Synthesis of densely substituted pyridine derivatives from nitriles by a non-classical [4+2] cycloaddition/1,5-hydrogen shift strategy


 A novel strategy has been established to assemble an array of densely substituted pyridine derivatives from nitriles and o-substituted aryl alkynes or 1-methyl-1,3-enynes via a non-classical [4 + 2] cycloaddition along with 1,5-hydrogen shift process. The well-balanced affinities of two different alkali metal salts enable the C(sp3)-H bond activation as well as the excellent chemo- and regioselectivities. This protocol offers a new guide to construct pyridine frameworks from nitriles with sp3-carbon pronucleophiles, and shows potential applications in organic synthesis and medicinal chemistry.


Introduction
Compounds containing pyridine core structures, not only widely exist in natural products, pharmaceuticals, and functional materials, 1-7 but also serve as useful and valuable building blocks for metal ligands. 8,9 For instance, pyridine derivative Actos I 10 is a famous drug for the treatment of type 2 diabetes; Bi-(or tri-)pyridines II [11][12][13][14][15] are often used as ligands in metal-catalyzed reactions; Kv1.5 antagonist III 16 and alkaloid papaverine IV 17 are two representative isoquinolines as a promising atrialselective agent and a smooth muscle relaxant, respectively (Fig. 1). Consequently, many synthetic methods have been developed to access this important class of N-heterocycles, including metal-free multicomponent syntheses, 18 transition metal-catalyzed [2 + 2 + 2] cycloaddition reactions [19][20][21][22][23][24][25] for pyridines, and transition metal-catalyzed annulations involving alkynes [26][27][28][29][30] for isoquinolines. Despite these advances, the lack of chemo-and regioselectivities in multicomponent reactions and requirement of complex functionalized substrates as well as harsh reaction conditions 22,31−33 often limits further exploitation of the synthetic potential of these approachs. The development of novel and e cient strategies toward pyridine derivatives is still highly desirable.
Herein we disclose our recent results on this non-classical [4 + 2] cycloaddition reaction, which provides a direct and atom-economic synthesis of densely substituted pyridines, isoquinolines, and benzothienopyridines with high chemo-and regioselectivity.

