Increased brightness of fluorescent uridine, qU, inside single- and double-stranded RNA

Abstract

Fluorescent base analogues enable studies of nucleic acids and by expanding their range of properties the scope of potential applications broadens. The recent photophysical characterisation of quadracyclic uridine (qU) in its monomeric ribonucleotide form revealed a high fluorescence quantum yield of 27%, with an interesting dual-band character emission. In this study, we show that incorporation of qU into RNA results in loss of the dual-band character but with further increased quantum yields and longer fluorescence lifetimes, in both single- and double-stranded RNA. Additionally, we find that neighbouring bases of qU affect its photophysical properties, with quantum yields varying between 33 and 68% in single-strands and 30–51% in duplexes. Furthermore, circular dichroism and duplex melting analyses of the qU-labelled duplexes were compared to unlabelled counterparts, with results suggesting that A-form RNA conformation is maintained, albeit with destabilisation caused by qU. Spectral features indicate that incorporated qU predominantly exists in its iminol tautomer, incapable of Watson-Crick base-pairing, which likely strongly contributes to the observed destabilisation. Overall, we report one of the highest brightness values (reaching 6 000 M^− 1 cm^− 1) to date for an RNA-incorporated fluorescent nucleobase, highlighting qU as a valuable label for applications such as fluorescence microscopy that require strong emission.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-43188-2.

Introduction

RNA plays a crucial role in living organisms in a wide range of cellular processes. Beyond its well-known role in gene expression as a coding template, discovered in 1961¹, the last decades of RNA research have unveiled its diverse roles beyond being a messenger (mRNA) between the DNA code and protein synthesis. TransferRNA (tRNA) and ribosomal RNA (rRNA) are vital biomolecules of protein synthesis^2,3 whereas other RNAs, including ribozymes and small nuclear (snRNA) or nucleolar (snoRNA) RNAs play a role in RNA processing^4,5. Furthermore, long non-coding RNA (lncRNA), microRNA (miRNA) and riboswitches are important non-coding RNAs for the regulation of gene expression^6–8. Improved understanding of the diverse and essential roles of different types of RNA has contributed to the emergence of RNA as a versatile therapeutic modality, with the potential to treat or even cure diseases previously considered untreatable. Its significance in endogenous biological processes and clinical as well as therapeutic innovation has expanded the need for improved methods in both basic and applied RNA research.

Fluorescence-based techniques are versatile for RNA research in multiple disciplines, from fundamental studies on RNA folding and intermolecular interactions, to applied research assessing for instance pharmacokinetics and drug formulations⁹. Central to all these applications is the need for reliable and efficient methods to label RNA with fluorescent dyes while preserving its biological function. Covalent labelling of RNA with conventional fluorophores such as fluorescein, rhodamine or cyanine provides bright emission for microscopy and spectroscopy. However, these dyes are chemically very different from nucleic acids, such that their physicochemical properties may alter the behaviour of the labelled RNA, for instance by changing its interactions or intracellular trafficking^10–13. Consequently, RNA research efforts the past decades have concurrently focused on the development of small, base-mimicking dyes that are internal to the base-stack of the RNA and, thus, potentially better at preserving the natural properties and function of the RNA of interest^12–15. Such fluorescent base analogues (FBAs) mimic not only the overall size and pi-stacking capability of natural nucleobases but are often also designed to maintain the Watson-Crick base pairing capability. It has been demonstrated that fluorescent base analogues can be used to spatiotemporally study RNA in cells^10,16, and to investigate immediate drug-target interactions and conformation studies in vitro^17–19. Nevertheless, most base analogues have relatively low brightness (< 500 M^− 1 cm^− 1) and absorb in the blue range, making them less straightforward to use in, for example, image-based applications. Therefore, development and characterisation of new fluorescent base analogues represent an important field of research to enable new and improved investigations in nucleic acid research.

Known as the first fluorescent base analogue, 2-aminopurine (2AP) was synthesised already in the late 1950 s and is nowadays commercially available and widely used in biochemical studies as an adenine analogue, capable of base-pairing with thymine but also in wobble base-pair with cytosine. Its 100-fold decrease in quantum yield (QY) upon double strand formation and general sensitivity to local environment can be wisely used to study important molecular interactions yet also implies limitations to the use of 2AP inside nucleic acid contexts where high brightness is crucial for the experiment²⁰. Moreover, its lowest-energy absorption maximum at 300 nm is close to that of natural RNA and outside the excitation range of many instruments used for biochemical applications, further limiting its use. Over the last decades, new and promising base analogues with properties in several aspects more useful than those of 2AP, have been developed^12–15,21.

Highly emissive fluorescent base analogues such as 8vG (λ_max = 277 nm)²², 8vA (λ_max = 290 nm)²³ and a broader family of RNA-mimicking fluorescent base analogues known as the isomorphic RNA alphabet have greatly expanded the scope of fluorescent base analogues to study nucleic acids²⁴. Among these isomorphic analogues, ^thG (λ_max = 321 nm) stands out as a highly efficient guanosine mimic with high QY in both DNA and RNA^25,26. ^thG has been employed in various applications including monitoring DNA conformational dynamics, base flipping, DNA-protein interactions, and siRNA cellular activity^26–29. Another notable and commercially available analogue, pyrrolo-dC³⁰, has recently undergone detailed characterisation in nucleic acid contexts³¹. This environment-sensitive analogue has proven useful in studies of DNA repair³².

Efforts in the design of fluorescent base analogues have also yielded several thymine and uracil analogues^33,34. For instance, furan-modified uracil (FU, λ_max = 316 nm) and thiophene-modified uracil (TU, λ_max = 314 nm) are environment-sensitive base analogues that can be incorporated in DNA and RNA while preserving the native secondary structures^35,36. FU has been particularly valuable for probing nucleic acid groove polarity and mismatch detection, and has demonstrated high fidelity in enzymatic incorporation^37,38. A similar strategy has led to the development of 5-modified thymine analogues^39,40, among which benzofuran-modified uracil (BFU) serves as an effective real-time reporter of DNA polymerase activity during incorporation³⁹.

