Characterization and sequencing of a respiratory burst-inhibiting acid phosphatase from Francisella tularensis.

Acid phosphatases (Acp) of intracellular pathogens have recently been implicated as virulence factors that enhance intracellular survival through suppression of the respiratory burst. We describe here the identification, purification, characterization, and sequencing of a novel burst-inhibiting acid phosphatase from the facultative intracellular bacterium, Francisella tularensis. Similar to other the burst-inhibiting Acps, F. tularensis Acp (AcpA) is tartrate-resistant and has broad substrate specificity. The AcpA enzyme is unique, however, in that it is easily released from the bacterial cell in soluble form, is a basic enzyme, suppresses the respiratory burst of not only fMet-Leu-Phe but also phorbol 12-myristate 13-acetate-stimulated neutrophils and does not fit into any of the three currently recognized classes of acid phosphatase. We also report the complete nucleotide sequence of the gene acpA, encoding AcpA, and the deduced primary structure of its encoded polypeptide. Comparative sequence analyses of AcpA is discussed. To our knowledge, this is the first report describing the cloning and sequencing of a burst-inhibiting acid phosphatase.

Acid phosphatases (EC 3.1.3.2) are a ubiquitous class of enzymes that catalyze the hydrolysis of phosphomonoesters at an acidic pH. In addition to mobilization of phosphate, some members of this class of enzymes perform many essential biological functions including regulation of metabolism, energy conversion, and signal transduction. These enzymes have been identified and characterized from many eukaryotic and prokaryotic sources and comprise several distinct subgroups based on substrate specificity, molecular weight, and sensitivity to known inhibitors.
In the past decade, a new emphasis has been placed on understanding the role acid phosphatases may play in microbial pathogenesis. Comprehensive studies of acid phosphatases purified from Leishmania donovani (1) and Legionella micda-dei (2) suggest that members of a class of tartrate-resistant, nonspecific acid phosphatases (TRAPs) 1 may play a crucial role in the survival of intracellular pathogens within a host's phagocytic cells. An exciting discovery in these studies was that TRAPs purified from these organisms suppressed the respiratory burst of activated human neutrophils (3,4). Although information is now becoming available about some of the enzymatic, biochemical, and biophysical properties of the burstinhibiting TRAPs, unequivocal proof of the role of these enzymes as virulence factors in vivo has yet to be obtained. Progress toward this goal is currently limited by the lack of protein or gene sequence information and the absence of isogenic TRAP mutants.
Francisella tularensis is the etiologic agent of the potentially fatal human disease tularemia and is capable of survival and multiplication within a host's professional phagocytes as well as nonphagocytic cells (5,6). Although many studies have been conducted into the host's immune response to Francisella infection, until recently relatively little attention has been focused on biochemical characterization of purified macromolecules which may function as virulence factors in these organisms (7). In initial studies, we found a particular strain of F. tularensis (ATCC 6223, B38) to be enriched in acid phosphatase activity. The Acp specific activity in this strain was greater than previously reported for any other bacterial or protozoan organism. It was also easily solubilized in the absence of detergents allowing relatively large amounts of enzyme to be purified to apparent homogeneity. We describe here the identification, purification, and characterization of some of the unique properties of this burst-inhibiting acid phosphatase (AcpA) as well as its complete primary structure derived from cloning and nucleotide sequencing of the AcpA gene (acpA).

EXPERIMENTAL PROCEDURES
Bacterial Strains and Materials-F. tularensis strains (ATCC 6223 and 29684) were purchased from American Type Culture Collection (Rockville, MD), and strain NDBR 101 LVS was obtained from The National Drug Company (Philadelphia, PA). Francisella novicida was purchased from the ATCC (15482). Strains of Mycobacteria were provided by Dr. John Urbance (University of Illinois, Urbana, IL). Bacteriological media including Bacto Cystine Heart agar (CHA) and Iso-Vitalex were obtained from Baxter (McGraw Park, IL). All other chemicals, unless stated otherwise, were purchased from Sigma and were of the highest purity available. Chromatography resins were pur-chased from Pharmacia Biotech Inc. Protein electrophoresis reagents and ampholytes were obtained from Bio-Rad Laboratories. SDS-PAGE molecular weight standards were obtained from Integrated Separation Systems (Hyde Park, MA) or NOVEX (San Francisco, CA). Heteropolymolybdate complexes were gifts from Dr. Robert Glew (University of New Mexico School of Medicine, Albuquerque, NM).
Culture Conditions-F. tularensis strains 6223, 29684, NDBR 101 LVS, and F. novicida were cultured on hemoglobin-enriched Bacto Cystine Heart agar for 1-5 days at 37°C. The organisms were passaged once after being received from ATCC; aliquots were then frozen at Ϫ80°C and used for inoculation of CHA for purification of the enzyme. Bacteria were harvested by scraping the cultures from the agar. Harvested material was suspended in 100 ml of buffer A (50 mM sodium acetate buffer, pH 6.0).
Screening of F. tularensis Hydrolase Activities-Bacterial cultures from CHA were resuspended to a protein concentration of 1 mg/ml, 200-l aliquots were added to api-ZYM® strips (bioMerieux Vitek, Inc., Hazelwood, MO), the strips were incubated for 12 h at 37°C and then analyzed for semiquantitation of F. tularensis hydrolase activities according to the manufacturer's instructions.
