Decreased Deposition of Beta-Amyloid 1-38 and Increased Deposition of Beta-Amyloid 1-42 in Brain Tissue of Presenilin-1 E280A Familial Alzheimer’s Disease Patients

Familial Alzheimer’s Disease (FAD) caused by Presenilin-1 (PS1) mutations is characterized by early onset, cognitive impairment, and dementia. Impaired gamma secretase function favors production of longer beta-amyloid species in PS1 FAD. The PS1 E280A mutation is the largest FAD kindred under study. Here, we studied beta-amyloid deposits in PS1 E280A FAD brains in comparison to sporadic Alzheimer’s disease (SAD). We analyzed cortices and cerebellum from 10 FAD and 10 SAD brains using immunohistochemistry to determine total beta-amyloid, hyperphosphorylated tau (pTau), and specific beta-amyloid peptides 1-38, 1-40, 1-42, and 1-43. Additionally, we studied beta-amyloid subspecies by ELISA, and vessel pathology was detected with beta-amyloid 1-42 and truncated pyroglutamylated beta-amyloid antibodies. There were no significant differences in total beta-amyloid signal between SAD and FAD. Beta-amyloid 1-38 and 1-43 loads were increased, and 1-42 loads were decreased in frontal cortices of SAD when compared to FAD. Beta-amyloid species assessment by ELISA resembled our findings by immunohistochemical analysis. Differences in beta-amyloid 1-38 and 1-42 levels between SAD and FAD were evidenced by using beta-amyloid length-specific antibodies, reflecting a gamma secretase-dependent shift in beta-amyloid processing in FAD cases. The use of beta-amyloid length-specific antibodies for postmortem assessment of beta-amyloid pathology can differentiate between SAD and PS1 FAD cases and it can be useful for identification of SAD cases potentially affected with gamma secretase dysfunction.


INTRODUCTION
Alzheimer's disease (AD) is the most common form of dementia in the elderly. The prevalence is estimated to be about 24 to 35 million people worldwide (Reitz and Mayeux, 2014). Clinically, it is characterized by progressive cognitive impairment and behavioral changes, which eventually lead to dementia and death (Giaccone et al., 2011). Neuropathological hallmarks are brain atrophy, beta-amyloid (Aβ) deposits forming plaques in the extracellular space and the accumulation of intraneuronal hyperphosphorylated tau protein (pTau) as neurofibrillary tangles (NFTs) (Alzheimer's Association, 2010). Age is one of the main risk factors for developing AD (Berr et al., 2005). AD can be divided into two groups: early onset AD (EOAD) occurring before 65 years of age, while 95% of all AD cases are late onset AD (LOAD) and occur after 65 years of age (Alzheimer's Association, 2010). The presence of allele 4 of Apolipoprotein E (ApoE4) is the main genetic risk factor for LOAD; ∼40% of these patients show at least one ApoE4 allele (Myers et al., 1996;Bettens et al., 2013).
Given that abnormal Aβ processing into longer Aβ species is a known feature in PS1 FAD (Chávez-Gutiérrez et al., 2012) and in PS1 E280A (Van Vickle et al., 2008), we studied morphology and distribution profiles of different Aβ species in brain regions of PS1 E280A FAD patients to make a direct comparison with SAD brain tissue. We found a specific Aβ 1-38/1-42 immunohistochemical profile associated with FAD when compared with SAD and decreased Aβ 1-42/3pE-x ratio of affected vessels in FAD.

Patients and Human Brain Samples
Presenilin-1 E280A genealogy was identified 30 years ago and mutation carriers have been followed up since then. Carriers for the present study were identified and affected patients went through neurological and neuropsychological characterization and were followed up using the CERAD protocol, NINCDS-ARDA, and DSM-IV criteria until end-stage dementia and death (Acosta-Baena et al., 2011;Sepulveda-Falla et al., 2011). For neuropathological studies, we collected postmortem brain tissue of 10 PS1-E280A patients and matched them with 10 SAD patients ( Table 1) showing corresponding CERAD and Braak stages (Braak and Braak, 1991;Gearing et al., 1995) (neuropathological analysis were performed by MG and DSF, Table 1). Brain donation and procedures were performed following ethical approval from respective institutions (Universidad de Antioquia, UKE). Formalin-fixed samples from frontal, temporal, parietal, occipital cortex, and cerebellum were selected for morphological studies. Further studies of Aβ species were conducted in frontal cortices.
Supernatant was collected and probed with the ELISA kit for each antigen. Samples were measured at 450 nm in a BioTek mQuant spectrophotometer (Winooski, VT, United States) and expressed as pmol/mg of total protein.

