NrnA Is a Linear Dinucleotide Phosphodiesterase with Limited Function in Cyclic Dinucleotide Metabolism in Listeria monocytogenes

ABSTRACT

Listeria monocytogenes produces both c-di-AMP and c-di-GMP to mediate many important cellular processes, but the levels of both nucleotides must be regulated. c-di-AMP accumulation attenuates virulence and diminishes stress response, and c-di-GMP accumulation impairs bacterial motility. An important regulatory mechanism to maintain c-di-AMP and c-di-GMP homeostasis is to hydrolyze them to the linear dinucleotides pApA and pGpG, respectively, but the fates of these hydrolytic products have not been examined in L. monocytogenes. We found that NrnA, a stand-alone DHH-DHHA1 phosphodiesterase, has a broad substrate range but with a strong preference for linear dinucleotides over cyclic dinucleotides. Although NrnA exhibited detectable cyclic dinucleotide hydrolytic activities in vitro, NrnA had negligible effects on their levels in the bacterial cell, even in the absence of the c-di-AMP phosphodiesterases PdeA and PgpH. The ΔnrnA mutant had a mammalian cell infection defect that was fully restored by Escherichia coli Orn. Together, our data indicate that L. monocytogenes NrnA is functionally orthologous to Orn, and its preferred physiological substrates are most likely linear dinucleotides. Furthermore, our findings revealed that, unlike some other c-di-AMP- and c-di-GMP-producing bacteria, L. monocytogenes does not employ their hydrolytic products to regulate their phosphodiesterases, at least at the pApA and pGpG levels in the ΔnrnA mutant. Finally, the ΔnrnA infection defect was overcome by constitutive activation of PrfA, the master virulence regulator, suggesting that accumulated linear dinucleotides inhibit the expression, stability, or function of PrfA-regulated virulence factors.

IMPORTANCE Listeria monocytogenes produces both c-di-AMP and c-di-GMP and encodes specific phosphodiesterases that degrade them into pApA and pGpG, respectively, but the metabolism of these products has not been characterized in this bacterium. We found that L. monocytogenes NrnA degrades a broad range of nucleotides. Among the tested cyclic and linear substrates, it exhibits a strong biochemical and physiological preference for the linear dinucleotides pApA, pGpG, and pApG. Unlike in some other bacteria, these oligoribonucleotides do not appear to interfere with cyclic dinucleotide hydrolysis. The absence of NrnA is well tolerated by L. monocytogenes in broth cultures but impairs its ability to infect mammalian cells. These findings indicate a separation of cyclic dinucleotide signaling and oligoribonucleotide metabolism in L. monocytogenes.

INTRODUCTION

c-di-AMP and c-di-GMP are among the most widespread cyclic dinucleotide second messengers in bacteria. Together, they are produced by all bacterial phyla, but their phylogenetic distributions do not entirely overlap (1). These nucleotides mediate many important aspects of bacterial physiology and pathogenesis, and c-di-AMP is also essential for the growth of many firmicute species in rich laboratory media and in the infected hosts (2, 3). However, cyclic dinucleotide levels must be regulated to avoid toxic accumulation, in part through hydrolysis. Depending on the phosphodiesterases, c-di-AMP and c-di-GMP may be degraded into pApA and pGpG, respectively, or into AMP and GMP (2, 4, 5).

In addition to cyclic dinucleotide hydrolysis, pApA and pGpG are also products of RNA metabolism. During bacterial growth, mRNA decay, abortive transcription initiation, and transcription elongation all generate RNA fragments that are cleaved by various RNases into oligoribonucleotides of 2 to 5 nucleotides, including pApA and pGpG (6, 7). These oligoribonucleotides, also called nanoRNAs, must be further degraded by nanoRNases to avoid detrimental consequences on bacterial growth and physiology. Indeed, the accumulation of oligoribonucleotides is lethal in Escherichia coli (8). Although the mechanisms for their toxic effects are not fully understood, nanoRNAs of 2 to 4 nucleotides can prime and shift transcription start sites at a large number of promoters in Pseudomonas aeruginosa (9). These oligoribonucleotides likely also regulate CRISPR-associated gene expression in Mycobacterium species but do not globally alter the transcriptional profile (10).

For some c-di-AMP- and c-di-GMP-producing bacteria, there appears to be cross talk between cyclic dinucleotide signaling and oligoribonucleotide metabolism. In Staphylococcus aureus, pApA inhibits the c-di-AMP phosphodiesterase activity of GdpP (also called PdeA) (11), and in Pseudomonas aeruginosa, pGpG inhibits the c-di-GMP phosphodiesterase activity of EAL-domain protein RocR (12, 13). A subset of nanoRNases, including P. aeruginosa and Vibrio cholerae Orn, Bacillus subtilis and B. anthracis NrnA, B. subtilis NrnB, and Caulobacter crescentus NrnC, can readily degrade pGpG (14). Indeed, Orn has a strong preference for linear dinucleotides in bacterial cells over longer substrates (15). Accordingly, a P. aeruginosa Δorn mutant, which is severely diminished for pGpG degradation, also accumulates c-di-GMP (12, 13). Furthermore, several NrnA homologs, also called DhhP, can hydrolyze c-di-AMP and c-di-GMP (5). NrnA is the only identified hydrolase for both c-di-AMP and oligoribonucleotides in Mycobacterium species (16, 17).

