structural analysis of a multifunctional, tandemly repeated...

Structural Analysis of a Multifunctional, TandemlyRepeated Inositol Polyphosphatase

Robert J. Gruninger1, L. Brent Selinger2 and Steven C. Mosimann1⁎1Department of Chemistry andBiochemistry, University ofLethbridge, 4401 UniversityDrive, Lethbridge, Alberta,Canada T1K 3M42Department of BiologicalSciences, University ofLethbridge, 4401 UniversityDrive, Lethbridge, Alberta,Canada T1K 3M4

Received 29 April 2009;received in revised form26 May 2009;accepted 28 May 2009Available online3 June 2009

Mitsuokella multacida expresses a unique inositol polyphosphatase(PhyAmm) that is composed of tandem repeats (TRs). Each repeat possessesa protein tyrosine phosphatase (PTP) active-site signature sequence and fold.Using a combination of structural, mutational, and kinetic studies, we showthat the N-terminal (D1) and C-terminal (D2) active sites of the TR havediverged and possess significantly different specificities for inositol polypho-sphate. Structural analysis and molecular docking calculations identify stericand electrostatic differences within the substrate binding pocket of each TRthat may be involved in the altered substrate specificity. The implications ofour results for the biological function of related PTP-like phytases arediscussed. Finally, the structures and activities of PhyAmm and tandemlyrepeated receptor PTPs are compared and discussed. To our knowledge, thisis the first example of an inositol phosphatase with tandem PTP domainspossessing substrate specificity for different inositol phosphates.

© 2009 Published by Elsevier Ltd.

Edited by M. GussKeywords: inositol phosphatase; tandem repeat; protein tyrosine phospha-tase; phytase; myo-inositol

Introduction

Inositol polyphosphates (IPPs) are ubiquitous innature and their central role in eukaryotic cellularsignaling is well known.1 Myo-inositol 1,2,3,4,5,6-hexakisphosphate (InsP6) is the most abundantcellular inositol phosphate and has been implicatedin numerous important cellular processes includingDNA repair, RNA processing and export, develop-ment, apoptosis, and pathogenicity.2–7 The importantbiological role that InsP6 plays is exemplified by theobservation that deletion of enzymes involved inInsP6 biosynthesis has a lethal phenotype in mouseembryos.8,9 Inositol polyphosphatases (IPPases) thatdegrade InsP6 are ubiquitous in nature and have been

identified in prokaryotes, protists, fungi, animals, andplants and are generically referred to as phytases.10,11Four distinct classes of phytate-degrading IPPaseshave been identified to date and include the histidineacid phosphatases, β-propeller phytase, purple acidphosphatase, andprotein tyrosinephosphatase (PTP)-like phytase (PTPLP), also known as cysteinephytases.12–14 Phytases catalyze the stepwise removalofphosphates togenerate lower inositolphosphates ina highly ordered, specific manner.15 Although phy-tases are functionally similar, theyare structurally andmechanistically diverse.The recently described PTPLP class of IPPases

possess a PTP active-site signature sequence [CX5R(S/T)], are structurally similar to PTPs, and utilize aclassic PTP reaction mechanism.13,14 Interestingly,these enzymes display no catalytic activity againstclassic PTP substrates due to several unique struc-tural features that confer specificity for IPPs.13,16–18Although the biological function of these enzymes isunclear, they have been found in a wide range ofbacteria including plant and human pathogens.19,20A PTPLP identified in the rumen and gastrointestinalinhabitant Mitsuokella multacida is unique from allother PTPLPs because it is composed of tandemlyrepeated PTPLP domains. This tandem arrangementis commonly observed in receptor PTPs (RPTPs)21but not in IPPases. Interestingly, only the N-terminal

⁎Corresponding author. E-mail address:[email protected] used: IPP, inositol polyphosphate;

IPPase, inositol polyphosphatase; InsP6, myo-inositolhexakisphosphate; PhyAmm,Mitsuokella multacida IPPase;PhyAsr, Selenomonas ruminantium IPPase; LAR, leukocyteantigen related; PTEN, phosphatase and tensinhomologue; PTP, protein tyrosine phosphatase; PTPLP,protein tyrosine phosphatase-like phytase; RPTPs,receptor protein tyrosine phosphatases; ASU, asymmetricunit; TR, tandem repeat; TRI, tandem repeat interface.

doi:10.1016/j.jmb.2009.05.079 J. Mol. Biol. (2009) 392, 75–86

Available online at www.sciencedirect.com

0022-2836/$ - see front matter © 2009 Published by Elsevier Ltd.

(D1) repeat in many RPTPs is catalytically active.The function of the inactive C-terminal (D2) repeat inRPTPs is not fully understood but has been shown tobe involved in regulation and substrate recognitionin some RPTPs.22–25In this work, we describe the structure of the

tandemly repeated InsP6-degrading IPPase from M.multacida (PhyAmm). Using a combination ofmacromolecular docking, site-directed mutagenesis,and kinetic assays, we demonstrate that each repeathas a unique substrate specificity. The resultspresented here suggest that PhyAmm may utilizea novel mechanism of dephosphorylating IPPs.Furthermore, these results may have implicationsfor our understanding of the function of the D2repeat in RPTPs.

Results

Structure of PhyAmm

The structure of PhyAmm has been determined toa resolution of 2.3 Å by the multiwavelengthanomalous dispersion method. Statistics for the

data collection and refinement of PhyAmm areshown in Table 1. The electron density used to buildthe model was of excellent quality (Fig. S1). The finalmodel was refined to an R-factor and Rfree of 20.8%and 25.4%, respectively. The model displays goodstereochemistry as assessed by PROCHECK,26 with85.8% and 13.5% of the residues in the mostfavorable and additionally allowed regions of theRamachandran plot, respectively. Continuous elec-tron density is observed for amino acids 46–636,with the remaining residues (11–45 and 636–640)located at the termini (the expression tag is encodedby residues 11–34).PhyAmm crystallizes in space group P21 with a

homodimer in the asymmetric unit (ASU) and anapproximate 2-fold noncrystallographic symmetryaxis between the chains making up the dimer(Fig. 1a). The PISA (Protein Interfaces, Surfaces, andAssemblies) server calculates the surface area buriedfrom formation of the PhyAmm dimer to be 4040 Å2

and indicates that the dimer in the ASU is stable.27The dimer interface consists of an extensive networkof hydrogen bonds and van der Waals contacts. Mostof these contacts are made between structuralfeatures that are unique to PTPLPs, including an

Table 1. Data collection and refinement statistics

Native WO42− derivative

Data collectionSpace group P21 P21Unit cella, b, c (Å) 74.39, 73.30, 161.31 74.25, 74.00, 160.83β (°) 93.7 93.9

