Mycobacterial RNase E cleaves with a distinct sequence preference and controls the degradation rates of most Mycolicibacterium smegmatis mRNAs

Abstract

The mechanisms and regulation of RNA degradation in mycobacteria have been subject to increased interest following the identification of interplay between RNA metabolism and drug resistance. Mycobacteria encode multiple ribonucleases predicted to participate in mRNA degradation and/or processing of stable RNAs. RNase E is hypothesized to play a major role in mRNA degradation because of its essentiality in mycobacteria and its role in mRNA degradation in gram-negative bacteria. Here, we defined the impact of RNase E on mRNA degradation rates transcriptome-wide in the nonpathogenic model Mycolicibacterium smegmatis. RNase E played a rate-limiting role in degradation of the transcripts encoded by at least 89% of protein-coding genes, with leadered transcripts often being more affected by RNase E repression than leaderless transcripts. There was an apparent global slowing of transcription in response to knockdown of RNase E, suggesting that M. smegmatis regulates transcription in responses to changes in mRNA degradation. This compensation was incomplete, as the abundance of most transcripts increased upon RNase E knockdown. We assessed the sequence preferences for cleavage by RNase E transcriptome-wide in M. smegmatis and Mycobacterium tuberculosis and found a consistent bias for cleavage in C-rich regions. Purified RNase E had a clear preference for cleavage immediately upstream of cytidines, distinct from the sequence preferences of RNase E in gram-negative bacteria. We furthermore report a high-resolution map of mRNA cleavage sites in M. tuberculosis, which occur primarily within the RNase E-preferred sequence context, confirming that RNase E has a broad impact on the M. tuberculosis transcriptome.

Keywords: Mycobacterium smegmatis; Mycobacterium tuberculosis; Mycolicibacterium smegmatis; RNA degradation; RNA processing; RNase E.

Figure 1

Knockdown of rne expression causes growth cessation and altered transcript abundance in Mycolicibacterium smegmatis.A, promoter replacement strategy to construct a strain in which rne expression is repressed by addition of ATc. B, rne transcript levels were reduced in the repressible rne strain following 3 h of exposure to ATc. ∗∗∗∗p < 0.001, two-tailed t test. C, growth of the repressible rne strain slowed approximately 15 h after addition of ATc. D, 8 h after addition of ATc or vehicle, rifampicin was added to block new transcription, and mRNA levels of the indicated genes were measured at several time points by qPCR to determine their half-lives. ∗∗p < 0.01, pair-wise comparisons by linear regression. ATc, anhydrotetracycline; qPCR, quantitative PCR.

Figure 2

Knockdown of rne expression causes stabilization of most of the Mycolicibacterium smegmatis transcriptome, with leadered transcripts tending to be stabilized more than leaderless transcripts. Eight hours after addition of ATc (or vehicle) to knock down (or not) rne, rifampicin was added to block new transcription, and mRNA levels were measured transcriptome wide at several time points by RNA-Seq to determine half-lives. A, dots represent transcripts with measurable half-lives in both conditions. B, the distribution of fold change in half-life for the transcripts shown in A. C, the median fold change in half-life upon rne knockdown was higher for leadered transcripts than for leaderless transcripts (left). The median abundance of leadered transcripts was higher prior to rne knockdown (right). D, only transcripts with 10<log₂ abundance<14 were considered, which reversed the difference in abundance trend between leadered and leaderless transcripts. The median fold change in half-life upon rne knockdown was still higher for leadered transcripts than for leaderless transcripts. ATc, anhydrotetracycline.

Figure 3

Knockdown of rne impacts mRNA abundance both directly and indirectly.A, each dot represents a gene for which log₂ fold change in transcript abundance upon rne repression is shown as a function of log₂ fold change in half-life. The solid line shows the linear regression fit where y = 0.2350 ∗ x + 0.6406. The dashed line shows the expected relationship between log₂ fold change half-life and log₂ fold change abundance if transcription rate were unchanged. B, estimated transcription rates were calculated from the measured mRNA half-lives and steady-state abundance. The same genes shown in A are shown here. C, for each gene, the expected change in abundance was calculated as a function of change in half-life according to the equation in A. The differences between expected and observed changes in abundance were then calculated, and genes with large differences were considered more likely to be subject to active regulation. Gene set enrichment analysis was performed on the observed/expected log₂ fold change abundance, and the gene categories with statistically significant enrichment or depletion are shown. Genes in the categories with positive enrichment scores had larger than expected increases in transcript abundance, and genes in the categories with negative enrichment scores had lower than expected increases (or had decreases) in transcript abundance. The q value is a p value corrected for multiple comparisons.

