Discovery of a two protease DNA damage checkpoint recovery mechanism

The DNA damage response is a signaling pathway found throughout biology. In many bacteria the DNA damage checkpoint is enforced by inducing expression of a small, membrane bound inhibitor that delays cell division providing time to repair damaged chromosomes. How cells sense successful DNA repair and promote checkpoint recovery is unknown. By using a high-throughput, forward genetic screen, we identified two unrelated proteases, YlbL and CtpA, that promote DNA damage checkpoint recovery in Bacillus subtilis. Deletion of both proteases leads to accumulation of the checkpoint protein YneA. DNA damage sensitivity and increased cell elongation in protease mutants depends on yneA. Further, expression of YneA in protease mutants was sufficient to inhibit cell proliferation. Finally, we show that one of the two proteases, CtpA, directly cleaves YneA in vitro. With these results, we report the mechanism for DNA damage checkpoint recovery in bacteria that use membrane bound cell division inhibitors.


Introduction 28
The DNA damage response (DDR, SOS response in bacteria) is an important pathway for 29 maintaining genome integrity in all domains of life. Misregulation of the DDR in humans can 30 result in various disease conditions (1, 2), and in bacteria the SOS response has been found to be 31 important for survival under many stressors (3)(4)(5). The DNA damage response in all organisms 32 results in three principle outcomes: a transcriptional response, DNA repair, and activation of a 33 DNA damage checkpoint (6)(7)(8). In eukaryotes, the G1/S and G2/M checkpoints are established 34 by checkpoint kinases, which transduce the signal of DNA damage through inactivation of the 35 phosphatase Cdc25 (7). Checkpoint kinase dependent inhibition of Cdc25 leads to accumulation 36 of phosphorylated cyclin dependent kinases, which prevents cell cycle progression (7). In 37 phenotypes ( Fig 2C). The variants and the wild-type proteases were ectopically expressed to the 151 same level in vivo (Fig 2D), suggesting that the lack of complementation is not due to instability 152 caused by the amino acid changes. With these results, we conclude that protease activity is 153 required for YlbL and CtpA to function in response to DNA damage. 154 ylbK disruption results in a polar effect on ylbL 155 We noticed that ylbK, the gene upstream of ylbL, had a phenotype similar to ylbL in the Tn-seq 156 experiments (Table S2 & S3). We tested whether ylbKL functioned together in the DNA damage 157 response. Deletion of ylbK resulted in sensitivity to MMC ( Fig S3A). Ectopic expression of ylbK 158 failed to complement the Δ ylbK phenotype (Fig S3A). Given that ylbK is upstream of ylbL we 159 attempted to complement the Δ ylbK phenotype using ylbL and found that sensitivity to MMC 160 was rescued (Fig S3A). Closer examination of the ylbKL locus revealed that a putative ribosome 161 binding site (RBS) for ylbL translation was present within the 3' end of ylbK ( Fig S3B). Thus, a 162 second deletion of ylbK was made (ΔylbK-2) which included deletion of the codons for all but 163 the first 3 and the last 14 amino acids, leaving the RBS for ylbL intact (Fig S3B). This deletion 164 was not sensitive to MMC (Fig S3A). Western blotting revealed that the initial Δ ylbK construct 165 did not express YlbL, whereas Δ ylbK-2 did ( Fig S3C). As a result, we conclude that disruption of 166 ylbK results in a polar effect on ylbL, indicating that YlbL functions independently of YlbK. 167 In order to better understand the prevalence of false positives in Tn-seq experiments we 168 attempted to validate the MMC phenotypes of the forty genes with the lowest relative fitness 169 values in the second growth period of the MMC experiment. Intriguingly, we found that seven 170 additional genes, queA, ylmG, lgt, ylmE, sdaAB, cymR, and ywrC, resulted in no sensitivity to 171 MMC when deleted (Table 1). We also found that the genomic loci of queA, ylmG, ylmE, and 172 sdaAB, were proximal genes with validated phenotypes (Table 1). Taken together, our results 173 underscore the importance of validating results from forward genetic screens. 174

