Abstract
A designer enzyme consisting of an abiological molecule incorporated into a natural protein has been developed as an exceptionally chemoselective catalyst, highlighting that the internal space of proteins is highly beneficial for enhancing catalytic performance. However, other features of proteins have received less attention in designer enzymes, for e.g., their use as ligands to construct abiological (multinuclear) metal centers and their intrinsic functions that have often been traded off for a new function. Here, grafting a synthetic trinuclear zinc complex inside a human cytokine macrophage migration inhibitory factor (MIF) scaffold using solely amino-acid side chains leads to a designer multi-metalloenzyme with extrinsic and intrinsic functions. The crystal structure of the designer tri-zinc enzyme verifies the accuracy of our design process based on geometry optimizations and quantum-chemical calculations. The extrinsic catalytic performance of this designer enzyme is of the highest class and comparable to that of previously reported designer zinc hydrolases. Importantly, an intrinsic function of MIF, i.e., its tautomerase activity, is maintained in this designer tri-zinc enzyme. Considering that cytokines are originally expressed in response to in vivo events, this cytokine-based designer metalloenzyme holds promising potential as a synthetic biological tool for the self-adaptive regulation of life phenomena.
Similar content being viewed by others
Introduction
In nature, a wide repertoire of chemical transformations are facilitated by enzymes. The use of metal ions in proteins further expands the reaction repertoire. For example, soluble methane monooxygenase, a diiron enzyme, facilitates methane hydroxylation, whilst nitrogenase, which possesses a cofactor consisting of eight iron atoms and a molybdenum atom, catalyzes nitrogen fixation. Furthermore, the oxygen-evolving complex (OEC) of photosystem II, which has a multinuclear metal cluster comprised of four manganese atoms and a calcium atom, oxidizes water to produce oxygen.
In such metalloenzymes, the protein scaffold plays two roles. Firstly, the amino-acid side chains act as ligands for fine control of the reactivity of the metal ions. Secondly, a well-defined internal space is created around the metal center to form a reaction compartment that enhances the reaction rate and chemoselectivity.
While synthetic model complexes have been developed with the aim of understanding such natural (multi)metalloenzymes1,2, reproducing the well-defined enzymatic internal space responsible for high reactivity and selectivity remains challenging3,4. The significance of a protein as a reaction compartment has been elucidated in the studies on designer (metallo)enzymes that are constructed by incorporating catalytically active synthetic molecules, including metal complexes, into a protein. Thanks to the reaction compartment provided by the protein scaffold, enhanced reaction rates and high chemoselectivity have been achieved, while the incorporated synthetic molecule dictates the reaction type5,6,7,8,9,10,11,12.
In contrast, the use of proteins as ligands currently lags behind their use as reaction compartments. This can be attributed to the difficulties associated with designing coordination-chemistry interactions between metal ions and amino-acid side chains at the atomic level. Hence, the development of designer mononuclear metalloenzymes has primarily relied on a more classical metal-substitution approach13,14,15,16, meaning that the design of de novo metal-binding sites is even more challenging17,18,19,20,21,22. Although some excellent studies on the construction/mutualization of multinuclear metal centers using proteins and peptides as ligands have been reported23,24,25,26,27,28,29,30,31,32, there are only a small number of examples of multinuclear metal centers that exhibit catalytic activity and the variety of the successfully used peptide/protein scaffolds is limited.
In addition to the exploitation of proteins as a ligand, the intrinsic functions of proteins have been afforded less consideration when constructing designer metalloenzymes. The primary consideration has usually been whether the protein possesses an internal space with appropriate size and shape, which is often responsible for its intrinsic function. Therefore, in hitherto reported designer metalloenzymes, the intrinsic function of the natural protein has often been traded off for a new function. On the other hand, nature has evolved proteins that possess multiple and independent functions, which are referred to as moonlighting proteins33. These moonlighting proteins exhibit different functions depending on their cellular localization, such as in the nucleus, cytoplasm, or extracellular space. The prevalence of these proteins among highly evolutionarily conserved enzymes suggests that they have evolved secondary functions over time. In other words, the acquisition of secondary functions by moonlighting proteins likely occurred not through a trade-off mechanism with primary functions, but rather through additive gains while maintaining their primary functions. The continuing utilization of the protein solely as a structural framework whilst overlooking the intrinsic function of the protein may reduce the rationale for the use of natural proteins in designer metalloenzymes, as compared to the rapidly emerging de novo proteins34.
Nevertheless, prospective advantages of using natural proteins in designer metalloenzymes can be foreseen in their in celluo/vivo applications, especially where the designer enzyme orthogonally regulates life processes35,36. By enabling the installation of a new function into a protein while keeping its original function, similar to naturally evolved moonlighting proteins, designer metalloenzymes can be created that adaptatively exhibit an extrinsic functionality in relation to intrinsic biological processes, leading to promising tools for synthetic biology.
In this study, we aim to address two major challenges in the development of designer metalloenzymes: (i) the rational construction of a multinuclear metal center using a protein solely as a ligand, and (ii) preserving the intrinsic function of the scaffold protein.
Results and discussion
Building blocks and construction strategy for designer multi-metalloenzymes
We focused on zinc as the metal center in this study because it is one of the most available metal ions in living systems37. Besides being a fulcrum point for the tertiary structure of the protein scaffold, the zinc ion functions as a catalytically active site in natural enzymes.
Mononuclear zinc active sites represent one of the most advanced developments in designer metalloenzymes18,19,21,38,39,40,41. Given that the construction of designer enzymes that bear mononuclear zinc sites in E. coli has been successfully demonstrated17, designing non-natural multinuclear structures relying on zinc ions may be advantageous for future in celluo/vivo applications.
Some natural enzymes have a multinuclear zinc center, where the cooperative action of multiple zinc ions was found to be important for their catalytic activity42. However, compared to mononuclear zinc centers in designer enzymes, there are limited examples of functional multinuclear zinc centers, and in particular there are no examples with more than three centers.
Therefore, we aimed to graft a synthetic multinuclear zinc center into a protein scaffold, and thus we selected a catalytically active synthetic trinuclear zinc complex (Fig. 1a, b)43. In this complex three zinc ions are bridged by CO32–, a structural motif that is not found in natural enzymes.
a Schematic and b crystal structures of the synthetic trinuclear zinc complex (CCDC 931956)43, and c crystal structure of human MIF (PDB code: 1MIF), which is also known as a moonlighting protein, displaying the active sites for its multiple functionalities. The structural illustrations in this report were generated using CueMol2 (http://www.cuemol.org/en/). In (b), zinc, carbon, oxygen, and nitrogen atoms are shown in dark purple, cyan, red, and blue, respectively. c depicts oxygen in red and nitrogen in blue, while the carbon atoms are color-coded according to their functionality. Residues with carbon atoms in orange and yellow correspond to the CD74 binding site. Residues with the carbon atoms in cyan indicate the nuclease-active site, whereas those with carbon atoms in pink represent the tautomerase-active site.
