An Exhaustive Multiple Knockout Approach to Understanding Cell Wall Hydrolase Function in Bacillus subtilis

Most bacteria are surrounded by their cell wall, containing a highly crosslinked protective envelope of peptidoglycan. To grow, bacteria must continuously remodel their wall, inserting new material and breaking old bonds. Bond cleavage is performed by cell wall hydrolases, allowing the wall to expand. Understanding the functions of individual hydrolases has been impeded by their redundancy: single knockouts usually present no phenotype. We used an exhaustive multiple-knockout approach to determine the minimal set of hydrolases required for growth in Bacillus subtilis. We identified 42 candidate hydrolases. Strikingly, we were able to remove all but two of these genes in a single strain; this “Δ40” strain shows only a mild reduction in growth rate, indicating that none of the 40 hydrolases are necessary for growth. The Δ40 strain does not detectably shed old wall, suggesting that turnover is not essential for growth. The remaining hydrolases in the Δ40 strain are LytE and CwlO, previously shown to be synthetically lethal. Either can be removed in Δ40, indicating that either hydrolase alone is sufficient for cell growth. Screening of environmental conditions and biochemistry revealed that LytE activity is inhibited by Mg2+ and that RlpA-like proteins may stimulate LytE activity. Together, these results suggest that the only essential function of cell wall hydrolases in B. subtilis is to enable cell growth by expanding the wall and that LytE or CwlO alone is sufficient for this function. These experiments introduce the Δ40 strain as a tool to study hydrolase activity and regulation in B. subtilis. IMPORTANCE In order to grow, bacterial cells must both create and break down their cell wall. The enzymes that are responsible for these processes are the target of some of our best antibiotics. Our understanding of the proteins that break down the wall – cell wall hydrolases – has been limited by redundancy among the large number of hydrolases many bacteria contain. To solve this problem, we identified 42 cell wall hydrolases in Bacillus subtilis and created a strain lacking 40 of them. We show that cells can survive using only a single cell wall hydrolase; this means that to understand the growth of B. subtilis in standard laboratory conditions, it is only necessary to study a very limited number of proteins, simplifying the problem substantially. We additionally show that the Δ40 strain is a research tool to characterize hydrolases, using it to identify 3 ‘helper’ hydrolases that act in certain stress conditions.

candidate hydrolases. Strikingly, we were able to remove all but two of these genes in a 24 single strain; this "∆40" strain shows only a mild reduction in growth rate, indicating that 25 none of the 40 hydrolases are necessary for growth. The ∆40 strain does not detectably 26 shed old wall, suggesting that turnover is not essential for growth. The remaining 27 hydrolases in the ∆40 strain are LytE and CwlO, previously shown to be synthetically 28 lethal. Either can be removed in ∆40, indicating that either hydrolase alone is sufficient 29 for cell growth. Screening of environmental conditions and biochemistry revealed that 30 LytE activity is inhibited by Mg2+ and that RlpA-like proteins may stimulate LytE activity. 31 Together, these results suggest that the only essential function of cell wall hydrolases in 32 B. subtilis is to enable cell growth by expanding the wall and that LytE or CwlO alone is 33 sufficient for this function. These experiments introduce the ∆40 strain as a tool to study 34 hydrolase activity and regulation in B. subtilis. 35

IMPORTANCE 36
In order to grow, bacterial cells must both create and break down their cell wall. 37 The enzymes that are responsible for these processes are the target of some of our 38 best antibiotics. Our understanding of the proteins that break down the wall -cell wall 39 ∆40 strain is a research tool to characterize hydrolases, using it to identify 3 'helper' 46 hydrolases that act in certain stress conditions.

INTRODUCTION
verified deletion of all modified loci by PCR. After all knockouts had been combined into 115 a single strain, whole-genome sequencing was used to confirm all deletions and to 116 identify any genomic rearrangements or mutations that occurred during the construction 117 process. Despite the multiple rounds of transformation and loop outs this strain was 118 subjected to, with up to 4 resistance cassettes removed simultaneously multiple times, 119 we found no evidence of genomic rearrangements based on read coverage from DNA 120 extracted during exponentially growing cells ( Figure S1) (14), and only 8 SNPs leading 121 to 5 point mutations in genes involved in unrelated processes (Table S1). 122 Ultimately, this effort produced a strain lacking 40 hydrolases, which we termed 123 "∆40". The ∆40 strain is lacking all the identified hydrolases that met our criteria save 124 two -LytE and CwlO, two synthetically lethal endopeptidases previously shown to be 125 essential for growth (10). We were able to further knock out either lytE or cwlO in the 126 ∆40 strain, but not both, due to their synthetic lethality. 127

