An Asymmetric sp³–sp³ Cross-Electrophile Coupling Using ‘Ene’-Reductases

Summary paragraph:

The catalytic asymmetric construction of Csp³–Csp³ bonds remains one of the foremost challenges in organic synthesis.¹ Metal-catalyzed cross-electrophile couplings (XEC) have emerged as a powerful tool for C–C bond formation.^2–5 However, coupling two distinct Csp³-electrophiles with high cross- and stereoselectivity continues as an unmet challenge. Here, we report a highly chemo- and enantioselective Csp³–Csp³ XEC between alkyl halides and nitroalkanes catalyzed by flavin-dependent ‘ene’-reductases. Photoexcitation of the enzyme-templated charge-transfer complex between an alkyl halide and flavin cofactor enables the chemoselective reduction of alkyl halide over the thermodynamically favored nitroalkane partner. The key C–C bond-forming step occurs via the reaction of an alkyl radical with an in situ generated nitronate to form a nitro radical anion that collapses to form nitrite and an alkyl radical. An enzyme-controlled hydrogen atom transfer affords high levels of enantioselectivity. This reactivity is unknown in small molecule catalysis and highlights the potential for enzymes to use new mechanisms to address long-standing synthetic challenges.

Catalytic cross-couplings to forge Csp²–Csp² bonds have revolutionized organic synthesis, enabling the rapid construction of molecules for the pharmaceutical and agrochemical industries.^6,7 As target compounds become more complex — correlating with a higher percentage of sp³-hybridized atoms and stereocenters — there is a need for technologies to forge Csp³–Csp³ bonds stereoselectively.^8,9 Cross-electrophile couplings (XEC) involving two distinct Csp³-electrophiles are an attractive alternative to the traditional cross-couplings because they have broad functional group tolerance and avoid the need for sensitive organometallic reagents.^10–12 However, these reactions often form homo-coupled products because the metal catalysts struggle to distinguish between the two Csp³-electrophiles. This issue can be diminished using alkyl halides that react at different rates with the metal catalyst.^13,14 Moreover, while there has been significant progress toward catalytic asymmetric Csp²–Csp³ XEC,^4,5 stereoselective Csp³–Csp³ XEC is underdeveloped (Fig. 1a).^15,16 To overcome these limitations, previously unappreciated mechanistic steps and catalytic strategies need to be explored.^17–19

Fig. 1. Photoenzymatic asymmetric cross-electrophile coupling reactions.

a, Challenges associated with sp³–sp³ XEC. b, Proposed photoenzymatic asymmetric XEC. c, Two stereocomplimentary ‘ene’-reductases catalyze the model reaction at pH 9.0. Enantiomeric ratio (er) refers to as the ratio of (S)- to (R)-enantiomer. ERED, ‘ene’-reductase; GDH, glucose dehydrogenase; NADP⁺, nicotinamide adenine dinucleotide phosphate.

We questioned whether an enzyme could catalyze an asymmetric Csp³–Csp³ XEC. The high level of selectivity associated with biocatalytic reactions makes them attractive scaffolds for this challenge.^20,21 However, as natural enzymes do not catalyze cross-coupling reactions, we needed to develop a novel XEC mechanism that is compatible with existing enzymatic machinery.^22,23 Nitroalkanes are unique and ubiquitous reagents in organic synthesis but are not used as electrophiles for cross-couplings.²⁴ We recognized that the reactivity of nitronates could be used for a biocatalytic XEC. Nitronates react with open-shell electrophiles to forge a C–C bond and a nitro radical anion.^25,26 If this intermediate were to cleave mesolytically, the resulting radical could be quenched via hydrogen atom transfer (HAT) to afford the cross-coupled product.²⁷ The key to achieving this previously unknown reaction is identifying an enzyme to facilitate C–C bond formation, C–N bond mesolytic cleavage, and HAT (Fig. 1b). We envision a mechanism where reduction of the alkyl halide 1 forms an alkyl radical 4 that can react with an in situ generated nitronate 5 to forge a new C–C bond and a nitro radical anion 6. Enzyme-mediated homolytic cleavage of the C–N bond generates nitrite and an alkyl radical 7 that can be terminated via HAT to afford the cross-coupled product 3 (Fig. 1b). The proposed mechanism is attractive because the orthogonal reactivity of nitroalkanes and alkyl halides avoids undesired dimerization products.

