Real Time Monitoring of Peptidoglycan Synthesis by Membrane-Reconstituted Penicillin 2 Binding Proteins 3 Víctor M

1 Real time monitoring of peptidoglycan synthesis by membrane-reconstituted penicillin 2 binding proteins 3 Víctor M. Hernández-Rocamora1*, Natalia Baranova2*, Katharina Peters1, Eefjan Breukink3, 4 Martin Loose2#, Waldemar Vollmer1# 5 6 1 Centre for Bacterial Cell Biology, Biosciences Institute, Newcastle University, Richardson 7 Road, Newcastle upon Tyne, NE2 4AX, UK. 8 2Institute for Science and Technology Austria (IST Austria), Klosterneuburg, Austria 9 3Membrane Biochemistry and Biophysics, Bijvoet Centre for Biomolecular Research, 10 University of Utrecht, Padualaan 8, 3584 Utrecht, The Netherlands. 11 12 * Contributed equally. 13 # Correspondence: [email protected]; [email protected]. 14 15

16 ABSTRACT 17 Peptidoglycan is an essential component of the bacterial cell envelope that surrounds the 18 cytoplasmic membrane to protect the cell from osmotic lysis. Important antibiotics such as β- 19 lactams and glycopeptides target peptidoglycan biosynthesis. Class A penicillin binding 20 proteins are bifunctional membrane-bound peptidoglycan synthases that polymerize glycan 21 chains and connect adjacent stem peptides by transpeptidation. How these enzymes work in 22 their physiological membrane environment is poorly understood. Here we developed a novel 23 FRET-based assay to follow in real time both reactions of class A PBPs reconstituted in 24 liposomes or supported lipid bilayers and we demonstrate this assay with PBP1B homologues 25 from Escherichia coli, Pseudomonas aeruginosa and Acinetobacter baumannii in the presence 26 or absence of their cognate lipoprotein activator. Our assay allows unravelling the mechanisms 27 of peptidoglycan synthesis in a lipid-bilayer environment and can be further developed to be 28 used for high throughput screening for new antimicrobials.

29 INTRODUCTION 30 Peptidoglycan (PG) is a major cell wall polymer in bacteria. It is composed of glycan strands 31 of alternating N-actetylglucosamine (GlcNAc) and N-acetylmuramic acid (MurNAc) residues 32 interconnected by short peptides. PG forms a continuous, mesh-like layer around the cell 33 membrane to protect the cell from bursting due to the turgor and to maintain cell shape 34 (Vollmer et al., 2008). The essentiality and conservation of PG in bacteria make peptidoglycan 35 metabolism an ideal target of antibiotics. 36 Class A penicillin-binding proteins (PBPs) are bifunctional PG synthases, which use 37 the precursor lipid II to polymerize glycan chains (glycosyltransferase reactions) and crosslink 38 peptides from adjacent chains by DD-transpeptidation (Goffin & Ghuysen, 1998). 39 Moenomycin inhibits the glycosyltransferase and -lactams the transpeptidase function of class 40 A PBPs (Sauvage & Terrak, 2016, Macheboeuf et al., 2006). In E. coli, PBP1A and PBP1B 41 account for a substantial proportion of the total cellular PG synthesis activity (Cho et al., 2016) 42 and they are tightly regulated by interactions with multiple proteins (Egan et al., 2015, Typas 43 et al., 2012, Egan et al., 2020, Egan et al., 2017), including the outer membrane anchored 44 activators LpoA and LpoB (Egan et al., 2018a, Typas et al., 2010, Jean et al., 2014). 45 Historically, in vitro PG synthesis assays have been crucial to decipher the biochemical 46 reactions involved in PG synthesis and determine the mode of action of antibiotics (Izaki et al., 47 1968). However, these studies were limited by the scarcity of lipid II substrate and the inability 48 to purify a sufficient quantity of active enzymes. Lipid II can now be synthesized chemically 49 (VanNieuwenhze et al., 2002, Schwartz et al., 2001, Ye et al., 2001) or semi-enzymatically 50 (Breukink et al., 2003, Egan et al., 2015), or isolated from cells with inactivated MurJ (Qiao et 51 al., 2017). Radioactive or fluorescent versions of lipid II are also available to study PG 52 synthesis in the test tube. However, there are several drawbacks with currently available PG 53 synthesis assays. First, most assays are end-point assays that rely on discrete sampling and 54 therefore do not provide real-time information about the enzymatic reaction. Second, some 55 assays involve measuring the consumption of lipid II or analysing the reaction products by 56 SDS-PAGE (Egan et al., 2015, Barrett et al., 2007, Qiao et al., 2014, Sjodt et al., 2018) or 57 HPLC after digestion with a muramidase (Bertsche et al., 2005, Born et al., 2006). These 58 laborious techniques make assays incompatible with high through-put screening and hinder the 59 determination of kinetic parameters. A simple, real-time assay with dansyl-labelled lipid II 60 substrate overcomes these problems but is limited to assay GTase reactions (Schwartz et al., 61 2001, Offant et al., 2010, Egan et al., 2015).

62 Recently two types of real-time TPase assays have been described. The first uses non- 63 natural mimics of TPase substrates such as the rotor-fluorogenic 470 D-lysine probe Rf470DL, 64 which increases its fluorescence emission upon incorporation into PG (Hsu et al., 2019). The 65 second assay monitors the release of D-Ala during transpeptidation in coupled enzymatic 66 reactions with D-amino acid oxidase, peroxidases and chromogenic or fluorogenic compounds 67 (Frere et al., 1976, Gutheil et al., 2000, Catherwood et al., 2020b). Coupled assays are often 68 limited in the choice of the reaction conditions, which in this case must be compatible with D- 69 amino acid oxidase activity. Hence, each of the current assays has its limitations and most 70 assays exclusively report on either the GTase or TPase activity, but not both activities at the 71 same time. 72 Another major drawback of many of the current assays is that they include detergents 73 and/or high concentration (up to 30%) of the organic solvent dimethyl sulfoxide (DMSO) to 74 maintain the PG synthases in solution (Offant et al., 2010, Biboy et al., 2013, Huang et al., 75 2013, Lebar et al., 2013, Qiao et al., 2014, Egan et al., 2015, Catherwood et al., 2020b). 76 However, both detergents and DMSO have been shown to affect the activity and interactions 77 of E. coli PBP1B (Egan & Vollmer, 2016). Importantly, a freely diffusing, detergent- 78 solubilised membrane enzyme has a very different environment compared to the situation in 79 the cell membrane where it contacts phospholipids and is confined in two dimensions (Gavutis 80 et al., 2006, Zhdanov & Höök, 2015). Here we sought to overcome the main limitations of 81 current PG synthesis assays. We established the sensitive Förster Resonance Energy Transfer 82 (FRET) detection technique for simultaneous monitoring of GTase and TPase reactions. The 83 real-time assay reports on PG synthesis in phospholipid vesicles or planar lipid bilayers. We 84 successfully applied this assay to several class A PBPs from pathogenic Gram-negative 85 bacteria, demonstrating its robustness and potential use in screening assays to identify PBP 86 inhibitors. 87 88 RESULTS 89 Real time assay for detergent-solubilised E. coli PBP1B 90 To develop a FRET-based real time assay for PG synthesis using fluorescently labelled lipid 91 II, we prepared lysine-type lipid II versions with high quantum yield probes, Atto550 (as FRET 92 donor) and Atto647n (as FRET acceptor), linked to position 3 (Figure 1 – figure supplement 93 1A-B) (Mohammadi et al., 2014, Egan et al., 2015). For assay development we used E. coli 94 PBP1B (PBP1BEc) (Egan et al., 2015, Bertsche et al., 2005, Biboy et al., 2013) solubilized 95 with Triton X-100 and a lipid-free version of its cognate outer membrane-anchored lipoprotein

96 activator LpoB (Typas et al., 2010, Egan et al., 2014, Egan et al., 2018a, Lupoli et al., 2014a, 97 Catherwood et al., 2020b). 98 PBP1BEc can utilize fluorescently labelled lipid II to polymerize long glycan chains 99 only when unlabelled lipid II is also present in the reaction (van't Veer, 2016). We therefore 100 included unlabelled meso-diaminopimelic acid (mDAP)-type lipid II into reactions of PBP1BEc 101 with lipid II-Atto550 and lipid II-Atto647n (Figure 1A). Both probes were incorporated into 102 the produced PG or glycan chains as indicated by SDS-PAGE analysis (Figure 1B, i). After the 103 reaction, fluorescence spectra taken at the excitation wavelength of the donor fluorophore

104 (Atto550, λabs=552 nm) showed a reduced donor emission intensity (λfl=580 nm) and an

105 increased emission of the acceptor fluorophore (Atto647n, λfl=665 nm) (Figure 1C, i) indicative 106 of FRET between the two fluorophores. Analysis of the fluorescence spectra allowed to 107 calculate FRET efficiencies which we found to be 29 ± 6 % (Figure 1D, Figure 1 – figure 108 supplement 2). Ampicillin, which inhibits the TPase, blocked the formation of cross-linked PG 109 (Figure 1B, ii) and reduced the FRET efficiency by a third (Figure 1C, ii, 1D). Moenomycin, 110 which blocks the GTase and, indirectly, TPase activities completely abolished the 111 incorporation of fluorescent lipid II and the associated signal (Figure 1B, iii, 1D) (Bertsche et 112 al., 2005). These results demonstrate that incorporation of the labelled probes into PG by 113 PBP1BEc, produces a signal that depends on the GTase and TPase activity, with the latter being 114 the major contributor. 115 Next, we monitored reactions in real time by measuring fluorescence emission of the

116 donor and acceptor fluorophores (FIdonor and FIacceptor, respectively) after excitation of the donor 117 (540 nm) in a microplate reader for 60 min (Figure 1E, Figure 1 – figure supplement 3A). The

118 ratio between both signals (FIacceptor/FIdonor) was used as indicative of FRET. Without LpoB, 119 FRET appeared after ~5 min and slowly increased until it plateaued after 50-60 min (Figure 120 1E, left panel). By contrast, reactions with LpoB(sol) showed an immediate and rapid increase 121 in FRET which reached the plateau after 10-20 minutes, consistent with faster PG synthesis 122 (Figure 1E, left panel). In agreement with the end-point analysis described above, we found no 123 FRET in samples containing moenomycin (Figure 1E, middle panel), and ampicillin generally 124 reduced the final fluorescence ratio level by ~3-fold (Figure 1E, middle panel). Analysis of 125 reaction products by SDS-PAGE also confirmed cross-linked PG was only produced in the 126 absence of antibiotics, while the presence of ampicillin still allowed the formation of glycan 127 chains (Figure 1 – figure supplement 3B). 128 The GTase reaction began after a lag phase, consistent with previously published data 129 (Schwartz et al., 2002, Egan et al., 2014), which is likely caused by a slower initiation of glycan

130 chain synthesis compared to the rate of polymerization. We measured relative reaction rates by 131 calculating the slope of FRET reaction curves during the linear raise in signal after the lag 132 phase (when present) and compared the slopes with or without activator. LpoB activated 133 PBP1B ~10 or ~20-fold in the absence or presence of ampicillin, respectively (Figure 1F), 134 which is comparable to the ~10-fold activation of the GTase rate by LpoB measured with 135 dansyl-lipid II (Egan et al., 2014, Egan et al., 2018a). 136 137 Intra-chain versus inter-chain FRET 138 Because ampicillin substantially reduced the FRET signal we hypothesized that FRET arises 139 mainly between fluorophores on different glycan chains of a cross-linked PG product (Figure 140 1A). To determine the relative contribution of intra-chain versus inter-chain FRET, we digested 141 PG produced in the presence of labelled lipid II with either the DD-endopeptidase MepM, 142 which cleaves crosslinks between glycan chains (Singh et al., 2015, Singh et al., 2012), or the 143 muramidase cellosyl, which cleaves the β-(1,3)-glycosydic bond between MurNAc and 144 GlcNAc producing muropeptides (structures 1, 2, and 3 in Figure 2C) (Rau et al., 2001) (Figure 145 2A-B). As a control, glycan chains produced by PBP1BEc in the presence of ampicillin were 146 also digested with both hydrolases. SDS-PAGE analysis confirmed that MepM substantially 147 reduced the amount of crosslinked PG in the samples while cellosyl digested the PG into 148 muropeptides (Figure 2A). Next, we measured the FRET efficiency after digestion. MepM 149 digestion had a negligible effect on the FRET efficiency of glycan chains produced in the 150 presence of ampicillin but reduced the FRET efficiency by ~2-fold for crosslinked-PG samples 151 (Figure 2B). This confirms that inter-chain FRET is a major contributor to the final FRET 152 signal. MepM did not reduce FRET efficiency to the same value as ampicillin, presumably 153 because of incomplete digestion of the labelled PG. Finally, cellosyl completely abolished 154 FRET for both glycan chains and cross-linked PG (Figure 2B). 155 To study more in detail the contribution of intra-chain FRET, we performed reactions 156 with only labelled lipid II (Figure 1E, right panel), where cross-linking is not possible. PBP1BEc 157 was unable to use lipid II-Atto550 and lipid II-Atto647n for polymerization in the absence of 158 LpoB (Figure 1E, Figure 1 – figure supplement 3B), confirming a previous study (van't Veer, 159 2016). In the presence of LpoB(sol), PBP1BEc produced short, non-crosslinked individual PG 160 chains (Figure 1E, Figure 1 – figure supplement 3B) that gave rise to a slow but large increase 161 in FRET (Figure 1E, right panel; Figure 1 – figure supplement 3A), indicating that 162 polymerisation of labelled lipid II in the absence of unlabelled lipid II occurred.

163 To confirm that the formation of peptide crosslinks is required to produce substantial 164 FRET in the absence of LpoB, we analysed the PG synthesised by PBP1BEc from radioactively 165 labelled lipid II and the two fluorescent lipid II analogues (Figure 2C-E). We monitored the 166 reaction at different time points by fluorescence spectroscopy (FRET measurements) and 167 digested aliquots with cellosyl before separating the resulting muropeptides by HPLC. The 168 monomers and cross-linked muropeptide dimers were quantified by scintillation counting using 169 an in-line radiation detector attached to the HPLC column (Figure 2C). FRET increased over 170 time and correlated well with the formation of cross-linked muropeptide dimers, but not the 171 rate of lipid II consumption (peak 2) (Figure 2D-E). Overall, we conclude that FRET can arise 172 from GTase activity alone (intra-chain FRET), but the overall the contribution from the TPase 173 activity (inter-chain FRET) is dominant. 174 175 FRET assay to monitor PG synthesis in liposomes 176 To establish the FRET assay for membrane-embedded PG synthases we reconstituted a version 177 of PBP1BEc with a single cysteine at the cytoplasmic N-terminus into liposomes prepared from 178 E. coli polar lipids (EcPL) (Figure 3 – figure supplement 1A). The liposome-reconstituted 179 PBP1BEc became accessible to a sulfhydryl-reactive fluorescent probe only after disrupting the 180 liposomes with detergent (Figure 3A), showing that virtually all PBP1B molecules were 181 oriented with the N-terminus inside the liposomes. This suggests that the large, extracellular 182 portion of PBP1BEc is not transferred through the membrane during the reconstitution into 183 liposomes (Rigaud & Lévy, 2003). Next, we reconstituted unmodified PBP1BEc and tested its 184 activity by adding radioactive lipid II. In contrast to the detergent-solubilized enzyme, the 185 liposome-reconstituted PBP1BEc required the absence of NaCl from the reaction buffer for 186 improved activity (Figure 3 – figure supplement 1B-E), suggesting that ionic strength affects 187 either the structure of PBP1BEc in the membrane, the properties of EcPL liposomes or the 188 delivery of lipid II into the liposomes. 189 We next aimed to adapt the FRET assay to study PG synthesis on liposomes (Figure 3, 190 Figure 3 – figure supplement 2). As PBP1BEc did not accept Atto550- or Atto647-derivatised 191 lipid II for GTase reactions in the absence of unlabelled lipid II (Figure 1E), we reconstituted 192 PBP1BEc in liposomes along both Atto-labelled substrates and initiated the reaction by adding 193 unlabelled lipid II (Figure 3B). PBP1BEc reaction rates in liposomes were slower than in the 194 presence of Triton X-100 for all conditions tested (compare curves on Figure 3C, measured at 195 37°C with the ones on Figure 1E, measured at 25 °C) and there was a longer lag time before 196 FRET started to increase (Figure 3C, left panel). Ampicillin or moenomycin blocked the

197 increase in FRET (Figure 3C, middle panel). For an unknown reason, the FRET signal with 198 moenomycin was initially higher than without moenomycin and then decreased to initial values 199 without moenomycin (Figure 3C, middle panel), independent of the class A PBP used (see 200 below) but not in empty liposomes (Figure 3 – figure supplement 3). LpoB(sol) produced a 201 ~10-fold increase in the initial slope, measured as explained above, (Figure 3D) and the 202 resulting final FRET was much higher (Figure 3C, left panel). Interestingly, in the presence of 203 ampicillin and LpoB(sol), FRET increased rapidly at the start of reactions, but then decreased 204 slowly, reaching a lower FRET value than in the presence of LpoB(sol) alone (without 205 ampicillin) (Figure 3C, middle panel). The decrease in FRET in the presence of ampicillin 206 suggests the spectroscopic properties of the incorporated probes change over time, presumably 207 by moving them further away from the lipid end of the growing glycan chains. Liposomes 208 without unlabelled lipid II produced a low FRET signal only in the presence of LpoB(sol) 209 (Figure 3C, right panel). The analysis of the final products by SDS-PAGE confirmed that both 210 Atto550 and Atto647n were incorporated into glycan chains or cross-linked peptidoglycan 211 during the reaction in liposomes (Figure 3C, right side, Figure 3 – figure supplement 2B). 212 In summary, we established a FRET-based assay that allows to monitor the activity of 213 membrane-reconstituted PBP1B in real time and showed that the FRET signal was sensitive to 214 the presence of PG synthesis inhibitors (moenomycin and ampicillin). 215 216 Activities of other membrane-bound class A PBPs 217 To demonstrate the usefulness of the FRET assay to study class A PBPs of potential therapeutic 218 interest, we next tested two PBP1B homologues from Gram-negative pathogens, Acinetobacter 219 baumannii (PBP1BAb) and Pseudomonas aeruginosa and PBP1BPa). We set up reactions in the 220 presence or absence of a soluble version of the lipoprotein activator LpoPPa(sol) for PBP1BPa 221 (Greene et al., 2018b). There is currently no reported activator of PBP1BAb, but next to the 222 gene encoding PBP1BAb we identified a hypothetical gene encoding a lipoprotein containing 223 two tetratricopeptide repeats (Uniprot code D0C5L6) (Figure 3 – figure supplement 4) which 224 we subsequently found to activate PBP1BAb (see below, Figure 3 – figure supplement 5). We 225 named this protein LpoPAb and purified a version without its lipid anchor, called LpoPAb(sol). 226 We were able to monitor PG synthesis activity by FRET for both PBPs in the presence or 227 absence of their (hypothetical) activators, using the Triton X-100-solubilized (Figure 3 – figure 228 supplements 6 and 7) or liposome-reconstituted proteins (Figure 3E-H, Figure 3 – figure 229 supplement 2C-D). Our experiments revealed differences in the activities and effect of 230 activators between both PBP1B-homolgoues which we discuss in the following paragraphs.

