Visible Light-Induced Intermolecular Dearomatization for Constructing Angularly Fused Tetracyclic Scaffolds Authors: Fan Lin, Ting-Ting Song, Li-Ming Zhang, Zhi-Yuan Ding, Yilitabaier Julaiti, Yanhua Lu, Boshun Wan, Yaguang Sun, Qing-An Chen Published: 2026 ABSTRACT AbstractVisible light-induced dearomative cycloaddition has emerged as a powerful strategy for constructing complex polycyclic frameworks from simple aromatic precursors. Angularly fused polycyclic systems, which exhibit distinctive structural rigidity and tunable physicochemical properties, represent privileged scaffolds in bioactive natural products and pharmaceuticals. Herein, we report a robust photocatalytic protocol for the direct assembly of angularly fused tetracyclic scaffolds through intermolecular dearomatization of benzothiophene/indole derivatives with N-aryl cyclopropylamines. This transformation is achieved via a mechanistically intriguing cascade involving photocatalytic dearomative [3 + 2] cycloaddition and subsequent cyclization. This work demonstrates the synthetic feasibility of photocatalytic intermolecular dearomative [3 + 2] processes, providing direct access to structurally distinct angularly fused tetracyclic cores.Subject terms: Photocatalysis, Homogeneous catalysis, Synthetic chemistry methodology Introduction Characterized by their distinct three-dimensional architectures, angularly fused polycyclic scaffolds represent a privileged structural motif in bioactive natural products and pharmaceuticals with diverse biological activities (Fig. 1a)1,2. As a natural indole alkaloid, strychnine has been used as a potent rodenticide due to its potent neurotoxic activity mediated through glycine receptor antagonism in the central nervous system3,4. The monoterpene indole alkaloid aspidospermidine demonstrates dual pharmacological activities as both a respiratory stimulant and antimicrobial agent5. Schizozygine, a bioactive alkaloid from schizozygia caffaeoides, shows antimicrobial effects relevant to its traditional dermatological applications6. Similarly, alstonlarsine A has been identified as a selective DRAK2 kinase inhibitor with therapeutic potential7. Therefore, the construction of angularly fused tetracyclic scaffolds is of great interest. Intermolecular dearomative cycloaddition for constructing angularly fused tetracyclic scaffolds. a Representative natural polycyclic scaffolds molecules. b Previous work: Construction of angularly fused tetracyclic scaffolds via intramolecular dearomatization. c The reaction pattern of CPAs in photochemical reactions. d Design plan: Intermolecular dearomatization. e This work: Visible light-induced intermolecular dearomatization for constructing angularly fused tetracyclic scaffolds. Over the past decade, visible light-induced dearomative cycloadditions8–16, have emerged as a powerful strategy for the expedient synthesis of strained polycyclic architectures17–29 that are challenging to access through traditional methods. Particularly, You30–33, Oderinde34, and other groups35–39 have demonstrated pioneering work showing that triplet-state electron-deficient indoles readily engage in intramolecular dearomative [2 + 2], [4 + 2], and [5 + 2] cycloadditions to construct diverse angularly fused polycyclic scaffolds via an energy transfer mechanism (Fig. 1b). Among these, only one case involving a single electron transfer process was reported by Yang et al.40. While intramolecular variants have been well-explored, the development of intermolecular dearomative cycloadditions–especially those forming angularly fused systems–remains underexplored and represents a significant synthetic challenge. Meanwhile, cyclopropylamine (CPA) derivatives have emerged as intriguing synthons in modern organic synthesis, particularly for the construction of valuable cyclic amines41–43. Their synthetic utility stems from a characteristic ring-opening pathway initiated by single-electron oxidation to form reactive nitrogen radical cations (Fig. 1c)41. This distinctive reactivity enables CPAs to participate in cycloadditions with alkenes44–50, alkynes51,52, or