Controlling Regioselectivity in the Enantioselective N-Alkylation of Indole Analogs Catalyzed by Dinuclear Zinc-ProPhenol

Abstract

The enantioselective N-alkylation of indole and its derivatives with aldimines is efficiently catalyzed by a zinc-ProPhenol dinuclear complex under mild conditions to afford N-alkylated indole derivatives in good yield (up to 86%) and excellent enantiomeric ratio (up to 99.5: 0.5 e.r.). This protocol tolerates a wide array of indoles, as well as pyrrole and carbazole, to afford the corresponding N-alkylation products. The reaction can be run on gram scale with reduced catalyst loading without impacting the efficiency. The chiral aminals were further elaborated into various chiral polyheterocyclic derivatives. The surprising stability of the chiral N-alkylation products will open new windows for asymmetric catalysis and medicinal chemistry.

COMMUNICATION

Acetals, hemiaminals and aminals share a structure possessing a carbon bearing two heteroatoms prone to undergo ionization as shown in eq 1.¹

(eq. 1)

While they are broadly present both in synthetic and natural product chemistry, they are commonly constrained in ring systems to enhance their intrinsic stability.²

An exciting recent illustration of this phenomenon is the development of a potentially sensitive hemiaminal, elbasvir 1 (see figure 1), which is a highly potent and selective NS5A inhibitor of the hepatitis C virus. The potential sensitivity of this structural unit is mitigated by its incorporation in a ring. Indeed, its stability is such that the synthetic intermediate 2 can be synthesized asymmetrically using a novel C-N coupling.³ Indeed, this method extended to form the cyclic aminal 3 asymmetrically. Also there are many methods available for the synthesis of N,O-aminals and utilized for several organic transformations.¹² On the other hand, attempts to develop more traditional indole moieties containing asymmetric aminal syntheses as shown in eq. 2 have been less effective.⁴ These results may derive from the sensitivity of the aminal. In fact, a recent study in the stability of cyclic aminals noted and synthesized asymmetrically,⁵ “for the preparation and workup of these valuable intermediate and natural products, appropriate conditions have to be chosen and for application as drug molecules their sensitivity towards hydrolysis has to be taken into account.”

Figure 1.

Elbasvir and related hemiaminals and aminals

Non-acyclic aminals make the challenge even more daunting. Recent studies have shown that aminals 4–6 have significant antibiotic properties (Figure 2).^6a–b Indole and its derivatives have been extensively investigated as medicinally active molecules.⁷

Figure 2.

Non-cyclic aminals as antibiotics

Thus incorporating indoles into such aminals is made interesting. The only reports of asymmetric alkylation of indole are at C3 position.⁸ There are only two reports of the alkylation of the indole nitrogen to form the acyclic aminal, but none asymmetrically.⁹ Antilla et. al. reported the synthesis of chiral aminals by addition of sulfonamides and imide to imines with excellent enantioselectivity (Eqn 4a).^6c–d Due to the lack of methods available for the asymmetric synthesis of N,N-acyclic aminals, these compounds have not been utilized to date. Indeed, the reported use of Zn-ProPhenol¹⁰ led only to asymmetric C3-alkylation using the N-Ts imine as the aminoalkylating agent (Scheme 1).^11a One explanation for this regioselectivity could be that attack at the indole N occurs reversibly, whereas attack at C3 was irreversible. Due to the increased synthetic versatility of carbamate-protected amines relative to sulfonamides, we were intrigued to see what type of behaviour would be observed with N-Boc and N-Cbz imines. We began our studies with imine 7a and indole 8a in the presence of our zinc-ProPhenol L1 catalyst in THF at room temperature for 12 h. To our surprise, the N-alkylation product 9a was obtained in 61% yield with 98.5:1.5 e.r., while the C3-alkylation product was observed in 11% yield with 77:33 e.r. Clearly, the carbamate-protected imines gave us access to the reaction pathway in which C3-alkylation is suppressed.

