Main

Since the first study of the bicyclo[1.1.1]pentane (BCP) analogue of (S)-(4-carboxyphenyl)glycine in 19961, linear three-dimensional 1,3-disubstituted BCPs have emerged as promising bioisosteres for para-substituted benzenes in drug development2. For example, the BCP analogue of the γ-secretase inhibitor avagacestat, developed by Pfizer in 2012, exhibited comparable biological activity to the parent drug, while also displaying increased passive permeability and aqueous solubility3. Currently, mono- and difunctionalizations of [1.1.1]propellane via radical or anionic pathways are the most efficient and widely adopted strategies for synthesizing BCP derivatives (Fig. 1a(1))4,5,6,7,8,9,10,11,12,13,14. Although synthetically versatile for difunctionalization, [1.1.1]propellane is volatile and cannot be stored long term, even at −20 °C (refs. 4,14). Alternatively, cross-coupling reactions of stable BCP-based reagents, such as BCP iodides15,16,17,18,19, BCP boronates20,21, BCP redox-active esters22,23, BCP sulfonates24 and BCP thianthrenium reagents25,26, can be used for BCP introduction. Yet, most available BCP reagents only have a single reaction site and are typically used as end groups in coupling reactions, which necessitates the preparation of a new BCP reagent with specific substituents for each 1,3-disubstituted BCP analogue synthesis (Fig. 1a(2)). To date, the preparation of BCP analogues of pharmaceuticals still relies on lengthy and cumbersome de novo synthesis3,27. Although Uchiyama’s development of a silaborated BCP reagent provided an approach for 1,3-disubstituted BCPs, the transformations available with BCP–Si are limited to oxidation reactions28. The development of a versatile BCP core structure with two modification sites would significantly streamline the synthesis of sought-after 1,3-disubstituted BCPs (Fig. 1a(3)).

Fig. 1: Synthesis of BCP analogues of benzyl compounds.
figure 1

a, State-of-the-art of 1,3-disubstituted bicyclo[1.1.1]pentanes. b, Representative benzyl-containing drugs. c, Modular synthesis of methylene bicyclo[1.1.1]pentanes enabled by a thianthrenium-based reagent. OTf–, trifluoromethanesulfonate; Bpin, pinacol boronic ester; TMS, trimethylsilyl; Nu, nucleophile; 2c–3e, two-centre, three-electron bond.

Full size image

As important structural motifs containing benzene rings, benzyl groups are widely found in pharmaceuticals, for example, in donepezil29, imatinib30 and darapladib31. Of the top 200 top-selling drugs in 2023, about 45% contain benzyl structures (Fig. 1b)32. However, the current approach to their methylene BCP analogues involves multistep syntheses starting from the [1.1.1]propellane3,27. Transformations at the methylene site of stable BCP-based reagents are limited due to the steric hindrance imposed by the neopentyl-like group. We have previously reported BCP-based thianthrenium salts that can be used as efficient alkylating reagents in C–O, C–N and C–C cross-coupling reactions25,26. Given the cationic nature of thianthrenium salts, coupled with the excellent leaving-group ability of thianthrene (TT), we hypothesized that the use of TT-based neopentyl-like reagents could overcome the steric challenges typically encountered in SN2 reactions. We developed a stable iodobicyclo[1.1.1]pentylmethyl thianthrenium (IBM-TT+) reagent with dual reactivity that serves as a general platform for the synthesis of a wide range of BCP methylene derivatives (Fig. 1c). Distinct from reported BCP-based reagents, the incorporation of a thianthrenium group can exhibit electrostatic interactions with nucleophiles, thereby overcoming the challenge of steric hindrance typically encountered in SN2 reactions involving neopentyl-like electrophiles, such as BCP methylene iodide33,34. After nucleophilic displacement of thianthrene, the iodo group at the 1 position on the BCP scaffold can be transformed into many other functional groups directly, or engage in photoredox or transition-metal-catalysed radical reactions to access high-value structures.

