Coumarin-Based Photolabile Solid Support Facilitates Nonchromatographic Purification of RNA Oligonucleotides

We report the synthesis and application of a coumarin-based photocleavable anchor for solid-phase synthesis and nonchromatographic purification of RNA oligonucleotides. Unlike conventional nitroaryl anchors, the system enables photocleavage using visible light (456 nm), avoiding RNA damage associated with UV irradiation.

The approach was optimized using a 20-nt poly-U RNA and successfully applied to structured and unstructured RNAs containing phosphodiester and phosphorothioate backbones. Purification of a 103-nt sgRNA demonstrated preserved functional activity in CRISPR experiments, confirming the method’s suitability for sensitive and complex RNA constructs.

The original article

Coumarin-Based Photolabile Solid Support Facilitates Nonchromatographic Purification of RNA Oligonucleotides

Ian McClain, Hilal Dagci, Bhoomika Pandit, Maksim Royzen*

J. Org. Chem. 2025, 90, 46, 16326–16333

https://doi.org/10.1021/acs.joc.5c01528

licensed under CC-BY 4.0

Selected sections from the article follow. Formats and hyperlinks were adapted from the original.

The rapid growth of biomedical RNA research created a strong need for innovation in chemical RNA synthesis. Such need is driven by the FDA approval of multiple siRNA-based drugs, including patisiran (2018), givosiran (2019), lumasiran (2020), and inclisiran (2022). (1) The active ingredients of these drugs are currently manufactured using classical solid-phase synthetic procedures and purified by preparative HPLC. In recent years, we have witnessed a rapid development of CRISPR-Cas9-based research tools and therapeutics. (2) The key ingredient of these technologies is single-guide RNA (sgRNA), which is about 100-nt long and contains multiple modifications that improve its in vivo stability. (3−5) These sgRNAs are especially difficult to synthesize and purify using classical approaches due to their length and structural complexity.

Currently, HPLC is the most common approach to purify synthetic RNA oligonucleotides. Out of many different chromatographic methods, reverse-phase (RP) and anion-exchange chromatography are the most prevalent. (6) The former is typically carried out using oligonucleotides protected with a highly hydrophobic dimethoxytrityl (DMT) group to further differentiate their polarity from many impurities that accumulate during solid-phase synthesis. The DMT group is subsequently chemically cleaved post-purification. On the other hand, anion-exchange chromatography leverages the anionic nature of multiple phosphate groups on the backbone of RNA to build strong attraction forces with the positively charged amine ligands on the solid support. (7) A number of innovative approaches to facilitate chromatographic oligonucleotide purification have been reported. For example, scientists at Berry and Associates developed fluorous oligonucleotide tags to facilitate fluorous affinity chromatography. (8) Also, DNA-affinity chromatography has been reported using DNA-modified silica chromatographic column. (9) Despite considerable progress in terms of methods and modern instrumentation, long oligonucleotides remain difficult to purify chromatographically because of structural similarities between the target strand and the failure sequences. For RNAs that are over 100-nt, the target strand often has to be purified from a crude mixture containing more than 90% impurities.

A number of innovative nonchromatographic purification methods have emerged in recent years to address these challenges. (10) For example, Fang and co-workers developed acrylated phosphoramidites for selective modification of either the target or failure sequences. (11,12) Nonchromatographic purification was accomplished by polymerization, which facilitates facile differentiation of target and failure sequences. The method has been applied to isolate a 1728-nt-long oligonucleotide. (13) Bergstrom reported a purification approach that entails biotinylation of the 5′-end of synthetic RNAs. (14,15) Postsynthetically, the target strands were captured with NeutrAvidin-coated microspheres. The Minakawa group described a “catch and release” oligonucleotide purification strategy that combined strain-promoted alkyne–azide cycloaddition and photocleavage. (16) Our lab also reported a nonchromatographic purification procedure, which utilizes inverse electron demand Diels–Alder chemistry between trans-cyclooctene (TCO) and tetrazine (Tz), illustrated in Scheme 1. (17)

