ANTI-PATHOGEN AND ANTI-TARGET-BASED SCREENS, SUCCESSES AND DISAPPOINTMENTS

During the 1980s it became relatively easy to identify genes that encode proteins and to produce those proteins in quantities sufficient to enable screening of series of chemical compounds for inhibitory activity. This process accelerated with the advent of whole genome sequencing in the late 1990s and the expansion of those technologies into the 2000s driven by the completion of genome sequences for Mycobacterium tuberculosis (Cole et al. Reference Cole, Brosch, Parkhill, Garnier, Churcher, Harris, Gordon, Eiglmeier, Gas, Barry, Tekaia, Badcock, Basham, Brown, Chillingworth, Connor, Davies, Devlin, Feltwell, Gentles, Hamlin, Holroyd, Hornsby, Jagels, Krogh, McLean, Moule, Murphy, Oliver, Osborne, Quail, Rajandream, Rogers, Rutter, Seeger, Skelton, Squares, Squares, Sulston, Taylor, Whitehead and Barrell1998), Plasmodium falciparum (Wirth, Reference Wirth2002) and Leishmania and Trypanosoma species (El-Sayed et al. Reference El-Sayed, Myler, Blandin, Berriman, Crabtree, Aggarwal, Caler, Renauld, Worthey, Hertz-Fowler, Ghedin, Peacock, Bartholomeu, Haas, Tran, Wortman, Alsmark, Angiuoli, Anupama, Badger, Bringaud, Cadag, Carlton, Cerqueira, Creasy, Delcher, Djikeng, Embley, Hauser, Ivens, Kummerfeld, Pereira-Leal, Nilsson, Peterson, Salzberg, Shallom, Silva, Sundaram, Westenberger, White, Melville, Donelson, Andersson, Stuart and Hall2005). It was often quoted that all potential drug targets were now in reach, and miniaturized and robotic systems made the screening of millions of compounds feasible.

The scientific literature of the 1990s into the 2000s is replete with descriptions of proteins whose activity could be inhibited by a range of molecules, few of which had the physicochemical properties compatible with potential for use as drugs. Molecular structures of proteins made it possible to predict which chemical ligands could bind within active sites and a multitude of studies were performed to find such molecules and use them as leads to generate more compounds. In the trypanosomatids, enzymes involved in the metabolism of the unusual thiol compound, named trypanothione (Fairlamb et al. Reference Fairlamb, Blackburn, Ulrich, Chait and Cerami1985), were seen as particularly useful drug targets given their exclusive presence in these protozoa and absence in man. However, in spite of a multitude of screening and structure-based efforts to identify selective inhibitors this system, like many others, failed to yield credible compounds for development as drugs (Gilbert, Reference Gilbert2013). The development of genetic methods that enabled testing of enzyme essentiality appeared to offer a short cut to ensure the value of particular targets. A seminal paper from investigators at GSK in 2007, however, reported that the systematic expression of scores of target proteins and their screening against libraries of hundreds of thousands of compounds yielded absolutely nothing in terms of new antimicrobials (Payne et al. Reference Payne, Gwynn, Holmes and Pompliano2007). Other companies had been suffering similarly frustrating times and the academic laboratories involved in rather lower-level throughput screening could report the same. Target-based screening has to date failed to deliver its full potential with a few notable exceptions including new inhibitors of dihydrofolate reductases, DNA topoisomerases and C14-α-sterol demethylases in fungi. Notwithstanding, today within the MMV portfolio there are compounds whose activity was aimed at specific targets including dihydroorotate dehydrogenase (Booker et al. Reference Booker, Bastos, Kramer, Barker, Skerlj, Sidhu, Deng, Celatka, Cortese, Guerrero Bravo, Crespo Llado, Serrano, Angulo-Barturen, Jiménez-Díaz, Viera, Garuti, Wittlin, Papastogiannidis, Lin, Janse, Khan, Duraisingh, Coleman, Goldsmith, Phillips, Munoz, Wirth, Klinger, Wiegand and Sybertz2010). The electron transport chain, essential in Plasmodium pyrimidine biosynthesis, has also been targeted with potent quinolone-based compounds (Stocks et al. Reference Stocks, Barton, Antoine, Biagnini, Ward and O'Neill2013). Developing more sophisticated modelling methods continues to contribute towards lead compound identification and refinement against specific targets but also increasing the use of physicochemical constraints to enhance the probability of compounds emerging with drug-like features (Cunningham et al. Reference Cunningham, McPhillie, Johnson and Fishwick2013). In the trypanosomatids, protein farnesyl transferase and N-myristoyl transferase were shown to be targets against which very potent compounds could be designed (Frearson et al. Reference Frearson, Brand, McElroy, Cleghorn, Smid, Stojanovski, Price, Guther, Torrie, Robinson, Hallyburton, Mpamhanga, Brannigan, Wilkinson, Hodgkinson, Hui, Qiu, Raimi, van Aalten, Brenk, Gilbert, Read, Fairlamb, Ferguson, Smith and Wyatt2010). However, in these latter cases too there has been a difficulty in adapting compounds that are excellent inhibitors of the enzymes and potent killers of the pathogens to drugs (Tate et al. Reference Tate, Bell, Rackham and Megan2013). These difficulties relate to the problems surrounding pharmacokinetics and pathogen tropisms within hosts, but optimism remains that as our knowledge of drug interactions within the mammalian system increases, so potent inhibitors can be adapted to allow them to reach their targets.