Listeria monocytogenes is a Gram-positive firmicute bacterium that produces both c-di-AMP and c-di-GMP (18, 19). c-di-AMP directly binds several protein targets and mediates many important cellular processes, such as central metabolism, osmolyte transport, and cell wall integrity (20–22). c-di-AMP homeostasis is critical to L. monocytogenes growth, stress response, and pathogenesis. Whereas c-di-AMP is essential for growth in standard laboratory media and during infection, its accumulation also greatly attenuates virulence (23, 24). Similarly, c-di-GMP levels must be regulated, since its buildup promotes L. monocytogenes cell aggregation and impairs mammalian cell invasion (19, 25). While substantial progress has been made on the molecular targets and regulatory mechanisms of cyclic dinucleotide signaling in L. monocytogenes, much less is known about the fate of their hydrolytic products, pApA and pGpG.

Among the nanoRNases that have been identified in other bacteria, L. monocytogenes encodes only NrnA, which is a stand-alone DHH-DHHA1 phosphodiesterase. An L. monocytogenes nrnA::Himar1 mutant was previously shown to be defective for beta-hemolytic activity and diminished for virulence in the mouse infection model, but the mechanisms are unclear (26). Here, we found that NrnA degrades several substrates in vitro but has a strong preference for linear dinucleotides over cyclic dinucleotides and 3′-phosphoadenosine-5′-phosphate (pAp). Consistent with biochemical data, the ΔnrnA mutant accumulated pApA and pGpG but was unaltered for c-di-AMP and c-di-GMP levels. The absence of NrnA conferred defects in biofilm formation and mammalian cell infection, and both phenotypes were restored by Orn, indicating analogous functions between these proteins. Taking the data together, we conclude that the ΔnrnA mutant phenotypes are likely caused by accumulation of linear dinucleotides.

RESULTS

L. monocytogenes NrnA has several substrates in vitro, with a strong preference for linear dinucleotides.

Like other bacteria in the Firmicutes phylum, L. monocytogenes does not encode an Orn ortholog. Instead, it harbors a single NrnA homolog, encoded by lmo1575, with a stand-alone DHH-DHHA1 catalytic domain that belongs to the DHH phosphodiesterase superfamily (27). L. monocytogenes NrnA is highly similar in amino acid sequence to B. subtilis NrnA, with 54% sequence identity and 63% sequence similarity, and the two sequences are colinear (see Fig. S1 in the supplemental material). The recombinant L. monocytogenes NrnA-6×His protein exhibited readily detectable phosphodiesterase activity toward the model substrate bis-p-nitrophenylphosphate (bis-pNPP). The enzymatic activity was optimal at pH 8.0 and dependent on Mn²⁺, consistent with findings reported for several DHH-DHHA1 phosphodiesterases (28–31) (Fig. S2).

L. monocytogenes produces both c-di-AMP and c-di-GMP, which are hydrolyzed into pApA and pGpG, respectively (19, 23, 24). To understand L. monocytogenes NrnA functions in cyclic dinucleotide metabolism, we first examined its enzymatic activities toward these nucleotide substrates. Under our assay conditions, L. monocytogenes NrnA was active against all these cyclic and linear dinucleotides (Fig. 1; Fig. S3), but activity was much more robust toward pApA and pGpG than toward c-di-AMP and c-di-GMP. Using pApG as a proxy for other oligoribonucleotides unrelated to cyclic dinucleotide signaling in L. monocytogenes, we found that NrnA also robustly hydrolyzed this substrate at affinity and efficiency similar to those of pApA and pGpG (Fig. 1). Finally, because the B. subtilis NrnA homolog also degrades 3′-phosphoadenosine-5′-phosphate (pAp) (32), we tested L. monocytogenes NrnA toward this substrate and observed a readily detectable activity, albeit at a much lower efficiency than that for linear dinucleotides.

FIG 1.

L. monocytogenes NrnA has several substrates in vitro with a strong preference for linear dinucleotides. (A) Enzymatic activities toward all tested substrates. Abbreviations: cdA, c-di-AMP; cdG, c-di-GMP. Reactions were stopped and quantified for substrates and products by HPLC, using a standard curve for each compound. Specific activities are averages from at least three independent experiments. (B) NrnA activities toward pAp, c-di-AMP, and c-di-GMP, replotted from panel A on a different y axis scale.

