Several hypotheses have been proposed and competed in trying to explain the underlying cause of Alzheimer’s disease (AD). The dominant hypothesis, since 1991, is the amyloid hypothesis that implicates amyloid-β (Aβ) deposits as the cause of this common neurodegenerative disorder. Extracellular deposits of Aβ protein and the intracellular accumulation of phosphorylated tau protein are the basis of the neuropathological characterization of AD [1–3]. Contemporary AD research has been driven forward through the use of advanced molecular biology tools. One key discovery, the isolation and sequencing of the gene encoding the larger amyloid precursor protein (APP) [4], was made possible by the biochemical analysis of β-amyloid containing blood vessels (CAA, cerebral amyloid angiopathy) [5] and amyloid plaques consisting of Aβ [6].

The “Amyloid Hypothesis” was proposed, therefore, based on the previously mentioned discovery [3, 7]. Since then, however, amyloid plaque load in the brain and cognitive impairment in suffering patients [8] or even in transgenic mouse models for AD [9, 10] have not been found to be consistently correlated. This gave rise to considerable controversy in the field.

Memory loss in AD is the most prominent clinical manifestation of the disease. To that end, Haass and Selkoe [11] have recently evaluated the concept that soluble oligomers of Aβ, acting as diffusible assemblies, are capable of interfering with synaptic function and integrity. This has provided a gateway for understanding the basis of memory loss in AD. They debated that while insoluble plaque deposits might function as reservoirs of the pathological oligomers, the small soluble oligomers affect synaptic structure and plasticity. A modified amyloid hypothesis was brought forth, wherein it has been suggested that intraneuronal Aβ accumulation precedes the extracellular formation of Aβ plaques and other AD pathological events [12]. It is now well accepted that the pathologically inert amyloid fibrils, which are found in plaques, originate from a nearly irreversible reaction driven by monomeric Aβ peptide through toxic protofibrillar intermediates. For instance, the fact that amyloid plaques possibly are major sources of soluble toxic Aβ-aggregates that could readily be activated by exposure to biological lipids, has been credibly demonstrated by Martins et al. [13].

In addition to Aβ₁ starting with aspartate as the first amino acid, several N-truncated and modified Aβ species have been characterized [14–16]. In fact, various N- and C-terminal variants have been described in conjunction with in vitro and in vivo analysis of amyloid deposits in AD [14, 17, 18]. The toxicity of Aβ was further promoted due to enhanced aggregation and deposition brought on by the increase in C-terminal length of Aβ (from Aβ_x-40 to Aβ_x-42) and by N-terminal truncation [19–21]. Among Aβ species present in AD plaques, Lewis et al. [22] reported that Aβ_4-42 is a relatively abundant species in AD, aged controls and vascular dementia patients.

Mori and colleagues discovered that approximately 15-20% of Aβ peptides carried a pyroglutamate residue at their N-terminus [23]. This ignited a spark of interest in the temporal and spatial deposition of pyroglutamate Aβ, which has increased ever since. For instance, Saido et al. demonstrated by immunohistochemistry and biochemical assays Aβ_pE3-x is present in equivalent or larger amounts than full-length Aβ in senile plaques. The suggestion that Aβ_pE3-x precedes the deposition of unmodified Aβ (Aβ_1-x) was also proposed by the authors based on their analysis of brain tissue from Down syndrome cases [24]. Saido et al. furthermore suggested that, due to their limited degradation, Aβ_pE3 and other modified Aβ species accumulate unhindered [16]. The aggregation tendency and stability of the Aβ_pE3-x peptides is due to the formation of the lactam ring and loss of two negative charges and one positive charge [16]. The stability of the peptide is further increased by the formation of the moiety-terminal pyroglutamate that is resistant to degradation by peptidases. He and Barrow [19] reported that, as compared to full-length Aβ, Aβ_pE3-x peptides exhibited enhanced β-sheet formation and aggregation propensity in aqueous and hydrophobic media. They proposed that a reduction of the level of unfavorable charge repulsion between strands, brought on by the loss of the three charged groups, facilitates and stabilizes β-sheet formation. Using immunoprecipitation in combination with mass spectrometry, Portelius and colleagues [25] showed that Aβ_1-40, Aβ_1-42, pyroglutamate Aβ_pE3-42 and Aβ_4-42 can be detected in the hippocampus and cortex of AD patients. Interestingly, it has been demonstrated that N-terminal deletions enhance Aβ aggregation when comparing Aβ_4-42 with Aβ_1-42 [21].

