Extended Data Fig. 4: Biophysical and biochemical characterization of the constructs.

a, Individual A₂₆₀ UV melting profiles for the constructs used here. The miR-34a–mSirt1 duplex, miR-34a–mSirt1 trapped ES duplex, miR-34a–complementary-strand duplex and miR-34a single-stranded RNA (ssRNA) were each measured as three technical independent replicates (shown in different colours; n = 1). Individual technical replicates are plotted. T_m values are shown as means ± s.d. of fitted T_m values in individual technical replicates (n = 3). The other ssRNAs (bottom row) were measured and plotted as individual technical replicates; fitted T_m values are shown with associated confidence intervals of 95% (n = 1) as an estimate of the experimental error. Normalized differential melting curves (δA₂₆₀/δT) are plotted as a function of temperature (in K) (circles) and fitted to Supplementary equation (1a) or (1b) (curves), depending on the molecularity of the system. b, EMSA titration profiles for the miR-34a–mSirt1 duplex, miR-34a–mSirt1 trapped ES duplex and miR-34a–complementary-strand duplex, measured as three independent technical replicates. The ratio of bound to total miR-34a 3′-Cy3 is plotted as a function of titrand concentration (circles) and fits a standard binding isotherm (line) (Supplementary equation (2)). The plot centre is the mean; error bars represent 1 s.d. from the three independent replicates. Fitted K_d values along with confidence intervals of 95% are shown as an estimate of the experimental error (n = 3). Gel images were acquired by detection of Cy3 fluorescence. During the titration, miR-34a 3′-Cy3 was kept at a constant concentration of 24 nM, setting the sensitivity limit for estimating K_d (Supplementary Fig. 1a–c). mSirt1 and its trapped-ES counterpart are equivalent in their ability to form a stable RNA–RNA duplex with miR-34a. Tighter binding is observed for the complementary strand (48.4 ± 9.5 nM) than for the mSirt1 (124.3 ± 21.7 nM) and trapped-ES mSirt1 (110.3 ± 23.0 nM), providing a control for the dynamic range of K_d estimation. c, Equilibrium FBA profiles for mSirt1, mSirt1 trapped ES and a scrambled control, binding to miR-34a-loaded Ago2. The three targets were each measured as three independent replicates and fitted to a standard binding isotherm (line) (Supplementary equation (2)). The plot centre is the mean; error bars represent 1 s.d. from three independent replicates. Fitted K_d values are shown with confidence intervals of 95% (an estimate of the experimental error). As in c, mSirt1 and mSirt1 trapped ES are equivalent in their ability to form a stable ternary complex within RISC. The simulated data set (dotted lines) indicate curves corresponding to K_d values ten times lower (red) or ten times higher (green) than the average value for mSirt1 and mSirt1 trapped ES, providing a frame for the amplitude of our experimental error. d, Top, northern blot showing the detection of miR-34a loaded in Ago2. Bottom, a standard calibration curve (using naked miR-34a), used to obtain an estimate of miR-34a in RISC. The centre calibration curve was used to calculate R². The two outer curves indicate the 95% confidence interval of the calibration-line fit (from a single repeated experiment). The average ratio of Ago2 and miR-34a-loaded Ago2 (both in pmole) was used to obtain the fraction of Ago2 loaded with our guide (roughly 1.5%). The complete lists of fitted parameters for UV melting, EMSA titration, FBA titration and northern blot are in Supplementary Table 1a–d. The complete fitting analyses of UV melting, EMSA titration and FBA titration are in Supplementary Tables 7–9.