A feedback control mechanism governs the synthesis of lipid-linked precursors of the bacterial cell wall

SUMMARY Many bacterial surface glycans such as the peptidoglycan (PG) cell wall, O-antigens, and capsules are built from monomeric units linked to a polyprenyl lipid carrier. How this limiting lipid carrier is effectively distributed among competing pathways has remained unclear for some time. Here, we describe the isolation and characterization of hyperactive variants of Pseudomonas aeruginosa MraY, the essential and conserved enzyme catalyzing the formation of the first lipid-linked PG precursor called lipid I. These variants result in the elevated production of the final PG precursor lipid II in cells and are hyperactive in a purified system. Amino acid substitutions within the activated MraY variants unexpectedly map to a cavity on the extracellular side of the dimer interface, far from the active site. Our structural evidence and molecular dynamics simulations suggest that the cavity is a binding site for lipid II molecules that have been transported to the outer leaflet of the membrane. Overall, our results support a model in which excess externalized lipid II allosterically inhibits MraY, providing a feedback mechanism to prevent the sequestration of lipid carrier in the PG biogenesis pathway. MraY belongs to the broadly distributed polyprenyl-phosphate N-acetylhexosamine 1-phosphate transferase (PNPT) superfamily of enzymes. We therefore propose that similar feedback mechanisms may be widely employed to coordinate precursor supply with demand by polymerases, thereby optimizing the partitioning of lipid carriers between competing glycan biogenesis pathways.

6 minimal medium (Vogel-Bonner minimal medium, VBMM) 18 . Spontaneous suppressors 139 supporting the growth of the PBP1a-only mutant on VBMM medium were isolated to uncover 140 new insights into PG synthesis regulation. Several of these mutants were found to encode 141 variants of Pa PBP1a, and we previously reported that they bypass the Pa LpoA requirement for DlpoA cells on VBMM than the tagged wild-type protein (Fig. S2). Thus, the suppression 164 activity of the Pa MraY(T23P) variant is not due to increased accumulation of the enzyme. 165 Rather, the results suggest that the T23P change alters MraY activity to promote the growth of 166 the aPBP deficient strain on VBMM and impair growth of both mutant and wild-type strains on 167 LB when it is overexpressed. 168 169 E. coli also encodes aPBPs, Ec PBP1a and Ec PBP1b, controlled by OM lipoprotein activators 170 Ec LpoA and Ec LpoB, respectively ( Fig. 1C) 19,21 . We previously described an E. coli strain 171 lacking Ec PBP1a and Ec LpoB that relies on a LpoB-bypass variant of Ec PBP1b 172 [ Ec PBP1b(E313D)] as its only aPBP (Fig. 1C) 22 . Like the P. aeruginosa DponB DlpoA strain, 173 this E. coli mutant has a conditional growth defect caused by a deficit in aPBP activity. It grows 174 on LB without added NaCl (LBNS) but is inviable on LB with 1% NaCl. Overproduction of E. 175 coli MraY(T23P) [ Ec MraY(T23P)] but not wild-type Ec MraY suppressed the growth defect of this 176 aPBP-deficient E. coli strain on LB 1% NaCl (Fig. 1D). Therefore, an MraY(T23P) variant 177 suppresses an aPBP defect in two distantly related gram-negative bacteria, suggesting that its 178 properties are conserved. 179 180

MraY(T23P) is activated and increases lipid II accumulation in cells 181
MraY uses UM5 and C55P to form the first lipid-linked PG precursor lipid I, which is then 182 converted to the final precursor lipid II by MurG. We reasoned that MraY(T23P) might 183 overcome the aPBP-deficiency in mutants of P. aeruginosa and E. coli by increasing the 184 concentration of the synthase substrate lipid II to compensate for the poorly activated aPBP in 185 these cells. To investigate this possibility, we measured the concentration of lipid II in P. 186 aeruginosa and E. coli cells overproducing MraY(WT) or MraY(T23P). Exponentially growing 187 8 cultures were normalized by optical density, and the cells were harvested and extracted for 188 lipid-linked PG precursors ( Fig. 2A). The extract was subjected to acid hydrolysis to release 189 the disaccharide-pentapeptide from undecaprenyl-pyrophosphate (C55PP), and the soluble 190 disaccharide-pentapeptide was subsequently detected by liquid chromatography/mass 191 spectrometry (LCMS) as a measure of lipid II concentration (Fig. 2B-E). In both the wild-type 192 and aPBP deficient mutant backgrounds, MraY(WT) overproduction led to an approximately 193 twofold increase in lipid II levels relative to an empty vector control (Fig. 2C and 2E). The 194 increase was another twofold higher for cells overproducing MraY(T23P) (Fig. 2C and 2E). We 195 observed similar trends monitoring lipid I levels, but the increase in lipid I levels in cells 196 producing MraY(T23P) relative to MraY(WT) was not nearly as pronounced compared to the 197 change in lipid II levels (Fig. S3). These results suggest that the altered MraY enzyme is more 198 active than wild-type and that the ability to promote the accumulation of higher lipid II levels 199 indeed underlies the suppression of aPBP defects. affinity purified for biochemical assays. The reaction was followed by monitoring the production 204 of uridine derived from alkaline phosphatase treatment of the UMP product (Fig. 2F). Using 205 this assay, the Pa MraY(T23P) variant was found to be significantly more active than 206 9 The results thus far suggest that Pa MraY(T23P) makes more lipid-linked PG precursors than 213 normal, leading to their hyperaccumulation. In the PBP1a-only P. aeruginosa strain, this 214 increase in substrate supply suppresses the lethal aPBP deficiency. We wondered whether 215 excess lipid II production and the resulting sequestration of C55P in this building block might 216 also indirectly impede the synthesis of other surface glycans built on the lipid carrier like O-Ag. 217 A clue that this was the case came from the growth defect on LB medium of the wild-type 218 PAO1 strain caused by Pa MraY(T23P) but not Pa MraY(WT) overproduction ( Fig. 1B and Fig.  219   S1). Notably, this strain produces R2-pyocin, a lethal phage tail-like bacteriocin that uses a 220 receptor located within the LPS core to engage target cells 23,24 . P. aeruginosa is resistant to 221 killing by its own R2-pyocin because it decorates its LPS with O-Ag that masks the R2-pyocin 222 receptor. Defects in the O-Ag synthesis pathway therefore result in susceptibility to R2-pyocin 223 self-killing 25 . The connection between O-Ag and R2-pyocin activity suggested to us that the 224 growth phenotype induced by Pa MraY(T23P) overproduction on LB medium may be caused by 225 a decrease in O-Ag production and increased R2-pyocin self-intoxication. To test this 226 possibility, we examined the effect of Pa MraY(T23P) overproduction in a strain deleted for the 227 R2-pyocin gene cluster (PA0615-PA0628). Strikingly, unlike wild-type cells, the mutant 228 incapable of making R2-pyocin was largely unaffected by the overproduction of Pa MraY(T23P) 229 ( Fig. S4A), indicating that the growth defect caused by the altered enzyme was largely due to 230 R2-pyocin killing. This result suggested that O-Ag synthesis is reduced when lipid II synthesis 231 is hyperactivated in cells producing Pa MraY(T23P). Analysis of the LPS produced by these 232 cells confirmed that they indeed have reduced levels of O-Ag. They made approximately 30% 233 less O-Ag compared to cells expressing Pa MraY(WT) (Fig. S4B-C). These results suggest that 234 10 238

