Densely-functionalized bicyclic cyclopentanones by combined photoinduced 6-endo-trig Giese additions and mild aldol cyclizations

Abstract

Polycyclic scaffolds are central to numerous natural products and pharmaceuticals, motivating concise, stereocontrolled routes to their construction. We report a photoredox-enabled synthesis of trans-fused [n.3.0] bicyclic ketones (n = 4, 5, 10) from malonate-derived enol ethers. α-Brominated intermediates, formed by acylation with 2-bromo-2-methylpropanoyl bromide, undergo radical cyclization under two complementary conditions: (i) acridinium orange (AOH⁺) with Hantzsch ester (HE) at 455 nm, or (ii) photoexcited HE alone at 365 nm. Both modes trigger 6-endo-trig Giese addition, C–O bond fragmentation, and hydrogen-atom transfer to α-branched cyclic ketones that ring-close under mild Brønsted or Lewis acid activation to trans-fused products with exclusive junction control. Mechanistic studies (Stern–Volmer, UV–Vis, electrochemistry) support two activation pathways—AOH⁺* quenching by HE or direct HE excitation—each generating the same radical intermediates in fine. DFT calculations validate mechanistic pathways and regioselectivity in favor of philicity matching of the electrophilic radical and the polar electron-rich enol ether. The method accommodates ring-size diversity, accesses trans-hydrindanone architectures, and outcompetes 5-exo-trig spirocyclization.

---

Introduction

Polycyclic natural products and medicinal drugs are ubiquitous generating hence a continuous demand for novel cyclization strategies in academic and industrial laboratories alike.¹ Methodologies for ring closures are thus crucial and very diverse approaches are available, from polar bond formations to concerted, organometallic or radical (vide infra) pathways.² Cyclopentenones,³ and related saturated cyclopentanones,⁴ are key structural subunits often included in medicinal precursors of drug candidates,⁵ or as part of polycyclic natural products, such as przewalskin B,⁶ norspiculoic acid A or nitidasin⁷ (Scheme 1i). In the cyclopentanone series, generation of highly-substituted (hindered) skeletons remains a general challenge requiring, in addition, the stereocontrol of trans- or cis-fused junctions.⁸

Scheme 1 Polycyclic pentenones and pentanones. Photoreductive radical cyclizations. Synthetic strategy toward highly substituted [n.3.0] bicyclic derivatives. AOH⁺ = acridinium orange photocatalyst.

As mentioned earlier, radical-mediated synthesis of polycyclic molecules is a recognized strategy using, most often, intramolecular Giese reactions of carbon-centered radicals onto neighboring olefin(s).⁹ Conditions of radical formation and the size of the resulting ring(s) are key parameters that control addition regio- and stereoselectivity.¹⁰ Cyclizations leading to the formation of 5- or 6-membered rings have been particularly studied.¹¹ Reliable prognostics on such cyclizations are possible thanks to many seminal contributions and Beckwith's radical rules in particular.¹² Also, visible light-induced and photoredox catalyzed reactions have transformed the field of radical-mediated processes leading to a large array of novel synthetic methodologies.¹³ In this regard, formations of cyclized products by intramolecular additions after photoreduction of C–X bonds (X = halogen, chalcogen, etc.) are key.^13b,14 Most often, n-exo-trig regioselectivity is observed primarily, over n + 1-endo-trig pathways (Scheme 1ii).¹⁵ Of note, Blakey et al. reported a switchable regioselective 6-endo/5-exo photoredox catalyzed cyclization.¹⁶ The selectivity arose from different HAT rates; the 5-exo product being obtained with a polarity-matched thiol HAT process, while using Hantszch ester (HE) gave the 6-endo product.

Previously, using cyclic ketones 1 and α-diazomalonates¹⁷ as reactants (Scheme S1), a direct synthesis of malonate enol ethers 2 was reported under [CpRu(N≡CCH₃)₃][BAr_F] catalysis;¹⁸ the resulting adducts being readily functionalized under basic conditions to afford malonate alkylation products.¹⁹ Herein, to harness regioselective radical reactivity and later achieve densely-functionally cyclopentanones, the acylation of enol ethers 2 with 2-bromo-2-methylpropanoyl bromide 3 was pursued to produce substrates of type 4. Subsequent photoredox conditions afford α-branched ketones that cyclize under mild acidic conditions to form densely-functionalized [n.3.0] bicyclic products 5 (n = 4, 5, 10) (Scheme 1iii). This novel cyclization sequence, involving two independent C–C bond forming reactions and one original C–O cleavage, generates trans-fused bicyclo[4.3.0]nonanes exclusively. Mechanistic studies reveal a dichotomic nature of the photoredox initiation and a possible competition between 5-exo-trig and 6-endo-trig pathways, generally in favor of the second by virtue of radical philicity matching.²⁰ An unusual C–O fragmentation and hydrogen atom transfer (HAT) reactions liberate α-branched cyclic ketones of type 6 that are ideally suited for intramolecular aldol/ring closure reactions happening under mild acidic conditions with trans-selectivity only.

