Abstract
Alzheimer’s disease (AD) is a growing neurodegenerative disease without effective treatments or therapies. Despite the use of different approaches and an extensive variety of genetic amyloid based models, therapeutic strategies remain elusive. AD is characterized by three main pathological hallmarks that include amyloid-β plaques, neurofibrillary tangles, and neuroinflammatory processes; however, many other pathological mechanisms have been described in the literature. Nonetheless, the study of the disease and the screening of potential therapies is heavily weighted toward the study of amyloid-β transgenic models. Non-transgenic models may aid in the study of complex pathological states and provide a suitable complementary alternative to evaluating therapeutic biomedical and intervention strategies. In this review, we evaluate the literature on non-transgenic alternatives, focusing on the use of these models for testing therapeutic strategies, and assess their contribution to understanding AD. This review aims to underscore the need for a shift in preclinical research on intervention strategies for AD from amyloid-based to alternative, complementary non-amyloid approaches.
Alzheimer’s disease (AD) is a neurodegenerative disease characterized by amyloid-β (Aβ) plaques, tau tangles, and neuroinflammation, which have been found to alter structure and function of dendritic spines. Synaptic failure is associated to cognitive, memory, and learning dysfunction in AD [1, 2]. Two types of AD have been described: familial AD, accounting for ∼3– 5% of all AD worldwide, and sporadic Alzheimer’s disease (sAD), the more common (∼95%) and less well understood [1]. Importantly, while familial AD is due to genetic variants like mutations in amyloid precursor protein (APP), presenilin 1 (PS1), and/or presenilin 2 (PS2) [3], sAD has environmental and genetic risk factors that contribute to disease burden [4]. Environmental risk factors can include metabolic disorders, chronic diseases, aluminum toxicity, and head injuries [5]. Whereas, the two major genetic risk factors of sAD include two or more copies of the Apolipoprotein E4 (APOE4) gene and the TREM-2 gene, which encode the triggering receptor expressed on myeloid cells 2 [6]. Although familial AD and sAD share common aspects of the pathophysiology such as Aβ toxicity, neuroinflammation, oxidative stress, and tau tangles, there are multiple differences associated to the development and presence of clinical symptoms, genetic impact, variances in hippocampal volume loss, white matter abnormalities, and parietal atrophy [7].
Despite multiple efforts to develop therapeutic strategies for AD, these have fallen short in clinical trials [8]. One factor that has contributed to these shortcomings is that AD pre-clinical research has largely focused on transgenic models of disease. These models have been developed from genetic mutations described in familial AD [9]. Although advantageous in many ways, these models represent pathologies associated with familial AD and are at best incomplete models for sAD. One strategy that may aid in the evaluation of potential therapeutics is the use of non-transgenic animal models in preclinical studies.
Non-transgenic models include the study of the conventional AD pathological biomarkers, amyloid and tau, but also allow for the modeling of other pathological states, such as oxidative stress, synaptic dysfunction, and apoptosis, as well as alternative theories of sAD etiology and pathogenesis, such as type 3 diabetes (insulin resistance state), neuroinflammation, gut microbiota-brain axis alterations, lipid metabolism abnormalities, autophagy dysfunction, and metal ion disorders [7, 10]. Animal models have been used to study all aspects of sAD including non-rodent models like the cholesterol-fed rabbit model [11], the aged dog and non-human primates (for a review, see [12]). However, we will focus on rodents since these are the most used and represent a cost-efficient alternative [13].
In this review we will critically evaluate non-transgenic experimental rodent models of AD, and focus on testing of potential therapeutics. We compiled studies for the well-established non-transgenic animal models such as Aβ peptide infusion, STZ-icv model, high fat diet model, D-galactose aging induced model, aluminum toxicity induced AD model, and Adeno-associated virus (AAV) delivery for studying AD, evaluated them as amyloid or non-amyloid based models, and classified findings in different therapeutic domains (i.e., antidiabetics, antioxidants, anti-inflammatories, Chinese medicine, agonist and antagonist receptors). We propose that non-transgenic experimental animal models are a suitable approach to the evaluation of therapeutic strategies aimed toward treatment of AD symptoms and/or pathology.
