Purification and Molecular Characterization of cGMP-dependent Protein Kinase from Apicomplexan Parasites

The trisubstituted pyrrole 4-[2-(4-fluorophenyl)-5-(1methylpiperidine-4-yl)-1H-pyrrol-3-yl]pyridine (Compound 1) inhibits the growth of Eimeria spp. both in vitro and in vivo. The molecular target of Compound 1 was identified as cGMP-dependent protein kinase (PKG) using a tritiated analogue to purify a 120-kDa protein from lysates of Eimeria tenella. This represents the first example of a protozoal PKG. Cloning of PKG from several Apicomplexan parasites has identified a parasite signature sequence of nearly 300 amino acids that is not found in mammalian or Drosophila PKG and which contains an additional, third cGMP-binding site. Nucleotide cofactor regulation of parasite PKG is remarkably different from mammalian enzymes. The activity of both native and recombinant E. tenella PKG is stimulated 1000-fold by cGMP, with significant cooperativity. Two isoforms of the parasite enzyme are expressed from a single copy gene. NH2-terminal sequence of the soluble isoform of PKG is consistent with alternative translation initiation within the open reading frame of the enzyme. A larger, membrane-associated isoform corresponds to the deduced fulllength protein sequence. Compound 1 is a potent inhibitor of both soluble and membrane-associated isoforms of native PKG, as well as recombinant enzyme, with an IC50 of <1 nM.


The trisubstituted pyrrole 4-[2-(4-fluorophenyl)-5-(1methylpiperidine-4-yl)-1H-pyrrol-3-yl]pyridine (Compound 1) inhibits the growth of Eimeria spp. both in vitro
and in vivo. The molecular target of Compound 1 was identified as cGMP-dependent protein kinase (PKG) using a tritiated analogue to purify a ϳ120-kDa protein from lysates of Eimeria tenella. This represents the first example of a protozoal PKG. Cloning of PKG from several Apicomplexan parasites has identified a parasite signature sequence of nearly 300 amino acids that is not found in mammalian or Drosophila PKG and which contains an additional, third cGMP-binding site. Nucleotide cofactor regulation of parasite PKG is remarkably different from mammalian enzymes. The activity of both native and recombinant E. tenella PKG is stimulated 1000-fold by cGMP, with significant cooperativity. Two isoforms of the parasite enzyme are expressed from a single copy gene. NH 2 -terminal sequence of the soluble isoform of PKG is consistent with alternative translation initiation within the open reading frame of the enzyme. A larger, membrane-associated isoform corresponds to the deduced fulllength protein sequence. Compound 1 is a potent inhibitor of both soluble and membrane-associated isoforms of native PKG, as well as recombinant enzyme, with an IC 50 of <1 nM.
Protozoan parasites of the genus Eimeria are the causative agents of the intestinal disease known as coccidiosis. Coccidiosis occurs in several domesticated and wild animal species, but of major economic importance is the impact that Eimeria spp. have on the poultry industry. During acute infections, these parasites cause significant morbidity and mortality in broiler breeds of chicken (reviewed in Ref. 1). Anticoccidial compounds have been and continue to be used prophylactically in the majority of poultry operations today. The most successful anticoccidials have been the polyether ionophores, a family of compounds that continues to be the industry standard since their introduction nearly 30 years ago (2). Not surprisingly, reports of resistance development due to the extended and constant chemotherapeutic pressure exerted by this class of compounds are not uncommon (3,4). Since that time no novel anticoccidials with efficacy and economic features that approach the ionophore class have been introduced into the poultry industry. The need to identify and develop new drugs for the control of coccidiosis is critically important. In this report we describe the chemotherapeutic efficacy of a novel anticoccidial reagent. Data from biochemical purification and molecular cloning efforts predict that the therapeutic target of this class of compounds in Eimeria is a cGMP-dependent protein kinase (PKG). 1 PKG transfers the ␥-phosphate of ATP in a cGMP-dependent reaction to serine and/or threonine residues of several cellular proteins (5). Cyclic GMP is a ubiquitous intracellular messenger that has a role in several aspects of signal transduction that potentially regulate a myriad of physiological processes (reviewed in Ref. 6). cGMP also modifies the activity of proteins other than PKG, including cGMP-gated ion channels and cGMP-regulated phosphodiesterases (6).
Cyclic nucleotide-dependent protein kinases from unicellular organisms such as Paramecium to humans have been biochemically characterized and/or cloned (6 -9). Members of this group of kinases share sequence homology in both their regulatory and catalytic domains. The most striking feature that distinguishes cAMP-dependent (PKA) from cGMP-dependent protein kinases is that PKA exists as a heterotetramer in its inactive conformation, composed of two identical regulatory and two identical catalytic subunits, while PKG is in most cases a homodimeric enzyme (6,10). The regulatory and catalytic subunits of PKA are distinct gene products. Activation of PKA by cAMP occurs as a result of a conformational change in the enzyme initiated by the binding of two molecules of the cyclic nucleotide to each regulatory subunit. The conformational change releases the regulatory dimer from the inhibited complex, thereby activating the catalytic dimer (11,12). Unlike PKA, the nucleotide-binding/regulatory and catalytic domains of PKG are part of the same protein. In the absence of cGMP, PKG assumes a conformation that is autoinhibited (13). Binding of cGMP to two nucleotide-binding domains within each subunit of the homodimer causes an intramolecular conformational change that activates kinase activity. While the two families of enzymes have significant structural differences, the mechanisms of autoinhibition and activation by the respective cyclic nucleotide cofactors are similar.
