Directed evolution of the rRNA methylating enzyme Cfr reveals molecular basis of antibiotic resistance

  1. Kaitlyn Tsai
  2. Vanja Stojković
  3. Lianet Noda-Garcia
  4. Iris D Young
  5. Alexander G Myasnikov
  6. Jordan Kleinman
  7. Ali Palla
  8. Stephen N Floor
  9. Adam Frost
  10. James S Fraser
  11. Dan S Tawfik
  12. Danica Galonić Fujimori  Is a corresponding author
  1. Department of Cellular and Molecular Pharmacology, University of California San Francisco, United States
  2. Department of Biomolecular Sciences, Weizmann Institute of Science, Israel
  3. Department of Bioengineering and Therapeutic Sciences, University of California San Francisco, United States
  4. Department of Biochemistry and Biophysics, University of California San Francisco, United States
  5. Helen Diller Family Comprehensive Cancer Center, University of California San Francisco, United States
  6. Department of Cell and Tissue Biology, University of California San Francisco, United States
  7. Quantitative Biosciences Institute, University of California San Francisco, United States
  8. Department of Pharmaceutical Chemistry, University of California San Francisco, United States

Abstract

Alteration of antibiotic binding sites through modification of ribosomal RNA (rRNA) is a common form of resistance to ribosome-targeting antibiotics. The rRNA-modifying enzyme Cfr methylates an adenosine nucleotide within the peptidyl transferase center, resulting in the C-8 methylation of A2503 (m8A2503). Acquisition of cfr results in resistance to eight classes of ribosome-targeting antibiotics. Despite the prevalence of this resistance mechanism, it is poorly understood whether and how bacteria modulate Cfr methylation to adapt to antibiotic pressure. Moreover, direct evidence for how m8A2503 alters antibiotic binding sites within the ribosome is lacking. In this study, we performed directed evolution of Cfr under antibiotic selection to generate Cfr variants that confer increased resistance by enhancing methylation of A2503 in cells. Increased rRNA methylation is achieved by improved expression and stability of Cfr through transcriptional and post-transcriptional mechanisms, which may be exploited by pathogens under antibiotic stress as suggested by natural isolates. Using a variant that achieves near-stoichiometric methylation of rRNA, we determined a 2.2 Å cryo-electron microscopy structure of the Cfr-modified ribosome. Our structure reveals the molecular basis for broad resistance to antibiotics and will inform the design of new antibiotics that overcome resistance mediated by Cfr.

Editor's evaluation

The paper addresses an important unresolved mechanism of antibiotic resistance caused by a RNA-modifying enzyme Cfr, a protein that confers resistance to multiple ribosome-targeting antibiotics due to methylation of the rRNA residue A2503 on the large ribosomal subunit. Tsai et al. identify the mutational hotspots that increase Cfr activity and show how a single methyl group on A2503 can hamper the antibiotic binding. The paper is an important contribution to our understanding of antibiotic resistance and is of broad interest to readers in the field of antibiotic resistance, biochemistry, structural biology, and medicinal chemistry.

https://doi.org/10.7554/eLife.70017.sa0

eLife digest

Antibiotics treat or prevent infections by killing bacteria or slowing down their growth. A large proportion of these drugs do this by disrupting an essential piece of cellular machinery called the ribosome which the bacteria need to make proteins. However, over the course of the treatment, some bacteria may gain genetic alterations that allow them to resist the effects of the antibiotic.

Antibiotic resistance is a major threat to global health, and understanding how it emerges and spreads is an important area of research. Recent studies have discovered populations of resistant bacteria carrying a gene for a protein named chloramphenicol-florfenicol resistance, or Cfr for short. Cfr inserts a small modification in to the ribosome that prevents antibiotics from inhibiting the production of proteins, making them ineffective against the infection. To date, Cfr has been found to cause resistance to eight different classes of antibiotics. Identifying which mutations enhance its activity and protect bacteria is vital for designing strategies that fight antibiotic resistance.

To investigate how the gene for Cfr could mutate and make bacteria more resistant, Tsai et al. performed a laboratory technique called directed evolution, a cyclic process which mimics natural selection. Genetic changes were randomly introduced in the gene for the Cfr protein and bacteria carrying these mutations were treated with tiamulin, an antibiotic rendered ineffective by the modification Cfr introduces into the ribosome. Bacteria that survived were then selected and had more mutations inserted. By repeating this process several times, Tsai et al. identified ‘super’ variants of the Cfr protein that lead to greater resistance.

The experiments showed that these variants boosted resistance by increasing the proportion of ribosomes that contained the protective modification. This process was facilitated by mutations that enabled higher levels of Cfr protein to accumulate in the cell. In addition, the current study allowed, for the first time, direct visualization of how the Cfr modification disrupts the effect antibiotics have on the ribosome.

These findings will make it easier for clinics to look out for bacteria that carry these ‘super’ resistant mutations. They could also help researchers design a new generation of antibiotics that can overcome resistance caused by the Cfr protein.

Introduction

A large portion of clinically relevant antibiotics halt bacterial growth by binding to the ribosome and inhibiting protein synthesis (Tenson and Mankin, 2006; Wilson, 2009; Arenz and Wilson, 2016). Since antibiotic binding sites are primarily composed of ribosomal RNA (rRNA), rRNA-modifying enzymes that alter antibiotic binding pockets are central to evolved resistance (Vester and Long, 2013; Wilson, 2014). The rRNA-methylating enzyme Cfr modifies an adenosine nucleotide located within the peptidyl transferase center (PTC), a region of the ribosome essential for catalyzing peptide bond formation and consequently, a common target for antibiotics (Schwarz et al., 2000; Kehrenberg et al., 2005). Cfr is a radical SAM enzyme that methylates the C8 carbon of adenosine at position 2,503 (m8A2503, Escherichia coli numbering) (Jensen et al., 2009; Kaminska et al., 2010; Yan et al., 2010; Yan and Fujimori, 2011; Grove et al., 2011). Due to the proximal location of A2503 to many antibiotic binding sites, introduction of a single methyl group is sufficient to cause resistance to eight classes of antibiotics simultaneously: phenicols, lincosamides, oxazolidinones, pleuromutilins, streptogramin A (PhLOPSA), in addition to nucleoside analog A201A, hygromycin A, and 16-membered macrolides (Long et al., 2006; Smith and Mankin, 2008; Polikanov et al., 2015). Among rRNA modifying enzymes, this extensive cross-resistance phenotype is unique to Cfr and presents a major clinical problem.

Cfr emergence in human pathogens appears to be a recent event, with the first case reported in 2007 from a patient-derived Staphylococcus aureus isolate (Toh et al., 2007; Arias et al., 2008). Since then, the cfr gene has been identified across the globe in both gram-positive and gram-negative bacteria, including E. coli (Shen et al., 2013; Vester, 2018) and has been associated with several clinical resistance outbreaks to the oxazolidinone antibiotic, linezolid (Morales et al., 2010; Locke et al., 2010; Bonilla et al., 2010; Cai et al., 2015; Layer et al., 2018; Lazaris et al., 2017; Dortet et al., 2018; Weßels et al., 2018). The vast spread of Cfr is attributed to its association with mobile genetic elements and relatively low impact on bacterial fitness, suggesting that cfr can be rapidly disseminated within bacterial populations (LaMarre et al., 2011; Schwarz et al., 2016).

Due to the ability of Cfr to confer resistance to several antibiotics simultaneously, it is critical to understand how bacteria may adapt under antibiotic pressure to enhance Cfr activity and bolster protection against ribosome-targeting molecules. Identification of Cfr mutations that improve resistance will also be critical for informing clinical surveillance and designing strategies to counteract resistance. A major limitation in our current understanding of Cfr-mediated resistance is the lack of structural insight into changes in the ribosome as a result of Cfr modification. Steric occlusion of antibiotic binding has been proposed as a model to rationalize altered antibiotic susceptibility (Polikanov et al., 2015). Additionally, the observation that A2503 can adopt both syn and anti-conformations in previously reported ribosome structures suggests that methylation may regulate conformation of the base, as previously proposed (Toh et al., 2008; Schlünzen et al., 2001; Tu et al., 2005; Stojković et al., 2020). However, direct evidence for how m8A2053 alters antibiotic binding sites to inform the design of next-generation molecules that can overcome Cfr resistance is lacking.

In this study, we identified mechanisms that enhance antibiotic resistance by performing directed evolution of a cfr found in a clinical MRSA isolate under antibiotic selection (Barlow and Hall, 2003). The obtained highly resistant Cfr variants show increased rRNA methylation, driven primarily by robust improvements in Cfr cellular levels, achieved either by higher transcription or increased translation and improved cellular stability. In particular, mutation of the second Cfr amino acid to lysine strongly enhances translation and resistance. Finally, we used an evolved variant which achieves near-stoichiometric rRNA methylation to generate a high-resolution cryo-electron microscopy (EM) structure of the Cfr-modified E. coli ribosome. The obtained structural insights provide a rationale for how m8A2503 causes resistance to ribosome antibiotics.

Results

Evolved Cfr variants confer enhanced antibiotic resistance

To perform directed evolution of Cfr, we used error-prone PCR (EP-PCR) to randomly introduce 1–3 mutations into the cfr gene obtained from a clinical MRSA isolate (Toh et al., 2007), herein referred to as CfrWT (Figure 1a). Mutagenized cfr sequences were then cloned into a pZA vector where Cfr was expressed under tetracycline-inducible promoter Ptet introduced to enable precise control of Cfr expression (Wellner et al., 2013). The resulting library of ~107 E. coli transformants was selected for growth in the presence of increasing amounts of tiamulin, a pleuromutilin antibiotic to which Cfr confers resistance. During each round, a subset of the surviving colonies was sequenced to identify new mutations. After two rounds of evolution, wild-type Cfr was no longer detected, indicating that the introduced mutations provide enhanced survivability in the presence of tiamulin. After five rounds of mutation and selection, we performed two rounds of selection without mutagenesis, and with high tiamulin concentrations, thus leading to fixation of mutations that provide robust resistance.

Figure 1 with 2 supplements see all
Evolved variants of Cfr exhibit improved resistance to PhLOPSA ribosome antibiotics.

(a) Evolution of Cfr under selection by the PTC-targeting antibiotic tiamulin. (b) Cfr homology model based on RlmN generated by I-TASSER server (Yang and Zhang, 2015) with mutagenic hotspots in red. N-terminal domain (NTD) is labeled. Active site denoted by S-adenosylmethionine (SAM, gray) and [4Fe-4S] cluster (orange). (c) Evolved variants containing Cfr mutations were selected for further study. Ptet* indicates alterations to promoter sequence. (d) Promoter architecture of CfrV6 and CfrV7 where pPtet designates a partial Ptet promoter sequence and Ins designates a variable insertion sequence. (e) Fold improvement in MIC resistance value for PhLOPSA antibiotics and trimethoprim compared to empty pZA vector control determined from three biological replicates by microbroth dilution method. Trimethoprim is a negative control antibiotic that does not target the ribosome. LZD testing was performed against Escherichia coli BW25113 lacking efflux pump, acrB. Numerical MIC values are displayed in Figure 1—source data 1.

Analysis of surviving cfr sequences from the final rounds of selection revealed notable trends (Supplementary file 1). Three positions were primarily mutated: N2, I26, and S39. By homology modeling, these mutational hotspots appear distal from the enzyme active site (>12 Å; Figure 1b). In fact, these mutations reside in what has been predicted to be an N-terminal accessory domain separate from the radical-SAM catalytic domain (Kaminska et al., 2010). Second, ~28% of sequences contained alterations to the promoter. These alterations consist of either Ptet duplication, or insertion of a partial Ptet sequence (Supplementary file 1).

We selected seven evolved Cfr variants, referred herein as CfrV1–V7, as representative mutational combinations for further characterization (Figure 1c). All selected Cfr variants contain mutations in the cfr open reading frame (ORF) while CfrV6 and CfrV7 also harbor Ptet alterations (Figure 1d). Compared to CfrWT, these variants confer ~2-fold to ~16-fold enhanced resistance to PhLOPSA antibiotics and hygromycin A, with no changes in susceptibility to trimethoprim, an antibiotic that does not inhibit the ribosome (Figure 1e, Figure 1—figure supplement 1). Interestingly, the promoter alterations enable CfrV7 to be expressed and confer resistance to tiamulin in the absence of inducer (Figure 1—figure supplement 2). The robustness of resistance, and the absence of active-site mutations, suggests Cfr variants do not act as dominant-negative enzymes that inhibit C-2 methylation of A2503, as observed in a previous directed evolution experiment (Stojković et al., 2016). Furthermore, the specificity of resistance to PhLOPSA antibiotics suggests that these Cfr variants elicit their effects through PTC modification rather than triggering a stress response that confers global resistance.

Variants exhibit increased rRNA methylation and Cfr protein levels

To test the hypothesis that Cfr variants mediate higher resistance by increasing the fraction of ribosomes with m8A2503, we evaluated the methylation status of A2503 by mass spectrometry. Specifically, we expressed Cfr in E. coli and used oligonucleotide protection to isolate a 40-nt fragment of 23S rRNA containing A2503 (Andersen et al., 2004; Stojković and Fujimori, 2015). The isolated fragment was then enzymatically digested and analyzed by MALDI-TOF mass spectrometry (Figure 2a, Figure 2—figure supplement 1). As expected, an empty vector produces a 1013 m/z fragment corresponding to the mono-methylated m2A2503, modification installed by the endogenous enzyme RlmN. Upon expression of Cfr, we observe a reduction in the 1013 m/z peak and the emergence of a new peak at 1027 m/z, corresponding to m2A2503 conversion into hypermethylated m2m8A2503. CfrWT is able to convert less than ~40% of m2A2503 into the hypermethylated m2m8A2503 product. In contrast, the evolved variants achieve ~50%–90% methylation of A2503, indicating that variants are more active than CfrWT in vivo.

Figure 2 with 4 supplements see all
Cfr variants cause increased methylation of 23S rRNA at A2503, correlating with enhanced production of Cfr protein.

(a) Endogenously modified (m2A2503) and Cfr-hypermodified (m2m8A2503) rRNA fragments correspond to m/z values of 1013 and 1027, respectively. MALDI-TOF mass spectra of 23S rRNA fragments isolated from Escherichia coli expressing CfrWT, and evolved Cfr variants V2, V4, and V7. Ψ is pseudouridine, m2A is 2-methyladenosine, is m2m8A is 2,8-dimethyladenosine. (b) Relative protein expression of full-length Cfr variants compared to full-length CfrWT detected by immunoblotting against a C-terminal FLAG tag and quantification of top Cfr bands. Signal was normalized to housekeeping protein RNA polymerase β-subunit. Data are presented as the average of four biological replicates with standard deviation on a log2 axis. Asterisks denote N-terminally truncated versions of Cfr that do not contribute to resistance. Em = empty vector control. Original uncropped blot images are provided in Figure 2—source data 1. (c) Relative transcript levels for variants compared to CfrWT determined from three biological replicates with standard deviation. Statistical analysis was performed using a two-tailed t-test on log2 transformed data. (d) Percentage of total Cfr expression attributed to the production of full-length Cfr protein, presented as the average of four biological replicates with standard deviation. (e) Doubling times for E. coli expressing empty plasmid, CfrWT, or Cfr variants were determined from three biological replicates with standard error. Numerical data and exact p values where relevant for panels (b–d) are provided in Figure 2—source data 2.

The ability of evolved Cfr variants to achieve enhanced ribosome methylation in vivo could be attributed to enhanced enzymatic activity and/or higher levels of functional enzyme. To test the hypothesis that Cfr variants achieve higher turnover number, we anaerobically purified and reconstituted CfrWT and a representative evolved variant, CfrV4. We then evaluated the ability of CfrWT and CfrV4 to methylate a 23S rRNA fragment (2447–2625) in vitro by monitoring the incorporation of radioactivity from [3H-methyl] S-adenosylmethionine (SAM) into RNA substrate under saturating conditions (Bauerle et al., 2018). However, no significant difference in kcat between CfrWT (3.45×10–2±3.2×10–3 min–1) and CfrV4 (2.25×10–2±1.3×10–3 min–1) was observed (Figure 2—figure supplement 2).

Given these findings, we hypothesized that the variants might alter protein levels. To monitor Cfr protein levels, we inserted a flexible linker followed by a C-terminal FLAG tag, which does not alter resistance (Supplementary file 1). Interestingly, immunoblotting against FLAG revealed that in addition to full-length Cfr, N-terminally truncated Cfr proteins are also produced (Figure 2b). The truncations result from translation initiation at internal methionines but do not contribute to resistance (Figure 2—figure supplement 3), indicating that they are non-functional enzymes unable to methylate A2503. The smaller molecular weight truncation is present in higher levels for all Cfr variants compared to CfrfWT (Figure 2—figure supplement 4). Interestingly the larger molecular weight truncation is present only in CfrV1/V4/V6 and is generated by the I26M mutation introduced during directed evolution. Quantification of resistance-causative, full-length Cfr proteins alone revealed that the evolved variants achieve ~20–100-fold higher steady-state protein levels than CfrWT (Figure 2b).

