Introduction

Listeria monocytogenes is a facultative human pathogen that can cause serious infections after ingestion. To establish an infection, the bacterium first crosses the intestinal-blood barrier by invasion of gut epithelial cells and subsequent transcytosis to the basolateral side of infected cells, where it is released to the bloodstream (Quereda et al., 2021). The liver is then the main primary replicative niche of the pathogen, where the bacterium invades hepatocytes, replicates intracellularly and spreads from cell to cell (Koopmans et al., 2023). From there, the bacterium can spread hematogenously to the placenta of pregnant women, where it invades syncytiotrophoblasts to cross the placental barrier and ultimately causes fetal infections (Charlier et al., 2020). L. monocytogenes can also invade the brain, possibly achieved by a mechanism similar to that used to cross the intestinal barrier or by infected macrophages that circulate in the blood and are able to cross the blood brain barrier (Disson & Lecuit, 2012). Intra-axonal transport of L. monocytogenes from peripheral sites to the brainstem is another proposed mechanism of brain invasion (Bagatella et al., 2022).

The bacterium replicates inside the cytoplasm and exploits polymerization of host cell actin to drive intracellular locomotion and spread from cell to cell (Pizarro-Cerda & Cossart, 2018, Quereda et al., 2021). To ensure rapid intracellular replication, L. monocytogenes has adapted to the specific nutrient availability within the host cell cytoplasm using specific uptake systems such as the hexose phosphate transporter Hpt, enzyme I of the phosphoenolpyruvate:sugar phosphotranferase system, the Opp oligopeptide permease as well as the Cta and Tcy cysteine transporters (Borezee et al., 2000, Chico-Calero et al., 2002, Xayarath et al., 2009, Freeman et al., 2025), illustrating the specific importance of certain carbon and nitrogen sources for intracellular nutrition. Likewise, several genes required for the biosynthesis of different cellular building blocks such as purines, aromatic amino acids or menaquinone are required for intracellular replication, while their deletion is tolerated during growth in complex laboratory medium (Stritzker et al., 2004, Faith et al., 2012, Smith et al., 2021, Fischer et al., 2022). The conditional essentiality of biosynthetic genes in the host cell cytoplasm shows that certain limitations in the nutrient availability exist intracellularly that L. monocytogenes does not encounter in complex laboratory media.

Folate is one of the compounds, which apparently becomes limiting in the host cell as some folate biosynthesis genes are essential for intracellular growth, although they are dispensable during growth in BHI broth (Zhang et al., 2022, Feng et al., 2023, Stamm et al., 2024). The biologically active form of folate is tetrahydrofolate (THF) that transfers one carbon units (1C) to various substrates. 1C-THF species are important for biosynthesis of purines and pyrimidines as well as for the formation of serine, methionine and N-formylmethionine (Green & Matthews, 2007).

Three different pathways ensure 1C-THF formation in L. monocytogenes: (i) a two enzyme reaction mediated by the formate-THF ligase Fhs and the bifunctional N5,N10-methylene-THF dehydrogenase/cyclohydrolase FolD (Feng et al., 2023), (ii) the serine hydroxymethyltransferase GlyA (Schirch et al., 1985), and (iii) the aminomethyltransferase GcvT from the glycine cleavage system (Fujiwara et al., 1984) (Fig. 1A). The important role of 1C-THF biosynthesis for intracellular nutrition is reflected by the strong attenuation of a folD mutant in macrophages (Feng et al., 2023). Likewise, a mutant lacking the gcvPAB genes, encoding the two subunits of glycine dehydrogenase, which is another crucial component of the glycine cleavage system, shows impaired replication in macrophages (Fischer et al., 2022). The glycine cleavage system (GCS) is a multi-enzyme system that catalyzes the breakdown of glycine, in which the GcvPAB enzyme complex mediates the oxidative decarboxylation of glycine as the first step (Fig. 1A). An important function of this pathway is the generation of N5,N10-methylene-THF to replenish the 1C-THF pool (Kikuchi et al., 2008).