231 PBP1BAb showed GTase activity in the presence of Triton X-100 (Figure 3 – figure 232 supplement 5A) and was stimulated ~3.3-fold by LpoPAb(sol) (Figure 3 – figure supplement 233 5B); LpoPAb(sol) also accelerated the consumption of lipid II-Atto550 and glycan chain 234 polymerization (Figure 3 – figure supplement 5C). We measured a low FRET signal for PG 235 produced by the detergent-solubilised enzyme in the FRET assay (Figure 3 – figure supplement 236 6A) and poor production of cross-linked PG (Figure 3 – figure supplement 6C), unlike in the 237 case of the other PBPs. However, the liposome-reconstituted PBP1BAb displayed a higher 238 TPase activity than the detergent-solubilised enzyme (compare gels on Figure 3E, right panel 239 and Figure 3 – figure supplement 6C). In addition, the final FRET signal was substantially 240 higher in liposomes than in detergents (Figure 3E, Figure 3 – figure supplement 6A). 241 Moenomycin completely blocked FRET development, whilst ampicillin had a negligible effect 242 on the final FRET levels in detergents and only a small effect in liposomes (~1.2-fold 243 reduction), indicating that intra-chain FRET is the major contributor to FRET (Figure 3E; 244 Figure 3 – figure supplement 6A). LpoPAb(sol) stimulated PBP1BAb, with a higher effect in 245 detergents (12.3-fold increase) than liposomes (~2.5-fold increase) (Figure 3E, F; Figure 3 – 246 figure supplement 6A-B). 247 PBP1BPa displayed robust TPase activity in detergents and liposomes (Figure 3G, right 248 panel; Figure 3 – figure supplement 7C) and ampicillin reduced the final FRET signal by ~1.8- 249 fold in Triton X-100 and by ~1.5-fold in liposomes, indicating a substantial contribution of 250 inter-chain FRET to the FRET signal (Figure 3G, Figure 3 – figure supplement 7A). The 251 addition of LpoPPa(sol) resulted in an increase in the final FRET by ~2.2-fold in the membrane 252 and by ~2.1-fold in detergents (Figure 3G, Figure 3 – figure supplement 7A), accelerated initial 253 slopes by ~4.2-fold in the membrane and by ~11.5-fold in detergents (Figure 2H, Figure 2 – 254 figure supplement 7B); lipid II consumption was increased under both conditions (Figure 3G, 255 right panel; Figure 3 – figure supplement 7C). Overall, these results indicate that LpoPPa(sol) 256 stimulates both GTase and TPase activities in agreement with a recent report (Caveney et al., 257 2020). 258 259 PG synthesis on supported lipid bilayers 260 As we were able to successfully reconstitute active class A PBPs in membranes and monitor 261 their activity in real time, we next aimed to characterise the behaviour of these enzymes in the 262 membrane in more detail by reconstituting them on supported lipid bilayers (SLBs). SLBs are 263 phospholipid bilayers formed on top of a solid support, usually a glass surface and they allow

264 for studying the spatial organization of transmembrane proteins and their diffusion along the 265 membrane by fluorescence microscopy at high spatio-temporal resolution. 266 We optimized the reconstitution of PBP1BEc in SLBs formed with EcPL and used the 267 optimized buffer conditions for activity assays on liposomes. To support lateral diffusion and 268 also improve stability of the proteins incorporated into SLBs, we employed glass surfaces 269 coated with polyethylene glycol (PEG) end-functionalized with a short fatty acid (Roder et al., 270 2011) to anchor the EcPL bilayer (Figure 4A). We noticed a decrease in membrane diffusivity 271 and homogeneity at a high surface density of PBP1BEc (Figure 4 – figure supplement 1). To 272 prevent disturbing the SLB structure by the inserted protein we reduced the density of PBP1BEc 273 on SLBs from ~10-3 mol protein/mol lipid in liposomes to a range of 10-6 to 10-5 mol 274 protein/mol lipid. Using a fluorescently-labelled version of PBP1BEc reconstituted in SLBs, we 275 were able to track the diffusion of single PBP1B molecules in the plane of lipid membrane in 276 the presence or absence of substrate lipid II by TIRF microscopy (Figure 4B, 4D, Video 1). Ec 2 277 PBP1B diffused on these supported bilayers with an average Dcoef of 0.23±0.06 µm /s. 278 Addition of lipid II slowed down PBP1BEc diffusion (Figure 4C), resulting in a lower average 2 279 Dcoef of 0.10±0.06 µm /s. Upon addition of lipid II, we could not detect a prolonged confined 280 motion within particle tracks (Figure 4D), however the average length of displacements 281 between two sequential frames was reduced (Figure 4E). Thus we successfully reconstituted 282 diffusing PBP1BEc in SLBs and we observed that lipid II-binding slowed down the diffusion 283 of the synthase. 284 Next we wanted to confirm that PBP1BEc remained active to produce planar bilayer- 285 attached PG. We incubated SLBs containing PBP1BEc with radioactive lipid II and digested 286 any possible PG produced with a muramidase and analysed the digested material by HPLC. 287 Due to the low density and amount of PBP1BEc on each SLB chamber we expected a small 288 amount of PG product; hence, we included LpoB(sol) to boost the activity of PBP1BEc. Under 289 these conditions about 12% of the added radiolabelled lipid II was incorporated into PG after 290 an overnight incubation (Figure 4 – figure supplement 2A). However, products of both the 291 GTase and TPase activities of PBP1BEc were detected and these products were absent in the 292 presence of moenomycin (Figure 4 – figure supplement 2B). After overnight PG synthesis 293 reactions with radioactive lipid II, about 32% of the radioactivity remained in the membrane 294 fraction after washing (PG products and unused lipid II) and 68% was in the supernatant. The 295 analysis of the membrane and wash fractions by HPLC (Figure 4 – figure supplement 2C-D) 296 revealed that SLB-reconstituted PBP1BEc produced crosslinked PG while, importantly, the 297 wash fraction contained no PG products, confirming that the PG synthesis occurred on the

298 SLBs and this PG remained attached to the bilayer. The fraction of membrane-attached 299 radioactivity was almost the same (33%) when PBP1BEc was not present in the bilayer, 300 indicating that PBP1BEc did not affect lipid II-binding to the bilayer. 301 302 FRET assay on supported bilayers 303 Next, we adapted the FRET assay to SLBs and TIRF microscopy taking advantage of the 304 photostability and brightness of the Atto550 and Atto647n probes. Our aim was to visualize 305 PG synthesis by class A PBPs at high resolution as a first step towards understanding PG 306 synthesis at a single molecule level. We used a similar approach as for liposomes, where both 307 Atto550- and Atto647n-labelled lipid II were co-reconstituted with PBP1BEc on supported lipid 308 bilayers and PG synthesis was triggered by the addition of unlabelled lipid II (Figure 3A). To 309 measure any change in FRET due to PG synthesis, we took advantage of the fact that upon 310 photobleaching of the acceptor probe in a FRET pair, the emitted fluorescence intensity of the 311 donor increases as absorbed energy cannot be quenched by a nearby acceptor (Loose et al., 312 2011, Verveer et al., 2006). Indeed, we detected an increase in lipid II-Atto550 fluorescence 313 intensity upon photobleaching of the Atto647n probe after the addition unlabelled lipid II and 314 LpoB(sol), indicating the presence of FRET (Figure 5A, Figure 5 – figure supplement 1A). 315 When we bleached the acceptor at different time points of the reaction, we found the FRET 316 signal to increase after a lag phase of ~8 min. Importantly, there was no FRET increase in the 317 presence of ampicillin (Figure 5B, Figure 5 – figure supplement 1A, Video 2) or when a GTase- 318 defective PBP1BEc version (E233Q) was used (Figure 5C). In addition, the FRET signal was 319 abolished when the muramidase cellosyl was added after the PG synthesis reaction (Figure 320 5C). These results imply that the FRET signal detected by microscopy is primarily due to the 321 transpeptidase activity of PBP1BEc, in agreement with the results obtained on liposomes 322 (Figure 5C). 323 324 PG synthesised on supported lipid bilayers 325 As our experiments confirmed that the PG synthesized by PBP1BEc on SLBs remained attached 326 to the bilayer, we next analysed the lateral diffusion of lipid II-Atto647n and its products during 327 PG synthesis reactions. We first analysed the recovery of fluorescence intensity after 328 photobleaching to monitor the diffusion of lipid II-Atto647n during PG synthesis (Figure 5D). 329 Only when crosslinking was permitted (absence of ampicillin), the diffusion coefficient of lipid 330 II-Atto647n decreased 2 to 3-fold in a time-dependent manner. The time needed to reach the 331 minimum diffusivity value (~10 min) was similar to the lag detected in the increase of FRET

332 efficiency (Figure 5B). The fraction of immobile lipid II-Atto647n did not change significantly 333 in the presence or absence of ampicillin (13% ± 2% or 18% ± 6%, respectively, p-value = 0.15) 334 (Figure 5E), indicating that the crosslinked PG was still mobile under these conditions, but 335 diffused more slowly. We also compared the diffusion of lipid II-Atto647n during the PG 336 synthesis reaction with that of an AlexaFluor 488-labelled membrane-anchored peptide in the 337 presence or absence of ampicillin (Figure 5F, Figure 5 – figure supplement 2B). The inhibition 338 of TPase by ampicillin only affected the diffusivity of lipid II (2.9 ± 0.4µm2/s with ampicillin 339 and 0.67 ± 0.1 µm2/s without), while that of the lipid probe remained unchanged (1.6 ± 0.65 340 µm2/s with ampicillin and 1.94 ± 0.62 µm2/s without). This shows that the membrane fluidity 341 was not altered by the PG synthesis reaction and therefore was not the cause of the change in 342 lipid II diffusivity upon transpeptidation. As the immobile fraction of labelled lipid II did not 343 increase after PG synthesis and the diffusion was reduced only 2 to 3-fold, we concluded that 344 lipid II-Atto647n was incorporated into small groups of crosslinked glycan chains which can 345 still diffuse on the bilayer. 346 In summary, we report the incorporation of active PBP1BEc into supported lipid 347 bilayers, where we could track a decrease in the diffusion of the protein and its substrate during 348 PG synthesis reactions. Using this system we detected an increase in FRET upon initiation of 349 PG synthesis, only occurring when transpeptidation was not inhibited. 350 351 DISCUSSION 352 Although class A PBPs are membrane proteins and PG precursor lipid II is embedded in the 353 bilayer, few studies have provided information about the activity of these important enzymes 354 in a membrane environment. Here we developed a new assay that reports on PG synthesis by 355 these enzymes in detergents, on liposomes or on supported lipid bilayers. 356 357 Intra-chain vs inter-chain FRET 358 For all PBPs and conditions tested, FRET increased when only the GTase domain was active 359 (i.e. when FRET occurred between probes incorporated along the same strand), but the FRET 360 signal was always higher when transpeptidase was active (Figures 1, 2, 3 and Figure 3 – figure 361 supplements 6 and 7). For detergent-solubilised PBP1BEc, the FRET curve closely followed 362 the rate of the production of cross-linked PG as determined by HPLC analysis of the products 363 (Figure 2C-E), and the FRET of PBP1BEc-produced labelled PG decreased substantially upon 364 digestion with an endopeptidase (Figure 2A-B). These results indicate that inter-chain FRET 365 (arising from both fluorophores present on different, adjacent glycan chains) was a main

366 component of the total FRET signal. Why is this the case? FRET depends on the distance and 367 orientation of the two probes. It might be sterically unfavourable that two large Atto550 and 368 Atto647n containing lipid II molecules simultaneously occupy the donor and acceptor sites in 369 the GTase domain (van't Veer, 2016), preventing the incorporation of probes (and high FRET) 370 at successive subunits on a single glycan chain. Indeed, for all PBPs tested either in detergents 371 or liposomes, the incorporation of labelled lipid II into glycan chains was more efficient when 372 unlabelled lipid II was present and for most enzymes an activator was required to polymerize 373 glycan chains using labelled lipid II in the absence of unlabelled lipid II. We thus hypothesize 374 that the TPase activity brings glycan chains to close proximity, reducing the distance between 375 probes sufficiently to produce high levels of FRET (Figure 6). 376 377 Coupled reactions in class A PBPs and their activation 378 Our assay revealed the effect of Lpo activators on PBP1B analogues from 3 bacteria. 379 P. aeruginosa uses LpoP to stimulate its PBP1B (Greene et al., 2018b, Caveney et al., 2020). 380 Here, we identified an LpoP homologue in A. baumannii and showed that it stimulated its 381 cognate PBP1B. All three PBP1B homologues started the reaction after a lag phase, which was 382 abolished by the addition of the cognate activator (Egan et al., 2014, Caveney et al., 2020) 383 Considering the recently described role of PBP1B in repairing cell wall defects (Vigouroux et 384 al., 2020, Morè et al., 2019), the slow start in polymerisation and its acceleration by Lpo 385 activators could be an important mechanism to start PG synthesis at gaps in the PG layer where 386 the activators can contact the synthase. 387 To distinguish the effects of an activator on the TPase and GTase rates requires to use 388 different assays to measure GTase only or GTase/TPase, because ongoing glycan chain 389 polymerisation is required for transpeptidation to occur (Bertsche et al., 2005, Gray et al., 390 2015a). An elegant recent report (Catherwood et al., 2020b) described the use of a coupled D- 391 Ala release assay to determine the kinetic parameters of the TPase activity of PBP1BEc and the 392 effect of LpoB on this rate. Based on their observation that PBP1BEc had barely any TPase 393 activity in the absence of LpoB, the authors concluded that the LpoB-mediated TPase 394 activation explains the essentiality of LpoB for PBP1B function in the cell (Catherwood et al., 395 2020b). However, the assay used an enzyme concentration that is too low to support GTase 396 activity in the absence of LpoB, as demonstrated previously (Pazos et al., 2018b, Muller et al., 397 2007b). Therefore the essentiality of LpoB can be readily explained by its primary effect, the 398 >10-fold stimulation of the GTase rate (Egan et al., 2014). Our results provide an alternative 399 explanation for PBP1BEc essentiality. Activation by LpoB was much more needed when

400 PBP1B was embedded in the membranes of liposomes and supported bilayers, compared to 401 detergent-solubilized enzyme, supporting the idea that cellular PBP1B strictly requires LpoB 402 for GTase activity. In vitro, LpoB also stimulated the TPase causing PBP1BEc to produce a 403 hyper-crosslinked PG (Typas et al., 2010, Egan et al., 2018a) and the same was observed for 404 LpoPPa and PBP1BPa (Caveney et al., 2020). The GTase and TPase contribute both to the signal 405 in our FRET assay and the relative contribution of intra-chain FRET (due to the GTase) and 406 inter-chain FRET (due to the TPase) can be modified by an activator that enable the 407 incorporation of two adjacent probe molecules on the same glycan chain. Therefore, to untangle 408 the effects of activators on each of the activities requires a single quantitative model accounting 409 for the GTase and TPase rates and including parameters for the initiation, elongation and 410 termination of glycan chain synthesis of membrane embedded enzymes. Currently, such a 411 model is not available and our assay could help to develop such a model in the future. 412 413 Class A PBP activities in the membrane 414 Remarkably, we found slower reaction rates in liposomes than in detergents for all enzymes 415 tested. Several possible factors can explain this, including a slow incorporation of the added 416 unlabelled lipid II into liposomes, a limited capacity of the liposomes to incorporate the 417 unlabelled lipid II or the accumulation of the undecaprenyl pyrophosphate by-product that has 418 been showed to inhibit PBP1B activity (Hernández-Rocamora et al., 2018). None of these 419 factors should change in the presence of LpoB. Hence we favour the alternative explanation 420 that the membrane-embedding of PBP1B hinders lipid II binding, slowing down the reaction. 421 Remarkably, PBP1BAb showed higher TPase activity in liposomes than in detergents. This 422 observation highlights again that detergents can affect the activity of membrane proteins and 423 that experimental conditions in PG synthesis assays should be as close as possible at the 424 physiological conditions. 425 426 Towards single-molecule PG synthesis 427 We also adapted the FRET assay to supported lipid bilayers and super resolution 428 microscopy to study how PBP1BEc polymerizes PG on SLBs (Figure 5). As with the liposome 429 assays, we detected an increase in FRET signal upon triggering PG synthesis that correlated 430 with transpeptidation. Importantly we could follow the diffusion of the reaction products, 431 which indicates that PBP1BEc does not completely cover the surfaces with a layer of PG but 432 instead produced smaller patches of cross-linked glycan chains. We attribute this to the fact 433 that PBP1BEc was reconstituted at a very low density in order to ensure the homogeneity and

434 stability of the SLBs. Remarkably, we detected a reduction of PBP1BEc diffusivity in the 435 presence of lipid II (Figure 4). Previous in vivo single-molecule tracking of fluorescent-protein 436 tagged class A PBPs reported the presence of two populations of molecules, a fast diffusing 437 one and an almost immobile one with a near-zero diffusing rate which was assumed to be the 438 active population (Cho et al., 2016, Lee et al., 2016, Vigouroux et al., 2020). Our result 439 supports this interpretation, although more experiments are required to further explore this 440 point. 441 Several real time methods to study PG synthesis in vitro are described in the literature. 442 However, most of these report on either the GTase or TPase reaction, but not both at the same 443 time, and most available methods are not applicable to the membrane. The scintillation 444 proximity assay by Kumar et al. reports on PG production in a membrane environment and in 445 real time, but it is rather crude in that it uses membrane extract instead of purified protein and 446 relies on the presence of lipid II synthesizing enzymes present in the extract (Kumar et al., 447 2014). Moreover, it uses radioactivity detection and is not amenable to microscopy, in contrast 448 to methods based on fluorescently-labelled substrates. An important advantage of our new 449 assay over other real-time PG synthesis assays is that it uses natural substrates for 450 transpeptidation, i.e. nascent glycan strands, instead of mimics of the pentapeptide, and its 451 ability to measure the activities in a natural lipid environment. 452 Our new FRET assay can potentially be adopted to assay PG synthases in the presence 453 of interacting proteins, for example monofunctional class B PBPs in the presence of 454 monofunctional GTases (cognate SEDS proteins or Mtg proteins) or interacting class A PBPs 455 (Meeske et al., 2016, Bertsche et al., 2006, Sjodt et al., 2020, Derouaux et al., 2008, Banzhaf 456 et al., 2012, Sjodt et al., 2018, Taguchi et al., 2019). In addition, our assay has the potential to 457 be adopted to high throughput screening for new antimicrobials. 458

459 MATERIALS AND METHODS 460 Chemicals 461 [14C]GlcNAc-labelled lipid II and the lysine or mDAP forms of lipid II were prepared as 462 published (Breukink et al., 2003, Bertsche et al., 2005). Lipid II-Atto550 and Lipid II-Atto647n 463 were prepared from the lysine form of lipid II, prepared as described previously (Egan et al., 464 2015), and Atto550-alkyne or Atto647n-alkyne (Atto tec, Germany) in two steps: (1) 465 conversion of lysine form of lipid II to azidolysine form and (2) labelling of azidolysine lipid 466 II via click-chemistry. The protocol is extensively detailed elsewhere (Mohammadi et al.,

467 2014). The advantage of using this methodology over directly attaching the probes to the amine 468 group is the higher yield of click chemistry reactions, allowing the use of a smaller excess of 469 the reactive florescent probes (van't Veer, 2016). All lipid II variants were kept in 2:1 470 chloroform:methanol at -20 °C. Before enzymatic assays, the required amounts of lipid II were 471 dried in a speed-vac and resuspended in water (for assays in detergents) or the appropriate 472 buffer (for liposome and SLB assays). Polar lipid extract from E. coli (EcPL), 1,2-dioleoyl-sn- 473 glycero-3-phosphocholine (DOPC), 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-(1'-rac- 474 glycerol) (POPG) and tetraoleoyl cardiolipin (TOCL) were obtained from Avanti Polar Lipids 475 (USA). Lipids were resuspended in chloroform:methanol (2:1) at a concentration of 20 g/L, 476 aliquoted and stored at -20°C. Triton X-100, ampicillin, phenylmethylsulfonyl fluoride 477 (PMSF), protease inhibitor cocktail (PIC) and β-mercaptoethanol were from Merck. n-dodecyl- 478 beta-D-maltopyranoside (DDM) was purchased from Anatrace (USA). Moenomycin was 479 purchased from Hoechst, Germany. All other chemicals were from Merck. 480 481 Cloning 482 Construction of overexpression vector pKPWV1B – The plasmid pKPWV1B was constructed 483 for overexpression of full-length A. baumannii PBP1B (PBP1BAb: aa 1-798) with a cleavable