Hydrolase activity is greatly reduced in the ∆40 strain 128
To assess whether any other unidentified hydrolases remained in the ∆40 strain, 129 we conducted PG profiling of both wild type (WT) cells and the ∆40 strain (15), allowing 130 us to determine the abundance of hydrolase products in their cell walls ( Figure 2, Table  131 unambiguously assign specific PG products to D,D-endopeptidases in these 137

experiments. 138
We compared the relative abundance of different PG hydrolase products in the 139 ∆40 and WT strains ( Figure 2B, Table S3). The ∆40 strain showed a very small amount 140 of amidase activity (~20-fold reduction vs WT, 0.8% vs 0.04% of all muropeptides, 141 p=0.0161, unpaired t test) and a reduction of glucosaminidase activity (~2-fold reduction 142 vs WT, 0.8% vs 0.5% of all muropeptides, p=0.0279, unpaired t test), indicating that 143 these classes of hydrolases had been successfully reduced in the ∆40 strain. The WT backgrounds, consistent with previous reports (12). Furthermore, given that 207 removing cwlO increases the cell width coefficient of variation in the ∆40 background 208 but does not increase the width variation when deleted from WT cells, other hydrolases 209 must also have a role in width homeostasis. 210 We then quantified the chain length for the ∆40 strain and derivatives. Individual In contrast, the ∆40 strain had a significant increase in the average chain length 222 ( Figure 4C, 120 µm), nearly as long as the longest observed WT chain. The ∆40 ∆cwlO 223 strain was similar to the ∆40 strain, while the ∆40 ∆lytE strain had a large increase in 224 the average chain length (250 µm), almost double that of the longest observed WT 225 chain. Because the ∆40 ∆lytE strain lacks all cell separation hydrolases, the remaining 226 cell separation in this strain was likely due to mechanical tearing of cells under the 227 vigorous shaking conditions needed to be able to measure accurate culture OD 600 for these experiments; the ends of the chains had visible, phase-light debris resembling 229 torn cells still attached visible by TEM ( Figure S3). In gentler culture conditions on a 230 roller drum, this strain grows as a large clump of cells visible to the naked eye. 231 Additionally, we tested the ability of the ∆40 strain to sporulate. Hydrolases are 232 involved in both entry into sporulation, as well as exit from the spore during germination 233 (23-25). We found that the ∆40 was not able to sporulate ( Figure 2B, p=0.3859, one 234 sample t test vs efficiency of 0), likely because it lacks SpoIID and SpoIIP, causing a 235 block at the engulfment stage of sporulation (25). 236

∆40 cells do not detectably turn over their cell wall 237
Hydrolases are involved in cell wall turnover, where old PG material is shed from 238 the cell wall (26). We measured the rate of cell wall turnover of both WT and ∆40 cells 239 using pulse-chase labeling with the radioactive cell wall precursor 3 H-N-240 acetylglucosamine ( 3 H-GlcNAc). This revealed that, while WT cells turn over PG at a 241 rate of about 50% per generation in agreement with previous work (26), turnover in ∆40 242 strain was not detectable, with a rate not significantly different from zero ( Figure 5A, 243 p=0.4837, one sample t test vs. rate of 0). These results suggests that LytE and CwlO, 244 the only identifiable remaining hydrolases in the ∆40 strain, likely do not contribute to 245 cell wall turnover. Furthermore, this data suggests that cell wall turnover is not an 246 essential process: cell growth only requires the cleavage of bonds so the cell can 247

