CORDIS

Executive Summary:
PDE4NPD brought together efforts to tackle kinetoplastid diseases (human African Trypanosomiasis (sleeping sickness), leishmaniasis, Chagas’ disease) and one major helminth disease (schistosomiasis). The consortium has established a generic cyclic nucleotide phosphodiesterase (PDE) drug discovery platform to tackle these (and other) Neglected Parasitic Diseases (NPDs). The approach builds on insights and technologies that have been developed in the highly successful therapeutic targeting (e.g. Viagra®, Daxas® or Otezla®) of various members of the 11 human PDE families in the human genome. The PDE4NPD platform has cloned all genes encoding PDEs in studied (de)validated a number of parasitic PDEs as drug target via both in vitro and in vivo parasitology approaches. PDE4NPD has screened PDE-focused and fragment libraries by employing various target-centric biochemical, biophysical and pharmacological PDE studies. To support various medicinal chemistry programs PDE4NDP has invested strongly in structure-based approaches, resulting in the generation of 89 new PDE x-ray structures, next to a X-CHEM fragment screening campaign using one of the optimized x-ray systems.
Complementary to these molecular approaches, the consortium has also performed phenotypic screening on a large number of parasites, including the malaria parasite P. falciparum. The phenotypic testing of PDE4NPD compounds (> 1000) has resulted in a number of in vivo active compounds that are currently actively pursued for further development.
In conclusion, PDE4NPD has established a generic PDE drug development platform for tackling a wide variety of parasitic diseases and delivering a range of PDE-based drug development candidates

Project Context and Objectives:
2: Aims and objectives of PDE4NPD
PDE4NPD brought together efforts to tackle kinetoplastid diseases (human African trypanosomiasis (sleeping sickness), leishmaniasis, Chagas’ disease) and one major helminth disease (schistosomiasis). The consortium aimed to establish a generic cyclic nucleotide phosphodiesterase (PDE) drug discovery platform to tackle these (and other) Neglected Parasitic Diseases (NPDs). PDE4NPD aimed to screen PDE-focused and fragment libraries by employing various target-centric biochemical, biophysical and pharmacological PDE studies. Complementary to these molecular approaches, the consortium aimed to perform phenotypic screening on the respective parasites, including the malaria parasite Plasmodium falciparum.

The approach builds on insights and technologies that have been developed in the highly successful therapeutic targeting (e.g. Viagra®, Daxas® or Otezla®) of various members of the 11 PDE families in the human genome. PDEs are ideal for target-centric approaches, since (1) most state-of-the-art drug discovery technologies (e.g. computer-aided drug design, biophysical screening, structural biology, etc.) are applicable to PDEs; (2) there is vast experience in developing selective and safe human PDE inhibitors; and (3) PDEs display low functional redundancy. PDEs make such good drug targets because (1) PDE enzymes have a unique architecture at their active site, allowing the development of selective inhibitors and (2), in contrast to e.g. kinases, endogenous substrate levels of cAMP and cGMP are low (< 1-10 µM) and PDEs can therefore relatively easily be blocked with low concentrations of PDE inhibitors.

PDE4NPD had the main objective to establish a generic PDE drug development platform for tackling a wide variety of parasitic diseases in order to ultimately be able to support the discovery of PDE-based drug clinical candidates

PDE4NPD has focused on human African trypanosomiasis, leishmaniasis, Chagas’ disease and Schistosomiasis. Each of the parasite-specific programs were at a different stage of development at the start of the project and faced different challenges, as reflected in Table 1.

The overall objectives of PDE4NPD as listed in the original Description of Work are:
(1) Establish a generic PDE drug discovery platform that is applicable and available for combatting a wide variety of parasitic diseases. This will be achieved by establishing technologies, procedures, understanding and publicly accessible (chemogenomics) databases, that can be used for efficient PDE-based drug discovery for NPDs.
(2) Demonstrate the effectiveness of the drug discovery platform by delivering several PDE inhibitors as drug candidates. These inhibitors will target T. brucei TbrPDEB1 and TbrPDEB2, and T. cruzi TcrPDEC that have already been validated as targets by pharmacological and/or genetic means.

(3) Utilize the generic PDE drug discovery platform to tackle unexplored parasite PDE enzymes (e.g. Schistosoma spp., Leishmania spp., or other relevant T. brucei or T. cruzi PDE subtypes), e.g. by efficiently delivering chemical biological probes that can be used for PDE target validation in the respective parasites, while further developing chemical biological tools and molecular genetics technologies for validated parasite PDE target enzymes.

(4) Accumulate unique knowledge on targeting PDEs in pathogenic protozoa and helminths. This PDE4NPD knowledge will be captured in an annotated database and made available for the scientific community via open data sharing resources.

(5) To successfully disseminate and exploit the generic PDE drug discovery platform results and inhibitors through expansion of the current collaborations, attracting additional funding and stimulating an open-innovation platform that is able to efficiently hand promising chemical and biological leads to the international scientific community, SMEs and industry.

(6) develop a worldwide PDE4NPD platform. The platform will encourage other groups (academia, SMEs, major pharmaceutical industry) to tap into the knowledge and expertise of the platform, developed during the course of this project. Several pharmaceutical and biotech companies are willing to make compounds, biological materials and other assets available to PDE4NPD.

(7) To link with potential end users, like the DNDi, in order to progress drug development candidates resulting from PDE4NPD.

Project Results:
3: Project results
The PDE4NPD project was managed by Prof. Rob Leurs of Vrije Universiteit Amsterdam with support of Lygature, whereas the research was performed by 10 different groups from EU (Prof. Augustyns and Prof. Maes – University of Antwerp, Belgium, Dr. Bailey and Dr. England – IOTA, Cambridge (UK), Prof. Brown – University of Kent (UK), Dr. Gil and prof. Martinez– CSIC (Spain), Dr. Gul – Fraunhoffer (Germany), Dr. De Koning – Unversity of Glasgow (UK), Prof. Leurs and prof. De Esch – Vrije Universiteit Amsterdam (Netherlands)), Brasil (Dr. Soeiro – Fiocruz, Rio de Janeiro) and Egypt (prof. Botros – TBRI, Cairo), supported by a Scientific Advisory Board, consisting of Dr. Michael Pollastri (USA), Dr. David Manallack (Australia), Dr. Eric Chatelain (DNDi), Dr. Julio Martin-Plaza (GSK Tres Cantos Open Lab Foundation) and Prof. Tom Seebeck (Switzerland).
In the next sections the project results are presented with a focus on the establishment of the PDE4NPD platform and the application of the platform in drug discovery efforts for human African trypanosomiasis, leishmaniasis, Chagas’ disease and schistosomiasis.

3.1. PDE4NPD platform
PDE4NPD has focused a lot on establishing an efficient drug discovery platform for parasitic PDEs with a focus on PDE cloning and expression strategies, PDE structural biology, a PDE fragment-based and high-throughput screening approach and a phenotypic screening platform for efficient evaluation of new chemical matter of the PDE4NPD partners. The various established platforms are each presented in the next sections.

