E.
Klemencic
a,
R. C.
Brewster
a,
H. S.
Ali‡
a,
J. M.
Richardson
b and
A. G.
Jarvis
*a
aEaStCHEM School of Chemistry, University of Edinburgh, Joseph Black Building, David Brewster Road, The King's Buildings, Edinburgh, EH9 3FJ, UK. E-mail: amanda.jarvis@ed.ac.uk
bSchool of Biological Sciences, University of Edinburgh, Swann Building, Edinburgh, EH9 3BF, UK
First published on 29th January 2024
Artificial metalloenzymes (ArMs) have emerged as a promising avenue in the field of biocatalysis, offering new reactivity. However, their design remains challenging due to the limited understanding of their protein dynamics and how the introduced cofactors alter the protein scaffold structure. Here we present the structures and catalytic activity of novel copper ArMs capable of (R)- or (S)-stereoselective control, utilizing a steroid carrier protein (SCP) scaffold. To incorporate 2,2′-bipyridine (Bpy) into SCP, two distinct strategies were employed: either Bpy was introduced as an unnatural amino acid (2,2′-bipyridin-5-yl)alanine (BpyAla) using amber stop codon expression or via bioconjugation of bromomethyl-Bpy to cysteine residues. The resulting ArMs proved to be effective at catalysing an enantioselective Friedel–Crafts reaction with SCP_Q111BpyAla achieving the best selectivity with an enantioselectivity of 72% ee (S). Interestingly, despite using the same protein scaffold, different attachment strategies for Bpy at the same residue (Q111) led to a switch in the enantiopreference of the ArM. X-ray crystal structures of SCP_Q111CBpy and SCP_Q111BpyAla ArMs with bound Cu(II) ions unveiled crucial differences in the orientation of the catalytic centre. Combining structural information, alanine scanning studies, and computational analysis shed light on the distinct active sites of the ArMs, clarifying that these active sites stabilise the nucleophilic substrate on different sides of the electrophile leading to the observed switch in enantioselectivity. This work underscores the importance of integrating structural studies with catalytic screening to unravel the intricacies of ArM behaviour and facilitate their development for targeted applications in biocatalysis.
Site-specific coordination of synthetic metal complexes to proteins plays a crucial role in the development of stereoselective artificial metalloenzymes (ArMs), wherein the protein scaffold forms a secondary coordination sphere around the reactive centre providing stereocontrol.12,13 To synthesise ArMs, researchers have used many different approaches from simply leveraging native metal-binding activity found in certain proteins14 to combining synthetic chemistry and proteins, including using variants of natural cofactors, such as the iron-binding protoporphyrin IX, with alternative transition metals.15,16 More synthetic approaches to site-specific incorporation of metal complexes include the utilisation of supramolecular binding, with several different tethered transition metal complex systems explored to date, the most notable being the biotin–streptavidin (Sav) system.17 Bioconjugation of metal complexes via covalent attachment with unique reactive amino acid residues such as cysteine or azidophenylalanine is another widely used approach to prepare ArMs.18In vivo ligand incorporation offers an attractive alternative where metal-binding unnatural amino acids such as (2,2′-bipyridin-5-yl)alanine (BpyAla) can be selectively introduced to the protein scaffold directly using genetic code expansion technologies, allowing for more streamlined enzyme engineering.19,20
Within these different approaches to ArM design, several scaffolds have been repeatedly used with various attachment strategies for transformation into novel catalysts. These scaffolds have been described as ‘privileged’ scaffolds, in a manner analogous with privileged ligands in homogeneous catalysis. The most prominent examples are the LmrR scaffold, which allows supramolecular, bioconjugation, and incorporation of unnatural amino acids,21 and Sav, which has been explored in dative coordination,22 alongside the more common supramolecular approach.17 Despite these advances, few comparative studies have been conducted which included direct comparisons of the modification strategy for site-selective metal coordination. This makes it difficult to discern the role of protein modification upon ArM reactivity and predict the most effective route to ArM assembly for a desired application.