Recently, single- and dual-emissive thymine analogues have been developed for FRET-based applications^41–44. These analogues exhibit strong absorption (~ 40,000 M⁻¹ cm⁻¹) in the visible range and moderate QYs in aqueous environments upon excitation at 390 nm. When incorporated in DNA, they cause only minor reductions in melting temperature and offer excellent ratiometric responses, though their QYs remain relatively low and sequence-dependent^41,42. A highly emissive thymine analogue (ABN) suitable for single-molecule DNA detection has since been introduced^45,46. ABN offers exceptional brightness and visible-range absorption, but its incorporation can significantly destabilise duplexes and shows limited base-pairing selectivity. Notably, bright base analogues absorbing in the visible range (TCU, FCU, ABN) have thus far exclusively been developed for oligonucleotide labelling of DNA specifically.

Our group have developed several fluorescent base analogues with a particular emphasis on tricyclic cytosines (tCs) and quadracyclic adenines^14,15. tC and tC^O exert minimal perturbation on the secondary structure and Watson-Crick base-paring with guanine, while also exhibiting beneficial photophysical properties including high and relatively environment-insensitive brightness values^47,48. With a broad absorption band centred at 360 and 375 nm, respectively, they are both easily excitable outside the nucleic acid absorption band and, moreover, sufficiently bright to visualise RNA inside cells in confocal microscopy using the 405 nm laser line common in many microscopy set-ups^10,16. tC^O together with the non-emissive tC-derivative tC_nitro has been presented and used as the first interbase FRET pair for detailed DNA and RNA conformation studies^17,49,50. More recently, tC^O has also been synthesized in its triphosphate form and used in in vitro transcription to produce long bright RNAs that have found applications in real-time monitoring of cellular uptake of RNA therapeutics in live cells¹⁶. This demonstrated that tC^O is accepted as a good C analogue by several enzymes.

The family of quadracyclic adenines (qAs) has been reported as beneficial base analogues of adenine, and several of them exhibit significant to high brightness values for being base-mimicking dyes^18,51,52. The most interesting of our adenine analogues to date is 2CNqA, which, unlike tC^O and tC, displays fluorescence lifetime and QY slightly sensitive to its microenvironment¹⁸. We recently used fluorescence lifetime imaging (FLIM) to show that free 2CNqATP could be distinguished from RNA-incorporated 2CNqA, and therefore, for example, be useful for RNA metabolism studies⁵³. Like the tricyclic cytosines tC^O and tC, 2CNqA has a broad absorption band centred around 370 nm^18,51, and can be visualised in microscopes using the 405 nm laser^16,54. Furthermore, together with the FRET acceptor qA_nitro, 2CNqA has been developed into the first interbase adenine FRET pair¹⁸. More recently, it was found that 2CNqATP is able to spontaneously enter non-engineered cultured human cells and be metabolically incorporated into their endogenous RNA⁵⁴. This was the first report of such unassisted metabolic fluorescence labelling of any emissive label, and showed that 2CNqA is tolerated as an A analogue by a range of enzymes⁵⁴.

In efforts to design a base analogue with high brightness, more redshifted absorption and possible pH dependent emission, our group recently published the synthesis and photophysical characterisation of a quadracyclic uracil, qU. We unveiled its promising brightness with a molar absorptivity for the low-energy band at 380 nm of 9 600 M^− 1 cm^− 1 and QY of 27% in aqueous buffer at pH 7⁵⁵. The redshifted absorption of qU compared to many fluorescent base analogues makes qU a suitable fluorescence label for the 405 nm laser line. Moreover, the same study revealed that the photophysical properties of qU were pH dependent. With such promising photophysical characteristics of qU as a monomer, this quadracyclic uridine has arisen as a versatile candidate for internal fluorescence labelling of RNA.

Photophysical characterisation, NMR-spectra and quantum chemical calculations revealed that the qU chromophore free in aqueous solution predominantly exists in its non-Watson-Crick base pairing iminol tautomer at neutral pH (Fig. 1). However, previously developed base analogues have shown tautomerisation being highly dependent on nucleic acid-incorporation and -hybridisation⁵⁶. This indicates a possibility that qU incorporated into RNA is pushed towards its amide tautomer, resulting in the capability to base-pair with adenine. (Fig. 1)⁵⁵. We herein report the synthesis of the phosphoramidite of qU, its use and behaviour in standard oligonucleotide solid-phase synthesis and explore its photophysical behaviour when incorporated into short RNA sequences with varying neighbouring bases. In addition, we hybridise the qU-labelled oligonucleotides with their complementary RNA sequences to explore its uridine-mimicking capability in terms of base-pairing properties and effect on secondary structure and find a duplex-destabilising effect of qU. Moreover, we characterise variations in photophysical behaviour upon hybridisation and demonstrate that the high brightness of qU as a monomer is further considerably increased in all single- and double-stranded RNA contexts investigated.

Fig. 1.

Molecular structure of natural Watson Crick base-pairing between adenine and uracil (left), and between adenine and qU in its amide tautomer (middle). qU in its iminol tautomer shown to the right. R = RNA ribose-phosphate backbone. The modification of uridine to yield qU is highlighted in cyan.

Results

Synthesis of the qU phosphoramidite and preparation of oligonucleotides

With the aim of investigating the photophysical behaviour of qU inside oligonucleotides and its impact on the RNA structure and stability, we first needed to develop an optimized synthesis route towards the phosphoramidite of qU (Fig. 2, details in SI, Fig. S11 and forward) which is the required derivative for solid-phase synthesis of RNA oligonucleotides. Starting from the diastereomeric mixture 3 from our previously reported synthesis of qU⁵⁵, the amide functionality was initially protected using TIPS-OTf.

