Investigation of Grignard Reagent as an Advanced Base for Aza-Claisen Rearrangement

Employing iPrMgCl as an advanced base instead of lithium hexamethyldisilazane (LHMDS) resulted in dramatic improvements in aza-Claisen rearrangement. This advance is considered responsible for the increased bulkiness of the alkoxide moiety (including magnesium cation and ligands), followed by a resultant conformational change of the transition state. To support this hypothesis, various substrates of aza-Claisen rearrangement were prepared and screened. In addition, a molecular dynamic simulation study was performed to investigate and compare the structural stability of reaction intermediates.


Introduction
Azacycles, nitrogen containing cyclic molecules, has important biological activity and synthetic utility [1]. Various conversions of azacycle skeletons have contributed to the construction of alkaloid frameworks and the development of important synthetic methodologies (the aza-Cope rearrangement [2], transannulation [3], Diels-Alder cycloaddition [4], and so on [5]). However, their synthetic applications require further development to improve chemical yields, handling, and substrate generality. Aza-Claisen rearrangement (ACR) is one of these methodologies [6]. The [3,3]-sigmatropic rearrangement of nitrogen containing a diene moiety serves as a robust platform to introduce various alkaloid skeletons into natural products or active pharmaceutical ingredients (APIs) [7]. More importantly, with ACR employed, chiral communication and induction of remote stereogenic centers were also reported in a highly selective manner [8,9] (Figure 1). However, this type of rearrangement also requires harsh reaction conditions and at times results in low yields [10].

Introduction
Azacycles, nitrogen containing cyclic molecules, has important biological activity and synthetic utility [1]. Various conversions of azacycle skeletons have contributed to the construction of alkaloid frameworks and the development of important synthetic methodologies (the aza-Cope rearrangement [2], transannulation [3], Diels-Alder cycloaddition [4], and so on [5]). However, their synthetic applications require further development to improve chemical yields, handling, and substrate generality. Aza-Claisen rearrangement (ACR) is one of these methodologies [6]. The [3,3]sigmatropic rearrangement of nitrogen containing a diene moiety serves as a robust platform to introduce various alkaloid skeletons into natural products or active pharmaceutical ingredients (APIs) [7]. More importantly, with ACR employed, chiral communication and induction of remote stereogenic centers were also reported in a highly selective manner [8,9] (Figure 1). However, this type of rearrangement also requires harsh reaction conditions and at times results in low yields [10].  ACR proceeds through deprotonation, [3,3]-sigmatropic rearrangement, and protonation, as shown in Figure 2. For this process, both a strong base for deprotonation of α-hydrogen in an amide ACR proceeds through deprotonation, [3,3]-sigmatropic rearrangement, and protonation, as shown in Figure 2. For this process, both a strong base for deprotonation of α-hydrogen in an amide group and a high temperature for thermal rearrangement are required [11,12]. However, this reaction condition also permits sigma-bond rotation in amide enolate to hamper generation of the desired conformation. If the undesired conformation exists as a major form, ACR would be impossible (low conversion), and an ensuing side reaction would occur (side product). Because the oxygen-metal bond may play an important role in this enolate conformation, it can be presumed that the cation of a base determines a successful ACR process. Actually, it was shown that lithium hexamethyldisilazane (LHMDS) and iPrMgCl gave different results in the ACR of an API synthesis [13]. Inspired by this base-dependent ACR, we tried to apply this methodology to a more general substrate. It was expected that controlling an amide enolate conformation's equilibrium through various enolate moieties, including metal cations and ligands, would mitigate the undesired side reaction in accelerated ACR processes. Based on this hypothesis, a simple ACR substrate was designed to screen bases.