484 N-terminal oligo-histidine tag (His6 tag). Therefore, the gene mrcB was amplified using the 485 Phusion high fidelity DNA polymerase and the oligonucleotides PBP1B.Acineto-NdeI_f and 486 PBP1B.Acineto-BamHI_r and genomic DNA of A. baumannii 19606 (ATCC) as template. The 487 resulting PCR fragment and the Plasmid DNA of the overexpression vector pET28a(+) 488 (Novagen) were digested with NdeI and BamHI, ligated and transformed into chemical 489 competent E. coli DH5α cells with kanamycin selection. Plasmid DNA of transformants was 490 isolated and send for sequencing using following oligonucleotides: 491 Seq1_rev_PBP1B_Acineto, Seq2_fwd_PBP1B_Acineto, Seq3_fwd_PBP1B_Acineto, 492 Seq4_fwd_PBP1B_Acineto. 493 Construction of overexpression vector pKPWVLpoP – The sequence of the hypothetical 494 PBP1B activator of Acinetobacter baumannii 19606 (LpoPAb: NCBI reference number: 495 WP_000913437.1) contains a TPR fold and was found by blast analysis through its homology 496 to Pseudomonas aeruginosa LpoP (30% identity). The plasmid pKPWVLpoP was purchased 497 from the company GenScript. The gene was synthesized without the first 51 nucleotides 498 (encoding the 17 amino acids of the signal peptide) and with codon optimization for 499 overexpression in Escherichia coli. The codon optimized gene was subcloned in the

500 overexpression vector pET28a(+) using the cloning sites NdeI and BamHI enabling the 501 overexpression of the protein with an N-terminal oligo-histidine tag. 502 MGC-64PBP1B-his C777S/C795S – This fusion protein contains PBP1B with the substitution 503 of the N-terminal cytoplasmic tail for residues MGC and the addition of a hexahistine tag at 504 the C-terminus. To obtain this construct, the regions coding for aminoacids 64 to 844 of PBP1B 505 were amplified from genomic DNA using oligonucleotides PBP1B-MGC-F and PBP1B- 506 CtermH-R. The resulting product was cloned into pET28a+ vector (EMD Biosciences) after 507 digestion with NcoI and XhoI. C777S and C795S mutations were introduced using the 508 QuikChange Lightning mutagenesis kit (Agilent) using oligonucleotide primers C777S-D, 509 C777S-C, C795S-D and C795S-C The resulting plasmid was called pMGCPBP1BCS1CS2. 510 511 Purification and labelling of proteins 512 The following proteins were purified following published protocols: PBP1BEc 513 (Bertsche et al., 2006), LpoB(sol) (Egan et al., 2014), PBP1BPa (Caveney et al., 2020), 514 LpoPPa(sol) (Caveney et al., 2020), MepM (Singh et al., 2012). All chromatographic steps were 515 performed using an AKTA PrimePlus system (GE Healthcare). 516 E. coli PBP1B – The protein was expressed as a fusion with an N-terminal hexahistidine tag in 517 E. coli BL21(DE3) pDML924 grown in 4 L of autoinduction medium (LB medium 518 supplemented with 0.5% glycerol, 0.05% glucose, and 0.2% α-lactose) containing kanamycin 519 at 30 °C for ~16h. Cells were harvested by centrifugation (10,000 × g, 15 min, 4 °C) and the 520 pellet resuspended in 80 mL of buffer I (25 mM Tris-HCl, 1 M NaCl, 1 mM EGTA, 10% 521 glycerol, pH 7.5) supplemented with 1× protease inhibitor cocktail (PIC, Sigma-Aldrich), 100 522 µM phenylmethylsulfonyl fluoride (PMSF, Sigma-Aldrich) and DNase I. After disruption by 523 sonication on ice, membrane fraction was pelleted by centrifugation (130,000 × g for 1 h at 524 4 °C) and resuspended in buffer II (25 mM Tris-HCl, 1 M NaCl, 10% glycerol, 2% Triton X- 525 100, pH 7.5) by stirring at 4 °C for 24 h. Extracted membranes were separated from insoluble 526 debris by centrifugation (130,000 × g for 1 h at 4 °C) and incubated for 2h with 4 mL of Ni2+- 527 NTA beads (Novagen) equilibrated in buffer III (25 mM Tris-HCl, 1 M NaCl, 20 mM 528 imidazole, 10% glycerol, pH 7.5). Beads were washed 10 times with 10 mL of buffer III and 529 the protein was eluted with 3 mL buffer IV (25 mM Tris-HCl, 0.5 M NaCl, 20 mM imidazole, 530 10% glycerol, pH 7.5). His-PBP1B containing fractions were pooled and treated with 2 U/mL 531 of thrombin (Novagen) for 20 h at 4 °C during dialysis against dialysis buffer I (25 mM Tris- 532 HCl, 0.5 M NaCl, 10% glycerol, pH 7.5). Protein was then dialysed in preparation for ion

533 exchange chromatography, first against dialysis buffer II (20 mM sodium acetate, 0.5 M NaCl, 534 10% glycerol, pH 5.0); then against dialysis buffer II with 300 mM NaCl; and finally against 535 dialysis buffer II with 100 mM NaCl. Finally, the sample was applied to a 1 mL HiTrap SP 536 column (GE Healthcare) equilibrated in buffer A (20 mM sodium acetate, 100 mM NaCl, 10% 537 glycerol, 0.05% reduced Triton X-100, pH 5.0). The protein was eluted with a gradient from 0 538 to 100% buffer B (as A, with 2 M NaCl) over 14 mL PBP1B-containing fractions were pooled 539 and dialysed against storage buffer (20 mM sodium acetate, 500 mM NaCl, 10% glycerol, pH 540 5.0) and stored at −80 °C. 541 A. baumannii 19606 PBP1B – The protein was expressed in E. coli BL21 (DE3) freshly 542 transformed with plasmid pKPWV1B using the same protocol as PBP1BEc. Cells were 543 harvested by centrifugation (6,200 × g for 15 min at 4 °C) and resuspended in 120 mL of Ab 544 PBP1B buffer I (20 mM NaOH/H3PO4, 1 M NaCl, 1 mM EGTA, pH 6.0) supplemented with 545 DNase I, PIC (1:1,000 dilution) and 100 µM PMSF. After disruption by sonication on ice, the 546 membrane fraction was pelleted by centrifugation (130,000 × g for 1 h at 4 °C) and resuspended Ab 547 in PBP1B extraction buffer (20 mM NaOH/H3PO4, 1 M NaCl, 10% glycerol, 2% Triton X- 548 100, pH 6.0) supplemented with PIC and PMSF by stirring at 4 °C for 16 h. Extracted 549 membranes were separated from insoluble debris by centrifugation (130,000 × g for 1 h at 4 °C) 550 and incubated with 4 mL of Ni2+-NTA beads equilibrated in PBP1BAb extraction buffer 551 containing 15 mM imidazole. Beads were washed 10 times with 10 mL of PBP1BAb wash buffer

552 (20 mM NaOH/H3PO4, 10% Glycerol, 0.2% Triton X-100, 1M NaCl, 15 mM Imidazole, pH 553 6.0) and the protein was eluted with 3 mL buffer IV PBP1BAb elution buffer (20 mM

554 NaOH/H3PO4, 10% Glycerol, 0.2% Triton X-100, 1 M NaCl, 400 mM Imidazole, pH 6.0). 555 PBP1BAb-containing fractions were pooled and dialyzed in preparation for ion exchange 556 chromatography, first against PBP1BAb dialysis buffer I (20 mM sodium acetate, 1 M NaCl, 557 10% glycerol, pH 5.0), then against PBP1BAb dialysis buffer II (20 mM sodium acetate, 300 558 mM NaCl, 10% glycerol, pH 5.0) and finally against PBP1BAb dialysis buffer III (10 mM 559 sodium acetate, 100 mM NaCl, 10% glycerol, pH 5.0). The sample was centrifuged for 1 h at 560 130,000 × g and 4 °C and the supernatant was applied to a 5 mL HiTrap SP HP column 561 equilibrated in PBP1BAb buffer A (20 mM sodium acetate, 100 mM NaCl, 10% glycerol, 0.2% 562 Triton X-100, pH 5.0). The protein was eluted from 0 to 100% PBP1BAb buffer B (20 mM 563 sodium acetate, 2 M NaCl, 10% glycerol, 0.2% Triton X-100, pH 5.0) over 70 mL. PBP1BAb- 564 containing fractions were pooled and dialysed against PBP1BAb storage buffer (10 mM sodium 565 acetate, 500 mM NaCl, 0.2% Triton X-100, 20% glycerol, pH 5.0) and stored at -80 °C.

566 P. aeruginosa PBP1B – The protein was expressed on E. coli BL21(DE3) freshly transformed 567 with plasmid pAJFE52 which encodes PBP1BPa as a fusion with an N-terminal hexahistidine 568 tag in E. coli BL21(DE3). Cells were grown in 4 L of LB at 30 °C and expression was induced Pa 569 for 3 h with 1 mM IPTG when the culture reached an OD578 of 0.6. PBP1B was extracted and 570 purified using the same protocol as for E. coli PBP1B with the exception that only 2 mL of 571 Ni2+ beads were used. 572 MGC-64PBP1B-his C777S/C795S – This protein was expressed in E. coli BL21(DE3) freshly 573 transformed with plasmid pMGCPBP1BCS1CS2 and subsequently purified using the same 574 protocol as for the WT protein, except for the addition of 1 mM TCEP to all purification 575 buffers. The protein was labelled with Dy647-maleimide probe (Dyomics, Germany) following 576 instructions from the manufacturer. Briefly, 10.2 µM protein was incubated with 100 µM probe 577 and 0.5 mM TCEP for ~20 h at 4 °C and free probe was removed by desalting using a 5 mL 578 HiTrap desalting column (GE Healthcare). 579 LpoB(sol) – The protein was expressed on E. coli BL21(DE3) transformed with pET28His-

580 LpoB(sol). Cells were grown in 1.5 L of LB plus kanamycin at 30 °C to an OD578 of 0.4–0.6 581 and expression was induced with 1 mM of IPTG for 3 h at 30 °C. Cells were pelleted and

582 resuspended in buffer I (25 mM Tris-HCl, 10 mM MgCl2, 500 mM NaCl, 20 mM imidazole, 583 10% glycerol, pH 7.5) plus DNase, PIC and PMSF. Cells were disrupted by sonication on ice 584 and centrifuged (130,000 × g, 1 h, 4 °C) to remove debris. The supernatant was applied to a 585 5 mL HisTrap HP column (GE Healthcare) equilibrated in buffer I. After washing with buffer 586 I, the protein was eluted with a stepwise gradient with buffer II (25 mM Tris-HCl, 10 mM

587 MgCl2, 500 mM NaCl, 400 mM imidazole, 10% glycerol, pH 7.5). Fractions containing the 588 protein were pooled and the His-tag was removed by addition of 2 U/mL of thrombin while 589 dialysing against buffer IEX-A (20 mM Tris-HCl, 1000 mM NaCl, 10% glycerol, pH 8.3). 590 Digested protein was applied to a 5 mL HiTrap Q HP column (GE Healthcare) at 0.5 mL/min. 591 LpoB(sol) was collected in the flow through, concentrated and applied to size exclusion on a 592 Superdex200 HiLoad 16/600 column (GE Healthcare) at 1 mL/min in a buffer containing 593 25 mM HEPES-NaOH, 1 M NaCl, 10% glycerol at pH 7.5. Finally, the protein was dialysed 594 against storage buffer (25 mM HEPES-NaOH, 200 mM NaCl, 10% glycerol at pH 7.5) and 595 stored at −80 °C. 596 A. baumannii 19606 LpoP(sol) – The protein was expressed on E. coli BL21(DE3) transformed 597 with plasmid pKPWVLpoP. Cells were grown over night at 30 °C in 4 L of autoinduction 598 medium. Cells were pelleted by centrifugation (6,200 × g for 15 min at 4 °C) and resuspended

599 in 80 mL of buffer I (25 mM Tris/HCl, 10 mM MgCl2, 1 M NaCl, 20 mM Imidazole, pH 7.5)

600 supplemented with DNase I, PIC (1:1,000 dilution) and 100 µM PMSF. Cells were disrupted 601 by sonication on ice and centrifuged (130,000 × g for 1 h at and 4 °C) to removed debris. The 602 supernatant was incubated for 1h with 6 mL Ni-NTA beads preequilibrated in buffer I at 4 °C 603 with gentle stirring. The resin was split in 2 columns, each washed 10 times with 5 mL wash

604 buffer (25 mM Tris/HCl, 10 mM MgCl2, 1 M NaCl, 20 mM Imidazole, pH 7.5) and the protein

605 was eluted 7 times with 2 mL of elution buffer (25 mM Tris/HCl, 10 mM MgCl2, 1 M NaCl, 606 400 mM Imidazole, pH 7.5). The best fractions according to SDS-PAGE analysis were pooled 607 and dialyzed stepwise against increasing percentage of dialysis buffer I (25 mM

608 HEPES/NaOH, 10 mM MgCl2, 200 mM NaCl, 10% glycerol, pH 7.5). Thrombin (9 units) was

609 added to the protein to cleave the N-terminal His6 tag over night at 4 °C. The successful

610 cleavage of the N-terminal His6 tag was confirmed by SDS-PAGE. The protein was diluted 2×

611 with 25 mM HEPES/NaOH, 10 mM MgCl2, 10% glycerol, pH 7.5 to reduce the amount of 612 NaCl down to 100 mM. The protein was applied to a 5 mL HiTrap SP HP column and washed

613 with buffer A (25 mM HEPES/NaOH, 10 mM MgCl2, 100 mM NaCl, 10% glycerol, pH 7.5). 614 The protein was then eluted with a gradient of 100 mM to 1 M NaCl over 50 mL at 1 mL/min

615 using increasing percentage of buffer B (25 mM HEPES/NaOH, 10 mM MgCl2, 1 M NaCl, 616 10% glycerol, pH 7.5). Fractions were collected and analysed by SDS-PAGE. The best 617 fractions were pooled, dialysed against 25 mM HEPES/NaOH, 200 mM NaCl, 10% Glycerol,

618 10 mM MgCl2, pH 7.5 and the protein were stored at -80 °C. 619 P. aeruginosa LpoP(sol) – The protein was expressed on E. coli BL21(DE3) freshly Pa 620 transformed with from plasmid pAJFE57, encoding His6-LpoP (sol). Cells were grown on 1.5

621 L LB at 30ºC to an OD578 of 0.5 and expression was induced for 3h by addition of 1 mM IPTG. 622 After harvesting, cells were resuspended in 80 mL of 25 mM Tris-HCl, 500 mM NaCl, 20 mM 623 imidazole, 10% glycerol at pH 7.5. After addition of PIC and 100 µM PMSF, cells were 624 disrupted by sonication on ice. Debris was removed by centrifugation (130,000 × g, 1 h, 4 °C) 625 and the supernatant was applied to a 5 mL HisTrap column equilibrated in resuspension buffer. 626 After washing with 25 mM Tris-HCl, 1 M NaCl, 40 mM imidazole, 10% glycerol at pH 7.5, 627 protein was eluted with 25 mM Tris-HCl, 500 mM NaCl, 400 mM imidazole, 10% glycerol at 628 pH 7.5. Fractions containing His-LpoPPa(sol) were pooled and the His-tag was removed by 629 addition of 4 U/mL of thrombin while dialysing against 20 mM Tris-HCl, 200 mM NaCl, 10% 630 glycerol at pH 7.5 for 20 h at 4 ºC. The sample was concentrated and further purified by size 631 exclusion column chromatography at 0.8 mL/min using a HiLoad 16/600 Superdex 200 column 632 equilibrated in 20 mM Hepes-NaOH, 200 mM NaCl, 10% glycerol at pH 7.5. LpoPPa- 633 containing fractions were pooled, concentrated, aliquoted and stored at -80ºC.

634 635 PG synthesis assays in the presence of detergents 636 In vitro peptidoglycan synthesis assay using radiolabelled lipid II in detergents – To assay the 637 in vitro PG synthesis activity of PBP1BEc with radiolabelled lipid II substrate in the presence 638 of detergent we used a previously published assay (Banzhaf et al., 2012, Biboy et al., 2013).

639 Final reactions included 10 mM HEPES/NaOH pH 7.5, 150 mM NaCl, 10 mM MgCl2 and 0.05 640 % Triton X-100. The concentration of PBP1BEc was 0.5 µM. Reactions were carried out for 1 641 h at 37°C. Reactions were stopped by boiling for 5 min. Digestion with cellosyl, reduction with 642 sodium borohydride and analysis by HPLC were performed as described (Biboy et al., 2013). 643 FRET-based in vitro peptidoglycan synthesis assay in detergents – For assays in detergents,

644 samples contained 50 mM HEPES/NaOH pH 7.5, 150 mM NaCl, 10 mM MgCl2, and 0.05% 645 Triton X-100 in a final volume of 50 µL. PBP1BEc, PBP1BAb or PBP1BPa were added at a 646 concentration of 0.5 µM. When indicated, activators LpoB(sol), or LpoPAb(sol), or LpoPPa(sol) 647 were added at a concentration of 2 µM. Reactions were started by the addition of an equimolar 648 mix of lipid II, lipid II-Atto550 and lipid II-Atto647n, each at 5 µM and monitored by 649 measuring fluorescence using a Clariostar plate reader (BMG Labtech, Germany) with 650 excitation at 540 nm and emission measurements at 590 nm and 680 nm. Reactions were 651 incubated at the indicated temperature for 60 or 90 min. After the reaction emission spectra 652 from 550 to 740 nm were taken in the same plate reader with excitation at 522 nm. When 653 indicated ampicillin was added at 1 mM and moenomycin was added at 50 µM. After plate 654 reader measurements, reactions were stopped by boiling for 5 min, vacuum-dried using a 655 speed-vac desiccator and analysed by Tris-Tricine SDS-PAGE as previously described (Van't 656 Veer et al., 2016). 657 FRET reactions in the presence of radiolabelled lipid II described in Figure 1E-F were 658 performed using the same buffer and substrate and enzyme concentrations as for the plate 659 reader assay but in a final volume of 350 µL. Samples were incubated at 25 °C with shaking 660 using an Eppendorf Thermomixer. 50 µL aliquots were taken out at the indicated times and 661 reactions were stopped by addition of 100 µM moenomycin. Samples were then transferred to 662 a 96-well plate to measure FRET as described above. Finally, samples were transferred back 663 to Eppendorf tubes, digested with cellosyl and reduced with sodium borohydride as previously 664 described (Biboy et al., 2013). 665 Analysis of FRET reaction curves – Reaction curves were obtained by calculating the ratio 666 between fluorescence intensity at 680 nm and 590 nm monitored at every well. This maximises

667 the amount of information captured from the change in the spectrum due to FRET and 668 normalises the intensity removing any non-specific jumps in the signal due to bubbles in the 669 reaction well or lamp instability. The slope of reactions curves obtained by the FRET assay 670 was calculated when the ratio started to raise, avoiding the lag phase when present. Only the 671 linear phase of each curve was used. For example, for PBP1BEc in detergents, slopes were 672 calculated from 10 to 15 min in the absence of LpoB and within the first minute in the presence 673 of activator. To compare our results with prior reports, we report the fold-change in the slope 674 in the presence of the corresponding Lpo activator, i.e. the ratio between the slope in a condition 675 with activator and the slope at the same condition without activator. 676 Determination of FRET efficiency – For the determination of FRET efficiency, reactions were 677 prepared in the same conditions as for the plate reader assays but they were incubated at 37°C 678 for 1 h in Eppendorf tubes instead, and boiled for 5 min afterwards. For every antibiotic 679 condition, 4 samples were prepared: Sample 1 (DA reaction) contained 5 µM each of unlabelled 680 lipid II, lipid II-Atto550, lipid II-Atto647n; sample 2 (D reaction) contained 10 µM unlabelled 681 lipid II plus 5 µM lipid II-Atto550; sample 3 (A reaction) contained 10 µM unlabelled lipid II 682 plus 5 µM lipid II-Atto647n; and sample 4 (BG reaction) contained 15 µM unlabelled lipid II. 683 For digestion with hydrolases, 50 µL of the PG synthesis reactions were prepared as described 684 above and split into three aliquots. Either 5 µM MepM, 0.05 mg/mL cellosyl, or buffer were 685 added to a final volume of 20 µL and samples were incubated overnight at 37°C and boiled for 686 5 min to stop reactions. 687 Samples were measured in Cary Varian fluorimeter using a 1.5 mm light-path quartz 688 cuvette. For samples 1, 3 and 4, two spectra were measured, one with excitation at 552 nm 퐷 689 (휆푒푥) and emission collected from 560 to 750 nm (ds spectrum) and another one with excitation 퐴 690 at 650 nm (휆푒푥) and emission collected from 660 to 750 nm (as spectrum). For sample 2, only 691 the ds spectrum was measured. All spectra were taken with 5 nm slits for emission and 692 excitation at the same detector voltage settings (850 V).

