This is an open access article published under an ACS AuthorChoice License, which permits copying and redistribution of the article or any adaptations for non-commercial purposes. Article Cite This: ACS Omega 2019, 4, 4437−4447 http://pubs.acs.org/journal/acsodf Functionalized Artificial Bidomain Proteins Based on an α‑Solenoid Protein Repeat Scaffold: A New Class of Artificial Diels−Alderases Thibault Di Meo,†,‡,∥ Kalani Kariyawasam,†,∥ Wadih Ghattas,† Marie Valerio-Lepiniec,‡ Giuseppe Sciortino,§ Jean-Didier Mareć hal,§ Philippe Minard,‡ Jean-Pierre Mahy,*,† Agathe Urvoas,*,‡ ́ y Ricoux*,† and Rem † Institut de Chimie Moléculaire et des Matériaux d’Orsay, UMR 8182 CNRS, Laboratoire de Chimie Bioorganique et Bioinorganique, Bât. 420, Univ. Paris-Sud, Université Paris-Saclay, 91405 Orsay Cedex, France ‡ Institute for Integrative Biology of the Cell (I2BC), CEA, CNRS, Univ. Paris-Sud, Université Paris-Saclay, 91198 Gif-sur-Yvette See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles. Cedex, France § Departament de Química, Universitat Autònoma de Barcelona, Edifici C.n., 08193 Cerdanyola del Vallés, Barcelona, Spain * S Supporting Information Downloaded via 83.56.229.240 on May 21, 2021 at 11:16:30 (UTC). ABSTRACT: αRep is a family of entirely artificial repeat proteins. Within the previously described αRep library, some variants are homodimers displaying interdomain cavities. Taking advantage of these properties, one of these homodimers called αRep A3 was converted into entirely artificial single chain bidomain metalloenzymes. A non- mutated A3 domain was covalently linked with an A3′ domain bearing a unique cysteine on a chosen mutated position (F119C or Y26C). This single mutation ensured the covalent coupling of a 1:1 copper(II)/phenanthroline or copper(II)/terpyridine complex as a catalytic center within the interdomain cavity which was maintained large enough to accommodate two substrates of the Diels−Alder (D−A) reaction. This allowed us to obtain four new artificial Diels−Alderases that were fully characterized by matrix-assisted laser desorption ionization time-of-flight mass spectrometry, UV−vis spectroscopy, and size exclusion chromatography analyses and were then further used for the catalysis of the D−A reaction. They were found to be able to catalyze the enantioselective D−A reaction of azachalcone with cyclopentadiene with up to 38% yield and 52% enantiomeric excess, which validates the proposed strategy. Moreover, the data were rationalized with a computational strategy suggesting the key factors of the selectivity. These results suggest that artificial metalloenzymes based on bidomain A3_A3 proteins modified with nitrogen donor ligands may be suitable for further catalyst optimization and may constitute valuable tools toward more efficient and selective artificial biocatalysts. ■ INTRODUCTION Artificial metalloenzymes have emerged as a powerful new phenanthroline moiety. The choice of these two positions was based on the crystal structure of αRep A3.3 For these two approach to expand the repertoire of chemical reactions positions, the side chains point into the inter-subunit crevice performed by enzymes.1 Engineering such new biohybrid quite deep in the crevice and these positions are among a set of catalysts requires a stable host protein displaying a cavity wide positions highly tolerant to side chain substitutions. In enough to accommodate a metal complex responsible for the addition, models of two resulting biohybrids, αRep catalytic process and accessible for the substrates of the A3F119C-phenanthroline (A3-F119C-Phen) and αRep catalyzed reaction. αRep are artificial proteins especially well A3Y26C-phenanthroline (A3-Y26C-Phen), suggested that the adapted to the development of a new class of hybrid catalysts. dimer could accommodate two phenanthroline ligands in the These highly thermostable proteins based on an idealized α- crevice. Finally, the distance between the two β-carbon atoms helical repeated motif2,3 are devoid of cysteine residues and of symmetrical Tyr26 residues was longer than the distance present a hypervariable surface that can be evolved to create between the two β-carbon atoms of symmetrical Phe119 new protein−protein interactions.2,4 Recently, described αRep- residues, suggesting that the resulting biohybrids would based biohybrid catalysts were designed using a homodimeric present different catalytic properties. The best resulting αRep named αRep A3 as protein scaffold. These new artificial biohybrid A3F119C-Phen−Cu(II) was found to be able to Diels−Alderases were obtained by the covalent grafting of a copper(II)−phenanthroline moiety into cysteine-mutated Received: December 9, 2018 αRep A3 variants. Positions 26 and 119 were mutated into Accepted: February 13, 2019 cysteine and then used for the covalent coupling of the Published: February 28, 2019 © 2019 American Chemical Society 4437 DOI: 10.1021/acsomega.8b03448 ACS Omega 2019, 4, 4437−4447 ACS Omega Article Figure 1. Successive steps for the preparation of biohybrids based on the (A3_A3′)F119C and (A3_A3′)Y26C variants functionalized with -Phen− and -Terpy−Cu(II) complexes: (A) genetic constructions encoding the bidomain A3_A3′ proteins obtained by site-directed cysteine mutations F119C or Y26C. The resulting plasmids were transformed in E. coli and the two proteins (A3_A3′)F119C and (A3_A3′)Y26C were produced and purified. (B) Covalent ligand coupling of phenanthroline or terpyridine functionalized with a maleimide moiety followed by a Cu(II) metallation step. (C) Models for the four resulting biohybrids based on the crystallographic structure of A3 homodimer (3LTM). enantioselectively catalyze the Diels−Alder cycloaddition of Herein, we report the successful production of single-chain azachalcone with cyclopentadiene with up to 62% enantio- bidomain proteins in which a cysteine-free domain was linked meric excess (ee) but with only 11% yield. Such a low yield with a domain bearing either the F119C mutation [(A3_A3′)- could be explained by the presence, as the catalytic species, of F119C] or the Y26C mutation [(A3_A3′)Y26C]. Further the 1:2 copper(II) bis-phenanthroline complex buried in the chemical coupling of either a phenanthroline ligand (Phen) or crevice formed by the protein dimer, which could block the a bulkier terpyridine ligand (Terpy) followed by insertion of a access of the two substrates to the catalytic pocket.3 The copper(II) ion led to four new artificial metalloenzymes formation of such a species was due to the close proximity of ((A3_A3′)F119C-Phen−Cu(II), (A3_A3′)F119C-Terpy− the two phenanthroline ligands covalently attached to the two Cu(II), (A3_A3′)Y26C-Phen−Cu(II), and (A3_A3′)Y26C- Cys119 residues facing each other in the symmetric dimer. Terpy−Cu(II)) that were characterized by matrix-assisted laser Then, this suggested developing a strategy to anchor only one desorption ionization time-of-flight (MALDI-ToF) mass copper(II) complex as the catalytic center in the catalytic spectrometry (MS) and size exclusion chromatography. This pocket formed by the dimer’s interface. In addition, such an new class of artificial metalloenzymes was found to be able to approach should maintain a large catalytic pocket in order to catalyze the enantioselective Diels−Alder (D−A) reaction of accommodate both D−A substrates and thus improve the azachalcone with cyclopentadiene presenting up to 38% yield catalytic properties of the biohybrid catalysts. Such an and ee up to 52%, which validates the proposed strategy. asymmetric biohybrid cannot be obtained from a homodimeric Molecular modeling allowed the identification of the protein and therefore needs to be created from one protein preferential