Identification and characterization of the pyridoxal 5’-phosphate allosteric site in Escherichia coli pyridoxine 5’-phosphate oxidase
2021; Elsevier BV; Volume: 296; Linguagem: Inglês
10.1016/j.jbc.2021.100795
ISSN1083-351X
AutoresAnna Barile, Theo Battista, Annarita Fiorillo, Martino L. di Salvo, Francesco Malatesta, Angela Tramonti, Andrea Ilari, Roberto Contestabile,
Tópico(s)Metabolism and Genetic Disorders
ResumoPyridoxal 5'-phosphate (PLP), the catalytically active form of vitamin B6, plays a pivotal role in metabolism as an enzyme cofactor. PLP is a very reactive molecule and can be very toxic unless its intracellular concentration is finely regulated. In Escherichia coli, PLP formation is catalyzed by pyridoxine 5'-phosphate oxidase (PNPO), a homodimeric FMN-dependent enzyme that is responsible for the last step of PLP biosynthesis and is also involved in the PLP salvage pathway. We have recently observed that E. coli PNPO undergoes an allosteric feedback inhibition by PLP, caused by a strong allosteric coupling between PLP binding at the allosteric site and substrate binding at the active site. Here we report the crystallographic identification of the PLP allosteric site, located at the interface between the enzyme subunits and mainly circumscribed by three arginine residues (Arg23, Arg24, and Arg215) that form an "arginine cage" and efficiently trap PLP. The crystal structure of the PNPO–PLP complex, characterized by a marked structural asymmetry, presents only one PLP molecule bound at the allosteric site of one monomer and sheds light on the allosteric inhibition mechanism that makes the enzyme-substrate–PLP ternary complex catalytically incompetent. Site-directed mutagenesis studies focused on the arginine cage validate the identity of the allosteric site and provide an effective means to modulate the allosteric properties of the enzyme, from the loosening of the allosteric coupling (in the R23L/R24L and R23L/R215L variants) to the complete loss of allosteric properties (in the R23L/R24L/R21L variant). Pyridoxal 5'-phosphate (PLP), the catalytically active form of vitamin B6, plays a pivotal role in metabolism as an enzyme cofactor. PLP is a very reactive molecule and can be very toxic unless its intracellular concentration is finely regulated. In Escherichia coli, PLP formation is catalyzed by pyridoxine 5'-phosphate oxidase (PNPO), a homodimeric FMN-dependent enzyme that is responsible for the last step of PLP biosynthesis and is also involved in the PLP salvage pathway. We have recently observed that E. coli PNPO undergoes an allosteric feedback inhibition by PLP, caused by a strong allosteric coupling between PLP binding at the allosteric site and substrate binding at the active site. Here we report the crystallographic identification of the PLP allosteric site, located at the interface between the enzyme subunits and mainly circumscribed by three arginine residues (Arg23, Arg24, and Arg215) that form an "arginine cage" and efficiently trap PLP. The crystal structure of the PNPO–PLP complex, characterized by a marked structural asymmetry, presents only one PLP molecule bound at the allosteric site of one monomer and sheds light on the allosteric inhibition mechanism that makes the enzyme-substrate–PLP ternary complex catalytically incompetent. Site-directed mutagenesis studies focused on the arginine cage validate the identity of the allosteric site and provide an effective means to modulate the allosteric properties of the enzyme, from the loosening of the allosteric coupling (in the R23L/R24L and R23L/R215L variants) to the complete loss of allosteric properties (in the R23L/R24L/R21L variant). Pyridoxal 5'-phosphate (PLP) acts as cofactor for over 150 enzymes (1Percudani R. Peracchi A. A genomic overview of pyridoxal-phosphate-dependent enzymes.EMBO Rep. 2003; 4: 850-854Crossref PubMed Scopus (326) Google Scholar, 2Percudani R. Peracchi A. The B6 database: A tool for the description and classification of vitamin B6-dependent enzymatic activities and of the corresponding protein families.BMC Bioinformatics. 