693 FRET efficiency (E) was calculated according to the (ratio)A method described in

694 (Vámosi & Clegg, 1998). Briefly, (ratio)A is a normalised measure of the enhancement of the 695 acceptor emission due to FRET,

퐷 퐷 퐷 퐴 퐷 퐷 퐹퐴(휆푒푥,휆푒푚) [휖퐷(휆푒푥)퐸+휖퐴(휆푒푥)]훷 (휆푒푚) 휖퐷(휆푒푥)퐸+휖퐴(휆푒푥) 696 (ratio)A = 퐴 = 퐴 퐴 = 퐴 (1) 퐹퐴(휆푒푥,휆푒푚) 휖퐴(휆푒푥)훷 (휆푒푚) 휖퐴(휆푒푥) 퐴 퐷 697 where 훷 (휆푒푚) is a shape function of the acceptor emission spectrum, 퐹퐴(휆푒푥, 휆푒푚) is the 퐷 퐴 698 emission of the acceptor (only the acceptor) when excited at 휆푒푥 and 퐹퐴(휆푒푥, 휆푒푚) is the 퐷 699 emission of the acceptor when excited at 휆푒푥, both in the sample containing both donor and

퐷 700 acceptor. FRET efficiency is normalised by the extinction coefficients of the donor at 휆푒푥 퐷 -1 -1 퐷 퐴 퐷 -1 - 701 (휖퐷(휆푒푥) = 120000 M cm ) and of the acceptor at both 휆푒푥 and 휆푒푥 (휖퐴(휆푒푥) = 6000 M cm 1 퐴 -1 -1 702 , and 휖퐴(휆푒푥) = 150000 M cm , respectively). 푑푠 푑푠 703 In order to calculate (ratio)A from the ds spectrum, 3 spectral contributions (훿 , 훼 and 704 훽푑푠) were fitted in the ds spectra, 퐹푑푠(휆), and 2 spectral contributions (훼푎푠 and 훽푎푠) were 705 fitted in the as spectra, 퐹푎푠(휆): 푑푠 푑푠 푑푠 푑푠 푑푠 푑푠 푑푠 706 퐹 (휆) = 훿 퐹퐷푟푒푓(휆) + 훼 퐹퐴푟푒푓(휆) + 훽 퐹퐵푟푒푓(휆) (2) 푎푠 푎푠 푎푠 푎푠 푎푠 707 퐹 (휆) = 훼 퐹퐴푟푒푓(휆) + 훽 퐹퐵푟푒푓(휆) (3) 푑푠 708 were 퐹퐷푟푒푓(휆) is the background-free spectra from the donor-only reference sample exited at 퐷 푑푠 709 휆푒푥; 퐹퐴푟푒푓(휆) is the background-free spectra from the acceptor-only reference sample exited at 퐷 푎푠 710 휆푒푥; 퐹퐴푟푒푓(휆) is the background-free spectra from the acceptor-only reference sample exited at 퐴 푑푠 푎푠 퐷 퐴 711 휆푒푥 ; and 퐹퐵푟푒푓(휆) and 퐹퐵푟푒푓(휆) are the background spectra obtained at 휆푒푥 amd 휆푒푥 ,

712 respectively. (ratio)A was then calculated from Equation 4, integrating at wavelengths common 푑푠 푑푠 713 in both 퐹퐴푟푒푓(휆) and 퐹퐴푟푒푓(휆) (from 660 to 750 nm).

푑푠 푑푠 훼 퐹퐴푟푒푓(휆) 714 (ratio)A = 푎푠 푎푠 (4) 훼 퐹퐴푟푒푓(휆) 715 All calculations were implemented in Excel. 716 Continuous glycosyltransferase (GTase) assay using dansylated lipid II – Continuous 717 fluorescence GTase assays using dansylated lipid II and A. baumannii PBP1B were performed 718 as previously described (Schwartz et al., 2001, Offant et al., 2010, Egan & Vollmer, 2016).

719 Samples contained 50 mM HEPES/NaOH pH 7.5, 105 mM NaCl, 25 mM MgCl2, 0.039% 720 Triton X-100 and 0.14 µg/µL cellosyl muramidase in a final volume of 60 µL. PBP1BAb was 721 added at a concentration of 0.5 µM. When indicated, LpoPAb(sol) was added at a concentration 722 of 0.5 µM. Reactions were started by addition of dansylated lipid II to a final concentration of 723 10 µM and monitored by following the decrease in fluorescence over 60 min at 37°C using a 724 FLUOstar OPTIMA plate reader (BMG Labtech, Germany) with excitation at 330 nm and 725 emission at 520 nm. The fold-increase in GTAse was calculated against the mean rate obtained 726 with PBP1BAb alone at these reaction conditions, at the fastest rate. 727 Time-course GTase assay by SDS-PAGE followed by fluorescence detection – PBP1BAb at a 728 concentration of 0.5 µM was incubated with 5 µM lipid II-Atto550 and 25 µM unlabelled lipid 729 II in the presence or absence of 1.5 µM LpoPAb(sol). Reactions contained 20 mM HEPES,

730 150 mM NaCl, 10 mM MgCl2, 0.06% TX-100 and 1 mM Ampicillin to block transpeptidation. 731 Aliquots were taken after 0, 2, 5, 10, 30 and 60 min incubation at 37°C, boiled for 10 min to

732 stop reactions and analysed by Tris-Tricine SDS-PAGE followed by fluorescence detection as 733 previously described (Van't Veer et al., 2016). 734 735 PG synthesis in liposomes 736 Reconstitution of class A PBPs in liposomes – Proteoliposomes containing class A PBPs were 737 prepared as described previously with some modifications (Egan et al., 2015, Rigaud & Lévy, 738 2003, Hernández-Rocamora et al., 2018). The appropriate lipid or mixture of lipids were dried

739 in a glass test tube under stream of N2 to form a lipid film followed by desiccation under 740 vacuum from 2 h. When labelled lipid II was co-reconstituted with the indicated class A PBP, 741 they were added at 1:200 mol:mol phospholipid to each lipid II-Atto550 and lipid II-Atto647n. 742 Resuspension into multilamellar vesicles (MLVs) was achieved by addition of 20 mM 743 Tris/HCl, pH 7.5 with or without 150 mM NaCl as indicated in each experiment and several 744 cycles of vigorous mixing and short incubations in hot tap water. The final lipid concentration 745 was 5 g/L. To form large unilamellar vesicles (LUVs), MLVs were subjected to 10 freeze-thaw 746 cycles and then extruded 10 times through a 0.2 µm filter. LUVs were destabilised by the 747 addition of Triton X-100 to an effective detergent:lipid ratio of 1.40 and mixed with proteins 748 in different protein to lipid molar rations (1:3000 for PBP1BEc and PBP1BPa, and 1:2000 for 749 PBP1BAb). After incubation at 4°C for 1 h, prewashed adsorbent beads (Biobeads SM2, 750 BioRad, USA; 100 mg per 3 µmol of Triton X-100) were added to the sample to remove 751 detergents. Biobeads were exchanged after 2 and 16 h, followed by incubation with fresh 752 Biobeads for a further 2 h. After removal of Biobeads by short centrifugation at 4,000×g, 753 liposomes were pelleted at 250,000×g for 30 min at 4°C. The pellet containing proteoliposomes 754 was resuspended using the appropriate buffer. The resuspension was done in a 43% smaller 755 volume than the volume added of lipid II, so that the final concentration of lipids was 11.6 g/L. 756 Samples were then centrifuged for 5 min at 17,000×g and 4°C to remove any possible 757 aggregates. The supernatant was then used in the appropriate assays. Liposomes were analysed 758 by SDS-PAGE and, only for liposomes without labelled lipid II, also by bicinchoninic acid 759 assay (Pierce BCA Assay Kit, ThemoFisher Scientific, USA) to determine protein 760 concentration. The concentration of protein for liposomes with labelled lipid II was calculated 761 by densitometry of the samples in SDS-PAGE gels, after reactions were carried out. 762 PBP1BEc orientation assay. To assess the orientation of liposome-reconstituted PBP1BEc, 763 MGC-64PBP1B-his C777S C795S mutant containing a single cysteine in the N-terminal region 764 was reconstituted in liposomes with EcPL as described above. The accessibility of the cysteine 765 was determined using sulfhydryl-reactive fluorescent probe AlexaFluor555-maleimide.

766 Reactions containing 0.5 µM protein, 10 µM AlexaFluor555-maleimide, 0.2 mM TCEP were 767 incubated for 16 h at 4°C in the presence or absence 0.5% Triton X-100. Reactions were 768 stopped by addition of 5 mM DTT and boiling for 5 min. Samples were loaded in a 10% 769 acrylamide gel and, after electrophoresis, gels were first scanned using an Amersham Typhoon 770 Trio with excitation at 533 nm and a 40 nm-wide band-pass emission filter at 580 nm. The gel 771 was then stained by Coomassie. 772 In vitro peptidoglycan synthesis assay using radiolabelled lipid II in liposomes – The same 773 methodology as in detergents was used to assay the in vitro PG synthesis activity of PBP1BEc 774 in liposomes, with minor modifications. To start reactions, 1.5 nmol [14C]-labelled lipid II were 775 dried in a 0.5 mL glass tube using a vacuum concentrator, resuspended in 5 µL of the

776 appropriate liposome buffer, and mixed with liposomes, buffer and MgCl2 to a total volume of Ec 777 50 µL. Final reactions contained 0.5 µM PBP1B , 30 µM lipid II and 1 mM MgCl2 in 20 mM 778 Tris/HCl pH 7.5 with or without 150 mM NaCl as indicated for each experiment. Samples were 779 incubated for 90 min at 37°C with shaking at 800 rpm. Reactions were stopped by boiling for 780 5 min. Digestion with cellosyl, reduction with sodium borohydride and analysis by HPLC were 781 performed as described (Biboy et al., 2013). 782 FRET-based in vitro peptidoglycan synthesis assay in liposomes – For assays with liposomes,

783 samples contained 20 mM Tris pH 7.5, 1 mM MgCl2 in a final volume of 50 µL. In this case, 784 the same volume for each liposome preparation was added to the reactions, 10 µL, so that the 785 total amount of labelled lipid II was present in every reaction. In these conditions, concentration 786 of lipid II-Atto550 and lipid II-Atto647n would be 14.5 µM each, assuming no loss of lipids 787 during sample preparation. The final concentration of enzymes for the reactions shown in 788 Figure 3, determined by densitometry of SDS-PAGE gels, were ~0.59 µM for PBP1BEc, ~0.81 789 µM for PBP1BAb, and ~0.53 µM for PBP1BPa. When indicated, activators LpoB(sol), 790 LpoPAb(sol), or LpoPPa(sol) were added at a concentration of 2 µM. Reactions were started by 791 the addition of lipid II at 12 µM and monitored by measuring fluorescence over a period of 60 792 min (or 90 min for PBP1BPa liposomes) at 37°C using a Clariostar plate reader (BMG Labtech, 793 Germany), with emission measurements at 590 nm and 680 nm after excitation at 522 nm. 794 When indicated ampicillin was added at 1 mM and moenomycin was added at 50 µM. Activity 795 assays were performed immediately after preparation of liposomes was finished, as we noticed 796 that some proteins could slowly start polymerization using the labelled lipid II. After reactions, 797 samples were analysed by Tris-Tricine SDS-PAGE as indicated for detergents.

798 Analysis of FRET reaction curves – Slopes were calculated as indicated in the FRET assay in 799 the presence of detergent. As it is not possible to precisely adjust the final amount of enzyme 800 in different liposome preparations, there could be differences in activities measured due to 801 different enzyme amounts. Therefore, we calculated the ratio of the slope with activator over 802 the slope without activator for every liposome preparation and then averaged the values 803 (instead of averaging the different measurements from every sample). At least two independent 804 liposome preparations were prepared for every class A PBP. 805 806 Assays in supported lipid bilayers 807 Preparation of small unilamellar vesicles (SUVs) and proteoliposomes for SLB formation – 808 Liposomes of EcPL lipids and proteoliposomes with reconstituted PBP1BEc were prepared as 809 described previously by addition of beta-cyclodextrin to the solution of lipids and Triton X- 810 100 detergent (DeGrip et al., 1998, Roder et al., 2011). Briefly, a thin lipid film of E. coli polar

811 lipid extract was prepared by N2 assisted chloroform evaporation. After 2h of drying under 812 vacuum the lipid film was re-hydrated to 5 mM (total phosphorus concentration) in 150 mM 813 NaCl, 10 mM Tris-HCl, pH 7.4 supplemented with 20 mM Triton X-100. The suspension of 814 lipids/detergent was extensively vortex, freeze/thawed for 5 cycles and sonicated using a water- 815 bath sonicator for 10 min (on ice, to avoid lipids overheating upon sonication). To prepare 816 proteoliposomes, full length PBP1B produced as described above and containing 0.05% Triton 817 X-100 was mixed with a lipid-detergent suspension at the indicated ratio, usually 1:25000 818 (protein:lipids) and incubated for 10 min at RT. Incorporation of PBP1BEc into liposomes was 819 achieved by addition of 2× excess of beta-cyclodextrin solution for 5 min (at RT) with 820 subsequent 20-fold dilution in 20 mM Hepes, pH 7.4. The rapid depletion of detergent by 821 addition of beta-cyclodextrin leads to formation of very small unilamellar vesicles with an 822 average diameter of 18-25 nm and narrow size distribution (Roder et al., 2011). 823 To prepare liposomes with fluorescently labelled lipid II the extract of E. coli polar 824 lipids was supplemented with 2 mol% solution of either lipid II-Atto550 or lipid II-Atto647n. 825 The lipid film was treated similar as the film for preparation of proteoliposomes. Liposomes 826 were also prepared by cyclodextrin-assisted extraction of Triton X-100. 827 Formation of polymer-supported lipid bilayers (SLBs) and reconstitution of PBP1BEc into a 828 supported lipid membrane – To form polymer-supported lipid membranes the coverslips were 829 functionalised beforehand with a dense PEG film, where the ends of the polymer brush were 830 covalently modified with palmitic acid, which served as a linker to capture liposomes as

831 described elsewhere (Roder et al., 2011). To perform a FRET assay on supported lipid 832 membrane empty EcPL liposomes (i), liposomes with 2 mol% of either lipid II-Atto550 (ii) or 833 lipid II-Atto647n (iii), and PBP1BEc proteoliposomes (iv) were mixed at equimolar ratio; and 834 diluted by 20-fold with the 10 mM Tris pH 7.5 buffer directly in the reaction chamber. After 835 30 min of incubation at 37 °C the reaction chamber was washed 5 times by solution exchange. 836 Proteoliposomes adsorbed on the surface were fused by the addition of 10% (w/v) PEG 8kDa 837 solution (in water). The fusion reaction was carried for 15 min at 37°C, afterwards PEG 838 solution was rigorously washed out. Fluidity and homogeneity of the lipid membrane were

839 checked either with PE-Rhodamine dye (Avanti) or by addition of a His6-tagged (on the C- 840 terminus) neutral peptide (CMSQAALNTRNSEEEVSSRRNNGTRHHHHHH) labelled with 841 a single Alexa 488 fluorophore on its only Cys residue at the N-terminus to the EcPL membrane 842 containing 0.1 mol% dioctadecylamine (DODA)-tris-Ni-NTA (Beutel et al., 2014). 843 FRET-based in vitro peptidoglycan synthesis assay in supported lipid bilayers using TIRF 844 microscopy – Peptidoglycan synthesis reactions were carried out at 10 mM Tris pH 7.5

845 supplemented with 1 mM MgCl2, with or without 1 mM Ampicillin and in the presence of 4 µM 846 LpoB(sol). The reaction was started by addition of 4 µM of unlabelled lipid II. TIRF 847 microscopy, using a set up described elsewhere(Baranova et al., 2020) was used to monitor an 848 increase in FRET efficiency and spatial reorganization of FRET signal over the time course of 849 PG synthesis. To detect real-time FRET on supported lipid membranes we used the so-called 850 “acceptor photobleaching approach” where a region of interest of about 10×10 µm was 851 photobleached in the acceptor channel (lipid II-Atto647n) and the increase in fluorescence 852 intensity of the donor (lipid II-Atto550) was recorded within a delay of 1s. We found that in 853 our experiments, photobleaching of the acceptor dye was the only process that contributed to 854 the recorded increase in the donor fluorescence signal. Accordingly, the relative increase in 855 donor fluorescence can be used a direct readout for the FRET efficiency and could therefore 856 be calculated as described (Loose et al., 2011, Verveer et al., 2006). Briefly, donor intensity 857 levels were calculated before (ID) and after photobleaching (ID,pb) using intensity measurements 858 in ImageJ. FRET efficiency was calculated using Equation 5: 859 E = (ID,pb – ID) / ID,pb (5) 860 For time-course measurements (Figure 4D) the acceptor signal (lipid II-Atto647n) was 861 photobleached every minute after the initiation of the reaction (the data point at time 0 862 corresponds to addition of unlabelled lipid II). For each time point a new region of interest in 863 the same chamber was photobleached, and the change in the donor intensity was recorded to 864 calculate FRET efficiency using Eq.1.

