Tertiary structure of the Clivia–NBSI complex

As shown in Fig. 1b, the predicted secondary structure of Clivia contains two stems, P1 (orange) and P2 (green), that are connected by one big internal loop, J12 (purple). To facilitate the crystallization of the Clivia complex bound to NBSI, the variable apical loop L2 of stem P2 was replaced with the stable tetraloop or protein-binding loop (cocrystallization with the RNA-binding protein) (Fig. 1b), in which the GAAA tetraloop replacement generated crystals with high diffraction quality. The structure of the Clivia–NBSI complex was refined at a resolution of 1.6 Å with Rwork/Rfree values of 0.19/0.22 (Supplementary Table 1). Each asymmetric unit contains two molecules of the Clivia–NBSI complex and minor molecular interactions were found between them (Supplementary Fig. 1a,b). Then, we used size-exclusion chromatography to detect the solution state of the Clivia–NBSI complex and found that it existed homogeneously as a monomer in solution (Supplementary Fig. 1c,d). Superposition of two molecules of the Clivia–NBSI complex in one asymmetric unit generated a root mean squared deviation (r.m.s.d.) value of 1.09 Å (ref. 23). To simplify the description, we focus on the structure of one molecule of the complex below. The contacts described below had a distance cutoff of 3.5 Å.

The tertiary fold of the Clivia–NBSI complex is shown schematically in Fig. 1c and its cartoon representation is shown in Fig. 1d. The overall structure of the complex adopts a coaxial long helix fold and is stabilized by continuous stacking interactions from the bottom stem P1 to the apical stem P2 with the ligand NBSI intercalated in the zipped junction region at the interface of stems P1 and P2 (Fig. 1c,d). A15 and G28 from the internal loop form a noncanonical base pair and elongate stem P2 adjacent to the terminal base pair U16–A27 of stem P2. A6 and A31 from the internal loop form another noncanonical base pair and stack on the northern top of stem P1. U8 pairs with A29 in the internal loop below the bound dye, NBSI. The consecutive nucleotides from G9 to C14 (shown in purple) bend downward to interact with the minor groove of stem P1 and form a stable multilayered platform for NBSI binding. The NBSI-binding pocket is located in the central junction of the overall structure between stems P1 and P2 (Fig. 1c,d).

Stabilization interaction of the internal loop

Two fully hydrated Mg2+ ions (M1 and M2) were identified near the NBSI-binding pocket (Fig. 1d and Extended Data Fig. 1a), confirmed by the anomalous signal collected using Mn2+-soaked crystals (Supplementary Fig. 2). They form extensive hydrogen-bonding interactions with the sugar–phosphate backbone of stem P2 and the junction region, which may contribute to the stabilization of the NBSI-binding pocket (Extended Data Fig. 1b).

In addition to forming the noncanonical base pairs between stems P1 and P2, the consecutive nucleotides G9-U10-A11-A12-A13-C14 in internal loop J12 form a compact cap-like structure along the minor groove of stem P1 (Fig. 1c,d), in which U10-A11-A12-A13 forms continuous stacking interactions parallel to stem P1 and the two terminal residues G9 and C14 stack partially on each other in a direction tilted toward the long helix (Extended Data Fig. 1b,c). The Watson–Crick edge of G9 forms several hydrogen bonds with the 2′-OH group and the nonbridging phosphate oxygen of C14 and the bridging and nonbridging phosphate oxygens of A13 (Extended Data Fig. 1c,d). Furthermore, the 4-NH2 group of C14 forms another direct hydrogen bond with the 4′-O group of G9 (Extended Data Fig. 1c,d). It is notable that both G9 and C14 adopt a C2′-endo ribose sugar pucker conformation. The 2′-OH group of U10 forms a direct hydrogen bond with the N7 atom of A12 and the N3 atom of U10 forms another hydrogen bond with the nonbridging phosphate oxygen between A12 and A13 (Extended Data Fig. 1c).

