In order to more thoroughly understand human diseases, most biologists rely heavily on the use of analogy, i.e., researchers will observe or conduct controlled experiments in non-human animal models in hopes of elucidating information that is relevant for human disease. How researchers approach animal models has been shaped by the textbook definition that defines an animal model “as a living organism with an inherited, naturally acquired, or induced a pathological process that in one or more respects closely resembles the same phenomenon occurring in man." [3] This definition was then expanded upon and adopted by the American National Research Council Committee on Animal Models for Research and Aging to include any “living organism in which normative biology or behavior can be studied, or in which the phenomenon in one or more respects resembles the same phenomenon in humans or other species of animals." [4] The choice of animal disease model typically rests on whether the model is suitable in terms of face validity (similar disease symptom presentation), predictive validity (how well the model predicts human treatment outcome) or construct validity (both the symptoms and pathology mimic the human condition) [5]. Currently, biomedical research relies heavily on rodent disease models as they are mammalian, share a high degree of genetic identity with humans and have been utilized for a long period of time. This reliance has effectively established rodents as the “gold standard” model. However, many would argue that no animal model is ideal nor universally fit for all aspects of the study and that a diversified approach strengthens mechanistic conservation arguments [6].

Toward that end, we will discuss two degenerative diseases that have been studied extensively using various animal models : Huntington's disease and Parkinson's disease. We will highlight some of the strengths and weaknesses of each, keeping in mind some of the criteria that researchers consider when selecting an animal model that will be used to address their specific experimental question (Table 1).

Huntington’s Disease (HD) is a debilitating neurodegenerative disorder caused by an autosomal dominant expansion of cytosine-adenine-guanine (CAG) repeats in the first exon of the huntingtin (Htt) gene [7]. Typically an individual with 35 repeats or less will remain asymptomatic, with those having 40-70 repeats developing the disease; individuals with greater than 70 repeats will invariably develop HD during childhood [8]. This abnormality leads to selective degeneration of specific neuronal populations, eventually causing behavioral and physical movement defects [9]. The incidence of the disease is relatively rare in terms of affected individuals worldwide but is progressive, with disease onset often occurring during middle age and lasting for 15-20 years [10, 11]. Presently, tetrabenazine (marketed as “Nitoman” by Roche) is the only FDA-approved drug available that specifically treats HD symptoms [12], although there is some concern about the true efficacy of the drug [13, 14]. Currently, there are no curative therapeutics on the market.

A major obstacle to developing therapeutic treatments is the lack of understanding in regards to HD pathogenesis. For this reason, researchers have focused on animal models that closely mimic various aspects of the disease. However, there are no animal models to date that encompass all known HD features. The most popular models are the transgenic rodent models, in particular the R6/2 mouse [15-17] ; approximately 50% of all polyglutamine disease research are conducted with this particular strain [18]. For example, Pido-Lopez J et al evaluated the inhibition of tumour necrosis factor alpha by etanercept treatment in the R6/2 mice [19].

R6/2 express the N-terminal fragment of Htt’s first exon and display some of the progressive behavioral/pathological defects associated with the disease within 2-3 months of age (see Fig. 1 [20] ). As with other HD animal models, regardless of species, the R6/2 mice display no gross morphological problems (Fig. 1). They have the additional benefit of low strain variability, allowing researchers to utilize as little as ten mice to detect outcome differences as low as 10% [21]. However, the R6/2 model is not a precise replicate of all symptoms found in HD patients. For example, the relatively rapid onset of the disease more closely correlates to juvenile HD while the R6 mice display symptoms more in keeping with the adult HD form. Additionally, R6/2 display visual-spatial learning deficits as measured by Morris water maze tests [22] ; however, this does not necessarily correlate with HD patients who display cognitive defects in areas such as executive function due to striatum degeneration [23]. Lastly, there may be neuropathological differences in that the proteolytic mechanism possibly involved in producing mutant Htt fragments, a hallmark of human HD, does not appear to be required in the transgenic rodent models [24]. Taken together, this data suggest that these mice should be considered disease-relevant models based on their face versus construct validity.

