The composition and functional protein subsystems of the human nasal microbiome in granulomatosis with polyangiitis: a pilot study

Abstract

Background: Ear, nose and throat involvement in granulomatosis with polyangiitis (GPA) is frequently the initial disease manifestation. Previous investigations have observed a higher prevalence of Staphylococcus aureus in patients with GPA, and chronic nasal carriage has been linked with an increased risk of disease relapse. In this cross-sectional study, we investigated changes in the nasal microbiota including a detailed analysis of Staphylococcus spp. by shotgun metagenomics in patients with active and inactive granulomatosis with polyangiitis (GPA). Shotgun metagenomic sequence data were also used to identify protein-encoding genes within the SEED database, and the abundance of proteins then correlated with the presence of bacterial species on an annotated heatmap.

Results: The presence of S. aureus in the nose as assessed by culture was more frequently detected in patients with active GPA (66.7%) compared with inactive GPA (34.1%). Beta diversity analysis of nasal microbiota by bacterial 16S rRNA profiling revealed a different composition between GPA patients and healthy controls (P = 0.039). Beta diversity analysis of shotgun metagenomic sequence data for Staphylococcus spp. revealed a different composition between active GPA patients and healthy controls and disease controls (P = 0.0007 and P = 0.0023, respectively), and between healthy controls and inactive GPA patients and household controls (P = 0.0168 and P = 0.0168, respectively). Patients with active GPA had a higher abundance of S. aureus, mirroring the culture data, while healthy controls had a higher abundance of S. epidermidis. Staphylococcus pseudintermedius, generally assumed to be a pathogen of cats and dogs, showed an abundance of 13% among the Staphylococcus spp. in our cohort. During long-term follow-up of patients with inactive GPA at baseline, a higher S. aureus abundance was not associated with an increased relapse risk. Functional analyses identified ten SEED protein subsystems that differed between the groups. Most significant associations were related to chorismate synthesis and involved in the vitamin B₁₂ pathway.

Conclusion: Our data revealed a distinct dysbiosis of the nasal microbiota in GPA patients compared with disease and healthy controls. Metagenomic sequencing demonstrated that this dysbiosis in active GPA patients is manifested by increased abundance of S. aureus and a depletion of S. epidermidis, further demonstrating the antagonist relationships between these species. SEED functional protein subsystem analysis identified an association between the unique bacterial nasal microbiota clusters seen mainly in GPA patients and an elevated abundance of genes associated with chorismate synthesis and vitamin B₁₂ pathways. Further studies are required to further elucidate the relationship between the biosynthesis genes and the associated bacterial species.

Keywords: ANCA; GPA; Metagenomics; Microbiome; Staphylococcus; Vasculitis; rRNA sequencing.

Fig. 1

Hierarchical clustering and taxonomic annotation of bacterial 16S rRNA marker gene sequenced species. Bacterial 16S sequence data were available from 59 samples including seven active GPA, 31 inactive GPA, two disease controls (EGPA), seven unrelated healthy controls, four healthy household controls, and eight longitudinal samples. a Hierarchical clustering with heatmap presentation was done with R package Heatplus (v 2.20.0, Author: Alexander Ploner). For the heatmap presentation, we removed species with less than 5 % as their maximum relative abundance in five samples which resulted in the inclusion of 34 oligotype species. b A stacked bar chart showing the distribution of the top 14 species (minimum abundance of 1% covering 93.16% of all reads) is placed next to the heatmap. c A stacked bar chart showing the distribution of the next top 17 species (minimum abundance between 0.1% and 1% covering 5.56% of all reads)

Fig. 2

Hierarchical clustering and taxonomic annotation of shotgun-sequenced Staphylococcus species. For the heatmap analysis, we removed species with less than 5% as their maximum relative abundance in five samples, which retained 20 species for easier presentation in the heatmap. The same 20 species were used for the stacked bar chart. Left over black bars represent other species not present in the top 20 species

Figure 3

Differences in nasal Staphylococcus species composition between sample groups. a Differences in shotgun-sequenced nasal Staphylococcus species composition between sample groups were visualised using non-metric multi-dimensional scaling (NMDS plot) and correspondence analysis (CA plot). The significance of the separation between the different sample groups was further assessed by PERMANOVA test (statistical test for bacterial beta diversity). The overall group comparison was statistically different (P = 0.0031). The individual group comparisons revealed statistical differences in beta diversity between aGPA patients and HC (P = 0.0007) and between aGPA patients and disease controls (P = 0.0023). Beta diversity was also statistically different between the HC and inGPA patients (P = 0.0168) and between HC and HHC (P = 0.0168). b Scatter dot plot presentation of statistically associated S. epidermidis and S. aureus. S. epidermidis was found at statistically higher abundance in the HC group compared with aGPA patients. S. aureus was found at statistically higher abundance in aGPA patients compare to DC patients and the HC groups. c The direction of the Spearman’s correlation coefficient value (positive or negative value on the y-axis) determines whether S. epidermidis and S. aureus are either positively or negatively associated with the different sample groups. aGPA, active granulomatosis with polyangiitis (GPA); inGPA, inactive GPA; DC, disease controls (eosinophilic GPA and microscopic polyangiitis); HC, unrelated healthy controls; HHC, healthy household controls; PERMANOVA, permutational multivariate analysis of variance

Fig. 4

Taxonomic annotation of longitudinal case studies of shotgun-sequenced Staphylococcus species. Shotgun-sequenced Staphylococcus species were analysed in 13 longitudinal case studies together with healthy controls. The individual case studies were grouped together with follow-up samples 1 month and 3 months later (where available) and with or without healthy household controls at the time of initial sampling and 1 month later for one case study. The x-axis shows the proportional abundance of the top 25 species with a minimum abundance of 0.1% across the patient samples, which covers 97.85% of all Staphylococcus reads within the longitudinal cohort. The matching species of the healthy controls are presented at the bottom of Additional file 4: Figure S4 for comparisons. aGPA, active granulomatosis with polyangiitis (GPA); inGPA, inactive GPA; HC, unrelated healthy controls, HHC, healthy household controls

Fig. 5

Statistically significant functional SEED annotation pathway. Shotgun sequences were used for the analysis of SEED functional protein subsystems. Ten SEED functional protein subsystems were statistically associated with the four sample groups and are shown in Fig. 5. GPA, granulomatosis with polyangiitis (GPA), DC, disease controls; HC, healthy controls; HHC, healthy household controls

Fig. 6

Correlation between metagenomic species and SEED functional protein subsystems. Most abundant shotgun metagenomic species were correlated with the ten statistically associated SEED functional protein subsystems. The cuth parameter in the dendrogram was set in such a way that it identified five clusters which are colour coded. The cuth parameter sets the height at which to cut through the dendrogram to define groups of similar features/samples. A distance metric was generated with R function “vegist” from the VEGAN package using the “bray” method and Hclust R function from the VEGAN package using the ward. D method was used to cluster the distance matrix. The heatmap was generated with the Heatplus package from R, version 2.26.0