Single-step discovery of high-affinity RNA ligands by UltraSelex

a, Conventional SELEX scheme. The input dsDNA pool (round 0) is in vitro transcribed into RNA to yield the initial RNA pool, which is incubated with the immobilized target. After removal of unbound RNA by washing, the bound fraction is eluted, reverse-transcribed into cDNA and PCR-amplified. The obtained enriched dsDNA pool is used as input for the next iteration, and this cycle is repeated until binders dominate. Binders are then identified, based on their abundance, by sequencing. b, Design of the partially structured RNA pool used in all UltraSelex experiments reported here, and stepwise processing to yield the barcoded sequencing library. The RNA pool contained two constant primer-binding regions (purple and blue)¹⁷, two randomized stretches of 26 nt each (N26) and a constant 12 nt internal hairpin (H12) thought to increase the abundance of high-affinity aptamers in the starting pool¹⁶. An aliquot of the same RNA pool was used previously for the isolation of SiR-binding aptamers by conventional SELEX¹⁸. The offset PCR system was designed to reduce bias in the sequencing of templates with variable and constant regions, thereby achieving balanced per-base nucleotide compositions⁴⁶. Equimolar amounts of seven forward primers (Frw) and seven reverse primers (Rev) containing 0–6 inserted offset nucleotides (golden) were mixed and used for PCR amplification, generating length-heterogenous sequencing libraries. After offset PCR with the primer mixture, the Illumina primers, indices and barcodes were attached in another PCR reaction (barcode PCR). Before sequencing, a custom balancer oligonucleotide was added. For sequences, see Supplementary Table 2. c, Detailed architecture of the complete sequencing template.

a, Number of HTS reads per RNA species in the UltraSelex input library. Three independent replicates. b, Decreasing sequence diversity with increasing group number e1–11 in the SiR-B UltraSelex library. The x axis indicates the log₂ of the number of HTS reads per species, while the y axis denotes the number of RNA species with the specified abundance. The partitioning eluate (e) number is individually denoted. c, Percentage of RNA species with ≥2 HTS reads in the different partitioning groups (e1–11) of UltraSelex library SiR-B. The dashed gray line represents 1% as a reference. d, Formula and simulation of γ functions, assigning γ_i values to different partitioning groups g_i for i ≥ 2. c = 0.5, eluate_nums, number of partitioning eluates (here 11). Colors in the line plot represent different exponents θ from 0.1 to 10. e, Illustration of the SGREELI calculation protocol, comprising seven steps. After sequencing and quality control, fold-change (FC) values were calculated for each eluate (e2–11 for FC2–FC11) by normalizing read counts to their abundance in eluate e1. An RNA species that ranks at exactly the top 1% by FC value in each eluate is defined as the reference RNA (FC_ref), with this step performed for each eluate group (ref2–ref11). The fold-change values of these reference RNAs are log₂-transformed (log₂ FC_ref). Each eluate group is then assigned a specific γ value, here based on a linear function with exponent θ = 1, reflecting the expected elution patterns of strong binders in later fractions. The log₂ FC_ref value is then divided by γ to derive a normalization factor for each eluate group. All RNA species in a given eluate group have their FC values log₂ transformed and adjusted by this normalization factor, yielding intra- and inter-group adjusted fold-change (aFC) values. Finally, the area under the curve (AUC) is computed by integration of the aFC values for each species, using the corresponding γ values as the x-axis parameters.

