Crystal structures and kinetic studies of a laboratory evolved aldehyde reductase explain the dramatic shift of its new substrate specificity
aDepartment of Chemistry – BMC, Uppsala University, SE-751 23 Uppsala, Sweden, bFaculty of Biochemistry and Molecular Medicine, University of Oulu, PO Box 5400, Oulu FI-90014, Finland, and cBiocenter Oulu, University of Oulu, PO Box 5000, Oulu FI-90014, Finland
*Correspondence e-mail: email@example.com, firstname.lastname@example.org
The Fe2+-dependent E. coli enzyme FucO catalyzes the reversible interconversion of short-chain (S)-lactaldehyde and (S)-1,2-propanediol, using NADH and NAD+ as cofactors, respectively. Laboratory-directed evolution experiments have been carried out previously using phenylacetaldehyde as the substrate for screening catalytic activity with bulky substrates, which are very poorly reduced by wild-type FucO. These experiments identified the N151G/L259V double mutant (dubbed DA1472) as the most active variant with this substrate via a two-step evolutionary pathway, in which each step consisted of one point mutation. Here the crystal structures of DA1472 and its parent D93 (L259V) are reported, showing that these amino acid substitutions provide more space in the active site, though they do not cause changes in the main-chain conformation. The catalytic activity of DA1472 with the physiological substrate (S)-lactaldehyde and a series of substituted phenylacetaldehyde derivatives were systematically quantified and compared with that of wild-type as well as with the corresponding point-mutation variants (N151G and L259V). There is a 9000-fold increase in activity, when expressed as kcat/KM values, for DA1472 compared with wild-type FucO for the phenylacetaldehyde substrate. The crystal structure of DA1472 complexed with a non-reactive analog of this substrate (3,4-dimethoxyphenylacetamide) suggests the mode of binding of the bulky group of the new substrate. These combined structure–function studies therefore explain the dramatic increase in catalytic activity of the DA1472 variant for bulky aldehyde substrates. The structure comparisons also suggest why the active site in which Fe2+ is replaced by Zn2+ is not able to support catalysis.
Directed evolution is a powerful experimental tool to taylor the catalytic properties of natural enzymes to use them for applications (Savile et al., 2010; Huffman et al., 2019; Rosenthal et al., 2022) and to create model enzymes for targeted analyses of structure–function relationships. Classic examples of such studies involve the optimization of the KEMP eliminase activity of non-natural enzymes, where subsequent crystal structure determinations show how the active sites evolve to achieve significant new catalytic activity (Khersonsky et al., 2012; Blomberg et al., 2013). Here it is shown which structural adaptations were introduced by directed evolution experiments that changed the catalytic properties of the Escherichia coli (E. coli) (S)-lactaldehyde reductase [EC 18.104.22.168 (Blikstad et al., 2013, 2014)]. This enzyme, referred to as FucO, is also known as an Fe2+, NAD+-dependent group-III alcohol dehydrogenase (Reid & Fewson, 1994) and is involved in the catabolism of rare sugars in E. coli (Caballero et al., 1983; Baldomà & Aguilar, 1988; Blikstad & Widersten, 2010). FucO catalyzes the reversible interconversion of (S)-lactaldehyde and (S)-1,2-propanediol, using NADH and NAD+ as cofactors, respectively [Fig. 1(a)]. Its proposed physiological substrate is (S)-lactaldehyde [compound 2 in Fig. 1(b)], produced in the metabolism of L-fucose (Blikstad & Widersten, 2010) with the resulting (S)-1,2-propanediol as the ultimate fermentation product. FucO has, by virtue of its strict regiospecificity, been of interest as a potential biocatalyst for the stereospecific oxidation of 1,2-vicinal diols into chiral α-hydroxyaldehydes (Blikstad et al., 2013).
FucO is a dimeric enzyme consisting of two identical subunits (Fig. 2). Each subunit has two domains, the N-terminal NAD-binding domain and the C-terminal Fe2+-binding domain (Fig. 2). The pyrophosphate moiety of NAD is bound at the N-terminal end of the α-helix starting with Gly97. The catalytic site, near the bound nicotinamide moiety of NAD, is shaped by regions of both domains and recent structural studies have shown that the subunit can exist in both open and closed states in which the relative position of these two domains with respect to each other is different (Zavarise et al., 2023). It is proposed that in the closed state the competent ternary complex is formed. In this complex the substrate and the nicotinamide moiety tightly interact with each other, such that the active site Fe2+ ion stabilizes the negatively charged carbonyl oxygen generated after the hydride transfer from NADH to the carbonyl carbon of the aldehyde substrate. The reduction reaction [Fig. 1(a)] is completed by the transfer of a proton to the carbonyl oxygen from the catalytic water, by which the alcohol product is formed.
