As mentioned above, over-expression of MdrM results in enhanced triggering of type I interferon response in a manner that is dependent on c-di-AMP secretion (Woodward et al., 2010; Yamamoto et al., 2012). To better understand the physiological pathways involved in MdrM's function, we performed an unbiased genetic screen to identify L. monocytogenes genes that differentially modulate this MdrM-dependent type I interferon response. To achieve this goal a L. monocytogenes Himar-mariner1 transposon library was generated in the background of the ΔmarR strain (deleted of MdrM's repressor gene, marR), which highly expresses MdrM. As shown previously, this strain induces 2 to 3-fold higher levels of IFN-β in comparison to wild type (WT) bacteria upon infection of macrophage cells (Crimmins et al., 2008). Briefly, we used a previously described screening approach whereby IFN-β is measured in the supernatants of L. monocytogenes-infected cells by an ISRE-type I interferon reporter cell line that expresses luciferase in response to type I interferons (Jiang et al., 2005; Crimmins et al., 2008). Screening of ~6000 L. monocytogenes ΔmarR::marinerTn mutants in 96-well format uncovered 12 and 16 low and high inducers of IFN-β, respectively, as compared to the ΔmarR parental strain (Table 1). Among the low inducers, we identified mutants in genes encoding the flagellar system, the virulence master regulator PrfA and the LLO toxin, all factors that are known to mediate adhesion and invasion into host cells. We also identified genes related to cell wall and purine metabolism as well as unknown genes (Table 1A). Among the high inducers, we identified genes associated with lipoteichoic acid (LTA) biosynthesis, amino acid metabolism, cell division, transport and surface proteins (Table 1B). Notably, within this list, three genes were identified to be directly involved in LTA biosynthesis. LMRG_01692.6 (lmo2555) and LMRG_01693.6 (lmo2554) encode LTA anchor formation proteins, LafA and LafB respectively, and are both cytoplasmic glycosyltransferases involved in the synthesis of the LTA glycolipid anchor Gal(α1-2)Glc(α1-3)-diacylglycerol (Gal-Glc-DAG). The third LTA biosynthesis gene, LMRG_00310.2 (lmo0644), encodes the LTA primase, LtaP that transfers the first glycerolphosphate group onto the Gal-Glc-DAG anchor (Figure 1A) (Webb et al., 2009). Since these genes represent the first three sequential and critical steps of the LTA biosynthesis pathway, we decided to further investigate their role in triggering of the type I interferon response.

First, to validate the high-inducer phenotype, bone marrow derived (BMD) macrophage cells were infected with each of the LTA mutants, ΔmarR/lafA::Tn, ΔmarR/lafB::Tn, ΔmarR/ltaP::Tn, and real time quantitative PCR (RT-qPCR) analysis of IFN-β expression was performed at 6 h post infection (h.p.i.). As shown in Figure 1B, all the three mutants triggered higher levels of IFN-β relative to the parental strain ΔmarR and WT bacteria. Next, we examined whether these mutants are indeed defective in LTA synthesis. The mutants were grown in brain heart infusion (BHI) medium, in which each grew like WT bacteria (data not shown), and their LTA was extracted at mid-exponential phase as previously described (Webb et al., 2009). LTA was visualized via Western blot analysis using a polyglycerolphosphate-specific antibody (Webb et al., 2009). In line with previous studies (Webb et al., 2009), bacteria-bearing transposons in the lafA or lafB genes exhibited severely reduced LTA synthesis, whereas bacteria bearing a transposon in the ltaP gene displayed a reduced and modified LTA production (Figure 1C). This analysis established that the transposon mutants are indeed deficient in LTA biosynthesis. Next, the ability of the LTA mutants to grow intracellularly in BMD macrophages was evaluated. Remarkably, we observed that although the mutants hardly produce LTA, they grow efficiently in macrophage cells like WT bacteria (Figure 1D).

