Preparation of α-amino acids via Ni-catalyzed reductive vinylation and arylation of α-pivaloyloxy glycine.

This work emphasizes easy access to α-vinyl and aryl amino acids via Ni-catalyzed cross-electrophile coupling of bench-stable N-carbonyl-protected α-pivaloyloxy glycine with vinyl/aryl halides and triflates. The protocol permits the synthesis of α-amino acids bearing hindered branched vinyl groups, which remains a challenge using the current methods. On the basis of experimental and DFT studies, simultaneous addition of glycine α-carbon (Gly) radicals to Ni(0) and Ar-Ni(ii) may occur, with the former being more favored where oxidative addition of a C(sp2) electrophile to the resultant Gly-Ni(i) intermediate gives a key Gly-Ni(iii)-Ar intermediate. The auxiliary chelation of the N-carbonyl oxygen to the Ni center appears to be crucial to stabilize the Gly-Ni(i) intermediate. This journal is © The Royal Society of Chemistry.

Introduction

Ni-catalyzed cross-electrophile coupling has emerged as a powerful tool for expeditious creation of C(sp3)–C bonds,[1] in which alkyl radical intermediates are generally involved. However, to the best of our knowledge, application of this strategy to the reductive coupling of α-glycinyl electrophiles, which enables direct decoration of the α-carbon of glycine derivatives, remains unexplored.[2] A prominent obstacle may be the rapid conversion of α-glycinyl electrophiles to imino or iminium esters that are well-established for addition reactions.[3,4] Reduction of an α-glycinyl electrophile to generate a glycinyl α-C(sp3)-radical was hitherto unknown. Realization of such a radical forming process may invoke a reductive coupling protocol for facile structural enrichment of α-amino acids.[2]

Unusual and unnatural α-vinyl and -aryl amino acids have seen a broad range of applications in drug discovery, biomaterials, and protein engineering.[5-7] Selected examples include amoxicillin,[8] forphenicine,[9] the trypsin inhibitor radiosumin,[10] the phytotoxin rhizobitoxine,[11] butadienylglycine (found in the defensive secretion of beetles),[12] and 2-amino-3-(3,4-dihydroxyphenyl)but-3-enoic acid as an antidepressant (Scheme 1a).[13] Numerous synthetic methods have been developed to access α-aryl and -vinyl amino acids based on two-electron addition (e.g., R-Li, -Zn and -B and electron-rich arenes) to glycine cation equivalents such as iminoesters (Scheme 1b).[14-16] Nevertheless, decoration of glycine α-carbons with branched vinyl groups bearing multiple substituents remains a challenge.[17]

Scheme 1

(a) Examples of bioactive compounds consisting of unusual amino acids. (b) C(sp2)-functionalization of glycine cation equivalents using 2e addition strategies. (c) Ni-catalyzed reductive coupling strategy for the preparation of unusual α-amino acids.

Herein, we describe the Ni-catalyzed reductive coupling of N-carbonyl-protected α-pivaloyloxy glycine derivatives with aryl and vinyl halides/triflates to afford unusual α-aryl/vinyl amino acids. The use of α-pivaloyloxy glycine proved crucial for the coupling reaction probably because it converts in situ into the active α-iodoglycine or iminium ester at low concentrations. Upon SET reduction or halide abstraction, a glycinyl (Gly) radical intermediate was produced. DFT studies suggest that the Gly radical may simultaneously add to Ni(0) and Ar–Ni(ii), but with the former being more favored to give a highly stable Gly–Ni(i) chelate wherein the auxiliary chelation of the N-carbonyl oxygen to the Ni center appears to be pivotal. Oxidative addition of an aryl halide to the Ni(i) species forms a key Gly–Ni(iii)–Ar intermediate (Scheme 1c). The necessity of the N-carbonyl protecting groups may provide useful information about coupling reactions of functionalized alkyl precursors which are currently essential for Ni-catalyzed stereoselective synthesis of C–C bonds.[18]