The substrate scope of wild-type FucO is limited to short-chain aldehydes and α-hydroxy aldehydes and their corresponding primary alcohols and vicinal diols (Blikstad & Widersten, 2010). However, an enzyme variant that efficiently catalyzes the reduction of phenylacetaldehyde (compound 3) [Fig. 1(b)] was obtained, following iterative saturation mutagenesis of active site residues and subsequent functional selection by screening (Blikstad et al., 2013, 2014). This FucO variant, dubbed DA1472, was isolated from a laboratory evolution campaign with the aim to obtain FucO variants with a broader substrate specificity, being an aldehyde reductase/alcohol dehydrogenase capable of catalyzing the reduction of the more bulky substituted aryl-aldehydes and the oxidation of the corresponding substituted arylated alcohols. Phenylacetaldehyde (compound 3) was employed as the selection substrate in the directed evolution protocol. Its use was motivated by the notion that the wild-type enzyme displays measurable, albeit very low, catalytic activity with this aldehyde (Table 2) (Blikstad et al., 2013), thus demonstrating catalytic competence, but leaving ample room for improvement. Also, the thermodynamic drive of the reductive reaction direction increased the sensitivity of the screening.
The isolated DA1472 variant is a double point mutation variant, N151G and L259V. It exhibits profoundly shifted substrate selectivity, compared with the wild-type FucO, and the aim of the current study was to extend the available kinetic data (Blikstad et al., 2013, 2014) of the DA1472 variant as well as of its parents, the D93 (L259V) and the A5 (N151G) variants with a systematic determination of their kcat and Km values for a range of aldehyde substrates. Furthermore, we have complemented these enzyme kinetic studies with X-ray crystallography of the DA1472 and D93 variants to obtain a comprehensive description of the structure–function relationship of the new DA1472 variant. Four new structures are reported including structures of the DA1472 variant complexed with a non-reactive analog of the bulky phenylacetaldehyde substrate.
Buffer components and other bulk chemicals were of the highest purity commercially available. Propanal (compound 1, Fig. 1; ≥98%, Sigma–Aldrich W292312), (S)-lactaldehyde (compound 2; 1 M solution in H2O, Sigma–Aldrich 47014) and 3,4-dimethoxyphenylacetamide (compound 7, Sigma–Aldrich CDS011402) were used in assays and reaction mixtures as provided by the manufacturer. Phenylacetaldehyde (compound 3, ≥95%, Sigma–Aldrich W287407) was column purified before use. The 4-methoxy-, 3-methoxy- and 3,4-dimethoxy-substituted aldehydes (compounds 4−6) were synthesized and purified as described (Al-Smadi et al., 2018).
2.3. Crystallization, structure determination and structure refinement.
The crystals were obtained by cocrystallization at room temperature, using the sitting-drop vapor diffusion method. The results of the crystallization experiments were monitored by the IceBear research data-management software (Daniel et al., 2021). The details of the crystallization and crystal treatment protocols are described in Table S1 of the supporting information. In all cases, the datasets were collected from frozen crystals. The data collection and data-processing methods are listed in Table 1. The structures were solved by molecular replacement using the expert mode of PHASER (McCoy et al., 2007) as implemented in CCP4i2 (Potterton et al., 2018) by splitting the search molecule (PDB entry 1rrm; unpublished work) into two parts, consisting of residues 1–185 and 186–383. In all structures there was one dimer per asymmetric unit. The structures were refined using alternately REFMAC5 (Kovalevskiy et al., 2018) for refinement calculations and Coot (Casañal et al., 2020) for manual improvement of the protein structure by inspecting the corresponding electron density maps, for identifying water molecules and for building ligands. The quality of the structures was checked using the validation tools of Coot and PDB validation reports. The final refinement statistics are listed in Table 1. In the complexes obtained by cocrystalization with the nucleotides NADH or NAD+, the density of the ADP-ribose part of the nucleotide was always clearly defined by the electron density map, whereas the electron density for the nicotinamide moiety was usually weaker, suggesting that either the nicotinamide ring is partially disordered, or NAD+ or NADH is partially hydrolysed, or both. In the structures with poor density for the nicotinamide ring the nucleotide was modeled as ADP-ribose (ADPR), as also described previously (Zavarise et al., 2023).
‡ADPR refers to ADP-ribose; 7 is compound 7 (3,4-dimethoxyphenylacetamide).
§Williams et al. (2018).
The dataset of the binary complex of D93 (L259V) and NADH (PDB entry 7qlg) was collected at beamline I24 of the Diamond Light Source (DLS, Oxford, UK), and XDS (Kabsch, 2010; Sparta et al., 2016) was used for data processing at a resolution of 2 Å. The merged intensities were obtained using AIMLESS (Evans & Murshudov, 2013). The nucleotide ligand was built as NADH, with the ligand name NAI.
The dataset of the binary complex of FucO DA1472 (N151G/L259V) and NADH (PDB entry 7qnh) was collected using the Bruker Microstar X8 diffractometer home X-ray source, integrated using SAINT and scaled using SADABS of the PROTEUM3 suite. The unmerged intensity dataset, provided by XPREP, with a resolution limit of 2.2 Å, was subsequently input to the CCP4i2 X-ray data-reduction and analysis pipeline and merged with AIMLESS. TLS refinement was used to improve the structure. The TLS groups were chosen manually, following the domain structure of FucO, such that two groups were identified; the N-terminal domain for TLS group 1 (residues 1–185) and the C-terminal domain for TLS group 2 (residues 186–383). The fit of NADH in the corresponding electron density map is shown in Fig. S1 of the supporting information.
