- Email: [email protected]
The 1.9 Å Crystal Structure of Heat-labile Shrimp Alkaline Phosphatase
doi: 10.1016/S0022-2836(02)00035-9 available online at http://www.idealibrary.com on
w B
J. Mol. Biol. (2002) 318, 1265–1274
˚ Crystal Structure of Heat-labile Shrimp The 1.9 A Alkaline Phosphatase Maaike de Backer1, Sean McSweeney1, Hanne B. Rasmussen2 Bjørn W. Riise3, Peter Lindley1 and Edward Hough3* 1
European Synchrotron Radiation Facility, BP 220 F-38043 Grenoble Cedex France 2 Novo Nordisk A/S, Novo Alle 6B2.90, DK-2880 Bagsvaerd, Denmark 3
Department of Chemistry Faculty of Science, University of Tromsø N 9037 Tromso, Norway
Alkaline phosphatases are non-specific phosphomonoesterases that are distributed widely in species ranging from bacteria to man. This study has concentrated on the tissue-nonspecific alkaline phosphatase from arctic shrimps (shrimp alkaline phosphatase, SAP). Originating from a cold-active species, SAP is thermolabile and is used widely in vitro, e.g. to dephosphorylate DNA or dNTPs, since it can be inactivated by a short rise in temperature. Since alkaline phosphatases are zinc-containing enzymes, a multiwavelength anomalous dispersion (MAD) experiment was performed on the zinc K edge, which led to the determination of the structure to a resolution ˚ . Anomalous data clearly showed the presence of a zinc triad in of 1.9 A the active site, whereas alkaline phosphatases usually contain two zinc and one magnesium ion per monomer. SAP shares the core, an extended b-sheet flanked by a-helices, and a metal triad with the currently known alkaline phosphatase structures (Escherichia coli structures and a human placental structure). Although SAP lacks some features specific for the mammalian enzyme, their backbones are very similar and may therefore be typical for other higher organisms. Furthermore, SAP possesses a striking feature that the other structures lack: surface potential representations show that the enzyme’s net charge of 2 80 is distributed such that the surface is predominantly negatively charged, except for the positively charged active site. The negatively charged substrate must therefore be directed strongly towards the active site. It is generally accepted that optimization of the electrostatics is one of the characteristics related to cold-adaptation. SAP demonstrates this principle very clearly. q 2002 Elsevier Science Ltd. All rights reserved
*Corresponding author
Keywords: structure; X-ray crystallography; phosphatase; metalloprotein; cold-adaptation
Introduction Alkaline phosphatases (AP, EC 3.1.3.1) are homodimeric metalloenzymes that catalyze the hydrolysis and, in the presence of a phosphate acceptor, transphosphorylation of a wide variety of phosphate monoesters. The enzymatic reaction proceeds through a covalent phosphoseryl interThis article is dedicated to the memory of Bjørn W. Riise Abbreviations used: AP, alkaline phosphatase; SAP, shrimp AP; ECAP, Escherichia coli AP; PLAP, human placental AP; MAD, multiwavelength anomalous dispersion. E-mail address of the corresponding author: [email protected]
mediate to produce inorganic phosphate or to transfer the phosphoryl group to alcohols.1 Substrates include proteins, DNA and small organic molecules. The product, inorganic phosphate, is also an inhibitor. The phosphatases are abundant in prokaryotic and eukaryotic species, where they often occur in several isoforms. In humans, tissuenonspecific genes are widely expressed, in contrast to the localized, tissue-specific expression of the placental, intestinal and germ-cell genes.2 Although their physiological roles are not clear, alkaline phosphatases are probably involved in the mineralization of bone and, for Escherichia coli, in the acquisition of phosphate. They are utilized as clinical diagnostic tools for various diseases; e.g. deficient activity of the tissue-nonspecific
0022-2836/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved
˚ Crystal Structure of Heat-labile Shrimp Alkaline Phosphatase 1.9 A
1266
Table 1. Residues that constitute the metal-binding sites P. borealis
H. sapiens
E. coli
1 (Zn)
Asp315 His319 His432
2 (Zn)
Asp37 Ser86 Asp356 His357 Asp37 His149 Thr151 Glu310 Water
Asp316 His320 His432 PO32 4 Asp42 Ser92 Asp357 His358 Asp42 (His153) Ser155 Glu311
Asp327 His331 His412 PO32 4 Asp51 Ser102 Asp369 His370 Asp51 (Asp153) Thr155 Glu322 PO32 4
Metal-binding site
3 (Mg/Zn)
Ligands for the metal triad of shrimp (P. borealis ), human placental (Homo sapiens ) and E. coli AP, for PDB entries 1EW2 and 1ALK for the human and the bacterial enzyme, respectively. Differences in residue type are underlined. The residues corresponding to SAP-His149 are in parentheses, since they are not liganding the metals.
isozyme is the biochemical hallmark of hypophosphatasia (bone disorder).2 In addition, various tests relate alkaline phosphatase levels to lesions (including tumours), liver disease, vitamin D deficiency, etc.3 – 8 There is a long history of biochemical and biophysical studies on alkaline phosphatases, mainly concentrated on the E. coli enzyme (ECAP). It was shown decades ago that the active site of each monomer contains three metal-binding sites, which are usually occupied by two zinc and one magnesium cation.9 The magnesium-binding site may be occupied by zinc. The metal ions constitute a catalytic triad similar to that in phospholipase C from Bacillus cereus10 and P1 nuclease from Penicillium citrinum.11 A detailed hypothesis about the reaction mechanism has been formulated for the E. coli enzyme, which involves a double in-line displacement and the participation of all three metal ions,1,12 – 18 which has been reviewed by Holtz & Kantrowitz19 and updated by Stec et al.,12 In E. coli, the zinc ions stabilize the activated state of the enzyme and interact with the substrate and leaving group. The magnesium ion in the third metal-binding site plays a structural role and is involved in catalysis. It positions a hydroxide ion, a general base, which accepts a proton from the catalytic serine residue before the first nucleophilic attack on the phosphorus atom. Although this hypothesis is based on structural and biochemical data from the E. coli enzyme, it is assumed to be generally valid, since alkaline phosphatases have a reasonable cross-species sequence similarity. There must, however, be subtle differences, since the mammalian phosphatases are more efficient than their prokaryotic counterparts. The structures of alkaline phosphatase from two species have been determined. The extensively studied ECAP structure was solved initially in ˚ ,9 was later re-deter1973 to a resolution of 7.7 A mined to a higher resolution20,21 and has been
˚ with entry code 1ALK in the deposited at 2.0 A RCSB Protein Data Bank.1 Since then, several structures with various mutations and substrates have been deposited. Recently, the first mammalian ˚ : human placental structure was solved at 1.8 A alkaline phosphatase, PLAP, with PDB entry code 1EW2.22 The major functional differences between mammalian enzymes and their homologues from lower organisms are the improved catalytic activity, reduced heat stability and a shifted pH optimum (to a higher pH in a pH versus activity profile). Sequence alignments between E. coli and mammalian alkaline phosphatases show a 25– 30% identity. Here, we describe the structure of the thermolabile tissue-nonspecific Pandalus borealis (Arctic shrimp) alkaline phosphatase (SAP). The enzyme is used extensively in vitro to dephosphorylate DNA or dNTPs, and can be inactivated by a short rise in temperature to 65 8C. The sequence of this protein has been published.23 The sequences of SAP and ECAP24 have 24% identical, 16% strongly similar and 11% weakly similar residues. SAP and PLAP25 share 41% of their residues, 19% strongly similar and 12% weakly similar. The activesite residues are conserved among these three sequences, except for three residues; Asp153 and Lys328 are histidines in SAP (His149 and His316) and PLAP (His153 and His317), and PLAP has a T155S substitution (ECAP-Thr155, SAP-Thr151, PLAP-Ser155; see Table 1 for an overview). By solving the structure of SAP, we hope to identify structural differences between heat-labile SAP, bacterial ECAP and mammalian PLAP, which explain the differences in catalytic efficiency. Cold-Adaptation The SAP sample that was used for this study was obtained from shrimps living in the Barents Sea off the coast of Northern Norway, where the annual mean water temperature is 5 8C. Species that live in such environments must somehow be adapted to survive at these reduced temperatures. A lower temperature has a number of implications for the functioning of proteins. A number of physiological changes, including the solubility of salts, gases and small organic molecules, affect the charge distribution of the protein directly and therewith its solubility. The increased viscosity of water, which reduces the diffusion rates of substrate and product, leads to lower reaction rates. Reduced thermal motions affect the enzyme’s overall stability, kinetics and ligand binding. A number of studies have been devoted to the relation between cold-adaptation and structural characteristics. However, the number of cold-active enzymes where the structure is known is small. Until now, only seven structures have been solved: Antarctic bacterial citrate synthase,26 cod pepsin,27 salmon trypsin,28 salmon elastase,29 Vibrio marinus triose-phosphate isomerase,30 Alteromonas haloplanctis a-amylase,31 Aquaspirillium malate dehydrogenase.32
1.9 A˚ Crystal Structure of Heat-labile Shrimp Alkaline Phosphatase
1267
Figure 2. Superposition of the Ca-traces of SAP (in yellow) with the E. coli (red) and human placental (blue) structures, showing the conservation of the core (the bsheet and flanking a-helices) and the presence of a crown domain for PLAP and SAP. The structures were superimposed with O59 and prepared with Quanta (Accelrys).
Figure 1. The structure of the functional dimer (monomer A in green, monomer B in red) showing the large extended b-sheets (in white for both monomers), flanked by a-helices, the crown domain (on top of the b-sheet, coloured light green for monomer A), the N-terminal helix (light green for monomer A) and the zinc triad (represented by yellow van der Waals spheres) located on top of the b-sheet, next to the crown domain. (a) A front view, and (b) a top view of the structure. The Figure was prepared with Quanta (Accelrys).
In general, cold-active enzymes have an increased catalytic efficiency at lower temperature, which seems to be accompanied by a reduced thermal stability and increased molecular flexibility. The flexibility should prevent the enzymes from becoming overly rigid at low temperatures. There are, however, no common structural features that account for these findings; each cold-adapted enzyme (family) uses a different array of structural adjustments to gain the flexibility that results in increased catalytic efficiency and reduced stability. Apparently, the origin of increased activity and reduced thermal stability resides in particular regions or domains rather than general characteristics.33 Factors affecting protein stability include hydrophobic interactions (which are weakened
at lower temperatures), electrostatic interactions (which are stabilized at decreasing temperatures) and reduced content of secondary structure. While the enzyme as a whole is more flexible, it should still maintain a good binding-site geometry for the ligand. Moreover, cold-active enzymes seem to optimize their catalytic activity by optimizing the electrostatics at and around their binding sites.33 Cold-active APs from Atlantic cod,34,35 Marine Vibrio sp.36, Arctic strain TAB537 and SAP23 have been studied. This study investigates to what extent structural features that indicate adaptation to cold can be identified in SAP.