To further investigate the impact of aberrant LTA synthesis on triggering of the host innate immune system, a series of in-frame deletion mutants were generated in lafA, lafB, and ltaP genes, as well as in the lafC gene that was also shown to be involved in LTA production (Figure 1A) (Webb et al., 2009). Gene deletions were generated in the background of both WT and ΔmarR bacteria (Table 2). As demonstrated in Figure 2, all mutants grew similarly to WT bacteria in BHI medium and intracellularly in BMD macrophage cells (Figures 2A,B). Nevertheless, the deletion mutants exhibited an LTA profile similar to that described by Webb et al. (2009), namely the lafC and ltaP mutants exhibited high-molecular weight LTA structures detectable by Western blots using a polyglycerolphosphate-specific antibody, whereas lafA and lafB mutants hardly produced detectable LTA (Figure 2C). Of note, introducing a copy of lafA, lafB, lafC, and ltaP genes on the integrative plasmid pPL2 (resulting in pPL2lafA, pPL2lafB, pPL2lafC, and pPL2ltaP) (Table 2) completely restored LTA production in the corresponding LTA mutants and in the ΔmarR isogenic strains (Figure 2D). Taken together, these data validate the previously described roles of lafA, lafB, lafC, and ltaP genes in LTA synthesis, but more importantly demonstrate that LTA is dispensable for L. monocytogenes extracellular growth in broth and intracellular growth in mammalian cells. These findings were somewhat surprising, as LTA is known to be an important polymer in the cell wall of Gram-positive bacteria and therefore expected to influence bacterial virulence (Weidenmaier and Peschel, 2008; Reichmann and Grundling, 2011).

To gain further insight into the innate immune response elicited upon infection with the various LTA mutants, BMD macrophage cells were infected with ΔlafA, ΔlafB, ΔlafC, and ΔltaP mutants in WT or the ΔmarR background, and IFN-β, IL-6 (which is also part of the type I interferon response to L. monocytogenes infection) and TNF-α transcript levels were analyzed at 6 h.p.i. using RT-qPCR. As shown in Figure 3, each LTA mutant triggered increased level of IFN-β and IL-6 in comparison to his parental strain, without any detectable effect on TNF-α transcription (Figures 3A–E). Further, IFN-β induction by the LTA mutants could be restored by complementing their respective genes on a pPL2 plasmid (Figures 3A,B). Notably, the transcription levels of IFN-β and IL-6 were greater in cells infected with the ΔmarR-LTA mutants than in cells infected with the LTA mutants in a WT background (~2-fold). These results demonstrate that although deletion of each LTA biosynthesis gene is sufficient to trigger an enhanced type I interferon response, an additive effect is observed when combined with the ΔmarR deletion (which leads to over-expression of MdrM). Overall the data supports that aberrant LTA synthesis in L. monocytogenes triggers activation of the type I interferon response and not of other pro-inflammatory responses, such as those mediated by TNF-α. It is surprising that although LTA is a known immuno-stimulatory pathogen-associated microbial pattern (PAMP) (Takeuchi and Akira, 2001; Draing et al., 2008), we found that reduced bacterial LTA synthesis is associated with enhanced activation of the innate immune system. One possible explanation for this host-pathogen phenotype is that aberrant production of LTA triggers a bacterial response which, in turn, triggers activation of innate immune signaling pathways.

Since we observed that aberrant LTA synthesis and MdrM over-expression each trigger enhanced type I interferon responses (and trigger additively in combination), we examined if each induce IFN-β independently or converge on the same mechanism. To this end, ΔlafA was combined with several MDR deletion mutants, ΔmdrM, ΔmdrMTA (a triple mutant of mdrM, mdrT, and mdrA genes), and ΔmdrMTAC (a quadruple mutant of mdrM, mdrT, mdrA, and mdrC genes), which were shown to mediate most of the IFN-β response to L. monocytogenes infection (Kaplan Zeevi et al., 2013). BMD macrophage cells were infected with the combination mutants, ΔmdrM/ΔlafA, ΔmdrMTA/ΔlafA, and ΔmdrMTAC/ΔlafA (Table 2) and with the respective parental mutant strains and WT bacteria, and IFN-β transcript levels were analyzed at 6 h.p.i. As shown in Figure 4A, the enhanced induction of IFN-β by the ΔlafA mutant was completely dependent on the MDR transporters, as it was abolished in the background of the MDR deletion mutants. The data indicates that deletion of mdrM alone is enough to reduce most of the IFN-β response triggered by the ΔlafA mutant, though deletion of more MDR genes reduced this response further (Kaplan Zeevi et al., 2013). Similar results were obtained with ΔlafB, ΔlafC, and ΔltaP mutants, as shown in Figure 4B in combination with the quadruple ΔmdrMTAC mutant. Notably, these combined mutants hardly trigger IFN-β though grow normally in macrophage cells like WT bacteria (Figure 4C). Overall, these results suggest that the MDR transporters mediate the enhanced immuno-stimulatory phenotype of the LTA mutants.

