Chiral N,N′-dioxide/Mg(OTf)₂ complex-catalyzed asymmetric [2,3]-rearrangement of in situ generated ammonium salts

Abstract

Catalytic enantioselective [2,3]-rearrangements of in situ generated ammonium ylides from glycine pyrazoleamides and allyl bromides were achieved by employing a chiral N,N′-dioxide/Mg^II complex as the catalyst. This protocol provided a facile and efficient synthesis route to a series of anti-α-amino acid derivatives in good yields with high stereoselectivities. Moreover, a possible catalytic cycle was proposed to illustrate the reaction process and the origin of stereoselectivity.

The Lewis acid catalyzed asymmetric [2,3]-rearrangement of quaternary ammonium ylides formed in situ from glycine pyrazoleamides and allyl bromides.

Introduction

[2,3]-Rearrangements have been regarded as a class of synthetically powerful organic transformations due to their inherently high efficiency.¹ In particular, [2,3]-rearrangement of ammonium ylides has been extensively investigated for rapid construction of valuable nitrogen-containing molecules.² It is highly attractive to disclose asymmetric versions of such intriguing rearrangement,^3–5 but only a few examples concerning the catalytic enantioselective [2,3]-rearrangement of ammonium ylides have been reported to date.⁵ In 2014, Smith and co-workers demonstrated the first example of chiral isothiourea-catalyzed [2,3]-rearrangement of allylic ammonium ylides to gain optically enriched syn-configured α-amino acid derivatives (Scheme 1a).^5a In 2017, they developed an elegant tandem in situ protocol utilizing Pd/chiral isothiourea relay catalysis, which provides a direct method for the synthesis of syn-α-amino acid derivatives from N,N-disubstituted glycine aryl esters and allylic phosphates (Scheme 1b).^5e Recently, the group of Song reported an interesting study on an isothiourea catalyzed asymmetric [2,3]-rearrangement reaction of propargyl ammonium salts, which allows access to optically active allenyl α-amino amides.^5f Despite these impressive advances, there is still room for further development. For instance, chiral isothiourea is a unique catalyst with which syn-α-amino acid derivatives were preferentially afforded in current reports.^5a–e Given the wide and versatile use of α-amino acids in organic synthesis and pharmaceutical chemistry,⁶ it is highly desirable to search for new catalytic systems for [2,3]-rearrangement of ammonium ylides in terms of the catalyst, substrate scope and the method of ammonium ylide formation, as well as stereodivergence of products.⁷

Scheme 1. [2,3]-Rearrangements of allylic ammonium ylides.

Inspired by the achievements in enantioselective [2,3]-rearrangement^1–5,8 and our ongoing interest in synthesis of unnatural α-amino acid derivatives,⁹ we envisaged that chiral N,N′-dioxide-metal complex catalysts¹⁰ developed by our group would be potentially useful in promoting asymmetric [2,3]-rearrangement of allylic ammonium ylides upon rationally introducing pyrazoleamide groups.¹¹ As depicted in Scheme 1c, pyrazoleamide ammonium salts generated from glycine pyrazoleamides 1 and allyl bromides 2 were selected as the precursors of [2,3]-rearrangement. It was thought that such allylic ammonium salts could be activated by bidentate coordination with the chiral N,N′-dioxide–metal complex and then subjected to deprotonation with the assistance of an external base to afford ammonium ylides, which subsequently undergo [2,3]-rearrangement to deliver non-racemic α-amino acid derivatives. There are some difficulties associated with the catalytic asymmetric [2,3]-rearrangement, such as the compatibility of all reactants with the catalyst^5a,f and the background reaction in the presence of the external base.¹² Herein, we wish to disclose our effort toward one-pot asymmetric [2,3]-rearrangement of in situ formed allylic ammonium ylides. Chiral N,N′-dioxide/Mg(OTf)₂ (ref. 13) was found to promote the diastereo- and enantioselective rearrangement efficiently, and various anti-α-amino acid derivatives¹⁴ were readily obtained in good yields with high stereoselectivities (up to 95% yield, >19 : 1 anti : syn and 98.5 : 1.5 er) from easily available glycine pyrazoleamides and allyl bromides.

