Synthesis of 18O-labeled RNA for application to kinetic studies and imaging

Radioisotopes and fluorescent compounds are frequently used for RNA labeling but are unsuitable for clinical studies of RNA drugs because of the risk from radiation exposure or the nonequivalence arising from covalently attached fluorophores. Here, we report a practical phosphoramidite solid-phase synthesis of 18O-labeled RNA that avoids these disadvantages, and we demonstrate its application to quantification and imaging. The synthesis involves the introduction of a nonbridging 18O atom into the phosphate group during the oxidation step of the synthetic cycle by using 18O water as the oxygen donor. The 18O label in the RNA was stable at pH 3–8.5, while the physicochemical and biological properties of labeled and unlabeled short interfering RNA were indistinguishable by circular dichroism, melting temperature and RNA-interference activity. The 18O/16O ratio as measured by isotope ratio mass spectrometry increased linearly with the concentration of 18O-labeled RNA, and this technique was used to determine the blood concentration of 18O-labeled RNA after administration to mice. 18O-labeled RNA transfected into human A549 cells was visualized by isotope microscopy. The RNA was observed in foci in the cytoplasm around the nucleus, presumably corresponding to endosomes. These methodologies may be useful for kinetic and cellular-localization studies of RNA in basic and pharmaceutical studies.


INTRODUCTION
RNA-based drugs such as antisense oligomers, aptamers, short interfering RNA (siRNA) and microRNA (miRNA) have been under investigation for more than a decade. Since the key discoveries in RNA interference (RNAi) by Fire et al. (1), the pace of basic and clinical research on RNA drug candidates has been accelerating rapidly. To develop RNA as therapeutic drugs, effective analytical methods are needed for use in studies on the absorption, distribution, metabolism and excretion of RNA. However, new analytical methods for detecting labeled RNA drugs have not kept pace with research.
The conventional methods of labeling RNA are fluorescent and radioisotope labeling. While both of these methods have advantages, they also have disadvantages. Thus, fluorescent labeling is in principle sensitive enough to allow detection of a single molecule (2), but it has the disadvantage that the labeled RNA is not chemically equivalent to its unlabeled counterpart because of the covalently attached fluorophore. For example, blocking the 5 0 -hydroxyl terminus of the antisense strand of siRNA can lead to a loss of RNAi activity (3). Radioisotope labeling also has a high sensitivity and the label can be easily introduced into the molecule, for example, by transfer of the terminal phosphate of [5 0 -32 P]ATP to the 5 0 -hydroxyl terminus of an RNA oligomer catalyzed by T4 polynucleotide kinase. The disadvantage of radioisotope labeling, however, is the risk associated with increased radiation exposure as well as the short half-life of some radioactive isotopes, in particular phosphorus.
With these considerations in mind, we set out to label RNA oligomers with the stable oxygen isotope 18 O, which would avoid the disadvantages of fluorescent and radioisotope labeling outlined above. The properties of chemical equivalence and lack of radioactive decay make 18 O well suited for labeling molecules for use as tracers in basic and clinical studies. Except for the labeling of proteins for quantitative proteomics (4,5) and the labeling of RNA for comparative mass-spectrometric analyses (6,7), 18 O labeling has not been widely used in the life sciences. Several approaches to the labeling of nucleic acid oligomers with stable oxygen isotopes have been reported. These include the synthesis of 2 0 -18 Olabeled uridine RNA (8) and the incorporation of 18 O into the 3 0 -phosphate group of RNA on endonuclease digestion (6,7). In addition, 18 O has been incorporated into the phosphate groups of nucleic acid dimers during liquid-phase synthesis by using 18 O water as the oxygen donor to phosphate (9,10), and 18 O has been incorporated into dTdT and 17 O into one phosphate group of a deoxyribonucleotide octamer during solidphase synthesis (11).
Here, we report the synthesis of 21mer RNA with 18 O incorporated into all of the phosphate groups by using 18 O water in the oxidation step during solid-phase synthesis. We demonstrate the stability of the 18 O label under physiological conditions and the equivalence of 18 O-labeled and unlabeled RNA. We then go on to show that isotope ratio mass spectrometry (IRMS) and isotope microscopy can be used for the quantification and imaging of 18 O-labeled RNA in biological material.