Results
Investigation of reaction conditions. To validate the hypothesis, we commenced our studies by evaluating the reaction between 1-methyl-2-(phenylethynyl)benzene (1a) and benzonitrile (2a) in THF at 100 o C in the presence of different alkali metal salts (Table 1). Several metal amides in combination with t-BuOK were initially tested (entries 1−3), and LiN(SiMe 3 ) 2 turned out to be the best choice to give product 3a in 35% yield. Subsequently, screening of the reaction temperatures identi ed the optimal 120 o C (entries 4−6). The examination of other parameters including additive, solvent, and reaction time (entries 7−15) revealed that the reaction proceeded well with t-BuOK as additive in CPME for 24 h, improving the yield of 3a to 65%. Next, investigation of detailed molar ratios of all components indicated that excess nitrile was required due to the consumption of self-cyclotrimerization, and the well-balanced interaction between LiN(SiMe 3 ) 2 and t-BuOK is the key to the success of reaction (entries 16 and 17, see the Supporting Information for details). Finally, the reaction of 1a: 2a: LiN(SiMe 3 ) 2 : t-BuOK in a molar ratio of 1: 3: 2: 1.5 gave 3a in 80% isolated yield (entry 17).
Substrate scope. Having established the optimal reaction conditions, the scope of this [4+2] cycloaddition reaction was then explored (Table 2). Pleasingly, o-substituted aryl alkynes 1 with electron-donating substituents, such as -Me, -t-Bu, and -OMe at the 4-position of aromatic ring Ar 1 , were well compatible in this reaction, and the desired products (3b-3d) were isolated in 60-78% yields. o-Substituted aryl alkyne bearing 3-Me substituted Ar 1 also reacted with 2a to afford 3e in a slightly descending yield, while the substrates containing halogens were relatively limited (e.g., 3f). The reaction proceeded as well with a larger aromatic Ar 1 , 2-naphthyl, providing the corresponding adduct 3g in 65% yield. Moreover, when using a series of functionalized nitriles 2 as reactants, including p-Me, p-t-Bu, p-OMe, p-Ph, and m-t-Bu substituted benzonitriles, isoquinoline derivatives (3h-3l) with moderate to good yields (48-82%) were obtained The structure of 3i was con rmed unambiguously by X-ray diffraction. The substituent R 1 on arylmethane ring could be varied as well (3m-3p) under the standard conditions. When R 2 was replaced by a cyano group, contributing to relatively acidic benzyl protons, the cycloadditions occurred smoothly in the absence of lithium amide (3q-3r). In addition, the introduction of a sterically demanding phenyl at the benzylic position enabled the release of larger conjugated isoquinoline derivative in 45% yield (3s).
Benzylic C-H bond with a methyl group at the benzylic position could also be functionalized in this reaction albeit in 28% yield (3t). However, the substrate with a phenoxy group R 2 failed to undergo the desired [4 + 2] cycloaddition reaction (3u).
To further expand the generality of this method, a series of functionalized heteroaromatic substrates were utilized to test the reactivity ( Table 3). The result indicated 3-alkynyl-2-methylbenzo[b]thiophene (4) with different substituents (H, OMe, N,N-2Me) on the aromatic ring Ar 1 were well tolerated, converting to the fused heterocycle benzothienopyridine derivatives 5a-5c in 71-90% yields. Gratifyingly, halogens, such as 4-F, 4-Cl, 4-Br, 3-Cl and 3,5-2F, could also be installed on the same aromatic ring Ar 1 , and the corresponding products 5d-5h were obtained in acceptable to good yields. The generated heteroaryl halides provided versatile synthetic handles for further derivatization. Moreover, the synthesis of compounds 5j-5n with 53-71% yields suggested that the compatibility of substituents on nitrile has been extended to -Cl, -CF 3 , -OMOM, and -1,3-dioxolane derived fromaldehyde group, thus providing a platform for additional functionalization. 3,5-Disubstituted benzonitrile and 2-naphthonitrile were also applicable in this system, and the desired products 5o and 5p were delivered in 72% and 66% yields, respectively. It should be noted that the pincer-type or dipyridyl-type nitrogen ligands 5i, 5q and 5r could be constructed in good yields. More importantly, this protocol was also applied to the late-stage functionalization of drug molecule citalopram, which is a selective serotonin-uptake inhibitor, and the compound 5s was provided in 67% yield, exhibiting the potential in pharmaceutical research.
To further illustrate the value of this method, gram-scale experiments and postderivatizations of the newly formed products were conducted (Fig. 3). When the reaction proceeded at a 5 mmol scale, the desired products 3c and 5a were obtained in almost unaffected yields even with reduced amount of nitrile. Additionally, inspired by the importance of organic uorophores in a number of elds including materials science and biology, two new solid-state-emitting organic molecules 8 and 9 were synthesized. Tetraarylethylene 8 was obtained by a single Suzuki-Miyaura coupling of 5f with 60% yield, while the successive hydrolysis and condensation reaction of 5n gave compound 9 in 95% total yield. Studies on the photophysical propertie showed the absorption maxima of 8 and 9 were about 336 nm and 385 nm in DMSO solution, respectively, and the corresponding intense emission peaks were at 404 nm and 587 nm in powder, which might nd applications in construction of functional materials.
Mechanistic investigations. To shed light on the reaction mechanism, several control experiments were performed. Firstly, when 15 N-labeled benzonitrile was emplyed under the standard reaction conditions, 3c( 15 N) and 3c( 14 N) were detected respectively with the ratio of 49: 51 by HRMS analysis, elucidating that the nitrogen atom in the pyridine framework came not only from nitrile but also LiN(SiMe 3 ) 2 (Fig. 4a).
Then, the studies of introducing exogenous halogens PhX (X = F, Cl, or Br) to the reaction system showed that uorobenzene slightly reduced the yield of 3c, while chlorobenzene and bromobenzene greatly suppressed the transformation, indicating that the coordination of PhX to the alkali metals might account for the inhibitory effect (Fig. 4b). 83 Next, a Hammett study was carried out using various substituted benzonitriles to investigate a rate dependence on the electronic effect of nitrile (Fig. 4c). A linear relationship with negative ρ value of -2.95 was observed when relative rates with -OMe, -Me, -H, -Cl, and -CF 3 substituted benzonitriles [log(k X /k H )] were plotted against the substituent constant (σ). These results suggested that the electron-donating group should facilitate the intermolecular addition step, which was consistent with the reactivity of the nitrile substrates reported in Tables 2-4. Finally, the kinetic order experiments (Fig. 4d) showed a rst-order dependence on concentrations of 2a, which indicated that the formation of metalated imine by quickly capturing benzylic−metal species with nitrile should be the ratelimiting step. However, the benzyl substrate 1c, theoretically accelerating an equilibrium formation of benzylic−metal species, exhibited the negative-order kinetic effect in the reaction. Moreover, the rates of both t-BuOK and LiN(SiMe 3 ) 2 could not be described by a single power function. These observations aligned with the optimization data possibly portend the well-balanced interactions between base and additive for benzylic or allylic C(sp 3 )-H activation. This cooperative interaction was further proved with hydrogen/deuterium exchange experiment of 1c-d (Fig. S1).
Based on the above investigation and literature reports, 82-90 a possible mechanism is proposed in Fig. 5.
Initially, the C(sp 3 )-H bond of 1a is activated by the interaction of LiN(SiMe 3 ) 2 and t-BuOK to afford a nucleophilic benzylic−metal species A, which is quickly trapped by nitrile 2a to form the metalated imine intermediate B. Next, subsequent addition of intermediate B to the alkyne achieves the cyclization. The resulting alkenyl-metal species C undergoes protonation and 1,5-hydrogen shift process to deliver the desired 3a. Alternatively, the reaction could proceed by either protonation or silylation of intermediate B with HN(SiMe 3 ) 2 from the initial deprotonation equilibrium, generating imine D. 82 In the presence of Li/KN(SiMe 3 ) 2 , the imine D would undergo a reversible addition, silyl migration, and elimination pathway to release nitrogen-exchanged intermediate F, 91,92 followed by the further silyl transfer and cyclization to give nitrogen-exchanged intermediate G.

