Rapid Degradation of Hfq-Free RyhB in Yersinia pestis by PNPase Independent of Putative Ribonucleolytic Complexes

The RNA chaperone Hfq in bacteria stabilizes sRNAs by protecting them from the attack of ribonucleases. Upon release from Hfq, sRNAs are preferably degraded by PNPase. PNPase usually forms multienzyme ribonucleolytic complexes with endoribonuclease E and/or RNA helicase RhlB to facilitate the degradation of the structured RNA. However, whether PNPase activity on Hfq-free sRNAs is associated with the assembly of RNase E or RhlB has yet to be determined. Here we examined the roles of the main endoribonucleases, exoribonucleases, and ancillary RNA-modifying enzymes in the degradation of Y. pestis RyhB in the absence of Hfq. Expectedly, the transcript levels of both RyhB1 and RyhB2 increase only after inactivating PNPase, which confirms the importance of PNPase in sRNA degradation. By contrast, the signal of RyhB becomes barely perceptible after inactivating of RNase III, which may be explained by the increase in PNPase levels resulting from the exemption of pnp mRNA from RNase III processing. No significant changes are observed in RyhB stability after deletion of either the PNPase-binding domain of RNase E or rhlB. Therefore, PNPase acts as a major enzyme of RyhB degradation independent of PNPase-containing RNase E and RhlB assembly in the absence of Hfq.


Introduction
Small regulatory RNAs (sRNAs) function as posttranscriptional regulators by altering translation or stability of the target mRNA, which increases their applicability in different physiological processes in bacteria [1]. The RNA chaperone Hfq is hypothesized to facilitate the access of sRNAs to their mRNA targets and stabilize sRNAs by protecting them from the attack of RNase E [2]. Given that the increasing amount of available information on sRNA-induced mRNA decay is accumulating [3][4][5][6], the sRNA degradation processes and RNases that catalyze such activities must be investigated. The multienzyme assembly of RNA degradosome is important for mRNA decay and processing in Escherichia coli. RNase E and polynucleotide phosphorylase (PNPase) are two major components of the RNA degradation process [7,8]. RNase E is also responsible for the rapid degradation of sRNAs and competes with Hfq in accessing the same RNA sequences [9][10][11]. Hfq recruits RNase E by directly interacting with the RhlB-recognition region, which is hypothesized to cause the coupled cleavage of mRNA and sRNA [6,12]. PNPase plays the protective role in the RNase E-dependent degradation in the presence of Hfq [13,14]. Recent studies show that Hfq has a limited access to RNAs under wildtype conditions considering the dynamic interactions of Hfq with sRNAs [15][16][17]. A transient Hfq-free state of sRNAs may also be observed. A recent study shows that sRNAs are preferably degraded by the major exoribonuclease PNPase upon release from Hfq [14]. PNPase usually cooperates with RNase E in RNA degradation complexes [18]. RNA helicase RhlB usually facilitates RNA degradation by manipulating RNA structure and remodeling ribonucleoprotein complexes in the presence or absence of RNase E [19]. However, the relationship between the PNPase activity in Hfq-free sRNAs and RNA degradation complexes remains unknown. The well-characterized sRNA RyhB was used as a model sRNA for this study. RyhB is an Hfq-binding sRNA that maintains iron homeostasis in bacteria [20,21]. Besides Hfq, RyhB also becomes very stable when the overall mRNA transcription is stalled in E. coli [6]. Two RyhB homologs possessing the conserved core and rho sequences in E. coli [20] have also been characterized in S. typhimurium [22]. RyhB1 and RyhB2 are upregulated in the infected lungs of mice upon intranasal inoculation of Yersinia pestis, which indicates that they may serve as important functions during Y. pestis pathogenesis. The stability of RyhB1 and RyhB2 is differentially Hfq-dependent in Y. pestis grown under nutrient-limiting conditions [23]. This study constructs single or combined hfq mutant strains that lack various RNases or ancillary enzymes and monitors the expression level and degradation speeds of RyhB to investigate the effect of these enzymes on the degradation of Hfq-free RyhB.

Bacterial Strains and Growth
Conditions. All strains are derivatives of Y. pestis strain 201, a newly established biovar, the Microtus [24]. Table 1 shows the bacterial strains that are used in this study. Except for the RNase E mutants, all mutant strains were constructed by replacing the entire gene with an antibiotic cassette via -Red homologous recombination. RNase E is essential for viability in bacteria, but deleting the C-terminal half (CTH) of this enzyme is not lethal [26]. The CTH after the 910th containing putative PNPase-binding site (1190-1221aa corresponding to 1021-1061aa in E. coli RNase E) [26] was deleted and designated as 910 . Bacteria were grown to midexponential phase ( 620 ≈ 1.0) in BHI medium at 26 ∘ C. Iron depletion was induced by adding 100 M 2 ,2 -dipyridyl (DIP) for 20 min. Antibiotics were added when needed at the following concentrations: 34 g/mL chloramphenicol, 50 g/mL kanamycin, 100 g/mL ampicillin, 20 g/mL gentamicin, and 20 g/mL streptomycin.

