Effect of Seed Length and Binding Motifs on Hfq-Mediated sRNA-mRNA Annealing Analyzed Using Single-Molecule FRET

Summary

A single-molecule FRET assay enables quantitative analysis of Hfq-mediated annealing between bacterial sRNAs and mRNAs, allowing measurement of interaction kinetics and stability. Representative results demonstrate how seed length, Hfq binding motifs, and motif spacing affect RNA duplex formation.

Abstract

Defining the rules that govern annealing between bacterial small RNAs (sRNAs) and their mRNA targets is challenging because these RNAs are highly heterogeneous in sequence and structure. To disentangle the contribution of individual features, we systematically varied key determinants, including seed strength, Hfq-binding motifs, and the spacing between them. The effects of these features on sRNA-mRNA duplex stability were tested using single-molecule Förster resonance energy transfer (smFRET), which enables monitoring of individual RNA molecules on the millisecond timescale and provides access to dynamic events often obscured in ensemble assays.

We demonstrate that short seed regions (4-5 bp) form unstable, short-lived complexes, whereas longer seeds (8-10 bp) produce stable duplexes. Extending Hfq-binding motifs or introducing a spacer between the binding sites on mRNA further increased interaction lifetimes, highlighting the importance of motif strength and spatial arrangement in Hfq-mediated annealing efficiency. Moreover, this approach provides a robust and reproducible framework for dissecting RNA-RNA interaction dynamics at single-molecule resolution and for probing RNA chaperone mechanisms.

Introduction

Bacterial gene expression is extensively regulated at the post-transcriptional level by small RNAs. These short (typically 50-300 nt) RNAs act by base-pairing with target mRNAs¹. In many cases, sRNAs anneal to ribosome-binding sites to repress translation or trigger mRNA decay by recruiting RNases^2,3. Although these interactions have been extensively studied using genetic and biochemical assays, most available approaches infer regulation indirectly from steady-state effects on RNA levels or protein output, making it difficult to resolve the underlying binding dynamics.

Efficient target regulation often requires the RNA chaperone Hfq, a small hexameric Sm-like protein^4,5. Hfq extends sRNAs half-lives by shielding them from nucleases⁶ and promotes annealing with target mRNAs^7,8. Hfq achieves these functions via three distinct RNA-binding surfaces with different preferences: the distal face recognizes AAN motifs usually found in mRNAs, the proximal face binds U-rich sequences typical of sRNA terminators, and the rim interacts electrostatically with RNA backbones⁹. Hfq, however, is present in much lower abundance than the total pool of interacting RNAs^10,11. Despite this apparent limitation, newly synthesized sRNAs rapidly find their targets and trigger regulation within minutes of transcription¹². This paradox suggests that RNA-Hfq interactions are highly dynamic, with RNAs continuously associating and dissociating rather than remaining bound in stable complexes^13,14. Understanding how such transient interactions produce reliable regulation requires experimental access to individual binding events rather than population averages.

Interactions between sRNAs and mRNAs are mediated by imperfectly complementary seed sequences, which must locate their matching sites within mRNAs that are often long and highly structured. In addition, the distance between the sRNA seed region and the Hfq-binding site on the mRNA can vary considerably¹⁵, further complicating the annealing process. Hfq helps overcome these challenges by flexibly transferring sRNAs between potential binding sites on the target, thereby increasing the likelihood of encountering the correct pairing region¹⁶. Hfq-mediated pairing can also remodel local mRNA structures, transiently unwinding base-paired regions and exposing otherwise inaccessible sequences. Such remodeling, however, can also lead to nonproductive or abortive annealing events. These transient interactions can sometimes be stabilized by strong Hfq-binding motifs located near the pairing site¹⁷. Despite this flexibility, not all sRNA-mRNA encounters result in stable duplex formation, and the factors that determine annealing efficiency remain poorly understood. In particular, it remains unclear how individual sequence determinants contribute quantitatively to binding lifetimes and complex stability.

