Regulation of the hetero-octameric ATP phosphoribosyl transferase complex from Thermotoga maritima by a tRNA synthetase-like subunit

Authors


  • Present address: California Institute of Technology, BMB 147-75, Pasadena, CA 91125, USA; Universität zu Köln, Institut für Biochemie, Otto-Fischer-Strasse 12-14, D-50674 Köln, Germany. §These authors have contributed equally to this work.

E-mail wilmanns@embl-hamburg.de; Tel. (+49) 40 89902 126; Fax (+49) 40 89902 149.

Summary

The molecular structure of the ATP phosphoribosyl transferase from the hyperthermophile Thermotoga maritima is composed of a 220 kDa hetero-octameric complex comprising four catalytic subunits (HisGS) and four regulatory subunits (HisZ). Steady-state kinetics indicate that only the complete octameric complex is active and non-competitively inhibited by the pathway product histidine. The rationale for these findings is provided by the crystal structure revealing a total of eight histidine binding sites that are located within each of the four HisGS–HisZ subunit interfaces formed by the ATP phosphoribosyl transferase complex. While the structure of the catalytic HisGS subunit is related to the catalytic domain of another family of (HisGL)2 ATP phosphoribosyl transferases that is functional in the absence of additional regulatory subunits, the structure of the regulatory HisZ subunit is distantly related to class II aminoacyl-tRNA synthetases. However, neither the mode of the oligomeric subunit arrangement nor the type of histidine binding pockets is found in these structural relatives. Common ancestry of the regulatory HisZ subunit and class II aminoacyl-tRNA synthetase may reflect the balanced need of regulated amounts of a cognate amino acid (histidine) in the translation apparatus, ultimately linking amino acid biosynthesis and protein biosynthesis in terms of function, structure and evolution.

Introduction

The first committed reaction of the histidine biosynthesis pathway is catalysed by N-1-(5′-phosphoribosyl)-ATP transferase (ATP–PRTase, EC 2.4.2.17) (Aberg et al., 1997). It comprises the reversible, Mg2+-dependent condensation of 5-phosphoribosyl-α1-pyrophosphate (PRPP) and ATP to yield N-1-(5′-phosphoribosyl)-ATP (PR-ATP) and inorganic pyrophosphate (PPi) (Fig. 1). Histidine biosynthesis is one of the most energy-consuming anabolic pathways, and is thus subject to an elaborated multilevel regulation system (Alifano et al., 1996). In Escherichia coli and Salmonella typhimurium, the ATP–PRTase activity is regulated by the pathway product histidine and by the substrate analogues AMP and ADP, which act as competitive inhibitors to both ATP and PRPP (Martin, 1963; Kleeman and Parsons, 1976; Morton and Parsons, 1976; 1977a,b). Moreover, the steady-state concentration of histidine is modulated at the levels of transcription initiation and elongation (Dall-Larsen, 1988; Alifano et al., 1996). The multilevel regulation of ATP–PRTase activity allows rapid adjustment of the amount of histidine available for consumption in other metabolic processes such as histidyl-tRNA aminoacylation for protein biosynthesis.

Figure 1.

Reaction catalysed by tmATP–PRTase. The 5-phosphoribosyl moiety of PRPP is transferred to ATP to yield N-1-(5′-phosphoribosyl)-ATP (PR-ATP) and inorganic pyrophosphate (PPi). The reaction is Mg2+ dependent and reversible.

Comparative genomics and phylogenetic studies have shown that histidine prototrophic organisms can be classified into two groups that differ in the molecular architecture of their ATP–PRTases (Sissler et al., 1999; Bond and Francklyn, 2000). In the first group, which subsumes most bacteria including some proteobacteria, the HisG polypeptide (HisGL) contains 280–310 residues and is sufficient to catalyse the ATP–PRTase reaction. In the second group, which comprises the Archaea, the Eucarya and several proteobacteria, the HisGS subunit comprises only 205–215 residues, lacking 65–105 residues at the C-terminus (Fig. 2A). In contrast to HisGL, HisGS associates with a second polypeptide, termed HisZ, which is required to form a functional ATP–PRTase (Bovee et al., 2002). HisZ is found neither in genomes encoding HisGL nor in organisms lacking the capability to synthesize histidine. Based on sequence comparisons, distant similarities between HisZ proteins and class II aminoacyl-tRNA synthetases (aaRS) have been identified (Fig. 2B), suggesting an evolutionary link between these enzymes (Sissler et al., 1999).

Figure 2.

Figure 2.

