Prebiotic synthesis of dihydrouridine by photoreduction of uridine in formamide

In this report, we show that a very common modification (especially in tRNA), dihydrouridine, was efficiently produced by photoreduction of the canonical pyrimidine ribonucleoside, uridine in formamide. Formamide not only acts as a solvent in this reaction, but also as the reductant. The other three components of the canonical alphabet (C, A, G) remained intact under the same conditions, suggesting that dihydrouridine might have coexisted with all four canonical RNA nucleosides (C, U, A, G) at the dawn of life.

were recorded with an Agilent Technologies 6130 Quadrupole LC-MS using positive and negative Electron Spray Ionisation.

Photoreaction procedures
Photoreaction of uridine in formamide in different concentrations.
Uridine (16 mM, 32 mM, 60 mM or 100 mM) in degassed formamide (1 mL) was transferred to a Suprasil quartz NMR tube and irradiated at 254 nm for 2-14 hours until the starting material was all consumed.The tube was periodically removed from the reactor to record 1 H NMR spectra.Yields were determined based on relative integrations of the signals referred to an internal standard (150 or 250 ul of 50 mM sodium succinate in H2O) in 1 H NMR spectra.Yields for the various products are listed in Table 1.

Photoreaction of uridine in water.
Uridine (16 mM) in degassed water (0.5 mL) was transferred to a Suprasil quartz NMR tube and irradiated at 254 nm for 2 hours until the starting material was all consumed.The tube was periodically removed from the reactor to record 1 H NMR spectra.The water was removed by evaporation and the residue was taken up in formamide (0.5 mL) again.The solvent was removed by evaporation under high vacuum at 60 ºC before monitoring the reaction mixture with 1 H NMR.About 46% of uridine was reformed from uridine hydrate according to the observations on 1 H NMR spectrum (Figure S56).

Photoreaction of uridine in water in the presence of sodium formate.
Uridine (2 mg, 0.008 mmol) and sodium formate (7 mg, 0.10 mmol) were mixed in degassed water (0.5 mL), then transferred to a Suprasil quartz NMR tube and irradiated at 254 nm for 2 hours until the starting material was all consumed.The tube was periodically removed from the reactor to record 1 H NMR spectra (Figure S57).UAA trimer (2 mg) in degassed formamide (0.5 mL) was transferred to a Suprasil quartz NMR tube and irradiated at 254 nm for 2 hours.No reaction was observed from 1 H NMR spectra (Figure S59).

UV irradiation of trimer UAC at 254 nm.
UAC trimer (2 mg) in degassed formamide (0.5 mL) was transferred to a Suprasil quartz NMR tube and irradiated at 254 nm for 2 hours.No reaction was observed from 1 H NMR spectra (Figure S60).
6 ml (98.26 mmol) of d1-methyl formate (99 atom % D, Sigma -Aldrich) was mixed with 100 ml of 7 M NH3 in methanol in a sealed tube.The mixture was stirred at room temperature for 30 mins.Solvent was removed by evaporation under high vacuum to give 4.14 g (91.6% yield) of 1-d1-formamide which was used without further purification.

Preparation of N-d2-formamide.
5 ml (125.89mmol) of formamide was mixed with 10 ml of D2O for 30 mins before removing the solvents (H2O, HDO, D2O) at lower boiling point under high vacuum.The procedure was repeated for three times to obtain 5 g of N-d2-formamide (90 atom% D, 85 % yield) which was used without further purification.

Photoreaction of uridine in 1-d1-formamide or N-d2-formamide.
15 mg (0.06 mmol) of uridine in 1ml of 1-d1-formamide or N-d2-formamide was irradiated at 254 nm until starting material was all consumed.The solvent was removed under high vacuum.The residue was dissolved in D2O to record the crude 1 H NMR spectrum, then subjected to preparative reverse phase HPLC, using acetonitrile in water as the eluent, to afford pure partially deuterated dihydrouridine and 5, 6-dihydro-6(S, R)-carboxamido-uridine. (Fig. S21-27)

Photoreaction of other ribonucleosides (C, A, G, I) in formamide.
Cytidine, adenosine, guanosine or inosine (16 mM) in degassed formamide (0.5 mL) was transferred to a Suprasil quartz NMR tube and irradiated at 254 nm for 2 hours, a time period during which uridine is all converted to dihydrouridine.The solvent was then removed under high vacuum.The residue was dissolved in D2O to record the crude 1 H NMR spectra.No reaction from these ribonucleosides was observed in 1 H NMR spectra.

Photoreaction of mixtures of cytidine and uridine in formamide.
The mixture of cytidine and uridine (16 mM each) in degassed formamide (0.5 mL) was transferred to a Suprasil quartz NMR tube and irradiated at 254 nm for 6 hours, during which uridine is all converted to dihydrouridine.The solvent was then removed under high vacuum, and the residue was dissolved in D2O to record the crude 1 H NMR spectra.No reaction from cytidine was observed in 1 H NMR spectrum.

Photoreaction of mixtures of all four canonical ribonucleosides (U, C, A, G) in formamide.
A mixture comprising 1 mg each of uridine, cytidine, adenosine and guanosine in formamide (0.5 mL) was transferred to a Suprasil quartz NMR tube and irradiated at 254 nm for 5 hours in total.The solvent was then removed under high vacuum.The residue was dissolved in D2O to record the crude 1 H NMR spectrum.78% of uridine was converted to dihydrouridine while all other three ribonucleosides remained intact.

Photoreaction of uridine-5'-phosphate in formamide.
6 mg (0.016 mmol) of uridine-5'-phosphate in 1ml of formamide was transferred to a Suprasil quartz NMR tube and irradiated at 254 nm for 2 hours.The solvent was then removed under high vacuum.The residue was dissolved in D2O to record the crude 1 H NMR and 31 P NMR spectra.60 uL of 50 mM sodium succinate was added to the mixture to determine the yield of dihydrouridine-5'-phosphate by relative integration (72% yield).