Here, we established a minimal reconstituted single-molecule system that enables stepwise manipulation of sequence determinants and direct measurement of annealing kinetics using smFRET. In this system, we designed RNA pairs with initially inefficient annealing and systematically altered key features, including seed strength, Hfq-binding motif strength, and the spacing between these sites, to dissect how individual parameters influence duplex stability. By isolating defined sequence features in a controlled in vitro system, this approach enables the mechanistic dissection of sRNA targeting rules that are difficult to disentangle in vivo due to network complexity and indirect regulatory effects.

Protocol

This protocol describes the single-molecule FRET assay to monitor sRNA-mRNA annealing. It includes procedures for model RNA design, synthesis, slide passivation and functionalization, single-molecule imaging, and quantitative data analysis.

RNAs preparation
Design the sRNA and mRNA sequences in silico. Each sRNA should terminate with a transcription terminator and contain a U-rich Hfq-binding motif as well as a seed sequence complementary to the target mRNA. The corresponding mRNA should include AAN repeats serving as an Hfq-binding motif and a sequence complementary to the sRNA seed region. Sequences outside the Hfq-binding are randomized while avoiding AAN motifs and U-rich tracts to prevent unintended Hfq interactions. Add a tether-complementary sequence to the 3' end of the mRNA to enable immobilization on the surface during imaging. Predict RNA secondary structures excluding tether-complementary sequence using RNAstructure via the web server or locally¹⁸. For web-based analysis, upload FASTA sequences to the RNAstructure Web Server (https://rna.urmc.rochester.edu/RNAstructureWeb) and select Predict a Secondary Structure. Use default parameters (temperature 37 °C, no folding constraints). After computation, inspect the structure diagram and the probability of folding. Regions were classified as occluded if nucleotides within the seed or Hfq-binding motif had predicted base-pairing probabilities >60%. Sequences exceeding this threshold were redesigned. 

Prepare DNA templates for in vitro transcription by extending overlapping oligonucleotides using DNA polymerase, following the manufacturer's protocol. Primer concentration was 3 µM each. The thermocycler program was used as recommended by the manufacturer, except that the number of amplification cycles was reduced to 8, which was sufficient to extend the primers.

Transcribe RNA at 37 °C for 3 h to overnight using T7 RNA polymerase in 1× transcription buffer (80 mM Tris-HCl, pH 8.0, 2 mM spermidine, 10 mM NaCl), supplemented with 1 mM each NTP, 40 mM DTT, and 30 mM MgCl₂. Stop the reaction by adding EDTA to a final concentration of 30 mM. For RNA precipitation, add 1/10 volume of 3 M NaOAc (pH 5.4), followed by 2.5-3 volumes of ≥99% ethanol. Mix thoroughly and incubate at -80 °C for 1 h. Pellet RNA by centrifugation at ≥20,000 × g for 35 min at 4 °C. Remove the supernatant, wash the pellet with 70% ethanol, and air-dry briefly. Dissolve the RNA pellet in 20 µL of 8 M urea and heat at 90 °C for 2 min. Purify the RNA on 8% polyacrylamide gels containing 8 M urea. Excise the bands of interest and elute the RNA overnight in the elution buffer (0.3 M sodium acetate, pH 5.4, 1 mM EDTA). Recover the RNA by ethanol precipitation (follow instructions for DNA precipitation) and dissolve the pellet in nuclease-free water.

To prepare RNAs for fluorescent labeling, perform in vitro transcription as described, supplementing the reaction with 32 mM GMP to generate transcripts carrying a 5' monophosphate. Treat the RNA with 1-ethyl-3-[3-dimethylaminopropyl] carbodiimide hydrochloride¹⁹ (EDC) and imidazole to introduce an amine group at the 5' end. Incubate the sRNA in carbonate-bicarbonate buffer (pH 8.5) with either Alexa Fluor 555 or Cy3 NHS ester, following the manufacturer's instructions. Purify the modified RNA using chromatography spin columns, precipitate with ethanol, and dissolve the pellet in nuclease-free water. Determine labeling efficiency spectrophotometrically by measuring absorbance at 260 nm and at the dye-specific absorption maximum using a Nanodrop or equivalent spectrophotometer. Calculate RNA and dye concentrations using the Beer-Lambert law and the corresponding extinction coefficients provided by the fluorophore manufacturer. Labeling efficiency is defined as the molar ratio of dye to RNA and was routinely approaching 100%. Labeled RNAs can be stored at -20 °C.