Enzyme Assays-Acp activity was measured fluorometrically using an Aminco-Bowman spectrophotofluorometer. The 0.3-ml standard assay mixture contained 0.2 M sodium acetate buffer, pH 6.0, 1.0 mM 4-methylumbelliferyl phosphate (MUP), and varying amounts of enzyme. The mixtures were incubated at 37°C for 15 min and 1.2 ml of 0.5 M glycine, pH 10, was added to stop the reaction. Under these conditions, enzyme activity was linear with the amount of enzyme added. During kinetic experiments, enzyme activity was linear with time for at least 60 min. Only initial rates (slopes within the first 15 min) were used for calculation of enzyme activity and associated kinetic parameters. One unit of enzyme activity is defined as the amount of enzyme required to convert 1 nmol of substrate to product per h. Assays to determine the pH optimum were performed using either 0.2 M MES or 0.2 M HEPES as the buffer, and the final substrate concentration was 1.0 mM. Determination of the Michaelis-Menten constant for MUP and tyrosine phosphate was performed using 0.06 unit of AcpA and a wide range (K m /10 to 5 K m ) of each substrate. Replicates of five were tested at each substrate concentration. Data were analyzed using a nonlinear, least squares regression computer program (8) graciously supplied by Dr Stephen P. J. Brooks, Carleton University, Ottawa, Canada. Phospholipase C (PLC) activity was measured by monitoring the hydrolysis of p-nitrophenylphosphorylcholine as described previously (9).
Substrate Specificity Assays-Substrate specificity was determined by measuring the release of inorganic phosphate from phosphomonoester substrates (including MUP) using the method of Lanzetta et al. (10). This assay was also used for the determination of the pH optimum of AcpA for phosphomonoesters other than MUP. Phosphatidylinositol phosphates were assayed in the presence of 1.0% Triton X-100.
Peptide-tyrosine Phosphatase Activity of AcpA-A synthetic peptide p60 src (TEPQpYQPGE) containing a single phosphorylated tyrosine was synthesized by the University of Illinois Genetic Engineering facility according to a previously described method (11). Purity of the peptide was assessed by reversed-phase HPLC on a Vydac 218TP54 analytical column, and the product was found to be 98% pure. Mass spectrometry analysis of the peptide gave the expected molecular ion, and the amino acid analysis was within 5% of the expected values in all cases. AcpA catalyzed dephosphorylation of the monophosphorylpeptide and determination of kinetic parameters were performed as described above.
Purification of F. tularensis Acid Phosphatase (Acp)-All procedures were conducted at 4°C unless otherwise noted. The bacterial culture (16 g obtained by scraping bacteria growth from 100 CHA plates (150 mm)) was suspended in buffer A and homogenized using a motor driven Potter-Elvehjem homogenizer. An equal volume of an extraction buffer consisting of buffer A containing 2 M NaCl, 0.5% sodium cholate, 0.2 mM EDTA, 0.2 mM dithiothreitol, 75 g/ml phenylmethylsulfonyl fluoride, and 5 g/ml Pepstatin A was added to the homogenate, the mixture was stirred for 12 h and centrifuged at 200,000 ϫ g for 1.5 h. The supernatant, at a protein concentration of 5 mg/ml, was dialyzed for 12 h at 4 -6°C against three changes (6 liters each) of buffer A. This dialyzed supernatant, designated supernatant I, was again centrifuged at 200,000 ϫ g to remove a precipitate which had formed during dialysis. This second supernatant, containing 97% of the starting activity, was designated supernatant II. Supernatant II (210 ml) was applied to a S-Sepharose cation exchange column (3 ϫ 18 cm) pre-equilibrated with buffer A. The column was washed with 500 ml of buffer A and a 0 -0.5 M linear NaCl gradient (600 ml) in buffer A was applied to the column at 0.5 ml/min. A single peak of phosphatase activity was eluted between 0.17 and 0.26 M NaCl. Active fractions were pooled and concentrated by ultrafiltration. The concentrated sample was then applied and eluted (0.2 ml/min) from a Sephadex G-100 superfine column (1.5 ϫ 95 cm) equilibrated in buffer A containing 0.3 M NaCl. The sample eluted as a single peak, and fractions containing Acp activity were pooled and concentrated as described above. The sample (1.2 ml) was then applied in four separate 0.3-ml aliquots to a Superdex 75 HR 10/30 FPLC column and eluted with buffer A containing 0.3 M NaCl at 0.5 ml/min. Fractions were collected, analyzed for Acp activity, and monitored for protein purity by SDS-PAGE. Enzymatic activity in fractions other than those two containing the highest activities were contaminated and thus not pooled. The purification results are summarized in Table II.
