Structural basis for the catalytic mechanism and α-ketoglutarate cooperativity of glutamate dehydrogenase

Glutamate dehydrogenase (GDH) is a key enzyme connecting carbon and nitrogen metabolism in all living organisms. Despite extensive studies on GDHs from both prokaryotic and eukaryotic organisms in the last 40 years, the structural basis of the catalytic features of this enzyme remains incomplete. This study reports the structural basis of the GDH catalytic mechanism and allosteric behavior. We determined the first high-resolution crystal structures of glutamate dehydrogenase from the fungus Aspergillus niger (AnGDH), a unique NADP+-dependent allosteric enzyme that is forward-inhibited by the formation of mixed disulfide. We determined the structures of the active enzyme in its apo form and in binary/ternary complexes with bound substrate (α-ketoglutarate), inhibitor (isophthalate), coenzyme (NADPH), or two reaction intermediates (α-iminoglutarate and 2-amino-2-hydroxyglutarate). The structure of the forward-inhibited enzyme (fiAnGDH) was also determined. The hexameric AnGDH had three open subunits at one side and three partially closed protomers at the other, a configuration not previously reported. The AnGDH hexamers having subunits with different conformations indicated that its α-ketoglutarate–dependent homotropic cooperativity follows the Monod–Wyman–Changeux (MWC) model. Moreover, the position of the water attached to Asp-154 and Gly-153 defined the previously unresolved ammonium ion-binding pocket, and the binding site for the 2′-phosphate group of the coenzyme was also better defined by our structural data. Additional structural and mutagenesis experiments identified the residues essential for coenzyme recognition. This study reveals the structural features responsible for positioning α-ketoglutarate, NADPH, ammonium ion, and the reaction intermediates in the GDH active site.

Enzymes are important biological macromolecules, and their catalytic functions govern a number of biological activities in all living organisms. Visualization of the active site of an enzymesubstrate complex or an enzyme bound to the catalytically competent reaction intermediate provides direct proof of the reaction mechanism (1). Allosteric regulation of enzymes is one of the most fundamental processes that control several cellular activities. Obtaining quantitative molecular description of enzyme allostery has remained a central focus in biology (2). However, trapping the various structural intermediate states to gain detailed understanding about the kinetic properties of allosteric enzymes has remained very challenging (2,3). Glutamate dehydrogenase (GDH) 2 is an oxidoreductase important for ammonia metabolism in archebacteria, eubacteria, and eukaryotes (4,5). We have extensively studied this enzyme to discern the structural basis of unique properties related to its catalytic mechanism and allosteric behavior. GDH catalyzes the reversible oxidation of L-glutamate to ␣-ketoglutarate and serves as a coupler between carbon and nitrogen metabolism. Depending on the coenzyme specificity, GDHs can be classified as follows: (a) NADP ϩ -dependent; (b) NAD ϩ -dependent; and (c) NAD ϩ /NADP ϩ -dependent (or dual-specific) (6 -8). Generally, the GDHs involved in ammonium assimilation are NADP ϩ -specific, whereas the NAD ϩ -dependent enzymes are involved in glutamate catabolism (9). The mammalian GDHs possessing dual specificities can use either NAD ϩ or NADP ϩ with comparable efficiency and are allosterically regulated (10,11). NADP ϩ and NAD ϩ are identical except that NADP ϩ has an extra phosphate group attached to the 2Ј-hydroxyl of the adenosine. This specific difference is structurally remote to the reactive nicotinamide groups of these two coenzymes. Accordingly, the redox potentials of these two coenzymes are almost identical (8). Nature has developed multiple classes of GDHs that can efficiently discriminate between these two coenzymes, but how they accomplish this differentiation is unclear.
The NADP ϩ -specific bacterial/fungal GDHs and the dual coenzyme-specific mammalian GDHs are hexameric (12). The NAD ϩ -dependent bacterial/fungal enzymes are either homohexamers (13) or homotetramers (14,15). Despite numerous studies in the last 40 years to identify the intermediates formed during the reaction catalyzed by GDHs (16 -20), the structural basis of the reaction mechanism of this enzyme remains unresolved. A number of medium/low-resolution crystal structures of GDHs have been determined as complexes with substrates (␣-ketoglutarate or glutamate) and coenzymes (NADP ϩ or NAD ϩ ) (7,12,13,16,(21)(22)(23). However, these structures fail to explain the reaction mechanism and the role of different activesite residues in catalysis, mainly due to their limited resolution and unavailability of any intermediate-bound, catalytically competent structures of this enzyme. The binding site of ammonia has yet to be identified in the GDH active site. Structural determinants for the recognition of NADP ϩ in NADP ϩspecific GDHs have remained ambiguous. Significantly, no representative structure of a fungal enzyme is available so far.
We report the first high-resolution crystal structures of fungal GDH. The structures of AnGDH were solved as apoenzyme, catalytically competent ternary complexes, as well as complexes with reaction intermediates and an inhibitor. The structures of forward-inhibited AnGDH (fiAnGDH) have also been determined as apoenzyme and complexed with ␣-ketoglutarate (AKG). Analysis of structures complemented with functional characterization demonstrates the structural basis of the coenzyme specificity and kinetic cooperativity of this enzyme. Our data provide direct proof for some of the reaction intermediates of the catalytic mechanism. The results presented here are broadly applicable to all GDHs studied so far, and some aspects extend to dehydrogenases in general.

