Mutation of the Transcriptional Regulator YtoI Rescues Listeria monocytogenes Mutants Deficient in the Essential Shared Metabolite 1,4-Dihydroxy-2-Naphthoate (DHNA)

Listeria monocytogenes, a Gram-positive, facultative intracellular pathogen, survives and replicates in the cytosol of host cells. Synthesis of 1,4-dihydroxy-2-naphthoate (DHNA), an intermediate of menaquinone biosynthesis, is essential for cytosolic survival of L. monocytogenes independent from its role in respiration. Here, we demonstrate that DHNA is essential for virulence in a murine model of listeriosis due to both respiration-dependent and -independent functions.

ABSTRACT

Listeria monocytogenes, a Gram-positive, facultative intracellular pathogen, survives and replicates in the cytosol of host cells. Synthesis of 1,4-dihydroxy-2-naphthoate (DHNA), an intermediate of menaquinone biosynthesis, is essential for cytosolic survival of L. monocytogenes independent from its role in respiration. Here, we demonstrate that DHNA is essential for virulence in a murine model of listeriosis due to both respiration-dependent and -independent functions. In addition, DHNA can be both secreted and utilized as an extracellular shared metabolite to promote cytosolic survival inside host macrophages. To understand the role(s) of DHNA in L. monocytogenes intracellular survival and virulence, we isolated DHNA-deficient (ΔmenD strain) suppressor mutants that formed plaques in monolayers of fibroblasts. Five ΔmenD suppressor (mds) mutants additionally rescued at least 50% of the cytosolic survival defect of the parent ΔmenD mutant. Whole-genome sequencing revealed that four of the five suppressor mutants had independent missense mutations in a putative transcriptional regulator, ytoI (lmo1576). Clean deletion and complementation in trans confirmed that loss of ytoI could restore plaquing and cytosolic survival of DHNA-deficient L. monocytogenes. RNA-seq transcriptome analysis revealed five genes (lmo0944, lmo1575, lmo1577, lmo2005, and lmo2006) expressed at a higher level in the ΔytoI strain than in the wild-type strain, whereas two genes (lmo1917 and lmo2103) demonstrated lower expression in the ΔytoI mutant. Intriguingly, the majority of these genes are involved in controlling pyruvate flux. Metabolic analysis confirmed that acetoin, acetate, and lactate flux were altered in a ΔytoI mutant, suggesting a critical role for regulating these metabolic programs. In conclusion, we have demonstrated that, similar to findings in select other bacteria, DHNA can act as a shared resource, and it is essential for cytosolic survival and virulence of L. monocytogenes. Furthermore, we have identified a novel transcriptional regulator in L. monocytogenes and determined that its metabolic regulation is implicated in cytosolic survival of L. monocytogenes.

INTRODUCTION

Listeria monocytogenes is a facultative intracellular pathogen and the causative agent of listeriosis. Periodic outbreaks of listeriosis, especially in at-risk populations such as pregnant women, the elderly, and immunocompromised people, can be deadly, with fatality rates up to 30% (1). L. monocytogenes is exquisitely adapted to invade and colonize the cytosol of both phagocytic and nonphagocytic host cells (2). L. monocytogenes utilizes a well-characterized arsenal of virulence factors to access the cytosol, proliferate, and spread from cell to cell (3). Importantly, access to the cytosol alone is not sufficient for bacterial replication as mislocalization of non-cytosol-adapted bacteria does not facilitate replication but in many cases leads to cytosolic killing of the bacteria (4–8). Taken together, these data suggest that L. monocytogenes and other cytosol-adapted bacteria possesses unique adaptations to efficiently utilize host nutrients and tolerate cytosolic stresses.

Recent reports have identified bacterial genetic determinants for L. monocytogenes and Francisella sp. survival in host cells, and disruption of these pathways leads to bacteriolysis during cytosolic replication (9–14). Many of the bacterial determinants of cytosolic survival appear to be responsible for key metabolic adaptations during cytosolic replication (11, 12, 14). One such pathway is the synthesis of 1,4-dihydroxy-2-naphthoate (DHNA) which, through a yet undefined mechanism, protects L. monocytogenes from cytosolic stresses and prevents intracellular bacteriolysis of L. monocytogenes (12). L. monocytogenes mutants that are deficient in DHNA are also completely attenuated for virulence (12, 15–17).

DHNA is a quinone intermediate produced during biosynthesis of menaquinone (MK), the sole lipid electron carrier in L. monocytogenes (18). MK transfers electrons between NADH dehydrogenases and terminal cytochrome oxidases to generate a membrane potential and drive aerobic respiration (19). MK is synthesized from chorismate through a series of enzymatic steps encoded by men genes (see Fig. S1 in the supplemental material). In later steps of the pathway, O-succinylbenzoyl-coenzyme A (CoA) is converted to DHNA by MenB and a predicted thioesterase. DHNA is then prenylated by MenA, a DHNA prenyl transferase, to make demethylmenaquinone (DMK), and finally DMK is methylated by MenG to produce MK (18). Since DHNA production is intimately linked to MK biosynthesis, some phenotypes previously associated with MK/respiration deficiency in men mutants may instead stem from loss of DHNA. Indeed, DHNA is a shared resource that is produced and secreted by Lactobacillus spp. and Propionibacterium spp. and can be scavenged by Bifidobacterium spp. and Streptococcus spp. in complex communities (20–22). In addition, DHNA suppresses proinflammatory cytokine production and can protect interleukin-10 (IL-10)-deficient mice from colitis, further demonstrating the importance of DHNA as more than simply a biosynthetic intermediate (23). Finally, our previous work suggests that L. monocytogenes requires DHNA for cytosolic survival and virulence, independent of its role in respiration and the electron transport chain (12).

How DHNA contributes to cytosolic survival and virulence, independent of respiration, in L. monocytogenes is unknown. In this report, we demonstrate a role for DHNA, independent of MK, for growth in a minimal defined medium and show that, similar to other organisms, L. monocytogenes can both share and utilize DHNA as a shared resource through metabolite exchange. Utilizing a plaque-based suppressor screen, we identified a series of suppressor mutations that restored plaquing and cytosolic survival to DHNA-deficient L. monocytogenes. Whole-genome sequencing revealed that four of the five most potent suppressors had independent mutations in lmo1576 (also known as ytoI in Bacillus subtilis or spxR in Streptococcus pneumoniae), a putative transcriptional regulator of unknown function. RNA-seq transcriptome analysis and subsequent metabolic analysis demonstrated that YtoI is a regulator of fermentative flux and that its homolog in S. pneumoniae, SpxR, is required for virulence (24). These data suggest that defects in fermentative metabolism may be responsible for poor cytosolic survival of DHNA-deficient L. monocytogenes.

