The single cyclic nucleotide-specific phosphodiesterase of the intestinal parasite Giardia lamblia represents a potential drug target

Background Giardiasis is an intestinal infection correlated with poverty and poor drinking water quality, and treatment options are limited. According to the Center for Disease Control and Prevention, Giardia infections afflict nearly 33% of people in developing countries, and 2% of the adult population in the developed world. This study describes the single cyclic nucleotide-specific phosphodiesterase (PDE) of G. lamblia and assesses PDE inhibitors as a new generation of anti-giardial drugs. Methods An extensive search of the Giardia genome database identified a single gene coding for a class I PDE, GlPDE. The predicted protein sequence was analyzed in-silico to characterize its domain structure and catalytic domain. Enzymatic activity of GlPDE was established by complementation of a PDE-deficient Saccharomyces cerevisiae strain, and enzyme kinetics were characterized in soluble yeast lysates. The potency of known PDE inhibitors was tested against the activity of recombinant GlPDE expressed in yeast and against proliferating Giardia trophozoites. Finally, the localization of epitope-tagged and ectopically expressed GlPDE in Giardia cells was investigated. Results Giardia encodes a class I PDE. Catalytically important residues are fully conserved between GlPDE and human PDEs, but sequence differences between their catalytic domains suggest that designing Giardia-specific inhibitors is feasible. Recombinant GlPDE hydrolyzes cAMP with a Km of 408 μM, and cGMP is not accepted as a substrate. A number of drugs exhibit a high degree of correlation between their potency against the recombinant enzyme and their inhibition of trophozoite proliferation in culture. Epitope-tagged GlPDE localizes as dots in a pattern reminiscent of mitosomes and to the perinuclear region in Giardia. Conclusions Our data strongly suggest that inhibition of G. lamblia PDE activity leads to a profound inhibition of parasite proliferation and that GlPDE is a promising target for developing novel anti-giardial drugs.


Introduction
Giardia lamblia is a protozoan parasite that causes giardiasis, an intestinal disease with symptoms such as diarrhea, nausea, and malabsorption [1].Trophozoites are the disease-causing stage and colonize the upper small intestine of humans and other vertebrates.They form cysts, which are shed into the environment via the fecal route and which are then orally transmitted, mostly via contaminated water.Giardiasis occurs worldwide, predominantly in resource-poor countries with low standards of sanitation, and represents a major cause of non-bacterial diarrhea with 280 million symptomatic human cases every year [2].In developing countries, infection rates of 10% to 30% are common, though rates of 40% and higher have been reported in some instances [3,4].Chronic or recurrent giardiasis in early childhood is associated with poor cognitive function and failure to thrive [5].
Metronidazole (commercially known as Flagyl) and other nitroimidazoles are being used as a therapy of choice since the 1960s.However, resistance against metronidazole has been described [6,7].Consequently, substitution therapies including the benzimidazole albendazole, the acridine derivative quinacrine-or the aminoglycoside paromomycin, alone or in combination with metronidazole [8,9], are of increasing importance.New therapies are urgently needed because current treatments (i) depend on repeated dosing schedules (suboptimal for developing countries), (ii) have adverse effects, (iii) are ineffective in up to 20% of cases and (iv) clinical or laboratory-induced resistance has been reported for most of the current antigiardial drugs [10,11].
Phosphodiesterases (PDEs) are key enzymes of cyclic nucleotide signaling.They constitute the only enzymes for hydrolyzing the signaling molecules cAMP and cGMP and thus are crucially important regulators of the temporal and spatial shape of the cyclic nucleotide signals.Three structurally distinct classes of PDEs have been described [12].Thereof only class I enzymes have been identified in protozoan parasites and their mammalian hosts so far.Human PDEs (hPDEs) comprise eleven class I families (hPDE1-11), which differ with respect to substrate-specificity, regulation and distribution in tissues as well as in intracellular compartments.The catalytic domains of class I PDEs are highly conserved at the level of their three-dimensional structures, though the different families share only 20-50% amino acid sequence identity within their catalytic domains (S1 Table ).Small differences in structure and sequence of their catalytic pockets account for substrate selectivity (cAMP versus cGMP) and-most importantly -have allowed the development of family-specific PDE inhibitors [13].Most hPDE families are being actively studied as potential drug targets against a wide range of medical conditions and a number of PDE inhibitors are currently marketed for various conditions such as chronic obstructive pulmonary disease, psoriatic arthritis or erectile dysfunction [14,15].
The extensive available knowledge on PDE structure, physiology and pharmacology has prompted the study of PDEs as potential targets for the treatment of infectious diseases.In Trypanosoma brucei, the causative agent of African sleeping sickness, targeting PDEs by genetic or pharmacological means led to rapid parasite death in vitro and to the elimination of infections in vivo [16].Two inhibitors developed against the trypanosomal PDEs TbrPDEB1 and TbrPDEB2, namely NPD-001 and VUF13525, exhibit potencies in the nanomolar range [17,18].Recent studies in the malaria parasite Plasmodium falciparum have demonstrated that a PDE-inhibitor with specificity to a PDE expressed in the merozoite stage of the parasite led to the premature egress of merozoites, thus interrupting the multiple cycles of reinvasion and multiplication [19].In addition, genetic deletion of PDEγ in the rodent malaria pathogen Plasmodium yoelii blocked sporozoite motility and parasite transmission [20].In Giardia, cAMP has been suggested to be involved in excystation [21] and encystation [22], and a potential role in regulating trophozoite attachment and locomotion was hypothesized based on the intracellular localization of the protein kinase A (gPKA) [21].Thus, PDEs could play a vital role in the biology of this parasite.
In this study, we show that the repertoire of PDEs in the genome of the intestinal parasite G. lamblia is restricted to a single gene coding for a class I PDE (GlPDE).We expressed GlPDE lacking its N-terminal transmembrane helix region in yeast and characterized its enzymatic activity.We also show that the inhibition profile of a set of well-characterized inhibitors for human PDEs shows a good correlation between enzymatic GlPDE inhibition and anti-parasitic activity, and we have indication that GlPDE is associated with mitosomes, which are relic organelles in G. lamblia.Our results suggest that GlPDE represents a novel, and potentially promising drug target that should be further exploited for the development of novel and urgently needed medication against giardiasis.