Sequences of the catalytic tmHisGS and regulatory tmHisZ subunits.
A. Structure-based sequence alignment of tmHisGS and mtHisGL showing the complete sequence covered by the structure of tmHisGS. Numbering refers to the tmHisGS sequence. Secondary structure elements are shown on top, coloured according to tmHisGS domain organization, with domain I in khaki (residues 1–84 and 169–208) and domain II in yellow (residues 85–168) (Figs 3 and 4). Residues that are strongly conserved (>90%) in multiple sequence alignments of each of the two HisG families (data not shown) are in bold. Residues involved in the histidine-binding pockets 1 and 2 are marked with green asterisks (*) and green pound or hash signs (♯) respectively (Fig. 7). Residues that interact with the phosphate group in the P-loop site are marked with violet asterisks (*), and those that interact with the second phosphate group are marked with violet hash signs (♯) (Fig. 5). The ATP–PRTase PRPP-binding fingerprint sequence (ProSite PDOC01020) is underlined (residues 134–155). Residues forming homodimerization interfaces (tmHisGStmHisGS) are underlined by red line segments, and those that are involved in heterodimerization interfaces (tmHisGStmHisZ) are underlined by blue line segments.
B. Structure-based sequence alignment of tmHisZ and Thermus termophilus ttHisRS, showing the complete sequence of tmHisZ. Numbering refers to the tmHisZ sequence. Secondary structure elements are shown on top, coloured according to tmHisZ domain organization, with the core domain in grey (residues 1–24, 51–132 and 201–275), the proximal subdomain in cyan (residues 25–50) and the distal subdomain in violet (residues 133–200). Residues strongly conserved (>90%) in multiple sequence alignments of the HisZ family and the HisRS family (data not shown) are in bold. Residues involved in the histidine-binding pockets 1 and 2 are marked with green asterisks (*) and green hash signs (♯) respectively (Fig. 7). Fingerprint motifs of the HisRS family (Aberg et al., 1997) are boxed (residue ranges according to the ttHisRS sequence): motif 1 (23–44); motif 2 (101–133); motif 3 (299–310); histidine-binding motif 1 (259–264); and histidine-binding motif 2 (285–290). The two key catalytic arginines are marked with a black asterisk each (*). Residues forming homodimerization interfaces (tmHisZ–tmHisZ) are underlined by red line segments, and those that are involved in heterodimerization interfaces (tmHisGStmHisZ) are underlined by blue line segments.

Figure 2.

Figure 2.

Sequences of the catalytic tmHisGS and regulatory tmHisZ subunits.
A. Structure-based sequence alignment of tmHisGS and mtHisGL showing the complete sequence covered by the structure of tmHisGS. Numbering refers to the tmHisGS sequence. Secondary structure elements are shown on top, coloured according to tmHisGS domain organization, with domain I in khaki (residues 1–84 and 169–208) and domain II in yellow (residues 85–168) (Figs 3 and 4). Residues that are strongly conserved (>90%) in multiple sequence alignments of each of the two HisG families (data not shown) are in bold. Residues involved in the histidine-binding pockets 1 and 2 are marked with green asterisks (*) and green pound or hash signs (♯) respectively (Fig. 7). Residues that interact with the phosphate group in the P-loop site are marked with violet asterisks (*), and those that interact with the second phosphate group are marked with violet hash signs (♯) (Fig. 5). The ATP–PRTase PRPP-binding fingerprint sequence (ProSite PDOC01020) is underlined (residues 134–155). Residues forming homodimerization interfaces (tmHisGStmHisGS) are underlined by red line segments, and those that are involved in heterodimerization interfaces (tmHisGStmHisZ) are underlined by blue line segments.
B. Structure-based sequence alignment of tmHisZ and Thermus termophilus ttHisRS, showing the complete sequence of tmHisZ. Numbering refers to the tmHisZ sequence. Secondary structure elements are shown on top, coloured according to tmHisZ domain organization, with the core domain in grey (residues 1–24, 51–132 and 201–275), the proximal subdomain in cyan (residues 25–50) and the distal subdomain in violet (residues 133–200). Residues strongly conserved (>90%) in multiple sequence alignments of the HisZ family and the HisRS family (data not shown) are in bold. Residues involved in the histidine-binding pockets 1 and 2 are marked with green asterisks (*) and green hash signs (♯) respectively (Fig. 7). Fingerprint motifs of the HisRS family (Aberg et al., 1997) are boxed (residue ranges according to the ttHisRS sequence): motif 1 (23–44); motif 2 (101–133); motif 3 (299–310); histidine-binding motif 1 (259–264); and histidine-binding motif 2 (285–290). The two key catalytic arginines are marked with a black asterisk each (*). Residues forming homodimerization interfaces (tmHisZ–tmHisZ) are underlined by red line segments, and those that are involved in heterodimerization interfaces (tmHisGStmHisZ) are underlined by blue line segments.

To provide evidence for this suggested link and to reveal the mechanism of regulation of the first committed step of histidine biosynthesis, we have determined the crystal structure of the hetero-octameric (HisGS)4HisZ4tmATP–PRTase complex from Thermotoga maritima at 2.5 Å resolution (Fig. 3). The structure reveals how two separate catalytic HisGS dimers are bound into a pair of large cavities that are formed by a tetrameric HisZ core complex with a total of eight histidine sites within four equivalent HisGS–HisZ interfaces. Its regulatory, non-enzymatic tmHisZ subunit shares a structural relationship to class II aminoacyl-tRNA synthetases while the catalytic HisGS subunit is unrelated to any other known PRTases except members of HisGL ATP–PRTases. Taking into account conserved and non-conserved structural features of the tmATP–PRTase complex, an evolutionary model for members of the (HisGS)4HisZ4-type ATP–PRTase family has been constructed that considers maintenance, loss and gain of several structural elements from a precursor enzyme from which extant enzymes participating both in amino acid biosynthesis and protein biosynthesis may have evolved.

Figure 3.

Structure of the hetero-oligomeric HisZ4[(HisGS)2]2tmATP–PRTase complex.
A. Ribbon drawings of the tmATP–PRTase complex with the bound histidine and phosphate ligands in space filling presentation.
B. Space filling presentations of the tmATP–PRTase complex.
The tmATP–PRTase complexes shown in (A) and (B) to the right have been rotated by 180° around a horizontal axis with respect to the same panels to the left. Colour codes: HisZ core domain, grey; HisZ proximal subdomain, cyan; HisZ distal subdomain, violet; HisGS domain I, yellow; HisGS domain II, ochre. The bound phosphate and histidine ligands are shown in CPK representation. Atom colour codes: carbon, green; nitrogen, blue; oxygen, red; phosphorus, magenta.
C. Schematic representation of a possible assembly process of the HisZ4[(HisGS)2]2tmATP–PRTase complex, as suggested from the crystal structure.