As a protein scaffold, we selected macrophage migration inhibitory factor (MIF), a cytokine involved in various physiological processes in vivo (Fig. 1c). We reasoned that the trimeric structure of MIF, specifically its internal pore, could be suitable for the transplantation of the trinuclear synthetic zinc complex. Moreover, the rigid secondary structure, comprised of β-sheets in the subunits that form the pore can be expected to facilitate the grafting of metal centers. Furthermore, MIF is an attractive scaffold to use here due to its multiple intrinsic functions as explained below44,45,46.
MIF is encoded by the human mif gene and functions as an important regulator of innate immunity44. Upon stimulation with bacterial antigens or glucocorticoids, MIF is expressed and released from leukocytes. At the putative site shown in Fig. 1c, MIF binds to the CD74 receptor on other immune cells to activate a signaling pathway, which leads to gene transcription and the production of pro-inflammatory cytokines and chemokines for cell proliferation47.
As shown in Fig. 1c, MIF is considered a moonlighting protein due to its multiple functions beyond its role as a cytokine. It exhibits phenylpyruvate tautomerase activity, whose physiological function remains enigmatic45. Furthermore, recent studies have revealed that MIF possesses nuclease activity. Upon association with the apoptosis-inducing factor, MIF degrades genomic DNA into large fragments within the nucleus, resulting in chromatinolysis. During this process, MIF is thought to primarily bind to 5′ unpaired bases in single-stranded DNA (ssDNA) with stem-loop structures46. The crucial sites for these functions are mainly located on the surface of MIF (Fig. 1c), suggesting that constructing a multinuclear zinc center within the pore is unlikely to compromise these intrinsic capabilities45,47.
The binding sites for the zinc ions on the monomeric MIF were designed via the procedure outlined below. Each zinc ion of the synthetic trinuclear zinc complex is coordinated by three nitrogen atoms. Therefore, we selected a histidine residue as the sole ligand.
To analyze the optimal Cα atom positions for placing histidine residues in the pore of MIF, we conducted a systematic computational search. As shown in Fig. 2a, the z-axis was defined along the vertical axis of the pore. The three extracted zinc ions from the crystal structure of the synthetic complex were rotated in the x-y plane in 2° increments and translated along the z-axis in 0.5 Å increments whilst the distances, L1, L2, and L3, between the Zn ions and the Cα atoms were cataloged. This comprehensive approach generated a total of 1281 different geometric configurations for evaluation. In this search, we confined the analysis to the upper region of the pore based on the preliminary hole-size analysis by MD simulations (Supplementary Fig. 1). Given that the entrance at the bottom of the pore is relatively larger than that at the top, we reasoned that the bottom entrance is more suitable for substrate uptake.
a Schematic illustration of the computational search for the optimal positions of the Cα atoms to be substituted by histidine residues to coordinate the trinuclear zinc center. Spheres in deep purple represent Zn while those in orange or white represent Cα. The Cα sites in orange were selected for screening. Selected combinations of Cα and tri-zinc centers identified via a geometric search (Cα atoms and zinc shown as spheres). b Corresponding DFT-optimized structures with the selected histidine residues (zinc: purple; nitrogen: blue; oxygen: red). Structures represent S61H1-H631-Y100H1 (carbon: light red), S61H2-H631-Y100H1 (carbon: orange), S61H2-H631-Y100H2 (carbon: yellow), S61H1-H631-N98H1 (carbon: light green), S61H2-H631-N98H2 (carbon: green), H631-N98H1-Y100H1 (carbon: cyan), H631-N98H2-Y100H1 (carbon: blue), and H631-N98H2-Y100H2 (carbon: pink). The subscript numbers denote the monomer of MIF on which these residues are located. Estimated energy differences are displayed under the DFT-optimized structures. The colors of the bars in the energy diagram correspond to the carbon colors defined in the DFT-optimized structures described above.
The distances between the zinc ions and the Cα atoms of the coordinating histidine residues of natural zinc-binding proteins (PDB codes: 1AI0, 2CBA, 1IAG, 1TON, 1PQ4), which have three histidine residues as the sole ligands, often fall within the range of 6.0–6.7 Å. In case that one of the calculated distances (L1, L2, or L3) fell outside of this range, the corresponding set of trinuclear zinc and Cα distances was excluded from the list. Thus, only triplets of Cα where all three Zn-Cα distances were within this biomimetic range were considered for selection. Candidate triplets for mutation were further filtered based on their geometric consistency, specifically requiring that the maximum difference among the pairwise distances (|L1–L2|, |L2–L3|, and |L1–L3|) was constrained to be <0.2 Å. Among the selected triplets, the following eight unique sets of Cα combinations appeared repeatedly: S611-H631-Y1001, S612-H631-Y1001, S612-H631-Y1002, S611-H631-N981, S612-H631-N982, H631-N981-Y1001, H631-N982-Y1001, and H631-N982-Y1002, where the subscript numbers denote the monomer on which these residues are located (Fig. 2a and Supplementary Table 1). Notably, His63 was consistently included in all of the selected triplets, despite not being explicitly required.
To further evaluate which combinations of Cα atoms are optimal for substitution with a histidine residue, we performed density-functional-theory (DFT) calculations for all eight triplets. According to the root mean square fluctuation (RMSF) analysis of wtMIF based on MD simulations prior to the DFT calculations, the residues selected through geometrical screening were found to be remarkably rigid (Supplementary Fig. 2). This observation suggests that consideration of backbone flexibility is not essential for subsequent DFT calculations aimed at identifying geometrically permissible Cα atom combinations to form a trinuclear site. Consequently, optimized models were calculated using three zinc ions, the CO₃²⁻ bridge, and nine histidine side chains without including the entire protein structure in order to keep computational expenses low. The coordinates of the main-chain atoms (N, Cα, and C) belonging to the β-sheet framework were extracted from the PDB (PDB code: 1MIF) and fixed, while the rest of the structure remained unconstrained. These model systems contained approximately 100 atoms in total.
As shown in Fig. 2b, energy differences obtained from the DFT-optimized structures suggested S612-H631-Y1002, with two histidine residues introduced at the positions S61 and Y100 on the same monomer, and the original H63 on the other monomer, would yield the thermodynamically most stable trinuclear zinc center (Fig. 2b and Supplementary Table 2). The energy advantage of approximately 54.6 kcal/mol compared to the next most stable structure indicates this combination offers the optimal zinc-ion/histidine-residue interaction among the eight tested candidates.
Validation and optimization of the design via crystal-structure analysis
Henceforth, the designer zinc enzyme will be denoted as Znx-MIF, where x represents the number of zinc ions in the trimeric MIF. To validate our design strategy, we aimed to confirm the formation of the trinuclear zinc center via X-ray crystal-structure analysis.
The rationally designed MIF(S61H/Y100H) variant was expressed in E. coli BL21(DE3) and purified by ammonium-sulfate fractionation and anion-exchange column chromatography (Supplementary Figs. 3, 4, and Supplementary Table 3). The purified sample was crystallized using previously reported conditions48, except for the addition of ZnCl2. The X-ray crystal-structural analysis revealed that three zinc ions were bound in the pore of MIF (Supplementary Fig. 5a, d, g, and Supplementary Table 4). Furthermore, the presence of a bridging CO32– ion was also confirmed, indicating that Zn3-MIF(S61H/Y100H) had captured CO32– ions dissolved in the solution under atmospheric conditions, given that CO32– ions had not been used during the crystallization procedure. H61 of MIF(S61H/Y100H) is located slightly too far away from the zinc ion in order to coordinate with it when low-zinc-concentration conditions (10.5 eq per MIF monomer) were used for crystallization (Fig. 3c and Supplementary Fig. 5b, e, h). On the other hand, H63 and H100 coordinate with the zinc ion. While H61 and H63 maintain their original configurations with apo-MIF(S61H-Y100H), H100 flips to coordinate to the zinc ion.
