Metagenomic analyses of the gut microbiota associated with colorectal adenoma

Recent studies have suggested an association between certain members of the Fusobacterium genus, especially F. nucleatum, and the progression of advanced colorectal carcinoma (CRC). We assessed such an association of the gut microbiota in Japanese patients with colorectal adenoma (CRA) or intramucosal CRC using colonoscopy aspirates. We analyzed samples from 81 Japanese patients, including 47 CRA and 24 intramucosal CRC patients, and 10 healthy subjects. Metagenomic analysis of the V3-V4 region of the 16S ribosomal RNA gene was performed. The linear discriminant analysis (LDA) effect size (LEfSe) method was used to examine microbial dysbiosis, revealing significant differences in bacterial abundances between the healthy controls and CRA or intramucosal CRC patients. In particular, F. varium was statistically more abundant in patients with CRA and intramucosal CRC than in healthy subjects. Here, we present the metagenomic profile of CRA and intramucosal CRC and demonstrate that F. varium is at least partially involved in the pathogenesis of CRA and intramucosal CRC.


Introduction
Increasing evidence suggests that the human gut microbiota contributes to chronic inflammation and plays important roles in the early stage of carcinogenesis of colorectal adenocarcinoma (CRC) and colorectal adenoma (CRA) through interference with various intestinal functions [1,2]. Chronic inflammation induced by the microbiota is associated with altered interactions between the host and the microbiota, microbial imbalance (dysbiosis), and infections with specific pathogens in patients with CRC or CRA [3]. Studies recently suggested that Fusobacterium spp. overall (Pan-fusobacterium), especially Fusobacterium nucleatum (F. nucleatum), are abundant in advanced CRC tissue and may contribute to invasion and metastatic proliferation [4][5][6]. Indeed [4][5][6][7]. This evidence suggests an association of the gut microbiota with the early stage of carcinogenesis of CRC following microbial dysbiosis. Moreover, the procarcinogenic activities of specific pathogens and certain microbiota-derived metabolites [8] act on colonic epithelial cells during CRC genesis. Further research may be necessary to better understand the mechanisms that underlie the association between gut microorganisms and the early stage of carcinogenesis of CRC and CRA. In addition to F. nucleatum, F. varium [9] has also recently gained notoriety as a gastrointestinal pathogen [10,11]. Indeed, links between the enrichment of Fusobacterium spp. and the development of advanced CRC have been described [4,[12][13][14]. These reports suggest an association between F. varium or F. nucleatum infections and the genesis of CRC. However, little is known about microbiome profiles during the transition from normal colonic mucosae and CRA to intramucosal CRC. In this study, we focused on microbiome profiles in CRA and intramucosal CRC patients compared to those in healthy subjects. We present a metagenomic profiling study of the microbiomes of CRA and intramucosal CRC patients using colonoscopy aspirates, which can serve as a substitute for the gut microbiota in tissues [15].

Study design and enrolled subjects
The study was reviewed and approved by the ethics committee of the Jikei Institutional Ethical Board, Jikei University School of Medicine, and the clinical study committee of Jikei University Kashiwa Hospital [No. 23-277 (6738)] on February 6, 2012. Written informed consent was obtained from each patient included in the study. All procedures were performed in accordance with the Helsinki Declaration. Eighty-one consecutive Japanese people with no previous personal history of cancer were prospectively enrolled in this study and underwent regular colonoscopy at Jikei Kashiwa Hospital. After a complete colonoscopic examination, these subjects were newly classified as healthy (n = 10), CRA patients (n = 47), or intramucosal CRC patients (n = 24). Intramucosal CRC is in its earliest stage (stage 0) and is also known as carcinoma in situ or intramucosal carcinoma. Intramucosal CRC has not yet grown beyond the inner mucosal layer of the colorectum [16,17]. Among the 81 enrolled subjects, there were no significant differences in characteristics (sex, age, body mass index, diabetes, hypercholesterolemia, hypertension, antibiotic treatment, past history, and reason for colonoscopy) between the groups (healthy, CRA, and intramucosal CRC) ( Table 1).
GCTCTTCCGATCTCTG-3' and Rev 5'-CAAGCAGAAGACGGCATACGAGAT XXXXXXXX GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCTGAC-3', where X represents a barcode base. The PCR was prepared and run, and the 16S V3-V4 amplicons were purified. The amplicon libraries were pooled in equimolar concentrations, and the V3-V4 region of the 16S rRNA gene was sequenced using a MiSeq Reagent Kit on the Illumina MiSeq platform (Shallowater, US) for paired-end sequencing as described previously [19]. The obtained sequences were analyzed using Quantitative Insights Into Microbial Ecology (QIIME) version 1.8.0, which is software that performs microbial community analysis and taxonomic classification of microbial genomes [20,21]. Potential chimeric sequences were removed using UCHIME, and the remaining sequences were assigned to operational taxonomic units (OTUs) using ppen-reference OTU picking with a 97% threshold of pairwise identity and then classified taxonomically using the Greengenes reference database (http://greengenes.secondgenome.com/downloads/ database/13_5) [22]. The microbial diversities, Shannon diversity index, and weighted and unweighted UniFrac distances were estimated using QIIME version 1.8.0 software.

