Animal Models for Neural Diseases

Abstract

“Animal Models of Neural Disease” was the focus of General Session 5 at a 2010 scientific symposium that was sponsored jointly by the Society of Toxicologic Pathology (STP) and the International Federation of Societies of Toxicologic Pathologists (IFSTP). The objective was to consider issues that dictate the choice of animal models for neuropathology-based studies used to investigate neurological diseases and novel therapeutic agents to treat them. In some cases, no animal model exists that recapitulates the attributes of the human disease (e.g., fibromyalgia syndrome). Alternatively, numerous animal models are available for other conditions, so an essential consideration is selecting the most appropriate experimental system (e.g., Alzheimer’s disease). New technologies (e.g., genetically engineered rodent models) promise the opportunity to generate suitable animal models for syndromes that currently lack any in vivo animal model, while in vitro models offer the opportunity to evaluate xenobiotic effects in specific neural cell populations. The complex nature of neurological disease requires regular reassessment of available and potential options to ensure that animal-derived data sets support translational medicine efforts to improve public health.

General Session 5 at the 2010 scientific symposium, sponsored jointly by the Society of Toxicologic Pathology (STP) and the International Federation of Societies of Toxicologic Pathologists (IFSTP), covered some of the considerations in selecting and using animal models for safety and efficacy studies that include neuropathology. Several nuances were discussed, including diseases where suitable animal models do not exist for a human disease (fibromyalgia syndrome), situations where animal models are numerous and selection of the most appropriate one is critical (Alzheimer’s disease), and circumstances where an appropriate animal model does not currently exist but could be constructed (e.g., genetically engineered rodent models). In some cases, neurotoxicity itself may need to be modeled using xenobiotics to probe the reactions of cultured neural cells. This minireview summarizes the major points made regarding these four scenarios by the four speakers in this session.

Dr. Gary Jay (Pfizer Global Research, New London, CT) presented a seminar describing fibromyalgia syndrome (FMS), a limbically activated pain disorder. In humans, FMS is a generalized increase in the perception of sensory stimuli with central sensitization, allodynia (pain due to a sensory stimulus that does not normally provoke pain, such as electrical or mechanical or thermal), and hyperalgesia. It is defined by stiffness, easy fatigability, general achiness, diffuse skeletal pain, a sensation of joint swelling, mood disorder (depression and anxiety in 20-40% of patients), nonrestorative sleep, nocturnal myoclonus, cognitive difficulties, and poor quality of life. FMS is more commonly observed in women than in men. The trigger for this syndrome remains unknown. Similarly, FMS does not appear to exhibit a consistent set of neuropathologic changes (Jay 2007a, 2007b, 2009).

The pathoetiology of FMS is believed to be related to “central hypersensitivity” caused by biological-psychological-sociological problem(s). Abnormal facets of this syndrome include psychophysiologic, musculoskeletal, and cerebral abnormalities (prolonged distress, myofascial pain, “pain behavior,” anxiety and depression, central sensitization), and neuroendocrine and autonomic nervous system dysfunctions. The combination of allodynia and hyperalgesia indicates that both peripheral and central nociceptive abnormalities contribute to the syndrome. Peripheral nociceptive systems in the skin and musculature change significantly, with sensitization of vanilloid receptors, acid-sensing ion channel receptors, and purino-receptors. Tissue modulators of inflammation and nerve growth factors can excite these receptors, leading to significant changes in pain sensitivity. In FMS patients, however, there is no consistent evidence of soft tissue inflammation, leading the search for the initial pathoetiology to the central nervous system (CNS). While there are no “gold standard” tests that can be performed to confirm the diagnosis of FMS, there are enough known abnormalities in the neuroendocrine system, the autonomic nervous system, and the neurotransmitter/neuropeptide systems that improved diagnostic tools based on biochemical or molecular aberrations may be available in the near future (Jay 2007a, 2007b, 2009).