Order DNA oligonucleotides carrying a 3′-biotin moiety and a 5′-amino C6 linker. Precipitate DNA using ethanol and sodium acetate as described above. Label the amine group with Cy5 NHS ester in carbonate-bicarbonate buffer (pH 8.5) according to the manufacturer's instructions. Determine the concentration of labeled DNA spectrophotometrically as described above. Labeled DNAs can be stored at -20 °C.

Hfq purification
Grow Escherichia coli BL21 (DE3) Δhfq::cat-sacB cells carrying the Hfq expression plasmid pET21b-EcHfq, in which the E. coli hfq gene is expressed from an IPTG-inducible promoter. Hfq is expressed without an affinity tag; the native protein contains sufficient surface-exposed histidine residues to allow purification by Ni²⁺ affinity chromatography. Grow cells in 1 L of LB-Miller medium supplemented with 100 µg/mL ampicillin at 37 °C to an OD₆₀₀ of 0.6. Induce Hfq expression by adding IPTG to a final concentration of 1 mM and incubate for 4 h at 37 °C. Harvest the cells by centrifugation at 5,000 × g for 10 min, then resuspend the pellet in 50 mL of lysis buffer (50 mM Tris-HCl, pH 8.0, 500 mM NH₄Cl, 250 mM MgCl₂, 1 mM 2-mercaptoethanol, Caution: 2-mercaptoethanol is toxic, handle in a fume hood). Lyse the cells by sonication on ice using a pulse setting of 50% duty cycle. Apply three 40-s pulses with 1-minute cooling intervals on ice between pulses. Centrifuge the lysate at 27,000 × g for 20 min at 4 °C. Heat the supernatant at 85 °C for 45 min, then centrifuge again under the same conditions. Treat the resulting supernatant with RNase A (30 µg/mL) and DNase I (5 U/mL) for 1 h at room temperature with gentle shaking, followed by filtration through a 0.45 µm filter. Apply the clarified lysate to an affinity chromatography column pre-charged with NiSO₄. Wash the column with buffer (50 mM Tris-HCl, pH 8.0, 500 mM NH₄Cl, 0.5 mM 2-mercaptoethanol). Elute the protein using the same buffer with a linear imidazole gradient concentration from 10 mM to 1 M. Pool the fractions containing Hfq and load them onto a HiLoad 16/600 Superdex 200 size-exclusion chromatography column equilibrated with HB buffer (50 mM Tris-HCl, pH 7.5, 250 mM NH₄Cl, 1 mM EDTA, 10% glycerol). Operate the column according to the manufacturer's instructions. A representative SDS-PAGE analysis of Hfq purification is shown in Supplementary Figure 1. Concentrate the purified protein, if required, using centrifugal filters (3 kDa MWCO). Purified Hfq aliquots can be stored at -80 °C.

Single-molecule experiments
Passivation of quartz slides²⁰
Prepare quartz slides with five parallel holes (0.8 mm diameter) drilled near each long edge. Place quartz slides and coverslips in a Hellendahl staining jar (slide holder) and sonicate in acetone for 20 min (Caution: acetone is flammable and volatile; work in a fume hood). Repeat the cleaning step using methanol (Caution: methanol is toxic and flammable; handle in a fume hood). Rinse the slides thoroughly with water and sonicate in 1 M KOH for 1 h (Caution: KOH is caustic, wear gloves and eye protection). Rinse the slides with water, then sonicate sequentially in acetone and methanol for 20 min each. Repeat the KOH cleaning using 5 M KOH for 1 h, followed by thorough water rinsing (Caution: Handle concentrated KOH with appropriate protective gear). Air-dry the slides. Prepare a clean set of Hellendahl staining jars and rinse them with hexane (Caution: hexane is highly flammable; handle in a fume hood), then allow them to air-dry. Place the slides in the slide holder and rinse the slides twice with hexane. Add 75 mL of hexane to the slide holder and carefully inject 50 µL of dichlorodimethylsilane (DMDCS) beneath the hexane surface (NOTE: avoid exposing DMDCS to air). Incubate the slides with gentle rocking for 1.5 h. Rinse and sonicate the slides in hexane for 1 min and repeat this step three times to remove residual DMDCS. Air-dry the slides and vacuum-seal them in 50 mL tubes (e.g., Falcon or conical tubes). Prepared slides can be stored at −20 °C until further use.