Results

Structure of the catalytic tmHisGS subunit

The tmHisGS subunit of the (HisGS)4HisZ4tmATP–PRTase complex folds into two domains (Fig. 4). Domain I comprises residues 1–84 and 170–208 consisting of a seven-stranded β-sheet (β1–β6, β10), which is wrapped by five α-helices (α1–α3, α6, α7). Domain II (residues 85–169) is topographically inserted into domain I. It comprises a five-stranded β-sheet (β7–β10) and two α-helices (α4, α5). The two domains are connected by a long, two-stranded β-sheet (β6, β10), and the domain I and II interface is formed by helices α3 and α4. In domain II, C117 (helix α4) and C124 (strand β9) are connected by a disulphide bridge. Part of the tmHisGS subunit comprises a twofold repeat structure, consisting of the N-terminal segment of domain I (residues 1–84) and domain II (residues 85–169). The two segments can be superimposed with a root mean square deviation (r.m.s.d.) of 2.4 Å using main-chain atoms, suggesting a common ancestry by gene duplication (Fig. 4E). In the present structure, the four tmHisGS subunits bind either one or two phosphate ions within their active sites (Fig. 4A and B). The phosphate ion that is present in all four HisGS subunits interacts with the conserved P-loop (residues 150–154), connecting strand β9 and helix α5 in domain II. Each tmHisGS dimer is formed by a twofold head-to-tail arrangement of domains I and II.

Figure 4.

Structure of the catalytic tmHisGS subunit.
A. Ribbon representation of the monomeric tmHisGS subunit. The bound phosphate ions and histidine ligands are shown in space filling presentation (colours as in Fig. 3). The phosphate ions are bound to the active sites of tmHisGS subunits, thus providing the active site locations with respect to the overall structure of each tmHisGS subunit. Histidine ligands are labelled according to the panel of Fig. 6 they correspond to, A or B. The disulphide bridge linking residues C117 and C124 is shown in yellow orange. The two domains I and II, the N- and C-termini of the tmHisGS polypeptide chain and the secondary structural elements are labelled.
B. Ribbon representation of the functional (tmHisGS)2 dimer.
C and D. Surface representation of the monomeric tmHisGS subunit in two different orientations, rotated by 180° around a vertical axis. The interface for the second tmHisGS subunit, generating the functional (tmHisGS)2 dimer, is shown in red. The interfaces with the different parts of the tmHisZ subunits are shown in the tmHisZ colours used in Fig. 3. The binding surfaces for the two active site phosphate ions and for the two histidine molecules are shown in magenta and green respectively.
E. Superposition of domain I and domain II of tmHisGS in ribbon presentation. The colours are as in (B), except for those parts of domain I that do not match with residues of the smaller domain II (in grey). The termini of the polypeptide segments are labelled and numbered.
F. Superposition of the α-carbon traces of mtHisGL (Cho et al., 2003) and tmHisGS. The traces are in red and yellow respectively. The bound histidine ligands are shown in the atom type colours used in Fig. 3. However, to discriminate the histidine bound to the C-terminal histidine binding domain of mtHisGL its carbon atoms are displayed in cyan. The termini of the two polypeptide chains are labelled. The active sites of tmHisGS and mtHisGL are basically identical and their common location is indicated by one of the two phosphate ions bound to the active site of each tmHisGS subunit (in atom type colours). The tmHisGS active site phosphate ion superimposes with the active site sulphate ion in mtHisGL (Cho et al., 2003).

Its molecular architecture is essentially identical to that of the catalytic domain of members of (HisGL)2 ATP–PRTases (Cho et al., 2003; Lohkamp et al., 2004) (Fig. 4F). One of the two phosphate binding sites of HisGS superimposes with the phosphate group of AMP bound in the structure of a member of the (HisGL)2 ATP–PRTase family (Cho et al., 2003), supporting a similar active site structure of the catalytic domains and subunits of members of the (HisGL)2 and (HisGS)4HisZ4 ATP–PRTase families respectively (data not shown).

Structure of the regulatory tmHisZ subunit

The tmHisZ subunit of the octameric tmATP–PRTase complex comprises an unusual, arc-like structure (Fig. 5). The core of the domain consists of an eight-stranded mostly anti-parallel β-sheet, except for β8 that is parallel to stand β7. The topology of the sheet is β1-β5-β6-β11-β10-β9-β7-β8. It is inserted between two peripheral subdomains (residues 25–50 and 133–200). The first subdomain, referred to as proximal subdomain (see below), is formed by a small anti-parallel three-stranded β-sheet (β2-β4-β3), which is followed by an α-helix that is arranged with part of the large cavity formed by the central β-sheet. The second one, referred to as distal subdomain, is formed by six α-helices (α4–α9).The two remaining α-helices (α1, α3) pack against the outer, convex surface of the central β-sheet. Together with helices α8 and α9, they form an outer wheel wrapping the central β-sheet.

Figure 5.