Our procedure takes advantage of the free 5′–OH group on the target strand, and we can selectively tag the target with Tz (step 1). The key element of this process is a photolabile linker, which facilitates cleavage from the solid support in step 2, while preserving all the protecting groups on the RNA. The target strand can subsequently be captured using solid support-immobilized TCO and purified from the failure strands, which will be dissolved and washed away in the supernatant (step 3). The target RNA strand is isolated using standard cleavage, deprotection, desilylation, and ethanol precipitation steps. In our original report, we utilized an o-nitroaryl photocleavable linker developed by Greenberg et al. (18) The photocleavage step entailed exposure to 350 nm of light for 2 h. In the current work, we developed a coumarin-based photolabile linker that requires irradiation with visible light (456 nm) for a shorter period of time. This linker is particularly useful for RNAs that contain groups that are sensitive to UV light, such as the phosphorothioate backbone.

Results and Discussion

To optimize the photocleavage step (Scheme 1, step 2), we synthesized a model 20-nt poly-U RNA strand, termed RNA1. While on support, the oligonucleotide was tagged with the freshly prepared Tz anhydride following the protocol described in our previous work (Scheme 1, step 1). (17) The photocleavage was carried out using a Kessil lamp (PR160L) at 75% power. We tested 15, 30, and 60 min of irradiation, which was followed by deprotection, desilylation, and EtOH precipitation. The HPLC analysis of isolated RNAs is shown in Figure S1A. The highest isolated yield of RNA was obtained after 30 min of photocleavage. As illustrated in Figure S1B, relative to the AMA treatment, 30 min of photoirradiation enabled cleavage of 93% of RNA1. After optimization, the photocleavage step was repeated using a fresh sample and the cleaved oligonucleotide was captured by TCO-modified CPG beads using the previously described procedure (Scheme 1, step 3). (25) During this step, the failure strands lacking the Tz tag were removed by filtration. The target RNA1 was obtained upon treatment with a 1:1 aqueous methylamine-ammonia solution (AMA), followed by desilylation with triethylamine trihydrofluoride and EtOH precipitation. Figure 1A shows the HPLC analysis of the isolated RNA1 strand. Its identity was verified using a reference mixture of 17, 18, 19, and 20-nt poly-U RNA strands that were chromatographed using the same HPLC conditions (Figure 1B). Purity of isolated RNA1 was determined to be 86.2%. The isolated RNA1 strand was also analyzed by ESI-MS. The deconvoluted mass spectrum, shown in Figure 1C, matched the calculated m/z value of 6056.52. Based on the nanodrop measurements, the isolated yield of the target RNA was 30%.

J. Org. Chem. 2025, 90, 46, 16326–16333: Figure 1. (A) HPLC analysis of the isolated 20-nt poly-U RNA oligonucleotide. (B) Reference HPLC spectrum of a mixture containing 17, 18, 19, and 20-nt poly-U RNA strands. (C) Deconvoluted ESI-MS analysis of the isolated 20-nt poly-U RNA oligonucleotides.

To confirm that our approach is general, we applied the nonchromatographic RNA purification procedure toward the isolation of 20-nt stem-loop RNA (RNA2) with the sequence: 5′-ACCUGGCUUUCACCCAGGUT-3′ and a nonstructured 20-nt RNA (RNA3) with the sequence: 5′-GAUCCUGCCGACUACGCCAT-3′. Both RNAs were synthesized using 6 and were purified using the process described in Scheme 1. Figure 2A,B shows the HPLC spectra of purified RNA2 and RNA3. Both spectra show dominant target peaks with trace numbers of failure sequences. Purities of the isolated RNA2 and RNA3 were determined to be 84.6 and 86.7%, respectively. The isolated RNA2 and RNA3 strands were also analyzed by ESI-MS. The deconvoluted mass spectra are shown in Figure 2C,D. The observed m/z value of 6274.88 for RNA2 matched the calculated value of 6274.83. The observed m/z value of 6296.91 for RNA3 matched the calculated value of 6296.89. Based on the nanodrop measurements, the isolated yields of RNA2 and RNA3 were 17 and 18%, respectively.

