CvfA Protein and Polynucleotide Phosphorylase Act in an Opposing Manner to Regulate Staphylococcus aureus Virulence*

Background: Production of 3′-phosphorylated RNA by CvfA affects S. aureus virulence gene expression. Results: Disrupting pnpA-encoding exonuclease suppressed the cvfA-deleted mutant phenotype. Purified PNPase did not degrade 3′-phosphorylated RNA. Conclusion: CvfA-produced 3′-phosphorylated RNA inhibits PNPase-induced RNA degradation, resulting in hemolysin production by S. aureus. Significance: Altering the nucleotide structure at the RNA 3′ terminus regulates S. aureus virulence. We previously identified CvfA (SA1129) as a Staphylococcus aureus virulence factor using a silkworm infection model. S. aureus cvfA-deleted mutants exhibit decreased expression of the agr locus encoding a positive regulator of hemolysin genes and decreased hemolysin production. CvfA protein hydrolyzes a 2′,3′-cyclic phosphodiester bond at the RNA 3′ terminus, producing RNA with a 3′-phosphate (3′-phosphorylated RNA, RNA with a 3′-phosphate). Here, we report that the cvfA-deleted mutant phenotype (decreased agr expression and hemolysin production) was suppressed by disrupting pnpA-encoding polynucleotide phosphorylase (PNPase) with 3′- to 5′-exonuclease activity. The suppression was blocked by introducing a pnpA-encoding PNPase with exonuclease activity but not by a pnpA-encoding mutant PNPase without exonuclease activity. Therefore, loss of PNPase exonuclease activity suppressed the cvfA-deleted mutant phenotype. Purified PNPase efficiently degraded RNA with 2′,3′-cyclic phosphate at the 3′ terminus (2′,3′-cyclic RNA), but it inefficiently degraded 3′-phosphorylated RNA. These findings indicate that 3′-phosphorylated RNA production from 2′,3′-cyclic RNA by CvfA prevents RNA degradation by PNPase and contributes to the expression of agr and hemolysin genes. We speculate that in the cvfA-deleted mutant, 2′,3′-cyclic RNA is not converted to the 3′-phosphorylated form and is efficiently degraded by PNPase, resulting in the loss of RNA essential for expressing agr and hemolysin genes, whereas in the cvfA/pnpA double-disrupted mutant, 2′,3′-cyclic RNA is not degraded by PNPase, leading to hemolysin production. These findings suggest that CvfA and PNPase competitively regulate RNA degradation essential for S. aureus virulence.

duces 3Ј-monophosphorylated RNA (22). It remained unclear, however, how modification of the 3Ј-terminal structure of RNA by CvfA affects the expression of S. aureus virulence genes.
In this study, we searched for a gene that genetically interacts with cvfA to reveal the molecular mechanism of virulence gene regulation by CvfA. Our findings revealed that disruption of pnpA-encoding PNPase with exonuclease activity suppressed the phenotype of the cvfA-deleted mutant. Furthermore, RNA degradation activity of PNPase was affected by the structure of the 3Ј-terminal nucleotide of the RNA substrate.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Culture Conditions-The E. coli JM109 strain was used to host pET-11a, pND50, and their derivatives. E. coli strains transformed with the plasmids were aerobically cultured in the presence of 100 g/ml ampicillin or 25 g/ml chloramphenicol. S. aureus strains were aerobically cultured in tryptic soy broth at 37°C. To transform the S. aureus strain with plasmids, 10 g/ml erythromycin, 20 g/ml phleomycin, or 12.5 g/ml chloramphenicol was added to tryptic soy broth. Details of the bacterial strains and plasmids are listed in Table 1.
DNA Manipulation-Transformation of E. coli, extraction of plasmid DNA, and polymerase chain reaction (PCR) were performed according to Sambrook et al. (23). S. aureus was transformed using electroporation (24). Introduction of point mutations into plasmid DNA was performed according to Li et al. (25).
Construction of the pnpA-disrupted Mutant and Plasmids Carrying Mutated pnpA Genes-DNA fragments containing the internal region (ϩ84 to ϩ584 bp) of pnpA (ϩ1 as the first nucleotide of the open reading frame) was amplified by PCR using oligonucleotide primers ( Table 2) and genome DNA of NCTC8325-4 strain as the template. The amplified fragment was inserted into EcoRI and BamHI sites of pMutinT3, resulting in a targeting plasmid. S. aureus RN4220 strain was electroporated with the targeting plasmid, resulting in a strain resistant to erythromycin. The plasmid that was integrated into the pnpA gene was transferred to the NCTC8325-4 strain using phage 80␣, which resulted in the pnpA-disrupted mutant. Disruption of pnpA in NCTC8325-4 strain was confirmed by Southern blot analysis.