expand. 248
As hydrolase-deficient mutants have been shown to have altered cell wall 249 thickness (27, 28) we measured the cell wall thickness of the ∆40 strain using 250 transmission electron microscopy (TEM) and atomic force microscopy (AFM). We found that the wall was significantly thicker only in dehydrated samples measured using AFM 252 ( Figure 5B, center, p=0.0007, unpaired t test); measurements on TEM images or on 253 hydrated AFM samples showed no significant differences (TEM: Figure 5B, left, 254 p=0.1382, unpaired t test; hydrated AFM: Figure 5B, right, p=0.2887, unpaired t test). 255 The WT cell wall was more uniform in appearance in both TEM and AFM images, while 256 the ∆40 strain had more heterogeneity in density and thickness, in particular on the 257 outer face of the wall, with an increase in the presence of "ruffles" on the outer face of 258 the cell wall in the ∆40 strain ( Figure 5C,D). These "ruffles" may represent the additional 259 old cell wall material present due to the strongly reduced turnover rate. The internal face 260 of the cell wall appeared denser than the external face of the wall in both the WT and 261 ∆40 strains ( Figure 5D). The internal face of the cell wall in the ∆40 strain appeared to 262 have both a denser meshwork and an increased number of larger pores as compared 263 with the WT strain. 264 Substantial changes to the cell wall ultrastructure occur during sample prep for 265 TEM, especially in the outer layers of the cell wall (29-31); it would be interesting to 266 apply additional, less perturbative EM modalities such as cryo-electron microscopy to 267 the ∆40 strain to help clarify the exact nature of the changes to the ∆40 cell wall. It is 268 possible that both TEM and hydrated AFM highlights mostly the denser, newer cell wall 269 material, while dehydrated AFM allows visualization of all the cell wall, including the 270 more loosely-bound older wall material. 271

∆40 ∆cwlO cells are sensitive to various stresses, including ionic stress 272
Although the ∆40 strain grew mostly normally under our standard lab conditions, 273 we wondered whether the absence of so many hydrolases would sensitize cells to stress conditions. We used a spot dilution assay to measure the viability of our strains 275 under a variety of stress conditions: temperature, ionic stress, pH, and osmotic stress 276 ( Figure 6). In all conditions, including our control (37˚C), ∆40 cells had fewer CFUs than 277 WT. This is expected because ∆40 cells grow in long chains, and thus cells cannot 278 readily separate into individual CFUs. In all stress conditions ∆40 cells were similarly 279 viable to WT cells, as were ∆lytE, ∆cwlO, and ∆40 ∆lytE cells. However, ∆40 ∆cwlO 280 cells were susceptible to multiple stresses, including low pH, low temperature, and ionic 281

stress. 282
We were particularly intrigued by the susceptibility of ∆40 ∆cwlO to Mg 2+ . Mg 2+ is 283 coordinated between PG and teichoic acids (32), and this Mg 2+ binding is thought to 284 give structural stability to the cell wall (31, 33). High levels of Mg 2+ are often protective 285 against cell wall perturbations, including knockouts of hydrolases, PBPs, or components 286 of the Rod complex (10, 34); thus, the Mg 2+ sensitivity of the ∆40 ∆cwlO strain seemed 287 counterintuitive. Our experiments indicated ∆40 ∆cwlO cells were sensitive to both Ca 2+ 288 and Mg 2+ ; growth was inhibited by the addition of 10 mM MgCl 2 , 10 mM MgSO 4 , and 10 289 mM CaCl 2 , but not by the addition of 20 mM NaCl, suggesting that the growth inhibition 290 was not due to changes in ionic strength or chloride ions. We did observe growth 291 inhibition due to ionic stress at far higher salt concentrations (500 mM NaCl). Notably, 292 cells were not sensitive to an equivalent osmotic stress (1M sorbitol), indicating the 293 sensitivity is to ionic stress, not osmotic stress. 294 As ∆cwlO mutants in the WT background were Mg 2+ insensitive, we sought to 295 identify which hydrolases caused cells to be sensitive to Mg 2+ when they were removed. 296 To find these hydrolases, we returned to intermediate strains used to construct the ∆40 strain, which are missing subsets of hydrolases. We transformed a cwlO knockout into 298 these intermediate strains, then screened these crosses for the same small colony 299 phenotype and the Mg 2+ sensitivity seen in the ∆40 ∆cwlO strain. This identified two 300 genes: yabE and ydjM. Notably, during construction of the ∆40 strain, we had noticed 301 that yocH seemed significant -at several intermediate verification steps, a WT copy of 302 yocH had reintegrated itself during transformation with genomic DNA from single KO 303 strains -we therefore used PCR product for all transformations after this. Furthermore, 304 a ∆ydjM ∆yocH ∆cwlO mutant was previously demonstrated to be sick, with short and 305 sometimes anucleate cells (10). Because yabE, ydjM, and yocH have similar hydrolase 306 domains, and because yocH and ydjM had been identified previously to be involved in a 307 synthetic sick interaction with cwlO, we additionally tested whether the removal of yocH 308 contributed to the ∆40 ∆cwlO Mg 2+ sensitivity phenotype, and found that it did. 309 In total, we identified three genes, yabE, ydjM, and yocH, whose absence in a 310 ∆cwlO background caused the Mg 2+ sensitivity: A ∆yabE ∆ydjM ∆yocH ∆cwlO strain 311 showed a similar stress profile to ∆40 ∆cwlO, including sensitivity to MgCl 2 and CaCl 2 312 ( Figure 6A). yabE, ydjM, and yocH are 3 uncharacterized RlpA-like superfamily domain-313 containing proteins expressed during exponential growth. Like lytE and cwlO, yocH and 314 ydjM are in the walR regulon, while yabE is regulated by sigA (Table 1). All are likely 315 lytic transglycosylases: yocH has been shown to have lytic activity and has homology to 316 the E. coli lytic transglycosylase mltA (35), and all three share a similar catalytic domain. 317 Because yabE, ydjM, and yocH all contain a RlpA-like protein domain, we refer to these 318 genes collectively as RLPAs, and to the triple deletion of all three genes as ∆RLPAs.