Many of the compounds developed as antimicrobials emerged well before the advent of molecular targets for drugs and even before the luxury of effective in vitro cultivation methods. The compounds were found by screening of chemicals against pathogens within animal models, for example around 300 000 compounds were screened as potential antimalarials in the 1970s (Ockenhouse et al. Reference Ockenhouse, Magill, Smith and Milhous2005). Those screening programmes, by necessity in the absence of the reductionist approaches that came later, took the entire system into account. That is to say the chemicals they used needed to inhibit targets within pathogens, themselves distributed to infected tissues within the host. Hence any compounds which did not possess pharmacological properties compatible with suitable stability and distribution within the host would never be selected. Today we are reverting towards such a ‘systems pharmacology’ (Iyengar, Reference Iyengar, Zhao, Chung, Mager and Gallol2012) taking into account all of the parameters required to produce a suitable drug. As available compound numbers have increased, large-scale in vivo screening is not viable, hence it is usually necessary to determine PK properties of drugs independently of potency and then to marry datasets prior to choosing those which are selected for in vivo screening.

There is now a general acceptance that the best antimicrobial agents have emerged by screening compound libraries against whole organisms in vitro and then taking hits to optimize for their in vivo pharmacology. Thus the antimalarial pipeline currently contains over a dozen compounds, most of which came through screening against the whole organism rather than by aiming at a particular target. For tuberculosis, the kinetoplastid protozoa and many bacteria, the same is true. A challenge then is to identify targets within the parasites that are functional in the mammalian stages of the life-cycle. The selection of resistant strains can help to identify the target with whole genome sequencing pointing to mutations that lead to resistance (provided the mutations are within the target rather than affecting other aspects of drug-pathogen interaction such as entry into or efflux from cells). The use of high-throughput RNAi screens in trypanosomes has proven hugely successful in identifying loss of function mutations that can allow parasites to grow in the presence of drugs and thus point to potential modes of action (Horn, Reference Horn2013). Another route has been to measure the metabolome of pathogens treated with compounds in attempts to identify whether enzyme products and substrates change in ways pointing to selective inhibition (Creek and Barrett, Reference Creek and Barrett2013). The phenotypic whole-organism screens also pick out compounds that may act against multiple targets, and ‘polypharmacology’ where multiple targets are hit by a single drug certainly contributes to the superiority of the phenotypic screening route; clearly exampled in kinases (Knight et al. Reference Knight, Lin and Shokat2010). However, many compounds that are highly potent in phenotypic screens still fail as drugs as the other part of the ‘system’ (i.e. interactions within the host's biochemical environment) must also be surmounted. It is the appreciation of these pharmacological criteria that is now guiding the development of improved antimicrobial agents. Expression of target genes in highly tractable organisms like yeast offers another route to seek inhibitors of enzymes that function within a cellular context, and has been successfully used to identify new inhibitors of inositol phosphorylceramide synthase from Leishmania (Norcliffe et al. Reference Norcliffe, Alvarez-Ruiz, Martin-Plaza and Denny2013).

