Skip to main content
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2018 Aug 6.
Published in final edited form as: Nat Chem Biol. 2017 Jan 16;13(3):290–294. doi: 10.1038/nchembio.2273

Lysine relay mechanism coordinates intermediate transfer in vitamin B6 biosynthesis

Matthew J Rodrigues 1,2,#, Volker Windeisen 1,3,#, Yang Zhang 4, Gabriela Guédez 3, Stefan Weber 3, Marco Strohmeier 3, Jeremiah W Hanes 4,5, Antoine Royant 6,7, Gwyndaf Evans 2, Irmgard Sinning 3, Steven E Ealick 4, Tadhg P Begley 8, Ivo Tews 1,3,*
PMCID: PMC6078385  NIHMSID: NIHMS979246  PMID: 28092359

Abstract

Substrate channeling has emerged as a common mechanism for enzymatic intermediate transfer. A conspicuous gap in knowledge concerns the use of covalent lysine imines in the transfer of carbonyl-group-containing intermediates, despite their wide use in enzymatic catalysis. Here we show how imine chemistry operates in the transfer of covalent intermediates in pyridoxal 5΄-phosphate biosynthesis by the Arabidopsis thaliana enzyme Pdxl. An initial ribose 5-phosphate lysine imine is converted to the chromophoric l320 intermediate, simultaneously bound to two lysine residues and partially vacating the active site, which creates space for glyceraldehyde 3-phosphate to bind. Crystal structures show how substrate binding, catalysis and shuttling are coupled to conformational changes around strand β6 of the Pdxl (βα)8-barrel. The dual-specificity active site and imine relay mechanism for migration of carbonyl intermediates provide elegant solutions to the challenge of coordinating a complex sequence of reactions that follow a path of over 20 Å between substrate-and product-binding sites.


Pyridoxal 5΄-phosphate (PLP) is an active form of vitamin B6, which functions as an enzyme cofactor and antioxidant1,2. More than 140 different PLP-dependent enzymes catalyze reactions in prokaryotic amino acid and carbohydrate metabolism, and an estimated 1.5% of the genes in microbial genomes encode PLP-dependent enzymes3. The PLP synthase complex, which consists of the enzymes Pdxl and Pdx2, is conserved in all domains of life4,5. In striking contrast, the related Escherichia coli pathway, used only by a small number of bacteria, requires six enzymes to achieve the same biosynthesis4,6.

The Pdxl enzyme requires three substrates to catalyze the formation of PLP79. Ribose 5-phosphate (R5P) reacts with ammonia, which is produced by a separate Pdx2 glutaminase domain, to produce the chromophoric I320 intermediate with an absorbance maximum at 320 nm (ref. 10). I320 formation is followed by addition of the third substrate, glyceraldehyde 3-phosphate (G3P) (Fig. 1a)9. Pdxl has a (βα)8-barrel fold in a dodecameric assembly; each subunit contains two phosphate-binding sites, named P1 and P2, that define the active site architecture11,12 (Fig. 1b). P1 and P2 are separated by 21 Å (phosphorus to phosphorus) and are known to bind the phosphate groups of R5P, but the mechanism by which Pdxl transfers intermediates between the sites is unknown.

Figure 1|. The reaction catalyzed by the Pdxl and Pdx2 subunits of the PLP synthase complex and the structure of Pdxl.

Figure 1|

(a) Pyridoxal 5-phosphate synthase catalyzes the complex condensation reaction between ribose 5-phosphate, glyceraldehyde 3-phosphate and ammonia. (b) The overall structure of the Pdxl core. Pdxl forms a dodecamer from two interlocked hexameric rings. This core complex is shown in two orientations, with subunit boundaries and the two phosphate-binding sites for each protein monomer indicated. The phosphate-binding sites P1 and P2 are required for substrate and product binding, respectively, and are separated by 21 Å (phosphorus to phosphorus). A cartoon representation for one monomer illustrates the (βα)8-barrel fold of Pdxl.

The crystallographic studies described here reveal five Pdxl intermediate structures, and they have been used to propose a mechanism of intermediate transfer chemistry that is not evident from previous characterization of the I320 structure by MS analysis, phosphine trapping, and NMR analysis9,13. The I320 intermediate is simultaneously bound to two lysine residues, which allows the transfer between catalytic centers, and we describe structural changes to support intermediate formation and transfer.

RESULTS

The I320 intermediate is central to PLP biosynthesis

Various bacterial and eukaryotic Pdxl proteins were screened to identify the Arabidopsis Pdxl (ref. 14) crystal system that provided the Pdxl-intermediate complexes reported here (Fig. 2; Supplementary Results, Supplementary Fig. 1 and Supplementary Table 1). Although other crystal systems such as that in Thermus thermophilus yielded higher resolution diffraction (1.6 Å), they were not robust enough to survive crystal soaking experiments.

Figure 2|. Crystallographic structures of five covalent intermediates in PLP biosynthesis.

Figure 2|

Pdxl is shown in cartoon representation, and catalytic lysine side chains and intermediate atoms are shown in stick representation with carbon atoms in green (for Lys98 and Lys166) and orange (for intermediates), nitrogen atoms in blue, oxygen atoms in red, and phosphorus atoms in purple. 2Fo-Fc electron density maps are shown at 1 σ for the complexes. (a) Binding of R5P (first substrate) uses P1 and occurs by covalent attachment through Schiff base formation with Lys98. (b,c) Addition of ammonia (second substrate) leads to the formation of an intermediate in P1 in the K166R mutant (b), which converts to the I320 species in wild-type enzyme through formation of a second Schiff base with Lys166 (c). (d) Incorporation of G3P (third substrate) leads to a covalent complex with I320, with the G3P phosphate bound in the P1 site. (e) The PLP complex shows PLP covalently bound to the enzyme through Schiff base formation with Lys166, with its phosphate bound in the P2 site.