Inflection Peak High energyWavelength (Å) 1.1159 1.2150 1.2146 1.1579Resolution (Å) 80–2.3 80–2.8 80–2.8 80–2.8Observed reflections 241,259 210,895 170,071 169,776Unique reflections 74,432 53,991 43,324 43,324Completeness (%) 96.5 (91.4) 100 (100) 100 (100) 100 (100)Redundancy 3.9 (4.0) 3.9 (3.8) 3.9 (4.0) 3.9 (4.0)Rmerge 0.088 (0.293) 0.050 (0.203) 0.046 (0.157) 0.049 (0.194)I/σI 10.0 (2.1) 18.2 (5.4) 19.8 (7.3) 18.6 (6.1)

Refinement statisticsResolution (Å) 20–2.3No. of reflectionsWork set 72,395Test set 2256Rwork (%) 20.8Rfree (%) 25.4Protein atoms 9613Solvent atoms 745PO4

2− 4Wilson B (Å2) 42.7Average B protein (Å2) 39.2Average B solvent (Å2) 44.2Average B PO4

2− (Å2) 58.3RMSDBonds (Å) 0.011Angles (°) 1.46

Ramachandran distributionMost favored (%) 85.8Additionally allowed (%) 13.5Generously allowed (%) 0.4Disallowed (%) 0.3

Values in parentheses are for the highest-resolution shell (2.42–2.30 Å for native and 2.95–2.80 Å for the derivative).

76 Structure of a Tandemly Repeated Inositol Polyphosphatase

extended Ω loop, a small partial β-barrel domain(IPP domain), and an extended C-terminal helix.These structures have been proposed to be involvedin conferring substrate specificity in the relatedenzyme from Selenomonas ruminantium, PhyAsr.13,14Size-exclusion chromatography indicates thatPhyAmm also behaves as a dimer in vitro, suggestingthat the crystallographic dimer is also formed insolution (data not shown).Each PhyAmmmonomer that makes up the dimer

in the ASU is composed of tandemly repeated PTPdomains, D1 (residues 35–338) and D2 (residues347–640; according to RPTP convention),21 con-nected by a flexible nine-residue linker. Each PTPcore domain has a central five-stranded β-sheet(β2/2′, β3/3′, β10/10′, β4/4′, and β9/9′) with twoα-helices on one side (α1/1′ and α3/3′) and fiveα-helices (α4/4′, α5/5′, α6/6′, α7/7′, and α8/8′)on the other (Fig. 1b). The highly conserved strand-loop-helix active-site motif found in all PTP super-family members is formed by β10/10′ and α5/5′.The IPP domain that is unique to PTPLPs iscomposed of a highly twisted, five-stranded partialβ-barrel (β1/1′, β5/5′, β6/6′, β7/7′, and β8/8′) anda single α-helix (α2/2′).A search for structural homologues was carried

out using the DALI server28 and identified severalmembers of the PTP superfamily, including therelated PTPLP, PhyAsr (2b4u, Z score=35.1)13;phosphatase and tensin homologue (PTEN) (1d5r,Z score=12.1)29; and leukocyte antigen related(LAR) PTP (1lar, Z score=9.5),30 but no membersof the three other phytase classes. A similar DALI

search using only the IPP domain (residues 46–63and 132–202 in D1 and residues 344–356 and 431–500 in D2) identified a single homologous structurecorresponding to the equivalent domain in theclosely related PTPLP, PhyAsr (2b4u, Z=5.3),suggesting that this domain is unique to this proteinfamily.

The D1 and D2 tandem repeats adopt thesame fold

The sequence identity of D1 and D2 is 36%internally and 34% and 48% with PhyAsr, respec-tively. Least-squares superposition of the main-chain atoms of D1 and D2 with PhyAsr results in aroot-mean-square deviation (RMSD) of 1.46 Å over258 residues and 1.36 Å over 272 residues, respec-tively. A similar superposition between the repeatsresults in an RMSD of 1.44 Å over 257 matchingresidues (Fig. 2a). The regions that show the greatestdeviation are unique to the PTPLP family and areinvolved in conferring specificity for IPPs in theseenzymes. These regions include β7/7′ and β8/8′ ofthe IPP domain, the Ω loop connecting β2/2′ andβ3/3′ (Ωβ2–β3), and the N-terminal end of α7/7′.The structural variability in the β7/7′ and β8/8′ β-hairpin is due to a two-residue insertion in D1 andan antiparallel G1 β-bulge involving D183, N187,and V188 that results in D184 and N187 pointingdeep into the binding pocket, while K185 and K186point away from the binding pocket. The equivalentregion of D2 also forms a β-hairpin, but the β-strands are much shorter and K483-K485 point

Fig. 1. Structure of the tandemly repeated IPPase PhyAmm. (a) Quaternary structure of theM. multacida PTPLP dimerobserved in the ASU. The D1 repeats are colored orange and yellow and the D2 repeats are colored light blue and blue forchains A and B, respectively. (b) Tertiary structure of PhyAmm. The TRs of chain A are shown rotated by ∼45° relative totheir orientation in (a). The core PTP domains of D1 and D2 are shown in light blue and orange, respectively. The IPPdomain is colored green; the linker is shown in grey; and phosphate molecules found in the active site of both repeats areshown as yellow sticks.

77Structure of a Tandemly Repeated Inositol Polyphosphatase

toward the binding pocket to make contacts with thenegatively charged phosphates on InsP6. In D1,Ωβ2–β3 is shorter than in D2 due to a seven-residuedeletion, and it adopts an extended conformationdue to intermolecular contacts within the PhyAmmhomodimer in the ASU. In D2, Ωβ2–β3 adopts thesame conformation as seen in PhyAsr and contri-butes to the substrate binding pocket with K379pointing into the active site. The variation in the N-terminal end of α7 in D1 is due to the involvement ofD295, K298, N299, and Y300 in the dimer interface ofthe PhyAmm homodimer and a three-residueinsertion in the loop connecting α6 and α7.

Comparison of D1 and D2 active sites

The size of the active site is an importantdeterminant of substrate specificity in the PTPsuperfamily.33 Similar to PhyAsr, both the D1 andD2 repeats of PhyAmm have a very large, electro-positive active site to enable the binding of thelarge, highly negatively charged substrate, InsP6.Interestingly, comparison of the surface potentialscalculated using GRASP32 shows that the D1 repeatis less electropositive than the D2 repeat (Fig. 2b).PDBsum34 was used to compare the sizes of theactive sites of the two repeats and the related

Fig. 2. Comparison of the TRs of PhyAmm. (a) Structural alignment of the D1 and D2 repeats colored according toRMSD (dark blue: b0.5 Å → red: N4.0 Å). The alignment utilized all Cα atoms and was carried out with SPDBV.31 Imagewas created using SPDBV. (b) Electrostatic surface potential (±10 kT·e−1) of the PhyAmm TR calculated using GRASP2.32Electropositive regions are colored in blue and electronegative regions are colored in red. (c) Superposition of the activesite of the D1 (yellow) and D2 (blue) repeats of PhyAmm. Residues showing conformational variability or that aremutated between repeats are shown as sticks. Image is shown in divergent stereo mode.


