Design and synthesis of 3,3’-triazolyl biisoquinoline N,N’-dioxides via Hiyama cross-coupling of 4-trimethylsilyl-1,2,3-triazoles

Abstract

A new strategy to effectively lock the conformation of substituents at the 3,3’-positions of axial-chiral biisoquinoline N,N’-dioxides was developed based on the strong dipole–dipole interaction between 1,2,3-triazole and pyridine N-oxide rings. The crystal structure and the DFT calculations of 3,3’-bis(1-benzyl-1H-1,2,3-triazole-4-yl)-1,1’-biisoquinoline N,N’-dioxide (3a) provided strong support for this strategy. Furthermore, we successfully demonstrated that readily available 4-trimethylsilyl-1,2,3-triazoles are viable nucleophiles for Hiyama cross-coupling.

Graphical Abstract

Introduction

1,1’-Binaphth-2-ol (BINOL) is among the most successful chiral scaffolds ever developed.¹ Its strength comes from the substituents at its 3,3’-positions that have proven particularly effective to increase asymmetry around the reaction site (i.e., space proximal to its hydroxyls). In much the same way, axial-chiral 3,3’-functionalized biisoquinoline (or bipyridine) N,N’-dioxides (1) have emerged as an important class of chiral Lewis base catalysts for the activation of chlorosilanes (e.g., SiCl₄).² Their 3,3’-substituents can be readily extended by adding a phenyl linker,³ but such a modification leads to a widemouthed cavity,^3a projecting 3,3’-substituents away from the reaction space (2). Therefore, as part of our longstanding interests in the development of new chiral Lewis bases,^2b,4 we conceived of biisoquinolines bearing a deep chiral cavity (3). We envisioned that we could control the conformation of 3,3’-substituents by taking advantage of large dipole moments of a pyridine N-oxide ring and a 1,2,3-triazole ring (4.24 D and 4.38 D, respectively).^5,6 More specifically, the anti-conformation with respect to the directions of two dipole moments shown in Figure 1b is expected to be strongly favored over the syn-conformation.⁶ Herein, we report the synthesis of 3,3’-triazolyl biisoquinoline N,N’-dioxides via Hiyama cross-coupling with readily available 1-substituted-4-trimethylsilyl-1H-1,2,3-triazoles, the crystal structure of a triazolyl biisoquinoline, and the computational conformational analysis of itself and its SiCl₄ complex.

Figure 1.

The design of new Lewis bases based on dipole–dipole interaction.

A sequence of Sonogashira coupling of 3,3’-diiodo-1,1’-biisoquinoline N,N’-dioxide (4, Scheme 1) followed by the Cu(I)-catalyzed azide-alkyne cycloaddition (CuAAC)^7,8 appeared a most reliable approach since we previously developed a method to halogenate the 3,3’-positions of biisoquinoline N,N’-dioxide.^2b However, Sonogashira coupling of (S)-3,3’-diiodo-1,1’-biisoquinoline N,N’-dioxide and trimethylsilylacetylene turned out too elusive to move forward in our hands (see Supporting Information (SI) for details). As such, we considered a cross-coupling reaction of (S)-3,3’-dihalo-1,1’-biisoquinoline N,N’-dioxides and 4-metallo-1-substituted-1,2,3-triazoles. The Pd- or Cu-catalyzed direct arylations of 1-substituted-1,2,3-triazoles (i.e., 4, 5-unsubstituted) are exclusively C5-selective.⁹ 4-Boronyl triazoles¹⁰ remain an elusive substrate for Suzuki-Miyaura cross-coupling just like 2-pyridyl boronates (heterocyclic boronic acids and esters in which boron is attached to a carbon center adjacent to the ring heteroatom are often unstable due to their susceptibility to hydrolytic protodeboronation).¹¹ Although 4-stannyl triazoles¹² were shown to undergo Stille cross-coupling, only thermal azide-alkyne cycloaddition reactions are known for commercially available tributylstannylacetylene that was reported unreactive for some azides.^12f Negishi cross-coupling is reported only for 1,5-disubstituted-4-zincio-1,2,3-triazoles and 1-substituted-5-zincio-1,2,3-triazoles.¹³ Kumada cross-coupling of 4-magnesio-1,2,3-triazoles,¹⁴ and Hiyama cross-coupling of 4-silyl-1,2,3-triazoles¹⁵ are not known to the best of our knowledge. Underdevelopment of the methods to prepare 1,4-disubstituted-1,2,3-triazoles via C4 arylation is presumably because this motif can be readily prepared by CuAAC from the corresponding azides and acetylenes. In this context, we opted to investigate the feasibility of the corresponding Hiyama cross-coupling with readily available 1-substutited-4-trimethylsilyl-1H-1,2,3-triazoles (6) because of the clear benefits of organo(trialkyl)silanes that include low toxicity, high stability, good solubility and easy handling.¹⁶ It should be mentioned that in contrast to heteroatom substituted silanes, aryl(trialkyl)silanes have drawn little attention as cross-coupling nucleophiles and have remained a long-standing important challenge.¹⁶

Scheme 1.

