The bacterial strains, plasmids and cell lines used in this study are listed in Table 1 . To amplify the CRISPRi backbone PLJR962 (Addgene #115162) and for cloning, E. coli was cultured in Luria-Bertani medium (Merck), supplemented with 25 μg/mL of kanamycin (NZYTech) when appropriate. M. smegmatis strains were cultured in Middlebrook 7H9 broth (BD™ Biosciences) supplemented with 0.2% glycerol (ThermoScientific), 0.5% glucose (Sigma-Aldrich) and 0.05% tyloxapol (Sigma-Aldrich) or 7H10 agar (BD™ Biosciences) supplemented with 0.5% glycerol. When necessary, the medium was supplemented with 25 μg/mL of kanamycin. Bacteria were grown at 37°C with (broth) or without (agar) shaking at 180 rpm. To induce sgRNA and dCas9_Sth1 expression, 100 ng/mL of ATc were added to mycobacterial cultures every 24 h. The J774.A1 cells were cultured in DMEM media (Gibco) supplemented with 10% fetal bovine serum, 1% sodium pyruvate, 1% L-Glutamine, and 1% HEPES buffer. The growth of J774.A1 cells occurred at 37°C, with 5% CO₂. Stocks of amoxicillin (AMX), cefotaxime (CTX), ethambutol (EMB), meropenem (MEM) and isoniazid (INH) (Sigma-Aldrich) were prepared with purified MilliQ water. ATc (Sigma-Aldrich) was prepared at a stock concentration of 10 mg/mL in dimethyl sulfoxide (DMSO) cell culture grade (AppliChem). Potassium clavulanate (CLA) (Sigma-Aldrich) was prepared in phosphate buffer pH 6.0, 0.1 M.

The cloning procedures were executed as previously optimized (Rock et al., 2017; Marques, 2020; Silveiro, 2020; Wong and Rock, 2021). To achieve efficient repression, either the non-template strand or the promoter region of the target genes murT (MSMEG_6276), gatD (MSMEG_6277) and namH (MSMEG_6410) were targeted. Various PAM sequences were identified on the template strand of the target genes, chosen for differential repression regarding PAM strength and target site, and accordingly guided sgRNA design ( Table S1 ). The CRISPRi backbone PLJR962 was amplified in E. coli JM109 and 2 μg of plasmid were digested with BsmBI (ThermoScientific) in the presence of 1 mM DTT (ThermoScientific) for 4 h at 37°C in a PCR block (Applied Biosystems Veriti™). After, the digestion reactions were purified using the QIAquick PCR Purification Kit (Qiagen). To clone the sgRNA targeting sequences into the sgRNA scaffold, two complementary primers were designed ( Table S2 ), synthesized (Eurofins Genomics), annealed, and ligated into PLJR962 using T4 DNA ligase (400 U/µL; NEB) at 22°C for 30 min. To obtain the sgRNA-containing CRISPRi plasmids, E. coli chemically competent cells were transformed with the ligation mixture using a double heat-shock approach, followed by plasmid extraction from the obtained recombinants with the NZYMiniprep kit (NZYTech). The obtainment of the desired sgRNA-containing CRISPRi plasmids was confirmed by Sanger Sequencing. To express the recombinant clones as knockdown mutants in M. smegmatis, competent cells were transformed with 500 ng of recombinant plasmid by electroporation (using 2 mm cuvettes and a Bio-Rad Gene Pulser™ Electroporation System, with pulse settings of 2.5 kV, 1000 Ω resistance and 25 μF capacitance) in the presence of glycerol 10% (Goude and Parish, 2009; Marques, 2020; Silveiro, 2020).

To evaluate differences in cell viability after CRISPRi induction, spotting assays were performed as previously described (Marques, 2020; Silveiro, 2020). Briefly, cultures of M. smegmatis were grown to log phase, normalized to a theoretical OD₆₀₀ of 0.001 and, serially diluted in a 96-well plate. A minimal quantity (5 µL) of each bacterial suspension was plated on solid media with and without ATc and incubated at 37°C for 2 days. At least three biological replicates were performed.

