Fusobacterium nucleatum infection correlates with two types of microsatellite alterations in colorectal cancer and triggers DNA damage

Fn is a common resident in the human gut mucosa and is an anaerobic bacterium that colonizes CRC tumors more frequently than adjacent normal mucosa. To date, most epidemiological studies using 16s rRNA sequencing or metagenomic sequencing methods have detected an increased level of Fn DNA and/or RNA in colorectal adenoma/carcinoma tissues or stools from tumor bearing patients, as compared with normal controls [1]. Furthermore, Fn infection is associated with specific subtypes of CRC that exhibits CIMP or MSI-H [2, 3]. These observations might suggest that Fn infection may contribute to a serrated pathway of CRC development [4]. On the other hand, tumor tissue infected with Fn exhibits an inflamed tumor microenvironment, rich in inflammatory factors such as IL6 or reactive oxygen species [5], leading to the assumption that Fn infection might also contribute to the generation of IAMA or L/E positive CRC [6‐8]. Despite a strong association between Fn infection and colorectal cancer, there has been no evidence of Fn infection damaging the DNA of colon tissues. In this study, we show evidence that a degree of Fn infection may determine molecular characteristics of CRC, and that Fn infection may be carcinogenic.

A total of 304 cases of unselected sporadic CRC from North Carolina [9, 10] were analyzed for MSI-H, MSI-L and EMAST [11, 12]. The amount of Fn DNA per nanogram of tumor tissue DNA was also determined by qPCR (see Additional file 1: Additional Materials and Methods). Thirty-eight cases (12.5%), 129 cases (42.4%) and 137 cases (45.1%) exhibited MSI-H, L/E and MSS, respectively. Fn DNA was detected in 116 of 304 (38%) CRC tumor tissues, ranging from 0.002 to 880 pg/ng of tissue DNA. When the quantity of Fn DNA was compared among MSI-H, L/E and MSS CRC, the Fn DNA load in MSI-H was the highest (MSI-H > L/E, p = 0.028; MSI-H > MSS, p = 0.000085) and the Fn DNA load in L/E was higher than in MSS (L/E > MSS, p = 0.028) (Fig. 1a). We then determined whether Fn infection was associated with MSI-H and/or L/E compared to MSS using a logistic regression model. In univariate analysis, Fn infection was associated with MSI-H at an odds ratio (OR) of 4.21 (p < 0.001) and was also associated with L/E at an OR of 1.74 (p = 0.03). When adjusted for sex, age, tumor location and tumor stage, MSI-H (OR = 3.99, 95%CI 1.85–8.9, p < 0.001) and L/E (OR = 1.68, 95% CI 1.00–2.84, p = 0.05) were independently associated with Fn infection (Fig. 2). To validate the above results, we analyzed 174 cases of CRC from Mie, Japan. Thirteen (7.4%), 69, (39.7%) and 92 cases (52.9%) exhibited MSI-H, L/E and MSS, respectively. Fn DNA was detected in 131 of 174 (75%) tumors, ranging from 0.0003 to 200 pg/ug tissue DNA. The quantity of Fn DNA was highest in MSH-H compared to L/E (p = 0.02) or MSS (p = 0.0005), and the Fn load was higher in L/E than MSS (p = 0.015) (Fig. 1b). Fn infection was associated with MSI-H at OR = 13.83 (p = 0.007) and with L/E at OR = 2.35 (p = 0.02) in univariate logistic regression analysis. Multivariate analysis adjusted by sex, age. tumor location and stage showed that MSI-H (OR = 13.67, 95% CI 1.63–1789.13, p = 0.01) and L/E (OR = 2.23, 95% CI 1.05–4.95, p = 0.04) were significantly associated with Fn infection compared to MSS (Fig. 2).

To explore whether Fn infection causes cellular DNA damage (see Additional file 1: Additional Materials and Methods), we first determined the ability of human colon cancer cells to support infection. When each of 16 human colon cancer cell lines was co-cultured with Fn in 5% CO₂/21% O₂ conditions, Fn grew aerobically in 12 of 16 cell lines (Fig. 3a), but not in 4 cell lines (Fig. 3b). Furthermore, there was a difference in the ability to support aerobic growth of Fn among the 12 cell lines. Some cell lines such as WIDR required less Fn (MOI of 0.001) to initiate successive Fn growth whereas SNU503 required Fn at MOI of 10 (Fig. 3a). The supernatants from co-cultures between WIDR and Fn, where Fn grew, induced γ-H2AX in various colon cancer cell lines (Fig. 3c, d), a hallmark of DNA double strand breaks (DSBs) and suggesting that Fn infection may be carcinogenic to infected tissues, whereas the supernatants from co-cultures between HCEC-1CT and Fn, where Fn growth was not permissive, did not induce γ-H2AX in the exposed colon cancer cell lines (Fig. 3e). Inclusion of the antibiotic metronidazole [13] in co-cultures between WIDR and Fn inhibited Fn growth (Fig. 3f) and abolished the supernatants’ ability to induce γ-H2AX in WIDR cells (Fig. 3g). Finally, the bacterial culture medium where Fn was anaerobically grown induced γ-H2AX in WIDR cells (Fig. 3h), indicating that Fn produces a factor that may cause DNA DSBs in mammalian cells.

Here we provide the initial report showing that heavy or moderate loads of Fn DNA are associated with MSI-H and L/E CRC, respectively. We have also identified evidence that Fn infection may cause DNA damage in infected colon tissues. It remains to be determined whether different degrees of Fn infection directly or indirectly impair DNA mismatch repair differently, resulting in MSI-H or L/E, or whether Fn opportunistically heavily infects MSI-H compared to L/E or MSS CRC. The host cell-dependent aerobic growth of Fn observed in this study may give further clues to answer these questions. However, our observation that Fn infection may trigger cellular DNA damage strongly suggests that Fn infection causes genetic and/or epigenetic alterations, initiating and/or promoting colorectal carcinogenesis. The identity and origin of the DNA damaging factor generated by Fn infection will need to be investigated.