- Email: [email protected]
Structure of alkaline phosphatases
Clinica Chimica Acru, 186 (1989) 175-188 Else&r
175
CCA 04643
Structure and Function of Alkaline Phosphatases
Structure of alkaline phosphatases Eunice E. Kim and Harold W. Wyckoff Department of Molecular Biophysics and Biochemistry, Yale Universify, New Haven CT (USA) (Received 14 June 1989; revision received and accepted 18 September 1989) Key WOW%: Alkaline phosphatase; Three dimensional structure; Structure refinement; Prediction
The crystal structure of alkaline phosphatase (AP) from Escherichia coii, which is a prototype fpr rn~~~ APs, has been refined to a c~s~o~ap~c R-factor of 0.184 at 2.0 A resolution. During the course of the refinement residues 380 to 410 were retraced and I90 to 200 were shifted by one residue, and substantial changes in the active site of the enzyme were made. Based on the refined structure and the sequences of mammalian enzymes (25-30% strict homology) we have modelled the core of the three dimensional structures of the mammalian alkaline phosphatases. Considerable circ~t~ti~ evidence suggests that this is valid despite the fact that the mammalian enzymes are larger, contain carbohydrate and are membrane associated through a phosphatidylinositol moeity. The active site of the molecule is highly conserved but specific changes in the secondary ligands to bound phosphate and the Mg metal are observed. Introduction Alkaline phosphatase (E.C. 3.1.3.1) is a nonspecific phosphomonoesterase that hydrolyzes phosphate monoesters, ROP, at approximately the same rate regardless of the nature of the R group or the pK, of the leaving group, ROH. Although its specific function in mammals is not known at the present time, the fact that the enzyme is ubiquitous - found in organisms from humans to bacteria - and that it is found attached to the plasma membranes where extensive transport takes place, indicate that alkaline phosphatase is involved in fundamental biological processes. In humans, there are at least four distinct forms of alkaline phosphatase and changes of their levels in blood serum have been used for many years as a diagnostic tool for various diseases (see [l-6] for review).
Correspondence to: Dr. E.E. Kim, Department University, New Haven, CT 06511, USA. ~-8981/89/~3.5~
of Molecular Biophysics and Biochemistry,
@ 1989 Eisevier Science Publishers B.V. (Biomedical Division)
Yale
176
In the past three years several mammalian isozymes have been cloned and sequenced [7-121, but X-ray diffraction structure data are only available for Escherichia coli alkaline phosphatase [13,14]. The E. coli enzyme has been studied most extensively by a variety of physico-chemical techniques and is thus useful as a prototype. This enzyme is synthesized as a precursor with a signal peptide on loosely attached inner membrane ribosomes and is transported into the periplasmic space, where it ultimately appears in the dimeric form (M, = 94 kDa) with the signal peptide removed [15,16] and three metal ions sequestered in each of the two active site regions. The function of the E. coli enzyme is clearly part of the phosphate acquisition and transport system in the cell, induced along with many other proteins, by deprivation of free inorganic phosphate. The previously reported three-dimensional structure was at 2.8 A resolution from this laboratory [13,14]. However, there have been long standing uncertainties about the metal-ligand constellations due to poorly defined electron density in the active site region. Also the top portion (see Fig. 2a) of the molecule (residues 380 to 410) was essentially undefined in the electron density map. Recently we have extended the resolution to 2.0 A with completely new data collected with vastly improved technology. The structure has been revised and refined. Based on the refined structure of the E. cofi enzyme we have examined possible three dimensional structures of mammalian alkaline phosphatases employing the reported sequences. Experimental
section
Data collection Alkaline phosphatase from E. coli was purified and crystallized as described earlier [13,17]. All the diffraction data reported here have been collected using the San Diego Multiwire System area detectors at Yale. Data were collected out to 2.0 A resolution on two crystals soaked in stabilizing solution containing 2 mmol/l inorganic phosphate, 100 mmol/l Tris, 10 mmol/l Zn2’ and 10 mmol/l Mg2+ at pH 7.5. The inorganic phosphate was added primarily because it is a product and a competitive inhibitor of the enzyme, but also to enhance the metal binding. A total of 78,398 unique reflections was obtained from 422,395 actual measurements with an average five-fold redundancy including crystallographic symmetry. This is over 92% of possible reflections for this resolution and the R sym, the mean agreement between the equivalent measurements, was 4.7%. Refinement Refinement was carried out using both the stereochemically restrained least-squares refinement method (PROFFT [18]) and the simulated annealing refinement method (XPLOR [19,20]). The later involves molecular dynamics simulation at elevated temperature using diffraction data in combination with energy functions to overcome local minima. Progress was slow and frequent model-building interpretation (human intervention using computer graphics presentations of electron density maps) using the program FRODO on Evans and Sutherland PS300 graphics systems was necessary to correct misplaced atoms. Details of refinement will be reported elsewhere.
Fig. 1. Comparison of amino acid sequence to yield maximum number of identities. If involved in metal binding areindicated by (A) and cysteines (0)
of alkaline phosphatases. Deletions and insertions are made conserved in all five only E. coli sequence is given. Residues 0. Also putativeglyeosylationsites(A),hltron-exonjunctions in the mammalian sequences are shown.
Sequence comparison The amino acid sequence of human term placental [7-91, human intestinal [lo], human liver/bone/kidney [ll] and rat osteosarcoma [12] alkaline phosphatases deduced from cDNA sequences have been aligned with the sequence of the E. coli enzyme [21,22] allowing appropriate deletions and insertions (see Fig. 1). With only 25-3041 identity between E. coli and the mammalian sequences there exist several ways of alignment. Our version tries to avoid disruption of secondary structures since a single addition or deletion will switch subsequent residues to opposite sides of a P-sheet or rotate residues 100” along an a-helix. Results
Results of both the E. cob AP refinement and modelling of the mammalian enzymes are described below in each section dealing with the overall structure, the main &sheet, the active site of the enzyme and other various points on the mammalian enzymes. Results of refinement The final model has converged to a conventional R-factor (fractional residual discrepancy between observed and calculated structure amplitudes) of 0.184 with good stereochemistry (r.m.s. deviations: 0.015 A in bond
178
(b)
179
distances and 2.9 ’ in bond angles). During this refinement residues 380 to 410 were retraced and residues 190 to 200 were shifted by one residue, in addition to numerous other minor modifications. Also changes were made at the active site region of the protein. A total of 296 water molecules was located as well as a tightly bound phosphate in each active site of the enzyme. Most of the electron density map is now clearly interpretable except for 4 residues at the amino-end of the polypeptide chain and another 4 or 5 residues on the top portion (see Fig. 2a) of the molecule (404-408). Some sidechains, lysines in particular, are not well defined in the map but they are mostly on the surface of the molecule, sporadic and not in crucial positions. The map in the active site region is now clear and the metal-ligand constellations are well defined. The hydroxyl of Ser102, which is phosphorylated and dephosphorylated, is disordered. Overall structure Although the structure has been modified during the refinement, most the earlier description of the structure [16,23] remains the same. The enzyme is a basically 2-fold symmetrical dimer (97 X 47 X 52 A3) with two active sites located about 30 A from each other. Figure 2a shows the C, trace of the complete dimer with metals shown as spheres. Both the amino- and carboxyl-ends are at the bottom of the molecule as indicated in Fig. 2a. Each subunit (449 amino acid residues) shows typical a/P topology with a ten-stranded p-sheet at the center (see Fig. 2b) flanked by 15 helices with various lengths. In contrast to earlier descriptions residues 325 to 334 form a well-behaved a-helix while residues 425 to 430 are not helical. After retracing the chain the top portion contains a three-stranded sheet and a short helix. Therefore the structure consists of the central ten-stranded B-sheet and a minor three-stranded P-sheet on the top as well as sixteen helices. This constitutes about 55% of the protein while the rest of the molecule forms loops, coils, hairpins and other hard-to-describe regions. Further aspects of the structure are discussed below in conjunction with the placental enzyme modelling. Result of sequence comparison The results of the sequence comparisons are similar to what have been ‘observed by others: the eukaryotic isozymes show high homology amongst themselves, i.e. 2 50% overall and 2 80% in subgroups, but when compared to the E. coli enzyme they show 25-30% identities [11,24]. Unless specifically mentioned sequences numbers of the E. coli enzyme are given in the following comparisons. The conserved residues are distributed mainly in the central a/P portion of the molecule while only few are found in the interfacial region of the molecule as seen in Fig. 2b.