Structure of the (tmHisZ2)2 tetramer.
A. Ribbon representation of the monomeric tmHisZ subunit. The tmHisZ core domains and the two peripheral subdomains are coloured as in Fig. 3. The subdomains, the N- and C-termini of the tmHisZ polypeptide chain and the secondary structural elements are labelled.
B. Superposition of the α-traces of tmHisZ (in blue) and Thermus termophilus HisRS, ttHisRS, in complex with histidine (in red, PDB code 1adj) (Aberg et al., 1997). The superposition was calculated with align (Cohen, 1997) with an r.m.s.d. of 2.4 Å over main-chain atoms of 210 matching residues. The C-terminal anti-codon binding domain of ttHisRS has no equivalent in tmHisZ. The histidine ligands are shown in CPK, with green carbon atoms for tmHisZ ligands and cyan carbon atoms for ttHisRS.
C and D. Surface representation of the monomeric tmHisZ subunit in two different orientations, rotated by 180° around a vertical axis. The interface surface for the second tmHisZ subunit, generating the functional tmHisZ dimer, is shown in red. Interface regions with the second HisZ dimer and with HisGS subunits are shown in the colours of the respective domains or subdomains of these subunits as chosen in Fig. 3. The histidine binding pockets are in green.
E. Ribbon representation of the (tmHisZ2)2 tetramer (same view as in Fig. 3A). The interface of each dimer is formed by intermolecular β-sheet interactions between the two proximal subdomains of each tmHisZ subunit. The locations of the two large cavities for each (tmHisGS)2 are indicated by arrows. The two tmHisZ homodimers are rotated 108° around a vertical axis (marked by two black segments).
F. Surface representation of the (tmHisZ2)2 tetramer. Colour coding of the interfaces is as in (C) and (D). tmHisZ subunits are shown in either light grey or dark grey to highlight the boundaries between the homodimer and homotetramer interfaces. To improve the visualization of the HisGS binding sites within the (tmHisZ2)2 tetramer shown in (F) has been rotated by about 40° around a vertical axis with respect to its orientation in (E).

The tmHisZ2 dimer is assembled by interactions of the central β-sheet, the small β-sheet of the proximal subdomain, loops connecting the two β-sheets, and helix α1 of each tmHisZ subunit (Fig. 5A). While strands β5 and β6 of the central β-sheet and two extended loops form a symmetric two-layered structure with the same residue segments of the other protomer, the peripheral sheet of the proximal subdomain is connected via regular β-sheet hydrogen bond interactions connecting N(L41)-O(L41′) and O(L41)-N(L41′), thus forming an intermolecular β-sheet with β2-β4-β3-β3′-β4′-β2′ topology. At the opposite end of the HisZ2 dimer interface, there are extensive stacking interactions from aromatic residues of helix α1 and structurally adjacent sequence segments. The overall shape of tmHisZ2 dimer is that of a double arc in which the ‘roof’ is formed by an extensive layer of α-helices α9-α8-α3-α1-α1′-α3′-α8′-α9′ from the two tmHisZ subunits.

Assembly of the histidine-liganded octameric (tmHisGS)4tmHisZ4 complex

The structure of tmATP–PRTase constitutes the first structural representative of (HisGS)4HisZ4-type ATP–PRTases. Its overall dimensions are in the order of 110 Å × 100 Å × 90 Å (Fig. 3). The core of the complex is assembled by an X-shaped tmHisZ4 dimer of dimers. Each tmHisZ dimer is connected via a two-layered structure that is generated by the intermolecular β-sheet (β2-β4-β3-β3′-β4′-β2′) (Fig. 5A and E). Compared with the extensive tmHisZ2 dimer interface (overall area, 1370 Å2), the tetrameric tmHisZ dimer–tmHisZ dimer interface is small (overall area, 480 Å2). The observations are supported by gel filtration and dynamic light scattering experiments, indicating a concentration-dependent equilibrium of dimers and tetramers for tmHisZ (data not shown).

In the complete octameric (HisGS)4HisZ4tmATP–PRTase complex, two separate head-to-tail tmHisG2 dimers (Fig. 3) are accommodated by the two large cavities that are formed by the tmHisZ4 tetramer (Fig. 5E and F). Domain I of each tmHisGS subunit interacts with the distal α-helical subdomain of a tmHisZ subunit belonging to one tmHisZ dimer, while domain II mostly binds to the proximal β-sheet of a tmHisZ subunit belonging to the other tmHisZ dimer. In each tmHisGS dimer unit, one set of interactions of domains I and II from two different tmHisGS subunits are with the same tmHisZ subunit. Thus, the pattern of domain–domain interactions indicates that the stable (tmHisGS)4tmHisZ4 octameric complex is constituted by tightening the apparently weak dimer–dimer interface of the tmHisZ4 tetramer by two sets of tmHisGS dimer-mediated interactions between the two sets of tmHisZ2 dimers (Figs 3 and 5). Steady-state kinetics measuring the ATP–PRTase catalytic activity of the (tmHisGS)4tmHisZ4 complex confirm that only the complete complex is active (kcat/KmPRPP = 6.1 min−1 mM−1, at pH 8.0 and 20°C), while separate tmHisGS dimers do not display measurable catalytic activity.

Furthermore, histidine acts as a non-competitive inhibitor of tmATP–PRTase (KI = 0.35 ± 0.02 mM), suggesting that there is a histidine binding site that is different from the tmHisGS active site. These observations are matched by the presence of two histidine molecules within each of the four tmHisGStmHisZ interfaces of the octameric tmATP–PRTase complex (Figs 3 and 6). Each tmHisZ subunit binds the two histidines via helices α5, α6 and α7 of its distal subdomain. In turn, each tmHisGS subunit interacts with one of the two histidines by residues from the C-terminal helices α6 and α7 and the loop connecting helix α3 and strand β6 in the larger domain I. The second histidine is bound to the face of the tmHisGS hinge region, connecting domains I and II, that is opposite to the active site of the tmHisGS subunit. At the atomic level, both histidines are bound by several specific polar interactions involving its side-chain imidazole ring as well as the charged carboxylate and amino groups (Fig. 6). In particular, the location of the second histidine molecule in each tmHisGStmHisZ interface in close proximity to the catalytic centre suggests a model whereby the catalytic activity of the (HisGS)4HisZ4tmATP–PRTase is regulated by conformational changes that may be triggered by the absence/presence of histidine.

