structureof francisellatularensis acpafaculty.missouri.edu/~tannerjj/tannergroup/pdfs/acpajbc.pdfis...

Structure of Francisella tularensis AcpAPROTOTYPE OF A UNIQUE SUPERFAMILY OF ACID PHOSPHATASESAND PHOSPHOLIPASES C*

Received for publication, July 5, 2006, and in revised form, August 7, 2006 Published, JBC Papers in Press, August 9, 2006, DOI 10.1074/jbc.M606391200

Richard L. Felts‡, Thomas J. Reilly§, and John J. Tanner‡¶1

From the ‡Department of Chemistry, the §Department of Veterinary Pathobiology and Veterinary Medical Diagnostic Laboratory,and the ¶Department of Biochemistry, University of Missouri, Columbia, Missouri 65211

AcpA is a respiratory burst-inhibiting acid phosphatase fromthe Centers for Disease Control and Prevention Category A bio-terrorism agent Francisella tularensis and prototype of a super-family of acid phosphatases and phospholipases C. We reportthe 1.75-A resolution crystal structure of AcpA complexed withthe inhibitor orthovanadate, which is the first structure of any F.tularensisprotein and the first for anymember of this superfam-ily. The core domain is a twisted 8-stranded �-sheet flanked bythree �-helices on either side, with the active site located abovethe carboxyl-terminal edge of the �-sheet. This architecture isunique among acid phosphatases and resembles that of alkalinephosphatase. Unexpectedly, the active site features a serinenucleophile and metal ion with octahedral coordination. Struc-ture-based sequence analysis of the AcpA superfamily predictsthat the hydroxyl nucleophile and metal center are also presentin AcpA-like phospholipases C. These results imply a phospho-lipaseC catalyticmechanism that is radically different from thatof zinc metallophospholipases.

Francisella tularensis is a highly infectious intracellular bac-terial pathogen and the cause of tularemia (1). The organismcan be isolated from numerous rodent hosts and arthropodvectors, readily grown in broth culture, andmechanically aero-solized. It is one of the most infectious pathogenic agentsknown, requiring fewer than ten organisms to establish infec-tion (1). Inhalation of aerosolized F. tularensis can result inpneumonic tularemia (2), which has a case fatality rate of up to

30% if untreated (1). The U.S. Centers for Disease Control andPrevention considers F. tularensis to be a Category A bioterror-ism agent, which has led to renewed interest in identifyinggenes and pathways that underlie virulence to facilitate devel-opment of new antimicrobial drugs and vaccines (1, 3, 4).Acid phosphatase A from F. tularensis (AcpA)2 is a highly

expressed 57-kDa polyspecific periplasmic acid phosphatase(ACP) (5). AcpA hydrolyzes a variety of substrates, includingp-nitrophenylphosphate (pNPP), p-nitrophenylphospho-rylcholine (pNPPC), peptides containing phosphotyrosine, ino-sitol phosphates, AMP, ATP, fructose 1,6-bisphosphate,glucose and fructose 6-phosphates, NADP�, and ribose5-phosphate (5, 6). The enzyme is inhibited by the metal oxya-nions orthovanadate, molybdate, and tungstate. Based onamino acid sequence analysis, AcpA is distinct from histidineACPs (7) and purple ACPs (8), as well as class A, B, and Cbacterial nonspecific ACPs (9).Purified AcpA inhibits the respiratory burst of stimulated

neutrophils, which suggests that AcpA helps the pathogenelude the host oxidative defense system during the initial stagesof macrophage infection (5). Furthermore, proteomics studieshave shown that AcpA is expressed at a higher level in virulentF. tularensis strains compared with the nonvirulent vaccinestrain (10). Most recently, it has been shown that a mutantstrain of F. tularensis subspecies novicida lacking a functionalacpA gene is less virulent in mice than the wild-type strain dueto a defect in phagosomal escape.3 Thus, AcpA appears to beimportant for survival of themicrobe at two critical junctures ofinfection: colonization and intracellular survival.Amino acid sequence alignments show that AcpA belongs to

a superfamily of bacterial enzymes that includes ACPs andphospholipasesC (PLCs) froma variety ofmicrobial pathogens,including Pseudomonas aeruginosa,Mycobacterium tuberculo-sis, Bordetella pertussis, and several Burkholderia species (11).AcpA is the only characterized enzyme from theACP branch ofthe superfamily. PLCs from this superfamily are important vir-ulence factors in P. aeruginosa (12) and M. tuberculosis (13)infections, with the hemolytic PLC from P. aeruginosa (PlcH)being the best characterized example from the PLC branch ofthe superfamily (14). PlcH is particularly interesting, because it

* This work was supported by National Institutes of Health Grant U54AI057160 to the Midwest Regional Center of Excellence for Biodefense andEmerging Infectious Diseases Research and the University of MissouriResearch Board. The Advanced Light Source is supported by the Director,Office of Science, Office of Basic Energy Sciences, Materials Sciences Divi-sion, of the U.S. Dept. of Energy under Contract No. DE-AC03-76SF00098 atLawrence Berkeley National Laboratory. Beamline 8.3.1 was funded by theNational Science Foundation, the University of California, and HenryWheeler. Use of the Argonne National Laboratory Structural Biology Cen-ter beamlines at the Advanced Photon Source was supported by the U.S.Dept. of Energy, Office of Energy Research, under Contract No. W-31-109-ENG-38. The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby marked“advertisement” in accordance with 18 U.S.C. Section 1734 solely to indi-cate this fact.

The atomic coordinates and structure factors (code 2D1G) have been depositedin the Protein Data Bank, Research Collaboratory for Structural Bioinformat-ics, Rutgers University, New Brunswick, NJ (http://www.rcsb.org/).

1 To whom correspondence should be addressed: Dept. of Chemistry, Uni-versity of Missouri, 125 Chemistry Bldg., 601 S. College Ave., Columbia,MO 65211. Tel.: 573-884-1280; Fax: 573-882-2754; E-mail: [email protected].

2 The abbreviations used are: AcpA, acid phosphatase A from F. tularensis;ACP, acid phosphatase; pNPP, p-nitrophenylphosphate; pNPPC, p-nitro-phenylphosphorylcholine; PLC, phospholipase C; PlcH, hemolytic PLCfrom P. aeruginosa; ASA, human arylsulfatase A; AlkP, alkalinephosphatase.

