Listeria monocytogenes 6-Phosphogluconolactonase Mutants Induce Increased Activation of a Host Cytosolic Surveillance Pathway

Bacterial strains.The L. monocytogenes strains used and generated in this study were all in the 10403S or the 10403S mdrM deletion mutant (DP-L5444) background (6). The Tn917 mutant with insertion between lmo0558 and lmo0559 (DP-L5524), the lmo0559-deleted strain (DP-L5509), the pgl-deleted strain (DP-L5507), and the pgl mdrM double deletion strain (DP-L5532) are described in Results. In-frame deletions of L. monocytogenes genes were generated using splice overlap extension-PCR and allelic exchange, as previously described (3, 30).

Source of mice.C57BL/6 mice were obtained from The Jackson Laboratory. MyD88/Trif^−/− mice were obtained from G. Barton, University of California, Berkeley.

Infections and analysis of gene expression in macrophages.A total of 4 × 10⁶L. monocytogenes organisms were used to infect 0.7 × 10⁶ macrophage cells per well, seeded on 35-mm six-well tissue culture plates. These numbers resulted in infection of one to two bacteria per cell. Thirty minutes after addition of bacteria, macrophage monolayers were washed three times with phosphate-buffered saline (PBS) and fresh medium was added. At 1 hour postinfection (hpi), gentamicin was added to 50 μg/ml to kill extracellular bacteria. Unless indicated otherwise, RNA was collected at 4 hpi for further analysis. Induction of IFN-β by macrophages analyzed by real-time quantitative reverse transcription-PCR (qRT-PCR), as described previously (11). t tests (with equal variances assumed) were used for statistical analysis and were performed using the Microsoft Excel analysis package add-in feature, as previously described (16). P values are noted in the figure legends.

Bacterial intracellular growth curves.Bacterial intracellular growth curves were determined as described previously (26). Briefly, 2 × 10⁶ bone marrow-derived macrophages were infected with 4 × 10⁵ CFU of L. monocytogenes from an overnight culture. Thirty minutes after addition of bacteria, macrophage monolayers were washed with PBS. At 1 hpi, gentamicin was added to 50 μg/ml to kill extracellular bacteria. At different time points postinfection, three coverslips were taken and washed with water to lyse host cells. Bacteria recovered from each coverslip were plated on brain heart infusion (BHI) plates, and the number of CFU was determined.

In vivo infections.C57BL/6 mice (The Jackson Laboratory) were infected with 1 50% lethal dose (LD₅₀) (1 × 10⁵ CFU) of wild-type (wt) L. monocytogenes, the pgl deletion mutant, or the pgl deletion mutant complemented with pgl on pPL2, and organs were harvested at 48 hpi, as previously described (1).

Growth of L. monocytogenes in minimal medium.wt and pgl deletion mutant L. monocytogenes were grown in a defined minimal medium, as outlined by Phan-Thanh and Gormon (12, 24), with the one exception that 2.5 mg/liter of adenine was used. This medium is referred to in this paper as “excess-glucose” medium. Glucose-limiting medium was identical to excess-glucose medium except with a 10-fold-lower concentration of glucose (1 g/liter instead of 10 g/liter). For the addition of “excess glucose” to a glucose-limiting culture, glucose was added to a final concentration of 10 g/liter.

Sensitivity of L. monocytogenes to oxidative stress.BHI agar plates were spread with 100 μl of a mid-log-phase BHI bacterial culture and allowed to dry. Filter paper disks (7-mm diameter) were soaked with defined amounts of hydrogen peroxide, diamide, or carbenicillin (see Table 2). The presoaked filter disks were then applied to the plates described above (9) at four per plate, and the plates were incubated at 37°C. The diameter of the zone of growth inhibition around each filter was measured after 24 h of growth. t tests (with equal variances assumed) were used for statistical analysis and were performed using the Microsoft Excel analysis package add-in feature, as previously described (16).