RESULTS

DHNA is required for in vitro growth and virulence independent of the electron transport chain.

Synthesis of the MK intermediate DHNA in L. monocytogenes is crucial for cytosolic survival and virulence of L. monocytogenes in vivo (12). Disruption of early men genes abolishes synthesis of both DHNA and the downstream product MK and likely has pleiotropic effects on L. monocytogenes physiology and virulence (12). To understand specific contributions of DHNA relative to those of MK, we compared phenotypes of ΔmenD mutants which are missing both DHNA and MK to those of ΔmenA mutants, which lack only MK. To test the hypothesis that the virulence defects of DHNA-deficient mutants are due to more than loss of respiration, we used an acute murine listeriosis infection model. As expected, growth of respiration-defective mutants lacking menA was highly attenuated in vivo. Importantly, however, growth of ΔmenD mutants lacking DHNA was 2 logs more attenuated than that of ΔmenA strains (Fig. 1).

FIG 1.

DHNA is required for virulence in vivo. Bacterial burdens from the spleen (A) and liver (B) were enumerated at 48 h postinfection. Data are representative of results from two biological replicates. Mann-Whitney statistical analysis was performed to measure statistical significance in comparison to results for the ΔmenD mutant. Horizontal bars represent limits of detection. **, P ≤ 0.01.

To further dissect the respiration-independent roles of DHNA, we compared in vitro growth of ΔmenD mutants to that of ΔmenA mutants. When grown aerobically in either brain heart infusion (BHI) or minimal defined medium, both ΔmenD and ΔmenA mutants demonstrated growth defects relative to growth of the wild-type, as expected due to their inability to respire (Fig. 2A and B). Importantly, in both media the ΔmenD mutant displayed more pronounced defects, particularly in minimal defined medium, in which a ΔmenA mutant grew almost as well as the wild-type whereas a ΔmenD mutant was almost completely impaired for growth (Fig. 2A and B); this suggests that BHI medium may supply other factors important for growth of DHNA-deficient strains. As expected, supplementation with either MK or DHNA rescued growth of the ΔmenD mutant in vitro, confirming that this mutant is starved for both DHNA and MK.

FIG 2.

DHNA is required for growth of L. monocytogenes in vitro independent of the electron transport chain. The wild-type, ΔmenD, or ΔmenA strain was grown in aerated cultures of BHI medium (A) or minimal medium (B) at 37°C. The OD₆₀₀ was monitored for 12 to 24 h. Data represent one of three biological replicates. (C) The wild-type, ΔmenD, or ΔmenA strain was grown in aerated BHI cultures at 37°C until mid-late logarithmic phase and then examined for membrane potentials. Data represent the means ± standard errors of the means of three biological replicates normalized to the wild-type level (100%). Where indicated, cultures were supplemented with 5 μM DHNA or 5 μM MK. *, P ≤ 0.05; ***, P ≤ 0.001.

Since MK is an important component of the electron transport chain (ETC), we next examined the functionality of the ΔmenD and ΔmenA mutants’ ETC by measuring their membrane potentials. As previously reported (12), neither the ΔmenD nor ΔmenA mutant generated a membrane potential, consistent with the fact that MK is required for electron transport chain function and that there are no functional MenA or MenD homologs (Fig. 2C). MK supplementation to cultures restored the ability of both the ΔmenD and ΔmenA strains to generate membrane potentials; however, supplementation with DHNA did not rescue the ΔmenD mutant’s ability to generate a membrane potential, consistent with its inability to fully restore growth to wild-type levels under aerated conditions (Fig. 2A). Together, these data highlight the respiration-independent functions of DHNA and suggest that, unlike other organisms which capture extracellular DHNA for DMK/MK biosynthesis (22), L. monocytogenes cannot efficiently synthesize MK from extracellular DHNA to reestablish the respiratory chain. Since the ΔmenB mutant, which has a deletion in the final step of DHNA synthesis (Fig. S1), exactly phenocopies the ΔmenD mutant in growth and virulence (Fig. S2A) (12), the ΔmenD mutant was used for the remainder of this study.

DHNA is released by L. monocytogenes both in vitro and in the host cytosol.

DHNA produced and released by Propionibacterium freudenreichii ET-3 acts as a growth factor for Bifidobacterium (20). Similarly, both Gram-positive and Gram-negative bacteria release DHNA into the culture medium and cross-feed neighboring bacteria (21, 22). Therefore, we asked whether DHNA produced by L. monocytogenes could stimulate growth of DHNA-deficient strains. During growth of the ΔmenD mutant in minimal medium, we supplemented cultures with the cell-free supernatants from wild-type or ΔmenA cultures grown in minimal medium. Supernatants of both the wild-type and ΔmenA strains, but not of the ΔmenD mutant, could rescue growth of ΔmenD mutants in vitro (Fig. 3A and Fig. S2B). This suggests that DHNA, or possibly a factor derived from DHNA, is released by L. monocytogenes during growth in vitro and can, in turn, be used by neighboring bacteria. Although it is possible that both MK and DHNA are released by L. monocytogenes, rescue of the ΔmenD mutant with the supernatant of the ΔmenA mutants suggests that release of DHNA itself is sufficient to complement growth of the ΔmenD mutant.

FIG 3.

L. monocytogenes releases DHNA into the culture supernatant and during invasion of the macrophage cytosol. (A) The wild type or the ΔmenD mutant was grown in minimal medium at 37°C and monitored for growth (OD₆₀₀) over 24 h. Cultures were supplemented with 1% concentration of cell-free supernatants (SN) from wild-type or ΔmenA cultures grown in minimal medium. Data represent one of three biological replicates. (B) The wild-type, ΔyvcK, ΔmenD, or ΔmenA strain carrying the bacteriolysis reporter pBHE573 was used to infect macrophages at an MOI of 10. Infected macrophages where then infected with the wild-type, ΔmenA, or ΔmenD strain at an MOI of 1. At 6 h postinfection infected cells were examined for intracellular bacteriolysis of L. monocytogenes. Data represent the means ± standard errors of the means of four biological replicates normalized to the level of wild-type bacteriolysis. **, P ≤ 0.01; ***, P ≤ 0.001.

We next examined whether DHNA metabolite exchange can also occur during infection of macrophages. To test this hypothesis, we first infected macrophages with strains of L. monocytogenes carrying a bacteriolysis reporter plasmid (10) and then coinfected macrophages with a second unmarked strain of L. monocytogenes. In this setup, we exclusively monitored intracellular survival of the strain carrying the bacteriolysis reporter plasmid. As a positive control, we used a ΔyvcK mutant which was previously shown to lyse in the cytosol of macrophages (10). As expected, coinfection with the wild type or the ΔmenA mutant did not significantly rescue cytosolic survival of the ΔyvcK mutant (Fig. 3B). However, coinfection of ΔmenD mutants with either the wild type or a ΔmenA mutant restored cytosolic survival to wild-type levels. As seen previously, the ΔmenA mutant, which can make DHNA, was not defective for intracellular survival under any condition (Fig. 3B) (12). Taken together, these data suggest that L. monocytogenes participates in metabolite exchange, releasing DHNA or an MK-independent DHNA derivative which can be utilized by DHNA-deficient L. monocytogenes to support replication and cytosolic survival.

