Cloning and characterization of a cAMP-specific phosphodiesterase (TbPDE2B) from Trypanosoma brucei

Abstract

Here we report the cloning, expression, and characterization of a cAMP-specific phosphodiesterase (PDE) from Trypanosoma brucei (TbPDE2B). Using a bioinformatic approach, two different expressed sequence tag clones were identified and used to isolate the complete sequence of two identical PDE genes arranged in tandem. Each gene consists of 2,793 bases that predict a protein of 930 aa with a molecular mass of 103.2 kDa. Two GAF (for cGMP binding and stimulated PDEs, Anabaena adenylyl cyclases, and Escherichia coli FhlA) domains, similar to those contained in many signaling molecules including mammalian PDE2, PDE5, PDE6, PDE10, and PDE11, were located N-terminal to a consensus PDE catalytic domain. The catalytic domain is homologous to the catalytic domain of all 11 mammalian PDEs, the Dictyostelium discoideum RegA, and a probable PDE from Caenorhabditis elegans. It is most similar to the T. brucei PDE2A (89% identity). TbPDE2B has substrate specificity for cAMP with a K_m of 2.4 μM. cGMP is not hydrolyzed by TbPDE2B nor does this cyclic nucleotide modulate cAMP PDE activity. The nonselective PDE inhibitors 3-isobutyl-1-methylxanthine, papaverine and pentoxifyline are poor inhibitors of TbPDE2B. Similarly, PDE inhibitors selective for the mammalian PDE families 2, 3, 5, and 6 (erythro-9-[3-(2-hydroxynonyl)]-adenine, enoximone, zaprinast, and sildenafil) were also unable to inhibit this enzyme. However, dipyridamole was a reasonably good inhibitor of this enzyme with an IC₅₀ of 27 μM. cAMP plays key roles in cell growth and differentiation in this parasite, and PDEs are responsible for the hydrolysis of this important second messenger. Therefore, parasite PDEs, including this one, have the potential to be attractive targets for selective drug design.

Parasites of the Trypanosomatidae family are exposed to very different environments in their vertebrate host and insect vectors. During their life cycle, they adapt by developing a variety of morphological and biochemical changes. The signal transducers that control these extraordinary adjustments include cAMP, which is a mediator in the processes of cell transformation and proliferation. The intracellular levels of cAMP are significantly different depending on the life cycle stage both in Trypanosoma brucei and Trypanosoma cruzi (1, 2). The trypanosome, T. brucei, differentiates from long, slender bloodstream forms into short, stumpy forms that are infectious to the insect (3). This process, as well as the transformation of Leishmania donovani amastigotes into promastigotes (1), can be inhibited by nonspecific phosphodiesterase (PDE) inhibitors (such as caffeine, theophylline, or papaverine) and by cAMP analogs. Additionally, these agents have been shown to promote the in vitro differentiation of noninfectious insect stage epimastigotes into infectious metacyclic trypomastigotes in T. cruzi (4).

A process related to cell density seems to be responsible for triggering the slender to stumpy differentiation in bloodstream forms of T. brucei in culture (5). The signal responsible for cell cycle arrest in the G₁/G₀ phase that induces a quick and efficient differentiation is mediated by a low molecular weight factor termed SIF (Stumpy Induction Factor). This factor has been isolated from the conditioned growth media and its effect is mimicked by the mammalian PDE4 inhibitor etazolate or membrane-permeable cAMP derivatives, indicating again the likely involvement of cAMP in the regulation of metabolism in this parasite (6).

Cyclic nucleotide PDEs work as key components in the regulation of intracellular levels of cAMP by catalyzing its hydrolysis, and together with the adenylyl cyclases ultimately control the biological responses mediated by this messenger molecule. Eleven different mammalian PDE gene families have been described based on their distinct kinetic and substrate characteristics, inhibitory profiles, allosteric activators and inhibitors, and amino acid sequence (7). As a group they have been designated as class I PDEs. Additional forms of PDEs have been described in Saccharomyces cerevisiae (8), Dictyostelium discoideum (9), Vibrio fisheri (10), and Candida albicans (11), which exhibit very little amino acid sequence identity to the class I enzymes. These enzymes, currently designated as class II PDEs, likely have a different evolutionary origin because, in contrast to all other eukaryotic PDEs, they have catalytic domains unlike those in mammalian class I enzymes (8).

