MiaA-mediated tRNA modifications couple tryptophan attenuation and changes in tRNA abundance to complex phenotypes in Pseudomonas aeruginosa

Abstract

Transfer RNA (tRNA)-modifying enzymes are emerging as key regulators of bacterial physiology. MiaA, a tRNA isopentenyltransferase, is well studied in model organisms, but its role in the opportunistic pathogen Pseudomonas aeruginosa remains unclear. Using LC–MS, nanopore tRNA sequencing, as well as transcriptional, translational, and proteomic profiling, we mapped MiaA-dependent tRNA modifications and revealed unexpected effects of MiaA loss. Impaired translation of MiaA-sensitive codons reduced quorum-sensing-controlled virulence gene expression and attenuated pathogenicity in Galleria mellonella. Ribosome stalling at trp codons in miaA mutants overrides the attenuation-controlled repression of tryptophan biosynthesis, causing overproduction of tryptophan, along with upregulation of cognate tRNAs, thereby linking translation to global metabolic adaptation. MiaA is tightly regulated and is so central to bacterial physiology that its expression level correlates directly to virulence in clinical isolates, highlighting its role as a hub connecting translation, transcription, metabolism, and pathogenicity. These findings position MiaA as a key integrator of cellular processes critical for pathogen fitness and host interactions.

Graphical Abstract

Graphical Abstract.

Introduction

Transfer RNAs (tRNAs) are essential components of the translational machinery, decoding messenger RNA (mRNA) codons into amino acids during protein synthesis [1]. Beyond this canonical role, tRNAs undergo numerous post-transcriptional modifications that ensure translational accuracy, efficiency, and cellular fitness [2–6]. Modifications at positions 34 (the wobble base) and 37 (adjacent to the anticodon) are particularly important: position 34 affects codon pairing, whereas position 37 stabilizes codon–anticodon interactions and enhances decoding fidelity [7–9].

The effects of tRNA modifications depend on which tRNAs are modified, as each influences the translation of specific codons and proteins [10, 11]. Primary translation defects can trigger secondary effects that propagate through regulatory networks. Misfolded proteins can overwhelm quality control systems, triggering stress responses, such as heat shock or stringent responses [12–14], while altered translation of key regulators, including transcription factors, sigma factors, or signaling proteins [15–18], can amplify downstream gene expression changes. Moreover, altered tRNA modification impacts metabolism both directly, by consuming metabolic precursors required for tRNA modification pathways [19, 20], and indirectly, by modulating expression of biosynthetic operons. Hypomodified tRNAs can also impair translational fidelity through increased translational errors and frameshifting [21, 22] and reduce tRNA stability, further amplifying translational defects and downstream effects [23].

The tRNA-modifying enzyme MiaA is highly conserved in bacteria with well-documented effects on physiology and pathogenicity. MiaA catalyzes isopentenylation of adenosine at position 37 (i⁶A₃₇) in specific tRNAs decoding UNN codons [24, 25] (e.g. tRNA^Phe, tRNA^Leu, tRNA^Ser; Fig. 1a). The i⁶A derivative can be further methylthiolated by MiaB to form ms²i⁶A [26] and hydroxylated by MiaE to yield io⁶A or ms²io⁶A [27, 28]. tRNAs lacking these modifications show reduced codon-binding affinity and are prone to ribosomal P-site slippage in Escherichia coli and Salmonella Typhimurium [22, 29]. MiaA-dependent modifications regulate stress responses, virulence, and antibiotic resistance, partly by affecting translation of the stationary phase sigma factor RpoS [17, 18]. MiaA deficiency also reduces expression of virulence-associated phenotypes in extraintestinal pathogenic E. coli and Shigella flexneri by impairing translation of key regulatory factors [16, 18, 30, 31]. In Pseudomonas aeruginosa, MiaA-dependent i⁶A₃₇ modification has recently been characterized in a high-throughput epitranscriptomic profiling approach describing the tRNA modification landscape [32]. Furthermore, MiaA has been implicated in the control of virulence in this organism [33] and MiaB has been shown to connect environmental cues to activation of the type III secretion system (T3SS) [34].

Figure 1.

MiaA-dependent modification of adenosine at position 37 (A37) in P. aeruginosa tRNAs. (a) Pathway of MiaA-dependent modification of A37 in a subset of tRNAs decoding UNN codons. MiaA catalyzes the isopentenylation of adenosine at position 37 to i⁶A. This modification can be further methylthiolated by MiaB to form ms²i⁶A. Additionally, MiaE hydroxylates i⁶A or ms²i⁶A in an oxygen-dependent manner, generating io⁶A or ms²io⁶A. DMAPP = dimethylallyl pyrophosphate; SAM = S-adenosylmethionine; Cys = cysteine. (b) Targeted liquid chromatography–mass spectrometry (LC–MS/MS) quantification of MiaA-dependent tRNA modifications in total tRNA isolated from P. aeruginosa PA14 wild type (WT). Modified nucleosides are depicted as percentages relative to canonical adenosine, determined by external standard calibration. Statistical significance was determined by Student’s t-test (two-tailed, ***P < .001). ND, not detected. (c) Modification-induced mismatch profiles of PA14 and ΔmiaA tRNA^TrpCCA derived from total tRNA pooled sequencing using nano-tRNAseq. Positions with mismatches to the PA14 WT reference sequence are indicated by colors (green = A, blue = C, orange = G, red = T, dark gray = N; N indicates insertions or deletions). Close-up views of the anticodon region of the tRNA are shown on the right.

MiaA also links translation to metabolism. Reduced tRNA^Trp charging or hypomodification promotes ribosome stalling at the trp leader peptide, activating tryptophan biosynthesis genes [35–37]. A recent review emphasized that leader peptide-dependent attenuation is particularly sensitive to defects in tRNA modification and global ribosome pausing [38]. This reinforces the idea that altered decoding can reprogram amino acid homeostasis at the transcriptional level. Because MiaA uses dimethylallyl pyrophosphate, a terpenoid intermediate derived from acetyl-CoA, its activity connects tRNA modification to central metabolism and growth [20].

The widespread presence of MiaA-dependent tRNA modifications across diverse bacterial pathogens suggests a conserved role in tRNA function and regulatory processes with broad physiological and pathogenic implications [27]. Here, we combined LC–MS/MS, nanopore tRNA sequencing, transcriptomic, ribosome profiling, and proteomic data to define MiaA-dependent regulation in the opportunistic pathogen P. aeruginosa. Our findings reveal that MiaA modulates ribosome stalling in tryptophan attenuation, reshaping aromatic amino acid production, tRNA abundance, and complex phenotypes, including pathogenicity.

Materials and methods

Bacterial strains, media, and growth conditions

Strains, plasmids, and primers used in this study are listed in Supplementary Tables S1 and S2. Experiments were performed in standard lysogeny broth (LB) or M9 minimal medium (1 mM MgSO₄, 0.1 mM CaCl₂, 0.01 mM FeSO₄), M9 salt solution (42.2 mM Na₂HPO₄*2 H₂O, 22 mM KH₂PO₄, 18.7 mM NH₄Cl, 8.6 mM NaCl) supplemented with 20 mM glucose at 37°C. For plasmid maintenance, gentamicin was used at a final concentration of 50 μg/ml and streptomycin and kanamycin at a final concentration of 500 μg/ml for P. aeruginosa and 50 μg/ml for E. coli, carbenicillin and ampicillin were used at 400 μg/ml for P. aeruginosa and 100 μg/ml for E. coli, respectively. Unless stated otherwise, planktonic cultures were inoculated from stationary overnight cultures (ONC) with a starting OD of 0.05 and incubated in an orbital shaker (180 rpm). Growth curves were recorded in a BioTek Synergy H1 platereader over the course of 20 h. Growth parameters were calculated using QurvE [39].

Unless stated otherwise, all experiments were performed with at least three biological replicates.

Genetic engineering

To generate deletion and single-nucleotide polymorphism mutants in P. aeruginosa, we used a CRISPR–Cas9-assisted recombineering toolkit previously developed in our laboratory [40]. Deletion of miaA has been verified by polymerase chain reaction (PCR) and sequencing. For all miaA mutants, we ensured to keep the promoter and sequence of the downstream hfq gene intact to avoid polar effects.

For complementation of the miaA mutant in P. aeruginosa, we fused the promoter sequence upstream of the operon (NOG445/446) to the coding region of miaA (NOG447/448) by SOE PCR. The insert was cloned in the mini-CTX1 plasmid [41] by classic digestion/ligation procedure using SpeI/BcuI enzymes. The plasmid was introduced in P. aeruginosa by conjugation and selected on Pseudomonas Isolation Agar containing 200 µg/ml tetracycline. The resistance cassette was further excised using the pFLP2 plasmid, which was cured by sucrose-selection [42]. The integration was confirmed by PCR and Sanger sequencing using primers NOG409/410.

Construction and utilization of translation reporters

The translation reporters pCDN1 (pCoDoN) and pCDN2 were constructed as follows. A DNA fragment containing mCherry was PCR amplified from pSEVA247R using primer pairs AAR264/AAR265, and subsequently cloned into the EcoRI/KpnI sites of pSEVA2513, generating pSEVA2513-mCherry. Next, the monomeric superfolder green fluorescent protein (msfGFP) without its start codon and with a 5′-located Eco31I restriction site (used for codon cloning) was PCR amplified using primers AAR266/AAR267, SalI/HindIII digested, and cloned downstream of mCherry into pSEVA2513-mCherry, thereby generating pCDN1. For construction of pCDN2, a PCR fragment of msfGFP without start codon and with a flexible linker for protein/peptide fusion was amplified using primers AAR267/AAR268, SalI/HindIII digested and ligated into the same restriction sites of pSEVA2513-mCherry.

