Chiral N-salicylidene vanadyl carboxylate-catalyzed enantioselective aerobic oxidation of α-hydroxy esters and amides

Abstract

A series of chiral vanadyl carboxylates derived from N-salicylidene-l-α-amino acids and vanadyl sulfate has been developed. These configurationally well defined complexes were examined for the kinetic resolution of double- and mono-activated 2° alcohols. The best chiral templates involve the combination of l-tert-leucine and 3,5-di-t-butyl-, 3,5-diphenyl-, or 3,4-dibromo-salicylaldehyde. The resulting vanadyl(V)-methoxide complexes after recrystallization from air-saturated methanol serve as highly enantioselective catalysts for asymmetric aerobic oxidation of α-hydroxyl-esters and amides with a diverse array of α-, O-, and N-substituents at ambient temperature in toluene. The asymmetric inductions of the oxidation process are in the range of 10 to >100 in terms of selectivity factors (k_rel) in most instances. The previously undescribed aerobic oxidation protocol is also applicable to the kinetic resolution of C-13 taxol side chain with high selectivity factor (k_rel = 35). X-ray crystallographic analysis of an adduct between a given vanadyl complex and N-benzyl-mandelamide allows for probing the stereochemical origin of the nearly exclusive asymmetric control in the oxidation process.

The oxidation of alcohols normally requires stoichiometric use of DMSO-based reagents or metal oxides of high oxidation state (1). Advances on their aerobic oxidations with catalytic metal oxides [e.g., RuO₂–H₂O (2), V₂O₅–K₂CO₃ (3), and OsO₄–Cu(O₂CR)₂ (4)], homogeneous metal complexes [e.g., Co(OAc)₂-N-hydroxyphthalimide (5), Ru(III)NOCl(salen) (6), RuCl₂(PPh₃)₃-hydroquinone-K₂CO₃ (7), CuCl-Phen-DBADH₂ (8), RuCl₂(PPh₃)₂-TEMPO (9)/CuCl-TEMPO (10), Pd(OAc)₂-pyridine (11, 12), and Pd₄Phen₂(CO)(OAc)₄ (13)], bimetallic complexes [e.g., RuCl₃-Co(OAc)₂-aldehyde (14) and MoO₂(acac)₂-CuNO₃] (14, 15), or heterogeneous metal complexes [e.g., Ru(III) hydroxyapatite (16), polyaniline-supported MoO₂(acac)₂ (17), and Pd(OAc)₂-hydrotalcite-pyridine (18)] have been documented. Notably, additives and/or bases are often needed to increase the catalyst reactivity and/or to facilitate the turnover process. In addition, the targeted substrates are somewhat limited to primary, benzylic, allylic, and propargylic alcohols. Recently, the asymmetric variants of the aerobic catalytic process have attracted a lot of attention (19, 20). So far, Pd(II)-sparteine (21, 22), photoly activated RuNOCl(salen) (23), and Mn(III)(salen)PF₆ (24) have been developed with fair to high enantioselectivity toward kinetic resolutions of 2° benzylic alcohols. In marked contrast, the asymmetric aerobic catalytic process with α-hydroxycarboxylic acid derivatives is relatively unexplored (25–31).

Optically pure α-hydroxycarboxylic acid and mandelic acid derivatives are important precursors toward enantioselective synthesis (ref. 32; for a leading application, see also ref. 33), drug development (34–38), and biologically active (e.g., antibacterial) compounds (39–42). Tremendous endeavors have been devoted to enzymatic kinetic resolutions of the racemic substrates by hydrolysis (43), acylation [including a dynamic process by (RuCp(CO)₂)₂H (44, 45)], and aerobic oxidation (46), albeit with limited substrate scopes. Based on our experiences of using vanadyl and oxometallic species in catalyzing C–C and C–X bond forming (47–51), aerobic oxidative coupling (52, 53), and photoinitiated DNA cleavage (54) events, we thought to examine the feasibility of direct asymmetric, aerobic oxidation of α-hydroxy esters and amides at ambient temperature by chiral vanadyl complexes (55, 56).

Results and Discussions

Effects of Templates in Vanadyl Complexes.