3.1.1. PDE gene cloning
In order to study the PDEs in isolation for structural studies, pharmacological screening, kinetic and functional characterisation, assessment of genetic variations, etc., it is necessary to obtain the DNA encoding them, preferably for the actual strains being used in the screening process. DNA was thus obtained from TBRI for S. mansoni (Cairo strain), from Fiocruz for T. cruzi (Y and Colombiana strains) and from UA for L. donovani and L. infantum in order to clone the PDEs of each organism at UGLA, which also provided the DNA for T. brucei (Lister 427 strain). Each of the kinetoplastid species contained 5 PDEs in their DNA, only some of which had been cloned before: PDE-A, B1, B2, C and D. Except for T. brucei PDE-A and PDE-C, which were already known not to be drug targets (see table 1), all were amplified by PCR from their respective genomic DNA and their sequences determined. For S. mansoni, the genomic database, which as yet is only partially annotated, was scoured for PDE-related open reading frames (ORFs); this was hampered by the presence of multiple very large introns in each ORF. Ten apparent PDE genes were initially identified, with one more identified in the last months of the project after the release of an updated version of the genome online. In this case, cloning was attempted from cDNA, obtained by reverse transcriptase from mature mRNA from the Cairo strain of S. mansoni, and was successful for 9 out of 11 predicted genes, with the other two apparently not expressed in the human-infective forms from which the RNA was obtained, or not present at all in the genome of this strain. Each of the nine successfully amplified genes were re-synthesised in the regular codon usage of the cells in which they had to be expressed: T. brucei and Saccharomyces cerevisiae for functional complementation, and insect cells and E. coli for protein production/purification. All DNA sequences either have been or soon will be deposited at GenBank. The S. mansoni genes were assigned names according to the homology to human PDEs in a phylogenetic analysis. Two of them, PDEs 8 and 9C, had unique and substantial insertions in their catalytic domains, of 31 and 170 amino acids, respectively.
For many of the kinetoplastid PDEs and all the S. mansoni PDEs, it had not yet been established whether these were indeed functional PDEs and whether they were specific for cAMP or cGMP (or both). In order to address this question, two complementation systems were utilised in parallel; both based on the ‘rescue’ of a cell lacking other cAMP PDEs. The S. cerevisiae system had been described before (Kunz et al. 2004) and is based on complementation of the thermosensitivity of the Δpde1/2 strain. Expressing a cAMP-metabolizing PDE in these cells restores growth after exposure to elevated temperature. The system was used for the successful complementation of S. mansoni PDEs 1, 4A, 8 (original), 8 (insert removed), 9A and 11 and also provided the first evidence that that the PDE-D enzymes of the kinetoplastids are functional cAMP-metabolizing PDE, as L. major and T. brucei PDE-D both complemented in this yeast strain. Lysate from the yeast cells was also used for a first characterisation of the PDE-D activity. In addition, TbrPDE-D was shown to be essential in that all efforts to create a gene deletion failed, unless a ‘rescue copy’ under a tetracycline-inducible promotor was first introduced. This means that PDE-D is now an additional, potential drug target in T. brucei and possibly other kinetoplastids, with an emerging pharmacology.
In addition to the yeast complementation, PDE4NPD has developed a novel system in T. brucei to study complementation, inhibitor profiles and function of single parasite PDEs in live protozoan cells. In this system, a single allele of the essential TbrPDE-B1/B2 locus was replaced with an antibiotic resistance cassette, after which the heterologous PDE gene to be studied is introduced, under control of the tetracycline-inducible promotor, followed by the deletion of the second B1/B2 allele in the presence of tetracycline. Successful complementation would be demonstrated by the continued growth of these cells in the presence of tetracycline, and the death of the culture upon withdrawal of tetracycline, demonstrating complete dependence on the expression of the heterologous PDE for cell survival. This system was successfully employed for SmPDE-4A, TbrPDEB1, TbrPDE-B2, TcrPDE-B1, Tcr-PDEB2, Tcr-PDE-C, and L. infantum PDE-B2 and constructs for most other PDEs have been made and introduced into this system at this point. The successful complementation has allowed comparative studies with multiple cell lines, each expressing a single PDE gene upon which its survival depends, giving the effectiveness of a range of inhibitors for each PDE in a relevant protozoan cell system. Combination of these datasets with inhibitor studies of the purified enzymes, where available, has given valuable insights into the extent that cellular penetration limits the anti-parasite activity of some compounds. Moreover, these cell lines can be used to study inhibitor profiles of full-length PDEs without the need for protein production and purification as in the example below, which shows that some compounds (e.g. NPD-010 and NPD-024) are simply inactive due to the lack of cellular uptake, whereas others including NPD-335, NPD-226 and NPD-001 display substantially higher activity against SmPDE-4A than against the T. brucei PDE-Bs (but somewhat selective against B2 over B1).

3.1.2 PDE4NDP structural biology

3.1.2.1 Recombinant Protein Production
The primary approach for recombinant protein production utilized both E. coli and insect cell systems (Figure 1). For E. coli constructs of varying length were designed based on domain boundaries and cloned into pET28a vector, in-frame with a hexa-his tag. The recombinant vectors were used to transform competent T7 express E. coli cells and protein production was induced at 18°C with 0.5-1 mM Isopropyl ß-D-1 thiogalactopyranoside (IPTG). Full length PDEs were expressed and purified from S. frugiperda cells. To achieve this, coding sequences were cloned into either pFastBAC-HTA or pOPIN vectors and recombinant Bacmids were generated. Baculovirus particles were produced by transfecting sf9 cells with recombinant bacmids, which were then used to transfect sf21 cells for protein production. The 6xHis tagged proteins were purified using the same purification strategy as used for various shorter length constructs.
We have also explored the N-terminal GAF domains of T. brucei with the aim to gain mechanistic insight into the PDE activity in these parasites. Towards this end, constructs spanning GAF-A and GAF-B together as well as in isolation were designed and proteins were produced in E. coli.
Table 2 summarises the most suitable reagents generated for usage by PDE4NPD partners. It does not capture the fact that in many cases multiple constructs of varying length and tags were produced to find these “optimal” reagents.
Where the generic approaches only yielded insoluble (or very low yields of soluble protein we also examined use of more exotic systems such as LEXYS (Eukaryotic protein expression system in L. tarentolae) and a number of cold shock protein expression vectors.
The PDED isoforms from each of the parasites studied were particularly problematic and the highly conservative sequences of PDED from T. cruzi and T. brucei showed significant challenges in protein expression which meant we had to explore more exotic systems which included the use of chaperones. Many constructs were originally tried in order to produce soluble material. We resorted to two expression vectors pColdI and pColdTF that promote expression under an action of cold shock protein, which can assist protein refolding and solubility and hence increases the final protein yield. A second technique that was used was co-expression of PDED with chaperones from the Takara chaperone plasmid set. Plasmid pGro7 contains groEs and groEL chaperones that promoted protein folding and, in the end soluble material was obtained. The protein yield in pColdI vector was 4 mg/1L, while after fusion with trigger factor the protein yield was higher at 10 mg/L.
As with all proteins produced we ran an enzymatic assay in order to analyze the activity of PDED to catalyze cAMP. This data verified that it was a true phosphodiesterase albeit with significantly lower activity than other parasitic PDEs.

3.1.2.2 Provision of X-ray crystal structures
Provision of X-ray crystal structures of PDEs for the purpose of providing high resolution atomic models of the parasitic PDEs and human PDE’s in complex with ligands to guide Structure Based Drug Design of anti-parasite agents. This has been done by the cloning of the catalytic domains of Trypanosoma brucei PDE-B1 (TbrPDEB1), Trypanosoma brucei PDE-B2 (TbrPDEB2), T. cruzi Y strain PDE-C (TcrPDEC), Schistosoma mansoni PDE-4A (Sm PDE4A), human PDE4D (hPDE4D), human PDE4B (hPDE4B) and also a chimeric PDE4 loop swapped construct of human PDE5. All structures solved exhibited the well-known all α-helical fold (Figure 2). Each of the proteins were cloned and expressed fused to an N-terminal 6xHis tag, allowing purification by affinity chromatography (see Table 2), which was subsequently removed prior to crystallization and subsequent structure determination by X-ray crystallography in the presence of ligands of interest (Figure 3).
All structures solved have been deposited in the project version of the PDEstrian database and a number of them have been included in manuscripts, both submitted and in preparation, and have also been deposited in the Protein Data Bank (see Table 3) for use by the scientific community. Below we indicate for each crystallised PDE some relevant details

TbrPDEB1 (catalytic domain): A fragment of 942 bp (314 amino acids) of the TbrPDEB1 gene, cloned in to the vector pET28a included the complete catalytic domain with at least a dozen amino acid residues flanking at both ends, following the same protocol that lead to the production of the crystal structure of the Leishmania major PDEB1 (Wang et al 2007). The construct was used to transform competent T7 express E. coli cells and protein production was induced at 18 °C with 1 mM Isopropyl ß-D-1 thiogalactopyranoside (IPTG). The pET28 vector fuses the PDE to a His tag, allowing rapid purification on a Ni-NTA column. A high level of purification was achieved, after which the His-tag was removed using thrombin digestion. Final yield was 2.5 mg protein/L bacterial culture. Catalytic activity of the produced protein was confirmed using the PDELight™ HTS cAMP Phosphodiesterase Assay Kit (Lonza), and an initial inhibitor profile was obtained testing the compounds of the so-called PDE toolbox. To date, 36 crystal structures with inhibitors have been produced and are helping to inform Hit and Lead Development.

