Mechanism, kinetics and selectivity of selenocyclization of 5-alkenylhydantoins: an experimental and computational study

Summary The mechanism and selectivity of a bicyclic hydantoin formation by selenium-induced cyclization are investigated. The proposed mechanism involves the intermediates formed by an electrophilic addition of the selenium reagent on a double bond of the starting 5-alkenylhydantoin prior the cyclization. These intermediates are readily converted into the more stable cyclic seleniranium cations. A key step of the mechanism is an intramolecular cyclization which is realized through an anti-attack of the internal nucleophile, the amidic nitrogen, to the seleniranium cation yielding the intermediate imidazolinium cations. Their deprotonation is followed by the formation of the fused bicyclic reaction products. Important intermediates and key transition states are studied by using density functional theory (DFT) methods. The pathways of the reaction are investigated in detail. There are two regioselective pathways related to 5-exo and 6-endo products. Theoretical calculations and the monitoring of the cyclization reaction using 1H NMR spectroscopy are in a good agreement with the proposed mechanism and are consistent with our experimental results. The preferred pathway for formation of 5-exo products is confirmed.

In our previous work, we reported the experimental results on the selenocyclization of 5-alkenylhydantoins leading to fused bicyclic products [24]. To our best knowledge, no previous theoretical work has reported on this reaction. In this paper, density functional theory (DFT) [25][26][27][28] is utilized to understand the mechanism and selectivity of this key step in the synthesis of fused bicyclic hydantoins. 1 H NMR spectroscopy monitoring, as well as 1 H NMR chemical shifts prediction [29][30][31][32][33][34] is used as a useful tool in the search of the most probable intermediates in the reaction. The theoretical results are discussed and compared with our experimental observations.

Results and Discussion
We have recently reported a three-step reaction sequence to cisfused bicyclic hydantoins involving selenium-induced cyclization of 5-alkenylhydantoins as the key step [24]. The reaction is chemo-and regiospecific giving only the five-membered bicyclic hydantoins, 5-exo products, in good to excellent yields (56-96%). The cyclization is stereoselective obtaining a separable diastereomeric mixture, and the products with the bridgehead substituent and the CH 2 SePh group in cis relationship are formed predominately.
To gain deeper insight into the cyclization process leading to the formation of fused bicyclic hydantoins, a computational (DFT) study was carried out. The mechanism of the reaction of 5-alkenylhydantoins with phenylselenyl chloride was examined using the reaction with model substrate 1. The proposed fivestep mechanism, starting with an electrophilic addition of PhSeCl to the double bond, is presented in Scheme 1, while the 5-exo pathway with all stereoisomers is presented in Scheme 2. The first step of the proposed mechanism producing the seleniranium cation, followed by anti-attack of the external nucleophile (Cl − ) to the double bond [35] of 5-alkenylhydantoin is supposed to be reversible, and leads to the formation of two theoretically possible pairs of regioisomeric intermediates, (S,R/ S,S)-INT1 (Markovnikov products) and (S,R/S,S)-INT1' (anti-Markovnikov products) (Scheme 1). A similar mechanism was proposed earlier for cyclofunctionalization of olefinic urethanes [36].
The formation of addition products as intermediates in the second step was proposed after monitoring of the reaction by using 1 H NMR spectroscopy ( Figure 1). Also, in this experiment a regioselective formation of Markovnikov-type products of the addition of selenium reagent to the double bond was noticed. To further verify the nature of these intermediates we have computed 1 H chemical shifts by means of DFT GIAO computations.
In the first spectrum of time-dependent experiment, a signal corresponding to the olefinic proton of starting 5-alkenylhydantoin, triplet of septets at 5.04 ppm (Supporting Information File 1, Figure S1), disappeared completely, while the characteristic signal of the C(5)H proton (see Table 1 for label assignments) of the product 2, doublet of doublets at 3.83 ppm began to appear. This observation prompts us to propose the above mentioned mechanism which includes the product of the addition of PhSeCl on a double bond as intermediate     Starting 5-alkenylhydantoin 1 was synthesized from 6-methylhepten-2-on [24] by Bucherer-Bergs reaction which is not stereoselective and thus obtained as racemic mixture. All calculations were done for the enantiomers with S configuration at the quaternary carbon atom of the hydantoin nucleus. The selected 1 H chemical shifts of these intermediates, calculated and experimental, are listed in Table 1 while correlation plots are given in Supporting Information File 1, Figure S2.
The simulated chemical shifts for the C(5)H proton of (S,R)-  Table S1) it was possible to distinguish a configuration of these isomers, because calculated 1 H chemical shifts for (S,S)-INT1 had a better correlation with the experimental values. Moreover, calculations of relative free energies of these intermediates showed that the S-isomer is more stable (Figure 2 and Figure 3).
In the third step, these intermediates are readily converted into the more stable cyclic seleniranium cations (    From the charge analysis (data given in Supporting Information File 1, Table S2) it is possible to deduce that the C(6) atom is more electrophilic than C(5) due to the positive inductive effect of the two methyl groups. However, on the other hand these methyl groups hinder the approach of the nucleophilic nitrogen N(1) to C(6). The selenium ion binds to the C(5) atom, which becomes positively charged and susceptible to the nucleophilic attack of N(1) atom. Subsequent ring closure is the ratedeterming step and also the step in which the cyclization mode and distribution of the products is determined (Figure 3). This process was followed in 1 Table 1). It is well known that oxygen could compete with nitrogen as internal nucleophile leading to different heterocycles [37,38]. Disappearing of the signal of N(1)H is a clear proof that the amidic nitrogen from the hydantoin ring is an internal nucleophile.
The last step of the reaction mechanism is the deprotonation of the intermediate imidazolinium cations by the Cl − anion followed by the formation of the fused bicyclic products. The calculated geometries and relative free energies of products are depicted in Figure 6. It is obvious that the products 2' are less All parameters were calculated in acetonitrile as solvent.
stable than (S,S)-2 due to the repulsive 1,3-diaxial interactions between the two methyl groups. The simulated chemical shifts for C(5)H proton of (S,R)-2 and (S,S)-2 have excellent correlation to the experimental values (Table 1 and Supporting Information File 1, Figure S2).
Our investigations showed that the activation energies for the formation of (S,R)-INT-3' and (S,R)-INT3 (trans-diastereoisomers) are mutually very similar, whereas that for the formation of (S,S)-INT3' and (S,S)-INT3 (cis-diastereoisomers) are significantly lower (Table 2). But the free energy of both sixmembered imidazolinium cations INT3' are very similar and match higher than those for five-membered INT3. The facts that six-membered imidazolinium cations INT3' are very unstable and that the activation energies for the inverse reactions are very low (about 10 kJ/mol) could be an explanation for the regioselective formation of five-membered bicyclic products 2. On the basis of the calculated rate constants it is clear that the formation of (S,S)-INT3 is most favorable.
All these findings show that the formation of (S,S)-2 is both kinetically and thermodynamically favored, which is in agreement with the experimental results that (S,S)-2 is the major product of the reaction [24]. The formation of (S,R)-2 is thermodinamically controlled. A possible reason for the pronounced instability of (S,R)-2 can be the destabilizing steric interaction between the hydantoin ring and the SePh group. Much lower formation of this product is in accordance with experimental results where the diastereomeric ratio cis/trans = 81:19 is obtained [24].
With intent to verify the above proposed mechanism, we explored it on the other model substrate 3 with a different alkenyl moiety (Scheme 3).
The optimized structures of all possible intermediates are presented in Figure 7. From calculated free energies for all intermediates it could be concluded that the formation of fivemembered products is favored again. Also, cis-diastereoisomers are more stable than trans-diastereoisomers, but there is no significant difference (Figure 8). It is in accordance with experimental results where diastereomeric ratio cis/trans = 64:36 is obtained [24].

