Benzotetramisole-Catalyzed Dynamic Kinetic Resolution of Azlactones

Abstract

Enantioselective acyl transfer catalyst benzotetramisole (BTM) has been found to promote dynamic kinetic resolution of azlactones providing di(1-naphthyl)methyl esters of α-amino acids with up to 96% ee.

Dynamic kinetic resolution (DKR)¹ of azlactones² by way of their enantioselective alcoholysis (Figure 1) provides an attractive approach to the asymmetric synthesis of α-amino acid derivatives. Enzymatic variant of this transformation,³ while often successful, suffers from one inherent drawback: since enzymes are available in only one enantiomeric form, the reversal of enantioselectivity in a given reaction requires identification of a different enzyme, which is not always a trivial matter. Therefore, the development of nonenzymatic alternatives is of significant practical interest. Several mechanistically different approaches to activating azlactones toward alcoholysis have been employed. In 1997, Seebach et al. reported a Lewis acid-catalyzed version of this reaction using Ti(IV) TADDOLate 3 producing moderate ee’s (up to 68%).⁴ Promising levels of enantioselectivity (up to 78% ee) were achieved by Fu et al. in 1998 using enantioselective acyl transfer catalyst 4 (Figure 2). However, the reaction was extremely slow (50% conversion/1 week).⁵ Low enantioselectivities (<39% ee) were reported by Hua et al. in 1999 using a combination of a chiral diketopiperazine, cyclo-[(S)-His-(S)-Phe] 5, with diisopropyl L-tartrate, presumed to activate azlactones via hydrogen bonding.⁶ More recently (2005–2008), encouraging results have been obtained by Berkessel et al.⁷ and Connon et al.⁸ using bifunctional catalysts 6 (72–95% ee) and 7 (78–88% ee), respectively.

Figure 1.

Dynamic kinetic resolution of azlactones.

Figure 2.

Catalysts previously employed for the DKR of azlactones.

Surprisingly, apart from Fu’s seminal study, there have been no other reported attempts to achieve DKR of azlactones using enantioselective acyl transfer catalysis. Over the past several years, our group has developed amidine-based catalysts 8–11 (Figure 3), which display high enantioselectivity in the acylation of several classes of alcohols and oxazolidinones.⁹ Recently, Shiina et al.¹⁰ have demonstrated the utility of BTM 10 in the kinetic resolution (KR) of α-arylpropionic acids via enantioselective alcoholysis of their mixed anhydrides. We have found that HBTM 11 is also effective in the KR of α-aryloxy- and arylthioalkanoic acids via their symmetrical anhydrides.¹¹ These results have encouraged us to re-examine the possibility of DKR of azlactones via the acyl transfer mechanism.

Figure 3.

Amidine-based catalysts used in this study.

An equimolar combination of HBTM and benzoic acid was found to promote the methanolysis of substrate (±)-1a without any appreciable enantioselectivity (Table 1, entry 1). Switching to benzyl alcohol resulted in a modest ee (entry 2). The reaction with diarylcarbinols was extremely slow (entries 3 and 4). After these disappointing first results, we were pleased to discover that BTM is much more effective in this reaction. Encouraging results were obtained even using methanol (entry 5); however, the bulky di(1-naphthyl)methanol^10,11 was required to bring the enantiomeric excess to a respectable 85% (entry 11). A single recrystallization from ethyl acetate produced completely enantiopure material (>99.5% ee). The earlier imidazoline-based catalysts, 8 and 9, proved to be competent but less active and less enantioselective than BTM (entries 12–14).

Table 1.

Catalyst and Alcohol Screening

entry	catalyst (mol %)	time, d	R2	% convna	% ee
1	11 (5)	1	Me	54	<3
2	11 (5)	1	PhCH2	47	−25
3	11 (5)	1	Ph2CH	5	ND
4	11 (10)	1	1-Np2CH	<5	ND
5	10 (5)	2	Me	91b	34b
6	10 (5)	2	PhCH2	94b	48b
7	10 (5)	2	1-NpCH2	97b	51b
8	10 (5)	2	2-NpCH2	96b	47b
9	10 (5)	2	Me2CH	<5	ND
10	10 (5)	2	Ph2CH	91b	75b
11	10 (5)	2	1-Np2CH	96b	85b
12	10 (10)	0.4	1-Np2CH	92b	80b
13	8 (10)	2	1-Np2CH	47	59
14	9 (10)	2	1-Np2CH	47	−52

^aConversion was determined by ¹H NMR, unless indicated otherwise.

^bReported % isolated yields and % ee’s are averages of two runs.

Changing the amount of benzoic acid relative to the catalyst did not have a significant effect on the reaction rate or the enantioselectivity, although in the absence of the acid promoter the reaction did not proceed at all, consistent with Fu’s original report⁵ (entries 1–4, Table 2). Lower catalyst loadings, down to 2 mol %, were still effective, although required prolonged reaction times (entries 5 and 6). Decreased temperatures proved to be detrimental to the enantioselectivity, while higher temperatures resulted in a higher rates, but the same ee (entries 7–9). Solvents other than chloroform were less effective, in line with our earlier experience with the KR of alcohols^9c (entries 10–13).

Table 2.

