Kairui
Yang
ab,
Min
Kou
a,
Zicheng
Zhao
ab,
Jinyang
Li
ac and
Xianglei
Kong
*ab
aState Key Laboratory of Elemento-Organic Chemistry and Frontiers Science Center for New Organic Matter, College of Chemistry, Nankai University, Tianjin, 300071, China. E-mail: kongxianglei@nankai.edu.cn
bTianjin Key Laboratory of Biosensing and Molecular Recognition, College of Chemistry, Nankai University, Tianjin 300071, China
cSchool of Physics and Electronic Information, Anhui Normal University, Wuhu 241002, China
First published on 12th February 2025
The stable generation and structural characterization of sodium cationized nucleic acid radicals at the molecular level have always been a difficult problem to solve. Herein, we produced the radical cation of [Urd + Na − H]˙+ through ultraviolet photodissociation (UVPD) of the precursor ion of [I − Urd + Na]+ in the gas phase and further studied its infrared multiphoton dissociation (IRMPD) spectrum in the region of 2750–3850 cm−1. The comparison between the IRMPD spectra of the precursor and radical cations shows their common features at both 3445 and 3705 cm−1 peaks, as well as the difference in the 3628 cm−1 peak that exists only in the case of the latter. By combining with theoretical calculations, it is indicated that the bidentate coordination structure M–B(O2,O2′)-1 and the tridentate coordination structure R–T(O2,O′,O5′)–(C5H–C1′)-1 are dominantly populated for the precursor and the radical cations, respectively. After the homo-cleavage of the C–I bond using a UV laser, a multi-step hydrogen transfer process started from the C1′ position, followed by a rotation of the intramolecular C–N bond, resulting in the formation of the most stable isomer, characterized by its radical position at C1′ and its tridentate coordination mode. This result indicates that the generation of free radicals of metal cationized nucleic acids by UVPD may result in the hydrogen transfer from the sugar ring, as well as the accompanied change of its coordination mode of the attached metal ions.
To study such transient species formed during the reaction process, various methods for generating nucleoside free radicals have been developed in the past few years.8–15 Compared to the liquid phase, the gas phase provides a unique surrounding in which molecular ions are not affected by solvent and counterion effects.16,17 Typically, mass spectrometry-based experimental methods, including collision-induced dissociation (CID), ion mobility, and UV–vis/IR photodissociation action spectroscopy, can be combined with the design of suitable precursor ions, such as the charge-tagging method, to provide powerful means for the characterization of special radicals.18–25 By comparing experimental results with theoretical calculations, their structures and kinetic information of radical ions in the gas phase can be deeply investigated at the molecular level.
On the other hand, the coordination between nucleic acids and the ubiquitous sodium ion is an unavoidable issue. It has been reported that sodium cations can affect the replication and cleavage of DNA and RNA.26–31 The effects of sodium cationization on nucleosides have been studied in the gas phase in the past few years. The method of IRMPD spectroscopy has been applied in the structural study of such complexes.20,32 For instance, Zhu et al. compared the obtained IRMPD and theoretical spectra of [Urd + Na]+ and [dUrd + Na]+ in detail. For [Urd + Na]+, the most stable isomer is characterized by its tridentate coordination mode, while the most populated ion in the ESI process has a bidentate coordination mode. The reason is that the former isomer is more stabilized by hydration in aqueous solution and the corresponding microsolvation environment and is kinetically trapped in the desolvation process due to the energy barrier.33
However, studies on sodium cationized nucleic acid radicals, as well as the effects of the formation of radicals on the coordination modes, have been rarely reported. One of the problems may be the difficulty in stably generating and systematically studying them at the molecular level. We have previously reported the experimental way to generate metalized nucleoside radical cations in the gas phase with the aid of the iodine-labelled precursor and the method of UVPD to cleave the C–I bond. Herein, by combining a Fourier transform ion cyclotron resonance (FT ICR) mass spectrometer with tunable infrared and ultraviolet lasers,19,34 the radical cations of [Urd + Na − H]˙+ were generated and studied by IRMPD spectroscopy. Further comparisons between the experimental and calculational spectra for both the radical cations and the precursor ions reveal not only their structures in detail, but also the transition of the coordination modes of the sodium cations accompanied by the formation of the radical cations.
