Dihydrodipicolinate reductase (DHDPR) catalyses the second reaction in the diaminopimelate pathway of lysine biosynthesis in bacteria and plants. In contrast with the tetrameric bacterial DHDPR enzymes, we show that DHDPR from Vitis vinifera (grape) and Selaginella moellendorffii are dimeric in solution. In the present study, we have also determined the crystal structures of DHDPR enzymes from the plants Arabidopsis thaliana and S. moellendorffii, which are the first dimeric DHDPR structures. The analysis of these models demonstrates that the dimer forms through the intra-strand interface, and that unique secondary features in the plant enzymes block tetramer assembly. In addition, we have also solved the structure of tetrameric DHDPR from the pathogenic bacteria Neisseria meningitidis. Measuring the activity of plant DHDPR enzymes showed that they are much more prone to substrate inhibition than the bacterial enzymes, which appears to be a consequence of increased flexibility of the substrate-binding loop and higher affinity for the nucleotide substrate. This higher propensity to substrate inhibition may have consequences for ongoing efforts to increase lysine biosynthesis in plants.

Introduction

Enzymes of the diaminopimelate (DAP) pathway of lysine biosynthesis are of interest as metabolic engineering targets for the prospect of increasing the nutritional value of crops, given that lysine is a nutritionally limiting amino acid [16]. Inhibitors of these enzymes are also of interest for the development of new antibiotics, as products of this pathway are key components in the peptidoglycan layer of the bacterial cell wall [79]. The enzyme encoded by the dapB gene, dihydrodipicolinate reductase (DHDPR), catalyses the second step of this pathway, and hence is one such target.

DHDPR is now recognised to possess dehydratase activity [10], catalysing the conversion of (4S)-4-hydroxy-2,3,4,5-tetrahydro-(2S)-dipicolinic acid (HTPA), as opposed to dihydrodipicolinate (DHDP), to 2,3,4,5-tetrahydrodipicolinate (THDP). The mechanism by which this essentially irreversible reaction occurs has not been fully elucidated, but elimination of the hydroxyl group on HTPA to give a conjugated diene (DHDP), followed by reduction in the double bond utilising a pyridine nucleotide cofactor, is speculated to be the most plausible possibility [10]. Binding events at DHDPR occur via a compulsory-order ternary complex mechanism, where binding of the nucleotide cofactor necessarily precedes binding of the HTPA substrate [1114]. DHDPR is unusual among oxidoreductases in that it shows little cofactor specificity, utilising either phosphorylated or unphosphorylated nucleotides (NADPH or NADH) relatively readily [14,15]. Some DHDPR orthologues are inhibited by high concentrations of HTPA in a nucleotide-specific manner, and this is proposed to be the result of substrate prematurely binding to enzyme that still has oxidised nucleotide bound in the active site following THDP release from the previous catalytic cycle [11].

The family of DHDPR enzymes populate either obligate dimeric or tetrameric states. The widely studied bacterial DHDPR from Escherichia coli [13,16], Mycobacterium tuberculosis [15,17], Thermotoga maritima [18], Anabaena variabilis [19], and Staphylococcus aureus [12,20,21] have previously been shown to be tetrameric. In contrast, the two previously studied plant enzymes from maize (Zea mays) and Arabidopsis (Arabidopsis thaliana) are dimeric [22,23].

Crystal structures have been solved for DHDPR from a total of nine bacterial species, whereas no plant DHDPR structures were previously available. Each monomeric subunit consists of a Rossmann-like N-terminal domain, where the nucleotide cofactor binds, and a mixed α + β C-terminal substrate-binding domain. The N- and C-terminal domains are connected by two short hinge regions which allow a conformational change from an open to a closed conformation bringing the cofactor and substrate-binding sites together, enabling catalysis to take place [8,1517,20]. The four β-strands from the C-terminal domain of each subunit come together to form a 16-stranded β-barrel in the tetramer, which is stabilised by hydrogen bonding and hydrophobic interactions between the β-strands and the surrounding loops [12,15,16]. The length and orientation of the C- and N-terminal loops varies between different species [15,18,20] and have been postulated to affect the stability of the enzyme [18].

In the present paper, we sought to determine the structural arrangement of the dimeric DHDPR from plants and to elucidate the basis behind substrate inhibition of DHDPR. To extend previous studies of plant DHDPR, we have characterised DHDPR from grape (Vitis vinifera) and Selaginella moellendorffii. S. moellendorffii is a representative of an early lineage of vascular plants, which may provide information about when the plant DHDPR structure diverged from the bacterial structure. Given that different bacterial DHDPR enzymes do not show substrate inhibition [13,15], show intermediate substrate inhibition [11,19], or are strongly inhibited by the substrate [18], DHDPR from the pathogenic bacteria, Neisseria meningitidis (Nme-DHDPR) was also characterised. In addition to being of interest as an antibiotic target [8], Nme-DHDPR serves to increase the understanding of substrate inhibition in bacterial enzymes. We have utilised analytical ultracentrifugation, small-angle X-ray scattering (SAXS) and X-ray crystallography to demonstrate how small elements of secondary structure are able to prevent the assembly of the plant enzymes into a tetramer. Additionally, we use kinetic assays and an analysis of the structural flexibility of the proteins to suggest reasons for the differences seen in the level of substrate inhibition.

Experimental procedures

Overexpression and purification

The dapB gene from N. meningitidis MC58 (serotype B, ATCC) was amplified from genomic DNA by PCR, and the product was inserted into a pET151-D expression vector (Invitrogen). Genes encoding the dapB genes from S. moellendorffii (accession no. XP_002966640) and V. vinifera (accession no. XP_002280402), exclusive of the chloroplast transit peptide identified by ChloroP [24], were synthesised (Epoch Life Sciences) and cloned into the pET28a expression vector containing a tobacco etch virus cleavage site following a His-6 tag. Plasmids were transformed into E. coli BL21 (DE3) cells and expression was induced with 0.1 mg ml−1 IPTG for ∼16 h at 25°C. Harvested cells were resuspended in lysis buffer (30 mM NaHPO4, 500 mM NaCl, and 30 mM imidazole, pH 8.0). Protein was purified as described previously [23], with the addition of a final size-exclusion chromatography step using a Superdex 200 10/300 column equilibrated with 20 mM Tris–HCl, 150 mM NaCl (pH 8.0). Ath-DHDPR was purified as previously described [23]. Selenomethionine (SeMet)-derivatised protein of Smo-DHDPR (DHDPR from S. moellendorffii) was produced by the inhibition method. Briefly, cells were grown using the PASM-5052 media [25]; however, lactose was omitted and protein expression was induced by the addition of 0.1 mM IPTG and purified in the same manner as described above.

Analytical ultracentrifugation

Sedimentation velocity experiments were performed in a Beckman Coulter Model XL-I analytical ultracentrifuge. Samples (0.5 mg ml−1) were prepared in 20 mM Tris–HCl and 150 mM NaCl (pH 8.0). Proteins were centrifuged at 50 000 rpm at 20°C, and radial absorbance data were collected at appropriate wavelengths in the continuous mode without averaging. Data were fitted to a continuous mass coefficient distribution [c(m)] model using the program SEDFIT [26,27]. The partial specific volume of the proteins, buffer density, and buffer viscosity were computed using the program SEDNTERP [28].