In silico sequence analyses
The HMMER software package (version 3.1b1) was used to search the GiardiaDB databases (v 4.0) of annotated proteins and translated open reading frames (ORFs) for the presence of PDEs belonging to either of the currently established three PDE classes I, II and III [27].The profile hidden Markov models (HMMs) were generated from multiple sequence alignments of each PDE class: The HMM of class I PDEs was built from 24 catalytic domain sequences comprising one member of all 11 human PDE families, PDEs from kinetoplastid and plasmodial parasites, as well as PDEs from Drosophila melanogaster, Caenorhabditis elegans and Saccharomyces cerevisiae.The HMM of class II PDEs is based on all 7 manually annotated sequences in the Swiss-Prot database (release2014-04; entries CPDP_ALIFS, CPDP_YERPE, PDE1_CA-NAX, PDE1_DICDI, PDE1_SCHPO, PDE1_YEAST, PDE7_DICDI).To identify potential class III PDEs, two HMMs were created: (a) by using the twenty most diverse members of the NCBI CDD family PRK11148 (http://www.ncbi.nlm.nih.gov/cdd);(b) by using a seed alignment of 49 sequences contained in the HAMAP profile MF_00905 [28].All hits above the significance threshold (e < 0.0001) were analyzed manually.

Giardia lamblia cell culture
G. lamblia WBC6 (ATCC catalogue number 50803) trophozoites were grown under microaerophilic conditions in 11 ml culture tubes (Nunc, cat.156758) containing TYI-S-33 medium supplemented with 10% bovine serum and bovine bile according to standard protocols [40].Parasites were harvested by chilling the tubes on ice for 30 minutes to detach adherent cells and collected by centrifugation (900 x g, 10 minutes, 4˚C).Encystation was induced using the two-step method as described previously [41] by cultivating the trophozoites in bile-free medium for 44 hours and thereafter in medium (pH 7.85) containing porcine bile.

Yeast complementation assay
The heat-shock assay to detect complementation of the PDE-deficient phenotype of the S. cerevisiae strain PP5 was carried out exactly as described [43].Yeast strains were streaked onto SCleu plates (selective medium lacking leucine) and grown for 2 days at 30˚C.The colonies were then replica-plated onto YPD plates prewarmed to 55˚C, and were incubated for another 15 min at 55˚C.Plates were then cooled to room temperature and incubation was continued at 30˚C for 18-36 h.

