Patient consent for publication Not applicable.

Ethics approval The study was approved by the Institutional Review Board of the NYU Grossman School of Medicine (#s14– 00487, #S12– 00831). Participants gave informed consent to participate in the study before taking part.

Provenance and peer review Not commissioned; externally peer reviewed.

Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

Data availability statement

Data are available in a public, open access repository. All sequence data will be made publicly available on publication at the NIH Sequence Read Archive (https://www.ncbi.nlm.nih.gov/sra). Additionally, code that was used to perform computational analyses in this manuscript will be made available at https://github.com/scher-lab.

REFERENCES

  • 1.Armstrong AW, Read C. Pathophysiology, clinical presentation, and treatment of psoriasis: a review. JAMA 2020;323:1945–60. [DOI] [PubMed] [Google Scholar]
  • 2.Armstrong AW, Mehta MD, Schupp CW, et al. Psoriasis prevalence in adults in the United States. JAMA Dermatol 2021;157:940–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Boehncke W-H, Schön MP. Psoriasis. Lancet 2015;386:983–94. [DOI] [PubMed] [Google Scholar]
  • 4.Greb JE, Goldminz AM, Elder JT, et al. Psoriasis. Nat Rev Dis Primers 2016;2:16082. [DOI] [PubMed] [Google Scholar]
  • 5.Tervaert WC, Esseveld H. A study of the incidence of haemolytic streptococci in the throat in patients with psoriasis vulgaris, with reference to their role in the pathogenesis of this disease. Dermatologica 1970;140:282–90. [DOI] [PubMed] [Google Scholar]
  • 6.Skov L, Baadsgaard O. Bacterial superantigens and inflammatory skin diseases. Clin Exp Dermatol 2000;25:57–61. [DOI] [PubMed] [Google Scholar]
  • 7.Lande R, Botti E, Jandus C, et al. The antimicrobial peptide LL37 is a T-cell autoantigen in psoriasis. Nat Commun 2014;5:5621. [DOI] [PubMed] [Google Scholar]
  • 8.Tomi NS, Kränke B, Aberer E. Staphylococcal toxins in patients with psoriasis, atopic dermatitis, and erythroderma, and in healthy control subjects. J Am Acad Dermatol 2005;53:67–72. [DOI] [PubMed] [Google Scholar]
  • 9.Rudramurthy SM, Honnavar P, Chakrabarti A, et al. Association of malassezia species with psoriatic lesions. Mycoses 2014;57:483–8. [DOI] [PubMed] [Google Scholar]
  • 10.Zanvit P, Konkel JE, Jiao X, et al. Antibiotics in neonatal life increase murine susceptibility to experimental psoriasis. Nat Commun 2015;6:8424. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Gao Z, Tseng C-hong, Strober BE, et al. Substantial alterations of the cutaneous bacterial biota in psoriatic lesions. PLoS One 2008;3:e2719. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Fahlén A, Engstrand L, Baker BS, et al. Comparison of bacterial microbiota in skin biopsies from normal and psoriatic skin. Arch Dermatol Res 2012;304:15–22. [DOI] [PubMed] [Google Scholar]
  • 13.Alekseyenko AV, Perez-Perez GI, De Souza A, et al. Community differentiation of the cutaneous microbiota in psoriasis. Microbiome 2013;1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Langan EA, Künstner A, Miodovnik M, et al. Combined culture and metagenomic analyses reveal significant shifts in the composition of the cutaneous microbiome in psoriasis. Br J Dermatol 2019;181:1254–64. [DOI] [PubMed] [Google Scholar]
  • 15.Quan C, Chen X-Y, Li X, et al. Psoriatic lesions are characterized by higher bacterial load and imbalance between cutibacterium and corynebacterium. J Am Acad Dermatol 2020;82:955–61. [DOI] [PubMed] [Google Scholar]
  • 16.Tett A, Pasolli E, Farina S, et al. Unexplored diversity and strain-level structure of the skin microbiome associated with psoriasis. NPJ Biofilms Microbiomes 2017;3:14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Yerushalmi M, Elalouf O, Anderson M, et al. The skin microbiome in psoriatic disease: a systematic review and critical appraisal. J Transl Autoimmun 2019;2:100009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Ritchlin CT, Colbert RA, Gladman DD. Psoriatic arthritis. N Engl J Med 2017;376:957–70. [DOI] [PubMed] [Google Scholar]
  • 19.Taylor W, Gladman D, Helliwell P, et al. Classification criteria for psoriatic arthritis: development of new criteria from a large international study. Arthritis Rheum 2006;54:2665–73. [DOI] [PubMed] [Google Scholar]
  • 20.Human Microbiome Project Consortium. A framework for human microbiome research. Nature 2012;486:215–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Scher JU, Joshua V, Artacho A, et al. The lung microbiota in early rheumatoid arthritis and autoimmunity. Microbiome 2016;4:60. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Parada AE, Needham DM, Fuhrman JA. Every base matters: assessing small subunit rRNA primers for marine microbiomes with mock communities, time series and global field samples. Environ Microbiol 2016;18:1403–14. [DOI] [PubMed] [Google Scholar]