AMYLOID-BASED MODELS
Amyloid-based models have largely relied on the use of transgenic animals for the expression of mutations associated to familial AD in the genes for APP, PS1, and PS2. These mutations lead to alterations in amyloid expression and other pathologies [14]. However, non-transgenic amyloid based models have also been reported and are discussed below.
Aß infusion model
The Aβ infused model is mainly based on the amyloid cascade hypothesis. It is characterized by changes in Aβ metabolism, like increases in total Aβ production and reductions in Aβ clearance, leading to Aβ accumulation and oligomerization (for a review, see [15]). These changes have been reported to lead to increases in inflammatory response, oxidative stress, progressive synaptic dysfunction, hyperphosphorylation of tau, and cell death [15]. The most reported hallmarks of AD instantiated in this model are Aβ markers, neuroinflammation, and oxidative stress (Table 1). To mimic Aβ pathology in rodent animal models, different protocols have been reported (for a protocol of how to generate and Aβ infused model, see [16, 17]). Aβ1–42 and Aβ25–35 are the most common peptides reported to be infused in vivo models due to their high toxicity [18, 19]. The targeted areas include the CA1 and dentate gyrus regions of the hippocampus [20] since these areas are implicated in learning and memory and mostly affected by neurodegeneration in AD [21]. Also, the lateral ventricles are often targeted to achieve general diffusion of the peptide [22, 23].
Previous studies using different therapeutic strategies in Aβ infused model
Aβ, amyloid-β protein; ET, endothelin; GABA, gamma-Aminobutyric acid; PCAF, P300/CBP-associated factor; MWM, Morris water maze test; PA, passive avoidance test; NOR, novel object recognition test; BM, Barnes maze; OTC, operant test chamber.
This model has two major limitations. First, during the procedure of infusion, the needle could cause more damage than expected if surgery is not done properly, leading to additional cell death and inflammation. Second, the injection site limits the area were the peptide is infused, thus, Aβ effects are only produced in one specific region [16]. To overcome the previous disadvantages, the combination of different non transgenic models in one animal to express different aspects of the disease is suggested, such as the use of high fat diet to induce metabolic disorders and Aβ infusion [24, 25].
The advantages of using Aβ infused model are related to reduced action time and replacement of transgenic models. The main advantage is that neuronal death occurs near the injection site within one week [26]. Memory impairment and cognitive deficits have been reported also in short periods of time, as early as seven days post infusion [27, 28] making this model a timely alternative to transgenics which require breeding time. Moreover, it obviates the possible side effects (i.e., compensatory or homeostatic changes due to transgene expression) from the mutations introduced in transgenic rodent lines [17]. Lastly, this method allows the assessment of oligomerization on Aβ associated pathology. Studies that have employed this model suggest that soluble oligomers, and not monomers or fibrils, are most toxic (for a review, see [29]). However, caution is warranted in that often the state of oligomerization is unknown and dilution protocols poorly reported, although reported methods for achieving different oligomerization states are published (see examples, Walsh 2009 [30], Stine 2011 [31]).
Overall, the use of the Aβ-infused model is a reliable choice for testing different therapeutic strategies and assessing their effects in molecular, cellular, and behavioral aspects of the early physiopathology in sAD. Antioxidant, anti-inflammatory, mixed action antioxidant and anti-inflammatory, anti-diabetic compounds, and a diverse list of other therapies have been evaluated. Most often these compounds reduce neuroinflammation, oxidative stress and enhance memory (Table 1). This model also allows further research into the biological role of amyloid in the pathology of sAD as many recent clinical trials using amyloid-based strategies have failed [8].
NON-AMYLOID BASED MODELS
Non-amyloid based models include diverse strategies that lead to the symptoms and/or pathologies associated with sAD. These include memory loss, amyloid markers, phospho-tau, neuroinflammation, production of reactive oxygen species (ROS), cholinergic or synaptic dysfunction, and cell death. Because sAD is characterized by a complex interaction between genes, environment, and lifestyle, other non-transgenic models have been established to provide alternative approaches.