A cross-coupling approach.

We began our investigation based on the Hiyama cross-coupling reactions of 2-trimethylsilylpyridines reported independently by the groups of Gros, Whittaker, and Marples.¹⁷ We chose 1-benzyl-4-trimethylsilyl-1H-1,2,3-triazole (6a) and (S)-4 or 5 as our model substrates and tested them with Pd(PPh₃)₄ in the presence of TBAF·3H₂O and Ag₂O in THF. In contrast to those previous reports, neither desired 3a nor the corresponding mono-coupled by-product formed in isolable quantities (Table 1, entries 1 & 2). Therefore, we screened several commercially available Pd catalysts and found that PdCl₂(dppe), PdCl₂(dppp), and PdCl₂(BINAP) provided the desired product (3a) in 12%, 15% and 7% yields, respectively along with trace amount of the mono-coupled by-product (entries, 4, 5 & 8). We then evaluated different solvents, fluoride sources using PdCl₂(dppp) and Ag₂O, and in turn, metal additives with PdCl₂(dppp) and TBAF·3H₂O, and found that THF, TBAF·3H₂O and Ag₂O were best out of all tested (see SI for details). We next probed the reaction temperatures and catalyst loading, and found that 40 °C and 40 mol % catalyst loading were optimum (entry 12) which gave 3a in 41% yield. Since our main objective of this study was to develop a preparative method for catalyst 3, we proceeded to test the model substrates on a 1 mmol scale and obtained the desired product in 46% yield with only trace amount of the corresponding mono-coupled product. As we used 40 mol % of a Pd catalyst, 46% yield corresponds to the catalyst turnover of 2.3 times (two coupling sites in 5). We evaluated three more triazoles on a 1 mmol scale. 1-Mesityl-4-trimethylsilyl-triazole (6b) worked in a similar efficiency (54% yield). 1-Benzhydryl-4-trimethylsilyl-triazole (6c) provided the desired product in 31% yield with trace amount of the mono-coupled product and 1-(1-adamantyl)-4-trimethylsilyl-triazole (6d) afforded the desired product in 23% and the mono-coupled product in 19%. These results represent the first example of the Hiyama cross-coupling of 4-trimethylsilyl-1,2,3-triazoles, and demonstrated that the transformation was feasible. It should be mentioned that 5-unsubstituted 1,2,3-triazoles have proven useful as hydrogen bond donors and have been incorporated into organocatalysts, anion-binding receptors, etc.^7,18,19 These molecules were made by a sequence of Sonogashira and CuAAC reactions. Since Sonogashira reactions are often limited to aryl iodides²⁰ that are not always readily available, the present method could possibly complement the currently adapted synthetic sequence as demonstrated herein (vide supra).

Table 1.

Hiyama cross-coupling of 4-TMS-triazoles.

Entry	ArX	6	Catalyst	mol %	Temp (°C)	Yield (%)a
1	4	6a	Pd(PPh3)4	20	rt	3a, trace
2	5	6a	Pd(PPh3)4	20	rt	3a, trace
3	5	6a	Pd2(dba)3/P(o-tol)3	20	rt	3a, 0
4	5	6a	PdCl2(dppe)	20	rt	3a, 12
5	5	6a	PdCl2(dppp)	20	rt	3a, 15
6	5	6a	PdCl2(dppb)	20	rt	3a, trace
7	5	6a	PdCl2(dppf)	20	rt	3a, 0
8	5	6a	PdCl2(BINAP)	20	rt	3a, 7
9	5	6a	Pd2(dba)3/Xantphos	20	rt	3a, trace
10	5	6a	PdCl2(dppp)	20	60	3a, 24
11b	5	6a	PdCl2(dppp)	40	60	3a, 42
12b	5	6a	PdCl2(dppp)	40	40	3a, 41
13b,c	5	6a	PdCl2(dppp)	40	40	38,46d
14 b,c	5	6b	PdCl2(dppp)	40	40	3b, 54d
15b,c	5	6c	PdCl2(dppp)	40	40	3c, 31d
16b,c	5	6d	PdCl2(dppp)	40	40	3d, 23d

0.1 mmol of 4 or 5 was used for entries 1 – 12. 6a(R = Bn), 6b (R = Mes), 6c (R = Bzh) & 6d (R = 1-ad).