Results The overall structure Shrimp alkaline phosphatase is a homodimer ˚ £ 65 A ˚ £ 50 A ˚ with dimensions of about 95 A (Figure 1). The two monomers in the working asymmetric unit do not represent the functional dimer; which is formed by one monomer and its symmetry-related partner, as is evident from the packing in the unit cell and by comparison with ECAP and PLAP (Figure 2). The core of the monomer is composed of a ten-stranded b-sheet, sandwiched between a set of helices of various lengths, three on one side, and five on the other side of the sheet. The b-strands are parallel, except for one anti-parallel insertion. In the functional dimer, the monomers associate such that their individual b-sheets form one extended sheet. On top of the dimeric sheet, perpendicular to the strands, a small domain is situated (360 –430, referred to as the “crown domain” in analogy
1268
˚ Crystal Structure of Heat-labile Shrimp Alkaline Phosphatase 1.9 A
Figure 3. Details of metal-binding sites of SAP for (a) Zn1, (b) Zn2 and (c) Zn3, showing the zinc ions as yellow van der Waals spheres and the liganding residues (which are listed in Table 1). A 2Fo 2 Fc map is plotted in blue at a level of 1.0s and an Fo 2 Fc map is plotted in green at þ3s. Note the unassigned density adjacent to Zn1. This Figure was prepared with Bobscript.60
with PLAP), which is composed of six small bstrands. As this domain is part of the dimer interface, each monomer contributes three small bstrands (360 –364, 386– 391, and 419– 421) to form a three-stranded double layer with the other monomer. One of the connecting loops points outwards to the solvent; others embrace the crown domain or interact with the core. The N-terminal helix points away from the core of the monomer. In the functional dimer, this helix embraces its neighbouring monomer and is part of the dimer interface. Two disulphide bridges are present per monomer; Cys115, located near a glycosylation site at the tip of the b-sheet, bridges with Cys180, which is part of a loop connecting two small helices that are flanking the b-sheet. The second bridge is formed between Cys467 and Cys475, within a small loop at the C terminus. Although the sequence contains the glycosylation motif N-X-T/S twice, extended density was found at only one of these sites, Asn114, located at the tip of the large b-sheet. No significant residual density was found at the second potential glycosylation site, Asn166. Although one N-acetyl-glucosamine molecule could easily be modelled near Asn114, a second sugar molecule was harder to fit. There was clearly 2Fo 2 Fc (1 to 2s ) as well as Fo 2 Fc (higher than 3s ) density for a second ring, although it was clear that the second sugar molecule could not form a 1 –4 glycosidic bond, as is expected for a second N-acetyl-glucosamine linkage. Although alternative sugars would fit the density well, they were not included in the final model, since there is uncertainty about the nature of the sugar moieties. The active site Anomalous scattering data clearly showed that the three metal ions in the active site are all zinc, although the third zinc site had a somewhat lower occupancy. The occupancies have been refined with SHELXL38 to 0.97, 0.97 and 0.88 for respectively, Zn1, Zn2 and Zn3 of monomer A and to 0.95, 0.92 and 0.88 for monomer B. The zinc triad, which is located near the carboxyl top of the b-sheet, is readily accessible for substrates. Since there is only space to accommodate the phosphate
group in the active site, and there is no apparent binding site for the leaving group of the substrate, the location and shape of the active site explain the non-specificity of the enzyme. Unassigned 3s density in an Fo 2 Fc map (Figure 3), associated with Zn1 and Zn2 in the active sites of both monomers, could not be interpreted unambiguously. Fitting with the most likely small molecules, Tris, sulfate, phosphate and carbonate, was attempted, but the results after refinement were not convincing. Therefore, no ligand was included during refinement of the final structure. The three metal ions are relatively close, with distances between Zn1 and Zn2 and between Zn2 ˚ , whereas Zn1 and Zn3 are 7.1 A ˚ and Zn3 of 4.6 A apart. Zn1 interacts with the imidazole nitrogen atoms of His319 and His432, with one carboxyl oxygen atom of Asp315. Zn2 has four ligands: a carboxyl oxygen atom of Asp37 and of Asp356, an imidazole nitrogen atom of His357 and the hydroxyl oxygen atom of Ser86. Zn3 binds to the second carboxyl oxygen atom of Asp37, to an imidazole nitrogen atom of His149, the hydroxyl group of Thr151, to Glu310 and to a water molecule. The ligands of the metals are listed in Table 1.
Model statistics The final model converged to an R-factor of 19.91% for the working set and a free R-value of 22.75% for a randomly chosen test set of 5% of the reflections, with a highest-resolution shell of 1.99 – ˚ . It contains 8108 non-hydrogen protein 1.92 A atoms; 476 amino acid residues per monomer, three zinc ions, one N-acetyl-glucosamine unit, two sulfate groups and one maleic acid molecule per monomer. A stereochemical check by PROCHECK39 shows root-mean-squared (r.m.s.) ˚ in deviations from ideal geometry of 0.028 A covalent bond lengths and 2.08 in bond angles. A Ramachandran plot shows that 89.2% of the residues lie in the most favoured regions, 10.3% in the additional allowed regions, 0.2% in the generously allowed regions and 0.2% in the disallowed regions.
1.9 A˚ Crystal Structure of Heat-labile Shrimp Alkaline Phosphatase
1269
Coordination of the zinc ions in the metal triad
Figure 4. Superposition of SAP (coloured by atom) and PLAP (yellow), showing Arg162 in the active site, which has a different conformation in the two structures. The Figure was prepared with MOLSCRIPT.
Discussion Comparison between SAP and other AP structures The overall structures of SAP, ECAP and PLAP are similar, with the core and active-site architecture preserved (Figure 2). Moreover, the backbones of SAP and PLAP superimpose with an ˚ for 455 Ca atoms, whereas the r.m.s.d. of 1.07 A superposition of SAP with 1ALK results in an ˚ for 343 Ca atoms. SAP and PLAP r.m.s.d. of 1.33 A share the crown and N-terminal domains, which are probably present in all mammalian enzymes. PLAP, however, contains a fourth metal-binding site, where a magnesium ion is bound.22 Although SAP shares some of the residues that comprise this binding site, no extra metal ion was detected. In the E. coli structure 1ALK, phosphate is bound in the active site with two of the phosphate oxygen atoms coordinated by Zn1 and Zn2, and the other two oxygen atoms bound by the guanidinium group of Arg166, which in turn is hydrogen bonded to Asp101 and a water molecule. The phosphate group is further bound to the amide group of Ser102 and a water molecule. The corresponding arginine residue in SAP (Arg162) is oriented differently; it points away from the metal ions (Figure 4). It has been shown that ECAPArg166 is primarily responsible for substrate binding, as it interacts directly with the substrate in the non-covalent enzyme – phosphate complex.1 Modification of the arginine residue leads to a lower substrate affinity, but not to a significant decrease of the activity.40,41
Although the zinc atoms in SAP are coordinated to ECAP and PLAP in a similar way, substitutions in the metal-binding sites seem to influence metal affinity. Zn1 in SAP is held by an aspartate and two histidine residues, like the other known structures. There is a possibility that residual density close to the zinc ions could be a further but unidentified ligand (see above). In 1ALK and PLAP the phosphate contributes two oxygen atoms, resulting in a penta-coordinated zinc ion. Zn2 has four ligands in SAP: two aspartate residues, a histidine residue, and the nucleophilic serine residue; in both ECAP and PLAP, Zn2 is also coordinated tetrahedrally by two aspartate residues, a histidine residue and an oxygen atom of the phosphate inhibitor. While SAP-Zn3 binds to an aspartate, a threonine residue, a glutamate residue, a histidine residue and a water molecule, the corresponding metal sites in ECAP and PLAP are occupied by a magnesium ion, which is coordinated octahedrally by an aspartate residue, a glutamine residue, a threonine residue (ECAP) or a serine residue (PLAP) and three water molecules. APs usually contain a magnesium ion in the third metal-binding site, while a zinc ion is observed in SAP. The histidine residue that ligates Zn3 in SAP (H149) replaces an aspartate residue in E. coli (D153), which does not interact with the metal ion directly. Since histidine is a common ligand for zinc, the third metal-binding site of SAP probably has a higher affinity for zinc than the corresponding site in ECAP. A mutant E. coli enzyme with the aspartate residue substituted with histidine also contained zinc at this site.42 Sequence alignments show that this histidine residue is conserved in mammalian sequences, including PLAP, although a magnesium ion is reported to occupy this site.22 Indeed, the structure of PLAP shows that this histidine residue is too far away from the zinc ion to function as a ligand. Biochemical data suggests that a fully zincoccupied enzyme is less active. In fact, a number of mammalian enzymes need the addition of magnesium in order to achieve maximal activity.43 Since this effect is time-dependent and independent of the concentration of magnesium, a conformational change has been proposed for PLAP44 and similar effects have been reported recently for AP from Scylla serrata.45 This probably involves the exchange of zinc with magnesium, which leads to a more open conformation of the active site, since magnesium is coordinated octahedrally. This effect has been observed for E. coli mutants D153H and for a double mutant D153H/K328H, which mimic mammalian enzymes.42,46 Stec and co-workers12 propose that a magnesium-coordinated hydroxide ion functions as a general base and accepts a proton from the nucleophilic serine residue. The activated serine residue is stabilized by coordination to Zn2. A metal ion other than magnesium (i.e. zinc) would
˚ Crystal Structure of Heat-labile Shrimp Alkaline Phosphatase 1.9 A
1270
Table 2. Selected number of active-site residues of alkaline phosphatase structures Structure SAP PLAP ECAP_DM ECAP_P ECAP_woP ECAP_VO3
Comment
PDB entry
Residue
Residue
Residue
– – D153H, K328H Phosphate present Without phosphate VO22 3 present
1K7H 1EW2 1ANI 1ALK 1ED8 1B8J
His149 His153 His153 Asp153 Asp153 Asp153
His316 His328 His328 Lys328 Lys328 Lys328
Ser86 Ser102 Ser102 Ser102 Ser102 Ser102
Active-site residues of SAP (P. borealis ), PLAP (H. sapiens ), 1ANI (E. coli; double mutant D153H and K358H), 1ALK (E. coli; containing phosphate), 1ED8 (E. coli; no phosphate present), and 1BJ8 (E. coli; containing vanadate).
distort the local geometry and therewith the action of the enzyme. It is somewhat surprising that SAP is fully occupied by zinc, since both Zn2þ and Mg2þ were present in the crystallization buffer. There should be enough time to exchange metals during the crystallization process. After dissolving the crystals in a solution containing both zinc and magnesium, the activity of SAP increases. This may be explained by the exchange of zinc for magnesium in the third metal-binding site. The presence of zinc in the SAP structure can be explained by a higher affinity for zinc than for magnesium in the third metal-binding site, or by a more stable conformation when fully occupied with zinc. We have therefore probably crystallized the enzyme in a low-activity state.
Comparison of active sites The active site of SAP has been compared with PLAP and with several E. coli structures; 1ALK, which contains phosphate in the active site (referred to as “ECAP_P”); 1ED8 without phosphate (“ECAP_woP”); 1ANI, a double mutant with the substitutions D153H and K328H (“ECAP_DM”), which are characteristic for mammalian APS and present in SAP and PLAP, and 1B8J, which has vanadate bound in the active site, as a model for the covalent phosphoenzyme structure (“ECAP_VO3”, see Table 2 for an overview). Understanding the differences between these structures may shed light on the mechanism of SAP. The first obvious difference concerns substrate binding. While the phosphate molecule in PLAP and ECAP_P binds in a monodentate manner to both Zn1 and Zn2, the phosphate molecule in ECAP_DM has shifted towards the solvent, losing interactions with the zinc ions. Although vanadate is penta-coordinated, it binds to the metal ions in a fashion similar to that of ECAP_P. The arginine residue in the active site seems to be influenced by the type of substrate present and is known to be involved in substrate binding.1,41 In ECAP_P, Arg166 holds two phosphate oxygen atoms with its guanidinium group (Figure 4), whereas in ECAP_VO3, the arginine residue points towards the substrate vanadate in a similar conformation.