LytE is inhibited by Mg 2+ in vitro and in vivo, and RLPAs suppress Mg 2+ lethality in 320 vivo 321
Finally, we sought to identify the source of Mg 2+ growth inhibition in the ∆RLPAs 322 ∆cwlO background. Because LytE is essential in the absence of CwlO, we hypothesized 323 that the sensitivity of the ∆40 ∆cwlO strain to Mg 2+ (and, by extension, the sensitivity of 324 the ∆RLPAs ∆cwlO strain to Mg 2+ ) could be explained by Mg 2+ inhibition of LytE. To 325 investigate this, we first characterized the response of ∆cwlO cells to the removal of 326 LytE. We constructed an otherwise wildtype strain with cwlO knocked out and lytE 327 under inducible control and monitored its growth by time-lapse phase-contrast 328 microscopy. When lytE was induced, cell growth was normal (Movie S1). When lytE 329 induction was removed, cell growth initially slowed, followed by a period of 'stuttery' of Mg 2+ has no effect on cell viability or growth -growth is only inhibited in the absence 338 of the RLPAs. Thus, the RLPAs appear to allow LytE to maintain its activity in the 339 presence of Mg 2+ . This ∆RPLAs ∆cwlO strain additionally had a similar environmental To test whether LytE activity is directly inhibited by Mg 2+ , we overexpressed and 342 purified both full-length LytE and a truncated LytE protein with only its catalytic domain. 343 In vitro activity assays with and without the addition of Mg 2+ showed that indeed LytE 344 activity is inhibited by Mg 2+ ( Figure 7B). Additionally, we reasoned that if the Mg 2+ -345 sensitivity phenotype was due to direct inhibition of LytE by Mg 2+ , increasing the levels 346 of LytE should protect cells from death by increasing the total amount of LytE activity. 347 Indeed, overexpression of LytE allowed the ∆RLPAs strain to survive in the presence of 348 higher levels of Mg 2+ , although 100 mM MgCl 2 still inhibited growth ( Figure 7C). 349 Thus, we conclude that LytE activity is inhibited by Mg 2+ both in vivo and in vitro. The ∆RLPAs ∆cwlO strain also has increased sensitivity to ionic stress and low 360 temperatures, suggesting RLPAs might stimulate LytE activity under those conditions as 361 well.