Codons cloned into pCDN1 were synthetically designed as oligonucleotide pairs containing three consecutive codons, phosphorylated and annealed as previously described [40], and integrated into the Eco31I-digested plasmid by ligation using the T4 DNA ligase (Thermo Fisher). All oligomers used for cloning (Supplementary Table S2) were ordered from Sigma. In frame codon and DNA insertions were verified by Sanger sequencing (Microsynth). Pseudomonas aeruginosa strains were electroporated with pCDN constructs and plated on LB agar supplemented with 500 μg/ml kanamycin. Isolated colonies were picked and grown overnight in 3 ml LB + 500 μg/ml kanamycin. Aliquots of 1 ml from each subculture were washed with phosphate-buffered saline (PBS) and grown in 10 ml LB without antibiotic in 50 ml Erlenmeyer flask. To avoid carry-over of overnight-saturated fluorescent proteins from pre-cultures, the initial OD₆₀₀ was adjusted to 0.005. Samples were taken in early exponential phase (OD₆₀₀ 0.5), appropriately diluted in filtered PBS (0.22 µm pore size) containing 30 μg/ml chloramphenicol to ~10⁶ cells/ml, and measured on a BD FACS Fortessa flow cytometry system. msfGFP was excited using a 488-nm laser and detected with a 525/50 band-pass filter, whereas mCherry was excited using a 561-nm laser and detected with a 610/20 band-pass filter. Calibration of the experiments was performed as follows: forward and side scatter density plots were used to identify the bacterial cell population and to exclude debris. Moreover, detection of the fluorescent populations was calibrated using non-fluorescent PA14 WT and a PA14 pCDN1-ATG bearing strain, producing both mCherry and msfGFP. All flow cytometry data were processed using FlowJo, excluding the cells that did not produce mCherry signal (i.e. dead cells or cells that have lost the pCDN derivative). The msfGFP and mCherry signals were determined using population mean fluorescence intensities and used to calculate the msfGFP/mCherry fluorescence ratio. Data visualization and statistical analyses were performed in GraphPad Prism (v10.5.0). Relative fluorescence ratios for each codon were compared between strains using two-way ANOVA.

Galleria mellonella infection assay

The Galleria mellonella model was used to determine P. aeruginosa virulence as described previously [43]. ONC were washed, bacteria were serially diluted in PBS (10 mM Na₂HPO₄, 137 mM NaCl, 1.8 mM KH₂PO₄, 2.7 mM KCl, pH 7.4), and 20 µl containing 100 colony-forming units (CFU) were injected into the last proleg of the larvae. The infected larvae were incubated in the dark at 37°C. To assess virulence, mortality rates of 10 replicate larvae were monitored for 48 h. Mortality was asserted by visual observation of melanization of the cuticle and lack of movement after stimulation. Furthermore, to assess the bacterial load in G. mellonella, the hemolymph of 5–7 infected larvae was harvested at the indicated timepoints, serially diluted and spotted on agar plates for CFU counting.

To analyze the link between virulence and miaA expression in clinical isolates, virulence data obtained in our previous research were used [44].

Cytotoxicity

RAW264.7 (ATCC number: TIB-71TM) cells were routinely grown in Dulbecco’s modified Eagle’s medium (DMEM) (4.5 g/l glucose) and supplemented with 10% FCS, 20 mM HEPES, and 1% bovine serum albumin (BSA) at 37°C, 5% CO₂. All cell cultures used in this study were tested and confirmed as mycoplasma-free. Infection assays were performed as described previously [45]. Briefly, 5 × 10⁵ RAW 264.7 cells were seeded in DMEM in a 24-well plate. Before the start of the experiments, the supernatant of each well was discarded and replaced with 400 μl of fresh DMEM without phenol red. Eukaryotic cells were infected at an MOI (multiplicity of infection) of 1 with P. aeruginosa strains grown to exponential phase, washed with PBS, and adjusted to a final volume of 100 μl in fresh DMEM. The number of CFUs in the inoculum dose was verified by serial plating on LB plates. Immediately after start of the infection, a centrifugation step at 1200 × g for 5 min was used to ensure cell-to-cell contact. One-hour post-infection (p.i.), 100 μl of DMEM containing gentamicin was added to the final concentration of 50 μg/ml. Fifty microliters of supernatant of infected RAW264.7 cells was collected 6 h p.i. and lactate dehydrogenase (LDH) levels were measured using the CytoTox 96 Non-radioactive Cytotoxicity Assay (Promega) according to the manufacturer’s protocol. PBS served as negative control and the lysis solution of the manufacturer was used as killing control. Values are expressed as % of the value obtained with the PA14 WT strain infection.

Motility

Swimming, swarming and twitching were performed as described previously [46]. Briefly, swimming was assessed by stabbing sub-cultures with an OD₆₀₀ of 2.0 into semi-solid BM2 agar plates (0.3% w/v agar) using a sterile toothpick. Plates were incubated for 18 h at 37°C. Swarming motility was assessed by spotting 2 µl of a sub-culture with an OD₆₀₀ of 2.0 onto semi-solid BM2 agar plates (0.5% w/v agar). Plates were incubated for 18 h at 30°C in a humid chamber. Twitching motility was assessed by stabbing sub-cultures with an OD₆₀₀ of 1.5 into LB agar plates using a sterile toothpick, followed by incubation at 37°C for 48 h. After removing the agar, twitching area was stained with 0.1% (w/v) crystal violet for 10 min and then washed with ddH₂O until the background was clear. All plates were imaged with the Epson Perfection V800 Photo scanner and the area was calculated using FiJi [47]. All values are the ratio of the area (or diameter for swimming) normalized to the value of the WT parental strain.

Biofilm formation

Crystal violet biofilm formation was assessed according to a previously published method [48]. Hundred microliters of the inoculum (OD₆₀₀ = 0.02) were added to the wells of a flexible, non-treated U bottom PVC 96-well plate (Corning Inc., NY, USA). After 24 h of incubation at 37°C in a humid atmosphere, wells were washed with water prior to the addition of 150 µl of the crystal violet staining solution (0.1% w/v in water). After 30-min incubation, the staining solution was removed and wells were again washed and air-dried. For de-staining, 200 µl of 96% (v/v) ethanol was added to each well and the plate was incubated for 30 min at RT. 125 µl of the solution was transferred to a fresh 96-well flat bottom plate before absorbance was measured at 550 nm. Each strain was tested in five biological replicates with eight technical replicates each.

Biofilm structures were examined by confocal microscopy as described in a previously published method [49]. Biofilms were cultivated for 48 h in a microtiter plate (Greiner half-area microtiter plate) under static conditions in a humid atmosphere with a starting OD₆₀₀ of 0.002 [49, 50]. After 24 h, pre-grown biofilms were stained with the LIVE/DEAD^®  BacLight™ Bacterial Viability Kit (Molecular Probes, Life Technologies, CA, USA, final concentrations of 2.1 µM Syto9 and 12.5 µM propidium iodide). Stacks of 48-h-old biofilms with a total height of 60 µm (20 focal planes; z-step size 3 µm) were acquired by using an automated confocal laser scanning microscope (SP8 System, Leica, Germany) with an HC PL APO 40×/1.10 W motCORR CS2 water immersion objective. Imaris 7.6 (Bitplane, UK) was used for 3D reconstructions of biofilm structures.

Disk diffusion assays for ciprofloxacin and H₂O₂ susceptibility testing

Pre-cultures of P. aeruginosa strains were grown overnight in LB, and subcultured to exponential phase (OD₆₀₀ ≈ 0.3). Cells were adjusted to 10⁸ CFU/ml in LB, and 1 ml was layered on 25 ml LB-agar plates. The plates were dried in the sterile hood for 1 h, and a disk containing either 5 µg of ciprofloxacin or 10 µl of a 30% H₂O₂ solution (v/v, Sigma) was placed in the center. After 18 h, the plates were scanned and the area of inhibition measured using FiJi [47].

Antimicrobial susceptibility testing in biofilms

Pseudomonas aeruginosa biofilms were grown in LB medium for 24 h in 96-well plates as described previously [51] and treated with increasing concentrations of ciprofloxacin for an additional 24 h, before the surviving bacteria were determined by counting the CFUs.

Pyocyanin production

Pyocyanin production was determined as described elsewhere [52, 53]. Briefly, planktonic cultures were grown for 8 h shaking before harvesting 5 ml bacterial suspension. After centrifugation, pyocyanin was extracted from the supernatants by addition of an equal volume of chloroform. 3 ml of the organic phase was mixed with 1 ml 0.2 M hydrochloric acid (HCl) and the absorbance at 520 nm was determined for the aqueous phase. Pyocyanin concentrations (µg/ml supernatant) were calculated by multiplication with the correction factor 17.072 and normalized to growth (OD₆₀₀).

