Artificial Metalloenzyme-Catalyzed Enantioselective Amidation via Nitrene Insertion in Unactivated C(sp3)–H Bonds

Enantioselective C–H amidation offers attractive means to assemble C–N bonds to synthesize high-added value, nitrogen-containing molecules. In recent decades, complementary enzymatic and homogeneous-catalytic strategies for C–H amidation have been reported. Herein, we report on an artificial metalloenzyme (ArM) resulting from anchoring a biotinylated Ir-complex within streptavidin (Sav). The resulting ArM catalyzes the enantioselective amidation of unactivated C(sp3)–H bonds. Chemogenetic optimization of the Ir cofactor and Sav led to significant improvement in both the activity and enantioselectivity. Up to >700 TON and 92% ee for the amidation of unactivated C(sp3)–H bonds was achieved. The single crystal X-ray analysis of the artificial nitrene insertase (ANIase) combined with quantum mechanics-molecular mechanics (QM-MM) calculations sheds light on critical second coordination sphere contacts leading to improved catalytic performance.


INTRODUCTION
Nitrogen-containing motifs are prevalent in natural products, functional materials, and pharmaceuticals. 1 Various strategies for synthesizing nitrogen-containing compounds, such as nucleophilic substitution, condensation, reductive amination, and hydroamination, have been developed in the past decades. 2 These strategies, however, mostly rely on functional group interconversion, often leading to significant waste generation. Methods for constructing C−N bonds enabled by transition metal catalysts were developed to provide direct access to these functionalities. Among these, the Pd-catalyzed Buchwald− Hartwig amination was one of the most representative examples and has been widely applied in academia and industry. 3,4 However, this reaction only applies to substrates containing a (pseudo)halide, thus requiring pre-functionalized substrates.
Developing more direct and atom-economic strategies to construct C−N bonds has attracted increasing attention recently. 3 Pioneering work was reported as early as 1983 by Breslow, relying on Fe(III)-or Rh(II)-catalyzed synthesis of oxathiazolidine using hypervalent ylides as sulfonylnitrene precursors. 5 Capitalizing on this work, many research groups have exploited the potential of metal-nitrene chemistry to create C−N bonds, relying on either homogeneous catalysts or repurposed enzymes. 2,3,6−15 Recently, Chang and co-workers reported an elegant Ir-catalyst for synthesizing γ-lactams, using dioxazolones as nitrene precursors. 16 Since then, Chang, 17 Yu, 18 Chen, 19 and Meggers 20 reported enantioselective variations of this C−H amidation, see Scheme 1a. All four groups obtained excellent enantioselectivity with their catalytic systems. Except for the system reported by Meggers, high catalyst loadings were typically required for the enantioselective amidation of unactivated, purely aliphatic C(sp 3 )−H bonds, affording modest turnover numbers, Scheme 1a. More recently, another highly efficient catalytic system was reported by Chang and co-workers, and up to 47000 TON was obtained for the racemic amidation of C−H bonds. 21 To complement homogeneous catalysis, repurposed enzymes have attracted increasing attention in the context of enantioselective nitrene insertion. 14,22,23 Compared to homogeneous catalysis, enzyme catalysis displays some distinct advantages, including high specificity, mild (aqueous) reaction conditions, high turnover numbers, and compatibility with biological systems. During the preparation of this paper, Fasan and co-workers reported repurposed myoglobin-catalyzed stereoselective construction of β-, γ-, and δ-lactams using dioxazolones as substrates. Utilizing their engineered enzymatic catalysis system, they achieved high enantioselectivity and TON for both benzylic and allylic C−H amidation reactions. 24 With the aim of combining the advantages of both homogeneous and enzymatic catalysis, 25,26 artificial metalloenzymes (ArMs) have emerged as an attractive means to endow organometallic catalysts with an evolvable genotype. ArMs result from the incorporation of a catalytically competent metallocofactor into a genetically encoded protein. Since the first example of ArMs reported by Wilson and Whitesides in the late 1970s, 27 several protein scaffolds have proven versatile for assembling and optimizing of such hybrid catalysts. These include the following: human carbonic anhydrase II, 28 hemoproteins, 29−32 proline oligopeptidase, 33 lactococcal multiresistance regulator, 34 nitrobindin, 35 four-helix bundles, 36 streptavidin, 37−40 etc. 41−51 Regarding the advantages of ArMs, the introduction of metallocofactors endows the host protein with new-to-nature catalytic activity, thus potentially contributing to expand the catalytic repertoire of (natural) enzymes. The presence of a well-defined second coordination sphere provided by the host protein may enable the achievement of high levels of selectivities. Importantly, the ArMs' performance can be improved by combining both chemical optimization (i.e., modification of the cofactor and linker structures) and genetic optimization (directed evolution of the host protein). 52 Based on our experience with Cp*Ir-pianostool cofactors, 41 we set out to evaluate the potential of [Cp*Ir(amidoquinoline)Cl] and [Cp*Ir(aminosulfonamide)Cl]-derived cofactors, to engineer an ANIase based on the biotin−streptavidin technology, Scheme 1b.