Bacterial cell growth requires the action of PG hydrolases, but previous in vivo 364
hydrolase studies have been impeded by their diversity and redundancy. We 365 constructed and validated a B. subtilis strain lacking all hydrolases potentially involved 366 in cell growth besides LytE and CwlO. These deletions constitute 40 genes in total, 367 representing 10% of secreted proteins and 1% of all genes. The resulting ∆40 strain 368 enables the investigation of given hydrolases and the cellular contexts in which they 369 function, and in this work, allowed several new discoveries regarding their sufficiency, 370 regulation, and genetic interplay. 371 First, we found that the ∆40 strain is viable. This demonstrates that LytE and 372 CwlO alone can function to expand the cell wall to allow cell growth. Furthermore, as 373 single knockouts of LytE and CwlO in the ∆40 strain are viable and allow growth (albeit 374 at a somewhat reduced rates with some shape defects), this demonstrates B. subtilis 375 requires only one of these two hydrolases to grow. 376 Our minimal hydrolase strain allowed us to show that RlpA-like lytic 377 transglycosylases enhance LytE activity in vivo and that this enhancement can be 378 important for growth under conditions where LytE activity is inhibited, including the 379 presence of divalent cations, ionic stress, and cold. Although the mechanism for LytE 380 enhancement is unclear, we hypothesize that RlpAs stimulate LytE activity via a direct 381 interaction, as has been observed for similar proteins in M. smegmatis (37). Synthetic 382 lethal or synthetic sick interactions are straightforward to identify and characterize in the 383 ∆40 strain, giving a useful tool to interrogate genetic relationships between different 384 hydrolases or between hydrolases and other genes of interest -such as those involved 385 in cell wall synthesis. Surprisingly, the growth rate of the ∆40 strain is only slightly impaired under 387 standard lab conditions. What, then, is the function of these 40 hydrolases, and why 388 does B. subtilis encode so many of them? This multitude of hydrolases likely arises from 389 the fact that hydrolases are involved in other processes aside from cell growth such as 390 sporulation (4) and cell motility (38). Additionally, some hydrolases might be only be 391 needed under nutrient conditions not tested here, such as during phosphate limitation 392 where teichoic acids are not produced, where cells may require hydrolases that are not 393 regulated by teichoic acids (39-41). Finally, these other hydrolases may be important 394 during non-exponential growth states such as during stationary phase, where the 395 recycling of cell wall turnover products, lacking in the ∆40 strain, reduces cell lysis (42). 396 Thus, a broader screen of the sensitivity of the ∆40 strain in different nutrient and 397 environmental conditions will allow the determination of which hydrolases are useful for 398 which conditions. 399 In summary, the ∆40 minimal hydrolase strain provides a powerful experimental 400 background to investigate the function, regulation, and interplay of hydrolases, 401 improving our understanding of precisely how these enzymes conduct their cellular 402 tasks. In the future, individual hydrolases can be reintroduced into the ∆40 strain to 403 investigate their specific activities in the absence of confounding contributions from the 404 other 39 genes. Using the ∆40 strain, PG profiling can determine the biochemical 405 activity of hydrolases. Uncovering synthetic genetic interactions between hydrolases 406 and other genes of interest -now easy to do for all 40 hydrolases at once -will allow us 407 to flesh out our understanding of bacterial cell growth. Understanding the function of cell wall hydrolases is essential for a complete understanding of how bacteria grow, and the 409 ∆40 strain will allow rapid progress to this end. 410

ACKNOWLEDGEMENTS 411
We would like to thank Carl Wivagg, Alex Bisson, Matthew Holmes, Ferran 412 Garcia-Pichel, and Susanne Neuer for helpful advice and discussions, and Georgia 413 Squyres for both helpful advice, discussions, and reading of the manuscript. We thank 414

Strains, media, and growth conditions 426
Glycerol stocks stored at -80˚C were struck onto LB agar plates. For strain 427 bSW61 (lytE::pSpac-lytE, ∆cwlO), these plates were additionally top spread with 1 mM 428 IPTG. After incubation overnight at 37˚C, colonies were inoculated into 1 mL media (the 429 specific media used depended on the experiment, see figure legends for details) and 430 grown on a roller at 37˚C until they reached mid-exponential-phase growth (OD 600 ~0.2). 431 Cells were diluted 1:10 in prewarmed media and again grown until mid-exponential 432 phase; this process was repeated until the start of the experiment. Alternately, a 1:10 433 dilution series of cells were grown overnight in media on a roller at 25 ̊ C. The next day, 434 the culture whose OD 600 was nearest to 0.2 was diluted 1:10 and grown in media at 435 37˚C as above. S7 50 AA indicates S7 50 media with added amino acids as in (43) Table S2, and 460 identified putative membrane-bound/cytoplasmic proteins using UniProt (45). 461