Results and discussion

In the context of cyclization strategies, it was then enticing to use compounds 2 and their malonate subunits to introduce halogenated chains under basic conditions (substrates 4, Table 1 and Scheme 2) and study subsequent reductions under photoredox conditions. Upon generation of electrophilic α-keto radicals, ring formation ought to occur onto the nucleophilic end of the enol ether moiety providing an effective discrimination and a preferred 6-endo regioselectivity. Preliminary studies were performed using enol ether 4a made from 2a by treatment with KHMDS (2.0 equiv.) at 25 °C. After rapid and quantitative deprotonation of the malonate group, addition of 2-bromo-2-methylpropanoyl bromide 3 (1.5 equiv.) afforded adduct 4a (64%). Then, under blue LED irradiation (λ_max 455 nm) using Ir(ppy)₃ (2.5 mol%) as photocatalyst (PCat) and diisopropylethylamine (DIPEA) as reductant and hydrogen donor (10 equiv.), the targeted reactivity was immediately observed in acetonitrile at 25–28 °C (Table 1, entry 1). ¹H NMR analyses of crude reaction mixtures were initially ambiguous as large differences were observed between spectra measured before and after a short filtration over a SiO₂ plug.²¹ In the latter case, spectra revealed an excellent conversion of 4a to major product 5a obtained as a single stereoisomer. This product was isolated upon silica gel chromatography (43%). Yet, NMR investigations of 5a were not agreeing with structures derived from simple 5-exo-trig or 6-endo-trig cyclization pathways. The connectivity indicated the occurrence of a rearrangement that could only be untangled upon X-ray analysis (CCDC 2058766). Single crystals of 5a were obtained by diffusion in a mixture of methylene chloride and pentane. The structural determination revealed a [4.3.0] bicycle with two trans-fused carbocycles; a cyclohexane and a densely-functionalized cyclopentanone. Care was then taken to unravel the reactivity prior to the silica gel treatment, i.e. before the cyclization to 5a. All crude NMR data pointed toward α-branched cyclohexanone 6a, which cannot be isolated. Further analysis of the crude mixture indicated, in retrospect, the presence of a minor component corresponding to spiro bicycle 7a (4–18%, Table 1). The origin of this moiety will be later discussed during the mechanistic analysis but such an occurrence was expected in view of the usual predominance of 5-exo-trig cyclization pathways and also results previously-obtained with malonate enol ethers.¹⁹ Given the unusual skeletal rearrangement of 4a to 6a and then trans-5a, the reactivity was investigated further. First, a screen of experimental conditions (Table S1) was performed to improve conversions and yields, an excerpt is detailed in Table 1. Substitution with [Ir(dFCF₃ppy)₂(bpy)]PF₆ as photocatalyst increased the yield up to 50% (entry 2). Additional amounts of HE (5 equiv.) were beneficial (5a, 60%, entry 3). In effect, HE (10 equiv.) could substitute DIPEA entirely (5a, 72%, entry 4). Classical Ir(ppy)₃ could be used with only 2.5 equivalents of HE to form 5a in 70% (entry 5). Various organic dyes absorbing in blue-green domains were also tested as photocatalysts (entries 6–8, and Table S1), including rose Bengal or eosin Y giving yields of 60% and 68%, respectively (entries 6 and 7). With mildly-reducing acridine orange (AO), to a bit of our surprise, bicycle 5a was still obtained in an effective yield (72%, entry 8); a rationale will be provided later (vide infra). Further variations of different Hantzsch esters, solvents, catalyst loading and substrate concentration, did not lead to any additional improvements (Table S1). As expected, reactions do not proceed in the absence of HE or blue light (entries 9 and 10). Without AO, 5a is nevertheless obtained in a surprising 19% yield, suggesting that HE can also play a photoreducing role (entry 11); this reactivity will be later exploited. Nonetheless, entry 8 combining (i) visible light photoredox catalysis and (ii) a silica gel filtration represented the most effective conditions to 5a, which were selected for the remainder of the study.

Scheme 2 (A) Acylated malonate enol ethers 4a to 4m. (B) Hydroxylated trans-bicycles 5a to 5m. ^a Yields determined by ¹H-NMR spectroscopy using 1,3,5-trimethoxybenzene as internal standard. Isolated yield of the major diastereomer in parenthesis. In grey, lack of product formation or open-armed product obtained.