The extracytoplasmic side of the MraY dimer interface may be a regulatory site 239
MraY is a polytopic membrane protein with ten transmembrane helices and an N-out, C-out 240 topology 26 . The structure of the enzyme from Aquifex aeolicus revealed that it forms a dimer 241 with most of the monomer-monomer contacts made between the N-and C-terminal helices 26 . 242 Notably, the T23 residue lies near the dimer interface on the extracytoplasmic side of MraY. 243 We therefore wondered whether other substitutions in this area might also activate the 244 enzyme. To test this possibility, a mutagenized copy of Pa mraY under the control of an IPTG 245 inducible promoter was transformed into the DponB DlpoA P. aeruginosa strain. The resulting 246 transformants were then selected on VBMM in the presence of IPTG to identify MraY variants 247 that rescue the aPBP deficiency. Twenty-one suppressing clones were isolated that each 248 contained a single point mutation in the plasmid-borne copy of mraY (Fig. 3A). The positions 249 of these substitutions were mapped onto a model of the Pa MraY structure generated using 250 AlphaFold 27,28 . Strikingly, all changes were located proximal to the dimer interface, with a 251 majority positioned on the extracytoplasmic side of the protein far from the active site, which is 252 located on the cytoplasmic side of the enzyme ( Fig. 3B-C , Table S1). Overall, our genetic and 253 biochemical results implicate the extracytoplasmic region of MraY near the dimer interface as a 254 potential regulatory site for the enzyme. 255

A potential binding site for flipped lipid II within the MraY dimer interface 257
Both the A. aeolicus and Enterocloster boltae MraY crystal structures revealed the presence of 258 a cavity located at the dimer interface that is lined by hydrophobic residues 26,29 . The authors 259 concluded that electron density within this tunnel could accommodate a cylindrical molecule 260 that is too long to be detergent from the sample preparation 26 . Instead, they suggested that this 261 electron density could accommodate one or more lipid molecules. Although it has been 262 11 speculated to be C55P 26 , the identity of the lipid has remained unclear. Additionally, a recent 263 study identified lipid molecules co-purifying with MraY using native mass-spectrometry 30 . The 264 most abundant species were the C55P substrate and lipid I product, but peaks corresponding 265 to C55PP, cardiolipin, and lipid II were also detected 30 . Thus, MraY likely binds a lipid molecule 266 within the dimer interface near residues we have implicated in controlling the activity of the 267 enzyme.  containing wild-type Ec MraY was recently reported 31 , and these methodologies were used to 273 obtain the structure of Ec MraY(T23P) within the same complex (Fig. S5, Table S2). In both 274 cases, electron density was observed at the MraY dimer interface. Focused refinement of 275 MraY alone in the Ec MraY(T23P) complex significantly improved the potential lipid density at 276 the MraY dimer interface (Fig. 4A-B). As in previous A. aeolicus and E. boltae MraY 277 structures, this electron density fills the hydrophobic cavity found at the MraY dimer interface. 278 However, we uniquely observed this electron density extending into the periplasmic space 279 above the MraY molecules where the environment is more hydrophilic (Fig. 4A-B). Although 280 structural refinement alone could not conclusively identify the lipid within the dimer, the size of 281 the electron density extending into the periplasmic space is consistent with a large head-group 282 such as the disaccharide-pentapeptide found on lipid II. 283

284
To assess whether a lipid II molecule could enter the hydrophobic cavity of the MraY dimer, we 285 used molecular dynamics (MD) simulations. In the first set of simulations, we used the structure 286 of the E. coli MraY dimer from the YES complex (PDBID 8G01) 31 embedded in a lipid bilayer 287

