Oligodeoxynucleotides modified with 5-fluoro-2’-O-methyl-4-thiouridine (FSU) (ODN 1–3) (Table 1) were synthesized using "ultra-mild" DNA synthesis, as described previously.45
Table 1
Sequences of oligodeoxynucleotide (5’→3’) containing 5-fluoro-2’-O-methyl-4-thiouridine (FSU).
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 |
ODN 1 | C | G | A | T | A | C | G | A | FSU | A |
ODN 2 | A | FSU | A | G | C | A | T | A | G | C |
ODN 3 | A | T | A | G | C | A | FSU | A | G | C |
ODN 1–3 have the same nucleosides composition. In ODN 1, FSU is located as 9th nucleoside in the sequence (near 3’ end of strand), while in ODN 2 (it’s reversed sequence), FSU is placed as 2nd nucleoside of oligonucleotide (near 5’ end of strand). ODN 3 has the same sequence as ODN 2 but positions of FSU and T are converted.
Aqueous solutions of ODN 1–3 in 0.1M phosphate buffer, pH 7.0, were irradiated at 5℃, 20℃ and 35℃ and the progress of reaction was monitored by HPLC (Fig. 2). Under these conditions (λ = 355 nm) selective excitation of FSU is achieved. For all oligonucleotides, after irradiation with 355 nm for 150s the formation of the same photoproducts (a-d) with appropriate nucleosides sequence 1–3 (corresponding to ODN 1–3) was observed but with different yields (Fig. 2A, 2B, 2C). Since ODNs 1–3 have the same nucleoside composition, photoproducts (a-d) obtained from ODNs 1–3, respectively, have similar UV absorption spectra (Fig. 1S). For ODN 1, fluorescent photoproducts 1a and 1b were formed as the main photoproducts (Fig. 2A), while for ODN 2, only trace amounts of photoproducts 2a and 2b were formed (Fig. 2B). For fluorescence spectra of 1a and 1b see Supplementary Fig. 2S. Non-fluorescent photoproduct 2c was observed in the mixture after irradiation of ODN 2 as the dominant one (Fig. 2B), while the product with the same spectroscopic properties was also detected in the case of ODN 1 (1c) but with significantly lower yield. Moreover, this type of product was not observed after ODN 3 irradiation (Fig. 2C). The formation of photoproduct 2d or 3d was noticed in the mixture after irradiation of ODN 2 or 3, respectively (Fig. 2B, 2C). However, it was not formed in the case of ODN 1.
In the case of ODN 3 (Fig. 2C) the formation of intermediate photoproducts with retention times between 15–17 minutes was observed. Their absorptions spectra are presented in Fig. 3S. After performing an additional experiment of irradiation of ODN 3 for 10 minutes at room temperature, we noticed the total conversion of starting ODN 3 and intermediate photoproducts to tree photoproducts 3a, 3b and 3d (Fig. 2D).
The main photoproducts after irradiation of ODN 1–3 were isolated and separated by HPLC and their UV absorption (Fig. 3), fluorescence (Fig. 2S) and MALDI TOF MS spectra (Table 1S, Fig. 4S-7S) were recorded. Photoproducts 1a, 1b, 2a, 2b, 3a, 3b have the same absorption and emission spectra characteristic for previously observed fluorescent photoadduct of FSU with T,35 and were identified as two diastereomeric (a,b) intrastrand photocrosslinks of FSU with T (Fig. 3).
The UV absorption spectra of photoproducts 2d and 3d have a maximum at 260 nm and lack of the band with a maximum at 340 nm (characteristic for thiocarbonyl group) is observed (Fig. 1S). In MALDI TOF MS spectrum of the product 3d (Fig. 7S) the signal corresponding to a mass reduced by 16 compared to ODN 3 is present. This prompt us to claim that 3d and 2d are the products of photooxidation of FSU to 5-fluorouridine (Fig. 3). To identify photoproduct 2c, an additional experiment was performed by irradiating FSU in the presence of cytidine in a phosphate buffer, pH 7.0. The triple molar equivalent of cytidine was used to minimize the formation of photoadducts of FSU itself.46 Comparing the absorption spectra of the products obtained after irradiation FSU + C, a similarity of the UV spectrum of one of the products to the spectrum of photoproduct 2c formed after irradiation of ODN 2 was noticed (Fig. 8S). This product was isolated from the reaction mixture and based on the result obtained from its ESI MS spectrum (Fig. 9S) and MALDI TOF MS spectrum of photoproduct 2c (Fig. 5S) we suggest that in the case of ODN 2 the intrastrand photocrosslink with C is formed (Fig. 3).
