The bacterial strains and plasmids used in this work are listed in Table 1. Listeria monocytogenes Scott A was used as the wild-type (WT) strain and acquired from the International Life Sciences Institute (ILSI) North America (Fugett et al., 2006). Strain asnB::Himar1 was identified as a sensitive mutant from a random Scott A transposon mutant collection in a screen with the natural antimicrobial t-CIN (Rogiers et al., 2017). Escherichia coli DH5α (Grant et al., 1990) was employed as host for cloning constructs, and E. coli S17-1 λpir (Simon et al., 1983) as the donor strain for conjugational plasmid transfer. Listeria monocytogenes strains were grown at 30 or 37°C in Brain Heart Infusion (BHI, Oxoid, Hampshire, United Kingdom) medium. Escherichia coli strains were grown in Luria-Bertani (LB, 10-g/L tryptone, 5-g/L yeast extract, and 5-g/L NaCl) medium at 37°C. Growth media were supplemented with 50-μg/ml erythromycin (Em; Acros Organics, Fair Lawn, NJ, United States) or 50-μg/ml kanamycin (Km; AppliChem GmbH, Darmstadt, Germany) when appropriate.

For genetic complementation of the asnB::Himar1 mutant, the asnB gene was amplified using primers asnB_NcoI and asnB_SalI (Table 2) and cloned in pIMK2 (Monk et al., 2008) digested with NcoI and SalI. The construct was verified with Sanger sequencing and then conjugated from E. coli S17-1 λpir into the asnB::Himar1 mutant. Successful integration was confirmed via PCR with primers asnB_NcoI and NC16 (which anneals near, and points toward, the plasmid integration site) and Sanger sequencing with primers pIMK_FW and pIMK_REV, which point toward the asnB gene from both sides of the pIMK2 cloning site. The complemented strain was designated asnB/pIMK2-asnB. Control strains were constructed by integration of the empty pIMK2 plasmid into the WT and asnB::Himar1 strains and were designated as WT/pIMK2 and asnB/pIMK2, respectively.

Growth curves of strains WT/pIMK2, asnB/pIMK2, and asnB/pIMK2-asnB were established by turbidity measurement (OD₆₃₀) of cultures growing in an automated temperature-controlled microplate reader (Multiskan Ascent®, Thermo Fisher Scientific). Inocula were prepared by adjusting overnight 4-ml BHI broth cultures to the same turbidity (OD₆₀₀≈2) using an Ultrospec™ 10 Cell Density Meter (Biochrom, Cambridge, United Kingdom), and then diluting the suspensions 1,000-fold in BHI or BHI with 3- or 4-mM t-CIN (Acros Organics). Subsequently, 200-μl aliquots were transferred into a 96-well microplate, which was then covered with a transparent adhesive foil (Greiner Bio-One, Frickenhausen, Germany) and incubated at 30 or 37°C in the microplate reader. Every 15min, the plates were shaken at 960rpm and OD630 was measured. The Excel add-in package DMFit (Quadram Institute Bioscience, Norwich, United Kingdom) was used to determine the maximum growth rate (μ_max), the lag phase time (λ), and the maximal cell density (OD_max) value at stationary phase based on the Baranyi and Roberts microbial growth model (Baranyi and Roberts, 1994).

Lysozyme sensitivity was evaluated by a disk diffusion assay and broth growth inhibition assay as previously described (Rae et al., 2011) with minor modifications. For the disk diffusion assay, 5μl of an overnight culture was evenly spread on a BHI agar plate, and a 6-mm sterile Whatman paper disk was placed in the center and impregnated with 10μl of a 10-mg/ml lysozyme (Sigma Aldrich, Saint Louis, MO, United States) solution in 10-mM potassium phosphate buffer (PPB) pH 7.0. The size of the formed inhibition halo was measured after incubation at 30°C for 24h. For the broth inhibition assay, overnight cultures were diluted 1,000-fold in BHI broth supplemented with 1-mg/ml lysozyme and 200-μl aliquots of the cell suspension were loaded into 96-well microplates. The OD₆₃₀ was monitored with 15-min intervals in the microplate reader at 30°C. Additionally, a lysozyme lysis assay was conducted using cells from overnight cultures that were washed and resuspended in 10-mM PPB pH 7.0 to an OD₄₅₀ of 0.6–0.8. Two hundred and seventy microliter of the suspension were then dispensed into a 96-well microplate and 30μl of a 1,000-fold diluted lysozyme stock (1mg/ml or water for control) was added. The OD₄₅₀ was recorded with 30-min intervals in the microplate reader at 25°C.