[enlarge]

Figure 1. Murine Hungtinton’s Disease Model. A. R6/2 transgenic mice (Jackson Laboratory Huntington Disease model 006494 B6CBA-Tg(HDexon1)62Gbp/3J; image courtesy of Jackson Laboratory). As with other HD animal models, there are no discernible gross morphological problems with the R6/2 mice. B. Htt accumulation in murine Huntington Disease model. Coronal sections from 14 weeks old WT and R6/2 mouse brain stained for Htt detection.(from [1] )

Although most studies are conducted in R6/2 mice, other rodent models (transgenics expressing fragments or full-length mutant Htt) are being utilized [21], like Hdh^Q111/Q7 [25], zQ175 mice [16], or YAC128 mice [17], or Hdh^Q111/Q111, Hdh^F7Q/+ and Hdh^F140Q/+ mice [26]. Cross-breeding of BACHD mouse strain with full-length mutant HTT [27] and YAC18 mouse strain with full-length human wild-type HTT [28] on the Hdh−/− background has generated a fully humanized transgenic mouse model of Huntington disease Hu97/18, which re-generated the precise genetic makeup of human HD patients and displayed similar motor, psychiatric and cognitive deficits as well as canonical neuropathological abnormalities as in HD [29]. Many of these strains differ in regards to disease onset; while R6 mice develop HD symptoms rather quickly (see above), other models may not demonstrate strong HD-relevant behavioral problems until two years of age [30]. In addition, there may be other pharmacological differences. For example, N171-82Q mice contain a larger N-terminal fragment of the Htt gene compared to R6/2 mice and present a similar but less severe phenotype than R6/2 [31]. The N171-82Q strain was originally created in an effort to more closely mimic the pathologic Htt accumulation typically seen in humans, structures that were not originally reported in R6 mice but have been reported more recently (Fig. 1 [1, 15, 32, 33] ). Nevertheless, these mice are not utilized as often as R6/2 due to a weaker phenotype and higher strain variability. However, in head to head experiments where researchers assayed the therapeutic benefit of rapamycin, the researchers observed that the compound alleviated symptoms in N1717-82Q but had no effect in R6/2 [34]. These differences could simply be attributed to the fact that N171-82Q present a less severe phenotype than R6/2 and therefore are more susceptible to therapeutic treatments. However, due to the pathology and symptom onset of N171-82, some argue that these mice are a more predictive model of HD and that these types of preclinical results should be given further weight [21].

Aside from obvious disparities between the various transgenic strains as illustrated above, many subtle differences such as genetic background or sex may impact results. Due to the method by which R6/2 were initially generated, these mice may also be homozygous for the rd1 mutation (rd1 affects the rod beta-subunit of the cGMP phosphodiesterase); typically the full R6/2 genotype is not reported. However, researchers have observed that whether the mice are rd1 homozygous or heterozygous can affect motor function outcomes and reduce differences seen between experimental and control, making an accurate interpretation of the results difficult [35]. In the rat HD model, which contains a truncated Htt and develops many motor defects / neuropathology typical of HD within a year, it was recently demonstrated that there are sex-related differences in regards to the number of neurons affected and cognitive deficits [36-38]. These differences were further exaggerated by whether the rats were homozygous or heterozygous for the mutated Htt transgene [39].

With the wide variety of factors that can affect experimental outcomes, it is not surprising that none of the rodent models are considered a “gold standard” for preclinical studies; to date, none of these models have proven to be predictive of clinical trial success. This becomes problematic in light of the cost of drug discovery for disease-relevant therapeutics.

Due to recent trends that have increased the cost of delivering a drug to market (estimated cost $800M - $1B), pharmaceutical companies are seeking ways to increase efficiency and decrease costs in the drug development process [40]. New advanced methods (in silico screens, combinatorial chemistry, etc.) are applied at earlier stages, leading to an increased number of leads being generated. In the case of HD specifically, there were approximately 250 therapeutic targets in the exploratory phase of target validation in 2011; many of these were initially identified by isolating proteins that interacted with either full-length or partial Htt fragments in yeast. But biological validation of lead compounds in rodents is both costly and time-consuming, leading to a sharp decline in the number of promising therapeutics that can be studied at a given time; as of 2011, there were only twenty compounds that were being studied at the preclinical in-vivo stage of HD drug development. This problem is compounded by a large number of late-stage failures that have risen over the past decade, offsetting any values gained by current high-volume early-stage approaches [40]. In essence, the pharmaceutical industry needs better methodologies for screening large numbers of compounds directly in living animal models.

