Toll-Interacting Protein and Altered Lung Microbiota in Idiopathic Pulmonary Fibrosis

To the Editor:

Idiopathic pulmonary fibrosis (IPF) is a progressive chronic disease of the respiratory tract that has no known cure and results in significant global morbidity and mortality. Clinical observations have demonstrated an association among lung microbiota, alveolar inflammation, disease progression, and death in patients with IPF (1, 2). TLRs (Toll-like receptors) are highly conserved canonical innate immune receptors that recognize a diverse repertoire of molecular patterns derived from microbes (3). TLR signaling may regulate interactions with lung microbiota in transgenic animal models (4). TOLLIP (Toll-interacting protein) is a complex protein, widely expressed in the respiratory tract and a key regulator of TLR signaling. The rs5743890 SNP of the TOLLIP gene is associated with increased mortality in IPF (5, 6). However, the role of this TOLLIP SNP in regulating lung microbiota is unknown.

Lung dysbiosis and the rs5743890 TOLLIP SNP are independently associated with mortality in IPF. However, it is not known whether an association exists between lung dysbiosis and the rs5743890 polymorphism. We hypothesized that patients with IPF carrying the rs5743890 minor allele would demonstrate altered lung microbiota compared with wild-type subjects. To determine the association between the TOLLIP rs5743890 SNP and respiratory tract microbiota, we examined data from patients with IPF enrolled in the COMET (Correlating Outcomes with Biochemical Markers to Estimate Time-Progression in Idiopathic Pulmonary Fibrosis) study (6). Lung microbial communities were characterized using 16S ribosomal RNA amplicon sequencing from BAL fluid, and TOLLIP rs5743890 genotyping was undertaken as previously reported (5, 6). Patients were included in the analysis if genotype status for the rs5743890 TOLLIP polymorphism and BAL 16S data were available. Genotype status was stratified as wild-type major allele carriers or minor allele carriers. BAL bacterial burden was measured as previously reported (1). To study differences in microbial community composition (β-diversity), we used principal-component analysis and negative binomial generalized linear models (GLMs) (mvabund R package [https://www.r-project.org/]) of multivariate abundance data. We also examined community composition using the Bray Curtis dissimilarity score. We determined the α-diversity of microbial communities using the Shannon diversity indices. This is a secondary analysis of previously published COMET data (1, 6). Sequencing data are available via the National Center for Biotechnology Information Sequence Read Archive (accession numbers PRJNA515255 and PRJNA515279).

To test our hypothesis, we stratified our cohort by homozygote major and heterozygote minor allele carriage. A total of 60 patients were included, and the clinical characteristics and demographics of subjects are reported in Table 1. There were no significant differences in clinical characteristics between cohorts. The presence of radiographic honeycombing may be associated with altered community composition of lung microbiota (7); however, we report no differences in the presence of honeycombing between our cohorts. Initially we examined the community composition using principal-component analysis. We found a significant difference in community composition, with homozygote major and heterozygote minor cohorts clustering separately (P = 0.02, permutational multivariate ANOVA) (Figure 1A). We also used a negative binomial GLM and report significant community differences between genotype cohorts (overall P = 0.005). We next used rank abundance analysis to compare lung microbiota across homozygote major (Figure 1B) and heterozygote minor (Figure 1C) cohorts, noting an absence of operational taxonomic unit (OTU) 0878 Tropheryma in the heterozygote minor allele cohort. The GLM identified several OTUs that differed in abundance between genotype cohorts, including OTU0878 Tropheryma (P = 0.01), OTU1037 Lachnospiraceae unclassified (P = 0.005), OTU1192 Cryomorphaceae unclassified (P = 0.006), and OTU0870 Flavobacterium (P = 0.01); however, after controlling conservatively for multiple comparisons, none of the observed associations in the model remained statistically significant. We then used a complementary technique to study community composition. We calculated within-group similarities in genotype cohorts using the Bray-Curtis dissimilarity score (whereby a score of 1 means no shared OTUs and a score of 0 means that communities are identical). We found that minor allele carriers demonstrated less within-group similarity compared with the respiratory tract communities of homozygote major allele carriers (P < 0.05) (Figure 1D). We found no significant difference in lung microbiota burden (P = 0.45) measured by droplet digital PCR between genotype cohorts or α-diversity measured by the Shannon diversity index (P = 0.32) (data not shown). We did not find any associations between TOLLIP rs5743894 and rs111521887 SNPs and altered lung microbiota (data not shown) in this cohort. We concluded that the TOLLIP rs5743890 SNP promotes alterations in respiratory tract community composition in patients with IPF.

Table 1.