In this report we demonstrate that Apicomplexan parasite PKG has several features that distinguish it from PKG homologues in other organisms. Unlike mammalian and invertebrate PKG enzymes which are typically homodimers, the E. tenella enzyme is a monomeric protein. The E. tenella enzyme has remarkably different cGMP activation kinetics, characterized by a minimal basal kinase activity in the absence of nucleotide cofactor which is induced by as much as 1000-fold upon addition of cGMP. Moreover, the activation profile shows strong cooperativity, a feature that is not as striking in mammalian PKG. The size of the parasite enzyme, demonstrated biochemically and confirmed by cDNA cloning of five Apicomplexan PKGs, is 30 -40% larger than most PKG enzymes that have been described. Consistent with the architecture of other PKGs, the nucleotide-binding/regulatory domain of the parasite enzymes is located toward the amino-terminal end of the protein and the catalytic domain is positioned C-terminal. A conserved parasite signature sequence of nearly 300 amino acids resides between the regulatory and catalytic domains of the protein. Within this region we have identified a third nucleotide-binding site, yet another feature that is unique to, but shared among, the collection of Apicomplexan parasite enzymes.

EXPERIMENTAL PROCEDURES
Parasite Preparation and in Vivo Studies-Chickens were infected orally with 7.5 ϫ 10 4 Eimeria tenella LS18 sporulated oocysts. The unsporulated oocysts were harvested from the ceca 7 days post-infection and purified according to the method of Schmatz et al. (14), and then sporulated by continual agitation for 36 h at 29°C. For the in vivo studies drug was administered in feed at 50 ppm on day 1. Birds were infected with 3.5 ϫ 10 3 E. tenella sporulated oocysts on day 2 and 3.6 ϫ 10 3 Eimeria acervulina sporulated oocysts on day 3. Medicated feed was continuously available for the duration of infection. The experiment was terminated on day 8 and efficacy was estimated by performing oocyst counts.
Determination of in Vitro Antiprotozoal Activity-Conditions for the in vitro culture of parasites and determination of IC 50 and or minimal inhibitory concentrations (defined as the lowest concentration (ng/ml) at which parasite growth was fully inhibited) for compounds were conducted according to previously described methods for E. tenella (14), Toxoplasma gondii (15), and Besnoitia jellisoni (16). The Neospora caninum cell based assay (17) was adapted for use by increasing length of assay from 5 days to 7 days. 2 The N. caninum and T. gondii strains expressing ␤-galactosidase constructs were obtained from David Sibley (Washington University) and John Boothroyd (Stanford University), respectively.
Ligand Binding Assay-Soluble extracts for binding studies were prepared by vortexing 2 ϫ 10 9 E. tenella unsporulated oocysts with an equal volume of buffer (10 mM HEPES pH 7.4, 1 mM sodium orthovanadate, 20% glycerol, 0.1 mg/ml Bacitracin, and 0.5% Sigma protease inhibitor mixture P8340), and an equal volume of 4-mm glass beads for 20 min. The resulting homogenate was centrifuged (100,000 ϫ g, 1 h) and the supernatant (S100) used directly. E. tenella protein (10 -25 l) was assayed in 100 l at a final concentration of 75 mM Tris, pH 7.5, 12.5 mM MgCl 2 , 1.5 mM EDTA, and 2 nM [ 3 H]Compound 1 (60 Ci/mmol). Nonspecific counts were estimated using a 1000-fold molar excess of unlabeled Compound 1. Samples were incubated for 1 h at 25°C, and either filtered through Whatman GF/B glass fiber filters (presoaked in 0.6% polyethyleneimine for 1 h at 25°C), or filtered through prepackaged gel filtration columns (800 l, Edge Biosystems). The filters were washed with 100 mM NaCl, 10 mM Tris, pH 7.4, dried, and radioactivity determined by scintillation counting using Ready-SAFE scintillation mixture. The void volume was collected from the gel filtration columns according to the manufacturers instructions and mixed with Ready-SAFE prior to counting.
Purification of E. tenella PKG-Oocyst lysates and S100 fractions were prepared as described above. The S100 was dialyzed against 30 mM sodium phosphate, pH 7, 20% glycerol, 1 mM dithiothreitol, 1 mM EDTA, 10 mM sodium fluoride, and 0.1 mM sodium orthovanadate (Buffer A). The dialysate (ϳ1 g of protein) was applied to a HiLoad 26/10 Q-Sepharose column (Amersham Bioscience) and eluted with a salt gradient (0 -1 M NaCl) in Buffer A. Aliquots were taken from fractions for binding assays. Fractions that bound to the column and contained Compound 1 binding activity were dialyzed against Buffer B (30 mM sodium phosphate, pH 7.0, 1 mM dithiothreitol, 10 mM NaF, 1 mM EDTA, 0.1 mM sodium orthovanadate) that is supplemented with 1 M ammonium sulfate. The dialysate was applied to a butyl-Sepharose column (Amersham Bioscience, two tandem 5-ml columns, pre-treated with bovine serum albumin) and the column was eluted with a reverse salt gradient (from 1 to 0 M (NH 4 ) 2 SO 4 in Buffer B). Fractions were analyzed as above using the Compound 1 binding assay and active fractions dialyzed against Buffer C (10 mM sodium phosphate, pH 7.0, 1 mM dithiothreitol, 10 mM NaF, 0.1 mM sodium orthovanadate). The dialysate was loaded onto two tandem 5-ml hydroxyapatite columns (Bio-Rad) and protein was eluted with a linear salt gradient (in Buffer C) up to 0.4 M sodium phosphate. Compound 1 binding fractions from the hydroxyapatite column were pooled, dialyzed against Buffer A, and applied to an anion-exchange MonoQ HR 5/5 column (Amersham Bioscience) which had been equilibrated in Buffer A with 100 mM NaCl. The column was washed with 100 mM NaCl in Buffer A and then eluted with a linear gradient to 1 M NaCl which was started once the unbound proteins had been eluted. The Compound 1 binding activity, which did not bind to the column in 100 mM NaCl, was dialyzed against Buffer A and re-applied to the same column. After unbound proteins had been eluted in the absence of NaCl and the absorbance had returned to baseline, column bound proteins were eluted with a NaCl gradient to 1 M (Fig. 2B) in Buffer A. Fractions were collected and aliquots were assayed for binding activity and also for purity on polyacrylamide gels. In some cases fractions with ligand binding activity were then applied to a Superdex 200 column in Buffer A and analyzed as described above. SDS-polyacrylamide gel electrophoresis was performed on gradient gels (Novex). Western blots were carried out after transfer to nitrocellulose and immunoreactive proteins were detected using the ECL procedure (Amersham Bioscience). Protein sequence analysis of the purified proteins was performed on Coomassie Blue-stained gel slices. The slices were digested with trypsin, peptides separated by high performance liquid chromatography on C18 reverse phase chromatography, and sequence performed on isolated peptides using Edman degradation.