Radioiodination-Iodination of AcpA was performed using IODO-GEN (Pierce). Ten g of pooled Acp from the Superdex 75 column was added to an IODOGEN-coated tube containing 10 l of 0.5 M Tris buffer, pH 7.5, and 0.5 mCi of 125 I. The reaction was incubated at room temperature for 3 min, after which 200 l of a 10 mg/ml solution of KI was added to stop the reaction. Labeled enzyme was separated from unincorporated 125 I by desalting on a GF-5 Excellulose column (Pierce Chemical Co.) pretreated with 1.0 ml of a 10 mg/ml suspension of BSA and equilibrated in 0.5 M Tris, pH 7.5. Void volume fractions containing radioactivity were pooled and analyzed by SDS-PAGE and autoradiography.
Preparation of Rabbit Anti-F. tularensis Acp (AcpA) Antisera-Purified AcpA (719 g) was dialyzed against 0.9% NaCl, filter-sterilized, and emulsified in complete Freund's adjuvant. The immunogen was then injected subcutaneously at multiple sites into a New Zealand White rabbit. Twenty six days after primary immunization, the immune response was boosted by a single subcutaneous injection with 200 g of purified AcpA emulsified in Ribi Adjuvant (Ribi Biologicals). Serum was collected by ear vein puncture 7 days following the second injection.
Purification of Anti-Acp Antibodies-Monospecific anti-AcpA antibodies (IgG) were purified from anti-AcpA antisera by repeated absorption and centrifugation with nonrelevant antigen as described previously (12). Nonrelevant antigen used was either pellet I obtained following removal of supernatant I during purification of AcpA as described above or an E. coli Y1090 freeze-thaw extract. Anti-AcpA IgG was then purified from the adsorbed antiserum by protein A-Sepharose affinity chromatography.
Polyacrylamide Gel Electrophoresis and Detection of Acid Phosphatase by Western Blot-Sodium dodecyl sulfate-PAGE was performed as described by Laemmli (13). Polyacrylamide gels were 3% T stacking and 7.5% T resolving. Molecular weight of acid phosphatase was estimated using NOVEX Mark 12 molecular weight standards and GelReader for Macintosh Version 2.0 software (University of Illinois National Center for Supercomputing Applications). Western blot detection of AcpA was performed as described previously (14). Purified rabbit anti-F. tularensis AcpA antibody was used as the primary antibody (1:12,000 dilution), and goat anti-rabbit IgG (H ϩ L) was conjugated to alkaline phosphatase as the secondary antibody (1:1000 dilution).
Mass Spectrometry of F. tularensis Acid Phosphatase-The purified acid phosphatase was subjected to matrix-assisted laser desorption time of flight mass spectrometry using a VG TofSpec mass spectrometer. Approximately 5 pmol of AcpA was embedded in a matrix of sinapinic acid and irradiated at 337 nm. The instrument is equipped with a 33 nm nitrogen laser with a 5-ns maximum pulse width, a 50-J minimum output, and a 150 ϫ 250 micron spot size. Data acquisition and processing were performed by a VG OPUS data system and a VAXstation 4000 computer.
Isolation of Neutrophils-Neutrophil-enriched cell fractions were isolated from freshly collected normal porcine blood (50 ml) as described previously (15). The neutrophil fraction was resuspended in HEPES/ NaCl buffer (200 mM HEPES, 0.9% NaCl, pH 7.3) to 1 ϫ 10 7 cells/ml and stored on ice until use (within 2 h of isolation) in respiratory burst assays.
Measurement of Respiratory Burst in Neutrophils-Respiratory burst activity of isolated porcine neutrophils was measured in the presence and absence of AcpA by following the production of superoxide using modifications of a previously described method (16). Briefly, the superoxide dismutase-inhibitable reduction of ferricytochrome c at 550 nm was continuously measured at 37°C using either a Beckman DU-50 spectrophotometer or an Aminco dual-beam recording spectrophotom-eter (DW 2000). The standard assay (0.4 ml) was performed in HEPES/ NaCl buffer containing 25 mM HEPES, 150 mM NaCl, 0.90 mM CaCl 2 , and 0.50 mM MgCl 2 , pH 6.8 instead of the modified Dulbecco's phosphate-buffered saline medium. O 2 Ϫ production was initiated by either the addition of 1 l of PMA (1 g/l in dimethyl sulfoxide) or 5 l of fMLP (100 M in dimethyl sulfoxide). The purified acid phosphatase sample used in these studies was also analyzed for superoxide dismutase activity (17), catalase activity (18), and as a direct scavenger of superoxide using a xanthine oxidase assay (19).
Cyanogen Bromide Cleavage of AcpA-Cyanogen bromide cleavage of AcpA was done as described previously (20). Briefly, 50 g of AcpA was digested in the dark at 20°C for 20 h. The resultant peptide sample was analyzed by gel electrophoresis using 10 -20% polyacrylamide gradient gels. Separated peptides were electroblotted to PVDF using standard conditions. The membrane was then partially destained, air-dried and submitted to the University of Illinois Molecular Genetics Facility for sequence analyses.
N-terminal Amino Acid Sequence-The N-terminal amino acid sequence of AcpA was performed using automated Edman degradation and a model 470A Applied Biosystems gas phase Sequencer equipped with a 120A phenylthiohydantoin-amino acid analyzer by the Genetic Engineering Facility of the University of Illinois Biotechnology Center.
