SpoVG Is a Conserved RNA-Binding Protein That Regulates Listeria monocytogenes Lysozyme Resistance, Virulence, and Swarming Motility

ABSTRACT In this study, we sought to characterize the targets of the abundant Listeria monocytogenes noncoding RNA Rli31, which is required for L. monocytogenes lysozyme resistance and pathogenesis. Whole-genome sequencing of lysozyme-resistant suppressor strains identified loss-of-expression mutations in the promoter of spoVG, and deletion of spoVG rescued lysozyme sensitivity and attenuation in vivo of the rli31 mutant. SpoVG was demonstrated to be an RNA-binding protein that interacted with Rli31 in vitro. The relationship between Rli31 and SpoVG is multifaceted, as both the spoVG-encoded protein and the spoVG 5′-untranslated region interacted with Rli31. In addition, we observed that spoVG-deficient bacteria were nonmotile in soft agar and suppressor mutations that restored swarming motility were identified in the gene encoding a major RNase in Gram-positive bacteria, RNase J1. Collectively, these findings suggest that SpoVG is similar to global posttranscriptional regulators, a class of RNA-binding proteins that interact with noncoding RNA, regulate genes in concert with RNases, and control pleiotropic aspects of bacterial physiology.

tifying lysozyme resistance suppressor mutations via wholegenome sequencing. These suppressor strains were derived in the pgdA mutant background in order to circumvent the identification of mutations that upregulated pgdA. Upon genome sequencing, we observed that four of the five individually derived strains contained an identical mutation in the promoter of an operon encoding two copies of the gene spoVG, resulting in its significant downregulation. spoVG is broadly conserved, especially among Gram-positive bacteria (12), and spoVG mutants display remarkable phenotypes in many species, including reduced methicillin resistance, decreased capsule production, and decreased enzyme secretion in Staphylococcus aureus (13)(14)(15) and altered asymmetric cell division, decreased hemolysin production, and sporulation phenotypes in Bacillus subtilis (16)(17)(18). Additionally, our lab identified spoVG in a separate suppressor screen for mutants that rescued virulence defects of (p)ppGpp-deficient L. monocytogenes (19). Despite these phenotypes and despite being initially characterized nearly 40 years ago (20), the function of the spoVGencoded protein has remained unclear. In this study, we determined that SpoVG is an RNA-binding protein that interacts with noncoding RNAs, regulates genes in cooperation with RNases, and controls pleiotropic aspects of bacterial physiology, including motility, carbon metabolism, and virulence. Together, these characteristics are similar to those of posttranscriptional gene regulators such as CsrA, a class of RNA-binding molecules that are fundamental for synchronizing environmental cues with gene regulation in order to adapt bacteria to their diverse ecological domains (11). Considering that these proteins have been primarily characterized in Gram-negative bacteria, we hypothesize that SpoVG may act as a functionally conserved counterpart to these molecules in Gram-positive organisms.

RESULTS
Suppressor analysis of lysozyme-sensitive mutants. Previous attempts to select lysozyme-resistant suppressor mutants in the rli31 mutant background identified mutations that upregulated pgdA (7). To identify other genes involved with lysozyme resistance, here we generated five independently derived lysozyme-resistant suppressor strains in the pgdA mutant background. Wholegenome sequencing and variant analysis identified differences between these suppressor strains and the parental pgdA strain (Ta-ble 1). All five strains encoded mutations in the essential walRK two-component system (TCS) operon, which upregulates expression of autolysins and other cell wall components (21,22). Three mutations mapped to the response regulator walR, one mapped to the histidine kinase walK, and one mapped to walI, a negative regulator of walRK that we previously characterized as a lysozyme-sensitive mutant in L. monocytogenes (7). These data suggest that increased activation of the WalRK TCS leads to lysozyme sensitivity, while reduced activation leads to increased lysozyme resistance. It is unlikely that WalRK is a direct Rli31 target, however, as walI mutants display gross morphological cell wall phenotypes, such as susceptibility to antibacterial peptides and ␤-lactam antibiotics (7).