Animal models proven useful for drug discovery are available in the HD field, in particular C. elegans (worms), D. melanogaster (flies) and D. rerio (zebrafish) [41-46]. These models are typically transgenic in nature, expressing various lengths of Htt. None of these systems, as with the HD-relevant rodents, may be considered true construct valid models but all three are amenable to forward genetic screening. Forward screening, as opposed to the more traditional reverse screening, allows enhancers and suppressors of deficient Htt to be identified without prejudiced of prior knowledge, thus, useful for ascertaining new genes and signaling pathways that may underpin the disease. In addition, available methodologies in both flies and worms allow researchers to perform genome-wide screens for genes that regulate mutant Htt aggregation (for examples, see [47, 48] ); as a corollary, chemical suppressor screens aimed at inhibiting Htt aggregation can be conducted in zebrafish [49]. Moreover, all three model systems are well suited for large-scale drug screening efforts; they are relatively inexpensive to house, highly fecund and amenable to both quantifiable robotic/automated assay protocols as well as forward chemical screening methodologies [50]. In these models, multiple variables relevant for HD pathology can be analyzed such as neuronal cell loss (see Fig. 2 for example), behavioral defects, protein aggregations and shortened lifespan [51]. Examples of how positive hits in a “lower” HD animal model obtained similar results in mammals and success with FDA approved drugs in whole organism screening suggests relevance for human HD therapeutics [2, 52, 53]. In addition, testing drug candidates in multiple models strengthens the argument that the drugs in question will be relevant for higher animals by establishing a “conserved mechanism” model [6]. For example, one study isolated novel HD therapeutic targets by first conducting a screen in an in vitro cell model system, then confirming the results in worms, flies and zebrafish, suggesting that they will be more clinically relevant for humans [54].

[enlarge]

Figure 2. Drug testing in HD worm model. Compounds lithium chloride (LiCl) and mithramycin (MTR) were tested for their ability to increase neuronal survival, protecting against polyglutamine mediated death in worms. (from [2] )

It is possible that some candidate drugs may go directly from preclinical testing in lower animal models to humans, in particular, if the drugs were already FDA approved for alternative treatments. However, if the therapeutics mechanism were to involve gene therapy, it would be necessary to examine the various aspects such as the site of injection delivery, the extent of brain tissue affected etc in higher mammalian models that have more comparable brain neuroanatomy to humans [55, 56]. Currently, sheep models are being verified for these purposes [57]. While large domesticated HD animal models are in the process of being established, most researchers interested in conducting non-rodent mammalian studies utilize non-human primates such as capuchin, cynomolgus and rhesus monkeys plus baboons [58, 59]. Research from this subfield has already given rise to Phase I clinical trials involving the use of polymer-encapsulated cells engineered to secrete human ciliary neurotrophic factor [60]. Typically HD is induced pharmacologically in primates (e.g 3-nitropropionic acid [61], ibotenic acid [62], quinolinic acid [63] ) or through lentiviral transfer of a mutant HTT transgene [59]. Treated primates then present with symptoms and neuronal loss patterns similar to humans. As with all previously mentioned models, it is unclear whether the disease etiology in primates is the same as in humans and therefore, these models do not serve as “construct valid” models of HD. For example, when capuchin monkeys were treated with two varying types of pharmacological agents (quinolinic acid and 3-nitropropionic acid) to induce HD, they presented with similar behavioral characteristics although the lesion patterns were different [61].

A number of new mouse models of HD have recently been described. In addition to the transgenic HD mice discussed above, a number of knock-in mice have been developed in which various numbers of CAG repeats encoding glutamine (Q) are inserted into the mouse Htt gene (for example, Hdh140Q knock-in mice [64] ). These mice develop milder versions of the behavioural phenotypes observed in transgenic HD mouse models. These knock-in mice have provided evidence that proteolysis of full-length mutant Htt, to generate the corresponding N-terminal fragments, may be a key step in the age-dependent pathogenesis of HD [65].