Scheme 1.

Enantioselective alkylation of Aldimines with indoles and sulfonamides

Given the rarity of chiral aminals in the literature, as well as their potential applications, we were excited to explore this reaction further. Hoping to improve the selectivity for the N-alkylation product, a solvent screen was performed. Relative to THF, all of the solvents examined gave inferior yield and enantioselectivity of the desired N-alkylation product (Table 1). In fact, all of the other solvents screened gave the C3-alkylation adduct as the major product. Other ProPhenol catalysts L2–L6 were also screened and the catalyst L1 produced the best yield and enantioselectivity of N-alkylation product (Table 1). While we were unable to completely suppress the C3-alkylation, the N-alkylated compounds were typically the major product in THF, were formed with good yield and enantioselectivity and were easily separable from the C3-alkylation product. With optimized conditions in hand, we sought to explore the scope of the reaction by first varying the indole (See the scheme 2).

Table 1.

Optimization of the reaction conditions^a,^b

entry	Solvent/ Ligand	N-alk. Yield	N-alk. e.r.	C3-alk. Yield	C3-alk. e.r.
1	PhMe/L1	15 %	81:19	60 %	57.5:42.5
2	DCM/L1	Trace	77:23	70 %	42:58
3	THF/L1	61 %	98.5:1.5	11%	77:33
4	PhH/L1	14 %	65:35	65 %	74:26
5	ACN/L1	12 %	90:10	70%	89.5:10.5
6	DME/L1	7%	96:4	42%	61.5:38.5
6	THF/L2	35 %	82:18 (−)	34 %	56:44(−)
7	THF/L3	Trace	70:30	14 %	68:32
8	THF/L4	17 %	98:2	54 %	85:15
9	THF/L5	35 %	98:2	14 %	64:36
10	THF/L6	37 %	96:4	30%	65.5:34.5

^aReaction conditions: 1 equiv of indole, 1.1 equiv of imine, 10 mol % of Zn-ProPhenol (Generated using 1M Et₂Zn in hexanes and ligand (2:1)) at rt in solvent (0.3 M) for 12 h.

^bRemaining starting material has been recovered.

Scheme 2.

Scope of the reaction with Boc imines. [a] <12% of C3-alkylation product obtained with low e.r. [b] Reaction conditions: 1 equiv of indole, 1.1 equiv of imine, 10 mol % of Zn-ProPhenol (Generated using 1M Et₂Zn in hexanes and ligand (2:1)), at rt, in THF (0.3 M) for 12 h.

When the indole C3 position was blocked, higher yields of the desired compounds were obtained. Skatole produced the N-alkylation product 9b in 77% yield with e.r. of 94.5:5.5. A heterocyclic imine such as 2-furyl substituted 7c gave the N-alkylation product 9c in 66% yield with 92.5 : 7.5 e.r. (Scheme 2). To determine the effect of steric crowding of the nucleophilic site, the reaction was screened with 2-substituted indoles. With an alkyne at the 2-position, the N-alkylated indole derivative 9d was formed exclusively with excellent enantioselectivity (95:5 e.r.). Higher temperatures and longer reaction times did not improve the yield. Similarly, methyl indole-2-carboxylate also produced exclusively the N-alkylation product 9e in 94:6 e.r. The scope of the reaction further extended to the other imines, such as phenyl imine 7g and 3-methylphenyl imine 7h; the desired products were obtained in 61% yield with 97:3 e.r. and 64% with 95:5 e.r., respectively.