865 To have a control on the lipid membrane integrity during PG synthesis the phospholipid 866 DODA-tris-Ni-NTA (Beutel et al., 2014) was included during reconstitution at a 0.1 mol%

867 ratio. DODA-tris-Ni-NTA was then visualized using a His6-containing peptide 868 (CMSQAALNTRNSEEEVSSRRNNGTRHHHHHH) labelled with Alexa488 on its single 869 Cys residue, which we added in the same experiment in which we performed FRET analysis. 870 To compare the fluidity and the immobile fraction of lipid II-Atto647n before and after 1 h of 871 the synthesis reaction with the fluidity of phospholipids in the lipid membrane, the same region 872 of interest was photobleached with a laser first at 640 nm and afterwards at 480 nm. 873 In vitro peptidoglycan synthesis assay using radiolabelled lipid II on supported lipid bilayers 874 – To assay PG synthesis on supported lipid bilayers (SLBs) using radioactively labelled lipid 875 II, we first reconstituted PBP1BEc on SLBs containing E. coli polar lipid extract and a 1:105 876 PBP1BEc to lipid molar ratio, as described above. Due to the low density of the enzyme, several 877 1.1 cm2 chambers were assayed for every condition in order to accumulate a measurable signal. 878 In every chamber, reactions were started by addition of 10 µM [14C]-labelled lipid II and 4 µM 879 LpoB(sol) in a total volume of 100 µL per chamber. The synthesis reaction was carried out in

880 10 mM Tris pH 7.5, 1 mM MgCl2. The chambers were incubated overnight (~16h) at 37°C, 881 covered with parafilm. Reactions were stopped by addition of 100 µM moenomycin. To digest 882 the produced peptidoglycan, cellosyl was added at 0.05 g/L, in the presence of 0.3% triton X- 883 100. After 1h incubation at 37°C, samples from 6 Chambers were pooled in an Eppendorf tube, 884 concentrated using a speed-vac evaporator, reduced using sodium borohydride and analysed 885 by HPLC as described above. For the experiment to determine lipid II incorporation and the 886 localisation of the produced PG, before addition of moenomycin, chambers were washed by 887 removal of 50 µL of buffer and addition of 50 µL of fresh buffer while mixing. This was 888 repeated 5 times. The removed volume from each wash was pooled and treated the same as the 889 samples left in the chamber. 890 Single molecule tracking and analysis – To perform single molecular tracking, MGC-64PBP1B- 891 his C77S C795S was labelled with the photostable far-red dye Dy647N as described above and 892 then reconstituted into a polymer-supported lipid membrane as described elsewhere (Roder et 893 al., 2011, Roder et al., 2014). Single molecule tracking experiments were performed at a low 894 protein to lipid molar ratio (1:10-6). At this ratio, supported lipid membrane was largely 895 homogeneous with the lowest immobile fraction from all the ratios tested (Figure 3 – figure 896 supplement 1). The single-molecule motion of PBP1B was measured prior and after the 897 addition of 1.5 µM lipid II after 15 min ex situ incubation, in the presence of 10 mM Hepes pH

898 7.4, 150mM NaCl, 1 mM MgCl2 buffer and in the absence of LpoB(sol). The localization and

899 tracking of PBP1B particles was performed by the SLIMfast software (Serge et al., 2008). To 900 ensure that non-specifically stuck PBP1B particles did not contribute to the measured diffusion 901 coefficient, the immobile particles were excluded using the DBSCAN spatial clustering 902 algorithm (Sander et al., 1998) with the following clustering parameters: a search area of 100 903 nm, the minimal time window of 30 frames at 65 ms/frame acquisition time. The displacement 904 distributions for active PBP1B (in the presence of lipid II) was compared to the displacement 905 distribution of PBP1B before lipid II addition by fitting the two-component Rayleigh 906 distribution and comparing the weighted contribution of each population. The mean-squared 907 displacement was fitted to each individual trajectory longer than 650 ms (10 frames). Each 908 MSD curve was fitted with a linear fit considering max 30% of the lag-time for each trajectory. 909 FRAP analysis – To control membrane fluidity upon the reconstitution of the transmembrane 910 PBP1B (Figure 3 – figure supplement 1 and Figure 4 – figure supplement 1) and fluidity of 911 lipid II Atto-647n during peptidoglycan synthesis (Figure 4E-F) we used a Matlab-based GUI 912 frap_analysis (Jönsson, 2020) in details described elsewhere (Jönsson et al., 2008). This code 913 allows to quantify the contribution of the immobile fraction to the estimated diffusion 914 coefficient, and particularly suitable for the analysis of 2D diffusion with the photobleaching 915 contribution during the recovery. 916 917 ACKNOWLEDGEMENTS 918 We thank Alexander Egan (Newcastle University) for purified proteins LpoB(sol) and 919 LpoPPa(sol), Federico Corona (Newcastle University) for purified MepM, Oliver Birkholz and 920 Jacob Piehler for their help with PBP1B reconstitution into polymer-SLBs and initial guidance 921 on single particle tracking. We also acknowledge Changjiang You for providing tris-DODA- 922 NTA reagent. This work was funded by the BBSRC grant BB/R017409/1 (to W.V.), by the 923 European Research Council through grant ERC-2015-StG-679239 (to M.L.), and long-term 924 fellowships HFSP LT 000824/2016-L4 and EMBO ALTF 1163-2015 (to N.B.). 925 926 CONFLICT OF INTEREST 927 The authors declare no competing financial interests. 928 929 REFERENCES 930 (!!! INVALID CITATION !!! {}). DOI: 931 Banzhaf, M., van den Berg van Saparoea, B., Terrak, M., Fraipont, C., Egan, A., Philippe, J., 932 Zapun, A., Breukink, E., Nguyen-Disteche, M., den Blaauwen, T., and Vollmer, W.

933 (2012) Cooperativity of peptidoglycan synthases active in bacterial cell elongation. 934 Molecular Microbiology 85: 179-194. DOI: 10.1111/j.1365-2958.2012.08103.x 935 Baranova, N., Radler, P., Hernández-Rocamora, V.M., Alfonso, C., López-Pelegrín, M., 936 Rivas, G., Vollmer, W., and Loose, M. (2020) Diffusion and capture permits dynamic 937 coupling between treadmilling FtsZ filaments and cell division proteins. Nature 938 Microbiology. DOI: 10.1038/s41564-019-0657-5 939 Barrett, D., Wang, T.S., Yuan, Y., Zhang, Y., Kahne, D., and Walker, S. (2007) Analysis of 940 glycan polymers produced by peptidoglycan glycosyltransferases. Journal of 941 Biological Chemistry 282: 31964-31971. DOI: 10.1074/jbc.M705440200 942 Bertsche, U., Breukink, E., Kast, T., and Vollmer, W. (2005) In Vitro Murein 943 (Peptidoglycan) Synthesis by Dimers of the Bifunctional Transglycosylase- 944 Transpeptidase PBP1B from Escherichia coli. Journal of Biological Chemistry 280: 945 38096-38101. DOI: 10.1074/jbc.M508646200 946 Bertsche, U., Kast, T., Wolf, B., Fraipont, C., Aarsman, M.E., Kannenberg, K., von 947 Rechenberg, M., Nguyen-Disteche, M., den Blaauwen, T., Holtje, J.V., and Vollmer, 948 W. (2006) Interaction between two murein (peptidoglycan) synthases, PBP3 and 949 PBP1B, in Escherichia coli. Molecular Microbiology 61: 675-690. DOI: 950 10.1111/j.1365-2958.2006.05280.x 951 Beutel, O., Nikolaus, J., Birkholz, O., You, C., Schmidt, T., Herrmann, A., and Piehler, J. 952 (2014) High-fidelity protein targeting into membrane lipid microdomains in living 953 cells. Angewandte Chemie International Edition 53: 1311-1315. DOI: 954 10.1002/anie.201306328 955 Biboy, J., Bui, N.K., and Vollmer, W., (2013) In vitro peptidoglycan synthesis assay with 956 lipid II substrate. In: Bacterial Cell Surfaces: Methods and Protocols. A.H. Delcour 957 (ed). Totowa, NJ: Humana Press, pp. 273-288. DOI: 10.1007/978-1-62703-245-2_17 958 Born, P., Breukink, E., and Vollmer, W. (2006) In vitro synthesis of cross-linked murein and 959 its attachment to sacculi by PBP1A from Escherichia coli. Journal of Biological 960 Chemistry 281: 26985-26993. DOI: 10.1074/jbc.M604083200 961 Breukink, E., van Heusden, H.E., Vollmerhaus, P.J., Swiezewska, E., Brunner, L., Walker, 962 S., Heck, A.J., and de Kruijff, B. (2003) Lipid II is an intrinsic component of the pore 963 induced by nisin in bacterial membranes. Journal of Biological Chemistry 278: 964 19898-19903. DOI: 10.1074/jbc.M301463200 965 Catherwood, A.C., Lloyd, A.J., Tod, J.A., Chauhan, S., Slade, S.E., Walkowiak, G., Galley, 966 N.F., Punekar, A., Smart, K., Rea, D., Evans, N.D., Chappell, M.J., Roper, D.I., and

967 Dowson, C.G. (2020a) Substrate and stereochemical control of peptidoglycan 968 crosslinking by transpeptidation by Escherichia coli PBP1B. J Am Chem Soc. DOI: 969 10.1021/jacs.9b08822 970 Catherwood, A.C., Lloyd, A.J., Tod, J.A., Chauhan, S., Slade, S.E., Walkowiak, G.P., Galley, 971 N.F., Punekar, A.S., Smart, K., Rea, D., Evans, N.D., Chappell, M.J., Roper, D.I., and 972 Dowson, C.G. (2020b) Substrate and stereochemical control of peptidoglycan cross- 973 linking by transpeptidation by Escherichia coli PBP1B. Journal of the American 974 Chemical Society 142: 5034-5048. DOI: 10.1021/jacs.9b08822 975 Caveney, N.A., Egan, A.J.F., Ayala, I., Laguri, C., Robb, C.S., Breukink, E., Vollmer, W., 976 Strynadka, N.C.J., and Simorre, J.-P. (2020) Structure of the peptidoglycan synthase 977 activator LpoP in Pseudomonas aeruginosa. Structure 28: 643-650.e645. DOI: 978 https://doi.org/10.1016/j.str.2020.03.012 979 Cho, H., Wivagg, C.N., Kapoor, M., Barry, Z., Rohs, P.D.A., Suh, H., Marto, J.A., Garner, 980 E.C., and Bernhardt, T.G. (2016) Bacterial cell wall biogenesis is mediated by SEDS 981 and PBP polymerase families functioning semi-autonomously. Nature Microbiology 982 1: 16172. DOI: 10.1038/nmicrobiol.2016.172 983 DeGrip, J.W., VanOostrum, J., and Bovee-Geurts, P.H.M. (1998) Selective detergent- 984 extraction from mixed detergent/lipid/protein micelles, using cyclodextrin inclusion 985 compounds: a novel generic approach for the preparation of proteoliposomes. 986 Biochemical Journal 330: 667. DOI: 10.1042/bj3300667 987 Derouaux, A., Wolf, B., Fraipont, C., Breukink, E., Nguyen-Disteche, M., and Terrak, M. 988 (2008) The monofunctional glycosyltransferase of Escherichia coli localizes to the 989 cell division site and interacts with penicillin-binding protein 3, FtsW, and FtsN. 990 Journal of Bacteriology 190: 1831-1834. DOI: 10.1128/jb.01377-07 991 Egan, A.J., Biboy, J., van't Veer, I., Breukink, E., and Vollmer, W. (2015) Activities and 992 regulation of peptidoglycan synthases. Philosophical transactions of the Royal Society 993 of London. Series B, Biological sciences 370. DOI: 10.1098/rstb.2015.0031 994 Egan, A.J., Cleverley, R.M., Peters, K., Lewis, R.J., and Vollmer, W. (2017) Regulation of 995 bacterial cell wall growth. The FEBS Journal 284: 851-867. DOI: 10.1111/febs.13959 996 Egan, A.J., and Vollmer, W. (2016) Continuous fluorescence assay for peptidoglycan 997 glycosyltransferases. Methods in Molecular Biology 1440: 171-184. DOI: 998 10.1007/978-1-4939-3676-2_13

999 Egan, A.J.F., Errington, J., and Vollmer, W. (2020) Regulation of peptidoglycan synthesis 1000 and remodelling. Nature Reviews Microbiology 18: 446-460. DOI: 10.1038/s41579- 1001 020-0366-3 1002 Egan, A.J.F., Jean, N.L., Koumoutsi, A., Bougault, C.M., Biboy, J., Sassine, J., Solovyova, 1003 A.S., Breukink, E., Typas, A., Vollmer, W., and Simorre, J.-P. (2014) Outer- 1004 membrane lipoprotein LpoB spans the periplasm to stimulate the peptidoglycan 1005 synthase PBP1B. Proceedings of the National Academy of Sciences of the United 1006 States of America 111: 8197-8202. DOI: 10.1073/pnas.1400376111 1007 Egan, A.J.F., Maya-Martinez, R., Ayala, I., Bougault, C.M., Banzhaf, M., Breukink, E., 1008 Vollmer, W., and Simorre, J.P. (2018a) Induced conformational changes activate the 1009 peptidoglycan synthase PBP1B. Molecular Microbiology 110: 335-356. DOI: 1010 10.1111/mmi.14082 1011 Egan, A.J.F., Maya-Martinez, R., Ayala, I., Bougault, C.M., Banzhaf, M., Breukink, E., 1012 Vollmer, W., and Simorre, J.P. (2018b) Induced conformational changes activate the 1013 peptidoglycan synthase PBP1B. Mol Microbiol 110: 335-356. DOI: 1014 10.1111/mmi.14082 1015 Frere, J.M., Leyh-Bouille, M., Ghuysen, J.M., Nieto, M., and Perkins, H.R. (1976) 1016 Exocellular DD-carboxypeptidases-transpeptidases from Streptomyces. Methods in 1017 Enzymology 45: 610-636. DOI: 10.1016/s0076-6879(76)45054-1 1018 Gavutis, M., Jaks, E., Lamken, P., and Piehler, J. (2006) Determination of the two- 1019 dimensional interaction rate constants of a cytokine receptor complex. Biophysical 1020 Journal 90: 3345-3355. DOI: 10.1529/biophysj.105.072546 1021 Goffin, C., and Ghuysen, J.M. (1998) Multimodular penicillin-binding proteins: an enigmatic 1022 family of orthologs and paralogs. Microbiology and Molecular Biology Reviews 62: 1023 1079-1093. DOI: 1024 Gray, A.N., Egan, A.J., Van't Veer, I.L., Verheul, J., Colavin, A., Koumoutsi, A., Biboy, J., 1025 Altelaar, A.F., Damen, M.J., Huang, K.C., Simorre, J.P., Breukink, E., den Blaauwen, 1026 T., Typas, A., Gross, C.A., and Vollmer, W. (2015a) Coordination of peptidoglycan 1027 synthesis and outer membrane constriction during Escherichia coli cell division. Elife 1028 4: e07118. DOI: 10.7554/eLife.07118 1029 Gray, A.N., Egan, A.J., Van't Veer, I.L., Verheul, J., Colavin, A., Koumoutsi, A., Biboy, J., 1030 Altelaar, A.F., Damen, M.J., Huang, K.C., Simorre, J.P., Breukink, E., den Blaauwen, 1031 T., Typas, A., Gross, C.A., and Vollmer, W. (2015b) Coordination of peptidoglycan

1032 synthesis and outer membrane constriction during Escherichia coli cell division. Elife 1033 4. DOI: 10.7554/eLife.07118 1034 Greene, N.G., Fumeaux, C., and Bernhardt, T.G. (2018a) Conserved mechanism of cell-wall 1035 synthase regulation revealed by the identification of a new PBP activator in 1036 Pseudomonas aeruginosa. Proc Natl Acad Sci U S A 115: 3150-3155. DOI: 1037 10.1073/pnas.1717925115 1038 Greene, N.G., Fumeaux, C., and Bernhardt, T.G. (2018b) Conserved mechanism of cell-wall 1039 synthase regulation revealed by the identification of a new PBP activator in 1040 Pseudomonas aeruginosa. Proceedings of the National Academy of Sciences of the 1041 United States of America 115: 3150-3155. DOI: 10.1073/pnas.1717925115 1042 Gutheil, W.G., Stefanova, M.E., and Nicholas, R.A. (2000) Fluorescent coupled enzyme 1043 assays for D-alanine: application to penicillin-binding protein and vancomycin 1044 activity assays. Analitical Biochemistry 287: 196-202. DOI: 10.1006/abio.2000.4835 1045 Hernández-Rocamora, V.M., Otten, C.F., Radkov, A., Simorre, J.-P., Breukink, E., 1046 VanNieuwenhze, M., and Vollmer, W. (2018) Coupling of polymerase and carrier 1047 lipid phosphatase prevents product inhibition in peptidoglycan synthesis. The Cell 1048 Surface 2: 1-13. DOI: 1049 Hsu, Y.P., Hall, E., Booher, G., Murphy, B., Radkov, A.D., Yablonowski, J., Mulcahey, C., 1050 Alvarez, L., Cava, F., Brun, Y.V., Kuru, E., and VanNieuwenhze, M.S. (2019) 1051 Fluorogenic D-amino acids enable real-time monitoring of peptidoglycan biosynthesis 1052 and high-throughput transpeptidation assays. Nature Chemistry 11: 335-341. DOI: 1053 10.1038/s41557-019-0217-x 1054 Huang, S.H., Wu, W.S., Huang, L.Y., Huang, W.F., Fu, W.C., Chen, P.T., Fang, J.M., 1055 Cheng, W.C., Cheng, T.J., and Wong, C.H. (2013) New continuous fluorometric 1056 assay for bacterial transglycosylase using Förster resonance energy transfer. Journal 1057 of the American Chemical Society 135: 17078-17089. DOI: 10.1021/ja407985m 1058 Izaki, K., Matsuhashi, M., and Strominger, J.L. (1968) Biosynthesis of the peptidoglycan of 1059 bacterial cell walls: XIII. Peptidoglycan transpeptidase and D-alanine 1060 carboxypeptidase: penicillin-sensistive enzymatic reactions in strains of Escherichia 1061 coli. Journal of Biological Chemistry 243: 3180-3192. DOI: 1062 Jean, N.L., Bougault, C.M., Lodge, A., Derouaux, A., Callens, G., Egan, A.J., Ayala, I., 1063 Lewis, R.J., Vollmer, W., and Simorre, J.P. (2014) Elongated structure of the outer- 1064 membrane activator of peptidoglycan synthesis LpoA: implications for PBP1A 1065 stimulation. Structure 22: 1047-1054. DOI: 10.1016/j.str.2014.04.017

1066 Jönsson, P. (2020) frap_analysis 1067 https://www.mathworks.com/matlabcentral/fileexchange/29388-frap_analysis 1068 (accessed 2020) 1069 Jönsson, P., Jonsson, M.P., Tegenfeldt, J.O., and Höök, F. (2008) A method improving the 1070 accuracy of fluorescence recovery after photobleaching analysis. Biophysical Journal 1071 95: 5334-5348. DOI: 10.1529/biophysj.108.134874 1072 Kumar, V.P., Basavannacharya, C., and de Sousa, S.M. (2014) A microplate assay for the 1073 coupled transglycosylase-transpeptidase activity of the penicillin binding proteins; a 1074 vancomycin-neutralizing tripeptide combination prevents penicillin inhibition of 1075 peptidoglycan synthesis. Biochemical and Biophysical Research Communications 1076 450: 347-352. DOI: 10.1016/j.bbrc.2014.05.119 1077 Lebar, M.D., Lupoli, T.J., Tsukamoto, H., May, J.M., Walker, S., and Kahne, D. (2013) 1078 Forming cross-linked peptidoglycan from synthetic gram-negative Lipid II. Journal of 1079 the American Chemical Society 135: 4632-4635. DOI: 10.1021/ja312510m 1080 Lee, T.K., Meng, K., Shi, H., and Huang, K.C. (2016) Single-molecule imaging reveals 1081 modulation of cell wall synthesis dynamics in live bacterial cells. Nature 1082 Communications 7: 13170. DOI: 10.1038/ncomms13170 1083 Loose, M., Fischer-Friedrich, E., Herold, C., Kruse, K., and Schwille, P. (2011) Min protein 1084 patterns emerge from rapid rebinding and membrane interaction of MinE. Nature 1085 Structural & Molecular Biology 18: 577-583. DOI: 10.1038/nsmb.2037 1086 Lupoli, T.J., Lebar, M.D., Markovski, M., Bernhardt, T., Kahne, D., and Walker, S. (2014a) 1087 Lipoprotein activators stimulate Escherichia coli penicillin-binding proteins by 1088 different mechanisms. Journal of the American Chemical Society 136: 52-55. DOI: 1089 10.1021/ja410813j 1090 Lupoli, T.J., Lebar, M.D., Markovski, M., Bernhardt, T., Kahne, D., and Walker, S. (2014b) 1091 Lipoprotein activators stimulate Escherichia coli penicillin-binding proteins by 1092 different mechanisms. J Am Chem Soc 136: 52-55. DOI: 10.1021/ja410813j 1093 Macheboeuf, P., Contreras-Martel, C., Job, V., Dideberg, O., and Dessen, A. (2006) 1094 Penicillin binding proteins: key players in bacterial cell cycle and drug resistance 1095 processes. FEMS Microbiology Reviews 30: 673-691. DOI: 10.1111/j.1574- 1096 6976.2006.00024.x 1097 Meeske, A.J., Riley, E.P., Robins, W.P., Uehara, T., Mekalanos, J.J., Kahne, D., Walker, S., 1098 Kruse, A.C., Bernhardt, T.G., and Rudner, D.Z. (2016) SEDS proteins are a

