Bacterial strains, plasmids, and oligonucleotides used in this study are listed in Tables 1–3, respectively. Staphylococcus aureus strains were grown in Tryptic Soy Broth (Pronadisa) supplemented with 0.25% glucose (TSBg) or, when indicated, in modified chemically defined medium (Toledo-Arana et al., 2005). To prepare S. aureus and E. coli competent cells, B2 (casein hydrolysate, 10 g L⁻¹; yeast extract, 25 g L⁻¹; NaCl, 25 g L⁻¹; K₂HPO₄, 1 g L⁻¹; glucose, and 5 g L⁻¹; pH 7.5), and SuperBroth (tryptone, 30 g L⁻¹; yeast extract, 20 g L⁻¹; and MOPS, 10 g L⁻¹; pH 7) media were used, respectively. For selective growth, media were supplemented with the appropriate antibiotics at the following concentrations: ampicillin (Amp), 100 μg ml⁻¹ for E. coli transformants; erythromycin (Erm), 1,5 μg ml⁻¹ or 10 μg ml⁻¹ for S. aureus cells harboring the pMAD or pCN plasmids, respectively.

Most of the plasmids used in this study were engineered as previously described (Caballero et al., 2018; Menendez-Gil et al., 2020). pMAD plasmids used for chromosomal deletions were constructed by amplifying flanking sequences (AB and CD) of the target regions using primers A/B and C/D (Table 3). PCR fragments were digested and ligated into pMAD through a double-fragment ligation process using BamHI, EcoRI, and KpnI or NheI (Tables 2, 3).

The green fluorescent protein (GFP) reporter plasmids were constructed using the Listeria monocytogenes pAD-cGFP plasmid as a template (Balestrino et al., 2010). To build pGFP, the hly 5’UTR and GFP sequences were amplified with primers Sal-GFP-fw and BcuI-TT-BamHI-GFP-rvs (Table 3) and the resulting PCR fragment was cloned into the pEW plasmid. The 3’UTR of ftnA was amplified using primers BamHI-EcoRI-3UTR-ftn-fw and SmaI-3UTR-ftn-rvs (Table 3) and inserted downstream of the gfp gene using restriction sites BamHI and SmaI. The pGFP-Δ3’UTR^ftnA was constructed using the pGFP-3’UTR^ftnA as a template and primers SalI-GFP-fw and KpnI-D3UTR-term-ftn (Table 3). The amplification product was ligated into the pGFP plasmid using SalI and KpnI.

The plasmids expressing ^3XFFtnAΔ3’UTR^19-56 and ^3XFFtnAΔ3’UTR^57-93 were constructed using the p^3xFFtnA plasmid backbone. For the ^3XFFtnAΔ3’UTR^19-56 plasmid, an overlapping PCR was performed using oligonucleotide pairs +1-ftn and 3’UTR-ftn-19-56-fw and 3’UTR-ftn-19-56-rv and term-ftn. Analogously, oligonucleotides +1-ftn and 3UTR-ftn-term-1/2 were used for the ^3XFFtnAΔ3’UTR^57-93 plasmid (Table 3). The resulting amplicons were then inserted into pEW using BamHI and KpnI (Table 2).

Precultures were grown in 5 ml of TSBg supplemented with Erm (TSBg+Erm) and incubated overnight (ON) at 37°C and 200 rpm. Bacterial concentrations of the precultures were estimated by measuring their optical density (OD₆₀₀). Erlermeyer flasks containing fresh TSBg+Erm were inoculated with precultures to reach starting OD₆₀₀ of 0.02. For rifampicin mRNA stability assays, test cultures were grown at 37°C and 200 rpm until an OD₆₀₀ of 0.5 (exponential phase). Six aliquots of 20 ml of the culture were transferred to 50 ml Falcon tubes containing 300 μg ml⁻¹ of rifampicin and incubated at 37°C for 0, 2, 4, 8, 15, and 30 min. Then, 5 ml of stop solution (95% ethanol and 5% phenol) were added to the samples and centrifuged for 2 min at 4,400 g. Pellets were frozen in liquid nitrogen and stored at −80°C. RNA extraction and Northern blot analysis were performed as described in Toledo-Arana et al. (2009) and Menendez-Gil et al. (2020). Radiolabeled riboprobes were synthesized from a PCR carrying the T7 promoter (Table 3) using the MAXIscript T7 transcription kit (Ambion) and [α32P]-UTP, following the manufacturer’s recommendations. These riboprobes were designed to target the ftnA or the ^3XFftnA mRNAs. The mRNA levels were quantified by densitometry of Northern blot autoradiographies using ImageJ.¹ Each of the mRNA levels was normalized to the levels of the sample at 0 min of rifampicin incubation.