To further expand the substrate scope and to improve the yield of the reaction, Cbz imines were also examined. Due to decreased steric demands of the Cbz group relative to Boc, we anticipated that Cbz imines would be more reactive. Indeed, the reaction proceeded even at 4 °C, which allowed improvement in a number of examples. With methyl indole-2-carboxylate, the Cbz imine 10i was quite differential and gave the desired product 11i in 71% yield with 97.5:2.5 e.r., compared to 28% yield and 94:6 e.r. for the Boc N-alkylation product 9e. This supports our hypothesis that the increased yield of these substrates is due to the decreased steric demand of the Cbz imines. Since indole-2-carboxylates are useful for the synthesis of polycyclic molecules, the reaction was also carried out with other imines, such as 3-methylphenyl imine 10j, which gave the N-alkylation product 11j in 55% yield (BRSM 87%) with excellent enantiomeric ratio (99:1). Skatole formed the N-alkylation product 11b in good yield (86%) with 90.5:9.5 e.r. The 5-methoxyindole produced the desired N-alkylation product 11c in 62% yield with 96:4 e.r. To explore the peri effect, the reaction was carried out with 7-methylindole. To our delight, the methyl group did not negatively affect the reaction, and the N-alkylation product 11d was obtained in 63% yield with 98:2 e.r. even at 4 °C. The scope of the reaction was further analyzed with halogen substituted indoles. With 5-bromoindole, the corresponding N-alkylation product 11e was obtained in 61% yield with 70:30 e.r. To improve these results, the reaction was carried out at 4 °C, which increased the selectivity up to e.r 90:10 with no loss of yield. The reaction tolerated a carbamate NH, which was not competitive with the indole nitrogen; the indole N-alkylation product 11f was obtained in 85:15 e.r. at room temperature. Pleasingly, the enantioselectivity increased to 96:4 e.r. by reducing the temperature to 4 °C. The reaction even tolerated a nitrile-bearing indole, and the desired product 11g formed in 81% yield with 91:9 e.r. This observation is noteworthy, as the Lewis basic nitrile could have acted as a poison for the Zn-ProPhenol catalyst. Due to the increased reactivity of imines relative to aldehydes in ProPhenol reactions, the reaction could even be applied to aldehyde-substituted indoles. With 4-indole carboxaldehyde, the N-alkylation product 11h was obtained in 56% yield with 90:10 e.r (Scheme 3).

Scheme 3.

Scope of the reaction with Boc imines [a] less than 10% of C3-alkylation product obtained with low enantioselectivity. [b] Reaction conditions: 1 equiv of indole, 1.1 equiv of imine, 10 mol % of Zn-ProPhenol (Generated using 1M Et₂Zn in hexanes and ligand (2:1)), at rt, in THF (0.3 M) for 12 h. [c] L4 ProPhenol ligand used (optimization table given in SI). [d] 40% of C3 alkylation product formed.

The scope of the reaction further extended to the 3-methyphenyl imine with indole, where the product 11k was obtained in 58% yield with 91.5:8.5 e.r. Cyclopropyl imine were also tolerated, and the corresponding N-alkylated indole product 11l was formed in 59% yield with 70:30 e.r. Because of the lower enantioselectivity, the reaction was screened with other ProPhenol ligands (see Supporting Information for details). To our delight, the electron rich 4-methoxyphenyl derived ProPhenol (L4) produced the product with a greatly enhanced enantomeric ratio of 95:5 with 59% yield. A sterically crowded imine bearing an alkene substituent at the 2-position was also a viable substrate, and the N-alkylation product 11m was obtained in 37% yield with 98:2 e.r. An imine bearing a 2-thiophene group (10n) gave the N-alkylation product 11n in 62% yield with excellent enantioselectivity (99.5: 0.5 e.r.). The 4-methoxyphenyl substituted imine 10o with indole produced the N-alkylation product in 71% yield with 96:4 e.r. (Scheme 3).

To demonstrate the scalability and practicality of this method, we performed a gram scale reaction using 8a and 1.05 equivalent of imine 10a. Notably, we could reduce the catalyst loading from 10 mol % to 2 mol % without impacting the yield or enantioselectivity of the reaction (Scheme 4).¹³

Scheme 4.

