Introduction

Pseudomonas aeruginosa is a highly adaptable opportunistic pathogen. It exhibits notable metabolic flexibility, genomic variability and phenotypic diversity1. Due to its ability to survive in diverse environments, along with its acquired and innate antibiotic resistance mechanisms, and its capacity to form biofilms, there is an urgent need to discover new therapeutic strategies1,2. In this context, one promising therapeutic target is ribonucleotide reductase (RNR), a family of metalloenzymes responsible for the de novo synthesis of deoxyribonucleotides (dNTPs) through a radical-based chemistry mechanism3. Three main classes of RNRs have been described: class I (subtypes Ia, Ib, Ic, Id and Ie)4, Class II and Class III3. This classification is based on structural differences, oxygen requirements, metallocofactor dependencies, and the mechanisms used for radical generation3,5,6. P. aeruginosa is one of the few organisms known to encode all three main classes, reflecting the finely tuned gene expression control systems of this microorganism6. Although class I RNRs are among the best-characterized enzymes, little is known about class Ia in P. aeruginosa7. This RNR class consists of two homodimeric subunits (α and β), encoded by the nrdAB operon. To date, two transcriptional factors (AlgR and NrdR) have been identified as regulators of the operon6,8,9. However, key aspects of the expression control mechanisms of nrdAB operon remain unclear.

Several years ago, a 5’ untranslated region (5’ UTR) upstream nrdA was predicted through bioinformatic analysis8, but further experimental validation is needed to confirm its structure and regulatory role. The 5’ UTRs of bacterial mRNAs have increasingly been recognized as crucial expression-modulating elements that influence gene expression at multiple levels. Rather than being passive sequences between promoters and start codons, 5’ UTRs harbour a wide variety of regulatory elements that actively control translation efficiency, transcript stability and transcription termination10,11. Their functional diversity includes ribosome binding sites (RBS) or Shine-Dalgarno (SD) sequences, upstream open reading frames (uORFs), binding motifs for small non-coding RNAs (sRNAs), and dynamic secondary structures capable of responding to environmental or metabolic signals, among others12,13. Changes in 5’ UTR sequence or structure can alter gene expression by affecting ribosome accessibility, modulating mRNA degradation, or engaging in complex networks with trans-acting elements such as sRNAs and RNA-binding proteins14,15. For example, riboswitches within 5’ UTRs bind metabolites, ions, or coenzymes to control downstream gene expression11,16. These spatial and temporal mechanisms highlight the role of 5’ UTRs not only in transcriptional regulation but also as central elements in post-transcriptional control11. Notably, 5’ UTRs do not function independently, as recent studies suggest that riboswitches, sRNAs, antisense RNAs, and 5’ UTRs themselves can regulate one another, forming intricate regulatory networks15. Despite their importance, many 5’ UTRs remain poorly characterized, and their potential as tunable expression modules in bacterial gene expression is only beginning to be uncovered13.

Therefore, this study aims to characterize the predicted 5’ UTR of the nrdAB operon in P. aeruginosa, and to assess its potential role in transcript stability, relative protein levels, and gene expression dynamics under different conditions.

Results and discussion

The nrdAB operon encodes a long 5’ untranslated region (5’ UTR) with potential regulatory function

The oxygen-dependent class Ia ribonucleotide reductase (RNR) of Pseudomonas aeruginosa is encoded by two cotranscribed genes, nrdA and nrdB17. In a previous study conducted by our group, several regulatory features of the nrdAB operon were identified. Transcriptional binding sites for AlgR and NrdR transcription factors were mapped6,8 (Fig. 1a), and bioinformatic analyses revealed that the transcription start site (TSS) of nrdAB is located approximately 400 bp upstream of the translational start codon (ATG), indicating the presence of a long 5’ UTR6 (Fig. 1a-b, yellow triangle). Based on this, the putative − 10 and − 35 promoter boxes of the nrdAB operon were mapped (Fig. 1a), consistent with recognition by the housekeeping sigma factor σ⁷⁰.

Fig. 1
figure 1

Analysis of the nrdA upstream region and identification of long 5’ untranslated region (5’ UTR). (a) Schematic representation of the nrdAB promoter, 5’ UTR, and its genetic context. The AlgR transcriptional factor binding site is shown in green, and NrdR binding sites are indicated in red. The ribosome binding site (RBS), translational start codon (Met, ATG), and the − 35 and − 10 promoter regions are underlined. Transcriptional start sites (TSSs) identified by different methods are marked with coloured triangles, as detailed in panel (b). The long 5’ UTR of nrdA (408–411 bp) is highlighted in light orange. (b) Identification of TSSs using various approaches, including bioinformatic analysis, primer extension and 5’rapid amplification of cDNA ends (5’ RACE) performed in our laboratory, as well as transcriptomic data from a previously published differential RNA-seq (dRNA-seq) experiment (PRJNA169508)59. Coloured triangles correspond to the TSS identified by each method and are mapped in panel (a). (c) Primer extension analysis showing the identified TSS (blue triangle in panels (a-b)), marked with an asterisk (*). (d) 5’RACE analysis of the nrdA, with the identified TSS (pink triangle in panels (a,b)). The percentage indicates the proportion of sequenced colonies that matched the same region, with the number in parentheses corresponding to the exact number of colonies showing that specific base pair. The original and cropped version of the gel are provided in the Supplementary Material Figure S6 and are also available in the public repository (https://doi.org/10.34810/data2361).

Full size image

To experimentally validate the TSS of the nrdAB mRNA, we combined laboratory data with genome annotation databases, which consistently placed the TSS between 408 and 411 bp upstream of the translational start codon (Fig. 1a-b). A primer extension assay confirmed that the 5’ end of the nrdAB transcript begins with a cytosine nucleotide located 411 bp upstream of the predicted translational start codon (Fig. 1a-c, blue triangle), while 5’Rapid alignment of cDNA ends (5’RACE) positioned the TSS 409 bp upstream of the predicted translational start codon (Fig. 1a-b, d, pink triangle), corroborating the presence of a long 5’ UTR.

Since bacterial 5’ UTRs are typically short (20–30 nucleotides), the presence of such an extended region suggests a potential regulatory function18. Motivated by its unusual length, we aimed not only to investigate whether the 5’ UTR contributes to the modulation of nrdA expression within the nrdAB operon, but also to explore the nature of this region by analysing the presence of potential regulatory elements.

The 5’ UTR is unlikely to act as a regulatory element responsive to coenzyme B12

Long 5’ UTRs have been associated with various regulatory elements18,19, among which riboswitches are particularly notable. The vast majority of riboswitches are located within the 5’ UTRs of mRNA20. A bioinformatic analysis using Riboswitch Scanner21 identified a potential cobalamin riboswitch within the 5’ UTR of nrdA (Supplementary Material Figure S1). All riboswitch classes available in the software’s database were tested, but only the cobalamin riboswitch yielded a positive prediction. Interestingly, the nrdABS operon, which encodes the class Ia RNR in Streptomyces, is regulated by a cobalamin riboswitch22. This is not an isolated case; several RNRs, including the class Ia enzyme from Bacillus halodurans, are also regulated by B12 riboswitches23. Given this context, we investigated whether a similar regulatory element might be present in the 5’ UTR of the P. aeruginosa nrdA gene. Furthermore, a known connection exists between coenzyme B12 and RNR activity, as demonstrated by the class II RNR from Pseudomonas aeruginosa, which, like other class II RNRs, relies on the availability of S-adenosylcobalamin (coenzyme B12)24.

