The three-dimensional structure of the nudix enzyme diadenosine tetraphosphate hydrolase from Lupinus angustifolius L1

The three-dimensional structure of the nudix enzyme diadenosine tetraphosphate hydrolase from Lupinus angustifolius L1

doi:10.1006/jmbi.2000.4085 available online at http://www.idealibrary.com on J. Mol. Biol. (2000) 302, 1165±1177 The Three-dimensional Structure of ...

653KB Sizes 0 Downloads 47 Views

doi:10.1006/jmbi.2000.4085 available online at http://www.idealibrary.com on

J. Mol. Biol. (2000) 302, 1165±1177

The Three-dimensional Structure of the Nudix Enzyme Diadenosine Tetraphosphate Hydrolase from Lupinus angustifolius L James D. Swarbrick1, Tanya Bashtannyk1, Danuta Maksel1 Xiu-Rong Zhang2, G. Michael Blackburn2, Kenwyn R. Gayler1 and Paul R. Gooley1* 1

Department of Biochemistry and Molecular Biology University of Melbourne Parkville, Victoria 3010, Australia 2

Department of Chemistry University of Shef®eld Shef®eld, S3 7HF, England

The solution structure of diadenosine 50 , 5000 -P1, P4-tetraphosphate hydrolase from Lupinus angustifolius L., an enzyme of the Nudix family, has been determined by heteronuclear NMR, using a torsion angle dynamics/simulated annealing protocol based on approximately 12 interresidue NOEs per residue. The structure represents the ®rst Ap4A hydrolase to be determined, and sequence homology suggests that other members will have the same fold. The family of structures shows a wellde®ned fold comprised of a central four-stranded mixed b-sheet, a twostranded antiparallel b-sheet and three helices (aI, aIII, aIV). The rootmean-squared deviation for the backbone (C0 , O, N, Ca) of the rigid parts Ê . Several (residues 9 to 75, 97 to 115, 125 to 160) of the protein is 0.32 A regions, however, show lower de®nition, particularly an isolated helix (aII) that connects two strands of the central sheet. This poor de®nition is mainly due to a lack of long-range NOEs between aII and other parts of the protein. Mapping conserved residues outside of the Nudix signature and those sensitive to an Ap4A analogue suggests that the adenosineribose moiety of the substrate binds into a large cleft above the fourstranded b-sheet. Four conserved glutamate residues (Glu55, Glu58, Glu59 and Glu125) form a cluster that most likely ligates an essential magnesium ion, however, Gly41 also an expected magnesium ligand, is distant from this cluster. # 2000 Academic Press

*Corresponding author

Keywords: hydrolase; pyrophosphatase; Nudix; solution structure; Ap4A

Introduction Diadenosine 50 , 5000 -P1, P4-tetraphosphate (Ap4A) hydrolase from Lupinus angustifolius L., is an 18.5 kDa member of a subfamily of ApnA hydrolases (Dunn et al., 1999), which belong in turn to the Nudix (nucleoside diphosphate linked to x) enzyme family (Bessman et al., 1996). Ap4A hydrolases have been isolated and characterized from a variety of eukaryote and prokaryote organisms and tissues. Such hydrolases cleave Ap4A asymmeAbbreviations used: Ap4A, diadenosine 50 , 5000 -P1, P4tetraphosphate; DSS, 3-(trimethylsilyl)-1-propanesulfonic acid; HSQC, heteronuclear single quantum coherence; NOE, nuclear Overhauser effect; rmsd, root-meansquare deviation; TOCSY, total correlated spectroscopy. E-mail address of the corresponding author: [email protected] 0022-2836/00/051165±13 $35.00/0

trically into ATP ‡ AMP, and are thus distinguished from the unrelated and symmetrically cleaving Escherichia coli apaH gene product (Guranowski et al., 1983). Many functions have been suggested for both the enzyme and Ap4A in cells (McLennan, 1999). Ap4A is a potential by-product of aminoacyl tRNA synthesis, and accumulation of Ap4A in cells has been implicated in a range of biological events, including DNA replication, cellular differentiation, heat shock, metabolic stress and even apoptosis (McLennan, 1999). The importance of Ap4A hydrolases is highlighted by their recently established role in the invasive phenotype of pathogenic bacteria. The gene IalA from Bartonella bacilliformis encodes an asymmetric Ap4A hydrolase (Conyers & Bessman, 1999; Cartwright et al., 1999). Transformation of E. coli both with this gene and with the IalB gene resulted in an invasive phenotype on otherwise minimally invasive # 2000 Academic Press

1166 strains, and suggested that asymmetric Ap4A hydrolase played an important role in invasion by B. bacilliformis (Mitchell & Minnick, 1995). Homology searches suggest that similar Ap4A hydrolases may be a common feature of many invasive bacteria and a potential target for inhibition of their invasion (Cartwright et al., 1999). The threedimensional structure of the enzyme and complexes with its substrate or analogues are sought as a prerequisite for design of appropriate inhibi-

Structure of Ap4A Hydrolase

tors; however, no structure for the IalA enzyme exists. Ap4A hydrolase from L. angustifolius, with 35 % sequence identity (McLennan, 1999) is one of the sequences most closely related to IalA (Cartwright et al., 1999) (Figure 1); by comparison, the human analogue has only 18 % identity (McLennan, 1999). Searches of data bases for other Nudix proteins, uncover a family of approximately 300 members, each with a highly conserved Nudix signature: GX5EX7REUXEEGU, where U is a

Figure 1. Alignment of Ap4A hydrolase from L. angustifolius with the ten best matches from a BLAST search. Sequences and accession numbers are L. angustifolius (lupin) (O04841), Arabidopsis thaliana (mouse-ear cress) (Q9SQZ2), Hordeum vulgare (barley) (O24005), Zymomonas mobilis (Q9RH11), Rickettsia prowazekii (Q9ZDT9), B. bacillifomis (P35640), Campylobacter jejuni (CAB75217), Neisseria meningitidis (AAF42031), Helicobacter pylori (Q9ZJZ8), Pseudomonas aeruginosa (Q9X4P2) and Haemophilus in¯uenzae (Q57045). The location of secondary structure elements, bstrands (A-F) and helices (I-IV), determined for Ap4A hydrolase from lupin are indicated and fully described in the text. Conserved residues are indicated, and residues within the Nudix signature are shaded. Notably, Tyr77 of lupin is conserved in all Ap4A hydrolases that are known (Dunn et al., 1999).

