Synthesis of 3-Aryl-1-aminopropane Derivatives: Lithiation–Borylation–Ring-Opening of Azetidinium Ions

Abstract In situ generated 2-phenyl-azetidinium ylides react with boronic esters to form acyclic γ-dimethylamino tertiary boronic esters. The transformation is believed to involve the formation of a zwitterionic boronate, which subsequently undergoes ring-opening 1,2-migration, which is promoted by the relief of ring strain. Owing to the configurational instability of the initially formed ylides, which appear to be in equilibrium with the open-chain carbene form, the reaction is not stereospecific. The C–B bond of the γ-dimethylamino tertiary boronic esters can be transformed into a variety of functional groups (C–OH, C–vinyl, C–H, C–BF3), thus giving a diverse selection of 3-aryl-1-aminopropanes, which represent a privileged motif among drug molecules.

Boronic esters are arguably the most versatile of organic functional groups. This group can be transformed to introduce C-O, C-N, C-C, or C-X bonds under mild conditions, a characteristic that makes boronic esters (and boronic acids) extremely valuable late-stage intermediates in medicinal chemistry programs. 1 Although this value has mostly been demonstrated in the context of sp 2 -hybridised boron-bearing carbon centres (the transformation of aryl and vinyl boronic esters and acids through Suzuki-Miyaura cross-coupling), 2 the diversification of sp 3 -hybridised boron-bearing carbon centres is much less common, despite the identified need to populate compound libraries with such 3D molecules. 3 However, the relatively recent development of robust and generally applicable methods for the enantioselective preparation of secondary and tertiary alkyl boronic esters, 4 together with stereospecific methods for the subsequent transformation of the C-B bonds, 5 is expected to lead to a step-change in the use of organoboron chemistry in the pharmaceutical industry.
One of the most privileged structural motifs among marketed drugs and drug candidates is the 3-aryl-1-aminopropane unit (Figure 1). 6 Bedaquiline, an antituberculosis agent, is one of the more high-profile and structurally complex members of this class of drug molecule (Figure 1). 7 This motif can be introduced through the addition of aryl metal reagents to β-aminoketones, 8 the hydroformylation/reductive amination of styrenes, 9 the Heck-Matsuda addition of aryl halides/pseudohalides to allyl amines 10 and the electrophilic addition of amine-containing electrophiles to diarylmethyl anions. 11 We were particularly interested in accessing a diverse selection of 3-aryl-1-aminopropanes 1 through C-B functionalisation of γ-dimethylamino tertiary boronic esters (Scheme 1). We envisioned that these boronic ester intermediates could be accessed through lithiation-borylation of Previous experience within our laboratory gave us cause to be both optimistic and hesitant. For example, we have shown that such ring-opening lithiation-borylation reactions can be carried out with both epoxides 7 and N-Boc aziridines 9 to give the corresponding products 8 and 10 with high levels of enantiospecificity (Scheme 2). 12,13 However, this type of reaction failed with N-Boc azetidines 11, presumably owing to the ring-opening 1,2-migration step being slow (Scheme 2). 14 However, the recent work of Couty and David suggested that ring-opening of the corresponding azetidinium ions might be more facile. 15 Specifically, they showed that cyano-substituted azetidinium ions 3a can be deprotonated with LiHMDS to give ylides 4a that can be trapped with electrophiles; when the electrophile is an aldehyde, ketone, or acrylate, the resulting alkoxide or enolate undergoes an intramolecular S N 2 reaction to open the azetidinium ion (Scheme 2). The ylides were chemically unstable, even at cryogenic temperatures, necessitating that the electrophile be present during their formation. These studies also showed that these stabilised azetidinium ylides are configurationally unstable, even within the short lifetime imposed by in situ trapping. However, the level of con-figurational stability of less stabilised ylides, such as those derived from phenyl-substituted azetidinium ions, was unclear.

Scheme 2 Previous studies from our laboratory and that of Couty and David
To test our reaction, we first prepared the triflate salt of phenyl-substituted azetidinium ion 3b in four steps from commercially available 3-chloro-1-propiophenone (13) (Scheme 3). Ketone 13 was first reduced to the alcohol 16 and then converted into the corresponding dichloride. 14b A ring-closing double displacement reaction with methylamine gave azetidine 14, which was subsequently N-alkylated in good yield to give azetidinium ion 3b. 14b,15b