Figure 6.

Binding of histidine to the tmATP–PRTase complex.
A and B. The two histidine binding sites, which are displayed in (A) and (B), respectively, are located in the interface between tmHisG domain I and the tmHisZ distal subdomain. The histidine ligand and residues of the two histidine-binding pockets are shown in a ball-and-stick representation, using the colour codes of previous figures. The histidine ligands are modelled in a σA-weighted 2Fo-Fc electron density map contoured at 1.0 σ. Specific interactions are shown in dashed lines, and the distances between the involved atoms are in Å.
C. Native polyacrylamide gel electrophoresis of the octameric tmATP–PRTase in the absence of effectors (lane 3), after incubation with 14 mM ATP (lane 4), 14 mM AMP (lane 5), 20 mM histidine (lane 6), or 20 mM histidine and 14 mM AMP (lane 7). Control lanes contain tmHisG only (lane 1) or tmHisZ only (lane 2). No detectable disassembly of the complex is observed.

Discussion

(HisGL)2 and (HisGS)4HisZ4-type ATP–PRTases are regulated by different types of histidine feedback inhibition

The catalytic activities of the members of both ATP–PRTase families are tightly regulated by several molecules that mimic compounds involved in the ATP–PRTase reaction and by the pathway end-product histidine (Kleeman and Parsons, 1976; Morton and Parsons, 1977a,b). Our structure of the (HisGS)4HisZ4tmATP–PRTase complex confirms previous predictions (Sissler et al., 1999) that the mechanism of activity regulation by histidine, however, may be unrelated in the two ATP–PRTase families. Only the sequences of (HisGL)2-type ATP–PRTases contain a C-terminal domain that binds histidine. Structures of two representatives in the absence/presence of histidine revealed that histidine binding leads to a conformational change within the additional C-terminal domain that results in the stabilization of a postulated catalytically inactive hexamer state with respect to the catalytically active dimeric state (Cho et al., 2003; Lohkamp et al., 2004). In contrast, in the (HisGS)4HisZ4tmATP–PRTase complex, histidine binds to two sites within each of the four tmHisGStmHisZ interfaces, which are formed by surface patches of the catalytic core domains I and II of each catalytic tmHisG subunit and the distal α-helical subdomain of each tmHisZ subunit (Figs 3 and 6). A structural comparison of the HisGS subunit from the tmATP–PRTase complex indicates a closer relation with the inhibited form of the catalytic domain of the (HisGL)2-type ATP–PRTase from Mycobacterium tuberculosis (r.m.s.d. 1.38 Å, PDB entry 1nh8) than with its active conformation (r.m.s.d. 1.57 Å, PDB entry 1nh7), suggesting that the crystal structure of the (HisGS)4HisZ4tmATP–PRTase is in an inhibited conformation. However, the interpretation remains tentative in the absence of another tmATP–PRTase structure in the active conformation.

Our structural data match the previous model of the Lactococcus lactis ATP–PRTase complex, which suggested that the basic structural units of both HisGS and HisZ are dimers (Bovee et al., 2002). However, the structure of the tmATP–PRTase comprises a HisZ4 tetramer core with two separate (HisGS)2 dimers (Fig. 3) rather than a complex with a mixed HisGS/HisZ tetramer core as was suggested for L. lactis ATP–PRTase (Bovee et al., 2002). In the (HisGS)4HisZ4-type ATP–PRTase complex from L. lactis, analytical ultracentrifugation data revealed small changes in the sedimentation coefficient in the presence of ATP, AMP and histidine, suggesting conformational changes and, additionally or alternatively, changes in subunit composition (Bovee et al., 2002). However, for the hetero-octameric tmATP–PRTase complex, we were not able to detect significant changes of its hydrodynamic radius upon the addition of different effector ligands, using dynamic light scattering (data not shown).

Although our data, along with previous data, show unambiguous evidence for a common ancestry of the catalytic domains and subunits in (HisGL)2 and (HisGS)4HisZ4-type ATP–PRTases respectively (Fig. 4F), no overall similarity to other PRTases can be detected. Like a few other unrelated PRTases, for example, quinolate PRT (Eads et al., 1997) and anthranilate PRT (Mayans et al., 2002), the catalytic HisGS subunit appears to have evolved independently, in contrast to the large number of known class I PRTases that have evolved from a common ancestor (Craig and Eakin, 2000; Sinha and Smith 2001). It is particularly remarkable that the two PRTases from the tryptophan and histidine biosynthesis pathways are unrelated, in light of evidence by sequence, structure and function for a common ancestry of several enzyme types found in both pathways (Lang et al., 2000; Hocker et al., 2001; Mayans et al., 2002; Vega et al., 2003).