Statistical Analysis
Data were analyzed using IBM SPSS Statistics 22 software (IBM/SPSS Inc., Armonk, NY, United States) and GraphPad Prism 6 (GraphPad Software, Inc., La Jolla, CA, United States). Analyses included Kolmogorov-Smirnov and Shapiro-Wilk tests for normal distribution assessment. For non-parametric comparisons, we applied Mann-Whitney U (given as Z) for two group comparisons and Chi square test was applied to categorical variables. Correlation analysis was performed using Spearman Rho's test. Statistical significance of all analyses was determined with * p ≤ 0.05, * * p ≤ 0.01, and * * * p ≤ 0.001.

Similar Morphology and Distribution of 6E10 Detected Aβ Pathology in SAD and PS1 E280A FAD
Detected Aβ immunosignal in FAD did not differ significantly from SAD in all five brain regions studied (Figures 1A,B). Both SAD and FAD cases showed diffuse, core, and neuritic Quantification of Aβ immunosignal present in frontal cortex, temporal cortex, parietal cortex, occipital cortex, and cerebellum. No statistically significant differences in all evaluated areas were observed between SAD and FAD cases. (C) Radar chart for the 6E10 percentage of immunosignal distribution according to brain regions in SAD (blue) and FAD (red) cases. Both groups present a FC-predominant distribution pattern.
(D) Radar charts for individual 6E10 immunosignal distribution pattern for all SAD (blue) and FAD (red) cases studied. FC predominant distribution pattern was identified among both groups, with seven FAD cases and five SAD cases. TC and PC patterns were identified in SAD and FAD cases, respectively, with two cases for each. Some cases presented dissimilar patterns in both groups, with three SAD cases and one FAD case.
Aβ plaques. FAD cases present with abundant diffuse plaques, as previously described (Shepherd et al., 2009). SAD cases also showed abundant diffuse plaques, particularly in frontal and temporal cortices. There was high variability in the size and distribution of plaques in both groups and studied group regions ( Figure 1A). The highest average immunosignal was detected in the frontal cortex of FAD as well as of SAD patients, followed by temporal cortex, parietal cortex, and occipital cortex, with the lowest Aβ signal being detected in the cerebellum ( Figure 1B and Table 2), creating a predominantly frontal aggregation pattern ( Figure 1C). When individual aggregation patterns were analyzed, we found that 5/10 SAD and 7/10 FAD cases showed frontal cortex predominance. Two out of 10 SAD showed temporal cortex predominance while 2/10 FAD showed parietal cortex predominance ( Figure 1D).

Differential pTau Pathology and
Increased AT8 Immunosignal in PS1 E280A FAD It has been previously suggested that Aβ aggregation drives pTau pathology in AD (Stancu et al., 2014). Both SAD and FAD groups presented with abundant pTau AT8 diffuse signal Frontiers in Aging Neuroscience | www.frontiersin.org and aggregates (Figure 2A). We found significantly elevated levels of pTau in FAD compared to SAD in occipital cortex, which represented the most affected brain region in both groups. Lowest pTau signals were observed in cerebellum, whereas previously described  FAD cases showed significantly more signal than SAD cases ( Figure 2B and Table 2). As with Aβ, we analyzed pTau pathology distribution patterns, with average patterns showing a similar predominance of temporal and occipital cortices signal (Figure 2C), Regarding individual patterns, 3/10 SAD and 2/10 FAD cases showed parietal predominance. Four out of 10 SAD and 1/10 FAD cases showed temporo-occipital predominant pattern, while temporal (3/10) and occipital (4/10) predominant patterns were identified in FAD. Interestingly, two FAD cases showed high deposits in all cortices but in the temporal cortex ( Figure 2D). Given that general deposition patterns for Aβ and pTau did not show clear differences between SAD and FAD, we analyzed the correlations between both signals for each of the studied regions. Besides the correlation between Aβ in cerebellum and pTau in parietal cortex, no significant correlations between 6E10 and AT8 immunosignals for any other evaluated regions were observed (Supplementary Figure S1 and Supplementary Table S1).