We measured transcript levels for all variants to assess the contribution of altered transcription to increased protein levels. For Cfr variants with promoter alterations, enhanced production of the Cfr transcript is a large contributor to Cfr protein expression, as CfrV6 and CfrV7 exhibit ~6-fold and ~10-fold enhancement in Cfr mRNA levels compared to CfrWT, respectively (Figure 2c). We also observe a ~2–3-fold increase in mRNA levels for CfrV1-5. Of note, increased Cfr transcript levels likely explain the higher expression of the larger molecular weight truncation observed for all variants (Figure 2c, Figure 2—figure supplement 4). Despite the observed increase in mRNA levels for CfrV1-5, this alone cannot explain the multi-fold improvement in protein expression and indicates that these variants also boost protein levels through a post-transcriptional process. This is further supported by the expression profiles for CfrV1-5, which are dominated by the full-length protein (Figure 2d). Interestingly, enhanced production of Cfr protein correlates with larger fitness defects in E. coli, with an increase in doubling time of ~4 min for CfrV7 compared to empty vector in the absence of antibiotics (Figure 2e).

Promoter and second position mutations drive Cfr resistance

Given that the evolved variants achieve robust enhancement in Cfr expression we sought to elucidate the mechanism(s) by which this occurs. To evaluate the importance of promoter alterations, we generated a construct where the Ptet* promoter sequence from CfrV6 was inserted upstream of CfrWT ORF, herein referred to as Ptet*V6. The insertion of Ptet* alone was sufficient to elicit improvement in Cfr expression (Figure 3a). Furthermore, E. coli expressing Ptet*V6 resembled CfrV6 in its ability to survive in the presence of chloramphenicol (Figure 3b). Taken together, these results suggest the altered promoter drives expression and resistance for CfrV6.

Figure 3 with 2 supplements see all
Mutations to the second amino acid and promoter are the largest contributors to Cfr expression and resistance.

(a) Effect of Cfr mutations and promoter alteration on relative Cfr protein expression was assessed by immunoblotting against a C-terminal FLAG tag. Quantification was performed for full-length Cfr protein normalized to housekeeping protein RNA polymerase β-subunit. Data are presented as the average of four biological replicates with standard deviation on a log2 axis. Asterisks denote N-terminally truncated Cfr protein products that do not contribute to resistance and were not included in quantification. Em = empty vector control. Original uncropped blot images are shown in Figure 3—source data 1. (b) and (c) Dose-dependent growth inhibition of Escherichia coli expressing pZA-encoded CfrWT, CfrV6 (panel (b)), CfrV3 (panel (c)), and individual mutants that comprise these variants toward chloramphenicol (CHL) presented as an average of three biological replicates with standard error. Numerical data for all figure panels are provided in Figure 3—source data 2.

To investigate the contributions of mutations within the Cfr protein, we generated constructs containing Cfr mutations N2K/I, I26M, and S39G in isolation. Interestingly, we observe that mutations at the second position, N2K and N2I, display the largest enhancements in expression, ~27-fold and ~12-fold, respectively (Figure 3a). Similarly to the evolved variants, in addition to the full-length Cfr protein, we also observe expression of the truncation that results from initiation at M95 (Figure 3—figure supplement 1). The dominance of the second position mutants in driving antibiotic resistance is further manifested by E. coli expressing CfrN2K, but not I26M or S39G, exhibiting survival similar to that of the triple mutant, CfrV3, in the presence of chloramphenicol (Figure 3c). Similarly, E. coli expressing CfrN2I also exhibits increased resistance to chloramphenicol when compared to the corresponding directed evolution variant, CfrV5, albeit weaker than CfrN2K (Figure 3—figure supplement 2a). Taken together, these results suggest that the second position mutations drive the robust expression and resistance observed for CfrV1-5. Of note, ribosome methylation by the produced Cfr does not impact the translation of CfrN2K, as this mutant and its corresponding catalytically inactive double mutant protein CfrN2K_C338A are similarly highly expressed (Figure 3—figure supplement 2b-c).

Mutations impact Cfr translation and degradation

The Cfr coding mutations drive enhanced steady-state protein levels of Cfr protein through a post-transcriptional process. However, because levels at steady-state reflect the net effect of protein synthesis and degradation, we sought to evaluate how Cfr mutations impact both processes, especially since the nature of N-terminal amino acids and codons can greatly influence both translation and degradation in bacteria (Gottesman, 2003; Bhattacharyya et al., 2018; Bentele et al., 2013; Tuller et al., 2010; Verma et al., 2019; Goodman et al., 2013; Boël et al., 2016; Looman et al., 1987; Stenström et al., 2001b; Stenström et al., 2001b; Stenström and Isaksson, 2002; Sato et al., 2001).

To test the hypothesis that second position mutations enhance translation of mutants, we used polysome profiling to evaluate the relative abundance of Cfr mRNA in polysome fractions. Polysome profiles derived from 10% to 55% sucrose gradients appear similar across biological conditions, suggesting expression of CfrWT and its evolved mutants do not affect global translation (Figure 4a–b). CfrWT transcripts migrate with low polysomes (fractions 10 and 11) (Figure 4c). In contrast, CfrV4 transcripts are strongly shifted toward high polysomes (fractions 16 and 17), which indicate that CfrV4 mRNA is associated with a large quantity of ribosomes and is better translated than CfrWT (Figure 4d). Further support that CfrV4 is well-translated is the observation that CfrV4 mRNA co-migrates with mRNA of the well-translated housekeeping gene, recA (Li et al., 2014; Figure 4—figure supplement 1). At least in part, this is due to the N2K mutation which shifts transcripts to higher polysomes fractions (fractions 12 and 13) (Figure 4c). The recA control mRNA shows excellent reproducibility across biological samples, indicating that the observed shift of mutant Cfr transcripts toward higher polysomes is due to introduced mutations (Figure 4b). Taken together, these results suggest that enhanced translation is a cumulative effect of N2K and other ORF mutations obtained by directed evolution.

Figure 4 with 5 supplements see all
Directed evolution mutations impact Cfr translation and degradation.

(a) Sucrose gradient fractionation of polysomes from Escherichia coli expressing empty vector or CfrWT/N2K/V4 denoting fractions corresponding to low- and high-density polysomes. (b) mRNA distribution of well-translated, housekeeping gene recA across polysome profiles. (c) mRNA distribution of Cfr transcripts expressing CfrWT or CfrN2K. (d) mRNA distribution of Cfr transcripts expressing CfrWT or CfrV4. For (b–d), transcript levels for each fraction were determined by RT-qPCR and normalized by a luciferase mRNA control spike-in. Values presented as the average of three biological replicates with standard error. (e) Protein degradation kinetics of CfrWT, single mutations CfrN2K/N2I/S39G/I26M, and evolved variant CfrV3 in E. coli after halting expression by rifampicin treatment. Percentage of Cfr protein remaining over time was determined by immunoblotting against C-terminal FLAG tag and presented as the average of three biological replicates with standard error. Original uncropped blot images are shown in Figure 4—source data 1. Numerical data for all figure panels are provided in Figure 4—source data 2.

To further interrogate the role of second position mutations in Cfr translation, we determined the second codon identity for all sequenced variants from the final rounds of evolution (Supplementary file 1). Interestingly, all N2K mutations were encoded by an AAA codon, while AUU encoded all N2I mutations. In E. coli, the tRNA molecules that decode K(AAA) and I(AUU) are slightly more abundant than the wild-type N(AAU), accounting for 3.0% and 5.4% of the tRNA pool compared to 1.9%, respectively (Dong et al., 1996). To test if tRNA abundance and codon sequence contribute to enhanced translation, we evaluated the impact of synonymous codons on protein expression. Lysine codons AAA and AAG are decoded by the same tRNALys in E. coli. Interestingly, mutating CfrN2K from AAA to AAG, which increases G/C content, did not significantly impact expression (Figure 4—figure supplement 2). The isoleucine AUA codon is decoded by the low-abundant tRNAIle2 (Del Tito et al., 1995; Nakamura et al., 2000). Mutation of N2I from AUU to the AUA rare codon resulted in a ~2-fold decrease in Cfr expression, supporting tRNA abundance as a contributing factor (Figure 4—figure supplement 2).

To evaluate the impact of mutations introduced during directed evolution on protein half-life, we monitored changes in protein abundance over time after halting expression with rifampicin (Figure 4e, Figure 4—figure supplement 3). While CfrWT is rapidly degraded with a half-life of ~20 min, CfrN2K/I exhibit increased half-lives of ~60 min. These results suggest that mutation of the second amino acid to lysine or isoleucine contributes to improved steady-state expression both by enhancing translation and stability of Cfr in the cell. CfrS39G also exhibits an increased half-life of ~60 min. The half-life increase is the most pronounced for the I26M single point mutant and similar to that of the triple-mutant, CfrV3 (>100 min for both proteins). Together, these results suggest that evolved variants achieve higher expression through mutations that enhance translation and decrease the degradation of mutant Cfr proteins.

Evolved Cfr enables understanding of the structural basis of resistance

Molecular understanding of Cfr-mediated resistance to antibiotics necessitates structural insights into methylated ribosomes. However, obtaining the structure of a Cfr-modified ribosome has been so far hampered by moderate methylation efficiency of S. aureus Cfr, a challenge that can be addressed by the improved methylation ability of directed evolution variants. Of all characterized evolved variants, CfrV7 achieves the highest levels of antibiotic resistance and methylation of rRNA, providing a unique tool for structural determination. Relative peak quantification of the MALDI spectra revealed that CfrV7 achieved near-stoichiometric (~90%) m8A2503 methylation (Figure 2—figure supplement 1).

Ribosomes were purified from E. coli expressing CfrV7 to obtain a 2.2 Å cryo-EM structure of the Cfr-modified 50S ribosomal subunit (Figure 5a, Table 1, Figure 5—figure supplement 1). The high-resolution cryo-EM density map enabled modeling all known modified nucleotides including the novel C8 methylation of A2503 (Figure 5b). Furthermore, comparison of the Cfr-modified ribosome with the high-resolution cryo-EM structure of unmodified, wild-type E. coli ribosome we published previously (Stojković et al., 2020) allowed us to identify with high confidence any structural changes due to the presence of m8A2503. Importantly, modification of A2503 by Cfr does not affect the conformation or position of the A2503 nucleotide. The adenine ring remains in the syn-conformation and places the newly installed C8-methyl group directly into the PTC to sterically obstruct antibiotic binding (Figure 5c–d).

Figure 5 with 5 supplements see all
Near-stoichiometric ribosome methylation by CfrV7 enables structural understanding of Cfr-mediated resistance to antibiotics.

(a) Cfr-modified 50S ribosomal subunit highlighting adenosine 2503 (A2503) within 23S rRNA and the binding site of PTC-targeting antibiotics. Cfr methylates A2503 at the C8 carbon to produce m2m8A2503. (b) Cryo-EM density maps of adenosine 2503 in 23S rRNA contoured to 3σ. Cfr-modified (m2m8A2503) in cyan. Wild-type (m2A2503) in orange; PDB 6PJ6. (c) Close-up view of 23S rRNA nucleotides in the 50S ribosomal subunit. Cfr-modified Escherichia coli ribosome in cyan. Wild-type E. coli ribosome in orange; PDB 6PJ6. (d) Structural overlay of Cfr-modified E. coli ribosome (cyan) and Haloarcula marismortui 50S ribosome in complex with pleuromutilin antibiotic tiamulin (purple, PDB 3G4S) highlighting steric clashes between m8A2503 and the antibiotic. EM, electron microscopy.

Table 1
Cryo-EM data collection, refinement, and validation statistics.
Data collection and processing
Electron microscopeKrios
Magnification29,000
Number of micrographs2055
Number of particles picked from good micrographs162,713
Number of particles used in final reconstruction141,549
Pixel size (Å)0.822
Defocus range (μm)–0.2 to –1.5
Voltage (kV)300
Electron dose (e-/Å2)80
Map refinement
Model resolution (Å)2.2
FSC threshold0.143
Model resolution range (Å)2.2–20
Map sharpening B-factor (Å2)–55.86
Refinement and model statistics
Clashscore, all atoms2.23
Protein geometry
MolProbity score1.29
Rotamer outliers (%)0.92
Cβ deviations > 0.25 Å (%)0.32
Ramachandran (%)
- Favored95.79
- Allowed4.01
- Outliers0.2
Deviations from ideal geometry
- Bonds (%)0.03
- Angles (%)0.08
Nucleic acid geometry
Probably wrong sugar puckers (%)0.84
Bad backbone conformations (%)12.86
Bad bonds (%)0.07
Bad angles (%)0.08

Strikingly, beyond the addition of a single methyl group to the substrate nucleotide, the presence of m8A2503 does not result in any additional structural changes to the PTC region of the ribosome (Figure 5c). Furthermore, the increased resistance provided by CfrV7 appears to be mediated specifically by improved methylation of A2503. No off-target activity of the evolved variant was observed as manual inspection did not reveal density that could correspond to additional C8-methyl adenosines within the high-resolution regions of the 50S ribosomal subunit. This result was cross-validated using our qPTxM tool (Stojković et al., 2020), which identified only A2503 and A556 as possible C8-methyl adenosines. Closer examination of A556 reveals it registered as a false positive (Figure 5—figure supplement 2a-d).

Contrary to previous reports, we do not observe changes to methylation of C2498, a distal PTC nucleotide whose endogenous 2′-O-ribose modification has previously been reported to be suppressed by Cfr methylation of A2503 and hypothesized to alter the PTC through long-range effects (Kehrenberg et al., 2005; Jensen et al., 2009; Purta et al., 2009). Although it is unclear what percentage of C2498 retains the native modification in our structure, we observe clear density for the methyl group and the nucleotide conformation is unaltered. The density for the methyl group is slightly off of the rotameric position, but the dropoff in density along the methyl bond matches the expected shape (Figure 5—figure supplement 2e-g). Taken together, the results do not indicate that conformational changes to C2498 are involved in Cfr-mediated resistance.

Although Cfr has been identified in animal-derived E. coli isolates (Wang et al., 2012; Deng et al., 2014; Liu et al., 2017; Ma et al., 2021), the resistance gene has primarily been identified clinically in staphylococcal organisms such as S. aureus (Vester, 2018). However, given the high sequence and structural conservation within the PTC region (Figure 5—figure supplement 3), structural impacts of the Cfr m8A2503 modification within E. coli and S. aureus ribosomes are likely conserved.

Structural superposition of the Cfr-modified ribosome with ribosomes in complex with PhLOPSA antibiotics, hygromycin A, nucleoside analog A201A, and 16-membered macrolides enables direct identification of chemical moieties responsible for steric collision with m8A2503 for these eight antibiotic drug classes (Figure 5—figure supplement 4, Figure 5—figure supplement 5). For example, overlay of a bacterial ribosome in complex with the pleuromutilin derivative tiamulin, the selection antibiotic used during directed evolution, reveals steric clashes between the C10 and C11 substituents of the antibiotic with the Cfr-introduced methyl group (Figure 5d). The pleuromutilin class of antibiotics has recently regained interest for their applications as antimicrobial agents in humans but existing molecules remain ineffective against pathogens with Cfr (Goethe et al., 2019). Given recent synthetic advances that enable more extensive modification of the pleuromutilin scaffold (Murphy et al., 2017; Farney et al., 2018), the structural insights we obtained will inform the design of next-generation antibiotics that can overcome Cfr-mediated resistance.

Discussion

By relying on directed evolution under antibiotic selection, we identified strategies that increase the ability of a multi-antibiotic resistance determinant Cfr to cause resistance. Enhanced resistance is associated with improved in vivo methylation of rRNA at the C8 position of A2503. The positive correlation between extent of rRNA modification and resistance aligns with previous studies that investigated linezolid resistance caused by mutation of rRNA, where the severity of linezolid resistance was proportional to the number of 23S rRNA alleles harboring the resistance mutation (Besier et al., 2008; Ebihara et al., 2014; Lobritz et al., 2003). While alteration of the antibiotic binding site through mutations and enzymatic modification of 23S rRNA are functionally distinct, dependence on the extent of rRNA modification provides parallels between the two mechanisms. Although Cfr-mediated methylation is an enzymatic process, the ability of Cfr to confer resistance is restricted by ribosome assembly. Since the A2503 is only accessible to Cfr prior to incorporation of 23S rRNA into the large ribosomal subunit (Yan et al., 2010), the extent of resistance correlates with the ability of the enzyme to methylate 23S rRNA prior to its incorporation into the 50S subunit. The results of our evolution experiment indicate that increasing the intracellular concentrations of Cfr, rather than improving catalysis of an enzyme with a complex radical mechanism (Yan and Fujimori, 2011; McCusker et al., 2012; Grove et al., 2011; Bauerle et al., 2018) is the preferred strategy to increase the proportion of ribosomes with the protective m8A2503 modification.

