2.2. RNA I Displays a Sub-Micromolar Binding Affinity for Hfq Protein In Vitro

Next, we determined the binding affinity of Hfq for RNA I. Hfq–RNA binding affinities were determined by the gel-shift assay, as described in the Section 4 Materials and Methods. Here, we compared the binding affinity of RNA I to DsrA (87 nucleotides), one of the earliest identified and well-characterized sRNAs that can bind Hfq [48]. The single-stranded AU-rich region between stem loops I and II of DsrA primarily serves as an Hfq binding site [49], with both the proximal face and the outer rim of Hfq interacting with DsrA, as mutations within these surfaces reduce the respective amounts of DsrA co-immunoprecipitated with Hfq and its steady-state levels [50]. Our in vitro gel-shift assay on DsrA and Hfq complexes revealed that the first band-shift appeared at a monomeric Hfq concentration of 86.9 nM (Figure 3A, lane 4), with a second band-shift arising at an Hfq concentration of 695 nM (Figure 3A, lane 7). Quantification showed that the first DsrA binding affinity (K₁) was 122 ± 12 nM Hfq monomer (Figure 3C). The binding affinities of DsrA reported in the literature are 221 nM Hfq monomer for K₁ and 4 μM Hfq monomer for K₂ [51], with another group reporting 21 ± 1 nM Hfq₆ for DsrA K₁ and 94 ± 5 nM Hfq₆ for K₂ (i.e., about 126 nM Hfq monomer and 564 nM Hfq monomer, respectively) [52]. Thus, our binding affinity data were consistent with previous studies.

Open in a separate window

Figure 3

RNA I displays a sub-micromolar binding affinity for Hfq protein in vitro. (A) Gel-shift mobility assay of DsrA–Hfq interactions. Free-form (DsrA) and the Hfq-bound form (DsrA-Hfq) are as indicated. The Hfq monomer concentration is shown at the top of the gel. Samples were resolved by 6% native PAGE. (B) Gel-shift mobility assay of RNA I–Hfq interactions. (C) Quantification of RNA–Hfq binding percentages. The Y-axis shows the percentage of Hfq-bound RNA relative to total RNA, setting lane 1 as 0% binding. The X-axis indicates the Hfq monomer concentration (nM) in log scale. Quantification was based on at least three technical repeats.

Under the same reaction conditions, we determined the binding affinity of RNA I for Hfq. Similar to our results for DsrA, RNA I also displayed two gel-shift signals in the in vitro gel-shift assay. The first band-shift appeared at a monomeric Hfq concentration of 173.8 nM (Figure 3B lane 5), and a second band-shift appeared at an Hfq concentration of 1390 nM (Figure 3B lane 8). Quantification revealed the first RNA I binding affinity (K₁) to be 225 ± 21 nM Hfq monomer (Figure 3C). Thus, RNA I has a lower binding affinity for Hfq compared to DsrA, yet our results demonstrated that RNA I exhibits a binding affinity for Hfq in the sub-micromolar range that is comparable to previously reported binding affinities for other sRNAs [51,53].

2.3. Proximal Face Mutants of Hfq Decrease Both DsrA and RNA I Binding, and These Two RNAs Compete with Each Other for Hfq Binding

Hfq has three RNA-binding surfaces [2], i.e., the proximal face that preferentially binds single-stranded U-rich sequences [15], a distal face that preferentially binds A-rich sequences [16], and a rim that helps RNA–RNA annealing with low binding sequence specificity [17]. Class I sRNAs bind to the proximal face and the outer rim of Hfq, whereas class II sRNAs interact with both the proximal and distal faces of Hfq protein [2]. We determined which Hfq face (proximal or/and distal face) interacts with RNA I using Hfq proximal face (Q8A, K56A, H57A, and K56A/H57A) and distal face (Y25D*) mutants, following Mikulecky et al. [52]. We used DsrA as a control to validate the effects of Hfq mutations on Hfq–sRNA binding. Based on the binding affinity results shown in Figure 3A, we selected 43 nM and 174 nM Hfq monomer concentrations to analyze binding of DsrA to wildtype or mutant Hfq (Figure 4). In the gel-shift assays, we observed that the proximal face mutants (Q8A, K56A, H57A, and K56A/H57A) disrupted DsrA binding to a greater extent than wildtype (with the highest disruption identified for the K56A/H57A double-mutant, binding of 13.2 ± 6.9% relative to binding of wildtype Hfq, Hfq_WT), as evidenced by more free-form RNA appearing at the bottom of the gel (Figure 4A, and relative binding percentage shown in Table 1). In contrast, the distal face mutation (Y25D*) slightly enhanced DsrA binding (Figure 4A, lanes 6 and 7) relative to Hfq_WT (Figure 4A, lanes 2 and 3). This higher binding to the Y25D* mutant is consistent with a previous report by Mikulecky et al., showing that the distal face Y25D* mutant has a higher binding affinity for DsrA (K_d of 16 nM of Hfq_Y25D* hexamer, i.e., 96 nM of Hfq_Y25D* monomer, compared to the K_d of 21 nM of Hfq_WT hexamer, i.e., 126 nM of Hfq_WT monomer) [52].

