Analysing the fitness cost of antibiotic resistance to identify targets for combination antimicrobials

Aviram Rasouly; Yosef Shamovsky; Vitaly Epshtein; Kayan Tam; Nikita Vasilyev; Zhitai Hao; Giulio Quarta; Bibhusita Pani; Lingting Li; Carmen Vallin; Ilya Shamovsky; Shankarling Krishnamurthy; Aaron Shtilerman; Samantha Vantine; Victor J Torres; Evgeny Nudler

doi:10.1038/s41564-021-00973-1

. Author manuscript; available in PMC: 2022 Oct 1.

Published in final edited form as: Nat Microbiol. 2021 Oct 25;6(11):1410–1423. doi: 10.1038/s41564-021-00973-1

Analysing the fitness cost of antibiotic resistance to identify targets for combination antimicrobials

Aviram Rasouly ^1,^2,^#, Yosef Shamovsky ^1,^#, Vitaly Epshtein ¹, Kayan Tam ³, Nikita Vasilyev ¹, Zhitai Hao ¹, Giulio Quarta ⁴, Bibhusita Pani ¹, Lingting Li ¹, Carmen Vallin ¹, Ilya Shamovsky ¹, Shankarling Krishnamurthy ¹, Aaron Shtilerman ¹, Samantha Vantine ¹, Victor J Torres ³, Evgeny Nudler ^1,^2,^*

PMCID: PMC9389595 NIHMSID: NIHMS1828175 PMID: 34697460

Abstract

Mutations in the rifampicin (Rif)-binding site of RNA polymerase (RNAP) confer antibiotic resistance and often have a global effect on transcription which compromise fitness and stress tolerance of resistant mutants. We hypothesized that the non-essential genome, through its impact on the bacterial transcription cycle, might represent an untapped source of targets for combination antimicrobial therapies. Using transposon sequencing (Tn-seq), we carried out a genome-wide analysis of fitness cost in a clinically common rpoB H526Y mutant. We find that genes whose products enable increased transcription elongation rates compound the fitness costs of resistance whereas genes whose products function in cell wall synthesis and division mitigate it. We validate our findings by showing that the cell wall synthesis and division defects of rpoB H526Y result from an increased transcription elongation rate that is further exacerbated by the activity of the uracil salvage pathway and unresponsiveness of the mutant RNAP to the alarmone ppGpp. We applied our findings to identify drugs that inhibit more readily rpoB H526Y and other Rif^R alleles from the same phenotypic class. Thus, genome-wide analysis of fitness cost of antibiotic resistant mutants should expedite discovery of new combination therapies and delineate cellular pathways that underlie molecular mechanisms of cost.

Rifampicin (Rif), an inhibitor of bacterial RNA polymerase (RNAP), is part of the first-line combination therapy for patients with tuberculosis, leprosy, meningococcal and Legionnaires’ diseases ¹. Rif resistance (Rif^R) is exclusively acquired de novo with more than half of Rif^R clinical isolates carrying either the H526Y or the S531F/L amino acid substitution in the RNAP beta subunit (RpoB) ^2,3. Both residues are highly conserved among bacterial pathogens, as are the majority of other residues that constitute the Rif-binding site located deep within the DNA/RNA channel of the enzyme ^4-6.

In E.coli, phenotypic profiling of a comprehensive panel of Rif^R mutants established a severe fitness cost under a subset of stress conditions in most mutants ^7,8. Importantly, the sensitivity profile of H526Y and S531F is distinct. The apparent physiological differences may emanate from idiosyncratic biochemical properties of each mutant enzyme. The H526Y RNAP is less prone to pausing during transcription elongation and is often referred to as a “fast” RNAP mutant^9-11. In contrast, the S531F substitution emulates the effect of ppGpp on RNAP and is thus considered a “stringent” RNAP ¹².

In previous studies, fitness cost of antibiotic resistance was leveraged to restrict, and even reverse resistance in bacterial populations¹³. Collateral sensitivity studies established that cycling between two or more antibiotics can keep resistance at bay, by virtue of higher sensitivity of antibiotic resistant mutants to antibiotics from a different class^14-16. For example, independent Rif^R lineages of E. faecalis, evolved through serial passages of cells with increasing Rif concentrations, are highly sensitive to a number of cell wall synthesis inhibitors ¹⁵. In addition to collateral sensitivity, a suppressive interaction between two antibiotics enables combination therapy that selects against resistance, and so does induced synergy between pairs of antibiotics^17,18. Of note, none of these elegant approaches aim to discover non-essential cellular factors which affect the fitness cost of antibiotic resistance.

We hypothesized that the cost of rifampicin resistance depends on non-essential genes. For example, the essential core of RNAP interacts with dozens of non-essential transcription factors at any given moment¹⁹. In addition, RNAP responds to signaling molecules and substrate nucleotides, some of which are synthesized by non-essential enzymes as well. Therefore, genome-wide analysis of Rif^R cost would provide a wealth of information pertaining non-essential genes with either a compounding or mitigating effect on cost, and lend itself for mechanistic insights.

In this study, we employed transposon-sequencing (Tn-seq) for genome wide mapping of cost affecting genes in the Rif^R H526Y mutant. We report that cost-compounding genes function in the pathways that further fuel the inherently high transcription elongation rate of H526Y RNAP: uracil salvage, (p)ppGpp degradation, and ribosome biogenesis. Cost-mitigating genes function in cell wall synthesis, cell division, and ppGpp synthesis. Indeed, cell wall synthesis inhibitors and uracil analogs prevent growth of Rif^R mutants from the H526Y class at concentrations that are sub-inhibitory for growth of Rif^S cells. Moreover, we discover that this drug-drug combination, which is antagonistic in wild type cells, is in fact additive towards Rif^R H526Y class mutants. As a proof of concept, we prevent selection in vitro and reduce the tissue load in vivo of Rif^R H526Y mutants. Mechanistically, we demonstrate that de facto ppGpp indifference of the mutant RNAP and superfluous activity of the uracil salvage pathway account for cell wall synthesis and division defects of H526Y Rif^R mutants.

Genome-wide analysis of fitness cost in Rif^R rpoB H526Y

Tn-seq is a powerful tool for analysis of fitness across the genome (Extended Data Fig. 1a-c) ^20,21. With the aim of mapping genes that are either compounding or mitigating fitness cost of Rif^R, we calculated gene fitness values from pools of transposon mutants generated in the wild type and its rpoB H526Y derivative (Fig. 1a-d and Supplementary Table 1). Cost-compounding genes have a higher fitness value in the rpoB H526Y mutant pool than in the wild type pool, and therefore the corresponding transposon mutants have a net fitness-gain effect. Within this category, there are non-essential ribosome biogenesis factors and the uracil salvage enzyme Upp (Fig. 1c,d, and Extended Data Fig. 1e,g). Cost-mitigating genes are depleted from the rpoB H526Y pool with a net fitness-loss effect and are far greater in number. Within this group, there is a significant enrichment for non-essential genes with a function in cell wall synthesis and division (Fig. 1c,d, and Extended Data Fig. 1e,f). In rpoB H526Y, but not in the wild type background, gene knock-outs (KO) of all top hits present a long recovery lag following conjugation with the donor used for transposon delivery (Fig. 1e-f). The longest recovery lag is seen with the KO of yebA, zapA and zapB. Interestingly, under normal growth conditions none of the gene KO have a significant growth defect, neither in the wild type nor rpoB H526Y background (Extended Data Fig.1h). Therefore, the cell envelope injury that recipient cells sustain during conjugation enabled a serendipitous discovery of genetic interaction between rpoB H526Y and all top hits. This insight spurred subsequent testing of genetic interaction between rpoB H526Y and double KO combinations of top hits. We hypothesized that a culminating cell wall synthesis and division defect in a double KO reflects synthetic lethal (SL) interaction with rpoB H526Y. Indeed, whereas the growth rate of yebAzapA, a double KO of the top two hits, is normal (Fig. 1g-h), microscopy imaging demonstrates the elongated cells (Extended Data Fig. 2). Transduction of the rpoB H526Y allele into yebAzapA was successful only when either zapA or yebA are complemented in trans (Extended Data Fig. 3a-b). Without continuous expression of either zapA or yebA, the yebAzapA rpoB H526Y strain (henceforth the SL strain) had a reduced plating efficiency of at least four orders of magnitude (Fig. 1g), and failed to grow in liquid (Fig. 1h) with an ensuing cell division block (Extended Data Fig. 2).

High sensitivity of Rif^R rpoB H526Y to cell-wall synthesis inhibitors and uracil analogs

The synthetic lethality between rpoB H526Y and yebAzapA suggests a differential sensitivity of rpoB H526Y cells to inhibitors of cell wall synthesis and division. Leveraging on previously reported sensitivity of zapA to A22 ^22,23, we investigated the A22 sensitivity of cells harboring the rpoB H526Y mutation, either alone or in combination with zapA (Fig. 2a). A22 inhibits the essential protein MreB, which orchestrates cell shape and growth by mobilizing cell wall synthesis factors from within the cell. It also has a role in the transfer of cell wall synthesis proteins to the division septum ^24
25. While individually H526Y and zapA cells are mildly sensitive to A22, zapA H526Y cells are extremely sensitive (Fig. 2a). Furthermore, a previous study⁸ reported that rpoB H526Y is highly sensitive to the uracil analogs 5-fluoro-uracil (5FU) and 5-fluoro-uridine (5FUd), which we confirmed (Fig. 2b). The anti-bacterial mechanism of action of uracil analogs is complex, and the high level of incorporation of the analogs into DNA, RNA, and peptidoglycan precursors has been observed²⁶. Interestingly, the high level resistance to both analogs²⁶ is mediated by the loss-of-function mutations in the uracil salvage enzyme Upp, one of the top fitness-compounding hits from our Tn-seq analysis. Our analysis, therefore, points to a suboptimal activity of uracil salvage in rpoB H526Y that confers higher sensitivity to uracil analogs. Strikingly, whereas a combination of A22 and 5FUd is antagonistic towards the wild type cells, the two drugs have an additive effect on growth inhibition of rpoB H526Y (Fig. 2c). Because neither A22 nor 5FUd are FDA approved drugs, we tested the effect of the two FDA approved drugs Mecillinam and 5FU as well (Fig. 2d). In contrast to the wild type, where 5FU has a suppressive effect on mecillinam potency, the same drug combination is additive towards rpoB H526Y. The altered interaction type between uracil analogs and cell-wall synthesis inhibitors towards rpoB H526Y renders this combination favorable for inhibiting the growth of rpoB H526Y. Moreover, the combination of A22 and 5FUd with Rif does not interfere with the activity of the latter towards the wild type (Fig. 2E) and is potent enough for the complete growth inhibition of rpoB H526Y (Fig. 2F).

Fig.2∣ — a, Sensitivity of the indicated strains to the cell wall synthesis and division inhibitor A22. Plating efficiency is based on colony forming unit (CFU) values. Mean and standard error are based on three biological replicates. *** p-value<0.001, **** p-value<0.0001. p-values are based on the student t-test (two-sided, equal variance) b, same as in (a) but with the uracil analog 5-fluorouridine (5FUd) c, growth inhibition across two-dimensional A22-5FUd concentration space for wild type and *rpoB* H526Y cells. d, Same as in (c) but with a combination of mecillinam and 5FU. e, The effect of a A22+5FUd combination in the presence of Rif. Growth of indicated strains across a Rif concentration gradient alone, with A22 (1μg/ml) and 5FUd (1μg/ml) separately or in combination.

Specific targeting of rifampicin resistant alleles

Selection for Rif^R mutants typically produces a rich genotypic landscape comprising different point mutants, a high proportion of which is the rpoB H526Y allele (Fig. 3a). The results are dramatically different when Rif^R was selected in yebAzapA cells (Fig. 3b,e), in which fewer Rif^R allele types were selected and the frequency of Rif^R was much lower (Fig. 3d). As expected, rpoB H526Y mutants were undetectable; we have never isolated either rpoB H526Y or S522F, which is another “fast” Rif^R mutant¹⁰, in the yebAzapA genetic background. Other Rif^R alleles, such as the highly abundant D516G, sharply decreased in numbers. The only exception is the increased proportion of the extremely slow growing T563P cells. This mutant belongs to the same class as S531F, where the RNAP mutant emulates the properties of ppGpp-bound RNAP¹². The overall frequency of Rif^R mutants in yebAzapA cells was ~2.3-fold lower than in wild type cells (Fig. 3d). The driving force for this reduced frequency is the selection against “fast” Rif^R mutants in yebAzapA.