THE GROWTH OF PK, PD AND ADME-TOX-BASED APPROACHES

It is critical to optimize pharmacological features to ensure that drugs are not toxic to host animals at doses required to kill pathogens. Moreover, compounds must be distributed to the tissues in which their target pathogens have localized and they must be stable enough and remain available in sufficient quantities to exert their anti-pathogen effect. Several examples exist whereby compound selection was driven primarily through optimization of PK/PD parameters. For example, the benzoxaborole class of compounds were shown to have some activity against Trypanosoma brucei in vitro (albeit of relatively low potency) – but in vivo they were shown to be efficacious against the T. brucei acute mouse model (Jacobs et al. Reference Jacobs, Nare, Wring, Orr, Chen, Sligar, Jenks, Noe, Bowling, Mercer, Rewerts, Gaukel, Owens, Parham, Randolph, Beaudet, Bacchi, Yarlett, Plattner, Freund, Ding, Akama, Zhang, Brun, Kaiser, Scandale and Don2011). Since the standard model for Stage 2 HAT involves a six month follow-up post-treatment to seek parasites in blood, SCYNEXIS pursued a lead optimization programme to determine levels of compound in brain and compared these to drug doses needed to kill and the time of exposure required. SCYX-7158, which emerged from this work is now in phase 1 clinical trials (Wring et al. Reference Wring, Gaukel, Nare, Jacobs, Beaudet, Bowling, Mercer, Bacchi, Yarlett, Randolph, Parham, Rewarts, Platerner and Don2013).

For Plasmodium, MMV has implemented a very rigorous series of protocols to enable PK and PD profiling of compounds, with plasma half-life, protein binding and volume of distribution, as well as the PD property of rate of kill (Anthony et al. Reference Anthony, Burrows, Duparc, Moehrle and Wells2012) seen as key criteria to advance compounds. Only compounds that combine satisfactory PK/PD relationships are now considered within the MMV target product profile (TPP) of single-dose drugs for radical cure (Jimanez-Diaz et al. Reference Jimanez-Diaz, Viera, Fernandez-Alvaro and Angulo-Barturen2013). For leishmaniasis and tuberculosis it has proven remarkably difficult to identify compounds that are efficacious against pathogens within the acidic environment of the phagolysosomes in which they thrive in host macrophages. The presence of these pathogens within granulomas that further restrict the local accumulation of drugs (Kjellsson et al. Reference Kjellsson, Via, Goh, Weiner, Low, Kern, Pillai, Barry and Dartois2012) has yet to be surmounted in investigative PK studies. No available animal models give a reliable reflection of the human host environment which contributes to the difficulties that have been so problematic in identifying suitable drugs for these organisms. As we better appreciate these issues, so the need for better, more predictive in vitro and in vivo models for drug development becomes ever more clear.

