Wei
Zeng‡
ab,
Ai-Wen
Chen‡
ab,
Ming-Jie
Yan
ab and
Jie
Wang
*abc
aSchool of Chinese Materia Medica, Nanjing University of Chinese Medicine, Nanjing 210023, China. E-mail: jiewang@simm.ac.cn
bState Key Laboratory of Chemical Biology, Shanghai Institute of Materia Medica, Chinese Academy of Sciences, Shanghai 201203, China
cUniversity of Chinese Academy of Sciences, Beijing 100049, China
First published on 22nd November 2023
Construction of the sterically demanding Csp2(oS)–Csp3(T) bond was achieved by carrying out the Pd-catalyzed carboxylate-directed Mizoroki-Heck reaction under extra-ligand-free aqueous conditions. The cooperative role of the presence of water with the absence of phosphine ligand was proposed to accelerate the migratory insertion process considerably, delivering a broad substrate scope.
Fig. 1 (A) Directing group strategy for intermolecular Csp2(oS)–Csp3(T) bond formation. (B) Construction of the Csp2–Csp3 bond via the directivity of carboxylic acid: precedent and this work. |
Our reaction conditions were optimized with γ,δ-unsaturated carboxylic acid 1 and aryl halide 2 as model substrates (Table 1). Initially, the conditions reported by Shenvi and coworkers were attempted with 2-bromoanisole as the aryl coupling partner. However, the desired product 3 was not detected even in the slightest amount (entry 1).10e To our delight, by simply changing 2-bromoanisole to 2-iodoanisole (2a), 1 was delivered in 30% yield under the otherwise same conditions (entry 2). Ligand screening revealed that inclusion of PPh3 and DavePhos increased the yields slightly (entry 3) while P(o-Tol)3 and JohnPhos provided product in less than 10% yield (entry 4). Bidentate phosphine ligands were also tested and proved to be ineffective for this chemistry (see ESI† for details). Interestingly, we found that the reaction also proceeded to some extent in the absence of phosphine ligand, with a low yield of 15% obtained (entry 5). Under these extra-ligand-free conditions, other single solvents such as DMA, 1,4-dioxane, and DMSO all performed poorly (entry 6). Gratifyingly, a thorough solvent screen led us to discover H2O to be a crucial co-solvent for increasing the reaction yield significantly (entries 7–9), and the optimal DMF/H2O ratio was identified to be 5:1, providing 3 in 60% yield (entry 9). The H2O promotion phenomenon was also observed to be general, for several other solvents tested (entries 10–12 vs. entry 6). DMSO/H2O gave the same yield as did DMF/H2O (entry 12 vs. entry 9) and showed comparable efficiencies for other substrates except for strongly electron-deficient ones (see ESI† for a detailed list of substrates tested). Under the newly developed conditions, addition of phosphine ligands exhibited a deleterious effect on the reaction yields (entries 13 and 14). Finally, changing the temperature (entry 15), palladium source (entry 16) and base (entry 16) all delivered inferior results.
Entry | Ligand | Solvent | Yieldb |
---|---|---|---|
a Reaction conditions: 1 (0.1 mmol), 2a (1.5 equiv.), Pd2(dba)3·CHCl3 (3 mol%), ligand (12 mol%) if used, and K2CO3 (2.5 equiv.) were dissolved in solvent (0.6 mL) and heated to 100 °C under an Ar atmosphere. b Yield determined from a crude 1H NMR analysis using 1,3,5-trimethoxybenzene as an internal standard. c 2-Bromoanisole was used instead of 2a. d DMF (0.3 mL) and H2O (0.3 mL). e DMF (0.4 mL) and H2O (0.2 mL). f Isolated yield. g Reaction temperature was 90 or 110 °C. h Pd(OAC)2, PdCl2, Pd(PPh3)2Cl2, or Pd(PPh3)4 was used instead of Pd2(dba)3·CHCl3. i NaHCO3, Na2CO3, Cs2CO3, tBuONa, or Et3N was used instead of K2CO3. See ESI† for more details. | |||
1c | TrixiePhos | DMF | 0% |
2 | TrixiePhos | DMF | 30% |
3 | PPh3/DavePhos | DMF | 39%/33% |
4 | P(o-Tol)3, JohnPhos | DMF | <10% |
5 | — | DMF | 15% |
6 | — | DMA, 1,4-dioxane, DMSO | <5% |
7d | — | DMF/H2O (1/1) | 44% |
8e | — | DMF/H2O (2/1) | 52% |
9 | — | DMF/H 2 O (5/1) | 60% (58%) |
10 | — | DMA/H2O (5/1) | 32% |
11 | — | 1,4-dioxane/H2O (5/1) | 30% |
12 | — | DMSO/H2O (5/1) | 60% |
13 | TrixiePhos | DMF/H2O (5/1) | 36% |
14 | PPh3/DavePhos | DMF/H2O (5/1) | 39%/39% |
15g | — | DMF/H2O (5/1) | 47–51% |
16h | — | DMF/H2O (5/1) | 12–45% |
17i | — | DMF/H2O (5/1) | 15–45% |
With the optimized conditions in hand, we set out to investigate the substrate scope. It turned out that the new conditions developed above were generally effective across a range of substrates (>40 examples) as shown in Scheme 1. The aryl coupling partner scope was first investigated (Scheme 1A). Various ortho-substituents ranging from electron-rich to electron-poor ones including MeO (3, 4, 5, 7), Me (6), F (8), and carbonyl (9) were well tolerated. Even a highly hindered bis-ortho-substituted substrate provided product in 33% yield (7). Commercial aryl halides with other substitution patterns and different electronic characteristics were then evaluated. While a similar efficiency was observed for electron-rich and electron-neutral arenes (11–20) as was observed for the Shenvi conditions, a significant improvement in yields for electron-poor ones was obtained.10e For example, aryl bromides with strongly electron-withdrawing groups in the para position were previously reported to be unreactive, while that in the meta position was delivered in low yield. In contrast, under our conditions, these substrates all provided products in moderate to good yields (21–23). Moderately electron-deficient F- and Cl-substituted arenes also worked well, providing 24 and 25 in 54% and 71% yields, respectively. Various heteroaromatic substrates were also tested and showed improved efficiency. In this regard, electron-rich heterocycles such as benzothiophene (26), benzofuran (27), and either 2- or 3-thiophene with or without ortho-substitution (28–30) all gave moderate yields. Electron-poor ones such as quinoline (31) and chromone (32) also proved to be viable substrates under our conditions.