2009; 10: 273Crossref PubMed Scopus (166) Google Scholar) involved in a number of metabolic pathways such as the synthesis, transformation, and degradation of amines and amino acids; supply of one carbon units; transsulfuration; synthesis of tetrapyrrolic compounds (including heme) and polyamines; and biosynthesis and degradation of neurotransmitters. In the cell, PLP is supplied to apoenzymes through either biosynthesis or recycling of B6 vitamers coming from the environment and protein turnover. The latter route is the only one available in organisms that cannot synthesize PLP, such as humans, and they obtain it through a salvage pathway (Fig. 1) catalyzed by the enzymes pyridoxal kinase, pyridoxine 5'-phosphate oxidase (PNPO; EC 1.4.3.5), and either specific or nonspecific phosphatases (3di Salvo M.L. Contestabile R. Safo M.K. Vitamin B(6) salvage enzymes: Mechanism, structure and regulation.Biochim. Biophys. Acta. 2011; 1814: 1597-1608Crossref PubMed Scopus (137) Google Scholar). By using FMN as cofactor, PNPO catalyzes the oxidation of both pyridoxine 5'-phosphate (PNP) and pyridoxamine 5'-phosphate to PLP, reducing molecular oxygen to hydrogen peroxide. With the Escherichia coli PNPO (ePNPO), the specificity constant for PNP is 50-fold higher than that for pyridoxamine 5'-phosphate (4Zhao G. Winkler M.E. Kinetic limitation and cellular amount of pyridoxine (pyridoxamine) 5'-phosphate oxidase of Escherichia coli K-12.J. Bacteriol. 1995; 177: 883-891Crossref PubMed Google Scholar). In E. coli, which synthesizes PLP through the so-called DXP-dependent pathway, ePNPO (encoded by the pdxH gene) plays the dual role of salvage pathway enzyme and of last enzyme in PLP biosynthesis (Fig. 1). This is the reason why the reaction catalyzed by ePNPO is a key point in PLP homeostasis. We have recently demonstrated (5Barile A. Tramonti A. di Salvo M.L. Nogués I. Nardella C. Malatesta F. Contestabile R. Allosteric feedback inhibition of pyridoxine 5'-phosphate oxidase from Escherichia coli.J. Biol. Chem. 2019; 294: 15593-15603Abstract Full Text Full Text PDF PubMed Scopus (5) Google Scholar) that ePNPO undergoes an allosteric feedback inhibition by PLP, through a mechanism in which binding of substrate at the active site and binding of PLP at an allosteric site negatively affect each other to a large extent (Fig. 2A). Such a strong allosteric coupling is also reflected in the lack of catalytic competence of the enzyme–PNP–PLP ternary complex (PES in Fig. 2A) and in the lack of capability by the enzyme–PLP complex (PE, in which PLP is bound at the allosteric site) to bind a second PLP molecule at the active site (at least in the range of PLP concentration used in kinetics measurements). Since binding of PLP at the allosteric site takes place with higher affinity than binding at the active site, PLP binding at the active site of the free enzyme does not occur. PLP allosteric inhibition is also observed with human PNPO (6Barile A. Nogués I. di Salvo M.L. Bunik V. Contestabile R. Tramonti A. Molecular characterization of pyridoxine 5'-phosphate oxidase and its pathogenic forms associated with neonatal epileptic encephalopathy.Sci. Rep. 2020; 10: 13621Crossref PubMed Scopus (4) Google Scholar). However, in this case the coupling between the allosteric site and the active site is weaker, so that the PES complex maintains a partial catalytic activity and PLP can also bind at the active site, forming an enzyme complex with two PLP molecules (PEP in Fig. 2B). The PLP allosteric site of ePNPO most likely corresponds to the secondary "PLP tight binding site" previously revealed by other authors (7Yang E.S. Schirch V. Tight binding of pyridoxal 5'-phosphate to recombinant Escherichia coli pyridoxine 5'-phosphate oxidase.Arch. Biochem. Biophys. 