THE CHANGING DRUG DEVELOPMENT LANDSCAPE, PRODUCT DEVELOPMENT PARTNERSHIPS, RE-ENGAGING PHARMA AND OPEN SOURCE DRUG DEVELOPMENT

By the late 1990s, a series of mergers within the pharmaceutical industry, accompanied by business models dependent upon identifying block-buster drugs that could be sold at high prices for the chronic diseases common in the affluent west, were impacting adversely on anti-pathogen drug development. Short doses of cheap curative drugs were the norm for most infectious diseases within the developed world. For the cohort of diseases that afflict predominantly the world's poorest people, the required investments reaching many millions of dollars for development, could not be put forward with no prospect of a return. It was to this back-drop that in 1998 the Global Alliance for Tuberculosis was founded followed by MMV in 1999. In 2003, the Drugs for Neglected Diseases initiative emerged with a remit to seek new agents for HAT, leishmaniasis and Chagas disease (Don and Ioset, Reference Don and Ioset2013). Other PDPs, such as the Consortium for Parasitic Drug Development (CPDD), received substantial backing from the Bill and Melinda Gates Foundation to take the orally available diamidine pro-drug, pafuramidine, forward for HAT (Paine et al. Reference Paine, Wang, Generaux, Boykin, Wilson, De Koning, Olson, Pohlig, Burri, Brun, Murilla, Thuita, Barrett and Tidwell2010) and the Institute of One World Health (iOWH) took paromomycin forward for visceral leishmaniasis (Sundar et al. Reference Sundar, Jha, Thakur, Sinha and Bhattacharya2007). Elsewhere academic-based laboratories, for example the Drug Discovery Unit in Dundee and the Center for Discovery and Innovation in Parasitic Diseases at UCSF, have dedicated themselves to drug discovery with the PDPs offering natural points for handing over optimized leads for clinical development. The growth of the PDPs and increased level of sophistication in academic drug discovery has also seen a new acceptance within the pharmaceutical industry that competence is now available for them to engage in the area. Pharmaceutical companies have made compound libraries available for large screens against HAT, leishmania, Chagas disease, malaria, tuberculosis and other infections, and significant hits have emerged. Moreover, several large pharmaceutical companies have also re-launched their own programmes in drug discovery for tropical diseases. GSK Tres Cantos (Spain) is one of the leaders, making available the data from their malaria screening programme. In addition the GSK Open lab offers opportunities for academics and small biotech companies to work within a GSK development environment (Jimanez-Diaz et al. Reference Jimanez-Diaz, Viera, Fernandez-Alvaro and Angulo-Barturen2013). Novartis established a tropical disease research facility in Singapore in 2004, with a current focus on malaria (Smith et al. Reference Smith, Diagana and Yeung2013), dengue and HAT. The adoption by industry of pre-competitive models for drug discovery emulates many of these changes and offers opportunities to extend public private interactions into new disease areas. Another interesting growth area has been the emergence of open-source drug discovery where ideas on chemical synthesis can be exchanged on line to accelerate the resolution of problems in chemical synthesis as was successfully demonstrated in separating enantiomers of the anti-schistosomal drug praziquantel (Robertson et al. Reference Robertson, Ylioja, Woelfle, Robins, Williamson, Badiola, Olliaro, Wells and Todd2013).

A largely overlooked area of anti-pathogen chemotherapy has been the development of novel drugs against helminth infections. Given that billions of people the world over are afflicted with one or more parasitic helminth species it has been extraordinary that so little effort has gone into developing novel agents against the human helminthiases (Keiser and Utzinger, Reference Keiser and Utzinger2010). The fact that several orally available, relatively safe and efficacious drugs, developed initially for the veterinary market, exist has been a disincentive towards development, as has the fact that many of the helminthiases are not in themselves life threatening. However, with the realization that these neglected tropical diseases do have a huge impact on human populations, interest in this area has been reinvigorated (Hotez et al. Reference Hotez, Brindley, Bethony, King, Pearce and Jacobson2008). With it comes a new series of challenges around targeting organisms that pass through multiple life-cycle stages in their host (targeting adult vs larval worm stages requires different compounds in many cases) (Taylor et al. Reference Taylor, Hoerauf, Townson, Slatko and Ward2013). The presence of relatively large organisms, with multiple organ systems and a protective cuticle is presenting a new wave of challenges for those interested in developing drugs for these diseases.