Results
To prove the efficiency of the cation-dependent ACR, large amounts of substrate were pursued in a short time. In addition, cyclic ACR substrate was favored over acyclic because the ACR of cyclic substrates proceeds with fewer entropic/geometric issues [12]. Commercially available methyl pipecolinate 1 was subjected to the sequential Schotten-Baumann amidation [14] to create amide 2, which was converted to aldehyde 3 via diisobuylaluminum hydride (DIBAL) reduction. Finally, Wittig methylenation afforded allyl amide 4 an excellent yield, as depicted in Scheme 1.  Inspired by this base-dependent ACR, we tried to apply this methodology to a more general substrate. It was expected that controlling an amide enolate conformation's equilibrium through various enolate moieties, including metal cations and ligands, would mitigate the undesired side reaction in accelerated ACR processes. Based on this hypothesis, a simple ACR substrate was designed to screen bases.

Results
To prove the efficiency of the cation-dependent ACR, large amounts of substrate were pursued in a short time. In addition, cyclic ACR substrate was favored over acyclic because the ACR of cyclic substrates proceeds with fewer entropic/geometric issues [12]. Commercially available methyl pipecolinate 1 was subjected to the sequential Schotten-Baumann amidation [14] to create amide 2, which was converted to aldehyde 3 via diisobuylaluminum hydride (DIBAL) reduction. Finally, Wittig methylenation afforded allyl amide 4 an excellent yield, as depicted in Scheme 1. ACR proceeds through deprotonation, [3,3]-sigmatropic rearrangement, and protonation, as shown in Figure 2. For this process, both a strong base for deprotonation of α-hydrogen in an amide group and a high temperature for thermal rearrangement are required [11,12]. However, this reaction condition also permits sigma-bond rotation in amide enolate to hamper generation of the desired conformation. If the undesired conformation exists as a major form, ACR would be impossible (low conversion), and an ensuing side reaction would occur (side product). Because the oxygen-metal bond may play an important role in this enolate conformation, it can be presumed that the cation of a base determines a successful ACR process. Actually, it was shown that lithium hexamethyldisilazane (LHMDS) and iPrMgCl gave different results in the ACR of an API synthesis [13]. Inspired by this base-dependent ACR, we tried to apply this methodology to a more general substrate. It was expected that controlling an amide enolate conformation's equilibrium through various enolate moieties, including metal cations and ligands, would mitigate the undesired side reaction in accelerated ACR processes. Based on this hypothesis, a simple ACR substrate was designed to screen bases.