A11, A12 and A13 stack on each other in the minor groove side of stem P1, consequently forming three consecutive tiers (Fig. 1c–g and Extended Data Fig. 1e). In the first tier, A11 interacts with the sugar edge of U33 along its Watson–Crick edge and forms an A4–U33·A11 base triple (Fig. 1e). In the second tier, A12 forms two hydrogen bonds with the base of G5 and one hydrogen bond with the 2′-OH group of G5. Additionally, the 2′-OH group of U10 was found to interact with the N7 atom of A12 in the same tier (Fig. 1f). The third tier is formed by three residues (A6, A13 and A31) from the internal loop, in which A6 forms a reverse base pair with A31 along their Watson–Crick edges and A13 interacts with the Hoogsteen edge of A6 from the minor groove side of stem P1 (Fig. 1g). It is notable that the ribose of A6 adopts a 2′-endo sugar pucker conformation in the structure (Fig. 1g). In addition, hydrogen-bonding interactions are found between different tiers (Extended Data Fig. 1e). The 2′-OH group of C30 in the first tier forms a hydrogen bond with the 6-NH2 group of A31 in the second tier. Additionally, the 2′-OH group of A6 in the second tier forms a hydrogen bond with the phosphate between A4 and G5 (Extended Data Fig. 1e).

Composition of NBSI-binding pocket and recognition of NBSI

The NBSI-binding pocket is located at the interface of stems P1 and P2 in the tertiary fold of the Clivia aptamer (shown in surface representation with NBSI shown in ball-and-stick representation; Fig. 2a). The major aromatic moieties of NBSI intercalate into the binding pocket, while the terminal (2-hydroxyethyl)(methyl)amino group of NBSI points outward from the binding pocket (Fig. 2a). Three groups of successive residues (C14-A15, U8-G9 and G28-A29-C30) from internal loop J12 constitute a three-sided box and encompass the bound NBSI in the binding pocket (Fig. 2b,c).

Fig. 2: Structural alignment of NBSI-binding pocket of the Clivia aptamer.

a, Surface representation of the Clivia aptamer bound to NBSI shown in stick representation. b, The planar moiety of ligand NBSI (shown in ball representation) is surrounded by three groups of consecutive junction residues U8-G9, C14-A15 and G28-A29-C30. A15 and G28 form one base pair and stack above NBSI. U8, A29 and C30 form one base triple and stack below NBSI. G9 and C14 stack on each other and bracket the side of NBSI. c, NBSI intercalates into the binding pocket and is sandwiched by two adjacent bases, in which the composite omit map of NBSI contoured at level 1.0σ is shown. d, The 2-NH2 group of G28 forms two hydrogen bonds with the N7 atom and nonbridging phosphate oxygen of A15. e, The Watson–crick edge of U8 forms a pairing interaction with the Hoogsteen edge of A29. The N3 atom of A29 forms an additional hydrogen bond with the 4-NH2 group of C30 and the 6-NH2 group of A29 forms an additional hydrogen bond with the N3 atom of C14.

As shown in Fig. 2b,c and Extended Data Fig. 2a, NBSI is sandwiched between the upper two contacting bases A15·G28 and the lower base triple U8·A29·C30, anchored by the sugar–phosphate RNA backbone between G28 and A29 and capped by two adjacently stacked residues C14 and G9. The composite omit map of NBSI contoured at level 1.0σ is shown in Fig. 2c and Extended Data Fig. 2a. The 2-NH2 group of G28 forms a hydrogen bond with the N7 atom of A15 and a hydrogen bond with the nonbridging phosphate oxygen between A15 and C14 above the bound NBSI (Fig. 2d). The Hoogsteen edge of A29 forms a pairing interaction with the Watson–Crick edge of U8 and the N3 atom in the sugar edge of A29 forms another hydrogen bond with the 4-NH2 group of C30 below the bound NBSI. In addition, it was found that the 6-NH2 group of A29 forms a hydrogen bond with the N3 atom of the capped residue C14 (Fig. 2e). Here, U8 and A29 also adopt a 2′-endo sugar pucker conformation.