Fig.3∣ — **a-c,** Normalized frequency of Rif^R alleles after selection on Rifampicin plates (100 μg /ml) and genotyping mutants from wild type (a), *yebAzapA* (b) or wild type cells on plates additionally supplemented with A22 and 5FUd, each at 1μg/ml (c). Each allele is color-coded based on the corresponding amino acid substitution. The position of amino acid 526 is highlighted in bold. d, Fluctuation analysis of a general Rif^R rate in three different conditions, corresponding to panels a-c. For each condition, n=3 biologically independent replicates of fluctuation assays. Error bars denote 95% confidence intervals around the mean. *p-value <0.05, *** p-value <0.001, ns not significant. p-values are based on one-way ANOVA with a post-hoc Tukey’s multiple comparison **e-f**, Correlation between the normalized frequency of Rif^R alleles in wild type and *yebAzapA* (e) cells, or wild type and wild type cells on plates additionally supplemented with A22 and 5FUd (f). Alleles colored in blue are undetectable in either one of the conditions. g, Two distinct phenotypic classes of Rif^R mutants. h, Murine peritonitis infection model. Bacterial load in the peritoneal lavage and liver in mice 24h after infection with either wild type (triangle shaped symbols) or H526Y cells (circle shaped symbols). For each strain, the two groups of mice (n=10 mice/group) received either no drug (clear symbols) or 20 mg/kg A22 one-hour post infection (filled symbols). The median for each group is indicated with a solid horizontal line. Dotted line indicate limit of detection. * p-value <0.05, ** p-value <0.01, n.s. not significant. p-values are based on one-way ANOVA with Kruskal-Wallis test to correct for multiple comparisons i, Heatmap with MIC values of the indicated strains for A22, mecillinam (MEC), 5FU, 5FUd and seven additional drugs identified by the Biolog panel described in Extended Data Fig. 8: vancomycin (VAN), oxacillin (OXA), cefamandole (CEF), tetracycline (TET), paromomycin (PAR), nitrofurantoin (NIT), and norfloxacin (NOR).

Next, we repeated the same Rif^R selection scheme in wild type cells on plates supplemented with a combination of 5FUd and A22, both at sub-inhibitory concentration (Fig. 3c,f). The genotypic landscape of Rif^R generated under these conditions is similar to that generated in a yebAzapA genetic background with no selection of H526Y or S522F alleles. Moreover, the sharp decrease of the abundant allele D516G under these conditions, together with its sensitivity to A22, 5FUd and high temperature (Extended Data Fig. 4a), unequivocally groups it in the same class as H526Y and S522F (Fig. 3g). Overall, the results show that a drug combination reproduces the effect of a synthetic lethal genetic background.

Because antibiotic resistance is inevitable, directed eradication of pre-existing resistant mutants during infection is an auspicious anti-resistance strategy. Therefore, we measured wild type and rpoB H526Y cells growth in the presence of A22 and human serum containing complement. Serum supplemented with A22, at a concentration sub-inhibitory for both wild type and rpoB H526Y cells in broth, specifically reduced the growth rate of rpoB H526Y cells (Extended Data Fig. 4b). This result implies that differential growth inhibition of rpoB H526Y cells with A22 is tenable in the serum environment, and requires an even lower concentration of the drug.

To investigate whether fitness cost observed in rpoB H526Y cells in the presence of A22 and complement also leads to bacteria clearance in vivo, we established a murine peritonitis infection model to examine the effects of A22 during peritonitis and bacterial dissemination (Fig. 3h and Extended Data Fig. 4c). Mice infected with rpoB H526Y and treated with a single dose of A22 at 20 mg/kg one-hour post-infection had the lowest median load of bacteria in the peritoneal lavage, spleen, and liver 24-hour post-infection. Moreover, we found a significant statistical interaction between A22 treatment and genotype of the infecting strain which accounts for the low load of rpoB H526Y in the liver, and in the pooled data from all three organs (Supplementary Table 3). This analysis suggests that A22 is more potent against rpoB H526Y than against the wild type in vivo.

In order to expand the list of antimicrobials that are highly potent against rpoB H526Y, we screened a commercial panel of 240 small molecules, including 81 antibiotics and antimicrobials (Supplementary Table 4). Out of those, 56 inhibited growth of either wild type or rpoB H526Y (Extended data Fig. 4d). Next, we validated top candidates in a separate set of assays to determine minimal inhibitory concentration values (Fig. 3i and Supplementary Table 5). The data indicate that rpoB H526Y is highly sensitive to a wide range of cell wall synthesis and division inhibitors. As expected, rpoB D516G and rpoB S522F, but not S531F, share much of the sensitivity profile of rpoB H526Y. In genetic background of either zapA or yebA KO, MIC values for both cell wall synthesis inhibitors and uracil analogs are further decreasing.

Sub-optimal ppGpp signaling and uracil salvage underlie the fitness cost of rpoB H526Y

According to our Tn-seq analysis, whereas the ppGpp synthetase RelA is a fitness-mitigating hit, the pppGpp phosphatase Gpp is a fitness-compounding hit (Fig. 1b-c, Extended Data Fig. 1f-g). The data suggest that fitness of rpoB H526Y is positively correlated with concentration of ppGpp at steady state. Interestingly, pivotal cell-division factors, including the A22 target MreB, are no longer essential in strains overexpressing relA ²⁷. Therefore, we tested whether the high concentration of ppGpp suppresses cell wall synthesis and division defects of rpoB H526Y as well. A moderate overexpression of relA does not inhibit growth of rpoB H526Y (Extended Data Fig. 5a-c) and suppresses both the synthetic lethality between rpoB H526Y and yebAzapA (Fig. 4a) and the hypersensitivity of zapA rpoB H526Y to A22 (Fig. 4b).

Surprisingly, a moderate overexpression of relA entirely suppresses the two additional phenotypes of rpoB H526Y: the sensitivity to 5FUd (Fig. 4C) and high temperature (Fig. 4D) ⁸. Likewise, ppGpp overexpression restored stress tolerance to D516G and S522F Rif^R mutants (Extended Data Fig. 5d-e).

Of note, a mild overexpression of ftsQAZ - a known suppressor of cell wall synthesis and division defects - suppresses the sensitivity of rpoB H526Y to A22 (Fig. 4b) and its synthetic lethality with yebAzapA (Fig. 4a). However, it does not suppress the sensitivity of rpoB H526Y to either 5FUd or high temperature, implying that the latter are not the manifestations of rpoB H526Y cell wall synthesis and division defect. Therefore, rpoB H526Y impairs tolerance to distinct types of stress, and high ppGpp appears to restore stress tolerance via a direct effect on mutant RNAP. In other words, the steady state concentration of ppGpp is not high enough to compensate for the defect of rpoB H526Y RNAP. When ppGpp concentration is high, either under conditions of starvation induced by treatment with the amino acid analog Serine hydroxamate (Fig. 4e) or during growth in minimal medium (Extended Data Fig. 6b), rpoB H526Y is once again fully responsive to ppGpp.

Our data further suggest that ppGpp binding to RNAP orchestrates tolerance to all three stresses: A22, 5Fud, and high temperature. If true, it follows that a relAspoT should be sensitive to the same set of stresses. Indeed, relAspoT mutant, which does not make ppGpp, is much more sensitive to A22, 5FUd and high temperature than wild type cells (Fig. 4f and Extended Data Fig. 6a). Moreover, the sensitivity of relAspoT is suppressed by the rpoB S531F mutation, which is known to emulate the properties of ppGpp-bound RNAP (Fig. 4f).

This fitness cost of rpoB H526Y cells is not driven by ppGpp deficiency. First, the sensitivity of the relAspoT rpoB H526Y triple mutant to A22, 5FUd and high temperature is greater than that of either rpoB H526Y or relAspoT mutant (Extended Data Fig. 6a). Second, direct measurements of ppGpp in cell lysates confirm that the level of ppGpp is the same in wild type and rpoB H526Y under normal growth conditions (Extended Data Fig. 6d).

In agreement with previous reports²⁸, our proteomic analysis identified dozens of differentially expressed proteins in relAspoTand rpoB H526Y cells, compared to wild type cells (Extended Data Fig. 6c and Supplementary Table 6). There is a partial, yet statistically significant, overlap between the sets of differentially expressed proteins. However, none of these proteins have a known function in cell wall synthesis and division pathways.

Alternatively, the high transcription elongation rate could account for the fitness cost and reduced stress tolerance of rpoB H526Y cells. If true, the decreasing of the transcription elongation rate in vivo by means other than ppGpp should provide a similar phenotypic suppression. Our Tn-seq analysis, the uracil analogs sensitivity, and the lower levels of de novo pyrimidine biosynthesis enzymes CarAB and PyrBI (Supplementary Table 6) all point to unnecessary high activity of uracil salvage in rpoB H526Y cells.

Indeed, in a genetic background of upp gene knockout (encoding uracil phosphoribosyl transferase), rpoB H526Y regains much of its A22 tolerance (Fig. 4g). The dominance of salvage over de novo pathways in controlling ribonucleotide pools within the cell is also evident from the tolerance to uracil analogs in carAB, which is indistinguishable from that of the wild type strain (Fig. 4h). Collectively, the data strongly support a model where inherently high transcription elongation rates are fueled by sub-optimal basal levels of ppGpp and activity of an uracil salvage enzyme. Consequently, rpoB H526Y is sensitive to cell wall synthesis inhibitors and uracil analogs, alone and in combination (Fig. 4i).

Rifampicin resistance mediated by rpoB H526Y desensitizes elongating RNAP to ppGpp

The effects of H526Y and ppGpp on transcription elongation were mostly studied separately and in vitro²⁹. The combined effect of the two opposing forces on the properties of RNAP in vivo is unknown. For example, it is not clear whether the high concentration of ppGpp can slow the H526Y enzyme globally in vivo.

Using nascent elongating transcript sequencing (NET-seq), we profiled in vivo pausing of RNAP on a genome-wide scale (Fig. 5). As expected, the average pause frequency of rpoB H526Y RNAP is significantly lower (Fig. 5c) than that of wild type RNAP. In particular, rpoB H526Y cells have many more genes that had a low pause frequency compared to wild type cells (Fig. 5a). Notably, the consensus pause sequence is identical in wild type and H526Y cells (Fig. 5d), implying that the majority of pauses in rpoB H526Y cells are at sequences similar to this consensus pause sequence. In relAspoT cells, the average pause frequency of RNAP is also decreased (Fig. 5c), demonstrating that even without stress, basal ppGpp levels have a significant role in regulating transcription elongation. Of note, we found no correlation between pausing of RNAP in relAspoT and H526Y cells at transcription termination sites (Fig. 5g). The overexpression of relA in rpoB H526Y cells partially increases genome-wide average pause frequency (Fig. 5c) and the number of genes with higher pause frequency (Fig. 5b), demonstrating that high ppGpp can indeed reduce the global transcription elongation rate of rpoB H526Y RNAP in vivo.

ppGpp might indirectly affect the elongation rate of RNAP in vivo. Therefore, we purified wild type and H526Y RNAPs to measure the transcription elongation rate with or without ppGpp (Fig. 5e). When ppGpp is added to reactions with wild type RNAP, it induces pausing at multiple sites along the DNA template. In contrast to the potent effect that ppGpp has on elongation by the wild type RNAP, the addition of an equal amount of ppGpp has little effect on the H526Y enzyme. It takes a 10-fold greater concentration of ppGpp to significantly slow the rate of H526Y RNAP (Extended Data Fig. 7). As in living cells, H526Y RNAP is desensitized to ppGpp during elongation in vitro.

Next, we investigated whether the equivalent Rif^R mutant in M. tuberculosis, rpoB H445Y, assumes the same idiosyncratic properties, i.e. faster elongation rate and compromised sensitivity to ppGpp. The H445Y enzyme is clearly faster than the wild type enzyme (Extended Data Fig. 8). The attenuating effect of ppGpp on transcription elongation of M. tuberculosis RNAP is weaker, and requires a higher ppGpp concentration. Notwithstanding its milder effect on elongating wild type RNAP, the H445Y enzyme is evidently desensitized to ppGpp.

Finally, we tested the RNAP pause frequency in the upp and the rpoB H526Y upp double mutant (Extended Data Fig. 9). With no activity of the uracil/cytosine salvage pathway, the intracellular UTP concentration is expected to be lower with a concomitant increase in RNAP pausing (Fig. 4h) ⁹. As expected, the RNAP pause frequency is significantly higher in upp than in the wild type strain. Remarkably, in a genetic background of upp, rpoB H526Y has a pause frequency similar to that of the wild type (Extended Data Fig. 9).