The weak correlation between the severity of dementia and the density and localization of amyloid plaques in the brain of AD patients is one of the major flaws in the amyloid hypothesis. Even before the primary signs of plaque deposition, memory impairment and pathological changes already appear in many AD mouse models [26]. Soluble oligomers are low molecular weight non-fibrillar structures, which are stable in aqueous solution and remain soluble even after high speed centrifugation [26]. Occurring more often than their proliferation inside the extracellular space, Aβ oligomers develop preferentially within neuronal processes and synapses [27, 28]. Results from several labs led to the proposition of these oligomers as the missing link in the amyloid hypothesis. While Aβ plaques are poor correlates for the clinical symptomatology in AD and Down syndrome patients, soluble oligomers are suggested to be good predictors for synaptic loss [29], neurofibrillary tangles [30] and clinical phenotype [31, 32]. Tomiyama et al. generated APP transgenic mice expressing the E693Δ mutation, which causes neuronal cell death and cognitive impairment by enhanced intracellular Aβ oligomerization without plaque formation [33].

Although Aβ_4–42 is highly abundant in AD brains and was the first N-truncated peptide discovered [14] its possible role in AD pathology has been largely overlooked. We have recently shown that Aβ_4–42 rapidly forms aggregates and possesses a high aggregation propensity [34]. In vitro and in vivo exposure indicated that Aβ_4-42 is as toxic as Aβ_pE3-42 and Aβ_1–42. In addition, we have generated transgenic mice expressing Aβ_4-42 (Tg4-42 transgenic line) that developed a massive CA1 pyramidal neuron loss in the hippocampus [34]. Interestingly, as assessed using the Morris water maze test, the hippocampus-specific expression of Aβ_4–42 alone correlated with age-dependent spatial reference memory deficits [34]. In the present report, we developed a novel antibody specific for N-truncated Aβ and characterized it using a toxicity assay, 5XFAD transgenic mice, sporadic and familial AD cases.

The amyloid-β hypothesis has been the most influential hypothesis in coining the molecular pathology of AD [2]. According to the initial hypothesis, amyloid fibrils, which are large insoluble polymers of Aβ found in senile plaques, are the major trigger of neuron loss and dementia that are typical for AD. While Aβ plaques are poor correlates for the clinical symptomatology in AD and Down syndrome patients, soluble oligomers are suggested to be good predictors for synaptic loss [29], neurofibrillary tangles [30] and clinical phenotype [46]. Furthermore, memory impairment and pathological changes in many AD mouse models occur well before the onset of plaque deposition [47]. Albeit there are convincing genetic, biochemical and cell biological data pointing to a major role of Aβ in AD, growing evidence points towards soluble Aβ oligomers rather than Aβ precipitated in plaques. Blennow et al. [48] for example discussed whether Aβ deposition is the cause or consequence of neurodegeneration in sporadic AD, but also whether the transgenic mouse models are at all accurate models for sporadic AD.