1167

Structure of Ap4A Hydrolase

hydrophobic residue (Dunn et al., 1999). The only known structure of a member of this family to have been solved is the solution structure of the MutT enzyme, an 8-oxo-guanosine hydrolase from E. coli, in both the free and inhibitor-bound states (Frick et al., 1995; Lin et al., 1996; Lin et al., 1997). The fold of the protein includes a central ®vestranded b-sheet and two helices, of which one is a catalytic helix that contains part of the Nudix signature. Site-directed mutagenesis, metal titrations and kinetic analysis (Lin et al., 1997; Harris et al., 2000) show that within the Nudix motif are residues essential for catalysis and for the ligation of an enzyme-bound divalent cation. Site-directed mutagenesis of other Nudix hydrolases supports the essential role of these residues (Safrany et al., 1998; Dunckley & Parker, 1999; Yang et al., 1999). Recently, we reported the 1H, 13C and 15N NMR resonance assignment and secondary structure characterization of the Ap4A hydrolase from lupin (Swarbrick et al., 2000). These data suggested a structure similar to that of MutT (Lin et al., 1997); however, some differences were noted, including two additional helices and possibly additional bstrands. Here, we describe the 3D solution structure of the metal-free Ap4A hydrolase from lupin. Further, we describe changes to the 1H, 15N spectra of the complex of magnesium and Ap4A hydrolase on the addition of a diadenosine tetraphosphate analogue, P1,P4-dithio-P2,P3-monochloromethylene diadenosine 50 ,5000 -P1,P4-tetraphosphate (Chan et al., 1997).

Results and Discussion Structure determination Using standard triple resonance procedures the H, 13C and 15N spectra of lupin Ap4A hydrolase have been assigned (97.5 % of all 1H, excluding labile side-chain 1H). Ap4A hydrolase contains a large number of aromatic residues; seven Trp, eight Phe, four Tyr and one His. As many of the aromatics are clustered together in the structure, assignment could not depend solely on NOE data to connect the resonances of the aromatic rings to their CaH-CbH2 groups. For example, two pairs of tryptophan residues are near each other, Trp129 is near Trp131, and Trp91 is near Trp95, complicating the assignment of these residues. To accomplish this, we used 2D (Hb)Cb(CgCd)Hd and (Hb)Cb(CgCdCe)He, 2D 1H,1H TOCSY and 2D 1H TOCSY-relayed ct-[13C,1H]-COSY. The ®rst two experiments established through-bond connectivities from the rings to their CbH2 atoms, while the remaining experiments completed the ring connectivities. With these experiments, 95.6 % of all aromatic 1H have been assigned. A total of 4114 unique NOE peaks have been assigned in the 3D 13 C and 15N-edited NOESY spectra, of which 1999 are meaningful interresidue peaks (Table 1). Although a large number of intraresidue NOEs have been assigned, only a minimal number have 1

been included in the structure calculations (Table 1). These NOEs are between protons more than ®ve bonds apart. The structure of Ap4A hydrolase was calculated in an iterative fashion, starting with approximately 500 interresidue NOEs. Initial calculations employed the torsion angle dynamics program DYANA (GuÈntert et al., 1997) using NOE and experimental angle constraint data. Additional unambiguous NOEs were added on the basis of agreement with trial structures using the ASNO functionality within DYANA (GuÈntert et al., 1993). At a later stage in the calculations when the NOE constraint list was approaching completion, we used CNS (BruÈnger et al., 1998), including a cartesian stage of re®nement, and supplementing the experimental data with 3JHNHa, 13Ca/13Cb chemical shift and 1DNH residual dipolar coupling restraints. The inclusion of the dipolar coupling restraints did not improve the precision of the structure, instead they highlighted several ambiguous and erroneous NOEs, and thus assisted in improving the accuracy of the structure. In our last few cycles of calculations, 40 hydrogen bonds were added. The inclusion of these bonds was assessed from NH exchange data and their presence in more than eight of 25 accepted models calculated without these bonds. The ®nal structure calculation used 2006 NOE distance, 40  2 hydrogen bond, 219 dihedral angle, 280 13Ca/13Cb, 93 dipolar coupling, 94 3JHNa coupling restraints (Table 1). Of the ®nal 100 structures calculated, 25 structures with the lowest energy, and no experimental distance vioÊ and experimental dihelated by more than 0.4 A dral angle violated by more than 4  , were selected to represent the family of solution structures of Ap4A hydrolase. Structural statistics for the ®nal ensemble are given in Table 1 and the structural data are summarized in Figure 2. The calculation shows the convergence to a well-de®ned structure Ê for the most rigid with a backbone rmsd of 0.32 A regions (Table 1). Although we have assigned more than 12 interresidue NOEs per residue, the NOE data are not evenly distributed, thus several regions, including a helix, show markedly poor de®nition. Description of the solution structure The fold of Ap4A hydrolase comprises a curved four-stranded mixed central b-sheet, a twostranded antiparallel b-sheet and four, predominantly a, helices, forming an a-b-a sandwich (Figure 3). Residue numbering is based on the native sequencing, where the ®rst Met is residue 1. The main-chain fold can be described as follows. The N terminus of the protein starts in the back of protein (Figure 3(b)). The ®rst ®ve residues (ÿ5 to ÿ1, Gly-Pro-Leu-Gly-Ser, not included in Figure 1) are non-native, arising from the PreScission protease cleavage site, and appear somewhat disordered. However, they do show contacts to residues 73 to 77 of the outer b-strand (bD) of the central

1168

Structure of Ap4A Hydrolase

Table 1. Summary of structural statistics for the family of 25 accepted structures A. NMR experimental constraints Distance restraints All Intraresidue Sequential Short-range Long-range Hydrogen bond Dihedral angle restraints All f c w1 Dipolar coupling restraints (1DNH) 13 a 13 b C / C shifts Three-bond coupling (3JHNa) B. Structure statistics Restraint violations Ê) Max. distance violation (A Ê Number of distance violations >0.2 A Ê) Mean rmsd from distance restraints (A Max. angle violation (deg.) Number of angle violations >2  Mean rmsd from dihedral angle restraints ( ) Deviation from idealized covalent geometry Ê) Bonds (A Angles (deg.) Impropers (deg.) Lennard-Jones energy (kcal molÿ1)a Ramachandran plotb Residues in most-favoured regions Residues in additional allowed regions Residues in generously allowed regions Residues in disallowed regions C. rmsd to the mean structure All residuesb Backbone atoms (C0 ,O,N,Ca) All non-hydrogen atoms Ordered residuesc Backbone atoms (C0 ,O,N,Ca) All non-hydrogen atoms

2086 7 611 406 982 80 219 81 84 54 93 280 92 Family

Best

0.38 6.76  2.65 0.0316  0.0019 3.76 0.96  1.01 0.2510  0.0666

0.26 8 0.031 1.46 0 0.199

0.0028  0.0003 0.4387  0.0181 0.3477  0.0204 ÿ406.10  61.04

0.0024 0.413 0.309 ÿ526.38

72.9 22.5 2.8 1.9

71.1 25.2 3.0 0.7

0.90  0.22 1.28  0.21 0.32  0.04 0.79  0.06

Lowest energy conformers obtained from 100 starting structures; calculations in DYANA using the same experimental dihedral Ê 2. and distance restraints yielded the best 25 structures with target functions of 5.0 to 6.4 A a The Lennard-Jones van der Waals energy was calculated within CNS and is not included in the target function for simulated annealing. b Only native residues were considered, residues 1-160. c Ordered residues include 9-75, 97-115, 125-160.