Streptozocin-induced AD model (STZ-icv)
Streptozotocin (STZ, 2-deoxy-2– (3-(methyl-3- ni-trosoureido)-D-glucopyranose) is a compound de-rived from Streptomycetes achromogenes. Hoyer in 1998 proposed the hypothesis that sporadic late-onset AD is caused by brain insulin resistance state [33]. For this purpose, an experimental animal model was developed by infusing STZ intracerebroventricularly in lower doses compared to the systemic diabetes experimental model [34–36]. STZ-icv infusion is relatively consistent in surgical technique but time post infusion is variable. STZ is infused mostly in doses of 3.0 mg/kg, uni-or bi-laterally into the lateral cerebral ventricles, with a waiting period of 21 days before starting molecular and behavioral assays [37–39].
Intracerebroventricularly administration of STZ produces morphological changes in brain regions such as ventricle enlargement and corpus callosum thickening [40]. These changes are accompanied by reduced neurogenesis and increased neuroinflammation in the periventricular areas and the dorsal hippocampus [41]. STZ has also been shown to produce oxidative stress, deficiency in cholinergic transmission, alterations of insulin receptor (IR) signaling, and memory impairments [42, 43].
This model has been widely used to study the insulin mechanism involved in the brain and its relationship to AD. Dysfunction in brain glucose/energy metabolism and insulin signaling have been documented to play a role in early sAD pathology [44]. This evidence has led to the proposal of a “type 3 diabetes” hypothesis, which can be grossly understood as a state of brain insulin resistance [45–47]. Intracerebroventricular injection of STZ causes central insulin resistance, mainly by decreasing IRs signaling and phosphorylating IRs by protein kinase C [48]. In addition, it has been reported that IRs are commonly found in the olfactory bulb, hippocampus, and hypothalamus [49]. In this regard, the hippocampus is one of the main areas affected by insulin resistance in the brain, as a result, alterations of insulin signaling in the hippocampus affects cognitive functions [50]. However, the molecular and cellular mechanisms of brain insulin resistance are not completely understood. To overcome this statement, Norwitz et al., 2019 has proposed that brain insulin resistance state in AD might be a network of interactions, feedback mechanisms, and cannot only be explained by IRs dysfunction [46].
Although STZ-icv model is relevant for studying impaired glucose/energy metabolism in sAD, female rodents have exhibited high resistance to STZ, thus sex bias can be expected [7]. Nevertheless, strengths of the STZ-icv model include extensive literature in rodent and monkeys to understand the molecular and cellular basis of early sAD [42], but also to evaluate therapeutic strategies that can be useful to study AD treatments [51]. Approved treatments for AD, donepezil and memantine, improve STZ-induced phenotypes and pathologies [52–54], perhaps an indication of predictive validity for drug efficacy. Therapeutic groups tested in this model include antioxidants, anti-inflammatories, anti-diabetics, modulators, and other diverse classes (see Table 2). Effects of these classes are most often reported in reductions of oxidative stress, neuroinflammation, enhanced synaptic/cholinergic activity and memory (Table 2).
Previous studies using different therapeutic strategies in STZ-ICV model
Aβ, amyloid-β protein; PPAR, peroxisome proliferator-activated receptor; hCG, human chorionic gonadotropin; RAS, renin– angiotensin system; mTOR, mammalian Target of Rapamycin; CB, cannabinoid receptor; GLP, glucagon-like peptide; GIP, glucose-dependent insulinotropic polypeptide; NMDA, N-methyl-D-aspartate; GABA, gamma-aminobutyric acid; AChE, acetylcholinesterase; MWM, Morris water maze test; PA, passive avoidance test; RAM, radial arm maze; HB, hole-board task; NOR, novel object recognition test; BM, Barnes maze; EPM, elevated plus maze.