The initial biochemical descriptions of PDEs in trypanosomatids were from the group of Walter (12–14). Since then, two reports (15, 16) have been published on this topic in T. cruzi, with somewhat contradictory results. The first reports for cloning and expression of a PDE from T. brucei were the recent descriptions of TbPDE2A and TbPDE1 (17–19). We also recently have characterized a soluble cAMP-specific PDE from Leishmania mexicana (20). This enzyme, as well as the PDE2A from T. brucei (19), shows a much lower sensitivity to selective and nonselective PDE inhibitors than mammalian isozymes.

Taking advantage of the T. brucei gene sequencing project, we identified two expressed sequence tag (EST) clones homologous to, but distinct from, TbPDE2A. We report here the full gene sequence for a PDE from T. brucei, TbPDE2B, as well as the expression and characterization of its cAMP-specific hydrolytic activity.

Materials and Methods

Database Searching.

The amino acid sequences of mammalian PDEs (PDEs 1–10) were used as queries to search the EST database. The program used was blast (21), accessed from the database search and analysis search launcher (22).

Other Databases or Programs.

The GAF (for cGMP binding and stimulated PDEs, Anabaena adenylyl cyclases, and Escherichia coli FhlA) and catalytic domain boundaries were identified both by Hidden Markov Modeling (HMM) searches of the smart database and pfam: multiple alignments and profile HMMs of protein domains release 5.1 (Washington University, St. Louis). Alignment of GAF domains were constructed by using CLUSTAL W 1.8 and refined by visual alignment of known signature sequences. Pairwise sequence alignments were made by using the sim-Local Similarity Program accessed from the bcm search launcher. For K_m calculations enzyme activity data were analyzed with the prism program (GraphPad, San Diego) by using the one-site, nonlinear regression fit.

Primers.

Primers were designed by using the program amplify (23) and were purchased from Operon Technologies (Alameda, CA). Their sequences and designations are as follows: AA06.1s (GGAGCTGTTCCAAACCTTCTCTATGTTTG), AA06.2s (CTGGCGCCTCACTACGTAACTGTCGTATC), AA06.1as (GTTGTTTGTCAACTCACGGTTGAAGCG), AA06.2As (CCTGGTACGCGTCCTGAATATTCTCACC), W8.1s (GAAGTTAAGAAGCACCGTAATGTCCC), W8.1as (GATTCCGGATCAGAGAGGATCTCAAC), W8.2as (GCAAGGTTGCAGTGATGCACCTCAAG), AA.c5 (GTAAGATTTGTACATACTTCCGTGAAGGC), GAF.1s (GCTGGGAAAGACAGAGACAGATGACAC), AP1 (GTAATACGACTCACTATAGGGC), and AP2 (ACTATAGGGCACGCGTGGT).

DNA Sequencing and Sequence Assembly.

All PCR products were subcloned into the PCRII-TOPO vector (Invitrogen). Plasmid DNA was prepared by using the SNAP kit (Invitrogen). Sequencing was done by using an Applied Biosystems PRISM dye terminator cycle-sequencing kit (Perkin–Elmer), and sequencing reactions were purified by using Centri-sep columns (Princeton Separations, Adelphia, NJ). Sequences were assembled by using the program SEQUENCHER 3.0 (Gene Codes, Ann Arbor, MI).

Sequence Amplification.

Advantage Genomic Polymerase PCR mix was purchased from CLONTECH. Reactions were set up as follows: 0.2 μg of T. brucei DK-4 Istar 1.1 genomic DNA, 0.2 μM W8.1s primer, 0.2 μM AA06.1as primer, 2.5 μl of 10× reaction buffer (supplied with Advantage polymerase), 0.2 mM dNTP, 0.5 μl Advantage genomic polymerase mix, in a final volume of 25 μl, with the following cycling protocol on a GeneMate Genius PCR machine (ISC BioExpress, Kaysville, UT): 94°C for 30 s, 5 cycles of 94°C for 5 s, 72°C for 3 min; 5 cycles of 94°C for 5 s, 70°C for 3 min; 30 cycles of 94°C for 5 s, 68°C for 3 min. To obtain the missing 5′ and 3′ ends of the ORF the Universal Genome Walker Kit (CLONTECH) was used to produce five Genome Walker genomic “libraries” with the set of restriction enzymes: DraI, EcoRV, PvuII, ScaI, and StuI. Each batch of digested genomic DNA was ligated separately to the Genome Walker Adapter. PCRs were set up as follows: 1 μl of each genomic DNA library (template), 0.2 μM W8.1as or AA06.1s primer, 0.2 μM AP1 primer, 5 μl of 10× Tth PCR buffer (supplied with Advantage Tth polymerase), 0.2 mM dNTP, 1 μl Advantage Tth polymerase mix (50×), in a final volume of 50 μl, with the following cycling protocol: 94°C for 1 min, 7 cycles of 94°C for 25 s, 72°C for 3 min; 32 cycles of 94°C for 25 s, 67°C for 3 min; 1 cycle of 67°C for 7 min. A second PCR amplification was carried out by using the first PCR products, diluted 1:50, as template, and the primers W8.2as or AA06.2s and the AP2 primer with the same cycling protocol.