Fig. 2.

Synthetic route towards the qU phosphoramidite (8) required for further RNA oligonucleotide synthesis.

This protection facilitated the complete separation of the two diastereoisomers, allowing for the isolation of compound 4 with the desired β-conformation, as confirmed by NOESY analysis (see NOESY spectra in SI). Subsequent treatment of compound 4 with a catalytic amount of MeONa in MeOH yielded the TIPS-protected nucleoside 5 quantitatively, without the need for purification. Compound 6 was obtained via standard 5’-DMT protection of 5 in 70% yield. In the following step, regioisomers bearing 2’- and 3’-TBDMS protections were formed, from which compound 7 was isolated in 53% yield. Finally, phosphoramidite 8 was synthesised from compound 7 using established procedures. Overall, the desired phosphoramidite was obtained from starting material 1 in 11% yield over a ten-step procedure.

Using phosphoramidites of the natural nucleosides and 8, we synthesised short RNA sequences with varying neighbouring bases to qU (see Methods and SI, Fig. S11 and forward, for details) to enable investigations of the effect of microenvironment on qU photophysics. qU-modified sequences have the general sequence 5’-CGCAXqUYAUCG-3’ where X and Y denote different nearest neighbours to qU (details in SI, Table S1). These strands were characterised as single- and double-strands, i.e. hybridised with their complementary strands or with mismatched sequences (Table S1) where the opposite base to qU is exchanged from A to C, G or U, respectively.

Absorption and emission spectra of qU incorporated into RNA

Absorption and emission of RNA-incorporated qU at neutral pH were measured to compare qU’s spectral features with the monomer at neutral pH published by Le et al.⁵⁵. As shown in Fig. 3, incorporation of qU into an RNA strand overall maintained its absorption spectral features, with a 5 nm redshift of the 380 nm peak maximum.

Fig. 3.

Absorption (solid lines) and emission (dashed lines) spectra of qU-modified sequences with varying neighbouring nucleobases. Sample name in legend denotes qU as X and indicates the direct neighbouring bases. (A) single-stranded and (B) double-stranded sequences (legend at top). (C) mismatches where the UXC sequence was hybridised with the fully complementary sequence (A opposite to qU, Match) or with mismatch sequences (Mis U, Mis C, and Mis G, where A is exchanged to the indicated base). qU ribonucleoside (Monomer) at neutral pH shown in black for comparison, taken from Le et al.⁵⁵. Absorption spectra are normalised to the peak absorption at 380 and 385 nm of monomer and incorporated qU, respectively, and emission spectra are normalised to the emission peak at ca. 450 nm. Sequence details in SI.

Overall, the emission maxima remain at 440 nm but lack the emission band at 530 nm that is present for the qU monomer and was suggested in Le et al. to originate from the amide and deprotonated form of qU⁵⁵. Figure 3A shows that for single-stranded sequences, varying the neighbouring bases does not affect the absorption spectra of qU substantially. However, it should be noted that the long-wavelength absorption (> 425 nm) present for the monomer, and which originates from amide and deprotonated species⁵⁵, is substantially supressed or missing for qU incorporated in RNA. Upon hybridisation with their complementary RNA strands, some of the qU absorption above 425 nm reappears. This is especially pronounced when qU is surrounded by two purines (A or G) as nearest neighbours. The position of this absorption band, at 440 nm, corresponds better to the deprotonated than the amide species present in the qU monomer under similar conditions⁵⁵. As mentioned above, the 530 nm emission band that is present for the monomer, is highly supressed inside RNA single- and double-strands irrespective of immediate neighbouring bases yet slightly less supressed in the case of surrounding pyrimidines (Fig. 3A, B).

The UXC sequence was also hybridised to sequences in which the opposite base to qU is a mismatch according to Watson-Crick base-pairing. The absorption of the C and U mismatches show similar trends to the matched sequences (Fig. 2C). However, a mismatch with G results in increased absorption at 440 nm, similar to that of the qU monomer. Compared to the monomer, mismatching cause a substantial decrease in the 530 nm emission, like hybridisation with the matched sequences do (Fig. 2B).

Fluorescence lifetimes and quantum yields of qU incorporated into RNA

By recording time-correlated single photon counting (TCSPC), absorption and steady-state emission, we observe that qU not only maintains its high brightness from the monomeric form but also further increases the quantum yield when incorporated into RNA (Table 1).

Table 1.

Average fluorescence lifetimes (τ_avg), fluorescence quantum yields (QY) and melting temperatures (T_m) of strands with varying nearest neighbouring bases to qU, indicated from 5’ to 3’ in the first column. Full sequences and melting curves can be found in SI. ΔT_m is the difference in melting temperature of qU-modified (mod) and -unmodified (unmod) double strands as determined by absorption spectroscopy. Mismatch sequences are UXC with the qU-pairing base in the complementary strand exchanged from A to the indicated base (C, G or U). Average (avg) values for single- (ss) and/or double-stranded (ds) RNA with their complementary strand (match) given in the bottom along with the average values for mismatched (mis) sequences. Strands are sorted from highest to lowest average fluorescence lifetime in single strands. Presented are mean values ± standard deviations.