Photoreaction of uridine-3'-phosphate in formamide.
6 mg (0.016 mmol) of uridine-3'-phosphate in 1ml of formamide was transferred to a Suprasil quartz NMR tube and irradiated at 254 nm for 2 hours.The solvent was then removed under high vacuum.The residue was dissolved in D2O to record the crude 1 H NMR and 31 P NMR spectra.60 uL of 50 mM sodium succinate was added to the mixture to determine the yield of dihydrouridine-3'-phosphate (64% yield).

Photoreaction of uridine-2',3'-phosphate in formamide.
2 mg (0.006 mmol) of uridine-2',3'-phosphate in 0.5ml of formamide was transferred to a Suprasil quartz NMR tube and irradiated at 254 nm for 2 hours.The solvent was then removed under high vacuum.The residue was dissolved in D2O to record the crude 1 H NMR and 31 P NMR spectra.60 uL of 50 mM sodium succinate was added to the mixture to determine the yield of dihydrouridine-2',3'-phosphate (69% yield).

Computational methods
To locate the most stable geometries of uridine in a polar solvent, the conformational space was studied using the Conformer-Rotamer Ensemble Sampling Tool 1 (CREST) at the semi-empirical GFN2-xTB 2 level of theory.The conformational space was explored using the iMTD-GC scheme together with the analytical linearized Poisson-Boltzmann (ALPB) implicit solvent model, assuming a water solvent.The optimized ground-state structures and the corresponding harmonic vibrational frequencies for isolated uridine and the uridine and formamide complex were located using the Kohn-Sham density functional theory (KS-DFT) and the range-separated hybrid ωB97X-D functional including dispersion correction. 3These calculations were performed using the def2-TZVP 4 basis set.To estimate solvent effects exerted by formamide on our model systems, the polarizable continuum model 5,6 (PCM) with the integral equation formalism variant (IEFPCM) was applied.The level of theory is further described as the PCM/ωB97X-D/def2-TZVP abbreviation.The KS-DFT and PCM computations were performed by means of the Gaussian 7 16 package.
Vertical excitation energies of the uridine and formamide complex were obtained, using the spin-component scaled variant 8 of the algebraic diagrammatic construction to the second order method 9,10 [SCS-ADC(2)] with the aug-cc-pVDZ 11 basis set (SCS-ADC(2)/aug-cc-pVDZ).For these calculations we used the S0 geometries optimized at the PCM/ωB97X-D/def2-TZVP level of theory.The molecular orbital character of lowest-lying excited states was characterized through the analysis of natural transition orbitals (NTOs).The charge transferred between the explicit formamide molecule and the nucleoside was evaluated using the one-electron transition density matrix (1TDM) and the Löwdin style analysis.The 1TDM and NTOs analyses were performed employing the TheoDore 12 3 package.The crucial points on excitedstate potential energy surfaces of the studied system were found at the SCS-ADC(2)/aug-cc-pVDZ level of theory.The minimum-energy crossing point (MECP) between the first excited and ground state was optimized using the sequential penalty constrained optimization method implemented in the CIOpt 13 package.The MECP was optimized utilizing the energies and analytical gradients obtained using the MP2/ADC(2)/aug-cc-pVDZ.The CIOpt and Turbomole 14 7.3 programs were interfaced to perform the MECP optimization.The potential energy profiles (PEPs) of the photochemical reaction were constructed using the image dependent pair potential (IDPP) interpolation 15 between the optimized S0 structure and the S1 minimum-energy geometry.The PEPs, between the S1 minimum and the S1 transition-state structure, were built with geometries generated utilizing the reaction path (RP) optimization method 16 implemented in the Turbomole 14 7.3 program.To obtain the PEPs from the S1 transition-state structure towards the S1/S0 minimum-energy, the intrinsic reaction coordinate method 17 (IRC) was employed.The RP and IRC methods allowed us to obtain smooth potential energy profiles between the crucial excited-state points.All optimized structures and the energies of ground and excited states were located using the SCS-MP2/aug-cc-pVDZ and SCS-ADC(2)/aug-cc-pVDZ methods, respectively.All excited-state calculations were performed using the Turbomole 14 7.3 package.The IDPP method was used in the ORCA 18 5.0.3 program.
To benchmark the single-reference SCS-ADC(2) method which was used for the optimization of excited-state minimum-energy structures and the S1/S0 minimumenergy crossing point, the critical excited-state geometries were reoptimized with the extended multi-state complete active space second-order perturbation theory 19,20 (XMS-CASPT2), based on the state-averaged complete active space self-consistent field (SA-CASSCF) method with the cc-pVDZ basis set.To find the S1 minimumenergy structure at the XMS-CASPT2 level, the complete active space (CAS) was built of molecular orbitals (MOs) having natural orbital occupations in the range of 0.02-1.98, 21and contained one σ, three π, and one nO occupied MOs alongside with three π * and one σ * virtual MOs (overall 10 electrons were correlated in 9 orbitals, as shown in Fig. S48, left panel).The SA-CASSCF wave function was averaged over three lowestlying electronic states.The S1/S0 minimum-energy conical intersection was optimized assuming the CAS constructed of the three π, nO, and four π * MOs (10 electrons in 9 orbitals, shown in Fig. S48, right panel).In both cases, the SA-CASSCF wave function was averaged over three lowest-lying electronic states.The vertical shift, i.e., the empirical correction applied to the zeroth-order Hamiltonian, was equal to 0.4 Hartree, and the XMS-CASPT2 calculations were conducted using the BAGEL 22 1.2.0 package.
The ground-state structures of substrates, transition states and products of studied chemical reactions were obtained at the PCM/ωB97X-D/def2-TZVP level of theory, assuming the IEFPCM model of bulk formamide.To evaluate the kinetic isotope effect for a given chemical process, we recalculated the thermal corrections to Gibbs free energy for the corresponding substrate and transition-state structures, replacing a hydrogen atom with deuterium at a given site, to estimate a new energy barrier.Then, the following formula kH/kD =  !(∆$ !‡ !∆$ $ ‡ )/'( was used, in which the ∆ ) ‡  ∆ , ‡ corresponding to the relative difference Gibbs free energies between the substrate and transition state structures for formamide and deuterated formamide molecules, respectively.In addition, the R and T variables are the gas constant and temperature (298 K), respectively.The ratio kH/kD expresses the change of the reaction rate when the studied system is isotopically substituted with deuterium.