Methyl and benzyl mandelates were first used as test substrates (for the biological activity of mandelic acid, see ref. 57; see also refs. 58 and 59) for a diverse array of N-naphthalidene (1–6), N-salicylidene (7–14), and N-ketopinidene-based (15) vanadyl complexes bearing the optimal l-tert-leucine chiral template. So far, arenes and CCl₄ are the best solvents of choice. Therefore, the test aerobic oxidations were carried out in toluene under oxygen atmosphere at ambient temperature in the presence of 5 mol % of the individual catalysts 1–15 (Fig. 1).

Fig. 1.

A list of chiral vanadyl(V) methoxides examined.

In general, the asymmetric oxidations of benzyl mandelate are more selective than those of methyl mandelate (Table 1). It also was found that the enantiocontrol for the kinetic resolution of methyl and benzyl mandelates (16a and 16d) highly depends on the sterics of the C-3 substituents in the catalyst templates. Within the same N-naphthalidene family, the selectivity factors follow the order of 4 (G = Ph₂CH; k_rel = 26/94) > 5 (G = Ph₂COH; k_rel = 12/28) ≥ 1 (G = 3,4-benzo; k_rel = 12/14) > 6 (G = OCH₂Ph; k_rel = 5/10) > 3 (G = 3,4-benzo-5,6-benzo; k_rel = 4/6) > 2 (G = H; k_rel = 1/1). An essentially similar trend (except in the oxidation of benzyl mandelate by 13) was observed in the N-salicylidene family, where the selectivity factors follow the order of 8 (R¹ = t-Bu; R² = OMe; k_rel = 50/76), 10(R¹ = R² = t-Bu; k_rel = 29/>100), 9 (R¹ = t-Bu, R² = NO₂; k_rel = 26/>100), and 7 (R¹ = t-Bu; R² = H; k_rel = 26/76) > 11 (R¹ = adamantyl; R² = CH₃; k_rel = 20/32) > 13 (R¹ = R² = Ph; k_rel = 36/8) > 12 (R¹ = R² = Br; k_rel = 10/15).

Table 1.

Effects of catalysts on the asymmetric aerobic oxidation of racemic methyl and benzyl mandelates 16a and 16d

Catalyst	Time, h	Conversion,* %	%ee,† (yield,‡ %)	krel§
1	43/36	55/53	83 (41)/80 (44)	12/14
2	63/58	48/52	2 (44)/7 (43)	1/1
3	87/130	51/53	48 (43)/61 (43)	4/6
4	9/13	52/50	87 (44)/93 (43)	26/94
5	13.5/15	52/50	75 (42)/83 (43)	12/28
6	75/90	52/52	56 (43)/72 (45)	5/10
7	22/14.5	52/52	87 (46)/97 (46)	26/76
8	88/124	54/52	98 (45)/97 (45)	50/76
9	22/14.5	52/50.5	87 (47)/97 (50)	26/167
10	23/19	54/50	93 (40)/98 (46)	29/458
11	22/26	51/53	81 (44)/92 (45)	20/32
12	14/12	55/56	78 (43)/87 (40)	10/15
13	16/14	55/60	97 (44)/82 (37)	36/8
14	133/128	71/66	45 (26)/47 (29)	2/2
15	86/240	49/51	6 (44)/7 (46)	1/1

*Determined by ¹H NMR analysis of the reaction mixture.

^†Determined by HPLC analysis on Chiralpak AD-H or AS column.

^‡Isolated, purified material for the alcohol by column chromatography.

^§k_rel = ln[(1 − C)(1 − ee)]/ln[(1 − C)(1 + ee)], where C = conversion and ee = enantiomeric excess.

Notably, within the 3-tert-butyl N-salicylidene family, catalysts 7–10 are the most selective ones, and the electronic effect of the C5 substituent (R²) plays a role on the reaction rate. Vanadyl complex-8 bearing electron-donating C5-group (R² = OMe) exhibits the slowest rates of oxidation (88–124 h) but with the best enantioselectivities (k_rel = 50/76). On the contrary, vanadyl complex-9 bearing electron-withdrawing C5-group (R² = NO₂) exerts the fastest rates of oxidation (14.5–22 h) but with slightly poorer enantioselectivities (k_rel = 26/>100). Finally, very poor enantioselectivities (k_rel = 1–2) were observed with 14 and 15 as the catalysts.

