Effect of tRNA Maturase Depletion on Levels and Stabilities of Ribosome Assembly Cofactor and Other mRNAs in Bacillus subtilis

The passage of ribosomes across individual mRNAs during translation can have different effects on their degradation, ranging from a protective effect by shielding from ribonucleases to, in some cases, making the mRNA more vulnerable to RNase action. We recently showed that some mRNAs coding for proteins involved in ribosome assembly were highly sensitive to the availability of functional tRNA. ABSTRACT The impact of translation on mRNA stability can be varied, ranging from a protective effect of ribosomes that shield mRNA from RNases to preferentially exposing sites of RNase cleavage. These effects can change depending on whether ribosomes are actively moving along the mRNA or stalled at particular sequences or structures or awaiting charged tRNAs. We recently observed that depleting Bacillus subtilis cells of their tRNA maturation enzymes RNase P and RNase Z led to altered mRNA levels of a number of assembly factors involved in the biogenesis of the 30S ribosomal subunit. Here, we extended this study to other assembly factor and non-assembly factor mRNAs in B. subtilis. We additionally identified multiple transcriptional and translational layers of regulation of the rimM operon mRNA that occur in response to the depletion of functional tRNAs. IMPORTANCE The passage of ribosomes across individual mRNAs during translation can have different effects on their degradation, ranging from a protective effect by shielding from ribonucleases to, in some cases, making the mRNA more vulnerable to RNase action. We recently showed that some mRNAs coding for proteins involved in ribosome assembly were highly sensitive to the availability of functional tRNA. Using strains depleted of the major tRNA processing enzymes RNase P and RNase Z, we expanded this observation to a wider set of mRNAs, including some unrelated to ribosome biogenesis. We characterized the impact of tRNA maturase depletion on the rimM operon mRNA and show that it is highly complex, with multiple levels of transcriptional and posttranscriptional effects coming into play.

gene and a protein subunit encoded by the rnpA gene (4). Their 39 ends are matured either by the endo/exoribonuclease RNase Z or by a number of redundant 39 exoribonucleases, depending nominally on whether the tRNA gene carries the CCA motif required for aminoacylation (5). RNase Z processes the one-third of B. subtilis tRNAs lacking an encoded CCA motif through stimulation of its endoribonuclease activity about 200-fold by a uracil residue that naturally occurs #2 nucleotides (nt) downstream of the so-called discriminator nucleotide (nt 75) of each of these tRNA precursors (6). The enzyme's 39exoribonuclease activity is required to trim back to nt 75 to allow addition of the CCA motif by nucleotidyltransferase (NTase, or CCAse). Both RNase P and Z are essential in B. subtilis, and depletion of either enzyme inhibits cell growth, presumably due to a lack of functional tRNAs for translation.
Translation can also be inhibited by antibiotics that target the ribosome, such as chloramphenicol (Cm), which targets the peptidyltransferase center (PTC) located on the large ribosomal subunit. Although it was once thought that Cm blocked translation randomly, recent ribosome profiling experiments have shown that Cm preferentially causes ribosomes to stall at particular sites, in particular when alanine (Ala) or serine (Ser) residues have just been incorporated into the nascent peptide (7).
We recently showed that depletion of tRNA maturase activity affects ribosome assembly, leading to a specific 30S subunit late assembly defect (8). While this defect was mostly explained by a RelA-dependent accumulation of the stringent response alarmone (p)ppGpp and inhibition of GTP-dependent assembly factor activity, we also observed that the levels of several mRNAs encoding ribosome assembly cofactors were affected. Notably, the steady-state levels of transcripts encoding the GTPases Era and YqeH were upregulated during tRNA maturase depletion, whereas mRNAs encoding the GTPase CpgA and the RNA chaperone RimM were downregulated. Because RNase P is thought to have very few direct mRNA targets, and because RNase Z depletion had comparable effects on the expression of these mRNAs, we considered it unlikely that the effects observed were directly due to RNase P or RNase Z cleavages in each of these mRNAs. We therefore wished to better understand by which mechanism(s) tRNA maturase depletion affected the levels of the cofactor encoding mRNAs. Since the late 30S ribosome assembly defect observed in tRNA maturase depletion strains was very similar to that observed in both E. coli and B. subtilis DrimM mutants, we put additional focus on exploring the decrease in rimM expression under these conditions.