Three animal models in rodents have been used to study FMS. The first consists of multiple intramuscular (IM) injections of acidic saline, which is thought to induce central sensitization; increase descending facilitation; enhance glutamate release; and produce widespread cutaneous, muscle, and visceral pain. The second model uses subcutaneously administered reserpine to induce biogenic amine depletion, which leads to chronic muscular pain, tactile allodynia, and depression. The third model is the Otsuka Long-Evans Tokushima Fatty (OLETF) rat, which has increased exploratory behavior, increased anxiety, impaired learning, and disturbances of regulation of body temperature. These animal models are mechanistic in nature, and for the most part model separate parts of the syndrome, but not FMS in its entirety. While there may be some neurochemical similarities, the animal models appear to be too simplistic to replicate the complex changes that characterize the human syndrome. The limbic system contributes significantly to FMS (since the ability to undertake learned behavior, especially with emotional aspects, is limbically oriented). This behavior, it appears, most typically results from a negative experience such as trauma (sexual, physical, and/or emotional), especially if it occurs during childhood. The rat’s limbic system is poorly developed when compared with the human limbic system, so this aspect of human FMS cannot be explored fully in animal models.

Dr. Ronald B. Demattos (Eli Lilly and Company, Indianapolis, IN) presented considerations surrounding selection among the numerous animal models of Alzheimer’s disease (AD) to assist drug development. While no animal model fully recapitulates all the features of AD, various animal models are useful to preclinically investigate selected genetic, mechanistic, or safety features of the disease. Available models may be categorized into non-transgenic (spontaneous) or transgenic classes. The non-transgenic category includes primate models (vervets, lemurs, cotton-top tamarins, rhesus macaques, squirrel monkeys, and chimpanzees), which deposit amyloid beta (Aβ) in the brain parenchyma and cerebral vascular walls (i.e., cerebral amyloid angiopathy [CAA]); canine models, which have deposition of Aβ without neurofibrillary tangles; and other models in rabbits and rodents (guinea pigs, rats, mice, and hamsters). Most animal models of AD currently in use are mouse transgenic models, which express various human mutations of amyloid precursor protein (APP). Such amyloidogenic mutations have been described for gamma-secretases and presenilins, beta-secretase, APP central domain, and the apolipoprotein E (ApoE) and J (ApoJ) sites. Expression of these transgenes results in Aβ deposits in neural parenchyma, cerebral vessel walls (CAA), or both. However, the Aβ-producing mouse models do not develop neurofibrillary tangles and typically lack neuronal loss. In contrast, tauopathy mouse models have been generated in which human tau protein is expressed and hyperphosphorylated, but these do not develop Aβ deposits or CAA. To get closer to the human entity, in which accumulation of both Aβ and hyperphosphorylated tau is accompanied by neuronal loss, mice transgenic for 3 or 5 mutant human proteins have been developed. These mice develop robust deposits of Aβ and hyperphosphorylated tau and have some neuronal loss.

One well-characterized transgenic mouse model of AD is the PDAPP (platelet-derived growth factor promoter expressing amyloid precursor protein) transgenic mouse (Games et al. 1995). This mouse overexpresses human APP^v717F and, starting at 7 to 8 months of age, robustly deposits Aβ-42 in its hippocampus and elsewhere, and lower levels of Aβ-40 in the cerebral parenchyma and cerebral vasculature. However, similar to most transgenic mouse models of AD, PDAPP mice fail to develop prominent neuronal degeneration or neurofibrillary tangles. One attractive feature of this animal model is that levels of cerebrospinal fluid (CSF) and plasma Aβ match those of humans fairly closely; some other transgenic mouse models (e.g., Tg2576 mice) have levels well in excess of those found in AD patients. The PDAPP model was used in Lilly’s development of an anti-Aβ monoclonal antibody (LY2062430) designed either to capture Aβ peripherally and stimulate its efflux from the brain or to cross into the brain parenchyma and bind to soluble monomers of Aβ. As a biomarker of functionality of the former mechanism, elevated plasma and CSF Aβ-42 levels occurred in PDAPP mice given this monoclonal antibody, and these correlated with reductions in immunohistochemically detectable Aβ in mouse brain microscopic sections. Similar elevations of plasma and CSF Aβ-40 were seen, suggesting that LY2062430 had functionality for both the 1-42 and 1-40 forms of Aβ.

