Yeast-based systems for tropical disease drug discovery

KEYWORDS:

• Yeast-based systems
• drug discovery
• high throughput screening
• Neglected Tropical Diseases
• Leishmania
• Trypanosoma
• Plasmodium
• pathogenic worms

1. Early drug discovery for neglected tropical diseases

The term ‘neglected tropical diseases’ (NTDs) is coined to identify a diverse group of communicable diseases caused by various infectious and parasitic agents that affect more than one billion people worldwide. They mainly prevail among people living in poverty, with poor sanitary conditions and in close contact with livestock. For most of these diseases, the therapeutic arsenal is limited, unsafe [or] difficult to manage, which hampers patient adherence to treatment. Frequent treatment discontinuation facilitates the transmission of the disease and the emergence of resistant strains. These facts fully justify the increasing interest shown by public and private initiatives to find new drugs against these scourges.

Massive screening of active compound libraries by high-throughput techniques (HTS) is currently a powerful methodology to identify new molecules against these diseases. Target-based (biochemical) and nontarget (phenotypic) approaches are valid in vitro methods to face this task, but they are based on different paradigms.[1] On the one hand, target-based screenings require the identification of a validated protein, which should be essential for pathogen’s metabolism. Once established, the protein should be purified – frequently in recombinant form – and crystallized in order to obtain 3D models through X-ray diffraction analyses. This approach enables in silico studies of chemical modifications of the hit compound that improve its binding to the target protein. After this step, selected hit compounds must be optimized either to facilitate their entry into the host cell or to increase their stability within the host. Unfortunately, this approach does not warranty that picked-up compounds can have more than one non-identified target.[2]

On the other hand, phenotypic screenings are cell-based assays that determine the ability of one compound either to kill or to prevent the proliferation of the pathogen. Phenotypic assays provide an unbiased approach to identify compounds that are active against the pathogen, but provide no information about the mechanism of action.[3] Furthermore, phenotypic screenings can be performed only against pathogens that can be cultured.[4] This requirement is particularly problematic when designing screenings for antiparasitic compounds, since the culture of parasites outside the animal host may be extremely difficult or even impossible. For example, there is no assay for the hypnozoite stage of Plasmodium vivax. This pathogen, which is responsible for malaria, cannot be continuously maintained in vitro cultures and consequently, assays for HTS with the parasite have not been developed.[5] This problem is even worse in case of metazoan parasites, such as nematode or trematode worms, due to the presence of different larval stages. During the last years, a plethora of HTS methods based on genetically modified cultivable parasites have been developed. They are aimed to assess the killing effect of compounds from libraries by means of easy optical readouts.[6] Among them, those that best resemble host/parasite interactions in vitro are preferred for drug screening.[7,8]

2. Approaches to target identification

The identification of molecular target[s] underlying the observed phenotypic response is useful for the following reasons: (1) to understand the biology of the disease, (2) to allow for rational chemical redesign during SAR studies, and (3) to predict toxicity and side effects of the studied molecules. For this purpose, the identification of putative molecular targets is based on the interaction affinity between selected hit compounds with cellular components, on the premise that physical interaction is the prerequisite to produce negative effects on the pathogen.

Target deconvolution is a complex of retrospective in vitro strategies designed to identify compound mechanism of action.[9,10] The identification of targets is decided on a case-by-case basis due to the lack of standardized procedures in use. Three distinct, but complementary approaches, are currently proposed: (1) direct biochemical methods, (2) genetic interaction methods, and (3) computational inference methods.

Direct biochemical methods involve labeling of the small molecule of interest, incubation with cellular components and direct detection of binding, usually following some type of stringent wash procedure. The direct way to identify targets within the pathogen is the affinity chromatography and further characterization of the small molecules immobilized in a matrix by mass spectrometry. The facts that identified proteins retain their intact tertiary structure and can be analyzed by X-ray diffraction after crystallization represent an important advantage of this method. However, immobilization and/or labeling of the drug may alter the interaction parameters with the target. This strategy has been useful to identify ATP-binding proteins that interact with the protein kinase inhibitor staurosporine in Leishmania donovani – a protozoan responsible for serious visceral disorders in the Old World.[11]