Although phosphate is not present in the structure of ECAP_woP, the arginine residue has the same orientation as in ECAP_P and ECAP_VO3. The double mutant has phosphate bound at a position different from that in ECAP_P. Hence, the arginine residue points towards this phosphate molecule and has a conformation different from that of the other structures. Arg162, the corresponding arginine residue in SAP, has a conformation different from that of ECAP_DM or any of the other structures (Figure 4), possibly due to the presence of the unidentified ligand. Asp153, which is substituted for a histidine residue in mammalian alkaline phosphatase sequences, is not a direct ligand for magnesium in ECAP_P, ECAP_woP or ECAP_VO3. However, the aspartate residue has been shown to be important for magnesium binding, as it interacts with the metal ion via two water molecules.46,47 In a manner similar to that of SAP, the double mutant ECAP_DM contains a zinc ion in the third metal site, in contrast to the other E. coli structures. The mutated D153H is directed to the metal in a manner similar to that of SAP-H149, the corresponding ligand for the third zinc ion. The presence of the histidine residue must increase the affinity for zinc at this binding site. On the other hand, PLAP contains the histidine residue, but still has magnesium in the active site. This may be due to different crystallization conditions. In ECAP_P, ECAP_woP and ECAP_VO3, Lys328 forms a salt-bridge to Asp153. This salt-bridge is lost in mammalian enzymes. For a K328H mutant, a decrease in phosphate affinity and an increase in Km have been observed because of the loss of a water-mediated interaction between the lysine residue and the substrate.42,46 The side-chain of the corresponding His316 in SAP has an orientation similar to that of the lysine residue in ECAP_P. In one of the monomers of the double mutant ECAP_DM-H328, the histidine residue adopts a conformation similar to that of SAP-H316, it has a different conformation in the second monomer. The final difference between the five structures is subtler, the hydroxyl group of ECAP_P-Ser102 is not a ligand for Zn2, since it binds to the phosphate group. In ECAP_DM-Ser102 and SAP-Ser86, the hydroxyl group points slightly away from Zn2. In ECAP_woP, the hydroxyl group has a
1.9 A˚ Crystal Structure of Heat-labile Shrimp Alkaline Phosphatase
1271
has an aspartate residue (ALK-D153) near the third metal-binding site increases the affinity for zinc, which explains the presence of the zinc triad. At present, it is difficult to speculate about similarities or differences of its enzymatic mechanism compared to E. coli. Future experiments with inhibitor soaks and metal variation will hopefully shed more light on this. Cold adaptation
Figure 5. Surface potential representation, prepared with GRASP,48 of (a) SAP, (b) PLAP and (c) ECAP, with potentials ranging from 2 15 (blue) to þ 15 (red).
different conformation and points in the opposite direction, while in ECAP_VO3, the serine residue binds to vanadate and forms an intermediate analogue. There are many differences in detail among the structures of the AP family, but the largest difference is perhaps the bound substrate and the consequences thereof in terms of orientations of specific residues. At present, therefore, it is hard to draw conclusions on the mechanism of SAP. The currently known structures show that the architecture of the active site has been conserved in evolution, as most active-site residues are identical among a broad range of species. The presence of a histidine residue (SAP-H149), where E. coli
SAP was examined for features that could result in cold adaptation. Compared to its mesophilic homologues, SAP has a surplus of negatively charged amino acid residues and a relatively low number of proline residues, while the frequency of aromatic residues is higher.23 Moreover, the overall charge of 2 80 for the functional dimer (assigning histidine neutral) is distributed such that the protein surface is predominantly negatively charged (see Figure 5 for a potential surface representation of GRASP)48 and will be highly receptive for hydrogen bonded water. The active site is the only clear positively charged patch on the surface. As a result, the negatively charged substrate (a phosphomonoester) must be strongly directed towards the active site. Neither ECAP nor PLAP has this feature. Since electrostatic optimization is a way for cold-active proteins to adjust themselves to their environment, the surface of SAP may have evolved to optimize this direction of the substrate to the active site. A similar observation was found for malate dehydrogenase.32 The potential surfaces of a psychrophilic and a thermophilic enzyme, whose sequences were 90% similar, showed remarkable differences: the cold-adapted enzyme was dominated by negative surface potentials and increased positive potentials around the active site, whereas the thermophilic enzyme showed much weaker potentials. Similar findings were found for trypsin28,49 and citrate synthase.26 It appears clear that optimization of surface potentials is one of the strategies of cold adaptation. SAP has the enzymatic characteristics of coldactive proteins; it is efficient and heat-labile. The crystal structure suggests that this has been
Table 3. Data processing for the MAD data sets at the peak wavelength (PK), inflection point (IP), and a remote wavelength (RM) and for a high-resolution data set (High Res) Data set
Refl. measured
Redundancy
Total reflections
Completeness (%) (unique refl.)
Completeness (%) (Friedel pairs)
MAD-PK MAD-IP MAD-RM High Res.
824,294 647,899 963,147 1,131,014
8.6 7.0 7.8 2.6
43,304 42,653 54,982 103,370
99.0 (90.5) 98.0 (80.9) 98.6 (87.3) 96.8 (99.2)
98.9 (89.9) 97.1 (72.3) 98.6 (87.0)
Rmerge
kIl/ksIl
7.1 (29.6) 7.2 (23.1) 5.8 (24.2) 7.6 (46.5)
30.0 (5.9) 24.5 (3.5) 31.9 (6.8) 11.8 (2.2)
The number of reflections measured is indicated in column 2, the total number of unique reflections in column 4 and the completenesses for the unique reflections and for the Friedel pairs are listed in columns 5 and 6, respectively. Rmerge is defined as {Shkl Si lIi ðhklÞ2 , IkðhklÞll}={Shkl Si Ii ðhklÞ}: Numbers in parentheses are the values for the highest resolution shells. The highest resol˚ ; for the inflection point it is 2.58–2.50 A ˚ ; for the remote wavelength it is 2.37– ution shell for the peak wavelength is 2.57–2.49 A ˚ , and for the high-resolution data set it is 1.99–1.92 A ˚. 2.29 A
˚ Crystal Structure of Heat-labile Shrimp Alkaline Phosphatase 1.9 A
1272
Table 4. Phasing statistics from SHARP for the MAD data set
Rcullis Phasing power
Data set
Acentric
Anomalous
Centric
RM PK IP RM PK IP
0.00 0.58 0.64 0.00 2.39 2.15 0.87 0.95 0.47
0.92 0.79 0.90 1.16 1.73 1.26 0.00 0.00
0.00 0.63 0.69 0.00 1.74 1.64 0.77 0.89 0.47
Lack of closure Lack of isomorphism Figure of merit
Data for acentric, anomalous and centric reflections for the remote (RM), peak (PK) and inflection point (IP) wavelengths.
achieved by optimization of the surface charge distribution.
Materials and Methods Crystallization SAP was kindly provided by Biotec ASA (Tromsø, Norway). The protein solution had a concentration of 11 mg/ml and contained 0.1 mM Zn2þ and 1.0 mM Mg2þ. Crystals were obtained with the hanging drop vapour-diffusion method. The crystallization buffer was 42.5% (w/v) ammonium sulphate, 10% (v/v) glycerol, 100 mM Tris – maleate (pH 5.6). Before data collection, the crystals were soaked for a few minutes in a solution with the same composition as the crystallization buffer, but without Zn2þ or Mg2þ (to wash away excess metal ions) and with an increased concentration (30%) of glycerol for cryo-protection.
determined the positions of six zinc atoms, which formed two triads. This was interpreted as the active sites of the two monomers in the asymmetric unit. The zinc substructure was refined by SHARP,55 which calculated the initial protein phases. The B-factors of the zinc atoms were refined anisotropically in order to obtain the best-quality initial maps. Phasing statistics are listed in ˚ Table 4. The phases, calculated to a resolution of 2.5 A ˚ by solvent flattening with were extended to 1.9 A SOLOMON,56 with a solvent content of 47%, using the high-resolution data set. This resulted in phases with a figure of merit of 0.85. Several secondary structural elements could be identified immediately in the initial electron density maps. Model building and refinement
A multiwavelength anomalous dispersion (MAD) data set was collected around the zinc K edge at beamline ID14-EH4 at the European Synchrotron Radiation Facility (E.S.R.F.) at a temperature of 100 K, using an ADSC Q4R CCD based detector. A fluorescence scan showed the presence of zinc in the sample. Data were collected at the peak of the white-line with a wavelength ˚ , at the inflection point with a wavelength of of 1.2820 A ˚ and at a remote wavelength (0.9310 A ˚ ) to a 1.2825 A ˚ for the peak and inflection point resolution of 2.5 A ˚ for the remote wavelength. In wavelengths and to 2.3 A ˚ was collected addition, a high-resolution data set to 1.9 A at ID14-EH2 at a temperature of 40 K at a wavelength of ˚ , with helium cooling and a similar detector. The 0.933 A data sets were indexed and integrated by DENZO, and further processed by SCALEPACK50 and programs of the CCP4 program suite:51 SCALEIT52 and TRUNCATE.53 Details of the data processing are shown in Table 3. The protein crystallizes in space group P43212 with and c ¼ 84:322 A: unit cell dimensions of a ¼ 171:069A; Sixteen protein modules fit in the unit cell, one dimer per asymmetric unit. This corresponds to a Matthew’s coefficient of 2.8, with a calculated molecular mass of 53 kDa, corresponding to a solvent content of 47% (v/v).
The backbone of the protein was traced by ARP/ wARP.57 Side-chains were added manually with QUANTA 2000 (Accelrys). The model was refined with REFMAC58 against the map obtained from SOLOMON using the high-resolution data set with extended phases ˚ and all reflections. After an initial rigid body to 1.9 A refinement (without NCS) the overall R-factor was 0.38. In the following cycles of refinement, the atom positions and B-factors were refined isotropically. When the overall structure was built, including termini, water molecules and sugar moieties, there was still residual density (Fo 2 Fc . 3s ) around Zn1 in both monomers. The zinc ion had only three ligands, while it prefers a coordination of four or five ligands. The residual density had a flat shape and could fit a small molecule. Several approaches were taken in order to interpret this density: omit maps (excluding the active-site residues) have been calculated, all reflections were included (including the 5% of reflections that were used for calculating Rfree) and anomalous difference maps were calculated. In addition, successively collected data sets were inspected for radiation damage. Finally, various small molecules, which could have been present during the preparation of the enzyme or crystallization were modelled, but none was completely satisfactory. The possibility remains that the site is partially occupied by different ligands. Further residual density, which clearly did not belong to the protein, was fitted with either sulphate or maleate ions, which were both present in the crystallization buffer. Water molecules were added with the X-SOLVATE module of Quanta 2000 (Accelrys).
Phasing
Protein Data Bank accession number
The anomalous Pattersons showed significant and well-resolved peaks, up to a level of 23s. SOLVE54
Final atomic coordinates have been deposited in the RCSB Protein Data Bank with the entry code 1K7H.
Data collection
1.9 A˚ Crystal Structure of Heat-labile Shrimp Alkaline Phosphatase
Acknowledgments We thank the E.S.R.F. for providing synchrotron radiation facilities and Raimond Ravelli for assistance with data collection at beamlines ID14-EH4 and ID14EH2. We thank the staff of the Swiss-Norwegian beamlines (BM01) at the E.S.R.F., where many preliminary data collections were done, and Dr Inge W. Nilsen at the Norwegian Institute of Fisheries and Aquaculture, who kindly provided us with the gene sequence before publication.
References 1. Kim, E. E. & Wyckoff, H. W. (1991). Reaction mechanism of alkaline phosphatase based on crystal structures. Two-metal ion catalysis. J. Mol. Biol. 218, 449–464. 2. Moss, D. W. (1992). Perspectives in alkaline phosphatase research. Clin. Chem. 38, 2486– 2492. 3. Burlina, A., Rubin, D., Secchiero, S., Sciacovelli, L., Zaninotto, M. & Plebani, M. (1994). Monitoring skeletal cancer metastases with the bone isoenzyme of tissue unspecific alkaline phosphatase. Clin. Chim. Acta, 226, 151– 158. 4. Demers, L. M., Costa, L. & Lipton, A. (2000). Biochemical markers and skeletal metastases. Cancer, 88, 2919– 2926. 5. Henthorn, P. S. & Whyte, M. P. (1992). Missense mutations of the tissue-nonspecific alkaline phosphatase gene in hypophosphatasia. Clin. Chem. 38, 2501–2505. 6. Millan, J. L. (1992). Alkaline phosphatase as a reporter of cancerous transformation. Clin. Chim. Acta, 209, 123– 129. 7. Mornet, E. (2000). Hypophosphatasia: the mutations in the tissue-nonspecific alkaline phosphatase gene. Hum. Mutat. 15, 309– 315. 8. Narayanan, S. (1991). Serum alkaline phosphatase isoenzymes as markers of liver disease. Ann. Clin. Lab. Sci. 21, 12 – 18. 9. Knox, J. R. & Wyckoff, H. W. (1973). A crystallo˚ graphic study of alkaline phosphatase at 7– 7 A resolution. J. Mol. Biol. 74, 533– 545. 10. Hough, E., Hansen, L. K., Birknes, B., Jynge, K., Hansen, S., Hordvik, A. et al. (1989). High-resolution ˚ ) crystal structure of phospholipase C from (1.5 A Bacillus cereus. Nature, 338, 357– 360. 11. Volbeda, A., Lahm, A., Sakiyama, F. & Suck, D. (1991). Crystal structure of Penicillium citrinum P1 ˚ resolution. EMBO J. 10, 1607– 1618. nuclease at 2.8 A 12. Stec, B., Holtz, K. M. & Kantrowitz, E. R. (2000). A revised mechanism for the alkaline phosphatase reaction involving three metal ions. J. Mol. Biol. 299, 1303–1311. 13. Holtz, K. M., Stec, B. & Kantrowitz, E. R. (1999). A model of the transition state in the alkaline phosphatase reaction. J. Biol. Chem. 274, 8351– 8354. 14. Stec, B., Hehir, M. J., Brennan, C., Nolte, M. & Kantrowitz, E. R. (1998). Kinetic and X-ray structural studies of three mutant E. coli alkaline phosphatases: insights into the catalytic mechanism without the nucleophile Ser102. J. Mol. Biol. 277, 647– 662. 15. Chaidaroglou, A. & Kantrowitz, E. R. (1989). Alteration of aspartate 101 in the active site of Escherichia coli alkaline phosphatase enhances the catalytic activity. Protein Eng. 3, 127– 132.
