Introducing Human APOE into Aβ Transgenic Mouse Models

Apolipoprotein E (apoE) and apoE/amyloid-β (Aβ) transgenic (Tg) mouse models are critical to understanding apoE-isoform effects on Alzheimer's disease risk. Compared to wild type, apoE−/− mice exhibit neuronal deficits, similar to apoE4-Tg compared to apoE3-Tg mice, providing a model for Aβ-independent apoE effects on neurodegeneration. To determine the effects of apoE on Aβ-induced neuropathology, apoE−/− mice were crossed with Aβ-Tg mice, resulting in a significant delay in plaque deposition. Surprisingly, crossing human-apoE-Tg mice with apoE−/−/Aβ-Tg mice further delayed plaque deposition, which eventually developed in apoE4/Aβ-Tg mice prior to apoE3/Aβ-Tg. One approach to address hAPOE-induced temporal delay in Aβ pathology is an additional insult, like head injury. Another is crossing human-apoE-Tg mice with Aβ-Tg mice that have rapid-onset Aβ pathology. For example, because 5xFAD mice develop plaques by 2 months, the prediction is that human-apoE/5xFAD-Tg mice develop plaques around 6 months and 12 months before other human-apoE/Aβ-Tg mice. Thus, tractable models for human-apoE/Aβ-Tg mice continue to evolve.


Introduction
Alzheimer's disease (AD) is the most common form of dementia and represents a serious economic and social burden worldwide. The familial form of AD (FAD) is caused by autosomal dominant mutations that increase levels of the 42 amino acid isoform of the amyloid-beta 42 (Aβ42) peptide [1,2]. The primary genetic risk factor for AD is inheritance of the APOE4 gene for apolipoprotein E (apoE), compared to APOE3, with APOE2 reducing risk [3][4][5]. The mechanism(s) by which apoE and Aβ affect pathogenesis of the disease is unclear (reviewed [3][4][5]). However, evidence suggests that apoE may isoform-specifically modulate Aβ-induced neurotoxicity [4,5]. To address potential mechanisms in vivo, several transgenic (Tg) mouse models have been developed to assess the structural and functional interactions between apoE and Aβ. However, each of these models has potential drawbacks that affect the interpretation and physiological relevance of the results. This paper will summarize the current models resulting from the introduction of human-APOE into Aβ-Tg mice. The continued development and characterization of both apoE and apoE/Aβ-Tg mouse models is critical to understanding the apoE-isoform effects on AD pathology.

Amyloid-β.
Traditional diagnosis of AD is based on pathology that includes extracellular amyloid plaques, composed primarily of Aβ42, intraneuronal neurofibrillary tangles, hyperphosphorylated tau, neuroinflammation, and neuronal cell loss. Aβ is a 39-43 amino acid (4 kDa) peptide produced via sequential proteolysis of amyloid precursor protein (APP) by β-secretase/BACE followed by γ-secretase (composed of presenilins [PS] 1 and 2), to produce Aβ peptides primarily 40 and 42 amino acids in length. Genetic and experimental evidence indicates that Aβ42 is the cause of AD pathogenesis [1,2]: (1) FAD, although only 3-5% of all AD cases is caused by autosomal dominant mutations in APP, PS1, or PS2 that increase levels of Aβ42 or the Aβ42 : 40 ratio; (2) Down syndrome is caused by trisomy of chromosome 21 (the location of the APP gene), and is characterized by plaque deposition and dementia by the age 40; (3) Aβ42 is neurotoxic in vitro and in vivo. Aβ, particularly the more toxic Aβ42, aggregates to form a variety of higher-order assemblies including oligomers, protofibrils, fibrils, and amyloid plaques [6]. Amyloid plaques themselves exist in different conformations including compact plaques (composed of a dense Thioflavin-S-(ThioS-) positive core), neuritic plaques (identified as ThioS-positive plaques surrounded by a ring of dystrophic neurites), and diffuse plaques (characterized by amorphous wisps of amyloid that lack a central core and are not neurotoxic) [7][8][9][10]. While total plaque burden does not directly correlate with dementia [11], it is an indication of increased Aβ42 levels, and compact or dense core plaques may be disease relevant [7,12].