Taken together, our biochemical data demonstrate that a combination of inherently high transcription elongation rate and ppGpp indifference contribute to high transcription elongation rates of rpoB H526Y in both E. coli and M. tuberculosis (Fig. 5f). The superfluous activity of the uracil salvage pathway further fuels the high transcription elongation rates in this Rif^R mutant. The high elongation rate of this mutant underlies its sensitivity to cell wall synthesis inhibitors and uracil analogs (Fig. 4i).

Discussion

The fitness cost of antibiotic resistance serves as the basis for approaches which aim to limit the spread of resistant mutants ^13-16,18. For instance, collateral sensitivity and drug-drug interaction studies present antibiotic combinations that leverage on fitness cost and favor growth of sensitive over resistant bacterial cells. Indeed, the collateral sensitivity to cell wall synthesis inhibitors of some Rif^R mutants has been established in E. faecalis and M. tuberculosis ^15,30. In the current study, we provide genome-wide mapping of fitness cost across the non-essential genome in a model of rifampicin resistance in E. coli. Our discovery of cost-compounding and cost-mitigating genes (Fig. 1) complements previous studies on the cost of rifampicin resistance with new levels of information. First, our data show that the inactivation of cost-mitigating genes potentiates collateral sensitivities of resistant mutants. In the case of rpoB H526Y, we report that non-essential factors ZapA and YebA provide strong protection against antibiotics which target cell wall synthesis and division (Figs. 2 and 3). This means that future inhibitors of non-essential cost-mitigating genes should potentiate the collateral sensitivity of antibiotic resistant mutants. Moreover, the discovery of cost-mitigating genes, in combination with genome-wide antibiotic sensitivity profiles from gene knockout collections^22,31, lends itself for selection of antibiotics most suited to exploit collateral sensitivities. Indeed, the reported sensitivity of zapA and yebA to A22 and mecillinam²², respectively, pointed to high sensitivity of rpoB H526Y to both drugs (Figs. 2 and 3). These findings underscore the value in expanding the collateral sensitivity screening against a much wider collection of small molecules. Second, the discovery of cost-compounding genes and pathways illuminate loci across the non-essential genome where fitness compensatory mutations are expected to evolve. We discovered that manipulation of either ppGpp signaling or uracil salvage restore stress tolerance in rpoB H526Y (Fig. 4). Therefore, it is possible that mutations in genes within these pathways evolve and mitigate fitness cost in clinical settings as well.

Third, our data provide insights pertaining the molecular basis that underlies fitness cost in general and collateral sensitivity in particular. A major conclusion from the current study is that the fitness cost of rpoB H526Y is a consequence of abnormally high transcription elongation rate in vivo (Figs. 4 and 5), and that fitness-compounding genes exacerbate this defect. Specifically, lowering transcription elongation rates, either with high concentration of ppGpp or by shutting down the uracil salvage pathway, restores tolerance to cell wall synthesis inhibitors and uracil analogs in rpoB H526Y (Fig. 4). Interestingly, the antagonistic interaction between cell wall synthesis inhibitors and uracil analogs in wild type cells is additive in rpoB H526Y cells (Fig. 2). It is possible that higher incorporation of analogs to RNA during transcription decreases its incorporation to DNA and cell wall precursors, thereby affecting the sensitivity of rpoB H526Y to the analogs alone and in combination with cell wall synthesis inhibitors. In M. tuberculosis, 5FU indeed incorporates simultaneously to DNA, RNA and cell wall precursors²⁶.

Fourth, we and others¹⁵ demonstrate that collateral sensitivity is not uniform among Rif^R mutants Our work establishes that rpoB H526Y, S522F, and D516G share a similar collateral sensitivity profile. However, the “stringent” RNAP S531F, L533P and T563P make a distinct group that awaits further phenotypic characterization. Collateral sensitivity is therefore genotype, and not antibiotic, specific.

With the advent of genomic medicine, rapid and culture-free genome sequencing of pathogens from clinical samples is now within reach^32,33. Our genome-wide analysis of fitness cost should prove instrumental in tailoring potent drug combinations against antibiotic resistant mutants.

Methods

Selection of rifampicin resistance

Rifampicin resistant mutants were selected following plating of overnight cultures of wild type MG1655 on LB plates supplemented with 100 μg/ml Rifampicin (Sigma, cat. number R3501, from 50 mg/ml stock in DMSO) at 30°C or 37°C. After re-streaking on selection plates and genotyping by Sanger sequencing, mutants were stored at −80°C. The H526Y mutant was independently isolated three times in MG1655, MG1655 rpoC-3X-Flag::Kn and MG1655 rpoC-6X-His. All three isolates are phenotypically indistinguishable.

Transposon sequencing

Random transposon mutagenesis

Delivery of the Hi-Mar transposon by conjugation to either recipient wild type or H526Y cells, readily generates saturated pools of transposon mutants. The donor strain, MFDpir pSC189 ^34,35, is 2,6-Diaminopimelic acid (DAP) auxotroph and, therefore, is grown in the presence of 0.3mM DAP (Sigma cat. number D1377; stock solution is 0.1M in 1N HCl.) ON cultures of donor and recipients were washed once with an equal volume of fresh medium in order to remove incompatible antibiotics. Then 33μl of donor cells and 66μl of recipient cells were combined in a single tube, mixed and spotted on LB plates supplemented with DAP and incubated for 2h at 30°C. Typically, 50-60 reactions are prepared per mutant pool. Cells were scraped from the plates and resuspended in LB without DAP to counter select the donor. Following 1h recovery in LB, cells were diluted and transposon mutants selected on LB plates supplemented with 50 μg/ml Kanamycin without DAP.

Construction of Tn-seq libraries

Genomic DNA was prepared with the Qiagen genomic-tip kit (cat. number 10243). Genomic DNA (1ug) was sheared to 100-200 bp fragments with a Covaris sonic device. Before adapter ligation, DNA was end-repaired, dA-tailed using the NEB Next end repair/dA tailing module (cat. number E7442), and cleaned with the Zymo DNA clean and concentrator kit (cat. number D4013). 50 μM dsDNA adapter was prepared by annealing of 100μM of the single strand oligos: 5’TTCCCTACACGACGCTCTTCCGATCT[Index][phase]T3’, 5’[phase][index]AGATCGGAAGAGCGTCGTGTAGGGAA3’. The index is a library-specific four nucleotide long sequence. Also, for different libraries, adapters were further phased to increase read diversity during Illumina sequencing. The phasing sequence is typically 0-5 nucleotide long. dsDNA adapters with a T-overhang (5μM) were ligated to fragmented A-tailed genomic DNA (up to 1μg) using NEB T4 ligase (cat. number M0202) for 16h at 16°C in a thermocycler. Adapter-ligated DNA was cleaned and amplified in a two-step PCR. The cycling program for the first PCR: 98°C 1min, 10 cycles of 98°C 10sec, 60°C 1min, 72°C 1min and final extension at 72°C 2min. First PCR primers: 5’AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCT3’ and 5’CAAGCAGAAGACGGCATACGAGATTTCTAGAGACCGGGGACTTATCAGCCAAC3’. After PCR cleanup, DNA was further amplified with a second PCR consisting of 15 cycles of the same program and the following primers: 5’ AATGATACGGCGACCACCGAGAT3’ and 5’CAAGCAGAAGACGGCATACGAGAT3’. 125-200 bp PCR products were size selected after separation on 8% NOVEX TBE gels (Invitrogen cat. Number: EC6215BOX) and gel extraction. Sequencing of the 4nM pool of indexed libraries was performed on Illumina MiSeq with a single read setup and 150-nucleotide read length.

Tn-seq data processing and analysis

All data analysis (with the exception of read alignment) was performed with in-house python written codes. All codes are available upon request. Reads were de-multiplexed according to phase and index and then trimmed to remove Hi-Mar transposon sequences. Reads were aligned to the E.coli genome using Bowtie1 ³⁶ with -m 1 -v 0 flags in order to exclude mismatched or non-unique reads. For each read, the transposon insertion site corresponds the TA dinucleotide at the end of the read. For each gene, we defined an IPKM value as the total number of transposon insertions along the gene per number of TA sites in a 1-kilobase sequence per one million reads. In addition, we determined for each gene the saturation of transposon insertion by calculating the number of TA sites hit compared to the total number of TA sites in the sequence of each specific gene. The statistical significance of differential IPKM value of a given gene between the wild type and H526Y mutant pool was determined by computational simulations. Each simulation round consisted of a randomized selection of transposon insertion sites equal in number to the number of TA sites of the tested gene. The randomized differential IPKM was then compared to the actual differential IPKM. The p-value corresponds to the number of simulation rounds, where the simulated differential IPKM was larger than the actual IPKM value, divided by the total simulation number (up to 100,000). Gene Ontology (GO) enrichment analysis was performed using the web-based Gene Ontology pathway analysis tool ³⁷(http://geneontology.org), a detailed description is provided in Supplementary Table 1. Genes included in Supplementary Table 1 have a minimal IPKM value of 280 in either wild type or the rpoB H526Y libraries. The threshold IPKM value was chosen based on the IPKM value distribution presented in Extended Data Fig. 1c. The entire Tn-seq experiment was repeated a second time with a different transposon mutagenesis system. The linear correlation of IPKM values among top genes (Supplementary Table 2) is presented Extended Data Fig. 1d. Also, Supplementary Table 2 presents the list of genes from the second transposon mutagenesis run.

Strain construction and genetic manipulations

For all experiments, wild type E. coli MG1655 was used as the parental strain. Transient expression of the Lambda-red recombineering system from the pKD46 plasmid facilitated the introduction of gene knock-out cassettes to selected chromosomal loci as previously described ³⁸. In order to derive clean gene deletions from gene knockouts, selectable markers of either Kanamycin (plasmid pKD4) or Chloramphenicol (plasmid pKD3) resistance were removed by FLP recombinase expressed from the pCP20 plasmid. P1-phage mediated generalized transduction allowed combination of various double gene knock-outs, as previously described ³⁹. For the NET-seq experiments, the same lambda-red recombineering approach enabled the addition of the nucleotide sequence corresponding to the 3X-Flag epitope to the chromosomal 3’ end of rpoC, thereby yielding the Flag-tagged beta-prime subunit of RNAP. Insertion of the nucleotide sequence corresponding to 6X-Histidine at the 3’ end of rpoC was mediated by the no-SCAR system, where selection is Cas9-dependent and no selectable markers are required, as previously described ⁴⁰. The expression of zapA, relA455 and ftsQAZ in-trans was driven from a low-copy number plasmid with an anhydrotetracycline (aTc) regulated promoter, weak Shine-Dalgarno sequence motif and a carbenicillin resistance selectable marker. This plasmid was constructed using Gibson-assembly ⁴¹, with the vector backbone of pCAS9-CR4, as previously described ^40,41. Induction of in-trans expression of relA455 or zapA from this plasmid set depends on 20ng/μl of aTc. For expression of the ftsQAZ operon, no inducer was added because an increased expression level was toxic.

Drug and stress sensitivity experiments

The susceptibility of different strains to the MreB inhibitor A22, the PBP2 inhibitor Mecillinam and the fluoro-analog 5-fluoro-uridine (5FUd) was assayed by plating 10-fold serial dilutions of overnight cultures on LB-agar plates at 37°C, as follows: A22 (Sigma, cat. Number 475951) stock is 20 mg/ml in DMSO. Sensitivity to A22 on plates was tested in the range of 0.25 μg/ml to 2μg/ml. Mecillinam (Sigma, cat.Number 33447) stock is 10mg/ml in DMSO. Sensitivity to Mecillinam on plates was tested in the range of 0.1-0.5 μg/ml. 5FUd (Sigma, cat.Number F5130) stock is 10 mg/ml in water. Sensitivity to 5FUd on plates was tested in the range of 0.5-20 μg/ml. High-temperature sensitivity was tested by incubating plates at 44.5°C. Amino acid auxotrophy of different strain was tested by re-streaking strains on MOPS-agar plates at 37°C without amino-acid supplement. For ZapA depletion experiments in liquid LB, ON cultures of the yebAzapA H526Y paTc-zapA strain supplemented with 20 ng/μl aTc were diluted 1/4000-fold into fresh LB without aTc and grown at 37°C either in a Bioscreen-C device with automatic readout of OD600 as a proxy for cell-growth, or in 6ml LB in a conventional temperature-controlled shaker at 300 RPM. Typically, complete growth arrest occurred after 6-8 generations of growth. For ZapA depletion experiments in LB-agar plates, ON cultures of the yebAzapA H526Y paTc-zapA strain supplemented with 20 ng/μl aTc were serially diluted and plated on LB agar plates without aTc. No aTc was added in the experiments with paTc-ftsQAZ due to the growth inhibitory effect of strong overexpression of ftsQAZ. Experiments were repeated at least three times. Error bars represent standard error. The student t-test (two-sided, equal variance) was used to calculate p-values.