However, UV irradiation shows quite different results, which also depend on the wavelength of the UV laser applied. With the UV laser at 218 nm, only the fragment ion with m/z 238.09 was observed. When a 245 nm UV laser was applied, both fragments at m/z 238.09 and 266.06 were observed. The ion at m/z 266.06 can be clearly identified as the product ion formed by the homolytic cleavage of the C–I bond, which is the radical cation with the form of [Urd + Na − H]˙+. Interestingly, while the 282 nm UV laser was used, only the radical ion at m/z 266.06 was observed. The fundamental aspects underlying this phenomenon of the wavelength-dependent ultraviolet photodissociation are important. But a clear understanding about it is also challenging, since the dissociation channels are relative to both thermodynamics and dynamic processes, and possible intersystem crossing may make the situation more complicated.36
MS3 tandem mass spectra of the radical ion [Urd + Na − H]˙+ generated by the 282 nm UVPD were also obtained after its selection (Fig. 1g). Both CID and IRMPD result in the fragment ion at m/z 238.09, which is formed by the loss of the unit of CO from the radical cation (Fig. 1h and i), while the 5 s UV irradiation at 280 nm results in no observed fragment ion (Fig. 1j). For the m/z 238 fragment ion observed in Fig. 1h and i, MS4 tandem mass spectrometry shows the only peak at m/z 198.06 (Fig. S1 in ESI†), which was formed by the loss of the unit of C2H2N.
To test if the product radical ions of [Urd + Na − H]˙+ have different structures that are relative to the wavelength applied in the previous UVPD step for its generation, MS3 tandem mass spectrometry experiments based on IRMPD and CID were also performed for the radical cations generated by UVPD with different wavelengths, including 250, 260, and 290 nm. The resultant mass spectra are all very similar to those shown in Fig. 1h and j, indicating their structural difference is indistinguishable, if there is any.
To determine the wavelength-dependent yield of the radical ions, the yield spectrum of [Urd + Na − H]˙+ as a function of applied UV wavelength was recorded, as well as the UVPD spectrum of the precursor ion. As shown in Fig. 2a/b, the absorption range of the parent ion [I − Urd + Na]+ is wide between 218 and 314 nm, with a maximum value at 278 nm. Meanwhile, the radical ions of [Urd + Na − H]˙+ can only be generated within the range of 234–314 nm, while the range of the ion m/z 238.08 is 218–278 nm. The changes in dissociation channels result in the generation of fragments at m/z 238 or 266 at low or high wavelengths, respectively.