Results and discussion

Development of the BCP-based thianthrenium reagent

Atom transfer radical addition of [1.1.1]propellane has proven to be an efficient approach to functionalized BCPs17,18,19. Based on atom transfer radical addition reactivity, we synthesized iodomethyl thianthrenium reagent 1 from methylene iodide and thianthrene in 90% yield (Fig. 2a). The incorporation of the cationic TT+ group to methyl iodide lowers the C–I bond dissociation energy from 63.2 kcal mol−1 to 58.7 kcal mol−1 (Supplementary Table 7). The subsequent synthesis of iodobicyclo[1.1.1]pentylmethyl thianthrenium reagent 2 can therefore be realized in just 3 min with 390-nm irradiation of 1 and [1.1.1]propellane. The reaction also proceeds with a 95% yield when using 405-nm visible light, albeit with a slightly longer reaction time of 5 min. Control experiments revealed that other light sources and thermal conditions are not as efficient. In addition, deuterium-labelled reagent d2-2 can be readily prepared through hydrogen isotope exchange in 2 (Supplementary Information, p. 11)35. Compound 2 can be synthesized on at least a 15-mmol scale as a non-hygroscopic, free-flowing, off-white powder with a melting point of 130 °C. It remains stable under ambient conditions for at least 6 months without observable decomposition. Although compound 2 is prepared from [1.1.1]propellane, its centralized production and potential commercial availability would minimize the need for practitioners aiming to prepare BCP derivatives to directly handle [1.1.1]propellane. Attempts to synthesize derivatives of 2, substituted at the methylene position, have thus far not been successful (Supplementary Fig. 4).

Fig. 2: Synthesis of iodobicyclo[1.1.1]pentylmethyl thianthrenium reagent 2.
figure 2

a, Discovery and optimization of 2. Carbon, grey; hydrogen, white; sulfur, yellow; iodine, purple. b, Computational reaction profile (PBE0-D3/x2c-TZVPall/CPCM(MeCN)) of the photolysis of 1 and subsequent insertion of [1.1.1]propellane, with key dissociation, ISC and phosphorescence rates given. c, The dominant highest occupied and lowest unoccupied natural transition orbital (HONTO and LUNTO) of the S1 state of 1 at the S0 geometry (isovalue, 0.02 a.u.); the σ*(C–I) and σ*(C–S) NBOs, their contributions to the LUNTO, and the LUNTO after projecting out σ*(C–I) and σ*(C–S) contributions are shown for comparison. a15-mmol scale. b5 min. c60 °C. ISC, intersystem crossing; LED, light-emitting diode; DCM, dichloromethane; DMF, N,N-dimethylformamide; ΔG, Gibbs free energy change, ΔE, energy difference; vert., vertical; equil., equilibrium; dissoc., dissociation; phosp., phosphorescence.

Full size image

Mechanism analysis

The reaction of 1 with [1.1.1]propellane is a chain reaction with a quantum yield of Φ = 76. To elucidate the initiation step, we determined the bond dissociation free energies (computed with ORCA36) of the C–I (50.4 kcal mol−1) and the C–S (homolytic, 33.4 kcal mol−1; heterolytic (yielding TT and ICH2+), 44.7 kcal mol−1) bonds (Supplementary Table 7). However, optimization of the structures of the S1 and T1 states of 1, with its S0 equilibrium geometry as an initial guess, both converged to structures with completely broken C–I bonds. The iodine atom forms a two-centre, three-electron bond with the second sulfur atom of TT (S–I bond length, 2.962 Å; Supplementary Table 8), again demonstrating the unusual and synthetically useful reactivity of the thiathrene scaffold (Fig. 2b). Natural transition orbitals of the S1 state at the S0 geometry reveal that the S1 state is dominated by local excitation on the TT group, plus charge transfer from TT to the C–I σ* orbital (26%). Contributions to the C–S σ* orbital can also be seen but are less pronounced than those of the C–I bond (13%, Fig. 2c). These observations indicate that the electronic excitation of 1 weakens both the C–I bond and the C–S bond, with the bond to iodide weakened more, which is also reflected by the Wiberg bond orders (Fig. 2b). The preferential weakening of the C–I bond can be explained by a lower natural bonding orbital (NBO) energy expectation value of the C–I σ* orbital (+0.15 eV) compared with the C–S σ* orbital (+2.34 eV), which leads to more electron density being accepted by the C–I σ* orbital upon excitation. Compound 1 is thus a rare example of where a stronger bond (here the CH2–I bond) is selectively photolysed in the presence of a weaker bond (here the S–CH2 bond). At the optimized S1 geometry, the S1 → T1 intersystem crossing (ISC) is extremely fast (k = 3.8 × 1013 s−1) and irreversible, and dominates over all other decay routes of the S1 state, which can be attributed to large contributions of the heavy iodine atom to the excited states as well as the El-Sayed-allowed (n(I) → σ*(C–I))37 character (Supplementary Figs. 12 and 13) of the ISC transition. The fast ISC rate indicates that the dissociation of the iodine atom occurs exclusively on the T1 manifold. After reaching the T1 state equilibrium, 1 can either decay to the ground state S0 or undergo S–I bond cleavage to yield the products TT-CH2· (A) and I·. Once radical A is formed, a radical chain reaction takes place through addition of electrophilic A to the fragile central C–C σ-bond of [1.1.1]propellane to form BCP radical B, followed by abstraction of an iodine atom from another molecule of 1 by B, to regenerate A and furnishing product 2. Notably, both steps are barrierless at the PBE038-D339,40/x2c-TZVPall41/CPCM42,43(MeCN) level, and the more accurate DLPNO-CCSD(T)44/x2c-TZVPall/SMD43,45(MeCN) method predicts only a 0.6 kcal mol−1 barrier of the former step and no barrier for the latter step (Supplementary Fig. 14), which are consistent with the experimentally observed high chain propagation efficiency. Finally, the chain propagation is likely terminated by the radical coupling between I· and B.