Validation of top hits from Tn-seq data analysis

Over-night cultures of the MFDpir pSC189 donor and various recipients (gene knock-out of top hits from the Tn-seq analysis in either wild type or Rif^R H526Y background) were washed once with an equal volume of fresh medium in order to remove incompatible antibiotics. Then 33μl of donor and 66μl of recipient were combined in a single tube, mixed and spotted on LB plates supplemented with DAP and incubated for 2h at 30°C. Cells were scraped from the plates and resuspended in LB without DAP to counter select the donor. Following a 30min recovery in LB without DAP, cells were diluted in LB and growth was monitored for 16h, 37°C in a Bioscreen C machine (Growth curves USA). The no-conjugation control growth curves included the same set of strains: ON cultures were diluted 1/1000 and growth was monitored for 16h, 37°C in a Bioscreen C machine.

Phenotypic array plate screening

The full collection of chemical sensitivity array plates (Biolog) includes the following 10 plates: PM11C, PM12B, PM13B, PM14A, PM15B, PM16A, PM17A, PM18C, PM19, and PM20B. At the beginning of each experiment, plates were equilibrated at room temperature (RT), and 80 μl of LB were added in order to dissolve the lyophilized drugs in each well. Over-night cultures of wild type and Rif^R H526Y were diluted 1:200 in fresh LB and again 1:5 into each well (100 μl final), for an overall dilution of 1:1000. Bacterial growth (OD600) in the reconstituted plates was recorded on a Cytation 5 Cell Imaging Multi-Mode Reader (Biotek) at 15min time intervals for 16 hours, 37°C, with continuous double orbital Shaking. The area under the curve (AUC) corresponding to the first 8h of growth was extracted from each growth curve by numeric integration with Simpson’s rule. Each of the 240 drugs in this collection is provided at four different concentrations. We, therefore, removed from subsequent analysis any drug that did not decrease the AUC by at least two-fold in one of its concentrations in comparison to a control grown in LB. There were no drugs for which all four concentrations totally inhibited growth of both wild type and Rif^R H526Y cells. Out of a total of 240 drugs, we defined 81 compounds that are either known antibiotics or are antimicrobials with a clear mode of action towards bacteria. Although we did not pursue further the rest of the compounds, we provide this data as well in Supplementary Table 4. The 81 antimicrobials and antibiotics were stratified by cellular target into 10 groups. These include: 10 Aminoglycosides (AG), 25 Cell wall synthesis and division inhibitors (CW), 8 Quinolones (QUI), 4 genotoxic compounds (GNTX), 10 DNA polymerase starving molecules (REP), 8 Tetracyclines (TETC), 8 Macrolides (MCLD), 2 rifamycins (RIF), 2 Antimicrobial peptides (AMP), and 4 in a miscellaneous group (MICEL). Of the 81 compounds, 56 passed our requirement of growth inhibition by a magnitude of at least 2-fold reduction in AUC, in at least one of the two strains. Finally, the AUC ratio between the two strains, for each concentration of each drug was determined, and data pertaining the drug concentration with the maximal differential in AUC is presented in Extended Data Fig.8 and Supplementary Table 4.

Microscopy

Heat fixed bacteria smears were imaged on a Zeiss Axio Observer with a 60X objective. For cell length analysis, p-values were calculated with the student t-test (two-sided, equal variance).

Fluctuation assays

All fluctuation experiments were performed on more than 150 independent cultures of either wild type or the yebAzapA double mutant. Cultures were grown ON in LB at 37°C instrument with 0.2ml volume each. Individual cultures were then plated on LB agar plates supplemented with Rifampicin (100 μg/ml) and incubated at 37°C for 24h. For the experiments testing the effect of drug combinations on the genotypic landscape of rifampicin resistance, the Rifampicin plates were supplemented with A22 (1 μg/ml) and 5FUd (1 μg/ml) as well. The plating efficiency of both strains was 100% on LB plates without selection. The number of Rifampicin resistant colonies on each plate was determined and a single colony from each plate was used for Sanger sequencing of the Rifampicin resistance region of rpoB using the primers: 5’CGGCAACCGTCGTATCCGTT3’ and 5’ CAACACCGTCGGTCACTTTACGATA3’. The general mutation rate (μ) was calculated using the MSS-maximum likelihood method as previously described ⁴². An ordinary one-way ANOVA with a post-hoc Tukey’s multiple comparison test was performed to determine significance.

Growth in the presence of human complement

The pooled human serum used in this study was obtained from blood drawn from consenting healthy human donors and prepared as described in (PMID: 30602580). Briefly, blood was allowed to clot at room temperature for 15 mins. The coagulated blood was centrifuged at 4000 rpm at 4 °C for 10 mins to obtain serum. The serum was stored in −80 °C until use. Wild type or H526Y were sub-cultured at a 1:1000 dilution from an overnight culture of 3 single colonies in LB +/− 0.75 μg/ml A22 and the indicated w/v of human serum. Growth curves were obtained by measuring O.D. at 15-minute intervals for 12 hours using a Bioscreen-C device.

Human serum

The pooled human serum used in this study was obtained from blood drawn from consenting healthy human donors.

Mouse intraperitoneal infection

8-week-old C57BL/6J females (Jackson Laboratory) were used. All mouse experiments were reviewed and approved by the Institutional Animal Care and Use Committee of New York University Langone Health (NYULH). All experiments were performed according to NIH guidelines, the Animal Welfare Act, and U.S. federal law. Wild type or H526Y (MG1655) cells were sub-cultured at a 1:100 dilution in LB from an overnight culture of 3 single colonies for 3 hours. The bacteria were washed 2 times with PBS and then O.D. normalized to ~1 x 10⁹ CFU/ml in PBS. 8-week-old female C57BL/6J (Jackson Laboratory) were infected intraperitoneally with 3 x 10⁸ CFU. 1-hour post infection, the mice were treated with either PBS + 10% DMSO or 20 mg/kg A22 by intraperitoneal injection. 1-day post infection, the mice were sacrificed by CO₂ inhalation. Peritoneal lavages were performed using 3ml PBS to rinse the peritoneal cavity. Other organs were homogenized in 1 ml PBS. Statistical significances were calculated using one-way ANOVA with Kruskal-Wallis test to correct for multiple comparisons.

Quantification of rRNA synthesis rate using RT-qPCR

Exponential cultures of wild type, H526Y and relAspoT were grown in LB at 37°C to OD600=0.3, samples were then harvested for a no-treatment control and serine hydroxamate (Sigma, cat. number S4503) was added to the remaining cultures at a final concentration of 0.2 mg/ml in order to induce the stringent response. After 10 minutes, samples were collected for total RNA preparation using the RNA protect and RNeasy kit (Qiagen, cat. numbers: 74104, 76506). Reverse transcription of total RNA to cDNA was performed with the ImProm reverse transcriptase system (Promega, cat. number: A3802). As a proxy for the rate of rRNA synthesis, the level of the short-lived 5’ end of the ribosomal RNA poly-cistronic operon was monitored during amplification in a SYBR green-based Real time PCR reaction with the following oligos: 5’ TGACACGGAACAACGGCAAACACG3’ and 5’ TGCATAATACGCCTTCCCGCTACA3’ as previously described ⁴³. Error bars represent the standard error. The statistical significance was calculated with the student t-test (two-sided, equal variance).

MIC assays

Over-night cultures of the indicated strains grown in LB 30C were diluted 10,000-fold and growth in 0.2ml of LB was monitored for 16h, 37°C in a Bioscreen C machine (Growth curves USA). For each drug, growth was tested over a 500-fold concentration range in 2-fold increments. MIC values were defined as the minimal concentration for 85% growth inhibition using the formula: 1 – [AUC drug/AUC control]. Both raw and processed data are given in Supplementary Table 5.

ppGpp quantification in cell lysates

ppGpp extraction was performed as described previously⁴⁴. Briefly, single colonies of E.coli MG1655 and its rpoB H526Y derivative were grown overnight in MOPS minimal media supplemented with 0.4% glucose, 0.15% casamino acids, and 2 mM potassium phosphate (KH₂PO₄). Cultures were diluted 100-fold in MOPS minimal media containing 0.4% glucose, 0.15% casamino acids, and 0.4 mM phosphate. [³²P] H₃PO₄ was added to a final concentration of 150 μCi/ml when OD600 reached ~0.05. When OD600 ~ 0.6, cultures were centrifuged at 10,000 x g for 5 min at 4 °C and the cell pellet was resuspended in 25 μl of 13M formic acid. Metabolites were extracted by freeze-thaw four times on dry ice. Supernatants were spotted onto 20cm by 20 cm, 100-μm polyethyleneimine-cellulose plates (Selecto Scientific) and separated by thin-layer chromatography in 1.5 M KH₂PO₄ (pH 3.4). After chromatography, nucleotides were visualized by autoradiography. The integrated density value function in Adobe Photoshop 2020 was used to quantify the relative ppGpp concentrations. Data from two independent experiments were collected and the relative abundance was calculated relative to the density of the origin.

LC-MS analysis

Proteins extraction, alkylation and digestion

Bacterial pellets collected from 1.5-ml mid-log cultures were resuspended in 90ul of extraction buffer (100 mM ammonium bicarbonate, 10 mM DTT, 2% SDS) and incubated at 95°C for 5 min with vigorous vortexing every 1 min. After cooling at room temperature for 10 min, 10 μl 0.5 M iodoacetamide (freshly dissolved in water) was added and samples were incubated at room temperature in the dark for 30 min. To remove SDS and iodoacetamide, proteins were precipitated with 5 vol of acetone at −20C for 1 hour. Acetone pellets collected by 10-min centrifugation at 16000xg were rinsed with 80% acetone, dried in air and dissolved in 20 μl of denaturing buffer (50 mM ammonium bicarbonate, 8 M urea). For digestion, 5 μl from each sample were mixed with 45μl 50 mM ammonium bicarbonate containing 20 ng/μl trypsin/LysC mixture (Promega) and incubated at 25°C for 18 hours. Digestion reactions were quenched by mixing with 5 μl 20% trifluoroacetic acid. Peptides from quenched reactions were desalted using C18 spin tips (ThermoFisher Scientific) according to manufacturer instructions. Desalted peptides were dried under vacuum and re-dissolved in 10 μl 0.1% formic acid prior to a LC-MS run.

LC-MS

Peptides were analyzed in an Orbitrap Fusion Lumos mass spectrometer (ThermoFisher Scientific) coupled to a Dionex UltiMate 3000 (ThermoFisher Scientific) liquid chromatography system. During each run 1.5-2 ug of peptides were resolved on a 50-cm long EASY-Spray PepMap RSLC C18 column using a 150-min gradient from 96% buffer A (0.1% formic acid in water) to a 40% buffer B (0.1% formic acid in acetonitrile) followed by a 98% buffer B over 5 min with a flow rate of 300 nl/min. A data-dependent acquisition method was performed as described elsewhere ⁴⁵.

Data processing

Raw data was processed using the MaxQuant v.1.6.0.1 software suite ⁴⁶. The protein sequence database included the E. coli MG1655 proteome downloaded from UniProt (https://www.uniprot.org) combined with a list of known protein contaminants. A peptide search engine was used with default parameters with the exception that up to three missed trypsin cleavages were allowed, variable modifications were set to include methionine oxidation and protein N-terminus acetylation, and a fixed modification of cysteine carbamidomethylation. Advanced identification options "Second peptide" and "Match between runs" were enabled. Peptides and proteins quantitation were done using a label-free quantitation method.

Data analysis

Statistical analysis was performed using R with MaxQuant results imported from the "proteinGroups.txt" file. All proteomic data is available in Supplementary Table 5. Statistical significance of the overlap between two groups of proteins was based on Cumulative distribution function (CDF) of the hypergeometric distribution and calculated by a program available at: https://systems.crump.ucla.edu/hypergeometric/

Nascent elongating transcript sequencing (NET-seq) experiments

Growth conditions

Over-night cultures of wild type rpoC-3X-Flag, H526Y rpoC-3X-Flag and H526Y rpoC-3X-Flag carrying the paTc-relA455 were diluted in 150ml of LB to an OD600=0.04 and grown to OD600=0.4. For the H526Y paTc-relA455, aTc was added at 100 ng/ml 25 minutes before harvest in order to induce expression of relA455. Cells were collected on a filter, scraped with a scoopula and flash frozen in N2. A second set of NET-seq libraries was prepared from cultures of the wild type rpoC-3X-Flag, upp rpoC-3X-Flag and rpoB H526Y upp rpoC-3X-Flag cells under the same conditions of growth and with the same methodology for cell pellet collection.