The Kamer and Jarvis groups have extensively studied the human steroid carrier protein (SCP-2L),23 a single-domain 13.5 kDa protein containing a hydrophobic tunnel, as a scaffold for ArM design.24 It has been exploited in the design of ArMs for selective hydroformylation using Rh–phosphine organometallic complexes,6 which showed high activity and selectivity in the production of long chain linear aldehydes, under aqueous conditions. Moreover, engineering the scaffold for improved thermostability incurred a five-fold increase in TON compared to the wild type.7 This protein has also been used in ArMs for the selective oxidation of lignin model compounds,9 for enantioselective Cu-catalysed Diels–Alder reactions,25 and artificial photoenzyme design.26 Other carrier proteins such as adipocyte lipid binding protein (ALBP),27 ferric hydroxamate uptake protein component A (FhuA),28 and maltose binding protein29 have been used in the design of ArMs suggesting that proteins with a carrier function may also serve as privileged scaffolds. In this work, the potential of SCP-2L to serve as a privileged protein scaffold is explored, through studying the copper-catalysed 1,4-addition of indole to enones.
Bipyridine (Bpy) is a well-studied ligand for transition metals and is utilised in the coordination sphere of many catalytic complexes.30 It can be introduced into a protein via the bioconjugation of bromomethylbipyridines at cysteine residues (Fig. 1). Bpy is also one of very few metal ligands that has been incorporated into the genetic code by amber stop codon suppression. Here two methods for site-selective incorporation of bipyridine into the SCP-2L scaffold were compared, and the copper-catalysed 1,4-addition of indole to enones was used to analyse ArM activity. Differences in stereoselectivities were identified with results rationalised using ArM crystal structures and DFT simulations.
Fig. 1 A) SCP-2L and the residues (orange) used for site-selective ligand attachment (PDB 1IKT). B) Comparison of modification approaches for SCP_Cu ArM design: (upper panel) in vitro bioconjugation of unique Cys residues with 1 before addition of Cu2+ ions in a two-step protocol; (lower panel) one-step in vivo site-selective incorporation of the unnatural amino acid BpyAla 2. |
In comparison to bioconjugation, genetic code expansion using stop codon suppression provides a method of directly incorporating Bpy as the amino acid BpyAla 2 during protein expression (Fig. 1B).32 The synthesis of BpyAla was optimised to allow BpyAla production on scales of 10 grams without the need for column chromatography (see ESIa† for full details).
The SCP-2L gene was codon-optimised for E. coli and prepared by gene synthesis. Amber stop codons were introduced at A100, V83 or Q111 by site-directed mutagenesis. E. coli BL21 (DE3) cells were co-transformed with the pEVOL-BpyAla plasmid, which contains the orthogonal MjTyrRS/MjtRNATyr genes,32 and the pET28 plasmid carrying the SCP gene with a C-terminal TEV cleavable His6 tag. The genes were expressed in the presence of 0.5 mM BpyAla20 and the resulting proteins purified by Ni-affinity chromatography with the His6 tag removed using TEV-protease. Typical yields were 5–15 mg L−1 in LB media, representing a decrease of 20–80% compared to the yields of SCP-2L without BpyAla. Proteins containing unnatural amino acids are known to have lower expression yields, due to incomplete stop codon supression giving truncated protein.19 Mass spectrometry on the purified proteins confirmed the successful incorporation of BpyAla at residue positions 100, 111, or 83 (Fig. 2 and ESIa† Fig. S2).
To obtain the Cu-metalloproteins, one equivalent of Cu(NO3)2 was added directly to the Bpy-containing SCP proteins. Copper binding was confirmed by UV-vis spectroscopy which clearly showed a red shift in the π–π* transition of the Bpy ligand in the presence of Cu(II), consistent with previous reports on copper binding to Bpy-containing proteins.20 Titration experiments confirmed that, at the concentrations of interest (20–100 μM), Cu(II) bound to the Bpy-containing SCP proteins with ∼1:1 stoichiometry (see ESIa,† Fig. S5 to S10). Copper binding was also confirmed by ICP-MS analysis (see ESIa†).