The DHH motif is invariant among DHH superfamily proteins and is indispensable for NrnA catalytic activity, since the middle His residue of this motif coordinates divalent metals for hydrolysis (28, 30). Consistent with this requirement, L. monocytogenes NrnA activity was completely abolished when the DHH motif was replaced with AAA residues (Fig. S2A and B).

L. monocytogenes NrnA hydrolyzes pApA and pGpG but not c-di-AMP and c-di-GMP in L. monocytogenes broth cultures or during infection.

Since NrnA degraded pApA and pGpG in biochemical assays, we next quantified these nucleotides in L. monocytogenes cultures (Fig. 2A). Compared to the wild type (WT), the ΔnrnA mutant accumulated, on average, ∼20- and ∼13-fold higher pApA and pGpG, respectively. Both nucleotides were restored to at least the WT level upon complementation with an nrnA⁺ allele but not with an nrnA_AAA allele, consistent with the enzymatic function of NrnA in vitro (Fig. 2B and C).

FIG 2.

L. monocytogenes NrnA hydrolyzes pApA and pGpG but has minimal c-di-AMP and c-di-GMP phosphodiesterase activity in L. monocytogenes. (A) Schematic diagram of c-di-AMP and c-di-GMP metabolism in L. monocytogenes. (B to E) Nucleotide levels in bacterial cells grown in BHI medium to mid-exponential phase were quantified by LC-MS/MS, with C¹³N¹⁵-c-di-AMP as an internal standard. In each experiment, the quantified nucleotide of each strain was normalized to the average of WT concentrations, set at 100. Cytoplasmic c-di-AMP and c-di-GMP concentrations for the WT strain were 7.6 ± 0.7 μM and 12 ± 7 μM (standard deviation), respectively. The dacA strain is a genetic depletion of the dacA gene. For all panels, data are averages from at least three independent experiments, and error bars represent standard deviations. For panels B and C, statistical analyses were performed by one-way analysis of variance (ANOVA), followed by Dunnett’s multiple comparisons of each strain with the WT. For panels D and E, statistical analyses were performed by paired t tests comparing strains from each experiment. ns, nonsignificant; ****, P < 0.0001.

In some c-di-AMP- and c-di-GMP-producing bacteria, a feedback regulatory loop has been reported between cyclic dinucleotide levels and their hydrolytic products (11–13). Thus, we sought to determine whether L. monocytogenes NrnA was involved in the regulation of cyclic dinucleotide homeostasis. L. monocytogenes synthesizes c-di-AMP by the diadenylate cyclase DacA and degrades it by at least two phosphodiesterases, PdeA and PgpH (23, 24) (Fig. 2A). In the WT, dacA-depleted, and pdeA pgpH null strains, nrnA deletion had no effect on c-di-AMP levels (Fig. 2D). Additionally, in the pdeA pgpH deletion mutant, which accumulates 4-fold c-di-AMP compared to the WT, the expression of an additional ectopic nrnA⁺ allele did not reduce c-di-AMP levels. Altogether, these data indicate that NrnA has a minor or negligible role in c-di-AMP hydrolysis in L. monocytogenes. Our data also suggest that, in the absence of NrnA, the accumulated pApA did not significantly inhibit c-di-AMP hydrolysis.

L. monocytogenes is an intracellular pathogen that can invade many mammalian cell types and replicate in the infected cell cytosol (33). c-di-AMP secreted by L. monocytogenes in the infected cell cytosol activates the type I interferon response (type I IFN) (18). Thus, we quantified type I IFN to assess c-di-AMP synthesis and secretion in infected macrophages (Fig. 3). As expected, the Δhly mutant, which is trapped in the phagosome, did not induce type I IFN, whereas the tetR::Tn917 mutant, which hypersecretes c-di-AMP, induced a robust response (18). The WT and ΔnrnA mutant elicited similar type I IFN responses, suggesting that these two strains had similar c-di-AMP levels. As also previously shown, the pdeA pgpH null mutant hyperinduced type I IFN, and we found this response to be unaffected by nrnA deletion or expression of an additional ectopic copy (Fig. 3). Thus, these assays indicate that NrnA did not significantly degrade c-di-AMP in cytosolic L. monocytogenes.

FIG 3.

L. monocytogenes NrnA has insignificant c-di-AMP phosphodiesterase activity during infection. (A) L. monocytogenes induces type I interferon response (type I IFN) by c-di-AMP. (B) Type I IFN was quantified by interferon-sensitive response element assays. Bone marrow-derived macrophages were infected with L. monocytogenes, cell supernatants were incubated with ISRE-L929 cells, and type I IFN was quantified based on bioluminescence, corrected for background level induced by reagents and media. Type I IFN responses were normalized to the level detected in uninfected cells in each experiment. Δhly and tetR::Tn917 strains were examined as negative and positive controls, respectively. Data are averages from three independent experiments. Error bars represent standard deviations. Statistical analyses were performed by one-way ANOVA, followed by Dunnett’s multiple comparisons for the indicated pairs. ns, nonsignificant.