The mutants generated in this study (Table 1) were obtained as previously described (Valle et al., 2003) by a two-step homologous recombination that exchanges a specific chromosomic region by the mutant allele present in the pMAD plasmid (Arnaud et al., 2004). The marker-less mutants were verified by PCR using oligonucleotides E and F (Table 3) and Sanger sequencing.

Bacteria were grown as described above. At OD₆₀₀ 0.5, 30 ml samples were taken for protein extraction as described by Menendez-Gil et al. (2020). Western blotting was performed as previously described (Caballero et al., 2018). The 3xFLAG tagged protein samples were incubated with mouse monoclonal anti-FLAG M2-Peroxidase (HRP) antibodies (Sigma) diluted 1:1,000, whereas the GFP samples were incubated with mouse monoclonal anti-GFP antibodies 1:5,000 (Living Colors, Clontech) and peroxidase-conjugated goat anti-mouse immunoglobulin G and M antibodies 1:2,500 (Pierce-Thermo Scientific). Membranes were developed using the SuperSignal West Pico Chemiluminiscent Substrate kit (Thermo Scientific). Mean intensities of developed protein bands were quantified by densitometry of Western blot images using ImageJ and plotted as arbitrary units (A.U.). Statistical significances were calculated by running a paired t-test in GraphPad Prim; asterisks (*) indicate p-values lower than 0.05 (p < 0.05) while ns indicate not significant differences.

Precultures were grown in 5 ml of modified chemically defined medium without iron (MM^wo/Fe) overnight at 37°C and 200 rpm (Toledo-Arana et al., 2005). Since all glass material contains iron traces, bacteria can grow in this medium. In order to eliminate the remaining free iron, we used 2′2-dipyridil (DIP; Sigma) as an iron chelator. Precultures were normalized to OD₆₀₀ 0,1 and 5 μl of these aliquots were diluted in 195 μl of modified chemically defined medium containing different concentrations of DIP in 96-well microtiter plates. The growth curve was monitored using the SpectraMax 340 PC Microplate Reader (Molecular Devices). OD₆₅₀ measurements were performed every 30 min at 37°C for a period of 20 h.

In a previous study, we showed that deletion of the ftnA 3’UTR increased both the ftnA mRNA and ferritin protein levels in S. aureus (Menendez-Gil et al., 2020). To evaluate whether the ftnA 3’UTR deletion affected ftnA mRNA stability, we performed rifampicin mRNA stability assays and half-life determinations. To that end, we transformed the S. aureus 15981 ΔftnA strain with the p^3xFFtnA and p^3xFFtnAΔ3’UTR plasmids, which expressed the WT and Δ3’UTR ftnA mRNAs under the control of the P_blaZ constitutive promoter, respectively. This allowed us to exclusively monitor the plasmidic ftnA gene using a strand-specific riboprobe. The resulting strains were grown until exponential phase and their total RNAs extracted at different time points after rifampicin addition. Northern blots revealed that the half-life of the ftnA mRNA was higher in the p^3xFFtnAΔ3’UTR strain (2.5 min) when compared to the p^3xFFtnA strain (0.8 min; Figure 1). Note that since the decrease in concentration of the two mRNAs was not exponential for all time points, we used only the first three time points (0–4 min) for the WT mRNA and the first four time points (0–8 min) for the Δ3’UTRftnA deletion mutant to calculate their half-life (Figure 1B). These results indicated that the ftnA 3’UTR may be targeted by unknown RNases to modulate FtnA expression.