Figure 1.

We here have continued our previously initiated investigation on the attenuated phenotype of a L. monocytogenes ΔgcvPAB mutant (Fischer et al., 2022). Our results indicate that attenuation of this mutant results from limited 1C-THF availability. Isolation of ΔgcvPAB suppressor mutants restoring attenuation led to the discovery of a mutation that reactivates the fhs/folD 1C-THF biosynthesis pathway, which is naturally inactivated by a premature stop codon in the fhs gene of L. monocytogenes strain EGD-e, a widely used laboratory strain. Furthermore, the three 1C-THF generating pathways were found to be synthetic lethal, further illustrating the importance of folate biosynthesis for growth and virulence of L. monocytogenes.

Results

In vitro virulence of a L. monocytogenes ΔgcvPAB mutant

We have demonstrated previously that a L. monocytogenes ΔgcvPAB mutant replicates with reduced growth rate in mouse macrophages and is impaired in cell-to-cell spread in 3T3 mouse fibroblasts (Fischer et al., 2022). To further support this observation, we here complemented the ΔgcvPAB mutant with an IPTG-inducible gcvPAB copy and re-analyzed cell-to-cell spread. In agreement with our previous results, plaque formation was strongly reduced in 3T3 cells infected with the ΔgcvPAB mutant (15±19% of wild type level) and small plaques were formed by the complemented strain in the absence of IPTG (50±19%). However, plaque formation was restored when IPTG was added (105±24%, Fig. 1B-C), as expected.

To further study this virulence defect, dissemination of a ΔgcvPAB strain expressing the red fluorescent protein DsRed-Express in infected 3T3 cultures was analyzed microscopically. The wild type and the ΔgcvPAB mutant were found disseminated throughout the cytoplasm of infected cells and neighbor cells also contained bacteria (Fig. S1). In contrast, the ΔactA mutant, which cannot spread due to the absence of the actin tail nucleating ActA protein (Kocks et al., 1992), formed concentrated foci of fluorescent bacteria in the cytoplasm of infected cells and neighbor cells were usually not infected (Fig. S1). As the phenotypes of the ΔgcvPAB and ΔactA mutants were different in this assay, the plaque formation defect of the ΔgcvPAB mutant must either be caused by an inability to invade the cells or to replicate within them. To discriminate between these possibilities, we next quantified invasion and intracellular replication of the ΔgcvPAB mutant in comparison to well characterized mutants lacking either the hly or actA genes in 3T3 cells. As can be seen in Fig. 2A, invasion into 3T3 cells was not impaired, however, intracellular growth was retarded compared to wild type and to the ΔactA mutant, which cannot spread at all but otherwise grows normally (Kocks et al., 1992). Similarly, intracellular growth of the ΔgcvPAB mutant was delayed in J774 macrophages (Fig. 2B), however, no delay was detected in HepG2 hepatocytes (Fig. 2C). The ΔgcvPAB mutant was as hemolytic (Fig. 2D-E) and as resistant against lysozyme as the wild type (Fig. 2F). Differences in hydrogen peroxide sensitivity were also not found as the minimal inhibitory H2O2 concentrations was 3.1 mM for both strains. Taken together, the ΔgcvPAB mutant shows delayed intracellular growth in fibroblasts, explaining the plaque formation defect, and the delayed growth of the ΔgcvPAB mutant observed in macrophages is not related to common pathogen defense strategies of macrophages.

Figure 2.

