Chiral α-Stereogenic Oxetanols and Azetidinols via Alcohol-Mediated Reductive Coupling of Allylic Acetates: Enantiotopic π-Facial Selection in Symmetric Ketone Addition

Abstract

Iridium-tol-BINAP-catalyzed reductive coupling of allylic acetates with oxetanones and azetidinones mediated by 2-propanol provides chiral α-stereogenic oxetanols and azetidinols. As illustrated in 50 examples, complex, nitrogen-rich substituents that incorporate the top 10 N-heterocycles found in FDA-approved drugs are tolerated. In addition to 2-propanol-mediated reductive couplings, oxetanols and azetidinols may serve dually as reductant and ketone proelectrophiles in redox-neutral C-C couplings via hydrogen auto-transfer, as demonstrated by the conversion of dihydro-1a and dihydro-1b to adducts 3a and 4a, respectively. The present method delivers hitherto inaccessible chiral oxetanols and azetidinols, which are important bioisosteres.

Graphical Abstract

Catalytic enantioselective addition of C-nucleophiles to ketones remains a formidable challenge in chemical synthesis.^1,2 Ketones are more stable than aldehydes and reversible addition can erode kinetic stereoselectivity. To enforce irreversible addition, non-stabilized premetalated C-nucleophiles are often required and, in the case of catalytic enantioselective ketone allylation,^3,4,5 allylmetal reagents based on B,^3a–j Si,^3k,l and Sn^3m–s have been developed. Catalytic reductive coupling precludes the need for discrete allylmetal reagents,⁴ but as exemplified by Nozaki-Hiyama-Kishi ketone allylations,⁵ zero-valent metals are often required as reductants. Recently reported enantioselective copper-catalyzed reductive ketone allylations represent a significant advance, as tractable allene or diene pronucleophiles may be used with silane reductants.⁶ Allylations mediated by inexpensive feedstock reductants (H₂, HCO₂H, 2-PrOH)⁷ are even more ideal, but applications to ketone electrophiles remain limited.⁸ A second challenge associated with catalytic asymmetric ketone allylation is that a steric or electronic bias between groups flanking the carbonyl is typically required to enforce enantioselectivity.^2,9 For symmetric ketones, this mode of enantiodiscrimination is unavailable, requiring enantiotopic π-facial discrimination of the σ-allylmetal nucleophile.¹⁰ Perhaps due to this issue, systematic studies of catalytic asymmetric additions to symmetric ketones are exceptionally uncommon. Isolated examples of allylation^3f,j,6a and aldol addition exist,¹¹ but, to our knowledge, only one systematic study of catalytic enantioselective additions to symmetric ketones has been described, which employs diene pronucleophiles (Figure 1).^8c

Figure 1.

Enantiotopic π-facial discrimination in the asymmetric allylation of non-symmetric and symmetric ketones.

Here, using an air and water stable iridium-tol-BINAP catalyst, we report 2-propanol-mediated reductive couplings of racemic branched allylic acetates 2a-2x (and phthalimido-allene 2x) with oxetanone 1a and N-benzhydryl azetidinone 1b to form highly enantiomerically enriched chiral α-stereogenic oxetanols and azetidinols. Despite formation of congested acyclic vicinal tertiary-quaternary centers,¹² complete branched regioselectivities are observed. This method enables entry to unique chiral oxetanes¹³ and azetidines,¹⁴ which serve as metabolically stable bioisosteres in pharmaceutical and agrochemical ingredients (Figure 2).^15,16 Functional group compatibility and relevance to drug discovery is underscored by the use of allyl pronucleophiles incorporating the top 10 N-heterocycles found in FDA-approved drugs.¹⁷

Figure 2.

Commercial oxetanes and azetidines.

