Main

Ring systems are key pharmacophores in drugs and bioactive natural products and have found high prevalence among more than 95% of marketed drugs1,2. Two-dimensional (2D) aromatic rings such as benzene and pyridine rank among the most commonly occurring scaffolds3,4. The three-dimensional (3D) aliphatic 3-, 4-, 5- and 6-membered rings follow behind but have recently attracted increasing attention in drug discovery (Fig. 1a) due to their modest molecular weight, improved solubility and physicochemical profile5,6. Furthermore, this poorly populated chemical space offers unmatched opportunities in skeletal diversification at the core of diversity-oriented synthesis7. In this context, fused ring systems with hybrid 2D and 3D fragments, possessing unique structural and physicochemical properties, have emerged as important scaffolds in medicinal chemistry (Fig. 1b). For example, varenicline, based on a hybrid 2D/3D fused 6−6−5−6 ring, has been marketed to treat smoking addiction and MK-8886, based on a hybrid 2D/3D fused 6−5−3 ring, has been invented by Merck for the treatment of type 2 diabetes mellitus and has recently entered clinical trials8. Additionally, many other compounds containing fused 2D/3D rings are currently under pre-clinical validation or clinical trials9,10,11,12,13,14. However, the preparation of such polycyclic molecules—especially those featuring high ring strain—often exhibits low efficiency and often requires elaborate synthetic efforts, which is a rate-limiting factor in drug discovery and development15,16. To this end, the advancement of general and straightforward synthetic methods towards fused 2D/3D rings featuring readily available feedstock resources, good generality and high reaction efficiency/selectivity is of great interest.

Fig. 1: Overview of 2D/3D scaffolds.
figure 1

a, Occurrence of selected rings in drugs. b, Examples of bioactive compounds and natural products with fused 2D/3D rings. c, Comparison between conventional cycloadditions and cascade dearomative cycloaddition/rearrangement (CDACR) reactions. Int., intermediate; Pro., product; Sub., substrate; FDA, US food and drug administration.

Full size image

The de novo synthesis of ring systems is often based on cycloaddition reactions. The exergonic cycloadditions of 1,3-dienes with alkenes, capable of generating various ring systems, are amongst the most fundamental transformations in synthetic chemistry17,18. Recently, this classical realm has witnessed rapid advancements owing to the introduction of aromatics as reactants by the visible-light energy-transfer process19,20,21,22,23,24,25,26,27,28,29,30,31,32,33. Stemming from our reported [4 + 2]-dearomative cycloaddition between aza-arenes and alkenes29, we questioned whether the established reactivity could be diverted towards a cascade dearomative cycloaddition/rearrangement reaction forming unusual fused 2D/3D rings (Fig. 1c). In this context, quinolines are recognized as suitable reactants due to the potentially divergent ortho/para/meta cycloaddition reactivity and the innate presence of pharmaceutically relevant N-heterocycles. More specifically, the carbocyclic ring of quinolines is initially proposed to couple with alkenes in a [2 + 2] fashion under energy-transfer conditions. This endergonic process will lead to an enthalpy increased and more reactive vinylpyridine intermediate, which could easily undergo further cycloadditions or rearrangements by a consecutive energy transfer, thus leading to diverse sp3-carbon-rich ring systems34. This proposed cascade dearomative cycloaddition/rearrangement (CDACR) reaction offers an ideal approach towards pyridine-fused 2D/3D rings. It is worth noting that, despite the fact that the intramolecular photochemical [2 + 2] cycloaddition of a benzenoid ring has been disclosed35,36,37 and its great potential was demonstrated in natural product synthesis38,39,40,41, the intermolecular variants—that are obviously more challenging—are rarely investigated in modern synthetic chemistry. In addition, provided that the transformation could proceed, this reaction will further encounter major regio- and diastereo-selectivity issues, as inferior diastereocontrol is a general problem in the intermolecular [2 + 2] cycloaddition of various aromatics, such as benzothiophene22 and indole42,43. Overall, from the perspective of both synthetic method development and its potential contribution to drug discovery, employing aromatics in cycloadditions and harnessing a consecutive transformation could lead to intriguing domains of chemical space, in compliance with the idea of diversity-oriented synthesis and skeletal diversity44,45,46,47,48,49,50. In this Article, we report the realization of this concept by two types of highly efficient and selective intermolecular cascade dearomative cycloaddition/rearrangement reaction of quinoline derivatives with alkenes. These reactions commence with an energy transfer (EnT)-enabled [2 + 2] cycloaddition of quinolines, with chloro or ester substitutions at the benzenoid rings, and are followed by either a second EnT-enabled cyclopropanation or an EnT-enabled cyclobutane rearrangement. The two consecutive EnT processes provide a straightforward approach for accessing synthetically challenging pyridine-fused polycyclic rings. The pharmaceutically relevant 3D 6-, 5-, 4- or 3-membered rings could be directly elaborated through one-shot but cascade transformations.