2000; 377: 109-114Crossref PubMed Scopus (30) Google Scholar) when they observed that, if the enzyme is incubated with PLP and then passed through a size exclusion chromatography (SEC) column, it retains PLP. This observation is due to the fact that PLP binds tightly at a secondary site, distinct from the active site, as demonstrated by a quadruple ePNPO variant whose active site is impaired and that preserves this property (5Barile A. Tramonti A. di Salvo M.L. Nogués I. Nardella C. Malatesta F. Contestabile R. Allosteric feedback inhibition of pyridoxine 5'-phosphate oxidase from Escherichia coli.J. Biol. Chem. 2019; 294: 15593-15603Abstract Full Text Full Text PDF PubMed Scopus (5) Google Scholar). In the cell, PLP tightly bound to this secondary site of ePNPO is protected from the environment where, owing to the high reactivity of its aldehyde group, it would readily combine with all sorts of amines and thiols present in the cell; at the same time, it can be transferred to PLP-dependent apoenzymes, suggesting that ePNPO may play a role in the intracellular PLP delivery process (7Yang E.S. Schirch V. Tight binding of pyridoxal 5'-phosphate to recombinant Escherichia coli pyridoxine 5'-phosphate oxidase.Arch. Biochem. Biophys. 2000; 377: 109-114Crossref PubMed Scopus (30) Google Scholar). A crystal structure of ePNPO (Protein Data Bank [PDB] code: 1G79), obtained from crystals soaked in a solution containing a large concentration of PLP (40 mM), revealed a putative secondary PLP-binding site located at the protein surface, about 11 Å from the active site (8Safo M.K. Musayev F.N. di Salvo M.L. Schirch V. X-ray structure of Escherichia coli pyridoxine 5'-phosphate oxidase complexed with pyridoxal 5'-phosphate at 2.0 A resolution.J. Mol. Biol. 2001; 310: 817-826Crossref PubMed Scopus (36) Google Scholar). However, so far the real correspondence of this binding site to the PLP tight binding site has not been experimentally confirmed. In this crystal structure, in which PLP is also bound at the active site, PLP at the secondary site is present in two positions, with an occupancy ratio of 0.7 and 0.3, respectively. In the higher occupancy position, the PLP ring is sandwiched between the side chains of Phe177 and Lys145, the pyridine nitrogen interacts with Asn84, and the phosphate group interacts with the amino group of Lys145 (8Safo M.K. Musayev F.N. di Salvo M.L. Schirch V. X-ray structure of Escherichia coli pyridoxine 5'-phosphate oxidase complexed with pyridoxal 5'-phosphate at 2.0 A resolution.J. Mol. Biol. 2001; 310: 817-826Crossref PubMed Scopus (36) Google Scholar) (Fig. S1). PLP in the lower occupancy position has a much weaker interaction with the protein and might result from crystal packing (8Safo M.K. Musayev F.N. di Salvo M.L. Schirch V. X-ray structure of Escherichia coli pyridoxine 5'-phosphate oxidase complexed with pyridoxal 5'-phosphate at 2.0 A resolution.J. Mol. Biol. 2001; 310: 817-826Crossref PubMed Scopus (36) Google Scholar). Here we demonstrate, through site-directed mutagenesis experiments, that this putative site is not involved in the tight, allosteric binding of PLP. On the other hand, we employed a quadruple ePNPO variant (K72I/Y129F/R133L/H199A), with an impaired active site, to obtain a crystal structure of the enzyme in which PLP is bound at a novel secondary site. We show that this coincides with the allosteric PLP tight binding site. The actual involvement of the putative secondary site identified by