The pentaglycine bridges of Staphylococcus aureus peptidoglycan are essential for cell integrity

Bacterial cells are surrounded by cell wall, whose main component is peptidoglycan (PG), a macromolecule that withstands the internal turgor of the cell. PG composition can vary considerably between species. The Gram-positive pathogen Staphylococcus aureus possesses highly crosslinked PG due to the presence of cross bridges containing five glycines, which are synthesised by the FemXAB protein family. FemX adds the first glycine of the cross bridge, while FemA and FemB add the second and the third, and the fourth and the fifth glycines, respectively. Of these, FemX was reported to be essential. To investigate the essentiality of FemAB, we constructed a conditional S. aureus mutant of the femAB operon. Depletion of femAB was lethal, with cells appearing as pseudomulticellular forms that eventually lyse due to extensive membrane rupture. This deleterious effect was mitigated by drastically increasing the osmolarity of the medium, indicating that pentaglycine crosslinks are required for S. aureus cells to withstand internal turgor. Despite the absence of canonical membrane targeting domains, FemA has been shown to localise at the membrane. To study its mechanism of localisation, we constructed mutants in key residues present in the putative transferase pocket and the α6 helix of FemA, possibly involved in tRNA binding. Mutations in the α6 helix led to a sharp decrease in protein activity in vivo and in vitro but did not impair correct membrane localisation, indicating that FemA activity is not required for localisation. Our data indicates that, contrarily to what was previously thought, S. aureus cells do not survive in the absence of a pentaglycine cross bridge.


Results and Discussion
the femAB operon is essential for the viability of S. aureus. Previous S. aureus femAB null mutants likely had compensatory mutations 16,18,19 . To evaluate the essentiality of femAB, as well as the phenotypes resulting from FemAB depletion in a background without the existence of compensatory mutations, we constructed a conditional femAB mutant. The femAB operon of the clinically relevant CA-MRSA strain MW2 was placed under the control of the IPTG inducible Pspac promoter. As the Pspac promoter is known to be leaky 20 , a plasmid-encoded lacI repressor gene was provided to decrease the basal transcription of femAB. The resulting strain was named MW2-iFemAB.
Growth of MW2-iFemAB in liquid medium supplemented with IPTG at 0, 10, 25 and 500 µM was followed for 10 hours. In the presence of 500 µM of IPTG, growth of the conditional mutant was similar to the parental strain MW2 (Fig. 1a), therefore this concentration of inducer was used in subsequent assays. The growth rate of MW2-iFemAB decreased with decreasing IPTG concentrations and, surprisingly, no bacterial growth was observed in the absence of IPTG, indicating that this operon is essential for survival (Fig. 1a), contrarily to what was previously thought. In order to confirm the essentiality of femAB, we measured colony forming units (CFUs) in cultures of MW2-iFemAB incubated in the absence of IPTG, and found severe loss of viability when compared to IPTG-induced cultures or to parental MW2 ( Supplementary Fig. 1). To evaluate the possibility of polar effects on genes downstream of femAB, we introduced a plasmid encoding femAB under the control of the cadmium-inducible promotor Pcad into MW2i-FemAB. Expression of plasmid-encoded femAB was able to restore growth in cells of MW2-iFemAB background depleted of endogenous femAB ( Supplementary Fig. 2). We also tested the essentiality of the femAB operon in an HA-MRSA background -strain COL -and likewise no growth of COL-iFemAB was observed in the absence of IPTG ( Supplementary Fig. 3).
To assess the effect of loss of FemAB activity on PG composition, we analysed the cell wall of MW2-iFemAB cells incubated with 500, 10 or 0 µM of IPTG until bacterial growth was arrested in the non-induced culture (see Methods). As expected, the muropeptide profiles in cells depleted of FemAB show a massive accumulation of peak 4 (Fig. 1b, [IPTG] 0 µM, see Supplementary Fig. 4 for peak assignment), which was previously identified as the monomeric pentapeptide substituted with a single glycine residue 21 , the substrate of FemA. This was in contrast to cells where femAB expression was fully induced (Fig. 1b, [IPTG] 500 µM), where the major monomeric form present was the pentaglycine substituted monomer (peak 5). To confirm the composition of these muropeptide species, peaks 4 and 5 were analysed by mass spectrometry and found to have molecular weights of 1024.48 Da and 1252.55 Da, respectively, consistent with the expected weight of a monoglycyl-(1024.49 Da) and a pentaglycyl-(1252.58 Da) substituted GlcNAc-MurNAc-pentapeptide. Lack of FemAB also prevented the formation of pentaglycine crosslinked forms such as dimers (peak 11), trimers (peak 15), tetramers (peak 16) and higher order forms which co-elute near the end of the run (Fig. 1b, arrow). Low FemAB expression levels, just enough to sustain bacterial growth ([IPTG] 10 µM), resulted in the presence of some pentaglycine crosslinked forms (peaks 11, 15, 16, etc.). This degree of peptidoglycan structural organisation might be the minimum to ensure cell viability.