Research Design and Methods

Optimization experiments initially focused on the evaluation of axially chiral chelating phosphine ligands I-VII in the reaction of oxetanone 1a with allylic acetate 2a (Table 1, entries 1–7). A promising result was obtained using the cyclometalated π-allyliridium-C,O-benzoate complex derived from 4-cyano-3-nitrobenzoic acid and (S)-tol-BINAP, (S)-Ir-VII, which delivered the desired oxetanol 3a in 61% yield and 99% ee (Table 1, entry 7). Upon variation of the electronic properties of the C,O-benzoate moiety (Table 1, entries 7–9), a comparable isolated yield of 3a was observed using the more electron-deficient 3,4-dinitrobenzoate (Table 1, entry 9). As 3,4-dinitrobenzoic acid is commercially available and 3-nitro-4-cyanobenzoic acid is not, further optimization was conducted using the 3,4-dinitrobenzoate. A survey of different bases revealed that for reactions conducted using K₂CO₃ (100 mol%) the isolated yield of 3a was elevated to 79% (Table 1, entry 12). Finally, as loss of the C,O-benzoate ligand represents a catalyst decomposition pathway, 3,4-dinitrobenzoic acid was employed as an additive in the hope of extending the lifetime of the catalyst. Indeed, in the presence of 3,4-dinitrobenzoic acid (5 mol%), 3a was formed in 96% yield and 99% ee. Notably, these cyclometalated π-allyliridium-C,O-benzoate catalysts were ineffective in corresponding reactions using diene pronucleophiles, which required a cyclometallated PhanePhos catalyst.^8c The observed difference in reactivity may be due to the axial vs planar chirality of the ligands, which endow the catalysts used in each study with distinct chiral-at-iridium topologies. Whereas the diene pronucleophiles used in the prior study give rise to C2-substituted allyliridium species, the allylic acetate pronucleophiles used in the present study give rise to C1-substituted allyliridium species.

Table 1.

Selected optimization experiments in the enantioselective iridium-catalyzed reductive coupling of oxetanone 1a with allyl acetate 2a to form oxetanol 3a.^a

Entry	Base	(S)-Ligand	X	Additive	3a (Yield, ee)
1	Cs2CO3 (20 mol%)	I	CN	---	44%, 98%
2	Cs2CO3 (20 mol%)	II	CN	---	37%, 98%
3	Cs2CO3 (20 mol%)	III	CN	---	42%, 99%
4	Cs2CO3 (20 mol%)	IV	CN	---	44%, 99%
5	Cs2CO3 (20 mol%)	V	CN	---	40%, 97%
6	Cs2CO3 (20 mol%)	VI	CN	---	46%, 99%
7	Cs2CO3 (20 mol%)	VII	CN	---	61%, 99%
8	Cs2CO3 (20 mol%)	VII	OMe	---	20%, 98%
9	Cs2CO3 (20 mol%)	VII	NO2	---	64%, 98%
10	Cs2CO3 (100 mol%)	VII	NO2	---	23%, 98%
11	K2CO3 (20 mol%)	VII	NO2	---	30%, 99%
12	K2CO3 (100 mol%)	VII	NO2	---	79%, 99%
➡13	K2CO3 (100 mol%)	VII	NO 2	3,4-(NO2)2BzOH	96%, 99%

^aYields of material isolated by silica gel chromatography. Enantioselectivities were determined by chiral stationary phase HPLC analysis. See Supporting Information for further details.

It was of interest to determine whether conditions optimized for the enantioselective iridium-catalyzed reductive coupling of oxetanone 1a with allyl acetate 2a were transferrable to the related ketones azetidinone 1b and cyclobutanone 1c (Scheme 1). In the event, azetidinone 1b was converted to the azetidinol 4a in 65% yield and 98% ee, and at longer reaction times (36 hr) azetidinol 4a could be obtained in 82% yield. In contrast, cyclobutanone 1c did not participate in reductive coupling. As determined by electron transmission spectroscopy, there exists a substantial difference in LUMO energies between oxetanone 1a (14 Kcal/mol) and cyclobutanone 1c (24 Kcal/mol) that is not attributed to n→π* interactions.¹⁸ As revealed by crystallographic data,¹⁹ the shorter carbon-heteroatom bond lengths of oxetanone 1a (1.46 Å) and azetidinone 1b (1.50 Å) compared to cyclobutanone 1c (1.54 Å) appear to compress the angle between the C-C bonds: 88.9° vs 90.8° vs 93.2°, for 1a, 1b and 1c, respectively. Based on these data, we posit that increased angle strain and σ-inductive effects associated with heteroatom substitution account for the observed divergence in reactivity in reductive couplings to oxetanone 1a, azetidinone 1b vs cyclobutanone 1c.

Scheme 1.

Enantioselective iridium-catalyzed reductive coupling of allylic acetate 2a with oxetanone 1a, azetidinone 1b and cyclobutanone 1c, experimentally determined LUMO energies and selected single crystal X-ray diffraction data.^a

^aYields of material isolated by silica gel chromatography. Enantioselectivities were determined by chiral stationary phase HPLC analysis. See Supporting Information for further details.

