Modular arene functionalization by differential 1,2-diborylation

Abstract

Aromatic rings, also known as arenes, containing two or more adjacent different substituents are ubiquitously found in small molecule drugs¹. Strategies that can rapidly introduce diverse vicinal substituents to readily available precursors would greatly benefit generation of analogues of bioactive compounds², which, however, remain challenging to realize to date. The existing approaches for preparing vicinal difunctionalized arenes lack modularity, or regioselectivity, or generality. Here we report a nickel-catalyzed vicinal diborylation method that can directly install two chemically differentiated boryl groups in a regio- and site-selective manner using readily available aryl triflates or chlorides as substrates. This reaction operates under simple and mild conditions and is scalable. It also shows a broad substrate scope and excellent functional group tolerance. Given that each boryl group can be independently transformed into various functional groups, this method offers a modular, regioselective, and divergent approach to access diverse vicinal difunctionalized arenes, showing promise for constructing analogue libraries. The combined experimental and computational mechanistic studies reveal a highly unusual reaction pathway, involving the formation of a dearomatized gem-diboryl species and 1,2-boron migration. The site- and regioselectivity of this reaction are proposed to be controlled by steric interactions of the boryl groups with the nickel catalyst. The mechanistic insights gained in this investigation could have broad implications on developing other boron-mediated functionalization reactions.

By an order of magnitude, benzenoids are the most frequently encountered ring system in small molecule drugs, and nearly half of them contain vicinal substitution patterns^2,3 (Fig. 1a). Conventionally, the preparation of arenes containing adjacent different substituents relies on employing substrates with one existing functional group to introduce the second one at the vicinal position step-by-step (Fig. 1b). Typically, this approach lacks modularity or flexibility to alter the existing substituent, therefore inconvenient for constructing analogue libraries. Alternatively, the aryne-mediated approach allows for direct and modular vicinal difunctionalization of arenes; however, it is generally challenging to control the regioselectivity with electronically unbiased arynes^4,5. Additionally, the emerging palladium/norbornene (Pd/NBE) cooperative catalysis, also known as Catellani-type reactions, offers regioselective functionalizations of both ipso and ortho positions of various arenes^6–10. Nevertheless, this type of catalysis at this stage still faces a number of limitations. One of the most prominent weaknesses is that the kinds of functional groups that can be introduced by the Pd/NBE catalysis remain limited, especially for functionalizing the arene ortho position^9,11.

Fig. 1|. Construction of vicinal difunctionalized arenes.

a, Vicinal disubstituted arenes in small molecule drugs. b, Conventional methods to prepare vicinal disubstituted arenes. c, Ni-catalyzed regioselective formation of differentially 1,2-diborylated arenes leads to modular and diverse post-functionalization. ATP, adenosine triphosphate; R, any substituents; FG, functional groups; X, halides or other leaving groups; Nu, nucleophile; E, electrophile; NBE, norbornene; pin, pinacolate; dan, 1,8-naphthalenediamino.

On the other hand, borylated arenes are highly versatile precursors for introducing various functional groups through well-established reactions, such as Suzuki–Miyaura coupling, Chan–Lam coupling, Matteson-type reactions, etc¹². One can imagine that methods that can directly introduce two chemically differentiated boryl groups in a regio- and site-selective manner from readily available aryl substrates would be highly valuable¹³, as each boryl group can be converted into diverse functional groups independently (Fig. 1c). As a result, a library of vicinal difunctionalized arenes can be furnished through divergent post derivatizations of a common diborylated intermediate, which in turn should assist analogue synthesis and lead optimization in drug discovery. Despite scattered examples of diborylation of arenes^14–17, to the best of our knowledge, simultaneous installation of two different boryl groups remains unachieved yet. Here we report the development of a nickel-catalyzed regioselective differential diborylation of readily available aryl triflates or chlorides through an unprecedented mechanism, which provides a unique platform for preparing diverse vicinal functionalized arenes.

