Fluorogenic and Bioorthogonal Modification of RNA Using Photoclick Chemistry.

A bromoaryltetrazole-modified uridine was synthesized as a new RNA building block for bioorthogonal, light-activated and postsynthetic modification with commercially available fluorescent dyes. It allows “photoclick”-type modifications by irradiation with light (300 nm LED) at internal and terminal positions of presynthesized RNA with maleimide-conjugated fluorophores in good yields. The reaction was evidenced for three different dyes. During irradiation, the emission increases due to the formation of an intrinsically fluorescent pyrazoline moiety as photoclick product. The fluorogenecity of the photoclick reaction was significantly enhanced by energy transfer between the pyrazoline as the reaction product (poor emitter) and the photoclicked dye as the strong emitter. The RNA-dye conjugates show remarkable fluorescent properties, in particular an up to 9.4 fold increase of fluorescence, which are important for chemical biology and fluorescent imaging of RNA in cells.


Introduction
Bioorthogonal reactions are widely employed for the labeling of biomolecules, including proteins, carbohydrates and nucleic acids, with fluorescent markers and probes [1][2][3]. The widely used Cu(I)-catalyzed azide-alkyne cycloaddition (CuAAC) offers not only a broad substrate scope, but also high selectivity and fast reaction kinetics [4,5]. Although a huge variety of modifications of biomolecules by CuAAC have been reported based on Cu-chelating ligands [6], and even a few in living cells, the use of copper(I) is problematic because of its cytotoxicity, and therefore alternatives are needed [7][8][9]. Considering the reaction kinetics, both (i) the inverse electron-demand Diels-Alder reaction (iEDDA) between 1,2,4,5-tetrazines [10,11] or 1,2,4-triazines [12] and strained olefins or alkynes, and (ii) the photoclick reaction between in situ generated nitrile-imines and electron-deficient alkenes offer second order rate kinetics that are comparable to CuAAC [13,14].
The great advantage of the photoclick reaction compared to iEDDA reactions is the spatiotemporal control by the use of light to initiate the reaction [15,16]. An additional and important characteristic of the photoclick reaction is the intrinsic fluorogenecity of the reaction (Figure 1). Fluorogenecity is an increasingly important feature of bioorthogonal labeling strategies that allows to reduce undesired background fluorescence by using light-up probes [16]. This is particularly important if molecular imaging is performed not only with fixed cells but also with living cells, because excess dye labels cannot be simply washed away [17,18]. During irradiation, the emission increases due to the formation of an intrinsically fluorescent pyrazoline moiety as photoclick product. This work focuses on enhancing the fluorogenic properties of the photoclick reaction by energy transfer between the pyrazoline as the reaction product, which is a poor emitter, and the photoclicked dye as the strong emitter. Recently, our group reported such photoclick reactions on tetrazole-modified DNA, whereas this work focuses Figure 1. Principle of fluorogenic RNA photoclick labeling using the tetrazole-modified uridine (marked in blue, for building block structure see Scheme 1) as a photoreactive building block in singlestranded RNA and maleimide-conjugated dyes (maleimide marked in blue, dye in red). The energy transfer between the pyrazoline as a photoclick product and the attached dye yield a fluorogenic photoclick reaction.

General
All Reagents were purchased from commercial sources and used without further purification. The compounds were purified via silica gel column chromatography (230-400 mesh). The completion of the reactions was monitored via t.l.c. on silica gel F254-coated plates and under a 254 nm lamp. Nucleosides were stained with 3% H2SO4 in methanol followed by heating with a heat gun. 1 H and 13 C-NMR spectra were recorded either on a Bruker Ascend 500 MHz or Bruker Ascend 400 MHz spectrometer in DMSO-d6. Chemical shifts δ are given in ppm relative to tetramethylsilane, the spectra were referenced to the solvent peak (DMSO-d6, 2.50 ppm). Coupling constants J are given in Hz. For multiplicities, the following abbreviations are used: s (singlet), d (doublet), dd (doublet of doublets), ddd (doublet of doublets of doublets), t (triplet), dt (doublet of triplets), q (quartet), m (multiplet).