Investigations into expression levels of CfrWT and its respective mutants revealed that, in addition to full-length protein, a smaller Cfr isoform of ~30 kDa is also produced (Figures 2b and 3a). The truncated product is observed when the expression is driven by both the non-native Ptet and native Pcfr promoter (Figure 6, Figure 6—figure supplement 1). The smaller product likely results from translation initiation at an internal start codon, as mutation of Met at position 95 abolishes its production. The sequence upstream of M95 is A-rich, which has been demonstrated to promote translation initiation (Saito et al., 2020). However, why an internal region of the Cfr ORF would be recognized as an initiation signal is unclear. Truncation of the first 38 residues of Cfr would eliminate a significant portion of the protein, including the N-terminal accessory domain which is likely involved in substrate recognition (Boal et al., 2011; Schwalm et al., 2016). Elimination of this domain provides rationale for why the smaller Cfr isoform does not contribute to resistance, as the protein would likely exhibit perturbed binding to rRNA. Thus, while the truncated product does not contribute to resistance, the potential function of the smaller protein remains elusive and requires further study.

Figure 6 with 1 supplement see all
Start codon selection as a proposed mechanism of Cfr inducibility.

(a) Sequence upstream of cfr from pSCFS1 (Accession: AJ579365) resistance plasmid from an animal-derived Staphylococcus sciuri isolate. The upstream region contains two overlapping upstream ORFs (uORFs) followed by an RNA structural element. Proposed antibiotic-induced ribosome stalling at uORF1/2 and RNA rearrangement could reveal the occluded RBS, allowing translation to initiate at AUG(–3), adding an MKE polypeptide to the N-terminus of Cfr. (b) Addition of an N-terminal MKE peptide to Cfr in the context of the pZA plasmid where expression is controlled by the non-native tetracycline-inducible promoter, Ptet. Growth inhibition of Escherichia coli with pZA-encoded MKE-Cfr in the presence of CHL was determined from two biological replicates with standard error. (c) Relative protein expression of full-length MKE-Cfr compared to full-length CfrWT detected by immunoblotting against a C-terminal FLAG tag and quantification of top Cfr bands. Signal was normalized to housekeeping protein RNA polymerase β-subunit. Data are presented as the average of four biological replicates with standard deviation on a log2 axis. Em = empty vector control. Numerical data for figure panels are provided in Figure 6—source data 1.

The evolved variants improve the expression of resistance-causative, full-length Cfr using two mechanisms. Improved Cfr expression for CfrV6/7 is driven by increased transcription (Figure 2c) due to alterations to the Ptet promoter likely introduced by primer slippage during the EP-PCR step of directed evolution. CfrV6 contains a full duplication of Ptet, providing two sites for transcription initiation, likely responsible for enhanced cfr transcript levels. Interestingly, this result parallels a clinical instance of high Cfr resistance discovered in an S. epidermidis isolate where transcription of cfr was driven by two promoters (LaMarre et al., 2013) and highlights transcriptional regulation as an important mechanism for modulating the in vivo activity of Cfr.

Improved expression for evolved variants CfrV1-5 is mediated by mutations that improve both translational efficiency and protein stability in vivo. Of the tested mutations, I26M provides the largest improvement in stability. Of note, the N-terminally truncated Cfr derived from translation initiation at I26M is rapidly degraded, as no detectable protein is observed after 60 min (Figure 4e). However, these results indicate that the costly production and clearance of this nonfunctional protein is offset by the improved cellular stability of the full-length Cfr carrying the I26M mutation. We also observe modest improvements in protein stability with N2K/I mutants (Figure 4e). In bacteria, the identity of N-terminal residues are important determinants of degradation through N-degron pathways (Dougan et al., 2012; Tobias et al., 1991). During protein synthesis, the N-terminus is co-translationally processed by two enzymes, peptide deformylase to remove the formyl group from Met (fM) and methionine aminopeptidase (Koubek, 2021). Based on previous biochemical work, it is unlikely that CfrWT and CfrN2K/I would have different N-terminal processing, since fMN… and fMK/I… are likely to be efficiently de-formylated (Ragusa et al., 1999) and resistant to methionine excision (Hirel et al., 1989; Frottin et al., 2006; Xiao et al., 2010). Although the precise mechanism by which N2K/I improves Cfr stability remains elusive, these mutations may alter recognition by other enzymes important for degradation, such as endopeptidases or L/F-tRNA-protein transferase (Izert et al., 2021; Ottofuelling et al., 2021).

Of the mutations investigated, N2K is the largest contributor to enhanced Cfr expression and resistance. Although N2K contributes to cellular stability, our results suggest that improved Cfr translation is the dominant role of this mutation (Figure 4c). The effect of N-terminal residues (and thus codons near the start site) on early stages of translation has been well documented. Previous work has demonstrated that minimal secondary structure near the start codon (Kudla et al., 2009; Goodman et al., 2013; Bentele et al., 2013; Boël et al., 2016; Bhattacharyya et al., 2018) and presence of A/U-rich elements downstream of the translation start site (Cifuentes-Goches et al., 2019; Saito et al., 2020) are important factors for efficient translation initiation. RNA secondary structure predictions of the region proximal to the start codon suggest that the N2K mutation (AAU to AAA) could disrupt base pairing between the N2 (AAU) codon and the downstream T7 (ACA) codon (Figure 4—figure supplement 4). However, the base pair between the second and seventh codon is predicted to be retained when N2K is encoded by the AAG isocodon (Figure 4—figure supplement 4), which was also able to increase Cfr protein levels (Figure 4—figure supplement 2) and suggests that alternative mechanisms may be responsible for improved translation. While initiation is a major rate-limiting step in protein synthesis, rates of elongation have also been demonstrated to impact translation efficiency, with several proposed models on how the interconnected factors of codon bias, mRNA structure/sequence, and interactions between the ribosome and nascent chain are involved in modulating protein synthesis (Rodnina, 2016; Choi et al., 2018; Samatova et al., 2020). For example, recent work investigating the role of codons 3–5 identified that both mRNA sequence and amino acid composition are key determinants of proper elongation at the N-terminus (Verma et al., 2019). Although the mechanism is poorly understood, previous studies have discovered that presence of an AAA lysine codon after the start site is associated with improved translation efficiency in certain contexts (Looman et al., 1987; Stenström et al., 2001a; Stenström et al., 2001b; Stenström and Isaksson, 2002; Sato et al., 2001). Our results indicate that the effect of N2K on early steps of translation elongation may be mediated, at least in part, by tRNA abundance, but the exact impact of Lys2 on translation requires further study. Interestingly, the observed internal translation start sites (I26M and M95) that are responsible for producing Cfr truncations (Figure 2b, Figure 2—figure supplement 3) contain a lysine immediately after methionine, further highlighting the putative role for lysine codons in early steps of translation.

To date, only a few S. aureus Cfr variants have been reported and no mutations matching those obtained from directed evolution have been found in clinical isolates. However, enhanced expression through positioning of Lys as the second amino acid of Cfr can be recapitulated by accessing an upstream translational start site found in a native sequence context of cfr (Figure 6). In the specific case of the pSCFS1 resistance plasmid, the sequence upstream of the annotated start codon, which we validated as the start site under the experimental conditions tested (Figure 6—figure supplement 1), contains regulatory elements that have been proposed to modulate Cfr expression (Schwarz et al., 2000; Kehrenberg et al., 2007). It is plausible that in response to antibiotics, the upstream start codon is used to add three amino acids (MKE) to the N-terminus of Cfr and thus placement of a lysine (K) at position 2 of the newly expressed protein, analogous to the N2K mutation. Although start codon selection requires further investigation, N-terminal addition of MKE to Cfr expressed under non-native Ptet promoter phenocopies the N2K directed evolution mutation, resulting in increased expression and resistance compared to CfrWT (Figure 6). Since our assessment of the evolved variants indicates that an increase in Cfr expression is accompanied by a decrease in fitness (Figure 2e), start site selection in response to antibiotic pressure would mitigate detrimental impact on fitness while enabling higher resistance when acutely needed.

Interestingly, mutations obtained through directed evolution have been observed in Cfr homologs that share less than 80% sequence identity with Cfr. Methionine (M) at position 26 is observed for the functionally characterized Cfr homologs Cfr(B) (Deshpande et al., 2015; Marín et al., 2015; Hansen and Vester, 2015) and Cfr(D) (Pang et al., 2020), which have been recovered from human-derived isolates and share 74% and 64% amino acid identity with Cfr, respectively (Schwarz et al., 2021; Figure 4—figure supplement 5). We also observe lysine (K) at position 2, methionine (M) at position 26, and glycine (G) at position 39, akin to N2K, I26M, and S39G mutations, for a number of Cfr homologs that clade with functional Cfr or Cfr-like genes (Stojković et al., 2019). While the precise roles of these residues within less well-characterized and more distantly related Cfr proteins requires further study, these observations indicate that directed evolution accessed sequence space that is already being exploited by proteins that are, or are hypothesized to be, functional Cfr resistance enzymes.

In addition to identifying mechanisms that increase Cfr-mediated resistance, directed evolution of Cfr also provided an indispensable reagent that enabled structural determination of the Cfr-modified ribosome. The high-resolution cryo-EM structure revealed that broad resistance is due to steric effects of the judiciously positioned methyl group within the shared binding site of PTC-targeting antibiotics. Lack of notable changes in position or orientation of A2503 or surrounding PTC nucleotides upon Cfr methylation suggests that the resulting modification does not obstruct the translation capabilities of the ribosome. This absence of PTC disruption is consistent with the observation that the fitness cost of Cfr acquisition is not due to ribosome modification, but rather results from expression of the exogenous protein (LaMarre et al., 2011). Importantly, overlay with existing structures containing PTC-targeting antibiotics provides direct visualization of chemical moieties that are sterically impacted by m8A2503 and will inform the design of antibiotic derivatives that can overcome resistance mediated by Cfr.

Materials and methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Strain, strain background (E. coli)ER2267Tawfik lab stock
Strain, strain background (E. coli)BW25113Keio collection
Strain, strain background (E. coli)BW25113 acrB::KanKeio collection
Strain, strain background (E. coli)Rosetta2 BL21(DE3) pLysSNovagen
Strain, strain background (E. coli)MRE600ATCC
Gene(Staphylococcus aureus)cfr geneAccession: EF450709.1
Gene(Staphylococcus sciuri)Sequence upstream cfr geneGenscriptAccession: AJ579365
Recombinant DNA reagentpZA (plasmid)Wellner et al., 2013; Stojković et al., 2016
Recombinant DNA reagentpET28a (plasmid)Fujimori lab stock
Recombinant DNA reagentpKK3535 (plasmid)Fujimori lab stock
Chemical compound, drugAnhydrotetracycline hydrochlorideSigma-Aldrich37919
Chemical compound, drugTiamulinWako Chemicals328-34002
Chemical compound, drugClindamycinTCI AmericaC2256
Chemical compound, drugChloramphenicolAllstar Scientific480-045
Chemical compound, drugLinezolidAcros460592500
Chemical compound, drugHygromycin ADr. Kim Lewis
Chemical compound, drugTrimethoprimSigma-Aldrich92131
Chemical compound, drug[3H-methyl] S-adenosylmethioninePerkinElmerNET 155 H001MC
Chemical compound, drugRifampicinSigma-AldrichR3501
AntibodyAnti-FLAG(mouse monoclonal)Sigma-AldrichF31651:2000
AntibodyAnti-RNA polymerase beta(rabbit monoclonal)Abcamab1915981:2000
AntibodyAnti-rabbit IgG cross-absorbed DyLight 680 (goat polyclonal)Thermo Fisher Scientific355671:10,000
AntibodyAnti-mouse IgG cross-absorbed IRDye 800CW(goat polyclonal)Abcamab2167721:10,000
AntibodyAnti-GAPDH(mouse monoclonal)Abcamab1252471:2000
Sequence-based reagentcfrIntegrated DNA TechnologiesFor RT-qPCR. The sequence of this oligonucleotide can be found in Supplementary file 1
Sequence-based reagentrecAIntegrated DNA TechnologiesFor RT-qPCR. The sequence of this oligonucleotide can be found in Supplementary file 1
Sequence-based reagentlucIntegrated DNA TechnologiesFor RT-qPCR. The sequence of this oligonucleotide can be found in Supplementary file 1
Sequence-based reagentLuciferase control RNAPromegaL456AControl RNA-spike in for polysome analysis
Commercial assay or kitRNAprotect Bacteria ReagentQIAGEN76506
Commercial assay or kitRNeasy Mini KitQIAGEN74104
Commercial assay or kitiScript cDNA Synthesis KitBio-Rad170-8891
Commercial assay or kitSsoAdvanced Universal SYBRGreen SupermixBio-Rad172-5721
Peptide, recombinant proteinRNaseT1Thermo Fisher ScientificEN0541
Peptide, recombinant proteinRQ1 RNase-free DNase IPromegaM610A
Software, algorithmImage Lab SoftwareBio-Rad
Software, algorithmMotionCor2Zheng et al., 2017
Software, algorithmGCTFZhang, 2016
Software, algorithmCryoSPARC v2.0Punjani et al., 2017
Software, algorithmRelion 3.0Scheres, 2012
Software, algorithmcisTEMGrant et al., 2018
Software, algorithmSharpen3DGrant et al., 2018
Software, algorithmPHENIXAdams et al., 2010
Software, algorithmeLBOWMoriarty et al., 2009
Software, algorithmMolProbityChen et al., 2010
Software, algorithmPymol Molecular Graphics SystemSchrödinger, LLCVersion 2.4.1

E. coli strains and plasmids

Request a detailed protocol

E. coli ER2267 expressing Cfr from a pZA vector (Wellner et al., 2013; Stojković et al., 2016) was used in directed evolution experiments. Antibiotic resistance, fitness, in vivo RNA methylation, and protein/transcript expression, polysome analysis, and protein degradation experiments were conducted with E. coli BW25113 expressing Cfr protein from a pZA vector under the Ptet promoter (or Pcfr promoter where noted). E. coli BW25113 acrB::kan, where the efflux pump acrB was replaced with a kanamycin cassette, was used for antibiotic susceptibility testing of the oxazolidinone antibiotic linezolid and hygromycin A. For experiments for which tagless versions of evolved Cfr variants were used, comparisons were made to the wild-type Cfr protein to which the original C-terminal His tag had been removed. E. coli Rosetta2 BL21(DE3) pLysS was used for overexpression of N-His6-SUMO-tagged Cfrs from a pET28a vector. E. coli MRE600 was used for the preparation of Cfr-modified ribosomes for structural studies.

Cfr mutagenesis and selection scheme

Request a detailed protocol

The wild-type cfr gene (accession: EF450709.1) with a C-terminal His6-tag, or pooled cfr genes from the previous round of evolution, were randomly mutagenized by error-prone polymerase chain reaction as described previously (Stojković et al., 2016). The mutagenized cfr gene pool was then recloned into a pZA vector and transformed into E. coli ER2267. The frequency of mutations was determined by sequencing randomly selected library variants and was on average 1–3 mutations per gene. E. coli transformants were then subjected to selection by plating cells on LB agar containing tiamulin (Wako Chemicals USA), in addition to 100 µg/ml ampicillin for plasmid maintenance and 20 ng/ml anhydrotetracycline (AHT, Sigma-Aldrich) for induction of Cfr expression. For each round of evolution, the E. coli transformants were divided equally and plated on 4–5 plates of LB agar containing different concentrations of tiamulin and grown at 37 oC for up to 48 hr. The tiamulin concentration was increased in 50–100 μg/ml increments. For example, in the first round of evolution, the transformation was plated on the 150, 200, 250, and 300 μg/ml tiamulin plates, in the last round, we selected on 250, 350, 450, and 550 μg/ml tiamulin plates. About 2 ml was plated on tiamulin deficient plates in order to determine transformation efficiency. In general, colonies isolated from tiamulin plates in which the ≤10% of the transformants grew were taken for the next round. After five rounds of mutagenesis and selection, two rounds of enrichment (selection without mutagenesis) using high tiamulin concentrations (400–1500 µg/ml) was conducted. After each round of selection or enrichment, 5–10 randomly selected colonies were sequenced from each plate.