Reaction discovery and optimization

Our original design was to use the Pd/NBE catalysis to realize ortho, ipso-diborylation with Bpin-Bdan (2) as the reagent^18–21 (Fig. 2a). Considering that Bpin (LUMO coefficient: 0.33) is more electrophilic than Bdan (LUMO coefficient: 0.28), regioselective diborylation—installing Bpin at the ortho position and Bdan at the ipso position—was expected to be achieved. Interestingly, although the Pd catalysis failed to yield any diborylated product, the use of a Ni catalyst with an aryl triflate substrate²² (1a) led to the formation of differentially diborylated arene (3a) with Bpin at the ipso position and Bdan at the ortho position, which was unambiguously characterized by X-ray crystallography. The reaction can take place without NBE, suggesting a vastly different mechanism from the Catellani-type reactions. After further optimization of the reaction conditions, the diborylated product (3a) was obtained in 97% yield using Ni(cod)₂/PPhCy₂ as the catalyst, 1,4-diaza-bicyclo[2.2.2]octane (DABCO) as the base, and cyclohexane as the solvent (Fig. 2b, entry 1). The ipso Bdan-borylated product (3a’), likely formed through the conventional Miyaura-borylation pathway^23,24, was found as the major side-product. Control experiments indicated that the Ni complex, the phosphine ligand, and the base all play essential roles in this reaction (entries 2–4). Notably, DABCO is a particularly effective base, as other organic and inorganic bases were incompetent and/or gave more Miyaura-borylation product (3a’), though the exact reason remains to be uncovered (entry 5). Regarding the solvent effect (entry 6), less polar solvents generally worked better than polar ones. Various ligands have also been surveyed (entry 7), revealing that monodentate phosphines, such as PPh₃ and PCy₃, gave much higher yields than bidentate phosphines or N-heterocyclic carbene (NHC) ligands, with PPhCy₂ being optimal. It is noteworthy that reducing the Ni catalyst loading to 5 mol% still afforded excellent yield and selectivity (entry 8).

Fig. 2|. Reaction discovery and condition optimization.

a. The discovery of diborylation of arenes using Bpin-Bdan (2). b. Selected condition optimization and control experiments of diborylation of aryl triflate 1a. c. Selected condition optimization for using aryl halides as substrates. For experimental details, see the Supplementary Information, Sections 2 and 3. NBE, norbornene; BTMG, 2-tert-butyl-1,1,3,3-tetramethylguanidine; TBME, tert-butyl methyl ether; IMes, 1,3-bis(2,4,6-trimethylphenyl)imidazol-2-ylidene; TIPS, triisopropylsilyl. ^a.Yields were determined by crude ¹H NMR spectroscopy with CH₃NO₂ as an internal standard. ^bReaction time was 24 hours.

Simply replacing aryl triflates with aryl halides did not yield any desired diborylation product. We hypothesized that the triflate anion might play an essential role in this Ni catalysis. To test this hypothesis, additives containing triflate anions were employed (Fig. 2c). Indeed, the desired diborylated product (3a) can be accessed from the corresponding aryl chloride when adding a metal triflate, such as Zn(OTf)₂. The yield can be substantially increased by using more soluble TIPS-OTf as the additive. Besides aryl chlorides, the corresponding aryl bromide can also deliver the target product in moderate yield under the same condition.

Scope investigation

With the optimized reaction conditions (Condition A) in hand, the scope of the 1,2-diborylation of a series of aryl triflates was explored (Fig. 3). First, aryl triflates with different substitution patterns, including ortho, meta, para, and poly-substituted substrates, all showed excellent reactivity (3a to 3f). Less sterically hindered aryl triflates, such as simple phenyl triflate (3b) and para- and meta-substituted triflates (3c, 3d), were also suitable substrates. Their yields were substantially improved under a slightly modified condition (Condition B), in which the addition of TIPS-OTf and excess base can likely scavenge trace amounts of water, thereby suppressing undesired side reactions, such as ipso protonation and further Suzuki-coupling between the aryl triflate substrate and the Bpin moiety in the diborylated product. Interestingly, high C6 selectivity was observed using the meta-methyl substituted substrate (3d), which is possibly governed by the steric preference. In contrast, the major products from the aryl triflates with a meta-electron-withdrawing group contain the Bdan group at the C2 position (3w, 3x) (for a discussion of the C6 versus C2 selectivity, see Supplementary Information, Section 6.6). Owing to the mild reaction conditions, the reaction can tolerate a wide range of functional groups, such as fluoro group (3e), trifluoromethyl group (3m, 3w), trimethylsilyl group (3p), ether (3q, 3t), ester (3s), ketone (3u), trifluoromethoxy group (3v), pyridine (3aa), phthalimide (3ab), and trisubstituted alkene (3r). Besides monocyclic benzenoids, aryl triflates bearing a fused ring (3ac), indole (3ad), dihydrobenzofuran (3ae), naphthalene (3ac) and carbazole (3ag), can all undergo effective diborylation in a site- and regio-selective manner. It is noteworthy that a challenging substrate based on unprotected 3,4-dihydro-2(1H)-quinolone (3ah) proved amenable to the transformation under a slightly modified condition B (see Supplementary Information). Moreover, this method holds potential for late-stage functionalization of complex bio-relevant molecules, e.g., estrone (3ai), ezetimibe (3aj) and acetaminophen (3ak), in which the two chemically differentiated boryl groups could be sequentially transformed into various functional groups (vide infra).