Oligonucleotide synthesis was performed on a H-6 synthesizer by K&A Laborgeräte (Schaafheim, Germany), the coupling time of the artificial building block was increased to 24 min. After cleavage, the oligonucleotides were purified on a semi-preparative reversed-phase HPLC system (RP-C18 column, A = NH4HCO3 buffer, B = acetonitrile). The purified oligonucleotide strands were quantified photometrically using a NanoDrop ND-1000 spectrometer and the extinction coefficient of the strands (ε260 = 165 807 L mol −1 cm −1 ). Photoclicked oligonucleotides were purified via illustra TM NAP-5 columns (GE Healthcare, Chicago, USA) to remove excess dyes. MS were measured on a Shimadzu Axima Confidence using 3-hydroxypicolinic acid as matrix substance. Irradiation experiments were performed in 10 mm quartz glass cuvettes and irradiated with 300 nm LEDs for defined time spans in 10 mM Na-Pi buffer, 250 mM NaCl with 2.5 μM RNA and 3.75 μM dyemaleimide conjugates at 20 °C. Spectroscopic measurements were carried out on a Cary 100 Scan UV/vis spectrometer by Varian and a Fluoromax-4 spectrofluorometer by Horiba Jobin-Yvon (Kyoto, Japan).
Synthesis of compound 2 Figure 1. Principle of fluorogenic RNA photoclick labeling using the tetrazole-modified uridine (marked in blue, for building block structure see Scheme 1) as a photoreactive building block in single-stranded RNA and maleimide-conjugated dyes (maleimide marked in blue, dye in red). The energy transfer between the pyrazoline as a photoclick product and the attached dye yield a fluorogenic photoclick reaction.

General
All Reagents were purchased from commercial sources and used without further purification. The compounds were purified via silica gel column chromatography (230-400 mesh). The completion of the reactions was monitored via t.l.c. on silica gel F 254-coated plates and under a 254 nm lamp. Nucleosides were stained with 3% H 2 SO 4 in methanol followed by heating with a heat gun. 1 H and 13 C-NMR spectra were recorded either on a Bruker Ascend 500 MHz or Bruker Ascend 400 MHz spectrometer in DMSO-d6. Chemical shifts δ are given in ppm relative to tetramethylsilane, the spectra were referenced to the solvent peak (DMSO-d6, 2.50 ppm). Coupling constants J are given in Hz. For multiplicities, the following abbreviations are used: s (singlet), d (doublet), dd (doublet of doublets), ddd (doublet of doublets of doublets), t (triplet), dt (doublet of triplets), q (quartet), m (multiplet).
Oligonucleotide synthesis was performed on a H-6 synthesizer by K&A Laborgeräte (Schaafheim, Germany), the coupling time of the artificial building block was increased to 24 min. After cleavage, the oligonucleotides were purified on a semi-preparative reversed-phase HPLC system (RP-C18 column, A = NH 4 HCO 3 buffer, B = acetonitrile). The purified oligonucleotide strands were quantified photometrically using a NanoDrop ND-1000 spectrometer and the extinction coefficient of the strands (ε 260 = 165 807 L mol −1 cm −1 ). Photoclicked oligonucleotides were purified via illustra TM NAP-5 columns (GE Healthcare, Chicago, USA) to remove excess dyes. MS were measured on a Shimadzu Axima Confidence using 3-hydroxypicolinic acid as matrix substance. Irradiation experiments were performed in 10 mm quartz glass cuvettes and irradiated with 300 nm LEDs for defined time spans in 10 mM Na-Pi buffer, 250 mM NaCl with 2.5 µM RNA and 3.75 µM dye-maleimide conjugates at 20 • C. Spectroscopic measurements were carried out on a Cary 100 Scan UV/vis spectrometer by Varian and a Fluoromax-4 spectrofluorometer by Horiba Jobin-Yvon (Kyoto, Japan).