Cobalamin riboswitches constitute a diverse family characterized by a highly conserved secondary structure and a sequence signature known as the B12 box25. To investigate the presence of a putative B12 box, a multiple sequence alignment was performed between the predicted P. aeruginosa nrdA cobalamin riboswitch and several bioinformatically predicted cobalamin riboswitches in P. aeruginosa, obtained from the RegPrecise database26. As shown in Fig. 2a, the consensus consistency score (cons) was 47, obtained using T-coffee alignment tool27. Given that all the sequences in this alignment were bioinformatically predicted and correspond to different genes from P. aeruginosa, high sequence conservation was not expected, and the relatively low alignment score is not surprising. However, the nrdA sequence contains a region that partially matches the B₁₂ box consensus motif as well as other conserved features annotated in Fig. 2a, which may support its classification as a putative cobalamin riboswitch.

Fig. 2
figure 2

Analysis of the cobalamin riboswitch in the nrdA 5’ UTR: motif-alignment and experimental exploration. (a) Multiple sequence alignment between PA1156 (nrdA, catalytic component of ribonucleotide reductase class Ia) and several bioinformatically predicted cobalamin riboswitches in P. aeruginosa: PA2911 (btuB, probable TonB-dependent receptor), PA1271 (btuB2, Probable TonB-dependent receptor), PA2946 (cbtA, hypothetical protein), PA2906 (cobG, probable oxidoreductase) and PA2945 (cobW, conserved hypothetical protein). Positions are numbered relative to the translational start codon (ATG). Predicted secondary structure elements, including loops and hairpins, are annotated based on structural alignments described in a previously published study29. Sequence conservation across signature helices (0 to 6 arrows) is colour-coded. The underlined region corresponds to the sequence in Supplementary Material Figure S1. All sequences are based on computational predictions; alignments reflect potential motif conservation rather than exact sequence identity. Consistency scores generated by T-Coffee are shown and represent the reliability of base-to-base alignment across sequences. Scores range from 0 to 100, with higher values indicating greater confidence in the alignment at a given position. These scores reflect alignment quality only and do not imply structural or functional conservation. Individual scores for each gene are provided. (b,c) Transcriptional and translational GFP reporter assays of nrdA with full 5’ UTR (PnrdA) and truncated 5’ UTR (PnrdA-Δ5’ UTR) in P. aeruginosa PAO1 wild-type, using pETS130-GFP derivatives. Assays were performed under varying concentrations or absence of coenzyme B12. Relative fluorescence units (RFU), normalized to OD600 (Optical density (OD) was measured at 600 nm) were recorded over a 10-hours period following coenzyme B12 addition. Symbols in the graph represent different concentrations of coenzyme B12: □ 1 µg/mL, ○2.5 µg/mL and △10 µg/mL).

Full size image

To validate the presence of a functional B12 riboswitch, transcriptional and translational GFP fusions of nrdA were constructed, and GFP reporter assays were performed. These constructs were used as B12 sensors to evaluate the presence of a functional riboswitch and the effect of coenzyme B12 on downstream gene expression. For simplicity, we refer to the construct containing the nrdA promoter, 5′ UTR region, and a short fragment of the coding sequence as PnrdA, and to the version with a truncated 5′ UTR as PnrdA-Δ5’ UTR. Although we observed differences in expression from PnrdA at increasing concentrations of coenzyme B12 in both the transcriptional (Fig. 2b) and translational (Fig. 2c) fusions, similar changes were also observed even in the PnrdA-Δ5’ UTR. This suggests that the observed effect is not mediated by the predicted 5′ UTR but rather reflects broader physiological changes induced by coenzyme B₁₂24,28. To confirm the responsiveness of our system to coenzyme B12, we included a transcriptional GFP fusion of PA2945 (cobW), a gene bioinformatically predicted to contain a cobalamin riboswitch (RF00174). The cloned region comprises cobW promoter, the predicted cobalamin riboswitch sequence29, and a short fragment of the cobW coding sequence, and is referred to as PcobW throughout the text. In this case, we detected a significant decrease in expression upon coenzyme B₁₂ addition during both exponential (EX) and stationary growth phases (ST) (Supplementary Material Figure S2), supporting the sensitivity of our system at the transcriptional level. In parallel, we also analysed PnrdA and PnrdA-Δ5’ UTR under the same conditions and time points. No significant decrease in expression was observed upon B₁₂ addition. Additional control constructs were tested to evaluate B12-dependent regulation at the translational level, but these did not yield conclusive results, representing a limitation of the study.

Although translational regulation appears to be the predominant mechanism among cobalamin riboswitches in γ-Proteobacteria29, the precise mechanisms of coenzyme B12 binding and downstream regulation remain poorly understood30. Additionally, cobalamin riboswitches are a complex family capable of sensing and responding to a range of corrinoid compounds, although some respond only to specific subsets of these compounds31. Furthermore, as previously discussed, the sequence and structure of the nrdA putative riboswitch region did not perfectly align with other bioinformatically predicted B12 riboswitches, which exhibited a higher consensus score (Fig. 2a).

In conclusion, despite the lack of validated riboswitch-positive controls for P. aeruginosa in the literature and the inherent difficulty of experimentally confirming riboswitch function, our findings do not support the existence of a functional B₁₂-responsive riboswitch within the 5′ UTR of the nrdAB operon. Under the tested conditions, expression patterns observed at the transcriptional level suggest that this long 5′ UTR does not mediate regulation in response to coenzyme B₁₂. However, we cannot entirely rule out the possibility of regulation occurring at the translational level, particularly given the limitations in control strains and tools available for reliably evaluating this mechanism in P. aeruginosa.

Exploration of additional regulatory elements within and acting upon the 5’ UTR of nrdAB

Given that our data indicate the nrdA 5′ UTR does not function as a cobalamin riboswitch—at least at the transcriptional level—we investigated alternative regulatory elements that may reside within or act upon this region. We explored upstream open reading frames (uORFs), small non-coding RNAs (sRNAs) encoded within the 5’ UTR or elsewhere in the genome.

uORFs are short open coding sequences often involved in post-transcriptional regulation through ribosome-mediated translational attenuation11. To assess the non-coding nature of the nrdAB 5’ UTR, both bioinformatic and experimental approaches were employed. As shown in Fig. 3a, a translated BLAST (tBlastn) using the entire 5’ UTR sequence against the core nucleotide database (core nt) yielded less than 30% coverage to hypothetical proteins from various P. aeruginosa strains, further indicating this 5’ UTR sequence is unlikely to encode a functional uORF. Similarly, using the CPC2 tool32, the full 411 bp 5’ UTR sequence was predicted to encode a short peptide of 49 amino acids although its coding potential was predicted to be low. To further evaluate this CPC2 prediction, a BLASTp33 search was performed against the non-redundant protein database (nr), which revealed no significant similarity to known proteins, supporting the notion that the predicted peptide is unlikely to be functional. To experimentally validate these bioinformatic predictions, we tested the possibility of translation in both the 5′–3′ and reverse orientations from the 5′ UTR. The full 5′ UTR sequence was cloned into a plasmid under the control of an arabinose-inducible promoter in both orientations. Based on this nucleotide length, the full 5’ UTR sequence could theoretically encode a protein of approximately 15 kDa, although CPC2 predicted a smaller peptide corresponding to less than 10 kDa. As shown in Fig. 3b, arabinose induction did not lead to detectable protein bands at either the expected sizes (~ 10 or ~ 15 kDa), and the overall protein expression profiles remained similar between induced and non-induced conditions in both orientations. Although the resolution of the gel in the low molecular weight range (Fig. 3b) is limited and does not entirely rule out the presence of very small peptides, the combined bioinformatic and experimental evidence suggests that the 5’ UTR is unlikely to encode a functional uORF.