The regulatory HisZ subunit comprises a tRNA synthetase-like fold

A search for similar structures using the tmHisZ co-ordinates as a template reveals structural relations to class II aminoacyl-tRNA synthetases (aaRS). The closest similarity is observed for available Histidyl-tRNA synthetase (HisRS) structures that use the histidine biosynthesis product as a cognate amino acid in aminoacylation (Ames et al., 1961) (Fig. 7). Most of the secondary structural elements of tmHisZ, except helices α6, α7 and α8 of the distal subdomain, match with those in these HisRS structure. On the other hand, in HisRS, there is no central large cavity as in the arc-like structure of tmHisZ (Fig. 5A). Instead, the region near helix α2 is filled by several additional secondary structural elements that provide each HisRS subunit with a more globular shape than tmHisZ. In addition, each HisRS subunit comprises a C-terminal anti-codon binding domain that is not present in tmHisZ. It is also noted that the dimeric arrangement of HisRS is unrelated to that found in the dimeric HisZ2 unit. Furthermore, although both HisRS and tmHisZ bind histidine, the structure of the tmATP–PRTase complex does not confirm previous predictions of a common histidine binding site in HisRS and (HisGS)4HisZ4 ATP–PRTases (Sissler et al., 1999).

Figure 7.

Structure-derived model of (HisGS)4HisZ4-type ATP–PRTase evolution. Evolutionary relationships of HisRS, asparagine synthetase A (AS-A), HisZ, HisGL and HisGS. The loss, gain or modification of functional sites or domains are highlighted in green, red and blue, respectively, while the remaining parts are shown in grey. The suggested common ancestor to HisRS, HisZ and AS-A is shown as a functional class II aaRS precursor. In the evolution of AS-A, the precursor active site (AS) architecture is modified (blue), while the active site is lost during the evolution of HisZ and new histidine binding sites (BS) have emerged (red). In the evolution of both proteins, the anti-codon binding (ACB) domain of the class II aaRS precursor (green) is lost. The suggested common ancestor to HisGL and HisGS is a HisGS-like protein. BD, binding domain.

The unambiguous similarities of HisRS and HisZ in terms of sequence and structure (Figs 2B and 5B) suggest an evolutionary link between the biosynthesis of a specific amino acid (histidine) and its use in protein biosynthesis by its cognate aminoacyl-tRNA synthetase (HisRS) (Fig. 7). Remarkably, in our model of HisZ evolution, the catalytic function of HisRS coupled with its capability to bind histidine within the active site has been lost while new histidine binding sites evolved in HisZ. Although the precise conformational changes associated with the regulation of (HisGS)4HisZ4-type ATP–PRTase activity have not yet been determined, the ‘reinvented’ histidine binding sites in the tmHisGStmHisZ interface, as observed in the tmATP–PRTase complex, seem to be more suited to regulate (HisGS)4HisZ4-type ATP–PRTase activity by providing direct contacts to surface patches close to the active sites of the HisGS subunit, rather than the buried histidine binding site of the HisRS active site.

Assuming that aminoacyl-tRNA synthetases have been among the earliest proteins to emerge, possibly descending from ribozymes (Wong, 1975; Schimmel and Ribas De Pouplana, 2000), it appears plausible to assume that (HisGS)4HisZ4-type ATP–PRTase catalytic activity has evolved subsequently. In this scenario, its regulatory HisZ subunit would have evolved from a HisRS precursor enzyme while its HisGS subunit would have emerged independently from a yet unidentified precursor. A number of other enzymes, involved in processes such as translation regulation, stimulation of DNA polymerase activity and indirect pathways of amino acid biosynthesis (Schimmel and Ribas De Pouplana, 2000; Francklyn, 2003; Roy et al., 2003), have also been shown to be evolutionarily related to aaRS. Generally, these tRNA synthetase (RS)-like proteins share the loss of the N-terminal tRNA anti-codon binding domain. The best documented case is on the common ancestry of asparagine synthetase A (AS-A) and AsnRS (Nakatsu et al., 1998; Roy et al., 2003), in which most of the active site architecture is maintained, thus establishing a relationship that is closer than that between HisZ and HisRS. Although a distant relationship between HisZ and AS-A can be detected by the dali software when searching against the Protein Data Bank (z-score, 9.4; rank, 10), the closest similarity for both enzymes is with class II aminoacyl-tRNA synthetases, supporting an evolutionary model of aaRS being the precursor of both AS-A and HisZ.

In summary, the relationship of HisRS and the regulatory HisZ subunit of (HisGS)4HisZ4-type ATP–PRTases provides the first direct structural evidence for an evolutionary relationship of an extant aaRS and the regulatory component of an enzyme that is involved in the biosynthesis of its own cognate amino acid. At present, it remains, however, speculative whether there may have been a functional interaction between a tRNA synthetase precursor and an ATP–PRTase precursor. Our data support an earlier hypothesis that the histidine biosynthesis pathway may have emerged from an ancient tRNA-dependent version of its biosynthesis (Francklyn, 2003), ultimately linking the extant protein world with its emergence from the RNA world.

Experimental procedures

Cloning

The full-length hisG and hisZ genes were cloned from T. maritima genomic DNA using the following polymerase chain reaction (PCR) primers: 5′-GGGAAACATATGCTGA AACTGG-3′ (hisG 5′  primer); 5′-sGAGCAAGCTTTTATCC CCGG-3′ (hisG 3′ primer); 5′-CTGGGGCCATGGATTTCT TGG-3′ (hisZ 5′ primer); and 5′-CCTTTGGGAAT TCCCAGTTTCAG-3′ (hisZ 3′ primer). Expression plasmids were constructed by digesting the hisG PCR fragment (643 bp) with NdeI and HindIII and inserting the product into pET21, which harbours an ampicillin resistance cassette. Likewise, the hisZ PCR fragment (841 bp) was digested with NcoI and EcoRI and ligated into pET28, which contains a kanamycin resistance cassette.