The datasets of the two ternary complexes of DA1472 (N151G/L259V), obtained by cocrystallization with compound 7 and NAD+ or NADH (PDB entries 7qlq and 7qls, respectively) were collected at the BioMAX beamline of MAX IV (Lund, Sweden). The datasets of these crystals were processed with DIALS (Winter et al., 2018) and XDS respectively, and scaled and merged using AIMLESS. Towards the end of the refinement, TLS and NCS refinements were used to improve the structure further, using the same TLS groups as outlined above. In both these structures the bound ligands in the active site are the substrate analog (compound 7) and ADPR, with the ligand names E9I and APR, respectively. The restraints of the model of the substrate analog, as used in the refinement calculations, were obtained from the Global Grade Server (https://grade.globalphasing.org/cgi-bin/grade/server.cgi). The fit of this ligand in the corresponding electron density map is shown in Fig. S1.
The four new structures were compared with each other and with three FucO structures (available in the PDB), with a wild-type catalytic site, as well as with two previously described structures of point-mutated FucO variants (PDB entries 7qnf and 7r0p; Zavarise et al., 2023). The structures used to study the wild-type active site are the 1rrm structure (PDB entry 1rrm), refined at 1.60 Å resolution, complexed with Zn2+ and ADPR; the 5br4 structure (PDB entry 5br4; Cahn et al., 2016), refined at 0.91 Å resolution, and complexed with Zn2+ and NAD+; as well as the 2bl4 structure (PDB entry 2bl4; Montella et al., 2005) refined at 2.85 Å, complexed with Fe2+ and NAD+. The 1rrm and 2bl4 structures have the wild-type sequence, whereas the 5br4 structure involves a point-mutation variant in which a residue at the binding site of the adenine moiety of NAD+ (far from the catalytic site) was mutated. This mutation causes only minor, local structural changes. The conformations of the main chain traces of the 1rrm, 5br4 and 2bl4 structures around the catalytic site are the same.
The two previously reported structures of point-mutated FucO variants used for comparison are the structures of the double mutant N151G/L259 (PDB entry 7qnf, refined at 2.14 Å resolution, complexed with Fe2+) as well as the D47 variant, which has the F254I point mutation (PDB entry 7r0p, refined at 1.48 Å resolution, complexed with Fe2+). In the latter two structures, the active sites are also complexed with a nucleotide, ADP-ribose and NAD+. The active site of subunit B of the 7qnf structure is also complexed with the substrate ethyleneglycol, interacting directly with the active site Fe2+ ion via one of its oxygen atoms. Using the D47 variant the kinetic and structural properties of the F254I point mutation have been probed. Like for D93, the catalytic properties of D47 for the oxidation of propanol-1 have not changed very much with respect to the wild-type (Blikstad et al., 2014), and there are only minor structural rearrangements of the main chain for residues 253–254–255, being located near the edge of the substrate specificity pocket.
In the four new crystal structures described here, the bound active site nucleotide is NADH (PDB entries 7qnh, 7qlg) or ADPR, and compound 7 (PDB entries 7qls, 7qlq) and each of these active sites are also complexed with Fe2+. In each of these four structures, FucO was crystallized with a dimer in the asymmetric unit and in these structures the two domains of each subunit adopt the closed conformation. For structural comparisons the subunits were superimposed on each other with the SSM protocol (Krissinel & Henrick, 2004), as implemented in Coot. The figures, which visualize the structural information, were made using PyMOL (The PyMOL Molecular Graphics System, Version 2.3, Schrödinger, LLC).
Aldehyde reduction was assayed spectrophotometrically by the time-dependent concomitant oxidation of NADH, followed at 340 nm in a 0.5 cm cuvette in a 0.1 M sodium phosphate buffer, pH 7.5 at 30°C. The given concentrations of the aldehyde substrates included in the activity assays refer to the sum of the carbonyl and hydrated forms of the aldehydes. The Michaelis–Menten model [equation (1)] or a reparametrized version thereof [equation (2)] was fitted to the initial rates to extract steady-state parameters, using the programs MMFIT or RFFIT, respectively, of the Simfit package (https://www.simfit.org.uk). In cases where substrate inhibition was observed (e.g. see Fig. S2), a model including reversible formation of a dead-end complex [equation (3)] was fitted using the QNFIT program of the Simfit package. The parameter KIS is the equilibrium dissociation constant of the non-productively bound aldehyde substrate.
