Mycobacteria recycle their peptidoglycan via a novel pathway which influences antimicrobial resistance and limits proliferation in macrophages

Affiliations: 7 1Institute of Microbiology and Infection 8 School of Biological Sciences 9 University of Birmingham 10 Birmingham, UK, B15 2TT 11 12 2i3SInstituto de Investigação e Inovação em Saúde 13 Universidade do Porto, Porto, Portugal 14 15 3IBMC Instituto de Biologia Molecular e Celular 16 Universidade do Porto, Porto, Portugal 17 18 4University of Leicester, 19 Leicester, UK, LE1 7RH 20 21 *To whom correspondence should be addressed: 22 Patrick J. Moynihan, Ph.D., 23 Institute of Microbiology and Infection 24 School of Biological Sciences 25 University of Birmingham 26 Birmingham, UK, B15 2TT 27 email: p.j.moynihan@bham.ac.uk 28 29

integrity of this macromolecule must be maintained under most growth conditions and its 51 rupture leads to lysis and cell death 1 . As a result of this essentiality, it is vital that cells are able 52 to withstand their own internal turgor pressure and still be able to cleave the cell wall to allow 53 for division, growth and the insertion of macromolecular structures such as secretion systems 1 . 54 Throughout this process, the activity of lytic enzymes or through the attack of host agents like 55 lysozyme, the sacculus is cleaved with the resulting generation of small PG metabolites known 56 as muropeptides 2 . 57 In Gram-positive bacteria muropeptides are typically released from the cell wall through 58 the action of lysozyme-like hydrolytic enzymes, whereas in Gram-negative bacteria, lytic 59 transglycosylases generate 1,6-anhydroMurNAc products 3,4 . These metabolites have been 60 shown to be important in many aspects of host-pathogen interactions. For example, tracheal 61 cytotoxin produced by Bordetella pertussis is the product of lytic transglycosylases 5 . Release 62 of a similar molecule has also been shown to be involved in tissue damage during Neisseria 63 gonorrhoeae infection and in the closure of the light-organ of the bobtail squid 6,7 . In many 64 to recycle their PG at all 10 . 79 The cell wall of M. tuberculosis is built upon a foundation of PG. The remainder of this 80 structure is formed by the modification of muramic acid residues with an arabinogalactan 81 polymer that is in turn esterified by mycolic acids 11 . This waxy coating contributes to drug 82 resistance in M. tuberculosis, but is also the target of several mycobacteria-specific 83 antibiotics 11 . The challenge of multi-and extensively-drug resistant M. tuberculosis has not 84 adequately been met by drug discovery efforts, however recent reports suggest that β-lactams 85 are effective at treating these drug-resistant infections [12][13][14] . Despite their therapeutic promise, 86 we know relatively little about the turn-over of PG in mycobacteria, which is the eventual target 87 of β-lactam antibiotics. 88 cartridges as previously described 20 . Purified fractions were evaporated to dryness and the 163 concentration of reducing sugars in the pool of soluble muropeptides was assessed using the 3-164 methyl-2-benzothiazolinone hydrazone (MBTH) assay 21 . 165

Synthesis of 4MU-D-lactate 166
Instead of the 2-or 3-step protocols published for the synthesis of 4MU-D-lactate previously, 167 we used a simplified one step method 22,23 . 1.5 g of (s)-(-)-bromopropionic acid was added to 1 168 g of 4-methylumbelliferone stirring in 40 mL anhydrous dimethylformamide and 0.75 g 169 Cs2CO3. This was stirred at room temperature over-night and the product was extracted three 170 times with water/ethyl-acetate and the organic phase was dried over sodium sufate. The organic 171 phase was then filtered and evaporated to dryness. The product was subsequently purified using 172 silica chromatography and was dried as a crystalline white solid. 173