J. Org. Chem. 2025, 90, 46, 16326–16333: Figure 2. (A) HPLC analysis of the isolated RNA2 oligonucleotide. (B) HPLC analysis of the isolated RNA3 oligonucleotide. (C) Deconvoluted ESI-MS analysis of the isolated RNA2 oligonucleotide. (D) Deconvoluted ESI-MS analysis of the isolated RNA3 oligonucleotide.

An important advantage of the coumarin-based photolabile linker is that it is much more compatible with RNAs containing a phosphorothioate backbone. Prolonged exposure of the latter to 350 nm light can potentially cause photo-cross-linking, thus damaging the RNA’s structure and functions. Meanwhile, the phosphorothioate backbone is an important feature of many therapeutic RNAs known for enhancing in vivo stability against nuclease degradation. (26) We tested our nonchromatographic RNA purification procedure toward the isolation of RNA4 and RNA5 that have the same sequences as RNA2 and RNA3 but contain a phosphorothioate backbone. Both RNAs were synthesized using 6 and were purified using the process described in Scheme 1. Figure 3A,B shows the HPLC spectra of purified RNA4 and RNA5, respectively. Both spectra show dominant target peaks with trace amounts of failure sequences. Purities of isolated RNA4 and RNA5 were determined to be 97.4 and 97.2%, respectively. The isolated RNA4 and RNA5 strands were also analyzed by ESI-MS. The deconvoluted mass spectra are shown in Figure 3C,D. The observed m/z value of 6582.9 for RNA4 perfectly matched the calculated one. The observed m/z value of 6604.9 for RNA5 matched the calculated value of 6605.0. Based on the nanodrop measurements, the isolated yields of RNA4 and RNA5 were 29 and 26%, respectively.

J. Org. Chem. 2025, 90, 46, 16326–16333: Figure 3. (A) HPLC analysis of the isolated RNA4 oligonucleotide. (B.) HPLC analysis of the isolated RNA5 oligonucleotide. (C) Deconvoluted ESI-MS analysis of the isolated RNA4 oligonucleotide. (D) Deconvoluted ESI-MS analysis of the isolated RNA5 oligonucleotide.

Conclusion

In conclusion, this report describes a new coumarin-based photolabile anchor that was utilized for the solid-phase synthesis of RNA oligonucleotides. Photocleavage was carried out using 456 nm light, which is compatible with RNAs containing phosphodiester and phosphorothioate backbones. Synthetic RNAs were isolated using a nonchromatographic purification process that we developed in our prior work. (17,25) The photocleavage step was optimized using a model 20-nt poly-U oligonucleotide. Subsequently, the nonchromatographic purification procedure was applied to more complex RNAs, including 100-nt sgRNA for CRISPR experiments. In comparison to the standard RNA purification procedures using either RP-HPLC or anion-exchange chromatography, our process is faster and does not require expensive instrumentation and columns. By contrast, chromatographic purification typically entails lengthy optimization and preparation. However, it can facilitate the isolation of oligonucleotides in higher purity. In comparison to our previous work, we did not achieve significant improvement in terms of isolated yields. The main advantage of the new coumarin-based photolabile anchor is that it is more compatible with RNAs containing a phosphorothioate backbone. We have shown that the latter is stable to 456 nm light but sensitive to prolonged irradiation with UV light. The reported purification process, however, has a number of limitations in its current form. The process is not compatible with RNAs containing modifications at the 5′-end, such as a phosphate group. Also, the isolated oligonucleotides were not pure RNAs because their first nucleotide was deoxythymidine. In the future, we plan to adopt the reported synthetic procedures to make a uridine analog of compound 5. We also plan to further optimize the process to obtain better isolated yields. The experiments shown in Figure S1 indicate that the photocleavage yields are comparable to those of the AMA treatment. Therefore, there are inefficiencies in other steps of the purification process, such as tagging or capture steps. We plan to improve them in our future work using more reactive Tz and TCO compounds, (28,29) as well as alternative solid support materials to capture the photocleaved RNA. (30) Furthermore, we plan to work on improving the purity of nonchromatographically isolated RNA. This will entail modification of the capping step of the solid-phase synthesis to ensure that all but 5′-hydroxyl groups are protected. To achieve that, we plan to explore alternative capping reagents. Our ultimate goal is for the coumarin-based photolabile anchor and the nonchromatographic purification process to be useful for the synthesis of therapeutic RNAs for various biomedical applications.