orientations of the two copper complexes in the binding pocket. The computation allowed suggesting the key scaffold containing two different domains linked together in recognition factors of the stereo determining step of the order to obtain an asymmetric bidomain protein. The metal reaction. ■ complex can then be grafted on a single cysteine located on one of the two domains while the second domain contains no cysteine residue. We recently described the construction of the RESULTS AND DISCUSSION single chain bidomain protein A3_A3 in which two A3 Engineering of the Host Protein. As the first αRep- domains were encoded by different genetic sequences and based biohybrid design was based on the αRep A3 homodimer, linked by a flexible (SGGGG)2 peptide.5 the introduction of a single cysteine in the polypeptide chain 4438 DOI: 10.1021/acsomega.8b03448 ACS Omega 2019, 4, 4437−4447 ACS Omega Article Figure 2. MALDI-ToF spectra of the (A3_A3′)Y26C (A) and (A3_A3′)F119C (B) protein variants measured before (blue) and after functionalization with terpyridine (red) and phenanthroline (green). resulted in the presence of two cysteine residues within the maleimido group, respectively, which could react with the thiol dimer’s cavity and therefore necessarily ended up with a function of the cysteinyl residues of the mutant proteins symmetrical coupling of two chemical ligands. In order to (Figure 1). The reactivity and accessibility of the thiol function explore the effect of a single metal complex anchored in the of the cysteinyl residues was then assessed, thanks to the time- cavity of this protein, we needed to mutate one residue of only resolved Ellman’s assay using sulfhydryl reagent 5,5-dithio-bis- one domain of the dimer into cysteine. The differentiation of (2-nitrobenzoic acid) (DTNB) that reacts with sulfhydryl the two domains of the dimer can be achieved by mutation if groups, releasing 2-nitro-5-thiobenzoate (TNB−) which dis- they were expressed as in a bidomain monomeric protein. We plays a maximum absorption at 412 nm.8 First, for each one of recently described the genetic construction of a plasmid the two variants, the labeled cysteine concentration at the end encoding such an A3_A3 bidomain monomeric protein, in of the reaction was equal to the protein concentration. This which a synthetic nonrepeated gene encoded the first domain indicated that all cysteine residues were reactive and did not and the repeated sequence contained in the αRep library form any disulfide bridge even after several weeks’ storage encoded the second domain. The two domains are covalently without any reducing agent. Second, the kinetic of the reaction linked on a single polypeptide chain by a flexible linker was different for the two variants: (A3_A3′)Y26C showed very sequence. The X-ray structure of this nonmutated bidomain fast labeling kinetics and led rapidly to a stable maximum A3_A3 protein (PDB ID 6FT5) showed that the linker did not absorbance at 412 nm: (t1/2 = 30 s), while (A3_A3′)F119C affect the global structure of the monomer and functional reacted more slowly with DTNB and the absorbance at 412 studies indicated that the linker reinforced the closed form of nm reached a plateau value only after several minutes: (t1/2 = 3 the bidomain protein.5 Taking advantage of the nonrepeated min 30 s). sequence of the first domain, mutations could selectively be This denoted in (A3_A3′)F119C either a sterically hindered introduced in this domain. In order to compare the activities of environment or a shift in the pKa of Cys119. Thus, the thiol the resulting biohybrids with the previously described dimer’s function appeared to be more accessible and more reactive in activities, the same positions as in the dimer were chosen for (A3_A3′)Y26C than in (A3_A3′)F119C. It is noteworthy cysteine mutation (F119 and Y26). The corresponding that, in both cases, those t1/2 values were longer than those synthetic genes obtained by site directed mutagenesis were measured for the symmetrical (A3Y26C)2 and (A3F119C)2 sub-cloned into a modified expression vector, which resulted in containing no linker (resp. 30 s vs instantaneous and 3 min 30 plasmids encoding the two bidomain variants (A3_A3′)F119C s vs 1 min 40 s).3 This indicated that in the single chain and (A3_A3′)Y26C. The proteins were produced in (A3_A3′)Y26C and (A3_A3′)F119C proteins the linker Escherichia coli and purified as described and yields were peptide stabilizes the closed form of the binding cavity. This equivalent to the ones obtained with the dimers (between 50 result is in accord with by the functional studies reported for and 70 mg of pure protein per liter of culture). Analytic size the nonmutated A3_A3, indicating that the interdomain exclusion chromatography profiles were equivalent to that of interactions are reinforced in the bidomain as compared to the A3 dimer and to that of the nonmutated A3_A3 bidomain the A3 homo-dimer.5 protein, indicating that the mutations did not affect the global Protein Functionalization. The two variants protein conformation. (A3_A3′)Y26C and (A3_A3′)F119C were then functionalized Bioconjugation. For this study, we selected two robust by incubation with 1.2 equiv of 2-bromo-N-(1,10-phenan- copper(II) ligands with different nitrogen donor motifs that is, throlin-5-yl)acetamide hydrobromide or 2 equiv of N-[2- phenanthroline and terpyridine, which are known to bind ([2,2:6,2-terpyridine]-4-yloxy)ethyl]maleimide overnight at 4 copper(II) with strong affinities to give Cu(II) complexes and °C, resulting in the formation of the cysteine−phenanthroline that are able to catalyze D−A reactions.6,7 Derivatives of these (Phen) or cysteine−terpyridine (Terpy) labeled proteins in ligands had to be prepared to enable their covalent attachment their apo form: (A3_A3′)Y26C-Phen, (A3_A3′)F119C-Phen, to the (A3_A3′)F119C and (A3_A3′)Y26C variants. 2- (A3_A3′)Y26C-Terpy, and (A3_A3′)F119C-Terpy. The Bromo-N-(1,10-phenanthrolin-5-yl)acetamide (1) and N-[2- complete functionalization was confirmed by Ellman’s assay, ([2,2′:6′,2″-terpyridin]-4′-yloxy)ethyl]maleimide (3) were with a complete loss of thiol reactivity (data not shown). prepared following already described procedures (see Exper- MALDI-ToF MS analysis was first used to characterize the imental Section) as they bear reactive bromoacetyl group and biohybrids (Figure 2). Peaks at 42 564 ± 5 and 42 706 ± 5 Da 4439 DOI: 10.1021/acsomega.8b03448 ACS Omega 2019, 4, 4437−4447 ACS Omega Article Figure 3. Metallation of the (A3_A3′)Y26C-Terpy and (A3_A3′)F119C-Terpy biohybrids with Cu(II). (A) Evolution of the UV−visible spectrum of (A3_A3′)Y26C-Terpy after the progressive addition of up to 1.6 equiv Cu(NO3)2 to a 5.6 μM solution of (A3_A3′)Y26C-Terpy in 20 mM MOPS 150 mM NaCl pH 7 at 20 °C. Inset: evolution of the absorbance at 320 nm as a function of the number of equiv. of Cu(NO3)2/ (A3_A3′)Y26C-Terpy. (B) Evolution of the UV−visible spectrum of (A3_A3′)F119C-Terpy after the progressive addition of up to 1.6 equiv Cu(NO3)2 to a 13.5 μM solution of (A3_A3′)Y26C-Terpy in 20 mM MOPS 150 mM NaCl pH 7 at 20 °C. Inset: evolution of the absorbance at 318 nm as a function of the number of equiv of Cu(NO3)2/(A3_A3′)F119C-Terpy. were observed respectively for (A3_A3′)Y26C-Phen and Cu(NO3)2 added (red dots), of the absorbance at 320 nm (A3_A3′)Y26C-Terpy that were in agreement with the which then remains about constant for higher concentrations covalent anchoring of one phenanthroline and one terpyridine of Cu(NO3)2 (blue dots). This is clearly the sign of the per protein (calculated mass 42 565 and 42 702 Da, formation, with a 1:1 stoichiometry, of a (A3_A3′)Y26C- respectively) (Figure 2A). Peaks at 42 585 ± 5 and 