12
[palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine (POPE), palmitoyl-2-oleoyl-sn-glycero-288 3-phosphoglycerol (POPG) and cardiolipin (CDL)] containing C55P, C55PP, lipid I, or lipid II with 289 hydrophilic head-groups oriented towards what would be the periplasmic side of the membrane. 290 Using course-grained MD simulations we observed that in almost all runs, lipid I and lipid II 291 molecules spontaneously entered the central cavity, where typically two molecules would occupy 292 the cavity (Fig. 4C, Movie S1 lipid I is not found in the periplasmic leaflet of the inner membrane. Therefore the simulations 301 with lipid I are not likely to reflect a physiologically relevant binding event. Instead, lipid II is the 302 best candidate for the native ligand due to its strong and long-lasting interaction. Notably, the 303 bound lipid II molecules in the simulations make extensive contacts with the MraY dimer, with 304 many residues contacting the bound lipids for 100% of the MD simulations (Fig 4D, Fig. S6). 305 These residues include several that were identified in the mutational analysis as being 306 hyperactive ( Fig. 3A-B). To investigate the interaction in more detail, a pose of the E. coli MraY 307 dimer with two bound lipid II molecules was converted to an atomistic description for further MD 308 analysis. The data show that the lipid II molecules are stable in the central cavity with the 309 isoprenyl chains adopting a curved orientation. The result predicts contacts between the MurNAc 310 sugar and MraY that include several residues where substitutions were identified in our screen 311 (Y21, L22, T23, W217, F224, Y227, and K358) (Fig. 4E-F). Together, these data indicate that 312 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made Bacterial surfaces contain multiple types of glycan and other polymers that are required for 320 cellular integrity and/or barrier function. Although most of the proteins involved in the synthesis 321 of major surface components are known, how the biogenesis of these molecules is regulated 322 to efficiently distribute shared precursors like the C55P lipid carrier among competing 323 synthesis pathways remains poorly understood. In this report, we uncover a mechanism 324 governing the activity of MraY, the essential enzyme catalyzing the first membrane step in the 325 PG synthesis pathway in which C55P is consumed to form lipid-linked PG precursors. This 326 regulation is likely to play an important role in the efficient distribution of C55P among glycan 327 biogenesis pathways that utilize the limiting carrier. 328 329 The first clue that MraY is regulated came from the discovery that an mraY(T23P) mutant can 330 suppress an aPBP deficiency in both P. aeruginosa and E. coli. The aPBP deficient strains 331 encode a single aPBP lacking its required activator. Prior work with these strains suggests that 332 their conditionally lethal growth phenotypes are caused by poor PG synthesis efficiency 333 resulting from the synthase having a reduced affinity for lipid II in the absence of its activator 20 . 334 Accordingly, we infer that MraY(T23P) suppresses this problem by raising the steady state 335 level of lipid II to overcome the substrate binding limitations of the unactivated aPBP. The 336 ability of the altered MraY to increase lipid II levels indicates a role for the enzyme in regulating 337 14 the maximum level of lipid II in cells. We propose that this control is mediated via feedback 338 inhibition of MraY by externalized lipid II (Fig. 5). 339

340
In support of the feedback inhibition model, the biochemical results with purified enzymes 341 indicate that the observed regulation is intrinsic to MraY and does not require additional 342 proteins. The MraY(T23P) variant, which is apparently less sensitive to regulatory control, 343 showed much greater activity in vitro than the wild-type enzyme. At first glance, this result may 344 seem incompatible with the proposed feedback control given that the product of the reaction is 345 lipid I with its head-group in the cytoplasm, not externalized lipid II. However, because the 346 reactions are performed in detergent, the lipid I formed in the reaction is likely capable of 347 reorienting in the micelles to mimic a periplasmic orientation. Although externalized lipid I is not 348 observed in vivo, the MD simulations predict that both flipped lipid I and lipid II are capable of 349 binding at the MraY dimer interface. It is therefore reasonable to interpret the biochemical 350 results in the context of a feedback inhibition model with MraY(WT) activity leveling off early in 351 the time course due to feedback control. By contrast, we infer that MraY(T23P), with its 352 substitution in the proposed binding site for flipped lipid II, is insensitive to feedback control 353 and therefore displays robust activity in the assay. Another factor that is likely to contribute to 354 the biochemical results is the co-purification of lipid II with the purified enzymes, which 355 according to the model would be expected to further reduce the activity of MraY(WT) relative to 356 MraY(T23P). Importantly, the activity for the wild-type enzyme was already so low that it was 357 (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted August 1, 2023. ; https://doi.org/10.1101/2023.08.01.551478 doi: bioRxiv preprint