Table 1 Optimization of the reaction conditions^a

Entry	Photocatalyst (PCat)	Additive	Yield 5a (%)	Ratio (6a/7a)^b
a Reaction conditions: 4a (0.05 mmol), PCat (2.5 mol%), additive, MeCN (0.1 M), 25–28 °C, blue LEDs, 3 h. Yields determined by ^1H-NMR spectroscopy using 1,3,5-trimethoxybenzene as internal standard. DIPEA: N,N-Diisopropylethylamine.b Ratios between 6a and 7a were determined by ^1H NMR analysis.c Isolated yields in parenthesis.d Without light. n.r.: no reaction.
1	Ir(ppy)3	DIPEA (10 equiv.)	48 (43)^c	85/15
2	[Ir(dFCF3ppy)2(bpy)]PF6	DIPEA (10 equiv.)	50	90/10
3	[Ir(dFCF3ppy)2(bpy)]PF6	DIPEA (10 equiv.), HE (5 equiv.)	60	82/18
4	[Ir(dFCF3ppy)2(bpy)]PF6	HE (10 equiv.)	72	91/9
5	Ir(ppy)3	HE (2.5 equiv.)	70	92/8
6	Rose bengal	HE (2.5 equiv.)	60	96/4
7	Eosin Y	HE (2.5 equiv.)	68	91/9
8	Acridine orange	HE (2.5 equiv.)	72 (66)^c	92/8
9	Acridine orange	None	Traces	—
10	Acridine orange	HE (2.5 equiv.)	n.r.^d	—
11	None	HE (2.5 equiv.)	19	92/8

---

In effect, a series of α-bromo gem-dimethyl ketones 4a to 4m were prepared from the corresponding malonate enol ethers (Scheme 2A). Overall, the acylation was effective (52–95% yields) under irreversible deprotonation conditions (KHMDS). Then, using conditions from Table 1 (entry 8), trans-fused bicyclic compounds 5 were obtained in low to good yields (15–66%, Scheme 2B).²² First, with 4a, the reaction was conducted in 1 mmol scale and proceeded without loss of conversion to afford 5a with a comparable isolated yield (66%). However, performed on a gram scale of starting bromide, the isolated yield of 5a decreased to 50%. Benzyl malonate enol ether 4b reacted similarly to give 5b in slightly lower 53% yield, a possible consequence of the increased steric hindrance. 7-Membered ring 5c formed efficiently (58%). The trans ring fusion was ascertained by X-ray diffraction analysis again (CCDC 2481640, Scheme 2). In the case of cyclopentanone-derived 5d, which was unproductive, the elevated strain associated with a trans-fused [3.3.0] structure²³ is probably responsible for the lack of ring closure observed.²⁴ 4,4′-Difluorinated enol ether 4e and dioxolane 4f reacted to yield 5e and 5f in 51% and 49% yields, respectively. However, gem-disubstituted 4g and 4h afforded the corresponding dimethylated and dicarboxyethylated products 5g and 5h in lower 38 and 26% (NMR yields), respectively. Overall, despite a lower productivity for diester 5g, the presence of (inductive) electron-withdrawing groups at 4,4′-positions seems more favorable. However, strong steric effects preclude cyclization as hexamethylated 5i could not be formed. Instead of trans-fused rings, 5-exo-trig cyclization is preferred forming the spirocycle adduct 7i (30% NMR and 23% isolated yields). The diastereoselectivity of the cyclization was briefly studied with 4-trifluoromethyl and 4-methyl-substituted enol ethers 5j and 5k. trans-Bicycles 5j (44% NMR yield) and 5k (55% NMR yield) were obtained as mixtures, with respective 3 : 1 and 2 : 1 diastereomeric ratios (dr). Only the major isomers could be isolated upon chromatography (14% and 18%, respectively); their structures (CCDC 2481641 and 2481639) presenting the 4-CH₃ or 4-CF₃ groups in axial positions. A rationale for the observed stereochemistry is provided in the SI (Scheme S4).²⁵ However, when the substrate is again strongly hindered or biased sterically like 5l with the tert-butyl substituent, a lack of desired reactivity is obtained. Finally, of importance for the mechanism, [10.3.0] bicycle 5m was obtained from 12-membered 4m in a low ¹H-NMR yield (15%) but alongside 6m (63%) as major product. This derivative 6m is not prone to enolization and cyclization. Yet, this observation of 6m led us to check how general was the formation of open-arm molecules. In the six-membered ring series, crude mixtures from reactions of 4a and 4f were verified immediately after irradiation, only to confirm the presence of α-functionalized cyclohexanones in all instances. The photoredox cyclization therefore stops at the C–O bond fragmentation and the formation of adducts of type 6. With these mixtures in hand, a small amount of silica gel (pipette filtration) was sufficient to provoke the intramolecular aldol to 5a and 5f.²⁶ Nevertheless, condensations were more reliable under soft-enolization Lewis acid–base conditions.²⁷ In fact, full conversion was obtained with either MgCl₂/Et₃N²⁸ or LiCl/DBU^27a to yield 5a (66%) in 15 and 30 minutes from 6a, respectively. Finally, care was taken to evaluate the reactivity of the trans-fused products of type 5 only to find that the tertiary alcohol group is unreactive under dehydration (e.g., Burgess reagent²⁹) or alkylative conditions (e.g., benzyl 2,2,2-trichloroacetimidate³⁰). However, upon addition of ^tBuOK at 25 °C, hydroxydecarboxylative elimination occurs. Mechanistically, it is proposed that β-lactone [8a] is formed and leads to 9a (85%) upon strain release and CO₂ cleavage.

