Crystal structure of the tetrameric inositol 1-phosphate phosphatase (TM1415) from the hyperthermophile, Thermotoga maritima Kimberly A. Stieglitz 1 , Mary F. Roberts 1 , Weizhong Li 2 and Boguslaw Stec 2 1 Merkert Chemistry Center, Boston College, Chestnut Hill, MA, USA 2 The Burnham Institute for Medical Research, La Jolla, CA, USA The extended inositol monophosphatase (IMPase) ⁄ fructose-1,6-bisphosphatase (FBPase) family provides an interesting case in which the changes in 3D archi- tecture and structural flexibility can be linked to increased specialization and emergence of new func- tion. This structural family was identified at the begin- ning of the 1990s [1,2]. Since then, the rapid progress in genome sequencing has added many new members. At present, the family contains around 1000 sequences, classified in three Pfam5 subfamilies, of which  600 are in the IMPase subfamily. Despite having the same architectural features at the single monomer level, the proteins in this family show increasingly complex oligomeric organization. The inositol polyphosphate 1-phosphatases and 3¢-phosphoadenosine-5¢-phosphate phosphatases are monomeric. Most of the IMPases are dimeric, although the Escherichia coli enzyme (also known as SuhB) can be monomeric as well as dimeric [3]. All of the eukaryotic FBPases are tetrameric, and some are allosterically regulated by AMP [4] or oxida- tion of disulfides [5]. The IMPases in higher eukaryotes are involved in secondary messenger signaling by regenerating the myo-inositol pool, whereas the bacterial and archaeal counterparts may be involved in osmolyte synthesis, as has been postulated for the IMPase from Thermoto- ga maritima (TM1415) [6]. Members of this family show reactivity towards several substrates, including poly phosphorylated inositol species and phosphorylat- ed nucleotides. A few IMPase enzymes from hyper- thermophilic organisms also catalyze the hydrolysis of the 1-phosphate of FBP; these have specific FBPase, as well as IMPase, activities [7]. We have previously solved two structures from this superfamily – Metha- nocaldoccocus jannashii IMPase (MJ0109) [7,8] and Archaeoglobus fulgidus IMPase (AF2372) [9] – and in Keywords fructose-1, 6-bisphosphatase; inositol monophosphatase; protein folding; thermophile; Thermotoga maritima Correspondence B. Stec, The Burnham Institute, 10901 N. Torrey Pines Road, La Jolla, CA 92037, USA Fax: +1 858 7139948 Tel: +1 858 7955257 E-mail: bstec@burnham.org (Received 15 December 2006, revised 5 March 2007, accepted 8 March 2007) doi:10.1111/j.1742-4658.2007.05779.x The structure of the first tetrameric inositol monophosphatase (IMPase) has been solved. This enzyme, from the eubacterium Thermotoga maritima, similarly to its archaeal homologs exhibits dual specificity with both IMPase and fructose-1,6-bisphosphatase activities. The tetrameric structure of this unregulated enzyme is similar, in its quaternary assembly, to the allosterically regulated tetramer of fru ctose-1,6-bi sphosph atase. The indi vidual dimers are similar to the human IMPase. Additionally, the structures of two crystal forms of IMPase show significant differences. In the first crystal form, the tetrameric structure is symmetrical, with the active site loop in each subunit folded into a b-hairpin conformation. The second form is asymmetrical and shows an unusual structural change. Two of the subunits have the active site loop folded into a b-hairpin structure, whereas in the remaining two subunits the same loop adopts an a-helical conformation. Abbreviations AF2372, IMPase from Archaeoglobus fulgidus; FBPase, fructose-1,6-bisphosphatase; IMPase, inositol 1-phosphate phosphatase; TM1415, IMPase from Thermotoga maritima; MJ0109, IMPase from Methanocaldococcus jannaschii. FEBS Journal 274 (2007) 2461–2469 ª 2007 The Authors Journal compilation ª 2007 FEBS 2461 this report we present the structure for TM1415, the IMPase ⁄ FBPase from T. maritima. The proteins in the family share a common mono- mer architecture and have common principles of chem- ical reactivity. A three metal ion-assisted catalytic mechanism was proposed to function in the entire fam- ily [8,10]. It has also been suggested that the enzymes might have a common origin in a postulated ancient cyclitol phosphatase [7]. One of the common elements of this architecture is a loop that closes the active site once the substrate binds. Many of the eukaryotic members of this family (including Homo sapiens IMPase, inositol polyphosphate 1-phosphatase, and