Equilibrium ground-state structure of uridine and its complex with formamide
To find the equilibrium ground-state geometry of β-uridine in formamide solution, we conducted the semi-empirical GFN2-xTB 2 calculations to explore the conformational space of the canonical pyrimidine nucleoside.Subsequently, we selected the most stable C2'-endo and C3'-endo structures of β-uridine, and then we reoptimized these structures (Fig. S41) at the PCM/ωB97X-D/def2-TZVP level of theory, assuming the continuum model of bulk formamide, and the corresponding harmonic vibrational frequencies were also computed.Our Gibbs free energy computations show that the C2'-endo and C3'-endo conformers of uridine are almost isoenergetic, and the corresponding energy difference amounts to -0.03 kcal/mol.This result indicates that both conformers of b-uridine can coexist in an equilibrium in formamide solution.The key structural difference between the considered conformers, apart from the conformational arrangement of the sugar part, is an intramolecular hydrogen bond (1.93 Å) between the hydroxyl group at the C2' atom (sugar fragment) and the carboxyl oxygen atom at the C2 atom (nucleobase part) that occurs only in the C2'-endo structure (see Fig. S41).Since the C2=O moiety of the nucleobase can create the intramolecular hydrogen bond with the C2'-endo sugar, we anticipate that this carbonyl group might only periodically interact with neighbouring formamide molecules, when it is not involved in the intramolecular hydrogen bond.Thus, we added an explicit quantum-mechanical formamide molecule at the C4=O carbonyl group of the chromophore and optimized the geometry of this complex (Fig. S42) using the PCM/ωB97X-D/def2-TZVP method.In the complex (Fig. S42), there is a hydrogen bond of 1.93 Å between the uridine C4=O carbonyl oxygen and the amino group of formamide.Importantly, an explicit formamide solvent molecule can also interact in an unconstrained manner with the C3'-endo conformer at the C4=O site, thus we assume that both conformers should exhibit similar photochemical reactivity in formamide solution.In this work, our aim was to investigate the active participation of a single neighbouring formamide molecule in the photochemistry of β-uridine in order to elucidate the photoreduction mechanism of uridine to dihydrouridine.

Photophysical properties of uridine and formamide complex
For the optimized S0 structure of the uridine and formamide complex, we calculated vertical excitation energies and corresponding oscillator strengths for the three lowest-energy excited states at the SCS-ADC(2)/aug-cc-pVDZ level of theory, and the results are presented in Tab.S1.
According to these calculations, the first excited state at 5.16 eV (240.3 nm) is an optically bright state with a substantial oscillator strength of 0.299 and can be assigned as a ππ* transition (see Fig. S43).The second excited state at 5.236 eV (236.8 nm) can be considered as a mostly dark state with insignificant contribution to the overall absorption given the oscillator strength of 0.015.In the case of the S2 state, the electron is mainly promoted (see Fig. S43) from the lone electron pair orbital (nO), localized at the C4=O site, to the π* molecular orbital located on the aromatic ring of uridine.Both of these excited states lie energetically close to each other and can be easily populated in the Franck-Condon region of the investigated system.The third excited state is located at higher energy (5.656 eV and 219.2 nm) and is characterized by the nOπ* excitation (Fig. S43) occurring within the formamide molecule.
In our work, we irradiated samples using 254 nm wavelength (4.88 eV) to perform the photochemical reduction of uridine to dihydrouridine.In the previous paragraph, we pointed out that the photoexcitation of formamide, assuming 254 nm light, is very unlikely owing to significant energy separation of the S3 state from the S1 and S2 states.Thus, the presented photophysical properties of the uridine and formamide complex indicate that irradiation at 254 nm populates excited states associated with the pyrimidine ring.Our next step was to investigate the photochemistry of the uridine-formamide complex based on explorations of excitedstate potential energy surfaces beyond the Frank-Condon region.

