Post-transcriptional labeling by using Suzuki–Miyaura cross-coupling generates functional RNA probes

Abstract Pd-catalyzed C-C bond formation, an important vertebra in the spine of synthetic chemistry, is emerging as a valuable chemoselective transformation for post-synthetic functionalization of biomacromolecules. While methods are available for labeling protein and DNA, development of an analogous procedure to label RNA by cross-coupling reactions remains a major challenge. Herein, we describe a new Pd-mediated RNA oligonucleotide (ON) labeling method that involves post-transcriptional functionalization of iodouridine-labeled RNA transcripts by using Suzuki–Miyaura cross-coupling reaction. 5-Iodouridine triphosphate (IUTP) is efficiently incorporated into RNA ONs at one or more sites by T7 RNA polymerase. Further, using a catalytic system made of Pd(OAc)2 and 2-aminopyrimidine-4,6-diol (ADHP) or dimethylamino-substituted ADHP (DMADHP), we established a modular method to functionalize iodouridine-labeled RNA ONs in the presence of various boronic acid and ester substrates under very mild conditions (37°C and pH 8.5). This method is highly chemoselective, and offers direct access to RNA ONs labeled with commonly used fluorescent and affinity tags and new fluorogenic environment-sensitive nucleoside probes in a ligand-controlled stereoselective fashion. Taken together, this simple approach of generating functional RNA ON probes by Suzuki–Miyaura coupling will be a very important addition to the resources and tools available for analyzing RNA motifs.


INTRODUCTION
Understanding of RNA structure and function, and its use in therapeutics are greatly aided by recent developments in the nucleic acid functionalization strategy based on bioorthogonal chemical reactions (1)(2)(3)(4). Traditional approaches like solid-phase ON synthesis and enzymatic methods are very useful in installing variety of probes onto RNA for various biophysical investigations. However, in several instances, elaborate chemical manipulations to synthesize the functionalized monomers (e.g. phosphoramidites and triphosphates) and challenges associated with their incorporation (e.g. stability under reaction conditions, poor coupling and enzymatic incorporation efficiency) limit the applications of these methods (4). In this context, postsynthetic modification of RNA by using bioorthogonal reactions is proving as a valuable tool to generate functional RNA probes. In this method, an RNA ON is labeled with a small reactive handle by using solid-phase ON synthesis protocol or by using the substrate promiscuity of RNA polymerases and certain RNA processing enzymes (e.g. transferases, [4][5][6]. Following this step, a chemoselective reaction with the cognate reactive partner is performed to introduce the desired functional modification into the RNA. Reactions like azide-alkyne cycloaddition (7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22), Staudinger ligation (21), inverse electron demand Diels-Alder (23)(24)(25)(26)(27)(28), to name a few, have emerged as valuable tools to label, image and profile RNA in cell-free and cellular environments. These methods often use bulky activated building blocks (e.g. cyclooctyne, tetrazine, norbornyl etc.) to promote efficient post-synthetic reaction under mild conditions. However, the synthesis of many of these building blocks is tedious involving multiple steps, and if commercially available are very expensive (29)(30)(31). Therefore, establishment of new post-synthetic RNA modification strategies that allow direct introduction of various functionalities by using easily accessible tags and reporters remains a high priority.