Crystallisation and cryoprotection

Protein crystallisation experiments for Smo-DHDPR were performed at the CSIRO Collaborative Crystallisation Centre (www.csiro.au/C3), Melbourne, Australia using the PACT Suite and the JCSG+ suite of crystal screens at 281 and 293 K. The screens were set up using the sitting-drop vapour-diffusion method. Crystals used for data collection were obtained at 293 K from a 300 nl drop formed from 150 nl Smo-DHDPR solution (14.5 mg ml−1 in 20 mM Tris–HCl, pH 8.0) and 150 nl reservoir solution [PACT condition D7; 20% (w/v) PEG 6000, 200 mM sodium chloride, 100 mM Tris–HCl, pH 8.0, including 0.02% (w/v) sodium azide]. Nme-DHDPR was crystallised by the sitting-drop vapour-diffusion method at 293 K by mixing 150 nl of protein (10 mg ml−1) with 150 nl of reservoir solution [0.03 M CaCl2, 0.03 M MgCl2, 20% (v/v) PEG 500 MME, 10% (w/v) PEG 20 000, and 0.1 M Tris (base)/BICINE (pH 8.5)]. Ath-DHDPR and SeMet-Smo-DHDPR were crystallised by the vapour-diffusion method at 291 K at a protein concentration of ∼20 mg ml−1 using the same drop size and ratio, as previously described. NADH (1 mM) and 2,6-pyridinedicarboxylic acid (PDC, 1 mM) were added to each protein prior to setting the drops. SeMet-Smo-DHDPR and Ath-DHDPR were crystallised in 2 days following mixing with reservoir solutions containing 0.2 M CaCl2 and 20% (w/v) PEG 3350, and 1.6 M NH4SO4 (pH 5). Cryoprotection was achieved by soaking the crystal in a solution prepared by the addition of 20% glycerol to the reservoir solution prior to freezing in liquid nitrogen.

Data collection and processing

All data were collected on the MX1 and MX2 beamlines at the Australian synchrotron. Data were processed using XDS [29] and AIMLESS [30]. Data collection and refinement statistics are presented in Supplementary Table S1.

Structure determination and refinement

Nme-DHDPR was solved by molecular replacement with PHASER MR, using 1ARZ as a search model. As molecular replacement attempts for Smo-DHDPR and Ath-DHDPR were unsuccessful, selenium-derivatised protein crystals were prepared for Smo-DHDPR. Subsequently, SAD methods, implemented in the SHELX C/D/E suite [31] using the anomalous scattering of the selenium-derivatised methionines (18 selenium sites), were used to determine the structure and an initial model was autobuilt using ARP/wARP [32]. This model was then used for molecular replacement of both the native Smo-DHDPR and Ath-DHDPR. All models were iteratively refined and built in REFMAC5 [33] and COOT [34], respectively.

Activity assays

DHDPR enzyme activity was measured in 100 mM HEPES, pH 8.0, using a coupled assay as previously described [35]. Stock solutions of (S)-ASA, pyruvate, (S)-lysine, NADPH, and NADH were prepared fresh for each experiment. Assay temperature was regulated by the use of a circulating water bath, and assays were performed at 25°C. Initial rate data were typically reproducible within 10% and were analysed using the non-linear regression software (GraphPad Prism). Cuvettes were pre-incubated with an excess of DHDPS (20–100 µg ml−1) for 60 s to generate the HTPA substrate, before assays were initiated by the addition of DHDPR (typically 0.6 µg ml−1).

Small-angle X-ray scattering

SAXS data were collected at the Australian Synchrotron SAXS/WAXS beamline [36] using a camera length of 1576 mm, as described previously [23]. Protein samples (50 μl, 4–10 mg ml−1) were loaded onto a Superdex 200 Increase 5/150 SEC column (GE Healthcare) and eluted with buffer containing (20 mM Tris–HCl, pH 8.0, and 150 mM NaCl). Scattering data were background-corrected, averaged, and scaled using the ScatterBrain software (written and provided by the Australian Synchrotron; available at http://www.synchrotron.org.au/). The ATSAS software package was used for all analyses [37]. Comparison of scattering data with crystallographic models was performed using CRYSOL [38].

Results and discussion

Crystal structures show that plant and bacterial DHDPR share a largely conserved tertiary structure and reveal molecular basis for divergent quaternary structure

Ath-DHDPR has previously been observed to form a dimer in solution; however, the arrangement of the two subunits could not be elucidated without a high-resolution model and it was unclear whether the dimer was formed by two adjacent monomers as part of the intra-sheet or inter-sheet interface observed in the tetrameric bacterial DHDPR structures [23]. Novel DHDPR from plants S. moellendorffii (Smo-DHDPR) and V. vinifera (Vvi-DHDPR), and from the pathogenic bacteria, N. meningitidis (Nme-DHDPR), were successfully overexpressed in E. coli and purified using nickel affinity chromatography to a high level of homogeneity. Analytical ultracentrifugation (Figure 1) showed that Smo-DHDPR and Vvi-DHDPR exist as homodimers in solution, with calculated masses of 64.2 and 62.9 kDa, respectively, while Nme-DHDPR exists as a homotetramer in solution, with a calculated mass of 113 kDa. Given that the dimeric arrangement of DHDPR previously observed for Arabidopsis is also found in grape and Selaginella (an ancient lineage of ancestral plants) DHDPR, this suggests that a dimeric structure is conserved among plant species.

Analytical ultracentrifugation of DHDPR shows dimeric or tetrameric species.

Figure 1.
Analytical ultracentrifugation of DHDPR shows dimeric or tetrameric species.

Sedimentation velocity experiments were carried out for Nme-DHDPR, Vvi-DHDPR, and Smo-DHDPR (0.5 mg ml−1) and data were fitted to a continuous mass coefficient distribution c(m) model.

Figure 1.
Analytical ultracentrifugation of DHDPR shows dimeric or tetrameric species.

Sedimentation velocity experiments were carried out for Nme-DHDPR, Vvi-DHDPR, and Smo-DHDPR (0.5 mg ml−1) and data were fitted to a continuous mass coefficient distribution c(m) model.

To determine the arrangement of the dimeric DHDPR enzyme, the structures of Nme-DHDPR, Ath-DHDPR, and Smo-DHDPR were determined by X-ray crystallography. Consistent with the oligomeric states measured by analytical ultracentrifugation, Nme-DHDPR formed a homotetramer in the asymmetric unit, while Smo-DHDPR and Ath-DHDPR assembled as dimers in their respective crystal lattices (Figure 2, top row). Comparison of the Nme-DHDPR structure to the representative structure of the E. coli DHDPR (PDBID: 1ARZ) gives an RMSD of 1.15 over 263 residues (PDBeFOLD, [39]), demonstrating the highly conserved nature of the bacterial DHDPR structure. The N-terminal domain is a three-layer (α–β–α) sandwich forming a Rossmann-like fold that contains the nucleotide-binding site (Figure 2). The C-terminal domain is a mixed α + β fold with a central four-stranded β-sheet that facilitates oligomerisation and HTPA binding. The pairing of β11 from two adjacent monomers creates a continuous, curved, eight-stranded mixed parallel/antiparallel β-sheet (‘intra-sheet’ interface), which assembles in a sheet-to-sheet manner (‘inter-sheet’ interface), forming a 16-stranded flattened β-barrel in the tetramer (Figure 2). The length and orientation of the long C-terminal loop that wraps around the β-barrel is similar to that of E. coli DHDPR.