PG purification, HPLC conditions, and MS data analysis 462
PG purification was conducted as in (46), with an HF treatment step instead of 463 HCl to remove teichoic acids and the addition of a protein digestion step. Cells were 464 grown in a baffled flask to an OD 600 of ~0.5 in 50 mL of CH media. Cells were mixed 465 50/50 with 50 mL of boiling 10% SDS and boiled for 15 min in a water bath, then 466 pelleted at 5000x g and washed 5x with ddH2O. Cells were then resuspended in 2 mL 467 DNAse/RNase buffer (10 mM Tris pH 7.5, 2.5 mM MgCl 2 , 0.5 mM CaCl 2 ) with 20 µL 468 DNAse I and 20 µL RNAse A, then incubated overnight at 37˚C and washed 3x with ddH2O to remove nucleic acids. Next, cells were resuspended in 2 mL Proteinase K 470 buffer (10 mM Tris pH 7.5, 1 mM CaCl 2 ) with 20 µL Proteinase K, incubated overnight at 471 45 0 C, and washed 3x with ddH2O to remove proteins. Next, cell walls were treated with 472 48% (v/v) hydrofluoric acid on ice for 24H, then washed twice with 100 mM Tris pH 8 473 and 4 times with ddH2O. Then, the PG was resuspended in 12.5 mM NaHPO 4 pH 5.5 474 with 5000 units of mutanolysin and digested overnight (16h) at 37˚C on a roller to yield 475 soluble muropeptides. Undigested material was pelleted by spinning at 16000x g for 5 476 mins and the supernatant was transferred to a new tube. Soluble muropeptides were 477 reduced with sodium borohydride (1 mg/mL) for 30 mins and the reaction was stopped 478 by adding 10 µL 30% phosphoric acid. The pH was adjusted to 4-6 using NaOH, and Feature detection was performed on the raw MS data using Dinosaur (47). 485 Feature detection was done separately on both the positive and negative mode scans 486 with default parameters. Feature data were analyzed using a custom MATLAB program, 487 available at https://bitbucket.org/garnerlab/wilson_40_2020/. We first filtered feature 488 data for charge < 3. Next, we filtered for the top 10 features present during each scan. 489 For each of these features, theoretical m/z values were compared with observed m/z 490 with a cutoff of 10 ppm. We required that a compound be present on both the positive 491 and negative scans, and consolidated features matching the same compound within a retention time within 1 min. Finally, we filtered out compounds corresponding to in-493 source decay (loss of glucosamine without a change in retention time), and compounds 494 present at less than 0.1% of all muropeptides. Retention times shown in Table S3 were 495 analyzed manually. 496

Growth rates 497
Cells were grown to an OD 600 of ~0.3-0.5 on a roller drum at 37˚C and diluted to 498 an OD 600 of ~0.05 in baffled flasks in a water bath shaker at 37˚C. Samples were 499 withdrawn at 5 min intervals and OD 600 was measured in a plastic cuvette using a 500 Biowave Cell Density Meter CO8000. T vs. OD 600 curves were fit to a single exponential 501 (OD 600 = Ae BT ) to extract a growth rate (B). 502

Autolysis rates 503
Cells were grown to an OD 600 of ~0.3-0.5 on a roller drum at 37˚C and diluted to 504 an OD 600 of ~0.025 into prewarmed CH in baffled flasks at 37˚C. Once cells reached an 505 OD 600 of 0.5, sodium azide (75 mM final) or ampicillin (100 µg/mL final) was added to 506 part of culture and transferred to a prewarmed 96 well plate. OD 600 was measured every 507 2 minutes for 24H at 37˚C using a BioTek Epoch 2 Microplate Spectrophotometer. The 508 plate was shaken at maximum RPM in between measurements. 509

Sporulation efficiency 510
Sporulation was induced by resuspension according to (44). Cells were grown to 511 an OD 600 of ~0.3-0.5 in CH media, pelleted, and resuspended in resuspension medium. 512 Sporulation efficiency was assessed by measuring the number of heat-resistant CFUs 513 per mL of culture after 36H. The cultures were heated to 80˚C for 20 mins, then plated. 514 CFU counts were then done after 24H of incubation at 37˚C.

Turnover rates 516
Turnover rates were measured as in (43)  Waltham, MA, USA) and radioactivity was measured using a scintillation counter (Tri-528 Carb 2100 TR, PerkinElmer). Decays/min vs. OD 600 plots were fit to a single exponential 529 (DPM = Ae BT ) to extract a turnover rate (B). 530

Cell dimensions 531
Cells were grown to an OD 600 of ~0.3-0.5 in a water bath shaker at 37˚C. 1 mL of 532 culture was stained with FM 5-95 and concentrated to 100 µL by centrifugation at 2000x 533 g and resuspension. 5 µL of concentrated cells were spotted under 2% (w/v) agarose 534 pads in CH containing 0.5 ug/mL FM 5-95. Images were collected on a Nikon Ti-E 535 microscope using a Nikon CFI Plan Apo DM Lambda 100X Oil objective, 1.45 NA, 536 phase ring Ph3 using an ORCA-Flash4.0 V2 sCMOS camera. Analysis was performed 537 using Morphometrics v1.1 (48). Zero length or width cells were discarded, as well as any cells with width greater than length. Outliers were removed using Graphpad Prism 539 ROUT with default parameters (1%). 540