As mentioned earlier, there is interest within the scientific community in developing large (non-primate) animal models of HD. One of the drivers is the lack of marked neurodegeneration observed in mouse models which fail to recapitulate the situation in HD patients [65]. Sheep have proved attractive as a model since the development of brain regions most affected by HD are similar in sheep and humans. A transgenic sheep model has been established in which expression of full-length Htt protein containing 73 polyQ repeats is driven by the human promoter [66]. The long life-span of sheep (>10 years) opens the opportunity for studying the long-term development of HD pathology whilst the large brain size makes PET and MRI analysis possible. Thus, it may be possible to study the relationship between the progression of disease symptoms and anatomical changes within the brain [66].

Another development is the generation of conditional mouse HD models that express mutant Htt in specific cell types (as opposed to the global expression in the models described above). Transgenic mice have been generated that express full-length human mutant Htt containing 97 polyQ repeats. These transgenic mice are constructed such that it is possible to genetically reduce expression of mutant Htt in either stratum or cortex or in both of these brain regions and observe the effects of the reduced expression on the HD phenotype [67].

Pigs are closer to humans than small mammals in regard to anatomic structures and physiology. Pigs have a relatively quick breeding period versus non-human primates, and have an average of 7 piglets per litter, which is advantageous when utilizing them as a model of human disease. Finally, genetic tools such as CRISPR/Cas9 and somatic cell nuclear transfer have provided a way to modify endogenous genes in pigs and create non-chimeric first-generation offspring. For these reasons, pig models of human diseases are being developed and utilized in many areas of research. Previously described pig models of HD have lacked the desired expression of full-length mutant huntingtin or neurodegeneration necessary to utilize the models to study pathogenesis and treatment options. Yet the benefits of a non-primate large animal model ensure a continued push for the development of a pig model. Recently, Yan et al published details of the development and characterization of an HD knockin pig model that does express full length mutant Htt at endogenous levels and shows neurodegeneration similar to that seen in the human disease [68]. This knockin model is unique in that CRISPR/Cas9 was used to edit fetal fibroblast cells with a contain heterozygous expanded human HTT exon1 with 150 CAG repeats within the pig HTT gene. The fibroblasts underwent SCNT and embryos were put into surrogate pigs, ultimately yielding six piglets carrying the human HTT exon1 with repeats. Subsequent mattings have yielded 15 F1 pigs and 10 F2 pigs, all positive for the mutant HTT but with varying CAG repeat numbers. The further characterization of these K1 pigs by Yan et al demonstrated the model exhibits multiple symptoms typically found in human HD but not in mouse models, including neurodegeneration of the medium spiny neurons and striatal neurons, respiratory problems and movement disorders [68]. This model of HD shows great promise as a tool in the development of HD therapies and provides several advantageous over mouse models of the disease.

Using multiple species to model human diseases has the advantage of highlighting evolutionarily conserved mechanisms which in turn can strongly suggest that therapeutics aimed at these conserved nodes will be successful during human clinical trials. However, as has been discussed, each animal model comes with caveats, never fully mirroring the human disease. In addition, certain species are better suited for varying types of experimental designs. These statements should not be used as an argument to do away with animal research. Instead, they should serve as a reminder that experiments must be carefully controlled/designed from the onset and examined for assumptions prior to data interpretation. The scientific community is continually developing new animal models of both HD and PD. This reflects both a desire to generate better models to explore specific aspects of these neurodegenerative diseases and an on-going drive to have available animal models that better replicate the human diseases. The hope is that such models will give greater insight into the precise biological mechanisms underlying the pathogenesis of these devastating neurodegenerative disorders. It is also hoped that better models will be better able to predict the clinical efficacy of potential new therapeutics and help deliver much needed new treatments.

Dr. Jennifer L. Walker contributed the discussion on pigs in December 2018. The original article "Animal Models for Huntington's and Parkinson's Diseases" was separated into two articles "Huntington's Disease Animal Models" and "Parkinson's Diseases Animal Models" in November 2019.