Photochemistry of uridine and formamide complex
Exploring the S1 excited-state potential energy surface of the uridine and formamide complex (Fig. S44) at the SCS-ADC(2)/aug-cc-pVDZ level of theory, we found that our system, after photoexcitation, can easily reach the S1 excited state having the nOπ* molecular orbital character.As demonstrated by previous experimental and theoretical studies, high-energy vibrational motions of the UV-excited molecule can force the elongation of the C4=O carbonyl bond, [23][24][25][26] which lowers the energy of the nOπ* excited state.Since the S1(nOπ*) excited-state region could be readily populated, we optimized a local S1 minimum that is shown in Fig. S44.This S1(nOπ*) minimumenergy structure is characterized by an excited-state intermolecular O⋯O interaction involving lone electron pair orbitals (Fig. S44) originating from the uridine and the formamide carbonyl oxygen atom, and the corresponding C4-O⋯O=C distance amounts to 2.22 Å.This intriguing excited-state oxygen-oxygen interaction replaced the ground-state hydrogen bond between uridine and formamide, and can significantly stabilize the UV-excited complex by lowering the energy of the system by 1.38 eV when compared to the vertical excitation energy of the optically bright state.Furthermore, the wavefunction analysis of the S1 minimum having the nOπ* transition (Fig. S44) indicated a charge-transfer (CT) process of 0.20 e -from the nO orbital of formamide to the π* molecular orbital of uridine.Thus, the S1 minimum-energy structure is an excited-state charge-transfer complex, in which the formamide molecule acts as an electron donor when both molecules are in an appropriate arrangement.It is worth adding that this kind of excited-state intermolecular interaction was also found and described for other chromophores in the presence of water molecules, including 7H-adenine, 27 methylated cytosine 23 and guanosine 24 .
We next performed SCS-ADC(2) calculations of excited-state potential energy surfaces which show the plausible photochemical mechanism involving the CT complex as the key reactive species (see Fig. S45).After photoexcitation of the uridine and formamide complex to the optically bright excited state (orange line, at 5.16 eV), the system can easily populate the 1 nOπ* excited state (green line) and then reach the charge-transfer S1 minimum-energy structure at 3.78 eV.Having the excess energy originating from the electronic excitation, our model system can subsequently pass through the S1 transition-state structure (at 4.11 eV), which involves an excited-state hydrogen atom transfer from the formamide carbon atom to the uridine carbonyl oxygen atom resulting in the formation of the biradical system.This photochemical reaction leading to the production of hydrouridyl and formamide radicals also leads to the S1/S0 state crossing (at 3.49 eV) that can serve as a radiationless deactivation channel to the electronic ground state.We have demonstrated that the formation of an excited-state intermolecular CT complex in the S1 minimum can entail the hydrogen atom transfer process leading to the photoreduction of uridine by solvent formamide molecule.It is worth adding that the critical point for the whole photoreduction process is the S1 transition-state geometry (at 4.11 eV) which lies about 1 eV below the vertical excitation energy of the 1 ππ* excited state (in the Franck-Condon region).
In addition to the easily accessible photoreduction process described in the previous paragraph, we found another plausible photochemical pathway in a different region of the S1(nOπ * ) hypersurface (see Fig. S46).After photoexcitation of the uridine and formamide complex, the system can readily reach the 1 nOπ * excited-state potential energy surface as shown in Fig. S45.Subsequently, the model system can follow a distinct virtually barrierless photochemical pathway (see Fig. S46, green line) to the S1(nOπ * )/S0 intersection seam at 4.55 eV, corresponding to the S1(nOπ * ) transition-state excited-state geometry on the S1/S0 crossing seam.In the optimized S1(nOπ * ) structure (see inset in Fig. S46) occurs an excited-state intermolecular O⋯O interaction engaging lone electron pair orbitals from the uridine and the formamide carbonyl oxygen atom, and the corresponding C4-O⋯O=C distance is of 2.02 Å.The excited-state intermolecular interaction facilitates a charge-transfer process of 0.41 e -from the nO orbital of formamide to the π * molecular orbital of uridine in the S1 state.Together with the excited-state charge transfer process, there is also a significant elongation of the N-H bond (by 0.23 Å) in the amino group towards the C5 atom of uridine.Since an electron and proton are being transferred from formamide to uridine during the single step of the photochemical mechanism, we classify the process as a concerted protoncoupled electron transfer mechanism (PCET).Consequently, in the vicinity of the S1/S0 crossing seam, the concerted PCET enables the hydrogen atom transfer from the amino group of formamide to the C5 atom of UV-excited uridine, yielding the nitrogencentered formamide and uridine radicals.The former radical is energetically unstable (see Fig. S54) and should tautomerize to the carbon-centered formamide radical.It is worth emphasizing that the S1 transition-state structure can serve as a pathway for radiationless deactivation to the electronic ground state, as it is located at the S1/S0 intersection seam.
Both discussed photochemical processes demonstrate that this first crucial step of the photoreduction mechanism might be easily initiated photochemically and can occur via a stepwise or concerted mechanism.The latter involves the hydrogen atom transfer from the amino group to the C5 atom and is marked by a higher energetically photochemical pathway compared to the two-stepwise hydrogen atom transfer from the C-H site to the carbonyl oxygen atom of uridine (Fig. S45 and S46).Nevertheless, both excited-state processes are expected to play an important role in the photochemistry of uridine dissolved in a formamide solution.
Furthermore, our quantum-chemical calculations show that photoinduced reduction chemistry can be exclusively initiated by the electronic excitation of uridine.There is no need to excite solvent formamide molecules, for releasing solvated electrons, to conduct the photoreduction process.
Based on these static quantum-chemical calculations, we have found another example, in which ground-state solvent molecules can actively participate in the photochemistry of the chromophore.