In addition to c-di-AMP, L. monocytogenes produces c-di-GMP, which is degraded into pGpG by three EAL-domain phosphodiesterases (19) (Fig. 2A). We found cytoplasmic c-di-GMP and c-di-AMP concentrations to be comparable in brain heart infusion (BHI)-grown L. monocytogenes, at approximately ∼7 μM. The ΔnrnA mutant had a c-di-GMP level similar to that of the WT, again indicating that NrnA had a minor role in cyclic dinucleotide hydrolysis in L. monocytogenes (Fig. 2E).

The L. monocytogenes ΔnrnA mutant is diminished for biofilm formation.

The L. monocytogenes ΔnrnA mutant had no growth defect in BHI broth (a rich medium) or Listeria synthetic medium (LSM). We also examined this mutant for growth under NaCl stress and antibiotics targeting the bacterial cell wall (β-lactams), transcription (rifampin), and translation (erythromycin). The ΔnrnA mutant was indistinguishable from the WT under these culture conditions (Fig. S4).

The P. aeruginosa Δorn mutant is increased for biofilm formation due to an accumulation of c-di-GMP (12–14). At least for exponential-phase cultures, the L. monocytogenes ΔnrnA mutant had a c-di-GMP level similar to that of the WT, so we did not anticipate any change in biofilm formation. Interestingly, we found the ΔnrnA strain to be diminished for biofilm formation in modified LSM, even though its total cell growth had the same density as the WT (Fig. 4A).

FIG 4.

L. monocytogenes ΔnrnA mutant is defective for biofilm formation and mammalian cell infection due to the loss of NrnA enzymatic activity. (A) Biofilm formation assays were performed for bacterial cultures grown in modified Listeria synthetic medium for 40 h under static incubation at 30°C. Biofilm was quantified by staining with 0.3% crystal violet solution, read for absorbance at 595 nm, and normalized to the total culture density (OD₆₀₀) after thorough mixing of biofilm and planktonic growth biomass. In each experiment, biofilm/planktonic growth was calculated as a percentage of WT. (B) Plaque formation at 4 to 6 days postinfection of L2 fibroblasts. Plaque sizes were quantified by ImageJ. In each experiment, the average plaque size by each strain was normalized to the average WT plaque size, set at 100. For both panels A and B, data are averages from at least three independent experiments. Error bars represent standard deviations. Statistical analyses were performed by one-way ANOVA followed by Dunnett’s multiple comparisons with the WT. *, P < 0.05; **, P < 0.01; ****, P < 0.0001.

The L. monocytogenes ΔnrnA mutant has an infection defect, likely due to impaired PrfA-regulated virulence factors.

As an intracellular pathogen, L. monocytogenes replicates in the infected cell cytosol and spreads to neighboring cells to avoid innate immune killing (33). Following host cell entry, the bacterium is enclosed in a phagosomal vacuole and must escape into the cell cytosol through the activities of listeriolysin O (LLO; encoded by the hly gene). Cytosolic bacteria can rapidly replicate and spread to neighboring cells, mediated by ActA, which recruits and polymerizes host cell actin (33). Together with several virulence factors, LLO and ActA are transcriptionally regulated by the master virulence regulator PrfA, which is transcriptionally and allosterically activated during infection (34).

The L. monocytogenes intracellular life cycle can be assessed by plaque formation upon infection of murine fibroblasts (L2 cells). As also previously reported (26), the ΔnrnA mutant exhibited a significant plaque formation defect, with an average plaque size at ∼60% of that formed by the WT (Fig. 4B). Consistent with this plaque defect, an nrnA::Himar1 mutant has also been shown to be diminished for virulence in the mouse infection model (26).

The ΔnrnA plaque formation defect might be due to diminished intracellular growth or cell-to-cell spread. To assess intracellular growth, we monitored the ΔnrnA mutant for replication in bone marrow-derived macrophages and found that it was similar to the WT strain (Fig. 5A and B). We next evaluated cell-to-cell spread by examining actA expression. Because PrfA regulon expression is very weak in BHI broth, we exposed L. monocytogenes cultures to LB broth supplemented with glucose-1-phosphate and charcoal, a condition that activates PrfA through unknown mechanisms (35, 36). Both the WT and ΔnrnA strains upregulated actA expression under this condition (Fig. 5C). Of note, hly gene expression was also similar for both strains, despite the observation that the ΔnrnA mutant is diminished for beta-hemolysis (26).

FIG 5.