sheet and to residues in the C-terminal helix, aIV. Residues 1 to 10 are poorly de®ned, and are located below the central b-sheet and helix aII, making contact with residues 74 to 77 and 95 to 99. The ®rst b-strand (bA, residues 11 to 17), one of the internal strands of the central sheet, runs parallel with bE and antiparallel to bC of the central sheet. After this strand, residues, 18 to 21 are involved in a type I reverse turn, which precedes strand bB (residues 23 to 27) of the antiparallel two-stranded b-sheet. Residues 28 to 31 also form a type I reverse turn, which is followed by a wellde®ned loop and strand that is not a part of a bsheet. After this strand the main-chain enters strand bC (residues 38 to 41), an outer strand of the central sheet, running antiparallel to bA. In agreement with NH exchange data, the structure shows a hydrogen bond between residues 26 and

36, and consequently the strand bB is not completely separate from the central sheet. However, the turn-loop-strand between strands bB and bC disrupts any further contact between bB and bA, making the central four-stranded sheet independent from the two-stranded antiparallel sheet. At the C terminus of strand bC are Gly40 and Gly41, the former is the ®rst conserved Gly of the Nudix motif. Strand bC is followed by a small loop that precedes the catalytic helix (aI, residues 48 to 60). This loop-helix motif contains the remainder of the Nudix box residues, including several conserved glutamate residues (Glu55, Glu58 and Glu59). Helix aI is below and orthogonal to the central b-sheet, lying in the natural twist of the sheet. The main-chain reverses as a well-de®ned loop and then forms strand bD (residues 65 to 70), the other outer strand of the central sheet, which runs anti-

1169

Structure of Ap4A Hydrolase

Figure 2. Plots of (a) the distribution of observed and assigned NOE data for lupin Ap4A hydrolase: ®lled black bars are for longrange NOEs, gray bars for shortrange NOEs, open bars for sequential NOEs. (b) The distribution of calculated long-range NOEs for the best conformer of the family of solution structures (for protons nearer Ê , belonging to residues than 5 A more than ®ve apart in the sequence). (c) The rmsd from the mean structure for the backbone (C0 , N, O, Ca) atoms. The location of helices (I to IV) and strands (A to F) are indicated. (d) The steadystate {1H}-15N NOE.

parallel to strand bE. After bD the backbone moves away from strand bE, forming a bulge between residues 71 and 74. Residues 75 and 76 in most conformers exist as a short strand (bD0 ) antiparallel to bE, which is supported by backbone chemical shifts (Swarbrick et al., 2000), interstrand NOEs between the NH protons of residues 75 and 100, and the slowly exchanging NH of residue 100. The continuation of this strand involves a poorly de®ned helix (aII, residues 81 to 91) that essentially appears disordered with respect to the core of the protein. Nevertheless, superposition over the region 78 to 97 (Figure 3(c)) and analysis by PROCHECK-NMR shows the presence of a mix of 310 and a-helix spanning residues 81 to 91, in agreement with the Chemical Shift Index data previously reported (Swarbrick et al., 2000). The second internal strand of the sheet (bE, residues 100 to 107) follows and runs parallel with strand bA and antiparallel to bD. The main-chain crosses over the sheet as a large disordered loop positioned at the C-terminal end of aI and then enters into the second strand of the two-stranded antiparallel b-sheet (bF, residues 126 to 132). A short helix (aIII, residues 134 to 140), precedes and runs antiparallel to a larger C-terminal helix (aIV, residues 146 to 155). Helices aIII and aIV are on the opposite side of the b-sheet to the catalytic helix, aI, and are orthogonal to the sheet. The helix aIV sits above the central b-sheet, making many contacts with residues in that sheet; however, aIII sits close

to the edge and near the two-stranded antiparallel sheet, making contacts with residues in this small sheet. The protein ends with a possible small piece of 310 helix that spans residues 157 to 159. The poorly defined regions Despite the large number of NOEs, a considerable portion of the protein is less well de®ned in two regions: 74 to 96 and 117 to 124 (Figure 2(c)). Both regions may have some functional importance. The ®rst region may play a role in substrate and inhibitor binding (see below) and the second contains the putative metal ligand, Glu125, and is near the C-terminal end of the catalytic helix. We have addressed whether it is a lack of data or mobility, or both, by comparing the distribution of experimental NOE data with possible interproton Ê and acquired heteronuclear distances less than 5 A NOE data as a part of a protein dynamics study. Figure 2(a) and (b) show that the relative number of NOEs assigned is representative of the calculated long-range distances (NOEs between residues greater than ®ve apart in the sequence). All observable NOEs for helix aII (residues 81 to 91) have been assigned to give a total of 82 NOEs (36 sequential, 41 medium-range, four long-range, one intraresidue), where the long-range NOEs are between residues 84 to 96 and 91 to 97. These data compare favorably to the average of 12 interresidue NOEs per residue, but unfavorably to the

1170

Structure of Ap4A Hydrolase

Figure 3. Main-chain and ribbon models of the solution structure of Ap4A hydrolase. (a) Superposition (residues 975, 97-115, 125-160) of the 25 conformers used to represent the structure of Ap4A hydrolase. The four b-strands of the mixed central sheet are colored cyan; the two b-strands of the antiparallel sheet in magenta; helices in red; the Nterminal region in green. (b) Same as (a) but rotated around y by about 50  . Helix a1 has an interhelical angle of approximately 60  to helix aIII, and 130  to aIV. Helix aIII and helix aIV are approximately antiparallel, with an interhelical angle of 170  . All three helices are approximately orthogonal to the central sheet. (c) Ribbon representation of the lowest-energy conformer of the family of structures. The structure is in the same orientation as in (a). Helices are labelled aI to aIV; strands A-F. The continuation of strand D is indicated as D0 , which follows a bulge. Small pieces of 310 helix are observed between strands E and F, and at the end of aIV. (d) The region 73 to 97 (superposition over C0 , O, N, Ca for residues 78-97, rmsd 0.89  0.28) showing the helical nature of this region. The apparent disorder of this helix is partly due to an absence of long-range distance constraints from this helix to other regions.

average of six long-range NOEs per residue for the whole protein (Table 1). However, the relative number of expected long-range NOEs is consistent with the observed, supporting that it is a lack of observable long-range NOEs from aII to other parts of the protein that is partly the cause of poor de®nition for this region. Notably, the heteronuclear NOE data suggest that the region 74 to 91 is

more mobile than other secondary structural elements. As expected for an isolated helix, aII is relatively less stable than other secondary structural elements. We assessed NH exchange qualitatively by observing NH resonances in 15N,1HHSQC spectra after sample preparation (about six hours at 4  C, pH 6.5). While most NH resonances persisted for protons that are expected to stabilize