TbrPDEB2 (catalytic domain): In order to established an alternative structural biology platform for TbrPDE-focused drug discovery, TbrPDEB2, an isoform of TbrPDEB1, was assessed. This was an essential step since it is only the inhibition of both B1 and B2 that kills T. brucei, and it therefore must be established that their binding site is indeed equivalent and furnishes the same pharmacology. A gene segment coding for its catalytic domain residues 600-918 was cloned in frame with an N-terminal 6xHis tag that was succeeded by an engineered TEV protease cleavage site in the pET28a(+) vector. Protein expression and purification were carried out as described for TbrPDEB1 and a yield of 6 mg/L of E. coli cells was achieved. A high cAMP hydrolysis activity of the protein was confirmed using the PDELight™ HTS cAMP Phosphodiesterase Assay. As expected, a similar inhibitor profile was obtained as for TbrPDEB1 when the protein was screened against PDE-toolbox as described before. Finally, the protein was used in crystallisation trials employing commercially available reagents and its structure was determined at 1.1 Å resolution. Its noteworthy here that this is the highest resolution structure obtained for any PDE so far. The crystal system was used to determine inhibitor-bound structures and a total of 5 inhibitors have been resolved at excellent resolutions.

TcrPDEC (catalytic domain): The catalytic domain of TcrPDEC (Y-strain) was produced using a similar method of E. coli expression and affinity purification as described above for TbrPDEB1, to a protein yield of 2 mg/ml, and catalytic activity confirmed using the same commercially available kit. Several inhibitors were identified from the PDE4NPD toolbox series of inhibitors; To date 4 crystal structures have been solved including that of a non-hydrolysable cAMP and a PDE inhibitor to 2.6 Å resolutions.

SmPDE4A (catalytic domain): The catalytic domain boundaries of Schistosoma mansoni phosphodiesterase 4A (SmPDE4A) were determined by sequence comparison with other known PDEs and a gene segment coding for residues 303-671 was cloned into the pOPIN-F vector. Protein was expressed in E. coli cells, purified to a yield of 3 mg/L and found to be active in cAMP activity assays. The protein was subsequently subjected to crystallisation trials yielding crystals in space group P3 which diffracted to a useful resolution of 2.3 Å. The structure was determined by molecular replacement using hPDE4D catalytic domain as the search template and revealed a fold similar to other known PDEs. Additionally, an AMP-bound structure was also obtained at 2.0 Å and showed a conserved mode of interaction, highly similar to those of as known for other cAMP-specific PDEs.

Human PDE4D (catalytic domain): In addition to the two highest priority parasite PDEs, the catalytic domain of human phosphodiesterase 4D (hPDE4D), as the closest human homologue, has been similarly produced in E. coli, to a final yield of 5 mg purified protein/L culture. This is essential for the lead optimization strategy, which aims to improve the parasite-to-human PDE selectivity while retaining or enhancing the inhibition of parasite PDEs. Crystal structures of 27 inhibitors (predominantly matched pairs with TBrPDEB1 structures) have been delivered.

Human PDE4B (UCR2+ catalytic domain): In parallel with hPDE4D, crystallisation and ligand-bound structure determination efforts have been undertaken on human phosphodiesterase 4B (hPDE4B), a subtype of the human phosphodiesterase 4. A regulatory domain construct of hPDE4B, containing N-terminal UCR2 and catalytic domain residues, was cloned in frame with an N-terminal 6xHis tag and expressed and purified to a final yield of 5 mg/L of insect cell culture. Efforts on hPED4B were kept low priority as the hPDE4D system was routinely yielding high resolution crystal structures with inhibitors of interest; however, a total of 4 inhibitor bound structures were determined.

Human PDE5LS: The catalytic domain coding segment of human PDE5 gene, where a section encoding one loop was replaced by the equivalent of hPDE4B for better stability, was cloned into the pFastBac insect cell expression vector. The expressed protein was purified by NI-NTA and ion exchange chromatography and used to produce high quality diffraction crystals. The latter were used to determine inhibitor-bound structures and a total of 12 such structures were determined successfully that assisted in our drug discovery pipeline.

Biophysical assays for PDE4NPD drug discovery
PDE4NPD partners devised and refined protein immobilization, SPR and ITC assays for measurement of kinetic and thermodynamic data respectively. These data were utilized in providing additional understanding of binding of the inhibitors developed in the project. Moreover, the University of Kent also undertook fragment screening by SPR, NMR and X-ray of two fragment files: a small 450 compound Fluorine fragment file and a fragment library from the Structural Genomics Consortium. The Fluorine compound file was tested against the human PDE5LS cGMP-binding PDE by SPR and NMR, with subsequent “hits” being followed up by crystal structure determination. This was followed by X-ray fragment screening of the SGC library and a small example of PDE-biased fragment set (PDE actives from CHEMBL- selected by IOTA) by X-ray utilizing the XCHEM platform at DIAMOND light source which was performed towards the end of the project and has yielded a number of interesting hits for future follow up (see Figure 4).
3.1.3 PDE4NPD fragment and HTS screening platform
The PDE4NPD partners involved were responsible for hit finding using biochemical, fragment-based and virtual screening, along with structural approaches, to identify novel chemical structures to initiate medicinal chemistry lead generation against new PDE targets, and to provide new structural classes to aid established medicinal chemistry programs.
The following activities have been successfully completed:
• Screening of the IOTA Fragment Library of ~1,500 compounds against six parasite PDE enzymes: TbrPDEB1 & PDEB2, TcrPDEB1 & PDEC, SmaPDE4A and LmPDEB1
• Cheminformatics analysis of the hits from the fragment screens showing clusters of chemically-related active fragments, plus some single actives.
• The information from the fragment screens was used by the medicinal chemistry programs for inhibitors of LmPDEB1, SmPDE4A and TbrPDEB1
• Crystallography has been performed on selected hit compounds with TbrPDEB1, and docking modes determined.
• The PDE activity assay was transferred to European Lead Factory, and a library of 450,000 compounds has been screened against a trypanosome PDE. A number of potent hits which are at least 10-fold selective against human PDE4D have been identified.

Fragment Screening Assay
The IOTA fragment library contains ~1,500 compounds of which > 90% are compliant with the “Rule of 3” criteria widely used for selection of fragments for initial drug discovery screening. The library has a broad diversity and novelty and has been screened successfully against many potential therapeutic targets including a number of enzymes.
Fragment screening was a joint effort of PDE4NPD partners IOTA and Fraunhoffer. All PDE screening has been conducted using the Perkin-Elmer LANCE® assay, which is a time-resolved fluorescence resonance energy transfer (TR-FRET) assay measuring the concentration of cyclic AMP remaining after incubation with the PDE. This assay format was chosen because it eliminates many of the problems of standard fluorescence assays and also does not use any following enzymes, which were known from previous work to give a high false-positive hit rate with the IOTA fragment library.
The concentration of cyclic AMP used in the assay was 8 nM, which is well below the Km for all the PDE enzymes. This makes the assay very sensitive to detecting weak hits, important when screening fragments which are expected to have low affinities for the enzymes. For each PDE enzyme, the assay was optimized such that the Z’ factor (a measure of assay reliability) was >0.6. Fragments were initially screened against each enzyme at a final concentration of 200 µM. Table 4 shows the hit rates obtained for the six different PDE enzymes. The most active compounds in each screen were selected for dose-response curves, and Ki values determined for between 30 and 100 compounds depending on the primary assay hit rate (Table 5). Data from both the primary assay and the Ki determinations, show that there are significant differences in the ability of the fragments to inhibit the different parasitic PDE enzymes. Of particular note is the observation that for the PDEB1 enzyme family fewer fragments with potent Ki values are obtained. It therefore appears that the active site structures of the PDEB1 enzymes may be more restricted or less flexible than the other two PDEs, resulting in fewer fragments gaining access to the cAMP binding pocket.