Results and discussion

Optimization of the reaction conditions

We began our studies by investigating the coupling of benzoyl protected α-pivaloyloxy glycine ethyl ester 1a and (2-bromovinyl)benzene 2 (E/Z = 5 : 1). An extensive survey of experimental parameters led us to identify the optimal reaction conditions at 60 °C comprising NiBr2/dtBBpy in combination with the reductant Zn, the additives MgCl2 and TBAI, and the solvent 1,4-dioxane.[19] The vinyl glycine product 3a was isolated in 75% yield consisting only of the E-isomer (Table 1, entry 1).[20] Control experiments indicated that NiBr2, Zn, MgCl2, and TBAI all were essential reagents (entries 2–5). Other nickel sources, ligands or temperatures gave inferior results (entries 6–10 in Table 1). The reaction could be performed on a gram scale with slightly diminished efficiency (entry 11). The need for 5 mol% ligand (L1) was tentatively explained by the inhibited formation of hydro-reduction product 1a-H. The remaining non-ligand chelated 5 mol% Ni may be stabilized by weak coordination with 1a, which could ensure the formation of a decent amount of Ni–L1 complex for efficient catalysis, provided formation of inactive Ni species is possible over the course of the reaction.

Optimization for the reaction of 1a with 2

Entry	Variation from the standard conditions	Yield%a
1	No changes	65 (75)b
2	w/o Ni	Trace
3	w/o Zn	Not detected
4	w/o TBAI	Trace
5	w/o MgCl2	Trace
6	NiCl2	10
7	L2 instead of L1	Trace
8	L3 instead of L1	62
9	L4 instead of L1	18
10	L5 instead of L1	Not detected
11	1a (8 mmol)	63c

NMR yield using 2,5-dimethylfuran as the internal standard.

Isolated yield.

The reaction was run on a gram scale.

Substrate scope

The established vinylation conditions (Table 1, entry 1) proved to be general for various vinyl halides (Fig. 1). The substituents on the phenyl rings of styrenyl bromides include both electron-donating and -withdrawing groups as evidenced by 3b–3f. For 3d, 20% of the enamine tautomer was also detected.[19] The naphthyl, anthranyl, furyl, and thiyl-conjugated vinyl bromides all delivered the corresponding vinyl glycines (4–7) in good to excellent yields. The dienyl glycine 8 was obtained in 75% yield, which can be used for further transformations (e.g., Diels–Alder reactions). The alkyl-substituted 1-vinyl bromides (E and Z mixtures) en route to 9 and 10 were also viable, wherein conversion of the Z-vinyl bromides to the E-products was much less effective than that observed for 2 (Table 1). The 2,2-disubstituted vinyl bromides appeared to be slightly more effective than the mono-substituted ones (e.g., 11–15). An outstanding feature of this vinylation protocol was its competency in the cases of α-vinyl bromide substrates, which produced the phenyl, alkyl, and silyl-substituted vinyl glycines 16–20 in moderate to good yields. The low yield for 18 was due to homocoupling of the α-vinyl bromide. The 1,2-disubstituted Z-alkenyl bromides were compatible with the coupling conditions, affording 21–24 in moderate to good yields. To our delight, the trimethyl substituted vinyl bromide delivered 25 in 44% yield, which is difficult to access with concurrent methods.

Fig. 1

Coupling of 1a with vinyl bromides and triflates, and aryl halides using the conditions in Table 1, entry 1 with isolated yields. aYield of isomerization to α-enamine determined by 1H NMR analysis after purification. bA mixture of E/Z vinyl–Br in a ratio of 1.8 : 1 was used. cA mixture of E/Z vinyl–Br in a ratio of 10 : 1 was used. dNiCl2(DME).