PLP formation is initiated with formation of the Pdxl-R5P complex, observed here at 1.9-Å resolution (Fig. 2a), and previously observed for Plasmodium berghei and Geobacillus stearothermophilus Pdxl (refs. 15,16). A covalent imine forms between the ε-ΝΗ2 of Lys98 and the R5P Cl, with the R5P phosphate bound in P1 (refs. 10,1517) (Fig. 2a). The proximity of the ε-ΝΗ2 group of Lys98 to the side chains of Aspll9 and Serl21 (Supplementary Fig. 2a,b) of the totally conserved DESE sequence motif in PLP synthases suggests that Aspll9 or Serl21 may play a role in Schiff base formation, but as of yet we are unable to clearly assign the base catalyzing this step.

Transient association with Pdxl activates Pdx2 to catalyze glutamine hydrolysis, releasing ammonia as a nitrogen source for PLP biosynthesis8,18,19. The ammonia produced by Pdx2 passes through a hydrophobic tunnel in the center of the Pdxl (βα)8-barrel to the Pdxl P1 active site; channeling of intermediates prevents their diffusion into the bulk solvent and is common in glutamine amidotrans-ferase enzymes such as PLP synthase11,15,20. Ammonia reacts with Pdxl-R5P to form the chromophoric I320 intermediate10,21. Since the product, PLP, binds in the P2 site22, we hypothesized that formation of the I320 intermediate is central to the migration of reaction intermediates from Lys98 (P1) to Lysl66 (P2).

Soaking Pdxl-R5P crystals with ammonium chloride generated Pdxl-I320 (Fig. 2c). While the addition of ammonium salts decouples Pdxl-I320 formation from Pdx2-dependent glutamine hydrolysis10, the change in osmotic pressure caused many crystals to become disordered. More than 1,000 crystals were tested to optimize the protocol used to provide both high-resolution diffraction and homogeneous accumulation of Pdxl in the I320 state. Online UV-Vis (visible) microspectrophotometry23 was performed to ensure that crystals used in diffraction experiments contained Pdxl predominantly in the I320 state (Fig. 3a).

Figure 3|. The dual-specificity binding site P1 and the product-binding site P2.

Figure 3|

(a) The UV-Vis spectrum of a Pdx1-I320 crystal shows an absorption maximum at 280 nm, corresponding to protein, and near 320 nm for the I320 intermediate. (b) The intermediate I320 is covalently bound to both Lys98 and Lys166, shown in two views rotated by 90°. The overlay with the R5P complex (faded) shows the different positioning of Lys166 and adjacent residues Thr165 and Glyl67. Selected amino acid atoms and the I320 atoms are shown in stick representation, with carbon atoms in green (for amino acids) and orange (for intermediates), nitrogen atoms in blue, oxygen atoms in red, and phosphorus atoms in purple. Arrows indicate the different conformations of the Lys166 side chains and the Lys166-Gly167 peptide between Pdxl-I320) and Pdx1-R5P complexes. Pep-flip, peptide flip. (c) The I320-G3P complex horseshoe-like intermediate. The intermediate is covalently attached to both Lys98 and Lys166. Formation of the G3P binding site requires prior formation of the I320 adduct to dissociate the C3-C5 atoms from the binding site, which allows the G3P phosphate to bind in the P1 site. The overlay with the R5P complex (faded) shows where the ribose and triose portions match. Atoms colored as described for Figure 3b. (d) The PLP adduct is covalently bound by a Schiff base to Lys166. 2Fo - Fc electron density map of the refined complex shown at 1 σ. Atoms colored as described for Figure 3b.

Pdxl-I320 crystals typically diffracted to 2.3–2.0-Å resolution and showed continuous electron density between Lys98 and Lysl66 and loss of the phosphate group (Figs. 2c and 3b). Interpretation of the highest-resolution data set at 1.7-Å resolution was aided by the earlier MS and NMR data showing that I320 retains all five carbons of the R5P substrate10, with Cl covalently bound to a nitrogen atom13, observed here to be the ε-ΝΗ2 of Lys98 (Fig. 3b).

Comparison of the 13C-NMR spectra of I320 reconstituted in vitro using 14N- and 15N-labeled ammonium chloride showed that the nitrogen incorporation occurred at the C2 atom13, which allowed assignment of nitrogen and oxygen atoms bound to C2 and C3 of the intermediate, respectively (Figs. 2c and 3b). The C3-bound oxygen atom of I320 is the sole oxygen that remains from the R5P substrate13. The C5 attachment of I320 to a nitrogen atom was previously interpreted as being caused by I320 decomposition under the denaturing and acidic conditions in the NMR experiment13; however, the crystallographic analysis now resolves the nature of the C5-N interaction, observed here to be a covalent bond between C5 and the ε-ΝΗ2 of Lysl66. Thus, the intermediate is not released from Lys98 before I320 formation as previously thought13, but is simultaneously bound to two lysine residues.