Conclusion
In summary, this work has provided the first theoretical study of the selenocyclization of 5-alkenylhydantoins. By 1 H NMR monitoring of the reaction we proposed a five-step mechanism which involves the intermediate formed by the addition of selenium reagent on starting 5-alkenylhydantoin. The Markovnikov-type product is favored. These intermediates are readily converted into the more stable cyclic seleniranium cations. The key step is a selenium-induced intramolecular cyclization which is accomplished through an anti-attack of the internal nucleophile, amidic nitrogen, to the seleniranium cation yielding the intermediate imidazolinium cations. The last step of the reaction is deprotonation followed by the formation of the fused bicyclic hydantoins. DFT calculations support a plausible reaction pathway where the ring closure is the rate-determining step. The preferred 5-exo cyclization is confirmed. Elucidation of regio-and stereoisomerism of all intermediates and products were done on the basis of excellent correlation of experimental and calculated 1 H chemical shifts as well as relative free energies. Theoretical calculations are in good agreement with our experimental results.

Experimental
All commercials were used as received without further purification. Model compounds 1 and 3 and the corresponding products 2 and 4, respectively, were synthesized according the reported procedures [24] and fully characterized by spectroscopic methods.

Computational methods
All calculations were conducted using Gaussian 09 [39] with the B3LYP functional [40,41] and the split-valence triple-zeta basis set 6-311G [42,43]. To attain better description of the delocalization effects that are crucial for the geometry and electronic structure of the investigated molecules, diffuse functions were added to the heavy atoms. The p and d polarization functions were also used. Full geometry optimizations, without any symmetry constraints, and frequency calculations were performed for all species in acetonitrile as solvent (dielectric constant = 35.688) using the SMD solvation model [44]. SMD is a continuum solvation model based on the quantum mechanical charge density of a solute molecule interacting with a continuum description of the solvent. Frequency calculations were performed to confirm that the optimized structures are energetic minima (no imaginary frequencies). A natural bond orbital (NBO) analysis [45][46][47] was performed for all species.
Relative free energies were calculated at T = 298 K. Rate constants were calculated using Transition state theory as implemented in TheRate program [48] and 1 M standard state as: where k B , h and R stand for the Boltzman, Planck, and the gas constant, and ΔG ≠ is the free activation energy, which is calculated as the G (Gibbs free energy) difference between transition states and reactants.
The 1 H NMR spectra of the intermediates and products in acetonitrile were simulated. The geometry of TMS in acetonitrile was optimized using the B3LYP/6-311+G(d,p) and SMD models. We chose to use the B3LYP functional based on previously demonstrated good NMR performance for containing systems like those studied here [49]. The nuclear magnetic shielding tensors were calculated for TMS, intermediates and products using the Gaussian GIAO (gauge independent atomic orbital) method. The values for all hydrogen atoms in intermediates and products were subtracted from the value for hydrogen in TMS (31.91). All compounds belong to the C 1 point group, and, thus, all hydrogens show different chemical shifts. For this reason the corresponding mean values were taken to represent the chemical shifts of the protons bonded to a certain carbon.