Assembly and functionalization of flow channels
A simple flow channel can be assembled using a quartz slide, glass coverslip, double-sided tape, pipette tips, and epoxy. Apply thin strips of double-sided tape between the holes on the quartz slide and place the coverslip on top. Seal the longer edges with epoxy. Cut approximately 5 mm segments from a pipette tip (both the narrow tip and the wider collar) and insert one into a hole to serve as an outlet and the other into a parallel hole to form an inlet reservoir. Secure both with epoxy. The inlet reservoir should retain liquid without leaking prior to injection. To draw solutions through the channel, connect tubing from a 1-mL syringe to the outlet.

Pretreat the channels sequentially with the following reagents to functionalize the surface: biotinylated BSA (0.2 mg/mL) for 5 min, Tween-20 (0.2%) for 10 min, and NeutrAvidin (0.1 mg/mL) for 1 min. Approximately 100 µL of each solution is sufficient per channel. Avoid pulling more liquid than is present in the inlet reservoir, as this will introduce air into the channel and generate bubbles. Leave a small volume of solution in the reservoir between steps to prevent the channel from drying. Rinse thoroughly with 1× TNK buffer (10 mM Tris-HCl, pH 7.5, 50 mM NaCl, 50 mM KCl) between each step. This treatment establishes a biotin-NeutrAvidin surface for RNA immobilization.

Prepare mRNA-tether duplexes immediately before immobilization by mixing 40 nM mRNA with 20 nM biotinylated Cy5-labeled DNA tether. Denature the mixture at 75 °C for 5 min in 1× TNK buffer (10 mM Tris-HCl, pH 7.5, 50 mM NaCl, 50 mM KCl), then refold at 37 °C for 15 min and equilibrate at 20 °C.

Dilute the sample 50-fold in imaging buffer (10 mM Tris-HCl, pH 7.5, 50 mM NaCl, 50 mM KCl, 4 mM 6-hydroxy-2,5,7,8-tetramethylchroman-2-carboxylic acid, 0.01% octaethylene glycol monododecyl ether, 0.8% glucose, 2 U/mL RNase inhibitor) and inject it into the flow channel to immobilize the mRNAs. Incubate for 1 min, then wash with imaging buffer supplemented with an oxygen scavenging system (165 U/mL glucose oxidase and 2170 U/mL catalase) to reduce photobleaching.

Prepare sRNA-Hfq complexes by mixing the components in a 1:1 ratio to a final 250 nM concentration in 1× TNK buffer and incubate for 5-15 min at room temperature. Dilute the complexes 50-fold in the imaging buffer containing the oxygen scavenging system and inject them into the flow channel immediately before imaging.

Single-molecule data acquisition
Set up the prism-based total internal reflection fluorescence (TIRF) microscope by powering on the EMCCD camera and lasers. Mount the prepared slide with assembled flow channels on the microscope stage and place the prism on top. Use Single software (https://github.com/Ha-SingleMoleculeLab) to acquire data. In the software, determine the background level by selecting AutoScale and copy the obtained value into the Background field. Set the acquisition frame rate to 100 ms. Excite Alexa Fluor 555 or Cy3 using a 532 nm laser and Cy5 using a 633 nm laser. Collect emissions from both channels simultaneously through a 60× water-immersion objective. For each movie, begin with ten frames of 633 nm excitation to localize Cy5-labeled mRNAs in the field of view. Continue recording for 5 min with 532 nm excitation. End the acquisition with 1 s of 633 nm excitation to verify Cy5 photobleaching. This will produce the .pma movie file, which can be used for analysis.