Fig. 2. a. C, trace of the complete dirner of E: coli alkaline phosphatase (drawn using SZAZAM developed by A. Perlo). Tube is drawn with 0.5 A radius and the three metals are shown as spheres. Strand G of the central beta-sheet (see Fig. 3) is indicated. The molecule is viewed with its noncrystallographic two-fold axis vertical and the maximum dimension of the molecule horizontal. This is referred to as front view in the text. C,-C, distances are 3.8 A. b. Conserved residues in all five (E. coli, human placental, human intestinal, human liver/kidney/bone and rat osteosarcoma alkaline phosphatases are shown as 0 on C,, trace of monomer, and the metals are shown as stippled spheres.
180
sequence comparison J
B
i
H
A
in P-sheet G
c
HIS
PRO T/S
DIR
D
F
E
HIS T/A
“s6THR
GLY
$$I++-~
W
HIS
GLY
ASP
GLN
Fig. 3. Residues in the central &sheet (shown inside the box) and ones involved in metal binding (shown in bold face) are compared. Strands are named A through 3 in sequence order, while they are shown in their physical order. All are parallel except one (i). If residues are conserved three letter codes are given while one letter codes (in the fotm of E. c&/pIacental) are given othetwise. Additionally boxed residues are to highlight changes discussed in the text.
All alkaline phospbatases are synthesized with a leading signal peptide which is removed after transport of the enzyme either to the periplasmic space in E, culi or on route to the cytoplasmic membrane in the case of enkaryutic enzymes. However, unlike the E. coli enzyme, mammalian alkaline phosphatases are anchored to the cytoplasmic membrane. Recently the placental enzyme has been shown to be anchored to the membrane at position Asp484 by a mechanism involving a phosphatidylinositol-glycan anchoring moiety with the carboxyl terminal sequence removed in post translational processing [25,26]. Others are also believed to be anchored in the similar manner. As shown in Fig. 2a both the carboxyl- and the amino-ends of the molecule are pointing outwards, well away from the active site as well as the core of the molecule. Thus extensions of the chain at either end, such as in the case of mammals at the carboxyl end or fusion proteins [27] at the amino end, should not affect the structure or function grossly. a/@ portion of the naolecule A comparison of the residues in the 10 stranded &sheet is shown in Fig. 3. If they are identical, three letter codes are given; otherwise one letter codes are given in the order of E. coli/placental enzymes. Residues inthe &sheet are shown inside the box and the ones involved in the metal binding are shown in bold face. Strands are named A through J in sequence order,
181
Fig. 4. Active site of the E. coli alkalinephosphatase. Some residues and water molecules are omitted for clarity. In the mammalian sequences all the residues involved in the metal binding are conserved except for Thr155, which is replaced by serine in some’cases. Others in this region are also conserved as well except for a few, e.g. 153 (Asp to His) and 328 (Lys to His) which are shown as single lines in modelled positions.
while they are shown in their physical order. The strands are parallel except for one antiparallel insertion (strand i). The sheet is largely classical in that it is connected in a right-handed sense as one proceeds from one strand to the next while the twist of the sheet is left-handed. The structure is boat shaped, narrow at the bottom and broader at the top both in and perpendicular to the plane of Figs. 2a and 3. The monomer is considerably thicker on one side of the &sheet as seen in the top views of Fig. 5. The active site is located at the carboxyl end of the &sheet, and a detailed view is shown in Fig. 4. As seen in Fig. 3, the strands in the core of the molecule, i.e. strands A and G, have larger number of identities than the shorter strands on the surface of the molecule, i.e. strands B and E. The p-sheet can be considered as having horizontal rows of residues facing one side of the sheet alternating with rows of residues facing the opposite side (as indicated by shading in Fig. 3). When one considers these two surfaces separately, there are only 16 (52%) and 14 (40%) identities on each surface.
182
(b)
Fig. 5. C, trace of monomer. Panels on the left are drawn with the non-crystallographic two-fold axis vertical (referred to as front view) while the ones on the right are drawn looking down the two-fold axis (referred to as top view). a. Insertions/deletions sites in the marmnak sequences. Insertion sites are shown as 0 while the segments deleted are shown with heavier lines b. Intron-exon junctions of the mammalian enzymes are shown as 0 on C, trace of one subunit with central &strands indicated by heavier lines. c. Disulfides in E. coli are indicated by heavy lines and labekd a and b. Putative glycosylation sites at Asn positions are shown by 0. Ala161 is shown as a stippled sphere.