Soluble oligomers are low molecular weight non-fibrillar structures, which are stable in aqueous solution and remain soluble even after high speed centrifugation. Aβ oligomers develop preferentially within neuronal processes and synapses rather than in the extracellular space [27, 28]. At high concentrations, vesicular full-length Aβ aggregates form high molecular weight oligomers which are capable of seeding amyloid fibril growth [49]. Results from several labs propose these oligomers to be the missing link in the amyloid hypothesis. Just like in the human brain, studies using AD mouse models support the pathogenic role of oligomers. In the Tg2576 mouse model, the appearance of Aβ dodecamers coincided with the onset of spatial memory impairment. Interestingly, injection of these purified oligomers into the ventricle of wildtype rats caused a dramatic drop in spatial memory performance [50]. With regard to short-term effects, oligomers have been shown to impair synaptic plasticity by blocking long term potentiation and reinforcing long term depression [51]. Another hint was reported by Tomiyama et al. [33], who generated APP transgenic mice expressing the E693Δ mutation causing neuronal cell death and cognitive impairment by enhanced intracellular Aβ oligomerization without plaque formation. Loss of Aβ clearance instead of increased Aβ generation has been considered to be involved in the pathology of the sporadic variant of AD [52]. The above mentioned thoughts consider full-length Aβ_{1–40/1–42} as the major culprit in AD pathology.

Of note, full-length Aβ peptides are physiological molecules produced throughout the life of a human being. The generation of N-truncated Aβ peptides has been suggested to increase toxicity [53]. Pike et al. [21] compared Aβ peptides with initial residues at positions 1, 4, 8, 12, and 17 and ending with residue 40 or 42 and showed that N-terminal deletions enhance Aβ aggregation in relation to full-length Aβ. Furthermore they reported that Aβ peptides exhibiting aggregation showed circular dichroism spectra consistent with predominant β-sheet conformation, fibrillar morphology under transmission electron microscopy, and significant toxicity in cultures of rat hippocampal neurons.

We have recently extended these observations and showed that soluble aggregates have specific features responsible for their neurotoxicity [34]. Aβ_4-40, Aβ_4-42, Aβ_1-42 and Aβ_pE3-42 were unstructured in the monomeric state [34]. However, upon heating the Aβ variants showed a high propensity to form folded structures, in particular the three most toxic variants Aβ_pE3-42, Aβ_1-42 and Aβ_4-42. In addition, monomeric Aβ_4-42 and Aβ_pE3-42 were rapidly converted to soluble aggregated species. Both N-truncated variants exhibited similar biochemical properties, which opens the discussion which one of them might be more important in AD pathology [34].

In the present report we endeavored to address this question. We succeeded to develop an antibody differentiating between full-length Aβ and the two other major N-truncated variants, Aβ_4-x and Aβ_pE3-x. In combination with two other antibodies exclusively reacting with Aβ_1-x (IC16) or Aβ_pE3-x (1–57), we were able to show that Aβ_4-x preceded Aβ_pE3-x accumulation in the brain of 5XFAD transgenic mice. More importantly, Aβ_4-x was detected together with Aβ_1-x in the intraneuronal compartment of cortical neurons prone to degenerate in 5XFAD mice at 12 months of age [39, 40]. Early and transient intraneuronal accumulation of Aβ correlated with subsequent neuron loss also in diverse APP/Aβ transgenic mouse models and brain regions [34, 41, 54–57]. Interestingly, such a transient appearance of intraneuronal Aβ_x-42 has also been described by Mori et al. [58] studying the brain of Down syndrome patients between 3 to 73 years. Using an antibody against the N-terminus of Aβ_pE3-x, no intraneuronal staining was reported [58] corroborating our observation of a lack of intraneuronal accumulation of Aβ_pE3-x in 6 week-old 5XFAD mice in the present study.

