High-resolution landscape of an antibiotic binding site

The study investigates how comprehensive, unbiased mutational analysis of the rifampicin (Rif) binding site in the β subunit of Escherichia coli RNA polymerase (RNAP) reveals functions beyond classical resistance. Although high-resolution structures and resistant mutants have defined the Rif binding site, prior work is constrained by positive selection and covers a small fraction of the possible single-residue variants. The authors hypothesize that saturating, selection-independent mutagenesis can uncover mutations that modulate Rif binding affinity and RNAP function, potentially converting Rif from bacteriostatic to bactericidal, altering transcription speed, and reshaping antibiotic susceptibility and bacterial physiology. The study aims to map drug–enzyme interactions at single-residue resolution, delineate how mutations affect Rif action and RNAP kinetics, and relate laboratory phenotypes to natural rpoB sequence diversity.

Antibiotic resistance mutations in essential enzymes such as RNAP and the ribosome have historically illuminated core biological mechanisms and regulatory features. Structural work established the mechanism of Rif inhibition and the positions in rpoB that mediate resistance. Classical rifampicin-resistant mutants display pleiotropic phenotypes affecting transcription initiation, elongation, termination, and antitermination. However, known resistant alleles sample less than 5% of the non-redundant mutational space of the Rif site. Multiplex automated genome engineering (MAGE) combined with deep sequencing has previously enabled dense mutational scans in essential genes (infA, rpoD), overcoming selection biases and low spontaneous mutation rates. Prior studies also link RNAP elongation speed to Rho-dependent termination efficiency, and show that DNA damage repair mutants are sensitized to certain antibiotics including Rif. These strands of literature motivate an unbiased, saturated analysis of the Rif binding pocket to uncover functional diversity beyond resistance alleles.

• Mutant library construction and fitness screening: Generated a saturated library of 760 single-codon substitutions (20 substitutions × 38 positions) covering rpoB positions 510–537 and 563–572 using MAGE. Post-recombineering, deep sequencing quantified variant abundances (timepoint 0) and following growth (timepoint 1) to compute fitness via log fold change. Stop codon depletion and a bimodal distribution of fitness effects were observed, with an inviable threshold around zero fold-change and approximately 58% substitutions inviable.
• Drug sensitivity screens: The pooled mutants were screened under rifampicin (Rif), bicyclomycin (BCM; Rho inhibitor), and 5-fluorouracil (5FU; nucleotide analogue). Drug-specific effects were isolated by normalizing drug-treated abundances to fitness control (log ratio). Z-scores and positional averages assessed mutability and drug sensitivity/resistance at each site.
• Reconstruction and MIC assays: Selected mutants with strong pooled phenotypes were reconstructed in E. coli MG1655. Minimal inhibitory concentrations (MICs) were measured for Rif, BCM, and 5FU to validate pooled screening results.
• In vitro Rif binding persistence: Purified RNAP containing mutant L521Y (21Y) was tested in a modified runoff transcription assay. Polymerase was saturated with Rif, washed, and allowed to resume transcription; delayed recovery indicated tighter Rif binding. Control 3-mer synthesis rates assessed catalytic competence under Rif block.
• In vivo occupancy after Rif: RNAP ChIP-seq was performed after 1 h Rif exposure and 1 h recovery to test promoter persistence; meta-analysis around transcription start sites compared wild-type to mutant T525D (25D).
• Bactericidal assays and DNA damage: Viability after Rif exposure was assayed by serial dilution plating in LB. DNA fragmentation was measured by TUNEL staining following Rif treatment. SOS response was assessed via expression of SOS genes and LexA protein cleavage. ROS involvement was probed by respiration/ROS assays.
• DSB repair dependence: Deletions of recA or recBCD were introduced into Rif hypersensitive backgrounds to test dependence on DSB repair for Rif killing.
• Localization of DNA breaks: Semi-long-range qPCR (SLR-qPCR) quantified double-strand breaks (DSBs) at promoters versus gene bodies of highly expressed σ70-dependent genes after Rif.
• Replication dependence of killing: Replication was inhibited by dCas9 targeting oriC and by a temperature-sensitive dnaC2 allele (PC2). Effects on Rif-induced DSBs and viability were measured with and without active replication.
• RNAP kinetic profiling: For “fast” mutants (identified by screen correlations), transcription elongation speed was measured using multiprobe qPCR along lacZ following induction; pausing was quantified by NET-seq as mean pause frequency per kilobase across highly expressed genes.
• Nucleotide pool measurements: LC-MS quantified nucleoside triphosphates (notably UTP and CTP) in fast mutants compared to wild type and an upp mutant (uracil salvage deficient). Thymidine supplementation tests assessed rescue of 5FU sensitivity and growth defects.
• Thymineless death and trimethoprim sensitivity: Fast mutants were tested in a thyA background for resistance phase duration upon thymidine withdrawal, and for sensitivity to trimethoprim, which limits dTMP synthesis.
• Temperature-dependent growth: Growth of fast mutants was compared to wild type at 25 °C, 37 °C, and 42 °C; subinhibitory BCM was used to assess whether reduced termination confers low-temperature growth advantages.
• Natural variation analysis: Approximately 4,055 rpoB orthologues from KEGG were aligned. Evolutionary conservation was quantified by Jensen–Shannon divergence. Substitution counts and frequencies within the 38-aa Rif site were tabulated by phylum/class. Natural variant frequencies were compared to fitness and drug-response phenotypes from the screen to highlight impactful natural substitutions.

