In addition to full-length Aβ peptides starting with an aspartate in position 1 (Aβ
1–x), a variety of N-terminally truncated and posttranslationally modified Aβ peptides have been detected in human AD brains [
14,
16,
17]. N-terminal truncations were shown to promote the aggregation propensity of Aβ peptides [
18]. However, the presence of N-truncated peptides has been demonstrated mainly by mass spectrometry following immunoprecipitation with generic Aβ antibodies such as 4G8 or 6E10 [
14,
15,
21], and their genesis in vivo is mostly unresolved. Consequently, the functional role of N-truncated Aβ peptides in the pathogenesis of AD has remained unclear, including fundamental questions about the abundance and distribution of N-truncated isoforms. This shortcoming has been particularly obvious for the Aβ
4–x peptides starting with Phe in position 4, which have been proposed to be an abundant Aβ species in AD brains but for which no specific antibodies have been generated to date. We have previously reported a monoclonal antibody, NT4X-167, that preferentially detected Aβ
4–x peptides and protected primary cortical neurons from the toxicity of Aβ
4–42 peptides in vitro [
30]. However, in addition to monomeric and oligomeric Aβ
4–x peptides, NT4X-167 was shown to recognize Aβ
pE3–x peptides. Hence, this antibody is not suitable for accurate measurement of the abundance and distribution of Aβ
4–x peptides. We have now raised polyclonal antibodies by immunizing guinea pigs with the six-amino acid peptide (FRHDSG) corresponding to residues 4–9 of the Aβ peptide sequence. The specificity of these antibodies for Aβ
4–x peptides was confirmed by CIEF immunoassay and urea SDS-PAGE, and no cross-reactivity for Aβ
1–40, Aβ
2–40, Aβ
3–40, Aβ
pE3–40, and Aβ
5–40 was observed. Furthermore, in immunohistochemical staining, the immunoreactivity could be entirely blocked by preabsorption with Aβ
4–40 but not Aβ
1–40 peptides. Two independent animals were immunized and yielded two antisera, 029-1 and 029-2, with nearly identical immunoreactivity, indicating that Aβ
4–9 might be a reliable immunogen to raise Aβ
4–x-specific antibodies in guinea pigs. Compared with the well-established antibody IC16, which preferentially detects full-length Aβ peptides, the immunohistochemical staining patterns of the newly generated Aβ
4–x-specific antibodies were quantitatively and qualitatively different. In brain sections of both patients with sporadic AD and two AD mouse models, the distribution of Aβ
4–x peptides was restricted largely to amyloid plaque cores and CAA, whereas diffuse amyloid deposits were negative. The presence of Aβ
4–x peptides in amyloid plaque cores raises the question whether these truncated species are critical in the very early stages of the pathology. We have not yet conducted a comprehensive longitudinal study comparing different animal ages and time points before or after the onset of amyloid deposition. However, using two-dimensional Western blotting combined with mass spectrometry, N-terminally truncated Aβ peptide species starting at position 4 or 5 have already been detected at 2.5 months of age in the APP/PS1KI line, indeed indicating a very early appearance of these truncated species [
26]. In good agreement and using a similar experimental approach, Sergeant and colleagues reported that Aβ aggregates at the first stages of amyloid deposition in nondemented individuals with amyloid and tau pathologies are composed predominantly of N-truncated variants, including Aβ
4–x peptides [
31]. Antibodies raised in guinea pigs are especially useful for colocalization studies because most high-quality antibodies against other Aβ species or APP fragments have previously been generated in either mice or rabbits. Indeed, double-immunofluorescence staining demonstrated Aβ
4–x-positive amyloid plaque cores decorated by APP-positive dystrophic neurites with no overlap in the fluorescent signals. In line with this observation, no intraneuronal staining was observed for Aβ
4–x peptides in mice at the ages of 8–10 months. However, it could be worth studying younger animals because intraneuronal Aβ accumulation is most prominent in young mice prior to amyloid plaque formation [
32]. Overall, the distribution of Aβ
4–x peptides was also substantially different from the staining pattern reported for other N-truncated species, including Aβ
2–x [
7] or Aβ
5–x [
10,
11], which were not or less confined to cored neuritic plaques. Previous studies have demonstrated that Aβ
4–x peptides rapidly formed soluble oligomers and fibrillar higher-molecular-weight aggregates [
33]. This biochemical property might explain not only the confined localization of Aβ
4–x peptides to amyloid cores but also their high neurotoxicity in vitro and in vivo. Short-term exposure of primary cortical neuron cultures to Aβ
4–40 and Aβ
4–42 peptides resulted in a concentration-dependent cytotoxic effect with comparable effect sizes to Aβ
1–42. Furthermore, the expression of Aβ
4–42 under the control of a neuronal promotor caused age-dependent behavioral deficits and hippocampal neuron loss in a transgenic mouse model (Tg4-42) [
33].
