A Color‐Shifting Near‐Infrared Fluorescent Aptamer–Fluorophore Module for Live‐Cell RNA Imaging

Abstract Fluorescent light‐up RNA aptamers (FLAPs) have become promising tools for visualizing RNAs in living cells. Specific binding of FLAPs to their non‐fluorescent cognate ligands results in a dramatic fluorescence increase, thereby allowing RNA imaging. Here, we present a color‐shifting aptamer‐fluorophore system, where the free dye is cyan fluorescent and the aptamer‐dye complex is near‐infrared (NIR) fluorescent. Unlike other reported FLAPs, this system enables ratiometric RNA imaging. To design the color‐shifting system, we synthesized a series of environmentally sensitive benzopyrylium‐coumarin hybrid fluorophores which exist in equilibrium between a cyan fluorescent spirocyclic form and a NIR fluorescent zwitterionic form. As an RNA tag, we evolved a 38‐nucleotide aptamer that selectively binds the zwitterionic forms with nanomolar affinity. We used this system as a light‐up RNA marker to image mRNAs in the NIR region and demonstrated its utility in ratiometric analysis of target RNAs expressed at different levels in single cells.


Introduction
RNAl abeling methods combined with advanced fluorescence microscopy enable imaging the complex dynamics of RNAi nl iving cells with ah igh temporal and spatial resolution. Thec onventional methods to label RNAe mploy fluorophore-labeled hybridization probes [1] and molecular beacons, [2] yet they suffer from impermeability and heterogeneous distribution of the probes. [3] Fluorescent proteintagged RNA-binding proteins (RBP) are still the current gold standard for live-cell RNAi maging. [4] However,t hese RBPbased tagging systems including ar ecently developed CRISPR-Cas system for endogenous RNAi maging [5] have limitations such as high background fluorescence and potential alterations in RNAproperties due to the bulky tag. [3] Thecurrently emerging field of fluorescent light-up RNA aptamers (FLAPs) provides powerful tools for live-cell RNA imaging.Non-covalent binding of cell-permeable,fluorogenic dyes to their cognate aptamers induces af luorescence turnon, enabling RNAd etection with high signal-to-background ratios.A mong literature-reported FLAP systems,t he origin of dye fluorescence quenching can be based on: i) vibrational and rotational motions (e.g. MG aptamer, [6] Spinach [7] and its relatives, [8] Pepper [9] ); ii) ground-state complex formation (e.g. SRB-2, [10] DNB, [11] Riboglow, [12] o-Coral [13] and Rho-BAST [14] ); iii) spirolactonization (e.g.SiRA [15] ). SiRA was an important milestone towards in vivo RNAimaging because of the lack of photostable and bright FLAPs that fluoresce in the NIR region.
Thefluorogenic nature of FLAPs enables imaging RNAs of interest (ROIs) in living cells in the presence of their cognate ligands.H owever,d ifferences in the cellular uptake of the dye between different cells,h eterogeneous probe distribution within as ingle cell, probe instability due to photobleaching,variations in the cell morphology such as cell thickness and focal plane during an imaging experiment could cause signal fluctuations and consequently misinterpretation of data. [16] An effective way to solve this problem is to use ar atiometric system that enables imaging both the free dye and the RNA-fluorophore complex simultaneously.Unfortunately,all FLAPs reported so far are based on as ingle-color fluorescence turn-on.
In this work, to address these issues,w er eport the evolution, characterization, and application of an ovel colorshifting NIR-fluorescent aptamer-fluorophore module based on spirolactamization of fluorophores for RNAi maging in living cells ( Figure 1). Thef ree fluorophore is cyan fluorescent in solution while the aptamer-bound form fluoresces in the NIR region. We demonstrate the application of this colorshifting FLAP system as agenetically encoded tag to visualize mRNAs in the NIR region using its light-up feature as well as its capability for ratiometric analysis of RNAe xpression levels in single cells.