PTPLP PhyAsr. The active site in the D1 repeat(volume of 4862 Å3 and depth of 13 Å) is smallerthan the active sites of D2 (volume of 7333 Å3 anddepth of 17 Å) and PhyAsr (volume of 5370 Å3 anddepth of 14 Å). The front and sides of the bindingpocket are formed by Ωβ2–β3, α7/7′, and the IPPdomain, respectively. These structures are alsoobserved in PhyAsr and make several specificcontacts to the substrate.14The base of the active site is formed by the general

acid (GA) loop (equivalent to the WPD loop inPTP1B) and the highly conserved phosphate bind-ing loop (P-loop). In both repeats, the GA loop is in a

closed conformation placing the general acid (D221and D519) in a position equivalent to the generalacid in PhyAsr and PTEN. The P-loop contains thecatalytic cysteine (C250 and C548) and the invariantarginine that coordinates the scissile phosphate fornucleophilic attack (R256 and R554). While the D1 P-loop (HC250YAGMGR256T) differs in sequence fromthe D2 P-loop (HC548QAGAGR554T), least-squaressuperposition of the Cα atoms results in an RMSD of0.22 Å, indicating that the P-loop conformations areessentially identical (Fig. 2c). There are smalldifferences in some of the P-loop side-chain con-formations and the GA loop. There is a 0.45-Å shift

Fig. 3. Comparison of the TRs in PhyAmm and RPTPs. (a) Structure of the RPTP LAR TR.30 (b) Structure of thePhyAmm TR. Close-up of the boxed regions in (a) and (b) showing the TRI of (c) LAR and (d) PhyAmm. In both (a) and(b), the catalytic cysteines are shown as spheres to illustrate the relative rotation of the D1 and D2 repeats. For clarity, theclose-up of the LAR dimer interface is rotated ∼45° about the long axis of the repeat relative to the orientation shown in(c). Residues making productive interactions are shown as sticks. Stabilizing interactions are shown as dashed lines. In allimages, the D1 repeat is colored yellow, the D2 repeat is colored blue, and the repeat linker is shown in red.


in H222 Cα and a rotation aboutχ2 of 13° that resultsin a 0.81-Å movement of Nδ1 away from thecatalytic site in D1. In addition, Asp221 χ2 is rotatedby 26°, resulting in a 0.64-Å movement of Oδ1 intothe D1 active site. Despite these small differences,overall the structures of the active sites are essen-tially identical.

The PhyAmm tandem repeat is unique from theRPTP tandem repeats

We compared the orientation of the tandem repeat(TR) in PhyAmm to the structure of all tandemlyrepeated RPTPs in the Protein Data Bank (PDB).Only RPTPs whose structure had been determinedwith both the D1 and D2 repeats were used in thecomparison; these included CD45 (1ygu), LAR(1lar), RPTPσ (2fh7), RPTPγ (2nlk), and RPTPɛ(2jjd). The orientation of the dimer is the same in allof the RPTPs with the active sites oriented ∼90°relative to one another about the long axis of the TR.A six-residue linker is located at the interfacebetween the repeats and oriented almost perpendi-cular to the long axis of the TR (Fig. 3a). The TRinterface (TRI) covers a surface area of ∼1300 to1500 Å2 and is made up of an extensive network ofsalt bridges, hydrogen bonds, and van der Waalscontacts. The conserved threonine and glutamatewithin the domain linker [consensus sequence G(D/E)TE(V/I/L)] are involved in specific, conservedhydrogen bonds and salt bridges to the D2 repeat30(Fig. 3c). In PhyAmm, the linker is orientatedparallel with the long axis of the TR, with the D1and D2 active sites located on the same side (Fig. 3b).The TRI of PhyAmm covers a surface area of∼1100 Å2 and is composed of van der Waalscontacts, salt bridges, and hydrogen bonds betweenα1 and α7′ (Fig. 3d). Unlike RPTPs, no residues inthe linker (N339SWEPDYN) make specific contactswith either repeat at the TRI.Structural superposition of the D1 and D2 repeats

of PhyAmm onto the RPTP domains reveals thatPhyAmm cannot adopt the RPTP tandem dimerdue to significant steric clashes between Ωβ2–β3and α7. The related PTPLP, PhyAsr, also adopts adimer both in solution (data not shown) andcrystallographically13; however, unlike PhyAmm,the PTPLP domains in PhyAsr are not covalentlylinked. The PhyAsr dimer is unique from the RPTPand PhyAmm TRs due to steric clashes betweenΩβ2–β3 and the extended C-terminal helix (equiva-lent to α7) in the RPTP TR and steric clashesbetween the N terminus and the IPP domain in thePhyAmm TR. Additionally, the N and C termini ofthe repeats of PhyAmm are 46 Å apart in the PhyAsrdimer, which would require a large insertion toenable a covalent linkage between the repeats.

Catalytic activity of PhyAmm D1 and D2 repeats

In many RPTPs, only the N-terminal (D1) repeat iscatalytically active.21 To test whether both PhyAmmrepeats possess catalytic activity, we used site-

directed mutagenesis to change the catalytic cysteineto a serine in the D1 and D2 repeats (C250S andC548S, respectively) and then tested the mutants forcatalytic activity against InsP6. The kcat and Km valuesfor the hydrolysis of InsP6 by wild-type PhyAmm are1109±64 s−1 and 347±48 μM, respectively. Mutationof the catalytic cysteine in D1 (C250S) had little effecton the rate of InsP6 hydrolysis (kcat, 1048±48 s−1) butresulted in a slight increase in Km (459±35 μM).Mutation of the cysteine in D2 (C548S) completelyabolished enzyme activity. Even after increasedincubation times with the C548S mutant, there wasno detectable catalytic activity above the limit ofdetection for our assay system (0.003 μmol ofphosphate). These results indicate that the D2 repeatis solely responsible for hydrolyzing InsP6 andsuggest that the D1 repeat is catalytically inactive. Itis noteworthy that in contrast to PhyAmm, the D2repeat is catalytically inactive in RPTPs, and that thelack of activity is due to mutation of catalyticresidues. All of the catalytic residues are conservedin both PhyAmm repeats, indicating that the lack ofactivity against InsP6 in the D1 repeat is due to aunique feature of this enzyme.