To assess the scope of this process, optimal conditions identified for the iridium-tol-BINAP-catalyzed reductive coupling of allylic acetate 2a with oxetanone 1a or azetidinone 1b were applied to diverse allylic acetates 2a-2v, which incorporate the 10 most frequently encountered N-heterocycles in FDA-approved drugs (beyond β-lactams) (Table 2).¹⁷ To our delight, the coupling of oxetanone 1a and azetidinone 1b to allylic acetates bearing a variety of substituted aryl (2a-2d) and heteroaryl (2e-2l) groups, alkyl (2n-2r, 2u, 2v) and cycloalkyl (2m, 2s, 2t) groups, delivered oxetanols 3a-3v and azetidinols 4a-4v in good yield with excellent levels of enantioselectivity. In particular, the formation of adducts bearing pinacol boronates (3i, 3r, 4i, 4r) and those derived from the FDA approved drugs indomethacin (3n, 4n) and losartan (3v, 4v) highlights the exceptional functional group tolerance of this method, and its suitability for late-stage functionalization of clinical candidates.²⁰ Catalyst-directed diastereoselectivity is illustrated by the conversion of chiral allylic acetate 2w (derived from (+)-α-pinene) to oxetanols 3w and iso-3w and azetidinols 4w and iso-4w (Table 3). The absolute stereochemistry of adducts 3a-3v and 4a-4v was assigned in analogy to compounds 3a and 4a, which were determined by single-crystal X-ray diffraction analysis. As illustrated in reactions of oxetanol dihydro-1a and azetidinol dihydro-1b, ketone allylation can be performed from the alcohol oxidation level via hydrogen auto-transfer, although such processes are slightly less efficient (eq. 1 & 2). Finally, beyond couplings to allylic acetates, phthalimido-allene 2x is a competent pronucleophile, as illustrated by the enantioselective formation of adducts 3x and 4x (eq. 3 & 4).

Table 2.

Iridium-tol-BINAP-catalyzed reductive coupling of oxetanone 1a or azetidinone 1b with racemic allylic acetates 2a-2v to form enantiomerically enriched oxetanols 3a-3v and azetidinols 4a-4v, respectively.^a

^aYields of material isolated by silica gel chromatography. Enantioselectivities were determined by chiral stationary phase HPLC analysis. Standard conditions: 0.2 mmol scale, 18 h for 1a, 36 h for 1b. See Supporting Information for further experimental details.

^b24 h

^c48 h

^d(S)-Ir-tol BINAP (7.5 mol %),

^e(S)-Ir-SEGPHOS (5.0 mol%).

^fDerivatized for ee% determination.

Table 3.

Catalytic-directed diastereoselectivity in iridium-tol-BINAP-catalyzed reductive couplings of oxetanone 1a or azetidinone 1b with non-racemic allylic acetate 2w.^a

^aYields and diastereomeric ratios are of material isolated by silica gel chromatography. See Supporting Information for further experimental details.

(eq. 1)

(eq. 2)

(eq. 3)

(eq. 4)

A general catalytic mechanism and stereochemical model for iridium-tol-BINAP-catalyzed reductive coupling of oxetanone 1a or azetidinone 1b with racemic allylic acetates 2a-2v mediated by 2-propanol is proposed in analogy with prior computational studies (Figure 3).²¹ Entry into the catalytic cycle occurs via protonolysis of the π-allyl precatalyst by 2-propanol. β-Hydride elimination of the resulting alkoxide releases acetone and generates an iridium hydride, which upon deprotonation forms an anionic square planar iridium(I) species, ablating stereochemistry at the iridium center. Ionization of the allylic acetate is enantiodetermining as it provides an allyliridium(III) complex that is stereogenic at the iridium center. Ketone addition occurs through a six-centered transition structure. Exchange of the resulting homoallylic alkoxide with 2-propanol releases product and closes the catalytic cycle. The enantiofacial selectivity observed in the present ketone additions is consistent with the indicated stereochemical model based on prior DFT calculations for corresponding aldehyde allylations.²¹

Figure 3.

General catalytic mechanism and proposed stereochemical model.

In summary, we report the first enantioselective additions of allylic acetate pronucleophiles to symmetric ketones, as illustrated in iridium-tol-BINAP-catalyzed reductive couplings with commercially available oxetanone 1a and N-benzhydryl azetidinone 1b mediated by 2-propanol. This method provides access to chiral nonracemic oxetanols and azetidinols bearing structurally complex, nitrogen-rich substituents, including the top 10 N-heterocycles found in FDA-approved drugs.¹⁷ Given the importance of oxetanes and azetidines as metabolically stable bioisosteres,^15,16 it is our hope this method will accelerate small-molecule drug discovery campaigns.²²