Results and discussion

In the initial study, N,N-dimethylglycine pyrazoleamide (1a) and cinnamyl bromide (2a) were selected as the model substrates to optimize the reaction conditions. The preliminary study indicated that the tandem ammonium salt formation/[2,3]-rearrangement took place well in the presence of an external base, and the desired product 3a was isolated in 91% yield with a 1 : 1 anti : syn ratio by using diisopropylamine as the base (see page 9 in the ESI† for more details). This result showed the difficulty in achieving the asymmetric version of such one-pot transformation. For the purpose of determination of the er value, compound 3a was converted to 4a quantitatively with MeOH at 60 °C for further analysis. Next, different metal salts coordinated with chiral N,N′-dioxide L-PrMe2 were examined in CH₂Cl₂ at 30 °C (Table 1, entries 1–4). Mg(OTf)₂ performed better than other metal salts, giving the desired rearranged product 3a in 91% yield, 1.8 : 1 anti : syn, and 67.5 : 32.5 er for the major diastereomer (Table 1, entry 3). The complex of Mg(NTf)₂ provided a comparable result (Table 1, entry 4). Subsequently, the amide moiety in the ligand was evaluated, and it was found that sterically bulky 1-adamantyl amine derived L-PrAd afforded better stereoselectivities (Table 1, entry 7 vs. entries 5 and 6). Further investigations on the chiral backbone in the ligand showed that the l-ramipril-derived L-RaAd was superior to the l-proline-derived L-PrAd and (S)-pipecolic acid-derived L-PiAd in terms of enantioselectivity (91% yield, 3 : 1 anti : syn, and 81 : 19 er; entries 7–9). Other reaction parameters were investigated and the solvent was proven to play a significant role in the reaction. With MeCN as the solvent, the amino acid derivative 3a was produced in 99% yield with 13 : 1 anti : syn and 93.5 : 6.5 er for the major diastereomer (entry 10). The enantioselectivity could be further enhanced when the reaction was performed at decreased temperature (−20 °C)¹⁵ while a slightly lower yield was obtained (82% yield, 12 : 1 anti : syn, 95.5 : 4.5 er; entry 11). To our delight, the addition of NaBAr^F₄ as an additive and preparation of the catalyst in EtOAc produced optimized results (entry 12; 94% yield, >19 : 1 anti : syn, 97 : 3 er). Moreover, when the product with a low anti : syn ratio was subjected to the reaction conditions, it was found that both anti : syn ratios and er values did not change. This result implied that an ultimately high anti : syn ratio was obtained during the [2,3]-rearrangement rather than epimerization of the product during the reaction (for further details, see ESI,† page 11).

Optimization of the reaction conditions^a.

Entry	Metal salt	Ligand	Yield of 3ab (%)	anti : syn of 3ac	er of 4ad
1	Sc(OTf)3	L-PrMe2	77	1 : 1	race/race
2	Ni(OTf)2	L-PrMe2	90	1.1 : 1	54 : 46/57.5 : 42.5
3	Mg(OTf)2	L-PrMe2	91	1.8 : 1	67.5 : 32.5/67.5 : 32.5
4	Mg(NTf)2	L-PrMe2	93	1.6 : 1	65.5 : 34.5/64.5 : 35.5
5	Mg(OTf)2	L-PrEt2	99	3 : 1	61 : 39/80 : 20
6	Mg(OTf)2	L-PrtBu	90	2 : 1	56.5 : 43.5/52.5 : 47.5
7	Mg(OTf)2	L-PrAd	98	5 : 1	75.5 : 24.5/57 : 43
8	Mg(OTf)2	L-PiAd	94	2 : 1	52.5 : 47.5/race
9	Mg(OTf)2	L-RaAd	91	3 : 1	81 : 19/54.5 : 45.5
10e	Mg(OTf)2	L-RaAd	99	13 : 1	93.5 : 6.5
11e,f	Mg(OTf)2	L-RaAd	82	12 : 1	95.5 : 4.5
12e,f,g	Mg(OTf)2	L-RaAd	94	>19 : 1	97 : 3

^aUnless otherwise noted, all reactions were carried out with 1a (0.10 mmol), 2a (0.10 mmol), iPr₂NH (0.15 mmol) and metal salt/ligand (1 : 1, 10 mol%) in CH₂Cl₂ (1.0 mL) at 30 °C for 14 h.

^bIsolated yield of 3a.

^cDetermined by ¹H NMR.

^dDetermined by HPLC on a chiral stationary phase.

^eCarried out in MeCN. The metal salt/ligand was pretreated in CH₂Cl₂.

^f−20 °C for 24 h.

^gNaBAr^F₄ {NaB[3,5-(F₃C)₂C₆H₃]₄} (20 mol%) was added. The metal salt/ligand/NaBAr^F₄ was pretreated in EtOAc instead of CH₂Cl₂. Tf = trifluoromethanesulfonyl.