X-ray crystallographic structures determined of Nme-, Smo-, and Ath-DHDPR enzymes.

Figure 2.
X-ray crystallographic structures determined of Nme-, Smo-, and Ath-DHDPR enzymes.

All structures comprise an N-terminal domain containing a Rossmann-fold and a C-terminal domain that facilitates oligomerisation. Nme-DHDPR is shown in a tetrameric assembly, while Smo- and Ath-DHDPR are dimers. The structures are largely conserved; however, Smo-DHDPR includes an insertion of two short α-helices (α-A and α-B, purple), and Ath-contains a two-stranded antiparallel β-sheet insertion (β-A and β-B, yellow).

Figure 2.
X-ray crystallographic structures determined of Nme-, Smo-, and Ath-DHDPR enzymes.

All structures comprise an N-terminal domain containing a Rossmann-fold and a C-terminal domain that facilitates oligomerisation. Nme-DHDPR is shown in a tetrameric assembly, while Smo- and Ath-DHDPR are dimers. The structures are largely conserved; however, Smo-DHDPR includes an insertion of two short α-helices (α-A and α-B, purple), and Ath-contains a two-stranded antiparallel β-sheet insertion (β-A and β-B, yellow).

During the evolution of quaternary structures, the more ancient interfaces tend to be larger than the more recently formed interfaces [40], and nearly all bacterial DHDPR enzymes have a larger inter-strand interface (Table 1); thus, it was expected that this would be the ancestral interface found in the plant enzymes. However, crystal structures of both Ath-DHDPR and Smo-DHDPR determined here allow the quaternary structure of dimeric DHDPR proteins to be observed for the first time and reveal that the plant enzymes form a dimer using the intra-strand interface (Figure 2). This dimerisation interface has had extensive modifications in plant DHDPR, increasing the size of the interface from 1100 to 1400 Å in the tetrameric DHDPR to 1600–1900 Å in plant DHDPR (Table 1).

Table 1
Relative sizes of the intra- and inter-strand interfaces of different DHDPR structures, as calculated by PISA
 PDB ID Intra-sheet interface (Å2Inter-sheet interface (Å2
N. meningitidis 5UGJ1 1224 1532 
A. thaliana 5UA01 1890 NA 
S. moellendorffii 5U5N1 1582 NA 
A. variabilis 5KT0 1112 1683 
Corynebacterium glutamicum 5EER 1318 1484 
Pseudomonas aeruginosa 4YWJ 1369 1523 
Burkholderia thailandensis 4F3Y 1307 1521 
S. aureus 3QY9 1170 1339 
Bartonella henselae 3IJP 1503 1325 
Thermotoga maritima 1VM6 1059 778 
M. tuberculosis 1C3V 1127 1506 
E. coli 1ARZ 1354 1563 
 PDB ID Intra-sheet interface (Å2Inter-sheet interface (Å2
N. meningitidis 5UGJ1 1224 1532 
A. thaliana 5UA01 1890 NA 
S. moellendorffii 5U5N1 1582 NA 
A. variabilis 5KT0 1112 1683 
Corynebacterium glutamicum 5EER 1318 1484 
Pseudomonas aeruginosa 4YWJ 1369 1523 
Burkholderia thailandensis 4F3Y 1307 1521 
S. aureus 3QY9 1170 1339 
Bartonella henselae 3IJP 1503 1325 
Thermotoga maritima 1VM6 1059 778 
M. tuberculosis 1C3V 1127 1506 
E. coli 1ARZ 1354 1563 
*

Structures determined as part of this work.

The Ath- and Smo-DHDPR models share a fundamentally conserved tertiary fold with Nme- and previously solved bacterial DHDPR structures, with RMSD values against E. coli DHDPR of 2.38 over 204 residues and 2.15 over 217 residues, respectively [39]. However, a key difference between plant and bacterial DHDPR structures is the structural arrangement of residues in the region between α6 and β10 (Figure 2 and Supplementary Material S1). The inter-sheet interface, which is formed by β9 and the C-terminal loops in Nme-DHDPR, is rearranged in the dimeric plant structures, resulting in a structural occlusion that prevents two dimers coming together to form a tetramer in the same manner as bacterial DHDPR. The structural occlusion to tetramerisation takes two different forms — a two-stranded (β-A and β-B) antiparallel β-sheet in the Ath-DHDPR model and two short α-helices (α-A and α-B) in Smo-DHDPR (Figure 2, bottom row) — despite 90% sequence similarity in this region (compared with 72% overall). This region includes highly conserved residues, some of which are found in vastly different three-dimensional positions in the two different crystallographic conformations. For example, Glu226, which is located on strand β-B in Ath-DHDPR and participates in the dimerisation interface, is situated more than 20 Å away on helix α-B in the structure of Smo-DHDPR (Glu218), where the residue is not involved in any protein interactions. In Ath-DHDPR, the β-sheet insertion is involved in intra- and inter-subunit interactions, thus increasing the dimer interface area, whereas the α-helical structure in Smo-DHDPR does not contribute to inter-subunit interactions. The intra-sheet interface in these dimeric structures is slightly larger than the equivalent interface in the bacterial tetramer, comprising 1889 Å2 (15% of total area) for Ath-DHDPR and 1582 Å2 (12%) for Smo-DHDPR, compared with 1224 Å2 (9%) for Nme-DHDPR (Table 1).

Another variation in the plant DHDPR structures is that the β-sheet formed by dimerisation consists of only six complete strands, with β9 from Nme-DHDPR being replaced with the single turn of α7 and a truncated β9. This β-sheet is less curved compared with the equivalent eight-stranded β-sheet in Nme-DHDPR, and the reduced curvature pushes the flanking α-helices closer together, decreasing the distance between adjacent α5 helices from ∼7.5 Å in Nme-DHDPR to ∼5.5 Å in the plant DHDPR structures. There is also a notable difference in orientation of helices α5 and α6 in Ath- and Smo-DHDPR compared with Nme-DHDPR, most markedly for helix α5, which is splayed outward in the plant structures, by virtue of a longer tethering loop between α5 and β8, rather than following the curve of the β-sheet in front, as in the bacterial structures.

Smo-DHDPR showed density for a molecule of NADH bound in a similar conformation to that previously observed in Tma-DHDPR, Eco-DHDPR, and Mtb-DHDPR structures (Figure 3) [1518], except that the amide is twisted out of coplanarity with the pyridine ring. Interestingly, in addition to the many conserved interactions between Smo-DHDPR and other DHDPR with NADH, a unique interaction between the adenosyl ribose of NADH and residue of Asp20 is seen. In contrast with commonly observed bifurcated hydrogen-bonding contacts between the adenosyl ribose hydroxyls of NADH and the carboxylate side chain of a characteristic acidic residue in NADH-binding Rossmann-fold oxidoreductases, this interaction in Smo-DHDPR involves only a single carboxylate oxygen of the Asp20 side chain and the ribose 3′ hydroxyl, and the carboxylate approaches the 3′-hydroxyl from behind.