Excited-state key structures at XMS-CASPT2 and SCS-ADC(2) methods
In the previous section, we discussed the plausible radiationless deactivation pathway as a photoreduction mechanism of UV-excited uridine utilizing formamide at the SCS-ADC(2) level.Recently, it was shown that the SCS-ADC(2) method 8 can offer accurate results for the description of the photochemistry of carbonyl-containing chromophores. 25In particular, the method can physically represent the elongation of the carbonyl bond in heteroaromatic chromophores, while the original ADC(2) method 9,10 offers much less accurate excited-state potential energy surfaces and energies for this group of chromophores. 25However, the single-reference SCS-ADC(2) method 8 still has some limitations, which should always be considered when a new photochemical mechanism is postulated at this level of theory.The reference electronic ground state is described by the single-reference SCS-MP2 method, 28 which cannot correctly describe multireference systems, and consequently, the S1/S0 surface crossing might be nonphysical in some cases.Therefore, we reoptimized the S1 and S1/S0 minimum-energy structures (shown in Fig. S45), obtained using the SCS-ADC(2)/augcc-pVDZ method, employing the multireference XMS-CASPT2/SA-3-CASSCF(10,9)/cc-pVDZ level of theory.The active spaces used in the XMS-CASPT2 calculations are presented in Fig. S48.The superimposed XMS-CASPT2 and SCS-ADC(2) minimum-energy geometries are shown in Fig. S47 as silver and orange structures, respectively.To evaluate the similarity of the corresponding structures, we computed root-mean-square deviations (RMSDs), which amount to 0.113 and 0.223 Å for the S1 and S1/S0 minimum-energy structures, respectively.It demonstrates that the SCS-ADC(2) and XMS-CASPT2 geometries are virtually identical.Thus, the SCS-ADC(2) method can be used for the studied system.
The key structural difference between the SCS-ADC(2) and XMS-CASPT2 S1(nOπ*) minimum-energy geometries is the intermolecular C4-O⋯O=C distance equal to 2.22 and 2.50 Å, respectively.It is worth noting though that the excited-state 1 nOπ* potential energy surface for increasing the O⋯O distance is very flat at the multireference XMS-CASPT2 level.More specifically, comparing the S1(nOπ*) minimum (O⋯O distance is 2.50 Å) with the S1(nOπ*) structure having the O⋯O distance equal to 2.28 Å, the corresponding energy difference only amounts to -0.74 kcal/mol at the XMS-CASPT2 level.Furthermore, our excited-state geometry optimizations of the model system were conducted in the gas phase without any solvent models.Thus, we cannot unequivocally assess the excited-state equilibrium intermolecular O⋯O distance that would be relevant in the presence of bulk formamide solvent.To estimate the charge-transfer character of the XMS-CASPT2 S1 minimum, having the O⋯O distance equal to 2.50 Å, we performed the wavefunction analysis at the SCS-ADC(2) level of theory, indicating the intermolecular electron transfer is 0.10 e -.It is worth adding that the excited-state CT complex should be energetically stabilized in a polar formamide solution 29 because the dipole moment of the S1 CT state in implicit COSMO solvent model is larger by approximately 3 D compared to the gasphase structure.Therefore, we conclude that since the XMS-CASPT2 approach has predicted the existing charge-transfer intermolecular O⋯O interaction, the SCS-ADC(2) method offers a reliable description of this S1 minimum.
The SCS-ADC(2) S1/S0 minimum-energy crossing point (MECP) was predicted and confirmed by the XMS-CASPT2 calculations.The main structural difference is only associated with the position of the formamide radical with regard to the hydrouridine radical.The formamide radical is noticeably shifted towards the hydrouridine radical at the XMS-CASPT2 level compared to the corresponding SCS-ADC(2) structure.Considering the energetic availability of the S1/S0 MECPs, we found that the ÄE between S1 minimum and S1/S0 MECP at the XMS-CASPT2 and SCS-ADC(2) levels amounts to 0.84 and 0.29 eV, respectively.It indicates that both methods predicted a peaked S1/S0 conical intersection that might be readily energetically accessible.
In order to assess the optimized S1(nOπ * ) transition-state excited-state geometry on the S1/S0 crossing seam, at the SCS-ADC(2)/aug-cc-pVDZ level of theory (presented in Fig. S46), we performed a single-point calculation using the XMS-CASPT2/SA-2-CASSCF(10,9)/cc-pVDZ method.Our analysis revealed that the energy difference between the S1 and S0 states amounts to 0.57 eV at the XMS-CASTP2 level.Thus, the multireference calculations confirmed that the S1 structure is situated close to the S1/S0 crossing seam and might be responsible for the radiationless deactivation mechanism.
In conclusion, the performed benchmark of the SCS-ADC(2) method with the multireference XMS-CASPT2 approach has shown that the SCS-ADC(2) results are valid and the single-reference method can be applied for the investigated system in this paper.

Excited-state complex of cytidine and formamide
Uridine and cytidine exhibit similar structural motifs in the chromophore part of their nucleosides.In particular, both contain the carbonyl oxygen atom attached to the aromatic structure, which possesses a photochemically reactive lone electron pair that could be involved in excited-state CT complexes (see above).Furthermore, the cytidine molecule has another lone electron pair associated with the N3 nitrogen atom that is a part of the pyrimidine ring.Since both molecules are characterized by similar structural features, we decided to also perform UV irradiation experiments with cytidine in a formamide solution.Under such conditions, we have not observed any photoreduction of cytidine to dihydrocytidine experimentally, as demonstrated for uridine.
To explain the striking difference in photochemical reactivity of both nucleosides, we performed quantum-chemical calculations for a complex of cytidine with a single formamide molecule at the SCS-ADC(2)/aug-cc-pVDZ level of theory, and the results are shown in Fig. S49.Exploring the S1 excited-state potential energy surface of the complex, we couldn't find a similar S1 minimum that was found in the uridine and formamide complex presented in Fig. S44.Instead, we optimized another the S1(nπ*) minimum-energy structure (Fig. S49) in which there is an excited-state intermolecular interaction between lone electron pair orbitals of the cytidine nitrogen atom and the formamide carbonyl oxygen atom.Although a similar excited-state complex between these molecules is formed in the S1(nπ*) state, the corresponding molecular n and π* orbitals are mainly localized on the formamide molecule.Consequently, our wavefunction analysis of the optimized S1 structure revealed no intermolecular electron transfer between the chromophore and formamide.The lack of intermolecular charge-transfer mechanism necessary for the photoreduction process explains why cytidine cannot undergo a photochemical reduction in formamide solution.