3 J. S. Gunn, personal communication.

THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 281, NO. 40, pp. 30289 –30298, October 6, 2006© 2006 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in the U.S.A.

OCTOBER 6, 2006 • VOLUME 281 • NUMBER 40 JOURNAL OF BIOLOGICAL CHEMISTRY 30289

at UN

IVE

RS

ITY

OF

MIS

SO

UR

I on October 28, 2006

ww

w.jbc.org

Dow

nloaded from

http://www.jbc.org

is a multifunctional enzyme that displays sphingomyelin syn-thase activity in addition to PLC activity (15). PLCs of theAcpA/PlcH superfamily share no sequence homology with thewell studied zinc metallophospholipases Clostridium perfrin-gens �-toxin and Bacillus cereus phosphatidylcholine-prefer-ring PLC (16), which suggests that PLCs of the AcpA/PlcHsuperfamily have a novel, and as yet uncharacterized, catalyticmechanism. To gain insights into the structural basis of thecatalytic activity of enzymes of the AcpA/PlcH superfamily, wehave determined the crystal structure of AcpA bound to thecompetitive inhibitor orthovanadate.

EXPERIMENTAL PROCEDURES

Crystallization and X-ray Diffraction Data Collection—Ex-pression of recombinant F. tularensis AcpA in Escherichia coli,protein purification, and the growth of three different crystalforms were described previously (17). Structure determinationutilized crystal form III, which was obtained by incubatingthe enzyme with the competitive inhibitor sodium orthovana-date (Na3VO4, 5mM) prior to crystallization and using polyeth-ylene glycol 1500 as the precipitating agent (17). These crystalshave space group C2221 with unit cell dimensions a � 112 Å,b � 144 Å, c � 124 Å, two molecules per asymmetric unit, and43% solvent content.The derivative used for phasing was produced by soaking an

AcpA/orthovanadate crystal in 40 mM Sm(C2H3O2)3 for 10

min. Diffraction data extending to 2.4-Å resolution were col-lected from the Sm derivative at Advanced Photon Sourcebeamline 19-ID using � � 1.6531 Å, which corresponds to anenergy between the L-I and L-II absorption edges of Sm. Dataprocessing was done with HKL2000 (Table 1). Anomalous dif-ference Patterson maps showed several strong features on theu � 0 Harker section.

The data set used for phase extension and refinement calcu-lations at 1.75-Å resolution was collected from an AcpA/or-thovanadate crystal atAdvancedLight Source beamline 8.3.1. Asecond data set, which was used for anomalous difference Fou-rier analysis of the active site metal center, was collected fromanother AcpA/orthovanadate crystal at beamline 8.3.1. Thisdata set was collected at low energy (� � 1.74 Å) to enhance theanomalous signal of the metal ion. Both data sets were pro-cessed with HKL2000 (18). See Table 1 for a summary of dataprocessing statistics.Phasing and Refinement Calculations—The structure was

solved using single wavelength anomalous diffraction phasing.SnB (19) was used to identify a 10-atom anomalous constella-tion for the Sm derivative, which was input to SHARP (20) forsingle wavelength anomalous diffraction phase calculationsand solvent flattening. The resulting SHARP phases had a fig-ure of merit of 0.84 for reflections to 2.4-Å resolution. An elec-tron density map calculated from the SHARP phases clearly

TABLE 1Data collection and refinement statisticsValues for the outer resolution shell of data are given in parentheses.

Orthovanadate (high energy) Orthovanadate (low energy) Sm derivativeWavelength (Å) 1.1271 1.740 1.6531Space group C2221 C2221 C2221Unit cell dimensions (Å) a � 112.1, b � 144.4, c � 123.9 a � 111.2, b � 142.6, c � 125.6 a � 112.2, b � 144.2, c � 123.9Diffraction resolution (Å) 50-1.75 (1.81-1.75) 50-2.20 (2.28-2.20) 50-2.40 (2.48-2.40)No. of observations 397,516 295,644 572,236No. of unique reflections 98,137 51,195 39,618Redundancy 4.1 (4.0) 5.8 (4.9) 14.4 (13.4)Completeness (%) 97.2 (94.9) 99.9 (99.9) 99.9 (99.9)Average I/� 26.6 (2.2) 16.8 (3.4) 38.6 (23.8)Rsym (I) 0.047 (0.357) 0.098 (0.443) 0.065 (0.118)No. of non-hydrogen atoms 8,182No. of residues in chain A 481No. of residues in chain B 472No. of water molecules 550Rcryst 0.198 (0.244)Rfree

a 0.231 (0.291)Root mean square deviationbBond lengths (Å) 0.012Bond angles (deg.) 1.7

Ramachandran plotcFavored (%) 96.4Allowed (%) 3.1

Average B-factors (Å2)Protein 25Orthovanadate 27Active site metal ion 16Water 28

PDB accession code 2D1Ga A 5% random test set.b Compared to the Engh and Huber force field (50).c The Ramachandran plot was generated with RAMPAGE (51).

Structure of F. tularensis AcpA

30290 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 281 • NUMBER 40 • OCTOBER 6, 2006

at UN

IVE

RS

ITY

OF

MIS

SO

UR


ww

w.jbc.org

Dow

nloaded from

http://www.jbc.org

showed features resembling protein secondary structural ele-ments. A partial backbone tracing consisting of a few �-helicesand�-strandswas obtainedwith the automatedmodel buildingprogramMAID (21). TheMAID tracing was used to determinethe noncrystallographic symmetry transformation relating thetwo protein molecules in the asymmetric unit. In preparationfor noncrystallographic symmetry averaging, the programsMAMA (22) and CNS (23) were used to create a mask thatcovered one of the molecules in the asymmetric unit. TheSHARP phases were then improved and extended to 1.75-Åresolutionwith 2-fold noncrystallographic symmetry averagingand solvent flipping in CNS. The 1.75-Å resolution density-modified phases were input to ARP/wARP (24) for automatedelectron density map interpretation. The best model fromARP/wARP included the backbone for 97% of the expected res-idues in the asymmetric unit and 83% of the expected sidechains. The model was improved with several rounds of modelbuilding in COOT (25) followed by refinement with REFMAC5(26). Refinement statistics are listed in Table 1.The asymmetric unit includes 953 amino acid residues

belonging to twoAcpAmolecules (chains labeledA andB). The

following sections of the polypep-tide chains are disordered: A1–A4,A15–A18, A490–A498, B1–B5,B12–B18, B129–B137, and B490–B494. The root mean square differ-ence between chains A and B is 0.27Å for C� atoms and 0.55 Å for allatoms, which indicates that the twochains have nearly identical con-formations. Each AcpA moleculecontains one orthovanadate ion(HVO4

