Bacterial strains and plasmids used in this study are listed in Supplementary Table S1. P. aeruginosa and E. coli strains were routinely grown in Luria-Bertani broth (LB) at 37°C. For selective E. coli growth, ampicillin, gentamicin, and chloramphenicol were added at 100, 20, and 25μg/ml, respectively. For selective Pseudomonas growth, carbenicillin and gentamicin were added at 300 and 60μg/ml, respectively. For P_BAD induction in vector plasmid pGM931, arabinose was added to a final concentration of 10mm.

Anaerobic cultivations of P. aeruginosa strains were performed in Oxoid anaerobiosis jars at 37°C, using agar plates prepared with Brain Heart Infusion (BHI)-rich medium supplemented with 100mm KNO₃ to allow anaerobic respiration or without KNO₃ to test arginine fermentation. For the anaerobic growth assay, cell cultures of P. aeruginosa RP73 (Bianconi et al., 2015) and RP73 ΔersA (Ferrara et al., 2020), PAO1 (Stover et al., 2000), and PAO1 ΔersA (Ferrara et al., 2015) with an OD₆₀₀ of 1 (corresponding to 1×10⁹CFU/ml) were serially diluted until 10⁻⁶; 2μl of each dilution was spotted and incubated for 72h. For total RNA extraction from anaerobically grown cell cultures of PAO1 and PAO1 ΔersA (Ferrara et al., 2015), the strains were plated at the confluence and incubated for 72h. The anaerobic atmosphere was induced by the Oxoid AnaeroGen sachet. Control testing of anaerobiosis was performed using the Oxoid Anaerobic indicator in the jar as a visual check that anaerobic conditions have been achieved and maintained.

The oligonucleotides used in this study are listed in Supplementary Table S2. Plasmids pBBR1-anr::sfGFP and pBBR1-dnr::sfGFP expressing anr::sfGFP and dnr::sfGFP translational fusions, respectively, under the PLtetO-1 constitutive promoter were constructed as follows. A DNA fragment including the 31-nt UTR along with the first 22 codons (66nt) of the anr open reading frame (66nt) was PCR amplified from PAO1 genomic DNA with oligos 2/3, digested with NsiI/NheI, and cloned into the sfGFP reporter vector pXG10-SF (Corcoran et al., 2012) giving rise to plasmid pXG10-anr::sfGFP. With the same procedure, a DNA fragment including the 116-nt UTR and the first 32 codons (96nt) of the dnr open reading frame was amplified with oligos 4/5, digested NsiI/NheI, and cloned into the sfGFP reporter vectors pXG10-SF giving rise to plasmid pXG10-dnr::sfGFP. The DNA fragments spanning from the PLtetO-1 promoter to the end of the sfGFP reporter gene were amplified by PCR, respectively, from pXG10-anr::sfGFP and pXG10-dnr::sfGFP with oligos 6/7, digested ClaI/XbaI, and cloned into the low-copy number shuttle vector pBBR1-MCS5 (Kovach et al., 1995) giving rise to constructs pBBR1-anr::sfGFP and pBBR1-dnr::sfGFP, respectively. All constructs were verified by sequencing (Eurofins Genomics) using either oligos 7 or 8.

Purified RNA for RNA/RNA interaction assays was prepared by T7 RNA polymerase transcription of gel-purified DNA fragments. DNA fragments for anr mRNA and ErsA RNA preparations were amplified from P. aeruginosa PAO1 genomic DNA with oligo pairs 9/10 and 11/12, respectively. Each transcription reaction was performed with the Riboprobe^® System-T7 (Promega) with 300ng of DNA template. Synthesized RNA was purified using the RNeasy MinElute Cleanup Kit (Qiagen). Purified RNA was checked by denaturing polyacrylamide gel electrophoresis and quantified using Eppendorf Biospectrometer. Electrophoretic Mobility Shift Assay to analyze ErsA/anr mRNA interactions was performed in 10μl of reactions containing 1×RNA-binding buffer (10mm Tris–HCl, pH 7, 100mm KCl, 10mm MgCl₂, and 10% glycerol), purified ErsA RNA and increasing amounts of purified anr mRNA, or yeast tRNA (Ambion). Binding reactions were incubated at 37°C for 20min, then loaded into a native 6% polyacrylamide gel (acrylamide-bis ratio 29:1) in 0.5×TBE buffer (45mm Tris-borate, pH 8.0, 1mm EDTA), and electrophoresed using a Mini-Protean Electrophoresis System (Bio-Rad) at 4°C and 180V for 90min. RNAs were transferred into a GeneScreen plus nylon hybridization transfer membrane (Perkin Elmer) using a semi-dry electroblotting (Fastblot B33, Biometra) set at 25V, 400mA for 1h, and UV-crosslinked to the membrane with a Stratalinker 1800 UV Crosslinker (Stratagene). The blotting membrane was hybridized with a biotinylated anti-ErsA probe (oligo 1) using the North2South Chemiluminescent Hybridization and Detection kit (Thermo Scientific) according to the manufacturer’s instructions. After the addition of the Streptavidin-HRP conjugate, the ErsA bands were visualized by mixing equal volumes of luminol/enhancer solution and stable peroxide solution and acquiring images with a ChemiDoc Touch Imaging System (Bio-Rad) using ImageLab analysis software.