Fig. 3
figure 3

The nrdA 5’ UTR region as a potential reservoir of upstream open reading frames (uORFs), small non-coding RNAs (sRNA) and sRNA binding sites. (a) In silico analysis using the Coding potential calculator (CCP2) and BLAST tools. A 49-amino acid peptide predicted by CCP2 was analysed via Blastp, and the entire 5’ UTR sequence was examined using tBlastn. (b) SDS-PAGE (4–20% acrylamide) of crude extracts from P. aeruginosa PAO1 WT harbouring pJN106 (5’-3’ orientation from the 5’ UTR) and pJN107 (3’-5’ orientation from the 5’ UTR), before (not-induced) and after induction with 0.4% L-arabinose. (c) Bioinformatic, bibliographic and experimental exploration of potential regulatory elements across the nrdA upstream and downstream regions. These include putative promoters or σ70 binding sites (P1 and P2, arrows), transcription start sites (TSSs, coloured triangles), Rho-independent terminators (hairpin structure), small RNAs (sRNA, curved red arrow) and sRNA binding sites (curved purple line with tick marks)43,44,46,47. Positions are shown relative to the translation start codon (ATG, + 1) with original genome coordinates also provided. The original and cropped version of the gel are provided in the Supplementary Material Figure S6 and are also available in the public repository (https://doi.org/10.34810/data2361).

Full size image

In addition to investigating uORFs, we next focused on small RNAs (sRNAs), another regulatory element that may reside within or exert influence upon the nrdAB 5’ UTR. They can be cis-encoded (encoded on the antisense strand of their targets) or trans-encoded (originating from distant genomic loci)34. sRNAs are typically transcribed from independent or nearby promoters and often terminate via Rho-independent mechanisms34,35,36. They can modulate gene expression at multiple levels, including transcription, translation, mRNA stability, and even DNA maintenance or silencing11,37. Most sRNAs act through base-pairing interactions with coding sequences or UTRs. However, alternative mechanisms have also been described34,35,36,37. As part of this analysis, we examined the upstream and downstream regions of the nrdA coding sequence for potential transcriptional and post-transcriptional regulatory elements, including transcription start sites (TSSs), σ70 binding sites (promoters), Rho-independent terminators, annotated sRNAs, and their putative interactions. These analyses aimed to uncover regulatory features that might indicate the presence of previously uncharacterized small non-coding RNAs beyond those described in public databases or prior transcriptomic studies. Promoter prediction tools such as PromoterHunter38 and BPROM39 revealed two potential σ70 binding sites: one previously characterized (Fig. 1a, purple underlined sequences; Fig. 3c, P1 black arrow), and a second located 226 bp upstream of the translational start codon, located within the 5’ UTR (Fig. 3c, P2 grey arrow). Analysis using ARNold software40 predicted a single Rho-independent terminator at the end of the nrdAB operon (Fig. 3c). Although searches in Rfam41 and RNAcentral42 databases did not reveal annotated non-coding RNAs matching this region, RNA-seq data identified a putative sRNA named pant12543. This sRNA maps to the negative strand between coordinates 1,254,329–1,254,720 and overlaps the nrdAB 5’ UTR region (Fig. 3c). It was initially annotated as intergenic, likely due to annotation criteria that may underestimate the regulatory potential of 5’ UTRs. Notably, no additional TSSs or terminators were found near the pant125 locus (Fig. 3c), raising the possibility that pant125 may originate from the nrdA transcript itself. However, this hypothesis remains speculative and would require experimental validation, such as strand-specific RNA detection or Northern blotting, to confirm its transcriptional origin and potential regulatory role.

We next investigated whether this 5’ UTR could instead serve as a regulatory target of trans-encoded sRNAs through base-pairing interactions. Using PresRAT software44, RNAcentral database42, and literature mining, only two sRNA candidates were identified: P2445 and PhrS46,47. P24, detected via PresRAT between coordinates 1,254,295 to 1,254,261 of P. aeruginosa PAO1 genome, shared partial similarity in RNAcentral database42 but was discarded due to failure to align to the expected coordinates on PAO1 genome. The second sRNA, PhrS, was reported in a RIL-seq dataset to interact with nrdA in an Hfq-dependent manner, specifically between coordinates 1,251,452–1,251,176. This interaction showed an odds ratio of 8.4, which is indicative of biological relevance, even though the interaction site lies outside the annotated 5′ UTR of nrdA47. Notably, the same study reported approximately 800 RNA-RNA interactions involving nrdA, suggesting a complex network of possible regulators47. In P. aeruginosa, approximately 150 distinct sRNAs have been identified, although estimates from different approaches point to potentially higher numbers11. Despite advances in discovery technologies, identifying sRNAs, elucidating their biological functions, and determining their target genes remain significant challenges11,47,48.

In summary, our bioinformatic and experimental analyses did not reveal any definitive regulatory elements within the 5’ UTR region of nrdA under the approaches and conditions tested. Importantly, our data suggest that the 5’ UTR is unlikely to encode a functional peptide, thereby reinforcing its characterization as non-coding RNA element. Nonetheless, we cannot entirely exclude the hypothesis that pant125 or another sRNA derived from the nrdA transcript could play a regulatory role, as previously reported43,49. Moreover, although transcriptomic predictions have proposed hundreds of potential sRNA–mRNA interactions involving nrdA47, no studies have yet demonstrated any specific interaction between trans-encoded sRNA and the 5’ UTR capable of modulating gene expression. This remains an unexplored regulatory possibility and will require dedicated investigation. These findings underscore the complexity of bacterial regulatory RNA networks and highlight the need for complementary functional and transcriptomic approaches in future studies to fully uncover the role of long 5’ UTRs and their associated elements. This reinforces the importance of integrating computational predictions with experimental validation to fully understand the regulatory landscape of bacterial 5’ UTRs.

The 5’ UTR is involved in reducing nrdA transcript levels

Given the complexity of 5’ UTR-mediated regulation —including elements such as riboswitches, sRNAs, and uORFs—we investigated whether the 5′ UTR of nrdA modulates its expression beyond the mechanisms previously attributed to the transcriptional regulators AlgR and NrdR6,8.

To assess the functional relevance of the 5’ UTR, transcriptional GFP fusions were constructed containing the nrdA upstream region, including the promoter, either with the full-length 5’ UTR (PnrdA, pETS134) or with truncated version of the 5′ UTR (PnrdA-Δ5′ UTR, pETS257). As shown in Fig. 4a, under aerobic conditions, the presence of the 5’ UTR (PnrdA) led to reduced nrdA expression compared to PnrdA-∆5’ UTR, which exhibited significantly higher levels. Interestingly, this repressive effect was also observed under anaerobic conditions (Fig. 4a, right), despite class Ia RNR being enzymatically active primarily under aerobic conditions3,5. Expression differences associated with growth phase were also evident. Constructs containing the 5′ UTR (PnrdA) showed reduced expression during the stationary phase compared to exponential growth under both oxygen conditions. In contrast, the PnrdA-Δ5′ UTR construct exhibited increased expression during the aerobic stationary phase, while under anaerobic conditions, its expression pattern resembled that of the full-length construct (Fig. 4a, blue bars).

Fig. 4
figure 4

Analysis of nrdA expression using GFP reporter assay and quantitative reverse transcription PCR (qRT-PCR) under aerobic and anaerobic conditions during exponential and stationary growth phases. (a) nrdA expression measured as relative fluorescence units (RFU), normalized to OD600 (RFU/OD600) from GFP transcriptional reporter fusions (pETS130-GFP derivatives) containing either the full-length 5’ UTR (PnrdA) or truncated version of 5’ UTR (PnrdA- Δ5’ UTR), under aerobic (left) and anaerobic conditions (right), during exponential (EX) and stationary (ST) growth phases. (b) Fold change in nrdA transcript levels measured by qRT-PCR in P. aeruginosa CRISPR-Cas9-engineered chromosomal 5’ UTR mutant (PAO1 Δ5’ UTR) compared to the wild-type strain (PAO1 WT), under aerobic (left) and anaerobic (right) conditions during both exponential and stationary growth phases. Comparisons were made between genotypes within each growth phase, and data were normalized accordingly to the WT sample in each condition to emphasize intra-phase differences. Dashed lines indicate the separation between exponential and stationary phases. (c) Fold change in nrdA transcript levels measured by qRT-PCR in P. aeruginosa PAO1 WT and the isogenic Δ5′ UTR strains, comparing exponential (EX) and stationary (ST) under aerobic (left) and anaerobic (right) conditions. Data were normalized to the exponential (EX) phase sample of each genotype to highlight intra-genotype variation. Dashed lines indicate the separation between WT and Δ5′ UTR strains. Significance: p-value < 0.0001(****). Bars are grouped by condition and genotype to facilitate intra-group comparisons.