Single-molecule data analysis
Analyze raw .pma files using custom IDL scripts to perform donor-acceptor channel mapping and extract fluorescence intensity trajectories. First, record a calibration movie using multicolor fluorescent beads detectable in both donor and acceptor channels upon excitation with either the 532 nm or 633 nm laser. Use this calibration movie to generate a mapping file that aligns donor and acceptor channels. Apply the resulting mapping file to all experimental recordings to identify corresponding donor-acceptor pairs, detect fluorescent spots, and export intensity time traces for downstream FRET analysis. Detailed step-by-step instructions are provided together with the script (https://github.com/Ha-SingleMoleculeLab).

Process intensity traces in MATLAB using the script s_tr_E.m. Launch the script and enter the directory containing the .traces and .pks files, followed by the file index and the time resolution. Select the frame window used for spot identification when prompted. The script then displays donor and acceptor intensities per single spot. Navigate through molecules using keyboard commands (b to go back, g to jump to a specified molecule). Define donor and acceptor baselines (k) and correct leakage (l) by clicking on the plots when needed. Save accepted traces individually as .dat files containing time, donor, and acceptor intensities by pressing s. Use the exported files for downstream quantitative analysis.

Use the peak-finder program to analyze individual smFRET traces and extract binding events for quantification, including dwell-time analysis and FRET state distributions. Launch the program and open the folder containing the exported .dat trace files using the File menu. Set the baseline values and donor-to-acceptor leakage values determined previously using the MATLAB script. Next, set the intensity thresholds to guide automatic peak detection such that all visually verified binding events lie above the selected values. If noise causes a single event to be split into multiple peaks, manually merge them using the built-in merge function. Click Save peaks to export detected events as .csv files containing start time, end time, and mean FRET efficiency for each event. Output files are saved in the peaks directory, while frame-resolved FRET values for each event are written to the peak-fret directory. FRET efficiency is calculated by dividing the acceptor fluorescence intensity by the sum of donor and acceptor intensities, using values corrected for background and donor leakage into the acceptor channel. Inspect traces manually for Cy5 photobleaching, identified by an abrupt decrease in FRET efficiency accompanied by an increase in Cy3 intensity. When photobleaching occurs, assume binding persists as long as Cy3 fluorescence remains detectable, and correct peak durations by editing the corresponding .csv files.

Import frame-resolved FRET efficiency values from the peak-fret files into GraphPad Prism. Generate FRET histograms using the Frequency Distribution function and fit the distributions with Gaussian models to identify the dominant FRET population.

Determine dwell times for individual binding events by subtracting association from dissociation time points using a spreadsheet. Calculate the fraction of transient events (<10 s) for each experimental repeat. Statistical analyses were performed in GraphPad Prism. Comparisons between two conditions were evaluated using unpaired t-tests, and comparisons among three or more conditions were analyzed using one-way ANOVA with Tukey's post-hoc test.

Results

Seed length sets the dynamic regime of sRNA-mRNA pairing
We established a minimal single-molecule assay to test how sequence features govern Hfq-mediated annealing. Model RNAs were designed following previously described architectures of Hfq-dependent sRNA^-mRNA pairs^16,17 and carried an adenosine-rich Hfq-binding motif on the mRNA and a uridine-rich rim-binding site on the sRNA (Figure 1A). The sRNA was labeled with Alexa Fluor 555 (donor) at the 5ʹ end and the mRNA-DNA tether with Cy5 (acceptor), allowing detection of base-pairing-induced proximity through FRET.

Varying the seed complementarity from 4, 5, 8, to 10 bp produced distinct dynamic behaviors. With a 4-bp seed, interactions were predominantly transient and exhibited heterogeneous FRET efficiencies, suggesting incomplete base-pairing (Figure 1B-D). In contrast, 8-bp and 10-bp seeds yielded stable duplexes that often persisted through the observation window (Figure 1C). The interactions produced the expected high FRET state (E_FRET ^≈ 0.79-0.80, Figure 1D), consistent with full base pairing between RNAs. A 5-bp seed produced short-lived yet well-defined binding events (E_FRET ^≈ 0.66), providing a workable dynamic range for subsequent modifications in RNA sequence (Figure 1B-D). We therefore used 5-bp complementarity for all downstream tests.