Since these MDR transporters are considered to facilitate secretion of immuno-stimulatory molecules, as shown for c-di-AMP, we investigated if deficient LTA production triggers enhanced secretion of such molecules. To examine this hypothesis, the immuno-stimulatory potency of ΔlafA and WT bacteria culture supernatants were analyzed. Briefly, bacteria were grown to mid-exponential phase in minimal-defined medium (MDM), their supernatants were collected and added to BMD macrophage cells and IL-6 transcription levels were analyzed 6 h later. In line with our hypothesis, the supernatant derived from the ΔlafA mutant activated a higher IL-6 response as compared to that derived from WT bacteria (Figure 4D). Similarly, the supernatant of the double mutant ΔmarR/ΔlafA was even more immune-stimulatory than that of ΔmarR or ΔlafA alone (Figure 4D). Together these data support the model that defective LTA production leads to secretion of immune-stimulatory bacterial moieties that can be sensed by the innate immune system. Unfortunately, at this point we could not confirm nor exclude c-di-AMP as the secreted molecule.

To further explore the relationship between LTA synthesis and MDR transporters, we performed an independent experiment that evaluated LTA production in the various MDR deletion mutants. To this end, ΔmdrM, ΔmdrMTA, and ΔmdrMTAC mutants were grown to mid-exponential phase in BHI medium at 37°C, and their LTAs were extracted and compared with those of ΔmarR and WT bacteria using Western blot analysis. Remarkably, we found that while most of the mutants exhibited an LTA profile similar to that of WT bacteria, ΔmdrMTAC mutant exhibited an altered profile, specifically missing a high molecular weight LTA band in the gel (Figure 5A). Careful examination of the LTA Western blots indicated that indeed two separate LTA bands appear (marked with asterisks), one that is thin with a high molecular weight and the other in many cases much thicker slightly lower in the gel. These bands, which can be easily missed if the LTA samples are overloaded, potentially represent different LTA variants.

In this regard, distinct LTA variants harboring different structures have been reported previously for L. monocytogenes strains (Hether and Jackson, 1983; Uchikawa et al., 1986; Dehus et al., 2011). For example, a recent report by Dehus at el. demonstrated a growth temperature-dependent expression of two LTA structural variants, named LTA1 and LTA2 (Dehus et al., 2011). LTA1 was shown to be produced at both 30°C and 37°C and comprises the canonical LTA structure. In contrast, LTA2 was shown to be produced predominantly at 37°C and harbors two diacyl-glycerol moieties, instead of one, linked to the Gal-Glc disaccharide (Dehus et al., 2011). To characterize in more detail the LTA deficiency associated with MDR transporters, WT and ΔmdrMTAC bacteria were grown at 30 or 37°C temperatures and their LTAs were extracted and analyzed. As shown in Figure 5B, WT bacteria grown at 37°C exhibited a more intense upper LTA band in comparison to bacteria grown at 30°C, in accordance with LTA2 production. Notably, ΔmdrMTAC mutant was completely missing the upper LTA band at 30°C, and demonstrated a weak band at 37°C in comparison to WT bacteria. These results suggest that the MTAC transporters contribute to the production of this LTA variant, which appears to correspond to LTA2. Of note, WT and ΔmdrMTAC bacteria exhibited comparable levels of the lower LTA variant.

To better delineate between the two LTA variants, the LTAs of WT and ΔmdrMTAC bacteria (grown at 37°C) were further isolated using a water/butanol extraction method and then separated by reverse phase HPLC using an analytic C8 column (Morath et al., 2001; Draing et al., 2006). Western blot analysis of the HPLC fractions clearly demonstrated that in the ΔmdrMTAC mutant one LTA variant is observed across the HPLC fractions. In contrast, WT bacteria produce several variants differing in molecular weight, which are distributed throughout the fractions (Figure 5C). Taken together, these results demonstrate that the ΔmdrMTAC mutant exhibits an altered LTA profile in comparison to WT bacteria, largely missing one or more distinct LTA variants.