Isolation of virulence suppressors.

Menaquinone synthesis-independent roles for DHNA in bacterial physiology and pathogenesis have not been carefully investigated. To uncover novel functions of DHNA in L. monocytogenes, we performed a suppressor screen to identify mutations that suppress defects of the ΔmenD mutant. The ΔmenD mutants were incapable of forming plaques in fibroblast monolayers (Fig. 4A) (12, 15, 16), a classic ex vivo model of L. monocytogenes virulence. Importantly, much of this virulence phenotype was independent of respiration since ΔmenA mutants, terminal cytochrome oxidase mutants, and F_oF₁ ATP synthase mutants form plaques, albeit not always at wild-type levels (12).

FIG 4.

Isolation of ΔmenD suppressor strains that rescue plaquing. (A) The wild type or the ΔmenD mutant, as indicated, was examined for plaque formation in L2 fibroblast cells at 6 days postinfection. Representative images are shown of plaque formation from L2 fibroblasts infected with the ΔmenD mutagenized library (mds 1° screen) or an mds strain during the secondary (2°) screen. (B) All mds strains were examined for aerobic growth in BHI medium at 37°C. Cultures were monitored (OD₆₀₀) for 12 h. Data represent one of two biological replicates. (C) All mds strains were examined for plaque formation in L2 fibroblasts at 6 days postinfection and for intracellular survival in macrophages at 6 h postinfection. Data are plotted for each mutant as plaque size relative to that of the wild type (100%) and as bacteriolysis relative to that of the wild type. Vertical dotted and dashed lines denote wild-type and ΔmenD levels of intracellular bacteriolysis, respectively. Gray circles highlight mds18, mds71, mds175, mds182, and mds188, which displayed the lowest levels of intracellular bacteriolysis and whose genomes were sequenced.

We generated a chemical mutant library by treating the ΔmenD mutant with ethyl methanesulfonate (EMS) to induce random transition mutations (25) and selected for suppressor mutants able to form plaques (Fig. 4A). Bacteria were plaque purified from 229 individual plaques and then screened again in a secondary plaquing assay to verify their phenotypes. A total of 143 isolates formed plaques in the secondary plaquing assay and were carried forward for further analysis. Analysis of these 143 ΔmenD suppressor (mds) mutants in aerobically grown cultures (Fig. 4B) suggests that the suppressor mutations were restoring neither the ability of the ΔmenD mutant to synthesize MK/DHNA nor the electron transport chain. These results support the hypothesis that the ΔmenD mutant’s plaquing/virulence defect is in part due to respiration-independent functions of DHNA.

Since DHNA-deficient strains were also defective for intracellular survival in macrophages (12), we next examined each mds isolate for intracellular bacteriolysis. Most mds strains were as defective for intracellular survival as the parental ΔmenD mutant (Fig. 4C), suggesting that plaquing and intracellular survival phenotypes are not strictly linked. A subset of mds isolates, mds18, mds71, mds175, mds182, and mds188, lysed at approximately 50% or less than the ΔmenD mutant; therefore, these mutants were selected for further examination.

Identification of DHNA suppressor mutations.

To identify the specific mutations that rescued plaquing and cytosolic survival phenotypes of the ΔmenD mutant, we sequenced the genomes of mds18, mds72, mds175, mds182, mds188, and ΔmenD mutants. Each mds isolate contained 3 to 6 single nucleotide polymorphisms (SNPs) in its genome in comparison to the sequence of the ΔmenD mutant (Table 1). As expected, most mutations were either G-to-A or C-to-T transition mutations. Almost all of the SNPs were predicted to cause missense mutations. SNPs in lftR resulted in a nonsense mutation and might disturb expression of the adjacent gene in the lftRS operon (26). Other SNPs in lafC, dxr, and lmo0540 are predicted to be silent mutations. Interestingly 4 out of 5 of the mds isolates had distinct missense mutations in a putative transcriptional regulator, ytoI. Additionally, missense mutations in prs (lmo0199), encoding a putative phosphoribosyl pyrophosphate synthetase, were found in two independent mds isolates.

TABLE 1.

Locations of single nucleotide polymorphisms in ΔmenD suppressor isolates in comparison to the sequence of the parental ΔmenD mutant^b

Isolate	Gene	Locus	SNP	Variant	Function/description
mds18	rplL	lmo0251	G19A	E7K	50S ribosomal protein, L7/L12
	ytoI	lmo1576	G1277A	G426E	Unknown
	lafC	lmo2553	C72T	Silent	Putative integral membrane protein
mds71	lftR	lmo0719	C289T	Nonsense	PadR-like transcriptional regulator
	ytoI	lmo1576	G1276A	G426R	Unknown
	lmo1593	lmo1593	G535A	A179T	Similar to nifs; iron-sulfur cofactor synthesis
	gpsA	lmo1936	G354A	M118I	Glycerol-3-phosphate dehydrogenase
mds175	prs	lmo0199	G101A	G34E	Phosphoribosyl pyrophosphate synthetase
	lmo0947	lmo0947	C959T	S320L	Putative transport protein
	dxr	lmo1317	C375T	Silent	Similar to deoxyxylulose 5-phosphate reductoisomerase
	ytoI	lmo1576	G71A	R24Q	Unknown
mds182	lmo0090	lmo0090	C628T	H210Y	Similar to ATP synthase alpha chain
	rpoC	lmo0259	C2201T	A734V	RNA polymerase beta subunit
	lmo0540	lmo0540	C1110T	Silent	Penicillin binding protein
	ytoI	lmo1576	A34T	I12F	Unknown
			C2251176Ta		Intergenic region between lmo2202 and lmo2203
	lmo2735	lmo2735	C1273T	H425Y	Sucrose phosphorylase
mds188	prs	lmo0199	G170A	C57Y	Phosphoribosyl pyrophosphate synthetase
	lmo1851	lmo1851	G529A	D177N	Similar to carboxy-terminal processing proteinase
			C2547268Ta		Intergenic region between comfa and lmo2514

^aPosition on the L. monocytogenes 10403s genome.

^bShading represents SNPs found in ytoI among the different suppressor mutants.