N-benzyl-mandelamide was next examined with the more selective vanadyl complexes 1 and 3–13 under the standard reaction conditions (Table 2). To our surprise, the enantiocontrols (k_rel = 2–6) in the oxidation drop significantly by using the N-naphthalidene family 1 and 3–6 as the catalysts. Conversely, the selectivity factors (k_rel = 44 to >100) in the 3-tert-butyl N-salicylidene family 7–10 are generally higher than those from the aerobic oxidations of methyl and benzyl mandelates (k_rel = 26–50; 76- >100). Similarly, the resultant enantioselectivity trend exerted by the electronic effect at C-5 group also was observed in the 3-tert-butyl N-salicylidene family 7–10 (compare the results from catalysts 8 and 9). Notably, significant increase of asymmetric inductions was attained by the 3,5-dibromo- and 3,5-diphenyl-N-salicylidene-based vanadyl complexes 12 and 13. Both the selectivity factors are >100. Based on the selectivity and reactivity profiles in these three representative substrate studies with varying templates in the chiral vanadyl complexes, 3-tert-butyl, 3,5-di-tert-butyl, 3,5-dibromo, and 3,5-diphenyl-N-salicylidene-based vanadyl complexes (7, 10, 12, and 13) were selected for subsequent substrate survey regarding the effects of the pendant O- and N-alkyl (phenyl) groups and α-substituents in the α-hydroxy esters and amides on the asymmetric oxidation. Ultimately, vanadyl complex 10 was preferred for the oxidation of α-aryl-α-hydroxy esters and amides in terms of selectivity concerns. Conversely, vanadyl complex 12 was selected for the oxidation of α-alkyl-α-hydroxy esters and amides in terms of faster reaction rate concerns.

Table 2.

Effects of catalyst templates on the asymmetric aerobic oxidation of racemic N-benzyl mandelamide-17c

Catalyst	Time, h	Conversion,* %	% ee,† (yield,‡ %)	krel§
1	168	52	58 (44)	6
3	254	51	28 (43)	2
4	264	46	21 (50)	2
5	240	47	41 (52)	4
6	192	52	56 (44)	5
7	20	50	98 (48)	458
8	60	50	98 (50)	458
9	13	52	93 (43)	44
10	25	50	99 (47)	1,057
11	240	48	62 (48)	9
12	9	50	99 (46)	1,057
13	8	51	97 (47)	119

Footnotes are the same as those in Table 1.

Effects of the Pendant O- and N-Substituents.

Vanadyl complex 10 was chosen for the asymmetric oxidations of mandelates (16a–f) and mandelamides (17a–f) with varying O-alkyl, O-phenyl, N-alkyl, and N-phenyl appendages. The best enantioselectivities were achieved by using substrates bearing O-/N-benzyl (k_rel > 100) and -diphenylmethyl (k_rel = 60 to >100) groups (Table 3). Conversely, there is no consistent selectivity trend by increasing the steric bulk of the pendant alkyl groups from methyl, ethyl, isopropyl to tert-butyl. The worst scenarios go to substrates possessing N-tert-butyl (k_rel = 1), O-ethyl (k_rel = 12), and O-/N-phenyl (k_rel = 10–13) groups. Reasonably high selectivity factors (k_rel = 29 and 66) were observed in the case of methyl and isopropyl mandelates-16a and c. However, very poor selectivity was resulted from N-isopropylmandelamide-17a (k_rel = 1). The results indicate that a delicate balance among the steric, conformational (s-cis vs. s-trans), and electronic (π–π interaction) factors in a given substrate is essential during asymmetric discrimination event in the incipient adduct formed between the catalyst and the substrate (see below).

Table 3.