RESULTS
tRNA maturase depletion alters assembly factor mRNA levels. We previously showed that depletion of RNase P or RNase Z results in altered mRNA levels of four key 30S assembly cofactors (Era, YqeH, RimM, and CpgA) (8). The effects of depleting the RNA subunit of RNase P (RnpB) were more severe than those of depleting the protein subunit (RnpA), presumably because the RNA component of RNase P is more rapidly depleted than the protein subunit once transcription is shut off. To ask whether this applied to other mRNAs involved in ribosome biogenesis, we extended this analysis to the expression of several other cofactor and ribosomal protein mRNAs, using xylose (Pxyl-rnpA) or IPTG (isopropyl-b-D-thiogalactopyranoside)-dependent (Pspac-rnpB and Pspac-rnz) promoter constructs to deplete the protein and RNA subunits of RNase P and RNase Z, respectively. The two control transcripts, yqeH and era, and three new transcripts, ydaF and yjcK (encoding two potential homologs of the Escherichia coli RimJ acetylase) and rpsU (encoding r-protein S21), were globally increased under conditions of tRNA maturase depletion (Fig. 1A), while rimM and cpgA (controls) and yfmL (encoding a DEAD box helicase) transcripts all showed decreased expression, with a visible accumulation of degradation intermediates for yfmL (Fig. 1B). Expression of the rbfA and ylxS/rimP mRNAs was relatively unchanged (Fig. 1C), showing that tRNA maturase depletion does not cause nonspecific perturbation of the expression of all B. subtilis ribosome assembly cofactor genes. Although the primary focus of this study was on assembly factor mRNAs because of the link to a defect in 30S biogenesis, we also asked whether effects of tRNA maturase depletion on other mRNAs could be seen. Indeed, mRNAs from the yrzI and bmrCD operons, encoding multiple peptides of unknown function and a multidrug resistance pump, respectively, were also upregulated upon RNase P or RNase Z depletion (Fig. 1D), suggesting that this phenomenon is not confined to mRNAs with ribosome-related functions.
tRNA maturase depletion and the translation inhibitor chloramphenicol alter mRNA stability in a similar manner. To determine whether tRNA maturase depletion impacted mRNA expression at the transcriptional or posttranscriptional level, we measured the stability of several of these mRNAs after rifampicin treatment in RNase P (RnpA or RnpB)-depleted cultures. The upregulated transcripts (yqeH, era, ydaF, and yjcK) were all stabilized during RnpA and RnpB depletion ( Fig. 2A and B; also, see Fig. S1 in the supplemental material), suggesting that they are affected by RNase P depletion at the posttranscriptional level. There is evidence that the lack of functional tRNAs can increase ribosome stalling on translated mRNAs (9). Thus, the increased stability of these transcripts could be due to ribosome stalling and the blocking of RNase access to cleavage sites on these mRNAs. To test this hypothesis, we sought to recapitulate the effect by pausing translation in a different manner, using the translation elongation inhibitor Cm. Indeed, the addition of a subinhibitory concentration (2.5 mg/mL) and the MIC (5 mg/ mL) of Cm to wild-type (WT) cells also increased the levels of the yqeH, era, ydaF, and yjcK mRNAs (Fig. 2C), suggesting that the stabilization of these transcripts in tRNA maturase depletion strains is most likely due to the lack of mature tRNA and ribosome stalling.