Structural fold of AnGDH
We determined the first crystal structures of GDH from the fungus kingdom (Fig. 1). The structures of AnGDH were solved as apoenzyme as well as its complex with substrate (␣-ketoglutarate)-coenzyme (NADPH), reaction intermediates (␣-iminoglutarate (AIG)-2-amino-2-hydroxyglutarate (AHG)), the fiAnGDH-AKG complex, and inhibitor (isophthalate) at resolutions of 2.8, 1.8, 1.75, 2.25, and 1.9 Å, respectively ( Table 1). The hexamer of fiAnGDH has three subunits complexed with ␣-ketoglutarate and the rest of the subunits in an unliganded form. All the structures are of high quality as reflected by their low R-factors and good stereochemical parameters (Table 1). Only the apo-AnGDH structure has relatively high R-factors, mainly due to the lower resolution and poor redundancy of the diffraction data. All the bound ligands are unambiguously defined in the active site of AnGDH by clear electron densities. AnGDH has high sequence identity (Fig. S2) with other NADP ϩ /NAD ϩ -specific GDHs and a few extra amino acid insertions. The overall structural fold of AnGDH is similar to the previously determined structures of Escherichia coli GDH (EcGDH) and other GDHs (7,12,(21)(22)(23). Each subunit of AnGDH consists of two domains separated by a deep cleft (Fig. 1, a and b). Domain I consists of residues 1-190 and 437-460, and domain II consists of residues 191-436. These domains are mainly composed of ␣-helices and ␤-sheets, which are numbered as H 1 -H 16 and ␤ 1 -␤ 13 , respectively. Domain I plays a significant role in subunit assembly and combines protomers leading to formation of a hexamer with 32 symmetry (Fig. 1c). Domain I is mainly engaged in binding of ␣-ketoglutarate, whereas domain II facilitates binding of NADP(H). Domain II has seven ␣-helices and seven ␤-strands folded in a modified Rossmann fold (28,29) with ␤ 7 H 8 ␤ 8 ␤ 9 H 9 and ␤ 11 H 12 ␤ 12 H 13 ␤ 13 H 7 as first and second motifs, respectively. These two motifs are connected by H 10 H 11 ␤ 10 . Analysis of the domain II of AnGDH-AKG-NADPH complex shows that the residues present in the loops "␤ 8 -loop-␤ 9 " and "H 9 -loop-H 10 " are responsible for NADP(H) recognition (discussed below). The interface between the two domains is formed by the interactions provided by long helices (H 15 and H 16 ). Like other GDHs, AnGDH forms a hexameric structure that may be considered to be composed of two trimers or three dimers. The dimeric interface is mainly formed by ␤ 1 , ␤ 2 , H 1 , and H 16 of the substrate-binding domain I, whereas the trimeric interface formation is mediated by the residues located at the N terminus of H 14 and the C terminus of H 15 . Similar interfaces have also been reported for hexameric structures of EcGDH and Plasmodium falciparum GDH (PfGDH) (9).

Conformational flexibility in AnGDH structure
The apo-AnGDH and fiAnGDH-AKG complexes were crystallized with a hexamer in the asymmetric unit (Fig.  1c). Structural superpositions of both (apo-AnGDH and fiAnGDH-AKG) show lower root mean square deviation (r.m.s.d.) values (0.6 Å) for equivalent monomers, indicating almost identical conformation of these two structures. AnGDH hexamer has a cylindrical shape with approximate height and diameter of 110 and 96 Å, respectively. Interestingly, each subunit of AnGDH hexamer has a different conformation (Fig. 1d) Fig. 1e) between the C␣ atoms of Lys-122 and Arg-280. These results indicate that three protomers at one side of the hexamer are in an open conformation, and three subunits at the opposite side are in a closed one. Such symmetric opening and closing of trimers in a hexameric assembly of a GDH were observed for the first time. Interestingly, the closed subunits of apo-AnGDH and only an ␣-ketoglutaratebound form of fiAnGDH have identical cleft opening, indicating that the subunits in the AnGDH hexamer may remain in both open

Structural insights into the catalytic properties of GDH
(three) and closed (three) states in the absence of any ligand. The opening of the binding cleft in AnGDH-AKG-NADPH complex is much smaller, with a distance of 6.0 Å ( Table 2 and Fig. 1, e and f), and similar values were also observed for other AnGDH ternary complexes reported in this study. Previously, the ternary complex structures of bovine GDH (21), Burkholderia thailandensis (BtGDH), and C. glutamicum (CgGDH) (30) have been reported with the opening cleft distances of 11.4, 9.9, and 9.4 Å, respectively. Hence, AnGDH ternary complex structure is the first representative with super-closed conformation  Table 1 Data collection and refinement statistics The following abbreviations are used: apo-AnGDH, AnGDH structure without any ligand in the active site; AnGDH-AKG-NADPH: AnGDH structure complexed with AKG and NADPH; fiAnGDH-AKG, forward-inhibited AnGDH structure complexed with AKG; AnGDH-AIG-NADP ϩ , AnGDH structure complexed with reaction intermediates AIG/AHG and NADP ϩ ; AnGDH-IPT-NADPH, AnGDH structure complexed with inhibitor IPT and NADPH.

Active site of AnGDH complexed with ␣-ketoglutarate and NADPH
The structure of the ternary complex of AnGDH with ␣-ketoglutarate and NADPH has been determined at 1.8 Å; the electron density for the substrate and coenzyme in the active site was unambiguous (Fig. 2a). This complex represents the first non-mammalian GDH showing correct orientation of the adenosine 2Ј-phosphate group of NADPH in the catalytically competent enzyme active site. The domain closure facilitates appropriate positioning of NADPH and ␣-ketoglutarate as well as interactions of the ligands with active-site residues. ␣-Ketoglutarate is bound via several polar interactions involving residues Lys-78, Gln-99, Lys-102, Lys-114, Asp-154, Arg-193, and Asn-346 ( Fig. 2b and Fig. S3a). The distance between the reactive carbonyl carbon (C2) of ␣-ketoglutarate and hydride donating/accepting carbon (C4) of the nicotinamide group of the coenzyme in AnGDH is 2.8 Å, and it is 3.8, 4.11, and 4.14 Å, respectively, in the bovine GDH, BtGDH, and CgGDH. In AnGDH, a water molecule located close to the ␣-carbon atom of ␣-ketoglutarate forms a short hydrogen bond (1.9 Å) with the side chain of Lys-114, implying its importance in catalysis. The coenzyme is held in the active-site cleft via interactions with the residues primarily from domain II and a few others from domain I. The adenine ring is anchored inside a pocket formed by His-84, Ile-155, and Thr-321 ( Fig. 2c and Fig. S3b). The ribose of adenosine is placed in a groove formed by the side chains of Ser-229, Asp-252, and Ala-320. The residues Gly-228 -Ala-233 forming the GXGXX(G/A) motif (32) provide hydrogen-bonding interactions to the coenzyme. Ser-229, a part of this motif, forms a hydrogen bond with the 3Ј-hydroxyl group of adenosine. To position the 2Ј-phosphate group of NADPH, the Asp-252 carboxylate group is pointing away from the ribose sugar and forms a salt bridge with the side chain of Lys-277. In NAD ϩ -dependent dehydrogenases, the position of Asp-252 is generally occupied by an Asp or Glu located at the C terminus of the second ␤-strand of the ␤␣␤-fold; these residues form important hydrogen bonds with the 2Ј-hydroxyl group of adenosine (33,34). In AnGDH, the 2Ј-phosphate group is anchored primarily via direct hydrogen bonds with the side chains of Ser-253 and Gln-282 of domain II and Lys-122 of domain I. Lys-277 and His-84 side chains also form water-mediated hydrogen-bonding interactions with the 2Ј-phosphate group. Alanine mutants of these five residues were generated, and the observed deviation of the measured kinetic parameters (Table 3 and Fig. S4) of the mutants confirms involvement of these residues in NADPH binding. The alanine mutants of Ser-253, Lys-277, and Gln-282 lose significant amount of the NADPH-dependent enzymatic activity as compared with the native enzyme. Notably, none of these mutants as well as the WT enzyme show any measurable NADH-dependent activity. Because of the disruption of the polar interactions by alanine mutation of the polar residues, the apparent K m values for NADPH binding are increased to varying extents in the single mutants. The highest K m value was observed for the K277A mutant, which also showed weak positive cooperativity toward NADPH saturation. A significant decrease in k cat is observed for S253A and Q282A mutants with latter having the lowest value. The catalytic efficiency (k cat /K m ) of S253A, K277A, and Q282A has decreased drastically (220, 40, and 3300 times, respectively) as compared with the WT enzyme. The catalytic efficiency lost in H84A and K122A mutants was 1.7-and 3.0fold, respectively, and is not that significant. The measured kinetic parameters of the NADPH-dependent activities indicate primary involvement of Ser-253, Lys-277, and Gln-282 in binding the 2Ј-phosphate group of the cofactor.
The pyrophosphate group of NADPH is hydrogen-bonded to the main chain of the AnGDH GXGXX(G/A) motif (formed by residues Gly-228 -Ala-233 of domain II) and is also hydrogenbonded with well-defined water molecules nearby. The 2Ј-hydroxyl group of the ribose sugar has hydrogen-bonding interactions with the side chains of Arg-82, Asp-154, and Asn-346