Due to the practical limitations of the spectrophotometric assay, for example a high starting absorbance of NADH which limits the maximal concentration that can be added to the assay mixtures, and the limited solubility of the tested aldehyde substrates, the non-varied substrate could not be present at saturating concentrations in all cases. Pseudo-first-order reaction conditions could therefore not be fully met in these cases. Hence, values of kcat and kcat/KM should, in noted cases (see Table 2 footnotes), be considered as lower estimates. NADH concentration was kept at 0.4 mM, which corresponds to 20 × KM for the wild-type and A5 enzymes, 5 × KM for D93 and 0.62 × KM for DA1472. Aldehydes were added at concentrations of 2–50 mM (compounds 1 and 2), 0.1–4.5 mM (compound 3), 0.05–1 mM (compound 4), 0.025–0.75 mM (compound 5) and 0.025–1.5 mM (compound 6). Kinetic parameters were also determined in the presence of varying concentrations of NADH (20–800 µM) and either 30 mM (wild-type, 14 × KM and DA1472, 2.1 × KM) or 50 mM (A5, 4.6 × KM and D93, 17 × KM) of aldehyde 1. The ratios of the aldehydic versus the hydrated states of the aldehyde substrates were not determined.
‡Determined in the presence of 50 mM of compound 1.
§Determined in the presence of 30 mM of compound 1.
¶Underestimated due to the relatively high KMNADH.
††Unable to reach enzyme saturation within accessible substrate concentration. The > sign indicates the lower range for the parameter value, limited by the concentration range used.
‡‡Blikstad et al. (2013).
§§Determined after fitting a model of reversible substrate inhibition [equation (3)].
The reduction rates of compound 3, expressed as kcat or kcat/KM, were determined at pH values between 6–8, as described above, in 0.1 M sodium phosphate buffers, in the presence of 0.4 mM NADH and at 30°C.
Initial velocities of the DA1472 catalyzed reduction of 0.02–0.75 mM compound 5 were recorded in 0−30%(w/v) sucrose in 0.1 M sodium phosphate buffer, pH 7.0 at 30°C. Translation of sucrose concentrations into relative viscosity was carried out as described previously (Blikstad & Widersten, 2010). The degree of dependency of kcat and kcat/KM on the relative viscosity (ηrel) was estimated by fitting a linear model using LINFIT in the Simfit package [Table S2, Fig. S2(c)].
The DA1472 variant emerged from two separate but converging evolutionary pathways (Blikstad et al., 2013, 2014). One route produced the single mutant L259V (dubbed variant D93). D93 displays features of a `generalist' enzyme regarding substrate selectivity, retaining reasonable activities with compound 1 and its physiological substrate compound 2 (Table 2) and, in addition, being able to catalyze the reduction of the more bulky compound 3 with 13-fold improved catalytic efficiency (kcat/KM) compared with the wild-type (Table 2). In a parallel search, FucO variant A5 (N151G) was isolated. A5 catalyzes the reduction of its physiological substrate (compound 2) with a similar kcat value as the wild-type, but it requires higher substrate concentrations to reach saturation (KM of compound 2 has increased 100-fold; Table 2). The substantial increase in KM for compound 2 reflects the importance of the acetamide side chain of Asn151, being hydrogen bonded with the 2-hydroxyl moiety of compound 2 (Zavarise et al., 2023). However, the activity with the more bulky substrate, compound 3, as judged by kcat/KM, has increased by 150-fold with respect to the wild-type (Table 2).
The double mutant (DA1472) was subsequently isolated from a second-generation library parented by the D93 variant. The observed improvement in activity with compound 3 of DA1472 requires both substitutions to be present (Table 2). Comparing the activities of A5 and D93 with those of DA1472, it is clear that DA1472 is closer to A5 in the activity profile, demonstrating the important contribution of the N151G substitution to the shift in substrate selectivity, although combined epistatic effects can be observed. For example, the substantially lower KM of compound 3 shown by DA1472 (approximately ninefold, compared with the wild-type, Table 2) is not the result of step-wise additive effects by the individual mutations, but rather depends on the combination of substitutions.
In order to further analyze the substrate scope of DA1472 and its parent variants, regarding arylated aldehydes and to assess possible effects on the physiological reaction caused by the introduced mutations, an extended set of phenylacetaldehyde derivatives [compounds 4, 5 and 6; Fig. 1(b)] were also investigated as substrates (Table 2). The results illustrate a dramatic shift in substrate scope. The preference of the wild-type enzyme for short-chain aliphatic 2-hydroxyaldehydes has been shifted to favor aryl-substituted, non-hydroxylated aldehydes. kcat/KM for phenylacetaldehyde (compound 3) has increased approximately 9000-fold whereas the activity with the native (S)-lactaldehyde (compound 2) has decreased approximately 80-fold (Table 2). Furthermore, the accentuated (S)-selectivity of wild-type FucO in the oxidation of 1,2-propanediol [320-fold in favor of (S)-1,2-propanediol over the (R)-enantiomer] is lost in this variant (Blikstad et al., 2014). The activities displayed by DA1472 with the methoxy-substituted aldehydes 4, 5 and 6 are notably high and the kcat/KM values for the monosubstituted compounds 4 and 5 even exceed that of the wild-type enzyme with the physiological substrate 2 (Table 2). Different degrees of substrate inhibition were observed in the reactions with all phenylacetaldehyde derivatives and a model incorporating the reversible formation of a dead-end quaternary complex [equation (3)] was fitted to the steady-state initial rates. An example saturation curve of the DA1472 catalyzed reduction of compound 3 is shown in Fig. S2, together with the proposed reaction scheme. As judged by the estimated values of the apparent dissociation constants of these proposed dead-end complexes (KIS in Table 2), the para-methoxy substituted 4 and 6 appear to be less prone to bind in a non-productive mode.