Structure of Ap4A Hydrolase

the b-sheets and helices aI, aIII and aIV, no slowly exchanging NH resonances were observed for aII. The lack of stability must be due to the relatively isolated nature of this helix, preventing the formation of stabilizing interactions characteristic of helices in globular proteins. The hydrophobic core and surface The majority of the hydrophobic residues are located above the b-sheet, with only a few hydrophobic residues below the sheet (for example, Phe103 and Phe105) packing with the catalytic helix aI (Ile53 and Leu56) (Figure 4). Helix aIV shows a hydrophobic face (Val148, Tyr149, Val152, Val155 and Phe156) that is in contact with Met37, Trp102 and Leu104 of the central b-sheet (strands bC and bD). The side-chain of Trp102 is lying ¯at across this sheet, causing a number of signi®cant ring current shifts, including a 4 ppm up®eld shift of a CaH of Gly13. Another core of hydrophobic residues comprises residues from the two-stranded b-sheet, including Phe24 (bB), Phe126, Trp129 and Trp131 (which are orthogonal to each other) from bF, Leu16 from bA, Leu56 from aI and residues located prior to the long ¯exible loop near a1, including Ile114 and Leu116. Most of these residues are poorly conserved, although their hydrophobic nature is maintained (Figure 1). Of the seven tryptophan residues in lupin Ap4A hydrolase, only two are highly conserved, Trp35 and Trp74, both are not in these cores. Indeed, both tryptophan residues would appear distant from the substrate-binding site (see below) and therefore their conservation must be for structural reasons.

1171 Trp35 is located prior to the Nudix region, in another conserved region (Figure 1). The side-chain of Trp35 is near the protein surface with its indole nitrogen atom oriented into the solvent. Trp74 is aligned face-on with Pro4 and may provide some packing for the N-terminal region. Surprisingly, Trp102, which has a central role in the packing of the protein, is not conserved. Trp91 and Trp95 are in the ill-de®ned aII helix, but in most conformers are stacked with each other (Figure 4(c)). Strong NOEs between the indole Ne1H and water indicate that these residues are quite accessible. Ap4A hydrolase has 17 glutamate and 11 aspartate residues, of which ®ve glutamate residues are strictly conserved (Figure 1). There is a concentration of negatively charged residues along aI and the two loops positioned at either end (Figure 5) of this helix. In addition to the strictly conserved glutamate residues (Glu46, Glu55, Glu58 and Glu59) of the Nudix box, there are additional glutamate and aspartate residues in this region, Asp43, Glu44 and Asp47. In the long loop there are Glu122, Glu125 and Glu128, where Glu125 is strictly conserved in this family. Most of the conserved glutamate residues have been shown to be catalytically essential in lupin Ap4A hydrolase and other Nudix enzymes (Harris et al., 2000; Dunckley & Parker, 1999; Yang et al., 1999; Safrany et al., 1998; D.M. et al., unpublished results). Comparison to the structure of the ternary complex of inhibitor[Mg2‡]2-MutT complex (Lin et al., 1997) suggests that in Ap4A hydrolase Glu55, Glu58, Glu59 and Glu125 are candidates for ligating a magnesium ion. Glu55, Glu59 and Glu125 are all clustered together, supporting their roles in metal ligation,

Figure 4. Distribution of hydrophobic residues of Ap4A hydrolase. (a) Aromatic residues: tryptophan purple; tyrosine green; phenylalanine blue; and the sole His159 orange. (b) Methyl-containing residues: leucine green; valine blue; isoleucine orange; methionine and the single Cys15 purple. Cys15 obscures Met37 and Leu104. (c) The loop region including helix aII. Despite the lack of de®nition and possible mobility of this region, there are several hydrophobic residues, including Tyr77, which is conserved in all the Ap4A hydrolases. Analogous to MutT (Lin et al., 1997), the adenosine ring of the substrate is expected to pack between Tyr77 and Phe79.

1172

Structure of Ap4A Hydrolase

Figure 5. (a) Electrostatic potential mapped onto the surface of the lowest-energy model of Ap4A hydrolase. The surface was generated using the default parameters of GRASP. (b) Positions of the proposed metal ligand glutamate residues (Glu55, Glu58, Glu59 and Glu125), and other negatively charged residues that may participate in the ligation of a second metal (Conyers et al., 2000). The positions of the carbonyl groups of Gly40 and Gly41 are indicated. The orientation of the structure is similar to that in (a).

which is consistent with site directed mutations of these residues resulting in 102 to 105-fold decreases in kcat (D.M. et al., unpublished results). Glu58 is also near this cluster, but similar mutation of this residue results in only two to tenfold reduction in kcat, suggesting that either it does not ligate magnesium or its ligation is not essential. The carbonyl group of the conserved Gly (Gly40 in Ap4A hydrolase) of the Nudix signature is expected to ligate magnesium; however, a magnesium titration of lupin Ap4A hydrolase (D.M. et al., unpublished results) suggested that Gly41, a residue that is not strictly conserved in the Nudix family (Dunn et al., 1999), is a metal ligand. The structure presented here shows that the carbonyl group of Gly41 is distant from the glutamate cluster, suggesting that it cannot ligate the same metal ion. Gly40 is nearer, but cannot be involved in ligating metal as its carbonyl group is oriented towards strand bA and away from the catalytic cluster of glutamate residues (Figure 5). The 3D 15 N-edited NOESY-HSQC data collected on Ap4A hydrolase complexed with and without magnesium show no detectable difference in NOE patterns. Despite changes to several chemical shifts, particularly the peptide NH of Ile42, the NOEs from Gly41 and Ile42 to residues on strand bA are retained, suggesting that within this region the backbone does not readjust on the addition of magnesium. Recently, magnetic resonance studies on the B. bacilliformis Ap4A hydrolase (Conyers et al., 2000) suggested that these enzymes differ from those of the MutT family in that they have Ê of each two enzyme-bound metal ions within 6 A other. Consequently, it is tempting to suggest that Gly41 ligates a second metal ion. Other residues near Gly41 that are possible new metal ligands include Asp43, Glu46 and Asp47. Mutation of Glu46, however, had no effect on kcat, indicating that this residue is not catalytically essential (D.M. et al., unpublished results). Again, small chemical shift differences are observed for the peptide NH

