Chemical-genetic interrogation of RNA polymerase mutants reveals structure-function relationships and physiological tradeoffs

The multi-subunit bacterial RNA polymerase (RNAP) and its associated regulators carry out transcription and integrate myriad regulatory signals. Numerous studies have interrogated the inner workings of RNAP, and mutations in genes encoding RNAP drive adaptation of Escherichia coli to many health- and industry-relevant environments, yet a paucity of systematic analyses has hampered our understanding of the fitness benefits and trade-offs from altering RNAP function. Here, we conduct a chemical-genetic analysis of a library of RNAP mutants. We discover phenotypes for non-essential insertions, show that clustering mutant phenotypes increases their predictive power for drawing functional inferences, and illuminate a connection between transcription and cell division. Our findings demonstrate that RNAP chemical-genetic interactions provide a general platform for interrogating structure-function relationships in vivo and for identifying physiological trade-offs of mutations, including those relevant for disease and biotechnology. This strategy should have broad utility for illuminating the role of other important protein complexes.

Introduction medium or artificially inducing the stringent response leads to high-level resistance to 234 mecillinam and A22 in E. coli (Bendezú and de Boer, 2008). We therefore explored 235 whether β-P153L resistance arises from a gene expression program locked into a 236 stringent-like state. We measured differential gene expression in β-P153L and its 237 parental strain without and with induction of the stringent response. We achieved 238 induction of the stringent response by expressing a constitutively active allele of RelA 239 (relA*) and compared our results to a published dataset that used the same method  Figure 2C). 258 Thus, despite in vitro behaviors of M + mutants that mimic ppGpp binding (Rutherford et 259 al., 2009;Zhou and Jin, 1998), the steady-state transcriptional program of β-P153L in 260 vivo is largely distinct from the canonical stringent response. 261 β-P153L protects against death caused by loss of rod shape in rich media respectively, two components of the cell wall elongation machinery that directs lateral 264 cell wall insertion and maintains rod shape in E. coli. They are essential during rapid 265 growth (e.g. in LB), but dispensable in nutrient-poor environments (e.g. M9) (Bendezú 266 and de Boer, 2008). As the stringent response is not obviously responsible for 267 resistance in β-P153L, we sought to understand the origin of resistance by determining 268 the full range of resistance responses and morphological changes associated with 269 growth in the antibiotics. 270 We compared liquid growth curves in LB for β-P153L and its parental control over a with mecillinam revealed that β-P153L cocci retained a cell wall (Supplemental Figure   281 3B), rather than forming cell-wall-less spheroplasts. 282 These results predicted that β-P153L should also render the genes encoding PBP2 and 283 MreB (mrdA and mreB) non-essential during rapid growth conditions. We constructed 284 ∆mreB and ∆mreB β-P153L mutants under permissive conditions (minimal medium, 285 30˚C), and tested growth of the double mutant after shifting to non-permissive 286 conditions (LB, 37˚C). The ∆mreB β-P153L double mutant exhibited essentially normal 287 growth, while the ∆mreB control quickly halted growth after the transfer ( Figure 6C). 288 Whole-genome resequencing confirmed that the strains did not contain second-site 289 suppressors (Supplemental Table 4).We conclude that β-P153L renders mreB non-290 essential in rich media by preventing lysis after a loss of rod shape.

M + mutants have growth rate-independent reductions in cell size
It has been proposed that the irreversible step toward death in A22-treated cells is the expansion of cell width beyond a limit at which division can no longer occur, leading to 294 run-away cell width and eventual lysis (Bendezú and de Boer, 2008). According to this 295 model, the small size of β-P153L cells during treatment could keep the mutant below 296 the non-reversible threshold and prevent death. However, the basis for the small size of 297 β-P153L cells was not immediately clear. E. coli and many other rod-shaped bacteria 298 have a log-linear relationship between cell size and growth rate when the nutrient 299 content of the medium is varied (Schaechter et al., 1958;Taheri-Araghi et al., 2015). 300 This relationship, termed the Growth Law, suggested that the smaller size of β-P153L in 301 LB might have been simply due to its lower growth rate.

302
To test this idea, we measured cell size and growth rate across four media with different 303 nutritional contents. If the small size of β-P153L had been due to a growth rate defect 304 alone, the overall relationship between growth rate and cell size would have been 305 indistinguishable between the two strains. Instead, we found that β-P153L was 306 significantly smaller than its parental strain across all growth rates ( Figure 7A). To ask 307 whether our conclusions could be generalized to other M + mutants, we chose 6 308 additional M + mutants and measured the relationship between cell size and growth rate. 309 We found that only a subset of M + mutants had a slow growth phenotype in LB, but all 310 M + mutants had reduced size, with even the most subtle M + mutant exhibiting a 27% 311 reduction in cross-sectional area ( Figure 7B). We conclude that M + mutants exhibit a 312 spectrum of reduced sizes and that their size reduction is not due solely to slower 313 growth.