DNA Manipulations and Cloning of Parasite PKG cDNAs-DNA manipulations were performed according to standard procedures (18). Plasmid DNA was purified either using Promega Wizard Miniprep kits (Promega) or the Qiagen Maxi kit (Qiagen). DNA sequence was obtained using an ABI Prism sequencer and fluorescent sequencing reagents (PerkinElmer Life Science) and analyzed using Vector NTI TM Suite software (Informax). Oligonucleotides were purchased from Invitrogen or Integrated DNA Technologies. Enzymes used for the polymerase chain reaction (PCR) were purchased from PerkinElmer Life Science. PCR products were gel purified using Qiaex II (Qiagen) and subcloned directly into the TA-cloning vector pGEM-T Easy (Promega). Gel purified DNA fragments to be used as hybridization probes were random-primer labeled with [␣-32 P]dCTP (3000 Ci/mmol, Amersham Bioscience). Restriction and modifying enzymes were purchased either from Invitrogen or New England Biolabs. Electroporation competent bacterial cells were from Invitrogen.
Seven tryptic peptide sequences generated from the purified E. tenella ligand-binding protein were used to design several degenerate oligonucleotides. These were used in coupled reverse transcriptase-PCR reactions with mRNA from E. tenella sporozoites as template. PCR products of interest were cloned and sequenced. A partial cDNA clone generated by RT-PCR was used as a hybridization probe to isolate full-length cDNA clones from an E. tenella unsporulated oocyst cDNA library (19). Clones coding for T. gondii PKG were isolated from a cDNA library obtained from the NIH AIDS Research and Reference Reagent Program (Bethesda, MD, catalog number 1896). Clones were identified by heterologous screening at a reduced hybridization stringency using a 2 B. Nare, unpublished observations. portion of the E. tenella cDNA as probe. A Cryptosporidium parvum expressed sequence tag containing a fragment of a PKG homologue was identified in a preliminary survey sequence analysis of the Cryptosporidium parvum genome (expressed sequence tag AQ083827 (20)). To obtain the complete C. parvum PKG gene, a PCR fragment derived from this expressed sequence tag was used to probe a C. parvum genomic library (number 1644, NIH AIDS Research and Reference Reagent Program). A PKG open reading frame spanning two overlapping EcoRI genomic clones was subsequently identified by DNA sequence analysis. An identical un-annotated open reading frame (Contig number 1655) was also located in a BLAST search of the partial C. parvum genome survey data base, made accessible through NCBI by the University of Minnesota C. parvum sequencing project (www.ncbi.nlm.nih.gov/cgibin/Entrez/genom_table_cgi (21)). An alternative strategy was employed to isolate PKG cDNA clones from both Eimeria maxima and Plasmodium falciparum parasites. The first step involved a coupled RT-PCR to generate partial PKG cDNAs for each parasite. Using an alignment of the deduced amino acid sequence from cDNA clones coding for E. tenella and T. gondii PKG proteins, areas of sequence identity within the presumptive cGMP-binding (peptides a, VKFFEML; and b, GEYFGERAL) and catalytic domains (peptides c, RDLKPENI; and d, HYMAPEV) were identified. Total RNA purified from E. maxima sporulated oocysts and a trophozoite-enriched preparation of P. falciparum was first converted into cDNA using reverse transcriptase. The respective populations of cDNA products were then used as template in PCR reactions using degenerate oligonucleotide primers from peptides a and d. Primary PCR reaction products were then used as templates in secondary nested PCR reactions using degenerate primers from peptides b and/or c. PCR reaction products produced in the secondary nested reactions were subcloned, DNA sequence confirmed, and then used as hybridization probes to screen E. maxima and P. falciparum cDNA libraries to isolate full-length cDNA clones.
Parasite cDNAs encoding PKG open reading frames were modified prior to subcloning by appending an NH 2 -terminal or COOH-terminal FLAG epitope (Kodak) via PCR amplification (Pfu polymerase, Stratagene). DNA fragments encoding FLAG epitope-tagged PKGs were placed in an expression vector under the control of a Toxoplasma ␣-tubulin promoter (23). Following electroporation, stable transgenic Toxoplasma lines were selected with chloramphenicol (24,25) and clones expressing recombinant PKG were identified by immunofluorescence analysis with FLAG M2 antisera (Sigma). Recombinant FLAGtagged Eimeria PKG was purified from parasite lysates by FLAG immunoaffinity chromatography as described for the recombinant Toxoplasma PKG. 3 Production of Polyclonal Antisera-Antisera were raised in rabbits to the following peptides based on amino acid sequence of the EtPKG full-length clone: EDTQAEDARLLGHLEKREKT (TR3) and EED-EGIELEDEYEWDKDF (TR6). Both peptides were conjugated to Keyhole Limpet hemocyanin prior to immunization. Antibody production was performed at Covance Research Laboratories, Denver, PA.