A network of polar interactions between side chains of Aspll9 and Glul22 of the DESE sequence motif and Argl64 (all residues completely conserved in PLP synthases) is in close proximity to the intermediates in the P1 site. Previous studies demonstrated that replacing Arg l64 in G. stearothermophilus Pdxl with alanine causes a 95% reduction in the rate of I320 formation16, and Argl64 is suitably positioned to activate the ε-ΝΗ2 group of Lysl66 for Schiff base formation in the Pdxl-I320 structure (Supplementary Fig. 2c). This conserved network of polar interactions is reminiscent of acetoacetate decarboxylase, a well-studied example of the electrostatic perturbation model for enzyme activation24.

Several examples of crystallographic structures with imine-bound cofactors are known to suffer from site-specific radiation damage25,26. Radiation damage may lead to artifacts in electron density maps that are not representative of the protein structure before X-ray exposure27. UV-Vis spectra of Pdxl-I320 in crystallo during X-ray irradiation were collected. The spectra of the crystals changed in response to irradiation, with 20% of the changes occurring in the first 245 kGy. While their occupancy is below 20%, the species generated by X-ray irradiation are not expected to have a substantial effect on the interpretation of electron density maps28. We constructed a 2.2-Å resolution multi-crystal data set using a protocol similar to that first described for investigation of site-specific radiation damage in horseradish peroxidase29. The X-ray dose absorbed during data collection was calculated using RADDOSE-3D30; BLEND was used to merge the diffraction data collected from each crystal below the dose threshold of 245 kGy into a single data set31. The resulting structure confirmed the nature of the bridging structure of I320 (Supplementary Fig. 3).

Conformational changes supporting intermediate transfer

Formation of Pdxl-I320 requires conformational changes in strand β6 of the Pdxl (βα)8-barrel, which results in reorientation of Lysl66 toward P1; this residue points toward the P2 site in the Pdxl-R5P structure (Fig. 3b; Supplementary Fig. 4a). We investigated the Pdxl K166R variant10,16 to better understand whether the structural transition is linked to ammonia incorporation. Sequential exposure of Pdxl K166R to R5P and ammonium chloride results in a complex termed Pdxl-pre-I320 (Fig. 2b). Like Lysl66 in the Pdxl-I320 complex, Argl66 reorients toward P1, and a peptide flip between Arg l66 and Gly l67 causes reorientation of the Thrl65 and Argl66 side chains as Pdxl transitions from the Pdxl-R5P state to the Pdxl-pre-I320 and Pdxl-I320 complexes (Fig. 3b; Supplementary Fig. 4b,c).

An earlier analysis predicted such changes for the wild-type enzyme by showing that the β6 strand lacks hydrogen bond stabilization from adjacent β-strands, allowing for additional flexibility in this region12. The recent structural analysis of G. stearothermophilus Pdxl also revealed that Lysl66 adopts these two conformations16.

These conformational changes are likely linked to the passage of ammonia through the transient hydrophobic tunnel in the (βα)8-barrel11,15. Notably, the absolutely conserved methionine residue in this tunnel, Met l62, is located on strand β6 together with the catalytic residue Lys l66 (ref. 15). In the previous study, the equivalent residue to Met l62 in Plasmodium berghei, Met l48, was exchanged for leucine, causing an increase in the catalytic rate of Pdx2-dependent PLP biosynthesis with no effect on the rate of NH4Cl-dependent catalysis, suggesting a coupling of Pdxl synthase and Pdx2 glutaminase activities involving Met l48. We propose that the disruption of Met l62 on the β6 strand by the passage of ammonia couples glutamine hydrolysis to the conformational changes required for I320 formation (Supplementary Fig. 4d).

Elimination of the R5P phosphate group bound in the P1 site immediately precedes the formation of the I320 intermediate9. This phosphate-binding site is located in the typical position for the (βα)8-fold at the C-terminal face of the β-barrel, and makes additional contacts with the β6-α6 loop that carries Lys l66 (refs. 11,12,32). The Pdxl-pre-I320 structure highlights the role of this phosphate group as a rigid anchor that, together with the Lys98 imine, ensures correct positioning of the intermediate for ammonia incorporation and covalent attachment to Lysl66 in the wild-type protein (Fig. 2b). Although incorporation of the ammonia nitrogen atom occurs in the Pdxl K166R variant, the different functionality of the Argl66 guanidino group prevents the catalysis of phosphate elimination10,16. The retention of the phosphate group explains the different UV-Vis absorption spectrum of the pre-I320 complex, which has a lower absorbance compared to Pdxl-I320 (ref. 10).

Dual-specificity binding site for pentose and triose

The structural rearrangements that are associated with I320 formation result in the displacement of carbon atoms C3, C4 and C5 of the intermediate away from the P1 phosphate-binding site (Fig. 3b), and the elimination reaction allows the R5P phosphate group to diffuse out of the P1 site. These actions create space for the binding of G3P in the P1 site. In the Pdxl-I320-G3P complex, G3P is covalently bound to the C2 nitrogen atom of I320, with the G3P phosphate in the P1 binding site (Fig. 2d and 3c). Surprisingly, the intermediate is covalently attached to both Lys98 and Lysl66, forming a horseshoe-like structure.

In solution, addition of G3P to Pdxl -I320 allows the reaction to proceed to the PLP product state; however, this did not occur in crystallo. The total loss of diffraction upon prolonged G3P exposure provides strong evidence for extensive structural changes that accompany the final steps of PLP formation. The C terminus of Pdxl is disordered in crystal structures and is known to participate in the final stages of PLP biosynthesis3234. The packing of Pdxl in the crystalline state may prevent the C terminus from adopting the conformation required for Pdxl to complete catalysis of PLP biosynthesis.

