Characterization of TbPDE2A, a Novel Cyclic Nucleotide-specific Phosphodiesterase from the Protozoan Parasite Trypanosoma brucei *

This study reports the identification and characterization of a cAMP-specific phosphodiesterase from the parasitic hemoflagellate Trypanosoma brucei. TbPDE2A is a class I phosphodiesterase. Its catalytic domain exhibits 30–40% sequence identity with those of all 11 mammalian phosphodiesterase (PDE) families, as well as with PDE2 fromSaccharomyces cerevisiae, dunce fromDrosophila melanogaster, and regA fromDictyostelium discoideum. The overall structure of TbPDE2A resembles that of human PDE11A in that its N-terminal region contains a single GAF domain. This domain is very similar to those of the mammalian PDE2, -5, -6, -10, and -11, where it constitutes a potential cGMP binding site. TbPDE2A can be expressed in S. cerevisiae, and it complements an S. cerevisiae PDE deletion strain. Recombinant TbPDE2A is specific for cAMP, with aK m of ∼2 μm. It is entirely resistant to the nonselective PDE inhibitor 3-isobutyl-1-methylxanthine, but it is sensitive to trequinsin, dipyridamole, sildenafil, and ethaverine with IC50 values of 5.4, 5.9, 9.4, and 14.2 μm, respectively. All four compounds inhibit proliferation of bloodstream form trypanosomes in culture, indicating that TbPDE2A is an essential enzyme.

Cyclic nucleotide-specific phosphodiesterases represent a large and divergent group of enzymes. In eukaryotes, two classes of phosphodiesterases (PDEs) 1 have been recognized (1,2). Class I includes all currently known families of mammalian PDEs (see below), as well as a number of PDEs from lower eukaryotes, such as PDE2 from the yeast Saccharomyces cerevisiae or the product of the regA gene of Dictyostelium discoideum. All class I enzymes show a considerable extent of amino acid sequence conservation within their catalytic domains. In contrast, the class II PDEs show little sequence similarity to the class I enzymes, and their K m are generally much higher than those of the class I enzymes. Class II PDEs were identified (e.g. in S. cerevisiae (1), D. discoideum (3), and Candida albicans (4)).
Class I PDEs have been studied particularly well in mammals, where 11 distinct PDE families have been identified to date, based on their substrate specificities, kinetic properties, allosteric regulators, inhibitor sensitivities, and amino acid sequences. Each family exhibits Ͼ50% amino acid sequence identity within the conserved catalytic domain of about 250 amino acids length, while sequence identity between different families is only 30 -40% in the same region (5). Family members also share extensive similarity in regions outside the catalytic domain, while no significant similarity of that region can be detected between different families. Besides their amino acid sequences, the different families also display distinctive pharmacological properties, which form the basis for the development of family-specific PDE inhibitors for clinical use (5)(6)(7)(8)(9). Genetic defects in PDE genes have been suspected as the underlying causes of a number of diseases (10 -13), although the relationship between PDE genes and disease is clearly established only between PDE6 and some forms of retinal degenerative disease (11).
PDEs have become highly attractive targets for drug development over the last few years. A growing number of familyspecific and subtype-specific inhibitors have been developed despite the considerable sequence conservation between the catalytic domains of different families. Several PDE inhibitors are being used or are under exploration for ailments as diverse as autoimmune disease (14,15), arthritis (16), asthma (17,18), and impotency (19,20) and as anti-inflammatory agents (21). The ongoing development of new and ever more subtype-specific inhibitors holds great promise for achieving more specific drug action with fewer side effects as well as new areas for the application of PDE inhibitors. In view of the potential of PDE inhibitors as chemotherapeutics, it is surprising how little is currently known about PDEs of parasites as possible drug targets.