Results

Reaction development

In line with the previous report on dearomative [4 + 2] cycloaddition of quinolines with alkenes29, this work started with 6-chloroquinoline (1a) and 2-chloropropene (2a), in the presence of the photosensitizer [Ir(dF(CF3)ppy)2(dtbbpy)][PF6] (Ir-F, 2 mol%) (dF(CF3)ppy, 3,5-difluoro-2-[5-(trifluoromethyl)-2-pyridinyl-N]phenyl-C; dtbby, 4,4′-di-tert-butyl-2,2′-bipyridyl) and HCl (2 equiv.), using 1,1,1,3,3,3-hexafluoropropan-2-ol (HFIP) as solvent51, under the irradiation of blue light-emitting diodes (LEDs). As a result, a structurally unique pyridine-fused 6−5−4−3 ring system (3a), which features five stereocentres—of which two are contiguous quaternary—was produced in good yield (78% yield) and with high diastereoselectivity (92:8 diastereomeric ratio (d.r.)), whereas the [4 + 2] cycloaddition product was not observed (Fig. 2a). Notably, the substrates 1a and 2a and the photosensitizer Ir-F are all commercially available; conversely, the produced highly strained fused ring 3a is hardly accessible through conventional synthetic methods. To enhance the reproducibility of this reaction and improve the user-friendliness, key reaction parameters were examined to assess their influence on the yields52. Accordingly, the concentration of substrates, moisture and scale-up did not show any notable influence on the outcome. By contrast, the reaction was negatively affected with high oxygen levels, high temperature or low light intensity. Next, methyl quinoline-8-carboxylate (1b) and vinyl acetate (2b) were subjected to the same reaction conditions thus furnishing a new type of polycyclic product, a fused 6−4−6 ring system (4) with excellent yield (97%) and as a single diastereoisomer (Fig. 2b). Two contiguous all-carbon quaternary centres as the bridgeheads of the 3D 4- and 6-membered rings are efficiently generated in this reaction. Control experiments revealed that both the visible light and the photosensitizer were mandatory for these two transformations.

Fig. 2: Dearomative cycloadditions of quinoline derivatives with alkenes.
figure 2

a, Cascade dearomative [2 + 2] cycloaddition/cyclopropanation reaction and the rapid assessment for its key reaction parameters. b, Cascade dearomative [2 + 2] cycloaddition/rearrangement reaction. [Ir-F] = Ir[dF(CF3)ppy]2[dtbbpy](PF6) (CAS No. 870987-63-6). HFIP = 1,1,1,3,3,3-hexafluoropropan-2-ol. One representative enantiomer of the racemic product is presented for all throughout the text. Ac, acetyl; [A], acid additive; c, concentration; d.r., diastereomeric ratio; EnT, energy transfer; I, light intensity; T, temperature.

Full size image

Mechanistic investigation

With respect to the reaction in Fig. 2a, this transformation was likely to be initiated by an energy-transfer-mediated dearomative [2 + 2] cycloaddition between 1a and 2a (ref. 53). This gave the cinnamyl chloride analogue intermediate I (Fig. 2a), which was also detected by NMR spectroscopy. A subsequent homolytic bond dissociation of C−Cl by a second energy-transfer event furnished a triplet radical pair (from II to III). The following C−C and C−Cl bond formations furnished the terminal product 3a. Using 6-fluoroquinoline instead of the 6-chloro congener 1a, a double [2 + 2] cycloaddition product 3b was isolated, which provided strong evidence for the cascade energy-transfer process (Fig. 2a). Notably, all the ring formation processes proceeded in a highly diastereoselective fashion.

We investigated the mechanism of the [2 + 2] cycloaddition/cyclopropanation sequence by performing density functional theory calculations on the reaction between 6-chloroquinoline (1a) and 2-chloropropene (2a) (Fig. 3). Energy transfer from the excited photosensitizer to protonated quinoline 1A yields the triplet 3A, which readily reacts with 2-chloropropene (2a, labelled as B in Fig. 3 for clarity) via 3TS-1 on the triplet surface to form the first C–C bond with an activation free energy of 12.0 kcal mol−1. As the open-shell singlet (OSS) surface lies slightly (1–2 kcal mol−1) below the triplet (T) surface in energy in the vicinity of the resulting biradical intermediate 3C, 3C is most likely to undergo intersystem crossing (ISC) to the OSS biradical 1C. The OSS surface crosses with the closed-shell singlet (CSS) surface at two minimum-energy crossing points (MECPs): MECP-1, which leads back to separated reactants and MECP-2, which furnishes the formal [2 + 2] cycloadduct. The chemo- and diastereo-selectivity of the reaction are therefore controlled by the rate of ISC and the OSS→CSS MECPs.