The Pdxl-PLP state can be observed by directly soaking Pdxl crystals with PLP. The 1.6-Å resolution electron density reveals an imine attachment of PLP to Lysl66 by the carbon atom that was initially the C5 of R5P (Figs. 2e and 3d). A noncovalent yeast Pdxl-PLP complex was previously reported22; however, examination of the deposited structure (PDB ID 3o05) shows conclusively that the PLP is covalently bound. The covalent nature of this complex is supported by UV-Vis spectroscopy of a Pdxl-PLP crystal that shows the characteristic absorption maximum at 408 nm as is typical for covalently bound PLP rather than 388 nm as is typical for free PLP (Supplementary Fig. 5)13.

The structural data presented here have been combined with the published biochemical biophysical data from Pdxl into a mechanistic proposal for catalysis and intermediate transfer (Fig. 4; Supplementary Fig. 6). The schematic shows the central role of the I320 intermediate, linking between side chains Lys98 and Lysl66, for transfer of the reaction between substrate- and product-binding sites.

Figure 4|. The central role of I320 intermediate transfer in vitamin B6 biosynthesis.

Figure 4|

The proposed Pdx1 reaction mechanism is shown as a schematic. Lys98 points toward P1, and Lys166 points toward P2 in the free enzyme. Structural transitions support reorientation of the side chain of Lys166 toward the P1 site at the time of ammonia incorporation, leading to formation of I320, creation of the G3P binding site, and the subsequent formation of the covalent I320-G3P complex. The reversion of the structural transitions around Lys166 lead to release of the I320-G3P intermediate from covalent attachment to Lys98 and observation of covalently bound PLP in the P2 site.

DISCUSSION

Pdxl is the first enzyme found to use a dual transamine intermediate to transfer a reaction across an extended active site within a (βα)8-barrel enzyme35, in addition to coupling delivery of ammonia from Pdx2 to Pdxl to I320 formation. Pdxl enzymes that have been studied thus far span bacterial species11,12,16,17, unicellular eukaryotes15,19,22,36,37 and plants14. The mechanism of I320 formation and intermediate transfer is assumed to be conserved given the high overall conservation of the Pdxl enzyme5, the consistency of findings across all kingdoms of life and, in particular, the conservation of Lys98 and Lysl66. Central to the function of Pdxl is the bridging structure of the I320 intermediate between these conserved lysine residues (Figs. 2c and 3b).

The structures reveal a mechanism for how Pdxl uses covalent tethers to prevent loss of intermediates to surrounding solvent, and explain how the enzyme maintains a high local concentration of substrate and protects the reactive I320 species. The use of covalent tethers is reminiscent of the protein-bound phospho-pantothenoyl thioesters that are used to transfer intermediates between active sites in the assembly line enzymology of fatty acid synthases38, polyketide synthases39 and non-ribosomal polypeptide synthases40. Pdxl uses a similar transfer strategy to those found in the glycine cleavage system41 and the classical pyruvate dehydrogenase complex42, where lipoic acid is used to chaperone intermediates between active sites. Similarly to Pdxl, these enzymes use covalent tethers to prevent the loss of substrates or intermediates to surrounding solvent and maintain the local concentration of substrate. In contrast to the previous examples, Pdxl transfers covalent intermediates within a single catalytic domain. The intricate relay mechanism displayed by the Pdxl subunit of PLP synthase allows the enzyme to maintain precise control of the complex reaction performed across multiple active sites.

A conspicuous absence in Nature’s repertoire is the use of lysine imines to channel carbonyl-group-containing intermediates; this is surprising because lysine imines are used extensively in enzymatic catalysis, and transimination reactions are facile. In this study we described PLP synthase as the first example of an enzyme that uses two lysine residues to shuttle intermediates between two active sites. Lys98 anchors intermediates in the P1 site for the first phase of PLP assembly. Lysl66 then adds and facilitates intermediate transfer to the P2 site, where completion of PLP formation occurs. This transfer strategy enables a single enzyme to catalyze the complex assembly process required for the formation of the pyridine ring of PLP and stands in contrast to the use of six enzymes in E. coli to achieve the same biosynthetic goal.

Note: Another X-ray structure of the I320 complex43 appeared while this paper was under revision.

METHODS

Methods, including statements of data availability and any associated accession codes and references, are available in the online version of the paper.

ONLINE METHODS

Molecular biology and protein production.

Pdxl.l and Pdxl.3, the two alleles of active Pdxl from Arabidopsis thaliana, were cloned as C-terminal his-tagged proteins in E. coli BL21 (DE3) cells as previously described, using the NdeI/XhoI restriction sites of the pET21a expression plasmid (Novagen)14. The expressed proteins (including tag) have molecular weights of 33,532 Da (Pdxl.l) and 33,886 Da (Pdxl.3). Cells were grown in 1 L cultures at 37 °C to an optical density (OD) of 0.6. Protein expression was induced using 60 ml of 25% (w/v) lactose and then grown for a further 16 h at 30 °C.

Protein purification.

Cells were lysed by sonication, and centrifuged for 1 h at 140,000 × g. Proteins were purified from the supernatant using 1 ml Immobilized Metal Affinity Chromatography HiTrap columns (GE Healthcare Life Sciences) loaded with nickel and equilibrated with lysis buffer (50 mM Tris-Cl, pH 7.5, 500 mM sodium chloride, 10 mM imidazole, 2% glycerol). The column was washed with wash buffer (50 mM Tris-Cl, pH 7.5, 500 mM sodium chloride, 50 mM imidazole, 2% glycerol) and the protein was eluted in elution buffer (50 mM Tris-Cl pH 7.5, 500 mM sodium chloride, 500 mM imidazole, 5% glycerol). For subsequent size-exclusion chromatography a Superdex 26/60 column (GE Healthcare Life Sciences) was equilibrated with gel filtration buffer (20 mM Tris-Cl pH 8.0, 200 mM KCl).