314
Z-ring stabilization is sufficient for A22 resistance 315 We noticed that the parental strain of β-P153L, BW25113 rpoBC-cat, which differs from 316 the precursor BW25113 by the presence of a chloramphenicol marker between rpoB 317 and rpoC, exhibited a cell-elongation phenotype during log phase in rich media with 318 longer than normal cells that sometimes filamented to >10 µm. We hypothesize that the 319 elongation phenotype is related to the cat insertion as the strain has no other mutations 320 (Supplemental Table 4). Given this observation, we reasoned that the smaller size of 321 β-P153L and the other M + mutants could be related to suppression of filamentation.
Interestingly, filamentation in another rpoB mutant was previously found to be 323 suppressed by the stringent response (Vinella and D'ari 1994). Moreover, filamentation 324 due to the ftsZ ts allele ftsZ84 is suppressed by both induction of the stringent response 325 and an M + mutation in rpoD (Powell, 1998), with increased levels of FtsZ84 in both 326 cases (Powell, 1998). Moreover, mreB essentiality is suppressed by overexpression of 327 the central components of the division machinery FtsQAZ (Bendezú and de Boer, 328 2008). We therefore considered whether increased expression of FtsZ underlays both 329 suppression of filamentation and resistance to A22 and mecillinam in β-P153L.

330
Using single-cell fluorescence microscopy, we quantified the concentration and spatial 331 distribution of an FtsZ-msfGFP translational fusion in β-P153L and its parent. FtsZ 332 concentration was virtually identical (0.1% difference) between the two strains (p=0.9, 333 two sample Kolmogorov-Smirnov test) ( Figure 7C), providing strong evidence that β-334 P153L does not alter FtsZ concentration. In wild-type cells, the fraction of cells with a 335 detectable FtsZ-ring is directly proportional to nutrient-determined growth rate (Weart 336 and Levin, 2003). Surprisingly, β-P153L increased the population of cells with an FtsZ-337 ring ( Figure 7D), despite having a lower growth rate ( Figure 7A) and the same FtsZ 338 concentration as its parent ( Figure 7C). 339 While the filamentous phenotype of the parental rpoBC-cat strain indicates that it has 340 the opposite effect of β-P153L, the effects of the β-P153L mutation are clearly distinct 341 from simply correcting for the division deficiency of the parent. The wild-type BW25113 342 and the parental strain have equivalent MICs for mecillinam and A22 (Supplemental 343 Figure 4A,B). Thus, if β-P153L simply corrected the cell elongation phenotype of its 344 parental strain, it would be sensitive to the antibiotics as well. Instead, we propose that 345 β-P153L stabilizes FtsZ-rings independent of FtsZ expression, and that this phenotype 346 suppresses filamentation during normal growth and allows for continued division under 347 the action of A22 and mecillinam.

348
Since the other 6 M + mutants also exhibited a range of lengths shorter than the parental 349 strain, we sought to test whether cell length was a proxy for FtsZ-ring stabilization that 350 gives rise to A22 resistance. We grew all seven M + mutants in LB with above-MIC 351 concentrations of A22 and mecillinam and compared the normalized maximum OD600 in the presence of the drugs to cell size and growth rate parameters. The normalized 353 ODmax in the presence of A22 was strongly negatively correlated with cell length 354 (R 2 =0.74, p=0.006), but not with cell width (R 2 =0.14,p=0.36) or growth rate (R 2 =0.17, 355 p=0.31) (Figure 7E, Supplemental Figure 4C,D). 356 Interestingly, resistance to mecillinam had a different origin. The normalized ODmax in 357 the presence of mecillinam was correlated with growth rate (R 2 =0.53, p=0.04) but not 358 cell size (R 2 =0.16, p=0.32). We conclude that FtsZ-ring stabilization is not sufficient to 359 confer resistance to mecillinam and that another aspect of regulation by the stringent 360 response, likely a correlate of the growth rate phenotype, is necessary. These results 361 are consistent with previous findings that mild FtsZ overexpression is sufficient for 362 survival in A22 but not mecillinam (Cho et al., 2014). We conclude that FtsZ-ring 363 stabilization rather than FtsZ upregulation determines cell length and A22 sensitivity in 364 M + mutants, and that a reduction in growth rate is neither sufficient nor necessary.

