Catalytic Asymmetric Hydrogenation of 3-Substituted Benzisoxazoles

A variety of 3-substituted benzisoxazoles were reduced with hydrogen using the chiral ruthenium catalyst, {RuCl(p-cymene)[(R,R)-(S,S)-PhTRAP]}Cl. The ruthenium-catalyzed hydrogenation proceeded in high yield in the presence of an acylating agent, affording α-substituted o-hydroxybenzylamines with up to 57% ee. In the catalytic transformation, the N–O bond of the benzisoxazole substrate is reductively cleaved by the ruthenium complex under the hydrogenation conditions. The C–N double bond of the resulting imine is saturated stereoselectively through the PhTRAP–ruthenium catalysis. The hydrogenation produces chiral primary amines, which may work as catalytic poisons, however, the amino group of the hydrogenation product is rapidly acylated when the reaction is conducted in the presence of an appropriate acylating agent, such as Boc2O or Cbz-OSu.

The benzisoxazoline products can be transformed into optically active -substituted o-hydroxybenzylamines, because N-O bonds are known to break through heterogeneous catalysis under hydrogenation conditions [32,33]. Enantiomeric benzylamines are often used as chiral auxiliaries [34,35] and constituents of catalysts for asymmetric synthesis [36][37][38]. Furthermore, the structural motives are seen in many isoquinoline alkaloids [39][40][41][42]. Chiral amines are typically prepared through enzymatic [43] or chemical resolution of their racemates [44]. The diastereoselective nucleophilic additions to imines have been applied to the asymmetric synthesis of the chiral amines [45,46]. However, to our knowledge there have been only a few reports on the enantioselective synthesis of -substituted o-hydroxybenzylamines [47,48]. In this paper, we report a catalytic asymmetric hydrogenation of 3-substituted benzisoxazoles to yield chiral -substituted o-hydroxybenzylamines. The asymmetric reaction proceeded through the PhTRAP-ruthenium catalysis, which transformed the benzo-fused heteroaromatics into -substituted o-hydroxybenzylamines in high yields and up to 57% ee.

Optimization of Reaction Conditions
In our initial attempts, 3-ethylbenzisoxazole (1a) was treated with pressurized hydrogen gas (50 atm) in toluene or isobutyl alcohol at 80 °C for 4 h in the presence of {RuCl(p-cymene)[(R,R)-(S,S)-PhTRAP]}Cl (2) ( Table 1, entries 1 and 2) [25]. A small amount of 1a reacted with hydrogen, but no saturation of the C-N double bond was observed in either reaction, which afforded only the achiral imine 3a, formed through the hydrogenolytic cleavage of the N-O bond of 1a [49,50]. The ruthenium catalyst failed to reduce the C-N double bond even in the presence of N,N,N',N'tetramethylguanidine (TMG) (entries 3 and 4), even though in our previous reports the base additive brought about a remarkable acceleration of hydrogenations of heteroaromatics [22][23][24][25]28,30,31]. We were pleased that the hydrogenation of the benzisoxazole was accompanied by the reduction of the C-N double bond in the presence of stoichiometric Boc 2 O, which afforded N-Boc-protected (R)-1-(2hydroxyphenyl)-1-propylamine 4a with 25% ee (entry 5). To our surprise, no formation of 4a was observed in the reaction using both Boc 2 O and TMG (entry 6). The base additive might inhibit the hydrogenation of imine 3a. Various aprotic solvents were evaluated for the asymmetric hydrogenation (entries 7-10). As a result, 4a was obtained with the highest ee value when the hydrogenation was conducted in an ethereal solvent, such as THF or cyclopentyl methyl ether (CPME). The benzisoxazole was fully converted to the N-protected amine 4a with 44% ee after 24 h (entry 11). The hydrogenation of 1a was conducted by using other amino group protecting agents in place of Boc 2 O. Carboxylic anhydrides, Ac 2 O and Bz 2 O, also worked as the acylating agent in the asymmetric reduction of the benzisoxazole to give the corresponding chiral amine with 43% and 47% ee, respectively (entries 12 and 13). Use of sulfonic anhydride resulted in a complex mixture, which contained a small amount of 3a (entry 14). Various N-acylating agents other than acid anhydrides were applied to the asymmetric reactions. No conversion of 1a took place when the reaction was conducted with Boc-ON (entry 15). Use of O-alkoxycarbonyl-N-hydroxysuccinimide led to a remarkable increase in the reaction rate and some improvement of the stereoselectivity (entries [16][17][18]. The reactions with Cbz-OSu and Fmoc-OSu in THF gave the desired products 8a and 9a with 52% and 56% ee, respectively. The hydrogenation product 4a was formed under 10 atm of hydrogen, but the lower hydrogen pressure caused a significant decrease in the reaction rate (entry 19). Substrate 1a and imine 3a completely disappeared from the reaction mixture at 24 h.  Various chiral ligands other than PhTRAP were evaluated for the hydrogenation of 1a (Table 2). Before the evaluation of ligands, we attempted the catalytic asymmetric reactions by means of a few in-situ-generated PhTRAP-ruthenium catalysts. In our previous study on the asymmetric hydrogenation of oxazoles, the chiral catalyst was generated in situ from Ru( 3 -methallyl) 2 (cod) and the chiral bisphosphine [30]. However, the chiral ruthenium complex failed to catalyze the conversion of 1a to imine 3a (entry 1). No reduction of the C-N double bond was observed in the reaction mixture. The hydrogenation of 1a was conducted in the presence of the crude ruthenium complex, which was prepared by mixing [RuCl 2 (p-cymene)] 2 and (R,R)-(S,S)-PhTRAP in CH 2 Cl 2 -EtOH (1:2) at 50 °C for 1 h and then removing the solvent in vacuo. The resulting residue worked as the chiral catalyst as with the isolated complex 2, akthough its catalyst efficiency was lower than that of 2 (entry 2). By using the procedure for preparing the chiral catalyst from [RuCl 2 (p-cymene)] 2 , the hydrogenation of 1a was conducted with a series of chiral bisphosphines (entries 3-8). As a result of the ligand evaluation, PhTRAP was found to be the most effective ligand for the asymmetric hydrogenation of benzisoxazoles. The ruthenium catalysts prepared from BPPFA [51,52], Josiphos [53], and Chiraphos [54] could cleave the N-O bond in 1a to give imine 3a under the hydrogenation conditions, while they are ineffective for the conversion of 3a into 8a. Furthermore, the enantiomeric excesses of the chiral products were lower than 20% ee. In contrast, [RuCl 2 (p-cymene)] 2 -BINAP [55,56] might be competent to reduce the C-N double bond of 3a, but the catalyst failed to cleave the N-O bond of 1a with sufficient reaction rate. Furthermore, no conversion of 1a was observed in the reactions using DIOP [57] and Me-DuPhos ligands [58].