In parallel, insight into the photocatalytic mechanism (4a 1.0 equiv., AO 2.5 mol% and HE 2.5 equiv.) was looked for (Fig. 1). Optical and electrochemical data of the reactants provided possible energy pathways upon light excitation (see SI).³¹ Absorption and emission spectra of AO (Fig. S1) afforded an optical energy gap (E₀₀) of 2.62 eV, with a reported reduction potential in acetonitrile of −2.0 V vs. SCE.³² As a consequence, the excited state AO* has E^red = +0.62 V (vs. SCE) and cannot perform the necessary oxidation of HE (E^ox = +0.93 V vs. SCE) by single electron transfer (SET). The overall process is unproductive (ΔG₀ > 0, see SI).³³ After this observation, the UV-Vis absorption spectrum of the crude reaction was analyzed further and it revealed a lower energy signal upon reaction of AO with HE in excess, which was matched with protonated acridinium AOH⁺ (see Fig. 1IIa). This cationic species possesses all the necessary properties in its excited state to perform photocatalysis (E₀₀ = 2.44 eV, E^red (AOH⁺*/AOH˙) = +1.72 V and E^ox (AOH⁺⁺˙/AOH⁺*) = −1.24 V vs. SCE) (Tables S2 and S3). In effect, by irradiating at 455 nm, AOH⁺ is efficiently excited and two quenching pathways are then energetically possible, either (i) a SET involving the oxidation of HE or (ii) the direct reduction of enol ether 4a (E^red = −1.0 V vs. SCE). In support, with Stern–Volmer experiments, quenching of AOH⁺ luminescence (λ_em = 525 nm, Φ = 37% and τ = 2.5 ns in acetonitrile) is detected for both HE and enol ether 4a with kinetic quenching constants (k_q) of 1.1 × 10¹⁰ M⁻¹ s⁻¹ and 1.6 × 10⁹ M⁻¹ s⁻¹, respectively (Fig. 1IIb). While quenching of HE is 10-fold faster than that of the enol ether 4a, quenching efficiencies (η) are similar with values of 28% and 20%, taking into account the available concentrations of HE and 4a in solution during the photocatalytic reaction (Fig. S10–S13).³⁴ Overall, path i (Fig. 1) remains more favorable than path ii (Scheme S2) and the difference in efficiencies increases over time as 4a is progressively consumed (Fig. S12). In any case, satisfactorily, both paths i and ii lead to the same product. With path i, shown in Fig. 1I, after formation of AOH˙ by quenching of AOH⁺* with HE, reduction of substrate 4a by SET leads to α-keto radical A.⁴⁴ With path ii, displayed in Scheme S2, intermediate A is formed directly by reaction of 4a with AOH⁺*. Then, independent of its origin, A undergoes a preferred 6-endo-trig cyclization to form B. Subsequent C–O cleavage occurs and affords stabilized tertiary malonyl C. Such homolytic C–O cleavage seems to be original to this transformation. Then, the highly electrophilic radical cannot readily cyclize to the intramolecular δ-ketone; preferred HAT of C with either HE or protonated HE⁺ (ref. 35) must occur and yield open-chain products 6 observed in crude mixtures, for 5m in particular. In terms of regioselectivity, confirmation was brought by computational analysis of the preferred 6-endo-trig (A → B) over the 5-exo-trig (A → D) cyclization pathways, both kinetically (ΔG^‡ = −2.9 kcal mol⁻¹) and thermodynamically (ΔG° = −3.2 kcal mol⁻¹) (Fig. 1III).

Fig. 1 (I) Synthesis of open-chained ketone 6a and spiro derivative 7a via 6-endo-trig (major, black) and 5-exo-trig (minor, grey) closures under photocatalysis (path i, AOH⁺, λ_exc 455 nm). (II) (a) Absorption and normalized emission spectra of reaction components in air-equilibrated acetonitrile 20–23 °C with concentrations 1 × 10⁻⁵ to 5 × 10⁻⁶ M and (b) Stern–Volmer plot of the luminescence quenching of AOH⁺ upon increasing concentration of quencher. Quenching experiments were conducted in dry acetonitrile under N₂ atmosphere with AOH⁺ (1 × 10⁻⁵ M), the emission intensity was followed at λ = 529 nm; I₀ is the intensity before and I after adding the quencher, [Q] stands for concentration of the quencher, in which Q is either HE (path i, this figure) or 4a (path ii, Scheme S2). (III) Gibbs free energy analysis of 5-exo vs. 6-endo cyclizations from first radical A to bicyclic B or spiro D radicals, and respective transition states F and E. (IV) Comparison of A with theoretical A1, A2 and A3 radicals, and subsequent regioselectivity (ΔΔG^‡, kcal mol⁻¹).

