Helix-7 in Argonaute2 shapes the microRNA seed region for rapid target recognition

Abstract

Argonaute proteins use microRNAs (miRNAs) to identify mRNAs targeted for post-transcriptional repression. Biochemical assays have demonstrated that Argonaute functions by modulating the binding properties of its miRNA guide so that pairing to the seed region is exquisitely fast and accurate. However, the mechanisms used by Argonaute to reshape the binding properties of its small RNA guide remain poorly understood. Here, we identify a structural element, α-helix-7, in human Argonaute2 (Ago2) that is required for speed and fidelity in binding target RNAs. Biochemical, structural, and single-molecule data indicate that helix-7 acts as a molecular wedge that pivots to enforce rapid making and breaking of miRNA:target base pairs in the 3' half of the seed region. These activities allow Ago2 to rapidly dismiss off-targets and dynamically search for seed-matched sites at a rate approaching the limit of diffusion.

Keywords: Argonaute; RNA silencing; microRNA; seed; target search.

Figure 1. Helix‐7 dynamically interacts with the miRNA seed region

1. Cartoon representation of the Ago2–miRNA crystal structure. Ago2 is colored gray except for helix‐7, which is yellow. miRNA guide colored red.

2. Crystal structure of the Ago2–miRNA–target RNA ternary complex. Target RNA is colored blue.

3. Close‐up view of the unpaired complex shows helix‐7 breaks nucleobase stacking in the miRNA seed region by intercalating between g6 and g7.

4. Close‐up view of the Ago2–miRNA–target complex shows helix‐7 docks into the minor groove of the guide:target duplex, directly contacting base pairs at positions g6 and g7.

Figure EV1. ∆Helix‐7 Ago2 adopts a near‐native fold

1. SDS–PAGE of purified wild‐type and Δhelix‐7 Ago2 samples. The Δhelix‐7 mutant was indistinguishable from wild type during recombinant expression and purification.

2. Cartoon schematic of heat inactivation experiment. Ago2–miR122 complexes were heated to various temperatures for 10 min, then cooled on ice, and then measured for endonuclease activity at 37°C.

3. Denaturing PAGE of a miR‐122 target cleaved by wild‐type and ∆helix‐7 Ago2–miR122 complexes heated to indicated temperatures.

4. UV CD spectra of wild type (dashed black line) and ∆helix‐7 (blue line) at 25°C are nearly superimposable.

Figure 2. Helix‐7 is not required for target recognition

1. Plot of g2‐g8 target RNA bound (0.1 nM) versus Ago2–miR122 concentration for wild‐type (WT), helix‐7 double point mutant (MI‐AA), and Δhelix‐7 Ago2. Average values from at least three independent experiments ± standard deviation (SD) are plotted. Top panel shows target RNA paired to seed region of the guide RNA (miRNA‐122).

2. Dissociation of a ³²P‐labeled target RNA (0.1 nM) from the Ago2–miR122 complex (1 nM) was monitored in the presence of unlabeled target RNA (100 nM). Fraction of the target RNA bound to Ago2–miR122 is plotted as a function of time for WT and Δhelix‐7 Ago2. Average values from at least three independent experiments ± SD were fit to single‐exponential decays.

Figure 3. Helix‐7 minimizes dwell time on off‐targets

1. Cartoon schematic of our single‐molecule FRET assay.

2. Guide:off‐target pairs examined listed with increasing values of N (the length of the matched region between miRNA and target RNA). N = 7 is the fully seed‐paired target.

3. Representative fluorescence time traces (with a resolution of 100 ms for wt, 300 ms for MI‐AA and Δhelix‐7) show docking and dissociation of wild‐type (top), MI‐AA (middle), and Δhelix‐7 (bottom) Ago2–miRNA at single spots in the microfluidic chamber.

4. Dwell time histograms obtained for N = 4, 5, and 6, with wild‐type Ago2; histograms were fit with single‐exponential decay. The first column of the data was not included in the analysis to avoid potential artifacts arising from the limit of the time resolution. The bin size is 0.5 s for N = 4 and 1 s for N = 5 and 6.

5. Bar plot showing the dependence of dwell time on N for wild‐type Ago2.

6. Dwell time histogram for different N's with the MI‐AA helix‐7 mutant. The dwell time histogram for N = 6 fit with double‐exponential decay (green; R ² = 0.98); the dwell time for N = 6 with single exponential fit was 15.2 ± 3.41 s, R ² = 0.96.

