Methodology design
To overcome the inherent limitation in traditional pyridine synthetic methods—direct functionalization of the pyridine ring and cyclization of acyclic intermediates, we propose a novel condensation approach involving 3-formyl(aza)indoles and various enamines. Enamines, critical intermediates generated in situ from terminal acetylenes using ammonium acetate, play a key role in this process (Fig. 2a). In generating enamine intermediates, electron-withdrawing groups (EWG) on terminal acetylenes serve as activators, promoting ammonia addition and enamine formation by stabilizing the partial negative charge at the α-carbon28. Meanwhile, our previous study indicates that EWG does not solely determine enamine formation29. NMR studies demonstrated that in situ-generated enamine preferred the cis-conformer over the trans-conformer, possibly rooted in the formation of resonance-assisted hydrogen bonding (RAHB) within the β-amino acrylate structure. RAHB, proposed by Gilli et al., involves intramolecular hydrogen bonding between proton donor and acceptor groups connected through a conjugated π-bonds system, allowing for additional stabilization of the entire structure30–32. Various synthetic methodologies have harnessed RAHB for controlling reactivity and facilitating synthon formation33. Our hypothesis posits that RAHB can expedite the formation of crucial enamine intermediates from terminal acetylenes with substituents possessing hydrogen bonding acceptors within the resonance system, a pivotal step in the synthesis of di-meta-substituted pyridines (Fig. 2b left).
To validate this hypothesis, we subjected 4 classes of terminal acetylenes to the model reaction condition utilizing 3-formyl-7-azaindole (1a) as the representative substrate (Figure S1). Extended reaction time (72 hours) was adopted to seize any minimal reaction conversion. Acetylenes having EWGs without RAHB potency (phenyl, 4-trifluoromethylphenyl, 4-nitrophenyl, 3-pyridyl, and 5-pyrimidyl; Figure S1a) mostly showed no or marginal reaction conversion. Even strong EWG, 4-nitrophenyl acetylene, showed only a limited conversion of 37%. On the other hand, acetylenes having EWGs with RAHB potency (2-pyridyl, 2-pyrimidyl, benzoyl, and diethyl phosphoryl; Figure S2b) fully converted 1a to the desired di-meta-substituted pyridines. In a comparison, acetylenes having substituents possessing hydrogen bonding acceptors without resonance (dimethyl aminomethyl, methoxymethyl, diethoxylmethyl, and 1-benzotriazolyl methyl; Figure S1c) did not undergo the desired transformation. It was not a big surprise to find acetylenes having electron-donating/neutral groups did not show any reactivity (4-methoxylphenyl, 4-dimethylaminophenyl, cyclohexen-1-yl, trimethylsilyl; Figure S1d). Notably, as shown in (Fig. 2b right), acetylenes with structurally analogous substituents exhibited divergent outcomes according to their RAHB potency; 3-pyridyl acetylene showed minimal conversion, whereas RAHB-forming 2-pyridyl acetylene exhibited complete conversion to the desired meta-pyridyl pyridine. Similar outcomes were observed in the case of 5-pyrimidyl vs. 2-pyrimidyl acetylenes. Our findings confirmed the necessity of RAHB in the designed methodology.
Acetylene substrate scope
In the pursuit of optimal reaction conditions, we employed 3-formyl-7-azaindole (1a) and diethyl ethynyl phosphonate (2f) as model substrates, exploring various parameters such as catalyst, substrate stoichiometry, and reaction time. The effectiveness of different catalysts was evaluated, focusing on Lewis acids, which could potentially enhance enamine formation by coordinating two Lewis bases within a 6-membered H-bonding network of enamine intermediates instead of the proton (Figure S2)34. A range of Lewis acids (entries A–E, Table S1), acetylene-activating metal catalysts (entries F–H)35, and Brønsted acids (entries I, J) were assessed, and zinc triflate emerged as the most efficient catalyst. Further exploration involving the equivalence of acetylene, catalyst loading, and reaction time determined that optimal conversion required two equivalents of acetylene at an elevated temperature for 16 hours (Table S2). Interestingly, increased zinc triflate loading did not improve the reaction conversion (entries I–K). The finalized optimized reaction conditions involved a 0.1 M ethanolic solution of 3-formyl(aza)indoles with 2.0 equivalents of terminal acetylene, 5.0 equivalents of ammonium acetate, and 10 mol% zinc triflate at 120 ℃ for the corresponding reaction time.