L. monocytogenes ΔnrnA mutant has no defect for intracellular growth in infected macrophages or expression of actA, which mediates cell-to-cell spread. (A) Representative intracellular growth curves during infection of bone marrow-derived macrophages. Data are averages from two independent experiments, each with technical duplicates. (B) Relative growth, calculated as (CFU counts at 5 hpi or 8 hpi)/(CFU counts at 2 hpi). Data are averages from 6 independent experiments. (C) Expression of actA and hly genes under PrfA-activating conditions (exposure to glucose-1-phosphate and activated charcoal). In each experiment, gene expression was normalized to that of the housekeeping gene rplD and represented as relative to the WT grown in BHI culture, set at 1. Data are averages from three independent experiments. (D) Plaque formation assays, performed in separate experiments than those for Fig. 4. PrfA* denotes PrfA G145S mutant, which is constitutively active. Data are averages from three experiments. For all panels, error bars represent standard deviations. Statistical analyses in panels B and C were performed by paired t tests for strains in each experiment. Statistical analyses in panel D were performed by one-way ANOVA followed by Dunnett’s multiple comparisons with the WT. ns, nonsignificant; *, P < 0.05; ****, P < 0.0001.

In addition to hly and actA, PrfA also upregulates several other virulence genes essential for the intracellular life cycle (33). Although PrfA expression and activity are controlled by multiple mechanisms, there are several PrfA variants that are constitutively active, independent of these regulations (37, 38). The ΔnrnA plaque defect was completely rescued upon allelic replacement of wild-type PrfA with PrfA G145S (denoted PrfA*), which is constitutively active (Fig. 5D). This indicates that the ΔnrnA strain is defective for PrfA activation or diminished for the expression, stability, or activity of virulence factors within the PrfA regulon.

The ΔnrnA mutant phenotypes are likely due to oligoribonucleotide accumulation.

For both biofilm formation in modified LSM and plaque formation during L2 cell infection, the ΔnrnA phenotypes were restored upon complementation with the WT nrnA⁺ allele but not the nrnA_AAA (DHH→AAA) mutant allele, indicating that the biological function of NrnA is associated with its enzymatic activity (Fig. 4, Fig. S5). Among different substrates tested in vitro, our analyses revealed that NrnA likely does not degrade c-di-AMP and c-di-GMP in L. monocytogenes. Thus, we sought to examine the physiological relevance of oligoribonucleotides and pAp.

In E. coli, M. tuberculosis, and B. subtilis, pAp is a by-product of sulfate assimilation, encoded by cysH, cysDNC, or analogous genes and operons (39–41). L. monocytogenes does not harbor any of these genes, but we could not exclude the possibility that it could still produce pAp from other unidentified pathways. Several bacteria encode CysQ that specifically degrades pAp, whereas Orn is specific for oligoribonucleotides (15, 41, 42). To distinguish these substrates in the bacterial cells, we complemented the ΔnrnA mutant with the orn or cysQ gene from E. coli. In both biofilm and plaque assays, Orn completely restored ΔnrnA defects, whereas CysQ had no detectable effect (Fig. 6). These data suggest that L. monocytogenes NrnA is functionally analogous to Orn. Thus, the ΔnrnA phenotypes were caused by an accumulation of oligoribonucleotides, most likely linear dinucleotides such as pApA and pGpG.

FIG 6.

L. monocytogenes NrnA is functionally analogous to Orn. (A and B) Biofilm formation assays (A) and plaque formation assays (B), performed as for Fig. 4. The orn and cysQ genes were cloned from E. coli MG1655. All complementing genes were expressed under a strong constitutive promoter (P_hyper). Data are averages from at least three independent experiments and were normalized to the WT level in each experiment. Error bars represent standard deviations. Statistical analyses were performed by one-way ANOVA followed by Dunnett’s multiple comparisons with the WT; ns, nonsignificant; *, P < 0.05; **, P < 0.01; ****, P < 0.0001.

DISCUSSION

By utilizing c-di-AMP and c-di-GMP signaling, L. monocytogenes must regulate the levels of these second messengers. The previously identified phosphodiesterases for these cyclic dinucleotides degrade them into their corresponding hydrolytic products, pApA and pGpG (19, 23, 24). Their subsequent degradation to mononucleotides has not been examined in this bacterium. Among the reported oligoribonucleotidases, L. monocytogenes encodes a single homolog, NrnA. Here, we found that L. monocytogenes NrnA robustly degrades pApA and pGpG into the mononucleotides AMP and GMP, respectively. Although the purified L. monocytogenes NrnA enzyme can degrade c-di-AMP and c-di-GMP, the catalytic efficiencies toward these nucleotides were at least 200- to 300-fold lower, suggesting that these cyclic dinucleotides are unlikely physiological substrates in L. monocytogenes. The ΔnrnA mutant exhibited biofilm formation and infection defects, most likely due to the accumulation of dinucleotides such as pApA and pGpG.