Fig. 3: Calculated energy diagram.
figure 3

Values are in kcal mol−1, obtained at the ωB97X-D/def2-TZVPP, SMD(HFIP)//ωB97X-D/def2-SVP, SMD(HFIP) level of theory for the [2 + 2] cycloaddition/cyclopropanation reaction sequence between 6-chloroquinoline and 2-chloropropene. T, triplet; CSS, closed-shell singlet; OSS, open-shell singlet; MECP, minimum-energy crossing point; alt, alternative diastereomer.

Full size image

The second C–C bond formation through MECP-2 yields [2 + 2] cycloadduct 1D, which is 1.4 kcal mol−1 more stable than its diastereomer 1D-alt. Energy transfer to adduct 1D under the reaction conditions gives rise to the triplet 3D, which can either fragment via 3TS-2 to revert back to 3C or undergo homolytic C–Cl cleavage via 3TS-3 (ref. 54). The latter process is favoured with a low activation free energy of 5.4 kcal mol−1, compared to 12.4 kcal mol−1 for the C–C fragmentation. The resulting triplet radical pair 3E is more stable than its diastereomer 3E-alt as the chlorine atom prefers to stay at the less sterically congested convex face of the fused 6-4 ring system (Supplementary Fig. 13 and Supplementary Table 5). Finally, 3E collapses to the CSS product 1F through MECP-3, forming both the cyclopropane ring and the C–Cl bond. Due to the loss of aromaticity and the strained cyclopropane structure, 1F is 29.8 kcal mol−1 higher in free energy than the starting materials 1A and B. Despite this, the formation of 1F is irreversible as the triplet state from 1F is too high in energy to be formed with the triplet sensitizer used for these reactions.

The reaction in Fig. 2b of methyl quinoline-8-carboxylate (1b) with vinyl acetate (2b) was proposed to proceed through a conceptually similar energy-transfer-mediated dearomative [2 + 2] pathway, leading to a vinylcyclobutane (IV). A cascade energy-transfer process triggers the ring rearrangement to produce a kinetically stable fused ring 4.

Scope and limitation

Next, we evaluated the generality of these two cascade dearomative [2 + 2] cycloaddition/rearrangement reactions. With respect to the [2 + 2] cycloaddition/cyclopropanation, commercially available 2-chloropropene derivatives were first examined and provided the corresponding fused 6−5−4−3 rings (3, 5 and 6) in good yields and diastereoselectivities (Fig. 4). An array of single-step prepared 2-chloroalkenes are compatible, demonstrating an excellent functionality tolerance, such as esters (7–11), an ether (12), fluorinated alkyl and aryl moieties (11–13), a sulfonamide derivative of the drug probenecid (14), a biphenyl (15), a benzenesulfonyl compound (16), a malonate (17), a pyridine (18) and an amide (19). 1,1-Dichloroethylene and trichloroethylene also worked smoothly thus providing polychlorinated products 20 and 21 with good results, whereas no conversion was observed using tetrachloroethylene (Supplementary Fig. 14). Comparable reaction outcomes were observed by using bromo-substituted alkenes (22–25). As shown in Fig. 5, the quinoline scope of this [2 + 2] cycloaddition/cyclopropanation reaction was further evaluated. As a result, 6-chloroquinolines with additional substituents at any of the 2-, 3- or 4-positions exhibited excellent compatibility (26–36). The topology of the fused ring products was expanded by using tricyclic quinoline-derivative substrates (37–39). Notably, introducing one more substituent at the 5-, 7- or 8-position of 6-chloroquinoline did not hamper the reaction efficiency, thus furnishing the fused 2D/3D rings with three (40 and 41) or four (42) quaternary centres. For instance, 42 with one all-carbon quaternary centre and three fluorinated/chlorinated ones was obtained in good yield and high diastereoselectivity. 7-Chloroquinoline proved amenable to this [2 + 2] cycloaddition/cyclopropanation reaction (product 43) while 5- or 8-chloroquinoline did not lead to detectable amounts of products (Supplementary Fig. 14). Pleasingly, compounds 6 and 41 afforded suitable crystalline specimens and the corresponding X-ray structures are displayed to illustrate the configuration of the fused rings. The scope of the cascade dearomative [2 + 2] cycloaddition/rearrangement is presented in Fig. 6. The two adjacent quaternary centres (44 and 45) were formed in a highly efficient and syn-diastereoselective fashion, even in the presence of trifluoromethyl (44) or chloro (45) groups. Substitution at the pyridyl fragment was also compatible (46–50), targeting the desired product in moderate to good yields (58–82%). Good functional group tolerance was also observed in this transformation (51–59): alkenes (52, 57 and 59), an alkyne (53), a chlorinated arene (54) and a polyfluorinated alkyl (55) proved compatible with the protocol.

