In a previous metagenomic analysis, we identified colonization of potential pathogens, including Fusobacterium nucleatum, in gastric cancer-associated microbiota. Among the eleven cancer biopsies we collected in the study, we found that four specimens were colonized by Fusobacterium nucleatum, suggesting that Fusobacterium nucleatum infection is a frequent event in the gastric cancer patients of the southwestern region of Taiwan. To provide additional support to this observation, we obtained gastric cancer tissue specimens from the Tissue Bank of Chiayi Chang Kung Hospital. These specimens are resected gastric cancer tissues from patients who underwent partial or total gastrectomy. Patients with localized cancer undergo gastrectomy so that the main lesion can be effectively removed and this is expected to offer a better prognosis for the patients. As a result, the patient cohort in this study has a better overall five-year survival rate than the survival rate reported by Cancer Registry Annual Report, Taiwan. Although this cohort represents a subgroup of gastric cancer patients, it is well suited to examine the frequency of Fusobacterium nucleatum infection.
In this study, we developed a nested PCR method in order to determine the presence of Fusobacterium nucleatum in the specimens. The purpose is to increase the sensitivity as well as the specificity of the detection assay. In previous studies using the conventional PCR method, Fusobacterium nucleatum infection in gastric cancer specimens was generally reported at low frequency [34], while our result obtained by metagenomic profiling suggests that the frequency is likely much higher than previously thought. There are two possibilities which could be the cause of this discrepancy. One possibility is that the frequency of Fusobacterium nucleatum infection is particularly high in this region in Taiwan, making it a unique phenomenon in this region. However, low frequency of detection may be due to the limitation of methodology, which subsequently leads to inability to detect low number of bacteria (Fusobacterium nucleatum) in the specimens. This is the likely scenario, especially when the specimens are removed from resected cancer tissues and contain little surface mucosa. By employing a nested PCR detection method, we hope to achieve higher detection sensitivity for identifying Fusobacterium nucleatum infection.
The detection target is highly conserved nusG gene of Fusobacterium nucleatum [17]. The size of the first- and second-stage PCR products is expected at 175 and 124 bp, respectively. The identity of amplified products was confirmed by Sanger's sequencing. We first tested the nested PCR detection method on DNA specimens from a previous study. The result showed that the nested PCR protocol is able to identify the majority of Fusobacterium nucleatum-positive specimens, except those specimens with very low percentage (data not shown). Although metagenomic analysis determines only the ratio but not the absolute number of bacteria (Fusobacterium nucleatum) in these specimens, we calculated the positive identification rate at approximately 90% while no specimen was falsely identified. We, subsequently, employed the nested PCR method followed to determine the presence of Fusobacterium nucleatum in the resected gastric cancer tissue specimens (Fig. 1A). The result showed that 19 of the 60 specimens examined are positively identified for Fusobacterium nucleatum infection. The other 41 specimens either are negative or show only marginally recognizable PCR products. All these specimens are grouped as negative for Fusobacterium nucleatum. The ratio of Fusobacterium nucleatum-positive patients is about one-third of the cohort, a result similar to that reported in our previous study using upper endoscopic-collected gastric biopsies.
The risk of Fusobacterium nucleatum infection was analyzed against clinical characteristics and cancer stage of the cohort (Fig.1B). The result of analysis indicates that older age (≥ 75) does not change the risk of Fusobacterium nucleatum infection. However, Helicobacter pylori infection appears to be positively associated with a higher risk of Fusobacterium nucleatum infection. In addition, the risk increases in the late-stage gastric cancer patients, suggesting the microenvironment becoming more suitable for colonization and growth of Fusobacterium nucleatum during the progression of cancer. Interestingly, there is a significant decrease in the risk of Fusobacterium nucleatum infection in male patients. We have no explanation for this observation at present, and it is necessary to examine larger or independent cohort to see if this phenomenon is also present in other cohorts.