PKG Catalytic Assay-Kinase activity was detected using a peptide substrate and [␥-33 P]ATP. An aliquot containing enzyme (1 l) was mixed with a reaction mixture (10 l) whose composition is as follows: 25 mM HEPES pH 7.4, 10 mM MgCl 2 , 20 mM ␤-glycerophosphate, 5 mM ␤-mercaptoethanol, 10 M cGMP, 1 mg/ml bovine serum albumin, 400 M Kemptide, or 9 M myelin basic protein, 2 M [␥-33 P]ATP (0.1 mCi/ml). The reaction was allowed to proceed for 1 h at room temperature and then terminated with the addition of phosphoric acid to a final concentration of 25 mM. Labeled peptide was captured on P81 filters or on Millipore 96-well plates (MAPH-NOB). In both cases filters were washed with 75 mM phosphoric acid, dried, and 33 P-labeled phosphopeptide was detected by liquid scintillation counting.
Kinetic Characterization of E. tenella PKG-AMP-PCP and guanethidine (1-(2-guanidinoethyl)octahydroazocine) were obtained from Sigma. Kinase assays were performed in 50-l reaction volumes containing 25 mM HEPES pH 7.0, 10 mM MgCl 2 , 20 mM ␤-glycerophosphate, 1 mM dithiothreitol, 0.1 mg/ml bovine serum albumin, 10 M cGMP, between 2 and 12 Ci of [␥-33 P]ATP, and varying levels of ATP, peptide substrate (biotinyl-e-aminocaproyl-GRTGRRNSI-OH), and inhibitor. For the two substrate pattern, ATP and peptide were each varied at four concentrations between 0.5 and 3 times their respective K m values. For inhibition patterns, one substrate concentration was fixed at 10 M while the other was varied at four concentrations between 0.5 and 3 times its K m ; the inhibitor concentration was varied between 0 and 3 times its K i . Reactions were initiated with 5 l of enzyme (or buffer for the background) and incubated for 30 min in a heating block at 30°C. The assays were terminated by the addition of 25 l of 8 M guanidine-HCl solution (Pierce) before spotting 15 l onto a SAM 2® streptavidin membrane (Promega). The membrane was washed twice with 1 M NaCl and twice with 1 M NaCl, 1% H 3 PO 4 on a rotating mixer for 20 min. The membrane was then rinsed successively with water and ethanol and dried under a heat lamp. The individual assays were then separated, placed in scintillation vials containing 2 ml of Ultima Gold mixture (Packard), and counted in a Packard TriCarb 2500 liquid scintillation counter. After subtracting the appropriate background for each assay point, the data was fit to the appropriate equation using GraFit (Erithacus Software) ) for noncompetitive inhibition. K m and K i values are reported with their standard errors derived from the fit. Activation experiments were also performed as described above except that cGMP concentration was varied and ATP and peptide concentrations were fixed at 20 M. Data was fit to the following modified Hill equation using Kaleida- Values for K 50 , the concentration for half-maximal activation by the allosteric activator cGMP, and h, the Hill coefficient, are reported with their standard errors derived from the fit.
In-gel Kinase Assay-In-gel kinase assays were performed using the In-gel Protein Kinase Assay Kit (Stratagene, number 206020) according to the manufacturer's recommendation, except that myelin basic protein was used as the substrate and cGMP was added to 20 M in the kinase assay buffer along with [␥-32 P]ATP.
cGMP-agarose Affinity Chromatography-Chromatography on cGMP-agarose was performed according to the manufacturers instructions (Biolog, A019). Briefly the 0.6-ml column was equilibrated with Buffer G (50 mM HEPES pH 7.4, 10% glycerol, 10 mM sodium fluoride, 0.1 mM sodium orthovanadate, 1 mM EDTA). The sample (crude S100 or purified protein) was mixed with an equal volume of Buffer G and applied to the column, which was then washed with 10 ml of the same buffer. The column was then washed with 10 ml of Buffer G containing 1 mM GMP. Proteins were eluted with 10 ml of Buffer G containing 15 mM cGMP.
Triton X-114 Phase Separation-Phase separation was performed according to the method of Bordier (27). Fig. 1A), prevents the in vitro development of several intracellular Apicomplexan parasites including E. tenella, T. gondii, N. caninum, and B. jellisoni (Fig. 1B). Moreover, Compound 1 is orally active against E. tenella and E. acervulina in parasite-infected chickens at a dose of 50 ppm in the feed (Fig. 1C) and is also efficacious in a mouse model of Toxoplasmosis (28).

Antiparasitic Activity of Compound 1-The
Biochemical Purification of the Compound 1-binding Protein from E. tenella-To identify the molecular target of this compound, a binding assay was developed using a tritiated version of Compound 1 as ligand. Saturable binding to an S100 extract of E. tenella can be competed with unlabeled ligand with an IC 50 of 10 nM (data not shown). The ligand-binding protein was purified from the E. tenella S100 fraction by conventional chromatography using a four-column protocol that is summarized in Fig. 2A. There is only a single peak of binding activity detected in three of the four column profiles. The one exception to this is a small amount of activity detected in the flow through from the first column in the series, HiLoad Q-Sepharose. Fig. 2B illustrates the profile of binding activity following the final chromatographic step on MonoQ anion exchange. Gel electrophoresis of proteins in fractions that correspond to the peak of binding activity from MonoQ are visualized by silver staining in Fig. 2C. Two proteins with apparent molecular weights of 120,000 and 150,000 are coincident with the peak of binding activity. Subsequent gel filtration chromatography of a pool of fractions 34 and 35 was able to enrich for the 120-and 150-kDa pair, but was not able to resolve the two proteins (data not shown).