Under the optimized conditions, various acetylenes were employed in the designed transformation with 3-formyl-7-azaindole (1a) (Fig. 2c). Terminal acetylenes with keto analogs―ketone (2a), carboximidate (2b), sulfone (2c), sulfonamide (2d), phosphinate (2e), phosphonate (2f)―successfully underwent a transformation to yield the desired di-meta-substituted pyridines (3aa–3af). In the case of heteroaryl acetylenes, a complete reaction required 72 hours, except for pyrimidine analogs (2i, 2j, and 2j'). The prolonged reaction time could be attributed to the weaker electron-withdrawing capacities of heteroaryls compared to keto analogs36. Terminal acetylenes with 2-pyridyl (2g), 2-pyrimidyl (2i), and 4-pyrimidyl (2j) moieties facilitated RAHB formation upon enamine generation, successfully undergoing the desired transformation to yield corresponding meta-substituted pyridines (3ag, 3ai, and 3aj). In contrast, analogs lacking RAHB capability (2h and 2k) showed no conversion even after 72 hours. Remarkably, the acetylene (2j') containing 2-amino-4-pyrimidine, widely used in Imatinib analog syntheses37, produced the desired pyridine analog (3aj') in an excellent yield despite its 2-amino moiety. Terminal acetylenes containing electron-rich 5-membered heteroarenes―thiazole (2l), benzothiazole (2m), and benzimidazole (2n)―also proved compatible with our methodology, yielding the desired pyridines (3al–3an). The scalability of our methodology was validated through the gram-scale synthesis of 3ai, affirming its utility.
Comparatively, traditional methods involve complex steps for introducing functional groups at the meta position of pyridines. Our innovative methodology offers a streamlined approach, enabling the robust synthesis of diverse di-meta-substituted pyridines in a single step from corresponding terminal acetylenes—eliminating the need for precious transition metal catalysts or corrosive reagents38. While conventional approaches rely on heteroarene cyclization37 or palladium-mediated C–C bond formation39 for introducing heteroaromatics at the meta position of pyridines, our methodology provides a more efficient and practical alternative.
(Aza)Indole substrate scope
Considering reagent scalability and reaction kinetics, we focused our substrate scope study on two model acetylenes―phosphinate (2e) and 2-pyrimidine (2i)―in the context of 3-formyl(aza)indoles (Fig. 3). We utilized 2.0 equivalents for 2e but 1.2 equivalent for 2i, since the 1.2 equivalent of 2i was enough for the reaction completion within 16 hours of reaction time (Table S3). The condensation reaction of all regioisomers of 3-formylazaindoles (1a–1d) and 3-formylindole (1e) with 2e and 2i resulted in the formation of 3ae–3ee and 3ai–3ei, respectively, in moderate to excellent yields. Notably, all indole isomers featuring bromo substituents (1f–1i) demonstrated efficient conversion to the desired products (3fe–3ie and 3fi–3ii) in excellent yields. Emphasizing the utility of bromo substituents as sites for further modifications, the bromo group is challenging to keep intact in conventional metal-catalyzed coupling reactions, signifying the utility of this methodology.
Both electron-withdrawing nitro group (1j–1m) and electron-donating methoxy group (1n–1q) on various indole isomers proved amenable to this methodology, yielding the desired products (3je–3qe and 3ji–3qi) in good to excellent yields. Subsequently, we explored the substrate scope of N-substituted-7-azaindoles, encompassing electron-deficient benzene sulfonyl group (1r) to benzyl groups with diverse substituents (1s–1u). All substrates exhibited clean conversion to the desired products (3re–3ue and 3ri–3ui) in excellent yields. Remarkably, even direct N-pyrimidyl 3-formyl-7-azaindoles (1v) and N-ethyl 3-formyl-7-azaindole (1w) demonstrated excellent conversion to the desired products (3ve, 3vi, 3we, and 3wi) under the optimized reaction conditions. While no specific pattern of yields based on substrate electronics emerged, a trend indicated that 5- or 6-substituted indoles generally yielded lower than their regioisomers.
To validate the electronic effect of substrate reactivity, we conducted a competitive kinetic study between 5-nitro-3-formylindole (1l) and 5-methoxy-3-formylindole (1p) using ethynylphosphinate (2e) (Figure S3). The competitive reaction resulted in a full conversion of 1l to 3le, while 1p exhibited only a 10% conversion to 3pe, suggesting that electron-deficient (aza)indoles exhibit faster reaction kinetics than their electron-rich counterparts.
Synthetic applications
In our pursuit of broadening the applications of our developed methodology, we explored its potential in diverse synthetic scenarios, showcasing its versatility and utility (Fig. 4).
Late-Stage Core Remodeling of Natural Products: Utilizing our methodology's late-stage core remodeling capability, we targeted the transformation of dehydroanhydrolycorine, an indole-fused natural product (Fig. 4a). Dehydroanhydrolycorine was converted to the 3-formylindole derivative (4a) via the Vilsmeier-Haack reaction. Treatment of 4a with 2c, 2d, and 2e under standard conditions led to successful transformations, yielding new skeletons (5a, 5b, and 5c) in moderate yields. This approach demonstrates the potential to generate novel core skeletons while preserving the inherent biorelevance of natural products40.
Drug–Natural Product Conjugation: The developed methodology enabled introducing diverse functional groups at the meta position of pyridine while retaining a biologically relevant aniline moiety and prompted its application in drug–natural product conjugation (Fig. 4b). We exemplified this by directly conjugating benzocaine41, a local anesthetic, with myosmine, a naturally occurring nicotine alkaloid9, yielding their direct conjugate (5d). This strategy allows the creation of compounds with dual bio-relevant features in a single structure.