Protein expression and purification

For expression of the native genes, E. coli BL21(DE3) RIL cells were transformed with the hisGS or hisZ expression constructs, and plated out on Luria–Bertani (LB) plates containing 50 µg ml−1 ampicillin or 25 µg ml−1 kanamycin, respectively, at 37°C. Before the final volume expression experiments, 25–50 ml of precultures were grown from a single colony overnight in LB at 37°C. A 10 ml aliquot of this preculture was used as an inoculum to 1 l of expression culture in a 5 l flask. For native expression, LB medium supplied with the appropriate antibiotics was used as growth medium. Isopropyl-β- d-thiogalactopyranoside (IPTG) (1 mM) was added to start induction at an OD600 of ≈0.6–0.7 and the cultures were harvested after another 4–6 h. The typical yield for tmHisG and tmHisZ was ≈15–25 mg of purified protein per litre of culture.

For production of the seleno- l-methionine (SeMet) incorporated protein, methionine auxotroph E. coli strain B834(DE3) cells were transformed with the hisGS and hisZ expression constructs, and plated out as described above. The following media and stock solutions were used: M9 minimal medium, 10× stock (0.55 M Na2HPO4, 0.30 M KH2PO4, 85 mM NaCl, 60 mM NH4Cl), medium A (1× M9 supplied with 20 mM d-(+)-glucose, 1 mM MgCl2, 0.3 mM CaCl2, 4 µM biotin, 2.7 µM thiamine), starvation medium (medium A plus antibiotics, without methionine), trace element solution, 100× stock (30 mM FeCl3·6H2O, 6.0 mM ZnCl2, 0.8 mM CuCl2·4H2O, 0.4 mM CoCl2·6H2O, 1.6 mM H3BO3, 0.07 mM MnCl2·6H2O), 50 mg ml−1l-methionine (1000× stock) and 50 mg ml−1 SeMet (1000× stock). A single colony of freshly transformed cells was used to inoculate a 25–50 ml preculture in medium A plus methionine, which was grown overnight at 37°C until saturation (OD600 ≈ 2–3). Next, the preculture was added to 1 l of medium A with methionine pre-incubated at 37°C and grown until OD600 of 1.0. The cell suspension was centrifuged twice at 4000 r.p.m. in a Sorvall SLA-3000 rotor and washed in 1× PBS. The cell pellet was then resuspended in 1 l of starvation medium and incubated at 37°C for 4–6 h. Thirty minutes before induction, 1 ml of SeMet was added. The cell culture was induced by the addition of 1 mM IPTG and expression proceeded for 6 h before harvesting. The typical yield for SeMet tmHisG and tmHisZ was ≈5–10 mg of purified protein per litre of culture.

Both the native and the SeMet proteins were purified in an identical way, with the exception of the addition of 1 or 5 mM DTT to all buffers respectively. The expression cultures were centrifuged at 4500 r.p.m. in a Sorvall SLA-3000 rotor, and the cell pellets washed with 1× PBS. In order to obtain a 1:1 protein complex, the wet cell pellets of tmHisGS and tmHisZ were mixed in a ≈1.5:1 (w/w) ratio. The combined cell pellets were then thoroughly resuspended in 25 ml of stabilizing buffer per litre of culture for heat shock treatment at 80°C for 10 min. The composition of the stabilizing buffer was as follows: 50 mM Tris-Cl, pH 8.0, containing 0.4 mM histidine, 4 mM K3PO4, 100 mM KCl, 5 mM DTT. The lysate was then clarified by centrifugation at 23 000 r.p.m. in a Sorvall SS-34 rotor and filtered through a 0.22 µm membrane. The binary tmHisGStmHisZ complex was further purified in 20 mM Tris/HCl, pH 7.5, by anion exchange chromatography on a Pharmacia HiLoad Q-Sepharose (HR 10/26), eluting in a NaCl linear gradient at about 300 mM NaCl. The elution fractions were pooled, concentrated and dialysed against a suitable buffer for gel filtration (20 mM Tris-Cl, pH 7.5, 40 mM NaCl, 0.4 mM histidine, 5 mM DTT, 0.6 mM AMP). Gel filtration was carried out on a Pharmacia Superdex 200 (HR 26/60) column, and the complex eluted with an estimated molecular mass of ≈200 kDa, corresponding to the (tmHisGS)4–(tmHisZ)4 octameric complex. The elution fractions matching the peak were pooled and concentrated up to 50 mg ml−1, and 0.6 mM AMP was added. Part of the purified ATP–PRTase complex was flash-frozen in liquid nitrogen and stored at −80°C. Protein samples for biochemical assays were produced by the same procedures but without histidine and AMP.

Enzyme kinetics

Steady-state kinetics were performed to determine the kinetic parameters of the tmATP–PRTase complex and the inhibition constant of the pathway product histidine. The steady-state assay was based on the increase of A290 after the formation of the product PR-ATP (Voll et al., 1967). Reactions contained 1.2 µM of enzyme (molarity calculated for a tmHisG–tmHisZ dimer) in 50 mM Tris (pH 8.0), 10 mM MgCl2, 150 mM KCl, inorganic pyrophosphatase at 2 µg ml−1, and saturating concentrations (5 mM) ATP, in a final volume of 120 µl. PRPP was added last (from 0.5 to 8.7 mM final concentration) to start the reaction, which was monitored over 10 min every 5 s in a UVIKON 922 spectrophotometer (Kontron) at 20°C. To measure the inhibitory effect of histidine, reactions were set up as described and histidine was added to a final concentration of 0.1, 1, 5 or 10 mM. These samples were incubated with histidine during 1 h before adding PRPP (Bell et al., 1974). The kcat and KmPRPPvalues were determined from saturation curves by non-linear least squares regression. KIHis was determined from the relation v0app = Vmapp[S]/(Kmapp+[S]), where Vmapp = k3[E0]/(1 + [I]/KI) and Kmapp = Km.