Loss of FemAB activity leads to membrane damage. The morphology of cells depleted of FemAB was analysed by super resolution structured illumination microscopy (SIM). MW2-iFemAB was grown with or without IPTG (500 µM). Following growth arrest of the non-induced culture ( Supplementary Fig. 5, arrow), cells were stained with membrane dye FM 4-64, PG dye Van-FL and DNA dye Hoechst 33342. In the presence of IPTG, MW2-iFemAB cells divided normally with DNA segregation preceding the synthesis of a division septum at mid-cell (Fig. 2a, top row). Cells containing multiple septa were rarely observed (0.3 ± 0.6%, N = 300 septating www.nature.com/scientificreports www.nature.com/scientificreports/ cells - Fig. 2b, left panel). In contrast, FemAB depleted cells often appeared as pseudomulticellular forms with two or more perpendicular septa (46.7 ± 9.9%, N = 300 septating cells), suggesting that a second round of division starts before daughter cell separation is completed (Fig. 2b, right panel arrows). Furthermore, the nucleoid morphology of FemAB depleted cells was altered, with the presence of cells containing condensed DNA (Fig. 2a, bottom row asterisks, 43.7 ± 5.5% in FemAB − vs 0.7 ± 0.6% in FemAB + , N = 300 cells for each condition). Our results are in agreement with previous reports showing that suppressed fem mutants show irregular placement of cross walls and retarded cell separation 16 . This phenotype can be a consequence of either multiple septation or defective cell splitting.
When cells depleted of FemAB were incubated for longer periods, we noticed a decrease in culture density, suggesting cell lysis ( Supplementary Fig. 5). We therefore imaged cells 2 hours after growth arrest and observed extensive membrane damage, characterised by bulges and invaginations (Fig. 3a, arrow, 68.9 ± 2.0% in FemAB − vs 4.0 ± 2.6% in FemAB + , N = 300) and the presence of anucleate cells, indicative of loss of viability (Fig. 3a, asterisks, 48.0 ± 8.8% in FemAB − vs 1.0 ± 1.0% in FemAB + , N = 300). These results suggest that the inability of S. aureus to survive with shortened crossbridges could be because the three-dimensional structure of this alternate PG does not confer sufficient robustness and/or flexibility to bear the internal osmotic pressure, in these conditions, causing the cells to rupture. In order to test this hypothesis we incubated MW2-iFemAB in the absence of IPTG (to deplete FemAB expression) with increasing concentrations of NaCl added to the medium, to alleviate turgor. MW2-iFemAB was able to grow in the presence of NaCl in a dose dependent manner (Fig. 3b), confirming that in the absence of pentaglycine crosslinks, the PG layer is not able to withstand the internal pressure exerted on the membrane. Supplementing the medium with a different osmolyte, sucrose, also rescued cells depleted of FemAB ( Supplementary Fig. 6). These results are in accordance with data from Hübscher and colleagues 18 , who showed by transcriptome analysis that the femAB null mutant AS145 adapted to the FemAB deficit by tuning its metabolic pathways, presumably to reduce turgor. It is likely that monoglycyl-substituted muropeptides are not suitable substrates for transpeptidation by S. aureus PBPs in vivo and thus crosslinking of glycan chains is stalled after FemAB depletion. Accordingly, solid-state NMR data obtained by Kim et al. 22 indicated that monoglycyl crossbridges would be too short to connect the glycan chains of the S. aureus PG, and that crosslinking with such a reduced bridge length would require a major rearrangement of the tertiary structure of PG 22 .