Generation of dihydrouridine in the electronic ground state
Our photochemical computational studies concerning the uridine and formamide complex have shown that the system can undergo an excited-state intermolecular hydrogen atom transfer leading to the formation of the hydrouridyl and formamide radicals in the vicinity of the S1/S0 surface crossing (Fig. S45).UVirradiation experiments of uridine in a formamide solution resulted in very efficient generation of dihydrouridine.Therefore, our next step was to elucidate the molecular mechanism of subsequent photoreduction of the hydrouridyl radical.
The hydrouridyl radical can undergo a tautomerization process in which a proton can be transferred through solvent formamide molecules from the C4-OH site to the C6 atom of the nucleoside (Fig. S50) yielding a tautomer more stable by 9.3 kcal/mol when compared to the enol form resulting from the S1/S0 state crossing.According to our KS-DFT ground-state calculations for this proton transfer process, the corresponding transition state is characterized by the energy barrier of 9.9 kcal/mol.In turn, the protonation at the C6 atom (Fig. S51, at -2.5 kcal/mol) leads to the more stable hydrouridyl radical tautomer and the enol tautomer of formamide is formed.It is worth adding that solvent molecules act as a proton relay during the tautomerization through a formamide molecule.Consequently, the solvent changes its tautomeric form from keto to enol or vice versa in the case of formamide.Therefore, to characterize the proton transfer process, we added two explicit formamide molecules to the hydrouridyl radical (Fig. S51).To avoid the unneeded generation of the enol form of formamide during the tautomerization reaction, we included one formamide molecule in the enol form and the other in the keto form.The formamide-assisted proton transfer entails two consecutive tautomerizations of the involved formamide molecules.During this process, one formamide molecule changes its form from enol to keto and the other from keto to enol.This approach allowed for maintaining the tautomeric balance of the solvent molecules that is unchanged during the hydrouridine radical tautomerization process.It is worth adding that single tautomerization of formamide from its keto to enol form is associated with ΔE of 16.39 kcal/mol.
Our theoretical explorations showed that the proton transfer process, allowing for tautomerization of the hydrouridine radical, is the lowest-energy mechanism that could occur shortly after the deactivation of the model system to the ground state.Subsequently, the most stable hydrouridyl radical can abstract a hydrogen atom from a nearby neutral formamide molecule, and the corresponding hydrogen atom abstraction transition state was found at 13.0 kcal/mol (Fig. S52).Passing through this modest barrier results in the formation of neutral dihydrouridine and formamide radical (Fig. S52, 1.4 kcal/mol).It is worth noting here, that this pathway leading to the formation of a single dihydrouridine molecule involves the generation of two formamide radicals, one generated photochemically and one in the subsequent dark photoreaction.These formamide radicals may next undergo radical recombination to yield a single oxamide molecule which is also observed among the products of our irradiation UV experiments.
To further elaborate on the uridine photoreduction process and analyze the mechanism, we conducted additional UV-irradiations experiments in deuterated formamide, such as N-d2-formamide or 1-d-formamide.The photochemical reaction of uridine in N-d2-formamide yielded dihydrouridine and the formamide C6-adduct of uridine.Deuterium was found in the C5 and C6 positions of dihydrouridine, whereas only the C5 atom was deuterated in the formamide C6-adduct of uridine.In the N-d2formamide solution, a proton transfer mechanism, resulting in the formation of the most stable tautomer of the hydrouridine radical (Fig. S51) should be noticeably slower since the amino group of formamide which is involved in the proton relay process leading to the tautomerization possesses deuterium atoms.Based on our calculations, hydrouridine tautomerization is not the rate-limiting step, albeit we expected this H to D substitution on the amino group to result in a noticeable primary kinetic isotope effect on this particular step of the reaction.Slower tautomerization of the hydrouridine radical, could enable more efficient radical recombination process between the formamide radical and the C6 position of hydrouridyl radical, leading to the formation of formamide-uridine adduct.To estimate the extent of this effect in the case of N-d2formamide, we recomputed the Gibbs free energy values for the hydrouridine radical, and the corresponding transition state associated with the enol to C6 tautomerization (Fig. S51).The corresponding transition state with the N-d2-formamide molecule has higher energy by approximately 1 kcal/mol compared to the same transition state structure involving non-deuterated formamide molecules.Consequently, our estimated kinetic isotope effect, that is the ratio of kH/kD is equal to 5.9.It indicates that the proton transfer process in an N-d2-formamide solution should be 5.9 times slower versus a formamide solution.It is worth adding that in the entire tautomerization process of hydrouridine radical, occurring from the C4-OH to C6 site, requires a few proton transfers through N-d2-formamide molecules.Consequently, since the proton transfer from the C4-OH to C6 site is significantly slower in an N-d2-formamide solution, the photochemically generated formamide radical could have enough time to be attached to the C6 atom of hydrouridyl radical and form the formamide C6-adduct (Fig. S53).This explains why the C6-adduct is observed when the formamide solution is replaced by N-d2-formamide.Furthermore, the proposed proton transfer mechanism also elucidates the incorporation of deuterium in the C6 atom of uridine.According to Fig. S52, the final step of dihydrouridine formation occurs through hydrogen atom transfer from the formamide molecule to the C5 atom of uridine.However, this process would not lead to the deuteration in the C5 position.The deuterated C5 site can be explained through a direct photochemical hydrogen atom transfer from the Nd2 site of N-d2formamide to the C5 atom of uridine (Fig. S46).In other words, upon UV excitation of uridine, the Nd2 site can act as an active donating group, allowing for deuteration at the C5 position, which is in line with our experimental results.
The photochemical reduction of uridine was also performed in 1-d-formamide solution, which led to the formation of dihydrouridine having only a partially deuterated C5 position.In the proposed mechanism of dihydrouridine generation, only the final step occurring in the dark (Fig. S52) might be altered by 1-d-formamide since the final intermolecular hydrogen atom transfer takes place preferentially from the formamide C1 site to the uridine C5 atom, yielding the energetically most stable formamide radical at the carbon atom (see Fig. S54).Therefore, assuming the 1-d-formamide molecule, we recomputed the ground and transition states to evaluate the kinetic isotope effect (Fig. S52), and the ratio of kH/kD equal to 4.97 for this particular step.However, it is worth noting that this final intermolecular hydrogen atom transfer is not the rate determining step and the effect of deuteration on the overall reaction should be negligible.Consequently, the formation of the uridine C5 formamide adduct is highly unlikely because any nearby 1-d-formamide molecule can still efficiently donate its hydrogen atom with reaction rate that should exceed the diffusion rate of the formamide radical formed in previous steps.According to this mechanism, generated dihydrouridine should be selectively deuterated in the C5 position which very well matches the experimental results.
In summary, our mechanistic rationale proposed based on quantum chemical calculations is entirely consistent with the additional irradiation experiments performed in N-d2-formamide and 1-d-formamide solvents since and it explains the selective deuteration of the photoproduct as well as formation of the formamide adduct and oxamide.
Scheme S1 Photochemical mechanism of uridine reduction in 1-d-formamide (Panel A) or N-d2-formamide (Panel B and C).i) Proton transfer following electron transfer; ii) solvent-assisted tautomerization; iii) hydrogen atom abstraction from formamide;