of these residues on complexing metal, but there is no signi®cant change to the NOE patterns. Inhibitor titration of Ap4A hydrolase The preferred substrate for Ap4A hydrolase is Ap4A; however, it is not known whether one or two adenosine rings of this molecule bind to the protein. Analogous to the structure of the MutTinhibitor complex (Lin et al., 1997), a binding site for a purine ring is predicted to be above the central sheet, and near to the poorly de®ned helix, aII (Figure 6). The structure of the complex of MutT with a,b-methyleneadenosine triphosphate bound showed contacts between the adenosine-ribose group of the inhibitor and hydrophobic residues of strands A (Leu4, Ile6), C (Ile80) and D (Tyr73, Phe75) (equivalent to bA, bD and bE in Ap4A hydrolase, respectively) (Lin et al., 1997). The NdH2 side-chain of Asn119 in MutT is expected to make contact with O-6 of the guanine ring. These residues are not conserved in Ap4A hydrolase, except for Tyr77 (Dunn et al., 1999), which is located prior to aII (Figures 1 and 6). Residues in bA and bD of Ap4A hydrolase that are in positions approximately similar to those of Leu4, Ile6, and Ile80 of MutT are Asn11 (bA), Gly13 (bA) and Gln100 (bD), respectively, and therefore we may expect some differences in inhibitor binding. Located above and at either end of the b-sheet are clusters of conserved glutamine and arginine residues. Arg9, Gln98 and Gln100 form one group, which are near the analogous MutT binding site. A second group consists of Arg28, Gln36 and Gln39 at the opposite end of the central sheet. Near these groups are conserved tyrosine and phenyalanine residues; Tyr77 is near the ®rst group, and Phe126, Phe144 and Tyr149 are near the second group. We have titrated 15N-labelled lupin Ap4A hydrolase with the inhibitor P1,P4-dithio-P2,P3-monochloromethylene diadenosine 50 ,5000 -P1,P4-tetraphosphate (Chan et al., 1997) and followed the effect that the

1173

Structure of Ap4A Hydrolase

Figure 6. Residues that showed 1H,15N chemical shift differences with the titration of an Ap4A analagoue (Chan et al., 1997). (a) and (b) are the same but (b) is rotated 90  around x. Residues that are conserved in the Ap4A hydrolase family (Figure 1) and are perturbed by the addition of inhibitor are colored green and marked. Residues that are not conserved, but are perturbed by the addition of inhibitor are colored blue. Tyr77, a residue strictly conserved in the Ap4A hydrolases (Dunn et al., 1999), is colored magenta. The effect of the inhibitor on the NH resonance of this residue is not clear, as it is overlapping. A number of residues in aII are sensitive to the inhibitor, suggesting this region may play a signi®cant role in binding substrate. The region near Tyr77 and Phe79 is expected to bind an adenine ring, analogous to the inhibited structure of MutT (Lin et al., 1997).

addition of this inhibitor has on the peptide and side-chain NH groups of the protein. This inhibitor is a mixture of stereoisomers (RR0 , SS0 , RS0 , and SR0 ), and the hope was for selective binding of one isomer in slow exchange. Unfortunately, this was not the case and, while many resonances disappeared, these were replaced by numerous, but weak peaks suggesting multiple bound stereoisomers (Figure 7). Nevertheless, plotting the residues whose peptide NH resonances clearly disappeared, suggesting proximity to the inhibitorbinding site, onto the structure, points to a substrate-binding cleft above the central b-sheet (Figure 6). It is apparent that these residues include conserved residues from the two groups described above. These include Arg9 and Gln100 from the ®rst group of conserved residues, and Arg28, Gln36, Phe144 and Tyr149 from the second group. It is of some interest that residues in aII are affected by the inhibitor, including Asn88, Gly92, Ser93 and Ne1H of Trp91 and Trp95. None of these residues is conserved (Figure 1). The peptide NH of Tyr77, the strictly conserved tyrosine residue, overlaps with other resonances and thus the effect of the inhibitor on this NH is not clear; however, the NH of Phe79 is substantially attenuated with the addition of inhibitor. The peptide NH, and not the Ne1H indole resonance of Trp35 was affected, which is consistent with this residue being oriented

away from the inhibitor-binding site. Notably, the addition of the inhibitor has minimal effect on residues in the Nudix box and other residues located below the central b-sheet, with no effects on the residues of helix aI. As the residues that are affected by the addition of the inhibitor form a large continuous cleft, we cannot conclude whether one or two adenine rings bind to the enzyme. This knowledge will be obtained only through the analysis of a stable, single-inhibited conformer.

Conclusion We have described the ®rst high-resolution solution structure of an Ap4A hydrolase. Although the structure is similar to that of MutT, the only other enzyme from the Nudix family to be solved, clear differences are observed. These include the presence of an ill-de®ned helix, aII, which is near the expected substrate-binding site, a well-de®ned helix aIII, and the position of the two-stranded bsheet, which is more independent of the central sheet. The position of the glutamate residues that ligate metal appear similar to MutT, but the carbonyl group of Gly41, an apparent metal ligand analogous to MutT, is noticeably distant from these glutamate residues. This Gly may therefore be a potential ligand to a second metal ion. Although the Ap4A analogue did not form a discrete com-

1174

Structure of Ap4A Hydrolase

Figure 7. Part of the [15N, 1H]-HSQC spectra of Ap4A hydrolase, pH 6.5, 20 mM MgCl2 20 mM imidazolium, titrated with the inhibitor P1,P4-dithio-P2,P3-monochloromethylene diadenosine 50 ,5000 -P1,P4-tetraphosphate. Protein concentration was 0.3 mM and the inhibitor (a) 0 mM, (b) 0.19 mM, (c) 0.39 mM. Squares indicate resonances that have been bleached from the spectrum. (*) in (c) indicates a new resonance. Peptide NH resonances of other residues that were clearly affected by the addition of the inhibitor, but are not included in these spectra are: Tyr8, Arg9, Asn11, Val12, Arg28, Leu29, Trp35, Gln36, Met37, Gly40, Gly41, Trp74, Thr76, Asp78, Phe79, Gly92, Ser93, Gln100, Ile138, Glu143, Phe144, Lys145, Val148, Tyr149, and Val155. All residues affected by the inhibitor are shown in Figure 6.

plex with the enzyme, several conserved residues outside of the Nudix signature, but located near or in the central sheet were identi®ed as possibly important for substrate and inhibitor recognition. These ®ndings have formed the basis for further inhibitor binding and dynamic studies to develop our understanding of the structure and mechanism of Ap4A hydrolases and other Nudix enzymes.