Fig. 4: Substrate scope of [2 + 2] cycloaddition/cyclopropanation with respect to the halogenated alkenes.
figure 4

Reaction conditions: quinolines (0.2 mmol), alkenes (0.4 mmol or 1.0 mmol), HCl (0.4 mmol, 4 M in 1,4-dioxane) and Ir-F (2 mol%) in HFIP (1 ml) were stirred for 16 h under argon and irradiated with 30 W blue LEDs (λmax = 450 nm). For experimental details, see Supplementary Methods. d.r., diastereomeric ratio; pTol, para-tolyl.

Full size image
Fig. 5: Substrate scope of [2 + 2] cycloaddition/cyclopropanation with respect to quinolines.
figure 5

Reaction conditions: quinolines (0.2 mmol), alkenes (0.4 mmol or 1.0 mmol), HCl (0.4 mmol, 4 M in 1,4-dioxane) and Ir-F (2 mol%) in HFIP (1 ml) were stirred for 16 h under argon and irradiated with 30 W blue LEDs (λmax = 450 nm). For experimental details, see Supplementary Methods. Ac, acetyl; d.r., diastereomeric ratio. Orange and green circles denote specific substituted carbon atoms to help visualizing the skeletal modification events.

Full size image
Fig. 6: Substrate scope of [2 + 2] cycloaddition/rearrangement.
figure 6

Reaction conditions: quinolines (0.2 mmol), alkenes (1.0 mmol), HCl (0.4 mmol, 4 M in 1,4-dioxane) and Ir-F (2 mol%) in HFIP (1 ml) were stirred for 48 h under argon and irradiated with 30 W blue LEDs (λmax = 450 nm). For experimental details, see Supplementary Methods. Ac, acetyl; d.r., diastereomeric ratio. Green circles highlight the formal C=C bond cleavage between C7 and C8. Light blue circles represent the substitution at the vinyl ester coupling partners.

Full size image

Synthetic application

Gram-scale reactions and derivatization of the obtained fused 6−5−4−3 and 6−4−6 rings were conducted (Fig. 7). Starting from the commercial feedstock 6-chloroquinoline and 2-bromopropene, 1.23 g of 22 was produced in 72% yield, which was comparable with the outcome of the corresponding small-scale reaction. In the presence of Pd(PPh3)4, under the irradiation of blue LEDs, 22 was converted to a rearomatization product (60). A nucleophile, lithium morpholin-4-ide could also trigger a rearomatization process by providing 61. These two-step sequences from a halogenated quinoline and an alkene to 60 and 61 can be recognized as formal cross-electrophilic couplings. The strategy was applied towards the synthesis of an advanced intermediate (62) of CXCR7 receptor chemokine antagonists, offering a straightforward route towards highly decorated quinolines55,56. Compound 24 can also be synthesized in gram scale and, after treatment with tBuOK, was converted into 63 featuring a spirocyclic epoxypropane moiety. The meta-chloroperbenzoic acid-mediated chemoselective oxidation of 44 gave the pyridine N-oxide 64 without competitive epoxidation of the cyclohexene moiety. A cyano moiety was introduced (65) by a subsequent Reissert–Henze reaction. Hydrolysis of 51 provided solid 66 whose structure was resolved by X-ray analysis, thus confirming the 6-4–6 tricyclic structure.