Precise control over the chemoselectivity of the electron transfer events is required for the proposed reaction. Reduction of the nitroalkanes is thermodynamically favored by comparison to all but the most electronically activated alkyl halides (nitroalkanes E_p/2 = ~ −0.9 V vs. saturated calomel electrode, SCE;²⁸ alkyl halides E_p/2=−1.1 V to −2.5 V vs. SCE²⁹).³⁰ To achieve the desired reaction, we require a catalyst that will preferentially reduce alkyl halides instead of the nitroalkanes. We and others recently demonstrated that flavin-dependent ‘ene’-reductases (EREDs) can reduce alkyl halides using protein templated charge-transfer (CT) complexes.^31–33 Protein templated complexes provide the opportunity for substrate binding to override the inherent thermodynamic preference in electron transfer events. If the protein only forms the CT complex with the alkyl halide, it would be selectively reduced over the nitroalkane. Finally, EREDs can precisely control the radical terminating HAT step, enabling the formation of products with high levels of enantioselectivity.^34,35

We initiated our studies by exploring the photoenzymatic coupling of α-chloroamide 1a (E_p/2 = −1.65 V vs. SCE)³¹ with 1-nitroethylbenzene 2a (E_p/2 = −0.89 V vs. SCE)²⁸ catalyzed by a panel of EREDs under cyan light irradiation (λ_max = 497 nm) (Supplementary Table 1). To our delight, many of the enzymes provided the desired cross-coupled product 3a, with none providing the nitrated product (Supplementary Table 1). The most promising catalyst was the ‘ene’-reductase from Caulobacter segnis (CsER), providing product 3a with moderate yield (28%) but excellent enantioselectivity, with 95:5 enantiomeric ratio (er) of (S)- to (R)-enantiomer of the product. We hypothesized that the modest yield is likely due to inadequate concentration of the nitronate at pH 8.0 (nitroalkane/nitronate = 200:1, pKa of 2a is 10.3).³⁶ Indeed, in moving to the more basic reaction conditions at pH 9.0 (nitroalkane/nitronate = 20:1), the desired product is formed in high yield and excellent enantioselectivity (92% yield, 95:5 er) for the (S)-enantiomer, outperforming other tested EREDs (Fig. 1c and Supplementary Table 1). The ERED variant from Gluconobacter oxydans (GluER-T36A) favors the formation of the (R)-enantiomer of the product (51% yield, 10:90 er), providing a complimentary catalyst for accessing both enantiomers of the product (Fig. 1c). Control experiments confirmed that ERED, cyan light, and NADPH regeneration system (GDH/NADP⁺/glucose) are crucial for the desired reactivity (Supplementary Table 2). Notably, this reaction can be run on a preparative scale and afford product 3a in 72% isolated yield from a 0.1 mmol-scale reaction with no changes in enantioselectivity. We solved the crystal structure of wild-type CsER, and the docking model of 3a with CsER suggests the (S)-preference of CsER in this reaction (Supplementary Fig. 4). Notably, the coupled product is not formed when the same reaction is attempted using photoredox catalysts.³⁷ Instead, we observed nitroalkane reduction to the oxime using Ir(ppy)₃ as a photoredox catalyst, highlighting that this reactivity is unique to biocatalysis.

With the optimized reaction parameters in hand, we sought to explore the scope and limitations of this photoenzymatic transformation (Fig. 2). A variety of α-aryl nitroalkanes are well accepted as XEC partners with α-chloroamide 1a. α-Aryl nitroethanes possessing electron-donating or electron-withdrawing substituents at the meta- and para-position were efficiently converted to the desired enantioenriched β-stereogenic amide products (9-16) in yields of 80–98% with excellent enantioselectivity (>95:5 er). Ortho-substituted α-aryl nitroalkanes were less tolerated in the reaction, only the ortho-fluoro substituted nitroalkane was accepted to provide product 8 in 28% yield and 94:6 er (Fig. 2, and Supplementary Fig. 2). In addition, the larger substrate α-naphthalenyl nitroethane was well tolerated in this reaction, providing the corresponding product 17 with 58% yield and excellent enantioselectivity (99:1 er). This enzyme, however, was limited to relatively small alkyl substituents at the α-position. While the ethyl group was well accepted (19, 64% yield and 98:2 er), larger groups, such as n-propyl, were poorly reactive. Pleasingly, CsER could also accommodate heterocycles, including the electron-rich α-thiophene and the electron-deficient α-pyridine nitroethanes, providing the respective β-heterocycle substituted chiral amide products (22-24) in 40–95% yield and high enantioselectivity (up to 99:1 er). The compatibility of this photoenzyme was further highlighted by the acceptance of α-nitroester as a coupling partner, giving β-stereogenic 1,4-dicarbonyl product 25 in 96% yield and 82:18 er (Fig. 2).