Discussion
In summary, we have developed a non-classical [4 + 2] cycloaddition reaction of benzylic and allylic prenucleophiles with nitriles, leading to a range of N-heterocycles bearing pyridine frameworks and various functional groups. This protocol hinges on the cooperative interaction of alkali metals to trigger benzylic and allylic C-H bond activation, simultaneously avoiding the inherent intramolecular cyclization.
The Hammett experiment and kinetic analysis show that the addition of benzylic-or allylic − metal species to nitrile is the rate-limiting step. Notably, this protocol provides a new guide for the synthesis of functionalized pyridines from sp 3 -carbon pronucleophiles with nitriles, which implies the potential value in organic synthesis and medicinal chemistry.

Method
Procedure for the synthesis of product 3. In a glovebox, an oven-dried Schlenk tube equipped with a magnetic stirring bar was added a mixture of o-substituted aryl alkyne 1 (0.2 mmol), nitrile 2 (0.6 mmol), base (0.4 mmol), additive (0.3 mmol) dissolved in 1.0 mL of dry CPME. The Schlenk tube was sealed with a plug and removed from the glovebox. The mixture was allowed to stir at 120°C under N 2 for 24 h. The reaction mixture was cooled to room temperature and quenched by 10 mL water. The product was extracted with ethyl acetate (10 mL × 3). The organic phase was dried over Na 2 SO 4 and concentrated in vacuum. The resulting residue was puri ed by silica gel column chromatography using petroleum ether/ethyl acetate (100:1-10:1 v/v) as eluent to afford the corresponding product 3.
Procedure for the synthesis of product 5. In a glovebox, an oven-dried Schlenk tube equipped with a magnetic stirring bar was added a mixture of o-substituted aryl alkyne 4 (0.2 mmol), nitrile 2 (0.6 mmol), base (0.4 mmol), additive (0.24 mmol) dissolved in 0.5 mL of dry CPME. The Schlenk tube was sealed with a plug and removed from the glovebox. The mixture was allowed to stir at 120°C under N 2 for 24 h.
The reaction mixture was cooled to room temperature and quenched by 10 mL water. The product was extracted with ethyl acetate (10 mL × 3). The organic phase was dried over Na 2 SO 4 and concentrated in vacuum. The resulting residue was puri ed by silica gel column chromatography using petroleum ether/ethyl acetate (100:1-10:1 v/v) as eluent to afford the corresponding product 5.
Procedure for the synthesis of product 7. In a glovebox, an oven-dried Schlenk tube equipped with a magnetic stirring bar was added a mixture of 1-methyl-1,3-enyne 6 (0.2 mmol), nitrile 2 (0.6 mmol), base (0.4 mmol), additive (0.24 mmol) dissolved in 0.5 mL of dry 1,4-dioxane. The Schlenk tube was sealed with a plug and removed from the glovebox. The mixture was allowed to stir at 120°C under N 2 for 12 h.
The reaction mixture was cooled to room temperature and quenched by 10 mL water. The product was extracted with ethyl acetate (10 mL × 3). The organic phase was dried over Na 2 SO 4 and concentrated in vacuum. The resulting residue was puri ed by silica gel column chromatography using petroleum ether/ethyl acetate (100:1-50:1 v/v) as eluent to afford the corresponding product 7.

Data availability
The authors declare that the data supporting the ndings of this study are available within the article and Supplementary Information, as well as from the authors upon reasonable request. The X-ray crystallographic coordinate for structure 3i reported in this study has been deposited at the Cambridge Crystallographic Data Centre (CCDC), under CCDC 2051390. The data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data request/cif.

Competing interests
The authors declare no competing interests.
Author contributions D.H. and W.W. designed the project. D.H., H.J., and W.W. co-wrote the manuscript, analyzed the data, discussed the results, and commented on the manuscript. D. H., K.D., and Y.Z. performed the experiments. All authors contributed to discussions.