183
However, one can consider many of the replacements as structurally conservative, and some of these are boxed. Replacements of Gln to Arg at 83 and Arg to Ala at 418 can be considered paired conservative changes since they are adjacent on the same side of the p-sheet. Again replacements of Ile to Leu at 365 and Leu to Ile at 47 are another pairwise conservative couple. If one includes these and other conservative replacements, i.e. Leu to Val or Phe and Asp to Glu, then the homology fractions on the two surfaces increase to 77% and 71%, respectively. The cluster of replacements in the region at 320, 145, 143, and 203 is particularly interesting since the somewhat anomalous hydrophilic patch on the E. coli enzyme is replaced with more classical hydrophobic surface. Active site The active site region can be considered as AsplOl-Ser102-Ala103 and the metal triplet (two Zn and one Mg) and their ligands as well as Arg166 and others in the vicinity. This is shown in Figure 4 where some water molecules and residues are omitted for clarity. Phosphate and water molecules found in this region during the refinement are involved in an extensive hydrogen bonding scheme. Znl is penta-coordinated by the imidazole nitrogen atoms of two histidines, His331 and His412, both carboxyl oxygens of Asp227 and one of the phosphate oxygens with average metal-ligand distance of 2;P7 A. His372 which was originally thought to be a direct ligand to Znl is not (3.8 A away from Znl), but is hydrogen bonded to one of the carboxyl oxygens of Asp327. Zn2 is coordinated tetrahedrally by the imidazole nitrogen of His370, one of the carboxyl oxygens of AspSl(OD1) and of Asp369(ODl) and one of the phosphate oxygens. The average metal-ligand distance is 2.00 A. In the absence of phosphate the hydroxyl (OG) of the SerlO2 is a ligand to this metal. Mg-coordination can be described as a slightly distorted octahedron with the second carboxyl oxygen of Asp51(OD2) and one of the carboxyl oxygens of Glu322(OEl) and hydroxyl of Thr{55(OG) and three water molecules as ligands. The average Mg-0 distance is 2.12 A. Asp153 is not a direct ligand to this metal as previously indicated, but is an indrect ligand hydrogen-bonded to two of the three water molecules that are coordinating Mg. Phosphate is coordinated to both Znl and Zn2, and the other two oxygens are tightly held by two amino functions of the guanidinium group of ArgZ66, which is in turn further H-bonded to Asp101 and a water that is held by Asp153 and Tyr169. The phosphate is further H-bonded to the amide of Ser102 and a water molecule that is coordinated to Mg and another water that is bridging to Lys328. The van der Waals surface on this region shows that the active site pocket is just big enough to hold the phosphate end of the substrate thus leaving the remainder of the molecule in the solvent region. The residues 101 to 112 form a helical stem that is buried between the B-sheet and other helices and loops and this provides a firm anchor for the active Ser102. The hydroxyl group of SerlO2 is disordered. The carboxyl group of Asp101 is positioned such that there is no direct contact with either serine or the phosphate but it appears that it is functional in positioning the guanidinium group of Arg166 and providing part of the electrostatic environment.
184
In the case of the mammalian enzymes the Asp-Ser-Ala/Gly sequence at the phosphoryl acceptor is conserved. This sequence also exists in serine proteases in which the serine is an acyl acceptor. All the residues serving as direct ligands to the metals (bold faced in Fig. 3) are conserved except for fully acceptable replacement at position 155 where Thr is replaced by Ser in some cases. Also Arg166, which complexes to phosphate in the active site, is conserved. Thus the structure of the active site region of the enzyme by implication appears highly conserved even in the case of Mg-coordination whose role in catalysis is not clear. This suggests a real importance of this site as well. However, it is interesting to note that both Lys 328 and Asp 153 which are above the Mg-coordination sphere are replaced by histidines in all mammals as shown in Fig. 4. The mammalian enzymes are 20- to 30-fold more active than the E. coli enzyme when they are at their pH optima and under conditions of substrate saturation. Insertion and deletion sites In the mammalian enzymes there are several places where insertions and deletions take place, and these are shown in Fig. 5a. Many of these occur on the surface loop regions of the molecule away from the active site. There are three major insertions at the positions 125, 404 and 411 involving 8, 16 and 13 residues respectively. The insertion sites 87, 404 and 411 are near the active site. Residue 411 is next to His412 which is a direct ligand to Znl. These insertions at 404 and 411 considerably expand the upper region of the molecule and the variability in this region among isozymes might indicate different roles. All the deletion sites are located at least 20 A away from the metal triplet. The deletion of 10 residues between 32 and 41 (See Fig. 5a) occurs at the bottom of the interface region that includes part of helix 29-37, which is followed by the important central strand (strand A) of the P-sheet. However, this is not as serious as one might expect since the region between 16 and 23 in the placental sequence (and others) is strongly predicted to be a helix, based on the algorithm of Gamier et al. [28]. This might replace structurally the 29-37 helix of the E. coli leaving a slightly longer connecting loop (either 6 or 7 residues instead of 5 residues) before the first /3 strand. The deficit then would be at the amino terminal which could be readily accommodated. Intron-exon structure In humans there are at least three genes: placental, intestinal and liver/kidney/bone. In all three genes the intron-exon junctions occur at analogous positions, although the gene for liver/kidney/bone is larger than the other two [29]. As seen in Fig. 5b all the intron-exon junctions occur at either on or near the surface residues and at or near the residues where changes in secondary structure take place. Similar observations have been made in other proteins [30,31]. The ones at residues 52 and 323 are near the active site of the enzyme. Glycosylation sites Unlike the E. coli enzyme the mammalian enzymes are glycosylated and there are five putative sites (Asn-xxx-Tbr/Ser) reported in the liver/bone/kidney enzyme [11,12] while only two are retained in the placental and intestinal sequences. All except one of the putative glycosylation sites occur on the
185
surface of the molecule (see Fig. 5c) and that one is in a poorly conserved region which could easily be distorted. Disulfides All four cysteines in the E. coli enzyme are involved in forming two disulfide bridges, both away from the active site of the enzyme as seen in Fig. 5c. All four are replaced in the mammalian enzymes. There are five cysteines in mammalian enzymes indicating the presence of at least one free cysteine. The C, positions of 111 and 129 in the E. coli enzyme, equivalent to 101 and 121 in the placental sequence, are at a distance of possible disulfide bridge formation. Hypophosphatasia mutants A number of hypophosphatasia patients suffering from severe to mild skeletal and related disorders are reported to have mutations in the bone alkaline phosphatase gene. One has been clearly shown to be a single point mutant at the residue corresponding to Ala161 in the E. coli enzyme, changing it to threonine [32]. Residues 161 and 162 are conserved (see Fig. 1). The structure of this region is shown in Fig. 6 and is located at the end of a short helix away from the active site as shown in Fig. 5c, yet the mutation produces a molecule with little activity. The C, of Ala161 is in a hydrophobic core and His162 is mostly inaccessible with well restrained hydrogen bonds. Ala161 is between residues 155 and 166 which are both involved in the active site. The local distortion needed to accommod-
Fig. 6. Environment of Ala161 of E. coli enzyme. His162 is in a depression on the surface. with little solvent accessibility. Ala161 is inaccessible and C, is pointing back into a hydrophobic cluster.