Structural insights into the catalytic properties of GDH
( Fig. 2c). A well-defined electron density (Fig. 2a, inset) indicates that the C4 atom of the nicotinamide ring has tetrahedral geometry, implying the presence of a reduced form of coen-zyme in the enzyme active site. The nitrogen atom of the amide group is hydrogen-bonded to one of the oxygen atoms of the NADPH pyrophosphate moiety and the side chain of Asn-231   (Fig. 2c). The distance between the ␣-carbonyl carbon (C2) of ␣-ketoglutarate and the C4 atom of nicotinamide ring is 2.8 Å, indicating that reactive states of the substrate and coenzyme are trapped in the enzyme active site. Such a close interaction between the coenzyme and substrate has never been captured in crystal structures of GDHs reported previously.

Reaction intermediates in the AnGDH active site
The enzymatic reaction was carried out during the crystallization process following the scheme (Fig. 3a) in the presence of NADP ϩ so that ␣-iminoglutarate (AIG) formed in the enzyme active site does not get reduced to form L-glutamate. The high resolution (1.75 Å) electron density map (Fig. 3, b, panel i, and c) shows the presence of reaction intermediates and NADP ϩ in the active site. Initially, NADP ϩ and ␣-iminoglutarate were refined (Fig. 3b, panel ii) in the active site. However, a positive residual F o Ϫ F c electron density (Fig. 3b, panel ii) remained connected with ␣-iminoglutarate. Refinement of 2-amino-2hydroxyglutarate with partial occupancy could satisfy the remaining positive electron density (Fig. 3b, panel iii).
This complex presents the first structural proof of formation of the ␣-iminoglutarate and 2-amino-2-hydroxyglutarate as intermediates during the reaction catalyzed by a GDH. Notably, the binding mode of ␣-iminoglutarate and ␣-ketoglutarate is identical. The ␣-imino group has polar interactions with the carboxylate group of Asp-154 and the main chain carbonyl group of Gly-153 ( Fig. 3d and Fig. S5b). The ␣-hydroxyl group of 2-amino-2-hydroxyglutarate is hydrogen-bonded to the Lys-114 side chain ( Fig. 3d and Fig. S5a). The carbonyl group of Gly-153 adopts two alternative conformations in this structure. The amino group of Lys-114 side chain acquires different conformations in AnGDH-AKG-NADPH and AnGDH-AIG-NADP ϩ . The distance between the ␣-carbon atom of ␣-iminoglutarate and the C4 atom of the nicotinamide moiety of NADP ϩ in AnGDH-AIG-NADP ϩ structure is 3.0 Å. The binding mode of NADP ϩ in the AnGDH-AIG-NADP ϩ complex is almost identical as that observed for NADPH in the AnGDH-AKG-NADPH complex. The only difference is that in the ␣-iminoglutarate-bound structure the 2Ј-phosphate group has moved closer to Lys-277, and the side chain of this residue directly interacts with the phosphate oxygen atom (Fig. S5c). These structural data support the accommodation of the tetrahedral intermediate for the first time, an entity postulated before but with no such direct structural evidence. Our data directly implicate the formation of an ␣-iminoglutarate and 2-amino-2-hydroxyglutarate as the reaction intermediates in the catalytic mechanism of glutamate dehy-

Binding mode of an inhibitor isophthalate in the AnGDH active site
This is the first structure of an isophthalate (IPT)-bound GDH (Fig. 4a). In fact, there is no isophthalate complexed protein structure available in the PDB. Isophthalate occupies the same binding pocket where ␣-ketoglutarate binds in the AnGDH active site ( Fig. 4b and Fig. S6). All eight carbons of isophthalate are present in the plane of the benzene ring, but the oxygen atoms of the carboxylate groups are out of this plane. In contrast, the crystalline form of free isophthalate is reported to form a planar structure with all the atoms residing in one plane (35). The two carboxylates of isophthalate make identical interactions as those observed for ␣-ketoglutarate. The active site-bound NADPH has an identical conformation in both AnGDH-AKG-NADPH and AnGDH-IPT-NADPH complexes. Although the two carboxylate groups of the substrate and the inhibitor occupy conformationally similar positions, their other carbon atoms do not. Interestingly, due to the presence of a hydrophobic inhibitor, the water molecule observed close to Lys-114 in the AnGDH-AKG-NADPH is displaced in the AnGDH-IPT-NADPH complex. Notably, isophthalate has caused ϳ58.9°rotation of the peptide bond between Gly-153 and Asp-154 as compared with the ␣-ketoglutarate-bound structure (Fig. 4b). The conformational change of Gly-153 main-chain carbonyl group is enough to position the aromatic ring of isophthalate in the binding pocket, suggesting the plasticity of the AnGDH active site. This plasticity may be essential for binding the reaction intermediates during catalysis.