2�) and one metal ion boundin the active site. The metal ion wasmodeled as Ca2� for purposes ofcrystallographic refinement butappears with atom name X1 andresidue name UNK in the coordi-nate file deposited in the ProteinData Bank (PDB (27)) to indicatethat the identity of the metal isunknown at this time. The solventstructure includes 550 orderedwater molecules and 4 bound poly-ethylene glycol fragments. There isalso a decavanadate ion (V10O28

6�)bound in a crystal contact region,where it interacts with the carboxyl-terminal histidine affinity tag of oneof theAcpAmolecules. Coordinatesand structure factor amplitudeshave been deposited in the PDBunder accession code 2D1G.Site-directed Mutagenesis and

Activity Assays—AcpA mutant Ser-1753 Ala was generated using theQuikChangemutagenesis kit (Strat-agene), and the mutation was veri-

fied by DNA sequencing. The mutant enzyme was expressedand purified using methods employed for AcpA (17). SDS-PAGE analysis showed that Ser-175 3 Ala had the expectedmolecular weight, and Western blots using rabbit anti-AcpApolyclonal and anti-His tag antibodies were positive. Enzymaticactivities of AcpA and Ser-1753 Ala were measured using adiscontinuous colorimetric assay with pNPP and pNPPC assubstrates (6).

RESULTS

Overall Structure of AcpA—The structure of AcpA com-prises three domains and has approximate dimensions of 60 Å� 48 Å� 66 Å. The core domain is a highly twisted, 8-stranded�-sheet flanked by three �-helices on either side (Fig. 1A). Thestrand order of the�-sheet is 12, 2, 11, 10, 1, 9, 3, then 8, with allbut strand 11 in parallel (Fig. 2). There are two smaller domainslocated above the carboxyl-terminal edge of the 8-stranded�-sheet. One of these small domains consists of residues47–147 and features four short �-helices (labeled A–D) con-nected by rather long loops (Fig. 1, blue domain). This domainhas a disulfide bond linking Cys-102 and Cys-138 (Figs. 1A and

FIGURE 1. Overall structure of AcpA. A, stereographic ribbon drawing of AcpA chain A. The vanadate inhibitor(magenta/red) and bound metal ion (yellow) are drawn in sphere mode. Strands and helices of the core �-sheetdomain are colored green and red, respectively. The two smaller domains are colored blue (residues 47–147)and orange (flap domain, residues 258 –283). Selected strands and helices are labeled as in the topologydiagram (see Fig. 2). Disulfide-bonded Cys side chains are drawn as gray spheres. The thin dashed line indicatesdisordered residues 15–18. B, surface topography of the active site entrance. The orthovanadate inhibitor(green/red) and the four water molecules in the trough (red) are drawn as spheres. The �-sheet core domain iscolored white. The two smaller domains are colored blue (residues 47–147) and orange (residues 258 –283).Residues lining the active site trough are indicated in the schematic diagram on the right. This figure, andothers, were created with PyMol (W. L. DeLano (2002) The PyMOL Molecular Graphics System, www.pymol.org).



at UN

IVE

RS

ITY

OF

MIS

SO

UR


ww

w.jbc.org

Dow

nloaded from

http://www.jbc.org

2). As discussed below, this domain forms part of the dimerinterface. The other small domain (residues 258–283) consistsof a pair of 2-stranded anti-parallel �-sheets (�4–�7), whichresembles a flap (Fig. 1, orange domain), and there is a disulfidebond that links Cys-269 of this domain to Cys-216 of the�-sheet core domain.

Purified AcpA forms an apparent dimer according to analyt-ical ultracentrifugation and gel-filtration chromatography data(6). The two proteins chosen for the asymmetric unit (Fig. 3A)form the largest intermolecular surface between any two pro-

teins in the crystal lattice, based on analysiswith PISA (28). Thisinterface buries 2398 Å2 of surface area, whereas the next larg-est interface buries only 932 Å2 of surface area. Furthermore,this interface had the highest possible PISA complexation sig-nificance score (1.0), compared with 0 for all other possibleinterfaces. It is concluded that the pair of protein molecules inthe asymmetric unit represents the AcpA dimer in solution.The small helical domain (residues 47–147) and the �12 face

of the �-sheet core domain form the dimer interface (Fig. 3A).Secondary structural elements involved in dimerization include

�B and its adjacent loops (residues73–87), the loop following �C (res-idues 116–119), a 10-residue sec-tion of the loop connecting �10 and�11 (394–404), �12 and its adja-cent loops (residues 425–433), andresidues in a loop near the car-boxyl terminus (residues 459–466).Together, these residues form a flatsurface (Fig. 3B) that spans 40 Å inone direction and 30 Å in the other.The dimer interface is highly

hydrophilic, and hydrogen bondingappears to play amajor role in dimerstability. There are 14 direct inter-subunit hydrogen bonds (Table 2)but no ion pairs. Hydrogen-bondingside chains in the interface includeAsn-74, Thr-79, Gln-81, Asn-116,Gln-401, Asp-404, and Tyr-428.Note that the intersubunit hydro-gen bonds display 2-fold symmetry(Table 2). In addition, there are 16interfacial water molecules thatmediate 20 intersubunit hydrogenbonds (Fig. 3C). As with the inter-subunit hydrogen bonds, the 16

FIGURE 2. Secondary structure topology diagram of AcpA. �-Helices are shown as rectangles labeled A–J,and �-strands are shown as arrows numbered 1–12. Cys residues are represented in yellow with disulfidesbridges shown as dashed lines. The green boxes denote active site residues, with red numbering for residuescoordinating to the bound metal and blue numbering for residues interacting with the vanadate inhibitor.