186
ate the mutant might propagate to Arg166, since site directed mutations of Arg166 are known to have substantial effects on activity in the E. coli enzyme [33,34]. An alternative explanation would be that proper folding is prevented resulting in a more extensive disruption. Conclusion Based on above observations one can conclude that although there is only 25-3048, strict sequence homology between mammalian and E. coli alkaline phosphatases it appears that the mammalian enzymes retain the core of the three-dimensional structure of the E. coli enzyme. The active site region is highly conserved with some specific changes in the secondary ligands. Many changes in the ten-stranded b-sheet are structurally conservative. Identities are clustered in the core of the structure while surface residues and segments further from the active site are less homologous. None of the insertions, deletions, carbohydrate attachment sites or exon boundaries pose serious problems. Of course confirmation of these observations and the ramifications of the differences have to wait until a detailed crystal structure analysis of the mammalian enzymes is available. Acknowledgements
We thank Drs. J.E. Coleman for providing the protein, A.T. Brtinger for extensive use of his program XPLOR, S.K. Kattti for helpful discussions during refinement, A. Perlo for use of his program SZA-ZAM and J. Ponder for use of his version of program predicting the secondary structure, and J. Mouning for help in preparing the manuscript. This work was supported by Public Health Service grant GM22778 from National Institute of General Medical Science. References 1 2 3 4 5 6 7
8 9
10
McComb RB, Bowers ON, Posen S. Alkaline phosphatases. New York: Plenum, 1979. Reid TW, Wilson IW. E. co/i alkaline phosphatase. Enzymes 1971;4:373-416. Femley HN. Mammalian alkaline phosphatases. Enzymes 1971;4:417-447. Coleman JE, Gettins P. Alkaline phosphatase, solution structure and mechanism. Adv Enzymol Relat Areas Mol Biol 1983;55:381-452. Harris H. Multilocus enzyme systems and the evolution of gene expression: the alkaline phosphatases as a model example. Harvey Lecture Ser 1982;76:95-124. Moss DW. Genetics, expression, and modification in the human alkaline phosphatases. Isozymes 1987;16:67-80. Kam W, Clauser E, Kim YS, Kan YW, Rutter WJ. Cloning, sequencing, and chromosomal localization of human term placental alkaline phosphatase cDNA. Proc Nat1 Acad Sci USA 1985;82:8715-8719. Mill&n JL. Molecular cloning and sequence analysis of human placental alkaline phosphatase. J Biol Chem 1986;261:3112-3115. Henthom PS, Knoll BJ, Raducha M, et al. Products of two common alleles at the locus for human placental alkaline phosphatase differ by seven ammo acids. Proc Nat1 Acad Sci USA 1986;83:5597-5601. Berger J, Garattini E, Hua J-C, Udenfriend S. Cloning and sequencing of human intestinal alkaline phosphatase cDNA. Proc Nat1 Acad Sci USA 1987;84:695-698.
187 11 Weiss MJ, Henthom
12
13 14
15 16 17 18 19 20 21 22
PS, Lafferty MA, Slaughter C, Raducha M, Harris H. Isolation and characterization of a cDNA encoding a human liver/bone/kidney-type alkaline phosphatase. Proc Nat1 Acad Sci USA 1986;83:7182-7186. Thiede MA, Yoon K, Golub EE, Noda M, Rodan GA. Structure and expression of rat osteosarcoma (ROS 17/2.8) alkaline phosphatase: product of a single copy gene. Proc Natl Acad Sci USA 1988;85:319-323. Sowadski JM, Handschumacher MD, Murthy HMK, Foster BA, Wyckoff HW. Refined structure of alkaline phosphatase from Escherichiu coli at 2.8 A resolution. J Mol Biol 1985;186:417-433. Wyckoff HW. Structure of Escherichia coli alkaline phosphatase determined by X-ray diffraction in Phosphate metabolism and cellular regulation in microorganisms American Society for Microbiology 1987;118-126. Chang CN, Inouye H, Model P, Beckwith J. Processing of alkaline phosphatase precursor to the mature enzyme by and Escherichia coli inner membrane preparation. J Bacterial 1980;142:726-728. Inouye H, Barnes W, Beckwith J. Signal sequences of alkaline phosphatase of Escherichia coli. J Bacterial 1982;149:434-439. Appleburry ML, Johnson BP, Coleman JE. Phosphate binding to alkaline phosphatase. J Biol Chem 1970;245:4968. Finzel BC. Incorporation of fast Fourier transforms to speed restrained least-squares refinement of protein structures. J Appl Cryst 1987;20:53-55. Brtinger AT, Kuriyan J, Kaplus M. Crystallographic R factor refinement by molecular dynamics. Science 1987;235:458-460. Briinger AT, Karplus M, Petsko GA. Crystallographic refinement by simulated annealing: Application to crambin. Acta Cryst 1989;A45:50-61. Bradshaw RA, Cancedda F, Ericsson LH, et al. Amino acid sequence. of Escherichia coli alkaline phosphatase. Proc Nat1 Acad Sci USA 1981;78:3473-3477. Chang CN, Kuang W-J, Chen EY. Nucleotide sequence of the alkaline phosphatase gene of
Escherichiu coli K-12. Gene 1986;44:121-125. 23 Wyckoff HW, Handschumacher MD, Murthy HMK, Sowadski JM. The three dimensional structure of alkaline phosphatase from E. co& Adv Eruymol Relat Areas Mol Biol 1983;55:453-480.
24 Mill&n JL. Gncodevelopmental expression and structure of alkaline phosphatase genes. Anticancer Res 1988;8:995-1004. 25 Low MG, Saltiel AR. Structural and functional roles of glycosylphosphatidylinositol in membranes. Science 1988;239:268-275. 26 Micanovic R, Bailey CA, Brink L, Gerber L, Pan Y-CE, Hulmes JD, Udenfriend S. Aspartic acid-484 of nascent placental alkaline phosphatase condenses with a phosphatidylinositol glycan to become the carboxyl terminus of the mature enzyme. Proc Nat1 Acad Sci USA 1988;85:1398-1402. 27 Hoffman CS, Wright A. Fusions of secreted proteins to alkaline phosphatase: an approach for studying protein secretion. Proc Nat1 Acad Sci USA 1985;82:5107-5111. 28 Gamier J, Osguthorpe DJ, Robson B. Analysis of the accuracy and implications of simple methods for predicting secondary structure of globular proteins. J Mol Biol 1978;120:97-120. 29 Weiss MJ, Ray K, Henthom PS, Lamb B, Kadesch T, Harris H. Structure of the human liver/bone/kidney alkaline phosphatase gene. J Biol Chem 1988;263:12002-12010. 30 Craik CS, Sprang S, Fletterick R, Rutter WJ. Intron-exon splice junctions map at protein surfaces. Nature 1982;299:180-182. 31 Bajaj M, Blundell T. Evolution and tertiary structure of proteins. Arm Rev Biophys Bioeng 1984;13:453-492. 32 Weiss MJ, Cole DEC, Ray K, Whyte MP, Lafferty MA, Mulivor RA, Harris H. A m&sense mutation in the human liver/bone kidney alkaline phosphatase gene causing a lethal form of hypophosphatasia. Proc Nat1 Acad sci USA 1988;85:7666-7669. 33 Chaidaroglou A, Brezinski DJ, Middleton SA, Kantrowitz ER. Function of arginine-166 in the active site of Escherichia coli alkaline phosphatase. Biochemistry 1988;27:8338-8343. 34 Butler-Ransohoff JE, Kendall DA, Kaiser ET. Use of site-directed mutagenesis to elucidate the role of arginine-166 in the catalytic mechanism of alkaline phosphatase. Proc Natl Acad Sci USA 1988;85:4276-4278.