The DNA fragment containing intact pnpA gene was amplified by PCR using oligonucleotide primers ( Table 2) and genome DNA of NCTC8325-4 strain as the template. The amplified DNA fragment was inserted into EcoRI and BamHI sites of pND50, resulting in ppnpA carrying intact pnpA. DNA fragments containing the mutated pnpA gene were amplified by PCR using oligonucleotide primers and ppnpA as the template. The amplified DNA fragments were treated with DpnI to degrade the template ppnpA, and the amplified DNA was selfligated, resulting in plasmid carrying a mutated pnpA gene (D96G, R402A/R403A, H407D, R413D, D496G, or ⌬RBD). The desired pnpA mutation was confirmed by sequencing. The RN4220 strain was transformed with ppnpA and plasmids carrying mutated pnpA genes. The plasmids were transferred to the cvfA/pnpA double-disrupted mutant by phage 80␣.
Measurement of Hemolysin Production-S. aureus overnight culture (2 l) was spotted onto tryptic soy agar plates containing 5% sheep erythrocytes and incubated for 12 h at 37°C. The clear zone around the S. aureus colony was evaluated. Reporter Assay-S. aureus strains were transformed with reporter plasmids carrying the agr P2, agr P3, and hla promoter. Overnight cultures of the transformed strains were inoculated to 100-fold amounts of fresh tryptic soy broth and aerobically cultured at 37°C. The cultured cells were collected by centrifugation and lysed in a buffer (20 mM KH 2 PO 4 (pH 7.8), 0.04% Triton X-100, 0.1 mM DTT, 10 g/ml lysostaphin, 1 tablet of protease inhibitor (Roche Applied Science)). Cell lysate supernatant was incubated with luciferase substrate, and luminescence was measured using a luminometer (Berthold Technologies, Bad Wildbad, Germany). The promoter activity was calculated as luminescence units/mg of protein.
Microarray Analysis-Overnight cultures of S. aureus strains were inoculated to 100-fold amounts of fresh tryptic soy broth and aerobically cultured to A 600 ϭ 4 at 37°C. The cultured cells were collected by centrifugation and treated with RNAprotect Bacteria Reagent (Qiagen). The cells were washed with phosphate-buffered saline and lysed by lysostaphin. Total RNA was extracted using an RNeasy mini kit according to the manufacturer's protocol (Qiagen), and any remaining DNA was degraded with RQ1 RNase-free DNase (Promega). cDNA was synthesized from the RNA using Superscript II reverse transcriptase. cDNA treated with NaOH was purified using a QIAquick PCR purification kit (Qiagen) and digested by DNase I. Fragmented cDNA was biotinylated using a biotin-ddUTP kit (Affymetrix), and hybridized on Staphylococcus aureus GeneChips (Affymetrix). The GeneChips were washed, and the 570-nm signal was read. Signal intensities were analyzed with GeneChips operating software and normalized with GeneSpring 4.0. Genes with a signal intensity that had a greater than 2-fold difference from the parent strain based on two independent experiments were identified as affected genes.
Quantitative Real Time PCR Analysis-Total RNA was collected from S. aureus cells as described above. cDNA was synthesized from the RNA using Multiscribe Reverse Transcriptase (Applied Biosystems). Quantitative real time PCR was performed using cDNA as a template, SYBR Premix ExTaq (Takara Bio), and primers ( Table 2). The signals were detected using an ABI PRISM 7700 sequence detector (Applied Biosystems).
Purification of PNPase-S. aureus pnpA gene fused with His 6 tag was inserted into pET11a, resulting in pN-His-pnpA. E. coli BL21(DE3) harboring pLysS was transformed with pN-His-pnpA. The transformed strain was aerobically cultured in 100 ml of Luria-Bertani broth at 37°C to A 600 ϭ 0.5; 1 mM isopropyl ␤-D-1-thiogalactopyranoside was added, and the mixture was cultured overnight at 16°C. The cells were collected by centrifugation and lysed by freezing and thawing, followed by sonication. The sample was centrifuged, and ammonium sulfate was added to the supernatant at a final concentration of 100%. The resulting precipitate was dissolved in buffer A (50 mM Tris-HCl (pH 8.0), 500 mM NaCl, 20% glycerol, 1 mM imidazole) and subjected to a Ni-NTA column (ProBond Resin, Invitrogen). The column was washed several times with buffer B (50 mM Tris-HCl (pH 8.0), 500 mM NaCl, 20% glycerol, 67 mM imidazole), and the proteins were eluted with buffer C (50 mM Tris-HCl (pH 8.0), 500 mM NaCl, 20% glycerol, 1 M imidazole). The amount of protein in each fraction was determined by the Bradford assay. To obtain an antibody against PNPase, purified PNPase (0.2 mg) was subcutaneously injected into a Japanese white rabbit five times at 2-week intervals. Blood was collected from the rabbit and used for IgG purification by protein G-Sepharose.