Variation of Reaction Conditions^a

entry	mol % 10/mol % BzOH	time d	temp °C	solvent	% convn	% ee
1	10/0	1	23	CDCl3	<5	ND
2	10/5	0.4	23	CDCl3	90	83
3	10/10	0.4	23	CDCl3	92	80
4	10/20	0.4	23	CDCl3	93	82
5	5/5	2	23	CDCl3	96	85
6	2/2	4	23	CDCl3	94	84
7	5/5	2	0	CDCl3	95	72
8	10/5	2	−20	CDCl3	94	55
9	5/5	1	45	CDCl3	91	84
10	5/5	2	23	C6D6	98	80
11	5/5	2	23	CH2Cl2	95	74
12	5/5	2	23	CH3CN	85	66
13	5/5	2	23	THF	<5	ND

^aConversion in all cases was determined by ¹H NMR.

DKR of other azlactones bearing primary alkyl substituents 1b–e produced uniformly good yields and ee’s in the 80–90% range (Table 3, entries 1–5). Isopropyl-substituted substrate 1f proved resistant to alcoholysis under the same conditions (entry 6). On the other hand, excellent enantioselectivity (94% ee) was obtained in the case of 2,4-diphenylazlactone 1g (entry 7). This result was all the more remarkable given the fact that the highest ee previously reported for either enzymatic^3a or nonenzymatic^7a DKR of this substrate (or any other 4-aryl-substituted azlactones) has been 75%. Variation of electronic properties of the C4-aryl substituent had only a slight effect on the enantioselectivity or reaction rate (entries 8–11). DKR of the 1-naphthyl-substituted substrate 1l, however, proceeded rather slowly and with lower enantioselectivity (entry 12). Substrate 1m was completely unreactive, presumably due to the presence of an ortho-substitutuent (entry 13).

Table 3.

Structural Variation of Azlactone Substrates^a

entry	substrate	time d	R1	% yieldb	% eeb
1	1a	0.4	Me	90	83
2	1b	7	Me2CHCH2	97	80
3	1c	3	MeSCH2CH2	92	90
4	1d	4	PhCH2	94	83
5	1e	4	CH2=CHCH2	88	83
6	1f	4	Me2CH	<5	ND
7	1g	2	Ph	89	94
8	1h	2	p-MeOC6H4	87	91
9	1i	2	p-ClC6H4	88	96
10	1j	2	p-BrC6H4	90	95
11	1k	2	2-naphthyl	88	91
12	1l	7	1-naphthyl	46	76
13	1m	2	2,4-(MeO)2C6H3	0	ND
14c	1h	2	p-MeOC6H4	86	92
15c	1i	2	p-ClC6H4	86	97
16c	1n	2	p-FC6H4	90	95

^aNa₂SO₄ was added to prevent loss of catalytic activity of BTM during prolonged reactions (see ref 9c).

^bReported % isolated yields and % ee’s are averages of two runs.

^cAzlactone was generated in situ. The yield is based on the N-benzoyl-α-arylglycine starting material (see Scheme 1).

Initially, azlactones 1a–m used in this study were synthesized in a separate step and purified by recrystallization before subjecting them to the DKR conditions. Later, however, we found that in situ cyclization of N-benzoyl-α-amino acids with DCC followed by addition of di-(1-naphthyl)methanol and the catalyst works just as well as the earlier, more time-consuming protocol (cf. entries 14–16 vs 8 and 9, Table 3). Coupled with the one-step synthesis of starting materials via the Ben-Ishai amidoalkylation¹² of simple arenes, this procedure provides an attractive route to enantioenriched arylglycine derivatives¹³ illustrated in Scheme 1. Hydrogenolysis of dinaphthylmethyl ester 2i proceeded without appreciable erosion of enantiomeric purity.

Scheme 1.

Asymmetric Synthesis of Arylglycines

It is of interest to analyze the mechanism of enantiodifferentiation in the DKR of azlactones. Berkessel et al.^7a proposed that the thiourea-catalyzed version of this process occurs via irreversible, hydrogen bonding-assisted attack of the alcohol on the less hindered face of the substrate, which results in the highest enantioselectivity being observed in the case of the bulkiest C4-substituents (cf. general structure 1, R¹ = isobutyl (1b), isopropyl (1f), or tert-butyl). The situation is clearly different in our case. The strong dependence of the enantioselectivity on the alcohol nucleophile indicates that enantiodifferentiation occurs in the second step of the catalytic cycle (Figure 4). The absolute sense of asymmetric induction and structure-selectivity trends can be explained by transition state model 15, which invokes hydrogen bonding between the benzamide group and the reacting carbonyl. The involvement of π–π interactions with the C2-phenyl group on the catalyst and/or lower steric repulsion may explain the higher enantioselectivity generally observed in the DKR of the aryl-substituted azlactones 1g–k compared to the alkyl-substituted ones 1a–e. The low reactivity of substrate bearing bulky C4 substituents (cf. 1f, 1l, and 1m) is also consistent with this proposal.

Figure 4.

Proposed catalytic cycle and transition state (benzoate anion omitted for clarity).

In conclusion, we have developed a new, highly enantioselective method for the DKR of azlactones. It is especially suited for the C4-aryl-substituted substrates, thus complementing the previously available enzymatic and nonenzymatic protocols. Further exploration of its substrate scope is under active investigation in our laboratory.