The African trypanosome Trypanosoma brucei is a eukaryotic microorganism that causes the fatal human sleeping sickness (22), as well as Nagana, a devastating disease of domestic animals in large parts of sub-Saharan Africa. Chemotherapy of human trypanosomiasis is in a dismal state (23). New drugs and drug targets are urgently required, and the cyclic nucleotide-specific PDEs may constitute a class of new drug targets. cAMP signaling in trypanosomes is still largely unexplored (24,25). cAMP is involved in the regulation of differentiation of bloodstream form trypanosomes from the proliferative "long slender" forms to the insect-preadapted, nonproliferative "short stumpy" forms (26). Several gene families for adenylyl cyclases have been identified in T. brucei (24,27,28). 2 Even less is known about trypanosomal phosphodiesterases. An early study demonstrated PDE activity in cell lysates of bloodstream form T. brucei (29), and experiments with PDE inhibitors suggested that interference with PDE activity might affect cell differentiation (26,30). A recent report (31) described a PDE activity in cultured Leishmania mexicana. This enzyme is partly soluble and partly particulate, and its characteristics indicate that it may represent a class II PDE.
The current study describes the identification and characterization of a cAMP-specific PDE from T. brucei (TbPDE2A) with considerable amino acid sequence similarity to the class I PDEs. This classification is further supported by its low K m for cAMP as a substrate. In its N terminus, TbPDE2A contains a single GAF domain (32), which may represent a noncatalytic cGMP-binding site. TbPDE2A is a representative of a small, dispersed family of similar, but nonidentical genes. It can be expressed in the yeast S. cerevisiae, and it complements the heat-shock susceptibility phenotype of a PDE1/PDE2 deletion strain. Inhibitor studies showed that TbPDE2A is fully resistant to most inhibitors tested, including broad spectrum inhibitors such as IBMX. However, the enzyme is sensitive to the family 5 inhibitors dipyridamole and sildenafil, the family 3 inhibitor trequinsin, and the calcium-channel blocker ethaverine.
TbPDE2A Constructs-Screening of the T. brucei EST data base for class I PDE homologs identified plasmid pT2928 (accession number W84103) as coding for a potential T. brucei PDE. The cDNA fragment of pT2928 was completely sequenced and shown to represent the catalytic domain and the complete 3Ј-untranslated region of a candidate phosphodiesterase. This gene was termed TbPDE2A. The cDNA of pT2928 was then used to screen a genomic library of T. brucei in EMBL4 (37), and a 6.3-kilobase pair genomic DNA fragment was isolated and sequenced. The fragment contained the TbPDE2 gene, flanked by an unrelated gene on either side (see "Results"). Full-length TbPDE2A for recombinant expression was constructed as follows: The 3Ј-end of the open reading frame of TbPDE2A was amplified from the cDNA plasmid pT2928 using the forward primer pde2tyfor (5Ј-ATGACAATGGATG-GATGTGCTTAT-3Ј) and the reverse primer pde2tirev (5Ј-CTTCTCGA-GGGATCCCTATCCATGGGCAGACGAAGCCCCTGTACTC-3Ј), containing XhoI, BamHI, and NcoI sites (underlined) and a stop codon (boldface italic type). The resulting PCR fragment (366 bp) was cloned into pGEM-T-Easy (Promega) and verified by sequencing. The fragment was then excised by digestion with EcoRV and XhoI and was inserted into pT2928 digested with the same enzymes. This step removed the 3Ј-untranslated region and introduced an NcoI site immediately before the stop codon and resulted in plasmid pTPDE23U.
The 5Ј-end of the open reading frame was amplified from a fragment of genomic DNA, using the forward primer pde2gf2 (5Ј-GAGAAT-TCAAACATGTATGTGCACGACGTACGCATGTTC-3Ј), containing an EcoRI site (underlined) followed by a Kozak sequence and the start codon (boldface type, underlined), and the reverse primer pde2gr (5Ј-TTCAACCCCATATGATCAAGATCATGCACCAG-3Ј). The PCR product (804 bp) was cloned into pGEM-T-Easy, verified by sequencing, and then excised by digestion with EcoRI and NdeI and cloned into pTPDE23U cut with the same enzymes. This step generated a fulllength copy of TbPDE2A (pTPDE2A) containing an NcoI site immediately before the stop codon. The aim of these manipulations was to obtain a TbPDE2A gene with an unaltered open reading frame but with restriction sites introduced upstream (EcoRI) and downstream (XhoI, BamHI) of the start and stop codons, respectively, to facilitate further cloning steps. In addition, a Kozak sequence for efficient expression in S. cerevisiae was introduced in front of the start codon, as well as an NcoI site immediately before the stop codon for the later insertion of immunological tags (see below).