In this regard, Pd-mediated C-C bond formation, which is applied in almost all facets of chemistry, is proving useful as a valuable chemoselective transformation for synthetic modification of biomacromolecules (32)(33)(34)(35)(36). This has been possible due to the development of new Pd-ligand catalytic systems, which appreciably accelerate the coupling reaction in aqueous buffer (37)(38)(39). Manderville group first demonstrated the usefulness of Suzuki-Miyaura reaction in the post-synthetic functionalization of DNA oligonu- cleotides (ON, (40)). ONs containing 8-bromoguanosine were reacted with arylboronic acids in the presence of a catalytic system made of Pd(OAc) 2 and a water-soluble triphenylphosphan-3,3 ,3 -trisulfonate ligand, which was used previously for nucleotide modification (41,42). Using a similar method, a diarylethene photoswitch capable of undergoing reversible electrocyclic rearrangement was introduced into DNA ONs (43). This catalytic system requires elevated temperature (>70 • C), long reaction time and alkaline conditions to generate coupled ON products in moderate yields. Meantime, Davis group used a combination of Pd(OAc) 2 and 2-aminopyrimidine-4,6-diol (ADHP) or dimethylamino-substituted ADHP (DMADHP), which was originally developed for labeling proteins by Suzuki and Sonogashira reactions (44)(45)(46), to step up a milder route to directly install functional labels onto DNA ONs by using Suzuki-Miyaura reaction (47).
Despite these successes with protein and DNA, functionalization of RNA by Pd-mediated coupling reactions remains a major challenge as methods developed for protein and DNA mostly do not work for RNA due to inherently low stability of RNA (28,(48)(49). Therefore, we embarked on establishing a milder and efficient method to modify RNA by first incorporating a halogenated nucleotide analog into RNA by transcription reaction, followed by a posttranscriptional Suzuki-Miyaura reaction in the presence of a cognate reactive partner labeled with a desired biophysical reporter or tag ( Figure 1). Here, we demonstrate a posttranscriptional modification method to generate functional RNA probes by using Suzuki-Miyaura reaction under benign conditions (37 • C and pH 8.5). This method is modular, and offers direct access to RNA labeled with fluorogenic environment-sensitive nucleoside analogs for nucleic acid structure and recognition analysis, fluorescent probes for microscopy and an affinity tag for pull-down and immunoassay ( Figure 1).

MATERIALS AND METHODS
Experimental procedure for the synthesis of 5-Iodouridine triphosphate (IUTP) and boronic ester substrates are provided in the Supplementary Data.

Incorporation of IUTP by in vitro transcription reaction
Radiolabel experiment. The promoter-template duplexes (5 M) were assembled by heating a 1:1 mixture of DNA promoter of T7 RNA polymerase consensus sequence and DNA ON templates T1-T5 in annealing buffer (10 mM Tris-HCl, 1 mM ethylenediaminetetraacetic acid (EDTA), 100 mM NaCl, pH 7.8) at 90 • C for 3 min. The solution was allowed to attain room temperature slowly and then kept in an ice bath for 20 min followed by storing at −40 • C. The transcription reactions were carried out at 37 • C in 40 mM Tris-HCl buffer (pH 7.8) containing 250 nM annealed promoter-template duplexes, 10 mM MgCl 2 , 10 mM NaCl, 10 mM of dithiothreitol (DTT), 2 mM spermidine, 1 U/l RNase inhibitor (Riboblock), 1 mM guanosine triphosphate (GTP), cytidine triphosphate (CTP), uridine triphosphate (UTP) and or IUTP 2, 20 M adenosine triphosphate (ATP), 5 Ci ␣-32 P ATP and 3 U/l T7 RNA polymerase in a 20 l reaction volume. The reaction was quenched after 3.5 h by adding 20 l of loading buffer (7 M urea in 10 mM Tris-HCl, 100 mM EDTA, 0.05% bromophenol blue, pH 8). Each sample was heated for 3 min at 75 • C and then cooled in an ice bath. The samples (4 l) were loaded on an 18% denaturing polyacrylamide gel and were electrophoresed. The radioactive bands were phosphorimaged and then quantified by using GeneTools software from Syngene to determine the relative transcription efficiencies. Percentage incorporation of IUTP 2 is reported with respect to transcription efficiency in the presence of natural NTPs. All reactions were performed in duplicate and the errors in yields were <2%.