In addition to the walRK mutations, four of the five suppressor strains contained an identical mutation in the promoter of the spoVG operon, 14 nucleotides (nt) upstream of the transcriptional start site. This mutation led to a 27-fold decrease in mRNA abundance compared to wild-type (WT) bacteria, as determined by quantitative real-time PCR (qPCR) (data not shown). The spoVG operon encodes two paralogs (with 84% identity to one another) of the gene spoVG (lmo0196 and lmo0197). The function of spoVG and its relationship to Rli31 were unclear, but we observed that the spoVG 5=-untranslated region (UTR) contained 14/14 nucleotides of perfect complementarity to Rli31. Therefore, we chose to focus on understanding the relationship between SpoVG and Rli31.
Mutation of spoVG increases L. monocytogenes lysozyme resistance. A spoVG deletion mutant was constructed that lacked both spoVG paralogs and the 5=-UTR in the WT as well as the ⌬pgdA and ⌬rli31 mutant backgrounds. These mutants were assayed for lysozyme sensitivity along with their parental strains. Deletion of spoVG significantly increased lysozyme resistance of the pgdA mutant (Fig. 1A) and completely restored lysozyme resistance for the ⌬rli31 strain (Fig. 1B). Deletion of spoVG in an otherwise-WT background increased lysozyme resistance to a level greater than that in WT bacteria (Fig. 1C). These data suggest that Rli31 and SpoVG each regulate lysozyme resistance in the absence of the other, but deletion of both genes leads to a neutral, WT phenotype.
Given the intriguing complementarity between Rli31 and the spoVG 5=-UTR, we examined whether the spoVG 5=-UTR affected lysozyme resistance. A chromosomal mutation in the spoVG 5=-UTR was introduced that disrupted the region's complementary to Rli31. However, this mutation did not alter the degree of lysozyme resistance observed in the parental strain and did not affect other observable phenotypes (data not shown). In-frame deletion of the spoVG open reading frames (ORFs), which left the 5=-UTR intact (here referred to as the spoVG ORF mutant) (19), was similar to the spoVG mutant lacking the UTR (Fig. 1C). These findings suggest that the spoVG mutant phenotype is caused by mutation of spoVG and not by the spoVG 5=-UTR.
To investigate the role of each spoVG paralog, premature stop codons were introduced into spoVG I and spoVG II, and these strains were assayed for lysozyme resistance. Mutation of either paralog in the rli31 mutant background did not significantly change lysozyme sensitivity of the rli31 mutant (data not shown), suggesting that the function of the paralogs is redundant. This likely explains why suppressor mutations were identified in the promoter of the operon, which reduced expression of both paralogs, rather than a single spoVG ORF.

Mutation of spoVG increases virulence in vivo.
To determine if spoVG contributed to growth in vivo, infections in mice were performed using ⌬spoVG, ⌬spoVG ⌬rli31, and ⌬spoVG ⌬pgdA strains. Deletion of spoVG significantly increased in vivo growth of the ⌬pgdA strain in both spleens (40-fold rescue) and livers (174fold rescue). Deletion of spoVG also restored virulence of the rli31 mutant (5-fold below WT) back to WT levels in both spleens and livers. In addition, a spoVG mutant in a WT background was 5-fold more virulent than WT bacteria (Fig. 1D). These data demonstrated that deletion of spoVG increases the growth in vivo of lysozyme-sensitive and WT L. monocytogenes.