bicyclo[1.1.0]butanes53, providing efficient access to structurally diverse cyclic amine scaffolds. Inspired by these precedents and based on our research interests in photocatalytic dearomatization54–57, we envisioned that combining the photochemical excitation of benzothiophenes or indoles with the radical reactivity of N-aryl cyclopropylamines could enable distinctive dearomative cycloadditions (Fig. 1d). In this context, it is challenging to suppress the self-dimerization of aromatic compounds without disturbing photocatalysis. The design of the substrates and the selection of the photosensitizer are critical factors in the transformation process. Herein, we developed a robust photocatalytic protocol for the direct assembly of angularly fused tetracyclic scaffolds through intermolecular dearomatization of benzothiophene/indole derivatives with N-aryl cyclopropylamines (Fig. 1e). This protocol not only expands the toolbox for dearomatization but also provides rapid access to complex heterocycles relevant to medicinal chemistry and materials science. Results The evaluation of reaction conditions commenced with methyl benzothiophene-2-carboxylate (1a) and N-phenyl cyclopropylamine (2a) as the model substrates (Table 1, see Supplementary Information for details). Pleasingly, under blue LEDs irradiation for 20 h, the saturated bicyclic amine 4a could be obtained in 50% yield with [Ir(dF(CF3)2(ppy)2(dtbbpy)]PF6 (PC-I, 1 mol%) in acetone (0.1 M) (entry 1), accompanied by the inevitable formation of dimer (3a) of 1a. Dimer 3a was obtained with high diastereoselectivity ( > 95:5), and the configuration was unambiguously assigned by single-crystal X-ray diffraction (CCDC: 2526038). In addition, both diastereomers of the [3 + 2] cycloadduct 4a (trans) and 4a’ (cis) were successfully isolated. The configuration of 4a’ (cis) was also confirmed by single-crystal X-ray diffraction (CCDC: 2526040). Screening of solvents revealed that THF and DCM resulted in relatively low yields (entries 1–3). Photosensitizers [Ir(ppy)2(dtbbpy)]PF6 (PC-II), Ir(dFppy)2(pic) (PC-III), TXT (PC-IV), and 5CzBN (PC-V) proved to be less efficient (entries 4–7). Yet, through extensive conditional optimization, attempts to suppress dimerization failed (Please see Supplementary Tables 1–4 for details). The control experiments with the isolated compounds 4a (trans) and 4a’ (cis) demonstrated that none of the tested conditions (e.g. photocatalysis, thermal activation, or base promotion) facilitated their cyclization to the lactam 5a. Instead, under either photochemical or basic conditions, 4a (trans) and 4a’ (cis) underwent mutual isomerization to establish an equilibrium (Please see Supplementary Tables 22 and 23 for details). This impasse highlighted the need for a more reactive acylating reagent capable of driving the cascade to completion. We therefore pivoted to a substrate-design strategy, hypothesizing that kinetically accelerating the [3 + 2] step could outcompete dimerization. Inspired by precedents employing pyrazole derivatives as superior leaving groups in dearomative photocycloadditions14,20,29, we evaluated a series of benzothiophene substrates bearing acyl pyrazoles (entries 8–12). Gratifyingly, all such substrates successfully delivered product 5a, confirming the activated group’s role as an integral “cyclization driver.” Notably, when bulky aryl substituents were introduced on the pyrazole moiety, the formation of the dimer could be effectively suppressed. In particular, the phenyl-substituted substrate afforded the target product 5a in an isolated yield of 71%. Corresponding [3 + 2] cycloadduct 4d’ (cis) could be isolated in other solvent systems (DMSO or MeNO2, entries 13 and 14), whose configuration was similarly established by single-crystal X-ray diffraction (CCDC: 2526041). Finally, control experiments established the indispensability of both light and the photocatalyst (entries 15 and 16). Table 1Optimization of the reaction conditionsaEntry1SolventPhotocatalyst3 (%)b (dr)4 (%)b (dr)5a (%)b11aAcetonePC-I63 ( > 95:5)50 (67:33)N.D.21aTHFPC-I43 ( > 95:5)16 (70:30)N.D.31aDCMPC-I74 ( > 95:5)6 (62:38)N.D.41aAcetonePC-II51 ( > 95:5)30 (67:33)N.D.51aAcetonePC-III53 ( > 95:5)22 (75:25)N.D.61aAcetonePC-IVc82 ( > 95:5)10 (67:33)N.D.71aAcetonePC-V85 ( > 95:5)2 (57:43)N.D.81bAcetonePC-I35 ( > 95:5)<56791cAcetonePC-I20 ( > 95:5)<561101dAcetonePC-I11 ( > 95:5)<574 (71)e111eAcetonePC-I23 ( > 95:5)<557121fAcetonePC-I18 ( > 95:5)<566131dDMSOPC-I6 ( > 95:5)39 ( < 5:95)24141dMeNO2PC-I22 ( > 95:5)12 ( < 5:95)Trace151dAcetonenone00016 d1dAcetonePC-I000aReaction conditions: 1 (0.15 mmol), 2a (0.10 mmol), and photocatalyst (1 mol%), solvent (1.0 mL), blue LEDs, RT,20 h.bYields were determined by GC-FID. Yield of 3 was calculated based on substrate 1 (0.15 mmol).cPhotocatalyst (5 mol%).dIn the absence of light irradiation.eIsolated yield. Optimization of the reaction conditionsa aReaction conditions: 1 (0.15 mmol), 2a (0.10 mmol), and photocatalyst (1 mol%), solvent (1.0 mL), blue LEDs, RT, 20 h. bYields were determined by GC-FID. Yield of 3 was calculated based on substrate 1 (0.15 mmol). cPhotocatalyst (5 mol%). dIn the absence of light irradiation. eIsolated yield. With the optimized reaction condition established, we investigated the scope of aromatic substrates and N-aryl cyclopropylamines (Fig. 2). First, substitutions on the N-aryl cyclopropylamines were evaluated. Pleasingly, the para-substituted N-aryl cyclopropylamines bearing various electron-donating groups (-Me, -tBu, and -OMe) performed well, affording the products (5b–5d) in 56–68% yields. In addition, angularly fused polycyclic products (5e–5g) with electron-withdrawing substituents could also be obtained in a reasonable yield. The -OMe group at the meta position of the arylamine was also applicable in this protocol, affording the desired product 5h. Furthermore, 2-substituted N-aryl cyclopropylamines performed well, delivering polycyclic products (5i and 5j) in high yields. The pyridyl, 1-naphthyl, and 2-naphthyl groups could also be tolerated in this strategy (5k–5m). Furthermore, N-aryl cyclobutylamines bearing various groups have also been explored. The [4 + 2] cycloaddition reaction generally tolerated both electron-donating (-tBu, -OMe, and -OCF3) substituents on the aryl ring, forming angularly fused tetracyclic products (5n–5q) in 60–69% yields. However, attempts to employ other known cyclopropanamine precursors, such as N-tosyl cyclopropylamines58 or N-cyclopropylureas59 under the developed conditions proved unsuccessful (Please see Supplementary Tables 11–14 for details). Substrate scope of benzothiophene derivatives. Conditions: 1 (0.30 mmol), 2 (0.20 mmol), and PC-I (1 mol%), acetone (2 mL), blue LEDs (24 W), RT, 20 h. a1 (0.20 mmol) and 2 (0.20 mmol) were used. Isolated yields were given. Next, we examined the scope of benzothiophenes for this dearomative cycloaddition reaction. The 5-position Me-substituted benzothiophene substrate worked smoothly, thus providing polycyclized product 5r. Then, the applicability of aromatic substrates bearing various electron-withdrawing groups (-X and -CF3) at the 5-position was further investigated. The desired products 5s–5v could also be accessed in yields of 53–62%. The 5-methoxy-substituted substrate underwent successful transformation, yielding the desired product 5w. Moreover, naphthyl as well as heterocyclic substituents such as pyrrole and thiophene were well tolerated in this dearomative cycloaddition (5x–5z). Substrates with electron-withdrawing groups at the 6-position of the phenyl ring were likewise compatible with this protocol (5aa and 5ab). The structure of 5ab was further confirmed by single-crystal X-ray crystallography (CCDC: 2468281). The reaction proceeded efficiently with 7-substituted benzothiophenes, affording products 5ac and 5ad. Benzothiophene bearing a chlorine atom at the 4-position was successfully converted to product 5ae in moderate yield. Furthermore, substituted benzothiophenes (-Me and -F) participated readily in the reaction with N-Ph cyclobutylamine, delivering target products (5af and 5ag) in good yields. The structures of 5ad and 5af were further confirmed by single-crystal X-ray crystallography (CCDC: 2526042 and 2468282). A naproxen-derived