175
CCA 04643
Structure and Function of Alkaline Phosphatases
Structure of alkaline phosphatases Eunice E. Kim and Harold W. Wyckoff Department of Molecular Biophysics and Biochemistry, Yale Universify, New Haven CT (USA) (Received 14 June 1989; revision received and accepted 18 September 1989) Key WOW%: Alkaline phosphatase; Three dimensional structure; Structure refinement; Prediction
The crystal structure of alkaline phosphatase (AP) from Escherichia coii, which is a prototype fpr rn~~~ APs, has been refined to a c~s~o~ap~c R-factor of 0.184 at 2.0 A resolution. During the course of the refinement residues 380 to 410 were retraced and I90 to 200 were shifted by one residue, and substantial changes in the active site of the enzyme were made. Based on the refined structure and the sequences of mammalian enzymes (25-30% strict homology) we have modelled the core of the three dimensional structures of the mammalian alkaline phosphatases. Considerable circ~t~ti~ evidence suggests that this is valid despite the fact that the mammalian enzymes are larger, contain carbohydrate and are membrane associated through a phosphatidylinositol moeity. The active site of the molecule is highly conserved but specific changes in the secondary ligands to bound phosphate and the Mg metal are observed. Introduction Alkaline phosphatase (E.C. 3.1.3.1) is a nonspecific phosphomonoesterase that hydrolyzes phosphate monoesters, ROP, at approximately the same rate regardless of the nature of the R group or the pK, of the leaving group, ROH. Although its specific function in mammals is not known at the present time, the fact that the enzyme is ubiquitous - found in organisms from humans to bacteria - and that it is found attached to the plasma membranes where extensive transport takes place, indicate that alkaline phosphatase is involved in fundamental biological processes. In humans, there are at least four distinct forms of alkaline phosphatase and changes of their levels in blood serum have been used for many years as a diagnostic tool for various diseases (see [l-6] for review).
Correspondence to: Dr. E.E. Kim, Department University, New Haven, CT 06511, USA. ~-8981/89/~3.5~
of Molecular Biophysics and Biochemistry,
@ 1989 Eisevier Science Publishers B.V. (Biomedical Division)
Yale
176
In the past three years several mammalian isozymes have been cloned and sequenced [7-121, but X-ray diffraction structure data are only available for Escherichia coli alkaline phosphatase [13,14]. The E. coli enzyme has been studied most extensively by a variety of physico-chemical techniques and is thus useful as a prototype. This enzyme is synthesized as a precursor with a signal peptide on loosely attached inner membrane ribosomes and is transported into the periplasmic space, where it ultimately appears in the dimeric form (M, = 94 kDa) with the signal peptide removed [15,16] and three metal ions sequestered in each of the two active site regions. The function of the E. coli enzyme is clearly part of the phosphate acquisition and transport system in the cell, induced along with many other proteins, by deprivation of free inorganic phosphate. The previously reported three-dimensional structure was at 2.8 A resolution from this laboratory [13,14]. However, there have been long standing uncertainties about the metal-ligand constellations due to poorly defined electron density in the active site region. Also the top portion (see Fig. 2a) of the molecule (residues 380 to 410) was essentially undefined in the electron density map. Recently we have extended the resolution to 2.0 A with completely new data collected with vastly improved technology. The structure has been revised and refined. Based on the refined structure of the E. cofi enzyme we have examined possible three dimensional structures of mammalian alkaline phosphatases employing the reported sequences. Experimental
section
Data collection Alkaline phosphatase from E. coli was purified and crystallized as described earlier [13,17]. All the diffraction data reported here have been collected using the San Diego Multiwire System area detectors at Yale. Data were collected out to 2.0 A resolution on two crystals soaked in stabilizing solution containing 2 mmol/l inorganic phosphate, 100 mmol/l Tris, 10 mmol/l Zn2’ and 10 mmol/l Mg2+ at pH 7.5. The inorganic phosphate was added primarily because it is a product and a competitive inhibitor of the enzyme, but also to enhance the metal binding. A total of 78,398 unique reflections was obtained from 422,395 actual measurements with an average five-fold redundancy including crystallographic symmetry. This is over 92% of possible reflections for this resolution and the R sym, the mean agreement between the equivalent measurements, was 4.7%. Refinement Refinement was carried out using both the stereochemically restrained least-squares refinement method (PROFFT [18]) and the simulated annealing refinement method (XPLOR [19,20]). The later involves molecular dynamics simulation at elevated temperature using diffraction data in combination with energy functions to overcome local minima. Progress was slow and frequent model-building interpretation (human intervention using computer graphics presentations of electron density maps) using the program FRODO on Evans and Sutherland PS300 graphics systems was necessary to correct misplaced atoms. Details of refinement will be reported elsewhere.
Fig. 1. Comparison of amino acid sequence to yield maximum number of identities. If involved in metal binding areindicated by (A) and cysteines (0)
of alkaline phosphatases. Deletions and insertions are made conserved in all five only E. coli sequence is given. Residues 0. Also putativeglyeosylationsites(A),hltron-exonjunctions in the mammalian sequences are shown.
Sequence comparison The amino acid sequence of human term placental [7-91, human intestinal [lo], human liver/bone/kidney [ll] and rat osteosarcoma [12] alkaline phosphatases deduced from cDNA sequences have been aligned with the sequence of the E. coli enzyme [21,22] allowing appropriate deletions and insertions (see Fig. 1). With only 25-3041 identity between E. coli and the mammalian sequences there exist several ways of alignment. Our version tries to avoid disruption of secondary structures since a single addition or deletion will switch subsequent residues to opposite sides of a P-sheet or rotate residues 100” along an a-helix. Results
Results of both the E. cob AP refinement and modelling of the mammalian enzymes are described below in each section dealing with the overall structure, the main &sheet, the active site of the enzyme and other various points on the mammalian enzymes. Results of refinement The final model has converged to a conventional R-factor (fractional residual discrepancy between observed and calculated structure amplitudes) of 0.184 with good stereochemistry (r.m.s. deviations: 0.015 A in bond
178
(b)
179
distances and 2.9 ’ in bond angles). During this refinement residues 380 to 410 were retraced and residues 190 to 200 were shifted by one residue, in addition to numerous other minor modifications. Also changes were made at the active site region of the protein. A total of 296 water molecules was located as well as a tightly bound phosphate in each active site of the enzyme. Most of the electron density map is now clearly interpretable except for 4 residues at the amino-end of the polypeptide chain and another 4 or 5 residues on the top portion (see Fig. 2a) of the molecule (404-408). Some sidechains, lysines in particular, are not well defined in the map but they are mostly on the surface of the molecule, sporadic and not in crucial positions. The map in the active site region is now clear and the metal-ligand constellations are well defined. The hydroxyl of Ser102, which is phosphorylated and dephosphorylated, is disordered. Overall structure Although the structure has been modified during the refinement, most the earlier description of the structure [16,23] remains the same. The enzyme is a basically 2-fold symmetrical dimer (97 X 47 X 52 A3) with two active sites located about 30 A from each other. Figure 2a shows the C, trace of the complete dimer with metals shown as spheres. Both the amino- and carboxyl-ends are at the bottom of the molecule as indicated in Fig. 2a. Each subunit (449 amino acid residues) shows typical a/P topology with a ten-stranded p-sheet at the center (see Fig. 2b) flanked by 15 helices with various lengths. In contrast to earlier descriptions residues 325 to 334 form a well-behaved a-helix while residues 425 to 430 are not helical. After retracing the chain the top portion contains a three-stranded sheet and a short helix. Therefore the structure consists of the central ten-stranded B-sheet and a minor three-stranded P-sheet on the top as well as sixteen helices. This constitutes about 55% of the protein while the rest of the molecule forms loops, coils, hairpins and other hard-to-describe regions. Further aspects of the structure are discussed below in conjunction with the placental enzyme modelling. Result of sequence comparison The results of the sequence comparisons are similar to what have been ‘observed by others: the eukaryotic isozymes show high homology amongst themselves, i.e. 2 50% overall and 2 80% in subgroups, but when compared to the E. coli enzyme they show 25-30% identities [11,24]. Unless specifically mentioned sequences numbers of the E. coli enzyme are given in the following comparisons. The conserved residues are distributed mainly in the central a/P portion of the molecule while only few are found in the interfacial region of the molecule as seen in Fig. 2b.