FIGURE 3. Dimeric structure of AcpA. A, the AcpA dimer is drawn in ribbon representation and covered with a semi-transparent molecular surface. Thesurfaces are colored blue for one subunit and yellow for the other subunit. The orientation of the yellow subunit is nearly identical to that of Fig. 1A. For eachsubunit, the coloring scheme of the ribbon is the same as that used in Fig. 1. The red spheres represent interfacial water molecules. The arrow denotes thenon-crystallographic molecular 2-fold symmetry axis. B, this is the same as panel A except that the yellow subunit has been removed to show the flatness of thedimer interface. C, this is the same as panel A except that the blue subunit has been removed and the yellow subunit has been rotated 90° so that the interfacialsurface points toward the viewer. Note the 2-fold symmetry in the constellation of interfacial water molecules.



at UN

IVE

RS

ITY

OF

MIS

SO

UR


ww

w.jbc.org

Dow

nloaded from

http://www.jbc.org

bridging water molecules obey the 2-fold symmetry of thedimer (Fig. 3C). Although hydrogen bonding is prominent inthe interface, a few nonpolar residues contribute significantsurface area to the interface. For example, Leu-82 packs againstLeu-119, whereas Leu-433 from one subunit packs against Leu-433 of the opposite subunit at the centroid of the dimer.

Active Site Architecture and Implications for CatalyticMechanism—The location of the active site was clearly indi-cated by a strong electron density feature corresponding to thebound orthovanadate inhibitor (Fig. 4A). The active site islocated above the carboxyl-terminal edge of the 8-stranded�-sheet near�1 and�F (Fig. 1A). The inhibitor binds in one endof a 12-Å long trough, which is located in a broad, shallowdepression formed by residues from all three domains (Fig. 1B).Note that four water molecules are bound in the trough (Fig.1B). The shape of the trough suggests that it may be involved inbinding the leaving group of the substrate. This idea was testedby modeling pNPP in the active site. We found that the nitro-phenyl group of pNPP fits edgewise into the trough (waterremoved) without causing steric clash.Surprisingly, there is a metal ion bound in the active site,

based on the observation of a very strong electron density fea-ture that could not be assigned to the protein, inhibitor, orsolvent (Fig. 4A). Four lines of evidence suggest that this featurerepresents a metal ion. First, it is surrounded by an octahedralarray of six oxygen ligands (Fig. 4, A and B): Glu-43, Asn-44,Ser-175, Asp-386, Asp-387, and the vanadate inhibitor. Threeof the six coordinating ligands are carboxyl groups, which issuggestive of a bound metal ion with a charge of at least �2.

FIGURE 4. Active site of AcpA. A, stereographic drawing of the AcpA metal center covered by two electron density maps. The cyan cage represents a simulatedannealing �A-weighted mFo � DFc electron density map (3 �). The metal ion, orthovanadate, and surrounding residues within 3.9 Å were omitted prior tosimulated annealing refinement and map calculation. The red cage represents an anomalous difference Fourier map (3.5 �) calculated with phases from thefinal model and anomalous differences from the low energy orthovanadate data set. The orthovanadate inhibitor is shown in magenta/red, and the metal ionis colored yellow. Protein residues appear in white. B, stereographic drawing of the AcpA active site highlighting protein-inhibitor electrostatic interactions(dashed lines). The orthovanadate inhibitor is shown in magenta/red, and the metal ion is colored yellow. Protein residues appear in white. C, schematic diagramof the active site.

TABLE 2Intersubunit hydrogen bonds

Hydrogen bond partnersDistance

Chain A Chain BÅ

Asn-74 (N�2) Asp-404 (O�1) 3.1Asn-74 (N) Asp-404 (O�2) 3.0Asp-404 (O�1) Asn-74 (N�2) 2.8Asp-404 (O�2) Asn-74 (N) 3.1Thr-79 (O�1) Gly-117 (O) 2.6Gly-117 (O) Thr-79 (O�1) 2.7Gln-81 (N�2) Asn-116 (O�1) 3.3Asn-116 (O�1) Gln-81 (N�2) 3.1Gln-401 (N�2) Tyr-428 (OH) 3.2Gln-401 (N�2) Val-429 (O) 3.0Gln-401 (O�1) His-431 (N) 2.8Tyr-428 (OH) Gln-401 (N�2) 3.2Val-429 (O) Gln-401 (N�2) 3.1His-431 (N) Gln-401 (O�1) 2.9



at UN

IVE

RS

ITY

OF

MIS

SO

UR


ww

w.jbc.org

Dow

nloaded from

http://www.jbc.org

Second, the proposed metal site corresponded to the highestpeak in an anomalous difference Fourier map calculated fromdiffraction data collected at low energy (� � 1.74Å). The anom-alous difference density feature was quite prominent whenviewed at the 3.5 � contour level (Fig. 4A, red cage) andremained visible even at 10�. Third, the proposedmetal ion sitewas the second strongest binding site of the Sm derivative usedfor single wavelength anomalous diffraction phasing. We notethat lanthanides readily replace metal ions of protein active/binding sites, including Mg2�, Ca2�, and first row transitionmetal ions (29). Fourth, as described in the next section, struc-tural homologs of AcpA, such as arylsulfatase A (ASA) andalkaline phosphatase (AlkP), have metal ions bound in theactive site at locations that are structurally analogous to theproposed AcpA metal site.

Electron densitymaps were analyzed to gain insights into theelemental identity of the bound metal ion. The anomalous dif-ference Fourier peak corresponding to the metal ion was muchstronger than that of the vanadate (Fig. 4A, red cage), whichimplies that the metal ion is a stronger anomalous scattererthan the V atom of the inhibitor. This result is consistent withthe active site metal ion being a first row transitionmetal. Also,several simulated annealing refinements were performedagainst the 1.75-Å data set with differentmetal ionsmodeled inthe active site. The resulting difference electron density maps(�A-weighted mFo � DFc) suggested that the metal ion has atleast the number of electrons of Ca2� and is more likely a firstrow transition metal ion. Therefore, the metal ion was conser-vatively modeled as Ca2� in the current structure pending fur-ther biochemical and analytical studies of the metal content ofAcpA.The orthovanadate inhibitor exhibits distorted trigonal bipy-