1099 widespread family of bacterial cell wall polymerases. Nature 537: 634-638. DOI: 1100 10.1038/nature19331 1101 Mohammadi, T., Sijbrandi, R., Lutters, M., Verheul, J., Martin, N.I., den Blaauwen, T., de 1102 Kruijff, B., and Breukink, E. (2014) Specificity of the transport of lipid II by FtsW in 1103 Escherichia coli. Journal of Biological Chemistry 289: 14707-14718. DOI: 1104 10.1074/jbc.M114.557371 1105 Morè, N., Martorana, A.M., Biboy, J., Otten, C., Winkle, M., Serrano, C.K.G., Montón Silva, 1106 A., Atkinson, L., Yau, H., Breukink, E., den Blaauwen, T., Vollmer, W., and Polissi, 1107 A. (2019) Peptidoglycan remodeling enables Escherichia coli to survive severe outer 1108 membrane assembly defect. mBio 10: e02729-02718. DOI: 10.1128/mBio.02729-18 1109 Muller, P., Ewers, C., Bertsche, U., Anstett, M., Kallis, T., Breukink, E., Fraipont, C., Terrak, 1110 M., Nguyen-Disteche, M., and Vollmer, W. (2007a) The essential cell division protein 1111 FtsN interacts with the murein (peptidoglycan) synthase PBP1B in Escherichia coli. J 1112 Biol Chem 282: 36394-36402. DOI: 10.1074/jbc.M706390200 1113 Muller, P., Ewers, C., Bertsche, U., Anstett, M., Kallis, T., Breukink, E., Fraipont, C., Terrak, 1114 M., Nguyen-Disteche, M., and Vollmer, W. (2007b) The essential cell division 1115 protein FtsN interacts with the murein (peptidoglycan) synthase PBP1B in 1116 Escherichia coli. Journal of Biological Chemistry 282: 36394-36402. DOI: 1117 10.1074/jbc.M706390200 1118 Offant, J., Terrak, M., Derouaux, A., Breukink, E., Nguyen-Disteche, M., Zapun, A., and 1119 Vernet, T. (2010) Optimization of conditions for the glycosyltransferase activity of 1120 penicillin-binding protein 1a from Thermotoga maritima. The FEBS Journal 277: 1121 4290-4298. DOI: 10.1111/j.1742-4658.2010.07817.x 1122 Pazos, M., Peters, K., Casanova, M., Palacios, P., VanNieuwenhze, M., Breukink, E., 1123 Vicente, M., and Vollmer, W. (2018a) Z-ring membrane anchors associate with cell 1124 wall synthases to initiate bacterial cell division. Nat Commun 9: 5090. DOI: 1125 10.1038/s41467-018-07559-2 1126 Pazos, M., Peters, K., Casanova, M., Palacios, P., VanNieuwenhze, M., Breukink, E., 1127 Vicente, M., and Vollmer, W. (2018b) Z-ring membrane anchors associate with cell 1128 wall synthases to initiate bacterial cell division. Nature Communications 9: 5090. 1129 DOI: 10.1038/s41467-018-07559-2 1130 Qiao, Y., Lebar, M.D., Schirner, K., Schaefer, K., Tsukamoto, H., Kahne, D., and Walker, S. 1131 (2014) Detection of lipid-linked peptidoglycan precursors by exploiting an

1132 unexpected transpeptidase reaction. Journal of the American Chemical Society 136: 1133 14678-14681. DOI: 10.1021/ja508147s 1134 Qiao, Y., Srisuknimit, V., Rubino, F., Schaefer, K., Ruiz, N., Walker, S., and Kahne, D. 1135 (2017) Lipid II overproduction allows direct assay of transpeptidase inhibition by 1136 beta-lactams. Nature Chemical Biology 13: 793-798. DOI: 10.1038/nchembio.2388 1137 Rau, A., Hogg, T., Marquardt, R., and Hilgenfeld, R. (2001) A new lysozyme fold: crystal 1138 struture of the muramidase from Streptomyces coelicolor at 1.65 Å resolution. 1139 Journal of Biological Chemistry 276: 31994-31999. DOI: 10.1074/jbc.M102591200 1140 Rigaud, J.-L., and Lévy, D. (2003) Reconstitution of Membrane Proteins into Liposomes. 1141 Methods in Enzymology 372: 65-86. DOI: https://doi.org/10.1016/S0076- 1142 6879(03)72004-7 1143 Roder, F., Waichman, S., Paterok, D., Schubert, R., Richter, C., Liedberg, B., and Piehler, J. 1144 (2011) Reconstitution of membrane proteins into polymer-supported membranes for 1145 probing diffusion and interactions by single molecule techniques. Analytical 1146 Chemistry 83: 6792-6799. DOI: 10.1021/ac201294v 1147 Roder, F., Wilmes, S., Richter, C.P., and Piehler, J. (2014) Rapid transfer of transmembrane 1148 proteins for single molecule dimerization assays in polymer-supported membranes. 1149 ACS Chem Biology 9: 2479-2484. DOI: 10.1021/cb5005806 1150 Sander, J., Ester, M., Kriegel, H.-P., and Xu, X. (1998) Density-based clustering in spatial 1151 databases: the algorithm GDBSCAN and its applications. Data Mining and 1152 Knowledge Discovery 2: 169-194. DOI: 10.1023/A:1009745219419 1153 Sauvage, E., and Terrak, M. (2016) Glycosyltransferases and transpeptidases/penicillin- 1154 binding proteins: valuable targets for new antibacterials. Antibiotics (Basel) 5. DOI: 1155 10.3390/antibiotics5010012 1156 Schwartz, B., Markwalder, J.A., Seitz, S.P., Wang, Y., and Stein, R.L. (2002) A kinetic 1157 characterization of the glycosyltransferase activity of Eschericia coli PBP1b and 1158 development of a continuous fluorescence assay. Biochemistry 41: 12552-12561. 1159 DOI: 10.1021/bi026205x 1160 Schwartz, B., Markwalder, J.A., and Wang, Y. (2001) Lipid II: total synthesis of the bacterial 1161 cell wall precursor and utilization as a substrate for glycosyltransfer and 1162 transpeptidation by penicillin binding protein (PBP) 1b of Escherichia coli. Journal of 1163 the American Chemical Society 123: 11638-11643. DOI: 10.1021/ja0166848

1164 Serge, A., Bertaux, N., Rigneault, H., and Marguet, D. (2008) Dynamic multiple-target 1165 tracing to probe spatiotemporal cartography of cell membranes. Nature Methods 5: 1166 687-694. DOI: 10.1038/nmeth.1233 1167 Singh, S.K., Parveen, S., SaiSree, L., and Reddy, M. (2015) Regulated proteolysis of a cross- 1168 link–specific peptidoglycan hydrolase contributes to bacterial morphogenesis. 1169 Proceedings of the National Academy of Sciences of the United States of America 1170 112: 10956-10961. DOI: 10.1073/pnas.1507760112 1171 Singh, S.K., SaiSree, L., Amrutha, R.N., and Reddy, M. (2012) Three redundant murein 1172 endopeptidases catalyse an essential cleavage step in peptidoglycan synthesis of 1173 Escherichia coli K12. Molecular Microbiology 86: 1036-1051. DOI: 1174 10.1111/mmi.12058 1175 Sjodt, M., Brock, K., Dobihal, G., Rohs, P.D.A., Green, A.G., Hopf, T.A., Meeske, A.J., 1176 Srisuknimit, V., Kahne, D., Walker, S., Marks, D.S., Bernhardt, T.G., Rudner, D.Z., 1177 and Kruse, A.C. (2018) Structure of the peptidoglycan polymerase RodA resolved by 1178 evolutionary coupling analysis. Nature 556: 118-121. DOI: 10.1038/nature25985 1179 Sjodt, M., Rohs, P.D.A., Gilman, M.S.A., Erlandson, S.C., Zheng, S., Green, A.G., Brock, 1180 K.P., Taguchi, A., Kahne, D., Walker, S., Marks, D.S., Rudner, D.Z., Bernhardt, T.G., 1181 and Kruse, A.C. (2020) Structural coordination of polymerization and crosslinking by 1182 a SEDS-bPBP peptidoglycan synthase complex. Nature Microbiology. DOI: 1183 10.1038/s41564-020-0687-z 1184 Taguchi, A., Welsh, M.A., Marmont, L.S., Lee, W., Sjodt, M., Kruse, A.C., Kahne, D., 1185 Bernhardt, T.G., and Walker, S. (2019) FtsW is a peptidoglycan polymerase that is 1186 functional only in complex with its cognate penicillin-binding protein. Nature 1187 Microbiology 4: 587-594. DOI: 10.1038/s41564-018-0345-x 1188 Typas, A., Banzhaf, M., Gross, C.A., and Vollmer, W. (2012) From the regulation of 1189 peptidoglycan synthesis to bacterial growth and morphology. Nat Rev Microbiol 10: 1190 123-136. DOI: 10.1038/nrmicro2677 1191 Typas, A., Banzhaf, M., van den Berg van Saparoea, B., Verheul, J., Biboy, J., Nichols, R.J., 1192 Zietek, M., Beilharz, K., Kannenberg, K., von Rechenberg, M., Breukink, E., den 1193 Blaauwen, T., Gross, C.A., and Vollmer, W. (2010) Regulation of peptidoglycan 1194 synthesis by outer-membrane proteins. Cell 143: 1097-1109. DOI: 1195 10.1016/j.cell.2010.11.038

1196 Vámosi, G., and Clegg, R.M. (1998) The helix−coil transition of DNA duplexes and hairpins 1197 observed by multiple fluorescence parameters. Biochemistry 37: 14300-14316. DOI: 1198 10.1021/bi9727601 1199 van't Veer, I., (2016) Peptidoglycan synthesis in Escherichia coli from a PBP1b perspective. 1200 In: Bijvoet Center for Biomolecuar Research. Utrecht, The Netherlands: Utrecht 1201 University, pp. DOI: 1202 Van't Veer, I.L., Leloup, N.O., Egan, A.J., Janssen, B.J., Martin, N.I., Vollmer, W., and 1203 Breukink, E. (2016) Site-specific immobilization of the peptidoglycan synthase 1204 PBP1B on a surface plasmon resonance chip surface. Chembiochem 17: 2250-2256. 1205 DOI: 10.1002/cbic.201600461 1206 VanNieuwenhze, M.S., Mauldin, S.C., Zia-Ebrahimi, M., Winger, B.E., Hornback, W.J., 1207 Saha, S.L., Aikins, J.A., and Blaszczak, L.C. (2002) The first total synthesis of lipid 1208 II: the final monomeric intermediate in bacterial cell wall biosynthesis. Journal of the 1209 American Chemical Society 124: 3656-3660. DOI: 10.1021/ja017386d 1210 Verveer, P.J., Rocks, O., Harpur, A.G., and Bastiaens, P.I.H. (2006) Imaging Protein 1211 Interactions by FRET Microscopy: FRET Measurements by Acceptor 1212 Photobleaching. Cold Spring Harbor Protocols 2006: pdb.prot4598. DOI: 1213 10.1101/pdb.prot4598 1214 Vigouroux, A., Cordier, B., Aristov, A., Alvarez, L., Özbaykal, G., Chaze, T., Oldewurtel, 1215 E.R., Matondo, M., Cava, F., Bikard, D., and van Teeffelen, S. (2020) Class-A 1216 penicillin binding proteins do not contribute to cell shape but repair cell-wall defects. 1217 Elife 9: e51998. DOI: 10.7554/eLife.51998 1218 Vollmer, W., Blanot, D., and de Pedro, M.A. (2008) Peptidoglycan structure and architecture. 1219 FEMS Microbiology Reviews 32: 149-167. DOI: 10.1111/j.1574-6976.2007.00094.x 1220 Xie, Z.R., Chen, J., and Wu, Y. (2014) Linking 3D and 2D binding kinetics of membrane 1221 proteins by multiscale simulations. Protein Sci 23: 1789-1799. DOI: 1222 10.1002/pro.2574 1223 Yakovlieva, L., and Walvoort, M.T.C. (2020) Processivity in Bacterial Glycosyltransferases. 1224 ACS Chemical Biology 15: 3-16. DOI: 10.1021/acschembio.9b00619 1225 Ye, X.Y., Lo, M.C., Brunner, L., Walker, D., Kahne, D., and Walker, S. (2001) Better 1226 substrates for bacterial transglycosylases. Journal of the American Chemical Society 1227 123: 3155-3156. DOI: 10.1021/ja010028q

1228 Zhdanov, V.P., and Höök, F. (2015) Kinetics of enzymatic reactions in lipid membranes 1229 containing domains. Physical Biology 12: 026003. DOI: 10.1088/1478- 1230 3975/12/2/026003

1231 1232

1233 FIGURE LEGENDS 1234 Figure 1. FRET assay to monitor peptidoglycan synthesis in real time. (A) Scheme of the 1235 reactions of a class A PBP (GTase-TPase) with unlabelled lipid II and the two versions of 1236 labelled lipid II, yielding a PG product that shows FRET. (B) SDS-PAGE analysis of PG 1237 products by PBP1BEc (0.5 µM) reactions with unlabelled lipid II, Atto550-labelled lipid II and 1238 Atto647n-labelled lipid II at a 1:1:1 molar ratio (each 5 µM), in the absence of antibiotics (I, 1239 red) or in the presence of 1 mM ampicillin (II, blue) or 50 µM moenomycin (III, yellow). 1240 Samples were incubated for 1 h at 37°C and boiled for 5 min. (C) Representative fluorescence 1241 emission spectra taken after reactions performed as described in B and following the same 1242 labelling pattern. (D) FRET efficiency for PBP1BEc reactions carried out as indicated in B,

1243 calculated using the (ratio)A method (see Materials and methods). Values are mean ± SD of at 1244 least 3 independent samples. (E) Representative reactions curves from FRET assays of 1245 detergent-solubilised PBP1BEc. The same components as indicated in B were incubated in the 1246 presence or absence of 2 µM LpoB(sol). Reactions were performed in the absence of antibiotic 1247 (left panel), with 1 mM ampicillin (Amp) or 50 µM moenomycin (Moe) (middle panel), or by 1248 omitting unlabelled lipid II (right panel). The numbers indicate the corresponding lane of the 1249 gel in Figure 1 – figure supplement 2D. Samples were incubated for 1 h at 25°C. (F) Averaged 1250 initial slopes from reaction curves obtained by the FRET assay for detergent-solubilised E. coli 1251 PBP1B in the presence (blue) or absence (red) of LpoB, and in the presence or absence of 1252 ampicillin. Values are normalised relative to the slope in the absence of activator for each 1253 condition and are mean ± SD of 2-3 independent experiments. 1254 1255 Figure 2. The FRET signal arises from both the glycosyltransferase and transpeptidase 1256 reactions. (A) PG synthesised in reactions of PBP1BEc in the presence or absence of 1 mM 1257 ampicillin was incubated with no PG hydrolase (U), DD-endopeptidase MepM (M), or 1258 muramidase Cellosyl (C), and aliquots were analysed by SDS-PAGE. Reaction conditions 1259 were the same as indicated in Figure 1B-D. (B) FRET efficiency for samples prepared as

1260 indicated in A, calculated using the (ratio)A method (see Materials and methods). Values are 1261 mean ± SD of at least 3 independent experiments. (C) PBP1BEc (0.5 µM) was incubated with 1262 5 µM each of lipid II-Atto647n, lipid II-Atto550 and 14C-labelled lipid II. At indicated time 1263 points, aliquots were taken and reactions were stopped by addition of moenomycin. After 1264 measuring fluorescence (see D), the PG was digested with the muramidase cellosyl, and the 1265 resulting muropeptides were reduced with sodium borohydride and separated by HPLC. The 1266 structures of muropeptides corresponding to peaks 1-3 are shown below the chromatograms.

1267 (D) Fluorescence spectra taken with excitation at 522 nm for the samples described in C. (E) 1268 Quantification of peak 2 (GTase product, blue), peak 3 (GTase+TPase, black) or the sum of 1269 both 2 and 3 (yellow) from chromatograms in C, along with the FRET signal (red) calculated 1270 as ratio of acceptor emission over donor emission from data in D. 1271

1272 Figure 3. The FRET assay for PG synthesis can be adapted for reactions on liposomes. 1273 (A) Class A PBPs were reconstituted in E. coli polar lipid liposomes. To assess the orientation 1274 of the liposome-reconstituted PBPs, MGC-64PBP1B-his C777S C795S containing a single 1275 cysteine in the N-terminal region was reconstituted as in A. The accessibility of the cysteine 1276 was determined by staining with sulfhydryl-reactive fluorescent probe, AlexaFluor555- 1277 maleimide, in the presence or absence of Triton X-100 (TX). Samples were analysed by SDS- 1278 PAGE with fluorescence scanning to detect labelled protein followed by Coomassie staining. 1279 (B) To perform activity assays in liposomes, class A PBPs were reconstituted along a 1:1 molar 1280 ratio mixture of Atto550-labelled lipid II and Atto647n-labelled lipid II in liposomes as in A. 1281 Reactions were started by addition of unlabelled lipid II in the presence or absence of 1282 lipoprotein activators (lpo). Using this methodology, we monitored the activity of PBP1BEc 1283 (C-D), PBP1BAb (E-F) and PBP1BPa (G-H). Representative reactions curves are shown. 1284 Reactions were carried out in the presence (blue lines) or absence (red lines) of the lipoprotein 1285 activators (LpoB(sol) for PBP1BEc, LpoPAb(sol) for PBP1BAb and LpoPPa(sol) for PBP1BPa), 1286 and either in the absence of antibiotic (left) or in the presence of 1 mM ampicillin (Amp) or 50 1287 µM moenomycin (Moe, black and yellow lines) (middle). For PBP1BEc, control reactions in 1288 the absence of unlabelled lipid II (right panel) are also shown. Products were analysed by SDS- 1289 PAGE followed by fluorescence scanning at the end of reactions (right side). Curves are 1290 numbered according the corresponding lane on the SDS-PAGE gels. PBP1BEc, PBP1BAb and 1291 PBP1BPa were reconstituted in EcPL liposomes containing labelled lipid II (0.5 mol% of lipids, 1292 1:1 molar ratio mixture of atto550-labelled lipid II and Atto647n-labelled lipid II), at protein 1293 to lipid molar ratios of 1:3000, 1:2000 and 1:3000, respectively. Reactions were started by 1294 adding unlabelled lipid II (final concentration 12 µM) and incubated at 37°C for 60 min 1295 (PBP1BEc and PBP1BAb) or 90 min (PBP1BPa) while monitoring fluorescence at 590 and 680 1296 nm with excitation at 522 nm. (D), (F) and (H) show averaged initial slopes from reaction 1297 curves obtained by the FRET assay for liposome-reconstituted PBP1BEc, PBP1BAb and 1298 PBP1BPa, respectively, in the presence (blue) or absence (red) of lipoprotein activators and in

1299 the presence or absence of ampicillin. Values are normalised relative to the slope in the absence 1300 of activator and are mean ± variation of 2 independent experiments. 1301 1302 Figure 4. Addition of lipid II slows down diffusion of PBP1B on supported lipid bilayers. 1303 (A) Schematic illustration of the approach (not to scale). A single-cysteine version of PBP1BEc 1304 (MGC-64PBP1B-his C777S C795S) labelled with fluorescent probe Dy647 in its single Cys 1305 residue (PBP1BEc-Dy647) was reconstituted into a polymer-supported lipid membrane formed 1306 with E. coli polar lipids and its diffusion was monitored using TIRF microscopy in the presence 1307 or absence of substrate lipid II. (B) Single-molecule TIRF micrograph of PBP1BEc-Dy647 1308 diffusing in the lipid membrane in the presence of 1.5 µM lipid II (corresponding to Video 1).