ApoE.
ApoE is the only apolipoprotein synthesized within the blood-brain barrier (BBB) and is the primary apolipoprotein associated with lipoprotein particles in the central nervous system (CNS), as peripheral apoE is unable to cross the BBB or blood-cerebrospinal fluid barrier. While the majority of apoE in the CNS is secreted by glial cells, particularly astrocytes, neuronal production of apoE has been observed under specific pathological conditions [13]. CNS lipoproteins are critical for lipid homeostasis, particularly as cholesterol and phospholipids are required for neuronal growth, repair, and synaptogenesis (reviewed [3][4][5]). In humans, three apoE polymorphic alleles exist (APOE2, APOE3, and APOE4) which encode three protein isoforms (apoE2, apoE3, and apoE4). Although human-apoE (h-apoE) is a 299 amino acid protein, the three apoE-isoforms differ by a single amino acid substitution at residues 112 or 158: apoE2 contains Cys 112,158 , apoE3 contains Cys 112 Arg 118 , and apoE4 contains Arg 112,158 [5]. In the general population, APOE3 is the most common allele (77%), followed by APOE4 (15%) and APOE2 (8%). In contrast, 40% of AD patients express at least one APOE4 allele. Compared to APOE3, inheritance of one or two copies of the APOE4 allele increases the risk for developing AD by 4-and 12-fold, whereas APOE2 decreases risk by 2-and 4-fold [5,14]. Amyloid plaque deposition is greatest in AD patients with an APOE4 allele(s) [14]. In addition to AD, APOE4 is a risk factor for cerebral amyloid angiopathy (CAA; amyloid deposition in blood vessels) and impairs recovery from cerebral insults such as stroke, cerebral hemorrhage, and brain injury [4]. While apoE knock-out mice (apoE −/− ) and h-apoE-Tg models demonstrate that apoE affects neuronal viability independent of Aβ-induced neurotoxicity, the focus of this paper is on the synergism between the h-apoE isoforms and Aβ on neuropathology. To address the latter in vivo, apoE/Aβ-Tg models were subsequently developed to specifically address the isoform-specific effects of h-apoE on Aβ deposition.

Transgenic Mice Expressing
Human ApoE (Table 1) Several approaches have been used to make Tg mouse models to assess the function of apoE. ApoE −/− Tg-mice were initially used to help understand the role of apoE in the brain [15,16] although the homology between mouse (m-) and h-apoE is 70%, and mice express only a single isoform, comparable to apoE4 at residues 112 and 158 [54]. ApoE −/− mice have been used as the background for several h-APOE-Tg mouse lines. Heterologous promoters have been used to drive the expression of h-apoE in glia or neurons. Examples include glial fibrillar acidic protein (GFAP; glial) [22,[55][56][57], transferrin (neuronal) [58], platelet-derived growth factor (PDGF; neuronal) [59], neuron-specific enolase (NSE; neuronal) [15], and thymocyte differentiation antigen 1 (Thy1; neuronal) [59]. However, these models have limitations inherent in the use of a heterologous promoter and specific to apoE: (1) the expression of protein by a heterologous promoter is not regulated as it would be by the endogenous promoter; (2) the inserted copy number of the transgene cannot be regulated by a traditional Tg mouse approach; (3) while the cell-specific expression of apoE in the brain is controversial [60][61][62][63], the majority of evidence suggests that glia, not neurons, are the primary cell type to express apoE [13,[64][65][66][67]; (4) by using the m-apoE −/− background and inducing apoE expression via CNS-specific promoters, lack of peripheral apoE becomes a variable of potential importance when interpreting results from these mice; (5) evidence suggests that in both humans and apoE-TR mice, apoE4 levels are significantly lower than apoE2 or apoE3 [68][69][70][71][72][73]. Knockin or targeted-replacement (TR) mice were developed that express h-apoE under the control of the endogenous mouse promoter and provide an alternative to heterologous expression of h-apoE. In apoE-TR mice, the coding domain for each of the h-APOE isoforms replaces the coding domain for m-APOE. Thus, in apoE-TR mice, glial cells express h-apoE in a native conformation at physiologically regulated levels, and in the same temporal and spatial pattern as endogenous m-apoE. Thus, the interpretation of apoE isoform-specific results is determined by the nature of the apoE-Tg mouse model. As discussed, a number of apoE-Tg mouse models have been developed with apoE expression under the control of different promoters [74]. The general phenotypes of apoE −/− mice and three examples of the most widely studied apoE-Tg mice (GFAP-apoE, NSE-apoE, and apoE-TR) are briefly discussed (Table 1).