Dietary patterns for studying AD
Dietary patterns have been investigated as one of the modifiable risk factors for cognitive decline, dementia, and other pathologies [4]. Clinical studies have assessed different dietary patterns (e.g., Mediterranean diet, ketogenic diet, DASH diet) and their implications in AD pathology, suggesting that low sugar consumption, high consumption of omega-3 polyunsaturated fatty acids and antioxidant compounds could be useful in preventing cognitive decline [55–57]. Conversely, people who follow an unhealthy diet could increase the risk of AD [56]. In a cross-sectional study, adherence to junk food was a significant predictor of cerebral Aβ deposition [58].
Consumption of high-fat diet (HFD) is a major cause of overweight and obesity, this leads to high levels of circulating free fatty acids which can result in low-grade inflammation by increasing the release of pro-inflammatory cytokines, promoting the activation of downstream signaling, including the serine kinases, Ikkb and JNK1, this entails to the inhibition of IRs signaling, therefore, enhancing insulin resistance, cognitive impairment, and neurodegeneration [49, 59]. Furthermore, high fat and high fructose consumption are related with diabetes and hyperinsulinemia, two risk factors for developing sAD [5]. In fact, a link between obesity and AD has been suggested where insulin resistance leads to regulation of inflammatory processes, adipokine dyshomeostasis, oxidative stress, and mitochondrial dysfunction [60].
Non-transgenic animal models have been useful for studying the molecular and cellular basis of different dietary patterns and their relationship with sAD [61, 62]. For example, Busquets et al. reported Aβ formation in C57BL/6J mice fed with HFD in a long-term period which also exhibited dysregulations in autophagy, apoptosis, and exacerbation in inflammatory processes [63]. Furthermore, HFD and high fructose diet are used to induce metabolic disorders and obesity in animal models [64]. HFD induces brain insulin resistance by decreasing tyrosine phosphorylation of insulin receptor and increasing serine phosphorylation of IRS-1 [65], which can lead to dysregulation in proteins implicated in the insulin signaling pathway [66]. HFD also exacerbates neuroinflammation and oxidative stress and diminishes long term potentiation in the hippocampus [62, 67].
Limitations of this model are related with different contents of diets reported and the variety of time points to evaluate molecular pathways. For example, in an 8-week HFD consumption period, studies report blood– brain barrier (BBB) damage associated with microglia activation, increased oxidative stress and proinflammatory cytokines [68], and memory impairment [69–71]. While 30 weeks of HDF consumption increases the protein expression of phosphorylated-tau protein, amyloid-β protein precursor (AβPP) and β-site APP cleaving enzyme [72], leading to the formation of Aβ plaques and neurofibrillary tangles. Conversely, a short-term consumption period (3, 7, 10 days) of HFD modulates several markers of inflammation, endoplasmic reticulum stress, and apoptosis in the hippocampus of young mice [73]. Therefore, molecular and cellular changes through time in this model are not completely understood and may provide temporally relevant biological insight.
Lastly, the HFD model has been used to test different compounds with antioxidant, anti-inflammatory, anti-glycemic and lipid lowering properties (Table 3). This model is particularly useful in that HFD consumption mimics modern Western diets [59], providing strong face validity for studying the role of nutrition in AD.
Previous studies using different therapeutic strategies in the HFD model
Aβ, amyloid-β protein; MWM, Morris water maze test; PA, passive avoidance test; NOR, novel object recognition test.
D-galactose aging induced model
The primary risk factor for developing AD is age [74]. Age is associated with increased production of ROS. D-galactose is an aldohexose which through the catalysis of galactose oxidase can result in the generation of ROS [75] and is commonly used for studying biological changes implicated in AD, especially oxidative stress (Table 4). Furthermore, it has been reported that generation of ROS can lead to the activation of two pathways: an intrinsic pathway in the cell which affects mitochondrial function, therefore, enhancing apoptosis; and an extrinsic pathway which increases inflammatory response [76]. D-galactose aging induced model is often produced by chronic administration via intraperitoneal and subcutaneous routes [77–79]. However, in low doses and short-term administration it has been reported to have positive effects on learning and memory [80]. The administration of D-galactose in rodents may induce brain maturation comparable to human brain aging; and can also develop hallmarks observed in sAD, including mitochondrial dysfunction [77], oxidative stress [81], neurodegeneration [82], and learning and memory impairment [83].