Full size image

To validate these findings in a physiological context, nrdA expression was compared between P. aeruginosa PAO1 wild-type (PAO1 WT) and a CRISPR-Cas9-engineered chromosomal 5’ UTR mutant (PAO1 ∆5’ UTR) across all growth conditions. Figure 4b focuses on genotype-specific comparisons within each condition, revealing consistently higher nrdA expression in the mutant under both aerobic (Fig. 4b, left) and anaerobic (Fig. 4b, right) environments in agreement with the GFP reporter assay results (Fig. 4a). Figure 4c explores expression dynamics across growth phases within each genotype. Specifically, comparisons were made between exponential and stationary phases for both WT and Δ5′ UTR strains under aerobic and anaerobic conditions. In the WT strain, nrdA levels declined during the stationary phase, whereas in the Δ5′ UTR mutant, expression remained relatively stable regardless of growth phase or oxygen availability, mirroring the trends observed with the GFP reporter system.

Together, these data underscore the regulatory complexity of class Ia RNR in P. aeruginosa, influenced by both growth phase and oxygen availability. The results suggest that the 5′ UTR of nrdA contributes to transcript repression, particularly during the stationary phase, when dNTP synthesis demand is reduced due to limited DNA replication.

The 5’ UTR modulates the stability and decay of nrdA transcript

Having established that the nrdA 5’ UTR contributes to reduced transcript levels by repressing gene expression (Fig. 4), we next examined its role in post-transcriptional level. Given that 5’ UTRs often influence mRNA stability, we hypothesized that its presence may promote transcript degradation11,14.

To test this, we employed two complementary strategies based on the same methodological framework. The first relied on a plasmid-based system (pETS130) which includes the nrdA promoter and either the full (PnrdA, pETS134) or truncated (PnrdA-∆5’ UTR, pETS257) 5’ UTR, transcriptionally fused to cat (chloramphenicol acetyltransferase) and gfp. In this setup, the genetic background remained constant (P. aeruginosa PAO1 WT), with the only variable being the 5’ UTR sequence within each plasmid construct (PnrdA or PnrdA-∆5’ UTR). To specifically assess the impact of the 5’ UTR on transcript stability, we used cat mRNA levels as an indirect measure of nrdA mRNA decay. This strategy was necessary due to technical limitations: the chromosomal nrdA copy —wild-type and thus containing the full 5′ UTR—prevented independent evaluation of truncated variant. Under these conditions, cat served as the only reliable proxy for comparing transcript decay between constructs differing solely in 5’ UTR content. The second approach focused on measuring the endogenous nrdA mRNA half-life directly from the chromosome in both P. aeruginosa PAO1 WT and PAO1 Δ5’ UTR strains. All qRT-PCR data were normalized using 16S rRNA as the internal reference gene.

As shown in Fig. 5a, transcript from the construct with truncated 5’ UTR (WT + PnrdA-Δ5’ UTR) exhibited a slower degradation rate (K = 0.07 min− 1; 95% CI: 0.03768 to 0.1352 min− 1) compared to the construct with the full 5’ UTR (WT + PnrdA: K = 0.21 min− 1; 95% CI: 0.1876 to 0.2329). These values corresponded to estimated half-lives of 9.53 min (95% CI: 5.13–18.39 min) for the truncated 5′ UTR transcript, and 3.32 min (95% CI: 2.98–3.70 min) for the full-length 5′ UTR transcript, indicating a three-fold increase in stability (Fig. 5c). In the chromosomal setup (Fig. 5b), nrdA mRNA in the Δ5’ UTR strain showed a reduced decay rate (K = 0.83 min− 1; 95% CI: 0.7624 to 0.8988 min− 1) compared to the WT strain (K = 1.5 min− 1; 95% CI: 1.404 to 1.603 min− 1), with half-lives of 0.84 min (95% CI: 0.7712 to 0.9091 min) for the Δ5’ UTR and 0.46 min (95% CI: 0.4323 to 0.4938 min) for the WT, indicating a two-fold increase in transcript stability (Fig. 5d).

Fig. 5
figure 5

Analysis of nrdA and nrdJ mRNA stability using transcriptional shut-off in plasmid-based and chromosomal systems. (a) Decay curves and retrotranscribed transcript levels of nrdA and nrdJ following rifampicin treatment at 0, 5 and 20 minutes in P. aeruginosa PAO1 WT carrying pETS130-GFP derivatives. Constructs include either the full 5’UTR (blue, WT + PnrdA) and truncated 5’ UTR (pink, WT + PnrdA- Δ5’ UTR). In this plasmid-based system, cat mRNA, located downstream of the promoter and 5’ UTR, was used as proxy to estimate nrdA transcript stability. This approach was necessary due to technical limitations in directly quantifying plasmid-derived nrdA. nrdJ levels were measured directly from the chromosomal copy and are shown in green and yellow (WT + PnrdA and WT + PnrdA-Δ5’ UTR, respectively). (b) Decay curves and retrotranscribed transcript levels of nrdA and nrdJ following rifampicin treatment at 0, 2-, 6-, 10- and 20-min in P. aeruginosa wild-type (WT) and PAO1 Δ5’ UTR (Δ5’ UTR). (c,d) One-phase exponential decay model fitted to the data to calculate best-fit mRNA half-life values (in minutes, min) and decay rate constant (k, in minutes, min−1). In panel (c), nrdA values were derived from cat mRNA (proxy), while nrdJ was measured directly from the chromosome (endogenous). In panel (d), both nrdA and nrdJ were quantified from their endogenous chromosomal copies. All qRT-PCR data were normalized to the endogenous control 16S rRNA. The dashed line represents the 50% threshold used to determine mRNA half-life. Decay parameters were calculated using nonlinear regression with a one-phase exponential model, as described in Materials and Methods.

Full size image

To complement the analysis and analyse the specificity of the 5’ UTR effect on nrdA, we examined the decay profile of nrdJ, a chromosomally encoded class II RNR not affected directly by the 5’ UTR manipulation. Although nrdJ was not part of the plasmid-based system, its decay profile was evaluated in both experimental systems to verify that 5’ UTR-driven regulation did not produce secondary effects on unrelated transcripts. In contrast to the chromosomal comparison of nrdA and nrdJ between WT and Δ5′ UTR strains, the plasmid-based setup compared constructs in an identical genetic background (PAO1 WT), providing a consistent reference point. As shown in Fig. 5a and b, the decay curves for nrdJ were nearly identical across all strains and constructs (Fig. 5a and b, green and yellow lines), and the best-fit half-life values were similarly consistent between strains using both approaches. In the plasmid-based system (Fig. 5c), decay rates were 0.44 min− 1 (95% CI: 0.4361 to 0.4499 min− 1) for the WT + PnrdA and 0.30 min− 1 (95% CI: 0.2618 to 0.3158 min− 1) for the WT + PnrdA-Δ5’ UTR. A similar trend was observed in the chromosomal context (Fig. 5d), with decay rates of 1.96 min− 1 (95% CI: 1.818 to 2.149 min− 1) for WT and 2.1 min− 1 for Δ5’ UTR (95% CI: 1.263 to 2.913 min− 1). Half-life values followed the same pattern: 1.57 min (95% CI: 1.541 to 1.589 min) for WT + PnrdA and 2.42 min (95% CI: 2.195 to 2.648 min) for WT + PnrdA-Δ5’ UTR (Fig. 5c) in the plasmidic system; and 0.35 min (95% CI: 0.3225 to 0.3813 min) for WT and 0.33 min (95% CI: 0.2379 to 0.5490 min) for Δ5’ UTR in the chromosomal system (Fig. 5d).