Open in a separate window

Figure 4

The proximal, but not distal, face of Hfq mediates binding of both RNA I and DsrA, and the two RNA species compete with each other for Hfq binding. (A) Gel-shift mobility assay of DsrA with Hfq mutant variants. Free-form RNA (DsrA) and the Hfq-bound form (DsrA-Hfq) are indicated. Hfq variants and their monomer concentrations (nM) are indicated at the top of the gel. Samples were resolved by 6% native PAGE. (B) In vitro gel-shift mobility assay of RNA I with Hfq mutant variants. (C) Competition assay between DsrA and RNA I. Free-form RNA (labeled as either DsrA or RNA I) and the respective Hfq-bound forms (labeled as DsrA–Hfq or RNA I–Hfq) are indicated. ³²P-labeled RNA (labeled as either hot DsrA or hot RNA I) and non-isotope-labeled RNA (labeled as either cold DsrA or cold RNA I) are as indicated at the top of the gel. Reactions lacking Hfq monomer are indicated. Increasing concentrations of cold DsrA, ranging from 25 to 200 nM, and cold RNA I, ranging from 850 nM to 6.8 μM, are represented by white triangles. Competition reactions were resolved by 6% native PAGE.

Using these same Hfq mutants under identical reaction conditions, we analyzed the surface by which RNA I binds to Hfq. Since RNA I has a lower binding affinity for Hfq than DsrA, for this experiment we used 130 nM and 260 nM Hfq monomer concentrations. Whereas the percentage binding of RNA I was reduced to varying degrees for proximal face mutants (also with the highest decrease observed for the K56A/H57A double-mutant, with 13.8 ± 8.6% relative to binding of Hfq_WT; Figure 4B and Table 1), the distal face mutant slightly increased the RNA I binding percentage (Figure 4B, lanes 6 and 7), indicating that like DsrA, RNA I interacts with the same face of Hfq.

To validate this result that RNA I and DsrA interact with the same Hfq surface, we used in vitro competition experiments to determine if these two RNAs compete with each other for Hfq binding in the presence of a limited amount of Hfq protein. We used non-isotope-labeled (cold) RNA as a competitor against an isotopically labeled (hot) RNA species. Although DsrA exhibits a higher binding affinity for Hfq than RNA I, we observed that by increasing the amount of cold RNA I (from 850 ng to 6.8 μg), it gradually outcompeted DsrA for Hfq binding, as revealed by more free-form (unbound) DsrA signal detected in the bottom of the gel (Figure 4C, left panel, lanes 2–6). Likewise, cold DsrA could outcompete hot RNA I for binding of Hfq (Figure 4C, right panel, lanes 8–12), albeit at a lower cold RNA concentration (from 25 ng to 200 ng), consistent with its higher binding affinity for Hfq. Taken together, these data demonstrate that plasmid-borne RNA I can bind to the proximal face of Hfq with a high binding affinity and is able to compete with a chromosome-borne sRNA (DsrA) for Hfq binding in vitro.