Effects of O-/N-substituents on the asymmetric aerobic oxidation of racemic mandelates 16a–f and mandelamides 17a–e

X–R	Time, h	Conversion,* %	% ee,† (yield,‡ %)	krel§
OCH3 (16a)	23	54	93 (40)	29
OCH2CH3 (16b)	22.5	50	71 (48)	12
OCH(CH3)2 (16c)	19	49	88 (49)	66
OCH2Ph (16d)	19	50	98 (46)	458
OCHPh2 (16e)	35	49	91 (50)	117
OPh (16f)	14	53	79 (45)	13
NHCH(CH3)2 (17a)	104	50	7 (44)	1
NHCH2Ph (17b)	25	50	99 (47)	1057
NHCHPh2 (17c)	52	51	93 (47)	60
NH-t-Bu (17d)	119	49	8 (48)	1
NHPh (17e)	82	51	70 (46)	10

Footnotes are the same as those in Table 1.

Effects of the α-Substituents in the Benzyl 2-Hydroxyesters.

Based on the screening of the pendant O-substituents above, a series of benzyl α-hydroxyesters bearing different α-aryl groups were further examined under the optimal aerobic oxidation protocol catalyzed by vanadyl complex 10 (Table 4). For 4-substituted-phenyl (i.e., mandelate) derivatives, 18–21, the selectivity factors in the kinetic resolution process range from 10 to 37. In general, substrates bearing electron-withdrawing para groups (e.g., 4-Cl and 4-Br) are more reactive (15–24 h) and more enantioselective (k_rel 28–37) than those with electron-donating (e.g., 4-CH₃; 95 h; k_rel = 14) and/or coordinating (e.g., 4-CH₃O; 57 h; k_rel = 10) groups. The electronic influence on enantiocontrol is even more pronounced for 2-substituted-phenyl derivatives 22–24. The selectivity profiles follow the order of 2-Cl (k_rel = 70) > 2-CH₃ (k_rel = 21) > 2-NO₂ (k_rel = 15). For naphthyl-containing benzyl mandelate analogs, 1-naphthyl (i.e., 2,3-/o,m-benzo-fused) system 25 (k_rel = 18) is more selective than 2-naphthyl (i.e., 3,4/-m,p-benzo-fused) one-26 (k_rel = 9), indicating a somewhat similar detrimental effect of the para-substituent on the enantiocontrol. For heteroaryl-containing systems, the enantioselectivity decreases dramatically with increasing coordination ability of the heteroatom. By comparing 16d with 27 and 28, the selectivity factors of the aerobic oxidations follow the order of 16d (k_rel > 100) > 28 (for the biological activities of the acids from the corresponding 28, see ref. 60) (k_rel = 43) > 27 (for the biological activities of the acids from the corresponding 27, see ref. 57) (k_rel = 5).

Table 4.

Effects of α-aryl groups on the asymmetric aerobic oxidation of racemic benzyl mandelate derivatives 18–28 by catalyst-10

Ar	Time, h	Conversion,* %	% ee,† (yield,‡ %)	krel§
C6H5 (16d)	12	50	98 (46)	458
4-CH3C6H4 (18)	95	58	90 (40)	14
4-CH3OC6H4 (19)	57	51	70 (49)	10
4-ClC6H4 (20)	23.5	50	86 (47)	37
4-BrC6H4 (21)	15	52	88 (43)	28
2-CH3C6H4 (22)	365	52	84 (46)	21
2-ClC6H4 (23)	63	51	94 (48)	70
2-NO2C6H4 (24)	50	53	81 (45)	15
1-Np (25)	73	51	80 (48)	18
2-Np (26)	24.5	59	81 (40)	9
(27)	58	51	55 (45)	5
(28)	4	53	95 (47)	43

Footnotes are the same as those in Table 1.