As the regioselectivity depended most probably on a philicity matching of first radical A with the enol ether moiety, care was taken to evaluate theoretically the influence of the substituents adjacent to the radical carbon center of A. DFT calculations using hypothetical intermediates A1 (gem-Me omitted), A2 (α-carbonyl omitted) and A3 (both elements missing) were executed and results compared to that of original A (Fig. 1IV). Energy barriers were calculated for 6-endo (As → Bs) or 5-exo (As → Ds) pathways via transition states Fs and Es, respectively (Fig. S15). As expected, in view of the polarization of the electron-rich alkene and its more nucleophilic terminal end, electrophilic α-carbonyl radicals A and A1 favor the formation of B and B1 via transition states F and F1 (ΔΔG^‡_F/E = −2.9 and −0.6 kcal mol⁻¹), respectively.³⁶ Then, changing the nature of radicals from electrophilic to nucleophilic, A2 and A3 formed preferentially spiro 5-membered rings through transitions states E2 and E3 (ΔΔG^‡_E/F = −3.0 and −3.5 kcal mol⁻¹), respectively. Then, with a clear explanation for the formation of open-chained derivatives 6, ring closures to bicyclic trans-fused 5 were also investigated computationally using mild Brønsted acid conditions as model (Scheme 3).³⁷ Direct cyclization of ketone 6a, used as reactant, through a concerted transition state with proton-transfer and C–C bond formation occurring at the same time was unsuccessful.³⁸ Then, we proceeded via all possible enols derived from 6a. Intermediates 6Aax and 6Aeq model most-stable enols with either axial or equatorial orientation of the pendant arm, and intermediates 6Bax and 6Beq, necessary for the reactivity, are the enols formed on the side of the malonate chains. The lower reaction pathway requires the formation of enol 6Beq prior to the intramolecular aldol. Its geometry favors a pre-organization of the arm leading to a late 6-membered transition state (TS2-trans) that achieves the trans-fused bicycle 5a by the lower activation barrier computed so far (ΔG^‡ = +29.9 kcal mol⁻¹).³⁹ In fact, without the presence of enol 6Beq, the cyclization from 6Aeq was located at +50.6 kcal mol⁻¹ as it models a less favored 7-membered ring transition state (Scheme S3). For good measure, analogous transition state TS2-cis, which would be necessary to reach a cis-fused bicycle, was located at 41.0 kcal mol⁻¹, more than 10 kcal mol⁻¹ higher than TS2-trans, in agreement with experimental observation. Additionally, formation of precursor 6Bax is strongly unfavored due to the required enol geometrical isomerization upon proton transfer from 6Aax. This motion breaks the π-delocalization and the H⋯O interaction, via TS1-cis (ΔG^‡ = +38.3 kcal mol⁻¹) which is also far too high. Preferred formation of trans-fused bicycle 5a is hence justified computationally.

Scheme 3 Computed reaction paths from synthetic intermediate 6a to trans-fused 5a and putative 5a_cis. Gibbs free energies in kcal mol⁻¹.

Finally, literature analysis indicated that HE can directly promote photoreductive processes, including selective debromination of α-bromo ketones.⁴⁰ In this context, reactions were tested without photocatalyst under strict anaerobic conditions but employing a LED of higher energy. Satisfactorily, productive results were obtained while irradiating at 365 nm as product 5a was isolated in 68% yield (Fig. 2I).⁴¹ In fact, with enol ether 4a as substrate, HE (E₀₀ = 3.06 eV and E^ox = +0.93 V vs. SCE) acts as photoreductant (HE* having E*^ox = −2.13 V). As it could be expected, Stern–Volmer analysis with gradual additions of 4a revealed an emission quenching of HE (Φ = 1.7% and τ = 0.32 ns in DMSO, of comparable polarity to acetonitrile). The SET process is characterized by a quenching constant of k_q = 4.2 × 10¹⁰ M⁻¹ s⁻¹ (Fig. 2II), corresponding to an efficiency η of 57% indicating a faster and more effective activation than the reactions derived from AOH⁺ as photocatalyst. In the present case, full reduction of 4a is enacted with 2.5 equivalents of HE in 1.5 hours only. Further investigation demonstrated a lack of photodecarbonylation after prolonged UV irradiation.⁴²

Fig. 2 (I) Cyclization with HE under UV light (365 nm) followed by soft enolization/cyclization. Yield determined by ¹H-NMR spectroscopy using 1,3,5-trimethoxybenzene as internal standard. (II) Stern–Volmer plot of the luminescence quenching of HE upon increasing concentration of 4a. Quenching experiments were conducted in dry acetonitrile under N₂ atmosphere with HE (5 × 10⁻⁵ M), the emission intensity was followed at λ = 455 nm; I₀ is the intensity before and I after adding the quencher 4a.

Conclusions

Highly functionalized trans-fused [n.3.0] bicycles (5) were synthesized (n = 4, 5, 10). The synthetic method combines photoinduced 6-endo-trig Giese additions with mild intramolecular aldol cyclizations. In practice, two distinct photoredox methods were considered. On one hand, photocatalytic conditions were developed using acridinium orange as a mild photocatalyst in the blue range (λ_exc = 455 nm), in combination with Hantzsch ester HE, which serves as both the reducing agent and hydrogen donor. On the other hand, HE was used for its own photoreductive properties at higher energy (λ_exc = 365 nm). Overall, these cyclization methods allow for the formation of unusually-dense trans-fused hydrindanones derivatives. While the process was mostly demonstrated for enol ethers made from cyclohexanone backbone, extension to other ring size remains a possibility. Finally, DFT calculations validate all the mechanistic pathways and regioselectivity in favor of a philicity matching of the electrophilic radical and the polar electron-rich nature of the malonate enol ether.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data that support the findings of this study are openly available in yareta.unige.ch at https://doi.org/10.26037/yareta:ehkvr2be4za3bchnlyjv6pgeie. It will be preserved for 10 years.

