An enantioselective artificial Suzukiase based on the biotin–streptavidin technology

Introduction of a biotinylated monophosphine palladium complex within streptavidin affords an enantioselective artificial Suzukiase. Site-directed mutagenesis allowed the optimization of the activity and the enantioselectivity of this artificial metalloenzyme. A variety of atropisomeric biaryls were produced in good yields and up to 90% ee.


Results and discussion
With the aim of identifying suitable cross-coupling reaction conditions, we evaluated ve different biotinylated catalyst precursors 1-5 in the presence of either avidin or streptavidin (Avi and Sav hereaer). For this purpose, the reaction of 1iodonaphthalene 6c with 2-methoxy-1-naphthaleneboronic acid 7c was selected as a model system (Scheme 1). Initial experiments revealed that protonolysis of the boronic acid led to modest cross-coupling yields. Systematic variation of the base, pH, and organic co-solvent led to the identication of suitable reaction conditions: sodium hydroxide in DMSO : water (1 : 9) proved to be particularly effective in preventing protonolysis. However, competing deboronation required the use of excess boronic acid (1.5 equivalents vs. aryl halide) to achieve full conversion. Gel electrophoresis of the reaction mixture revealed that Sav remained largely tetrameric and active toward biotin binding even aer an SMC performed at 50 C (see ESI Fig. S3 †).
Having identied suitable reaction conditions, we screened biotinylated phosphine-and NHC ligands 1-5 combined with WT (strept)avidin (Table 1). This screening led to the identication of complexes 1, 2, and 3 as the most promising catalysts in combination with WT Sav. Despite the structural similarity between avidin and streptavidin, the result obtained with avidin was inferior to those obtained with Sav. We thus focused on Sav for further optimization studies. Past experience with other ArMs based on the biotin-streptavidin technology suggest that mutations at positions S112 and K121 (which lie closest to the biotinylated metal cofactor) offer a versatile means to rapidly optimize the ArM's performance. Complexes 1-3 were thus screened with a small library of Sav mutants. The results of the chemogenetic optimization for the synthesis of 2-methoxy-1,1 0binaphthyl 8b are displayed as a ngerprint in Fig. 1, and selected results are collected in Table 2.
From these data, the following features emerge: (a) Compared to the free cofactor, higher conversions are observed with the articial Suzukiase. This demonstrates that the Sav host protein exerts a benecial inuence on both the activity (i.e. protein accelerated catalysis) 36 and the selectivity.
(b) The electron-donating properties and bulkiness of the phosphine play a key role in determining the activity of the corresponding articial Suzukiase. Accordingly, the (t-Bu) 2 Pbearing catalysts outperform the (Ph) 2 P-systems. Strikingly, both ligands 1 and 3 ( Table 2, entries 10-13) show similar activity in case of S112M and S112A mutants.
(c) The enantioenriched nature of the cross-coupled product 8b strongly supports the hypothesis that the SMC is indeed catalysed by a homogeneous protein-embedded Pd-cofactor rather than Pd-nanoparticles. 12,25,26,37 (d) Varying the spacer between the biotin anchor and the P(t-Bu) 2 from ethyl to propyl affords opposite enantiomers of biaryl 8b for a given Sav mutant. The best ees in favour of (S)-8b were obtained with 2$Sav K121A and 2$Sav S112M ( Table 2, entries 8 and 9). These ndings illustrate the versatility of chemogenetic optimization strategy: upon varying the length of the spacer the Pd-moiety experiences a very different second coordination sphere environment which is reected in the enantioselectivity.
highest turnover with good ee ( Table 2, entry 22). This nding highlights the non-additive nature of multiple mutations. 38,39 Decreasing the temperature to 4 C leads to an improvement in enantioselectivity (86% ee, Table 2, entry 24) albeit at the cost of a slower rate. (b) Increasing the ratio of complex 1 vs. Sav tetramer (S112Y-K121E mutant) from one to four leads to a gradual erosion of enantioselectivity (from 80 to 69% ee, see Table S4, ESI †). This suggests that an empty biotin binding site adjacent to a complex 1 within Sav is favorable for selectivity.
(c) On a preparative scale, up to 90% ee (R)-8b, were obtained using 0.50 mol% catalyst loading and 0.25 mol% Sav (tetramer) loading at 4 C ( The articial Suzukiase performed well on a range of hydrophobic substrates, leading to the following observations (Table 3, Fig. 2): (a) The nature of the aryl halide does not inuence the enantioselectivity but to lower conversions on going from iodide to bromide to chloride 6a-c (Table 3, entries 6-8).
(b) Substitution of the methoxy group by either a methyl and ethoxy group on the arylboronic acid (compare 7c to 7b and 7d) leads lower enantioselectivities (Table 3, entries 7, 10 and 12). Depending on the nature of the ortho-substituent of the aryl moiety, the CIP priorities are inverted, yielding the opposite absolute conguration (e.g. compare 8b and 8d) for the same spacial arrangement. With the exception of the biphenylbinaphthyl 8h, all SMC products afford the same spacial arrangement using 1$S112Y-K121E Sav. We speculate that upon binding to the palladium cofactor, the bulky biphenyl group 7h forces the catalyst into a different second coordination sphere environment, leading to the opposite enantiomer (and spacial arrangement) of the coupled product.
(c) The system affords the highest enantioselectivity for binaphthyls or naphthyl-o-biphenyl products (albeit with the opposite chirality) in the presence of one smaller coupling partner, (e.g. 6h), the enantioselectivity is signicantly lower.

