Introduction

Cutaneous Leishmaniasis (CL) is a poverty-associated disease that presents as lesions in the form of ulcers, nodules, papules, and plaques on exposed body parts (arms, legs, face). There are approximately 1 million new cases of CL each year; however, it is difficult to estimate the true incidence of leishmaniasis due to a lack of surveillance systems in remote areas and marginalized populations where the disease occurs, and because it is not a reportable disease within the health systems of many endemic countries. CL has a wide geographical range, with the majority of cases occurring in Afghanistan, Algeria, Colombia, Brazil, Iran, Syria, Ethiopia, Sudan, Costa Rica, and Peru. It is estimated that there are 20 species of Leishmania that cause CL, each with different epidemiology and natural history. Some species of Leishmania are more common than others, and their incidences vary widely amongst their geographical locations. The most common Old World species that cause CL are L. tropica, L. aethiopica, and L. major and the New World parasites include L. mexicana, L. amazonensis, L. braziliensis, L. panamensis, and L. guyanensis [1]. It must be noted that hybrid Leishmania have been isolated in nature and conjugation (transfer of genetic materials) has been demonstrated in the sandfly vector [2].

Leishmania infections typically originate from the bite of sandflies belonging to either Phlebotomus spp (Old World) or Lutzomyia spp (New World), which transmit the parasites into the skin of the host. The first sign of infection appears as a small erythema that develops at the site of the sandfly bite and advances within a few weeks to months. Lesions often become chronic after the initial erythema develops into a papule, then a nodule that may ulcerate over a period of two weeks to six months before becoming the characteristic lesion [3]. In strict CL, each lesion is representative of an independent fly bite, but, in rare cases, the disease can manifest as a disseminated disease (diffuse CL [DCL]) [4].

There is currently no satisfactory treatment for any form of CL and a need to develop short, safe, efficacious, affordable, and field-friendly treatments for the disease. The Drugs for Neglected Diseases initiative (DNDi) is a collaborative, patients’ needs-driven, nonprofit drug research and development (R&D) organization that is developing new treatments for neglected diseases, including drug discovery against CL. DNDi decided to focus on development of treatment for CL that is predominately caused by the L. tropica and L. braziliensis species. This is due to the severity and public health importance of these particular species; however, it is likely that these new treatments would be efficacious against CL caused by other parasites, too, as described in the 2007 consultative meeting to develop a strategy for treatment of cutaneous leishmaniasis [5]. Here, they suggest that these parasites will be targeted primarily because of the public health importance of the disease they produce. L. tropica is an anthroponotic parasite; therefore, treatment of cases is essential for controlling transmission. L. braziliensis is associated with mucosal leishmaniasis (ML) more often than other New World Leishmania, and one risk factor for developing ML is duration of infection. Hence, again, early treatment is important for control of ML. Furthermore, it was suggested that highly active drugs against L. tropica and L. braziliensis may be active against others (such as amphotericin-B). Decisions on how to treat CL are individualized. Treatment approach depends on a number of factors, including Leishmania spp. and/or geographic incidence as well as clinical characteristics and symptoms. For example, the oral agent miltefosine has FDA-approved indications that are limited to L. braziliensis, L. panamensis, and L. guyanensis.

Animal models can be used to evaluate drug candidates, with the goal of developing new drugs that offer a safe, effective, and shorter-course treatment for CL, as well as an acceptable safety profile. Therefore, it is essential to review the current animal models, which are highly diverse, and provide guidance on which models are the best for the discovery and utility of new anti-leishmanial drugs.