High-resolution crystal structures were obtained of the D93 (L259V) and the DA1472 (N151G/L259V) variants cocrystallized with NADH (at 2.0 and 2.2 Å resolution, respectively, Table 1). NADH is bound in the groove shaped by the N-terminal NAD-binding domain and the C-terminal Fe2+-binding domain (Fig. 2). The mode of binding of NADH in the active site, including the nicotinamide moiety, is well defined by the electron density map (Fig. S1), being the same in both structures. Also, crystal structures were obtained of the DA1472 variant in the presence of the bulky substrate analog, compound 7, cocrystallized in the presence of either NAD+ or NADH (at 2.6 and 2.4 Å resolution, respectively, Table 1). In the structures of the latter two complexes, the mode of binding of the nicotinamide moieties of NAD+ and NADH is not defined by the electron density map and therefore, the nucleotide was modeled as ADP-ribose (ADPR, as described in the Materials and methods). The mode of binding of the substrate analog is the same in both structures and well defined by the electron density map (Fig. S1).
The structures of the protein regions with the mutated residues are well defined by the electron density maps and the conformations of these regions are the same as in the wild-type. The comparison of the structures of the active sites of the wild-type and the DA1472 variant shows that the shorter side chains of the L259V (valine instead of leucine) and N151G (glycine instead of asparagine) amino acid substitutions provide more space in the region where the substrate binds (Figs. 3 and 4). The substrate-binding region is identified by the mode of binding of the substrate ethyleneglycol, as observed in PDB entry 7qnf, also shown in Figs. 3 and 4. Fig. 3 also includes the mode of binding of NADH to the active site of the DA1472 variant (PDB entry 7qnh). This is the same as observed for NAD+ bound in the wild-type active site (PDB entry 5br4; Fig. 3).
The mode of binding of compound 7 (3,4-dimethoxyphenylacetamide) as provided by the structures obtained from the cocrystallization experiments in the presence of NADH or NAD+, is shown in Fig. 4. The observed mode of binding of compound 7 is possible because of the N151G and L259V amino acid changes of DA1472. In particular, the side chain of Asn151 of the wild-type active site clashes with the mode of binding of compound 7 (Fig. 4). The oxygen atom of the amide group of compound 7 interacts with the Fe2+ ion and the amide nitrogen is hydrogen bonded to O (Thr144) (Fig. 5). The phenyl ring of compound 7 fits in a hydrophobic pocket and has a presumably favorable (Serrano et al., 1991; Chatterjee et al., 2019; Chakrabarti & Bhattacharyya, 2007) edge-to-face interaction with the phenyl ring of Phe254 (Fig. 5). The two amino acid changes allow the more bulky phenylacetaldehyde substrates (compounds 3–6) to be efficiently reduced (Table 2) by the DA1472 variant. The structures of the DA1472 variant complexed with this substrate analog (Figs. 4 and 5) suggest the mode of binding and the favorable interactions of the aromatic phenylacetaldehyde substrates in the active site of this FucO variant.
Two other enzymological properties of the DA1472 variant were investigated: the dependency of the kinetic properties on medium viscosity as well as the pH dependency of the reduction reaction. The viscosity of the solvent affects the kinetic rates of diffusion-controlled steps, including association with the substrate (in the case of diffusion-controlled reactions), as well as the dissociation of product(s) and/or conformational changes that are rate limiting (Schurr, 1970; Demchenko et al., 1989; Eser & Fitzpatrick, 2010). The dependency on viscosity was tested with the meta-substituted compound 5 as the aldehyde substrate, showing complete (unit) dependency for kcat and kcat/KM [Table S2, Fig. S2(c)]. These properties are in line with rate-limiting product release, presumably associated with conformational changes, as was also proposed for the wild-type enzyme, when using propanal (compound 1) as the aldehyde substrate (Blikstad & Widersten, 2010). The pH dependency of kcat and kcat/KM in the reduction of compound 3 is relatively flat (Fig. S3), as observed for the wild-type enzyme for the reduction of compound 1 (Blikstad & Widersten, 2010), which is in agreement with the proposed reaction mechanism, whereby a catalytic water molecule provides the proton for the reduction reaction (Zavarise et al., 2023).
Structure comparisons suggest that the geometry of the DA1472 catalytic site is also the same as seen for the wild-type. For example, the superpositioning of the structures of DA1472 (7qnf, complexed with ethyleneglycol, refined at 2.14 Å) and the high-resolution structure of wild-type FucO (PDB entry 1rrm, chain A, complexed with (S)-1,2-propanediol, refined at 1.6 Å) shows no structural differences (Fig. 6). In these two structures, the modeled nucleotide is ADPR and in both active sites a substrate is bound. In the active site of DA1472 the metal ion is Fe2+, whereas in the 1rrm structure the bound metal ion is Zn2+. The enzyme is only active in the presence of Fe2+ and is inactivated in the presence of Zn2+, Cu2+ and Cd2+ (Montella et al., 2005). The Fe2+ ion is coordinated by a carboxylate oxygen of the Asp196 side chain and the NE2 nitrogen atoms of His200, His263 and His277. In the 7qnf structure the Fe2+ ion is also coordinated by an additional water (WAT232) as well as an oxygen atom of the substrate (Fig. 6). The latter oxygen atom is replaced by a water in the case where there is no substrate, substrate analog or nicotinamide moiety bound in the active site. This coordination geometry has also been observed in the structures of the unliganded active sites of homologous dehydrogenases, such as the bacterial alcohol dehydrogenase of Zymomonas mobilis (Moon et al., 2011) and Klebsiella pneumonia (Marçal et al., 2009).
