Stimulated by the challenges of our synthetic target macrolactams, we first examined the module assembly by optimizing unactivated Csp3‒H olefination. 1-Bromo-2-(tert-butyl)benzene and 3-(4,4-dimethylpent-2-yn-1-ylidene)pentane-2,4-dione were selected as the model substrates and a number of reaction parameters such as base, ligand, Pd catalyst, and solvents were screened. After considerable experimentation, we were pleased to discover that a simple cocktail containing [PdCl(allyl)]2 (5 mol%), tBuXphos (30 mol%), and NaOAc in DMF at 100 °C in 1,4-dioxane established the reaction conditions, affording compound 3aa in 76% isolated yield with high stereoselectivity after 4 h (Table 1, entry 1). This Csp3‒H olefination is distinctive from Martin and co-workers’ recent work,42 in which they described an interesting Pd-catalyzed [4+1] cycloaddition of diazo esters. A series of control experiments were also conducted to validate the role of each parameter. Not surprisingly, the examined parameters were all essential for this transformation. The use of either DIPEA or KOAc did not further improve the yield of the desired product 3aa (Table 1, entries 2 and 3). Notably, the ligand appears to be important, as replacing tBuXphos with Xphos or Brettphos provided 3aa in a much lower yield and no reaction occurred in the absence of ligand (Table 1, entries 6, 7, and 8). In addition, a diminished yield was observed when Pd(MeCN)4(OTf)2 or Pd(OAc)2 was employed (Table 1, entries 9 and 10). The influence of the solvents was also investigated. While similar efficiency was obtained using DMA, only traces of product were obtained in THF and no detectable amount of 3aa could be found in acetonitrile (Table 1, entries 13 and 14).
After determining the optimal reaction conditions, we turned our attention to evaluate the scope of this Pd-catalyzed intermolecular unactivated Csp3‒H bond insertion reaction with ene-yne-ketones as donor/donor carbene precursors. As shown in Table 2, our Csp3‒H carbene olefination method turned out to be widely applicable regardless of the electronic variations at the para and meta positions on the aromatic ring of the aryl bromides (3aa-3ea). Likewise, the naphthyl bromide employed for the synthesis of 3fa served well as a partner in the reaction. Gratifyingly, functional groups on the tertiary alkyls including cyano and ester are compatible (3ga-3ha), although aryl, secondary, and primary alkyls are not reactive probably due to steric hindrance or β-H elimination.63-66 Particularly interesting was the observation that the aryl bromide substrate substituted with free amine did not interfere, providing 3ja in a good yield without traces of the N‒H bond carbene insertion product being observed. Remarkably, the ene-yne-ketones containing ketone, ester, and heterocyclic ring can be successfully transformed into corresponding products (3ab-3ad) in good to excellent yields (77-91%).
To evaluate the generality of the protocol, alternatively, we investigated this Csp3‒H carbene olefination process using allenyl ketones as donor/donor carbene precursors.47 As illustrated in Table 3, a wide range of allenyl ketones with electron-donating or -withdrawing substituents were well tolerated and a series of alkenes derivatives substituted with dihydrofurans were obtained. Generally, reactions of allenyl ketones with electron-donating substituents attached to a phenyl ring proceeded in higher yields than those having electro-withdrawing groups (5ab, 5ac, 5ad, 5ah). Moreover, the relative configuration of 5ab was unambiguously assigned by the X-ray crystal structure analysis. Particularly, substrates bearing furanyl and thienyl functional groups were also amenable to the standard conditions, which provided the pharmaceutical bis-heterocyclic compounds in decent yields with excellent stereoselectivities.
The identification of lead compounds greatly benefits from fragment-based drug design and the ability to directly modify the privileged scaffolds. Therefore, to highlight the potential application of these Csp3‒H bond carbene coupling reactions in medicinal chemistry, late-stage cyclization/olefination of complex and bioactive molecules was subjected to our established protocol. Remarkably, the alkenes substituted with furans and dihydrofurans products derived from Repaglinide, Isoxepac, Mycophenolic acid, Adapalene and Dehydrocholic acid were synthesized in moderate to excellent yields (Table 4). For example, Repaglinide, an antidiabetic drug used to control blood sugar in type 2 diabetes mellitus, had also been installed with 1-bromo-2-(tert-butyl)benzene and subjected into this protocol, gave access to the product 3ka in an excellent 92% yield. Notably, starting from Isoxepac, a non-steroidal anti-inflammatory drug with analgesic activity, which was successfully converted to new furan or dihydrofuran-containing Isoxepac (3la, 5lc) in 84% and 74% yield, respectively.
