The Intersection between Oral Microbiota, Host Gene Methylation and Patient Outcomes in Head and Neck Squamous Cell Carcinoma

Abstract

The role of oral microbiota in head and neck squamous cell carcinoma (HNSCC) is poorly understood. Here we sought to evaluate the association of the bacterial microbiome with host gene methylation and patient outcomes, and to explore its potential as a biomarker for early detection or intervention. Here we performed 16S rRNA gene amplicon sequencing in sixty-eight HNSCC patients across both tissue and oral rinse samples to identify oral bacteria with differential abundance between HNSCC and controls. A subset of thirty-one pairs of HNSCC tumor tissues and the adjacent normal tissues were characterized for host gene methylation profile using bisulfite capture sequencing. We observed significant enrichments of Fusobacterium and Peptostreptococcus in HNSCC tumor tissues when compared to the adjacent normal tissues, and in HNSCC oral rinses when compared to healthy subjects, while ten other bacterial genera were largely depleted. These HNSCC-related bacteria were discriminative for HNSCC and controls with area under the receiver operating curves (AUCs) of 0.84 and 0.86 in tissue and oral rinse samples, respectively. Moreover, Fusobacterium nucleatum abundance in HNSCC cases was strongly associated with non-smokers, lower tumor stage, lower rate of recurrence, and improved disease-specific survival. An integrative analysis identified that enrichment of F. nucleatum was associated with host gene promoter methylation, including hypermethylation of tumor suppressor genes LXN and SMARCA2, for which gene expressions were downregulated in the HNSCC cohort from The Cancer Genome Atlas. In conclusion, we identified a taxonomically defined microbial consortium associated with HNSCC that may have clinical potential regarding biomarkers for early detection or intervention. Host-microbe interactions between F. nucleatum enrichment and clinical outcomes or host gene methylation imply a potential role of F. nucleatum as a pro-inflammatory driver in initiating HNSCC without traditional risk factors, which warrants further investigation for the underlying mechanisms.

Keywords: Fusobacterium; HNSCC; host-microbiome interaction; methylation; microbiome.

Figure 1

Oral microbiota dysbiosis associated with head and neck squamous cell carcinomas (HNSCCs). (A) Principal coordinates plot using Bray–Curtis dissimilarity. (B) Comparison of oral microbiota diversity summarized at the genus level. (C) Comparison of oral microbiota between HNSCC tumor (tumor) and adjacent normal (AN) tissues, with bacterial genera that best discriminated HNSCCs and controls by a linear discriminant analysis (LDA) for effect size (LEfSe). The bar length represents log10 LDA score. Difference in the relative abundance tested by pairwise Tukey HSD post hoc is shown on the right panel. (D) Comparison of oral microbiota between HNSCC oral rinses and healthy subjects. Adjusted Mann–Whitney U-test p value is marked with * if ≤ 0.05, ** ≤ 0.01, or *** ≤ 0.001.

Figure 2

A refinement of twelve “core” bacterial genera discriminating HNSCCs from controls. (A) A SparCC correlation analysis (r ≥ 0.3, p≤ 0.01) showing connections between pathogen–pathogen, nonpathogen–nonpathogen, and pathogen–nonpathogen bacteria. Grey dots indicate bacterial genera with no significant differences in abundance between HNSCC and controls. Red and green dots indicate bacterial genera enriched in HNSCCs and controls, respectively. Connections with red and blue lines indicate collaborative and inhibitive relationships, respectively. Hierarchical cluster analyses of tissues (B) and oral rinses (C) using distance matrix of 12 discriminative bacterial genera classified oral microbial communities into two clades, namely, high-risk and low-risk, based on the dendrogram topologies. The receiver operating characteristic (ROC) analysis of combined twelve genera achieved AUCs of 0.84 and 0.86 in tissues and oral rinses, respectively. Clustering between the two groups was evaluated using a two-tailed Fisher exact test.

Figure 3

Association of Fusobacterium abundance with HNSCC patient outcomes. (A) High abundance of Fusobacterium in HNSCC tumor tissues was positively associated with non-smokers, low T classification, better disease specific survival, and lower recurrence of disease. (B) Both univariate and multivariate regression analyses demonstrated that the high amount of Fusobacterium was an independent predictor of HNSCC disease specific survival. (C) A higher relative abundance of Fusobacterium in tumor tissue was correlated with improved 3-year disease-specific survival (DSS) and disease-free survival (DFS). (D) Quantitative PCR (qPCR) targeting the nusG gene confirmed the high abundance of Fusobacterium nucleatum in HNSCC tumor tissue.

Figure 4

RNA ISH analysis of F. nucleatum in HNSCC tumors. (A) and (B) demonstrate an F. nucleatum biofilm-like lining on the surfaces of the tumor cells at 200×; the black arrows point to the biofilm. (C) and (D) demonstrate the intracellular localization of F. nucleatum in tumor cells located adjacent to keratin whorls of squamous cell carcinoma at 200×. (E) and (F) further demonstrate the localization of F. nucleatum in the intracellular compartments of tumor cells in an oral cavity squamous cell carcinoma at 200×.

Figure 5

Bisulfite DNA capture sequencing demonstrating differentially methylated human gene promoters in Fusobacterium-enriched HNSCC tumor tissues. (A) A heat map of methylation profiles of host gene promoters between Fusobacterium-high (F-High) and Fusobacterium-low (F-Low) HNSCC tumor tissues. Negative binomial generalized linear regression model z values (glm.nb, p ≤ 0.05) are shown in the right panel of the heat map, with colors indicating gene expression changes (log2FC) in the TCGA HNSCC cohort. (B) Positive correlation between Fusobacterium enrichment and TSG-LXN promoter hypermethylation in HNSCC tumor tissues. (C) Positive correlation between Fusobacterium enrichment and TSG-SMARCA2 promoter hypermethylation in HNSCC tumor tissues. AN: adjacent normal tissues. Tumor: HNSCC tumor tissues.