Fusobacterium-targeted analysis
A primer set specific for sequencing the DNA of Fusobacterium spp. by MiSeq was designed based on the internal transcribed spacer (ITS) region between the 16S rRNA and 23S rRNA genes (S1 Fig). The forward primer was 5'-CGCTCTTCCGATCTCTGGGWACCRMGTGAACTG AAACATC-3', and the reverse primer was 5' -CGCTCTTCCGATCTGACCCTTAYGAGAT WTGGTCCTC-3'. Each 1-μl sample of DNA was amplified in triplicate using the following protocol: preheating at 94˚C for 3 min, 20 cycles of denaturation at 94˚C for 30 s, annealing at 50˚C for 30 s and extension at 72˚C for 30 s, with a final terminal extension at 72˚C for 10 min. The steps from the second amplification to the assignment and to OTU picking were the same as mentioned above, and sequences were then classified taxonomically using BLASTN against the NCBI nonredundant database.

Statistical analysis
Differences in the Shannon diversity index between the 2 groups were analyzed by t-test. Intergroup differences at the phylum, class, order, family, genus and species level in each cluster were analyzed by the linear discriminant analysis (LDA) effect size (LEfSe) method [23] with default settings on the website https://huttenhower.sph.harvard.edu/galaxy/root. LEfSe uses the two-tailed nonparametric Kruskal-Wallis test to evaluate the significance of differences in OTUs in 2 groups. A set of pairwise tests among 2 groups was performed using the unpaired Wilcoxon test. Finally, LDA was performed to estimate the effect size of each differentially abundant OTU [23,24]. The results are expressed as the mean ± SEM. A strength of the LEfSe method compared with standard statistical approaches is that in addition to providing p values, it estimates the magnitude of the association between each OTU and the grouping categories, such as CRA, intramucosal CRC, and healthy subjects [23,24]. For stringency, the gut microbiotas were considered significantly different if their differences had a p value < 0.05 and an LDA score (log10) > 3, i.e., one order of magnitude greater than the default of the LEfSe method [23,24]. The Kruskal-Wallis and Cramer V tests were used to determine associations between the groups of subjects and characteristics. StatView software, version J 5.1 (SAS Institute Inc., Cary, N.C., USA), was used for all analyses.

Overview of gut microbiota composition in patients and healthy subjects
A total of 740,273 high-quality paired sequences were obtained from the 81 samples, with a mean of 9,139 ± 3,185. We first calculated UniFrac distances to determine the similarity between microbial communities. Weighted UniFrac PCoA (principal coordinate analysis) indicated that no distinct aggregation was observed between CRA and intramucosal CRC patients, whereas healthy subjects were found to be aggregated on the lower side ( Fig 1A), suggesting a difference in gut microbiota composition between these patients and healthy subjects. Two distinct aggregations were observed in the data of Unweighted UniFrac PCoA; however, the subject in each group was not biased (Fig 1B), suggesting that the balance of gut microbes is more important than the existence of certain bacterial members in the CRA or intramucosal CRC patient gut. The Shannon index demonstrated a significant difference between CRA patients and healthy subjects (p = 0.019, S2 Fig). Moreover, there was no statistically significant

LEfSe analysis and LDA based on OTUs characterize microbiomes in patients with CRA and healthy subjects
We then performed LEfSe analysis to compare the estimated phylotypes of patients with CRA and healthy microbiotas. The gut microbial communities in patients with CRA were diverse compared to those in healthy subjects. The results indicated differences in the phylogenetic distributions of the microbiotas of patients with CRA and those of healthy subjects at the OTU level (Fig 2A). A histogram of the LDA scores was computed for features that showed differential abundance between healthy subjects and CRA patients. The LDA scores indicated that the relative abundances of Fusobacterium, Parvimonas, and Atopobium were much more enriched in patients with CRA than in healthy subjects ( Fig 2B). The most differentially abundant bacterial taxon in patients with CRA was Fusobacterium spp. (LDA score [log 10] > 3), whereas the healthy microbiome was characterized by a preponderance of Lachnobacterium, Salmonella, and Moraxellaceae (LDA score [log10] > 3) (Fig 2B).