15
Although additional experiments are required to further investigate the possible feedback 363 regulation of MraY, it is a compelling model because it suggests a mechanism by which cells 364 can balance the supply of flipped lipid II precursor with the activity of the PG synthases that 365 use it to build the cell wall (Fig. 5). We propose that when PG synthases are highly active, the 366 steady state level of lipid II remains low such that MraY is functioning near its maximum activity 367 to continue supplying lipid-linked PG precursors (Fig. 5, left panel). However, when the supply 368 of lipid II exceeds the capacity of the PG synthases to use it, either transiently or due to a 369 change in growth conditions, the steady state level of lipid II will rise such that it begins binding 370 MraY dimers to inhibit their activity and reduce flux through the lipid stages of PG precursor 371 production until supply more closely matches demand (Fig. 5, right panel). Such feedback 372 control would prevent excess C55P from being sequestered in PG precursors when they are 373 not needed, making more of the lipid carrier available to other glycan synthesis pathways for 374 their efficient operation. Accordingly, P. aeruginosa cells with an activated MraY variant, which 375 is presumably less sensitive to feedback control, display reduced ability to make O-Ag, 376 rendering them susceptible to self-intoxication by their encoded pyocins (Fig. S4B, C). MraY within the YES complex 31 , we observe an MraY dimer with electron density at this 385 interface as observed in prior X-ray crystal structures 26,29 . However, in our structure, this 386 density not only fills the pocket but also extends into the extracytoplasmic opening. This 387 16 density in the extracytoplasmic space is large enough to correspond to a head-group of flipped 388 lipid II. Accordingly, MD simulations indicate the capacity of MraY dimers to bind two 389 molecules of flipped lipid II with contacts between the protein and the MurNAc sugar that likely 390 provide specificity for externalized lipid II binding over C55PP or C55P. Notably, the head-391 groups of the lipid II binding substrates remain relatively flexible in the simulations (Movie S1, 392 The MD simulations predict conformational changes in the MraY dimers associated with lipid II 396 binding that increase the distance between the 6 th transmembrane helix (TM6) of each 397 monomer in the dimeric structure and alter the position of the 9 th transmembrane helix (TM9) 398 ( Fig. S8A-D). Similarly, the distance between a periplasmic helix (residues 221-228) from each 399 monomer is also increased (Fig. S8C-F). These changes are reminiscent of the conformational 400 difference between MraY in the YES complex relative to the free MraY structure from A. 401 aeolicus 26 . When the structures are aligned on one monomer, the second monomer in the YES 402 complex 31 is tilted relative to its partner in the A. aeolicus dimer 26 resulting in the opening the 403 periplasmic cavity and tightening the interface at the cytoplasmic side of the enzyme where the 404 active site is located (Fig. S9). Because MraY in the YES complex is inhibited by the phage 405 lysis protein, this opened conformation likely represents the inhibited state. The similarities 406 between the conformational changes in MraY observed in the YES complex and upon lipid II 407 binding in the MD analysis indicate that it is feasible for lipid II binding on the periplasmic side 408 of the enzyme to be communicated to the active site via an alteration of the dimer interface. 409 Accordingly, an increased mobility of TM9 on the cytoplasmic-face is also observed in the MD 410 analysis when lipid II is bound (Fig. S8B). How the T23P and other changes that presumably 411 activate MraY by reducing the sensitivity of the enzyme to inhibition by lipid II are not yet clear. 412 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted August 1, 2023. ; https://doi.org/10.1101/2023.08.01.551478 doi: bioRxiv preprint However, electron density corresponding to the lipid is still observed at the dimer interface 413 between MraY(T23P) protomers in the variant YES complex. Although this result may be 414 affected by the enzyme being stuck in an inhibited state by the phage inhibitor, it suggests that 415 T23P and other changes in MraY may affect the conformational response of the enzyme to 416 lipid II binding rather than the binding event itself. Consistent with this possibility, tyrosine at 417 position 21 has an altered conformation in the MraY(T23P) structure in which its hydroxyl 418 group forms a hydrogen bond network with Y227 and K358 on the opposing monomer (Fig.  419   S10). Substitutions within these residues were also identified in the screen for hyperactive 420 MraY enzymes, and Y227 is in the periplasmic helix that was found to be altered in the MD 421 analysis upon lipid II binding. Thus, alterations affecting interactions in this region may be 422 responsible for the regulation of MraY activity and its potential modulation by lipid II binding. (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted August 1, 2023. ; https://doi.org/10.1101/2023.08.01.551478 doi: bioRxiv preprint 18 438 In summary, we provide evidence that the essential and broadly conserved MraY step in PG 439 synthesis is subject to a previously unknown regulatory mechanism. Mutational and structural 440 evidence identified the likely regulatory site on the enzyme. Importantly, this site is accessible 441 by small molecules from the extracytoplasmic side of the membrane unlike the active site, 442 which is in the cytoplasm. This regulatory site therefore represents an attractive new target for 443 the development of small molecule inhibitors of MraY for potential use as antibiotics. 444 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted August 1, 2023. ; https://doi.org/10.1101/2023.08.01.551478 doi: bioRxiv preprint

446
We would like to thank all the members of the Bernhardt, Clemons, Stansfeld, and Rudner 447 Labs for their thoughtful discussions and advice throughout this project. We are also grateful to  molecules (highlighted as green, gold and pink spheres) freely enter the MraY cavity during 514 unbiased MD simulations. In 8/9 repeats, 2 or 3 lipid I or II molecules bind the cavity. In the last 515 repeat, one lipid II and one C55P molecule bind. (D) Lipid II contacts with MraY residues that 516 interact with lipid II for over 60% of atomistic MD simulations. Error bars represent standard 517 error from 5 repeats. Darker green bars represent residues altered in hyperactive 518 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made lipid II is consumed at a rate proportional to its production such that steady-state levels of the 528 precursor remains low and MraY activity is unimpeded. Right: When PG polymerase activity is 529 reduced due to changes in growth conditions or other perturbations, lipid II will be produced 530 faster than it is consumed, resulting in the accumulation of elevated levels of flipped lipid II. 531 Higher levels of the precursor promote its binding to MraY dimers, reducing their activity in (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted August 1, 2023. ; https://doi.org/10.1101/2023.08.01.551478 doi: bioRxiv preprint 22 left) with corresponding 2D classes observed in the dataset. Arrows denote the methodology 568 order, following several rounds of heterogeneous refinement. The number of particles sorted is 569 shown below the densities. The masked volume of MraY (green, top right) used for particle 570 subtraction is shown overlayed with the density (purple) of the entire YES complex. The final 571 model is colored by resolution using the viridis color scheme. The unmodeled density at the 572 dimer-interface is isolated for clarity and shown in a dotted box. 573 574 Supplementary Figure 6. MraY residues contacting lipid II in the MD simulations. Lipid II 575 contacts with MraY residues from atomistic MD simulations. Error bars represent standard 576 error from 5 repeats. Darker green bars represent residues altered in hyperactive variants. 577 Dashed line at x=0.6 represents cutoff for interactions shown in Figure 4C. The structure is shown from the top, lipid II is hidden, and helices with notable differences are 588 indicated. (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made      . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted August 1, 2023. ; https://doi.org/10.1101/2023.08.01.551478 doi: bioRxiv preprint  . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted August 1, 2023. ; https://doi.org/10.1101/2023.08.01.551478 doi: bioRxiv preprint Left: When PG polymerase activity is high, flipped lipid II is consumed at a rate proportional to its production such that steady-state levels of the precursor remains low and MraY activity is unimpeded. Right: When PG polymerase activity is reduced due to changes in growth conditions or other perturbations, lipid II will be produced faster than it is consumed, resulting in the accumulation of elevated levels of flipped lipid II. Higher levels of the precursor promote its binding to MraY dimers, reducing their activity in order to bring lipid II supply back in balance with demand by the polymerases. See text for details. Abbreviations: C55P, undecaprenylphosphate; UM5, UDP-MurNAc-pentapeptide; UG, UDP-GlcNAc; PG, peptidoglycan.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  P. aeruginosa cells expressing the indicated plasmid were grown to mid-log, normalized for optical density, and extracts were prepared for immunoblotting. Protein was detected using α-VSVG antibody.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made LI hydrolysis product LII hydrolysis product     . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted August 1, 2023. ; https://doi.org/10.1101/2023.08.01.551478 doi: bioRxiv preprint 3 SI Table 3

