Introduction

DNA cytosine methyltransferases [1] deaminate cytosine (C, 1) to uracil (U, 3), and cytosine deamination also can be effected by nitrosation [2, 3]. C-to-U damage can be repaired via enzymatic base excision [4] and, if left unrepaired, causes the G:C→A:T mutation [5, 6]. The nitrosative C-to-U process had been thought to occur via the transient cytosinediazonium ion 2 and its hydrolysis by direct nucleophilic aromatic substitution (Fig. 1), which is very much SN1Ar-like [7]. However, theoretical study of the unimolecular dediazonation of cytosinediazonium ion [8] showed that the classical diazonium ion 2 (5←N≡N) is merely a shallow minimum and less stable than free 5 and N2 and that an electrostatic complex 4 (5 N≡N) is bound by only 5.4 kcal mol−1. The HO-tautomer of cytosinediazonium ion 2′ (not shown in Fig. 1) is 3.7 kcal mol−1 more stable than 2 itself. 2′ has a classical diazonium-ion structure and it is bound by 10.8 kcal mol−1 relative to 5′ and N2. Hence, cytosine deamination essentially produces a free cation, 5 or 5′. More recently, it has been shown that the ions 5 or 5′ are best described as cyclic nitrilium ions 6 and 6′ and that the dative bond easily breaks to form the more stable ring-opened cations 7 and 7′. These results seriously put in question whether uracil is the only product [9] of nitrosative cytosine deamination.

Fig. 1
figure 1

Nitrosative deamination of cytosine and cytidine

If the R-group is ribose or 2′-deoxyribose, 6 and 7 are the appropriate models for the deamination of cytidine, its nucleotides CMP, CDP, and CTP, and their 2′-deoxy derivatives. Ubiquitous water may react with 7 by addition, substitution (R≠H), or deprotonation (R=H). Water addition forms of carbamic acid [10] and subsequent decarboxylation [11] leads to (Z)-3-aminoacrylonitrile 9. (Z)-3-aminoacrylonitrile is susceptible to base- [12, 13] and acid-catalyzed [14] nucleophilic addition to the C=C and C≡N bonds and this chemistry also might lead to DNA adducts. Alternatively, deglycation by substitution (R≠H) or deprotonation (R=H) would lead to (Z)-3-isocyanatoacrylonitrile 10. Unsaturated isocyanates are toxic [15] and 3-isocyanatoacrylonitrile can form adducts and cross-links [16, 17].

Here, we report the results of a study of the formation of 10 by deprotonation of 7 and 7′, of the E/Z-preferences of 3-isocyanatoacrylonitrile 10 and 11 and of their conformers 12 and 13, and of C−N rotation–inversion in both E/Z-isomers via TS1 and TS3 (Fig. 2).

Fig. 2
figure 2

Scope of the present study of 3-isocyanatoacrylonitrile

Computational methods

Geometry optimizations and vibrational analyses were carried out at the MP2(full)/6-31G** level [18] with the program Gaussian03 [19]. We employed perturbation theory deliberately because it is preferable to (hybrid) density functional theory for the present purpose. Structure (Z,s-cis)-10 benefits from through-space interaction and such dispersion interactions are not properly accounted for by the standard DFT models [20]. Atomic charges were calculated with the natural bond orbital (NBO) method [21, 22] at the same level. Molecular models of the minima and major structural data and atom and fragment charges are reported in Figs. 3 and 4, respectively. Molecular models of the transition-state structures are shown in Fig. 5. Coordinates of all stationary structures are available as Supporting Information.

Fig. 3
figure 3

MP2(full)/6-31G** structures of 713

Fig. 4
figure 4

NBO analysis of 713. Numbers printed in bold are fragment charges

Fig. 5
figure 5

Molecular models of the MP2(full)/6-31G** structures of TS1 and TS3

Total energies E (in Hartrees), vibrational zero-point energies VZPE (kcal mol-1), thermal energies TE (kcal mol−1, 298.15 K), and entropies S (cal mol−1 K−1) are reported in Table 1. In Tables 2 and 3 are reported the values ΔE, ΔE 0 = ΔE + ΔVZPE, ΔH 298 = ΔE + ΔTE + ΔRT, and ΔG 298 = ΔH 298 − 0.29815·ΔS. The discussion refers to Gibbs free energies unless otherwise noted.

Table 1 Total energies and thermodynamical data
Table 2 Relative energies, trans-preferences, Z-preferences and proton affinities (kcal mol−1)
Table 3 Activation barriers for C−N rotation–inversion (kcal mol−1)

Results and discussion

Stereochemical preferences

Studies of α,β-unsaturated isocyanates are scarce [23, 24] and the stereochemistry of vinyl isocyanates with additional functionality remains unexplored. Structures 10-13 (Fig. 3) are almost isoenergetic (Table 2). (Z,s-cis)-10 is the most stable isocyanatoacrylonitrile and the relative energies of isomers and conformers only are 0.1–0.9 kcal mol−1. Z-preferences occur for the s-cis- and s-trans-conformers and they are higher for the s-cis-structures. If steric interactions dominated, one would expect preferences for the E-configuration and the s-trans conformation; (E,s-trans)-13. The calculated relative energies of 0.87 kcal mol−1 for (E,s-cis)-11 vs. (E,s-trans)-13 and of 0.28 kcal mol−1 for (Z,s-trans)-11 vs. (E,s-trans)-13 are in line with expectation based on steric demand. Assuming additivity, one would thus estimate an energy of about 1.15 kcal mol−1 for (Z,s-cis)-10 relative to (E,s-trans)-13 for steric reasons. Yet, (Z,s-cis)-10 is 0.05 kcal mol−1 more stable than (E,s-trans)-13 and, hence, 10 benefits from through-space neighboring group attraction of about 1.2 kcal mol−1.