Determination of antibiotic resistance

Request a detailed protocol

Antibiotic resistance experiments by broth microdilution followed established protocols (Wiegand et al., 2008). In brief, 2 ml of LB media with selection antibiotic was inoculated with a freshly transformed colony containing either empty plasmid, CfrWT, or Cfr mutants. Cultures were grown at 37°C with shaking for approximately 2.5 hr. After measuring the OD600 value, cultures were diluted to 106 cells and 50 µl of this dilution was dispensed into 96-well plates containing 50 µl of LB media with antibiotic of interest, ampicillin (100 µg/ml), and AHT (30 ng/ml). Antibiotic resistance of evolved Cfr variants was evaluated using a twofold serial dilution of antibiotic with the following concentration ranges: tiamulin (50–6400 µg/ml, Wako Chemicals); clindamycin (50–6400 µg/ml, TCI America), chloramphenicol (0.5–64 µg/ml, AllStar Scientific), linezolid (1–256 µg/ml, Acros), hygromycin A (2–1024 µg/ml, gifted from Dr. Kim Lewis), and trimethoprim (0.125–0.2 µg/ml, Sigma-Aldrich). Chloramphenicol resistance of single Cfr mutations was evaluated using concentrations of 1, 2–12 µg/ml (in 2 µg/ml step increments), followed by 16–64 µg/ml (twofold dilution). The minimum inhibitory concentration (MIC) required to inhibit visible bacterial growth was determined after incubating plates at 37°C with shaking for 18 hr. Plate OD600 values were also recorded with a microtiter plate reader (SpectraMax M5, Molecular Devices). Antibiotic resistance determination on LB agar plates was conducted as described previously (Stojković et al., 2016; Wiegand et al., 2008). In brief, 3 µl of 108, 106, and 104 dilutions E. coli harboring Cfr were spotted on LB agar plates containing various concentrations of tiamulin. LB agar plates also contained ampicillin (100 µg/ml) and AHT (30 ng/ml). LB agar plates were incubated at 37°C for 24–48 hr.

Oligo-protection of rRNA and MALDI-TOF analysis

Request a detailed protocol

E. coli expressing empty plasmid or Cfr were grown at 37°C to an OD600 of 0.4–0.6 with shaking by diluting an overnight culture 1:100 into LB media containing ampicillin (100 µg/ml) and AHT inducer (30 ng/ml). Total RNA purification, oligo-protection of the 23S rRNA fragment C2480-C2520, and RNaseT1 digestion were performed as described previously (Stojković and Fujimori, 2015; Andersen et al., 2004). Mass spectra were acquired in positive ion, reflectron mode on an AXIMA Performance MALDI TOF/TOF Mass Spectrometer (Shimadzu). Relative peak intensity values were calculated using the Shimadzu Biotech MALDI-MS software.

Expression and purification of Cfr

Request a detailed protocol

CfrWT and CfrV4 were expressed, purified, and reconstituted using modified published protocols (Yan et al., 2010; Stojković and Fujimori, 2015). In brief, N-His6-SUMO-tagged CfrWT/V4 was overexpressed in minimal media conditions with 800 µM IPTG and 1,10-phenanthroline to obtain enzyme lacking a [4Fe-4S] iron-sulfur cluster. Minimal media also contained selection antibiotics kanamycin (50 µg/ml) and chloramphenicol (34 µg/ml). All purification steps were performed in a glovebox (MBraun, oxygen content below 1.8 ppm) that was cooled to 10°C. Cfr was purified by Talon chromatography (Clontech) from clarified lysates. Following chemical reconstitution of the [4Fe-4S], the N-His6-SUMO-tag was cleaved by incubating the fusion protein with SenP1 protease (prepared in-house, 1 mg SenP1:100 mg Cfr) for 1 h at 10 °C in buffer containing 50 mM EPPS (pH 8.5), 300 mM KCl 15 % glycerol, and 5 mM DTT. The cleaved protein was purified away from SenP1 and the N-His6-SUMO-tag by FPLC on a Mono Q 10/100 GL anion exchange column (GE Healthcare Life Sciences) using buffers containing 50 mM EPPS (pH 8.5), 50 mM or 1 M KCl (low-salt or high-salt), 15% glycerol, and 5 mM DTT. Protein was eluted using a linear gradient of 100% low-salt to 100% high-salt buffer over eight column volumes. Fractions containing apo-reconstituted, tag-less Cfr were combined, concentrated using a concentrator cell (Amicon Ultra- 0.5 ml, 30 MWCO), and stored at –80°C. Protein concentration was determined by Bradford assay (Bio-Rad).

Preparation of rRNA substrate

Request a detailed protocol

The E. coli 23S rRNA fragment 2447–2624 used for in vitro experiments was prepared using modified published protocols (Stojković and Fujimori, 2015). The desired DNA fragment was amplified from plasmid pKK3535 using previously established primers (Yan et al., 2010) and used as the template in the in vitro transcription reaction. Following DNase treatment and purification, RNA was precipitated overnight at −20°C by addition of 1/10th volume of 3 M NaOAc, pH 5.5, and 3 volumes of ethanol (EtOH). The RNA was then pelleted and washed with 70% aqueous EtOH, dried, and resuspended in nuclease-free water to obtain a final concentration of ~6 mg/ml. The rRNA fragment was refolded and purified by size exclusion chromatography. To refold the RNA, the sample was heated at 95°C for 2 min and then cooled to 65°C over 5 min. MgCl2 was subsequently added to a final concentration of 10 mM prior to a final cooling step at room temperature for at least 30 min. After removing insoluble debris, RNA was purified by FPLC on a HiLoad 26/60 Superdex 200 column (GE Healthcare Life Sciences) using buffer containing 50 mM HEPES (pH 7.5), 10 mM MgCl2, and 50 mM KCl. Fractions containing the desired rRNA product were combined and precipitated overnight at −20°C by addition of 1/10th volume of 3 M NaOAc, pH 5.5, and 3 volumes of EtOH. The RNA was then pelleted and washed with ice-cold 80% aqueous EtOH, dried, and resuspended in nuclease-free water. After confirming RNA purity on a denaturing 5% TBE, 7 M Urea-PAGE gel, the RNA sample was concentrated to ~450 mM using a SpeedVac Vacuum Concentrator prior to storage at –80°C.

Cfr kinetic assay

Request a detailed protocol

Methylation activity of CfrWT and CfrV4 were assessed by monitoring radioactivity incorporation into RNA. Flavodoxin and flavodoxin reductase enzymes were prepared as described previously (McCusker et al., 2012). Prior to assembling reaction components, the RNA substrate was refolded as described above. Reactions were conducted in 52 μl volumes in an anaerobic chamber (MBraun, oxygen levels less than 1.8 ppm) under the following conditions: 100 mM HEPES (pH 8.0), 100 mM KCl, 10 mM MgCl2, 2 mM DTT, 50 µM Flavodoxin, 25 µM Flavodoxin reductase, 100 µM rRNA substrate, 2 mM [3H-methyl] S-adenosylmethionine (175.8 dpm/pmol), and 5 µM apo-reconstituted Cfr. Reactions were equilibrated at 37°C for 5 min and subsequently initiated by addition of NADPH (Sigma-Aldrich, final concentration 2 mM). The reaction was allowed to proceed at 37°C and timepoints at 0, 2, 4, 6, and 8 min of 10 µl volume were quenched by the addition of H2SO4 (50 mM final concentration). For each time point, the RNA volume was brought up to 100 µl with nuclease-free water and was purified away from other reaction components by an RNA Clean & Concentrator kit (Zymo Research) by following the manufacturer’s instructions. Purified RNA eluate was added to Ultima Gold scintillation fluid, and the total amount of radioactivity incorporated in the product was detected using a Beckman–Coulter LS6500 scintillation counter. Amount of product generated at each time point was calculated by subtracting background radioactivity (t=0 min) and taking into account that two of the three tritium atoms from [3H-methyl] S-adenosylmethionine would be incorporated into the final methylated RNA product (Yan and Fujimori, 2011; Bauerle et al., 2018).

Evaluation of Cfr protein expression by quantitative western blot

Request a detailed protocol

E. coli expressing empty plasmid, CfrWT, or Cfr mutants were grown at 37°C to an OD600 of ~0.4 with shaking by diluting an overnight culture 1:100 into 10 ml LB media containing ampicillin (100 µg/ml) and AHT inducer (30 ng/ml). Cells were harvested by centrifugation. Cell pellets were lysed for 15 min using B-PER Bacterial Protein Extraction Reagent (Thermo Fisher Scientific) containing DNase I (New England Biolabs) and 1× cOmplete, EDTA-free protease inhibitor cocktail (Roche). Whole-cell lysate samples containing 4 µg of protein were fractionated using a 4–20% SDS-PAGE gel (Bio-Rad). Proteins were transferred to a 0.2-µm nitrocellulose membrane using a Trans-Blot Turbo transfer system (Bio-Rad) with a 7 min, mixed MW protocol. Membranes were incubated with TBST-Blotto buffer (50 mM Tris-pH 7.5, 150 mM NaCl, 0.1% Tween-20, 5% w/v Bio-Rad Blotting Grade Blocker) for 1 hr at room temperature, followed by TBST-Blotto containing two primary antibodies: monoclonal mouse anti-FLAG M2 (1:2000 dilution, Sigma-Aldrich) and monoclonal rabbit anti-RNA polymerase beta (1:2000 dilution, Abcam) for 1 hr at room temperature. After washing 3× for 5 min with TBST, membranes were then incubated overnight at 4°C with TBST-Blotto containing two secondary antibodies: goat anti-rabbit IgG cross-absorbed DyLight 680 (1:10,000 dilution, Thermo Fisher Scientific) and goat anti-mouse IgG cross-absorbed IRDye 800CW (1:10,000 dilution, Abcam). Membranes were rinsed 3× for 5 min with TBST and allowed to dry completely prior imaging using a Bio-Rad ChemiDoc Molecular Imager. Quantification was performed using Image Lab Software (Bio-Rad) within the linear range of detection. The housekeeping protein RNA polymerase beta, which was stably expressed in all experimental conditions, was used as an internal loading control.

Determination of E. coli growth rate

Request a detailed protocol

E. coli expressing empty plasmid, CfrWT, or Cfr variants were grown at 37°C with shaking by diluting a 50 µl of an overnight culture into 10 ml of LB media containing ampicillin (100 µg/ml) and AHT inducer (30 ng/ml). OD600 values were recorded every 20 min with a microtiter plate reader (SpectraMax M5, Molecular Devices).

qPCR Primer Design and Validation

Request a detailed protocol

qPCR primer sequences for cfr, recA, and luc were designed using NCBI Primer Blast. Template accession numbers, amplicon length, and primer sequences are described in Supplementary file 1. Primer sequences for rrsA were used as published previously (Zhou et al., 2011). For each primer pair primer, qPCR was performed on a tenfold dilution series of desired samples. Amplification efficiency was calculated from the slope of the graph of Cq values plotted against log10 of the at least four template concentrations. Primers for recA: Y=–3.238*X+38.46, R2=0.9992, PCR efficiency=103.6%. Primers for luc: Y=–3.316*X+34.52, R2=0.9967, PCR efficiency=100.2%. Primers for cfr: Y=–3.254*X+37.52, R2=0.9960, PCR efficiency=102.9%. Primers for rrsA: Y=–3.629*X+32.24, R2=0.9965, PCR efficiency=90.0%.

Determination of Cfr mRNA expression by RT-qPCR

Bacterial growth

Request a detailed protocol

E. coli expressing empty plasmid control, CfrWT, or Cfr variants were grown at 37°C with shaking by diluting an overnight culture 1:100 into 5 ml of LB media containing ampicillin (100 µg/ml) and AHT inducer (30 ng/ml). When cells reached an OD600 of ~0.4, RNAprotect Bacteria Reagent (QIAGEN) was added to the culture following the manufacturer’s instructions. Cells were then harvested by centrifugation for 10 min at 5000×g at 4°C and frozen on dry ice.

Total RNA isolation and DNase treatment

Request a detailed protocol

Pellets were then thawed and resuspended in 200 µl of lysis buffer containing 30 mM Tris-HCl (pH 8.0), 0.1 mM EDTA, 15 mg/ml lysozyme, and Proteinase K (New England Biolabs). Following lysis for 10 min at room temperature, total RNA was isolated using an RNeasy Mini Kit (QIAGEN) following the manufacturer’s instructions. Yield and purity of isolated RNA were assessed by NanoDrop UV spectrophotometer (Thermo Fisher Scientific). RNA integrity was assessed by performing 1% TBE agarose gel electrophoresis with samples that had been boiled for 95°C for 5 min in RNA loading dye (New England Biolabs). Genomic DNA was eliminated by incubating 2 µg of RNA with 2 U of RQ1 RNase-free DNase I (Promega) for 30 min at 30°C. The DNase reaction was halted by the addition of RQ1 Stop Solution (Promega) and incubation for 10 min at 65°C.

cDNA synthesis

Request a detailed protocol

Reverse transcription was performed using the iScript cDNA Synthesis Kit (Bio-Rad) following the manufacturer’s instructions with tenfold diluted DNase-treated RNA. In brief, reactions of 20 µl volume were prepared by combining 4 µl 5× iScript buffer, 1 µl iScript RNase H + MMLV reverse transcriptase, 11 µl nuclease-free water, and 4 µl of RNA. Reactions were incubated for 5  min at 25°C, followed by 20  min at 42°C and 1  min at 95°C. If not used immediately, cDNA was stored at –20°C.

RT-qPCR

Request a detailed protocol

SsoAdvanced Universal SYBR Green Supermix (Bio-Rad) was used for 10  μl qPCR reactions. Each reaction contained 5 µl of 2× Supermix, 0.3 µM of each forward and reverse primer, and 4 µl of diluted cDNA. The cDNA was diluted 40-fold for reactions with cfr primers and 4000-fold for reactions rrsA primers. Reactions were prepared in a 96-well PCR Plate (Bio-Rad, MLL9601) and run on a Bio-Rad CFX qPCR Machine. The thermal cycling conditions were as follows: 98°C for 30 s, followed by 35 cycles of 98°C for 10 s and 60°C for 45 s with plate read, ending with melt curve analysis using 5 s, 0.5°C increment steps from 65°C to 95°C. A no template control and no reverse transcription control were included on each plate for each primer pair. Cq values were determined using CFX Maestro Software using a single threshold method. For each sample, the average of three triplicate Cq values was used for further analysis. Relative transcript expression was calculated using the Pfaffl method (Pfaffl, 2001). Expression was normalized to rrsA transcripts which is stably expressed in E. coli BW25113 (Zhou et al., 2011) and across our experimental conditions.

Polysome analysis

Lysate preparation and sucrose gradient fractionation were adapted from previously published protocols with modification (Mohammad and Buskirk, 2019; Li et al., 2014).

Lysate preparation

Request a detailed protocol

E. coli expressing empty plasmid control, CfrWT, or Cfr mutants were grown at 37°C with shaking by diluting an overnight culture 1:100 into 400 ml of LB media containing ampicillin (100 µg/ml) and AHT inducer (30 ng/ml). Cells were harvested at an OD600 ~0.4–0.5 in 200 ml batches by rapid filtration at 37°C followed by flash freezing in liquid nitrogen as described previously (Li et al., 2014). Each frozen cell pellet was combined with 650 µl lysis buffer as frozen droplets containing 20 mM Tris (pH 8.0), 10 mM MgCl2, 100 mM NH4Cl, 5 mM CaCl2, 0.4% Triton X-100, 0.1% NP-40, 100 U/ml RNase-free DNase I (Roche), and 10 U/ml SUPERase-In (Invitrogen). Cells with lysis buffer were pulverized in a 10-ml jar containing a 12-mm grinding ball using a TissueLyser II (QIAGEN) by performing five rounds of 3 min at 15 Hz. Canisters were pre-chilled by submersion in liquid nitrogen for at least 1 min prior to each round of pulverization. Lysates were recovered from the frozen jars using a spatula pre-chilled in liquid nitrogen and stored at –80°C until further use.

Sucrose gradient fractionation

Request a detailed protocol

Pulverized lysates were thawed at 30°C for 2 min followed by an ice-water bath for 20 min. Lysates were clarified by centrifugation at 20,000×g for 10 min at 4°C. The RNA concentration of the clarified lysate was measured by NanoDrop UV spectrophotometer (Thermo Fisher Scientific) and diluted to 2.5 mg/ml with lysis buffer. Ribosome and mRNA components were separated on a linear, 12 ml, 10–55% (w/v) sucrose gradient containing 20 mM Tris (pH 8.0), 10 mM MgCl2, 100 mM NH4Cl, 2 mM DTT, and 10 U/ml SUPERase-In. Sucrose gradients were generated using a Bio-Comp Gradient Master with the following program: Time=1:58 s; Angle=81.5o, Speed=16 rpm. For each biological sample, 190 µl (~0.5 mg RNA) of clarified lysate was loaded onto sucrose gradients in duplicate. Ultracentrifugation was performed using a SW Ti41 rotor (Beckman Coulter) for 201,000×g for 2.5 hr at 4°C. Gradients were fractionated using a Bio-Comp Fractionator in 20 fractions at a speed of 0.25 mm/s where absorbance at 260 nm was continuously monitored.