RESULTS AND DISCUSSION
Inspired by the seminal publication of Wilson and Whitesides, 27 several groups have capitalized on the biotin-Sav technology to develop ArMs that catalyze a variety of transformations including: hydrogenation, 56 transfer hydrogenation, 5557,58 cross-coupling, 59 olefin metathesis, 60,61 C−H activation, 62,63 hydroxylation, hydroamination, etc. 33,37−41,64−68 The versatility of the biotin-Sav technology for the assembly of ArMs can be traced back to the high affinity of biotin for Sav (K d < 10 −13 M), as well as the remarkable stability of Sav against chaotropic agents, including organic solvents, temperature, pH, etc. With the aim of identifying the most promising biotinylated Cp*Ir cofactor, we tested both 8-amidoquinoline and aminosulfonamide bidentate ligands. In total, eight biotinylated cofactors 3− 10 were tested in the presence of wild-type mature streptavidin (WT Sav). 69 We selected dioxazolone 1 as the model substrate for the chemical optimization of ANIase activity, as shown in Table 1. To our delight, varying amounts of the γ-lactam 2 were detected by chiral GC analysis for all of the cofactors embedded within Sav WT. Notably, the biotin's anchoring position and the bidentate ligand's nature have a marked influence on the ANIase's performance. The best performing ArM [Cp*Ir(Boc-AQ-biot)Cl] 10 · Sav WT included a bulky carbamate moiety ( Table 1, entry 8). In all cases, perfect regioselectivity for the nitrene insertion was observed in favor of γ-lactam 2. However, the enantioselectivity (ee) was modest, varying between 3% and 30%, as shown in Table 1. Importantly, the free cofactor afforded significantly lower TONs than the corresponding ANIase [Cp*Ir(Boc-AQ-biot)Cl] 10 · Sav WT, highlighting the beneficial influence of the protein scaffold (see Table 1, entries 8−9). Having identified a promising ANIase, we set out to further optimize the catalytic performance by genetic means in the presence of [Cp*Ir(Boc-AQ-biot)Cl] 10.
To identify the position of the cofactor upon incorporation in Sav WT, we determined the crystal structure of [Cp*Ir(Boc-AQ-biot)Cl] 10 · Sav WT (PDB: 8BY1), Figure 1. Both Ir (R)and Ir (S) -configurations of [Cp*Ir(Boc-AQ-biot)Cl] 10 were observed, and the position of the metal center was slightly different, depending on its absolute configuration. The closest lying residues include S112 and K121 (Cβ 112 −Ir = 5.6 and 6.7 Å and Cβ 121 −Ir = 7.3 and 8.5 Å and 7.2 and 7.0 Å for the neighboring Sav monomer).