Figure 4

Cytidines are enriched in regions of RNase E-dependent mRNA cleavage in both Mycolicibacterium smegmatis and Mycobacterium tuberculosis.A, overview of the method for relative quantification of mRNA cleavage events using standard RNA-Seq data. Illumina RNA-Seq data for both M. smegmatis (this work) and M. tuberculosis (16) were used. Both datasets included an rne knockdown condition and multiple control conditions. Cleavage events result in a lower proportion of reads in the immediate vicinity of the cleavage site compared with uncleaved regions of the transcript. Read depth (coverage) for each coordinate within each coding sequence across the genome was determined for each sample, then normalized by the average read depth within that gene in that sample. For each coordinate, the log₂ ratio of coverage in the rne knockdown compared with a control (or two distinct controls compared with each other) was determined. The median log₂ ratio should be approximately zero for all comparisons because of the method of normalization. Coordinates at or near RNase E cleavage sites are expected to have high ratios in the rne knockdown/control comparison. The regions surrounding coordinates with log₂ ratios in the top 5% and middle 5% were then assessed for base composition bias (A, U, C, and G frequency). The bases at each position within 20 coordinates upstream and downstream of coordinates of interest (those having log₂ ratios in the middle 5% or top 5%) were determined. B, Log₂ ratios from the M. smegmatis control strain in the presence and absence of ATc, which is not expected to affect RNase E activity. C, the base frequencies in 41-nt regions centered on coordinates with log₂ ratios in the middle 5% or top 5% of the distribution shown in B. D, Log₂ ratios from the M. smegmatis-repressible rne strain in the +ATc condition (rne repressed) versus the no-ATc condition (rne expressed). E, the base frequencies in 41-nt regions centered on coordinates with log₂ ratios in the middle 5% or top 5% of the distribution shown in D. Coordinates with log₂ ratios in the top 5% are expected to be enriched for RNase E cleavage site–containing regions. F, Log₂ ratios from two M. tuberculosis strains that are expected to have similar RNase E activity (a WT strain and a strain expressing a CRISPRi system with a nontargeting sgRNA). G, the base frequencies in 41-nt regions centered on coordinates with log₂ ratios in the middle 5% or top 5% of the distribution shown in F. H, Log₂ ratios from an M. tuberculosis strain expressing an sgRNA to knock down expression of rne versus a strain with a nontargeting sgRNA. I, the base frequencies in 41-nt regions centered on coordinates with log₂ ratios in the middle 5% or top 5% of the distribution shown in H. Coordinates with log₂ ratios in the top 5% are expected to be enriched for RNase E cleavage site–containing regions. ATc, anhydrotetracycline; sgRNA, single guide RNA.

Figure 5

RNase E cleaves 5′ of cytidines in vitro.A, SYBR Gold–stained Tris–borate–EDTA–urea gels revealing cleavage of the RNA substrate shown in B (300 ng) upon incubation for 1 h with 80 ng purified and recombinant Mycolicibacterium smegmatis RNase E catalytic domain (residues 146–824, with an N-terminal FLAG-His tag). The D694R, D738R mutant is predicted to be catalytically dead. Untreated RNA was not incubated with reaction buffer, whereas mock reactions contained RNA and buffer in the absence of enzyme. Cleavage sites are numbered 1 to 4, and resulting fragments visible on the gel are designated with numerals i–iv, as shown schematically in B. Red arrows denote cleavage fragments. Blue arrows indicate the longer strand of the partial-duplex substrate or the annealed substrate, and purple arrows indicate the shorter strand of the partial-duplex substrate. Note that while all samples were heated with formamide prior to loading the gels, the partial-duplex substrate did not fully denature following incubation with reaction buffer in the mock reaction or enzyme-containing reactions. B, schematic (not to scale) of the partial duplex RNA substrate used in A. Red arrows indicate RNase E cleavage sites mapped by 5′ or 3′ RACE on cleavage products extracted from the gels shown in A. The thin lines indicate the sizes of the extracted cleavage products (not to scale), with red dots indicating the ends that were mapped by RACE. C, a 50 nt region of the single-stranded portion of the RNA substrate shown in A was synthesized. The expected products from cleavage at sites 1 and/or 2 are shown. The bolded “C” was mutated to G in E. D, RNase E cleavage reactions using the substrate shown in C and enzyme that was repurified using a more stringent wash protocol to remove contaminating E. coli RNases. Reactions contained 80 ng RNase E and 150 ng of RNA and were incubated for 2 h. The expected cleavage products shown in C are indicated with red arrows. Black arrows indicate the positions of MW standards. The blue arrow indicates the full-length substrate. E, cleavage reactions were done as in D with the addition of a substrate with a C to G mutation at the position 3′ of cleavage site 2 (indicated with a red “G”). The indicated MW standards were combined in the first lane. Bands labeled “nonspecific” are unidentified byproducts of the MW standard synthesis reactions. Red arrows denote the expected cleavage products observed in D. Orange arrows indicate bands that appeared or shifted in position when the substrate had the C to G mutation at cleavage site 2. All gels are representative of at least three independent experiments. MW, molecular weight; RACE, rapid amplification of complementary DNA ends.

Figure 6

A transcriptome-wide mRNA cleavage site map in Mycobacterium tuberculosis reveals sequence and secondary structure preferences consistent with RNase E and greater cleavage site frequency in 5′ UTRs and intergenic regions.A, WebLogo (3.7.4) generated from the complete set of mapped M. tuberculosis cleavage sites aligned by cleavage site position. Cleavage occurs between positions −1 and 1 as indicated by the scissor icon. B, RNA cleavage typically occurs within regions of lower secondary structure. The minimum free energy secondary structure was predicted for sliding 39 nt windows across 200 nt of sequence spanning each RNA cleavage site. For each coordinate, the mean (solid line; interquartile range, dashed lines) predicted free energy (ΔG) of secondary structure formation of all 2983 cleaved RNAs was determined. C, the frequencies of RNA cleavage sites in various genomic regions were determined: coding sequences, 5′ UTRs, and between adjacent genes on the same strand. Regions between genes on the same strand were separated according to whether the gene pair was predicted to be transcribed in an exclusively operonic (polycistronic) fashion. Gene pairs were considered to be transcribed exclusively in operons if only the first gene had a mapped transcription start site (TSS). Gene pairs were considered to be not transcribed exclusively in operons if each gene had its own TSS. In the latter case, genes may be transcribed as a mixture of monocistronic and polycistronic transcripts. 5′ UTRs were included only if the next upstream gene was on the opposite strand. The observed frequencies of cleavage sites in each region were compared with the frequencies that would be expected if cleavage sites were distributed among these regions without bias. ∗∗∗∗p < 0.0001 by binomial test comparing the observed versus expected frequencies.