Vinyl triflates that could be readily obtained from ketones or alkynes were viable (Fig. 1). By replacing NiBr2 with NiCl2(Py)4, coupling of 1a with cyclohexenyl triflate produced 26 in 80% yield. Products 27 and 28 bearing cycloheptenyl and cyclooctenyl groups were obtained in moderate yields, while the formation of cyclododecenyl amino acid 29 was of lower yield. The conjugated 3,3,5-trimethyl-4-oxocyclohexa-1,5-dien-1-yl, sterically hindered 2,3,3a,4,7,7a-hexahydro-1H-4,7-methanoindenyl and phenyl-conjugated vinyl triflates within a cycloheptene ring resulted in 30–32 in moderate yields. While 2-hexenyl glycine 33 was obtained in 62% yield, the more hindered heptenyl glycine 34 was produced in 20% yield.

The vinylation method was also extended to the generation of arylated glycines 35–40 (Fig. 1). The reactions exhibited excellent functional group compatibility, retaining such functionalities as aldehyde in 35c and vinyl in 35f and 35g. The naphthyl and anthrenyl bromides afforded 36 and 37 in high yields. Electron-rich heteroaromatics such as 3-bromofuran and thiophene gave 38 and 39 in low yields. This problem could be solved using the iodo analogs, as exemplified by 40 and 35j.

The N-carbonyl protecting groups ranging from benzoyl groups decorated with electron-donating and -withdrawing substituents, to thiophenyl, Cbz, and Boc were suitable as evidenced in 41–45 (Fig. 2). In the case of arylation, Fmoc was more effective than Boc and Cbz groups (see 46–48). In contrast, N-tert-butylsulfinyl was ineffective. Moreover, hydrolysis of 26 and 35k with hydrochloric acid afforded the amino acid salts 50 and 51 in good yields (eqn (1)).

Fig. 2

Variation of amine-protecting groups. aNiCl2 was used in place of NiBr2.

Mechanistic and computational studies

Possible in situ formation of an α-haloglycine or iminium salt

The use of pivaloyloxy glycine is crucial for this coupling event.[21] Reaction of bromobenzene with unstable iminoester 52 (with or without 1 equiv. of PivOH) or unstable α-bromo glycine 53 only resulted in a trace amount of 35a (eqn (3)). The formation of hydro-deoxygenative glycine 1a-H as the major product (eqn (2)) implied that pre-formed iminoester or α-halo glycine was not suitable for this coupling event.[19]

Exposure of 1a to TABI and/or MgCl2 resulted in nearly full recovery of 1a even after 48 h, indicating that halide substitution was unable to effectively cleave the C–O bond in 1a (eqn (3)). However, in the presence of Zn and TBAI, 1a was consumed and majorly reduced to 1a-H within ∼1 hour, while the use of Zn and MgCl2 was ineffective. It was noted that in the standard catalytic coupling reaction, 1a went to completion within ∼1 h (Fig. S1†).[19] We reason that TBAI triggers the conversion of 1a to an imino/iminium ester or α-iodoglycine which acts as the actual coupling partner. This process could be equilibrated or slow so that low concentrations of the active species were maintained. When Zn was present, fast reduction of the in situ generated active species facilitated the consumption of 1a and ensured a matched coupling reactivity with the C(sp2)-partners (eqn (4)). No deuterium incorporation into 1a-H was detected when D2O was added to eqn (3), implying that a radical process may take place in the reduction.

Formation of an α-glycinyl carbon radical intermediate

More importantly, α-cylcopropyl imine 54 resulted in the radical ring-opening product 55 under Zn/MgCl2/TBAI conditions (eqn (5)). The catalytic reaction as in Table 1, entry 1 was completely inhibited upon introduction of 2 equiv. of TEMPO.[19] These results suggest that the catalytic reaction involves generation of a glycine α-carbon radical via single-electron reduction/halide abstraction of the in situ-formed iminium intermediate or α-iodoglycine by MgCl2-activated Zn, or L–NiI (eqn (6)).[22-24] The L–NiI species can be generated from reduction of L–NiII by Zn or from the reductive elimination of a Gly–(L)NiIII–Ar intermediate (Scheme 1c).