Compound 1-binding Protein Is a cGMP-dependent Protein Kinase-The structural similarity between Compound 1 and protein kinase inhibitors that have been described in the literature (29,30) prompted us to assay for and detect kinase activity (using the peptide substrate Kemptide) in the ligand binding fractions during the purification protocol. However, protein kinase activity was marginal (data not shown), even in the final fractions from the MonoQ column pictured in Fig. 2C.
Following electrophoretic resolution of the column purified ligand binding activity, both the 120-and 150-kDa proteins were submitted for sequence analysis. Seven tryptic peptide sequences were generated from the 120-kDa protein (Table I), but no sequence information was retrieved from the larger protein. Each of the peptides was used as a query to search in silico translations of nucleotide data bases as well as protein data bases, but no significant matches were found. Several degenerate oligonucleotides were designed based on the first four peptides and used in various combinations in coupled RT-PCR reactions with E. tenella sporozoite mRNA as template. One primer set consisting of oligonucleotides from peptides 3 and 4 generated a PCR product that was successfully nested in a secondary reaction with a second oligonucleotide from peptide 3 along with the original peptide 4 oligonucleotide. The primary PCR product from this series was cloned and its 452-nucleotide insert has an open reading frame that is 31% identical to both Drosophila melanogaster PKG (PID:g140293) and the regulatory subunit of Saccharomyces cerevisiae PKA (PID:g172689).
Based on the prediction that the 120-kDa ligand-binding protein is a cyclic nucleotide-dependent protein kinase, enzyme activity in the purified fractions was assayed in the presence of cyclic nucleotides. While cAMP only provided a 2-3-fold stimulation (data not shown), cGMP stimulated kinase activity by ϳ500 -1000-fold (Table II). Moreover, Compound 1 inhibited the cGMP-dependent activity with an IC 50 of ϳ0.6 nM. These data are consistent with the conclusion that the antiparasitic activity of Compound 1 is due, at least in part, to inhibition of PKG.  2. Purification of the E. tenella Compound 1-binding protein. A, starting with an S100 fraction from E. tenella unsporulated oocysts, Compound 1 binding activity was enriched in a 4-column chromatographic purification scheme that is summarized. B, Compound 1 binding activity of fractions (20-l aliquots of each) following MonoQ chromatography, the final step in the purification protocol. C, a silverstained gel (2.5-l aliquots) of fractions from the MonoQ column with peak Compound 1 binding activity. An assay of nucleotide-dependent kinase activities in fractions of crude E. tenella S100 resolved by HiLoadQ chromatography is shown in Fig. 3. As noted earlier using the ligand binding assay, there was a small amount of nucleotide-dependent kinase activity that was not retained by this matrix. There was only a single peak of PKG activity in this initial purification step, and this activity was inhibited by Compound 1 (IC 50 of 0.6 nM). The peak of ligand binding activity co-eluted with the PKG activity (data not shown). In addition, there were two areas of PKA activity in this chromatographic profile. Both of these were resolved from the cGMP dependent activity and the cAMP dependent activity in these fractions was considerably less sensitive to Compound 1 (IC 50 of Ͼ 600 nM).
To further demonstrate that the purified 120-kDa ligandbinding protein is parasite PKG, in-gel kinase assays were performed. The MonoQ purified fractions pictured in Fig. 2C were separated in this gel system. The autoradiographs shown in Fig. 4A demonstrate that kinase activity was detected in the area of 120 kDa in the presence of cGMP. No activity was detected in the absence of cGMP. Nucleotide-dependent in-gel kinase activity was inhibited by Compound 1 (data not shown). No activity was detected in the area corresponding to the 150-kDa protein that co-purified with E. tenella PKG through the 4-column protocol. The peak of Compound 1 binding activity following MonoQ ion exchange chromatographic fractionation as shown in Fig. 2, was further characterized using cGMPagarose affinity chromatography. Elution of PKG activity is coincident with ligand binding activity (Fig. 4B). A silverstained gel of the cGMP eluate highlights an enrichment of the 120-kDa protein (Fig. 4C). Taken together, these data strongly support the conclusion that the purified 120-kDa ligand-binding protein from E. tenella is a PKG.
Kinetic Characterization of E. tenella PKG-A detailed analysis of the kinetic mechanism of purified native E. tenella PKG (EtPKG) was undertaken using a biotinylated PKI-derived peptide substrate. Systematic variation of both substrate concentrations yielded an intersecting double reciprocal plot (not shown) with K m values of 12 Ϯ 2 M for ATP and 19 Ϯ 1 M for the peptide (Table II). These results establish that EtPKG employs, in common with all protein kinases, a sequential kinetic mechanism wherein both substrates are bound to the enzyme prior to the release of product. Inhibition patterns with dead-end inhibitors were used to further elucidate the binding events. As indicated in Table III, the non-hydrolyzable ATP analog AMP-PCP was a competitive inhibitor versus ATP and noncompetitive with respect to peptide substrate. Conversely, guanethidine (a peptide competitive inhibitor) was competitive with the peptide substrate and noncompetitive versus ATP. These symmetrical inhibition patterns are consistent with the random addition of substrates to the enzyme. Compound 1 was also established to be a very potent ATP-competitive and peptide-noncompetitive inhibitor by a similar analysis.