Cheminformatics Analysis of Active Fragments
The active compounds from the fragment screen of the IOTA library were subjected to cheminformatics analysis, and the following broad observations are made:
1. A few fragments were active against all six PDE enzymes, while around 20 more were active against four or five PDEs (see figure 5). Given that all the enzymes have related active sites which bind cyclic AMP this was an expected observation. However, quite a few compounds inhibited only one or two of the enzymes, particularly SmPDE 4A, TcrPDEC and LmPDEB1, which are the enzymes with the highest hit rates and therefore presumably the more “open” active sites.
2. Many of the hits fall into families of related chemical structure allowing the determination of structure-activity relationships. Six families containing at least four fragments each were identified as preferred fragments for the parasitic PDEs: alkylphthalamides, aromatic tricyclics and bicyclics, imidazoles, sulphones and aromatic alkenes.
3. The IOTA fragments had considerable physicochemical similarity to the fragments identified as PDE inhibitors in the ChEMBL database (see figure 5).
4. In addition, a search of the literature databases showed that some of these fragments were similar to known human PDE inhibitors, but a few compounds had much less similarity, having Tanimoto coefficients <0.7 against all known PDE inhibitors in the ChEMBL database.

High Throughput Compound Screening
In order to identify a larger range of novel PDE inhibitor structures, particularly larger molecules than the fragments in the IOTA library, the LANCE® assay was transferred to the European Lead Factory for screening of a trypanosome PDE against a library of 450,000 drug-like compounds. The assay was miniaturized to run in 1,536 well microtitre plates and was shown to be robust in terms of statistical quality and minimal interference from the test compounds. The primary screen was carried out at 10µM compound concentration. Based on a strict statistical selection criterion 5,731 compounds were determined to be hits, but a large number gave a weak effect and therefore 2,812 compounds were taken forward for Ki determination against the trypanosome PDE and human PDE4D as a selectivity counter-screen. This led to 67 compounds being identified as having a pKi >5.5 against trypanosome PDE, and >10-fold selectivity over hPDE4D (see Figure 6).

3.1.4 PDE4NPD toolbox
The homology between class I, eukaryotic, PDEs is as extensive as between human PDEs only, which implies that any parasitic PDE has a high homology with a human PDE. It also implies that parasitic PDE hit finding can be successfully performed with a collection of human PDE inhibitors. Therefore, PDE4NPD has assembled from their own libraries, commercial vendors and by re-synthesis, a series of human PDE inhibitors with sub-micromolar potencies against all of these enzymes with high chemical diversity. The latest version of this collection, PDE4NPD toolbox2.2 contains 51 individual compounds which are available for early pharmacological characterization of PDEs. Throughout the project, this compound toolbox was used for hit finding both on parasites, for the phenotypical approach and biochemically on purified PDEs for the structure-based approach. The PDE4NPD toolbox was also used for a parasite program outside the direct focus of the PDE4NPD project, Giardia lamblia, and resulted in low and submicromolar hits and the identification of a single PDE in G. lambia as promising drug target (Kunz et al. 2017).

3.1.5 PDE4NDP parasitology platform
Every PDE4NPD synthesized compound has been tested biochemically at the relevant parasitic PDE(s) of interest in a parasitology platform, localized at LMPH of the University of Antwerp (Belgium), Instituto Oswaldo Cruz (Rio de Janeiro, Brasil) and the Theodor Bilharz Research Institute (Cairo, Egypt).
Integrated in vitro screening at LMPH was pivotal for the primary phenotypic activity characterization of the many compounds that were part of the project and consisted of the following panel of assays: T. brucei (extracellular haemoflagellates), T. cruzi (intracellular amastigotes), L. infantum (intracellular amastigotes), P. falciparum (intracellular schizonts). To assess selectivity of action, cytotoxicity on MRC-5 (human fibroblasts) and primary peritoneal mouse macrophages (PMM) was assessed in parallel (see Table 6 for overview).
Stringent activity/selectivity criteria were adopted before a compound was advanced to secondary evaluation (at LMPH, FIO, TBRI). In total, 1004 compounds were screened in 32 screening sessions, including an independent repeat (total = 2008 assays in full dose-titration at least involving five 4-fold compound dilutions in the dose range 64 µM → 0.25 µM or lower). Several molecules were found active with different levels of activity as indicated in Table 7.
The next step for promising in vitro ‘hits’ was confirmation in the corresponding in vivo model in laboratory rodents (mostly mouse). Because oral dosing remains the preferred route of administration, in vitro phase-I and phase-II metabolic stability was assessed first using male mouse liver microsomes (S9). The threshold was at least 50% of parent compound remaining after 30 minutes. A few compounds falling just under this threshold were nevertheless advanced for in vivo evaluation, but then with co-administration of the non-specific CYP450 inhibitor ABT (1-aminobenzotriazole). In total, 34 compounds were evaluated for in vitro metabolic stability, information that was also used to steer the SAR.
Compounds fullfilling the acceptance criteria for metabolic stability were subsequently evaluated in the primary in vivo model aiming to maximize drug exposure: high dose level (50-100 mg/kg) in a daily dose regimen for 5 days starting very shortly after the experimental infection (= no head start for the infection) or after establishing parasitemia (in acute T.cruzi mouse model). Compounds showing significant activity were subsequently dose-titrated adopting the same dosing regimen.
- Sleeping sickness: T. brucei mouse model: parasitaemia at day 4 + survival
- Chagas disease: T.cruzi Y strain mouse model: parasitemia at day 8 + survival Leishmaniasis: L. infantum hamster or Balb/c mouse model
- Malaria: P. berghei mouse model: parasitaemia at day 4 + survival

Chagas disease
In depth, phenotypic studies on T. cruzi were performed at Fiocruz using a flow chart based on the Target Product Profile for Chagas Disease (see http://www.dndi.org/; Romanha et al., 2010, Guedes-da-Silva et al., 2016). The 75 molecules (including pyrazolones, phthalazinones, isocyanoethyl aryl ethers, imidazoles among others compound classes) were selected based on a primary screening conducted at LMPH. At Fiocruz, their potency was screened on different parasite forms and strains that belong to distinct DTUs that are relevant for human infection (as DTUs VI, I and II), besides evaluating the effect on some strains that are naturally resistant to nitro-derivatives, such as Y and Colombiana. Cardiac cell cultures and fibroblast cell line L929 were used for the cytotoxic analysis by the colorimetric assays Presto-Blue and Alamar-Blue, respectively, and thus for determination of the selectivity indexes. The proof-of-concept regarding anti-parasitic efficacy and preliminary drug safety of the most promising NPD compounds were performed using male Swiss mice (18-20g) subjected to (i) compound escalating schemes (for acute toxicity and exclusion of toxic concentrations) followed (ii) by NPD administration of non-toxic doses in acute T. cruzi infected-animals (exploring parasitological and clinical parameters). For each compound, 2-4 experiments were run (duplicate/triplicate for in vitro and n ≥5 for in vivo) including the reference drug benznidazole (Bz at optimal dose) in parallel
The findings on intracellular forms of Tulahuen transfected β-galactose strain (96 h of compound incubation, DTU VI) showed that 25 out of 75 molecules presented anti-T.cruzi activity ≥ as Bz (2.7 ± 0.4 µM), exhibiting EC50 values ranging 0.17-3.4 µM, being the most effective, (NDP228) about 17-fold more potent than Bz. Most of those active inhibitors presented low toxicity towards mammalian cells (>200 µM), and 19 displayed comparable or higher selectivity index (SI = 48-599) than Bz (SI = 51).