Generation of the Complete ORF.

To obtain the ORF sequence of the T. brucei PDE, we used the same protocol as for the sequence amplifications described above but with the primers AA.c5 and GAF.1s. This reaction was repeated three times, and each PCR product was subcloned and sequenced separately to avoid PCR artifacts.

Expression of T. brucei PDE.

The ORF sequence for T. brucei PDE2B was subcloned into the pcDNA 3.1-TOPO vector (Invitrogen) according to the manual (Eukaryotic TOPO TA Cloning, version C) and plasmid DNA was purified as described above. Human embryonic kidney 293 (HEK293) cells were transfected with 12 μg of DNA in 60 μl of GenePORTER Transfection Reagent (Gene Therapy Systems, San Diego) in 100-mm dishes and kept at 37°C in 5% CO₂ for 24 h. After this period fresh medium was added and incubated under the same conditions for an additional 24 h. The same amount of pcDNA vector containing the sequence for the green fluorescent protein was transfected under identical conditions as a positive control for expression and as a negative control for PDE activity. Cells from two plates were harvested at a time and homogenized with 1 ml of buffer containing 40 mM Tris⋅HCl (pH 7.5), 15 mM benzamidine, 15 mM 2-mercaptoethanol, 1 μg/ml pepstatin A, 1 μg/ml leupeptin, and 5 mM EDTA. The cell suspension was immediately subjected to sonication (3 × 5 s) on ice. One volume of glycerol and 1 mg/ml of BSA were added immediately to the homogenate. A pool from 10 plates was stored at −70°C in aliquots and did not lose appreciable activity over 1.5 months.

Sacchromyces cerevisiae Methods.

The yeast strain JBS21.51 (mat a; ade2–1oc; can1–100; his3–11,15; leu2–3,112; trp1–1; ura3–1; pde1∷HIS3; pde2∷Kan^r) was generated from Cry1 (mat a; ade2–1oc; can1–100; his3–11,15; leu2–3,112; trp1–1; ura3–1), a generous gift of T. N. Davis (University of Washington), using standard techniques of PCR-based gene replacement. The plasmid JBS52.19, containing the TbPDE2B entire ORF on a BstXI fragment was cloned into the SmaI site of p424 (2 μm origin, GPD promoter, TRP1 selection) (24). Sequencing of the splice junctions confirmed the plasmid construction. Strains JBS67.1 and JBS75 contain p424 in JBS21.51 or Cry1, respectively. Strain JBS67.2 is derived from JBS21.51 and JBS52.19. All transformations were carried out with the lithium acetate method of Gietz et al. (25). Strains with TRP1 plasmids were maintained on selective media. Heat shock was performed by replica plating cells to prewarmed (55°C) plates after 2 days of growth at 30°C. Plates were maintained at 55°C for 10 min to 2 h and allowed to cool to room temperature. After 2 days at 30°C plates were scored for growth. Soluble extracts were obtained from yeast according to the method of Atienza and Colicelli (26).

PDE Assay.