benzothiophene also underwent smooth cycloaddition with N-phenyl cyclopropylamine to deliver the target compound 5ah. These results collectively demonstrate that the electronic nature of the substituent is not a limiting factor for this transformation, with both electron-donating and -withdrawing groups being well tolerated. However, C2-substituted benzofuran substrates proved incompatible with our developed photocatalytic protocol (Please see Supplementary Tables 8–10 for details). We attribute the lack of reactivity with benzofuran substrates to their unfavorable electronic properties. Even with an electron-withdrawing acyl pyrazole at C2, the strong π-donating ability of oxygen in benzofuran keeps the C2 = C3 bond insufficiently electrophilic, which hinders its reaction with nucleophilic radical species, leading to low reactivity. Subsequently, the substrate scope of the dearomative reaction from the indoles and N-aryl cyclopropylamines was investigated (Fig. 3a). Initial study with 6a was carried out to provide the corresponding product 7a in 73% isolated yield. Notably, the addition of Et3N is necessary to facilitate diastereomeric interconversion, thereby enabling more efficient elimination of the pyrazole moiety and subsequent cyclization. A variety of substituents on the phenyl ring of N-aryl cyclopropylamines, including -Me, -OMe, and -F, were well tolerated, delivering desired products (7b–7d) in reasonable yields. The structure of 7c was further confirmed by single-crystal X-ray crystallography (CCDC: 2468232). Moreover, the 5-bromoindole substrate underwent the reaction smoothly, affording angularly fused tetracyclic product 7e. The 6-bromo-substituted product 7f was also successfully isolated by subjecting the intermediate formed in the previous step to continued irradiation at 390 nm. In the reaction with N-phenylcyclobutane, the [4 + 2] cycloaddition product could not be isolated when acetone was used as the reaction solvent. Instead, an interesting tetrahydrofuran-involved hydroalkylation product 7g’ was obtained (THF as solvent). The structure of 7g’ was further confirmed by single-crystal X-ray crystallography (CCDC: 2526043). This transformation likely proceeds via a photo-promoted hydrogen atom transfer (HAT) process, followed by reaction with alkyl radical (from THF) to afford the dearomative hydroalkylation product. Notably, the unprotected indole substrate 6h failed to react. This is likely because the strong π-donation from the unprotected N–H (vs electron-withdrawing Boc group) indole nitrogen excessively enriches the C2 = C3 bond, reducing its electrophilicity and hindering productive reaction with a nucleophilic radical partner. To our delight, benzothiophene and benzofuran with an acyl pyrazole group at the 3-position proved amenable to this protocol, yielding the corresponding angularly fused tetracyclic products 7i and 7j. To further delineate the structural requirements for this transformation, we examined a variety of heterocyclic substrates beyond the successful 2-acylpyrazole-substituted benzothiophenes and N-Boc-protected 3-acylpyrazole-substituted indoles. As illustrated in Supplementary Fig. 2, N-methylindole, thiophene, and coumarin derivatives all exhibited poor reactivity. In other cases, such as with substrates containing methyl esters (CO₂Me) at distinct positions, the reactions were sluggish. These unsuccessful attempts are compiled in the Supplementary Information to provide a complete picture of the reaction scope and limitations. The substrate scope of indole derivatives and one-pot substrate scope. a The substrate scope of indole derivatives. Conditions: I) 6 (0.20 mmol), 2 (0.30 mmol), and PC-I (1 mol%), Et3N (2.0 eq.), acetone (2.0 mL), blue LEDs (24 W), RT, 20 h. II) DBU (2.0 eq.), 60 °C, 6 h. aTHF was used instead of acetone. bStep 2: 390 nm, RT, 16 h. Isolated yields were given. N.D.: Not detected. N.R.: No reaction. b One-pot substrate scope. Conditions: I) 1’ (0.40 mmol), 3-phenyl-1H-pyrazole (0.60 mmol), EDCI (0.80 mmol), DMAP (20 mol%), DCM (1.0 mL), RT, 12 h. II) 2 (0.20 mmol), PC-I (1 mol%), acetone (1.0 