YlbL and CtpA have overlapping functions 175
The similarity in phenotypes led us to hypothesize that YlbL and CtpA have overlapping 176 functions. To test this we performed a cross-complementation experiment using spot-titer assays 177 for MMC sensitivity. Over-expression of YlbL, but not YlbL-S234A, complemented a ctpA 178 deletion (Fig 3A & B). Similarly, over-expression of CtpA, but not CtpA-S297A, complemented 179 a ylbL deletion (Fig 3A & B). In addition, deletion of both proteases rendered B. subtilis 180 hypersensitive to MMC, even more so than loss of uvrA, which codes for the protein responsible 181 for recognizing MMC adducts as part of nucleotide excision repair (Fig 3C;37,38). To further 182 test the hypothesis that YlbL and CtpA have overlapping functions, we over-expressed each of 183 the proteases separately in the double protease mutant background and observed a complete 184 rescue of MMC sensitivity upon expression of the wild type (WT), but not the serine variants 185 ( Fig 3D & E). 186

DNA damage-dependent cell division delay is increased in protease deletions 187
The experiments performed thus far cannot distinguish between sensitivity to MMC resulting 188 from cell death, growth inhibition or both. To determine whether sensitivity arises from cell 189 death, we performed a survival assay using an acute treatment of MMC. We detected a slight 190 decrease in percent survival as MMC concentration increased in the Δ ylbL and the double mutant 191 strain ( Fig S4A). We compared the decrease in percent survival in single and double protease 192 mutants to a Δ uvrA strain, which has been shown previously to be acutely sensitive to MMC 193 (39). The strain lacking uvrA was very sensitive to an acute treatment of MMC (Fig S4A), 194 whereas, the double protease deletion strain was significantly less sensitive to acute exposure 195 compared with Δ uvrA (compare Fig 3C & S4A). Taken together, we conclude that MMC 196 sensitivity of the protease mutants observed in spot-titer assays is primarily caused by growth 197 inhibition. 198 We hypothesized that sensitivity to DNA damage resulting from growth inhibition could 199 also be explained by inhibiting cell proliferation, or inhibiting cell division rather than cell 200 growth. To distinguish between these two possibilities, we measured cell length, because 201 inhibition of proliferation should be observed as an increase in cell length, consistent with a 202 failure in checkpoint recovery. Thus, we designed a MMC recovery assay, reasoning that 203 following treatment with MMC, cells lacking YlbL, CtpA, or both, would remain elongated 204 showing slower checkpoint recovery relative to the WT strain. We grew cultures either in a 205 vehicle control or in the presence of MMC. After a two hour treatment, the MMC containing 206 media was removed and cells were washed. Cells were then transferred to fresh media without 207 MMC and allowed to continue growing to assay for checkpoint recovery. Although cells 208 appeared to be elongated in the Δ ylbL and double mutant strains, there was heterogeneity in the 209 population ( Fig 4A). As a result, we measured the cell length of at least 900 cells for each 210 genotype and each condition and plotted the cell length distributions as histograms ( Fig 4B). 211 There was no difference in the vehicle control cell length distributions (Fig 4B). The MMC 212 treatment of all strains resulted in a rightward shift in the distribution for all strains (Fig 4B,  213 compare upper panels). When comparing the protease deletions to the WT, the difference in 214 distribution could be visualized by considering the percentage of cells greater than 6.75 μ m in 215 length, which is about three cell lengths of 2.25 μ m each. We found that deletion of ylbL resulted 216 in an increase in the percentage of cells longer than 6.75 μ m in MMC treated cultures and after 217 both 2 hours and 4 hours of recovery ( Fig 4C). Deletion of ctpA, however, resulted in a very 218 slight, though significant (p-value = 0.0142 for one-tailed Z-test), increased percentage of cells 219 longer than 6.75 μ m after 4 hours of recovery ( Fig 4C). The double mutant resulted in a 220 percentage of cells slightly greater than Δ ylbL alone after both 2 hours (p-value = 0.0001 for one-221 tailed Z-test) and 4 hours (p-value = 0.0088 for one-tailed Z-test) of recovery ( Fig 4C). 222 Taken together, we conclude that YlbL is the primary protease under these conditions, 223 with CtpA also contributing. We also conclude that cells lacking YlbL or both YlbL and CtpA 224 take longer to divide following exposure to MMC, which is consistent with DNA damage 225 sensitivity resulting from inhibition of cell proliferation. Further, the observation of inhibition of 226 cell proliferation suggests that YlbL and CtpA proteases could be important for DNA damage 227 checkpoint recovery (see below). 228