Cloning and Recombinant Expression of Parasite PKG cDNAs-The partial cDNA clone generated as a RT-PCR product was used as a hybridization probe to screen an E. tenella unsporulated oocyst cDNA library and several clones were carried to plaque purity. The largest cDNA clone in this group (PKG7) is 4283 nucleotides in length with a deduced open reading frame of 1003 amino acids, capable of coding for a protein of nearly 113 kDa. Clone PKG7 has a 614-nucleotide 5Ј-untranslated region and a 657-nucleotide 3Ј-untranslated region. Six additional cDNA clones were also full-length, each capable of coding for the same 1003-amino acid open reading frame. Each of the seven tryptic peptide sequences from the purified 120-kDa protein are contained within this open reading frame. The deduced amino acid sequence of the full-length protein still most closely resembles Drosophila PKG; 31% identity to the cGMP-binding domains in the amino-terminal half of the protein and 45% identity to the catalytic domain at the carboxyl-terminal end of the kinase. PKG clones from E. maxima, T. gondii, C. parvum, and P. falciparum have also been isolated and the deduced amino acid sequences of the parasite proteins are aligned in Fig. 5A. The parasite PKG proteins are similar in size and, except at the extreme amino-terminal end, they share considerable sequence identity across their entire length.
The parasite proteins are also aligned in Fig. 5A with a human homologue, cGKII. It is immediately apparent that the parasite PKG proteins are considerably larger than the human enzyme and all other PKGs that have been described in the literature (generally 75-85 kDa). One notable exception is a splice variant of Drosophila PKG (31) that is predicted from cDNA clones to code for a 120-kDa protein. A schematic comparison of parasite and human PKG in Fig. 5B calls attention to a length of nearly 300 amino acids between the nucleotidebinding and catalytic domains that is found in the parasite PKG sequences, but is noticeably absent from human cGKII. The 300-amino acid signature sequence in the parasite proteins accounts for the bulk of the increase in size relative to other FIG. 3. Cyclic nucleotide-dependent kinase activity of E. tenella S100 proteins resolved by HiTrap Q column chromatography. An aliquot of 540 mg of E. tenella S100 protein was fractionated by HiTrap Q column chromatography. Kinase activity in fractions eluted with a NaCl gradient were assayed in the presence or absence of cyclic nucleotide (10 M cGMP or cAMP) and Compound 1 (20 nM) as described under "Experimental Procedures" using myelin basic protein as substrate.  Fig. 2). c Recombinant PKG expressed in T. gondii (see "Experimental Procedures").
PKGs. Closer inspection of the amino acid sequence in this region has tentatively identified a third cGMP-binding site, which is absent in the human enzyme. Stable heterologous expression of the E. tenella PKG cDNA clone with an amino-terminal FLAG-epitope tag ( FLAG EtPKG) has been accomplished in the related Apicomplexan parasite T. gondii. 3,4 Native and recombinant PKG enzyme preparations are indistinguishable in so far as K m for ATP and peptide substrates (summarized in Table II). Inhibition of native and recombinant enzyme activity by Compound 1 is also equivalent. The activity of both enzymes in the absence of cyclic nucleotide activator was extremely low and difficult to measure. The fold activation, which represents the quotient of the maximal rate and the unactivated rate, thus cannot be determined with great precision. This is unlike bovine PKG, 4 which has detectable enzyme activity in the absence of nucleotide cofactor. The cGMP nucleotide cofactor activates both native EtPKG and recombinant FLAG EtPKG dramatically and to a similar extent, typically by 500 -1000-fold (Table II). Analysis of activity in a cGMP titration revealed a strongly cooperative activation for both native and recombinant enzymes, with Hill coefficients of 2.3 and 1.8, respectively (Table II). The cyclic nucleotide-dependent activation kinetics of native and recom-binant EtPKG are indistinguishable and yet are considerably different from mammalian PKG enzymes (33,34). 4 Membrane-associated and Soluble Isoforms of Parasite PKG-Polyclonal antisera were raised against two peptides derived from the deduced E. tenella PKG sequence. In an S100 fraction prepared from E. tenella, each antisera detects two immunoreactive proteins and one of these co-migrates with the purified 120-kDa protein (Fig. 6A). The antisera also crossreacted with a 120-kDa protein in the peak binding fractions throughout the purification protocol (Fig. 6B). Upon final purification following MonoQ column chromatography, a second faster migrating immunoreactive band is evident. As this protein is not present in the S100 or in earlier fractions during purification, it is thought to represent a breakdown product. The 150-kDa protein that co-purifies with the 120-kDa protein does not react with either antibody.
An E. tenella P100 fraction was also analyzed by Western blot analysis. Following extensive washes of the P100 fraction with buffer lacking detergent (see "Experimental Procedures"), only a single immunoreactive band was detected. This protein co-migrates with the slower migrating bands in the S100 fraction (Fig. 6C). In a Triton X-114 phase separation experiment, the slower migrating band partitioned into the detergent-enriched layer, while the smaller protein was found in the detergent-depleted layer (Fig. 6D). Both of these experiments suggest that the less mobile immunoreactive protein is membrane  Fig. 2). Following renaturation, an in-gel kinase assay was performed (see "Experimental Procedures") in the presence or absence of 10 M cGMP. B, affinity chromatography using cGMP-agarose was performed using a 200-l fraction of highly purified E. tenella PKG (equivalent to fraction number 35 in Fig. 2B). Fractions of ϳ200 l were collected from the column flowthrough (FT), column washes with both binding buffer (W), and 1 mM GMP (G), and finally after elution with 15 M cGMP (cG). Aliquots of the respective fractions were assayed for Compound 1 binding activity and kinase activity using Kemptide as substrate. C, a silver-stained gel of the two cGMP-agarose chromatographic fractions with peak binding and PKG activities (7.5 l/lane).

Purification and Molecular Characterization of cGMP-dependent Protein Kinase
associated. Since each of the proteins binds to and is specifically eluted from a cGMP-agarose affinity matrix (data not shown), it appears that they represent two isoforms of E. tenella PKG.
The soluble E. tenella PKG isoform has been purified in sufficient quantity and subjected to amino-terminal sequence analysis. The NH 2 -terminal sequence begins with serine 49 within the open reading frame deduced from the cDNA clone.