FBPase) are inhibited by submillimolar concentrations of lithium ions [11–13]. Li + therapy is an important remedy against common mental disorders [14], and inhibition of some of the human IMPase family enzymes is believed to be responsible for these thera- peutic effects. We have proposed that the mobile loop is the determinant of sensitivity to Li + [9]. In this report we provide an interesting addition to this already intriguing protein family – the crystal structure of an IMPase (from the hyperthermophile T. maritima) that has a tetrameric arrangement and is also the most active IMPase known [6]. Additionally, a fragment of this IMPase (TM1415) shows unusual structural duality. The active site loop (residues 22–42) comes into the direct lattice contact with itself. In one subunit it has an a helical conformation, whereas its partner is in a b conformation. Because this fragment plays a crucial role in catalytic activity and may also be involved in Li + inhibition, the structural duality supports the idea by Dunker & Wright [15,16], that flexible protein fragments might be involved in devel- oping new functionalities in proteins. Results and Discussion Structure solution TM1415 was expressed in E. coli and purified as des- cribed previously [6]. The purified enzyme was crystal- lized in two crystal forms, characterized by space groups P2 1 and P2 1 2 1 2 1 . The structure was solved by molecular replacement (molrep) [17]. The structure of both crystal forms showed a tetrameric arrangement of this enzyme (Fig. 1). More specifically, both crystal forms contained a single tetramer in the asymmetric unit. The corresponding models were refined using cns [18] at 2.2 and 2.4 A ˚ resolution for forms 1 and 2, respectively. Refinement resulted in an R-factor of 0.209 and 0.178 (Rfree, 0.278 and 0.267) for forms 1 and 2, respectively (Table 1). Stereochemistry meas- ures, such as the Ramachandran plot and backbone and side-chain statistics, were well within the expected values, as indicated by procheck [19]. The electron density was very good throughout both models (sup- plementary Fig. S1) with the exception of loop regions, and in particular two catalytic loops (22–42,60–72) on one side of the tetramer in the second crystal form. The C-terminal region appeared to be much more mobile, as reflected by increased temperature factors. The model for crystal form 1 contains the terminal Lys256, whereas the model for crystal form 2 was ter- minated at Gly254. The difference in electron density suggested the pres- ence of multiple water molecules, as well as a single Mg 2+ in the active site of crystal form 1 (supplement- ary Fig. S1). This metal ion is found in a classical octahedral coordination with Asp79, Glu65 side-chain carboxyls, the carbonyl of Ile81 and three water mole- cules as direct ligands. In crystal form 2, the difference in electron densities suggested the presence of a single tartaric acid molecule per active site. The tetramer of crystal form 1 is quite symmetrical, whereas the tetramer of crystal form 2 is highly asymmetrical (Table 2). The refined models of both forms contain  300 water molecules. Monomer architecture: common family structure As with other enzymes in this family, including FBPases, the prote in has a well-preserved l ayered struc- ture of a and b secondary-structure elements (a-b-a-b-a) (Fig. 1, left panels), falling into two domains. The lower domain consists of a layer of two a-helices and an extensive seven-stranded b-sheet containing the act- ive site. The lower domain is reminiscent of the AMP- binding domain of eukaryotic FBPases (supplementary Fig. S2). The active site is located at the bulge of the C-terminal end of the third strand with a characteristic DPXD motif, where both aspartates participate in metal ion binding. The strand is followed by a two- turn a-helix that forms a phosphate-binding site. The second domain has a three-layer motif with two layers of a-helices flanking the second b-sheet. This domain corresponds to the FBP-binding domain of FBPases. The two helices of the first domain are connected by a mobile active site loop that previously was suggested to modulate activity by co-ordinating the third metal ion needed for catalysis [8,10]. This loop was also sug- gested to be involved in Li + inhibition [9]. The average temperature factors for both domains oscillate below 40 A ˚ 2 , with helices being much more ordered than the connecting loops. (See supplementary Fig. S3 for the temperature