mL), blue LEDs (24 W), RT, 20 h. Isolated yields were given. Without presynthesizing acyl pyrazole in advance, we successfully developed a one-pot synthetic strategy that enabled the direct conversion of benzothiophene-2-carboxylic acids into angularly fused tetracyclic products (Fig. 3b). This further highlights the versatility and mild reaction conditions of our protocol. When employing thianaphthene-carboxylic acid as the substrate, various N-aryl cyclopropylamines underwent smooth transformation to afford the corresponding products (5a, 5b, 5d, 5e, and 5n). Furthermore, substituted benzothiophene-2-carboxylic acids also proved to be suitable substrates for this transformation through in situ activation of the carboxylic acid group, as demonstrated by the successful synthesis of compounds 5r, 5s, and 5ae. To shed light on our mechanistic hypothesis, a series of mechanistic experiments were conducted (Fig. 4). First, when 1ai was subjected to the standard reaction conditions, no target product was produced, ruling out the possibility of 1ai being a reaction intermediate (Fig. 4a). Second, the kinetic profile of a standard reaction mixture of 1d and 2a showed that the reaction proceeded very rapidly and did not have obvious induction periods, with the majority of the product formed within 5 h (Fig. 4b). Next, light on/off experiments demonstrated that the transformation was interrupted in the absence of light (Fig. 4c). It suggests the crucial role of light irradiation and rules out a long radical chain process. Moreover, the photocatalyst PC-I was found to be the only light-absorbing species in the reaction near the excitation wavelength from ultraviolet–visible (UV-vis) absorption spectroscopy (Fig. 4d). Then, Stern−Volmer luminescence quenching experiments were performed using d, 2a, and 6a, respectively (Fig. 4e). The results demonstrate that both substrate 1d and cyclopropylamine 2a are competent quenchers of the excited photocatalyst PC-I*, with 1 d exhibiting significantly higher quenching efficiency than 2a. It implies that the intertwined EnT/SET or SET pathway is likely involved in the formation of 5. While no obvious quenching was observed for indole 6a, this indicates that radical addition is involved in the formation of 7. Mechanistic studies. a Investigation of involved intermediates. b Kinetic profile. c Light on/off experiments. d UV-vis absorption analysis. e Stern-Volmer quenching studies. f Radical inhibition experiments. g Diradical capture experiments. h The role of adding Et3N. i Intermediate conversion experiments. The dearomative process was suppressed when adding radical inhibitors TEMPO or BHT (Fig. 4f, entries 1–3), which suggests the reaction proceeds via a radical mechanism. And the observation of TEMPO-adduct by ESI-HRMS suggests that the radical intermediate is likely involved in the reaction. On the other hand, the standard reaction was significantly inhibited in the presence of known triplet quenchers, such as O2 or 1,1-diphenylethylene (Fig. 4f, entries 4 and 5). When using 2,5-dimethylhexa-2,4-diene as a diradical scavenger under standard conditions, the [2 + 2] adduct [8a + 8a’] was detected and isolated in 46% yield (Fig. 4g). The structure of 8a was confirmed by single-crystal X-ray crystallography (CCDC: 2526044). The detection of a diradical intermediate provides evidence for the involvement of triplet-state species, suggesting that an energy transfer mechanism may be involved in this dearomatization process. To understand the nature of the reaction, the ΔG(T1 − S0) values of substrates 1a (55.8 kcal mol-1), 1d (54.4 kcal mol-1), and 6a (59.6 kcal mol−1) were calculated. As a result, the yield of the corresponding product did not exhibit a straightforward linear dependence on the triplet energy of photocatalysts (Supplementary Table 2). In addition, electrochemical studies also show that it is thermodynamically feasible for the excited photocatalyst (Ep/2III*/II = +1.21 V vs SCE in CH3CN) to oxidise substrate 2a (Ep/2ox = +1.19 V vs SCE in CH3CN) via SET (Supplementary Fig. 10)41,44,60. However, for substrates 1d (Ep/2ox = +0.75 