Late-Stage Modification for Structure-Activity Relationship (SAR) Studies: Late-stage functionalization of bioactive molecules is pivotal for SAR studies. Our methodology's high reactivity and excellent functional group tolerance make it suitable for late-stage modifications. Inspired by the reactivity of 4-pyrimidyl acetylene (2j'), we designed 2p as a terminal acetylene for late-stage conjugation of Nilotinib, a medication for treating chronic myelogenous leukemia. The condensation of 2p with 1a yielded an unprecedented Nilotinib analog (5e). Furthermore, the methodology was applied to 5-methyl-3-formyl-7-azaindole (4c) and 3-formyl benzofuran (4d), resulting in compounds (5f and 5g) with potential bioactive properties. In particular, 2-amino-5-methyl-pyridine moiety in 5f is an important starting material for synthesizing a well-known insomnia treatment, Zolpidem42, which can be a promising example of drug–drug conjugation. To note, 3-formyl benzofuran, another possible masked diformylmethane analog, successfully underwent the desired transformation and introduced the 2-hydroxyphenyl group on the Nilotinib, which proves the expandability of the designed transformation from diverse masked diformylmethane analogs.
Generation of Unprecedented Regioisomer of a Bioactive Molecule: To showcase the applicability of our methodology in regioisomer generation, we targeted CZC24832, a selective PI3Kγ inhibitor with a di-meta-substituted pyridine core11. Despite extensive SAR studies on CZC24832 and its analogs, the regioisomer 6h had never been achieved due to challenges in coupling ortho-aniline at the meta position of pyridines. The condensation reaction of 5-fluoro-3-formyl-7-azaindole 4e with sulfonamidyl acetylene 2d enabled the synthesis of the functionalized meta-bipyridine 5h, which underwent subsequent reactions to furnish the unprecedented regioisomer of CZC24832 (6h) in moderate yields (Fig. 4d). This compound retains all the bio-relevant motifs of CZC24832, offering a potentially bioactive molecule with a distinct substituent orientation.
Discovery of an anti-inflammatory agent
Inflammation, a crucial innate defense mechanism, safeguards the body against infections triggered by pathogens, allergens, and toxins, thereby contributing to overall homeostasis34. However, when inflammation becomes chronically abnormal, it poses a significant threat, causing severe tissue damage and contributing to various diseases such as sepsis43, neurodegenerative disorders44, autoimmune conditions45, and even certain cancers46. Consequently, the urgent need for effective anti-inflammatory agents with innovative mechanisms of action is evident. Non-steroidal anti-inflammatory drugs (NSAIDs), encompassing small-molecule drugs exhibiting anti-inflammatory effects, often feature an aniline moiety in their structures47. Meanwhile, pyridines are found in several anti-inflammatory drugs like epibatidine, piroxicam, and niflumic acid, which further underscores the potential of these heterocyclic compounds in combating inflammation48,49.
Motivated by the structural attributes of meta-substituted pyridines conjugated with anilines or aminopyridines, we conceived the exploration of novel anti-inflammatory agents with constructed bi-heteroaryl compounds (Figure S4a–d). Employing the Griess assay, a method for assessing nitric oxide (NO) production in live cells50, we screened 66 synthesized compounds in mouse macrophage RAW264.7 cells. This screening identified initial hit compounds (3gi, 3si, and 3ti) exhibiting NO inhibition in response to lipopolysaccharide (LPS), a powerful inflammation inducer. These hit compounds were named SB2031, SB2032, and SB2033, respectively, initiating a subsequent SAR study. Notably, SB2037, derived from 6-fluoroindole, emerged as the most potent NO inhibitor with minimal cytotoxicity (Figs. 5a and S4e,f).
To gain insight into the molecular mechanism of SB2037, we investigated the cellular inflammatory signaling pathway. Toll-like receptors (TLRs), particularly TLR4, play a pivotal role in recognizing pathogen-associated molecular patterns (PAMPs)51. TLR4 responds to gram-negative bacterial infections by recognizing LPS, triggering acute inflammation via the activation of mitogen-activated protein kinases (MAPKs). This cascade results in the nuclear translocation of nuclear factor kappa B (NF-κB) and the production of pro-inflammatory molecules, including NO, IL-6, and IL-1β (Fig. 5b). Confirming the cellular anti-inflammatory effects of SB2037, our study demonstrated reductions in IL-6 and IL-1β at both translational and transcriptional levels (Figs. 5c, 5d, and S5a–c). Additionally, SB2037 treatment effectively diminished the production of reactive oxygen species (ROS), as quantified by flow cytometry using the cell-permeable non-fluorescent probe, dichlorodihydrofluorescein diacetate (DCFH-DA) (Figs. 5e and f). Subsequent investigation into the molecular signaling pathway highlighted SB2037's capability to suppress the activation of MAPKs, including Jun N-terminal kinase (JNK), extracellular signal-regulated kinase (ERK), and p38, as evidenced by their decreased phosphorylation levels (Fig. 5g), followed by inhibiting NF-κB nuclear translocation (Figs. 5h and S5d,e). Our findings demonstrate SB2037's potent ability to restore cellular homeostasis by modulating TLR4-mediated acute inflammatory signaling pathways in macrophages.