Even though no direct hydrogen-bonding interaction was identified between NBSI and the Clivia aptamer in the binding pocket, water molecule W1 was found to coordinate with the N1 atom of the imidazole moiety of NBSI and form extra hydrogen bonds with the neighboring residues, potentially facilitating the specific recognition of NBSI by the Clivia aptamer (Extended Data Fig. 2b). Additionally, one Mg2+ ion (M1) was observed in vicinity of the binding pocket, with one hydrated water WM1 forming two hydrogen bonds with the 2′-OH group of U8 and the O6 atom of G28, aiding in the formation of the binding pocket (Extended Data Fig. 2c).

Investigation of RNA folding and recognition of NBSI

On the basis of the architecture of the Clivia–NBSI complex, we introduced residue-specific substitutions to detect their contributions to RNA folding and NBSI recognition (Fig. 3a). As shown in Fig. 1e–g, A11-A12-A13 formed three consecutive stacked base triples below the binding pocket. Substitutions of A11-A12-A13 to G11-G12-G13, C11-C12-C13 and U11-U12-U13 all disrupted binding activity (Fig. 3a). Replacement of U8–A29, located below the bound NBSI, with either U8C–A29G or U8A–A29U resulted in a complete loss of binding activity (Fig. 3a). Then, we substituted the noncanonical base pair G28–A15, located above the bound NBSI. The binding activity of G28A–A15G and G28–A15C was completely abolished, while G28U–A15 resulted in severely reduced binding activity (Fig. 3a). It was noted that U7 protrudes from the main helical architecture of the Clivia monomer (Extended Data Fig. 3a) and participates in the molecular interaction of mol A and mol B in the asymmetry unit by partially stacking on the variable loop of stem P2 (Supplementary Fig. 1b). The deletion of U7 (Del-U7) resulted in a complete loss of NBSI-binding activity (Fig. 3a) but the substitution of U7 with C, G or A was tolerated (Extended Data Fig. 3b). As mentioned before, the NBSI-binding pocket is sandwiched between the coaxially stacked stems P1 and P2 in the tertiary fold. To assess the effect of stem length on NBSI-binding activity, stem P1 and stem P2 were elongated and shortened by two base pairs, respectively. Normalized fluorescence experiments showed that the binding activity and binding affinity of these mutants were comparable to those of the wild-type (WT) construct (Fig. 3a,b).

Fig. 3: Fluorescence assay of the Clivia aptamer and mutants.

a,b, Fluorescence assay of the Clivia aptamer (a) and structure-guided mutants (b). Here, Del-2bp-P1 indicates that G2-A3 and U34-C35 were removed from stem P1, Ad-2bp-P1 indicates that G(−1)-G0 and C37-C38 were added to stem P1, Del-2bp-P2 indicates that U16-G17 and C26-A27 were removed from stem P2 and Ad-2bp-P2 indicates that C-C and G-G were added to the apical end of stem P2. The activated fluorescence of NBSI by Clivia mutants from three independently repeated experiments is normalized for comparison with the WT Clivia aptamer. Data represent the mean ± s.d. from three replicates. c, The 3D fluorescence assay of fluorophore NBSI in the presence of H2O (blue), buffer (green) or Clivia (red). In the presence of water or buffer, free NBSI exhibits two excitation and emission wavelength peaks at approximately (460 nm, 600 nm) and (560 nm, 600 nm), while Clivia-bound NBSI exhibits single excitation (529 nm) and emission (584 nm) wavelength peaks.

Source data

As shown in Fig. 1d and Supplementary Fig. 2, two bound Mg2+ cations (M1 and M2) were observed adjacent to the NBSI-binding pocket. To evaluate the contributions of the cations to the structural folding of Clivia and the ligand recognition of NBSI, we tested the binding activities of Clivia with NBSI in the existence of different concentrations of Mg2+ (Extended Data Fig. 3c) and other cations, including Ca2+, K+ and Ba2+ (Extended Data Fig. 3d). In the absence of Mg2+, the binding affinity of Clivia and NBSI was 225 nM (Extended Data Fig. 3c). In the presence of Mg2+ with concentration varied from 2 to 40 mM, the binding affinity became 50–61 nM (Extended Data Fig. 3c). The binding activity was not affected when the cations were changed to 20 mM Ca2+ or 1 M K+. However, when replaced with 20 mM Ba2+, the binding affinity was decreased to 143 nM (Extended Data Fig. 3d).