V, +1.86 V, and Ep/2red = −1.12 V, −1.52 V vs SCE in CH3CN) and 6a (Ep/2ox = +1.82 V, and Ep/2red = −1.10 V, −1.85 V vs SCE in CH3CN), no obvious correlation between the oxidation/reduction potentials and the reactivity was found. For the indole substrate 6a, the addition of NEt3 led exclusively to the formation of the trans-configuration [3 + 2] cycloadduct 7aa (trans). In the absence of the base, both diastereomers 7aa (trans) and 7aa’ (cis) were isolated successfully with 64:36 dr (Fig. 4h). In addition to routine NMR analysis, variable-temperature 1H NMR experiments were performed to assess the conformational stability of both compounds (7aa and 7aa’) (Supplementary Figs. 12 and 13). Control experiments demonstrated that both 7aa (trans) and 7aa’ (cis) could be converted to the target product 7a under base-promoted conditions. Under photochemical conditions, conversion of 7aa (trans) was inhibited in the presence of NEt3 (Supplementary Table 27). However, under prolonged irradiation in the absence of NEt3, 7aa (trans) was efficiently transformed into lactam 7a. Short-term monitoring of the photochemical process revealed an initial formation of the cis-isomer 7aa’ (cis) (Fig. 4i, eq. 1). Furthermore, the cis-configuration pyrazole-substituted adducts 4d’ (cis) or 7aa’ (cis) were efficiently transformed into lactam 5a or 7a via either photochemical or base-mediated pathways (Fig. 4i, eqs. 2 and 3). Collectively, these results provide compelling evidence that the trans-configuration [3 + 2] cycloadducts first undergo isomerization to their cis-counterparts, which subsequently serve as the direct precursors for the final cyclization step to afford the desired lactam products (5 or 7). Based on DFT calculations and mechanistic experiments, a plausible mechanism is postulated as outlined (Fig. 5, please see Supplementary Information and Source Data for details). First, visible light excitation of the photocatalyst yields an excited state complex which sensitizes 1d via the EnT process, affording diradical intermediate A. Radical addition of species A to substrate 1d furnishes intermediate B. Subsequent intramolecular radical-radical coupling of B leads to the undesired dimer 3d. In the meantime, the photoexcited species IrIII* undergoes a single-electron reduction by cyclopropylamine (2a) to deliver IrII and radical cation species C. A subsequent ring opening via β-scission delivers distonic radical cation D. Next, radical addition with 1d with radical cation species D (path I) furnishes another distonic radical cation E. The direct combination of two relatively low-concentration excited-state species (A and D, path II) is kinetically unlikely. Subsequently, species E undergoes cyclization, providing access to the radical cation F. At this stage, F could be reduced by the IrII species generated in the reductive quenching photoredox cycle, delivering [3 + 2] cycloadducts trans-G and cis-G. At the same time, the alternative pathway (path III: reduction–direct cyclization) remains a possibility. Proposed mechanism. Energy profile and proposed mechanistic pathways. The target angular-fused lactam 5a can be generated through two distinct pathways (photo-induced or base-promoted pathway, Fig. 5). Under photo-induced conditions, the carbonyl group of trans-G is photoexcited to generate the diradical species trans-H, which then undergoes β-scission via transition state trans-TS5 to afford intermediate I. Subsequent radical addition yields the excited-state species cis-J, thereby achieving configurational inversion. The species cis-J can also originate from the excitation of the cis-G. Intermediate cis-J undergoes a 1,5-H transfer to form cis-K, which further cyclizes to give intermediate L. Subsequent deprotonation followed by pyrazole elimination affords the target product 5a. In an alternative pathway, intermediate trans-G undergoes deprotonation under basic conditions (DBU) to generate the anionic intermediate trans-N. The elimination (retro-Mannich reaction) of trans-N affords ring-opening species O via transition state trans-TS7 (with an activation energy of 13.2 kcal mol−1), thereby