Source data

a, Spike-in SiR binder RNA and non-binding RNA were enriched differently under various washing conditions. Sample preparation followed Fig. 2b. Washing condition: ‘1× ASB’ (20 mM HEPES (pH 7.4), 125 mM KCl and 5 mM MgCl₂), ‘HEPES only’ (20 mM HEPES (pH 7.4)); and ‘HEPES + EDTA’ (20 mM HEPES (pH 7.4) and 5 mM EDTA¹⁸). Partitioning included 14 wash eluates (e1–14) and a final release eluate (e15). RT–qPCR signals were normalized to initial abundance prior to selection. Continuous bands denote standard deviation, with the mean at the center (n = 3). b, Box plot showing the abundance of previously SELEX-identified aptamers in UltraSelex libraries. The x axis indicates partitioning groups, while the y axis denotes species abundance in reads per million qualified reads (RPM). Each colored dot or triangle represents a previously identified aptamer: red dots indicate RNA species with higher abundance than SiR1, while gray triangles represent those lower or equal to SiR1. The percentage of gray dots in each eluate is shown above. SiR-A (n = 39), SiR-B (n = 37) and SiR-C (n = 38). The box plots are defined as in Fig. 2f. c, Abundance of three known SiR-binding aptamers in UltraSelex sequence data from SiR-A and SiR-C, as in Fig. 2d. d, Plot of the aFC values of the same three aptamers, as in Fig. 2c. e, Rank distribution of turn-on ('On') and non-turn-on ('Non') SELEX RNA aptamers in UltraSelex libraries. The y axis represents aptamer rank in each library. Turn-on and non-turn-on groups are shown in purple and gray, respectively. SiR-A (n = 39), SiR-B (n = 37) and SiR-C (n = 38). Box plot definition follows b. f, Rank analysis of SiR-A and SiR-C UltraSelex and SELEX libraries, following Fig. 2f. SiR-A: e1–11 and AUC (n = 39), On (n = 19), Non (n = 20). SiR-C: e1–11 and AUC (n = 38), On (n = 19), Non (n = 19). SELEX: r4–14 and AUC (n = 39), On (n = 19), Non (n = 20). The box plots are defined as in Fig. 2f.

Source data

a, Fluorescence turn-on of a silicon rhodamine dye by the 25 top-ranked RNA aptamers from UltraSelex SiR-A (left) and SiR-C (right). Error bars represent the mean ± SD from technical triplicates. All other details are as in Fig. 3d. b, The measured turn-on fluorescence and K_D values of six UltraSelex SiR-B RNA aptamers with binding affinities in the nanomolar range. c, Quantification of the SiR fluorescence intensity of 50 aptamer-transfected HEK-239T cells with plasmids encoding self-circularizing (Tornado) overexpressed RNA constructs of SiR-B7 and SiR-B15 aptamers and SELEX-derived SiR-binding aptamer¹⁸ in comparison with a SiR-unrelated insert⁴⁷. The confocal imaging procedure was similar to that in Fig. 3f, except the cells were incubated with 1 µM SiR-(PEG)₂-NH₂ for 45 min. The cytosolic fluorescence intensity of aptamer-transfected cells (red dots) was compared with that of control cells (gray dots). Control cells were transfected with the same plasmid type, carrying the SiR-unrelated RNA insert. Fluorescence values were compared between the groups by p values: ****p(SiR-B7/control) = 2.42 × 10⁻¹² and ****p(SiR-B15/control) = 0.0001 and ****p(SELEX/control) = 1.73 × 10⁻¹⁷ using a two-sided Mann–Whitney U test. The box plots are defined as in Fig. 2f. d, Confocal images of live HEK-239T cells transfected with plasmids encoding the aptamers in c. The punctuate fluorescence is likely due to lysosomal accumulation of the dye over time. The image channels were identical to those in Fig. 3f. Scale bars = 10 µm. e, Sankey plot displaying the enrichment route of the top-ranked SiR-A and SiR-C UltraSelex aptamers. All details are as in Fig. 3g.

Source data

a, BLI K_D measurement for the reaction mixture between RNA starting pool and nsp12 proteins. b, Percentage of non-stochastic RNA species (≥2 HTS reads) in the different partitioning groups (e1–11) of UltraSelex libraries Nsp-A, Nsp-B and Nsp-C. PCR cycles were normalized. The dashed gray line represents 1% as a reference. c, Plot of the aFC values of three representative nsp12 aptamers, calculated using a top percentile value of 1.0% and a linear γ function. All details are as in Fig. 2e, but using nsp12 UltraSelex RNA aptamers. d, Sankey plot displaying the enrichment route of the top-ranked nsp12 UltraSelex aptamers. In the upper part, column AUC represents the top 0.001% of the sequences ranked by SGREELI (golden bucket), while columns e1–11 rank sequences by read abundance (golden: top 0.001%; orange: sequence observed, but ranked below 0.001%, dark green: sequences not observed in that group. The gray lines indicate the flow of the sequences between the nodes. In the magnified bottom part, the yellow bucket stands for the top 25 sequences, the gold for the top 25–top 0.001% and the orange and dark green buckets are as in the upper part.