1309 Calculated particle tracks are overlaid. (C) Histograms of diffusion coefficients (Dcoef) of Ec 1310 PBP1B -Dy647 particles in the presence (red) or absence (black) of lipid II. The average Dcoef 1311 decreased from 0.23±0.06 µm2/s to 0.1±0.04 µm2/s upon addition of lipid II. Values are mean 1312 ± SD of tracks from 3 independent experiments. (D) Representative tracks for diffusing 1313 PBP1BEc-Dy647 particles in the absence (black, top) or presence of lipid II (red, bottom), 1314 showing the absence of confined motion in the presence of lipid II. (E) Displacement 1315 distributions of PBP1BEc-Dy647 particles (solid lines) in the absence (left) or presence (right) 1316 of lipid II were analysed using a Rayleigh model incorporating two populations of particles, a 1317 fast-diffusing one (grey dashed lines) and a slow-diffusing one (black dashed lines). In the 1318 absence of lipid II, only 8±5% of the steps were classified into the slow fraction (121±6nm 1319 average displacement), while the majority of steps were of 257±6 nm (fast fraction). The slow 1320 fraction increased upon addition of lipid II to 37±5% of the steps, with an average displacement 1321 of 132±16 nm. 1322

1323 Figure 5. FRET assay on a planar lipid membrane. (A) FRET acquisition by TIRF 1324 microscopy. PBP1BEc was reconstituted into a polymer supported lipid membrane to preserve 1325 its lateral diffusion. A supported lipid membrane was formed from E. coli polar lipid extract 1326 supplemented with 0.5 mol% of labelled lipid II (Atto550 and Atto647n at 1:1 ratio). To initiate 1327 PG polymerization unlabelled lipid II (10 µM) and of LpoB(sol) (4 µM) were added from the 1328 bulk solution. An increase in FRET efficiency was recorded by dual-colour TIRF microscopy: 1329 the acceptor (lipid II-Atto647n) was photobleached and the concomitant increase in the donor 1330 intensity (lipid II-Atto550) was recorded within a delay of 1 s. (B) FRET kinetics of PG 1331 polymerization and cross-linking. Inhibition of PBP1BEc TPase activity with 1 mM ampicillin

1332 did not produce any changes in the donor intensity, confirming that FRET signal is specific to 1333 cross-linked PG. A sigmoid (straight lines) was fitted to the data to visualise the lag in the 1334 increase of FRET signal. (C) FRET efficiency was measured after a round of PG synthesis 1335 before and after digestion with the muramidase cellosyl. After cellosyl digestion, FRET 1336 efficiency decreased by 2.5-fold, resulting in a FRET signal comparable to the one of a control 1337 surface with a GTase-defective PBP1BEc(E233Q), performed in parallel. Each dot corresponds 1338 to a different surface area within the same sample. (D) Quantification of the diffusion 1339 coefficient of lipid II-Atto647n over the time course of PG polymerization (left panel) from the 1340 experiment presented in B, calculated from the dynamics of the recovery of lipid II-Atto647n 1341 signal within the photobleached ROI. (E) Quantification of the fraction of immobile lipid II- 1342 Atto647n from several experiments as the one depicted in B, each dot represents the value from 1343 a different experiment. (F) Diffusion of lipid II-Atto647n or a phospholipid bound probe 1344 labelled with Alexa 488 (SLB) was recorded in a FRAP assay, using a 1 s delay and dual- 1345 colour imaging, 30 min after initiation of PG synthesis by addition of lipid II and LpoB(sol).

1346 Only the diffusion of lipid II, but not of a fluorescently labelled, His6-tagged peptide attached 1347 to dioctadecylamine-tris-Ni2+-NTA, was affected by the presence of ampicillin during the PG 1348 synthesis reaction. 1349

1350 Figure 6. PG synthesis with labelled lipid II versions and detection of FRET. (A) A mixture 1351 of Atto550-lipid II, Atto647n-lipid II and unlabelled lipid II is utilized by a class A PBP with 1352 or without inhibition of the TPase activity by a β-lactam. FRET can only occur between 1353 fluorophores within the same glycan strand in linear glycan chains produced in the presence of 1354 a β-lactam (left reaction, dashed arrows). When the TPase is active (right reaction) FRET can 1355 occur either between probes within the same strand (dashed arrows) or between probes on 1356 different strands of the cross-linked PG product (solid arrows). We hypothesize that at any time 1357 only one labelled lipid II molecule occupies the two binding sites in the GTase domain and that 1358 therefore two probes within the same strand are separated by at least one subunit. As a result, 1359 average distances between probes in different strands may be shorter than between probes 1360 within the same strand and thus inter-chain FRET contributes stronger to the total FRET signal 1361 than intra-chain FRET. (B) Lipoprotein-stimulated PBPs produced short chains when labelled 1362 lipid II versions were incubated in the absence of unlabelled lipid II (e.g., Figure 1B and Figure 1363 1- figure supplement 1C). In this situation crosslinking does not occur due to the attachment of 1364 the probe to the mDAP residue in the pentapeptide. Within these short strands intra-chain FRET

1365 is stronger than within the long glycan strands depicted in (A), due to a shorter average distance 1366 between the probes. 1367 1368 Figure 1 - figure supplement 1. Fluorescent lipid II analogues to monitor PG synthesis in 1369 real time. (A) Chemical structures of lipid II analogues used for the FRET assay. R 1370 corresponds to Atto550n (donor) or Atto647n (acceptor) in the corresponding analogue. The 1371 chemical structures of alkyne versions of Atto550 and Atto647n probes that were used for 1372 derivatization are not published. Therefore the carboxylic variants are depicted here with an 1373 asterisk indicating where the alkyne versions diverge. (B) Absorbance (dashed lines) and 1374 fluorescence emission (solid lines) spectra for Atto550 (red lines) and Atto647n (blue lines). 1375 1376 Figure 1 - figure supplement 2. Analysis of fluorescence spectra to calculate FRET 1377 efficiency. Examples of deconvolution the fluorescence spectra of PG samples prepared in the 1378 presence of Lipid II-Atto550, Lipid II-Atto647n and unlabelled Lipid II, obtained from a 1379 reaction without antibiotics (A) or in the presence of ampicillin (B) or moenomycin (C). FRET

1380 efficiencies were calculated using the (ratio)A method, in which the enhancement of emission 1381 of the acceptor due to the donor is calculated by comparing the emission of (only) the acceptor 1382 when exciting at the donor excitation with the emission of the acceptor when exciting only the 1383 acceptor (Vámosi & Clegg, 1998). For this, two spectra were taken for every sample, either 1384 exciting at 552 nm (donor excitation) or at 650 nm (acceptor excitation). To process the spectra 1385 and separate the emission of the acceptor from that of the donor in the spectra taken at the 1386 donor excitation, reference spectra were measured from (1) reactions containing Lipid II- 1387 Atto550 and unlabelled Lipid II (donor reference), (2) reactions containing Lipid II-Atto647n 1388 and unlabelled Lipid II (acceptor references at both excitation wavelengths) and (3) reactions 1389 containing only unlabelled Lipid II (background references at both excitation wavelengths). 1390 Reference samples were prepared for every antibiotic condition measured. The reference 1391 spectra were then used to analyse the spectrum containing both donor and acceptor probes 1392 (black dots). Spectra taken with donor excitation was deconvolved into 3 components: donor 1393 (blue), acceptor (yellow) and background (black), while the spectrum taken with acceptor 1394 excitation was deconvolved into 2 components: acceptor (yellow) and background (black). The 1395 fitted spectra are shown in red, and the residuals of the fit are shown below each spectrum. 1396 1397 Figure 1 - figure supplement 3. FRET assay to monitor PG synthesis in real time. (A) 1398 Fluorescence emission spectra taken at the end (t=1 h) of the reactions of E. coli PBP1B shown

1399 in Figure 1E (t=60 min). (B) Aliquots at the end of the reactions shown in Figure 1E were 1400 boiled and analysed by SDS-PAGE using fluorescence detection, lanes are labelled with the 1401 reaction numbers as in Figure 1E. 1402 1403 Figure 3 - figure supplement 1. Activity of membrane-reconstituted PBP1BEc is optimal 1404 in E. coli polar lipids at low ionic strength. (A) Representative SDS-PAGE analysis of the 1405 reconstitution of PBP1BEc in liposomes made of E. coli polar lipids at a 1:3000 mol:mol 1406 protein:lipid ratio. After reconstitution, proteoliposome samples (lane 1) were centrifuged at 1407 low speed to remove aggregates and both pellet and supernatant samples were analysed (lanes 1408 2 and 3, respectively). The supernatant was subsequently used for PG synthesis reactions. A 1409 gradient of PBP1BEc (0.25, 0.41, 0.62, 0.82, 1.23 and 1.65 µg) was loaded as a standard to 1410 estimate protein concentration by densitometry. (B), (C) and (D) Representative 1411 chromatograms showing the muropeptide analysis of PG produced by detergent-solubilised 1412 PBP1BEc (B) or liposome-reconstituted PBP1BEc in the presence or absence of NaCl (C and 1413 D, respectively). The concentration of PBP1BEC was 0.5 µM and, if added, that of LpoB(sol) 1414 was 2 µM LpoB(sol). The reaction buffer contained 150 mM NaCl in B and C. Samples were 1415 incubated at 37 °C for 60 min in B and 90 min in C and D. The labelled peaks correspond to 1416 the muropeptides shown in Figure 1E. (E) Quantification of the total amount of radioactivity 1417 incorporated into PG (left) or the ratio between the radioactivity of peaks 3 and 2 (indicative 1418 of the degree of crosslinking of the PG, right) for activity assays for PBP1BEc in liposomes in 1419 the same conditions as in D. Values are mean ± SD (or variation) of at least two reactions. 1420 1421 Figure 3 - figure supplement 2. The FRET assay for PG synthesis can be adapted for 1422 reactions on liposomes. (A) Spectra corresponding to E. coli PBP1B reactions shown in Figure 1423 2C, taken at t=60 min. (B) The same gels depicted in Figure 2C, but scanned using the donor 1424 fluorescence (Atto550n). (C) Spectra corresponding to A. baumannii PBP1B reactions shown 1425 in Figure 2E, taken at t=60 min. (D) Spectra corresponding to P. aeruginosa PBP1B reactions 1426 shown in Figure 2G, taken at t=90 min. 1427 1428 Figure 3 - figure supplement 3. Moenomycin does not affect FRET on liposomes with lipid 1429 II-Atto550 and lipid II-647 in the absence class A PBPs. (A) EcPL liposomes incorporating 1430 an equimolar amount of lipid II-Atto550 and lipid II-Atto647n at 0.5%mol of the total lipid 1431 contents where incubated in the presence of 12 µM lipid II and in the presence (black line) or 1432 absence (red line) of 50 µM moenomycin for 60 min at 37 ºC while monitoring FRET as

1433 indicated in materials and methods. (B) Fluorescence spectra for the samples described in A at 1434 the end of the incubation period (t=60 min). 1435 1436 Figure 3 - figure supplement 4. Amino acid sequence comparison between LpoP 1437 homologues from A. baumannii and P. aeruginosa. (A) In the genomes of A. baumanni and 1438 P. aeruginosa the gene encoding LpoP is present within the same operon as the gene encoding 1439 their cognate PBP1B. Both LpoP proteins are predicted lipoproteins with a disordered region 1440 between the N-terminal Cys and the C-terminal globular domain containing the 1441 tetratricopeptide repeats (TPR). LpoPAb has a shorter disordered linker than LpoPPa. (B) 1442 Sequence comparison between the globular regions of LpoPAb (Ab) and LpoPPa (Pa). Proteins 1443 sequences (minus the signal peptides) were aligned using T-COFFEE EXPRESSO and the 1444 resulting alignment was visualized using JALVIEW. Residues conserved in both proteins are 1445 highlighted in a darker colour. 1446 1447 Figure 3 – figure supplement 5. LpoPAb stimulates the glycosyltransferase activity of 1448 PBP1BAb. (A) Real-time glycosyltransferase activity assays using dansyl-lipid II and 1449 detergent-solubilised A. baumannii PBP1B (PBP1BAb). PBP1BAb (0.5 µM) was mixed with 10 1450 µM dansyl-lipid II in the presence or absence of soluble 0.5 µM A. baumannii LpoP 1451 (LpoPAb(sol)). A control was performed by adding 50 µM moenomycin (black). Each data point 1452 represeants mean ± SD of 3 independent experiments. (B) Averaged initial slopes from reaction 1453 curves in A. Values are normalised relative to the slope in the absence of activator and are mean 1454 ± SD of 3 independent experiments. (C) Time-course GTase assay by SDS-PAGE followed by 1455 fluorescence detection. Detergent-solubilised PBP1BAb was incubated with 5 µM lipid II- 1456 Atto550 and 25 µM unlabelled lipid II in the presence or absence of 1.5 µM LpoPAb(sol). 1457 Reactions contained 1 mM Ampicillin to block transpeptidation. Aliquots were taken at the 1458 indicated times (in min), boiled and analysed by SDS-PAGE. A control in which only 1459 LpoPAb(sol) was present is also shown. 1460 1461 Figure 3 - figure supplement 6. PG synthesis activity of A. baumannii PBP1B in the 1462 presence of Triton X-100 followed by FRET. (A) Representative FRET curves for activity 1463 assays using detergent-solubilised A. baumannii PBP1B (PBP1BAb). PBP1BAb (0.5 µM) was 1464 mixed with unlabelled lipid II, Atto550-labelled lipid II and Atto647n-labelled lipid II at a 1:1:1 1465 molar ratio (5 µM of each), in the presence or absence of 2 µM soluble A. baumannii LpoP 1466 (LpoPAb(sol)). Controls were performed by adding 50 µM moenomycin in the absence (black)

1467 or presence (yellow) of LpoPAb(sol). Reactions were performed without antibiotic (left), with 1468 1 mM ampicillin (middle), or in the absence of unlabelled lipid II (right). The numbers indicate 1469 the corresponding lane of the gel in C. Samples were incubated for 60 min at 30°C. (B) 1470 Averaged initial slopes from reaction curves obtained by the FRET assay for detergent- 1471 solubilised PBP1BAb, in the presence (blue) or absence (red) of LpoP, and in the presence or 1472 absence of ampicillin. Values are normalised relative to the slope in the absence of activator 1473 for each condition and are mean ± SD of 2 independent experiments. (C) Aliquots after 1474 reactions in A were boiled and analysed by SDS-PAGE followed by fluorescence detection. 1475 (D) Fluorescence emission spectra taken after reactions in A (t=60 min). 1476 1477 Figure 3 - figure supplement 7. PG synthesis activity of P. aeruginosa PBP1B in the 1478 presence of Triton X-100 followed by FRET. (A) Representative FRET curves for activity 1479 assays using detergent-solubilised P. aeruginosa PBP1B (PBP1BPa). PBP1BPa (0.5 µM) was 1480 mixed with unlabelled lipid II, Atto550-labelled lipid II and Atto647n-labelled lipid II at a 1:1:1 1481 molar ratio (5 µM of each), in the presence or absence of 2 µM soluble P. aeruginosa LpoP 1482 (LpoPPa (sol)). Controls were performed by adding 50 µM moenomycin in the absence (black) 1483 or presence (yellow) of LpoPPa(sol). Reactions were performed without of antibiotic (left 1484 panel), with 1 mM ampicillin (middle panel), or in the absence of unlabelled lipid II (right 1485 panel). The numbers indicate the corresponding lane of the gel in C. Samples were incubated 1486 for 90 min at 37°C. (B) Averaged initial slopes from reaction curves obtained by the FRET 1487 assay for detergent-solubilised PBP1BPa, in the presence (blue) or absence (red) of LpoP, and 1488 in the presence or absence of ampicillin. Values are normalised relative to the slope in the 1489 absence of activator for each condition and are mean ± SD of 2-3 independent experiments. 1490 (C) Aliquots after reactions in A were boiled and analysed by SDS-PAGE followed by 1491 fluorescence detection. (D) Fluorescence emission spectra taken after reactions in A (t=90 1492 min). 1493 1494 Figure 4 - figure supplement 1. Control of membrane fluidity and integrity upon 1495 reconstitution of E. coli PBP1B. (A) The fluidity of supported lipid bilayers is reduced when 1496 increasing PBP1BEc density. The diffusion of phospholipid probe DOPE-rhodamine in the 1497 polymer-supported SLB was monitored by FRAP at different densities of PBP1B. The fluidity 1498 of the membrane decreased (black line) while the immobile fraction increased (orange line) 1499 with higher protein densities. 1500

1501 Figure 4 - figure supplement 2. E. coli PBP1B is active after reconstitution in supported 1502 lipid bilayers. (A) and (B) PBP1BEc was reconstituted on supported lipid bilayers prepared 1503 with E. coli polar lipid extract in 1.1 cm2 chambers. The protein to lipid ratio was 1:105 1504 (mol:mol). Reactions were started by adding 1 nmol of radiolabelled lipid II per chamber, in 1505 the presence of LpoB(sol) (4 µM) moenomycin (100 µM). Three chambers were prepared for 1506 each condition and samples were combined before the analysis. Chambers were incubated 1507 overnight at 37 °C and the reaction was stopped by adding moenomycin. Cellosyl and Triton 1508 X-100 were added to solubilize the membranes and digest the PG product. The resulting 1509 muropeptide samples were concentrated, reduced with sodium borohydride and analysed by 1510 HPLC. Full chromatograms are shown in A, while zoomed-in chromatograms are shown in B. 1511 (C) and (D) PG synthesis occurs only in the membrane fraction of SLBs. PBP1BEc was 1512 reconstituted on SLBs as in A and B. In addition, control chambers were prepared without 1513 PBP1B. Chambers were incubated over night to allow for PG synthesis and then washed with 1514 fresh buffer. The washes and chambers (membranes) were treated and analysed as described 1515 for A and B. Five chambers were combined for reactions with PBP1BEc, and four chambers for 1516 control reactions. Full chromatograms are shown in C, while zoomed-in chromatograms are 1517 shown in D. The labelled peaks in all chromatograms correspond to the muropeptides shown 1518 in Figure 1F. 1519 1520 Figure 5 - figure supplement 1. Control of membrane fluidity and integrity during the 1521 FRET assay. (A) Fluorescence intensity profiles 1s after photobleaching taken from the 1522 images depicted on Figure 4B. (B) Montage comparing the recovery of fluorescence after

1523 photobleaching of a tracer (DODA-tris-Ni-NTA plus a His6-tagged peptide labelled with 1524 AlexaFluor 488) with the one of lipid II-Atto647n on a supported lipid bilayer containing 1525 PBP1B at a 1:105 protein:lipid (mol:mol) ratio. The assay was performed after a PG synthesis 1526 reaction carried out for 1.5 h. The fact that fluorescence is recovered for both, indicates that 1527 the membrane remains fluid while lipid II stays diffusive after the synthesis reaction. 1528 1529 Video 1. Single-molecule imaging of PBP1B on supported lipid bilayers. PBP1BEc-Dy647 1530 was reconstituted in EcPL SLBs at a 1:106 (mol:mol) protein to lipid ratio and was tracked 1531 using single-molecule TIRF before or after the addition of 1.5 µM lipid II. Images were taken 1532 with a rate of 62 ms per frame. 1533

1534 Video 2. FRET assay on supported lipid bilayers. PBP1BEc was reconstituted in EcPL SLBs 1535 at a 1:105 (mol:mol) protein to lipid ratio along lipid II-Atto647, lipid II-Atto550. Membranes 1536 were incubated with 5 µM lipid II in the presence or absence of 1 mM ampicillin. To detect 1537 FRET, the fluorescence of the acceptor Atto647n was bleached within a region. In the 1538 subsequent frame the fluorescence of Atto550 increased indicating the presence of FRET. In 1539 the presence of ampicillin this increase did not happen. 1540 1541 Supplementary File 1: table of oligonucleotides used in this study. 1542

Key Resources Table

Reagent type Designation Source or Identifiers Additional (species) or reference information resource strain, strain BL21(DE3) New England C2527 background Biolabs (Escherichia coli)

Recombinant pDML219 Bertsche et Expression of DNA reagent al., 2006 N-terminal His-tagged E. coli PBP1B.

Recombinant pKPWV1B This paper Expression of DNA reagent N-terminal His-tagged A. baumannii 19606 (ATCC) PBP1B.

Recombinant pAJFE52 Caveney et Expression of DNA reagent al., 2020 N-terminal His-tagged P. aeruginosa PBP1B.

Recombinant pMGCPBP1BC This paper Expression of DNA reagent S1CS2 E. coli PBP1B version with a single Cys residue in the N-terminus and C- terminal His- tag.

Recombinant pET28His- Egan et al., Expression of DNA reagent LpoB(sol) 2014 soluble version of E. coli LpoB with an N- terminal His- tag.