In the P. aeruginosa Δorn mutant, the accumulated pGpG inhibits EAL-domain phosphodiesterases, thereby elevating c-di-GMP levels (12, 13). Similarly, the S. aureus Δpde2 mutant (analogous to L. monocytogenes ΔnrnA) accumulates both pApA and c-di-AMP (11). The accumulation of c-di-AMP might be attributed to the inhibitory effect that pApA exerts on GdpP, the major c-di-AMP phosphodiesterase in this bacterium. Additionally, this phenotype likely also reflects the hydrolytic activity of Pde2 toward c-di-AMP in S. aureus, since the ΔgdpP Δpde2 mutant is further elevated for c-di-AMP compared to the ΔgdpP strain. In contrast, L. monocytogenes appeared to separate cyclic dinucleotide and oligoribonucleotide metabolism. We found L. monocytogenes NrnA to have negligible effects on c-di-AMP levels, even in the absence of both PdeA and PgpH, indicating that NrnA is specialized for oligoribonucleotide hydrolysis. Furthermore, unlike P. aeruginosa and S. aureus, the L. monocytogenes ΔnrnA mutant did not exhibit elevated c-di-AMP or c-di-GMP levels despite a substantial accumulation of pApA and pGpG. This might reflect the absence of product inhibition on cyclic dinucleotide phosphodiesterases. Alternatively, L. monocytogenes might harbor additional oligoribonucleotideases, and nrnA deletion alone might not accumulate enough pApA and pGpG for discernible inhibition of PdeA and EAL-domain phosphodiesterases in the bacterial cell, respectively. Among previously characterized nanoRNAses, L. monocytogenes also encodes YhaM. However, previous evidence suggests that YhaM is unlikely a physiological phosphodiesterase for dinucleotides. B. subtilis YhaM degrades 5-mer nanoRNA into 2-mer products and appears to have minimal activity toward dinucleotides (14, 43). Consistent with biochemical studies, B. subtilis YhaM only weakly complements the E. coli Δorn mutant growth defect and P. aeruginosa Δorn biofilm formation phenotype (14, 43).

Both c-di-GMP and c-di-AMP have been shown to regulate bacterial biofilm formation. Elevated c-di-GMP levels promote the motile-sessile transition through complex regulations of the bacterial flagellar motor and extracellular polysaccharide synthesis (2). The P. aeruginosa Δorn strain exhibits increased bacterial cell aggregation and exopolysaccharide production, but these phenotypes are attributed to c-di-GMP accumulation, since accumulated pGpG inhibits the EAL-domain RocR phosphodiesterase (13). The mechanisms by which c-di-AMP regulates biofilm formation are much less studied and appear divergent among species (44–47). We found the L. monocytogenes ΔnrnA mutant to be diminished for biofilm formation. Consistent with our finding, nrnA gene expression is upregulated in L. monocytogenes biofilm (48). For exponential-phase L. monocytogenes cultures, we found no discernible function for NrnA in c-di-AMP and c-di-GMP hydrolysis, but it remains possible that this activity is more prominent under biofilm-forming conditions. Additionally, in the absence of NrnA, the accumulated oligoribonucleotides might inhibit the expression of genes for extracellular matrix biosynthesis. Indeed, oligoribonucleotides have been shown to likely regulate the transcription of CRISPR-associated genes in Mycobacterium species (10).

The L. monocytogenes nrnA::Himar1 mutant was previously shown to have a mouse virulence defect, and we made similar observations here using plaque formation assays in L2 cells (26). Because the ΔnrnA mutant was fully complemented by the E. coli orn⁺ allele, we hypothesize that L. monocytogenes NrnA is functionally analogous to Orn and, therefore, preferably degrades linear dinucleotides such as pApA and pGpG (15). Short oligoribonucleotides have been shown to cause global shifts in transcriptional start sites in P. aeruginosa (9). It is unclear whether a similar phenomenon occurs in the L. monocytogenes ΔnrnA mutant, since it exhibited no obvious growth or stress response defects in broth cultures, and at least hly and actA gene expression was unaffected under PrfA-activating conditions. Additionally, the ΔnrnA plaque defect was completely rescued upon PrfA* expression, indicating that it is caused by impaired PrfA function or diminished virulence factors within the PrfA regulon. Future studies examining the effects of oligoribonucleotides on the transcription, translation, and posttranscriptional regulations of these virulence factors will yield exciting novel roles for oligoribonucleotides in bacterial pathogenesis.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

L. monocytogenes strains used in this study are listed in Table S1 in the supplemental material. L. monocytogenes cultures were grown in brain heart infusion (BHI) broth or Listeria synthetic medium (LSM) at 37°C (21). Complementation of the ΔnrnA mutant was accomplished by expressing L. monocytogenes nrnA, E. coli MG1655 orn, and cysQ genes under a P_hyper promoter in the pPL2 plasmid, which stably integrates into the L. monocytogenes chromosome (49). Gene deletions were performed by allelic exchange using pKSV7 as previously described (24).