Experimental Section

Thin-layer chromatography (TLC) was performed on SiliaPlate silica gel TLC plates with 250 μm thickness (Silicycle Inc., QC, Canada). Preparative TLC was performed using SiliaPlate silica gel TLC plates with 1000 μm thickness. ¹H and ¹³C NMR spectroscopy was performed on a Bruker NMR instrument at 500 MHz (¹H) and 126 MHz (¹³C). All ¹³C NMR spectra were proton decoupled. High-resolution ESI-MS spectra were acquired using an Agilent Technologies 6530 Q-TOF instrument. The RNA and DNA samples were analyzed on a Thermo Fisher Scientific (West Palm Beach, CA) LTQ Orbitrap Velos Mass spectrometer using quartz capillary emitters. To facilitate spray optimization, 10% isopropyl alcohol was added to each sample prior to MS analysis. Purification of all synthetic oligonucleotides was characterized by denaturing urea polyacrylamide gel electrophoresis (PAGE) (15 wt %, Acrylamide:Bis-acrylamine = 29:1, 1× TBE buffer). Gels were then stained with Ethidium Bromide (EthBr, 1 μg/mL) and visualized with a ChemDoc Imaging System. Low-range ssRNA ladder (New England Biolabs cat# N0364S) was used as shown in Figure 4.

All oligonucleotide solid-phase syntheses were performed on a 0.5 μmol scale using the Oligo-800 synthesizer (Azco Biotech, Oceanside, CA, USA). Solid-phase syntheses were performed on a control-pore glass (CPG 1000) purchased from Glen Research (Sterling, VA, USA). Other oligonucleotide solid-phase synthesis reagents were obtained from ChemGenes Corporation (Wilmington, MA, USA). Phosphoramidites (TBDMS as the 2′-OH protecting group): rA was N-Bz protected, rC was N-Ac protected, and rG was N-iBu protected. The coupling step was done using 5-ethylthio-1H-tetrazole solution (0.25 M) in acetonitrile for 12 min. The 5′-detritylation step was done using 3% trichloroacetic acid in CH₂Cl₂. Oxidation was done using I₂ (0.02 M) in a THF/pyridine/H₂O solution. CPG modifications were carried out using native amino lcaa CPG 1000 Å, purchased from ChemGenes (Wilmington, MA, USA), Cat.# N-5100-10.

Standard Oligonucleotide Purification Procedure

HPLC analysis of RNA1, RNA2, and RNA3 was performed on a Shimadzu LC-20 Instrument, equipped with a DNAPac PA200 BioLC analytical column (Thermo Scientific). Running buffer A contained 20 mM TRIS Base, acidified with conc. HCl to a final pH of 8. Running buffer B contained 20 mM TRIS Base and 1.25 M NaCl, pH 8. The gradient (0–90% buffer B) employed at 1.2 mL per min flow rate was utilized with an oven temperature of 60 °C.

HPLC analysis of the RNA4 and RNA5, containing phosphorothioate backbone RNA was performed on Shimadzu Nexera 40 Instrument, equipped with Kinetex 2.6 μm XB-C18 100 Å column (Phenomenex) using previously reported methods. (31,32) Running buffer A contained 14.3 mM triethylamine and 114 mM hexafluoroisopropanol dissolved in water containing 2.5% v/v CH₃OH. Running buffer B contained 14.3 mM triethylamine and 114 mM hexafluoroisopropanol dissolved in a 3:2 solution of CH₃OH:H₂O (v/v). The gradient (0–80% buffer B) employed at 0.6 mL per min flow rate was utilized with a Kinetex 2.6 μm XB-C18 100 Å column and an oven temperature of 65 °C.