For generating an N-terminally truncated form of TbPDE2A without the noncatalytic cGMP-binding domain (starting at Met 124 of the full sequence), the corresponding region was amplified from genomic DNA using the forward primer pde2gf1 (5Ј-GAGAATTCAAACATGGAAGT-TAACGAACACCGAGCAACATTG-3Ј), containing an EcoRI site (underlined) followed by a Kozak sequence and the codon for Met 124 (boldface type, underlined), and the reverse primer pde2gr (see above). The PCR product (475 bp) was cloned and sequenced as indicated above. Finally, it was excised by digestion with EcoRI and NdeI and inserted into the corresponding sites of pTPDE23U to generate pTPDE2AT. The various constructs are outlined in Fig. 1.
For inserting a hemagglutinin tag (amino acid sequence: YPYDVP-DYAGIPM) at the C terminus of both constructs, two complementary oligonucleotides, Htfor (5Ј-CATGGTTACCCATACGATGTCCCAGAT- open BamHI site underlined), were annealed and then inserted into pTPDE2A and pTPDE2AT digested with NcoI and BamHI. The resulting tagged constructs (pTPDE2Ahm and pTPDE2AThm) were finally verified by sequencing. Similar constructs were also made that contain the TY-1 tag (38) instead of the hemagglutinin tag at their C termini.
For expression in S. cerevisiae, the hemagglutinin-tagged genes were introduced into the yeast expression vector p425CYC1 containing an attenuated CYC1 promotor (39) and into pLT1, which allows high level expression from a strong TEF2 promotor. pLT1 was derived from p425CYC1 by replacing its expression cassette with the TEF2 promotor, including the original TEF2 Kozak sequence. The initiation codon is followed by two restriction sites, which allow cloning of the gene to be expressed. The resulting sequence of the expression site is as follows: TEF2 promotor (Ϫ412 through Ϫ7), followed by 5Ј-CTAAACAT-GAGTCGACCTCGAGT-3Ј (Kozak sequence in boldface type, startcodon in boldface type and underlined, SalI site underlined, XhoI site in italic type). Protein expression and stability of the enzyme under assay conditions were monitored by immunoblotting, using a monoclonal antibody against the hemagglutinin tag (Roche Molecular Biochemicals).
Yeast Cell Lysis-Yeast cells grown to midlog to end log phase in SC-leu medium were collected, resuspended quickly in the original volume of prewarmed YPD medium, and incubated for an additional 3 h at 30°C to maximize protein expression. Cells were then harvested and washed once in H 2 O and once in HBB buffer (Hank's balanced salt solution, containing 50 mM HEPES, pH 7.5). The washed cell pellet was suspended in an equal volume of HBB containing a protease inhibitor mixture (Complete™; Roche Molecular Biochemicals). Cells were lysed by grinding with glass beads (425-600 m; Sigma) in 2-ml Sarstedt tubes using a FastPrep FP120 (3 ϫ 45 s at setting 4). After cell breakage, a hole was punched in the bottom of the tube with a needle, the tube was placed on top of a 5-ml plastic tube, and it was centrifuged in an SS34 rotor for 6 min at 6000 rpm. This step left the glass beads in the Sarstedt tube, while the cell lysate was collected in the plastic tube, where unbroken cells and large cell fragments formed a pellet. The supernatant was transferred to a fresh tube and was clarified by centrifugation for 15 min at 15,000 ϫ g. To the resulting supernatant, glycerol was added to a final concentration of 25% (v/v), and it was stored at Ϫ70°C. Under these conditions, TbPDE2A activity is stable for at least several months.