Scheme 3 Synthesis of phenyl-substituted azetidinium ion 3b
With azetidinium ion 3b in hand, we subjected it to the conditions similar to those established by Couty and David 15b -LiHMDS (1.7 equiv), THF, -78 °C, 1 hour, then warming to room temperature-in the presence of EtBpin. Under these conditions, γ-dimethylamino tertiary boronic ester 2b was isolated in 41% yield (Table 1, entry 1). When the putative ylide was generated in the absence of the boronic ester, which was subsequently added, the desired product was not observed, thus highlighting the instability of the ylide. Increasing the amount of base to 3.0 equivalents, adding the base at -20 °C, or subsequently warming the reaction mixture to reflux (to promote 1,2-migration) did not result in improved yields (  Having established optimum conditions for this transformation, we explored the scope of the methodology by testing a range of boronic esters (5a-i) (Scheme 4). The use of primary (5a-c) and secondary boronic esters (5d and 5e) gave moderate to good yields of the corresponding γ-dimethylamino tertiary boronic esters 2ba-be. Additionally, the use of allylic boronic ester 5f gave the corresponding product in 45% yield. Aryl boronic esters proved to be more challenging. When electron-rich boronic esters, such as 5g, was employed the expected product 2bg could be isolated in good yield. However, when more electron-poor aryl boronic esters were used, such as phenyl-and 2-thienyl-boronic ester, the corresponding boronic ester products could not be isolated owing to facile protodeboronation. Pleasingly, the unstable boronic ester products could be functionalised in situ, prior to work-up, to give more stable derivatives; the addition of aqueous H 2 O 2 /NaOH to the reaction mixture led to the corresponding tertiary alcohols 15bh and 15bi being isolated in good yields. Interestingly, in our experience, similar tertiary boronic esters, not containing a dimethylamino group, do not undergo protodeboronation so readily, 4n,17 suggesting that complexation of the boronic ester with the proximal amino group promotes fragmentation. Because protodeboronation can be desirable-γ-amino diarylmethines are prominent members of this family of therapeutics ( Figure 1)-we sought conditions to effect this transformation more efficiently. Considering our previously reported conditions for the protodeboronation of diarylalkyl boronic esters, 5c upon lithiation-borylation of 3b and 5j, the reaction mixture was warmed to room temperature and CsF (1.5 equiv) and H 2 O (1.1 equiv) were added sequentially; within 1 hour of stirring the resulting mixture at room temperature, the tertiary boronic ester was completely consumed and, subsequently, diarylmethine 16bj could be isolated in 71% yield (Scheme 5).

Entry
Base Equiv of base T (°C) Yield of 2b (%)

Paper Syn thesis
To further demonstrate the versatility of these tertiary boronic esters, we subjected 2ba to a variety of conditions to effect functionalisation of the C-B bond (Scheme 6). In situ oxidation of 2ba using H 2 O 2 /NaOH gave the corresponding tertiary alcohol 15ba in 58% yield. Other oxidizing conditions (NaBO 3 ·4H 2 O, NaBO 3 ·4H 2 O/CH 3 COOH, 18 Oxone/acetone, 19 TMANO·2H 2 O 20 ) were less effective and led to partial oxidation of the tertiary amine group. Olefination of isolated 2ba under modified Zweifel conditions 21 with vinyl lithium gave alkene 17ba in 52% yield. Protodeboronation of isolated 2ba using TBAF·3H 2 O gave the desired γ-dimethylamino aryl-dialkyl methine 16ba in 73% yield. 5c We also attempted to transform 2ba into the corresponding trifluoroborate salt using standard conditions (KHF 2 , MeOH). 22 However, the trifluoroborate moiety underwent partial ligand exchange with the pendant amine to give a mixture of the desired trifluoroborate salt and intramolecularly complexed difluoroborane 18ba (2:1, 82% overall yield), as determined by 1 H and 19 F NMR analysis. 23 Pleasingly, when the crude mixture was heated at reflux in MeCN, complete conversion into 18ba was effected (67% yield). We then set out to understand the mechanism of the transformation. First, we prepared pyrrolidinium ion 19a and subjected it to the lithiation-borylation conditions to investigate the contribution of relief of ring strain in the putative 1,2-migration step. The substrate, 19a, was prepared through reductive alkylation/alkylation of commercially available 2-phenylpyrrolidine. Upon treatment of a THF solution of 19a and EtBpin at -78 °C with LDA (2 equiv) with subsequent warming to room temperature, 1,2,3,4,7,8-hexahydroazocine 20a (48%) was initially isolated, a species that in solution isomerised over a short period of time into the corresponding benzo-fused 1,2,3,4,5,8hexahydroazocine 21a; the desired tertiary boronic ester was not observed (Scheme 7). Hexahydroazocine 20a presumably arises from a Sommelet-Hauser rearrangement: proton transfer of the initially formed benzylic ylide to the methylenic ylide followed by a 2,3-sigmatropic rearrangement. 24 The transformation suggests that either the Sommelet-Hauser rearrangement is faster than the trapping of the benzylic ylide with EtBpin, or that trapping is indeed efficient but that the subsequent 1,2-migration of the boronate is slow, thus allowing fragmentation back to the ylide. Operation of the latter scenario would suggest that the relief of ring-strain in the 1,2-migration of the azetidinium boronates is an important contributor to the success of the transformation.
Presumably, lithium-stabilised ylide 4b-Li, should it be an intermediate, would undergo solvent-mediated dissociation into the ylide 4b (deprotonation might lead to 4b directly), which if pyramidalised, undergoes rapid inversion, a process that could occur via the ring-open carbene form of 4b (Scheme 8). 14c, 26 We surmised that in a less-coordinating solvent, such as TBME, 4b-Li might be more stable. However, when the lithiation-borylation-oxidation reac-