previous crystallographic studies (8Safo M.K. Musayev F.N. di Salvo M.L. Schirch V. X-ray structure of Escherichia coli pyridoxine 5'-phosphate oxidase complexed with pyridoxal 5'-phosphate at 2.0 A resolution.J. Mol. Biol. 2001; 310: 817-826Crossref PubMed Scopus (36) Google Scholar) in the allosteric binding of PLP was tested through site-directed mutagenesis. A double variant (K145A/F177A) and two triple variants (N84A/K145A/F177A and N84W/K145A/F177A) of ePNPO (Fig. S1) were produced and characterized. Initially, these variant forms were analyzed with respect to their structure, thermal stability, and catalytic properties, to confirm that the introduced replacements had not significantly altered the native structure of the enzyme. Far-UV CD spectra (Fig. S2) indicated that no substantial alterations in the secondary structure were caused by the variations. Differential scanning fluorimetry (DSF) measurements were carried out to determine and compare melting temperature (Tm) values of variant and wildtype enzymes. The obtained denaturation curves (Fig. S3A) were fitted to a sigmoidal equation (Equation 1) to obtain Tm values. The wildtype enzyme shows a Tm of 70.0 °C ± 0.04 deg. C. The double variant form shows a similar Tm value, whereas the melting temperature of the triple variants is about 2 °C lower (Table 1). These results indicate that the variations did not drastically alter the overall structure of the enzyme.Table 1Parameters obtained from the characterization of ePNPO variantsEnzyme formTm (°C)aDetermined in Tris buffer from kinetic analysis.KM (μM)aDetermined in Tris buffer from kinetic analysis.kCAT (s−1)bDetermined from fluorimetric PLP binding equilibrium measurements.KDPLP (μM)cDetermined from PLP retention measurements.PLP (%)Wildtype70.0 ± 0.041.6 ± 0.30.25 ± 0.030.15 (5)69 ± 19K145A/F177A70.9 ± 0.052.4 ± 0.50.17 ± 0.010.47 ± 0.0672 ± 8N84A/K145A/F177A66.7 ± 0.053.2 ± 0.10.07 ± 0.020.36 ± 0.0573 ± 10N84W/K145A/F177A66.6 ± 0.063.1 ± 0.20.05 ± 0.010.33 ± 0.0773 ± 19R23L70.0 ± 0.11.7 ± 0.70.09 ± 0.020.58 ± 0.1238 ± 11R215L69.0 ± 0.092.6 ± 0.80.05 ± 0.021.28 ± 0.2222 ± 7R23L-R24L71.1 ± 0.042.1 ± 0.40.19 ± 0.031.00 ± 0.0930 ± 8R23L/R215L68.9 ± 0.033.2 ± 0.70.05 ± 0.021.18 ± 0.1718 ± 9R23L-R24L-R215L68.0 ± 0.053.4 ± 0.20.09 ± 0.011.03 ± 0.1819 ± 3a Determined in Tris buffer from kinetic analysis.b Determined from fluorimetric PLP binding equilibrium measurements.c Determined from PLP retention measurements. Open table in a new tab The catalytic properties of the ePNPO variants were analyzed in 50 mM Tris-HCl buffer, pH 7.6, at 37 °C, using PNP as substrate. In these conditions, the produced PLP readily forms an aldimine complex with Tris, whose absorbance at 414 nm is followed to determine the initial velocity of the PNPO reaction (9Kwon O. Kwok F. Churchich J.E. Catalytic and regulatory properties of native and chymotrypsin-treated pyridoxine-5-phosphate oxidase.J. Biol. Chem. 1991; 266: 22136-22140Abstract Full Text PDF PubMed Google Scholar). Therefore, in Tris buffer PLP does not accumulate in the solvent and does not inhibit the enzyme activity by binding at the allosteric site (5Barile A. Tramonti A. di Salvo M.L. Nogués I. Nardella C. Malatesta F. Contestabile R. Allosteric feedback inhibition of pyridoxine 5'-phosphate oxidase from Escherichia coli.J. Biol. Chem. 2019; 294: 15593-15603Abstract Full Text Full Text PDF PubMed Scopus (5) Google Scholar). Initial velocity data as a function of substrate concentration were analyzed using the Michaelis–Menten equation, determining kCAT and KM values compared in Table 1. Although KM values obtained with the variant enzymes are similar to that of wildtype ePNPO, kCAT values are somewhat lower; in particular, in the case of the triple N84W/K145A/F177A variant, kCAT