Oligonucleotide Synthesis and Gene Cloning-A nondegenerate oligonucleotide (5Ј-ACI GAT GTI AAT AAT III AAA CCI AAT GAT TAT GG-3Ј) was prepared (Applied Biosystems 319 DNA Synthesizer) by reverse translation of the N-terminal peptide sequence. The codon usage in the valAB locus of F. novicida (21) was used as a guide in designing the oligonucleotide. The oligonucleotide was 3Ј-end-labeled (ECL 3Ј-oligolabeling system, Amersham) as per manufacturer's instructions and used to screen a F. tularensis ATCC 29684 genomic library of partial Sau3AI fragments cloned into the BamHI site of the phagemid vector pTZ18U (Bio-Rad). One of the hybridizing clones contained a 1.3-kilobase DNA insert. Partial sequencing of this insert revealed one open reading frame (ORF) with a deduced amino acid sequence identical with the N-terminal amino acid sequence and to the sequence of an internal CNBr-generated peptide. This insert was used as a probe to identify a 3.1-kilobase HindIII fragment of F. novicida DNA that was cloned into pUC18 (22). BLASTP and BLASTX (23) were used to search for amino acid sequence similarities among the data bases available on-line throughout the National Center for Biotechnology Information. Pairwise alignments were done using FASTA (24) and modified by inspection. A 1798-base pair region was sequenced on both strands using a commercial T7 DNA polymerase (Sequenase, U. S. Biochemical Corp.) or Taq DNA polymerase (TaqTrack, Promega) using both universal and custom-designed primers. The gene encoding AcpA was designated acpA and was assigned the GenBank accession number L39831.
We chose to sequence the F. novicida acpA gene to facilitate future genetic experiments which can most easily be done in F. novicida. Although 16 S RNA and DNA relatedness (25) studies clearly identify F. novicida as a F. tularensis strain (26), biohazard rules place strictures on the transfer of genes between F. novicida and F. tularensis.
Protein Determination-Protein concentrations were determined using bicinchoninic acid (Micro BCA Protein Assay Reagent, Pierce) as described previously (27). Human albumin/␥-globulin protein standard (Sigma) was used as a standard.

Detection of Acid Phosphatase Activity in F. tularensis-Acid
phosphatase specific activities varied markedly between species of Francisella and among strains of F. tularensis. F. tularensis strain 6223 displayed the highest specific activity. It was generally in excess of 18,000 units/mg (13,000 to 30,000) and represents, to our knowledge, the highest specific activity ever reported for a bacterial or protozoan acid phosphatase. In comparison to other Acp-enriched intracellular pathogens (Table I), F. tularensis strain 6223 Acp specific activity is greater than 10 times that of L. micdadei (28), more than 4 times that of Coxiella burnetii strain PRS Q177 strain (29), and about twice that of the protozoan parasite, L. donovani (3). Acp specific activity in strain NDBR 101 was 550 to 3089 units/mg, whereas strain 29684 Acp specific activity was only 100 units/mg. The Acp specific activity of F. novicida was approximately 1700 units/mg.
The rather wide variation in acid phosphatase specific activ-ity among members of the Francisella genus may correlate with the passage history of individual strains. During experiments aimed at optimizing expression of Acp, we observed a large decrease in Acp specific activity upon repeated passage of strain 6223 on CHA (data not shown). A loss of almost 90% (8 -10-fold reduction) of the starting Acp specific activity was seen following 9 passages. The reduction was most likely not due to the accumulation of reversible inhibitors since washing the cells in physiological saline followed by extraction of Acp failed to increase Acp specific activity, and mixing of extracts from passaged cultures with purified AcpA did not result in the inhibition of the activity of the purified enzyme. Furthermore, detection of AcpA by Western blot analysis indicated a marked reduction in anti-AcpA reactive material following 9 passages as compared to that found in initial cultures (data not shown). Therefore, single passage F. tularensis (6223) was selected as the source for enzyme purification. AcpA Purification-In initial attempts to solubilize the enzyme, we found at least 70% of the phosphatase activity could be extracted with 1 M NaCl alone; including sodium cholate in the extraction buffer resulted in complete solubilization of the enzyme. Furthermore, essentially no difference in total AcpA activity was observed in the extracted material compared to the activity exhibited by intact bacteria (Table II). All of the enzymatic activity detected in intact bacteria was solubilized by the cholate NaCl extraction buffer and remained in the supernatant following extensive dialysis and two centrifugations (200,000 ϫ g, 1.5 h). The soluble AcpA was completely retained during loading at pH 6 on cation exchange resins S-Sepharose and Mono S and eluted as a single peak of activity between 0.17  M and 0.26 M NaCl (Fig. 1A). AcpA eluted in the breakthrough volume, however, during attempted anion exchange chromatography on either Q-Sepharose or Mono Q at pH 7.3. The material recovered from cation exchange chromatography was enriched 32-fold in acid phosphatase activity and contained 94% of the starting activity. Gel filtration chromatography through Sephadex G-100 superfine (Fig. 1B) resulted in an additional 13-fold increase in specific activity with 83% recovery of the applied activity. Final purification of the en-zyme was achieved by gel filtration FPLC (Fig. 1C). This step resulted in a further 1.7-fold increase in specific activity with 26% of the sample recovered in a single protein peak coincident with AcpA activity. The apparently low recovery from the FPLC column is explained by the conservative pooling of AcpA active fractions as described under "Experimental Procedures." The actual recovery was approximately 75%, but only the two fractions containing the highest AcpA activity were pooled for further analyses. Overall, AcpA was purified 713-fold over that in intact bacteria (Table II). The purification behavior of AcpA from strains 6223, NDBR 101, and 29684 and the results of comparative molecular weight ( Fig. 2A) and immunoreactivity with rabbit anti-Ft (6223) AcpA IgG (Fig. 2B) suggested the enzyme is very similar in all strains of F. tularensis. Also, the enzyme activity chromatographed as a single entity throughout FIG. 1. Purification steps of F. tularensis acid phosphatase. For A-C, AcpA activity (q) and protein concentration (E). A, S-Sepharose cation exchange chromatography of Supernatant II containing AcpA using a 0 to 0.5 M NaCl linear gradient (O) as described under "Experimental Procedures." Twenty-one 6.0-ml fractions (38 -58) found to contain AcpA activity eluted between 0.17 and 0.26 M NaCl. B, Sephadex G-100 Superfine chromatography of pooled and concentrated AcpA from S-Sepharose (5.3 ml, 6.7 mg/ml protein). Application and elution of AcpA to this gel filtration resin was performed as described under "Experi- all purification steps suggesting that multiple acid phosphatases may not exist in F. tularensis in contrast to the results reported for some other facultative intracellular organisms (1,2).