42 720 ± Terpy−Cu(II) complex in which the Terpy ligand displayed a 5 Da were also observed in the mass spectra of (A3_A3′)- high affinity for the copper(II) ion. The same experiment was F119C-Phen and (A3_A3′)F119C-Terpy, respectively (Figure performed with (A3_A3′)F119C-Terpy at a concentration of 2B), that were in agreement with the covalent anchoring of 13.5 μM and similar results were obtained (Figure 3B), the one phenanthroline and one terpyridine ligand per protein isosbestic point being at 303 nm and the maximum absorbance (calculated mass 42 581 and 42 718 Da, respectively). wavelength of the bands being 318 and 330 nm in this case. When the functionalization experiment was performed on Moreover, the calculation of molar extinction coefficient of the the A3_A3 protein bearing no cysteine residues, the MALDI- complex at 318 nm resulted in a value of 6500 M−1·cm−1 for ToF MS analysis showed only a peak at 42 346 Da (A3_A3′)Y26C-Terpy−Cu(II) and 9500 M−1·cm−1 for corresponding to the noncoupled protein, which indicated (A3_A3′)F119C-Terpy−Cu(II). These shifts can be caused that the coupling reaction only occurred in the presence of by either the small differences between the spectral properties cysteine residues in the variants and that no other reactive of the two variants, or by a different environment of the group was modified (data not shown). terpyridine moiety inside the (A3_A3′)Y26C and (A3_A3′)- Insertion of Copper(II) into the (A3_A3′)Y26C and F119C pocket. (A3_A3′)F119C Variants Functionalized with Phen and It is noteworthy that when the solution was subjected to -Terpy. The (A3_A3′)Y26C-Phen−Cu(II) and (A3_A3′)- chromatography on a column of molecular sieves, which F119C-Phen−Cu(II) biohybrids did not present any charac- removed the excess of copper nitrate, the A3sA3Y26Terpy− teristic absorbance in the 300−400 nm range that could allow Cu(II) complex was recovered unchanged, as shown by its us to follow the binding of the copper(II) ion to the ligand. On UV−visible spectrum. This confirmed the high affinity of the contrary, for (A3_A3′)Y26C-Terpy−Cu(II) and (A3_A3′)F119C-Terpy−Cu(II) biohybrids, the insertion of (A3_A3′)Y26C-Terpy for Cu(II). However, it was possible to Cu(II) into the terpyridine ligand was accompanied by the remove Cu(II) from the Terpy ligand by incubating the appearance of two bands at 320 and 330 nm assigned to the π purified complex with a large excess of ethylenediaminetetra- → π * transition of the Terpy−Cu(II) complex (Figure 3). It acetic acid (EDTA) (5 mM). was therefore possible to carry out a titration of the Consequently, for further use as catalysts for the D−A (A3_A3′)Y26C-Terpy and (A3_A3′)F119C-Terpy biohybrids reaction, 100 μM solutions of (A3_A3′)F119C-Phen−Cu(II), by increasing amounts of copper(II) nitrate by following the (A3_A3′)Y26C-Phen−Cu(II), (A3_A3′)F119C-Terpy−Cu- absorbance at 320 nm as a function of the Cu(NO3)2/ (II), or (A3_A3′)Y26C-Terpy−Cu(II) in buffer A were (A3_A3′)Y26C-Terpy and (A3_A3′)F119C-Terpy ratio. prepared by adding 1 equiv of copper nitrate (10 mM in In a typical experiment, the effect of the successive additions water) to the corresponding biohybrids solutions, as described of 0.2 equiv of Cu(NO3)2, from 0 to 1.6 equiv, to a 5.6 μM in the Experimental Section. The final solutions were solution of (A3_A3′)Y26C-Terpy in 20 mM MOPS 150 mM chromatographed on molecular sieves prior to use in order NaCl pH 7, was followed by UV−visible spectroscopy (Figure to remove any excess of Cu(II) that could bind nonspecifically 3A). This allowed us to follow the gradual appearance of the to the protein. bands at 320 and 330 nm characteristic of the (A3_A3′)Y26C- Analytic Size−Exclusion Profiles of the Biohybrids. In Terpy−Cu(II) complex and the formation of an isosbestic order to ensure that phenanthroline and terpyridine grafting on point at 306 nm. When plotting the absorbance at 320 nm as a the (A3_A3′)Y26C and (A3_A3′)F119C variants did not function of the number of equiv of Cu(II) added to disrupt the closed conformation of A3_A3 in solution, the (A3_A3′)Y26C-Terpy (Figure 3A, inset), a typical saturation oligomeric states of the biohybrids copper complexes of curve was obtained with a linear increase, till to 1 equiv of (A3_A3′)F119C-Phen, (A3_A3′)Y26C-Phen, (A3_A3′)- 4440 DOI: 10.1021/acsomega.8b03448 ACS Omega 2019, 4, 4437−4447 ACS Omega Article F119C-Terpy, and (A3_A3′)Y26C-Terpy were investigated by with this activity present a high potential for synthetic gel filtration (Figure 4). applications.12 As a consequence, in the last decade, many artificial Diels−Alderases have been reported for the catalysis of the benchmark reaction of azachalcone with cyclo- pentadiene and showed variable endo/exo selectivities and enantioselectivities.7,13−15 The reaction of 2-azachalcone (4) with cyclopentadiene (5) (Scheme 1) was thus investigated in the presence of either the unmodified A3_A3 protein or the (A3_A3′) cysteine mutant- derived biohybrids, in the presence or not of copper(II)− phenanthroline or −terpyridine complexes. It is noteworthy that this reaction can result in up to four isomer products: two major endo enantiomers (6a and 6b) and two minor exo enantiomers (6c and 6d, Scheme 1). Consequently, in all the cases, as the exo products were negligible versus endo products, ee values were calculated for the endo products. In all the cases where an enantiomeric excess was observed, the major enantiomer was the 1S,2R,3R,4R endo product 6a with respect to the 1R,2S,3S,4S endo product 6b as determined by circular dichroism analysis.23 Figure 4. Size-exclusion profiles of the biohybrids. Solutions The D−A reactions were performed under the following containing sample proteins were injected on Superdex 75 10/300 GL column equilibrated in buffer A and elution profiles were recorded conditions: 34 mM 5 and 1 mM 4 were allowed to react in using the absorbance at 280 nm. For all the samples, an excess of buffer A, and no co-solvent added apart from the substrate copper(II) per protein monomer was pre-incubated before injection solution (1% acetonitrile) at 4 °C, in the presence of 30 μM of on the column. Normalized elution profiles are shown for A3_A3 catalyst (Table 1). (black), (A3_A3′)F119C (red), (A3_A3′)Y26C (blue), and the corresponding biohybrids (A3_A3′)F119C-Phen−Cu(II) (red, Table 1. Compared ee, Yields, and Endo/Exo Ratio dashed line), (A3_A3′)F119C-Terpy−Cu(II) (red, dotted line), Measured after the D−A Reaction of Cyclopentadiene 5 (A3_A3′)Y26C-Phen−Cu(II) (blue, dashed line) and (A3_A3′)- with 2-Azachalcone 4 in MOPS Buffer Catalyzed by Either Y26C-Terpy−Cu(II) (blue, dotted line). A3_A3, (A3_A3′)Y26C-Phen, (A3_A3′)Y26C-Terpy, (A3_A3′)F119C-Phen, or (A3_A3′)F119C-Terpy First, the elution volumes of the two variants (A3_A3′)- Biohybrids in the Presence of Copper(II)nitratea F119C (10.84 mL) and (A3_A3′)Y26C (10.72 mL) were very catalyst ee (%) yield (%) close to the elution volume of wild type A3_A3 (10.87 mL) in Cu(NO3)2 0±1 38 ± 2 the presence of Cu(II), indicating that the cysteine mutations A3_A3 15 ± 3 7±1 did not affect the monomeric structure and overall closed A3_A3 + Cu 3±2 22 ± 3 conformation of the single chain bidomain protein. Then, the A3_A3 + Phen−Cu 31 ± 1 9±1 elution volumes observed for the four biohybrids (10.70 mL) A3_A3 + Terpy−Cu 19 ± 1 7±1 for (A3_A3′)F119C-Phen−Cu(II), 10.77 mL for (A3_A3′)- (A3_A3′)Y26C-Phen 12 ± 1 8±2 Y26C-Phen−Cu(II), 10.76 mL for (A3_A3′)F119C-Terpy− (A3_A3′)Y26C-Terpy 16 ± 1 9±2 Cu(II), and 10.78 mL for (A3_A3′)Y26C-Terpy−Cu(II) (A3_A3′)F119C-Phen NDb 4±1 remained very similar reflecting a very close hydrodynamic (A3_A3′)F119C-Terpy 16 ± 1 6±1 radius value for all the proteins. This