Results and Discussion
Design of aColor-Shifting Aptamer-Fluorophore Module. Rhodamine dyes have been widely used for biological imaging owing to their high quantum yield, excellent photostability,a nd good cell permeability. [13,15,17] Furthermore,t he rational design of new fluorescent probes that modulate the spirocyclization of rhodamine dyes is becoming ap owerful method for background-free fluorescence microscopy in vivo. [18] Thespirocyclic(lactone or lactam) form of rhodamine is colorless and non-fluorescent due to the disruption of the conjugated p-system, whereas the open form is colored and fluorescent due to the extended p-conjugation. This fluorescence switching mechanism has also been applied to NIRfluorescent silicon rhodamines,e nabling precise modulation of the ring-open and -closed forms. [18,19] However,these NIR fluorescent rhodamine scaffolds can only be used for fluorescence turn-on probes.
Here,w ea imed to design ac olor-shifting aptamerfluorophore module that enables simultaneous imaging of the RNA-fluorophore complex and the free fluorophore. Moreover,itisdesirable to have the emission of the complex in the NIR region due to its better live cell compatibility.T o this end, we chose an environmentally sensitive hybrid fluorophore consisting of benzopyrylium and coumarin moieties (BC)a st he core structure of our probes.S imilar to rhodamine dyes, BC fluorophores exist in an equilibrium between as pirocyclica nd az witterionic form, however the closed form is cyan-and the open form is NIR-fluorescent. [20] We envisioned that the color-switching property of BC fluorophores could be exploited for ratiometric RNAimaging provided that ah igh-affinity aptamer binding exclusively to the open form can be generated. Moreover,the fluorescence light-up feature of the aptamer can be used for imaging RNAs in the NIR region. Thei deal color-shifting BC compound should exist mainly as ac yan fluorescent spirocyclicf orm in the unbound state;however,itshould efficiently switch to the NIR fluorescent zwitterionic form upon binding to the aptamer.
In order to increase the chances to evolve an aptamer binding preferentially to the NIR fluorescent open form, we planned to use a BC fluorophore that mainly exists in the zwitterionic form in aqueous solution at physiological pH. This fluorophore was immobilized on as olid support as bait during the in vitro evolution of aptamers.After the discovery of the aptamer,the structure of the BC scaffold was minimally modified in order to fine tune the open-closed ratio.F inally, the color-shifting efficiencyo ft he aptamer upon binding to as eries of BC analogs was investigated to find the best aptamer-fluorophore pair for ratiometric RNAi maging.
In vitro Evolution of Aptamers for BC1. We performed SELEX (Systematic Evolution of Ligands by EXponential enrichment) to find RNAaptamers binding specifically to the BC zwitterion. [21] BC1 was chosen as the ligand for in vitro selection because it predominantly exists in the NIR fluorescent zwitterionic form [20b] (Figure 2A). To attach ligands to asolid support for SELEX, amine-functionalized BC1 (BC1-NH 2 ,F igure 2A)w as synthesized and immobilized on N-Hydroxysuccinimide-activated sepharose beads (Scheme S1). [15, 20b] We started the selection with an RNAl ibrary containing % 3 10 15 different sequences.T he sequence of the RNA library consisted of a5 '-forward primer binding site,a26nucleotide (nt) random region, a1 2-nt constant sequence forming astable stem-loop,another 26-nt random region and a3 '-reverse primer binding site [22] (Figure 2B). RNAt ranscripts were first incubated with am ock resin to remove sepharose-binding RNAs equences,a nd then with sepharose beads functionalized with BC1 ( % 4mMonthe resin). After washing the beads,b ound RNAw as nonspecifically eluted with formamide solution, precipitated, reverse-transcribed and PCR-amplified. Theo btained DNAw as in vitro transcribed and the enriched RNApool served as an input for the next round of selection ( Figure S1). After 5r ounds of SELEX, the fraction of eluted RNAi ncreased to 35 %, indicating the successful selection of BC1-binding aptamers ( Figure 2C). In order to select for high affinity binders,w e increased the selection pressure by gradually decreasing the ligand concentration on the beads.F or this purpose,w e synthesized abiotinylated BC1 ligand containing alinker with ad isulfide bond (BC1-SS-Biotin,F igure 2A)a nd used streptavidin-conjugated beads as solid supports (Scheme S1). This way,the concentration of BC1-SS-Biotin on streptavidinbeads could be precisely and easily controlled and decreased during the course of SELEX. Furthermore,b ound RNA could be specifically eluted with DTT by reducing the disulfide bond between the BC1 and biotin. Selection stringencyw as increased gradually by decreasing the ligand concentration from 4mMt o1 0mM, lowering RNAi nput concentration from 100 to 10 mM, increasing the washes from 6t o3 0c olumn volumes,a nd eluting resin-bound RNAwith DTT ( Figure 2C). After eleven iterative rounds,t he eluted RNAp ool was reverse-transcribed, PCR-amplified, cloned into plasmids that were subsequently transformed into bacteria. Single colonies of bacteria were picked and the plasmids were sequenced by Sanger sequencing.