Turn-over of 4MU reporter compounds by M. bovis BCG 174
To test turn-over of 4MU-GlcNAc or 4MU-D-lactate by whole cells, 100 µL of a mid-175 exponential culture (OD600 = 0.6) was added to a sterile 96 well plate in Sauton's minimal media 176 supplemented with 0.05% Tween and 1% glycerol in addition to 1 mM 4MU-D-lactate or 4MU-177 GlcNAc. Similar controls lacking cells or the reporter compound were included as well. This 178 was incubated at 37 °C and mixed at 300 r.p.m. Each day the fluorescence of the sample was 179 read on a BMC PolarStar microplate reader (Ex. 355 nm; Em 460 nm) with a constant gain 180 setting. 181

Turn-over of M. bovis BCG PG in vitro 182
Cultures of M. bovis BCG wild-type, ∆lpqI, and ∆lpqI::lpqI were grown to an OD600 of 183 0.6 in the presence of 10 µCi 3 H meso-diaminopimelic acid (DAP), at which point they were 184 collected by centrifugation, washed 3 times with sterile media and diluted to 0.01 in fresh 185 culture flasks. Periodically a sample of 0.5 mL was taken, and the cells were collected by 186 centrifugation. The spent medium was mixed with 10 mL scinitilation fluid and counted using 187 a liquid scintillation counter. The cell pellet was re-suspended in 10% SDS, boiled for 20 min, 188 and centrifuged again. The cell-wall material was then resuspended in 1 mL scintillation fluid 189 and the material was counted in a liquid scintillation counter. The counts of the cell wall and 190 the media were added together to give total 3 H DAP in each culture and the data is presented as 191 a percentage of that total. During the course of the experiment the OD600 of the culture was 192 monitored daily. All measurements are from three biological replicates. [pRv0237] grown in Terrific Broth to an OD600 of 0.6, chilled to 20 °C and induced with 1 mM 198 IPTG and grown for a further 18 h before being collected by centrifugation. Cells were 199 resuspended in 25 mM Tris-HCl, 300 mM NaCl, 10 mM imidazole pH 7.8 and lysed via three 200 passages through a French pressure cell. The protein was purified using standard IMAC 201 procedures with washes of lysis buffer, lysis buffer including 50 mM imidazole and finally 202 eluted with 500 mM imidazole in lysis buffer. Eluted protein was dialysed exhaustively against 203 25 mM Bis-Tris, 100 mM NaCl pH 7.8 in the presence of recombinant Ulp1 protease which 204 specifically cleaves the His6-SUMO tag. Digested protein was subjected to a second IMAC 205 column (1 mL HisTrap FF, GE Healthcare) and the flow-through fraction was found to contain 206 pure, un-tagged Rv0237. Purified protein was dialysed into 25 mM Bis-Tris pH 6.5, 100 mM 207 NaCl. 208

Crystallography 209
Prior to crystallization, LpqI was concentrated to 20 mg•mL -1 in 25 mM Bis-Tris pH 210 7.5. LpqI crystals were grown by the sitting-drop vapour diffusion method by mixing an equal 211 volume of protein solution with 1.1 M sodium malonate, 0.1M HEPES, 0.5% v/v Jeffamine 212 ED-2001 (pH 7.0). Crystals were cryo-protected with a saturated solution of sodium malonate 213 and plunge frozen in liquid nitrogen. X-ray data was collected at the Diamond Light Source, 214 Oxford. Data were processed using XiaII and file manipulations were performed using the 215 CCP4 suite of programs. The structure was phased by molecular replacement using the 216 previously released, but unpublished M. smegmatis LpqI structure (PDB: 4YYF) using the 217 program PHASER. The structure was subsequently auto-built in PHENIX and the remaining 218 parts were built in COOT with further refinement using PHENIX and PDB-REDO. 219

Kinetic characterisation of Rv0237 220
Purified Rv0237 was evaluated for glycoside hydrolase activity using a variety of 221 substrates. As an initial screening assay, Rv0237 was incubated at 1 µM with either 4-222 methylumbeliferyl or p-nitrophenyl derivatives of a variety of sugars as listed in Figure 4a in 223 Bis-Tris pH 7.5, 100 mM NaCl at 37 °C. The release of p-nitrophenol was followed by change 224 in absorbance at 420 nm while production of 4-menthylumbelliferone was monitored by 225 fluorescence as above in a BMG Polarstar spectrophotometer. Kinetic characterisation of 226 Rv0237 was conducted using varying concentrations of 4MU-GlcNAc. The raw data were 227 compared to standards of 4-methylumbelliferone. All data were analysed using GraphPad Prism 228