Materials and Methods Sample preparation Expression and puri®cation of residues 1-160 of Ap4A hyrolase has been described (Swarbrick et al., 2000; Maksel et al., 1998). In brief, the protein was expressed as a GST fusion from pGEX-6P-3 (Pharmacia Biotech, Inc) in a 2 l fermenter (B. Braun Biotech. International) and isotopically enriched with [15N]ammonium chloride and [13C]glucose (Cambridge Isotopes Labs.), following established procedures (Cai et al., 1998). After puri®cation, cleavage with PreScission protease and repuri®cation on glutathione-Sepharose (Pharmacia Biotech, Inc) and Mono-Q column (Pharmacia Biotech, Inc) a residual protease contaminant was observed to be present. This contaminant resulted in minor degradation over 24 hours, to considerable degradation over several days. As the addition of 1-3 mM EDTA inhibited degradation, we have determined the full structure of the apoenzyme. Recently, we determined the isoelectric point of the enzyme by capillary isoelectrofocussing (Beckman P/ ACE System 5010) to be 4.3, and thus have included a ®nal step of puri®cation: a Mono-S column (Pharmacia Biotech, Inc.) in 25 mM Mes (pH 5.5), which gives pure enzyme free of protease contaminants. NMR spectroscopy Samples of Ap4A hydrolase (0.55 ml, 0.7 to 1.2 mM) were prepared for NMR spectroscopy. Samples for the sequence-speci®c assignment of resonances and NOEs

were prepared in a 50 mM sodium phosphate buffer (pH 6.5), 1.5 mM dithiothreitol, 3 mM EDTA, 0.02 % (w/ v) sodium azide, and either 100 % 2H2O or 90 % H2O/ 10 % 2H2O. All samples contained 20 ml of a protease inhibitor solution made from one tablet of CompleteTM EDTA-free (Boehringer Mannheim) dissolved in 5 ml of H2O or 2H2O. Samples for inhibitor titration and preliminary characterization of the binary complex of Mg2‡enzyme were prepared in 20 mM imidazolium buffer (pH 6.5), 20 mM MgCl2. Using a Varian 600 Inova, operating at 25  C and equipped with a 5 mm 1H, 13C, 15N single z-axis gradient probe, the following spectra including those reported elsewhere (Swarbrick et al., 2000) were acquired on samples in phosphate buffer: 3D 15N-edited NOESYHSQC (100 ms mixing time) (Zhang et al., 1994), 13C-edited NOESY-HSQC (100 ms mixing time, 13C carrier set to 46 ppm, in 100 % 2H2O and in 90 %/10 % H2O/2H2O) (Muhandiram et al., 1993), 13C-edited NOESY-HSQC (100 ms mixing time, 13C carrier set to 125 ppm in 100 % 2 H2O), 2D (Hb)Cb(CgCd)Hd, 2D (Hb)Cb(CgCdCe)He (Yamazaki et al., 1993), 2D 1H TOCSY-relayed ct[13C,1H]-COSY (Zerbe et al., 1996), 3D HNHB (Archer et al., 1991), 3D HACAHB (Grzesiek et al., 1995), 3D HNHA (Vuister & Bax, 1993), [13C,1H]-HSQC on 10 % 13 C-labelled enzyme, for stereoassignment of the methyl groups of leucine and valine residues (Senn et al., 1989), and 2D TOCSY (100 % 2H2O, 40 ms mixing time). A 0.5 mM sample was prepared in phosphate buffer with 12 mg/ml of the ®lamentous bacteriophage pf1, prepared as described (Hansen et al., 1998). Residual dipolar couplings were determined from [15N,1H]-HSQC-IPAP experiments (Ottiger et al., 1998). To analyze changes on complexing magnesium, 3D 15N-edited NOESY-HSQC spectra (100 ms mixing time) were acquired on the binary complex of enzyme-Mg2‡. Samples of 15Nlabelled Ap4A hydrolase (0.3 mM) in 20 mM imidazolium buffer (pH 6.5), 20 mM MgCl2 were titrated with 02 mM P1,P4-dithio-P2,P3-monochloromethylene diadenosine 50 ,5000 -P1,P4-tetraphosphate (Chan et al., 1997) dissolved in 20 mM imidazolium (pH 6.5), 20 mM MgCl2. Spectral changes were followed by 2D [15N, 1H]-HSQC

1175

Structure of Ap4A Hydrolase spectra. The steady-state {1H}-15N NOE were determined as described (Farrow et al., 1994). A relaxation delay of two seconds prior to three seconds of saturation was employed for the NOE spectra, and a ®ve seconds relaxation delay for the reference spectrum. The data were acquired in an interleaved manner. Data were processed with NMRPipe (Delaglio et al., 1995). All constant-time data sets were linear-predicted in the constant-time acquired dimensions using the mirror-image method to double the number of data points prior to apodization. The indirect dimensions of 3D 13C or 15N-edited NOESY-HSQC were not linear-predicted, whereas other non-constant-time indirect dimensions for all other experiments were linear-predicted by the forward-backward method to double the number of data points. Typical window functions used were cosine bells in the directly detected dimension and cosine-squared bells in indirectly detected dimensions. Data were converted by SPSCAN{ and analyzed with XEASY (Bartels et al., 1995). 1H chemical shifts were referenced to DSS at 0 ppm, and 13C and 15N chemical shifts calculated indirectly from the 1H spectrometer frequency. Structure calculations Distance restraints were derived from the 3D 15N and C separated NOESY spectra. NOE peaks were integrated with the peakint function in XEASY, and the NOE volumes were converted to distances using CALIBA (GuÈntert et al., 1991), using expected distances in secondary structural elements as a guide for calibration. Volumes of NOEs of methyl groups, degenerate methylene groups and aromatic resonances were divided by the number of participating protons. NOEs that involved backbone atoms were calibrated expecting an rÿ6 dependence, and non-backbone atoms to an rÿ4 dependence. The minimum and maximum upper bound distances Ê , respectively; the lower bound diswere 2.4 and 5.5 A Ê , the sum of van der Waals radii of tance was set at 1.8 A two protons. Dihedral angles were determined with the following procedures and experiments: f,c pairs were derived from 13Ca chemical shifts (LuginbuÈhl et al., 1995); w1 angles were determined by considering HNHB and HACAHB experiments and assigned to either ÿ60, ‡60 or 180  30  . The three bond couplings 3JHNa from HNHA experiments were tabulated for direct re®nement (Garrett et al., 1994), assuming an error of 2 Hz and all couplings 6 Hz<3JHNa < 8 Hz were excluded. Similarly, 13 a C and 13Cb chemical shifts were tabulated for direct re®nement (Kuszewski et al., 1995). Measured residual dipolar couplings (1DNH) ranged from ÿ16.8 to 12.3 Hz. As the data are incomplete, the axial (ÿ7.5 Hz) and rhombicity (0.22) terms were determined and redetermined on trial structures during the late phase of structure determination (Clore et al., 1998). Residual dipolar couplings from regions having heteronuclear {1H}-15N NOE ratios less than 0.7 were excluded in the re®nement, (residues ÿ5 to 9, 77 to 95 and 116 to 122). Errors of  2 Hz were assumed. Trial structures (100 conformers), using NOE and dihedral angle data, were calculated in either DYANA v1.4 or v1.5 (GuÈntert et al., 1997) on an SGI O2 workstation or CNS v0.9 (BruÈnger et al., 1998) on dual-Pentium III 450 and 550 MHz workstations operating with Linux 13