Strand	τavg, ss	τavg, ds	QY, ss	QY, ds	Tm, mod	Tm, unmod	ΔTm
	ns	ns	%	%	°C	°C	°C
CXA	9.5 ± 0.3	3.5 ± 0.4	68 ± 1	47 ± 3	49 ± 2	55 ± 1	−6 ± 2
AXA	9.3 ± 0.2	9.8 ± 0	66 ± 2	51 ± 1	37 ± 1	47 ± 1	−11 ± 1
UXA	9.1 ± 0.1	5.4 ± 0.2	62 ± 3	49 ± 3	41 ± 1	47 ± 2	−6 ± 2
AXC	8.3 ± 0.2	6 ± 0	60 ± 1	45 ± 2	56 ± 1	56 ± 1	−14 ± 2
CXC	7.7 ± 0.1	6.4 ± 0	54 ± 1	42 ± 2	53 ± 1	60 ± 5	−7 ± 5
UXC	7.1 ± 0.1	6 ± 0	55 ± 2	37 ± 1	46 ± 2	55 ± 1	−9 ± 2
GXG	7 ± 0.1	7.8 ± 0.1	51 ± 1	51 ± 5	52 ± 2	61 ± 1	−9 ± 2
GXU	5.7 ± 0	7.3 ± 0.1	43 ± 1	49 ± 0	40 ± 1	54 ± 1	−14 ± 2
UXG	5.5 ± 0.1	5.6 ± 0	42 ± 1	30 ± 2	48 ± 1	54 ± 1	−6 ± 2
AXU	4.8 ± 0.1	9.4 ± 0	34 ± 1	40 ± 0	46 ± 1	35 ± 2	−12 ± 2
UXU	4.5 ± 0.1	4.3 ± 0.1	33 ± 1	32 ± 0	39 ± 1	46 ± 1	−7 ± 2
Mismatch C	-	7.7 ± 0	-	69 ± 2	47 ± 1	41 ± 1	6 ± 1
Mismatch G	-	6.5 ± 0.1	-	57 ± 0	36 ± 1	51 ± 1	−15 ± 2
Mismatch U	-	8.1 ± 0.1	-	72 ± 3	47 ± 2	41 ± 1	6 ± 2
Avg, match	7.1 ± 1.7	6.5 ± 1.9	51.6 ± 11.6	43 ± 7	45.9 ± 5.9	51.6 ± 7.2	−9.1 ± 3
Avg, mis	-	7.4 ± 0.7	-	66 ± 6	43.4 ± 5.4	44.2 ± 4.8	−0.9 ± 10

The increased brightness of qU upon RNA incorporation is an uncommon but useful property of internal RNA fluorescence labels. Most fluorescent base analogues emit considerably less inside nucleic acid contexts^12–15,21. Our investigations show a longer average fluorescence lifetime compared to the monomer at pH 7 (τ_{avg, monomer} = 2.44 ns), irrespective of the neighbouring bases, in single- and as double-stranded RNA. As expected, the increase in average fluorescence lifetime is accompanied with higher QY for qU in all RNA contexts compared to qU as a monomer at pH 7 (QY = 27%). With adenine on the 3’ side, the average fluorescence lifetimes are longer than 9 ns and the QYs are higher than 60% in single strands. Three sequences with a C on the 3’ side of qU have the following longest fluorescence lifetimes and highest QYs, whereas the 3’-trend is not as clear for neighbouring G and U. Furthermore, we observe an overall trend that the fluorescence lifetime and QY of qU agree well between different single-stranded oligomers, as expected. However, upon hybridisation to duplexes, the trend is less clear. For instance, CXA as a single strand gives rise to the longest average fluorescence lifetime (9.5 ns) but after hybridisation results in one of the shortest (3.5 ns) whereas its QY remains amongst the highest as a duplex (47%).

Interestingly, we note that hybridising qU with a sequence where the opposite base in the complementary strand is exchanged from A to one of the other natural bases, the QYs increase from 37% (match) to 57% (mismatch G), 69% (mismatch C), and 72% (mismatch U), which are the highest reported for qU. Furthermore, the two highest QYs and longest fluorescence lifetimes of double-stranded qU-labelled RNA, namely mismatch C and U, result in higher melting temperature (vide infra) than their unmodified counterparts, meaning that these two mismatch cases stabilise the duplexes modified with qU compared to the mismatched duplexes containing natural U. Mismatch G also has a higher QY than the matched sequences, yet on the contrary also the most destabilising effect with qU (ΔT_m = − 15 °C, vide infra).

RNA duplex melting

By recording UV absorption at 260 nm over a temperature range and calculating the melting temperature (T_m), we observe that qU decreases the melting temperature for all duplexes with varying neighbours of qU, implying that exchange from U to qU destabilises RNA duplexes on average by 9 ± 3 °C (Table 1). The decreased melting temperature varied between 6 and 14 °C, depending on qU neighbours. We find that a purine on the 5’ side of qU generally results in larger decreases in melting temperature (ΔT_m of − 9 to − 14 °C). Notably, U on the 5’ side of qU seems to cause not only higher non-radiative decays (vide supra) but also less destabilising effect on duplex formation (ΔT_m of − 6 to − 9 °C). Interestingly, the duplex hypochromicity is smaller for sequences modified with qU (see melting curves in SI) compared to unmodified duplexes, suggesting that qU induces a less perfect duplex formation, which may be explained as local melting inside the duplex.

Circular Dichroism

To further investigate the structural impact of exchanging a natural U with a qU in a 10-mer RNA duplex, we measured circular dichroism (CD) spectra for all modified duplexes and their natural unmodified counterparts. All CD spectra display a positive band around 250–300 nm, a shoulder at 210 nm, and a strong negative band around 245 nm with no CD signal outside the nucleic acid wavelength range and long-wavelength qU absorption band (Fig. 4 and Fig. S8-9). This CD fingerprint indicates that the duplex exists in A-form RNA. Furthermore, it supports the conclusion from the melting temperature measurements that the RNA forms duplexes also when one U is exchanged with a qU in a 10-mer duplex. It should also be noted that there are differences in amplitudes and slight shifts in the CD spectra when comparing the modified to the unmodified duplexes, which will be discussed below.

Fig. 4.