In vivo virulence of the L. monocytogenes ΔgcvPAB mutant

In order to confirm the virulence attenuation of the ΔgcvPAB mutant in vivo, C57BL/6J mice were infected intravenously (i.v.) with the EGD-e or ΔgcvPAB strains. At day 3 post infection (p.i.), we determined the CFUs within different organs (Fig. 2G). Here, we detected a trend of reduced CFU numbers in the spleen, liver and brain from mice that were infected with the ΔgcvPAB mutant, but this did not reach statistical significance. Importantly, however, mice infected with the EGD-e strain showed a more severe body weight loss compared to mice infected with the ΔgcvPAB mutant (Fig. 2H). In line with this, the more pronounced enlargement of the spleens observed in mice infected with EGD-e compared to those infected with the ΔgcvPAB mutant indicated a reduced pathogenicity of the ΔgcvPAB mutant strain (Fig. 2I). Together, our data indicate a moderately ameliorated disease progression in mice infected with the ΔgcvPAB mutant compared to mice infected with the EGD-e reference strain.

Growth defect of a L. monocytogenes ΔgcvPAB mutant in synthetic medium

To further investigate the phenotype of the ΔgcvPAB mutant, we aimed at the identification of extracellular growth conditions that would mimic the intracellular growth defect. In the course of this search we noticed that the ΔgcvPAB mutant had a remarkable growth defect in Listeria synthetic medium (LSM). LSM broth is a chemically defined medium containing all components required for growth at a defined concentration (Whiteley et al., 2017). When cultivated in LSM broth at 37°C, growth of the ΔgcvPAB mutant was strongly retarded, which was in stark in contrast to BHI broth, where no growth defect was apparent (Fig. 3A). This growth defect was complemented as strain LMS311 carrying an IPTG-inducible gcvPAB copy grew as slow as the parental mutant in the absence and as fast as the wild type in the presence of IPTG (Fig. 3B). This demonstrates that the decarboxylase component of the glycine cleavage system is also required for normal growth in synthetic LSM medium.

Figure 3.

High glycine concentrations are toxic for the ΔgcvPAB mutant

Breakdown of glycine in the GCS generates 1C-THF, which serves as an important one carbon unit donor in various biosynthesis pathways (Fig. 1A). If glycine cannot be catabolized by the GCS, it might be re-routed towards serine formation via serine hydroxymethyl transferase GlyA, even though this would even further deplete the cell for 1C-THF. We therefore considered the possibility that glycine might become toxic in the absence of the GCS as observed in a GCS mutant of the cyanobacterium Synechocystis (Eisenhut et al., 2007) due to depletion of 1C-THF pool. To test this, we determined the growth of the wild type and the ΔgcvPAB mutant in the presence of varying glycine concentrations. As can be seen in Fig. 4A, the wild type was able to grow without glycine and a tenfold increase of the glycine concentration also had no effect. Apparently, glycine can be generated, presumably from serine via GlyA, when it is not supplied externally, and does not become toxic in the presence of a functional GCS. In contrast, the ΔgcvPAB mutant exhibited delayed growth in LSM with standard glycine concentration, but growth was accelerated when the glycine concentration was halved and even reached wild type level, when it was further reduced (Fig. 4B). In the complete absence of glycine, the growth of the ΔgcvPAB mutant was almost unaffected, whereas a tenfold increase in glycine concentration significantly impaired growth (Fig. 4B). In contrast, alterations in the serine concentrations had no effect on growth of either strain (Fig. S2). This demonstrates that glycine or a metabolite of glycine becomes toxic in the absence of a functional GCS. That the ΔgcvPAB mutant can even grow without glycine also reinforces the idea that glycine must be made from serine through GlyA, since glycine formation through the GCS running in a reverse reductive mode would not be possible in the absence of the GcvP component (Yishai et al., 2018).

Figure 4.