Yeast cell extracts
Yeast cell pellets were created exactly as described previously [44].Briefly, yeast was grown at 30˚C in SC-leu medium to end log phase and thereafter again for 3.5 h in YPD to maximize protein expression.Washed cells were frozen in liquid nitrogen and stored at -70˚C.Frozen cell pellets were thawed on ice and resuspended in an equal volume of ice-cold extraction buffer (50 mM HEPES, pH 7.5, 100 mM NaCl, 1x Complete protease inhibitor cocktail without EDTA (Roche Applied Science)).Cell lysates were produced by three passages through a high pressure homogenizer (French press at 20'000 psi, 4˚C).Unbroken cells and cell debris were removed by centrifugation for 20 min at 16'000 RCF at 4˚C. 25% glycerol was added to the cleared supernatant.Aliquots were snap-frozen in liquid nitrogen and stored at -70˚C for subsequent PDE activity assays.

PDE assay
PDE activity was determined by a modification of the two-step procedure of Thompson and Appleman [45] as described previously [42].In all reactions no more than 20% of the substrate was hydrolyzed.Assays were always carried out in triplicates and at 1 μM cAMP concentration.Inhibitors were dissolved and diluted in dimethyl sulfoxide (DMSO), with 1% final DMSO concentration in the reaction mixes.Control reactions with DMSO alone were always included.Data were analyzed using the GraphPad Prism software package (GraphPad, San Diego).

G. lamblia trophozoite proliferation assay
Test compounds were dissolved in DMSO and serially diluted in Giardia culture medium.G. lamblia WBC6 trophozoites were grown until adherent cells reached a confluency close to 100%.Cells were then chilled on ice for 20 min and adherent parasites were detached from the culture tube.Cultures were diluted to 5x10 3 cells/ml in growth medium (room temperature) and 100 μl aliquots were immediately added to wells of 96-well plates preloaded with the same volume of test compound dilutions.DMSO (solvent) controls were always included.Plates were incubated for 72 h at 37˚C in a humid chamber with an anaerobic atmosphere.Adherent cells were then washed twice with 200 μl PBS before adding 200 μl of PBS containing 10 mg/ml resazurin and 1% glucose.After incubation for 4 h at 37˚C, plates were read on a fluorescence plate reader (Wallac 1420 VICTOR 2 , PerkinElmer) with excitation and emission wavelengths of 544 nm and 590 nm, respectively.Data were analyzed with the GraphPad Prism software package using a sigmoidal dose-response model for regression.All assays were done in quadruplicate and at least three independent experiments were performed.

Transient expression of HA-tagged GlPDE in Giardia trophozoites
C-terminally HA-tagged GlPDE was expressed under its endogenous promotor in G. lamblia trophozoites as follows: The entire open reading frame of GlPDE (gene ID GL50803_14058) and 43bp of the 5'-upstream region were PCR amplified from genomic DNA using the primer pair 5'-CGTCTAGATTTGTGTCATAAGCAAGGTAA-3' and 5'-CGTTAATTAACTA CGC GTAGTCTGGGACATCGTATGGGTACTCCTTCTTGTCCAGCTCAA-3'.These primers introduce a C-terminal HA-tag (underlined) and the two restrictions sites XbaI (bold) and PacI (bold italic) into the amplicon.The XbaI/PacI-digested PCR product was cloned into vector PAC-CHA [46] and Giardia WBC6 trophozoites were transfected by electroporation using 15 μg of plasmid DNA as input (device and settings: Bio-Rad Gene Pulser, 350V, 960 μF, 800 O).Stable transfectants were selected with the antibiotic puromycin (Sigma, cat.7699111) at a concentration of 77 μM.Expression of the PDE is driven by its endogenous promoter in the upstream region.The transiently transfected cells, expressing HA-tagged GlPDE under its endogenous promoter, were used in all localization studies.
Immunofluorescence analysis was performed on a Nikon Eclipse 80i microscope using the software Openlab 5.5.2 (PerkinElmer) for picture acquisition.

RNA preparation and reverse transcription
RNA was isolated from wild type cells using an RNAeasy kit (QIAGEN Cat.74104) following the 'Animal Cells Spin' protocol.A DNase I digestion was included to remove residual genomic DNA according to the manufacturer's protocol.RNA was eluted in 50 μl of RNase-free water.Reverse transcription was performed using a Qiagen Omniscript RT kit (Cat.205111) and cDNA was amplified by PCR using different combination of primers.