Characterization of the Rli31 secondary structure. The spoVG 5=-UTR contained a higher degree of complementarity to Rli31 (14/14 nucleotides) than anywhere else on the L. monocytogenes chromosome ( Fig. 2A). To better understand this relationship, we characterized the predicted Rli31 secondary structure. Expression of WT rli31 with its endogenous promoter fully complemented lysozyme sensitivity of the ⌬rli31 mutant (Fig. 2B). Mutations that disrupted the 5= hairpin (mutant A or B) or the 3= hairpin (mutants C or D) were then introduced into rli31, which failed to complement the ⌬rli31 mutant ( Fig. 2B and C). Mutations in the 5= apical loop (mutant E) also failed to complement the ⌬rli31 strain. Compensatory mutations that restored the 5= hairpin (mutant AϩB) or the 3= transcriptional terminator (mutant CϩD) restored lysozyme resistance of ⌬rli31 to WT levels (Fig. 2B). These results support the predicted structure of Rli31 and suggest that Rli31 is composed of a long 5= hairpin and a 3= transcriptional terminator. The complementary region between Rli31 and the spoVG 5=-UTR is encoded in a C-rich apical loop of Rli31.
Rli31 binds the spoVG 5=-UTR in vitro but does not regulate SpoVG mRNA or protein abundance. Given the complementarity between Rli31 and the spoVG 5=-UTR, we asked whether these RNAs interacted in vitro. RNA-RNA electrophoretic mobility shift assays (EMSAs) were performed using 32 P-labeled in vitrotranscribed (IVT) WT and mutant Rli31 and the unlabeled spoVG 5=-UTR. Addition of the spoVG 5=-UTR caused a migration difference of WT Rli31 at a molar ratio of 1:1 but did not alter migration of Rli31 mutant E at a molar excess of 5:1 (Fig. 2C). The spoVG 5=-UTR also altered migration of Rli31 mutant AϩB, while Rli31 mutant A showed an altered migration pattern compared to The indicated strains were grown to mid-exponential phase with shaking in BHI and treated with 1 mg/ml lysozyme. CFU were measured at 1, 2, and 4 h post-addition of lysozyme. A two-tailed P value is reported for each spoVG mutant in comparison to its parental strain. ***, P Ͻ 0.0001; ns, no significant difference. (D) CFU in organs of CD-1 mice that were infected i.v. for 48 h. The data are a combination of the results of three separate experiments totaling at least 14 mice, with the exception of the ⌬pgdA and ⌬pgdA ⌬spoVG strains, which were two separate experiments with a total of 10 mice each. The dotted line indicates the limit of detection, and black bars represent median values. A two-tailed Mann-Whitney t test was used for statistical analysis for each group. ns, no significant difference; *, P Ͻ 0.05; **, P Ͻ 0.01; ***, P Ͻ 0.0001. WT Rli31 (Fig. 2C). Another RNA of similar size (Rli109) was also tested as a negative control, and addition of the spoVG 5=-UTR to Rli109 did not cause a migration difference when we used a molar excess of 15:1 (Fig. 2C, right panel). These data suggest that the spoVG 5=-UTR and the Rli31 apical loop specifically interact in vitro.
We hypothesized that Rli31 regulates spoVG transcript or protein abundance. To evaluate whether Rli31 affects spoVG tran-script stability, qPCR was performed using RNA from WT and ⌬rli31 bacteria. However, mRNA abundance of spoVG was unaltered in the rli31 mutant, and rli31 RNA abundance was unaltered in spoVG mutants (data not shown). To test if rli31 affected spoVG protein abundance, Western blot assays were performed using WT and rli31 mutant bacteria. Surprisingly, no difference in SpoVG protein abundance was observed between WT and rli31 mutants (Fig. 2D). Genes associated with lysozyme resistance are  (51). The red dotted line indicates complementarity between the RNAs. (B) The mutations indicated in panel A were introduced into rli31 on the pIMK plasmid and integrated into ⌬rli31 strain bacteria. Cultures were treated with 1 mg/ml lysozyme at mid-exponential phase, as indicated by the red arrow. Turbidity was monitored at 10-min intervals. Data are representative of results from at least three separate experiments. (C) Approximately 400 ng of the indicated 32 P-labeled in vitro-transcribed RNA was incubated with unlabeled spoVG 5=-UTR RNA at 25°C for 30 min. EMSAs were performed using native PAGE. Molar ratios of unlabeled 5=-UTR to 32 P-labeled RNA are indicated. (D and E) L. monocytogenes lysates were collected from the indicated strains, separated by gel electrophoresis, and imaged by Western blotting using an antibody specific for SpoVG (17). A nonspecific band that reacted with an antibody to LLO was used as a loading control. (E) Lysozyme (200 g/ml) was added to mid-logarithmically growing bacteria in BHI cultured with shaking at 37°C. Bacteria were then harvested at the indicated times, and Western blot assays were performed as described for panel D.