Complex preparation.

Pdxl-R5P was prepared using crystals of Pdx1.3 grown in 0.5 M sodium citrate buffered with 0.1 M HEPES, pH 7.5. These crystals were soaked in 2.5 mM R5P. K166R-pre-I320 was prepared using pre-formed crystals of K166R Pdxl.l grown in 100 mM sodium cacodylate pH 6.5 and 15% (w/v) PEG 8000.0.5 μΐ of 1:1 100 mM R5P and mother liquor were then added to 1 μl drops containing the crystals, leading to a final concentration of about 15 mM R5P in the drop. The well was closed for 5 min, allowing for equilibration of the sample. 0.5 μl of 1:1 1 M NH4Cl and mother liquor were added to the same drop resulting in a final drop size of 1.5 μΐ and final NH4C1 concentration of 125 mM. The well was sealed and allowed to equilibrate for 2 h. Pdxl-I320 was prepared using pre-formed crystals of Pdxl.3 grown in 100 mM Tris-Cl pH 8.5 and 12.25% PEG 4000 (w/v). 0.5 μl of 50 mM R5P in mother liquor was added to 2 μl drops containing Pdxl.3 crystals resulting in a final concentration of 10 mM R5P in the drop. The well was sealed and equilibrated for 30 min. 0.5 μl of 500 mM NH4Cl in mother liquor were then added to the same well with Pdxl.3 crystals, resulting in a final NH4C1 concentration of 100 mM in the drop. The well was sealed and equilibrated for 4 d. Pdxl -I320-G3P was prepared starting with crystals of Pdxl-I320 described above. After 4 d, the Pdxl-I320 crystals were transferred to a solution containing mother liquor containing 10 mM G3P and allowed to equilibrate for 150 s. Cryoprotectant buffer containing 10 mM G3P was then added to stabilize the crystals before flash cooling in liquid nitrogen. The Pdxl-PLP complex was prepared by adding 10 mM PLP to preformed Pdxl.3 crystals grown in 0.1 M Tris-Cl, pH 8.25, containing 8% PEG 8000.

In all cases except for the I320-G3P complex, cryobuffer containing the well solution and an additional 20% glycerol was added, and crystals were either flash cooled in liquid nitrogen (R5P, I320, PLP complexes) or directly transferred into the cryostream (pre-I320 complex).

Data collection and structure determination.

Offline and online UV-Vis absorption spectra were recorded at the European Synchrotron Radiation Facility (ESRF), France, on beamline ID14–4 (ref. 44) and at the ID29S-Cryobench laboratory23,45. Diffraction data for the Pdxl-R5P data set were collected on beamline ID14–1 (ref. 46) of the ESRF at a wavelength of 0.9760 Å. Diffraction data for the Pdxl K166R-pre-I320 data set were collected on beamline ID29 (ref. 47) of the ESRF, at a wavelength of 0.933 Å. Diffraction data for the Pdxl-I320 crystal single crystal data set were collected on beamline I04–1 of the Diamond Light Source, UK, at a wavelength of 0.9173 Å. Diffraction data for the Pdxl-I320 multi-crystal data set were collected on beamline ID23–1 (ref. 48) of the ESRF, at a wavelength of0.9763 Å. Diffraction data for the Pdxl-I320-G3P crystal were collected on beamline ID23–1 of the ESRF at a wavelength of 0.9763 Å. Diffraction data for the Pdxl-PLP data set were collected on beamline ID14–1 of the ESRF, at a wavelength of 0.9334 A. All data sets were collected at a temperature of 100 K. Data integration and scaling was carried out with XDS, DIALS and AIMLESS4951. Structure determination by molecular replacement used bacterial Pdxl from Bacillus subtilis4 as the search model (PDB ID 2NV1) and MOLREP52. Iterative model building and refinement were carried out using COOT53, REFMAC5 (ref. 54) and PHENIX55. Pdxl crystallized in a R3 unit cell with four monomers in the asymmetric unit. All monomers were first refined individually, i.e., without non-crystallographic symmetry (NCS) constraints, and inspected. The crystals selected in this study represent structures that were uniform with respect to accumulation of intermediates over all four monomers. For I320 multi-crystal and I320-G3P, the final round of refinement used the Cartesian protocol in PHENIX with Chain A as reference and NCS constraints over Lys98, Lysl66 and the intermediate. Dictionary files defining ligand geometry restraints for refinement were created using J Ligand56. All further data manipulations were carried out by the CCP4 suite of programs57. The refined crystal structures include Ramachandran favored residues (outliers) as Pdxl-R5P 98.8% (0.0%), Pdxl K166R-pre-I320 98.7% (0.0%), single crystal Pdxl-I320 98.8% (0.0%), multi-crystal Pdxl-I320 97.6% (0.0%), Pdxl-I320-G3P 98.7% (0.0%), Pdxl-PLP 98.3% (0.0%). The illustrations in the main text were prepared using PyMOL (Schrödinger, LLC). Data collection and refinement statistics are shown in Supplementary Table 1. Coordinates and structure factors have been deposited in the Protein Data Bank under the accession codes 5LNS (Pdxl-R5P), 5LNT (Pdxl K166R-pre-I320), 5LNU (Pdxl-I320-G3P), 5LNW (Pdxl-I320-G3P), 5LNR (Pdxl-PLP).