Paper Syn thesis
tion was performed in TBME [the nonaflate salt of (R)-3b was used owing to the poor solubility of the corresponding triflate salt in TBME], the tertiary alcohol was again isolated as the racemate. The insolubility of the azetidinium substrate frustrated attempts to use even less-coordinating solvents (e.g., hexanes). The use of LDA in place of LiHMDS or the use of the less-sterically hindered neopentylglycol boronic ester, EtBNeo, did not lead to enantiomerically enriched product. The intermediacy of a carbene was supported by the isolation of cyclopropane 22 when using vinyl boronic ester 5k (Scheme 8). Presumably, the desired boronate does not form or undergo 1,2-migration owing to steric hindrance and instead the carbene reacts with the alkene moiety. 27 In conclusion, when 2-phenyl-azetidinium ions are converted into azetidinium ylides, through deprotonation with LDA in the presence of boronic esters, they undergo ringopening carboboration to give γ-dimethylamino tertiary boronic esters. The transformation presumably involves the complexation of the boronic ester with the carbanion of the ylide to form a boronate, which then undergoes ring-opening 1,2-migration, a process that is promoted by the relief of ring-strain of the azetidinium ion. This strain also contributes to the configurational instability of the in situ formed ylides, which appear to be in equilibrium with the ringopened carbene form. The C-B bond of the products can be transformed into a range of functional groups to give a selection of highly functionalised 3-aryl-aminopropanes, which are attractive targets for the pharmaceutical industry.
Reaction mixtures were stirred magnetically. Air-and moisture-sensitive reactions were carried out in flame-dried glassware under a nitrogen atmosphere using standard Schlenk manifold techniques. Fine chemicals were purchased from Acros Organics, Alfa Aesar, Inochem-Frontier Scientific, Sigma-Aldrich, TCI Europe or Santa Cruz Biotechnology and used as received unless otherwise stated. The following pinacol boronic esters were purchased from commercial suppliers: 5e (Frontier Scientific), 5f (Sigma-Aldrich), 5h (Sigma-Aldrich). n-BuLi was received from Acros Organics as a 1.6 M solution in hexane. Lithium diisopropylamide (LDA) was freshly prepared from n-BuLi and distilled diisopropylamine immediately before use. Et 3 N and diisopropylamine were distilled over CaH 2 before use. Anhydrous MeCN, CH 2 Cl 2 , Et 2 O, THF and toluene were obtained from a purification column composed of activated alumina and stored subsequently over 3 Å molecular sieves. Analytical TLC was carried out on aluminiumbacked silica plates (Merck, Silica Gel 60 F254, 0.25). Flash column chromatography was carried out on silica gel (Aldrich, Silica Gel 60, 40-63 μm). Microwave reactions were performed using a Biotage Initiator EXP EU microwave synthesiser. Infrared (IR) spectra were recorded on neat compounds using a PerkinElmer Spectrum One FT-IR spectrophotometer, irradiating between 4000 cm -1 and 600 cm -1 . Only strong and selected absorbance values (ν max ) are reported. 1 H NMR spectra were acquired using a Joel ECS 300, Joel ECS 400 or Varian 400-MR Fourier transform spectrometer for samples in CDCl 3 or CD 3 OD at 301 or 400 MHz as indicated. Chemical shifts (δ H ) are expressed in parts per million (ppm) and are referred to the residual protio solvent signals of CHCl 3 (7.26 ppm) or MeOH (3.31 ppm). 1 H NMR coupling constants are expressed in hertz (Hz) and are quoted as apparent multiplicities (s = singlet, br s = broad singlet, d = doublet, t = triplet, q = quartet, quin = quintet, sept = septet, m = multiplet, dd = doublet of doublets, ps = pseudo). 13 C NMR spectra were recorded at 101 MHz; chemical shifts (δ C ) and are expressed in ppm. Carbon atoms attached to boron or to bromine are usually not observed due to quadrupolar relaxation. See the Supporting Information for proton and carbon assignments and molecule numbering. 11 B NMR spectra were measured using Norell S-200-QTZ quartz tubes at 128 MHz with complete proton decoupling. Some γ-dimethylamino tertiary boronic esters show two or even three signals, the upfield signals indicative of amine-assisted complexation of CD 3 OD (the solvent) and/or H 2 O. 19