Large-scale transcription reaction using template T1 and IUTP 2. Transcription reaction was performed in 250 l reaction volume using 2 mM of ATP, GTP, CTP and IUTP 2, 20 mM MgCl 2 , 10 mM DTT, 0.40 U/l RNase inhibitor (Riboblock), 300 nM promoter-template duplex and 800 units of T7 RNA polymerase. The reaction mixture was incubated at 37 • C for 6 h. The reaction volume was made one-third by speed vac followed by addition of 50 l denaturing loading buffer (7 M urea in 10 mM Tris-HCl, 100 mM EDTA, pH 8) and the sample was loaded on a preparative 20% denaturing polyacrylamide gel. After running the gel for 6 h, appropriate band was marked by UV shadowing and excised from the gel. In order to isolate the RNA, the band was extracted with 0.3 M sodium acetate and desalted using a Sep-Pak classic C18 cartridge. Under these conditions, an average of 10 nmole of the iodo-modified RNA transcript 4 was isolated (ε 260 = 84300 M −1 cm −1 ). The purity and identity of the iodo-labeled RNA transcript was confirmed by matrix-assisted laser desorption/ionizationtime of flight (MALDI-TOF) mass analysis (Supplementary Figure S2). mmol) was added and the mixture was stirred at 65 • C for 1 h. Solution was taken in a 5 ml volumetric flask and the volume was adjusted to 5 ml with autoclaved H 2 O to give a final stock solution of 50 mM.

Suzuki-Miyaura reaction between transcript 4 and boronic
Analytical-scale reaction with boronic ester 9 and 10. To a solution of iodo-modified RNA transcript 4 (200 M, 1 equivalent) in 10 l of Tris-HCl buffer (50 mM, pH 8.5) was added boronic ester 9 or 10 (50 or 100 equivalent) dissolved in dimethyl sulfoxide (DMSO). The coupling reaction was initiated by adding Pd(OAc) 2 (L1) 2 catalyst (1 or 2 equivalent). Final reaction volume was 50 l containing 20% DMSO (v/v). The reaction mixture was incubated at 37 • C in thermoshaker. Aliquots of reaction mixture (16 l) were taken at 3, 6 and 9 h, and 10 l of denaturing loading buffer (7 M urea in 10 mM Tris-HCl, 100 mM EDTA, pH 8) was added to each aliquot. Samples were loaded on an analytical 20% denaturing polyacrylamide gel. Bands corresponding to the products were visualized by UV-shadowing method (short wave UV 254 nm and long wave UV 365 nm).
Large-scale reaction. To a solution of iodo-modified RNA transcript 4 (5 nmol, 100 M, 1 equivalent) in 10 l of Tris-HCl buffer (50 mM, pH 8.5) was added boronic acid/ester dissolved in DMSO (250 nmol, 5 mM, 50 equivalent). The reaction was initiated by adding catalyst Pd(OAc) 2 (L) 2 (10 nmol, 0.2 mM, 2 equivalent). The final reaction volume was adjusted to 50 l with water and DMSO such that the percentage of DMSO was 20% v/v. The reaction was incubated at 37 • C for 6−12 h. The reaction mixture was filtered using a spin filter (0.45 m pore size) and was further washed with 40 l of water.
In order to setup an RNA functionalization method by Suzuki-Miyaura reaction, we chose to incorporate 5iodouridine 5 -triphosphate (IUTP 2) into RNA transcripts by in vitro transcription reaction (65,66). Iodo-modified nucleoside phosphoramidites can also be used to incorporate the halogen label into RNA ONs by solid-phase method. IUTP was prepared by phosphorylating IU (1) using POCl 3 and bis-tributylammonium pyrophosphate (Figure 2). The efficiency of IUTP incorporation by bacteriophage T7 polymerase was evaluated by performing in vitro transcription reactions with a series of T7 promoter-template DNA duplexes ( Figure 2). The templates were designed to guide the incorporation of monophosphate of IUTP into RNA at one or two sites. The templates also contained a single dT residue at the 5 -end of the coding region so that a reaction performed in the presence of UTP/IUTP, GTP, CTP and ␣-32 P ATP, if successful, would result in the formation of the full-length transcripts containing a radioactive ␣-32 P A label at the 3 -end.