ramidal geometry and is bound by six side chains and themetalion (Fig. 4, B and C). The inhibitor axial oxygen atom interactswith His-106 and His-350, whereas the equatorial oxygenatoms bind to Asn-44, His-287, His-288, Asp-208, His-350,and the metal ion. Asp-208 appears to share a proton with theinhibitor.The location of Ser-175 relative to the inhibitor andmetal ion

suggests that it plays the role of nucleophile that attacks thesubstrate P atom. The hydroxyl oxygen atom of Ser-175 is 1.8 Åfrom the metal ion and 2.2 Å from the inhibitor V atom (Fig.4C). Ser-175 appears to be in an ideal location for backsidenucleophilic attack at the substrate P atom. Thus, the active sitestructure strongly suggests that Ser-175 is the enzyme nucleo-phile, and the role of the metal ion is to activate Ser-175 fornucleophilic attack (Fig. 5A, step 2). This hypothesis impliesformation of a covalent Ser-175-phosphoryl intermediate dur-ing catalysis (Fig. 5A, step 2).We engineered the Ser-1753Ala mutant to test the impor-

tance of this residue for catalysis. The mutant exhibited nodetectable activity using either pNPPor pNPPCas the substrateeven at enzyme concentrations �0.01 mM and substrate con-centrations up to 20mM.Thus, Ser-175 plays an essential role incatalysis, which is consistent with our hypothesis that it is theenzyme nucleophile.Hydrolysis of the Ser-175-phosphoryl intermediate (Fig. 5A,

step 3) presumably requires a general base to activate a watermolecule. Residues that bind the inhibitor and that are locatedon the solvent side of the active site are possible candidates forthis role. Asp-208 is, perhaps, the most likely candidate,because aspartic acid residues serve as the general base in otherphosphatases, such as protein tyrosine phosphatase (30), andthe carboxyl of Asp-208 forms a hydrogen bond (2.8 Å) with awater molecule (Wat-285) in our structure.Comparison to Other Protein Structures—To understand the

relationship of AcpA to other phosphatases, we searched thePDB for structural homologs of AcpA using the programDALI(31). Surprisingly, the closest homolog was not a phosphatasebut was human arylsulfatase A (ASA, PDB code 1AUK, DALIZ-score � 16), followed by phosphoglycerate mutase (PDBcode 1EJJ, Z � 15), phosphonoacetate hydrolase (PDB code1EI6, Z� 10), and E. coli alkaline phosphatase (AlkP, PDB code

FIGURE 5. Summaries of the proposed catalytic mechanisms of AcpA andASA. A, the proposed mechanism of AcpA. The steps shown are: 1, substratebinding (see Fig. 4 for more details); 2, nucleophilic attack at the substrate Patom by Ser-175, which is activated by a metal ion; and 3, hydrolysis of theenzyme-phosphoryl intermediate to release the phosphate product andregenerate the active site. B, the ASA mechanism proposed previously (33,40). The steps shown are: 1, formation of a gem-diol from the reaction of waterwith FGly-69; 2, nucleophilic attack at the substrate S atom by one of thegem-diol hydroxyl groups and release of ROH; 3, elimination of the productsulfate facilitated by proton transfer to His-125; and 4, regeneration of theactive site.



at UN

IVE

RS

ITY

OF

MIS

SO

UR


ww

w.jbc.org

Dow

nloaded from

http://www.jbc.org

1B8J, Z � 9). All four enzymes belong to the AlkP superfamily,which has been described in detail (32). AcpA shares a common�-sheet core domain and active site location with AlkP super-family members. The shared secondary structural elementsconsist of the middle six strands of the central �-sheet alongwith the six flanking �-helices (Fig. 6A).We note that DALI didnot identify a single ACP with structural similarity to AcpA.Structural homology of AcpA to AlkP enzymes, especially

ASA and AlkP, extends to details of the active site. ASA uses aformylglycine (FGly-69) as the nucleophile and binds a singlemetal ion in the active site (Ca2� orMg2�) (33). The array ofmet-al-binding ligands in ASA is remarkably similar to that of AcpA(Fig. 6B). In both enzymes, themetal ion has octahedral coordina-tion with three carboxyl groups, an asparagine side chain, thenucleophilic oxygen atom, and the inhibitor/phosphoryl. Recog-nition of the substrate phosphoryl is also similar in the twoenzymes. For example,His-287 andHis-350ofAcpAare structur-ally analogous to His-125 and His-229, respectively, of ASA (Fig.6B). Moreover, imidazole nitrogen atoms of AcpA His-106 andHis-288 overlap nearly perfectly with the �-amino groups of ASALys-302 and Lys-123 (Fig. 6B). The onlymajor difference betweenthe twoactive sites, besides thenucleophile, appears tobeAsp-208of AcpA, which is replaced by Val-91 in ASA.Like AcpA, AlkP has a Ser nucleophile (34), but AlkP binds

threemetal ions: two Zn2� (ZnI and ZnII) andMg2�. TheAcpAmetal ion and AlkP ZnII occupy analogous locations in theirrespective structures, although ZnII has tetrahedral coordina-

tion and the AcpA metal ion has octahedral coordination (Fig.6C). Despite the difference in coordination geometry, Glu-43and Asp-386 of AcpA are analogous to Asp-51 and Asp-369 ofAlkP. There are also similarities between the two enzymes interms of binding the substrate phosphoryl group. For example,AcpA His-106 is analogous to AlkP His-412, whereas the sidechains of AcpAHis-287 and AlkP Arg-166 occupy similar loca-tions in their respective structures (Fig. 6C). One notable dif-ference between the two enzymes is that AlkP does not have anacidic residue equivalent to AcpA Asp-208.Conservation of Active Site Residues in the AcpA/PlcH

Superfamily—Analysis of available sequence databases usingBLAST (35) shows that close homologs of AcpA are present inother bacteria, including Burkholderia mallei, Corynebacte-rium jeikeium, andBradyrhizobium japonicum. These proteinshave 520–648 residues and share 38–46% global amino acidsequence identity with AcpA. The 10 residues that contact van-adate or the metal ion in AcpA are identically conserved inthese proteins:Glu-43,Asn-44,His-106, Ser-175,Asp-208,His-287, His-288, His-350, Asp-386, and Asp-387 (Fig. 7, upper sixprotein sequences). Moreover, residues that form a disulfidebond (Cys-216 to Cys-269) and an ion pair (Asp-393 to Arg-414) are also identically conserved (Fig. 7, upper six sequences).The conserved ion pair links residues within the long loopbetween �10 and �11. This loop also contains metal-bindingresidues Asp-386 and Asp-387, and thus the conserved ion pairmay be crucial for stabilizing the metal-binding site.