Mutations suppressing glycine sensitivity of the ΔgcvPAB mutant

We next exploited glycine sensitivity to screen for mutants suppressing the ΔgcvPAB phenotype. For this, the ΔgcvPAB mutant was cultivated in LSM medium containing 100-fold the amount of glycine as present in the standard recipe. Cells were grown for 24 hours and then plated on LSM plates containing 100-fold the standard amount of glycine, a condition under which the ΔgcvPAB mutant would not grow. Colonies that could grow on these plates were isolated and their genomes were sequenced. Glycine insensitive suppressors carried mutations in the codY (lmo1280), fhs (lmo1877), folK (lmo0226) and glyA (lmo2539) genes. Suppressor LMSF3 carries a G236E substitution in the codY gene encoding the transcriptional repressor of the CodY regulon. LMSF8 had a G82R exchange in folK coding for 7,8-dihydro-6-hydroxymethylpterin pyrophosphokinase that acts in folate biosynthesis. LMSF10 had a G62S substitution in glyA (encoding the gene for serine hydroxymethyltransferase for serine/glycine interconversion) in addition to the codY G236E exchange that is also present in LMSF3. However, the most remarkable suppressor mutation was found in strain LMSF15, in which the N-and C-terminal parts of the fhs pseudogenes that are separated by a premature stop codon and thus inactivated in EGD-e are reunited by the introduction of four nucleotides at the end of the N-terminal fhs pseudogene lmo1877. Here, introduction of a quadruplet (GTGG) restores the complete fhs reading frame as it is found in strain 10403S but with one extra valine inserted at the fusion site (Fig. 5A). All four suppressors grew as the wild type in BHI and LSM broth without glycine. However, all suppressors grew better than the parental ΔgcvPAB strain in LSM broth containing 100-fold the glycine concentration of the standard recipe. None of the suppressors restored wild type-like growth, but restoration of full length fhs (named fhs⁺ here) had the strongest effect (Fig. 5B).

Figure 5.

Specific suppression of the ΔgcvPAB virulence defects by Fhs restoration

In order to determine to what degree suppression of glycine toxicity also repairs the intracellular growth defect in the four suppressor mutants, their growth inside J774 cells was measured. Growth of the ΔgcvPAB suppressors with mutations in codY, folK and glyA was as retarded as observed in the parental ΔgcvPAB mutant. However, restoration of full length fhs suppressed this defect and the ΔgcvPAB suppressor with the restored fhs gene grew as fast the wild type (Fig. 5C). To further confirm these observations, we generated a plasmid that allows restoration of full length fhs in strain EGD-e and its descendants. Using this plasmid, full length fhs was generated in the ΔgcvPAB mutant and the growth of the resulting strain in mouse macrophages was determined. As can be seen in Fig. 5D, fhs repair in the ΔgcvPAB background restored wild-type like intracellular replication. Moreover, the ΔgcvPAB mutant with the repaired fhs gene also formed plaques as the wild type (Fig. 5E). The fhs gene was also repaired in EGD-e, but this did not further enhance plaquing efficiency or intracellular replication (Fig. 5D-E).

Three pathways feed 1C-THF generation to support growth and virulence

In order to determine the contribution of the three 1C-THF generating pathways, i. e. the GCS, GlyA and the Fhs/FolD pathway, to growth of L. monocytogenes, we sought to generate mutants lacking all three pathways simultaneously. For this, a ΔglyA mutant was constructed in EGD-e first. The ΔglyA mutant could not grow in LSM broth not containing glycine. However, normal growth was observed, when LSM was supplemented with high glycine concentrations (Fig. S3A-B). Thus, GlyA is needed for glycine generation from serine when glycine is not supplied externally. In contrast, the ΔglyA mutant could grow normally in the absence of serine (Fig. S3C-D), which showed that serine can be generated in the absence of GlyA, presumably from pyruvate via serine dehydratase.

We next tried to generate a ΔglyA ΔgcvPAB double deletion in the fhs⁻ EGD-e background, but all attempts failed, suggesting that at least one pathway for 1C-THF generation must be present for viability. Because of this observation, deletion of glyA and gcvPAB was tried in the EGD-e fhs⁺ background, which turned out to be possible. The resulting fhs⁺ ΔglyA ΔgcvPAB strain exhibited wildtype growth in BHI medium, but was unable to grow in synthetic medium (Fig. 6A-B). Thus, Fhs activity is sufficient to maintain growth in complex medium but not in synthetic medium. Following this, an fhs⁻ ΔglyA igcvPAB mutant was generated in the EGD-e background. This strain lacks GlyA and Fhs activity and GCS activity is IPTG-dependent. This mutant could barely grow without IPTG even in BHI broth, while normal growth was observed with IPTG (Fig. 6A). Thus, at least one pathway for 1C-THF generation is essential for growth.