The structure of Cu(II)-bound SCP_Q111CBpy was determined to 1.52 Å resolution, and clearly confirmed the incorporation of 2,2′-bipyridine at C111, which is positioned at the centre of the C-terminal helix, α5. The bipyridine rings are sandwiched in a parallel fashion between the amide side-chain atoms of Q108 (one turn up on α-helix 5) and Q90 on α-helix 4 (Fig. 3C). The outer and inner pyridine rings are 3.4 Å from Q108 and Q90 respectively, a distance that favours formation of stabilising aromatic π–amino electrostatic interactions. α-Helix 5 is slightly shifted relative to its position in the wt SCP_2L structure (PDB ID 1IKT) (Fig. S14 in ESIa†) and is more flexible than the remainder of the structure, with B-factors ranging from 52–96 Å2 from the N- to the C-terminus of the helix (N103-L120), compared to 75 Å2 for the protein overall. The B-factor of the Bpy moiety is 92 Å2, suggesting flexibility in its position. Additional electron density was observed in the 2Fo–Fc map around the solvent-exposed face of the Bpy moiety, and a single peak in an anomalous difference map confirmed this is a bound Cu(II) ion (Fig. 3C). The Cu(II) atom lies in the same plane as the two pyridine rings and is 2 Å from the two pyridine ring nitrogens, and thus we assume that the Cu(II) ion adopts an octahedral coordination geometry. The high B-factor (108 Å) of the Cu(II) is consistent with its position in the flexible α-helix 5. Only one water molecule could be modelled around the Cu(II) ion, but the positions of Q108 and Q90 suggest that their amide side-chain atoms may play a role in coordinating the other (unmodelled) waters in the coordination sphere.
The structure of Cu(II)-bound SCP_Q111BpyAla was determined to 2.51 Å resolution (Fig. 3B and D) and revealed a different orientation of the bipyridine moiety compared to that seen in the Cu(II)-bound SCP_Q111CBpy structure. Despite the lower resolution, the electron density for the BpyAla bipyridine rings, α-helix 5 and the additional C-terminal amino acids (120–128) was very clear, indicative of a more rigid conformation for these parts of the protein. The co-planar BpyAla bipyridine rings are wedged into a shallow pocket formed by the turn between G85 and P89 at the edge of the apolar tunnel. This conformation is stabilised by van der Waals interactions between C1 of BpyAla and both G85 Hα (3.35 Å) and D88 Hα (3.36 Å), C2 of BpyAla and Hδ of P89 (2.87 Å: 3.36 Å to the Cδ) and C3 of BpyAla and Hγ of P89 (2.74 Å: 3.26 Å to the Cγ) (Fig. 3D). In addition, the side-chains of D88 and K115 (one turn down on α-helix 5) lie on either face of the bipyridine and may stabilise the BpyAla conformation by aromatic–π electrostatic interactions.
The 2Fo–Fc electron density map showed a single intense peak adjacent to the BpyAla bipyridine corresponding to a Cu(II) ion coordinated to N1 and N2 of BpyAla (Cu–N bond distance of 2 Å) and four water molecules in an octahedral arrangement (Fig. S15 in ESIa†). The B-factors for the Bpy moiety and the Cu(II) ion (43 Å2 and 57.91 Å2, respectively) were lower in this structure compared to the bipyridine rings in the Cu(II)-bound Q111CBipy structure, indicating that the catalytic environment is more rigid.
Taken together the structural analyses confirm the incorporation of Bpy at position 111 and the binding of Cu(II) to the Bpy moieties in both SCP ArMs. They also revealed marked differences in the environments of the Bpy and bound Cu(II) ions in the SCP_Q111BpyAla and SCP_Q111CBpy structures. Moreover, the Bpy moiety at position 111 is more rigid when incorporated as the unnatural amino acid BpyAla than via bioconjugation at a cysteine.
Entry | Ligand/protein | Yield of 5a (%) | e.r.b (R:S) |
---|---|---|---|
Conditions: The reaction was carried out using 9 mol% of Cu(NO3)2 (90 μM), a small excess of 1.25 equivalents of SCP (112.5 μM) to ensure that all Cu(II) ions are bound, 1 mM of 3 and 2.5 mM of 4 in MES buffer (20 mM, 150 mM NaCl, pH = 6.0) for 3 days at 4 °C giving 5. All data are averages of experiments done in triplicate. Standard deviations for each enantiomer is shown in brackets.a Yield obtained by HPLC using 2-phenylquinoline as internal standard.b e.r. determined using chiral HPLC. n.d. not determined.c Cu(NO3)2 72 μM.d Cu(NO3)2 63 μM.e rt reaction. S and R were assigned using chiral HPLC.34 | |||
1 | None | 76 (±6) | 50:50 (±0) |
2 | BpyAla(rac) | 50 (±2) | 50:50 (±0) |