factor profiles of both models.) Structure of TM1415 IMPase K. A. Stieglitz et al. 2462 FEBS Journal 274 (2007) 2461–2469 ª 2007 The Authors Journal compilation ª 2007 FEBS Dimer architecture: similarity to human IMPase Despite being a bacterial enzyme and having signifi- cant sequence divergence, the overall architecture of the dimer is reminiscent of the human IMPase (Fig. 1, right panel). However, the dimer interface is quite dif- ferent from the human enzyme and other nonarchaeal IMPases. The superposition of the second subunit of the MJ0109 structure (1G0H) after superposition of the first subunit requires a rotation of  15° along the longest axis of the dimer in one direction, whereas superposition onto the second subunit of the human enzyme (1AWB) requires rotation of  15° in the opposite direction. The rotation around the axis that intersects the self-contacting residue, Ile173, is denoted as the dimer twist-angle. To superpose the second sub- unit on the structure of the eukaryotic FBPase takes a 30° rotation in the twist direction of the archaeal enzyme. In comparing the dimer orientation of all these IMPases, we see that the human IMPase has the extreme dimer twist-angle that is almost 45° off from the dimer position in FBPases, whereas the TM1415 and MJ0109 constitute intermediate twists spaced 15° off each other (central panel of Fig. 1). Tetramer architecture: similarity to FBPase TM1415, besides being the most active IMPase ( 100-fold more active than its archaeal, bacterial or eukaryotic counterparts) [6], is also a very efficient FBPase [7]. It is notable that the TM1415 structure represents a tetrameric organization for an IMPase (Fig. 1) reminiscent of the eukaryotic FBPase organ- ization. Each tetramer is organized as a dimer of dimers (Fig. 1). Despite the fact that TM1415 lacks the additional helix (FBPases have three helices at the tetramer interface), the helical layers in both enzymes directly interact. This dimer–dimer interface was shown to be crucial for the transition of the allosteric signal in pig kidney FBPase [20]. The rotation around the axis that relates dimers is termed the tetramer twist-angle. The TM1415 tetramer, despite having a Fig. 1. Schematic of the transformation of TM1415 into the eukaryotic FBPase. The transformation requires  25° rotation of the dimers around the tetramer axis and  30° rotation of the monomers in individual dimers. In the center there is a schematic of the mutual relationship of the monomers, in selected members of the family, as indi- cated by the twist angle of the dimer. TM1415 constitutes one of the intermedi- ates between the human IMPase and FBPase dimer organization. In the surround- ing panels, clockwise, starting at the top left, are the structure of the tetrameric TM1415, the superposition of TM1415 with the Homo sapiens (human) IMPase (in gold), the superposition of Sus scrofa (pig) FBPase with the MJ0109 (in blue), and the tetramer- ic pig FBPase. The secondary structure ele- ments are marked in red (helices) and yellow (b-sheets). K. A. Stieglitz et al. Structure of TM1415 IMPase FEBS Journal 274 (2007) 2461–2469 ª 2007 The Authors Journal compilation ª 2007 FEBS 2463 similar quaternary arrangement to eukaryotic FBPases, is more twisted, and the lower dimer requires  25° rotation to superpose with the corresponding dimer of the R-state FBPase. It can be compared with an over- rotated T state (the eukaryotic T state is rotated only 17° away from the R state, Fig. 1) [21]. As indicated in the central panel of Fig. 1, a rotation of  25° in tetr- amer twist, and an additional dimer twist in both dimers of  30°, are required to transform TM1415 into the FBPase. This nonregulated tetrameric organization of the T. maritima IMPase ⁄ FBPase might represent an evolu- tionary intermediate between dimeric IMPases and the allosterically regulated tetrameric eukaryotic FBPases. As a reminder, not all eukaryotic FBPases are alloster- ically regulated by AMP (e.g. chloroplast FBPase) [5]. However, the quaternary structures of these IMPase and FBPases suggest that quaternary organization may be critical in creating regulatory abilities. Crystal packing Both crystal forms of TM1415 have a tetramer present in the asymmetric unit with similar packing of individ- ual tetramers. However, in crystal form 2, the packing is not as tight as in crystal form 1. In crystal form 1 (P2 1 ), the solvent content is 44.3% and the tetramers have multiple contacts with their neighbors. Crystal packing contacts