We then analyzed the survival rate of patients who tested positive and negative for Fusobacterium nucleatum. The result showed that Fusobacterium nucleatum-infected patients appear to have a poorer outcome although a statistical significance (p < 0.05) was not attained (Fig. 2A). When the stage I patients who have nearly 100% five-year survival rate are excluded in survival analysis, the result is still similar in that Fusobacterium nucleatum-positive patients seem to have a poorer outcome (Fig. 2A). As noted above, patients who underwent gastrectomy have an overall better treatment outcome, hence minimizing the effect of other unfavorable factors. Nevertheless, despite of the lack of statistical significance, our analysis showed that Fusobacterium nucleatum infection could have an adverse impact on the treatment outcome.
In our cohort, approximately one third patients are tested positive for Helicobacter pylori infection. Our previous study suggested that Fusobacterium nucleatum is likely a secondary settler to the gastric microbiota after Helicobacter pylori colonizes gastric epithelium. It is possible that consecutive or simultaneous infection of Helicobacter pylori and Fusobacterium nucleatum could have synergistic effect to drive cancer progression. To examine this hypothesis, we analyzed the impact of Fusobacterium nucleatum infection to the survival of Helicobacter pylori positive patients. The result clearly demonstrated that patients infected by both Helicobacter pylori and Fusobacterium nucleatum have a poorer outcome than those infected with only Helicobacter pylori (Fig. 2B). Similar result is observed when stage I patients were excluded from the analysis (Fig. 2B). Our data provides an important clinical evidence indicating the impact of Fusobacterium nucleatum infection to the prognosis of gastric cancer patients.
The clinical data indicates that Fusobacterium nucleatum infection is frequent in patients with advanced-stage gastric cancer and may exert an unfavorable impact on the survival. We hypothesize that the interaction of Fusobacterium nucleatum with the host cells promotes cancer progression and leads to poorer outcome. In order to provide experimental support to this hypothesis, we investigated the pathogenic molecular effect of Fusobacterium nucleatum on the gastric cancer cells through an in vitro coculture system. For this study, we isolated a gastric cancer cell line derived from the resected gastric cancer tissue from a patient who was confirmed to have Fusobacterium nucleatum infection. Fusobacterium nucleatum strain ATCC25586 was cultured under anaerobic condition and harvested when the coculture experiment was performed. The bacteria was resuspended in DMEM and added to the cell culture with multiplicity of infection (MOI) at 10 and 100. The coculture was done at 37°C and 5% CO2 for 24 and 72 hours. The cells were then extensively washed and collected for subsequent RNA expression analysis by next-generation sequencing. During the incubation period, an increase in the number of bacteria (Fusobacterium nucleatum) was observed under microscope although we did not determine the precise doubling time of Fusobacterium nucleatum during the assay.
The RNA sequencing analysis revealed that the presence of Fusobacterium nucleatum led to a change in the gene expression profile in a dosage- and time-dependent manner. After 24 hours of coculturing with Fusobacterium nucleatum at a low MOI (10), only a limited number of marginally expressed genes (TPM<10) showed more than fourfold change in expression level (Figure 3A). In contrast, expression of a specific set of genes was drastically unregulated by a high MOI (100) treatment (Figure 3B). After 72 hour of coculturing, low MOI treatment led to significantly a higher number of genes, displaying more than fourfold expression increase in expression level (Figure 3A). On the other hand, the number of drastically upregulated genes in the high MOI treated cells decreases after longer incubation, but there are an increased number of genes with more than fourfold decrease in expression level (Figure 3B). Taken together, our result indicates a rapid and drastic cellular response to a high number of bacteria (Fusobacterium nucleatum). Additionally, it appears that Fusobacterium nucleatum infection, regardless the MOI, elicits a chronic response from the gastric cancer cells.