Supplementary information (SI) is available. See DOI: https://doi.org/10.1039/d5qo01635e.

CCDC 2058766 and 2481639–2481641 contain the supplementary crystallographic data for this paper. These data can be be obtained free of charge from The Cambridge Crystallographic Data Centre via https://www.ccdc.cam.ac.uk/structures.^43a–d

Acknowledgements

We thank the University of Geneva and the Swiss National Science Foundation for financial support (Grant 200020-207539 to J.L.). We thank Prof. Philippe Renaud (Bern) and Dr Fabrice Dénès (Nantes) for helpful discussions. The computations were performed at University of Geneva (organic chemistry department). Open access funding was provided by the University de Geneva.

References

1. (a) P. S. Baran, J. Am. Chem. Soc., 2018, 140, 4751–4755; (b) K. C. Nicolaou and T. Montagnon, Molecules that changed the world: a brief history of the art and science of synthesis and its impact on society, Wiley-VCH, Weinheim, 2008; (c) W. Liu, B. Hong, J. Wang and X. Lei, Acc. Chem. Res., 2020, 53, 2569–2586; (d) D. J. Newman and G. M. Cragg, J. Nat. Prod., 2020, 83, 770–803.
2. (a) N. Müller, T. Magauer and O. Kováč, J. Org. Chem., 2025, 90, 5083–5092; (b) S. J. Kwon and D. Y. Kim, Chem. Rec., 2016, 16, 1191–1203; (c) I. Omae, Coord. Chem. Rev., 2011, 255, 139–160; (d) J. Liao, X. Yang, L. Ouyang, Y. Lai, J. Huang and R. Luo, Org. Chem. Front., 2021, 8, 1345–1363; (e) S. H. Kyne, M. Clémancey, G. Blondin, E. Derat, L. Fensterbank, A. Jutand, G. Lefèvre and C. Ollivier, Organometallics, 2018, 37, 761–771.
3. J. Jose and T. V. Mathew, Tetrahedron, 2024, 150, 133747.
4. D. T. Hog, P. Mayer and D. Trauner, J. Org. Chem., 2012, 77, 5838–5843.
5. (a) C. Djerassi, L. Miramontes, G. Rosenkranz and F. Sondheimer, J. Am. Chem. Soc., 1954, 76, 4092–4094; (b) H. Li, H. J. Lee, Y. H. Ahn, H. J. Kwon, C.-Y. Jang, W.-Y. Kim and J.-H. Ryu, Biochem. Biophys. Res. Commun., 2014, 443, 132–137.
6. G. Xu, A.-J. Hou, Y.-T. Zheng, Y. Zhao, X.-L. Li, L.-Y. Peng and Q.-S. Zhao, Org. Lett., 2007, 9, 291–293.
7. (a) F. Berrue, O. P. Thomas, R. Laville, S. Prado, J. Golebiowski, R. Fernandez and P. Amade, Tetrahedron, 2007, 63, 2328–2334; (b) N. Kawahara, M. Nozawa, D. Flores, P. Bonilla, S. Sekita, M. Satake and K.-I. Kawai, Chem. Pharm. Bull., 1997, 45, 1717–1719; (c) N. A. Eddy and P. Ichalkaranje, in Molecules, 2016, vol. 21; (d) M. J. Brienne, A. Heymes, J. Jacques, G. Snatzke, W. Klyne and S. R. Wallis, J. Chem. Soc. C, 1970, 423–432.
8. P. Jankowski, S. Marczak and J. Wicha, Tetrahedron, 1998, 54, 12071–12150.
9. (a) B. Giese, Angew. Chem., Int. Ed. Engl., 1983, 22, 753–764; (b) P. Renaud, Chimia, 2001, 55, 1045.
10. T. RajanBabu, Acc. Chem. Res., 1991, 24, 139–145.
11. (a) K. J. Romero, M. S. Galliher, D. A. Pratt and C. R. J. Stephenson, Chem. Soc. Rev., 2018, 47, 7851–7866; (b) H. Y. Song, J. M. Joo, J. W. Kang, D.-S. Kim, C.-K. Jung, H. S. Kwak, J. H. Park, E. Lee, C. Y. Hong, S. Jeong, K. Jeon and J. H. Park, J. Org. Chem., 2003, 68, 8080–8087; (c) H. Zhang, S. Ma, Z. Xing, L. Liu, B. Fang, X. Xie and X. She, Org. Chem. Front., 2017, 4, 2211–2215; (d) D. P. Curran and C. T. Chang, J. Org. Chem., 1989, 54, 3140–3157; (e) S. Pradhan and S. Das, Helv. Chim. Acta, 2025, 108, e202400206; (f) J. Xuan and A. Studer, Chem. Soc. Rev., 2017, 46, 4329–4346.
12. (a) A. L. J. Beckwith and C. H. Schiesser, Tetrahedron, 1985, 41, 3925–3941; (b) I. V. Alabugin and K. Gilmore, Chem. Commun., 2013, 49, 11246–11250.