Genetic modifications of easy-to-manipulate organisms (such as the budding yeast Saccharomyces cerevisiae) can also be used to identify protein targets by modulating presumed targets in cells, thereby changing small-molecule sensitivity. The search for heterozygous deletions or silenced ORFs in libraries of those yeast strains with increased sensitivity to a selected lead compound, can be useful to identify their encoded target protein.[12] Genetically identified targets need further biochemical confirmation using rescue (add-back) experiments in order to exclude the possibility that mutations in other genes can produce similar effects on drug sensitivity.[13] Several advantages of this methodology include the unbiased approach and no requirement of compound tagging for immobilization. In addition, research on global changes in gene expression induced by the potential drug, permits the elaboration of ‘connectivity maps,’ which contribute to understand the mechanism of action using transcriptomics analysis of a specific proteomic profile. However, despite the use of a minimum living environment, several limitations have been indicated: (1) not all ORFs from pathogens are represented in yeast; (2) sensitivity to drug candidates does not respond to mRNA down-regulation; and (3) yeast models are hard to extrapolate to metazoan parasites like pathogenic helminths, thus recommending the use of alternative models, such as the free living nematode Caenorabdhitis elegans.[14]

The availability of genomic sequences for most of the potential targets in major parasites, as well as the crystal structure of some major targets, enable docking calculations of the interactive likelihood of a molecule with its putative target by computational inference. When 3D structures are available for the protein that has been identified, structural analysis should be performed in silico to confirm the protein–compound interaction. Despite this is an inexpensive growing-use approach, it only generates predictions that require mandatory experimental validation to confirm the direct drug/target interaction in vitro and their ultimate functional consequences in vivo. This approach is being exploited to develop families of potential inhibitors of robust targets in trypanosomas and leishmanias.[15]

3. Yeast-based systems for drug discovery

Yeast-based screening platforms were proposed as inexpensive powerful tools for the detection and characterization of drug interactions with molecular components in a contextualized biological environment. As a ‘minimum’ eukaryote that preserves the entire basal metabolic network, yeast represents the system of choice to elucidate mechanisms of drug action in early phases of drug discovery. Due to its easy genetic manipulation, yeast-deficient mutants can be complemented with genes encoding heterologous proteins, thus enabling automated, target-based screening within an axenic in-cell living environment. However, the foreign transfected target gene must be validated within the metabolic yeast context, growth conditions, level of expression, post-translational modifications, and, probably more important, in the absence of the host environment.

This approach may be useful to identify novel compounds that specifically inhibit the activity of foreign proteins pointed as targets for anti-parasite drugs. In addition, it provides an alternative to pathogens with unfeasible or hard-to-culture life-stage forms.[16] Nevertheless, in spite of the feasibility in gene manipulation and expression of foreign proteins, yeasts differ from their pathogenic counterparts in the composition of their plasma membrane. This peculiarity can compromise the results in further steps of drug discovery. Despite yeast cell wall does not limit permeability of druggable small molecules, efflux pumps are responsible for the relatively low intracellular concentrations of foreign compounds. In yeasts, pumping out processes can be dramatically reduced by deleting ABC transporters, such as the major multidrug export pump encoded by the Pdr5p gene.[17]

The pioneering work that applied this strategy in parasites for the first time, focused on the ornithine decarboxylase (ODC) gene – a validated target responsible for polyamine synthesis – from the pathogenic helminth Haemonchus contortus – a nematode responsible for serious digestive disorders in ruminants. ODC-deficient yeasts complemented with the nematode–ODC gene were used to test a 90,000-small-molecule library with the unexpected result of just a single and wrong hit, stilbamidine, which is a well-known inhibitor of S-adenosylmethionine decarboxylase – an important target of the polyamine biosynthesis pathway – but not of ODC.[18]