Overnight stationary cultures were diluted 100-fold in BHI medium and grown to exponential phase (2–3h) at 30°C with shaking. Crystal violet flagellar staining was performed as described (Upadhyay et al., 2012), and cells were observed with a Leica SFL4000 microscope. Swimming motility was evaluated by picking colonies from a BHI plate and stab-inoculating them into BHI soft agar (0.2%) with a toothpick (Gründling et al., 2004). Plates were incubated at 30°C for 24h, and motility was assessed by measuring the migration distance of bacteria from the center to the periphery of the colony.

Cells were harvested from BHI broth at OD₆₀₀≈1, washed once with 0.1-mM PPB pH 7.0, and diluted appropriately. Fifty microliter of cell suspension was applied to a coverslip mounted to an AI stub by carbon adhesive disks and dried in the oven at 37°C. HR SEM was performed on a Nova NanoSEM450 (FEI) scanning electron microscope.

To measure the cell dimensions, 1μl of an appropriately diluted exponential culture (OD₆₀₀≈1) was applied to 2% agarose pads deposited on a microscopy slide. A Gene Frame (Thermo Fisher Scientific) was used to mount a cover glass on the microscopy slide. Observations were performed with an Eclipse Ti-E inverted microscope (Nikon Instruments Europe BV, Netherlands) in phase contrast modus at a total magnification of 100x and images were acquired using NIS-elements software (Nikon). Image analysis (cell width and cell length) was conducted with the MicrobeTracker software (Sliusarenko et al., 2011), with manual curation to remove false segmentation and tracking.

The amino acid sequences of AsnB from L. monocytogenes Scott A and its homologs from other bacteria were acquired from the National Center for Biotechnology Information database (NCBI) and listed in Supplementary Table S1. Amino acid sequence alignment was conducted with MUSCLE (Edgar, 2004), and a phylogenic tree was constructed with CLC Genomic Workbench (QIAGEN, Hilden, Germany) using the Neighbor-joining (NJ) method with 100 bootstrap replicates.

PG was purified as described (Courtin et al., 2006) with minor modifications. Overnight cultures were diluted 100-fold in 0.5-L BHI broth and grown to OD₆₀₀≈1 at 30°C. After cooling in ice water (20–30min), cells were collected by centrifugation (5,000rpm, 10min, 4°C), resuspended in 40-ml cold H₂O, boiled (10min), cooled again, and centrifuged. After suspending the cell pellet in 1-ml H₂O, 1-ml SDS solution (10% SDS, 100-mM Tris-HCl pH 7.0) at 60°C was added and the suspension was boiled (30min) and centrifuged (10min, 14,000rpm, RT). The pellet was resuspended in 2-ml lysis solution (4% SDS, 50-mM Tris-HCl pH 7.0), boiled (15min), and washed 6 times with H₂O preheated to 60°C. Afterward, the pellets were treated with 2mg/ml pronase from Streptomyces griseus (Roche, Basel, Switzerland) in 50-mM Tris-HCl pH 7.0 for 1.5h at 60°C, and with 10-μg/ml DNase (Thermo Fisher Scientific), 50-μg/ml RNase (Thermo Fisher Scientific), and 50-μg/ml lipase from Aspergillus niger (Sigma Aldrich) in 20-mM Tris-HCl pH 7.0, 1-mM MgCl₂, and 0.05% sodium azide for 4h at 37°C. Then, the suspensions were washed with H₂O and treated with 200-μg/ml trypsin (Sigma Aldrich) in 20-mM Tris-HCl pH 8.0 at 37°C with agitation overnight. Finally, after inactivating trypsin by boiling for 3min, the suspensions were incubated with 48% hydrofluoric acid (Merck, Kenilworth, NJ, United States) overnight at 4°C. After centrifugation (10min, 14,000rpm, RT), the pellet was washed twice with 250-mM Tris-HCl (pH 7.0) and four times with H₂O to reach a pH close to 5. The extracted PG was eventually lyophilized and resuspended in H₂O to 20mg/ml.