Recombinant pKPWVLpoP This paper Expression of DNA reagent N-terminal His-tagged A. baumannii 19606 (ATCC) LpoP.

Recombinant pAJFE57 Caveney et Expression of DNA reagent al., 2020 soluble version of P. aeruginosa LpoP with an N-terminal His-tag. sequence- PBP1B.Acineto This paper PCR cloning AGATATCAT based -NdeI_f primers ATGATGAAG reagent TTTGAACGT GGTATC GGTTTCTTC sequenced- PBP1B.Acineto This paper PCR cloning GCGGGATC based -BamHI_r primers CTTAGTTGT reagent TATAACTAC CACTTGA AATG sequenced- Seq1_rev_PBP This paper PCR cloning AGGTTCTAA based 1B_Acineto primers ACGGGCAA reagent CTC sequence- Seq2_fwd_PB This paper PCR cloning TGGTTATGG based P1B_Acineto primers ATTGGCCTC reagent TC sequenced- Seq3_fwd_PB This paper PCR cloning CTGGGCAA based P1B_Acineto primers GCCAGATTG reagent AAG sequenced- Seq4_fwd_PB This paper PCR cloning ACAATTACG based P1B_Acineto primers CCAGACACC reagent AG sequence- PBP1B-MGC-F This paper PCR cloning CATCATCCA based primers TGGGCTGTG reagent GCTGGCTAT GGCTACTGC TA sequence- PBP1B- This paper PCR cloning CATCATCTC based CtermH-R primers GAGATTACT reagent ACCAAACAT ATCCTT sequence- C777S-D This paper PCR AACTTTGTT based mutagenesis TCCAGCGGT reagent primers GGC

sequence- C777S-C This paper PCR GCCACCGCT based mutagenesis GGAAACAAA reagent primers GTT sequence- C795S-D This paper PCR CAATCGCTG based mutagenesis TCCCAGCAG reagent primers AGC sequence- C795S-C This paper PCR GCTCTGCTG based mutagenesis GGACAGCG reagent primers ATTG

Chemical [14C]GlcNAc- Breukink et compound labelled Lipid II al., 2003 (mDAP) Bertsche et al., 2005

Chemical Lipid II (mDAP) Egan et al., compound 2015

Chemical Lipid II (Lys) Egan et al., compound 2015

Chemical Lipid II-dansyl Egan et al., compound 2015

Chemical Lipid II-Atto550 Mohammadi compound et al., 2014 van’t Veer et al., 2016

Chemical Lipid II- Mohammadi compound Atto647n et al., 2014 van’t Veer et al., 2016

Chemical Polar lipid Avanti Polar 100600P compound extract from E. Lipids coli (EcPL)

Chemical 1,2-dioleoyl-sn- Avanti Polar 850375P compound glycero-3- Lipids phosphocholin e (DOPC)

Chemical 1-palmitoyl-2- Avanti Polar 840457P compound oleoyl-sn- Lipids glycero-3-

phospho-(1'- rac-glycerol) (POPG)

Chemical Tetraoleoyl Avanti Polar 710335P compound cardiolipin Lipids

Chemical Dy647P1- Dyomics 647P1-03 compound maleimide probe

Chemical Alexa Fluor ThermoFisher A10254 compound 488 C5 Scientific Maleimide

Chemical Alexa Fluor ThermoFisher A20346 compound 555 C2 Scientific maleimide

Chemical Triton X-100 Roche 10789704001 compound

Chemical Moenomycin Sigma 32404 compound

Chemical Ampicillin Sigma A9518 compound

Chemical Methyl-β- Sigma-Aldrich 332615 compound cyclodextrin

Chemical Poly(ethylene Sigma-Aldrich 81323 compound glycol) methyl ether

average Mn 5,000

Chemical 1,2-dioleoyl-sn- Avanti Polar 810150C compound glycero-3- Lipids phosphoethan olamine-N- (lissamine rhodamine B sulfonyl) (DOPE- Rhodamine)

Chemical dioctadecylami Beutel et al., compound ne (DODA)- 2014 tris-Ni-NTA

Chemical cOmplete™, Roche 5056489001 compound EDTA-free Molecular Protease Biochemicals

Inhibitor Cocktail

Chemical Phenylmethyls Sigma-Aldrich P7626 compound ulfonylfluoride (PMSF)

Chemical Ni-NTA Qiagen 1018142 compound superflow resin

Chemical Bio-Beads SM- Bio-Rad 1523920 compound 2 resin

Commercial Pierce BCA ThermoFisher 23227 assay, kit Protein Assay Scientific Kit

Commercial HiTrap SP HP GE 17115101 assay, kit column, 1 mL biosciences

Commercial HiTrap GE 17140801 assay, kit Desalting biosciences column, 5 ml

Commercial Prontosil 120- BISCHOFF 1204F184P3 assay, kit 3-C18 AQ Chromatogra reversed- phy phase column

Peptide, DNase ThermoFisher 90083 recombinant Scientific

Peptide, Cellosyl Hoescht Mutanolysin recombinant (Germany) from Sigma protein (M9901) can also be used.

Peptide, MepM Federico recombinant Corona, protein following protocol in Singh et al., 2012

Chemical His6-tagged BioMatik CMSQAALNTR compound (on the C- NSEEEVSSRR terminus) NNGTRHHHHH neutral peptide H

Software, Fiji https://fiji.sc algorithm

Software, Matlab MathWorks https://www. algorithm mathworks.c om

Software, frap_analysis Jönsson, algorithm 2020 1543 1544 1545

1546

A GTase TPase + + FRET

Lipid II Lipid II-Atto550 Lipid II-Atto647n (donor) (acceptor) cross-linked peptidoglycan

B Atto550 Atto647n C D fluorescence fluorescence Donor emission I II III I II III Crosslinked PG 800 no AB (I) 35 +Amp (II) a.u ) +Moe (III) 30 600 25

Linear PG Acceptor 20 400 chains emission 15

200 FRET efficiency (%) 10

Fluorescence ( intensity 5 0 Lipid II 0 600 650 700 750 no AB +Amp +Moe Wavelength (nm) E PBP1BEc +/- LpoB F 1.8 no AB Amp / Moe Only labelled Lipid II 30 PBP1BEc +LpoB (V) +LpoB 1.6 25 1.4 +LpoB/Moe (VII) +LpoB/only 20

donor 1.2 labelled (VIII)

+Moe (III) FI PBP1B (I) / 1.0 15 +LpoB/Amp (VI)

acceptor 0.8 10 FI 0.6 Fold change in initialslope 5 0.4 +Amp (II) Only labelled (IV) 0 0 20 40 60 0 20 40 60 0 20 40 60 no AB +Amp Time (min)

A B Normalised fluorescence ( ( ) fluorescence Normalised

Atto550 Atto647n 1.0 1.0 0.8 0.8 0.6 0.6 Atto550 Atto647n 0.4 0.4

0.2 0.2 0.0 0.0 Normalised absorption ) ( 400 500 600 700 800 Wavelength (nm)

A Atto550 excitation Atto647n excitation B Atto550 excitation Atto647n excitation (552 nm) (650 nm) (552 nm) (650 nm) 400 300 700 600

350 a.u ) a.u ) 250 600 300 500 500 200 150 400 400 150 200 300 300 150 100 200 100 200 50

Fluorescence intensity ( 100 Fluorescence intensity ( 50 100 0 0 0 0 20 10 10 0 0 0 0 -10 Residuals Residuals -10 -20 600 650 700 750 650 675 700 725 750 600 650 700 750 650 675 700 725 750 Wavelength (nm) Wavelength (nm) Wavelength (nm) Wavelength (nm)

C 700 800 600 a.u )

600 500 400 400 300

200 200

Fluorescence intensity ( 100

0 0 20 20 0 0

Residuals -20 600 650 700 750 650 675 700 725 750 Wavelength (nm) Wavelength (nm) A PBP1B (I) +LpoB (V) +Amp (II) +LpoB/Amp (VI) 80

40 a.u )

0 100 +Moe (III) +LpoB/Moe (VII) Only labelled +LpoB/only (IV) labelled (VIII) 80 Fluorescence intensity ( 60

0 550 600 650 700 550 600 650 700 550 600 650 700 550 600 650 700 Wavelength (nm)

B Atto550 fluorescence Atto647n fluorescence I II III IV V VI VII VIII I II III IV V VI VII VIII Crosslinked PG

Linear PG chains

Lipid II A no AB +Amp B U M C U M C Crosslinked Undigested PG 30 MepM Cellosyl 25

20 Linear PG chains 15

10 FRET efficiency (%) 5

Lipid II / 0 digested MP no AB +Ampicillin

C 2 Donor emission 3 60 min D 1 300

40 min a.u ) 250 00 2 200 20 min Acceptor

150 emission 10 min 100

5 min Fluorescence ( intensity 0 560 580 600 620 640 660 680 700

Radioactivity (CPM) Radioactivity Wavelength (nm) 0 min E

Peaks 2+3 0 10 20 30 40 50 60 70 80 3 Elution time (min)

1 2 3 60 FI acceptor G - M-P G - M(r) G - M(r) Peak 2 2

L-Ala L-Ala L-Ala 40 / FI

D-Glu D-Glu D-Glu D-Ala FRET donor 1

m-Dap m-Dap m-Dap D-Ala 20

D-Ala D-Ala D-Ala m-Dap % radioactivity in peaks Peak 3 D-Ala D-Ala D-Glu 0 0 L-Ala 0 10 20 30 40 50 60 Time (min) G - M(r) A MGC-64PBP1BC777S/C795S-his B peptidoglycan +/- Lpo activator TP Class A PBP Alexa555-maleimide GTase & TPase GT -TX +TX lipid II

-Cys Coomassie

Fluorescence lipid II lipid II Atto550 Atto647

C E. coli PBP1B liposomes +/- LpoB Atto647n fluorescence D Ec no AB Amp / Moe Only labelled Lipid II I II III IV V VI VII VIII 35 PBP1B 2.0 +LpoB 30 +LpoB (II) 1.5 25 donor

20 / +LpoB/Moe (IV)

1.0 +LpoB/Amp (III) 15 acceptor

FI PBP1B (I) 10 +LpoB (VI) +Moe (VIII) 0.5 PBP1B (V) 5

+Amp (VII) Fold change in initialslope 0

0 20 40 60 0 20 40 60 0 20 40 60 no AB +Amp

Time (min)

E A. baumannii PBP1B liposomes +/ - LpoP F no AB Amp / Moe I II III IV V VI PBP1BAb Crosslinked 4 1.8 +LpoPAb PG

1.6 slope +LpoP/Amp (V) 3 +LpoP (IV)

donor 1.4 FI

1.2 Linear PG 2

1.0 PBP1B (I) +LpoP/Moe (VI) chains acceptor

FI 0.8 +Amp (II) 1 0.6

+Moe (III) Fold change in initial 0.4 Lipid II 0 0 20 40 60 0 20 40 60 no AB +Amp

Time (min)

G P. aeruginosa PBP1B liposomes +/ - LpoP H Pa 1.2 no AB Amp / Moe I II III IV V VI Crosslinked 12 PBP1B PG +LpoPPa +LpoP (IV) 10 1.0 +LpoP/Amp (V)

donor 8

FI 0.8

/ Linear PG chains 6 0.6 +Moe (III)

acceptor PBP1B (I) 4

FI +LpoP/Moe (VI) 0.4 +Amp (II) 2 Lipid II Fold change in initialslope 0.2 0 0 20 40 60 80 0 20 40 60 80 no AB +Amp Time (min) Ec A 1 2 3 PBP1B standard 170 130– 100– 70– 55–

B 3 C 1

2 3

400 2 400 1 +LpoB +LpoB

2 3 1

PBP1B 2 3 PBP1B liposomes Radioactivity (CPM) Radioactivity Radioactivity (CPM) Radioactivity

0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 Elution time (min) Elution time (min) D E 3 100 2.5 PBP1B (A) +LpoB (B)

200 2 80 2.0 1 +LpoB 60 1.5

40 1.0 PBPB Peak 3 / Peak 2 liposomes % PG products 20 0.5

(no salt)

Radioactivity (CPM) Radioactivity 0 0.0 0 10 20 30 40 50 60 70 A B A B Elution time (min) A PBP1BEc (I) +LpoB (II) +Moe (VIII) +LpoB/Moe (IV) B Atto550 fluorescence 200 I II III IV V VI VII VIII

150

100 a.u ) 50

0 200 +Amp (VII) +LpoB/Amp (III) Only labelled +LpoB / (V) Only labelled (VI) 150 Fluorescence ( intensity 100

0 550 600 650 700 550 600 650 700550 600 650 700550 600 650 700 Wavelength (nm) C 180 PBP1BAb (I) +LpoPAb (IV) +Moe (III) +LpoPAb/Moe (VI) +Amp (II) +LpoPAb/Amp (V)

a.u ) 160 140 120 100

80 60 40

Fluorescence ( intensity 20 0 550 600 650 700 550 600 650 700 550 600 650 700 550 600 650 700 550 600 650 700 550 600 650 700 D Wavelength (nm) 180 PBP1BPa (I) +LpoPAb (IV) +Moe (III) +LpoPPa/Moe (VI) +Amp (II) +LpoPPa/Amp (V)

a.u ) 160 140 120 100

80 60 40

Fluorescence ( intensity 20 0 550 600 650 700 550 600 650 700 550 600 650 700 550 600 650 700 550 600 650 700 550 600 650 700 Wavelength (nm) A B 0.38 LUVs +Moe 100 0.36 a.u ) 80 0.34 donor +Moe FI

/ 0.32 60

0.30

acceptor 40 FI 0.28 LUVs 20 0.26 Fluorescence ( intensity 0.24 0 0 10 20 30 40 50 60 550 600 650 700 550 600 650 700 Wavelength (nm)

A mrcB (PBP1B) lpoP

13 17 disordered 115 145 154 180 A. baumannii 1 MVGC TPR TPR 183

16 20 disordered 190 219 229 255 P. aeruginosa 1 MVGC TPR TPR 259

B A 5 100 B

90 4 +moe 80 3 PBP1BAb 2 70

60 1 Relative fluorescence (%) +LpoP Fold change in initialslope 50 0 0 10 20 30 40 50 60 PBP1B +LpoP Time (min)

C PBP1BAb PBP1BAb + LpoPAb LpoPAb 0 2 5 10 30 60 0 2 5 10 30 60 0 2 5 10 30 60 time (min)

Linear PG chains

Lipid II Acinetobacter baumannii PBP1B +/- LpoP A B PBP1BAb 0.9 no AB Amp / Moe Only labelled Lipid II 16 +LpoPAb 14 0.8 12 0.7 donor 10

FI +LpoP (V) +LpoP/only

+LpoP/Moe (VII) / labelled (VIII) 0.6 +LpoP/Amp (VI) 8 PBP1B (I) 6 acceptor +Amp (II)

FI 0.5 4 Only labelled (IV) 0.4 Fold change in initialslope 2 +Moe (3) 0 0 20 40 60 0 20 40 60 0 20 40 60 no AB +Amp Time (min)

C Atto550 fluorescence Atto647n fluorescence I II III IV V VI VII VIII I II III IV V VI VII VIII Crosslinked PG

Linear PG chains

Lipid II

D PBP1BAb (I) +LpoPAb (V) +Moe (III) +LpoPAb/Moe 100 (VII)

40 a.u )

0 100 +Amp (II) +LpoPAb/Amp (VI) Only labelled +LpoPAb/only (IV) labelled (VIII) 80

Fluorescence ( intensity 60

0 550 600 650 700 550 600 650 700 550 600 650 700 550 600 650 700 Wavelength (nm)

A Pseudomonas aeruginosa PBP1B +/- LpoP B 30 Pa 1.8 no AB Amp / Moe Only labelled Lipid II PBP1B Pa 25 +LpoP 1.6 +LpoP (V)

1.4 20

1.2 15

1.0 PBP1B (I) +Moe (III) +LpoP/Moe (VII) Only labelled (IV) 0.8 10 +LpoP/Amp (VI) +LpoP/only 0.6 labelled (VIII) Fold change in initialslope 5

FRET signal / donor signal +Amp(II) 0.4 0 0 20 40 60 0 20 40 60 0 20 40 60 no AB +Amp Time (min)

C Atto550 fluorescence Atto647n fluorescence I II III IV V VI VII VIII I II III IV V VI VII VIII Crosslinked PG

Linear PG chains

Lipid II

D PBP1BPa (I) +LpoPPa (V) +Moe (III) +LpoPPa/Moe 100 (VII)

a.u ) 40

0 +Amp (II) +LpoPPa/Amp Only labelled +LpoPPa/only 100 (VI) (IV) labelled (VIII) 80 Fluorescence ( intensity

0 550 600 650 700 550 600 650 700 550 600 650 700 550 600 650 700 Wavelength (nm)

A B C peptidoglycan

PBP1BEc-Dy647 GTase 1.2 PBP1B+lipid II TPase TP 1.0

0.8 PBP1B GT lipid II 0.6 PDF

0.4

PEG 0.2 film 0.0 0.01 0.1 1 D (µm2 /s) TIRF microscopy 10 µm coef

D E Fast Fast

Ec Slow Slow PBP1B -Dy647 PBP1B PBP1B + lipid II 4

3 ) - 3 10 × PBP1BEc-Dy647 + Lipid II 2 PDF ( 1

0 5 µm 0 0.25 0.5 0.75 1.0 0 0.25 0.5 0.75 1.0 Displacement (µm)

100

3.0 (%) fraction Immobile 80 2.5 /s) 2 60 2.0 (µm 40

coef 1.5 D 1.0 20

0.5 0 10-6 10-5 10-4 10-3 ratio PBP1B:EcPL A 1 B 1

+Moe 100 1000 1 +Moe

2 Radioactivity (CPM) Radioactivity Radioactivity (CPM) Radioactivity 3

PBP1B + PBP1B + 2 LpoB 3 LpoB

0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 Elution time (min) Elution time (min)

C 1 D 1

PBP1B - wash

PBP1B - wash 1000 1 2

100 PBP1B - 3 membrane

PBP1B - 2 3 membrane 1

Control - wash Radioactivity (CPM) Radioactivity Radioactivity (CPM) Radioactivity Control - wash

1 Control - Control - membrane membrane

0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 Elution time (min) Elution time (min) A Atto647n Atto550 B C (acceptor) (donor) PBP1B/LpoB 0.20 0.5

0.4 0.15

0.3 0.10 Before FRAP

0.2

0.05 FRET efficiency FRET efficiency PBP1B/LpoB +Amp 0.1 0.0

after FRAP 0.0

st 0 5 10 15 20 25 1 Time (min)

D E F PBP1B/LpoB PBP1B/LpoB + Amp 3.5 PBP1B/LpoB +Amp 35 3.5 3.0 30 3.0

2.5 25 2.5 /s) /s) 2 2.0 20 2 2.0 (µm (µm 1.5 15 1.5 coef coef D

PBP1B/LpoB D 1.0 10 1.0 Immobilefraction (%) 0.5 5 0.5

0.0 0 0.0 0 5 10 15 20 25 PBP1B PBP1B lipid II SLB lipid II SLB Time (min) +Amp probe probe A B Supported lipid membrane probe Lipid II-Atto647n 1000 8000 900 7000

800 6000 intensity ( Atto647n fluorescence 5000 a.u .) 700 4000 600 3000

500 2000 Before photobleaching

2200 6000 2000 5500 5000 1800 4500 .) a.u

Atto550 fluorescenceAtto550 ( intensity 1600 4000 1400 3500 1 s 1200 3000 1000 2500 0 5 10 15 20 25 Distance (µm) 65 s A

FL-Lipid II versions + unlabelled Lipid II β-lactam GTase (only GTase) & TPase

+ +

Glycan strands Lipid II Lipid II- Lipid II- Crosslinked (low FRET) Atto550 Atto647n peptidoglycan (donor) (acceptor) (high FRET) B Only FL-Lipid II GTase

Lipid II- Lipid II- Short glycan chains Atto550 Atto647n (high FRET) (donor) (acceptor)