Fig. 2. a. C, trace of the complete dirner of E: coli alkaline phosphatase (drawn using SZAZAM developed by A. Perlo). Tube is drawn with 0.5 A radius and the three metals are shown as spheres. Strand G of the central beta-sheet (see Fig. 3) is indicated. The molecule is viewed with its noncrystallographic two-fold axis vertical and the maximum dimension of the molecule horizontal. This is referred to as front view in the text. C,-C, distances are 3.8 A. b. Conserved residues in all five (E. coli, human placental, human intestinal, human liver/kidney/bone and rat osteosarcoma alkaline phosphatases are shown as 0 on C,, trace of monomer, and the metals are shown as stippled spheres.
180
sequence comparison J
B
i
H
A
in P-sheet G
c
HIS
PRO T/S
DIR
D
F
E
HIS T/A
“s6THR
GLY
$$I++-~
W
HIS
GLY
ASP
GLN
Fig. 3. Residues in the central &sheet (shown inside the box) and ones involved in metal binding (shown in bold face) are compared. Strands are named A through 3 in sequence order, while they are shown in their physical order. All are parallel except one (i). If residues are conserved three letter codes are given while one letter codes (in the fotm of E. c&/pIacental) are given othetwise. Additionally boxed residues are to highlight changes discussed in the text.
All alkaline phospbatases are synthesized with a leading signal peptide which is removed after transport of the enzyme either to the periplasmic space in E, culi or on route to the cytoplasmic membrane in the case of enkaryutic enzymes. However, unlike the E. coli enzyme, mammalian alkaline phosphatases are anchored to the cytoplasmic membrane. Recently the placental enzyme has been shown to be anchored to the membrane at position Asp484 by a mechanism involving a phosphatidylinositol-glycan anchoring moiety with the carboxyl terminal sequence removed in post translational processing [25,26]. Others are also believed to be anchored in the similar manner. As shown in Fig. 2a both the carboxyl- and the amino-ends of the molecule are pointing outwards, well away from the active site as well as the core of the molecule. Thus extensions of the chain at either end, such as in the case of mammals at the carboxyl end or fusion proteins [27] at the amino end, should not affect the structure or function grossly. a/@ portion of the naolecule A comparison of the residues in the 10 stranded &sheet is shown in Fig. 3. If they are identical, three letter codes are given; otherwise one letter codes are given in the order of E. coli/placental enzymes. Residues inthe &sheet are shown inside the box and the ones involved in the metal binding are shown in bold face. Strands are named A through J in sequence order,
181
Fig. 4. Active site of the E. coli alkalinephosphatase. Some residues and water molecules are omitted for clarity. In the mammalian sequences all the residues involved in the metal binding are conserved except for Thr155, which is replaced by serine in some’cases. Others in this region are also conserved as well except for a few, e.g. 153 (Asp to His) and 328 (Lys to His) which are shown as single lines in modelled positions.
while they are shown in their physical order. The strands are parallel except for one antiparallel insertion (strand i). The sheet is largely classical in that it is connected in a right-handed sense as one proceeds from one strand to the next while the twist of the sheet is left-handed. The structure is boat shaped, narrow at the bottom and broader at the top both in and perpendicular to the plane of Figs. 2a and 3. The monomer is considerably thicker on one side of the &sheet as seen in the top views of Fig. 5. The active site is located at the carboxyl end of the &sheet, and a detailed view is shown in Fig. 4. As seen in Fig. 3, the strands in the core of the molecule, i.e. strands A and G, have larger number of identities than the shorter strands on the surface of the molecule, i.e. strands B and E. The p-sheet can be considered as having horizontal rows of residues facing one side of the sheet alternating with rows of residues facing the opposite side (as indicated by shading in Fig. 3). When one considers these two surfaces separately, there are only 16 (52%) and 14 (40%) identities on each surface.
182
(b)
Fig. 5. C, trace of monomer. Panels on the left are drawn with the non-crystallographic two-fold axis vertical (referred to as front view) while the ones on the right are drawn looking down the two-fold axis (referred to as top view). a. Insertions/deletions sites in the marmnak sequences. Insertion sites are shown as 0 while the segments deleted are shown with heavier lines b. Intron-exon junctions of the mammalian enzymes are shown as 0 on C, trace of one subunit with central &strands indicated by heavier lines. c. Disulfides in E. coli are indicated by heavy lines and labekd a and b. Putative glycosylation sites at Asn positions are shown by 0. Ala161 is shown as a stippled sphere.