• Saturated mutagenesis of the rifampicin binding site (760 single-residue mutants across 38 positions) revealed a bimodal fitness landscape with approximately 58% of substitutions inviable and a subset of highly mutable positions, notably the α-helix at positions 523–527.
• Rif resistance landscape: Many substitutions at positions 516, 526, and 572 confer high-level Rif resistance; specific substitutions such as R529K and P564K also cause strong resistance. The spectrum of resistant substitutions is broader than previously appreciated.
• Identification of Rif hypersensitive mutants: Substitutions L521Y (21Y) and T525D (25D), among others at positions 514, 521, 525, render cells Rif hypersensitive without intrinsic fitness defects, decreasing Rif MIC and increasing persistence of RNAP at promoters after Rif exposure (ChIP-seq). In vitro runoff assays show tighter Rif binding for 21Y.
• Conversion of Rif from bacteriostatic to bactericidal: In hypersensitive mutants, Rif reduces viability by at least an order of magnitude, a Rif-specific effect (other bacteriostatic drugs remain bacteriostatic). TUNEL staining increases to ~35% positive cells (WT ~15%) after Rif.
• DNA damage mechanism and replication dependence: Rif induces promoter-specific DSBs (SLR-qPCR) and activates the SOS response (LexA cleavage, DSB-repair gene upregulation). Deleting recA or recBCD greatly increases Rif lethality in hypersensitive backgrounds; WT recA deletion also increases sensitivity. Blocking replication (dCas9 at oriC or dnaC2 non-permissive temperature) reduces DSBs and abolishes Rif bactericidal effects in hypersensitive mutants, implicating transcription–replication conflicts at promoters.
• Fast RNAP mutants: Mutants with high fitness and sensitivity to both 5FU and BCM (e.g., 15G, 15Y, 23P, 23W, 26V, 27W) have increased elongation rates and reduced pausing; conversely, mutants such as 12K, 67C, 67V, 72L are resistant to both 5FU and BCM.
• Nucleotide pool depletion: Fast RNAP mutants are depleted of pyrimidine triphosphates (UTP, CTP) to levels comparable to an upp mutant; 15Y also shows purine depletion. Thymidine supplementation rescues 5FU sensitivity (and a growth defect in 15Y) and fast mutants show shortened resistance phases in thymineless death assays; slow RNAP 513P is 5FU-resistant with prolonged resistance phase. Fast mutants are also more sensitive to trimethoprim.
• Temperature-dependent physiology: Fast RNAP mutants grow faster and reach higher density at 25 °C, show no advantage at 37 °C, and some are temperature-sensitive at 42 °C. Subinhibitory BCM improves WT growth at 25 °C, consistent with poor termination contributing to low-temperature growth advantage.
• Natural diversity: Despite high conservation, functional substitutions are common in nature. S522A is frequent (~10% of surveyed species). Notable natural variants include 524L (fast-like), 526N (fast-like with high Rif resistance and 5FU sensitivity; enriched in Tenericutes), and 567Q (resistance to 5FU and BCM; found in Epsilonproteobacteria/Aquificae), mirroring phenotypes observed experimentally.

Comprehensive, selection-independent mutagenesis of the Rif binding site reveals that this RNAP domain is highly multifunctional: it governs antibiotic binding affinity, influences RNAP kinetics, and shapes global cellular physiology. The identification of an α-helix (positions 523–527) where single substitutions either strengthen or weaken Rif binding provides a fine-grained genetic map of ligand–enzyme contacts, informing strategies to enhance Rif potency. Mutations that strengthen Rif binding convert Rif from bacteriostatic to bactericidal by prolonging RNAP arrest at promoters, which leads to replication-dependent promoter-proximal double-strand breaks and SOS induction. This mechanism is distinct from ROS-mediated killing and clarifies how Rif can cause DNA damage even in wild-type cells without overt lethality. Separately, “fast” RNAP alleles increase elongation rate, reduce pausing, and deplete nucleotide pools, making cells hypersensitive to nucleotide biosynthesis inhibitors (5FU, trimethoprim) and to BCM due to termination defects, while conferring a growth advantage at low temperature via increased readthrough. Mapping natural variation shows that several impactful substitutions exist in environmental and pathogenic bacteria, suggesting evolutionary tuning of RNAP speed and drug sensitivity, and that some naturally occurring substitutions confer resistance with lower fitness costs than commonly observed clinical mutations.

This study delivers a high-resolution functional map of the rifampicin binding site in E. coli RNAP. It expands the Rif resistance landscape, identifies an α-helix critical for Rif interactions, and discovers single-residue mutants that increase Rif affinity sufficiently to render Rif bactericidal through replication-dependent promoter DSBs. It also defines a class of fast RNAP mutants that deplete nucleotide pools, rewire antibiotic susceptibility, and alter optimal growth temperatures. The natural occurrence of functionally similar variants underscores the ecological and evolutionary relevance of these phenotypes. Future work should test whether enhancing Rif affinity can broaden its spectrum and shorten treatments in diverse bacteria, dissect the generality of promoter-centric transcription–replication conflicts in antibiotic action, and explore therapeutic combinations exploiting RNAP speed–nucleotide availability coupling.

• Organismal scope: Most experiments were performed in E. coli MG1655; applicability to other Gram-negative or Gram-positive species, including pathogens, remains to be established.
• Locus coverage: Saturating mutagenesis covered a defined 38-amino-acid window of rpoB; effects of distal RNAP regions and multi-site interactions were not assayed.
• Pool-screen constraints: Detection of strongly slow-growing but highly drug-resistant mutants can be challenging in pooled formats, potentially underestimating certain resistance phenotypes.
• Environmental conditions: Assays were conducted under laboratory growth conditions (LB medium, defined temperatures); physiological responses may differ in host or environmental contexts.
• Mechanistic depth: While data support replication-dependent DSBs at promoters as the basis for Rif bactericidal activity in hypersensitive mutants, detailed molecular intermediates and universality across species require further validation.