In addition to detecting the structural folding features of Clivia, we investigated the fluorescence characteristics of NBSI under different conditions, including in water, in buffer and in the presence of Clivia. As shown in the three-dimensional (3D) fluorescence spectra (Fig. 3c and Extended Data Fig. 4), NBSI exhibited two distinct excitation peaks at approximately 460 and 560 nm in both water and buffer environments. Interestingly, upon the introduction of Clivia, the excitation spectrum of NBSI showed a single peak centered around 530 nm. Moreover, the comparison of the emission intensity of Clivia-bound NBSI (100 nM) to that of free NBSI at the same concentration revealed that the emission of free NBSI was negligible in comparison (Fig. 3c and Extended Data Fig. 4). These observations collectively suggest that the interaction between NBSI and Clivia induces a notable alteration in the intricate fluorescence attributes of NBSI.

Photophysical properties of Clivia-bound NBSI derivatives

Our previous findings demonstrated that modifying the fluorophore structure can effectively alter its photophysical properties, resulting in a broader range of fluorescence wavelengths, including a larger Stokes shift19,22. As shown in the binding pocket of NBSI (Fig. 2a), the terminal (2-hydroxyethyl)(methyl)amino group of NBSI points outward from the ligand-binding pocket of the Clivia aptamer, providing ample space for NBSI modification.

In the reported NBSI derivative, NBSI571 (2)22, this terminal (2-hydroxyethyl)(methyl)amino group is replaced by a minor hydroxyl group at position 4, along with two additional fluoride atoms at adjacent positions 3 and 5 of the benzylidene moiety (Fig. 4a). The fluorescence binding experiments revealed that the binding affinity of NBSI571 (Kd = 58 nM) remained unaffected by the substitutions compared to NBSI (Kd = 55 nM) (Extended Data Fig. 5a,b)22. Next, 3D fluorescence spectra were generated for both free NBSI571 in buffer and Clivia-bound NBSI571 (Fig. 4b). Free NBSI571 in buffer exhibited two excitation wavelength peaks at 382 and 554 nm, whereas Clivia-bound NBSI571 displayed only one excitation wavelength peak at 497 nm. The binding of Clivia resulted in a notable alteration in the fluorescence characteristics of NBSI571. Then, we performed cocrystallization of the Clivia aptamer with NBSI571. The binding pocket of the Clivia–NBSI571 complex structure revealed that the three newly attached groups (3,5-F and 4-OH) of NBSI571 project outward in a manner similar to the (2-hydroxyethyl)(methyl)amino group of NBSI (Figs. 2a,b and 4c,d).

Fig. 4: Structure of the Clivia aptamer binding with modified NBSI analogs.

a, The chemical structure of NBSI analogs NBSI571, NBSI565 and NBSI618. b, The 3D fluorescence assay of free (green) and Clivia-bound (red) NBSI571. In the presence of buffer, free NBSI571 exhibits two excitation and emission wavelength peaks at (382 nm, 582 nm) and (554 nm, 594 nm), while Clivia-bound NBSI571 exhibits single excitation (497 nm) and emission (572 nm) wavelength peaks. c, Surface representation of the Clivia aptamer bound to NBSI571 shown in stick representation. d, Binding pocket composition of NBSI571. The binding pocket of NBSI571 in the Clivia aptamer suggests that there is room for additional modifications on the phenyl moiety of NBSI571 to enhance dye properties. e,f, Binding pocket composition of NBSI565 (e) and NBSI618 (f).

Source data

Two additional reported derivatives of NBSI, designated NBSI565 (3) and NBSI618 (4), were generated by introducing a 4-OH or 4-CN group to the styryl moiety of NBSI571, respectively (Extended Data Fig. 5a)22. NBSI565 had an excitation maximum at 490 nm and an emission maximum at 565 nm, while its binding affinity was Kd = 22 nM (Extended Data Fig. 5b)22. Our 3D fluorescence spectra of free NBSI565 in buffer and Clivia-bound NBSI565 revealed that free NBSI565 in buffer exhibited two excitation wavelength peaks at 382 and 552 nm, whereas Clivia-bound NBSI565 displayed only one excitation wavelength peak at 491 nm (Extended Data Fig. 5c). Structure determination of the Clivia–NBSI565 complex demonstrated a remarkable resemblance of the binding pocket architecture to that of NBSI571 and NBSI. NBSI565 was stacked in the binding pocket with the attached 4-OH group residing at the terminus of the binding pocket (Fig. 4e).