Discussion on deuteration data inferred from theoretical calculation.
The photoreductions in deuterated formamide are initiated by the formation of an excited-state charge transfer (CT) complex 18.Electron transfer allows immediate proton transfer from C-D (panel A) or C-H (panel B) of deuterated formamide.The hydrouridyl radical 20a or 20 subsequently tautomerizes to its more stable form 21 or 21a via solvent assisted hydrogen atom transfer.Hence, the deuteration of 6 in the C6 position is observed solely in the N-d2-formamide reaction (6b in panel B).Moreover, the formamide molecules act as bridges for proton relay.In the photoreaction in N-d2formamide, H to D exchange results in a six-fold decrease of the tautomerization rate, resulting in a longer lifetime of the enol hydrouridyl radical 20, which may more easily undergo radical coupling with the formamide radical 23 yielding enol 24.After deuterium exchange between enol 24 proton and deuterium of N-d2-formamide, C5 deuterated formamide adducts of uridine 26 could be formed via 25 after keto-enol tautomerization.In the reaction with 1-d-formamide, final abstraction of another deuterium atom from another 1-d-formamide molecule to the C5 position of the partially hydrogenated pyrimidine ring yields 6a.Selective deuteration of 6a at C5 position observed in the 1-d-formamide reaction corroborates the computational suggestion that the final deuterium atom abstraction yields the carbon-centered radical of formamide 19, which could dimerize to furnish byproduct oxamide 22.In addition to the stepwise (proton transfer following charge transfer) mechanism, a concerted proton coupled electron transfer (PCET) mechanism (panel C) can also occur and play a role in the photochemistry of uridine in formamide solution.The concerted PCET enables the hydrogen atom transfer from the amino group of formamide to the C5 atom of uridine, producing nitrogen centered formamide radical 27 and uridine radical 28.
The energetically unstable radical 27 either tautomerizes to the carbon-centered formamide radical 23 or abstracts a proton from the C5 atom of uridine radical 28, yielding C5 deuterated uridine 29, which could be isolated together with natural uridine 2 in an uncompleted photoreaction.Prolonged irradiation in N-d2-formamide would convert 29 to 5,6-dideutero-uridine 6c.Recombination of radicals 28 and 23 produces formamide adduct 26.Hydrouridyl radical 28 could also abstract a hydrogen atom from N-d2-formamide to give 6a.

Energy [eV]
Reaction coordinates    Fig. S51 -The Gibbs free energy profiles, obtained at the PCM/ωB97X-D/def2-TZVP level, show a proton transfer from the C4-O to the C6 site of the hydrouridine (U) radical through two formamide molecules, one in the enol and the other keto form, leading to more stable radical.The presented proton relay transition state was computed for N-h2-formamide or N-d2-formamide molecules to estimate the primary isotopic effect of the proton relay mechanism, that is kH/kD = 5.9.
Fig. S52 -The Gibbs free energy profiles, obtained at the PCM/ωB97X-D/def2-TZVP level, demonstrate a hydrogen atom abstraction mechanism in which either 1-hformamide or 1-d-formamide molecule donates the hydrogen atom to the C5 position of the nucleoside resulting in the formation of dihydrouridine and formamide radical.The presented transition state was computed for 1-h-formamide or 1-d-formamide molecules to evaluate the primary isotopic effect of the hydrogen atom abstraction process, that is kH/kD = 4.97.

Fig. S11 1 H
Fig. S11 1 H NMR spectra of the mixture after photoreaction of uridine (60 mM) in formamide, and the spiking experiment with dihydrouridine.a) 1 H NMR spectrum for the reaction mixture after 7 hours of irradiation; b) as a), spiked with dihydrouridine.

Fig. S12 1
Fig. S12 1 H NMR spectrum of the mixture after photoreaction of uridine (16 mM) in formamide with added sodium succinate (150 uL of 50 mM solution).

Fig. S13 1 H
Fig. S13 1 H NMR spectrum of the mixture after photoreaction of uridine (32 mM) in formamide with added sodium succinate (150 uL of 50 mM solution).

Fig. S14 1
Fig. S14 1 H NMR spectrum of the mixture after photoreaction of uridine (60 mM) in formamide with added sodium succinate (250 uL of 50 mM solution).

Fig. S15 1
Fig. S15 1 H NMR spectrum of the mixture after photoreaction of uridine (100 mM) in formamide with added sodium succinate (250 uL of 50 mM solution).

Fig. S16 1 H
Fig. S16 1 H NMR spectra of the mixture after photoreaction of uridine (16 mM) in formamide/water mixtures.a) Reaction in 10% formamide in water (containing 13 % of remaining uridine); b) Reaction in 25% formamide in water; c) Reaction in 50% formamide in water; d) Reaction in 25% water in formamide; e) Reaction in 10% water in formamide.

Fig. S28 13 C
Fig. S28 13 C NMR spectrum of the mixture after photochemical reaction of uridine in formamide (enlarged high field region) and the spiking spectrum with oxamide.a) 13 C NMR spectra of the reaction mixture in DMSO-d6; b) as a), spiked with oxamide.

Fig. S29 1 H
Fig. S29 1 H NMR spectra of cytidine and the mixture after photoreaction of cytidine (16 mM) in formamide.a) 1 H NMR spectrum of cytidine in D2O; b) 1 H NMR spectrum of cytidine photoreaction in formamide.

Fig. S30 1
Fig. S30 1 H NMR spectra of adenosine and the mixture after photoreaction of adenosine (16 mM) in formamide.a) 1 H NMR spectrum of adenosine in D2O; b) 1 H NMR spectrum of adenosine photoreaction in formamide.

Fig. S31 1 H
Fig. S31 1 H NMR spectra of adenosine and the mixture after photoreaction of guanosine (16 mM) in formamide.a) 1 H NMR spectrum of guanosine in D2O; b) 1 H NMR spectrum of guanosine photoreaction in formamide.

Fig. S32 1 H
Fig. S32 1 H NMR spectra of inosine and the mixture after photoreaction of inosine (16 mM) in formamide.a) 1 H NMR spectrum of inosine in D2O; b) 1 H NMR spectrum of inosine photoreaction in formamide.