The situation with the downregulated transcripts was more complicated. The major rimM (5, 3, and 1.8 kb) and cpgA (5 and 4.5 kb) transcripts were strongly destabilized in RNase P RNA subunit (RnpB)-depleted cells ( Fig. 3B; Fig. S2), suggesting that the decrease in expression also occurs at a posttranscriptional level in this strain. A similar decrease in FIG 1 Depletion of tRNA processing enzymes results in perturbed expression of some mRNAs encoding proteins involved in 30S subunit assembly. Northern blots showing (A) up-regulated mRNAs, (B) down-regulated mRNAs, (C) unaffected mRNAs, and (D) mRNAs unrelated to ribosome assembly, present in total mRNA isolated in the presence or absence of inducer as indicated. Note that the basal level of the bmrCD transcript, encoding a multidrug transporter, is higher in the Pspac-rnz and Pspac-rnpB strains because of the presence of erythromycin in the medium for stable maintenance of the construct. 16S rRNA levels (ethidium bromide stained) are shown as a loading control. Series of blots where a single loading control is shown were stripped and reprobed. The membrane used for panel B is the same as in panel C. The blots for era, yqeH, rimM, and cpgA were regenerated as in reference 8 with independent RNA preparations, with permission granted by the publisher for reuse of previously published data. Numbers of repetitions are as follows: yqeH, 3; era, 3; rpsU, 4; ydaF, 2; yjcK, 2; rimM, 3; cpgA, 3; rbfA, 2; ylxS, 2; yfmL, 2; bmrCD, 2; yrzI, 2.
tRNA Maturase Impact on mRNA Levels and Stability Microbiology Spectrum expression was seen after 30 min at high Cm concentration (MIC) in WT cells for rimM and rapidly upon exposure to Cm for cpgA (Fig. 3C), suggesting that this phenomenon is also linked to ribosome stalling. For both cpgA and yfmL, the major transcripts were processed to shorter forms in the absence of RnpB or in the presence of Cm (Fig. 3B). One possibility is that, in contrast to the upregulated mRNAs, when nonfunctional tRNA  tRNA Maturase Impact on mRNA Levels and Stability Microbiology Spectrum precursors accumulate to high levels in the rnpB depletion strain, ribosomes eventually stall at sites that preferentially allow RNase access. Another possibility is that the RNAs are largely unoccupied by ribosomes.
In the less severely depleted rnpA strain, the full-length (5-kb) rimM transcript and the two major cpgA mRNAs were stabilized (or showed little effect), rather than being destabilized, as seen for rnpB (Fig. 3A). These results suggest that downregulation of rimM and cpgA arises from a mixture of transcriptional (down) and posttranscriptional (up initially, then down) effects and that one or the other effect predominates depending on the severity of RNase P depletion. Indeed, upon close inspection of Fig. 3C, it can be seen that rimM and cpgA mRNA levels initially increase at 15 min and then decrease after further exposure to Cm at both subinhibitory and MIC doses. Thus, the Cm effect globally tracks the effect of tRNA depletion, with the weak Cm dose (2.5 mg/mL) mimicking the weak effect of depleting RnpA and the strong Cm dose (5 mg/mL) mimicking the strong effect of depleting RnpB, consistent with the notion of opposing responses to severe versus less severe levels or duration of translation inhibition.
Identification of rimM-containing transcripts sensitive to RNase P depletion. Because the DrimM phenotype closely fitted the 30S late assembly defect observed in strains depleted of RNase P or RNase Z (8), we attempted to narrow down the determinants of the downregulation of this operon. The rimM gene is located in a large operon containing several genes encoding components of the translation machinery: ribosomal protein genes rpsP and rplS (encoding S16 and L19, respectively), signal recognition particle component genes (ffh and ylxM), and trmD (encoding a tRNA methyltransferase) (Fig. 4A). To identify the gene composition of the three rimM-containing transcripts, we performed Northern blots with probes located in open reading frames (ORFs) of the neighboring genes (Fig. S3). In all, six different transcripts originate from this locus (Fig. 4A). Promoters upstream of ylxM (P 1 ) and rplS (P 3 ), and terminators downstream of ylqC and rplS (T 1 and T 3 , respectively) were identified earlier by transcriptome analysis (10). Our Northern blot analysis suggested that two transcripts originate from P 1 : the full-length mRNA (5 kb) (Fig. 4A, highlighted in purple), which terminates at T 3 , and a shorter transcript (2.5 kb) (Fig. 4A, highlighted in orange) that terminates at T 1 and does not contain the rimM ORF. The smallest species identified (0.5 kb) (Fig. 4A, highlighted in yellow) corresponds to the monocistronic rplS transcript (P 3 to T 3 ). Using end enrichment RNA sequencing (Rend-seq), DeLoughery et al. identified a third transcription start site (P 2 ) located just upstream of rpsP and only 18 nt downstream of an RNase Y cleavage site in the ffh-rpsP intergenic region, in addition to a potential terminator/attenuator (T 2 ) within the trmD ORF (11). The three remaining transcripts (0.7, 1.8, and 3 kb [ Fig. 4A, highlighted in green, cyan, and pink, respectively, and marked with asterisks]) therefore correspond to either P 2 primary transcripts or RNase Y-processed transcripts originating from P 1 , which terminate at T 1 , T 2 , and T 3 , respectively. Interestingly, of the six transcripts encoded by this locus, only the three containing both the ylqD and rimM ORFs were downregulated upon RNase P depletion (Fig. 4B).
