Ligand-Enabled Reactivity and Selectivity in a Synthetically Versatile Aryl C–H Olefination

Abstract

The Mizoroki–Heck reaction, which couples aryl halides with olefins, has been widely used to stitch together the carbogenic cores of numerous complex organic molecules. Given that the position-selective introduction of a halide onto an arene is not always straightforward, direct olefination of aryl C–H bonds would obviate the inefficiencies associated with generating halide precursors or their equivalents; however, methods for carrying out such a reaction have suffered from narrow substrate scope and low positional selectivity. Here we report an operationally simple, atom-economical, carboxylate-directed Pd(II)-catalyzed C–H olefination reaction with phenylacetic acid and 3-phenylpropionic acid substrates, using oxygen at atmospheric pressure as the oxidant. The positional selectivity can be tuned by introducing amino acid derivatives as ligands. We demonstrate the versatility of the method through direct elaboration of commercial drug scaffolds and efficient synthesis of 2-tetralone and naphthoic acid natural product cores.

Unactivated carbon–hydrogen (C–H) bonds are among the simplest and most common structural motifs in naturally occurring organic molecules, and as such, they are ideal targets for chemical transformations. Though C–H bonds are generally unreactive, during the past several decades transition metal catalysis has emerged as an effective means of converting unactivated C–H bonds into carbon–heteroatom and carbon–carbon (C–C) (1–5). This technology has proven to be valuable in natural products synthesis, where several distinct C–H functionalization strategies have been exploited (6–12).

Traditionally, C–C bonds between aryl and olefinic fragments have been forged through the Pd-catalyzed Mizoroki–Heck reaction, which couples aryl halides or triflates with olefins (Figure 1.a). Considering the prominence of this transformation in organic synthesis (13), Pd-catalyzed olefination of aryl C–H bonds has the potential to emerge as a powerful platform for more direct access to carbogenic cores of complex molecules (Figure 1.a and 1.e), particularly in cases in which the position-selective introduction of a halide is problematic. However, to date the few pioneering examples of Pd-catalyzed C–H olefination in total synthesis are restricted to specific cases, generally including electron-rich heterocycles, such as indoles and pyrroles, and/or stoichiometric palladium (14–17). The development of Pd-catalyzed arene C–H olefination reactions to provide general routes to commonly occurring carbogenic motifs thus remains an outstanding challenge. This limitation is rooted in two interrelated problems. Firstly, the substrates that are typically effective in palladium-catalyzed C–H activation are synthetically restrictive, either because they are limited to electron-rich arenes or heterocycles, or because they posses impractical chelating functional groups to promote metalation. These directing groups include those that are irremovable and recalcitrant to undergo further synthetic elaboration, such as pyridine, and those that are removable, but require several steps for installation and detachment, such as oxazoline (5). Secondly, methods for effecting position-selective C–H activation on multiply substituted arenes (18, 19), particularly via ligand control, remain underdeveloped.

Figure 1.

(a) Comparison of the Mizoroki–Heck reaction and arene C–H olefination. (b) Schematic depiction of our position-selective C–H activation approach. (c) Substrates that were found to be unreactive under our original conditions but could be efficiently olefinated in the presence of amino acid ligands. (d) Non-steroidal anti-inflammatory drugs (NSAIDs) that can be directly ortho-olefinated. (e) Natural product components that can readily be synthesized using position-selective C–H olefination.

Herein, we report a catalytic system that addresses these problems. A practical Pd-catalyzed reaction using molecular oxygen as the terminal oxidant has been developed to perform C–H olefination with synthetically useful phenylacetic acid and 3-phenylpropionic acid substrates. This reaction has remarkably broad substrate scope, which is enabled in part by the use of amino acid ligands to enhance the reactivity of the catalytically active Pd(II) species. Moreover, two distinct methods to control the positional selectivity of C–H activation with multiply substituted arenes have been demonstrated: substrate control, by appending a protecting group (PG) to modulate the intrinsic steric bias; and catalyst control, by using ligands to alter the steric and electronic properties around the metal center (Figure 1.b). The versatility of this olefination reaction is demonstrated by direct functionalization of commercially available drugs (Figure 1.d) and by atom- and step-economic syntheses of 2-tetralone derivatives (2) (key synthetic intermediates for tetraline based natural products) and two challenging naphthoic acid components in neocarzinostatin (1) and kedarcidin (3), highly active antibiotics (Figure 1.e).

As part of the overarching goal of developing practically useful C–H functionalization reactions, our laboratory has focused on discovering reactivity with broadly useful substrates, in which all moieties present in the starting material can be used for subsequent synthetic applications in an atom-economical manner. Following this philosophy, we sought to develop a Pd-catalyzed olefination protocol (20) for phenylacetic acid substrates, as the resulting carbon skeletons are well-established platforms for the synthesis of 2-tetralones and naphthoic acids. Although carboxy-directed C–H activation involving six atoms in the coordination assembly is rare, we hypothesized that our K⁺-promoted Pd insertion procedure for benzoic and phenylacetic acids (21, 22), which promotes C–H activation through the complex-induced proximity effect (CIPE) (23), could be exploited for subsequent olefination.