229
To evaluate the ability of the enzyme to degrade fragments derived from PG, M. 230 smegmatis PG was digested with mutanolysin and soluble fragments were prepared and 231 quantified as above. Reactions including 1 µM Rv0237, 0.5 mM PG fragments in 25 mM 232 ammonium acetate buffer pH 6.5 were incubated for 18h at 37 °C. In parallel reactions were 233 carried out using pNP-GlcNAc in order to monitor enzyme activity visually. The reactions were 234 then evaluated by TLC (Silica 60 F254, Merck, Germany) using a mobile phase consisting of 1-235 butanol, methanol, ammonium hydroxide and water at a ratio of 5:6:4:1. TLCs were stained 236 with α-naphthol and developed by charring. 237

Infection of bone marrow-derived macrophages 238
BMDM were differentiated from bone marrow cells obtained from femurs and tibiae of 239 C57BL/6 mice cultured in the presence of L-cell conditioned medium, as described before 24 .

Peptidoglycan Recycling Genes in Mycobacteria 251
The genome of M tuberculosis encodes many lytic enzymes, including at least five vitro 18,28 . We hypothesized that a recycling system for these muropeptides is likely to also exist 266 in mycobacteria and analyzed the genomes of several mycobacteria for known PG-recycling 267 systems including the recently discovered systems of Pseudomonas putida and Tannerella 268 forsythia (Table S1, Figure 1 (Table S1). The M. tuberculosis, M. leprae, C. 276 glutamicum and M. bovis BCG genomes do, however, appear to encode orthologs of NagA and 277 NagZ. These enzymes are predicted to be an N-acetylglucosamine-6-phosphate N-deacetylase 278 and a GH3-family β-N-acetylglucosaminidase respectively 31,32 . NagZ in particular is typically 279 associated with PG recycling, whilst NagA is typically associated with the assimilation of 280 GlcNAc regardless of the source. 281

Utilisation of peptidoglycan components by mycobacteria 282
Prior research has shown that most mycobacteria are unable to use GlcNAc as a sole 283 carbon source, with M. smegmatis being one of the notable exceptions 33 . Furthermore, amino 284 acids including L-Ala, L-Glu, and L-Asp have previously been shown to serve as nitrogen 285 sources for M. tuberculosis H37Rv 34 . To our knowledge, recycling of GlcNAc or MurNAc has 286 not been reported, nor has recycling been evaluated for soluble PG fragments. To evaluate this, 287 M. bovis BCG was cultured in minimal media supplemented with glycerol (1% v/v) or MurNAc 288

Mechanism of MurNAc metabolism 299
The ability of M. tuberculosis and M. bovis BCG to grow on MurNAc was surprising 300 and so we evaluated the biochemical processing steps associated with MurNAc utilization. 301 MurNAc is a combination of GlcNAc and D-lactate joined by an ether linkage. This suggests 302 that the bacterium is likely either using the GlcNAc moiety for glycolysis, or shunting the lactate 303 derived from MurNAc into the TCA cycle. We tested this inhibiting by glycolysis with 2-304 deoxyglucose (2DG) in cultures grown using MurNAc, glucose and glycerol as sole carbon 305 sources ( Figure S1). These data suggested that the pathway of MurNAc utilization did not 306 require glycolysis and indicated that lactate instead was likely serving as a carbon source. 307 Consistent with this, when used as a sole carbon source, growth on L-lactate and MurNAc was 308 O2 dependent while D-lactate was better utilized under static, 5% CO2 culture conditions, where 309 MurNAc could not be used as a carbon source (Figure 2c). 310 These data allow us to hypothesize a mechanism by which M. bovis BCG metabolises 311 MurNAc. Given that metabolism of L-lactate and MurNAc are O2-dependent, we anticipate that 312 use of MurNAc follows cleavage of the D-lactate from MurNAc via an inverting mechanism to 313 produce L-lactate and GlcNAc. In this case, the O2-dependency on MurNAc metabolism is 314 likely the result of an O2-depenedent lactate monooxygenase. Consistent with this, two O2-