Results
To prove the efficiency of the cation-dependent ACR, large amounts of substrate were pursued in a short time. In addition, cyclic ACR substrate was favored over acyclic because the ACR of cyclic substrates proceeds with fewer entropic/geometric issues [12]. Commercially available methyl pipecolinate 1 was subjected to the sequential Schotten-Baumann amidation [14] to create amide 2, which was converted to aldehyde 3 via diisobuylaluminum hydride (DIBAL) reduction. Finally, Wittig methylenation afforded allyl amide 4 an excellent yield, as depicted in Scheme 1. Various conditions were screened to effectively convert allyl amide 4 to the ring-expanded lactam 5, as shown in Table 1. As expected, bulky Grignard reagents, such as iPrMgCl or tBuMgBr, exhibited greater efficiency than LHMDS, the standard base for ACR [15,16] (entries 1, 2, 4). In contrast, sp 2 -hybridized carbanion, such as 2-mesitylMgBr, produced a lower yield than LHMDS [17] (entry 3). Relatively small Grignard reagents were also not effective, likely because of the nucleophilic substitution of Grignard reagents. Although sodium and potassium cations are larger than lithium cations, employing NaHMDS or KHMDS as a base also generated worse results than LHMDS (entries 7,8): in these cases, rather than the ACR adduct 5, an unidentified side product was obtained soon after the base was added. By contrast, iPrMgCl improved the ACR process, but other reaction conditions, such as the solvent and temperature, needed to be further optimized. The screening of representative solvents (benzene, tetrahydrofuran (THF), n-decane) is depicted in entries 9-11. The use of benzene improved results slightly; however, polar or extremely nonpolar solvents did not. Lastly, an ACR carried out at room temperature with benzene yielded no reaction (entry 12). Replacing iPrMgCl with iPrMgBr showed similar results, as shown in entry 13. Various conditions were screened to effectively convert allyl amide 4 to the ring-expanded lactam 5, as shown in Table 1. As expected, bulky Grignard reagents, such as iPrMgCl or tBuMgBr, exhibited greater efficiency than LHMDS, the standard base for ACR [15,16] (entries 1, 2, 4). In contrast, sp 2 -hybridized carbanion, such as 2-mesitylMgBr, produced a lower yield than LHMDS [17] (entry 3). Relatively small Grignard reagents were also not effective, likely because of the nucleophilic substitution of Grignard reagents. Although sodium and potassium cations are larger than lithium cations, employing NaHMDS or KHMDS as a base also generated worse results than LHMDS (entries 7,8): in these cases, rather than the ACR adduct 5, an unidentified side product was obtained soon after the base was added. By contrast, iPrMgCl improved the ACR process, but other reaction conditions, such as the solvent and temperature, needed to be further optimized. The screening of representative solvents (benzene, tetrahydrofuran (THF), n-decane) is depicted in entries 9-11. The use of benzene improved results slightly; however, polar or extremely nonpolar solvents did not. Lastly, an ACR carried out at room temperature with benzene yielded no reaction (entry 12). Replacing iPrMgCl with iPrMgBr showed similar results, as shown in entry 13. The optimized ACR conditions were applied to various substrates, as shown in Scheme 2. The reduction of known ester 6 [18] provided primary alcohol 7, which was protected by a tertbutyldiphenylsilyl (TBDPS) group. The silyl ether 8 was treated with trifluoroacetic acid (TFA) to generate amine salt, which was transformed uneventfully into butyryl amide 9. To verify the bulkiness of allyl side chain 9, TBDPS was converted into tert-butyldimethylsilyl (TBS) or triethylsilyl (TES) through tetra-n-butylammonium fluoride (TBAF)/AcOH deprotection followed by a corresponding silyl protection. This straightforward synthesis enabled TBS or TES to substitute for ACR substrates 11 and 12 in an excellent yield. Amide substitution was also attempted. After acidic deprotection of the tert-butyloxycarbonyl (Boc) group of carbamates 8, as described above, a concomitant acylation with alkynyl or aromatic functionality was also executed, employing N-ethyl-N′-(3-dimethylaminopropyl)carbodiimide hydrochloride(EDCI)/4-dimethylaminopyridine (DMAP) amidation conditions. After two uneventful procedures, desired substrates 13 and 14 were produced. iPrMgCl benzene c no reaction 13 iPrMgCl benzene 81 a Isolation yield; b Unidentified side product was obtained in 1 min.; c Reaction was carried out at the room temperature.
The optimized ACR conditions were applied to various substrates, as shown in Scheme 2. The reduction of known ester 6 [18] provided primary alcohol 7, which was protected by a tert-butyldiphenylsilyl (TBDPS) group. The silyl ether 8 was treated with trifluoroacetic acid (TFA) to generate amine salt, which was transformed uneventfully into butyryl amide 9. To verify the bulkiness of allyl side chain 9, TBDPS was converted into tert-butyldimethylsilyl (TBS) or triethylsilyl (TES) through tetra-n-butylammonium fluoride (TBAF)/AcOH deprotection followed by a corresponding silyl protection. This straightforward synthesis enabled TBS or TES to substitute for ACR substrates 11 and 12 in an excellent yield. Amide substitution was also attempted. After acidic deprotection of the tert-butyloxycarbonyl (Boc) group of carbamates 8, as described above, a concomitant acylation with alkynyl or aromatic functionality was also executed, employing N-ethyl-N -(3-dimethylaminopropyl)carbodiimide hydrochloride(EDCI)/4-dimethylaminopyridine (DMAP) amidation conditions. After two uneventful procedures, desired substrates 13 and 14 were produced. Amide substituent derivatization was also carried out, as shown in Scheme 3. Williamson etherification of the primary alcohol 7, followed by sequential Boc deprotection/amidation, afforded differently substituted amides 16-18 in a straightforward manner. These three derivatives were expected to prove the substituent effect of amide moiety. Amide substituent derivatization was also carried out, as shown in Scheme 3. Williamson etherification of the primary alcohol 7, followed by sequential Boc deprotection/amidation, afforded differently substituted amides 16-18 