Stronger Hfq-binding motifs increase the stability of the sRNA-mRNA complex
We next asked how Hfq-binding strength modulates binding lifetimes at fixed 5-bp complementarity. Extending the mRNA motif from AAN4 to AAN6 significantly increased the probability of forming stable complexes (p < 0.001). Mean lifetimes rose from 7.5 s to 18 s, and the fraction of unstable events < 10 s decreased from 80 ± 2% to 68.3 ± 2.9% (Figure 2A-C).

We then tested the effect of the Hfq-binding motif on the sRNA. Shortening the rim-binding site from UUAUUUUUUU to CUUC (sRNA-SR, "short rim") produced highly unstable interactions (mean lifetime 1.9 s with AAN4), and strengthening the mRNA motif in that context had only a modest effect (mean 3.0 s with AAN6, not significant, Figure 2B,C). Thus, both partners contribute to the stability of the produced complex, but weak sRNA-Hfq contacts limit the gains from a stronger mRNA motif.

A spacer between Hfq- and seed sites on the mRNA promotes longer-lived complexes
Hfq can remain bound to the AAN region while using its rim surface to facilitate the search for optimal sRNA pairing site¹⁶. This property suggested that additional spacing between the two binding regions on the mRNA might influence the interaction stability.

To test this, we introduced a 15-nt single-stranded spacer between the Hfq- and sRNA-binding sites (N15). This modification significantly increased the lifetime of stable annealing events from 7.5 s (AAN4) to 47 s (AAN4-N15) (p < 0.01) (Figure 3A,B). The percentage of binding events lasting less than 10 s also decreased from 80 ± 2% to 44 ± 6% (Figure 3C). These data indicate that added flexibility between binding modules can stabilize annealing intermediates and promote longer-lived duplexes.

DATA AVAILABILITY:
All data supporting the findings of this study, including single-molecule FRET trajectories, processed datasets, and analysis scripts, are publicly available in the Mendeley Data repository at: doi:10.17632/3z8fng2c8g.1

Figure 1: Effect of sRNA-mRNA complementarity on binding efficiency. (A) smFRET assay for sRNA-mRNA annealing. Base pairing between the Alexa Fluor 555-labeled sRNA and the immobilized mRNA-Cy5-DNA tether complex produces a high FRET signal. Model mRNAs contain an Hfq-binding site (AAC)₄ (blue) and an sRNA-binding site (S5, red), followed by five nucleotides and the RNA-DNA hybrid region. sRNAs include a rim-binding site (orange). Panels in (B-D) are grouped by experimental condition (seed length).(B) Representative sRNA binding event on a single mRNA molecule. Fluorescence intensity traces of Alexa Fluor 555-labeled sRNA (donor, green) and Cy5-DNA tether (acceptor, red) are shown. Cy5 was directly excited during the first and last ten frames of the recording. (C) Rastergrams depicting 50 representative mRNA molecules bound by sRNA-Hfq. Each horizontal bar represents a single sRNA-mRNA interaction. (D) Histograms of FRET efficiencies for interactions between mRNAs and sRNAs containing 4 bp (N = 416 mRNAs), 5 bp (N = 513 mRNAs), 8 bp (N = 191 mRNAs), or 10 bp (N = 105 mRNAs) complementary regions. FRET efficiencies were calculated from baseline- and leakage-corrected donor and acceptor intensities for individual binding events in 0.1 s intervals. Gaussian fits (blue) are shown to indicate the dominant FRET population and were not used to calculate E_FRET values. Please click here to view a larger version of this figure.