are numerous across the entire tetra- mer surface and include some weak contacts in the loop region (residues 22–42). The crystal contacts in form 1 leave sufficient room for unhindered packing of the loops in the b-hairpin conformation. In crystal form 2, the contacts are much more sparse, despite having almost the same solvent content (43.6%) as crystal form 1. Two important lattice Fig. 2. Schematic representation of the crystal packing in crystal form 2. Areas in contact are enclosed in the boxes. Boxes in broken lines represent symmetry-related contacts to those enclosed by the solid boxes. The contact on the left hand side comprises the active site loop (residues 22–42), shown in detail in Figs 3 and 4. At the other side of the tetramer, the active site loops are not in direct lattice contact, resulting in much higher temperature factors. Both loops in the upper dimer, regardless of the packing contact, are in a helical conformation, whereas the lower dimer loops are in an extended conformation. Fig. 3. The view of the active site of both superposed TM1415 models (crystals forms 1 and 2) and that of human IMPase (PDB code 1AWB). The model of the human enzyme (in light gray) has three metal ions bound (yellow spheres) and the substrate ( D-inosi- tol-1-phosphate, gray). The model of the TM1415 crystal form 1 is in purple and the crystal form 2 in green (including the tartrate molecule). The conformational change in loop 22–42 is indicated by green arrows, and the change in the 60s loop harboring Glu65–66 directly co-ordinating metal ions, is shown in red arrows. The con- tact of the active site loop (22–42) in the helical (in green, crystal form 2), as well as in the extended conformation (in purple, crystal form 1), to the symmetry mate in extended conformation is marked by a blue arrow. Structure of TM1415 IMPase K. A. Stieglitz et al. 2464 FEBS Journal 274 (2007) 2461–2469 ª 2007 The Authors Journal compilation ª 2007 FEBS contacts occur around t he loop region (residues 22–42). The loops on one side of the tetramer are in direct and intimate lattice contact (two direct hydrogen bonds), whereas the loops on the other side of the tetramer are practically unhindered, without direct contact (Fig. 2). Even though the crystal packing in both forms is dif- ferent, in form 2 there is still sufficient space around the loops so that the lattice could accommodate the tetramer with all loops in b-hairpin conformation,with- out significant structural changes (Fig. 3). Active site loops The direct comparison of both crystal forms provides unexpected insights into the principles of protein design and shows an unusual structural plasticity. The tetramer in crystal form 1 (P2 1 ) is very symmetrical (the rmsds between individual subunits are listed in Table 2), whereas the tetramer in the crystal form 2 is very asymmetrical. In the following discussion, we use the nomenclature suggested in Fig. 4B, which identifies the upper and lower dimers in the tetrameric structure. In crystal form 2, the lower dimer has the active site loops in the b-hairpin conformation, whereas in the upper dimer the loops are folded into two short a-heli- ces (Figs 3 and 4). The color brackets in Fig. 4 indicate the same fragment folded in different confor- mations in lower and upper dimers. Despite lacking any significant contacts, loops in the upper and lower dimers of crystal form 2 have distinct and different conformations (Fig. 4). The loop (resi- dues 62–72) shows significantly changed architecture from a more extended conformation (b-sheet in form 1) to the more coiled conformation in form 2. The act- ive site loop (residues 22–42), and the loop (residues 62–72) in TM1415 are required for the formation of the active site in IMPase and FBPase [8,10]. Figure 3 shows the superposition of both crystal forms of TM1415 and the human IMPase with a full comple- ment of metal ions and the substrate. The structures superpose well at the active site region and the figure shows the change in conformations of those two active site loops (green and red arrows). Both loops harbor the residues involved in metal ion binding and there- fore the subunits with alternative loop conformation would be incapable of binding the full set of metal ions. Additionally, residue Asp44, which plays a cru- cial role in the catalytic mechanism (creation of the nucleophile) in both upper subunits, has an alternative conformation in which it loses a direct contact to Thr84 Oc that is critical for water