X-ray structure
To gain structural insight on the second coordination sphere around the Pd moiety, crystals of Sav S112Y-K121E were grown and soaked with a solution of 1 in DMSO (for details, see ESI †). The resulting crystals were subjected to X-ray diffraction at the beamline X06DA at the Swiss Light Source. In the resulting structure of complex 1$S112Y-K121E Sav, there is strong residual electron density in the 2F o À F c difference map in the biotin-binding vestibule. The Pd atom was unambiguously assigned as a strong peak in the difference map (12s) with a corresponding peak in the anomalous difference map (8s). The density map clearly highlights the presence of the phosphine ligand, the palladium, and the chloride (see ESI Fig. S5 †). However, the cinnamyl ligand could not be resolved, likely due to disorder caused by ligand uxionality. The amide linker exhibits two H-bonding interactions with the protein backbone: the amide nitrogen exhibits an H-bonding interaction with the side-chain of S88, and the carbonyl oxygen exhibits an Hbonding interaction with the backbone nitrogen of N49. The palladium atoms of two symmetry related-cofactors (face-toface) are separated by 11.7Å. The palladium was found in the vicinity of the O atoms of S112Y and K121E (Pd/O distance 6.4 A and 7.1Å respectively). The average B-factor of Pd is 36.10, suggesting that the complex may be uxional or located in a shallow energy minimum (Fig. 3).
In the synthesis of chiral biaryls by the SMC, enantioselectivity is dictated by the conformation of the Pd intermediate just prior to reductive elimination. 40,43 In previous DFT studies of enantioselective reductive eliminations in the SMC, it was shown that weak interactions between substrates and substrate-ligand interactions were largely responsible for biasing certain conformations and thus leading to enantioinduction. 6,40,44 By contrast, in the 1$Sav system, it is more likely that the origin of the enantioselectivity observed is from the outer-sphere ligand effect of the protein. Inspection of the X-ray structure of the active site of 1$S112Y-K121E shows that the Pdphosphine ligand is effectively "locked" into place by the steric environment of the Sav protein, which is evidenced by the discrete atomic position of the Pd and P and their relatively low B-factors despite proximity to the solvent pocket. The restricted conformational space around the Pd atom limits certain conformations. The nearby mutated residues play an additional role. In particular, mutation of S112 to the larger tyrosine provides a signicant steric barrier on the lower face of the Pd atom and is most likely responsible for the modest improvement in enantioselectivity upon mutation. Furthermore, the conformation of Y112 is effectively locked in place by an Hbonding interaction of the sidechain oxygen with the backbone carbonyl of E121 (distance 2.88Å), which would not be the case with the similarly large S112W mutation ( Table 2, entry 18). Further X-ray studies of 1$Sav with various mutations are currently underway.
It should be emphasized however that, while the X-ray structure sheds light on the position of the biotinylated catalyst precursor, we cannot exclude that the loaded catalyst, bearing the two bulky aromatic coupling partners, adopts a different position within the biotin-binding vestibule. As demonstrated recently for an articial imine reductase, QM-MM studies may shed light on the enantioselection mechanism and the precise geometry of the transition state. 41 In this context, reducing the cofactor exibility by introducing stabilizing interactions with the protein has proven useful. 42

Conclusions
In summary, incorporation of an electron-rich phosphinopalladium moiety within Sav affords an articial Suzukiase for the synthesis of enantioenriched binaphthyls (up to 90% ee and 50 TONs for 2-methoxy-binaphthyl 8b). While it is straightforward to identify aminoacids close to the cofactor, it is remains Fig. 3 Close-up view of the structure of the artificial Suzukiase 1$S112Y-K121E Sav (pdb code: 5CSE). The solvent-excluded surfaces of two symmetry-related Sav monomers are highlighted in grey and magenta respectively. The biotinylated Pd-cofactor 1 and the mutated residues S112Y and K121E are displayed as stick. No electron density for the cinnamyl-moiety (turquoise) could be detected and was thus modelled, minimizing steric-clashes. extremely challenging to predict or rationalize why a given mutation leads to an improvement in catalytic performance. The chemogenetic optimization strategy allows to address this challenge with a limited effort: combining a small library of m mutants with a small library of n biotinylated ligands affords an m$n diversity matrix. Current efforts are aimed at performing catalytic asymmetric SMC with articial Suzukiases in vivo.