Using an in vitro toxicity assay, we were able to demonstrate that NT4X-167 is particularly protecting against Aβ_4–42 and that the binding to Aβ_pE3-42 has no therapeutic consequence. The mechanism(s) of the diverging biological effects are not clear. The Western blot analysis might not accurately reflect the difference in affinity of NT4X-167 between Aβ_4–42 and Aβ_pE3-42. The data from the in vitro toxicity assay provides evidence that NT4X-167 preferentially binds Aβ_4–42. On the other side it could also be that NT4X-167 does not efficiently bind to some toxic aggregate(s) of Aβ_pE3-42 as it did not significantly detect the aggregate at 50 kDa as compared to 1–57 antibody under native conditions. We have previously shown that passive immunization of 5XFAD mice with 9D5, a monoclonal antibody specifically detecting low molecular weight Aβ_pE3-x aggregates, significantly reduced overall Aβ plaque load and Aβ_pE3-x levels, and normalized behavioral deficits [59].

While amyloid plaques were observed using NT4X-167 in 5XFAD transgenic mice, it barely reacted with plaques in the brain of sporadic AD patients and familial cases with the Arctic, Swedish and the presenilin-1 mutation PS1∆9. These data are corroborated by a previous work by Kuo et al. [60]. They analyzed Aβ pathology using chemical and morphological approaches comparing the plaques of APP23 transgenic mice and human AD brain. The authors concluded that despite an apparent overall structural resemblance to AD pathology, the chemical analyses revealed that the amyloid plaque cores in APP23 transgenic mice were completely soluble in buffers containing SDS [60]. Human AD plaque cores were highly resistant to chemical and physical disruption accounting for the extreme stability of AD plaque cores [60]. Moreover, the corresponding lack of post-translational modifications such as N-terminal degradation, isomerization, racemization, pyroglutamyl formation, oxidation, and covalently linked dimers in transgenic mouse Aβ, provides an explanation for the differences in solubility between human AD and the APP23 mouse plaques [60]. NT4X-167 preferably stained Aβ in blood vessels in human specimens, in which Aβ_x-40 is a major component. The Aβ plaques in PS1Δ9 AD cases are characterized by cotton wool morphology composed by Aβ_x-42 aggregates. The lack of cotton wool plaque staining using NT4X-167 further strengthens the possibility that it may prefer binding to Aβ_4-40 as compared to Aβ_4-42 aggregates.

Selkoe and others reported that toxic Aβ oligomers are primarily dimers and trimers of Aβ [28, 61, 62]. Haass and Selkoe argued that small molecules that can specifically inhibit the formation of Aβ oligomers and/or prevent their binding to and stabilization on neuronal membranes is at the top in the search for an AD therapy [11]. More recently, De Strooper [63] discussed that it is more likely that several of the identified oligomeric species (derived from full-length Aβ) have similar or overlapping properties. They conclude that coexistence of several oligomeric populations that do or do not propagate into fibrils is possible. Despite the differences in structure, stability and concentration, all oligomers may contribute to Aβ toxicity. They further discussed some technical issues defining oligomers like the apparent ‘SDS resistance’ [63]. Bitan et al. [64] have demonstrated that SDS can artificially induce oligomerization of Aβ. Hepler et al. [65] were able to isolate monomers, trimers and tetramers as major bands derived from full-length Aβ oligomers, Aβ fibrils and Aβ monomers after SDS-PAGE separation. Our data are well in line with these previous observations. Under reducing conditions Aβ_pE3-40 and Aβ_4–40 generated monomers and dimers, while Aβ_pE3-42 and Aβ_4–42 in addition produced trimers and tetramers as previously shown [14, 59]. Using native conditions, Aβ_1–42 and Aβ_pE3-42 appeared as aggregates of different sizes with higher molecular weight aggregates. In contrast Aβ_4–40 and Aβ_4–42 ran as a single band at approx. 50 kDa.