Phosphodiesterase Assay-PDE assays were done using the ZnSO 4 / Ba(OH) 2 precipitation method of Schilling et al. (40). This assay was chosen because it can potentially be upgraded to high throughput screening and because it produces less liquid radioactive waste than other procedures. The assay was validated and found to provide results essentially identical to those with other assay formats (40). The reaction contained 50 mM HEPES, pH 7.5, 0.5 mM EDTA, 10 mM MgCl 2 , and [ 3 H]cAMP or [ 3 H]cGMP (50,000 dpm/reaction) in a total volume of 100 l. Incubation was at 30°C for 20 min. Reactions were stopped by the addition of 50 l of 21.5 mM ZnCl 2 , followed by 50 l of 9 mM Ba(OH) 2 and incubated on ice for 30 min. The precipitates were filtered through GF-C glass fiber filters, and filters were washed three times with 1 mM NaOH, 100 mM NaCl. The filters were dried and counted in liquid scintillation fluid (4 g/liter omnifluor in toluene). All assays were carried out in triplicates and with three independent enzyme preparations. Controls for the efficiency of precipitation of cAMP and of AMP were always included. AMP was precipitated with an efficiency in the range of 55%, while the contamination of the precipitate with cAMP was in the range of 0.6 -0.8% of the input. The assay worked reliably over all cAMP concentrations used, up to 50 M. When assaying yeast cell extracts, control lysates from the PDE deletion strain transfected with empty vector were used as background controls. These control lysates never exhibited detectable activity beyond the background produced without lysate. The efficiency of 5Ј-AMP precipitation remained unchanged when yeast lysates were incubated for up to 2 h under assay conditions, indicating that the assay is not affected by a potential conversion of 5Ј-AMP to adenosine or adenine. Inhibitor studies were done at a cAMP concentration of 1 M, i.e. close to the K m of TbPDE2A, so that the IC 50 values should approximate the K i . Inhibitors were dissolved in Me 2 SO or ethanol, and the final concentration of the solvent never exceeded 1% in the assay reaction. Incubation times and enzyme concentrations were always adjusted so that less than 30% of the input substrate was hydrolyzed (2-5 g of total protein/100-l assay). IC 50 values were calculated by curve fitting on a four-parameter dose-response model with variable slope, using the Prism software package of Graph Pad Inc. (San Diego, CA).
Cytotoxicity Determination-Cytotoxicity of PDE inhibitors was determined for bloodstream forms in culture by determining acid phosphatase activity as described (41). Exponentially growing monomorphic bloodstream forms MiTat 1.2 were transferred into colorless medium (42) (cell density 3 ϫ 10 5 cells/ml culture) and were seeded into microtiter wells (199 l/well) containing 1 l of inhibitor or solvent control. Plates were incubated for both 20 and 40 h at 37°C in a humidified incubator with a 5% CO 2 atmosphere. At the end of the growth period, cells were lysed by the addition of 20 l of lysis/substrate buffer (20 mg/ml p-nitrophenylphosphate in 1 M sodium acetate, pH 5.5, 1% Triton X-100), and the incubation was continued for another 4 h at 37°C. Production of p-nitrophenol was determined at 405 nm on a microtiter plate reader. To control for intrinsic absorbance by the inhibitors, control series containing inhibitor dilutions but no cells were run for every experiment, and the resulting absorbance values were subtracted as background from the experimental readings. All assays were run in triplicates.
cAMP Determination-Intracellular cAMP concentrations were determined using an enzyme-linked immunoassay kit (BioMol). Bloodstream form trypanosomes (5 ϫ 10 6 /ml) were incubated with inhibitors at the concentrations and for the times specified. At the end of the incubation, 1-ml aliquots were centrifuged at 3,000 rpm for 8 min to collect the cells. The cell pellet was suspended in 150 l of 0.1 N ice-cold HCl. After incubation in ice for 20 min, the suspension was centrifuged at 12,000 rpm for 10 min. The HCl precipitate was processed for protein determination, while 100 l of the cleared supernatant was transferred to a fresh tube and used to assay the cAMP content, following the instructions of the supplier of the kit. Assays were performed in triplicates, and two independent experiments using different batches of cells were performed. The cAMP content of control cells as determined with this procedure was 66 Ϯ 14 pmol/10 9 trypanosomes, corresponding well with values obtained with different trypanosome strains and a different cAMP assay (26).