Previous studies using different therapeutic strategies in D-Galactose aging model
Aβ, amyloid-β protein; MWM, Morris water maze test; PA, passive avoidance test.
This model is used due to its mild side effects and the high survival rate throughout the experimental period [84]. It is often combined with other non-transgenic models to exacerbate different pathologies implicated in sAD and test therapeutic strategies. It has been commonly combined with aluminum chloride (AlCl3) which has exhibited cognitive impairment, hippocampal neurodegeneration, and high expression of p-tau [85–87]. D-galactose has also been reported with Aβ peptide and streptozotocin infusion which exhibit other hallmarks like Aβ accumulation, neuroinflammation, and insulin resistance [88–90]. Lastly, there has been an interest in the combination between long-term D-galactose injection and ovariectomy which has emerged as a novel rodent model for AD [91], especially to investigate the interplay between female hormonal changes and aging. Insofar, this model has largely reported the effects of few antioxidant and anti-inflammatory treatments on Aβ markers, oxidative stress, and learning and memory assays (Table 4).
Aluminum induced AD neurotoxicity model
Environmental exposure to heavy metals such as lead (Pb), mercury (Hg), aluminum (Al), cadmium (Cd), and arsenic (As) have a high degree of toxicity [92]. Heavy metals are known to induce multiple organ damage that can cause cardiovascular diseases, developmental abnormalities, neurologic and neurobehavioral disorders, diabetes, hearing loss, hematologic and immunologic disorders, and are considered human carcinogens according to the US Environmental Protection Agency and the International Agency for Research on Cancer [93]. It has been reported that oral or intraperitoneal administration of AlCl3 induces neurotoxicity [94]; however, there is no clear and well-established reference dose. This model mimics some aspects of the pathogenesis of sAD. It enhances loss of dendritic connectivity in CA1, CA2, and CA3 neurons in the hippocampus [95]; and exacerbates oxidative stress and synaptic activity in the brain [96, 97].
Recently, a novel non-transgenic model that orally provides a metal mixture (As: 3.80 ppm, Cd: 0.98ppm, and Pb: 2.22 ppm) from gestation (prenatal day 5) through early adulthood (postnatal day 90) has exhibited pathological hallmarks of AD such as overexpression of Aβ42, oxidative stress, inflammatory markers, and neuronal apoptosis in frontal cortex and hippocampus [98]. Although the use of heavy metals has not been as vastly reported as other models, it may be a good choice for studying sAD, especially in combination with other non-transgenic experimental manipulations to test different therapeutic strategies (Table 5).
Previous studies using different therapeutic strategies in aluminum chloride model
Aβ, amyloid-β protein; MWM, Morris water maze test; PA, passive avoidance test; RAM, radial arm maze.
Adeno-associated virus (AAV) delivery for studying AD
Wild-type AAVs or adeno satellite viruses are classified under the family Parvoviridae and the genus Dependoparvovirus, and they have a non-pathogenic nature and a single stranded DNA [99]. However, they need a helper virus, like adenovirus, to perform infection and replication [99]. Due to their ability to transduce neurons, their nonpathogenic nature, long-term transgene expression, mild immune responses and ability to cross the BBB, AAVs have been widely used to study neurodegenerative diseases, especially AD (for a review, see [100]). Furthermore, different routes of AAV delivery to the nervous system (e.g., intraparenchymal, intranasal, intracerebroventricular, intravenous) facilitate the study of different aspects of the disease [101].
Available gene delivery mechanisms have added new ways to study early and late onset of AD. One of the reasons AAVs models are an excellent choice for studying AD is that they can express different genes implicated in AD pathology based on different hypotheses, such as: Aβ cascade [102], neuroinflammation [103], and tau pathology [104]. Tau pathology is one of the main hallmarks studied with AAV delivery [105]. For instance, one study delivered AAV-ShRNA Tau in C57Bl6/SJL wild-type mice and found that knockdown of tau in the hippocampus significantly impaired motor coordination and spatial memory [106]. This can be beneficial for studying tau pathology effects in a short period of time compared to other models like Aβ infused model or HFD model where tau pathology is often observed at a later timepoint.