activation. All of these conformational changes make the upper dimer catalytically incompetent. The catalytic loops in crystal form 1 have tempera- ture factors slightly higher than, but comparable with, the rest of the structure (Table 1). The stability of the loops is confirmed with the quality of the electron A B Arg24B α β Arg24B Asp42B Asp42B Lys31B Lys31B Lys30C Asp42C Asp42C Arg24CArg24C Lys30C Fig. 4. (A) The stereo-diagram of the active site loop (residues 22– 42) in contact with the same sequence symmetry-related fragment (residues 22–42) covered with the sigma-weighted 2Fo-Fc electron density, contoured at the 1.1 r level. The helical (residues 22–42) fragment (blue model) is covered with the purple map, whereas the symmetry-related 22–42 fragment (purple model) is covered with a yellow map. The blue and red brackets denote the same sequence in both models. (B) Comparison of both crystal forms. The crystal form 1 is in purple and the crystal form 2 is in green. Please note significant displacement of residues 22–42 (loop 1) and 62–72 (loop 2) in the upper dimer. The fragments enclosed in boxes (with colors corresponding to the colors of the electron den- sity maps in (A) are in contact in the crystal form 2, as presented above. K. A. Stieglitz et al. Structure of TM1415 IMPase FEBS Journal 274 (2007) 2461–2469 ª 2007 The Authors Journal compilation ª 2007 FEBS 2465 density (Fig. 4 and supplementary Fig. S1). The tem- perature factors for the loops in direct contact in crys- tal form 2 (one in b- the second in a helical conformation) are slightly higher than, but comparable to, the rest of the structure; the remaining two loops have substantially higher temperature factors (supple- mentary Fig. S4) and much weaker electron densities. The symmetrical crystal form 1 has a single metal ion bound at the active site. This metal contributes to the stabilization of active site architecture. Choe et al. [10] concluded that this particular metal ion-binding site is the most frequently occupied and has the highest affinity to metal ions in FBPase. The metal ion does not make direct contact with either of the two mobile loops. In crystal form 2, we found no metal ions, but a single molecule of tartaric acid in the active site cav- ity. This ligand does not interact with any elements of the active site loop. Its location is somewhat reminis- cent of the sugar moiety of the substrate (Fig. 3). Therefore, it would be difficult to argue that the tar- trate or the metal ion is the cause of the asymmetry. The asymmetry must originate from two differently organized dimers present in solution that are dynamic- ally incorporated into the lattice during crystallization (supplementary Fig. S5). Evolutionary link The active site loop sequence (residues 22–42) is parti- ally similar to the human enzyme and partially similar to the sequences of M. jannaschii and A. fulgidus IMPases (Fig. 5). In Fig. 5 we present the alignment of the TM1415 sequence with sequences resulting from the BLAST searches and clustering of all 786 inde- pendent sequences into 13 functional families. Every sequence in Fig. 5 represents a single protein family that has distinct functionality. Figure 5 shows a small fragment of the alignment, including the mobile loop sequences. Judging by the clustering of sequences, TM1415 is closer to the human IMPase. The mobile loop sequence fits the environment less than other fragments (as measured by psqs, data not shown). The results of servers ranking this fragment propensity for order-disorder (disprot) [22] clearly suggest that this fragment has a tendency to be disordered. There- fore, this fragment has evolved to be mobile and less structured. The characteristic element for our sequence is that, when aligned structurally with other members of the family, the fragment has low sequence conservation (Fig. 5). It has variable length as well as amino acid content, suggesting that the fragment is rapidly evol- ving. Even when crucial positions are preserved, differ- ent residues fulfill the same structural role (Figs 4 and 5). The other fragment (160s loop), predicted to be unstructured (supplementary Fig. S4), participates in the assembly of the dimer and also of the tetramer. As Table 2. Comparison of the rmsds between different subunits within individual crystal forms (off-diagonal terms), as well as between corresponding subunits of both crystal forms (diagonal terms). Please note the symmetry of crystal form 1 and asymmetry of crystal form 2 