Bacterial cultures were grown to log phase, normalized to a theoretical OD₆₀₀ of 0.001 and, incubated at 37°C with shaking until an OD₆₀₀ of 0.1 was reached (Marques, 2020; Silveiro, 2020; Wong and Rock, 2021). After an incubation period of 6 h with or without ATc, the cultures were harvested by centrifugation at 3000 g for 5 min, ressuspended in 500 µL of RNAprotect Bacteria Reagent (Qiagen) and, stored at -20°C. When convenient, the cells were thawed and pelleted at 5000 g for 5 min at 4°C. Lysis was promoted by the addition of 350 µL of buffer NR (NZYTech) and 3.5 µL of beta-mercaptoethanol (Merck), followed by intense vortexing. The cells were then disrupted in a BeadBug™ (Benchmark Scientific) by 4 cycles at maximum speed for 1 min, interspersed by resting periods of 1 min on ice. After a spin-down, the lysate was applied into the NZYSpin Homogenization column, and the subsequent steps were performed according to the instructions of the NZYTotal RNA Isolation Kit (NZYTech). The RNA was eluted in 40 µL of DEPC water (Invitrogen) at 60°C. Genomic DNA contamination was avoided by performing a rigorous treatment with TurboDNase (Invitrogen). The quantity and purity of the obtained RNA was determined using Nanodrop™ and the absence of significant DNA contamination was confirmed by performing a PCR on the housekeeping gene sigA using NZYTaq DNA polymerase (NZYTech) followed by 1.5% agarose gel electrophoresis.

Purified RNA (100 ng) was used to synthesize cDNA (Singh et al., 2016) by reverse transcription using the NZY First-Strand cDNA Synthesis Kit (NZYTech). Subsequently, the cDNA levels were quantified by qPCR using the NZYSupreme qPCR Green Master Mix (2x), ROX (NZYTech) and a QuantStudio™ 7 Flex Real-Time PCR System (Applied Biosystems). The qRT-PCR primers were designed using Primer 3 ¹ to amplify a specific product between 100-250 bps ( Table S3 ). The PCR reaction (40 cycles) proceeded as follows: 95°C for 10 min, 95°C for 15 s, 60°C for 30 s. The amplification efficiency of the primers was verified to be 80-110%. For each run, at least two technical replicates and a negative control were produced, and the mean C_T was calculated. Data were analyzed using the Comparative C_T method (ΔΔC_T method; the fold change is calculated as 2^-ΔΔCT) with M. smegmatis WT as the calibrator sample and sigA as a reference gene (Choudhary et al., 2015; Rock et al., 2017; Wong and Rock, 2021). Two biological replicates were utilized for the quantification of the target sequences (Singh et al., 2016).

The antibiotic susceptibility determination assays were performed for three beta-lactams (amoxicillin, cefotaxime and meropenem), with and without the beta-lactamase inhibitor clavulanate, and for two antimycobacterial agents (isoniazid and ethambutol), as described previously (Marques, 2020; Silveiro, 2020). The antibiotics were prepared at the desired initial concentration (512 μg/mL for beta-lactams and 128 μg/mL for EMB/INH) while clavulanate was used at a fixed concentration of 2.5 μg/mL. Next, the antibiotics were serially diluted 1:2 in supplemented 7H9 media. M. smegmatis cultures were grown to log phase, diluted to a theoretical OD₆₀₀ of 0.002 and split into two equal volumes with and without 100 ng/mL of ATc. Subsequently, 100 µL of each bacterial suspension was added to the respective wells and, the plates were incubated at 37°C for 3 days. After incubation, the wells were observed for bacterial growth, usually indicated by turbidity or the presence of a bacterial pellet in the bottom of the wells. The dilution at which no bacterial growth was observed corresponds to the annotated MIC value. This assay was performed at least three times for each culture of interest and the median MIC values were calculated. Whenever mentioned, the “breakpoints” refer to the EUCAST non-species related PK/PD breakpoints ² (version 12) because there are not any currently defined beta-lactam breakpoints for mycobacteria.