{ R. W. Glaser and K. WuÈthrich, http:// www.mol.biol.ethz.ch/wuthrich/software/spscan/

Redhat 6.0 or 6.2. The ®nal stages of re®nement used CNS and all data described above. In the ®nal Cartesian SA re®nement step in CNS, the following force constants Ê ÿ2 for experimental distance were used: 40 kcal molÿ1 A ÿ1 restraints, 200 kcal mol radÿ2 for dihedral angle restraints, 0.5 kcalÿ1 ppmÿ2 for 13Ca and 13Cb restraints, 0.5 kcal molÿ1 Hzÿ2 for 1DNH restraints, and 1.0 kcal molÿ1 Hzÿ2 for 3JHNa couplings. Structures were analysed with PROCHECK-NMR (Laskowski et al., 1996), MOLMOL (Koradi et al., 1996), and GRASP (Nicholls et al., 1991). Protein Data Bank accession number The ®nal ensemble of 25 conformers has been deposited with the RCSB Protein Data Bank; accession number 1F3Y, release date 6th June, 2001.

Acknowledgements We are grateful to Dr Lewis Kay and Dr Ranjith Muhandiram, University of Toronto, and Dr Walter Zhang, St. Jude Children's Research Hospital, Memphis, for providing some pulse sequences; Dr Richard Pau for preparing the pf1 bacteriophage; Edward D'Auvergne for assistance in analysing the {1H}-15N NOE data; Zlatan Trifunovic for the capillary isoelectrofocussing; Dr Jamie Fletcher for assisting with the BLAST search; the Australian Research Council and the National Health & Medical Research Council for ®nancial support.

References Archer, S. J., Ikura, M., Torchia, D. A. & Bax, A. (1991). An alternative 3D NMR technique for correlating backbone 15N with side-chain Hb resonances in larger proteins. J. Magn. Reson. 95, 636-641. Bartels, C., Xia, T., Billeter, M., GuÈntert, P. & WuÈthrich, K. (1995). The program XEASY for computer-supported NMR spectral analysis of biological macromolecules. J. Biomol. NMR, 5, 1-10. Bessman, M. J., Frick, D. N. & O'Handley, S. F. (1996). The MutT proteins or ``Nudix'' hydrolases, a family of versatile, widely distributed, ``housecleaning'' enzymes. J. Biol. Chem. 271, 25059-25062. BruÈnger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-Kunstleve, R. W., Jiang, J.-S., Kuszewski, J., Nilges, M., Pannu, N. S., Read, R. J., Rice, L. M., Simonson, T. & Warren, G. L. (1998). Crystallography & NMR system: a new software suite for macromolecular structure determination. Acta Crystallog. sect. D, 54, 905-921. Cai, M., Huang, Y., Sakaguchi, K., Clore, G. M., Gronenborn, A. M. & Craigie, R. (1998). An ef®cient and cost-effective isotope labeling protocol for proteins expressed in Escherichia coli. J. Biomol. NMR, 11, 97-102. Cartwright, J. L., Britton, P., Minnick, M. F. & McLennan, A. G. (1999). The IalA invasion gene of Bartonella bacilliformis encodes a (di)nucleoside polyphosphate hydrolase of the MutT motif family and has homologs in other invasion bacteria. Biochem. Biophys. Res. Commun. 256, 474-479. Chan, S. W., Gallo, S. J., Kim, B. K., Guo, M. J., Blackburn, G. M. & Zamecnik, P. C. (1997). P1, P4-dithio-P2,P3-monochloromethylene diadenosine

1176 50 ,5000 -P1,P4-tetraphosphate - a novel antiplatelet agent. Proc. Natl Acad. Sci. USA, 94, 4034-4039. Clore, G. M., Gronenborn, A. M. & Tjandra, N. (1998). Direct structure re®nement against residual dipolar couplings in the presence of rhombicity of unknown magnitude. J. Magn. Reson. 131, 159-162. Conyers, G. B. & Bessman, M. J. (1999). The gene, ialA, associated with the invasion of human erythrocytes by Bartonella bacilliformis, designates a nudix hydrolase active on dinucleoside 50 -polyphosphates. J. Biol. Chem. 274, 1203-1206. Conyers, G. B., Wu, G., Bessman, M. J. & Mildvan, A. S. (2000). Metal requirements of a diadensoine pyrophosphatase from Bartonella bacilliformis: magnetic resonance and kinetic studies of the role of Mn2‡. Biochemistry, 39, 2347-2354. Delaglio, F., Grzesiek, S., Vuister, G. W., Zhu, G., Pfeifer, J. & Bax, A. (1995). NMRPipe - a multidimensional spectral processing system based on unix pipes. J. Biomol. NMR, 6, 277-293. Dunckley., T. & Parker, R. (1999). The DCP2 protein is required for mRNA decapping in Saccharomyces cerevisiae and contains a functional MutT motif. EMBO J. 18, 5411-5422. Dunn, C. A., O'Handley, S. F., Frick, D. N. & Bessman, M. J. (1999). Studies on the ADP-ribose pyrophosphatase subfamily of the nudix hydrolases and tentative identi®cation of trgB, a gene associated with tellurite resistance. J. Biol. Chem. 274, 32318-32324. Farrow, N. A., Muhandiram, R., Singer, A. U., Pascal, S. M., Kay, C. M., Gish, G., Shoelson, S. E., Pawson, T., Forman-Kay, J. D. & Kay, L. E. (1994). Backbone dynamics of a free and a phosphopeptide-complexed src homology 2 domain by 15N NMR relaxation. Biochemistry, 33, 5984-6003. Frick, D. N., Weber, D. J., Abeygunawardana, C., Gittis, A. G., Bessman, M. J. & Mildvan, A. S. (1995). NMR studies of the conformations and location of nucleotides bound to the Escherichia coli MutT enzyme. Biochemistry, 34, 5577-5586. Garrett, D. S., Kuszewski, J., Hancock, T. J., Lodi, P. J., Vuister, G. W., Gronenborn, A. M. & Clore, G. M. (1994). The impact of direct re®nement against three-bond HN-CaH coupling constants on protein structure determination by NMR. J. Magn. Reson. ser. B, 104, 99-103. GuÈntert, P., Braun, W. & WuÈthrich, K. (1991). Ef®cient computation of three-dimensional protein structures in solution from nuclear magnetic resonance data using the program DIANA and the supporting programs CALIBA, HABAS and GLOMSA. J. Mol. Biol. 217, 531-530. GuÈntert, P., Berndt, K. D. & WuÈthrich, K. (1993). The program ASNO for computer-supported collection of NOE upper distance constraints as input for protein structure determination. J. Biomol. NMR, 3, 601606. GuÈntert, P., Mumenthaler, C. & WuÈthrich, K. (1997). Torsion angle dynamics for NMR structure calculation with the new program DYANA. J. Mol. Biol. 273, 283-298. Grzesiek, S., Kuboniwa, H., Hinck, A. P. & Bax, A. (1995). Multiple-quantum line narrowing for measurement of Ha-Hb J couplings in isotopically enriched proteins. J. Am. Chem. Soc. 117, 5312-5315. Guranowski, A., Jakubowski, H. & Holler, E. (1983). Catabolism of diadenosine 50 ,5000 -P1,P4-tetraphosphate in procaryotes. Puri®cation and properties of diadenosine 50 ,5000 -P1,P4-tetraphosphate (symmetri-