One potential safety risk of removing Aβ is cerebral hemorrhage (Nicoll et al. 2003). It is hypothesized that increased CAA is a risk factor for such hemorrhage, that preclinical species can adequately model the likelihood of such CAA-associated hemorrhage, and that assessment of the safety risks of immunotherapy in a CAA-bearing model would be a suitable means to assess this potential. To address this risk, Lilly chose to identify an animal model possessing significant CAA, characterize the age when CAA would be in its developing stages, and assess the safety of its passive immunotherapy in that model. To this end, they determined that 20- to 24-month-old PDAPP mice develop significant CAA, and that there was significant Aβ-40 in vessel walls of such mice. A 6-week study in such aged PDAPP mice was selected and successfully used to preclinically characterize the lack of hemorrhagic potential following mobilization of Aβ.

Dr. Edward J. Weinstein (Sigma-Aldrich, St. Louis, MO) presented a talk titled “Unmet Needs for Animal Models” highlighting the systematic generation of knock-out (KO) and knock-in rat models for neurologic conditions. The advantages of rats as models for neural diseases include that they are easy to rear, manageable and of convenient size for the laboratory, inexpensive to maintain, have reasonably short life and reproductive cycles, and are large enough for repeated physiological measurements. Furthermore, the neurobiological attributes of rats as well as the ability to selectively manipulate their genome are standardized enough that results in this species that can be generalized to other mammalian species with minimal difficulty. The large database of rat-based toxicologic studies provides the final impetus for generating new animal models of neurological disease in this species. The Sigma-Aldrich genetic engineering laboratory has developed KO rats via embryo microinjection of zinc-finger nucleases (ZFN). This technique uses artificial restriction enzymes generated by fusing a zinc finger DNA-binding domain to a DNA-cleavage domain. Zinc finger domains can be engineered to target a desired DNA sequence, which enables zinc-finger nucleases to target unique gene sequences within a large, complex genome. By taking advantage of endogenous DNA repair machinery, these enzymes can be used to precisely alter the rat genome. The time from gene sequence to identification of founder animals can take up to 12 weeks, which is substantially faster than KO fabrication by traditional homologous recombination methods. Creation of KO rats for neurology conditions has included targeted deletion of ApoE or APP (for AD), leptin (appetite disorders), brain-derived neurotropic factor (BDNF, for depression, schizophrenia, obsessive-compulsive disorder, AD, Huntington’s disease, Rett syndrome, dementia, anorexia and bulimia nervosa), Disrupted-in Schizophrenia 1 (DISC1, for schizophrenia), and multiple proteins that participate in Parkinson’s disease (e.g., PTEN-induced putative kinase 1 [Pink1], parkin, α-synuclein, DJ-1 and, Lrrk2), Fmr1 (autism), and Mecp2 (Rett syndrome), among other molecules/targets.

Dr. Martin Philbert (University of Michigan, Ann Arbor, MI) concluded the afternoon’s session by discussing the use of 1,3-dintrobenzene (DNB) as a tool to model neurotoxicologic mechanisms in mammalian cells in vitro. This neurotoxicant causes acute energy deprivation and induces oxidative stress, especially in proteins in mitochondria, which results in vacuolation of neural parenchyma (Phelka, Beck, and Philbert 2003). Using astrocytes as a model system to examine such stresses, DNB has been shown to induce changes in the permeability of cell and mitochondrial membranes. These morphological changes, such as mitochondrial swelling, can be quantified following DNB exposure. These data indicate that the biophysical modifications induced by neurotoxicants, including changes in both mitochondrial shape and membrane potential, may be just as important in determining the scope and progression of neurotoxicity as are the overt biochemical alterations that are typically imputed to be the primary causes of neurotoxicity.

In summary, General Session 5 covered one example for each of the four scenarios: where suitable animal models do not exist for a human disease; where animal models are numerous, and selection of the most appropriate one is critical; where an appropriate animal model does not currently exist but could be constructed; and where neurotoxicity itself needed to be modeled. These four topics were chosen as representatives of four common scenarios that experimental neuropathologists should consider when asked to select or evaluate animal models of human neurological diseases.