Figure 6.

Next, the individual contribution of the three 1C-THF forming pathways to intracellular growth in macrophages was measured. Of the three mutants that still possessed only one of the three biosynthetic pathways, normal intracellular proliferation was observed only in mutants that still possessed either the GCS or the Fhs/FolD pathway (Fig. 6C).

It has been shown that mutants with 1C-THF biosynthesis defects are auxotrophic for purines (Feng et al., 2023). We therefore tested whether addition of adenine could restore growth of ΔgcvPAB, mutant. As can be seen in Fig. 6D, addition of excess adenine restored normal growth of the ΔgcvPAB mutant, demonstrating that reduced 1C-THF biosynthesis that occurs in the absence of a functional GCS causes partial purine auxotrophy.

Discussion

Here we have elucidated the cause of the virulence defect of the L. monocytogenes ΔgcvPAB mutant, which shows delayed intracellular growth in the cytoplasm of macrophages and fibroblasts, the latter also explaining the spreading defect. Mice infected with this mutant lose less weight than the wildtype even though statistically significant effects on bacterial replication were not observed. Due to the lack of the two glycine decarboxylase subunits GcvPA and GcvPB, the glycine cleavage system (GCS) is not functional in the ΔgcvPAB mutant. This system feeds the 1C-THF pool and a shortage of one carbon units needed for several anabolic reactions is the reason for the attenuated phenotype of the ΔgcvPAB strain. The observation that reactivation of the naturally inactive fhs gene reverses the phenotype of the ΔgcvPAB mutant is the most important piece of evidence for this conclusion. The fhs gene is split into two fragments by a premature stop codon in L. monocytogenes EGD-e. Formate tetrahydrofolate ligases are proteins with three domains. Their active center is located in the larger first domain (domain A), while domains B and C are probably used for oligomerisation (Radfar et al., 2000, Kim et al., 2020). The premature stop codon in EGD-e fhs disconnects the entire third domain, most likely inactivating the protein. Either the GCS or serine hydroxymethyltransferase activity provided by GlyA is needed for viability in the fhs⁰ EGD-e background, whereas gcvPAB and glyA together can only be deleted in a strain with a restored full-length fhs gene. The reactions mediated by these three pathways all replenish the 1C-THF pool and therefore we conclude that fhs reactivation compensates 1C-THF depletion in the ΔgcvPAB mutant. This effect is particularly evident in synthetic medium, where the growth defect of the ΔgcvPAB mutant could also be reversed by adenine supplementation, and during intracellular growth. Both observations together point towards a limited availability of folates or folate depending metabolites such as adenine in the host cell cytoplasm.

That glycine is toxic for the ΔgcvPAB mutant is another argument for the 1C-THF depletion hypothesis: 1C-THF cannot be generated from glycine in the ΔgcvPAB mutant and 1C-THF formation by GlyA is prevented in the presence of excess glycine at the same time. This is because the GlyA-mediated reaction is reversible (Schirch & Szebenyi, 2005) and therefore shifted towards the 1C-THF consuming formation of serine (Fig. 1A). The reactivation of Fhs partially neutralizes the toxic effect of glycine, as this counteracts the 1C-THF depletion resulting from increased serine biosynthesis. Limitations in 1C-THF availability also explains the contradiction between the reported essentiality of the folD gene in EGD-e (Fischer et al., 2022) and the successful folD deletion in strain 10403S (Feng et al., 2023): N10-formyl-THF, a 1C-THF species needed for purine and N-formylmethionine biosynthesis can only be generated by Fhs or FolD and therefore, a fhs folD double mutant has pronounced growth defects in laboratory media and during infection (Feng et al., 2023). That is why transposon insertion mutants in the folD gene were likely counter-selected in our recent Tn-Seq study that was performed in the fhs⁰ EGD-e background (Fischer et al., 2022).