YlbL and CtpA levels are not regulated by DNA damage 229
A potential model to regulate YlbL and CtpA in response to DNA damage is to increase protein 230 levels following exposure to DNA damage. Increased protease levels in response to DNA 231 damage could promote the DNA damage checkpoint recovery when needed. To test this model, 232 we monitored YlbL and CtpA protein levels via Western blotting over the course of the MMC 233 recovery assay. YlbL and CtpA protein levels did not change relative to the loading control 234 DnaN throughout the course of the experiment (Fig S4B & C). As a positive control, we 235 performed the same experiment and monitored RecA protein levels and found that, indeed, RecA 236 protein levels increased (Fig S4B & C), as expected because recA is induced as part of the SOS 237 response (41, 42). We conclude that YlbL and CtpA protein levels are not regulated by DNA 238 damage. 239

The cell division inhibitor YneA accumulates in protease mutants 240
The data presented thus far led us to hypothesize that in the absence of YlbL and CtpA, a protein 241 accumulates, resulting in inhibition of cell division (Fig 5A). To identify the accumulating 242 protein, we performed an analysis of the entire proteome of WT and double protease mutant cell 243 extracts. We chose to analyze the proteomes of cells after two hours of recovery, because the cell 244 length distributions differed most between WT and the double protease mutant (Fig 4B). We 245 found that the normalized spectral count data had similar distributions for both WT and the 246 double mutant, which were approximately log normal ( Fig S5A). We verified that the 247 distribution of the test statistic (the difference in double mutant average and WT average) was 248 normally distributed (Fig S5B), thus allowing a t-test to be used. We also performed a principle 249 component analysis and found that WT replicates and double mutant replicates each clustered 250 together ( Fig S5C). 251 In total, 2329 proteins were detected, and 183 proteins were found to be differentially 252 represented (p-value < 0.05) in the double mutant relative to WT (Table S4). Of the proteins 253 differentially represented in the double mutant, 104 had a fold change greater than one ( Fig 5B,  254 red points). There are three major mechanisms that have been reported in B. subtilis to inhibit 255 cell division: 1) Noc dependent nucleoid occlusion (43), 2) FtsL depletion (44, 45), and 3) 256 expression of YneA (22). One possibility was that Noc protein levels were higher in the double 257 mutant, but we observed no difference in Noc levels ( Fig S5D). Another possibility was that 258 FtsL or the protease RasP, which degrades FtsL, was affected in the protease mutant (45). We 259 found no difference in relative protein abundance of FtsL or RasP (Fig S5D), ruling out the FtsL 260 RasP pathway. Among the top 10 proteins that were more abundant in the double mutant was 261 YneA, the SOS-dependent cell division inhibitor (Table S4). We asked if the enrichment of 262 YneA was simply because it is SOS induced. We analyzed the relative abundance of several 263 other proteins that are known to be SOS induced, including RecA,UvrA,UvrB,DinB,and YneB 264 (41), which is in an operon with YneA (22). We found that none of these other proteins were 265 enriched in the double mutant ( Fig S5E). These results suggest that YneA accumulation is not a 266 result of increased SOS activation, and regulation of YneA accumulation is likely to be post 267 translational, because the protein levels of another member of the operon, YneB were 268 unchanged. Taken together, our proteomics data suggest that YlbL and CtpA promote DNA 269 damage checkpoint recovery through regulating YneA protein abundance. 270 We directly tested for YneA accumulation in protease mutants throughout the MMC 271 recovery assay using Western blotting. YneA accumulated in all protease deletion strains after 2 272 hours and 4 hours of recovery, though YneA accumulation in Although accumulation of YneA fit our data well, we considered that the other proteins enriched 280 greater than five-fold in the double mutant may have contributed to the DNA damage sensitivity 281 phenotype. To test this, we constructed deletions of each gene in WT and the double mutant and 282 tested for MMC sensitivity. We found that no single deletion of each of the 10 genes resulted in 283 sensitivity to MMC (Fig S6A). In the double mutant, only deletion of yneA was able to rescue 284 the sensitivity to MMC (Fig S6A). We verified that deletion of yneA could rescue MMC 285 sensitivity in all protease mutant backgrounds ( Fig 6A). We examined cell length in the DNA 286 damage recovery assay. As expected, deletion of yneA resulted in less severe cell elongation 287 relative to WT (compare WT in Fig 4B and Δ yneA∷loxP in Fig 6B). In addition, deletion of 288 ylbL, ctpA, or both no longer changed the cell length distribution in the absence of yneA at the 289 two hour recovery time point (Fig 6B, 6C, and S6B). In the MMC treatment, we did observe a 290 slight increase (p-value = 0.0004 for one-tailed Z-test) in the percentage of cells greater than 6.75 291 μ m in the double protease deletion strain compared to WT ( Fig 6C). Given that MMC sensitivity 292 and most cell elongation in protease mutants depends on yneA, we hypothesized that expression 293 of YneA alone would be sufficient to inhibit growth to a greater extent in the protease mutants. 294 Indeed, strains lacking YlbL, CtpA or both were more sensitive to over-expression of yneA from 295 an IPTG inducible promoter than WT ( Fig 6D). Further, we show that YneA accumulated in the 296 protease mutant strains following yneA ectopic expression ( Fig 6E). We conclude that YneA 297 accumulation results in severe growth inhibition in cells lacking YlbL and CtpA. 298