ApoE −/−
. Compared to wild type, apoE −/− mice have decreased excitatory transmission, spine density, and dendritic length [16][17][18]. These changes may underlie behavioral deficits, as apoE −/− mice are cognitively impaired [15,19,20]. However, lack of peripheral apoE can have profound effects on plasma lipid homeostasis, potentially leading to a number of confounding variables, including metabolic disease and increased risk for cardiovascular disease, compromising the ability to compensate for oxidative stress or inflammation, deficits that can effect the vasculature of the brain. In addition, the relevance of these mice is unclear as there are no apoE −/− humans. ApoE −/− provides the background for a number of the h-apoE-Tg mice.  show increased CA1 cellular atrophy and decreased spine density compared to GFAP-apoE3 mice [75]. In addition, compared to GFAP-apoE3 mice, GFAP-apoE4 mice demonstrate impaired cognition and increased anxiety; common symptoms in AD patients [17,23,24]. Interestingly, cognitive impairment is evident at an earlier age in female compared to male GFAP-apoE4 mice [23,24]. A limitation of this model is that the effect of apoE isoform on neuroinflammatory responses cannot be interpreted, as apoE is expressed under the control of a promoter that is induced by neuroinflammation.

Neuronal apoE Expression.
Neuronal apoE expression remains controversial, although expression has been identified under conditions of stress [25]. To investigate the role of neuronal expression, apoE-Tg mice expressing h-apoE under the control of the NSE promoter were crossed with apoE −/− mice [25][26][27]. Kainic acid was used to induce apoE expression in these mice and resulted in the protection of NSE-apoE3 mice from age-and kainic acid-induced presynaptic and dendritic degeneration compared to NSE-apoE4 mice [25]. In addition, female NSE-apoE4 mice demonstrate impaired cognition compared to female NSE-apoE3 mice [25]. In contrast, male NSE-apoE4 mice do not exhibit cognitive deficits [15]. Thus, the effect of apoE4 on cognition may be modulated by gender. This is similar to the early development of cognitive deficits in GFAP-apoE4.

GFAP-apoE/Aβ-Tg.
The initial studies demonstrating that the lack of m-APOE delayed plaque deposition in Aβ-Tg mice led to the question of what would be the effect of introducing h-APOE into apoE −/− /Aβ-Tg mice ( Table 2). To initially address this issue, GFAP-apoE mice was crossed with apoE −/− /PDAPP mice [41,42]. Surprisingly, the introduction of h-APOE did not result in the expected reduction in the age of onset of Aβ pathology, rather the presence of h-APOE further delayed Aβ deposition. Amyloid accumulation is delayed from 6 to 12 months in apoE +/− /PDAPP mice, and to 15 months in GFAP-apoE +/− /PDAPP mice (Table 2). Once plaque pathology returned, the greatest plaque burden was found in GFAP-apoE4/PDAPP mice, compared with GFAP-apoE2/PDAPP and GFAP-apoE3/PDAPP mice. One potential confounding factor in these mice, as well as the NSE-apoE mice described below, is the equal expression of the h-APOE isoforms, in contrast to human data where inheritance of an APOE4 allele results in lower apoE levels [68,69,73].