1273
16. Chen, L., Neidhart, D., Kohlbrenner, W. M., Mandecki, W., Bell, S., Sowadski, J. & AbadZapatero, C. (1992). 3-D structure of a mutant (Asp101 ! Ser) of E. coli alkaline phosphatase with higher catalytic activity. Protein Eng. 5, 605– 610. 17. Butler-Ransohoff, J. E., Rokita, S. E., Kendall, D. A., Banzon, J. A., Carano, K. S., Kaiser, T. & Matlin, A. R. (1992). Active-site mutagenesis of E. coli alkaline phosphatase: replacement of Serine-102 with nonnucleophile amino acids. J. Org. Chem. 57, 142 –145. 18. Ghosh, S. S., Bock, S. C., Rokita, S. E. & Kaiser, E. T. (1986). Modification of the active site of alkaline phosphatase by site-directed mutagenesis. Science, 231, 145– 148. 19. Holtz, K. M. & Kantrowitz, E. R. (1999). The mechanism of the alkaline phosphatase reaction: insights from NMR, crystallography and site-specific mutagenesis. FEBS Letters, 462, 7 – 11. 20. Sowadski, J. M., Handschumacher, M. D., Murthy, H. M., Kundrot, C. E. & Wyckoff, H. W. (1983). Crystallographic observations of the metal ion triple in the active site region of alkaline phosphatase. J. Mol. Biol. 170, 575– 581. 21. Sowadski, J. M., Handschumacher, M. D., Murthy, H. M., Foster, B. A. & Wyckoff, H. W. (1985). Refined structure of alkaline phosphatase from Escherichia coli ˚ resolution. J. Mol. Biol. 186, 417– 433. at 2.8 A 22. Le Du, M. H., Stigbrand, T., Taussig, M. J., Menez, A. & Stura, E. A. (2001). Crystal structure of alkaline ˚ resophosphatase from human placenta at 1.8 A lution. Implication for a substrate specificity. J. Biol. Chem. 276, 9158– 9165. 23. Nilsen, I. W., Overbo, K. & Olsen, R. L. (2001). Thermolabile alkaline phosphatase from Northern shrimp (Pandalus borealis ): protein and cDNA sequence analyses. Comp. Biochem. Physiol. B Biochem. Mol. Biol. 129, 853– 861. 24. Bradshaw, R. A., Cancedda, F., Ericsson, L. H., Neumann, P. A., Piccoli, S. P., Schlesinger, M. J. et al. (1981). Amino acid sequence of Escherichia coli alkaline phosphatase. Proc. Natl Acad. Sci. USA, 78, 3473 –3477. 25. Kam, W., Clauser, E., Kim, Y. S., Kan, Y. W. & Rutter, W. J. (1985). Cloning, sequencing, and chromosomal localization of human term placental alkaline phosphatase cDNA. Proc. Natl Acad. Sci. USA, 82, 8715 –8719. 26. Russell, R. J., Gerike, U., Danson, M. J., Hough, D. W. & Taylor, G. L. (1998). Structural adaptations of the cold-active citrate synthase from an Antarctic bacterium. Structure, 6, 351– 361. 27. Karlsen, S., Hough, E. & Olsen, R. L. (1998). Structure and proposed amino-acid sequence of a pepsin from Atlantic cod (Gadus morhua ). Acta Crystallog. sect. D, 54, 32 –46. 28. Smalas, A. O., Heimstad, E. S., Hordvik, A., Willassen, N. P. & Male, R. (1994). Cold adaption of enzymes: structural comparison between salmon and bovine trypsins. Proteins: Struct. Funct. Genet. 20, 149–166. 29. Berglund, G. I., Willassen, N. P., Hordvik, A. & Smalas, A. O. (1995). Structure of native pancreatic ˚ resoelastase from North-Atlantic samlon at 1.61 A lution. Acta Crystallog. sect. D, 51, 925– 937. 30. Alvarez, M., Zeelen, J. P., Mainfroid, V., RentierDelrue, F., Martial, J. A., Wyns, L. et al. (1998). Triose-phosphate isomerase (TIM) of the psychro-
1274
31.
32.
33. 34.
35.
36.
37.
38. 39.
40.
41.
42.
43.
44.
˚ Crystal Structure of Heat-labile Shrimp Alkaline Phosphatase 1.9 A
philic bacterium Vibrio marinus. Kinetic and structural properties. J. Biol. Chem. 273, 2199– 2206. Aghajari, N., Feller, G., Gerday, C. & Haser, R. (1998). Structures of the psychrophilic Alteromonas haloplanctis alpha-amylase give insights into cold adaptation at a molecular level. Structure, 6, 1503– 1516. Kim, S. Y., Hwang, K. Y., Kim, S. H., Sung, H. C., Han, Y. S. & Cho, Y. (1999). Structural basis for cold adaptation. Sequence, biochemical properties, and crystal structure of malate dehydrogenase from a psychrophile Aquaspirillium arcticum. J. Biol. Chem. 274, 11761– 11767. Smalas, A. O., Schroder-Leiros, H.-K., Os, V. & Willassen, N. P. (2000). Cold adapted enzymes. Biotechnol. Ann. Rev. 6, 1 – 57. Asgeirsson, B., Hartemink, R. & Chlebowski, J. F. (1995). Alkaline phosphatase from Atlantic cod (Gadus morhua ). Kinetic and structural properties which indicate adaptation to low temperatures. Comp. Biochem. Physiol. 110B, 315– 329. Asgeirsson, B., Hauksson, J. B. & Gunnarsson, G. H. (2000). Dissociation and unfolding of cold-active alkaline phosphatase from Atlantic cod in the presence of guanidinium chloride. Eur. J. Biochem. 267, 6403– 6412. Hauksson, J. B., Andresson, O. S. & Asgeirsson, B. (2000). Heat-labile bacterial alkaline phosphatase from a marine Vibrio sp. Enzyme Microb. Technol. 27, 66 –73. Rina, M., Pozidis, C., Mavromatis, K., Tzanodaskalaki, M., Kokkinidis, M. & Bouriotis, V. (2000). Alkaline phosphatase from the Antarctic strain TAB5. Properties and psychrophilic adaptations. Eur. J. Biochem. 267, 1230– 1238. Sheldrick, G. M., Dauter, Z., Wilson, K. S. & Sieker, L. C. (1993). Acta Crystallog. sect. D, 49, 18 – 23. Laskowski, R. A., MacArthur, M. W., Moss, D. S. & Thornton, J. M. (1993). PROCHECK: a program to check the stereochemical quality of protein structures. J. Appl. Crystallog. 26, 283– 291. Butler-Ransohoff, J. E., Kendall, D. A. & Kaiser, E. T. (1988). Use of site-directed mutagenesis to elucidate the role of arginine-166 in the catalytic mechanism of alkaline phosphatase. Proc. Natl Acad. Sci. USA, 85, 4276– 4278. Chaidaroglou, A., Brezinski, D. J., Middleton, S. A. & Kantrowitz, E. R. (1988). Function of arginine-166 in the active site of Escherichia coli alkaline phosphatase. Biochemistry, 27, 8338– 8343. Murphy, J. E., Tibbitts, T. T. & Kantrowitz, E. R. (1995). Mutations at positions 153 and 328 in Escherichia coli alkaline phosphatase provide insight towards the structure and function of mammalian and yeast alkaline phosphatases. J. Mol. Biol. 253, 604– 617. Cathala, G. & Brunel, C. (1975). Bovine kidney alkaline phosphatase. Catalytic properties, subunit interactions in the catalytic process, and mechanism of Mg2þ stimulation. J. Biol. Chem. 250, 6046– 6053. Hung, H. C. & Chang, G. G. (2001). Differentiation of the slow-binding mechanism for magnesium ion
45.
46.
47.
48. 49.
50. 51. 52. 53. 54. 55. 56.
57.
58.
59.
60.
activation and zinc ion inhibition of human placental alkaline phosphatase. Protein Sci. 10, 34 – 45. Zhang, R.-Q., Chen, Q.-X., Xiao, R., Xie, L.-P., Zeng, X.-G. & Zhou, H.-M. (2001). Inhibition kinetics of green crab (Scylla serrata ) alkaline phosphatase by zinc ions: a new type of complexing inhibition. Biochim. Biophys. Acta, 1545, 6– 12. Janeway, C. M., Xu, X., Murphy, J. E., Chaidaroglou, A. & Kantrowitz, E. R. (1993). Magnesium in the active site of Escherichia coli alkaline phosphatase is important for both structural stabilization and catalysis. Biochemistry, 32, 1601– 1609. Murphy, J. E., Xu, X. & Kantrowitz, E. R. (1993). Conversion of a magnesium binding site into a zinc binding site by a single amino acid substitution in Escherichia coli alkaline phosphatase. J. Biol. Chem. 268, 21497– 21500. Nicholls, A., Sharp, K. & Honig, B. (1991). Proteins: Struct. Funct. Genet. 11, 281. Schroder Leiros, H. K., Willassen, N. P. & Smalas, A. O. (2000). Structural comparison of psychrophilic and mesophilic trypsins. Elucidating the molecular basis of cold-adaptation. Eur. J. Biochem. 267, 1039– 1049. Otwinowski, Z. & Minor, W. (1997). Methods Enzymol. 276, 307– 326. Collaborative Computational Project Number 4 (1994). The CCP4 suite: programs for protein crystallography. Acta Crystallog. sect. D, 50, 760– 763. Howell, P. L. & Smith, G. D. (1992). Identification of heavy-atom derivatives by normal probability methods. J. Appl. Crystallog. 25, 81 – 86. French, G. S. & Wilson, K. S. (1978). On the treatment of negative intensity oberservations. Acta Crystallog. sect. A, 34, 517. Terwilliger, T. C. & Berendzen, J. (1999). Automated MAD and MIR structure solution. Acta Crystallog. sect. D, 55, 841– 861. La Fortelle, E. d. & Bricogne, G. (1997). Methods Enzymol. 276, 472– 494. Abrahams, J. P. & Leslie, A. G. W. (1996). Methods used in the structure determination of bovine mitochondrial F1 ATPase. Acta Crystallog. sect. D, 52, 30 –42. Perrakis, A., Sixma, T. K., Wilson, K. S. & Lamzin, V. S. (1997). wARP: improvement and extension of crystallographic phases by weighted averaging of multiple refined dummy atomic models. Acta Crystallog. sect. D, 53, 448– 455. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. (1997). Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallog. sect. D, 53, 240– 255. Jones, T. A., Zou, J. Y., Cowan, S. W. & Kjelgaard, M. (1991). Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallog. sect. A, 47, 110 – 119. Esnouf, R. M. (1997). An extensively modified version of MolScript that includes greatly enhanced coloring capabilities. J. Mol. Graph. 15, 132– 134.