In the FucO structures of the point-mutation variants reported here, as well as in previous work (Zavarise et al., 2023), the active sites are complexed with Fe2+, like in the wild-type 2bl4 structure (Montella et al., 2005), and the Fe2+ position is the same in all these structures. In the 1rrm and 5br4 crystal structures this Fe2+ is replaced by Zn2+. The Fe2+ ion interacts with the reactive oxygen atom of the substrate (Fig. 6). This interaction stabilizes the intermediate of the reaction cycle that has the negatively charged oxyanion, generated in the reduction by the transfer of the hydride ion from NADH to the carbonyl carbon atom of the substrate. In the FucO active site, the Fe2+ ion tightly interacts with the nicotinamide moiety, as the distance between the Fe2+ ion and the C5 atom of the nicotinamide group is about 3 Å. The Fe2+ position is the same in the presence of bound ADPR, NADH or NAD+ (Fig. 7). The position of the nicotinamide moieties of the latter two structures is well defined by the corresponding electron density maps (Fig. S1), being hydrogen bonded to residues of the NAD-binding domain. The oxygen atom of the amide group of the nicotinamide part of NAD is hydrogen bonded to OG1 of Thr149, whereas its nitrogen atom is hydrogen bonded to the peptide oxygen atom of residue 151 (Fig. 7) and a carboxylate side-chain oxygen atom of Asp102. Further studies are required to understand why, in the presence of Zn2+ (instead of Fe2+), the enzyme is inactive; but in this respect, it is of interest to note that in the Zn2+ active site, complexed with NAD+ (as captured in the atomic resolution 5br4 structure), the Zn2+ mode of binding is shifted significantly away from the canonical Fe2+ position (Fig. 7), such that the distance between the Zn2+ ion and the C5 atom of the nicotinamide is 4 Å (instead of 3 Å). Also, the side chains of Asp196, His200 and His263, which interact with the Zn2+ ion of the 5br4 structure, have shifted somewhat in the same direction as the Zn2+ ion (Fig. 7). In the Fe2+-complexed active sites such a shift is not observed in any of the available structures. These structural differences between the Zn2+/NAD+- and the Fe2+/NAD+-complexed active sites could be related to the notion that FucO is inactive in the presence of Zn2+.
The 30-fold increase in KM for NADH displayed by DA1472 (but not by A5 and D93; Table 2) cannot be explained directly from the available structures of the complexes with bound NAD+ or NADH, which suggest that the amino acid changes do not affect the mode of binding of the nucleotide. It is proposed that in solution FucO exists in both open and closed states (Zavarise et al., 2023). Indeed, it has been demonstrated that group-I alcohol dehydrogenases exhibit such dynamic nucleotide binding behavior (Plapp, 2010). The existence of closed/open states of the binary E·NAD+(H) complexes of the FucO enzyme, where one of the conformations would exhibit lower affinity for the coenzyme, can possibly provide an explanation to why the mutations in DA1472 result in this behavior. If the substitutions affect the relative stabilities of such states, then an apparent dissociation constant such as KM could be affected as a consequence. This hypothesis is in line with current theories on the functional evolution of enzymes which state that conformational sub-states and their relative stabilities can form the basis for a given functional profile (James & Tawfik, 2003; Tokuriki & Tawfik, 2009; Richard, 2022). Perturbation of the native state by amino acid changes may shift the relative sub-state landscape, resulting in new traits. In any case, KM for NADH will depend on the rate constants of the individual reactions, including the off-rates of alcohol and NAD+, and changes in these rates can give rise to higher KM values, as observed for NADH for the DA1472 variant (Table 2).
The N151G/L259V double substitution of FucO results in a variant with a shifted substrate scope that does not overlap with the wild-type enzyme and this variant can be viewed as a specialist aryl-aldehyde reductase. The 9000-fold shift in catalytic efficiency as calculated from the increase of the kcat/KM values from wild-type FucO to DA1472 for the substrate phenylacetaldehyde (compound 3) (Table 2) corresponds to an estimated change in transition state stabilization of 5.4 kcal mol−1, as calculated from equation (4) (Fersht et al., 1985):
The structural data show that the two amino acid changes (N151G and L259V) do not cause any change of main-chain conformation, but do provide more space in the substrate-binding groove. Also, the geometry of the catalytic site is not affected (Figs. 6 and 7) and the enzymological data suggest that the reaction mechanism of the DA1472 variant is also the same. Thus, these two amino acid changes lead to a dramatic modification in function, illustrating the malleability of this active site by the introduction of structurally non-destructive substitutions, obtained by targeted functional selection.