Top view of the crystal structures (a, d) and close-up side view (b, e) of Zn3-MIF(S61H/Y100H) under high zinc concentration (PDB code: 9JIZ; carbon: cyan) and Zn3-MIF(Y100H) (PDB code: 9JJ0; carbon: cyan). Superimposed images of the crystal structures: c MIF(S61H/Y100H) (PDB code: 9JIT; carbon: pink), Zn3-MIF(S61H/Y100H) under low zinc concentration (PDB code: 9JIY; carbon: orange), and Zn3-MIF(S61H/Y100H) under high zinc concentration; f MIF(Y100H) (PDB code: 9JIV; carbon: pink), and Zn3-MIF(Y100H) under low zinc concentration. Superimposed images of the DFT-optimized trinuclear zinc center, S61H2-H631-Y100H2, (carbon: yellow) with g the synthetic zinc complex, h Zn3-MIF(S61H/Y100H) under high zinc concentration, and i Zn3-MIF(Y100H) under low zinc concentration (carbon: cyan). Carbon atoms are colored as described above. Zinc atoms are displayed in orange and deep purple for Zn3-MIF(S61H/Y100H) under low zinc concentration and Zn3-MIF(S61H/Y100H) under high zinc concentration, respectively. Other atoms are colored as follows: nitrogen (blue), oxygen (red), and chloride (green). Spheres represent Cα atoms, and the matching colors correspond to those of the same monomer.
Upon increasing the zinc concentration (26.0 eq to MIF), H61 flips to coordinate to the zinc ion (Fig. 3a–c and Supplementary Fig. 5b). These results indicate that H63 and Y100H are key residues for the coordination of zinc, and moreover, the presence of at least two histidine ligands would allow the formation of a trinuclear zinc complex. The three zinc ions, the CO32– ion, as well as the H61, H63, and H100 residues in the crystal structure could be superimposed on the calculated structure with an RMSD of 0.709 Å, which validates our design strategy (Fig. 3g, h).
Given the flexibility of the coordination modes of the zinc ion and the aforementioned observation that H61 is not necessarily essential, we prepared a variant with a single substitution, MIF(Y100H). Here, we assumed the presence of an abundant anion species such as OH–, which can act as an active species for the hydrolysis reaction, and Cl– to bind to the zinc center as an external ligand instead of the internal H61 ligand to compensate for the charge balance of the trinuclear zinc center.
Even when the low-zinc-concentration conditions were used, the corresponding crystal-structure analysis confirmed the presence of a trinuclear zinc center in Zn3-MIF(Y100H) (Fig. 3d–f and Supplementary Fig. 5c, f, i). The configuration of H100 of Zn3-MIF(Y100H) is identical to that of Zn3-MIF(S61H/Y100H) under high-zinc-concentration conditions. As can be expected, the Cl– ions bind to the zinc ion in Zn3-MIF(Y100H). The trinuclear zinc center of Zn3-MIF(Y100H) aligned well with the calculated structure with an RMSD of 0.900 Å (Fig. 3i).
Based on the fact that a trinuclear zinc center was confirmed to be present in MIF(Y100H), the H61 residue in Zn3-MIF(S61H/Y100H) and the chloride ion in Zn3-MIF(Y100H) could be labile ligands for the generation of a vacant site on the zinc ion during catalysis. Hence, we postulated that Zn3-MIF(Y100H) could potentially be a better catalyst than Zn3-MIF(S61H/Y100H), which has a lower propensity to coordinate to zinc and a narrower space around the trinuclear zinc center where H61 likely hampers substrate access from the bottom (Fig. 3b, e).
Investigation of the Zinc-Binding Behavior of MIF(Y100H) in Solution
Next, we investigated the structure of Zn3-MIF(Y100H) in solution. While the wild-type MIF (wtMIF) is thought to be a stable trimer in solution, dynamic-light-scattering (DLS) measurements were conducted to confirm whether MIF(Y100H) and Zn3-MIF(Y100H) also exist as trimers (Fig. 4a and Supplementary Fig. 6). In the range of concentrations typically used for catalytic reactions of designer metalloenzymes, there was almost no difference between wtMIF and MIF(Y100H) in the presence or absence of zinc ions. The observed diameters are in good agreement with the crystal structures. These results suggest that MIF(Y100H) exists as a trimer and binds with a zinc ion in solution, as observed in the crystal structure.
a DLS measurements of wtMIF and the MIF(Y100H) variant. Experimental conditions: 100 μM MIF (monomer) and 100 μM ZnCl2 in 10 mM MOPS buffer (pH = 7.9). b Binding isotherm of wtMIF and MIF(Y100H) as a function of the monomer protein to zinc ion concentration ratio. The fitting curve is displayed as a dashed black line. Experimental conditions: 150 μM MIF (monomer) was titrated with 1.7 mM ZnCl2 in 10 mM MOPS buffer (pH = 7.9). Inset: thermodynamic parameters for the binding of Zn to MIF(Y100H).
To determine the stoichiometry and dissociation constant between the zinc ion and MIF(Y100H), isothermal titration calorimetry (ITC) was carried out (Fig. 4b). The corresponding results confirmed once more that three zinc ions bind with trimeric MIF(Y100H), which is consistent with the crystal structure of Zn3-MIF(Y100H). The obtained titration curve fitted well to a typical 1:1 binding model, suggesting no cooperativity upon zinc-binding in trimeric MIF(Y100H). The dissociation constant between one zinc ion and monomeric MIF(Y100H) was determined to be 8.9 μM. In contrast to MIF(Y100H), no heat signal was observed in the case of wtMIF, suggesting that it lacks the ability to bind to zinc.
The thermodynamic parameters obtained for the binding of zinc to MIF(Y100H) suggest that the zinc binding was promoted by small enthalpy and large entropy changes (inset in Fig. 4b). This finding is in good agreement with those of a previously reported de novo copper enzyme29.