Edited by R. Huber (Received 26 October 2001; received in revised form 25 January 2002; accepted 28 January 2002)
w B
J. Mol. Biol. (2002) 318, 1265–1274
˚ Crystal Structure of Heat-labile Shrimp The 1.9 A Alkaline Phosphatase Maaike de Backer1, Sean McSweeney1, Hanne B. Rasmussen2 Bjørn W. Riise3, Peter Lindley1 and Edward Hough3* 1
European Synchrotron Radiation Facility, BP 220 F-38043 Grenoble Cedex France 2 Novo Nordisk A/S, Novo Alle 6B2.90, DK-2880 Bagsvaerd, Denmark 3
Department of Chemistry Faculty of Science, University of Tromsø N 9037 Tromso, Norway
Alkaline phosphatases are non-specific phosphomonoesterases that are distributed widely in species ranging from bacteria to man. This study has concentrated on the tissue-nonspecific alkaline phosphatase from arctic shrimps (shrimp alkaline phosphatase, SAP). Originating from a cold-active species, SAP is thermolabile and is used widely in vitro, e.g. to dephosphorylate DNA or dNTPs, since it can be inactivated by a short rise in temperature. Since alkaline phosphatases are zinc-containing enzymes, a multiwavelength anomalous dispersion (MAD) experiment was performed on the zinc K edge, which led to the determination of the structure to a resolution ˚ . Anomalous data clearly showed the presence of a zinc triad in of 1.9 A the active site, whereas alkaline phosphatases usually contain two zinc and one magnesium ion per monomer. SAP shares the core, an extended b-sheet flanked by a-helices, and a metal triad with the currently known alkaline phosphatase structures (Escherichia coli structures and a human placental structure). Although SAP lacks some features specific for the mammalian enzyme, their backbones are very similar and may therefore be typical for other higher organisms. Furthermore, SAP possesses a striking feature that the other structures lack: surface potential representations show that the enzyme’s net charge of 2 80 is distributed such that the surface is predominantly negatively charged, except for the positively charged active site. The negatively charged substrate must therefore be directed strongly towards the active site. It is generally accepted that optimization of the electrostatics is one of the characteristics related to cold-adaptation. SAP demonstrates this principle very clearly. q 2002 Elsevier Science Ltd. All rights reserved
*Corresponding author
Keywords: structure; X-ray crystallography; phosphatase; metalloprotein; cold-adaptation
Introduction Alkaline phosphatases (AP, EC 3.1.3.1) are homodimeric metalloenzymes that catalyze the hydrolysis and, in the presence of a phosphate acceptor, transphosphorylation of a wide variety of phosphate monoesters. The enzymatic reaction proceeds through a covalent phosphoseryl interThis article is dedicated to the memory of Bjørn W. Riise Abbreviations used: AP, alkaline phosphatase; SAP, shrimp AP; ECAP, Escherichia coli AP; PLAP, human placental AP; MAD, multiwavelength anomalous dispersion. E-mail address of the corresponding author: [email protected]
mediate to produce inorganic phosphate or to transfer the phosphoryl group to alcohols.1 Substrates include proteins, DNA and small organic molecules. The product, inorganic phosphate, is also an inhibitor. The phosphatases are abundant in prokaryotic and eukaryotic species, where they often occur in several isoforms. In humans, tissuenonspecific genes are widely expressed, in contrast to the localized, tissue-specific expression of the placental, intestinal and germ-cell genes.2 Although their physiological roles are not clear, alkaline phosphatases are probably involved in the mineralization of bone and, for Escherichia coli, in the acquisition of phosphate. They are utilized as clinical diagnostic tools for various diseases; e.g. deficient activity of the tissue-nonspecific
0022-2836/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved
˚ Crystal Structure of Heat-labile Shrimp Alkaline Phosphatase 1.9 A
1266
Table 1. Residues that constitute the metal-binding sites P. borealis
H. sapiens
E. coli
1 (Zn)
Asp315 His319 His432
2 (Zn)
Asp37 Ser86 Asp356 His357 Asp37 His149 Thr151 Glu310 Water
Asp316 His320 His432 PO32 4 Asp42 Ser92 Asp357 His358 Asp42 (His153) Ser155 Glu311
Asp327 His331 His412 PO32 4 Asp51 Ser102 Asp369 His370 Asp51 (Asp153) Thr155 Glu322 PO32 4
Metal-binding site
3 (Mg/Zn)
Ligands for the metal triad of shrimp (P. borealis ), human placental (Homo sapiens ) and E. coli AP, for PDB entries 1EW2 and 1ALK for the human and the bacterial enzyme, respectively. Differences in residue type are underlined. The residues corresponding to SAP-His149 are in parentheses, since they are not liganding the metals.
isozyme is the biochemical hallmark of hypophosphatasia (bone disorder).2 In addition, various tests relate alkaline phosphatase levels to lesions (including tumours), liver disease, vitamin D deficiency, etc.3 – 8 There is a long history of biochemical and biophysical studies on alkaline phosphatases, mainly concentrated on the E. coli enzyme (ECAP). It was shown decades ago that the active site of each monomer contains three metal-binding sites, which are usually occupied by two zinc and one magnesium cation.9 The magnesium-binding site may be occupied by zinc. The metal ions constitute a catalytic triad similar to that in phospholipase C from Bacillus cereus10 and P1 nuclease from Penicillium citrinum.11 A detailed hypothesis about the reaction mechanism has been formulated for the E. coli enzyme, which involves a double in-line displacement and the participation of all three metal ions,1,12 – 18 which has been reviewed by Holtz & Kantrowitz19 and updated by Stec et al.,12 In E. coli, the zinc ions stabilize the activated state of the enzyme and interact with the substrate and leaving group. The magnesium ion in the third metal-binding site plays a structural role and is involved in catalysis. It positions a hydroxide ion, a general base, which accepts a proton from the catalytic serine residue before the first nucleophilic attack on the phosphorus atom. Although this hypothesis is based on structural and biochemical data from the E. coli enzyme, it is assumed to be generally valid, since alkaline phosphatases have a reasonable cross-species sequence similarity. There must, however, be subtle differences, since the mammalian phosphatases are more efficient than their prokaryotic counterparts. The structures of alkaline phosphatase from two species have been determined. The extensively studied ECAP structure was solved initially in ˚ ,9 was later re-deter1973 to a resolution of 7.7 A mined to a higher resolution20,21 and has been
˚ with entry code 1ALK in the deposited at 2.0 A RCSB Protein Data Bank.1 Since then, several structures with various mutations and substrates have been deposited. Recently, the first mammalian ˚ : human placental structure was solved at 1.8 A alkaline phosphatase, PLAP, with PDB entry code 1EW2.22 The major functional differences between mammalian enzymes and their homologues from lower organisms are the improved catalytic activity, reduced heat stability and a shifted pH optimum (to a higher pH in a pH versus activity profile). Sequence alignments between E. coli and mammalian alkaline phosphatases show a 25– 30% identity. Here, we describe the structure of the thermolabile tissue-nonspecific Pandalus borealis (Arctic shrimp) alkaline phosphatase (SAP). The enzyme is used extensively in vitro to dephosphorylate DNA or dNTPs, and can be inactivated by a short rise in temperature to 65 8C. The sequence of this protein has been published.23 The sequences of SAP and ECAP24 have 24% identical, 16% strongly similar and 11% weakly similar residues. SAP and PLAP25 share 41% of their residues, 19% strongly similar and 12% weakly similar. The activesite residues are conserved among these three sequences, except for three residues; Asp153 and Lys328 are histidines in SAP (His149 and His316) and PLAP (His153 and His317), and PLAP has a T155S substitution (ECAP-Thr155, SAP-Thr151, PLAP-Ser155; see Table 1 for an overview). By solving the structure of SAP, we hope to identify structural differences between heat-labile SAP, bacterial ECAP and mammalian PLAP, which explain the differences in catalytic efficiency. Cold-Adaptation The SAP sample that was used for this study was obtained from shrimps living in the Barents Sea off the coast of Northern Norway, where the annual mean water temperature is 5 8C. Species that live in such environments must somehow be adapted to survive at these reduced temperatures. A lower temperature has a number of implications for the functioning of proteins. A number of physiological changes, including the solubility of salts, gases and small organic molecules, affect the charge distribution of the protein directly and therewith its solubility. The increased viscosity of water, which reduces the diffusion rates of substrate and product, leads to lower reaction rates. Reduced thermal motions affect the enzyme’s overall stability, kinetics and ligand binding. A number of studies have been devoted to the relation between cold-adaptation and structural characteristics. However, the number of cold-active enzymes where the structure is known is small. Until now, only seven structures have been solved: Antarctic bacterial citrate synthase,26 cod pepsin,27 salmon trypsin,28 salmon elastase,29 Vibrio marinus triose-phosphate isomerase,30 Alteromonas haloplanctis a-amylase,31 Aquaspirillium malate dehydrogenase.32
1.9 A˚ Crystal Structure of Heat-labile Shrimp Alkaline Phosphatase
1267
Figure 2. Superposition of the Ca-traces of SAP (in yellow) with the E. coli (red) and human placental (blue) structures, showing the conservation of the core (the bsheet and flanking a-helices) and the presence of a crown domain for PLAP and SAP. The structures were superimposed with O59 and prepared with Quanta (Accelrys).
Figure 1. The structure of the functional dimer (monomer A in green, monomer B in red) showing the large extended b-sheets (in white for both monomers), flanked by a-helices, the crown domain (on top of the b-sheet, coloured light green for monomer A), the N-terminal helix (light green for monomer A) and the zinc triad (represented by yellow van der Waals spheres) located on top of the b-sheet, next to the crown domain. (a) A front view, and (b) a top view of the structure. The Figure was prepared with Quanta (Accelrys).
In general, cold-active enzymes have an increased catalytic efficiency at lower temperature, which seems to be accompanied by a reduced thermal stability and increased molecular flexibility. The flexibility should prevent the enzymes from becoming overly rigid at low temperatures. There are, however, no common structural features that account for these findings; each cold-adapted enzyme (family) uses a different array of structural adjustments to gain the flexibility that results in increased catalytic efficiency and reduced stability. Apparently, the origin of increased activity and reduced thermal stability resides in particular regions or domains rather than general characteristics.33 Factors affecting protein stability include hydrophobic interactions (which are weakened
at lower temperatures), electrostatic interactions (which are stabilized at decreasing temperatures) and reduced content of secondary structure. While the enzyme as a whole is more flexible, it should still maintain a good binding-site geometry for the ligand. Moreover, cold-active enzymes seem to optimize their catalytic activity by optimizing the electrostatics at and around their binding sites.33 Cold-active APs from Atlantic cod,34,35 Marine Vibrio sp.36, Arctic strain TAB537 and SAP23 have been studied. This study investigates to what extent structural features that indicate adaptation to cold can be identified in SAP.