result showed that the (A3_A3′)Y26C-Phen−Cu 22 ± 1 21 ± 2 closed form of the A3_A3 protein was retained for the four (A3_A3′)Y26C-Terpy−Cu 14 ± 1 16 ± 2 biohybrids and that the covalent attachment of both Terpy− (A3_A3′)F119C-Phen−Cu 5±1 38±2 Cu(II) and Phen−Cu(II) complexes to the (A3_A3′)Y26C (A3_A3′)F119C-Terpy−Cu 52 ± 1 15 ± 1 and (A3_A3′)F119C variants did not alter the overall a In all experiments, 30 μM monomer protein was used after conformation of the protein. incubation with 1 equiv of copper(II)nitrate. bee value could not be Catalysis of a Diels−Alder Reaction. The Diels−Alder determined in this case due to the low yield. (D−A) cycloaddition is a very useful carbon−carbon bond- forming reaction for the synthesis of organic building blocks, and it has been widely studied.9,10 Very few enzymes are Under these conditions, in the absence of copper(II) nitrate, known to catalyze this reaction,11 therefore artificial enzymes very low yields were obtained, endo products 6a−b being Scheme 1. Diels−Alder (D−A) Reaction 4441 DOI: 10.1021/acsomega.8b03448 ACS Omega 2019, 4, 4437−4447 ACS Omega Article Table 2. Selected GoldScore Values for the Docking of Metal Free Cofactor, Metalated Cofactor + Substrate, and Metalated Cofactor + Endo-6a and 6b Productsa (A3_A3′) variant isomer products yieldexp Fmax Fmax (Cu + 4) eeexp (%) Fmax (Cu + 6) ΔFmax F119C-Phen 6a 38 ± 2 26.6 (9/50, I) 30.0 (45/50, II) 5 55.1 3.8 6b 51.3 Y26C-Phen 6a 21 ± 1 26.6 (6/50, IV) 29.8 (46/50, I) 22 33.6 5.6 6b 28.0 F119C-Terpy 6a 15 ± 1 40.3 (33/50, III) 42.0 (23/50, I) 52 61.7 7.2 6b 54.5 Y26C-Terpy 6a 16 ± 2 41.0 (21/50, II) 37.9 (1/50, IIX) 14 54.1 5.2 6b 48.9 a The data are compared with relevant experimental yield and ee data. In parenthesis, the position in ranking and the population of the specific cluster are reported. obtained in about 7 ± 1% yield, even in the presence of A3_A3 (2) ligand exchange leading to the formation of the co- with an enantiomeric excess of 15 ± 3% (Table 1). Copper(II) substrate-bound cofactor [docking of Cu(cofactor)(4)], and nitrate alone was found to be able to catalyze the D−A (3) metal containing cofactor bound to endo-6a, -6b and exo- reaction, a 38 ± 2% yield in products 6 being obtained in the 6c, -6d D−A products to model the final step of the reaction presence of 30 μM Cu(NO3)2, but no enantiomeric excess was [docking of Cu(cofactor)(6)]. Because transition states full observed. When both A3_A3 and copper(II) nitrate were characterization are not achievable by docking methods, and associated, a 22 ± 1% yield was obtained, but with a negligible accurate QM/MM calculations are out of the scope of the enantiomeric excess of about 3 ± 2%. When Phen−Cu and present paper, simulation with product-like geometry were Terpy−Cu were noncovalently inserted into A3_A3, to expected to provide the main information concerning protein− generate the (A3_A3)-Phen−Cu and (A3_A3)-Terpy−Cu ligand recognition factors that could define the stereospecificity biohybrids following the “host−guest” strategy, low yields in of the biohybrid enzymes.24 Details on the results and their endo products were formed (resp. 7 and 9 ± 1%) but higher ee analysis are reported in the Supporting Information. of respectively 31 ± 1 and 19 ± 1% were measured. When comparing the docking on the different structural When the reactions were run in the presence of either the models of the reaction taking place with the biohybrids, four biohybrids, (A3_A3′)Y26C-Phen or -Terpy, and interesting structural tendencies appear. The calculations on (A3_A3′)F119C-Phen or -Terpy, but in the absence of metal-free Phen and Terpy biohybrids show substantial copper(II), rather low yields ranging from 6 ± 1 to 9 ± 1% cofactor flexibility in the low energy solutions with the were obtained. cofactor being able to go from subpockets to different ones for When the (A3_A3′)Y26C derived biohybrids were used in (A3_A3′) F119C-Phen and for (A3_A3′)F119C-Terpy. The the presence of copper(II), quite similar results were observed, cofactor goes from the subpockets involving residues S147, (A3_A3′)Y26C-Phen−Cu leading both to slightly higher yield W143, D186, Q146, Y211′, W235′, D242′, Q246′ and W143, (21 ± 2%) and enantiomeric excess (22 ± 1%) than Q146, S147, Y26′, G29′, K30′, W50′, A54′, A58′, Q61′, Q246′ (A3_A3′)Y26C-Terpy−Cu (16 ± 2% yield and 14 ± 1% ee). to subpockets involving residues A116, I115, F112, Q34, A116, When the reactions were catalyzed by (A3_A3′)F119C V88 and W81, W143, W81′, Q84′, V88′, S85′, R53′, D57′, derived biohybrids, two different results were observed. First, W111′, F112′, I115′, for (A3_A3′)F119C-Phen and when (A3_A3′)F119C-Phen−Cu(II) was used as catalyst, a (A3_A3′)F119C-Terpy, respectively. This scattering is though higher yield in product 6 (38 ± 2%) was obtained but almost largely reduced upon the loading of the metal and its aquo no enantiomeric excess was detected (5 ± 1%). On the ligands leading to low energy solutions with the cofactors and contrary, when the Phen ligand was replaced by the more Cu(II) directed toward the center of the hydrophobic binding bulky ligand terpyridine (A3_A3′)F119C-Terpy−Cu(II), the pocket (Table S1C). The major interactions for the lower enantiomeric excess increased to reach 52 ± 1% but the yield energy solutions are between the water molecules and side in product 6 decreased to 15 ± 1%. chains of S85′, D57′, R53′, and R238′ as well as hydrophilic Molecular Modeling. A computational approach was contacts of the cofactor with W143, Q146, W50′, Y26′ , and subsequently performed to rationalize these data. Our strategy W143, Q146, W235′, Q246′, V273′ for (A3_A3′)F119C- consisted in (i) building the two models of the biohybrid Phen−(H2O)2 and (A3_A3′)F119C-Terpy−(H2O)2, respec- variants (A3_A3′)Y26C and (A3_A3′)F119C; (ii) character- tively (see Supporting Information). These low energy izing throughout density functional theory (DFT) calculations orientations are consistent with those obtained when the 2- on cluster models the geometries of intermediates and azachalcone substrate binds to the copper showing that the products along the copper mediated D−A reaction; and (iii) metalated biohybrids present a good pre-organization of the inserting those geometries into the variant models by binding site for the first co-substrate. Finally, docking optimized protein−ligand docking. The complementarity of simulations with the D-A products endo-6a, -6b, exo-6c, and the catalytically relevant structures into the context of the -6d all show good complementarities with the binding site catalysis was then discussed. More concretely, the modeling (more than 50.0 GoldScore units) though at this state we can consisted in building the biohybrid structures of the metal free already observe that endo-6a products have better comple- cofactors covalently linked to (A3_A3′)Y26C and (A3_A3′)- mentarity values than endo-6b with at least 5.2 GoldScore F119C following the three different reaction steps: (1) the fitness units (Table S2C). resting state of the artificial metalloenzymes produced by In order to better understand the key recognition factors metalation of the hybrids (docking of Cu(cofactor)(H2O)2), involved in the enantiomeric selectivity, the dockings for the 4442 DOI: 10.1021/acsomega.8b03448 ACS Omega 2019, 4, 4437−4447 ACS Omega Article (A3_A3′)F119C-Phen−Cu-6a and 6b (5% ee) and (A3_A3′)- 2%). On the contrary, in the case of (A3_A3′)F119C-Terpy− F119C-Terpy−Cu-6a and 6b (52% ee) systems were repeated Cu(2-azachalcone), the bulkier maleimide linker and Terpy considering the flexibility in solution of the side chains in the ligand lead to stronger van der Waals stabilizations of the binding pocket. All the