FemA activity is not required for its membrane localisation. The Fem proteins are non-ribosomal peptidyl transferases which use dedicated amino acid charged tRNA molecules as substrates, an interesting www.nature.com/scientificreports www.nature.com/scientificreports/ activity seldom seen in nature 12 . The mechanism of this transfer is still poorly understood, as binding sites for entering tRNA molecules have not been identified. In the case of Fem proteins which transfer two amino acids, such as FemA (and FemB), the transfer of both glycines to lipid II appears to occur simultaneously rather than sequentially, judging from in vitro data, which may indicate that these proteins act as homodimers 13,23 . We have recently reported that all three Fem proteins of S. aureus localise to the membrane throughout the entire cell cycle, which was unexpected given that these proteins lack canonical transmembrane domains 24 . Therefore a possible mechanism of Fem localisation to the membrane could be through protein activity, which is dependent on interactions with both the substrate lipid II and glycyl-charged tRNA molecules.
In order to investigate the mechanism of localisation of FemA, we identified possible key regions in FemA required for activity, based on the known crystal structure of FemA 25 and on homology to the FemX protein from Weisella viridescens 26,27 . We decided to focus on the putative transferase pocket that contains Arg220, Phe224 and Tyr327, which are conserved across the Fem family 25,27 . We also mined the sequence of FemA for regions which could bind DNA/RNA using DP-Bind 28,29 , in order to identify the putative tRNA-binding site. We found that the region with the highest probability of binding to RNA corresponded to the α6 helix (aa 176-188) of FemA, rich in Lys/Arg residues with polar and charged sidechains exposed to the solvent 25 , which could stabilise the entering tRNA. Specifically, amino acids Lys180 and Arg181 showed >96% probability of binding DNA/RNA in each of three individual prediction algorithms performed by DP-Bind (see Methods), and therefore were selected for mutagenesis. www.nature.com/scientificreports www.nature.com/scientificreports/ To assess if the selected mutations had an effect on FemA transferase activity, we cloned wild-type femA into the pET-24b expression vector and performed site-directed mutagenesis on femA residues to obtain FemA mutants where the target residues were replaced by alanines. In this way, we constructed pET-FemA RF220AA and pET-FemA Y327A , in order to express mutants in the transferase domain and pET-FemA KR180AA to express a mutant of the predicted tRNA binding helix. We purified recombinant FemA wt , FemA KR180AA , FemA RF220AA and FemA Y327A with C-terminal histidine tags and synthesised the FemA substrate lipid II-Gly 1 in vitro (see Methods). As recombinant FemA RF220AA was very unstable and readily precipitated, we could not use it for further assays. Lipid II-Gly 1 was trapped in Triton X-100 micelles and incubated with either FemA wt , FemA KR180AA or FemA Y327A in the presence of [U-14 C]-glycine charged tRNA. After 30, 60 or 90 minutes the lipid fraction was extracted and separated by thin layer chromatography and radioactive glycine transfer to lipid II-Gly 1 was measured. Both mutants FemA KR180AA and FemA Y327A transferred less [U− 14 C]-glycine to their substrate than FemA wt , consistent with a reduction of enzyme activity ( Fig. 4a and Supplementary Fig. 7).
In order to both confirm decrease of FemA activity in vivo and study protein localisation, we used the backbone of pFemAB wt , a replicative vector encoding a femA-mCherry fusion followed by femB (both under the control of a cadmium inducible promoter), to generate femA-mCherry alleles with the mutations described above. These expression plasmids were transformed into MW2-iFemAB, allowing us to deplete native femAB expression (in the absence of IPTG) and express mutant alleles (in the presence of cadmium).
We were able to complement the lack of femAB expression from the native locus by expressing femA-mCherry-femB from pFemAB wt in the presence of cadmium (0.1 µM), as assessed by growth rates, morphology, lysostaphin (an enzyme that cuts pentaglycine bridges 30 ) and oxacillin MICs and muropeptide composition ( Table 1). Expression of the catalytic site mutants FemA RF220AA and FemA Y327A caused a reduction of the pentaglycine substituted monomer content in peptidoglycan (Fig. 4b), although morphology was similar to wild-type and no major differences in lysostaphin and oxacillin MICs were observed (Table 1). In contrast, the double mutation in the α6 helix of FemA caused severe loss of FemA activity. The FemA KR180AA mutant showed a marked reduction in growth rate, increased lysostaphin and decreased oxacillin resistances and a pseudomulticellular morphology when observed by microscopy (Table 1), similar to what was observed when depleting femAB expression. Furthermore, analysis of the muropeptide content in this mutant revealed a pronounced accumulation of monoglycyl substituted pentapeptides and concomitant reduction in pentaglycine crosslinked species (Fig. 4b and Table 1). Nevertheless, FemA KR180AA still localised to the membrane, similarly to FemA wt , indicating that FemA localisation is independent of protein activity (Fig. 4c). Because loss of activity in FemA KR180AA is likely due to a www.nature.com/scientificreports www.nature.com/scientificreports/ deficit in tRNA binding, it is possible that FemA KR180AA could still localise to the membrane through recognition of the lipid-linked peptidoglycan precursor.