![]() | ||
Fig. 2 (a) The UVPD spectrum of the precursor ion of [I − Urd + Na]+; (b) the yield spectrum of the radical ion of [Urd + Na − H]˙+. Both spectra were obtained in the range of 210–350 nm. |
To explore their structures, IRMPD spectroscopy experiments were performed for both ions. In such a study, the intensity of the spectra was calculated based on IIR = −ln(Ip/∑(Ifi + Ip)), where Ip and Ifi are the intensities of the parent ion and the i-th product ion, respectively, in the corresponding experiment. As shown in Fig. 3, the IRMPD spectrum of the precursor ion [I − Urd + Na]+ is characterized by sharp peaks at 3445 and 3705 cm−1, which is consistent with that of [Urd + Na]+ previously reported by Zhu et al.33 For the radical ion [Urd + Na − H]˙+ generated with the UVPD of [I − Urd + Na]+ at 282 nm, its spectrum shows some difference. The two peaks centered at 3438 and 3710 cm−1, which are very close to those observed in the case of the precursor ion, still dominate the spectrum, but they have noticeably widened. Besides, two weak peaks centered at 2972 and 3628 cm−1 can be clearly identified, the latter of which also has a peak width similar to that of the peak at 3710 cm−1. To investigate whether the radicals generated at different wavelengths of ultraviolet laser irradiation have different structures, infrared photodissociation experiments on the mass-selected radical ions generated by UVPD at 250 and 290 nm were conducted separately at ten wavenumbers that are selected according to the spectral characteristics observed in Fig. 3b. The results show that the infrared dissociation mass spectra of these free radical ions generated at different wavelengths have very similar characteristics, indicating that they do not have significant structural differences. In other words, the previously observed wavelength-dependent generation of radical isomers has not been observed here.32
![]() | ||
Fig. 3 A comparison between the experimental IRMPD spectra of the precursor ion of [I − Urd + Na]+ and the radical cation of [Urd + Na − H]˙+. |
![]() | ||
Fig. 4 Structures of the top isomers of the precursor ions. Their relative energies and Gibbs free energies are shown in the parentheses in the unit of kcal mol−1. |
The top two isomers of [I − Urd + Na]+ have structures very similar to those of [Urd + Na]+ previously reported by Zhu et al.33 The most stable isomer, M–T(O2,O′,O5′)-1, has a triply coordinated sodium ion and a syn-orientation of the nucleobase, which has an energy 0.3 kcal mol−1 lower than the second stable isomer, M–B(O2,O2′)-1, which is characterized by the double coordination and the anti-orientation of the nucleobase. Tautomers with tridentate or bidentate binding typically have much higher energies (Fig. S2†). The found best tautomer, M–T(O2,O′,O5′)–O4H-1, for example, is found to be 8.8 kcal mol−1 higher in energy (Fig. 4).
Fig. 5 compares the experimental IRMPD with the theoretically calculated IR spectra of the three isomers shown in Fig. 4. For the most stable isomer, M–T(O2,O′,O5′)-1, the predicted peak at 3573 cm−1, due to the O2′–H stretch mode, is absent in the experimental spectrum. Similar to that, the predicted peak at 3563 cm−1 of M–T(O2,O′,O5′)–O4H-1 is also absent. It is found that the spectrum of the second stable isomer M–B(O2,O2′)-1 is characterized by the two peaks at 3455 cm−1 and 3678 cm−1, corresponding to the stretching vibrations of N3–H/O2′–H and O5′–H, respectively, which are in good agreement with the experimental IRMPD spectrum. The energies and IR spectra of other structures including different tautomers show that their contributions to the observed ions are negligible (Fig. S3†). Thus, the M–B(O2,O2′)-1 is believed to be dominantly populated in the experiments reported here.
This result is consistent with the previous work on sodiated uridine,33 in which the calculated spectrum of the bidentate structure B(O2,O2′) matches the experimental spectrum much better than the tridentate structure T1(O2,O4′,O5′), which has an energy 1.5 kcal mol−1 lower than that of the former isomer. Their further calculations explained the reasons behind the result.33 The bidentate structure B1(O2,O2′) is more stabilized by hydration in aqueous solution than its tridentate counterpart and is kinetically trapped during the ESI process, resulting in it as the dominant conformer of [Urd + Na]+ in the gas phase. The explanation can also be applied to the experimental results reported here. As shown in Fig. S4,† the results show that the binding of two water molecules can further stabilize the isomer M–B(O2,O2′)-1, since M–B(O2,O2′)-1–2W is found to be 1.9 kcal mol−1 more stable than M–T(O2,O′,O5′)-1–2W.
Based on these considerations, the relevant isomers were constructed, optimized, and screened. The top isomers belonging to different structural types are shown in Fig. S5–9.† For simplicity, these structures are named according to their structural styles defined in the following way. First, these structures are divided into bidentate or tridentate cases according to their coordination modes and named with the letters of B or T, along with the coordinated atoms shown in the parentheses. Second, they are classified into two categories based on the presence or absence of the radical transfer. For those isomers with the transfer, the positions of the transferred hydrogen atom before and after the transfer are shown in the following parentheses. If there is no radical transfer, this part will be omitted. Third, the tautomerization, if there is one, is identified in the following parentheses. In the last step, the isomers are labelled with the corresponding numbers according to their energy order with the same structural types.