Schistosomiasis
Phenotypic studies on S. mansoni with selected compounds were performed at TBRI. The selection of these compounds was based on previous screening conducted at LMPH, which include cytotoxicity on MRC-5 cells. Only non-toxic compounds from this assay were tested for their potential antischistosomal activity. Promising compounds showing worm insult at 100 µM (as either worm killing or sluggish worm movement, worm uncoupling, and oviposition in vitro) were re-examined at lower concentrations to verify the reproducibility of worm killing, as well as to test the effects of these compounds on worm coupling and ovipositing capacity. Moreover, compounds showing >50% killing of adult mature schistosomes (6 weeks old) at 100 µM were further examined against different schistosome maturity stages, early mature (4 weeks old) and immature (3 weeks old) and in combination. The potency relative to positive control praziquantel (PZQ) was assessed using EC50 calculations.

Based on in vitro findings, the most promising compounds were selected for in vivo studies using S. mansoni infected mice. Compounds were orally administered in a dose of 20 mg/kg/day for 5 days and the percentages of worm reduction, egg developmental stages and load of eggs in hepatic and intestinal tissues were determined. Moreover, administration of selected compounds in combination with PZQ in a dose of 10 mg/kg/day for 5 days was also planned to check synergy with PZQ.

3.2. PDE4NPD parasite programs
The various PDE4NPD platforms have been used within the PDE4NPD program to evaluate new chemical matter as PDE inhibitors and/or phenotypic hits for the four parasite prioritized parasite programs. Below developments in each of the parasite programs is detailed.


3.2.1. T. brucei
3.2.1.1 PDE expression analysis
The expression levels of all T. brucei PDEs were assessed in standard long-slender bloodstream forms of strain Lister 427 and in procyclic (insect-stage) forms of the same strain. As shown, in the figure 7, TbrPDE-B1 and TbrPDE-B2 were expressed to the highest level, particularly in bloodstream forms, consistent with these being known to be the most crucial and essential phosphodiesterases of T. brucei. TbrPDE-A and TbrPDE-C, known not to be drug targets were expressed at a very low level. In bloodstream forms, PDE-D now know also to be essential and a functional cAMP PDE (see section 3.1.1) was expressed at an intermediate level. Interestingly, TbrPDE-D, like TbPDE-B1 was mostly localized to the parasite’s flagellum, which seems the essential compartment for PDE activity and thus for cAMP signaling in this species.

3.2.1.2 Target-based approach
TbrPDEB1 has been one of the main targets in PDE4NPD in view of the earlier proof-of-concept of TbrPDEB1/B2 inhibition as therapeutic approach for human African Sleeping Sickness. Moreover, a number of effective assays (biochemical, x-ray, SPR) have been developed within PDE4NPD to support the target-based drug discovery effort. At the start of the project PDE4NDP already had access to some chemical series that could inhibit TbrPDEB1. Within PDE4NPD it was decided to focus on two main, target-based research lines; 1) obtaining a TbrPDEB1/B2 selective inhibitor for human application, focusing on selectivity over the human off-target hPDE4 and 2) obtaining a proof-of-concept compound to be used in animal models of trypanosomiasis. Compounds studied in animal models should fulfill some properties such as good in vitro potency, metabolic stability, solubility, etc. In the first part of the project, we have obtained around 60 different compounds, with different structures trying at the same time to keep high TbrPDEB1 inhibition, a high in vitro potency against T. brucei and to improve the pharmacokinetic properties, with the ultimate aim to serve as a compound that can serve as a proof-of-concept compound. After all the different structural modifications and measurements of metabolic stability, a molecule showed good potency in the in vitro phenotypic assay on T. brucei. At the same time, this derivative presented good stability in mouse liver microsomes, which makes this compound suitable for further in vivo proof-of-concept studies.
The search for selective TbrPDEB1/B2 inhibitors has been heavily supported by the PDE4NPD protein expression and x-ray crystallography platform (see 3.1.2.2). The TbrPDEB1 and B2 x-ray structures have been solved in combination with a number of ligands. In comparison with the human PDEs, the parasitic PDEs have an additional pocket, P-pocket, just adjacent to the substrate binding site. This pocket can be exploited to obtain selectivity for the TbrPDEs over the off-target human PDE4. So far, all know TbrPDEB1 inhibitors, like e.g. NPD-001 have a much higher potency at human PDE4 and this precludes further development. With the various co-crystals at hand PDE4NPD embarked on a structure-based approach to selective TbrPDEB1 inhibitors. A first series of NDP-analogs was synthesised with the idea to make the molecule more rigid in order to allow a side chain to go into the P-pocket. To that end, various substituted biphenyl linkers were combined with the phtalizone core scaffold of NPD-001, resulting in a series of potent (Ki = 100 nM) and at least 10-fold selective for human PDE4 (Blaazer et al., 2018). X-ray crystallography studies show that these compounds indeed target the P-pocket, providing clear evidence that targeting of the parasite-specific subpocket is These compounds also increase the levels of cAMP in the parasites, but they are however not very effective against the T.brucei parasites, most likely by a lack of reduced penetration into the parasites. The PDE4NPD chemists therefore continued to develop SAR of the scaffold and tried various options to replace the rigid biphenyl spacer. Many of these replacements were not effective and lead to reduced activity at the enzyme, by a new series of alkynamides again combined P-pocket targeting (as prove by x-ray crystallography) with selective inhibition of TbrPDEB1 and B2 over human PDE4. Fortunately, these molecules also show nice phenotypic activity against T.brucei and this new class of molecules will be investigated further.
Moreover, the TbrPDE inhibitor program will be continues after the PDE4NPD project and can build upon the qualified hit list of the European Lead Factory (expected May 2018) and the current X-CHEM x-ray fragment screen. Both approaches will most likely offer new scaffolds to target in PDE4NPD.
3.2.1.3 Phenotypic approach
Despite the fact that TbrPDEB1 and B2 have been receiving much attention as target for T.brucei targeting, the PDE4NPD program has always closely monitored the phenotypic screening results from LMPH. As can be seen in Table 7, the phenotypic screening revealed 46 good hits. From these hits, NPD-2975 showed a most interesting profile with no toxicity towards human cells and an IC50 value of 100nM. To develop some clear SAR a number of analogs have been prepared and some derivatives show similar good phenotypic activities. As NPD-2975 is based on a scaffold of known PDE-inhibitors, the compound has been tested as inhibitor for both TbrPDEB1 and the newly characterized TbrPDED, but at neither of the enzymes inhibitory activities could be identified. Moreover, the phenotypic hit does not affect HERG or CYP enzymes, does not interact with any proteins in a Safety panel of diverse enzymes and receptors, has proper ADMET properties and is negative in an Ames test. In vivo testing of NPD-2975 led to very encouraging results in an acute infection model in mice. Infection could be completely blocked at 50 mg/kg, b.i.d. p.o and rescued animals were not infective anymore after blood transfer to naïve mice. Currently, NPD-2975 is evaluated in different in vivo models, whereas at the same time the SAR of this chemical series is further developed.

3.2.2. T. cruzi program
3.2.2.1 PDE expression analysis
Expression analysis for all T. cruzi PDE genes was performed using cDNA obtained by quantitative reverse transcriptase PCR (qRT-PCR) on mRNA from amastigotes (intracellular forms collected from the cardiac cell cultures supernatant), trypomastigotes (bloodstream forms) and epimastigotes (insect stage, axenic cultures) of Y strain. In the chart shown, the expression is normalised to TcrPDE-C. It is clear that, unlike T. brucei, all TcrPDEs are in fact quite highly expressed, especially in the medically-important amastigotes and bloodstream forms. In trypomastigotes and epimastigotes, TcrPDE-A is substantially under-represented, however, and this may thus have a particular function in the intracellular communication pathways. The expression profile did not give a clear indication as to which TcrPDE might be essential in these parasites (Figure 8).