2.4. Hfq protects RNA I from RNase E-Mediated Degradation Predominantly through Its Proximal but Not Distal Face in E. coli

As shown in Figure 2, Hfq binding protects RNA I from RNase E cleavage in vitro. Moreover, RNA I binds Hfq through its proximal, but not distal, face (Figure 4). Since RNase E cleavage at the 5′ end of RNA I is the rate-limiting step for RNA I degradation, we explored if Hfq binding could stabilize RNA I in vivo and whether the resulting stabilization effect preferentially operates through the Hfq proximal face. To do so, we ectopically expressed Hfq_WT, a proximal face double-mutant (Hfq_K56A/H57A), or a distal face mutant (Hfq_Y25D*) in Escherichia coli lacking the chromosomally encoded hfq gene (i.e., hfq deletion), and then measured RNA I stability to compare how these mutations affect RNA I stability (i.e., protection of RNA I from degradation). We employed rifampicin to stop de novo RNA synthesis and then determined the amounts of RNA remaining at different time points after rifampicin treatment. RNA I stability for the Hfq_Y25D* distal face mutant (Δhfq/pBAD-Hfq_Y25D*) was equivalent to that for the Hfq_WT strain (Δhfq/pBAD-Hfq_WT; half-life >8 min, calculated as 25.7 ± 6.6 min and 28.9 ± 5.4 min, respectively; Figure 5A,B), indicating that the distal face mutation (Y25D*) does not impair the ability of Hfq to protect RNA I from RNase E cleavage. Notably, the Hfq_WT-expressing strain generated more RNA I_FL than the Hfq_Y25D*-expressing strain, which accumulated more RNA I_-5 (Figure 5A), as discussed further below. In contrast to the distal face mutant, the Hfq proximal face double-mutant (Δhfq/pBAD-Hfq_K56A/H57A) significantly shortened the RNA I half-life from >25 min to 6.1 ± 0.3 min (Figure 5A,B), supporting that mutations in the proximal face limit Hfq from adequately protecting RNA I from degradation. Therefore, the protective ability of Hfq for RNA I predominantly operates through its proximal face.

Open in a separate window

Figure 5

Hfq protects RNA I from RNase E-mediated degradation primarily through its proximal face. (A) Northern blot of RNA I. Full-length RNA I (RNA I_FL) and its degradation product (RNA I_-5) are indicated. Methylene blue staining of tRNA was used as a loading control. Time after rifampicin treatment and the strains from which total RNA were isolated are shown. Average half-lives were calculated from at least three biological repeats. (B) RNA I degradation rate. The Y-axis indicates the signals of remnant RNA I from each lane after rifampicin treatment compared to time 0. Signal from time 0 was set as 100%. Fraction (%) of remnant RNA I is plotted in log scale on the Y-axis. The X-axis indicates the time before and after rifampicin treatment (min). Δhfq/pBAD-Hfq_WT is shown in black, Δhfq/pBAD-Hfq_Y25D* in blue, and Δhfq/pBAD-Hfq_K56A/H57A in orange. Half-life was determined as described in the Section 4 Materials and Methods.

3.1. Hfq Might Have Other Binding Sites in Addition to the 5′ end Single-Stranded AU-Rich Region of RNA I

Many sRNAs are stabilized by binding of Hfq since both Hfq and RNase E share similar target sequences (i.e., AU-rich single-stranded sequences) for either binding or cleavage [21]. ColE1 plasmid-borne antisense RNA I is a well-known target of RNase E that is cleaved between the fifth and sixth nucleotides of its 5′ end single-stranded AU-rich region [43,44]. Our in vitro Hfq binding and protection of RNase E cleavage assay demonstrated binding of Hfq to that region of RNA I, thereby reducing RNase E cleavage, and our in vivo RNA I half-life data showed that Hfq expression prevents RNA I from degradation (Figure 2 and Figure 6). However, despite using a high-micromolar concentration of Hfq (16 μM monomer, which proved sufficient to form the Hfq–RNA I complex, Figure 3B) in our N-RNE cleavage assay (Figure 2E, lane 6), it still could not completely protect RNA I_FL from N-RNE cleavage (at an N-RNE concentration of 750 nM; Figure 2E, lane 6). The intracellular Hfq level in E. coli is ~55,000 monomer (or ~9170 hexamer) molecules per cell (corresponding to ~91 μM Hfq monomer, or ~15.2 μM Hfq hexamer) under exponential growth in LB medium at 37 °C [55], i.e., a concentration that is ~5.7-fold higher than the concentration we used in our in vitro assays, indicating that Hfq-mediated protection of RNA I can occur in vivo.