often upregulated in response to lysozyme (23,24), so we also tested whether Rli31 regulated spoVG in response to lysozyme. A time course of lysozyme treatment revealed that SpoVG abundance was similar between WT and ⌬rli31 strain bacteria at all time points (Fig. 2E). These data suggest that, despite their sequence complementarity and despite interacting in vitro, Rli31 does not regulate spoVG mRNA or protein abundance.
SpoVG weakly and nonspecifically interacts with singlestranded DNA in vitro. L. monocytogenes SpoVG was previously reported to bind to DNA (25), and the crystal structure of B. subtilis SpoVG contained two positively charged grooves of similar widths (measured in Angstroms using PyMOL) (Fig. 3A). To assess DNA binding, SpoVG (Lmo0196) carrying a C-terminal sixhistidine epitope tag was purified from Escherichia coli, and EMSAs were performed using 32 P-labeled DNA. EMSAs were then performed with SpoVG and the cap41 promoter from S. aureus (25) and the pgdA promoter from L. monocytogenes (oligonucleotide sequences are described in Table S3 in the supplemental material). The migrations of protein-DNA complexes were plotted against total protein concentrations, and nonlinear regression was used to calculate the apparent K d (dissociation constant) value for single-stranded cap41 binding, which was 1.2 M. Significant binding was not observed for double-stranded DNA (Fig. 3B). spoVG bound to various pgdA promoter probes with similar affinities as cap41 and also bound to a probe corresponding to the pgdA ORF (Fig. 3C). These results suggested that DNA binding was not specific, and we observed that spoVG bound to various scrambled DNA probes with similar affinities as the cap41 and pgdA probes (Fig. 3C).
SpoVG binds multiple RNAs in vitro, including Rli31. Given the genetic interactions between spoVG and rli31 and that SpoVG weakly bound DNA, we next asked if SpoVG interacted directly with Rli31. To assess RNA binding, rli31 was in vitro transcribed and EMSAs were performed using SpoVG-6-His purified from E. coli. To ensure that SpoVG-6-His was not contaminated with Hfq, an RNA-binding protein that adheres to immobilized metal affinity columns (26), a two-step purification was performed that removed all detectable amounts of Hfq, as observed by Western blotting (see Fig. S1 in the supplemental material). EMSAs revealed that SpoVG bound Rli31 with an apparent K d of 273 nM. To test for specificity, five other L. monocytogenes noncoding RNAs were evaluated. RNAs were chosen that had wellestablished 5= and 3= ends and were of similar size to Rli31 (27,28). These probes were 6S, Rli32, signal recognition particle (SRP) RNA, RliI, and Rli109. Two of these molecules, 6S RNA and SRP RNA, are well-described protein-binding RNAs, whereby the 6S RNA specifically interacts with proteinaceous RNA polymerase components (29) and the SRP RNA is part of the highly conserved signal recognition particle (30). Rli32, RliI, and Rli109 are uncharacterized sRNAs in L. monocytogenes (28). These RNAs were in vitro transcribed, EMSAs were performed, and gel shift assays showed a range of affinities for the various sRNAs (Fig. 4A).  Table S3 in the supplemental material.) SpoVG bound to Rli109 with an apparent K d of 129 nM and to Rli32 with a K d of 273 nM. Very weak binding was observed for RliI, and no gel shifts were observed using the RNAs with known protein-binding partners, 6S RNA and SRP RNA (Fig. 4A).
These data suggested that SpoVG bound to RNA with a nearly 10-fold greater affinity than DNA. To directly compare RNA and DNA binding, competition assays were performed using radiolabeled RNA and unlabeled RNA/DNA competitors. SpoVG binding to 32 P-labeled RNA (Rli109) was competed away using a 65: 1 molar excess of unlabeled RNA (Rli109). However, a 750: 1 molar excess of unlabeled DNA competitor (cap41) did not affect RNA binding of SpoVG (Fig. 4B).