We hypothesized that most SNPs may result in loss-of-function mutations; therefore, to validate suppressor mutations in each mds strain, we complemented in trans wild-type versions of putative suppressor genes back into each mds isolate and tested these strains in plaquing and intracellular bacteriolysis assays (Fig. 5A and B). Among mutations found in the mds strains, expression of wild-type rpoC abolished plaquing and increased bacteriolysis of mds182, suggesting that mutations in rpoC suppress DHNA deficiency. In mds182, mutations in both rpoC and ytoI appear to act synergistically to restore virulence and intracellular survival of the ΔmenD mutant. Indeed, mds182 forms the largest plaques of all the mds isolates and appears to lyse at wild-type levels (Fig. 5A and B). In contrast, overexpression of lafC (27) or dxr (28) significantly decreased plaquing of the ΔmenD mutant but did not influence cytosolic survival. Demonstrating a role for lafC and dxr was surprising as the mutations identified in our genetic screen were predicted to be silent mutations. This suggests that the silent mutations in lafC and dxr may target regulatory elements in these genes. In contrast, constitutive expression of the wild-type lftRS operon, encoding an unknown transcriptional regulator and gene of unknown function, respectively, worsened intracellular survival but did not affect plaque formation.

FIG 5.

Identification of suppressor mutations in mds18, mds71, mds175, mds182, and mds188. mds strains were complemented either with an empty vector or in trans with genes identified in Table 1 and then examined for plaque formation in L2 fibroblasts at 6 days postinfection (A) or intracellular bacteriolysis in macrophages at 6 h postinfection (B). Data represent the means ± standard errors of the means of three biological replicates normalized to wild-type plaque size (100%) or level of intracellular bacteriolysis. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.

Complementation with ytoI in mds18, mds71, and mds175 significantly decreased plaquing and increased cytosolic lysis of these suppressor mutants. These data suggest that inactivation of ytoI in the ΔmenD mutant rescues both plaquing and intracellular survival. Restoration of wild-type ytoI expression significantly decreased plaquing but did not increase cytosolic survival of mds182, suggesting that additional mutations in this strain mask the phenotype of the ytoI mutation. Given the importance of ytoI in suppressing DHNA deficiency, we sequenced this gene in an additional 135 mds suppressors and found 6 strains with nonsense mutations and an additional 30 mds strains with missense mutations in ytoI, further highlighting its importance in bacterial physiology in the absence of DHNA (Fig. S3).

Finally, other mutations in the mds strains do not appear to be important for suppression of DHNA deficiency although we cannot exclude the possibility that these mutations are dominant negative or even gain-of-function mutations. Taken together, these data highlight that DHNA deficiency is multifactorial and that multiple independent pathways can overcome loss of DHNA. Importantly, mutations in ytoI were isolated disproportionately among our suppressors, suggesting that this putative transcriptional regulator is closely linked with DHNA function.

YtoI deletion rescues both virulence and intracellular survival.

Multiple independent mutations in ytoI rescued both plaquing and intracellular survival defects of DHNA-deficient L. monocytogenes; therefore, we hypothesized that these mutations represented loss-of-function mutations. To test this hypothesis, we generated a clean deletion of ytoI in the wild type and in a ΔmenD background. The single ΔytoI mutant was indistinguishable from the wild type during replication in BHI or minimal medium, suggesting that mutation in ytoI alone neither diminishes nor enhances replication of L. monocytogenes (Fig. 6A and B). Deletion of ytoI in a ΔmenD background did not improve growth in BHI medium; however, the double ΔmenD ΔytoI mutant consistently, though minimally, displayed improved growth in minimal medium in comparison to growth of the ΔmenD mutant (Fig. 6A and B).

FIG 6.

The unknown transcriptional regulator YtoI suppresses plaquing and intracellular survival of ΔmenD mutant but not in vitro growth. The wild type, ΔmenD, ΔytoI, and ΔmenD ΔytoI strains were grown in BHI medium (A) or minimal medium (B) and monitored (OD₆₀₀) for growth for 12 or 24 h, respectively. DHNA (5 μM) was supplemented to cultures as indicated. Data represent one of three biological replicates. (C) The wild-type, ΔmenD, ΔytoI, and ΔmenD ΔytoI strains were examined for plaque formation in L2 fibroblasts at 6 days postinfection. Data represent the means ± standard errors of the means of three biological replicates normalized to wild-type plaque size (100%). (D) The wild-type, ΔmenD, ΔytoI, and ΔmenD ΔytoI strains were tested for intracellular bacteriolysis in macrophages at 6 h postinfection. Data represent the means ± standard errors of the means of three biological replicates normalized to the level of wild-type intracellular bacteriolysis. (E and F) Aerobic cultures of the wild type, ΔytoI mutant, ΔytoI mutant complemented with ytoI, ΔmenD mutant, and ΔmenD ΔytoI mutant were grown in BHI medium at 37°C until mid-logarithmic phase. Expression levels of genes in the ytoI-lmo1575 operon were examined by qRT-PCR. Data represent the relative expression (RQ) ± standard errors of the means of three biological replicates normalized to the level of the wild type (E) or the ΔmenD mutant (F) ND, not detected. (G) Overexpression of Lmo1575 does not suppress ΔmenD intracellular bacteriolysis. The wild type, ΔmenD mutant, and ΔmenD mutant expressing lmo1575 were examined for intracellular bacteriolysis in macrophages. Data represent the means ± standard errors of the means of three biological replicates normalized to the level of wild-type intracellular bacteriolysis. ***, P ≤ 0.001; NS, not significant.

We next examined these strains for plaquing and intracellular survival (Fig. 6C and D). The ΔytoI mutant had no noticeable change in plaquing or intracellular survival in comparison to that of the wild type. In contrast, the ΔmenD ΔytoI mutant formed plaques and displayed lower levels of bacteriolysis than the ΔmenD mutant, providing further evidence that loss of ytoI was responsible for rescuing plaquing and intracellular survival of the ΔmenD suppressor mutant. Neither plaquing nor intracellular bacteriolysis returned to wild-type levels, suggesting that mutation of ytoI alone does not rescue all physiological and virulence defects of DHNA-deficient L. monocytogenes. Interestingly, loss of ytoI also rescues the plaquing defect of a ΔcydAB mutant (Fig. S4). cydAB encodes the cytochrome bd oxidases of the ETC (29). The ΔcydAB mutant is the only ETC mutant with an intracellular bacteriolysis phenotype in macrophages (12), perhaps suggesting that CydAB contributes to DHNA functions independent of respiration.