As shown in Figure 1, the RNA I 5′ end deletion variant (RNA I_-8) still binds Hfq at a concentration of 173.8 Hfq monomer (i.e., it does not completely lack binding of the chaperone; Figure 2B), indicating other site(s) binding is obviously involved. It is possible that Hfq binding to these sites is also involved in preventing RNase E cleavage. RNase E is known to have multiple cutting sites in RNA I [56], and a recent study by Richards and Belasco proposed a scanning model by which RNase E searches its downstream cleavage sites by scanning linearly from the 5′ end monophosphate along the single-stranded regions of the RNA body [57]. Obstacles, such as bound proteins or sRNAs, but not stem-loop structures between the 5′ end and the internal RNase E cleavage sites, hinder cleavage of the downstream sites [57]. Therefore, certain Hfq binding sites within the RNA I body, apart from in its 5′ end, may also interfere with the scanning activity of RNase E and cleavage of those internal RNase E sites, consequently affecting RNA I stability in vivo.

3.2. Hfq Binding Face for RNA I

By using Hfq proximal and distal face mutants, we have shown that antisense RNA I predominantly interacts with the proximal and not distal face of Hfq (Figure 4A,B). However, we cannot rule out the possibility that RNA I might interact with other sites of Hfq, such as the rim or C-terminal domain. A recent study using synchrotron radiation-based circular dichroism (SRCD) reported that the C-terminal domain of Hfq facilitates annealing of the first stem-loop fragment of RNA I to its perfectly complementary stem-loop fragment in RNA II [46]. Since RNA–RNA annealing is known to be catalyzed by the arginine patch on the outer rim of Hfq [17], it is possible that the rim surface of Hfq is also involved in binding to and catalyzing RNA I–RNA II annealing.

3.3. How Mutations of Hfq Binding Surfaces Affect RNA I Stability

A previous study has shown that degradation of RNA I_-5 intermediates depends on 3′ polyadenylation [42], with RNA I_-5 rendered very stable in the absence of pcnB (which encodes poly(A) polymerase I, PAP I), resulting in a very low plasmid copy number [42,58]. In addition, PAP I was found to co-purify with both Hfq and PNPase [59]. In our study, we detected very low amounts of RNA I_-5 in an hfq deletion strain (Δhfq/pBAD), slightly more in the hfq wildtype strain (MG1655/pBAD), higher again in the Hfq_WT overexpression strain (MG1655/pBAD-Hfq), and highest of all in the distal face Hfq_Y25D* mutant overexpression strain (Δhfq/pBAD-Hfq_Y25D*; Figure 5A and Figure 6A). We also showed that Hfq_Y25D* binds RNA I more strongly than Hfq_WT, consistent with other proximal face-binding sRNAs (Figure 4B) [52]. This higher binding capacity of RNA I for the Hfq_Y25D* mutant is correlated with greater accumulation of RNA I_-5. One possibility is that the Y25D* mutant reduces the interaction between Hfq protein and PAP I, thereby inhibiting polyadenylation and stabilizing RNA I_-5. Moreover, the increased accumulation of RNA I_-5, in accordance with increased expression/binding of Hfq, indicates that the presence of Hfq may impact the degree to which RNA I is polyadenylated by influencing the structure of RNA I, representing a well-known function of the chaperone protein. Although the polyadenylation activity of PAP I is reduced in an hfq deletion strain [59,60], our data indicated that Hfq-mediated stimulation of PAP I-driven polyadenylation might depend on the target RNA species and/or target RNA structure.

4.2. Protein Expression and Purification

For the in vitro RNase E cleavage assay, Flag-tagged N-terminal RNase E (N-RNE, encoding aa 1 to 597, carried by the pflagRE1 plasmid [71]) was purified by using the BL21(DE3) rne131 bacterial strain as a host. The bacterial culture was grown in LB medium supplemented with 50 μg/mL of ampicillin with 160 rpm shaking at 32 °C to A₆₀₀ = 0.5. N-RNE was induced by 1 mM IPTG for 2 h, and then the bacterial culture was harvested at 5000× g, 4 °C, for 5 min. The N-RNE protein was purified, as previously described [71]. To lyse bacterial cells, the bacterial culture was resuspended in Flag-binding buffer (20 mM Tris-HCl pH 8.0, 10 mM KCl, 10 mM MgCl₂, 20 mM EDTA, 200 mM NaCl), and then passed through a French Press One-Shot Model (Constant Systems Limited, Daventry, UK) to obtain total cell lysate. Cell debris was removed by centrifuging the cell lysate (12,000 rpm, 4 °C, for 10 min) and passage through a 0.45 µm filter. The N-RNE was purified from cleared total lysate using ANTI-FLAG^® M2 Affinity Gel (Sigma-Aldrich, Burlington, MA, USA). Buffer A (50 mM Tris-HCl pH 8.0, 0.15 M NaCl) was used to wash non-specific proteins from the M2 affinity column, and then 1 mL of Flag-peptide (1 mg/mL) was used to elute N-RNE. The eluted N-RNE was further dialyzed to remove Flag-peptide, as described previously [71]. In brief, the purified N-RNE was placed in a dialysis bag (MWCO 6000–8000) and dialyzed twice into storage buffer (50 mM Tris-HCl pH 8.0, 0.15 M NaCl, 1 mM EDTA) for 12 h at 4 °C. The dialyzed N-RNE protein was supplemented with a final concentration of 10% glycerol and stored at −20 °C until needed.