In NBSI618, a 4-CN group was incorporated into the styryl moiety and resulted in an excitation maximum at 510 nm and an emission maximum at 618 nm, thereby achieving a substantial Stokes shift of 108 nm (Extended Data Fig. 5a,b)22. The distinct 3D fluorescence spectra of free NBSI618 in buffer and Clivia-bound NBSI618 clearly demonstrated that the binding of Clivia induced a notable change in the fluorescence properties of NBSI618, as evidenced by a shift in the excitation wavelength (Extended Data Fig. 5d). Structure determination of the Clivia–NBSI618 complex revealed that NBSI618 retained the ligand conformation and interaction observed for NBSI (Fig. 2), NBSI571 (Fig. 4c,d) and NBSI565 (Fig. 4e). The elongated 4-cyano-benzylidene moiety of NBSI618 was capsuled within the binding pocket of Clivia (Fig. 4f). Detailed analysis revealed a substantial increase in the buried area of the fluorophore within Clivia, expanding from 383.2 Å2 for NBSI571 to 412.1 Å2 for NBSI618. This enhancement highlights the robust ligand-binding pocket of the Clivia aptamer, potentially contributing to the remarkably large Stokes shift observed in NBSI derivatives. It is worth noting that the largest Stokes shift observed for NBSI618 binding to Clivia may also partially stem from the intrinsic molecular properties of NBSI618.

Orthogonal application of Clivia and alternate aptamers

The chemical structure of NBSI consists of an electron donor, an electron acceptor and a bridged π conjugate structure (Fig. 5a). We found that the bridged π conjugate substructure present in NBSI and its derivatives also exists in the well-known conventional fluorophores DFHBI (5) and DFHO (6) (Fig. 5a), which are capable of being activated by various RNA aptamers, including Spinach24, iSpinach25, Corn11 and Broccoli26. In addition, NBSI571 contains a hydroxyl group at position 4 and two fluoride atoms at adjacent positions 3 and 5 of the benzylidene moiety present in DFHBI and DFHO.

Fig. 5: Structural basis underlying the orthogonal use of Clivia and alternate aptamers.

a, Chemical structure comparison of NBSI, DFHBI, DFHO and NBSI571. b, Docking of NBSI571 into the binding pocket of Spinach bound to DFHBI (PDB 4TS0). c, Docking of NBSI571 into the binding pocket of Corn fluorogenic aptamer bound to DFHO (PDB 5BJP) results in a steric clash of the phenyl moiety of NBSI571 with the base of A11 and the sugar of G12 from Corn. d, In vitro fluorescence assay of the Clivia and Broccoli aptamers binding to NBSI, NBSI571 and DFHBI. The in vitro fluorescence activation of NBSI, NBSI571 and DFHBI by the Broccoli aptamer is normalized for comparison with the Clivia aptamer. All tests were performed three times independently. Data represent the mean ± s.d. from three replicates. e, Activation of NBSI, NBSI571 and DFHBI by the Clivia and Broccoli aptamers in HEK293T cells. Scale bars, 20 μm.

Source data

Given the strong structural resemblance of NBSI571 with DFHBI and DFHO (highlighted in blue and red, respectively, in Fig. 5a), we conducted a detailed comparison of the binding patterns of these fluorophores with their corresponding RNA-based fluorogenic aptamers (Fig. 5b,c and Extended Data Fig. 6). It was observed that both DFHBI and DFHO are recognized by the same molecular moiety, including the benzylidene substructure (shown in bule) and the bridged π conjugate substructure (shown in red), present in NBSI571 (Fig. 5a and Extended Data Fig. 6a). However, our attempts to dock NBSI571 into the binding pockets of Spinach (Protein Data Bank (PDB) 4TS0), iSpinach (PDB 5OB3), Corn (PDB 5BJP) and Beetroot (PDB 8EYV) revealed severe steric clashes with the phenyl group of NBSI571 (Fig. 5b,c and Extended Data Fig. 6). These results indicate the unique fluorophore recognition pattern of Clivia, implying that NBSI and its derivatives, including NBSI571, can be used in an orthogonal manner for the simultaneous tracking of diverse RNA molecules alongside other fluorophores such as DFHBI and DFHO.