Paper Syn thesis
observed using a Bellingham + Stanley Ltd. ADP220 polarimeter at 589 nm (Na D-line) in a cell with path length of 1 dm. GC-MS experiments were carried out using an Agilent 6890 apparatus (column: Supelco SLBTM-5ms capillary column 15 m × 0.25 mm × 0.25 μm).

Synthesis of Pinacol Boronic Esters from Boronic Acids; General Procedure (GP1)
A mixture of boronic acid (1.0 equiv), pinacol (1.0 equiv) and anhydrous MgSO 4 (4.0 equiv) in Et 2 O (0.5 M) was stirred at r.t. for 16 h. The reaction mixture was filtered and the solvent removed in vacuo. The crude material was purified by distillation or flash column chromatography to give the pure boronic ester.

Lithiation-Borylation of 1,1-Dimethyl-2-phenylazetidin-1-ium Trifluoromethanesulfonate (3b) To Give the Tertiary 3-Dimethylamino-Boronic Ester; General Procedure (GP2)
To a solution of diisopropylamine (2.0 equiv) in anhydrous THF (2.0 M) was added n-BuLi (2.0 equiv) at -78 °C. After stirring for 30 min, the solution was added dropwise to a mixture of azetidinium salt 3b (1.0 equiv) and the boronic ester (1.2 equiv) in dry THF (0.03 M) at -78 °C. The reaction mixture was stirred at -78 °C for 1 h and then allowed to warm to r.t. The solvent was removed in vacuo and the crude residue was taken up with H 2 O and extracted with CH 2 Cl 2 (3 times). The combined organic layers were dried over MgSO 4 and concentrated under reduced pressure to afford the crude tertiary boronic ester, which was purified by chromatography on silica gel (EtOAc/ Et 3 N = 100:0.5) to afford the γ-dimethylamino tertiary boronic ester.

Lithiation-Borylation of 1,1-Dimethyl-2-phenylazetidin-1-ium Trifluoromethanesulfonate (3b) with in situ Oxidation; General Procedure (GP3a)
To a solution of diisopropylamine (2.0 equiv) in anhydrous THF (2.0 M) was added n-BuLi (2.0 equiv) at -78 °C. After stirring for 30 min, the solution was added dropwise to a mixture of azetidinium salt 3b (1.0 equiv) and the boronic ester (1.2 equiv) in dry THF (0.03 M) at -78 °C. The reaction mixture was stirred at -78 °C for 1 h and then allowed to warm to r.t. The reaction mixture was cooled to 0 °C and a 2:1 mixture of aq NaOH (2.0 M) and 30% H 2 O 2 was added under vigorous stirring. The cooling bath was removed and the reaction mixture was stirred at r.t. for 1 h. The solvent was removed in vacuo and the residue was partitioned between H 2 O and CH 2 Cl 2 . The phases were separated and the aq layer was re-extracted with CH 2 Cl 2 (2 times). The combined organic layers were washed with brine, dried over Mg-SO 4 , filtered and concentrated under reduced pressure. The residue was purified by flash chromatography on silica gel (EtOAc/Et 3 N = 100:0.5) to afford the pure tertiary alcohol.