To assess which region of Rli31 was bound by SpoVG, the Rli31 mutants described in Fig. 2 were in vitro transcribed and assayed for binding. Mutation of the Rli31 hairpin (mutants A and B) did not affect binding, while mutation of the apical loop (mutant E) resulted in 2-to 4-fold reduced binding (Fig. 4C). Collectively, these results revealed that L. monocytogenes SpoVG is an RNAbinding protein that interacts with the Rli31 5= apical loop in vitro.
spoVG mutant strains of L. monocytogenes are pleiotropic. spoVG mutant strains of S. aureus are defective for secretion of extracellular enzymes (14). To characterize secreted proteins from ⌬spoVG L. monocytogenes strains, supernatants from exponentially growing WT and ⌬spoVG mutant bacteria were precipitated, separated by electrophoresis, and visualized by Coomassie staining. One band appeared for the WT that was lost in the ⌬spoVG mutant, and mass spectrometry identified this protein as flagellin (see Fig. S2A in the supplemental material). Deletion of flaA in L. monocytogenes did not alter lysozyme resistance (data not shown); however, these data suggested that spoVG may be required for proper expression of motility genes. ⌬spoVG appeared motile in liquid culture as observed by phase-contrast microscopy, but during growth in soft agar the spoVG mutant displayed a severe defect in swarming motility, spreading to only 6.6% of the area of WT bacteria (Fig. 5A). After prolonged incubation at 30°C, suppressors were observed that swarmed away from the original colony. These swarming suppressor mutants were isolated and reinoculated into soft agar. After 3 days, the suppressors displayed the smooth colony morphology and increased the swarming area to an average of 60% of WT (Fig. 5B). DNA from six individually derived swarming suppressors was purified and subjected to whole-genome sequencing. Four of the six strains contained unique point mutations in the gene encoding a major RNase, RNase J1 (Table 2). Two other mutations occurred in genes encoding RNA turnover machinery proteins, including the transcription termination protein Rho and the termination factor NusG ( Table 2). The swarming suppressor mutations did not affect lysozyme resistance (see Fig. S2B), suggesting that spoVG independently regulates motility and lysozyme resistance.
SpoVG regulates carbohydrate import genes. We sought to determine what genes were regulated by SpoVG in L. monocytogenes. Transcriptome sequencing (RNA-seq) was performed using RNA from ⌬spoVG that was collected from mid-exponentialphase cultures growing at 37°C in brain heart infusion (BHI) medium. Compared to transcripts from WT bacteria, 577 genes were regulated greater than 2-fold and 101 genes were regulated greater than 4-fold in the spoVG mutant (see Table S1 in the supplemental material). Five of the 16 most upregulated genes encoded phosphotransferase system (PTS) components predicted to import carbohydrates (31). Of these, four genes were predicted to import N,N=-diacetylchitobiose, the disaccharide breakdown product of chitin. In summary, spoVG is required for proper expression of hundreds of L. monocytogenes genes, many of which are involved with carbohydrate metabolism.
Global posttranscriptional regulators, including the CsrA/Rsm family, are a class of RNA-binding proteins that coordinate various aspects of bacterial physiology and are primarily described in Gram-negative bacteria (11,35). In this study, we observed a number of similarities between SpoVG and this class of regulators. For example, both spoVG and csrA mutants are pleiotropic and exhibit defects in motility, carbohydrate metabolism, and virulence (11). CsrA interacts with the noncoding RNA CsrB, which antagonizes its function (36). Similarly, SpoVG bound the apical loop of Rli31 in vitro. While it remains unclear how SpoVG regulates RNA transcripts, CsrA binds mRNA near the RBS to occlude ribosome recruitment. The binding affinity for SpoVG for Rli31 in vitro was weaker than that of CsrA (37); however, we note that Rli31 was among the most abundant RNAs in L. monocytogenes (7) and that the affinity of SpoVG for mRNA transcripts may be higher than for the arbitrarily chosen RNAs described in this report. Future studies will be required to determine the consensus binding motif and affinity of SpoVG for mRNA transcripts.
