Introduction
Long-standing evidence shows that progressive cerebral deposition of Aβ plays a seminal role in the pathogenesis of Alzheimer’s disease (AD). There has been great interest, therefore, in understanding the proteolytic processing of APP and the enzymes responsible for cleaving at the N- and C-termini of the Aβ region. Besides Aβ peptides starting with an aspartate at position 1, a variety of different N-truncated Aβ peptides has been identified in AD brains. Ragged peptides with a major species beginning with phenylalanine at position 4 of Aβ have been reported as early as 1985 by Masters et al. [
36]. This finding has been disputed, as no N-terminal sequence could be obtained from cores purified in a sodium dodecyl sulfate-containing buffer, suggesting that the N-terminus is blocked [
16,
59]. In 1992, Mori et al. first described the presence of Aβ
3(pE) using mass spectrometry of purified Aβ protein from AD brains, explaining the difficulties in sequencing the amino-terminus [
41]. They reported that only 10–15% of the total Aβ isolated by this method begins at position 3 with Aβ
3(pE). Saido et al. [
52] showed for the first time that Aβ
3(pE) represents an important fraction of Aβ peptides in AD brain, which was later verified by other reports using AD and Down’s syndrome post-mortem brain tissue [
18,
19,
23,
25,
28,
29,
38,
45,
46,
49,
53,
63]. N-terminal deletions in general enhance aggregation of β-amyloid peptides in vitro [
47]. Aβ
3(pE) has a higher aggregation propensity [
20,
55], and stability [
30], and shows an increased toxicity compared with full-length Aβ [
51].
Mouse models mimicking AD-typical pathology, such as deficits in synaptic transmission [
24], changes in behavior, differential glutamate responses, and deficits in long-term potentiation are typically based on the overexpression of full-length amyloid precursor protein (APP) [
39]. Learning deficits [
2,
21,
43,
48] were evident in different APP models, however, Aβ-amyloid deposition did not correlate with the behavioral phenotype [
22]. In the past, Aβ has been regarded as acting extracellularly, whereas recent evidence points to toxic effects of Aβ in intracellular compartments [
64,
68]. In addition, another concept favors that toxic forms of Aβ are soluble oligomers and β-sheet containing amyloid fibrils [
26,
58]. It has been demonstrated that soluble oligomeric Aβ42, but not plaque-associated Aβ correlates best with cognitive dysfunction in AD [
37,
42]. Oligomeres are formed, preferentially, intracellularly within neuronal processes and synapses rather than extracellularly [
60,
66]. Previously, we have reported that intraneuronal Aβ rather than extracellular plaque pathology correlates with neuron loss in the hippocampus [
8], the frontal cortex [
11], and the cholinergic system [
10] of APP/PS1KI mice expressing transgenic human mutant APP751 including the Swedish and London mutations on a murine knock-in (KI) Presenilin 1 (PS1) background with two FAD-linked mutations (PS1
M233T and PS1
L235P). The APP/PS1KI mice exhibit robust and learning deficits at the age of 6 months [
67], age-dependent axonopathy [
71], neuron loss in hippocampus CA1 together with synaptic deficits, and hippocampus atrophy coinciding with intraneuronal aggregation of N-terminal-modified Aβ variants [
4].
The APP/PS1KI mouse model exhibits a large heterogeneity of N-truncated Aβ
x–42 variants [
8]. It is impossible to decipher the true toxic peptides from those, which might only co-precipitate. Therefore, we generated a new mouse model expressing only N-truncated Aβ
3(pE) in neurons, and demonstrate for the first time that this peptide is neurotoxic in vivo inducing neuron loss and an associated neurological phenotype.
