The specificity and structure of DNA cross-linking by the gut bacterial genotoxin colibactin

Abstract

Colibactin is a chemically unstable genotoxic gut bacterial natural product that is linked to colorectal cancer (CRC). Though it has eluded isolation and structural characterization, colibactin is proposed to contain two cyclopropane “warheads” capable of forming DNA interstrand cross-links (ICLs) connected by a reactive central scaffold of unresolved structure. The discovery of distinctive mutational signatures arising from colibactin exposure and their detection in cancer genomes suggest that colibactin influences CRC. However, we lack direct information regarding the specificity and structure of the colibactin-DNA ICL, limiting our understanding of how this natural product targets DNA and the origins of mutations arising from this DNA damage.Though prior studies had revealed colibactin’s DNA alkylating activity and implicated adenine (A)– and thymine (T)–rich sequences as likely sites for ICL formation, the precise nature of colibactin’s interactions with DNA, the exact sites of alkylation, and its sequence specificity were unknown. To address these gaps in knowledge, we sought to experimentally elucidate the specificity and structure of the colibactin-DNA ICL using biochemical assays, advanced mass spectrometry (MS), and nuclear magnetic resonance (NMR) spectroscopy approaches.We first investigated the reactivity of colibactin toward DNA oligonucleotides in vitro using a newly developed MS-based assay, overcoming the challenge of its chemical instability by leveraging in situ bacterial production. We observed ICL formation of bis-N3-adenine ICLs within a preferred motif of 5′-WAWWTW-3′ (where the adenines bolded and opposite the underlined thymine are alkylated, and W represents either A or T). This preference for AT-rich sequences is consistent with the locations of colibactin-derived mutational signatures. Additional experiments suggested that colibactin binds and alkylates in the minor groove.To gain initial insights into the structure of the colibactin-DNA ICL, we further applied MS to characterize the intact lesion. Unexpectedly, we observed a mass consistent with ICL formation arising from a proposed colibactin structure containing a chemically unstable central α-ketoimine.To obtain more detailed structural information, we produced the colibactin-DNA ICL on a large enough scale to enable solution-state NMR studies. The structure that we obtained verifies the sites and locations of colibactin DNA alkylation identified in our in vitro assays. Analysis of the structure revealed chemical features of colibactin that are important for DNA binding and alkylation and explain its sequence specificity. Most notably, the positively charged central α-ketoiminium of colibactin makes extensive electrostatic and hydrogen bonding interactions with the floor of the minor groove. The results of calculations and experiments with a synthetic colibactin analog further support the importance of this unstable central functional group to the specificity of colibactin-DNA ICL formation.Our study reveals the specificity and structure of the colibactin-DNA ICL by combining MS and NMR. Colibactin’s preference for alkylating AT-rich sequences sheds light on the origins of mutational signatures. These results also help resolve the structure of colibactin’s unstable central region and implicate it as a key determinant of sequence specificity. Together, our findings reveal a strategy for DNA alkylation distinctive among natural products, enhancing our understanding of colibactin’s chemical structure, its recognition of and reaction with DNA, and its downstream effects on the host genome.