A determinant for downregulation of the rimM operon is located within the ylqD ORF. To further narrow down which ORF was responsible for downregulation of rimM operon expression, we subcloned the ylqD-rimM or rimM-only parts of the operon under the control of a Pspac promoter, rendered constitutive by deleting the lac operator [Pspac(con)] (Table S3), with an artificial terminator hairpin to provide a defined 39 end. The constructs were integrated into the chromosome at the amyE locus, and levels of the ectopic transcript were analyzed by Northern blotting in RNase P-depleted cells using a probe specific for rimM. The steady-state levels of the synthetic ylqD-rimM transcript were downregulated in response to RNase P depletion, albeit not as dramatically as the native operon (1.3-versus 2.3-fold), suggesting that a determinant involved in downregulation is still included in this shorter construct ( Fig. 5A and B). Two degradation intermediates (;0.5 and ;0.4 kb) of the ylqD-rimM transcript also accumulated, suggesting that this transcript is cleaved twice under conditions of RnpB depletion. It is possible that the weaker effect of RnpB depletion on the full-length transcript and the accumulation of visible degradation intermediates is explained by the presence of a stabilizing terminator hairpin at the 39 end of each of these species that is not present in the native mRNA. Intriguingly, the construct containing only the rimM ORF (with the same 39 terminator) was upregulated in response to RNase P depletion ( Fig. 5C and D). In combination, these results suggest that the region responsible for posttranscriptional downregulation of rimM-containing transcripts upon depletion of RNase P is located primarily within the ylqD ORF.
Since unprocessed tRNAs accumulate in RNase P-and RNase Z-depleted cells, we wondered whether they could act as potential posttranscriptional regulators of target mRNAs by base pairing to their targets via their single-stranded 59 and 39 extensions. Using TargetRNA2 (12), a prediction program used for identifying targets of small RNAs (sRNAs) in bacteria, we identified an 11-nt region within the ylqD ORF that could potentially base pair with the 59 immature extension of unprocessed trnD-Tyr tRNA (Fig. 5B). To test whether this sequence was involved in downregulation of the ylqD-rimM construct in cells depleted of RNase P, we weakened the putative base-pairing interaction by introducing mutations into the ylqD mRNA sequence (while maintaining the YlqD amino acid sequence as much as possible) (Fig. 5B). The mutant ylqD M -rimM construct was downregulated and processed similarly to the WT under conditions of RNase P depletion, suggesting that 59-extended trnD-Tyr does not act as a posttranscriptional regulator of this operon. For the moment, the sequence element(s) within ylqD responsible for downregulation of the rimM operon under conditions of tRNA maturase depletion remains unknown. Downregulation of rimM expression under physiological conditions resulting in reduced rnpA expression is independent of immature-tRNA accumulation. We next asked whether the downregulation of the rimM operon we observed during RNase P depletion would also occur under physiological conditions where RNase P expression is reduced. The level of expression of the rnpB RNA is relatively constant in tiling array experiments in over 100 conditions tested, whereas rnpA mRNA levels decrease upon ethanol addition and during stationary phase in both complex and minimal media (Fig. 6A) (10). We confirmed that rnpA RNA expression was reduced to levels below detection in these three conditions in comparison with exponential growth in the respective medium, by Northern blotting (Fig. 6B). Ethanol treatment did not affect rnpB RNA levels; however, they were reduced during stationary phase in both minimal and complex media, in contrast to the tiling array data. The expression of rimM varied similarly to that of rnpA under the conditions tested (Fig. 6B).