For Hfq protein purification, the MG1655Δhfq bacterial strain with either wildtype hfq plasmid or one of the various hfq mutant plasmids (pBAD-His-Hfq, pBAD-His-Hfq _(Q8A), pBAD-His-Hfq _(Y25D*), pBAD-His-Hfq _(K56A), pBAD-His-Hfq _(H57A), and pBAD-His-Hfq _(K56A/H57A)) was cultured at 32 °C in LB medium supplemented with 50 μg/mL of ampicillin and 160 rpm shaking to A₆₀₀ = 0.3. His-tagged Hfq and its mutant variants were induced by a final concentration of 0.1% arabinose for 2 h, as described previously [71]. Bacterial culture (500 mL equivalents) was then harvested by centrifugation and resuspended in 15 mL of His-binding buffer (20 mM sodium phosphate, 0.5 M NaCl, 40 mM imidazole) containing 1 mM of PMSF. The cells were lysed by passage through a French Press (Aminco Model FA-073) twice at 10,000 psi. Cell debris was removed, as described above. The cleared cell lysate was then incubated with 1 mL of Ni Sepharose^TM 6 Fast Flow (GE Healthcare, Chicago, IL, USA) for 1 h at room temperature in a Poly-Prep chromatography column (Bio-Rad, Hercules, CA, USA). The column was then washed with His-binding buffer until the pass-through buffer reached A₂₈₀ < 0.05 mg/mL. To elute the His-tagged Hfq protein, six elutions were prepared, as follows: E1/E2 (200 mM imidazole), E3/E4 (400 mM imidazole), and E5/E6 (600 mM imidazole). His-tagged Hfq protein was eluted with 1 mL of each elution buffer (20 mM sodium phosphate, 0.5 M NaCl, plus 200 mM, 400 mM, or 600 mM imidazole). Purified His-tagged Hfq protein was concentrated using a Vivaspin 500 MWCO 3000 system (GE Healthcare) and dialyzed against PBS buffer (137 mM NaCl, 2.7 mM KCl, 8 mM Na₂HPO₄, 1.46 mM KH₂PO₄). The His tag was removed by AcTEVTM protease (Invitrogen) and confirmed by 15% SDS-PAGE. The resulting purified Hfq protein was stored at −80 °C with 5% glycerol. The purified Hfq was used to generate rabbit polyclonal antibody and for in vitro binding or competition assays.