RNA extraction and DNase treatment

Request a detailed protocol

Fractions 1+ 2, 3+4, 16+17, and 18+19 were combined. RNA was extracted from uniform volumes of each fraction or combination of fractions. RNA extraction was performed by adding one volume of TRIzol reagent (Invitrogen), mixing until homogeneous, and incubating at room temperature for 5 min. Samples were then incubated at room temperature for another 5 min following the addition of 0.4 volumes of chloroform. After centrifugation for 15 min at 12,000×g at 4°C, the aqueous supernatant was transferred to a new tube to which 250 pg of a luciferase control RNA spike-in (luc, Promega). RNA was precipitated overnight at –20°C by the addition of 1 volume of isopropanol and 2 µl of GlycoBlue (15 mg/ml, Invitrogen). RNA was pelleted by centrifugation, washed two times with 75% ice-cold, aqueous ethanol, and allowed to dry at room temperature for ~30 min. The RNA was then resuspended in 20 µl of nuclease-free water. RNA quality and concentration were assessed by a NanoDrop UV spectrophotometer (Thermo Fisher Scientific). Genomic DNA was eliminated by incubating 10 µl of isolated RNA with 1 U of RQ1 RNase-free DNase I (Promega) for 30 min at 30°C. The DNase reaction was halted by the addition of RQ1 Stop Solution (Promega) and incubation for 10 min at 65°C.

cDNA synthesis and RT-qPCR

Request a detailed protocol

Reverse transcription was performed using the iScript cDNA Synthesis Kit (Bio-Rad) following the manufacturer’s instructions. In brief, reactions of 20 µl volume were prepared by combining 4 µl 5× iScript buffer, 1 µl iScript RNase H + MMLV reverse transcriptase, 5 µl nuclease-free water, and 10 µl of DNase-treated RNA. Reactions were incubated for 5  min at 25°C, followed by 20  min at 42°C and 1  min at 95°C. SsoAdvanced Universal SYBR Green Supermix (Bio-Rad) was used for 10  μl qPCR reactions in a 96-well plate as described above. Each reaction contained 5 µl of 2× Supermix, 0.3 µM of each forward and reverse primer, and 4 µl of tenfold diluted cDNA. Reactions containing cfr, recA, and luc primers (Supplementary file 1) were performed for each fraction, including a no template control and no reverse transcription control for each primer set on each plate. The average of three triplicate Cq values was used for further analysis.

Data analysis

Request a detailed protocol

Normalized mRNA distribution profiles for the target mRNAs were calculated as described previously (Pringle et al., 2019). In brief, the relative abundance of each target mRNA normalized to luciferase RNA spike-in. The percentage of target mRNA found across gradient fractions was calculated by dividing the amount of target mRNA detected in one fraction by the sum of the target mRNA detected in all fractions.

Protein degradation assay

Bacterial growth and rifampicin treatment

Request a detailed protocol

E. coli expressing CfrWT or Cfr mutants were grown at 37°C with shaking by diluting an overnight culture 1:100 into 25 ml of LB media containing ampicillin (100 µg/ml) an AHT inducer (30 ng/ml). When cells reached an OD600 ~0.4–0.5, rifampicin (Sigma-Aldrich) was subsequently added to a final concentration of 100 µg/ml, and cultures were allowed continued incubation at 37°C with shaking. Time points at 0, 20, 40, 60, 80, and 100 min were harvested by centrifuging 3 ml of the culture at 8000 rpm at 4°C for 10 min, decanting the supernatant, and immediately flash-freezing the pellet in liquid nitrogen. Cell pellets for each time point were lysed using B-PER Bacterial Protein Extraction Reagent as described above.

Western blot

Request a detailed protocol

Whole-cell lysate samples containing 5 µg of protein were fractionated on a 4–20% SDS-PAGE gel and transferred onto a 0.2-µm nitrocellulose membrane as described above. Membranes were stained with Ponceau S stain (0.1% w/v Ponceau S, 5% v/v acetic acid) and imaged using a Bio-Rad ChemiDoc Molecular Imager. After blocking in TBST-Blotto buffer for 1 hr at room temperature, membranes were incubated with TBST-Blotto containing primary monoclonal mouse anti-FLAG M2 antibody (1:2000 dilution, Sigma-Aldrich) or monoclonal mouse anti-GAPDH antibody (1:2000 dilution, Abcam) for 1 hr at room temperature. After washing 3× for 5 min with TBST, membranes were incubated overnight at 4°C with TBST-Blotto containing a secondary antibody, goat anti-mouse cross-absorbed IRDye 800CW (1:10,000 dilution, Abcam). Membranes were rinsed and imaged as described above.

Data analysis

Request a detailed protocol

Quantification was performed using Image Lab Software (Bio-Rad) within the linear range of detection. The Ponceau S total protein stain was used to normalize for differential sample loading. Percentage (%) of initial Cfr protein remaining was calculated by dividing the amount of Cfr protein at a given time point after rifampicin treatment by the amount of Cfr protein prior to rifampicin treatment (t=0 min).

Purification of Cfr-modified E. coli ribosome

Request a detailed protocol

Cfr-modified, 70S ribosomal subunit was purified from E. coli MRE600 expressing CfrV7 variant using previously published protocol with modification (Mehta et al., 2012; Stojković et al., 2020). In short, E. coli transformed with pZA-encoded CfrV7 were grown to an OD600 of 0.5 in LB media containing ampicillin (100 µg/ml) and AHT inducer (30 ng/ml) at 37°C with shaking. Cells were harvested by centrifugation, washed, and lysed by using a microfluidizer. The lysate was clarified by ultracentrifugation at 30,000×g 30 min at 4°C using a Ti45 rotor (Beckman Coulter) two times. The recovered supernatant was applied to a 32% w/v sucrose cushion in buffer containing 20 mM Hepes-KOH (pH 7.5), 500 mM NH4Cl, 20 mM Mg(OAc)2, 0.5 mM EDTA, 6 mM β-mercaptoethanol, 10 U/ml SuperASE-In and was ultracentrifuged at 100,000×g for for 16 hr at 4°C in a SW Ti41 rotor (Beckman Coulter). After removing the supernatant, the pellet was resuspended slowly at 4°C over 1 hr in Buffer A containing 20 mM Hepes-KOH (pH 7.5), 200 mM NH4Cl, 20 mM Mg(OAc)2, 0.1 mM EDTA, 6 mM β-mercaptoethanol, and 10 U/ml SuperASE-In. Particulates that were not resuspended were removed by centrifugation at 10,000 rpm for 10 min at 4°C. Sample concentration was determined by NanoDrop UV spectrophotometer (Thermo Fisher Scientific), where A260=1 corresponds to 24 pmol of 70S ribosome. Tight-coupled 70S ribosomes were purified as described previously (Khusainov et al., 2017). In brief, 70S ribosomes were purified on a 15–30% w/v sucrose gradient in Buffer A. Sucrose gradients were generated using a Bio-Comp Gradient Master. 300–400 pmol of 70S ribosomes were loaded on each sucrose gradient. Ultracentrifugation was performed using a SW Ti41 rotor (Beckman Coulter) for 75,416×g for 16 hr at 4°C. Gradients were fractionated using a Bio-Comp Fractionator in 20 fractions at a speed of 0.25 mm/s where absorbance at 260 nm was continuously monitored. Fractions corresponding to 70S ribosomes were combined and precipitated by slow addition at 4°C of PEG 20,000 in Buffer A to a final concentration of 9% w/v. Ribosomes were isolated by centrifugation for 10 min at 17,500×g. After removing the supernatant, ribosomes were slowly resuspended overnight at 4°C in buffer containing 50 mM Hepes-KOH (pH 7.5), 150 mM KOAc, 20 mM Mg(OAc)2, 7 mM β-mercaptoethanol, 20 U/ml SuperASE-In.

Cryo-EM analysis

Request a detailed protocol

Purified 70S ribosomal subunits were diluted from 2 to 0.5 mg/ml in Buffer A, applied to 300-mesh carbon coated (2 nm thickness) holey carbon Quantifoil 2/2 grids (Quantifoil Micro Tools) and flash-frozen as described in Khatter et al., 2015. Data were collected using serialEM on the in-house Titan Krios X-FEG instrument (Thermo Fisher Scientific) operating at an acceleration voltage of 300 kV and a nominal underfocus of Δz=0.2–1.5 μm at a nominal magnification of 29,000 (calibrated physical pixel size of 0.822 Å). We recorded 2055 movies using a K2 direct electron detector camera in super-resolution mode with dose fractionation (80 individual frames were collected, starting from the first one). Total exposure time was 8 s, with a total dose of 80 e- (or 1 e-/Å2/frame). Images in the stack were aligned using the whole-image motion correction and patch motion correction (5×5 patches) methods in MotionCor2 (Zheng et al., 2017). Before image processing, all micrographs were checked for quality and 1531 best were selected for the next step of image processing. The contrast transfer function of each image was determined using GCTF (Zhang, 2016) as a standalone program. For particle selection, we have used Relion 3.0 autopicking procedure (Scheres, 2012). For the first steps of image processing, we used data binned by a factor of 8 (C8 images). During the first round of 2D classification, we removed only images with ice or other contaminants. Subsequently, the initial structure was generated using the ab initio procedure in CryoSPARC v2.0. Following this step, we performed Relion 3D classification with bin by four data (C4) in order to exclude bad particles. The resulting 141,549 particle images of ribosomes were used for subsequent classification and refinement procedures. For the initial refinement, we used a spherical mask, which was followed by further refinement using a mask around the stable part of 50S (excluding L1 stalk, L7/L12 region). A further improved cryo-EM map was obtained by using CTF-refinement procedure from Relion 3.0. The post-processing procedure implemented in Relion 3.0 (Scheres, 2012) was applied to the final maps with appropriate masking, B-factor sharpening (automatic B-factor estimation was –55.86) and resolution estimation to avoid over-fitting (final resolution after post-processing with 50S mask applied was 2.7 Å). Subsequently, the stack of CTF-refined particles was processed in a new version of CryoSPARC v2.0 (Punjani et al., 2017). After homogeneous refinement, the same stack of particles was additionally refined in cisTEM (Grant et al., 2018). After Auto-Refine (with automasking within cisTEM), we performed local refinement using 50S mask (the same one used for refinement in Relion) and also applied per particle CTF refinement as implemented in cisTEM software. After such refinement, the resolution was improved to 2.2 Å (Figure 5—figure supplement 1). This map after Sharpen3D (Grant et al., 2018) was used for model building and map interpretation.

Atomic model building and refinement

Request a detailed protocol

The final model of the 50S subunit was generated by multiple rounds of model building in Coot (Emsley et al., 2010) and subsequent refinement in PHENIX (Adams et al., 2010). The restraints for the novel m2m8A nucleotide for the atomic model fitting and refinements were generated using eLBOW (Moriarty et al., 2009). The atomic model of the 50S subunit from the E. coli ribosome structure (PDB 6PJ6) (Stojković et al., 2020) was used as a starting point and refined against the experimental cryo-EM map by iterative manual model building and restrained parameter-refinement protocol (real-space refinement, positional refinement, and simulated annealing). Final atomic model comprised of ∼92,736 atoms (excluding hydrogens) across the 3015 nucleotides and 3222 amino acids of 28 ribosomal proteins. Proteins L7, L10, L11, and L31 were not modeled in. In addition, 179 Mg2+, 2716 water molecules, 1 Zn2+, and 1 Na+ were included in the final model. Prior to running MolProbity (Chen et al., 2010) analysis, nucleotides 878–898, 1052–1110, 2101–2189 of 23S rRNA, and ribosomal protein L9 were removed, due to their high degree of disorder. Overall, protein residues and nucleotides show well-refined geometrical parameters (Table 1). Figures were prepared using Pymol Molecular Graphics System, Version 2.4.1 unless otherwise noted.

qPTxM analysis of post-transcriptional modifications

Request a detailed protocol

The final model and map were run through qPTxM (Stojković et al., 2020) with default parameters except for d_min=2 and cc_threshold=0.5 to search for evidence of posttranscriptional modifications. Of a total of 39 sites with density suggesting possible modifications, 2 were C8-methyl adenosines, A556 and A2503. None of the identified sites were 2′O-methyl cytosines. To calculate expected density dropoff curves for methylated and unmethylated nucleotides, the phenix.fmodel (Adams et al., 2010) tool was used to generate noise-free maps from models of a single nucleotide in each state, and scripts modified from qPTxM were used to collect measurements of the density at 0.1 Å intervals along the vector of the proposed methylation. Means and standard deviations were calculated for densities at the four positions tested by qPTxM on each nucleotide, from which Z-scores were then calculated for selected nucleotides. To measure densities for both the best tested rotamer of m(2′O)C 2498 and the modeled rotamer, densities along the 2′O-methyl bond were compared between the files generated by qPTxM run two times as described above, once with prune=True (removing the modeled methyl group and placing the rotameric methyl with the strongest density) and once with prune=False (leaving the modeled methyl group intact).

Data availability

Atomic coordinates have been deposited in the Protein Data Bank under accession number 7LVK, and the density map has been deposited in the EMDB under accession number 23539.

The following data sets were generated
    1. Stojkovic V
    2. Myasnikov AG
    3. Frost A
    4. Fujimori DG
    (2021) RCSB Protein Data Bank
    ID 7LVK. Cfr-modified 50S subunit from Escherichia coli.
    1. Stojkovic V
    2. Myasnikov AG
    3. Frost A
    4. Fujimori DG
    (2021) Electron Microscopy Data Bank
    ID EMD-23539. Cfr-modified 50S subunit from Escherichia coli.
The following previously published data sets were used
    1. Stojkovic V
    2. Myasnikov A
    3. Frost A
    4. Fujimori DG
    (2020) RCSB Protein Data Bank
    ID 6PJ6. High resolution cryo-EM structure of E.coli 50S.

References

    1. Emsley P
    2. Lohkamp B
    3. Scott WG
    4. Cowtan K
    (2010) Features and development of Coot
    Acta Crystallographica. Section D, Biological Crystallography 66:486–501.
    https://doi.org/10.1107/S0907444910007493
  1. Book
    1. Mehta P
    2. Woo P
    3. Venkataraman K
    4. Wali Karzai A
    (2012) Ribosome Purification Approaches for Studying Interactions of Regulatory Proteins and RNAs with the Ribosome
    In: Keiler Kenneth C, editors. Bacterial Regulatory RNA: Methods and Protocols. Totowa, NJ: Humana Press. pp. 273–289.
    https://doi.org/10.1007/978-1-61779-949-5_18
  2. Book
    1. Vester B
    2. Long KS
    (2013)
    Antibiotic Resistance in Bacteria Caused by Modified Nucleosides in 23S Ribosomal RNA
    Landes Bioscience.
    1. Wilson DN
    (2009) The A–Z of bacterial translation inhibitors
    Critical Reviews in Biochemistry and Molecular Biology 44:393–433.
    https://doi.org/10.3109/10409230903307311

Decision letter

  1. Marina V Rodnina
    Reviewing Editor; Max Planck Institute for Biophysical Chemistry, Germany
  2. Philip A Cole
    Senior Editor; Harvard Medical School, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Thank you for submitting your article "Directed evolution of the rRNA methylating enzyme Cfr reveals molecular basis of antibiotic resistance" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by a Reviewing Editor and Michael Marletta as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

The referees agreed that the experiments were carefully performed and interpreted. The primary experimental revision (point 1) relates to providing a suitable control that tRNA abundance determines the higher translational efficiency of the mutant Cfr constructs. There are also suggestions for further experiments that would strengthen the paper (points 2, 4, 5), but we feel this is optional and could be viewed as beyond the scope of this particular study. I therefore leave it to you whether you perform these additional experiments (which might well be interesting).

1. Directed evolution yielded protein isoforms that are better expressed and are more stable. The authors show that mutant mRNAs are shifted towards heavier polysomes and suggest that this may be due to the tRNA abundance effect (Figure 4 – supplement 1). However, this experiment lacks a control. If tRNA abundance is crucial, overexpression of the tRNA decoding the wt AAU codon should also increase protein expression.

2. The authors find two shortened Cfr isoforms that apparently arise from incorrect initiation on internal methionine codons on cfr mRNA (I26M, M95) and show that translation can start at an initiation codon upstream of M1. This is by itself an exciting story in the Cfr story. The regulatory mechanism due to an uORF is plausible and well documented in Figure 4- supplement 3 and 4. However, the appearance of a (native) internal initiation site starting from M95 is surprising. Given that this is a very prominent translation product (well over 50% with the native Pcfr, Figure 4 – supplement 4), this aspect of the paper raises questions. How does the initiation from M95 depend on antibiotics? The authors clearly show that the shorter isoform does not confer resistance and is rapidly degraded. So, what is the functional role of the shorter isoform? To address these questions the authors could control how much Cfr-short isoform (M95) is present in the absence of antibiotics and how it is affected upon antibiotic treatment. Figures 4-supplementary 3 and 4 are so interesting that I would suggest including them in the main text. The expression of Cfr short (M95) seems to depend on the N2 mutation. This should be quantified and explained.

