Radical Addition to Strained σ-Bonds Enables the Stereocontrolled Synthesis of Cyclobutyl Boronic Esters.

While radical additions to π-bonds are well established, additions to σ-bonds are far less explored. We have found that electron deficient radicals derived from alkyl iodides under visible light irradiation add to the central strained bond of bicyclobutyl (BCB)-boronate complexes and lead to 1,3-alkyl disubstituted cyclobutyl boronic esters in high yields, with full stereospecificity and high levels of stereoselectivity. Novel cyclobutyl-substituted structures, including peptide and steroid boronic ester derivatives can be accessed. Additionally, although the use of electron-rich alkyl iodides as radical precursors was found to be ineffective, an alternative route involving alkylsulfonylation of the BCB-boronate followed by reductive desulfonylation provided access to simple alkyl substituted cyclobutane products.

The field of radical chemistry 1 has witnessed a renaissance over the last decade, 2 triggered largely by the introduction of photoredox methods (electron transfer photocatalysis) 3-4 for the generation of open-shell species under mild conditions. The radicals generated have been utilized in a broad array of reactions including addition to π-systems (e.g. olefins, alkynes, aromatics), 2-4 radical-metal recombination 5 and radical-radical cross-coupling. 6 We and others recently showed that electrophilic radicals readily add to π-bonds of unsaturated boronate complexes 1 (Scheme 1a). 7 Subsequent one-electron oxidation of the resulting intermediate radical anion 2 triggered a 1,2-metallate rearrangement leading to densely functionalized boronic esters 3. We questioned whether less activated σ-bonds, embedded in a strained molecule, 8 could undergo analogous radical reactivity as this would significantly enhance the scope of both boron and radical chemistry. In support of this hypothesis, the addition of radicals to exceptionally strained σ-bonds has been observed, 9 although studies in this area remain limited. Specifically, we sought to add radicals to the central σ-bond of a strained bicyclobutyl (BCB)-boronate complex 4 (Scheme 1b), which we recently showed was capable of reacting with electrophilic palladium(II) intermediates. 10 Following radical addition, one-electron oxidation 7 would trigger a 1,2metallate rearrangement 11 leading to cyclobutyl boronic esters 6. Not only does this proposal present novel reactivity between boronate complexes and radicals, but it also offers a mechanistically distinct method to prepare challenging polysubstituted cyclobutanes. 12 Such entities have the potential to open up considerable chemical space due to the three readily diversifiable positions (radical precursor, 4 boronic ester substituent, and the boron atom 13 itself). Furthermore, cyclobutanes are of increasing interest in medicinal chemistry, 14 adding significant motivation for develop-ing this methodology. Herein, we show that radicals can indeed add to the central σ-bond of a strained BCB-boronate complexes, thereby providing a stereocontrolled synthesis of 1,3-dialkyl substituted cyclobutyl boronic esters 6.

Scheme 1. a) Radical addition to π-bonds of unsaturated boronate complexes. b) Proposed radical addition to σbonds of BCB-boronate complexes.
We began our studies by focusing on developing a radical trifluoromethylation process, 2c as the trifluoromethyl group is a highly sought-after moiety in medicinal chemistry. 15 For the optimization studies, the BCB-boronate complex 3a was generated in situ by adding t-BuLi to a mixture of sulfoxide 7 and cyclohexyl boronic acid pinacol ester 8a in THF. An excess (5 equiv.) of trifluoromethyl iodide-DMSO complex and 1 mol% of the photocatalyst Ru(bpy)3 2+ were then added under blue LED irradiation. After 16 h we were delighted to find that the expected boronic ester 6a was formed in good yield and moderate diastereoselectivity (Table 1, entry 1). A brief solvent screen showed that THF was superior to DMSO and MeCN (entries 2-3). Performing the reaction at -78 °C 16 was found to significantly improve the process, providing the desired compound 6a in high yield and as a single diastereoisomer (entry 4; stereochemistry established by 1 H-NMR NOE analysis, see SI for details). In line with our previous experience with vinyl boronates, 7b we found that the photocatalyst was unnecessary for the process (entry 5). Finally, the loading of radical precursor could be decreased to 3 equivalents, and the reaction time shortened to 1 hour, without any loss in reaction efficiency (entry 6). A control experiment highlighted the importance of light, as only traces of 6a were observed in the absence of blue LED irradiation (entry 7). Finally, performing the reaction in the presence of a radical inhibitor (TEMPO) led to complete inhibition (entry 8), and to the detection of the TEMPO-CF3 adduct (see SI), which is indicative of the involvement of radical species in the process. used. e 1 mol% of Ru(bpy)3Cl2 used. f Reaction carried out with 3 equiv. of CF3I·2DMSO, reaction time: 1 hour. g Reaction carried out without blue LED irradiation. h Reaction carried out in the presence of 1 equiv. of 2,2,6,6-tetramethylpiperidine 1-oxyl (TEMPO).
With the optimized conditions in hand, we explored the scope of the radical trifluoromethylation reaction by varying the structure of the starting boronic ester. The reactions were found to occur with high diastereoselectivity, and simple chromatographic separation of the isomers allowed isolation of the products as single diastereoisomers. A variety of alkyl groups could be accommodated on the boronic ester starting material, from a simple Me group up to a t-Bu group, showing that the reaction tolerates the full spectrum of steric demand and affords product 6b-d with high stereoselectivity (Scheme 2). Aromatic boronic esters were also successfully employed, although only moderate levels of diastereoselectivity were achieved (6e-g). 17 A cyclobutyl boronic ester was also explored, giving the strained boronic ester 6h, featuring two adjacent cyclobutyl rings with high diastereoselectivity. Enantioenriched boronic esters featuring a chiral center in the αposition gave 6i-k with complete retention of stereochemistry. The pyrrolidine boronic ester substrate is especially noteworthy as this is normally a poor migrating group 18 but worked well under our standard conditions giving 6k in good yield and with high stereocontrol. In addition, cyclobutyl boronic esters bearing complex chiral structures and biologically relevant substituents could be obtained through this methodology with high stereocontrol (6l-o).