achieving racemization of the stereocenter next to the sulfur atom. A subsequent addition partially yields the cis-configured intermediate cis-N, which can also be afforded from cis-G in the presence of DBU. The anionic cis-N then undergoes addition–cyclization onto the carbonyl group to form intermediate M, followed by pyrazole elimination to furnish the desired lactam 5a. Additionally, computational results indicate that, compared to the acylpyrazole-substituted substrate 1d, the intermediate cis-N1a derived from the methyl formate-substituted substrate 1a needs to overcome a higher energy barrier (9.4 kcal mol−1) in the subsequent cyclization step (cis-N1a to M1a). Moreover, due to the stronger nucleophilicity of the OMe, which makes it a poorer leaving group, 3 + 2 cyclization product 4a is a thermodynamically stable product (Supplementary Fig. 37). To further demonstrate the synthetic potential of this protocol, we carried out scale-up synthesis of angularly fused polycyclic products (5a and 7a) under standard conditions, and the efficiencies did not erode significantly (Fig. 6a). These compounds serve as versatile building blocks, enabling further structural diversification through subsequent transformations (Fig. 6b). Initially, with the addition of LiAlH4, β-lactam 5a could be reduced to yield azetidine 9 in 62% yield. Subsequently, oxidation of azetidine 9 with mCPBA yielded the corresponding sulfoxides 10 and 10’, with concurrent azetidine ring opening. This ring cleavage, likely driven by ring strain release or nucleophilic attack-induced C–N bond cleavage, resulted in the formation of N-phenyl amino alcohol. Furthermore, sulfone 11 was created from 9 through further oxidation with H2O2. X-ray structures of products 10, 10’, and 11 were presented to confirm their configurations (CCDC: 2473229–2473231). The chemoselective oxidation of 5a by CAN was successfully carried out to give sulfoxides 12 and 12’ in 80% total yield with 4:1 diastereoselectivity due to the chirality of sulfur. Oxidative amidation of sulfoxide 12 with hypervalent iodine reagent could afford sulfoximine 13 in high yield61. Corresponding sulfone 14 was obtained from 5a by further oxidation with mCPBA. Besides, the deprotection of the N-Boc group occurred smoothly upon the treatment of 7a with trifluoroacetic acid, affording 15 in 80% yield. The free NH group in 15 offered flexible handles for further functional group elaboration. In the presence of trifluoroacetic anhydride and Et3N, N-protection of 15 with COCF3 delivered compound 16 in 83% yield. X-ray structures of products 12, 12’, 14, 15, and 16 were presented to confirm the configuration of these products (CCDC: 2468285–2468289). Scale-up experiments and synthetic transformations. a Scale-up experiments. b Synthetic transformations. In conclusion, we have developed an efficient visible-light photocatalytic strategy for the direct construction of angularly fused tetracyclic scaffolds via intermolecular dearomatization of benzothiophene/indole derivatives with cyclopropylamines. This transformation proceeds through a mechanistically distinct cascade involving a photocatalytic dearomative [3 + 2] cycloaddition followed by cyclization, offering a streamlined approach to structurally complex polycyclic systems. The present work not only expands the synthetic utility of photocatalytic intermolecular dearomative cycloadditions but also provides direct access to structurally distinct angularly fused tetracyclic cores, which are privileged motifs in bioactive molecules. This methodology thus opens an avenue for the rapid assembly of rigid polycyclic architectures with potential applications in medicinal chemistry and natural product synthesis. Methods General procedures for the synthesis of product 5 To an oven-dried 4 mL vial with a PTFE-coated stirring bar was added 1 (0.30 mmol, 1.5 eq.), 2 (0.20 mmol, 1.0 eq.), PC-I (0.002 mmol, 1 mol%), and acetone (2.0 mL, 0.1 M) in a nitrogen glove box. The vial was capped with a septum and wrapped with parafilm. The reaction mixture was stirred under irradiation with a light box at room temperature