POTENTIAL RISKS TO A DECADE OF PROGRESS

For the first time in history a robust pipeline of rationally selected compounds for development against malaria has been developed (Burrows et al. Reference Burrows, Burlot, Campo, Cherbuin, Jeanneret, Leroy, Spangenberg, Waterson, Wells and Willis2013). Elsewhere, less robust pipelines, but pipelines nevertheless, have emerged for treatment of human African trypanosomiasis, leishmaniasis, Chagas disease, tuberculosis and other neglected diseases. The paradigms that have emerged to create these pipelines are now being applied to other parasitic and bacterial diseases. Comparing the status of the different pipelines indicates that the tropism and behaviour of pathogens within the host is a main determinant with the ease by which we may target drugs to them. From an optimist's perspective our increased appreciation of the barriers that selectively hinder development of drugs for different pathogens encourages the belief that the dissection of these problems and adoption of pharmacological approaches to breach the blood-brain barrier, granuloma barriers or the harsh acidic environment of the macrophage mean a way ahead is possible.

The focus must now be to sustain the successes of the past decade to ensure the tools are available for more effective treatment of patients, control of diseases in populations and eventual elimination or eradication. For some diseases, for example malaria, the demands are high with need for different types of drugs, exhibiting a variation of mechanisms of action and pharmacokinetic properties, to be used for different purposes in treatment and control (Burrows et al. Reference Burrows, Burlot, Campo, Cherbuin, Jeanneret, Leroy, Spangenberg, Waterson, Wells and Willis2013). For malaria, this has led to new definitions of target clinical profiles and illustrates the need for several new drugs, and safe effective multi-stage combinations of these drugs. Combination therapy has long been a complexity for those developing drugs for tuberculosis where issues of drug–drug interaction and pharmaceutical formulation have to be addressed in parallel with clinical development. Other demands face neglected diseases. In the case of HAT, the incidence of the disease has declined substantially over the past decade, with fewer than 7000 new cases reported in 2011 (Simarro et al. Reference Simarro, Cecchi, Franco, Paone, Diarra, Ruiz-Postigo, Fèvre, Mattioli and Jannin2012). This has led many funding agencies to question whether there is a need to continue investing in new tools when existing tools, such as the nifurtimox–eflornithine combination (NECT), can be demonstrably effective when used correctly, despite the facts that these existing tools are generally highly unsatisfactory as drugs. Resistance to eflornithine is very readily selected (Vincent et al. Reference Vincent, Creek, Watson, Kamleh, Woods, Wong, Burchmore and Barrett2010) and reports of treatment failure to eflornithine monotherapy are already emerging from parts of Africa. The pipeline is not yet sufficiently robust to cover loss of drugs to resistance. For leishmaniasis too, the existing drugs suffer many limitations and the realization that for every clinically recognized case of leishmaniasis there are probably 10 undiagnosed, asymptomatic infections offering a potential reservoir that will seriously hinder any elimination programme. Liposomal amphotericin B and new combinations have been hailed as a tool for elimination and offer the potential advantage of shorter course treatments against visceral leishmaniasis where cost, delivery to populations, and stability plus differences in efficacy between India and Sudan remain key issues (Croft and Olliaro, Reference Croft and Olliaro2011).

There must be a degree of pragmatism among the communities involved in developing drugs for these pathogens. The paradigm of bringing compounds into PDPs, where robust pharmacological data combines with potency studies to drive development, clearly works. Further expansion of the role of pharmaceutical companies, integration of the activities of existing PDPs, support for innovation in the academic and biotech sectors, enhanced sharing of information and an insistence on global standards for the development of new compounds for these diseases is the way ahead. The implementation of new drugs in programmes for the treatment, control and elimination/eradication of these diseases also requires more forethought, planning and commitment of funds and support for health systems.