Fluorescence measurements in P. aeruginosa strains PAO1 wild type (Stover et al., 2000), PAO1 ΔersA (Ferrara et al., 2015), and PAO1 hfq⁻ (Sonnleitner et al., 2003) carrying the sfGFP translational fusions with target genes were carried out as follows. To test the sRNA/target interaction in planktonic cells of aerobically grown cultures, strains carrying the reporter pBBR1-anr::sfGFP or pBBR1-dnr::sfGFP alone, or combined with either pGM931 (Qiu et al., 2008; Delvillani et al., 2014; Ferrara et al., 2020) or pGM-ersA (Ferrara et al., 2015), were inoculated in 15ml tubes filled with 5ml of LB at an OD₆₀₀ of 0.1 and grown at 37°C in a rotatory shaker. Samples were taken after 6 and 24h (Ferrara et al., 2015). For the testing of anaerobically grown cultures, strains were inoculated at an OD₆₀₀ of 0.4 in 50ml flasks filled with 10ml of LB medium supplemented with 100mm KNO₃. The anaerobic atmosphere was induced by Oxoid AnaeroGen sachet. Control testing of anaerobiosis was assessed by the Oxoid Anaerobic Indicator. Samples were taken after 2days of static anaerobic growth at 37°C. The collected samples were centrifuged, washed twice, and resuspended in PBS (10mm Na₃PO₄, 150mm NaCl) to OD₆₀₀ of 1. Samples were then serially diluted 1.33-fold (corresponding to OD₆₀₀ of 0,75, 0.5, and 0.25). 200μl of aliquots was transferred to black polystyrene 96-well microplates with a clear, flat bottom (Corning). At least three biological replicates were used for every experimental set. The absorbance (Abs₅₉₅) and fluorescence polarization (FP_485/535) were measured in an EnSight Multimode Plate Reader (PerkinElmer) using Kaleido data acquiring software. GFP activity was expressed in arbitrary units (AU) as FP_485/535/Abs₅₉₅.

To test fluorescence in surface-grown cells, the strains were spotted on agar plates of BHI supplemented with or without 100mm KNO₃ and incubated at 37°C for 72h. Cells were collected from spots, resuspended in PBS, and assayed as described above.

Quantitative RT-PCR analysis (qRT-PCR) was performed on total RNA extracted from P. aeruginosa PAO1 wild type and ΔersA surface-grown cells under anaerobic conditions for 72h. Immediately after opening the jar, cells were removed from plates, resuspended in RNAprotect Cell Reagent (Qiagen), incubated, for 5min at room temperature, pelleted by centrifugation, and stored at −80°C until use. RNA extraction was performed as described previously in Ferrara et al. (2015). The quality and concentration of the extracted RNA were assessed by a Biospectrometer (Eppendorf).

cDNA was synthesized from 1μg of total purified RNA using Superscript III Reverse Transcriptase (Invitrogen) according to the manufacturer’s instructions. qRT-PCR was performed in triplicate using SYBR Green PCR Master Mix (Bio-Rad) on a CFX Connect Real-Time System (Bio-Rad). Oligo pairs 13/14 (Vitale et al., 2008), 15/16 (Tribelli et al., 2019), and 17/18 (Jackson et al., 2013) were used for amplification of 16S, anr, and dnr, respectively. The reaction procedure involved incubation at 95°C for 5min and 39cycles of amplification at 95°C for 15s, 58°C for 20s, and 72,5°C for 30s. The calculation of the relative expression of anr and dnr genes in the PAO1 ΔersA mutant vs. the wild-type strain was performed first normalizing mRNA amounts to 16S ribosome RNA (ΔC_T) and then relating the ΔC_T in the ΔersA mutant to the wild type (ΔΔC_T).

The transcriptome profiling of the aerobically grown PAO1 wild-type strain vs. the corresponding ΔersA mutant showed more than 160 differentially expressed genes (DEGs) (Falcone et al., 2018). These data were collected from bacterial cells incubated at 37°C until the onset of the stationary phase (OD₆₀₀ =2.7; Falcone et al., 2018).