Animal Models for Different Species of Leishmania

It must be noted that there is no validated animal model for CL, although these models are routinely used in several laboratories. Animal model validity refers to the similarity between the model and the human condition. Standardized animal models are used in laboratories that screen drug compounds against Leishmania strains to assure reproducibility. However, the predictive validity of the animal models is often low due to incomplete correlation between animal and human disease mechanisms. The philosophy is that if an agent was ineffective in these models, then no further development would be justified. Besides, some preliminary data on dose response and toxicity may be collected. Certain parameters must be considered when determining the similarity of an animal model to the human disease. These parameters include, but are not limited to, the clinical manifestations and presentations of the disease, disease pathogenesis, and the immunological responses that occur. For an animal model to be considered “validated” and used as part of preclinical development studies of an agent for therapy or prophylaxis, it must be similar to humans with respect to etiology, pathophysiology, symptomatology, and response to the therapeutic or prophylactic agents. For example, the BALB/c–L. major model is far from being a validated model for human CL based on the above criteria. Immunosuppression renders BALB/c mice resistant to L. major infection [6], whereas it has the opposite effect in humans. However, it is convenient, with easy read out using conventional methods. The inbred, resistant strains have many similarities with human disease, but don’t have the same pathological features and disease manifestations, such as secretions from the lesions (classically known as wet lesions), chronic persistent lesions, and the variety of disease presentation. Being inbred, these mice don’t show the variations in the presentation of the disease seen in humans; nevertheless, they are the most appropriate models available.

Ideally, outbred mice challenged by sandfly bites would be closer to human disease, particularly for vaccine studies, but it would require a large number of mice with complex endpoint evaluation (different disease manifestation) and a setup for sandfly challenge, which would be time-consuming and expensive and also require special expertise. It is therefore not practical for routine screening of drugs.

Many in vivo experimental models of both the New World and Old World CL have been developed for the study and testing of new compounds against the parasite, but few accurately reproduce the biological responses that would occur in humans. Inbred mice strains are most commonly used for experimental CL infections. However, other animal models, such as hamster and rat species (primary tests), dogs (secondary tests), and nonhuman primates (tertiary tests) have also been used. Examples of experimental CL involving dogs are lacking (though not unheard of [7–9]), and, as a result, this paper focuses solely on rodents and nonhuman primates as the potential models for drug evaluation. In vivo models for CL should aim to mimic the natural transmission of the disease—for example, the parasite load, presence of saliva, and site of inoculation—to enable accurate representation of disease progression. However, in many of the current models, this is not always the case.

The purpose of an animal model regarding CL is to investigate the parasitic cause of the human disease and use this information in drug studies to prevent increased risk to the human population. When exposed to the different Leishmania species, which each have individual pathogenic characteristics, the animal model should have pathological features and immunological responses that are similar or identical to humans. The model can then be used for the development and evaluation of new potential anti-leishmanial compounds and/or immunotherapy or therapeutic vaccines. There are many ethical issues that arise when using animal models, and thus their use should be avoided when possible. The three R’s of animal testing (Replacement, Reduction and Refinement), which were first described in 1955 by Russell and Burch [10], must always be considered as guiding principles for the more ethical use of in vivo models for the discovery of new drugs. The drug discovery algorithms for CL typically begin with the in vitro high throughput screening (HTS) during lead identification, which initially assesses the efficacy of large numbers of compounds against the parasites directly. The compounds that qualify as an in vitro hit are then assessed using in silico methods to determine their pharmacokinetic properties and toxicity during candidate identification. Assessing these properties via computer simulation has the potential to speed the rate of discovery without the need for expensive laboratory work. Drug discovery is a very innovative part of the drug development pipeline, with customised in vitro HTS and in silico packages, allowing for the identification of drugs with high on-target efficacy, low off-target toxicity, and favourable pharmacokinetics. Moreover, the in vitro results may potentially correlate with the in vivo studies [11] and the in silico ADME screening tools for pharmacokinetics and pharmacodynamics are gaining increasing accuracy and regulatory acceptance as well. Such weight is now given to these results that drug candidates with an unacceptable in silico and/or in vitro profile are excluded from further study by the drug developer. In vitro drug discovery has many advantages over in vivo testing, such as the very quick generation of results and the minute amount of test compound used. However, current drug screening methods in CL are laborious, especially when the intracellular amastigote model is used. In addition, the drug discovery test results of in vitro systems will always need to be verified in animals during drug development and before the human clinical trials of the compound begin. This is because the in vitro predictions for the compounds may not necessarily transfer to in vivo situations. Similarly, results of animal models may not be totally predictive of the response in humans, as evidenced by the low/lack of efficacy of allopurinol alone in humans, while it was shown to be active in vitro and in animal models [12–15].