RNA Extraction and Northern Blotting Analysis.
Pure bacterial cultures were mixed with RNAprotect Bacteria Reagent (Qiagen) to minimize RNA degradation. The total RNA was then extracted from Y. pestis using TRIzol Reagent (Invitrogen). Northern blotting analysis was performed by using a DIG Northern Starter Kit (Roche) according to the manufacturer's protocol described by Beckmann et al. [27]. RNA samples (3 g) were denatured at 70 ∘ C for 5 min, separated on 6% polyacrylamide-7M urea gel, and transferred onto Hybond N + membranes (GE) via electroblotting. The membranes were UV-crosslinked and prehybridized for 1 hr, and 3 -end DIG-labeled RNA oligonucleotides were added. The membranes were then hybridized overnight at 68 ∘ C in a DIG Easy Hyb. RNA was immunologically detected and scanned according to the instructions. Multiple exposures to X-ray film were taken to achieve the desired signal strength.

RNA Half-Life Determination.
Bacteria grown to exponential phase were treated with 250 g/mL rifampicin for RNA half-life determination. Culture samples were collected at 0, 5, 10, 20, 30, and 60 min and were subject to RNA extraction and Northern blotting. Films were scanned and RNA band intensity was measured using Quantity One software. The intensities were plotted and RNA half-lives were calculated using the slope from each plot.

Quantitative RT-PCR.
Total RNA was isolated from different Y. pestis strains grown to exponential growth phase (OD 620 = 1.2) in BHI by using Trizol Reagent (Invitrogen). DNA contaminants were removed by using DNA-free Kit  (Ambion), and the cDNA was converted by using random hexamer primers with the Superscript II system (Invitrogen). Real-time PCR was performed in duplicate for each RNA preparation by using the LightCycler system (Roche) with an appropriate dilution of cDNA as a template. Negative controls without reverse transcriptase enzyme were included in all experiments. Relative quantitative analysis across different cDNA templates was performed by using LightCycler 480 software (Bio-Rad) with the 16S rDNA as the normalized gene.

Influence of RNases and Ancillary RNA-Modifying
Enzymes on the Regulation of Hfq-Free RyhB. BHI was selected as the growth medium for bacterial culture because some mutants that were constructed in this study experienced a slow growth upon inoculation into TMH medium, which pose a challenge to our experiments. The expressions of RyhB1 and RyhB2 were monitored in multiple hfq mutants that lacked major RNases or ancillary RNA-modifying enzymes to validate the influence of endoribonucleases, exoribonucleases, and ancillary RNAmodifying enzymes on RyhB regulation in Y. pestis without Hfq (Figure 1). The expression levels of RyhB1 and RyhB2 slightly increased (∼1.8-fold) upon the deletion of PNPase, but no obvious changes were observed in the RNase E truncate and deletion strains of RNase G (rng), RNase II (rnb), or polyA polymerase (pcnB). In contrast, RyhB was rarely detected in the double mutants that lacked Hfq and RNase III (rnc).
The rne (910-1221aa), rng, pnp, and rnb genes were deleted from the double deletion mutant that lacked Hfq and RNase III to determine which RNases account for the degradation of RyhB1 and RyhB2, respectively ( Figure 2). RyhB in the hfq-rnc-pnp mutant reached a similar amount of that in the hfq mutant, which indicates that PNPase was the main contributor in the degradation of Hfq-free RyhB [14].
The degradation of Hfq-free RyhB by PNPase tends to occur in stationary phase rather than exponential phase in E. coli [14]. However, the inactivation of PNPase in this study increased the RyhB levels in Y. pestis grown to exponential phase. Therefore, PNPase may degrade the Hfq-free RyhB in different growth-phase-dependent manners in E. coli and in Y. pestis. However, such discrepancy may also be due to the different sample timing that was used in these two experiments. It would be helpful to make it clear if more timepoint samplings are included in these experiments.

