Uncanonical Semireduction of Quinolines and Isoquinolines via Regioselective HAT-Promoted Hydrosilylation

Abstract

Heterocycles are the backbone of modern medical chemistry and drug development. The derivatization of “an olefin” inside aromatic rings represents an ideal approach to access functionalized saturated heterocycles from abundant aromatic building blocks. Here, we report an operationally simple, efficient, and practical method to selectively access hydrosilylated and reduced N-heterocycles from bicyclic aromatics via a key diradical intermediate. This approach is expected to facilitate complex heterocycle functionalizations that enable access to novel medicinally relevant scaffolds.

Heterocycles form the backbone of modern pharmaceuticals. Saturated heterocycles, in particular, represent important building blocks in the practice of medicinal chemistry.^1-4 Despite their significance, the synthesis of saturated heterocycles, especially those having multiple substituents that are distal to a heteroatom, represents a formidable challenge.⁵ On the other hand, unsaturated heterocycles (e.g., heteroaromatics) are some of the most commercially abundant building blocks. Additionally, the πelectrons in the heteroaromatics are versatile functional handles for selective functionalizations.^5-23 Regiocontrolled reduction of heteroarenes 1 to their saturated or semisaturated congeners 2 represents an attractive approach to access a diverse array of heterocyclic moieties (Figure 1A). Toward this end, the classic Birch reduction remains the state-of-art method to reduce (hetero)arenes. These techniques convert aromatics into radical anion intermediate 4, which is then protonated to afford reduced product 5 (Figure 1B).^24-35 In recent years, other dearomatization strategies have also been reported.^36-41 However, most of these methods are limited by substrate scope or harsh conditions. Developing mild alternative strategies capable of breaking the strong aromaticity in Lewis basic heteroaryls and with regioselective control remains an outstanding challenge. Inspired by the elegant reports of photoinduced [4+2] cycloaddition between heterocycles and olefins^42-50 or triazolinediones,^51-54 we envisaged that photo-excited-state intermediates such as 7 might be leveraged to provide regioselective hydrosilylated products 8 and reduced products 9 from abundant feedstocks in a controlled manner (Figure 1C). The dearomative hydrosilylation of heterocycles, in particular, might provide unique opportunities to access novel medicinally relevant scaffolds in medicinal chemistry settings. Heteroarylsilane derivatives, such as 8, potentially offer improved lipophilicity, stability, or pharmacokinetic properties.^55-58 Moreover, the resulting allylsilane synthon would supply a synthetic handle for subsequent functionalization,^59,60 enabling library synthesis of multiple-substituted saturated heterocycles featuring a higher fraction of sp³ carbon atoms (Fsp³)^61,62 (Figure 1A).

Figure 1.

Regioselective saturation of bicyclic heterocycles. (A) Increasing aromatics saturation provides three-dimensional chemical space to escape from flatland; (B) classic Birch reduction breaks aromatics through radical anion intermediates; (C) proposed diradical synthon to access dearomatized hydrosilylation and reduction products from heterocycles.

Our dearomatization exploration commenced with the hypothesis that the photoexcited quinolinium intermediate might be reduced by a hydride reductant,^42-54 such as silane reagents,^63-69 to enable direct addition and reduction of the heteroarenes. During the solvent investigation, formation of an intriguing hydrosilylated product 11 was observed and subsequently optimized (Table 1, entry 1).⁷⁰ The dearomatized allylic silane 11 was found to be sensitive to acidic conditions. However, the protic solvent hexafluoroisopropanol (HFIP) provided an optimal balance between promoting the addition and efficiently preserving product (compare entries 1 and 2 and vide infra).^61,62 Other solvents led to lower yields or no product formation (entries 4–6). A series of photosensitizers were also investigated for this reaction, and only a small amount of 12 was observed (see Supporting Information for photosensitizer investigation). It is noteworthy that strict oxygen removal is not required for this reaction, with a product yield of 50% being observed when the reaction is conducted in the presence of air. The [2+2] photocycloaddition of allylic silane 11 was observed when prolonging the reaction time to 18 h (see SI-12, Supporting Information).

Table 1.

Saturation and Hydrosilylation Optimization of Quinolines

entry	deviation from above	yield of 11	12 (%)a
Hydrosilylation:
1	none	53	<5
2	TFA (2.0 equiv) as additive	26	30
3	Et3SiH (2.0 equiv)	42	0
4	CF3CH2OH instead of HFIP	33	<5
5	iPrOH, CH2Cl2, THF, DMF instead of HFIP	0	0
6	HFIP (0.2 M)	46	<5
Reduction:
7	none	0	75
8	no TFA or UV irridiation	0	0
9	AcOH, MsOH, Et2O.BF3 instead of TFA	0	0
10	PhSiH3 or (EtO)3SiH instead of Et3SiH	0	0
11	hν 254 nm instead of 300 nm	0	34
12	hν 365 nm instead of 300 nm	0	37

^aYield determined by NMR analysis with CH₂Br₂ as an internal standard; quartz tubes were used for 254 nm reactions, and Pyrex tubes were used for 300 and 365 nm reactions. HFIP, hexafluoro-2-propanol. See the Supporting Information for experimental details.