Discussion
Mice transgenic for the human APP gene have been proven valuable model systems for AD research. Early pathological changes, including deficits in synaptic transmission [
24], changes in behavior, differential glutamate responses, and deficits in long-term potentiation [
39] have been reported in several studies. In addition, learning deficits [
2,
15,
21,
43,
48] and reduced brain volume [
4] were evident in transgenic APP models. Interestingly, extracellular amyloid deposition did not correlate with the behavioral phenotype [
22,
67]. These deficits occurred well before plaque deposition became prominent and may, therefore, reflect early pathological changes, likely induced by intraneuronal APP/Aβ mistrafficking or intraneuronal Aβ accumulation (reviewed in [
1]). The coincidence of intracellular Aβ with behavioral deficits supporting an early role of intracellular Aβ has been recently demonstrated in a mouse model containing the Swedish and Arctic mutations [
27,
34]. In accordance with these findings, we have previously shown that intraneuronal Aβ accumulation precedes plaque formation in transgenic mice expressing mutant APP695 with the Swedish, Dutch, and London mutations in combination with mutant PS-1 M146L. These mice displayed abundant intraneuronal Aβ immunoreactivity in hippocampal and cortical pyramidal neurons [
69]. An even more pronounced phenotype was observed in another transgenic mouse model, expressing Swedish and London mutant APP751 together with mutant PS-1 M146L [
3]. In young mice, a strong intraneuronal Aβ staining was detected in vesicular structures in somatodendritic and axonal compartments of pyramidal neurons and an attenuated neuronal immunoreactivity with increasing age. The intraneuronal immunoreactivity declined with increased plaque accumulation [
70], a finding which was also reported in Down’s syndrome patients, where the youngest patients displayed the strongest immunoreactivity [
40]. The neuronal loss in CA1 of the hippocampus did not correlate with the amount of extracellular Aβ [
4,
8]. The same observation has been reported in the APP/PS1M146L model [
57]. Hippocampal neuron loss has also been reported in the APP23 mouse model [
7], however whether intraneuronal Aβ contributes to the neuron loss in this model is not clear. The triple-transgenic mouse model expresses mutant APP in combination with mutant PS-1 and mutant Tau protein. These mice displayed early synaptic dysfunction before plaque or tangle deposition was evident, together with early intraneuronal Aβ immunoreactivity preceding plaque deposition. Tau and Aβ immunoreactivity colocalized in hippocampal neurons, which might imply that early intraneuronal Aβ accumulation could affect Tau pathology [
44]. All these AD mouse models express full-length APP, and also C-terminal fragments and Aβ peptides after cleavage. It is, therefore, difficult to decipher the pathological function of specific Aβ peptides.
In order to specifically investigate the neurotoxicity of Aβ
3pE–42 generation in vivo, we have generated transgenic mice expressing Aβ
3Q–42 starting at position 3 with glutamine and ending with position 42 (TBA2 mouse line). Owing to the replacement of N-terminal glutamate by glutamine, the Aβ peptides are more prone to conversion into pyroglutamate [
54]. The severity of the neurological phenotype observed in TBA2 mice, accompanied with the Purkinje cell loss and premature mortality reflects the in vivo toxicity of Aβ
3(pE)–42. However, we cannot rule out that unprocessed Aβ
3Q–42 has been stabilized by Aβ
3(pE)–42 accumulation, and might also contribute to the observed neurological phenotype. In addition, the level of Aβ
3(pE)–42 in 6-month-old APP/PS1KI mice was comparable to that of 2-month-old TBA2 mice. However, 85% of Aβ peptides in the APP/PS1KI mice terminated at position 42, the N-terminus shows a large heterogeneity including Aβ
3(pE). The time point of high levels of Aβ
3(pE)–42 coincided with the onset of behavioral deficits in both mouse models. In addition, only a fraction of Purkinje cells showed abundant levels of intracellular Aβ
3(pE) accumulation leading to the assumption that only those are prone for degeneration (Fig.
4). We have previously shown that at 6 months, the APP/PS1KI mice exhibit a neuron loss in CA1 of the hippocampus [
4,
8], the frontal cortex [
11], and in distinct cholinergic nuclei [
10]. Recently, a transgenic mouse model expressing human APP with the 714 austrian mutation has been reported showing intraneuronal Aβ accumulation correlating with brain atrophy [
65]. Overall, the pathological events seen in the APP/PS1KI mouse model might be at least partly triggered by Aβ
3(pE)–42 accumulation however the TBA2 mouse model is the only one expressing Aβ
3(pE)–42 without any of the other Aβ peptides.