Biofilm formation assays.

Biofilm formation assays were performed for cultures grown in LSM containing 12.5 μg/ml cysteine (modified LSM) at 30°C as previously described (21, 50). Briefly, overnight cultures were inoculated into 200 μM modified LSM in a polystyrene 96-well plate and incubated statically for 40 h. Total growth was assessed by measuring the optical density at 595 nm (OD₅₉₅) of the cultures after careful resuspension of all biomass in the wells. For biofilm quantification, culture supernatants were removed and wells were gently washed with sterile MilliQ water. After the wells were air dried, biofilm was stained with 0.3% (wt/vol) crystal violet, solubilized with 90% ethanol, and quantified by measuring absorbance at 595 nm.

Protein expression and purification.

The nrnA gene (lmo1575) from L. monocytogenes 10403S was PCR amplified and cloned into pET20b vector with NdeI and XhoI restriction sites, generating an NrnA-6×His construct. The recombinant NrnA-6×His protein was expressed in E. coli Rosetta (DE3) strain and purified as previously described (51). After purification, NrnA-6×His was buffer exchanged using PD-10 columns (GE Healthcare) into storage buffer (40 mM Tris, [pH 8], 10 mM NaCl). Protein concentration was measured by Bradford assay (Bio-Rad).

Enzyme assays.

Enzyme assays were performed in reaction mixtures containing reaction buffer (100 mM Tris [pH 8], 20 mM KCl, 100 μM dithiothreitol [DTT], 2 mM MnCl₂), NrnA, and various substrate concentrations and incubated at 37°C. NrnA concentrations used for reactions with different substrates were 50 nM for bis-p-nitrophenylphosphate (bis-pNPP), 1 μM for c-di-AMP and c-di-GMP, 15 nM for pAp, and 0.2 nM for pApA, pGpG, and pApG. Activity toward bis-pNPP was monitored continuously by absorbance at 410 nm. For enzyme assays with varied divalent cations, 1 mM each metal was used. For enzyme assays under different pH conditions, 100 mM Tris-HCl or Tris base was used. Reactions with nucleotide substrates were stopped at 0, 5, 10, 15, and 30 min by heating at 95°C for 5 min and analyzed on a Dionex UltiMate 3000 high-performance liquid chromatograph (HPLC) equipped with a DAD-3000 diode array detector. For HPLC, 20 μl of the supernatant was injected into a reverse-phase column (C₁₈, 3.9 by 150 mm, 4 μm; Waters Nova-Pak) equilibrated with 100 mM potassium phosphate, pH 6. Nucleotides were eluted by a mobile phase containing 100 mM potassium phosphate, pH 6, over a gradient of up to 40% methanol. Concentrations of remaining substrates and formed products were calculated based on standard curves for each nucleotide.

L2 plaque assay.

Plaque assays were performed using L2 fibroblasts as previously described (52). Briefly, L2 cells were plated onto a 6-well dish at 1.2 × 10⁶ cells per well and infected with L. monocytogenes at a multiplicity of infection of 0.5. At 1 h postinfection, 50 μg/ml gentamicin was added to cell culture media to kill extracellular bacteria. At 4 to 6 days postinfection, cells were stained with 0.3% crystal violet and imaged for plaques. Areas of plaques were quantified with the ImageJ software (https://imageJ.nih.gov).

Quantitative reverse transcription-PCR.

L. monocytogenes cultures were grown in BHI broth at 37°C to an OD₆₀₀ of 0.7, washed, and resuspended in LB broth containing 25 mM glucose-1-phosphate and 0.2% activated charcoal. Cultures were harvested immediately before and 1 h following resuspension for RNA extraction. RNA samples were treated with Turbo DNase (Ambion, Thermo Fisher) and converted to cDNA using an iScript cDNA synthesis kit (Bio-Rad), and gene expression was quantified using iTaq universal SYBR green (Bio-Rad) with primers specific to each target, with rplD as the control housekeeping gene (see Table S2 in the supplemental material).

Intracellular growth curves.

C57BL/6 immortalized bone marrow-derived macrophages (iBMMs) were grown in Dulbecco’s modified Eagle’s medium (DMEM) with 20% heat-inactivated fetal bovine serum (FBS), 2 mM sodium pyruvate, 1 mM l-glutamine, and 1% β-mercaptoethanol. Infection and intracellular growth assays were performed as previously described (53). Briefly, iBMMs were plated onto a 24-well plate at 0.4 × 10⁶ cells per well and infected with L. monocytogenes at an MOI of 1. At 30 min postinfection, cells were washed with PBS and replenished with fresh medium containing 50 μg/ml gentamicin. At indicated time points, cell medium was removed, cells were lysed with MilliQ water, and serial dilutions of cell lysates were plated onto LB agar to assess bacterial burdens.