Results
The G. lamblia genome encodes only a single class I PDE The vast majority of genes in the Giardia Genome Database (www.GiardiaDB.orgversion 4.0) [49] were annotated based on automated annotation algorithms.However, Giardia is a highly diverged organism [50].To enhance the detection power for a putative PDE, we performed a comprehensive search in GiardiaDB using the HMMER software package (v 3.1b).To identify class I PDEs, a profile hidden Markov model (HMM) was built from the catalytic domain of 24 PDE sequences, including members of all eleven hPDE families as well as a selection from phylogenetically diverse species.This search identified a single class I PDE (GlPDE; evalue = 4e-90) in all G. lamblia isolates sequenced so far (gene IDs GL50803_14058, DHA2_150867, GL50581_303, GSB_153349, GLP15_4333).A putative PDE with similarity to GlPDE was also identified in the database of annotated proteins of the salmon parasite Spironucleus salmonicida, a diplomonad species such as G. lamblia.GlPDE and the Spironucleus PDE (SsPDE, GiardiaDB gene ID SS50377_13952) share a similar domain structure, a short stretch of weak homology in the N-terminal protein half (185 aa with 24% identity) and 35% identity in the C-terminal catalytic domain, indicating that both enzymes are distantly related members of the same PDE family (see S2 Fig) .To identify class II or class III PDEs [12] in G. lamblia, the database was queried with the appropriate profile HMMs (see Materials and methods), but no corresponding protein sequences could be detected.To validate the automated annotation of the GlPDE open reading frame, RNA from trophozoites in exponential growth phase was reverse transcribed, and the presence of the full length GlPDE transcript was confirmed.No substantial alterations in the RNA expression levels during in vitro-induced trophozoite-to-cyst differentiation were found in a recently published RNA-seq study [51].In contrast, the putative nucleotidyl cyclases identified in the GiardiaDB by means of profile HMM searches (S1A Fig), as well as the protein kinase A subunits (gPKAr and gPKAc) display a differential expression pattern during in vitro encystation (S1B Fig) .Domain organization of GlPDE.The open reading frame of GlPDE (GL50803_14058) encodes a protein of 1371 amino acids (molecular mass = 154.9kDa, pI = 7.4; see Fig 1A).The N-terminal region (aa 1-573) is predicted to contain eight or nine transmembrane helices (TMH).Interestingly, the first 71 residues, comprising two TMHs, are reminiscent for a predicted mitochondrial targeting signal (MitoProt II [52], p = 0.96).No significant similarity to any sequence in the TrEMBL database could be found within the 420 amino acids between the last TMH and the catalytic domain (aa 574-996).The highly conserved catalytic domain is located at the C-terminus (aa 997-1371; see Fig 1A).In accordance with average homology between Giardia genome assemblages A, B and E [50], GlPDE homologs show an overall sequence identity of 77 to 90% (S1 File).Lowest sequence conservation is found in the 140 aa -long region between TMH 5 and 6 (35 to 73% identity) and the highest one in the catalytic domain (86 to 94%).All residues of the substrate-binding pocket in the catalytic domain are Catalytic domain.The catalytic domain of GlPDE is predicted to conform closely to those of other class I PDEs of human or parasite origin.It is predicted to form 16 alpha helices that align well with those of human and kinetoplastid PDEs (Fig 1B).Sequence identity to the human PDEs varies from 21% (hPDE6A) to 32% (hPDE4D), which is in the range usually found between different PDE families (S1 Table ).All functionally important residues are highly conserved between GlPDE and the consensus sequence of human PDEs (see S3 Fig) .This includes the purine-scanning invariant glutamine (Gln-1317) and a pair of residues forming a hydrophobic clamp around the nucleobase (P-clamp [53,54] Leu-1279 and Phe-1329), as well as the eight strictly conserved residues of the metal-binding pocket at the site of catalysis (His-1068, His-1072, His-1143, Asp-1144, His-1147, Glu-1173, His-1176, Asp-1261).
However, GlPDE also contains some notable differences to the human enzymes.First, a 35 amino acid sequence is inserted into the loop between helix 6 and 7, reminiscent of an insert also found in hPDE3 isozymes, but the two sequences are not related.Secondly, the GlPDE catalytic domain contains a unique M-loop connecting helices 14 and 15 (Figs 1B and S5).The M-loop contributes to the catalytic pocket in human PDEs and is also involved in shaping the P-pocket of kinetoplastid PDEs and the related "selectivity pocket" of hPDE10, which both are exploited for inhibitor design [55][56][57].The M-loop of GlPDE is the least conserved part of the catalytic domain as determined by conservation scoring functions (trident/valdar01 [31]).
Interestingly, the H-loop contains a residue, Asp-1151, which is unique in GlPDE in terms of its physico-chemical properties since all hPDEs contain a polar but uncharged amino acid (Asn,Thr or Ser) at this position (S3 Fig) .In various human and kinetoplastid PDEs the corresponding residue forms a part of the catalytic pocket and has been recognized as an anchor point for inhibitors to increase isozyme selectivity [58][59][60][61][62][63].
A further prominent difference between GlPDE and the human PDEs is found outside of the catalytic cleft (residues Gly-1146 and Ala-1175).These two residues substitute a totally conserved acidic residue (usually aspartic acid) and an invariant histidine of human PDEs, respectively (S3 Fig), which have been implicated in forming a structure-stabilizing salt bridge [64] between the ends of the H-loop.Most importantly, the M-loop and the H-loop are completely conserved between the sequenced Giardia assemblages A, B and E (S1 File).