AcpA Purity and Molecular Weight-The purity of AcpA was assessed in several experiments. 1) SDS-PAGE of samples obtained throughout the purification procedure demonstrate the presence of an ϳ57-kDa protein which was continuously enriched as the purification proceeded and electrophoresed as a single Coomassie Blue or silver (data not shown)-stained band following recovery from the final FPLC gel filtration step (Fig.  1D). 2) In an effort to visualize minor protein contaminants or those which may be refractory to staining, the purified AcpA fraction (Superdex 75 fraction) was radioiodinated, subjected to SDS-PAGE, and the 125 I-labeled proteins were visualized by autoradiography. A single major band was seen on autoradiographs as increasing amounts of the radiolabeled AcpA fraction were applied to the SDS-PAGE gel (Fig. 3). This band, comprising 98% of the total signal as measured by quantitative densitometry, had a molecular weight of approximately 57,000. 3) N-terminal amino acid sequence analysis through the first 20 amino acids revealed the presence of a single threonine residue at the N terminus of the sequence (TDVNNSKPNDYGTLVKIEQK).
The molecular mass of the purified enzyme was determined by gel filtration chromatography, SDS-PAGE, and matrix-assisted laser desorption time of flight MS. Superdex 75 FPLC gel filtration chromatography gave a partition coefficient for AcpA of 0.09 (Fig. 4A). This value was compared to the regression line generated from the four molecular weight standards, and the K av corresponded to an apparent molecular weight of 56,000. A similar value, 57,000, was obtained with SDS-PAGE (Fig. 4B) using both reducing and nonreducing conditions (data not shown). Finally, mass spectrometry of AcpA indicated a singly charged species at 55,759 atomic mass units with a mass accuracy of 0.1% (Fig. 4C).
AcpA pH Optimum and Isoelectric Point-The purified AcpA behaved as an acid phosphatase (acid pH optimum) in all buffers tested (Fig. 5). Although the activity was slightly less in MES and HEPES than in acetate buffer, the optimal pH was 6.0 and activity was markedly diminished at 2 pH units to either side of this optimum. The pH optimum was independent of phosphomonoester substrates assayed including adenosine monophosphate (Fig. 5A), glucose 6-phosphate (Fig. 5B), tyrosine phosphate (Fig. 5D), and phosphorylated tyrosine residue   of p60 src (Fig. 7C, inset). When the purified AcpA was subjected to isoelectric focusing (pH 3-12), a single peak of activity was found at pH 9.2 (Fig. 6). The basic pI of this enzyme is consistent with its fractionation behavior during ion exchange chromatography (Fig. 1A). Substrate Specificity-AcpA has a broad in vitro substrate specificity (Table III). Sixteen of the 26 phosphomonoesters tested were hydrolyzed at greater than 50% the rate of MUP. The most active physiological substrates included tyrosine phosphate, AMP, ATP, and mannose 6-phosphate. Of the inositol phosphates tested, the monophosphates were preferred substrates. Inositol 1-monophosphate was hydrolyzed at near the same rate as MUP while inositol 4-phosphate was hydrolyzed at only 28% the rate of MUP. Inositol 1,4,5-trisphosphate (IP 3 ) was also recognized as a substrate, although it was hydrolyzed at only 15% the rate of MUP. Inositol cyclic phosphate was the most slowly hydrolyzed substrate and may be a conse-quence of its cyclic nature. In contrast to some of the inositol phosphates which were good substrates for AcpA, phosphatidylinositol phosphate derivatives, PIP and PIP 2 , were poor substrates. In general, these studies showed that small phosphomonoesters were more easily hydrolyzed than larger or multiphosphorylated compounds. For example, yeast mannan, phosvitin, and phytic acid were not recognized as substrates by AcpA. The acidic pH optimum of AcpA, and, more importantly, its inability to hydrolyze the thiophosphate substrate, cysteamine phosphate, which is an alkaline phosphatase-specific substrate, is consistent with the designation of AcpA as an acid phosphatase.