Lastly, we assessed the immuno-stimulatory potency of the LTA of ΔmdrMTAC vs. WT bacteria. Bacteria were grown to mid-exponential phase in BHI at 37°C and LTAs were extracted using water/butanol as described previously (Morath et al., 2001; Draing et al., 2006). Purified LTAs were then added to BMD macrophage cells, and IL-6 transcription levels were analyzed using RT-qPCR. Of note, IL-6 was utilized as the readout since L. monocytogenes cell wall preparations do not induce much IFN-β when added to macrophage cells extracellularly (unpublished data). As shown in Figure 5D, the LTA of ΔmdrMTAC mutant was much less immuno-stimulatory than the LTA of WT bacteria, demonstrating that ΔmdrMTAC LTA is not only different in its composition but also in its ability to activate the innate immune system. Importantly, for the first time these observations establish a relationship between MDR transporters and LTA synthesis and furthermore, implicate a role for MDR transporters in LTA production/regulation in bacteria.

L. monocytogenes 10403S strain was used as WT strain and as parental strain for all mutants generated in this work (Table 2). E. coli XL-1 blue strain (Stratagene) was used for vector propagation. E. coli SM-10 strain (Simon et al., 1983) was used for conjugative plasmid delivery to L. monocytogenes bacteria. L. monocytogenes strains were grown in BHI (Merck^©) medium or MDM (Phan-Thanh and Gormon, 1997) at 37°C and E. coli strains were grown in Luria Bertani (LB, Acumedia^©) medium at 37°C. For infection experiments L. monocytogenes bacteria were grown overnight in BHI at 30°C without agitation. Primary BMD macrophages were isolated from 6 to 8 weeks old female C57BL/6 mice (Harlan Laboratories Ltd, Israel) and cultured as described (Portnoy et al., 1989). ISRE-L929 reporter cell line was used to evaluate the levels of IFN-β secreted to the supernatants of infected macrophage cells (a gift from Bruce Butler, University of Texas Southwestern Medical Center) (Jiang et al., 2005).

L. monocytogenes library of Mariner1-Tn mutants was constructed as described in (Zemansky et al., 2009) using 10403S ΔmarR Listeria monocytogenes mutant as a background strain. Screening was performed as described in Crimmins et al. (2008). Briefly, a total of 6000 mutants were grown in BHI medium in 96-well plates overnight at 30°C. BMD macrophages from C57BL/6 mice were plated in 96-well plates, 4 × 10⁴ macrophages per well. Macrophages in each well were infected with 2 × 10⁶ bacterial cells. Thirty minutes post-infection, macrophages were washed and gentamicin was added (50 μg/ml) to kill extracellular bacteria. At 6 h post infection (h.p.i.), macrophage culture supernatants were taken and frozen at −80°C. The amount of IFN-β in the culture supernatants was detected using the type I interferon reporter cell line (ISRE-L929) (Jiang et al., 2005). Reporter cells were grown in 96-well plates and incubated with 10× dilution of the macrophage culture supernatants for 4 h. Cells were lysed using Glo Lysis Buffer (Promega) and luciferase activity was detected using the Beetlejuice D-Luciferine kit (PJK) and measured in a luminescence plate reader (Biotek Synergy HT). Transposon insertion sites were identified as described in Zemansky et al. (2009). Gene deletion (in frame) mutants were generated by standard techniques using pKSV7oriT derivate vector (pBHE261) (Lauer et al., 2008).

Intracellular L. monocytogenes growth was performed as described previously (Lobel et al., 2012). Briefly, 2 × 10⁶ BMD macrophage cells were seeded on a petri dish with glass cover slips and infected with 8 × 10⁶ bacteria. At 0.5 h.p.i. cells were washed and at 1 h.p.i. gentamicin (50 μg/ml) was added. At each time point, cells from 3 cover slips were lyzed, diluted, and plated on BHI agar plates. Bacterial CFUs were counted after overnight growth at 37°C. For extracellular growth, overnight L. monocytogenes cultures were adjusted to OD_600nm of 0.02 in 200 μl of fresh BHI medium and were further grown in a Synergy HT Biotek® plate reader at 37°C with continuous shaking.