Results The overall structure Shrimp alkaline phosphatase is a homodimer ˚ £ 65 A ˚ £ 50 A ˚ with dimensions of about 95 A (Figure 1). The two monomers in the working asymmetric unit do not represent the functional dimer; which is formed by one monomer and its symmetry-related partner, as is evident from the packing in the unit cell and by comparison with ECAP and PLAP (Figure 2). The core of the monomer is composed of a ten-stranded b-sheet, sandwiched between a set of helices of various lengths, three on one side, and five on the other side of the sheet. The b-strands are parallel, except for one anti-parallel insertion. In the functional dimer, the monomers associate such that their individual b-sheets form one extended sheet. On top of the dimeric sheet, perpendicular to the strands, a small domain is situated (360 –430, referred to as the “crown domain” in analogy
1268
˚ Crystal Structure of Heat-labile Shrimp Alkaline Phosphatase 1.9 A
Figure 3. Details of metal-binding sites of SAP for (a) Zn1, (b) Zn2 and (c) Zn3, showing the zinc ions as yellow van der Waals spheres and the liganding residues (which are listed in Table 1). A 2Fo 2 Fc map is plotted in blue at a level of 1.0s and an Fo 2 Fc map is plotted in green at þ3s. Note the unassigned density adjacent to Zn1. This Figure was prepared with Bobscript.60
with PLAP), which is composed of six small bstrands. As this domain is part of the dimer interface, each monomer contributes three small bstrands (360 –364, 386– 391, and 419– 421) to form a three-stranded double layer with the other monomer. One of the connecting loops points outwards to the solvent; others embrace the crown domain or interact with the core. The N-terminal helix points away from the core of the monomer. In the functional dimer, this helix embraces its neighbouring monomer and is part of the dimer interface. Two disulphide bridges are present per monomer; Cys115, located near a glycosylation site at the tip of the b-sheet, bridges with Cys180, which is part of a loop connecting two small helices that are flanking the b-sheet. The second bridge is formed between Cys467 and Cys475, within a small loop at the C terminus. Although the sequence contains the glycosylation motif N-X-T/S twice, extended density was found at only one of these sites, Asn114, located at the tip of the large b-sheet. No significant residual density was found at the second potential glycosylation site, Asn166. Although one N-acetyl-glucosamine molecule could easily be modelled near Asn114, a second sugar molecule was harder to fit. There was clearly 2Fo 2 Fc (1 to 2s ) as well as Fo 2 Fc (higher than 3s ) density for a second ring, although it was clear that the second sugar molecule could not form a 1 –4 glycosidic bond, as is expected for a second N-acetyl-glucosamine linkage. Although alternative sugars would fit the density well, they were not included in the final model, since there is uncertainty about the nature of the sugar moieties. The active site Anomalous scattering data clearly showed that the three metal ions in the active site are all zinc, although the third zinc site had a somewhat lower occupancy. The occupancies have been refined with SHELXL38 to 0.97, 0.97 and 0.88 for respectively, Zn1, Zn2 and Zn3 of monomer A and to 0.95, 0.92 and 0.88 for monomer B. The zinc triad, which is located near the carboxyl top of the b-sheet, is readily accessible for substrates. Since there is only space to accommodate the phosphate
group in the active site, and there is no apparent binding site for the leaving group of the substrate, the location and shape of the active site explain the non-specificity of the enzyme. Unassigned 3s density in an Fo 2 Fc map (Figure 3), associated with Zn1 and Zn2 in the active sites of both monomers, could not be interpreted unambiguously. Fitting with the most likely small molecules, Tris, sulfate, phosphate and carbonate, was attempted, but the results after refinement were not convincing. Therefore, no ligand was included during refinement of the final structure. The three metal ions are relatively close, with distances between Zn1 and Zn2 and between Zn2 ˚ , whereas Zn1 and Zn3 are 7.1 A ˚ and Zn3 of 4.6 A apart. Zn1 interacts with the imidazole nitrogen atoms of His319 and His432, with one carboxyl oxygen atom of Asp315. Zn2 has four ligands: a carboxyl oxygen atom of Asp37 and of Asp356, an imidazole nitrogen atom of His357 and the hydroxyl oxygen atom of Ser86. Zn3 binds to the second carboxyl oxygen atom of Asp37, to an imidazole nitrogen atom of His149, the hydroxyl group of Thr151, to Glu310 and to a water molecule. The ligands of the metals are listed in Table 1.
Model statistics The final model converged to an R-factor of 19.91% for the working set and a free R-value of 22.75% for a randomly chosen test set of 5% of the reflections, with a highest-resolution shell of 1.99 – ˚ . It contains 8108 non-hydrogen protein 1.92 A atoms; 476 amino acid residues per monomer, three zinc ions, one N-acetyl-glucosamine unit, two sulfate groups and one maleic acid molecule per monomer. A stereochemical check by PROCHECK39 shows root-mean-squared (r.m.s.) ˚ in deviations from ideal geometry of 0.028 A covalent bond lengths and 2.08 in bond angles. A Ramachandran plot shows that 89.2% of the residues lie in the most favoured regions, 10.3% in the additional allowed regions, 0.2% in the generously allowed regions and 0.2% in the disallowed regions.
1.9 A˚ Crystal Structure of Heat-labile Shrimp Alkaline Phosphatase
1269
Coordination of the zinc ions in the metal triad
Figure 4. Superposition of SAP (coloured by atom) and PLAP (yellow), showing Arg162 in the active site, which has a different conformation in the two structures. The Figure was prepared with MOLSCRIPT.
Discussion Comparison between SAP and other AP structures The overall structures of SAP, ECAP and PLAP are similar, with the core and active-site architecture preserved (Figure 2). Moreover, the backbones of SAP and PLAP superimpose with an ˚ for 455 Ca atoms, whereas the r.m.s.d. of 1.07 A superposition of SAP with 1ALK results in an ˚ for 343 Ca atoms. SAP and PLAP r.m.s.d. of 1.33 A share the crown and N-terminal domains, which are probably present in all mammalian enzymes. PLAP, however, contains a fourth metal-binding site, where a magnesium ion is bound.22 Although SAP shares some of the residues that comprise this binding site, no extra metal ion was detected. In the E. coli structure 1ALK, phosphate is bound in the active site with two of the phosphate oxygen atoms coordinated by Zn1 and Zn2, and the other two oxygen atoms bound by the guanidinium group of Arg166, which in turn is hydrogen bonded to Asp101 and a water molecule. The phosphate group is further bound to the amide group of Ser102 and a water molecule. The corresponding arginine residue in SAP (Arg162) is oriented differently; it points away from the metal ions (Figure 4). It has been shown that ECAPArg166 is primarily responsible for substrate binding, as it interacts directly with the substrate in the non-covalent enzyme – phosphate complex.1 Modification of the arginine residue leads to a lower substrate affinity, but not to a significant decrease of the activity.40,41
Although the zinc atoms in SAP are coordinated to ECAP and PLAP in a similar way, substitutions in the metal-binding sites seem to influence metal affinity. Zn1 in SAP is held by an aspartate and two histidine residues, like the other known structures. There is a possibility that residual density close to the zinc ions could be a further but unidentified ligand (see above). In 1ALK and PLAP the phosphate contributes two oxygen atoms, resulting in a penta-coordinated zinc ion. Zn2 has four ligands in SAP: two aspartate residues, a histidine residue, and the nucleophilic serine residue; in both ECAP and PLAP, Zn2 is also coordinated tetrahedrally by two aspartate residues, a histidine residue and an oxygen atom of the phosphate inhibitor. While SAP-Zn3 binds to an aspartate, a threonine residue, a glutamate residue, a histidine residue and a water molecule, the corresponding metal sites in ECAP and PLAP are occupied by a magnesium ion, which is coordinated octahedrally by an aspartate residue, a glutamine residue, a threonine residue (ECAP) or a serine residue (PLAP) and three water molecules. APs usually contain a magnesium ion in the third metal-binding site, while a zinc ion is observed in SAP. The histidine residue that ligates Zn3 in SAP (H149) replaces an aspartate residue in E. coli (D153), which does not interact with the metal ion directly. Since histidine is a common ligand for zinc, the third metal-binding site of SAP probably has a higher affinity for zinc than the corresponding site in ECAP. A mutant E. coli enzyme with the aspartate residue substituted with histidine also contained zinc at this site.42 Sequence alignments show that this histidine residue is conserved in mammalian sequences, including PLAP, although a magnesium ion is reported to occupy this site.22 Indeed, the structure of PLAP shows that this histidine residue is too far away from the zinc ion to function as a ligand. Biochemical data suggests that a fully zincoccupied enzyme is less active. In fact, a number of mammalian enzymes need the addition of magnesium in order to achieve maximal activity.43 Since this effect is time-dependent and independent of the concentration of magnesium, a conformational change has been proposed for PLAP44 and similar effects have been reported recently for AP from Scylla serrata.45 This probably involves the exchange of zinc with magnesium, which leads to a more open conformation of the active site, since magnesium is coordinated octahedrally. This effect has been observed for E. coli mutants D153H and for a double mutant D153H/K328H, which mimic mammalian enzymes.42,46 Stec and co-workers12 propose that a magnesium-coordinated hydroxide ion functions as a general base and accepts a proton from the nucleophilic serine residue. The activated serine residue is stabilized by coordination to Zn2. A metal ion other than magnesium (i.e. zinc) would
˚ Crystal Structure of Heat-labile Shrimp Alkaline Phosphatase 1.9 A
1270
Table 2. Selected number of active-site residues of alkaline phosphatase structures Structure SAP PLAP ECAP_DM ECAP_P ECAP_woP ECAP_VO3
Comment
PDB entry
Residue
Residue
Residue
– – D153H, K328H Phosphate present Without phosphate VO22 3 present
1K7H 1EW2 1ANI 1ALK 1ED8 1B8J
His149 His153 His153 Asp153 Asp153 Asp153
His316 His328 His328 Lys328 Lys328 Lys328
Ser86 Ser102 Ser102 Ser102 Ser102 Ser102
Active-site residues of SAP (P. borealis ), PLAP (H. sapiens ), 1ANI (E. coli; double mutant D153H and K358H), 1ALK (E. coli; containing phosphate), 1ED8 (E. coli; no phosphate present), and 1BJ8 (E. coli; containing vanadate).
distort the local geometry and therewith the action of the enzyme. It is somewhat surprising that SAP is fully occupied by zinc, since both Zn2þ and Mg2þ were present in the crystallization buffer. There should be enough time to exchange metals during the crystallization process. After dissolving the crystals in a solution containing both zinc and magnesium, the activity of SAP increases. This may be explained by the exchange of zinc for magnesium in the third metal-binding site. The presence of zinc in the SAP structure can be explained by a higher affinity for zinc than for magnesium in the third metal-binding site, or by a more stable conformation when fully occupied with zinc. We have therefore probably crystallized the enzyme in a low-activity state.
Comparison of active sites The active site of SAP has been compared with PLAP and with several E. coli structures; 1ALK, which contains phosphate in the active site (referred to as “ECAP_P”); 1ED8 without phosphate (“ECAP_woP”); 1ANI, a double mutant with the substitutions D153H and K328H (“ECAP_DM”), which are characteristic for mammalian APS and present in SAP and PLAP, and 1B8J, which has vanadate bound in the active site, as a model for the covalent phosphoenzyme structure (“ECAP_VO3”, see Table 2 for an overview). Understanding the differences between these structures may shed light on the mechanism of SAP. The first obvious difference concerns substrate binding. While the phosphate molecule in PLAP and ECAP_P binds in a monodentate manner to both Zn1 and Zn2, the phosphate molecule in ECAP_DM has shifted towards the solvent, losing interactions with the zinc ions. Although vanadate is penta-coordinated, it binds to the metal ions in a fashion similar to that of ECAP_P. The arginine residue in the active site seems to be influenced by the type of substrate present and is known to be involved in substrate binding.1,41 In ECAP_P, Arg166 holds two phosphate oxygen atoms with its guanidinium group (Figure 4), whereas in ECAP_VO3, the arginine residue points towards the substrate vanadate in a similar conformation.