Analysis of radiolabeled glucose metabolism in wt and pgl deletion mutant strains by TLC. L. monocytogenes cultures were synchronized and grown to an optical density at 600 nm (OD₆₀₀) of 1.6. Aliquots (1 ml) were resuspended in an equal volume of PBS and labeled with 2.5 μCi of either [1-¹⁴C]glucose or [U-¹⁴C]glucose for 1 h at 37°C. Cell-free supernatants were analyzed by thin-layer chromatography (TLC), eluting with water-isopropanol-ammonium hydroxide (2:6:1) as the developing solvent. Radioactivity was quantified using phosphorimaging followed by densitometry. TLC plates were glass-backed Silica Gel 60 HPTLC plates from EMD Chemicals (Gibbstown, NJ).

Mass spectrometry analysis of pgl deletion mutant-secreted metabolite. L. monocytogenes cultures were grown to an OD₆₀₀ of 2.0 in BHI broth. Cultures (35 ml) were transitioned to a solution of 0.5% glucose and incubated at 37°C for 2 h. Cell-free supernatants were lyophilized, resuspended in 300 μl deionized water, and analyzed by quadrupole time-of-flight mass spectrometry (Q-Tof Premier; Waters, Milford, MA).

Microarray analysis of L. monocytogenes gene expression.Oligonucleotides for L. monocytogenes arrays were synthesized by The Institute for Genomic Research (TIGR), and the arrays were printed at the UCSF Center for Advanced Technology. Exponentially growing BHI cultures of wt and pgl deletion mutant L. monocytogenes (OD₆₀₀ of 0.5) were filtered and frozen in liquid nitrogen. Bacteria were washed off the filter, and bacterial RNA was isolated using phenol-chloroform extraction. Bacterial RNA was amplified using the MessageAmp II Bacteria Prokaryotic RNA kit (Ambion). Microarrays were gridded using Genepix and SpotReader and analyzed using Acuity software (6, 16). Genes that showed at least a twofold and statistically significant difference (by significance analysis of microarrays [38] with a false-discovery rate of 1.3%) from wt gene expression were selected for further discussion. Each sample was analyzed in duplicate.

Deletion of a previously uncharacterized L. monocytogenes gene results in increased activation of a host cytosolic surveillance pathway.We recently performed a genetic screen to identify L. monocytogenes Tn917 mutants that triggered altered expression of IFN-β from infected host cells (6). One mutant identified in the screen had a Tn917 insertion between two predicted genes, lmo0558 and lmo0559 (Fig. 1a). An in-frame deletion of lmo0558 recapitulated the phenotype seen with the Tn917 mutant, whereas a mutant with an in-frame deletion of lmo0559 induced the same levels of IFN-β as wt L. monocytogenes (Fig. 1b). Complementation of lmo0558 cloned into an integration vector (15) restored the IFN-β induced by the lmo0558 deletion mutant to wt levels (Fig. 1c). Activation of IFN-β by wt L. monocytogenes and the lmo0558 mutant was largely independent of the Toll-like receptor adaptors MyD88 and TRIF (Fig. 1c) (19). These results indicated that the lmo0558 mutant induced increased activation of the cytosolic surveillance system.

Growth of the lmo0558 mutant in vitro.lmo0558 is homologous to E. coli K-12 pgl (ybhE) (38.9% similar and 22% identical), which encodes 6-phosphogluconolactonase (Pgl), the second enzyme in the oxidative branch of the PPP (Fig. 1d) (35). As one might expect from deletion of a prominent metabolic gene, the lmo0558 deletion mutant grew to a substantially lower OD than wt L. monocytogenes in glucose-limiting minimal medium (12, 24) (Fig. 2). This defect was rescued by the addition of glucose to the culture (Fig. 2c) or during growth in excess-glucose medium (Fig. 2b). Surprisingly, wt L. monocytogenes and the lmo0558 deletion mutant grew at identical rates in glucose-limiting minimal medium, but the mutant ceased replicating earlier than the wt (Fig. 2b). Abortive growth was not due to an inability of this mutant to grow to a higher OD, as the wt and the deletion mutant grew to identical ODs in medium containing excess glucose. Similarly to the case for the glucose-rich medium, growth of the lmo0558 deletion mutant was indistinguishable from that of wt bacteria inside of macrophages (Fig. 2d).

• Open in new tab
• Download powerpoint

FIG. 2.