LTA extraction was modified from a previously described method by Webb et al. (2009). Overnight cultures were adjusted to OD_600nm of 0.05 in 20 ml of BHI medium. Cultures were grown to OD_600nm of 0.5 then washed with Buffer-A (20 mM Tris-HCl pH-8, 0.5 M NaCl, and 1 mM EDTA) and resuspended with 1 ml of the same buffer. Samples were vortexed with 0.5 ml glass beads 425–600 μm (Sigma Aldrich) for 45 min, and then centrifuged for 1 min at 200 g to pellet the beads. Five hundred microliter of the supernatant were pelleted by centrifugation at 20,000 g for 15 min, at 4°C. The supernatants were discarded and the pellets were resuspended in 50 μl of Buffer-A with 2% SDS. The samples were boiled for 20 min and centrifuged at high-speed for 1 min. The supernatants were analyzed for total protein content by Lowry assay, and samples with equal amounts of total proteins were separated in a 15% SDS-polyacrylamide gel and probed with polyglycerolphosphate-specific LTA antibody (Clone 55, Hycult biotechnology) and HRP-conjugated goat anti-mouse IgG (Jackson ImmunoResearch, USA) used at 1:1000 and 1:20,000 dilutions, respectively. Western blots were developed by enhanced chemiluminescence reaction (ECL).

For further analysis, LTA was purified by water/butanol extraction protocol as described previously (Morath et al., 2001; Draing et al., 2006). In brief, bacteria cultivated overnight were adjusted to OD_600nm 0.1 in 200 ml of BHI medium, and grown to OD_600nm of 1 at 37°C. Bacterial cultures were then kept on ice for 45 min, and washed with 0.1 M sodium citrate (pH = 4.7). Bacterial pellets were suspended in 20 ml of the same buffer and lysed by French-Press (Stansted Fluid Power, UK) at 16,000 psi. Bacterial lysates were centrifuged at 17,000 g for 20 min and suspended in 10 ml of 0.1 M sodium citrate and in an equal amount of 1-butanol (Merck). Samples were stirred for 30 min and then centrifuged at 17,000 g for 20 min to achieve a two-phase separation. The lower aqueous phase was collected and a fresh phase of 10 ml sodium citrate 0.1 M was added. Extraction was repeated two more times, yielding a total of 30 ml of LTA extract. Samples were then evaporated by rotary evaporator, suspended in 10 ml of water and lyophilized.

Ten milligram of dry LTA was resuspended in 0.5 ml of chromatography start buffer (15% 1-propanol (Sigma Aldrich) in 0.1 M ammonium acetate, pH = 4.7) and centrifuged at 27,000 g for 60 min. The supernatant was subjected to reverse phase high-performance liquid chromatography (RP-HPLC) (JASCO) on a C8 analytic column (Symmetry C8, 5 μm, 3.9 × 150 mm, Waters) using a linear gradient of 15–60% of 1-propanol in 0.1 M ammonium acetate (pH = 4.7). Two milliliter fractions were collected lyophilized and resuspended again for Western blot analysis. The protocol was slightly modified from (Lehner et al., 2001; Grundling and Schneewind, 2007)

Overnight cultures were inoculated in a 20 ml of MDM (Phan-Thanh and Gormon, 1997) at OD_600nm 0.02. Cultures were grown at 37°C to OD_600nm 0.5, and then were centrifuged at 3800 rpm for 20 min at 4°C. Supernatants were filtered through a 0.2 μm filter device and stored at −80°C. 2 ml of supernatants were lyophilized and resuspended in 100 μl of sterile water from which 50 μl were added to BMD macrophage cells for 6 h.

RNA from infected macrophages or macrophages treated with bacterial supernatants or LTA extracts was obtained using Trizol reagent according to standard protocols. One microgram (1 μg) of RNA was reverse transcribed to cDNA using High Capacity reverse transcription kit® (Applied Biosystems). RT-qPCR was performed on 10 ng of cDNA using SYBER Green® with Step-one Plus RT-PCR system (Applied Biosystems). The transcription level of macrophage cytokines was normalized to the reference gene GAPDH. Statistical analysis was performed using the StepOne™ V2.1 software. Error bars represent a 95% confidence interval, i.e., the value is expected to fall within the bar range in 95% of repeat experiments. When the error bars of two samples do not overlap, the significance of the difference (p-value) is <<0.01.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.