Multi-crystal analysis to assess beam damage.

About 50 complete data sets for I320 complexes were collected at ID23–1, ESRF. Data collected before the 245 kGy threshold were integrated using XDS49; all X-ray doses were calculated using RADDOSE-3D58. BLEND was run in analysis mode to group data sets with similar unit cell dimensions31. Grouped wedges of data were input to synthesis mode to identify groups that merged well. A core group of five wedges with good merging statistics (RMEAS 16.5, RPIM 11.0, 62.3% complete, 2.0-Å resolution) was identified. BLEND was then run in combination mode to add wedges to the core group individually and assess whether any improvements in completeness could be made without significantly worsening merging statistics. This resulted in a 19-wedge data set from nine crystals (RMEAS 20.1, 99.5% complete, 2.2-Å. resolution) leading to a low-dose structure of the I320 complex. Coordinates and structure factors of the multi-crystal Pdxl-I320 complex have been deposited in the Protein Data Bank under the accession code 5LNV.

Supplementary Material

SI

Acknowledgments

We thank W.J. Anderson, H. Clarke, C. Phippen (Southampton), A. Wessling and N. Kwak (Heidelberg), for their experimental contributions; T. Fitzpatrick (Geneva, Switzerland) for the generous gift of the expression plasmid for Arabidopsis Pdxl.3; S. Findlow and C. Holes at the Macromolecular Crystallisation Facility, Centre for Biological Sciences, and P. Horton and S. Coles at the Southampton Diffraction Centre, Chemistry, both University of Southampton; J. Kopp and C. Siegmann from the crystallization platform of the Cluster of Excellence CellNetworks, Heidelberg; staff at the Diamond Light Source and the European Synchrotron Radiation Facility for access and excellent user support without which this project would not have been possible; P. Carpentier, M. Weik, G. Gotthard and D. von Stetten at ESRF for support during data collection and online spectroscopy and D. Flot and G. Leonard for flexible access to the ESRF beamlines; A. Douangamath, P. Aller, R. Owen and M. Walsh for support at Diamond beamlines; B. Kappes (Erlangen) and P. Macheroux (Graz) for critical discussions; and L. Kinsland for assistance in preparation of the manuscript. This work was supported in parts by grants by the European Commission (VITBIOMAL-012158) and by the Deutsche Forschungsgemeinschaft (DFG) (TE368) to I.T., by the NIH (DK44083) and by the Robert A. Welch Foundation (A-0034) to T.P.B., by ESRF Mxl461, Mxl732, RADDAM and Diamond Light Source Mx8891 to I.T. M.J.R. was supported by a joint studentship between Diamond Light Source and the University of Southampton.

Footnotes

Author contributions

M.J.R., V.W, G.G., S.W, M.S. and J.WH. performed protein expression, purification and enzymatic essays. M.J.R., V.W, G.G., S.W, M.S. and I.T. performed crystallization and X-ray diffraction experiments. M.J.R., V.W, A.R. and I.T. performed online spectroscopy experiments. M.J.R., VW, Y.Z., I.S., S.E.E., T.P.B. and I.T. performed crystallographic analysis and data deposition. M.J.R., A.R., G.E., T.P.B. and I.T. performed spectroscopic data analysis. M.J.R., G.E., I.S., S.E.E., T.P.B. and I.T. wrote the paper. M.J.R. and V.W contributed equally to this work.

Competing financial interests

The authors declare no competing financial interests.

Additional information

Any supplementary information, chemical compound information and source data are available in the online version of the paper. Reprints and permissions information is available online at http://www.nature.com/reprints/index.html. Correspondence and requests for materials should be addressed to I.T.

Accession codes. Coordinates and structure factors have been deposited in the Protein Data Bank under the accession codes 5LNS (Pdxl-R5P), 5LNT (Pdxl K166R-pre-I320), 5LNU (Pdxl-I320-G3P), 5LNV (Pdxl-I320 multi-crystal), 5LNW (Pdxl-I320-G3P), 5LNR (Pdxl-PLP).