Elastase activity

The elastolytic activity of secreted proteases was tested in an Elastin Congo Red (ECR) assay [54]. Cultures were harvested after 8 h and 24 h (PL) or 48 h (BF), respectively, and 100 µl of the supernatant was mixed with 900 µl ECR buffer (100 mM Tris-HCl, pH 7.5, 1 mM CaCl₂), supplemented with 22.5 mg/ml ECR (Sigma–Aldrich). The suspension was incubated for 3 h at 37°C and 900 rpm. After centrifugation, the absorbance of the supernatant was determined at 495 nm and normalized to growth (OD₆₀₀).

tRNA purification and identification of modifications

Extraction of total RNA

Total RNA was isolated by acidic phenol extraction as described previously [55]. Subsequent steps were carried out on ice. Briefly, bacterial liquid cultures (50 ml) were centrifuged (3800 × g) for 15 min at 4°C and the pellets suspended in resuspension buffer (0.3 M sodium acetate, pH 4.5, 10 mM Na₂ EDTA) followed by two extractions with equal volumes of cold acidic phenol. During the first extraction, cells were vortexed for periods of 30 s, 60 s, and 60 s, with 60 s pause intervals between each step followed by centrifugation (15 000 × g, 15 min, 4°C). The aqueous phase was transferred to a new tube containing 0.5 ml cold acidic phenol and vortexed for 60 s followed by centrifugation (15000 × g, 15 min, 4°C). The aqueous phase was transferred to a new tube, inverted with 2.5× volumes of cold ethanol [100%, high-performance liquid-chromatography (HPLC)-grade] and incubated on ice for 2 h. Total RNA was recovered by centrifugation (15 000 × g, 15 min, 4°C) and the pellet was resolved in 100 µl 0.3 M sodium acetate (pH 4.5) followed by a second precipitation with 2.5× volumes of cold ethanol (100%, HPLC-grade) at −20°C overnight. Total RNA was recovered by centrifugation (15 000 × g, 15 min, 4°C) and the pellet was washed twice with ethanol (70% v/v). The pellet was dried and resuspended in RNase-free H₂O. The concentration of RNA was determined by a Nanodrop spectrophotometer (DeNovix DS-11 FX+).

Purification of total tRNA

Total tRNA was isolated from total RNA by solid-phase chromatography using anion exchange columns with a quaternary ammonium modified silica membrane (Chromabond SB 1 ml, Macherey-Nagel) as described previously [56]. The column was placed on a 15-ml falcon tube and conditioned with sample buffer (20 mM Tris–HCl, pH 7.5, 0.2 M NaCl, 10 mM MgCl₂). The falcon tube containing the column on top was centrifuged after each subsequent step (300 × g, 4°C). After conditioning, 1 mg of total RNA was mixed with 6 ml sample buffer and applied to the column. The column was washed with sample buffer six times followed by elution of total tRNA in three consecutive steps with elution buffers containing increasing salt concentrations (20 mM Tris–HCl, pH 7.5, 0.5 M to 1 M and up to 1.3 M NaCl, 20 mM MgCl₂). Afterward, total tRNA was precipitated by adjusting the NaCl concentration (2.5 M) and by addition of 2.5× volumes of ethanol (100%, HPLC-grade) at −20°C overnight. Total tRNA was recovered by centrifugation (3.800 × g, 1 h, 4°C) and washed three times with ethanol (70% v/v). The pellets were dried and suspended in RNase-free H₂O. The concentration of tRNA was determined using a BioTek Synergy H1 platereader with a Take3 Microvolume Plate (Agilent) and RNA integrity was checked using the Agilent Bioanalyzer (RNA 6000 Nano Kit; Agilent).

Liquid chromatography–mass spectrometry analysis of ribonucleosides

tRNAs were enzymatically hydrolyzed prior to LC–MS/MS analysis as described [56]. Briefly, total tRNA (10 μg) was incubated for 4 h at 37°C in an RNA–hydrolysis solution [0.5 M Tris–HCl, pH 8, 50 mM MgCl₂, 1 mg/ml BSA, 15 U/μl benzonase nuclease, 0.004 U/μl phosphodiesterase-I (PDE-I), 30 U/μl alkaline phosphatase] with gentle shaking and mixed with an equal volume of the internal standard (IS) tenofovir (100 ng/ml) followed by enzyme removal using a pre-rinsed 5 kDa centrifugal filter device (Vivaspin500; Sartorius, Göttingen, Germany) for 45 min at 4°C and 15 000 × g. For the chromatographic separation of the ribonucleosides, a Shimadzu Nexera HPLC system (Shimadzu, Duisburg, Germany) with a Zorbax Eclipse XDB-C18 RP-HPLC column (50 × 4.6 mm, 1.8 μm particle size) (Agilent, Santa Clara, CA, USA) equipped with a column saver (2 μm) (Supelco, Bellefonte, PA, USA) and a C18 RP security guard (Phenomenex, Aschaffenburg, Germany) was used. The column oven was set to 30°C and the autosampler to 4°C. The injection volume was 10 μl for each sample and the flow rate 0.4 ml/min. The mobile phase A consisted of 0.1% formic acid dissolved in HPLC-grade H₂O and the mobile phase B of formic acid in HPLC-grade methanol. A gradient duration of 21.1 min in total with 5% mobile phase B at 0.3 min at the beginning followed by 95% at 13.9 min and 14.9 min and 5% at 15 min until 21 min was used. Equilibration was performed for 3 min.

For the MS/MS analysis of modified ribonucleosides, the Shimadzu Nexera HPLC system was coupled to a Linear Ion Trap Quadrupole QTRAP5500 mass spectrometer (Sciex, Framingham, MA, USA). The system was equipped with an ESI ion source for ionization of the samples. The analysis was performed in positive ionization mode and the MS system was operated with the following parameters: Source temperature, 400°C; ion spray and detector voltage, 5.5 kV; curtain gas, 30 psi; nitrogen collision gas, 10 psi; nebulizer and interface heater gas, 60 psi and 75 psi, respectively. For the quantification of modified ribonucleosides, external standard calibration with synthetic reference standards for each analyte was used. Commercially available synthetic reference standards were purchased or synthesized. The concentrations of the external standard mixtures ranged from 13.4 pM up to 5 μM with 2.5-fold serial dilutions. Raw data were analyzed using the Analyst 1.6 software (Sciex) with the following parameters: Smoothing width 3 points, noise percent 90%, and peak-splitting factor 2. For the calibration curve, quadratic regression with 1/x weighting and accuracy with 100 ± 20% was determined to detect the linear range. Based on the calibration curve of each reference standard, the lower limit of quantification and upper limit of quantification were determined. The modification amount was calculated by the ratio of the peak area from the modified ribonucleoside and the peak area from the IS.

Nano-tRNAseq library preparation and sequencing

Nano-tRNAseq of P. aeruginosa tRNA samples was performed as recently described [57]. Oligonucleotides used for nano-tRNAseq are listed in Supplementary Table S2. Briefly, for the nano-tRNAseq library preparation, tRNA samples were deacetylated and ligated to the pre-annealed 5′ and 3′ splint adapters. Next, 5′ and 3′ ligated tRNAs were ligated to the pre-annealed RTA adapters. Afterward, a reverse transcription master mix was added directly to the ligation reaction before ONT RMX sequencing adapters (Direct RNA Sequencing Kit, SQK-RNA002, Oxford Nanopore Technologies) were ligated to the tRNAs. MinION flow cells (FLO-MIN-106) underwent quality control, priming, and loading following the standard ONT SQK-RNA002 protocols. Nano-tRNAseq data processing and analysis was done according to [19]. Briefly, raw nanopore reads were basecalled using Guppy (v6.5.7), aligned to a non-redundant PA14 tRNA reference database [58] using BWA mem (v0.7.17), and filtered, sorted and indexed using SAMtools (v1.15) [59]. tRNA abundance estimates were obtained from mapped read counts and differential abundance analysis was performed using DESeq2 [60]. Modification-induced mismatch profiles were generated from aligned reads using bam-readcount and analyzed in R using custom scripts.

Transcriptomic and proteomic analyses

RNA extraction and sequencing

RNA was isolated using the RNeasy Mini Kit (Qiagen) together with Qiashredder™ columns following the protocol used in our earlier work [44]. Briefly, 10 ml bacterial cultures were grown planktonically with a starting OD₆₀₀ of 0.05 and harvested at an OD₆₀₀ of 2.0 and immediately stabilized by mixing with an equal volume of RNAprotect (Qiagen). RNA purification was carried out according to the manufacturer’s guidelines with minor adjustments. Residual genomic DNA was removed using the DNA-free™ Kit (Thermo Fisher Scientific). RNA quality and integrity were verified on an Agilent Bioanalyzer using the RNA 6000 Nano Kit. Ribosomal RNA depletion was performed with the Illumina Ribo-Zero Bacteria Kit, and libraries were sequenced on an Illumina NovaSeq 6000 platform in paired-end mode (2 × 50 bp).

Reads were aligned to the PA14 reference genome with bowtie2 [61], and gene-level counts were obtained using FeatureCounts [62]. Differential expression between PA14 WT and mutant strains was evaluated using the edgeR package [63]. Genes with insufficient expression were removed using the filterByExpr function, and normalization was performed via calcNormFactors (TMM method). Differential expression testing employed glmTreat with a fold change threshold of 1.2. P-values were adjusted for multiple testing using the Benjamini–Hochberg procedure, and genes with false discovery rate (FDR) ≤ 0.05 were designated as differentially expressed. Enrichment of Gene Ontology was assessed using a hypergeometric test (R function phyper) with FDR-corrected significance cutoffs (<0.05).

Clinical isolate transcriptome analysis

To assess miaA expression across 77 clinical P. aeruginosa isolates, we analyzed previously published transcriptional profiles recorded under planktonic (PL), as well as under biofilm (BF) growth conditions [50, 64, 65]. Reads were aligned to the PA14 reference genome using bowtie2 [61], and gene-level counts were generated using FeatureCounts [62], analogous to the workflow described above. Raw counts were transformed to counts per million (cpm) using the TMM normalization method implemented in edgeR [63].