We next asked whether tRNA maturation was affected in stationary phase or upon addition of ethanol using a probe for trnJ-Lys tRNA. Surprisingly, despite the decreased levels of the rnpA mRNA in all three conditions, and of rnpB in stationary phase, we did not observe an accumulation of pre-tRNAs (Fig. 6C). It is possible that very few new tRNA molecules are synthesized under these conditions and/or that the remaining cellular RNase P activity provided by the stable RnpA protein and RnpB RNA is sufficient to ensure the processing of any that are transcribed. In either case, these experiments suggest that the downregulation of rimM expression that accompanies the decrease in rnpA and rnpB expression in stationary phase or during ethanol stress is more related to growth arrest than an accumulation of immature tRNAs.
Downregulation of rimM in RNase P depletion strains depends partially on (p) ppGpp production. In bacteria, both stationary phase and ethanol stress are associated with increased production of (p)ppGpp, hyperphosphorylated guanosine derivatives that are known to globally reprogram transcription (13,14). Considering that tRNA maturase-depleted cells also trigger a RelA-dependent production of (p)ppGpp (8), we asked whether rimM downregulation in these cells was dependent on (p)ppGpp production, by measuring rimM expression in (p)ppGpp 0 strains depleted of RnpA or RnpB. The (p)ppGpp 0 strain lacks the three genes encoding (p)ppGpp-synthesizing enzymes in B. subtilis (yjbM, ywaC, and relA) (15). If (p)ppGpp were the key mediator, we would expect the effect of RNase P depletion on rimM expression to be reduced or abolished in the (p) ppGpp 0 background. Rather than simply abolishing the effect, rimM transcripts actually showed higher levels in the tRNA maturase-depleted (p)ppGpp 0 strains compared to the RnpA-and RnpB-depleted strains capable of making (p)ppGpp (Fig. 7A), suggesting that (p)ppGpp has an independent repressive effect on rimM mRNA levels.
We thus assessed whether the alarmone (p)ppGpp could downregulate rimM expression in the absence of a tRNA processing defect using an engineered strain that allows us to produce (p)ppGpp in the absence of immature tRNA accumulation or nutrient starvation (8). This (p)ppGpp 1 strain consists of an ectopic copy of the ywaC gene placed under the control of a Pxyl promoter in the (p)ppGpp 0 strain background. We used derepression of the CodY-regulated ywaA mRNA as a proxy to follow the increase in (p) ppGpp levels in this strain in vivo (Fig. 7B) (8). We observed that (p)ppGpp induction alone had no effect on the small (Y/P 2 -T 2 ) (Fig. 7B, cyan dot) rimM transcript, whereas the two larger species (Y/P 2 -T 3 and P 1 -T 3 ) (Fig. 7B, pink and purple dots, respectively) were downregulated as the expression of the (p)ppGpp reporter ywaA increased (Fig. 7B). Although (p)ppGpp production alone recapitulated what was seen in RnpB-depleted cells, the fact that only the two larger transcripts behaved as expected from the results obtained in the RNase P-depleted ppGpp 0 strain (Fig. 7A) suggests that regulation of the smallest transcript (Y/P 2 -T 2 ) is more complex than simple transcriptional repression by (p)ppGpp.
The fact that the Y/P 2 -T 2 rimM transcript could be downregulated independently of (p)ppGpp production led us to investigate the possibility of a further layer of regulation where the growth slowdown in tRNA maturase-depleted cells would also affect rimM expression by a mechanism independent of alarmone levels. To test this, we tRNA Maturase Impact on mRNA Levels and Stability Microbiology Spectrum sought to reproduce the growth rate defect by depleting for an unrelated essential enzyme. We therefore performed Northern blot analysis on total RNA extracted from both RNase III (rnc) depletion and deletion strains. The double-strand-specific endoribonuclease RNase III is essential in B. subtilis because it is required to silence expression of foreign toxin genes of two prophages (txpA in the prophage-like element skin and yonT in prophage SPb) (16). Whereas depletion of RNase III in a WT background leads to growth arrest, the rnc gene can be deleted in a strain lacking the two prophage toxins without a marked effect on growth rate. The depletion of RNase III in a toxin-WT background led to a very limited derepression of the CodY regulon in comparison with tRNA maturase-depleted strains (Fig. 8A) and did not result in accumulation of visible amounts of (p)ppGpp on thin-layer chromatography (TLC) (Fig. 8B). This validates the use of RNase III depletion strains to examine the effect of growth rate on rimM expression and to distinguish this from the effect of accumulating high levels of (p)ppGpp. While RNase III deletion (in the DtxpA DyonT background) had no effect on rimM expression, all three rimM-containing transcripts were strongly downregulated during RNase III depletion (in the toxin-WT background) (Fig. 8C), confirming that growth rate also plays a major role in the regulation of rimM expression, independently of (p)ppGpp and tRNA maturase depletion.