3 | wt SCP_2L | 43 (±5) | 50:50 (±0) |
4c | SCP_A100CBpy | 42 (±1) | 66:34 (±4) |
5c | SCP_V83CBpy | 28 (±1) | 52:48 (±2) |
6d | SCP_Q111CBpy | 25 (±2) | 64:36 (±4) |
7 | SCP_A100BpyAla | 45 (±3) | 51:49 (±0) |
8 | SCP_V83BpyAla | 45 (±8) | 63:37 (±1) |
9 | SCP_Q111BpyAla | 42 (±7) | 20:80 (±1) |
10 | SCP_Q111CBpy + Triton | 30 (±8) | 57:43 (±0) |
11 | SCP_Q111BpyAla + Triton | 32 (±3) | 23:77 (±3) |
12 | SCP_A100BpyAla + Triton | 21 (±3) | 52:48 (±1) |
13 | SCP_V83BpyAla + Triton | 16 (±5) | 57:43 (±3) |
14e | SCP_A100CBpy without Cu | 6.1 | n.d. |
15e | SCP_A100BpyAla without Cu | 1.9 | n.d. |
SCP ArMs with BpyAla showed a higher yield (42–45%) compared to the equivalent Cys-coupled Bpy SCP ArMs. Among the novel ArMs, SCP_Q111BpyAla displayed the greatest enantioselectivity with an e.r. of 20:80 (±1), favouring the S enantiomer, and a yield of 42% (Table 1, entry 9). The BpyAla moiety in this SCP ArM is situated at the centre of one side of the tunnel, in the middle of α-helix 5 (Fig. 1A and 3A). By contrast, the ArMs with BpyAla situated at either end of the tunnel showed either no enantioselectivity in the case of SCP_A100BpyAla (Table 1, entry 7), or lower enantioselectivity with a preference for the R enantiomer (Table 1, entry 8). We hypothesise that the lower selectivity at the entries to the tunnel could be due to increased flexibility and space leading to less defined binding pocket, compared to position Q111 in the centre of the tunnel. While enantioselectivity towards the R enantiomer was observed with all three Cys mutants, it was limited with the best result observed for SCP_A100CBpy which gave a 66:34 ratio (Table 1, entries 4–6).
Both enantiomers of 5 can be obtained in the Friedel–Crafts alkylation using SCP-ArMs with Bpy attached to residue 111, but by using different attachment strategies (Table 1, entries 6 and 9). The crystal structures of SCP_Q111BpyAla and SCP_Q111CBpy provide a structural indication for the observed differences in the enantioselectivities of these ArMs [almost 2× higher e.e., from 64:36 (R) for SCP_Q111CBpy to 20:80 (S) for SCP_Q111BpyAla]. The longer linker to the protein backbone in SCP_Q111CBpy, which has an additional –CH2–S– compared to SCP_Q111BpyAla, allows it to extend out from α-helix 5 and adopt a more exposed position on the surface of the protein, compared to the BpyAla which is closer to the scaffold protein. In addition, the longer linker confers greater flexibility (as reflected in the B-factors), explaining the lower enantiomeric excess. Alignment of the Cu(II)-bound SCP_Q111BpyAla and SCP_Q111CBpy structures (Fig. 3A and B) highlights the different positions and opposite orientations of the Bpy moieties, despite being incorporated at the same amino acid.
The first step of the reaction is the conjugate addition of indole 3 to enone 4 to form intermediate (Int1), via a transition state (TS1) (Scheme S1 ESIb†). The second step is the product formation by the protonation reaction via a second transition state (TS2). To investigate the enantioselectivity of the Friedel–Craft alkylation, we looked to prior work using cluster models for metalloenzyme-catalyzed reactions.35,36 The same QM-cluster model technique along with density functional theory (DFT) methodology was utilised to study our reaction. Three active site models were created for analysis (Fig. S1 ESIb†). Model A represents a reaction catalysed by Cu(II)-2,2′-bipyridine, in the absence of protein, whereas models B and C represent reactions catalysed by Cu(II)-bound SCP_Q111BpyAla and SCP_Q111CBpy, respectively. The computational analysis on Model A (Fig. S11 ESIb†) showed that the conjugate coupling step is both the rate-determining step as well as the enantioselective step, and the formation of the keto product is most likely. This agrees well with the experimental results (Table 2) as well as previous studies,37 therefore only the first step was modelled going forwards. To create models B and C, the indole 3 and enone 4 substrates, along with Cu(II) ions, were incorporated into the crystal structures of Cu(II)-bound SCP_Q111BpyAla and SCP_Q111CBpy, respectively. The PDB files of the apo SCP_ArMs were prepared as described in the ESIb,† and 3 and 4 were docked using AutoDock Vina,37 with the lowest energy docked poses used for molecular dynamics (MD) simulations (see ESIb† for detailed methods and results).