Condition-dependent miaA expression was determined by calculating log₂-fold changes (log₂FC) between biofilm and planktonic conditions (BF versus PL). Clinical isolates were ranked according to their miaA log₂FC values, ranging from isolates with higher miaA expression during planktonic growth (negative log₂FC) to isolates with elevated expression during biofilm growth (positive log₂FC). To visualize both relative and absolute expression patterns, cpm values for planktonic and biofilm conditions were plotted alongside the corresponding log₂FC values.

Virulence phenotypes previously determined in the G. mellonella infection model [44] were integrated as an aligned heatmap displaying % survival at 48 h p.i. for each isolate. Consistent isolate ordering across all data layers enabled direct comparison of miaA expression dynamics, biofilm-associated transcriptional regulation, and virulence phenotypes.

Protein extraction and quantification

Cell pellets were resuspended in a sodium dodecyl sulfate lysis buffer and mild sonication was used for cell lysis. Protein concentration was quantified using the Pierce Modified Lowry Protein Assay Kit (Thermo Fischer Scientific). Proteins were extracted by phenol-chloroform extraction. The precipitated proteins were dissolved in a digestion buffer, and samples were acidified to pH 2.3 with formic acid followed by centrifugation. The supernatant was transferred to a new tube and further evaporated at 30°C in a Speed-Vac^®. Peptides were desalted using the OASIS elution plate (Waters), with elution performed by using 60% acetonitrile (ACN) and 0.1% formic acid. Peptide concentration was determined using Pierce Quantitative Fluorometric Peptide Assay (Thermo Fischer Scientific). LC–MS analyses were conducted utilizing an Exploris 240 orbitrap mass-spectrometer (Thermo Fischer Scientific) with a resolution set to 60.000 and a scan range of 350–1900 m/z. Prior to MS acquisition, samples were separated on a Thermo Scientific COL-nano050G1B column (PharmaFluidics) using an ACN gradient ranging from 2.5% to 95%. Raw data were processed with MaxQuant (v2.4.10.0) [66] applying the integrated Andromeda search engine [67] for protein identification at a 1% FDR. Subsequent analyses were conducted in RStudio and included log₂ transformation of peptide intensities, removal of contaminants and non-unique peptides, and statistical comparison between conditions using two-sample t-tests on mean log₂ values.

Analysis of time-resolved changes in mRNA and protein levels

For comparison of miaA transcript and protein abundance dynamics over a 10-h timecourse, miaA transcript expression data were obtained from previously published RNA-seq datasets [68], while MiaA protein abundance was determined through proteomics (Supplementary data ES3b). Transcript and protein abundance values were z-score normalized and plotted over the 10-h period to facilitate comparison of relative expression changes.

Ribosome profiling

Ribosome profiling was performed as described previously [10]. Briefly, 200-ml cultures were grown to exponential phase (OD₆₀₀ = 0.4) and harvested by rapid filtration. For ribosome profiling, lysates were treated with MNase to generate ribosome-protected fragments. Monosomes were isolated by sucrose gradient centrifugation and RNA footprints were purified. Size-selected RNA (15–45 nt) was converted to complementary DNA libraries using a NEBNext Multiplex small RNA library prep for Illumina (E7300) using custom multiplex primers. Sequencing was performed on an Illumina NovaSeq 6000 platform as 50 bp paired-end sequencing at the Genome Analytics Research Group at the Helmholtz Centre for Infection Research (HZI) of Braunschweig.

The sequencing depth of the ribosomal footprints of PA14 WT and ΔmiaA was between 29 and 34 million reads. For gene-level comparison, data were processed the same way as transcriptome data. Ribosome density (“ribosomes per mRNA”) was calculated by comparing ribosomal footprints (RBF) versus transcription (TR) using the formula “ΔmiaA_RBF - ΔmiaA_TR” for ribosome density of ΔmiaA and “WT_RBF - WT_TR” for PA14 WT. Differences in ribosome density between the ΔmiaA mutant and PA14 WT were calculated using edgeR comparing (ΔmiaA_RBF - ΔmiaA_TR) - (WT_RBF - WT_TR). Reads per gene were filtered and normalized in the same way as the transcriptome data (filterByExpr, calcNormFactors). The replicates were analyzed using the edgeR function plotMDS. For codon enrichment analysis, genes were ordered by their adjusted P-value and the top 25% genes with the highest and the lowest ribosome occupancy (log₂-fold change higher/lower 0) in ΔmiaA were selected. Enrichment analysis was performed for each codon using the R function phyper and information about the presence/absence of codons within each gene. The codon enrichment score (fold enrichment of the number of genes bearing a codon compared to the number of genes expected by chance) was calculated using the formula Cs/Cb, where Cs is the proportion of these gene-subsets containing a codon and Cb is the codon proportion in the total number of genes used in the edgeR analysis. Data were plotted using the R library ggplot2.

Codon enrichment analysis

For each gene of P. aeruginosa PA14, the number of MiaA-associated codons (TCA, TCG, TGG, TAC, TAT, TGC, TGT, TTC, TTT) relative to the total codon count was compared to the genome-wide expected proportion (median proportion of MiaA codons) using a hypergeometric test, with P-values corrected for multiple testing (Benjamini–Hochberg). Genes with FDR <0.05 were considered significantly enriched. Analyses were performed in R.

Sequence conservation analysis across clinical strains

MMseqs2 (v13.45111) [69] was used to search for MiaA sequences in 6870 high quality assembled P. aeruginosa genomes of clinical isolates in our collection and an equal number of hits was returned. A total of 7 sequences were excluded based on their immediate proximity to contig edges. The remaining 6863 translated sequences were aligned using MAFFT (v7.526) [70], gaps were removed and identity relative to the PA14 reference was calculated for each amino acid position. Predicted substrate binding sites and tRNA interaction sites were obtained from UniProt and are indicated.

Targeted metabolomics of tryptophan and intermediates of aromatic amino acid metabolism

For the extraction of analytes, bacterial liquid cultures were harvested at an OD₆₀₀ of 2. All subsequent steps were carried out on ice or at 4°C to preserve sample integrity. A volume of 2 ml of bacterial suspension was collected and centrifuged at 2500 × g, 20 min at 4°C and the supernatant was discarded. Pellets were washed twice with 500 μl of cold PBS, centrifuged, and the supernatant was discarded. Pellets were resuspended in 300 µl of ice-cold extraction solvent [ACN, methanol (MeOH), H₂O, (2/2/1, v/v/v); 1.33 μM IS caffeic acid] to stop cellular metabolism. The suspension was incubated on ice for 15 min and centrifuged at 20 800 × g, 10 min at 4°C, and the resulting supernatant was carefully transferred into a new 2.0-ml tube. The extraction process was repeated twice using 200 μl of extraction solvent without the IS [ACN, MeOH, H₂O (2/2/1, v/v/v)]. The supernatants from all three extraction steps were combined, resulting in a total volume of ~700 μl, and stored at −20°C overnight for protein precipitation. On the next day, the samples were thawed and centrifuged at 20 800 × g, 10 min at 4°C and the resulting supernatant was then transferred to a new 2 ml tube. For sample preparation for LC–MS/MS analysis, the supernatant was evaporated at 30°C in a Speed-Vac^® to dryness. The dried samples were then dissolved in 150 μl of sample solvent without the IS (HPLC-grade H₂O, 0.1% formic acid), vortexed for 30 s, and centrifuged for 10 min at 20 800 × g at 4°C. A volume of 75 μl of the supernatant was transferred to an MS vial with insert. For further dilution, sample solvent with the IS (HPLC-grade H₂O, 0.1% formic acid, 1.33 μM caffeic acid) was used.

The residual pellet from the extraction steps was used for protein quantification using the Pierce™ Modified Lowry Protein Assay Kit (Thermo Fischer Scientific, 23240). For this, the pellet was resuspended in 800 μl of 0.1 M sodium hydroxide (NaOH). The samples were then heated at 95°C for 15 min until dissolution. Subsequently, the samples were centrifuged (20 800 × g, 10 min at 4°C), and 10 μl of the supernatant was used for the Lowry assay in triplicates, following the manufacturer’s instructions.

For chromatographic separation, a Shimadzu Nexera HPLC system (Shimadzu, Duisburg, Germany) with a Zorbax Eclipse XDB-C18 RP-HPLC column (50 × 4.6 mm, 1.8 μm particle size) (Agilent, Santa Clara, CA, USA) equipped with a column saver (2 μm) (Supelco, Bellefonte, PA, USA) and a C18 RP security guard (Phenomenex, Aschaffenburg, Germany) was used. The column oven was set to 30°C and the autosampler to 4°C. The injection volume was 10 μl for each sample and the flow rate 0.4 ml/min. The mobile phase A consisted of 0.1% formic acid dissolved in HPLC-grade H₂O and the mobile phase B of formic acid in HPLC-grade methanol. A gradient duration of 21.1 min in total with 5% mobile phase B at 0.3 min at the beginning followed by 95% at 13.9 min and 14.9 min and 5% at 15 min until 21 min was used. Equilibration was performed for 3 min. The MS analysis was performed with a linear Ion Trap Quadrupole QTRAP5500 mass spectrometer (Sciex, Framingham, MA, USA) equipped with an ESI ion source used in positive or negative ionization mode, depending on the analyte. The MS system was operated with the following parameters: Source temperature, 400°C; ion spray and detector voltage, 5.5 kV; curtain gas, 30 psi; nitrogen collision gas, 10 psi; nebulizer and interface heater gas; 60 psi and 75 psi, respectively. External standard calibration with synthetic reference standards was used for quantification of metabolite levels.