Circular dichroism spectra of double-stranded samples with varying neighbouring bases to qU for modified sequences (red, X indicates qU) and unmodified sequences (black, X corresponds to natural uridine). Sequence indicated in graph, compare Table S1 for sequence details. (A) Showing an example duplex where qU has induced a redshift in the peak around 260 nm, blue shift in the negative peak around 210 nm and a decreased amplitude, (B) showing an example of where CD of the modified and unmodified duplexes differ minimally, with just a slight decrease in amplitude around 260 nm. All circular dichroism spectra, in full range, can be found in SI.

pH effects on the photophysics of RNA-incorporated qU

Given that the qU monomer exhibits pH responsiveness⁵⁵, we examined whether this property persists when qU is incorporated into RNA. At pH 4, absorption and emission spectra were largely unchanged relative to neutral pH; however, fluorescence lifetime and quantum yield (QY) decreased substantially (Fig. S3 and Table S4-S5). Elevated pH similarly reduced QY and shortened lifetimes compared to pH 7 (Table S5). Under basic conditions (pH 10), single-stranded sequences showed a 20 nm redshift in the lowest-energy absorption band relative to pH 7 (Fig. S2A), consistent with the spectral signature of the anionic species reported previously⁵⁵. Notably, the emission of qU when flanked by U and C on both the 3’ and the 5’ sides displayed essentially no shift in emission maxima at pH 10 compared to at neutral pH whereas G or A neighbouring one or both sides caused a 40 nm redshift (Fig. S2 B). When qU was flanked by different bases, the neighbour on the 5’ side determined the emission maximum position of that RNA but there is also an influence of most pyrimidines in the 3’ position when there are purines in the 5’ position (Fig. S2 C).

To further probe these effects, we analysed three sequences (AXC, UXA, UXU) at pH 9. None displayed the 40 nm emission shift observed at pH 10 (example in Fig. S3A, B). For single strands, absorption featured peaks at 385 and 410 nm (example in Fig. S3A), indicative of a mixture of anionic and iminol forms. Upon hybridisation with complementary strands, the 410 nm peak disappeared (example in Fig. S3 B).

Discussion

In this work we investigated the characteristics of qU as internal fluorescence label in single- and double-stranded RNA. To this end, we developed a new synthetic route towards the phosphoramidite of qU (Fig. 2, compound 8) and incorporated it into RNA oligonucleotides using standard solid-phase synthesis. By performing CD and duplex melting studies, we investigated how an exchange from natural U to qU in those oligonucleotides affect structure and stability of qU-modified RNA double strands compared to the unmodified sequences. Finally, and most importantly, we investigated the photophysics of qU inside single- and double-stranded RNA using absorption and steady-state as well as time-resolved fluorescence spectroscopy.

Photophysics of qU incorporated into RNA

RNA-incorporated qU is to this date the brightest fluorescent uridine reported inside RNA. The quantum yields of qU with all neighbouring bases investigated in this study were high both for single (33–68%) and double strands (30–51%) (Table 1) and, as previously reported, qU has a high molar absorptivity for this class of labels (9600 M^− 1 cm^− 1 as a monomer at 380 nm)⁵⁵. In duplexes, the brightness reaches values of approximately 4000 M^− 1 cm^− 1 which matches and surpasses other bright fluorescent bases like tC^O, tC, pA, ^thG, and 2CNqA which have brightness values of between 310 M^− 1 cm^− 1 for ^thG and 2000 M^− 1 cm^− 1 for tC^O inside nucleic acid contexts^18,26,57. However, qU is plausibly insufficiently bright for single-molecule detection, unlike the fluorescent base ABN^45,46. Moreover, with a peak molar absorptivity at 385 nm (shifted 5 nm compared to the monomer, Fig. 3), qU inside RNA can be excited using instruments with a 405 nm laser line. As we have previously reported for two of our other fluorescent base analogues, tC^O and 2CNqA, the excitability at 405 nm is a crucial property for various biological applications, for example, in live-cell confocal microscopy^10,16,54. The slight redshift of qU is advantageous for such applications. As noted above, a single qU incorporation is unlikely to achieve single-molecule (read single RNA strand) brightness; however, incorporating multiple qU bases into the same RNA strand could enable detection of individual RNA molecules. In such future designs, careful spacing of qU residues will be essential to prevent self-quenching, which can be mitigated by separating qU labels with natural nucleotides. In our previous work on fluorescent base analogues with similar photophysical properties, we observed negligible self-quenching when labels were spaced by at least three natural bases (data not published), and we have seen comparable indications when introducing the analogues tC^O and 2CNqA into long RNAs using in vitro transcription^10,58.

On a more detailed level, we observe that for the fluorescence quantum yields and lifetimes (Table 1) the neighbouring bases affect qU in single strands to different extents. Additionally, not only the identity, but also their position, i.e. whether on the 5’ or 3’ of qU, affects the photophysical characteristics. For instance, single-stranded UXA has a quantum yield of 62% and a long average lifetime (9.1 ns) while AXU has a quantum yield of 34% with one of the shortest average lifetimes (4.8 ns) of all measured single strands. This phenomenon has been observed for other fluorescent base analogues¹⁵ and is attributed to the different geometrical arrangement of the surrounding bases in relation to the fluorescent base analogue, caused by the twist of the RNA. qU in single strands with A as its 3’ neighbour display high quantum yields (QY = 62–68%). Previous characterisations of the neighbouring effects on the adenine analogues 2CNqA and pA have unveiled that neighbouring A generally cause high QYs in duplexes^15,18. However, this is not a general trend for QYs of fluorescent base analogues in single strands. The time-resolved fluorescence decays measured for all single strands demand three exponentials for a good fit (Table S2), like the fluorescence decay of qU monomer, which was interpreted to be an effect of the different qU species; iminol, amide and deprotonated⁵⁵. However, the three lifetimes found here (typically 0.3 ns, 4 ns, and 11 ns; Table S2) are not similar to the ones reported for the monomer at pH 7 (0.6 ns, 3 ns, and 6 ns)⁵⁵, which indicates that the microenvironment of qU in single- and double-stranded RNA considerably changes the excited state processes of qU while maintaining the dominant iminol tautomer of qU (vide infra).