Discussion
As the enzyme responsible for bacterial transcription and the integrator of transcriptional 367 control, RNAP has been the focus of an enormous amount of research. In addition to 368 structural, biochemical, and evolutionary analyses, multiple studies have utilized RNAP-369 centric genetic approaches, including early work on resistance to RNAP-targeting drugs 370 such as rifampicin (Jin and Gross, 1988) and streptolydigin (Heisler et al., 1993),  Here, we show that a bottom-up approach based on unbiased, expansive screening and 381 clustering of the phenotypes of large numbers of RNAP mutations is a powerful tool for 382 functional discovery, illuminating structure-function relationships of RNAP at the single-383 residue level and systems-level connections between transcription and other cellular 384 processes. That our strategy was successful even though our library is overrepresented 385 in Rif R and M + mutants underscores the fact that mutations isolated under the same 386 selective pressure can have distinct, pleiotropic phenotypes. Importantly, our findings 387 also highlight the patterns behind these pleiotropic fitness trade-offs, hinting at the 388 possibility of general rules that govern the physiological effects of RNAP mutations. We 389 probed one such chemical interaction to discover an effect of M + mutants on cell-size 390 control by modulating FtsZ-ring stability. Our single-cell analyses revealed that M + mutants are shorter than the parental strain ( Figure 7A,B) and that β-P153L cells remain as small cocci during A22 or mecillinam 426 treatment ( Figure 6B), prompting the hypothesis that M + mutants have higher 427 expression of FtsZ. However, FtsZ concentration was unaffected (Figure 7C), 428 consistent with previous findings that neither the direct σ 70 promoters for ftsZ (Navarro  We thank Seth Darst for advice on the design of lineage-specific insertion/deletion 462 mutations. We thank Ann Hochschild for sharing mutants in σ 70 that were used in the 463 screen. We thank Wilma Ross and Rick Gourse for sharing the pALS13 and pALS14 464 plasmids. We thank Curtis Ross for his contributions to the data browser. The raw images and Iris data files for the chemical genetic screen along with the two 471 datasets generated in this work have been submitted for publication on Dryad. All oligonucleotides used in this study are listed in Supplemental Table 5. All plasmids used in this study are listed in Supplemental Table 5.

500
Mutagenesis using oligonucleotide recombineering 501 We generated some mutations reported in this study de novo using oligonucleotide  Screening for attenuation mutants 550 We screened for attenuation-enhancing mutations in a BW25113 rpoBC-cat ∆trpR::kan  , 1994;Yanofsky and Horn, 1981). Single colonies were isolated by streaking from 559 the non-selective patch, and the mutation was confirmed with Sanger sequencing.

560
Finally, the mutant was transduced into BW25113 using the genetically linked cat gene. Design and assembly of the 1536-colony array 606 We split biological replicates for each transcription mutation or gene deletion into two 607 sets (Array #1 and Array #2). We then arrayed the mutations within each set in triplicate  In the first step, we hierarchically clustered a randomized copy of the dataset. We then 676 calculated the smallest cophenetic distance in the randomized dendrogram. After 677 repeating these steps 30,000 times, we used the 5 th percentile of the distribution of 678 cophenetic distances as a cutoff to define significant clusters in the original dataset. The 679 cutoff represents the cophenetic distance that is closer than the closest distance in 95%

711
Options for GSEA included running the analysis for pre-ranked lists and the flags "-norm 712 meandiv", "-scoring_scheme weighted","-create_svgs false","-make_sets true", "-713 plot_top_x 20", "-rnd_seed 081889","-set_max 500","-set_min 3","-zip_report true", and 714 "-gui false". The output for each condition was then collected into a single and an orbital frequency of 237 cycles per minute. of each strain with no drug as a reference point. We extracted the maximum growth rate 754 from the curve, computed the time at which the growth rate first dropped below 10% of 755 this value, and added two hours to define the time t2 that determines the upper limit of the area to be measured for every drug concentration for that strain. This calculation sets the upper limit for the area of integration to two hours into the transition phase of 758 the strain when grown without stressors. The lower limit of the area to be integrated was 759 the initial time t1 that measurements started. The area integrated was the blanked OD600 760 between times t1 and t2. The OD600 was not log-transformed for this calculation. We then 761 normalized the mean AUC for every combination of strain and drug concentration to the 762 no-drug control. The no-drug control was included in every plate and every 763 measurement was compared to the control on the same plate. P1vir transduction and selected for the linked rpoBC-cat antibiotic resistance cassette.

770
The sequences of all transductants were verified with Sanger sequencing.

Whole genome resequencing of ΔmreB strains
The raw reads from whole genome resequencing were used as input into breseq After growing all cultures into log phase, each culture was split into two experiments. In 898 the first, culture densities were normalized to an OD600 of 0.1, then used to inoculate the 899 same media in a 96-well plate at a final volume of 200 µL and an initial inoculum with an 900 OD600 of 0.01. Growth curves were measured as described above, and maximum 901 growth rates were computationally extracted from the growth curves. For the data in 902 Figure 7E and Supplemental Figure 4C,D, in addition to growing the M + mutants in 903 LB, we also generated liquid growth curves in LB with 13.5 µg/mL of A22 or mecillinam.

904
The maximum OD600 was computationally extracted from the growth curves in the 905 presence of drug and normalized against the maximum OD600 of the same strain in LB 906 without antibiotic.
In the second experiment, cultures were directly spotted onto a phosphate buffer saline 909 1% (w/v) agarose pad and phase contrast images were acquired using a Ti-E microscope (Nikon) with a 100X (NA: 1.4) objective and a Zyla 5.5 sCMOS camera (Andor). Phase-contrast images were segmented and meshed using Morphometrics 912 (Ursell et al., 2017) and shape parameters were computationally extracted from the 913 mesh.           Figure 6A and are replotted here for comparison to BW25113. The 1403 parental strain halts growth earlier than BW25113 when challenged with A22, but 1404 the concentration at which A22 becomes lethal is identical between the two 1405 within the resolution of this experiment. By contrast to both wild-type strains, β-1406 P153L was highly resistant to A22.    Condition: the name of the stressor, NES: the normalized enrichment score as output