Results and Discussion
In principle, diaryltetrazoles are chemically stable but they are also bioorthogonally reactive groups when excited by UV-B light [21]. The RNA building block 9 bears the reactive tetrazole at its position 5. In order to keep the modified nucleotide as small as possible the uracil moiety replaces one of the aryl groups of conventional diaryltetrazoles. The bromo substituent at the opposite aryl group is needed for efficient photoclick chemistry [19] presumably by its heavy-atom effect on populating the photochemically active triplet state [22]. The synthetic route (Scheme 1) was adapted from the synthesis of a similar 2 -deoxyuridine building block for DNA recently published by our group [19] but contains improved synthetic procedures and a different protection scheme due to the presence of the 2 -hydroxy group. Accordingly, the 2 ,3 ,5 -tribenzoyl-protected 5-hydroxymethyluridine (2) was synthesized by coupling 5-hydroxymethyluracil that was prepared following published procedures [20] with the commercially available 1-acetate-2,3,5-tribenzoate of β-D-ribofuranoside using modified Vorbrüggen conditions in 70% yield [23]. The 5-hydroxymethyl group was oxidized by Dess-Martin periodinane in 72% yield [24]. The conversion of the aldehyde moiety of 3 to the hydrazone 4 was achieved in 82% yield by treatment with benzenesulfonyl hydrazide [25]. Upon addition of the commercially available 4-bromobenzenediazonium tetrafluoroborate, the tetrazole 5 was formed in 48% yield [25]. The next steps comprised the deprotection of all benzoyl groups of 5 by 7 M ammonia in methanol to obtain nucleoside 6, selective protection of the 5 -hydroxy group of 6 by the dimethoxytrityl group to get nucleoside 7, and protection of the 2 group of 7 with a tert-butyldimethyl silyl group. The latter reaction gave nucleoside 8 in 36% yield using silver nitrate as catalyst [26]. Finally, the hydroxy group at the 3 -position of 8 was phosphitylated by 2-cyanoethyl N,N-diisopropylchlorophosphoramidite to yield the final RNA phosphoramidite 9. The RNA building block 9 was then incorporated via solid-phase synthesis into two different RNA sequences with the tetrazole moiety either at an internal (RNA1) or at a 5 -terminal position (RNA2) of the sequences. The synthesized oligonucleotide strands were purified by semi-preparative HPLC, identified by MALDI-TOF mass spectrometry and quantified by their UV/vis absorbance at 260 nm.
As reaction partners for the photoclick labeling of RNA1 and RNA2, three commercially available maleimide-dye conjugates were chosen, including sulfo-Cy3 (Cy3), AlexaFluor555 (AF555) and AlexaFluor647 (AF647). During the 30 min reaction course, the UV/Vis absorbance and fluorescence changes were recorded (representatively shown for RNA1 and the AF555 modification in Figure 2, for the others see Figures S31-S41). The tetrazole absorbance is visible as tailing side band at 300 nm, which is at the border of the strong RNA absorbance. According to these measured UV/Vis spectra, 300 nm light from the appropriate LED was chosen to induce the photoclick reaction. The RNA building block 6 has an extinction coefficient of ε 300 =20.300 M −1 cm −1 which is significantly larger compared to the extinction coefficients of the natural RNA components A, C, G and U (ε 300 =60-260 M −1 cm −1 , see Supporting Information). Accordingly, in RNA1 and RNA2 the excitation selectivity for the tetrazole at 300 nm is approximately 10:1 that excludes an inner filter effect by the RNA. In the presence of 1.5 equiv. of the dye-maleimide conjugate, a new and broad band between 320 and 420 nm is formed, which can be assigned to the pyrazoline moiety formed in the RNA product, whereas the tetrazole absorbance between 280 and 320 nm concomitantly decreases. The UV/Vis spectra show distinct isosbestic points that indicate the clean conversions from the tetrazoles to the pyrazolines in all cases without the formation of a long-living intermediate. According to these UV/Vis absorbance changes, the reaction is finished after 30 min irradiation. As reaction partners for the photoclick labeling of RNA1 and RNA2, three commercially available maleimide-dye conjugates were chosen, including sulfo-Cy3 (Cy3), AlexaFluor555 (AF555) and AlexaFluor647 (AF647). During the 30 min reaction course, the UV/Vis absorbance and fluorescence changes were recorded (representatively shown for RNA1 and the AF555 modification in Figure 2, for the others see Figures S31-41). The tetrazole absorbance is visible as tailing side band presence of 1.5 equiv. of the dye-maleimide conjugate, a new and broad band between 320 and 420 nm is formed, which can be assigned to the pyrazoline moiety formed in the RNA product, whereas the tetrazole absorbance between 280 and 320 nm concomitantly decreases. The UV/Vis spectra show distinct isosbestic points that indicate the clean conversions from the tetrazoles to the pyrazolines in all cases without the formation of a long-living intermediate. According to these UV/Vis absorbance changes, the reaction is finished after 30 min irradiation. The oligonucleotide products were evidenced by MALDI-TOF mass spectrometric analysis ( Figures S42-S53). After chromatographic removal of excess and unreacted dyes (see SI for experimental details and Figures S54-S57) the sample concentrations were calculated based on their UV/Vis absorbances. The yields of the postsynthetic photoclick modifications were determined for each sample separately after the purification (The Cy3-conjugates show the lowest yields of 27% for RNA1 and 31% for RNA2, albeit its strong increase in fluorescence (vide infra). Exemplarily, the photoclick conjugation was repeated with 10.0 equiv. of the