CtpA specifically digests YneA in vitro 299
To test the hypothesis that YneA is a direct substrate of the proteases we purified YneA, CtpA, 300 and YlbL lacking their N-terminal transmembrane domains. We were unable to detect protease 301 activity from YlbL using YneA, lysozyme, or casein as substrates (unpublished observations; see 302 discussion). When purified CtpA was incubated with YneA, we observed digestion of YneA 303 over time, but no digestion was observed using CtpA-S297A ( Fig 7A). To test if CtpA activity 304 against YneA was specific we completed the same reaction using lysozyme as a substrate and 305 detected no activity ( Fig 7B). We conclude that YneA is a direct and specific substrate of CtpA. 306

Discussion 307
All organisms control cellular processes through regulated signaling. To regulate a cellular 308 process a signaling pathway must have mechanisms of activation and inactivation. Many bacteria 309 use a small membrane protein as an SOS-induced DNA damage checkpoint protein (21-24). The 310 mechanism of checkpoint recovery, however, for organisms using membrane protein checkpoints 311 has remained unclear. Our comprehensive study identified a dual protease mechanism of DNA 312 damage checkpoint recovery (Fig 7C). Proteases YlbL and CtpA are constitutively present in the 313 plasma membrane of cells even in the absence of DNA damage. After encountering DNA 314 damage, YneA expression is induced. We hypothesize that YlbL and CtpA activities become 315 saturated by increased YneA expression, which results in a delay of cell division. Following 316 DNA repair, expression of YneA decreases and YlbL and CtpA clear any remaining YneA 317 allowing cell division to resume. DNA damage checkpoints are of fundamental importance to 318 biology, and we have discovered the pathway responsible for checkpoint inactivation and cell 319 cycle re-entry in B. subtilis. With these results, we propose to rename YlbL to LmpA (Lon 320 membrane anchored protease A) to reflect its activity as a membrane anchored Lon protease. 321 Recovery from a DNA damage checkpoint is a critical process for all organisms. One 322 theme found throughout biology is the use of multiple proteins with overlapping functions. In 323 eukaryotes, the phosphorylation events that establish the checkpoint are removed by multiple 324 phosphatases (46,47). In E. coli, there are two cytoplasmic proteases, Lon and ClpYQ, that have 325 been found to degrade the cell division inhibitor SulA (15,(17)(18)(19). Our study further extends the 326 use of multiple factors in regulating checkpoint recovery to B. subtilis, by describing a 327 mechanism using two extracellular proteases. In eukaryotes, multiple proteins with overlapping 328 functions often exist due to spatial or temporal restrictions, which appears to at least partially 329 explain the use of multiple factors in checkpoint recovery (46,47). In E. coli, ClpYQ was found 330 to be important at higher temperatures in the absence of Lon (18), again suggesting that each 331 protease functions under specific conditions. In the case of YlbL and CtpA, however, there 332 appears to be a shared responsibility in rich media. Deletion of each protease results in DNA 333 damage sensitivity and