The PhyAmm D1 repeat is active against lowerinositol phosphates

Although D1 does not hydrolyze InsP6, there is noobvious reason that it cannot carry out catalysis andit may be possible that the specificities of the tworepeats differ. To examine whether the D1 repeat ofPhyAmm has catalytic activity against lower IPPs(i.e., less than six phosphates), we tested the catalyticactivity of PhyAmm against several of the numerousInsP5, InsP4, and InsP3 substrates (Table 2). Removalof the phosphate at the 2-position of InsP6 is su-fficient to observe catalytic activity in D1. Thisconfirms the structure-based prediction that essentialactive-site residues are in a catalytically competentconformation. This repeat displayed a specificactivity of 8 U/mg for Ins(1,3,4,5,6)P5, approxi-mately five times greater than the limit of detectionof the assay. In contrast, the specific activity for Ins(1,3,4,5,6)P5 compared to InsP6 is approximately 1.4-fold lower in the D2 repeat. Removal of anadditional phosphate resulted in a further increase

Table 2. Catalytic activity of wild-type and C250S (D1repeat) and C548S (D2 repeat) mutants of PhyAmmagainst lower-order inositol phosphates

SubstrateC250S(U/mg)

C548S(U/mg)

Wild type(U/mg)

Ins(1,4,5)P3 183±11 53±2 237±14Ins(1,3,4)P3 237±16 53±2 335±23Ins(1,2,4,5)P4 225±14 44±3 316±19Ins(3,4,5,6)P4 246±19 54±4 375±26Ins(1,3,4,5,6)P5 449±28 8±0.6 514±43Ins(1,2,3,4,5,6)P6 608±38 1±0.5 836±52

One unit (U) is equal to 1 μmol of phosphate released per minuteper milligram of enzyme.


in catalytic activity, with the D1 repeat exhibiting aspecific activity of 44 and 54 U/mg toward Ins(1,2,4,5)P4 and Ins(3,4,5,6)P4, respectively. Interest-ingly, the presence of a phosphate at the 2-position ofIns(1,2,4,5)P4 hinders the catalytic activity of D1,supporting the hypothesis that this position isimportant for the differential specificity of the D1and D2 repeats. Removal of an additional phosphateto InsP3 did not result in a further increase in thecatalytic activity of the D1 repeat. In all cases, the D2repeat displays catalytic activity against the lowerinositol phosphates, with the D1 repeat having atmost 22% relative activity against Ins(1,4,5)P3.Interestingly, the catalytic activity in wild-typePhyAmm does not equal the sum of the D1 andD2 activities possibly because the D1 and D2 worktogether in the wild-type enzyme. Despite the higherlevel of catalytic activity in D2, it is clear that the D1repeat is catalytically active and displays a markedpreference for lower phosphorylated inositols. Incontrast, the D2 repeat displays a preference forhighly phosphorylated inositols, indicating that theD1 and D2 repeats have evolved different substratespecificities.

Basis for the different specificity in D1 and D2inferred from docking calculations

We used an in silico approach to understand thedifferent substrate specificities of the D1 and D2repeats (Table 3). Docking InsP6, Ins(1,3,4,5,6)P5,and Ins(3,4,5,6)P4 into the active sites of both repeatssuggests that the D1 repeat cannot bind InsP6 in acatalytically competent conformation due to stericinteractions with the β7–β8 hairpin (Fig. S2a). TheInsP6 is placed near the edge of the binding pocket inD1, with the closest phosphate being 6.32 Å from thenucleophile. Docking of Ins(1,3,4,5,6)P5 and Ins(3,4,5,6)P4 placed the substrate 3.11 and 3.20 Åfrom the cysteine (Fig. S2b and c). In contrast, D2 canbind all three substrates in a conformation thatsupports catalysis (Fig. S2d–f). The binding energiescalculated by AutoDock agree with the substratespecificity observed in vitro (Table 2). The bindingenergy for InsP6 is 3 kcal/mol lower in the D2 repeat

than in the D1 repeat. This difference in bindingenergy likely reflects the different positioning ofInsP6 in the D1 and D2 repeats. When InsP5 wasused in the docking, both D1 and D2 show verysimilar binding energies (−6.2 versus −5.9 kcal/mol,respectively). In contrast, docking of InsP4 suggeststhat the D1 repeat shows an energetic preference forbinding InsP4 relative to the D2 repeat (−8.1 versus−7.2 kcal/mol, respectively).To examine if the β7–β8 β-hairpin or differences in

the P-loop sequence are responsible for preventinghydrolysis of InsP6 in the D1 repeat of PhyAmm, weconstructed four separate site-directed mutants (inall cases, the mutants had the D2 catalytic cysteinemutated to serine to abolish D2 activity). The Y251Qand M254A mutants restore residues observed inthe P-loop of D2; the deletion of the β7–β8 β-hairpinloop (residues 183–187) was aimed at increasing thesize of the D1 active site, and a double mutantcontaining both the loop deletion and the Y251Qmutation were produced. Only mutants deleting theβ7–β8 β-hairpin loop produce a detectable signal inour assay; however, the signal is sufficiently close tothe limits of detection that we cannot conclude if themutants are active toward InsP6. These resultssuggest that additional structural determinantsincluding the differences in electrostatic surfacepotential, mutations in the vicinity of the activesite, and steric constraints are likely all involved inthe altered substrate specificity of D1.

Discussion

The structure of PhyAmm

This work represents the first structural studycarried out on a tandemly repeated IPPase. Similarto the other known PTPLP (PhyAsr), PhyAmm hasunique structural features that contribute to theobserved specificity and structural differencesamong PTPLPs and between PTPLPs and othermembers of the PTP superfamily. Like PhyAsr,PhyAmm is a homodimer both crystallographicallyand in solution; however, due to steric constraints,the association of the PTPLP folds in these enzymesis distinct. This shows that the organization ofPTPLP folds within this enzyme family is flexible.Interestingly, tandemly repeated PTP domains arecommonwithin the RPTP family, and the occurrenceof this feature in PTPLPs provides an additionallevel of similarity between these classes of enzyme.In all RPTPs structurally characterized to date, theassociation of PTP domains in the TR is structurallyequivalent, but significantly different from thatobserved in PhyAmm and PhyAsr (Fig. 3). Interest-ingly, in both PhyAmm and PhyAsr, all of the activesites are presented on one side of their respectivedimers. Given that PhyAmm and PhyAsr associatewith the outer membrane of M. multacida and S.ruminantium in vivo,35 the arrangement of activesites may be functionally significant.

Table 3. Docking of IPPs into the D1 and D2 active sites

DockingBinding energy

(kcal/mol)Clustersize

S to Pdistance (Å)

D1 InsP6 −5.95 13 6.32D2 InsP6 −8.95 6 3.47D1 InsP5 −6.21 24 3.11D2 InsP5 −5.91 12 3.13D1 InsP4 −8.06 9 3.20D2 InsP4 −7.19 19 3.14

The binding energy was calculated by Autodock 4.0. The clustersize represents the number of docked structures that areequivalent (all-atom RMSD b2.0 Å) to the lowest energy dockingfrom 50 separate Lamarckian genetic algorithm trials. The bestligand–receptor structure for each of the docking simulations waschosen based on the lowest binding energy. The S to P distance isthe distance (Å) from the catalytic cysteine Sγ to the closestphosphate on the substrate.