Reactions performed with template T1 and UTP/IUTP produced full-length transcripts 3 and 4, respectively, with excellent efficiency and comparable yields (98%, Figure 3, lanes 1 and 2). Slower mobility of 4 compared to 3 indicated the incorporation of modified U into transcript 4. The labeling of IU in the full-length transcript was confirmed by mass measurement of the purified transcript prepared from a large-scale reaction (Supplementary Figure  S2). A control reaction in the absence of UTP and IUTP did not yield full-length transcript, indicating that there was no misincorporation during the transcription process (lane 3). Interestingly, in a reaction containing 1:1 molar ratio of UTP and IUTP, the RNA polymerase preferentially incorporated IUTP over UTP (lane 4). Reactions with other templates (T2−T5) indicated that IU can be introduced near the promoter region and at more than one site with very good efficiency (lanes 5−12).

Post-transcriptional Suzuki-Miyaura cross-coupling
Optimization of coupling reaction conditions. IU-labeled RNA ON 4 was subjected to Suzuki-Miyaura crosscoupling at different stoichiometries of the substrate and reagents so as to achieve good conversion with minimum degradation of the coupled product. We preferred to use a combination of Pd(OAc) 2 and ADHP (L1) or DMADHP (L2) as this system has been shown to be efficient in the Pdmediated functionalization of protein and DNA (44)(45)(46)(47). Coupling reaction was performed by incubating 4 (1 equivalent) with Pd(OAc) 2 L1 2 (1 equivalent) and nitrobenzofurazan (NBD)-labeled boronic ester 9/10 (commonly used dye in fluorescence imaging, 100 equivalent) in Tris-HCl buffer (50 mM, pH 8.5) at 37 • C (Figures 4 and 5). Aliquots of reaction mixture after 3, 6 and 9 h were resolved by analytical polyacrylamide gel electrophoresis under denaturing conditions, and analyzed by UV-shadowing. Rewardingly, reactions with boronic esters 9 and 10 resulted in the formation of respective coupled RNA product, which mi-  grated slower compared to the substrate 4 ( Figure 6A). A reaction with NBD dye (10) attached to boronic ester via a longer linker showed almost complete consumption of the substrate in 6 h as compared to 9, which remained partially consumed after 9 h. When the amount of boronic ester 10 was reduced to 50 equivalents, the reaction required a slightly higher loading of Pd-L1 (2 equivalent) to effect the coupling in 6 h ( Figure 6B). UV-shadowing the gel at longer wavelength (∼365 nm) further confirmed the fluorescence labeling of RNA with NBD. Importantly, under these conditions we observed no detectable degradation of the RNA product.
Post-transcriptional coupling of RNA with fluorescent and affinity tags. Based on these preliminary results, transcript 4 (5 nmol, 1 equivalent) was then subjected to posttranscriptional coupling reaction with boronic esters 9−11 (50 equivalent) using Pd-L1/L2 (2 equivalent) catalytic systems. The reaction was monitored by High performance liquid chromatography (HPLC), and the peak corresponding to the product was isolated and characterized by mass analysis (Supplementary Figure S3 and Table S1). While UVshadowing of the gel gave a qualitative understanding of the coupling reaction, HPLC analysis gave a better understanding of the reaction in terms of efficiency and isolated yields ( Table 1). A reaction with NBD-boronic ester 9, containing a short linker, in the presence of L1 gave 28% of the fluorescent RNA product 9a in 12 h. However, a reaction with NBD-boronic ester 10, containing a longer linker, afforded 30% of the fluorescent RNA product 10a within 6 h. Under similar conditions, biotin-coupled RNA product 11a was isolated in 28% yield from a reaction with biotinylated boronic ester substrate 11. HPLC chromatogram of reactions with substrates 9−11 revealed the formation of noticeable amount of deiodinated transcript along with unreacted transcript 4, and hence, longer reaction times were not attempted with these substrates (Supplementary Figure S3). Reactions using Pd-L2 system for above substrates were found to be less efficient as compared to Pd-L1 combination (Table 1).