3. There is a rich literature (cited on p. 21 of the discussion) that may suggest alternative explanations. The authors should identify which of these suggestions can be excluded based on their data and clearly describe those that could contribute to the observed expression effect.

The text on p. 21, lines 6-19 should be modified, as it does not exactly state what was actually suggested, there are too many citations without clearly stating what these papers say.

4. The increased stability of the mutated Cfr variants is fascinating, but not well addressed. What causes the increase in protein stability? What is actually the N-terminus of the protein: is it deformylated and the N-terminal Met cleaved as efficiently as the wt? Is there any correlation with the N-end rule? One way to answer these questions would be to check the N-terminus of proteins from the gels by mass spectrometry. Add a few sentences in the Discussion as to how a (presumably N-terminal) Lys or Ile can increase protein stability.

5. Cfr methylates C8 of A2503. Mass spectra clearly distinguish the peaks of m2m8A2503 and m2A2503. However, this appears to rely on the assumption that A2503 is completely m2-modified. Has this been checked? What would be the mass of the unmodified fragment and was it identified in the spectra? Can the authors exclude the appearance of m8A2503 species? Please comment on the modification completeness of m2A2503 and provide the expected m/z value for the respective unmodified RNA fragment.

6. Since the ribosome structure presented is of a modified E. coli ribosome and the cfr enzyme is originally from S. aureus, the sequence alignment of Cfr from S. aureus and E. coli should be presented (as a Supplement). The authors should discuss whether it is legitimate to use S. aureus cfr to modify E. coli ribosome. This would widen the scope of the paper by generalizing the conclusions for gram-positive and gram-negative bacteria. This could be addressed in the introduction (e.g., p. 4 line 9 and line 15) and in the Results section (e.g., p.16 line 4 and line 9). In Figure 5 – add to the legend that this is an E. coli ribosome. In addition, please relate to this point in the discussion (p. 20).

7. The authors superimposed the complex structure of an archeal ribosome and linezolid and tiamulin for explaining the resistance mechanism. However, this resistance mechanism was not found in archaea and thus these ribosomes are less relevant for clinical aspects. More appropriate models to be used are of bacterial ribosomes in complex with these antibiotics, i.e. D. radiodurans ribosome in complex with tiamulin (PDBID 1XBP) and the pathogen S. aureus ribosome in complex with linezolid (PDBID 4WFA).

8. The paper would benefit from an analysis of the natural variability of the evolved sites. It is mentioned that clinical isolates do not have these exact substitutions, but a supplemental multiple sequence alignment of homologues with these positions marked would anyway be interesting. For example, an N2K variant appears in the results of a quick blastp search in a Bacillus Cfr.

9. Another remaining question is how the novel variants affects susceptibilities to ribosome-targeting antibiotics in other structural classes, including nucleoside analogues and macrolides. In the introduction the authors mention that in addition to PhLOPSA antibiotics, Cfr can confer resistance to nucleoside analog A201A, hygromycin A, and 16-membered macrolides. However, this is not returned to in the paper. MIC experiments with an expanded selection of antibiotics would be really nice to see if possible. Similarly, does the structure explain the previously observed cross-resistance that goes beyond PhLOPS(A)?

Reviewer #1:

The paper by Tsai et al. describes the antibiotic resistance mechanism caused by an RNA-modifying enzyme Cfr. The paper is important, as Cfr confers resistance to eight classes of antibiotics (collectively known as PhLOPSA). The strength of the manuscript is in the combination of different experimental techniques. These include directed evolution of Cfr; mass spectrometry analysis of the modification site in the 23S rRNA; mutational studies with quantitative analysis of their effect on the mRNA and protein levels of Cfr in the cell and its link to resistance; and finally, a high-resolution of cryo-EM of the ribosome with the fully modified m2m8A2503, the product of Cfr activity. Using directed evolution, the authors identified two mutational hotspots that increased resistance, one due to a more active promotor and another due to a mutation of residue N2 in Cfr. They demonstrate that Cfr methylates C8 of m2A2503 and that Cfr mutants are more efficient in installing the m8 modification. However, the catalytic activity of the mutants is not affected. Rather, the cellular concentration of Cfr is increased due to higher translation efficiency and higher Cfr protein stability. Finally, the cryo-EM structure clearly shows the modified nucleotide, which provides explanations as to how a single methyl group causes resistance to so many antibiotics.

The work is elegantly designed, and the experiments are of high technical quality. The paper is very well written and most of the conclusions are based on solid data. However, there are some loose ends, which should be addressed either experimentally or at least in the discussion.

1. The authors find two shortened Cfr isoforms that apparently arise from incorrect initiation on internal methionine codons on cfr mRNA (I26M, M95) and show that translation can start at an initiation codon upstream of M1. This is by itself an exciting story in the Cfr story. The regulatory mechanism due to an uORF is plausible and well documented in Figure 4- supplement 3 and 4. However, the appearance of a (native) internal initiation site starting from M95 is surprising. Given that this is a very prominent translation product (well over 50% with the native Pcfr, Figure 4 – supplement 4), this aspect of the paper raises questions. How does the initiation from M95 depend on antibiotics? The authors clearly show that the shorter isoform does not confer resistance and is rapidly degraded. So, what is the functional role of the shorter isoform? These are very exciting questions that should be addressed.

2. Directed evolution yielded protein isoforms that are better expressed and are more stable. The authors show that mutant mRNAs are shifted towards heavier polysomes and suggest that this may be due to the tRNA abundance effect (Figure 4 – supplement 1). However, this experiment lacks a control. If tRNA abundance is crucial, overexpression of the tRNA decoding the wt AAU codon should also increase protein expression. There is a rich literature (cited on p. 21 of the discussion) that may suggest alternative explanations. The authors should identify which of these suggestions can be excluded based on their data and clearly describe those that could contribute to the observed expression effect.

3. The increased stability of the mutated Cfr variants is fascinating, but not well addressed. What causes the increase in protein stability? What is actually the N-terminus of the protein: is it deformylated and the N-terminal Met cleaved as efficiently as the wt? Is there any correlation with the N-end rule?

4. Cfr methylates C8 of A2503. Mass spectra clearly distinguish the peaks of m2m8A2503 and m2A2503. However, this appears to rely on the assumption that A2503 is completely m2-modified. Has this been checked? What would be the mass of the unmodified fragment and was it identified in the spectra? Can the authors exclude the appearance of m8A2503 species?

My suggestion to address points 1-4 is the following.

1. The authors could control how much Cfr-short isoform (M95) is present in the absence of antibiotics and how it is affected upon antibiotic treatment. Figures 4-supplementary 3 and 4 are so interesting that I would suggest including them in the main text. The expression of Cfr short (M95) seems to depend on the N2 mutation. This should be quantified and explained.

2. An experiment with overexpression of tRNA decoding the wt AAU codon would provide an important test for the authors' suggestion that the expression differences are related to tRNA abundance. The text on p. 21, lines 6-19 should be modified, as it does not exactly state what was actually suggested, there are too many citations without clearly stating what these papers say.

3. Check the N-terminus of proteins from the gels by mass spectrometry. Add a few sentences in the Discussion as to how a (presumably N-terminal) Lys or Ile can increase protein stability.

4. Comment on the modification completeness of m2A2503 and provide the expected m/z value for the respective unmodified RNA fragment.

Reviewer #2:

The manuscript deals with a critical resistance mechanism relevant to eight families of antibiotics that inhibit ribosome activity. In this mechanism, the enzyme Cfr methylates the rRNA nucleotide 2503, which is part of the typical binding pocket of these antibiotics. By modifying the binding pocket architecture, binding of the drug is inhibited, and resistance to bacterial antibiotics is achieved. The authors aimed to decipher the resistance mechanism by structural studies, but the moderate Cfr enzyme activity of S. aureus hampered the ability to study such methylated ribosomes.

The authors decipher how antibiotic resistance can be increased by introducing mutations into the modifier enzyme Cfr by using a directed evolution approach. Using this approach, a high percentage of mutated ribosomes suitable for structural study were achieved. They discover that the mutated Cfr enzyme is not more efficient in methylation, but rather that the mutations alter active intercellular protein levels, and therefore higher methylation ratio is achieved. In addition, they found that the introduced mutations in Cfr enhance its translation efficiency as it is shown to be part of well-translated mRNAs, and this is the leading cause for higher methylation levels of the mutated Cfr. For some of their Cfr mutants, it is demonstrated that the increase in the expression of mutated Cfr is due to an improved translation that is mediated by tRNA abundance. The Cfr mutants show increased half-life, due to enhanced translation and decreased degradation. This part of the paper is the major technical advance.

The cryo-EM 2.2 A structure of an E. coli ribosome with a C8-modified 2503 rRNA nucleotide was obtained by using the mutated stable Cfr, which enables ~90% of 2503 methylation. Furthermore, by superimposing few published available antibiotics-ribosome complex structures on the current new mutated ribosome structure, the authors provide a valid explanation of the resistance mechanism. They demonstrate how the addition of a methyl group at the C8 position of A2503 will cause a steric clash with a bound drug and preclude its binding. This single structural change in the ribosome upon methylation may indeed be used for drug design for overcoming antibiotic resistance.

1. It should be mentioned that the cfr mutations gained using directed evolution are located in sites of the proteins which are distant from its active site. The most effective mutation (in residue 2) is in a separate domain from the active site. Actually, all the mutations are mainly at the surface of the Cfr. This is clear from Figure 1b and should be added.

2. Since the ribosome structure presented is of a modified E. coli ribosome and the cfr enzyme is originally from S. aureus the sequence alignment of Cfr from S. aureus and E. coli should be presented (as a supp.) and discussed.

How similar they are? Is it legitimate to use S. aureus cfr to modify E. coli ribosome? Apparently, it is since the authors present evidence for it but some text to relate to this point should be added. It will also widen the scope by generalizing the conclusions for gram-positive and gram-negative bacteria.

In the text, this point should be addressed in the introduction (For example: P. 4 line 9. Cfr from MRSA and in line 15 E. coli ribosome.) and in the Results section (For example: p.16 line 4 – mention S. aureus ribosome but the same page line 9 – the ribosomes of which the structure was determined are EC ones. In Figure 5 – add to the legend that this is an EC ribosome. In addition, please relate to this point in the discussion (p.20)).

3. The authors superimposed the complex structure of an archeal ribosome and linezolid and tiamulin for explaining the resistant mechanism. Resistant mechanism was not found in archea and this ribosome is less relevant for clinical aspects. More appropriate models to be used are of a bacterial ribosome in complex with these antibiotics, i.e. D. radiodurans ribosome in complex with tiamulin (PDBID 1XBP) and the pathogen S. aureus ribosome in complex with linezolid (PDBID 4WFA).

Reviewer #3:

In this study, Tsai et al. set out out to generate Cfr variants through directed evolution that are more effective in rRNA methylation and therefore resistance. The authors achieved this aim and exploited the novel variants to generate modified ribosomes that were suitable substrates for cryo-EM structural analysis. The paper is clear, and the data well presented. Directed evolution of Cfr was clearly an excellent strategy for generating near-stoichiometrically modified ribosomes. The evolved sites of Cfr are not very interesting in respect to protein function per se, as their effect seems mostly about increasing expression levels rather than altering enzymatic activity. However, these variants were clearly invaluable for enabling the authors to fulfil their aim of generating a high-resolution structure. By comparing this structure to other ribosome structures that include bound antibiotics, the authors show clearly how the modification is incompatible with antibiotic binding.

Overall, the paper is a significant contribution to our understanding of the mechanistic basic of resistance to a PhLOPS(A) antibiotics as conferred by the clinically relevant resistance factor Cfr. I expect the structure will be used by many in the ribosome field.

https://doi.org/10.7554/eLife.70017.sa1

Author response

Essential revisions:

The referees agreed that the experiments were carefully performed and interpreted. The primary experimental revision (point 1) relates to providing a suitable control that tRNA abundance determines the higher translational efficiency of the mutant Cfr constructs. There are also suggestions for further experiments that would strengthen the paper (points 2, 4, 5), but we feel this is optional and could be viewed as beyond the scope of this particular study. I therefore leave it to you whether you perform these additional experiments (which might well be interesting).

1. Directed evolution yielded protein isoforms that are better expressed and are more stable. The authors show that mutant mRNAs are shifted towards heavier polysomes and suggest that this may be due to the tRNA abundance effect (Figure 4 – supplement 1). However, this experiment lacks a control. If tRNA abundance is crucial, overexpression of the tRNA decoding the wt AAU codon should also increase protein expression.

The reviewer brings up an excellent point. To address it, we have carried out an in-depth investigation. To perform tRNA overexpression experiments, we cloned the tRNAAsn gene, which decodes the WT AAU codon, into a pBSTNAV vector. The pBSTNAV vector is commonly used for overexpression of tRNAs in E. coli and is driven by a strong, constitutive lpp promoter (Dégut et al., 2016). To ensure that the pBSTNAV_tRNAAsn vector was suitable for coexpression with the pZA-encoded Cfr vectors, we swapped the pBSTNAV ampicillin resistance marker for kanamycin. The pBSTNAV and pZA vectors are also ori-compatible. The sequence architecture of the construct is outlined in Author response image 1:

Author response image 1
pBSTNAV_tRNAAsn construct.

To perform co-expression experiments, E. coli BW25113 was co-transformed with pZA (Empty, CfrWT-FLAG, or CfrN2K-FLAG) and pBSTNAV (Empty or tRNAAsn). We confirmed tRNA overexpression by northern blot (Author response image 2) . We designed the 5’-biotinylated DNA oligo probe (purchased from IDT) to be complementary to the region of tRNAAsn containing the anticodon stem loop. The sequence is as follows: 5’-Biotin-GATTAACAGTCCGCCGTTCTACCGAC. Of note, we observe additional bands in the tRNA overexpression samples, likely attributed to pre-processed or truncated tRNAAsn.

Author response image 2
Confirmation of tRNAAsn overexpression.

Trizol-extracted total RNA was extracted from bacterial lysates and separated on a denaturing 10% acrylamide 8M TBE-Urea gel. Em = empty vector and WT = CfrWT-FLAG. (a) RNA was stained with SYBR-safe. (b) For detection of tRNAAsn by northern blot, RNA was transferred onto a Hybond N+ nylon membrane (GE Healthcare) and crosslinked to the membrane using a Stratalink 1800 UV Crosslinker. After blocking for 2.5 h at 40oC in hybridization buffer (250 mM sodium phosphate buffer-pH 7.0, 1 mM EDTA, 7% (w/v) SDS, 0.5% (w/v) BSA, 4 µg/mL salmon sperm DNA), the membrane was incubated with denatured tRNA probe overnight at 40oC in hybridization buffer. After washing the membrane 3 X 15 min in 1X SSC, 0.1% SDS, tRNAAsn was detected using the Thermo Chemiluminescent Nucleic Acid Detection Module following manufacturer’s instructions.

To test how overexpression of tRNAAsn impacts Cfr protein levels, co-transformants were grown to mid-log phase in LB media containing selection antibiotics and AHT inducer (30 ng/mL). Cfr protein levels were detected by immunoblotting against the FLAG tag as described in materials and methods. While performing the experiment, it became evident that co-expression of tRNAAsn caused significant fitness defects. Cultures that co-expressed tRNAAsn required 20 min longer (2h 5 min vs. 1h 45 min) to reach mid-log phase than those expressing an empty pBSTNAV vector. This result is also evident in the western outlined in Author response image 3, where overexpression of tRNAAsn causes decreased expression of both CfrWT and CfrN2K (Author response image 3) .

Author response image 3
Overexpression of tRNAAsn decreases Cfr protein levels.

Cfr protein was detected by immunoblotting against the FLAG tag and RNA polymerase β subunit was used as the loading control. Em = empty vector, WT = CfrWT-FLAG, and N2K = CfrN2K-FLAG.

This result suggests that overexpression of tRNA alone may not be sufficient for proper tRNA production and modulation of protein synthesis. Maturation of tRNAAsn requires trimming of 3’ and 5’ pre-tRNA sequence by RNases, post-transcriptional modification by various enzymes, and charging by asparaginyl-tRNA synthetase (Shepherd and Ibba, 2015). Thus, overexpression tRNAAsn may overwhelm endogenous tRNA maturation machinery, leading to accumulation of potentially toxic pre-processed intermediates and partially degraded tRNAAsn (Author response image 2b) , along with uncharged tRNAAsn. Additionally, if there did exist an increased level of charged Asn-tRNAAsn in our overexpression system, previous work suggests that disrupting the balance of the charged tRNA pool competing for the ribosomal A-site negatively impacts the fidelity of protein synthesis (Kramer and Farabaugh, 2007). Together, these results indicate that while in theory overexpression of tRNAAsn would allow additional testing of our model, the fitness defects associated with this experimental setup makes interpretation of results challenging.