the double mutant has a more severe sensitivity. In contrast, during 334 growth in minimal media, YlbL appears to be the primary protease, as the cell elongation 335 phenotype is more pronounced in cells lacking ylbL. Still it is unclear how or when each protease 336 functions. Why isn't one protease sufficient to degrade YneA? Do the proteases occupy distinct 337 loci in the cell, requiring that each protease degrades a specific YneA pool? Another possibility 338 is that protease levels are constrained by another evolutionary pressure, such as substrates unique 339 to each protease. Thus, the cell cannot maintain the individual proteases at levels required to 340 titrate YneA as part of the DDR, because the levels of another substrate would be too low. 341 Another explanation is that using multiple factors is an evolutionary strategy that increases the 342 fitness of an organism. It is clear that checkpoint recovery is crucial, because the fitness of cells 343 lacking ylbL or ctpA is significantly decreased in the presence of DNA damage (Table S2). 344 Although the dual protease mechanism described here resolves an important step in the 345 DDR, our data also reveal the complexity of the system. After we exposed cells to MMC the 346 cells elongated. We noticed however, that not all elongation depended on yneA (see Fig 6B), 347 suggesting another mechanism for cell cycle control. In B. subtilis, there have been reports of 348 yneA-independent control of cell division following replication stress (44,48,49). The essential 349 cell division component FtsL has been reported to be unstable and depletion leads to inhibition 350 of cell division (49). Further, ftsL transcript levels were reported to decrease following 351 replication stress independent of the SOS response (44), thus linking depletion of the unstable 352 FtsL protein to cell division control following replication stress. A study using a replication 353 block consisting of the Tet-repressor bound to a Tet-operator array, observed cell division 354 inhibition independent of yneA, noc, and FtsL (48). Interestingly, recent studies of Caulobacter 355 crescentus uncovered two cell division inhibitors that are expressed in response to DNA damage, 356 with one SOS-dependent inhibitor and the other SOS-independent (21, 25). In B. megaterium, a 357 recent study found that the transcript of yneA is unstable following exposure to DNA damage 358 (50), suggesting yet another layer of regulation. No factor was identified to regulate yneA 359 transcripts in the previous study, though it is possible that one of the genes of unknown function 360 identified in our screens could regulate yneA mRNA. These studies highlight the complexity of 361 regulating the DNA damage checkpoint in bacteria. 362

Tn-seq 380
A transposon insertion library was constructed similar to as described (52) with modifications 381 described in the supplemental methods. Tn-seq experiments were designed with multiple growth 382 periods similar to a prior description (27), with a detailed description in the supplemental 383 methods. Sequencing library construction and data analysis were performed as described 384 previously (34, 52) with modifications described in the supplemental methods. 385