The ytoI gene encodes a putative transcriptional regulator conserved primarily in Firmicutes (30). YtoI structural predictions show a helix-turn-helix (HTH) domain, as well as two tandem DRTGG-CBS domains and a Hotdog fold (Fig. S3). ytoI is cotranscribed with lmo1575, a gene encoding a putative phosphodiesterase (26). In the absence of ytoI, expression of lmo1575 was significantly increased, a phenotype that was reversed when ytoI was provided in trans (Fig. 6E). The same expression patterns were observed in a ΔmenD mutant background when ytoI was deleted (Fig. 6F). Together, these data suggest that YtoI may act as a transcriptional repressor of its own operon. We next hypothesized that derepression of lmo1575 may be the mechanism by which loss of ytoI rescues intracellular survival of the DHNA-deficient L. monocytogenes. To test this hypothesis, we examined intracellular survival of the ΔmenD mutant expressing lmo1575 from a constitutive promoter. Overexpression of lmo1575 in the ΔmenD mutant did not rescue intracellular bacteriolysis (Fig. 6G), suggesting that regulation of this lmo1575 alone is not responsible for suppression of the ΔmenD strain but, instead, that other genes in the YtoI regulon may be responsible for suppressing intracellular survival of the ΔmenD strain.

YtoI controls fermentative by-products.

To test the hypothesis that additional YtoI-regulated genes were involved in DHNA suppression, we assessed differential expression of genes regulated by the YtoI homolog, SpxR, in S. pneumoniae D39 (24). Surprisingly, approximately half of the genes known to be SpxR regulated based on S. pneumoniae SpxR microarray analysis do not have homologs in L. monocytogenes (Table S1). Moreover, when we examined four SpxR-regulated genes with homologs in L. monocytogenes, we found that none of these putative SpxR-regulated homologs were differentially expressed in the L. monocytogenes ΔytoI mutant (Fig. S5A), suggesting that the functional regulon of YtoI and SpxR may not be highly conserved between L. monocytogenes and S. pneumoniae. For these reasons we have decided to continue with the ytoI (YtoI) nomenclature rather than change the name to spxR (SpxR).

As the regulon of YtoI was not highly conserved relative to that of SpxR, we set out to characterize the YtoI regulatory system using RNA-seq transcriptome analysis of cytosolic wild-type L. monocytogenes and the isogenic ΔytoI mutant during macrophage infection. We also attempted to examine differential expression levels between ΔmenD and ΔmenD ΔytoI mutants; however, we were unable to obtain sufficient read counts for analysis. Five genes (lmo0944, lmo1575, lmo1577, lmo2005, and alsS) were expressed at greater than 2-fold higher levels in the ΔytoI mutant than in the wild type, whereas two genes (pflA and eutD) demonstrated greater than 2-fold lower expression in the ΔytoI mutant (Table 2 and Fig. S5B). Intriguingly, many of these genes are involved in controlling pyruvate flux (pflA, alsS, and eutD) or the NAD/NADH ratio (lmo2005). We confirmed changes in relative expression by performing quantitative PCR (qPCR) on six genes in different fermentative pathways (Fig. S5C). In L. monocytogenes, the major aerobic carbohydrate catabolism end products are acetate, acetoin, lactate, and carbon dioxide (Fig. 7A). In contrast, the major end product of carbon metabolism is lactate anaerobically (31).

TABLE 2.

Genes differentially regulated in the ΔytoI mutant compared to expression levels in the wild-type strain^a

^aGenes upregulated and downregulated in the ΔytoI mutant are boxed in blue and red, respectively. RPKM, reads per kilobase per million.

FIG 7.

(A) Metabolism of pyruvate in L. monocytogenes. Lmo1917 (PflA) converts pyruvate to acetyl-CoA in the anaerobic glycerol metabolic pathway (denoted by a dotted line). Genes upregulated or downregulated in the ΔytoI mutant are shown in blue or red, respectively. The production of acetoin (B), acetate (C), and lactate (D) in the wild type, ΔmenD mutant, ΔytoI mutant, and ΔmenD ΔytoI mutant in BHI broth aerobically and anaerobically was determined. Data represent the means ± standard errors of the means of three biological replicates normalized to the OD₆₀₀ value of each strain. Open bars, aerobic cultures; filled bars, anaerobic cultures; ND, not detected. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.

While regulation of metabolism is certainly more complicated than transcriptional regulation of select genes, we measured production of acetoin, acetate, and lactate in the wild-type, ΔmenD, ΔytoI, and ΔmenD ΔytoI strains, both aerobically and anaerobically, in an attempt to validate our transcriptional analysis. Consistent with the increased expression of alsS in ΔytoI mutants, acetoin production was significantly higher in the ΔytoI mutant grown aerobically than in wild-type bacteria (Fig. 7B). In contrast, acetoin production was reduced in both the ΔmenD mutant and ΔmenD ΔytoI mutant grown aerobically, suggesting that ytoI deficiency could not overcome defects in acetoin metabolism. Moreover, consistent with previous studies, nearly no acetoin was detected under anaerobic conditions (Fig. 7B) (31). Similarly, consistent with decreased expression of pflA and eut, ΔytoI mutants demonstrate decreased production of acetate under anaerobic conditions when pflA would be predicted to be functional. Independently of ytoI, DHNA deficiency inhibits acetate production as both the ΔmenD and ΔmenD ΔytoI mutants produced significantly less acetate when grown aerobically (Fig. 7C). Finally, while no anaerobically grown strains differed in production levels of lactate, ΔytoI mutants produced less lactate than their wild-type controls under aerobic conditions, potentially due to increased flux to acetoin due to overexpression of als (Fig. 7D). Taken together, these data suggest that metabolic gene expression changes controlled by ytoI can influence pyruvate metabolism.

To test the hypothesis that specific YtoI-regulated genes contribute to cytosolic survival of DHNA-deficient strains, we constructed strains overexpressing YtoI-repressed genes and knockouts of YtoI-activated genes in a ΔmenD mutant background. Overexpression or deletion of individual YtoI-regulated genes did not rescue DHNA-deficient strains from cytosolic killing (Fig. S6A and B). Future studies reconstructing overexpression and/or knockout of complete pyruvate metabolism pathways regulated by YtoI will likely be necessary in the future to identify how alterations of fermentative flux overcome DHNA deficiency during infection. Nevertheless, together these data demonstrate that YtoI is a novel transcriptional regulator of pyruvate flux that can overcome DHNA deficiency during macrophage infection, suggesting a key role for DHNA in regulating pyruvate metabolism and redox balance during infection.

DISCUSSION

Regulation of bacterial metabolism is increasingly being appreciated as a key node determining the outcome of host-pathogen interactions (32, 33). More specifically, recent data suggest that in addition to its role as a biosynthetic intermediate in MK biosynthesis, DHNA is an important molecule for bacterial physiology and host-microbe interactions (12, 21, 23). Despite a growing appreciation for its importance, little is known about how DHNA influences host-pathogen interactions on a molecular level. In this report we show that L. monocytogenes DHNA biosynthesis is necessary for efficient growth in vitro, for intracellular survival, and for virulence independent of DHNA’s role in MK biosynthesis or the electron transport chain. We provide evidence that, similar to its role in other systems, DHNA is secreted by L. monocytogenes during infection and can be used either as a shared resource and/or as a mechanism to influence host responses. Finally, we identified a novel transcriptional regulator, YtoI, that regulates pyruvate flux and whose inactivation can partially overcome the essentiality of DHNA for cytosolic survival of L. monocytogenes (Fig. 8).