Two enzymatic steps in THF biosynthesis ahead of Fhs are inhibited by cotrimoxazole (Fig. 1A), an antibiotic that is recommended for the treatment of listeriosis (Karsaliakos & Mylonakis, 2023). Our experiments revealed that collective inactivation of the Fhs/FolD pathway, the GCS and GlyA strongly impaired growth, suggesting the absence of other 1C-THF generating pathways in L. monocytogenes. Therefore, it would be interesting to test whether inhibitors of any of these three pathways such as the pyrazolopyran compounds inhibiting serine hydroxymethyltransferase (Makino et al., 2022) would act synergistically with cotrimoxazole.

A certain hierarchy of 1C-THF-generating pathways can also be derived from our data. While each of the three pathways was sufficient to maintain growth in BHI medium (Fig. 6A), the GCS and GlyA had a greater impact on growth in synthetic medium than Fhs/FolD (Fig. 6B). In contrast, the GCS or Fhs/FolD were each sufficient to maintain growth in macrophages. Thus, the GCS is important for 1C-THF formation under all tested conditions, whereas GlyA and the Fhs/FolD are required under specific conditions only and are therefore of secondary importance.

Several genes frequently inactivated by premature stop codons are known in reference strains and isolates of L. monocytogenes, including the internalin gene inlA or the gadR acid resistance regulator gene (Nightingale et al., 2005, Wu et al., 2023). As far as we know, the fhs gene is the first example of such a cleaved and inactive pseudogene that can be reactivated by suppressor mutations if the selection conditions favor its restoration. This proves that the inactivation of genes by premature stop codons is not an evolutionary dead end, but can be a reversible regulatory event of transient nature. It would be important to find out whether similar effects can also be observed on inlA genes inactivated by premature stop codons, because this would have important implications on the risk assessment of strains with such mutations.

Materials and methods

Bacterial strains and growth conditions

All L. monocytogenes strains are listed in Table 1. Strains were generally cultivated in BHI broth or on BHI agar plates. LSM broth and LSM agar plates were used for cultivation under chemically defined conditions (Whiteley et al., 2017). Antibiotics and supplements were added at the following concentrations: erythromycin (5 µg/ml), kanamycin (50 µg/ml), X-Gal (100 µg/ml) and IPTG (1 mM). Escherichia coli TOP10 was used as standard cloning host (Sambrook et al., 1989).

Table 1.

General methods, manipulation of DNA and oligonucleotide primers

Standard methods were used for transformation of E. coli and isolation of plasmid DNA (Sambrook et al., 1989). Transformation of L. monocytogenes was carried out as described by others (Monk et al., 2008). Restriction and ligation of DNA was performed according to the manufactureŕs instructions. The oligonucleotides used in this study are listed in Table 2.

Table 2.

Construction of plasmids and L. monocytogenes strains

All plasmids are listed in Table 1. Plasmid pSF2 was generated for deletion of glyA. To this end, fragments up-and downstream of glyA were amplified by PCR using SF1/SF2 and SF3/SF4 as the primers, respectively. The resulting fragments were fused together by splicing by overlapping extension (SOE) PCR and then inserted into pMAD using NcoI/SalI as the restriction enzymes.

Plasmid pSF4 was constructed to transfer the fhs⁺ mutation of strain LMSF15 to other strain backgrounds. For this, the fhs⁺ region of strain LMSF15 was amplified using SF7/SF8 as the primers and the obtained fragment was inserted into pMAD using NcoI/SalI. The plasmid insertion/excision protocol of Arnaud et al. (Arnaud et al., 2004) was used for gene deletions and allelic exchange. Successful deletions and allelic exchanges were confirmed by PCR and genome sequencing (see below).