Fig. S33 1 H
Fig. S33 1 H NMR spectra of cytidine and uridine, photoreaction of the mixture (16 mM each) in formamide.a) 1 H NMR spectrum of uridine in D2O; b) 1 H NMR spectrum of cytidine in D2O; c) 1 H NMR spectrum of dihydrouridine in D2O; d) 1 H NMR spectrum of photoreaction of mixtures of uridine and cytidine in formamide.

Fig. S34 1
Fig. S34 1 H NMR spectra of photoreaction of mixtures of A, G, C, U in formamide.a) 1 H NMR spectrum of guanosine; b) 1 H NMR spectrum of adenosine; c) 1 H NMR spectrum of cytidine; d) 1 H NMR spectrum of uridine; e) 1 H NMR spectrum of photoreaction of mixture in formamide; f) 1 H NMR spectrum of dihydrouridine.

Fig. S43 -
Fig.S43-The natural transition orbitals involved in electronic transitions of the lowestlying three excited states for the complex of uridine and formamide molecules, assuming the equilibrium ground-state structure.

Fig. S44 -
Fig. S44 -In the first row, there are shown the optimized S0 and S1 structures obtained using the PCM/ωB97X-D/def2-TZVP and SCS-ADC(2)/aug-cc-pVDZ methods, respectively.Below are depicted natural transition orbitals, generated for the S1minimum-energy geometry, which participate in the nOp* transition.The wavefunction analysis revealed an intermolecular charge transfer of 0.20 e -from the formamide nO orbital to the uridine p* molecular orbital in the S1 state.

Fig. S45 -
Fig.S45-The non-radiative photorelaxation pathway demonstrates a high-energy chemical reaction that starts from the Franck-Codon region (5.24 eV), then the system reaches the S1 minimum at 3.78 eV, and subsequently, there is the S1 transition-state geometry (4.11 eV) enabling for the hydrogen atom transfer from formamide to the uridine carbonyl oxygen atom.The excited-state intermolecular hydrogen transfer leads to the biradical system, having hydrouridine and formamide radicals, and the formation of the S1/S0 minimum-energy crossing point (3.49), allowing for the deactivation to the electronic ground state.The presented photochemical mechanism possesses only one energy barrier between the S1 minimum and S1 transition-state structure, equal to 0.33 eV.The orange and green lines correspond to the 1 pp* and 1 nOp* excited states, respectively.The black line shows the electronic closed-shell ground state.The ground and excited states potential energy surfaces were obtained at the SCS-MP2/aug-cc-pVDZ and SCS-ADC(2)/aug-cc-pVDZ level of theories, respectively.

Fig. S46 -Fig. S47 -
Fig.S46-The radiationless photorelaxation channel shows a high-energy chemical reaction that initiates from the Franck-Codon region (5.24 eV).Then, the model system goes towards the S1(nOp*)/S0 conical intersection seam at 4.55 eV, which corresponds to the S1(nOp*) transition-state excited-state geometry.The photochemical reaction enables the hydrogen atom transfer from formamide to the uridine C5 atom in the concerted proton-coupled electron transfer manner (PCET).The concerted PCET results in forming the biradical system composed of hydrouridine and formamide radicals.Reaching the intersection seam (at 4.55 eV) allows for the deactivation to the electronic ground state.The green line corresponds to the 1 nOp* excited state.The black line depicts the electronic closed-shell ground state.The ground and excited state potential energy surfaces were obtained at the SCS-MP2/aug-cc-pVDZ and SCS-ADC(2)/aug-cc-pVDZ level of theories, respectively.

Fig. S49 -
Fig. S49 -The S1(np*) minimum-energy structure of the cytidine and formamide complex obtained at the SCS-ADC(2)/aug-cc-pVDZ level of theory.The shown natural transition orbitals represent the occupied n and virtual p* molecular orbitals involved in the S1 state.The wavefunction analysis of the structure demonstrated that there is no net charge transfer between the molecules.

Fig. S53 -
Fig. S53 -The Gibbs free energy profiles present the formation of the C6-adduct of the hydrouridine and formamide radicals, computed at the PCM/ωB97X-D/def2-TZVP level of theory.

Fig. S54 -
Fig. S54 -The Gibbs free energy profiles present the stability of formamide radicals obtained at the PCM/ωB97X-D/def2-TZVP level of theory.

Figure S56. Uridine 2
Figure S56.Uridine 2 was reformed by dehydration of uridine hydrate during evaporation of formamide at 60 ºC under high vacuum.a) 1 H NMR of uridine 2 standard; b) 1 H NMR of uridine hydrate 4 formed from photoreaction of uridine in H2O; c) 1 H NMR of the mixture after removing formamide from 4 formamide solution by evaporation.

Figure S57 .
Figure S57.Dihydrouridine was formed (in 50% yield) by photoreaction of uridine in water in the presence of sodium formate (12 equivalents).a) 1 H NMR of the mixture after 2 hour's irradiations; b) as a), spiked with dihydrouracil; c) as b), spiked with dihydrouridine.

Figure
Figure S58.UV irradiation of thymidine at 254 nm.a) 1 H NMR of thymidine standard; b) 1 H NMR of the mixture after 2 hour's irradiations of thymidine in formamide; c) 1 H NMR of the mixture after 2 hour's irradiations of thymidine in water in the presence of formate (12 equivalents).

Figure S59 .
Figure S59.UV irradiation of trimer UAA at 254 nm.a) 1 H NMR of UAA before irradiation; b) 1 H NMR of the mixture after 2 hour's irradiations of UAA in formamide.

Figure S60 .
Figure S60.UV irradiation of trimer UAC at 254 nm.a) 1 H NMR of UAC before irradiation; b) 1 H NMR of the mixture after 2 hour's irradiations of UAC in formamide.

Figure S61 .
Figure S61.Quantitative 13 C NMR spectrum of the mixture after irradiation of 60 mM of uridine in formamide.(The yield of oxamide (20%) was calculated by using DHU (54%) as internal standard.) -TZVP level of theory.