Background

The composition and function of the gut microbiome have been shown to potentially influence the initiation and progression of colorectal cancer (CRC) [1]. Patients with CRC have an unbalanced gut microbiome, or dysbiosis, which is characterized by a decrease in beneficial bacteria and an increase in pathobionts, such as colibactin-producing Escherichia coli and enterotoxigenic Bacteroides fragilis (ETBF) [2].

While gut microbiota contains commensal E. coli strains, some strains may carry a pathogenic potential [3]. The pks genomic island contains the colibactin (clb) gene cluster, which encodes the genes required for colibactin synthesis [4]. Colibactin is a genotoxin that causes inter-strand cross-links (ICLs) [5] and double-strand DNA breaks (DSBs), cell cycle arrest, senescence, and chromosomal abnormalities in mammalian cells [6]. Murine models of pks + E. coli mono-colonization [7, 8] and colonization of adenomatous polyposis coli multiple intestinal neoplasia (ApcMin/+) mice with colibactin producing E. coli [9] revealed a causal link between the presence of colibactin and intestinal tumorigenicity. Other Enterobacteriaceae species, such as Klebsiella, inherited the pks island and some genes of the cluster from horizontal transfer and can also produce colibactin [10, 11]. Colonization with colibactin-producing bacteria in humans occurs mainly during early life [12], and the presence of the phylogroup of pks + E. coli is steadily increasing worldwide [13, 14].

Bacteroides strains such as EBTF have also been associated with CRC. ETBF which produces Bacteroides fragilis toxin (bft), has been shown to contribute to colon carcinogenesis [15] through induction of colonocyte proliferation [16], inhibition of apoptosis and promotion of proinflammatory signaling [17, 18]. Accordingly, ETBF colonization in a murine model of colitis-induced CRC increased the number of tumours [19], while in the ApcMin/+ CRC mouse model, it promoted the development of colon adenomas [20], further confirming its carcinogenic potential.

In this study we assessed the prevalence of pks + bacteria and ETBF in a cohort of 94 CRC patients and 62 healthy individuals from the province of Québec, Canada.

Methods

Patient recruitment and sample collection

Patients with CRC and healthy individuals were recruited at the Centre hospitalier de l’Université de Montréal (CHUM) (Additional file 1: Table S1). Individuals with inflammatory bowel disease (IBD), polyps or antibiotic treatment 6 months prior to sampling were excluded from the control group. Participants were requested to provide a fresh fecal sample collected at home following the International Human Microbiome Standards procedure [21]. Samples were collected in hermetic containers with an anaerobic sachet (BD BBL™ GasPak™ anaerobic indicator, BD, ON, Canada) and stored at −80 °C upon arrival at the laboratory within 24 h of sampling.

DNA extraction and polymerase chain reaction

Total DNA was extracted from human fecal samples with the PowerSoil® DNA Extraction Kit (Qiagen Inc., Toronto, ON, Canada) and polymerase chain reaction (PCR) was performed using PowerUp™ SYBR™ Green Master Mix (Thermo Fisher Scientific, Waltham, MA, USA) in the RG 3000A R PCR machine (Qiagen Inc.) using the following cycling conditions; 50 °C for 2 min, 95 °C for 2 min, 38 cycles of 15 s at 95 °C, followed by 1 min at 60 °C. Simultaneous amplification of colibactin A gene (clbA) and E. coli 16S rRNA were done with the following primers for clbA: Fw 5ʹ-CTCCACAGGAAGCTACTAAC-3ʹ, Rv 5ʹ-CGTGGTGATAAAGTTGGGAC-3ʹ [4] and Ecoli 16S: Fw 5ʹ-GTTAATTTTGCTCATTGA-3ʹ, Rv 5ʹ-ACCAGGGTATCTAATCCTGTT-3ʹ[22], with a 1:1:1:1 ratio. For the detection of ETBF, we performed a simultaneous PCR of the bft gene and B. fragilis 16S rRNA with the following primers for bft: Fw 5ʹ-GAACCTAAAACGGTATATGT-3’, Rv 5ʹ-GTTGTAGACATCCCACTGGC-3ʹ [8] and Bfr: Fw 5ʹ-CTGAACCAGCCAAGTAGCG-3ʹ, Rv 5ʹ-CCGCAAACTTTCACAACTGACTTA-3ʹ [23], with a 5:5:1:1 ratio. We used the E. coli NC101 strain (EcNC101 (a gift from Dr. Christian Jobin, Cancer Microbiota & Host Response, UF Health Cancer Center, University of Florida)) as a positive control for the presence of the pks island, and the ETBF strain (a gift from Dr. Cindy Sears, Johns Hopkins University School of Medicine [8]) as the positive control for the bft gene. The PCR products were then visualized on a 1.8% agarose gel containing Eco-stain plus (Bio Basic Inc., Markham, ON, Canada). The expected product sizes were: 330 bp for E. coli 16S rRNA; 300 bp for clbA; 230 bp for B. fragilis 16S rRNA; and 370 bp for bft.