  • 23.Apprill A, McNally S, Parsons R, et al. Minor revision to V4 region SSU rRNA 806R gene primer greatly increases detection of SAR11 bacterioplankton. Aquatic Microbial Ecology 2015;75:129–37. [Google Scholar]
  • 24.Thompson LR, Sanders JG, McDonald D, et al. A communal catalogue reveals earth’s multiscale microbial diversity. Nature 2017;551:457–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Manasson J, Shen N, Garcia Ferrer HR, et al. Gut microbiota perturbations in reactive arthritis and postinfectious spondyloarthritis. Arthritis Rheumatol 2018;70:242–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Bolyen E, Rideout JR, Dillon MR, et al. Reproducible, interactive, scalable and extensible microbiome data science using QIIME 2. Nat Biotechnol 2019;37:852–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Callahan BJ, McMurdie PJ, Rosen MJ, et al. DADA2: high-resolution sample inference from illumina amplicon data. Nat Methods 2016;13:581–3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Yilmaz P, Parfrey LW, Yarza P, et al. The SILVA and "all-species living tree project (LTP)" taxonomic frameworks. Nucleic Acids Res 2014;42:D643–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Quast C, Pruesse E, Yilmaz P, et al. The silva ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res 2013;41:D590–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Glöckner FO, Yilmaz P, Quast C, et al. 25 years of serving the community with ribosomal RNA gene reference databases and tools. J Biotechnol 2017;261:169–76. [DOI] [PubMed] [Google Scholar]
  • 31.R Core Team. R: a language and environment for statistical computing; 2021. [Google Scholar]
  • 32.McMurdie PJ, Holmes S. phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data. PLoS One 2013;8:e61217. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Peschel S. NetCoMi: network construction and comparison for microbiome data. R package version 1.0.2; 2021. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Gabor C, Nepusz T. The igraph software package for complex network research. InterJournalComplex Systems 2006;1695. [Google Scholar]
  • 35.Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-Seq data with DESeq2. Genome Biol 2014;15:550. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Bisanz JE. qiime2R: importing QIIME2 artifacts and associated data into R sessions, 2018 [Google Scholar]
  • 37.Wickham H, Averick M, Bryan J, et al. Welcome to the tidyverse. J Open Source Softw 2019;4:1686. [Google Scholar]
  • 38.Wickham H, François R, Henry L. dplyr: a grammar of data manipulation. R package version 1.0.5; 2021. [Google Scholar]
  • 39.Wickham H. Reshaping data with the reshape package. J Stat Softw 2007;21:1–20. [Google Scholar]
  • 40.Ogle DH, Wheeler P, Dinno A. FSA: fisheries stock analysis. R package version 0.8.31, 2020 [Google Scholar]
  • 41.Wichmann HE. ggplot2: elegant graphics for data Analysi, 2016. [Google Scholar]
  • 42.Kassambara A. ggpubr: ’ggplot2’ Based Publication Ready Plots. R package version 0.4.0, 2020 [Google Scholar]
  • 43.Slowikowski K. ggrepel: Automatically Position Non-Overlapping Text Labels with ’ggplot2’. R package version 0.9.1; 2021. [Google Scholar]
  • 44.Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc Series B 1995;57:289–300. [Google Scholar]
  • 45.Lozupone C, Knight R. UniFrac: a new phylogenetic method for comparing microbial communities. Appl Environ Microbiol 2005;71:8228–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Yoon G, Gaynanova I, Müller CL. Microbial networks in spring - semi- parametric rank-based correlation and partial correlation estimation for quantitative microbiome data. Front Genet 2019;10:516. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Peschel S, Müller CL, von Mutius E, et al. NetCoMi: network construction and comparison for microbiome data in R. Brief Bioinform 2021;22. doi: 10.1093/bib/bbaa290. [Epub ahead of print: 20 Jul 2021]. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Le ST, Toussi A, Maverakis N, et al. The cutaneous and intestinal microbiome in psoriatic disease. Clin Immunol 2020;218:108537. [DOI] [PubMed] [Google Scholar]