Spot-titer assays 386
B. subtilis strains were struck out on LB agar and incubated at 30°C overnight. The next day, a 387 single colony was used to inoculate a 2 mL LB culture in a 14 mL round bottom culture tube, 388 which was incubated at 37°C on a rolling rack until OD 600 was 0.5-1. Cultures were normalized 389 to OD 600 = 0.5 and serial diluted. The serial dilutions were spotted (4 μ L) on the agar media 390 indicated in the figures and the plates were incubated at 30°C overnight (16-20 hours). All spot-391 titer assays were performed at least twice. 392

Survival assays 393
Survival assays using an acute treatment of mitomycin C were performed as previously 394 described (53). Cultures were grown to an OD 600 of about 1, and triplicate samples of 0.6 mL of 395 an OD 600 = 1 equivalent was taken and cells were pelleted via centrifugation: 10,000 g for 5 396 minutes at room temperature (all subsequent centrifugation steps were identical). Cells were 397 washed with 0.6 mL 0.85% NaCl (saline) and pelleted via centrifugation. Cell pellets were re-398 suspended in 0.6 mL saline, and 100 μ L aliquots were distributed for each MMC concentration. 399 MMC was added to each tube to yield the final concentration stated in the figure, and cells were 400 incubated at 37°C for 30 minutes. Cells were pelleted via centrifugation to remove MMC, re-401 suspended in saline, and a serial dilution yielding a scorable number of cells (about 30-300) was 402 plated on LB agar to determine the surviving fraction of cells. Each experiment was performed 403 three times in triplicate for each strain. 404

Antiserum production 405
Purified proteins (see below for purification protocols) were submitted to Covance for antibody 406 production using rabbits. Two rabbits were used in the 77 day protocol, and the serum with the 407 least background was used for experiments. 408

Western blotting 409
For YlbL, CtpA, RecA, and DnaN Western blots, a cell pellet equivalent of 1 mL OD 600 = 1 was 410 re-suspended in 100 μ L 1x SMM buffer (0.5 M sucrose, 0.02 M maleic acid, 0.02 M MgCl 2 , 411 adjusted to pH 6.5) containing 1 mg/mL lysozyme and 2x Roche protease inhibitors at room 412 temperature for 1 or 2 hours. Samples were then lysed by addition of 6x SDS loading dye (0.35 413 M Tris, pH 6.8, 30% glycerol, 10% SDS, 0.6 M DTT, and 0.012% bromophenol blue) to 1x. 414 Samples (12 μ L) were separated via 10% SDS-PAGE, and transferred to nitrocellulose using a 415 Trans-Blot Turbo (BioRad) according to the manufacturer's directions. Membranes were 416 blocked in 5% milk in TBST (25 mM Tris, pH 7.5, 150 mM NaCl, and 0.1% Tween 20) at room 417 temperature for 1 hour or at 4°C overnight. Blocking buffer was removed, and primary 418 antibodies were added in 2% milk in TBST (αYlbL, 1:5000 or 1:8000; α DnaN, 1:4000). Primary antibody incubation was performed at room temperature for 1 420 hour or overnight at 4°C. Primary antibodies were removed and membranes were washed three 421 times with TBST for 5 minutes at room temperature. Secondary antibodies (Licor; 1:15000) were 422 added in 2% milk in TBST and incubated at room temperature for 1 hour. Membranes were 423 washed three times as above and imaged using the Li-COR Odyssey imaging system. All 424 Western blot experiments were performed at least twice with independent samples. 425 For YneA Western blots, cell pellets, 10 mL OD 600 = 1 for MMC recovery assay and 25 426 mL OD 600 = 1 for over-expression, were re-suspended in 400 or 500 μ L, respectively, of 427 sonication buffer (50 mM Tris, pH 8.0, 10 mM EDTA, 20% glycerol, 2x Roche protease 428 inhibitors, and 5 mM PMSF), and lysed via sonication. SDS loading dye was added to 2x and 429 samples were incubated at 100°C for 7 minutes. Samples (10 μ L) were separated using 16.5% 430 Tris-Tricine-SDS-PAGE (BioRad) and transferred to a nitrocellulose membrane using a Trans-431 blot Turbo (BioRad) according to the manufacturer's directions. All subsequent steps were 432 performed as above with a 1:3000 primary antibody dilution. 433