Uptake of PG metabolites by mycobacteria 322
While our data strongly support metabolism of MurNAc by M. bovis BCG, confirmation 323 of PG-recycling requires demonstration of the uptake of muropeptides by the bacterium. To 324 investigate this, we generated radio-labelled muropeptides and tested them in whole-cell uptake 325 assays to determine if mycobacteria are competent for recycling this more complex substrate. 326 Muropeptides had to be generated in M. smegmatis due to the inability of M. bovis BCG to take-327 up 14 C GlcNAc under the conditions we tested. As shown in Figure 3a, M. bovis BCG was able 328 to incorporate approximately 4% of the muropeptide-associated 14 C radio-label added to the 329 culture. We next sought to determine if components of the stem-peptide were also recycled. 330 The above experiments were repeated using 3 H-DAP-labelled muropeptides. This material was 331 also incorporated into whole cells at a rate of approximately 7% of the added label (Figure 3a). 332 We next evaluated the turn-over of muropeptides in whole cells using 3 H-DAP due to the 333 inability of M. bovis BCG to incorporate 14 C GlcNAc into its cell wall. As shown in Figure 3c, 334 M. bovis BCG very slowly releases DAP to the culture media in vitro. Consistent with a PG-335 recycling system we also found that soluble PG could serve as a sole carbon source for M. bovis 336 BCG under aerated conditions (Figure 3d). Together, these results indicate that pathogenic 337 mycobacteria possess the biochemical capacity to recycle components of their cell wall. 338

Biochemical and structural characterisation of LpqI 339
In previously characterized PG-recycling systems free amino sugars are by a glycoside with GlcNAc-containing substrates. Critically, this sugar is only found in the backbone of PG 367 and a small amount in the linker unit (MurNAc-6-P-Rha-GlcNAc-galactan) between PG and 368 arabinogalactan. We then evaluated the Michaelis-Menten kinetics of LpqI using 4MU-GlcNAc 369 as a substrate with a similar kcat (2.8 x 10 -2 ± 0.04 x 10 -2 •s -1 ) and Km (106 ± 5 µM) as observed 370 for other NagZ enzymes using this substrate (Figure 4b) 39 . In a similar assay we were also able 371 to show that LpqI releases GlcNAc from soluble PG fragments (Figure 4c). While hydrolytic 372 activity has been reported for most NagZ-type enzymes, a recent report suggested that β-N-373 acetylglucosaminidases from the GH3 family are in fact phosphorylases 38 . Another GH3 β-N-374 acetylglucosaminidase was recently reported to lack this activity, suggesting that it may not be 375 a general property of the family 43 . We tested the activity of the enzyme under the same 376 conditions as reported previously for Nag3 from Celulomonas fimi and found that there was no 377 detectable difference with our observed hydrolytic activity. The product of the reaction also co-378 migrated with GlcNAc on TLCs and not GlcNAc-1-P ( Figure S3). 379 To further validate its role in PG-recycling we solved the 1.96 Å crystal structure of 380 LpqI (PDB code: 6GFV; Figure 4d, S4, Table S3). LpqI consists of a single TIM-barrel domain 381 similar to cytoplasmic Gram-negative orthologs but lacks the C-terminal domain associated 382 with extracellular NagZ enzymes from some Gram-positive bacteria ( Figure S4). Alignment of 383 LpqI with the NagZ/GlcNAc/1,6-anhydroMurNAc complex from Pseudomonas aeruginosa 384 (NagZPa; PDB:5G3R) or NagZ from Bacillus subtilis (PDB:4GYJ) resulted in a root-mean-385 square deviation of 0.96 Å and 1.01 Å respectively ( Figure S4). Superposition of the post-386 cleavage NagZPa complex with LpqI indicates that the appropriate coordinating residues for 387 substrate recognition are intact in LpqI, supporting its role in PG-recycling (Figure 4d). 388