data taken together could be complex with R53, Q84, S85, V87, W81, F112, I115, W143, rationalized considering the steric hindrance of the binding W81′, Q84′, V88′, F112′, I115′, and W143′ that hinder its sites and the stabilization of the products with a transition flexibility and the access of the cyclopentadiene to one side of state-like geometry (Table 2). the pocket (Figure 5c). Moreover, the docking analysis of In the case of (A3_A3′)F119C-Phen−Cu-(2-azachalcone), (A3_A3′)F119C-Terpy−Cu(endo-6a/6b) shows an equiva- it appeared that the Phen ligand and the substrate were lent position of the terpyridine implying the specular surrounded by a wide pocket showing van der Waals contacts orientation of the two products that leads to favor the 6a with V88, W143, Q84′, W99′, F112′, I115′, and W143′ enantiomer stabilization, in particular the interactions of the (Figure 5a). The relatively open pocket and the limited substrates with Q84, F112, I115, A116, W143, Q84′, V88′, F112′, I115′, F119′, and W143′ (ΔFmax = 7.2, Figures 5d and S2C). These two observations could explain both the increase in the enantiomeric excess (5−52%) and the decrease in the yield (38−15%). Concerning the (A3_A3′)Y26C variant, the docking analysis shows a significant stabilization of the intermediate Cu- (cofactor)(2-azachalcone) with residues Q61, Q84, V88, A116, I115, F119, W81′, Q83′, I115′, W143′, S148′ and Q84, V88, F112, A116, W111, I115, F119, W143, Q85′, V88′, I115′, A116′, F119′ for (A3_A3′)Y26C-Phen−Cu(4) and (A3_A3′)- Y26C-Terpy−Cu(4), respectively. The preferential formation of the 6a enantiomer is predicted obtaining fitness differences of 5.6 and 5.2 GoldScore units for Phen and Terpy variants, respectively. In analogy with the previous case, the (A3_A3′)- Y26C-Terpy−Cu(2-azachalcone) shows both a larger pocket and a higher variability in docking solutions. Accordingly, a yield in product 6 of 16 ± 2% and an ee of 14 ± 1% were Figure 5. Representation of best docking solutions for the reactants found for (A3_A3′)Y26C-Phen−Cu(II) and a 21 ± 2% yield (A3_A3′)F119C-Phen−Cu(4), (A3_A3′)F119C-Phen−Cu(6) and in product 6 for (A3_A3′)Y26C-Phen−Cu(II) with an ee of (A3_A3′)F119C-Terpy−Cu(4), (A3_A3′)F119C-Terpy−Cu(6) unit 22 ± 1%. Both species are stabilized by van der Waals contacts showing their environment. The linker is depicted in green and the with V88, Q84, Q61, V88′, I115′, F119′, W143′, S147′, D150′ side chains that change their conformation during binding are for (A3_A3′)Y26C-Phen−Cu(6a) and V88, Q84, W81, F119, highlighted in pink. W143, W81′, Q84′, V88′, A116′, F119′, W143′ for (A3_A3′)- Y26C-Terpy−Cu(6a) (Figure S1C). number of interacting residues could lead to a moderate Taken all together, the computation results suggest possible control of the relative positioning of both complex and further improvements to our system by, for example, stabilizing substrates, in fact, for (A3_A3′)F119C-Phen−Cu(endo-6), the intermediates with bulkier hydrophobic residues in their both isomers reach binding modes with similar scoring (ΔFmax direct vicinity or hider the reactants access to one side of the = 3.8, Table 2). These considerations could explain both the binding pocket favoring the selective formation of one low ee observed as well as the higher yield in product 6 (38 ± enantiomer. It is noteworthy that no significant effect of the Table 3. Compared Yields and ee Measured after the D−A Reaction of Cyclopentadiene 5 with 2-Azachalcone 4 in the Presence of Various Copper(II)-Biohybrid Catalysts Reported in the Literature biohybrid catalyst protein−liganda conversion/yield (%) endo/exo (%) ee (%) refs b LmrR_M89C-Phen 93 95:5 97 7 SB-Lys-GTL*/196-Phenb 98 93:7 25 17 HSA-phtalocyanin 89 91:9 85 18 NCS-Phenb 84 84:16 0 10,16 FhuA-NHCc 62 90:10 ND 19 FhuA-Tpyd 15 66:34 ND 10 A2A-Phend 28 82:18 19 14 NB-Tpyd 22 90:10 ND 6 A3F119C-Phenb (A3F119NPH) 11 86:14 62 3 (A3_A3′)F119C-Terpy 15 93/7 52 this work (A3_A3′)F119C-Phen 38 92/8 5 a Nomenclature for the various proteins used: LmrR, Lactococcal multidrug resistance regulator; SB-Lys-GTL*/196, Sepabeads Lys196 mutant of Geobacillus thermocatenulatus lipase; HSA, human serum albumin; A3, αRep A3 protein; SCP-2L, sterol carrier protein type 2-like domain; NCS, neocarzinostatin; NB, nitrobindin; FhuA, ferric hydroxamate uptake protein component A. A2A, A2 adenosine receptor. bPhen, phenanthroline. c NHC, N-heterocyclic carbene. dTpy, terpyridine. 4443 DOI: 10.1021/acsomega.8b03448 ACS Omega 2019, 4, 4437−4447 ACS Omega Article catalysis on regioselectivity was observed with either of the two also successful to increase the yield of the reaction because catalytic Cu(II) complexes in the A3_A3′ mutants as in all the with (A3_A3′)F119C-Phen−Cu(II) as catalyst a 38 ± 2% cases the endo/exo ratios were similar (92 ± 1:8 ± 1). yield could be obtained, but unfortunately in this case a very ■ DISCUSSION The characteristics of the D−A reaction catalyzed by weak ee was found (5 ± 1%) as a result of both a bad control of the relative positioning of the two substrates in the wide pocket surrounding the Phen−Cu(II) cofactor and a too high (A3_A3′)F119C-Terpy− and -Phen−Cu(II), yield, endo/exo flexibility of this cofactor. A better ee could be obtained with ratio, and enantiomeric excess, ee, are compared in Table 3 (A3_A3′)F119C-Terpy−Cu(II) (52 ± 1%), but this time a with those reported to date for the D−A reaction between 4 low yield was found (15 ± 1%) resulting both from a lower and 5 catalyzed by various artificial metalloenzymes. flexibility of the bulky Terpy−Cu(II) cofactor and from a Considering the endo/exo ratio, (A3_A3′)F119C-Terpy− smaller sized pocket surrounding it, allowing a better control of Cu(II) and (A3_A3′)F119C-Phen−Cu(II) (endo/exo = 93/7 the relative positioning of the two substrates. These results and 92/8 resp.) stand among the systems that follow the together with those reported earlier3 put in light that a good traditional endo rule with LmrR_M89C-Phen−Cu,7 SB-Lys- compromise between flexibility of the cofactor and size of the GTL*/196-Phen−Cu 17 (endo/exo = 95/5 and 93/7, pocket surrounding it remain to be found to improve both the respectively), whereas FhuA-Tpy favored the formation of yield and the enantioselectivity of the reaction. Another the exo products (endo/exo = 66/34).10 Concerning the yields important factor that has probably to be taken into account is in product 6, the best one that was obtained in this work, 38% the accessibility of the substrates to the catalytic pocket, as with (A3_A3′)F119C-Phen−Cu, was lower than those until now the A3 and A3_A3-derived biohybrids generally lead obtained with the most efficient systems LmrR_M89C- to much lower yields than some reported in the litera- Phen−Cu,7 SB-Lys-GTL*/196-Phen−Cu,17 and HSA-phtalo- ture.7,16−18 Indeed, the conformational study and ligand cyanin−Cu18 that ranged between 89 and 98%, but were binding study of A3_A3 shed light on the very tightly higher than those reported for NB-Tpy-Cu6 and FhuA-Tpy- stabilized closed conformation in the bidomain single chain Cu19 (22 and 15%, respectively). Finally, the best ee obtained, protein.5 Our results demonstrated that artificial metal- 52% with (A3_A3′)F119C-Terpy−Cu, was also lower than loenzymes based on the A3_A3 protein modified with nitrogen those reported with LmrR_M89C-Phen (ee = 97%)7 and donor ligands may be suitable for further catalyst optimization HSA-phtalocyanin (ee = 85%)18 but higher than that reported and constitute valuable tools toward more efficient and with SB-Lys-GTL*/196-Phen (ee = 25%).17 We recently selective artificial biocatalysts (Table 3). published the production of a NCS−phenanthroline−Cu(II) artificial metalloenzyme, by the