Figure 2: Effect of Hfq-binding motif length on sRNA-mRNA interactions. The base-pairing region in all variants is 5 bp long. (A) Rastergrams showing 50 representative mRNAs bound by sRNA-Hfq complexes. Each horizontal bar represents a single interaction. Cartoon diagrams above illustrate the corresponding RNA designs with changes in the number of (AAC) repeats in the mRNA and the length of the rim-binding region in the sRNA (SR = "short rim"). Coloring as in Figure 1A. (B) Scatter plots of sRNA-mRNA binding lifetimes. Each dot represents a single binding event.Mean lifetimes (red horizontal line): 7.5 s for mRNA-AAN4 + sRNA (N = 188 mRNAs), 18 s for mRNA-AAN6 + sRNA (N = 142 mRNAs),1.9 s for mRNA-AAN4 + sRNA-SR (N = 85 mRNAs), and 3 s for mRNA-AAN6 + sRNA-SR (N = 149 mRNAs). (C) Bar graph showing the percentage of binding events shorter than 10 s. Error bars represent the standard deviation from two independent datasets. Statistical analysis by one-way ANOVA with Tukey's post-hoc test for multiple comparisons. ***p < 0.001, ****p < 0.0001; ns, not significant. Please click here to view a larger version of this figure.

Figure 3: Effect of spacer length between Hfq- and sRNA-binding sites on sRNA-mRNA interactions. The base-pairing region in all variants is 5 bp long. (A) Rastergrams depicting 50 representative mRNAs bound by sRNA-Hfq complex. Each horizontal bar represents a single binding event. The cartoon diagrams above illustrate the corresponding RNA design. N15 = 15 nt spacer of random sequence. Coloring as in Figure 1A. (B) Scatter plots of sRNA-mRNA binding lifetimes. Each dot represents a single binding event. Mean lifetimes (red horizontal line): 7.5 s for mRNA-AAN4 + sRNA (N = 188 mRNAs) and 47 s for mRNA-AAN4-N15 + sRNA (N = 289 mRNAs). (C) Bar graph showing the percentage of binding events shorter than 10 s. Error bars represent the standard deviation from two independent datasets. Statistical analysis by unpaired t-test. **p < 0.01. Please click here to view a larger version of this figure.

Supplementary Figure 1:Purification of E. coli Hfq. SDS-PAGE analysis of Hfq expression and purification. (Top) Samples collected during purification on a Ni²⁺ affinity (HiTrap) column, including before induction, after induction, pellet, lysate, load, flow-through, wash fractions, and eluted peak fractions. (Bottom) SDS-PAGE of peak fractions after size-exclusion chromatography on a HiLoad 16/600 Superdex 200 column. Monomeric (~11 kDa) and hexameric (~66 kDa) forms of Hfq are indicated. Please click here to download this File.

Discussion

This study establishes a minimal single-molecule assay to dissect how specific RNA sequence features determine the efficiency of Hfq-mediated sRNA-mRNA annealing. By varying seed length, Hfq-binding motif strength, and spacing between binding modules, we directly quantified how each parameter contributes to duplex stability at single-molecule resolution. Unlike ensemble measurements, which report population-averaged effects, this approach directly resolves individual binding and dissociation events, enabling measurement of kinetic heterogeneity and transient interactions. Our results show that even short seed regions can engage transiently, whereas longer seeds produce stable duplexes with extended lifetimes. These findings are consistent with earlier in vivo and ensemble data demonstrating that natural seed regions in bacterial sRNAs typically span 6-12 nucleotides and rarely form fully complementary duplexes^21,22,23.

We also found that both the mRNA and sRNA Hfq-binding motifs strongly influence duplex stability. Extending the AAN motif on the mRNA increased binding lifetimes, while shortening the sRNA rim-binding site markedly destabilized them. This supports a cooperative model in which Hfq's distal face anchors adenosine-rich mRNA motifs, and the rim recruits uridine-rich sRNAs to promote efficient alignment of complementary regions^9,17. In contrast to genetic or reporter-based assays, which typically detect only productive regulatory outcomes, this system quantitatively separates binding stability from downstream effects, allowing direct comparison of how individual sequence features modulate physical interaction lifetimes. A strong rim-binding motif is particularly important when structural rearrangements are required for sRNA-mRNA pairing, as it enhances RNA retention on Hfq¹⁷. The rim is also thought to promote lateral scanning of mRNAs by sRNA-Hfq complexes¹⁶, a property that may be particularly important when target sites are distant from the Hfq-binding region. Our results with a 15-nt spacer between the binding sites support this view, showing that additional flexibility between the seed and Hfq motif enhances pairing stability, possibly by reducing steric constraints during strand exchange.