In summary, these results demonstrate that the 5’UTR specifically influences the half-life and decay of nrdA transcripts, as assessed indirectly via cat mRNA in plasmid-based constructs and directly through endogenous measurements—without altering the decay profile of nrdJ. While methodological differences and certain statistical limitations exist, the overall trend supports a destabilizing role for the 5′ UTR in nrdA regulation.

Bioinformatic and experimental assessment of the nrdA 5′ UTR: RNA folding and protein levels

5’ UTRs are known to adopt dynamic secondary structures that modulate key aspects of post-transcriptional control including ribosome accessibility, transcript stability and attenuation of translation11. As shown previously, its presence results in reduced mRNA transcripts levels (Fig. 4) and reduced transcript stability (Fig. 5). Building on these observations, we next assessed two independent—but potentially interconnected—dimensions: (i) in silico predictions of RNA secondary structure, and (ii) differences in relative NrdA protein levels.

RNAfold50 predictions revealed comparable minimum free energy (MFE) values of − 1950.10 kcal/mol for the wild-type nrdAB transcript and − 1819.60 kcal/mol for the Δ5′ UTR variant (Fig. 6a). However, mountain plot analysis (Fig. 6b) revealed local differences at the 5’ (blue arrow) and 3’ (green arrow) ends, with the wild-type transcript exhibiting more compact and intricate folding in these regions. Although this remains speculative and was not experimentally tested, the predictions showed no significant differences in ribosome binding site (RBS) accessibility (− 11.46 vs. −11.44 kcal/mol), indicating similar computational estimates of initiation potential.

Fig. 6
figure 6

RNA secondary structure prediction analysis and relative NrdA protein levels for nrdAB with full-length and truncated 5’ UTR variants. (a) Table summarizing minimum free energy (MFE) values for predicted secondary structures for nrdAB mRNA with full 5’ UTR (nrdAB) and with truncated 5’ UTR (nrdAB- Δ5’ UTR), along with free energy values for ribosome binding site (RBS) accessibility, calculated using RNAfold software50. (b) Mountain plots illustrating the predicted optimal secondary structures. The x-axis indicates the position of each base pair (Sequence, bp) and the y-axis (Height) represents the cumulative number of base pairs enclosing each position. Blue and green arrows mark the 5’ and 3’ ends of the mRNAs, respectively. (c) Western blot analysis of relative NrdA protein levels in P. aeruginosa PAO1 WT and PAO1 Δ5’ UTR (nrdA) strains during exponential (EX) and stationary (ST) phases under both aerobic and anaerobic conditions. Membranes were stained with Ponceau S prior to antibody incubation to confirm equal protein loading (Supplementary Material Figure S3). The original and cropped version of the gel are provided in the Supplementary Material Figure S6 and are also available in the public repository (https://doi.org/10.34810/data2361).

Full size image

Western-blot analysis (Fig. 6c) revealed increased NrdA protein levels in the Δ5’ UTR strain under both aerobic and anaerobic conditions. To ensure equal protein loading and confirm transfer efficiency, membranes were stained with Ponceau S prior to antibody incubation (Supplementary Material Figure S3).These data are consistent with the observed mRNA trends and suggest that elevated protein levels may be primarily driven by increased transcript abundance and stability (Figs. 4 and 5), rather than by structural influence on translation initiation. Importantly, the structural predictions were obtained through bioinformatics modelling, which provides in silico predictions but does not account for in vivo factors. Therefore, any potential structural contribution to translation remains purely hypothetical and should be interpreted with caution.

Taken together, these findings indicate that the 5’ UTR modulates nrdA expression through regulation of mRNA levels and transcript stability rather than exerting direct structural control over translation. This regulatory mechanism may contribute to the downregulation of nrdA expression during stationary growth, a phase in which the cellular demand for dNTP synthesis is reduced. Although the effect is not strictly translational, the increased mRNA levels (Fig. 4) and enhanced stability (Fig. 5) observed in the Δ5′ UTR variant likely contribute to the elevated NrdA protein abundance observed (Fig. 6c). The observed increase in protein levels is therefore best explained as a cumulative outcome of transcript-level effects, while any contribution from structural modulation of translation remains speculative and unconfirmed by the present data.

Impact of the nrdA 5′ UTR on gene expression and virulence in a Galleria mellonella infection model

Previous studies have reported strong upregulation of nrdJ during Galleria mellonella infection, whereas nrdA appears to play a less relevant role51. To investigate whether the increased nrdA transcript levels, enhanced mRNA stability, and elevated relative NrdA protein levels resulting from truncation of the 5′ UTR affect virulence, we performed infection experiments using the Galleria mellonella in vivo model.

Figure 7a shows Kaplan-Meier survival curves for larvae infected with P. aeruginosa PAO1 wild-type (PAO1 WT) and PAO1 Δ5’ UTR strains. The curves were significantly different according to the Log-rank (Mantel-Cox) test (*p < 0.05), indicating a statistically significant increase in median survival time (MST) for larvae infected with the PAO1 Δ5′ UTR strain (approximately 22 h) compared to those infected with PAO1 WT (around 20 h). These results suggest a modest reduction in virulence associated with the truncation of the 5’ UTR.

Fig. 7
figure 7

Virulence analysis of P. aeruginosa PAO1 WT and P. aeruginosa Δ5’ UTR strains and nrdA expression during Galleria mellonella infection. (a) Kaplan-Meier survival curves of Galleria mellonella larvae infected with P. aeruginosa PAO1 WT and PAO1 Δ5’ UTR (nrdA, chromosomal 5’ UTR truncation). Larvae injected with PBS served as negative controls. Survival was monitored over 15–20 h post-injection, with observations recorded at 15, 16, 17, 18, 19, and 20 h. MST = median survival time. (b) nrdA expression measured by qRT-PCR in larvae infected with P. aeruginosa PAO1 WT and its isogenic Δ5’ UTR (nrdA, chromosomal 5’ UTR truncation) strains at 16–17 h post-injection. The time point for sample collection is indicated by an arrow in panel (a).

Full size image

To explore the molecular basis of this phenotype, we analysed nrdA and nrdJ expression during infection using qRT-PCR at 16 hours post-infection. As shown in Fig. 7b, nrdA expression was elevated in the PAO1 Δ5’ UTR strain compared WT. Expression of nrdJ during infection was also assessed and is presented in Supplementary Material Figure S4. Although nrdJ expression appeared slightly reduced in the mutant strain during infection, the underlying regulatory relationship between nrdA and nrdJ in this context remains unclear. It is possible that this imbalance contributes to the observed reduction in virulence; however, it is unlikely to be the sole determinant. Additional regulatory elements and environmental factors may also play a role, and further studies are needed to elucidate the mechanisms underlying this interaction and its impact on pathogenesis.

Importantly, comparative growth analysis under standard laboratory conditions revealed no significant differences between PAO1 WT and the ∆5’ UTR strain (Supplementary Material Figure S5), suggesting that general fitness is not impaired. Therefore, the observed differences in infection outcomes are unlikely due to growth defects.

Overall, these findings suggest that truncation of the nrdA 5′ UTR may contribute to reduced virulence, potentially through changes in ribonucleotide reductase expression during infection. However, further in vivo studies would be necessary to fully elucidate the mechanistic link between 5′ UTR-mediated regulation and pathogenicity.