Measurement of Poly(A) Polymerization Activity and Phosphorolytic Activity-To determine poly(A) polymerization activity, PNPase protein was mixed with 5 mM ADP containing 0.3 M [2,8-3 H]ADP in a reaction buffer (50 mM Tris-HCl (pH 8.0), 5 mM MgCl 2 , 5 mM ADP) and incubated at 37°C for 15 min. The reaction was terminated by adding 0.1% perchloric acid, mixed with the same volume of 10% TCA, and incubated on ice for 10 min. The precipitated poly(A) was trapped by a glass filter (Whatman), and the radioactivity on the filter was measured by a scintillation counter (LC 5000TS, Beckman).

Target
Primer To determine the phosphorolytic activity, PNPase protein was mixed with 30 M poly(A) (Sigma) in a reaction buffer (50 mM Tris-HCl (pH 8.4), 5 mM MgCl 2 , 60 mM KCl, 10 mM sodium phosphate) at 37°C for 10 min. The reaction mixture was mixed with a 2.5 volume of ethanol and centrifuged. The A 260 of the supernatant was measured to calculate the amount of released ribonucleoside diphosphates.
Phosphorolytic Activity of PNPase against Different RNA Substrates-RNA substrates (5Ј-AAAAAAAAAAG-3Ј) with different 3Ј-terminal nucleotides were synthesized using the phosphoramidite method. The 3Ј-hydroxylated RNA and 3Ј-phosphorylated RNA were obtained from Hokkaido System Science, Sapporo, Japan. The 2Ј,3Ј-cyclic RNA was obtained from GeneDesign, Osaka, Japan. Structures of RNA substrates were confirmed by electrospray ionization time-of-flight mass spectrometry (microTOF, Bruker Daltonics). PNPase protein was mixed with 30 M RNA substrate in a reaction buffer (50 mM Tris-HCl (pH 8.4), 5 mM MgCl 2 , 60 mM KCl, 10 mM sodium phosphate) at 37°C for 10 min. The reaction was terminated by the addition of 2ϫ loading buffer (95% formamide, 0.025% SDS, 18 mM EDTA, 0.025% xylencyanol, 0.025% bromphenol blue) and boiling. The RNA sample was electrophoresed in a 7 M urea, 20% polyacrylamide gel and stained with SYBR Green. Images were analyzed using an image analyzer (Typhoon FLA9000, GE Healthcare). To measure the amount of the RNA degradation product, PNPase protein was mixed with different amounts of RNA substrate in the reaction buffer at 37°C for 10 min. The reaction mixture was then mixed with 2.5 volumes of ethanol and centrifuged. The A 260 of the supernatant was measured to calculate the amount of released ribonucleoside diphosphates. The K m and V max values of PNPase protein against different RNA substrates were determined by nonlinear regression analysis using Graph Pad Prism version 5.0c.
Detection of PNPase and CvfA-S. aureus cells were collected, and the cell walls were lysed in digestion buffer (30% raffinose, 50 mM Tris-HCl (pH 7.5), 145 mM NaCl, 100 g/ml lysostaphin, 10 units/ml DNase I) at 37°C for 30 min. The cells were collected by centrifugation and lysed in lysis buffer (50 mM Tris-HCl (pH 8.0), 100 units/ml DNase I). The protein concentration was determined by the Bradford assay. The protein was electrophoresed in 15% SDS-polyacrylamide gels. The proteins were blotted to a PVDF membrane (Immobilon-P, Merck). The membrane was treated with blocking buffer (TBST: 20 mM Tris-HCl (pH 7.6), 150 mM NaCl, 0.1% Tween 20, 5% Easy Blocker (GeneTex, Irvine, CA)) at room temperature for 1 h. The membrane was treated with blocking buffer containing 1:1000 anti-PnpA IgG or anti-CvfA IgY (22) at room temperature for 1 h. After washing with TBST, the membrane was treated with a blocking buffer containing 1:2000 anti-rabbit IgG conjugated with alkaline phosphatase or anti-chicken IgY conjugated with HRP at room temperature for 1 h. After washing with TBST, the membrane for detecting PNPase was reacted with a substrate for alkaline phosphatase (nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate, Roche Applied Science). The membrane for detecting CvfA was reacted with an HRP substrate (Western Lightning Plus ECL, PerkinElmer Life Sciences) and subsequently exposed to film (Hyperfilm ECL, GE Healthcare).