Metabolite quantification of shikimate, shikimate-3-phosphate, and chorismate

Bacterial liquid cultures were harvested at an OD₆₀₀ of 2. Biomass was extracted with 250 µl methanol supplemented with 1% U-13C-ribitol (0.2 mg/ml) for 15 min in an ultrasonic bath. Subsequently, 250 µl of H₂O was added and the samples were vigorously mixed. After addition of 500 µl dichloromethane, the samples were mixed and centrifuged (10 000 × g, 5 min). The polar phase was collected and dried under vacuum. Metabolite analysis was performed on an Agilent GC-MSD system (7890B coupled to a 5977 GC) equipped with a high-efficiency source and a PAL RTC system according to [71]. A two-step derivatization with a methoxyamine hydrochloride solution (20 mg/ml in pyridine) and N-methyl-N-(trimethylsilyl)-trifluoracetamide was automatically performed with the PAL RTC system. One microliter of the sample was injected into a multimode inlet in both split mode (split ratio of 10:1, split flow of 12 ml/min) and pulsed splitless mode (30 psi until 0.75 min, 50 ml/min at 1 min). Separation was conducted on an Agilent VF-5ms column with a helium flow of 1.2 ml/min. The oven temperature was held at 70°C for 6 min and then linearly increased with 6°C/min up to 325°C. Ions were detected in scan mode from 70 to 700 m/z with 2.3 scans/s. Data analysis of intracellular metabolites was performed as previously described [72, 73].

Untargeted metabolomics for the analysis of leucine levels

For untargeted LC–MS/MS analysis of the intracellular metabolome, PA14 WT and ΔmiaA planktonic cultures were grown in triplicate in M9 minimal medium to early stationary phase (OD₆₀₀ = 2). Cell pellets were processed and LC–MS analysis was performed as previously described [74]. Leucine was annotated by matching in-house tandem MS reference spectra and retention times of authentic standards measured on the same LC–MS system using Bruker DataAnalysis 5.3. Extracted ion chromatograms were generated, and peak areas for leucine were determined by peak integration using Bruker DataAnalysis 5.3. Peak areas were subsequently normalized to the corresponding peak area of glipizide, which was spiked into the extraction solvent as an IS.

Bioinformatic and statistical analyses

Bioinformatic and statistical analyses were performed using established software packages in R (v4.3) or with GraphPad Prism (v10.5.0 and v8.3.0) as described in the respective sections. Read mapping was performed with Bowtie2 (v2.2.5), differential expression analyses with edgeR (v3.4X.X), and gene-level quantification with FeatureCounts (v2.0.1). Protein identification and quantification were performed using MaxQuant (v2.4.10.0) with the integrated Andromeda search engine. Homolog searches were performed using MMseqs2 (v13.45111) and multiple sequence alignments were generated with MAFFT (v7.526). Nano-tRNAseq data were processed using Guppy (v6.5.7), BWA mem (v0.7.17), SAMtools (v1.15), bam-readcount, and DESeq2 [60]. Functional enrichment and codon-enrichment analyses were conducted in R using hypergeometric testing (phyper) with false-discovery-rate correction.

Flow cytometry data were analyzed using FlowJo (v10.10.0). Motility images and inhibition zones from disk diffusion susceptibility testing were quantified using Fiji (v1.50t). Bacterial growth parameters, including doubling times, were calculated using QurvE (v1.1).

Data visualization was performed in R using ggplot2 (v3.4.X) [75] or GraphPad Prism. Biofilm images were reconstructed with Imaris 7.6 (Bitplane, UK). Statistical analyses were performed in R or GraphPad Prism and statistical tests and significance thresholds are indicated in the corresponding figure legends. A complete list of software packages, versions, and key parameters is provided in Supplementary Table S3.

Results

MiaA-dependent tRNA modifications in P. aeruginosa

To investigate the P. aeruginosa PA14 MiaA homolog (PA14_65 320) and its function in tRNA modification and translation, we generated a PA14 ΔmiaA mutant using a CRISPR–Cas9-assisted recombineering system [40]. Since the miaA open reading frame overlaps the transcriptional start site of hfq—a gene critical for post-transcriptional regulation [76]—we deleted only the N-terminal segment of miaA (amino acids S₂–L₁₆₃). Transcriptomic and proteomic analyses confirmed unchanged hfq expression in the ΔmiaA mutant compared to PA14 WT (Supplementary data ES1 and ES3). We then isolated total tRNA from PA14 WT and the ΔmiaA strain, enzymatically hydrolyzed the samples to ribonucleosides, and quantified MiaA-dependent modifications using targeted LC–MS/MS with synthetic reference standards [56]. In PA14 WT tRNAs, we detected varying levels of i⁶A, io⁶A, and ms²io⁶A, with io⁶A being the most abundant, while ms²i⁶A levels were below the detection limit (Fig. 1b and Supplementary Fig. S1). None of these modifications were found in the ΔmiaA strain, consistent with previous studies demonstrating that MiaA is essential for biosynthesis of isopentenylated adenosine derivatives in P. aeruginosa tRNAs [32, 34].

To map the positions of MiaA-dependent modifications, we performed direct nanopore RNA sequencing on total tRNA pools and extracted modification-induced mismatch profiles of the representative tRNA^TrpCCA from both PA14 WT and ΔmiaA [19, 57] (Fig. 1c). In the WT sample, elevated mismatch frequencies were detected at positions 37 and adjacent nucleotides, whereas these signals were absent in the ΔmiaA strain. These patterns confirm the presence of MiaA-dependent modified ribonucleotides in the WT tRNA and highlight the sensitivity of nano-tRNAseq in detecting basecalling errors arising from non-canonical, low-abundance anticodon modifications.

MiaA-dependent phenotypes in P. aeruginosa

To evaluate the phenotypic impact of MiaA, we compared the ΔmiaA strain to the PA14 WT across several infection-relevant phenotypes. Pathogenicity was first assessed using the G. mellonella larvae infection model [77]. While PA14 WT killed 100% of larvae within the first 24 h, the ΔmiaA mutant was almost completely avirulent, with 90% larval survival at 48 h p.i. (Fig. 2a). Complementation of miaA in cis under control of its natural promoter (PA14 ΔmiaA::miaA) fully restored virulence. This result was further supported by in vitro infection assays using murine macrophages (RAW 264.7), where the ΔmiaA strain exhibited significantly reduced cytotoxicity at 6 h p.i. compared to the WT and the complemented strain (Fig. 2b). To assess the contribution of downstream i⁶A-derived tRNA modifications to pathogenicity, we additionally examined ΔmiaB, ΔmiaE, and ΔmiaABE in the G. mellonella infection model (Supplementary Fig. S2). While ΔmiaB and ΔmiaE exhibited only modest attenuation, the ΔmiaABE mutant remained completely avirulent. These findings suggest that MiaA-mediated isopentenylation is the major determinant of pathogenicity within this modification pathway in the PA14 type strain.

Figure 2.

Phenotypic characterization of the ΔmiaA mutant. (a) Virulence of PA14 WT, ΔmiaA, and the complemented strain (PA14 ΔmiaA::miaA) in the G. mellonella infection model. Ten G. mellonella larvae were infected with 100 CFUs per larva, and survival was monitored over 48 h (n = 3 biological replicates). The dashed gray line indicates the PBS control. (b) In vitro cytotoxicity toward RAW264.7 macrophages following infection at an MOI of 1. LDH release was measured 6 h p.i. and is shown relative to WT levels. (c) Growth kinetics of PA14 WT and ΔmiaA strains in LB medium monitored over 20 h. Data represent mean ± SD of biological replicates (n = 6). (d) Doubling times in LB calculated by linear regression using default settings in QurvE [39]. Quantification of secreted virulence factors pyocyanin (e) and elastase (f) after 8 h of growth in LB medium. Values were normalized to growth (OD₆₀₀). Motility phenotypes assessed on agar plates: swimming (g) and swarming (h) on semi-solid BM2 agar (after 18 h at 30°C), and twitching motility (i) following stab inoculation into solid LB agar (incubation for 48 h at 37°C). Biofilm formation assessed by crystal violet staining of 24 h biofilms (j) and by confocal microscopy of 48 h biofilms stained with the LIVE/DEAD BacLight™ viability kit (k). Susceptibility to oxidative stress (H₂O₂; l) and antibiotic stress (ciprofloxacin; m) determined by disk diffusion assays. Inhibition zone areas are shown relative to PA14 WT. (n) Biofilm-induced tolerance to ciprofloxacin. Pre-formed biofilms (24 h) were treated with increasing concentrations of ciprofloxacin for 24 h, followed by quantification of surviving bacteria by CFU counting. Each data point represents an independent biological replicate (n = 2–8). Statistical significance was determined by ANOVA (one-way) with ns = not significant; *P < .05, **P < .01, ****P < .0001.

