Leaving Your Mark

Nature reviews | CanCer Associations between the intestinal microbiome and colorectal cancer (CRC) development have been proposed. However, direct links between the presence of an individual bacterial species and the generation of mutations driving tumorigenesis have been harder to pinpoint. Certain strains of Escherichia coli present in the stool and tumour biopsy samples of patients with CRC harbour a pathogenicity island, pks, encoding a series of enzymes that produce a genotoxin known as colibactin. Only last year, colibactin–DNA adenine adducts unique to mammalian cells infected with pks+ E. coli were identified. Now with the application of organoid technology, PleguezuelosManzano et al. have discovered two co-occurring mutational patterns in DNA that arise following exposure to pks+ E. coli, firmly establishing causality. To examine the consequences of genotoxic E. coli exposure on host epithelial cells, the authors developed a co-culture system wherein a pks+ E. coli strain derived from a CRC biopsy sample was microinjected into the lumen of clonal human intestinal organoids. DNA damage, specifically double-strand breaks and interstrand crosslinks, characteristic of that previously seen to be induced by pks+ E. coli was observed. However, infection of intestinal organoids with an isogenic mutant strain knocked out for clbQ (pksΔclbQ E. coli), which encodes an enzyme involved in the biosynthetic pathway of colibactin, did not result in DNA damage, confirming the phenotype was specific to the activity of colibactin. Next, to investigate the long-term effects of colibactin exposure, single cell-derived organoids were repeatedly injected with pks+ E. coli or pksΔclbQ E. coli over 5 months before subclonal organoids were derived from individual cells within the original cultures. Whole genome sequencing (WGS) of the clonal organoids before and after exposure to the genotoxic pks+ E. coli revealed an increase in the presence of single base substitutions (SBSs) compared with those subclones treated with pksΔclbQ E. coli. Typically these substitutions were changes of T to any of the other three nucleotides and occurred preferentially in the middle base of ATA, ATT and TTT triplets. The authors defined this as a pks-specific SBS signature (SBS-pks) as it could not be detected in organoids injected with pksΔclbQ E. coli. A second mutational signature characterized by a small insertion and deletion (indel) was also identified (ID-pks). This particular mutational pattern took the form of a single deletion of T in long poly-T stretches. In addition, both mutational signatures were further characterized by the presence of adenine residues upstream of the mutated sites. Taken altogether, the SBS-pks, ID-pks and associated recurrent patterns, collectively known as the pks-mutational signature, is distinct from those induced by other known environmental mutagens. Moving away from the in vitro set-up to investigate the potential presence of this pks-signature in human tumours, the authors analysed WGS data from a collection of 3,668 solid tumour metastases, under the assumption that a mutation acquired in a primary tumour will be maintained in the corresponding metastases. This demonstrated that SBS-pks and ID-pks co-occurred in CRC metastases and were enriched relative to metastases from other cancer types. However, it was noted that the pks-signature was also present in metastases derived from one head and neck tumour and three urinary tract tumours (two cancer types originating in tissues that are common sites of E. coli infection), suggesting that the genotoxicity of E. coli could act in other organs outside of the colon. A second, independent dataset consisting of 2,208, mostly primary, CRC tumours confirmed the enrichment of SBS-pks and ID-pks motifs in patient samples. Reasoning that the pks-signature could be the source of oncogenic mutations, common driver mutations present in CRC tumours from seven independent cohorts were compared to the SBS-pks and ID-pks signatures. Interestingly, out of 4,712 driver mutations, 112 (2.4%) were shown to match the pks-signature with adenomatous polyposis coli (APC) – the most frequently mutated gene in CRC – having the highest number of exonic mutations corresponding to either the SBS-pks or ID-pks signatures. Furthermore, in two independent CRC metastases from the original collection, an identical driver mutation in APC causing a premature stop codon could be matched to the SBS-pks signature. Mutational signatures arising in healthy human colorectal crypts have recently been described by Stratton and colleagues. Specifically, the co-occurrence of two motifs, named SBS-A and ID-A, was observed in a subset of crypts and could be traced to an unknown mutagenic agent acting in early childhood. Intriguingly, SBS-pks and ID-pks from this study strongly matched SBS-A and ID-A, respectively, intimating that the mutagenic agent responsible for these signatures in the healthy colon of certain individuals is pks+ E. coli. This study not only suggests that screening for and early elimination of pks+ E. coli could have important implications for CRC prevention but calls for reconsidered use of off-the-shelf probiotics that contain genotoxic strains of E. coli.