Beyond natural systems, the results are relevant to the design of synthetic or engineered sRNAs. Most existing design strategies focus on optimizing a seed region or a general scaffold^23,24,25, but our data highlight additional determinants such as the spacing between the Hfq-binding motif and the target region. Because sequence features are introduced in a controlled and modular manner, the assay provides a systematic experimental framework for testing design rules. This principle could guide future construction of artificial sRNAs with improved regulatory efficiency.

While the approach enables quantitative and reproducible analysis of RNA-RNA interactions, it also has limitations. The assay investigates fully complementary regions, whereas most physiological sRNA-mRNA interactions contain mismatches and bulges that can affect recognition and regulation. The method measures sRNA-mRNA binding lifetimes but does not directly address downstream regulatory outcomes, such as translation inhibition or mRNA decay. Future adaptations of this assay could incorporate mismatched seed regions, varied GC content, or ribosomes and degradosomes to systematically probe how these factors shape binding kinetics.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
0.45 μm filter	Millipore	HABG04700
5 mL Hi-Trap column	Cytiva	GE17-0408-01
Acetone	Merck	270725	HPLC grade
Acrylamide:Bis-acrylamide solution	Merck	A2792-100ML
ALEXA555 NHS succinimidyl ester	Thermo Fischer	A20009
Amicon Ultra Centrifugal Filters (NMWL, 3 kDa)	Merck	UFC900308
Ammonium chloride	Merck	1330-500G
Ampicylin	Merck	A9518-5G
ATP	Linegal Chemicals	K054.1
Beta-mercaptoethanol	Merck	M3148-500ML
BIO-BSA	Merck	A8549-10MG
Bio-Spin Columns with Bio-Gel P-30	Bio Rad	7326231	For purifing labeled RNA
Catalase	Merck	C9322
CTP	Linegal Chemicals	K057.4
Cy3 HNS ester	Cytiva	PA13105
Cy5 NHS succinimidyl ester	Cytiva	PA15101
DDT	Thermo Fischer	R0861
Dichlorodimethylsilane	Merck	440272
DNA oligomers (tether) a 5′-amino linker C6	Thermo scientific	n/a
Dnase I	Thermo Scientific	18047019
EDC	Thermo scientific	PG82079
EDTA	Merck	ED2SS-1KG
EMCCD camera	Andor
Ethanol	Merck	1085430250
Glucose	Merck	D9434-1KG
Glucose oxidase	Merck	G2133-50KU
Glycerol	Merck	G5516-1L
GMP	Merck	G8377
GTP	Linegal Chemicals	K056.4
Hcl	Merck	1.00318
Hexane	Thermo Fischer	045652.M1
HiLoad 16/60 Superdex 200 size exclusion column	Cytiva	GE28-9893-35
IPTG	Merck	420322-1GM
KCl	Merck	P9541-1KG
KOH	Merck	484016-1KG
Magensium chloride	Merck	M8266-1KG
Methanol	Merck	34885-M	HPLC grade
Neutravidin	Thermo Fischer	31000
Nickel sulfate	Merck	1067261000
Nikkol	Merck	P8925-1G
Q5 Polymerase	NEB	M0491S	Used for production of in vitro transcription template
RNase A	Thermo scientific	EN0531
RNasin Plus	Promega	N2611
Sodium Acetate	Merck	S7545-100G
Sodium chloride	Sigma-Aldrich	S3014-1KG
Spermidine	Merck Life Science	S2626
Tris Base	Thermo Scientific	17926
Trolox	Merck	238813-1G
Urea	Merck	U1250-5KG
UTP	Linegal Chemicals	K055.3