Bacteria whose relative abundances differ significantly between patients with CRA and healthy subjects
Among the bacteria whose relative abundances differed significantly between patients with CRA and healthy subjects, both Fusobacteriales and Fusobacterium were mainly detected in patients with CRA (Fig 3). The relative abundances of Moraxellaceae, Parvimonas, Butyricimonas, Atopobium, Comamonas, Lachnobacterium, and Salmonella were also significantly different between patients with CRA and healthy subjects (p < 0.001); however, the percentage of these bacteria relative to all bacteria was extremely small (less than 1%). Both Fusobacteriales and Fusobacterium were therefore mainly associated with CRA patients, in contrast to healthy subjects.

Characterization of the microbiomes of patients with intramucosal CRC and healthy subjects by LEfSe analysis and LDA based on OTUs
LDA scores showed significant bacterial differences between healthy subjects and patients with intramucosal CRC (Fig 4A). The intramucosal CRC microbiome was characterized by a preponderance of Fusobacterium, Actinobacillus, Peptostreptococcus, Parvimonas, and Actinomyces (LDA score [log10] > 3), whereas the healthy microbiome was characterized by a preponderance of Megamonas and Sphingobium (LDA score [log10] > 3) (Fig 4B).

Bacteria with relative abundances that significantly differ between patients with intramucosal CRC and healthy subjects
The relative abundances of bacteria significantly differed between patients with intramucosal CRC and healthy subjects (Fig 5). Although the percentages of Megamonas, Peptostreptococcus, Parvimonas, Actinomyces, Actinobacillus, and Sphingobium in colonoscopy aspirates were significantly different between patients with intramucosal CRC and healthy subjects (p < 0.001), these bacteria were present in extremely small populations (less than 1%). The intramucosal CRC microbiome was also mainly characterized by a preponderance of Fusobacteriales and Fusobacterium compared to healthy subjects (p < 0.001) (Fig 5).

Characterization of Fusobacterium spp. in patients with CRA and healthy subjects
F. nucleatum and F. periodonticum were detected in approximately 60% of healthy subjects, significantly more than in patients with CRA (Fig 6). Strikingly, F. varium, appearing in approximately 80% of CRA patients, could significantly distinguish CRA patients from healthy subjects (Fig 6A and 6B).

Characterization of Fusobacterium spp. in patients with intramucosal CRC and healthy subjects
LEfSe analysis showed that F. varium was significantly more abundant in intramucosal CRC patients than in healthy subjects (Fig 7A and 7B). Moreover, approximately 60% of

Characterization of microbiomes in patients with CRA and intramucosal CRC by LEfSe analysis and LDA based on OTUs
Next, to compare the characteristic microbiomes in patients with CRA and intramucosal CRC, we performed LEfSe analysis. The LDA scores showed significant bacterial differences between patients with CRA and intramucosal CRC (Fig 8A). The intramucosal CRC microbiome was characterized by a preponderance of Collinsella, Deltaproteobacteria, and Desulfovibrionales (LDA score [log10] > 3), whereas the CRA microbiome was characterized by a preponderance of Erysipelotrichaceae and Veillonella (LDA score [log10] > 3) (Fig 8B). The LDA scores of both Fusobacteriales and Fusobacterium were not different between CRA and intramucosal CRC.

Bacteria whose relative abundances differ significantly between patients with CRA and intramucosal CRC
The relative abundances of Collinsella, Deltaproteobacteria, Desulfovibrionales, Synergistaceae; Other, Erysipelotrichaceae, and Veillonella were significantly different between patients with Fusobacterium varium in colorectal adenoma CRA and intramucosal CRC (p < 0.001); however, the percentage of these bacteria relative to all bacteria was extremely small (less than 1%) (Fig 9).

Discussion
The present study has demonstrated for the first time that F. varium may be associated with the development of CRA as well as intramucosal CRC, as analyzed by a metagenomic approach using next-generation sequencing (NGS).