Evaluation of the catalytic activity of the designer tri-zinc enzyme
To evaluate the catalytic activity of the trinuclear zinc center constructed in MIF(Y100H), we tested it in the hydrolysis of p-nitrophenyl acetate (pNA), which is a widely used benchmark reaction in designer zinc enzymes (Fig. 5a)19,39.
a Benchmark hydrolysis reaction used in this study. b Comparison of the initial reaction rates of pNA hydrolysis catalyzed by wtMIF, the MIF(S61H/Y100H) variant, and the MIF(Y100H) variant in the absence and in the presence of ZnCl2. Reaction conditions: 50 μM MIF (trimer), 150 μM ZnCl2, 2 mM pNA in 50 mM MOPS buffer (pH = 7.9) at 25 °C. c Hydrolytic activity of the MIF(S61H/Y100H) variant (orange) and the MIF(Y100H) variant (blue) as a function of the concentration of Zn2+. Reaction conditions: 50 μM MIF (trimer), 0–400 μM ZnCl2, 2 mM pNA in 50 mM MOPS buffer (pH = 7.9) at 25 °C. The black solid line represents the theoretical regression (for details, see the section ‘Estimation of the kinetic parameters’ in the Supporting Information). d, e Michaelis–Menten plots of the hydrolytic activity of wtMIF and the MIF(Y100H) variant in the absence and presence of ZnCl2. Reaction conditions: 50 μM MIF (trimer), 0 or 400 μM ZnCl2, 0.2–8 mM pNA in 50 mM MOPS buffer (pH = 7.9) at 25 °C. The blue lines in (d) represent the regression curve of the Michaelis-Menten equation. The solid and dashed lines in (e) represent the linear regression curve based on the approximation of the Michaelis-Menten equation at low substrate concentrations. In this figure, colored bar-plots and dot-plots represent mean values from three independent experiments, whilst the corresponding individual data points are displayed as black dots.
No activity enhancement was observed when zinc ions were added in the presence of wtMIF, whereas the MIF(S61H/Y100H) mutant accelerated the reaction rate in the presence of zinc ions (Fig. 5b). Thus, the trinuclear zinc center formed in the MIF pore is deemed to be responsible for the acceleration of the hydrolysis. Next, the MIF(Y100H) variant was investigated. Even without the addition of zinc ions, some enhanced activity was observed for MIF(Y100H) compared to wtMIF. This might be attributed to the potential hydrolytic activity of the histidine residue itself, which has already been reported49. The most pronounced rate acceleration upon the addition of zinc ions was observed in the case of MIF(Y100H), thus verifying our rational design, which was intended to facilitate substrate access to the active site by removing His61 from MIF(S61H/Y100H) (Fig. 5b).
To investigate the time required for the formation of the zinc center, the initial rates of the hydrolysis reactions were measured by varying the incubation time after mixing the MIF and zinc ions. Over a range from 10 min to 24 h of incubation, no obvious differences were observed, suggesting that the zinc active center is formed quickly and is stable for several hours (Supplementary Fig. 9).
In general, a zinc-hydroxy species is the active intermediate that is formed during hydrolysis by a zinc enzyme. Assuming that the coordination of the chloride ion to the zinc ion observed in the crystal structure might compete with the coordination of H2O and/or OH–, the initial rate of hydrolysis in the presence of various NaCl concentrations was measured. As expected, the initial rate of hydrolysis by MIF(Y100H) in the presence of zinc ions decreased with increasing concentration of chloride ions (Supplementary Fig. 10). However, in the absence of zinc ions, this deceleration was not observed. Considering the possibility of a counter-cation effect, KCl was used instead of NaCl, resulting in a similar deceleration. These results indicate that the formation of the hydroxyl group on the zinc ion is in competition with the coordination of Cl– to Zn.
The initial rates of hydrolysis catalyzed by MIF(S61H/Y100H) and MIF(Y100H) were measured in the presence of various equivalents of zinc ions (Fig. 5c). Consistent with the crystal-structure analysis discussed above, the bending points appeared at approximately 3 equivalents of zinc, indicating that both MIF(S61H/Y100H) and MIF(Y100H) bind with three zinc ions.
When trimeric MIF(Y100H) is used with three molar equivalents of zinc ions or less, the system produces four distinct species: MIF(Y100H); Zn1-MIF(Y100H); Zn2-MIF(Y100H); and Zn3-MIF(Y100H). Considering the Zn-dissociation constant, almost all the zinc-binding sites of MIF(Y100H) are thought to be occupied in the presence of eight equivalents of zinc relative to trimeric MIF(Y100H). Under this concentration of zinc ions, Zn3-MIF(Y100H) is the only species formed, and it is reasonable to conclude that this is when the initial rate reaches the plateau (Fig. 5c).
The initial rates of hydrolysis were measured again by changing the concentrations of the substrate. Using a comparison of the kcat/KM values, which is the slope of the linear fit of the activity plots in the low-substrate-concentration region, Zn3-MIF(Y100H) exhibited approximately 1900% of the activity of its parent, wtMIF (Fig. 5d, e).
Several designer zinc enzymes have already been reported to exhibit high catalytic activity under basic conditions (pH = 9.5). Although Zn3-MIF(Y100H) also shows a dependency on the pH in the range close to physiological conditions (pH = 6.7–7.9), it is not as pronounced as that observed for other designer zinc enzymes (Supplementary Fig. 11). The catalytic activity of Zn3-MIF(Y100H) at pH = 7.9, kcat/KM = 46.0 ± 0.4 M–1s–1, represents a best-in-class value when compared with other designer zinc enzymes at their optimum pH (pH > 9.0)18,19,20,21,38,39. Zn3-MIF (Y100H) retained 57% of its activity at pH = 7.9, even though the designer zinc enzymes reported in previous studies lose their activity at pH ≈ 7.0.
As mentioned in the ITC measurements, it was suggested that the binding of zinc to the three binding sites inside MIF is stochastic. Therefore, the ratios of four species at a fixed zinc-ion concentration can be mathematically estimated using the dissociation constant (Supplementary Fig. 12a). Individual kcat and KM values for both Zn1-MIF(Y100H) and Zn2-MIF(Y100H), which could not be isolated, can be estimated through regression analysis by using the initial rates of hydrolysis at different substrate concentrations in the presence of 0.5, 1.5, 3, and 8 equivalents of zinc, in combination with the calculated ratios of the four species (Supplementary Fig. 12b). This analysis suggests that Zn2-MIF(Y100H) and Zn3-MIF(Y100H) have smaller KM values than Zn1-MIF(Y100H) (Supplementary Fig. 12c, d), indicating that the second/third zinc atom might enhance the substrate binding as observed in natural multinuclear zinc enzymes. Thus, we have found evidence of the advantages that a multinuclear zinc site has compared to a mononuclear zinc site.
Evaluation of the intrinsic activity of MIF
Building on the success of imparting extrinsic function to MIF, we next investigated whether the intrinsic function of MIF is retained. As described above, MIF holds several independent functions related to life processes. As shown in Fig. 1c, most of the important amino-acid residues for these functions are placed on the surface of MIF. To confirm whether the newly installed tri-nuclear zinc center induces critical structural changes of these residues, which could potentially lead to complete impairing of their intrinsic functions, we focus here on one of the intrinsic functions, i.e., tautomerase activity, often assessed in structure-function-correlation studies, using isolated MIF and its variants (Fig. 6a)50.
a The benchmark tautomerization reaction used in this study. b Comparison of the reaction rates of tautomerization catalyzed by wtMIF and the MIF(Y100H) variant in the absence and presence of ZnCl2. Reaction conditions: 50 nM MIF (trimer), 50 μM ZnCl2, 2 mM 4-HPP in 1 M borate buffer (pH = 6.2) at 25 °C. The colored bar-plots and black dot-plots represent mean values and individual data from three independent experiments, respectively. c Superimposed crystal structures around the active sites for tautomerase activity in wtMIF (PDB code: 3DJH), MIF(Y100H), and Zn3-MIF(Y100H). Comparison of water-molecule networks in (d) wtMIF, (e) MIF(Y100H), and (f) Zn3-MIF(Y100H), which are presumed to be important for tautomerase activity48. Carbon atoms are colored by variant (yellow: wtMIF; pink: MIF(Y100H); cyan: Zn3-MIF(Y100H)); other atoms: zinc (deep purple), nitrogen (blue), oxygen (red), and chloride (green).