Comparison of chains from each structure and details of NADH binding in Smo-DHDPR.

Figure 3.
Comparison of chains from each structure and details of NADH binding in Smo-DHDPR.

(A) Alignment of the dimerisation domain for each chain of each enzyme highlights differences in the orientation of the N-terminal domain for Nme- and Smo-DHDPR. Ath-DHDPR show little difference between chains. Colouring of the chains is as follows: Nme-DHDPR, A (green), B (grey), C (orange), and D (yellow); Smo-DHDPR, A (blue) and B (grey); Ath-DHDPR, A (cyan), B (red), and C (grey). Residues forming the latch and catch have been shown as sticks. (B) The hydrogen bonds formed between residues from Smo-DHDPR and NADH are shown as dashed lines. Grey mesh is 2FoFc electron density contoured at 1.5σ.

Figure 3.
Comparison of chains from each structure and details of NADH binding in Smo-DHDPR.

(A) Alignment of the dimerisation domain for each chain of each enzyme highlights differences in the orientation of the N-terminal domain for Nme- and Smo-DHDPR. Ath-DHDPR show little difference between chains. Colouring of the chains is as follows: Nme-DHDPR, A (green), B (grey), C (orange), and D (yellow); Smo-DHDPR, A (blue) and B (grey); Ath-DHDPR, A (cyan), B (red), and C (grey). Residues forming the latch and catch have been shown as sticks. (B) The hydrogen bonds formed between residues from Smo-DHDPR and NADH are shown as dashed lines. Grey mesh is 2FoFc electron density contoured at 1.5σ.

Differing conformations of subunits within the crystals underscore domain movement and provide insights into hinge mechanism

DHDPR is understood to undergo a conformational change from an open to a closed conformation, involving rotation of the N-terminal domain relative to the C-terminal domain [15,16]. This movement is necessary to bring the nucleotide and substrate into close proximity for hydride transfer to take place. Crystal structures have captured various conformations between the open and closed state, providing snapshots of this domain motion [1517,20]. Since a closed conformation has been observed in the absence of bound ligand [20], it seems that neither nucleotide nor substrate binding is required to induce this conformational change as originally thought [15,16]; rather, the movement may be a low-frequency normal mode of DHDPR [17]. Consistent with this, an alignment of the C-terminal domain from each subunit of our apo Nme-DHDPR structure shows that the N-terminal domain of chain A is in a partially closed conformation, while chains B, C, and D are in similar open conformations (Figure 3A), indicating that there is some degree of flexibility between the two domains, independent of cofactor or substrate binding. The greatest difference in inter-domain angle is between that of chains A and B, and represents a rotation angle of 11.2° (Supplementary Table S2). A similar rotation angle of 11.5° is calculated for the domain movement observed in the NADH-bound Smo-DHDPR structure, in which chain B exists in a more closed conformation relative to chain A (Figure 3A,B). In contrast, the three chains present in the Ath-DHDPR asymmetric unit all have the same domain orientations. If the N-terminal domain of Ath-DHDPR adopted the same conformation as that of the Nme-DHDPR structure, the packing and orientation of helix α5 in the plant DHDPR structures would result in a severe clash with helix α9, and this may influence the magnitude or trajectory of the domain movement.

Inspection of the ‘hinge’ regions (see Supplementary Figure S1) reveals that domain closure results in the insertion of a conserved glutamine (plants) or asparagine (bacteria) between the top of C-terminal domain helices α5 and α6 (Figure 3). This movement brings the amide side chain (the ‘latch’) in H-bonding proximity to a conserved threonine (the ‘catch’) at the top of α6. The C-terminally adjacent hydrophobic residue (Met or Phe) in the hinge loop could be envisaged as a molecular tow bar, which pulls the N-terminal domain with the latch as it moves towards the catch. In many bacterial DHDPR, a serine residue (Ser127 in Nme-DHDPR) at the top of helix α5 has the potential to function as an additional ‘safety catch’. In the open conformation, as in all chains in the Nme-DHDPR structure solved here, the side chain of this serine is oriented away from the latching Asn residue, whereas in the closed conformations observed in E. coli DHDPR (PBDID: 1ARZ; chains B, C, and D) and T. maritima DHDPR (PBDID: 1VM6; chains A, B, and D), it can be seen that the serine side chain is rotated towards the latch.

Discrepancies between crystal structures and solution scattering are indicative of domain movements

SAXS was utilised to assess the arrangement of Ath-DHDPR, Smo-DHDPR, and Nme-DHDPR in solution. The envelope volume and radius of gyration (Rg) determined experimentally agreed well with those calculated from the crystal structures (Table 2). Theoretical scattering curves generated from the crystal structures were fitted to experimental scattering data using CRYSOL [38], and showed reasonable agreement, but with moderate deviation around the peaks and troughs, which are less pronounced in the experimental SAXS profile than what the crystal structures predict (Figure 4). For Nme-DHDPR, a better fit was achieved with a structure in which two subunits were manually modelled in a closed conformation (Supplementary Figure S2). The SREFLEX program, which utilises normal mode analysis to account for flexibility in the structure [41], also produced a model with altered domain conformations to improve the agreement with the experimental scatter for Nme-DHDPR (Supplementary Figure S2). Crystals were unable to be obtained for Vvi-DHDPR; however, the experimental SAXS profiles for the plant DHDPR enzymes all have a similar shape, indicating that Vvi-DHDPR is likely to have a similar arrangement to Ath-DHDPR (Figure 4).

SAXS of DHDPR.

Figure 4.
SAXS of DHDPR.

Data were collected for different DHDPR enzymes as they eluted from a size-exclusion column. Experimental scattering was overlaid with the scattering profiles calculated for Smo-, Ath-, and Nme-DHDPR structures using CRYSOL. Curves have been arbitrarily displaced along the y-axis for clarity.

Figure 4.
SAXS of DHDPR.

Data were collected for different DHDPR enzymes as they eluted from a size-exclusion column. Experimental scattering was overlaid with the scattering profiles calculated for Smo-, Ath-, and Nme-DHDPR structures using CRYSOL. Curves have been arbitrarily displaced along the y-axis for clarity.