. Oligonucleotide primers used in this study
Primer (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made GAGGAGGATACAT -3'). ). After digestion with EcoRI/XmaI, the PCR product was ligated into pPSV38 to generate pNG93. The final construct was sequence verified using primers 556 and 557.

pLSM116 [PT7::H-SUMO-FLAG-Pa-mraY] is a pCOLADuet derivative.
The gene encoding full-length P. aeruginosa mraY was amplified from pNG93 using the primers oLSM302 and oLSM303. Using pCOLADuet as a template, the backbone was amplified using oLSM301 and oLSM304. The fragments were joined using Gibson assembly and sequence verified using primers 34 and 2325.

pLSM117 [PT7::H-SUMO-FLAG-Pa-mraY T23P
] is a pCOLADuet derivative. The gene encoding full-length P. aeruginosa mraY T23P was amplified from pNG102 using the primers oLSM302 and oLSM303. Using pCOLADuet as a template, the backbone was amplified using oLSM301 and oLSM304. The fragments were joined using Gibson assembly and sequence verified using primers 34 and 2325.
pLSM124 [PlacUV5::Ec-mraY] is a pPSV38 derivative. The gene encoding full-length E. coli mraY was amplified from MG1655 genomic DNA using primers oLSM312 and oLSM313. Using pNG93 as a template, the backbone was amplified using oLSM311 and oLSM314. The fragments were joined using Gibson assembly. The final construct was sequence verified using primers 556 and 557.
pLSM125 [PlacUV5::Ec-mraY(T23P)] is a pPSV38 derivative. Using pLSM124 as a template, T23 was mutated to P using site directed mutagenesis (QuikChange Lightning, Agilent) using the primers oLSM315 and oLSM316. The final construct was sequence verified using primers 556 and 557. pLSM141 [Plac::Pa-mraY] is a pRY47 derivative. The gene encoding full-length P. aeruginosa mraY was amplified from pNG93 using primers oLSM372 and oLSM373. Using pRY47 as a template, the backbone was amplified using oLSM374 and oLSM368. The fragments were joined using Gibson assembly. The final construct was sequence verified using primers 556 and 48.
pLSM142 [Plac::Pa-mraY(T23P)] is a pRY47 derivative. The gene encoding full-length P. aeruginosa mraY (T23P) was amplified from pNG102 using primers oLSM372 and oLSM373. Using pRY47 as a template, the backbone was amplified using oLSM374 and oLSM368. The fragments were ligated using Gibson assembly. The final construct was sequence verified using primers 556 and 48.
pLSM143 [Plac::Ec-mraY] is a pRY47 derivative. The gene encoding full-length E. coli mraY was amplified from pLSM124 using primers oLSM375 and oLSM376. Using pRY47 as a template, the backbone was amplified using oLSM377 and oLSM368. The fragments were ligated using Gibson assembly. The final construct was sequence verified using primers 556 and 48.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted August 1, 2023. ; https://doi.org/10.1101/2023.08.01.551478 doi: bioRxiv preprint pLSM144 [Plac::Ec-mraY(T23P)] is a pRY47 derivative. The gene encoding full-length E. coli mraY (T23P) was amplified from pLSM125 using primers oLSM375 and oLSM376. Using pRY47 as a template, the backbone was amplified using oLSM377 and oLSM368. The fragments were ligated using Gibson assembly. The final construct was sequence verified using primers 556 and 48.

Materials
Unless otherwise indicated, all chemicals and reagents were purchased from Sigma-Aldrich. Restriction enzymes were purchased from New England Biolabs. Oligonucleotide primers were purchased from Integrated DNA Technologies.

Electroporation of P. aeruginosa
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted August 1, 2023. ; https://doi.org/10.1101/2023.08.01.551478 doi: bioRxiv preprint 8 P. aeruginosa strains were made competent using previously described methods 11 . For electroporation, 100 ng of plasmid DNA was added to 40 μL of competent P. aeruginosa cells. Transformation was achieved using standard protocols and transformants were selected for using 30 μg/mL Gent.

Viability assays
Overnight cultures of PAO1, PA686, or PA760 derivatives, containing vectors producing the indicated alleles of mraY expressed from an IPTG-inducible (PlacUV5) plasmid were normalized to an OD600 of 2.4 before being serially diluted. Aliquots (5 µL) of the dilutions were spotted onto LB Gent agar, VBMM Gent agar, with or without IPTG. Plates were incubated at 30°C for 24 h at which point the plates were imaged. A similar protocol was adapted for MG1655 and MM119 derivatives containing vectors producing the indicated alleles of mraY from an IPTG inducible (Plac) plasmid.