Another similarity between SpoVG and CsrA is the coregulation of genes with major RNases. CsrA regulates RNA transcripts by protecting targets from cleavage by RNase E, a major RNase in Gram-negative bacteria. CsrA can also remodel mRNA to unveil Rho-binding sites, leading to premature transcriptional termination (11). RNase E is not conserved in Firmicutes, including L. monocytogenes; however, the major RNases J1 and J2 are functional homologs (38). Interestingly, we found that point muta-  tions in RNase J1, Rho, and NusG suppressed the swarming defects of spoVG mutants. We also noted that SpoVG is structurally similar to CsrA, as both proteins are small (less than 100 amino acids), dimeric, and consist of 5 to 7 ␤-strands and a single ␣-helix (11). Lastly, the L. monocytogenes spoVG operon encoded two highly similar spoVG paralogs, and each paralog was redundant for regulating lysozyme resistance and swarming motility. Likewise, multiple paralogs of CsrA are encoded in Legionella pneumophila and they regulate the organism's ability to survive in unique habitats (39). Based on all these findings, we suggest that SpoVG may represent a global posttranscriptional regulator in Grampositive bacteria, similar to the Csr/Rsm family of proteins described in Gram-negative organisms. The Rli31 apical loop contained 14/14 nucleotides of perfect complementarity to the spoVG UTR. Not surprisingly, Rli31 interacted with the SpoVG 5=-UTR in vitro. Additionally, spoVG is regulated by noncoding RNAs in S. aureus (33,40); therefore, we hypothesized that Rli31 regulates SpoVG translation. However, we were surprised to observe that Rli31 did not regulate SpoVG mRNA or protein abundance ( Fig. 2D and E) and that mutating the spoVG 5=-UTR at the region of complementarity to Rli31 did not affect lysozyme resistance or swarming motility (data not shown). The function of the spoVG 5=-UTR remains unresolved; however, we speculate that the UTR negatively regulates Rli31 as an "RNA-RNA" decoy under certain growth conditions. In Salmonella and E. coli, RNA-RNA decoys sequester sRNAs through base-pairing, which regulates chitobiose import (41). Curiously, we observed that 4 of the 16 most upregulated genes in the ⌬spoVG mutant were chitobiose import proteins. Based on the similarities between Rli31/SpoVG and the chitobiose decoy system described in proteobacteria, we speculate that the UTR may function as a means to regulate Rli31 in relation to carbohydrate import. Clearly, the relationship between Rli31 and SpoVG is exceedingly complicated and multifaceted, which may allow L. monocytogenes to fine-tune growth in complex environments.
L. monocytogenes lives as both a saprophyte and a foodborne pathogen. Its pathogenic lifestyle is controlled by the master virulence regulator PrfA, which is a Crp family member that is activated in host cells. PrfA mutants that are locked in the active state (PrfA*) are hypervirulent but suffer during environmental growth (42,43). Since SpoVG regulates motility and carbohydrate metabolism, we suggest that it modulates L. monocytogenes physiology suitable for environmental survival. Additionally, we observed that spoVG mutants bind to the biofilm indicator dye Congo red significantly less than WT bacteria (data not shown), suggesting that SpoVG may positively regulate biofilm formation genes. Like PrfA* mutants, spoVG mutants were hypervirulent, and we suspect that, also like PrfA* mutants, they would suffer during environmental growth. The spoVG paralogs (lmo0196 and lmo0197) occupy the locus adjacent to the PrfA regulon (lmo0200 to lmo0207), which although possibly coincidental may suggest coregulation. Indeed, both genes are modestly regulated by SigB (44) and by PTS-dependent sugar abundance (45). We suggest that, whereas PrfA controls intracellular survival and pathogenesis, SpoVG tilts the balance toward environmental survival.