183
However, one can consider many of the replacements as structurally conservative, and some of these are boxed. Replacements of Gln to Arg at 83 and Arg to Ala at 418 can be considered paired conservative changes since they are adjacent on the same side of the p-sheet. Again replacements of Ile to Leu at 365 and Leu to Ile at 47 are another pairwise conservative couple. If one includes these and other conservative replacements, i.e. Leu to Val or Phe and Asp to Glu, then the homology fractions on the two surfaces increase to 77% and 71%, respectively. The cluster of replacements in the region at 320, 145, 143, and 203 is particularly interesting since the somewhat anomalous hydrophilic patch on the E. coli enzyme is replaced with more classical hydrophobic surface. Active site The active site region can be considered as AsplOl-Ser102-Ala103 and the metal triplet (two Zn and one Mg) and their ligands as well as Arg166 and others in the vicinity. This is shown in Figure 4 where some water molecules and residues are omitted for clarity. Phosphate and water molecules found in this region during the refinement are involved in an extensive hydrogen bonding scheme. Znl is penta-coordinated by the imidazole nitrogen atoms of two histidines, His331 and His412, both carboxyl oxygens of Asp227 and one of the phosphate oxygens with average metal-ligand distance of 2;P7 A. His372 which was originally thought to be a direct ligand to Znl is not (3.8 A away from Znl), but is hydrogen bonded to one of the carboxyl oxygens of Asp327. Zn2 is coordinated tetrahedrally by the imidazole nitrogen of His370, one of the carboxyl oxygens of AspSl(OD1) and of Asp369(ODl) and one of the phosphate oxygens. The average metal-ligand distance is 2.00 A. In the absence of phosphate the hydroxyl (OG) of the SerlO2 is a ligand to this metal. Mg-coordination can be described as a slightly distorted octahedron with the second carboxyl oxygen of Asp51(OD2) and one of the carboxyl oxygens of Glu322(OEl) and hydroxyl of Thr{55(OG) and three water molecules as ligands. The average Mg-0 distance is 2.12 A. Asp153 is not a direct ligand to this metal as previously indicated, but is an indrect ligand hydrogen-bonded to two of the three water molecules that are coordinating Mg. Phosphate is coordinated to both Znl and Zn2, and the other two oxygens are tightly held by two amino functions of the guanidinium group of ArgZ66, which is in turn further H-bonded to Asp101 and a water that is held by Asp153 and Tyr169. The phosphate is further H-bonded to the amide of Ser102 and a water molecule that is coordinated to Mg and another water that is bridging to Lys328. The van der Waals surface on this region shows that the active site pocket is just big enough to hold the phosphate end of the substrate thus leaving the remainder of the molecule in the solvent region. The residues 101 to 112 form a helical stem that is buried between the B-sheet and other helices and loops and this provides a firm anchor for the active Ser102. The hydroxyl group of SerlO2 is disordered. The carboxyl group of Asp101 is positioned such that there is no direct contact with either serine or the phosphate but it appears that it is functional in positioning the guanidinium group of Arg166 and providing part of the electrostatic environment.
184
In the case of the mammalian enzymes the Asp-Ser-Ala/Gly sequence at the phosphoryl acceptor is conserved. This sequence also exists in serine proteases in which the serine is an acyl acceptor. All the residues serving as direct ligands to the metals (bold faced in Fig. 3) are conserved except for fully acceptable replacement at position 155 where Thr is replaced by Ser in some cases. Also Arg166, which complexes to phosphate in the active site, is conserved. Thus the structure of the active site region of the enzyme by implication appears highly conserved even in the case of Mg-coordination whose role in catalysis is not clear. This suggests a real importance of this site as well. However, it is interesting to note that both Lys 328 and Asp 153 which are above the Mg-coordination sphere are replaced by histidines in all mammals as shown in Fig. 4. The mammalian enzymes are 20- to 30-fold more active than the E. coli enzyme when they are at their pH optima and under conditions of substrate saturation. Insertion and deletion sites In the mammalian enzymes there are several places where insertions and deletions take place, and these are shown in Fig. 5a. Many of these occur on the surface loop regions of the molecule away from the active site. There are three major insertions at the positions 125, 404 and 411 involving 8, 16 and 13 residues respectively. The insertion sites 87, 404 and 411 are near the active site. Residue 411 is next to His412 which is a direct ligand to Znl. These insertions at 404 and 411 considerably expand the upper region of the molecule and the variability in this region among isozymes might indicate different roles. All the deletion sites are located at least 20 A away from the metal triplet. The deletion of 10 residues between 32 and 41 (See Fig. 5a) occurs at the bottom of the interface region that includes part of helix 29-37, which is followed by the important central strand (strand A) of the P-sheet. However, this is not as serious as one might expect since the region between 16 and 23 in the placental sequence (and others) is strongly predicted to be a helix, based on the algorithm of Gamier et al. [28]. This might replace structurally the 29-37 helix of the E. coli leaving a slightly longer connecting loop (either 6 or 7 residues instead of 5 residues) before the first /3 strand. The deficit then would be at the amino terminal which could be readily accommodated. Intron-exon structure In humans there are at least three genes: placental, intestinal and liver/kidney/bone. In all three genes the intron-exon junctions occur at analogous positions, although the gene for liver/kidney/bone is larger than the other two [29]. As seen in Fig. 5b all the intron-exon junctions occur at either on or near the surface residues and at or near the residues where changes in secondary structure take place. Similar observations have been made in other proteins [30,31]. The ones at residues 52 and 323 are near the active site of the enzyme. Glycosylation sites Unlike the E. coli enzyme the mammalian enzymes are glycosylated and there are five putative sites (Asn-xxx-Tbr/Ser) reported in the liver/bone/kidney enzyme [11,12] while only two are retained in the placental and intestinal sequences. All except one of the putative glycosylation sites occur on the
185
surface of the molecule (see Fig. 5c) and that one is in a poorly conserved region which could easily be distorted. Disulfides All four cysteines in the E. coli enzyme are involved in forming two disulfide bridges, both away from the active site of the enzyme as seen in Fig. 5c. All four are replaced in the mammalian enzymes. There are five cysteines in mammalian enzymes indicating the presence of at least one free cysteine. The C, positions of 111 and 129 in the E. coli enzyme, equivalent to 101 and 121 in the placental sequence, are at a distance of possible disulfide bridge formation. Hypophosphatasia mutants A number of hypophosphatasia patients suffering from severe to mild skeletal and related disorders are reported to have mutations in the bone alkaline phosphatase gene. One has been clearly shown to be a single point mutant at the residue corresponding to Ala161 in the E. coli enzyme, changing it to threonine [32]. Residues 161 and 162 are conserved (see Fig. 1). The structure of this region is shown in Fig. 6 and is located at the end of a short helix away from the active site as shown in Fig. 5c, yet the mutation produces a molecule with little activity. The C, of Ala161 is in a hydrophobic core and His162 is mostly inaccessible with well restrained hydrogen bonds. Ala161 is between residues 155 and 166 which are both involved in the active site. The local distortion needed to accommod-
Fig. 6. Environment of Ala161 of E. coli enzyme. His162 is in a depression on the surface. with little solvent accessibility. Ala161 is inaccessible and C, is pointing back into a hydrophobic cluster.