Determination of Kinetic Parameters and Peptide-tyrosine Phosphatase (PTPase) Activity of AcpA-The K m of AcpA for MUP and tyrosine phosphate was estimated to be 0.25 mM and 0.27 mM, respectively (Fig. 7) at pH 6.0. In addition to tyrosine phosphatase activity, AcpA displayed readily measurable PTPase activity. The K m of the monophosphorylated peptide p60 src (determined by the release of inorganic phosphate) was 0.34 mM. The V max values were 9.6 ϫ 10 6 , 8.0 ϫ 10 6 , and 6.7 ϫ 10 6 nmol of P i released per h per mg of enzyme for MUP, tyrosine phosphate, and p60 src , respectively (Fig. 7, B and D).
Effect of Inhibitors-To further characterize and classify this new AcpA, we measured the effects of acid phosphatase inhibitors. As shown in Table IV, the enzyme is not inhibited by L-(ϩ)-sodium tartrate, sodium fluoride, okadaic acid, divalent cation chelators (EDTA, EGTA), or detergents (CHAPS, Triton X-100, Triton X-114). However, the enzyme was sensitive to the early transition metal oxyanions such as molybdate and vanadate. As is true of the acid phosphatases described for other intracellular pathogens (2,28), this enzyme was sensitive to the heteropolymolybdate complex E 2 . Monofunctional sulfhydryl group reagents such as mercury and silver inhibited the enzyme by 50% at 0.5 M and 290 M, respectively. Hydroxymercuriphenylsulfonate, a potent inhibitor of bovine liver acid phosphatase (30) inhibited AcpA by 50% at a concentration of 50 M. Zinc was also found to be an inhibitor of the enzyme; 50 M ZnCl 2 inhibited AcpA activity by 50%. Inorganic phosphate was found to be a competitive inhibitor with a K i of approximately 50 M (data not shown). Glycerol, serine, and threonine had no inhibitory effect.
AcpA-mediated Inhibition of the Respiratory Burst in Neutrophils-In preliminary experiments, we found that a high speed supernatant from a crude F. tularensis extract containing an intense, heat-labile acid phosphatase activity was capable of dose-dependent inhibition of fMLP-activated porcine neutrophils. When this supernatant was subjected to gel filtration chromatography, the AcpA and respiratory burst inhibitory activities eluted coincidentally. To determine if AcpA was responsible for burst inhibition, porcine neutrophils were treated with the purified enzyme prior to activation with either fMLP or PMA. Under these conditions, AcpA caused a dose-dependent inhibition of the respiratory burst when added to either fMLP-or PMA-activated porcine neutrophils (Figs. 8, A  and B). The inhibition was also seen when AcpA was added following PMA or fMLP addition but required larger amounts of enzyme, monitoring superoxide formation for longer times, and was seen only after a lag of 2-3 min following addition of AcpA except when the highest amounts of AcpA were used (data not shown). A greater inhibitory effect was obtained by preincubation of the neutrophils with AcpA prior to activation (Fig. 8C). Maximum burst inhibition was seen following preincubation for 15 min at 37°C. Heat-inactivated AcpA had no effect on the rate of superoxide formation in activated neutrophils (Fig. 8A). We also did not detect catalase or superoxide dismutase activities in AcpA (data not shown), and AcpA had FIG. 8. AcpA-mediated inhibition of the respiratory burst in porcine neutrophils. Isolated porcine neutrophils (1 ϫ 10 7 cells/ml) were incubated with purified AcpA prior to addition of either PMA or fMLP. Superoxide anion production was determined by continuous spectrophotometric measurements of the reduction of ferricytochrome c at 550 nm. A, each point (ⅷ) represents the mean of the rate of cytochrome c reduction Ϯ S.D. from 5 separate experiments by porcine neutrophils after a 15-min preincubation with AcpA and activated with PMA; f, heat-inactivated AcpA (100°C, 15 min). B, comparison of the amount of superoxide anion production as measured by reduction of ferricytochrome c by fMLP-stimulated porcine neutrophils after a 15min exposure to 1000 units of AcpA (OO) or without prior exposure to AcpA (---). C, effect of increasing preincubation times of porcine neutrophils with purified AcpA (8000 units) on production of superoxide anion in PMA-activated neutrophils. no inhibitory effect on the rate of xanthine oxidase-catalyzed generation of superoxide (data not shown). Thus, it is unlikely this enzyme affects the respiratory burst indirectly through electron scavenging.