The initial rate of the tautomerase reaction was measured using 4-hydroxyphenylpyruvate (4-HPP) as the substrate (Fig. 6b). The reaction was catalyzed by wtMIF, and the addition of zinc ions did not affect the reaction rate. The initial reaction rate of MIF(Y100H) was slightly decreased compared to that using wtMIF.
While the residual activity of Zn3-MIF(Y100H) decreased by 53% compared to that of wtMIF, the intrinsic function of the protein scaffold, which is typically lost in conventional designer metalloenzymes upon introduction of novel functionalities, was indeed preserved in the case of Zn3-MIF(Y100H).
For the observed tautomerase-activity changes, the previous study indicated that the size of the pore formed at the trimer interface and the hydrogen-bond network between Y100 and H63 mediated by a water molecule within it have an allosteric effect on tautomerase activity, while the configurations of catalytic residues located on the surface involved in tautomerase activity hardly change between MIF variants with different levels of activity45,48.
As shown in Fig. 6c, the positions and orientations of catalytic tautomerase residues in wtMIF, MIF(Y100H), and Zn3-MIF(Y100H) structures are nearly identical. On the other hand, however, a comparison of the crystal structures revealed that the water-mediating hydrogen-bonding network between H63 and Y100 present in wtMIF was preserved in MIF(Y100H) but disrupted in Zn3-MIF(Y100H) upon zinc-ion binding (Fig. 6d–f). These observations align well with findings from the previous research48.
Given that the biological role of tautomerase activity is still not fully understood, it is not clear whether the observed decrease in tautomerase activity is a critical dysfunction to maintain vital functions. However, the tautomerase activity observed in this study remains within the same order of magnitude as that of wtMIF. Thus, as designed, both the intrinsic and extrinsic functions have been successfully retained and implemented.
In summary, we have succeeded in reproducing the structure of a synthetic trinuclear zinc complex in the internal space of the trimeric macrophage migration inhibitory factor (MIF) protein based on intuitive computational design. Our computational pipeline, which does not require drastic quantities of computational resources, significantly reduces experimental efforts, given that we were able to focus on the most promising variant and prepare only two variants, i.e., the identified lead candidate and one additional derivative. Additionally, our simple pipeline can be easily adapted for the grafting of other multinuclear synthetic complexes.
The designer tri-zinc enzyme, Zn3-MIF(Y100H), shows a top-tier activity compared to previously reported designer mononuclear zinc hydrolases, one of the most well-developed designer metalloenzymes. In contrast to the conventional approaches of reproducing and miniaturizing metalloenzymes, this study aimed at demonstrating that proteins can be used like small synthetic molecular ligands by synthesizing an unnatural multinuclear metal center using solely amino-acid side chains.
In addition, our designer tri-zinc enzyme exhibits not only the extrinsic function derived from the trinuclear zinc center but also retains the intrinsic functions of the wild-type MIF, something that is often sacrificed in favor of the extrinsic function in other designer enzymes. Given that MIF is produced, secreted, and translocated between cells by sensing cellular environments, harnessing its inherent disposition could allow developing designer enzymes that are autonomously produced within the body in response to physiological conditions and synthesize pharmaceuticals and/or induce any perturbation to change the physiological condition. The key discovery in this study is that grafting a trinuclear zinc center has minimal impact on the surface-derived functionality. This finding is crucial as it enables us to proceed to the next stage of our research, i.e., the integration and utilization of this designer multi-metalloenzyme in concert with cellular functions, leading to a synthetic-biology tool for the self-adaptive regulation of life phenomena51. Consequently, this report demonstrates some advantages of using natural proteins as scaffolds for designer metalloenzymes, even in light of the rapidly growing field of de novo proteins.
While the tandem reaction between hydrolysis and tautomerase activity is interesting and newly installed hydrolysis activity can be potentially exploited for the deprotection of caged drugs, it is desirable that the reaction promoted by the designer enzyme is biorthogonal. Thus, we are currently exploring non-natural reactivity by constructing other multinuclear centers based on the approach presented in this study.
Methods
Mutation candidate identification
We used a Python script to narrow down mutation candidates based on distance analysis. From the wild-type structure (PDB code 1MIF), we extracted the coordinates of the Cα atoms of the amino-acid residues on the inner β-sheet, which we refer to as the MIF-Cαcoordinates. Additionally, we extracted the Zn coordinates from the crystal structure of the Zn tri-nuclear synthetic complex (CCDC 931956), referred to as the Tri-Zn coordinates. The MIF-Cα coordinates were rotated around the symmetry axis of the pore along the z-axis such that the center of mass was positioned at z = 0 and the N-terminal end was oriented toward the +z direction. The Tri-Zn coordinates were aligned parallel to the xy-plane and translated along the z-axis in 0.5 Å increments within the range of 0–10 Å. At each z-coordinate, the Tri-Zn coordinates were rotated in 2° increments within the range of 0–120°, resulting in a total of 1281 points where the distance between the Cα and Zn was calculated. For each Zn atom, the three closest Cα atoms were identified within a distance range of 6.0–7.0 Å. Among all possible triplets, those in which the difference between the maximum and minimum of the three Cα–Zn distances was <0.2 Å were selected as candidates to ensure geometric uniformity. Triplets containing duplicated residue numbers were excluded. Based on this screening, a total of eight representative coordination candidates, i.e., S611-H631-Y1001, S612-H631-Y1001, S612-H631-Y1002, S611-H631-N981, S612-H631-N982, H631-N981-Y1001, H631-N982-Y1001, and H631-N982-Y1002, were selected for further DFT calculations.
DFT calculation
Based on the results from the Python analysis, we performed structural optimizations using density-functional-theory (DFT) calculations for eight selected candidates, i.e., S611-H631-Y1001, S612-H631-Y1001, S612-H631-Y1002, S611-H631-N981, S612-H631-N982, H631-N981-Y1001, H631-N982-Y1001, and H631-N982-Y1002. The DFT calculations included three zinc ions, the CO₃²⁻ bridging group, and nine histidine side chains with hydrogen atoms added, but excluded the larger protein structure. The coordinates of the main-chain atoms (N, Cα, and C) belonging to the β-sheet framework were fixed to preserve the protein backbone structure, while the remaining atoms were fully relaxed. All of the geometry optimizations were performed at the PBE1PBE-GD3BJ/6–31 G(d,p) level in the Gaussian 16 package (version ES64L-G16RevC.01)52.