Asymmetric Hydrogenations of 3-Substituted Benzisoxazoles
As shown in Table 3, the PhTRAP-ruthenium catalyst 2 converted various 3-substituted benzisoxazoles 1 to the corresponding N-Cbz-protected o-hydroxybenzylamines 8 under the optimized conditions. Reaction rates of the catalytic transformations were affected by the bulkiness of the substituent at the 3-position in 1. As with 1a, 3-methylbenzisoxazole 1b was converted to 8b with moderate enantiomeric excess (entry 1). Chiral benzylamines 8c and 8d were obtained in high yields from the benzisoxazoles bearing 2-phenethyl and isopropyl groups at their 3-position, but complete conversions of 1c and 1d needed long reaction times as compared to 1a or 1b (entries 2 and 3). Furthermore, the reactions of 1c and 1d were less enantioselective. The asymmetric hydrogenation of aryl-substituted substrate 1e produced the desired chiral diarylamine 8e with 55% ee, but a small amount of achiral diarylmethane was formed through the undesired hydrogenolysis of the benzylic C-N bond (entry 4). The stereoselectivity of the asymmetric reaction may correlate with the electronic property of the substituent at the 5-position of benzisoxazole. Electron-donating groups in 1f or 1g slightly improved the enantioselectivity (entries 5 and 6). In contrast, the fluorine atom in 1h caused a decrease in the ee value of the hydrogenation product (entry 7). The substituent at the 6-position unsystematically affects the stereoselectivity. The reaction of 3,6-dimethylbenzisoxazole (1j) proceeded with a comparable stereoselectivity to 1b (entry 9). However, the selectivity deteriorated in the reaction of the substrate bearing either electron-donating or electron-withdrawing groups at the 6-position (entries 8 and 10). Steric hindrance of the methoxy group in 1l scarcely affected the yield of the hydrogenation product, but it did cause significant decrease in enantioselectivity (entry 11).