FIG 8.

Current model of DHNA and YtoI in L. monocytogenes metabolism and pathogenesis. DHNA, but not menaquinone (MK), is involved in L. monocytogenes intracellular survival and plaque formation. DHNA could be exported and imported through the L. monocytogenes membrane and cell wall through an unknown mechanism/transporter. Mutation of the novel regulator YtoI partially rescues the virulence and bacteriolysis of a DHNA-deficient strain. However, how DHNA interacts with YtoI is still unclear. The majority of YtoI-regulated genes are involved in controlling fermentative flux, suggesting a critical role for regulating these metabolic programs to promote L. monocytogenes cytosolic survival. ETC, electron transport chain.

Genome analysis suggests that >30% of prokaryotes make DHNA using the classical MK biosynthesis pathway, suggesting that DHNA is similarly likely ubiquitous in the environment (33). DHNA produced and secreted by Propionibacterium and Lactobacillus spp. enhances growth of the human gut commensal Bifidobacterium (20, 21). Hence, DHNA has been proposed as a prebiotic to stimulate populations of beneficial gut microorganisms and protect against intestinal inflammation (34). In addition, group B Streptococcus (GBS), an obligate fermentative bacterium, is missing genes for DHNA synthesis but possesses genes for late MK biosynthesis and cytochrome bd oxidase (22). Supplementation with MK or DHNA enhances both respiration and virulence of GBS (22, 35). It is hypothesized that capture of DHNA/MK from commensal bacteria may induce respiration and cause dissemination of GBS in vivo (36). Finally, exogenously added DHNA directly modulates innate immune function, inhibiting the production of proinflammatory cytokines by macrophages. DHNA and other naphthoate molecules are known agonists of aryl hydrocarbon receptor (AhR) (37), a transcriptional regulator of cell metabolism and immunity (38). Whether pathogens intentionally secrete DHNA to manipulate the immune system is currently unknown. Taken together, these studies establish DHNA as an extracellularly available shared resource and leave open the possibility that some pathogens or commensals might utilize DHNA to actively manipulate immune responses. Our data suggest that L. monocytogenes can similarly both secrete and acquire DHNA from the environment. Exogenous DHNA or supernatant from DHNA-sufficient bacteria rescues growth of DHNA-deficient bacteria in vitro (Fig. 3A). Perhaps more striking, coinfection of macrophages with DHNA-deficient and -sufficient strains results in the rescue of cytosolic survival of DHNA-deficient strains (Fig. 3B). Whether this is due to metabolite exchange-mediated transcomplementation or a direct effect of DHNA inhibiting host cell autonomous defenses in the cytosol is currently unknown. Identifying the mechanisms by which DHNA is secreted and in turn acquired from the environment will allow us to understand the mechanism(s) by which DHNA promotes cytosolic survival of L. monocytogenes.

Little is known about DHNA outside its role in MK synthesis and subsequently ETC function. The naphthoquinone ring of DHNA may act as a cofactor in reduction-oxidation (redox) reactions. Indeed, Shewanella oneidensis MR-1 uses a derivative of DHNA to reduce carbon tetrachloride, a common environmental contaminant (39). Our data suggest that efficient replication in defined minimal medium in vitro, survival in the cytosol, and virulence depend partially on MK/ETC-independent functions of DHNA (Fig. 1 and 2) (12). Intriguingly, our data also suggest that extracellularly supplied DHNA is sufficient to enhance growth of DHNA-deficient strains in vitro; however, in contrast to observations of previous reports in other organisms, this is not mediated by restoration of the ETC (Fig. 2C). One explanation may be that L. monocytogenes is unable to transport DHNA into the cell or that extracellular DHNA is quickly converted to another molecule which cannot restore ETC function. Future identification of DHNA binding proteins is necessary to determine if DHNA acts as a cofactor or allosteric regulator of enzymes and/or transcription factors to facilitate its functions in metabolism and virulence. Furthermore, identification of transporters and/or extracellular receptors of DHNA or its derivatives will define the molecular mechanisms of metabolite exchange. Ultimately, understanding the role of DHNA in L. monocytogenes virulence will allow us to determine how common MK-independent functions of DHNA are in other bacteria.

Our DHNA suppressor analysis identified several different pathways which rescue virulence of DHNA-deficient strains, potentially by enhancing resistance toward cytosolic stresses and/or facilitating metabolism of cytosolically available substrates (Fig. 4 and 5). Specific mutations in rpoC, encoding the RNA polymerase β-subunit, are well known to increase replication and resistance against antimicrobial effectors in various bacteria (40, 41). We propose that suppressor mutations in rpoC may trigger a stress response in L. monocytogenes to increase resistance against cytosolic stresses. Mutations in the putative integral membrane protein, lafC, alters the glycolipid composition and structures of lipoteichoic acids in L. monocytogenes (27). It is possible the lafC mutation enhances L. monocytogenes resistance toward extracellular insults by altering stability of the cell wall. L. monocytogenes uses both the classical mevalonate (MVA) and alternative, nonmevalonate (methylerythritol phosphate [MEP]) pathways to synthesize isoprenoid units, which form prenol lipids necessary for cell wall and MK biosynthesis (28). Accumulation of phosphorylated prenol lipids sensitizes Bacillus subtilis against bacitracin (42). We propose that the dxr suppressor mutation, which encodes an enzyme in the MEP pathway, may relieve toxicity of prenol lipid accumulation in L. monocytogenes and/or sensitivity toward cytosolic stresses. Finally, in L. monocytogenes LftR has been shown to repress the putative multidrug resistance transporter LieAB (43), suggesting that disruption of lftR may increase resistance toward cytosolic stresses by increasing expression of LieAB or other genes in the LftR regulon. Understanding the mechanism of suppression by the mutations found in our screen will help clarify what stresses are present in the cytosol and how DHNA metabolism in L. monocytogenes protects against these stresses.