Mitomycin C recovery assay 434
An LB agar plate grown at 30°C overnight was washed with pre-warmed S7 50 minimal media 435 and used to inoculate a culture of S7 50 minimal media at an OD 600 = 0.1. The cultures were 436 incubated at 30°C until an OD 600 of about 0.2 (2-2.5 hours). MMC was added to 100 ng/mL and 437 cultures were incubated at 30°C for 2 hours. Cells were pelleted via centrifugation (4,696 g for 7 438 minutes) and the media was removed. Cell pellets were washed in an equal volume of 1x PBS, 439 pH 7.4, and pelleted again via centrifugation as above. Cell pellets were re-suspended in an equal 440 volume of pre-warmed S7 50 minimal media and incubated at 30°C for four hours. Samples for 441 microscopy and Western blot analysis were taken after the two hour MMC treatment and at two 442 and four hours following recovery, as indicated in the figures. The vehicle control samples were 443 treated for 2 hours with an equivalent volume of the vehicle in which MMC was suspended (25% 444 (v/v) DMSO). 445

Microscopy 446
A 500 μ L sample from the MMC recovery assay above was taken and FM4-64 was added to 2 447 μ g/mL and incubated at room temperature for 5 minutes. Samples were then transferred to 1% 448 agarose pads made of 1x Spizizen's salts. Images were captured using an Olympus BX61 449 microscope. 450

Cell length analysis 451
Cells were scored for cell length using the measuring tool in ImageJ software. For each image 452 scored, all cells that were in focus were measured. The number of cells scored for each 453 strain/condition is stated in the figures (n=cells measured). The histograms were generated using 454 ggplot2 in R. All scoring was done using unadjusted images. Representative images shown in the 455 figures were modified in ImageJ by subtracting the background (rolling ball radius method) and 456 adjusting the brightness and contrast. Any adjustments made were applied to the entire image. 457

Proteomics experimental details 458
Samples (5 mL OD 600 = 1) were harvested from cultures grown as described in the MMC 459 recovery assay section at 2 hours recovery via centrifugation: 4,696 g at room temperature for 10 460 minutes. Samples were washed twice with 500 μ L 1x PBS, pH 7.4 and pelleted via 461 centrifugation: 10,000 g at room temperature for 5 minutes. Samples were frozen in liquid 462 nitrogen and stored at -80°C. Samples were submitted for mass-spectrometry analysis to MS 463 Bioworks. Further sample processing and data analysis was performed by MS Bioworks as 464 described in the supplemental methods.   Tables 695 Table 1 Tn-seq yields many false positive results. The forty genes with the lowest relative 696 fitness in the second growth period of MMC Tn-seq experiment are listed. Each gene was 697 deleted and the deletion mutants were tested for sensitivity to MMC using a spot titer assay and a 698 range of MMC concentrations. Genes labeled as not sensitive had no difference in growth 699 relative to the WT strain on MMC containing media, with the exception of ylbK, which resulted 700 in a polar effect on ylbL (see figure S3). The y-axis is the relative protein levels, which is the indicated protein level normalized to the 737 loading control, DnaN, and the no treatment measurement. 738 Table S1 Tn-seq data collected. OD 600 measurements, incubation times, viable cell counts, 755 growth rates estimated based on viable cell counts, number of generations estimated based on 756 viable cell counts and incubation times, sequencing sample IDs, sequencing reads, reads mapped, 757 and the number of Tn insertions with more than 10 reads for each sample are presented. 758 Table S2 Tn-seq relative fitness lists. The relative fitness values for each gene with sufficient 759 data in all three growth periods for all three Tn-seq experiments are presented along with the 760 adjusted p-value (BH method; see STAR methods). The gene names or locus tags are listed, and 761 intergenic regions are annotated as "ig" with a number. 762