3.2.2.2 Target-based approach
As mentioned, Trypanosoma cruzi possesses the genes for 5 different PDEs, TcrPDEA, TcrPDEB1, TcrPDEB2, TcrPDEC and TcrPDED. At the start of the project target validation was proposed for TcrPDEC (Antimicrob Agents chemotherapy, 2010, 3738). Screening of the PDE4NPD toolbox (see section 3.1.4) for inhibitors for TcrPDEC resulted in the identification of several nanomolar hits. Chemical derivatization of these hits resulted in a set of compounds with variations in Ki values from >10 µM till 10 nM. However, no correlation between inhibition of the TcrPDEC enzyme and phenotypic effects on the growth of the parasite could be observed (Figure 9), which does clearly not confirm the hypothesis that inhibition of TcrPDEC is a valid target for treatment of Chagas Disease.
Next, we did the same with TcrPDEB1 as described for TcrPDEC: screening compounds on the enzyme and trying to correlate these activities with their effect on the parasite. Also, in this case, no correlation between these 2 data sets was observed (Figure 9) and this deprioritizes TcrPDEB1 as a therapeutic target as well.
In T. brucei, the combined inhibition of TbrPDEB1 and TbrPDEB2 is validated as therapeutic target (Oberholzer et al. 2007Mar;21(3):720-31). It remains to be investigated if the same holds for the PDEB1 and PDEB2 from T.cruzi. Several compounds with nanomolar activity against TcrPDEB1 are equipotent on TcrPDEC, while being inactive in the phenotypic assay, indicating that the combined inhibition of TcrPDEB1 and TcrPDEC is most likely also not a therapeutic target.
In conclusion, we invalidated TcrPDEB1 and TcrPDEC, both in monotherapy as well in combination, as therapeutic target for the treatment of Chagas Disease. Potentially, one of the other TcrPDEs or a combination may prove to be a validated approach, but this remains open for the moment.

3.2.2.3 Phenotypic approach
In an alternative approach, we tested a series of human PDE inhibitors on T.cruzi in vitro, which yielded 3 different hit series, imidazoles, phenylpyrazolones and phenylphthalazinones with low- and submicromolar activity on the parasite. Optimization of the imidazoles, several of which showed low micromolar activity on both the intracellular form and the bloodstream trypamastigotes (BT), is ongoing. Stepwise optimization of the phenylpyrazolones resulted in NPD-227, which showed submicromolar activity against different T.cruzi strains, and life cycle forms (Table 8). This compound was tested on TcrPDEB1, TcrPDEC and several human PDEs but shows no significant activity on any of them. Unfortunately, NPD-227 and all its analogues are inactive against the BT form of the parasite. This finding and the poor metabolic stability precludes further development of this particular compound. Still, we see NPD-227 as a valuable starting point for further optimization, which is ongoing, and as a tool for target finding and target evaluation. In vivo experiments (infected mouse) with this compound, in the presence of aminobenztriazole to block liver metabolism, showed no effect on parasitemia, most likely due to its inactivity on the BT form of the parasite. Combination of NPD-227 with benznidazole (Bz) in vivo, resulted in a significant improved increase in survival of the infected animals, compared with Bz alone. In the phenylphthalazinones series, several compounds showed activity (~10 μM) against both the intracellular and the BT form, comparable with Bz. One compound from this series, NPD-040, showed, in combination with Bz, potential synergism (xΣ FICI 0.58). Further experiments with more potent analogs are needed to confirm this finding.
Pilot experiments with high dose of selected compounds from the 3 hit series showed, on exposure, an increase in intracellular cAMP concentration, indicating that a PDE might be involved in the mode of action of these compounds. Further experiments are needed to validate this MoA.

3.2.3. Leishmania program
3.2.3.1 PDE expression analysis
The expression pattern of L. mexicana PDEs was obtained using qRT-PCR on mRNA from an in-house strain. The clinically important parasite species is the amastigote, which resides in the host macrophage. Figure 10 also shows the PDE expression in the promastigote insect stages, normalised to the level of LmexPDE-B1. In both stages, LmexPDE-A and LmexPDE-B1 were most highly expressed, which appears to indicate a greater role for PDE-A than is likely in the other kinetoplastids, and this warrants further investigation. For this species, PDEs C and D were least expressed.
Although much progress was made with PDE4NPD, including validation of on-target activity by observing elevated cAMP levels after treatment, and in vivo activity, it is still unclear which PDE(s) is/are the actual target(s) in this parasite species. It could be speculated that localisation in the flagellum is key, as shown for T. brucei and we therefore investigated the cellular localisation of PDE-B1 and PDE-B2 in L. infantum. Figure 12 shows that B1 localised entirely to the cell body, in distinct organelles, whereas B2 is exclusively located in the flagellum.
3.2.3.2 Target-based approach
As mentioned before, Leishmania species showed the presence of the genes for 5 different PDEs. At the start of the PDE4NPD project, there was no real target validation for any of them. In this project, we cloned all LmjPDEA, LmjPDEB1, LmjPDEB2, LmjPDEC and LmjPDED and developed assays for initially the LmjPDEBs. The enzyme LmjPDEB1 was crystallized by the group of Hengming Ke and proposed as therapeutic target, in analogy with the PDEs from the parasite T. brucei (Mol. Microbiol, 2007, 66, 1029-1038). Screening of our PDE4NPD compound toolbox (a chemically and pharmacologically divers set of PDE inhibitors) revealed compounds with low nanomolar inhibiting potencies against this enzyme. However, plotting the LmjPDEB1 inhibiting potency against their effect on the parasite shows no hint of a correlation (Figure 12), indicating that this enzyme might not be a therapeutic target. Still this is no proof as the correlation might be obscured by the physicochemical properties of compounds in this experiment. As this parasite lives inside vacuoles in macrophages, these physicochemical properties are very important for targeting. However, detailed analysis of the chemical properties in connection with their inhibiting activities on both the parasite and the enzyme indicates that there are no real physicochemical properties that can be correlated to the measured activities in the assays used. Hence, we conclude that LmjPDEB1 is most likely not a valid therapeutic target for the treatment of leishmaniasis.
In T. brucei, the combined inhibition of TbrPDEB1 and TbrPDEB2 is a validated therapeutic target (Oberholzer et al., 2007). In an experiment with 6 LmjPDEB inhibitors from different chemical classes, we found a correlation coefficient (R2) of 0.95 between LmjPDEB1 and LmjPDEB2 inhibition, indicating the high homology between these 2 enzymes. Therefore, this finding of very high homology makes it very unlikely that combined inhibition of LmjPDEB1 and LmjPDEB2 might be a therapeutic target as well.
Next, we investigated LmjPDED as potential therapeutic target. From the PDE4NPD toolbox screening 5 compounds emerged as potent LmPDED inhibitors with pKi values of 9.1 6.3 6.3 6.7 and 6.9 respectively. All these 5 compounds are virtually inactive when tested on growth of the parasite (pIC50 of 4.5 or below), which also indicates that LmjPDED is most likely not a therapeutic target. Moreover, 4 of these 5 compounds are also potent LmjPDEB1 inhibitors, which even disqualifies the combined inhibition of LmjPDEB1, LmjPDEB2 and LmjPDED as therapeutic target.
In conclusion, research in the PDE4NPD consortium has devaluated the enzymes LmjPDEB1, LmjPDEB2 and LmjPDED as therapeutic target. In addition, the combined inhibition of these 3 enzymes has no effect on the parasite. Whether LmjPDEA or LmjPDEC might be useful as therapeutic target still needs to be investigated. Moreover, unpublished results from the consortium (exposure to PDE inhibitors resulted in a strong increase in extracellular cAMP concentration) points to export as an additional method by which this parasite might control its intracellular concentration of cAMP. As the Leishmania parasite seems to have the means to export excessive cAMP, it might well be that PDE inhibition has no potential as therapeutic target in Leishmaniasis, unless cAMP transport is inhibited simultaneously.
3.2.3.3 Phenotypic approach
Due to the set-up of the project, parallel biochemical screening for inhibitors for parasitic PDEs next to phenotypic evaluation of all PDE4NPD compounds against the parasite, during the project phenotypic hits emerged which showed no significant activity on any parasitic PDE that was tested. The phenotypic hits active against Leishmania belong to 2 different chemical classes, imidazoles and phenylpyridazinones. From each chemical class 1 compound, NPD-311 and NPD1168, was selected for in vivo evaluation based on their in vitro ADME profile and low in vitro toxicity. NPD-311 was tested orally 50 mg/kg b.i.d. for 5 days in the L.infantum infected mouse model, after which reductions of parasitemia of 36% in the liver and 57% in the spleen was observed. NPD-1168 is metabolically not stable enough for in vivo testing. To circumvent this problem and to get a proof of concept, the compound was tested in an L.infantum infected mouse which was also treated with aminobenzotriazole, which inactivates the metabolic enzymes in the liver. Using this method, treatment with NPD-1168, 50 mg/kg b.i.d. for 5 days resulted in parasitaemia reductions of 63% in the liver and 79% in the spleen. Though significant effects were observed with both compounds in the L.infantum infected mouse, the observed reduction in parasite are not considered good enough to warrant further development of these particular compounds. Still, as proof of concept was obtained in an in vivo model for efficacy, optimization in these chemical classes is continued with focus on solubility and metabolic stability.
3.2.4. S.mansoni program
The S. mansoni program was the least advanced of the four parasite-centred projects at the start of the PDE4NPD consortium, because no knowledge existed about the expression and/or role of SmPDEs; indeed, the sequences of the ORFs had not been verified beyond the published genomic sequence, showing ten potential PDE genes. The main objective of PDE4NPD was to validate (or discard) putative S. mansoni PDE enzymes as drug targets via both gene cloning and phenotypic screening of a set of PDE4NPD inhibitors. PDE4NPD activities started with some computational work in order to unravel the structural features of these parasitic PDEs. A homology model was built and a number of potential hits selected after carried out virtual screening (VS) with this model.