To address the contribution of PG modifications to the intracellular survival of mycobacteria, J774.A1 macrophages were infected with control and mutant strains, as described previously (Anes et al., 2006; Jordão et al., 2008). After several passages and 48 h before infection, the macrophages were seeded in 96-well plates at a density of 12.5 x 10³ cells/well. That day, bacterial cultures were diluted to the same OD₆₀₀ and, incubated at 37°C with shaking until an OD₆₀₀ of 0.1 was reached. At that point (approximately 24 h before infection), the cultures were incubated with or without 100 ng/mL of ATc. On the day of infection, the cultures were centrifuged at 3000 g for 5 minutes and washed with 5 mL of PBS 1 X (Gibco) (Anes et al., 2006; Jordão et al., 2008). To obtain a single-cell suspension, the cultures were resuspended in DMEM media, subjected to 5 min of ultrasonic bath, and centrifuged at 500 g for 1 min (Pires et al., 2017). The cell media was discarded from the seeding plates and the macrophages were infected with a single-cell suspension of mycobacteria at a multiplicity of infection (MOI; ratio of bacteria/cells) of 5 and, incubated at 37°C with 5% CO₂ for different timepoints (1 h; 4 h; 24 h) (Anes et al., 2006; Jordão et al., 2008; Marques, 2020; Silveiro, 2020). Following internalization, the cells were washed twice with PBS 1 X and resuspended in DMEM supplemented with 10 μg/mL gentamicin to kill extracellular bacteria and 100 ng/mL of ATc, when appropriate.

At timepoints 1 h, 4 h and 24 h, the media was discarded, and the macrophages were lysed after 15 min incubation with a 0.05% IGEPAL solution (Pires et al., 2017). Serial dilutions of the resulting lysates were prepared in sterile water and plated onto 7H10 plates, which were incubated for 2 days at 37°C with 5% CO₂ (Anes et al., 2006; Jordão et al., 2008). The colony forming units (CFUs) were then counted using an inverted microscope (Nikon TMS Inverted Phase Contrast Microscope). Three biological replicates, each counting three technical replicates, were performed.

To assess whether the genes encoding the enzymes responsible for the characteristic modifications of mycobacterial PG were conserved in Mtb, data was retrieved from a study recently published by our group that set to uncover beta-lactam susceptibility patterns in a collection of 172 clinical strains of Mtb (Olivença et al., 2022b). Briefly, for each strain, quality improved reads were individually mapped against Mtb H37Rv reference genome (GenBank accession number AL123456.3), and the identification of single nucleotide polymorphisms (SNPs) was performed as previously described (Olivença et al., 2022b). The resulting MIC values were used in this study to uncover susceptibility differences to AMX or MEM, with or without clavulanate, that could be associated with frequent SNPs (with a mutation frequency ≥ 3) in the proposed genes of interest.

The statistical analysis was preformed using GraphPad (version 8.4.3.686) and data are presented as mean ± SEM, except if stated otherwise. Normality testing for qRT-PCR or CFUs data was preformed using Q-Q plots or the Shapiro-Wilk Test, respectively. Multiple comparisons were made using one-way ANOVA, followed by pairwise comparisons of selected groups using the Holm-Sidak test. All the prerequisites of the tests were verified.

To construct the knockdown mutants, a previously developed ATc-inducible dCas9_Sth1-based CRISPRi system, which enables the efficient knockdown of both essential and non-essential genes in mycobacteria, was used (Rock et al., 2017; Wong and Rock, 2021). To achieve efficient repression, either the non-template strand or the promoter region of the target genes murT (MSMEG_6276), gatD (MSMEG_6277) and namH (MSMEG_6410) were targeted. The efficiency of CRISPRi-mediated gene silencing is determined by several factors, namely PAM strength and the PAM position along the targeted gene (Rock et al., 2017; Marques, 2020). Therefore, various PAM sequences were identified on the template strand of the target genes and chosen for differential repression regarding PAM strength and target site ( Figure 1A ). PAM strength is directly proportional to the silencing efficiency, indicated as fold repression, of the possible PAM variants for dCas9_Sth1-mediated gene silencing in mycobacteria (Rock et al., 2017; Wong and Rock, 2021). To avoid lethality effects when targeting allegedly essential genes, the essentiality classification of the respective orthologue genes in Mtb was considered. The gene local context was also taken into consideration due to the previously reported probability of downstream polar effects (Choudhary et al., 2015; Singh et al., 2016; Rock et al., 2017) and reverse polarity for operonic genes (Peters et al., 2016), as is the case of our target genes ( Figure 1B ).

To uncover synergistic effects between the action of beta-lactams and the depletion of the N-glycolylation of muramic acid and/or the D-iso-Glu amidation of PG, knockdown mutants were also constructed in previously modified strains, lacking BlaS, PM965 (M. smegmatis ΔblaS1) and PM979 (M. smegmatis ΔblaS1 ΔnamH) (Raymond et al., 2005).