The substrate class was further extended to benzyl α-hydroxyesters possessing α-alkenyl, α-alkynyl, and α-alkyl groups (Table 5). In general, the first two substrate classes as represented in 29 (R = trans-PhCHCH) and 30 (R = Ph–CC) are more reactive than the corresponding α-aryl analogs 16d and 18–27. Their aerobic oxidations catalyzed by 10 can be completed at ≈50% conversion in 5.5–6 h. In addition, the asymmetric induction for 29 (k_rel = 27) is a lot more enantioselective than that for 30 (k_rel = 7). Furthermore, the conjugated alkene moiety in 29 remains intact without any intervening epoxidation. In marked contrast, substrates 31–35 bearing α-alkyl groups are completely inert toward aerobic oxidation under the optimal catalytic conditions in the presence of 10. Nevertheless, the desired process can be effected by using the most reactive 3,5-dibromo-N-salicylidene-based vanadyl complex 12, albeit with prolonged reaction time (63–188 h). Notably, the selectivity factors of their oxidations catalyzed by 12 highly hinge on the steric effects of the α-alkyl groups. In comparison, the substrate 35 (61) bearing 2° α-cyclohexyl group (k_rel = 8) is more enantioselective than that bearing 1° α-methyl group (i.e., lactate-31; k_rel = 2). Furthermore, the enantiocontrol increases with increasing steric hindrance of the β-branching units in 1° α-alkyl-substituted substrates 31–34. Namely, the selectivity factors follow the order of benzyl [33 (for the biological activity of 2-hydroxy-3-methylbutanoic and 3-phenylpropanoic acids derived from 32 and 33, see ref. 62) and 34, k_rel = 97 and 35] > isopropyl [32 (32); k_rel = 14] > methyl [31; k_rel = 2). Notably, the kinetic resolution of N-benzoylated-3-phenylisoserine methyl ester-34 [i.e., the taxol C-13 side chain (63)] at 57% conversion led to the recovery of the desired natural fragment [33–35% yield, 99% enantiomeric excess (ee)] bearing (2R, 3S)-absolute stereochemistry. Unfortunately, the oxidized product was further degraded into benzamide (56% yield) and benzaldehyde under the reaction conditions. To suppress the extensive oxidation of the resultant α-keto ester, other N-substituted-3-phenylisoserine methyl esters would be required.

Table 5.

Effects of α-alkenyl, alkenyl, and alkyl groups on the asymmetric aerobic oxidation of racemic 2-hydroxyesters 29–35 by catalyst-10 and 12 (for R = alkyl only)

R	Time, h	Conversion,* %	% ee,† (yield,‡ %)	krel§
trans-PhCH = CH (29)	5.5	56	96 (40)	27
(30)‖	6	50 (49)**	60 (46)/50 (50)**	7/5**
CH3 (31)	83	60	37 (36)	2
(CH3)2CH (32)	63	49	71 (46)	14
C6H5CH2 (33)	89	51	96 (46)	97
syn-C6H5CHNHBz (34)¶>	112	57	99 (34)	35
c-C6H11 (35)	188	49	60 (49)	8

Footnotes ∗, †, ‡, and § are the same as those in Table 1.

^¶Methyl esters were used.

^‖Ethyl ester was used.

**The data were obtained by using Toste’s best catalytic system.

Effects of the α-Substituents in the N-Benzyl-2-Hydroxyamides.

In view of the screening results of the pendant N-substituents in Table 3, a series of N-benzyl-α-hydroxy-amides bearing the same set of α-aryl groups (as in Table 4) were also examined under the optimal conditions catalyzed by vanadyl complex-10 (Table 6). In all cases except 4-methoxy-, 2-methoxy-, and 2-chloro-phenyl substituted cases (37, 41, and 42), the selectivity factors of their aerobic oxidations are at least two times larger than those of the corresponding benzyl esters (compare Tables 4 and 6). In addition, only the 4-methoxy- and 2-methoxy-phenyl analogs-37 and 41 led to unsatisfactory result (k_rel = 7 and −3). Notably, the opposite (S)-enantiomer-41 was recovered in 34% ee. Presumably, the competing coordination of the 2-methoxyphenyl group to the vanadyl center with the amide carbonyl group during the chelation adduct formation event erodes and reverses the asymmetric differentiation (see below). Nevertheless, significant improvement was achieved particularly in the case of 2-furanyl (k_rel = 24 for 45 vs. 5 for 27).

Table 6.

Effects of α-aryl groups on the asymmetric aerobic oxidation of racemic N-benzyl mandelate derivatives 36–46 by catalyst-10

Ar	Time, h	Conversion,* %	% ee,† (yield,‡ %)	krel§
C6H5(17b)	25	50	99 (47)	1,057
4-CH3C6H4 (36)	130	60	99.6 (38)	24
4-CH3OC6H4 (37)	106	61	77 (38)	7
4-ClC6H4 (38)	70	50	97 (47)	278
4-BrC6H4 (39)	24	53	>99 (45)	>80
2-CH3C6H4 (40)	81	49	94 (45)	211
2-CH3OC6H4 (41)	108	50	34 (44)	−3¶
2-ClC6H4 (42)	18	50	92 (47)	79
1-Np (43)	130	53	99 (45)	>80
2-Np (44)	114	64	>99 (32)	17
(45)	164	52	86 (43)	24
(46)	4	53	99 (42)	80

Footnotes ∗, †, ‡, and § are the same as those in Table 1.