Clinical Characteristics and Demographics of Study Subjects

	A/A	A/G
n (%)	41 (68)	19 (32)
Age, y, mean (SD)	64.27 (7.4)	66.46 (6.5)
Male, n (%)	29 (70.7)	14 (73.7)
FVC, predicted %, mean (SD)*	72.09 (18.3)	66.09 (16.0)
DlCO, predicted %, mean (SD)*	46.04 (15.2)	39.42 (10.2)
Smoking status, n
Nonsmokers	12	6
Ever-smokers	29	13
Honeycombing, n (%)a
Yes	22 (55.0)	11 (61.1)
No	18 (45.0)	7 (38.9)

Definition of abbreviations: A/A = major allele carriers; A/G = minor allele carriers.

^*Missing data: for FVC, one value was missing in each group; for Dl_CO, three values were missing in the A/A group; and for honeycombing, one value was missing in each group.

Figure 1.

The functional rs5743890 polymorphism is associated with respiratory tract dysbiosis in patients with IPF. Sixty patients with IPF enrolled in the COMET (Correlating Outcomes with Biochemical Markers to Estimate Time-Progression in Idiopathic Pulmonary Fibrosis) study were genotyped for the TOLLIP (Toll-interacting protein) rs5743890 polymorphism, and BAL fluid was analyzed using 16S ribosomal RNA amplicon sequencing to study lung microbiota. (A) Principal-component analysis of Hellinger-transformed 16S lung microbiota data demonstrates significant separation of major allele carriers (A/A) and minor allele carriers (A/G). (B and C) Rank abundance featuring the 15 most abundant OTUs in A/A and A/G. (D) Bray-Curtis dissimilarity scores in A/A and A/G showing increased within-group dissimilarity in A/G (*P < 0.05, PERMANOVA). A/A = major allele carriers; A/G = minor allele carriers; IPF = idiopathic pulmonary fibrosis; OTU = operational taxonomic unit; PC = principal component; PERMANOVA = permutational multivariate ANOVA.

The interaction between lung microbiota and host innate immunity is incompletely understood. Studies have established the impact lung microbiota may have on local immunity, including defining immunomodulatory metabolite production (8). TOLLIP is proposed to mediate protection against lung fibrosis by attenuating mitochondrial reactive oxygen species generation and upregulating autophagy (9). Previous studies have correlated lung microbiota with TOLLIP signaling within peripheral blood mononuclear cells in patients with IPF, suggesting an association with the systemic host response (10). Our data demonstrate that the TOLLIP rs5743890 SNP may contribute to altered lung microbiota with important implications for homeostasis at the alveolar epithelium and distal airway. The TOLLIP SNPs rs111251887 and rs5743894 are associated with increased IPF susceptibility and significantly reduced respiratory tract expression but are not associated with an increased risk of mortality in IPF. There is no association between altered lung microbiota and rs111521887 and rs5743894 SNPs in this cohort. In contrast, rs5743890 is associated with altered lung microbiota and previously reported associations with enhanced mortality in IPF. It is possible that the rs5743890 SNP contributes to IPF progression by the promotion of respiratory tract dysbiosis through a yet undefined mechanism independent of TOLLIP expression. TOLLIP is a ubiquitous protein with varying and complex roles in the regulation of immunity, inflammation, autophagy, and cell apoptosis in respiratory disease. Our current understanding of the biological implications of rs5743890 on TOLLIP function is limited to a possible reduction in expression. However, TOLLIP executes an inhibitory role on TLR2 and TLR4, key TLRs involved in innate immune recognition of gram-positive and gram-negative bacteria. TOLLIP prevents autophosphorylation of IRAK1 (IL-1 receptor–associated kinase 1) and negatively regulates NF-κB (nuclear factor-κB), with important downstream results for inflammatory cytokine production (11). Given correlations between inflammatory cytokines and lung microbiota (1), alterations in the alveolar cytokine milieu may promote lung dysbiosis. In addition, key changes in TLR receptor signaling may promote the expansion of specific taxa over others through impaired innate immune recognition.

There are several limitations to this study. This is a hypothesis-generating study that, in conjunction with the current literature, supports a role for TOLLIP in regulating lung microbial communities. However, the analysis was undertaken in a retrospective fashion with limited numbers of subjects and may not have been powered to accurately detect differences in communities. We also sample a cohort without a comparison group of minor allele homozygotes, which would have allowed more robust analysis.

In summary, we hypothesized that the rs5743890 TOLLIP SNP, a risk factor for enhanced mortality in IPF, would be associated with identifiable dysbiosis of the respiratory tract microbiome. Our work demonstrates that this genetic variant may be associated with community effects within the respiratory tract microbiome. The association among functional variants of the TOLLIP gene, host immunity, and outcomes in IPF is poorly understood. Given the evolving role of the lung microbiome in IPF pathogenesis and the association among rs5743890, mortality, and lung dysbiosis, further validation in human studies and mechanistic work is required to understand these interactions.