We can first focus on the structures without radical transfers. For such kinds of isomers, the top three are shown in Fig. 6. For ease of comparison, the relative energy of the first structure, R–T(O2,O′,O5′)-1, is set to 0 kcal mol−1. This structure can be thought of as the direct product ion after the homo-cleavage of the C–I bond of the most stable structure of the precursor ion (M–T(O2,O′,O5′)-1). And another isomer with a similar structure, R–T(O2,O′,O5′)-2, has an energy 0.5 kcal mol−1 higher. Similarly, the structure R–B(O2,O2′)-1 can be treated as the direct product ion from the precursor ion of M–B(O2,O2′)-1, which has a relative energy of 0.4 kcal mol−1.
Calculation reveals that most structures formed after radical transfers have lower energies than that of R–T(O2,O′,O5′)-1. Since the radical transfer is achieved through the transfer of H atoms here, the corresponding isomers are classified and labeled using the original site of the hydrogen atom that undergoes transfer, as described above (Scheme 1). Taking R–T(O2,O′,O5′)–(C5H–C1′)-1 as an example, the first half T (O2,O′,O5′) means its coordination style, and the second half means the transfer of the H atom from the sugar ring C1′ to the pyrimidine ring C5 positions.
Fig. 6 also shows the top six structures that experienced hydrogen transfers. It can be observed that among these six structures, the 2nd and 4th are characterized by bidentate coordination, while the rest are characterized by tridentate coordination. The energy of the most stable structure R–T(O2,O′,O5′)–(C5H–C1)-1 is 30.7 kcal mol−1 lower than that of R–T(O2,O′,O5′)-1. For the second stable structure, it also has an H transfer from the C1′ position, but with a bidentate coordination of B(O2,O2′). It has an energy 0.9 kcal mol−1 higher than that of the most stable structure. For structures of R–T(O2,O′,O5′)–(C5H–C4′)-1, R–T(O2,O′,O5′)–(C5H–C2′)-1, and R–T(O2,O′,O5′)–(C5H–C3′)-1, the difference exists only in the origin of the H atom transferred. And they have energies of 2.4, 3.9, and 5.2 kcal mol−1 higher than that of the top isomer. The structure of R–B(O2,O2′)–(C5H–C5′)-1, which is characterized by the bidentate coordination and the C5′ H transfer, has an energy 4.4 kcal mol−1 higher than that of R–T(O2,O′,O5′)–(C5H–C1)-1. The results mean that both coordination styles and origins of the H atoms transferred can greatly affect the total energies of the radicals.
Fig. 7 compares the experimental IRMPD spectrum with the theoretical infrared spectra of various structures of the radical cations shown in Fig. 6 (the corresponding spectra of other structures, including different tautomers, are listed in Fig. S10–16†). Among all the structures, the calculated IR spectrum of the most stable isomer, R–T(O2,O′,O5′)–(C5H–C1′)-1, matches best with the experimental result. The predicted peak at 3443 cm−1 due to the stretching vibration of the pyrimidine ring N3–H matches the experimental peak at 3438 cm−1. The peaks at 3695 cm−1 and 3703 cm−1 due to the stretching modes of O5′–H and O3′–H match the experimental peaks at 3696 cm−1 and 3712 cm−1, respectively. Importantly, the peak due to the O2′–H stretching vibration matches the experimental peak at 3628 cm−1. For other isomers involving hydrogen transfer processes, although their predicted spectra include peaks at ∼3440 and ∼3703 cm−1, their O2′–H stretching vibration modes are located at different places, which do not agree with the experimental result. The isomers without hydrogen transfer, however, are much higher in energy compared to the most stable isomer. Thus, it can be believed that the main contribution of the experimentally observed radical cations is coming from the isomer R–T(O2,O′,O5′)–(C5H–C1′)-1.