In fact, analysis of amyloid deposits in AD brains revealed various N- and C-terminal variants [14, 17, 18]. The increased C-terminal length of Aβ (from Aβ_x-40 to Aβ_x-42) enhances its aggregation properties. Faster aggregation leads to earlier Aβ deposition, which is believed to promote its toxicity [20, 21, 66]. Recently, Aβ_1-43 was discovered as a novel toxic peptide in AD [67, 68]. Besides Aβ peptides starting with aspartate as the first amino acid (Aβ₁), several N-truncated and modified Aβ species have also been described [14–16, 69]. Aβ_4–42 being one of them is particularly interesting as its discovery dates back to 1985 by Masters et al. [14]. Lewis et al. [22] reported that Aβ_4-42 is a relatively abundant species in AD, aged controls and vascular dementia patients. Using immunoprecipitation in combination with mass spectrometry, Portelius and colleagues [25] showed that Aβ_1-40, Aβ_1-42, Aβ_pE3-42 and Aβ_4-42 can be detected in the hippocampus and cortex of AD patients. Moreover, it has been demonstrated that N-terminal deletions enhance Aβ aggregation comparing Aβ_4-42 with Aβ_1-42 [21]. Youssef et al. [38] showed that Aβ_1-42 and Aβ_pE3-42 exhibited similar effects on neuronal cytotoxicity in primary cortical neurons and on memory impairment after intracerebroventricular injection in wildtype mice. Aβ_pE3-42 is now an established factor contributing to AD pathology [53] and may even be aggravating the severity of the disease [70]. Sergeant et al. demonstrated that amino-truncated Aβ species represented more than 60% of all Aβ species, not only in full blown AD, but also, and more interestingly, at the earliest stage of AD pathology [71]. They concluded, that a vaccine specifically targeting these pathological amino-truncated species of Aβ_x-42 are likely to be promising, by inducing the production of specific antibodies against pathological Aβ products that are, in addition, involved in the early and basic mechanisms of amyloidosis in the human brain.

The importance of position four of Aβ is corroborated by Haupt et al. [72], who observed an N-terminal β-strand, previously assumed to be an unstructured region [73–77]. Using proline mutagenesis to probe the structural relevance of N-terminal residues, they demonstrated that mutations affecting residues 4 or 8, significantly increased the fraction of elongated aggregates indicating that disrupting the N-terminal β-strand favors protofibrils relative to oligomers [72]. The pathological impact of Aβ_4–42 is elucidated by the generation of transgenic mice (Tg4-42) expressing Aβ_4–42 [34]. The Tg4-42 mice develop a severe age-dependent spatial reference memory deficit and massive hippocampus neuron loss.

At present, the enzymes responsible for N-terminal truncation are not well studied. Aminopeptidase A contributes to the N-terminal truncation of Aβ peptide producing Aβ_2-x [78]. Saido et al. [16] suggested that mono- or dipeptidylaminopeptidases cleave Aβ_1-x producing N-terminal truncated Aβ_3-x. Aβ_pE3-x formation is catalyzed by glutaminyl cyclase [79–82]. Which enzymes are involved in the truncation steps to generate Aβ_3-x and Aβ_4-x is unknown.

The present report describes the binding properties of the novel antibody NT4X-167, which recognizes the N-terminus of N-truncated Aβ. NT4X-167 bound most efficiently to Aβ_4-x. Phenylalanine at position four of Aβ was imperative for NT4X-167 binding. In vitro toxicity experiments demonstrated that Aβ_4–42 induced neuron death was significantly rescued by NT4X-167 treatment. No rescue effect was observed for Aβ_1–42 or Aβ_pE3-42 toxicity. NT4X-167 detected only a minor fraction of plaques in brain from sporadic and familial AD patients and 5XFAD transgenic mice. It preferentially reacted with intraneuronal Aβ in young 5XFAD mice. The finding that Aβ_4-x precedes Aβ_pE3-x in the well accepted 5XFAD AD mouse model further underlines the significance of Aβ_4-x. NT4X-167 did not cross-react with aggregates typical for other major neurodegenerative disorders implicating that the recognized aggregates are specific for AD. Taking all observations together, NT4X-167 represents a novel tool for AD research and therapy.