Upon hybridisation with its complementary sequence, the fluorescence quantum yield of qU decreases for all sequences except GXG, GXU, AXU (Table 1). Moreover, an alteration in quantum yield going from single strands to duplexes is, in all cases but two (AXA and UXG), accompanied by a corresponding expected alteration in average fluorescence lifetime. The pattern of high fluorescence quantum yields with certain neighbouring bases in single strands is not obvious for duplexes. A possible explanation for this is that in single strands, qU can induce a certain geometry for its surrounding bases irrespective of their nature, while the duplex formation imposes an overall A-form of the entire complex, though the local A-form around qU may be incomplete (vide infra). The quantum yields of qU in duplexes with mismatching bases opposite to qU are generally higher (QY = 57–72%; Table 1, Mismatch C, G, U) than for all duplexes where qU faces an A in the opposite strand (QY = 30–51%), and instead more similar to the high values of single strands (QY: 33–68%; Table 1). This suggests that qU is less restricted in its geometry, i.e. it must not accommodate to the A-form, in these mismatched cases and, hence, would have the freedom to stack with surrounding bases more like in the single strands. The time-resolved fluorescence measured for duplexes need between two and four exponentials for a good fit (Table S3), where the long lifetime observed in single strands (ca. 11 ns) is also found for the duplexes. The second longest lifetime (around 3.5–6 ns) is observed in many cases whereas the shortest one (ca. 0.3 ns) is absent in most cases.

The long wavelength absorption and emission bands ascribed to the amide and anionic species of the qU monomer in the previous study⁵⁵ are generally supressed for qU incorporated in an RNA oligonucleotide, with further decrease upon duplex formation (Fig. 3). This considerable decrease indicates that the RNA context favours the non-base-pairing iminol tautomer of qU, with duplex formation further favouring the iminol, unlike what we expected. These observations go hand in hand with the structure and stability aspects of the duplexes (vide infra). It should be noted that the duplexes with the highest QYs (Table 1; GXG, AXA, UXA, and GXU) display slightly increased absorption at 440 nm (Fig. 3); an absorption region which, for the monomer, is attributed to the amide tautomer⁵⁵.

Interestingly, the absorption of qU in duplex mismatch G displays the closest resemblance with the monomer at long wavelengths (Fig. 3), hence suggesting the presence of the amide and deprotonated species. This duplex has a quantum yield of 57% which is higher than all matched duplexes, closer to the values for the single strands yet lower than those of the other mismatches (C: QY = 69%; U: QY = 72%).

We find that qU inside RNA is, like the monomer, responsive towards changes in pH, especially at pH 10, where also the spectral features of the absorption and emission change substantially. The reduction in quantum yield (QY) at pH 9 aligns with previous observations for the monomer at high pH⁵⁵. In contrast, while the monomer showed no QY decrease at low pH⁵⁵, qU in RNA exhibited a significantly lower QY at pH 4 compared to neutral conditions. We attribute this to RNA conformational changes driven by base protonation below pH 5. Additionally, fluorescence lifetimes decreased markedly under acidic conditions; for example, single-stranded AXC dropped from 8.2 ns at neutral pH to 2.1 ns at pH 4. Such dramatic lifetime changes suggest that qU could serve as a sensitive pH reporter in FLIM-based studies, such as tracking RNA uptake via endocytosis, where endosomal pH transitions from neutral to ~ 5–5.5 in late endosomes and ~ 4.5–5 in lysosomes⁵⁹.

Structure and stability effects of qU in RNA duplexes

As discussed in the photophysical section, the spectral features of qU inside RNA indicate an enhanced dominance of the iminol tautomer of qU upon incorporation into an RNA oligonucleotide (Fig. 1). Because the iminol form cannot base-pair with A, the presence of qU possibly destabilises the duplex formation, which agrees with the decreases in duplex melting temperatures (Table 1). The decrease in duplex melting temperature when exchanging U to qU is present in all studied cases, with an average decrease of 9 °C. This considerable destabilisation stands in contrast to slight FBA-induced stabilisation (1–4°C) we previously characterised in our group; pA¹⁵, 2CNqA¹⁸, tC⁶⁰ and tC^O61, but in line with observed destabilisation for other FBAs^12–15,21, including a 14 °C destabilisation for ABN^45,46. Like the quantum yields and fluorescence lifetimes, the destabilisation displays a sequence dependency. With a purine base on the 5’ side of qU, the duplexes are more destabilised ( °C) than with 5’ pyrimidine ( °C). Furthermore, the striking stabilisation of 6 °C increase caused by mismatch U and C (Table 1) suggests that qU, in contrast to U, is able to form hydrogen bonds with U and C. It could additionally indicate that with pyrimidines in the opposite RNA strand there is significant space for the large qU, with its four rings instead of one for natural U, to be accommodated in an energetically favoured pi-stacking geometry in the RNA base stack.

In the melting temperature experiments, a smaller hypochromicity at 260 nm is found upon formation of qU-containing duplexes compared to formation of the corresponding duplexes containing a natural U (melting curves Fig. S6). This suggests that qU-containing duplexes are not just less stable but additionally less well-stacked. Since the iminol tautomer is unlikely to base-pair with adenine, qU may have significantly increased base-flipping compared to U. This would disturb the base-stacking not only for the qU-position of the duplex but possibly also for the neighbouring base pairs, resulting in the observed lower hypochromicity for the modified duplexes. While signs of the Watson-Crick base-pairing amide tautomer of qU is found for certain sequences in both absorption (440 nm) and emission spectra (530 nm) (Fig. 3), these sequences do not display less destabilisation.