![]() | ||
Fig. 7 A comparison among the experimental IRMPD spectrum and the calculated IR spectra of the corresponding isomers of the radical ion of [Urd + Na − H]˙+. Their relative energies and Gibbs free energies are shown in parentheses in kcal mol−1. The structures of these isomers can be found in Fig. 6. All calculations were performed at the level of B3LYP/6-311+G(d,p). |
To understand the processes of hydrogen migration and coordination transition, the potential energy distribution of possible isomerization pathways from R–B(O2,O2′)-1 to R–T(O2,O′O5′)–(C5H–C1′)-1 is calculated. The results are shown in Fig. 8. In this experiment, the generation of the free radical was achieved through C–I bond homo-cleavage induced by ultraviolet laser irradiation, indicating that the isomer R–B(O2,O2′)-1 is a direct product in the first step. This product ion can be isomerized to form the isomer of R–B(O2,O2′)–(C5H–C6)-1 through a transition state of TS1 that has an energy barrier of 53.1 kcal mol−1. In the process, the H atom migrates along the edge of the pyrimidine ring, resulting in the structure of R–B(O2,O2′)–(C5H–C6)-1 with the radical position transferred to the position of C6. The subsequent step is characterized by the hydrogen transfer from the ribose ring (C1′), forming a much more stable isomer, R–B(O2,O2′)–(C5H–C1′)-1, through the transition state TS2 with a barrier of 32.7 kcal mol−1.
In the last step, through the rotation of the single C–N bond within the molecule, the structure R–B(O2,O2′)–(C5H–C1′)-1 isomerizes to the most stable isomer R–T(O2,O′,O5′)–(C5H–C1′)-1 and completes the transition from bidentate to tridentate coordination modes. Although a similar channel may exist in the case of the precursor ion, the energy barrier of the process blocks the rotation in the ESI source, resulting in the survivor as the isomer of M–T(O2,O′,O5′)-1, instead of the most stable isomer. To understand the energy barrier of this step, a relaxed potential energy scan connecting the two isomers was performed here with 5° dihedral angle increments. As shown in the upper right corner of Fig. 8, the calculated energy barrier for the internal rotation is 23.3 kcal mol−1. For the precursor ion of [I − Urd + Na]+, the corresponding energy barrier is much higher, mainly due to the existence of the iodine atom (Fig. S17†), which explains why the precursor ion maintained the structure with the bidentate coordination mode.
For the second point, our previous results show that the radical cations of cytidine generated by the UVPD at different wavelengths may have different structures, due to whether the isomerization energy barrier can be overpassed after the homo-cleavage of the C–I bond.32 In that case, the energy of the UV photon at 245 nm allows the isomerization to occur after the homolytic cleavage, resulting in the formation of the most isomer of R–C5 from the initial isomer of R–O2, while that of the UV photon at 280 nm cannot make the isomerization happen. Here, the isomer-selective generation via wavelength-regulated photodissociation has not been observed. That is, the structure of [Urd + Na − H]˙+ generated by UVPD does not depend on the UV wavelength in the region of 250–290 nm. This difference can be reasonably explained by comparing the energy barriers of two isomerization processes. The energy barrier in the isomerization process shown in Fig. 8 is 53.1 kcal mol−1, about 14% lower than that of [Cyt]˙+ (61.5 kcal mol−1). Further calculations indicate that even for the UVPD at 290 nm, the energy provided by the single photon is enough for the isomerization after the break of the C–I bond.
Footnote |
† Electronic supplementary information (ESI) available: Experimental details, additional calculation results and molecular coordinates. See DOI: https://doi.org/10.1039/d4dt03561e |
This journal is © The Royal Society of Chemistry 2025 |