non-covalent insertion of a testosterone−phenanthroline−Cu(II) conjugate into the bind- ■ EXPERIMENTAL SECTION Plasmid Construction. The genetic sequence coding for ing site of NCS following the “Trojan Horse” strategy.16 This the protein αRep A3 was issued from a combinatorial library of biohybrid was found able to catalyze the D−A reaction with a concatenated, repeated, and degenerated sequences. In order very good yield, 84%, but with no enantiomeric excess, which to mutate the first domain only, repetitions in nucleotide could be explained both by the wide binding site of NCS open sequences were minimized to simplify site-directed muta- to the solvent, with no control of the relative positioning of the genesis. The sequence was therefore recoded using different two D−A substrates, and the high flexibility of the catalytic codons for the same amino acid in consecutive repeats. The Phen−Cu(II) moiety due to its noncovalent anchoring to the new nucleotide sequence was designed using the GenScript protein. We then reported a new artificial Diels−Alderase algorithm dedicated to codon optimization for E. coli A3F119NPH-Cu(II) that was obtained by covalent anchoring expression as previously described.3 The bidomain genetic of a copper(II)−phenanthroline moiety into the F119C construction was recently described.5 Briefly, the synthetic mutant of an artificial protein dimer based on a thermostable gene of A3 was cloned by CPEC into pQE81 plasmid α-helical repeated motif.3 In this case, a good enantioselectivity containing the original repeated sequence of A3 gene (named (62% ee) was obtained but with the yield in product 6 was low A3′ in the mutated bidomain construction). The two genes (11%), which could be explained by the presence of a were separated by a linker sequence coding for repeated serine copper(II)bisphenanthroline complex, due to the close and glycine residues (SGGGGSGGGG). The gene also proximity of the two phenanthroline ligands covalently included a tobacco etch virus (TEV) protease cleavage site attached to the two Cys119 residues facing each other in the after the N-terminal hexahistidine tag. The final construct then symmetrical dimer, which blocked the access of the two encoded HisTag-TEVsite-A3-(SG4)2-A3′. substrates to the catalytic center.3 Consequently, in order to Directed Mutagenesis. Genes coding for the variants avoid this major drawback, we proposed to associate in a single (A3_A3′)F119C and (A3_A3′)Y26C were obtained by site- chain protein a single cysteine mutated A3 domain with directed mutagenesis using the QuikChange protocol with the another nonmutated A3 domain, in order to ensure to have a following primers (F119C-FOR: 5′-CGGCGGCGtgcGCAT- 1:1 copper(II)−phenanthroline complex as catalytic center TAGG; F119C-REV: 5′-CCTAATGCgcaCGCCGCCG; and to enlarge the catalytic pocket in order to accommodate Y41C-FOR: 5′-GCAGCGGCAtgcGCGCTGGG; and Y41C- both D−A substrates. The results presented in this paper show REV: 5′-CCCAGCGCgcaTGCCGCTG). The vector contain- that this strategy was successful for the production of four of ing the A3_A3′ gene was amplified by PCR with Phusion high- such new artificial metalloenzymes, (A3_A3′)Y26C and fidelity DNA polymerase (Thermo Scientific) as follows: initial (A3_A3′)F119C derived with Phen−Cu(II) and -Terpy− denaturation at 98 °C for 1 min; 20 cycles including a Cu(II), that were obtained by the covalent anchoring of either denaturation step at 98 °C for 20 s, an annealing step at 65 °C a copper(II)−phenanthroline cofactor or a copper(II)− for 15 s, and an extension step at 72 °C for 3 min. The final terpyridine cofactor to either a F119C mutant or a Y26C extension step was done at 72 °C for 10 min. The PCR mutant of the single chain A3_A3 protein. The strategy was product was treated with Dpn I and directly used for 4444 DOI: 10.1021/acsomega.8b03448 ACS Omega 2019, 4, 4437−4447 ACS Omega Article transformation in E. coli XL1-Blue electrocompetent cells. For Monitoring of Thiols Accessibility. The accessibility of each mutagenesis, plasmids from five independent clones were thiol groups of Cys119 and Cys26 in both variants was purified using Nucleospin Plasmid extraction kit (Macherey monitored using Ellman’s test.20 In a standard assay, 445 μL of Nagel). The sequences of the variants were confirmed by reaction buffer (0.1 M sodium phosphate pH 8.0, 1 mM sequencing both DNA strands. EDTA) was mixed with 5 μL of reagent solution [10 mM 5,5′- Protein Production. Chemocompetent E. coli M15 dithio-bis-(2-nitrobenzoic acid)] in a quartz cuvette. Absorb- (Qiagen) bacteria were transformed with the desired plasmid ance at 412 nm was monitored using a JASCO spectropho- and grown on 2YT-agar plates supplemented with ampicillin tometer. Absorbance was recorded in the kinetic mode after (100 μg/mL), kanamycin (50 μg/mL), and 1% glucose. One the addition of 50 μL of a 100 μM protein solution until clone was picked and grown overnight at 37 °C in liquid 2YT complete reaction. The final absorbance obtained at the medium supplemented with ampicillin (100 μg/mL) and plateau was noted Aeq and the initial absorbance measured kanamycin (50 μg/mL). The overnight preculture was used to before protein addition was noted A0. The reduced cysteine inoculate 1 L of fresh medium distributed into two 2 L flasks. content was then calculated from the difference signal Cells were grown at 37 °C until OD600nm reached 0.8, protein corrected by the dilution factor using an εM value of 14 150 expression was induced by the addition of 500 μM IPTG and M−1·cm−1 at 412 nm. cultures were further incubated at 37 °C for 4 h. Cells were Ligand Synthesis. Synthesis of 2-Bromo-N-(1,10-phe- harvested by centrifugation for 20 min at 5000g at 4 °C and nanthrolin-5-yl)acetamide Hydrobromide (1). Compound resuspended into 40 mL of washing buffer (50 mM sodium (1) Was Synthetized According to Literature.7 1,10- phosphate pH 8; 300 mM NaCl; 20 mM imidazole). phenanthrolin-5-amine (195 mg, 1 mmol) was solubilized in Suspended cells were stored at −80 °C. a 50 mL two-necked round-bottom flask filled with 30 mL of Protein Purification. Frozen cells from 1 L of bacterial anhydrous CH2Cl2 and equipped with magnetic stirring and a culture were thawed and a tablet of EDTA-free anti-protease condenser. Bromoacetyl bromide (104 μL, 1.2 mmol) was cocktail (Roche) was added. Cells were lysed by three 30 s diluted in 10 mL anhydrous CH2Cl2 and added dropwise to cycles of sonication on ice and centrifuged 30 min at 11 000g the phenanthroline solution. The mixture was then refluxed at 4 °C. His-tagged proteins were purified from the crude overnight and the resulting suspension was filtered. Recrystal- supernatant using nickel-affinity chromatography on a 4 mL lization was performed in MeOH, leading to the HBr salt of Ni−NTA column (Macherey Nagel). The column was washed compound (1) as brownish crystals (yield 70%). 1H NMR with 150 mL of washing buffer (50 mM sodium phosphate pH (300 MHz, DMSO-d6): δ 10.86 (s, 1H), 9.32 (dd, J = 4.5, 1.2, 8; 300 mM NaCl, 20 mM imidazole) and proteins were eluted 1H), 9.21 (dd, J = 5, 1.2, 1H), 9.09 (d, J = 8.5, 1H), 9 (d, J = in six 1 mL fractions in the elution buffer (50 mM sodium 8.5, 1H), 8.54 (s, 1H), 8.20 (t, J = 4.3, 1H), 8.17 (t, J = 4.3, phosphate pH 8; 300 mM NaCl, 300 mM imidazole). A typical 1H), 4.34 (s, 2H) ppm. yield of about 100 mg of protein per liter of culture was Synthesis of 2-(2,2′:6′,2″-Terpyridine-4′-yloxy) Ethyl- obtained. Eluted fractions were analyzed by SDS-PAGE and amine (2). Compound (2) Was Synthetized According to pure fractions of proteins were pooled, diluted five times with Literature.21 4′-Chloro-2,2′:6′,2″-terpyridine (159.7 mg) and 50 mM sodium phosphate buffer pH 8, 150 mM NaCl, and ethanolamine (60 μL) were stirred in a suspension of finally cleaved with His-tagged TEV protease obtained from powdered KOH (100 mg) in DMSO (2 mL) at 40 °C for 2 the pRK793 vector (Addgene): purified His-tagged TEV h. The reaction mixture was then added to 10 mL of CH2Cl2 protease and purified αRep bidomain protein with its cleavable and washed three times with water. The CH2Cl2 solution was His-Tag were mixed in a 1:100 ratio in a total volume of 50 mL dried over MgSO4 and solvents evaporated. 