Fig. 7: Gram-scale reactions and synthetic transformations.
figure 7

a, Generation of highly decorated quinolines via rearomatization. b, Exploitation of the rearomatization strategy forming a key intermediate towards a CXCR7 receptor antagonist. c, Gram-scale [2 + 2] cycloaddition/rearrangement reaction followed by epoxidation. d, Synthetic elaboration of the pyridine core via oxidation and Reissert–Henze reaction. e, Ester group hydrolysis and X-ray structural determination of 66. a Contains approximately 15% of an inseparable isomeric product. Reaction conditions: (1) Pd(PPh3)4 (5 mol%), PPh3 (10 mol%), tBuOK (2.0 equiv.), iPrOH (0.1 M), 450 nm LEDs, room temperature; (2) morpholine (2.0 equiv.), nBuLi (2.0 equiv.), tetrahydrofuran (THF) (0.4 M), 0 °C then 22 (1.0 equiv.), THF (0.2 M), 0 °C; (3) 1-benzyl-1,4-diazepane (2.0 equiv.), nBuLi (2.0 equiv.), THF (0.4 M), 0 °C then 22 (1.0 equiv.), THF (0.2 M), 0 °C; (4) tBuOK (2.0 equiv.), THF (0.17 M), −40 °C; (5) mCPBA (3.0 equiv.), CH2Cl2 (0.1 M), room temperature; (6) TMSCl (3.0 equiv.), BzCl (2.0 equiv.), CH2Cl2 (0.1 M), 0 °C to room temperature; (7) NaOH (5.0 equiv.), MeOH:H2O (4:1, 0.1 M), 50 °C. For experimental details, see Supplementary Methods. Bn, benzyl; Bz, benzoyl; d.r., diastereomeric ratio; mCPBA, meta-chloroperbenzoic acid; TMS, trimethylsilyl.

Full size image

Conclusions

We herein introduced two types of energy-transfer-mediated cascade dearomative [2 + 2] cycloaddition/rearrangement (CDACR) reaction which could provide facile access to pyridine-fused 2D/3D ring systems. Tailor-made substitution at the quinoline framework could divert the second energy-transfer event towards either cyclopropanation or cyclobutane rearrangement, delivering 5–4–3-membered fused or 4–6-membered fused rings, respectively. Extremely high structural complexity was directly elaborated from readily available quinolines and alkenes by means of two consecutive energy-transfer events, mediated by an iridium-based photosensitizer. High reaction efficiency and excellent diastereoselectivity, which are challenging within the intermolecular dearomative [2 + 2] cycloaddition arsenal, have been observed. Furthermore, compared to the conventional cycloaddition reactions, this method using aromatics as reactants and harnessing a consecutive transformation leads to an intriguing chemical space. Given the high prevalence of pyridine-fused 2D/3D rings in drug discovery, we anticipate this method will facilitate the efficient synthesis of such scaffolds.

Methods

Representative procedure for the [2 + 2] cycloaddition/cyclopropanation cascade

An oven-dried 5 ml Schlenk tube was charged with the appropriate 6-chloroquinoline (1.0 equiv.), the appropriate haloalkene (2.0 or 5.0 equiv.), HCl (2.0 equiv., 4 M in 1,4-dioxane) and [Ir(dF(CF3)ppy)2(dtbbpy)][PF6] (2 mol%) and 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP, 0.2 M). The reaction mixture was degassed using two freeze–pump–thaw cycles. After the mixture was thoroughly degassed and filled with argon, the Schlenk tube was tightly sealed and stirred while under irradiation with 30 W blue LEDs (λmax = 450 nm) for 14–24 h (monitored by thin-layer chromatography). The reaction was quenched with saturated aqueous NaHCO3 and extracted with CH2Cl2 (3 times, 10 ml each time). The organic phases were combined and concentrated under reduced pressure. 1H NMR analysis of the crude reaction mixture gave the d.r. values. The analytically pure product was obtained by flash chromatography on silica gel (n-pentane/EtOAc, n-pentane/Et2O or CH2Cl2/MeOH as eluent).

Representative procedure for the [2 + 2] cycloaddition/rearrangement cascade

An oven-dried 5 ml Schlenk tube was charged with the appropriate 8-quinoline ester (1.0 equiv.), the appropriate vinyl ester (5.0 equiv.), HCl (2.0 equiv., 4 M in 1,4-dioxane) and [Ir(dF(CF3)ppy)2(dtbbpy)][PF6] (2 mol%) and 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP, 0.2 M). The reaction mixture was degassed using two freeze–pump–thaw cycles. After the mixture was thoroughly degassed and filled with argon, the Schlenk tube was tightly sealed and stirred while under irradiation with 30 W blue LEDs (λmax = 450 nm) for 48 h. The reaction was quenched with saturated aqueous NaHCO3 and extracted with CH2Cl2 (3 times, 10 ml each time). The organic phases were combined and concentrated under reduced pressure. 1H NMR analysis of the crude reaction mixture gave the d.r. values. The analytically pure product was obtained by flash chromatography on silica gel (n-pentane/EtOAc or n-pentane/acetone as eluent).