Table 2
SAXS parameters
  Envelope volume (Å3Rg (Å) 
N. meningitidis Experimental 179 479 33.1 
Theoretical (5UGJ) 178 700 32.0 
S. moellendorffii Experimental 98 511 30.4 
Theoretical (5U5N) 92 230 28.8 
A. thaliana Experimental 102 141 31.4 
Theoretical (5U5I) 86 880 28.7 
V. vinifera Experimental 110 701 37.4 
  Envelope volume (Å3Rg (Å) 
N. meningitidis Experimental 179 479 33.1 
Theoretical (5UGJ) 178 700 32.0 
S. moellendorffii Experimental 98 511 30.4 
Theoretical (5U5N) 92 230 28.8 
A. thaliana Experimental 102 141 31.4 
Theoretical (5U5I) 86 880 28.7 
V. vinifera Experimental 110 701 37.4 

Plant DHDPR enzymes are prone to substrate inhibition

As lysine is a nutritionally limiting amino acid in many crops, much research is currently aimed at increasing the lysine concentration in plants [16]. Typically, the native plant DHDPS enzyme, which catalyses the first step of the DAP pathway, is feedback-inhibited by lysine [23]. Expression of a feedback-insensitive DHDPS enzyme (often from bacteria) increases flux through the pathway, resulting in enhanced levels of free lysine [5,6]. However, DHDPR, which catalyses the second step, has been shown to be inhibited by high concentrations of its substrate [11,18]; thus, increasing the activity of DHDPS may result in the reduced activity of DHDPR, causing a metabolic bottleneck in the pathway. Previous work had shown that Ath-DHDPR is moderately inhibited by HTPA when using NADH as a cofactor, but is only slightly inhibited by HTPA when using NADPH as a cofactor [23]. Given that the tendency of bacterial DHDPR enzymes towards substrate inhibition differs greatly between species, it was unclear whether the results obtained for Ath-DHDPR were representative of all plant DHDPR enzymes.

Nme-, Smo-, and Vvi-DHDPR were kinetically characterised in the present study and readily catalysed the reduction in DHDP using either NADH or NADPH as a cofactor, with kinetic parameters comparable with those of previously studied DHDPR enzymes [13,15,18,20] (Table 3 and Figure 5). Nme-DHDPR follows a similar trend to E. coli and M. tuberculosis DHDPR enzymes, with a lower Km for NADH and higher Vmax with this nucleotide [13,15]. Akin to the previously characterised Ath-DHDPR [23], Vvi-DHDPR and Smo-DHDPR were found to have a lower Km(HTPA) when utilising NADH, but a higher Vmax when NADPH was used. All three enzymes displayed some level of substrate inhibition when utilising NADH (Figure 5), but the degree of inhibition varied between orthologues; for Nme-DHDPR, it is only minor (Ki = 5.5 ± 1.5 mM), whereas for Smo-DHDPR it is particularly severe (Ki = 32 ± 9 μM). Vvi- and Smo-DHDPR additionally showed some substrate inhibition with NADPH, and inhibition constants were lower than those determined for Ath-DHDPR.

Kinetic activity of Nme-DHDPR, Vvi-DHDPR, and Smo-DHDPR, utilising NADH or NADPH as a cofactor.

Figure 5.
Kinetic activity of Nme-DHDPR, Vvi-DHDPR, and Smo-DHDPR, utilising NADH or NADPH as a cofactor.

Data are fitted to either a Michaelis–Menten or a substrate inhibition model (GraphPad Prism) and the kinetic parameters determined are presented in Table 3. Error bars represent the standard deviation of ≥2 measurements.

Figure 5.
Kinetic activity of Nme-DHDPR, Vvi-DHDPR, and Smo-DHDPR, utilising NADH or NADPH as a cofactor.

Data are fitted to either a Michaelis–Menten or a substrate inhibition model (GraphPad Prism) and the kinetic parameters determined are presented in Table 3. Error bars represent the standard deviation of ≥2 measurements.

Table 3
Kinetic parameters (±SE) determined for Nme-DHDPR, Vvi-DHDPR, and Smo-DHDPR, from data fitted to either a Michaelis–Menten or a substrate inhibition model (GraphPad Prism)
 Cofactor Km (HTPA) (µM) Ki (HTPA) (mM) kcat (s−1Km (NAD(P)H) (µM) 
N. meningitidis NADH 13 ± 2 5.5 ± 1.5 77 ± 4 0.5 ± 0.2 
NADPH 85 ± 6 — 43 ± 1 12 ± 0.1 
V. vinifera NADH 3.1 ± 0.4 0.76 ± 0.11 14 ± 1 4.7 ± 0.8 
NADPH 9.2 ± 0.9 1.8 ± 0.2 21 ± 1 16 ± 2 
S. moellendorffii NADH 12 ± 3 0.032 ± 0.009 49 ± 8 1.4 ± 0.1 
NADPH 58 ± 11 1.3 ± 0.2 63 ± 5 6.8 ± 0.5 
 Cofactor Km (HTPA) (µM) Ki (HTPA) (mM) kcat (s−1Km (NAD(P)H) (µM) 
N. meningitidis NADH 13 ± 2 5.5 ± 1.5 77 ± 4 0.5 ± 0.2 
NADPH 85 ± 6 — 43 ± 1 12 ± 0.1 
V. vinifera NADH 3.1 ± 0.4 0.76 ± 0.11 14 ± 1 4.7 ± 0.8 
NADPH 9.2 ± 0.9 1.8 ± 0.2 21 ± 1 16 ± 2 
S. moellendorffii NADH 12 ± 3 0.032 ± 0.009 49 ± 8 1.4 ± 0.1 
NADPH 58 ± 11 1.3 ± 0.2 63 ± 5 6.8 ± 0.5 

Our results show that both Smo-DHDPR and Vvi-DHDPR demonstrate substrate inhibition, indicating that this may be a general characteristic of plant DHDPR enzymes. Therefore, in order to maximise lysine levels in plants, a non-substrate-inhibited DHDPR enzyme may also be necessary in addition to expressing a feedback-insensitive DHDPS enzyme.

Substrate inhibition may be connected to flexibility in the active site

Substrate inhibition in DHDPR is proposed to be the result of substrate binding to an enzyme that still has oxidised nucleotide bound in the active site, following the release of THDP [11]. This mechanism suggests that THDP is more readily released than oxidised nucleotide. With relatively few DHDPR structures solved in complex with substrate analogues, it is difficult to analyse and rationalise substrate inhibition at the molecular level. However, the flexibility seen in the active sites of the plant DHDPR structures determined here might be connected to this phenomenon. While the substrate-binding loop has previously been found to be flexible in the bacterial DHDPR structures, this loop appears to be substantially more disordered in the plant DHDPR structures presented here, to the extent where it is unable to be modelled in some chains. In the substrate-binding site, the density for Lys175 was poorly defined in the Smo-DHDPR structure and many of the substrate-binding loop residues are highly flexible or unable to be modelled due to the poorly defined density in Ath-DHDPR. In chains A and C of the Ath-DHDPR structure, this loop is modelled as a helix which obstructs the substrate-binding pocket and Lys183 of chain A is oriented away from the active site, forming a sulphate-mediated crystal contact with the N-terminal domain of chain B.

Based on the proposed mechanism of substrate inhibition, a highly flexible substrate-binding loop may provide a ‘back door’, which allows the THDP product to exit and a new HTPA substrate to enter the active site before the oxidised cofactor has been released. In Nme-DHDPR, which displays little inhibition, the C-terminal long loop from the opposite subunit in the tetramer forms interactions with the substrate-binding loop; thus, we postulate that the C-terminal loop in bacterial DHDPR may serve not only for stabilisation of the tetramer, but also for structural reinforcement of the substrate-binding loop in order to combat substrate inhibition. Consistent with this idea, the tetrameric T. maritima DHDPR, which lacks a C-terminal loop, also exhibits substrate inhibition [18].