Cy3-maleimide and RNA2 to determine whether the yield could be further enhanced. In fact, the yield of this irradiation reaction was 70%, which of course caused a remarkably higher fluorescence. With respect to potential cell experiments, however, such an excess of dye might be critical due to the high background fluorescence of the dye. The highest yields of 78% for RNA1 and 84% for RNA2 could be achieved by reaction with the AF555 maleimide (1.50 equiv.), although these samples display the lowest absolute fluorescence (see Figures  S38 and S40 and Table 1). The Cy3-conjugates show the lowest yields of 27% for RNA1 and 31% for RNA2, albeit its strong increase in fluorescence (vide infra). Exemplarily, the photoclick conjugation was repeated with 10.0 equiv. of the Cy3-maleimide and RNA2 to determine whether the yield could be further enhanced. In fact, the yield of this irradiation reaction was 70%, which of course caused a remarkably higher fluorescence. With respect to potential cell experiments, however, such an excess of dye might be critical due to the high background fluorescence of the dye. The highest yields of 78% for RNA1 and 84% for RNA2 could be achieved by reaction with the AF555 maleimide (1.50 equiv.), although these samples display the lowest absolute fluorescence (see Figures S38 and S40). The oligonucleotide products were evidenced by MALDI-TOF mass spectrometric analysis ( Figures S42-S53). After chromatographic removal of excess and unreacted dyes (see SI for experimental details and Figures S54-S57) the sample concentrations were calculated based on their UV/Vis absorbances. The yields of the postsynthetic photoclick modifications were determined for each sample separately after the purification (The Cy3-conjugates show the lowest yields of 27% for RNA1 and 31% for RNA2, albeit its strong increase in fluorescence (vide infra). Exemplarily, the photoclick conjugation was repeated with 10.0 equiv. of the Cy3-maleimide and RNA2 to determine whether the yield could be further enhanced. In fact, the yield of this irradiation reaction was 70%, which of course caused a remarkably higher fluorescence. With respect to potential cell experiments, however, such an excess of dye might be critical due to the high background fluorescence of the dye. The highest yields of 78% for RNA1 and 84% for RNA2 could be achieved by reaction with the AF555 maleimide (1.50 equiv.), although these samples display the lowest absolute fluorescence (see Figures S38 and S40 and Table 1). The Cy3-conjugates show the lowest yields of 27% for RNA1 and 31% for RNA2, albeit its strong increase in fluorescence (vide infra). Exemplarily, the photoclick conjugation was repeated with 10.0 equiv. of the Cy3-maleimide and RNA2 to determine whether the yield could be further enhanced. In fact, the yield of this irradiation reaction was 70%, which of course caused a remarkably higher fluorescence. With respect to potential cell experiments, however, such an excess of dye might be critical due to the high background fluorescence of the dye. The highest yields of 78% for RNA1 and 84% for RNA2 could be achieved by reaction with the AF555 maleimide (1.50 equiv.), although these samples display the lowest absolute fluorescence (see Figures S38 and S40).
The excitation of the fluorescence was set to 358 nm, which is close to the absorption maximum of the pyrazoline moiety. The extinction of the applied dyes Cy3, AF647 and AF555 is very low at 358 nm, which nearly completely eliminates direct excitation of these dyes. Thus, the observed fluorescence can be assigned to the pyrazoline as direct photoclick product and the attached dyes that were coupled to the RNA and light up as a result of the energy transfer from the pyrazoline chromophore to these dyes. The recorded fluorescence spectra during the irradiations (Figure 3) show only minor increases of fluorescence intensity between 400 and 500 nm, which is the characteristic emission range of the pyrazoline chromophore, but strong increases of fluorescence intensity in the emission range of the attached dyes (with maxima at 564 nm for Cy3, 666 nm for AF647 and 565 nm for AF555). Based on the assumption that direct excitation of these dyes can be excluded, as mentioned above, this result clearly evidences the energy transfer between the pyrazoline and the dyes [19]. However, the gain of fluorescence intensity strongly depends on both the type of fluorophore (Cy3, AF647, AF655) and the position of the photoclick modification in the RNA strand (terminal, internal). In general, the terminally modified RNA2 displays a higher fluorogenicity than the internally modified sequence RNA1. In case of the Cy3 dye, remarkable 7.5-fold (RNA1) and 9.4-fold (RNA2) fluorescence intensity increases could be achieved. Even though Cy3 and AF555 are spectroscopically similar, the energy transfer between the pyrazoline moiety and Cy3 seems to be more efficient compared to AF555 because only 2.6-fold (RNA1) and 3.2-fold (RNA2) fluorescence intensity increases were yielded with AF555. Although structural details of both AF-maleimides are reported in literature [27], the exact structure was not provided by the manufacturer of the dyes. Therefore, the varying energy transfer efficiency cannot conclusively be explained, but we assume differing Förster radii or competing photochemical processes could be possible explanations for the lower efficiency. The lowest fluorescence gains were achieved with the AF647 dye. Table 1. Yields of the photoclick reaction. RNA1 or RNA2 (2.5 µM) were irradiated with 300 nm light (LED) in the presence of 1.50 equiv (3.75 µM) Cy3-maleimide, AF555-maleimide and AF647-maleimide, respectively, in 10 mM Na-P i buffer (250 mM NaCl, pH 7).