Growth of the lmo0558 mutant in vitro. (a) Summary of growth rates and stationary-phase OD₆₀₀s of wt and pgl deletion mutant L. monocytogenes grown in glucose-limiting or excess-glucose minimal medium (mean ± one standard deviation; n = 4). (b) Representative growth curves of wt (•) and pgl deletion mutant (▪) L. monocytogenes in glucose-limiting or excess-glucose medium (n = 2). (c) Excess glucose added (arrow) to wt (•) and pgl deletion mutant (▪) cultures 1 h after the cessation of growth in glucose-limiting medium (n = 2). (d) Intracellular growth curves of wt L. monocytogenes (•) and the pgl deletion mutant (▪) in bone marrow-derived macrophages. Error bars represent one standard deviation.

The lmo0558 deletion mutant accumulates and secretes a glucose-derived metabolite. E. coli pgl deletion mutants are unable to catalyze the conversion of 6-phosphogluconolactone into 6-phosphogluconate, and the flow of glucose through the PPP is arrested at 6-phosphogluconolactone (14, 35). This accumulated metabolite is dephosphorylated (presumably inside the bacterium), and the resulting gluconolactone is exported into the medium (14). To determine if the lmo0558 deletion mutant secreted a metabolite derived from glucose, we suspended growing cultures of the wt or the lmo0558 mutant in PBS that contained a trace amount of ¹⁴C-labeled glucose, either labeled on C-1 or universally labeled on each carbon. The lmo0558 deletion mutant had a glucose-derived metabolite in the cell pellet that was absent from the wt (Fig. 3). In addition, while wt L. monocytogenes did not secrete a detectable amount of this metabolite, the lmo0558 deletion mutant secreted it in abundance (Fig. 3). ¹⁴C labeling of the secreted metabolite was the same regardless of whether glucose was universally labeled or labeled only on C-1. Therefore, deletion of lmo0558 resulted in the accumulation and secretion of a glucose-derived metabolite that retained the original C-1 of glucose. This observation was specific to the lmo0558 mutant, as the previously described hyper-IFN-β-inducing marR deletion and tetR::Tn917 mutants did not exhibit enhanced secretion of a glucose derived metabolite (Fig. 3) (6).

• Open in new tab
• Download powerpoint

FIG. 3.

Accumulation and secretion of a glucose-derived metabolite by lmo0558 deletion mutant L. monocytogenes. Growing cultures of wt, pgl deletion mutant, and MDR-overexpressing tetR::Tn917 and marR deletion strains were labeled with glucose that was ¹⁴C labeled on carbon 1 (1-Glc) or universally labeled (UL-Glc), and supernatants and pellets were analyzed by TLC.

Identification of the secreted glucose-derived metabolite as gluconate and identification of lmo0558 as pgl.Mass spectrometry was used to identify the glucose-derived metabolite. While glucose was observed in supernatants from both the wt and lmo0558 deletion strains, a second molecule at m/z 195.1 was observed only in supernatant from the lmo0558 mutant strain (Fig. 4). Accurate mass measurements and collision-induced dissociation confirmed the identity of the metabolite at m/z 195.1 as gluconate (Fig. 4 and data not shown). Thus, the metabolic phenotype of the lmo0558 mutant mirrors the phenotype of an E. coli pgl deletion mutant. In the E. coli pgl deletion mutant, the accumulating 6-phosphogluconolactone is dephosphorylated and the resulting gluconolactone is exported. Subsequently, the vast majority of the labile gluconolactone spontaneously hydrolyzes to gluconate (14). Indeed, the half-life of gluconolactone is estimated to be only 10 min (5). Due to practical considerations (see Materials and Methods), the vast majority of gluconolactone in our samples would have hydrolyzed to gluconate; therefore, we were unable to differentiate between gluconolactone and gluconate. These data provide compelling evidence that lmo0558 encodes 6-phosphogluconolactonase, and it will now be referred to as pgl.

• Open in new tab
• Download powerpoint

FIG. 4.