It is also important to use recent isolates of Leishmania from the field for in vitro as well as in vivo tests. Some laboratories prefer to use old WHO reference organisms isolated in the 1960s–1970s for consistency and uniformity; however, Leishmania change over time and those circulating presently have different characteristics (Ana Rabello, personal communication). Changes have also been demonstrated in L. donovani parasites kept in culture (in vitro) compared to parasites freshly isolated out of rodents [16].

Nonhuman primate models are often used for the evaluation of potential anti-leishmanial compounds as a precursor to human clinical trials [17]. However, the results achieved from these nonhuman primate studies are often no better than those from the rodent models, and their use over rodent models may even be inferior [18]. The determination of parasite load within the animal model is also an important factor to consider when establishing the efficacy of a compound.

The requirement of an animal model capable of generating reproducible results is therefore crucial for the study of potential anti-leishmanial compounds. It is important to mention that, as of yet, the translation of results from preclinical to clinical is very poor, hence almost no compound has reached Phase III. With these factors in mind, it is important to determine the most effective animal model for CL drug discovery to help focus on robust systems and reduce the use of animals. Rodent models are most commonly used for studies of experimental CL infections, especially when studying the pathogenesis of disease or to test novel anti-leishmanial agents. This is because of the high availability of cellular markers and inbred, congenic, and now even transgenic mice, in addition to all other attributes that have made rodents a suitable animal for laboratory-based experiments. Mice in particular are susceptible to most strains and species of Leishmania in both the non-cure and self-cure models [19]. Much of the recent work towards the study of CL has been based on initial findings [20–22] deducing that different strains of mice, with differing phenotypes, vary in their susceptibility to different Leishmania species.

Some of the observations made in the rodent models of CL might not be similar or relevant to human hosts due to the distance in phylogeny. However, rodent models can provide a fast turnaround during the drug research program prior to testing in nonhuman primate models. Old World monkey species (macaques, baboons, mandrills, etc.) have the closest evolutionary relatedness to humans among the approachable nonhuman primate models [17]. Therefore, nonhuman primate models are generally used as the final experimental study of the safety and efficacy of drugs, with the hope that their relatedness to humans will confer a similar mechanism of CL infection and disease progression.

Standardization of Inoculum

It is well known that promastigotes go through developmental stages (metacyclogenesis) during growth in culture. Therefore, the composition and size of inoculum as well as the method, route, and site of inoculation greatly influences the outcome of infection. It is therefore important that all these parameters are standardized for obtaining reproducible results.

During the course of in vitro growth, Leishmania undergo some essential changes (similar to that within sandfly vector) from highly dividing but non-infective form in log phase to non-dividing but highly infective metacyclic form in stationary phase [23]. Therefore, most laboratories harvest the parasite during the late stationary phase of growth and employ peanut agglutinin (PNA) to remove non-metacyclic parasites before inoculation. The size of inoculum is also important since, up to a certain level, higher doses produce larger lesions, which develop faster. Generally, 1 × 10⁶ metacyclic-enriched parasites were injected subcutaneously in a volume of 40μl. However, with the highly purified infective form of parasites, a few thousands of parasites are injected to simulate delivery by infected sandflies, which are estimated to be <5,000 [24]. The site of injection is also important for development of lesions. Footpads of mice are usually used for drug screening and provide easy measurement of lesion and parasite load at the lesion. The base of the tail of mice and the forehead of monkeys have also been used for vaccine studies. Another interesting site for mouse studies is the ear, which has also been used for inoculation by infected sandflies to simulate natural infection. Needle injections primarily deliver parasites subcutaneously, whereas sandfly introduces the parasite intradermally. The course of infections and the immune responses that ensue are different, primarily due to the type of inflammatory cells that appear soon after the introduction of parasites at the different sites [25]. Ideally, ear infection by infected sandflies would be preferable, particularly for vaccine development, since it is closer to natural infection in humans and has been shown to be different from needle challenge in vaccinated mice [24]. It should be noted that this method should be validated for humans first before making it a prerequisite for testing safe candidate vaccines in human clinical trials. However, this method requires availability of sandflies and consistent maintenance of infection rates. For drug screening, needle injection of a standard inoculum given subcutaneously in the food pad of mice would make it possible for many laboratories to screen drugs, albeit not simulating natural infection.

