Toxin-mediated ribosome stalling reprograms the Mycobacterium tuberculosis proteome

Abstract

Mycobacterium tuberculosis readily adapts to survive a wide range of assaults by modifying its physiology and establishing a latent tuberculosis (TB) infection. Here we report a sophisticated mode of regulation by a tRNA-cleaving toxin that enlists highly selective ribosome stalling to recalibrate the transcriptome and remodel the proteome. This toxin, MazF-mt9, exclusively inactivates one isoacceptor tRNA, tRNA^Lys43-UUU, through cleavage at a single site within its anticodon (UU↓U). Because wobble rules preclude compensation for loss of tRNA^Lys43-UUU by the second M. tuberculosis lysine tRNA, tRNA^Lys19-CUU, ribosome stalling occurs at in-frame cognate AAA Lys codons. Consequently, the transcripts harboring these stalled ribosomes are selectively cleaved by specific RNases, leading to their preferential deletion. This surgically altered transcriptome generates concomitant changes to the proteome, skewing synthesis of newly synthesized proteins away from those rich in AAA Lys codons toward those harboring few or no AAA codons. This toxin-mediated proteome reprogramming may work in tandem with other pathways to facilitate M. tuberculosis stress survival.

Fig. 1

MazF-mt9 targets only tRNA^Lys43-UUU in vivo. a Histogram representing the ratio of cleavage by MazF-mt9 identified using 5′ RNA-seq at each nucleotide within the lysT gene (tRNA^Lys43-UUU) in M. tuberculosis H37Rv after 7 days of toxin induction. Genomic positions and the negative strand sequence are shown. b Representation of the only MazF-mt9 target, tRNA^Lys43-UUU (adapted from Schifano et al.). Anticodon (UUU), in red; cleavage site, yellow arrow. c Northern analysis of RNA from ±MazF-mt9 M. tuberculosis cells using the tRNA isoacceptor-specific oligonucleotides indicated. Uncropped Northern blot images are provided as a tab within the Source Data file

Fig. 2

5′ RNA-seq captures ribosome stalling events at AAA Lys codons. a Top mRNA hits detected by 5′ RNA seq, ±25 nts surrounding the cleavage site (scissor). The first nucleotide of each mapped read, orange; AAA Lys codons, green. Counts are represented as reads per million (rpm). b Nucleotide frequency logo (weblogo, see ref. ) at each position constructed from the top 50 of the 181 mRNA hits identified. Positions numbered relative to the cleavage site. c Schematic of a ribosome (gray) and the ~15 nt length (yellow arrow) from the 5′ end of transcript to the A-site. Source data are provided in Supplementary Data 1

Fig. 3

Ribo-seq confirms transcriptome-wide AAA Lys-specific ribosome stalling. a, b, e, f Ribo-seq read coverage in reads per million (rpm) of AAA Lys-containing M. smegmatis (a, b) or M. tuberculosis mc² 6206 (e, f) transcripts in ±MazF-mt9 cells (arrows indicate direction of translation). The location of in-frame AAA Lys codons and out-of-frame AAAs are indicated (red arrow/gray box and blue arrow head, respectively). Scale bars illustrate a 100 bp genomic distance. c, g, Average read counts (rpm) of mapped ribosomes surrounding AAA Lys codons in +MazF-mt9 (red line) or -MazF-mt9 (blue) cells; light shading represents the standard deviation. d, h Ratio of codon occupancy (MazF-mt9 induced vs. uninduced) at the predicted ribosomal A-site position identified in the mapped ribosomal footprints, grouped by amino acid or stop codon. Source data for d and h are provided in a separate tab within the Source Data file

Fig. 4

Ribo-seq shows ribosome stalling preference for early AAA codons. a–d Average read counts (reads per million, rpm) of mapped ribosome footprints surrounding the first (a), second (b), third (c), or fourth and higher (d) AAA Lys codon of M. smegmatis genes. Blue or red solid line represents uninduced cells or cells expressing MazF-mt9, respectively, with the light-colored shading denoting standard deviation

Fig. 5

M. tuberculosis RNase J cleaves 5′ of stalled ribosomes favoring expression of AAA-deficient transcripts. KpLogo graphs derived from E. coli 5′ RNA-seq + MazF-mt9 (a) or +MazF-mt9 + RNase J (b) datasets compared to -MazF-mt9 samples. The enriched k-mers in the datasets are stacked vertically at the position of their initial nucleotide. The positions are numbered relative to the cleavage site (dotted line), with those achieving statistical significance (Bonferroni corrected p-value < 0.01) shown in red

Fig. 6

Induction of MazF-mt9 leads to global proteomic shifts based on the AAA codon content of transcripts. Volcano plot showing the proteomic changes in newly synthesized proteins between induced (+MazF-mt9) and uninduced cells. The diameter and color intensity of each represented protein (circle) is proportional to the number of AAA codons in its coding sequence. Fold changes in mRNA levels obtained by RNA-seq are shown for three examples of upregulated and downregulated proteins (arrows). The source data with full list of detected proteins is presented in Supplementary Data 2

Fig. 7

mRNA levels correlate with protein expression. Heatmaps comparing fold changes in newly synthesized proteins (AHA-labeling Proteomics, top row) and mRNA levels (RNA-Seq, middle row) in ±MazF-mt9 cultures. Genes with at least a log2ratio of ±1 were considered. The number of AAA codons (bottom row) of each gene is shown. Source data are provided as a tab within the Source Data file

Fig. 8

MazF-mt9 translation inhibition is codon specific. MazF-mt9 was co-expressed with two different versions of mCherry: one containing its lysine codons mutated to AAA (AAA mCherry) and the other to AAG (AAG mCherry). mCherry fluorescence was measured 5 h after induction of MazF-mt9 and normalized against an uninduced (−MazF-mt9) control. Each individual data point is derived from a biological replicate (n = 3). Box plot center line, median; box limits, upper and lower quartiles; whiskers, max and min values

Fig. 9

MazF-mt9 mechanism of action. The toxin controls gene expression through the cleavage of tRNA^Lys43-UUU (left panel), causing ribosome stalling and subsequent degradation of the mRNAs by one or more RNases (middle panel). This cooperative action reduces the steady levels of AAA-rich proteins and favors an increase in AAA-deficient proteins (right panel)