Next, we monitored bacterial growth in rich medium (Luria Bertani, LB) and observed an increased doubling time of the ΔmiaA strain, while there were no significant differences in the final OD₆₀₀ between the ΔmiaA and the PA14 WT or the complemented mutant (Fig. 2c and d). Further analysis revealed that the ΔmiaA strain failed to produce pyocyanin, a redox-active pigment [78] (Fig. 2e), and exhibited reduced elastase activity (Fig. 2f). Motility assays demonstrated markedly diminished swarming and twitching, while swimming was unaffected (Fig. 2g–i). Biofilm formation was impaired, as evidenced by decreased crystal violet staining (Fig. 2j) and reduced biofilm volume in 96-well plates assessed by confocal laser scanning microscopy (Fig. 2k). Furthermore, the ΔmiaA strain exhibited increased sensitivity to oxidative stress (H₂O₂; Fig. 2l) and to the quinolone antibiotic ciprofloxacin (Fig. 2m), with enhanced ciprofloxacin susceptibility also observed under biofilm growth conditions [51] (Fig. 2n). Importantly, all phenotypes were restored to WT levels upon miaA complementation. Together, these results underscore the critical role of MiaA in P. aeruginosa virulence, stress resistance, and biofilm-associated antibiotic tolerance.

MiaA-mediated changes in the transcriptome, translatome and proteome

We next employed both transcriptional (RNA-seq) and ribosome profiling (Ribo-seq) approaches, the latter involving deep sequencing of ribosome-protected mRNA footprints after nucleolytic digestion [79]. Multidimensional scaling (MDS) analysis of the most variably expressed and translated genes (Fig. 3a) revealed low variability between biological replicates. While the absence of a functional miaA gene had a pronounced effect at the transcriptional level, its impact on overall translation was comparatively minor. Calculation of a codon enrichment factor for each transcript [10, 19] revealed decreased translation efficiency (ribosomal footprints/mRNA ratio) for individual MiaA-dependent codons alongside several MiaA-independent codons (Supplementary Fig. S3). Notably, several of the affected MiaA-independent codons belong to the same amino acid families as MiaA-dependent codons (e.g. Ser and Leu), suggesting that loss of MiaA induces broader translational effects beyond direct MiaA targets.

Figure 3.

Absence of MiaA leads to global transcriptomic, translatomic, and proteomic changes. (a) MDS analysis of the most variably expressed and translated genes in PA14 WT and ΔmiaA. (b) Flow cytometry analysis of translation efficiency using dual fluorescent reporters. The fluorescence ratio between the translational signal from test codon stretches (with three consecutive repeats) with msfGFP versus the constitutively expressed reporter mCherry is shown for PA14 WT, ΔmiaA and ΔmiaA::miaA. Data represent the means ± SD of three biological replicates. Statistical significance was determined by ANOVA (two-way) with Tukey’s multiple comparison test; ns = not significant **P < .01, ***P < .001, ****P < .0001. Volcano plots depicting log₂-fold changes in transcription (c), ribosome occupancy as ribosomal footprints relative to mRNA (d) and protein abundance (e) comparing the ΔmiaA mutant to PA14 WT (n = 3). Significantly upregulated genes and proteins are shown in red and significantly downregulated genes and proteins are shown in blue (|log₂FC| ≥ 1, −log₁₀(FDR) ≥ 1.3). Selected genes and proteins are annotated. (f) Codon enrichment analysis in protein groups with altered protein-to-mRNA ratios in ΔmiaA compared to PA14 WT. The top 25% of genes with higher protein-to-mRNA ratios in PA14 WT versus ΔmiaA (PA14 WT > ΔmiaA, gray) and vice versa (ΔmiaA > PA14 WT, blue) were analyzed. The codon enrichment factor (y-axis) displays the ratio of transcripts containing a given codon as compared to transcripts lacking that codon. MiaA-sensitive codons are highlighted in red. Significance was determined using hypergeometric testing. Asterisks indicate Benjamini-Hochberg adjusted P-values; *P < .05, **P < .01, ***P < .001.

To further examine codon-specific changes in translation efficiency in the absence of MiaA, we employed a translational reporter system [19] in which a strong P_EM7 promoter drives constitutive expression of mCherry, followed by an msfGFP reporter fused in-frame to selected MiaA-sensitive test codon stretches with three consecutive repeats. The corresponding reporter plasmids were introduced into PA14 WT, ΔmiaA and the complemented strain, and decoding efficiency was quantified by flow cytometry (Fig. 3b). Compared to PA14 WT, the ΔmiaA strain exhibited significantly reduced normalized msfGFP-to-mCherry signals for several tested codons, including Ser-UCA, Tyr-UAC, and UAU, as well as Trp-UGG, confirming that i⁶A derivatives have an impact on the decoding capabilities of the respective tRNAs. In contrast, no significant differences were observed for the serine codons UCC and UCU. Notably, these codons are decoded by tRNA^SerGGA, which is not a MiaA substrate despite containing the canonical A36-A37-A38 recognition motif [80].

We next compared differentially expressed genes between the PA14 WT and its ΔmiaA mutant at the transcriptional, translational, and proteomic levels. Transcriptomic analysis identified 545 differentially expressed genes (DEGs; |log₂FC| ≥ 1, FDR < 0.05), including 156 genes significantly upregulated and 389 genes downregulated upon miaA deletion (Fig. 3c and Supplementary data ES1). Following normalization of ribosomal footprints to mRNA levels in our Ribo-seq experiment, we identified 491 genes with a differential ribosome occupancy (|log₂FC| ≥ 1, FDR < 0.05), comprising 86 with increased and 405 with decreased ribosome occupancy in the ΔmiaA strain (Fig. 3d and Supplementary data ES2). Transcriptional regulators together with two-component systems represented one of the largest functional groups among genes with altered ribosome occupancy, suggesting that comparatively subtle translational perturbations may be transcriptionally amplified through downstream regulatory cascades. Proteomic profiling further identified 336 differentially abundant proteins (|log₂FC| ≥ 1, FDR < 0.05), including 191 more abundant and 145 less abundant in ΔmiaA compared to WT (Fig. 3e and Supplementary data ES3). Functional enrichment across all three omics layers revealed a pronounced upregulation of genes involved in aromatic amino acid metabolism—particularly L-tryptophan biosynthesis—in the ΔmiaA strain (Supplementary Fig. S4). Conversely, and consistent with our phenotypic observations, genes linked to virulence, such as phenazine biosynthesis, quorum sensing, biofilm formation, and the type VI secretion system (T6SS) were significantly downregulated at both the transcriptional and translational levels.

To assess whether the proteomic changes in ΔmiaA result from altered decoding of specific codons, we quantified MiaA-modification-sensitive codons within the coding sequences of proteins that differed in abundance between WT and ΔmiaA (Fig. 3f). While many codons showed some degree of enrichment, reflecting broad proteome-wide shifts, a striking pattern emerged: codons dependent on MiaA modifications were preferentially enriched in proteins that were less abundant in the ΔmiaA strain. This bias indicates that genes containing higher levels of MiaA-dependent codons are disproportionately affected when the modification is lost.

In addition, to more directly connect codon usage of MiaA-dependent codons with infection-relevant phenotypes, we performed a genome-wide codon-enrichment analysis in PA14 based on currently known MiaA-dependent codons. This identified 233 genes significantly enriched in MiaA-sensitive codons (Supplementary data ES4), including multiple factors involved in virulence-associated processes, such as secreted phospholipases (plcB, plcH), genes involved in lipopolysaccharide and O-antigen biosynthesis or modification (lpxO1, lpxO2, arnT, wzm, orfEJN, PA14_23 400), type IV pili-associated attachment (pilD, pilY1), and biofilm matrix components (alg8, algI, pelG, pslL). In addition, several genes involved in cellular respiration (e.g. ccoN, cioAB, cyoBE, coIII, nuoAHLM) and denitrification (narGH, narK1, narK2, norB) were significantly enriched in MiaA-dependent codons, likely contributing to bacterial fitness during infection.

Impact of MiaA-dependent tRNA modifications on the regulation of L-tryptophan biosynthesis

Our finding that the tryptophan biosynthesis genes were strongly affected by the absence of MiaA—both at the translational and the transcriptional levels—suggests a link between impaired translation and altered transcription. Indeed, the regulation of tryptophan biosynthesis provides a compelling example of how translation efficiency can influence gene transcription. In E. coli, transcription of the L-tryptophan operon is controlled by attenuation—a mechanism that involves a regulatory trpL leader sequence [81]. Translation of trpL can trigger premature transcription termination when intracellular levels of L-tryptophan are high. In contrast, under conditions of L-tryptophan starvation, the accumulation of uncharged tRNA^TrpCCA causes ribosomes to stall at a tandem tryptophan codon motif within trpL. This stalling prevents the formation of a transcription terminator structure, thereby allowing transcription of the downstream structural genes [82]. We therefore hypothesized that the hypomodified tRNA^TrpCCA in the ΔmiaA mutant functionally mimics a state of L-tryptophan starvation despite sufficient intracellular tryptophan levels, thereby triggering a regulatory response similar to that seen under nutrient-limited conditions, as previously shown in E. coli [35, 36, 38].

In contrast to E. coli, the P. aeruginosa trpE and trpGDC genes are transcribed independently [37, 83]. In the PA14 strain, we identified two putative trpL leader sequences: one located upstream of trpE (PA14_07 940; referred to as trpL1) and another upstream of trpGDC (PA14_08340–08 360; referred to as trpL2). Similar to E. coli trpL, both P. aeruginosa leader sequences are enriched in tryptophan codons within an arginine-rich region (RWRWRA encoded by the sequence CGT-TGG-CGC-TGG-CGC-GCC) (Fig. 4a). To determine whether the predicted trpL leader peptides are translated in PA14 WT and to assess MiaA-dependent effects on their translation efficiency, we constructed translational fusions of each trpL sequence, including their native promoter, to a fluorescent reporter (P_trpL1-trpL1:msfGFP and P_trpL2-trpL2::msfGFP). Both trpL1 and trpL2 fusions produced robust fluorescence in PA14 WT (Supplementary Fig. S5). Notably, deletion of miaA resulted in a ~60% reduction in fluorescence for trpL1 and ~50% for trpL2, consistent with decreased translation efficiency in the absence of MiaA.