PDE activities were assayed at different concentrations of [³H] cAMP or [³H] cGMP according to the method of Hansen and Beavo (27). The reactions were performed in a buffer containing 40 mM Mops (pH 7.5), 0.8 mM EGTA, 15 mM Mg acetate, 0.2 mg/ml BSA in a final reaction volume of 250 μl. Concentrations from 0.03–300 μM [³H]cAMP were used to determine the K_m value in HEK293 cell lysates and concentrations from 0.002 to 10 μM [³H]cAMP were used for K_m determinations in yeast cell extracts. Hydrolysis of substrate did not exceed 20% under these conditions and PDE activity was proportional to time and enzyme concentration. For inhibition studies, assays were performed in the presence of rolipram (Biomol, Plymouth Meeting, PA), Ro 20–1724 (Hoffman–La Roche), zaprinast (May & Baker, Dagenham, U.K.), enoximone (Merrell Dow Research Institute, Cincinnati), sildenafil (Pfizer Central Research, Sandwich, U.K.), cGMP, papaverine, 3-isobutyl-1-methylxanthine, erythro-9-[3-(2-hydroxynonyl)]-adenine), pentoxifylline, etazolate, or dipyridamole obtained from Sigma using 1 μM [³H]cAMP as substrate.

cGMP Binding Assay.

Binding assays were performed by incubating 10–15 μg of protein homogenates from HEK293 cells expressing mouse PDE5A, TbPDE2B, or β-galactosidase for 2 h with various concentrations of cGMP. The binding buffer contained 25 mM NaCl, 2 mM EDTA, 0.2 mM isobutyl-methyl-xanthine, and constant specific activity [³H]cGMP (≈25,000 cpm/pmol). The same amount of serum albumin was used as a negative control for nonspecific binding. At the end of the incubation reactions were diluted 25-fold into ice-cold Tris buffer (20 mM, pH 7.5) containing 100 mM NaCl, 5 mM EDTA, and 3.3 M (NH₄)₂SO₄ and immediately filtered through 25-mm Millipore nitrocellulose filters (0.45 μM). Filters were washed two times with 4 ml of the same buffer and counted in Filter-Count scintillation fluid (Packard).

Results

Cloning and Sequencing.

Searches of the EST databases using sequences from the first 10 previously cloned mammalian PDEs resulted in two probable T. brucei rhodesiense EST PDE sequences. The first one (clone ID AA063739) corresponded most closely to the noncatalytic domain of PDE10A, PDE2, PDE5, and PDE6. The second (clone ID W84103) was homologous to the catalytic domain of all class I PDEs, including the conserved YHN PDE catalytic domain motif (24). A schematic diagram describing the cloning intermediates is shown in Fig. 1. Oligonucleotide primers were synthesized based on the sequence of these two EST clones and combined to amplify by PCR the gene sequence, yielding a clone 4.7 kb in length (clone WA3). This clone contained the EST sequence W84103 on the 5′ end and AA063739as on the 3′ end, indicating the possibility of two PDE genes in tandem. To extend the 5′ and 3′ ends of each gene five “genomic libraries” were prepared by using a Gene Walker kit and screened with the adapter primer AP1 together with the W8.1as primer (clone GW1) for the 5′ end and the primer AA06.1s (clone GW2) for the 3′ end (Fig. 1). Sequence alignment of all these clones yielded a sequence with two identical ORFs in tandem separated by a pyrimidine rich intergenomic region of 1,390 bp (Fig. 1).

Figure 1.

Diagram of overlapping genomic clones for TbPDE2B. The bottom scale is based on the nucleotide sequence of the full-length genes and drawn approximately to scale. The ORFs are indicated by ATG and STOP marked on the scale and are shaded gray. A, EST AA063739; W, EST W84103; IR, intergenomic region. Arrows indicate the primers (as specified in Materials and Methods) used to amplify each clone. GW1, GW2 indicate genomic walk 1 and 2, respectively. WA3 indicates a PCR amplicon that spans both genes.

To confirm this result, primers AA06.1s and W8.1as were used for a new PCR with genomic DNA as a template to yield a single band of 1.2 kb (clone AW4). This product contained the sequences of the clones AA063739 and W84103 as flanking regions and a sequence of 500 bp in the middle corresponding to the 5′ and 3′ ends of the gene missing in the first amplification (Fig. 1).

The complete gene (2,793 bp) was amplified as described in Materials and Methods, and the ORF sequence predicts a 930-aa protein with a molecular mass of 103,253 Da (Fig. 2). A consensus PDE catalytic domain is located between amino acids 668 and 908. The homology of this domain to other PDE catalytic domains suggests that TbPDE2B is a novel member of the recently described TbPDE2 family of class I PDEs (Fig. 3) (7, 19). Two conserved GAF domains in tandem are also predicted between amino acids 234–379 and 407–552 similar to those found in PDE2, PDE5, PDE6, PDE10, and PDE11 (Figs. 2–4).