From the published genome sequence, a set of 10 potential PDE genes was initially identified, recently expanded with one additional ORF that was missed in the original annotation. Amplification from cDNA from the TBRI strain of S. mansoni was successful for 9 of these genes, and their sequences aligned with the well-characterised human PDEs; names were allocated based on phylogenetic analysis (Figure 13). In the phylogenetic tree all SmPDEs were verified by us except SmPDE4C and SmPDE2, which could not be amplified from cDNA generated from the adult or immature worms, or from schistosomulae, despite very substantial effort, leading to the conclusion that they are not expressed in these life cycle stages. Most of the cloned genes displayed significant differences with the published genome, particularly regarding intro-exon boundaries. Indeed, two of them (SmPDE9C and SmPDE8) had unique inserts in their catalytic domains of 170 and 31 amino acids, respectively. The cloned genes were resynthesized in a main-stream codon preference for expression in insect cells (protein production), yeast (functional complementation) and T. brucei (complementation and inhibitor studies). Expression studies using quantitative Reverse Transcriptase PCR showed higher expression in male worms and similar levels in mature and immature worms. Highest expression was for PDEs 4A, 4B, 9C and 11. Expression in T. brucei was successful for all cloned SmPDEs, with so far functional complementation demonstrated and used for pharmacological assessment for SmPDE4A. In the yeast system, functional complementation with SmPDEs 1, 4A, 8, 9A and 11 has been achieved; SmPDEs 4C and 9B poorly complemented and may be specific for cGMP rather than cAMP.
As cloning and expression of different SmPDEs have been successfully developed, biochemical screening of some of them has become available after the enzymatic assay set-up, with prioritisation of SmPDE4A, for which the crystal structure was also solved. This work has allowed the evaluation of the top ranked virtual screening hits (hit rate 3/23) and also the PDE-Toolbox screening (hit rate 8/55).
PDE4NPD activities initially focused strongly on phenotypic screening of the compounds developed by the consortium that were non-toxic in the MRC-5 cellular assay. Selected compounds were assessed for killing of adult/early and immature S. mansoni worms, worm coupling and ovipositing in vitro. In vitro studies in one or two repeat experiments at concentration of 100 µM revealed potential antischistosomal activities against adult mature schistosomes, expressed as worm killing/and or sluggish worm movement, unpairing and absence or reduction in egg number for 64% (188/294). 8% (15/188) revealed worm killing > 50%, sluggish worm movement for the survivors with worm unpairing and complete absence of eggs. 16% (30/188) showed worm killing < 50% with uncoupling and absence of eggs, 14% (26/188) showing no worm killing with intact couples revealed complete absence of eggs while 30% (56/188) showed no worm killing, unpairing with absence of eggs. Reduction in egg number ˃ 25% at 100 and 50 µM was recorded despite the presence of intact couples for 60 out 188 (32%) compounds.
From the total of 294 compounds phenotypically evaluated against S. mansoni whole worms, 4 chemical classes were identified for their potential against schistosomiasis. This allowed us to develop specific medicinal chemistry programs around them. Interestingly, most compounds have substantially stronger effects on males than on females and display also a much stronger effect on the production of eggs, early mature and immature parasite stages - the worrisome stages of schistosomiasis - than on worm viability.
Based on the in vitro findings, 4 compounds with remarkable effects on the reduction of egg numbers and worm unpairing were selected to be tested in vivo using S. mansoni-infected animals. Consistent with this result, the most prominent and reproducible activity against schistosomiasis observed in mice receiving PDE4NPD compounds (10 mg/Kg/day for 5 days) together with a low dose of PZQ, was on the developmental stages of the ova: observations included a significant decrease to complete absence of immature eggs, a decrease in mature eggs, and an increase in dead eggs for 3 of the 4 compounds tested. It should be noted that the dose of PDE inhibitors used in vivo was relatively low as we were guided by the dose that showed promising effects on unicellular parasites (trypanosomes and Leishmania); Yet, S. mansoni is a compex multicellular parasite and the doses might not yet be optimal. The effect on ovipositioning deserves further studies because, despite the limited effects on worm viability, the complete absence of eggs was recorded even in the presence of living intact couples – and the morbidity of schistosomiasis rely mainly on the schistosome eggs.


Potential Impact:
4. Impact and main dissemination activities/exploitation of results

4.1 Potential impact
Worldwide, more than 1.4 billion people suffer from a neglected parasitic disease (NPD), with consequences that can include blindness, disfigurement, long-term disability or death. Safe and effective treatments remain unavailable for most NPDs.
Funding from the EU has enabled the PDE4NPD consortium to establish an international network to discover and evaluate novel treatments for NPDs, while at the same time building a European infrastructure to support international efforts in this and similar therapy fields (see www.pde4npd.eu for further details).
Specifically, the Consortium has developed strong ties with centres of excellence in neglected diseases, spanning Africa, the United States and South America. Many of these collaborations will continue after the PDE4NPD programme has ended. For example, the collaboration on the PDEs with the group led by Professor Mike Pollastri at North-Eastern University in Boston has been particularly productive and will be used as the basis of ongoing resaetch in NPD medicinal chemistry and chemical biology. Mike is an opinion leader in rare tropical disease research, concentrating on trypanosomiasis (for a recent review of his research, see: https://news.northeastern.edu/2018/01/10/why-you-should-care-about-rare-tropical-diseases/).
Other nodes within the PDE4NPD network include our continuing connection with the GSK R&D campus at Tres Cantos in Spain (http://uk.gsk.com/en-gb/research/sharing-our-research/open-innovation/). The Open Innovation approach which we have adopted alongside GSK has given us the opportunity to expand, exchange and evaluate our compound collections in a wide variety of neglected diseases of the developing world.
Similarly, strong European connections have been established with the high-throughput and high-content drug screening community, both within the project (at the Fraunhofer IME in Hamburg) as well as outside the project (through export of our PDE screens to the Innovative Medicines Initiative: https://www.imi.europa.eu/). The latter screens promise to considerably enhance the scope and quality of our own in-house lead chemistry efforts and will continue to be the focus of ongoing research programmes beyond the end of the PDE4NPD Project.
One of the major successes and potential impacts of our work has been to validate our compounds as a “PDE Toolbox”: a set of tools and reagents to support further work on parasite PDEs as therapeutic targets. This includes the gene constructs encoding the parasite PDEs, the proteins encoded by them, and the biochemical and biophysical assays measuring their activities, as well as a range of chemical inhibitors, summarised in several recent publications (see for example the description of the PDE structural biology platform PDEStrIAn: https://www.ncbi.nlm.nih.gov/pubmed/26908025).
Results from this work have been published as a series of scientific papers by the Consortium (see below), and the tools themselves are also available as a set of physical reagents from the PDE4NPD consortium leaders, the VU University in Amsterdam. Importantly, these reagents have been exchanged with scientific leaders such as the Medicines for Malaria Venture, to become components of their Pathogen Box (https://www.pathogenbox.org/).
In summary, the PDE4NPD project has not only established an extensive platform through which to address NPDs but has also trained a future work force skilled in the application of drug discovery techniques to this and other therapeutic areas, an R&D program that has laid the foundation for future therapeutic initiatives.