Figure 4.

Global reprogramming of L-tryptophan metabolism in the absence of MiaA and restoration upon mutagenesis of trpL leader peptides. (a) Transcriptomic and proteomic analysis of ΔmiaA trpL mutants. The genomic organization of trp structural genes in P. aeruginosa is shown. Conserved MiaA-dependent tryptophan codons in the two putative trpL leader peptides were substituted with MiaA-independent alanine codons (individually or in combination). Key branches of chorismate metabolism leading to L-tryptophan and pyocyanin biosynthesis are indicated. Heatmaps show differentially expressed genes and proteins involved in L-tryptophan and pyocyanin biosynthesis in ΔmiaA and the different ΔmiaA trpL codon substitution strains compared to the corresponding trpL codon substitutions in the PA14 WT background. Quantification of L-tryptophan (b) and anthranilic acid (c) using targeted LC–MS/MS analysis. Metabolite levels are shown as pmol normalized to the total amount of protein (mg). Data represent means ± SD of biological replicates (n = 2 for PA14 ΔmiaA; n = 3 for all other strains). (d) Quantification of pyocyanin after 8 h of growth in LB medium. Data represent means ± SD of biological replicates (n = 3). Volcano plots showing differentially expressed tRNAs in ΔmiaA compared to PA14 WT (e) and ΔmiaA trpL1* trpL2* compared to PA14 WT (f) with n = 2–3, |log₂FC| ≥ 0.6, adj. P-value <.01. Differentially expressed tRNAs are annotated. Statistical significance was determined by ANOVA (one-way) with ns = not significant, *P < .05, ****P < .0001.

We next engineered mutant strains carrying codon substitutions in the trpL leader peptides—either individually or in combination—in both the PA14 WT and ΔmiaA backgrounds. Specifically, tryptophan codons were replaced with alanine codons (GCC) at positions W10A and W12A in trpL1 (referred to as trpL1*), and at W18A and W20A in trpL2 (trpL2*) (Fig. 4a). We then performed transcriptomic and proteomic analyses in the ΔmiaA trpL substitution strains and compared differentially expressed genes and proteins with the respective trpL substitutions in the PA14 WT background (Fig. 4a, Supplementary Fig. S6, and Supplementary data ES1 and ES3). In the PA14 WT background, substitution of tryptophan codons in trpL1 did not affect the transcriptome, while codon substitutions in trpL2 resulted in the upregulation of trpGDC compared to PA14 WT with native trpL leader peptides, consistent with a disruption of the attenuation mechanism due to the loss of regulatory tryptophan codons in the attenuator region (Supplementary Fig. S6c–e).

Of note, the increased expression of trpE and elevated TrpE levels in the ΔmiaA background were restored in the ΔmiaA trpL1* background to the levels of the PA14 trpL1* strain, while in ΔmiaA trpL2* trpGDC and trpBA expression on the gene and protein level returned to the baseline level (PA14 trpL2*). Vice versa, the reduced expression of genes and proteins involved in pyocyanin biosynthesis could be restored to levels observed in the WT background upon replacement of tryptophan codons in either leader peptide, and more markedly in ΔmiaA trpL1* trpL2* (compared to PA14 trpL1* trpL2*).

Based on these observations, we next assessed intracellular L-tryptophan concentrations, which we expected to be elevated in the ΔmiaA strain and normalized in the ΔmiaA trpL1* and trpL2* mutant strains. We used targeted LC–MS/MS metabolomics to quantify selected intermediates of aromatic amino acid metabolism (Fig. 4b and c, and Supplementary Figs S7 and  S8). Consistent with our transcriptomic and proteomic data, L-tryptophan levels were significantly elevated in the ΔmiaA strain, supporting the conclusion that hypomodified tRNA^TrpCCA disrupts attenuation, resulting in L-tryptophan overproduction. Chromosomal complementation with a functional miaA gene fully restored L-tryptophan levels to those of the PA14 WT strain. Moreover, substitutions of the regulatory tryptophan codons in the trpL leader sequences normalized tryptophan levels in the ΔmiaA background—partially in the single leader peptide substitution mutants, and more completely when both trpL1 and trpL2 were modified (Fig. 4b). Similar trends were observed for other metabolites, including the tryptophan precursor anthranilic acid (Fig. 4c), the aromatic amino acids L-phenylalanine and L-tyrosine, as well as upstream metabolites shikimate and shikimate-3-phosphate (Supplementary Fig. S7). Of note, these metabolite levels remained unchanged in the PA14 WT background carrying the trpL codon substitutions (Supplementary Fig. S8a–d). In line with our previous observations, pyocyanin production was restored in the ΔmiaA trpL codon substitution strains and unaffected in the PA14 trpL mutants (compared to PA14 WT). (Fig. 4d and Supplementary Fig. S8e)

Absence of MiaA-dependent tRNA modifications alters tRNA abundance

To investigate whether changes in tRNA isopentenylatation are reflected in changes in the overall tRNA abundance, we quantified tRNAs from PA14 WT and ΔmiaA using nano-tRNAseq. Biological replicates showed low variability in both strains (Supplementary Fig. S9). Interestingly, a substantial number of tRNAs were significantly upregulated in the absence of miaA (|log₂FC| > 0.6, adj. P < .01) (Fig. 4 and Supplementary data ES5). Notably, many of the most highly expressed tRNAs are known targets of MiaA-dependent modification, including tRNA^TyrGUA, tRNA^TrpCCA, tRNA^SerGCU, tRNA^SerCGA, tRNA^SerUGA, tRNA^LeuCAA, tRNA^SecUCA. A particularly striking observation was the coordinated upregulation of the full set of tRNAs decoding serine and leucine codons. Similar to the regulation of L-tryptophan, L-leucine biosynthesis is a known subject of attenuation-mediated regulation across proteobacteria [84]. Targeted LC–MS/MS analysis revealed not only increased L-tryptophan levels but also confirmed accumulation of L-leucine in the ΔmiaA strain, which was restored to WT levels in the ΔmiaA trpL1* trpL2* strain (Supplementary Fig. S8f).

To further determine whether the global changes in tRNA composition in ΔmiaA depend on leader peptide-mediated attenuation, we sequenced the tRNA pool from ΔmiaA trpL1* trpL2* (Fig. 4f). Intriguingly, we no longer observed a significant upregulation of any of the tRNAs targeted by MiaA. Instead, only a few tRNAs were modestly upregulated, none of which were associated with MiaA-mediated tRNA modification.

MiaA influences P. aeruginosa motility and virulence independently of tryptophan attenuation

While the levels of amino acids, phenazine production, and tRNA abundance of the ΔmiaA strain were restored to WT levels upon replacement of tryptophan codons in the trpL leader peptides, other phenotypes—such as motility (Supplementary Fig. S10) and virulence in the G. mellonella infection model (Fig. 5a)—were not. Thus, altered virulence and motility are likely not a direct consequence of disrupted aromatic amino acid levels resulting from the impaired attenuation mechanism in the ΔmiaA strain. Instead, the absence of a functional MiaA appears to have additional effects on the expression of virulence genes that are independent of intracellular tryptophan levels. Confirming this, numerous virulence-associated genes and proteins were still significantly downregulated in the ΔmiaA trpL1* trpL2* mutant compared to PA14 trpL1* trpL2* (Supplementary Fig. S6a and b, and Supplementary data ES1). Examples are components of the T3SS machinery (exsBC, exoT, exoY, pcr2, pcrG, pcrV, popBD, pscN, pscQ), QS-controlled virulence factors (lasAB, phzABC, phzG, phzH, lecB, rhlAB), and additional secreted virulences factors (alkaline protease aprA, aprI).

Figure 5.

Role and condition-dependent expression of miaA in clinical P. aeruginosa strains. (a) Virulence of ΔmiaA trpL* mutants in the G. mellonella infection model. Ten G. mellonella larvae were infected with 100 CFUs per larva, and survival was monitored over 48 h (n = 3 biological replicates). The dashed gray line indicates the PBS control. (b) Conservation of the MiaA protein sequence across 6863 clinical P. aeruginosa isolates. Frequencies of amino acid identity at each position is shown relative to the PA14 reference sequence. The L133Q substitution (58%) represents the canonical sequence for the PAO1 lineage. A28V (6.7%), E89K (4.8%), and A134S (8.8%) variants occur in minor subpopulations, all additional substitutions occur at very low prevalence (<1%). The substrate binding site is indicated in blue, interaction sites with substrate tRNAs are indicated in purple. (c) Temporal dynamics of miaA transcription and protein abundance in PA14 WT during 10 h of growth in LB medium. mRNA expression and protein abundance values were normalized and are shown as z-scores (n = 4 biological replicates). (d) miaA expression dynamics across 77 clinical isolates under planktonic (PL) and biofilm (BF) conditions, shown as cpm. Conditional transcriptional changes are expressed as log₂-fold change (BF versus PL; right y-axis) for each isolate. Values <−0.5 are shown in blue (higher expression in PL) and values >0.5 are shown in red (higher expression in BF). Survival of infected G. mellonella larvae is displayed as a heatmap aligned with the corresponding isolates along the x-axis (% survival at 48 h p.i.). Isolates are consistently ordered across all data layers according to log₂-fold change (BF versus PL), enabling direct comparison of condition-dependent miaA expression, biofilm-associated regulation, and virulence.