RNA synthesis
RNA oligomers were synthesized on commercially available controlled-pore glass solid supports with a pore size of 1000 Å (3-Prime, Aston, PA, USA) in columns installed in an Expedite model 8909 nucleic acid synthesizer (Applied Biosystems, Foster City, CA, USA). Synthesis was carried out on a 1-mmol scale from the tertbutyldimethylsilyl (TBDMS) amidites (ST Pharm, Seoul, Korea) as described by Bellon (12). The detritylation reagent was 3% trichloroacetic acid in dichloromethane. For the coupling reaction, TBDMS amidites were used as 0.05 M solutions in acetonitrile, and the activator was 0.25 M 5-benzylmercaptotetrazole in acetonitrile. The capping reagent was a mixture of 10% acetic anhydride in tetrahydrofuran (THF) and 10% N-methylimidazole in THF/pyridine (8:1), and the oxidizing agent was 0.01 M iodine in THF/pyridine/water (78:20:2). All reagents and solvents used were commercially available. 18 O-labeled RNA oligomers were synthesized in the same way except that the solvent for the oxidizing agent was THF/pyridine/ 18 O water (78:20:2). 18 O-enriched water (99.2 atom %) was purchased from Taiyo Nippon Sanso (Tokyo, Japan). After synthesis, the protecting groups of the RNA oligomers were removed. The purity of the crude RNA product was confirmed by reverse-phase high-performance liquid chromatography (HPLC) and liquid chromatography coupled with electrospray ionization quadrupole time-of-flight tandem mass spectrometry (LC-ESI-Q-Tof MS). After ethanol precipitation, RNA oligomers were purified by preparative reverse-phase HPLC. Purified RNA oligomers to be used for the preparation of siRNA were dissolved in 30 mM HEPES-KOH buffer, pH 7.4, containing 100 mM potassium acetate and 2 mM magnesium acetate. siRNAs were prepared by annealing the purified short complementary strands by incubation at 90 C for 10 min and then at room temperature for 1 h. Duplex formation was confirmed by 15% polyacrylamide gel electrophoresis followed by staining with ethidium bromide.

HPLC and LC-ESI-Q-Tof MS
HPLC was performed on a Shimadzu LC-10A system (Shimadzu, Kyoto, Japan) equipped with an XBridge Oligonucleotide Separation Technology (OST) C18 column (1.7 mm; 4.6 Â 50 mm; Waters, Milford, MA, USA). The following solvent system was used for analytical and preparative HPLC: buffer A (50 mM triethylammonium acetate, pH 7, in 5% acetonitrile) and buffer B (50 mM triethylammonium acetate, pH 7, in 90% acetonitrile). Elution was carried out with a 5-10% linear gradient of buffer B in buffer A. Mass spectra were obtained with a Waters Acquity ultra performance liquid chromatography (UPLC) system coupled with a Waters Synapt G2 ESI-Q-Tof mass spectrometer. The UPLC system was used with a Waters Acquity UPLC OST C18 column (1.7 mm; 2.1 Â 50 mm). The following solvent system was used for liquid chromatographymass spectrometry: solvent A (an aqueous solution of 15 mM triethylamine and 400 mM hexafluoroisopropanol, pH 7.9) and solvent B (50% methanol). Elution was carried out with a 5-50% linear gradient of solvent B in solvent A. The mass spectrometer was operated in negative ion mode. External calibration was performed with sodium iodide, and the Transform software (Waters) was used for spectral deconvolution.

Stability of 18 O label at various pH values
Solutions of 18 O-labeled uridine 20mer (20 mM) were prepared in 50 mM sodium phosphate buffer at pH 3, 4, 5.5, 7 and 8.5. Solutions were incubated at 37 C and samples were analyzed by LC-ESI-Q-Tof MS after 1, 3, 7, 14, 21 and 28 days.