Post-transcriptional coupling generates RNA labeled with fluorogenic environment-sensitive nucleoside analogs.
Several examples including the ones reported from our laboratory indicate that responsive fluorescent nucleoside analogs can be assembled by conjugating heterocyclic rings onto nucleobases (67)(68)(69)(70)(71)(72)(73)(74)(75). Such nucleosides incorporated into ONs by chemical or enzymatic means serve as excellent probes for studying nucleic acid structure and function, and in diagnostic applications. These strategies usually use modified nucleoside phosphoramidites or triphosphates, which involve elaborate chemical manipulations. Moreover, in several instances, the substrates (i) show poor coupling efficiency, (ii) do not survive the conditions used in the solidphase protocols and (iii) are not efficiently incorporated by polymerases (4). In this context, post-transcriptional coupling of IU-labeled RNA ONs with heterocycle-containing boronic acids/esters should provide direct access to RNA functionalized with responsive nucleoside probes. This approach will avoid cumbersome synthesis and challenges in the incorporation of individual modified amidites and nucleotides.
Next, we sought to use this labeling approach to introduce new fluorescent nucleoside modifications into RNA ONs. Although predicting the fluorescence outcome based on the structure is difficult, we envisioned that coupling heterocycles onto nucleosides via an extended system may impart interesting photophysical features to otherwise nonemissive nucleosides (81,82). In this regard, we chose to couple easily synthesizable heteroarylvinyl boronic esters 16 and 17 with IU-labeled RNA ON 4 ( Figure 4). Rewardingly, these substrates underwent facile coupling reaction to produce RNA ONs containing 5-(benzothiophen-2-yl)vinyl uridine and 5-(benzofuran-2-yl)vinyl uridine, respectively, in good yields ( Figure 5, Table 1 and Supplementary Table  S1). The HPLC chromatogram of the reaction revealed the presence of major (designated as 16a and 17a ) and minor (designated as 16a and 17a ) peaks having the same mass; probably corresponding to trans and cis isomers, respectively (Supplementary Figure S5, Figure S6 and Table S1). In order to ascertain the isomeric identity of the products formed, the major peak (17a ) was isolated and digested using a cocktail of enzymes, which would generate individual ribonucleosides. The HPLC chromatogram of the digest revealed the presence of native ribonucleosides (C, G and A) and trans form of 5-(benzofuran-2-yl)vinyl uridine, which matched well with the retention time of the authentic trans isomer (18) obtained from a control reaction between IU and ester 17 (Supplementary Figure S7). Remarkably, reaction in the presence of Pd-L2 catalytic system yielded only the trans products 16a and 17a , suggesting that this post-transcriptional modification is an example of a ligandcontrolled stereoselective alkenylation process (Table 1 and Supplementary Figure S5).