To this end, we began by extensively screening reaction conditions, and optimizing with respect to solvent, inorganic base, temperature, and oxidant (see supporting online material). Gratifyingly, we found that 4-methoxyphenylacetic acid could be coupled with ethyl acrylate in the presence of 5 mol% Pd(OAc)₂ to give the desired product in high yield (Figure 2, 6a₁). As a further practical advantage, 1 atm of O₂ could be used as the terminal oxidant, with 5 mol% benzoquinone (BQ) serving as a ligand to prevent minor formation of the meta- and di-ortho-olefinated products.

Figure 2.

C–H olefination of phenylacetic acid substrates with ethyl acrylate (6a₁, 6b₁, and 6c to 6s) and with other olefin coupling partners (6a₂ to 6a₅ and 6b₂).

A wide range of phenyl acetic acid substrates were found to be compatible with this protocol (Figure 2). Products containing chlorides (6g, 6h, 6j, 6k, and 6p), fluorides (6f and 6r), and ketones (6l) could be obtained in high yields. Notably, the tolerance for chlorides on the aromatic ring in this ortho-olefination offers an opportunity for subsequent cross-coupling, facilitating expedient synthesis of highly complex biaryl molecules. Several drug substrates, including ketoprofen (4l), ibuprofen (4n), and naproxene (4o) were found to be compatible with this protocol, affording the respective olefinated products 6l, 6n, and 6o. These results demonstrate the potential for applying C–H olefination to effect direct functionalization of existing bioactive molecular scaffolds in the interest of enabling drug diversification for medicinal chemistry.

A diverse array of substitution patterns at the α-position were tolerated, with a higher degree of α-substitution corresponding to higher activity due to the Thorpe–Ingold effect (6j to 6s). Optically pure compounds with chiral centers at the α-position were found to racemize slightly under our reaction conditions; for example 6o was formed in 72% ee. In this case, the use of Li₂CO₃ as the base prevented racemization, affording the product in 97% ee but also lowered the conversion to 24%.

A variety of different olefin coupling partners were found to react well using this catalytic system (Figure 2, 6a₂ to 6a₅ and 6b₂). Interestingly, using 1-hexene as the olefin, we found that non-conjugated product 6a₅ was formed predominantly (with a E:Z ratio of 5:1), over the thermodynamically favorable conjugated system (observed in only trace amounts). Mechanistically, this suggests that following 1,2 migratory insertion of the Pd–aryl moiety into the olefin, the resulting intermediate is conformationally restricted from undergoing subsequent β-hydride elimination with the benzylic hydrogen atoms. A possible explanation is that the carbonyl oxygen atom from the carboxylate group remains coordinated to palladium in an unusual eight-membered ring, which restricts the bond rotation necessary to align the metal for syn elimination. Synthetically, olefinated products such as 6a₅ represent a class of substrates beyond the scope of traditional Mizoroki–Heck-type chemistry.

With this highly efficient catalytic C–H olefination protocol in hand, we next sought to control the positional selectivity of this reaction with multiply substituted arenes that yielded isomeric product mixtures under our original conditions (Figure 1.b). For instance, 3-methyl-5-methoxyphenylacetic acid (7) reacted to form a mixture of positional isomers without significant bias (1.4:1, 8-A:8-B) (Figure 3.a, Entry 1). We hypothesized that tuning the steric properties of the metal center through the coordination of a ligand could provide the site-recognition that we sought. Extensive screening of ligands led us to discover that mono-N-protected amino acids were effective in this respect (24). During our screening efforts, we observed that the resulting isomeric distribution was substantially dependent on the structure of the amino acid side chain, with Boc-Leu-OH and Boc-Ile-OH giving the best selectivity (7:1 and 8:1, respectively) (Figure 3.a, Entries 5 and 6). We were pleased to find that positional selectivity could be further enhanced by varying the N-protecting group, with Formyl-Ile-OH as the best ligand, giving a 20:1 ratio of positional isomers (Figure 3.a, Entry 12). The conversion using this particular ligand was slightly lower than without it, but by increasing the catalyst loading to 7%, the conversion could be improved to 75%. Allowing the reaction to run for 96 h further improved the conversion to 89%.

Figure 3.

(a) Selected amino acid ligand screening data for position-selective C–H olefination. (The full ligand screening data are available in Table S8.) (b) Substrate scope for ligand-controlled position-selective C–H olefination. (c) Ligand-enabled C–H olefination with 2 mol% Pd(OAc)₂. (d) Amino acid ligand-enabled C–H olefination with problematic substrates. (Unless otherwise noted, the reaction conditions in 3.b to 3.d are identical to those described in 3.a.)