To gain insight into the substitution reactivity of the alkyl thianthrenium salt 2, comparative experiments between 2 and its diiodide analogue 2′ were conducted with different nucleophiles (Fig. 3a). The results demonstrate that thianthrenium reagent 2 effectively alkylates various nucleophiles, including sulfonamides (3), pyridines (4), carboxylates (5), phenolates (6) and thiols (7), with yields ranging from 75% to 96%. In contrast, the iodide analogue 2′ failed to produce the desired products with N- and O-based nucleophiles, possibly due to the steric hindrance of the neopentyl-like group. These observations are supported by density functional theory (DFT) calculations, where the barriers to SN2 displacement from 2 are about 5 kcal mol−1 lower than those from 2′ using acetate anion as the nucleophile (Fig. 3b). Based on quantum chemistry calculations, we attribute this difference in reactivity to the electrostatic attraction in the SN2 transition state (TS) of 2 compared to 2′. Specifically, in TS-TT, an electrostatic attraction between acetate anion and 2 leads to a 279.2 kcal mol−1 stabilization of the TS, compared to only 188.4 kcal mol−1 for TS-I. Although the stabilization is largely offset by solvation effects, which reduce the interaction energy of 2 and OAc− more than the analogous interaction of 2′ and OAc−, the electrostatic interaction dominates the barrier difference between TS-TT and TS-I. The electrostatic attraction of cationic TT+ with the nucleophile thus helps to compensate for the reaction barrier, even in highly polar solvents such as MeCN. We note that the nucleophile does not have to carry an overall negative charge to display this electrostatic attraction; the neutral nucleophile 4-dimethylaminopyridine (DMAP) also reacts faster with 2 than with 2′, which is attributed to electrostatic attraction between TT+ and the partially negatively charged pyridine nitrogen atom of DMAP (Supplementary Information, p. 60). An alternative explanation for the high reactivity of 2 would be the addition of nucleophiles to the positively charged sulfur centre of 2, creating a sulfurane intermediate with a four-coordinate sulfur centre, from which product could form through reductive elimination46,47,48. However, our computations failed to locate any intermediates where the nucleophile–sulfur distance is consistent with a covalent Nuc–S(TT) bond (Supplementary Figs. 9 and 11). Given the diversity of nucleophiles, the sulfurane/ligand-coupling pathway cannot be completely ruled out in other cases.

Fig. 3: Mechanistic studies of substitution of 2.
figure 3

a, Alkylation of different nucleophiles with 2 and 2′. Yields were based on 1H NMR spectroscopy. b, DFT calculations of substitution of 2 and 2′ with OAc− as nucleophile (kcal mol−1). Carbon, grey; hydrogen, white; sulfur, yellow; oxygen, red; iodine, purple.

Full size image

Modular synthesis of methylene bicyclo[1.1.1]pentanes

Due to the inherent reactivity differences between the thianthrenium and iodide substituents, chemoselective functionalization of the TT+ followed by iodide substitution enables the modular synthesis of differently substituted methylene BCPs. With the discovery of the S-adenosylmethionine (SAM) superfamily, a variety of alkyl sulfonium salts have been developed by chemists as alkylating reagents through one- or two-electron processes in organic synthetic chemistry25,26,49,50. Inspired by these advances, various nucleophiles can undergo substitution reactions with 2, such as cyanide anion (8), azide anion (11), amine (13), imide (14), sulfinate anion (15), carboxylic acids (5, 17), thiols (7, 19) and phenol (21) (Fig. 4). Due to the redox properties of the thianthrenium group, the IBM-TT⁺ reagent can also engage in copper-catalysed cross-coupling reactions via single-electron transfer, enabling the formation of BCP analogues of diarylmethanes—key structural motifs in pharmaceutically active compounds (12)51. However, in the absence of copper catalyst, the photoinduced radical reactions with reagent 2 were unsuccessful, probably due to the instability of the methylene radical adjacent to the BCP structure (Supplementary Fig. 6). The high reactivity of the thianthrenium reagent enables mild conditions applicable for late-stage functionalizations of natural and medicinal molecules, such as betahistine (13), sulfisoxazole (3), indomethacin (5) and ezetimibe (6).