Circular dichroism (CD) spectra of both unmodified and the corresponding qU-modified RNA duplexes display the general CD features of A-form RNA (Fig. S4) and comparing the spectra of the unmodified to the qU-modified indicate minimal perturbation to the secondary structure by qU. However, all qU-modified duplexes cause a weaker CD signal compared to their unmodified counterparts, despite qU having a higher molar absorptivity (Fig. S10)⁵⁵. This observation may suggest a reduced overall presence of A-form helices, particularly in terms of achieving a fully formed A-type structure across the entire 10-mer duplex, when qU is incorporated. It should also be noted that in the regions where only qU absorbs (> 300 nm), there is no CD signal present, hence, the helical environment does not seem to induce CD in the qU chromophore (Fig. S8-9). This suggests that qU is geometrically flexible with an increased base-flipping, which was also suggested by the T_m measurements (vide supra). However, there is no strong evidence to support the notion of increased base flipping since FBAs like tC^O61, pA¹⁵, and 2CNqA¹⁸, which are all good analogues of their natural counterparts, also lack such CD features.

Conclusion

In this study, we synthesised the phosphoramidite derivative of the fluorescent quadracyclic uridine, qU, which we subsequently incorporated into both single- and double-stranded RNA. Compared to the qU monomer at pH 7 (QY = 27%), incorporation into RNA results in significantly increased brightness with quantum yields of up to 70% and longer average fluorescence lifetimes. However, the neighbouring bases and whether qU is in a single or double strand is found to influence these fluorescence properties, with quantum yields ranging from 30% to 72% depending on the microenvironment. Furthermore, using both absorption and emission spectroscopy, we find that qU predominantly adopts its non-Watson-Crick base-pairing iminol form, regardless of the surrounding microenvironment. This tautomeric preference contributes to a destabilisation of RNA duplexes. Nevertheless, circular dichroism measurements show that qU-labelled RNA duplexes retain the overall native A-form helical structure. Despite inducing duplex destabilisation, qU remains a strong candidate for use as an internal RNA fluorescence label due to its high brightness. For instance, in FRET-based RNA conformation studies similar to what we have performed with our previously developed fluorescent base analogues, qU could be strategically placed in non-base-paired regions, such as hairpin loops, overhangs, or abasic sites, where Watson-Crick base-pairing is not essential. Furthermore, with the highest brightness ever reported for an RNA-incorporated fluorescent uridine and its ability to preserve the overall RNA secondary structure, qU holds strong potential as an RNA label that complements external fluorophores such as Cy or Atto dyes for applications requiring bright labels, e.g. fluorescence anisotropy and microscopy. Moreover, qU’s considerable brightness inside RNA in combination with its environment-sensitive fluorescence lifetimes may find applications in, for example, fluorescence lifetime imaging microscopy (FLIM).

Methods

Synthesis of oligonucleotides

Oligonucleotide synthesis was performed on a KnA synthesizer using a CPG support preloaded with rG(tac) (CPG 1000Å, 20–30 µmol/g) at a 1 µmol scale. The standard ribonucleoside amidites employed were rA(tac), rU, rC(Ac), and rG(tac), all bearing 5’-DMTr and 2’-TBDMS protections. Solid-phase synthesis followed standard protocols: 3% dichloroacetic acid (DCA) in dichloromethane (DCM) for detritylation, 0.25 M Activator 42^® for coupling, acetic anhydride in tetrahydrofuran (THF) and 1-methylimidazole in THF/pyridine for capping, and 0.02 M iodine solution for oxidation. All amidites, including compound 8, were dissolved in dry acetonitrile at a concentration of 0.1 M. Coupling times were set to 12 min for standard amidites and extended to 20 min for compound 8.

Following synthesis, cyanoethyl protecting groups were removed by treating the solid support with 20% dimethylamine in dry acetonitrile for 30 min. The labelled oligonucleotides were then cleaved and deprotected using 0.05 M potassium carbonate in dry methanol at room temperature for 8 h. Volatiles were removed under reduced pressure using a SpeedVac concentrator. The crude product was redissolved in DMSO (100 µL) at 65 °C, followed by the addition of triethylamine trihydrofluoride (125 µL). The mixture was incubated at 65 °C for 2.5 h, then cooled to room temperature. To precipitate the oligonucleotides, 25 µL of 3 M sodium acetate (pH 5.3) and 1 mL of n-butanol were added sequentially, and the mixture was stored at − 20 °C for at least 30 min. The precipitate was collected by centrifugation (10 min, 3900 rpm, 4 °C) and washed with 750 µL of 95% ethanol.

Final purification was carried out by ion-pair reversed-phase HPLC using an XBridge C18 column (5 μm, 19 × 150 mm). The mobile phases were: A—60 mM dibutylammonium acetate (DBuAA) in H₂O/ACN (95:5), pH 7; and B—60 mM DBuAA in acetonitrile. The gradient elution was as follows: 5% B for 1 min, 10–35% B over 12 min, and 35–80% B over 0.5 min.

All experimental details and analyses in SI.

Annealing of single strands were performed with 2.5 µM of each strand, with 10% excess of complementary strand in samples containing qU. Samples were hybridised in a water bath starting from room temperature (22 ˚C), heating to 85 ˚C for 10 min, holding at 85 ˚C for 15 min and cooling to 5 ˚C for 80 min. Samples were kept at 5 ˚C until stored in − 80 ˚C or used for measurements.