PhyAmm D1 active site

Many tandemly repeated RPTPs have a catalyti-cally inactive C-terminal (D2) repeat due to muta-tions in catalytic residues. In several cases, themutations are conservative and do not involve thecatalytic cysteine. RPTPα, RPTPɛ, RPTPσ, RPTPδ,and LAR have conservative mutations in twoessential motifs: (1) the aspartate in the WPD loopis mutated to glutamate and (2) a conserved tyrosinethat is important for substrate binding is mutated tovaline or leucine. Mutation of these residues back tothe consensus sequence restores full catalytic activ-ity to this repeat in RPTPα, RPTPɛ, and LAR.30,36,37In contrast to RPTPs, the different catalytic proper-ties of the PhyAmm repeats are likely a result ofsteric constraints on substrate binding caused by adecrease in the size of the binding pocket (Fig. S2).Interestingly, RPTPα, RPTPɛ, RPTPσ, and LAR alsohave smaller binding pockets in the inactive repeat.Additionally, differences in the electrostatic poten-tial of the D1 active site may contribute to the lack ofactivity towards InsP6 (Fig. 2b). Electrostatic inter-actions with substrate are important in PhyAsr13and likely play an equally important role in thecatalytic activity of PhyAmm. Interestingly, theelectrostatic surface potential of the inactive repeats

of RPTPα, RPTPɛ, and RPTPσ are also less electro-positive than that of the catalytically active repeats.The active site of PhyAmm is largely composed of

structural elements that are unique to PTPLPs. Thesesame regions are sites of an insertion (β7–β8 β-hairpin loop) and deletions (Ω loop and α7 helix) inan alignment of the D1 and D2 repeats (Fig. 4). Theseregions also display the largest deviations in aleast-squares structural superposition of the repeats(Fig. 2a). The inositol phosphatase activity of severalPTPLPs has recently been reported13,16–18 and canbe separated into low-activity and high-activityforms. Interestingly, each of the low-activity PTPLPs(S. ruminantium subsp. lactolytica and Selenomonaslacticifex) has a pattern of insertions and deletionsimilar to those observed in D1 (Fig. 4). Thissuggests that the substrate specificity of D1 andlow-activity PTPLPs depend on each of thesestructural elements.

PhyAmm D1 activity

While it is impractical to assess D1 activity againstall possible myo-inositol polyphosphates, it is clearthat D1 exhibits catalytic activity against several lessphosphorylated forms, in particular InsP4 and InsP3.At the same time, the D2 repeat preferentially

Fig. 4. ClustalW alignment of bacterial PTPLPs. Numbers at the beginning and end of each sequence represent theresidue numbers for the first and last amino acids in that sequence, respectively. Residues in PhyAsr that have beenidentified as being involved in substrate binding are highlighted in bold and the variable positions in the phosphatebinding loop are highlighted in light gray. Secondary structures of the D1 and D2 repeats of PhyAmm are shown on thetop and bottom of the alignment, respectively. The protein abbreviations, source, and GenBank accession number are asfollows: PhyAmmD1,M. multacidaD1 repeat, ABA18187; PhyAsrl, S. ruminantium subsp. lactylitica, ABC69359; PhyBsl, S.lacticifex, ABC69361; PhyAsl, S. lacticifex, ABC69367;Megasphaera elsdenii, ABC69358; PhyAsr, S. ruminantium, AAQ13669;PhyAmmD2, M. multacida D1 repeat, ABA18187. A complete alignment can be found in supplementary information(Fig. S3).


hydrolyzes InsP5 and InsP6, the highly phosphory-lated forms (Table 2). Docking studies suggest theaxial C2 phosphate is at least partly responsible forthe inability of D1 to hydrolyze InsP6. This issupported by the observation that this repeatdisplays catalytic activity against Ins(1,3,4,5,6)P5(Table 2). The eukaryotic IPPase PTEN shows highlevels of catalytic activity against Ins(1,3,4,5,6)P5 butno activity against InsP6 due to steric clashesbetween the protein and the axial phosphate at thetwo positions.38 This is an important finding andsuggests that the axial C2 phosphate position canserve to distinguish inositol phosphate pools in thecell, ensuring that InsP6, the most abundant cellularinositol phosphate, does not interfere with signalingpathways involving other inositol phosphates.Furthermore, the activity of PhyAmm toward myo-inositols lacking a C2 phosphate suggests thatrelated PTPLPs expressed by pathogenic bacteriamay directly affect myo-inositol signaling pathways.What function might the low IPP repeat (D1) play

in PhyAmm? Phytate-degrading IPPases hydrolyzeInsP6 by the sequential, stepwise removal ofphosphates to generate lower-order IPPs that canserve as substrate for the removal of additionalphosphates in an ordered pathway. Many acidphosphatases15 and all characterized PTPLPs13,16–18can remove all of the equatorial phosphates to yieldinorganic phosphate and Ins(2)P1. The presence ofrepeats with a specificity for both highly phos-phorylated (D2) and less phosphorylated (D1) myo-inositols suggests that this enzyme may haveevolved a “divide-and-conquer” approach to hydro-lyzing the numerous myo-inositol derivatives. In thismechanism, D2 targets and hydrolyzes highlyphosphorylated IPPs, including InsP6, in an orderedpathway. At the same time, D1 targets lessphosphorylated IPPs including IPP intermediatesreleased by D2 and unrelated InsP4, InsP3, and InsP2molecules present in the environment. This wouldenable the enzyme to hydrolyze multiple IPP speciessimultaneously, which presumably is more efficientthan hydrolyzing a single species at a time.Interestingly, a highly conserved eukaryotic enzymepossessing both IPP kinase and phytase domainshas recently been described.39 The presence of adual-domain enzyme with both inositol kinase andphosphatase activity and an enzyme with domainsthat exhibit unique IPP phosphatase activity sug-gests that dual-domain proteins may play importantroles in inositol phosphate metabolism.

PhyAmm and tandemly repeated RPTPs

What is the biological advantage of havingtandemly repeated PTPLP or PTP folds that exhibitdifferent specificities? In the case of PhyAmm, itmay function in a divide-and-conquer mechanism ofmyo-inositol hydrolysis. Due to the higher level ofactivity in the D2 repeat, the D1 repeat likely onlyfunctions catalytically when substantial levels oflower phosphorylated IPPs are present. Alterna-tively, the D1 repeat may function to target an

unidentified substrate in vivo or, like RPTPs, play arole in substrate recognition and/or recruitment.Given the association of PhyAmm with the outermembrane of M. multacida in vivo,35 the D1 repeatmay act as a membrane anchor or target amembrane phospholipid.In addition to substrate recognition, targeting, or

regulation,22–25 our results suggest another role forthe D2 repeat in RPTPs retaining all essentialcatalytic residues. These repeats may not be cataly-tically inactive but rather possess a unique specifi-city for substrates that have yet to be identified. Thedifferences between the active and inactive repeatsof RPTPs are remarkably similar to those observedbetween the less active and more active repeats ofPhyAmm. In particular, mutations, insertions, anddeletions within loops that form the active sitereduce the volume of the active site and significantlyalter the electrostatic surface potential, making it lesselectropositive. While these differences are likely toalter substrate specificity, the underlying catalyticmachinery remains intact and may well retain someform of activity against a substrate that has yet to beidentified.