The mutations in the substrate binding pocket had a minor effect on FemA localisation, since FemA Y327A and FemA RF220AA appeared dispersed in the cytoplasm in a fraction of the cell population (Fig. 4c, white arrows). Nevertheless, as these mutations did not seem to decrease protein activity in vivo to a great extent, we could not conclude that the mechanism of FemA localisation to the membrane is via substrate recognition. An alternative possibility is that the recruitment of FemA to the membrane is mediated by protein-protein interactions with the membrane associated eukaryotic-type serine/threonine kinase Stk, a global cell wall synthesis regulator, which was recently shown to interact with FemA and FemB by bacterial two hybrid 31 . However, the same study could not find interactions between Stk and FemX, which initiates Lipid II crossbridge synthesis, or between FemX and FemA/B 31 . Further experiments are necessary to clarify the interactions of Fem proteins with each other and with their substrates, in order to understand how the localisation and timing of PG crossbridge synthesis is modulated during the cell cycle. FemA KR180AA localised to the membrane in >95% of the cells, similarly to FemA wt . FemA Y327A and FemA RF220AA appeared dispersed in the cytoplasm in a fraction of the population (27% and 10%, respectively, white arrows). N = 400 cells for each strain.

Concluding Remarks
The structural features of the staphylococcal PG seem remarkably unique in nature, as pentaglycine crosslinks have not been observed outside of the genus. These long bridges likely confer high flexibility to S. aureus PG that allows a high level of PG crosslinking, which in turn allows the cell to withstand high internal turgor. Accordingly, femAB mutants isolated in the past adapted to life with shortened crossbridges by reducing metabolic activity 18 . Moreover, the nature and length of PG branching has been implicated in resistance to β-lactams, not only in S. aureus, but also in other bacteria such as Streptococcus pneumoniae 7,32-34 .
We have shown that depletion of FemAB is lethal in CA-MRSA strain MW2 and in HA-MRSA strain COL, leading to the disruption of the cell envelope and causing cells to lose viability. This suggests that monoglycyl-substituted muropeptides are not good substrates for transpeptidation in vivo, either because transpeptidases fail to recognise them or because different S. aureus glycan strands are too far apart to be crosslinked via crossbridges with only one glycine.

Construction of S. aureus strains.
In order to construct an S. aureus strain with the femAB operon under the control of an inducible promoter, a fragment containing the first 400 bp of femA was amplified from S. aureus MW2 DNA with primer pair spacfemab_P1 EcoRI/spacfemab_P2 BamHI (see Supplementary Table 2 for primer sequences), cut with EcoRI and BamHI restriction enzymes and cloned into pMUTIN4 35 , downstream of the Pspac promoter, giving plasmid pFemABi, which was sequenced. pFemABi was then propagated in DC10B cells, electroporated into electrocompetent RN4220 cells, and transduced (using phage 80α) to MW2 and COL, where it integrated in the femAB locus by homologous recombination. The resulting strains contain a truncated copy of femA under the control of the femAB native promoter, and the femAB operon under the control of Pspac. Multicopy plasmid pMGPII 36 , which encodes Pspac repressor LacI, was then transduced into these strains, giving rise to MW2-iFemAB and COL-iFemAB.