Novel animal models for studying sAD have been developed using AAVs. Iqbal et al. (2013) proposed a sporadic AD model by using adeno-associated virus serotype 1 (AAV1) vector containing I2NTF - CTF. This vector translocates to the neuronal cytoplasm and inhibits protein phosphatase activity and consequently evokes the abnormal hyperphosphorylation of tau [107]. The AAV1- I2NTF - CTF rats show neurodegeneration and cognitive impairment at 4 months and abnormal hyperphosphorylation and aggregation of tau and intraneuronal accumulation of Aβ at 13 months [107]. Moreover, AAVs have been used to study the role of ApoE in sAD. In a recent study, intracisternal delivery of ApoE2 (AAVrh.10hAPOE2-HA) in non-human primates with slightest invasive surgical intervention suggests an ideal procedure to convey vector-mediated human ApoE2 to CNS [108], this finding is promising in terms of translational research.
In summary, the use of AVVs in gene therapy has accomplished promising results in CNS diseases and neurodegenerative diseases (for a review, see [101]). AAV provides the advantage of testing gene functions and variants directly in established mouse models of AD, but also in wild types, without the need to first breed different genetic lines [100]. This facilitates the study of genetics in non-transgenic animal models and contributes to the understanding of sAD pathology. However, transgenic mouse models remain the major choice to test the relevance of AAVs in gene therapy [109–111]. Although useful for evaluating and establishing late molecular mechanism in sAD, they are not typically used for testing compounds.
THERAPEUTICS AND CONTRIBUTIONS TOWARD UNDERSTANDING AD
Non-transgenic rodent models as a potential option for studying sAD
Current treatments for AD are exceptionally limited, ameliorate only some dementia symptoms, and offer marginal improvements in cognition [112]. The need for new interventions aimed at symptom reduction and halted disease progression is paramount. Better interventions require a critical assessment of the process of drug development and discovery, especially in the pre-clinical phase where the bulk of research is clustered. Within this context, the use of non-transgenic animals in therapeutics represents a small fraction of the published literature relative to transgenic models [113]. Importantly, while non-transgenic models can rely on amyloid-based interventions, many are non-amyloid models which lead to not only the expression of amyloid markers but other pathophysiological indicators. In this line, Grillo et al. generated a model of hippocampal-specific IR knockdown using a lentiviral vector (LV-IRAS). LV-IRAS-treated rats exhibited decreased neuroplasticity, failed to evoke long term potentiation, reduced GluA1 receptor levels and displayed spatial learning impairment. Indeed, this non-amyloid approach suggests that insulin signaling, and IR play an important role in memory and learning independently of Aβ formation. This findings support the observation that some aspects of sAD can be related to Aβ; nevertheless, an amyloid centric hypothesis cannot explain the complex genetic and environmental interaction involved in sAD [114]. Thus, some efforts to address the complex disease etiology by mixed transgenic and non-transgenic models have been reported and certainly represent an interesting direction for research [115]. However, non-transgenic models remain under-utilized and can provide unique and relevant biological information as well as models for testing of novel therapeutics.
Biases in non-transgenic models
Studies analyzing the use of transgenic models of AD have found that there are important biases and inconsistent reporting of methodology [9, 113]. In reviewing publications of therapeutics in non-transgenic AD models, we observe similar gaps in the pathologies that are studied and reported for each potential therapeutic manipulation resulting in few dependent measures (findings) being reported for each compound or compound class. Evaluation of oxidative stress is most often reported in the STZ-icv, D-galactose, and aluminum models (Tables 2, 4, and 5), while amyloid markers are most often reported in HFD (Table 3), and neuroinflammation in the Aβ infusion model (Table 1). Memory assays are the most common assays used in transgenic mouse models of AD [9]. By reviewing the assays reported in non-transgenic amyloid and non-amyloid models, we found a similar pattern. The most reported assays are the Morris water maze (MWM), the novel object recognition task, and passive avoidance (see Tables 1– 5). These are assays that evaluate learning and memory (for a review, see [9]). Most studies do not analyze behavioral psychological symptoms of dementia, such as depression, anxiety and sleep disturbances, important components of clinical symptoms [8, 9]. These behaviors can be incorporated into a behavioral battery for animals undergoing non-transgenic experimental manipulations and would greatly add to our understanding of disease and to the treatment of multiple psychological symptoms.