Characterisation of a ∆lpqI mutant 389
To evaluate the role of LpqI in muropeptide recovery, we constructed a mutant strain of 390 M. bovis BCG lacking lpqI using specialized transduction and confirmed the mutant by PCR 16 . The order in which muropeptides are recycled, and the chemical structure of the 399 recycled material is critical for the immune sensing of the bacterium. To determine the order of 400 PG-recycling steps, we first determined the impact of the loss of lpqI on the recycling of cell 401 wall material. We repeated the radio-label incorporation assay described above with the mutant 402 and observed that the ∆lpqI was able to incorporate 3 H stem-peptides from soluble PG as 403 efficiently as the wild-type (Figure 3b). Consistent with these observations, when we followed 404 release of pre-labelled cells for release of 3 H DAP into the culture media, we observed no 405 significant differences between the wild-type and the ∆lpqI strain (Figure 3c). This experiment 406 reported on the recycling of stem-peptides, however it did not indicate if the mutant strain was 407 still recycling MurNAc. To test this directly we evaluated the ability of the ∆lpqI strain to grow 408 on MurNAc, glycerol and PG. The ∆lpqI strain was not deficient for growth on MurNAc or 409 glycerol, however unlike the wild-type strain it was unable to grow on PG as a sole-carbon 410 source (Figure 3d). Similarly, the ∆lpqI strain incorporated significantly fewer 14 C-GlcNAc-411 labelled muropeptides (Figure 3a). Together these data indicate that in vitro lpqI is required for 412 amino-sugar recycling, but is not necessary for stem-peptide recycling or release. 413 Given that NagZ-like proteins have been found to play a role in β-lactam sensitivity in 414 other bacteria we sought to determine the antibiotic sensitivity of the ∆lpqI strain. In contrast 415 to inhibition of P. aeruginosa NagZ, deletion of lpqI resulted in a significant increase in survival 416 for lysozyme and all cell-wall active antibiotics tested (Figure 5a-d) 44 . A smaller impact on 417 survival in the presence of the protein synthesis inhibitor chloramphenicol was observed 418 (Figure 5e). This increase in resistance is not likely due to a change in cell-wall permeability as 419 determined by ethidium bromide uptake (Figure 5f). 420

In vitro characterization of a ∆lpqI mutant 421
We next sought to determine the impact of the loss of lpqI on host responses to infection.