For structural analysis, 50μl of purified PG was digested by adding 50-μl 25mM NaHPO₄ pH 5.5 and 2-μl 10mg/ml mutanolysin from Streptomyces globisporus (Sigma Aldrich) and incubating overnight at 37°C with shaking. The resulting soluble muropeptides were reduced with sodium borohydride and separated by reverse phase-ultra high-pressure liquid chromatography (RP-UHPLC) with a 1,290 chromatography system (Agilent Technologies, Santa Clara, CA, United States) and a Zorbax Eclipse Plus C18 RRHD column (100 by 2.1mm; particle size, and 1.8μm; Agilent Technologies) at 50°C using ammonium phosphate buffer and methanol linear gradient as described previously (Courtin et al., 2006). One microliter of collected muropeptides was then spotted directly on the matrix-assisted laser desorption ionization (MALDI) target and thoroughly mixed with 1μl of α-cyano-4-hydroxycinnaminic acid solution (5mg/ml in 50% acetonitrile containing 0.1% trifluoroacetic acid). Muropeptides were analyzed by MALDI – time of flight mass spectrometry (MALDI-TOF MS) with an UltrafleXtreme instrument (Bruker Daltonics, Billerica, MA, United States; located at Université Paris-Saclay, CEA, INRAE, Médicaments et Technologies pour la Santé (MTS), MetaboHUB, Gif-sur-Yvette, France). MS spectra were acquired at 2kHz laser repetition rate in the positive reflector ion mode, with a 20-kV acceleration voltage and an extraction delay of 130ns. The spectra were obtained by accumulating 1,000–5,000 shots (depending on the samples) over the 500–5,000m/z range. MS/MS spectra were acquired in LIFT mode, at 1kHz laser repetition rate applying 7.5kV for initial acceleration of ions and 19kV for reacceleration of fragments in the LIFT cell.

Human JEG-3 placental cells (ATCC HTB-36) and Caco-2 intestinal cells (ATCC HTB-37) were grown following ATCC recommendations at 37°C in a humidified 10% CO₂ atmosphere. The invasion of L. monocytogenes strains was assessed by the gentamicin assay, as described (Bierne et al., 2002) with some modifications. To prepare bacterial inoculums, bacterial cultures were grown to early exponential or stationary phase at 37°C, washed with 10-mM phosphate-buffered saline (PBS) pH 7.4, and diluted in Eagle’s Minimum Essential Medium (MEM). JEG-3 and Caco-2 cells at 80% confluency were infected with bacterial inoculums at a multiplicity of infection (MOI) of about 0.05–0.1 bacteria per cell and centrifuged at 300 x g for 2min to synchronize bacterial entry. For JEG-3 cells, after 1-h incubation, cells were washed with MEM and incubated with complete culture media containing 25-μg/ml gentamicin for 3h to kill the extracellular bacteria. For Caco-2 cells, after 0.5-h incubation, cells were washed with MEM and incubated with 25-μg/ml gentamicin for 0.5h to kill the extracellular bacteria while minimizing the intracellular replication of the infected bacteria. Subsequently, infected cells were washed twice in MEM and lysed in cold distilled water. The number of bacteria in bacterial inoculums and cell lysates was determined by serial dilutions plated on BHI agar and counting colony-forming units (cfu) after 48-h incubation at 37°C. The relative entry efficiency was expressed as the ratio of cfu recovered after cell lysis to inoculated cfu.