Materials and Methods

Cloning and mutagenesis

TheM.multacida (PhyAmm) open reading frame (minusputative signal peptide) was amplified from genomicDNA using polymerase chain reaction (PCR) with Phu-sion DNA polymerase (Finnzymes). Amino acids werenumbered according to the complete coding sequence ofthe M. multacida protein sequence (ABA18187) includingthe putative signal peptide. The amplified product wasligated into the NdeI site of pET28b and transformed intoEscherichia coli DH5α. Mutagenesis was carried out usingcounter PCR amplification of the expression plasmid aspreviously described.40 All PCR products were sequencedat the University of Calgary Core DNA and ProteinServices facilities. Sequence data were analyzed with theprograms SEQUENCHER 4.0 (Gene Codes Corp.) andMacDNAsis 3.2 (Hitachi Software Engineering Co.). All ofthe primers used in this study were purchased fromIntegrated DNATechnologies.

Purification of PhyAmm

Protein expression was carried out with E. coli BL21(DE3) cells transformed with the pET28b expressionconstruct. Cells were grown to an optical density(600 nm) of 0.6 to 0.8 in Luria–Bertani broth supplementedwith 50 μg/mL kanamycin. Protein expression wasinduced by adding isopropyl-β-D-thiogalactopyranosideto the culture at a final concentration of 1 mM. Theoverexpression was carried out at 37 °C in an incubatingshaker for 18 h. Induced cells were harvested andresuspended in lysis buffer [20 mM KH2PO4 (pH 7.0),300 mM NaCl, 10 mM β-mercaptoethanol, 25 mMimidazole (pH 8.0)]. Cells were lysed by sonication, andcell debris was removed by centrifugation. PhyAmm waspurified to homogeneity by metal chelating affinity (Ni2+-NTA-agarose), cation exchange (Macro-Prep High S, Bio-


Rad), and size-exclusion chromatography. The purity ofthe protein was assessed by SDS-PAGE41 and Coomassiebrilliant blue R-250 staining. Protein concentration wasdetermined by measuring absorbance at 280 nm using theextinction coefficient calculated with PROT-PARAM.42 Thepurified protein was stored in 10 mM KH2PO4 (pH 7.0),100 mM NaCl, 10 mM β-mercaptoethanol, and 0.1 mMEDTA (ethylenediaminetetraacetic acid). Purified proteinwas used immediately or was flash-frozen in liquidnitrogen and stored at 193 K.

Crystallization

Crystallization experiments were conducted usingsitting-drop vapor diffusion with a drop ratio of 2 μl of 5to 10 mg/ml protein solution and 2 μl of reservoir.Crystals were grown at room temperature in 6% to 10%polyethylene glycol 8000, 1% to 5% ethylene glycol, and100 mM Tris (pH 8.0). After several rounds of streakseeding, large, fragile crystals were obtained. Crystalswere cryoprotected by gradually transferring them into asolution containing the crystallization reagents and 25%ethylene glycol, followed by flash-freezing in liquidnitrogen. Heavy-atom-substituted crystals were obtainedby soaking the crystals in a solution consisting of thecrystallization reagents and 20 mM sodium tungstate for1 h prior to freezing.

Data collection and structure determination

Raw diffraction data were processed with MOSFLM43

and scaled with SCALA44 within the CCP4 programsuite.45 Ellipsoidal truncation and anisotropic correctionof the data were performed by the diffraction anisotropyserver to account for anisotropy in the diffraction data.46Native and derivative data sets were collected at 100 K onbeamline 8.3.1 at the Advanced Light Source, and thestructure was solved by the multiwavelength anomalousdispersion method with a WO4

2− soaked crystal. A three-wavelength data set (1.2150, 1.2146, and 1.1579 Å) wascollected at the tungsten LIII edge. Phases (20–2.8 Å) weredetermined using SOLVE47 and density modification wascarried out using RESOLVE.48 SOLVE identified fourtungsten atoms and obtained phases with a figure of meritof 0.51. Solvent flattening with RESOLVE resulted inphases with a figure of merit of 0.76 and a readilyinterpretable map. The resulting electron density map wasused for automated model building with RESOLVE andARPWARP.49 This model served as a starting point foriterative cycles of simulated annealing refinement andpositional and B-factor refinement against a native data set(20–2.3 Å) in CNS 1.250 followed by manual modelbuilding with XFIT.51 Unless indicated otherwise, figureswere prepared with PyMOL (version 0.99).52 PROCHECKwas used throughout refinement to assess the stereo-chemistry of the model.26 Data collection and refinementstatistics are shown in Table 1.

Analysis of enzymatic activity

Analysis of enzymatic activity was carried out at 310 Kas previously described.53 All assays were carried out in50 mM sodium acetate (pH 5.0), the optimum pH of theenzyme, with the ionic strength adjusted to 200 mM by theaddition of NaCl. Activity assays were carried out intriplicate and repeated aminimum of two times. InsP6 was

purchased from Sigma-Aldrich. All other inositol phos-phates were purchased from Echelon Biosciences. Oneunit (U) of activity is defined as 1 μmol of phosphatereleased per minute. Steady-state kinetic data for thehydrolysis of InsP6 were fit to the Michaelis–Mentenequation using nonlinear regression (SigmaPlot 8.0, SystatSoftware Inc.).

Docking calculations

The InsP4, InsP5, and InsP6 models were docked withthe D1 and D2 repeats of PhyAmm using the Lamarckiangenetic algorithm provided by AutoDock version 4.0.54Flexible ligand models were derived from a refined InsP6model (1dkq). Hydrogen atoms and Gasteiger chargeswere added and rotatable bonds (all exocyclic bonds) weredefined within AutoDock. The resulting ligand models[Ins(3,4,5,6)P4, Ins(1,3,4,5,6)P5, and InsP6] have net chargesof −4, −5, and −6 with 12, 15. and 18 rotatable torsionangles, respectively. Polar hydrogen atoms and Gasteigercharges were added to the protein (receptor) model priorto calculation of grid maps in a 15-Å cube (60×60×60points with a 0.25-Å spacing) centered about the activesite. For each active site, 50 docking simulations werecarried out with an initial population of 300 individualsand a maximum of 2.5×106 energy evaluations. Usingthese parameters, the docking simulations for eachligand–receptor pair produced significant “clusters” orgroups of related structures, suggesting adequate sam-pling of both ligand conformation and the active site. Thebest ligand–receptor structure for each of the dockingsimulation was chosen based on the lowest energy.

