Genome-wide mapping of transcriptional start sites defines an extensive leaderless transcriptome in Mycobacterium tuberculosis

Abstract

Deciphering physiological changes that mediate transition of Mycobacterium tuberculosis between replicating and nonreplicating states is essential to understanding how the pathogen can persist in an individual host for decades. We have combined RNA sequencing (RNA-seq) of 5' triphosphate-enriched libraries with regular RNA-seq to characterize the architecture and expression of M. tuberculosis promoters. We identified over 4,000 transcriptional start sites (TSSs). Strikingly, for 26% of the genes with a primary TSS, the site of transcriptional initiation overlapped with the annotated start codon, generating leaderless transcripts lacking a 5' UTR and, hence, the Shine-Dalgarno sequence commonly used to initiate ribosomal engagement in eubacteria. Genes encoding proteins with active growth functions were markedly depleted from the leaderless transcriptome, and there was a significant increase in the overall representation of leaderless mRNAs in a starvation model of growth arrest. The high percentage of leaderless genes may have particular importance in the physiology of nonreplicating M. tuberculosis.

Graphical abstract

Figure 1

Mapping of TSSs in M. tuberculosis (A) Genome visualization shows the transcriptional map of M. tuberculosis. Moving from the outer to innermost ring, forward genes are indicated in green and reverse genes in purple; TSS mapping is in orange, and whole-transcriptome expression is in light blue. Circular map was generated using Circos (Krzywinski et al., 2009). (B) The 50 bp upstream sequences of the 1,779 primary TSSs detected were used for motif discovery using MEME. A conserved −10 sequence with consensus TANNNT was found in 73% of the upstream promoter sequences; 7% of these had an extended −10 motif of TGNTANNNT. (C) The extended −10 motif (TGNTANNNT) was associated with maximal promoter activity measured by TSS peak height. Box plots indicate median (horizontal line), interquartile range (box), and minimum and maximum values (whiskers). Statistically significant differences are indicated for p values of ^∗p < 0.05, ^∗∗p < 0.01, and ^∗∗∗p < 0.001. (D) Base preference at the transcription initiation start point is presented. The percent representation of each base is shown for positions −1, +1, and +2 among the 1,779 primary TSSs. See also Figure S1 and Table S1.

Figure 2

Leaderless Genes Are Expressed at a Lower Level during Exponential Growth (A) and (C) show box plots indicating median (horizontal line), interquartile range (box), and minimum and maximum values (whiskers). Statistically significant differences are indicated (^∗∗∗p < 0.001). (A) RNA levels (RPKM values) during exponential growth across leaderless (L), Shine-Dalgarno (SD), and UTR genes (identified on the basis of a primary TSS or operon organization) are shown. See also Figure S2 for a similar analysis using more stringent criteria for identification of Shine-Dalgarno genes. (B) Correlation between promoter expression (primary TSS strength) and the associated downstream gene (RPKM values) for leaderless (orange) and leader (gray) genes is shown. (C) Protein levels (measured as ion counts) during exponential growth across leaderless, Shine-Dalgarno, and UTR genes are shown. y Axis represents ion counts rescaled by dividing by 10⁶ in order to reduce axis values. (D) Correlation between protein abundance and mRNA expression is presented. The plot shows an overlay of genes encoded on leaderless (orange), Shine-Dalgarno (blue), and UTR (gray) transcripts. Plots for individual transcript categories are included in Figure S2. See also Figure S2 and Tables S1 and S4.

Figure 3

Leaderless and Shine-Dalgarno Transcripts Are Differentially Represented among Different Functional Gene Classes The distribution of genes encoded by leaderless and Shine-Dalgarno transcripts across different functional classes was compared among the expressed transcriptome representation. Values on the x axis represent a difference in percentage; positive values indicate overrepresentation of a particular functional class with respect to whole-transcriptome representation, whereas negative values indicate underrepresentation. All functional categories shown were statistically significant after chi-square test analyses, and patterned bars further denote the functional categories that remained significant after multiple testing correction. ND, no difference. See also Table S1.

Figure 4

Differential Expression of Leaderless mRNAs in Response to Starvation (A) There was a significant increase in the median level of expression of leaderless mRNAs after 24 hr starvation measured by RPKM for whole genes. (B)–(D) show Artemis traces illustrating TSS mapping for genes strongly upregulated in the starvation model. Bars record the normalized number of mapped reads; the maximum normalized read count is indicated on top of each bar. The position of the TSS corresponds to the left-hand edge of the bar for transcripts in the forward orientation (shown in blue) and to the right-hand edge for transcripts in the reverse orientation (red). In each panel, the top trace is from exponential growth, and the lower trace shows TSS profiles after 24 hr starvation. The genome location is shown at the bottom of each panel. (B) Transcription of sigE is driven by three TSSs in exponential phase; only the middle TSS—generating a leaderless mRNA—is upregulated in response to starvation. (C) Upregulation of lat (Rv3289c) is accompanied by upregulation of an antisense transcript to lrpA (Rv3290c). Closely juxtaposed divergent promoters are a common feature of the M. tuberculosis transcriptome. (D) PrpD (Rv1130) and its adjacent divergently oriented regulator (Rv1129c) are both upregulated in response to starvation; insets show mapping of TSS to Met 26 and Val 9 start codons, generating leaderless mRNAs. (E) Bar charts indicate the percentage of proteins with abundance ratios (starvation/exponential growth) greater than 1.1 and less than 0.9 for leaderless, Shine-Dalgarno, and UTR categories. ns, not significant. See also Tables S1, S4, S5, and S6.