436
In an attempt to develop diagnostic media for the identification of mycobacteria, several 437 groups in the 1960s observed that M. tuberculosis and most other mycobacteria could not 438 metabolise GlcNAc as a sole carbon-source 33,34 . This, along with the absence of known PG 439 recycling-associated genes lead to the assumption that PG recycling is absent in pathogenic 440 mycobacteria. Our sole-carbon source assays indicate that while the bacteria are unable to 441 metabolise GlcNAc, surprisingly they can metabolise MurNAc (Figure 2). This is despite the 442 fact that they lack an ortholog of the only known lactyl-etherase, MurQ which cleaves an 443 otherwise stable lactyl-ether in the cytoplasm of most model organisms (Figure 1). Our data 444 indicate that rather than using the GlcNAc portion of the sugar, the bacteria are cleaving the 445 lactyl-ether and capable of metabolising the liberated lactate. During our study we found that 446 M. bovis BCG was only able to grow on MurNAc under aerated conditions. This was also found 447 to be the case for L-but not D-lactate which served as a much better carbon source under O2 448 limiting conditions. As MurNAc is a combination of D-lactate and GlcNAc, we can predict that 449 the lactyl etherase acting on MurNAc is likely proceeding via an inverting mechanism. The 450 presence of a specific lactyl-etherase is supported by the turnover of a 4MU-D-lactate reporter 451 compound by M. bovis BCG. The O2 dependence of this growth is intriguing as N-glycolylation 452 is also an O2-dependent activity, suggesting significant alterations to PG metabolism in hypoxic 453 vs. aerobically growing mycobacteria 45 . Consistent with this observation, Rv0237 has a 2-fold 454 upregulation during re-aeration after re-activation from non-replicating persistence in the 455 Wayne hypoxia model 46 . 456 Autolytic enzymes that cleave the glycan backbone of PG such as glucosaminidases, 457 lytic transglycosylases and lysozymes generally produce disaccharides. As such, free MurNAc 458 is unlikely to be generated by the known complement of autolytic enzymes in TB. We therefore 459 sought to identify the biochemical source of free amino-sugars which would feed a PG-460 recycling system. To do this, we biochemically and structurally characterised the predicted 461 mycobacterial NagZ ortholog, LpqI demonstrating that it is an authentic β-N-462 acetylglucosaminidase which is active against PG fragments. Consistent with a role in PG-463 recycling, M. bovis BCG ∆lpqI is unable to grow on soluble PG as a sole carbon source, while 464 recycling of the stem-peptide is unaltered in this mutant (Figure 3). Furthermore, uptake of 465 radio-labelled stem peptides was unchanged in the ∆lpqI mutant whereas 14 C GlcNAc-466 muropeptides show a significant decrease in incorporation (Figure 3). Together, these data 467 demonstrate that M. bovis BCG and M. tuberculosis remove the stem-peptide from PG-468 fragments prior to disaccharide cleavage and lactyl-ether removal (Figure 7). The processing 469 of GlcNAc-MurNAc by LpqI prior to lactyl-ether cleavage is also supported by our LpqI crystal 470 structure in which the lactate-binding residue R67 from the P. aeruginosa structure is conserved 471 The fate of GlcNAc in this pathway remains unclear, although our data and prior 473 observations suggest that the bacteria do not re-use this sugar. This is surprising given the 474 conservation of the nagA (Rv3332) gene in mycobacteria, however it is possible that an 475 alternative pathway exists which involves intermediates not generated under the conditions we 476 have tested. This is hinted at with our 14 C-labelled muropeptides where incorporation of the 477 labelled-GlcNAc is not expected given the lack of GlcNAc utilisation by the cells. It is likely 478 that at least some portion of the labelled material is labelled at MurNAc rather than GlcNAc 479 and that the sugar moiety is in fact used as some alternative reaction product upon cleavage of 480 the lactyl-ether. Bacterial etherases comprise a diverse number of mechanisms and potential 481 reaction products and so a product other than free GlcNAc is entirely possible 47 . We are 482 currently trying to identify and characterise this enzyme. 483 The recycling of bacterial PG has immense implications for the host-pathogen 484 relationship. PG has been shown to be a pathogen-associated molecular pattern and is detected 485 by many different specialised host receptors 8,48 . Of most relevance to M. tuberculosis is the 486 NOD2 receptor which senses intracellular muramyl-dipeptide (MurNAc-L-Ala-D-isoGlu) as a 487 minimal motif 8 . The immunogenicity of Freund's complete adjuvant, for example, is driven by 488 the presence of mycobacterial PG and its N-glycolyl modification 49 . Despite this, Hansen and 489 colleagues observed that the detection of M. tuberculosis by the immune system via Nod2 is 490 weaker than expected, with equal preparations of dead bacteria having substantially more 491 NOD2-stimulatory activity than wild-type bacteria 50 . The authors of that study speculated that 492 this was either due to active repression of the immune system or a reduction in the amount of 493 free NOD2-stimulatory effectors in live bacteria. 494 In our work, we have shown that mycobacteria recycle their PG by first cleaving the 495 stem peptide from the glycan backbone, and subsequently recycle the MurNAc portion of the 496 glycan, removing the D-lactate. This step-wise activity, starting with stem peptide removal, 497 would dramatically reduce the release of NOD2-stimulatory molecules, especially given that 498 this activity is happening beneath the mycomembrane, where diffusion of muropeptides is 499 expected to be highly restricted. Heat-killing of these bacteria would allow host-derived 500 lysozymes to release muropeptides and for those muropeptides to be able to diffuse and 501 stimulate NOD2 and other receptors. In line with this, our preliminary analysis suggest that 502 absence of LpqI does not alter the production of cytokines by infected bone-marrow-derived 503 macrophages (data not shown). 504 Deletion of the lpqI gene from M. bovis BCG yielded several surprising observations. 505 Impaired PG recycling has resulted in a strain that is more resistant to both lysozyme and several 506 antibiotics while not affecting growth in vitro. We are currently investigating the mechanistic 507 basis for this, though it is not likely due to a change in permeability of the cell wall ( Figure 5). 508 In other bacteria cell wall damage can trigger various stress responses, and so it is likely that a 509 build-up of GlcNAc-MurNAc disaccharides may trigger a stress-like response in 510 mycobacteria 51 . Consistent with this lpqI is encoded adjacent to a universal stress response 511 protein in several mycobacteria ( Figure S5). 512 Loss of this gene has also resulted in a substantial increase in growth in bone-marrow 513 derived macrophages ( Figure 6) suggesting that cell-wall turnover may act as a growth-rate 514 modulator in vivo. Despite these apparent fitness advantages, the lpqI gene appears to be intact 515 in virtually all mycobacteria for which sequence data is publicly available, and observed 516 mutations are unlikely to impact catalysis ( Figure S5). This suggests that there is a fitness cost 517 to the inactivation of this gene and warrants further investigation, perhaps in whole organismal 518 models. One possibility is that under stress-conditions mycobacteria may be able to scavenge 519 PG fragments from nearby dead cells allowing a small population to re-grow following mass 520 lysis. This is consistent with the observation that PG can lead to resuscitation of dormant 521 mycobacteria 52 . Alternatively, ∆lpqI-driven excessive growth in the macrophage may prevent 522 the development of a stable, long-term infection. PG recycling has also been shown to be critical 523 for Gram-positive bacteria in stationary phase, though our data do not support this requirement 524 for M. bovis BCG, it is possible that it is more important in the host 53 . 525 In conclusion, we have identified for the first time a PG recovery pathway in pathogenic 526 mycobacteria. We have shown that this occurs in a step-wise fashion by removing stem-peptide 527 from PG and subsequently cleaving the PG-disaccharide and finally releasing the D-lactate from 528 free MurNAc, most likely via an inverting mechanism. Finally, recycling of PG by these 529 bacteria is important for lysozyme and antibiotic tolerance, while deletion of this system results 530 in a significant growth advantage for these bacteria in macrophages. 531