Molecular dynamics simulations
All-atom molecular dynamics (MD) simulations were performed with the CHARMM36m force field53 using the LAMMPS software54. Atomic coordinates of human MIF were taken from the Protein Data Bank (PDB code: 1MIF). The MIF structure was placed in a periodic simulation box and solvated by adding water. The salt concentration was set to 0.15 M NaCl, and the corresponding numbers of Na+ and Cl− ions were added. After the steepest-descent energy minimization, the systems were equilibrated for 1 ns in the NPT ensemble at 300 K and 1 atm. Production runs were then performed for 100 ns in the NPT ensemble at 300 K and 1 atm, and the trajectory data were collected every 10 ps. The temperature was maintained with a Nosé–Hoover thermostat55,56 and the pressure was controlled with a Parrinello–Rahman barostat57,58. The nonbonded interactions were calculated with a cutoff distance of 1.2 nm, and the particle−particle particle−mesh (PPPM) method59 was used to calculate long-range electrostatic interactions. The equations of motion were integrated using the Verlet algorithm60 with a time step of 2 fs, along with the SHAKE algorithm61 to constrain the lengths of bonds to hydrogen. The root mean square fluctuations (RMSFs) of Cα atoms for each residue were calculated with respect to the average structure obtained from the last 30 ns of the trajectories. The time-averaged pore radius and its standard deviations along the Z-coordinate were also computed over the last 30 ns using the HOLE program62. VMD63 was used to generate the images.
Protein expression, purification, and crystallization
As listed in Supplementary Table 3, expression plasmids for the wild-type MIF, the S61H/Y100H variant, and the Y100H variant were transformed into E. coli BL21(DE3). A single colony of the transformed cells was inoculated into 10 mL of Luria-Bertani broth (LB) containing 50 μg/mL of ampicillin and 1% (w/v) glucose and incubated at 37 °C until the OD600 reached 0.8. The resulting culture was then inoculated into 700 mL of LB medium containing 50 μg/mL of ampicillin to give a final OD600 of 0.002 and incubated aerobically with vigorous shaking at 37 °C. When the OD600 reached between 0.8 and 1.0, the culture was cooled on ice for 10 min. Isopropyl β-D-1-thiogalactopyranoside (IPTG) was added to a final concentration of 0.4 mM to induce protein expression. The resulting culture was then incubated at 20 °C overnight with shaking, and the cells were harvested by centrifugation (4000 × g at 4 °C for 15 min) and stored at −80 °C.
The frozen harvested cells from the 700 mL culture were resuspended in 30 mL of 20 mM Tris-HCl buffer (pH = 7.4, adjusted at 4 °C) and lysed by sonication. MgCl2 was added to the lysate to give a final concentration of 1 mM, followed by the addition of DNase I (20 units). After a 20 min incubation on ice, streptomycin sulfate was added to the mixture to give a final concentration of 1% (w/v). The resulting mixture was further incubated on ice for 30 min. Cell debris and precipitates were removed from the lysate by centrifugation at 45,000 × g at 4 °C for 30 min. Ammonium sulfate was added to the supernatant to a final saturation of 55%, and the solution was incubated on ice for 60 min. The precipitated protein was collected by centrifugation at 45,000 × g at 4 °C for 30 min. The obtained precipitate was dissolved in 4 mL of 20 mM Tris-HCl buffer (pH = 7.4, adjusted at 4 °C) and dialyzed against 2 L of 20 mM Tris-HCl buffer (pH = 7.4) three times at 4 °C for 4 h, 4 h, and 12 h. The resulting solution was diluted with 26 mL of 20 mM Tris-HCl buffer (pH = 7.4). After centrifugation at 45,000 × g at 4 °C for 10 min, the supernatant was passed through a 0.22 μm filter. The filtered solution was loaded onto a column packed with 30 mL of Bio-Rad Macro-Prep High Q anion exchange resin, which had been pre-equilibrated in 20 mM Tris-HCl buffer (pH = 7.4). The flow-through fraction was collected and buffer-exchanged into 10 mM MOPS buffer (pH = 7.9) by dialysis (8 kDa cutoff). The obtained sample was concentrated using an Amicon Ultra-15 (10 kDa cutoff) centrifugal filter unit to reach a final concentration of 300 μM MIF trimer. The purified protein solution was stored at −80 °C. The concentrations of the purified proteins were determined using the calculated absorption coefficients at 280 nm based on their amino acid sequences (38,850 M–1cm–1 for WT MIF trimer and 34,380 M–1cm–1 for the S61H/Y100H and Y100H variants). MIF variants were crystallized at 20 °C using hanging-drop vapor-diffusion method. Crystals were obtained using a reservoir solution containing 2.0 M ammonium sulfate, 20 mM Tris-HCl buffer (pH = 7.5), and 3% isopropanol. The hanging-drop solution contained 1.9 μL of the protein solution (1.0 mM) in 20 mM Tris-HCl buffer (pH = 7.0) and 1.9 μL of the reservoir solution. For Zn3-MIF(S61H/Y100H)-L (PDB code: 9JIY) and Zn3-MIF(Y100H) (PDB code: 9JJ0), 0.2 μL of 100 mM ZnCl2 (10.5 eq. per MIF monomer) was added to the hanging drop solution. For Zn3-MIF(S61H/Y100H) (PDB code: 9JIZ), 0.5 μL of 100 mM ZnCl2 (26.0 eq. per MIF monomer) was added to the hanging-drop solution.
DLS measurements
Protein solutions were placed in plastic DTS0012 cuvettes or ZMV1002 quartz cells (both from Malvern Panalytical Ltd.) and maintained at 20 °C for at least 3 min before light scattering was measured.
ITC measurements
MIF trimer (50 μM in 5 mM MOPS buffer (pH = 7.9)) was titrated with ZnCl2 (1.7 mM in 5 mM MOPS buffer (pH = 7.9)) using a Malvern Panalytical MicroCal PEAQ-ITC. The cell was filled with MIF, and a syringe was filled with a zinc solution. The zinc solution was injected 13 times in portions of 3 μL (6 s duration and 210 s spacing). The obtained data were analyzed using the Malvern Panalytical MicroCal PEAQ-ITC analysis software.
Reaction conditions for the hydrolysis experiments
Experiments were performed in reaction mixtures where the final concentrations of each component were: MIF trimer (50 μM); ZnCl2 (0-400 μM); p-nitrophenyl acetate (pNA) (0.2–8 mM); and NaCl or KCl (0-96 mM). In more detail, a Znx-MIF stock solution and pNA stock solution were freshly prepared using the conditions listed in Supplementary Tables 5 and 6, respectively. Using a 12-channel pipet equipped with a mixer (Supplementary Fig. 7), 220 μL of Znx-MIF stock solution and 220 μL of pNA stock solution were mixed and then loaded directly into an Eppendorf UVette® cuvette set in an Agilent Technologies Cary 3500 spectrophotometer equipped with a thermostatic cell holder maintained at 25 °C. The absorption changes at 405 nm as a function of time were recorded for 90 s at 0.1 s intervals. The concentrations of p-nitrophenol at each pH were determined based on the calibration curves (Supplementary Fig. 8).
Reaction conditions for the tautomerization reactions
Experiments were performed under reaction conditions where the final concentrations in the reaction mixture are: MIF trimer (50 nM); ZnCl2 (50 μM); and p-hydroxyphenylpyruvic acid (2 mM). In more detail, the Znx-MIF stock solution was prepared using the conditions listed in Supplementary Table 7. The mixtures listed in Supplementary Table 8 were prepared in an Eppendorf UVette® cuvette set in an Agilent Technologies Cary 3500 spectrophotometer equipped with a thermostatic cell holder maintained at 20 °C. The absorption changes at 306 nm as a function of time were recorded for 60 s at 0.1 s intervals. The concentrations of the product were determined based on the previously reported absorption coefficient at 306 nm (11400 M–1cm–1)50.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
The crystallography data generated in this study have been deposited in the Protein Data Bank under accession codes 9JIT, 9JIV, 9JIY, 9JIZ, and 9JJ0. All other data reported in this study are available in the main text, the Supplementary Information, or Source Data. Source data are provided with this paper.