With the optimized reaction conditions in hand, the substrate scope of [2,3]-rearrangement was screened (Table 2). Varying N-substituents in the glycine pyrazoleamides had no effect on the anti : syn ratio of the reaction (>19 : 1 in all cases). However, the reactivity and enantioselectivity dropped sharply with the increase of the ring size and steric hindrance of N-substituents (3a–3e). As shown in Table 2, the use of glycine pyrazoleamides bearing symmetrical N,N-dialkyl substituents, such as N,N-diethylglycine pyrazoleamide, gave the desired product 3b in 64% yield with a 95 : 5 er value. Cyclic N-piperidinyl and N-azepanyl substituents were also tolerated in this reaction, delivering the products 3c and 3d in moderate yields with satisfactory enantioselectivities. Lower yield (27% yield) with a good enantiomeric ratio (92 : 8 er) was obtained for product 3e bearing a N-heterocycle (morpholinyl) substituent. Next, the reaction of 1a with allyl bromide compounds 2 bearing different cinnamic aryl substituents was evaluated. Both the position and electronic properties of substituents had obvious effects on the reaction. Meta-substituted (E)-(3-bromopropenyl)benzenes afforded the corresponding products (3g and 3o) with better results than those with substituents at ortho- or para-positions (3f, 3h, 3n, 3p). Generally, the substrate with an electron-rich group transformed into the rearrangement product with a slightly higher anti : syn ratio than the substrate with an electron-deficient group (3h and 3p). Other para-halogen-substituted aryl rings, including 4-FC₆H₄, 4-BrC₆H₄, and 4-IC₆H₄ delivered the expected products (3j–3l) in 57–70% isolated yields of the major diastereomers with 90.5 : 9.5 to 96 : 4 er. The presence of the 4-F₃CC₆H₄ substituent led to acceptable outcomes (3m, 57% yield, 4 : 1 anti : syn and 92 : 8 er). The allyl bromide compounds bearing a 2-naphthyl group and heteroaromatic ring could also perform well to yield the desired products (3q–3s) with high diastereo- and enantioselectivities. Moreover, the allyl bromide compound 2 bearing a styryl functional group was suitable for the current reaction, producing the amino acid derivative 3t in 82% yield with >19 : 1 anti : syn and 97 : 3 er. In addition, this rearrangement process of N,N-dimethylglycine pyrazoleamide 1a with cinnamyl bromide bearing a methyl on the double bond gave the amino acid derivative 3u in excellent yields (95% yield, >19 : 1 anti : syn and 98.5 : 1.5 er). Unfortunately, when cyclohexyl-substituted allyl bromide was employed, only a trace amount of the desired product (3v) was detected even with a prolonged reaction time (7 days). The absolute configuration of product 4u was determined to be (2R,3S) by X-ray crystallography analysis,¹⁶ and the others were assigned by comparing their CD spectra with that of 3u (see pages 96–105 in the ESI† for more details).

Substrate scope for the [2,3]-rearrangement^a.

^aAll reactions were carried out with L-RaAd/Mg(OTf)₂/NaBAr^F₄ (1 : 1 : 2, 10 mol%), 1 (0.2 mmol), 2 (0.2 mmol), and iPr₂NH (0.3 mmol) in MeCN (2.0 mL) at −20 °C. Isolated total yield of product 3. The anti : syn ratio was determined by ¹H NMR analysis. The er value was determined by HPLC on a chiral stationary phase.

^bIsolated yield of the major diastereomer 3.

^cer value of product 4.

To illustrate the potential utility of the methodology, a scale-up synthesis of 3r was carried out under the optimized reaction conditions. As illustrated in Scheme 2a, 3 mmol of compound 1a reacted smoothly with equal amounts of allyl bromide 2n, furnishing the desired product 3r in 89% yield with >19 : 1 anti : syn and 98 : 2 er. Further transformations of the product 4a were conducted (Scheme 2b). Compound 4a was easily reduced to 5a in 88% yield with maintained stereoselectivities (>19 : 1 anti : syn, 96.5 : 3.5 er) by treatment with 10% Pd/C in methanol. Additionally, the reduction of 4a with LiAlH₄ generated the corresponding alcohol 6a in 71% yield (>19 : 1 anti : syn, 97.5 : 2.5 er).

Scheme 2. (a) Scale-up synthesis of 3r; (b) further transformation of product 4a.

Based on previous work,¹⁷ the proposed catalytic cycle and a possible working mode of the enantioselective [2,3]-rearrangements of ammonium ylides are depicted in Scheme 3. Initially, the reaction of N,N-dimethylglycine pyrazoleamide (1a) and cinnamyl bromide (2a) produced the corresponding ammonium salt I which was activated by bidentate coordination with the N,N′-dioxide–metal complex and subjected to deprotonation with the assistance of iPr₂NH to afford metal bonded ammonium ylide II. Due to the steric repulsion of the aryl group of the cinnamyl moiety of the substrate with the octahydrocyclopenta[b]pyrrole unit in the ligand L-RaAd as well as the pyrazoleamide unit in the exo-transition state,¹⁴ the rearrangement occurred preferentially to afford the anti-configured α-amino acid derivative (2R,3S)-3a, which was consistent with the experimental results.

Scheme 3. The proposed catalytic cycle and working mode.

Conclusions

We have successfully developed the first Lewis acid catalyzed asymmetric [2,3]-rearrangement of quaternary ammonium ylides formed in situ from glycine pyrazoleamides and allyl bromides. The N,N′-dioxide/Mg(OTf)₂ catalytic system benefited the rearrangement process efficiently, providing diverse chiral anti-α-amino acid derivatives in good yields with high stereoselectivities (up to 95% yield, >19 : 1 anti : syn and 98.5 : 1.5 er). Besides, the potential use of the current method was illustrated by gram-scale synthesis and further transformations of products. A possible catalytic cycle along with the working mode was proposed to elucidate the reaction process and chiral induction. Further investigations on other reactions enabled by chiral N,N′-dioxide–metal complex catalysts are ongoing in our laboratory.

Conflicts of interest

There are no conflicts to declare.