Silkworm Infection Experiment-The infection experiment using silkworms was performed according to the previously described method (26). Fertilized eggs were purchased from Ehime Sansyu (Ehime, Japan). Hatched larvae were raised to fifth instar larvae by feeding an artificial diet. S. aureus overnight cultures were centrifuged, and the cells were suspended in saline. 2-Fold serial diluted bacterial solutions were injected into the hemolymph of silkworms (n ϭ 10). Surviving silkworms were counted at 29 h after the injection. LD 50 values were determined from dose-survival curves.
Mouse Infection Experiment-Four-week-old female ICR mice were purchased from CLEA Japan, Inc. (Tokyo, Japan). S. aureus overnight cultures were centrifuged, and the cells were suspended in PBS containing 5% hog gastric mucin. 2-Fold serial dilutions of the bacterial solutions were injected into the peritoneal cavity in mice (n ϭ 5). Surviving mice were counted at 24 h after the injection. LD 50 values were determined from the dose-survival curves.

Decreased Hemolysin Production and Agr Expression in the cvfA Deletion Mutant Is Suppressed by the Disruption of pnpA-
To investigate the molecular mechanisms underlying the regulation of hemolysin production by CvfA, we searched for gene mutations that suppress the decreased hemolysin production of the cvfA-deleted mutant. Because CvfA has phosphodiesterase activity against 2Ј,3Ј-cyclic phosphodiester linkage at the 3Ј-terminal nucleotide of RNA, we hypothesized that modification of the RNA by CvfA would affect the sensitivity of the RNA against other RNA metabolic enzymes. To examine this possibility, we constructed double-disrupted mutants of cvfA and eight other genes encoding RNases, including RNase III (SA1076), SA0489, RNase HII (SA1087), SA1335, SA0450, PNPase (SA1117), RNase R (SA0735), and YhaM (SA1660), and examined their hemolysin production. Disruption of SA1076, SA1335, or SA1660 in the cvfA-deleted mutant caused slow growth or was not stable and thus not further evaluated. Disruption of SA0449 or SA0735 in the cvfA-deleted mutant was not successful. Disruption of SA0450 or SA1087 in the cvfAdeleted mutant had no effect on hemolysin production. cvfAdeleted mutants with disruption of pnpA-encoding PNPase produced greater amounts of hemolysins than cvfA-deleted mutants without disruption, which normally produce only small amounts of hemolysins (Fig. 1A). Thus, disruption of pnpA suppressed the decreased hemolysin production of cvfAdeleted mutants. The doubling speed of the cvfA/pnpA doubledisrupted mutant was indistinguishable from that of the parent strain and the cvfA-deleted mutant (Fig. 1B). In addition, the promoter activity of the hla gene encoding ␣-hemolysin in the cvfA/pnpA double-disrupted mutant was higher than that in the cvfA-deletion mutant (Fig. 1C). In the cvfA/pnpA doubledisrupted mutant, promoter activities of P2 and P3 of the agr locus, which positively regulate hla transcription, were higher than those in the cvfA-deleted mutant (Fig. 1, D and E). Therefore, we concluded that the decreased hemolysin production and agr expression in the cvfA-deleted mutant were suppressed by the disruption of pnpA and that cvfA genetically interacts with pnpA.