  • 49.Chang H-W, Yan D, Singh R, et al. Alteration of the cutaneous microbiome in psoriasis and potential role in Th17 polarization. Microbiome 2018;6:154. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Di Paola M, Cavalieri D, Albanese D, et al. Alteration of fecal microbiota profiles in juvenile idiopathic arthritis. associations with HLA-B27 allele and disease status. Front Microbiol 2016;7:1703. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Aggarwal A, Sarangi AN, Gaur P, et al. Gut microbiome in children with enthesitis-related arthritis in a developing country and the effect of probiotic administration. Clin Exp Immunol 2017;187:480–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Grice EA, Kong HH, Conlan S, et al. Topographical and temporal diversity of the human skin microbiome. Science 2009;324:1190–2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Tomida S, Nguyen L, Chiu B-H, et al. Pan-genome and comparative genome analyses of propionibacterium acnes reveal its genomic diversity in the healthy and diseased human skin microbiome. mBio 2013;4:e00003–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Fitz-Gibbon S, Tomida S, Chiu B-H, et al. Propionibacterium acnes strain populations in the human skin microbiome associated with acne. J Invest Dermatol 2013;133:2152–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Fyhrquist N, Muirhead G, Prast-Nielsen S, et al. Microbe-host interplay in atopic dermatitis and psoriasis. Nat Commun 2019;10:4703. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.Stehlikova Z, Kostovcik M, Kostovcikova K, et al. Dysbiosis of skin microbiota in psoriatic patients: co-occurrence of fungal and bacterial communities. Front Microbiol 2019;10:438. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57.Coyte KZ, Schluter J, Foster KR. The ecology of the microbiome: networks, competition, and stability. Science 2015;350:663–6. [DOI] [PubMed] [Google Scholar]
  • 58.Diani M, Cozzi C, Altomare G. Heinrich koebner and his phenomenon. JAMA Dermatol 2016;152:919. [DOI] [PubMed] [Google Scholar]
  • 59.Muizzuddin N, Maher W, Sullivan M, et al. Physiological effect of a probiotic on skin. J Cosmet Sci 2012;63:385–95. [PubMed] [Google Scholar]
  • 60.Paller AS, Kong HH, Seed P, et al. The microbiome in patients with atopic dermatitis. J Allergy Clin Immunol 2019;143:26–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 61.Di Marzio L, Centi C, Cinque B, et al. Effect of the lactic acid bacterium streptococcus thermophilus on stratum corneum ceramide levels and signs and symptoms of atopic dermatitis patients. Exp Dermatol 2003;12:615–20. [DOI] [PubMed] [Google Scholar]
  • 62.Guéniche A, Hennino A, Goujon C, et al. Improvement of atopic dermatitis skin symptoms by Vitreoscilla filiformis bacterial extract. Eur J Dermatol 2006;16:380–4. [PubMed] [Google Scholar]
  • 63.Blanchet-Réthoré S, Bourdès V, Mercenier A, et al. Effect of a lotion containing the heat-treated probiotic strain lactobacillus johnsonii NCC 533 on staphylococcus aureus colonization in atopic dermatitis. Clin Cosmet Investig Dermatol 2017;10:249–57. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 64.Nakatsuji T, Chen TH, Narala S, et al. Antimicrobials from human skin commensal bacteria protect against staphylococcus aureus and are deficient in atopic dermatitis. Sci Transl Med 2017;9. doi: 10.1126/scitranslmed.aah4680. [Epub ahead of print: 22 Feb 2017]. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 65.Knackstedt R, Knackstedt T, Gatherwright J. The role of topical probiotics in skin conditions: a systematic review of animal and human studies and implications for future therapies. Exp Dermatol 2020;29:15–21. [DOI] [PubMed] [Google Scholar]
  • 66.Yan D, Ahn R, Leslie S, et al. Clinical and genetic risk factors associated with psoriatic arthritis among patients with psoriasis. Dermatol Ther 2018;8:593–604. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 67.Saheb Kashaf S, Proctor DM, Deming C, et al. Integrating cultivation and metagenomics for a multi-kingdom view of skin microbiome diversity and functions. Nat Microbiol 2022;7:169–79. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 68.Johnson JS, Spakowicz DJ, Hong B-Y, et al. Evaluation of 16S rRNA gene sequencing for species and strain-level microbiome analysis. Nat Commun 2019;10:5029. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 69.Caporaso JG, Lauber CL, Walters WA, et al. Ultra-high-throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms. Isme J 2012;6:1621–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 70.Takemoto A, Cho O, Morohoshi Y, et al. Molecular characterization of the skin fungal microbiome in patients with psoriasis. J Dermatol 2015;42:166–70. [DOI] [PubMed] [Google Scholar]
  • 71.Sinha R, Abu-Ali G, Vogtmann E, et al. Assessment of variation in microbial community amplicon sequencing by the microbiome quality control (MBQC) project consortium. Nat Biotechnol 2017;35:1077–86. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

supplemental 1
NIHMS1915798-supplement-supplemental_1.pdf (127.2KB, pdf)
supplemental 3
NIHMS1915798-supplement-supplemental_3.pdf (136.6KB, pdf)
supplemental 2
NIHMS1915798-supplement-supplemental_2.pdf (226.9KB, pdf)

Data Availability Statement

Data are available in a public, open access repository. All sequence data will be made publicly available on publication at the NIH Sequence Read Archive (https://www.ncbi.nlm.nih.gov/sra). Additionally, code that was used to perform computational analyses in this manuscript will be made available at https://github.com/scher-lab.

RESOURCES