Fig. 4: The substrate scope of nucleophiles.
figure 4

General conditions: H–Nu (1.0 equiv.), 2 (1.0–1.3 equiv.), K2CO3 (1.0–2.0 equiv.), MeCN (0.1 M), 25 °C, 12 h. aTBANu (1.2 equiv.) was used, 10 min–12 h. b2 (1.0 equiv.), PhMgBr (2.0 equiv.), CuCl2 (2 mol%), THF (0.2 M), 0 °C, 15 min. c2 (1.0 equiv.), PhSO2Na (1.5 equiv.), DMF (0.2 M), 65 °C, 12 h. dd2-2 (93% D) was used. e(PhSe)2 (0.55 equiv.), NaBH4 (2.0 equiv.), then 2 (1.0 equiv.), ethanol/MeCN (0.1 M), 25 °C, 12 h. TBA, tetrabutylammonium; THF, tetrahydrofuran.

Full size image

The resulting functionalized BCP iodides can be used in a variety of different reaction manifolds for the second transformation. For example, they can undergo lithium–halogen exchange followed by reaction with electrophiles to yield BCP carboxylic acids (23), alcohols (24), aldehydes (25), boronates (27), silanes (28) and stannane (29) (Fig. 5(1)). In addition, deiodination of 6 can be realized under mild conditions using a halogen-atom transfer reagent and silanes as the hydrogen donor (30). The Anderson16,17,18 and Qing groups52 have demonstrated that BCP iodides can be used as radical precursors under photoredox conditions (Fig. 5(2)). Examples include C–S cross-coupling (31), Giese-type radical addition (32, 35) and C(sp3)–C(sp2) cross-coupling (36). Furthermore, BCP iodide can be used in transition-metal-catalysed cross-coupling reactions (Fig. 5(3)), for example, copper-catalysed C(sp3)–N cross-couplings (37, 38)53. The combination of such transformations with the new thianthrenium chemistry presented here provides access to a diverse set of high-value benzyl bioisosteres that are otherwise challenging to obtain. Due to the higher reactivity of the TT+ group under both substitution and photoredox conditions, functionalization of TT must precede functionalization of the iodine atom from compound 2 (Supplementary Fig. 4).

Fig. 5: The functionalization of BCP iodides.
figure 5

aBCP iodide (1.0 equiv.), tert-BuLi (2.2 equiv.), diethyl ether (0.28 M), −78 °C, 1 h, then FG-X (1.1–3.0 equiv.), −78 to 25 °C, 2 h; FG-X used in the formation of 22–29 (in order): MeOD, CO2, Bpin-OiPr/NaBO3, ethyl formate, PhCHO, Bpin-OiPr, TMSCl, Sn(nBu)3Cl. bBCP iodide (1.0 equiv.), BEt3 (40 mol%), 2-mercaptoethanol (2.0 equiv.), nBu3SnH (4.0 equiv.), THF (0.2 M), 25 °C, 12 h. cBCP iodide (1.0 equiv.), 2-(phenylthio)isoindoline-1,3-dione or 2-(phenylsulfonyl)benzo[d]thiazole (3.0 equiv.), Ir[dF(CF3)ppy]2(dtbbpy)PF6 (2.5 mol%), Na2CO3 (2.0 equiv.), TMS3SiOH (2.0 equiv.), DCM (0.15 M), blue LEDs, ∼30 °C, 24 h. dBCP iodide (1.0 equiv.), phenyl vinyl sulfone or 4-ethinylanisole (6.0 equiv.), Ir[dF(CF3)ppy]2(dtbbpy)PF6 (2.0 mol%), K3PO4 (2.0 equiv.), TMS3SiH (1.0 equiv.), cyclohexane (0.10 M), ∼30 °C, 24 h. eBCP iodide (1.0 equiv.), methyl 2-(bis(tert-butoxycarbonyl)amino)acrylate (6.0 equiv.), Ir[dF(CF3)ppy]2(dtbbpy)PF6 (2.5 mol%), Na2CO3 (2.0 equiv.), TMS3SiH (2.0 equiv.), methanol/H2O (0.15 M), ∼30 °C, 24 h. fBCP iodide (1.0 equiv.), Fe(acac)3 (20 mol%), TMEDA (40 mol%), Grignard reagent (1.6 equiv.), 25 °C, 2 h. gBCP iodide (2.0 equiv.), R2NH (1.0 equiv.), Cu(TMHD)2 (50 mol%), K3PO4 (3.0 equiv.), DMF (0.10 M), 100 °C, 24 h. FG, functional group; PC, photocatalyst; TM, transition metal; Bpin, boronic acid pinacol ester; TMS, trimethylsilyl; Boc, tert-butoxycarbonyl; dF(CF₃)ppy, 2-(2,4-difluorophenyl)-5-(trifluoromethyl)pyridine; dtbbpy, 4,4′-di-tert-butyl-2,2′-dipyridyl. Cy, cyclohexyl; acac, acetylacetonate; TMEDA, N,N,N′,N′-tetramethylethylenediamine; TMHD, 2,2,6,6-tetramethyl-3,5-heptanedione; SET, single-electron transfer; XAT, halogen-atom transfer.