Immunoblotting
For analysis of protein levels from strains producing MraY-VSVG variants, an overnight culture of each of the strains was allowed to grow in LB containing 30 μg/mL Gent at 37°C. The following day, the cultures were diluted to an OD600 of 0.01 and allowed to grow at 37°C in LB containing 30 μg/ml Gent. After 2 h, 1 mM IPTG was added and the cultures were allowed to grow for another 2.5 h. Cultures were normalized to an OD600 = 1.0 and cells were collected by centrifugation at 5,000 × g for 2 min. The cell pellet was resuspended in 200 μL of 2× Laemmli buffer, then centrifuged for 10 min at 21,000 x g. Samples were analyzed by SDS-PAGE followed by imunoblotting. Protein was transferred from the SDS-PAGE gel to a nitrocellulose membrane using wet transfer (30 min at 100V) in cold transfer buffer (192 mM glycine, 20% methanol, 25 mM Tris base). The membrane was blocked in 5% (w/v) skim milk powder in Tris-Buffered saline (10 mM Tris-HCl pH 7.5, 150 mM NaCl) containing 0.5% (v/v) Tween-20 (TBS-T) for 45 min at room temperature with gentle agitation. The α-VSVG antibody (V4888, Sigma) was added to the blocking buffer at a 1:5000 dilution for 1 h. The membrane was washed three times in TBS-T for 5 min each before incubation for 1 h with secondary antibody (anti-rabbit IgG HRP, 1:5000 dilution, 7074S, NEB) in TBS-T with 1% (w/v) skim milk powder. The membrane was then washed three times with TBS-T for 5 min each before developing using Clarity Max TM Western ECL Substrate (1705062; BioRad) and imaged using a BioRad ChemiDoc XRS+.

Error Prone PCR
Mutagenesis was adapted from Yang et al 12 . Four independent mutant plasmid libraries were constructed by mutagenizing mraY in plasmid pNG93 (PlacUV5::mraY) using Taq polymerase with Thermopol buffer (New England Biolabs, M0267L). The forward 5'-ACACTTTATGCTTCCGGCTC-3' and reverse 5'-ACTGTTGGGAAGGGCGATCAAA-3' primers were used to amplify mraY from pNG93. The resulting PCR products were purified using the Monarch® PCR & DNA Cleanup Kit (NEB, T1030) and used as "megaprimers" that are denatured and annealed to the original plasmid (pNG93) to amplify the vector backbone using Q5® High-Fidelity 2X Master Mix (NEB, M0492S). The reactions were then digested with DpnI to eliminate any remaining parental plasmid DNA. All four libraries were independently electroporated into NEB 10-beta electrocompetent cells (NEB, C3020K) and plated on LB agar supplemented with 15 μg/ml gentamicin at 37ºC overnight.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted August 1, 2023. ; https://doi.org/10.1101/2023.08.01.551478 doi: bioRxiv preprint Transformants were slurried in LB, and the resuspended cells were normalized to an OD600 = 10. Cells from 1 mL of resuspension were centrifuged and plasmid DNA was isolated from the cell suspension using the Monarch® Plasmid DNA Miniprep Kit (T1010). All four libraries were independently transformed into electrocompetent PA686 cells and plated on LBNS agar supplemented with 30 μg/ml Gent and grown overnight at 37ºC. The resulting transformant colonies from each of the libraries were slurried in LBNS supplemented with 30 μg/ml Gent. Samples of each were normalized to OD600 = 10 in LBNS + 10% (v/v) DMSO, and stored at -80ºC. A sample from each library was then thawed and serial dilutions were plated on VBMM 30 μg/ml Gent with or without IPTG [50 μM], and grown at 30ºC overnight. Individual colonies arising on the IPTG supplemented plates from each library were selected and re-streaked on VBMM with or without IPTG. Those that displayed IPTG dependence were further isolated, and the plasmids sent for sequencing. Clones identified to contain a single point mutation were further characterized. The mutated mraY genes were each amplified using Q5 High-fidelity polymerase (NEB) via colony PCR. The purified PCR product was digested with EcoRI and XmaI, and subsequently ligated into pPSV38 for validation of the suppression phenotype. All clones were sequence verified. MraY variants are listed in Table S1.

Lipid II extraction
Cultures of PAO1, PA686 and MG1655 were grown at 37°C overnight, and MM119 at 30°C overnight. The next day, cultures were diluted to an OD600 of 0.01 and allowed to grow for 2 h at the above specified temperatures whereupon 1mM IPTG was added to induce expression of MraY. Cells were collected when the OD600 reached ~0.5, and normalized to OD = 1 in a 1 mL volume. Pellets were collected by centrifugation at 21,000 x g and stored at -20°C until needed. Cells were resuspended in 1 mL LB and added to a mixture of 2:1 methanol : chloroform (3.5 mL total) in borosilicate glass tubes (16x100 mm, Fisher Scientific 1495935AA). Samples were vortexed for 1 min to form a single phase. Cell debris was collected by centrifugation for 10 min at 2000 × g, 21°C. The supernatant was transferred to a fresh borosilicate glass tube, and 2 mL of chloroform was added. The supernatant was acidified using 0.1N HCl to pH 1 as determined by pH indicator strips. The samples were vortexed for 1 min and centrifuged for 20 min at 2000 × g at 21°C to form a two-phase system. Using a glass pipette, as much of the aqueous upper layer was removed without disturbing the interface between the aqueous and organic phases and 1 mL methanol was subsequently added to form a single liquid phase upon vortexing. Samples were transferred to 1.5 mL Eppendorf tubes by glass pipette then dried by nitrogen stream at 40°C. Dried samples were dissolved in 150 μL of a mixture of methanol and chloroform (2:1) by vortexing then centrifuged at 21,000×g for 1 minute and dried by nitrogen stream at 40°C. This was repeated with 40 μL organic mixture, and finally crude lipid extracts were dissolved in 10 μL DMSO by vortexing. Extracts were stored at -20°C.