Although phosphate is not present in the structure of ECAP_woP, the arginine residue has the same orientation as in ECAP_P and ECAP_VO3. The double mutant has phosphate bound at a position different from that in ECAP_P. Hence, the arginine residue points towards this phosphate molecule and has a conformation different from that of the other structures. Arg162, the corresponding arginine residue in SAP, has a conformation different from that of ECAP_DM or any of the other structures (Figure 4), possibly due to the presence of the unidentified ligand. Asp153, which is substituted for a histidine residue in mammalian alkaline phosphatase sequences, is not a direct ligand for magnesium in ECAP_P, ECAP_woP or ECAP_VO3. However, the aspartate residue has been shown to be important for magnesium binding, as it interacts with the metal ion via two water molecules.46,47 In a manner similar to that of SAP, the double mutant ECAP_DM contains a zinc ion in the third metal site, in contrast to the other E. coli structures. The mutated D153H is directed to the metal in a manner similar to that of SAP-H149, the corresponding ligand for the third zinc ion. The presence of the histidine residue must increase the affinity for zinc at this binding site. On the other hand, PLAP contains the histidine residue, but still has magnesium in the active site. This may be due to different crystallization conditions. In ECAP_P, ECAP_woP and ECAP_VO3, Lys328 forms a salt-bridge to Asp153. This salt-bridge is lost in mammalian enzymes. For a K328H mutant, a decrease in phosphate affinity and an increase in Km have been observed because of the loss of a water-mediated interaction between the lysine residue and the substrate.42,46 The side-chain of the corresponding His316 in SAP has an orientation similar to that of the lysine residue in ECAP_P. In one of the monomers of the double mutant ECAP_DM-H328, the histidine residue adopts a conformation similar to that of SAP-H316, it has a different conformation in the second monomer. The final difference between the five structures is subtler, the hydroxyl group of ECAP_P-Ser102 is not a ligand for Zn2, since it binds to the phosphate group. In ECAP_DM-Ser102 and SAP-Ser86, the hydroxyl group points slightly away from Zn2. In ECAP_woP, the hydroxyl group has a
1.9 A˚ Crystal Structure of Heat-labile Shrimp Alkaline Phosphatase
1271
has an aspartate residue (ALK-D153) near the third metal-binding site increases the affinity for zinc, which explains the presence of the zinc triad. At present, it is difficult to speculate about similarities or differences of its enzymatic mechanism compared to E. coli. Future experiments with inhibitor soaks and metal variation will hopefully shed more light on this. Cold adaptation
Figure 5. Surface potential representation, prepared with GRASP,48 of (a) SAP, (b) PLAP and (c) ECAP, with potentials ranging from 2 15 (blue) to þ 15 (red).
different conformation and points in the opposite direction, while in ECAP_VO3, the serine residue binds to vanadate and forms an intermediate analogue. There are many differences in detail among the structures of the AP family, but the largest difference is perhaps the bound substrate and the consequences thereof in terms of orientations of specific residues. At present, therefore, it is hard to draw conclusions on the mechanism of SAP. The currently known structures show that the architecture of the active site has been conserved in evolution, as most active-site residues are identical among a broad range of species. The presence of a histidine residue (SAP-H149), where E. coli
SAP was examined for features that could result in cold adaptation. Compared to its mesophilic homologues, SAP has a surplus of negatively charged amino acid residues and a relatively low number of proline residues, while the frequency of aromatic residues is higher.23 Moreover, the overall charge of 2 80 for the functional dimer (assigning histidine neutral) is distributed such that the protein surface is predominantly negatively charged (see Figure 5 for a potential surface representation of GRASP)48 and will be highly receptive for hydrogen bonded water. The active site is the only clear positively charged patch on the surface. As a result, the negatively charged substrate (a phosphomonoester) must be strongly directed towards the active site. Neither ECAP nor PLAP has this feature. Since electrostatic optimization is a way for cold-active proteins to adjust themselves to their environment, the surface of SAP may have evolved to optimize this direction of the substrate to the active site. A similar observation was found for malate dehydrogenase.32 The potential surfaces of a psychrophilic and a thermophilic enzyme, whose sequences were 90% similar, showed remarkable differences: the cold-adapted enzyme was dominated by negative surface potentials and increased positive potentials around the active site, whereas the thermophilic enzyme showed much weaker potentials. Similar findings were found for trypsin28,49 and citrate synthase.26 It appears clear that optimization of surface potentials is one of the strategies of cold adaptation. SAP has the enzymatic characteristics of coldactive proteins; it is efficient and heat-labile. The crystal structure suggests that this has been
Table 3. Data processing for the MAD data sets at the peak wavelength (PK), inflection point (IP), and a remote wavelength (RM) and for a high-resolution data set (High Res) Data set
Refl. measured
Redundancy
Total reflections
Completeness (%) (unique refl.)
Completeness (%) (Friedel pairs)
MAD-PK MAD-IP MAD-RM High Res.
824,294 647,899 963,147 1,131,014
8.6 7.0 7.8 2.6
43,304 42,653 54,982 103,370
99.0 (90.5) 98.0 (80.9) 98.6 (87.3) 96.8 (99.2)
98.9 (89.9) 97.1 (72.3) 98.6 (87.0)
Rmerge
kIl/ksIl
7.1 (29.6) 7.2 (23.1) 5.8 (24.2) 7.6 (46.5)
30.0 (5.9) 24.5 (3.5) 31.9 (6.8) 11.8 (2.2)
The number of reflections measured is indicated in column 2, the total number of unique reflections in column 4 and the completenesses for the unique reflections and for the Friedel pairs are listed in columns 5 and 6, respectively. Rmerge is defined as {Shkl Si lIi ðhklÞ2 , IkðhklÞll}={Shkl Si Ii ðhklÞ}: Numbers in parentheses are the values for the highest resolution shells. The highest resol˚ ; for the inflection point it is 2.58–2.50 A ˚ ; for the remote wavelength it is 2.37– ution shell for the peak wavelength is 2.57–2.49 A ˚ , and for the high-resolution data set it is 1.99–1.92 A ˚. 2.29 A
˚ Crystal Structure of Heat-labile Shrimp Alkaline Phosphatase 1.9 A
1272
Table 4. Phasing statistics from SHARP for the MAD data set
Rcullis Phasing power
Data set
Acentric
Anomalous
Centric
RM PK IP RM PK IP
0.00 0.58 0.64 0.00 2.39 2.15 0.87 0.95 0.47
0.92 0.79 0.90 1.16 1.73 1.26 0.00 0.00
0.00 0.63 0.69 0.00 1.74 1.64 0.77 0.89 0.47
Lack of closure Lack of isomorphism Figure of merit
Data for acentric, anomalous and centric reflections for the remote (RM), peak (PK) and inflection point (IP) wavelengths.
achieved by optimization of the surface charge distribution.
Materials and Methods Crystallization SAP was kindly provided by Biotec ASA (Tromsø, Norway). The protein solution had a concentration of 11 mg/ml and contained 0.1 mM Zn2þ and 1.0 mM Mg2þ. Crystals were obtained with the hanging drop vapour-diffusion method. The crystallization buffer was 42.5% (w/v) ammonium sulphate, 10% (v/v) glycerol, 100 mM Tris – maleate (pH 5.6). Before data collection, the crystals were soaked for a few minutes in a solution with the same composition as the crystallization buffer, but without Zn2þ or Mg2þ (to wash away excess metal ions) and with an increased concentration (30%) of glycerol for cryo-protection.
determined the positions of six zinc atoms, which formed two triads. This was interpreted as the active sites of the two monomers in the asymmetric unit. The zinc substructure was refined by SHARP,55 which calculated the initial protein phases. The B-factors of the zinc atoms were refined anisotropically in order to obtain the best-quality initial maps. Phasing statistics are listed in ˚ Table 4. The phases, calculated to a resolution of 2.5 A ˚ by solvent flattening with were extended to 1.9 A SOLOMON,56 with a solvent content of 47%, using the high-resolution data set. This resulted in phases with a figure of merit of 0.85. Several secondary structural elements could be identified immediately in the initial electron density maps. Model building and refinement
A multiwavelength anomalous dispersion (MAD) data set was collected around the zinc K edge at beamline ID14-EH4 at the European Synchrotron Radiation Facility (E.S.R.F.) at a temperature of 100 K, using an ADSC Q4R CCD based detector. A fluorescence scan showed the presence of zinc in the sample. Data were collected at the peak of the white-line with a wavelength ˚ , at the inflection point with a wavelength of of 1.2820 A ˚ and at a remote wavelength (0.9310 A ˚ ) to a 1.2825 A ˚ for the peak and inflection point resolution of 2.5 A ˚ for the remote wavelength. In wavelengths and to 2.3 A ˚ was collected addition, a high-resolution data set to 1.9 A at ID14-EH2 at a temperature of 40 K at a wavelength of ˚ , with helium cooling and a similar detector. The 0.933 A data sets were indexed and integrated by DENZO, and further processed by SCALEPACK50 and programs of the CCP4 program suite:51 SCALEIT52 and TRUNCATE.53 Details of the data processing are shown in Table 3. The protein crystallizes in space group P43212 with and c ¼ 84:322 A: unit cell dimensions of a ¼ 171:069A; Sixteen protein modules fit in the unit cell, one dimer per asymmetric unit. This corresponds to a Matthew’s coefficient of 2.8, with a calculated molecular mass of 53 kDa, corresponding to a solvent content of 47% (v/v).
The backbone of the protein was traced by ARP/ wARP.57 Side-chains were added manually with QUANTA 2000 (Accelrys). The model was refined with REFMAC58 against the map obtained from SOLOMON using the high-resolution data set with extended phases ˚ and all reflections. After an initial rigid body to 1.9 A refinement (without NCS) the overall R-factor was 0.38. In the following cycles of refinement, the atom positions and B-factors were refined isotropically. When the overall structure was built, including termini, water molecules and sugar moieties, there was still residual density (Fo 2 Fc . 3s ) around Zn1 in both monomers. The zinc ion had only three ligands, while it prefers a coordination of four or five ligands. The residual density had a flat shape and could fit a small molecule. Several approaches were taken in order to interpret this density: omit maps (excluding the active-site residues) have been calculated, all reflections were included (including the 5% of reflections that were used for calculating Rfree) and anomalous difference maps were calculated. In addition, successively collected data sets were inspected for radiation damage. Finally, various small molecules, which could have been present during the preparation of the enzyme or crystallization were modelled, but none was completely satisfactory. The possibility remains that the site is partially occupied by different ligands. Further residual density, which clearly did not belong to the protein, was fitted with either sulphate or maleate ions, which were both present in the crystallization buffer. Water molecules were added with the X-SOLVATE module of Quanta 2000 (Accelrys).
Phasing
Protein Data Bank accession number
The anomalous Pattersons showed significant and well-resolved peaks, up to a level of 23s. SOLVE54
Final atomic coordinates have been deposited in the RCSB Protein Data Bank with the entry code 1K7H.
Data collection
1.9 A˚ Crystal Structure of Heat-labile Shrimp Alkaline Phosphatase
Acknowledgments We thank the E.S.R.F. for providing synchrotron radiation facilities and Raimond Ravelli for assistance with data collection at beamlines ID14-EH4 and ID14EH2. We thank the staff of the Swiss-Norwegian beamlines (BM01) at the E.S.R.F., where many preliminary data collections were done, and Dr Inge W. Nilsen at the Norwegian Institute of Fisheries and Aquaculture, who kindly provided us with the gene sequence before publication.
References 1. Kim, E. E. & Wyckoff, H. W. (1991). Reaction mechanism of alkaline phosphatase based on crystal structures. Two-metal ion catalysis. J. Mol. Biol. 218, 449–464. 2. Moss, D. W. (1992). Perspectives in alkaline phosphatase research. Clin. Chem. 38, 2486– 2492. 3. Burlina, A., Rubin, D., Secchiero, S., Sciacovelli, L., Zaninotto, M. & Plebani, M. (1994). Monitoring skeletal cancer metastases with the bone isoenzyme of tissue unspecific alkaline phosphatase. Clin. Chim. Acta, 226, 151– 158. 4. Demers, L. M., Costa, L. & Lipton, A. (2000). Biochemical markers and skeletal metastases. Cancer, 88, 2919– 2926. 5. Henthorn, P. S. & Whyte, M. P. (1992). Missense mutations of the tissue-nonspecific alkaline phosphatase gene in hypophosphatasia. Clin. Chem. 38, 2501–2505. 6. Millan, J. L. (1992). Alkaline phosphatase as a reporter of cancerous transformation. Clin. Chim. Acta, 209, 123– 129. 7. Mornet, E. (2000). Hypophosphatasia: the mutations in the tissue-nonspecific alkaline phosphatase gene. Hum. Mutat. 15, 309– 315. 8. Narayanan, S. (1991). Serum alkaline phosphatase isoenzymes as markers of liver disease. Ann. Clin. Lab. Sci. 21, 12 – 18. 9. Knox, J. R. & Wyckoff, H. W. (1973). A crystallo˚ graphic study of alkaline phosphatase at 7– 7 A resolution. J. Mol. Biol. 74, 533– 545. 10. Hough, E., Hansen, L. K., Birknes, B., Jynge, K., Hansen, S., Hordvik, A. et al. (1989). High-resolution ˚ ) crystal structure of phospholipase C from (1.5 A Bacillus cereus. Nature, 338, 357– 360. 11. Volbeda, A., Lahm, A., Sakiyama, F. & Suck, D. (1991). Crystal structure of Penicillium citrinum P1 ˚ resolution. EMBO J. 10, 1607– 1618. nuclease at 2.8 A 12. Stec, B., Holtz, K. M. & Kantrowitz, E. R. (2000). A revised mechanism for the alkaline phosphatase reaction involving three metal ions. J. Mol. Biol. 299, 1303–1311. 13. Holtz, K. M., Stec, B. & Kantrowitz, E. R. (1999). A model of the transition state in the alkaline phosphatase reaction. J. Biol. Chem. 274, 8351– 8354. 14. Stec, B., Hehir, M. J., Brennan, C., Nolte, M. & Kantrowitz, E. R. (1998). Kinetic and X-ray structural studies of three mutant E. coli alkaline phosphatases: insights into the catalytic mechanism without the nucleophile Ser102. J. Mol. Biol. 277, 647– 662. 15. Chaidaroglou, A. & Kantrowitz, E. R. (1989). Alteration of aspartate 101 in the active site of Escherichia coli alkaline phosphatase enhances the catalytic activity. Protein Eng. 3, 127– 132.