A 5 nmole reaction scale gave 1.4−4 nmole of the coupled RNA product, which is sufficient for subsequent biophysical analysis (Table 1). Similar yields are reported for postsynthetic click functionalization of RNA ONs (12,17,21). It is worth mentioning here that dehalogenation is a common side reaction in Pd-catalyzed cross-coupling reactions, including Suzuki-Miyaura reaction. The extent of deiodination of the substrate, and hence, the reduction in the reaction efficiency is known to vary with the reaction temperature, nature of the catalyst-ligand system and loading,  Table S1 for mass spectra and data. b ε 260 of coupled RNA ON products was determined by using OligoAnalyzer 3.1. In case of 9a−11a, ε 260 of 5-vinyluridine (28)   nature of boronic acid/ester substrates, solvent/buffer conditions and reaction time (40,43,47). Hence, for a given boronic acid/ester substrate the deiodination can be potentially minimized by optimizing the reaction conditions in terms of catalyst loading and reaction time. In case more amount of the RNA product is required then a batch reaction is recommended, which can be pooled before purification, thereby avoiding multiple HPLC runs (data not shown). Control reactions with ON 4 in the presence of boronic acids/esters or Pd(OAc) 2 or Pd-ligand alone did not yield the coupled product (Supplementary Figure S8). Similarly, an unmodified RNA ON 3 did not react under the coupling reaction conditions. These results indicate that Suzuki coupling on IU-labeled RNA is highly chemoselective. The ground-state electronic spectrum of 18 was not significantly affected by changes in solvent polarity. However, the nucleoside exhibited excellent fluorescence solvatochromism, wherein the emission maximum, Stokes shift, intensity and quantum yield were significantly influenced by changes in the polarity of the medium (Supplementary Figure S9 and Table S3). In water, 18 displayed a weak fluorescence band (λ em = 483 nm) corresponding to a quantum yield of 0.028. As the polarity of the medium was decreased from water to methanol to dioxane, a significant enhancement in fluorescence efficiency (5-fold) accompanied by a blue-shifted emission maximum was observed (λ em = 427 nm in dioxane). Encouraged by these results, we next sought to study the responsiveness of the nucleoside analog to changes in neighboring base environment.
RNA ONs 17a , 19a and 20a were hybridized with DNA ONs such that the emissive analog 18 was paired opposite to complementary or mismatched bases ( Figure 8A). Typical of a responsive nucleoside probe, the emission maximum and intensity of the nucleoside were found to be sensitive to neighboring base environment (Figure 8). 18 incorporated into single stranded RNA ONs and then into duplexes (matched or mismatched) showed a progressing increase in fluorescence intensity as compared to the free nucleoside analog. Notably, the emission maximum of duplexes (∼455 nm) was significantly blue-shifted as compared to the nucleoside (483 nm). The enhancement in fluorescence and blueshifted emission indicate that the micropolarity around the emissive analog in duplexes is significantly lower than water. A comparison of emission maximum of the free nucleoside analog in different solvents and in duplexes suggest that the modification at the 5-position of uridine, which is projected in the major groove, experiences a polarity more close to methanol (Supplementary Table S3). This result is consistent with the major groove polarity of the duplexes reported in the literature (83,84). Further, among the duplexes, the emissive analog placed in the vicinity of guanine showed lower fluorescence intensity as compared to other bases, which is likely due to the known quenching effect of guanine by electron transfer process (85). Taken together, these results underscore the potential of post-synthetic Suzuki coupling reaction in directly accessing RNA ONs labeled with fluorogenic and environment-sensitive nucleosides.
Pd contamination in the labeled RNA product could potentially interfere with its application in cell-based and in vivo experiments due to the toxicity of Pd (86). In the Suzuki-based protein labeling method, Davis et al. observed loss in signal in the mass spectra due to non-specific coordination of Pd to the protein (87). This effect was significantly reduced by using 3-mercaptopropionic acid as a scavenger, which strongly binds to Pd as compared to the protein. However, when labeling DNA by Suzuki coupling such an effect was not observed (47). Similarly, in our reaction conditions and purification method, we did not observe loss of mass signal from Suzuki-labeled RNA ON products. Nevertheless, it is suggested that a Pd scavenger can be used when preparing labeled RNA ONs for in vivo and therapeutic applications.

CONCLUSION
We have established an efficient method to label RNA ONs with functional probes by conceiving post-transcriptional Suzuki-Miyaura cross-coupling reaction under biocompatible conditions. This direct RNA labeling method can be used to install commonly used biophysical reporters as well as generate RNA ONs labeled with new fluorogenic and environment-sensitive nucleoside probes in a ligandcontrolled stereoselective fashion. Our results demonstrate that this RNA bioconjugation approach based on Suzuki-Miyaura coupling is a very powerful tool, which will complement existing methods to functionalize and study RNA ON motifs.

SUPPLEMENTARY DATA
Supplementary Data are available at NAR Online.