The composition of the inoculum with respect to the types of parasites (at various stages) should be determined and standardized. The PNA negative selection to enrich metacyclic L. major contains apoptotic parasites that can promote infection [26] and, ideally, their proportion in the inoculum should be known. The screening labs have their consistent method of growing the parasite, harvesting at a predetermined time, and preparing metacyclic-rich inocula within their own laboratories, in order to limit variation. However, comparison of results from different labs can be difficult because of the strain of parasite used, the age of the parasite since the isolation from the field, method of growth, separation of infective parasite, etc.

Old World: L. major, L. tropica, and L. aethiopica

New World: L. mexicana, L. amazonensis, L. braziliensis, L. panamensis, and L. guyanensis

Continuous Monitoring of the Parasite Load in the In Vivo Models, Using Reporter Molecules

Leishmaniasis affects many millions of people worldwide and hence there is an ever-growing need for a rapid screening assay to allow the high throughput screening of potential anti-leishmanial compounds. The current classical methods of in vivo parasite load determination are laborious, time consuming, and do no support automation. Furthermore, methods such as lesion size measurements are not always indicative of parasite killing due to immunological responses such as inflammation, which might also contribute to lesion size. Previously, methods required the animals to be euthanized at various times after infection. Quantification was then achieved by measuring the parasite load either by microscopy, limiting dilution assay, or q-PCR amplification of the parasite DNA. As Calvo-Alvarez et al. wrote, “Furthermore, this approach has some important limitations that must be overcome: (i) postmortem analysis of animals makes it impossible to track the space/time progression of the pathogen within the hosts; (ii) spread of the pathogen to unexpected anatomic sites can remain undetected; (iii) in order to achieve precise and relevant data, it is necessary to kill large numbers of animals” [78]. The measurements of lesion size and footpad and ear thickness are all common methods used to quantify the progression of the disease and hence the parasite number [35,66]. However, these methods require an incubation period of about two weeks before any visual observations of infection can be made. Thus, it is important to devise a way of detecting parasite numbers in the earlier stages of infection, before a visual phenotype is observed. More recently, the in vivo imaging system (IVIS) has been developed to allow the detection of measurable signals associated with cells in living organisms [79]. It is extremely important to quantify the parasite load present in an in vivo model in a live host, in order to determine the infectivity of the Leishmania parasite and/or to measure the efficacy of a potential anti-leishmanial drug.

Many reporter molecules have been devised for the quantification of protozoan parasites and each one comes with different degrees of sensitivity, advantages, and disadvantages. Reporter genes must encode a gene product that has a readily measurable phenotype that must also be distinguishable over a phenotypic background [80]. As described by Dube et al., 2009, “The ideal reporter gene should be (i) absent from the host; (ii) inert (and should not affect the physiology of the parasite cell); and (iii) should represent a simple, sensitive and inexpensive assay for quantification of reporter expression” [81]. Continuous monitoring of parasite load in vivo is an ideal way to monitor drug treatment outcome.