2-(2,2′:6′,2″- and incubated overnight at 10 °C after the addition of 0.5 mM Terpyridine-4′-yloxy) ethylamine (145 mg) was obtained as a EDTA and 1 mM DTT. The mixture was then diluted twice light-yellow solid in 83% yield and used subsequently without with 50 mM sodium phosphate buffer pH 8, 150 mM NaCl further purification. 1H NMR (360 MHz, CDCl3): δ 3.17 (t, J leading to a final 30 mM imidazole concentration and loaded = 5.11 Hz, 2H), 4.27 (t, J = 5.11 Hz, 2H), 7.33 (dd, J1 = 4.77 on another 4 mL Ni−NTA column equilibrated with the Hz, J2 = 0.92 Hz, 2H), 7.86 (td, J1 = 11.67 Hz, J2 = 2 Hz, 2H, washing buffer. His-tagged TEV protease and uncleaved His- H4,4″), 8.03 (s, 2H), 8.61 (d, J = 7.96 Hz, 2H), 8.69 (d, J = 3.85 tagged αRep were retained on the column, while the flow Hz, 2H) ppm. ESI-MS m/z: 293.13 (M), 315.12 (M + Na+). through containing cleaved αRep protein was recovered, and Synthesis of N-[2-([2,2:6′,2″-Terpyridin]-4-yloxy)ethyl]- concentrated resulting in a 5 mL solution that was loaded on a maleimide (3). 2-([2,2′:6′,2″-Terpyridin]-4′-yloxy)ethylamine gel filtration column (HiLoad 16/600 Superdex 75, GE (2) (136.4 mg) in 5 mL of acetone/water (2/1) was added Healthcare) equilibrated with filtered coupling buffer (50 dropwise to a solution of NaHCO3 (118 mg) and N- mM sodium phosphate pH 7.75, 150 mM NaCl). The methoxycarbonylmaleimide (127 mg) in 5 mL of acetone/ coupling buffer was previously incubated for 1 h with Chelex water: 2/1 at 4 °C, and the reaction mixture was stirred for 1 h 100 resin (Bio-Rad). Elution fractions after gel filtration were 30 min. The solution was added to water (3 mL) and stirred pooled and stored at 4 °C until chemical coupling. Purity was for 2 h 30 min at room temperature. The obtained precipitate checked by SDS-PAGE and concentrations were obtained was collected and washed with water. Yield: 158.2 mg (80%); from the absorbance at 280 nm measured on a Nanodrop 1 H NMR (360 MHz, CDCl3): δ 8.68 (d, J = 4.83 Hz, 2H), spectrophotometer (Thermo Scientific) using extinction 8.60 (d, J = 8.10 Hz, 2H), 7.98 (s, 2H), 7.84 (td, J = 7.72, 1.88 coefficients at 280 nm calculated from the protein sequence Hz, 2H), 7.34 (dd, J = 4.82, 1.22 Hz, 2H), 6.75 (s, 2H), 4.40 (https://web.expasy.org/protparam/) (58 900 M−1·cm−1 for (t, J = 5.57 Hz, 2H), 4.03 (t, J = 5.55 Hz, 2H) ppm. ESI-MS A3_A3, 58 900 M−1·cm−1 for (A3_A3′)F119C and 57 410 m/z: = 373.13 (M), 395.11 (M + Na+). M−1·cm−1 for (A3_A3′)Y26C). After the cleavage step, a Ligands Coupling. In both protein variants, the typical yield of 40 mg of pure protein ready for the coupling concentration of thiol groups obtained via Ellman’s test was reaction was obtained. used as the cysteine concentration and thus of protein, 4445 DOI: 10.1021/acsomega.8b03448 ACS Omega 2019, 4, 4437−4447 ACS Omega Article available for either phenanthroline or terpyridine coupling. sulfur atom of the mutated cysteines was used for each Then, 1.2 equiv of compound 1, or 2 equiv of compound 3 simulation. In all the calculations, the Trojan horses and the were diluted in DMSO and mixed with 100 μM protein linker were treated as flexible structures. solution in coupling buffer (50 mM Na-phosphate pH 7.75, The solutions were analyzed by means of GaudiView GUI 150 mM NaCl) for a final DMSO concentration of 1% in the extension for Chimera27 taking into account three main desired volume. The mixture was incubated at 4 °C overnight. criteria, the maximum value of the GoldScore Fitness (Fmax), The coupling reaction was then stopped by the addition of 0.5 the position in the global energy ranking, and the population of mM EDTA and 1 mM DTT. The protein solution was the solution clusters. concentrated to about 5 mL using a Vivaspin concentrator and General Procedure for the Catalysis of Diels−Alder loaded on a gel filtration column (HiLoad 16/600 Superdex Cyclization Reactions. All solutions were kept on ice and 75, GE Healthcare) equilibrated in buffer A (20 mM MOPS reaction runs were performed in duplicate in 2 mL Eppendorf 150 mM NaCl, pH7) previously incubated with Chelex 100 conical tubes. In these tubes, the following were added resin and filtered. The coupling efficiency was measured using successively: 150 μL of the 100 μM solution of copper(II)- Ellman’s test and MALDI-ToF MS. The molar extinction biohybrid prepared previously (final concentration 30 μM), 5 coefficient at 280 nm was calculated by adding the calculated μL of a stock solution of 100 mM azachalcone 4 (final molar extinction coefficient of protein variants with the one concentration 1 mM, 33 equiv), 5 μL of a 187 mM measured for the phenanthroline or the terpyridine moiety. diphenylsulfone solution used as an internal standard (final The latter was determined in a 70% ethanol solution of a concentration 1.87 mM) in acetonitrile, and finally 1.4 μL of known concentration of 1,10-phenanthroline and terpyridine, pure freshly distilled cyclopentadiene 5 (final concentration 34 giving a molar extinction coefficient of 22 4281 and 8000 M−1· mM, 1133 equiv) into buffer A for a final volume of 500 μL. cm−1, respectively. The tubes were then placed in the ThermoMixer comfort and Mass Spectrometry Analysis. Protein solutions were stirred at 500 rpm for 48 h at 4 °C. Ethyl acetate (3 × 500 μL) desalted on a Zeba Spin column (Thermo) against 10 mM was used for extracting the reaction mixture. After NH4HCO3 buffer pH 8 and then diluted in a sinapinic acid centrifugation, organic layers were collected and dried over matrix for MALDI-ToF MS analysis in linear mode by using a MgSO4. HPLC analyses were then performed at 20 °C on an MALDI-ToF/ToF UltrafleXtreme spectrometer (Bruker Dal- Agilent Technologies 1260 Infinity apparatus equipped with a tonics). Lux Cellulose-1 (250 × 4.6 mm, 5 μm) and n-hexane/ Metallation of the apo-Biohybrids. Aqueous Cu(NO3)2· isopropanol (99/1) as the mobile phase at a 0.5 mL·min−1 flow 3H2O (1 μL, 10 mM) was added to 100 μL of 100 μM rate (DAD: 270 nm for 6 and diphenylsulfone, 300 nm for 4). solution of either A3_A3, (A3_A3′)F119C-Phen, (A3_A3′)- The peaks corresponding to the four stereoisomers were F119C-Terpy, (A3_A3′)Y26C-Phen or (A3_A3′)Y26C-Terpy identified by comparison with those of authentic samples: in buffer A and gently yet thoroughly mixed by pipetting in retention times: 15.4 and 17.0 min (6 exo isomers), 18.7 and order to obtain a 1:1 copper(II)/protein ratio. The solutions 24.6 min (6 endo isomers), 29.1 min (starting material, 4), and were kept on ice in a cold room overnight. 