Lipid II hydrolysis
Crude lipid II (LII) extracts were added (5 μL) to 5 μL of 0.2 M HCl, for a final concentration of 0.1 M HCl. Samples were boiled at 100°C for 15 min and then cooled to 4°C in a thermocycler. 10 μL of sodium borate pH 9 was added followed by 1 μL 0.5M NaOH to neutralize the solution. 2 μl of 100 mg/ml sodium borohydride was added and the samples were allowed to incubate for 30 min at room temperature. Following the incubation, 2 μl of 20% phosphoric acid . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted August 1, 2023. ; https://doi.org/10.1101/2023.08.01.551478 doi: bioRxiv preprint was added to quench the reaction and the samples were mixed and immediately subjected to LC/MS analysis.

LC/MS
High-resolution LC/MS traces of soluble LII hydrolysis products were obtained using the following protocol. Briefly, the hydrolyzed samples were subjected to LC/MS analysis (ESI, positive mode). A Waters Symmetry Shield RP8 column (3.5 µm, 4.6 mm X 150 mm) was used to separate hydrolysis products using the following gradient (A, H2O + 0.1% formic acid; B, acetonitrile + 0.1% formic acid; 0.5 ml/min): 0% B for 5 min, followed by a linear gradient of 0%-20% B over 40 min. Data was obtained on an Agilent 6546 LC-q-TOF Mass Spectrometer. Expected ion masses were extracted with a tolerance of 0.01 mass units.

Purification of UDP-MurNAc pentapeptide
Accumulation of the precursor was performed as previously described 13 with the following modifications. Bacillus cereus ATCC 14579 was grown in LB-lennox medium at 37°C until the OD600 reached between 0.7-0.8, at which point 130 μg/mL of chloramphenicol was added. After 15 minutes of incubation, 5 μg/ml of vancomycin was added and the cells allowed to incubate for another 60 min at 37°C with shaking. The culture was then cooled on ice and harvested by centrifugation (4000 x g, 20 min, SLC-6000 rotor, 4°C). Cells were collected and stored at -20°C until required.
Cells were resuspended in water (0.1 g wet weight/mL) and stirred into boiling water in a flask with stirring. Boiling was allowed to continue for another 15 minutes at which point the flask was removed from heat and allowed to cool to room temperature with stirring. After approximately 20 minutes the resuspension was cooled on ice and the debris was pelleted at 200,000 x g for 60 min at 4°C. The supernatant was removed and lyophilized. The lyophilized material was resuspended in water and acidified to pH 3 using formic acid (1mL/L culture extracted), centrifuged to remove precipitate, and immediately subjected to reversed phase high pressure liquid chromatography (RP-HPLC).
UDP-MurNAc pentapeptide was isolated by RP-HPLC on a Synergi 4u Hydro-RP 80A (250x 10.0 mM). The column was eluted over a 30-min isocratic program (A, H2O + 0.1 % formic acid; B, acetonitrile + 0.1% formic acid; 4 ml/min), 4% B for 30 min at room temperature. The elution was monitored by UV at 254 nm. UDP-MurNAc-pentapeptide eluted approximately at 20 min in a single peak, which was verified by mass-spectrometry (1194.35 Da). Peak fractions were collected and lyophilized. The final product was resuspended in water for downstream use.

Expression and Purification of PaMraY
For expression of P. aeruginosa MraY or MraY T23P , E. coli expression strain LSM9 containing pAM174 and the expression plasmid (pLSM116 or pLSM117) was grown in 1L TB supplemented with 2 mM MgCl2, kanamycin, and chloramphenicol at 37°C with shaking until the OD600 was 0.7. The cultures were cooled to 20 °C before inducing protein expression with 1mM IPTG and 0.1% (w/v) arabinose. Cells were harvested 19h post induction by centrifugation (6,000 x g, 15 min, 4°C). To purify FLAG-MraY or FLAG-MraY T23P , the cells were resuspended in lysis buffer B (50 mM HEPES pH 7.5, 150 mM NaCl, 20 mM MgCl2, 0.5 mM DTT) and lysed by passage through a cell disruptor (Constant systems) at 25 kpsi twice. Membranes were collected by ultracentrifugation (100,000 x g, 1h, 4°C). The membrane . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted August 1, 2023. pellets were resuspended in solubilization buffer B (20 mM HEPES pH 7.0, 0.5M NaCl, 20% (v/v) glycerol, and 1% (w/v) DDM (Thermo Fisher)) and rotated end over end for 1h at 4°C before ultracentrifugation (100,000 x g, 1h, 4°C). The supernatant was supplemented with 2 mM CaCl2 and loaded onto a pre-equilibrated homemade M1 anti-FLAG antibody resin. The resin was washed with 25 column volumes (CVs) of wash buffer C (20 mM HEPES pH 7.0, 0.5M NaCl, 20% (v/v) glycerol, 2 mM CaCl2, 0.1% (w/v) DDM) and the bound protein was eluted from the column with five CVs of elution buffer (20 mM HEPES pH 7.0, 0.5M NaCl, 20% (v/v) glycerol, 0.1% (w/v) DDM, 5 mM EDTA pH 8.0, and 0.2 mg/mL FLAG peptide). Fractions containing the target protein were concentrated and the protein concentration was measured via the Bradford method. Proteins were aliquoted and stored at -80°C until required.