represented by low and high values of the rmsd, respectively. Subunit a Crystal form 2 (P2 1 2 1 2 1 ) ABCD Crystal form 1 (P2 1 ) A 0.931 0.739 3.937 3.335 B 0.368 0.498 3.705 3.133 C 0.274 0.370 3.175 2.253 D 0.402 0.381 0.534 2.841 a The rmsds were calculated on the 252 Ca positions of individual subunits. Table 1. Data collection and refinement summary. Crystal form 12 Space group P2 1 P2 1 2 1 2 1 dmin (A ˚ ) 2.2 2.4 Reflections unique 46 203 36 088 <I ⁄ r(I)> a 14.1 (2.1) 10.1 (1.1) Completeness (%) a 98.3 (95.5) 98.0 (96.5) Average redundancy a 3.5 (2.2) 2.5 (2.3) Unit cell (A ˚ ) a ¼ 62.02 a ¼ 83.81 b ¼ 103.87 b ¼ 97.94 c ¼ 80.51 c ¼ 122.16 (deg) b ¼ 101.83 R merge 0.058 (0.41) 0.063 (0.55) R-factor 0.209 0.178 Free R-factor (5%) 0.279 0.266 rmsd Bond lengths (A ˚ ) 0.007 0.006 Bond angles (A ˚ ) 1.37 1.25 Ramachandran plot (%) Most preferred 89 89 Allowed 10 10 Generously allowed 1 1 Disallowed 0 0 Ligands and solvent Ligands None 4 tartaric acid Metal ions 4 None Water molecules 303 299 Temperature factors (A ˚ ) Protein average 45.3 40.5 Ligands ⁄ metals 33.2 52.4 Loops 55.0 65 ⁄ 85 2 a Last shell data in parenthesis (2.2–2.27 A ˚ , crystal form 1; 2.4– 2.49 A ˚ , crystal form 2). b Loops in contact ⁄ loops not in contact. Structure of TM1415 IMPase K. A. Stieglitz et al. 2466 FEBS Journal 274 (2007) 2461–2469 ª 2007 The Authors Journal compilation ª 2007 FEBS presented above, the change in the aggregation state contributed to a change in enzyme functionality. The inhibition, by Li + , of several members of the family, must also be considered as an acquisition of a novel functionality. In summary, the low sequence conserva- tion and high variability of both loops when linked to their structural instability suggests evolution of novel functionalities being driven by changes in sequence of those fragments [15,16,23]. Experimental procedures Cloning and protein production The TM1415 protein was expressed in E. coli, as described previously [6]. After cell lysis and removal of the cell debris by centrifugation, the supernatant containing TM1415 was heated to 85 °C and the denatured E. coli proteins were removed by centrifugation. The enzyme was further purified by column chromatography on a Sepharose QFF column. The protein was > 96% pure, as measured by SDS ⁄ PAGE. Crystallography The protein was crystallized by the standing drop vapor diffusion method. Five microlitres of the enzyme, in 50 mm Tris buffer, pH 7.5, was mixed with 5 lL of the crystalliza- tion buffer containing 15% (w ⁄ v) PEG 4000, 15% (v ⁄ v) PEG 400, 0.2 m MgSO 4 , and 50 mm (NH 4 ) 2 SO 4 ,in50mm cacodylate buffer, pH 8. The crystals appeared in 1 week and had the morphology of chunky prisms with a lattice organized by P2 1 symmetry. Mixing 5 lL of the enzyme in 50 mm Tris, pH 7.5, with 5 lL of the crystallization buffer containing 15% (v ⁄ v) PEG 400 0.2 m sodium tartrate, and 0.2 m (NH 4 ) 2 SO 4 in 50 mm citrate, pH 5.6, resulted in crys- tals with the morphology of thin plates characterized by the P2 1 2 1 2 1 space group. The data were collected to 2.2 A ˚ reso- lution for the first crystal form, and to 2.4 A ˚ for the second form, on the RaxisIV++ area detector mounted on the Rigaku rotating anode at the X-ray facility at Rice Univer- sity (Houston, TX, USA). The structure was solved by molecular replacement (molrep), using the model construc- ted from the structures of several IMPases (human, MJ0109 and AF2372) as search probes. The sequence of TM1415 shares the highest similarity to human IMPase ( 26% identity) and therefore this molecule was used as a scaffold. However, there are three deletions (residues 70s, 160s and 210s), making it a bit shorter. Those crucial dele- tions are shared with MJ0109 and AF2372. Therefore, we used those molecules to construct the fragments not corres- ponding well to the human structure. The probe model was built from fragments of those proteins with the highest homology to TM1415. The full atom model was used for molecular replacement and later refined, using cns,toa standard R-factor of 0.209 and 0.178, as calculated on all reflections to 2.2 and 2.4 A ˚ resolution, respectively, for forms 1 and 2. The data collection and refinement statistics can be found in Table 1. Sequence versus structure analysis The iterated BLAST search was used to establish the mem- bers of the multifunctional IMPase family. The groupings followed the CDD classification of the NCBI (http:// www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=cdd). The