Nucleotide Sequencing and Deduced Primary Structure of acpA-To further characterize the structure and function of AcpA, we cloned and sequenced the AcpA structural gene (acpA). A nondegenerate oligonucleotide was prepared and used to screen a F. tularensis ATCC 29684 and subsequently a F. novicida genomic library (see "Experimental Procedures"). The complete acpA nucleotide sequence and derived primary structure is shown in Fig. 9. The first 21-amino acid sequence of the open reading frame, prior to the N-terminal Thr residue of AcpA, contains many of the functional elements of a standard Gram-negative signal peptide (31). The next 20-amino acid deduced sequence is identical with the N-terminal sequence (TDVNNSKPNDYGTLVKIEQK) determined by Edman degradation of the purified AcpA. Furthermore, the deduced sequence (MYPNAKNPEGE) at position 422-454 was identical with the peptide sequence determined by Edman degradation of a CNBr fragment of AcpA. The molecular weight of the signal peptide cleaved AcpA predicted from the nucleotide sequence (55,593) is in close agreement with the molecular weight of AcpA (55,759) determined by mass spectrometry. These data strongly indicate the nucleotide sequence presented in Fig. 9 contains the complete open reading frame of the acpA gene.
Comparative sequence analyses (Blast, National Center for Biotechnology Information) indicate acpA has no overall sequence similarity to other known acid phosphatases, but it is partially similar to bacterial phosphatidylcholine phospholipases (PLC-N and PLC-H) identified in Pseudomonas aeruginosa (32,33). The amino acid sequence of PLC-N is 40% homologous to PLC-H (33). The majority of this homology lies within the amino two-thirds of the proteins' sequence while the remaining one-third shows very little homology. AcpA shows an overall sequence identity of 16% to either PLC-N or PLC-H. For comparison, the sequence alignment of AcpA to PLC-N is shown in Fig. 10. Considering both identical and conserved amino acid residues, AcpA shows an overall sequence similarity to PLC-N of 51%. In preliminary experiments, phospholipase C activity was detected in AcpA using the synthetic substrate p-nitrophenylphosphorylcholine assayed at pH 7.3 but not at pH 6.0, the pH optimum for phosphomonoesterase activity. The phospholipase C specific activity of AcpA (610 nmol of p-nitrophenylphosphorylcholine hydrolyzed/h/mg), although comparable to that of a commercial Clostridium phospholipase (1040 nmol/h/mg (Sigma, Type XIV)), was approximately 3-4 orders of magnitude lower than its phosphomonoesterase specific activity assayed at pH 7.3 (1.5 ϫ 10 6 nmol of MUP/h/mg) and pH 6.0 (9.5 ϫ 10 6 nmol/h/mg). There was no detectable phosphomonoesterase activity, using MUP as a substrate, in the Clostridium PLC preparation. to the AcpA N-terminal sequence is the deduced amino acid sequence identical with a CNBr peptide sequence prepared from AcpA.

DISCUSSION
Members of the genus Francisella are facultative intracellular pathogens and were found to harbor varying amounts of acid phosphatase activity in crude extracts. One strain in particular, ATCC 6223, produced the highest levels of acid phosphatase thus far reported for a protozoan or bacterial pathogen and was chosen for purification of the enzyme. The Acps from all Francisella strains were examined and found to be indistinguishable in purification, molecular weight, and reaction with rabbit anti-AcpA polyclonal antibody. These data suggest F. tularensis, in contrast to L. donovani and L. micdadei which contain multiple Acp types (1,34), produce a single Acp polypeptide that is similar, if not identical, in all members of the genus. F. tularensis (strain 6223) is remarkable in that it is highly enriched in a respiratory burst-inhibiting acid phosphatase. Using a specific activity for the purified enzyme of 1 ϫ 10 7 units/mg and a molecular mass of 56,000 Da, we estimate there are approximately 50,000 AcpA molecules produced per viable bacterial cell when cultured on hemoglobin-enriched CHA. This number was, however, dependent on the strain and passage history.
The physical and chemical properties of AcpA indicate this enzyme is unique not only among burst-inhibiting acid phosphatases but also among acid phosphatases in general. AcpA, in contrast to burst-inhibiting Acps (1,2,29), is easily released from the bacterial cell in soluble form, is a basic enzyme, and suppresses the respiratory burst of not only fMLP but also PMA-stimulated neutrophils. AcpA is also much more sensitive to inhibition by molybdate compounds than other burst-inhibiting Acps. As shown in Table IV, these compounds inhibit 50% of the activity of AcpA at concentrations that are 100 and 1000 times lower than the I 50 values for either Leishmania or Legionella acid phosphatases (1,2).
The recognized classes of acid phosphatases include high and low molecular weight acid phosphatases, some protein phosphatases specific for phosphoserine or phosphothreonine and purple acid phosphatases (35). The purple acid phosphatases are readily distinguished from other acid phosphatases by their purple color in solution, which is due to the presence of a binuclear iron center or iron-zinc center (36). AcpA is not purple in solution, and preliminary x-ray diffraction and proton accelerator studies of AcpA crystals did not indicate the presence of any metal cofactors. 2 Results from our inhibitor studies also suggest the enzyme is not a serine/threonine-specific protein phosphatase. This class of protein phosphatases, consisting of groups 1, 2A, 2B, and 2C, is either acutely sensitive to okadaic acid or has an absolute requirement for divalent cations (37,38). AcpA is resistant to okadaic acid and retains full activity in 20 mM EDTA.
AcpA also does not fit into either the high or low molecular weight class of acid phosphatases. High molecular weight acid phosphatases differ in several respects from their low molecular weight counterparts. A comparison of the class-distinctive properties of the high and low molecular weight Acps to those of AcpA is shown in Table V. According to its molecular weight, AcpA should be classified as a high molecular weight Acp. However, it has broad substrate specificity and is resistant to tartrate and fluoride, which are common inhibitors of high molecular weight acid phosphatases.