13. (a) T. P. Yoon, M. A. Ischay and J. Du, Nat. Chem., 2010, 2, 527–532; (b) J. W. Tucker, J. D. Nguyen, J. M. R. Narayanam, S. W. Krabbe and C. R. J. Stephenson, Chem. Commun., 2010, 46, 4985–4987; (c) N. A. Romero and D. A. Nicewicz, Chem. Rev., 2016, 116, 10075–10166; (d) M. H. Shaw, J. Twilton and D. W. C. MacMillan, J. Org. Chem., 2016, 81, 6898–6926; (e) J. J. Douglas, M. J. Sevrin and C. R. J. Stephenson, Org. Process Res. Dev., 2016, 20, 1134–1147; (f) A. L. Gant Kanegusuku and J. L. Roizen, Angew. Chem., Int. Ed., 2021, 60, 21116–21149; (g) S. P. Pitre and L. E. Overman, Chem. Rev., 2022, 122, 1717–1751; (h) J. D. Bell and J. A. Murphy, Chem. Soc. Rev., 2021, 50, 9540–9685; (i) M. A. Ashley, C. Yamauchi, J. C. K. Chu, S. Otsuka, H. Yorimitsu and T. Rovis, Angew. Chem., Int. Ed., 2019, 58, 4002–4006.
14. (a) C.-J. Wallentin, J. D. Nguyen, P. Finkbeiner and C. R. J. Stephenson, J. Am. Chem. Soc., 2012, 134, 8875–8884; (b) R. K. Dhungana, A. Granados, V. Ciccone, R. T. Martin, J. Majhi, M. Sharique, O. Gutierrez and G. A. Molander, ACS Catal., 2022, 12, 15750–15757.
15. (a) C. Walling and A. Cioffari, J. Am. Chem. Soc., 1972, 94, 6059–6064; (b) Y. Yamashita, Y. Ogasawara, T. Banik and S. Kobayashi, J. Am. Chem. Soc., 2023, 145, 23160–23166.
16. M. C. Maust, C. M. Hendy, N. T. Jui and S. B. Blakey, J. Am. Chem. Soc., 2022, 144, 3776–3781.
17. S. P. Green, K. M. Wheelhouse, A. D. Payne, J. P. Hallett, P. W. Miller and J. A. Bull, Org. Process Res. Dev., 2020, 24, 67–84.
18. (a) T. Achard, L. Egger, C. Tortoreto, L. Guénée and J. Lacour, Helv. Chim. Acta, 2020, 103, e2000190; (b) E. P. Kündig and F. R. Monnier, Adv. Synth. Catal., 2004, 346, 901–904.
19. J. Viñas-Lóbez, G. Levitre, A. de Aguirre, C. Besnard, A. I. Poblador-Bahamonde and J. Lacour, ACS Org. Inorg. Au, 2021, 1, 11–17.
20. F. Parsaee, M. C. Senarathna, P. B. Kannangara, S. N. Alexander, P. D. E. Arche and E. R. Welin, Nat. Rev. Chem., 2021, 5, 486–499.
21. This observation will be later considered and explained. It will lead to the development of soft enolization conditions to promote the final cyclization from 6 to closed trans-fused 5.
22. As noted in ref. 15b, formation of bicyclic adducts upon reductive olefin radical addition seems to give lower yields than acyclic or monocyclic derivatives.
23. Of note, as it could be expected, previous calculations indicate a thermodynamic preference for the cis- over the trans-isomer in bicyclo[3.3.0]octanones adducts. See H. L. Gordon, S. Freeman and T. Hudlický, Synlett, 2005, 2911–2914.
24. Three main derivatives compose the crude reaction mixture including (i) the open-chain ketone derivative of type 6 made by 6-endo-trig cyclization and malonate cleavage and (ii) the spiro derivative made by 5-exo-trig cyclization. The third derivative was not identified. Concerning the open-chain ketone, soft enolization conditions (vide infra) could not afford the cyclized product.
25. The selectivity originates in the (moderately) selective first addition of the tertiary radical onto the α-face of the alkene when the 4-substituent (CF₃ or Me) occupies an equatorial position.
26. Activated silica was shown to promote intramolecular cyclization on α,β-unsaturated 1,2-diones. See; E. A. Uhrich, W. A. Batson and M. A. Tius, Synthesis, 2006, 2139–2142.
27. (a) M. A. Blanchette, W. Choy, J. T. Davis, A. P. Essenfeld, S. Masamune, W. R. Roush and T. Sakai, Tetrahedron Lett., 1984, 25, 2183–2186; (b) D. F. Netz and J. L. Seidel, Tetrahedron Lett., 1992, 33, 1957–1958; (c) D. A. Evans, M. J. Dart, J. L. Duffy and D. L. Rieger, J. Am. Chem. Soc., 1995, 117, 9073–9074.
28. M. W. Rathke and P. J. Cowan, J. Org. Chem., 1985, 50, 2622–2624.
29. A. Chandra, S. R. Cheekatla, V. K. Vishwakarma and D. Kumar, Asian J. Org. Chem., 2025, 14, e202500338.
30. P. Eckenberg, U. Groth, T. Huhn, N. Richter and C. Schmeck, Tetrahedron, 1993, 49, 1619–1624.
31. (a) V. Balzani, P. Ceroni and A. Juris, Photochemistry and photophysics: concepts, research, applications, Wiley-VCH, Weinheim, 2014; (b) N. G. Connelly and W. E. Geiger, Chem. Rev., 1996, 96, 877–910.
