3.1. Dysbiosis of conjunctival microbiome in allergic
conjunctivitis
The study cohort was composed of 39 individuals with AC and 48 healthy
subjects (Table S1). The group of AC comprised 14 patients with PAC, 7
with SAC, and 18 with VKC (Table S2). Table S1 and Table S2 show the age
at sample collection. All the patients were in active stage of AC with
inflamed ocular surface. We performed shotgun metagenomic sequencing and
obtained on average 0.32 million high quality non-human reads for each
individual. A total of 278 species passed our decontamination pipeline
(Methods).
Overall, bacteria accounted for the majority of the conjunctival
microbiota of both healthy and AC individuals (Figure 1A). We observed
the enrichment of eukaryotic virus in a subset of healthy individuals
(Kruskal-Wallis rank-sum test P = 0.0014). The prominent viral
species in these samples was human beta herpesvirus 7, which is
frequently detected and rarely pathogenic in immunocompetent
individuals.22 Malassezia Fungi (M.furfur in particular) were abundant in a fraction of patients
with SAC/PAC (but not VKC) compared with healthy individuals (Figure 1B;P = 0.0095). IgE mediated hypersensitivity to Malasseziaspecies correlates with the clinical severity of atopic dermatitis
(AD).23 Malassezia produces immunogenic
proteins that elicit IgE and thus induce pro-inflammatory cytokines and
auto-reactive T cells, which contributes to AD
pathogenesis.24 Notably, SAC/PAC is dominated by
IgE-mediated reactions in contrast to VKC.
The alpha diversity showed no significant difference between healthy and
AC individuals (Figure 2A), whereas the Bray-Curtis dissimilarities
within AC groups were slightly lower than healthy groups (Figure 2B).
The principal coordinates analysis (PCoA) of the species composition
showed a clear delineation between healthy and AC participants (Figure
2C), suggesting that dysbiosis of the conjunctival microbiome is
associated with AC.
To identify the species that accounts for the dysbiosis, we performed
LEfSe analysis on the species profiles and the species with LDA effect
size > 4 were displayed in Figure 2D. Numerous species were
overabundant on the ocular surface of AC patients compared to healthy
subjects (Figure 2E), such as members of Oxalobacteraceae includingJanthinobacterium sp . Marseille (P =
1.6×10-17), Herminiimonas arsenitoxidans(P = 2.3×10-14), and Herminiimonasarsenicoxydans (P = 5.2×10-15).
Interestingly, family Oxalobacteraceae was prevalently detected in the
conjunctival microbiota of lens wearers25 and contact
lens wearing is in turn associated with ocular
allergy.26 Rothia was also enriched in samples
from participants with AC, including Rothia aeria (P =
1.8×10-4) and Rothia dentocariosa (P =
0.0010). In addition, we detected the enrichment of Moraxella
catarrhalis in AC (P = 0.0033). Moraxella catarrhalis in
the upper airways is linked with the development or exacerbation of
allergic airway inflammation and IL-17 and TNF-α are involved in this
process.27 Colonization of M. catarrhalis in
the airways leads to a low-grade systemic inflammation that is
associated with established asthma.28 This implies
that common microbial mechanisms may underlie both ocular allergy and
allergic diseases prevalent at other sites.
We next searched for functional differences in the conjunctival
microbiome between healthy and AC individuals. Most of the gene families
identified using LEfSe were enriched in the metagenome from the AC
patients (Figure 3A). For instance, we detected the enrichment of the
gene families related to amino acid metabolism such as aspartate kinase
(Kruskal-Wallis rank-sum test P = 0.0058), tryptophan synthase
(P = 0.012), and glutamine synthetase (P = 0.026). In
addition, the gene families encoding thioredoxin (P = 0.012) and
thioredoxin reductase (P = 0.026) were more abundant in AC
patients compared to healthy individuals.
Exposure to antibiotics has been associated with many allergic
diseases.29 Therefore, we further examined the
prevalence of antibiotic resistance genes in the conjunctival metagenome
of AC patients (Figure 3B). In particular, we observed that the genes
resistant to tetracycline is more prevalent in AC patients than healthy
individuals, including tetA (Fisher’s exact test P =
0.045), tetB (P = 0.0043), optrA (P =
0.038), vmlR (P = 0.0026), and evgS (P =
0.042). Glycopeptide antibiotic resistance genes such as vanTG(P = 0.016) and vanG (P = 0.037) were detected
exclusively in AC samples. Fluoroquinolone antibiotic resistance genes
were also more prevalent in the samples from AC patients compared to
healthy subjects, including acrF (P = 0.037), efrB(P = 0.031), evgS (P = 0.042), lfrA(P = 0.037), and patA (P = 0.038). This warrants
further studies into the relationships between antibiotic use and the
risk of ocular allergy.