61 min (internal standard, diphenylsulfone). The stereo- Analytical Size Exclusion Chromatography. Analytical isomers were assessed after circular dichroism measurements.7 size exclusion chromatography was done essentially as described7 with an Ä KTA Purifier (GE Healthcare) system on a Superdex 75 10/300 GL column (flow rate 1 mL·min−1) ■ * ASSOCIATED CONTENT S Supporting Information equilibrated with buffer A. Solutions containing sample The Supporting Information is available free of charge on the proteins were injected and elution profiles were recorded by ACS Publications website at DOI: 10.1021/acsome- following the absorbance at 280 nm. ga.8b03448. Molecular Modeling. The geometry of the Trojan horse Molecular modeling; GoldScore values for different models (cofactor, substrate bound cofactor, and product docking simulations; and Cartesian coordinates of the models) were optimized with Gaussian 09 (Gaussian 09, DFT optimized structures (PDF) ■ revision D.01. Gaussian, Inc.: Wallingford, CT, 2010) using the wave function B3LYP-D322 and the SMD continuum model for water23 (see Supporting Information for further details). AUTHOR INFORMATION The dimeric bio-assembly of the X-ray structure of the Corresponding Authors A3_A3 bidomain protein (PDB ID 6FT5) was used to model *E-mail: jean-pierre.mahy@u-psud.fr (J.-P.M.). the different biohybrid variants (A3_A3′)F119C and *E-mail: agathe.urvoas@u-psud.fr (A.U.). (A3_A3′)Y26C by mutating the relevant side chains. The *E-mail: remy.ricoux@u-psud.fr (R.R.). positions of the bidomain A3_A3′ protein were numbered ORCID starting from residue number 3, as the first two residues of the Giuseppe Sciortino: 0000-0001-9657-1788 sequence were lacking in the crystal structure. The numbers of Rémy Ricoux: 0000-0002-3464-7895 the mutated positions were thus in accordance with those described previously for the A3 homodimer.3 All the structural Author Contributions ∥ handling performed at the beginning of the study was carried T.D.M. and K.K. contributed equally to this work. out by means of UCSF Chimera embedded software.24 Notes Covalent protein−ligand docking assays were performed The authors declare no competing financial interest. with the program GOLD (version 5.2)25 using the GoldScore scoring function26 and considering the bond between the cofactor and the sulfur of the mutated cysteine as the covalent ■ REFERENCES (1) Schwizer, F.; Okamoto, Y.; Heinisch, T.; Gu, Y.; Pellizzoni, M. anchoring. An evaluation sphere of 20 Å radius centered on the M.; Lebrun, V.; Reuter, R.; Köhler, V.; Lewis, J. C.; Ward, T. R. 4446 DOI: 10.1021/acsomega.8b03448 ACS Omega 2019, 4, 4437−4447 ACS Omega Article Artificial Metalloenzymes: Reaction Scope and Optimization (20) Eyer, P.; Worek, F.; Kiderlen, D.; Sinko, G.; Stuglin, A.; Strategies. Chem. Rev. 2018, 118, 142−231. Simeon-Rudolf, V.; Reiner, E. Molar Absorption Coefficients for the (2) Urvoas, A.; Guellouz, A.; Valerio-Lepiniec, M.; Graille, M.; Reduced Ellman Reagent: Reassessment. Anal. Biochem. 2003, 312, Durand, D.; Desravines, D. C.; van Tilbeurgh, H.; Desmadril, M.; 224−227. Minard, P. Design, Production and Molecular Structure of a New (21) Maayan, G.; Yoo, B.; Kirshenbaum, K. Heterocyclic Amines for Family of Artificial Alpha-helicoidal Repeat Proteins (αRep) Based on the Construction of Peptoid Oligomers Bearing Multi-Dentate Thermostable HEAT-like Repeats. J. Mol. Biol. 2010, 404, 307−327. Ligands. Tetrahedron Lett. 2008, 49, 335−338. (3) Di Meo, T.; Ghattas, W.; Herrero, C.; Velours, C.; Minard, P.; (22) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. A Consistent and Mahy, J.-P.; Ricoux, R.; Urvoas, A. ΑRep A3: A Versatile Artificial Accurate Ab Initio Parametrization of Density Functional Dispersion Scaffold for Metalloenzyme Design. Chem.Eur. J. 2017, 23, 10156− Correction (DFT-D) for the 94 Elements H-Pu. J. Chem. Phys. 2010, 10166. 132, 154104. (4) Guellouz, A.; Valerio-Lepiniec, M.; Urvoas, A.; Chevrel, A.; (23) Marenich, A. V.; Cramer, C. J.; Truhlar, D. G. Universal Graille, M.; Fourati-Kammoun, Z.; Desmadril, M.; van Tilbeurgh, H.; Solvation Model Based on Solute Electron Density and on a Minard, P. Selection of Specific Protein Binders for Pre-Defined Continuum Model of the Solvent Defined by the Bulk Dielectric Targets from an Optimized Library of Artificial Helicoidal Repeat Constant and Atomic Surface Tensions. J. Phys. Chem. B 2009, 113, Proteins (AlphaRep). PLoS One 2013, 8, No. e71512. 6378−6396. (5) Léger, C.; Di Meo, T.; Aumont-Nicaise, M.; Velours, C.; (24) Pettersen, E. F.; Goddard, T. D.; Huang, C. C.; Couch, G. S.; Durand, D.; Li de la Sierra-Gallay, I.; van Tilbeurgh, H.; Hildebrandt, Greenblatt, D. M.; Meng, E. C.; Ferrin, T. E. UCSF Chimera?A N.; Desmadril, M.; Urvoas, A.; Valerio-Lepiniec, M.; Minard, P. Visualization System for Exploratory Research and Analysis. J. Ligand-Induced Conformational Switch in an Artificial Bidomain Comput. Chem. 2004, 25, 1605−1612. Protein Scaffold. Sci. Rep. 2019, 9, 1178. (25) Jones, G.; Willett, P.; Glen, R. C.; Leach, A. R.; Taylor, R. (6) Himiyama, T.; Sauer, D. F.; Onoda, A.; Spaniol, T. P.; Okuda, J.; Development and Validation of a Genetic Algorithm for Flexible Hayashi, T. Construction of a Hybrid Biocatalyst Containing a Docking 1 1Edited by F. E. Cohen. J. Mol. Biol. 1997, 267, 727−748. Covalently-Linked Terpyridine Metal Complex within a Cavity of (26) Jones, G.; Willett, P.; Glen, R. C. Molecular Recognition of Aponitrobindin. J. Inorg. Biochem. 2016, 158, 55−61. Receptor Sites Using a Genetic Algorithm with a Description of (7) Bos, J.; Fusetti, F.; Driessen, A. J. M.; Roelfes, G. Desolvation. J. Mol. Biol. 1995, 245, 43−53. Enantioselective Artificial Metalloenzymes by Creation of a Novel (27) Rodríguez-Guerra Pedregal, J.; Sciortino, G.; Guasp, J.; Active Site at the Protein Dimer Interface. Angew. Chem., Int. Ed. Municoy, M.; Maréchal, J.-D. GaudiMM: A Modular Multi-Objective 2012, 51, 7472−7475. Platform for Molecular Modeling. J. Comput. Chem. 2017, 38, 2118− (8) Ellman, G. L. Tissue Sulfhydryl Groups. Arch. Biochem. Biophys. 2126. 1959, 82, 70−77. (9) Funel, J.-A.; Abele, S. Industrial Applications of the Diels-Alder Reaction. Angew. Chem., Int. Ed. 2013, 52, 3822−3863. (10) Gregoritza, M.; Brandl, F. P. The Diels-Alder reaction: A powerful tool for the design of drug delivery systems and biomaterials. Eur. J. Pharm. Biopharm. 2015, 97, 438−453. (11) Klas, K.; Tsukamoto, S.; Sherman, D. H.; Williams, R. M. Natural Diels-Alderases: Elusive and Irresistable. J. Org. Chem. 2015, 80, 11672−11685. (12) Nicolaou, K. C.; Hale, C. R. H.; Nilewski, C.; Ioannidou, H. A. Constructing Molecular Complexity and Diversity: Total Synthesis of Natural Products of Biological and Medicinal Importance. Chem. Soc. Rev. 2012, 41, 5185. (13) Filice, M.; Romero, O.; Gutiérrez-Fernández, J.; de las Rivas, B.; Hermoso, J. A.; Palomo, J. M. Synthesis of a Heterogeneous Artificial Metallolipase with Chimeric Catalytic Activity. Chem. Commun. 2015, 51, 9324−9327. (14) Ghattas, W.; Dubosclard, V.; Wick, A.; Bendelac, A.; Guillot, R.; Ricoux, R.; Mahy, J.-P. Receptor-Based Artificial Metalloenzymes on Living Human Cells. J. Am. Chem. Soc. 2018, 140, 8756−8762. (15) Himiyama, T.; Taniguchi, N.; Kato, S.; Onoda, A.; Hayashi, T. A Pyrene-Linked Cavity within a β-Barrel Protein Promotes an Asymmetric Diels-Alder Reaction. Angew. Chem., Int. Ed. 2017, 56, 13618−13622. (16) Ghattas, W.; Cotchico-Alonso, L.; Maréchal, J.-D.; Urvoas, A.; Rousseau, M.; Mahy, J.-P.; Ricoux, R. Artificial Metalloenzymes with the Neocarzinostatin Scaffold: Toward a Biocatalyst for the Diels- Alder Reaction. ChemBioChem 2016, 17, 433−440. (17) Filice, M.; Romero, O.; Gutiérrez-Fernández, J.; de las Rivas, B.; Hermoso, J. A.; Palomo, J. M. Synthesis of a Heterogeneous Artificial Metallolipase with Chimeric Catalytic Activity. Chem. Commun. 2015, 51, 9324−9327. (18) Reetz, M. T.; Jiao, N. Copper-Phthalocyanine Conjugates of Serum Albumins as Enantioselective Catalysts in Diels-Alder Reactions. Angew. Chem., Int. Ed. 2006, 45, 2416−2419. (19) Osseili, H.; Sauer, D. F.; Beckerle, K.; Arlt, M.; Himiyama, T.; Polen, T.; Onoda, A.; Schwaneberg, U.; Hayashi, T.; Okuda, J. Artificial Diels-Alderase based on the transmembrane protein FhuA. Beilstein J. Org. Chem. 2016, 12, 1314−1321. 4447 DOI: 10.1021/acsomega.8b03448 ACS Omega 2019, 4, 4437−4447