Lactonization as a general route to β-C(sp3)-H functionalization.

Functionalization of the β-C-H bonds of aliphatic acids is emerging as a valuable synthetic disconnection that complements a wide range of conjugate addition reactions1-5. Despite efforts for β-C-H functionalization in carbon-carbon and carbon-heteroatom bond-forming reactions, these have numerous crucial limitations, especially for industrial-scale applications, including lack of mono-selectivity, use of expensive oxidants and limited scope6-13. Notably, the majority of these reactions are incompatible with free aliphatic acids without exogenous directing groups. Considering the challenge of developing C-H activation reactions, it is not surprising that achieving different transformations requires independent catalyst design and directing group optimizations in each case. Here we report a Pd-catalysed β-C(sp3)-H lactonization of aliphatic acids enabled by a mono-N-protected β-amino acid ligand. The highly strained and reactive β-lactone products are versatile linchpins for the mono-selective installation of diverse alkyl, alkenyl, aryl, alkynyl, fluoro, hydroxyl and amino groups at the β position of the parent acid, thus providing a route to many carboxylic acids. The use of inexpensive tert-butyl hydrogen peroxide as the oxidant to promote the desired selective reductive elimination from the Pd(IV) centre, as well as the ease of product purification without column chromatography, render this reaction amenable to tonne-scale manufacturing.

Main Text

Alkyl carboxylic acids are ubiquitous and inexpensive reagents in organic chemistry – as such, they are privileged substrates for C–H activation reactions[4,5]. The scope of these transformations is often limited due to the incompatibility of certain reaction partners. Indeed, for C–C bond formations, alkylation reactions are limited to primary alkyl iodide or alkyl boron coupling partners[6-8], olefination reactions are limited to electron-deficient olefins[9,10], alkynylation reactions are limited to silyl acetylene bromide[11], and arylation reactions are only compatible with aryl iodides but not the more practical aryl bromides and chlorides[12,13] despite the design of various directing groups. Most importantly, C–heteroatom bond-forming reactions (fluorination, hydroxylation, amination etc.) based on β-C–H activation of free aliphatic acids have not yet been realized. Considering these persistent limitations of the conventional β-C–H activation approach, we turned to a one-for-all β-lactonization strategy (Fig. 1a). β-Lactones are strained heterocycles that have received significant attention as valuable synthetic intermediates in natural and unnatural products synthesis[14,15]. Due to their inherent ring strain, they readily react with a wide range of nucleophiles by either acyl C–O or alkyl C–O bond cleavage. The lack of precedent of this reaction is likely due to the highly unfavored four-membered lactonization transition state[15]. Notably, this β-lactonization could provide a strategy to synthesize carboxylic acids containing α-quaternary centers that are inaccessible by conjugate addition chemistry, and difficult to prepare via α-substitution[16].

Fig. 1.

β-C(sp3)–H functionalization.

a, Lactonization as a general and scalable route to β-C(sp3)–H functionalization. b, Challenges: multiple reductive elimination pathways of PdIV centers (Nu = nucleophile = acid, solvent, etc). L = ligand, TBHP = tert-butyl hydroperoxide, HFIP = 1,1,1,3,3,3-Hexafluoro-2-propanol.

A mixture of K2PtCl4, (17 mol%), K2PtC16 (33 mol%) can promote the formation of γ-lactones from aliphatic acids in 16% yield, accompanied by 2% β-lactone[17,18]. γ-Lactonization of benzylic C–H bonds has also been reported using Pd and Pt catalysts[19,20]. These observations indicate that β-lactonization is a highly disfavored process. Guided by previous work using a bystanding oxidant to promote C–H activation/cyclization reactions[21,22], we began to investigate catalysts and conditions to achieve an unprecedented β-C–H lactonization reaction. Compared to β-lactam formation, where a nucleophilic directing group can be employed to form a strong C–N bond[23,24], β-C–H lactonization poses an additional challenge due to the low nucleophilicity of the carboxylic acid, the strain generated in forming a four-membered ring, and the facile ring opening under C–H activation conditions. Most problematically, Pd(IV) intermediates could readily undergo conventionally favored reductive elimination to form non-cyclic C–O bond formation products such as the most common competing pathways acetoxylation and alkoxylation (Fig. 1b). We selected 2,2-dimethylbutyric acid 1a as a model substrate when searching for reactivity with a wide range of oxidants and catalysts. The exploratory studies using various common oxidants for Pd(II)/Pd(IV) chemistry, such as PhI(OAc)2, K2S2O8, and F+ reagents, consistently gave the undesired non-cyclic oxidation products (see Supplementary Information Table S3 and Mechanistic Studies section for details). To avoid the undesired reductive elimination pathway, we tested the sterically bulky oxidant TBHP[4] as well as the PdCl2 derived catalysts as the BuO and Cl anion is less prone to reductive elimination due to sterics and electronics. The desired β-lactone 2a was formed in 15% NMR yield using a combination of Pd(CH3CN)2Cl2, TBHP oxidant, CsHCO3 and HFIP solvent. Encouragingly, no γ-lactone or β-, γ-hydroxylated products were observed during the reaction. The unique role of TBHP for favoring β-lactone formation can be rationalized based on the studies on the oxidation of Pd(II) to Pd(IV) by benzoyl peroxide[25] and TBHP[4]. Following the oxidation of Pd(II) to Pd(IV) by TBHP, BuO- and HO- bound to Pd(IV) center are less likely to undergo rapid reductive elimination due to the strong Pd–OBu (OH) bond. Based on the principle of organometallic chemistry, the steric hindrance of BuO- could also enhance the reductive elimination of the carboxylate from substrate to generate β-lactone product.