In order to identify the RNases that could potentially target the ftnA 3’UTR, we transformed the most relevant non-essential RNase mutants (Bonnin and Bouloc, 2015) of the S. aureus 15981 WT strain with the p^3xFFtnA and p^3xFFtnAΔ3’UTR plasmids, which expressed the 3xFLAG-tagged ferritin protein (^3xFFtnA) from the WT and Δ3’UTR ftnA mRNAs, respectively. The selected RNase mutants included ∆rnc (RNase III, a dsRNA endonuclease; Lasa et al., 2011; Lioliou et al., 2012), ∆mrnc (mini-RNase III, a double-stranded RNA endonuclease paralog to RNase III, which was initially identified in B. subtilis; Olmedo and Guzmán, 2008, with an uncharacterized ortholog in S. aureus), ∆pnpA (PNPase, 3′-5′ exonuclease; Anderson and Dunman, 2009), ∆rnr (RNase R, 3′-5′ exonuclease; Oussenko et al., 2002), ∆rny (RNase Y, a single-stranded RNA endonuclease; Marincola et al., 2012), and ∆rnjA (RNase J1, a bifunctional RNase with endonuclease and 5′ to 3′ exonuclease activities; Linder et al., 2014). We then determined their ^3xFFtnA protein levels by Western blot using anti-FLAG antibodies (Figure 2A). As expected, the S. aureus 15981 WT strain that carried the p^3xFFtnAΔ3’UTR plasmid expressed higher ^3xFFtnA protein levels than the strain expressing the whole ftnA mRNA. Such protein increase was also obtained when ferritin was expressed from the p^3xFFtnAΔ3’UTR plasmid in the ∆rnr, ∆mrnc, and ∆rny mutant strains. Since ^3xFFtnA could not be detected in the ∆rnjA strain, we performed the same experiment but loading a higher amount of total protein (Figure 2B). The ∆rnjA mutant strains also showed an increase in ferritin expression when the 3’UTR was deleted (Figures 2A,B). These results suggested that RNase R, mini-RNase III, RNase Y, and RNase J1 were not involved in the 3’UTR-mediated processing of the ftnA mRNA. In contrast, the Western blots revealed that the ∆rnc mutant strains carrying the p^3xFFtnA and p^3xFFtnAΔ3’UTR plasmids expressed similar levels of the ^3xFFtnA protein regardless of the 3’UTR deletion from the 3xFLAG-ftnA mRNA. Similar results were obtained in the ∆pnpA mutant strains (Figure 2A). This suggests that RNase III and PNPase could be targeting the ftnA 3’UTR to process the ftnA mRNA.

mRNA decay is often initiated by endoribonucleases, including RNase Y and RNase III (Durand et al., 2015). To confirm the implication of RNase III in the ftnA mRNA decay, we evaluated the half-life of the WT and Δ3’UTRftnA mRNAs in S. aureus 15981 WT and its isogenic ∆rnc mutant strain. The rifampicin mRNA stability assays revealed that the ftnA mRNA half-life increased from 1.3 min in the WT strain to 7.0 min in the ∆rnc mutant (Figures 2C,D). This was higher than the half-life observed for the Δ3’UTRftnA mRNA mutant, indicating that the RNase III might target the ftnA mRNA through additional mechanisms. Note that the half-life of the ftnA mRNA and its Δ3’UTR mutant were similar when expressed from the ∆rnc mutant (7.0 vs. 5.8 min, respectively; Figures 2C,D). Taken together, these results indicate that RNase III promotes ftnA mRNA decay in a process in which the ftnA 3’UTR plays a critical role.

In order to investigate whether the ftnA 3’UTR had functional capacities on its own, we fused the ftnA 3’UTR downstream of the gfp gene, which encodes the GFP, thus, generating the pGFP-3’UTR^ftnA plasmid. As a control, we constructed a plasmid that included the transcriptional terminator (TT) of the ftnA mRNA downstream of the gfp gene (pGFP-∆3’UTR^ftnA; Figure 3A). Then, we transformed the S. aureus 15981 WT strain with these plasmids and determined the GFP levels by Western blot analysis. The results revealed that the pGFP-3’UTR^ftnA plasmid expressed lower GFP levels when compared to the pGFP-∆3’UTR^ftnA plasmid (Figure 3B). This confirmed that the ftnA 3’UTR alone worked as an independent module able to reduce the expression of a heterologous gene like gfp. To analyze whether such GFP expression reduction was mediated by RNase III and PNPase, we introduced the pGFP-3’UTR^ftnA plasmid into the ∆rnc and ∆pnpA mutant strains. Western blot results showed that the ∆rnc and ∆pnpA mutants expressed similar GFP levels to the ones produced by the strain carrying the pGFP-∆3’UTR^ftnA plasmid, indicating that the 3’UTR would still be targeted and processed by RNase III and PNPase regardless of the CDS (Figure 3C).