Fig. 3|. Substrate scope of vicinal diborylation of aryl triflates.

All yields are isolated yields. Experiments were typically run with aryl triflates (1.0 equiv.) on 0.2 mmol scale. Reactions employing ortho-substituted aryl triflates as substrates were run under Condition A with 1.05 equiv. of Bpin-Bdan. Reactions employing meta- or para-substituted aryl triflates as substrates were run under Condition B with 3.0 equiv. of Bpin-Bdan. Boc, tert-butyloxycarbonyl. ^a.Ratio of positional isomers (Bdan at C₆:C₂). ^b.Ratio of positional isomers (Bdan at C₅:C₇). ^c.Ratio of positional isomers (Bdan at C₂:C₄). ^d.Run without TIPS-OTf.

Arguably, aryl chlorides represent the most abundant and economical aryl halide source. Employing the optimized conditions from Fig. 2c, the scope of aryl chloride substrates was briefly examined (Fig. 4). Simple aryl chlorides containing a phenyl group (5b), a terminal alkene (5d), fluoride (5e, 5g), boronic ester (5f), silyl ether (5c, 5h), and morpholine (5i) are competent substrates. In addition to arenes, heteroarenes such as 2-methylbenzo[d]thiazole (5j) and thiophene (5k) are also effective substrates, resulting in highly site-selectivity diborylation. Unprotected nitrogen-containing heterocycles are more challenging substrates, although 1H-pyrrolo[2,3-b]pyridine (5l) still afforded the desired diborylation product, albeit in low yield. Notably, aryl chlorides bearing highly reactive functional groups—including unprotected piperidine (5m), phenol (5n), benzyl alcohol (5o), and phenethyl alcohol (5p)—all reacted smoothly to deliver the corresponding diborylated products in good to synthetically useful yields. Interestingly, four boryl groups were efficiently installed using a dichloride substrate (5q). Moreover, complex aryl chlorides, such as those derived from indomethacin (5r), a 3,4-dihydropyrimidinone scaffold, vitamin E (5t), fenofibrate (5u), and bezafibrate, are also suitable for the Ni-catalyzed differential 1,2-diborylation. Remarkably, unprotected urea (5s), secondary amide (5v), and free carboxylic acid (5v) moieties were all tolerated.

Fig. 4|. Substrate scope of vicinal diborylation of aryl chlorides.

All yields are isolated yields. Experiments were typically run with aryl chlorides (1.0 equiv.) on 0.2 mmol scale. Reactions employing ortho-substituted aryl chlorides as substrates were run under Condition C with 1.05 equiv. of Bpin-Bdan. Reactions employing meta- or para-substituted aryl chlorides as substrates were run under Condition D with 3.0 equiv. of Bpin-Bdan. TBS, tert-butyldimethylsilyl. ^a. Quenched with TBAF.