However, we would like to point out that while we were unable to increase WT expression by boosting levels of tRNAAsn, we have demonstrated that mutating the N2I directed evolution codon (AUU) to the Ile codon decoded by a rare tRNAIle (AUA) does diminish protein levels by ~2-fold (Figure 4 —figure supplement 2). This result suggests that tRNA abundance of the second codon may be a contributor to improved Cfr expression, and we state this in the text as follows:

“The isoleucine AUA codon is decoded by the low-abundant tRNAIle2 (Del Tito et al. 1995; Nakamura, Gojobori, and Ikemura 2000). Mutation of N2I from AUU to the AUA rare codon resulted in a ~2-fold decrease in Cfr expression, supporting tRNA abundance as a contributing factor (Figure 4 —figure supplement 2).”

2. The authors find two shortened Cfr isoforms that apparently arise from incorrect initiation on internal methionine codons on cfr mRNA (I26M, M95) and show that translation can start at an initiation codon upstream of M1. This is by itself an exciting story in the Cfr story. The regulatory mechanism due to an uORF is plausible and well documented in Figure 4- supplement 3 and 4. However, the appearance of a (native) internal initiation site starting from M95 is surprising. Given that this is a very prominent translation product (well over 50% with the native Pcfr, Figure 4 – supplement 4), this aspect of the paper raises questions. How does the initiation from M95 depend on antibiotics? The authors clearly show that the shorter isoform does not confer resistance and is rapidly degraded. So, what is the functional role of the shorter isoform? To address these questions the authors could control how much Cfr-short isoform (M95) is present in the absence of antibiotics and how it is affected upon antibiotic treatment. Figures 4-supplementary 3 and 4 are so interesting that I would suggest including them in the main text. The expression of Cfr short (M95) seems to depend on the N2 mutation. This should be quantified and explained.

To address how expression of the truncated product may be impacted by antibiotics, we expressed CfrWT using the Ptet or Pcfr promoter in the presence of sub-inhibitory concentrations of chloramphenicol. Similar results were also obtained with the florfenicol.

Interestingly, we find that levels of the truncated Cfr protein are similar or increased upon incubation with chloramphenicol for both promoter constructs (Author response image 4a, b). While this result is intriguing, it is likely that either increased plasmid copy number or increased Cfr mRNA levels are responsible for this result, as a similar trend is also observed for full-length Cfr. It has been well documented that inhibition of protein synthesis can increase plasmid yields (Begbie et al., 2005; Sambrrok et al., 1989). Furthermore, increased production of the truncation in this context does not impact resistance, as incubation with chloramphenicol increases resistance slightly, likely due to the marginally improved production of full-length Cfr (Author response image 4) . Thus, the functional role of the shorter Cfr isoform remains elusive.

Author response image 4
Impact of antibiotics on expression of full-length and truncated Cfr protein.

E. coli BW25113 were transformed with pZA constructs containing either CfrWT expressed using the Ptet (WT) or native Pcfr promoter (SCFS1) and plated on LB agar plates containing selection antibiotic and either 0, 0.5, 1, or 2 µg/mL of chloramphenicol (CHL). Overnight cultures were diluted into fresh media containing the respective concentration of CHL and grown to mid-log phase. (a) Cfr protein levels were assessed by immunoblotting against a C-terminal FLAG tag. Asterisks denote the truncated Cfr product. Em = empty vector control. (b) Relative protein expression of full-length Cfr (teal) and the truncation corresponding to translation initiation at Met95 (yellow) compared to CfrWT expression with no CHL. Signal was normalized to housekeeping protein RNA polymerase β-subunit. (c) Minimum inhibitory concentration of CHL required to inhibit bacterial growth of E. coli BW25113 transformants that had been pre-incubated with either 0, 0.5, 1, or 2 µg/mL of CHL.

To address how production of the truncation is impacted by Cfr mutations, we have also quantified expression of the truncation in both Cfr variants and in N2 mutants. For Cfr variants containing N2 mutations (CfrV1-5), we observe higher levels of the M95 truncation compared to CfrWT (~2-15 fold, new Figure 2 —figure supplement 4). However, higher expression of the truncation is likely explained by the increased transcript levels for these variants (~1.5-3 fold, Figure 2c). It is well documented that improved translation often results in improved mRNA stability due to increased ribosome occupancy and shielding of the transcript from degradation machinery (Braun et al., 1998; Deana and Belasco, 2005; Edri and Tuller, 2014). In our work, for example, the ~3 fold increase in the CfrV4 transcript level (Figure 2c) is likely explained by its dramatically improved translation observed in polysome analysis (Figure 4d, Figure 4 —figure supplement 1c).

For the N2 mutations alone, we observe higher levels of the M95 truncation for N2K mutants (encoded by AAA or AAG, ~2.5 and ~5 fold respectively, new Figure 3 —figure supplement 1, new Figure 4 —figure supplement 2). Analogous to the Cfr variants, higher expression of the truncation is likely explained by the increased transcript levels of N2K caused by improved translation (Figure 4c). Given that N2I (encoded by AUU or AUA) is not as robustly expressed, and likely not as well translated, the transcriptional impact would be less pronounced, resulting in truncation levels similar to CfrWT (new Figure 3 —figure supplement 1, new Figure 4 —figure supplement 2).

We have edited the text accordingly to incorporate these observations in the Results section:

“The truncations result from translation initiation at internal methionines but do not contribute to resistance (Figure 2 —figure supplement 3), indicating that they are non-functional enzymes unable to methylate A2503. The smaller molecular weight truncation is present in higher levels for all Cfr variants compared to CfrfWT (Figure 2 —figure supplement 4). […] Similarly, to the evolved variants, in addition to the full-length Cfr protein, we also observe expression of the truncation that results from initiation at M95 (Figure 3 —figure supplement 1)”

And in the Discussion section:

“Investigations into expression levels of CfrWT and its respective mutants revealed that, in addition to full-length protein, a smaller Cfr isoform of ~30 kDa is also produced (Figure 2b, Figure 3a). […] Thus, while the truncated product does not contribute to resistance, the potential function of the smaller protein remains elusive and requires further study.”

3. There is a rich literature (cited on p. 21 of the discussion) that may suggest alternative explanations. The authors should identify which of these suggestions can be excluded based on their data and clearly describe those that could contribute to the observed expression effect.

The text on p. 21, lines 6-19 should be modified, as it does not exactly state what was actually suggested, there are too many citations without clearly stating what these papers say.

We have edited the discussion to more thoroughly address the potential mechanisms by which N2K may influence translation. The text now reads as follows:

“Of the mutations investigated, N2K is the largest contributor to enhanced Cfr expression and resistance. Although N2K contributes to cellular stability, our results suggest that improved Cfr translation is the dominant role of this mutation. […] Interestingly, the observed internal translation start sites (I26M, M95) that are responsible for producing Cfr truncations (Figure 2b, Figure 2 —figure supplement 3) contain a lysine immediately after methionine, further highlighting the putative role for lysine codons in early steps of translation.”

4. The increased stability of the mutated Cfr variants is fascinating, but not well addressed. What causes the increase in protein stability? What is actually the N-terminus of the protein: is it deformylated and the N-terminal Met cleaved as efficiently as the wt? Is there any correlation with the N-end rule? One way to answer these questions would be to check the N-terminus of proteins from the gels by mass spectrometry. Add a few sentences in the Discussion as to how a (presumably N-terminal) Lys or Ile can increase protein stability.

To address reviewer’s comments, we performed IP-MS/MS analysis of CfrWT and evolved variant CfrV4 which contains the N2K mutation. In brief, we expressed C-terminally FLAG-tagged CfrWT and CfrV4 in E. coli and harvested cells at mid-log phase. Cfr protein was immunoprecipitated using anti-FLAG M2 affinity gel (Sigma) from cell lysate under anaerobic conditions. The protein band corresponding to full-length Cfr was excised from the SDS-PAGE gel and subjected to trypsin digestion and subsequent MS/MS analysis. Sequence coverage for CfrWT and CfrV4 are displayed in Author response images 5 and 6.

For CfrWT, we did not detect N-terminal (Nt) peptides with fMet or peptides lacking Met1, suggesting that the Nt Met is retained but deformylated. This result is in line with previous biochemical work demonstrating that residues at the P1’ position larger than valine (V), such as asparagine (N), are disfavored substrates for methionine aminopeptidase (Frottin et al., 2006; Hirel et al., 1989; Xiao et al., 2010)

Due to high lysine content at the N-terminus of CfrV4, we were unfortunately unable to detect Nt peptides for the variant protein. However, previous biochemical work has demonstrated that basic side-chains such as lysine (K) at position 2 are slightly favored substrates for peptide deformylase (Ragusa et al., 1999) and strongly disfavored for methionine aminopeptidase (Frottin et al., 2006; Hirel et al., 1989). Similarly, nascent peptides with isoleucine (I) at position 2 are also likely to be efficiency deformylated and resistance to Met excision (Frottin et al., 2006; Hirel et al., 1989; Xiao et al., 2010).

Author response image 5
MS/MS peptide coverage for CfrWT.
Author response image 6
MS/MS peptide coverage for CfrV4.

Although we cannot completely rule out its involvement in stability, these preliminary results and existing biochemical evidence suggest that differential co-translational processing of the N-terminus, which would invoke ”classic” N-end rule or fMet/N-degron pathways (Izert et al., 2021), is likely not responsible for the observed improved stability of N2K/I mutants.

Although the precise mechanism by which N2K/I improve Cfr stability remains elusive, these mutants could alter recognition by other enzymes that may be important in regulating degradation, such as endopeptidases or L/F-tRNA-protein transferase (LFTR). For example, it was discovered that LFTR can add primary destabilizing residues to Nt Met, creating an N-degron (Dougan et al., 2012), but its efficiency can depend on the identity of the second amino acid (Ottofuelling et al., 2021).

We have modified the text to include discussion on how N2K/I may be involved in protein stability, which now reads as follows.

“We also observe modest improvements in protein stability with N2K/I mutants (Figure 4e). In bacteria, the identity of N-terminal residues are important determinants of degradation through N-degron pathways (Dougan, Micevski, and Truscott 2012; Tobias et al. 1991). During protein synthesis, the N-terminus is co-translationally processed by two enzymes, peptide deformylase to remove the formyl group from Met (fM) and methionine aminopeptidase (Koubek et al. 2021). Based on previous biochemical work, it is unlikely that CfrWT and CfrN2K/I would have different N-terminal processing, since fMN… and fMK/I… are likely to be efficiently de-formylated (Ragusa et al. 1999) and resistant to methionine excision (Hirel et al. 1989; Frottin et al. 2006; Xiao et al. 2010). Although the precise mechanism by which N2K/I improves Cfr stability remains elusive, these mutations may alter recognition by other enzymes important for degradation, such as endopeptidases or L/F-tRNA-protein transferase (Izert, Klimecka, and Górna 2021; Ottofuelling et al. 2021).”

5. Cfr methylates C8 of A2503. Mass spectra clearly distinguish the peaks of m2m8A2503 and m2A2503. However, this appears to rely on the assumption that A2503 is completely m2-modified. Has this been checked? What would be the mass of the unmodified fragment and was it identified in the spectra? Can the authors exclude the appearance of m8A2503 species? Please comment on the modification completeness of m2A2503 and provide the expected m/z value for the respective unmodified RNA fragment.

The RNA fragment containing unmodified A2503 (AΨG) has a predicted m/z value of 999.14 (now included in Figure 2 —figure supplement 1). The mass of the unmodified fragment overlaps with an unrelated ACG fragment which has a predicted m/z value of 998.16 (see detailed m/z predictions obtained from ChemDraw, Author response image 7). Since the mass of the unmodified A2503 is near-identical to the mass + 1 isotope of the unrelated fragment (999.14 and 999.16, respectively), it cannot be accurately determined if unmodified A2503 is present our MALDI-MS samples. However, the predicted isotope abundance for the unrelated fragment (37%) is qualitatively similar in spectra obtained from wild type E. coli with endogenous RlmN only (Author response image 8) and suggests little-to-no presence of the unmodified AΨG species. Although we cannot conclude definitively, these results suggest that RlmN achieves complete or near-complete C2 methylation of A2503. Relative abundance of the isotope peak is also consistent in samples where CfrWT or Cfr variants were expressed, suggesting that there is likely little/no unmodified A2503 present in these samples either.

Author response image 7
m/z predictions obtained from ChemDraw for rRNA fragments derived from oligo-protection MALDI-TOF mass spectrometry.

The predicted m/z and abundance for the mass + 1 isotope peak is labeled in green.

Author response image 8
Annotated MALDI spectra for empty vector control, CfrWT/V4/V7.

The mass peak (m) is labeled in blue, while the isotope peak (m+1) and predicted relative abundance derived from ChemDraw is labeled in green.

While the monomethylated mass of 1013 could also correspond to m8A2503, this scenario is highly unlikely. This is because m8 modification would induce strong antibiotic resistance, as shown by our data, which is not what we observed in samples where 1013 predominates. Additionally, our previous work indicates that RlmN methylates A2503 in early stages of ribosome biogenesis.

6. Since the ribosome structure presented is of a modified E. coli ribosome and the cfr enzyme is originally from S. aureus, the sequence alignment of Cfr from S. aureus and E. coli should be presented (as a Supplement). The authors should discuss whether it is legitimate to use S. aureus cfr to modify E. coli ribosome. This would widen the scope of the paper by generalizing the conclusions for gram-positive and gram-negative bacteria. This could be addressed in the introduction (e.g., p. 4 line 9 and line 15) and in the Results section (e.g., p.16 line 4 and line 9). In Figure 5 – add to the legend that this is an E. coli ribosome. In addition, please relate to this point in the discussion (p. 20).

The cfr gene recovered from animal-derived E. coli isolates (Deng et al., 2014; Liu et al., 2017; Ma et al., 2021; Wang et al., 2012) is identical in sequence to the clinical S. aureus isolate used in this study. Conservation of the sequence, along with the observation that this Cfr sequence confers resistance in veterinary E. coli isolates, validates the use of E. coli as a model organism for our studies.

We have modified the text to clarify this point in the introduction accordingly:

“Since then, the cfr gene has been identified across the globe in both gram-positive and gram-negative bacteria, including E. coli (Shen, Wang, and Schwarz 2013; Vester 2018)…”

Although Cfr is predominantly found in Staphylococcal species, the 23S rRNA sequence is highly conserved between E. coli and S. aureus (Eyal et al., 2015). Specifically, the sequence and structure of the PTC is highly conserved, which we have now illustrated by including a structural overlay of the E. coli Cfr-modified ribosome and wild-type S. aureus ribosome as Figure 5 —figure supplement 3.

We have also added the following text within the Results section as follows:

“Although Cfr has been identified in animal-derived E. coli isolates (Wang et al. 2012; Deng et al. 2014; Liu et al. 2017; Ma et al. 2021), the resistance gene has primarily been identified clinically in staphylococcal organisms such as S. aureus (Vester 2018). However, given the high sequence and structural conservation within the PTC region (Figure 5 —figure supplement 3), structural impacts of the Cfr m8A2503 modification within E. coli and S. aureus ribosomes are likely conserved.”

7. The authors superimposed the complex structure of an archeal ribosome and linezolid and tiamulin for explaining the resistance mechanism. However, this resistance mechanism was not found in archaea and thus these ribosomes are less relevant for clinical aspects. More appropriate models to be used are of bacterial ribosomes in complex with these antibiotics, i.e. D. radiodurans ribosome in complex with tiamulin (PDBID 1XBP) and the pathogen S. aureus ribosome in complex with linezolid (PDBID 4WFA).

Previous SAR work has demonstrated that the C5 group of linezolid is likely responsible for steric collision with m8A2503. However, in the three published structures, the C5 group is modeled in dramatically different conformations (outlined in Author response image 9). Although the structure of the S. aureus ribosome in complex with linezolid (PDB: 4WFA, purple) represents a more biologically-relevant organism, the modeled conformation of the C5 group, for which there is weak density, does not sterically clash with m8A2503. For the remaining structures (PDB: 3CPW in yellow and PDB: 3DLL in pink), while both modeled C5 conformations would sterically clash with m8A2503, 3CPW exhibits superior density for the C5 moiety. We also recently obtained a structure of a linezolid-bound E. coli ribosome (Tsai et al., 2021), and the C5 group is best modeled in a conformation near-identical to 3CPW. Together, these results, combined with the observation that Cfr confers antibiotic resistance in E. coli, support our decision to use the H. marismortui ribosome (PDB: 3CPW) for structural comparison.