is about five times lower than that of wildtype ePNPO. Nevertheless, it is evident that the variations did not perturb the functional properties of the enzyme. Binding of PLP to ePNPO variants was also analyzed using the spectrofluorometric method previously employed for the wildtype enzyme (5Barile A. Tramonti A. di Salvo M.L. Nogués I. Nardella C. Malatesta F. Contestabile R. Allosteric feedback inhibition of pyridoxine 5'-phosphate oxidase from Escherichia coli.J. Biol. Chem. 2019; 294: 15593-15603Abstract Full Text Full Text PDF PubMed Scopus (5) Google Scholar), which is based on the resulting increase of FMN fluorescence emission. As we have previously demonstrated (5Barile A. Tramonti A. di Salvo M.L. Nogués I. Nardella C. Malatesta F. Contestabile R. Allosteric feedback inhibition of pyridoxine 5'-phosphate oxidase from Escherichia coli.J. Biol. Chem. 2019; 294: 15593-15603Abstract Full Text Full Text PDF PubMed Scopus (5) Google Scholar), PLP does not bind at the active site of wildtype ePNPO, most probably because binding at the allosteric site, which takes place with higher affinity, causes a conformational change that prevents binding at the active site. Therefore, the fluorimetric method only reveals binding of PLP at the allosteric site. PLP binding curves obtained with ePNPO variants (Fig. S5A) were fit to the quadratic Equation 2 (see Experimental procedures section), obtaining dissociation constant (KD) values (Table 1) that are moderately higher (0.33–0.47 μM) than that previously reported for wildtype ePNPO (0.15 μM) (5Barile A. Tramonti A. di Salvo M.L. Nogués I. Nardella C. Malatesta F. Contestabile R. Allosteric feedback inhibition of pyridoxine 5'-phosphate oxidase from Escherichia coli.J. Biol. Chem. 2019; 294: 15593-15603Abstract Full Text Full Text PDF PubMed Scopus (5) Google Scholar). This result is inconsistent with a role for the mutated residues in PLP binding at the allosteric site. It is known by previous experiments on wildtype ePNPO that, when the protein is incubated with PLP and then passed through an SEC column, it retains PLP tightly bound to it (7Yang E.S. Schirch V. Tight binding of pyridoxal 5'-phosphate to recombinant Escherichia coli pyridoxine 5'-phosphate oxidase.Arch. Biochem. Biophys. 2000; 377: 109-114Crossref PubMed Scopus (30) Google Scholar). This happens even when the active site is unable to bind PLP as a consequence of multiple variations (5Barile A. Tramonti A. di Salvo M.L. Nogués I. Nardella C. Malatesta F. Contestabile R. Allosteric feedback inhibition of pyridoxine 5'-phosphate oxidase from Escherichia coli.J. Biol. Chem. 2019; 294: 15593-15603Abstract Full Text Full Text PDF PubMed Scopus (5) Google Scholar), demonstrating that PLP binds tightly at an allosteric site. All ePNPO variants (100 μM) were incubated with an equimolar amount of PLP. Then, they were passed through an SEC column and the stoichiometry of protein-bound PLP with respect to protein subunits was determined. The percentage of protein-bound PLP was 72 ± 8, 73 ± 10, and 73 ± 19, with the K145A/F177A, N84A/K145A/F177A and N84W/K145A/F177A ePNPO variants, respectively (Table 1). Such values are similar to that obtained with the wildtype enzyme (69 ± 19%), strongly suggesting that the putative secondary PLP-binding site indicated by previous crystallographic studies does not correspond to the allosteric tight binding site. As explained in the introduction, a crystal structure of ePNPO in which PLP is bound at both the active site and a secondary site has been previously reported by other authors (8Safo M.K. Musayev F.N. di Salvo M.L. Schirch V. X-ray structure of Escherichia coli pyridoxine 5'-phosphate oxidase complexed with pyridoxal 5'-phosphate at 2.0 A resolution.J. Mol. Biol. 2001; 310: 