GlPDE functionally complements a PDE-deficient S. cerevisiae strain and hydrolyzes cAMP, but not cGMP in vitro
S. cerevisiae strains lacking both endogenous PDEs display heat-shock sensitivity as a phenotype [65].Functional complementation of this phenotype by heterologous PDE expression has proven to be a sensitive tool to validate PDE enzyme function [66,43,42].Since transmembrane domains tend to interfere with recombinant expression of proteins, a GlPDE construct was used that contained the full catalytic domain, but lacked the N-terminal transmembrane domains (amino acids Met-588 -Glu-1371) (see Fig 1A).The construct was expressed in the PDE-deficient yeast strain PP5 [66].As predicted from sequence analysis, the recombinant recombinant proteins expressed in yeast are marked with grey arrows.(B) Sequence alignment of the catalytic domains (α-helices 3-16) of GlPDE, human PDE4B, human PDE3B and T. brucei TbrPDEB1.Secondary structure features are shown in dark blue (α-helices) and light blue (3 10 -helices).α-helices are additionally numbered above the sequences (H3-H16).Residues of the substrate-binding subpocket are indicated with dots above the alignment.Thereof, the invariant glutamine is additionally marked with an @-sign (highlighted in yellow) and the hydrophobic P-clamp residues are tagged with green labels.Amino acids that are conserved in all 11 human PDE families and in GlPDE are shown in red letters and indicated with an asterisk below the alignment.Asterisks are underlined at the eight strictly conserved residues of the metal-binding pocket at the site of catalysis.The single histidine residue that is conserved among all hPDEs, but is substituted by an alanine in GlPDE (Ala-1175) is marked with a black triangle.Soluble lysates prepared from the recombinant yeasts expressing GlPDE(aa588-1371) hydrolyzed cAMP with standard Michaelis-Menten kinetics with a K m of 408 ± 37 μM (Fig 2B).The enzyme is cAMP-specific and does not accept cGMP as a substrate.No cAMP-hydrolyzing activity could be detected in lysates from the parental PP5 strain.

Effect of PDE inhibitors on GlPDE activity and on Giardia trophozoite proliferation
To pharmacologically validate GlPDE, a selection of 27 PDE inhibitors covering specific human and trypanosomal PDE families were assessed with cell lysates of recombinant yeast strain PP5 expressing GlPDE(aa588-1371).This profiling assay showed that most of the hPDE-family-specific inhibitors were essentially inactive against recombinant GlPDE, even at 100 μM concentration (Fig 3A).In contrast, the two trypanosomal PDE inhibitors NPD-001 and VUF13525, as well as the compounds dipyridamole, PFE-PDE9 (PubChem SID 15099396), roflumilast, piclamilast, and NEU222 showed promising activity against GlPDE.The same inhibitors were assessed for their activity against the proliferation of G. lamblia trophozoites in culture (Fig 3B).At 50 μM final concentration, most drugs showed no or only minor anti-giardial activity.However, those compounds that were active against recombinant GlPDE (roflumilast, piclamilast, PFE-PDE9, NPD-001 and VUF13525) also inhibited Giardia trophozoite proliferation, with the exception of the non-selective inhibitor dipyridamole.These data indicate a strong correlation between anti-giardial activity and GlPDE enzyme inhibition.To confirm our findings, the IC 50 values of the compounds found to be active as shown in Fig 3, and additionally the piclamilast analog NEU222, were determined against recombinant GlPDE (aa588-1371) and compared to the EC 50 values obtained in the cell proliferation assays.Again, there is a good agreement between their potency for enzyme inhibition and for anti-parasitic activity (Table 1).Representative dose-response curves of the most active compounds NPD-001 and VUF13525 are given in Fig 4.