Protein Data Bank accession number

The coordinates and structure factors for the structure ofPhyAmm have been deposited in the PDB with accessionnumber 3f41.

Acknowledgements

R.J.G. receives doctoral funding from the NaturalSciences and Engineering Research Council ofCanada (NSERC) and Alberta Ingenuity. L.B.S.and S.C.M. are supported by grants from NSERC,the Alberta Heritage Foundation for MedicalResearch (AHFMR), and the Canada Foundationfor Innovation. X-ray diffraction data were collectedat beamline 8.3.1 of the Advanced Light Source atLawrence Berkeley Lab, under an agreement withthe Alberta Synchrotron Institute (ASI). The ASIsynchrotron access program is supported by grantsfrom the Alberta Science and Research Authorityand AHFMR.

Supplementary Data

Supplementary data associated with this articlecan be found, in the online version, at doi:10.1016/j.jmb.2009.05.079


References

1. Irvine, R. F. & Schell, M. J. (2001). Back in the water:the return of the inositol phosphates. Nat. Rev. Mol.Cell Biol. 2, 327–338.

2. Hanakahi, L. A., Bartlet-Jones, M., Chappell, C.,Pappin, D. & West, S. C. (2000). Binding of inositolphosphate to DNA-PK and stimulation of double-strand break repair. Cell, 102, 721–729.

3. York, J. D., Odom, A. R., Murphy, R., Ives, E. B. &Wente, S. R. (1999). A phospholipase C-dependentinositol polyphosphate kinase pathway required forefficient messenger RNA export. Science, 285, 96–100.

4. Macbeth, M. R., Schubert, H. L., Vandemark, A. P.,Lingam, A. T., Hill, C. P. & Bass, B. L. (2005). Inositolhexakisphosphate is bound in the ADAR2 core andrequired for RNA editing. Science, 309, 1534–1539.

5. Tan, X., Calderon-Villalobos, L. I., Sharon, M., Zheng,C., Robinson, C. V., Estelle, M. & Zheng, N. (2007).Mechanism of auxin perception by the TIR1 ubiquitinligase. Nature, 446, 640–645.

6. Chatterjee, S., Sankaranarayanan, R. & Sonti, R. V.(2003). PhyA, a secreted protein of Xanthomonas oryzaepv. oryzae, is required for optimum virulence andgrowth on phytic acid as a sole phosphate source.Mol.Plant-Microbe Interact. 16, 973–982.

7. Majerus, P. W., Zou, J., Marjanovic, J., Kisseleva, M. V.&Wilson, M. P. (2008). The role of inositol signaling inthe control of apoptosis. Adv. Enzyme Regul. 48, 10–17.

8. Frederick, J. P., Mattiske, D., Wofford, J. A., Megosh,L.C., Drake, L. Y., Chiou, S. T. et al. (2005). An essentialrole for an inositol polyphosphate multikinase, Ipk2,in mouse embryogenesis and second messengerproduction. Proc. Natl Acad. Sci. USA, 102, 8454–8459.

9. Verbsky, J., Lavine, K. & Majerus, P. W. (2005).Disruption of the mouse inositol 1,3,4,5,6-pentaki-sphosphate 2-kinase gene, associated lethality, andtissue distribution of 2-kinase expression. Proc. NatlAcad. Sci. USA, 102, 8448–8453.

10. Mullaney, E. J., Daly, C. B. & Ullah, A. H. (2000).Advances in phytase research. Adv. Appl. Microbiol. 47,157–199.

11. Caffrey, J. J., Hidaka, K., Matsuda, M., Hirata, M. &Shears, S. B. (1999). The human and rat forms ofmultiple inositol polyphosphate phosphatase: func-tional homology with a histidine acid phosphataseup-regulated during endochondral ossification. FEBSLett. 442, 99–104.

12. Mullaney, E. J. & Ullah, A. H. (2003). The term phy-tase comprises several different classes of enzymes.Biochem. Biophys. Res. Commun. 312, 179–184.

13. Puhl, A. A., Gruninger, R. J., Greiner, R., Janzen, T. W.,Mosimann, S. C. & Selinger, L. B. (2007). Kinetic andstructural analysis of a bacterial protein tyrosinephosphatase-likemyo-inositol polyphosphatase. ProteinSci. 16, 1368–1378.

14. Chu, H. M., Guo, R. T., Lin, T. W., Chou, C. C., Shr,H. L., Lai, H. L. et al. (2004). Structures ofSelenomonas ruminantium phytase in complex withpersulfated phytate: DSP phytase fold and mechan-ism for sequential substrate hydrolysis. Structure, 12,2015–2024.

15. Konietzny, U. & Greiner, R. (2002). Molecular andcatalytic properties of phytate-degrading enzymes(phytases). Int. J. Food Sci. Technol. 37, 791–812.

16. Puhl, A. A., Greiner, R. & Selinger, L. B. (2009).Stereospecificity of myo-inositol hexakisphosphatehydrolysis by a protein tyrosine phosphatase-like

inositol polyphosphatase from Megasphaera elsdenii.Appl. Microbiol. Biotechnol. 82, 95–103.

17. Puhl, A. A., Greiner, R. & Selinger, L. B. (2008).Kinetics, substrate specificity, and stereospecificity oftwo new protein tyrosine phosphatase-like inositolpolyphosphatases from Selenomonas lacticifex. Biochem.Cell Biol. 86, 322–330.

18. Puhl, A. A., Greiner, R. & Selinger, L. B. (2008). Aprotein tyrosine phosphatase-like inositol polypho-sphatase from Selenomonas ruminantium subsp. lacti-lytica has specificity for the 5-phosphate of myo-inositol hexakisphosphate. Int. J. Biochem. Cell Biol.40, 2053–2064.

19. Nakashima, B. A., McAllister, T. A., Sharma, R. &Selinger, L. B. (2007). Diversity of phytases in therumen. Microb. Ecol. 53, 82–88.

20. Lim, B. L., Yeung, P., Cheng, C. & Hill, J. E. (2007).Distribution and diversity of phytate-mineralizingbacteria. ISME J. 1, 321–330.

21. Andersen, J. N., Mortensen, O. H., Peters, G. H.,Drake, P. G., Iversen, L. F., Olsen, O. H. et al. (2001).Structural and evolutionary relationships amongprotein tyrosine phosphatase domains. Mol. Cell Biol.21, 7117–7136.

22. Kashio, N., Matsumoto, W., Parker, S. & Rothstein, D.M. (1998). The second domain of the CD45 proteintyrosine phosphatase is critical for interleukin-2secretion and substrate recruitment of TCR-ζ in vivo.J. Biol. Chem. 273, 33856–33863.

23. Yang, J., Groen, A., Lemeer, S., Jans, A., Slijper,M., Roe,S. M. et al. (2007). Reversible oxidation of themembrane distal domain of receptor PTPα is mediatedby a cyclic sulfenamide. Biochemistry, 46, 709–719.