Intriguingly, suppressor mutations in ytoI were identified in multiple suppressor isolates and rescued growth in minimal medium, intracellular survival, and plaquing of the ΔmenD mutant. YtoI is a transcriptional regulator (YtoI family) conserved in several Firmicutes (30) and contains a helix-turn-helix (HTH) domain, tandem DRTGG-CBS domains, and a Hotdog fold. It is predicted that CBS domains bind adenosyl compounds, such as ATP or AMP (44), while Hotdog folds are commonly found in thioesterases in the fatty acid biosynthesis pathway and may bind CoA moieties (30). It is hypothesized that the YtoI family of regulators may control metabolism in Firmicutes by sensing the energy state of the cell. Indeed, SpxR, a homolog of YtoI in S. pneumoniae D39, was found in a genetic screen to modulate hydrogen peroxide production (24). SpxR influences the activity of pyruvate oxidase (SpxB), a hydrogen peroxide-producing enzyme in the acetate metabolism pathway, as well as several other metabolic genes and is required for S. pneumoniae virulence (24). Surprisingly, despite the conserved genomic architectures of SpxR and YtoI, they appear to have distinct regulons and influences on pathogen virulence in their respective species whereby SpxR is essential for virulence while YtoI appears dispensable (see Table S1 and Fig. S5 in the supplemental material). Identifying conditions that activate YtoI expression/activity and what, if any, allosteric regulators control its function will be critical to further understanding the role of YtoI in regulating metabolism, stress responses, and ultimately virulence.

Although YtoI alone is dispensable for growth and virulence, loss of YtoI may increase resistance when DHNA-deficient strains are in the cytosol. Loss of ytoI rescues the plaquing defect of DHNA-deficient strains and also that of the ETC ΔcydAB mutant (Fig. S4). One explanation is that the cytochrome bd oxidase may participate in a metabolic feedback loop to regulate DHNA levels in L. monocytogenes; thus, a ΔcydAB mutant is DHNA deficient. Alternatively, the cytochrome bd oxidase may be required for DHNA function, possibly as an electron sink for DHNA. Determining the mechanism of suppression by mapping the YtoI regulon and also deciphering the metabolic changes in ΔmenD and suppressor strains may help to elucidate the role of DHNA in intracellular survival of L. monocytogenes. Our results showed that YtoI is involved in L. monocytogenes pyruvate metabolism (Fig. 7); however, how these fermentative by-products are involved in L. monocytogenes virulence and intracellular survival remains to be clarified.

Replication and survival in the cytosol are essential to the virulence of L. monocytogenes and other cytosol-adapted pathogens. Metabolic adaptations to the cytosolic environment are essential for not only replication but also tolerance of cytosolic stresses (45, 46). Our work indicates that DHNA, independent of the ETC, is central to L. monocytogenes intracellular survival. Whether the function of DHNA during infection is intrinsic to the physiology of the bacterial cell or related to modulation of host responses remains to be determined. Importantly, however, manipulation of pyruvate metabolism, again, either through bacterial physiology or evasion of host recognition, can overcome many of the defects of DHNA deficiency during infection. As DHNA is a highly conserved and abundant shared resource, understanding its functions in host-pathogen interactions, in the context of both bacterial physiology and host response, is critically important. More specifically, understanding how DHNA protects L. monocytogenes during intracellular replication will provide insight into how the host cytosol protects itself against invaders and may help in the design of antimicrobial strategies against cytosolic pathogens.

MATERIALS AND METHODS

Bacterial strains, plasmid construction, and growth conditions in vitro.

All strains used in this study are listed in Table S2 in the supplemental material. L. monocytogenes strain 10403s is referred to as the wild-type strain, and all other strains are isogenic derivatives of this parental strain. L. monocytogenes strains were grown at 37°C in brain heart infusion (BHI) medium or minimal medium supplemented with glucose as the sole carbon source. Minimal medium is identical to minimal medium described by Phan-Thanh and Gormon (47), with the exception that phosphate concentrations are reduced 10-fold. To compensate for decreased phosphate buffering, 20.93 g/liter MOPS [3-(N-morpholino) propanesulfonic acid] was used to buffer the medium at a pH of 7.5. Escherichia coli strains were grown in Luria-Bertani (LB) broth at 37°C. Antibiotics were used at concentrations of 100 μg/ml carbenicillin, 10 μg/ml chloramphenicol, 2 μg/ml erythromycin, or 30 μg/ml kanamycin when appropriate. Medium, where indicated in Fig. 2, was supplemented with 5 μM 1,4-dihydroxy-2-naphthoate (DHNA) (281255; Sigma) or 5 μM menaquinone (MK) (V9378; Sigma).

In supernatant complementation experiments, overnight aerobic cultures of the wild-type strain or the ΔmenA mutant grown in minimal medium were centrifuged to remove bacteria. Then supernatants were filtered through a 0.2-μm-pore-size syringe filter. Filtered supernatants were added to growing cultures of L. monocytogenes at a final concentration of 1% (vol/vol).

Vectors were shuttled into L. monocytogenes by E. coli strain S17 or SM10 through conjugation (48). In-frame deletions of genes in L. monocytogenes were performed by allelic exchange (49) using the suicide plasmid pksv7-oriT as previously described (50). The integrative vector pIMK2 (51) was used for constitutive expression of L. monocytogenes genes. The phage integration vector pPL2 (52), encoding a theophylline-inducible riboswitch (53), was used for expression of lmo2006 and prkA as previously described (54).

Acute virulence assay.

All techniques were reviewed and approved by the University of Wisconsin—Madison Institutional Animal Care and Use Committee (IACUC) under the protocol M02501. Female C57BL/6 mice (6 to 8 weeks of age; purchased from Charles River) were injected intravenously with 1 × 10⁵ CFU (50% lethal dose [LD₅₀] for wild-type 10403s L. monocytogenes). At 48 h postinfection, spleens and livers were harvested and homogenized in 0.1% Nonidet P-40 in phosphate-buffered saline (PBS). Homogenates were then plated on LB plates containing streptomycin to enumerate CFU and quantify bacterial burdens.

Measuring bacterial membrane potential.

L. monocytogenes strains grown to mid-late logarithmic phase in BHI medium at 37°C were stained for membrane potentials using 3 mM membrane potential indicator dye DIO2(3) (3,3′-diethyloxacarbocyanine iodide) (320684; Sigma) for 30 min. The proton ionophore carbonyl cyanide 3-chlorophenylhydrazone (CCCP; 500 μM) (C2759; Sigma) was used to depolarize the bacterial membranes. Samples were analyzed on a BD LSR-II flow cytometer as previously described in Chen et al. (12).

Generating a chemical mutant library.

Chemical mutagenesis of the ΔmenD strain was carried out as outlined in Shetron-Rama et al. (55) with minor modifications. Overnight cultures of the ΔmenD mutant were inoculated (1:20) into aerated BHI cultures supplemented with 50 μg/ml of MK at 37°C. When cultures reached an optical density at 600 nm (OD₆₀₀) between 0.9 and 1.0, bacteria were washed in PBS twice (pelleted at 3,220 × g for 30 min) and then resuspended in 20 ml of tryptic soy broth (TSB). Ethyl methanesulfonate (EMS; 300 μl) (M0880; Sigma) was added to bacterial suspensions and incubated for 30 min at 37°C (no aeration). EMS-treated cultures were again washed twice in PBS (pelleted at 3,220 × g for 30 min) and then resuspended in 20 ml of BHI medium plus 40% glycerol and frozen at –80°C in 1-ml aliquots.