Cell lysis

Lysis of frozen cell pellets was performed for 5 min at room temperature in 0.5ml of the following lysis buffer: 10 mM Hepes pH7.5, 100 mM NaCl, 10 mM MnCl₂, 1mM EDTA, 0.2 mM CaCl₂, 0.1% IGEPAL CA-630, 0.4% Triton X-100, 100 U/ml of RNasIN (Promega, cat. number PR-N2611) and 500 μg/ml Human recombinant lysozyme (Sigma, cat. number: L1667). Then 0.5ml of lysis buffer (without lysozyme) supplemented with 0.25 mg/ml Heparin (Sigma, cat. number: H3393) and 100U/ml DNase I (Pierce, cat. number: 89836) was added and lysates were further incubated for 10 more minutes at room temperature. Shortly before clearing of lysates by centrifugation at 20,000 RPM for 3min, ribosomes and polyribosomes were released from nascent RNA by addition of 25 mM EDTA as previously described ⁴⁷.

Pull down of RNAP-nascent RNA complexes

0.3 ml of anti-3X-Flag affinity gel (Sigma, cat. number: A2220) pre-equilibrated with binding buffer (50 mM Hepes pH7.5 and 150 mM NaCl) was mixed with the cleared lysate at 4°C for 2h. The solution was then loaded on empty poly-Prep Chromatography Columns and allowed to settle. The affinity gel-bound RNAP-nascent RNA complexes were then washed with 80X volumes of wash buffer (10 mM Hepes pH7.5, 300 mM NaCl, 0.1% IGEPAL CA-630, 0.4% Triton X-100, 1mM EDTA, 50U/ml RNasIN). Elution of RNAP-nascent RNA complexes was performed by three successive of 15-minute incubations in 0.25ml of wash buffer supplemented with 0.2mg/ml 3X-Flag peptide.

Nascent RNA purification

Nascent RNA was purified with the TRIzol-LS reagent (Thermo-Fisher, cat. number: 10296010), residual DNA was removed by treatment with 2U of TURBO DNase (Thermo-Fisher, cat. number: AM2238) and 40U RNasIN for 1h at 37°C, and nascent RNA was decontaminated with RNA clean and concentrator (Zymo, cat. number: R1017).

NGS library construction and illumina sequencing

Sequencing libraries were prepared with the NEBNext® Multiplex Small RNA Library Prep Set for Illumina (NEB, cat. number: E7300) with the following modifications. First, after first linker ligation to the 3’ of nascent RNA, samples were decontaminated with the RNA clean and concentrator and then fragmented in an equal volume of 30ul 100 mM sodium carbonate pH9.2 for 25 minutes at 95°C. Samples were next precipitated over-night after addition of 438μl of 10 mM Tris-HCl pH7, 55ul sodium acetate pH5.2, 2ul GlycoBlue (Invitrogen, cat. number: AM9516) and 550ul Isopropanol. Concentrated samples were separated in 8% Novex TBE gels (Thermo-Fisher, cat. number: EC62155BOX) for 45 minutes on 180V and the size range of 50-200 nucleotides was selected for extraction from the gel. Cut gels was crushed by passing samples through gel-breaker tubes (IST engineering, cat. number: 3388-100) after 2 minutes centrifugation at 20,000 RPM. RNA was extracted from the gels by incubation at 70°C for 15 minutes in 0.5ml of 10 mM Tris-HCl pH7, 1 mM ETDA. Gel pieces were removed after filtering samples through Spin-X columns (Fisher, cat. number: 07-200-387) at 20,000 RPM for 3 minutes and size-selected RNA was concentrated by Isopropanol precipitation. Before the second adapter ligation, the 5’ end of nascent RNA was phosphorylated by T4 polynucleotide kinase (NEB, cat. number: M0201) in 1X T4 DNA ligase buffer (NEB, cat. number: M0202) plus 10 mM ATP for 1 hour at 37°C and decontaminated with RNA clean and concentrator. Subsequently, the protocol was identical to the protocol provided with the Library construction kit. Libraries were pooled at 4 nM and sequenced on an Illumina NextSeq sequencer with a NextSeq® 500/550 High Output Kit v2 (75 cycles) cartridge. The first sequenced nucleotide from Rd2 corresponds to the 3’ end of the nascent RNA. Each library had at least 20 million aligned reads.

Data processing and analysis

All data analysis (with the exception of read alignment) was performed with in-house python written codes. All codes are available upon request. Reads were de-mulitplexed according to index and then trimmed to a length of 20 nucleotides. Reads were aligned to the E. coli genome using Bowtie1 with -m 1 -v 0 flags in order to exclude mismatched or non-unique reads. After collecting the strand-specific count signal for each genomic position, RNAP pause sites were called using sliding windows - one hundred nucleotide long if the following two conditions were met. First, if the count signal of the position at the center of each sliding window was the maximal count signal within this window, and second, if the same position at the center of the sliding window, the count signal was five standard deviations above the mean count signal of the sliding window. For each CDS, the mean pause frequency was calculated as the number of pauses divided by the gene length. For the genome-wide aggregated mean pause-frequency, error bars represent standard errors. For RNAP pauses within coding sequences of the top 3500 expressed genes, the consensus pause sequence was determined with WebLogo ⁴⁸. For termination meta-analyses, termination sites were grouped based on termination type, according to previous work ⁴⁹. For each position in the termination window, the Z-score is presented.

E. coli RNAP purification

To purify wild type and H526Y mutant RNAP, pVS10 and pVS10-RpoB-H526Y were introduced into BL21(DE3). Recombinant protein expression was auto-induced for 16 h at 30°C ⁵⁰. Cells containing overexpressed recombinant proteins were harvested by centrifugation and stored as pellets at −80°C.

Cell pellets were resuspended in lysis buffer [50 mM Tris-HCl (pH 8), 5% (v/v) glycerol, 0.5 mM β-ME, 1M NaCl] supplemented with complete, EDTA-free protease inhibitor cocktail tablets (Roche Applied Science) and lysed by sonication. The cell lysate was clarified by centrifugation to remove insoluble debris. The supernatant was applied to a HisTrap column (GE Healthcare) equilibrated in HisTrap Buffer A [50 mM Tris-HCl (pH 8), 5% (v/v) glycerol, 0.5 mM β-ME, 500mM NaCl, 10mM imidazole]. The column was washed with 20 column volumes (cv) of HisTrap buffer A. Proteins bound to the column were eluted in Histrap Buffer B [50 mM Tris-HCl (pH 8), 5% (v/v) glycerol, 0.5 mM β-ME, 250mM NaCl, 250mM imidazole]. Fractions containing recombinant RNAP eluted from the HisTrap column were diluted 5 times in Hep A buffer [50 mM Tris-HCl (pH 8), 5% (v/v) glycerol, 20mM NaCl] and applied to Heparin column (GE Healthcare) equilibrated in Hep A buffer. The bound protein was eluted in a linear gradient from 0.02 to 1.5 M NaCl. RNAP in the peak fractions was then diluted and applied to Mono Q column (GE Healthcare). Recombinant core RNAP was eluted in a linear gradient from 0.15 to 0.5M NaCl. Finally, the protein sample was purified by chromatography on a Superose 6 Increased gel filtration column (GE Healthcare) equilibrated in [40mM Tris-HCl pH=8.0; 10mM MgCl₂; 100mM NaCl, 1mM DTT], and the peak fractions containing RNAP were frozen in liquid nitrogen prior to storage at −80°C.

Recombinant full-length σ70 was expressed as described previously ⁵¹ and purified using a combination of HisTrap, Q and Gel filtration chromatography (GE Healthcare). Purified DksA was a gift from Irina Artsimovitch, Ohio State University.

M. tuberculosis RNAP purification

The plasmids used to overexpress M. tuberculosis σ^A (pET28a-TEV-Mtbσ^A) and M. tuberculosis RNAP core enzyme wild type protein (pACYCDuet-Mtb-rpoA-rpoZ and pETduet-Mtb-rpoB-rpoC) were a gift of Yu Zhang (Institute of Plant Physiology and Ecology, Chinese Academy of Science). The plasmid pETduet-Mtb-rpoB(H445Y)-rpoC was generated through the Q5 site-directed mutagenesis kit (New England Biolabs) with the oligos 5’- GGGGTTGACCTACAAGCGCCGAC -3’ and 5’- GACAGCGGGTTGTTCTGGTCC-3’.

For preparation of M. tuberculosis σ^A, E. coli BL21(DE3) cells carrying pET28a-TEV-Mtbσ^A were cultured in Luria-Bertani broth (LB) at 37 °C, and the expression of N-terminal 6×His-tagged Mtb σ^A was induced at 18°C for 14 h with 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) at OD₆₀₀ of 0.8. Cells were harvested by centrifugation (8000 × g, 4°C), resuspended in lysis buffer (20 mM Tris-HCl (pH 8.0), 0.3 M NaCl, 5% (v/v) glycerol, 0.5 mM β-mercaptoethanol, and protease inhibitor cocktail) and lysed using an Avestin EmulsiFlex-C3 cell disrupter (Avestin, Inc.). The lysate was centrifuged (16,000 × g; 45 min, 4°C) and the supernatant was loaded onto a 5 mL packed HisTrap Fast Flow column (GE Healthcare Life Sciences.) with buffer A (20 mM Tris-HCl (pH 8.0), 0.3 M NaCl, 5% (v/v) glycerol, 0.5 mM β-mercaptoethanol, 5 mM imidazole) and buffer B (20 mM Tris-HCl (pH 8.0), 0.3 M NaCl, 5% (v/v) glycerol, 0.5 mM β-mercaptoethanol, 300 mM imidazole). The eluted fraction was digested by tobacco etch virus protease and dialyzed overnight in dialysis buffer (20 mM Tris-HCl (pH 8.0), 0.2 M NaCl, 1% (v/v) glycerol, and 0.5 mM β-mercaptoethanol). The sample was loaded onto a second 5 mL packed HisTrap Fast Flow column and the cleaved protein was retrieved from the flow through fraction. The sample was diluted to the dialysis buffer with 0.05 M NaCl and further purified through a HiTrap Heparin HP 5 mL column (GE Healthcare Life Sciences) with buffer A (20 mM Tris-HCl (pH 8.0), 0.05 M NaCl, 1% (v/v) glycerol, and 1 mM dithiothreitol (DTT)) and buffer B (20 mM Tris-HCl (pH 8.0), 1 M NaCl, 1% (v/v) glycerol, and 1 mM DTT). Fractions containing M. tuberculosis σ^A was concentrated to 3.5 mg/mL and stored at −80 °C. The M. tuberculosis RNAP core enzyme wild type or βH445Y protein were expressed and purified from E. coli BL21(DE3) carrying pETDuet-Mtb-rpoA-rpoZ and pACYCDuet-Mtb-rpoB-rpoC (or pETduet-Mtb-rpoB(H445Y)-rpoC) as described in ⁵². The protein sample was concentrated to 5 mg/mL and stored at −80 °C.

In vitro transcription reactions

5μl A1 trptT’ (5pmol) DNA were mixed with 1.5μl (~5pmol) of Histidine-tagged wild type RNAP or 1.3μl (~5pmol) Histidine-tagged H526Y RNAP plus 0.25μl (2.5mg) sigma70 and 20μL TB100 [40 mM Tris-HCl pH=8.0; 10 mM MgCl2; 100 mM NaCl]. Samples were incubated for 5 minutes at 37°C. AUC RNA primer was added to 10 mM and ATP and GTP were added, each to 25 mM. Incubation was continued at 37°C for 5 minutes. 2μl of CTP (α-32P) were added for 5 minutes at 22°C. Samples were mixed with 80μl TB100 each and split into two 50μl parts each. One part from each set was mixed with 5μl 1mM ppGpp and the other with 5μl TB100. Samples were incubated for 5 minutes at 22°C. One aliquot from each mix was withdrawn and quenched by 10μl of SB [1XTBE, 20mM EDTA; 8M Urea, 0.025% xylene cyanol, 0.025% bromophenol blue]. The remainder was chased by 1mM NTPs at 22°C. After 10, 20, 40 or 60 seconds, 10μL aliquots were withdrawn and quenched as above. The samples were heated for 5 minutes at 100°C in a dry bath and loaded on a 6% (20x20cm) (19:1) polyacrylamide gel with 7M Urea and TBE pre-run for 7 minutes. The gel was subjected to electrophoresis for 20 minutes at 50W and transferred to Whatman paper, dried at +80°C 45 minutes, and the dried gel was exposed to a phosphor screen overnight.