Source data

a, Presence of the 47 UltraSelex top aptamers (representing the top 25 from SiR-A, SiR-B and SiR-C each) in the final round of the conventional SELEX library¹⁸. Only 5/47 sequences were found (pie chart) with ranks (by read abundance) between 3 and 484 (box plot). The box pots are defined as in Fig. 2f. b, BLI K_D measurement for the reaction mixture between the RNA from SELEX round 11 pool and nsp12 proteins. The concentration of RNA pool was 1000 nM. c, The rank order distribution of the 76 high-affinity RNA aptamers in three UltraSelex libraries (all 76 species found) vs three SELEX libraries against nsp12 (9 species were absent in SELEX-A, 11 in SELEX-B and 11 species in SELEX-C, ranked by their abundance after 11 rounds of selection). The box pots are defined as in Fig. 2f.

Source data

a, Chemical structure of the SiR original (blue and black part) and mock (black part only) compound for RNA aptamer selection. b, High correlation of the normalized AUC values among the UltraSelex experimental replicates against the SiR mock target. The AUC values from the top 2000 species of each replicate were analyzed. The raw AUC values were normalized by the min–max method using (AUC − min_AUC)/(max_AUC − min_AUC). Each species is denoted by a blue dot, and the reference line x = y = z is represented by the dashed gray line. c, Fraction of identical sequences between the triplicate UltraSelex datasets from b with increasing top rank of aptamers. d, Comparison of the rank order distribution of turn-on ('On') and non-turn-on ('Non') SELEX RNA aptamers in the UltraSelex libraries before (w/o mock) and after (mock) computational filtering for mock-bead binding. The RNA species were ranked by the ratio of the normalized AUC in SiR-B to SiR mock target. The y axis represents the rank order of the RNA aptamer in the specified UltraSelex library. The group of turn-on and non-turn-on RNA aptamers are shown with a purple box and a gray box, respectively. SiR-A (n = 39), SiR-B (n = 37) and SiR-C (n = 38). The box pots are defined as in Fig. 2f. e, Venn diagram showing the intersection of the 25 top-ranked RNA aptamers derived from the three mock-bead-optimized UltraSelex experiments SiR-A, SiR-B and SiR-C.

Source data

a, Consensus secondary structure of the SiR-A-like species in the top 7500 UltraSelex SiR-B library, containing the tetranucleotide UGAA (purple) in the loop and UAUC (green) in the bulge. In total, 1293 species harbor both tetranucleotides. b, Revised consensus secondary structure of the predominant species with a three base-pair stem embedding the UAUC tetraloop. c, Distribution of lengths and base pairing types of the stem embedding the UAUC loop of the RNA species in a and b. ‘|’ indicates a base pair (A–U, G–C and G–U), while ‘.’ symbolizes unpaired bases. The predominant stem type with three base pairs and two/four base pairs are shown in light and dark green, respectively, while the minor variants are depicted in blue. d, Detailed sequence information for the subgroup with a 3 bp-stem in c. Wobble pairs (G–U) are denoted by red color. ‘#’ symbolizes the UAUC loop. e, Correlation between the length of the stem embedding the ‘UAUC’ and the measured turn-on fluorescence and K_D values. Seven UltraSelex SiR-B RNA aptamers containing the ‘UAUC’ motif are depicted with their ranking order. Symbols indicate different stem types. K_D values above 2000 nM are shown as ‘2000 nM’. f, Abundance distribution of the UltraSelex SiR-B top 7500 RNA species in the mutagenized focused UltraSelex library. The left box plot panel illustrates (in triplicates) the distribution of these 7500 progenitor species from the original synthesized DNA library (input DNA) to mutagenesis PCR products, Taq PCR products to in vitro transcribed RNA. The right box plot panel shows the abundance of these progenitor species in the focused UltraSelex partitioning eluates (e1–11). The box pots are defined as in Fig. 2f.