Circular dichroism spectra
Circular dichroism (CD) spectra of 10 mM siRNA solutions in 50 mM sodium phosphate buffer, pH 7.5, were recorded at 25 C with a JASCO J-720 W spectropolarimeter (JASCO, Tokyo, Japan) in a 1-cm-path-length quartz cell at a rate of 50 nm/min and digitized at a resolution of 0.1 nm. The bandwidth was 1 nm and the response time was 8 s. For each determination, at least five spectra were accumulated and the averaged spectrum was subjected to noise reduction.

Determination of melting temperature (T m )
Melting curves of siRNA in 50 mM sodium phosphate buffer, pH 7.5, were recorded with a UV-1800 UV-VIS spectrophotometer (Shimadzu Scientific Instruments, Kyoto, Japan). The absorbance at 260 nm was monitored as the temperature was raised from 25 to 95 C at a rate of 0.5 C/min. T m values were calculated with Tm Analysis Software (Shimadzu Scientific Instruments).

Cell culture
The cell lines A549 (a human lung adenocarcinoma epithelial cell line) and HCT116 (a human colorectal carcinoma cell line) were obtained from the European Collection of Cell Cultures (Salisbury, UK). The cells were cultured in Dulbecco's modified Eagle's medium (DMEM) and McCoy's 5A medium (Sigma, St. Louis, MO, USA), respectively, containing 10% heat-inactivated fetal bovine serum (MP Biomedicals Japan, Tokyo, Japan) and incubated at 37 C in a humidified 5% CO 2 atmosphere.  O-labeled control siRNA (100 or 500 mg) in water for injection was administered intravenously. Animals were sacrificed 5, 30 and 60 min after RNA administration, blood was collected in heparinized tubes and plasma was prepared by centrifugation at 900g. For the calibration curve, pooled plasma prepared from untreated mice was spiked with 18 O-labeled control siRNA (final RNA concentrations: 0, 10, 20, 40, 60, 80,100 and 200 mg/ml). For IRMS, plasma samples were lyophilized. Experiments were performed in triplicate. The relative difference in 18 O/ 16 O ratio between the sample and the internal standard, d 18 O, was expressed as parts per thousand according to the equation Imaging of cells transfected with 18 O-labeled RNA by isotope microscopy A549 cells were seeded on a piece of a silicon wafer (7 Â 7 mm) in a 35-mm dish (5 Â 10 5 cells/dish) and incubated at 37 C in humidified 5% CO 2 . After 24 h, the culture medium was replaced by 2 ml of fresh medium and the cells were transfected with unlabeled or 18 O-labeled control siRNA (490 ml) in Opti-MEM (final concentrations, 100 and 300 nM) by using Lipofectamine 2000 (10 ml). Twenty-four hours after transfection, the cells on the silicon wafer were washed with phosphate-buffered saline, immobilized with 2.5% glutaraldehyde in phosphate-buffered saline at 37 C for 1 h and dehydrated by soaking successively in 20, 40, 60, 70, 80, 90 and 100% ethanol followed by tert-butanol at room temperature for 10 min each. The transfected cells were frozen at 4 C for 24 h and dried under reduced pressure for isotope microscopy. Samples were coated with a 30-nm layer of gold to avoid the accumulation of positive charge due to the primary beam of the isotope microscope.
There are two approaches to obtaining an isotope map by secondary ion mass spectrometry (SIMS): scanning and direct imaging. In scanning, isotopic data are collected sequentially from each point of the analyzed area with a focused primary ion beam and images are reconstructed by computer. In direct imaging, stigmatic ion optics are used with a high-efficiency 2D stacked complementary metal-oxide-semiconductor (CMOS) active pixel sensor (SCAPS) detector. The SCAPS detector features a wide dynamic range, no dead time and direct detection of ions. Small structures in biological materials, such as organelles or vesicles, are not always exposed at the sample surface. Direct imaging is useful to reach targets on the order of a few tens of nanometers in size embedded in a matrix on the order of a few microns in size because a high-intensity primary beam can be used without degrading the spatial resolution.
In the present study, the Hokudai isotope microscope system (a Cameca ims-1270 ion microprobe equipped with a SCAPS detector; Cameca, Gennevilliers, France) was used to visualize the isotope distribution in the cells, a technique known as isotopography. The sample surface ($100 Â 100 mm 2 ) was homogeneously irradiated with a 20-keV Cs + primary beam with a beam current of $5 nA. Secondary ion images of 12 C 14 N, 16 O, 18 O and 16 O were obtained sequentially for the same field for each analytical sequence. The exposure times were 5 s for 12 C 14 N, 5 s for 16 O and 500 s for 18 O isotopography. A 50-mm contrast aperture was used, and the secondary ion contributions, except for objective isotopes, were cut by the exit slit. The typical spatial resolution under the conditions of isotopography was $0.3 mm, and the width of a pixel of SCAPS corresponds to 0.2 mm on the sample surface. An image-processing method with a moving average of 3 Â 3 pixels was applied to 18 O/ 16 O isotopography. The resulting spatial resolution of the isotope ratio image was 0.6 mm. Other analytical methods for isotopography were carried out as described elsewhere (13,14).