Next, we screened single saturation mutagenesis libraries resulting from randomization at position S112X or K121X′ to afford 38 corresponding single mutant ANIases. The screening results are summarized in Figure 2a−b. The following trends are apparent: i) Residues at position Sav S112X have a more pronounced influence on ANIase's activity and enantioselectivity than residues at position Sav K121X′. ii) Mutations at position Sav K121X′ have a modest influence on ANIase. Both lysine and arginine lead to slightly better performance, in terms of both activity and enantioselectivity, than other single mutants at K121X′. iii) Introduction of a hydrophobic residue at position S112X has the most pronounced positive effect on both activity and selectivity. The best performing single mutant ANIase [Cp*Ir(Boc-AQ-biot)Cl] 10 · Sav S112I affords (S)-2 in TON 171 and 79% ee. The structurally related, but smaller, Sav S112V-or Sav S112L-ANIases lead to a pronounced decrease in ee or TON, respectively Based on this initial screen, we selected [Cp*Ir(Boc-AQbiot)Cl] 10 · Sav S112I and introduced a second mutation at position K121X′. Unfortunately, only 13 double mutant variants were obtained after expression and purification, which were used for the screening, Figure 2c−d. From this screening, it appears that combining the best single-point mutants at both S112 and K121 (i.e., S112I with K121R) leads to significantly improved ANIase performance. Indeed, [Cp*Ir(Boc-AQ-biot)Cl] 10 · Sav S112I-K121R clearly outperforms its parent single mutant (TON = 308, 86% ee compared to TON 171 and 79% ee). Compared to the wild-type ANIase, the [Cp*Ir(Boc-AQbiot)Cl] 10 · Sav S112I-K121R double mutant displays 6-fold higher TONs; see Table S2 for a complete list of results.
To gain insight into the influence of the second sphere on ANIase performance, we determined the structures of ANIases [Cp*Ir(Boc-AQ-biot)Cl] 10 · Sav S112I and [Cp*Ir(Boc-AQbiot)Cl] 10 · Sav S112I-K121R by crystallography. Based on the  The residues S112 and K121 are displayed as purple sticks (nitrogen = blue, oxygen = red, and carbon = purple). The 2F o −F c difference map is displayed as a dark gray mesh (1σ), and the anomalous electron density is displayed as a red mesh (8σ). The occupancy of the Iridium was set to 50 (a) and 30% (b), respectively. ee for reactions using single mutants at positions S112X or K121X′; (c) TON and (d) ee for reactions using double mutants Sav S112I-K121X′. Results for reactions using single mutants Sav S112X, Sav K121X′ and double mutants Sav S112I-K121X′ are highlighted in blue, red, and violet, respectively. Double mutants Sav S112I-K121X′ are abbreviated as IX′ in (c) and (d).

Journal of the American Chemical Society
pubs.acs.org/JACS Article collected data sets, the absolute configurations of [Cp*Ir(Boc-AQ-biot)Cl] 10 · Sav S112I and [Cp*Ir(Boc-AQ-biot)Cl] 10 · Sav S112I-K121R could be determined unambiguously. In both cases, modeling of the cofactor into the residual electron density in the F o −F c map projected the iridium in the position of the anomalous density peak, and the configuration could be identified as Ir (S) -[Cp*Ir(Boc-AQ-biot)Cl] 10, Figure 3. The screening results reveal that the basicity of the residue at position K121 plays an important role in selectivity, next to steric bulkiness. The more basic arginine leads to higher enantioselectivity. Upon decreasing the temperature to 10°C, both the TON and the ee were positively affected: in the presence of [Cp*Ir(Boc-AQ-biot)Cl] 10 · Sav S112I-K121R, dioxazolone 1 was converted to γ-lactam (S)-2, in 363 TON and 89% ee, Table S3, entry 9. Next, we evaluated the substrate scope using a focused library of Sav mutants (including Sav S112I, Sav S112I-K121R, Sav S112V, and Sav S112V-K121R) combined with either [Cp*Ir(Boc-AQ-biot)Cl] 10 or [Cp*Ir(Meoc-AQ-biot)-Cl] 9, Table 2.
Various alkyldioxazolones bearing unactivated C(sp 3   The protein is represented as both a cartoon and transparent surface models. The monomers are color-coded in different shades of blue. The residues S112 and K121 are displayed as sticks (atoms are color-coded; nitrogen = blue, oxygen = red, and carbon = purple). The 2F o −F c difference map is displayed as a dark gray mesh (1σ), and the anomalous electron density is displayed as a red mesh (8σ). The occupancy of the iridium was set to 70%.