Hexameric structure of fiAnGDH with open and partially closed subunits
In the hexameric fiAnGDH structure, three monomers (of one trimer) are in partially closed conformation and complexed with ␣-ketoglutarate. The remaining unliganded three subunits (of other trimer) have open conformations ( Table 2). The covalent modification of Cys-141 in all six subunits is clearly visible in the electron density map (Fig. S7). The ␣-ketoglutarate molecules bound in the active site of partially closed fiAnGDH sub-units have different conformations when compared with those observed in AnGDH-AKG-NADPH complex (Fig. 5, a and b). The striking conformational difference is seen for the ␣-carbonyl group, which mainly forms hydrogen bonds with the side chain of Lys-114 and main chain -NH group of Gly-80 in fiAnGDH-AKG complex, whereas in AnGDH-AKG-NADPH complex, this group is primarily interacting with the side chain of Asp-154 as well as the main chain carbonyl group of Gly-153. The carboxylate group of Asp-154 interacts with the side chains of Arg-82. The spatial arrangement of the AnGDH active site suggests that Arg-82 might play a crucial role in maintaining the ionization state of Asp-154. Arg-82 is highly conserved among NAD ϩ -and NADP ϩ -dependent GDHs (Fig. S2). The K m , k cat , and k cat /K m values for NH 4 ϩ binding to the R282Q mutant are 22.3 Ϯ 1.1 mM, 10.3 Ϯ 0.5 s Ϫ1 , and 4.6 ϫ 10 Ϫ1 Ϯ 0.02 mM Ϫ1 s Ϫ1 , respectively (Fig. S8). For the native enzyme, the K m , k cat , and k cat /K m values for NH 4 ϩ are 1.4 Ϯ 0.2 mM, 106.5 Ϯ 0.9 s Ϫ1 , and 76.1 Ϯ 9.3 mM Ϫ1 s Ϫ1 , respectively. An almost 16-fold increase in K m value and a 165 times decrease in catalytic efficiency of R82Q mutant are consistent with the role of Arg-82 in AnGDH catalysis as proposed above.
In the AKG-bound fiAnGDH monomers, an active-site water molecule is visible, and it is hydrogen-bonded to the carboxylate group of Asp-154 and the main-chain carbonyl group of Gly-153 (Fig. 5a). This water molecule has not been observed before in any of the available GDH structures. Position of this water defines the space needed for ammonia binding during the catalytic conversion of ␣-ketoglutarate to L-glutamate (see "Discussion").

Intersubunit interactions in AnGDH structures
Kinetic measurements of AnGDH showed sigmoidal saturation with ␣-ketoglutarate and hyperbolic saturation with NADPH (Fig. 6, a and b). We performed careful analysis of the hexameric structures of this enzyme to decipher the structural basis of this allosteric feature. It is evident ( Table 2) that individual subunits undergo conformational changes in the presence and absence of substrate, coenzyme, and ligand. During the catalytic cycle, domain I of each subunit of the hexameric assembly remains almost unchanged. However, domain II moves closer to domain I, and it undergoes clockwise rotation (around 20°) with respect to the 3-fold axis of the hexamer. The Each subunit of the central hexamer has 3-fold symmetryrelated lateral interactions with two neighboring subunits from different hexamers present in the same horizontal plane (Fig. 6, c-e). On one side (with the A-C subunits) of the hexamer, the lateral interactions are mediated by the side chains of Asn-335 and the main chain carbonyl group of Thr-362. On the other side of the hexamer (with D-F subunits), the H 13 helix (residues 351-363) of one monomer is packed in the groove formed by the H 10 helix (residues 283-287) and ␤ 10 (residues 298 -302) from a subunit of the neighboring hexamer through hydropho-

Structural insights into the catalytic properties of GDH
bic and van der Waals interactions. The methyl groups of Thr-356 and Thr-362 side chains from one subunit have hydrophobic contacts with the side chains of Ala-302 and C␤ carbon atom of Ser-285, respectively, of the other subunit. Therefore the lateral interactions are mediated by Ser-285, Ala-302, Asn-335, Thr-356, and Thr-362.
The inter-subunit interactions between the hexamers in the vertical directions are asymmetric (Fig. 6, c, f, and g) Analysis of the crystal packing of the AnGDH ternary complex (AnGDH-AKG-NADPH) revealed that the crystallographichexamersarecomposedofsymmetricallyidenticalmonomers arranged in the crystal with a large (104 Å diameter) solvent cavity (Fig. S9). The ternary complexes in the crystallographic hexamer are in the super-closed conformations ( Table  2)

Discussion
GDH is an essential enzyme in all living organisms. Despite extensive studies (17,30) in the last 40 years on prokaryotic as well as eukaryotic GDHs, the structural basis of the coenzyme specificity and the mechanistic features of this enzyme remained incomplete. Also, no structural information was available for a fungal enzyme. This prompted us to perform structural studies on A. niger GDH (AnGDH). Another reason for this study was to understand the structural basis of the cooperative nature of this fungal enzyme.
This study presents the high resolution crystal structure of substrate-and cofactor-bound GDH Michaelis-Menten complex. We report the first fungal GDH structure. The hexameric AnGDH is unique with three open subunits at one side and three partially closed protomers at the other side. Such hexameric GDH structure has not been reported before. Partially closed subunits bind the substrate ␣-ketoglutarate. The position of the water attached to Asp-154 and Gly-153 defines the ammonium ion-binding pocket, which had remained unresolved. The binding pocket for the 2Ј-phosphate group of the coenzyme is better defined by our structural data. The structure of the AnGDH-AKG-NADPH complex provides a glimpse of the super-closed catalytically competent enzyme with substrate and coenzyme at a favorable distance for hydride transfer. The structure of AnGDH-AIG-NADP ϩ complex captures the formation of ␣-iminoglutarate and 2-amino-2-hydroxyglutarate during the reaction. AnGDH-IPT-NADPH complex is the first structure of a protein-isophthalate complex and reveals the plasticity of the enzyme active site. Implications of the elucidated structures to the GDH reaction mechanism, coenzyme recognition, and substrate cooperativity are discussed below.