Chain length 541
Cells were grown to an OD 600 of ~0.3-0.5 on a roller drum at 37˚C and diluted to 542 an OD 600 of ~0.025 into prewarmed CH in baffled flasks at 37˚C. Once cells reached an 543 OD 600 of 0.5, 1-5 µl of culture was spotted under a prewarmed 2.5% (w/v) agarose pad 544 in CH. A 10x10 image tile series was collected (~1.5 mm square). A custom MATLAB 545 program was used to register and stitch the images together, and then chain length was 546 measured manually with the assistance of a custom MATLAB program. 547

Electron microscopy and cell wall thickness measurements 548
Electron microscopy was performed as in (49). Briefly, exponentially growing 549 cells were fixed in 100 mM MOPS buffer pH 7 containing 2% (w/v) paraformaldayde,  Images collected were segmented (inner cell wall, outer cell wall) using DeepCell and outer cell wall was measured every 10 nm along a user-defined line, and the mean 562 of that measurement was taken to be the cell's cell wall thickness. 563

AFM imaging and cell wall thickness measurement 564
B. subtilis cells at mid-exponential phase were boiled rapidly to kill bacterial cells 565 & inactivate any potential hydrolase activity. Cells were broken by French Press and 566 FastPrep, then suspended in 5% (w/v) SDS and boiled for 25 min, and sacculi collected 567 by centrifugation at 20,000 g for 3 mins. The resulting pellets were washed with distilled 568 water to remove all traces of SDS, then re-suspended in Tris-HCl (50 mM, pH7) 569 containing 2 mg/ml pronase and incubated at 60°C for 90 mins. The resulting sacculi 570 were then re-suspended in LC-MS Chromasolv water for storage at -20°C. 571 Freshly cleaved mica discs were incubated with Cell-Tak™, (285 ml 100 mM 572 NaHCO 3 (pH 8) then 10 µl of Cell-Tak (Corning, 5% (w/v) in acetic acid) and 5 µl of 1 M 573 NaOH, covered and left for 20 minutes then washed five times with HPLC grade water) 574 to ensure attachment of sacculi on the glass surface. Sacculi were diluted in HPLC-575 grade water to appropriate concentration and dried onto mica using N 2 . These were 576 further washed and dried with N 2 again to remove any unattached sample. 577 All AFM data was taken on a JPK Nanowizard III in QI (quantitative imaging) 578 mode. Samples were imaged in HPLC Grade water using a FastScanD cantilever 579 (Bruker, Santa Barbara), nominal spring constant 0.25 N/m with a 256 x 256-pixel scan 580 region, driven at ~167 Hz with a typical Z length of ~ 300 nm using peak interaction 581 forces of 2-3 nN. Images were flattened to median of differences and first order planefit 582 using Gwyddion.
Cells were grown to an OD 600 of 0.5 and diluted 1:10 into 100 µL of LB media in a 585 96 well plate. A 1:10 serial dilution series was made, and 3 µL of each dilution was 586 spotted onto the plate using a multichannel pipettor. The plates were allowed to dry and 587 incubated in at 37˚C or 42˚C as indicated for 18h. Plates incubated at 25˚C or 18˚C 588 were left for additional time (24h and 48h, respectively). Plates were photographed 589 using a Canon SC1011 scanner with the lid open. 590 For the colony morphology assay in Figure 1, this protocol was followed except 591 that a colony of cells of each strain were simply resuspended in 100 µL of media using a 592 toothpick (omitting the broth culture step). 593

LytE purification 594
His-SUMO-tagged full length LytE missing the signal peptide (26-355) and His- Fractions contained cleaved protein were pooled and concentrated to a volume of 2 mL, 616 then stored in dialysis in cleavage buffer. Activity tests were performed using purified 617 PG in cleavage buffer plus 0.5% (w/v) Pluronic F-108 and PG from Sigma to an OD 600 618 of 0.25 at 37˚C. OD 600 was measured every 2 minutes for 24H using a BioTek Epoch 2 619 Microplate Spectrophotometer. The plate was shaken at maximum RPM in between 620 measurements. 621