Synthetic utility exploration

To enhance the practicality of this method, the air-sensitive Ni(cod)₂ catalyst can be replaced with Engle’s air-stable Ni(cod)DQ catalyst in this reaction²⁵, allowing convenient glove-box free operations (Fig. 5a). In addition, gram-scale synthesis was smoothly realized with the 1-admantyl-substituted substrate (1o). Next, to demonstrate that the Bpin and Bdan moieties can be individually transformed into diverse functional groups, some representative examples are included in Fig. 5b. First, both boron groups can be oxidized to chloride (6) and hydroxy groups (7) in high yields, offering a distinct approach to introduce 1,2-dichloride or catechol from a mono aryl chloride or phenol precursor. On the other hand, Bdan, considered as a protected or deactivated boronate²⁶, is typically much less reactive than Bpin, which provides an opportunity to develop “two-stage” functionalization of the diborylated products. For example, the Bpin can be selectively converted to alkynyl, aryl, alkyl, pyridyl, allyl, hydroxyl, and hydrogen groups via Suzuki couplings, oxygenation, or hydrogenation, respectively, with the Bdan group intact. Subsequently, the Bdan group in the resulting mono-functionalized intermediate can undergo further transformations, such as Suzuki coupling, oxygenation, chlorination, bromination, fluorination, Morken amination²⁷, and selenation²⁸. As a result, diverse vicinal difunctionalized analogues can be generated from the common diborylated intermediate, showcasing the versatility of this method. It is worth mentioning that most of the difunctionalized products are not trivial to access via benzyne chemistry or the Pd/NBE catalysis. Interestingly, simply by changing the sequence between Suzuki coupling and oxygenation, a pair of products with “swapped” functional groups (8b versus 9b) can be obtained. Notably, hydrogenation at the Bpin position can lead to the net functional group translocated product (10b).

Fig. 5|. Synthetic applications.

a, Using an air-stable Ni pre-catalyst to avoid glovebox operations and enabling gram-scale synthesis; b, “Two-stage” functionalization of the 1,2-diborylated products; c, Rapid, modular, and divergent syntheses of analogues of bioactive compounds. All yields are isolated yields. See Supplementary Information for experimental details.

The utility of this method was further explored in divergent and modular preparation of analogues of 2-arylpropionic acids, which are an important class of bioactive compounds and have been used for reversing AKR1C3 (aldo-keto reductase 1C3)-mediated drug resistance in cancer treatment²⁹, and inhibiting cyclooxygenase³⁰ and γ-secretase³¹ (Fig. 5c). The core of these compounds contains a 1,2,4-trisubstituted benzene ring. The prior approach requires a five-to-seven step process to prepare each analogue^29,30, in which the C2-substituent of the product—originated with the starting material—cannot be flexibly varied except through de novo syntheses. Using this method, the differentially diborylated intermediate (15) can be synthesized in one step from the corresponding commercially available aryl chloride, which then underwent assorted post-difunctionalization to allow divergent syntheses of six 2-arylpropionic acid analogues.

Mechanistic studies

The intriguing transformation and unusual regioselectivity motivated us to investigate the mechanism of the Ni-catalyzed diborylation. First, the reaction was not affected by light, and an excellent yield was obtained under the dark condition (Extended Data Fig. 1a). In addition, various efforts to trap or detect radical intermediates all failed (see Supplementary Information, Section 5.5). All these observations do not support a radical-involved mechanism. The oxidative addition of Ni(cod)₂ with the aryl triflate substrate (1a) took place at room temperature to afford an isolatable aryl-Ni(II) complex (22), which can efficiently yield the desired diborylation product either in a stoichiometric reaction or through the role as a catalyst (Extended Data Fig. 1b). These results suggest that aryl-Ni(II) complex 22 is likely an intermediate in the catalytic cycle. Moreover, the kinetic isotope effect (KIE) of the reaction between 1a and deuterated D-1a was measured (Extended Data Fig. 1c). The parallel experiments gave a KIE value of 1.1, while the value of the intermolecular competition KIE was 1.2. Interestingly, the intramolecular competition experiment using D-1b as the substrate also gave no obvious KIE. Altogether, these experiments indicate that C–H cleavage should not be involved in the turnover-limiting step (TLS) ³². Furthermore, the reaction of the vinyl triflate or chloride substrate (23) gave an unexpected ipso gem-diborylated product (24), which offers inspirations on the proposed mechanism (Extended Data Fig. 1d).