Amyloid precursor protein transgenic mouse models have been reported to show no [
29] or low Aβ
3(pE) levels [
18]. Maeda et al. have demonstrated that the localization and abundance of [11C]PIB autoradiographic signals were closely associated with those of amino-terminally truncated and modified Aβ
3pE deposition in AD and different APP transgenic mouse brains, implying that the detectability of amyloid by [11C]PIB-PET is dependent on the accumulation of specific Aβ subtypes [
35]. There is an interesting coincidence of considerable amounts of Aβ
3(pE) and massive neuron loss in the APP/PS1KI mouse model [
1,
8]. An emerging role of intracellular Aβ accumulation has been previously shown in human AD [
13,
17]. It has been observed that Aβ localizes predominantly to abnormal endosomes [
9], multivesicular bodies, and within pre- and postsynaptic compartments [
31,
61]. Takahashi et al. demonstrated that Aβ42 aggregates into oligomers within endosomal vesicles and along microtubules of neuronal processes, both in Tg2576 neurons with time in culture, as well as in Tg2576 and human AD brain [
60]. In good agreement with these reports, we observed that Aβ is also localized in the late endosomal/lysosomal compartment in the TBA2 model. Owing to the large heterogeneity of N-truncated Aβ
x–42 peptides in the APP/PS1KI model, it is impossible to study the role of a single Aβ variant. In the TBA2 model, however, we were able to demonstrate the intraneuronal Aβ
3(pE)–42 aggregation induced neuron loss without contribution of extracellular Aβ aggregation.
N-truncated Aβ
3(pE) peptides have been identified by several groups from AD brains [
18,
19,
23,
25,
28,
29,
38,
41,
45,
46,
49,
52,
53,
63]. In addition, other N-terminal truncated peptides have been identified such as Aβ
5–40/42 [
62], Aβ
11–40/42 [
32,
33], and Flemish and Dutch N-terminally truncated amyloid beta peptides [
14]. Cai et al. have demonstrated that secretion of Aβ
1–40/42 and Aβ
11–40/42 is abolished in BACE1−/− neurons establishing that BACE1 is the principal β-secretase for endogenous APP in neurons. Although Aβ
11–40/42 peptides have been observed in neuronal cultures and in the brains of patients with AD, the involvement of these peptides in its pathogenesis remains to be elusive [
6]. In general, N-terminal deletions enhance aggregation of β-amyloid peptides in vitro [
47]. Aβ
3(pE) has a higher aggregation propensity [
20,
55] and stability [
30], and shows an increased toxicity compared with full-length Aβ [
51]. It has been suggested that N-truncated Aβ peptides are formed directly by BACE and not through a progressive proteolysis of full-length Aβ
1–40/42 [
50].
In in vitro experiments Schilling et al. have shown that the cyclization of glutamate at position 3 of Aβ is driven enzymatically by glutaminyl cyclase (QC) [
54]. In addition, it has been shown that QC inhibition significantly reduced Aβ
3(pE) formation, emphasizing the importance of QC activity during the cellular maturation of pyroglutamate-containing peptides. The pharmacological inhibition of QC activity by the QC inhibitor P150, which significantly reduced the level of Aβ
3(pE) in vitro [
12] and in vivo [
56] suggests that QC inhibition might serve as a new therapeutic approach to rescue Aβ
3(pE) triggered neurodegeneration in human disorders.
Aβ accumulation has an important function in the etiology of AD with its typical clinical symptoms, such as memory impairment and changes in personality. However, the mode of this toxic activity is still a matter of scientific debate. Previously, we have shown that the APP/PS1KI mouse model develops severe learning deficits at 6 months of age correlating with a CA1 neuron loss and an atrophy of the hippocampus [
4,
8], together with a drastic reduction of long-term potentiation and disrupted paired pulse facilitation. This was accompanied by reduced levels of pre- and post-synaptic markers. We also observed that intraneuronal and not plaque-associated Aβ including N-modified Aβ
3(pE)–42 species increased and coincided well with CA1 neuron loss, however the dominant species was Aβ
1–42 in the APP/PS1KI model. In good agreement with this study, we could show for the first time that intraneuronal Aβ
3(pE)–42 accumulation is sufficient for triggering neuron death and inducing an associated neurological phenotype in a novel transgenic mouse model (TBA2 mice). The Aβ staining in the cerebellum was completely restricted to the intraneuronal compartment further supporting the notion that intraneuronal pathology is instrumental in neuron loss and that extracellular plaque deposition has no drastic effect on cell survival.