Identification of aH igh Affinity BC1-Binding Aptamer. After analyzing 60 individual colonies,w ei dentified 52 unique RNAs equences ( Figure S2), all of which were in vitro transcribed for screening according to their affinity towards BC1.S ince BC1 is predominantly in the NIRfluorescent zwitterionic form, binding of an aptamer to BC1 would not necessarily cause afluorescence or spectral change, making BC1 impractical for aptamer screening.T osolve this problem, we synthesized afluorescence turn-on probe of BC1 by attaching the contact quencher dinitroaniline (DN)t o yield BC1-DN ( Figure 2A and Scheme S2). DN effectively quenches the fluorescence of BC1 by forming an intramolecular ground state heterodimer. [10] In the presence of a BC1-binding aptamer,t he fluorophore would interact with the aptamer rather than the quencher,t hereby enhance the fluorescence.All RNAtranscripts were screened based on the fluorescence intensity enhancement of BC1-DN (100 nM) upon mixing with aptamers (10 mM) ( Figure S2). Out of 52 sequences,8were determined to be highly active showing > 6-fold fluorescence enhancement, 29 were moderately active (3to 6-fold), and 15 were less active (< 3-fold). After measuring dissociation constants (K D )between BC1-DN and the highly active sequences ( Figure S3), we discovered that RNA8 showed the best binding affinity (K D = 920 nM)a nd 10-fold fluorescence enhancement;t herefore,i tw as selected for further investigation.
Tr uncation of RNA8 and the stabilization of the terminal helix by inserting an extra G-C pair yielded am inimal 38-nt RNA8-2 aptamer,d ubbed as BeCA ( Figure 3A and Figure S4). BeCA binds BC1-DN with a K D of 230 nM, and displays a9 .7-fold fluorescence turn-on ( Figure 3B,C). Further truncations or deletion of the internal hairpin structure caused as ignificant loss of fluorogenicity ( Figure S4). Sequence alignment and secondary structure analysis of the aptamers with the highest turn-on values revealed that sequences GUGG and AGGAA in the main loop are conserved ( Figure 3A,F igure S5A). Additional mutational studies did not considerably improve the affinity and fluorescence turn-on of BeCA ( Figure S5B).
Fine-Tuning Spirocyclization of BC. The BeCA aptamer was designed and evolved to selectively bind BC1,w hich mainly exists in the zwitterionic form. To obtain ac olorshifting ligand for BeCA,weaimed to increase the propensity of BC1 to form the corresponding spirocyclic form in the unbound state by chemically altering its structure.H owever, these modifications should be minimal so that the specific interactions between BeCA and BC1 are not negatively affected. Forthis purpose,wetook two different approaches. Thefirst approach was based on the introduction of electronwithdrawing fluorine atoms to the aromatic groups of benzopyrylium and coumarin ( Figure 4). Decreasing the electron density in the p-conjugated system allows more efficient nucleophilic attack by the carboxyl group to form the spirolactone. [23] Thesecond approach was,however, based on increasing the nucleophilicity of the carboxyl group without modifying benzopyrylium and coumarin. Thec onversion of the carboxyl group to electron-deficient amides has recently been reported to shift rhodamines equilibrium to the spirocyclic configuration. [18a, 20d] Hence,e lectron-withdrawing fluorine atoms (BC 2-4)a nd electron-deficient amides (BC Exchanging the carboxylic acid group with an amide group (BC5)y ielded af luorophore with dramatically increased D 50 (@ 70), implying that BC5 is almost completely in the cyan fluorescent spirolactam form in water.However,the BC derivatives carrying cyanamide (BC6)a nd methylsulfonamide (BC7)h ave D 50 values of 51 and 62, respectively, lower than that of BC5 due to the decreased nucleophilicity of electron-deficient amides.Y et, both BC6 and BC7 displayed much higher D 50 values than BC1.