Synthesis of 18 O-labeled RNA
We designed a siRNA targeting the human GAPDH gene ('target siRNA') and a control siRNA that had no genesilencing target in the human genome, both 21 nucleotides in length ( Table 1). The sense and antisense strands of each siRNA were separately synthesized in the solid phase by the phosphoramidite method with 16 Figure S1), and their yield and purity estimated by their absorbance at 260 nm were nearly identical (Table 1). Thus, the use of 18 O water in the oxidation step achieved the same result as when 16 O water was used. LC-ESI-Q-Tof mass spectra of the RNA oligomers ( Figure 1) show that the mass difference between the 18 O-labeled and the unlabeled oligomer was 34-35 Da (Table 1). This is somewhat lower than the mass difference calculated on the basis of the incorporation of 20 atoms of 18 O into each 21mer allowing for the 99.2% isotopic purity of the 18 O water used (2 Â 20 Â 0.992 = 39.68). This somewhat low value may be explained by a combination of an 18 O/ 16 O kinetic isotope effect and reaction with water from moisture in the air. The discrepancy may be decreased by using 18 O water with a higher degree of isotopic purity and excluding moisture from the air. In any case, these results confirm that the 18 Olabeled RNAs were correctly synthesized and that one 18 O atom was incorporated into each phosphate group $85% of the time under our experimental conditions.
Limbach's group (6,7) has described the incorporation of 18 O into the 3 0 -phosphate groups of RNA during endonuclease digestion. However, this method can only be used on a laboratory scale, and the 18 O label is incorporated only into the 3 0 -phosphate group. In addition, Dai et al. (8) have described the solid-phase synthesis of 2 0 -18 Olabeled uridine RNA by using 2 0 -18 O uridine phosphoramidite derived from 2,2 0 -cyclouridine. However, the required phosphoramidite is not commercially available, and labeling is limited to uridine residues. The incorporation of 18 O into the phosphate group of nucleic acid oligomers by using 18 O water as an oxygen donor to phosphate has been reported by Potter's group in the liquid-phase synthesis of the dimers dGdA (9) and rUrA (10) for their studies on oxygen-chiral phosphate groups. Although 18 O has been incorporated into dTdT and 17 O into one phosphate group of a deoxyribonucleotide octamer by the phosphoramidite solid-phase method (11), our study is, to the best of our knowledge, the first in which a stable oxygen isotope has been incorporated into multiple phosphate groups in a nucleic acid oligomer.
By the method reported here, 18 O can be incorporated into all phosphate groups in an RNA oligomer or, by limiting the use of 18 O water to certain oxidation steps during synthesis, only into selected phosphate groups. The method could also be used to incorporate 18 O into DNA oligomers. Similarly, the method could be extended to the use of 34 S-labeled sulfurizing reagents such as 3H-1,2-benzodithiol-3-one 1,1-dioxide (Beaucage reagent) (15,16) or 10 B-labeled N,N-diisopropylethylamine/borane complex (17) in the oxidation step to allow the synthesis of 34 S-labeled phosphorothioate or 10 B-labeled boranophosphate oligomers. Thus, this labeling approach is not limited by the nature of the phosphoramidites, labeling position, nucleotide sequence or class of oligomer, and it should allow the practical preparation of labeled oligonucleotides with any solidphase DNA/RNA synthesizer. The approach is readily amenable to scaling up, and the handling of the stable isotopes is simple and safe.