in a straightforward manner. These three derivatives were expected to prove the substituent effect of amide moiety. Amide substituent derivatization was also carried out, as shown in Scheme 3. Williamson etherification of the primary alcohol 7, followed by sequential Boc deprotection/amidation, afforded differently substituted amides 16-18 in a straightforward manner. These three derivatives were expected to prove the substituent effect of amide moiety. Finally, acyclic substrate 21 was prepared from p-methoxybenzyl amines 19 through an amidation/crotylation sequence. Thus, with the various substrates in hand, iPrMgCl mediated ACR (Scheme 4). The ACR of various substrates is summarized in Scheme 5. As expected, the employment of iPrMgCl/benzene converted each of the substrates into the requisite lactams/amides more efficiently than in conventional conditions (LHMDS/toluene). Of note, superior results occurred in some cases, such as sterically demanded substrates 13 and 14, as well as labile TES-protected substrate 12. ACR under the treatment of acyclic amide 21 with iPrMgCl also yielded impressive results. Moreover, the LHMDS base did not produce the desired free amide 21' at all. In fact, a brief comparison of the alkene-substituted ACR substrate showed that the larger the substituent of the allylic side chain, the worse the product yield. Based on this tendency, the increased bulkiness of the side chain could cause the desired ACR process to occur. Similarly, the results from amide analogs 16-18 also supported this hypothesis. Consequently, increased repulsion of the allylic side chain with amide enolate is thought to be responsible for the low yield of ACR, but this obstacle can be overcome by employing iPrMgCl as a base. To verify the structure of macrolactams, ACR product 14' was crystallized. Figure 3 represents the structure of 14', which features the 1,2-anti-configured side chain and the 10-membered lactam framework. The crystal structure below shows that ACR proceeded as designed. The ACR of various substrates is summarized in Scheme 5. As expected, the employment of iPrMgCl/benzene converted each of the substrates into the requisite lactams/amides more efficiently than in conventional conditions (LHMDS/toluene). Of note, superior results occurred in some cases, such as sterically demanded substrates 13 and 14, as well as labile TES-protected substrate 12. ACR under the treatment of acyclic amide 21 with iPrMgCl also yielded impressive results. Moreover, the LHMDS base did not produce the desired free amide 21' at all. In fact, a brief comparison of the alkene-substituted ACR substrate showed that the larger the substituent of the allylic side chain, the worse the product yield. Based on this tendency, the increased bulkiness of the side chain could cause the desired ACR process to occur. Similarly, the results from amide analogs 16-18 also supported this hypothesis. Consequently, increased repulsion of the allylic side chain with amide enolate is thought to be responsible for the low yield of ACR, but this obstacle can be overcome by employing iPrMgCl as a base. The ACR of various substrates is summarized in Scheme 5. As expected, the employment of iPrMgCl/benzene converted each of the substrates into the requisite lactams/amides more efficiently than in conventional conditions (LHMDS/toluene). Of note, superior results occurred in some cases, such as sterically demanded substrates 13 and 14, as well as labile TES-protected substrate 12. ACR under the treatment of acyclic amide 21 with iPrMgCl also yielded impressive results. Moreover, the LHMDS base did not produce the desired free amide 21' at all. In fact, a brief comparison of the alkene-substituted ACR substrate showed that the larger the substituent of the allylic side chain, the worse the product yield. Based on this tendency, the increased bulkiness of the side chain could cause the desired ACR process to occur. Similarly, the results from amide analogs 16-18 also supported this hypothesis. Consequently, increased repulsion of the allylic side chain with amide enolate is thought to be responsible for the low yield of ACR, but this obstacle can be overcome by employing iPrMgCl as a base. To verify the structure of macrolactams, ACR product 14' was crystallized. Figure 3 represents the structure of 14', which features the 1,2-anti-configured side chain and the 10-membered lactam framework. The crystal structure below shows that ACR proceeded as designed. To verify the structure of macrolactams, ACR product 14' was crystallized. Figure 3 represents the structure of 14', which features the 1,2-anti-configured side chain and the 10-membered lactam framework. The crystal structure below shows that ACR proceeded as designed. Molecules 2019, 24, x 6 of 17    In the case of lithium enolate ( Figure 5), MD results clearly showed that the desired Li-conformer 17-A is very unstable: it did not stay in one state for long ( Figure 5A and 5B). In contrast, the conformation of 17-B was stable, maintaining the C-C-C-O dihedral angle at 60° ( Figure 5A and 5C).    In the case of lithium enolate ( Figure 5), MD results clearly showed that the desired Li-conformer 17-A is very unstable: it did not stay in one state for long ( Figure 5A and 5B). In contrast, the conformation of 17-B was stable, maintaining the C-C-C-O dihedral angle at 60° (Figure 5A and 5C).  However, in the case of magnesium enolate from amide 17, the structural stability of both conformations showed similar fluctuation patterns ( Figure 6A). The most common dihedral angles (i.e., conformation) of 16-C were 110 • ( Figure 6A,B), while those of 17-D were -110 • (Figure 6A,C). The instability of undesired conformer 17-D would explain the rapid conversion of desired conformer 17-C in the corresponding ACR product.
However, in the case of magnesium enolate from amide 17, the structural stability of both conformations showed similar fluctuation patterns ( Figure 6A). The most common dihedral angles (i.e., conformation) of 16-C were 110° ( Figure 6A and 6B), while those of 17-D were -110° ( Figure 6A and 6C). The instability of undesired conformer 17-D would explain the rapid conversion of desired conformer 17-C in the corresponding ACR product.