Structural basis of ␣-ketoglutarate cooperativity in AnGDH
Our structural data provide the possible explanation of the ␣-ketoglutaratedependent cooperativity in AnGDH. It is evident that due to the inter-subunit interactions, the protomers of a catalytically incompetent hexamer are locked either in an open or partially closed conformation. In the super-closed conformation, the enzyme gains a catalytically competent form. These different conformational states can be directly correlated to the cooperative behavior of AnGDH. The Monod-Wyman-Changeux (MWC) model (36) and the Koshland-Nemethy-Filmer (KNF) (37) model are the two generally accepted models used to explain the kinetic cooperativity in a multimeric allosteric enzyme. According to the MWC model, the subunits of the enzyme are present in a reversible equilibrium between a low-affinity tensed (T) and a high-affinity relaxed (R) state in the absence of the ligand. When added, the ligand would bind to the R-state, and the equilibrium would adjust by converting more of the T-state subunits to the R-state, thereby leading to positive cooperativity. In contrast, the KNF model indicates that ligand binding to one subunit can induce conformational changes in the other subunits resulting in an increase in ligand affinity (and hence positive cooperativity).
Analysis of the structural data collated so far suggests that the unliganded structure of apo-AnGDH represents the low-affinity state (resting-state), in which half of the hexameric enzyme (one trimer) remains as open/partially closed conformation (Fig. 7). Another low-affinity form (resting-state) of the enzyme with all its six subunits in the open conformation may well exist; however, we have not encountered such a structure so far. The hexameric unit with a trimer of three ␣-ketoglutaratebound, partially closed subunits at one side and the trimer with three open subunits on the other side represents the T-state and is

Structural insights into the catalytic properties of GDH
also not catalytically competent. Closing and opening of one trimer subunit at a time also correlates well with the measured Hill coefficient of 2.5 for ␣-ketoglutarate saturation (26). Binding of ␣-ketoglutarate and NADPH to all six subunits would generate a fully active enzyme species (R-state); this is consistent with the biphasic kinetic response to incubation with the product NADP ϩ and higher initial velocities with prior incubation of AnGDH with ␣-ketoglutarate and NADPH (26). Therefore, the homotropic ␣-ketoglutarate interactions of AnGDH may possibly be described by the mechanism (Fig. 7) following the MWC model.
Our data present the first structural evidence for the allosteric regulation in smaller GDHs. Based on the crystal structures of bovine GDH, the allosteric regulation in additional antenna domain-containing mammalian GDH has been previously reported to follow the KNF model (38). The MWC allosteric model observed in AnGDH may be applicable to smaller GDHs without the antenna domain, as this model nicely correlates with the measured Hill coefficient of 5.9 (39) of CsGDH. This implies that collective conformational transition of all six subunits in the hexamer contributes toward CsGDH positive cooperativity. The amino acid residues engaged in forming inter-subunit contacts in the AnGDH hexamer are replaced with other residues in AtGDH (sequence alignment; Fig. S1). Such interactions may account for the absence of ␣-ketoglutarate cooperativity in AtGDH. A systematic mutation analysis of these residues is expected to convert an allosteric AnGDH into a Michaelian enzyme and to allow further probing into the mechanistic details of the allosteric regulation in this enzyme as well as other smaller GDHs.

Structural basis of NADP(H) recognition by AnGDH
The precisely defined position of the coenzyme in the highresolution electron density map of our AnGDH structures rationalizes the structural basis of coenzyme recognition in this enzyme as well as in other NADP ϩ -dependent GDHs. Structural and mutagenesis data indicate that Lys-122, Ser-253, Lys-277, and Gln-282 are the primary determinants for coenzyme binding in AnGDH. In the AnGDH-complexed structures, the Lys-277 side chain makes a direct (Fig. 8a) or water-mediated ( Fig. 2c) contact with the 2Ј-phosphate group of coenzyme. This observation suggests that the ionization state of the 2Ј-phosphate group of NADP(H) might dictate its interactions with these residues (Fig. 8, a and b).
In the NADP ϩ -bound AnGDH-AIG-NADP ϩ complex structure, the 2Ј-phosphate group might be dibasic (Fig. 8b); conversely, in other coenzyme-bound structures this group might be monobasic. Lys-277 is conserved in all the NADP ϩdependent GDHs (Fig. 8c). Interestingly, only K277A AnGDH mutant shows mild positive cooperativity toward NADPH, whereas the other mutants and WT enzyme show no such kinetic behavior. This result makes Lys-277 of AnGDH a unique residue governing the binding of 2Ј-phosphate of NADP(H).
One of the oxygen atoms (O2) of the 2Ј-phosphate group might remain neutral in both forms due to its close proximity to the Asp-252 side chain (Fig. 8, a and b), which is possibly negatively charged as it interacts with the side chains of Gln-282 and Lys-277. Although in AnGDH the Asp-252 carboxylate group does not interact directly with the phosphate group, it might still play a crucial role in positioning Gln-282 and Lys-277 side chains (Fig. 8a). In AnGDH, Ser-253 and Gln-282 side chains form hydrogen bonds with O2 oxygen atom due to its close proximity. Similar interactions of NADP ϩ 2Ј-phosphate group oxygen atom with threonine and arginine side chains are also observed in Lactococcus lactis 6-phosphogluconate dehydrogenase (LlPGDH) (40).
Both Asp-252 and Ser-253 of AnGDH are strictly conserved in the NADP ϩ -dependent GDHs (41). Asp-252 is the signature C-terminal end residue of the second ␤-strand of the ␤␣␤-fold (Fig. 8d), and its negatively charged side chain has been so far believed to destabilize the binding of NADP ϩ 2Ј-phosphate (42,43). Our results depict the essentiality of Asp-252 for optimal orientation of Lys-277 and Gln-282, which facilitate NADP ϩ binding. The presence of Lys-122, Lys-277, and Gln-282 creates a positively charged surface (Fig. S10) for binding the 2Ј-phosphate of NADP(H) in AnGDH. In other NADP ϩ -dependent GDHs, this Gln-282 is replaced by an amino acid residue with a proton donor -NH group (Fig. S2). The lowest catalytic effi-

Structural insights into the catalytic properties of GDH
ciency of the AnGDH Q282A mutant points to the importance of this residue in coenzyme recognition. In EcGDH (12), CbGDH (30), the side chain of an arginine residue, might serve the same purpose.
The other oxygen atom (O4) of the 2Ј-phosphate group should remain negatively charged and form an ionic interaction with a suitable residue on the protein. The Lys-122 of AnGDH satisfying this ionic interaction is highly conserved in NADP ϩdependent GDHs. The similar role for an equivalent lysine residue in CgGDH has also been proposed (30,44). The O3 oxygen atom of the 2Ј-phosphate may remain neutral or be negatively charged. The existence of two different ionization states of the 2Ј-phosphate group of NADP(H) was proposed before (32) but without any relevant structural data. The interactions present in the NADP ϩ -and NADPH-bound AnGDH structures provide the first direct evidence justifying the proposed ionization states of the 2Ј-phosphate group. However, further high-resolution structures, systematic mutagenesis studies, kinetic parameter measurements, and quantum mechanics/molecular mechanics calculations would prove valuable.