The variation in the extent of substrate inhibition observed with phosphorylated versus unphosphorylated nucleotide, and between different DHDPR orthologues, is not explained by tetramerisation or the presence of a C-terminal loop. It is likely that each DHDPR orthologue is tuned slightly differently with regard to substrate, cofactor, and product affinity, as a result of unique features of the binding sites. In general, Km values for the nucleotide, with which substrate inhibition is most severe, are lower than those for the alternative nucleotide, which is consistent with tighter binding of that nucleotide.

Nucleotide binding and specificity of plant and bacterial DHDPR

The nucleotide specificity of DHDPR is remarkably promiscuous compared with many other oxidoreductase enzymes. Many DHDPR sequences contain an acidic residue positioned to interact with the adenosyl ribose hydroxyls of NADH, akin to other NADH-specific enzymes [14]. Some DHDPR sequences additionally contain an adjacent basic residue which is primed to interact with the phosphorylated nucleotide. In E. coli DHDPR, these two residues (Glu38 and Arg39) were shown to be in different conformations in the NADH- versus the NADPH-bound form [14], providing a convincing explanation for the dual specificity of this enzyme, but this reasoning cannot be extended to other orthologues that do not possess equivalent residues. A sequence alignment shows that the residues in the position equivalent to Glu38 and Arg39 in E. coli DHDPR are Glu36 and His37 in Nme-DHDPR (Supplementary Figure S1), and although the structure of Nme-DHDPR was solved in the apo form here, superposition of an NADH-bound E. coli DHDPR structure reveals that these residues are situated in an equivalent three-dimensional position, poised to interact with the adenosyl ribose. Nme-DHDPR was also found to readily utilise either nucleotide cofactor and bind them with similar affinities to E. coli DHDPR, while displaying a modest preference for NADH. In Ath-DHDPR, which crystallised in the absence of nucleotide, Glu58 is suitably placed to interact with the ribose hydroxyls of NADH, but since NADPH is the biologically relevant cofactor in plants, given that DHDPR is located in the chloroplasts [42], the functional significance of this potential interaction with NADH is unclear. An equivalent basic residue to potentially interact with the 2′ phosphate of NADPH could not be identified in the vicinity of the nucleotide-binding site in the Ath-DHDPR structure, although previous kinetic characterisation found NADPH to be the preferred cofactor for this enzyme [23].

In Smo-DHDPR, an acidic residue, situated further upstream in the 1D sequence (Asp20), appears to be in adequate proximity to form a hydrogen bond with the adenosyl ribose of the NADH with which it crystallised (Figure 3B). This particular aspartic acid residue is part of a highly unusual X-nonPro cis-peptide bond with Asn19. In the rare occasions when non-proline cis-peptides are observed, they are often found close to active sites, as is the case here, and are deemed to be important for function [43], and therefore, expected to be well conserved. However, Asp20 in Smo-DHDPR replaces the first glycine in the characteristic glycine-rich nucleotide-binding motif, which in DHDPR follows the pattern GXXGXXG. Again, given that NADPH is considered to be the biologically relevant cofactor for plant DHDPR enzymes, the biological significance of this cis-peptide and associated nucleotide interaction is not known. Unlike the typical bifurcated hydrogen-bonding interaction between an acidic residue and the ribose of NADH seen in other Rossmann-fold oxidoreductases, this backward approach of the Asp20 side chain involving only the 3′ hydroxyl is unlikely to interfere with a 2′ phosphate substituent, and so, it is possible that the interaction also occurs when NADPH is bound. The results of the kinetic analysis showed that Smo-DHDPR has a higher Km value for NADPH and is less severely substrate-inhibited when utilising this nucleotide, indicating that it is not bound as tightly.

Conclusions

The three novel crystal structures described here have contributed further evidence that bacterial DHDPR are tetrameric, whereas the plant orthologues are dimeric. While it had been expected that the dimeric form of plant DHDPR would involve the inter-sheet interface seen in bacterial DHDPR enzymes, based on the tendency of this interface to be the most extensive in the tetrameric structures, the crystal structures presented here reveal that the plant enzymes form a dimer using the intra-strand interface. The crystal structures show the molecular basis for this difference, but only the evolutionary rationale underlying the divergence in oligomeric state can be surmised. Increased oligomeric assembly has been suggested to have several different advantages, including the potential for allostery or cooperativity, increased stability due to a smaller surface-to-volume ratio, reduced sensitivity of catalytic sites to protein motion in higher-order structures, and increased encounter rate of the protein with a substrate [44,45]. It has been suggested that ligand-binding cooperativity operates in E. coli DHDPR [46], implying a functional interaction between the subunits. Cooperativity would certainly offer a compelling reason for oligomerisation, but this requires further investigation. It is difficult to determine whether the last common form of DHDPR was the prokaryotic tetrameric form, with subsequent reduction in oligomeric structure to form dimers in eukaryotes, or if the ancestral enzyme was dimeric, with the subsequent increase in oligomeric structure in prokaryotes.

Substrate inhibition of plant DHDPR may hinder efforts to increase the lysine production by the increased flux through the dap pathway. This inhibition appears to be a consequence of high affinity for the nucleotide substrate, while the reduced flexibility of the substrate-binding loop can alleviate this inhibition. The C-terminal long loop makes a substantial contribution to the inter-strand interface in Nme-DHDPR and other previously solved DHDPR structures. This loop is highly variable across bacterial DHDPR orthologues, differing in length, composition, and orientation, which is consistent with it independently evolving in each orthologue. We postulate here that this loop functions to reduce substrate-binding loop flexibility, and as such, may have convergently evolved to counteract substrate inhibition, with varying degrees of effectiveness.

Database (PDB) depositions

SeMet-Smo-DHDPR, 5U5I; Smo-DHDPR, 5U5N; Ath-DHDPR, 5UA0; Nme-DHDPR, 5UGJ.

Abbreviations

     
  • DAP

    diaminopimelate

  •  
  • DHDP

    dihydrodipicolinate

  •  
  • DHDPR

    Dihydrodipicolinate reductase

  •  
  • HTPA

    (4S)-4-hydroxy-2,3,4,5-tetrahydro-(2S)-dipicolinic acid

  •  
  • Nme-DHDPR

    DHDPR from Neisseria meningitidis

  •  
  • PEG 500 MME

    poly(ethylene glycol) methyl ether 500

  •  
  • RMSD

    root mean square deviation

  •  
  • SAD

    single-wavelength anomalous diffraction

  •  
  • SAXS

    small-angle X-ray scattering

  •  
  • SeMet

    selenomethionine

  •  
  • Smo-DHDPR

    DHDPR from S. moellendorffii

  •  
  • THDP

    2,3,4,5-tetrahydrodipicolinate

  •  
  • Vvi-DHDPR

    DHDPR from V. vinifera

Author Contribution

S.R.A.D. cloned the gene for Nme-DHDPR. S.A.J.W., J.R.K., E.R., S.R.A.D., and F.G.P. purified protein. S.A.J.W., J.R.K., E.R., and F.G.P. produced protein crystals. J.R.K. and D.C.G. determined the structures. S.A.J.W., S.R.A.D., and F.G.P. carried out SAXS, analytical ultracentrifugation, and kinetic assays. S.A.J.W., J.R.K., and F.G.P. wrote the manuscript, with contribution from all authors.

