Genome-Wide Spectra of Transcription Insertions and Deletions Reveal That Slippage Depends on RNA:DNA Hybrid Complementarity

Abstract

Advances in sequencing technologies have enabled direct quantification of genome-wide errors that occur during RNA transcription. These errors occur at rates that are orders of magnitude higher than rates during DNA replication, but due to technical difficulties such measurements have been limited to single-base substitutions and have not yet quantified the scope of transcription insertions and deletions. Previous reporter gene assay findings suggested that transcription indels are produced exclusively by elongation complex slippage at homopolymeric runs, so we enumerated indels across the protein-coding transcriptomes of Escherichia coli and Buchnera aphidicola, which differ widely in their genomic base compositions and incidence of repeat regions. As anticipated from prior assays, transcription insertions prevailed in homopolymeric runs of A and T; however, transcription deletions arose in much more complex sequences and were rarely associated with homopolymeric runs. By reconstructing the relocated positions of the elongation complex as inferred from the sequences inserted or deleted during transcription, we show that continuation of transcription after slippage hinges on the degree of nucleotide complementarity within the RNA:DNA hybrid at the new DNA template location.IMPORTANCE The high level of mistakes generated during transcription can result in the accumulation of malfunctioning and misfolded proteins which can alter global gene regulation and in the expenditure of energy to degrade these nonfunctional proteins. The transcriptome-wide occurrence of base substitutions has been elucidated in bacteria, but information on transcription insertions and deletions-errors that potentially have more dire effects on protein function-is limited to reporter gene constructs. Here, we capture the transcriptome-wide spectrum of insertions and deletions in Escherichia coli and Buchnera aphidicola and show that they occur at rates approaching those of base substitutions. Knowledge of the full extent of sequences subject to transcription indels supports a new model of bacterial transcription slippage, one that relies on the number of complementary bases between the transcript and the DNA template to which it slipped.

Keywords: Buchnera; Escherichia coli; RNA polymerase; elongation complex; error rates; mutation; transcription; transcription slippage.

FIG 1

Characteristics of transcription errors in bacterial genomes. (A) Rates of transcription insertions, deletions, and base substitutions in E. coli and Buchnera. Frequencies of each type of transcription error were computed for the same eight replicate samples of E. coli and for the same two replicate samples of Buchnera. (B) Error frequencies of Buchnera transcription insertions, Buchnera transcription deletions, and E. coli transcription insertions in homopolymeric runs. The y-axes follow those of panel A. Each error type shown follows a natural exponential function (Buchnera_insertions, r² = 0.739, P < 0.004; Buchnera_deletions, r² = 0.981, P < 0.001; E. coli_insertions, r² = 0.894, P < 0.003). (There were too few E. coli transcription deletions to test for this trend.) (C) Length distribution of transcription insertions and deletions in E. coli and Buchnera.

FIG 2

Compositional biases of transcription deletions. (A) Nucleotide composition of transcription deletions in E. coli (black) compared to that expected based on the nucleotide composition of all transcribed sequences (white) in each replicate. Comparisons were performed using pairwise Wilcoxon tests and data were subjected to the Benjamini-Hochberg correction. *, P < 0.05; ***, P < 0.001. (B) Nucleotide composition of transcription deletions in Buchnera (gray) compared to that expected based on the nucleotide composition of all transcribed sequences (white) in each replicate. Comparisons were performed using a Student’s t-test. The y-axes follow those in panel A. (C) Average proportion of each nucleotide at each of the 15 bases preceding and the 15 bases succeeding E. coli transcription deletions and in deletions as a whole (shaded gray region). Significant biases in nucleotide frequencies occur in the four bases before a deletion and one base after a deletion. (D) Dinucleotide frequencies of the two bases preceding transcription deletions in E. coli (black) compared to that expected based on the nucleotide composition of all transcribed sequences (white). (E) Dinucleotide frequencies of the two bases preceding transcription deletions in Buchnera (gray) compared to that expected based on the nucleotide composition of all transcribed sequences (white). Comparisons in panels C, D, and E were made using the Fisher exact test and data were subjected to the Benjamini-Hochberg correction. *, P < 0.05; ***, P < 0.001.

FIG 3

Dependence of transcription deletions on sequence complementarity in the RNA:DNA hybrid. (A) The 9-base RNA:DNA hybrids were reconstructed for transcription deletions (black rings) and for expected deletions based on the nucleotide composition of all transcribed sequences (white rings) in E. coli, and the extent of complementarity between the region preceding the end of a deletion and the RNA:DNA hybrids was computed. (B) The 9-base RNA:DNA hybrids were reconstructed for transcription deletions (gray rings) and for deletions expected based on the nucleotide composition of all transcribed sequences (white rings) in Buchnera, and the extent of complementarity between the region preceding the end of a deletion and the RNA:DNA hybrids was computed. Due to the small sample size with Buchnera, only 8 and 9 bases of RNA:DNA hybrid complementarity were significant. For both organisms, there were significant deviations from expectation (chi-square test, P < 0.001), indicating that transcription slippage is more likely to stop at regions of higher base complementarity than expected. Comparisons of the extent of RNA:DNA complementarity were performed using the Fisher exact test and data were subjected to the Benjamini-Hochberg correction. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

FIG 4

Transcription deletions in short sequence repeats. (A) Proportions of transcription deletions between 1 and 4 nucleotides in length occurring within repetitive sequences in E. coli. In all cases, deletion lengths correspond to the length of the repeat unit within a repetitive sequence, and there is a minimum of two repeat units for a sequence to be considered repetitive. (The wide error bars in single nucleotide deletions result from replicates with few or no deletions of that length.) (B) Proportions of deletions with successive complementary bases in the 3′ end of the RNA:DNA hybrid after slippage. Deletions of all lengths were included in this analysis. Comparisons in panels A and B were performed with pairwise Wilcoxon tests (n = 8 for each test), and data were subjected to the Benjamini-Hochberg correction. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

FIG 5

Model of transcription slippage resulting in deletions. Based on locations and sequence contents of deletions genome-wide, the degree of complementarity of the RNA:DNA hybrid after a transcription slippage event (I and II) determines whether transcription is aborted, producing a truncated transcript (III and IV), or resumed, producing a transcript containing a deletion (V and VI). Steps in the model use the following notations: template DNA is shown in black, transcript RNA and incoming ribonucleotides are in blue, the original RNA:DNA hybrid location is orange, the nontranscribed (i.e., deleted) region is shown in red, mismatched bases are shown as angled contacts between noncomplementary nucleotides, and the RNA polymerase transcription elongation complex is represented by a yellow bubble. In this model, normal transcription (I) becomes interrupted when the elongation complex and transcript lose register with the DNA template (II). Possible outcomes include the elongation complex slipping forward to a region of low complementarity (III), and in the example depicted, the elongation complex slips forward 5 bases, landing on a template location where 6 of the 9 bases in the RNA:DNA hybrid are not complementary. If transcription cannot resume due to the extent of mispairing in the RNA:DNA hybrid and/or fraying at the end of the transcript, the transcript is aborted (IV). Alternatively, if the elongation complex slips to template location with fewer mismatches (V), the 3′ end of the RNA bonds sufficiently to the DNA template, and transcription resumes (VI) after the skipped the region, generating a deletion.