Addition of the α-glycinyl carbon radical to ArNi(0) vs. ArNi(ii)

According to a well-established radical-chain mechanism, the putative Gly radical could be trapped by an Ar–Ni(ii) species (eqn (6) and Scheme S4†).[24] Thus, we performed stoichiometric reactions of the Ar1–NiII complex 56 with 1a with and without Zn (eqn (7)).[19,25] The reactions resulted in 35l in 30% and 60% yields, respectively, suggesting that C(sp2)–NiII complexes might engage in the coupling event.[24] However, in a deliberately designed competition reaction of 1a with an equimolar mixture of 55 and ethyl bromobenzoate (Ar2Br), 35b was formed in preference to 35l (2.3 : 1) (eqn (8)). Without Ni/L1, 35b remained the dominant product when Zn was used. The preferential formation of 35b should not be directed by addition of a Gly radical to the in situ formed Ar2–Ni(ii), which would instead give 35I as the major product. Thus, a competing process that bypasses the aryl–NiII species may occur and be more favored. This idea was further confirmed by the reaction profiles shown in Fig. 3. The formation of 35b becomes much faster than that of 35l after an induction period, during which Ni(0) species is presumably generated.[26] Likewise, the reaction cannot be dictated by Ar–Ni(i) that is most likely generated from single electron reduction of Ar–Ni(ii).[27,28] This would again warrant 35I as the major product, which opposes the observation in eqn (8).

Fig. 3

The reaction profile for the formation of 35l and 35b in eqn (8) in the presence of Zn. Yield based on 1 equiv. of 1a.

The seemingly unusual observation in eqn (8) led us to consider the viability of addition of a Gly radical to Ni(0), which would generate a Gly–Ni(i) species.[29] Subsequent oxidative addition of an aryl halide to Gly–Ni(i) would give the key Gly–Ni(iii)–Ar intermediate (Scheme 1). DFT computations provide support for this mechanistic proposal (Scheme 1 and Fig. S4†).[19] There is a large decrease of free energy of 66 kcal mol−1 as the Ni(0) catalyst binds the Gly radical carbon, forming the highly stable Gly–Ni(i)–L intermediate IM1 wherein the chelation of benzoyl oxygen to the Ni(i) center is prominent. Oxidative addition of bromobenzene to IM1 requires overcoming an energy barrier of 16.9 kcal mol−1 to give the Ni(iii) intermediate IM2.

The subsequent reductive elimination proved to be rate-determining with an activation energy of 17.3 kcal mol−1. In comparison, DFT studies indicate that the radical-chain mechanism as in Scheme S4† was disfavored, because the precursor complex I5 to the oxidative addition of bromobenzene to L–Ni(0) would be much less stable than complex IM1 (Fig. 4).[19] The Ni(0) catalyst would proceed to IM1, which is completely irreversible.

Fig. 4

Free energy profile comparing the favored Ni0 → NiI → NiIII → NiI pathway (black) and the Ni0 → NiII → NiIII → NiI pathway (red) with upper left superscripts indicating the spin states of open-shell, paramagnetic species.

Conclusions

In summary, we have demonstrated that α-pivaloyloxy glycine effectively coupled with vinyl and aryl halides/triflates to afford C(sp2)-functionalized α-amino acids under Ni-catalyzed reductive conditions. This method displays unique competency for incorporating hindered α- and tri-substituted vinyl moieties into the α-position of glycine, which is unattainable by previous methods. Mechanistically, a glycine α-carbon radical is thought to arise from the reduction of an in situ generated iminium or α-iodoglycine by Zn or a Ni(i) species, which then participates in the coupling process by addition to Ni(0) and Ar–Ni(ii). DFT calculations showed that addition of the glycinyl carbon radical to L–Ni0 to give L–NiI–Gly followed by oxidative addition to ArX is possibly a more favored process. The auxiliary chelation of N-carbonyl to the Ni center appears to play a profound role in stabilizing the Ni intermediate, which may become crucial in developing an asymmetric version of this amino-acid forming event, which is ongoing in our laboratory and will be reported in due course.

Conflicts of interest

There are no conflicts to declare.