DISCUSSION
This study began with the observation that depletion of tRNA maturase enzymes in B. subtilis led to a defect in 30S ribosome subunit assembly, which we showed was in part due to an accumulation of (p)ppGpp and an inhibition of the activity of 30S assembly GTPases (8). In our early attempts to understand the mechanism underlying this phenomenon, we studied the expression of several 30S assembly factors and discovered that most were either up-or downregulated at the mRNA level upon depletion of either RNase P or RNase Z.  Fig. 4. 16S rRNA levels (ethidium bromide stained) are shown as a loading control. Series of blots where a single loading control is shown were stripped and reprobed. Note that this Northern blot was generated by reprobing a membrane previously used in reference 8, with permission to reuse the ywaA control panel granted by the publisher.
tRNA Maturase Impact on mRNA Levels and Stability Microbiology Spectrum We initially focused on the rimM mRNA because the 30S ribosome assembly defect observed in tRNA maturase depletion mutants was very similar to that seen in a DrimM strain. Although we later showed that ectopic rimM expression could not correct the assembly defect in the rnpB-depleted strain (8), we were nonetheless curious about how rimM expression was affected by the decrease in the levels of mature tRNAs. Together, our data (summarized in Fig. S4) indicate that the downregulation of rimM transcript levels in tRNA maturase-depleted cells is the result of a complex mixture of transcriptional and posttranscriptional mechanisms, caused by a combination of effects mediated by a reduction in growth rate, (p)ppGpp production, and a translational defect due to lack of functional tRNAs, with each layer of regulation capable of functioning independently of the others and affecting the three rimM transcripts distinctly at different levels of severity.
We hypothesized that the accumulation of immature tRNAs during tRNA maturase depletion increases ribosome stalling. Stalled ribosomes are known to affect mRNA decay in bacteria (17), and a tRNA loss-of-function mutation leading to pre-tRNA processing defects was reported to induce ribosome stalling in mice (9). In agreement with our hypothesis, we observed that treatment with the translation elongation inhibitor chloramphenicol at the MIC recapitulates the effects of tRNA maturase depletion on the mRNA levels of several different assembly factor mRNAs tested, four of which were upregulated (era, yqeH, ydaF, and yjcK) and two downregulated (rimM and cpgA). Interestingly, Cm treatment at subinhibitory concentrations did not impact cofactor mRNA levels in the same way, with low Cm concentrations initially having transitory up-effects that were then reversed at longer incubation times. One possibility is that tRNA Maturase Impact on mRNA Levels and Stability Microbiology Spectrum short ribosome stalls transiently block access to cleavage sites by housekeeping RNases such as RNase Y, resulting in mRNA stabilization, while prolonged stalling could lead to mRNA destabilization by an enzyme such as Rae1, proposed to enter the A site of stalled ribosomes (2), or by leaving large stretches of mRNA unoccupied by ribosomes and vulnerable to cleavage by canonical degradation pathways. Another notable difference between the two conditions is that the stringent response is induced by Cm at the MIC, as evidenced by the increase in expression the ilvA mRNA from the CodY regulon (Fig. S5), which could also contribute to the increased severity of the response to higher Cm concentrations. Activation of the stringent response in Cmtreated B. subtilis was also previously observed (18), consistent with our results. This is a marked difference from E. coli, where Cm is a known inhibitor of stringent response induction (19,20). The mechanism still remains elusive in both cases. Beyond their canonical role in protein synthesis, tRNAs have been implicated in the regulation of several biological processes (for reviews, see references 21 and 22). A new class of small noncoding RNAs has emerged recently called tRNA-derived fragments (tRFs) or tRNA-derived small RNAs, whose biological roles are not yet well understood (23). Different types of tRFs differ in the cleavage position of the mature or precursor tRNA transcript. They have been particularly studied in humans, where they have been shown to be involved in regulation of a variety of cellular processes, including global translation, cellular proliferation, apoptosis, and epigenetic inheritance (24). Interestingly, a 39 tRF in human cells plays an essential role in fine-tuning ribosome biogenesis under normal physiological conditions by posttranscriptionally regulating translation of at least two r-protein mRNAs (25). Although tRFs have not yet been identified in B. subtilis, we asked whether pre-tRNAs could bind certain assembly factor mRNAs via their 59 or 39 extensions and cause some of the posttranscriptional effects observed in the tRNA maturase depletion strains. tRFs with 59 or 39 extensions (pre-tRFs) could similarly behave as a new pool of potential regulatory sRNAs. Although the potential base-pairing we identified between the 59 extension of trnD-Tyr and the rimM transcripts does not seem to play a role in the downregulation of rimM expression, this does not preclude the possibility that other pre-tRNAs or pre-tRFs could be involved in posttranscriptional regulatory events in B. subtilis.