X-ray structure determination

Crystals of the binary complex were grown by assembling 2 µl of 8 mg ml−1tmATP–PRTase with 2 µl of reservoir solution containing 22.5% (w/v) methylpentanediol (MPD), 0.2 M phosphate/citrate buffer, pH 4.2, using the sitting drop vapour diffusion technique. Both the native and the SeMet crystals grew under identical crystallization conditions. Crystals suitable for diffraction appeared after 2 days and reached a final size of 1.0 × 0.45 × 0.35 mm. A stepwise increase of MPD concentration (up to 25% in 1% steps) was used to cryoprotect the crystals, which were flash-frozen on a liquid nitrogen stream.

Several X-ray data sets were collected on crystals of the native and SeMet tmATP–PRTase complexes (Table 1). A three-wavelength multiple anomalous dispersion data set was collected in two 0.2° steps and two 0.4° oscillation angle wedges per wavelength (over a total angular range of 158°) to avoid spot overlaps according to the data collection strategy program best (Popov and Bourenkov, 2003). The data were merged during data processing and scaled with denzo and scalepack (Otwinowski and Minor, 1997). We were able to identify 23 out of 28 selenium atoms with the software solve v2.03 (Terwilliger and Berendzen, 1999), using the single anomalous diffraction mode for the peak wavelength (0.9788 Å). The solution was refined against the inflection and remote data sets found. The final list of sites included the three out of the four possible selenium atoms at the N-terminus of each tmHisZ. The five missing sites corresponded to the fourth tmHisZ N-terminus (M1) and all four tmHisGS N-termini that are disordered. The figure of merit (fom) was 0.35 in the resolution range of 30–3.5 Å. Density improvement by density modification and non-crystallographic symmetry (NCS) averaging with resolve v2.03 substantially improved the quality of the electron density maps, and raised the fom to 0.52, using X-ray data from 30 to 2.6 Å. A correct, but incomplete, ATP–PRTase model was built automatically using the warpNtrace mode of ARP/wARP (Perrakis et al., 1999) with experimental phases up to 2.6 Å. The free R factor decreased from 43% to 40% and the work R factor from 38% to 25%, and the fom rose from 0.52 to 0.71, coupled with a significant increase in the quality of the electron density. The wARP model had a final (initial) connectivity index of 0.89 (0.79) and the total number of amino acids built was 1484 out of 1932 (77%), arranged in 27 fragments. The remaining model could be built manually, succeeded by one further round of ARP/wARP in molrep mode (Perrakis et al. 1999). The free R factor, work R factor and fom of the subsequent model were 34%, 22% and 0.76 respectively. The model was refined with cns v1.1 (Brunger et al. 1998) until convergence to a free R factor of 28.6% and a work R factor of 20.3% was achieved. The final model includes nearly all residues (1911 out of 1932). Its co-ordinates and the structure factors have been deposited in the Protein Data Bank data bank (1USY). Further statistics are summarized in Table 2.

Table 1.  Crystallographic data and phasing statistics.
 NativePeakInflectionRemote
  • a,b

    . The figure of merit (fom) and mean phase error (mpe) are given for the resolution shells of 30.0–3.5a and 30.0–2.6b.

  • The native and the three-wavelength multiple anomalous dispersion data sets were collected at the synchrotron radiation beamlines BW6 (MPG-ASMB/DESY, Hamburg) and BW7A (EMBL/DESY, Hamburg) respectively. Values in parentheses pertain to the last resolution shell.

BeamlineBW7ABW6  
Wavelength (Å)0.97460.97880.97900.9000
Space groupP212121P212121  
Cell (Å)
 a100.7101.6  
 b133.3134.4  
 c152.5154.2  
Resolution (Å)30.0–2.5030.0–2.8030.0–2.8030.0–2.60
(2.60–2.50)(2.85–2.80)(2.85–2.80)(2.64–2.60)
No. of reflections
 Measured818 378930 472930 8291 133 033
 Unique62 20452 74352 83365 941
Multiplicity13.1517.6417.6117.18
Completeness (%)86.9 (88.7)98.4 (98.1)98.2 (95.3)95.3 (93.3)
〈I〉/σ(〈I〉)13.8 (5.1)18.2 (6.9)18.4 (6.4)16.0 (3.6)
R-sym (%)3.9 (21.3)5.5 (28.9)5.4 (33.0)6.0 (54.8)
R-anom (%) 4.4 (15.1)3.0 (16.6)3.8 (27.3)
fom (mpe, °) 0.34 (73.8)a0.40 (73.8)b0.52 (65.8)b
Table 2.  Refinement statistics of the tmATP–PRTase structure.
Resolution (Å)20.0–2.52
No. of reflections66 598
No. of reflections for validation3547 (5.1%)
R free (%)28.6
R work (%)20.3
No. of non-H atoms
 Protein15 244
 Ligands118
 Solvent274
Bond length r.m.s.d. (Å)0.016
Bond angle r.m.s.d. (°)1.80
Mean B-value (Å2)
 Protein32.5
 Ligands63.1
 Solvent28.7

Acknowledgements

The authors thank Gleb P. Bourenkov and Andrea Schmidt for their support during data collection and Frank Lehmann for technical assistance during protein production. The project has been supported by a grant by the Deutsche Forschungsgemeinschaft to M.W. (Wi 1058/5-3). We thank Areti Malapetsas for editing the manuscript.

Ancillary