Type I interferon assays.

IFN-β production by iBMMs was detected using the type I IFN reporter cell line, ISRE-L929, as previously described (18). Briefly, ISRE-L929 cells were grown in DMEM with 5% heat-inactivated FBS and plated onto a 96-well plate at 5 × 10⁵ cells per well. iBMMs were infected with L. monocytogenes as described for intracellular growth curves. At 5 h postinfection, iBMM cell culture supernatants were removed and incubated with ISRE-L292 cells for 4 h. Uninfected iBMM supernatant was used as the background control. Following incubation, L929 cells were lysed with 40 μl TNT buffer (20 mM Tris base [pH 8], 100 mM NaCl, 1% Triton X-100). Finally, L292 cell lysates were mixed with 40 μl of luciferase substrate solution (20 mM tricine, 2.67 mM MgSO4·7H₂O, 0.1 mM EDTA, 33.3 mM DTT, 530 μM ATP, 270 μM acetyl coenzyme A lithium salt, 470 μM luciferin, 5 mM NaOH, 265 μM magnesium carbonate hydroxide). Bioluminescence was measured using a Synergy H1 plate reader (BioTek).

Quantification of nucleotides by LC-MS/MS.

Quantification of c-di-AMP and c-di-GMP from L. monocytogenes cultures was performed as previously described (24). Briefly, L. monocytogenes strains were grown in BHI to mid-log phase (OD₆₀₀, ∼0.5). Bacterial cell pellets from 0.5-ml cultures were resuspended in 50 μl of 0.25 μM C¹³N¹⁵-c-di-AMP, used as an internal standard, and lysed by sonication. Fractions from methanol extraction were pooled, dried by evaporation, and resuspended in 50 μl of MilliQ H₂O. Liquid chromatography-tandem mass spectrometry (LC-MS/MS) was performed on an Agilent 6460 Triple Quad LC-MS/MS with a 1260 HPLC. Chromatographic separation was performed with an analytical Synergi 4u Hydro-RP 80A column (50 by 2 mm, 4 μm; Phenomenex). Mass transitions for c-di-AMP quantification were previously published (24). c-di-GMP was detected based on three transitions, +691/152, +691/540, and +691/248, and quantified using the +691/152 transition. For each nucleotide, the peak areas of the quantifier transition were divided by that of C¹³N¹⁵-c-di-AMP (+689/146) in the same sample.

pGpG and pApA were detected by ultrahigh-performance LC (UPLC)-MS/MS using an Acquity ultra performance LC system (Waters) coupled with a Xevo TQ-S mass spectrometer (Waters) with an electrospray ionization source in negative ion mode. The MS parameters were the following: capillary voltage, 1.0 kV; source temperature, 150°C; desolvation temperature, 400°C; cone gas, 120 L/h. Five microliters of each sample was separated in reverse phase using Acquity UPLC premier BEH C₁₈, 2.1 by 100 mm, 1.7 μm particle size, VanGuard FIT at a flow rate of 0.3 ml/min with the following gradients of solvent A (8 mM DMHA [N,N-dimethylhexylamine] plus 2.8 mM acetic acid in water, pH ∼9) to solvent B (methanol): t = 0 min, 100% A to 0% B; t = 10 min, 60% A to 40% B; t =10.5, 100% A to 0% B; t = 15 min, 100% A to 0% B (end of gradient). The conditions of the multiple reaction monitoring transitions for cone voltage (V) and collision energy (eV) were pApA, 677/136 and 677/410 (70/34 and 70/22); pGpG, 709/152 and 709/558 (70/34 and 70/22).

Western blotting.

The nrnA⁺ and nrnA_AAA alleles were PCR amplified with C-terminal FLAG tags and cloned into the pPL2 plasmid under the P_hyper promoter. Cultures were grown in BHI to mid-log phase (OD₆₀₀, ∼0.5). Cell pellets were washed in phosphate-buffered saline, resuspended in lysis buffer (20 mM Tris-HCl [pH 7.5], 150 mM NaCl, 0.1% Tween 20), and lysed by sonication. Following SDS-PAGE, proteins were transferred to a nitrocellulose membrane, which was then blocked with 5% nonfat milk for 1 h at room temperature. FLAG-NrnA was blotted using anti-FLAG M2 antibody (Sigma) at 1:2,500. The secondary antibody was horseradish peroxidase-conjugated anti-mouse polyclonal antibody (Promega). The protein ladder (Precision Protein; Bio-Rad) was detected by StrepTactin-horseradish peroxidase conjugate (Bio-Rad). The membrane was stained with a Clarity Western ECL kit (Bio-Rad) for visualization.