Code availability
The Python code of the geometric screening is available as Supplementary Code.
References
McWilliams, S. F. & Holland, P. L. Dinitrogen binding and cleavage by multinuclear iron complexes. Acc. Chem. Res. 48, 2059–2065 (2015).
Ohki, Y. et al. Nitrogen reduction by the Fe sites of synthetic [Mo3S4Fe] cubes. Nature 607, 86–90 (2022).
Cook, S. A. & Borovik, A. S. Molecular designs for controlling the local environments around metal ions. Acc. Chem. Res. 48, 2407–2414 (2015).
Fujisaki, H. et al. Selective methane oxidation by molecular iron catalysts in aqueous medium. Nature 616, 476–481 (2023).
Schwizer, F. et al. Artificial metalloenzymes: reaction scope and optimization strategies. Chem. Rev. 118, 142–231 (2018).
Chen, D. et al. An evolved artificial radical cyclase enables the construction of bicyclic terpenoid scaffolds via an H-atom transfer pathway. Nat. Chem. https://doi.org/10.1038/s41557-024-01562-5. (2024).
Ji, P., Park, J., Gu, Y., Clark, D. S. & Hartwig, J. F. Abiotic reduction of ketones with silanes catalysed by carbonic anhydrase through an enzymatic zinc hydride. Nat. Chem. 13, 312–318 (2021).
Longwitz, L., Leveson-Gower, R. B., Rozeboom, H. J., Thunnissen, A.-M. W. H. & Roelfes, G. Boron catalysis in a designer enzyme. Nature 629, 824–829 (2024).
Trimble, J. S. et al. A designed photoenzyme for enantioselective [2+2] cycloadditions. Nature 611, 709–714 (2022).
Sun, N. et al. Enantioselective [2+2]-cycloadditions with triplet photoenzymes. Nature 611, 715–720 (2022).
Oohora, K. & Hayashi, T. Myoglobins engineered with artificial cofactors serve as artificial metalloenzymes and models of natural enzymes. Dalton Trans. 50, 1940–1949 (2021).
Srivastava, P., Yang, H., Ellis-Guardiola, K. & Lewis, J. C. Engineering a dirhodium artificial metalloenzyme for selective olefin cyclopropanation. Nat. Commun. 6, 7789 (2015).
Yamamura, K. & Kaiser, E. T. Studies on the oxidase activity of copper(II) carboxypeptidase A. J. Chem. Soc., Chem. Commun. 830 https://doi.org/10.1039/c39760000830. (1976).
Okrasa, K. & Kazlauskas, R. J. Manganese-substituted carbonic anhydrase as a new peroxidase. Chem. Eur. J. 12, 1587–1596 (2006).
Fernández-Gacio, A., Codina, A., Fastrez, J., Riant, O. & Soumillion, P. Transforming carbonic anhydrase into epoxide synthase by metal exchange. ChemBioChem 7, 1013–1016 (2006).
Fujieda, N. et al. A Well-defined osmium–cupin complex: hyperstable artificial osmium peroxygenase. J. Am. Chem. Soc. 139, 5149–5155 (2017).
Song, W. J. & Tezcan, F. A. A designed supramolecular protein assembly with in vivo enzymatic activity. Science 346, 1525–1528 (2014).
Zastrow, M. L., Peacock, A. F. A., Stuckey, J. A. & Pecoraro, V. L. Hydrolytic catalysis and structural stabilization in a designed metalloprotein. Nat. Chem. 4, 118–123 (2012).
Rufo, C. M. et al. Short peptides self-assemble to produce catalytic amyloids. Nat. Chem. 6, 303–309 (2014).
Studer, S. et al. Evolution of a highly active and enantiospecific metalloenzyme from short peptides. Science 362, 1285–1288 (2018).
Jeong, W. J. & Song, W. J. Design and directed evolution of noncanonical β-stereoselective metalloglycosidases. Nat. Commun. 13, 6844 (2022).
Caldwell, S. J. et al. Tight and specific lanthanide binding in a de novo TIM barrel with a large internal cavity designed by symmetric domain fusion. Proc. Natl. Acad. Sci. USA 117, 30362–30369 (2020).
Yeung, N. et al. Rational design of a structural and functional nitric oxide reductase. Nature 462, 1079–1082 (2009).
Mirts, E. N., Petrik, I. D., Hosseinzadeh, P., Nilges, M. J. & Lu, Y. A designed heme-[4Fe-4S] metalloenzyme catalyzes sulfite reduction like the native enzyme. Science 361, 1098–1101 (2018).
Lombardi, A., Pirro, F., Maglio, O., Chino, M. & DeGrado, W. F. De novo design of four-helix bundle metalloproteins: one scaffold, diverse reactivities. Acc. Chem. Res. 52, 1148–1159 (2019).
Reig, A. J. et al. Alteration of the oxygen-dependent reactivity of de novo Due Ferri proteins. Nat. Chem. 4, 900–906 (2012).
Ulas, G., Lemmin, T., Wu, Y., Gassner, G. T. & DeGrado, W. F. Designed metalloprotein stabilizes a semiquinone radical. Nat. Chem. 8, 354–359 (2016).
Zhang, S.-Q. et al. De novo design of tetranuclear transition metal clusters stabilized by hydrogen-bonded networks in helical bundles. J. Am. Chem. Soc. 140, 1294–1304 (2018).
Pirro, F. et al. A de novo-designed type 3 copper protein tunes catechol substrate recognition and reactivity. Angew. Chem. Int. Ed. 62, e202211552 (2023).
Abe, S. et al. Polymerization of phenylacetylene by rhodium complexes within a discrete space of apo-ferritin. J. Am. Chem. Soc. 131, 6958–6960 (2009).
Choi, T. S. & Tezcan, F. A. Overcoming universal restrictions on metal selectivity by protein design. Nature 603, 522–527 (2022).
Hoffnagle, A. M. & Tezcan, F. A. Atomically accurate design of metalloproteins with predefined coordination geometries. J. Am. Chem. Soc. 145, 14208–14214 (2023).
Jeffery, C. J. Moonlighting proteins. Trends Biochem. Sci. 24, 8–11 (1999).
Krishna, R. et al. Generalized biomolecular modeling and design with RoseTTAFold All-Atom. Science 384, eadl2528 (2024).
Okamoto, Y. et al. A cell-penetrating artificial metalloenzyme regulates a gene switch in a designer mammalian cell. Nat. Commun. 9, 1943 (2018).
Chang, T., Vong, K., Yamamoto, T. & Tanaka, K. Prodrug activation by gold artificial metalloenzyme-catalyzed synthesis of phenanthridinium derivatives via hydroamination. Angew. Chem., Int. Ed. 60, 12446–12454 (2021).
Coleman, J. E. Zinc proteins: enzymes, storage proteins, transcription factors, and replication proteins. Annu. Rev. Biochem. 61, 897–946 (1992).
