Crystallographic approach to fragment-based hit discovery against Schistosoma mansoni purine nucleoside phosphorylase

Several Schistosoma species cause Schistosomiasis, an endemic disease in 78 countries that is ranked second amongst the parasitic diseases in terms of its socioeconomic impact and human health importance. The drug recommended for treatment by the WHO is praziquantel (PZQ), but there are concerns associated with PZQ, such as the lack of information about its exact mechanism of action, its high price, its effectiveness – which is limited to the parasite’s adult form – and reports of resistance. The parasites lack the de novo purine pathway, rendering them dependent on the purine salvage pathway or host purine bases for nucleotide synthesis. Thus, the Schistosoma purine salvage pathway is an attractive target for the development of necessary and selective new drugs. In this study, the purine nucleotide phosphorylase II (PNP2), a new isoform of PNP1, was submitted to a high-throughput fragment-based hit discovery using a crystallographic screening strategy. PNP2 was crystallized and crystals were soaked with 827 fragments, a subset of the Maybridge 1000 library. X-ray diffraction data was collected and structures were solved. Out of 827-screened fragments we have obtained a total of 19 fragments that show binding to PNP2. 14 of these fragments bind to the active site of PNP2, while five were observed in three other sites. Here we present the first fragment screening against PNP2. D ow naded rom http://pndpress.com /bchem j/article-oi/10.1042/BC J20210041/7/bcj-2021-0041.pdf by gest on 28 Sptem er 2021 Bchem al Jornal. This is an Acepted M ancript. ou re encuraged to se he Vrsion of R eord tat, w en puished, w ill relace his vesion. he m st up-tote-version is avilable at https://drg/10.1042/BC J210041


INTRODUCTION
Schistosoma mansoni is a parasitic trematode that causes the common intravascular infection schistosomiasis (1). Amongst the parasitic diseases, schistosomiasis ranks second in terms of social and economic impact, and public health importance (2). According to the WHO, schistosomiasis has been reported in 78 countries. In 2013 more than 40 million people were treated for schistosomiasis, 261 million required preventive treatment, and nearly 700 million are at risk of infection (1,3). This burden of infection makes schistosomiasis a major health problem, particularly in developing countries.
The Schistosoma genome project was established in 1992 to improve the understanding of Schistosoma biology, with a focus on the characterization of new genes, the discovery and development of new drug targets and vaccines, and the determination of mechanisms of drug resistance and antigenic variation that enable evasion of the host's immune system (4). The S. mansoni 363 megabases (MB) nuclear genome was published in 2009, and a new 364.5 MB version was made available in 2012. This genomic information revealed a total of 10,852 genes encoding over 11,000 proteins, 45 % of which remain without known or predicted function (5)(6)(7). The breakthrough in the availability of genomic data allows new opportunities for innovation in the control of schistosomiasis. These data offer a new pipeline for the identification of novel drug targets and vaccine candidates through a system-wide perspective (6,8,9).
Schistosoma lacks the capacity for de novo synthesis of purine nucleosides, and is dependent exclusively on the salvage pathway for their purine requirements (10)(11)(12). Accordingly, S. mansoni acquire purine nucleosides from the host via a purine salvage pathway; this brings attention to the enzymes of the S. mansoni salvage pathway as potential drug targets for novel chemotherapy. In the past, the use of purine and purine nucleoside analogues have been successfully exploited as drug targets against several other parasites (13)(14)(15). It has been shown Downloaded from http://portlandpress.com/biochemj/article-pdf/doi/10.1042/BCJ20210041/920687/bcj-2021-0041.pdf by guest on 28 September 2021 that inhibition of a salvage pathway enzyme, Hypoxanthine-guanine phosphoribosyltransferase (HGPRTase), via siRNA decreased the viability of Schistosoma (16). Therefore, the Schistosoma purine salvage pathway is a suitable target for the development of novel compounds for the combatting of schistosomiasis.
Purine nucleoside phosphorylase 1 (PNP1) is a member of the purine salvage pathway of S.
mansoni, and we reported its crystal structure in 2003 (10,17). PNP1 is involved in the conversion of the purine salvage pathway intermediate inosine to hypoxanthine (10). PNP1 catalyses the cleavage of the glycosidic bond in purine ribo-and deoxyribonucleosides, in the presence of inorganic phosphate as a second substrate generating a purine base and ribose-1phosphate. PNP1 also has been sudied for discovery of inhibitors (17)(18)(19).
Purine nucleoside phosphorylase 2 (PNP2) (PDB ID 5CXQ) is a new isoform of PNP, which was identified in the S. mansoni genome-wide Open Reading Frame search. The new isoform, which presents a high degree of conservation to PNP1, has three substitutions in the active site, making its catalytic site very different from the PNP1 (20). PNP2 is highly expressed in the cercariae, where PNP1 has its lowest expression level (6,7), which makes it a suitable drug target.
In the past fifteen years fragment-based drug design has been established as an effective alternative to high throughput screening for the identification of hit compounds in drug discovery. Fragment based screening and optimization have attained reliable success in numerous drug discovery project with one approved drug in the market and several other compounds in clinical trials (21,22). This approach allows the study of the interactions of very simple molecules (fragments) with protein targets, providing useful information for drug design. Structural methods allow us to rapidly and effectively explore the complementarity between a protein active site and drug-like molecules via the use of fragments (23). Recent advancements in the synchrotron facilities with collection of greater amounts of high-quality diffraction data, and automation in the data analysis has increased the utility of fragment screening (23,24).
PNP-based drug design is an apposite target as it has been explored for other diseases (25), and there are no reports of a high throughput fragment screening for S. mansoni. The reproducibility of PNP2 crystals is highly consistent and this makes it a tempting target for fragment screening.
Here, we report an extensive crystal optimization and dimethyl sulfoxide (DMSO) tolerance test for PNP2 crystals, and a subsequent fragment-based screening of 827 compounds (a subset of the Maybridge fragment library). This resulted in 19 new PNP2 crystal structures with bound fragments. The majority of the fragments were observed in the active site, including a fragment that explores a previously unidentified pocket closer to α6-and α8-helices.
Furthermore, fragments were observed in three other binding sites. Our findings reveal a great deal of atomic-resolution structural information regarding the interaction of fragments with PNP2, and a validated methodology to improve crystal quality for fragment screening campaigns.