^¶The (S)-enantiomer was obtained.

Among all of the N-benzyl-α-hydroxy-amides examined, the reactivity profile again follows the order of α-alkenyl ≥ α-aryl ≫ α-alkyl (Tables 6 and 7). For the amides 47–51 possessing α-alkenyl and α-alkyl groups, good to excellent kinetic resolutions (k_rel = 33 to >100) remain attainable except in the methyl case (i.e., 48) where poor selectivity factor (k_rel = 3) like that (k_rel = 2) for the analogous ester-31 was observed.

Table 7.

Effects of α-alkenyl and alkyl groups on the asymmetric aerobic oxidation of racemic N-benzyl-2-hydroxyamides 47–51 by catalyst-10 and 12 (for R = alkyl only)

R	Time, h	Conversion,* %	% ee,† (yield,‡ %)	krel§
trans-PhCH = CH(47)	9	51	>99 (47)	>211
CH3 (48)	92	50	33 (47)	3
(CH3)2CH (49)	142	51	95 (46)	81
C6H5CH2 (50)	102	52	95 (43)	56
c-C6H11 (51)	182	52	90 (45)	33

Footnotes are the same as those in Table 1.

During the x-ray structural identification of a catalyst–substrate adduct and the preparation of this work, a very recent elegant study by Toste and coworkers (64) unraveled the use of in situ-generated chiral vanadyl(V) isopropoxides by direct mixing of N-salicylidene-l-α-amino alcohols [i.e., Bolm’s ligands (65, 66)] with VO(OiPr)₃ for asymmetric aerobic oxidations of α-hydroxyesters in acetone with good enantiocontrols (k_rel = 13 to >50) except in the TMS-CC α-substituted case (k_rel = 6). Notably, the corresponding α-hydroxyamides except N-t-butyl-mandelamide were not examined, and the α-hydroxyesters they studied bear somewhat different α-substituents from ours. To probe whether their best system could improve the enantioselectivity of the worst substrate in our study, the aerobic oxidation of benzyl 4-phenyl-3-butynoate-30 was examined by using their best catalyst under their optimal reaction conditions. It was found that the resultant selectivity factor (k_rel = 5) is lower than that (k_rel = 7) carried out by our best system (entry 2, Table 5). VO(Oi-Pr)₃ is moisture sensitive. Anhydrous acetone needs to be used in Toste’s system. Therefore, our current chiral vanadyl(V) methoxide catalyst system represents a highly enantioselective, complementary, and water-tolerant alternative particularly toward the asymmetric oxidations of α-hydroxyamide substrate class.

Structural Studies of a Catalyst-Substrate Adduct.

To gain insights into the origin of enantiocontrol, we tried to get the single crystal of a stable vanadyl complex–substrate adduct. The slowest reacting, but highly enantioselective, catalyst 8 was combined with one equiv of the racemic N-benzyl-mandelamide 17b in toluene under anaerobic reaction condition (0.2 M), and the mixture was stirred at ambient temperature in argon for 40 min followed by heating at 100°C for 2 h. Some green precipitate was collected (14% yield) and recrystallized from degassed and argon-purged toluene to give a diastereomeric crystalline adduct 52 of x-ray quality. The filtrate contains mainly the (R)-17b (37%) in essentially optically pure form (99% ee) along with some α-ketoamide (17b′, 40%). Based on the crystallographic analysis of the adduct 52, the methoxide ligand in the original catalyst 8 has been replaced by the 2-alkoxide unit of the N-benzyl-mandelamide with release of MeOH (i.e., proton exchange) (Fig. 2). In addition, the amide carbonyl group is coordinated anti to the VO unit. Remarkably, only the (2R)-enantiomer is chelated by which the α-phenyl group stays away from the sterically congested 3-tert-butyl group in the template of the catalyst 8. In addition, the amide is in s-cis conformation, and the phenyl group of the N-benzyl unit in the substrate is positioned underneath the salicylidene template. More intriguingly, the phenyl group is sandwiched between the 3-tert-butyl and 5-methoxy groups in the template in a perpendicular π–π staking fashion.