Although AcpA was shown to have PTPase activity but it did not possess an unambiguous phosphate binding loop signature sequence, (H/V)C(X) 5 R(S/T)(G/A/P), present in Yop51 and more than 40 other PTPases (39). We did find a possible phosphate binding loop (C(X 5 )KSG) in AcpA (Fig. 10, residues 237-245) in which the critical arginine residue found in all PTPs is replaced by a lysine, and this may explain why AcpA still retains PTPase activity. P-loop motifs found conserved in GTP-and ATPbinding proteins also have the general sequence G(X) 4 GK(T/S) in which a lysine residue is conserved in all cases (40). It is tempting to speculate that AcpA has a diverged cysteine active site, phosphate binding loop in which an arginine has been conservatively replaced by a lysine. The lack of a consensus PTPase P-loop, however, precludes its classification as a PTPase.
Inhibition of AcpA activity by monofunctional sulfhydryl inhibitors including mercuric ions, silver, and hydroxymercuriphenylsulfonate suggests this enzyme may possess a cysteine active site and may therefore be classified as a "low molecular weight" acid phosphatase despite its high molecular weight. This is not without precedent since a cysteine active site, low molecular weight TRAP that has high molecular mass (35 kDa), has been described (41).
Interestingly, comparative nucleotide sequence analyses revealed partial homology to known phosphatidylcholine phospholipases (PLC) of P. aeruginosa but failed to reveal homology to any known acid phosphatase and did not detect the presence of any known acid phosphatase, protein-tyrosine phosphatase, or phospholipase signature motifs. In preliminary experiments, we were able to detect phospholipase C activity in the purified AcpA when assayed using a synthetic substrate, p-nitrophenylphosphorylcholine, at pH 7.0 but not at pH 6.0, the pH optimum for phosphomonoesterase activity. The phospholipase C specific activity of AcpA, although comparable to that of a commercial Clostridium phospholipase, was 3000 times lower 2 E. Garman, personal communication. than its phosphomonoesterase specific activity. The markedly higher rate of hydrolysis of monophosphate esters at acidic and neutral pH compared to phosphodiester substrates, including the p-nitrophenylphosphorylcholine phospholipase C substrate, supports the designation of AcpA as an acid phosphatase in spite of its partial sequence similarity to P. aeruginosa PLC. Unequivocal demonstration of PLC activity of AcpA must await further studies using natural substrates.
The mechanism(s) by which any acid phosphatase suppresses the respiratory burst has also not been determined. A proposed mechanism for Leishmania and Legionella Acp mediated inhibition of the fMLP-stimulated respiratory burst is Acp-catalyzed depletion of PIP 2 and IP 3 (4). In this mechanism, it is not clear, however, whether depletion of PIP 2 and IP 3 pools occurs by direct hydrolysis of these intermediates or whether Acp is somehow interfering with plasma membrane signal transduction mechanisms. It has yet to be shown that any burst-inhibiting Acp gains entry or accessibility to PIP 2 or IP 3 pools within the neutrophil or macrophage. In the case for AcpA, it seems unlikely that depletion of PIP 2 and IP 3 pools accounts for all the observed inhibition since PIP 2 and IP 3 are relatively poor substrates for AcpA, and AcpA also inhibits PMA-stimulated porcine neutrophils which is an PIP 2 /IP 3 independent superoxide anion production pathway (42). Furthermore, it is unlikely that AcpA gains access to the neutrophil cytoplasm. In preliminary experiments using radioiodinated, catalytically active AcpA, we found no evidence for uptake of exogenously added AcpA into neutrophils over a 2-h time period even though burst inhibition occurred within the first 15 min. Thus, it seems more likely that AcpA inhibits the respiratory burst by hydrolysis of neutrophil surface-exposed substrates that are involved in signal transduction pathways necessary for burst activation or maintenance.
The broad substrate specificity of AcpA including its tyrosine phosphatase (PTPase) and phospholipase C activities may provide clues to possible mechanisms of respiratory burst inhibition. Dephosphorylation of multiple targets including phosphatidylcholine, protein tyrosine phosphates, secondary messengers, or other low molecular weight substrates critical to phagocyte activation such as ribose 5-phosphate, NADPH, or ATP may explain why this particular acid phosphatase inhibits the respiratory burst of both fMLP or PMA-stimulated neutrophils.
Whether AcpA's burst-inhibiting activity is relevant to the pathogenicity of F. tularensis or secondary to even more important microbial physiologic processes remains to be determined. There is no unequivocal proof that any of the burst-inhibiting Acps function as virulence factors in vivo. In our opinion, identification of these enzymes as virulence factors must await construction and use of isogenic Acp-negative mutant strains in both in vitro and in vivo infectivity experiments. Until now, there has been no nucleotide sequence information reported for any burst-inhibiting Acp. The results of cloning and sequencing of the AcpA gene reported here should help in the design of experiments aimed at elucidating the physiological function of AcpA and to directly test its role, if any, in F. tularensis virulence.