^a1H NMR yield estimated with 1,1,2,2-tetrachloroethane as an internal standard unless otherwise noted.

^b4.8 equiv of 6, TBAF·3H₂O, and Ag₂O.

^c1 mmol scale reaction.

^dIsolated yield after flash chromatography on silica gel.

We were able to obtain crystals of 3a suitable for the X-ray structural analysis (Figure 2). In the solid state, 3a was found to adapt the anti-conformation in accordance with our hypothesis based on the dipole–dipole interaction (Figure 1). The dihedral angle between its triazole and pyridine N-oxide rings is 7.66 °. At least in the solid state, the 3,3’-substituents effectively embrace the reaction space, forming a deep chiral cavity. We also performed a computational conformational analysis of 3a using density functional theory. Calculations have been performed with the PBEh-3c method and the C-PCM solvation model with the dielectric constant of dichloromethane (DCM). The calculated most stable conformation of 3a turned out almost identical to its crystal structure (see Figure S1 in SI). The anti-conformation was found to be favored by 8.6 kcal/mol over the syn-conformation and the calculated average dihedral angle between its triazole and pyridine N-oxide rings is 7.75 °, providing further support for our hypothesis that the dipole-dipole interaction effectively locks the conformation of 3,3’-substituents. This energy difference roughly corresponds to the equilibrium ratio of 1,000,000:1 (= anti:syn).

Figure 2.

The crystal structure of 3a.

The binding geometry of 3a to SiCl₄ was computationally investigated (Figure 3). To our delight, 3a was found to bind to SiCl₄ through its two oxygen atoms with the binding energy of 30.8 kcal/mol in contrast to some 1,2,3-triazole bearing ligands that are known to coordinate to a metal via its triazole nitrogen atoms.^18f,21 Furthermore, the dipole–dipole interaction between the triazole and pyridine N-oxide rings appeared to dominate their relative conformation even after 3a complexed with SiCl₄ although those two rings slightly twisted out of the co-planarity (the dihedral angle in the complex is = 20.57 °). Furthermore, the anion–π-type interaction²² between chlorine atoms of a hypervalent chlorosilane and phenyl rings of the benzyl units was found in 3a–SiCl₄ complex, which brought the benzyl groups to embrace the reaction space, leading to a structurally well-defined, deep chiral pocket. It should be mentioned that a pileup of electron density occurs at the peripheral chlorine atoms of a hypervalent silicon complex of this kind.^{2b,23, 24}

Figure 3.

The most stable conformation and potential of 3a-SiCl₄ complex calculated with PBEh-3c/C-PCM(DCM).

Next, we conducted preliminary tests to probe the catalytic performance of 3a–d in comparison to a conventional Lewis base catalyst (2a)^2b,25 (Scheme 2). The asymmetric transfer hydrogenation of N-aryl ketimines with trichlorosilane is a proven testing ground for new chiral Lewis bases (eq. 1).²⁶ To our delight, 3a–d were found more enantioselective than 2a (44–56% ee vs. 24% ee). Furthermore, 3a was substantially more reactive than the others (71% yield). As 3a–d provided promising reactivity and enantioselectivity for the testing ground reaction, we preliminarily evaluated them for the direct catalytic asymmetric synthesis of α-chiral primary amine that is among the long-standing important challenges in organic synthesis (eq. 2).²⁷ Catalysts 3a, c and d enantioselectively catalyzed the reduction of amine salt 9 albeit with low yields while 2a and 3b were ineffective. Catalyst 3a was found distinctively more enantioselective than 3c or d. It should be mentioned that the Lewis base-catalyzed asymmetric transfer hydrogenation of ketimines with trichlorosilane currently remains limited to N-aryl and alkyl protected ones.²⁶ These preliminary observations clearly demonstrated that 3,3’-triazolyl biisoquinoline N,N’-dioxides are complementary to existing Lewis base catalysts, and bode well for the development of their applications.

Scheme 2.

Preliminary evaluation of catalysts (reaction conditions were not fully optimized). Yields refer to ¹H NMR yield determined with 1,1,2,2-tetrachloroethane as an internal standard (see SI for details).

In summary, we developed a new strategy to effectively control the conformation of substituents at the 3,3’-positons of axial-chiral biisoquinoline (or bipyridine) N,N’-dioxides on the basis of the strong dipole–dipole interaction between 1,2,3-triazole and pyridine N-oxide rings. The X-ray structure and the DFT calculations of 3a provided strong support for this strategy. Furthermore, we successfully demonstrated that readily available 4-trimethylsilyl-1,2,3-triazoles are viable nucleophiles for Hiyama cross-coupling reaction where aryl(trialkyl)silanes have drawn little attention as cross-coupling nucleophiles and have remained a long-standing important challenge.