Fig. 2.

ortep (www.ornl.gov/sci/ortep/ortep.html) drawing (ellipsoids are shown at 20% probability level) for the x-ray crystal structure of the catalyst–substrate adduct 52.

The oxygen-sensitive adduct 52 was then subjected to the oxidation in molecular oxygen in toluene again at normal concentration (0.2 M). The oxidation went to completion in ≈1 h with a half-life (t_1/2) of 20 min rather than 60 h as shown in Table 2, leading to the α-ketoamide in quantitative yield. The results indicate that one of the rate-limiting steps of the catalytic process is the deprotonation of the substrate by the methoxide in the catalyst. The thermodynamically more stable adduct 52 as shown in the x-ray analysis is a slower-reacting diastereomeric species toward oxidation (Fig. 3). On the contrary, the sterically more encumbered diastereomeric adduct 52′ is faster reacting for subsequent α-proton elimination process leading to α-ketoamide 17b′.

Fig. 3.

Reactivity difference for the diastereomeric adducts formed between a vanadyl(V) methoxide and racemic mandelamide.

The nearly exclusive enantiocontrol is further demonstrated by subjecting each enantiomeric N-benzyl-mandelamide under the same aerobic oxidation conditions catalyzed by 10. At 85% consumption of the (S)-enantiomer, the (R)-antipode-17b remains intact.

Mechanistic Proposal.

It was believed that the catalytic process proceeds through initial deprotonation of the substrate by chiral vanadyl(V) methoxide (e.g., 8 or 10) followed by ligand exchange, resulting in chelation of the substrate to the catalyst to give the diastereomeric adduct-52′ with extrusion of methanol (i.e., 2 equiv) (Fig. 4). α-Proton elimination of the kinetically favored adduct 52′ through a two-electron oxidation process with concomitant reduction of the vanadyl(V) species to the corresponding vanadium(III)OH would lead to the oxidized product (i.e., α-ketoamide-17b′) or the coupled adduct 53. The vanadium(III) hydroxide may disproportionate with the starting vanadyl methoxide catalyst to give two vanadyl(IV) complexes with release of N-benzyl benzoyl-formamide and CH₃OH (pathway I, Fig. 4). Alternatively, it may be oxidized by reaction with molecular oxygen to lead to a peroxo-dimer 54, which returns to the original catalyst but now with a hydroxide ligand instead of methoxide to complete the catalytic cycle (pathway II, Fig. 4).

Fig. 4.

Proposed mechanism.

In conclusion, we have documented a successful example of using N-salicylidene-l-α-amino acid-based vanadyl(V) complexes for highly enantioselective and chemoselective oxidations of racemic, functionalized α-hydroxy esters and amides under oxygen atmosphere at room temperature. Judicious selections of the C3,C5 substituents and pendant chiral groups in the template and solvents allow us to access the optimal vanadyl complexes. Kinetic experiments and x-ray crystallographic analysis of the more stable diastereomeric adduct indicate the involvement of vanadyl(V)-α-alkoxy ester/amide adduct responsible for subsequent two-electron oxidation event and clarify the origin of stereocontrol due to tight chelation of the substrate to the vanadyl catalyst. The current protocol works well for α-aryl-α-hydroxyesters and essentially all kinds of α-hydroxyamides and also can be applied to racemic taxol C-13 side-chain resolution, auguring well for its potential applications in the pharmaceutical industry.

Materials and Methods

Representative Preparation Procedure and Analytical Data for Vanadyl(V) Methoxide (or Hydroxide) Complexes 1–13.