Statistics

All data were analyzed using GraphPad Prism (Version 5.0, GraphPad Software, San Diego, CA, USA). χ2 tests were used to compare categorical variables, unless expected frequencies were  < 5, in which case Fisher’s exact test was used. P < 0.05 were considered statistically significant.

Results and discussion

The presence of colibactin-producing bacteria in stool samples collected from participants (Additional file 1: Table S1) was detected by PCR using specific primers targeting the clbA gene encoded in the pks island, required for colibactin production [4]. In addition, as a positive control for the PCR reaction, primers universal for all strains of E. coli were used [22] (Fig. 1a). We found that 42% of healthy donors and 46% of CRC patients were colonized by a pks + bacteria (Fig. 2, Table 1). Interestingly, pks + bacteria were more prevalent in late-onset (40 out of 79; 52%) compared to early-onset CRC (3 out of 15; 20%; P < 0.05) (Fig. 2, Table 1). Overall, the levels of pks + bacteria colonization in our CRC patients were within the range previously reported in literature with 68% [8] and 66.7% [7] in two cohorts from the USA, 56.4% in Sweden [24], 43% in Japan [25], 23% in Iran [26], and 16.7% in Malaysia [27]. As for the healthy population, they approached levels reported in a Japanese cohort (46%) [25], whereas lower levels were found in other healthy cohorts: 22% [8] and 20.8% [7] in the USA; 18.5% in Sweden [24]; 7.1% in Iran [26] and 4.35% in Malaysia [27]. These disparities in prevalence around the world could be attributed to dietary differences. For example, the so-called Western diet [28] has been linked to a higher incidence of colorectal cancer containing pks + E. coli [29]. Our study indicates that colibactin-producing bacteria are less prevalent in early-onset compared to late-onset CRC, although this finding should be confirmed in larger cohorts. While this could indicate that colibactin-producing bacteria may not be involved in the etiology of early-onset CRC, we cannot rule out that pks + E. coli and other colibactin-producing bacteria may have been present during childhood and subsequently eliminated, with the effects of early mutagenic exposure manifesting later in life [30].

Fig. 1
figure 1

clbA and bft detection. a clbA and b bft presence in DNA extracted from fecal samples were detected using conventional qualitative PCR. M: marker; negative (−) control: water (H2O); positive control ( +): DNA extracted from pks + EcNC101 or ETBF bacteria

Full size image
Fig. 2
figure 2

Prevalence of pks + and ETBF in the cohort. Controls – healthy participants; CRC – colorectal cancer patients. * P < 0.05, Fisher’s exact test

Full size image
Table 1 Prevalence of clbA and bft in controls and CRC patients
Full size table

Other possible explanations for the increasing incidence rate of CRC in the younger population [31] could be related to early exposures to a deleterious lifestyle, environmental pollutants, a western diet [32], diets high in sugar [33], metabolic diseases during adolescence [34] or other components of the gut microbiota, such as the genera Fusobacterium and Flavonifractor [35].

To detect the presence of ETBF among the cohort, PCR using specific primers targeting the bft gene [8] was performed. Additionally, as a positive control for the PCR reaction, primers universal for B. fragilis strains [23] were used (Fig. 1b). Bft was detected in 21% of healthy donors and 32% of CRC patients (Fig. 2, Table 1). Overall, the levels of ETBF colonization in our CRC patients were within the range previously reported from other cohorts with 6.1% in Japan [36], 31.6% [37] and 47% [38] in two Iranian cohorts, 38% in Turkey [16], 49.3% in New Zealand [39], and 60% in the USA [8]. Regarding healthy controls, prevalence in our cohort was higher than those reported in the Turkish cohort (12%) [16], and in two Iranian cohorts (3.8% and 8.3%) [37, 38], whereas higher levels were reported in a cohort from the USA (30%) [8].

Finally, double colonization with pks + bacteria and ETBF was detected in 8% of healthy individuals and 13% of CRC patients (Table 2). In a US cohort, higher levels of double colonization with pks + bacteria and ETBF were detected in the healthy population (22%), with even higher levels reported in patients with familial adenomatous polyposis (FAP) (52%) [8]. Of note, the presence of both pks + bacteria and ETBF may lead to higher colonic inflammation and tumorigenesis [8].

Table 2 Prevalence of double colonization in controls and CRC patients
Full size table

Conclusion

The prevalence of colibactin-producing bacteria and ETBF in CRC patients from our cohort was within the range reported in other studies. Nevertheless, we found that healthy controls had higher prevalence of pks + bacteria and ETBF when compared to most of the other cohorts. However, when comparing different reports, it should be taken into account that the type of tissue (mucosal vs. fecal samples) and measurement techniques (cultured vs. direct PCR) used to determine the prevalence of pro-carcinogenic bacteria may account for some of the variations between reported results. In any case, as these healthy individuals may be at a higher risk of developing CRC due to the potentially elevated levels of pks + bacteria and ETBF, it is critical to propose adapted dietary and medical interventions to regulate the abundance of these bacteria. A novel result of our study is the finding of a low prevalence of pks + bacteria in early-onset compared to late-onset CRC. Further studies are needed to understand the role of colibactin-bacteria in early-onset CRC.