Conclusions

The 5’ untranslated region (5’ UTR) of the nrdAB operon in Pseudomonas aeruginosa was identified as an unusually long non-coding sequence, prompting investigation into its potential regulatory role under the specific conditions tested. Experimental data show that truncation of this region increases nrdA mRNA levels, enhances transcript stability and leads to higher relative NrdA protein abundance under the conditions tested. These effects were consistent across both native and plasmid-based systems. The 5′ UTR appears to contribute to nrdA downregulation during stationary phase, a period when dNTP synthesis demand naturally decreases. In vivo assays using Galleria mellonella revealed reduced virulence of the Δ5′ UTR strain, accompanied by increased nrdA expression and no detectable growth defects under standard laboratory conditions. A slight reduction in nrdJ expression was observed, although its regulatory connection to nrdA in this context remains unclear.

Although the extended 5′ UTR of the nrdAB operon raises the possibility of regulatory interactions, no specific elements—such as riboswitches, sRNA, or other regulatory features—were detected under the tested conditions. The available data do not allow us to determine whether the observed effects on transcript abundance, stability, and protein levels are solely attributable to the 5’ UTR sequence itself or also involve additional trans-acting factors. Methodological limitations constrain this analysis, and further studies will be required to explore potential interactions and their contribution to the regulatory mechanisms identified.

These findings highlight the significant role of the 5’ UTR in fine-tuning nrdAB operon expression and underscore the potential importance of untranslated regions as regulatory elements. Further studies are needed to elucidate the molecular mechanisms underlying nrdAB modulation and its integration into broader regulatory networks.

Materials and methods

Strains, plasmids, and growth conditions

The bacterial strains and the plasmids used are listed in Supplementary Material Table S1. E. coli and P. aeruginosa strains were routinely grown in Luria-Bertani (Scharlab, Spain) medium or minimal medium (MM) as previously described52 at 37 °C. For riboswitch analysis, MM was specifically used to provide defined growth conditions and reduce metabolic interference. For anaerobic growth, LB medium supplemented with 10 g/L KNO3 (LBN) was used in screw-cap Hungate tubes24. Liquid cultures were incubated with shaking at 200 rpm. Antibiotics were added at the following concentrations: 50 µg/mL ampicillin (Ap) and 10 µg/mL gentamicin (Gm) for E. coli; and 100 µg/mL gentamicin (Gm) and 300 µg/mL carbenicillin (Cb) for P. aeruginosa. For transcriptional and translational assays, rifampicin (Sigma Aldrich, Merck) was used at a final concentration of 200 µg/mL. When required, adenosyl cobalamin (coenzyme B12; Sigma Aldrich, Merck) was added at variable concentrations, depending on the experimental setup.

DNA manipulation

DNA manipulation and plasmid constructions were performed using standard protocols53. All kits and molecular biology enzymes were obtained from Thermo Fisher Scientific (Spain) and New England Biolabs, and used according to the manufacturer’s instructions. DNA fragments were amplified using Phusion High-Fidelity DNA polymerase or DreamTaq Green PCR MasterMix with the oligonucleotides listed in Supplementary Material Table S2. DNA fragments were isolated and purified from agarose gels using a GeneJet Gel Extraction Kit (Thermo Fisher Scientific), or the Monarch® DNA Gel Extraction Kit (New England Biolabs) in the case of digested DNA. Plasmid DNA was extracted using a GeneJET Plasmid Miniprep kit (Thermo Fisher Scientific) and introduced into P. aeruginosa cells via electroporation with a Gene Pulser XCell electroporator (Bio-Rad) as previously described24. All the constructs obtained were verified by DNA sequencing (Eurofins Genomics).

Plasmid construction

The plasmids pETS257, pETS258, pETS259 and pETS260 were constructed as follows (see Supplementary Material Table S1). Briefly, the nrdA upstream region was amplified using the primers 1 and 5 for the nrdA translational fusion (pETS258); 1 and 3, and 2 and 4 for transcriptional nrdA-Δ5’ UTR fusion (pETS257); and 1 and 3, and 4 and 5 for nrdA-Δ5’ UTR translational fusion (pETS259). For nrdA-Δ5’ UTR constructs, two fragments were generated using the primers described above and used as templates for overlap extension PCR with primers 1 and 2 A total of 82 bp from the 5’ end of the 5’ untranslated region (5’ UTR) and 43 bp from the 3’ end of the 5’ UTR were maintained to streamline the construct while preserving essential regulatory elements. The deletion was carefully designed to avoid disrupting the transcription start site (TSS) and the ribosome binding site (RBS). Although the mutant is referred to as PnrdA-Δ5’ UTR, the construct represents a targeted truncation rather than a full deletion. For simplicity, we refer to the construct containing the nrdA promoter, 5′ UTR region, and a short fragment of the coding sequence as PnrdA throughout the text and the 5’ UTR truncated one as PnrdA-Δ5’ UTR. The GFP (green fluorescent protein) gene was amplified from pETS130 using 7 and 8 primers. For translational fusions, the gfp, nrdA and nrdA-Δ5’ UTR inserts were digested with the SmaI restriction enzyme and ligated with T4 ligase to generate nrdA and nrdA-Δ5’ UTR GFP translational fusions. For pETS260, cobW upstream region was amplified using the primers 9 and 10. The resulting fragments (nrdA-Δ5’ UTR, nrdA-Δ5’ UTR::gfp, nrdA::gfp and cobW were cloned separately to pJET 1.2 vector (Thermo Scientific) and transformed into E. coli DH5α. The resulting plasmids and pETS130-GFP vector were digested with BamHI-KpnI (translational GFP fusions), BamHI-SmaI or SacI- BamHI (transcriptional GFP fusion) and ligated using the T4 ligase enzyme. These plasmids were electroporated into P. aeruginosa PAO1 WT. All nrdA GFP transcriptional and translational fusions include the native promoter, the 5’ untranslated region (5’ UTR) — either complete or truncated — and short fragment of the coding sequence (CDS) as explained before.

The plasmids pJN106 and pJN107 were generated as described below (Supplementary Material Table S1). The 5’-3’ orientation from the 5’ UTR was amplified by primers 11 and 12, while the 3’-5’ orientation from the 5’ UTR was amplified by primers 13 and 14. The resulting fragments, named 5’ UTRfor and 5’ UTRrev, were cloned separately to pJET 1.2 and transformed into E. coli DH5α. The resulting plasmids and pJN105 vector were digested with EcoRI-PstI and ligated using T4 ligase enzyme. These plasmids were electroporated into P. aeruginosa PAO1 WT.

CRISPR-cas9 mutant construction

P. aeruginosa PAO1 Δ5’ UTR (nrdA) chromosomal mutant was generated using the CRISPR-cas9 toolkit described previously54, following the established protocol with some modifications. The recombineering ssDNA oligonucleotide consisted of 50 nucleotides downstream of the transcriptional starting site (located 409 bp upstream of the nrdA start codon-ATG), directly followed by 50 nucleotides downstream of the nrdA ribosome binding site (located 9 bp upstream of the start codon). The oligonucleotide was synthesized by Eurofins Genomics (Supplementary Material Table S2). P. aeruginosa PAO1 pSH124 cells were electroporated with both plasmids pS149 (pS148-Δ5’ UTR) and the recombineering oligonucleotide (5’ UTR_PAO1(+)-Rec) (Supplementary Material Table S1 and Table S2). A total of 82 bp from the 5’ end of the 5’ untranslated region (5’ UTR) and 43 bp from the 3’ end of the 5’ UTR were maintained to streamline the construct while preserving essential regulatory elements. Although the mutant is referred to as Δ5’ UTR, the construct represents a targeted truncation rather than a full deletion. As described, flanking sequences were preserved to avoid disrupting key regulatory elements, including the transcription starting site (TSS) and ribosome binding site (RBS). Screening of edited strains was performed using two consecutive colony PCR (cPCR): an initial cPCR was conducted on pools of colonies, and upon identification of a positive pool, individual colonies were subsequently tested by cPCR. Positive colonies were sequenced by Eurofins Genomics, and the plasmids were subsequently cured as described in the protocol54. To assess potential off-target effects from single guide RNA (sgRNA) within pS149, BLASTn33 analysis was performed using both forward and reverse sequences of the CRISPR-Cas9 guide RNA with PAM site against the Pseudomonas aeruginosa PAO1 genome (taxid: 208964), alongside the potential on-target effects analysis conducted using Benchling (https://www.benchling.com).