Entry | Mutant of SCP_Q111BpyAla | Yield of 5a (%) | e.r.b (R:S) |
---|---|---|---|
Conditions: reactions were carried out using 9 mol% of Cu(NO3)2 (90 μM), a small excess of 1.25 equivalents of SCP (112.5 μM) to ensure that all Cu(II) ions are bound, 1 mM of 3 and 2.5 mM of 4 in MES buffer (20 mM, 150 mM NaCl, pH = 5.0) for 3 days at 4 °C giving 5. All data are averages of experiments done in triplicate. Standard deviation for each enantiomer is shown in brackets.a Yield obtained by HPLC using 2-phenylquinoline as internal standard; standard deviations are shown in brackets.b e.r. determined using chiral HPLC. | |||
1 | None | 43 (±2) | 14:86 (±1) |
2 | Q108A | 38 (±7) | 28:72 (±0) |
3 | V82A | 33 (±2) | 24:76 (±1) |
4 | F34A | 21 (±1) | 31:69 (±0) |
5 | M112A | 33 (±4) | 19:81 (±0) |
6 | K115A | 41 (±5) | 27:73 (±1) |
7 | D88A | 54 (±10) | 18:82 (±1) |
For SCP_Q111BpyAla catalysed reactions (Model B), the reactant complexes in two conformations: proS and proR were selected and their geometries optimized to give ReproS,B and ReproR,B as shown in Fig. 4A and B. The ReproS,B structure is lower in energy by 5.4 kcal mol−1 (ΔE + ZPE) than ReproR,B and therefore ReproS,B represents the more favourable substrate binding orientation.
Next, the Friedel–Craft (FC) alkylation reaction catalysed by SCP_Q111BpyAla, using ReproS,B and ReproR,B reactants as the starting structures, was calculated. Firstly, the conjugate addition of indole 3 to enone 4, where substrate 3 configuration is either proS or proR was tested. The C–C coupling transition states (TS1proS,B and TS1proR,B) of 3 with 4 leads to an intermediate for each configuration: Int1proS,B and Int1proR,B respectively. The calculated potential energy landscape for the conjugate coupling reaction for Model B is represented, along with the optimized geometric structures of these transition states, in Fig. 5A. The energy barrier for TS1proS,B (10.1 kcal mol−1) is lower than TS1proR,B (15.5 kcal mol−1) (values relative to the stable configuration ReProS,B). The transition states were characterized by an imaginary frequency i308 cm−1 and i319 cm−1 for TS1proS,B and TS1proR,B respectively. Furthermore, in transition state TS1proS,B, indole 3 has strong water bridge interactions with D88 that make it more stable than the TS1proR,B, which has a week hydrogen bonding interaction with M112. Subsequently, the transition states are relaxed to their respective intermediates (Int1proS,B and Int1proR,B), which are characterized by lower relative energy values: 1.65 and 7.92 kcal mol−1, respectively. Thus, the proS configuration pathway is energetically favoured over the proR configuration pathway. Taken together these analyses explain the preferred enantioselectivity of SCP_Q111BpyAla (e.r. 14:86, R:S) that we observed experimentally (Table 2).
In a similar manner, the FC alkylation reaction mechanism catalysed by SCP_Q111CBpy (Model C) was explored. The reactant complexes in both proS and proR configurations were selected as the starting structures. The optimized geometries of these reactant complexes ReproS,C and ReproR,C are represented in Fig. 4C and D. The free energy of the ReproS,C structure is higher than the ReproR,C by 5.6 kcal mol−1 (ΔE + ZPE), indicating that the reactant complex of model C in proR configuration shows the strongest substrate-bound pose and hence signifies the favoured reactant orientation.
Finally, the FC reaction pathway of Cu(II)-bound SCP_Q111CBpy was explored, using ReproS,C and ReproR,C reactants as the starting structures. The calculated potential energy landscape for the conjugate coupling reaction for model C is represented, along with the optimized geometric structures of these transition states, in Fig. 5B. The lowest energy barrier (13.6 kcal mol−1) was obtained for the proR C–C coupling transition state (TS1proR,C) while the energy barrier for the proS transition state (TS1proS,C) was higher (15.4 kcal mol−1). The transition states are characterized by the presence of an imaginary frequencies i288 cm−1 and i273 cm−1 for TS1proR,C and TS1proS,C respectively. Furthermore, in the TS1proR,C transition state a hydrophobic interaction from M105 was observed with 3, while no such interaction of 3 was observed in TS1proS,C. The system then relaxed to the low energy state intermediates Int1proR,C and Int1proS,C, which are characterized in the local minima state by the presence of all real frequency values indicating stable structures. The Int1proR,C has lower energy (6.7 kcal mol−1) than the Int1proS,C configuration (9.8 kcal mol−1), which favours the proR configuration pathway over the proS pathway. This is consistent with the experimentally observed preference to produce the R enantiomer by SCP_Q111CBpy (Table 1).