While this could represent a proteolytic degradation product, Ser 49 is adjacent to Met 48 which in turn is in a good translation initiation context. Accordingly, the soluble PKG isoform might be the product of an internal translation initiation event. The membrane-associated PKG isoform, which is believed to be acylated at the amino terminus (35), has not been sequenced. The membrane-associated and soluble isoforms of native PKG are equally sensitive to inhibition by Compound 1, with IC 50 FIG. 5. Alignment of human and Apicomplexan parasite PKG sequences. A, alignment of the deduced amino acid sequence of PKG proteins cloned from the apicomplexan parasites E. tenella (EtPKG), E. maxima (EmPKG), T. gondii (TgPKG), P. falciparum (PfPKG), and C. parvum (CpPKG) with human cGKII (HscGKII) using the MultiClustal algorithm (20). Areas of amino acid residue identity are marked with a dark background. Sequences corresponding to the cGMP-binding domains are enclosed in the red boxes. The length of sequence that is unique to parasite PKG proteins is in green type. B, schematic comparison of a prototype apicomplexan parasite PKG to the human enzyme.
values of 0.86 and 0.6 nM, respectively. Likewise, FLAG EtPKG is inhibited by Compound 1 with an IC 50 of 0.6 nM.
E. tenella PKG Is a Monomeric Protein-While most PKG enzymes are homodimeric, dimerization is not required for catalytic activity (36). Dimer formation is dictated by a leucine zipper motif that is located at the amino-terminal end of PKG proteins (37,38). However, no such motif can be detected within the parasite PKG sequence. Native EtPKG purified as a soluble protein from S100 and analyzed by gel filtration chromatography appears monomeric, with an apparent molecular weight between 113,000 and 125,000 (data not shown). Since the soluble native protein lacks the initial 48 amino acids predicted from the cDNA clone, we cannot dismiss a potential role for these sequences in dimerization. However, the fulllength recombinant protein FLAG EtPKG also behaves as a monomer (data not shown). This demonstrates that the presence of the initial 48 amino acid residues of the open reading frame is not sufficient to cause dimerization. Experiments designed to estimate the molecular weight of the membrane-associated isoform of PKG are hampered by the association of the proteins with detergent micelles and are not reported here. DISCUSSION The results presented in this article clearly demonstrate the antiparasitic activity of Compound 1, both in cell based assays and animal models (Fig. 1). Compound 1 is structurally similar to known protein kinase inhibitors (29,30,39,40) and we show here that this compound does inhibit a kinase in the parasite, namely PKG. Like other protein kinase inhibitors, Compound 1 is competitive with respect to ATP. Structural conservation among protein kinases at the ATP-binding site predicts that inhibitor selectivity might represent a substantial obstacle. Purification of Compound 1 binding activity suggests otherwise. Only a single peak of ligand binding activity was detected throughout the purification scheme. While secondary binding sites with lower affinity or rapid off rates may not be detected using the assay conditions described here, we can conclude that PKG does represent the highest affinity and primary Compound 1-binding site in E. tenella.
The ability of Compound 1 to selectively inhibit parasite enzyme activity is a critical requirement for development as an antiparasitic compound. The area of greatest conservation between parasite and human PKG corresponds to the catalytic domain, where there is 44% identity (Fig. 5B). Modeling of the catalytic domain of mammalian PKG using the coordinates for the catalytic subunit of PKA, has called attention to several residues critical for ATP binding and catalytic activity (41). Each of these residues is conserved in the parasite PKG clones described here. Despite the amino acid identity of residues critical for ATP binding and catalytic activity, Compound 1 is a selective inhibitor of parasite enzyme activity. Chicken lung PKG, affinity purified by cGMP-agarose, and ion exchange chromatographies, is poorly inhibited by Compound 1(IC 50 of 9.3 M). 5 In the absence of cyclic nucleotide, PKG resides in an autoinhibited conformation (13,42) in which the amino-terminal end of the protein interacts with the catalytic domain. Even with this type of structural regulation, mammalian PKG activity can be detected without nucleotide cofactor. Addition of cGMP to the bovine cGKI isoform activates basal kinase activity up to 24-fold with little cooperativity. 4 In contrast, both native and recombinant preparations of E. tenella PKG have extremely low kinase activity in the absence of nucleotide cofactor. Upon addition of cGMP, the magnitude of stimulation of parasite kinase activity is up to 1000-fold and the Hill coefficient of ϳ2 predicts that cyclic nucleotide activation is a highly cooperative event (Table II).