24. Blanchetot, C. & den Hertog, J. (2000). Multipleinteractions between receptor protein-tyrosine phos-phatase (RPTP) α and membrane-distal protein-tyrosine phosphatase domains of various RPTPs.J. Biol. Chem. 275, 12446–12452.

25. Ng, D. H., Maiti, A. & Johnson, P. (1995). Pointmutation in the second phosphatase domain of CD45abrogates tyrosine phosphatase activity. Biochem.Biophys. Res. Commun. 206, 302–309.

26. Laskowski, R. A., MacArthur, M. W., Moss, D. S. &Thornton, J. M. (1993). PROCHECK: a program tocheck the stereochemical quality of protein structures.J. Appl. Crystallogr. 26, 283–291.

27. Krissinel, E. & Henrick, K. (2007). Inference ofmacromolecular assemblies from crystalline state.J. Mol. Biol. 372, 774–797.

28. Holm, L., Kaariainen, S., Rosenstrom, P. & Schenkel,A. (2008). Searching protein structure databases withDaliLite v.3. Bioinformatics, 24, 2780–2781.

29. Lee, J. O., Yang, H., Georgescu, M. M., Di Cristofano,A., Maehama, T., Shi, Y. et al. (1999). Crystal structureof the PTEN tumor suppressor: implications for itsphosphoinositide phosphatase activity and mem-brane association. Cell, 99, 323–334.

30. Nam, H. J., Poy, F., Krueger, N. X., Saito, H. &Frederick, C. A. (1999). Crystal structure of the tandemphosphatase domains of RPTP LAR. Cell, 97, 449–457.

31. Guex, N. & Peitsch, M. C. (1997). SWISS-MODEL andthe Swiss-PdbViewer: an environment for compara-tive protein modeling. Electrophoresis, 18, 2714–2723.

32. Petrey, D. & Honig, B. (2003). GRASP2: visualization,surface properties, and electrostatics of macromole-cular structures and sequences. Methods Enzymol. 374,492–509.

33. Yuvaniyama, J., Denu, J. M., Dixon, J. E. & Saper,M. A. (1996). Crystal structure of the dual spe-


cificity protein phosphatase VHR. Science, 272,1328–1331.

34. Laskowski, R. A., Hutchinson, E. G., Michie, A. D.,Wallace, A. C., Jones, M. L. & Thornton, J. M. (1997).PDBsum: a Web-based database of summaries andanalyses of all PDB structures. Trends Biochem. Sci. 22,488–490.

35. D'Silva, C. G., Bae, H. D., Yanke, L. J., Cheng, K. J. &Selinger, L. B. (2000). Localization of phytase inSelenomonas ruminantium and Mitsuokella multiacidusby transmission electron microscopy. Can. J. Microbiol.46, 391–395.

36. Buist, A., Zhang, Y. L., Keng, Y. F., Wu, L., Zhang, Z. Y.& den Hertog, J. (1999). Restoration of potentprotein-tyrosine phosphatase activity into the mem-brane-distal domain of receptor protein-tyrosinephosphatase α. Biochemistry, 38, 914–922.

37. Lim, K. L., Ng, C. H. & Pallen, C. J. (1999). Catalyticactivation of the membrane distal domain of proteintyrosine phosphatase epsilon, but not CD45, by twopoint mutations. Biochim. Biophys. Acta, 1434, 275–283.

38. Caffrey, J. J., Darden, T., Wenk, M. R. & Shears, S. B.(2001). Expanding coincident signaling by PTENthrough its inositol 1,3,4,5,6-pentakisphosphate 3-phosphatase activity. FEBS Lett. 499, 6–10.

39. Mulugu, S., Bai, W., Fridy, P. C., Bastidas, R. J., Otto,J. C., Dollins, D. E. et al. (2007). A conserved family ofenzymes that phosphorylate inositol hexakispho-sphate. Science, 316, 106–109.

40. Street, I. P., Coffman, H. R. & Poulter, C. D. (1991).Isopentenyl diphosphate isomerase. Site-directed muta-genesis of Cys139 using “counter” PCR amplification ofan expression plasmid. Tetrahedron, 47, 5919–5924.

41. Laemmli, U. K. (1970). Cleavage of structural proteinsduring the assembly of the head of bacteriophage T4.Nature, 227, 680–685.

42. Gill, S. C. & von Hippel, P. H. (1989). Calculation ofprotein extinction coefficients from amino acidsequence data. Anal. Biochem. 182, 319–326.

43. Leslie, A. G. W. (1992). Recent changes to theMOSFLM package for processing film and imageplate data. Joint CCP4 + ESF-EAMCB Newsletter onProtein Crystallography.

44. Evans, P. R. (1997). Scala. Joint CCP4 and ESF-EAMCBNewsletter 22–24.

45. CCP4. (1994). The CCP4 suite: programs for proteincrystallography. Acta Crystallogr., Sect. D: Biol.Crystallogr. 50, 760–763.

46. Strong, M., Sawaya, M. R., Wang, S., Phillips, M.,Cascio, D. & Eisenberg, D. (2006). Toward thestructural genomics of complexes: crystal structureof a PE/PPE protein complex from Mycobacteriumtuberculosis. Proc. Natl Acad. Sci. USA, 103, 8060–8065.

47. Terwilliger, T. C. & Berendzen, J. (1999). AutomatedMAD and MIR structure solution. Acta Crystallogr.,Sect. D: Biol. Crystallogr. 55, 849–861.

48. Terwilliger, T. C. (2001). Map-likelihood phasing. ActaCrystallogr., Sect. D: Biol. Crystallogr. 57, 1763–1775.

49. Perrakis, A., Morris, R. & Lamzin, V. S. (1999).Automated protein model building combined withiterative structure refinement. Nat. Struct. Biol. 6,458–463.

50. Brunger, A. T. (2007). Version 1.2 of the crystallographyand NMR system. Nat. Protoc. 2, 2728–2733.

51. McRee, D. E. (1999). XtalView/Xfit—a versatileprogram for manipulating atomic coordinates andelectron density. J. Struct. Biol. 125, 156–165.

52. DeLano, W. L. (2008). The PyMOL MolecularGraphics System. DeLano Scientific LLC, PaloAlto, CA.

53. Yanke, L. J., Bae, H. D., Selinger, L. B. & Cheng, K. J.(1998). Phytase activity of anaerobic ruminal bacteria.Microbiology, 144, 1565–1573.

54. Morris, G. M., Goodsell, D. S., Halliday, R. S., Huey,R., Hart, W. E., Belew, R. K. & Olson, A. J. (1998).Automated docking using a Lamarckain geneticalgorithm and an empirical binding free energyfunction. J. Comput. Chem. 19, 1639–1662.


structural analysis of a multifunctional, tandemly repeated...

Documents