Plasmid pSH572 was generated for complementation of the ΔgvcPAB mutant. To this end, the gcvPAB operon was amplified from EGD-e chromosomal DNA in a PCR using the oligonucleotides TE263/TE276. The resulting fragment was cut using NcoI/SalI and ligated to pIMK3 digested with the same enzymes. Plasmid pSH572 was transformed into strain LMS305 and kanamycin resistant clones were selected. Plasmid insertion at the attB tRNA^Arg site was confirmed by PCR.

pJEBAN6 for expression of DsRed-Express was electroporated in various recipient strains and erythromycin resistant clones were selected. Plasmid acquisition and concomitant DsRed-Express production was confirmed by fluorescence microscopy.

Genome sequencing

Chromosomal DNA of bacterial strains was isolated using the GenElute Bacterial Genomic DNA Kit (Sigma-Aldrich). Libraries were generated from 1 ng DNA using the Nextera XT DNA Library Prep Kit (Illumina). Sequencing was carried out on a NextSeq sequencer in paired-end sequencing mode with 2 x 150 bp read length. Reads were mapped to the L. monocytogenes EGD-e genome (NC_003210.1) [58] as the reference in Geneious (Biomatters Ltd.) and the alignment was analyzed using the Geneious SNP finder tool. Genome sequences of the ΔgcvPAB mutant and glycine resistant ΔgcvPAB suppressors were deposited at the European Nucleotide Archive (ENA, https://www.ebi.ac.uk/) under project accession number PRJEB94141.

Hemolysis assays

The CAMP test was used as a qualitative assay to record hemolysis (Fernandez-Garayzabal et al., 1996). For this, Staphylococcus aureus SG511 was streaked on Mueller-Hinton agar plates containing 5% sheep blood right-angled to the L. monocytogenes strains to be tested. Hemolysis became apparent after overnight incubation at 37°C.

For semi-quantitative determination of hemolysis activity, bacterial cultures were cultivated for 5 h at 37°C. Culture supernatants were prepared from these cultures by centrifugation of a 1 ml culture volume and a two-fold dilution series of the culture supernatant was prepared in phosphate-buffered saline (PBS) containing 6 mM cysteine. Triplicates of each dilution (100 µl) were pipetted into a 96-well microtiter plate and 1 µl of a 1% (v/v) human erythrocyte concentrate was added to each well. The microtiter plate was incubated for 30 min at 37°C in a static incubator. Hemolytic activity is expressed as the dilution factor of the last dilution showing complete hemolysis.

Assays to determine hydrogen peroxide and lysozyme sensitivity

To assess the sensitivity of L. monocytogenes to H2O2, strains were inoculated in BHI broth containing increasing concentrations of H2O2 (0.2-100 mM) in a microtiter plate at 37°C. The plate was incubated overnight at 37 °C in a static incubator. The minimum inhibitory concentration of hydrogen peroxide was determined the next morning by visual inspection.

For analysis of lysozyme sensitivity, L. monocytogenes strains were grown in BHI broth at 37°C until mid-logarithmic growth phase (OD600∼0.8). Cells were collected by centrifugation and resuspended in 50 mM Tris/HCl pH8.0 to an optical density of OD600=0.6. Lysozyme was added to a final concentration of 2.5 µg/ml and the cells were shaken at 37°C. Lysis was followed by measuring the decrease in optical density (λ=600 nm) every 15 min in a spectrophotometer.