Lithiation-Borylation of 1,1-Dimethyl-2-phenylazetidin-1-ium Trifluoromethanesulfonate (3b) with in situ Oxidation at Low Temperature; General Procedure (GP3b)
To a solution of diisopropylamine (2.0 equiv) in anhydrous THF (2.0 M) was added n-BuLi (2.0 equiv) at -78 °C. After stirring for 30 min, the solution was added dropwise to a mixture of azetidinium salt 3b (1.0 equiv) and the boronic ester (1.2 equiv) in dry THF (0.03 M) at -78 °C. The reaction mixture was stirred at -78 °C for 1 h and then allowed to warm to r.t. The reaction mixture was cooled to -40 °C and a 2:1 mixture of aq NaOH (2.0 M) and 30% H 2 O 2 was added under vigorous stirring. The cooling bath was removed and the reaction mixture was stirred at r.t. for 1 h. The solvent was removed in vacuo and the residue was partitioned between H 2 O and CH 2 Cl 2 . The phases were separated and the aq layer was re-extracted with CH 2 Cl 2 (2 times). The combined organic layers were washed with brine, dried over MgSO 4 , filtered and concentrated under reduced pressure. The residue was purified by flash chromatography on silica gel (EtOAc/ Et 3 N = 100:0.5) to afford the pure tertiary alcohol.
Lithiation-Borylation of 1,1-Dimethyl-2-phenylazetidin-1-ium Trifluoromethanesulfonate (3b) with in situ Protodeboronation; General Procedure (GP4) To a solution of diisopropylamine (2.0 equiv) in anhydrous THF (2.0 M) was added n-BuLi (2.0 equiv) at -78 °C. After stirring for 30 min, the solution was added dropwise to a mixture of azetidinium salt 3b (1.0 equiv) and the boronic ester (1.2 equiv) in dry THF (0.03 M) at -78 °C. The reaction mixture was stirred at -78 °C for 1 h and then allowed to warm to r.t. CsF (1.5 equiv) was added at r.t., followed by H 2 O (1.1 equiv) and the reaction mixture was stirred at r.t. for 1 h. The solvent was removed in vacuo and the residue was partitioned between H 2 O and CH 2 Cl 2 . The phases were separated and the aq layer was re-extracted with CH 2 Cl 2 (2 times). The combined organic layers were washed with brine, dried over MgSO 4 , filtered and concentrated under reduced pressure. The residue was purified by flash chromatography on silica gel (EtOAc/Et 3 N = 100:0.5) to afford the pure protodeboronated product.

3-Chloro-1-phenylpropan-1-ol (23)
Following a literature reported procedure, 16 NaBH 4 (5.7 g, 150 mmol) was added in small portions at 0 °C to a stirred solution of 3-chloro-1-propiophenone (13) (8.4 g, 50 mmol) in MeOH (104 mL). The mixture was stirred at r.t. for 18 h and was then quenched with H 2 O (90 mL). The solvent was removed in vacuo and the residue was extracted with Et 2 O (3 × 50 mL). The combined organic layers were dried over MgSO 4 and concentrated in vacuo to give the title compound as a colourless oil (8.4 g, 98% yield). The product was used in the next step without further purification. All analytical data matched that previously reported.

1-Methyl-2-phenylazetidine (14)
Following a procedure by Luisi, 14b to a solution of 3-chloro-1-phenylpropan-1-ol (23) (3.1 g, 18.2 mmol) in dry CH 2 Cl 2 (18 mL), a solution of SOCl 2 (4.0 mL, 54.6 mmol) in dry CH 2 Cl 2 (5.5 mL) was added dropwise at r.t. After stirring for 1 h, the reaction mixture was poured into H 2 O (20 mL) and aq NaOH (15% w/v) was added slowly to neutralise the excess of HCl until the pH of the solution was 7. The aq phase was extracted with CH 2 Cl 2 (3 × 30 mL) and the combined organic layers were dried over MgSO 4 , filtered and evaporated under vacuum to afford 1-phenyl-1,3-dichloropropane that was employed in the next step without further purification.
To a solution of 1-phenyl-1,3-dichloropropane in EtOH (23 mL) and Et 3 N (5.1 mL, 36.4 mmol) in a sealed flask, a solution of MeNH 2 (33% w/v in EtOH, 23 mL) was added at r.t. The reaction mixture was heated at 70 °C for 16 h and then allowed to cool to ambient temperature. The solvent was removed in vacuo and HCl (30 mL, 2.0 M) was added. The aq phase was extracted with CH 2 Cl 2 (3 × 40 mL) and subsequently basified by addition of aq NaOH (15% w/v) until the pH of the solution was >12. The basic aq phase was extracted with CH 2 Cl 2 (3 × 50 mL), and the combined organic layers were dried over MgSO 4 , filtered and concentrated in vacuo. The crude mixture was purified by flash silica gel column chromatography (EtOAc/Et 3 N = 100:0.5) to afford azetidine 14 (1.34 g, 50%, over two steps) as a colourless oil. The analytical data matched that previously reported.