The deletion of cvfA affects the expression of various genes in S. aureus (27) and S. pyogenes (20). We examined whether the disruption of pnpA restores the effects of cvfA deletion on gene expression other than agr and hla genes. First, we performed microarray analysis of the cvfA-deleted mutant. The cvfA deletion affected the expression of 20% of S. aureus genes (supple-mental Tables S1 and S2), consistent with previous reports (20,27). Among the genes with altered expression, we focused on the decreased expression of saeP, which is encoded by the sae locus that positively regulates S. aureus hemolysin production (28). The sae locus contains four genes, saeP, saeQ, saeR, and saeS. The saeP and saeQ genes are transcribed as a saeP-saeQ- Asterisks indicate a Student's t test p value of less than 0.05. F-H, S. aureus strains were cultured to late exponential phase (A 600 ϭ 2.5). The total RNA was extracted from the cells, and quantitative RT-PCR was performed using saeP, saeQRS, and adhE gene-specific primers. Vertical axis represents the relative amount of mRNA against that in the parent strain. Means Ϯ S.D. from three independent experiments are presented. Asterisks indicate a Student's t test p value of less than 0.05. I, overnight cultures of the S. aureus parent strain (NCTC8325-4), the pnpA-disrupted mutant (M1117NC), and the pnpA-disrupted mutant (M1117NC) transformed with an empty vector (pND50) or plasmid carrying intact pnpA (ppnpA) were spotted onto nutrient agar plates containing 5% sheep erythrocytes and incubated overnight at 37°C. RLU, relative light units. saeR-saeS polycistron and are required for saeRS function (29). It was recently suggested that the cvfA gene affects the processing of the saePQRS transcript (27). Quantitative reverse transcriptase (RT)-PCR analysis confirmed that the expressions of the saeP and saeQ genes in the cvfA mutant were decreased compared with the parent strain (Fig. 1, F and G). The expression of saeP and saeQ was higher in the cvfA/pnpA doubledisrupted mutant than in the cvfA-deleted mutant (Fig. 1, F and  G). In addition, in the cvfA-deleted mutant, the expression of various metabolic genes involved in amino acid biosynthesis, acetate metabolism in glycolysis, citric acid cycles, and nutrient transporter genes were altered (supplemental Tables S1 and S2). Quantitative RT-PCR analysis confirmed that expression of the adhE gene, which is a metabolic gene involved in acetate metabolism, was decreased in the cvfA-deleted mutant compared with the parent strain (Fig. 1H). The expression of adhE was higher in the cvfA/pnpA double-disrupted mutant than in the cvfA-deleted mutant. Therefore, the altered expression of saeP, saeQ, and adhE genes in the cvfA-deleted mutant was suppressed by the disruption of pnpA.
We further examined whether the decreased virulence of the cvfA-deleted mutant against animals was blocked by the disruption of pnpA. The LD 50 value of the cvfA-deleted mutant against silkworms or mice was larger than that of the parent strain ( Table 3). The LD 50 value of cvfA/pnpA double-disrupted mutant against silkworms or mice was almost the same as that of the cvfA-deleted mutant. Thus, the decreased virulence of the cvfA-deleted mutant against animals was not attenuated by disrupting pnpA. These findings suggest that the phenotype of the cvfA-deleted mutant was not totally suppressed by the disruption of pnpA.
To evaluate the role of pnpA in S. aureus virulence, we examined the virulence property of a single mutant of pnpA. Hemoly-  Table 4. C, poly(A) polymerization activity of purified PNPase was measured at 37°C for 15 min using ADP as a substrate. Vertical axis represents the amount of ADP incorporated into poly(A), and horizontal axis represents the amount of added PNPase protein. D, mutated PNPases were purified by the same method for wild-type PNPase. Purified proteins (1 g) were analyzed by SDS-PAGE stained with Coomassie Brilliant Blue. E, poly(A) polymerization activities of mutated PNPases were measured using the same method as for wild-type PNPase. F, phosphorolytic activities of wild-type PNPase and mutated PNPases were measured at 37°C using poly(A) as a substrate.

TABLE 3 Animal-killing ability of the S. aureus cvfA/pnpA double-disrupted mutant
The animal-killing abilities of S. aureus strains were examined using silkworm and mouse models. Two-fold serial dilutions of bacterial solutions were injected into the animals, and the cfu values causing 50% of the animals to die (LD 50 ) were determined. -Fold is the LD 50 of the mutant/LD 50 of the parent strain.

Strain