The deduced amino acid sequence of PKG from multiple Apicomplexan parasite cDNA clones (Fig. 5, A and B) contains a block of nearly 300 amino acids that does not align with mammalian or invertebrate PKGs. Located in this region of the parasite proteins is a putative third cGMP-binding site, a feature that distinguishes the parasite enzymes from all other PKGs. The consensus for nucleotide binding is not striking (e.g. PKG site I, Glu 238 -(X) 8 Fig. 2) were probed with anti-TR3 at a dilution of 1:1000. B, fractions corresponding to the peak of ligand binding activity (0.5% of each fraction) from each column during the PKG purification protocol; also included is an aliquot of the flow-through from the HiLoadQ column. This panel was probed with peptide affinity purified anti-TR3 at a dilution of 1:1000. C, an extensively washed P100 membrane-enriched fraction and an S100 fraction (15 g/lane) were probed with peptide affinity purified anti-TR3 at a dilution of 1:1000. D, a Triton X-114 detergent extract of E. tenella UO (20 g), the corresponding Triton X-114 detergent-depleted phase (20 g), and the Triton X-114 detergent-enriched phase (12.5 g), were probed with anti-TR6 at a dilution of 1:500. The bands indicated by * and ** represent the membrane bound and soluble forms of PKG, respectively. cGMP binding to PKA and PKG, respectively (43,44). The x-ray crystal structure of the PKA regulatory domain (11) calls attention to a glutamic acid residue that hydrogen bonds with the 2Ј-OH of the nucleotide ribose. This is followed nine residues later by a conserved arginine that interacts with the phosphodiester of the cyclic nucleotide. The following amino acid, either a serine or threonine in PKG but not PKA, forms a hydrogen bond with the C-2 amino group of guanine in cGMP. Site-directed mutagenesis of mammalian kinases to replace this conserved T/S residue, had marked effects on the relative cGMP versus cAMP binding affinities of PKG and PKA (45). Using a series of nucleotide analogs along with site-directed mutations engineered at each of the three parasite cGMPbinding sites, we now know that each site functionally contributes to the unique activation kinetics of the parasite kinases (35). 4 Biochemical analysis demonstrates that there are at least two isoforms of parasite PKG (Fig. 6), a characteristic feature of mammalian cGMP-dependent kinases. However, unlike mammals, analysis of E. tenella 6 and T. gondii genomic DNA 3 indicates that PKG is coded for by a single gene in these organisms. The two genes typically found in animals code for related but distinct PKG isoforms. In some cases a primary transcript is also differentially spliced to generate further isoform heterogeneity (42). Exhaustive screening of E. tenella and T. gondii cDNA libraries never identified differentially spliced clones that might explain the presence of the two parasite isoforms. Expression of the full-length EtPKG and TgPKG cDNA clones each produces two proteins with electrophoretic mobilities that align with the native isoforms (35). Aminoterminal sequence of the soluble native isoform of E. tenella PKG has identified Ser 49 as the first amino acid. This residue is adjacent to Met 48 , which in turn is in a good translational initiation context; a counterpart exists at Met 103 in TgPKG. These results suggest that the truncated soluble PKG isoform could result from a secondary translational initiation event at an internal methionine codon. This conclusion is supported by mutagenesis studies (35) which demonstrate that differential isoform expression can be modulated by amino acid substitutions that alter potential initiator methionine codons M1, M48 (EtPKG), and M103 (TgPKG).
The NH 2 -terminal amino acid sequence of EtPKG and Tg-PKG deduced from the respective cDNA clones share a dual acylation signal (MGAC(S/I)SK). Site-directed mutagenesis experiments conclusively demonstrate that both of these enzymes, when expressed as recombinants, are myristoylated at G-2 and palmitoylated at C-4 (35). Transfection studies have allowed us to conclude that acylation is capable of directing membrane association of the larger PKG isoform. Similarly, myristoylation of mammalian cGKII is required for membrane partitioning of this PKG isoform. Membrane association of this isoform is critically important, enabling it to regulate CFTR, a reversibly phosphorylated integral membrane Cl Ϫ channel (46).
The two forms of E. tenella PKG that can be detected immunologically (Fig. 6) have different properties presumably dictated in part by their acylation status. Triton X-114 phase separation experiments (Fig. 6D) demonstrate that the higher molecular weight isoform is hydrophobic and the lower molecular weight isoform is hydrophilic. Although it might be predicted that the larger isoform would preferentially remain with the P100 fraction, we have not found this to be the case (Fig.  6C). The S100 fraction always contains an amount of the hydrophobic isoform. Extended washes of the P100 fraction in the absence of detergent are able to remove all traces of the soluble isoform under conditions where at least some of the hydrophobic isoform remains membrane associated. Consistent with these observations, proteins that are acylated, often only weakly associate with membranes (reviewed in Ref. 47) and have been found in soluble biochemical fractions (46).
When the E. tenella S100 fraction is applied to an anionexchange column, Western blot analysis confirms that the hydrophobic isoform is found in the flow-through (Fig. 6B). If the initial parasite extract is prepared in and subjected to ion exchange chromatography in detergent, the two isoforms bind and co-elute in this fractionation step. The most obvious explanation for this result is that the hydrophobic isoform behaves anomalously in the absence of detergent. It fails to bind to the anion exchange matrix and is likely to be responsible for the ligand binding activity identified in the column flow-through.
While the amino-terminal end of PKGs are generally not well conserved, molecular analysis has demonstrated that this portion of the protein has several functional domains. The leucine zipper motif (37) found in most PKGs is responsible for homodimerization of this family of proteins. This motif is notably absent from each of the parasite PKGs described here. While the functional significance of a dimeric PKG remains unclear, dimerization has been proposed to play a role in the association of PKG with cellular targeting proteins (48). However, it is clear that monomeric PKG, generated by NH 2 -terminal proteolysis or by mutagenesis, is catalytically active in response to cGMP (14). We report data in this article which demonstrate that native soluble E. tenella PKG behaves as a fully functional monomeric enzyme. Since native soluble EtPKG lacks the NH 2terminal 48 amino acid residues of the full-length protein, it could be argued that the full-length protein is a homodimer. Because of its hydrophobic character and association with detergent micelles, full-length acylated PKG is technically difficult to characterize by gel filtration chromatography. The recombinant FLAG ETPKG is a full-length protein that is not acylated but does contain the first 48 amino acids of the translational open reading frame. This recombinant EtPKG co-migrates in gel filtration experiments with the native soluble PKG, suggesting that the initial 48 amino acids of EtPKG do not contain a dimerization motif. Based on these data we conclude that both parasite PKG isoforms are monomeric proteins.
The list of potential substrates for cGMP-dependent protein kinases in higher eukaryotes has grown considerably long in the past several years (32,49). A role for PKG has been implicated in multiple distinct metabolic processes including the control of blood pressure, intestinal fluid secretion, and even erectile function (26). Clearly, several of these physiologic processes do not have counterparts in the parasite. The biochemical role of PKG in the parasite is unknown. It is tempting to speculate that information on kinase substrates could lead to the identification of new or synergistic parasitic chemotherapeutic targets.