ApoE-TR/Aβ-Tg.
The physiologic advantages of using apoE-TR mice to study the function(s) of apoE in vivo led to the generation of apoE-TR/PDAPP [30] and apoE-TR/ Tg2576 mice [47] ( Table 2). The resulting data confirmed that h-APOE delayed plaque deposition. In apoE-TR/PDAPP mice, plaque deposition was delayed from 6 months to 18 months of age. An apoE isoform-specific effect on Aβ pathology was also observed, with ThioS staining, Aβ immunoreactivity, and Aβ levels in brain homogenates highest with apoE4 [30]. In Tg2576 mice, plaque deposition initiates at 9 months of age, while in apoE-TR/Tg2576 mice there is minimal plaque staining at 15 months. Interestingly, at 15 months, the isoform effect in these mice is primarily on CAA (E4 > E3), as amyloid deposition in the parenchyma is minimal [47].

Addressing h-APOE-Induced Delay in Aβ Pathology.
The major drawback to the apoE/Aβ-Tg crosses described thus far is the significant h-APOE-induced delay in Aβ pathology. For example, in apoE-TR/PDAPP mice, plaque deposition is not identified until mice are ≥18 months of age (Table 2). This substantial delay precludes timely analyses of apoE iso-form-specific effects on early aspects of Aβ pathology. One approach to address this temporal delay is to introduce an additional insult, such as traumatic brain injury (TBI) [82], kainic acid [25], nitric-oxide-synthase-2-(NOS2-) knock-out [83], or by blocking Aβ degradation via neprilysin in-hibition [84]. Although no amyloid deposition is present in 12-month-old GFAP-apoE/PDAPP mice, TBI at 9 months leads to amyloid deposition at 12 months [82], which is greater with apoE4 compared to apoE3. Kainic acid decreases synaptophysin and MAP-2 staining in apoE −/− and NSE-apoE Tg mice, with the effect more pronounced with apoE4 than apoE3. Nitric oxide, produced by inducible NOS (encoded for by the NOS2 gene), is an important signaling and redox factor that plays a key role in neuroinflammation and neurodegeneration [83]. NOS2 −/− mice have been crossed with multiple Aβ-Tg mouse models, and results demonstrate increased tau phosphorylation and neuronal loss in NOS2 −/− /Aβ-Tg mice [83]. In the Tg2576/NOS2 −/− mice, Aβ deposition is higher compared to Tg2576 mice. Inhibition of neprilysin, an enzyme that degrades extracellular Aβ, with thiorphan, induces fibrillization and deposition of Aβ and in wild-type mice. Thiorphan treatment of apoE-TR mice leads to aggregation of mouse Aβ 1 week after treatment, with higher Aβ deposition in apoE4-TR compared to apoE3-TR mice [84,85]. Thus, thiorphan treatment of h-apoE-Aβ-Tg mice represents a potential method of accelerating human Aβ deposition.

Concluding Remarks
ApoE is the greatest risk factor for AD. ApoE −/− mice exhibit neuronal and cognitive deficits. Human apoE-Tg mouse models demonstrate that, compared to apoE3, apoE4 increases markers of neurodegeneration and cognitive impairment. Initially, to determine the effect of apoE on Aβ pathology, Aβ-Tg mice were crossed with apoE −/− mice, resulting in a significant delay in plaque deposition. Surprisingly, the introduction of h-APOE to several apoE −/− /Aβ-Tg mouse models further delayed plaque deposition. This temporal delay restricts the usefulness of the current apoE/Aβ-Tg mice for investigating the process of Aβ accumulation and the resulting neurotoxicity. To accelerate Aβ deposition, current apoE/Aβ-Tg models could be treated with an additional insult such as TBI, crossed with other Tg models of neurodegeneration (NOS2 −/− ), or treated with drugs that decrease Aβ degradation. Alternatively, the development of Aβ pathology could be accelerated by crossing Aβ-Tg models with a rapid onset of Aβ pathology, such as 5xFAD mice with apoE-TR mice. Transgenic models such as these provide tractable models for identifying biomarkers and the efficient initial validation of therapeutic targets.