Introduction

N-Fused heterocycles are ubiquitous within crucial molecules, including biologically active natural products, pharmaceuticals, and functional materials (Figure 1) .

Figure 1: Representative examples of relevant N-fused heterocycles.

It has been assessed that almost one-third of the best-selling therapeutics contains fused heterocyclic structures . Among the N-heterocycles, imidazopyrazine structures , derived from amalgamation of privileged imidazole and pyrazine pharmacophores, are well represented in the area of medicinal chemistry since they possess pharmacological properties as mammalian target of rapamycin (mTOR) inhibitors , adenosine triphosphate (ATP) competitive inhibitors of the insuline-like growth factor 1 (IGF-1) receptor related to Ewing sarcoma , IGF-1 receptor inhibitors or act as ligands on corticotropin releasing hormone (CRH) , γ-aminobutyric acid (GABA) and melanocortin receptors .

Given the established potencies of this class of N-ring-fused compounds, planned syntheses that simplify their preparation by using small building blocks and that lead, through appropriate transformations, to a product that becomes a substrate for another complexity-generating reaction, merit investigation .

Herein, we report a 3-CR-based synthesis of new properly decorated (thio)hydantoin framework able to afford, by a chemospecific Lewis acid-catalyzed ring-closure protocol, valuable heterocyclic N,O-aminals (Scheme 1).

Scheme 1: Different acid-catalyzed six-membered ring cyclizations.

Results and Discussion

Since the direct functionalization of N-heterocycles offers an attractive entry to important molecular targets that might otherwise require lengthy synthetic procedures , our consolidated 3-CR strategy implicates a careful selection of the starting components that ensures the installation of functionalities to be converged, by regioselective control, in different ring-closing processes .

With these considerations in mind and with the aim of diversity-oriented synthesis of N-heterocycles via sequential multicomponent approaches, we envisioned that α-aminoacetals could act as bifunctional building blocks along with 1,2-diaza-1,3-diene (DD) coupling partners , in obtaining functionalized N-aminohydrazones as key intermediates.

Based on our previous findings , the initial nucleophilic addition of α-aminoacetals 2a,b as nitrogen source to the activated heterodiene system of 4-methoxycarbonyl-DDs 1a–f in dichloromethane (DCM) or ethanol (EtOH) at room temperature affords N-aminohydrazone derivatives I (Scheme 2), whose sequential acylation process by iso(thio)cyanates 3a–h gives rise to the asymmetric (thio)urea derivatives (intermediate II). The spontaneous nucleophilic attack of the (thio)amide nitrogen on the terminal methyl ester function at C-4 of the starting azo-ene system provides a regioselective heteroring closure, positioning appropriate functions both at N-3 and C-4 of the (thio)hydantoin frameworks 4a–r (30–81%) broadening their usable decorations (Scheme 2).

Scheme 2: Substrate scope for the assembly of suitably N-3-functionalized (thio)hydantoins 4a–r. aDCM was utilized as solvent with isothiocyanates 3a–f, while bEtOH was utilized with isocyanates 3g,h.

Recently, we reported that compounds 4a, 4f, and 4m undergo an intramolecular cyclization process through the involvement of the restored keto function of the hydrazone moiety and the open-chain hemiacetal or aldehyde hydrate in Brønsted acid medium to access 1H-imidazo[5,1-c][1,4]oxazine derivatives (Scheme 1) .

Considering that the hydrazone function at C-4 of 4a–r may exist in a tautomeric equilibrium with the corresponding ene-hydrazino form , we conceived the idea of reversing the reactivity of 4a–r in the six-membered cyclization process (N- vs O-annulation) through the generation of an electrophilic oxocarbenium cation intermediate from the acetal residue at N-3 of the (thio)hydantoin core. To pursue our goal, different Lewis acids (10 mol %) such as Zn(OTf)2, CuCl2, and FeCl3 were screened at room temperature in different solvents, employing compound 4a as the model substrate (Table 1).

Table 1: Optimization conditions for the Lewis acid-catalyzed intramolecular cyclization of 4a.

Entrya Lewis acid Solvent Time (h) 5a (yield %)b 6a (yield %)b 1 Zn(OTf)2 (10 mol %) DCM 42 38 22 2 CuCl2 (10 mol %) DCM 96 28 15 3 FeCl3 (10 mol %) DCM 86 44 31 4 FeCl3 (20 mol %) DCM 38 63 18 5 FeCl3 (30 mol %) DCM 2 73 8 6 FeCl3 (30 mol %) ACN 2 68 13 7 FeCl3 (30 mol %) THF 2 62 11 8 FeCl3 (30 mol %) EtOH 48 –c –c

aThe reactions were performed on a 0.5 mmol scale in 5 mL of solvent. bIsolated yields of products 5a and 6a based on starting 4a. cNot detected by TLC analysis.

From the set of data collected, both the formation of N,O-aminal 5a and corresponding hemiaminal 6a were observed (entries 1–7, Table 1). Similarly to what was observed by Yu and co-workers for the intramolecular cyclization of alkynyl aldehyde acetals , it was found that the use of FeCl3 provided the better result in terms of overall yield (entry 3, Table 1). Moreover, the choice of iron(III) seemed to have remarkable advantages such as an environmentally benign alternative to traditional transition-metal catalysis, a low cost, nontoxicity, good stability, and easy handling . Upon increasing the amount of FeCl3 to 20 mol %, the time of the reaction was reduced from 86 to 38 hours, and the yield of 5a was incremented with respect to 6a (entry 4, Table 1). Rising the amount of FeCl3 to 30 mol %, the reaction was complete in 2 hours, enhancing the yield of 5a (73%) and minimizing the yield of 6a (8%) (entry 5, Table 1). In reactions carried out in acetonitrile (ACN) or tetrahydrofuran (THF) the yield of 5a decreased, while utilizing ethanol the reaction proceeded slowly and produced a complicated mixture in which both 5a and 6a were not detected (Table 1, entries 6–8).