Discussion

To summarize, animal models that show the greatest potential for the evaluation of new potential anti-leishmanial drugs have not been certified, and this review goes some way in clarifying our understanding of the predictive validity and representativeness of the models for human disease. Currently, the BALB/c–L. major model is one of the most widely used. The idea of using BALB/c mice for drug studies is based on the notion that if the BALB/c mice can be protected or cured from the parasite disease, then so can the resistant mouse strains. However, using this model will possibly result in the inability to detect potentially useful compounds due to the stringencies and peculiarities associated with it. The risk of overlooking such compounds is too great, and thus, the model has limitations. Despite this, it should be noted that the choice of a rodent model strongly depends on the research question that is being addressed. The use of BALB/c–L. major can be justified in drug testing, as it is reproducible and generally accepted to be a rigorous test. CL progresses quickly in this non-healing model; hence, when a lesion size and parasite load reduction is obtained, it is likely to lead to reduction in other self-cure models. The in vivo model should be considered solely as a predictive tool in a complete set of experiments involving in vitro and additional in vivo work in which each experiment answers a different defined question, as demonstrated by Grogl et al., 2013.

It is usually difficult to refute a dogma. The BALB/c–L. major model has been used for testing drugs for decades [97,98], primarily because of extreme sensitivity of BALB/c to L. major infection and ease of outcome evaluation. It is now clear that the outcome of the infection is directly influenced by the immune responses of the host. It is also known that L. major infection in BALB/c triggers a strong Th-2 response to leishmanial antigens, leading to rapid lesion growth and generalized infection and, eventually, death. Hence, as mentioned elsewhere in the text, immune-suppressive agents, which control this Th-2 response, reduce the lesion size and parasite load. We would like to caution that reduction in lesion size and parasite burden may not be due only to the anti-parasitic activity of the agents being tested but also their effects on the immune responses. As of yet, the most comprehensive rodent models are the L. tropica–CsS-16 [48], L. amazonensis–CBA [18], and L. braziliensis–golden hamster [66], which each show promise for drug discovery (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. The recommendation of various animal models with Leishmania parasite–animal models for discovery of new anti-leishmanial drugs, based on the evidence and interpretation of the studies reviewed in this paper.

https://doi.org/10.1371/journal.pntd.0003889.t001

Many Leishmania species are ineffective at producing human-like immune and clinical responses in nonhuman primate models, or else have not been widely studied (L. tropica, L. mexicana, L. panamensis, and L. guyanensis). The L. major–vervets model is one that well represents the human form of disease and has the additional advantage of the animal breeding relatively well in captivity. The vervet monkey has also been shown to produce recognizable lesions after infection with L. aethiopica, but more research is needed to evaluate the models’ potential for drug discovery. The L. major–rhesus model also shows consistency with human cases of CL and has been used in vaccine evaluation. Both the rhesus monkey and Cepus apella develop clinical symptoms after infection with L. amazonensis, however, the former model is more representative of the human CL disease and thus may be better for drug trials. The rhesus monkey also develops characteristic lesional granulomas remarkably similar to those seen in humans infected with L. braziliensis. Therefore, the L. major–vervets (37), L. major–rhesus [40,41], and L. braziliensis–rhesus [68,69] models show the greatest potential for the evaluation of new potential anti-leishmanial drugs.