4.3. RNA Preparation, N-RNE In Vitro Cleavage, In Vitro Protein–RNA Gel-Shift Assay, and RNA–Protein Competition Assay

To prepare RNA species used for N-RNE cleavage assay and gel-shift assays, RNA species were synthesized in vitro using a T7 in vitro transcription kit (MAXIscript, Ambion, Glasgow, UK) that contained purified RNA T7 RNA polymerase, according to the manufacturer’s instructions. All in-vitro-synthesized RNA species were internally labeled with [α-³²P] UTP. In vitro RNA synthesis by T7 polymerase produces additional GGG at the 5′ end of the RNA, and it is known that this GGG does not affect RNase E cleavage on RNA I [44]. DNA templates used for in vitro transcription were generated by PCR using gene-specific primers, with the forward primers harboring the T7 promoter sequence. For full-length RNA I, pBAD plasmid was used as a template for PCR amplification with primers 14 and 15. To generate the 5′ end eight-nucleotide deletion RNA I variant (RNA I_-8) template, primers 15 and 16 were used. To generate DsrA as a T7 in vitro transcription DNA template (DsrA on chromosome: 43.61 centisomes; map position: 2,023,337 to 2,023,251), chromosomal DNA from the MG1655 strain was used as a PCR amplification template with primers 17 and 18. After in vitro transcription, full-length ³²P-labeled RNA transcripts were purified on a 6% polyacrylamide (19:1)/7 M urea gel and eluted in elution buffer (10 mM Tris pH 7.5, 1 mM EDTA, 0.1% SDS) at room temperature for 90 min. Eluted RNA transcripts were precipitated by adding 1 volume of isopropanol and 0.1 volume of 3 M sodium acetate (pH 7.8) and kept at −20 °C for at least 2 h to facilitate precipitation. Following precipitation, full-length RNAs were pelleted down at 13,000 rpm, 4 °C, for 15 min and washed with 70% ethanol. The RNA pellets were then re-suspended in 20 µL of DEPC H₂O and stored at −20 °C before further experimentation. For the competition assay, non-isotope-labeled (cold) RNA transcripts were prepared, as described above, except using non-radioactive UTP and without PAGE purification. BR13 corresponding to the first 13 mers, identical to the T7 synthesized RNA I 5′ end (5′-GGG ACA GUA UUU G-3′) [44] (shown in Figure 2A) carrying the fluorescein modification at the 3′ end, was obtained from a previous study [71].

To perform in vitro Hfq protection and RNase E cleavage assays, the procedure was adapted from a previous publication [44]. In brief, 30,000 cpm (counts per minute) of purified internally ³²P-labeled RNA I with 20 pmol of cold RNA I were pre-incubated with 0, 0.9, 1.8, or 3.6 μg of monomeric Hfq (final concentrations of 0, 4, 8, or 16 μM) based on optimized conditions in reaction buffer (20 mM Tris-HCl pH 8, 100 mM NaCl, 10 mM MgCl₂) at 37 °C for 20 min. Then, 1 μg of N-RNE (final concentration of 750 nM) was added into each reaction for in vitro cleavage at 37 °C for 1.5 h. To stop N-RNE in vitro cleavage, 1 volume of phenol:chloroform:isoamyl alcohol (25:24:1; Affymetrix, Santa Clara, CA, USA) was added into each sample, followed by brief vortexing and centrifugation at 13,000 rpm for 2 min. The aqueous phase, which contained the RNA species, was resolved on an 8% acrylamide/7 M urea gel at 11 W for 3 h, and the radioactivity signal was visualized using a Typhoon FLA 9000 imaging system (GE Healthcare). The Hfq protection and the percentage of N-RNE cleavage were calculated by measuring and comparing the signal intensity of un-cleaved RNA I (RNA I_FL) with the cleaved RNA I (RNA I_-5). The ratio of RNA I_-5 to RNA I_FL for each Hfq concentration was plotted using Prism (GraphPad Software version 9, Boston, MA, USA, www.graphpad.com).

To determine the binding affinity and competitive ability of RNA transcripts, we performed an in vitro protein–RNA mobility shift assay, as described previously [72]. In brief, purified sRNAs (obtained through in vitro transcription) or 5 pmol of BR13 were denatured by heating them at 90 °C for 3 min and then allowing them to refold into their natural structures by cooling them at room temperature. To determine binding affinities, an aliquot containing 30,000 cpm (counts per minute) of ³²P-labeled RNAs was incubated with 0, 21.7, 43.4, 86.9, 173.8, 347.5, 695, 1390, 2780, or 5560 nM of monomeric Hfq protein. For the competition assay, an aliquot containing 30,000 cpm (counts per minute) of ³²P-labeled RNA and 695 nM (determined from binding affinity data) of Hfq monomer were used. ³²P-labeled RNA was competed by adding increasing amounts of cold DsrA (25 ng, 50 ng, 100 ng, and 200 ng). Cold RNA I ranged from 850 ng, 1.7 μg, and 3.4 μg to 6.8 μg. To measure in vitro binding of Hfq mutants to DsrA and RNA I, a concentration of either 43 nM/174 nM or 130 nM/260 nM of Hfq monomer was used, respectively. sRNAs and Hfq were incubated for 20 min at 25 °C in binding buffer (20 mM Tris-HCl pH 7.6, 0.5 M NaCl, 1 mM EDTA, 1% glycerol), before being subjected to electrophoresis on 5% polyacrylamide (19:1) native gels in 0.5 X TBE running buffer at 250 V and 4 °C for 3.5 to 4 h until the xylene cyanol dye reached the middle of the gel. After electrophoresis, gels were dried on Whatman 3 MM filter paper using a vacuum gel dryer system. Radioactivity signals were captured on a phosphor imaging plate (FUJIFILM Imaging Plate) and visualized using the Typhoon FLA 9000 imaging system (GE Healthcare). Signal intensity was quantified in the Vision Works LS system (UVP, Upland, CA, USA). The percentage of binding was obtained by measuring the total RNA signal compared to the free-form isotope-labeled RNA signals and the band-shift signals under each condition (Figure 3C). Quantifications were obtained from at least three technical repeats. Binding affinity was calculated as the Hfq concentration required for 50% binding. Plotting was performed in Prism 9 (GraphPad Software, Boston, MA, USA, www.graphpad.com). All concentrations of RNAs and protein used were determined from preliminary experiments under identical conditions.