Characteristic Aβ Species Profile in the Frontal Cortex of PS1 E280A FAD and SAD Patients
Given that APP processing and Aβ production are thought to be main features in FAD (Moro et al., 2012), we analyzed deposition of specific Aβ species in frontal cortices of both AD groups (Figures 3A,B). Aβ 1-38 and Aβ 1-43 showed statistically significant higher levels, while Aβ 1-42 showed statistically significant lower levels in SAD compared to FAD. No differences between FAD and SAD were found for Aβ 1-40 (Figures 4A,B). It has been suggested that differences in Aβ ratios can be pathologically relevant (Pauwels et al., 2012). By calculating the Aβ 1-38/1-42, the Aβ 1-42/1-40, and the Aβ 1-40/1-43 ratios, we found significantly higher values in SAD for Aβ 1-38/1-42, and higher values in FAD than in SAD for Aβ (B) Quantification of pTau immunosignal in both groups. FAD patients showed significantly higher levels in occipital cortex and cerebellum. * p ≤ 0.05. (C) Radar chart for the AT8 percentage of immunosignal distribution according to brain regions in SAD (blue) and FAD (red) cases. Both groups present a TC/OC predominant distribution pattern. (D) Radar charts for individual AT8 immunosignal distribution pattern for all SAD and FAD cases studied. TC/OC predominant (in which both TC and OC were equally affected) patterns were identified for four SAD and one FAD case. Three PC predominant SAD cases and two FAD cases, one OC predominant SAD case and four FAD cases, one TC predominant case and two FAD cases, and one case of each group presented FC predominant distribution.

Characterization of Pyroglutamylated Aβ and CAA Pathology in PS1 E280A FAD
We tested one specific antibody against Pyroglutamylated Aβ 3 to 40-42 (Pyroglu. Aβ 3pE-x) in frontal cortex of SAD and FAD cases. We detected equally strong amyloid pathology in both groups (Figure 6A). The Pyroglu. Aβ 3pE-x antibody showed strong signal in the vessels of both SAD and FAD cases ( Figure 6B). Previously, it has been reported that FAD presents with higher CAA scores than SAD (Ringman et al., 2016). We evaluated CAA pathology using Aβ 1-42 and the Pyroglu. Aβ 3pE-x antibodies. We focused on the number of vessels that presented grade 3 CAA pathology assessed according to the Vonsattel grading system (Vonsattel et al., 1991). For both antibodies, there were no statistically significant differences in severe CAA pathology between groups (Figures 6B,C). When an Aβ 1-42/Pyroglu. Aβ 3pE-x ratio was calculated for severe CAA affected vessels, SAD cases presented a statistically significant higher CAA Aβ 1-42/Pyroglu. Aβ 3pE-x ratio when compared to FAD cases ( Figure 6C).