Figure 2.

Complete gene sequence of TbPDE2B. Boxed amino acid regions indicate the two GAF domains and the catalytic domain identified by pfam sequence similarity to known domains in other proteins (42). * indicates the stop codon. Underlined YHN and HDX₂HX₄N motifs indicate PDE catalytic domains (see text). Italicized text indicates noncoding regions. Note that nucleotides are numbered beginning with 1 at the start ATG.

Figure 3.

Graphical representation of pairwise alignments of the TbPDE2B catalytic domain with the catalytic domains of one member of each of the known human class I PDEs as well as class I PDEs from Dictyostelium (DdRegA), C. elegans, and T. brucei (TbPDE2A). Catalytic domains were defined by pfam (42). The catalytic domain for the A gene of each of the 11 human PDEs was compared with the catalytic domain of TbPDE2B by blastp (21).

Figure 4.

Multiple sequence alignment of the TbPDE2B GAF domains to the homologous regions of several other representative PDEs. Part of the total GAF domain defined by pfam (42) is shown. Alignments were made with clustal w and refined manually (31). The columns marked with * or # indicate residues that bind cGMP in the crystal structure of PDE2A (unpublished work) or those recognized as a GAF domain signature motif, respectively (39). Letters after the PDE name, if any, indicate which GAF domain, A or B, is shown. TbPDE2B (T. brucei PDE2B AF192755), TbPDE2A (T. brucei PDE2A, AF263280), and HsPDE5A (Homo sapiens PDE5A, AF043731).

Expression and Characterization of Recombinant T. brucei PDE Activity.

To confirm that the isolated gene encodes an active PDE, a plasmid containing the complete ORF was expressed in HEK293 cells. The cAMP-hydrolyzing activity was determined at 1 μM substrate concentration by using cells harvested at 48 h after transfection. cAMP PDE activity was increased on average 10-fold above cells transfected with either the same plasmid containing green fluorescent protein coding sequence or nontransfected cells. However, no increase in cGMP hydrolytic activity was observed (data not shown), indicating that this sequence encodes for a cAMP-specific PDE. A more detailed kinetic characterization of the enzyme showed a K_m of 2.4 μM (± 0.6), as the average of three separate experiments. The catalysis of cAMP was not stimulated or inhibited by cGMP, at concentrations up to 200 μM.

cGMP Binding by T. brucei PDE.

When tested for the ability to bind cGMP with high affinity by the Millipore filter binding assay, no binding activity above the β-galactosidase transfection control could be detected. This was despite the fact that under the same transfection and assay conditions, extracts from mouse PDE5A showed easily measurable binding activity.

Inhibitor Specificity of Recombinant T. brucei PDE Activity.

Analysis with a battery of mammalian PDE inhibitors (Table 1) showed an extremely low sensitivity to the nonspecific PDE inhibitors papaverine, pentoxifyline, and 3-isobutyl-1-methylxanthine. No inhibition was observed in presence of specific inhibitors of the mammalian cAMP-specific PDE (PDE4) rolipram and R0 20–1724 for the recombinant enzyme; however, the endogenous PDE activity from the HEK293 cells was completely abolished with the lowest concentration of these compounds used in the assay (10 μM). No IC₅₀ could be obtained even at very high concentrations for specific inhibitors of the PDE 2, 3, 5, and 6 families (erythro-9-[3-(2-hydroxynonyl)]-adenine, enoximone, zaprinast, and sildenafil). Only dipyridamole at a concentration of 27 μM was able to inhibit 50% of the total activity in the assay. This value is from 6–71 times higher than those obtained for PDE5, PDE6, PDE8, and PDE10 at similar substrate concentrations. Etazolate had a weak inhibitory effect with an IC₅₀ of 127 μM.

Table 1.