Ethics statement.
This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Research Council of the National Academy of Sciences (46). All protocols were reviewed and approved by the Animal Care and Use Committee at the University of California, Berkeley (MAUP R235-0815B).
Bacterial strains and microbiological assays. All strains of L. monocytogenes used in this study were in the 10403S background and cultured in BHI medium. For construction of ⌬spoVG mutant strain, a splice overlap extension product was generated with primers TB211/TB214 and cloned into pKSV7 (47), and spoVG was removed by allelic exchange. rli31 complement strains were constructed by amplifying rli31 and the rli31 promoter with TB140 and TB141. This fragment was introduced into L. monocytogenes by using pIMK (48), and mutations were introduced using primers TB14-TB23. Transductions were performed using U153 phage as previously described (7,49). Motility assays were performed as previously described (50). Hen egg white lysozyme (Sigma) was used for all lysozyme assays, and the assays were performed as previously described (7). Lysozyme-resistant pgdA suppressors were derived as described elsewhere (7) by passaging the ⌬pgdA strain with increasing concentrations of lysozyme, until all strains were resistant to 1 mg/ml lysozyme. Lists of all strains and oligonucleotides used in this study are provided in Tables S2  and S3, respectively, in the supplemental material.
In vivo infections. All in vivo infections were performed with Crl: CD1(ICR) (CD-1) mice from Charles River. Mice were infected intravenously (i.v.) with 10 5 logarithmically growing bacteria, and organs were harvested at 48 h postinfection. Organs were homogenized with 0.1% NP-40, and the indicated dilutions were plated onto LB agar.
qPCR. RNA was purified from 20 ml of logarithmically growing bacteria by phenol-chloroform extraction and ethanol precipitation. A 4.4-g aliquot of RNA was DNase treated and reverse transcribed with iScript (Bio-Rad). cDNA levels were measured using SYBR Fast (KAPA) and oligonucleotides specific for the target gene (see Table S3 in the supplemental material).
Western blot analysis. 10 ml cultures of the indicated strains were grown to mid-exponential phase with shaking at 37°C in BHI medium, collected by centrifugation, and lysed by bead beating followed by boiling in SDS loading buffer. Protein abundance was normalized to the optical density at 600 nm (OD 600 ), and soluble proteins were separated by denaturing gel electrophoresis. Membranes were probed using anti-SpoVG antibodies (Linc Sonenshein, Tufts University) and anti-listeriolysin O (LLO) antibodies. Membranes were then probed using the appropriate secondary antibodies (LI-COR), and fluorescence was visualized using an Odyssey imaging system (LI-COR).
Modeling of SpoVG. The crystal structure of SpoVG from B. subtilis (PDB accession number 2IA9) was manipulated using the PyMol molecular graphics system, version 1.7.4 (Schrödinger, LLC) to annotate positively charged residues (R/K/H) and negatively charged residues (E/D).
EMSAs and nucleotide probe preparation. For IVT reactions, PCR products containing T7 promoters were generated (see Table S3 in the supplemental material), and IVT was performed using [␣-32 P]ATP (PerkinElmer) and the MEGAscript T7 transcription kit (Life Technologies). IVT-derived RNAs were DNase treated, diluted in Tris-EDTA (TE buffer), and purified using MicroSpin G-25 columns (GE Life Sciences). EMSA reactions were performed with the indicated amounts of SpoVG in 25 to 30 l volumes, and all reaction mixtures contained nonspecific DNA, RNA, and protein competitors. Unless otherwise stated, between 250 and 500 ng of 32 P-labeled DNA/RNA was included for each reaction mixture. All reaction mixtures contained 2ϫ EMSA buffer (40 mM Tris-HCl, 2 mM MgCl 2 , 20% glycerol, 300 mM NaCl, 5 mM DTT; pH 8.0), 5 g yeast tRNA, 1 g bovine serum albumin, and 100 ng poly(dI-dC). Reaction mixtures were incubated at room temperature for 30 min before separation by 5% native PAGE.