Stability of the 18 O label in the RNA oligomer
The stability of the 18

Equivalence of 18 O-labeled and unlabeled RNA
For labeled RNA to be used as a tracer, it should ideally have the same physicochemical and biological properties as the corresponding unlabeled RNA. To demonstrate the equivalence of 18 O-labeled and unlabeled RNA, we prepared target and control siRNAs in which labeled and unlabeled sense and antisense strands were paired in all four combinations: 16 Figure S3). This shows that the 18 O-labeled RNA oligomers annealed normally to both 18 O-labeled and unlabeled complementary strands to form a duplex.
Melting curves were acquired for the four siRNA duplex combinations (data not shown) and used to calculate T m values (Supplementary Table S1). All of the target siRNA duplex combinations had the same T m value to within 0.6 C and all of the control siRNA duplex combinations to within 0.5 C. Thus, 18 O labeling of the backbone phosphate groups had little or no effect on the strength of base pairing.
Furthermore, the CD spectra of the four target and control siRNA duplex combinations were all similar, with a negative maximum at 210 nm and a positive maximum at 260 nm (Supplementary Figure S4). This CD spectral pattern typically indicates the A-form RNA duplex (19,20). Thus, the CD spectra show that the 18 O-labeled siRNA strands paired with either labeled or unlabeled strands to form an A-type helix that was conformationally indistinguishable from that formed by two unlabeled strands.
Finally, to compare the biological activity of 18 O-labeled and unlabeled siRNA, we measured the silencing effect of siRNAs targeting the GAPDH gene in the human cell line HCT116. These short double-stranded RNAs suppress the expression of their target genes by causing cleavage of the mRNA or inhibiting its translation (21)(22)(23)(24). As expected, the control siRNA had no silencing effect (data not shown). In contrast, the target siRNA had a silencing effect, and the effect of the 18 (Figure 2). In RNAi, the antisense strand is the main effective strand (25,26), and our 16 O/ 18 O siRNA, which had the 18 O label only in the antisense strand, also had a silencing effect that was indistinguishable from that of 16 Taken together, these results show that the physicochemical properties and biological activity of the 18 O-labeled RNA were essentially the same as those of the unlabeled RNA, increasing our confidence in the suitability of 18 O-labeled RNA as a tracer.

Quantification of 18 O-labeled RNA
IRMS, which can be used to determine the 18 O/ 16 O ratio in samples, finds application in food science (27) and sports drug testing (28). In the present study, we extended this technique to estimate the concentration of 18 O-labeled siRNA in biological samples. A calibration curve for 18 O-labeled control siRNA in mouse plasma was constructed in which d 18 O, the relative difference in 18 O/ 16 O ratio between the sample and the internal standard, was estimated by IRMS and plotted against the concentration of control siRNA as determined by its absorbance at 260 nm ( Figure 3A). The calibration curve showed good linearity over the entire concentration range of 0-200 mg/ml, with a correlation factor (R 2 ) of 0.9999. We next estimated the blood concentration of control siRNA after intravenous administration to mice. Naked RNA oligomer in the blood is rapidly degraded by ribonucleases and eliminated by the kidney. After administration of naked RNA oligomer (100 or 500 mg), the blood concentration-time curves ( Figure 3B) showed a rapid decay of the signal that presumably reflects the degradation of the RNA. These results give us confidence that 18 O labeling combined with IRMS allows determination of the blood concentration of RNA after administration, opening the way for the use of 18 O labeling in pharmacokinetic studies of RNA drugs.
To detect intact RNA after administration, it is necessary to protect it from degradation, for example, by chemical modification or a drug delivery system. In this study, the administered RNA was not protected, so it is assumed that the observed 18 O signal came not only from the intact RNA but also from its degradation products. However, even with chemical modification or a drugdelivery system, it is in general difficult to ensure the complete stability of RNA oligomers in vivo. In addition, it is difficult to judge the intactness of an RNA oligomer in vivo by any labeling method.
Isotope dilution mass spectrometry has previously been used for the quantification of a therapeutic monoclonal antibody containing 13 C and 15 N (29) and an antiretroviral drug containing 13 C (30), with the isotopically labeled compound as the internal standard in both studies. However, it is difficult to apply such isotopiclabeling methods to the analysis of large molecules such as proteins and oligonucleotides because the mass difference between the unlabeled compound and the isotopically labeled compound cannot be reliably determined without additional treatment such as enzymatic digestion or fragmentation (6,29). Immunoassay and radioisotope labeling are sometimes used for the quantification of oligonucleotides, but the former method requires the preparation of a specific antibody for each compound, while the latter carries the risks associated with increased radiation exposure. Although our method of detection of 18 Olabeled RNA may still need improvement in sensitivity and accuracy, the 18 O-labeling approach described here is likely to be useful in basic and clinical studies of RNA drugs.