Central to neurodegeneration and AD is cell dysfunction and loss [14]. Therefore, another component that we deem important in future studies is to ensure the evaluation of neuronal and synaptic dysfunction and/or loss as these may occur as a consequence to many pathologies including amyloid, tau, inflammation and ROS production and are central to neurodegenerative disease. In our current assessment of the literature we observe that few of the treatments evaluated across the models report apoptosis as an outcome variable or synaptic/cholinergic activity (Tables 1– 5). These may be useful pathologies to establish a model and extend its construct validity to sAD.
Non-pharmacological interventions
There are few instances where a specific treatment has been evaluated across multiple non-transgenic AD models. In evaluating non-pharmacological interventions like treadmill exercise, for example, is reported for the STZ-ICV and HFD models (see Tables 2 and 3). In both cases, authors report a decrease in amyloid markers and concomitant improvement in memory assays. A similar therapeutic strategy, swimming exercise, was reported to decrease tau pathology and neuroinflammation while also improving memory assays in the STZ-ICV model (Table 2); while reducing oxidative stress and improving memory in the MWM in the aluminum induced AD neurotoxicity model (Table 5). The outcome of exercise-based approaches remains to be elucidated for other non-transgenic models but has been evaluated extensively in human interventions with mixed results [116]. Identifying common molecular mechanisms, or minimum requirements for exercise-based interventions may benefit from the use of multiple non-transgenic models. Incomplete assessment of pathological findings clouds the ability to infer molecular mechanism/s of action that may underlie what consistently is reported as enhanced learning and memory in a variety of neurodegenerative models.
Pharmacological interventions
Similarly, when evaluating pharmacological interventions few compounds have been studied across different models. Caffeic acid reportedly studied in STZ-ICV and HFD models as an antioxidant and anti-inflammatory. Decreased amyloid, tau (HFD), neuroinflammation (STZ) and oxidative stress (both) markers are reported. Only HFD reports increased memory in MWM with caffeic acid. In this example, there are again incomplete assessments of the pathological hallmarks of sAD which lead to important future directions; does caffeic acid affect amyloid and tau pathology in the STZ model? Or neuroinflammation in the HFD model? Is there common mechanistic pathway for these sparse effects that may warrant further study in drug development? Luteolin, a flavonoid, has been studied in the Aβ infusion, STZ-icv, and HFD models. In all models, an increase in memory in MWM is reported. Effects on pathology are varied across the models or not reported (see Tables 1– 3). Like luteolin, many flavonoids have been studied: anthocyanins have been evaluated in the Aβ infusion, STZ-ICV, and D-galactose models. With anthocyanins, we did not find overlap in findings reported (see Tables 1, 2, and 4), although all reports are generally favorable. Other flavonoids studied in non-transgenic animal models include curcumin (Table 2), hesperidin (Table 4) and its aglycone, hesperitine (Table 1), and isoflavones (Table 5). All models reviewed have been used to evaluate the effect of flavonoids. This compound class may prove beneficial and could warrant further study to identify common effects and the molecular mechanisms by which these effects are observed. In general, these examples highlight the need for standardization in reporting of AD-associated pathology, as well as thorough pre-clinical evaluation in AD models aimed at understanding the benefits and mechanisms of therapeutic action.