Effect of different compounds on T. brucei PDEs

Inhibitor	PDE selectivity (IC50)*	TbPDE2A† IC50 (μM)	TbPDE2B‡ IC50 (μM)
IBMX	Nonselective (2–50 μM)	545	>1,000
Papaverine	Nonselective (5–25 μM)	ND	304 ± 19
Pentoxifylline	Nonselective (45–150 μM)	ND	>800
Rolipram	PDE 4 (2 μM)	>100	>300
Ro 20-1724	PDE 4 (2 μM)	ND	>300
Etazolate	PDE 4 (1.2 μM)	30.3	127 ± 4
Enoximone	PDE 3 (1 μM)	ND	>100
cGMP§	PDE 3	>100	>200
Zaprinast	PDE 5 (0.76 μM)	42.5	>50
	PDE 6 (0.15 μM)
Sildenafil	PDE 5 (0.0039 μM)	9.4	>100
EHNA	PDE 2 (1 μM)	ND	>180
Dipyridamole	PDE 5 (0.9 μM)	5.9	27 ± 3
	PDE 6 (0.38 μM)
	PDE 8 (4.5 μM)
	PDE 10 (1.1 μM)

ND, not determined. IBMX, 3-isobutyl-1-methylxanthine. EHNA, erythro-9-[3-(2-hydroxynonyl)]-adenine. 

^*From ref. 24. 

^† From ref. 19. 

^‡ Substrate concentration 1 μM [³H]-cAMP, mean ± SD, n = 3. 

^§ No inhibition or activation was observed. 

Complementation of S. cerevisiae PDE Deficiency.

To test whether this PDE retains activity in vivo, TbPDE2B was expressed in a PDE-deficient yeast strain (JBS21.51). Yeast cells lacking endogenous PDEs are sensitive to heat shock; they cannot survive incubation at 55°C (28). Several class I PDEs have been shown to complement this defect to varying degrees (19, 29). As shown in Fig. 5 expression of the TbPDE2B gene rescues this strain from heat shock sensitivity. TbPDE2B expressing yeast were tolerant of a strong (60 min at 55°C) heat shock, suggesting that the enzyme is highly active in yeast.

Figure 5.

S. cerevisiae rescue of phenotype. The S. cerevisiae heat shock sensitivity of cells lacking endogenous PDEs is rescued by a plasmid expressing TbPDE2B. JBS75 (PDE1 PDE2 containing p424), JBS67.2 (pde1 pde2 containing TbPDE2B on p424), and JBS67.1 (pde1 pde2 containing p424) were grown for 2 days at 30°C on selective plates, replica-plated to fresh selective plates, and held at 55°C or 30°C for 1 h before growing 2 days at 30°C. PDE represents cells that express PDEs; pde represents cells that do not express PDEs. Data are representative of three independent experiments.

Discussion

Through a bioinformatic approach, we have isolated a T. brucei gene encoding a cAMP-specific PDE. Sequence comparisons indicate that the T. brucei PDE2B is highly homologous to TbPDE2A and similar to other eukaryotic class I PDEs but has no extended homology to class II PDEs. TbPDE2B contains the conserved PDE catalytic domain initiating YHN motif, as well as the putative metal binding motif HDX₂HX₄N (30). Because this gene product is homologous to known PDEs, contains the signature PDE motifs, catalyzes the in vitro hydrolysis of cAMP when expressed in mammalian cells, and phenotypically rescues a S. cerevisiae PDE deficiency, it is apparent that this gene encodes an active PDE.

TbPDE2B and TbPDE2A appear to be recently diverged genes. The GAF and catalytic domains of the two genes match with >89% identity at the amino acid and DNA levels, but the genes are organized quite differently. TbPDE2A is a single gene flanked by two unrelated genes whereas TbPDE2B is arranged as two consecutive identical ORFs (19). TbPDE2B also encodes a second GAF domain, which is missing in TbPDE2A. It is plausible that TbPDE2A is a recent duplication of TbPDE2B, and that TbPDE2B duplicated even more recently to form a head-to-tail concatamer.