GDH reaction mechanism: Conversion of ␣-ketoglutarate to L-glutamate
The conversion of ␣-ketoglutarate to L-glutamate by GDH takes place in sequential steps as suggested through the extensive biochemical studies in last 4 decades. The identification of two active-site lysine residues (45-48) prompted the sugges-tion that the reaction mechanism proceeds through formation of a Schiff's base (49). However, this idea was ruled out due to lack of experimental evidence (50). Later, based on isotope exchange rates and spectroscopic studies, it was proposed that the reductive amination of ␣-ketoglutarate proceeds through an enzyme-bound ␣-iminoglutarate intermediate (13,19,50,51). The reaction mechanism involves the nucleophilic attack by ammonia on the ␣-ketoglutarate of the GDH-NADPH-␣ketoglutarate ternary complex. The ␣-ketoglutarate carbonyl oxygen was proposed to be protonated by a catalytic lysine forming the carbinolamine intermediate. Elimination of water from this carbinolamine generates the ␣-iminoglutarate intermediate, which was spectroscopically identified in bovine GDH (17). Proton transfers in this step are possibly assisted by a carboxylate group of an aspartate (Asp-165 in CsGDH) (13). The reduced coenzyme, stacking in close proximity to the newly formed imino group, completes the facile hydride transfer to form L-glutamate. Besides a catalytic carboxylate-containing residue, two active-site lysines (Lys-113 and Lys-125 in CsGDH) are also conserved in all NADP ϩ -dependent GDHs. Despite numerous mechanistic studies, the direct structural evidence to capture the reaction intermediates was unavailable, and the ammonia binding pocket was yet to be established.
The constellation of AnGDH active site residues and their binding mode with substrate, coenzyme, and reaction intermediates unraveled the snapshots of steps that occur during the

Structural insights into the catalytic properties of GDH
catalytic reductive amination of ␣-ketoglutarate to L-glutamate. Our structural analysis reveals NADPH binding induces a major conformational change that places the substrate and coenzyme in a catalytically correct orientation. A water molecule-forming hydrogen bond with Lys-114 in AnGDH structures implies its catalytic importance. The ionic interactions of Asp-154 with Arg-82 and Lys-114 suggest that two latter residues might play a crucial role in the catalysis. In a catalytically competent active site, the carboxylate group of Asp-154 is likely to remain negatively charged by interacting with the positively charged side chains of Arg-82 and Lys-114 (Fig. 9). Our kinetics data on the R82Q AnGDH mutant and strict conservation of this residue (Fig. S2) among other GDHs implicate involvement of Arg-82 in the catalytic reaction mechanism. The importance of Lys-114 in catalysis was demonstrated before in CgGDH (30). The Asp-154 side chain and the main chain of Gly-153 might play a crucial role in anchoring an ammonium ion. The position of a water molecule, hydrogen-bonded to these two groups in the AKG-bound subunits of the fiAnGDH structure (Fig. 5a), supports this view. Our structural analysis promotes the idea that the negatively charged carboxylate group of Asp-154 and the polarized main-chain carbonyl group of

Structural insights into the catalytic properties of GDH
Gly-153 are involved in positioning the NH 4 ϩ ion in the active site. Reported weaker ammonium ion affinity for EcGDH D165S as compared with the WT enzyme with almost unchanged affinity for NADPH and ␣-ketoglutarate (52) agrees with this proposition.
Our data provide direct structural evidence supporting the following AnGDH reaction mechanism (Fig. 9) similar to the schemes proposed for the GDH reaction in general. First, the Asp-154 carboxylate group deprotonates NH 4 ϩ and facilitates nucleophilic attack by ammonia onto the ␣-carbon of ␣-ketoglutarate. The generated oxyanion becomes protonated by the neighboring water molecule located near Lys-114; this in turn regenerates the water molecule by proton donation. A tetrahedral intermediate (2-amino-2-hydroxyglutarate) is formed in the active site, and the AnGDH-AIG-NADP ϩ structure directly supports formation of such an intermediate (Fig. 3b). Elimination of a water molecule leads to ␣-iminoglutarate formation. Subsequently, the hydride transfer from NADPH to ␣-iminoglutarate forms L-glutamate. The orientation of the nicotinamide ring of NADPH in the AnGDH-AKG-NADPH complex structure represents the catalytically competent conformation of the coenzyme capable of hydride transfer. Theoretical calculation on alcohol dehydrogenase reported (53) the required distance for hydride transfer between NAD ϩ and substrate to be around 2.7 Å; the corresponding distances observed in the catalytically competent AnGDH structures are consistent. On the whole, collated structural data amply illuminate the key features of the GDH reaction mechanism.

Conclusions
Our results provide the structural basis of three important aspects related to catalysis by GDH: (a) cofactor specificity, (b) allosteric regulation, and (c) reaction mechanism. We have determined the first crystal structures of a fungal glutamate dehydrogenase. The complexed structures of AnGDH present direct evidence of formation of ␣-iminoglutarate and 2-amino-2-hydroxyglutarate as reaction intermediates. The different conformational states of AnGDH structures suggest that the allosteric regulation in this enzyme follows the MWC model. The structural data reveal that the 2Ј-phosphate group of NADP(H) anchored to NADP(H)-GDH might have two possible ionization states. These findings resonate with other dehydrogenase mechanisms as well.