RESULTS
The TbPDE2A Locus-Upon searching the T. brucei expressed sequence tag data base for potential phosphodiesterase genes, an expressed sequence tag clone (pT2928) was identified. The corresponding plasmid was obtained (courtesy Philip Majiwa, ILRI) and sequenced. The cDNA fragment contained the 3Ј-part of a cDNA that unambiguously represented a phosphodiesterase gene, termed TbPDE2A according to the recently proposed rules for the nomenclature of trypanosomatid genes (43). Southern blot analysis of genomic DNA demonstrated that TbPDE2A is not a single gene but rather a member of a small gene family ( Fig. 2A). This was further confirmed by screening a procyclic cDNA library, 3 which resulted in the identification of several cDNA clones that represent different PDE2 family members (unpublished results). The cDNA fragment from pT2928 was then used to screen a genomic library of T. brucei, and the TbPDE2A locus was recovered on a 6-kilobase pair genomic EcoRI DNA fragment. The fragment was sequenced, as were several cDNA clones for TbPDE2A. The organization of the TbPDE2A locus (Fig. 2B) demonstrates that it contains three different, closely spaced genes. The first one is a RIME element (nucleotides 376 -876), a member of a family of abundant, highly transcribed, repetitive transposable elements (44). Within this element, nucleotides 868 -632 on the reverse strand represent the open reading frame coding for a RIME-associated protein. The RIME element is flanked by two 12-bp direct repeats (nucleotides 364 -375 and 877-888). The open reading frame for TbPDE2A extends from nucleotide 1770 to 3225 and codes for a protein of 485 amino acids. The predicted initiator methionine codon was functional, and the predicted open reading frame coded for an active protein when expressed in S. cerevisiae (see below). The coding region is followed by a long 3Ј-untranslated region of 1196 nucleotides, and the poly(A)-addition site is represented by nucleotide 4420. Downstream of the TbPDE2A gene, a gene for a member of the NHP2/RS6 family of nuclear proteins (45) is coded for by nucleotides 4635-5062. This is the first instance where such a gene was detected in T. brucei, and the role of the corresponding protein remains unknown. The presence of unrelated genes upstream and downstream of TbPDE2A demonstrated that the members of this PDE family are not closely linked.
Expression of TbPDE2A was analyzed both by Northern blot hybridization and by reverse transcriptase-polymerase chain reaction. Both approaches demonstrated that TbPDE2A is expressed both in the bloodstream and the procyclic (insect stage) form of the trypanosome life cycle (data not shown).
Predicted Amino Acid Sequence of TbPDE2A-The open reading frame of TbPDE2A predicts a protein of 485 amino acids, with a calculated molecular mass of 55,348 (Fig. 3). The N terminus of TbPDE2A contains a single GAF domain (Val 3 -Val 117 ; Ref. 32), which may function in cGMP binding. The presence of a single GAF domain in TbPDE2A is reminiscent of the human PDE11A, which also has a single GAF domain, while all other mammalian PDEs with such domains (PDE2, -5, -6, and -10) contain two of them in a closely spaced arrangement. The overall sequence identity between the single GAF domain of TbPDE2A and either of the corresponding domains of mammalian PDE2, -5, -6, -10, and -11 varies between 30 and 50%, with several residues (Leu 59 , Cys 60 , Pro 62 , Asn 77 , Lys 78 , Phe 88 , and Asp 91 ) strongly or absolutely conserved. For mammalian PDE5A, where cGMP-binding by the GAF domain was experimentally demonstrated, the interaction with cGMP was predicted to occur via Asn 77 , Lys 78 , and Asp 91 , all of which are strongly conserved (46).
The overall sequence conservation between catalytic domains of phosphodiesterases that belong to the same family is Ͼ50%, while between families, the extent of identity is less than 40% (6). When the sequence of the catalytic domain of TbPDE2A was compared with those of representatives of the 11 currently known mammalian PDE families, TbPDE2A exhibited no sequence identity of more than 40% with any of them. A comparison between TbPDE2A and class I PDEs from lower eukaryotes, such as PDE2 of S. cerevisiae, dunce of D. melanogaster, and regA of D. discoideum showed similarly low extents of sequence conservation.