(c) Superposition of both crystal structures (PDB: 8AQX and 8BY0). The [Cp*Ir(Boc-AQ-biot)Cl] 10 of Sav S112I is represented as green sticks with the Ir as a dark blue sphere, whereas in the case of Sav S112I-K121R, it is represented as purple sticks with the Ir as a dark blue sphere.
Analysis of the crude of the reaction by 1 H NMR revealed >90% yield for γ-lactam 14. The only side-product identified was the corresponding hydroxamate, resulting from nitrene insertion into water (see SI, p. S25). v) Strikingly, no conversion was observed for either allylic and propargylic C−H insertion ( To study the origin of the good enantioselectivity, transition states for the conversion of substrate 13 in the presence of ANIase [Cp*Ir(Boc-AQ-biot)Cl] 10 · Sav S112I were modeled by QM-MM, including solvation. Computational details are collected in the Supporting Information. The four possible transition states resulting from the two pseudo-enantiomers at Ir for [Cp*Ir(Boc-AQ-biot)Cl] 10 · Sav S112I with the two prochiral C−H bonds of substrate 13 were computed,  Figure 2. We hypothesize that residue Sav K121 (or K121R) forms a H-bond with the carbonyl moiety of the substrate in the lowest energy transition state leading to (S)-14. As this interaction is absent in the lowest energy transition state leading to (R)-14, we surmise that a H-bond between a cationic residue at position 121 and the carbonyl moiety of the substrate significantly contributes to favoring the formation of (S)-14 in high yield. For the lowestlying transition state (e.g., Ir (S) -pro-S) leading to (S)-14, close contacts between the protein and the nitrene-bound substrate include amino acids N49, A86, H87, S88, and I112. No such close contacts between the protein and the nitrene moiety are apparent in the Ir (R) -pro-R transition state, leading to (R)-14. Interestingly, any mutation at Sav G48 markedly reduces the TON, as shown in Table S13. We hypothesize that bulkier amino acids at this position prevent the γ-CH 2 group of the

CONCLUSION
In summary, we have developed a versatile ANIase that is highly efficient for nitrene insertion into unactivated aliphatic primary, secondary, and tertiary C(sp 3 )−H bonds. Strikingly, and despite a lower BDE, allylic and propargylic C−H bonds are not subject to functionalization. By screening a small library of biotinylated Cp*Ir-cofactors in the presence of Sav WT, we identified [Cp*Ir(Boc-AQ-biot)Cl] 10 as the most efficient. Iterative saturation mutagenesis led to the identification of Sav S112I-K121R as a versatile host for the artificial nitrene insertase. Interestingly, the single crystal X-ray structure of [Cp*Ir(Boc-AQ-biot)Cl] 10 · Sav S112I-K121R revealed the preferential incorporation of the Ir (S) -cofactor. QM-MM calculations suggest that this pseudo-enantiomer leads to the most stable transition state to afford enantioenriched (S)-γ-lactam 14.
The ANIase presented herein catalyzes the enantioselective intramolecular nitrene insertion into various C(sp 3 )−H bonds, with BDEs up to 100 kcal/mol. Accordingly, it complements Irbased homogeneous catalysts which perform best with benzylic C−H bonds. Current efforts aiming toward intermolecular C−H amidation will contribute to expand the potential applications of ANIase. We surmise that second coordination sphere interactions between the substrates and the host protein may increase the effective molarity of both substrates to afford enantioenriched high-added value amides.  The cofactor is displayed as color-coded sticks (nitrogen = blue, oxygen = red, and carbon = green) with Ir displayed as dark blue sphere. The nitrene moiety is represented by color-coded sticks (carbon = cyan). Close-lying residues that interact with the transition states are represented as color-coded sticks (for residues interacting with substrate and cofactor, carbon = yellow and magenta respectively). Critical interactions between the host protein and the four diastereotopic transition states are presented in Figure S1.
Journal of the American Chemical Society pubs.acs.org/JACS Article