AnGDH expression and purification
The expression and purification of recombinant AnGDH were performed as described previously (5,26,27), with minor modifications. A single colony of E. coli BL21 (DE3) with the AnGDH expression construct was grown overnight at 37°C in LB broth medium with ampicillin (100 g/ml). The culture (1% v/v) was re-inoculated into the LB medium containing ampicillin (100 g/ml) and grown at 37°C until an optical density of 0.5 at 600 nm, and then protein expression was induced by addition of isopropyl 1-thio-␤-D-galactopyranoside (400 M). Protein expression was done by growing the culture at 22°C for 12 h. The cell pellet obtained from 1 liter of culture was suspended in a lysis buffer (Buffer A: 100 mM potassium phosphate buffer, pH 7.5, 1 mM EDTA, and 1ϫ protease inhibitor mixture). After cell disruption using ultrasonication, the cell-free extract was prepared by centrifugation at 12,000 rpm, and the supernatant was collected. Ammonium sulfate saturation was performed, and the pellet obtained after 30 -70% saturation was dissolved in Buffer B (20 mM potassium phosphate buffer, pH 7.5, and 1 mM EDTA). The sample was further desalted using HiPrep 26/10 desalting column. Next, the protein sample was loaded onto a 50-ml CR-12 dye affinity column (Novacron Red LS-BL coupled through an epoxy spacer arm to Sepharose), and elution was performed by a linear gradient of potassium chloride. The eluted sample was desalted and loaded onto a DEAE-Sepharose (5 ml, HiTrap) column, and bound proteins were eluted by a linear gradient of potassium chloride. The final purification of AnGDH was performed using a Superdex 200 16/60 gel-filtration column. The purity of the protein was analyzed on SDS-PAGE (54). Protein concentrations were measured using Bradford's method (55).

Enzyme assay
The enzyme activity of pure AnGDH was assayed as described previously (5). Reductive amination of ␣-ketoglutarate to L-glutamate is accompanied by oxidation of NADPH to NADP ϩ . Therefore, the enzyme activity was measured by the initial rate of disappearance of NADPH absorbance at 340 nm. The change in absorbance (⌬A 340 ) was recorded per min. Forward reaction was carried out in 1 ml of total reaction mixture containing 100 mM Tris buffer, pH 8.0, 10 mM ␣ketoglutarate, 10 mM ammonium chloride, and 0.1 mM NADPH. One enzyme activity unit refers to the amount of enzyme required to oxidize 1 mol of NADPH per min under the standard assay conditions.

Site-directed mutagenesis
PCR-based site-directed mutagenesis was performed to generate Ala-substituted AnGDH mutants at residues His-84, Lys-122, Ser-253, Lys-277, and Gln-282 using the pET43.1bAnGDH expression vector. R82Q mutant was prepared in a similar way. The primers used for generating these mutants are listed in Table S1. Mutations were confirmed with DNA sequencing, and the expression constructs of these mutants were transformed into ⌬GDH E. coli BL21 (DE3) (56). Expression, purification, and activity assays of the mutant enzymes were performed following the same procedures as described for recombinant WT AnGDH. Correctness of folding of the mutants was confirmed by circular dichroism (CD) measurements.

Kinetics of WT and mutant AnGDH
NADPH saturation of the WT AnGDH and its mutant forms (H84A, K122A, S253A, K277A, and Q282A) was performed using reductive amination assay. This reaction was followed at pH 8.0 with varying substrate (NADPH) concentrations. The standard assay (as mentioned under "Enzyme assay") was suitably modified. The substrate conversion was maintained below 10% to achieve initial velocity. The NADPH concentration was varied, and the enzyme and other substrate concentrations

Structural insights into the catalytic properties of GDH
were kept constant. The saturation of NADH was performed in parallel.
The assay was suitably modified for the ammonia saturation kinetics of WT AnGDH and its R82Q mutant. NADPH and AKG concentrations were kept fixed, and ammonium chloride concentration was varied with a suitable amount of the enzyme. All these experiments were performed in triplicate at room temperature.

Fluorescence quenching experiment
Purified AnGDH (1 mg/ml) in 20 mM phosphate buffer, pH 7.5, was used for measuring the substrate-and cofactor-induced fluorescence intensity change. All the experiments were performed in a fluorescence spectrophotometer (JASCO). For the tryptophan quenching in AnGDH, the excitation wavelength of 295 nm was used, and the emission wavelength was in the range of 300 -500 nm. The excitation and emission band widths were kept at 2.5 nm, and slit width used was 2.5 nm. The reaction mixtures contained a fixed concentration (3 M) of AnGDH and variable concentrations of ␣-ketoglutarate in a total reaction volume of 0.5 ml assembled in a quartz cuvette, and a path length of 10 mm was used. Different concentrations of ␣-ketoglutarate used for the measurements were 2, 5, 7, 9, 15, and 20 mM. In separate experiment, a fixed concentration (6 M) of AnGDH was used with NADPH of varying concentrations (0.2, 0.5, 1, 2, 4, 6, 8, 12, 15, and 18 M). The observed fluorescence intensities were corrected for the inner filter effect in the experiments. All the spectral measurements were carried out at 25°C and in triplicate.

Preparation of forward-inhibited AnGDH
The forward-inhibited AnGDH (fiAnGDH) was prepared by treating the purified enzyme with 2 mM 2-HED and incubating the mixture at 37°C as described previously (27). Forward inhibition of the enzyme was confirmed by activity measurements as discussed above. The fiAnGDH preparation was further buffer-exchanged with Buffer B, concentrated to 12 mg/ml, and stored at 4°C.