Funding

Funding for travel to the Australian Synchrotron was provided by the NZ Synchrotron Group. DG was supported by a Rutherford Discovery Fellowship awarded by the NZ government and administered by the RSNZ.

Acknowledgments

Parts of this research were undertaken at the MX and SAXS beamline of the Australian Synchrotron, Victoria, Australia. We thank Janet Newman at the CSIRO Collaborative Crystallisation Centre (www.csiro.au/C3), Melbourne, Australia for the enthusiastic help with crystallography.

Competing Interests

The Authors declare that there are no competing interests associated with the manuscript.

References

References
1
Zhu
,
X.
and
Galili
,
G.
(
2003
)
Increased lysine synthesis coupled with a knockout of its catabolism synergistically boosts lysine content and also transregulates the metabolism of other amino acids in Arabidopsis seeds
.
Plant Cell
15
,
845
853
2
Zhu
,
X.
and
Galili
,
G.
(
2004
)
Lysine metabolism is concurrently regulated by synthesis and catabolism in both reproductive and vegetative tissues
.
Plant Physiol.
135
,
129
136
3
Falco
,
S.C.
,
Guida
,
T.
,
Locke
,
M.
,
Mauvais
,
J.
,
Sanders
,
C.
,
Ward
,
R.T.
et al. 
(
1995
)
Transgenic canola and soybean seeds with increased lysine
.
Biotechnology
13
,
577
582
PMID:
[PubMed]
4
Frizzi
,
A.
,
Huang
,
S.
,
Gilbertson
,
L.A.
,
Armstrong
,
T.A.
,
Luethy
,
M.H.
and
Malvar
,
T.M.
(
2008
)
Modifying lysine biosynthesis and catabolism in corn with a single bifunctional expression/silencing transgene cassette
.
Plant Biotechnol. J.
6
,
13
21
5
Lee
,
S.I.
,
Kim
,
H.U.
,
Lee
,
Y.-H.
,
Suh
,
S.-C.
,
Lim
,
Y.P.
,
Lee
,
H.-Y.
et al. 
(
2001
)
Constitutive and seed-specific expression of a maize lysine-feedback-insensitive dihydrodipicolinate synthase gene leads to increased free lysine levels in rice seeds
.
Mol. Breed.
8
,
75
84
6
Long
,
X.
,
Liu
,
Q.
,
Chan
,
M.
,
Wang
,
Q.
and
Sun
,
S.S.M.
(
2013
)
Metabolic engineering and profiling of rice with increased lysine
.
Plant Biotechnol. J.
11
,
490
501
7
Cox
,
R.J.
(
1996
)
The DAP pathway to lysine as a target for antimicrobial agents
.
Nat. Prod. Rep.
13
,
29
43
8
Cox
,
R.J.
,
Sutherland
,
A.
and
Vederas
,
J.C.
(
2000
)
Bacterial diaminopimelate metabolism as a target for antibiotic design
.
Bioorg. Med. Chem.
8
,
843
871
9
Mitsakos
,
V.
,
Dobson
,
R.C.J.
,
Pearce
,
F.G.
,
Devenish
,
S.R.
,
Evans
,
G.L.
,
Burgess
,
B.R.
et al. 
(
2008
)
Inhibiting dihydrodipicolinate synthase across species: towards specificity for pathogens?
Bioorg. Med. Chem. Lett.
18
,
842
844
10
Devenish
,
S.R.A.
,
Blunt
,
J.W.
and
Gerrard
,
J.A.
(
2010
)
NMR studies uncover alternate substrates for dihydrodipicolinate synthase and suggest that dihydrodipicolinate reductase is also a dehydratase
.
J. Med. Chem.
53
,
4808
4812
11
Dommaraju
,
S.R.
,
Dogovski
,
C.
,
Czabotar
,
P.E.
,
Hor
,
L.
,
Smith
,
B.J.
and
Perugini
,
M.A.
(
2011
)
Catalytic mechanism and cofactor preference of dihydrodipicolinate reductase from methicillin-resistant Staphylococcus aureus
.
Arch. Biochem. Biophys.
512
,
167
174
12
Dogovski
,
C.
,
Dommaraju
,
S.R.
,
Small
,
L.C.
and
Perugini
,
M.A.
(
2012
)
Comparative structure and function analyses of native and his-tagged forms of dihydrodipicolinate reductase from methicillin-resistant Staphylococcus aureus
.
Protein Expr. Purif.
85
,
66
76
13
Reddy
,
S.G.
,
Sacchettini
,
J.C.
and
Blanchard
,
J.S.
(
1995
)
Expression, purification, and characterization of Escherichia coli dihydrodipicolinate reductase
.
Biochemistry
34
,
3492
3501
14
Reddy
,
S.G.
,
Scapin
,
G.
and
Blanchard
,
J.S.
(
1996
)
Interaction of pyridine nucleotide substrates with Escherichia coli dihydrodipicolinate reductase: thermodynamic and structural analysis of binary complexes
.
Biochemistry
35
,
13294
13302
15
Cirilli
,
M.
,
Zheng
,
R.
,
Scapin
,
G.
and
Blanchard
,
J.S.
(
2003
)
The three-dimensional structures of the Mycobacterium tuberculosis dihydrodipicolinate reductase−NADH−2,6-PDC and −NADPH−2,6-PDC complexes. Structural and mutagenic analysis of relaxed nucleotide specificity
.
Biochemistry
42
,
10644
10650
16
Scapin
,
G.
,
Reddy
,
S.G.
,
Zheng
,
R.
and
Blanchard
,
J.S.
(
1997
)
Three-dimensional structure of Escherichia coli dihydrodipicolinate reductase in complex with NADH and the inhibitor 2,6-pyridinedicarboxylate
.
Biochemistry
36
,
15081
15088
17
Janowski
,
R.
,
Kefala
,
G.
and
Weiss
,
M.S.
(
2010
)
The structure of dihydrodipicolinate reductase (DapB) from Mycobacterium tuberculosis in three crystal forms
.
Acta Crystallogr. D Biol. Crystallogr.
66
,
61
72
18
Pearce
,
F.G.
,
Sprissler
,
C.
and
Gerrard
,
J.A.
(
2008
)
Characterization of dihydrodipicolinate reductase from Thermotoga maritima reveals evolution of substrate binding kinetics
.
J. Biochem.
143
,
617
623
19
Christensen
,
J.B.
,
Soares da Costa
,
T.P.
,
Faou
,
P.
,
Pearce
,
F.G.
,
Panjikar
,
S.
and
Perugini
,
M.A.
(
2016
)
Structure and function of cyanobacterial DHDPS and DHDPR
.
Sci. Rep.
6
,
37111
20
Girish
,
T.S.
,
Navratna
,
V.
and
Gopal
,
B.
(
2011
)
Structure and nucleotide specificity of Staphylococcus aureus dihydrodipicolinate reductase (DapB)
.
FEBS Lett.
585
,
2561
2567
21
Dommaraju
,
S.
,
Gorman
,
M.A.
,
Dogovski
,
C.
,
Pearce
,
F.G.
,
Gerrard
,
J.A.
,
Dobson
,
R.C.J.