Author response image 9

A2503 is modeled in the syn-conformation in the H. marismortui ribosome-tiamulin complex (PDB: 3G4S, purple), the same conformation as m8A2503 in the Cfr-modified ribosome. A2503 is modeled as anti- in the D. radiodurans structure (PDB: 1XBP, yellow). Thus, the archaeal structure was chosen to provide more straight-forward structural comparison. Superposition of both tiamulin-ribosome complexes reveals that m8A2503 sterically clashes with the C10 and C11 substituents of tiamulin in both cases.

Author response image 10

8. The paper would benefit from an analysis of the natural variability of the evolved sites. It is mentioned that clinical isolates do not have these exact substitutions, but a supplemental multiple sequence alignment of homologues with these positions marked would anyway be interesting. For example, an N2K variant appears in the results of a quick blastp search in a Bacillus Cfr.

We analyzed sequences of Cfr homologues with less than 80% sequence identity and did discover that directed evolution mutations are recapitulated in these sequences. We have added a supplementary figure and a section to the discussion which reads as follows:

“Interestingly, mutations obtained through directed evolution have been observed in Cfr homologues that share less than 80% sequence identity with Cfr. Specifically, methionine (M) at position 26 is observed for the functionally characterized Cfr homologues Cfr(B) (Deshpande et al. 2015; Marín et al. 2015; Hansen and Vester 2015) and Cfr(D) (Pang et al. 2020), which have been recovered from human-derived isolates and share 74% and 64% amino acid identity with Cfr, respectively (Schwarz et al. 2021) (Figure 4 —figure supplement 5). We also observe lysine (K) at position 2, methionine (M) at position 26, and glycine (G) at position 39, akin to N2K, I26M, and S39G mutations, for a number of Cfr homologues that clade with functional Cfr or Cfr-like genes (Stojković et al. 2019). While the precise roles of these residues within less-well characterized and more distantly related Cfr proteins requires further study, these observations indicate that directed evolution accessed sequence space that is already being exploited by proteins that are, or are hypothesized to be, functional Cfr resistance enzymes.”

9. Another remaining question is how the novel variants affects susceptibilities to ribosome-targeting antibiotics in other structural classes, including nucleoside analogues and macrolides. In the introduction the authors mention that in addition to PhLOPSA antibiotics, Cfr can confer resistance to nucleoside analog A201A, hygromycin A, and 16-membered macrolides. However, this is not returned to in the paper. MIC experiments with an expanded selection of antibiotics would be really nice to see if possible. Similarly, does the structure explain the previously observed cross-resistance that goes beyond PhLOPS(A)?

We agree with the reviewer and performed MIC testing of hygromycin A against E. coli BW25113 lacking efflux component acrB (Figure 1 —figure supplement 1). Analogous to PhLOPSA antibiotics, Cfr variants confer higher resistance to hygromycin A than CfrWT (4-fold higher resistance for CfrV3 and 8-fold higher resistance for CfrV7).

We also performed additional structural overlays of the Cfr-modified ribosome with existing structures of ribosomes in complex with hygromycin A, nucleoside analog A201A, and 16-membered macrolides tylosin and josamycin (Figure 5 —figure supplement 4). As expected, the Cfr modification introduces a steric clash between the C8 methylation mark and the antibiotics, providing molecular rationale for how Cfr confers resistance to, in total, 8 classes of ribosome-targeting antibiotics.

References

Begbie S, Dominelli G, Duthie K, Holland J, Jitratkosol M. 2005. The effects of sub-inhibitory levels of chloramphenicol on pBR322 plasmid copy number in Escherichia coli DH5α cells. J Exp Microbiol Immunol 7:82–88.

Braun F, Le Derout J, Régnier P. 1998. Ribosomes inhibit an RNase E cleavage which induces the decay of the rpsO mRNA of Escherichia coli. EMBO J 17:4790–4797.

Deana A, Belasco JG. 2005. Lost in translation: the influence of ribosomes on bacterial mRNA decay. Genes Dev 19:2526–2533.

Dégut C, Monod A, Brachet F, Crépin T, Tisné C. 2016. in vitro/in vivo Production of tRNA for X-Ray Studies. Methods Mol Biol 1320:37–57.

Deng H, Sun J, Ma J, Li L, Fang L-X, Zhang Q, Liu Y-H, Liao X-P. 2014. Identification of the multi-resistance gene cfr in Escherichia coli isolates of animal origin. PLoS One 9:e102378.

Dougan DA, Micevski D, Truscott KN. 2012. The N-end rule pathway: From recognition by N-recognins, to destruction by AAA proteases. Biochimica et Biophysica Acta (BBA) – Molecular Cell Research. doi:10.1016/j.bbamcr.2011.07.002

Edri S, Tuller T. 2014. Quantifying the effect of ribosomal density on mRNA stability. PLoS One 9:e102308.

Eyal Z, Matzov D, Krupkin M, Wekselman I, Paukner S, Zimmerman E, Rozenberg H, Bashan A, Yonath A. 2015. Structural insights into species-specific features of the ribosome from the pathogen Staphylococcus aureus. Proc Natl Acad Sci U S A 112:E5805–14.

Frottin F, Martinez A, Peynot P, Mitra S, Holz RC, Giglione C, Meinnel T. 2006. The Proteomics of N-terminal Methionine Cleavage* S. Mol Cell Proteomics 5:2336–2349.

Hirel PH, Schmitter MJ, Dessen P, Fayat G, Blanquet S. 1989. Extent of N-terminal methionine excision from Escherichia coli proteins is governed by the side-chain length of the penultimate amino acid. Proc Natl Acad Sci U S A 86:8247–8251.

Izert MA, Klimecka MM, Górna MW. 2021. Applications of Bacterial Degrons and Degraders — Toward Targeted Protein Degradation in Bacteria. Frontiers in Molecular Biosciences. doi:10.3389/fmolb.2021.669762

Kramer EB, Farabaugh PJ. 2007. The frequency of translational misreading errors in E. coli is largely determined by tRNA competition. RNA 13:87–96.

Liu X-Q, Wang J, Li W, Zhao L-Q, Lu Y, Liu J-H, Zeng Z-L. 2017. Distribution of cfr in Staphylococcus spp. and Escherichia coli Strains from Pig Farms in China and Characterization of a Novel cfr-Carrying F43:A-:B- Plasmid. Frontiers in Microbiology. doi:10.3389/fmicb.2017.00329

Ma Z, Liu J, Chen L, Liu X, Xiong W, Liu J-H, Zeng Z. 2021. Rapid Increase in the IS26-Mediated cfr Gene in E. coli Isolates with IncP and IncX4 Plasmids and Co-Existing cfr and mcr-1 Genes in a Swine Farm. Pathogens 10. doi:10.3390/pathogens10010033

Ottofuelling RD, Ninnis RL, Truscott KN, Dougan DA. 2021. Novel modification by L/F-tRNA-protein transferase (LFTR) generates a Leu/N-degron ligand in Escherichia coli. bioRxiv.

Ragusa S, Mouchet P, Lazennec C, Dive V, Meinnel T. 1999. Substrate recognition and selectivity of peptide deformylase. Similarities and differences with metzincins and thermolysin. J Mol Biol 289:1445–1457.

Sambrrok J, Fritsch EF, Maniatis T. 1989. Molecular cloning a laboratory manual 2nd edition.

Shepherd J, Ibba M. 2015. Bacterial transfer RNAs. FEMS Microbiol Rev 39:280–300.

Tsai K, Stojković V, John Lee D, Young ID, Szal T, Vazquez-Laslop N, Mankin AS, Fraser JS, Fujimori DG. 2021. Structural basis for context-specific inhibition of translation by oxazolidinone antibiotics. bioRxiv. doi:10.1101/2021.08.10.455846

Wang Y, He T, Schwarz S, Zhou D, Shen Z, Wu C, Wang Y, Ma L, Zhang Q, Shen J. 2012. Detection of the staphylococcal multiresistance gene cfr in Escherichia coli of domestic-animal origin. J Antimicrob Chemother 67:1094–1098.

Xiao Q, Zhang F, Nacev BA, Liu JO, Pei D. 2010. Protein N-terminal processing: substrate specificity of Escherichia coli and human methionine aminopeptidases. Biochemistry 49:5588–5599.

https://doi.org/10.7554/eLife.70017.sa2

Article and author information

Author details

  1. Kaitlyn Tsai

    Department of Cellular and Molecular Pharmacology, University of California San Francisco, San Francisco, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Visualization, Writing – original draft, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0605-2720
  2. Vanja Stojković

    Department of Cellular and Molecular Pharmacology, University of California San Francisco, San Francisco, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9885-3428
  3. Lianet Noda-Garcia

    Department of Biomolecular Sciences, Weizmann Institute of Science, Rehovot, Israel
    Contribution
    Data curation, Formal analysis, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
  4. Iris D Young

    Department of Bioengineering and Therapeutic Sciences, University of California San Francisco, San Francisco, United States
    Contribution
    Data curation, Formal analysis, Writing – original draft, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4713-9504
  5. Alexander G Myasnikov

    Department of Biochemistry and Biophysics, University of California San Francisco, San Francisco, United States
    Contribution
    Data curation, Formal analysis, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
  6. Jordan Kleinman

    Department of Cellular and Molecular Pharmacology, University of California San Francisco, San Francisco, United States
    Contribution
    Data curation, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7146-824X
  7. Ali Palla

    Department of Cellular and Molecular Pharmacology, University of California San Francisco, San Francisco, United States
    Contribution
    Data curation, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7380-3881
  8. Stephen N Floor

    1. Helen Diller Family Comprehensive Cancer Center, University of California San Francisco, San Francisco, United States
    2. Department of Cell and Tissue Biology, University of California San Francisco, San Francisco, United States
    Contribution
    Formal analysis, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9965-9694
  9. Adam Frost

    1. Department of Biochemistry and Biophysics, University of California San Francisco, San Francisco, United States
    2. Quantitative Biosciences Institute, University of California San Francisco, San Francisco, United States
    Contribution
    Formal analysis, Methodology, Resources, Supervision, Writing – review and editing
    Competing interests
    Reviewing editor, eLife
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2231-2577
  10. James S Fraser

    1. Department of Bioengineering and Therapeutic Sciences, University of California San Francisco, San Francisco, United States
    2. Quantitative Biosciences Institute, University of California San Francisco, San Francisco, United States
    Contribution
    Data curation, Formal analysis, Resources, Supervision, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5080-2859
  11. Dan S Tawfik

    Department of Biomolecular Sciences, Weizmann Institute of Science, Rehovot, Israel
    Contribution
    Formal analysis, Funding acquisition, Methodology, Resources, Supervision, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5914-8240
  12. Danica Galonić Fujimori

    1. Department of Cellular and Molecular Pharmacology, University of California San Francisco, San Francisco, United States
    2. Quantitative Biosciences Institute, University of California San Francisco, San Francisco, United States
    3. Department of Pharmaceutical Chemistry, University of California San Francisco, San Francisco, United States
    Contribution
    Conceptualization, Formal analysis, Funding acquisition, Project administration, Resources, Supervision, Writing – review and editing
    For correspondence
    danica.fujimori@ucsf.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4066-9417

Funding

National Institute of Allergy and Infectious Diseases (R01AI137270)

  • Danica Galonić Fujimori

National Science Foundation (1650113)

  • Kaitlyn Tsai

University of California, San Francisco

  • Kaitlyn Tsai

National Institute of General Medical Sciences (F32GM133129)

  • Iris D Young

National Institute of General Medical Sciences (R01GM123159)

  • James S Fraser

Sanghvi-Agarwal Innovation Award

  • James S Fraser

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

The authors thank members of the Fujimori lab for discussion and comments on the manuscript. The authors thank the UCSF Center for Advanced CryoEM, which is supported by the National Institutes of Health (S10OD020054 and 1S10OD021741) and the Howard Hughes Medical Institute (HHMI). The authors thank Professor Kim Lewis and Dr. Yu Imai of Northeastern University for kindly providing hygromycin A.

Senior Editor

  1. Philip A Cole, Harvard Medical School, United States

Reviewing Editor

  1. Marina V Rodnina, Max Planck Institute for Biophysical Chemistry, Germany

Version history

  1. Received: May 4, 2021
  2. Accepted: November 25, 2021
  3. Version of Record published: January 11, 2022 (version 1)

Copyright

© 2022, Tsai et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,534
    Page views
  • 239
    Downloads
  • 6
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Kaitlyn Tsai
  2. Vanja Stojković
  3. Lianet Noda-Garcia
  4. Iris D Young
  5. Alexander G Myasnikov
  6. Jordan Kleinman
  7. Ali Palla
  8. Stephen N Floor
  9. Adam Frost
  10. James S Fraser
  11. Dan S Tawfik
  12. Danica Galonić Fujimori
(2022)
Directed evolution of the rRNA methylating enzyme Cfr reveals molecular basis of antibiotic resistance
eLife 11:e70017.
https://doi.org/10.7554/eLife.70017

Further reading

    1. Biochemistry and Chemical Biology
    2. Cell Biology
    Maria Körner, Susanne R Meyer ... Alexander Buchberger
    Research Article Updated

    The ATPase p97 (also known as VCP, Cdc48) has crucial functions in a variety of important cellular processes such as protein quality control, organellar homeostasis, and DNA damage repair, and its de-regulation is linked to neuromuscular diseases and cancer. p97 is tightly controlled by numerous regulatory cofactors, but the full range and function of the p97–cofactor network is unknown. Here, we identify the hitherto uncharacterized FAM104 proteins as a conserved family of p97 interactors. The two human family members VCP nuclear cofactor family member 1 and 2 (VCF1/2) bind p97 directly via a novel, alpha-helical motif and associate with p97-UFD1-NPL4 and p97-UBXN2B complexes in cells. VCF1/2 localize to the nucleus and promote the nuclear import of p97. Loss of VCF1/2 results in reduced nuclear p97 levels, slow growth, and hypersensitivity to chemical inhibition of p97 in the absence and presence of DNA damage, suggesting that FAM104 proteins are critical regulators of nuclear p97 functions.

    1. Biochemistry and Chemical Biology
    2. Epidemiology and Global Health
    Takashi Sasaki, Yoshinori Nishimoto ... Yasumichi Arai
    Research Article

    Background: High levels of circulating adiponectin are associated with increased insulin sensitivity, low prevalence of diabetes, and low body mass index (BMI); however, high levels of circulating adiponectin are also associated with increased mortality in the 60-70 age group. In this study, we aimed to clarify factors associated with circulating high-molecular-weight (cHMW) adiponectin levels and their association with mortality in the very old (85-89 years old) and centenarians.

    Methods: The study included 812 (women: 84.4%) for centenarians and 1,498 (women: 51.7%) for the very old. The genomic DNA sequence data were obtained by whole genome sequencing or DNA microarray-imputation methods. LASSO and multivariate regression analyses were used to evaluate cHMW adiponectin characteristics and associated factors. All-cause mortality was analyzed in three quantile groups of cHMW adiponectin levels using Cox regression.

    Results: The cHMW adiponectin levels were increased significantly beyond 100 years of age, were negatively associated with diabetes prevalence, and were associated with SNVs in CDH13 (p = 2.21 × 10-22) and ADIPOQ (p = 5.72 × 10-7). Multivariate regression analysis revealed that genetic variants, BMI, and high-density lipoprotein cholesterol (HDLC) were the main factors associated with cHMW adiponectin levels in the very old, whereas the BMI showed no association in centenarians. The hazard ratios for all-cause mortality in the intermediate and high cHMW adiponectin groups in very old men were significantly higher rather than those for all-cause mortality in the low level cHMW adiponectin group, even after adjustment with BMI. In contrast, the hazard ratios for all-cause mortality were significantly higher for high cHMW adiponectin groups in very old women, but were not significant after adjustment with BMI.

    Conclusions: cHMW adiponectin levels increased with age until centenarians, and the contribution of known major factors associated with cHMW adiponectin levels, including BMI and HDLC, varies with age, suggesting that its physiological significance also varies with age in the oldest old.

    Funding: This study was supported by grants from the Ministry of Health, Welfare, and Labour for the Scientific Research Projects for Longevity; a Grant-in-Aid for Scientific Research (No 21590775, 24590898, 15KT0009, 18H03055, 20K20409, 20K07792, 23H03337) from the Japan Society for the Promotion of Science; Keio University Global Research Institute (KGRI), Kanagawa Institute of Industrial Science and Technology (KISTEC), Japan Science and Technology Agency (JST) Research Complex Program 'Tonomachi Research Complex' Wellbeing Research Campus: Creating new values through technological and social innovation (JP15667051), the Program for an Integrated Database of Clinical and Genomic Information from the Japan Agency for Medical Research and Development (No. 16kk0205009h001, 17jm0210051h0001, 19dk0207045h0001); the medical-welfare-food-agriculture collaborative consortium project from the Japan Ministry of Agriculture, Forestry, and Fisheries; and the Biobank Japan Program from the Ministry of Education, Culture, Sports, and Technology.