Subsequent ontological analysis revealed that these genes with altered expression levels are participating in specific cellular functions. Most prominently is the drastic increase of the genes in response to the interferon response genes. These genes, including MX1, MX2, IFI35, IFI44, IFI44L, IFIT1, IFIT2, IFIT3, IFITM1, IFITM3, IFIH1, IRF7, IFI2, and IFI6 [28], were drastically upregulated by high MOI of Fusobacterium nucleatum infection (Fig.4A). After prolonged incubation (72 hours), the level of these genes, except IFI27 and IFI6, has decreased considerably, nearly to the unstimulated basal levels. A similar induction pattern is observed for TRIM14 [8,30], ISG15 [38], USP18 [4], CD317 [6], OAS1, OAS2, OAS3, and OASL [16,31], which are all interferon-dependent genes participating in the antiviral activity of the cells (Fig. 4B). The rapid decrease indicates that the activation and regulation of the interferon response genes triggered by high number of bacteria (Fusobacterium nucleatum) are an immediate and short-term response. Moreover, low MOI treatment did not elicit the activation of the interferon response. Although the interferon response is apparently induced by a high MOI of Fusobacterium nucleatum, we only found a slight change in interferon genes, including IFNA1, IFNB1, IFND1. IFND2, and IFNE, throughout the experiment. Together, it can be concluded that Fusobacterium nucleatum at high MOI is able to activate the innate antiviral response in the gastric cancer cells.
In addition to the interferon response genes, a simultaneously increased expression of interleukins and chemokines, including IL6, IL8, IL32, CXCL1, CXCL2, and CXCL6, was observed for the high MOI treated cells at 24 hours (Fig. 4C). Similarly, the expression of these inflammatory cytokines returned to unstimulated levels at 72 hours and is apparently a part of the immediate response to the Fusobacterium nucleatum infection. However, distinct from the short burst expression of the interleukins and chemokines, CCL2 remained largely unchanged at 24 hours but was drastically upregulated at 72 hours (Fig. 4C). CCL2 is a chemoattractant of macrophages, and upregulation of CCL2 is shown to promote cancer progression [21,33]. Since both low and high MOI treatments induce the expression of CCL2 with similar strength, this suggests that the activation of CCL2 by Fusobacterium nucleatum infection is possibly done through an independent mechanism.
In addition to the genes of immunological functions, our analysis also discovered that genes involved in cell mobility and adhesion were deregulated by Fusobacterium nucleatum infection. Among these genes are EPSTI1 [12,24] and ICAM1 [9], which are both involved in cell-matrix interaction; they were upregulated at 24 hours by high MOI treatment (Fig. 5). Since the expression of these two genes returned to unstimulated levels at 72 hours, it is possible that EPSTI1 and ICAM1 were regulated as part of the interferon response. On the other hand, TNXB [2], the gene encoding tenascin-X, is transiently downregulated at 24 hours and returned to its unstimulated level at 72 hours (Fig. 5). However, suppressed by both low and high MOI treatment, regulation of TNXB is not dosage dependent and is distinct from all the other genes that respond to the interferon signaling.
Other cytoskeleton and adhesion genes are activated in a distinct pattern. Specifically, the expression of ACTA2, ACTC1, and ACTG2 is continuously increased by both low and high MOI treatments (Fig. 5). In addition to actins, CNN1, EDN1, and CEMIP were upregulated as well (Fig. 5C). CNN1 [22,26] and EDN1 [20] are involved in the endothelin pathway that modulates smooth muscle regulation and thereby exerts multiple functions to promote cancer progression. Increasing the expression of these genes was time-dependent, but not dosage-dependent, indicating that chronic infection with a small number of bacteria (Fusobacterium nucleatum) could lead to long-term and drastic effects on the cytoskeleton dynamic and cell mobility.
As stated above, there are a significantly increased number of genes suppressed by both low and high MOI treatment after 72 hours of incubation. The most prominently suppressed genes, including EGR1, NR1D1, ARC, RRAD, FOS, BHLHE41, PER1, and HES1, have diverse cellular functions (Fig. 6). Among these genes are NR1D1 [5,18], [3,37], and PER1 [14], which participate in circadian molecular mechanism but are implicated in cancer progression as well. It is possible that suppression of these genes and activation of actins and other cell adhesion genes are done through a common Fusobacterium nucleatum-dependent regulation mechanism. This putative regulatory pathway remains to be investigated.