Gene reporter assay experiments

Green fluorescent protein (GFP) reporter assay was carried out with two methodologies; first, P. aeruginosa bacterial pETS130-GFP derivatives were grown in LB medium (for aerobic cultures) and LBN medium (for anaerobic cultures) at 37 °C and 200 rpm, reaching the appropriate optical density (OD) at 600 nm wavelength (OD600). Cultures were collected at two growth phases: exponential phase (OD600 = 0.4–0.7) and stationary phase (OD600 ~ 2). At each point, three independent 1 mL samples were taken. Samples were centrifuged at 6000 rpm for 10 min and the resulting pellets were washed with phosphate-buffered saline (PBS) (Fisher Scientific) containing 2% formaldehyde and stored in the dark at 4 °C for 10 min. After a second centrifugation, the pellets were resuspended in 1 mL PBS. The fluorescence was measured in 96-well plates (Costar ® 96-Well Black polystyrene plate, Corning) using an Infinite 200 Pro Fluorescence Microplate Reader (Tecan, Switzerland).

Alternatively, P. aeruginosa bacterial pETS130-GFP derivatives growth and fluorescence were monitored simultaneously every 20–30 min in a 96-well plate (Costar ® 96-Well Black polystyrene plate, Corning) using a Spark Microplate Reader (Tecan, Switzerland) at 37 °C. For the riboswitch assay, P. aeruginosa strains carrying pETS130-GFP derivatives were grown in minimal medium (MM)52 until OD600 ~ 0.3–0.4. Coenzyme B12 was added at 1 µg/mL or at varying concentrations (1, 2.5 and 10 µg/mL) and GFP expression was measured in a 96-well plate (Costar ® 96-Well Black polystyrene plate, Corning) either every 20 min or once a specific OD600 was reached, using an Infinite 200 Pro Fluorescence Microplate Reader (Tecan, Switzerland).

Western immunoblot analysis

P. aeruginosa PAO1 WT and P. aeruginosa PAO1 Δ5’ UTR strain cultures were collected under aerobic and anaerobic conditions during exponential (OD600 = 0.4–0.7) and stationary (OD600 ~ 2) growth phases. Protein concentrations were adjusted to 14 µg for all the samples based on quantification using the Bradford Assay (Bio-Rad), with bovine serum albumin (BSA) as the standard. Western blotting was conducted as previously described, with some modifications5. After protein transfer to PVDF membranes, Ponceau S staining (0.1% Ponceau S in 5% acetic acid) was performed for 5 min at room temperature to verify transfer efficiency and uniform protein loading. Membranes were then rinsed with distilled water and destained using TBST (Tris-buffered saline with 0.1% Tween) prior to blocking and antibody incubation. A pre-stained protein ladder (Thermo Scientific™ PageRuler™ Plus Prestained Protein Ladder, 10–250 kDa) was included to estimate molecular weights. The membranes were not cut prior to antibody incubation; full-length blots are shown in Supplementary Material Figure S3, with membrane edges visible. A 1:650 dilution of polyclonal Anti-NrdA primary antibody (Agrisera, Sweden) was used, followed by detection with a donkey anti-rabbit horseradish peroxidase–conjugated secondary antibody (Bio-Rad) at a 1:5000 dilution. The antibody–antigen complexes were visualized using Amersham™ ECL™ Prime Western Blotting Reagent (GE Healthcare), according to the manufacturer’s instructions. Protein bands were visualized and analysed by the ImageQuant™ LAS 4000 Mini system (GE Healthcare). The resulting image was processed using Fiji (ImageJ, NIH, USA) for visualization purposes.

5’ UTR overexpression

P. aeruginosa PAO1 pJN106 and P. aeruginosa PAO1 pJN107 cultures were collected during the exponential phase, after 2 h of induction with 0.4% L-arabinose. Cell pellets were lysed with BugBuster® 1X (Sigma Aldrich) following the manufacturer’s instructions. Lysates were centrifuged at 16,000 × g for 20 min at 4 °C. Protein concentrations were determined using the Bradford assay (Bio-Rad) with bovine serum albumin (BSA) as a standard. A 4–20% Mini-PROTEAN® TGX™ Precast protein gel (Bio-Rad) was loaded, and the electrophoresis was performed at 40 mA.

RNA isolation, reverse transcription and real-time PCR (qRT-PCR)

P. aeruginosa PAO1 cultures were collected under aerobic and anaerobic conditions during exponential (OD600 = 0.4–0.7) and stationary (OD600 ~ 2) growth phases. Total RNA was extracted using the GeneJET RNA Purification Kit (Thermo Fisher Scientific) following the manufacturer’s instructions. To eliminate residual DNA contamination, RNA samples were treated with TURBO™ DNase (Thermo Fisher Scientific) following the manufacturer’s instructions and verified by PCR with the oligonucleotides listed in Supplementary Material Table S2. RNA concentration and purity were determined using a NanoDrop (Nanodrop spectrophotometer ND-1000). cDNA synthesis was performed using Maxima Reverse Transcriptase (Thermo Fisher Scientific) and Random hexamers (Thermo Fisher Scientific) according to the manufacturer’s instructions. Quantitative real-time PCR (qRT-PCR) was carried out using PowerUp™ SYBR−Green™ (Applied Biosystems) with the primers listed in Supplementary Material Table S2. Reactions were run on a StepOnePlus™ Real-Time PCR System (Applied Biosystems) following the manufacturer’s specifications. The gap gene was used as an internal reference unless otherwise specified. The results were analysed using the ΔΔCt method, unless stated differently. qRT-PCR data were normalized within each condition, relative to the corresponding control sample in each growth phase or genotype. This normalization strategy was applied to highlight relative differences within each experimental group. Specific details regarding the normalization approach and reference samples used are described in each corresponding experimental context.

Primer extension

Bacterial cells were grown in LB medium to an OD600 of 0.7, and the total RNA was extracted using the GeneJET RNA Purification Kit (Thermo Fisher Scientific) following the same protocol described in the Materials and Methods “RNA isolation, reverse transcription and real-time PCR (qRT-PCR)” section. SupersScript reverse transcriptase (Thermo Fisher Scientific) was used to reverse transcribe nrdA using extension primers 27 and 28 (Supplementary Material Table S2). The primer was radioactively labelled with [g-32P]ATP using T4 Polynucleotide Kinase (Thermo Fisher Scientific). A DNA fragment of 841 nucleotides corresponding to the sequence upstream of the nrdA coding region was amplified by PCR using High Expand Taq polymerase (Roche). The resulting PCR products generated a sequence ladder with the same oligonucleotide used for the extension reaction using the Sequenase kit (Promega) (Supplementary Material Table S2). The extension product was resolved on an 8% acrylamide–7 M urea gel alongside the sequence ladders.

