Structure-informed design of an ultrabright RNA-activated fluorophore

Abstract

RNA-based fluorogenic aptamers, such as Mango, are uniquely powerful tools for imaging RNA that activate the fluorescence of a weakly or non-fluorescent small molecule when bound. A central challenge has been to develop brighter, more specific and high-affinity aptamer-ligand systems for cellular imaging. Here we report an ultrabright fluorophore for the Mango II system discovered using a structure-informed, fragment-based small-molecule microarray approach. This dye-termed SALAD1 (structure-informed, array-enabled LigAnD 1)-exhibits subnanomolar aptamer affinity and 3.5-fold brighter fluorescence than Mango II-TO1-biotin pair, a widely used fluorogenic system. Performance was improved by modulating RNA-dye molecular recognition without altering the fluorophore's π-system. High-resolution X-ray structures reveal the binding mode for SALAD1, which exhibits improved pocket occupancy, a more defined binding pose and a unique bonding interaction with potassium. SALAD1 is cell-permeable and facilitates improved in-cell confocal RNA imaging. This work introduces an additional RNA-activated fluorophore demonstrating how fragment-based ligand discovery can be used to create high-performance ligands for RNA targets.

Fig. 1|. Structure and fragment-binding to the Mango aptamer.

a, Chemical structures of TO and TO1–biotin. b, Pocket analysis of the Mango II RNA aptamer modelled in the presence of TO. c, Fragment microarray-based screening using Cy5-labelled Mango II RNA (250 nM) with/without competing TO ($2.5\mu \text{M}$). Replicate screenings were performed for each sample. d, Z-score comparison of each fragment as a function of incubation conditions (Mango II versus Mango II + TO, Z-score cutoff = 3). The red line indicates Pearson correlation fitting. Fragments bind to the Mango II aptamer in both competitive (blue dots) and non-competitive modes (red dots).

Fig. 2|. Identification of fragments that co-bind RNA with TO.

a, Fluorescence intensity assay (${\lambda }_{\text{ex}}=510\text{nm};{\lambda }_{\text{ex}}=535\text{nm}$) using representative non-competitive fragments (F1–F3, F5, F6 and F10) and competitive fragments (F28–F30) discovered by microarray screening. Data are presented as mean ± s.d., based on three technical replicates ($n=3$). BRACO19 and PhenDC3, as classical G4-binders, were used as controls in the displacement study. The fragments were titrated into Mango II RNA in the presence of a saturating concentration (500 nM) of TO. The initial fluorescence of the Mango II–TO complex was normalized to 1. Fragment titrations show fluorescence enhancement and quenching effects. b,c, Chemical structure of F2 (b) and SMM screening results (Z-scores and spot images; c). d, SPR binding assay for TO, F2 and TO + F2 solutions, respectively. SPR sensorgrams were obtained in triplicate experiments for each sample. e, Quantitive analysis of binding using SPR. The SPR data are obtained from three technical replicate injections ($n=3$), and the data in the bar graph are presented as mean ± s.d. f, WaterLOGSY NMR assays demonstrating F2 binding with Mango II RNA in the presence or absence of TO. N–Me–Val–OH is used as an internal, non-binding control for comparison.

Fig. 3|. Linked fluorescent probes and their photophysical properties.

a, Secondary structure representation of Mango II complexed with a TO-based generic ligand. Lines with embedded arrowheads and Leontis–Westhof symbols denote connectivity and base pairs, respectively. b, Structures of the designed fluorescent probe (SALAD1) and three analogues (SALAD2–4). c, Excitation (${\lambda }_{\text{ex}}$) and emission (${\lambda }_{\text{em}}$) spectra of SALAD1 with and without Mango II. d, Emission spectra of TO-based ligands (free dye, ${\lambda }_{\text{ex}}=510\text{nm}$) and fluorescence intensity assay comparing RNA–dye complexes containing TO, TO1–biotin and designed analogues (${\lambda }_{\text{ex}}=510\text{nm}$; from left to right, TO, TO1–biotin, SALAD1–4). e, Emission spectra and fluorescence intensity comparison of the six RNA–dye complexes (${\lambda }_{\text{ex}}=510\text{nm}$; same colour theme and order as d). f, $K_{\text{D}}$ measurement of Mango II RNA–dye complexes. Mango II RNA was titrated into 1 nM dye solutions and the fitting was carried out based on the triplicate data points. Data are normalized to the highest signal of TO1–biotin. g, Selectivity profile comparing the fluorescence of SALAD1 (40 nM) in the presence of representative nucleic-acid structures. The experiment was carried out using a plate reader in a medium-throughput manner, based on triplicate samples. h, $K_{\text{D}}$ measurements of SALAD1 binding to four different Mango RNA aptamers. All data in e–h were acquired based on three independent measurements of each sample ($n=3$) and are presented as mean ± s.d. a.u., arbitrary units.

Fig. 4|. X-ray crystallographic study on the binding mode of dyes.

a-e, Top and side views of X-ray crystal structures of the Mango II aptamer binding site complexed with TO1–biotin (PDB 6C63; a), SALAD1 (PDB 8VXX; b), SALAD2 (PDB 8VXZ; c), SALAD3 (PDB 8VY0; d) and SALAD4 (PDB 8VY1; e). A22 is marked in yellow, and purple spheres represent K⁺. Dark grey meshes depict the |Fo|–|Fc| electron density map before building the fluorophores, contoured at $1.0\sigma$.

Fig. 5|. Confocal imaging of HEK293T cells transiently transfected with a plasmid expressing an mCherry-Mango II x24 construct.

a-h, Cells were treated with TO1–biotin (a–d) or SALAD1 (e–h). The Hoechst 33258 signal is blue, the SALAD1 or TO1–biotin signal is green, and the mCherry signal is red. Scale bars, $20\mu \text{M}$. i, Quantification of the fluorescence in b and f via the mean fluorescence intensity of SALAD1 (f) or TO1–biotin (b) in transfected cells (with Mango II) or non-transfected cells. In the box plots, the centre is the median, the top and bottom bounds of the box represent the 75th and 25th percentiles, the whiskers extend up to 1.5 times the height of the box, and the minima and maxima are the observed minima and maxima in the plots. The statistical analysis for each sample is based on the total number of cells within the imaging area. Number of cells used: SALAD1 + Mango II, $n=14$; SALAD1, $n=5$; TO1–biotin + Mango II, $n=10$; TO1–biotin, $n=5$. The statistics of difference significance are calculated based on a two-tailed Welch’s $t$-test. $****P<0.0001$. a.u., arbitrary units.