Another important unresolved issue is the abundance of Aβ
4–x peptides in relation to full-length Aβ peptides in both AD and transgenic mouse models of the disease. To start to address this issue and to evaluate the novel Aβ
4–x antibodies for quantitative analysis, we combined the 029-2 antibody with a C-terminus-specific Aβ
40 antibody [
29] in a sandwich ELISA. In the SDS-soluble brain fraction of 5-month-old heterozygous 5XFAD mice, this assay detected around 1 ng of Aβ
4–40 per milligram of tissue with approximately fivefold higher levels in homozygous 5xFAD mice of the same age. In a previous study, we had determined the levels of full-length Aβ
1–40 and Aβ
1–42 peptides in the same brain extracts of the same animal cohort with a comparable ELISA system and IC16 as a capture antibody [
28]. Combining the results from both studies indicates that Aβ
1–40 and Aβ
1–42 peptides are approximately 75-fold and 200-fold more abundant, respectively, than Aβ
4–40 peptides in the 5XFAD mouse model at 5 months of age. This also fits with a matrix-assisted laser desorption/ionization time-of-flight mass spectrometric analysis of formic acid brain extracts of 7-month-old 5XFAD mice. Although peptide concentrations cannot be deduced from peak heights in mass spectra, this analysis showed that the signal intensities generated by the Aβ
4–40 and Aβ
4–42 peptides were only a small fraction of the Aβ
1–40 and Aβ
1–42 peaks [
5]. Taken together, these results indicate that, quantitatively, N-truncated Aβ
4–x peptides are a minor Aβ species in the 5XFAD mouse model. However, evidence from human studies suggests that the proportions of N-truncated to full-length Aβ peptides might be substantially different in the brains of patients with AD. Already more than 30 years ago, it was shown by N-terminal sequencing analysis of Aβ peptides purified from amyloid plaque cores that only around 10% of peptides displayed an intact N-terminus, whereas > 60% started with the Phe residue in position 4 [
19]. Another sequencing study confirmed that Aβ peptides starting with Phe represented the major component of plaques, whereas full-length Aβ starting with Asp was detected predominantly in the vasculature [
20]. In addition, later studies using mass spectrometry have generally supported that Aβ
1–42, Aβ
pE3–42, and Aβ
4–42 belong to the Aβ peptide species with a high prevalence in AD brains [
14,
15,
21,
34]. Finally, comparative biochemical studies of amyloid plaques isolated from human AD brains and APP transgenic mouse models have also shown that N-truncated Aβ peptides are much more prevalent in patients with AD, and it has been proposed that the greater abundance of N-truncated Aβ peptides is at least in part responsible for the substantially lower solubility of amyloid plaque material from humans [
35,
36]. In any case, additional studies in human AD brains using genuinely quantitative methods to determine the abundance of Aβ
4–x peptides and their distribution in soluble and insoluble brain fractions are clearly warranted. How Aβ
4–x peptides are generated remains entirely unclear. Although there is evidence that enzymatic activities can facilitate N-terminal Aβ truncations such as Aβ
2–x [
37,
38], no enzymes have yet been identified that are able to generate Aβ
4–x peptides. Nonenzymatic generation of truncated Aβ peptides has also been proposed, and it has been shown that full-length Aβ peptides can spontaneously decompose into shorter N- and C-terminally truncated isoforms in vitro in the absence of proteases [
39,
40]. With respect to N-terminal truncations, spontaneous decomposition of full-length to Aβ
3–x peptides with subsequent cyclization of the N-terminal Glu to pGlu has been demonstrated, but there is only very limited evidence that a similar process could produce Aβ
4–x peptides [
39]. Our finding that Aβ
4–x peptides are confined largely to amyloid cores supports an important role of these N-truncated Aβ species in the process of amyloid plaque formation. Beyond that, further understanding of the pathological relevance of Aβ
4–x peptides will likely require clarification of their origin and the subsequent generation of genetic loss-of-function models.