4.4. RNA Half-Life Determination, Total RNA Extraction, and Northern Blot Analysis

Bacterial strains MG1655/pBAD, MG1655/pBAD-Hfq, MG1655Δhfq/pBAD, MG1655Δhfq/pBAD-Hfq, MG1655Δhfq/pBAD-Hfq_(Y25D*), and MG1655Δhfq/pBAD-Hfq _(K56A/H57A) were cultured at 37 °C with 180 rpm shaking in LB medium supplemented with an antibiotic (50 μg/mL of ampicillin). Hfq protein was induced by 0.2% arabinose at the early exponential growth phase and incubated for a further 2 to 2.5 h. Then, the bacterial samples were harvested for further RNA analysis.

To measure the RNA I half-life, de novo RNA synthesis was arrested by adding a 1/50 volume of rifampicin (50 mg/mL), and the remaining RNA I was determined by Northern blotting over time. Cultured cells were collected before and at 1, 2, 4, 6, and 8 min after rifampicin treatment, with 24 mL of cultured cells being collected at each time point and added into 4 mL of stop solution (95% EtOH, 5% saturated phenol, pH 6.6/7.9) to stop all enzymatic activities. The mixed samples were centrifuged at 4000× g for 10 min at 4 °C. Collected bacterial pellets were stored at −80 °C until total RNA had been prepared for half-life determination. Total RNA was extracted, as described in a previous publication [41]. Extracted RNA was separated through an 8% polyacrylamide (19:1)/7 M urea gel in 0.5 X TBE buffer using an Adjustable Height Vertical Electrophoresis System (C.B.S. Scientific, San Diego, CA, USA) at 11 W for ~3.5 h until the xylene cyanol dye had reached three-quarters the length of the gel. The RNA was transferred onto Zeta-Probe^® Blotting Membranes (Bio-Rad) at 400 mA, 100 min, at 4 °C in 0.5 X TBE buffer by Trans-Blot Cell (Bio-Rad) and crosslinked onto the membrane using a Stratalinker UV Crosslinker 2400 (Stratagene, San Diego, CA, USA) at a setting of 120,000 microjoules/cm².

To perform Northern blot detection of RNA I, an antisense DNA probe (5′-GTA ACT GGC TTC AGC AGA GCG CAG ATA CC-3′) was labeled with [γ-³²P] ATP at the 5′ end using T4 polynucleotide kinase (NEB). The radioactively labeled probe was purified via a MicroSpin G-25 column (GE Healthcare) before use. To detect RNA I, membranes were pre-blotted with ULTRAhyb-Oligo hybridization buffer (Ambion) for 2 h at 42 °C. The radioactively labeled probe was then added in hybridization buffer and further hybridized to the membrane at 42 °C for at least 6 h. The RNA I signal detected by the radioactive probe was visualized on an imaging plate (FUJIFILM) and detected using a Typhoon FLA 9000 imaging system (GE Healthcare). The intensity of either 5S rRNA or tRNA on the methylene blue-stained membrane was used as a loading control for normalization. Signal intensity was quantified using VisionWorks LS software version 8.20 (UVP). The RNA I half-life was determined from at least three biological repeats. The half-life of RNA I was determined from a least-squares linear fit to a semi-log plot of remaining RNA I (%) versus time.