Density functional theory (DFT) calculations were next performed to elucidate the mechanism of the Ni-catalyzed diborylation and the origin of regioselectivity. The computed reaction energy profile (Extended Data Fig. 2a) reveals that the reaction proceeds via an S_NAr-type oxidative addition^33,34 (TS1) of the aryl triflate substrate to a bisphosphine-ligated Ni(0) complex (IM1). After cis/trans isomerization, the more stable trans isomer of the bisphosphine (aryl)Ni(II)(OTf) complex IM2 is formed. Several potential reaction modes between Bpin-Bdan and IM2 were explored computationally (see Supplementary Information, Section 6, for details). The most favorable pathway was identified as the σ-bond metathesis between a monophosphine-ligated (aryl)Ni(II)(OTf) complex and Bpin-Bdan (TS2, Extended Data Fig. 2a) to form a π-arene complex between Ar–Bdan and (Bpin)Ni(II)(OTf) (IM3). From IM3, a relatively facile migratory insertion via TS4 leads to a gem-Bdan,Bpin-substituted dearomatized intermediate IM4, which is indirectly supported by the experiments with the vinyl substrates (Extended Data Fig. 1d). The regioselectivities in both the σ-bond metathesis and subsequent dearomative migratory insertion are controlled by steric effects that preferentially place the larger Bdan group further away from the Ni center to avoid repulsions with the PPhCy₂ ligand (see the Supplementary Information for steric clashes in the less favorable transition states). Subsequently, IM4 undergoes a Bdan-mediated 1,2-shift via TS5 (Extended Data Fig. 2a, ΔG^‡ = 16.9 kcal/mol relative to IM4). This concerted process allows simultaneous 1,2-shifts of the anti-configurated Ni and Bdan substituents to form vic-Bdan,Bpin-substituted dearomatized intermediate IM5. In contrast, a competing syn-1,2 boron shift (TS6), which would lead to a different regioisomer via 1,2-migration of Bpin^35–37, is less favorable by 5.7 kcal/mol due to steric repulsions between the syn-migrating Bpin and Ni. This is consistent with the experimentally observed product regioselectivity, where Bdan is added to the ortho position of o-tolyl triflate 1a. The catalytic cycle concludes with a DABCO-assisted deprotonation and rearomatization via TS7, requiring 19.2 kcal/mol, to yield the 1,2-diborylation product 3a. Overall, σ-bond metathesis (TS2) was identified as the TLS, which is consistent with the lack of primary KIE in all competition experiments (vide supra, Extended Data Fig. 1c). The proposed catalyzed cycle is summarized in Extended Data Fig. 2b. Moreover, the selectivity between the two ortho positions in the reactions with 5-indolyl triflate 1ad and 2-naphthyl triflate 1af was also explored by computation. Excellent site-selectivity was observed experimentally in which the Bdan is installed on the C4 and C3 positions, respectively, for these two substrates (vide supra, Fig. 3). The computed transition state energy differences in the 1,2-boron shift step fully corroborate with the experimentally observed site-selectivity (Extended Data Fig. 2c). A deeper analysis revealed that the selectivity is controlled by the stability of the dearomatized intermediates (see Supplementary Information, Section 6.5).

Conclusion

In summary, we have developed a Ni-catalyzed differential diborylation of common aryl triflates and chlorides, which allows site- and regio-selective installation of two different boron groups at vicinal positions. The reaction operates under mild conditions and shows excellent chemoselectivity. The two boron moieties can be individually transformed into diverse functional groups in a modular manner. It is expected that this method could be valuable for late-stage functionalization of advance bioactive compounds. The rare mechanism and unusual mode of activation discovered here could offer new insights into the reactivity of metalloboron species. Efforts on developing dearomative borylation through trapping the dearomatized intermediates are ongoing.

Extended Data

Extended Data Fig. 1|. Selected experimental mechanistic studies.

a. Control experiment under the dark condition. b. Stoichiometric reaction and catalytic reaction based on aryl-Ni(II) complex 22. c. Kinetic isotope effect (KIE) studies. d. Generation of gem-diborylated products 24 from vinyl triflate or chloride 23. dr, diastereomeric ratio.

Extended Data Fig. 2|. Selected computational mechanistic studies.

a. Free energy profile of the Ni-catalyzed diborylation of o-tolyl triflate 1a. b. Proposed catalytic cycle based on the computational studies. c. Regioselectivity in boron 1,2-shift. All energies were calculated at the M06/SDD–6–311+G(d,p), SMD(cyclohexane)//B3LYP-D3/LANL2DZ–6–31G(d) level of theory. L, ligand.

Data availability

All the data generated or analyzed during this study are included in this article and its Supplementary Information. Crystallographic data for the structures reported in this study have been deposited at the Cambridge Crystallographic Data Centre (CCDC), under deposition numbers CCDC 2415595 (3a). These data are provided free of charge by the joint Cambridge Crystallographic Data Centre and Fachinformationszentrum Karlsruhe Access Structures service www.ccdc.cam.ac.uk/structures.