For reactions with M. tuberculosis RNAP, 2μl A1 trptT' (14pmol) DNA were mixed with 5μl (~15pmol) of M. tuberculosis WT RNAP holoenzyme or 1μl (~9pmol) of M. tuberculosis H445Y RNAP holoenzyme and 20μl TB MT [10mM Tris-HCl pH=8.0, 10mM MgCl₂, 50mM NaCl, 0.1mM EDTA, 0.1mM DTT, 100μg/mL albumin]. Samples were incubated for 5 minutes at 37°C, and then AUC RNA primer (10μM), ATP, and GTP (each at 25μM) were added. Incubation continued at 37°C for 5 minutes. 2μl of CTP-α-P-32 were added for 5 minutes at 22°C. Samples were mixed with 55μl TB MT each, and two 40μl samples were taken from each. Half of the samples were mixed with 4μl ppGpp (10 mM final concentration) and half were mixed with 4μl water. Samples were incubated for 5 minutes at 22°C. One aliquot from each mix was withdrawn and quenched by 10μl of SB [1XTBE, 20mM EDTA, 8M Urea, 0.025% xylene cyanol, 0.025% bromophenol blue]. The remainders were chased by 100μM NTPs at 22°C. After 10, 20, or 40 seconds, 10μL aliquots were withdrawn and quenched as above. The samples were heated for 5 minutes at 100°C dry bath and loaded at 10% (20x20cm) (19:1) polyacrylamide gel with 7M Urea and TBE pre-run for 5 minutes. The gel was run 25 minutes at 50W, exposed to a phosphor screen, and subsequently visualized using ImageQuant software.

Statistical analysis

In the Tn-seq analysis, p-values for fold IPKM value of each gene between wild type and rpoB H526Y mutant pool are based on computational simulations as described in the relevant method section (Fig. 1b, and Supplementary Table 1). In all drug sensitivity assays (Fig.2, Fig.4, Extended Data Fig. 4a, and Extended Data Fig.5d-e), p-values are based on the student’s t-test (two-sided, equal variance). Standard error of the mean (SEM) from triplicate samples is shown as error bars in all figures unless indicated otherwise. In the mice experiments (Fig. 3h and Extended Data Fig. 4c), p-values are based on one-way ANOVA with Kruskal-Wallis test to correct for multiple comparisons. To test the significance of the interaction between genotype and A22 treatment, two-way ANOVA was used (Supplementary Table 3). p-values for mutation rates analysis (Fig. 3d) are based on one-way ANOVA with a post-hoc Tukey’s multiple comparison. The error bars in Fig.3d denote 95% confidence intervals. p-values for RT-qPCR analysis of rRNA leader transcript abundance (Fig. 4e) are based on student’s t-test (two-sided, equal variance). p-values for NETseq analysis of mean pause frequency (Fig. 5c and Extended Data Fig. 9) are based on student’s t-test (two-sided, equal variance). For microscopy analysis of cell length distribution (Extended Data Fig. 2), p-values were calculated with the student t-test (two-sided, equal variance). Statistical significance of the overlap between two groups of proteins (Extended Data Fig. 6c) was based on Cumulative distribution function (CDF) of the hypergeometric distribution and calculated by a program available at: https://systems.crump.ucla.edu/hypergeometric/. No statistical methods were used to pre-determine sample sizes but our sample sizes are similar to those reported in previous publications. No randomization or blinding of samples were applied in the experiments.

Extended Data

Extended Data Fig. 3∣ — a, Transduction of Rif^R *rpoB* H526Y to *yebA zapA* depends on a complementation plasmid expressing *zapA*. b, Poor co-transduction efficiency of Rif^R *rpoB* H526Y cells to multiple double knock-out combinations of top cost mitigating genes from the Tn-seq analysis. Donor P1 phage lysates were prepared either from Rif^R *rpoB* H526Y (with linked kanamycin resistance cassette downstream of the 3’ of *rpoC*) or from Rif^S strain with the same linked selectable marker. The efficiency of transducing the Km cassette to indicated recipients (wild type and selected double knockouts) was calculated by dividing the number of transductants to the number of recipient cells. c, serial dilution and plating of the indiciated strains. Unlike the synthetic lethal interaction between Rif^R *rpoB* H526Y with *yebA zapA* (ΔΔ) , the Rif^R *rpoB* S531F is compatible with the *yebA zapA* double deletion without any growth defect.

Extended Data Fig. 4∣ — Increased sensitivity of Rif^R cells harboring alleles H526Y, D516G, and S522F to A22 (1 μg/ml), 5FUd (5 μg/ml) and high temperature. The S531F is included as an example of a Rif^R allele that does not share the same phenotypic profile. Mean and standard errors of all plating efficiency data are based on colony forming unit values (c.f.u) from three biological triplicates. * p-value <0.05, ** p-value <0.01, *** p-value <0.001. p-values are based on the student t-test (two-sided, equal variance) b, Growth of wild type and *rpoB* H526Y cells in LB supplemented with increasing concentration of human serum, with or without sub-inhibitory concentration of A22. c, Bacterial load in the peritoneal lavage, spleen, and liver in mice 24h after infection with either wild type (triangle shaped symbols) or H526Y cells (circle shaped symbols). For each strain, the two groups of mice (n=10 mice/group) received either no drug (clear symbols) or 20 mg/kg A22 one-hour post infection (filled symbols). The median for each group is indicated with a solid horizontal line. Dotted lines indicate limit of detection. * p-value <0.05, ** p-value <0.01, n.s. not significant. d, Phenotypic profiling of wild type and *rpoB* H526Y cells against a collection of 56 antibiotics and antimicrobials with a known mode of action. Color code is according to the cellular target of each drug. Abbreviations of each class of drugs: Aminoglycosides (AG), Cell wall synthesis and division inhibitors (CW), Quinolones (QUI), genotoxic compounds (GNTX), DNA polymerase starving molecules (REP), Tetracyclines (TETC), Macrolides (MCLD), rifamycins (RIF), Antimicrobial peptides (AMP), and in a miscellaneous group (MICEL). Arrows point to antibiotics that were selected for MIC analysis across the nine different strains that are presented in Fig.3i.

Extended Data Fig. 5∣ — a, Degenerated Shine-Dalgarno (SD) sequence in the “weak” SD vector provides low translation rates in comparison to an optimal “strong” SD sequence. the two vectors are otherwise identical. b, Growth of wt and *rpoB* H526Y cells with the indicated concentration of inducer (aTc) for the expression of *relA* from a “weak”-SD plasmid. c, same as in (b) except that *relA* is expressed from a “strong”-SD plasmid. d, Plating efficiency values from serial dilution of overnight cultures of wt, *rpoB* D516G and *rpoB* D516G carrying prelA (for moderate over-expression of *relA*) on LB agar supplemented with either A22 (1 μg/ml) or 5fUd (15 μg/ml) for 16h at 37C. e, same as in (a) except with S522F and S522F prelA, and that 5FUd concentration is 5μg/ml. For both d and e, Mean and standard errors of all plating efficiency data are based on colony forming unit values (c.f.u) from three biological triplicates (n=3). * p-value <0.05, ** p-value <0.01, *** p-value <0.001. p-values are based on the student t-test (two-sided, equal variance).

Extended Data Fig. 6∣ — a, Sensitivity of the indicated strains to A22, 5FUd and high temperature. b, Robust growth of wild type and *rpoB* H526Y versus auxotrophy of *relAspoT* on minimal agar plates. c, Proteomic analysis of differentially expressed proteins between *rpoB* H526Y and *relAspoT* (ppGpp⁰). Venn diagram presenting the overlap between differentially expressed proteins in *rpoB* H526Y and *relAspoT* cells in comparison to the wild type strain. A full list of genes is given in Supplementary Table 6. The total number of proteins detected across all three strains is 2174. The overlap between differentially regulated proteins in *rpoB* H526Y and *relAspoT* cells, in comparison to wild type cells, has p-values of 1.1x10⁻⁴ and 1.29x10⁻⁸, respectively, for up- and down-regulated proteins. d, ppGpp concentration in lysates prepared from two biological replicates of wild type and *rpoB* H526Y.

Extended Data Fig. 7∣ — a, *In vitro* transcription reactions of *E. coli* Rif^R *rpoB* H526Y RNAP with increasing concentrations of ppGpp (0,0.1 and 1 mM). At the highest ppGpp concentration, reactions were performed with or without DksA. Stalled RNAP elongation complexes (20-mer, denoted with the red arrow) were chased to the end of the template (runoff, denoted with the blue arrow), and the labeled nascent RNA then separated using PAGE. For *in vitro* transcription experiments, n=3 independent replicates. b, Quantification plot of signal from gel scans as a function of distance from the initial site of stalling (right-end on the x-axis). Arrows (labeled 1 through 6) correspond to major pause sites.

Extended Data Fig. 8∣ — a, *In vitro* transcription reactions of *M. tuberculosis* wild type and Rif^R *rpoB* H445Y RNAP with or without ppGpp (10 mM). Stalled RNAP elongation complexes (20-mer, denoted with the red arrow) were chased to the end of the template (runoff, denoted with the blue arrow), and the labeled nascent RNA then separated using PAGE. For *in vitro* transcription experiments, n=3 independent replicates. b, Quantification plot of signal from the gel scans as a function of distance from the initial site of stalling (right-end on the x-axis). Arrows (labeled 1 through 5) correspond to major pause sites.

Extended Data Fig. 9∣ — NET-seq analysis of RNAP pausing in wild type, *upp*, and *upp*H526Y double mutant. For each gene, the mean number of pauses per kilobase gene (the pause frequency) was calculated. The mid-line in each box denotes the aggregated median frequency for the top 3500 expressed genes for each condition. The lower and upper ends of the box denote Q1 and Q3, respectively. The whiskers span 1.5*(Q3-Q1) from each side of the box. **** p-value <0.0001, N.S not significant. p-values are based on the student t-test (two-sided, equal variance). For each condition, n=2 biologically independent replicates of NET-seq experiments.

Extended Data Fig. 10∣ — a, Same image from Fig.5e with b, quantification plots. *In vitro* transcription reactions of wild type and *rpoB* H526Y RNAP with or without ppGpp (100 μM). Stalled RNAP elongation complexes (20-mer, denoted with the red arrow) were chased to the end of the template (runoff, denoted with the blue arrow), and the labeled nascent RNA then separated using PAGE. Numbering corresponds to time points 0, 10, 20, 40 and 60-seconds following the chase reaction. Arrows (labeled 1 through 5) correspond to major pause sites.

Supplementary Material

SupplTable5

NIHMS1828175-supplement-SupplTable5.xlsx^{(165.5KB, xlsx)}

SupplTable4

NIHMS1828175-supplement-SupplTable4.xlsx^{(794.5KB, xlsx)}

SupplTable6

NIHMS1828175-supplement-SupplTable6.xlsx^{(1.2MB, xlsx)}

SupplTables1_2_3_7

NIHMS1828175-supplement-SupplTables1_2_3_7.xlsx^{(310.9KB, xlsx)}

Acknowledgments

This study was supported by NIH grants T32 AI007180 (K.T.), R01 AI099394, R01 AI121244, R01 AI105129 (V.J.T.), R01 GM126891 (E.N), DoD grant PR171734 (E.N.), the Blavatnik Family Foundation, and by the Howard Hughes Medical Institute (E.N.). We thank Leonard Ash from the NYU Metabolomics Core Resource Laboratory for technical assistance.

Footnotes

Competing Interests Statement

The authors declare no competing interests.

Code availability statement

All codes used in this study are available upon request from the corresponding author.

Data Availability Statement

All sequencing data generated in the Tn-seq and NET-seq experiments are publicly available through the sequence read archive (SRA) database of NCBI. Tn-seq data (Fig.1, Extended Data Fig. 1 and Supplementary Table 1) are available through the SRA with the accession numbers: SRX7174171 and SRX7174170. NET-seq data are available through the SRA with the accession numbers: SRX7174164, SRX7174165, SRX7174166, SRX7174167 (for data presented in Fig.5) and SRX11385220, SRX11385221, SRX11385222 (For data presented in Extended Data Fig. 9). The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD027810 (Data presented in Extended Data Fig. 6c and Supplementary Table 6). All strains and plasmids constructed in this study are available upon request from the authors. Raw data for all experiments is provided in the source data of this study.