7. Bar plot for dwell times of MI‐AA Ago2 as a function of N.

8. Dwell time histogram for different N's with the ∆helix‐7 mutant. The dwell time histograms for N = 5 and N = 6 were fit with double‐exponential decays (green; R ² = 0.97 for N = 5 and R ² = 0.96 for N = 6); the dwell time for N = 5 and N = 6 with single exponential fit was 9.1 ± 0.91 s (R ² = 0.86) and 22.3 ± 5.08 s (R ² = 0.90), respectively.

9. Bar plot for dwell times of ∆helix‐7 Ago2 as a function of N.

Data information: For all bar plots (E, G, and I), error bars are the SD of three independent experiments carried out on different days. The dashed red lines indicate the observation time limit (300 s), which is constrained by photobleaching.

Figure EV2. Dwell times of a spectrum of off‐target sequences

1. Sequences of guide RNA (green) paired to mismatched target RNAs (red).

2. Dwell time histograms for wild‐type and ∆helix‐7 Ago2 on the off‐target RNAs.

3. Bar plot of average dwell times for indicated off‐target RNAs.

Figure 4. Helix‐7 discourages non‐Watson‐Crick pairing at g6 and g7

1. A–C

   Plots of target RNA (0.1 nM) bound to Ago2 versus Ago2–miR122 (wild type and Δhelix‐7) concentration for targets with an increasing number of G:U wobble pairs in the seed 3′ end.

2. D

   Plot of target RNA (0.1 nM) bound to Ago2 versus Ago2–miR122 concentration for a target with mismatches at positions g6–g8.

3. E

   Bar plots of the K _d of wild‐type (left) and Δhelix‐7 (right) Ago2–miR122 for indicated target RNAs. The dashed red lines indicate the upper limit (100 nM) for which a dissociation constant can be determined under our experimental regime.

Data information: Average values from at least three independent experiments ± SD are plotted.

Figure 5. ΔHelix‐7 Ago2 is sensitive to inhibition by off‐target RNAs

1. Ago2–miRNA122 (1 nM) was incubated with a ³²P‐labeled, seed‐matched target RNA (0.1 nM) in the presence of increasing concentrations of unlabeled a competitor RNA (top panel). Fraction target RNA bound to Ago2 is plotted as a function of off‐target competitor RNA concentration.

2. Fraction of target RNA bound to Ago2 in the presence of increasing concentrations of unlabeled total cellular RNA.

Data information. Average values from at least three independent experiments ± SD are plotted.

Figure 6. Helix‐7 mutants display reduced target binding rates

1. Ago2–miR122 complexes were mixed with a ³²P‐labeled seed‐matched target RNA; bound and free RNAs were then separated using a filter‐binding apparatus at various times. Representative time course for target RNA (0.1 nM) binding to wild‐type, MI‐AA, and Δhelix‐7 Ago2 (1 nM). Average values from at least three independent experiments ± SD are plotted.

2. Observed binding rates (k _on,obs) plotted as a function of Ago2–miR122 concentration. Average values ± SD are plotted.

3. Arrival time distribution of single Ago2–miRNA complexes binding immobilized targets in a microfluidic chamber. The value is the average of three independent measurements.

Figure EV3. Helix‐7 facilitates lateral diffusion

1. Schematic of target RNA containing two seed‐matched sites separated by a 15 nt linker. Guide RNA (green) is shown paired to site 1.

2. Representative fluorescence time traces and FRET efficiencies for wild‐type (top) and Δhelix‐7 (bottom) Ago2. FRET changes arise from a single Ago–miRNA complex shuttling between the two adjacent binding sites (Chandradoss et al, 2015).

3. Bar graph comparing wild‐type and Δhelix‐7 Ago2 shuttling rates. k _{shuttling,(obs)} = 0.28 and 0.006/s for wild‐type and ∆helix‐7 Ago2, respectively. For wild‐type Ago2, 86–90% of each stable binding event included at least one shuttle between site 1 and site 2. For ∆helix‐7 Ago2, only 13–16℅ of binding events displayed any shuttling. The error bars are the SD of three independent experiments.

4. Dwell time histogram for high FRET and low FRET states for wild‐type (top) and ∆helix‐7 (bottom) Ago2.

Figure 7. Structure of a helix‐7 Ago2 mutant

1. Superposition of wild‐type (white) and MI‐AA (gray) Ago2. Helix‐7 colored yellow; ordered region of the guide RNAs colored in red.

2. Close‐up view shows helix‐7 tilts away from the seed region in the MI‐AA mutant.

3. Side‐by‐side comparison of the seed region in wild‐type (left) and MI‐AA (right) Ago2 crystal structures.