1273
16. Chen, L., Neidhart, D., Kohlbrenner, W. M., Mandecki, W., Bell, S., Sowadski, J. & AbadZapatero, C. (1992). 3-D structure of a mutant (Asp101 ! Ser) of E. coli alkaline phosphatase with higher catalytic activity. Protein Eng. 5, 605– 610. 17. Butler-Ransohoff, J. E., Rokita, S. E., Kendall, D. A., Banzon, J. A., Carano, K. S., Kaiser, T. & Matlin, A. R. (1992). Active-site mutagenesis of E. coli alkaline phosphatase: replacement of Serine-102 with nonnucleophile amino acids. J. Org. Chem. 57, 142 –145. 18. Ghosh, S. S., Bock, S. C., Rokita, S. E. & Kaiser, E. T. (1986). Modification of the active site of alkaline phosphatase by site-directed mutagenesis. Science, 231, 145– 148. 19. Holtz, K. M. & Kantrowitz, E. R. (1999). The mechanism of the alkaline phosphatase reaction: insights from NMR, crystallography and site-specific mutagenesis. FEBS Letters, 462, 7 – 11. 20. Sowadski, J. M., Handschumacher, M. D., Murthy, H. M., Kundrot, C. E. & Wyckoff, H. W. (1983). Crystallographic observations of the metal ion triple in the active site region of alkaline phosphatase. J. Mol. Biol. 170, 575– 581. 21. Sowadski, J. M., Handschumacher, M. D., Murthy, H. M., Foster, B. A. & Wyckoff, H. W. (1985). Refined structure of alkaline phosphatase from Escherichia coli ˚ resolution. J. Mol. Biol. 186, 417– 433. at 2.8 A 22. Le Du, M. H., Stigbrand, T., Taussig, M. J., Menez, A. & Stura, E. A. (2001). Crystal structure of alkaline ˚ resophosphatase from human placenta at 1.8 A lution. Implication for a substrate specificity. J. Biol. Chem. 276, 9158– 9165. 23. Nilsen, I. W., Overbo, K. & Olsen, R. L. (2001). Thermolabile alkaline phosphatase from Northern shrimp (Pandalus borealis ): protein and cDNA sequence analyses. Comp. Biochem. Physiol. B Biochem. Mol. Biol. 129, 853– 861. 24. Bradshaw, R. A., Cancedda, F., Ericsson, L. H., Neumann, P. A., Piccoli, S. P., Schlesinger, M. J. et al. (1981). Amino acid sequence of Escherichia coli alkaline phosphatase. Proc. Natl Acad. Sci. USA, 78, 3473 –3477. 25. Kam, W., Clauser, E., Kim, Y. S., Kan, Y. W. & Rutter, W. J. (1985). Cloning, sequencing, and chromosomal localization of human term placental alkaline phosphatase cDNA. Proc. Natl Acad. Sci. USA, 82, 8715 –8719. 26. Russell, R. J., Gerike, U., Danson, M. J., Hough, D. W. & Taylor, G. L. (1998). Structural adaptations of the cold-active citrate synthase from an Antarctic bacterium. Structure, 6, 351– 361. 27. Karlsen, S., Hough, E. & Olsen, R. L. (1998). Structure and proposed amino-acid sequence of a pepsin from Atlantic cod (Gadus morhua ). Acta Crystallog. sect. D, 54, 32 –46. 28. Smalas, A. O., Heimstad, E. S., Hordvik, A., Willassen, N. P. & Male, R. (1994). Cold adaption of enzymes: structural comparison between salmon and bovine trypsins. Proteins: Struct. Funct. Genet. 20, 149–166. 29. Berglund, G. I., Willassen, N. P., Hordvik, A. & Smalas, A. O. (1995). Structure of native pancreatic ˚ resoelastase from North-Atlantic samlon at 1.61 A lution. Acta Crystallog. sect. D, 51, 925– 937. 30. Alvarez, M., Zeelen, J. P., Mainfroid, V., RentierDelrue, F., Martial, J. A., Wyns, L. et al. (1998). Triose-phosphate isomerase (TIM) of the psychro-
1274
31.
32.
33. 34.
35.
36.
37.
38. 39.
40.
41.
42.
43.
44.
˚ Crystal Structure of Heat-labile Shrimp Alkaline Phosphatase 1.9 A
philic bacterium Vibrio marinus. Kinetic and structural properties. J. Biol. Chem. 273, 2199– 2206. Aghajari, N., Feller, G., Gerday, C. & Haser, R. (1998). Structures of the psychrophilic Alteromonas haloplanctis alpha-amylase give insights into cold adaptation at a molecular level. Structure, 6, 1503– 1516. Kim, S. Y., Hwang, K. Y., Kim, S. H., Sung, H. C., Han, Y. S. & Cho, Y. (1999). Structural basis for cold adaptation. Sequence, biochemical properties, and crystal structure of malate dehydrogenase from a psychrophile Aquaspirillium arcticum. J. Biol. Chem. 274, 11761– 11767. Smalas, A. O., Schroder-Leiros, H.-K., Os, V. & Willassen, N. P. (2000). Cold adapted enzymes. Biotechnol. Ann. Rev. 6, 1 – 57. Asgeirsson, B., Hartemink, R. & Chlebowski, J. F. (1995). Alkaline phosphatase from Atlantic cod (Gadus morhua ). Kinetic and structural properties which indicate adaptation to low temperatures. Comp. Biochem. Physiol. 110B, 315– 329. Asgeirsson, B., Hauksson, J. B. & Gunnarsson, G. H. (2000). Dissociation and unfolding of cold-active alkaline phosphatase from Atlantic cod in the presence of guanidinium chloride. Eur. J. Biochem. 267, 6403– 6412. Hauksson, J. B., Andresson, O. S. & Asgeirsson, B. (2000). Heat-labile bacterial alkaline phosphatase from a marine Vibrio sp. Enzyme Microb. Technol. 27, 66 –73. Rina, M., Pozidis, C., Mavromatis, K., Tzanodaskalaki, M., Kokkinidis, M. & Bouriotis, V. (2000). Alkaline phosphatase from the Antarctic strain TAB5. Properties and psychrophilic adaptations. Eur. J. Biochem. 267, 1230– 1238. Sheldrick, G. M., Dauter, Z., Wilson, K. S. & Sieker, L. C. (1993). Acta Crystallog. sect. D, 49, 18 – 23. Laskowski, R. A., MacArthur, M. W., Moss, D. S. & Thornton, J. M. (1993). PROCHECK: a program to check the stereochemical quality of protein structures. J. Appl. Crystallog. 26, 283– 291. Butler-Ransohoff, J. E., Kendall, D. A. & Kaiser, E. T. (1988). Use of site-directed mutagenesis to elucidate the role of arginine-166 in the catalytic mechanism of alkaline phosphatase. Proc. Natl Acad. Sci. USA, 85, 4276– 4278. Chaidaroglou, A., Brezinski, D. J., Middleton, S. A. & Kantrowitz, E. R. (1988). Function of arginine-166 in the active site of Escherichia coli alkaline phosphatase. Biochemistry, 27, 8338– 8343. Murphy, J. E., Tibbitts, T. T. & Kantrowitz, E. R. (1995). Mutations at positions 153 and 328 in Escherichia coli alkaline phosphatase provide insight towards the structure and function of mammalian and yeast alkaline phosphatases. J. Mol. Biol. 253, 604– 617. Cathala, G. & Brunel, C. (1975). Bovine kidney alkaline phosphatase. Catalytic properties, subunit interactions in the catalytic process, and mechanism of Mg2þ stimulation. J. Biol. Chem. 250, 6046– 6053. Hung, H. C. & Chang, G. G. (2001). Differentiation of the slow-binding mechanism for magnesium ion
45.
46.
47.
48. 49.
50. 51. 52. 53. 54. 55. 56.
57.
58.
59.
60.
activation and zinc ion inhibition of human placental alkaline phosphatase. Protein Sci. 10, 34 – 45. Zhang, R.-Q., Chen, Q.-X., Xiao, R., Xie, L.-P., Zeng, X.-G. & Zhou, H.-M. (2001). Inhibition kinetics of green crab (Scylla serrata ) alkaline phosphatase by zinc ions: a new type of complexing inhibition. Biochim. Biophys. Acta, 1545, 6– 12. Janeway, C. M., Xu, X., Murphy, J. E., Chaidaroglou, A. & Kantrowitz, E. R. (1993). Magnesium in the active site of Escherichia coli alkaline phosphatase is important for both structural stabilization and catalysis. Biochemistry, 32, 1601– 1609. Murphy, J. E., Xu, X. & Kantrowitz, E. R. (1993). Conversion of a magnesium binding site into a zinc binding site by a single amino acid substitution in Escherichia coli alkaline phosphatase. J. Biol. Chem. 268, 21497– 21500. Nicholls, A., Sharp, K. & Honig, B. (1991). Proteins: Struct. Funct. Genet. 11, 281. Schroder Leiros, H. K., Willassen, N. P. & Smalas, A. O. (2000). Structural comparison of psychrophilic and mesophilic trypsins. Elucidating the molecular basis of cold-adaptation. Eur. J. Biochem. 267, 1039– 1049. Otwinowski, Z. & Minor, W. (1997). Methods Enzymol. 276, 307– 326. Collaborative Computational Project Number 4 (1994). The CCP4 suite: programs for protein crystallography. Acta Crystallog. sect. D, 50, 760– 763. Howell, P. L. & Smith, G. D. (1992). Identification of heavy-atom derivatives by normal probability methods. J. Appl. Crystallog. 25, 81 – 86. French, G. S. & Wilson, K. S. (1978). On the treatment of negative intensity oberservations. Acta Crystallog. sect. A, 34, 517. Terwilliger, T. C. & Berendzen, J. (1999). Automated MAD and MIR structure solution. Acta Crystallog. sect. D, 55, 841– 861. La Fortelle, E. d. & Bricogne, G. (1997). Methods Enzymol. 276, 472– 494. Abrahams, J. P. & Leslie, A. G. W. (1996). Methods used in the structure determination of bovine mitochondrial F1 ATPase. Acta Crystallog. sect. D, 52, 30 –42. Perrakis, A., Sixma, T. K., Wilson, K. S. & Lamzin, V. S. (1997). wARP: improvement and extension of crystallographic phases by weighted averaging of multiple refined dummy atomic models. Acta Crystallog. sect. D, 53, 448– 455. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. (1997). Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallog. sect. D, 53, 240– 255. Jones, T. A., Zou, J. Y., Cowan, S. W. & Kjelgaard, M. (1991). Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallog. sect. A, 47, 110 – 119. Esnouf, R. M. (1997). An extensively modified version of MolScript that includes greatly enhanced coloring capabilities. J. Mol. Graph. 15, 132– 134.
Edited by R. Huber (Received 26 October 2001; received in revised form 25 January 2002; accepted 28 January 2002)