et al. 
(
2010
)
Cloning, expression and crystallization of dihydrodipicolinate reductase from methicillin-resistant Staphylococcus aureus
.
Acta Crystallogr. Sect. F Struct. Biol. Cryst.
66
,
57
60
22
Tyagi
,
V.V.S.
,
Henke
,
R.R.
and
Farkas
,
W.R.
(
1983
)
Partial purification and characterization of dihydrodipicolinic acid reductase from maize
.
Plant Physiol.
73
,
687
691
23
Griffin
,
M.D.W.
,
Billakanti
,
J.M.
,
Wason
,
A.
,
Keller
,
S.
,
Mertens
,
H.D.T.
,
Atkinson
,
S.C.
et al. 
(
2012
)
Characterisation of the first enzymes committed to lysine biosynthesis in Arabidopsis thaliana
.
PLoS ONE
7
,
e40318
24
Emanuelsson
,
O.
,
Nielsen
,
H.
and
von Heijne
,
G.
(
1999
)
Chlorop, a neural network-based method for predicting chloroplast transit peptides and their cleavage sites
.
Protein Sci.
8
,
978
984
25
Studier
,
F.W.
(
2005
)
Protein production by auto-induction in high-density shaking cultures
.
Protein Expr. Purif.
41
,
207
234
26
Schuck
,
P.
(
2000
)
Size-distribution analysis of macromolecules by sedimentation velocity ultracentrifugation and Lamm equation modeling
.
Biophys. J.
78
,
1606
1619
27
Schuck
,
P.
,
Perugini
,
M.A.
,
Gonzales
,
N.R.
,
Howlett
,
G.J.
and
Schubert
,
D.
(
2002
)
Size-distribution analysis of proteins by analytical ultracentrifugation: strategies and application to model systems
.
Biophys. J.
82
,
1096
1111
28
Laue
,
T.M.
,
Shah
,
D.B.
,
Ridgeway
,
T.M.
and
Pelletier
,
S.L.
(
1992
) Computer-aided interpretation of analytical sedimentation data for proteins. In
Analytical Ultracentrifugation in Biochemistry and Protein Science
(
Harding
,
S.E.
,
Rowe
,
A.J.
and
Horton
,
J.C.
, eds), pp.
90
125
,
The Royal Society of Chemistry
,
Cambridge
29
Kabsch
,
W.
(
2010
)
XDS.
Acta Crystallogr. D Biol. Crystallogr.
66
,
125
132
30
Evans
,
P.R.
and
Murshudov
,
G.N.
(
2013
)
How good are my data and what is the resolution?
Acta Crystallogr. D Biol. Crystallogr.
69
,
1204
1214
31
Sheldrick
,
G.M.
(
2010
)
Experimental phasing with SHELXC/D/E: combining chain tracing with density modification
.
Acta Crystallogr. D Biol. Crystallogr.
66
,
479
485
32
Langer
,
G.
,
Cohen
,
S.X.
,
Lamzin
,
V.S.
and
Perrakis
,
A.
(
2008
)
Automated macromolecular model building for X-ray crystallography using ARP/wARP version 7
.
Nat. Protoc.
3
,
1171
1179
33
Murshudov
,
G.N.
,
Skubák
,
P.
,
Lebedev
,
A.A.
,
Pannu
,
N.S.
,
Steiner
,
R.A.
,
Nicholls
,
R.A.
et al. 
(
2011
)
REFMAC5 for the refinement of macromolecular crystal structures
.
Acta Crystallogr. D Biol. Crystallogr.
67
,
355
367
34
Emsley
,
P.
,
Lohkamp
,
B.
,
Scott
,
W.G.
and
Cowtan
,
K.
(
2010
)
Features and development of Coot
.
Acta Crystallogr. D Biol. Crystallogr.
66
,
486
501
35
Coulter
,
C.V.
,
Gerrard
,
J.A.
,
Kraunsoe
,
J.A.E.
and
Pratt
,
A.J.
(
1999
)
Escherichia coli dihydrodipicolinate synthase and dihydrodipicolinate reductase: kinetic and inhibition studies of two putative herbicide targets
.
Pestic. Sci.
55
,
887
895
36
Kirby
,
N.M.
,
Mudie
,
S.T.
,
Hawley
,
A.M.
,
Cookson
,
D.J.
,
Mertens
,
H.D.T.
,
Cowieson
,
N.
et al. 
(
2013
)
A low-background-intensity focusing small-angle X-ray scattering undulator beamline
.
J. Appl. Crystallogr.
46
,
1670
1680
37
Petoukhov
,
M.V.
,
Franke
,
D.
,
Shkumatov
,
A.V.
,
Tria
,
G.
,
Kikhney
,
A.G.
,
Gajda
,
M.
et al. 
(
2012
)
New developments in the ATSAS program package for small-angle scattering data analysis
.
J. Appl. Crystallogr.
45
,
342
350
38
Svergun
,
D.
,
Barberato
,
C.
and
Koch
,
M.H.J.
(
1995
)
CRYSOL – A program to evaluate X-ray solution scattering of biological macromolecules from atomic coordinates
.
J. Appl. Crystallogr.
28
,
768
773
39
Krissinel
,
E.
and
Henrick
,
K.
(
2005
)
Multiple alignment of protein structures in three dimensions
. In
Computational Life Sciences: First International Symposium, CompLife 2005
,
Konstanz, Germany
,
September 25–27, 2005
.
Proceedings (R. Berthold, M., Glen, R. C., Diederichs, K., Kohlbacher, O. and Fischer, I., eds)
, pp.
67
78
,
Springer
,
Berlin, Heidelberg
40
Dayhoff
,
J.E.
,
Shoemaker
,
B.A.
,
Bryant
,
S.H.
and
Panchenko
,
A.R.
(
2010
)
Evolution of protein binding modes in homooligomers
.
J. Mol. Biol.
395
,
860
870
41
Panjkovich
,
A.
and
Svergun
,
D.I.
(
2016
)
Deciphering conformational transitions of proteins by small angle X-ray scattering and normal mode analysis
.
Phys. Chem. Chem. Phys.
18
,
5707
5719
42
Ghislain
,
M.
,
Frankard
,
V.
and
Jacobs
,
M.
(
1990
)
Dihydrodipicolinate synthase of nicotiana sylvestris, a chloroplast-localized enzyme of the lysine pathway
.
Planta
180
,
480
486
43
Jabs
,
A.
,
Weiss
,
M.S.
and
Hilgenfeld
,
R.
(
1999
)
Non-proline Cis peptide bonds in proteins
.
J. Mol. Biol.
286
,
291
304
44
Lynch
,
M.
(
2013
)
Evolutionary diversification of the multimeric states of proteins
.
Proc. Natl Acad. Sci. U.S.A.
110
,
E2821
E2828
45
Marsh
,
J.A.
and
Teichmann
,
S.A.
(
2015
)
Structure, dynamics, assembly, and evolution of protein complexes
.
Annu. Rev. Biochem.
84
,
551
575
46
Ge
,
X.
,
Olson
,
A.
,
Cai
,
S.
and
Sem
,
D.S.
(
2008
)
Binding synergy and cooperativity in dihydrodipicolinate reductase: implications for mechanism and the design of biligand inhibitors
.
Biochemistry
47
,
9966
9980

Author notes

*

These authors contributed equally to this work