References

  • 1.Ehrenshaft M, Bilski P, Li MY, Chignell CF & Daub ME A highly-conserved sequence is a novel gene involved in de novo vitamin B6 biosynthesis. Proc. Natl. Acad. Sci. USA 96, 9374–9378 (1999). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Mooney S & Hellmann H Vitamin B6: killing two birds with one stone? Phytochemistry 71, 495–501 (2010). [DOI] [PubMed] [Google Scholar]
  • 3.Percudani 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 10, 273 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Mittenhuber G Phylogenetic analyses and comparative genomics of vitamin B6 (pyridoxine) and pyridoxal phosphate biosynthesis pathways. J Mol. Microbiol Biotechnol 3, 1–20 (2001). [PubMed] [Google Scholar]
  • 5.Das S et al. Biology’s new Rosetta stone. Nature 385, 29–30 (1997). [DOI] [PubMed] [Google Scholar]
  • 6.Fitzpatrick TB et al. Two independent routes of de novo vitamin B6 biosynthesis: not that different after all. Biochem. J 407, 1–13 (2007). [DOI] [PubMed] [Google Scholar]
  • 7.Raschle T, Amrhein N & Fitzpatrick TB On the two components of pyridoxal 5´-phosphate synthase from Bacillus subtilis. J. Biol. Chem 280, 32291–32300 (2005). [DOI] [PubMed] [Google Scholar]
  • 8.Burns KE, Xiang Y, Kinsland CL, McLafferty FW & Begley TP Reconstitution and biochemical characterization of a new pyridoxal-5´-phosphate biosynthetic pathway. J. Am. Chem. Soc 127, 3682–3683 (2005). [DOI] [PubMed] [Google Scholar]
  • 9.Hanes JW et al. Mechanistic studies on pyridoxal phosphate synthase: the reaction pathway leading to a chromophoric intermediate. J. Am. Chem. Soc 130, 3043–3052 (2008). [DOI] [PubMed] [Google Scholar]
  • 10.Raschle T et al. Reaction mechanism of pyridoxal 5´-phosphate synthase. Detection of an enzyme-bound chromophoric intermediate. J. Biol. Chem 282, 6098–6105 (2007). [DOI] [PubMed] [Google Scholar]
  • 11.Strohmeier M et al. Structure of a bacterial pyridoxal 5´-phosphate synthase complex. Proc. Natl. Acad. Sci. USA 103, 19284–19289 (2006). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Zhu J, Burgner JW, Harms E, Belitsky BR & Smith JL A new arrangement of (β/α)8 barrels in the synthase subunit of PLP synthase. J. Biol. Chem 280, 27914–27923 (2005). [DOI] [PubMed] [Google Scholar]
  • 13.Hanes JW, Keresztes I & Begley TP 13C NMR snapshots of the complex reaction coordinate of pyridoxal phosphate synthase. Nat. Chem. Biol 4, 425–430 (2008). [DOI] [PubMed] [Google Scholar]
  • 14.Tambasco-Studart M et al. Vitamin B6 biosynthesis in higher plants. Proc. Natl. Acad. Sci. USA 102, 13687–13692 (2005). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Guédez G et al. Assembly of the eukaryotic PLP-synthase complex from Plasmodium and activation of the Pdxl enzyme. Structure 20, 172–184 (2012). [DOI] [PubMed] [Google Scholar]
  • 16.Smith AM, Brown WC, Harms E & Smith JL Crystal structures capture three states in the catalytic cycle of a pyridoxal phosphate (PLP) synthase. J. Biol. Chem 290, 5226–5239 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Zein F et al. Structural insights into the mechanism of the PLP synthase holoenzyme from Thermotoga maritime. Biochemistry 45, 14609–14620 (2006). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Wallner S, Neuwirth M, Flicker Κ·, Tews I & Macheroux P Dissection of contributions from invariant amino acids to complex formation and catalysis in the heteromeric pyridoxal 5-phosphate synthase complex from Bacillus subtilis. Biochemistry 48, 1928–1935 (2009). [DOI] [PubMed] [Google Scholar]
  • 19.Gengenbacher M et al. Vitamin B6 biosynthesis by the malaria parasite Plasmodium falciparum: biochemical and structural insights. J. Biol Chem 281, 3633–3641 (2006). [DOI] [PubMed] [Google Scholar]
  • 20.Raushel FM, Thoden JB & Holden HM Enzymes with molecular tunnels. Acc. Chem. Res 36, 539–548 (2003). [DOI] [PubMed] [Google Scholar]
  • 21.Hanes JW, Keresztes I & Begley TP Trapping of a chromophoric intermediate in the Pdxl-catalyzed biosynthesis of pyridoxal 5´-phosphate. Angew. Chem. Int. Ed. Engl 47, 2102–2105 (2008). [DOI] [PubMed] [Google Scholar]
  • 22.Zhang X et al. Structural insights into the catalytic mechanism of the yeast pyridoxal 5-phosphate synthase Snzl. Biochem. J 432, 445–450 (2010). [DOI] [PubMed] [Google Scholar]
  • 23.von Stetten D et al. In crystallo optical spectroscopy (icOS) as a complementary tool on the macromolecular crystallography beamlines of the ESRF. Acta Crystallogr. D Biol. Crystallogr 71, 15–26 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Ho M-C, Ménétret J-E, Tsuruta H & Allen KN The origin of the electrostatic perturbation in acetoacetate decarboxylase. Nature 459, 393–397 (2009). [DOI] [PubMed] [Google Scholar]
  • 25.Borshchevskiy V et al. Low-dose X-ray radiation induces structural alterations in proteins. Acta Crystallogr. D Biol. Crystallogr 70, 2675–2685 (2014). [DOI] [PubMed] [Google Scholar]
  • 26.Dubnovitsky AP, Ravelli RB, Popov AN & Papageorgiou AC Strain relief at the active site of phosphoserine aminotransferase induced by radiation damage. Protein Sci. 14, 1498–1507 (2005). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Weik M et al. Specific chemical and structural damage to proteins produced by synchrotron radiation. Proc. Natl. Acad. Sci. USA 97, 623–628 (2000). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Pearson AR & Owen RL Combining X-ray crystallography and single-crystal spectroscopy to probe enzyme mechanisms. Biochem. Soc. Trans 37, 378–381 (2009). [DOI] [PubMed] [Google Scholar]
  • 29.Berglund GI et al. The catalytic pathway of horseradish peroxidase at high resolution. Nature 417, 463–468 (2002). [DOI] [PubMed] [Google Scholar]