Fig. 2. Scope of the photoenzymatic cross-electrophile couplings.

Reaction conditions: α-chloro carbonyl substrate (10 μmol, 2 equiv), nitroalkane (5 μmol, 1 equiv), GDH-105 (0.3 mg), NADP⁺ (0.05 μmol, 1 mol%), glucose (25 μmol) and purified ‘ene’-reductases (0.05 μmol, 1 mol% based on nitroalkane) in tricine buffer (100 mM, pH 9.0), with 6% DMSO as cosolvent, final total volume is 800 μL. Reaction mixtures were irradiated with cyan LEDs under anaerobic conditions at room temperature for 24 h. ^a Yields determined via LCMS relative to an internal standard (TBB). ^b Enantiomeric ratio (er) refers to as the ratio of (S)- to (R)-enantiomer, er determined by HPLC on a chiral stationary phase. ^c Isolated yields were based on 0.10 mmol-scale reaction. ^d 3 equiv of α-bromo ester (15 μmol) were used.

As for the alkyl halide scope, secondary amides with methyl or benzyl are well accepted when coupled with α-(p-CF₃)phenyl nitroethane (2j) respectively, affording the products (26-27) in 53-75% yields and good enantioselectivities (up to 95:5 er) (Fig. 2). Tertiary amides are also well tolerated by the reaction, with cyclic pyrrolidine, piperidine, morpholine, and linear Weinreb amides providing the corresponding products (28-33) in moderate to good yields and enantioselectivities (42%-95% yields, up to 99:1 er). Pleasingly, different classes of carbonyls as coupling electrophiles are also feasible, as exemplified by the coupling of α-halo ester and α-halo ketones with α-aryl nitroalkanes, giving the respective enantioenriched β-chiral substituted esters and ketones (34-37, Fig. 2). Notably, the β-chiral amide 3a can be reduced to the corresponding γ-chiral amine 41 with good yield and no erosion of stereoselectivity (70% yield, 95:5 er, Fig. 3). Additionally, the enzymatic product 16 can be hydrolyzed to give β-chiral acid 42 and further reduced to γ-chiral alcohol 43 in good yield and excellent stereoretention (99:1 er, Fig. 3).

Fig. 3. Derivatization of the enzymatic products.

Reaction conditions: ^a BH₃·Me₂S (3.0 equiv), THF, 0 to 65 °C, 5 h. ^b H₂SO₄ (4 M)/HOAc, 150 °C, 16 h. ^c BH₃·Me₂S (3.0 equiv), THF, 0 to 45 °C, 5 h.

Mechanistic studies were conducted to determine the mode of radical initiation. Specifically, we were interested in understanding why the less oxidizing α-chloroamide (1a, E_p/2 = −1.65 V vs. SCE) is reduced preferentially to the nitroalkane (2a, E_p/2 = −0.89 V vs. SCE). We hypothesized that an enzyme templated charge-transfer (CT) complex controlled the electron transfer events.^26,27 To probe this possibility, the cofactor FMN in CsER was entirely reduced to FMN_hq with sodium dithionite, which revealed negligible absorption around 500 nm (Fig. 4a). Upon the addition of chloroamide 1a, a new broad absorption band (λ_max = 480 nm) was observed, suggesting the formation of a CT complex between the FMN_hq and 1a. Interestingly, we found nitroalkane 2a can oxidize the ground state FMN_hq to generate a flavin feature with an absorption band around 450 – 500 nm. We attribute this feature to a mixture of flavin quinone and flavin semiquinone (Supplementary Fig. 10). Interestingly, when nitroalkane 2a is mixed with CsER and cofactor turnover mix under visible light irradiation, we do not observe reduction to the oxime, hydroxylamine, or hydrodenitrated products. This suggests that initial reduction of the nitroalkane can occur, in contrast to examples with photoredox catalysts,³¹ but subsequent electron transfers to form oximes and hydroxylamines do not transpire, making this single electron reduction event reversible under the reaction conditions. As the first irreversible step is the reduction of the alkyl halide, reversible nitroalkane reduction does not have a detrimental effect on the reaction.