5’ RACE (5’ rapid alignment of cDNA ends)

P. aeruginosa PAO1 cultures were collected under aerobic and anaerobic conditions during the exponential growth phase (OD600 = 0.5), and RNA extraction was performed as previously described in Materials and Methods “RNA isolation, reverse transcription and real-time PCR (qRT-PCR)” section. cDNA synthesis was performed using Maxima Reverse Transcriptase (Thermo Fisher Scientific), according to the manufacturer’s instructions. Each reaction contained 1.5 pmol of gene-specific primer 46 (Supplementary Material Table S2) and 0.3 µg of total RNA. The resulting cDNA was purified using the High Pure PCR Product Purification Kit (Roche Life Science), according to the manufacturer’s instructions. Poly-A tailing was performed using 40 U of Terminal Deoxynucleotidyl Transferase (Thermo Fisher Scientific) and 1.5 nmol dATP (Thermo Fisher Scientific), following the manufacturer’s instructions.

The first PCR (PCR1) was conducted using primers 47 and 26 (Supplementary Material Table S2) with Phusion High-Fidelity DNA polymerase (Thermo Fisher Scientific), using the poly-A-tailed cDNA as a template. PCR1 was optimised with 10 cycles of 40-second extension, followed by 25 cycles with a 20-second extension, increasing by 5 s per cycle. The PCR1 product was gel-purified using the GeneJet Gel Extraction kit (Thermo Fisher Scientific) and diluted 1:20. The second PCR (PCR2) was performed using primers 48 and 49 (Supplementary Material Table S2), with PCR1 as a template, using Phusion High-Fidelity DNA polymerase (Thermo Fisher Scientific) according to the manufacturer’s instructions. PCR2 products were purified and cloned into the pJET1.2 blunt vector, following the same protocol as described in Materials and Methods section “Plasmid construction”. Colonies were sequenced by Eurofins Genomics using the primer 49 (Supplementary Material Table S2).

Transcriptional shut-off assay

P. aeruginosa PAO1 wild-type, P. aeruginosa Δ5’ UTR, P. aeruginosa pETS134, and P. aeruginosa pETS257 strains were grown and treated with 200 µg/mL of rifampicin (a transcriptional blocking agent). Samples were collected at 0, 5, and 20 min after rifampicin addition to arrest transcription for P. aeruginosa pETS134 and pETS257, and at 0, 2, 6, 10, and 20 min for P. aeruginosa PAO1 wild-type, and Δ5’ UTR. Total RNA was extracted using the GeneJET RNA Purification Kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. Reverse transcription and subsequent quantitative real-time PCR (qRT-PCR) measurements were performed as described in Materials and Methods “RNA isolation, reverse transcription and real-time PCR (qRT-PCR)” section, with the modification that the 16S rRNA gene was used as an internal reference for normalization due to its previous use in related experiments55,56. nrdJ was included as a specificity control to confirm that the truncation of 5’ UTR did not affect unrelated transcripts. cat mRNA levels were used as an indirect indicator of nrdA mRNA decay, specifically in plasmid-based constructs designed to analyse transcript half-life. This strategy was necessary due to technical constraints: the chromosomal wild-type copy of nrdA, containing the full 5′ UTR, hindered independent assessment of the truncated variant. Under these conditions, cat provided the only reliable proxy for comparing transcript stability associated solely with the presence or truncation of the 5′ UTR.

Absolute PCR quantification was performed using a standard curve from target genes (cat, nrdA, nrdJ, 16S). Briefly, PCR amplification of genomic nrdA, nrdJ and 16S rRNA, and cat from pETS130-GFP plasmid was performed using the oligonucleotides listed on Supplementary Material Table S2. A ten-fold dilution of known quantities of amplicons was performed, and the number of copies was calculated by the following formula (Integrated DNA Technologies, IDT):

$$\:Number\:of\:copies=\frac{Quantity\:of\:DNA\:\left(ng\right)\times\:6.022\times\:{10}^{23}}{Length\:of\:DNA\:\left(bp\right)\times\:1\times\:{10}^{9}\times\:650}$$

In which \(\:6.022\times\:{10}^{23}\)is Avogadro number, 650 is the average weight (grams, g) of a base pair (bp) in Daltons and \(\:1\times\:{10}^{9}\) is to convert grams(g) to nanograms (ng).

Once the Ct values from each dilution of every target gene were determined, a standard curve was established by plotting the target DNA copy number against each Ct value. The standard curve was used to calculate the number of copies from each target gene in different samples. The percentage of retro-transcribed mRNA remaining was calculated using non-rifampicin treated sample or the before-rifampicin sample as the initial mRNA amount of each target gene.

Data were analyzed using nonlinear regression analysis in GraphPad Prism 10.1.1 (GraphPad Software). A one-phase decay model was applied assuming a decay from a shared initial value (Y0 = 100). The equation model was: \(\:Y=\left({Y}_{0}-Plateau\right)\times\:{e}^{-K\times\:X}+Plateau\), where X is time (minutes) and Y is percentage (%) of remaining cDNA that decays with one phase down to a Plateau, and K is the decay rate constant.

To ensure biological plausibility and stability of the fit, constraints were applied: K > 0.01 and in some cases, Plateau values were fixed when instability was detected in the fitting process. Key parameters extracted from the fit included the decay rate constant (K) and half-life (\(\:{t}_{1/2}=\frac{(\text{ln}2)}{K}\)). For the sample nrdA transcript sample within WT + PnrdA- Δ5’ UTR parameters determination also included fitting the plateau (Plateau = 6) to ensure that the calculated values accurately reflect the true behaviour of the system. Comparisons between the groups focused on differences in K and half-life as indicators of decay rate.

The goodness of the fit was evaluated using the R2 value and 95% confidence intervals (profile likelihood method). In some cases (Δ5’ UTR, nrdJ transcript), profile likelihood intervals could not be fully determined and were calculated using the symmetrical (asymptotic) method implemented in GraphPad Prism 10.1.1 (GraphPad Software). While this method assumes normality and may underestimate the uncertainty for some parameters, it provides a consistent basis for comparison.

Galleria mellonella infection

Galleria mellonella larvae were maintained and dose-infected as previously described51 with some modifications. Bacterial doses were plated in LB agar (Scharlab) plates to determine bacterial colony-forming units (CFUs), corresponding to approximately 20–40 CFUs per larva. For bacterial RNA extraction during infection, groups of 10 infected larvae were collected approximately 16–17 h post-infection and anesthetized on ice for 10 min. Bacterial sample collection from Galleria mellonella was performed following the previously described protocol51, and RNA extraction was carried out as described in Materials and Methods “RNA isolation, reverse transcription and real-time PCR (qRT-PCR)” section. For virulence analysis, groups of 10 larvae were infected and incubated at 37 °C, with periodic monitoring throughout the course of infection.

Bioinformatic analysis and data sourcing

Riboswitch prediction analysis was performed using Riboswitch Scanner21,57 with default settings. RegPrecise26 was used to identify cobalamin regulon genes in P. aeruginosa PAO1. Sequences obtained from RegPrecise database and putative B12 riboswitch sequences were aligned using T-coffee software27 with default parameters. The alignment was manually analysed, considering the predicted secondary structure29.

BLAST software33 and the Coding Potential Calculator 2 (CPC2) tool32 were used for protein-coding potential prediction, both with default settings. The Rfam41, RNAcentral42 and PresRAT44 databases were used to identify potential matches with non-coding RNA sequences. The presence of Rho-independent terminators was evaluated using ARNold software40 with standard settings. For RNA structure predictions were performed using RNAfold with the standard settings50.

Promoter prediction was conducted using PromoterHunter38 and BPROM39. Both scientific publications and data repositories were consulted, including Kyoto Encyclopaedia of genes and genomes58as well as raw data from several RNA-seq experiments, with the specific sources cited in the relevant sections of the manuscript.

Statistical analysis

Statistical analyses were performed using GraphPad Prism 10.1.1 (GraphPad Software). Single comparisons were performed using unpaired Student’s t-test. Statistical significance was indicated by p-values as follows: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. Error bars in figures represent the standard deviation (SD) between samples. Welch’s t-test was applied when unequal variances were detected.