The use of nonhuman primate models for the evaluation of potential anti-leishmanial compounds is often used as a precursor to human clinical trials. However, the results achieved from these nonhuman primate studies are often inferior to those that use rodent models. Nonhuman primates should only be used as a model for drug evaluation if rodent models are incapable of reproducing reliable results. As of January 1, 2013, Directive 2010/63/EU replaced Directive 86/609, stating that the use of nonhuman primates can only be used under strict circumstances, for example, for the benefit of human beings when it involves life-threatening or debilitating diseases. Nonhuman primates must only be used if there is scientific justification to the effect that the purpose of the procedure cannot be achieved by the use of species other than nonhuman primates. Their use should not be sanctioned during the early stages of research and is more appropriate for toxicology testing at the preclinical stage of development. Often, the rodent models are far better at assessing the efficacy of a drug, and some reproduce aspects of human infection to a far greater extent than the nonhuman primates. When considering in vivo models for drug discovery purposes, it should be noted that only small fractions of drugs are available. Relatively restricted amounts of drug are used in mice, being smaller than hamsters. Nonhuman primate models require much higher doses that are often not available at early stages. Animal models that show potential for studies of drug evaluation should have obtained the disease through a method that mimics natural infection and, additionally, should present similar clinical symptoms as are observed in humans. It is also important to choose the right animal model for the Leishmania species, since it has been demonstrated in multiple clinical trials that the efficacy of a given compound varies depending on the species of Leishmania causing the infection. These factors should be considered although it is acknowledged that such parameters are not always feasible. From this analysis, we conclude that L. major–vervet, L. major–rhesus, L. tropica–CsS-16, L. amazonensis–CBA, L. braziliensis–golden, or L. braziliensis–rhesus are the best models for CL drug discovery. However, the use of these nonhuman primate models should only be used under extreme cases whereby they conform to the new directive and are scientifically justified.

Neither β-galactosidase nor β-lactamase are effective reporter molecules for in vivo drug discovery. However, their use in vitro is widely used due to their simplicity and low cost. The use of the firefly luciferase reporter molecule in transgenic Leishmania species shows exciting results in its use for drug evaluation due to its high sensitivity and use in real-time analyses; however, the process is costly due to the need for the substrate luciferin. Alternatively, fluorescent reporter molecules can be used to determine parasite load and have the advantage that they do not require the use of exogenous substrates. Fluorescent proteins work intrinsically, allowing for the real-time analysis of the molecular events in living cells. Although the GFP and its variants have been used to quantify Leishmania parasites in in vivo models previously, there are concerns that the excitation maxima of these proteins are close to the UV range, causing damage to living cells. The recent mCherry + L. major model has been shown effective in a vaccination study in BALB/c mice and has greater sensitivity over the more commonly used GFP models. Additionally, the combined use of EGFP and firefly luciferase in the L. major–BALB/c model demonstrates a highly sensitive method for parasite quantification that has potential for drug evaluation.

It is important to emphasize the extra valuable information we can obtain using reporter molecules as tools for calculation of parasite load when compared to conventional methods. The conventional methods used to quantify parasite load, such as the direct measurement of parasite via microscopy or q-PCR amplification of parasite DNA, are not only laborious and time consuming, but also require the euthanasia of large numbers of animals. In addition, these methods cannot detect the spread of pathogens to unexpected anatomic sites or track their space/time progression. In contrast, reporter molecules provide a readily measurable phenotype that allows the continuous monitoring of parasite load in vivo. These tools are highly sensitive and often show potential for automation resulting in high-throughput quantification, especially when combined with computational methods.

Current animal models for the utility and discovery of new potential anti-leishmanial drugs differ vastly in their immunological response to disease and clinical presentation of symptoms. After the initial HTS of the drugs in vitro, animal models are used to ensure correlation of drug results in vivo before human clinical trials begin. Infected animal models with similarities to the human form of CL should be more representative when considering the effectiveness of a compound within a human. Models with dissimilar clinical presentations should be avoided. The accurate quantification of parasite load and, thus, effect of a drug must then be determined through the use of reliable quantification methods as described. It should also be noted that the overall curative effects of anti-leishmanial drugs in vivo depend not only on antiparasitic activities of the agents but also on immune responses of the host. Hence, understanding the immunopathology of the animal model is important when selecting agents for clinical development.

In conclusion, we found that the following models are the most suitable for the assessment of anti-leishmanial drugs: L. major–C57BL/6 mice or–vervet monkey, L. major–rhesus, L. tropica–CsS-16, L. amazonensis–CBA, L. braziliensis–golden or L. braziliensis–rhesus. We also found that the other models investigated here are not currently supported. The guidance provided in this report should help when considering a CL animal model for drug discovery, and we suggest that further work should focus on these recommended models.