FIGURE 6. Comparison of AcpA to ASA and AlkP. A, ribbon representation of the conserved cores of AcpA (red), ASA (green, PDB code 1N2K (40)), and AlkP(blue, PDB code 1B8J (34)). The orthovanadate inhibitor (magenta/red) and metal ion (yellow) of AcpA are drawn as spheres. Strands and helices of AcpA arelabeled as in the topology diagram (Fig. 2). B, stereographic drawing of the ASA active site (green) superimposed onto the AcpA active site (white). Orthovana-date of AcpA is colored magenta/red. The phosphoryl intermediate of ASA is colored green/red. The metal ions of each structure are colored yellow. Selectedresidues are labeled as AcpA/ASA. C, stereographic drawing of the AlkP active site (green) superimposed onto the AcpA active site (white). The orthovanadateinhibitors of both structures are colored magenta/red. The AcpA metal ion appears in yellow, and the zinc ions of AlkP are colored cyan. Selected residues arelabeled as AcpA/AlkP.



at UN

IVE

RS

ITY

OF

MIS

SO

UR


ww

w.jbc.org

Dow

nloaded from

http://www.jbc.org



at UN

IVE

RS

ITY

OF

MIS

SO

UR


ww

w.jbc.org

Dow

nloaded from

http://www.jbc.org

TheAcpA structure also provides insights into the active sitearchitectures of enzymes from the PLC branch of the AcpA/PlcH superfamily. PlcH has 730 residues, and the amino-termi-nal two-thirds of the enzyme shares 23% amino acid sequenceidentity with AcpA. The AcpA structure, however, suggeststhat the sequence homology between AcpA and PlcH is muchstronger within the active site. For example, five of the tenAcpA active site residues (Glu-43, Asn-44, His-106, His-350,and Asp-386) are identically conserved in PlcH (Fig. 7). More-over, PlcH residue Thr-178 aligns with AcpA nucleophile Ser-175, and PlcH residue Glu-358 aligns with AcpAmetal-bindingresidue Asp-387 (Fig. 7). In addition, ion pair residues Asp-393and Arg-414 of AcpA are also present in the PlcH sequence(Asp-364 andArg-401). All nine of these conserved residues arealso present in close homologs of PlcH (Fig. 7, lower nine pro-tein sequences). Thus, enzymes in the PLC branch of the AcpA/PlcH superfamily likely retain the essential hydroxyl nucleo-phile and octahedral metal-binding site of AcpA.

DISCUSSION

The AcpA Family of Phosphatases—The structure reportedhere shows that AcpA from F. tularensis is distinct from otherACPs in terms of overall fold and active site architecture andthat AcpA shares a common�/� corewith enzymes in theAlkPsuperfamily. Amajor result from our work is the discovery thatAcpA is a Ser-basedmetallophosphatase. This result was unex-pected because AcpA had been predicted to be a Cys-basedphosphatase based on a putative catalytic motif in residues216–224 (CX5KSG). Furthermore, there were no reports in theliterature showing that metal ion is required for activity. TheAcpA structure thus provides the framework for experimentsthat will establish new paradigms for the AcpA family of ACPs.The structural similarity between AcpA and AlkP enzymes

shedsnew light on the catalyticmechanismofAcpA. For example,the mechanism of AlkP involves two consecutive in-line nucleo-philic attacks at the phosphorous, which results in retention ofconfiguration at the P center (36–39). ZnII activates Ser-102 forthe first nucleophilic attack, andZnI activates awatermolecule forthe second attack. We propose that AcpA follows an AlkP-likemechanism with AcpA Ser-175 serving as nucleophile, the AcpAmetal ion playing the role of AlkP ZnII, and a protein side chain,possibly Asp-208, substituting for AlkP ZnI.

Interestingly, although the active site structures of ASA andAcpA are very similar (Fig. 6B), it is unlikely that they share acommon catalytic mechanism. The ASA mechanism proceedsthrough a gem-diol intermediate formedby reaction ofwaterwithFGly-69 (Fig. 5B, step 1). One of the hydroxyl groups of the gem-diol is activated fornucleophilic attackat the substrateSatom(Fig.5B, step 2). Proton transfer from the other hydroxyl to His-125facilitates release of the product sulfate (Fig. 5B, step 3). Becausethenucleophilic oxygenatomleaveswith theproduct, the reaction

occurs with overall inversion of configuration of the sulfate (33,40). Formation of a gem-diol intermediate from AcpA Ser-175 ischemically unfavorable. Thus it appears that AcpA andASA havequite different catalytic mechanisms despite having nearly identi-cal metal ion-binding sites, similar constellations of substrate-binding residues, and a common protein fold.Catalytic Mechanism of AcpA-like PLCs—The AcpA struc-

ture also provides new insights into the catalytic mechanism ofPLCs of theAcpA/PlcH superfamily. As discussed above, struc-ture-based sequence analysis suggests that these PLCs have ahydroxyl nucleophile (e.g.Thr-178 in PlcH) coupled to an octa-hedral metal center in the active site. Involvement of a threo-nine nucleophile implies a double-displacement catalyticmechanism for PlcH and related PLCs in which a covalentintermediate is formed between the nucleophilic threonine andthe phosphoryl head group of the substrate. This predictedmechanism is radically different from that of zincmetallophos-pholipases C. perfringens �-toxin and B. cereus phosphatidyl-choline-preferring PLC, which utilize a single nucleophilicattack on the phosphodiester substrate by an activated watermolecule without formation of a covalent intermediate (41).The AcpA structure provides a basis for designing experimentsto test this proposed mechanism.Role of AcpA in Virulence of F. tularensis—F. tularensis is a

facultative intracellular pathogen whose primary target ofinfection is themacrophage (1). An essential aspect of virulenceis the ability of F. tularensis to escape phagosomal containment,which leads to over 1000-fold replication of the pathogen in thecytoplasm and eventual apoptosis of the infected macrophage.Several proteins are thought to contribute to intramacrophagegrowth and survival of F. tularensis, including putative tran-scriptional regulators MglA and MglB (42, 43), phosphatasessuch as AcpA, and proteins with unknown functions IglC (44,45) and FTT0918 (46).A current challenge is to understand the role of AcpA in