In a 50-ml, two-necked, round-bottom flask was placed l-tert-leucine (5 mmol) and NaOAc–5H₂O (1.17 g, 10 mmol) in degassed water (10 ml). After having been stirred at 60°C for 10 min to effect their complete dissolution, the reaction mixture was treated dropwise with a solution of respective 2-hydroxy-benzaldehyde derivatives (5 mmol) in degassed EtOH (12.5 ml). The reaction mixture became homogeneous by heating at 80°C for 15 min and then gradually cooled to ambient temperature for 2 h. To the resultant Schiff base was added a solution of vanadyl(IV) sulfate trihydrate (1.08 g, 5 mmol) in degassed water (5 ml). Dark green complex started crashing out in 15 min. The reaction mixture was stirred for 2 h and then concentrated to half of the original solvent volume. The crude vanadyl(IV) complex collected by filtration was washed sequentially with water (5 × 25 ml) and cold ether (5 × 25 ml) and then dried in vacuo to furnish pure vanadyl(IV) catalyst. The corresponding analytically pure vanadyl(V) methoxide (or hydroxide) complexes were obtained by recrystallization from oxygen-saturated MeOH and were used for asymmetric aerobic oxidation experiments.

Representative Procedure for the Asymmetric Aerobic Oxidation of α-Hydroxy-Esters and Amides.

Into a 50-ml, two-necked, round-bottom flask was placed vanadyl catalyst-10 (0.05 mmol; 5 mol%) in oxygen-saturated toluene (3 ml) under oxygen atmosphere. The reaction flask was vacuum-evacuated at 15 torr for 20 sec and then filled with an O₂ balloon (150 ml). A solution of α-hydroxy-ester/amide (1 mmol) in oxygen-saturated toluene (2 ml) was added by cannula, and the resulting dark brown mixture was stirred at ambient temperature. The reaction progress was monitored by ¹H NMR spectroscopy for percent conversion (142 mg; i.e., 1 mmol of 2-methyl-naphthalene was used as an internal standard). The enantiomeric excess of the kinetically resolved product was determined by chiral HPLC analysis after filtration of the reaction aliquot (100 μl) over a short plug of silica gel (Et₂O or CH₂Cl₂ as eluent). Upon reaching optimal resolution of the asymmetric oxidation (50–64% conversion), the reaction was quenched by addition of silica gel (150 mg), and the mixture was concentrated under reduced pressure. The resulting residue was loaded directly on top of an eluent-filled silica gel column and purified by flash column chromatography. The enantiomeric excess of the pure, resolved α-hydroxy-ester/amide was analyzed again by chiral HPLC analysis.

Preperation of the Vanadyl(V) Complex-N-Benzyl-Mandelamide Adduct 52.

To a reaction tube (15-mm outer diameter; 100-mm length) equipped with a magnetic stirring bar was placed vanadyl(V) methoxide catalyst 8 (435 mg; 1 mmol) and racemic N-benzyl-mandelamide 17b (241 mg, 1 mmol) in freshly distillated argon-purged toluene (5 ml) under argon atmosphere. After it was stirred at ambient temperature for 40 min, the solution turned deep blue. The resulting deep blue solution then was heated at 100°C for 2 h under argon atmosphere. The resulting mixture was gradually cooled to ambient temperature and concentrated in vacuo to give a green needle solid. The needle solid was redissolved in degassed CH₂Cl₂ (5 ml) and was reprecipitated by slow addition of degassed hexane (5 ml). Green precipitate of 52 was obtained from the mixed solvents by slow decantation. The collected filtrate was loaded directly on top of an eluent-filled silica gel column and then purified by flash chromatography. The oxidized product N-benzyl-2-oxo-2-phenyl-acetamide-17b′ was isolated in 40% yield (96 mg) along with the resolved (R)-N-benzyl- mandelamide-17b in 37% yield (8 9 mg; 99%ee).

The green precipitate again was redissolved in degassed CH₂Cl₂ (3 ml), and argon-purged hexane (3 ml) was slowly diffused to the solution under argon atmosphere at ambient temperature for 3 days. The deep brown fine single crystals of 52 were collected in 12% yield (73 mg). The absolute stereochemistry for 52 was determined by x-ray crystallographic analysis.

Supporting Information.

Spectral data and characterization for all the vanadyl(V) methoxide complexes 1–15, kinetic resolution products 16–51 and oxidation products 16′–51′, and selected X-ray data of the adduct-52 are included as Appendices 1–5 and Data Sets 1 and 2, which are published as supporting information on the PNAS web site.

Abbreviation

ee

enantiomeric excess.