Scheme 2. Scope of boronic esters for the radical trifluoromethylation reaction a
a Unless noted otherwise, isolated yields of single diastereoisomer in reactions carried out under a nitrogen atmosphere, using 1.2 equiv. of 7, 1.0 equiv. 8 and 3 equiv. of CF3I·2DMSO using Schlenk techniques on a 0.2 mmol scale. Diastereomeric ratio measured by 19 F-NMR analysis of the crude reaction mixture and refers to stereochemistry across the cyclobutane ring. b Reaction time: 24 hours, propionitrile used as solvent. c Compound isolated as a mixture of diastereoisomers.
We then sought to explore the scope of radical precursors that could be employed. Longer perfluoroalkyl chains could be used resulting in similarly high yield and stereoselectivity (Scheme 3, compound 6p). A number of alkyl halides carrying electron withdrawing groupsincluding nitriles, esters, primary amides and sulfonesreacted with high yield and moderate to good diastereoselectivity (6q-t). The stereoselectivity of the process was observed to follow the trend CN < COOEt < SO2Ph, indicating that steric factors influence the outcome of the reaction. Remarkably, dipeptide 6u and tigogenin derivative 6v could also be efficiently synthesized starting from radical precursors derived from the corresponding biologically relevant molecule, showcasing the applicability of this process for late stage functionalizations.

Scheme 3. Scope of radical precursors a
a Isolated yields of mixtures of diastereoisomers in reactions carried out under a nitrogen atmosphere, using 1.2 equiv. of 7, 1.0 equiv. 8a and 2 equiv. of iodide 5 using Schlenk techniques on a 0.2 mmol scale. Diastereomeric ratio measured by 1 H-NMR analysis of the crude reaction mixture. b Ru(bpy)3(PF6)2 (1 mol%) was used. c Reaction time: 24 hours.
Unfortunately, non-activated alkyl iodides (e.g. benzyl iodide, methyl iodide) were found to be unreactive in the process, presumably because the polarity of the electron rich radical did not match with the polarity of the electron rich BCB-boronate, limiting the scope of the methodology. To overcome this limitation, we considered the use of sulfones as traceless activating groups. 19 Pleasingly, sequential radical alkylsulfonylation followed by reductive cleavage of the sulfone enabled access to alkylated products 9a and 9b in high yields (Scheme 4). Interestingly, the cyclobutylmethyl radical intermediate involved in the desulfonylation did not undergo ring opening [k=4.5·10 2 s -l ], 20 showing that reduction/protonation are faster processes.

Scheme 4. Strategy to introduce simple alkyl groups
The synthetic utility of the products was demonstrated by carrying out a number of functional group interconversions of the bo-ronic ester. For example, cyclobutane 6a underwent oxidation to alcohol 10a, formation of the potassium trifluoroborate salt 21 10b, Zweifel olefination 22 to give alkene 10c and Matteson homologation 23 to afford boronic ester 10d (Scheme 5).

Scheme 5. Product functionalization reactions
The mechanism of this photochemical radical reaction of BCBboronate complexes likely bears similarities to that previously reported for vinyl boronate complexes. 7b We propose that the light sensitive alkyl iodide undergoes photolytic initiation generating radical 11 which adds to the strained central σ-bond of BCBboronate 3 leading to the electron-rich radical anion 12 (Scheme 6). Single electron transfer to another molecule of alkyl iodide 5 regenerates radical 11 and zwitterionic species 13, which undergoes fast 1,2-metallate rearrangement to give the final product 6. 24 We believe that the single electron-transfer step is also the stereoselectivity determining step. As depicted in Scheme 6, alkyl iodide 5 approaches radical intermediate 12 from the less hindered face of the cyclobutane ring, away from the alkyl iodide-derived substituent. As the electron is transferred, 1,2-migration occurs on the opposite lobe of the p orbital before the molecule has time to undergo bond rotation around the C + -B bond. This accounts for the cis-relationship between the two alkyl groups in 6.

Scheme 6. Proposed reaction mechanism
In conclusion, we have shown that bicyclobutyl boronate complexes are excellent radical traps for electrophilic radicals, providing a novel visible light-mediated three-component coupling. Furthermore, the reaction can be conducted in the absence of photocatalysts. The reactions occur with moderate to excellent diastereoselectivity, are fully stereospecific, and allow the synthesis of complex chiral cyclobutanes, including peptide and steroid derivatives. The reactivity is induced by strain release-driven radical addition to σ-bonds, a pathway that will no doubt find further ap-plication in the synthesis of cyclobutyl-containing natural products and drug candidates

Supporting Information
The Supporting Information is available free of charge on the ACS Publications website.

AUTHOR INFORMATION Corresponding Author
*v.aggarwal@bristol.ac.uk

Notes
The authors declare no competing financial interests. ACKNOWLEDGMENT M.S. thanks the EU for H-2020 Marie Skłodowska-Curie Fellowship (grant no. 744242) and EPSRC (EP/R012695/1) for financial support. We thank Dr. Adam Noble for proofreading this manuscript.