General Information
Unless noted otherwise, all starting materials and reagents were obtained from commercial suppliers and were used without further purification. Tetrahydrofuran was distilled from sodium benzophenone ketyl. Dichloromethane was freshly distilled from calcium hydride. All solvents used for routine isolation of products and chromatography were reagent grade. Air-and moisturesensitive reactions were performed under argon atmosphere. Flash column chromatography was performed using silica gel 60 (230-400 mesh, Merck, Darmstadt, Germany) with the indicated solvents. Thin-layer chromatography was performed using 0.25 mm silica gel plates (Merck). 1 H-NMR data were reported in the order of chemical shift, multiplicity (s, singlet; d, doublet; t, triplet; q, quartet; m, multiplet and/or multiple resonance), number of protons, and coupling constant in hertz (Hz).

General Information
Unless noted otherwise, all starting materials and reagents were obtained from commercial suppliers and were used without further purification. Tetrahydrofuran was distilled from sodium benzophenone ketyl. Dichloromethane was freshly distilled from calcium hydride. All solvents used for routine isolation of products and chromatography were reagent grade. Air-and moisture-sensitive reactions were performed under argon atmosphere. Flash column chromatography was performed using silica gel 60 (230-400 mesh, Merck, Darmstadt, Germany) with the indicated solvents. Thin-layer chromatography was performed using 0.25 mm silica gel plates (Merck). 1 H-NMR data were reported in the order of chemical shift, multiplicity (s, singlet; d, doublet; t, triplet; q, quartet; m, multiplet and/or multiple resonance), number of protons, and coupling constant in hertz ( Hz).

Conclusions
iPrMgCl is an advanced base of ACR. In addition, it is superior to the classic LHMDS/toluene condition when the sterically demanded substrate is subjected to ACR. The improvement of the aspect of reactions is derived from the relative stability of the desired conformation for ACR. MD simulations were carried out to investigate the structural stability of the reaction intermediates. The results of C-C-C-O dihedral angle analysis clearly showed that the structural stability of the intermediates was highly related to the portion of the product in the reaction. More detailed mechanistic investigations using other substrates, as well as their applications to the synthesis of bioactive molecules or natural products, will be discussed in due course.