Genome-wide somatic mutation analysis via Hawk-Seq™ reveals mutation profiles associated with chemical mutagens

Abstract

It is difficult to identify mutagen-induced genome-wide somatic mutations using next generation sequencing; hence, mutagenic features of each mutagen and their roles in cancer development require further elucidation. We described Hawk-Seq™, a highly accurate genome sequencing method and the optimal conditions, for using it to construct libraries that would enable the accurate (c.a. 1 error/10 7-10 8 bp) and efficient survey of genome-wide mutations. Genomic mutations in gpt delta mice or Salmonella typhimurium TA100 exposed to methylnitrosourea (MNU), ethylnitrosourea (ENU), diethylnitrosamine (DEN), benzo[a]pyrene (BP), and aristolochic acid (AA) were profiled using Hawk-Seq™ to analyse positions, substitution patterns, or frequencies. The resultant vast mutation data provided high-resolution mutational signatures, including for minor mutational fractions (e.g. G:C>A:T by AA), which enabled the clarification of the mutagenic features of all mutagens. The 96-type mutational signatures of MNU, AA, and BP indicate their partial similarity to signature 11, 22, and 4 or 29, respectively. Meanwhile, signatures attributable to ENU and DEN were highly similar to each other, but not to signature 11, suggesting that the mechanisms of these agents differed from those of typical alkylating agents. Thus, Hawk-Seq™ can clarify genome-wide chemical mutagenicity profiles at extraordinary resolutions, thereby providing insight into mutagen mechanisms and their roles in cancer development.

Sequencing Chemically Induced Mutations in the Mutamouse Lacz Reporter Gene Identifies Human Cancer Mutational Signatures

“…The BaP mutation profile that we derived using our approach is consistent with previous work in vivo 27 and in vitro 28 that demonstrated the presence of SBS 4 after exposure to BaP. Indeed, the BaP mutation profile is consistent among the three studies (Pearson’s correlation of 0.80 and 0.71 with the in vivo and in vitro profile, respectively).…”

“…The mutation profile obtained with ENU, demonstrating a slight preponderance of T>A mutations over T>C mutations, is consistent (Pearson’s coefficient = 0.71) with that obtained in the bone marrow of gpt delta mice 27 , although the correlation is reduced when expanding the six possible base pair alterations to the 96 possible mutation types (Pearson’s coefficient = 0.49). This is mostly due to a deficiency of T>C mutations at CTN motifs with respect to gpt delta mice.…”

“…Thus, we developed a background signature derived from our empirical control data that can be integrated with COSMIC signatures to reduce the noise attributable to spontaneous mutation patterns. This in vivo control signature is a unique feature of our study, as there is currently no ‘background’ COSMIC signature and no in vivo control signature is reported in a recent study that generated chemical-specific signatures in vivo using a different approach 27 . Our results show that C>T transitions are the most common spontaneous mutations in vivo (Figure 5) and this was consistent among all tissues analyzed (Supplementary Figure 6).…”

“…Indeed, previous studies have demonstrated that mutational signatures observed in aflatoxin-induced cancers are observed in normal tissues long before tumour formation 25, 26 . Recent work in vivo 27 and in vitro 28 has shown that chemical-specific signatures detected shortly after exposure match signatures seen in human cancers. Thus, characterization of short-term mutational signatures in non-tumour tissues is a valuable approach to elucidate human-relevant mechanisms of carcinogenesis.…”

Detection of genome-wide low-frequency mutations with Paired-End and Complementary Consensus Sequencing (PECC-Seq) revealed end-repair derived artifacts as residual errors

“…Besides, this strategy also requires molecular barcodes to identify the PCR-amplified copies that are generated from same starting templates. Based on this strategy, many attempts have been made to improve the accuracy of standard NGS techniques (4,11), e.g., Safe-Sequencing System (Safe-SeqS) (12), circle sequencing (13), bottleneck sequencing system (BotSeqS) (14), duplex sequencing (DupSeq) (15) and hypothesis alignment with weak overlap (Hawk-Seq TM ) (16) for short-read platforms, as well as INC-seq (17) and circular consensus sequencing (CCS) (18) for long-read platforms. In short-read platforms, these consensus sequencing approaches have successfully improved the accuracy to a level of 10 −4 to 10 −7 or even lower (4,11).…”