The number of mutations per genome was estimated by measuring the rifampin resistance frequency of the library cultures. Cultures were thawed and then plated on BHI medium and BHI medium plus 50 μg/ml rifampin and enumerated for CFU. Sequences of the ΔmenD EMS mutagenesis library used in this work had an estimated 0.69 mutations per genome as determined using the equation described in Kari et al. (56).

Screening for the ΔmenD EMS library using a standard plaque assay.

Samples from the ΔmenD EMS mutagenesis libraries were thawed and combined with an equivalent volume of BHI medium. Suspensions were incubated at 37°C for 1 h and then used to infect L2 fibroblast cells at a multiplicity of infection (MOI) of 0.1. Standard plaque assays were performed as described in Chen et al. (12). Plaque areas were determined at 6 days postinfection using Fiji ImageJ analysis software (57). To verify ΔmenD suppressors, plaques were picked using sterile toothpicks and plated on BHI plates. Individual colonies were again examined for plaque formation in L2 fibroblast cells and again plaque purified.

Intracellular bacteriolysis assay.

Standard intracellular bacteriolysis assays were performed in immortalized Ifnar^−/− macrophages as previously described (12). For coinfection experiments, L. monocytogenes carrying the bacteriolysis reporter pBHE573 (10) was used to infect macrophages at an MOI of 10. Subsequently, these infected macrophages were infected with L. monocytogenes strains not carrying pBHE573 at an MOI of 1.

Genome sequencing and SNP determination.

Genomic DNA was purified using a MasterPure Gram-positive DNA purification kit (MGP04100; Epicentre). DNA sequencing libraries were generated using a Nextera XT DNA library preparation kit (FC-131-1024; Illumina) and Nextera XT index kit (FC-131-1001; Illumina) according to the manufacturer’s instructions. Sequencing of DNA libraries was performed at the University of Wisconsin Biotechnology Center DNA Sequencing Facility using an Illumina MiSeq platform (2- by 300-bp reads). Genome assembly and single nucleotide polymorphism (SNP) determinations in comparison to the sequence of the parental strain (ΔmenD mutant) were performed using Lasergene Seqman NGen software (DNASTAR).

RNA isolation and reverse transcription-PCR (RT-PCR).

L. monocytogenes strains were grown in aerated BHI cultures at 37°C until mid-late logarithmic phase. Bacterial cultures were mixed at 1:1 with RNAprotect bacteria reagent (76506; Qiagen) and incubated for 5 min at room temperature. Cultures were then pelleted and frozen at –80°C.

To isolate nucleic acids, frozen pellets were thawed and resuspended in 500 μl of AE buffer (10 mM Tris-Cl, 0.5 mM EDTA, pH 9) with 50 μl of 10% SDS and 500 μl of a 1:1 acidified phenol-chloroform mixture. To ensure complete lysis, bacteria were subjected to bead beating for 10 min at 1,500 rpm. Mixtures were incubated at 65°C for 10 min, and then the aqueous layer was removed and subjected to standard ethanol precipitation.

Total nucleic acids were quantified using a Nanodrop instrument. Two micrograms of nucleic acids was treated with Turbo DNase (AM2238; Ambion) for 1 h at 37°C to degrade contaminating genomic DNA. The DNase reaction was inactivated according to the manufacturer’s instructions. RNA was quantified, and then 250 ng of RNA was converted to cDNA using an iScript cDNA synthesis kit (Bio-Rad). cDNA was quantified by quantitative PCR (qPCR) on a StepOne real-time PCR System (Applied Biosystems).

RNA-seq analysis.

Cultures of L. monocytogenes were grown in LB medium at 30°C under static conditions. Bacteria were washed with PBS and then used to infect approximately 3 × 10⁷ immortalized murine macrophages at an MOI of 10. At 1 h postinfection, culture supernatant was replaced with medium containing 10 μg/ml gentamicin. At 4 h postinfection macrophages were washed three times with PBS and then lysed with cold 1% saponin. Lysed cells were combined with 10 ml of cold RNAprotect bacteria reagent (Qiagen). Samples were spun at 800 × g for 5 min to remove lysed macrophage debris, and then the remaining supernatant was spun at 8,000 × g for 5 min to collect bacteria for RNA extraction as described above.

RNA samples (obtained in triplicate) were submitted to the University of Wisconsin Biotechnology Center DNA Sequencing Facility for quality control, library preparation, and sequencing. Briefly, RNA quality was determined throughout sample preparation using an Agilent bioanalyzer. A RiboZero kit (Epidemiology) was used to remove rRNA. Additionally, oligo(dT) beads were used to remove poly(A) transcripts and enrich for bacterial RNA. RNA libraries were prepared using a TruSeq RNA Prep kit (Illumina) and run on an Illumina HiSeq2500 sequencer (1- by 100-bp reads).

RNA-seq analysis was performed using the DNASTAR Genomics Suite (https://www.dnastar.com/). Sequencing reads were aligned against the L. monocytogenes 10403s genome and then analyzed for differential gene expression (reads per kilobase per million [RPKM]) between samples. Genes identified for further analysis displayed greater than 2-fold changes in expression between wild-type and ΔytoI strains and P values of less than 0.3. Rockhopper 2 (58) (https://cs.wellesley.edu/~btjaden/Rockhopper/) software was used as a secondary transcriptome analysis tool.

Acetoin, acetate, and lactate assay.

Cultures were grown to stationary phase in BHI broth at 37°C, and the OD₆₀₀ of cultures was determined and adjusted to 1. One milliliter of culture was centrifuged and washed twice with BHI broth to remove the acetoin, acetate, and lactate in the supernatant and then inoculated at 1:100 into 5 ml of BHI broth. Acetoin, acetate, and lactate production in the supernatant was measured after 12 h (stationary phase) of L. monocytogenes culture growth under aerobic (shaking, 200 rpm) and anaerobic (in Hungate culture tubes) conditions, and the OD₆₀₀ was measured for normalization. The acetoin produced by L. monocytogenes cultures was measured with a Voges-Proskauer reaction (59). An Enzychrom acetate assay kit (Bioassay Systems) and Lactate-Glo assay kit (Promega) were used for acetate and lactate assays, respectively, according to the manufacturers’ supplied instructions.

Statistical analysis.

Statistical significance analysis (GraphPad Prism, version 6.0 h) was determined by one-way or two-way analysis of variance (ANOVA) with a Bonferroni posttest unless otherwise indicated in the figure legends. Bacteriolysis assay data were log transformed prior to statistical analysis.