  • 30.Zeldin OB, Gerstel M & Garman E E Optimizing the spatial distribution of dose in X-ray macromolecular crystallography. J. Synchrotron Radiat 20, 49–57 (2013). [DOI] [PubMed] [Google Scholar]
  • 31.Foadi J et al. Clustering procedures for the optimal selection of data sets from multiple crystals in macromolecular crystallography. Acta Crystallogr. D Biol. Crystallogr 69, 1617–1632 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Derrer B et al. Defining the structural requirements for ribose 5-phosphate-binding and intersubunit cross-talk of the malarial pyridoxal 5-phosphate synthase. FEBS Lett. 584, 4169–4174 (2010). [DOI] [PubMed] [Google Scholar]
  • 33.Raschle T et al. Intersubunit cross-talk in pyridoxal 5´-phosphate synthase, coordinated by the C terminus of the synthase subunit. J. Biol. Chem 284, 7706–7718 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Moccand C, Kaufmann M & Fitzpatrick TB It takes two to tango: defining an essential second active site in pyridoxal 5´-phosphate synthase. PLoS One 6, el6042 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Nagano N, Orengo CA & Thornton JM One fold with many functions: the evolutionary relationships between TIM barrel families based on their sequences, structures and functions. J. Mol. Biol 321, 741–765 (2002). [DOI] [PubMed] [Google Scholar]
  • 36.Reeksting SB et al. Exploring inhibition of Pdxl, a component of the PLP synthase complex of the human malaria parasite Plasmodium falciparum. Biochem. J 449, 175–187 (2013). [DOI] [PubMed] [Google Scholar]
  • 37.Neuwirth M et al. X-ray crystal structure of Saccharomyces cerevisiae Pdxl provides insights into the oligomeric nature of PLP synthases. FEBS Lett. 583, 2179–2186 (2009). [DOI] [PubMed] [Google Scholar]
  • 38.Jenni S et al. Structure of fungal fatty acid synthase and implications for iterative substrate shuttling. Science 316, 254–261 (2007). [DOI] [PubMed] [Google Scholar]
  • 39.Wu N, Tsuji SY, Cane DE & Khosla C Assessing the balance between protein-protein interactions and enzyme-substrate interactions in the channeling of intermediates between polyketide synthase modules. J. Am. Chem. Soc 123, 6465–6474 (2001). [DOI] [PubMed] [Google Scholar]
  • 40.Tanovic A, Samel SA, Essen LO & Marahiel MA Crystal structure of the termination module of a nonribosomal peptide synthetase. Science 321, 659–663 (2008). [DOI] [PubMed] [Google Scholar]
  • 41.Pares S, Cohen-Addad C, Sieker L, Neuburger M & Douce R X-ray structure determination at 2.6-A resolution of a lipoate-containing protein: the H-protein of the glycine decarboxylase complex from pea leaves. Proc. Natl. Acad. Sci. USA 91, 4850–4853 (1994). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Reed LJ Multienzyme complexes. Acc. Chem. Res 7, 40–46 (1974). [Google Scholar]
  • 43.Robinson GC, Kaufmann M, Roux C & Fitzpatrick TB Structural definition of the lysine swing in Arabidopsis thaliana PDX1: intermediate channeling facilitating vitamin B6 biosynthesis. Proc. Natl. Acad. Sci. USA 113, E5821–E5829 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.McCarthy AA et al. A decade of user operation on the macromolecular crystallography MAD beamline ID 14–4 at the ESRF J. Synchrotron Radiat 16, 803–812 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.McGeehan J et al. Colouring cryo-cooled crystals: online microspectrophotometry. J. Synchrotron Radiat 16, 163–172 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Wakatsuki S et al. ID 14 ‘Quadriga’, a beamline for protein crystallography at the ESRF. J. Synchrotron Radiat 5, 215–221 (1998). [DOI] [PubMed] [Google Scholar]
  • 47.de Sanctis D et al. ID29: a high-intensity highly automated ESRF beamline for macromolecular crystallography experiments exploiting anomalous scattering. J. Synchrotron Radiat 19, 455–461 (2012). [DOI] [PubMed] [Google Scholar]
  • 48.Nurizzo D et al. The ID23–1 structural biology beamline at the ESRF. J. Synchrotron Radiat 13, 227–238 (2006). [DOI] [PubMed] [Google Scholar]
  • 49.Kabsch W XDS. Acta Crystallogr. D Biol. Crystallogr 66, 125–132 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Evans PR & Murshudov GN How good are my data and what is the resolution? Acta Crystallogr. D Biol. Crystallogr 69, 1204–1214 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Gildea RJ et al. New methods for indexing multi-lattice diffraction data. Acta Crystallogr. D Biol. Crystallogr 70, 2652–2666 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Vagin A & Teplyakov A MOLREP: an automated program for molecular replacement. J. Appl. Crystallogr 30, 1022–1025 (1997). [Google Scholar]
  • 53.Emsley P & Cowtan K Coot: model-building tools for molecular graphics. Acta Crystallogr. D Biol. Crystallogr 60, 2126–2132 (2004). [DOI] [PubMed] [Google Scholar]
  • 54.Murshudov GN, Vagin AA & Dodson EJ Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr. D Biol. Crystallogr 53, 240–255 (1997). [DOI] [PubMed] [Google Scholar]
  • 55.Adams PD et al. The Phenix software for automated determination of macromolecular structures. Methods 55, 94–106 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.Lebedev AA et al. JLigand: a graphical tool for the CCP4 template-restraint library. Acta Crystallogr. D Biol. Crystallogr 68, 431–440 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57.Winn MD et al. Overview of the CCP4 suite and current developments. Acta Crystallogr. D Biol. Crystallogr 67, 235–242 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 58.Zeldin OB, Gerstel M & Garman EF RADDOSE-3D: time- and space-resolved modelling of dose in macromolecular crystallography. J. Appl. Crystallogr 46, 1225–1230 (2013). [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

SI

RESOURCES