To explore the potential orthogonal use of Clivia and alternate aptamers, we conducted an in vitro fluorescence assay of the Clivia and Broccoli aptamers binding to NBSI, NBSI571 and DFHBI (Fig. 5d). The results revealed that Clivia exclusively triggered the fluorescence of NBSI or NBSI571 in the presence of DFHBI, while Broccoli activated the fluorescence of only DFHBI (Fig. 5d). In live-cell imaging, after HEK293T cells were transfected with plasmids encoding either Clivia or Broccoli, distinct fluorescence responses were observed. When these cells were subjected to NBSI and NBSI571, the cells encoding Clivia emitted a vibrant red fluorescence. Conversely, when treated with DFHBI, the cells encoding the Broccoli aptamer displayed a vivid green fluorescence (Fig. 5e). Additionally, we performed an in vitro fluorescence assay of the Clivia and Corn aptamers binding to NBSI, NBSI571 and DFHO (Extended Data Fig. 7a). We observed that Clivia did not activate the fluorescence of DFHO, which was further supported by live-cell imaging experiments in HEK293T cells (Extended Data Fig. 7b). This stark contrast in fluorescence colors serves as a clear indicator of the specific interactions between these aptamers and their respective fluorophores, thus confirming Clivia’s unique ability to distinguish and activate fluorescence in response to NBSI and its derivatives.

Design of multivalent Clivia fluorogenic aptamers

Accurate spatiotemporal tracking of RNA molecules, particularly those that are present in low abundance, relies on sensitive detection and quantification. Furthermore, high concentrations of fluorophores may potentially interfere with normal cellular activities. To enhance the imaging sensitivity of individual RNA molecules in the presence of a controlled concentration of fluorophore NBSI, we embarked on the design of constructs containing multiple NBSI-binding pockets arranged in tandem within a single sequence.

Inspired by the coaxial tertiary architecture of the monomeric Clivia aptamer and the substitution results from Figs. 1d and 3a,b showing that the alteration of stem length had a minor effect on the fluorescence activation of Clivia, we used multivalency-based strategies27 to design a multivalent Clivia fluorogenic aptamer that houses repeated NBSI-binding modules from Clivia (CM modules) in a coaxial fold, termed Clivia_N (where N represents the copy number of CM modules) (Fig. 6a). Guided by the tertiary structure folding of monomeric Clivia (Fig. 1c,d), we used 3 bp from the north terminus of stem P1 and 2 bp from the south terminus of stem P2 as the linked stem (Fig. 6a). Therefore, the size of each NBSI-binding module in the multivalent Clivia fluorogenic aptamer was decreased from more than 36 nucleotides to 24 nucleotides (Figs. 1c and 6a).

Fig. 6: Design, fluorescence assay and structure of NBSI-binding modules in tandem arrays.

a, Design of multivalent Clivia fluorogenic aptamers containing multiple NBSI-binding modules (CM modules) in a tandem array. The linked stem was reduced to 5 bp (3 bp from stem P1 and 2 bp from stem P2). Each NBSI-binding module contained 24 nucleotides. b, Tertiary structure of Clivia_III with three NBSI-binding modules in a tandem array stacking continuously in a long helix. c, Composition of the first NBSI-binding pocket in Clivia_III. d,e, Live-cell imaging (e) and quantitation of the cellular fluorescence (d) for the multivalent Clivia fluorogenic aptamers containing 1–12 NBSI-binding modules. Scale bars, 10 μm. Data represent the mean ± s.d. (n = 150 cells).