Identification of gluconate in the pgl deletion mutant supernatant. Electrospray ionization mass spectra recorded in the negative ion mode for supernatants from wt (A) and pgl deletion mutant (B) Listeria monocytogenes are shown. The singly charged negative ion at m/z 195.1 corresponds to the (M − H)⁻ ion of gluconic acid (M = C₆H₁₂O₇). Mass spectral intensities are normalized to the intensity of the ion at m/z 215.0, which corresponds to the (M + Cl)⁻ ion of glucose (M = C₆H₁₂O₆). The peak at m/z 217.0 is part of the isotopic distribution of the latter ion (chlorine occurs naturally as a mixture of ³⁵Cl and ³⁷Cl).

pgl deletion results in increased expression of bacterial β-glucosidases but no change in expression of bacterial MDRs.To determine the global effect that deletion of pgl had on L. monocytogenes gene expression, we used microarray analysis to compare total gene expression of wt and pgl mutant mid-log-phase bacteria. The most striking finding was that the pgl deletion mutant had increased expression of several predicted β-glucosidases, which accounted for 4 out of the 10 most significant differences in gene expression (Table 1). Upregulation of β-glucosidases is consistent with the observation that 6-phosphogluconolactone is an inhibitor of β-glucosidases, and pgl mutant L. monocytogenes may have increased expression to compensate for the inhibitory effect of the accumulating metabolite. There was no increase in the expression of the bacterial MFS family transporters that were previously shown to contribute to the induction of IFN-β (6).

pgl deletion enhances activation of host IFN-β expression independently of mdrM, and L. monocytogenes mutants overexpressing MFS transporters do not secrete a glucose-derived metabolite.While the enhanced IFN-β induction by the pgl deletion mutant was not due to an increase in expression of either of the IFN-β inducing MFS transporter genes, mdrM or mdrT, it was still possible that gluconolactone and/or gluconate depended on these transporters for secretion. To address this possibility, we used ¹⁴C-labeled glucose to determine whether strains overexpressing MFS transporters (tetR::Tn917 and marR deletion strains) exported a glucose-derived metabolite. Figure 3 clearly shows that overexpression of these transporters did not result in increased secretion of any detectable glucose-derived metabolite.

We next used a genetic approach to more conclusively test for an interaction between pgl and L. monocytogenes mdrM. The only MFS transporter involved in the activation of IFN-β by wt L. monocytogenes, and the only one expressed at a detectable level, is mdrM. If mdrM was required for the increased induction of IFN-β by the pgl deletion mutant, then a mdrM pgl double deletion mutant would induce an amount of IFN-β equal to that of the mdrM deletion mutant alone. Instead, deletion of pgl in the mdrM deletion background rescued the IFN-β induction by the mdrM mutant to wt levels (Fig. 5).

• Open in new tab
• Download powerpoint

FIG. 5.

mdrM is not required for the increased IFN-β expression induced by pgl deletion mutant L. monocytogenes. qRT-PCR analysis of IFN-β gene expression in bone marrow-derived macrophages in response to infection with wt L. monocytogenes, the mdrM mutant, the pgl mutant, or the mdrM pgl double deletion mutant or in uninfected (un) bone marrow-derived macrophages is shown. All error bars represent one standard deviation (n = 4). *, P = 0.009; **, P = 0.002.

Growth of the pgl deletion mutant in mice and sensitivity to oxidative stress.To test the role of pgl in L. monocytogenes growth in vivo, we infected mice intravenously with wt or pgl mutant L. monocytogenes. At 48 h after infection of mice with 1 LD50, the pgl deletion mutant exhibited a 15- to 30-fold growth defect in the liver and spleen (Fig. 6). Interestingly, Salmonella enterica serovar Typhimurium mutants lacking a functional PPP also display decreased growth in vivo (18) and increased sensitivity to oxidative stress (18), likely due to the critical role that the PPP plays in regenerating reducing power in the form of NADPH (40). Therefore, we used a disc diffusion assay (9) to examine the sensitivities of wt and pgl mutant L. monocytogenes to hydrogen peroxide and the thiol oxidant diamide (37), or to the bactericidal antibiotic carbenicillin as a control. The results (Table 2) clearly show that the pgl deletion mutant was more sensitive to diamide and hydrogen peroxide but not to the antibiotic control.

• Open in new tab
• Download powerpoint

FIG. 6.

Growth of pgl deletion mutant L. monocytogenes in vivo. C57BL/6 mice were infected with 1 × 10⁵ CFU (1 LD₅₀) of wt L. monocytogenes, pgl deletion mutant, or complemented pgl deletion mutant bacteria. Organs were collected at 48 hpi, and bacterial numbers are represented as CFU per organ (n = 10 mice per strain). Statistical significance was determined by the nonparametric Mann-Whitney test (**, P = 0.0002; *, P = 0.002).