seq- uence to 3D structure fit was estimated by the program psqs, developed at the Burnham Institute by A. Godzik. Predictions of the disorder were performed at the web ser- ver of the disprot (Dunker) [21]. We also carried out a clustering experiment in which we investigated the entire IMPase ⁄ FBPase family. The PFAM clan ‘inositol polyphosphate 1 phosphatase like superfamily’ includes three PFAM domain families (a) Inositol_P(PF00459), (b) FBPase(PF00316) and (c) FBPase_glpX(PF03320). The Inositol_P family has 65 sequences in its seed alignment, representing 586 proteins; and FBPase and FBPase_glpX each has 14 and 6 seed sequences, representing 233 and 101 proteins. Fig. 5. Multisequence alignment of the loop fragments of all the family representatives, including T. maritima and Homo sapiens. Amino acids implicated in the catalytic mechanism are on a blue background. The green and the light blue background mark conserved residues. The secondary structure, as calculated by DSSP [26] is marked under the sequences, with red for the a-helix and orange for the b-sheet. The figure was prepared using the program, VISSA [27]. K. A. Stieglitz et al. Structure of TM1415 IMPase FEBS Journal 274 (2007) 2461–2469 ª 2007 The Authors Journal compilation ª 2007 FEBS 2467 All these 85 seed sequences were clustered into sub- families at 25% sequence similarity and one representative sequence was selected for each subfamily. Within each clus- ter, all the sequences were similar to the representative at 25% similarity or more, but similarities between all the rep- resentatives of different clusters were below 25%. The clus- tering was performed with program cd-hit [24]. These 85 sequences were clustered into 13 subfamilies. Three families represented singletons, and another 10 sub- families contained 2–22 members. Five subfamilies con- tained proteins with 3D structures deposited in PDB. The average identity within each cluster was 35%. Figure 4 shows the alignment of TM1415 (which belongs to the IMPA1_HUMAN cluster) with representative proteins from each cluster. The initial alignment was made using clustalw [25] and was edited according to the structural alignment of five present in the PDB proteins. Acknowledgements We would like to thank Dr Andrey Bobkov for help in protein production and gel filtration experiments. This work was supported by the NIH grant 1R01 G64481 (to BS), National Science Foundation grant MCB-9978250 (to MFR), and Department of Energy Biosciences DE-FG02–91ER20025 (to MFR). References 1 Zhang Y, Liang JY & Lipscomb WN (1993) Structural similarities between fructose-1,6-bisphosphatase and ino- sitol monophosphatase. Biochem Biophys Res Commun 190, 1080–1083. 2 York JD, Ponder JW & Majerus PW (1995) Definition of a metal-dependent ⁄ Li(+)-inhibited phosphomono- esterase protein family based upon a conserved three- dimensional core structure. Proc Natl Acad Sci USA 92, 5149–5153. 3 Chen L & Roberts MF (2000) Overexpression, purifica- tion and analysis of complementation behavior of E. coli SuhB protein: Comparison with bacterial and archaeal inositol monophosphatases. Biochemistry 39, 4145–4153. 4 Nimmo HG & Tipton KF (1975) The allosteric proper- ties of beef-liver fructose bisphosphatase. Eur J Biochem 58, 575–585. 5 Villeret V, Huang S, Zhang Y, Xue Y & Lipscomb WN (1995) Crystal structure of spinach chloroplast fructose- 1,6-bisphosphatase at 2.8 A ˚ resolution. Biochemistry 34, 4299–4306. 6 Chen L & Roberts MF (1999) Characterization of a tetrameric inositol monophosphatase from the hyperthermophilic bacterium Thermotoga maritima. Appl Environ Microbiol 65, 4559–4567. 7 Stec B, Yang H, Johnson KA, Chen L & Roberts MF (2000) MJ0109 is an enzyme that is both an inositol monophosphatase and the ‘missing’ archaeal fructose-1, 6-bisphosphatase. Nat Struct Biol 7, 1046–1050. 8 Johnson KA, Chen L, Yang H, Roberts MF & Stec B (2001) Crystal structure and catalytic mechanism of the MJ0109 gene product: a bifunctional enzyme with inosi- tol monophosphatase and fructose 1,6-bisphosphatase activities. Biochemistry 40, 618–630. 9 Stieglitz KA, Johnson KA, Yang H, Roberts MF, Seaton BA, Head JF & Stec B (2002) Crystal structure of a dual activity IMPase ⁄ FBPase (AF2372) from Archaeoglobus fulgidus. The story of a mobile loop. J Biol Chem 277, 22863–22874. 10 Choe JY, Fromm HJ & Honzatko RB (2000) Crystal structures of fructose 1,6-bisphosphatase: mechanism of catalysis and allosteric inhibition revealed in product complexes. Biochemistry 39, 8565–8574. 