Source data

To detect whether the repeated NBSI-binding pockets were affected by the tandem-arrayed pattern, we set out to solve the complex structure of the multivalent Clivia fluorogenic aptamer bound to NBSI and succeeded with the Clivia_III construct containing three CM modules (Fig. 6b and Extended Data Fig. 8). As shown in Fig. 6b and Extended Data Fig. 8a,b, the three NBSI-binding modules, CM1, CM2 and CM3, stacked continuously with the intercalated linked stem in the Clivia_III–NBSI complex structure. The 5-bp linked stem between every two NBSI-binding modules was involved in A-minor interactions (A42 and A41) with the NBSI-binding pocket situated above it (Fig. 6c and Extended Data Fig. 8a–d), as shown in the monomer structure (Fig. 1c,d). The residues lining each binding pocket were not affected by the tandem array of the NBSI-binding modules (Fig. 6b and Extended Data Fig. 8e,f). Then, we used size-exclusion chromatography to examine the solution state of the multivalent Clivia fluorogenic aptamers containing 1–6 repeated CM modules. Our analysis revealed that all aptamers maintained a homogeneous monomeric state in solution (Extended Data Fig. 9a). These observations establish the structural basis for the efficient application of multivalent Clivia fluorogenic aptamers for RNA localization and trafficking.

To assess the practical utility of multivalent Clivia fluorogenic aptamers, we conducted in vitro fluorescence assays using constructs with 1–12 repeated CM modules. The results exhibited enhanced fluorophore sensitivity and linearly increasing intensity with the growing number of CM modules in the same condition (Extended Data Fig. 9b). Furthermore, we transfected plasmids encoding multivalent Clivia fluorogenic aptamers with 1–12 CM modules into HEK293T cells. Upon incubation with the same concentration of NBSI, the intracellular fluorescence of the multivalent Clivia–NBSI complex exhibited a notable and consistent increase when up to eight NBSI-binding modules were arranged in tandem (Fig. 6d,e). However, intriguingly, when the copy number exceeded this point (10–12 CM modules), there was a decrease in fluorescence intensity. Therefore, we concluded that the optimal copy number of CM modules in the multivalent Clivia fluorogenic aptamer for maximal fluorescence enhancement is 8 (Fig. 6d,e).

Importance and comparison of base triples in RNA structures

As shown in Fig. 1, there are four non-G-quadplex tiers that constitute the platform below the bound ligand in the Clivia–NBSI complex structure, which are all composed of base triples. The top tier is composed of a noncanonical base triple formed by three residues (U8, A29 and C30) from junction J12 (Fig. 2e). Three additional tiers are formed by base triples involving three consecutive adenines A11, A12 and A13 from the junction region (Fig. 1e–g and Extended Data Fig. 1). Compared with other solved RNA fluorogenic molecule structures, we found that base triples also constitute the foundational tiers within the tertiary complex structure of the MG aptamer bound to TMR (7) dye (PDB 1F1T; Extended Data Fig. 10). As shown in Fig. 1d and Extended Data Fig. 10a,b, both the MG aptamer and the Clivia aptamer feature a helical scaffold with the fluorogenic dye binding at the center of the structure. Two base triples formed by A26 and A27 from the junction region and the minor groove of two base pairs (A22–U11 and G23–C10) from stem P2 in the MG aptamer show identical interactions to base triples featuring A11 and A12 in the Clivia aptamer (Fig. 1e–g and Extended Data Fig. 10c,d). Despite the high similarity in the overall helical structure and the platform of base triples, it is important to note that the specific recognition module and the bound fluorophore molecule structures are completely distinct (Fig. 2 and Extended Data Fig. 10e,h). A fluorescence assay of the MG aptamer in the presence of NBSI indicated that the MG aptamer could not activate the fluorescence of NBSI, contrasting with the activation behavior observed for Clivia (Extended Data Fig. 10i). This highlights the inherent versatility of RNA molecules and their specificity in recognizing small fluorophore molecules with diverse chemical structures. Additionally, base triples were identified as components of the drug-binding pocket in a riboswitch-like RNA structure from the internal ribosome entry site of hepatitis C virus (PDB 3TZR; Supplementary Fig. 3a,b). The ligand SS0 (8) was recognized by the G52–C111–A57 and U59–A109–A53 base triples (Supplementary Fig. 3c–e). This suggests that base triples may serve as common structural elements for interacting with small molecules.