Overall, minimal chemical modifications on the BC1 scaffold enabled fine-tuning of the equilibrium between spirocyclic and zwitterionic forms as demonstrated by the increased D 50 values,r anging from 23 to > 70. As ar esult, fluorophores BC 4-7 hold potential to function as colorshifting ligands for the BeCA aptamer in imaging applications.
Characterization of aC olor-Shifting Aptamer-Fluorophore Module. Thei deal color-shifting BC fluorophore for BeCA should be cyan fluorescent in solution but emit in the NIR region upon binding to the aptamer.T herefore,t he binding of BeCA to the ideal fluorophore would decrease the cyan fluorescence,increase the NIR fluorescence,and thereby increase the NIR/cyan ratio.F irst, we evaluated whether the chemically modified BC 4-7 analogs bind BeCA and cause afluorescence increase in NIR emission and adecrease in cyan fluorescence (Figure 4). Among these dyes, BC6 showed the highest fluorescence enhancement of 8.2-fold upon aptamer binding. As indicated by the D 50 value, BC6 displayed its maximum absorbance in the cyan region and existed mainly in the spirocyclicform ( Figure S6). Therefore, BC6 was chosen as the color-shifting ligand for BeCA. Furthermore,t he cyan and NIR fluorescence intensities of BC6 did not significantly change in the presence of total RNA, supporting the specific interaction between BC6 and BeCA ( Figure S7).
Next, we examined the photophysical features of BC6 in the presence of BeCA (Figure 5a nd Table S1). Titration of BC6 with the BeCA aptamer led to ad ecrease in the absorbance peak of the spirocyclicf orm (l = 420 nm), while an increase in the absorbance peak of the zwitterionic form (l = 665 nm), indicating the propensity of BC6 to favor the zwitterionic form when bound to BeCA.C onsistent with the absorbance change,t he cyan fluorescence (l = 478 nm) dropped while the NIR fluorescence (l = 684 nm) increased with increasing concentration of BeCA.T he color-shifting process of BC6,f rom unbound to BeCA-bound state, produced adynamic range (cyan/NIR emission ratio change) as high as 15-fold, substantially higher than the singlewavelength NIR fluorescence turn-on ( Figure 5).
BeCA binds the ligand BC6 with a K D of 220 nM ( Figure S8A), displaying essentially the same affinity as BC1-DN.N otably,t he fluorescence of BeCA-BC6 is independent of magnesium and potassium ions:9 0% of the maximum fluorescence was retained even in the absence of either of these cations,s uggesting that the folding of BeCA and its complexation with BC1 are independent of magnesium and potassium ( Figure S8B,C). An increase in temper- ature from 25 to 37 8 8Ccaused adecrease in the fluorescence of BeCA-BC6 by only 25 %, comparable to other literaturereported, thermally stable aptamer systems [8a,24] (Figure S8D).