11 Hallcher LM & Sherman WR (1980) The effects of lithium ion and other agents on the activity of myo-ino- sitol-1-phosphatase from bovine brain. J Biol Chem 255, 10896–10901. 12 Bone R, Springer JP & Atack JR (1992) Structure of inositol monophosphatase, the putative target of lithium therapy. Proc Natl Acad Sci USA 89, 10031–10035. 13 Patel S, Yenush L, Rodriguez PL, Serrano R & Blundell TL (2002) Crystal structure of an enzyme displaying both inositol-polyphosphate-1-phosphatase and 3¢-phos- phoadenosine-5¢-phosphate phosphatase activities: a novel target of lithium therapy. J Mol Biol 315, 677– 685. 14 Fieve RR (1999) Lithium therapy at the millennium: a revolutionary drug used for 50 years faces competing options and possible demise. Bipolar Disord 1 , 67–70. 15 Dunker AK, Brown CJ, Lawson JD, Iakoucheva LM & Obradovic Z (2002) Intrinsic disordered and protein function. Biochemistry 41 , 6573–6582. 16 Dyson HJ & Wright PE (2005) Intrinsically unstruc- tured proteins and their functions. Nat Rev Mol Cell Biol 6, 197–208. 17 Vagin A & Teplyakov A (1997) MOLREP: an auto- mated program for molecular replacement. J Appl Cryst 30, 1022–1025. 18 Brunger AT, Adams PD, Clore GM, DeLano WL, Gros P, Grosse-Kunstleve RW, Jiang JS, Kuszewski J, Nilges M, Pannu NS et al. (1998) Crystallography & NMR system: a new software suite for macromolecular struc- ture determination. Acta Crystallogr D 54, 905–921. 19 Laskowski RA, MacArthur MW, Moss DS & Thornton JM (1993) PROCHECK: a program to check the stereo- chemical quality of protein structures. J Appl Cryst 26, 283–291. 20 Lu G, Stec B, Giroux EL & Kantrowitz ER (1996) Evidence for an active T-state pig kidney fructose Structure of TM1415 IMPase K. A. Stieglitz et al. 2468 FEBS Journal 274 (2007) 2461–2469 ª 2007 The Authors Journal compilation ª 2007 FEBS 1,6-bisphosphatase: interface residue Lys-42 is important for allosteric inhibition and AMP cooperativity. Protein Sci 5, 2333–2342. 21 Zhang Y, Liang J-Y, Huang S & Lipscomb WN (1994) Toward a mechanism for the allosteric transition of pig kidney fructose-1,6-bisphosphatase. J Mol Biol 244, 609–624. 22 Dunker AK, Lawson JD, Brown CJ, Williams RM, Romero P, Oh JS, Oldfield CJ, Campen AM, Ratliff CM, Hipps KW et al. (2001) Intrinsically disordered protein. J Mol Graph Model 19, 26–59. 23 Siebert X & Amzel LM (2004) Loss of translational entropy in molecular associations. Proteins: Struct Funct Genet 54, 104–115. 24 Li W, Jaroszewski L & Godzik A (2001) Clustering of highly homologous sequences to reduce the size of large protein databases. Bioinformatics 17, 282–283. 25 Thompson JD, Higgins DG & Gibson TJ (1994) clustal w: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res 22, 4673–4680. 26 Kabsch W & Sander C (1983) Dictionary of protein sec- ondary structure: pattern recognition of hydrogen- bonded and geometrical features. Biopolymers 22, 2577–2637. 27 Li W & Godzik A (2006) VISSA: a program to visualize structural features from structure sequence alignment. Bioinformatics 22, 887–888. Supplementary material The following supplementary material is available online: Fig. S1. Sigma-weighted 2Fo-Fc electron density, con- toured at the 1.1 r level covering the active site loop (residues 22-42), the active site residues (65-66, 79-84) and the ligands bound. Fig. S2. Ribbon representation of a single monomer of the TM1415 (left) and the pig FBPase (right). Fig. S3. Temperature factor plots for both crystal forms. Fig. S4. Diagram representing the disorder-score, as calculated by the program disprot (Dunker) [22] . Fig. S5. The chromatogram obtained from gel filtra- tion experiments carried out in citrate buffer (pH 5.6) on a Superdex 200 column. This material is available as part of the online article from http://www.blackwell-synergy.com Please note: Blackwell Publishing is not responsible for the content or functionality of any supplementary materials supplied by the authors. Any queries (other than missing material) should be directed to the corres- ponding author for the article. K. A. Stieglitz et al. Structure of TM1415 IMPase FEBS Journal 274 (2007) 2461–2469 ª 2007 The Authors Journal compilation ª 2007 FEBS 2469 . Crystal structure of the tetrameric inositol 1-phosphate phosphatase (TM1415) from the hyperthermophile, Thermotoga maritima Kimberly. 2007) doi:10.1111/j.1742-4658.2007.05779.x The structure of the first tetrameric inositol monophosphatase (IMPase) has been solved. This enzyme, from the eubacterium Thermotoga maritima, similarly