Genome-wide surveillance of transcription errors in response to genotoxic stress

Abstract

Mutagenic compounds are a potent source of human disease. By inducing genetic instability, they can accelerate the evolution of human cancers or lead to the development of genetically inherited diseases. Here, we show that in addition to genetic mutations, mutagens are also a powerful source of transcription errors. These errors arise in dividing and nondividing cells alike, affect every class of transcripts inside cells, and, in certain cases, greatly exceed the number of mutations that arise in the genome. In addition, we reveal the kinetics of transcription errors in response to mutagen exposure and find that DNA repair is required to mitigate transcriptional mutagenesis after exposure. Together, these observations have far-reaching consequences for our understanding of mutagenesis in human aging and disease, and suggest that the impact of DNA damage on human physiology has been greatly underestimated.

Keywords: DNA damage; genotoxic stress; mutagenesis; transcription error.

Fig. 1.

MNNG causes a genome-wide increase in transcriptional mutagenesis. (A) Core concept of the circle-sequencing assay. (Left) Traditional sequencing approaches are capable of identifying transcription errors (red dots) present in isolated RNA fragments; however, during library preparation, reverse transcription errors introduce additional mutations into the complementary DNA (cDNA) (blue dots) that are indistinguishable from true transcription errors. Additional artifacts (green dots) are introduced during sequencing as well, which is highly error-prone. Right: To prevent these artifacts from confounding error measurements, RNA targets are circularized prior to reverse transcription. These circularized molecules are then reverse-transcribed in a rolling circle fashion to generate linear cDNA molecules that contain several tandem repeats of the original RNA fragment (orange strands). If a transcription error was present in the RNA template, this error will be detected in each of the repeats, while artifacts are only present in one repeat. (B) Dividing and nondividing cells experience >10-fold increase in transcription errors after 40 min of exposure to MNNG. These errors outnumbered the amount of mutations that arose in the genome. (C) Transcription errors are induced throughout the entire transcriptome of S. cerevisiae. −M indicates no MNNG treatment, +M indicates MNNG treatment, +6 indicates 6 h after MNNG treatment. Chromosomes are laid out in an end-to-end fashion, from chromosome I until XVI and the mitochondrial genome. (D) Each RNA polymerase has a unique error rate after MNNG treatment, reflecting the unique sensitivities and dynamics of different types of transcripts to mutagen exposure. (E) MNNG induces only single-base substitutions; no increase in deletions or insertions was detected. (F) The vast majority of single-base substitutions induced by MNNG are C→U transitions, errors that can be caused by O⁶-methyl-guanine. * denotes P < 0.05, unpaired two-tailed t test; error bar indicates SEM. For all samples, n = 3–13 biological replicates, except for transcription error measurements of dividing cells, for which n = 2.

Fig. 2.

Multiple mutagens cause transcriptional mutagenesis by RNAPII in multiple organisms. (A) MNNG, EMS, and ENU, but not MMS, UV light, or 4NQO induce transcription errors in nondividing cells. (B) Each dose was titrated to induce a ∼5-fold increase in genetic mutations. (C) Rpb9Δ cells display higher error rates upon MNNG exposure than WT cells. (D) MNNG induces transcription errors in C. elegans, D. melanogaster, and primary fibroblasts derived from adult mice. ns denotes nonsignificant differences. * denotes P < 0.05, unpaired two-tailed t test; error bar indicates SEM. For all samples, n = 3–6 biological replicates.

Fig. 3.

The kinetics of transcriptional mutagenesis in nondividing cells. (A) Increasing doses of MNNG induce increasing amounts of transcription errors. The orange datapoint indicates untreated cells, while the blue datapoints indicate treated cells. (B) After a single exposure to MNNG, cells experience prolonged transcriptional mutagenesis in transcripts derived from RNAPI, III, and mtRNAP. In contrast, the error rate of transcripts synthesized by RNAPII declines over time. (C) The DNA repair protein MGT1 is required for the recovery of the error rate of RNAPII and partially prevents increased error rates in RNAPI. ns indicates no significant difference. * denotes P < 0.05, unpaired two-tailed t test; error bar indicates SEM. For all samples, n = 3–4 biological replicates, except for the 2- and 6-h data points for mgt1Δ cells, for which n = 2.

Fig. 4.

The kinetics of transcriptional mutagenesis can be exploited to monitor DNA repair. (A–C) Observed C→U error rate as a function of neighboring nucleotide on the DNA template strand. The 5′ base flanking an O⁶-methyl-guanine lesion dictates the error rate of transcription. However, the 3′ neighbor is more important for the efficiency of DNA repair, as the error rate of bases flanked by a 3′ cytosine or adenine reduces faster than those flanked by a guanine or a thymine. The numbers above the bar graph indicate the average fold-difference between bases that are flanked on their 3′ side by a guanine and thymine, compared to bases that are flanked by adenine and cytosine, all of which have a guanine on their 5′ side. (D and E) Nucleosomes do not shield DNA from MNNG mutagen exposure and can accelerate DNA repair. (F–I) Error rate of transcription in undamaged cells. Several markers of inactive genes, as well as the transcription rate, correlate with increased error rates. Triangles (F–H) indicate increasing abundance of respective markers. The x axis of I indicates bins of increasing transcription levels. Bin 1 contains the genes with the lowest transcription levels (the lowest 30%), and bin 8 contains genes with the highest transcription levels (the highest 2%). (1 = 0–30%, 2 = 30–50%, 3 = 50–70%, 4 = 70–80%, 5 = 80–90%, 6 = 90–95%, 7 = 95–98%, and 8 = 98–100%). This binning pattern is based on our coverage of the yeast transcriptome and was constructed so that each bin contains an equal amount of sequenced bases. (J) The top 10% of transcribed genes display the lowest DNA repair rate after treatment with MNNG. * denotes P < 0.05, unpaired two-tailed t test; error bar indicates SEM. For all samples, n = 3–13 biological replicates, except for the 2- and 6-h data points for mgt1Δ cells, for which n = 2.