After this disappointing failure, yeast-based platforms were not resumed again to screen parasite targets until very recently. Dihydrofolate reductase (DHFR) is a recognized target that participates in thymidine, histidine, and methionine synthesis in several major parasites. A DHFR-deficient yeast strain was complemented with DHFR-encoding ORFs from Plasmodium spp., kinetoplastids (namely trypanosomas and leishmanias) and Schistosoma sp. (a trematode worm responsible for human schistosomiasis worldwide), showing that the well-characterized antimalarial drug pyrimethamine was able to inhibit the growth of the strain specifically complemented with the Plasmodium gene.[19] These authors further developed an HTS yeast platform to test a 14,400-small-molecule library of chemically diverse compounds to target DHFR, N-myristoyl transferase (enzyme that modifies membrane proteins) and phosphoglycerate kinase (a glycolytic enzyme in charge of maintaining ATP levels on hematic parasite stages) from the malaria parasite P. vivax. The system allowed multiplex detection in 384-well plates due to the tagging of each molecular target with a specific fluorochrome. The cytotoxicity of the hits was identified in yeast expressing the equivalent human ortholog. Under these circumstances, 36 specific hits were identified and re-tested against cell cultures of the bloodstream parasite T. brucei (the etiological agent of sleeping sickness in Africa), with 18 of them being confirmed to be also lethal for this pathogen.[20] This approach has been developed to test eight different Brugia malayi (a parasitic worm responsible of lymphatic filariasis) drug targets. Nine out of 400 compounds effectively inhibited some of the targets, and from those 9 compounds, 5 had antifilarial activity in vitro.[21]

A huge screening using a S. cerevisiae platform to assess the value of Leishmania inositol phosphorylceramide (IPC) synthase is currently on the way. Previous studies showed the suitability of IPC (a specific phosphosphingolipid of the parasite plasma membrane), pointing to this enzyme as a putative antiprotozoal drug target. Nowadays, using a collection of 1.8-million compounds, GlaxoSmithKline is carrying out a high-throughput yeast-based assay, which is grounded in the ability of the leishmanial enzyme to complement the lack of the S. cerevisiae gene ortholog AUR1p. The screening – performed under ultra HTS conditions – has identified so far >500 strong and selective hits (0.03% hit rate) in a primary screening.[22]

Yeast-based systems emerge as an inexpensive and simple HTS drug discovery tool for measuring the relevance of validated targets within a heterologous, yet cellular system, under axenic conditions. However, this method has limitations that may have been validated in the parasite, including the essentiality of the foreign protein within the yeast metabolic network, level of expression, the presence of cell wall or multiple export pumps, or the specific host/parasite interactions.

4. Expert opinion

The growing interest to fight NTDs manifested by all social stakeholders in the last decade has been accomplished in the 2015 Nobel Prize in Physiology or Medicine. Since no functional vaccines are currently available to prevent these diseases, chemotherapy is the only tool to deal with them. Nevertheless, most antiparasitic drugs currently used were synthesized several decades ago by serendipitous discovery, and have unknown or uncontroversial molecular targets. However, the lack of refinement of these drugs, along with their repeated administration, unpleasant side effects and severe toxicity, are in the origin of the emergence of resistant strains. One of the responsibilities of public administrations and private companies with the poorest layers of society is to find new, safer, and more efficient drugs to fully eliminate these devastating scourges in the coming years. In the early stages of drug discovery for NTDs, yeast-based systems have arisen as powerful tools that may complement the irreconcilable premises of target- and non-target-based paradigms of drug screening. Yeasts afford minimum axenic living conditions to harbor foreign proteins in order to study the killing effects of small molecules collections. On the grounds of providing HTS performance, multiplex optical detection, as well as easily cultivable cellular environment, several screenings of chemical libraries have been carried out in the last years on yeasts harboring robust parasite targets. Despite their undeniable advantages, yeast platforms are limited because of the biased selection of the molecular target, the presence of cell wall, or multiple export pumps that differ from those displayed by the parasite, the presence of a different metabolic context and maybe more importantly, the insurmountable gap of the role played by the host cells interacting with the parasite. These drawbacks hamper the use of this strategy as a substitute of phenotypic assays carried out with real infections ‘ex vivo’ (whenever they are feasible) before preclinical studies in experimental animals.

Declaration of interest

This research was supported by Ministerio de Economía y Competitividad (MINECO; AGL2010-16078/GAN and CYTED 214RT0482), Instituto de Salud Carlos III y Fondo Europeo de Desarrollo Regional FEDER (PI12/00104) and Junta de Castilla y León (grants Gr238, UIC108 and LE182U13). The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.