Numerous studies have shown that the formation of pyrimidine-pyrimidone photoadducts involves 2 + 2 photocycloaddition of C = S to C5 = C6 of another pyrimidine leading to the formation of a thermally unstable thietane in the first step.33,34,47,48 We propose that this process in the case of reaction of FSU or ClSU with T is followed by thietane ring opening leaving the thiol on C6 of the pyrimidine part and ring closure with concomitant loss of HF to form tricyclic structure of diastereomeric photocrosslinks a,b (Fig. 10S). We were able to observe the thietane experimentally for duplexes labeled with ClSU.32 The photoadduct formation of 4SU with C observed in E. coli tRNA is well described.33,34,47,48 We propose analogous pathway for the photoaddition of FSU to C with elimination of H2S and formation of non-fluorescent 4–5 photoadduct c (Fig. 10S).
Since short single stranded oligonucleotides show very high flexibility, it is difficult to unequivocally determine from the obtained results which of the cytosines C1 or C6 and C5 or C10 for ODN1 and ODN2 respectively, participates in the photocrosslinking reaction with FSU. To assess the conformational flexibility of ODN 1 and ODN 2 explicit-solvent molecular dynamics simulations were performed. The simulations helped to investigate the possibility of interaction between FSU and cytosines present in the tested oligonucleotides. Both the systems were run for 1000 ns and distances between the thiocarbonyl group of the FSU and the cytosine C = C double bond were measured. The simulations showed that cytosines located at the ends of the tested oligonucleotides are able to come into closer contact with FSU compared to the cytosine in the middle of the sequence (Fig. 4) and most likely they react with FSU to form non-fluorescent adduct.
Since temperature is one of the factors affecting the conformational flexibility of a single-stranded DNA molecules, we irradiated ODN 1–3 under different conditions (at 5℃ and 35℃). HPLC analyses of these experiments are presented in Fig. 2. Different conversion of the starting ODN was observed (Fig. 2E) after the same irradiation time (150 s). For all oligonucleotides, an increase in conversion was observed with increasing irradiation temperature. The highest conversion value was obtained for ODN 1 at 35℃ (95%), while at low temperature, at 5℃ for ODN 2 (86%). At room temperature, the highest reactivity was also observed for ODN 1. In the case of ODN 3, a clear difference in reactivity was observed depending on the temperature (from 50% at 5℃ to 93% at 35℃). Taking into account the reactivity of FSU toward intrastrand crosslink formation with T, the highest selectivity was noticed in the case of irradiation of ODN 1 in 35℃. Also for others oligonucleotides an increase in temperature resulted in an increase in the photocrosslinking with T, while lowering the irradiation temperature favored the photocrosslinking reaction of FSU with C. In the case of ODN 3, competition between the photocrosslinking reaction of FSU with T and the photooxidation of FSU was observed. In this case, the increase in temperature also caused an increase in the photocrosslinking yield.
To conclude, the results showed that the intrastrand photocrosslinking reaction can occur in short DNA fragments between bases which are not in close proximity. Regardless of the fact that FSU shows much higher reactivity towards thymine than cytosine, the factor determining the direction of the intrastrand photocrosslinking reaction in short single-stranded DNA oligonucleotides is their sequence and the position of FSU in the chain. The obtained results are valuable for the future designing of FSU-labeled probes for fluorescence-based detection of specific DNA sequences. Since the formation of fluorescent photocrosslink is only possible with T, substitution of T by dU can be considered when designing a probe with FSU in order to avoid a false-positive detection signal in the case of intrastrand photocrosslinking.