For the construction of the double mutants, having PM965 as a parental strain, the above knockdown mutants of murT, gatD and namH were considered. M. smegmatis PM965 was transformed with the following sgRNA-containing CRISPRi plasmids targeting the genes of interest: sgRNA1 or sgRNA2 for murT, sgRNA2 or sgRNA4 for gatD and, sgRNA1 or sgRNA3 for namH. To select mutants with medium-high repression vs. medium-low repression, the plasmids were chosen based on theoretical assumptions (differences in PAM strength and target site) (Choudhary et al., 2015; Rock et al., 2017) and on the spotting phenotypes of the single mutants ( Figure 2 ).

To facilitate the identification of synergies between the depletion of NamH and the MurT/GatD enzymatic complex, the triple mutants, using PM979 as a parental strain, were constructed. In this case, only the murT and gatD knockdown mutants were considered since PM979 already lacks the namH gene. The plasmids used for target repression were the same that were used for the double mutants.

To investigate the role of the D-iso-Glu amidation and of the N-glycolylation of PG in antibiotic susceptibility, the MICs of ethambutol, isoniazid and three beta-lactams, with or without clavulanate, were determined against the constructed mutants and respective control strains ( Figure 6 ). For these experiments, 100 ng/mL of ATc were used since the MIC of ATc against M. smegmatis mc² 155 WT has been defined as 500 ng/mL (Ehrt et al., 2005). Moreover, other studies have also used 100 ng/mL of ATc to induce CRISPRi-mediated silencing in M. smegmatis WT (Singh et al., 2016; Rock et al., 2017), including those employing the dCas9_Sth1-based system (Rock et al., 2017).

As expected, the observed MICs for the negative controls (M. smegmatis WT, M. smegmatis:PLJR962) were not considerably influenced by the addition of ATc (Marques, 2020; Silveiro, 2020). Notably, the repression of the murT/gatD operon and of the namH gene promoted increased susceptibility to all tested beta-lactams (Marques, 2020; Silveiro, 2020). The MICs of AMX remained mostly constant, as 2-fold differences can occur due to technical variability and are, thus, not considerable. Substantial susceptibility differences (≥ 4-fold) in AMX+CLA were found when gatD was repressed with sgRNA1-gatD or when namH was silenced with sgRNA1-namH. Hence, the correspondent induced knockdown mutants are susceptible to AMX+CLA according to the EUCAST breakpoints ( Table S4 ). In the case of the induced M. smegmatis:PLJR962:sgRNA1-gatD knockdown mutant, the susceptibility gain might be explained by a viability loss, as shown in the phenotyping assays. Unlike CTX or MEM, the addition of CLA to beta-lactams caused marked decreases in the MIC of AMX. The most notable susceptibility differences (ranging from 4 to 16-fold) occurred with CTX (+CLA), being present across all induced knockdown mutants, independently of the target gene. The most extensive increases in susceptibility to CTX (+CLA) occurred when gatD was repressed with sgRNA1-gatD or, when namH was silenced with sgRNA1-namH and sgRNA3-namH. Additionally, the fitness loss of the induced M. smegmatis:PLJR962:sgRNA1-gatD knockdown mutant also provoked a 4-fold decrease in the MIC of MEM (+CLA). In contrast, the repression of namH consistently promoted increased susceptibility to MEM (+CLA), thereby inferring a particular role of NamH in resistance to the carbapenem. Indeed, all knockdown mutants with a meropenem MIC ≤ 2 µg/mL are classified as susceptible to MEM according to the breakpoints ( Table S4 ). The MICs of EMB and INH remained essentially unchanged, with an interesting exception. The repression of murT with sgRNA1-murT and sgRNA3-murT caused a 4-fold decrease in the MIC of INH, which is most likely due to viability loss of the respective mutants.