817-826Crossref PubMed Scopus (36) Google Scholar). This structure was obtained by soaking the native crystals in a highly concentrated solution of PLP (40 mM). Under these conditions, it would be expected that PLP binds to the high-affinity allosteric site, the secondary site found by Safo and collaborators (8Safo M.K. Musayev F.N. di Salvo M.L. Schirch V. X-ray structure of Escherichia coli pyridoxine 5'-phosphate oxidase complexed with pyridoxal 5'-phosphate at 2.0 A resolution.J. Mol. Biol. 2001; 310: 817-826Crossref PubMed Scopus (36) Google Scholar) and the active site; therefore, it is surprising that the high-affinity allosteric site had not been revealed in previous crystallographic studies, unless the allosteric site was occluded by crystal packing. We carried out several cocrystallization trials with wildtype ePNPO and PLP in different conditions; however, we either did not obtain crystals or obtained crystals in which PLP was bound at the active site only. Therefore, we decided to change strategy and use a variant form of ePNPO in which the active site is impaired and not able to bind PLP. We recently showed that an ePNPO quadruple variant (K72I/Y129F/R133L/H199A; ePNPOqm), which has a drastically reduced capability to bind PNP (and therefore PLP) at the active site, is still able to bind PLP at the allosteric site (5Barile A. Tramonti A. di Salvo M.L. Nogués I. Nardella C. Malatesta F. Contestabile R. Allosteric feedback inhibition of pyridoxine 5'-phosphate oxidase from Escherichia coli.J. Biol. Chem. 2019; 294: 15593-15603Abstract Full Text Full Text PDF PubMed Scopus (5) Google Scholar). We crystallized this quadruple variant in both the absence of PLP (ePNPOqm) in P31 2 1 space group and presence of PLP (PLP-ePNPOqm) in P61 space group. The three-dimensional structures of both crystal forms were solved (Table 2). The final model of ePNPOqm was refined to 1.56 Å resolution; the asymmetric unit contains a 212-residue subunit (residues 7–218; the entire polypeptide chain is made by 218 residues), one FMN molecule, two sulphate ions, one phosphate ion, and 174 water molecules. The second subunit of the dimer was generated by the crystallographic 2-fold axis. The final model of the PLP-bound ePNPOqm (PLP-ePNPOqm) was refined to 2.42 Å resolution; the asymmetric unit contains a 353-residue dimer (residues/chain: 17/A–126/A; 176/A–218/A; 14/B–166/B; 170/B–218/B), one FMN molecule per subunit, one sulphate ion in subunit A, one PLP molecule with 0.81 occupancy in subunit B, and 45 water molecules. The quadruple mutant conserves the structural elements of the wildtype protein: five alpha helices, numbered from H1 to H5, and ten β strands. In addition, in the PLP-free ePNPOqm the N-terminal region is also visible, folded in an additional α helix (H0, residues 5–18). The subunit structure can be described, as reported by Safo et al., (10Safo M.K. Mathews I. Musayev F.N. di Salvo M.L. Thiel D.J. Abraham D.J. Schirch V. X-ray structure of Escherichia coli pyridoxine 5'-phosphate oxidase complexed with FMN at 1.8 A resolution.Structure. 2000; 8: 751-762Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar) in terms of two domains.The core of the larger domain is represented by a seven-stranded and a four-stranded antiparallel β sheet, structurally joined by the S4 strand (110–120) that is part of both sheets. The β sheets core region is flanked by H1 (and H0 in PLP-free ePNPOqm) on one side and H2 on the opposite side. The smaller domain packs adjacent to the β-barrel core and helix H2 and is made up by the three remaining α helices, H3, H4, and H5 (154–166) (Fig. 3A). In the PLP-ePNPOqm model, the subunit A backbone displays a major break of 49 residues (comprising residues 127–175), corresponding to the ePNPO smaller domain