Silkworm LD 50 Mouse LD 50 sin production was increased in the pnpA-disrupted mutant compared with the parent strain (Fig. 1I). The increase in the hemolysin production was blocked by the introduction of a plasmid carrying intact pnpA (ppnpA; Fig. 1I). Thus, pnpA negatively affects hemolysin production. Furthermore, the LD 50 value of the pnpA mutant against silkworms was smaller than that of the parent strain (Table 3). The LD 50 value of the pnpA mutant against mice was slightly smaller than that of the parent strain (Table 3). Thus, the pnpA gene has a negative role in S. aureus virulence. Mutated PNPase without RNA Degradation Activity Loses Complementation Activity against the cvfA/pnpA Double-disrupted Mutant-We constructed mutated PNPase proteins that lack phosphorolytic activity to determine whether the activity is required for the complementation activity of the increased hemolysin production of the cvfA/pnpA double-disrupted mutant. PNPase has two catalytic domains called PH-1 and PH-2 at the N-terminal region ( Fig. 2A). Analysis of PNPase of Streptomyces antibioticus revealed that amino acid substitutions in PH-1 and PH-2 lead to the loss of the phosphorolytic activity of PNPase (30). We constructed E. coli strains overproducing mutated PNPase proteins with substitutions in amino acids that are conserved among bacteria (D96G, R402A/ R403A, H407D, R413D, and D496G) and examined whether these mutated PNPases lose phosphorolytic activity. First, we purified wild-type recombinant PNPase protein from an E. coli strain expressing His 6 -tagged PNPase by ammonium sulfate precipitation and Ni-NTA resin column chromatography (Fig.  2B). The enzyme activity of the wild-type PNPase was measured by poly(A) polymerization, which is a reverse reaction of phosphorolysis (Fig. 2C) (3). The specific activity of the eluted fraction from the Ni-NTA column was 46 mol/15 min/mg of protein, which was 10 times higher than that of the ammonium sulfate precipitate fraction, and the recovery of activity was 33% ( Table 4). Analysis of SDS-PAGE revealed that the purity of the final protein sample was greater than 90% (Fig. 2B). Furthermore, we purified the mutated PNPase by the same method for wild-type PNPase (Fig. 2D). The mutated PNPase R402A/ R403A, H407D, and D496G did not show poly(A) polymerization activity, whereas D96G and R413D showed 10% poly(A) polymerization activity compared with wild-type PNPase (Fig.  2E). Amino acid substitution in enzymes may have different effects on forward and reverse reactions (30). We measured phosphorolytic activity of wild-type and mutated PNPases. The mutated PNPase R402A/R403A, H407D, and D496G showed a loss of phosphorolytic activity, whereas D96G and R413D showed 10% phosphorolytic activity compared with wild-type PNPase (Fig. 2F). Thus, R402A/R403A, H407D, and D496G lost both poly(A) polymerization activity and RNA degradation activity, whereas D96G and R413D retained both activities.
We then transformed the cvfA/pnpA double-disrupted mutant with plasmids expressing wild-type PNPase or mutated  PNPases. The cvfA/pnpA mutant transformed with plasmid expressing wild-type PNPase (ppnpA) decreased hemolysin production compared with the cvfA/pnpA mutant transformed with an empty vector pND50 (Fig. 3A). In contrast, the cvfA/ pnpA mutant transformed with plasmids expressing R402A/ R403A, H407D, and D496G, which showed a loss of RNA deg- radation activity, produced almost the same amount of hemolysins as the cvfA/pnpA mutant transformed with an empty vector pND50 (Fig. 3A). The cvfA/pnpA mutant transformed with plasmids expressing D96G and R413D, which retained RNA degradation activity, produced a smaller amount of hemolysins than the mutant transformed with an empty vector, and the production level was almost same as that of the mutant transformed with ppnpA expressing wild-type PNPase (Fig. 3A). We performed a Western blotting analysis to measure the expression of PNPase in the cvfA/pnpA mutant. Each mutated PNPase other than R413D was expressed at either an equal or greater level as wild-type PNPase in the cvfA/pnpA mutant (Fig. 3B). The R413D mutant PNPase expression was lower than wild-type PNPase in the cvfA/pnpA mutant (Fig.  3B). Thus, loss of complementation activities of R402A/R403A, H407D, and D496G in the cvfA/pnpA mutant is due to the loss of enzymatic activity and not to the expression level of the mutated PNPases. These results suggest that that RNA degradation activity of PNPase protein is necessary for its complementation activity for the phenotype of the cvfA/pnpA mutant. Mutated PNPase without the RNA Binding Domain Loses Complementation Activity against the cvfA/pnpA Double-disrupted Mutant-PNPase carries two RNA binding domains ( Fig. 2A). E. coli-mutated PNPase proteins without RNA binding domains retained more than half the poly(A) polymerization activity and phosphorolytic activity as wild-type PNPase (30). In contrast, the RNA binding domains were required for E. coli cell growth at low temperature. Crystal structural analysis indicated that the RNA binding domains are involved in the trimer formation of PNPase and accelerate the acquisition of substrate RNA (31,32). We examined whether the RNA binding domains of PNPase are required for complementation of the phenotype of the cvfA/pnpA mutant. First, we constructed an E. coli strain overproducing mutated PNPase without the RNA binding domain (residues 623-690) and obtained PNPase ⌬RBD in greater than 95% purity (Fig. 2D). The PNPase ⌬RBD retained poly(A) polymerization activity and phosphorolytic activity (Fig. 2, E and F). We then transformed the cvfA/pnpA mutant with a plasmid expressing PNPase ⌬RBD and examined hemolysin production. The transformed strain expressed PNPase ⌬RBD and produced the same levels of hemolysins as the cvfA/pnpA mutant transformed with an empty vector (Fig.  3, A and B). Thus, although PNPase ⌬RBD retained RNA degradation activity, it lost complementation activity against the phenotype of the cvfA/pnpA mutant. These results suggest that the RNA binding domain of PNPase is required for complementation of the cvfA/pnpA mutant phenotype.