Similarly, the median MICs of the double mutants were determined for the same antibiotics ( Figure 7A ). The MICs of M. smegmatis PM965 and PM965:PLJR962 remained unaltered with the addition of ATc. When compared to M. smegmatis WT, the parental PM965 strain presented substantially lower MICs to all beta-lactams because this strain lacks the major beta-lactamase of M. smegmatis, BlaS (Raymond et al., 2005). The absence of BlaS in conjunction with the repression of murT/gatD and namH promoted increased susceptibility to the tested beta-lactams, although with insubstantial differences in the MICs of AMX (+CLA) and MEM (+CLA). The lack of BlaS in PM965 aided the classification of the induced murT and namH knockdown mutants as susceptible to AMX, according to the breakpoints ( Table S4 ). Both induced M. smegmatis PM965:PLJR962:sgRNA1-murT and PM965:PLJR962:sgRNA1-namH are classified as susceptible to AMX+CLA according to the breakpoints. The most notable susceptibility differences, ranging from 4 to 16-fold in comparison with M. smegmatis WT and PM965, occurred with CTX (+CLA), being present across most induced knockdown mutants. Extensive susceptibility differences (≥16-fold) from the PM965 strain in CTX (+CLA) were found when murT was repressed with sgRNA1-murT or when namH was silenced with sgRNA3-namH. The increased susceptibility of the induced M. smegmatis PM965:PLJR962:sgRNA1-murT knockdown mutant to beta-lactams might be explained by fitness loss, as shown in the phenotyping assays. All induced knockdown mutants constructed in the BlaS-depleted strain except for induced M. smegmatis PM965:PLJR962:sgRNA4-gatD are susceptible to MEM according to the breakpoints ( Table S4 ). The MICs of EMB and INH remained essentially unchanged.

Similarly, the median MICs of the triple mutants, having M. smegmatis PM979 as a parental strain, were determined for the same antibiotics ( Figure 7B ). The MICs of M. smegmatis PM979 were not considerably altered by ATc addition. In fact, the PM979 strain was found to be very susceptible to beta-lactams when compared to the WT and PM965 strains, suggesting a major role for the N-glycolylation of muramic acid in resistance to beta-lactams. The MICs of AMX (+CLA), CTX (+CLA) and MEM+CLA against M. smegmatis PM979:PLJR962 suffered a 2-fold decrease when inducer was added, with the obtained MIC values for AMX (+CLA) and MEM+CLA being equal to those obtained with the PM979 strain in induced conditions. Therefore, the MIC differences observed for the knockdown mutants in CTX (+CLA) cannot be interpreted as a result of their knockdowns alone since these differ in 2-fold from the MIC values obtained with the induced PM979 strain. This reproducible effect is most likely due to dCas9_Sth1-associated toxicity as an exogenous protein to the bacterial system. Despite this remark, the additional repression of murT/gatD in the PM979 strain favoured even lower beta-lactam MICs. The cumulative effect of the absence of BlaS and NamH and of the depletion of the D-iso-glutamate amidation of PG provoked increased susceptibility to beta-lactams when compared to the parental strain PM979 (2 or 4-fold differences), and especially when compared to M. smegmatis WT and PM965 (4 to 128-fold differences). Although M. smegmatis PM979 is already very susceptible to beta-lactams when compared to the WT and PM965 strains, the additional repression of murT/gatD favored even lower beta-lactam MICs. When compared to the PM965 strain, all sgRNAs leading to depleted D-iso-Glu amidation caused considerable 4-fold decreases in the MIC of AMX, with comparable results for AMX+CLA. Most sgRNAs caused a 64 to 128-fold reduction in the MIC of CTX (+CLA) compared to PM965 and WT MIC values. Moreover, a noteworthy MEM MIC value of 0.5 µg/mL was obtained with all sgRNAs, differing in 8-fold from the PM965 and WT MIC values. Although the fold differences in the MIC of MEM between the triple mutants and the PM965 control strain seem to be less substantial, the PM965 MIC value for MEM is extremely low in comparison to that of AMX and CTX, making it difficult to attain great fold differences. Furthermore, all induced knockdown mutants are susceptible to AMX (+CLA) and MEM according to the breakpoints ( Table S4 ). Again, the MICs of EMB and INH remained essentially unchanged.

To elucidate the contribution of the D-iso-Glu amidation and of the N-glycolylation of PG to mycobacterial intracellular killing, J774 murine macrophages were infected with all strains at a MOI of 5 and, disrupted for the assessment of the intracellular survival of bacteria (in CFUs/mL) at 1 h, 4 h and 24 h post-infection ( Figure 8 ).