TbPDE2A Complements PDE-deficient S. cerevisiae-TbPDE2A was expressed, either as the full-size enzyme or as the truncated form without the N-terminal cGMP binding domain (amino acids 124 -485), in an S. cerevisiae strain from which both endogenous phosphodiesterase genes had been deleted (PP5; Ref. 35). PP5 is very sensitive to heat shock due to the absence of phosphodiesterase activity. Transformants were tested for heat shock resistance (Fig. 4). Both the full-size enzyme and the truncated form fully restored heat shock resistance of the indicator strain, indicating that TbPDE2A is active in S. cerevisiae and that the N-terminal domain is not required for the activity of the catalytic domain. Two promotors of different strengths were used for these expression experiments (an attenuated form of CYC1 as a weak promotor and TEF2 as a strong promotor), but essentially identical results were obtained. Thus, minimal amounts of TbPDE2A are apparently sufficient to rescue the heat shock resistance phenotype of the PP5 strain.
Characterization of TbPDE2A Activity-For the characterization of the catalytic activity of TbPDE2A, the enzyme was expressed in the PDE-deficient yeast strain PP5, using plasmid pLT1 with the strong TEF2 promotor. TbPDE2A was expressed either as the full-length enzyme or in its N-terminally truncated form (amino acids 124 -485), which lacks the GAF domain. To be able to monitor protein expression and stability, both constructs contained a hemagglutinin tag at their C termini. In vivo activity of all constructs was first assessed by analysis of the heat shock phenotype conferred to the host strain, and stability under assay conditions was monitored by immunoblotting with an anti-hemagglutinin antibody.
Both constructs exhibited very similar activities with cAMP as the substrate, with a K m in the range of 2 M and a V max of 1 mol ϫ min Ϫ1 /mg (Table I). These K m values are well within the range of other class I PDEs. With both constructs, cAMP hydrolysis was unaffected by the presence of a 100-fold excess of cGMP in the reaction (data not shown). This observation defines the catalytic activity of TbPDE2A as that of a cAMPspecific phosphodiesterase. In addition, it indicates that cGMP either does not bind to the GAF domain or that such a binding does not directly influence the catalytic activity of the enzyme under the conditions of our assay.
Inhibitor Profile of TbPDE2A-Inhibitor studies were performed on lysates from PP5 expressing the full-size TbPDE2A. For the initial screening, all inhibitors were used at a concentration of 100 M, with a substrate concentration of 1 M cAMP. Only a few of the inhibitors tested exhibited a significant effect on enzyme activity (trequinsin, ethaverine, sildenafil, and dipyridamole), even at the high concentration used for the screen. Several others (etazolate, erythro-9-(2-hydroxy-3-nonyl)adenine, zardaverine, pentoxyfylline, 8-methoxy-IBMX, and papaverine) inhibited the enzyme activity by about 50%. A third group, including the broad spectrum inhibitor IBMX (rolipram, theophylline, Ro-20 -1724, IBMX, zaprinast, cilostamide, and vinpocetine), showed no significant effect on enzyme activity. With respect to its resistance to IBMX, TbPDE2A is similar to the mammalian PDE9 family (6). The observation that ethaverine was significantly more effective as an inhibitor of TbPDE2A than its parent compound papaverine was unexpected, since this compound, the ethoxy derivative of papaverine, was so far only known as a calcium channel blocker (53).
Subsequently, IC 50 values were determined for several inhibitors, using yeast lysates expressing the full-size construct pTPDE2Ahm (Fig. 5). The concentration of cAMP as substrate was set at 1 M, i.e. the range of its K m . Several structurally unrelated inhibitors showed similar potency against TbPDE2A, with K i values in the low micromolar range. The   FIG. 4. Heat shock resistance. The heat shock-sensitive PDE deletion strain of S. cerevisiae, PP5, was transformed with plasmids containing a weak promotor (attenuated CYC1; series 1) or a strong promotor (TEF2; series 2) and expressing the following constructs: N-terminally truncated TbPDE2A containing a C-terminal hemagglutinin tag (a); full-size TbPDE2A, containing a C-terminal hemagglutinin tag (b); empty vector (c); full-size TbPDE2A containing a C-terminal TY-1 tag (d). A, control plate without heat shock; B, plate with heat shock. Two or three independent clones were tested for each construct. potency of these inhibitors toward TbPDE2A is not correlated with their family specificity for mammalian PDEs (Table II).