intramacrophage survival. Our structure-based sequence anal-ysis shows that essential elements of the AcpA active site areshared by PlcH-like PLCs, in particular, the hydroxyl nucleo-phile and mononuclear metal center. This structural similarityraises the possibility that AcpA exhibits PLC activity and sug-gests a new hypothesis that AcpA facilitates phagosomal escapeby hydrolyzing phospholipids of the phagosomal inner mem-brane. Interestingly, AcpA efficiently hydrolyzes the phospho-lipid-like substrate pNPPC (6). However, this is not necessarilyan accurate indicator of bona fide PLC activity, because pNPPClacks a hydrocarbon tail. We note that purified AcpA does notexhibit lecithinase activity on egg yolk agar, nor does it lyse redblood cells (data not shown). Thus, additional studies areneeded to determine whether AcpA exhibits true PLC activity.

FIGURE 7. Amino acid sequence alignment of AcpA/PlcH superfamily members. The upper six proteins of the alignment are AcpA and five close homologsof AcpA. These proteins share 37– 84% pairwise amino acid sequence identity. The lower nine proteins are PlcH and its close homologs. These proteins have31–75% pairwise identity. Numbers above the alignment correspond to AcpA residue numbering. Residues identically conserved in all sequences are denotedby white letters on red background. Other regions of high sequence conservation are indicated by boxed red letters on white background. Triangles below thealignment denote active site residues that contact either the metal ion or vanadate inhibitor in AcpA. Stars below the alignment indicate disulfide-bonded Cysin AcpA. Dotted lines connect disulfide-bonded Cys residues. Filled rectangles below the alignment indicate conserved residues Asp-393 and Arg-414 (AcpAnumbering), which form an ion pair in AcpA. This figure was created with ESPript (49).



at UN

IVE

RS

ITY

OF

MIS

SO

UR


ww

w.jbc.org

Dow

nloaded from

http://www.jbc.org

A second hypothesis about the role of AcpA in virulence isthat AcpAmight affect host signaling pathways by dephospho-rylation of host proteins, inositol phosphates, or phosphoi-nositides, the latter being critically important for phagosomeformation (47) and respiratory burst activation (48). The wideand relatively flat surface surrounding the active site (Fig. 1B) iscompatible with AcpA docking to a protein substrate.Finally, it is possible that AcpA functions in a phosphate

retrieval system that is activated upon phagosomal containment.AcpA would be an effective phosphate scavenger because of itsbroadsubstrate specificityandhighabundance.Central toall threehypotheses is the question of whether expression of AcpA is con-trolled byMglA/B, as has been suggested by Baron andNano (42).The AcpA structure provides a framework for exploring thesehypotheses and for designing AcpA and PlcH inhibitors.

Acknowledgments—We thank James Holton of Advanced LightSource beamline 8.3.1 and Steve Ginell of Advanced Photon Sourcebeamline 19ID for help with data collection. We thank MichaelCalcutt for help with gene sequencing.

REFERENCES1. Oyston, P. C., Sjostedt, A., and Titball, R.W. (2004)Nat. Rev.Microbiol. 2,

967–9782. Dennis, D. T., Inglesby, T. V., Henderson, D. A., Bartlett, J. G., Ascher,

M. S., Eitzen, E., Fine, A. D., Friedlander, A. M., Hauer, J., Layton, M.,Lillibridge, S. R., McDade, J. E., Osterholm, M. T., O’Toole, T., Parker, G.,Perl, T. M., Russell, P. K., and Tonat, K. (2001) J. Am. Med. Assoc. 285,2763–2773

3. Sjostedt, A. (2003) Curr. Opin. Microbiol. 6, 66–714. Larsson, P., Oyston, P. C., Chain, P., Chu, M. C., Duffield, M., Fuxelius,

H. H., Garcia, E., Halltorp, G., Johansson, D., Isherwood, K. E., Karp, P. D.,Larsson, E., Liu, Y., Michell, S., Prior, J., Prior, R., Malfatti, S., Sjostedt, A.,Svensson, K., Thompson, N., Vergez, L.,Wagg, J. K.,Wren, B.W., Lindler,L. E., Andersson, S. G., Forsman, M., and Titball, R. W. (2005)Nat. Genet.37, 153–159

5. Reilly, T. J., Baron, G. S., Nano, F. E., and Kuhlenschmidt, M. S. (1996)J. Biol. Chem. 271, 10973–10983

6. Reilly, T. J., Felts, R. L., Henzl, M. T., Calcutt, M. J., and Tanner, J. J. (2006)Protein Expr. Purif. 45, 132–141

7. Van Etten, R. L., Davidson, R., Stevis, P. E., MacArthur, H., and Moore,D. L. (1991) J. Biol. Chem. 266, 2313–2319

8. Vincent, J. B., and Crowder, M. W. (1995) Phosphatases in Cell Metabo-lism and Signal Transduction: Structure, Function, andMechanism of Ac-tion, R.G. Landes Company, Austin

9. Rossolini, G.M., Schippa, S., Riccio, M. L., Berlutti, F., Macaskie, L. E., andThaller, M. C. (1998) Cell. Mol. Life Sci. 54, 833–850

10. Hernychova, L., Stulik, J., Halada, P., Macela, A., Kroca, M., Johansson, T.,and Malina, M. (2001) Proteomics 1, 508–515

11. Stonehouse, M. J., Cota-Gomez, A., Parker, S. K., Martin, W. E., Hankin,J. A., Murphy, R. C., Chen, W., Lim, K. B., Hackett, M., Vasil, A. I., andVasil, M. L. (2002)Mol. Microbiol. 46, 661–676

12. Ostroff, R. M., Wretlind, B., and Vasil, M. L. (1989) Infect. Immun. 57,1369–1373

13. Raynaud, C., Guilhot, C., Rauzier, J., Bordat, Y., Pelicic, V., Manganelli, R.,Smith, I., Gicquel, B., and Jackson, M. (2002)Mol. Microbiol. 45, 203–217

14. Ostroff, R. M., Vasil, A. I., and Vasil, M. L. (1990) J. Bacteriol. 172,5915–5923

15. Luberto, C., Stonehouse,M. J., Collins, E. A., Marchesini, N., El-Bawab, S.,Vasil, A. I., Vasil, M. L., and Hannun, Y. A. (2003) J. Biol. Chem. 278,32733–32743

16. Titball, R. W. (1993)Microbiol. Rev. 57, 347–36617. Felts, R. L., Reilly, T. J., and Tanner, J. J. (2005) Biochim. Biophys. Acta

1752, 107–11018. Otwinowski, Z., and Minor, W. (1997)Methods Enzymol. 276, 307–32619. Weeks, C. M., and Miller, R. (1999) J. Appl. Crystallogr. 32, 120–12420. Bricogne, G., Vonrhein, C., Flensburg, C., Schiltz, M., and Paciorek, W.