The RNase-III-Inactivation-Induced mRNA Level Increase of PNPase May Be Partially Responsible for the Degradation of
Hfq-Free RyhB. Few amounts of micA could be also detected in the hfq-rnc double mutant of E. coli [14]. Andrade et al. explained this phenomenon as an impairment of RNase III activity that was caused by the decreased duplex in the absence of Hfq. However, this impairment cannot explain the obvious difference in RyhB expression between hfq and hfqrnc double mutant. RNase III can alter gene expression by cleaving dsRNA or by binding without cleaving RNA [28]. RNase III has been proved to involve in the autoregulation of PNPase in E. coli by cleaving the 5 end of pnp mRNA [29]. However, the unprocessed pnp mRNA is accumulated and can be translated into polynucleotide phosphorylase in E. coli rnc mutant [29]. To determine if the inactivation of RNase III affected the expression of PNPase, quantitative PCR was performed to estimate the relative amounts of pnp mRNA in different mutants (Figure 3). The pnp gene was upregulated from 1.9-to 3.3-fold in hfq-rnc double and triple mutants than in the hfq mutant, which further confirmed that PNPase was the main exoribonuclease responsible for the degradation of Y. pestis RyhB in the absence of hfq.
The RNase-III-inactivation-induced upregulation of PNPase could be partially responsible for the decreased expression of RyhB ( Figure 2). However, the effects of RNase III on RyhB stability could not be determined through other means.

PNPase Activity on RyhB in the Absence of RNase III Is
Dependent on the State of Hfq Binding. RNase III affects the stability of the Hfq-dependent sRNA, MicA, in Salmonella [30]. The expression patterns of single and double mutants of rnc and hfq were compared via Northern blotting to examine the effects of RNase III and Hfq inactivation on the rapid degradation of RyhB. RyhB was rarely detected after inactivating both RNase III and Hfq. However, the amount of RyhB could reach modest levels in the rnc and hfq single mutants as well as in the complementary strains that carried the corresponding plasmids. Therefore, the PNPase activity on RyhB in the absence of RNase III depends on the state of Hfq binding (Figure 4). RyhB was rapidly degraded by the increased levels of PNPase in the absence of Hfq because of the RNase III inactivation. Δhfq-rnc-rhlB in bacterial cells that were grown in minimal media (with ∼8 min half-life). The half-lives of both RyhB1 and RyhB2 exceeded 60 min in a WT strain that was grown exponentially in BHI medium (data not shown), which indicated that the nutrition conditions would influence the stability of Y. pestis RyhB in the absence of Hfq. The half-lives of RyhB in the hfq-pnp double mutant were investigated to verify the effects of PNPase on the degradation of Hfq-free RyhB ( Figure 5). The stability of RyhB slightly increased in the hfq-pnp double mutant rather than in the hfq single mutant, which confirmed the role of PNPase in the degradation of Hfq-free RyhB. The rnc deletion mutation produced insignificant effects on the stability of RyhB with half-lives of 20.2 min and 49.3 min ( Figure 5). However, the 14 min decrease in the half-life of RyhB2 in the hfq-rne 910 double mutant remains unclear. The half-lives of RyhB dramatically reduced to 3.8 min and 6.5 min in the hfq-rnc double mutant, whereas the deletion of the pnp gene increased the half-life of RyhB to >30 min ( Figure 5). Therefore, the RNase-III-induced PNPase increase might be responsible for the RyhB degradation in the absence of Hfq, and the PNPase served as the main enzyme in the degradation of Hfq-free RyhB.

Rapid Degradation of Hfq-Free RyhB by PNPase Is
PNPase usually forms multienzyme ribonucleolytic complexes with RNase E and/or RNA helicase RhlB during the degradation of the structured RNA [31,32]. RNase E serves as a "scaffolding" protein of RNA degradosome that contains the binding sites of three major degradosome components, namely, PNPase, DEAD-box helicase RhlB, and enolase [8,33]. RhlB facilitates the formation of single stranded RNA, which helps PNPase to engage in the 3 to 5 exoribonucleolytic degradation of RNA [15]. PNPase directly interacts with RhlB by forming the transient complex, which is not dependent on the formation of the degradosome [34]. Therefore, this study tries to determine if RNase E degradosome is involved in PNPase activity on Hfq-free RyhB. Given that the deletion of the rne gene in the encoding of RNase E is lethal, an rne mutant without PNP-binding domain was constructed in this study to produce an RNase E protein that was unassociated with PNPase.
The Northern blotting analysis revealed that the mutation of rne and rhlB had > 7min half-life in the hfq-rnc mutant, but its stability was substantially lower than that upon PNPase inactivation ( Figure 5). Therefore, the PNPasecontaining degradosome or exosome plays minor roles in Hfq-free RyhB decay, and PNPase might be involved in these processes by itself or through other unknown mechanisms. Therefore, the degradation of Hfq-free sRNAs is far more complex than what was previously expected. An extended analysis should be performed to check if these results could be applied to other sRNAs.