Trequinsin is an inhibitor of the PDE3 family; dipyridamole inhibits families 5, 6, 9, 10, and 11 (7); and sildenafil is quite specific for family 5. Ethaverine was not known so far as a PDE inhibitor.
The four compounds were further analyzed for their effects on cell growth in culture. Bloodstream form trypanosomes were grown in microtiter plates for 20 or 40 h in the presence of serial dilutions of the inhibitors (Fig. 6), and cell proliferation was determined by an acid phosphatase-based assay (41). All four compounds inhibited trypanosome growth with IC 50 values that were about 10-fold higher than those determined with the soluble enzyme. The Hill slopes of the dose-response curves were close to 1 for three of the compounds (dipyridamole: 1.38 Ϯ 021; sildenafil: 1.73 Ϯ 0.69; and trequinsin: 1.09 Ϯ 0.63), while it was 5.19 Ϯ 1.52 for ethaverine. This indicates that the observed inhibition of cell proliferation by the first three compounds is indeed due to the inhibitory effect of the compounds on PDE activity, while the inhibition by ethaverine may be due to the combined effects of calcium channel blocking and inhibition of PDE activity. The results obtained with the first three compounds indicate that the activity of TbPDE2A, and possibly other members of this family, is essential for trypanosome proliferation in culture.
Inhibition of TbPDE2A Raises the Intracellular cAMP Concentration-Inhibiting PDE activity in vivo should raise the intracellular cAMP concentration. To verify if the inhibitors tested above do in fact inhibit trypanosomal PDEs in vivo, trypanosomes were incubated with sildenafil (200 M), trequinsin (50 M), dipyridamole (100 M), or ethaverine (100 M) for 30 and 120 min. Control cells were incubated for the same lengths of time without drug. Cell morphology and motility were normal under these conditions. Cells were collected and processed for cAMP determination as detailed under "Experimental Procedures." cAMP concentrations were normalized to the values of the control cells at the respective time points. The results given in Fig. 7 demonstrate that trequinsin, sildenafil, and dipyridamole, but not ethaverine, lead to an increase in the intracellular cAMP concentration, although the extent of this increase varies between the different inhibitors. The cAMP concentration of control cells (66 Ϯ 14 pmol/10 9 trypanosomes) corresponds very well with the values obtained earlier with another assay procedure and in different trypanosome strains (26). DISCUSSION The current study reported the identification and characterization of a member of a small family of cAMP-specific phosphodiesterases from T. brucei. This is the first report of cloning  a gene for a phosphodiesterase from a parasitic protozoon. TbPDE2A is coded for by a gene that represents a small family of related but different genes. DNA sequence analysis of the locus revealed the presence of genes unrelated to phosphodiesterases upstream and downstream of the open reading frame for TbPDE2A, demonstrating that the genes of this PDE family are not clustered. The open reading frame predicts a protein consisting of 485 amino acids, with a molecular mass of 55,313. The predicted start codon is functional, as demonstrated by expression of the recombinant protein in S. cerevisiae, and no potential extension of the open reading frame upstream of this start codon is predicted from the DNA sequence. The open reading frame codes for a protein with a C-terminal catalytic domain with strong homology to class I PDEs. The extent of sequence conservation and the inhibitor profile unambiguously classify TbPDE2 as a new family of the class I PDEs. The N-terminal moiety contains a single, well conserved GAF domain (32), which is separated from the downstream catalytic domain by a linker region of about 80 amino acids. The GAF domain is very similar to those of the mammalian PDE families that contain such domains (families 2, 5, 6, 10, and 11). TbPDE2A only contains a single GAF domain, while the mammalian PDE2, -5, -6, and -10 all contain two such closely spaced domains. In this respect, it most closely resembles the mammalian PDE11A (7). The functional significance of this unusual architecture of TbPDE2A remains to be explored. The fact that GAF domains can potentially bind cGMP may indicate that cGMP signaling is also present in T. brucei, lending support to an earlier claim that cGMP signaling might exist in T. cruzi (54). The domain may serve as an integrator for cAMP-and a cGMP-mediated signaling cascades. On the other hand, GAF domains are representatives of a large family of domains that bind assorted small molecules other than cGMP (32). Thus, not every domain predicted from its amino acid sequence to be a cGMP binding domain may actually function by binding cGMP. For instance, several Escherichia coli proteins contain predicted cGMP-binding domains, although E. coli does not con-tain a guanylyl cyclase, and cGMP is unlikely to play a role in this organism.