Structure of Ap4A Hydrolase cal) pyrophosphohydrolase from Escherichia coli K12. J. Biol. Chem. 258, 14784-14789. Hansen, M. R., Mueller, L. & Pardi, A. (1998). Tunable alignment of macromolecules by ®lamentous phage yields dipolar coupling interactions. Nature Struct. Biol. 5, 1065-1074. Harris, T. K., Wu, G., Massiah, M. A. & Mildvan, A. S. (2000). Mutational, kinetic, and NMR studies of the roles of conserved glutamate residues and of lysine39 in the mechanism of the MutT pyrophosphohydrolase. Biochemistry, 39, 1655-1674. Koradi, R., Billeter, M. & WuÈthrich, K. (1996). MOLMOL: a program for display and analysis of macromolecular structures. J. Mol. Graph. 14, 51-55. Kuszewski, J., Qin, J., Gronenborn, A. M. & Clore, G. M. (1995). The impact of direct re®nement against 13Ca and 13Cb chemical shifts on protein structure determination by NMR. J. Magn. Reson. ser. B, 106, 92-96. Laskowski, R. A., Rullmann, J. A. C., MacArthur, M. W., Kaptein, R. & Thornton, J. M. (1996). AQUA and PROCHECK-NMR: Programs for checking the quality of protein structures solved by NMR. J. Biomol. NMR, 8, 477-486. Lin, J., Abeygunawardana, C., Frick, D. N., Bessman, M. J. & Mildvan, A. S. (1996). The role of Glu57 in the mechanism of the Escherichia coli MutT enzyme by mutagenesis and heteronuclear NMR. Biochemistry, 35, 6715-6726. Lin, J., Abeygunawardana, C., Frick, D. N., Bessman, M. J. & Mildvan, A. S. (1997). Solution structure of the quaternary MutT-M2‡-AMPCPP-M2‡ complex and mechanism of its pyrophosphohydrolase action. Biochemistry, 36, 1199-1211. LuginbuÈhl, P., Szyperski, T. & WuÈthrich, K. (1995). Statistical basis for the use of 13Ca chemical shifts in protein structure determination. J. Magn. Reson. ser. B, 109, 229-233. Maksel, D., Guranowski, A., Ilgoutz, S. C., Moir, A., Blackburn, M. G. & Gayler, K. G. (1998). Cloning and expression of diadenosine 50 ,5000 -P1, P4-tetraphosphate hydrolase from Lupinus angustifolius L. Biochem. J. 329, 313-319. McLennan, A. G. (1999). The MutT motif family of nucleotide phosphohydrolases in man and human pathogens (Review). Int. J. Mol. Med. 4, 79-89. Mitchell, S. J. & Minnick, M. F. (1995). Characterization of a two-gene locus from Bartonella bacilliformis associated with the ability to invade human erythrocytes. Infect. Immun. 63, 1552-1562. Muhandiram, D. R., Farrow, N. A., Xu, G.-Y., Smallcombe, S. H. & Kay, L. E. (1993). A gradient 13 C NOESY-HSQC experiment for recording NOESY spectra of 13C-labeled proteins dissolved in H2O. J. Magn. Reson. ser. B, 102, 317-321. Nicholls, A., Sharp, K. A. & Honig, B. (1991). Protein folding and association: insights from the interfacial and thermodynamic properties of hydrocarbons. Proteins: Struct. Funct. Genet. 11, 281-296. Ottiger, M., Delaglio, F. & Bax, A. (1998). Measurement of J and dipolar couplings from simpli®ed twodimensional NMR spectra. J. Magn. Reson. 131, 373378. Safrany, S. T., Caffrey, J. J., Yang, X., Bembenek, M. E., Moyer, M. B., Burkhart, W. A. & Shears, S. B. (1998). A novel context for the `MutT' module, a guardian of cell integrity, in a diphosphoinositol polyphosphate phosphohydrolase. EMBO J. 17, 6599-6607.

1177

Structure of Ap4A Hydrolase Senn, H., Werner, B., Messerle, B. A., Weber, C., Traber, R. & WuÈthrich, K. (1989). Stereospeci®c assignment of the methyl 1H NMR lines of valine and leucine in polypeptides by nonrandom 13C labelling. FEBS Letters, 249, 113-118. Swarbrick, J. D., Bashtannyk, T., Maksel, D., Pau, R. N., Gayler, K. R. & Gooley, P. R. (2000). 1H, 13C and 15 N backbone assignment and secondary structure of the 19 kDa diadenosine 50 ,5000 -P1, P4-tetraphosphate hydrolase from Lupinus angustifolius L. J. Biomol. NMR, 16, 269-270. Vuister, G. W. & Bax, A. (1993). Quantitative J correlation: a new approach for measuring homonuclear three-bond J(HNHa) coupling constants in 15Nenriched proteins. J. Am. Chem. Soc. 115, 7772-7777. Yamazaki, T., Forman-Kay, J. D. & Kay, L. E. (1993). Two-dimensional NMR experiments for correlating 13 Cb and 1Hd/e chemical shifts of aromatic residues

in 13C-labelled proteins via scalar couplings. J. Am. Chem. Soc. 115, 11054-11055. Yang, X., Safrany, S. T. & Shears, S. B. (1999). Sitedirected mutagenesis of diphosphoinositol polyphosphate phosphohydrolase, a dual speci®city NUDT enzyme that attacks diadenosine polyphosphates and diphosphoinositol polyphosphates. J. Biol. Chem. 274, 35434-35440. Zerbe, O., Szyperski, T., Ottiger, M. & WuÈthrich, K. (1996). 3D 1H-TOCSY-relayed ct-[13C,1H]-HMQC for aromatic spin system identi®cation in uniformly 13C labeled proteins. J. Biomol. NMR, 7, 99-106. Zhang, O., Kay, L. E., Olivier, J. P. & Forman-Kay, J. D. (1994). Backbone 1H and 15N resonance assignments of the N-terminal SH3 domain of drk in folded and unfolded states using enhanced-sensitivity pulsed ®eld gradient NMR techniques. J. Biomol. NMR, 4, 845-858.

Edited by M. F. Summers (Received 12 June 2000; received in revised form 31 July 2000; accepted 31 July 2000)