Gram Scale reaction¹⁴

Since indole derivatives were good substrates for this transformation, we sought to determine whether the reaction extended to other nitrogen-containing heterocycles. Pleasingly, methyl pyrrole-2-carboxylate with imine 10a produced the N-alkylation product 13 in 52% yield with 99.5:0.5 e.r (eq. 6). Carbazole with imine 10a solely produced N-alkylation product 15a in good yield with 98.5:1.5 e.r. at 4 °C. Similarly, 2-furyl imine 7c and 2-thienyl imine 10n both reacted with carbazole to give N-alkylation products 15b and 15c in 73 and 81% yield, respectively, with excellent enantioselectivity (eq. 7). As with indoles, a 2-substituted imine also reacted cleanly with carbazole, giving 16 in 82% yield and 96: 4 e.r (eq. 8, Scheme 5).

Scheme 5.

Reaction with imines with other nucleophiles¹⁴

The N-alkylated products that are produced efficiently using our method provide an ideal platform to construct new classes of valuable chiral heterocycles (Scheme 6). For example, the compound (9e) was cyclized onto the pendant ester with AlEt₃ to form a 5-membered ring lactam 17 in 82% yield with no erosion of enantiopurity (eq. 9). To further utilize the unique grouping of functionality in these molecules, a gold-catalyzed 6-endo-dig cyclization was performed on amino alkyne 9d to afford tetrahydropyrimidine 18 in 76% yield (eq. 10).

Scheme 6.

Synthetic applications

Furthermore, treatment of 16 with Cs₂CO₃ induced an aza-Michael addition to give rise to carbazole-substituted isoindoline derivative 19 in 52% isolated yield with no erosion of enantioselectivity (eq. 11). The structure of this molecule 19 was unambiguously determined through single crystal XRD analysis. The relative and absolute configurations were determined as 1S, 3R from the crystal structure of 19. By analogy, the same configuration was assigned to all the aminals reported herein (see the crystal structure 19). It should be noted that the assignment of S as the configuration of the aminal stereocenter in 19 corresponds to an assignment of R in the cyclization precursor. Thus, R is assigned to all the initial aminals.^11b These cyclizations are noteworthy, as they provide the first synthetic route to chiral 1,3-disubstituted isoindoline derivatives.

In conclusion, we have developed the first enantioselective synthesis of N-alkylated indoles, which proceeds with good yields with excellent enantioselectivities. Particularly noteworthy is the change in enantioselectivity with Zn-ProPhenol catalysts in going from the N-sulfonylimines (favoring C3) to the N-alkoxyacylimines. Scheme 7 offers a possible explanation.

Scheme 7.

Proposed catalytic cycle for the N-alkylation

N-Sulfonylimines should bind via one-point coordination of the imine nitrogen to zinc. Thus, the complex is rather flexible, which allows attack from both the nitrogen and C3 of the indole nucleophile. Moreover, since the attack at the nitrogen is presumably more reversible than the reaction at C3, C3-alkylation became the majority driven by thermodynamics. On the other hand, the N-alkoxyacyl imine undergoes two-point binding (Scheme 7). This more structurally rigid complex brings the nitrogen of the heterocycle and the carbon of imine to closer proximity, facilitating the observed asymmetric process. Exchange of the alkoxyacylimine with the bound product then liberates the adduct and initiates another cycle. This model predicts formation of the R enantiomer of the product as observed.

This methodology extended to pyrrole and carbazole derivatives, which afforded similarly desirable results. These chiral aminals are synthetically useful motifs that could be utilized for stereoselective lactam synthesis, and could also be transformed into a tetrahydropyrimidine. We also used these aminals—via an aza-Michael addition—to obtain the first reported examples of chiral 1,3-disubstituted isoindoline derivatives. It is noteworthy that all the transformations of the aminals occurred with no observable loss of e.r. Importantly this method for the first time opens the door for the utilization of the indole chiral N-alkylation products for further organic transformations and possible applications, notably for biological purposes as some aminals have already demonstrated antibiotic activity.