Acknowledgements 532
We wish to thank Sudagar S. Gurcha   functional groups that are typically conserved amongst closely related species, with two major 557 MurNAc recovery systems so far identified (AnmK/MurQ and AngK/MurU). b) For a complete 558 listing of gene conservation see Table S1. The PG recycling machinery is variable with respect 559 to the localisation of NagZ and the subsequent conversion to GlcNAc-1P or UDP-560 GlcNAc/MurNAc. All known MurNAc recovery systems that sustain bacterial growth (as 561 opposed to strictly recycling e.g. P. putida) terminate at MurQ in the cytoplasm. were incubated with 30,000 CPM of 14 C GlcNAc-labelled muropeptides or 100, 000 CPM of 586 3 H DAP-labelled muropeptides for 10 days after which the cell wall material was isolated and 587 subjected to liquid scintillation counting (n = 2). b) The same strains were incubated with 1 588 mM 4MU-GlcNAc in minimal media. After 3 days the fluorescence of the cultures were 589 measured (n = 3). c) M. bovis BCG WT and ∆lpqI were simultaneously evaluated for release of 590 cell wall peptides and growth (n = 3). d) The same strains were evaluated for their growth using 591 glycerol, MurNAc and PG as sole carbon sources using a resazurin assay (n = 3; *** = p < 592 0.001; ** = p < 0.005). 593 594 at a starting OD600 of 0.1. After 7 days incubation total growth was assessed using a resazurin 613 assay, where total fluorescence correlates with respiration and growth (n = 3). f) M. bovis BCG 614 WT, ∆lpqI and ∆lpqI::lpqI and ∆lpqI::EV were incubated with EtBr and the rate of EtBr was 615 monitored as an increase in fluorescence. No significant differences were found in pairwise t-616 tests across all strains (n = 3). 617 618 619 620 Figure 6. Loss of lpqI leads to increased growth in BMDMs. Freshly prepared BMDs were 621 infected with the indicated strains at a multiplicity of infection of 2 and incubated for at 37 °C. 622 At the indicated times the macrophages were lysed with saponin and CFUs were measured on 623 7H11 agar after 3 weeks incubation (n ≥ 4; **** = p < 0.0001).