Analysis of recombinant TbPDE2A demonstrated that it is a cAMP-specific phosphodiesterase with a K m for cAMP in the 2 M range. This K m is typical for many of the class I PDEs. It is also in agreement with the available estimates of the intracellular concentration of cAMP in T. brucei (1-10 M; Ref. 26). Recombinant proteins with or without the GAF domain exhibited similar activities with cAMP as a substrate, and the activity of both constructs was not affected by the presence of excess cGMP. These observations confirm that TbPDE2A is a cAMPspecific phosphodiesterase and that cGMP either does not bind to the GAF domain or that such a binding does not directly affect its catalytic activity. Thus, the GAF domain may be involved in the interactions with other components of the cell.
TbPDE2A displays a unique pharmacology that sets it apart from all previously characterized PDE families. IBMX and theophylline, two nonselective inhibitors of most PDEs, are not effective on TbPDE2A. Three compounds that were found to inhibit TbPDE2A at the low micromolar level are specific inhibitors of different mammalian PDE families. Trequinsin (IC 50 for TbPDE2A ϭ 5.4 M) is an inhibitor of family 3, and dipyridamole (IC 50 ϭ 5.8 M) is an inhibitor of the mammalian families 5 and 6, as is sildenafil (IC 50 ϭ 9.4 M). However, the IC 50 values of trequinsin and sildenafil for TbPDE2A are much higher than those observed with mammalian PDEs. Furthermore, at a concentration of 5 M, trequinsin is likely to also inhibit PDE3 and PDE4. In conjunction, the current data indicate a pharmacological profile of TbPDE2A that is unlike that of any of the mammalian PDE classes. TbPDE2A is completely insensitive to the methylxanthine inhibitors; it is inhibited by the class 3 inhibitor trequinsin, although at rather high concentrations, while it is not affected by other class 3 inhibitors such as cilostamide. In addition, TbPDE2A is sensitive to ethaverine (IC 50 ϭ 14 M), a derivative of the nonspecific inhibitor papaverine, which exerts only marginal activity against TbPDE2A. This finding was unexpected, since ethaverine is pharmacologically known only as a calcium channel blocker (53).
A similar pattern of inhibition was observed when cytotoxicity was determined with cultured bloodstream forms. Interestingly, the dose-response curve for ethaverine showed a very steep Hill slope (5.19 Ϯ 1.52), indicating that the effect of this compound on cell proliferation might be due to a combined effect of calcium channel blockage and PDE inhibition. In contrast to ethaverine, dipyridamole (which is not only a PDE inhibitor but also a potent inhibitor of adenosine transporters) showed a Hill slope of around 1 (1.37 Ϯ 0.21), with no sign of cooperative inhibition of cell proliferation. This suggests that even in the presence of dipyridamole, sufficient amounts of purines can be taken up by the trypanosomes to allow unconstrained proliferation in culture. The presence of the TbPDE2A inhibitors leads to a continuous increase in intracellular cAMP concentration.
The biological role of TbPDE2A is currently not known. The identification of inhibitors of this enzyme has now provided the necessary tools for its investigation and for the experimental dissection of cAMP signaling in trypanosomes. The observation that inhibitors of TbPDE2A prevent cell proliferation in culture demonstrates that TbPDE2A, or the TbPDE2 family as a whole, may be essential for cell proliferation. This is also supported by the observation that TbPDE2 mRNA is constitutively expressed. In conjunction, these data indicate that TbPDE2A and its isoenzymes may represent interesting targets for the development of a new generation of trypanocidal drugs, based on phosphodiesterase inhibitors. TbPDE2A and its relatives in T. brucei as well as in other protozoa may offer a new class of targets for the development of urgently required, novel and effective anti-protozoal drugs.