Full size image

Synthetic applications

Although 2 has so far been utilized primarily for the synthesis of BCPs containing a methylene group, the modular synthesis concept and its benefits are demonstrated through the successful preparation of two BCP analogues (Fig. 6). Photoinduced iridium-catalysed deiodination of 39 yielded the first BCP analogue 40 of donepezil—used to treat Alzheimer’s disease29—in 93% yield. In addition, the iodo atom of 41 was transformed into a carboxylic acid group via lithiation and electrophilic trapping with CO2, followed by amide condensation to produce BCP-imatinib 42. Compared to the previous eight-step approach to 4227, our method shows the versatility and efficiency in the preparation of BCP bioisosteres enabled by 2.

Fig. 6: Synthesis of BCP pharmaceutical analogues.
figure 6

Syntheses of (1) BCP-donepezil and (2) BCP-imatinib. HATU, hexafluorophosphate azabenzotriazole tetramethyl uranium; DIPEA, N,N-diisopropylethylamine.

Full size image

Conclusion

We developed a stable bifunctional iodobicyclo[1.1.1]pentylmethyl thianthrenium reagent, IBM-TT+. By exploiting the distinct reactivity of thianthrenium and iodine groups, this reagent enables the efficient and modular introduction of various functional groups onto BCP scaffolds that can now be used as core structure, to be functionalized at both termini, as opposed to currently available BCP end groups. We have efficiently synthesized BCP analogues of known benzyl-containing pharmaceuticals, highlighting the potential of reagent 2 in drug development. Moving forward, we envision that enhancing the practical application of our modular approach using reagent 2—through methods such as continuous-flow chemistry and the development of analogues of 2 that lack or incorporate substituted methylene groups—will be a promising avenue for future research.

Methods

General procedure for alkylation of nucleophiles with 2

Under nitrogen atmosphere, to a 4-ml borosilicate vial equipped with a magnetic stir bar were added nucleophiles (0.100 mmol, 1.00 equiv.), 2 (74.6 mg, 0.130 mmol, 1.30 equiv.), K2CO3 (27.6 mg, 0.200 mmol, 2.00 equiv.) and anhydrous MeCN (1.0 ml, 0.1 M). The vial was sealed with a septum cap. Then, the mixture was stirred for 12 h at 25 °C, and concentrated under reduced pressure. The residue was purified by flash column chromatography on silica gel to afford the desired product. Note that for nucleophiles without protons, no base is needed.

General procedure for lithium–halogen exchange of BCP iodides followed by reactions with electrophiles

Under an argon atmosphere, to a solution of BCP iodide (0.100 mmol, 1.0 equiv.) in anhydrous diethyl ether (0.35 ml, 0.28 M) at −78 °C was added tert-BuLi (0.13 ml, 1.7 M in pentane, 0.22 mmol, 2.2 equiv.) dropwise. The resulting yellow solution was stirred for 1 h at −78 °C, then electrophiles (0.11–0.30 mmol, 1.1–3.0 equiv.) in anhydrous diethyl ether (0.35 ml) were added at −78 °C. The resulting mixture was stirred at −78 °C for 1 h and 25 °C for 1 h. The mixture was then quenched with saturated aqueous NH4Cl (∼2 ml). The aqueous phase was extracted with ethyl acetate (3 × 3 ml). The combined organic phase was dried over Na2SO4, filtered and the solvent was removed under reduced pressure. The residue was purified by flash column chromatography on silica gel to afford the desried product.