The localization pattern of transiently overexpressed GlPDE is reminiscent of mitosomes and the perinuclear region
In order to investigate the localization of GlPDE in G.lamblia trophozoites, parasites were transfected with a plasmid encoding a C-terminally HA-tagged copy of GlPDE plus the 200 bp upstream region containing the putative endogenous promotor, resulting in the transiently transfected strain PDEp-PDEHA.The subcellular localization of the chimeric protein was detected by immunofluorescence staining in formaldehyde-fixed cells using an anti-HA antibody.The HA-tagged GlPDE was predominantly detected in dot-like patterns in the trophozoite periphery, and at distinct dots that are centrally located between the two nuclei.This localization pattern may indicate that GlPDE is associated with the various peripheral and the centrally located mitosomes.Further staining was found in the region surrounding the Giardia nuclei, which is reminiscent for the endoplasmic reticulum (see Fig 5) [67].
To exclude that GlPDE was localized in endocytic peripheral vesicles (PVs), the transiently transfected Giardia strain PDEp-PDEHA was first incubated in the presence of Dextran coupled to Alexa Fluor 594 (red), followed by fixation and staining with anti-HA antibodies and a secondary antibody conjugated to Alexa Fluor 488 (green) (see Fig 6).Microscopy clearly showed that the red PV staining and the green HA staining signals did not overlap.PVs were located more distally and were located closely underneath the plasma membrane, whereas the green anti-HA-labelled structures were found throughout the cell, but were less abundant in the cellular periphery.