In light of the recent advances in ligand-accelerated Pd(II)-catalyzed C–H activation[26], we next began to search for ligands that could significantly improve the reactivity of the catalyst. It is also possible an appropriate ligand could enhance the otherwise unfavored β-lactonization. Using the mono-N-protected α-amino acid (MPAA) ligand N-acetyl glycine L1, the yield was improved to 36%. Modification of the backbone of the α-amino acid ligand led only to minor improvements (L2 to L5). Considering the challenging reductive elimination of a strained four-membered ring from Pd(IV), we reasoned that switching the ligand binding mode from five- to six-membered chelation will increase the bite angle thereby favoring the desired reductive elimination. The β-amino acid-derived ligand N-acetyl β-alanine L6 under the same conditions improved the yield to 48%. Building on this promising finding, we then investigated the influence of substituents on the ligand’s side chain. Substituents at the β-position slightly reduced the reactivity (L7 to L10), suggesting that steric hindrance around the NHAc moiety was detrimental to reactivity. Meanwhile, substitution at the α-position proved beneficial (L11 to L13), with methyl-substituted L11 giving 65% yield. The isolated yield of β-lactone could be further improved to 73% when using TBHP in decane (see Supplementary Information for more optimization).

With the optimized ligand and conditions in hand, we next explored the scope of this methodology (Fig. 2). Aliphatic acids containing α-gem-dimethyl groups with various aliphatic chains including cyclobutanes (2f) were all compatible, affording the β-lactones (2a to 2f) in high yields. A range of functionalities such as fluoro (2g), chloro (2h), trifluoromethyl (2i), ketone (2j), and phosphoric ester (2k) were tolerated, with halogen (2h), ketone (2j) and phosphoric ester (2k) moieties serving as useful synthetic handles for subsequent derivatization. The lactone products containing a piperidine (2l) or a tetrahydropyran (2m) motif are especially valuable. Different protecting groups on the hydroxyl group including simple methyl (Me) (2n), benzyl (Bn) (2o), and methoxymethyl (MOM) (2p) were also well tolerated. Phenyl (2q to 2r) and phenyl ether (2s to 2v) groups were compatible with the TBHP system, and remained intact despite the potentially reactive aryl or benzylic C–H bonds. A range of substituents on the aryl ring from electron-donating (Me and O-alkyl) to electron-deficient (chloro, bromo, and nitro) groups were all well tolerated. Gemfibrozil (1v), an oral drug used to lower lipid levels[27], was converted to the corresponding β-lactone 2v in high yield. This lactone could serve as a versatile intermediate for library construction in medicinal chemistry (vide infra). Notably, the remaining α-methyl group from the above cases could then undergo further C–H functionalizations to afford greater structural diversity. Tertiary aliphatic acids containing a single α-methyl group (2w to 2ab) consistently afforded useful yields, in addition to those substrates containing α-hydrogens (2ac to 2ag).

Fig. 2.

Aliphatic acid scope for β-C(sp3)–H lactonization.

See Supplementary Information for details.