Since the desilylated compound 12 is also of high synthetic utility, we intended to develop a one-step procedure to access 12 via in situ protonic desilylation of 11. After extensive optimization, the combination of trifluoroacetic acid (TFA) and triethylsilane under 300 nm photoirradiation led to the formation of the reduced product 12 in 75% yield (Table 1, entry 7; see the Supporting Information for other conditions and details). Both TFA and triethylsilane were found to be crucial for this reduction, and modifying the conditions to employ alternative acidic additives or reductants was detrimental (entries 8–10). Various wavelengths were also investigated, and 300 nm was identified as the optimal wavelength for the reduction of quinoline substrate 10 (entries 11–12).

To evaluate the substrate scopes for both the reactivities, the robustness of the reduction and hydrosilylation reactions was demonstrated through the preparation of over 50 dearomatized products from commercial heterocycles (Figure 2). The investigation started with 1,2-hydrosilyation of isoquinoline (Figure 2A), where various silane reagents were explored in this reaction, yet only trialkylsilane reagents could provide a noticeable amount of the hydrosilylated products (13–20). The bulky silane reagents, such as tri-isopropylsilane (16) or dimethyl-tert-butylsilane (20), led to lower yields, presumably due to steric hindrance. Other silanes often afford the reduced product 54, or a large amount of starting material remains (see the Supporting Information for results of other silanes). A wide range of quinolines (21–30) and isoquinolines (31–36) were studied using triethylsilane as the reductant. The alkyl groups (11, 28, 29) and cyano (24), ester (25), fluoro (30), chloro (27), and even vinyl (22) groups were well tolerated. For quinolines and isoquinolines, the C6- and C7-silylated products are the major regioisomers, respectively. For the isoquinolines (33–36) with an electron-withdrawing group at the C3-position, a mixture of silylated products (C7 and C6) was isolated. Furthermore, this reaction could extend to other heterocycles (see the Supporting Information), such as quinazoline (37). This reaction was also scalable and provided comparable yields on a 10 mmol scale (11, 13).

Figure 2.

Substrate scope of hydrosilylation and reduction of heterocycles. Products are racemic mixtures, unless annotated. (A) Hydrosilylation; heteroaromatics (1.0 equiv), triethyl silane (3.0 equiv) in HFIP (0.1 M) were stirred under UV (300 nm) irradiation for 12 h; ^a10 mmol scale; ^bunder UV 254 nm; ^cTFA (1.0 equiv) as additive; ^dTFA (1.5 equiv) as additive. (B) Reduction; heteroaromatics (1.0 equiv), TFA (2.0 equiv), triethyl silane (2.2 equiv) in MeCN (0.2 M) were stirred under UV (300 nm) irradiation for 12 h; ^e20 mmol scale; ^funder UV 365 nm; ^g5 mmol scale; ^hTFA (1.0 equiv), PhMe₂SiH (3.0 equiv), MeCN (0.2 M), UV 300 nm; ⁱTFA (1.0 equiv), Et₃SiH (3.0 equiv), HFIP (0.1 M), UV 300 nm; ^jTFA (1.0 equiv), PhMe₂SiH (3.0 equiv), HFIP (0.1 M), UV 300 nm; ^kPhMe₂SiH (3.0 equiv), HFIP (0.1 M), UV 300 nm. See the Supporting Information for experimental details.

The substrate scope exploration of the reduction was initially evaluated on various quinolines (Figure 2B). Besides 5-substituted quinolines, other substitution patterns (38–53) were also found to be compatible in this reduction. Bromo (39, 43), primary benzylic alcohol (40, 50), ester (41, 44), cyano (42), methyl ether (45), chloro (47, 52), and boron pinacol ester (49) were all tolerated. Similarly, isoquinolines (54–56) were smoothly reduced under the standard conditions. The structures of compounds 40, 43, and 47 were unambiguously assigned by single-crystal X-ray analysis. The robustness of this operationally simple reaction was evaluated on multiple substrates by running the reduction at 5 to 20 mmol scales. The chemoselectivity of this transformation and applicability to medicinal-chemistry-relevant molecules were showcased by the successful reduction of deoxy-cinchonine (57) and fenazaquin (58).