Reaction Pathway of the Asymmetric Hydrogenation of Benzisoxazoles
As described above, the hydrogenation of benzisoxazoles 1 proceeds through the reductive N-O bond cleavage to form o-hydroxyphenyl imine 3, yielding -substituted N-Cbz-benzylamine 8 in the presence of Cbz-OSu. The acylating agent is indispensable for the reduction of the imine to amine. Furthermore, the absence of the ruthenium complex prevents the conversion of 1 to 3 as well as the formation of 8. In light of these observations, two reaction pathways are conceivable for the catalytic asymmetric hydrogenation of benzisoxazoles, as shown in Scheme 1. In both pathways, the ruthenium catalyst 2 initially cleaves the N-O bond in 1, giving imine 3 [49,50]. One involves the acylation of the imine NH with Cbz-OSu to form intermediate 10 [59,60], which is reduced to 8 by hydrogen through the ruthenium catalysis (path a). In the other possible pathway, the hydrogenation of C-N double bond in 3 occurs prior to the Cbz-protection of the nitrogen atom (path b) [61]. To ascertain these two possibilities, the following reactions were conducted by using the imine 3a as the substrate. First, the imine 3a was treated with Cbz-OSu at 80 °C for 4 h in the absence of catalyst 2 (Eq. 1). No formation of N-Cbz-imine 10a was observed in the resulting mixture, and the substrate 3a remained intact, hence path a can be ruled out. Next, the hydrogenation of 3a was conducted with the ruthenium catalyst in the absence of Cbz-OSu, but the reaction produced only a small amount of the expected primary amine 11 (Eq. 2). The reduction of the imine 3a, however, proceeded in the presence of Cbz-OSu, affording 8a in high yield (Eq. 3). Consequently, the present hydrogenation of the benzisoxazoles should occur through path b. Although the resulting primary amine 11 may strongly inhibit the catalysis of the PhTRAP-ruthenium complex, the generated amino group is rapidly protected by the coexistent Cbz-OSu under the hydrogenation conditions [62,63]. Thus, the rapid acylation effectively avoids the inhibition of the ruthenium catalyst by the free amino group of 11 [64,65].

Scheme 1.
Possible pathways for the asymmetric hydrogenation of benzisoxazoles.

Materials
Optical rotations and NMR spectra were measured with JASCO P-1020 polarimeter and Bruker AVANCE 400 (9.4 T magnet) spectrometer, respectively. In the 1 H-NMR spectra, chemical shifts (ppm) were referenced to internal tetramethylsilane (0.00 ppm in CDCl 3 ). In 13 C-NMR spectra, chemical shifts (ppm) were referenced to the carbon signal of the deuterated solvents used (77.0 ppm in CDCl 3 , 30.0 ppm in acetone-d 6 ). Elemental analyses were performed by Service Centre of Elementary Analysis of Organic Compounds. Flash column chromatographies were performed with silica gel 60 (230-400 mesh, Merck). Benzisoxazole substrates were synthesized from the corresponding o-hydroxyphenones through the oxime formation, followed by the O-acetylation of the oxime and the intramolecular nucleophilic substitution on the nitrogen atom [66,67].
{RuCl(p-cymene)[(R,R)-(S,S)-PhTRAP]}Cl was prepared according to the literature procedure [25]. Tetrahydrofuran (THF) was deoxygenated by purging with nitrogen for 30 min and was dried with an alumina and copper column system (GlassContour Co.). Other materials were purchased and used without further purification.

Conclusions
In this study, we proved that 3-substituted benzisoxazoles 1 react with hydrogen in the presence of the chiral ruthenium catalyst, {RuCl(p-cymene)[(R,R)-(S,S)-PhTRAP]}Cl (2). The rutheniumcatalyzed hydrogenation proceeds in the presence of an acylating agent such as Boc 2 O and Cbz-OSu, to afford -substituted N-protected o-hydroxybenzylamines 4-9 in high yields with moderate enantioselectivities. The conversion of 1 to the chiral amines proceeds through the imine intermediate 3, which is generated from the reductive cleavage of the N-O bond in the benzisoxazole ring. The C-N double bond of 3 is hydrogenated in moderate enantioselectivity by the PhTRAP-ruthenium catalyst, however, the resulting primary amine 11 causes deactivation of the catalyst 2. The deterioration of 2 can be avoided by the coexistent acylating agent, which rapidly reacts with the amino group.
Although the asymmetric hydrogenation of 1 with PhTRAP-ruthenium catalyst proceeds with moderate stereoselectivity at the current moment, various 3-substituted benzisoxazoles are transformed into the chiral o-hydroxybenzylamines in high yields. The products contain a molecular framework, which is often used as a chiral source in asymmetric synthesis and seen in various biologically active compounds. Consequently, the present catalytic asymmetric reaction may be potentially useful for organic synthesis. Further improvement of the stereoselectivity is in progress.