comprising the H3, H4, and H5 helices, whereas only a minor break of three residues is present in subunit B (residues 167–169), which could not be modelled because of the lack of information from the FO – FC map (Fig. 3B). No PLP is present at the active site of either structures (Fig. 3, A and B), although PLP-ePNPOqm was crystallized in the presence of excess PLP (see Experimental procedures section for details); this is evidently a consequence of the engineered active site variants (Fig. 3, C and D). In contrast to the symmetrical ePNPOqm structure, the PLP-ePNPOqm is the first ever characterized structure of the enzyme crystallized in a P61 space group, in which the dimer is asymmetric. This structural discrepancy resides in the major chain break contained in subunit A, comprising a portion of H3 (residues 123–130) and the whole H4 and H5 (residues 135–143 and 153–166, respectively) regions. The disorder of these elements in subunit A seems to occur as a consequence of PLP binding in a surface cleft, placed in part at the interface between subunits A and B, mainly circumscribed by three arginine residues (Arg23, Arg24, Arg215) belonging to subunit B and acting as an electrostatic "Arg cage" for PLP. The three Arg residues efficiently trap PLP in the cavity (Fig. 4, A and B). The binding driving force is represented by the phosphate group, as often occurs in PLP-binding proteins (11Denesyuk A.I. Denessiouk K.A. Korpela T. Johnson M.S. Functional attributes of the phosphate group binding cup of pyridoxal phosphate-dependent enzymes.J. Mol. Biol. 2002; 316: 155-172Crossref PubMed Scopus (58) Google Scholar), which is capable of establishing polar interactions with the amino groups of the side chains and backbone of Arg23, Arg215, and Arg24, respectively. In addition, a small network of hydrogen bonds takes place between Arg23 and the PLP phosphate group by means of a bridging water molecule, HOH419, whose role is apparently to contribute to stabilizing PLP lodging in the cavity; this effect is enhanced by the pyridine ring, able to coordinate Arg24 and Arg215 via the deprotonated nitrogen atom. This Arg cage was previously identified by Safo and collaborators (8Safo M.K. Musayev F.N. di Salvo M.L. Schirch V. X-ray structure of Escherichia coli pyridoxine 5'-phosphate oxidase complexed with pyridoxal 5'-phosphate at 2.0 A resolution.J. Mol. Biol. 2001; 310: 817-826Crossref PubMed Scopus (36) Google Scholar) (PDB code: 1G76, 1G77, 1G78, 1G79) as a binding site for a unique phosphate ion; the PLP phosphate group and this phosphate ion appear to bind in a very similar fashion, as only a slight displacement of Arg23 is detectable in the same binding site. The comparison between the corresponding Fo–Fc omit maps (contoured at 2.5σ) (Fig. 4A; Fig. S4) allows us to exclude the presence of a phosphate ion rather than PLP in this cavity of subunit B of the PLP-ePNPOqm structure. The superimposition of PLP-ePNPOqm and ePNPOqm revealed important differences between the two structures, highlighting the structural change induced by PLP binding to the allosteric site. Indeed, PLP binding causes the dismantling of a specific pattern of hydrogen bonds and electrostatic interactions anchoring H5 of subunit A to a short neighboring protein segment made of residues 192 to 200 of subunit B ("H-unc"; Fig. 5, A and B). In this structure of ePNPOqm without PLP, the interactions between two key residues of H5 (subunit A; Arg153 and Glu157) and residues Glu195, Leu198, and Asp200 placed on the 192 to 200 helix (subunit B) are particularly important to maintain these structural elements linked together and correctly folded (Fig. 5A). Binding of PLP to subunit A of ePNPOqm causes a displacement of the H-unc structural element (Cα of Glu195 moves 7.6 Å apart