Immunofluorescent staining was performed as described (Bierne et al., 2002). Briefly, overnight bacterial cultures were washed twice with PBS, fixed on a coverslip with 4% paraformaldehyde in PBS, and stained for 1h with a mixture of mouse monoclonal InlA antibodies L7.7 and G6.1 (Bierne et al., 2002) and a rabbit polyclonal L. monocytogenes antibody in 2% bovine serum albumin, at 1:500 dilution. Coverslips were then washed several times with PBS and incubated with Alexa 488-labeled goat anti-rabbit (1:400 dilution; Molecular Probes, Eugene, OR, United States) and Cy3-labeled goat anti-mouse (Jackson ImmunoResearch, West Grove, PA, United States) secondary antibodies for 1h, followed by mounting with 10μl of Fluoromount G (Interchim, Montluçon, France).

Data for the growth assay and intracellular infection assay are presented as means±SD from three independent repetitions. Differences of growth parameters were statistically analyzed by the Tukey’s honestly significant difference (Tukey’s HSD) test using GraphPad PRISM 7.0 (GraphPad, San Diego, CA, United States). Bacterial cell width and length are means±SD of 300 cells and the significance of mean differences was calculated by student’s t-test (two tailed). For the lysozyme lysis assay, linear regression was applied to fit straight lines through the data points and the slopes are presented as means±SD from five independent repetitions. The significance of mean differences was calculated by ANOVA (two-way ANOVA) using GraphPad Prism software. Values of p<0.05 were considered statistically significant.

Several mutants with increased sensitivity to t-CIN, an antimicrobial from cinnamon bark essential oil, were previously isolated from a random Himar1 transposon insertion library of L. monocytogenes Scott A (Rogiers et al., 2017). One mutant had the transposon inserted 22bp upstream of the start codon of asnB (NCBI accession no. and locus tag are CM001159 and LMOSA_25850), which is predicted to encode an asparagine synthetase, and this mutant was designated 5’asnB::Himar1 (Figure 1A). The transcription start site (TSS) of asnB has been reported to be located at −47bp relative to the start codon in L. monocytogenes EGD-e and at −50bp in Listeria innocua (Wurtzel et al., 2012), making it likely that transcription of asnB is disrupted in this mutant. The gene downstream of asnB encodes a putative rRNA methylase and is encoded on the opposite strand. An S-adenosylmethionine synthetase gene is located upstream of asnB and its stop codon is 136bp away from the start codon of asnB. Given their position and orientation, the expression of these flanking genes is unlikely to be strongly affected by the transposon. The genome of the 5’asnB::Himar1 mutant was sequenced and no other mutations were identified. To further study the function of the asnB gene, a genetically complemented mutant strain was constructed by chromosomal integration of a pIMK2 plasmid containing an intact asnB copy under control of a constitutive promotor (strain asnB/pIMK2-asnB), and the WT and asnB mutant strain were equipped with an empty pIMK2 vector at the same chromosomal insertion site. More detailed analysis of the t-CIN sensitivity indicated that the mutant displayed an extended lag phase (56 vs. 23h for the WT strain) but not a reduced exponential growth rate or stationary phase level (Figure 1C). Furthermore, t-CIN tolerance was restored to WT level (20-h lag phase) by genetic complementation (Figure 1C), but not by supplementation of the growth medium with Asn (Figure 1D), suggesting that t-CIN sensitivity of the asnB mutant is not related to an Asn deficiency that could be the result of impaired Asn synthetase activity.

Although the asnB mutant showed normal growth in BHI broth without t-CIN at 30°C, it grew at a lower rate and to a lower stationary phase level at 37°C (Figure 1B). Furthermore, both SEM and optical microscopy observations of exponential and stationary phase cells grown at 30°C revealed that the asnB mutant produced “fat rods” that were shorter and thicker than WT cells, and also the presence of some club-shaped cells (Figure 2). Both the growth at 37°C and cell morphology were largely restored by genetic complementation. The observed defects in growth and cell morphology are reminiscent of the phenotype of an asnB mutant of B. subtilis (Yoshida et al., 1999). Interestingly, B. subtilis has two paralogs of asnB, designated asnH and asnO. However, while the three paralogs could restore Asn prototrophy to an E. coli asnB mutant, none was essential for producing Asn in B. subtilis, and asnB was later shown to mediate amidation of the ε-carboxyl group of mDAP residues in the peptidoglycan peptide stem (Dajkovic et al., 2017). The role of asnB homologs in mDAP amidation has also been demonstrated in other Gram-positive bacteria, including L. plantarum (Bernard et al., 2011) and C. difficile (Ammam et al., 2020). However, unlike these bacteria, which have two or three asnB paralogs, L. monocytogenes contains only one. This unique constellation triggered us to further study the function(s) of asnB in L. monocytogenes, including its possible role in virulence, since such a role had not been previously reported in any pathogen.