Discussion
The present study reports on the identification and characterization of the single PDE from G. lamblia, GlPDE, and it addresses the question whether this enzyme could represent a valid drug target.The sequence characteristics unambiguously classify GlPDE as a new member of the class I PDE superfamily that includes all human and many protozoan parasite PDEs https://doi.org/10.1371/journal.pntd.0005891.g003[68,27].GlPDE exhibits many strongly conserved features with the human PDEs, but is also clearly distinct from them.
The N-terminal half of the protein is predicted to contain eight or nine putative transmembrane helices (TMH), of which the first two are predicted to represent a mitochondrial targeting signal.The TMH region is followed by a stretch of about 400 amino acids with no discernible domains, and a C-terminal catalytic domain with a high degree of similarity to those of the human PDEs.
The presence of such an extended N-terminal transmembrane domain containing two long extra-membranous regions (~140aa) is a feature that has not been described in other PDEs to date, though membrane-anchoring through a TMH bundle is also found in many other PDEs such as in hPDE3 [69] and in all five Plasmodium falciparum PDEs [70].
The prediction of an N-terminal mitochondrial targeting signal in GlPDE is in agreement with our finding, that HA-tagged GlPDE transiently expressed in Giardia trophozoites  suggests a localization at mitosomes.Mitosomes are organelles that are found in unicellular microaerophilic or anaerobic organisms such as Entamoeba, microsporidia and Giardia.They are believed to be derived from mitochondria, but lack the capability to gain energy from oxidative phosphorylation [71,72].A number of mitosome-associated proteins are closely related to those of mitochondria [73].Our tentative association of GlPDE with mitosomes may represent a further similarity with mitochondria of higher eukaryotes, where cAMP synthesis and hydrolysis is an important regulatory circuit for mitochondrial metabolism [74,75].Since mitosomes do not occur in mammalian cells, but are essential for Giardia and presumably regulated by cAMP signaling, future studies to verify the localization of GlPDE are certainly warranted.
With respect to the catalytic domain, all residues important for substrate recognition and catalysis are conserved between GlPDE and the consensus of human PDEs (S3 Fig) .Substrate specificity of PDEs is determined by a set of approximately twelve residues that define shape and chemical nature of the subpocket that accommodates the nucleobase [76,77,15].Concerning these residues, GlPDE conforms best to the cAMP-specific PDEs (S4 Fig).This is in full agreement with our experimental findings that recombinant GlPDE is cAMP-specific and does not accept cGMP as a substrate.The observed K m for cAMP of 408 μM is much higher when compared with hPDEs, where values do not exceed 20 μM for the preferred cyclic nucleotide [68].Although we cannot exclude an artifact of recombinant expression in yeast, class I PDEs with a high K m for their preferred substrate have been identified in other lower eukaryotes: For the PDEA orthologs of various kinetoplastid parasites, K m values between 190 and 600 μM have been reported [78,43,79].
The high K m of recombinant GlPDE(aa588-1371) for cAMP indicates that cAMP hydrolysis in Giardia occurs with first-order kinetics over a wide concentration range.The presence of a high-K m PDE as the sole enzyme for cyclic nucleotide degradation is somewhat unusual.However, a similar case is observed for the fission yeast Schizosaccharomyces pombe.Its genome codes only for a single class II PDE, Cgs2, that exhibits a high K m for cAMP.Nevertheless, this single PDE apparently is sufficient for all aspects of cAMP management, a situation akin to G. lamblia [80,81].
Other unicellular eukaryotes that express high-K m PDEs usually also express a low-K m enzyme, though their distinct roles are often not clearly understood.In the yeast S. cerevisiae, the high-K m , class II PDE, pde1, is the major regulator of agonist-induced cAMP signaling, while the low-K m , class I PDE, pde2, has only a limited effect on this pathway [82].In the slime mold Dictyostelium discoideum, several discrete pools of cAMP exist in the cell, and each is addressed by a specific pair of PDEs, one of high K m and low capacity, and the second of low K m and high capacity.In combination, they are thought to keep the average half-life of cAMP down, irrespective of fluctuating local concentrations [83].In the fungal rice pathogen, Magnaporthe oryzae, the high-K m PdeH and the low-K m PdeL play unique and non-redundant roles, with PdeH involved in conidial morphogenesis and PdeL in pathogenic development [84,85].
In order to pharmacologically evaluate the role of GlPDE, a panel of established PDE inhibitors with activity against human and/or trypanosomal PDEs was screened against recombinant GlPDE (aa 588-1371; lacking the N-terminal TMHs).We identified a number of compounds that inhibited GlPDE in the low micromolar or sub-micromolar concentration range.All these inhibitors, with the exception of dipyridamole, also impaired Giardia trophozoite proliferation.Interestingly, the compounds with good correlations of biochemical and cell-biological activities represent four different chemical scaffolds, and thus represent an excellent starting base for their further development into Giardia-specific PDE inhibitors and potential drug candidates.In addition, the amino acid sequence differences, and-by inference-the corresponding structural differences, between human and Giardia PDEs, indicate that a discrimination between human PDEs and GIPDE by appropriate structure optimization of drug candidates is feasible.
Not surprisingly, a few compounds of the original set exhibited an effect on trophozoite proliferation, but did not inhibit GlPDE enzyme activity, indicating off-target effects.It could also mean that such compound targets only the native full-length PDE in the parasite and not the heterologously expressed GlPDE(aa588-1371) used in the screenings.However, this is rather unlikely, since almost all known PDE inhibitors target the PDE catalytic pocket [15].In the case of BRL50481, the inhibitory effect on trophozoite proliferation may be attributed to the compound's nitro-group.Anaerobic organisms such as Giardia reduce nitro-groups to toxic nitroso derivatives, as it is the case for metronidazole, the current first-line drug against giardiasis [86].Off-target effects are also well known for established inhibitors of human PDEs.Vinpocetine, a classical hPDE1 inhibitor, also co-targets several proteins in addition to PDE1, such as voltage sensitive Na+ -channels and IKK, a "mediator" of NF-κB action [87][88][89].Thus, this compound is no longer considered a specific PDE inhibitor in whole-cell assay formats [90].
In conclusion, we show that G. lamblia expresses only a single PDE, which we here characterized on the molecular and functional level.A first pharmacological validation has shown that targeting GlPDE activity leads to significant impairment of G. lamblia throphozoite proliferation in vitro, which would be an important prerequisite to alleviate giardial disease symptoms.These promising findings warrant further studies in vitro and in vivo to exploit GlPDE as a potential target for the treatment of giardiasis.