BeCA as aN IR Light-up Aptamer for RNAI maging in Live Cells. First, we analyzed the utility of BeCA-BC6 as aN IR fluorescence turn-on tag for imaging mRNAs,w hich generally possess short half-lives,c omplex structures and have low abundance.T othis end, eight synonymous copies of BeCA without any stabilizing scaffold were introduced into the 3'-untranslated region of the green fluorescent protein (gfp)g ene. Escherichia coli (E. coli)e xpressing either gfp (control) or gfp-BeCA 8 mRNAwere imaged in the presence of BC6 (Figure 6and Figure S9). Thebright NIR fluorescence emission was detected predominantly at the poles of bacteria expressing gfp-BeCA 8 mRNA, but not in the control cells.The accumulation of mRNAa tt he bacterial poles demonstrated by the NIR signal was likely due to localization of the pET plasmid where mRNAt ranscription initiates. [24,25] Thea verage NIR fluorescence intensity at the bacterial poles of gfp-BeCA 8 -expressing cells was 6-fold higher than that of the control bacteria when 200 nM of BC6 was used. Furthermore, bacteria expressing gfp or gfp-BeCA 8 displayed similar GFP fluorescence intensities,suggesting that the BeCA tag did not significantly affect the transcription and translation processes.
BeCA was also expressed as synonymous repeats in mammalian cells and successfully visualized using the fluorophore BC6 in HEK293T cells ( Figure S10). This experiment demonstrated that BeCA folds correctly in mammalian cells and could be used to image other target RNAs utilizing its light-up feature and NIR fluorescence.
BeCA for Ratiometric RNAimaging in Live Bacteria. To investigate whether the BeCA-BC6 color-shifting aptamerfluorophore module could be used for ratiometric imaging of RNAs,proof-of-principle live-cell imaging experiments were performed. We expressed BeCA embedded in at RNA scaffold in E. coli and imaged bacteria in the presence of BC6 (500 nM) ( Figure S11). BeCA-expressing cells showed amuch higher fluorescence signal in the NIR channel than the tRNA-expressing control cells,i ndicating that BeCA selectively binds the BC6 zwitterion and lights up in the NIR channel. On the other hand, bacteria, whether expressing BeCA or not, showed similar cyan fluorescence due to the dynamic equilibrium between the intracellular and extracellular BC6.Increasing the concentration of BC6 from 500 nM to 1 mMinbacterial imaging did not significantly improve the NIR signal in bacteria expressing BeCA,y et it resulted in elevated NIR background fluorescence in control bacteria. As expected, using ah igher concentration of BC6 increased the cyan fluorescence in both BeCA-expressing and control bacteria. Even in the presence of as low as 200 nM of BC6, both cyan and NIR channel images of BeCA-expressing cells exhibited excellent image quality.
Next, we tested if we can detect variations in cyan and NIR emission intensities of BeCA-BC6 upon environmental changes such as exposure to high salt concentration. Upon addition of ammonium acetate to the BeCA-BC6 complex in vitro,w hich can replace mono and divalent cations in the aptamer,w eo bserved an increase in the cyan fluorescence (1.7-fold) and decrease in both NIR fluorescence (4.1-fold) and the ratio of NIR/cyan fluorescence (6.8-fold) intensities. This can be explained by unfolding of the aptamer followed by the dissociation of the dye from the aptamer (Figure S12A). Motivated by this result, we performed as imilar experiment with BeCA-expressing live bacteria that had already been incubated with BC6.After arapid treatment of the bacteria with ammonium acetate,c onfocal images indicated af luorescence increase in the cyan channel and adecrease in both NIR and ratiometric (NIR/cyan) channels as anticipated ( Figure S12B,C). Ratiometric images were obtained by pixel-to-pixel division of the fluorescence intensities in the NIR channel by the cyan channel. This experiment allowed us to visualize the unfolding of the BeCA aptamer in live bacteria exploiting the ratiometric features of BeCA-BC6.
We also used the color-shifting capability of BeCA-BC6 for the analysis of RNAe xpression levels in living cells.T he  transcription of BeCA can be effectively activated by isopropyl-b-d-thiogalactopyranoside (IPTG). [26] Cells expressing BeCA were treated with varying concentrations of IPTG,a nd imaged in both cyan and NIR channels simultaneously ( Figure 7A). Them ean NIR/cyan ratios of bacteria treated with 20, 50, 100, and 1000 mMo fI PTG were determined to be,r espectively,1 .7, 2.4, 3.0 and 3.7 fold higher than that of bacteria treated with 10 mMo fI PTG ( Figure 7B). These numbers correlate well with in vitro quantification of BeCA in total RNAs isolated from the bacteria treated with different concentrations of IPTG (Figure 7C and Figure S13). This result clearly shows that the ratiometric images obtained by our color-shifting FLAP system can be utilized to evaluate the expression levels of RNAt ranscripts.