Clinical trials
In relation to ongoing clinical trials, anti-amyloid therapies have dropped substantially, from ∼55% of trials in 2017 to ∼30% of trials in 2019 in phase III, while other therapies have seen a modest increment, with most dramatic increments observed for neuroprotective (∼25% increase from 2017) and anti-neuroinflammatory (∼2 fold increase since 2017) therapies in phase I [8]. Many active trials evaluate neurotransmitter-based approaches designed for behavioral psychological symptoms of dementia; an observation that highlights the dearth of research in non-transgenic animal models on other behavioral components, such as anxiety, depression, and sleep disturbances, that are also debilitating to individuals with sAD. Moreover, current clinical trials evaluate therapeutics aimed at reduction of neuroinflammation and oxidative stress as well as cognitive enhancers and neuroprotective effects, while few evaluate metabolic or insulin related approaches. Nonetheless, these potential therapeutics are in line with the classes of therapeutics evaluated in non-transgenic models and include interventions evaluated in these models such as aerobic exercise training (Tables 2, 3, and 5), vitamin D3 (Table 2), coconut oil (Table 3), and insulin (Table 4) which have had promising results. The direction in which therapeutics for sAD are moving is encouraging, and the studies reviewed suggest that rigorous preclinical evaluation in non-transgenic, as well as transgenic, models may aid in the evaluation of treatments prior to clinical trials.
Recent novel clinical trials like sodium oligomannate (GV-971), a mixture of acidic oligosaccharides derived from brown algae have suggested a possible link between gut microbiota and AD pathogenesis [117]. Lipopolysaccharides and amyloids produced by bacteria in the gut microbiome may play a role in AD pathogenesis, since gastrointestinal tract and BBB become more permeable during aging, allowing these molecules to indirectly trespass BBB. It is suggested that this can lead to prompt proinflammatory molecules, as consequence, microglia activation [118, 119]. In this regard, Wang et al. found that dysbiosis of gut microbiota evidenced by AD progression in 5xFAD mice, leaded to peripheral increase of phenylalanine and isoleucine which stimulated the differentiation and proliferation of pro-inflammatory T helper (Th1) 1 cells, thus, enhancing microglial activation [117]. Additionally, this study suggests a novel therapeutic strategy for AD, sodium oligomannate (GV-971), which is currently in phase III clinical trial in China. GV-971 suppressed gut dysbiosis and diminished neuroinflammation produced by microglia activation and reversed cognitive impairment [117]. Nevertheless, more studies will be necessary to understand GV-971 mechanism of action since gut-brain axis relationship is not completely elucidated. Finally, to the best of our knowledge, no studies have reported looking at GV-971 in non-transgenic rodent models.
LIMITATIONS
Our review is not without limitations. We excluded mixed models of transgenic and non-transgenic approaches, as well as limited our analysis in the STZ model to more recent literature but refer readers to a prior review [51]. Although we have considered treatment class, specific treatments and reported findings we have not meticulously analyzed sex of animals studied. As is the case in most neuroscience research, we suspect a strong male bias in these studies [120]. We also limited our review to experimental non-transgenic models as the literature on aged animals or species that develop similar pathology is currently very limited [12]. Lastly, further assessment of non-transgenic animal models should focus on evaluating study quality to identify risk bias and appraise efficacy of treatments [113].
CONCLUSION
We conclude that non-transgenic experimental AD models are a useful, complementary preclinical approach to the use of transgenic animal models. We highlight the strengths of these approaches in that they model overlapping but distinct mechanisms for neurodegeneration and AD. Moreover, the need for standardization in reporting therapeutic strategies, especially if they have exhibited promising results by not only using one animal model but rather testing the same therapeutic strategies in other models is crucial. The use of non-transgenic animal models allows for the modeling of risk factors to identify molecular mechanisms contributing to sAD, an area of research that is currently under-evaluated but that might prove beneficial in the development of early identification of people at risk and novel therapeutic strategies. Although current efforts in pre-clinical experimental research evaluating therapeutic strategies in non-transgenic animal models of sAD are few and incomplete, we propose that these models may benefit drug development in providing testing of treatment for the multiple pathologies associated to sAD more so than transgenic animal models, because they represent an inexpensive, accessible, and rapid research platform and because they are a tool to shift our focus on drug development from amyloid to novel approaches of pathology in AD.