In view of the observed cell shape defects of the asnB mutant and the possible role of AsnB in mDAP amidation, we analyzed some additional phenotypes that could be affected by cell wall perturbation. Swimming motility in 0.2% BHI agar at 30°C was completely lost for the asnB mutant, and flagellar staining indicated that this was due to the complete absence of flagella (Figures 3A,B). Motility and flagella production were almost fully restored by genetic complementation. Since flagellar motility was previously shown to play an essential role in L. monocytogenes biofilm formation (Lemon et al., 2007), we subsequently determined the biofilm-forming capacity of the asnB mutant using the crystal violet staining method. The data showed a complete loss of biofilm-forming capacity of the asnB mutant (Figure 3C).

Finally, we investigated the sensitivity of the asnB mutant to lysozyme, since mDAP amidation had been linked to lysozyme resistance in some bacteria (Levefaudes et al., 2015; Dajkovic et al., 2017). Both the disk diffusion assay and the broth growth inhibition assay revealed lysozyme sensitivity of the asnB mutant during growth, while the WT was fully resistant (Figures 4A,B). Furthermore, the asnB mutant was also more sensitive to lysozyme in a lysis assay with non-growing cells suspended in a buffer (Figure 4C). Genetic complementation restored WT, or almost WT, lysozyme tolerance in all these assays.

The aberrant cell shape, absence of flagellation, and lysozyme sensitivity suggested a role in mDAP amidation rather than in Asn biosynthesis for AsnB in L. monocytogenes. To further investigate this possible function, the AsnB amino acid sequence was compared with homologs in other Gram-positive bacteria (Figure 5A). The analysis revealed three major clades, of which one contains no representatives with proven mDAP amidation activity. Among the members of this clade are AsnH and AsnO from B. subtilis, but also homologs from Actinobacteria and AsnB from E. coli, which was included for comparison because it is known to be an Asn-producing enzyme. The other two clades, which are more closely related to each other than to the first clade, contain proteins from Actinobacteria and Firmicutes, respectively, and both contain representatives with demonstrated mDAP amidation activity. With 68% sequence identity, L. monocytogenes AsnB is most closely related to AsnB from B. subtilis (Figure 5B), which was recently shown to mediate PG mDAP amidation (Dajkovic et al., 2017). Another highly similar homolog is AsnB1 of Geobacillus stearothermophilus, whose mDAP in PG is amidated as well (Linnett and Strominger, 1974). In the same clade, AsnB1 of L. plantarum was also demonstrated to mediate mDAP amidation (Bernard et al., 2011). The same holds for AsnB1 from C. difficile (Ammam et al., 2020), but this sequence, together with that of its AsnB2 paralog, forms a small distinct clade.

To provide direct evidence for the involvement of AsnB in mDAP amidation, the PG structure of the WT/pIMK2, asnB/pIMK2, and asnB/pIMK2-asnB strains was analyzed by enzymatic hydrolysis followed by RP-HPLC and mass spectrometry. The muropeptide profiles of WT/pIMK2 and asnB/pIMK2-asnB by RP-HPLC were almost indistinguishable (Figure 6). In contrast, the profile of asnB/pIMK2 was strikingly different with the major peaks shifted toward lower retention times.