References

1.Furin J, Cox H & Pai M Tuberculosis. Lancet Lond. Engl (2019) doi: 10.1016/S0140-6736(19)30308-3. [DOI] [PubMed] [Google Scholar]
2.Guinn KM & Rubin EJ Tuberculosis: Just the FAQs. mBio 8, (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Berrada ZL et al. Rifabutin and rifampin resistance levels and associated rpoB mutations in clinical isolates of Mycobacterium tuberculosis complex. Diagn. Microbiol. Infect. Dis 85, 177–181 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Campbell EA et al. Structural mechanism for rifampicin inhibition of bacterial rna polymerase. Cell 104, 901–912 (2001). [DOI] [PubMed] [Google Scholar]
5.Molodtsov V, Scharf NT, Stefan MA, Garcia GA & Murakami KS Structural basis for rifamycin resistance of bacterial RNA polymerase by the three most clinically important RpoB mutations found in Mycobacterium tuberculosis. Mol. Microbiol 103, 1034–1045 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Lin W et al. Structural Basis of Mycobacterium tuberculosis Transcription and Transcription Inhibition. Mol. Cell 66, 169–179.e8 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Jin DJ & Gross CA Mapping and sequencing of mutations in the Escherichia coli rpoB gene that lead to rifampicin resistance. J. Mol. Biol 202, 45–58 (1988). [DOI] [PubMed] [Google Scholar]
8.Jin DJ & Gross CA Characterization of the pleiotropic phenotypes of rifampin-resistant rpoB mutants of Escherichia coli. J. Bacteriol 171, 5229–5231 (1989). [DOI] [PMC free article] [PubMed] [Google Scholar]
9.McDowell JC, Roberts JW, Jin DJ & Gross C Determination of intrinsic transcription termination efficiency by RNA polymerase elongation rate. Science 266, 822–825 (1994). [DOI] [PubMed] [Google Scholar]
10.Jin DJ et al. Effects of rifampicin resistant rpoB mutations on antitermination and interaction with nusA in Escherichia coli. J. Mol. Biol 204, 247–261 (1988). [DOI] [PubMed] [Google Scholar]
11.Svetlov V, Belogurov GA, Shabrova E, Vassylyev DG & Artsimovitch I Allosteric control of the RNA polymerase by the elongation factor RfaH. Nucleic Acids Res. 35, 5694–5705 (2007). [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Zhou YN & Jin DJ The rpoB mutants destabilizing initiation complexes at stringently controlled promoters behave like ‘stringent’ RNA polymerases in Escherichia coli. Proc. Natl. Acad. Sci. U. S. A 95, 2908–2913 (1998). [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Baym M, Stone LK & Kishony R Multidrug evolutionary strategies to reverse antibiotic resistance. Science 351, aad3292 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Imamovic L & Sommer MOA Use of collateral sensitivity networks to design drug cycling protocols that avoid resistance development. Sci. Transl. Med 5, 204ra132 (2013). [DOI] [PubMed] [Google Scholar]
15.Maltas J & Wood KB Pervasive and diverse collateral sensitivity profiles inform optimal strategies to limit antibiotic resistance. PLoS Biol. 17, e3000515 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Lázár V et al. Bacterial evolution of antibiotic hypersensitivity. Mol. Syst. Biol 9, 700 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Chait R, Craney A & Kishony R Antibiotic interactions that select against resistance. Nature 446, 668–671 (2007). [DOI] [PubMed] [Google Scholar]
18.Chevereau G & Bollenbach T Systematic discovery of drug interaction mechanisms. Mol. Syst. Biol 11, 807 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Epshtein V et al. UvrD facilitates DNA repair by pulling RNA polymerase backwards. Nature 505, 372–377 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Goodman AL et al. Identifying genetic determinants needed to establish a human gut symbiont in its habitat. Cell Host Microbe 6, 279–289 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
21.van Opijnen T, Bodi KL & Camilli A Tn-seq: high-throughput parallel sequencing for fitness and genetic interaction studies in microorganisms. Nat. Methods 6, 767–772 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Nichols RJ et al. Phenotypic landscape of a bacterial cell. Cell 144, 143–156 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Baba T et al. Construction of Escherichia coli K-12 in-frame, single-gene knockout mutants: the Keio collection. Mol. Syst. Biol 2, 2006.0008 (2006). [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Bean GJ et al. A22 disrupts the bacterial actin cytoskeleton by directly binding and inducing a low-affinity state in MreB. Biochemistry 48, 4852–4857 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Iwai N, Nagai K & Wachi M Novel S-benzylisothiourea compound that induces spherical cells in Escherichia coli probably by acting on a rod-shape-determining protein(s) other than penicillin-binding protein 2. Biosci. Biotechnol. Biochem 66, 2658–2662 (2002). [DOI] [PubMed] [Google Scholar]
26.Singh V et al. The complex mechanism of antimycobacterial action of 5-fluorouracil. Chem. Biol 22, 63–75 (2015). [DOI] [PubMed] [Google Scholar]
27.Bendezú FO & de Boer PAJ Conditional lethality, division defects, membrane involution, and endocytosis in mre and mrd shape mutants of Escherichia coli. J. Bacteriol 190, 1792–1811 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Aberg A, Fernández-Vázquez J, Cabrer-Panes JD, Sánchez A & Balsalobre C Similar and divergent effects of ppGpp and DksA deficiencies on transcription in Escherichia coli. J. Bacteriol 191, 3226–3236 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Kingston RE, Nierman WC & Chamberlin MJ A direct effect of guanosine tetraphosphate on pausing of Escherichia coli RNA polymerase during RNA chain elongation. J. Biol. Chem 256, 2787–2797 (1981). [PubMed] [Google Scholar]
30.Campodónico VL, Rifat D, Chuang Y-M, Ioerger TR & Karakousis PC Altered Mycobacterium tuberculosis Cell Wall Metabolism and Physiology Associated With RpoB Mutation H526D. Front. Microbiol 9, 494 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Liu A et al. Antibiotic sensitivity profiles determined with an Escherichia coli gene knockout collection: generating an antibiotic bar code. Antimicrob. Agents Chemother 54, 1393–1403 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Břinda K et al. Rapid inference of antibiotic resistance and susceptibility by genomic neighbour typing. Nat. Microbiol 5, 455–464 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Votintseva AA et al. Same-Day Diagnostic and Surveillance Data for Tuberculosis via Whole-Genome Sequencing of Direct Respiratory Samples. J. Clin. Microbiol 55, 1285–1298 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Ferrières L et al. Silent mischief: bacteriophage Mu insertions contaminate products of Escherichia coli random mutagenesis performed using suicidal transposon delivery plasmids mobilized by broad-host-range RP4 conjugative machinery. J. Bacteriol 192, 6418–6427 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Chiang SL & Rubin EJ Construction of a mariner-based transposon for epitope-tagging and genomic targeting. Gene 296, 179–185 (2002). [DOI] [PubMed] [Google Scholar]
36.Langmead B Aligning short sequencing reads with Bowtie. Curr. Protoc. Bioinforma Chapter 11, Unit 11.7 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
37.Ashburner M et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat. Genet 25, 25–29 (2000). [DOI] [PMC free article] [PubMed] [Google Scholar]
38.Datsenko KA & Wanner BL One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc. Natl. Acad. Sci. U. S. A 97, 6640–6645 (2000). [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Moore SD Assembling new Escherichia coli strains by transduction using phage P1. Methods Mol. Biol. Clifton NJ 765, 155–169 (2011). [DOI] [PubMed] [Google Scholar]
40.Reisch CR & Prather KLJ The no-SCAR (Scarless Cas9 Assisted Recombineering) system for genome editing in Escherichia coli. Sci. Rep 5, 15096 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
41.Gibson DG et al. Enzymatic assembly of DNA molecules up to several hundred kilobases. Nat. Methods 6, 343–345 (2009). [DOI] [PubMed] [Google Scholar]
42.Rosche WA & Foster PL Determining mutation rates in bacterial populations. Methods San Diego Calif 20, 4–17 (2000). [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Liu M et al. Global transcriptional programs reveal a carbon source foraging strategy by Escherichia coli. J. Biol. Chem 280, 15921–15927 (2005). [DOI] [PubMed] [Google Scholar]
44.Kamarthapu V et al. ppGpp couples transcription to DNA repair in E. coli. Science 352, 993–996 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Davis S et al. Expanding Proteome Coverage with CHarge Ordered Parallel Ion aNalysis (CHOPIN) Combined with Broad Specificity Proteolysis. J. Proteome Res 16, 1288–1299 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
46.Tyanova S, Temu T & Cox J The MaxQuant computational platform for mass spectrometry-based shotgun proteomics. Nat. Protoc 11, 2301–2319 (2016). [DOI] [PubMed] [Google Scholar]
47.Larson MH et al. A pause sequence enriched at translation start sites drives transcription dynamics in vivo. Science 344, 1042–1047 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Crooks GE, Hon G, Chandonia J-M & Brenner SE WebLogo: a sequence logo generator. Genome Res. 14, 1188–1190 (2004). [DOI] [PMC free article] [PubMed] [Google Scholar]
49.Dar D & Sorek R High-resolution RNA 3’-ends mapping of bacterial Rho-dependent transcripts. Nucleic Acids Res. 46, 6797–6805 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
50.Studier FW Protein production by auto-induction in high density shaking cultures. Protein Expr. Purif 41, 207–234 (2005). [DOI] [PubMed] [Google Scholar]
51.Borukhov S & Goldfarb A Recombinant Escherichia coli RNA polymerase: purification of individually overexpressed subunits and in vitro assembly. Protein Expr. Purif 4, 503–511 (1993). [DOI] [PubMed] [Google Scholar]
52.Banerjee R, Rudra P, Prajapati RK, Sengupta S & Mukhopadhyay J Optimization of recombinant Mycobacterium tuberculosis RNA polymerase expression and purification. Tuberc. Edinb. Scotl 94, 397–404 (2014). [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

SupplTable5

NIHMS1828175-supplement-SupplTable5.xlsx^{(165.5KB, xlsx)}

SupplTable4

NIHMS1828175-supplement-SupplTable4.xlsx^{(794.5KB, xlsx)}

SupplTable6

NIHMS1828175-supplement-SupplTable6.xlsx^{(1.2MB, xlsx)}

SupplTables1_2_3_7

NIHMS1828175-supplement-SupplTables1_2_3_7.xlsx^{(310.9KB, xlsx)}

Data Availability Statement

[R1] 1.Furin J, Cox H & Pai M Tuberculosis. Lancet Lond. Engl (2019) doi: 10.1016/S0140-6736(19)30308-3. [DOI] [PubMed] [Google Scholar]

[R2] 2.Guinn KM & Rubin EJ Tuberculosis: Just the FAQs. mBio 8, (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R3] 3.Berrada ZL et al. Rifabutin and rifampin resistance levels and associated rpoB mutations in clinical isolates of Mycobacterium tuberculosis complex. Diagn. Microbiol. Infect. Dis 85, 177–181 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R4] 4.Campbell EA et al. Structural mechanism for rifampicin inhibition of bacterial rna polymerase. Cell 104, 901–912 (2001). [DOI] [PubMed] [Google Scholar]

[R5] 5.Molodtsov V, Scharf NT, Stefan MA, Garcia GA & Murakami KS Structural basis for rifamycin resistance of bacterial RNA polymerase by the three most clinically important RpoB mutations found in Mycobacterium tuberculosis. Mol. Microbiol 103, 1034–1045 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R6] 6.Lin W et al. Structural Basis of Mycobacterium tuberculosis Transcription and Transcription Inhibition. Mol. Cell 66, 169–179.e8 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R7] 7.Jin DJ & Gross CA Mapping and sequencing of mutations in the Escherichia coli rpoB gene that lead to rifampicin resistance. J. Mol. Biol 202, 45–58 (1988). [DOI] [PubMed] [Google Scholar]

[R8] 8.Jin DJ & Gross CA Characterization of the pleiotropic phenotypes of rifampin-resistant rpoB mutants of Escherichia coli. J. Bacteriol 171, 5229–5231 (1989). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R9] 9.McDowell JC, Roberts JW, Jin DJ & Gross C Determination of intrinsic transcription termination efficiency by RNA polymerase elongation rate. Science 266, 822–825 (1994). [DOI] [PubMed] [Google Scholar]

[R10] 10.Jin DJ et al. Effects of rifampicin resistant rpoB mutations on antitermination and interaction with nusA in Escherichia coli. J. Mol. Biol 204, 247–261 (1988). [DOI] [PubMed] [Google Scholar]

[R11] 11.Svetlov V, Belogurov GA, Shabrova E, Vassylyev DG & Artsimovitch I Allosteric control of the RNA polymerase by the elongation factor RfaH. Nucleic Acids Res. 35, 5694–5705 (2007). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R12] 12.Zhou YN & Jin DJ The rpoB mutants destabilizing initiation complexes at stringently controlled promoters behave like ‘stringent’ RNA polymerases in Escherichia coli. Proc. Natl. Acad. Sci. U. S. A 95, 2908–2913 (1998). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R13] 13.Baym M, Stone LK & Kishony R Multidrug evolutionary strategies to reverse antibiotic resistance. Science 351, aad3292 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R14] 14.Imamovic L & Sommer MOA Use of collateral sensitivity networks to design drug cycling protocols that avoid resistance development. Sci. Transl. Med 5, 204ra132 (2013). [DOI] [PubMed] [Google Scholar]