for 20–30 h. The reaction mixture was concentrated in vacuo and purified by column chromatography on silica gel using petroleum ether and ethyl acetate to afford the corresponding products. General procedures for the synthesis of product 7 To an oven-dried 4 mL vial with a PTFE-coated stirring bar was added 6 (0.20 mmol, 1.0 eq.), 2 (0.30 mmol, 1.5 eq.), Et3N (0.40 mmol, 2.0 eq.), PC-I (0.002 mmol, 1 mol%), and acetone (2.0 mL, 0.1 M) in a nitrogen glove box. The vial was capped with a septum and wrapped with parafilm. The reaction mixture was stirred under irradiation with a light box at room temperature for the initial 20 h. The reaction mixture was concentrated in vacuo, then DBU (0.30 mmol, 1.5 eq.) and acetone (1.0 mL, 0.2 M) were added in the nitrogen glove box. The reaction mixture was stirred at 60 °C for 6 h. The reaction mixture was concentrated in vacuo and purified by column chromatography on silica gel using petroleum ether and ethyl acetate to afford the corresponding products. Supplementary information Supplementary Information Transparent Peer Review file Source data Source Data Acknowledgements We thank Prof. Zhi-Shi Ye (DUT) for helpful discussions and manuscript revisions. Author contributions Q.-A.C., Y.S. and T.-T.S. conceived and supervised the project. Q.-A.C., T.-T.S. and F.L. designed the experiments. F.L., T.-T.S., L.-M.Z., Z.-Y.D., Y.J., Y.L. and B.W. performed the experiments and analyzed the data. All authors discussed the results and commented on the manuscript. Peer review Peer review information Nature Communications thanks Qiang Fu and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available. Funding Financial support from the National Natural Science Foundation of China (22501278, T.-T.S.) is acknowledged. Data availability The X-ray crystallographic data for compounds generated in this study have been deposited in the Cambridge Crystallographic Data Centre (CCDC), under deposition numbers CCDC 2526038 (3a), 2526039 (3c), 2526040 (4a’), 2526041 (4 d’), 2468280 (5a), 2468281 (5ab), 2526042 (5ad), 2468282 (5af), 2468283 (7c), 2526043 (7 g’), 2468284 (7aa), 2526044 (8a), 2473229 (10), 2473230 (10’), 2473231 (11), 2468285 (12), 2468286 (12’), 2468287 (14), 2468288 (15), and 2468289 (16). These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif. Data relating to the characterization data of materials and products, general methods, optimization studies, experimental procedures, mechanistic studies, mass spectrometry, and NMR spectra are available in the Supplementary Information. All data is available from the corresponding author upon request. The source data generated in this study are provided as a Source Data file. Source data are provided with this paper. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Ting-Ting Song, Email: ttsong@dicp.ac.cn. Yaguang Sun, Email: sunyaguang@syuct.edu.cn. Qing-An Chen, Email: qachen@dicp.ac.cn. Supplementary information The online version contains supplementary material available at 10.1038/s41467-026-75078-6. References Associated Data Supplementary Materials Supplementary Information Transparent Peer Review file Source Data Data Availability Statement The X-ray crystallographic data for compounds generated in this study have been deposited in the Cambridge Crystallographic Data Centre (CCDC), under deposition numbers CCDC 2526038 (3a), 2526039 (3c), 2526040 (4a’), 2526041 (4 d’), 2468280 (5a), 2468281 (5ab), 2526042 (5ad), 2468282 (5af), 2468283 (7c), 2526043 (7 g’), 2468284 (7aa), 2526044 (8a), 2473229 (10), 2473230 (10’), 2473231 (11), 2468285 (12), 2468286 (12’), 2468287 (14), 2468288 (15), and 2468289 (16). These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif. Data relating to the characterization data of materials and products, general methods, optimization studies, experimental procedures, mechanistic studies, mass spectrometry, and NMR spectra are available in the Supplementary Information. All data is available from the corresponding author upon request. The source data generated in this study are provided as a Source Data file. Source data are provided with this paper.