‡The authors contributed equally to this work.
We thank the beamline scientists at MAX IV and DLS for their much appreciated expert help. The use of the facilities and expertise of the Biocenter Oulu Structural Biology core facility, a member of Biocenter Finland, Instruct-ERIC Centre Finland and FINStruct, is gratefully acknowledged.
This work was supported by grants of the Olle Engkvist Byggmästare Foundation (grant no. 194–0638 to MW) and of the Academy of Finland (grant no. 339894 to RKW).
Al-Smadi, D., Enugala, T. R., Norberg, T., Kihlberg, J. & Widersten, M. (2018). Synlett, 29, 1187–1190. CAS Google Scholar
Baldomà, L. & Aguilar, J. (1988). J. Bacteriol. 170, 416–421. PubMed Web of Science Google Scholar
Blikstad, C., Dahlström, K. M., Salminen, T. A. & Widersten, M. (2013). ACS Catal. 3, 3016–3025. Web of Science CrossRef CAS Google Scholar
Blikstad, C., Dahlström, K. M., Salminen, T. A. & Widersten, M. (2014). FEBS J. 281, 2387–2398. Web of Science CrossRef CAS PubMed Google Scholar
Blikstad, C. & Widersten, M. (2010). J. Mol. Catal. B Enzym. 66, 148–155. Web of Science CrossRef CAS Google Scholar
Blomberg, R., Kries, H., Pinkas, D. M., Mittl, P. R., Grütter, M. G., Privett, H. K., Mayo, S. L. & Hilvert, D. (2013). Nature, 503, 418–421. Web of Science CrossRef CAS PubMed Google Scholar
Caballero, E., Baldomá, L., Ros, J., Boronat, A. & Aguilar, J. (1983). J. Biol. Chem. 258, 7788–7792. CrossRef CAS PubMed Google Scholar
Cahn, J. K., Baumschlager, A., Brinkmann-Chen, S. & Arnold, F. H. (2016). Protein Eng. Des. Sel. 29, 31–38. Web of Science CAS PubMed Google Scholar
Casañal, A., Lohkamp, B. & Emsley, P. (2020). Protein Sci. 29, 1069–1078. Web of Science PubMed Google Scholar
Chakrabarti, P. & Bhattacharyya, R. (2007). Prog. Biophys. Mol. Biol. 95, 83–137. Web of Science CrossRef PubMed CAS Google Scholar
Chatterjee, K. S., Tripathi, V. & Das, R. (2019). J. Biol. Chem. 294, 6772–6784. Web of Science CrossRef CAS PubMed Google Scholar
Daniel, E., Maksimainen, M. M., Smith, N., Ratas, V., Biterova, E., Murthy, S. N., Rahman, M. T., Kiema, T.-R., Sridhar, S., Cordara, G., Dalwani, S., Venkatesan, R., Prilusky, J., Dym, O., Lehtiö, L., Koski, M. K., Ashton, A. W., Sussman, J. L. & Wierenga, R. K. (2021). Acta Cryst. D77, 151–163. Web of Science CrossRef IUCr Journals Google Scholar
Demchenko, A. P., Ruskyn, O. I. & Saburova, E. A. (1989). Biochim. Biophys. Acta, 998, 196–203. CrossRef CAS PubMed Web of Science Google Scholar
Eser, B. E. & Fitzpatrick, P. F. (2010). Biochemistry, 49, 645–652. Web of Science CrossRef CAS PubMed Google Scholar
Evans, P. R. & Murshudov, G. N. (2013). Acta Cryst. D69, 1204–1214. Web of Science CrossRef CAS IUCr Journals Google Scholar
Fersht, A. R., Shi, J. P., Knill-Jones, J., Lowe, D. M., Wilkinson, A. J., Blow, D. M., Brick, P., Carter, P., Waye, M. M. & Winter, G. (1985). Nature, 314, 235–238. CrossRef CAS PubMed Web of Science Google Scholar
Huffman, M. A., Fryszkowska, A., Alvizo, O., Borra-Garske, M., Campos, K. R., Canada, K. A., Devine, P. N., Duan, D., Forstater, J. H., Grosser, S. T., Halsey, H. M., Hughes, G. J., Jo, J., Joyce, L. A., Kolev, J. N., Liang, J., Maloney, K. M., Mann, B. F., Marshall, N. M., McLaughlin, M., Moore, J. C., Murphy, G. S., Nawrat, C. C., Nazor, J., Novick, S., Patel, N. R., Rodriguez-Granillo, A., Robaire, S. A., Sherer, E. C., Truppo, M. D., Whittaker, A. M., Verma, D., Xiao, L., Xu, Y. & Yang, H. (2019). Science, 366, 1255–1259. Web of Science CrossRef PubMed Google Scholar
James, L. C. & Tawfik, D. S. (2003). Trends Biochem. Sci. 28, 361–368. Web of Science CrossRef PubMed CAS Google Scholar
Kabsch, W. (2010). Acta Cryst. D66, 125–132. Web of Science CrossRef CAS IUCr Journals Google Scholar