Once the crucial connection of the aryl bromides and enynone building blocks was successfully established, we next selected different natural or unnatural amino acids and attempted to assemble them to the macrolactams via a short and modular biomimetic strategy. With 6-8 steps, eight novel polysubstituted alkene-embedded macrolactams (6a-6h) were efficiently assembled (Scheme 1). To explore whether these alkene-embedded macrolactams could successfully exhibit pharmacologically relevant features, the macrolactams 6a-6h were investigated for the inhibitory effects on inflammatory mediators by LPS-induced inflammatory responses in RAW 246.7 macrophages. The results showed that 6g exhibited prominent inhibitory effects on the production of TNF-α, IL-6, and IL-1β with IC50 values of 0.45, 1.59, and 0.59 μM, respectively. It should be noted that these pro-inflammatory cytokines are critically involved in the process of inflammation, immunity, cell survival and apoptosis, and metabolic diseases.67-69 Both 6g and 6h were approximately 10 times more potent in the inhibitory activity on IL-6 than the drug Dexamethasone, the widely used corticosteroid medication to relieve inflammation (see the Supporting Information). More importantly, they did not show obvious cytotoxicity at the indicated concentrations compared to Dexamethasone. We further examined the effects of 6g on the levels of phosphorylation of NF-κB and IκB-α induced by LPS in RAW 246.7 cells. As expected, 6g could abrogate the phosphorylation of NF-κB and IκB-α, an NF-κB inhibitory protein, whose phosphorylation results in its degradation and promotes subsequent translocation of NF-κB into nucleus and transcription of inflammatory genes (Figure 2B).70 The current preliminary pharmacological results indicated a promising prospect of 6g to be developed as a novel anti-inflammatory agent, with competitive potency and safety advantage.
Apart from the scope of these conversions and intriguing anti-inflammatory activities, we were also interested in the reaction mechanism. Two possible catalytic cycles are shown in Scheme 2. To gain insight into the proposed catalytic cycles and see which cycle is more favorable, we carried out DFT calculations to investigate the detailed mechanism. Although the similar mechanistic pathways have been proposed by others,42 the DFT calculations of Pd(II) shift are still uncovered to date.
Figure 3 shows the energy profiles calculated. Considering the sizes of ligand and substrate, we start with the complex A in which the Pd (0) metal center coordinates with the ligand L and the substrate aryl bromide [Figure 3(a)]. Oxidative addition (OA) followed by concerted metalation deprotonation (CMD) gives the key palladacycle intermediate IM4. The barrier for the OA is small while the barrier for the CMD process is 33.4 kcal/mol.
From the key palladacycle intermediate IM4, two possible paths (consideration of the two cycles shown in Scheme 2) were calculated (Figure 3(b)). Path A involves alkyne-activation cyclization followed by migratory insertion (Cycle A in Scheme 2), while Path B engages protonation first and then alkyne-activation cyclization (Cycle B in Scheme 2). Clearly, Path A requires to pass through a very high-lying transition state (TS6-7A) for the migratory insertion. The high-lying TS6-7A structure is a result of the unfavorable migration step that involves weakening/breaking of the two strong Pd-C bonds in the 5-membered ring of IM6A.
Figure 3(c) shows that when the migration insertion occurs on the carbene structure IM7B without a 5-membered ring moiety, a very small barrier of 7.5 kcal/mol is calculated. After the migratory insertion, which is highly exergonic, β-hydride elimination occurs easily (almost barrierless), followed by reductive elimination and ligand (aryl bromide) coordination to regenerate the active species A.
The calculation results suggest that Cycle B is favorable. From Figure 3, we can also see that the CMD transition state structure TS2-3 (Figure 3(a)) and the protonation transition state structure TS4-5B (Path B in Figure 3(b)) show similar stability, although the latter lies slightly higher than the former. On the basis of the results, TS4-5B (Path B in Figure 3(b)) is the rate-determining transition state, and therefore, the overall reaction barrier is 34.1 kcal/mol, corresponding to the energy difference between IM1 and TS4-5B. The calculated overall energy barrier is moderately high, which is understandable in view of the fact that the reaction temperature is 100 °C. In Figure 3, the series of transformation from IM2 to IM5B corresponds to a 1,4-Pd-shift.
Apart from all of the calculations mentioned above, we also calculated a pathway, which is closely related to Path A but starts from IM4B (instead of IM4) to react with 2a. The calculation results (Figure S1) indicate that this pathway is slightly favorable than Path A, but still less favorable than Path B.