In this study, we focused on the consequences of tRNA maturase depletion on ribogenesis cofactors mRNAs; however, the effects we uncovered are likely not restricted to this category of genes, as exemplified by the effects on the stabilities of the yrzI and bmrCD mRNAs. Along these lines, a recent study in E. coli showed that the abundance of 46% of transcripts were affected in a strain where the protein moiety of RNase P was heat denatured (26). The observations that the addition of chloramphenicol mimicked the effect of tRNA maturase depletion in B. subtilis for the upregulated mRNAs and that downregulation was the net result of a mixture of translational and transcriptional effects suggest that the effects seen on the E. coli transcriptome may substantially be the result of ribosome stalling on mRNAs due to lack of functional tRNAs, with differential impacts (up, down or neutral) on individual mRNA stabilities or transcription levels. More detailed studies are required to untangle these effects on a global level in both organisms.

MATERIALS AND METHODS
Strains and culture conditions. All B. subtilis strains used were derived from our laboratory strain SSB1002, a W168 trp 1 prototrophic strain. Strains are listed in Table S1, and details of strains and plasmid constructs are provided in Table S2 and S3, respectively. Oligonucleotides used are listed in Table S4.
Unless stated otherwise, B. subtilis strains were grown in 2Â YT liquid medium (1.6% peptone, 1% yeast extract, 1% NaCl) at 200 rpm at 37°C in #1/10 volume of the flask to ensure proper aeration. Overnight precultures were grown in the presence of appropriate antibiotics and inducer (1 mM IPTG or 2% xylose), in the case of depletion strains. Experimental cultures were grown in the absence of antibiotics, except where stated. For depletion strains, overnight induced cultures were washed three times with prewarmed 2Â YT medium and inoculated at an optical density at 600 nm (OD 600 ) between 0.02 and 0.2, depending on the strain, in fresh medium with or without inducer. Generally, induced cells were harvested for RNA preparation around an OD 600 of 0.6, and cells grown in the absence of the inducer were followed until they reach a plateau before being harvested. Inoculation and depletion conditions were determined empirically for each strain, such that the depleted cells were harvested between tRNA Maturase Impact on mRNA Levels and Stability Microbiology Spectrum OD 600 of 0.3 and 0.7. For RnpA depletion, cultures were inoculated at an OD 600 of 0.05 in the presence of 2% xylose (inducer) or 2% glucose to tighten repression of the Pxyl promoter, which typically led to a growth arrest (plateau) around an OD 600 of 0.6. For rnz and rnpB depletion strains, cultures were inoculated in presence or in the absence of 1 mM IPTG at OD 600 of 0.05 and 0.2, respectively. RNase Z and RnpB depleted cells typically plateau around OD 600 of 0.6 and 0.3, respectively. For rifampicin experiments, B. subtilis strains were grown in 2Â YT at 37°C with shaking as described above. At an OD 600 of 0.6 (or less for some depletion strains), rifampicin was added to a final concentration of 150 mg/mL in order to block new RNA synthesis. Samples were collected at different time points (e.g., 0, 2, 5, 10, 15, and 20 min) by mixing the cells with frozen 10 mM sodium azide (200 mL for 1.3 mL culture). Samples were vortexed until the sodium azide thawed, cells were pelleted by centrifugation at 4°C, and the pellet was conserved at 220°C until RNA extraction.