Khersonsky, O., Kiss, G., Röthlisberger, D., Dym, O., Albeck, S., Houk, K. N., Baker, D. & Tawfik, D. S. (2012). Proc. Natl Acad. Sci. USA, 109, 10358–10363. Web of Science CrossRef CAS PubMed Google Scholar
Kovalevskiy, O., Nicholls, R. A., Long, F., Carlon, A. & Murshudov, G. N. (2018). Acta Cryst. D74, 215–227. Web of Science CrossRef IUCr Journals Google Scholar
Krissinel, E. & Henrick, K. (2004). Acta Cryst. D60, 2256–2268. Web of Science CrossRef CAS IUCr Journals Google Scholar
Marçal, D., Rêgo, A. T., Carrondo, M. A. & Enguita, F. J. (2009). J. Bacteriol. 191, 1143–1151. Web of Science PubMed Google Scholar
McCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D., Winn, M. D., Storoni, L. C. & Read, R. J. (2007). J. Appl. Cryst. 40, 658–674. Web of Science CrossRef CAS IUCr Journals Google Scholar
Montella, C., Bellsolell, L., Pérez-Luque, R., Badía, J., Baldoma, L., Coll, M. & Aguilar, J. (2005). J. Bacteriol. 187, 4957–4966. Web of Science CrossRef PubMed CAS Google Scholar
Moon, J. H., Lee, H. J., Park, S. Y., Song, J. M., Park, M. Y., Park, H. M., Sun, J., Park, J. H., Kim, B. Y. & Kim, J. S. (2011). J. Mol. Biol. 407, 413–424. Web of Science CrossRef CAS PubMed Google Scholar
Plapp, B. V. (2010). Arch. Biochem. Biophys. 493, 3–12. Web of Science CrossRef PubMed CAS Google Scholar
Potterton, L., Agirre, J., Ballard, C., Cowtan, K., Dodson, E., Evans, P. R., Jenkins, H. T., Keegan, R., Krissinel, E., Stevenson, K., Lebedev, A., McNicholas, S. J., Nicholls, R. A., Noble, M., Pannu, N. S., Roth, C., Sheldrick, G., Skubak, P., Turkenburg, J., Uski, V., von Delft, F., Waterman, D., Wilson, K., Winn, M. & Wojdyr, M. (2018). Acta Cryst. D74, 68–84. Web of Science CrossRef IUCr Journals Google Scholar
Reid, M. F. & Fewson, C. A. (1994). Crit. Rev. Microbiol. 20, 13–56. CrossRef CAS PubMed Web of Science Google Scholar
Richard, J. P. (2022). Biochemistry, 61, 1533–1542. Web of Science CrossRef CAS PubMed Google Scholar
Rosenthal, K., Bornscheuer, U. T. & Lütz, S. (2022). Angew. Chem. Int. Ed. 61, e202208358. Web of Science CrossRef Google Scholar
Savile, C. K., Janey, J. M., Mundorff, E. C., Moore, J. C., Tam, S., Jarvis, W. R., Colbeck, J. C., Krebber, A., Fleitz, F. J., Brands, J., Devine, P. N., Huisman, G. W. & Hughes, G. J. (2010). Science, 329, 305–309. Web of Science CrossRef CAS PubMed Google Scholar
Schurr, J. M. (1970). Biophys. J. 10, 717–727. CrossRef CAS PubMed Web of Science Google Scholar
Serrano, L., Bycroft, M. & Fersht, A. R. (1991). J. Mol. Biol. 218, 465–475. CrossRef CAS PubMed Web of Science Google Scholar
Sparta, K. M., Krug, M., Heinemann, U., Mueller, U. & Weiss, M. S. (2016). J. Appl. Cryst. 49, 1085–1092. Web of Science CrossRef CAS IUCr Journals Google Scholar
Tokuriki, N. & Tawfik, D. S. (2009). Science, 324, 203–207. Web of Science CrossRef PubMed CAS Google Scholar
Williams, C. J., Headd, J. J., Moriarty, N. W., Prisant, M. G., Videau, L. L., Deis, L. N., Verma, V., Keedy, D. A., Hintze, B. J., Chen, V. B., Jain, S., Lewis, S. M., Arendall, W. B. III, Snoeyink, J., Adams, P. D., Lovell, S. C., Richardson, J. S. & Richardson, D. C. (2018). Protein Sci. 27, 293–315. Web of Science CrossRef CAS PubMed Google Scholar
Winter, G., Waterman, D. G., Parkhurst, J. M., Brewster, A. S., Gildea, R. J., Gerstel, M., Fuentes-Montero, L., Vollmar, M., Michels-Clark, T., Young, I. D., Sauter, N. K. & Evans, G. (2018). Acta Cryst. D74, 85–97. Web of Science CrossRef IUCr Journals Google Scholar
Zavarise, A., Sridhar, S., Kiema, T.-R., Wierenga, R. K. & Widersten, M. (2023). FEBS J., 290, 465–481. Web of Science CrossRef CAS PubMed Google Scholar
This is an open-access article distributed under the terms of the Creative Commons Attribution (CC-BY) Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original authors and source are cited.