The excitation of the fluorescence was set to 358 nm, which is close to the absorption maximum of the pyrazoline moiety. The extinction of the applied dyes Cy3, AF647 and AF555 is very low at 358 nm, which nearly completely eliminates direct excitation of these dyes. Thus, the observed fluorescence can be assigned to the pyrazoline as direct photoclick product and the attached dyes that were coupled to the RNA and light up as a result of the energy transfer from the pyrazoline chromophore to these dyes. The recorded fluorescence spectra during the irradiations (Figure 3) show only minor increases of fluorescence intensity between 400 and 500 nm, which is the characteristic emission range of the pyrazoline chromophore, but strong increases of fluorescence intensity in the emission range of the attached dyes (with maxima at 564 nm for Cy3, 666 nm for AF647 and 565 nm for AF555). Based on the assumption that direct excitation of these dyes can be excluded, as mentioned above, this result clearly evidences the energy transfer between the pyrazoline and the dyes [19]. However, the gain of fluorescence intensity strongly depends on both the type of fluorophore (Cy3, AF647, AF655) and the position of the photoclick modification in the RNA strand (terminal, internal). In general, the terminally modified RNA2 displays a higher fluorogenicity than the internally modified sequence RNA1. In case of the Cy3 dye, remarkable 7.5fold (RNA1) and 9.4-fold (RNA2) fluorescence intensity increases could be achieved. Even though Cy3 and AF555 are spectroscopically similar, the energy transfer between the pyrazoline moiety and Cy3 seems to be more efficient compared to AF555 because only 2.6-fold (RNA1) and 3.2-fold (RNA2) fluorescence intensity increases were yielded with AF555. Although structural details of both AFmaleimides are reported in literature [27], the exact structure was not provided by the manufacturer of the dyes. Therefore, the varying energy transfer efficiency cannot conclusively be explained, but we assume differing Förster radii or competing photochemical processes could be possible explanations for the lower efficiency. The lowest fluorescence gains were achieved with the AF647 dye.

Conclusions
In conclusion, we present an efficient synthetic route to the new tetrazole-modified phosphoramidite building block 9. The uridine substitutes one of the aryl groups of the photoactive diaryltetrazole. This modification was incorporated internally and 5 -terminally into two RNA strands using conventional automated solid-phase chemistry. Photoclick reactions were carried out as postsynthetic modifications by irradiation with 300 nm light (LED) in aqueous media and in the presence of three different commercially available dye-maleimide conjugates (Cy3, AF555 and AF647). The yields vary significantly for the different dye-maleimide conjugates: AF555 maleimide produces the highest yields of up to 84% (RNA2), whereas the Cy3-maleimide shows smaller yields down to 27% (RNA1). In contrast, Cy3 provides the highest fluorogenic effect by the 7.5-fold (RNA1) to 9.4-fold (RNA2) fluorescence intensity increase, which is nearly one magnitude of order increase of fluorescence intensity. The concept has a broad substrate scope, both with respect to the dye-maleimide conjugates and with respect to the position within RNA sequence. This photoclick chemistry is a highly promising candidate for cell applications. Although the 300 nm light is in principle harmful to biological cells, the short irradiation times may be tolerable. Due to the fact that cytotoxic copper salts are avoided and the application of time allows spatiotemporal resolution, this postsynthetic RNA chemistry is a new tool for fluorescent imaging of RNA in biological cells.