This ligand-controlled position-selective C–H olefination protocol could also be applied to other multiply substituted arene substrates including natural product precursors such as 9 and drug substrates such as flurbiprofen (11) (Figure 3.b). With substrate 9, because the two ortho-C–H bonds are approximately electronically equivalent, the outstanding positional selectivity observed with Boc-Ile-OH is likely a result of the catalyst’s recognition of the different steric environments. In contrast, both steric and electronic properties could be contributing to the improvement in positional selectivity with substrates 7, 11, 13, and 15; the mechanistic details in these cases remain to be fully elucidated.

To our delight, during this investigation, we also discovered that certain Boc-protected amino acid ligands could dramatically improve the yield in this olefination reaction (see supporting online material), with the optimal ligand choice highly dependent on the combination of substrate and coupling partner. For example, using Boc-Ile-OH, olefinated products 6t and 17 could be formed quantitatively, even when the catalyst loading was reduced to 1 to 2 mol% of Pd(OAc)₂ (Figure 3.c). Moreover, the use of amino acid ligands allowed for efficient di-olefination, fashioning product 17, for example, in quantitative yield. Thus, our parent C–H olefination protocol and the amino acid-ligated system offer complementary reactivity to access either the mono- or di-olefinated products, depending on which is more useful for a given synthetic application (Figures 2 and 3.c, respectively).

The strong ligand influence on reactivity in this system encouraged us to examine whether previously unreactive 3-phenylpropionic acid and electron-deficient phenylacetic acid substrates could now be ortho-olefinated in the presence of amino acid ligands. Indeed, we were pleased to find that an array of such substrates could be olefinated in moderate to good yields (Figure 3.d). Notably, the synthetic flexibility afforded by the extra methylene spacer in 18a and 18b greatly expands the range of core structures that can subsequently be accessed.

Finally, we demonstrated the synthetic utility of this highly versatile and scalable olefination reaction by concise syntheses of several natural product core structures, including 2-tetralones and naphthoic acids (Figure 4). The synthesis of natural products of this type is by no means simple, but due to impressive efforts by several groups (25–29), a basic roadmap for how to construct such molecules has emerged. In particular, olefinated arenes with structures analogous to 10-A (Figure 4.c) often serve as key intermediates, and the olefinic moieties are usually introduced via a Wittig reaction with the corresponding aldehyde or via bromination in the early stages to set up a late-stage Mizoroki–Heck reaction. Our C–H olefination reaction offers a departure from these approaches both in the retrosynthetic sense (in that it represents a different synthetic disconnection associated with a distinct pool of starting materials), and in the forward sense (in that the key transformations are highly step- and atom-economical).

Figure 4.

(a) Synthesis of 7,8-dimethoxytetalin-2-one. (b) Synthesis of the naphthoic acid component of neocarzinostatin (1). (c) Synthesis of the naphthoic acid component of kedarcidin (3).

Using our original protocol (Figure 2), commercially available 2,3-dimethoxyphenylacetic acid (4d) could be transformed to intermediate 6d₂ in high yield. A straightforward sequence of hydrogenation, ester formation, Dieckmann condensation, and decarboxylation yielded 2-tetralone derivative 19 (Figure 4.a) (29). Notably, nearly all substrates described in this work could be used to synthesize a diverse range of 2-tetralones.

Next, we demonstrated the utility of both protecting group- and ligand-controlled position-selective olefination by synthesizing the naphthoic acid components of both neocarzinostatin (1) and kedarcidin (3) (Figure 4). The use of a TIPS group afforded the desired positional selectivity for the preparation of 21-A (Figure 4.b). This key intermediate could then be cyclized in accordance with literature precedent (28) to give the naphthoic acid component 22 of neocarzinostatin (1) (26). On the other hand, the ligand-controlled position-selective olefination of 9 afforded 10-A in high yield (Figure 4.c). 10-A could then be converted into the desired naphthoic acid product (23) following a two-step procedure developed by Myers and Hirama: formation of the acid chloride and [6π]-electrocyclization (25, 27). In all three of these cases, the reaction sequences are among the highest yielding and most atom- and step-economical routes achieved to date for accessing these widely studied, commonly occurring core structures.

In this report we have attempted to demonstrate the importance of ligand development (30) for enabling unique reactivity and selectivity in C–H activation and at the same time, to illustrate two conceptual prerequisites for widespread application of Pd-catalyzed C–H activation in organic synthesis: (1) versatile substrates and coupling partners and (2) precise control of positional selectivity in C–H functionalization. We anticipate that this C–H olefination reaction and others grounded in this philosophy will find broad applicability in multifarious synthetic endeavors.