Crystallization
Active and forward-inhibited forms of AnGDH were crystallized using sitting or hanging drop vapor diffusion method at 22°C. Initial crystallization screens were set up using commercially available screen solutions: (a) JCSG Core-I suite (Qiagen), (b) PEGs suite (Qiagen), and (c) JCSG plus suite (Molecular Dimensions), with a Phoenix (Art Robins) crystallization robot available at the "Protein Crystallography Facility," Indian Institute of Technology, Bombay, India. Crystallization trials for apo-AnGDH with a protein concentration of 7 mg/ml were set up using factorial 1 screening conditions (57). The first crystals of apo-AnGDH were obtained in a condition containing 15% (w/v) PEG 3350, 0.1 M Tris-Cl, and 0.2 M NaCl within 1 week. Further optimization of this condition using several additives produced the best quality apo-AnGDH crystals in a mother liquor having 20% (w/v) PEG 3350, 0.1 M NaCl, 0.1 M Tris-Cl, pH 8.5, and 0.01 M BaCl 2 .
Crystallization of AnGDH complexed with ␣-ketoglutarate and NADPH (AnGDH-AKG-NADPH) was done as reported before (5). For complex formation, AnGDH (12 mg/ml) was mixed with ␣-ketoglutarate and NADPH for 30 min at 25°C achieving 0.6 mM final concentrations of the substrate and coenzyme. The crystallization screens were set up, and the initial crystals of the AnGDH-AKG-NADPH complex were observed within 1 week in a condition containing 0.1 M sodium citrate, pH 5.5, and 20% (w/v) PEG 3000. These crystals grew to their maximum size within 1 week and were further used for diffraction studies.
To obtain crystals of AnGDH complexed with isophthalate and NADPH (AnGDH-IPT-NADPH), the concentrated (12 mg/ml) protein solution was mixed with these compounds, and the mixture was incubated for 30 min at 25°C. The final concentrations of isophthalate and NADPH in the mixture were 0.8 and 0.6 mM, respectively. Crystallization screens of this complex were set up, and the best crystals were obtained in a condition containing 0.1 M MES, pH 6.0, 30% (v/v) PEG 200, and 5% (w/v) PEG 3000. The crystals grew to their maximum size within 2 weeks.
Preparation of AnGDH complex with ␣-iminoglutarate (AnGDH-AIG-NADP ϩ ) was done by incubating the mixture of enzyme (12 mg/ml) with ␣-ketoglutarate, NADP ϩ , and ammonium chloride at 25°C for 5 min. The final mixture contained 1 mM ␣-ketoglutarate, 1 mM NADP ϩ , and 0.5 M ammonium chloride. Crystallization screens of this complex were set up. After obtaining the first hit, the crystallization conditions were optimized, and the best crystals were obtained in a condition containing 40% (v/v) PEG 300, 0.1 M sodium cacodylate, pH 6.5, and 0.2 M calcium acetate hydrate. The crystals grew to their maximum size within 2 days.
The concentrated fiAnGDH (12 mg/ml) sample was incubated for 30 min at 25°C with ␣-ketoglutarate and NADH to prepare a complex. The final concentrations of ␣-ketoglutarate and NADH used for preparing this complex were 0.2 and 0.2 M, respectively. The crystallization screens for this complex were set up, and the best crystals appeared after 2 weeks in a condition containing 0.15 M potassium bromide and 20% (w/v) PEG 2000 MME, and grew to their maximum size in 2 weeks.

Data collection and processing
All the diffraction data were collected from the frozen crystals by the rotation method. The crystals were briefly transferred to their corresponding cryoprotectant solutions using a nylon loop and subsequently flash-frozen in the liquid nitrogen stream at 100 K. A dataset for the apo-AnGDH crystal was collected using CuK ␣ X-ray radiation source generated by a Bruker MICROSTAR diffractometer equipped with MAR345 detector at the Advanced Centre for Treatment, Research, and Education in Cancer, Navi Mumbai, India. The reservoir solution with 20% (v/v) glycerol was used as a cryoprotectant for freezing the apo-AnGDH crystal. Diffraction data from fiAnGDH crystals were also collected using CuK ␣ radiation generated by a Rigaku Micromax 007HF generator equipped with R-Axis IVϩϩ detector at the Protein Crystallography Facility, IIT Bombay, India. The crystals of AnGDH-AKG-NADPH, AnGDH-IPT-NADPH, and AnGDH-AIG-NADP ϩ complexes were first briefly transferred to the cryoprotectant solutions prepared from their respective mother liquors containing 30% (v/v) glycerol, and then subsequently flash-frozen in liquid nitrogen. The frozen crystals were then transferred to Structural insights into the catalytic properties of GDH the liquid nitrogen stream at 100 K for data collection. The diffraction data sets from these complexes were collected at the BM14 beamline of European Synchrotron Radiation Facility (ESRF), Grenoble, France, using a MarCCD detector. Indexing, integration, and scaling of all the data sets were performed by XDS (58). The intensities were converted to structure factors with the program modules F2MTZ and CAD of CCP4 (59). The data collection statistics are presented in Table 1.

Structure determination, model building, and refinement
The structure of the apo-AnGDH was determined by molecular replacement. The A subunit of E. coli glutamate dehydrogenase (EcGDH) crystal structure (PDB code 3SBO), which has an amino acid sequence identity of 55% with AnGDH, was used as the search model. Calculation of Matthews coefficient (2.7 Å 3 Da Ϫ1 ) (60) indicated the presence of six apo-AnGDH molecules in the asymmetric unit. The correct orientations of six subunits were identified by PHASER (61) using the search model. The hexameric unit of the model was refined for 10 cycles using REFMAC5 (62). The resulting electron density map was used for automated model building using Buccaneer software (63), which could correctly assign almost 60% residues of the hexameric AnGDH structure. The partially built model was used for subsequent manual model building by visual inspection in COOT (64) and refinement using REFMAC5. The solvent molecules and ions were progressively added at peaks of electron density higher than 3 in -A weighted F o Ϫ F c electron density maps while monitoring the decrease of R free and improvement of the overall stereochemistry. In the structure, subunits A-C have one and subunits D-F have two N-terminal residues missing, as they could not be built due to lack of features in the electron density.
The structures of AnGDH-AKG-NADPH complex and fiAnGDH were solved by molecular replacement using the coordinates of apoenzyme A subunit. The initial phases of the structures of other AnGDH complexes were obtained by the rigid body refinement of the protein part of the AnGDH-AKG-NADPH structure as all the complexed crystal forms had almost identical cell dimensions and belonged to the same space group. The first few cycles of refinement of only the protein molecules were performed by REFMAC5. Subsequently, the ligands were placed inside the -A weighted F o Ϫ F c electron density map, and further refinement cycles were carried out. The waters and other solvent molecules were added to the structures, and alternative conformations of residues were built using COOT. Convergence of the refinement process was monitored by the decrease of R free and improvement of the overall stereochemistry. The refinement statistics of all the structures presented in this study are reported in Table 1.
Author contributions-P. P. and P. B. data curation; P. P., N. S. P., and P. B. formal analysis; P. P. and P. B. validation; P. P., N. S. P., and P. B. visualization; P. P. and P. B. methodology; P. P. and P. B. writing-original draft; P. P., N. S. P., and P. B. writing-review and editing; N. S. P. and P. B. conceptualization; P. B. resources; P. B. software; P. B. supervision; P. B. funding acquisition; P. B. investigation; P. B. project administration.