MraY translocase in vitro assay
The assay was performed at 37°C in an assay buffer containing 20 mM HEPES pH 7.5, 500 mM NaCl, 20% (v/v) glycerol, and 0.1% (w/v) DDM, 10 mM MgCl2, 250 µM UDP-MurNAc pentapeptide, and 1.1 mM C55P (Larodan). Protein was added to initiate the reaction at a final concentration of 1.7 µM. At the appropriate time point the reaction was quenched by boiling for 3 min at 95°C. 1.5 units of alkaline phosphatase was added to the sample (NEB M0371L) and incubated at 25°C for 1h. The samples were heat quenched at 65°C to stop the reaction and were immediately loaded for analysis by LCMS. The samples were monitored by UV 254 and by MS (ESI, positive mode). A Thermo Fisher Hypersil Gold aQ C18 (150x4.6 mm 3 µm) HPLC column was used to separate the substrates and products using the following gradient program (A, H2O + 0.1% formic acid; B, acetonitrile + 0.1% formic acid; 0.4 ml/min): 4% B for 20 min. Data was obtained on an Agilent 6546 LC-q-TOF Mass Spectrometer.

Preparation of lipopolysaccharide and immunoblotting
To isolate LPS from the P. aeruginosa strains containing the indicated plasmids, overnight cultures of each of the strains were allowed to grow in LB at 37°C containing 30 μg/mL Gent. The next day, cultures were diluted to an OD600 of 0.01 and allowed to grow at 37°C in 25 mL LB containing 30 μg/ml Gent. After 2 h, 1 mM IPTG was added and the cultures were allowed to grow for another 2h until cultures reached mid-log. 20 mL of culture was pelleted at 4000 x g for 12 min, and cells were resuspended in 1 mL LB and the OD600 was measured. The cells were pelleted again at 12,000 x g for 2 min, and resuspended in 1X LDS buffer (Invitrogen, NP00008) + 4% BME to an OD600 = 20. Samples were boiled at 95 °C for 10 min. Each sample was subjected to the NI Protein Assay (G Biosciences, 786-005) to determine the protein content in each sample. The lysates (50 μl) were then incubated at 55°C with 1.25 μl proteinase K (NEB, P8107S). After 1h of incubation, samples were boiled at 95°C for 10 min, and then frozen at -20°C until required.
Volumes of lysates corresponding to 20 μg of protein were then run on a Criterion XT 4-12% Bis-Tris Precast Gel (Bio-Rad, 3450124) in MES running buffer (50 mM MES, 50 mM Tris base, 1 mM EDTA, 0.1% (w/v) SDS) for 1h 45 min at 100V constant. Glycan was transferred to nitrocellulose membranes as described above with the following differences: membranes were blocked for 1h at room temperature in 1% (w/v) skim milk, and were then incubated with antiserotype O5 B-band at a 1:1000 dilution overnight at 4 °C (gift from L. Burrows). After three 15-mL TBST washes, membranes were incubated with anti-mouse HRP antibody (1:5000, NEB 7076S) for 1h at room temperature. Blots were developed as described above.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted August 1, 2023. ; https://doi.org/10.1101/2023.08.01.551478 doi: bioRxiv preprint

Molecular dynamics simulations
For the coarse-grained MD, the structural model of the E. coli MraY dimer was aligned according to the plane of the membrane with memembed 14 , and then converted to the Martini 3 force field using the martinize protocol 15 . Bonds of 500 kJ mol -1 nm -2 were applied between all protein backbone beads within 1 nm. Proteins were built into 13 x 13 nm membranes composed of 40% POPE, and 10% each of POPG, CDL, lipid I, lipid II, C55-P, and C55-PP using the insane protocol 16 . Alternatively, membranes were built with 60% POPE, and 10% each of POPG, CDL, C55-P, and C55-PP. Lipid I, lipid II, C55-P and C55-PP parameters were from Orta et al. 3 .Systems were solvated with Martini waters and Na + and Clions to a neutral charge and 0.0375 M. Systems were minimized using the steepest descents method, followed by 1 ns equilibration using 5 fs time steps, then by 100 ns equilibration with 20 fs time steps, before 9 x 10 µs (complex membrane) or 5 x 10 µs (membrane without lipid I or lipid II) production simulations were run using 20 fs time steps, all in the NPT ensemble with the velocity-rescaling thermostat and semiisotropic Parrinello-Rahman pressure coupling 17,18 .
A pose of the E. coli MraY dimer with two lipid II molecules bound to the central cavity was selected for further analysis. All non-POPE lipids (except the two bound lipid II molecules) were deleted and the membrane allowed to shrink to 10 x 10 x 10.5 nm over 100 ns with positional restraints applied to the protein backbone. The resulting molecule was then converted to the atomistic CHARMM36m force field 19,20 using the CG2AT2 protocol 21 . Side chain pKas were assessed using propKa3.1 22 , and side chain side charge states were set to their default. Production simulations were run for 5 repeats of ca. 510 ns, using a 2 fs time step in the NPT ensemble with the velocity-rescale thermostat and semi-isotropic Parrinello-Rahman pressure coupling 17,18 .
All simulations were run in Gromacs 2021.3 23 . Images were made in VMD 24 . Kinetic analysis of protein-lipid interactions and binding site identification were performed using PyLipID 25 . Density and contact analyses of atomistic MD simulations were performed using MDAnalysis 26,27 . Contacts are defined as a distance of less than 4 Angstroms between Lipid II and MraY.

Expression and purification of the YES complex
The YES complex was expressed as described previously 3 . Briefly, ∆slyD BL21(DE3) competent cells were transformed with pET22b-SlyD1-154 and pRSFDuetEcMraY-EID21 and plated in LB-agar containing 35 µg/ml Kanamycin and 100 µg/mL Ampicillin. The culture was grown in 2xYT media at 37 • C, 225 r.p.m., and induced at an OD600 of 0.9 with 0.4mM IPTG at 18 • C overnight. The culture was harvested by centrifugation for 10 minutes at 9,000xg, 4 • C followed by flash freezing.