To construct S. aureus strains expressing mutated alleles of FemA-mCherry together with wild-type FemB, first a fragment containing femA-mCherry-STOP codon-femB was amplified from pMADfemAmch 24 using primers pcnfemab_P1 BamHI and pcnfemab_P2 EcoRI. This fragment was cut with BamHI and EcoRI and cloned into replicative vector pCNX 37 , under the control of Pcad, giving plasmid pFemAB wt . pFemAB wt was then used as the template for site-directed mutagenesis using Phusion polymerase (Thermo Scientific) following manufacturer's instructions. Primers fema_kr180aa_fw/ fema_kr180aa_rev were used to generate pFemA KR180AA B, encoding both K180A and R181A mutations; primers fema_rf220aa_fw/ fema_rf220aa_rev were used to generate pFemA RF220AA B, encoding both R220A and F224A mutations; and primers fema_Y327a_fw/ fema_Y327a_rev were used to generate pFemA Y327A B, encoding the Y327A mutation. Each plasmid was sequenced to confirm the presence of the mutations. pFemAB wt , pFemA KR180AA B, pFemA RF220AA B, pFemA Y327A B and empty vector pCNX were propagated in DC10B, electroporated into RN4220 and transduced to MW2-iFemAB, giving strains MW2pFemAB wt , MW2pFemA KR180AA , MW2pFemA RF220AA , MW2pFemA Y327A and MW2pCNX respectively.
Growth curves of S. aureus strains. To assess whether the femAB operon is essential for growth, or the effects of FemA mutations on growth rates, overnight cultures of MW2, MW2-iFemAB, COL, COL-iFemAB, MW2pFemAB wt , MW2pFemA KR180AA , MW2pFemA RF220AA , MW2pFemA Y327A were grown in TSB with 500 µM of IPTG, with the appropriate antibiotics (when applicable, at the concentrations described above), back-diluted 1:500 in the same medium and grown until the cultures reached an OD 600 of 0.7. At this point the cultures were washed three times to remove IPTG and back-diluted to an OD 600 of 0.007 in fresh TSB containing either 0, 10, 25 or 500 µM of IPTG, or 0.06, 0.12 or 0.25 M of NaCl, or 0.25% (w/v) of sucrose, in the case of MW2-iFemAB and COL-iFemAB. In the case of MW2pFemAB wt , MW2pFemA KR180AA , MW2pFemA RF220AA , MW2pFemA Y327A and MW2pCNX, cells were incubated without IPTG and cadmium chloride was added at 0.1 µM to drive the expression of either wild-type or mutant femA alleles from the pCNX-based plasmids.
Growth of all strains was monitored for 10 hours in a Bioscreen C Analyzer (Growth Curves USA), at 37 °C with shaking with OD 600 readings taken every 15 minutes. Growth curves were obtained from three independent experiments done with three biological replicates.

Viability Assays.
To assess the effect of femAB depletion on S. aureus cell viability, overnight cultures of MW2 and MW2-iFemAB grown with IPTG and the appropriate antibiotics were washed three times to remove the IPTG and back-diluted 1:2000 in fresh TSB supplemented with 500 µM of IPTG or no IPTG. Bacterial growth was monitored for 8 hours by measuring OD 600nm and 100 µl of appropriate dilutions were plated on TSA (MW2) or TSA supplemented with IPTG at 500 µM (MW2-iFemAB), at 1 hour intervals for 8 hours. Following overnight incubation, colony forming units (CFUs) on each plate were counted. In each experiment, CFUs were counted in a total of 6 plates, corresponding to three technical replicates for each of 2 relevant dilutions. These assays were done in triplicate.
www.nature.com/scientificreports www.nature.com/scientificreports/ Minimum inhibitory concentration (MIC) assays. MICs of lysostaphin and oxacillin were determined by broth microdilution in sterile 96-well plates. The medium used was TSB, containing a series of two-fold dilutions of each compound. Cultures of S. aureus strains and mutants were added at a final density of ~5 × 10 5 CFU ml −1 to each well. Wells were reserved in each plate for sterility control (no cells added) and cell viability (no compound added). Plates were incubated at 37 °C. Endpoints were assessed visually after 48 h and the MIC was determined as the lowest concentration that inhibited growth. All assays were done in triplicate.

Purification and analysis of S. aureus muropeptides.