For a more comprehensive understanding of the ErsA-dependent regulatory networks, we performed a new functional categorization of the above-mentioned DEGs. This analysis revealed 52 DEGs associated with growth or survival under anaerobic conditions (Table 1). Among these, some genes belong to the denitrification and nitrate metabolism (narK1, narH, narI, narJ, narL, and nirN), the fermentation pathways of arginine and pyruvate (arcD, ackA), and the universal stress response (uspL/PA1789, uspM/PA4328, uspN/PA4352, and uspO/PA5027). Out of 52 anaerobiosis-linked DEGs, 44 and 5 were already known to be up- or downregulated under anaerobic growth, respectively (Alvarez-Ortega and Harwood, 2007; Platt et al., 2008; Trunk et al., 2010; Crespo et al., 2017; Tribelli et al., 2019).

The majority of the DEGs known to be induced by anaerobic conditions was downregulated in the ΔersA strain. Vice versa, DEGs repressed in anaerobiosis and appeared upregulated in the ΔersA strain. Exceptions are the nirN gene for nitrite respiration, and the last genes of the nar operon (narH, J, and I) for nitrate respiration, which are upregulated by anaerobiosis, and appear also upregulated in the ΔersA mutant strain. Besides, as evidenced in Tables 1, 23 and 3 DEGs induced and repressed by anoxia, respectively, are under the transcriptional control of the master regulator of anaerobiosis, Anr (Trunk et al., 2010). Finally, the heat-shock protein IbpA and the transcription regulator PsrA (involved in the positive regulation of RpoS and the TTSS) were included among DEGs not belonging to the Anr regulon.

Taken together, this more accurate analysis of the comparative transcriptional profiling between PAO1 wild-type and ΔersA mutant strains suggested that the overall anaerobic metabolism in P. aeruginosa and particularly the denitrification processes could be co-regulated by ErsA. Despite the list of anaerobically regulated DEGs in Table 1 is reasonably not exhaustive since bacterial cells were grown in aerobic conditions before RNA extraction (Falcone et al., 2018), this evidenced that ErsA influences a peculiar set of genes belonging to the regulon of the transcription factor Anr. Specifically, the pattern of upregulated and downregulated genes in the PAO1 ΔersA strain seemed to be consistent with positive post-transcriptional regulation of anr mRNA by ErsA. This suggested that Anr expression could be directly targeted by ErsA. Furthermore, since Anr is the master regulator in anaerobiosis of a broad regulon of anoxia-responsive genes (Platt et al., 2008; Trunk et al., 2010) including also the transcription factor DNR (Supplementary Figure S1), which is indeed responsible for the transcription activation of the denitrification machinery (Trunk et al., 2010), we speculated that ErsA might additionally regulate DNR expression.

The results of the analysis described above suggested that anr and dnr mRNAs could be targeted by ErsA. To assess preliminarily the interaction between ErsA RNA and the two candidate target mRNAs, the web tool IntaRNA (Wright et al., 2014) was used. The modeling of the interaction between the full-length ErsA and anr mRNA comprehensive of 5' untranslated region (5'-UTR) predicted that the 34-nt long U-rich unstructured domain II of ErsA (Ferrara et al., 2015) could, from nt 31 to 54, extensively base-pair with the open reading frame region of anr mRNA, from nt +28 to +49 relative to the translational start site AUG (Figure 1A). To validate this sRNA/mRNA interaction in vivo, we used a robust two-plasmid reporter system suited for P. aeruginosa in our previous works (Ferrara et al., 2015, 2017). To this end, we generated a translational fusion between the first 22 codons of anr open reading frame linked to its 5'-UTR and the superfolder variant gene of the green fluorescent protein (sfGFP; Corcoran et al., 2012) under the control of the heterologous constitutive promoter PLtetO-1. We comparatively assayed the anr::sfGFP fusion in planktonic cells, grown in aerobic conditions, of wild-type and ΔersA mutant strains and observed no significant differences in fluorescence levels. Then, the fluorescence expressed by the anr::sfGFP fusion was assayed in P. aeruginosa PAO1 strains, still grown in planktonic and aerobic conditions, in the absence and presence of ErsA overexpression from the arabinose inducible pGM-ersA vector (Ferrara et al., 2015). As shown in Figure 1B, ErsA overexpression by pGM-ersA conferred an approximately 2-fold increase in GFP activity compared to the strain harboring the control vector pGM931 (Qiu et al., 2008; Delvillani et al., 2014). Since spurious activating interactions of ErsA with the sfGFP open reading frame were ruled out (Ferrara et al., 2015), these results strongly suggested a positive direct modulation of anr mRNA by ErsA. Since a non-planktonic growth, e.g., adherent to a surface in form of either biofilm or colony can strongly influence P. aeruginosa physiology and gene expression, we assayed the anr::sfGFP fusion in P. aeruginosa cells grown as extended colonies on agar plates, either in aerobic or anaerobic conditions. As shown in Figure 2A, no significant differences in fluorescence levels of the anr::sfGFP fusion between wild-type and ΔersA mutant strains were observed in both conditions. However, ErsA overexpression from the arabinose inducible pGM-ersA vector resulted in a significantly higher magnitude of induction of the anr::sfGFP fusion (Figure 2B) than the planktonic conditions (Figure 1B), namely, 5- (surface-grown cells) vs. 2-fold (planktonic cells). This enhanced effect of ErsA overexpression was independent of the presence of oxygen (Figure 2B). Overall, these results suggested that, above a certain abundance threshold, ErsA can enhance Anr mRNA translatability and the ErsA-mediated stimulation is more efficient when cells grow aggregated on a surface.