Source data

a, The CUUGA motif is the most abundant pentamer in the random region of 45 high-affinity Nsp aptamer (K_D < 10 nM). b, Consensus secondary structure of four main binding subtypes for nsp12 aptamers using LocARNA²⁶ and RNAalifold²⁷. Degenerate bases: R(A/G), Y(C/U), M(A/C), K(G/U), S(G/C), W(A/U), H(A/T/C), B(G/T/C), V(G/A/C) and D(G/A/U). Base-pair types are color-coded; lowercase letter indicates a gap at the alignment position with a low frequency. Subtype 1: Nsp-B5, B6, B19, B21, B43, B63, B137 and B138; subtype 2: Nsp-B9, B59, B71, B76, B100 and B113; subtype 3: Nsp-B2, B8, B28, B39, B58 and B69; subtype 4: Nsp-B16, B30, B32, B68 and B79. c, Predicted secondary structures of picomolar binder Nsp-B2 (K_D = 827 pM, RdRp inhibitor) and Nsp-B58 (K_D = 892 pM, RdRp non-inhibitor) using RNAfold³⁸ with default parameters. Both belong to subtype 3 in b. The right panel highlights a structural difference in Nsp-B58 caused by a single-nucleotide deletion at position 79 in Nsp-B2. d, Illustration of classical Nsp-B113 aptamer (subtype 2 in b) truncation. The head region (113-54H) is shaded in red, while the tail region (113-51T) is enclosed by a red circle frame. The overlap area contains both features. e, Nsp-B113 active truncation variant Nsp-B113-56H+T (56 nt), concatenating the head and tail region, retains nanomolar binding affinity (K_D = 3 nM). Removing parts of primer A (‘AGCUCAGCCUUCACUGC’, Nsp-B113-56H + T) or B (‘UCGGAUCCAC’, Nsp-B113-50H+T, K_D = 4 nM in Fig. 5f) maintains nanomolar affinity. Unlike SiR aptamers, where primer A generally pairs with primer B (Extended Data Fig. 8a,b), in nsp12 aptamers primer A and primer B typically pair with parts of the randomized region (d; Supplementary Fig. 6a). f, BLI K_D measurements for Nsp-113 truncation as described in Fig. 5c.

Source data

a, Inhibition of template extension by HIV-1 reverse transcriptase (RT, HXB2; Addgene, 153311) using UltraSelex-derived 2′-F-modified RNA aptamers. An 18-nt fluorescent 5′-Cy3-labeled DNA primer (10 nM) and 31-nt DNA template (100 nM)³¹ were incubated with the indicated 2′-F CTP/UTP-modified RNA aptamer (10 nM) and RT (10 nM) for 15 min. Controls included reaction without RT (-Enzyme), RNA aptamer (PC), or dNTP (-dNTP), while the reported unmodified strong broad spectrum of RT inhibitor unmodified RNA 70N89 (ref. ³¹) served as reference. Selected 2′-F-modified RNA aptamers containing GCCU and CGGG motifs³² were designated as the ‘cg’ group. b, Quantification of the inhibition effect from a. The gray dashed line represents the inhibition level of 70N89 aptamer. RNA aptamers with more than 10% inhibition (high) are marked with an asterisk (*), while those with similar inhibition to 70N89 were classified as ‘intermediate’ group. Error bars represent the mean ± s.d. from technical triplicates. c, Consensus secondary structure of RNA aptamers with ‘high’ inhibition in b using LocARNA²⁶ alignment with default parameters. Degenerated bases are annotated as in Extended Data Fig. 9b. d, Alignment annotated with the LocARNA²⁶ consensus structure of the ‘high’ (top) and ‘intermediate’ group (bottom) from b. The percentage of inhibition higher than 70N89 is shown in the last column. e, Predicted secondary structure of the RNA aptamer RT-F6, which exhibited the highest inhibition capability, using Mfold⁴² with default parameters. f, Box plot distribution of normalized AUC values of RT RNA aptamers from b. The AUC value with the weight of heptamer was min–max normalized into range between 0 and 1. The box plot is defined as in Fig. 2f.

Source data