(2003) Acta Crystallogr. Sect. D Biol. Crystallogr. 59, 2023–203021. Levitt, D. G. (2001) Acta Crystallogr. Sect. D Biol. Crystallogr. 57,

1013–101922. Kleywegt, G. J., and Jones, T. A. (1999) Acta Crystallogr. Sect. D Biol.

Crystallogr. 55, 941–94423. Brunger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P.,

Grosse-Kunstleve, R.W., Jiang, J. S., Kuszewski, J., Nilges,M., Pannu,N. S.,Read, R. J., Rice, L. M., Simonson, T., and Warren, G. L. (1998) ActaCrystallogr. Sect. D Biol. Crystallogr. 54, 905–921

24. Morris, R. J., Perrakis, A., and Lamzin, V. S. (2002)ActaCrystallogr. Sect. DBiol. Crystallogr. 58, 968–975

25. Emsley, P., and Cowtan, K. (2004) Acta Crystallogr. Sect. D Biol. Crystal-logr. 60, 2126–2132

26. Winn, M. D., Murshudov, G. N., and Papiz, M. Z. (2003) Methods Enzy-mol. 374, 300–321

27. Berman, H. M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T. N., Weissig,H., Shindyalov, I. N., and Bourne, P. E. (2000) Nucleic Acids Res. 28,235–242

28. Krissinel, E., and Henrick, K. (2005) in Computational Life Sciences: FirstInternational Symposium (Berthold, M. R., Glen, R., Diederichs, K., Kohl-bacher,O., and Fischer, I., eds) pp. 163–174, Springer, Konstanz,Germany

29. Turro, C., Fu, P. K.-L., and Bradley, P.M. (2003) inMetal Ions in BiologicalSystems (Sigel, A., and Sigel, H., eds) Vol. 40, pp. 323–353,Marcel Dekker,New York

30. Lohse, D. L., Denu, J. M., Santoro, N., and Dixon, J. E. (1997) Biochemistry36, 4568–4575

31. Holm, L., and Sander, C. (1993) J. Mol. Biol. 233, 123–13832. Galperin, M. Y., and Jedrzejas, M. J. (2001) Proteins 45, 318–32433. Lukatela, G., Krauss, N., Theis, K., Selmer, T., Gieselmann, V., von Figura,

K., and Saenger, W. (1998) Biochemistry 37, 3654–366434. Holtz, K. M., Stec, B., and Kantrowitz, E. R. (1999) J. Biol. Chem. 274,

8351–835435. Altschul, S. F., Gish,W.,Miller,W.,Myers, E.W., and Lipman, D. J. (1990)

J. Mol. Biol. 215, 403–41036. Coleman, J. E. (1992) Annu. Rev. Biophys. Biomol. Struct. 21, 441–48337. Vincent, J. B., Crowder, M. W., and Averill, B. A. (1992) Trends Biochem.

Sci. 17, 105–11038. Knowles, J. R. (1980) Annu. Rev. Biochem. 49, 877–91939. Hendry, P., and Sargeson, A.M. (1990) in Progress in Inorganic Chemistry:

Bioinorganic Chemistry (Lippard, S. J., ed) Vol. 38, pp. 201–258, JohnWiley & Sons, New York

40. Chruszcz, M., Laidler, P., Monkiewicz, M., Ortlund, E., Lebioda, L., andLewinski, K. (2003) J. Inorg. Biochem. 96, 386–392

41. Martin, S. F., and Hergenrother, P. J. (1999) Biochemistry 38, 4403–440842. Baron, G. S., and Nano, F. E. (1998)Mol. Microbiol. 29, 247–25943. Lauriano, C. M., Barker, J. R., Yoon, S. S., Nano, F. E., Arulanandam, B. P.,

Hassett, D. J., and Klose, K. E. (2004) Proc. Natl. Acad. Sci. U. S. A. 101,4246–4249

44. Lai, X. H., Golovliov, I., and Sjostedt, A. (2004)Microb. Pathog. 37, 225–23045. Santic, M., Molmeret, M., Klose, K. E., Jones, S., and Kwaik, Y. A. (2005)

Cell. Microbiol. 7, 969–97946. Twine, S., Bystrom, M., Chen, W., Forsman, M., Golovliov, I., Johansson,

A., Kelly, J., Lindgren, H., Svensson, K., Zingmark, C., Conlan, W., andSjostedt, A. (2005) Infect. Immun. 73, 8345–8352

47. Simonsen, A., Wurmser, A. E., Emr, S. D., and Stenmark, H. (2001) Curr.Opin. Cell Biol. 13, 485–492

48. Baggiolini,M., andWymann,M. P. (1990)Trends Biochem. Sci. 15, 69–7249. Gouet, P., Courcelle, E., Stuart, D. I., and Metoz, F. (1999) Bioinformatics

15, 305–30850. Engh, R. A., and Huber, R. (1991) Acta Crystallogr. Sect. A 47, 392–40051. Lovell, S. C., Davis, I. W., Arendall, W. B., 3rd, de Bakker, P. I., Word, J. M.,

Prisant, M. G., Richardson, J. S., and Richardson, D. C. (2003) Proteins 50,437–450



at UN

IVE

RS

ITY

OF

MIS

SO

UR


ww

w.jbc.org

Dow

nloaded from

http://www.jbc.org

structureof francisellatularensis acpafaculty.missouri.edu/~tannerjj/tannergroup/pdfs/acpajbc.pdfis...

Documents