Since RNase III is a double-stranded endoribonuclease, we looked for putative double-stranded RNA regions within the ftnA 3’UTR secondary structure. We used the RNAstructure version 6.2 software to predict the ftnA mRNA conformation.² We could not find any evident secondary structures susceptible to RNase III within the ftnA 3’UTR, nor a hypothetical 5’UTR-3’UTR interaction, as previously described for the icaR mRNA (de Los Mozos et al., 2013). A plausible alternative would be for the ftnA 3’UTR to be targeted by a trans-acting small RNA that, upon interaction, generated a double-stranded substrate for RNase III to process. This idea would require the putative paring region to be conserved among S. aureus strains and close relatives. Previous multiple sequence alignment analyses revealed that 8,193 out of the 10,000 S. aureus genomes available at the NCBI database presented an ftnA 3’UTR with a 100% of identity. The remaining ones showed just few nucleotide differences (Menendez-Gil et al., 2020). Although, such a high degree of conservation suggested an important role for the ftnA 3’UTR, it prevented us from identifying a putative functional region. Our previous analyses also showed that among other Staphylococcus species carrying the ftnA gene, the ftnA 3’UTR conservation only applied to S. argenteus (Menendez-Gil et al., 2020). Further nucleotide comparison analysis between the S. aureus and S. argenteus ftnA 3’UTRs revealed two conserved regions comprised between nucleotides 18–56 (conserved region I) and 68–89 (conserved region II), respectively, besides the putative TT (Figure 4A). To determine whether one of these regions could be involved in the ftnA mRNA processing, we constructed two plasmids expressing the 3xFLAG tagged ftnA mRNA carrying either a deletion between nucleotides 19 and 56 (p^3xFFtnA∆3’UTR^19-56), or 57 and 93 (p^3xFFtnA∆3’UTR^57-93), which selectively eliminated conserved regions I and II, respectively (Figure 4B). We used such plasmids to transform the S. aureus 15981 WT strain and evaluated their ^3xFFtnA protein expression. Western blot analyses revealed that both mutations produced similar ^3xFFtnA protein levels when compared to the full-length ftnA mRNA, suggesting that deletions of a few nucleotides are not enough to reproduce the effect generated by the ∆3’UTR mutant (Figure 4C). At the same time, it indicated that both conserved regions contribute to RNase III action to modulate ferritin expression.

To evaluate the biological relevance of the ftnA 3’UTR-mediated control of ferritin production, we constructed a chromosomal ftnA∆3’UTR mutant in the S. aureus 15981 genetic background. First, to control that the chromosomal mutant behaved as the plasmidic one, we performed Northern blot analyses to monitor the ftnA mRNA levels. Total RNAs were extracted from the WT and ∆3’UTR mutant strains grown until exponential phase in a rich medium (TSBg). Northern blot results showed that the chromosomal ftnA∆3’UTR mutant expressed higher ftnA mRNA levels than the WT strain as it occurred with the plasmidic mutant (Figure 5A).

Then, we aimed at comparing the capacities of the S. aureus 15981 WT and the chromosomal ftnA∆3’UTR mutant strains to grow under iron starvation conditions. We incubated microplates containing minimal medium lacking iron (MM^wo/Fe) at 37°C and measured bacterial growth by registering the optical density every 30 min. However, no growth differences were observed between the WT and ∆3’UTR mutant in the MM^wo/Fe (Figure 5B). Since iron traces could be still present in this medium, iron starvation might be difficult to achieve under laboratory conditions without the use of chelating agents (Pi and Helmann, 2017). Therefore, the MM^wo/Fe was complemented with increasing concentrations of 2,2′-dipyridyl (DIP), a strong iron chelator. Figure 5B shows that the addition of the DIP chelator at a concentration of 100 μM significantly affected the growth of the ftnA∆3’UTR mutant, while adding 500 μM of DIP completely impaired it. Altogether, these data portrayed the ftnA 3’UTR as an essential module to control the ftnA mRNA expression and maintain proper iron levels for S. aureus growth under iron starvation conditions.