To mimic amino acid starvation, we depleted charged arginine tRNAs by addition of arginine hydroxamate (RHX) at 250 mg/mL in cultures growing in 2Â YT at an OD 600 of 0.3.
To study the effect of translation pausing, we added the translation elongation inhibitor Cm at a subinhibitory concentration (2.5 mg/mL) or the MIC (5 mg/mL) to cells growing in 2Â YT at an OD 600 of 0.6. Cells were harvested just before Cm addition (t 0 ) and 15, 30, and 60 min after treatment.
To reproduce some growth conditions from the B. subtilis tiling array experiment (10) known to lead to a decrease in rnpA expression, ethanol was added to cultures growing in minimal medium (M9 with 0.5% glucose) at 4% (vol/vol) around an OD 600 of 0.4, and cells were harvested 10 min after treatment.
Plasmid constructs. The rimM gene was amplified by PCR using the oligonucleotide pair CC2034/ CC1986 and cloned between the BamHI and XhoI sites of the integrative pHM2-Pspac(con) vector (Table  S3). The bicistronic ylqD-rimM construct was amplified by PCR using the oligonucleotide pair CC1985/ CC1986 and cloned between the BamHI and SalI sites of pHM2-Pspac(con). The mutated construct ylqD M -rimM was obtained by two-fragment overlapping PCR. The upstream fragment was amplified with the forward primer CC1985 and the reverse primer CC2012, and the downstream fragment was amplified with the forward primer CC2011 and the reverse primer CC1986. The overlapping fragments were reamplified using the oligonucleotide pair CC1985/CC1986 and cloned between the BamHI and SalI sites of pHM2-Pspac(con). The integrative plasmids were linearized with XbaI before transformation, for integration into the amyE locus of the B. subtilis chromosome.
RNA extraction and Northern blots. RNA extraction was typically performed using the glass bead/ phenol protocol (adapted from reference 27) on 1 to 8 mL mid-log-phase B. subtilis cells growing in 2Â YT.
To perform Northern blotting, 5 mg total RNA was denatured for 5 min at 95°C in RNA gel loading dye (Thermo Scientific) before being separated on 1% agarose gels in 1Â Tris-borate-EDTA (TBE) (native) or on denaturing 5% acrylamide gels in 1Â TBE plus 7 M urea. RNA was transferred from agarose gels to a Hybond-N membrane (GE Healthcare) by capillary transfer for 4 h minimum in 1Â transfer buffer (5Â SSC [1Â SSC is 0.15 M NaCl plus 0.015 M sodium citrate], 0.01 M NaOH). For Northern blots of acrylamide gels, RNA was electrotransferred at 4°C in 0.5Â TBE for 4 h at 60 V or overnight at 12 V. RNA was cross-linked to the membrane by UV cross-linking at 120,000 mJ/cm 2 using HL-200 Hybrilinker UV cross-linker (UVP). Probes for Northern blots were usually 25-to 30-nt DNA oligonucleotides radiolabeled on their 59 ends by polynucleotide kinase. The cpgA mRNA was detected with a riboprobe by using a PCR fragment amplified using oligonucleotides CC2200 and CC2201 as the template. Membranes were preincubated in Ultra-Hyb (Life Technologies) for agarose blots or Roti-Hybri-Quick (Roth) for acrylamide blots for 1 h and hybridized with radiolabeled probes for a minimum of 4 h. Preincubation, hybridization, and wash steps were performed at 42°C in the case of 59-labeled oligonucleotides or at 68°C for riboprobes. Membranes were quickly rinsed once at room temperature in 2Â SSC-0.1% SDS to remove nonhybridized probe before being washed once for 5 min in the same buffer and then twice for 5 min in 0.2Â SSC-0.1% SDS. Northern blots were exposed to PhosphorImager screens (GE Healthcare), and the signal was obtained by scanning with a Typhoon scanner (GE Healthcare) and analyzed by Fiji (ImageJ) software.
TLC. TLC analysis was used to detect radiolabeled (p)ppGpp as described in reference 8.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 2.6 MB.