As expected, the intracellular survival of mycobacteria decreased over time (Marques, 2020; Silveiro, 2020). The single mutants’ results show that the mean CFUs/mL of the negative controls M. smegmatis WT and PLJR962 were not significantly influenced by ATc addition ( Figures 8A-C ). The repression of murT with sgRNA1-murT caused a highly significant decrease in the intracellular survival of these bacteria at all timepoints ( Figure 8A ; P < 0.0001, when compared to M. smegmatis WT ATc, PLJR962 ATc or the respective uninduced control). This observation is explained by a strong repression of the murT gene, causing a lethality effect that can be verified by the low viability of the correspondent induced murT ^- mutant in the phenotyping assays. Besides, the repression of murT with sgRNA3-murT provoked a significant decrease in intracellular bacterial survival at t = 24 h ( Figure 8A ; P = 0.004, comparing with the respective uninduced control) while retaining some viability in the phenotyping assays. A highly significant decrease in the intracellular survival of the induced gatD knockdown mutants at t = 24 h was only attained with sgRNA3-gatD ( Figure 8B ; P < 0.0001, compared to all controls). Furthermore, the knockdown of namH attained with sgRNA1-namH also caused a highly significant decrease in the intracellular survival of bacteria at t = 24 h ( Figure 8C ; P < 0.0001, compared to all controls). These results indicate that both the D-iso-Glu amidation and the N-glycolylation of PG contribute to the intracellular survival of mycobacteria during infection. Moreover, the PM979 strain, lacking both BlaS and NamH, undergoes a significantly faster clearance by J774 macrophages when compared to WT or PM965 ( Figure S6 ).

The mean CFUs/mL of the double ( Figures 8D, E ) and triple mutants ( Figure 8F ) were also determined. The controls of the double mutants (M. smegmatis PM965, M. smegmatis PM965:PLJR962) and of the triple mutants (M. smegmatis PM979, M. smegmatis PM979:PLJR962) did not display any differences in intracellular survival after the addition of inducer. Generally, the obtained results for the knockdown of murT and gatD in the PM965 and PM979 strains were identical to those observed with the single mutants ( Figure 8 ). Unfortunately, the effect of D-iso-Glu amidation on mycobacterial intracellular survival could not be shown using the best candidate sgRNAs targeting murT/gatD leading to significant changes in the CFUs/mL because the respective mutants were not available. Still, the repression of murT in both PM965 and PM979 strains with sgRNA1-murT caused a highly significant decrease in the intracellular survival of these bacteria at all timepoints (P < 0.0001, compared to all controls) ( Figures 8D, F ). Remarkably, the knockdown of namH in the PM965 strain with sgRNA1-namH caused a very significant decrease in the bacterial intracellular survival at t = 24 h when compared to PM965 ATc (P = 0.0018), PM965:PLJR962 ATc (P = 0.0048) or the respective uninduced mutant (P = 0.0008) ( Figure 8E ).

To uncover whether the genes encoding the proteins responsible for the characteristic modifications of mycobacterial PG are of conserved nature, we benefited from the data obtained for 172 Mtb isolates in a previous study (Olivença et al., 2022b), where the susceptibility of the strains to AMX and MEM, with and without clavulanate, was also evaluated.

By performing a SNP-based analysis against the H37Rv reference genome, we found the murT, gatD and namH genes to be highly conserved in this specific set of clinical strains of Mtb ( Figure 9A , Tables S5 , S6 ). When a certain SNP was detected in three or more strains, the geometric mean of the MIC of these strains to AMX and MEM was compared to the median MIC of the control strain Mtb H37Rv WT ( Figure 9B ). Among the 172 strains, only one SNP (4158032 snp A>C) was identified for the murT gene, occurring in the region coding for the C-terminal domain of MurT (DUF1727) ( Figure 9C ). Although synonymous mutations do not carry as much importance as missense variants or indels, three synonymous variants were found in gatD. One of these mutations (4158493 snp C>T) was found in 10/172 strains, all of which were classified as lineage 2 strains and had increased beta-lactam MIC values, namely in the case of AMX+CLA and MEM ( Figure 9B ). This mutation occurred in the GATase_3 domain of GatD, near its first catalytic site (C94) ( Figure 9D ). Another one of these mutations (4158865 snp G>A) was found in only 4/172 strains, all of which were classified as lineage 4 strains, however without any considerable MIC differences ( Figure 9B ). Four variants were found for the namH gene, but always at a low frequency so they could not be associated with MIC changes. Two of these mutations (4282589 snp T>C; 4282707 snp C>G) occurred in regions coding for the UlaG domain of NamH whereas the other two (4283319 snp G>A; 4283424 snp G>C) occurred in regions coding for the Rieske domain ( Figure 9E ).