Level of miaA gene expression correlates with virulence in clinical P. aeruginosa isolates

To assess the conservation and functional importance of miaA, we analyzed MiaA protein sequences in our collection of 6 863 fully sequenced P. aeruginosa isolates and found the amino acid sequence to be highly conserved across all isolates (Fig. 5b). Growth-phase-dependent analysis of miaA expression [68] and MiaA abundance in PA14 WT revealed temporal dynamics with an early activation of MiaA-dependent modification during exponential growth, followed by decreased activity in stationary phase (Fig. 5c). To further investigate the conditional importance of miaA across diverse genetic backgrounds, we examined miaA gene expression in a subset of 77 clinical isolates during planktonic growth (PL; early stationary phase) [65] and under biofilm conditions (BF) [50] (Fig. 5d). miaA expression varied among the strains, with both high- and low-expressing isolates particularly evident under biofilm growth conditions, whereas expression differences under planktonic conditions were more subtle. Moreover, an inverse trend was observed: isolates with higher miaA expression under biofilm conditions tended to show lower expression levels under planktonic conditions, and vice versa. Consistent with this trend, differential gene expression analysis (|log₂FC(BF versus PL)| > 0.5) revealed that miaA was upregulated during planktonic growth in ~30% of the strains (n = 25) and upregulated during biofilm growth in ~40% of the strains (n = 30). For the remaining ~30% of the strains (n = 22) miaA was unchanged between the two growth conditions.

We then integrated these expression data with virulence profiles from G. mellonella infection assays [44]. Virulent isolates were predominantly associated with elevated miaA expression under planktonic conditions, whereas avirulent strains clustered among strains with induced miaA expression during biofilm growth (Fig. 5d). This correlation between growth-state-specific miaA expression and virulence supports a model in which conditional miaA regulation contributes to isolate-specific infection strategies.

Discussion

In this study, we investigated the consequences of a miaA deletion in P. aeruginosa, revealing how the tRNA-modifying enzyme MiaA modulates bacterial phenotypes in this opportunistic pathogen. We characterized the tRNA modification landscape using LC–MS to quantify adenosine isopentenylations, and mapped the chemical modifications in a site-resolved manner using nanopore sequencing. Similar to Enterobacteriaceae, MiaA profoundly influenced P. aeruginosa behavior [16, 85, 86]. Infection-related phenotypes, including motility, biofilm formation and pathogenicity in the G. mellonella infection model, were significantly affected. To better understand how MiaA-dependent decoding defects might contribute to this pathogenicity attenuation, we performed a genome-wide codon-enrichment analysis based on MiaA-sensitive codons. This approach identified several genes significantly enriched in MiaA-dependent codons that are involved in virulence-associated processes, such as phospholipases, type IV pili-mediated attachment, biofilm matrix components, genes involved in lipopolysaccharide/O-antigen modification as wells as respiratory adaptation. These findings suggest that MiaA loss can directly impair translation of specific infection-relevant functions. At the same time, major quorum-sensing regulators and T3SS genes were not significantly enriched in MiaA-sensitive codons, indicating that the pronounced repression of virulence pathways at the transcriptome, proteome, and translatome levels cannot be explained solely by direct codon-dependent decoding defects in canonical virulence regulators.

To explore this link more directly, we examined the subset of translationally affected genes with a focus on regulatory elements, where subtle translational perturbations may be transcriptionally amplified through downstream regulatory cascades. While no statistically significant enrichment of regulatory categories was detected in our enrichment analysis, the PseudoCAP categories transcriptional regulators combined with two-component systems nevertheless represented one of the largest functional groups among translationally affected genes (7.2% of affected genes). The most prominent examples of global regulators include the stationary phase sigma factor RpoS, the Rhl QS regulator RhlR and the Las QS repressor RsaL.

The translational consequences of MiaA loss extended beyond the codons directly decoded by i⁶A-modified tRNAs. Translation efficiency was altered not only at MiaA-dependent codons but also at several MiaA-independent codons (Supplementary Fig. S3), and notably, several of the affected MiaA-independent codons belong to the same amino acid families as MiaA-dependent codons, such as Ser and Leu. Given the altered tRNA pools we observed for tRNA^Ser and tRNA^Leu families, competition between isoacceptor tRNAs for overlapping codon sets may account for these effects. Additionally, the profound changes in amino acid homeostasis in the ΔmiaA mutant could affect tRNA charging ratios more broadly, while ribosome stalling at MiaA-dependent codons may propagate to neighboring positions through ribosome queuing. Furthermore, recent large-scale tRNA profiling in P. aeruginosa revealed extensive connectivity between modification pathways and suggested that tRNA modifications form regulatory networks [32]. The observed substantial changes in the abundance of multiple tRNA species are in line with this study and indicate that perturbation of a single modification pathway can have broader consequences for tRNA homeostasis.

A striking downstream consequence of impaired MiaA-dependent decoding was the rewiring of tryptophan biosynthesis. In P. aeruginosa ribosome stalling at the trpL leader peptide led to a significant upregulation of aromatic compound production in the miaA mutant. Replacing the four tryptophan codons in both P. aeruginosa trpL leader peptides with alanine codons restored tryptophan production in ΔmiaA to WT levels. Elevated tryptophan and derivatives were accompanied by increased levels of the respective tRNAs, indicating that cells adjust tRNA abundance to altered translational demands. Consistent with this idea, in yeast under stress, the tRNA pool is dynamically rearranged, thereby selectively affecting translation of stress-related transcripts [87]. In a bacterial context, tRNA upregulation has been observed during extended sub-culturing under antibiotic exposure, suggesting that increased tRNA abundance can promote adaptation [88].

We propose the following model to explain how MiaA connects tRNA modification, amino acid levels, and tRNA abundance. Loss of i⁶A modification impairs decoding at cognate codons, causing ribosome stalling at leader peptides of amino acid biosynthetic operons, as demonstrated here for trpL and plausibly extending to leucine biosynthesis, given that attenuation-based regulation of leu operons is widespread in bacteria [89, 90] and we observed elevated intracellular leucine levels. The resulting derepression of biosynthesis elevates intracellular amino acid pools. In turn, elevated amino acid levels may contribute to high tRNA charging ratios, which protect tRNAs from degradation pathways that are normally activated when uncharged tRNAs accumulate during amino acid limitation [91], thereby raising their steady-state abundance. This interpretation is consistent with the observation that tRNA pools returned to WT levels upon restoration of tryptophan homeostasis through codon replacement in the trpL leader peptide, which breaks the cycle at the amino acid accumulation step. However, a subsequent methodological re-evaluation by the same research group questioned the extent of starvation-induced tRNA degradation and suggested that some of the reported effects may have resulted from RNA purification artifacts [92].

An alternative explanation is that the altered tRNA pools represent a compensatory response to impaired decoding of MiaA-dependent codons. A previous study demonstrated that overexpression of tRNA^LeuCAA partially suppresses the requirement for MiaA during translation of leucine UUX codons in E. coli, suggesting that increased tRNA abundance can compensate, at least in part, for the loss of i⁶A-dependent decoding efficiency [17]. The coordinated increase in several MiaA-targeted tRNAs observed here may therefore reflect a cellular adaptation aimed at maintaining translational capacity under conditions of impaired codon decoding. Whether transcriptional upregulation of tRNA genes is indeed compensatory remains, however, to be determined.

Loss of MiaA reshapes translation not only by affecting codon-biased genes but also by altering amino acid pools and tRNA availability, thereby modifying ribosome flux and potentially reinforcing stress responses and regulatory rewiring. Consistently, MiaA expression in P. aeruginosa was growth-phase dependent—high during early and exponential phases and reduced in stationary phase. Notably, virulence of clinical isolates correlated with miaA expression: highly virulent strains showed high planktonic miaA levels that dropped under biofilm conditions. Together, our findings highlight that the regulatory mechanisms defining the P. aeruginosa virulence phenotype extend beyond the mere expression of virulence genes: they also encompass dynamic adaptation of cellular metabolic states and reprogramming of translational landscapes. Our results establish MiaA as a critical factor for maintaining bacterial fitness and virulence, and, given its structural distinction from eukaryotic homologs [93, 94], highlights its potential as an attractive target for the development of novel pathoblockers.

Supplementary data

Supplementary data is available at NAR online.

Conflict of interest

None declared.

Funding

This work was supported by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany’s Excellence Strategy, EXC 2155 “RESIST”—Project-ID 390874280. S.H. received funding within the SFB/TRR-298-SIIRI—Project-ID 426335750 and in the SPP2389 (HA 3299/9-1, AOBJ: 687646), from the Ministry of Science and Culture of Lower Saxony (Niedersächsisches Ministerium für Wissenschaft und Kultur), BacData ZN3428, and from the Novo Nordisk Foundation (NNF 18OC0033946). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Funding to pay the Open Access publication charges for this article was provided by NNF 18OC0033946.

Data availability

RNA-seq and Ribo-seq data generated in this study have been deposited in the NCBI Gene Expression Omnibus under accession numbers GSE323095 and GSE322698. Previously published time-resolved expression profiles of PA14 WT [68] and RNA-seq data for clinical isolates [50, 65] are available under accession numbers GSE159698, GSE134231, and GSE123544. Proteomics have been deposited in the PRIDE (PRoteomics IDEntifications Database) under the accession number PXD076088, and nano-tRNA-seq data have been deposited in the European Nucleotide Archive (ENA) under the accession numbers PRJEB83028 (PA14 WT) and PRJEB111160 (ΔmiaA mutants), respectively. All data will be made publicly available upon publication.