Furthermore, we aimed to evaluate whether the activity of the RNA chaperone Hfq could influence the ErsA-mediated activation of anr expression. This is because the regulatory activity of ErsA was shown to be dependent on Hfq for other targets (e.g., algC, amrZ, and oprD; Ferrara et al., 2015; Falcone et al., 2018; Sonnleitner et al., 2020) and it has been deduced that Hfq stimulates the expression of anr by an unknown mechanism since Anr was less expressed in an hfq⁻ mutant of P. aeruginosa than in the corresponding wild type (Sonnleitner et al., 2011). The positive influence of Hfq on Anr expression was reconfirmed in our experimental system. As shown in Figure 1C, the fluorescence levels expressed by the translational fusion anr::sfGFP were significantly lower in the hfq⁻ than in the wild-type background. These results suggested that Hfq plays per se a direct and positive post-transcriptional regulatory role on anr.

Besides, differently from the wild type, the overexpression of ErsA from pGM-ersA in an hfq⁻ background (Sonnleitner et al., 2003) showed no effects on fluorescence levels generate by the anr::sfGFP fusion (Figure 1B). Despite Hfq contributes to ErsA stability (Ferrara et al., 2015), overexpression of ErsA from pGM-ersA in an hfq⁻ strain reaches levels similar to those of the wild type (Ferrara et al., 2015). Therefore, the Anr activation failure by ErsA in the absence of effective Hfq was a genuine effect which strongly indicated that the ErsA-mediated activation of anr expression is an Hfq-dependent mechanism.

To assess in vitro the ErsA/anr mRNA interaction, the whole ErsA RNA and the anr mRNA region spanning −31 (5'-UTR) to +66 were synthesized in vitro, mixed, and analyzed on native polyacrylamide gels. We mixed fixed amounts of ErsA RNA with increasing concentrations of target anr mRNA. As shown in Figure 1D, ErsA showed multiple electrophoretic forms, with two prevalent fast-migrating bands, #1 and #2. After the addition of increasing amounts of the target anr mRNA, bands #1 and #2 progressively decreased in intensity, and bands corresponding to the sRNA/mRNA complexes appeared accordingly. No extra-bands formed when increasing amounts of control tRNAs preparation were mixed with ErsA RNA. This indicated that ErsA and anr RNAs can specifically interact.

We also modeled a possible interaction of ErsA with dnr mRNA. The IntaRNA tool predicted that an interval similar to the above nt of the unstructured region of ErsA could couple with the open read frame of dnr, from nt 16 to 38 downstream AUG start codon (Supplementary Figure S2). To test in vivo this prediction, we generated a dnr::sfGFP translational fusion cloning the whole 5'-UTR of the dnr gene and 96nt of its open reading frame, corresponding to the first 32 amino acids of Dnr, fused to sfGFP. As in the case of anr, we compared the fluorescence of the dnr::sfGFP reporter in surface-grown cells of the wild-type and ΔersA strains under aerobic and anaerobic conditions, and no relevant differences were detected between the two strains (Figure 2C). However, differently from anr::sfGFP, ErsA overexpression had no effects on sfGFP expression (Figure 2D). We performed the same experiments with planktonic cells either in either aerobic or anaerobic conditions and again no effects of ErsA deletion or overexpression were detected, respectively (Figures 2E,F). Overall, these assays did not evidence any regulatory activity of ErsA on the translational fusion dnr::sfGFP.