With the optimized conditions in hand (Table 1, entry 5), a selection of N-3-functionalized (thio)hydantoins (4a–r, 1 mmol) were dissolved in DCM (10 mL), FeCl3 (30 mol %) added and magnetically stirred at room temperature. Within 2–30 h, the reactions went to completion (TLC monitoring), affording, at last, N,O-aminals 5a–r (42–82%) and the corresponding hemiaminals 6a–p (4–35%) after column chromatography (Scheme 3).

Scheme 3: Substrate scope of the iron(III)-catalyzed synthesis of functionalized heterocyclic N,O-aminals 5a–r and hemiaminals 6a–p.

An increased yield of 6 was observed alongside a decreased yield of 5, in all those cases that required prolonged reaction times (24–30 h). This event led us to suppose the formation of carbinolamine 6 from N,O-aminal 5 owing to the nucleophilic attack of a water molecule, probably caused by the enriched moisture content of the reaction environment during the time.

Then, to explain the related formation of 5 and 6, we hypothesized a plausible reaction mechanism in which iron is involved in two concomitant catalytic cycles (Scheme 4). Initially, FeCl3 forms an acid–base complex with one of the alkoxy groups of 4 providing intermediate A. The latter, by loss of a trichloro(alkoxy)ferrate(III) anion, generates a strong electrophile such as the oxocarbenium cation intermediate B. The released trichloro(alkoxy)ferrate(III) splits into FeCl3, which enters the catalytic cycle, and a free alkoxide, which acts as a base, promoting, via hydrazone–enamine tautomerization , the nucleophilic addition which concludes with the construction of the heterocyclic N,O-aminal 5 through the intramolecular N–C bond formation. The FeCl3 can also interact with the newly formed N,O-aminals 5, giving rise to the second parallel catalytic cycle. Similar to what was previously observed, the elimination of the trichloro(alkoxy)ferrate(III) anion from intermediate C provides the iminium ion D, susceptible to nucleophilic attack by a water molecule present in the reaction medium, leading to the carbinolamines 6. This latter synthesis represents an interesting example of auto-tandem catalysis in which FeCl3 promotes two subsequent reactions.

Scheme 4: Proposed mechanism for the formation of N,O-aminals 5 and hemiaminals 6.

For further confirmation to support our mechanistic hypothesis and in an attempt to switch the reaction toward the formation of hemiaminal 6a, we repeated the reaction of thiohydantoin 4j, (chosen as representative substrate), under the same previously optimized conditions, but extending the reaction time to 240 hours (experiment A in Scheme 5). In this case, the yield of hemiaminal 6a increased from 6% recorded after two hours at the complete conversion of 4j (Scheme 3) to 21% (experiment A, Scheme 5), in line with the values found for the slower reactions previously described (compounds 6b,d,g,i,n–p). By adding 500 μL of water to the medium the cyclization did not proceed, and the starting material 4j was recovered unchanged (experiment B, Scheme 5). This observation seems to suggest that the presence of a high water amount results in catalyst deactivation.

Scheme 5: Control mechanistic experiments.

Based on these results and what was observed in the optimization tests (Table 1, entry 6), we extended the reaction time but used ACN as solvent, which possesses a higher water content with respect to DCM (experiment C, Scheme 5). Gratifyingly, in this case, the formation of carbinolamine 6a becomes predominant (45%), despite a small quantity of N,O-aminal 5j (14%) is also produced, by virtue of the alcohol released from the starting acetal 4j. Probably, the higher water concentration in acetonitrile shifts the equilibrium in favour of 6a over the time (Scheme 4).

Within our proposed catalytic cycle, when compound 4j is utilized, methanol is released, due to the presence of a dimethyl acetal residue. Therefore, using molecular sieves (MS 4 Å) little alcohol molecules, such as MeOH, can be potentially trapped, allowing the insertion of a more encumbered alcohol, such as benzyl alcohol, which is not sequestered by MS 4 Å. As a matter of fact, the new benzylated N,O-aminal 7 was successfully obtained in 70% isolated yield as the sole product (experiment D, Scheme 5). In this latter case, the benzyl alcohol presumably reacts with the iminium ion D formed in the second catalytic cycle (Scheme 4), and its sole formation is ascribable to the capability of molecular sieves of sequestering MeOH eventually formed, shifting the equilibrium towards 7.

Noteworthy, in compounds 5a–r, 6a–p and 7, the newly created heterocyclic nucleus represents a new example of cyclic N,O-aminals and carbinolamine derivatives, an interesting class of organic compounds that are common structural motifs embedded within diverse biologically important natural products and pharmaceuticals . On the other hand, the N,O-aminals are stable and very practical synthetic intermediates commonly employed for the in situ generation of highly electrophilic iminium ions .

Conclusion

In summary, we planned the synthesis of decorated imidazo skeletons accessible through a judicious choice of the starting components of a 3-CR process and developed a catalytic system-controlled selective intramolecular N-annulation process for ring-fused biheterocyclic N,O-aminal derivatives as stable imine equivalents and useful tools for new bond formation in view of further fused-heterocylization processes. Moreover, control experiments corroborate our mechanistic hypothesis related to the formation of both N,O-aminals and corresponding hemiaminals. In particular, the domino reaction that leads to the carbinolamines represents an interesting example of “auto-tandem catalysis” in which the FeCl3 catalyzes two different chemical transformations in a single reactor, reducing the number of steps and the amount of waste with consequent benefits of cost and environmental impact .