3Ј-Phosphorylated RNA Is Resistant to Degradation by PNPase-Because CvfA cleaves the 2Ј,3Ј-cyclic phosphodiester linkage of 2Ј,3Ј-cyclic RNA and produces 3Ј-phosphorylated RNA (22), PNPase degrades RNA from the 3Ј terminus in the 5Ј direction (3). As revealed above, because the phenotype of the cvfA-deleted mutant was suppressed by the disruption of pnpAencoding PNPase, we hypothesized that the structural conversion of the RNA 3Ј terminus by CvfA prevents its degradation by PNPase. To test this hypothesis, we chemically synthesized 3Ј-hydroxylated RNA (3Ј-OH RNA), 2Ј,3Ј-cyclic RNA, and 3Ј-phosphorylated RNA using the phosphoramidite method ( Fig. 4, A-C), and examined whether the purified PNPase degrades these RNAs. In the case of 3Ј-OH RNA and 2Ј,3Јcyclic RNA, the RNA bands disappeared with increasing amounts of PNPase (Fig. 4D). In contrast, the 3Ј-phosphorylated RNA band did not disappear, even at the highest concentration of PNPase (Fig. 4D). Furthermore, we determined the K m and V max values of PNPase against these RNA substrates by measuring the dose-response curve of the substrate RNA concentration and RNA degradation activity (Fig. 4, E-G). V max values of PNPase against 3Ј-OH RNA, 2Ј,3Ј-cyclic RNA, and 3Ј-phosphorylated RNA were 35, 6, and 1 mol/min/mg protein, respectively (Table 5). In addition, the K m values against each RNA substrate were 22, 42, and 156 M, respectively (Table 5). These results suggest that 3Ј-phosphorylated RNA is resistant to degradation by PNPase.  Effect of Growth Phase and Cold Stress on the Expression of CvfA and PNPase-If CvfA and PNPase competitively regulate S. aureus gene expression, the expression ratio of CvfA and PNPase might change under different culture conditions. S. aureus exotoxin expression is stimulated in the stationary phase by the agr quorum-sensing system, whereas it is inhibited in the exponential phase (34). In E. coli, the expression of PNPase is activated at cold temperatures (35,36). We examined the effects of the growth phase and cold stress on the expression ratio of CvfA and PNPase. Both CvfA and PNPase were constantly expressed from the exponential phase (A 600 ϭ 0.2-1) to the stationary phase (A 600 ϭ 6), and the ratio of CvfA and PNPase did not change (Fig. 5A). In contrast, the amount of PNPase was decreased at 16°C compared with that at 37°C, whereas the amount of CvfA was increased at 16°C compared with that at 37°C (Fig. 5B). These results suggest that the ratio of CvfA and PNPase changes and affects gene expression under certain conditions.
Because the reciprocal expression of CvfA and PNPase was observed in the cold stress condition, we examined whether cvfA affects the expression of pnpA or vice versa. The amount of CvfA was increased in the pnpA-disrupted mutant (Fig. 5C). In contrast, the amount of PNPase was not altered in the cvfA-deleted mutant (Fig. 5C). These findings suggest that pnpA negatively regulates the expression of CvfA.
Expression of CvfA and PNPase in Clinical Isolates-We examined whether CvfA and PNPase are expressed in clinical isolates of S. aureus. S. aureus is a problematic pathogen due to its antibiotic-resistant capacity. We examined 10 clinical isolates of methicillin-resistant S. aureus (MRSA). All tested strains expressed both CvfA and PNPase (Fig. 5D). This finding suggests that the regulation by CvfA and PNPase is not specific to a laboratory strain but is conserved in most S. aureus strains.

DISCUSSION
The findings of this study indicated that the decreased hemolysin production and agr expression in the cvfA-deleted mutant was suppressed by disruption of pnpA-encoding PNPase with 3Ј-to 5Ј-exonuclease activity. The increased hemolysin production in the cvfA/pnpA mutant was complemented by the expression of wild-type PNPase, whereas it was not complemented by the expression of mutated PNPases without RNA degradation activity or without an RNA binding domain (Table 6). Therefore, both RNA degradation activity and the RNA binding domain of PNPase are required for the genetic interaction between cvfA and pnpA. Because the mutated PNPase without an RNA binding domain (PNPase ⌬RBD) retained the RNA degradation activity in vitro, PNPase ⌬RBD may be defective in capturing and degrading specific RNA substrates in vivo. Furthermore, we demonstrated that 2Ј,3Ј-cyclic RNA, a substrate of CvfA, is sensitive to PNPasemediated degradation, whereas 3Ј-phosphorylated RNA, a product of CvfA, is resistant to PNPase-mediated degradation (Fig. 4). These results suggest that an RNA essential for expression of the agr and hemolysin genes is modified to 3Ј-phosphorylated RNA by CvfA and escapes degradation by PNPase (Fig.  6). If the RNA is not modified by CvfA, the RNA will be degraded by PNPase. This model can explain why the cvfA-FIGURE 6. S. aureus hemolysin production via control of RNA stability by CvfA and PNPase. A specific RNA (3Ј-OH RNA) that is required for hemolysin production is cleaved by endonuclease activity of CvfA or other endonucleases and results in the production of 2Ј,3Ј-cyclic RNA. Next, the 2Ј,3Ј-cyclic RNA is converted to 3Ј-phosphorylated RNA by CvfA. 3Ј-OH RNA and 2Ј,3Ј-cyclic RNA are degraded by PNPase, whereas 3Ј-phosphorylated RNA is resistant to PNPase degradation.