32. T. T. Eisenhart and J. L. Dempsey, J. Am. Chem. Soc., 2014, 136, 12221–12224.
33. Formation of an electron donor acceptor (EDA) complex was furthermore ruled out by recording absorption spectra of AO while increasing the concentration of enol ether 4a.
34. As the quantity of HE is saturated in the reaction mixture ([HE]_sat = 0.02 M, SI), its concentration can be considered constant in time. Also, substrate 4a is quickly consumed lowering further its quenching efficiency (see SI, section S8).
35. S. Azizi, G. Ulrich, M. Guglielmino, S. le Calvé, J. P. Hagon, A. Harriman and R. Ziessel, J. Phys. Chem. A, 2015, 119, 39–49.
36. This theoretical analysis also explains the minimal amounts of spiro adduct 7a (ca. 7% NMR yield) in view of the relatively small difference in energy between transition states E and F.
37. (a) T. Kobayashi, H. Abe and H. Ito, J. Synth. Org. Chem., Jpn., 2019, 77, 1086–1095; (b) N. R. Cichowicz, W. Kaplan, Y. Khomutnyk, B. Bhattarai, Z. Sun and P. Nagorny, J. Am. Chem. Soc., 2015, 137, 14341–14348.
38. Several relaxed scans searching for this transition state to achieve the formation of the trans-fused bicycles 5a observed experimentally, where always going up in energy. The relative energy performing this transition state was over 70 kcal mol⁻¹, which was not in agreement with experimental observations and, therefore, this reaction path was discarded.
39. Currently, the activation barrier of 29.9 kcal mol⁻¹ is high in relation with the experimental room temperature conditions. Computationally, only one explicit proton is considered in the calculations, while experiments require general acidic conditions. This probably indicates that more than one proton is involved in the proton transfer or in the stabilisation of the TS via H-bonding. As a consequence, a lower energy barrier would result at room temperature.
40. (a) S. Fukuzumi, K. Hironaka and T. Tanaka, J. Am. Chem. Soc., 1983, 105, 4722–4727; (b) G. Li, R. Chen, L. Wu, Q. Fu, X. Zhang and Z. Tang, Angew. Chem., Int. Ed., 2013, 52, 8432–8436; (c) K. Nakajima, S. Nojima, K. Sakata and Y. Nishibayashi, ChemCatChem, 2016, 8, 1028–1032; (d) L. Buzzetti, A. Prieto, S. R. Roy and P. Melchiorre, Angew. Chem., 2017, 129, 15039–15043; (e) D.-L. Zhu, Q. Wu, H.-Y. Li, H.-X. Li and J.-P. Lang, Chem. – Eur. J., 2020, 26, 3484–3488; (f) G. Suresh Yedase, S. Venugopal, A. P. and V. Reddy Yatham, Asian J. Org. Chem., 2022, 11, e202200478; (g) S. Patel, A. Chakraborty and I. Chatterjee, Org. Lett., 2023, 25, 8246–8251; (h) N. Li, J.-L. Si and M.-Y. Xu, Chem. – Eur. J., 2025, 31, e202404116.
41. The reaction was performed in a one-pot procedure using an additional 2.5 equivalents of DBU in the initial reaction mixture. Under such conditions, the yield of 5a dropped to 15%. In addition, degradation of starting material 4a was observed by ¹H NMR spectroscopy along with other unidentified by-products.
42. J. E. Starr and R. H. Eastman, J. Org. Chem., 1966, 31, 1393–1402.
43. (a) CCDC 2058766: Experimental Crystal Structure Determination, 2026,  DOI:10.5517/ccdc.csd.cc2739th; (b) CCDC 2481639: Experimental Crystal Structure Determination, 2026,  DOI:10.5517/ccdc.csd.cc2p9bw7; (c) CCDC 2481640: Experimental Crystal Structure Determination, 2026,  DOI:10.5517/ccdc.csd.cc2p9bx8; (d) CCDC 2481641: Experimental Crystal Structure Determination, 2026,  DOI:10.5517/ccdc.csd.cc2p9by9.
44. In path i, the second single-electron transfer leading to A from 4a is not thermodynamically allowed stricto-senso but it could be promoted by an in-situ equilibrium between AOH^• and AO^•. Alternatively, excitation of AOH^• could occur. Both AO^• and AOH^•* are strongly reductive species.

---

Footnote

† These authors contributed equally to the work.

---