Cell culture infection experiments

Infection of J774A.1 mouse macrophages (ATCC® TIB-67^TM) and HepG2 human hepatocytes (ATCC® HB-8065^TM) with L. monocytogenes strains was performed as described earlier (Halbedel et al., 2019). Briefly, 10⁵ cells were seeded into the wells of a 24 multi well plate and cultivated in DMEM + 10% fetal calf serum (FCS) overnight before they were infected with 2 x 10⁵ bacteria. The bacteria were allowed to invade the cells during an incubation step at 37°C for 1 h. Extracellular bacteria were first washed off with PBS and the remaining extracellular bacteria were killed by gentamicin addition. Eukaryotic cells were lysed six hours post infection using ice-cold PBS containing 0.1% Triton X-100, serial dilutions were plated on BHI agar plates and incubated overnight at 37°C for quantification. Infection of 3T3-L1 mouse embryo fibroblasts (ATCC® CL-173T^M) was performed in the same way. Analysis of cell-to-cell spread using 3T3-L1 mouse embryo fibroblasts by plaque formation was carried out as described earlier (Halbedel et al., 2014). Here, 5 x 10⁵ fibroblast cells were seeded into the wells of a six well plate and cultivated in DMEM + 10% newborn calf serum (NCS). After three days of incubation, cells were infected with an inoculum of 2, 4 or 10 μl each containing 1 x 10⁶ bacteria. Plaques were visualised three days post infection using neutral red staining. Plaque areas were quantified using ImageJ.

Microscopy of infected 3T3 cells

5 x 10⁴ 3T3 mouse fibroblasts were seeded into the wells of a 24-well tissue culture plate containing coverslips and DMEM + 10% NCS as the culture medium and incubated in a 5% CO2 atmosphere at 37°C. Cells were infected as outlined above. 6 h after infection, cells were washed with PBS and then fixed with ice-cold methanol. Methanol was replaced by 500 μl PBS before the coverslips were removed from the wells and allowed to dry completely. Finally, a drop of ProLong Gold antifade reagent with DAPI (Invitrogen) was dropped onto a microscope slide and the coverslip was placed on top. Samples were dried overnight and examined using a Nikon Eclipse Ti fluorescent microscope.

In vivo infection in mice

80 male C57BL/6J mice (Janvier) were maintained in the animal facility at the university hospital of the OVGU Magdeburg under conditions that ensured precise temperature and humidity regulation, with a 12-hour day/night cycle. Mice at an age ranging from 10 to 18 weeks were infected with 1-20 x 10⁴ CFU in PBS via the tail vein. Following infection, the mice were monitored daily for body weight loss and other disease symptoms for the entire duration of the experiment. After euthanasia of mice by carbon dioxide (CO2) inhalation, the heart was punctured, the heart blood was taken and the heart was perfused with 10 mL PBS. All animal experiments were conducted according to the institutional guidelines, and the study was approved by local government agencies (Landesverwaltungsamt Sachsen-Anhalt; AZ 42502-2-1603 UniMD).

Determination of CFU in spleen, brain and liver

To determine the CFUs in the spleen, brain and liver, the organs were homogenised in 0.2% IGEPAL CA-630 (Sigma-Aldrich) lysis buffer. The livers were suspended in 2 ml IGEPAL buffer, while the brains and spleens were suspended in 1 ml IGEPAL buffer. The organs were homogenised at full speed (30000 rpm) using a POLYTRON® PT 3100 homogeniser (KINEMATICA AG). To prevent artefacts between samples, the homogeniser was washed with EtOH for 10 seconds three times and with PBS twice between every sample. Serial dilutions were plated onto BHI agar plates and incubated at 37 °C for 24 hours to quantify the colonies.

Data availability

Genome sequences of the ΔgcvPAB mutant and glycine resistant ΔgcvPAB suppressors were deposited at the European Nucleotide Archive (ENA, https://www.ebi.ac.uk/) under project accession number PRJEB94141.

Acknowledgements

We thank Birgitt Hahn for excellent technical assistance. We thank Nouria Jantz-Naeem, Anna Krone, Bianca Thoma, Hildburg Volkmann, Anne Hoffmann and Anja Sammt from the Kahlfuß lab for helping processing the samples following the in vivo infection model. This work was funded by the DFG (grants HA6830/2-1 and HA6830/5-1 to S. H.).

Additional files

Supplementary Figures S1-S3

Additional information

Funding

Deutsche Forschungsgemeinschaft (HA6830/2-1)

Deutsche Forschungsgemeinschaft (HA6830/5-1)