Fusobacterium varium in colorectal adenoma
Previous studies have found the enrichment of F. nucleatum to be associated with advanced CRC from frozen mucosal tissues [4,5] and fecal samples [25]. In clinical situations, analyses of the gut microbiota have limitations, including small frozen sample sizes and the use of fecal samples and undefined tissue sampling sites. However, colonoscopic technologies are suitable for diagnosing colorectal neoplasia, and colonoscopy aspirates are easily obtained for diagnosis. As it has been reported that colonoscopy aspirates can serve as a substitute for the gut Fusobacterium varium in colorectal adenoma microbiota in tissues [15], we analyzed the gut microbiota in a Japanese population using colonoscopy aspirates. We detected an abundance of Fusobacterium spp. in the colonoscopy aspirates of CRA or intramucosal CRC patients. Moreover, F. varium accounted for approximately 80% of the Fusobacterium spp. from CRA patients and 60% from intramucosal CRC patients. There was no difference in the Fusobacterium spp. between CRA and intramucosal CRC. F. nucleatum accounted for approximately 60% of the Fusobacterium spp. population in healthy controls, which was significantly higher than that in CRA patients. Previous studies have also reported that F. nucleatum can be detected in the stool and colon mucosa of healthy controls as well as CRC patients [26]. F. varium may show specificity for CRA and intramucosal CRC compared to healthy subjects. High copy numbers of F. nucleatum in advanced CRC tissues were associated with late stages and worse prognoses in a Japanese population [26]. F. nucleatum may mainly play a role in the progression of advanced CRC. The gut microbiome is characterized by dietary habits (rich in red meat relative to vegetables and fruits), living environments, and metabolic levels in different human races [27]. F. nucleatum was more abundant in patients with advanced CRC from Spain than in those from the United States and Vietnam. Moreover, a recent report demonstrated that the frequency of F. nucleatum positivity in Japanese patients with advanced CRC was much lower than that in similar patients in United States cohort studies [7]. The patient populations may be at least part of the reason why F. varium but not F. nucleatum was found to be enriched in CRA or intramucosal CRC patients compared to healthy subjects. Our current data, which indicate greater abundance of F. varium in patients with CRA and intramucosal CRC than in healthy subjects in a Japanese population, suggest that Fusobacterium spp. colonization may vary regionally. However, the relationship between the gut microbiota and human races is still unknown. Moreover, the different sample types analyzed from colonoscopy aspirates in this study should also be associated with the different results. In this study, we did not analyze advanced CRC; therefore, the association of F. varium with advanced CRC may also be possible. Future studies are needed to analyze the gut microbiota from colonoscopic aspirates in adenoma, intramucosal CRC, and advanced CRC in the same sets of experiments.
F. varium can invade colonic epithelial cells, activating early intracellular signaling systems to trigger host inflammatory reactions [11]. Adhering to the colonic mucosa is essential for the gut microbiota to deliver specific oncogenic molecules into the colonic epithelial cells, which results in inflammation and oncogenic signaling in the colonic epithelial cells [28]. We recently reported the complete genome sequence of F. varium Fv113-g1, which had been isolated from a patient with UC. Fv113-g1 possessed many accessary pangenome sequences with noteworthy multiple virulence factors, including FadA, in contrast to F. nucleatum [9]. Kasper's group also reported that F. varium suppresses Reg3 to avoid death induced by antimicrobial peptides (AMPs), promoting intestinal barrier breaks [29]. These results suggest that F. varium is pathogenic in human colorectal tumors. However, the association of F. varium with neoplasm proliferation is unknown. In this study, we analyzed not tissue samples but colonoscopic aspirates. Future studies of interactions between F. varium and neoplasm proliferation are needed to assess the possible association of F. varium with the initiation and/or progression of colorectal carcinogenesis. Outgrowth of Fusobacterium spp. such as F. varium and F. nucleatum following dysbiosis may be a more important cause of colorectal carcinogenesis. The human gut microbiota may contribute to the etiology of CRC, not only via the procarcinogenic activities of specific pathogens such as Fusobacterium spp. but also via the influence of the wider microbe-induced networks for metabolic pathway aberrations [1]. Mechanisms causing Fusobacterium abundance and oncogenic properties in other gastrointestinal tract tumors and pancreatobiliary tumors remain unknown.
The complex interactions of hosts and their gut microbiotas also contribute to the pathogenesis of CRC. In human host environments, F. nucleatum modulates tumor-associated inflammation, such as that involving regulatory T cells (Tregs), in the tumor microenvironment with consequences for the inhibition or promotion of CRC biology [24]. Moreover, F. nucleatum expands myeloid-derived immune cells, which inhibit T cell proliferation and induce T cell apoptosis in patients with CRC, resulting in immunosuppression [7]. Not only F. nucleatum but also F. varium may influence inflammation of the colorectum. A recent report indicated that F. varium decreased both CD4+ and CD8+ populations more strongly and caused a higher frequency of colonic CD4-CD8-T cells than any other microbe [30]. The mechanisms of interactions between hosts and F. varium and F. nucleatum should be further investigated.
As enhanced abundances of Fusobacterium spp. such as F. varium and F. nucleatum may also be associated with some patients with CRC, Fusobacterium spp. detection in colonoscopy aspirates may be sufficient for identifying patients with increased risks for CRA or CRC [4,5]. Moreover, high levels of serum anti-Fusobacterium spp. antibodies are also a potential biomarker for CRC diagnosis [31]. Targeting Fusobacterium spp. with antimicrobial interventions may be a potential treatment or method of prevention for patients with Fusobacterium-associated CRC. Treatment of a mouse model bearing colon cancer with the antibiotic metronidazole certainly reduced Fusobacterium amounts, the proliferating activity of cancer cells, and overall tumor growth [32]. Further investigation of antimicrobial interventions as a potential treatment for patients with Fusobacterium-associated CRC is also interesting.