The two GAF domain sequences at the N terminus of TbPDE2B are very similar to the two GAF domains found in mammalian PDE2, PDE5, PDE6, PDE10, and PDE11 (ref. 7 and unpublished work). An alignment of the most relevant parts of the GAF domains of TbPDE2B with TbPDE2A, human PDE2A and PDE5A is shown in Fig. 4. These motifs were first identified as cGMP binding domains in the PDE2s and the photoreceptor PDE6s (32), but the subsequent identification of a similar motif in Anabaena adenylate cyclases and E. coli Fh1A, organisms that do not make cGMP, required a more general name for these motifs (33). Homologous domains are also present in a number of other signaling molecules that include transcription regulators and sensory histidine kinases in bacteria, ethylene-responsive factors and phytochromes in plants, and nitrogen fixation proteins in Azotobacter (34). Because of the probable differences in ligand specificities of this domain in the many different enzymes containing GAF domains, there is no consensus function for these domains. However, in most of the other PDEs cGMP binding to the GAF domains acts as a means for regulation of the enzyme. For example, the phosphorylation state of PDE5 (35, 36) and the interaction between the catalytic subunits and the inhibitory γ-subunits and transducin in PDE6 (33, 37) are regulated by cGMP binding. Binding of cGMP to the noncatalytic GAF domain in PDE2 acts as a direct allosteric activator of catalytic activity (38). The crystal structure of the GAF domains from mouse PDE2A showing detailed topography of the cGMP binding site has recently been determined (unpublished work).

Based on the sequence alignments shown in Fig. 4 the TbPDE2B GAF domains contain most of the residues thought to be important for binding of cGMP (31). However, to date we have found no effect of cGMP on PDE activity. We also have been unable to demonstrate high affinity binding to the TbPDE2B expressed in HEK293 cells under conditions where binding to PDE5A is easily detected. It is possible that the affinity for cGMP is too low to be detected by the Millipore filter assay. Unfortunately, we have not yet been able to produce large enough quantities of TbPDE2B to reliably determine by equilibrium methods whether cGMP is bound to the enzyme with lower affinity. It is quite possible that the TbPDE2B GAF domains function as binding sites for a different small molecule such as formic acid as in the case of E. coli FhlA transcription factor (39).

The entire catalytic domains of TbPDE2A and TbPDE2B are very highly conserved, consistent with the similar K_m found for the two isozymes (2.4 ± 0.6 and 2.3 ± 0.6, respectively) and the fact that each is specific for cAMP hydrolysis. Additionally, both isozymes are relatively insensitive to mammalian PDE inhibitors (selective or nonselective), a finding in keeping with the other described PDEs from the Trypanosomatidae family.

Nonselective PDE inhibitors slightly affected PDE activities present in T. cruzi (16), T. gambiense (12), L. donovani (13), L. mexicana (20), and TbPDE2A (19). An extremely low inhibition by the selective PDE inhibitors for PDE3 (enoximone), PDE4 (rolipram, Ro 20–1724), and PDE5 has been also shown for L. mexicana PDEs (20) and TbPDE2A (19). The sequence differences between the catalytic domains of the two members of the TbPDE2 family, clustered between amino acids 787–819 for TbPDE2B (347–392 for TbPDE2A), likely account for the differences in sensitivity observed for TbPDE2B toward sildenafil, dipyridamole, zaprinast, etazolate, and 3-isobutyl-1-methylxanthine (Table 1). It is striking that there is a >10-fold difference in sensitivity to sildenafil, given the high homology between these isozymes. The significant differences between sensitivities of trypanosomatid PDEs and their mammalian counterparts make these enzymes potentially good targets for development of selective drugs.

The high IC₅₀ obtained with the PDE4 selective inhibitor etazolate (127 μM) for the TbPDE2B reported here does not support the idea of this enzyme being the target of the effects described for this compound in the induction of in vitro transformation of slender to stumpy forms of T. brucei because it occurs at concentrations of 1–2 μM etazolate (6). Therefore, the observed effect of etazolate in the differentiation process of these parasites could be through the inhibition of another PDE or perhaps through actions on some other target. For example in mammals, etazolate is also an adenosine receptor antagonist and can interact with γ-aminobutyric acid channels (40).

There are at least two copies of the gene coding for this cAMP-specific PDE in T. brucei. These genes are tandemly repeated in the genomic DNA, in contrast to the single copy of TbPDE2A (19). The presence of more than one copy of a gene at a single locus is common for genes that encode enzymes essential for normal metabolism in Trypanosomatids. For example, phosphoglucose isomerase, aldolase, and glycosomal glyceraldehyde phosphate dehydrogenases are all multiple copy genes in T. brucei (41). The fact that there are also multiple copies of this PDE gene may suggest that it is not a functionally redundant enzyme and has important functions in the life of the trypanosomatid.

Abbreviations

PDE

phosphodiesterase

EST

expressed sequence tag

HEK

human embryonic kidney