from its previous position), thereby disrupting the interactions of this structural element with H5 of subunit A, which loses its structure together with H3 and H4 (Fig. 5B). The displacement of H-unc upon PLP binding causes also the movement of Arg197 (placed on H-unc), which, in the wildtype ePNPO structure (PDB code: 1G76), contributes to the PLP binding to the catalytic site, establishing electrostatic interactions with both its aldehyde and hydroxyl groups. As shown in Figure 5C, in the ePNPOqm structure, the Arg197 displays a position similar to that of the wildtype protein. Upon PLP binding to the Arg-cage site, Arg197 shifts and its side chain rotates about 180° around its Cα with respect to its original position (Fig. 5C). The large conformational change of the 190 to 200 segment, and in particular the displacement of Arg197, alters the structure of the active site and therefore the affinity for both PLP and substrate, and as a consequence reduces the catalytic activity of the enzyme. Since the amino acid residues Arg23, Arg24, and Arg215 are crucial for the interaction with PLP in the PLP-ePNPOqm crystal structure, single variant forms (R23L and R215L) as well as two double (R23L/R24L and R23L/R215L) and a triple (R23L/R24L/R215L) ePNPO variants were produced through site-directed mutagenesis and characterized. All variants showed far-UV CD spectra superimposable to that of the wildtype ePNPO (Fig. S2), indicating that the variations did not affect the secondary structure of the protein. Data obtained from DSF measurements showed that all variants are quite similar to the wildtype enzyme with respect to thermal stability (Fig. S3B) (Table 1). With these variants, enzyme activity measurements yielded KM and kCAT values that, in some cases, are different with respect to those of wildtype ePNPO (Table 1); however, they testify to substantial preservation of the enzyme functional properties. PLP binding was analyzed through the same spectrofluorometric method described above (Fig. S5B). Noticeably, the PLP binding properties of all variants were found to be substantially altered, as shown by KD values that in some cases are 7- to 10-fold higher than that of the wildtype enzyme (Table 1). When ePNPO variants were incubated with PLP and passed through an SEC column, a considerably low percentage of PLP was retained: only 18% to 19% (with respect to protein subunits) in the case of the R23L/R215L and R23L/R24L/R215L variants, in comparison with the 70% value obtained with wildtype ePNPO (Table 1). As mentioned above, kinetic measurements of PNP oxidation into PLP catalyzed by PNPO are usually carried out in Tris buffer to prevent PLP accumulation and enzyme inhibition (9Kwon O. Kwok F. Churchich J.E. Catalytic and regulatory properties of native and chymotrypsin-treated pyridoxine-5-phosphate oxidase.J. Biol. Chem. 1991; 266: 22136-22140Abstract Full Text PDF PubMed Google Scholar). Studies carried out by our group showed that, when the reaction is carried out in Hepes buffer, a complex kinetics of PLP formation is observed as a consequence of PLP allosteric inhibition (5Barile A. Tramonti A. di Salvo M.L. Nogués I. Nardella C. Malatesta F. Contestabile R. Allosteric feedback inhibition of pyridoxine 5'-phosphate oxidase from Escherichia coli.J. Biol. Chem. 2019; 294: 15593-15603Abstract Full Text Full Text PDF PubMed Scopus (5) Google Scholar). Initially, upon mixing enzyme with PNP, the velocity of product formation decreases, resulting in a deceleration phase caused by the accumulation of PLP in the solvent, binding of PLP at the allosteric site, and the consequent onset of enzyme inhibition (Fig. 6A). The first deceleration phase is followed by a slower, approximately linear production of PLP. This compl
Referência(s)