Protein Expression and Purification
The PNP2 gene (Smp_179110) was identified in the S. mansoni genome (5

Crystallization Screening
The purified protein was concentrated to 4.6 mg/mL in purificatication buffer (20 mM

Fragment library screening and data analysis
The purified protein was subjected to the fragment screening procedure (Figure 1). Crystals crystals were mounted and stored in liquid nitrogen. X-ray diffraction data was collected at beamline I04-1 of DLS and autoprocessed. Dimple and REFMAC were used for initial structure refinement (31) and AceDRG for generation of ligand coordinates and restraints (32).
PanDDA was utilized for hit identification on the dimple-processed maps (33); Coot was then used for fragment fitting and subsequent model (re-)building (34). Subsequent refinements were performed with PHENIX (35) and analyzed with coot. The final models were submitted to Protein Data Bank (PDB) and assigned with the codes listed in Table 3.

Crystal Selection
The protein PNP2 has crystallized in 33 different conditions out of 6 crystallization kits ( Table   1). The extensive range of crystallization conditions provided a broad range of options of crystal optimization for fragment screening. 9 different conditions were selected for data collection based on the size of the crystals (0.50 µm on average), and X-rays diffraction data was collected ( Table 1). The resolution of the collected diffraction data was in the range of 1.34 Å to 2.51 Å. The crystals belonged to two space groups, 6 conditions were assigned to the space group P 21 3 and 3 conditions were indexed as F41 3 2.  µm -120 µm were obtained in 30 % PEG 3350 concentration. Protein concentration did not affect the crystal size. Crystals were only obtained in crystallization drops with a protein-toreservoir ratio of 1:1, while precipitation was observed in 1:2 and 2:1 ratios. The highestresolution diffraction data was collected from crystals produced in PEG 3350 30 % concentration, which consistently produced larger crystals (50 µm -105 µm) ( Table 3). The resolution of the diffraction data was improved, in the range of 1.50 Å -2.50 Å.

Fragment library soaking and hit detection
It was observed that larger crystals generally diffracted to higher resolution. In the fragmentsoaking condition containing 30 % DMSO and 60 mM of the fragment, soaking was performed using 827 fragments; diffraction data was subsequently collected from the resulting crystals and autoprocessed, with a resolution range of 1.60 -2.50 Å. Fragment hits were identified with PanDDA using the default settings and visually analyzed in COOT [33,34]. A total of 19 fragments were identified bound to PNP2 ( Table 4). Table 4. PDB IDs, chemical names and structures of each fragment observed bound to PNP2.
The number of DMSO molecules built and resolution of each structure are also listed.

Fragment binding sites
Of the 19 observed binders, a total of 14 fragments were found in the active site, and the other five fragments were observed in three additional sites (Figure 2).  suggest that the affinity for this interaction is moderate. Several residues differ between PNP2 and PNP1 in this region, namely Q174K, K178Q and R283K, indicating that this fragment binding would likely not be conserved for PNP1.  The active site of PNP2 has been described in the literature (20).  Interestingly, only one fragment, PDB ID 6BJ6, occupies the sugar-binding site, and is observed to hydrogen bond to Ser35, Asn117, and the carboxyl oxygen of Leu118, in addition to two water molecules (Figure 7). In comparison to the other structures presented here, Ser35 and His259 are slightly displaced, and Asn117 adopts a different rotamer. The binding of the sulfinyl oxygen atom of fragment 18 to Asn117 and to the main chain of Leu118 causes a small perturbation in the β-strand β5.  mansoni and other species of the genus Schistosoma that parasitize humans. The exact mechanism of the praziquantel is still not elucidated (38). Praziquantel is a WHO-approved drug, but a major limitation to schistosomiasis control has been the limited availability of There are no reports of a structure-based fragment screening for S. mansoni, and this approach is effective in the identification of hits which can be developed into lead compounds. Here we report the first high-throughput fragment screening process for S. mansoni. The fragment screening strategy that we describe is the robust technology available at Diamond Light Source, UK, which demands a conveniently short amount of time (typically less than 30 days) to screen a large library of fragments against a protein (Figure 2). The methodology employs several critical processing techniques that facilitate the otherwise prohibitive experimental and analytical workload. The most important of these is the use of the ECHO liquid handler, targeted with TeXRank, to perform the soaking procedure; this greatly simplifies sample handling, reducing the time required to perform the experiment, but also increasing the accuracy of the soaking. The autoprocessing pipeline for data collection and structure solution at DLS I04-1, including the identification of fragment hits with PanDDA(33), is another backbone of the fragment-screening methodology. The processing time of data collection and analysis is reduced to days rather than months; in our case, we were able to identify structurebound fragments for all 827 crystallographic datasets in two days.