Electronic structures and through-space interactions

The NBO analysis shows highly polar electronic structures and the main effect of the isocyanato group on the acrylonitrile [14] is an increase of the electron-deficiency of the CH group to which it is attached (Fig. 4). The CN groups in 1013 are highly dipolar but their overall charge is very small. On the other hand, the NCO groups in 1013 are charged significantly, they carry charges of about −0.2, and they are extremely quadrupolar [25]. These findings suggest that the attractive neighboring-group interaction in 10 is due in part to polarization of the cyano group in the field of the overall charge of the isocyanato group and in part to dipole–dipole attraction. The dipole-dipole attraction between the CN bond dipoles in the NCO and CN groups appears to dominate the repulsion between CO bond dipole of the NCO group and the CN group dipole. The NBO charge analysis thus provides a consistent rationale based on simple and widely accepted concepts of interaction and reactivity. The proposed interactions can be quantified via fragment-interaction analysis [26, 27] and recent progress in the development of orbital deletion analysis [22, 28] also might provide for interesting tests of the proposed rationale and any possible hyperconjugation.

The major effect of the protonation of the NCO group is a withdrawal of electron density from the alkene fragment; in the cations the HNCO and alkene fragments carry charges of about 0.6 and 0.4, respectively. Note that the protonation of either one of the heteroatoms of the NCO group does very little to increase the electrophilicity of the heterocumulene’s center. Instead, there is a rearrangement of electron density from one heteroatom to the protonated one.

Rotation-Inversion

A priori the s-cis/s-trans interconversion might involve in-plane N-inversion or rotation about the OCN–CH bond. The N-inversion would involve an N-hybridization change from sp2 to sp and shorten both bonds to N. Hybridization changes would be small along the rotation path and there would be some loss of π-conjugation.

The ideal inversion process would involve a planar transition-state structure. We searched for these transition-state structures under the constraint to C s -symmetry, and found the resulting structures C s -TS1 (i118, i101 cm−1) and C s -TS3 (i119, i81 cm−1) to be second-order saddle points on the potential-energy surface. The true transition-state structures TS1 and TS3 were located (Fig. 5) and the activation barriers are less than 3 kcal mol−1 (Table 3).

The ∠(C−C−N−C) dihedral angles β in TS1 and TS3 are 93.4° and 84.3°, respectively, and close to 90°. However, the ∠(C−N−C) angles α in TS1 and TS3 are 144.1° and 144.5°, respectively, and it is clear that these are not the structures of “pure” rotational transition states.

The rotation–inversion paths are illustrated schematically in Fig. 6. The “pure” rotation would leave α≈120° relatively unchanged as the rotation proceeds from β=0° to β=180°. The “pure” inversion would change α from α≈120° to α≈180° while β=0° and then from α≈180° to α≈120° while β=180°. The change in β at α=180° causes the discontinuity on the right in Fig. 6. In the present case, neither of these ideal paths exists and rotation–inversion paths occur instead.

Fig. 6
figure 6

Illustration of the rotation–inversion paths and the position of the transition state structures in the (α, β)-map. The “pure” rotation and inversion paths follow the green and purple arrows, respectively. (Horizontal axis is not drawn to linear scale)

Proton affinities

Isocyanate 10 shows a pronounced preference of about 10 kcal mol−1 for protonation at N rather than at O (Table 2). The proton affinity for N-protonation is 166.4 kcal mol−1 and very close to the proton affinity of water, which is about 167 kcal mol−1 [29]. Given that 7 is formed in aqueous solution, this result provides strong evidence that the reaction H2O + 710 + H3O + occurs under typical conditions of nitrosative deamination.

Conclusions

Pyrimidine ring-opening in nitrosative deamination of guanine, guanosine, and its nucleotide derivatives has been well established by theoretical study [20, 30, 31] and experimentation [32, 33]. The unimolecular dediazoniations of adeninediazonium and cytosinediazonium ions can proceed without ring-opening and it was realized only more recently that the cations which are produced by dediazoniation of adeninediazonium ion [34] and cytosinediazonium ion [9] can ring-open with very little kinetic hindrance.

The proton affinity of 10 suggests that 10 and its isomers and conformers are formed in nitrosative cytosine deamination in aqueous solution. The possibility for recyclization of 10 to uracils has been explored experimentally and it is not likely [35]. Thus, toxicological studies of nitrosative cytosine deamination need to direct attention at 3-aminoacrylonitrile 9 and 3-isocyanatoacrylonitrile 10. It remains to be seen whether 10 can be formed by hydrolytic deglycation from the respective cytidine derivatives.

Once 10 is formed, it can react with water to carbamic acid 8, with alcohols to carbamates 14, with amines to ureas 15, and with thiols to carbamothioates 16, etc. as shown in Fig. 7 for conformer 12 [36]. Note that this is a second path to 8 and on to 9, and this path differs from the sequence 789 in one very important way: 10 has some lifetime and 8 and 9 would be formed at some distance from the site of the deamination event while 7 survives merely until a nucleophile diffuses to it (usually water). The 3-isocyanatoacrylonitrile thus has the potential to cause biological damage in a larger region around the site of nitrosation and this neutral carcinogen can be much more selective in its reactions. In addition to the types of isocyanate adducts shown in Fig. 7 (and their isomers and conformers), there also exist further possibilities for additions to the acrylonitrile moiety [14], either alternatively or subsequently. Fig. 7 provides an example for a urea adduct formation with dG and subsequent intrastrand dG-to-dA cross-link formation. The exploration of the chemistry of 3-isocyanatoacrylonitrile promises to be interesting and complicated.

Fig. 7
figure 7

Possible products of nucleophilic addition to 12 and examples for adduct and cross-link formation