Fig. 4. Mechanistic experiments.

a, UV-Vis spectra of reduced CsER (FMNhq) in the presence of substrates. b, Feed experiment. c, Density functional theory (DFT) calculations of the model reaction.

Another exciting feature of these reactions is the minimal formation of the enzyme-dependent hydrodehalogenated product (Supplementary Table 2), suggesting that alkyl halide reduction occurs when the nitronate is present in the enzyme active site. However, oxidation of FMN_hq by the nitroalkane obscures the observation of a higher-order CT complex (Supplementary Fig. 10). To avoid this issue, nitrone 38, a close analog of nitronate 5, was used in the UV-Vis spectra experiments because 38 can also readily react with 1a (Fig. 4a), mimicking the radical initiation step of the model XEC reaction. Intriguingly, a further enhancement of the CT complex spectra was observed when nitrone 38 was added to a sample containing 1a and the reduced CsER, indicating a quaternary CT complex was formed to enable efficient radical formation and prolongation. No CT complex was observed using free FMN_hq with 1a and 38, suggesting the CT complex is formed within the enzyme active site (Supplementary Fig. 11).

Next, we were interested in understanding the denitration step of the reaction. While we propose a mechanism where the alkyl radical reacts with the nitronate to form the unstable radical anion, which rapidly undergoes mesolytic cleavage, we recognized the possibility of a two-step mechanism where the coupled nitroalkane is included in a redox-neutral process. The coupled nitroalkane intermediate is a substrate for reductive denitration in this scenario. To probe this possibility, we considered the reaction using nitromethylbenzene (2l), which under standard photoenzymatic conditions forms a 1:1 mixture of cross-coupling product 18 (28% yield) and compound 40 retaining a NO₂ group (29% yield) (Fig. 4b). When the nitroalkane product 40 is resubjected to the reaction conditions, no cross-coupling product 18 was observed, indicating that the nitroalkane is not an intermediate in the denitrative coupling reaction. As nitro radical anions are proposed intermediates in non-enzymatic radical reactions but are not reported to undergo mesolytic cleavage,²¹ we postulate the protein facilitates the mesolytic cleavage event.

The reaction with nitromethylbenzene 2l indicates a competition between denitration and electron transfer to FMN_sq under the reaction conditions. To better understand the distinctive feature of this reaction, we conducted density functional theory (DFT) calculations on the model reaction (Fig. 4c) and the one involving nitromethylbenzene 2l (Supplementary Fig. 13). We found the initial addition step of α-amidyl radical Int-1 to nitronate 5 is occurring rapidly with a free energy barrier of only 9.9 kcal mol⁻¹ for the model reaction. The resulting radical anion Int-2 readily undergoes irreversible denitration (a free energy barrier of 9.6 kcal mol⁻¹) to give the radical Int-3, which is terminated by HAT from FMN_sq, as supported by deuterium labeling experiments (Supplementary Fig. 8), to provide the final product 3a (Fig. 4c). Next, the same DFT calculations were conducted with nitromethylbenzene 2l (Supplementary Fig. 13). While a similar free energy barrier (10.0 kcal mol⁻¹) was observed for the initial addition step, we found a higher energy barrier (13.4 kcal mol⁻¹) for the denitration step, indicating a slower denitration step when compared to the model reaction with 2a. Thus, as a competitive pathway to the desired cross-coupling, the radical anion Int-2’ can be terminated by oxidation by FMN_sq to provide product 40 (Supplementary Fig. 13).

In summary, we have established an unprecedented photoenzymatic enantioselective sp³–sp³ cross-electrophile coupling between the readily available alkyl halides and nitroalkanes. This new enantioconvergent Csp³–Csp³ bond formation reaction is powered by EREDs, highlighting the unparallel capability of biocatalysts in differentiating Csp³ electrophile substrates and controlling stereoselectivity. By using non-traditional coupling partners and mechanisms, our work addresses the long-standing selectivity challenge in transitional-metal catalyzed cross-electrophile couplings by exploiting the promiscuous unnatural reactivity of EREDs, thus expanding the biocatalyst toolbox for asymmetric C–C bond formations.

Data and materials availability:

The data that support the findings in this study are available from the corresponding author upon reasonable request. Crystallographic models and structure factors have been deposited in the Protein Data Bank with accession number 7TNB for CsER.