Overall, the substrates' activation in FC alkylation mechanism with SCP_Q111BpyAla favours the formation of proS product over proR while with SCP_Q111CBpy favours proR product formation over proS. However, the products distribution was achieved via a competitive pathway, so a mixture of both enantiomers of the product is predicted, which agrees well with our experimental results. The larger difference in the relative energies for the SCP_Q111BpyAla TS intermediates matches the experimental observation of improved enantioselectivity when using SCP_Q111BpyAla as the catalyst over the use of SCP_Q111CBpy.
None of the mutations screened showed substantial variations (i.e. a complete drop in reactivity or selectivity). Indeed, the mutation with the biggest difference was F34A which was chosen due to its location nearby and within the protein hydrophobic pocket (Table 2, entry 4). SCP_Q111BpyAla F34A exhibited substantial precipitation suggesting that disruption to the protein core reduced stability, leading to the observed low activity and selectivity. A similar rationale was also used for the choice of V82A which gave lower activity and selectivity but not to the same extent (Table 2, entry 3 and 4). The crystal structure suggested that K115 and D88 may play a role in stabilising the bipyridine position within the protein through aromatic–π electrostatic interactions. Mutating D88 to alanine gave no meaningful change in enantioselectivity (Table 2, entry 7), whilst K115A showed a reduction in enantioselectivity (Table 2, entry 6). The more flexible nature of the lysine side chain means it is less clear if this can be attributed to structural changes as opposed to some role with substrate binding. The computational work revealed no interaction of the substrates with K115. In contrast, a close contact between the side chain of D88 and the copper atom was observed in the modelled transition states and D88 was shown to participate in water bonding networks with the substrates. Both these interactions would be disrupted on mutation of D88 to alanine, suggesting that either these interactions are of minor importance in catalysis or alternative residues such as Q108 could replace the hydrogen bonding interactions with water. Whilst Q108A was shown to reduce the enantioselectivity obtained with SCP-Q111BpyAla experimentally (Table 2, entry 2), no clear role was observed in the computational work.
The only amino acid shown to make direct contact with the substrate during the course of the reaction was M112: the methyl makes a weak hydrophobic interaction with 3 during substrate binding in the proS pathway (Fig. 6), whilst the amide backbone of M112 makes a weak hydrogen bond with 3 in the proR transition state (Fig. 5A). Mutating M112 to alanine gave a small reduction in enantioselectivity (Table 2, entry 5). Whilst alanine can also make hydrophobic interactions, its reduced chain length would preclude interactions with 3 in this instance and thus M112 could be helping to stabilise the substrates to a small extent. No difference would be expected for the proR pathway as alanine can still participate in amide backbone bonding.
The lack of side chain interactions with the substrates and no clear results from the alanine scanning suggests that a rational approach to designing the active site via mutagenesis may not lead to an improved enzyme. Indeed, the exposed nature of the active site on the side of the protein suggests that extensive backbone engineering to build up the protein bulk around the substrates may lead to more promising candidates for enantioselective catalysis.
The crystal structure obtained of SCP-Q111BpyAla_Cu is a rare example of a protein structure containing an unnatural amino acid. Our work shows that increasing our understanding of artificial metalloenzyme structures facilitates their development as catalysts, and moving forwards will be vital to enable future de novo design of metalloproteins.
Footnotes |
† Electronic supplementary information (ESI) available: ESIa: additional experimental details, materials, methods and data. ESIb: additional information on the computational studies. The X-ray crystal files of Cu(II)-bound SCP_Q111CBpy and SCP_Q111BpyAla have been deposited in the pdb: 8AF3 and 8AF2. See DOI: https://doi.org/10.1039/d3cy01648j |
‡ Current address: Chemistry Research Laboratory, Department of Chemistry and the INEOS Oxford Institute for Antimicrobial Research, University of Oxford, 12 Mansfield Road, Oxford OX1 3TA, UK. |
This journal is © The Royal Society of Chemistry 2024 |