To evaluate changes in the peptidoglycan composition caused by the depletion of the femAB operon, or caused by the expression of mutant FemA proteins, cells of MW2, MW2-iFemAB, MW2pFemA wt , MW2pFemA KR180AA , MW2pFemA RF220AA and MW2pFemA Y327A were first grown overnight in TSB supplemented with 500 µM of IPTG and the applicable antibiotics. Cultures were then washed three times to remove the IPTG and back-diluted 1:500 in fresh TSB with 0, 10 or 500 µM of IPTG, in the case of MW2 and MW2-iFemAB, or in the presence of cadmium chloride and absence of IPTG, in the case of MW2pFemA wt , MW2pFemA KR180AA , MW2pFemA RF220AA and MW2pFemA Y327A . Cells were collected at mid-exponential phase and PG was purified as described by Filipe et al. 38 . Muropeptides were prepared from PG samples by digestion with mutanolysin (0.135 U/µg of PG, from Sigma-Aldrich) and analysed by reverse phase HPLC using a Hypersil ODS (C18) column (Thermo-Fisher Scientific). Muropeptide species were eluted in 0.1 M sodium phosphate, pH 2.0, with a gradient of 5-30% methanol for 155 minutes and detected at 206 nm. The proportion of pentaglycine to monoglycine substituted monomers (Gly5/Gly1 ratio) in muropeptide preparations was calculated using the areas corresponding to half of the width of peaks 5 (Gly5) and 4 (Gly1) from the obtained chromatogram. To determine the molecular weights of muropeptides in peaks 4 and 5, muropeptide samples were separated by reverse phase HPLC using two Chromolith ® Performance RP-8 endcapped 100-4.6 monolithic columns (Merck Millipore) in tandem with a linear gradient from 0.01% (v/v) Trifluoroacetic acid (TFA) to 0.01% (v/v) TFA 15% (v/v) acetonitrile over 58 minutes followed by a wash step at 0.01% (v/v) TFA 50% (v/v) acetonitrile for 6 minutes at a flow of 1 mL/min at 30 °C. Sample elution was monitored by recording the absorbance at 206 nm. Muropeptide peaks 4 and 5 were collected, the eluent was evaporated at 30 °C using a SpeedVac and then resuspended in water, prior to MS analysis by MicroLC-MS using an Eksigent LC4500 coupled to TripleTOF 6600 with Dual spray ion source.

S. aureus imaging by fluorescence microscopy.
To evaluate changes in morphology caused by the depletion of the FemAB, or to investigate the localisation of mutant FemA proteins, cells of MW2, MW2-iFemAB, MW2pFemA wt , MW2pFemA KR180AA , MW2pFemA RF220AA and MW2pFemA Y327A were first grown overnight in TSB supplemented with 500 µM of IPTG and the applicable antibiotics. Cultures were then washed three times to remove the IPTG and back-diluted 1:500 in fresh TSB with 0, 25 or 500 µM of IPTG, in the case of MW2 and MW2-iFemAB, or in the presence of cadmium chloride and absence of IPTG, in the case of MW2pFemA wt , MW2pFemA KR180AA , MW2pFemA RF220AA and MW2pFemA Y327A . Cells were grown to an OD600 nm of 0.4-0.6, collected and then washed with phosphate buffer saline (PBS). In the case of MW2pFemA wt , MW2pFemA KR180AA , MW2pFemA RF220AA and MW2pFemA Y327A , cells were then mounted on microscope slides covered with a thin layer of 1% agarose in PBS and imaged by fluorescence microscopy. In the case of MW2 and MW2-iFemAB, to evaluate cell morphology, cells were resuspended in PBS and incubated with membrane dye Nile Red (10 µg/ ml, Invitrogen), Hoechst 33342 (10 µg/ml, Invitrogen) and a mixture containing equal amounts of vancomycin (Sigma) and a BODIPY FL conjugate of vancomycin (Van-FL, Molecular Probes) to a final concentration of 0.8 µg/ml, for 5 minutes at room temperature. Cells were then washed three times with PBS before being spotted on the agarose pads and imaged.
Super-resolution Structured Illumination Microscopy (SIM) imaging was performed using an Elyra PS.1 microscope (Zeiss) with a Plan-Apochromat 63×/1.4 oil DIC M27 objective. SIM images were acquired using five grid rotations, with 34 μm grating period for the 561 nm laser (100 mW), 28 μm period for 488 nm laser (100 mW) and 23 μm period for 405 nm laser (50 mW). Images were captured using a Pco.edge 5.5 camera and reconstructed using ZEN software (black edition, 2012, version 8.1.0.484) based on a structured illumination algorithm, using synthetic, channel specific optical transfer functions and noise filter settings ranging from −6 to −8.
Wide-field fluorescence microscopy was performed using a Zeiss Axio Observer microscope with a Plan-Apochromat 100×/1.4 oil Ph3 objective. Images were acquired with a Retiga R1 CCD camera (QImaging) using Metamorph 7.5 software (Molecular Devices).