4.2 Dissemination of results
To ensure optimal dissemination of the data generated, there has been a strong emphasis on developing an open innovation model for data sharing.
Starting from a communication strategy and plan, communication and dissemination activities were performed throughout all 4 years of the project.

4.2.1 Internal communication
For internal communication, we established a PDE4NPD Project Workspace (named “TI Plaza”), which was established in 2014, and which has (in 2016) been upgraded to “MyProjectPlaza”, a Microsoft Exchange Platform that offers many more possibilities for e.g. data storing and sharing and has been used until 2018 to optimize communication within the consortium.
A total of 5 PDE4NPD newsletters have been issued, informing PDE4NPD partners and SAB members on e.g. project progress, upcoming events and deliverables. Analytics indicated the the newsletters were well read.
PDE4NPD Poster and a PowerPoint templates were prepared (A0) for project members to use when presenting data from the PDE4NPD project. These were updated when necessary (change of partners etc.).
For the PDE4NPD young researchers, a total of 4 Workshops were organized, that were held during the Consortium Meetings. The themes were:
- 19 May 2015: Team building
- 6 November 2015: Social Media
- 3 June 2016: Research challenges
- 18 November 2016: Writing a good publication
4.2.1 External communication and dissemination
The Public PDE4NPD website, at www.pde4npd.eu went online early 2014, and was upgraded in 2016. It has been visited well: since December 2016, the new website has been visited by 1739 new users from all over the world (Figure 14). A PDE4NPD Project Brochure has been developed in 2014 and updated in 2016. The Brochure was distributed at many international conferences. Project results were frequently disseminated, both through oral and poster presentations. An overview of poster presentations at NPD conferences is provided on the PDE4NPD website. An overview of project publications is also available on the PDE4NPD website. More publications are currently being prepared or have already been submitted.
At Vrije Universiteit Amsterdam the project has been used very often as context for Bsc and Msc teaching in the programs of Pharmaceutical Sciences (Bsc, 80 students/yr), Science, Business & Innovation (Bsc, 80 students/yr), Biomolecular Sciences (Msc, 40 students/yr) and Drug Discovery & Safety (Msc, 40 students/yr). To exemplify, 2nd year Bsc students were asked for a practical course Organic Chemistry to each synthesize a specific analog of an active PDE4NPD compound, resulting in a large library of analogs quickly and a very motivated cohort of students.

4.3 Exploitation
4.3.1 Academic publication is the main route to scientific exploitation
Under our Open Innovation franchise, all the results of the Consortium have been or will be freely provided to academic collaborators in the scientific community through immediate publication. Little of the work has been deemed to be of sufficient commercial value for either patenting or commercial licensing.
4.3.2 Commercial benefits
Despite the open innovation nature of the project, this is not to say that there have been no commercial benefits from the program. At the beginning of the PDE4NPD program, there were two SME collaborators:
• European Screening Port (ESP) in Hamburg (Germany)
• IOTA Pharmaceuticals in Cambridge (UK)
Both companies have derived considerable benefit as participants in PDE4NPD, ESP through the further development of its capabilities in high-throughput and fragment-based screening, and IOTA through the optimization of its Fragment-based Drug Discovery platform and development of new PDE-centric human drug discovery initiatives.
Part-way through the program, ESP was effectively returned to the public domain by absorption into the German Fraunhofer Institute for Molecular Biology and Applied Ecology IME (https://www.ime.fraunhofer.de/en/presse_medien/fraunhofer-ime-sp.html) within which its mission is now to bridge the gap between basic academic research and the life sciences industry in the field of drug discovery and development, a role well-exemplified within the PDE4NPD program. Such collaborative activities have played a major role in translating Fraunhofer IME SP into a world-leading contract research organization for small-molecule screening, particularly for targets identified by academic partners.
Meanwhile, IOTA has shifted its emphasis away from parasite drug discovery towards the discovery of human therapeutics, exemplified through participation in the recently announced WINDOW Consortium for the discovery of drugs to cure the brain cancer glioblastoma (see http://www.iotapharma.com/news-details.aspx?id=20). PDEs are considered as potential targets in glioblastoma, and the experience with this gene family acquired during PDE4NPD has enabled IOTA to build a PDE-centric component within its rapidly evolving anti-cancer drug portfolio.

List of Websites:
www.pde4npd.eu

References
Blaazer et al 2018, Targeting a subpocket in Trypanosoma brucei phosphodiesterase B1 (TbrPDEB1) enables the structure-based discovery of selective inhibitors with trypanocidal activity, J Med Chem. 2018 Apr 19. doi: 10.1021/acs.jmedchem.7b01670

Guedes-da-Silva et al. 2016, Antitrypanosomal Activity of Sterol 14α-Demethylase (CYP51) Inhibitors VNI and VFV in the Swiss Mouse Models of Chagas Disease Induced by the Trypanosoma cruzi Y Strain, Antimicrob Agents Chemother, Accepted manuscript posted online 6 February 2017, doi:10.1128/AAC.02098-16 Antimicrob. Agents Chemother. April 2017 vol. 61 no. 4 e02098-16

Keller et al. 2010, Chemical Validation of Phosphodiesterase C as a Chemotherapeutic Target in Trypanosoma cruzi, the Etiological Agent of Chagas' Disease, Antimicrob. Agents Chemother. September 2010 54:3738-3745, Accepted manuscript posted online 12 July 2010 , doi:10.1128/AAC.00313-10

Kunz et al. 2004, TbPDE1, a novel class I phosphodiesterase of Trypanosoma brucei, Eur J Biochem. 2004 Feb;271(3):637-47

Kunz et al. 2017, The Single Cyclic Nucleotide-Specific Phosphodiesterase of the Intestinal Parasite Giardia Lamblia Represents a Potential Drug Target, PLoS Negl Trop Dis 11(9): e0005891. 2017 Sep 15 DOI: 10.1371/journal.pntd.0005891

Oberholzer et al 2007, The Trypanosoma brucei cAMP phosphodiesterases TbrPDEB1 and TbrPDEB2: flagellar enzymes that are essential for parasite virulence, The FASEB Journal. 2007 Mar;21(3):720-31. Epub 2006 Dec 13. https://doi.org/10.1096/fj.06-6818com
Romanha et al. 2010, In vitro and in vivo experimental models for drug screening and development for Chagas disease, Memórias do Instituto Oswaldo Cruz, March 2010, vol 105(2): 233-238, March 2010, http://dx.doi.org/10.1590/S0074-02762010000200022

Wang et al 2007, Crystal structure of the Leishmania major phosphodiesterase LmjPDEB1 and insight into the design of the parasite-selective inhibitors, Mol Microbiol. 2007 Nov;
66(4):1029-38. Epub 2007 Oct 17., DOI: https://doi.org/10.1111/j.1365-2958.2007.05976.x