To demonstrate the scalability and practicality of this transformation, we conducted a gram-scale β-lactonization of Gemfibrozil (1v) with 1 mol% Pd (Fig. 3). Pure product was obtained by a simple aqueous wash without chromatography. 1.0 g Gemfibrozil (1v) in HFIP, Pd(OAc)2 (1.0 mmol%), commercially available MPAA ligand L6 (2.0 mmol%), and NaOAc (1.0 eq) were added to a reaction tube, followed by TBHP (70% in water) (2.0 eq). After stirring at 60 °C for 24 hours, the HFIP solvent was removed by evaporation, followed by dissolution with ethyl acetate, and washing with saturated NaHCO3 solution to remove unreacted acid, ligand, and metal complex. Evaporation of solvent delivered the lactone product 2v in 92% yield. From a practical standpoint, this reaction has several key advantages over other C–H activation protocols: (1) the inexpensive oxidant TBHP; (2) tolerant of air and moisture; (3) reliably scale-up; (4) aqueous wash delivers the product without chromatography.

Fig. 3.

Gram-scale β-C(sp3)–H lactonization of Gemfibrozil with 1 mol% Pd and diverse transformations.

Nu = Grignard reagents (3a to 3j), alkynylaluminum reagent (3k), TBACN (3l), TBAF (3m), MgBr2 (3n), NaN3 (3o), NaNHNs (3p), KOH (3q), PhSNa (3r). See Supplementary Information for details.

As depicted in Fig. 3, β-lactone product 2v is a stepping stone for mono-selective installation of a range of alkyl, alkenyl, aryl, alkynyl, cyano, halogen, amino, hydroxyl, and thiophenyl groups[28-30]. Various alkyl (3a to 3e), alkenyl (3f to 3g), and aryl (3h to 3j) Grignard reagents in the presence of catalytic copper were able to successfully open the β-lactone to build new C–C bonds at the β-position of the parent aliphatic acids[28,29]. In particular, secondary alkyl structure motifs such as isopropyl (3c), cyclopropyl (3d), and cyclopentyl (3e) could be efficiently installed; in contrast, the analogous secondary alkyl iodides are usually incompatible in Pd-catalyzed C–H alkylation reactions. β-Vinyl aliphatic acids (3f to 3g) were directly accessible through reaction with their corresponding vinyl (3f) and isopropenyl (3g) Grignard reagents, which provided a strategy complementary to the Pd catalyzed β-C–H olefination of free acids and their derivatives, where only electron-deficient olefins were effective. β-Lactone 2v may also be expediently elaborated into corresponding β-arylated aliphatic acids (3h to 3j); this approach is particularly strategic in the case of 3i and 3j, as the corresponding iodides are often not a viable coupling partner. The use of Grignard reagents prepared from readily available aryl bromides or chlorides is also a practical advantage. Additionally, β-phenylacetylene aliphatic acids 3k could be successfully synthesized from β-lactone 2v on treatment with alkynyl aluminum reagent. Cyanide could also open the lactone to construct a new C–C bond, affording the corresponding β-cyano aliphatic acids (3l). The electrophilicity of the β-lactone carbonyl was further exploited by the addition of the weak fluoride nucleophile (3m) to introduce a CH2F fragment, a highly sought-after bioisostere in medicinal chemistry. By a similar β-lactone opening, MgBr2 delivered the formally β-brominated aliphatic acid (3n) in high yield, a versatile linchpin for further elaboration. Further manipulations of the β-lactone in the presence of hard nucleophiles NaN3 or NaNHNs afforded coveted β-amino acid scaffolds (3o to 3p) in consistently high yields. By making use of the β-lactone as a masked aldol adduct, mild hydrolysis afforded the β-hydroxyl acid 3q in high yield. Finally, the formal β-chalcogenation product 3r was obtained in near quantitative yield using thiophenol sodium salt as a nucleophile.

Methods

General Procedure for β-C(sp3)–H Lactonization

In the culture tube, Pd(CH3CN)2Cl2 (10 mol%, 2.6 mg), ligand L11 (20 mol%, 2.9 mg), CsHCO3 (0.5 eq, 9.7 mg), and carboxylic acid 1 (0.1 mmol) in order were weighed in air and placed with a magnetic stir bar. Then HFIP (1.0 mL) and TBHP (ca. 5.5 M in decane) (2.0 eq, 36 μL) were added. The reaction mixture was stirred at rt for 3 min, and then heated to 60 °C for 12 hours (600 rpm). After being allowed to cool to room temperature, the mixture was concentrated in vacuo, and the resulting mixture was purified by preparative thin-layer chromatography or diluted with EA and washed with saturated NaHCO3 solution. Full experimental details and characterization of new compounds can be found in the Supplementary Information.

Data availability

The data supporting the findings of this study are available within the article and its Supplementary Information Files.