Overexpression and purification of recombinant His-tagged proteins. Recombinant proteins were purified essentially as described by Rohrer et al. 39 , with some modifications. Single colonies of BL21 (DE3) expression strains containing either plasmid pET-FemA wt , pET-FemA KR180AA , pET-FemA RF220AA or pET-FemA Y327A were isolated from LA plates with kanamycin and used to inoculate LB (1 L) containing kanamycin. Cultures were grown to an OD 600nm of approximately 0.6 at which point IPTG was added (final concentration 1 mM) and incubated for 3 hours with shaking (150 rpm) at 30 °C. Cells were harvested by centrifugation and washed with 50 mM sodium phosphate buffer (pH 7.5) containing 300 mM NaCl and 20% glycerol. Afterwards, cells were suspended in the same buffer, containing phenylmethylsulfonyl fluoride (PMSF, final concentration, 0.1 mM) and lysozyme (final concentration, 1 mg/mL), and incubated on ice for 30 min. Cells were then disrupted three times in an ultrasonicator and centrifuged for 30 min at 4 °C to precipitate cell debris. The resulting supernatant was purified by affinity chromatography using a Ni-NTA column (Qiagen), following manufacturer's instructions. Protein concentration was assessed using a BCA Protein Assay Kit (Pierce).
Synthesis and purification of lipid II and lipid II-Gly 1 . Lipid II was prepared by reacting undecaprenyl phosphate (Larodan), UDP-MurNAc-pentapeptide from Staphylococcus simulans, UDP-GlcNAc (Sigma) www.nature.com/scientificreports www.nature.com/scientificreports/ and membrane proteins of Micrococcus luteus as previously described 13 . Monoglycyl lipid II was synthesised by reacting lipid I with tRNA preparations, in the presence of enzymes FemX and GlyS, according to the method described by Schneider et al. 13 . Lipid intermediates were extracted from reaction mixtures with an equal volume of butanol/pyridine acetate (2∶1; vol∶vol; pH 4.2). Extracts were then purified by anion-exchange chromatography using a Hi-Trap DEAE FF-agarose column (Amersham Biosciences) by reverse-phase HPLC and eluted in a linear gradient from chloroform-methanol-water (2∶3∶1) to chloroform-methanol-300 mM ammonium bicarbonate (2∶3∶1). The fractions containing lipid species were identified by thin layer chromatography with chloroform-methanol-water-ammonia (88∶48∶10∶1) as solvent 40 . The concentration of purified lipids was calculated by measuring inorganic phosphates released after the treatment with perchloric acid, as described previously 41 .
FemA enzymatic activity assay. In order to compare the activity of wild-type FemA to selected FemA mutants, enzymatic reactions were performed as described previously 13 . Briefly, 100 µl reactions were prepared containing 2.5 nmol of lipid II-Gly 1 , 10 µg of glycyl-tRNA synthetase (GlyS), 25 µg of tRNA, 2 mM ATP and 50 nmol [ 14 C]-glycine in Tris buffer (100 mM Tris-HCl, 20 mM MgCl 2 , pH 7.5, and 0.8% Triton X-100). Then 2.7 µg of wild-type FemA or FemA mutant protein was added and the reaction mixtures were incubated for 30,60 or 90 minutes at 30 °C. Lipid intermediates were then extracted and analysed by thin layer chromatography, as described above. Finally, the amount of [ 14 C]-glycine transferred to lipid II-Gly 1 was quantified using phosphoimaging in a STORM system (GE Healthcare). Enzymatic assays were done in triplicate.
Identification of FemA residues possibly involved in tRNA binding. Identification of FemA residues which could bind glycyl-charged tRNA was performed using DP-Bind 28,29 (http://lcg.rit.albany.edu/dp-bind/), a sequence-based web server which predicts DNA/RNA binding domains in proteins based on biochemical properties of amino acids and evolutionary information. Probability maps were generated using PSI-BLAST position-specific scoring matrix (PSSM) and three distinct machine learning methods that use evolutionary information: support vector machine (PSSM-SVM), kernel logistic regression (PSSM-KLR), and penalized logistic regression (PSSM-PLR). FemA residues K180 and R181 were identified as possibly part of DNA-binding domains based on strict consensus between the three methods. SwissPdb viewer/Deep view (http://www.expasy. org/spdbv/) was used to evaluate the structure of FemA, using file 1LRZ (doi: 10.2210/pdb1LRZ/pdb) deposited in the RCSB PDB by Benson et al.

Data Availability
Source data are available from the corresponding author upon reasonable request.