[R15] 15.Maltas J & Wood KB Pervasive and diverse collateral sensitivity profiles inform optimal strategies to limit antibiotic resistance. PLoS Biol. 17, e3000515 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R16] 16.Lázár V et al. Bacterial evolution of antibiotic hypersensitivity. Mol. Syst. Biol 9, 700 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R17] 17.Chait R, Craney A & Kishony R Antibiotic interactions that select against resistance. Nature 446, 668–671 (2007). [DOI] [PubMed] [Google Scholar]

[R18] 18.Chevereau G & Bollenbach T Systematic discovery of drug interaction mechanisms. Mol. Syst. Biol 11, 807 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R19] 19.Epshtein V et al. UvrD facilitates DNA repair by pulling RNA polymerase backwards. Nature 505, 372–377 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R20] 20.Goodman AL et al. Identifying genetic determinants needed to establish a human gut symbiont in its habitat. Cell Host Microbe 6, 279–289 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R21] 21.van Opijnen T, Bodi KL & Camilli A Tn-seq: high-throughput parallel sequencing for fitness and genetic interaction studies in microorganisms. Nat. Methods 6, 767–772 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R22] 22.Nichols RJ et al. Phenotypic landscape of a bacterial cell. Cell 144, 143–156 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R23] 23.Baba T et al. Construction of Escherichia coli K-12 in-frame, single-gene knockout mutants: the Keio collection. Mol. Syst. Biol 2, 2006.0008 (2006). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R24] 24.Bean GJ et al. A22 disrupts the bacterial actin cytoskeleton by directly binding and inducing a low-affinity state in MreB. Biochemistry 48, 4852–4857 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R25] 25.Iwai N, Nagai K & Wachi M Novel S-benzylisothiourea compound that induces spherical cells in Escherichia coli probably by acting on a rod-shape-determining protein(s) other than penicillin-binding protein 2. Biosci. Biotechnol. Biochem 66, 2658–2662 (2002). [DOI] [PubMed] [Google Scholar]

[R26] 26.Singh V et al. The complex mechanism of antimycobacterial action of 5-fluorouracil. Chem. Biol 22, 63–75 (2015). [DOI] [PubMed] [Google Scholar]

[R27] 27.Bendezú FO & de Boer PAJ Conditional lethality, division defects, membrane involution, and endocytosis in mre and mrd shape mutants of Escherichia coli. J. Bacteriol 190, 1792–1811 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R28] 28.Aberg A, Fernández-Vázquez J, Cabrer-Panes JD, Sánchez A & Balsalobre C Similar and divergent effects of ppGpp and DksA deficiencies on transcription in Escherichia coli. J. Bacteriol 191, 3226–3236 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R29] 29.Kingston RE, Nierman WC & Chamberlin MJ A direct effect of guanosine tetraphosphate on pausing of Escherichia coli RNA polymerase during RNA chain elongation. J. Biol. Chem 256, 2787–2797 (1981). [PubMed] [Google Scholar]

[R30] 30.Campodónico VL, Rifat D, Chuang Y-M, Ioerger TR & Karakousis PC Altered Mycobacterium tuberculosis Cell Wall Metabolism and Physiology Associated With RpoB Mutation H526D. Front. Microbiol 9, 494 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R31] 31.Liu A et al. Antibiotic sensitivity profiles determined with an Escherichia coli gene knockout collection: generating an antibiotic bar code. Antimicrob. Agents Chemother 54, 1393–1403 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R32] 32.Břinda K et al. Rapid inference of antibiotic resistance and susceptibility by genomic neighbour typing. Nat. Microbiol 5, 455–464 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R33] 33.Votintseva AA et al. Same-Day Diagnostic and Surveillance Data for Tuberculosis via Whole-Genome Sequencing of Direct Respiratory Samples. J. Clin. Microbiol 55, 1285–1298 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R34] 34.Ferrières L et al. Silent mischief: bacteriophage Mu insertions contaminate products of Escherichia coli random mutagenesis performed using suicidal transposon delivery plasmids mobilized by broad-host-range RP4 conjugative machinery. J. Bacteriol 192, 6418–6427 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R35] 35.Chiang SL & Rubin EJ Construction of a mariner-based transposon for epitope-tagging and genomic targeting. Gene 296, 179–185 (2002). [DOI] [PubMed] [Google Scholar]

[R36] 36.Langmead B Aligning short sequencing reads with Bowtie. Curr. Protoc. Bioinforma Chapter 11, Unit 11.7 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R37] 37.Ashburner M et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat. Genet 25, 25–29 (2000). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R38] 38.Datsenko KA & Wanner BL One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc. Natl. Acad. Sci. U. S. A 97, 6640–6645 (2000). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R39] 39.Moore SD Assembling new Escherichia coli strains by transduction using phage P1. Methods Mol. Biol. Clifton NJ 765, 155–169 (2011). [DOI] [PubMed] [Google Scholar]

[R40] 40.Reisch CR & Prather KLJ The no-SCAR (Scarless Cas9 Assisted Recombineering) system for genome editing in Escherichia coli. Sci. Rep 5, 15096 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R41] 41.Gibson DG et al. Enzymatic assembly of DNA molecules up to several hundred kilobases. Nat. Methods 6, 343–345 (2009). [DOI] [PubMed] [Google Scholar]

[R42] 42.Rosche WA & Foster PL Determining mutation rates in bacterial populations. Methods San Diego Calif 20, 4–17 (2000). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R43] 43.Liu M et al. Global transcriptional programs reveal a carbon source foraging strategy by Escherichia coli. J. Biol. Chem 280, 15921–15927 (2005). [DOI] [PubMed] [Google Scholar]

[R44] 44.Kamarthapu V et al. ppGpp couples transcription to DNA repair in E. coli. Science 352, 993–996 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R45] 45.Davis S et al. Expanding Proteome Coverage with CHarge Ordered Parallel Ion aNalysis (CHOPIN) Combined with Broad Specificity Proteolysis. J. Proteome Res 16, 1288–1299 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R46] 46.Tyanova S, Temu T & Cox J The MaxQuant computational platform for mass spectrometry-based shotgun proteomics. Nat. Protoc 11, 2301–2319 (2016). [DOI] [PubMed] [Google Scholar]

[R47] 47.Larson MH et al. A pause sequence enriched at translation start sites drives transcription dynamics in vivo. Science 344, 1042–1047 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R48] 48.Crooks GE, Hon G, Chandonia J-M & Brenner SE WebLogo: a sequence logo generator. Genome Res. 14, 1188–1190 (2004). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R49] 49.Dar D & Sorek R High-resolution RNA 3’-ends mapping of bacterial Rho-dependent transcripts. Nucleic Acids Res. 46, 6797–6805 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R50] 50.Studier FW Protein production by auto-induction in high density shaking cultures. Protein Expr. Purif 41, 207–234 (2005). [DOI] [PubMed] [Google Scholar]

[R51] 51.Borukhov S & Goldfarb A Recombinant Escherichia coli RNA polymerase: purification of individually overexpressed subunits and in vitro assembly. Protein Expr. Purif 4, 503–511 (1993). [DOI] [PubMed] [Google Scholar]

[R52] 52.Banerjee R, Rudra P, Prajapati RK, Sengupta S & Mukhopadhyay J Optimization of recombinant Mycobacterium tuberculosis RNA polymerase expression and purification. Tuberc. Edinb. Scotl 94, 397–404 (2014). [DOI] [PubMed] [Google Scholar]

PERMALINK

Analysing the fitness cost of antibiotic resistance to identify targets for combination antimicrobials

Aviram Rasouly

Yosef Shamovsky

Vitaly Epshtein

Kayan Tam

Nikita Vasilyev

Zhitai Hao

Giulio Quarta

Bibhusita Pani

Lingting Li

Carmen Vallin

Ilya Shamovsky

Shankarling Krishnamurthy

Aaron Shtilerman

Samantha Vantine

Victor J Torres

Evgeny Nudler

Abstract

Genome-wide analysis of fitness cost in RifR rpoB H526Y

Fig.1∣. Genome-wide fitness cost analysis of RifR rpoB H526Y in E. coli.

High sensitivity of RifR rpoB H526Y to cell-wall synthesis inhibitors and uracil analogs

Fig.2∣. High sensitivity of RifR rpoB H526Y to cell wall synthesis inhibitors and uracil analogs.

Specific targeting of rifampicin resistant alleles

Fig.3∣. Curtailing selection and undermining survival of RifR mutants via genetic or chemical stress of susceptible pathways.

Sub-optimal ppGpp signaling and uracil salvage underlie the fitness cost of rpoB H526Y

Fig.4∣. Steady-state levels of ppGpp are sufficient to protect wild type, but not RifR H526Y cells, from cell wall synthesis and division defects via binding to RNAP.

Rifampicin resistance mediated by rpoB H526Y desensitizes elongating RNAP to ppGpp

Fig. 5∣. Rifampicin resistance desensitizes elongating RNAP to ppGpp.

Discussion

Methods

Selection of rifampicin resistance

Transposon sequencing

Random transposon mutagenesis

Construction of Tn-seq libraries

Tn-seq data processing and analysis

Strain construction and genetic manipulations

Drug and stress sensitivity experiments

Validation of top hits from Tn-seq data analysis

Phenotypic array plate screening

Microscopy

Fluctuation assays

Growth in the presence of human complement

Human serum

Mouse intraperitoneal infection

Quantification of rRNA synthesis rate using RT-qPCR

MIC assays

ppGpp quantification in cell lysates

LC-MS analysis

Proteins extraction, alkylation and digestion

LC-MS

Data processing

Data analysis

Nascent elongating transcript sequencing (NET-seq) experiments

Growth conditions

Cell lysis

Pull down of RNAP-nascent RNA complexes

Nascent RNA purification

NGS library construction and illumina sequencing

Data processing and analysis

E. coli RNAP purification

M. tuberculosis RNAP purification

In vitro transcription reactions

Statistical analysis

Extended Data

Extended Data Fig. 1∣. Genome-wide fitness cost analysis in RifR rpoB H526Y.

Extended Data Fig. 2∣. Light Microscopy imaging of cell division defect in the synthetic lethal strain.

Extended Data Fig. 3∣. Specific synthetic lethal interaction between RifR rpoB H526Y and yebA zapA (ΔΔ).

Extended Data Fig. 4∣. Specific Targeting of the H526Y phenotypic class of RifR mutants.

Extended Data Fig. 5∣. Rescue of A22 and 5FUd sensitivity of additional RifR mutants with moderate over-expression of relA.

Extended Data Fig. 6∣. Indifference of rpoB H526Y cells to steady-state ppGpp concentration underlies its fitness cost.

Extended Data Fig. 7∣. High concentrations of ppGpp slow-down the “fast” H526Y RNAP in vitro.

Extended Data Fig. 8∣. Desensitization of M. tuberculosis H445Y RNAP to ppGpp in vitro.

Extended Data Fig. 9∣. Mean RNAP pause frequency in the upp genetic background.

Extended Data Fig. 10∣. Rifampicin resistance desensitizes elongating RNAP to ppGpp.

Supplementary Material

Acknowledgments

Footnotes

Data Availability Statement

References

Genome-wide analysis of fitness cost in Rif^R rpoB H526Y

Fig.1∣. Genome-wide fitness cost analysis of Rif^R rpoB H526Y in E. coli.

High sensitivity of Rif^R rpoB H526Y to cell-wall synthesis inhibitors and uracil analogs

Fig.2∣. High sensitivity of Rif^R rpoB H526Y to cell wall synthesis inhibitors and uracil analogs.

Fig.3∣. Curtailing selection and undermining survival of Rif^R mutants via genetic or chemical stress of susceptible pathways.

Fig.4∣. Steady-state levels of ppGpp are sufficient to protect wild type, but not Rif^R H526Y cells, from cell wall synthesis and division defects via binding to RNAP.

Extended Data Fig. 1∣. Genome-wide fitness cost analysis in Rif^R rpoB H526Y.

Extended Data Fig. 3∣. Specific synthetic lethal interaction between Rif^R rpoB H526Y and yebA zapA (ΔΔ).

Extended Data Fig. 4∣. Specific Targeting of the H526Y phenotypic class of Rif^R mutants.

Extended Data Fig. 5∣. Rescue of A22 and 5FUd sensitivity of additional Rif^R mutants with moderate over-expression of relA.