Next, several other γ,δ-unsaturated carboxylic acids were evaluated with o-OMe- or bis-o-OMe-substituted aryl coupling partners as shown in Scheme 1B. Yields of 37–55% were obtained consistently for all the 4–6-membered cyclic, heterocyclic, and noncyclic substrates (33–45). The ortho-benzyl ether protecting group was also tolerated, providing slightly lower yield than provided by the ortho-OMe counterpart (47% vs. 55% for 44 and 43). With a more hindered tetrasubstituted olefin (45), a useful yield of 37% was also obtained after slight modification to the original conditions including use of 10 mol% catalyst.
The successful implementation of this newly developed condition also relied largely on the judicious choice of aryl bromide or iodide precursors. Empirically, aryl iodides outperformed the corresponding aryl bromides for electron-rich and neutral substrates (1, 5, 6, 12, 13, 15, 17, 20, and 30), while aryl bromides worked better for electron-deficient ones (8, 23, 25). The generally moderate yields obtained throughout the investigation of substrate scope mainly stemmed from byproduct where the double bond migrated to be conjugated with the carboxylic acid through chain walking (typically 1/5–1/10 with regards to the shown products) and, in some cases, incomplete conversions.
While reactions with most of the substrates shown in Scheme 1 were performed on a 0.1 mmol scale, scale-up proved uneventful and was exemplified by substrates 8 and 44 on 0.5 and 1.0 mmol scales, respectively. They were then easily converted to known respective intermediates and natural products containing Csp2(oS)–Csp3(T) bonds, further demonstrating the utility of this new method (Schemes 2A and B). To this end, compound 47, a synthetic intermediate for S1P receptor modulator,11 was straightforwardly accessed from 8 in two steps and 64% overall yield by successive reduction of the alkene and carboxylic acid functional group. As another example, the sesquiterpenoid natural product himasecolone (49)12 was synthesized simply from 44 by carrying out a one-step Bn deprotection/alkene hydrogenation followed by transforming the carboxylic acid group to methyl ketone in 43% overall yield. In both cases, the strategic construction of the synthetically challenging Csp2(oS)–Csp3(T) moiety in an early stage provided extreme immediacy and simplicity for subsequent functional group manipulations.
Scheme 2 (A) Synthesis of drug intermediate 47 and (B) natural product himasecolone (49). (C) Proposed role of H2O and absence of bulky ligand. |
Based on previous reports of palladium-catalysed Mizoroki-Heck reactions in aqueous media,13 we arrived at an understanding of the mechanistic advantages of the current conditions (Scheme 2C). Oxidative addition and migratory insertion have often been regarded as rate-determining steps in Heck reactions.14 In the carboxylate-directed Heck reaction, we assumed the ArPd(II) carboxylate formation, coordination to alkene, and subsequent migratory insertion to be a relatively slow process, but one accelerated under our H2O-containing and ligand-free conditions.15 The H2O co-solvent not only increased the solubility of the inorganic base K2CO3 and the effective concentration of palladium carboxylate,16 but also promoted the migratory insertion to follow cationic mechanism by displacing the poisonous iodide or bromide anion from the coordination shell of palladium, the effect of which could be more pronounced under extra-ligand-free conditions.17 On the other hand, the omission of a bulky phosphine ligand also contributed significantly to this process by allowing for a free coordination site on the palladium center, reducing the steric congestion and thus increasing the rate of migratory insertion.18 While the sluggish migratory insertion step was accelerated substantially, the reaction showed better tolerance for the rate of oxidative addition, that is to say, a wide range of both aryl bromides and iodides with varied electronic characteristics could be used. In the case of electron-deficient substrates (9, 21–23, 31–32), our aqueous conditions free of phosphine ligand might have led to slower oxidative addition of aryl bromides,19 which probably better matched the accelerated downstream migratory insertion step and prevented the Ar-Pd(II) accumulation and deactivation process proposed by Shenvi and coworkers.10e
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3ob01784b |
‡ These authors contributed equally to this work. |
This journal is © The Royal Society of Chemistry 2024 |