The synthetic applicability of this dearomatization strategy was further showcased by derivatization of 11 and 12 to various saturated or partially saturated compounds, which are challenging to access otherwise (Figure 3A). For example, silylated compound 11 could undergo tetra-n-butylammonium fluoride (TBAF)-mediated allylic additions with aldehydes (59), benzyl bromides (60), propargyl bromides (61), or proton sources (12) to access C8-substituted dihydroquinoline products. Under DDQ oxidation conditions, the C6-silylated quinoline (62) was obtained in a high yield. Besides the intrinsic allyl silane reactivity, the remaining olefin moiety after dearomatization represents another versatile functional group for downstream derivatization. For example, Pd/C-catalyzed hydrogenation (65), [2+2]-photocycloaddition with dimethyl malonate (64), and hydroboration–oxidation (63) each produced the corresponding products in useful yields. In parallel, the partially reduced product 12 successfully underwent dihydroxylation (66), isomerization (67), N-oxide formation (68), addition into ketone (69), allylation (70), and reduction (71).

Figure 3.

Derivatization and synthetic application of dearomatized products. (A) Downstream derivatization of hydrosilylated compound 12 and reduced product 11; (B) synthetic application to access ORL1 antagonist 75; see the Supporting Information for experimental details.

This dearomative reduction method also enables more straightforward retrosynthetic disconnections for bioactive heterocycles by avoiding multistep and de novo syntheses of fused heterocycles. Cyclohexyl-fused pyridine 75 was developed as an opioid receptor-like 1 (ORL1) antagonist and was previously synthesized in 13 steps through de novo synthesis of the cyclohexyl ring (Figure 3B).⁷¹ Using the current method, we could simply access and subsequently employ quinoline reduced product 50 in a five-step sequence (alcohol protection with a triisopropylsilyl (TIPS) group, olefin isomerization, TBAF-assisted deprotection, installation of a p-toluenesulfonyl leaving group, Mukaiyama hydration, and S_N2 substitution) to access the same antagonist with higher efficiency. Notably, this strategy enables one to utilize a commercially available substituted quinoline followed by rapid and divergent functionalization.

In order to understand the reaction mechanism, a series of deuterium labeling and control experiments were conducted (Figure 4A). Under the standard reduction conditions, reactions with the Et₃Si-D (eq 1) and TFA-D (eq 2) indicated that the C5-proton (79) derives from the silane and the C8-proton is delivered by the acid (80/12). Deuteration of silylated product 11 with TFA-D afforded a similar 9:1 ratio of 80:12 (eq 3) to that observed in eq 2. This desilylation could also be thermally promoted at 60 °C (see Supporting Information). These results together suggested that the dearomative hydrosilylation is the initial step for the reduction. Despite the possibility of a radical mechanism, this reaction generated a noticeable amount of product in the presence of 4 equiv of TEMPO, albeit in lower yield, and no TEMPO adduct was detected by GC-MS or LC-MS (eq 4). In this context, TEMPO is less likely to act as a radical scavenger but instead might quench the photoexcited isoquinoline. We then turned our attention to understand the stereochemical outcome of this transformation. The deuterated-triethylsilane (eq 5) and C5-deuterated-quinoline (not shown) both resulted in a diastereomeric ratio of 9:1, favoring the syn hydrosilylation products (82 and 83). Crossover experiments with Et₃Si-D and nBu₃Si-H (eq 6) found no deuteron/proton scrambling between two different silanes. Based on experimental observations and mechanistic studies, we believe that the mechanism proceeds via a stepwise hydrogen atom transfer (HAT)/radical recombination mechanism (Figure 4B). The observed regioselectivities of the hydrosilylation are presumably due to the more electron-deficient radicals being photochemically generated at the 5-position of quinoline or 8-position of isoquinoline (intermediate B, Figure 4B). More detailed mechanistic studies, explanation for observed regioselectivity, and corresponding computational studies are ongoing and will be reported in due course.

Figure 4.

Mechanistic studies. (A) Mechanistic probe reactions: deuterium labeling experiments, TEMPO control experiment, hydrosilylation stereochemistry, and crossover experiment; (B) proposed mechanism.

In conclusion, we have developed a dearomatization method for the hydrosilylation and reduction of bicyclic heterocycles. As showcased in Figures 2 and 3, this operationally simple method enables rapid preparation of a variety of synthetically challenging saturated heterocycles in a chemoselective manner. In addition, the mechanistic studies demonstrate a stepwise HAT/radical recombination mechanism. We expect this method to have a substantial impact within the area of drug discovery through the combined derivatization and application of unique diradical synthons to access complex saturated heterocyclic building blocks.