Fig 1 .
Fig 1. Domain structure of GlPDE and alignment of its catalytic domain sequence with other PDEs.(A) Schematic representation of the domain structure.The predicted transmembrane helices (TMHs) between amino acids 1 and 573 are shown as red boxes.One segment is predicted with less confidence (predicted by 4 of 14 algorithms) and is depicted as a hatched box.The C-terminal catalytic domain is shown in blue and the GlPDE-specific 35aa-insert between helix 6 and 7 is indicated by a light-blue box.The N-terminal 71 amino acids that were recognized by the program MitoProtII as mitochondrial-targeting signal are indicated with an orange bar.GlPDE regions 588-1371 and 983-1371 corresponding to the https://doi.org/10.1371/journal.pntd.0005891.g001protein was catalytically active, and it fully restored heat-shock resistance of the mutant yeast strain, also in the presence of an N-terminal HA tag (see Fig 2A).Expression of a truncated version (see Fig 1A) comprising the catalytic domain solely (Gly-983 to Glu-1371) did not confer heat-shock resistance to the mutant strain.

Fig 3 .
Fig 3. Effect of human and trypanosomal PDE inhibitors on GlPDE activity and Giardia trophozoite proliferation.(A) Inhibitors were tested at 100 μM against recombinant GlPDE(aa588-1371) enzyme.(B) Trophozoite proliferation after 72h culture in the presence of 50 μM inhibitor.Compounds that are effective against trophozoites and also impair enzyme activity are indicated with green bars, those inhibiting only either throphozoite proliferation or GlPDE activity are presented with orange bars.Selectivity of the compounds for different PDE families is shown below the graphs using grey horizontal bars (non-sel.= non-selective; 1-10 = hPDE1-10; TbrPDEB = T.brucei TbrPDEB1/B2).

Table 1 .
Potency of selected PDE inhibitors.From those compounds exhibiting IC 50 <100 μM against recombinant GlPDE (aa588-1371) and EC 50 <50 μM against Giardia trophozoites in the benchmark screening (see in Fig3) exact dose responses were determined (values highlighted in green).PDE activity assays were done in triplicates and repeated at least once (SE of logIC 50 < = 0.1), and Giardia cell susceptibility assays were done in quadruplicates and repeated 2 to 5 times (SE of logEC 50 < = 0.15

Fig 4 .
Fig 4. Representative dose-response curves of the most active compounds.Panels (A) and (C) show IC 50 determinations for NPD-001 and VUF13525 against recombinant GlPDE(aa588-1371) enzyme (substrate concentration: 1 μM); Panels (B) and (D) show the dose-dependent effect of the same compounds on proliferation of G. lamblia trophozoites (WB, clone 6), determined after 72 h of anaerobic in vitro culture.Square symbols represent the mean values of determinations done in triplicate (A,C) or quadruplicate (B,D).Error bars represent the standard error of the means (SEM).https://doi.org/10.1371/journal.pntd.0005891.g004

Fig 5 .
Fig 5. Localization of GlPDE in Giardia trophozoites.C-terminally HA-tagged GlPDE was expressed under its endogenous promotor in Giardia lamblia trophozoites and detected with a FITC-conjugated anti-HA antibody.Cells of two independent transient transfections were analyzed by microscopy.An average of 50.6% of the cells were positively stained with the FITC-conjugated antibody and showed signal in discrete dots in the periphery of the cells and in the perinuclear region.In around 40% of the cells, the chimeric protein also aligned as-generally three-discrete dots between the two DAPI-stained nuclei, reminiscent of the central mitosomes.Upper row: Three representative transfected cells (1-3) and wildtype control cell (WB, clone C6).Lower row: Overlay of the same GlPDE-HA staining (green) with DAPI-stained nuclei (blue).https://doi.org/10.1371/journal.pntd.0005891.g005

Fig 6 .
Fig 6.Differential localization pattern of GlPDE-HA and internalized dextran in peripheral vesicles.Peripheral vesicles were loaded with Alexa Fluor 594-coupled dextran (red) and GlPDE-HA was detected with a FITC-conjugated anti-HA antibody (green).Nuclei were stained with DAPI (blue).The overlay of images shows completely different patterns of green and red signals.Left column represents a transfected cell, right column shows a non-transfected control cell.https://doi.org/10.1371/journal.pntd.0005891.g006 ). n.d., not determined.