MALDI-TOF MS analysis was conducted to identify the muropeptides corresponding to peaks in the chromatograms (Supplementary Table S2). MS analysis indicated that part of GlcNAc residues was deacetylated, and part of MurNAc residues was O-acetylated, which is consistent with previous studies (Boneca et al., 2007; Aubry et al., 2011; Burke et al., 2014). The dimer peaks in the muropeptide profile of asnB/pIMK2 (Figure 6B, peaks 7–13) are relatively smaller than those of WT/pIMK2 (Figure 6A, peaks f to k), indicating a reduced ratio of dimers to monomers, thus suggesting a reduced cross-linking of PG. In addition, the results indicated that most stem peptides in WT/pIMK2 muropeptides bear one amidation. In contrast, no amidation was detected in the asnB/pIMK2 mutant strain. Since amidation can not only occur on the ε-carboxyl group of mDAP, but also on the α-carboxyl groups of D-Glu, tandem mass spectrometry (MS/MS) was performed on the major monomer muropeptide (disaccharide tripeptide), which corresponds to peak b and peak 1 in the chromatogram of WT/pIMK2 and asnB/pIMK2, respectively (Supplementary Figure S1). This confirmed that only mDAP in the PG stem peptide is amidated in WT/pIMK2, whereas it is not in asnB/pIMK2 mutant. Altogether, these results provide convincing evidence that L. monocytogenes AsnB is an amidotransferase that amidates mDAP residues in the PG stem peptide.

Peptidoglycan N-deacetylation and O-acetylation are critical for virulence of L. monocytogenes by mediating escape of the immune response (Boneca et al., 2007; Aubry et al., 2011; Rae et al., 2011). However, the possible role of mDAP amidation in virulence has not yet been studied in L. monocytogenes, nor in any other pathogen, and was therefore addressed here by using a gentamicin cell invasion assay. Firstly, human JEG3 placental cells were infected with stationary phase bacteria for 1h, followed by incubation with gentamicin for an additional 3h, and the invasion efficiency was expressed as the ratio of the number of recovered bacteria to the number of bacteria initially applied. The results highlighted an important defect in cellular invasion of the asnB/pIMK2 mutant, with a reduction in invasion efficiency to less than 2% of the level of the WT strain, while complementation resulted in a restoration of invasion efficiency to about 30% of that of the WT strain (Figure 7A). To determine whether the attenuation of cell invasion caused by asnB inactivation also occurred in another cell type, we next performed invasion assays in human Caco-2 intestinal cells. The assay was modified to test bacteria in either stationary or exponential phase and to shorten the duration of the infection, in order to avoid the intracellular replication of bacteria, which might be impacted by AsnB deficiency. Caco-2 cells were thus exposed to bacteria for only 1h, of which 30min was allocated for entry of bacteria into the cells, and another 30min to the elimination of the extracellular bacteria with gentamicin. Entry efficiency was expressed as the ratio of bacteria recovered after 1h of infection to the initial number of bacteria applied (Figures 7B,C). Compared to the WT strain, the entry efficiency of the asnB/pIMK2 mutant at exponential and stationary phase was 10 and six times lower, respectively (Figures 7B,C). Furthermore, the entry efficiency was fully restored by complementation. Taken together, these results indicate that asnB inactivation impairs L. monocytogenes entry into epithelial cells.

Internalins are a family of leucine-rich repeat proteins that play an important role in the Listeria infection process (Bierne et al., 2007). The majority of internalins is LPXTG proteins covalently bound to the bacterial cell wall by the Sortase A (SrtA) enzyme, the best characterized being InlA, which mediates the adhesion and internalization of L. monocytogenes into epithelial cells (Gaillard et al., 1991; Mengaud et al., 1996). Interestingly, InlA is anchored to the bacterial surface by covalent linkage to mDAP in the peptidoglycan stem peptide (Dhar et al., 2000). Therefore, and because the asnB mutant has a defect in cell entry, we tested whether the asnB mutant is defective in cell wall anchoring of InlA. The presence of InlA on the bacterial surface was analyzed by immunofluorescence using InlA-specific monoclonal antibodies, as previously described for the study of a sortase A mutant (Bierne et al., 2002). In comparison with the WT strain, detection of InlA on the surface of the asnB/pIMK2 mutant cells was considerably reduced (Figure 8). In contrast, InlA was clearly detected on the surface of the complemented strain, indicating that expression of asnB in the mutant restored surface anchoring. Thus, amidation at the ε-carboxyl group of mDAP appears to be required for efficient cross-linking of InlA to mDAP.