186
ate the mutant might propagate to Arg166, since site directed mutations of Arg166 are known to have substantial effects on activity in the E. coli enzyme [33,34]. An alternative explanation would be that proper folding is prevented resulting in a more extensive disruption. Conclusion Based on above observations one can conclude that although there is only 25-3048, strict sequence homology between mammalian and E. coli alkaline phosphatases it appears that the mammalian enzymes retain the core of the three-dimensional structure of the E. coli enzyme. The active site region is highly conserved with some specific changes in the secondary ligands. Many changes in the ten-stranded b-sheet are structurally conservative. Identities are clustered in the core of the structure while surface residues and segments further from the active site are less homologous. None of the insertions, deletions, carbohydrate attachment sites or exon boundaries pose serious problems. Of course confirmation of these observations and the ramifications of the differences have to wait until a detailed crystal structure analysis of the mammalian enzymes is available. Acknowledgements
We thank Drs. J.E. Coleman for providing the protein, A.T. Brtinger for extensive use of his program XPLOR, S.K. Kattti for helpful discussions during refinement, A. Perlo for use of his program SZA-ZAM and J. Ponder for use of his version of program predicting the secondary structure, and J. Mouning for help in preparing the manuscript. This work was supported by Public Health Service grant GM22778 from National Institute of General Medical Science. References 1 2 3 4 5 6 7
8 9
10
McComb RB, Bowers ON, Posen S. Alkaline phosphatases. New York: Plenum, 1979. Reid TW, Wilson IW. E. co/i alkaline phosphatase. Enzymes 1971;4:373-416. Femley HN. Mammalian alkaline phosphatases. Enzymes 1971;4:417-447. Coleman JE, Gettins P. Alkaline phosphatase, solution structure and mechanism. Adv Enzymol Relat Areas Mol Biol 1983;55:381-452. Harris H. Multilocus enzyme systems and the evolution of gene expression: the alkaline phosphatases as a model example. Harvey Lecture Ser 1982;76:95-124. Moss DW. Genetics, expression, and modification in the human alkaline phosphatases. Isozymes 1987;16:67-80. Kam W, Clauser E, Kim YS, Kan YW, Rutter WJ. Cloning, sequencing, and chromosomal localization of human term placental alkaline phosphatase cDNA. Proc Nat1 Acad Sci USA 1985;82:8715-8719. Mill&n JL. Molecular cloning and sequence analysis of human placental alkaline phosphatase. J Biol Chem 1986;261:3112-3115. Henthom PS, Knoll BJ, Raducha M, et al. Products of two common alleles at the locus for human placental alkaline phosphatase differ by seven ammo acids. Proc Nat1 Acad Sci USA 1986;83:5597-5601. Berger J, Garattini E, Hua J-C, Udenfriend S. Cloning and sequencing of human intestinal alkaline phosphatase cDNA. Proc Nat1 Acad Sci USA 1987;84:695-698.
187 11 Weiss MJ, Henthom
12
13 14
15 16 17 18 19 20 21 22
PS, Lafferty MA, Slaughter C, Raducha M, Harris H. Isolation and characterization of a cDNA encoding a human liver/bone/kidney-type alkaline phosphatase. Proc Nat1 Acad Sci USA 1986;83:7182-7186. Thiede MA, Yoon K, Golub EE, Noda M, Rodan GA. Structure and expression of rat osteosarcoma (ROS 17/2.8) alkaline phosphatase: product of a single copy gene. Proc Natl Acad Sci USA 1988;85:319-323. Sowadski JM, Handschumacher MD, Murthy HMK, Foster BA, Wyckoff HW. Refined structure of alkaline phosphatase from Escherichiu coli at 2.8 A resolution. J Mol Biol 1985;186:417-433. Wyckoff HW. Structure of Escherichia coli alkaline phosphatase determined by X-ray diffraction in Phosphate metabolism and cellular regulation in microorganisms American Society for Microbiology 1987;118-126. Chang CN, Inouye H, Model P, Beckwith J. Processing of alkaline phosphatase precursor to the mature enzyme by and Escherichia coli inner membrane preparation. J Bacterial 1980;142:726-728. Inouye H, Barnes W, Beckwith J. Signal sequences of alkaline phosphatase of Escherichia coli. J Bacterial 1982;149:434-439. Appleburry ML, Johnson BP, Coleman JE. Phosphate binding to alkaline phosphatase. J Biol Chem 1970;245:4968. Finzel BC. Incorporation of fast Fourier transforms to speed restrained least-squares refinement of protein structures. J Appl Cryst 1987;20:53-55. Brtinger AT, Kuriyan J, Kaplus M. Crystallographic R factor refinement by molecular dynamics. Science 1987;235:458-460. Briinger AT, Karplus M, Petsko GA. Crystallographic refinement by simulated annealing: Application to crambin. Acta Cryst 1989;A45:50-61. Bradshaw RA, Cancedda F, Ericsson LH, et al. Amino acid sequence. of Escherichia coli alkaline phosphatase. Proc Nat1 Acad Sci USA 1981;78:3473-3477. Chang CN, Kuang W-J, Chen EY. Nucleotide sequence of the alkaline phosphatase gene of
Escherichiu coli K-12. Gene 1986;44:121-125. 23 Wyckoff HW, Handschumacher MD, Murthy HMK, Sowadski JM. The three dimensional structure of alkaline phosphatase from E. co& Adv Eruymol Relat Areas Mol Biol 1983;55:453-480.
24 Mill&n JL. Gncodevelopmental expression and structure of alkaline phosphatase genes. Anticancer Res 1988;8:995-1004. 25 Low MG, Saltiel AR. Structural and functional roles of glycosylphosphatidylinositol in membranes. Science 1988;239:268-275. 26 Micanovic R, Bailey CA, Brink L, Gerber L, Pan Y-CE, Hulmes JD, Udenfriend S. Aspartic acid-484 of nascent placental alkaline phosphatase condenses with a phosphatidylinositol glycan to become the carboxyl terminus of the mature enzyme. Proc Nat1 Acad Sci USA 1988;85:1398-1402. 27 Hoffman CS, Wright A. Fusions of secreted proteins to alkaline phosphatase: an approach for studying protein secretion. Proc Nat1 Acad Sci USA 1985;82:5107-5111. 28 Gamier J, Osguthorpe DJ, Robson B. Analysis of the accuracy and implications of simple methods for predicting secondary structure of globular proteins. J Mol Biol 1978;120:97-120. 29 Weiss MJ, Ray K, Henthom PS, Lamb B, Kadesch T, Harris H. Structure of the human liver/bone/kidney alkaline phosphatase gene. J Biol Chem 1988;263:12002-12010. 30 Craik CS, Sprang S, Fletterick R, Rutter WJ. Intron-exon splice junctions map at protein surfaces. Nature 1982;299:180-182. 31 Bajaj M, Blundell T. Evolution and tertiary structure of proteins. Arm Rev Biophys Bioeng 1984;13:453-492. 32 Weiss MJ, Cole DEC, Ray K, Whyte MP, Lafferty MA, Mulivor RA, Harris H. A m&sense mutation in the human liver/bone kidney alkaline phosphatase gene causing a lethal form of hypophosphatasia. Proc Nat1 Acad sci USA 1988;85:7666-7669. 33 Chaidaroglou A, Brezinski DJ, Middleton SA, Kantrowitz ER. Function of arginine-166 in the active site of Escherichia coli alkaline phosphatase. Biochemistry 1988;27:8338-8343. 34 Butler-Ransohoff JE, Kendall DA, Kaiser ET. Use of site-directed mutagenesis to elucidate the role of arginine-166 in the catalytic mechanism of alkaline phosphatase. Proc Natl Acad Sci USA 1988;85:4276-4278.