TABLE 6 Characteristics of the mutated PNPases
Complementation ϩ indicates that PNPase expression decreased the hemolysin production of the cvfA/pnpA double mutant. Phosphorolysis and Polymerization refer to the biochemical activities of PNPase in vitro.

Phosphorolysis Polymerization Complementation
deleted mutant exhibits decreased hemolysin production as well as why the cvfA/pnpA-disrupted mutant restores hemolysin production. Based on microarray and quantitative RT-PCR analysis, we found that the decreased expression of the sae locus in the cvfAdeleted mutant was suppressed by the disruption of the pnpA gene. Based on the report that the cvfA gene affects the processing of the saePQRS transcript (27), saePQRS mRNA might be a target of CvfA and PNPase. In addition, because expression of the sae locus is positively regulated by the agr locus (34) and cvfA and pnpA regulate the expression of the agr locus in an opposing manner, the effect of cvfA and pnpA on sae expression might be due in part to the altered expression of the agr locus. It is also possible that CvfA and PNPase directly target mRNAs encoding hemolysins and other virulence factors, whose expression is regulated by sae and agr. We also found that the decreased expression of the adhE gene encoding alcohol dehydrogenase, which is involved in acetate metabolism, was suppressed by the disruption of the pnpA gene. The expression of adhE was not affected by either agr (37) or saeRS (38). Thus, CvfA and PNPase target RNA genes other than agr and saeRS to control the expression of adhE, and those genes remain to be identified. As a whole, CvfA and PNPase regulate energy metabolism and virulence in S. aureus. We further demonstrated that the decreased killing abilities of the cvfA-deleted mutant against silkworms and mice were not attenuated by the disruption of pnpA ( Table 3). The finding that agr expression, which is required for virulence in both silkworms (26) and mice (39), was restored in the cvfA/pnpA double-disrupted mutant ( Fig. 1) suggests the presence of other factors that were not restored in the double mutant. These findings suggest that the target RNAs of CvfA are not totally same as those of PNPase. Further studies are needed to reveal the characteristics of the target RNAs of CvfA and PNPase and the mechanism of target recognition by each enzyme.
B. subtilis RNase Y, a homolog of CvfA, has endonuclease activity against mRNA (2,5,6). At present, it is unclear whether S. aureus CvfA has endonuclease activity. 2Ј,3Ј-Cyclic RNA is known to be produced by either endonucleolytic cleavage (40) or by RNA terminal cyclase (41). Bacterial tRNA ligase cleaves immature tRNA to produce 2Ј,3Ј-cyclic tRNA and further cleaves it to produce 2Ј-phosphorylated tRNA (42). Based on these reports, CvfA may be a phosphodiesterase against 2Ј,3Јcyclic RNA that is produced by RNA cleavage by CvfA or some other endonuclease or by an RNA terminal cyclase (Fig. 6).
RNA stability can be controlled by modifying the RNA structure by the addition of a 3Ј-poly(A) tail (43,44) and the formation of hairpin loop structures (45). To our knowledge, this study is the first to suggest that modification of the 3Ј-terminal nucleotide structure of RNA controls RNA stability and regulates bacterial virulence. Controlling RNA stability allows a faster response of gene expression to environmental changes than control of RNA transcription. S. aureus hemolysins function in various infectious stages, including lysis of host cells, escape from cellular immunity, and biofilm formation (46 -49). The regulation of RNA stability by CvfA and PNPase might be important for bacteria to quickly regulate the expression of hemolysin genes according to host environmental changes.
Control of RNA stability by modification of the 3Ј-terminal nucleotide structure of RNA requires little energy consumption and has little effect on the whole RNA secondary structure. Future studies are needed to develop methods to determine the 3Ј-terminal nucleotide structure of endogenous mRNA and to elucidate the biological significance of the control of RNA stability by altering the 3Ј-terminal nucleotide structure.