4.5. Plasmid Copy Number Determination

Plasmid copy numbers were measured according to a modified protocol from a previous study [41]. In brief, 300 μL of bacterial culture, as described above for our RNA I stability determination, was first pelleted and then resuspended in 60 μL of DNA loading dye (Thermo Fisher Scientific). Then, 120 μL of phenol:chloroform:isoamyl alcohol (25:24:1; Affymetrix) was added. To disrupt the cells and extract all nucleic acid content, the phenol/chloroform mixture was vortexed through a Thermomixer comfort (Eppendorf) at 1400 rpm for 10 min at room temperature and then centrifuged at 15,000× rpm for 10 min. Twenty microliters of the resulting aqueous phase was electrophoresed on a 0.6% agarose gel in 0.5 X TAE buffer at 70 V for 90 min. Since the Hfq-overexpressing strains (i.e., MG1655/pBAD-Hfq and hfq/pBAD-Hfq) have much lower plasmid copy numbers, higher amounts of DNA were loaded for quantification. DNA bands were visualized by staining the agarose gel with ethidium bromide for 5 min and UV illumination with a Gel Logic 212 PRO system (Carestream, Rochester, NY, USA). Chromosome DNA was used for normalization to obtain the relative plasmid copy number for each set of conditions, as described previously [41]. Signal intensity was measured using Fiji software version 2.3.0 [73].

4.6. DNA Isolation and Quantitative PCR (qPCR)

DNA isolation for further plasmid copy number determination was performed as described previously [71], with some modification. In brief, 1.5 mL of bacterial culture (grown as described above for our RNA I stability determination) was pelleted and then resuspended in 1 mL of lysis buffer (10 mM Tris-HCl pH 8.0, 1 mM EDTA pH 8.0, 0.6% SDS, 120 μg/mL proteinase K, 1 mg/mL RNase A). To lyse cells, the resuspended cell mixture was incubated at 37 °C for 1 h. DNA was further isolated with phenol:chloroform and precipitated in three volumes of EtOH at −20 °C for at least 2 h. The precipitated DNA was pelleted at 15,000 rpm, 15 min, 4 °C, and resuspended in 100 μL of TE buffer (10 mM Tris-HCl pH 8.0, 1 mM EDTA pH 8.0) for further experimental usage.

To measure the relative plasmid copy number, quantitative PCR (qPCR) was performed [74]. In brief, a 160 X dilution of isolated DNA from each strain was prepared for qPCR reaction with iQ™ SYBR^® Green Supermix (Bio-Rad, Hercules, CA, USA), according to the manufacturer’s instructions. qPCR was performed on a CFX Opus Real-Time PCR System (Bio-Rad, Hercules, CA, USA). Primers that generate similarly sized amplicons of either chromosome or plasmid sequences were designed: (1) for amplifying chromosome sequences to generate an 88 bp product that mapped to the mreB gene, or (2) for amplifying plasmid sequences to generate a 127 bp product that mapped to the ampA gene. Oligo primers (primers 20–23) used for qPCR (250 nM per reaction) are listed in Table 2. To obtain the relative expression of the plasmid-borne ampA gene representing the relative plasmid level in different strains, we adapted the 2^−ΔΔCt method [75]. In brief, the Ct value of the plasmid-borne ampA gene was first normalized against the chromosome-borne mreB gene to obtain ΔCt. The ΔCt value for each strain was then normalized against either MG1655/pBAD (Figure 6F) or hfq/pBAD-Hfq_Y25D* (Figure 6G) to obtain ΔΔCt. Relative fold-changes of plasmid levels in different strains were calculated as 2^−ΔΔCt. At least three biological repeats with triplicate technical repeats were performed for each target gene. Plotting was carried out in Prism 9 (GraphPad Software, Boston, MA, USA, www.graphpad.com).

We thank J. O’Brien for manuscript editing services. We thank D. Singh and Y.-C. Lee for useful discussions. We thank the IMB Animal Facility for support in generating the Hfq antibody (rabbit). We thank H.-C. Hung of the IBMS Medical Art Department for his assistance in drawing Figure 1 and Figure 7. We thank the editor and each reviewer for their insightful and critical comments and suggestions. This work fulfills partial requirements for the Ph.D. thesis of W.-S.W., Molecular and Cell Biology, Taiwan International Graduate Program, Academia Sinica and Graduate Institute of Life Science, National Defense Medical Center, Taipei, Taiwan.