Zastrow, M. L. & Pecoraro, V. L. Designing hydrolytic zinc metalloenzymes. Biochemistry 53, 957–978 (2014).
Zastrow, M. L. & Pecoraro, V. L. Influence of active site location on catalytic activity in de novo-designed zinc metalloenzymes. J. Am. Chem. Soc. 135, 5895–5903 (2013).
Basler, S. et al. Efficient Lewis acid catalysis of an abiological reaction in a de novo protein scaffold. Nat. Chem. 13, 231–235 (2021).
Der, B. S., Edwards, D. R. & Kuhlman, B. Catalysis by a de novo zinc-mediated protein interface: implications for natural enzyme evolution and rational enzyme engineering. Biochemistry 51, 3933–3940 (2012).
Weston, J. Mode of action of bi- and trinuclear zinc hydrolases and their synthetic analogues. Chem. Rev. 105, 2151–2174 (2005).
Liu, X., Du, P. & Cao, R. Trinuclear zinc complexes for biologically relevant μ3-oxoanion binding and carbon dioxide fixation. Nat. Commun. 4, 2375 (2013).
Calandra, T. & Roger, T. Macrophage migration inhibitory factor: a regulator of innate immunity. Nat. Rev. Immunol. 3, 791–800 (2003).
Lubetsky, J. B., Swope, M., Dealwis, C., Blake, P. & Lolis, E. Pro-1 of macrophage migration inhibitory factor functions as a catalytic base in the phenylpyruvate tautomerase activity. Biochemistry 38, 7346–7354 (1999).
Park, H. et al. PAAN/MIF nuclease inhibition prevents neurodegeneration in Parkinson’s disease. Cell 185, 1943–1959.e21 (2022).
Meza-Romero, R. et al. Modeling of both shared and distinct interactions between MIF and its homologue D-DT with their common receptor CD74. Cytokine 88, 62–70 (2016).
Pantouris, G. et al. Regulation of MIF enzymatic activity by an allosteric site at the central solvent channel. Cell Chem. Biol. 27, 740–750.e5 (2020).
Schoonen, L. et al. Alternative application of an affinity purification tag: hexahistidines in ester hydrolysis. Sci. Rep. 7, 14772 (2017).
Parkins, A. & Pantouris, G. Protocol for purification and enzymatic characterization of members of the human macrophage migration inhibitory factor superfamily. STAR Protoc. 4, 102375 (2023).
Scheller, L. & Fussenegger, M. From synthetic biology to human therapy: engineered mammalian cells. Curr. Opin. Biotechnol. 58, 108–116 (2019).
Frisch, M. J. et al. Gaussian 16 Rev. C.01. (2016).
Huang, J. et al. Charmm36m: an improved force field for folded and intrinsically disordered proteins. Nat. Methods 14, 71–73 (2017).
Plimpton, S. Fast parallel algorithms for short-range molecular dynamics. J. Comp. Phys. 117, 1–19 (1995).
Nosé, S. A. Molecular-dynamics method for simulations in the canonical ensemble. Mol. Phys. 52, 255–268 (1984).
Hoover, W. G. Constant-pressure equations of motion. Phys. Rev. A 34, 2499–2500 (1986).
Martyna, G. J., Tobias, D. J. & Klein, M. L. Constant pressure molecular dynamics algorithms. J. Chem. Phys. 101, 4177–4189 (1994).
Parrinello, M. & Rahman, A. Polymorphic transitions in single crystals: a new molecular dynamics method. J. Appl. Phys. 52, 7182–7190 (1981).
Hockney, R. W. & Eastwood, J. W. Computer simulation using particles. (Hilger: Bristol, 1989).
Verlet, L. Computer “experiments” on classical fluids. I. Thermodynamical properties of Lennard-Jones molecules. Phys. Rev. 159, 98–103 (1967).
Ryckaert, J.-P., Ciccotti, G. & Berendsen, H. J. C. Numerical integration of the Cartesian equations of motion of a system with constraints: molecular dynamics of N-alkanes. J. Comp. Phys. 23, 327–341 (1977).
Smart, O. S., Neduvelil, J. G., Wang, X., Wallace, B. A. & Sansom, M. S. P. Hole: a program for the analysis of the pore dimensions of ion channel structural models. J. Mol. Graph. 14, 354–360 (1996).
Humphrey, W., Dalke, A. & Schulten, K. Vmd: visual molecular dynamics. J. Mol. Graph. 14, 33–38 (1996).
Acknowledgements
Y.O., T.M., and T.I. acknowledge generous support from the Frontier Research Institute for Interdisciplinary Sciences (FRIS), Tohoku University. Y.O. and T.M. gratefully acknowledge support from the Tohoku Initiative for Fostering Global Researchers for Interdisciplinary Sciences (TI-FRIS) of MEXT’s Strategic Professional Development Program for Young Researchers. Y.O. gratefully acknowledges JST ACT-X (JPMJAX1913), JST PRESTO (JPMJPR22A4), and JSPS Grants-in-Aid for Scientific Research (21H05118). T.M. gratefully acknowledges the JST FOREST Program (JPMJFR212H). T.H. gratefully acknowledges JSPS Grants-in-Aid for Scientific Research (24K08609). Generous support from the FRIS CoRE, which is a shared research environment, as well as Prof. Shinsuke Niwa’s group at Tohoku University is acknowledged. A part of this work was conducted at the Institute for Molecular Science, supported by Advanced Research Infrastructure for Materials and Nanotechnology in Japan (JPMXP1223MS1070) of the Ministry of Education, Culture, Sport, Science and Technology (MEXT), Japan. Prior to submission, this manuscript has been revised by a professional language-editing service (www.mayerscientificediting.com). Synchrotron experiments were conducted at beamlines BL44XU, BL41XU, and BL45XU at SPring-8 (Hyogo, Japan) in collaboration with the Institute for Protein Research at Osaka University and the Japan Synchrotron Radiation Research Institute. This research was supported by Joint Research of the Exploratory Research Center on Life and Living Systems (ExCELLS). (ExCELLS program No, 25EXC604).
Author information
Authors and Affiliations
Contributions
Y.O. conceived, designed, and directed the overall project. A.U., F.T., and Y.O. designed and performed the experiments and analyzed the results, except for those specifically noted below. T.M. and Y.O. designed and performed the computational design of the designer multimetalloenzymes. T.H. performed the X-ray crystal-structure analysis and DLS measurements. T.I. designed and prepared a mixing unit for the enzymatic reactions. T.K. designed the mathematical models and analyzed the kinetic parameters. T.M., T.H., T.K., and Y.O. wrote the manuscript; all authors have given approval to the final version of the manuscript.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Nature Communications thanks Erik Donovan Hedegård, who co-reviewed with Gabriel Rodrigues; and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Source data
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Ueno, A., Takida, F., Kita, T. et al. A cytokine-based designer enzyme with an abiological multinuclear metal center exhibits intrinsic and extrinsic catalysis. Nat Commun 16, 6781 (2025). https://doi.org/10.1038/s41467-025-61909-5
Received:
Accepted:
Published:
Version of record:
DOI: https://doi.org/10.1038/s41467-025-61909-5