The results presented above with the anr::sfGFP translational fusion revealed no decrease in the translation of Anr in the ΔersA background under any oxygen conditions tested, either in planktonic or surface-grown cells. However, these tests may suffer from sensitivity when evaluating downregulations due to the stability of the reporter gene product vs. the detection of upregulation which is less affected by the half-life of the reporter. Therefore, to evaluate the ErsA-mediated regulation of Anr in physiological conditions of denitrification, we assessed whether the loss of ErsA and the ensuing expected decrease of translation rate could influence negatively the anr mRNA abundance in anaerobiosis. According to the well-characterized regulatory link between Anr and Dnr during anaerobic growth, i.e., the dnr gene is transcriptionally activated by Anr, the dnr mRNA levels were also expected to decrease in the ΔersA background. On these bases, we evaluated the mRNA levels of both anr and dnr expressed under anaerobic conditions in surface-grown cells of the PAO1 wild-type and ΔersA strains. Total RNAs were extracted from cell cultures grown for 3days in agar plates with BHI supplemented with 100mm KNO₃ in a jar for anaerobiosis and analyzed by qRT-PCR. As shown in Table 2, the expression of anr showed a 1.26-fold decrease in the ΔersA mutant relative to the wild-type strain. This was accompanied by a 1.53-fold decrease in dnr expression again in the ΔersA mutant.

These results are consistent with positive ErsA-mediated regulation of Anr in denitrification conditions. According to the well-known effect that translation potential affects mRNA stability, it is conceivable that the anr mRNA is more prone to degradation because of lower translatability in the absence of ErsA. Lower levels of Anr would lead to a lower degree of activation of the transcription of dnr and thus to a decrease in the abundance of dnr mRNA.

Although it was possible to detect the regulatory effects of ErsA on Anr and, subsequently, on Dnr during anaerobic respiration (Table 2), this ErsA role did not appear to be critical for P. aeruginosa PAO1 growth in these conditions. Since the experiments described above were not suited to quantitatively compare the growth efficiency between PAO1 wild-type and the ΔersA mutants, we repeated the bacterial cell plating on BHI supplemented with KNO₃ and incubated in anaerobiosis by spotting calibrated volumes of serial dilutions of quantified cell suspensions. We aimed also to assay in this way the other more virulent reference P. aeruginosa PA14 strain and the corresponding ΔersA mutant (Ferrara et al., 2015). As anticipated, no substantial differences in growth rate and efficiency under denitrification conditions were detected between both PAO1 (Figure 3) and PA14 and their corresponding ΔersA mutants. However, some important metabolic pathways occurring in low-oxygen conditions, such for instance denitrification, are perturbed by the competition for Hfq by the sRNA CrcZ, which can result in diminished anoxic growth and biofilm formation in P. aeruginosa (Pusic et al., 2016). It was indeed speculated that the Hfq sequestration-mediated function of CrcZ in limiting biofilm formation might be associated with the adaptive microevolution of P. aeruginosa for long-term persistence in the harsh environment of the CF airways (Pusic et al., 2016). Therefore, we wondered whether the regulatory effects of ErsA on Anr might be critical for the growth of a clinical strain of P. aeruginosa adapted for CF. To this end, we performed the calibrated bacterial cell plating on BHI supplemented with KNO₃ and incubation in anaerobiosis comparing the P. aeruginosa multi-drug resistant clinical isolate RP73 (Bianconi et al., 2015) with the corresponding ΔersA mutant strain (Ferrara et al., 2020). As shown in Figure 3, RP73 ΔersA is strongly impaired in growing under denitrification conditions compared to wild type as evidenced by colonies extremely smaller and plating efficiency at least 10-fold lower. In the absence of KNO₃ without oxygen, we also evaluated arginine fermentation for both PAO1 and RP73 and the corresponding ΔersA mutants. As shown in Figure 3, differently from PAO1, the loss of ErsA in RP73 strongly impairs also arginine fermentation, the other energy process for which Anr is critical. Despite the ersA gene deletion in the polA-engB intergenic region shows no polar effects on flanking genes (Falcone et al., 2018) and the mutagenesis protocols that we used to generate it are suited to hinder secondary off-site mutations (Ferrara et al., 2015), we evaluated the effects on denitrification capacity of the RP73 ΔersA strain following the reintroduction of the ersA gene expressed from pGM-ersA. As shown in Figure 3, pGM-ersA could complement the nitrate respiration in the RP73 ΔersA, further supporting the notion that ErsA regulation is key for denitrification in the RP73 CF-adapted strain. Taken together, this suggested that P. aeruginosa adaptation to CF lung might result in a higher dependence on ErsA for the regulation of anaerobic energy metabolism.