Photophysical characterisation

All measurements were performed in quartz crystal cuvettes with 10 mm optical path length at room temperature. Oligomers were measured in phosphate buffered saline, PBS (KCl 2.7 mM, KH₂PO₄ 1.5 mM, NaCl 138 mM, Na₂HPO₄*7H₂O 8 mM) at pH 7.4, or, when pH specified, in a McIlvaine pH 4 buffer (150 mM NaCl, 15 mM Na₂HPO₄, 12 mM citric acid, adjusted with HCl), pH 9 Tris buffer (150 mM NaCl, 30 mM Tris-Cl, adjusted with NaOH) or pH 10 buffer (10 mM sodium bicarbonate, 90 mM sodium carbonate, adjusted with NaOH). Unless otherwise stated, all measurements were performed at room temperature.

Concentrations of single-stranded oligomers were determined using the Beer-Lambert law and the absorbance value at 260 nm from absorption spectra measured on a Cary4000 or Cary3500 UV-vis Spectrophotometer (Agilent Technologies, CA, USA). Molar absorptivity values of the single-stranded oligomers were determined as the sum of molar absorptivity at 260 nm of constituent bases (adenine; 15 300 M^− 1 cm^− 1, cytosine; 7 400 M^− 1 cm^− 1, uracil; 9 300 M^− 1 cm^− 1, guanine; 1 800 M^− 1 cm^− 1, and quadracyclic uracil⁵⁵; 9 600 M^− 1 cm^− 1). Base-stacking hypochromicity was considered by multiplying the summed molar absorptivity with 0.9. Absorption spectra were recorded from 230 to 600 nm with 1 nm steps and 0.1 s integration time. Single- and double-stranded oligomers were measured at 2.5 µM and 5 µM strand concentration, respectively, to maintain a constant qU concentration (2.5 µM) throughout all samples. Each measurement was repeated two to three times.

Fluorescence quantum yields were determined as shown in Eq. 1 using the absorption value at 385 nm (A₃₈₅) and integrated emission () from 395 to 700 nm, with 2 nm slit width, 0.1 s integration and 1 nm steps, exciting at 385 nm through 2 nm slit width. Coumarin 102 in absolute ethanol was used as a reference, with a known fluorescence quantum yield () of 0.764⁶². Emission spectra were measured on a Spex Fluorolog-3 Spectrofluorimeter (JY Horiba). The refractive indices for sample (n_sample), and reference solvent (n_ref), were 1.334 and 1.333 respectively.

1

Fluorescence lifetime measurements were performed using time-correlated single-photon counting (TCSPC). Samples were excited with a 377 nm pulsed diode laser (PicoQuant, model LDH-P-C-375), featuring a spectral width (FWHM) of 1 nm and operated at a repetition frequency of 10 MHz via a PDL 800-B driver (PicoQuant). Emission detection was configured at 440 nm with a spectral bandwidth of 10 nm and an emission polariser oriented at the magic angle (54.9° relative to the excitation polarisation) to eliminate time-dependent anisotropy effects from the large biomolecule. Emitted photons were detected by a microchannel-plate photomultiplier tube (Hamamatsu R9809U 50 microchannels) and analysed using a LifeSpec II correlator (Edinburgh Analytical Instruments), employing 2048 channels with a resolution of 24.4 ps/channel. Data acquisition was set to stop upon reaching 10 000 counts in the peak channel. To characterise the instrument response function (IRF), the scattering from a water solution containing 0.01v% silica was measured. In the EasyTau2 software (PicoQuant), amplitude-weighted average fluorescence lifetimes (< τ >) were obtained by fitting IRF-deconvoluted multiexponential functions to the data (Eq. 2) in which τ represents the fluorescence lifetime of the n:th exponential component and α its amplitude. Fluorescence lifetime of double stranded samples was measured at 12 °C to assure complete hybridisation.

2

Melting curves were recorded on a Cary3500 (Varian Technologies) spectrophotometer using a 1 cm pathlength cuvette. Sample stirring in the cuvette was set to 500 rpm and temperature managed with a programmable temperature block monitored with a reference probe in phosphate buffer saline. Absorption at 260 nm was recorded every 1 ˚C, with integration time of 2 s. The samples’ temperatures were ramped up with a rate of 1 ˚C/min, kept at 90 ˚C for 5 min, cooled down to 5 or 20 ˚C at 1 ˚C/min, and kept for 5 min at the lowest temperature. This was repeated to achieve two full cycles, i.e. four temperature ramps. To avoid air bubbles, samples were degassed in the cuvettes under low pressure for 30 min prior to measurements. Nitrogen flow of 5 L/min was supplied to the instrument housing to prevent condensation outside the cuvettes. The first derivatives of the melting curves were smoothed in OriginLab with 5 ppt FFT filter and the melting temperatures defined as the peak maxima obtained from the peak analyser in OriginLab.

Circular dichroism spectra were recorded on a Chirascan CD spectrometer (Applied Photophysics) by integrating the intensity for 0.5 s at 1 nm intervals between 200 and 550 nm, with a total scan time of 5 min per spectrum, including instrument overhead. Phosphate buffered saline was recorded as baseline and used for background subtraction.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

A.F.E.K. performed fluorescence lifetime and quantum yield measurements. P.P. performed circular dichroism and absorption melting measurements. A.F.E.K. and P.P. analysed the data and wrote the manuscript under supervision of L.M.W. H.-N.L. and T.B. performed, under supervision of A.D., syntheses of the qU phosphoramidite and RNA oligonucleotides. A.D. and L.M.W. conceived the idea and outlined the project.

Funding

Open access funding provided by Chalmers University of Technology. This research was supported by and conducted as part of the FoRmulaEx research center for nucleotide delivery and with associated financial support to L.M.W. from the Swedish Foundation for Strategic Research (SSF) [IRC15-0065]; Swedish Research Council [VR, grant No. 2021–04409 to L.M.W.].

Data availability

Data available upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.