Besides the determination of the intracellular concentration of ap articular ROIa tt he population (bulk) level, the ratiometric images obtained from the BeCA-BC6 system could provide more precise information at the single-cell level. As shown in Figure 7D,t he amounts of RNAt ranscripts in single bacteria (cells a and b)donot always correlate well with the NIR fluorescence intensities.T his might be due to variations in intracellular probe concentration, cell thickness and focal plane.Owing to the dual-fluorescence nature of BeCA-BC6,wecan use the cyan channel to correct for these discrepancies.I ndeed, the ratiometric signals suggest that cells a and b in Figure 7D had similar levels of intracellular BeCA transcripts.E xtensive heterogeneity of RNAt ranscription at the single-cell level in E. coli due to stochasticity in transcriptional regulation could also be revealed in the ratiometric images especially at high IPTG concentration ( Figure 7D,cells c-e). [27] Conclusion In summary,w eh ave presented the development of acolor-shifting NIR fluorescent aptamer-fluorophore module BeCA-BC6 for live-cell RNAi maging. To achieve that, we exploited the intramolecular spirocyclization of an environmentally sensitive hybrid fluorophore, BC.I te xists in ad ynamic equilibrium between ac yan-fluorescent, spirocyclic, closed form and aN IR-fluorescent, zwitterionic, open form. In vitro selection, truncation, and mutation studies rendered a3 8-nt minimal aptamer BeCA,w hich binds selectively to the zwitterionic form of BC.B yi ntroducing electron-withdrawing fluorine atoms and electron-deficient amine groups to BC,weobtained aseries of BC analogs with various open-closed ratios.T he best probe BC6 exists primarily in the closed state with an emission maximum of 478 nm and emits at 684 nm when bound to BeCA,r epresenting the most NIR-shifted FLAP in the literature.T hus, BeCA-BC6 is av aluable addition to the RNAi maging toolbox due to the lack of FLAPs functioning in the NIR window where cells have much lower absorption (less phototoxicity), lower auto-fluorescence and deeper penetration.
Moreover,wedemonstrated that BC6 showed an emission ratio change (cyan/NIR) as high as 15-fold upon binding to BeCA.I nl ive-cell RNAi maging experiments,t he cyan fluorescence from unbound probes reveals the intracellular probe delivery and its distribution while the NIR fluorescence indicates the RNAlocation. BeCA-BC6 is the only aptamerfluorophore pair which allows simultaneous imaging of both free fluorophore and the complex. We used this feature to obtain ratiometric images of bacteria with different expression levels of BeCA.Ratiometric images,incontrast to singlecolor fluorescence images,d on ot suffer from problems associated with varying dye uptake,h eterogeneous probe distribution, probe instability,c ell morphology and fluctuations in focal plane.T hus,t he dual-color feature of BeCA-BC6 allowed us to more accurately analyze the expression levels of RNAt ranscripts and revealed the transcriptional heterogeneity at the single-cell level.
We also showed that multiple repeats of BeCA can be genetically fused to more complex and short-lived mRNAs to increase the signal-to-background ratio.C ombined with the ratiometric advantages, BeCA-BC6 can be used to precisely analyze the abundance of target RNAs,w hich could provide new insights in gene expression, regulation, developmental plasticity and disease diagnostics. [28] Thes pirocyclicp robes including BC6 have excellent membrane permeability compared to their zwitterionic counterparts. [20d] Further,s pirocyclicp robes are less likely to be switched to their zwitterionic forms by other cellular components,t hus resulting in less unspecific NIR staining inside the cells.A sd emonstrated in this study,a ptamer binding can significantly change their biophysical properties and modulate equilibrium dynamics.I nt he future,b y employing library mutagenesis and fluorescence activated cell sorting (FACS), it would be possible to evolve tailored color-shifting fluorophore-aptamer pairs with various colors, higher brightness,b etter affinity and improved thermal stability.