Imaging of 18 O-labeled RNA
Isotope microscopy is based on SIMS coupled with a special ion imager and is capable of 2D 18 O/ 16 O isotopic analysis (13,14). SIMS is used to analyze the composition of solid surfaces and thin films by irradiating the surface of the specimen with a focused primary ion beam and collecting and analyzing the ejected secondary ions. The secondary ions can be measured with a mass spectrometer to determine the elemental, isotopic or molecular composition of the surface. Isotope microscopy has been used mainly in cosmochemistry and geochemistry for the isotopic elemental analysis of extraterrestrial minerals such as primitive carbonaceous chondrites (31), lunar rocks (32) and material returned from the asteroid  Itokawa by the Hayabusa mission (33). These techniques have yet to find application in the pharmaceutical field. We set out to use isotope microscopy to generate a 2D image of the distribution of 18  As with the IRMS study, we assume that the observed 18 O signal was not only from intact RNA but also from degradation products. However, RNA introduced into cells with a transfection agent is reported to be detectable 48 h after transfection (35). Because we used the A : 12   transfection agent Lipofectamine, it is likely that the 18 O signals detected with SIMS represent at least some intact RNA. As far as we know, this is the first application of isotope microscopy to the imaging of biological samples containing 18 O-labeled compounds. Previous localization studies of RNA have usually been performed with fluorescently labeled RNA. However, fluorescence methods have disadvantages. One is the presence of covalently attached fluorophores, which not only make the labeled molecule chemically different from the unlabeled molecule, but in addition are usually bulky and hydrophobic and thus alter all the more the physicochemical properties of the labeled molecule. Another disadvantage is the possible loss of the fluorophore from the labeled molecule. In contrast, 18 O-labeled RNA appears to be both chemically and biologically equivalent to the corresponding unlabeled RNA. In addition, any desired subset of phosphate groups in the oligomer, or all of them, can be labeled at will. These represent distinct advantages of 18 O over fluorescent labels in cellular localization analysis. Imaging technology that combines 18 O labeling with isotope microscopy is expected to be a useful biological and pharmaceutical research tool.

CONCLUSIONS
We have prepared RNA labeled with one nonbridging 18 O atom in each phosphate group by using 18 O water as the oxygen donor to phosphate during the oxidation step of the synthetic cycle. The 18 O label was stable under physiological conditions for at least 28 days and the 18 O-labeled RNA was physicochemically and biologically indistinguishable from the corresponding unlabeled RNA by electrophoresis, CD, T m and RNAi activity. These properties of 18 O-labeled RNA make it suitable for use as a tracer in basic and clinical studies of RNA drugs such as antisense oligomers, siRNA, miRNA and aptamers. 18 O-labeled RNA lends itself to the quantification of RNA in blood by IRMS and to the imaging of the intracellular distribution of the RNA by isotope microscopy. These methods have the potential to avoid the present difficulties with the quantification and visualization of RNA in biological samples by radioisotope and fluorescent labeling, and we expect them to contribute the development of RNA therapeutics.

SUPPLEMENTARY DATA
Supplementary Data are available at NAR Online: Supplementary Table 1 and Supplementary Figures 1-4.

ACKOWLEDGEMENTS
We thank T. Ueda and T. Igawa (Graduate School of Pharmaceutical Sciences, Kyushu University) for help with the circular dichroism spectrometry, and T. Moromachi and R. Sato (SI Science Co., Ltd) for help with the isotope ratio mass spectrometry.

FUNDING
Funding for open access charge: BONAC Corporation.