DISCUSSION
Previously we reported specific neuropathological differences between FAD and early onset SAD, with increased 6E10 positive Aβ levels in the frontal cortex and cerebellum of PS1 E280A FAD cases , and some Aβ morphological differences in another subset of PS1 E280A patients (Trujillo-Rodríguez et al., 2014). In the present study, we observed no statistically significant differences in 6E10 positive signal from all brain cortices and cerebellum between PS1 E280A FAD cases and similarly, affected SAD cases. It should be kept in mind that 6E10 antibody also stains full length APP containing Aβ domain. However, since antibody positive signal referred specifically to plaques, only pathologic deposits were evaluated. It is possible that early-onset SAD cases might have lower Aβ levels when compared with later-onset SAD (Wolfe, 2007). Both AD groups showed increased Aβ pathology in the frontal cortex, also as a predominant pattern among individuals. This result is in agreement with previous findings regarding Aβ pathology in AD according to PET studies in SAD cases (Barthel et al., 2011) and may reflect region-specific susceptibility toward Aβ pathology (Braak and Del Trecidi, 2011;Thal et al., 2015). It has been reported that Aβ pathology in PS1 FAD is characterized by a diffuse pattern (Shepherd et al., 2009). Our set of severely affected SAD cases showed a similarly, extended diffuse pattern indistinguishable from the observed pattern in FAD when using 6E10A antibody.
For analyzing morphology and distribution of Aβ 1-38, 1-40, and 1-42 peptides, we used a set of antibodies developed by Janssen R&D [43] and characterized by high specificity and sensitivity. We identified a clear and complementary difference of Aβ 1-38 and Aβ 1-42 immunosignals in frontal cortex between FAD and SAD cases, suggesting that mutated PS1 favors Aβ 1-42 over Aβ 1-38 production, confirming previous findings (Szaruga et al., 2017). Even though increased Aβ 1-42 immunosignal in this population (Lemere et al., 1996) andin FAD in general (Kumar-Singh et al., 2006) has been widely reported, decreased 1-38 signal is a novel neuropathological finding in agreement with altered γ secretase function as described in vitro for PS1 FAD (Chávez-Gutiérrez et al., 2012;Szaruga et al., 2017) and in contrast to a recent report studying 1-38 pathology in APP, PS1, and PS2 FAD cases comparing them with SAD. In that report, neither PS1 mutant carriers nor SAD cases without CAA presented with Aβ 1-38 pathology. The PS1 E280A mutation was not included (Moro et al., 2012). In our study all severely affected SAD cases showed higher Aβ 1-38 loads when compared to PS1 E280A FAD. The differences between studies could be explained by the severity of studied SAD patients and specificity/sensitivity of the antibodies used.
Regarding Aβ 1-43 and 1-40 immunosignal, only Aβ 1-43 showed higher levels in SAD. Previous reports in human brain tissue indicated Aβ 1-43 to be more frequently observed in the core of Aβ plaques (Welander et al., 2009) and to accumulate more than Aβ 1-40 in SAD brains (Saito et al., 2011). It is to be noted that in the first study only one case of PS1 FAD was investigated. Furthermore, by evaluating Aβ 1-38/1-42 and Aβ 1-40/1-43 ratios, we confirm cell culture-based studies suggesting FAD-specific Aβ production (Chávez-Gutiérrez et al., 2012). Higher levels of Aβ 1-38 and Aβ 1-43 are favored in SAD while Aβ 1-42 production is specifically increased in FAD. Interestingly, the Aβ 1-42/1-40 ratio showed no difference between FAD and SAD cases, which is in opposition to previous findings (Kumar-Singh et al., 2006). This ratio could vary depending on PS1 mutation and on AD pathology severity of the SAD cases studied. Recently, the pathogenicity of Aβ 1-43 in FAD has come to the fore. It has been described that increased production of Aβ 1-43 in specific PS1 FAD mutations is a reflection of increased γ secretase dysfunction (Trambauer et al., 2020). Perhaps low Aβ 1-43 levels detected in our FAD samples are related with the  Figure 3). Both, SAD and FAD cases present with severe CAA, with similar percentage of vessels affected in two thirds of their perimeter (% of +++ vessels, n = 20). (C) Characterization of CAA using a Pyroglutamylated Aβ 3pE-x antibody (as seen in Figure 3). Both SAD and FAD cases present with severe CAA, with similar percentage of vessels affected in two thirds of their perimeter (% of +++ vessels, n = 20). relatively mild γ secretase dysfunction observed in PS1 E280A brain homogenates (Szaruga et al., 2015).
One possible objection for morphological differences in Aβ species detection could be that observed changes represent the endpoint of Aβ pathology (Serrano-Pozo et al., 2012). Thus, we evaluated Aβ species by ELISA. Aβ 1-38, 1-40, and 1-42 results resemble our histological findings. None of the calculated Aβ ratios (as measured by ELISA) were significantly different as opposed to histological findings and in contrast to some of our previous results obtained with less severely affected SAD cases (Sepulveda-Falla et al., 2012). Furthermore, Aβ as detected by ELISA may not be identical to Aβ detected by in situ methods. Aβ deposits can reflect long-term Aβ pathology better because only a continuous phenomenon could generate them, whether by increased production or impaired degradation. We propose that Aβ 1-38/1-42 and Aβ 1-40/1-43 ratios as assessed by immunohistochemistry are a sensitive enough tool for the neuropathological study of both AD groups. Our study shows Aβ species profile in only one PS1 mutation. Other PS1 mutations could present with different Aβ species profiles. However, among PS1 mutations, there are common features such as Aβ 1-38/1-42 ratio (Li et al., 2016;Szaruga et al., 2017) that could result in similar IHC findings.
Post-translational modifications of Aβ, like pyroglutamylated and truncated forms, deposit alongside other Aβ species (Cabrera et al., 2018). Pyroglu. Aβ 3pE-x have been identified as depositing early in humans and primates while depositing later in transgenic mouse models (Frost et al., 2013). It seems to be specifically associated with AD pathology in contrast to age-associated Aβ deposits (Moro et al., 2018), or to pure CAA (Gkanatsiou et al., 2019). In our analysis, both severely affected SAD and PS1 E280A FAD frontal cortices showed equally extensive Pyroglu. Aβ 3pEx plaque and vessel pathology. However, Pyroglu. Aβ 3pE-x antibody detected relatively more severely affected vessels in FAD than those detected with Aβ 1-42 antibody. This difference could be related with the temporal order in which different Aβ species are deposited in SAD and FAD. For FAD, higher Aβ 1-42 generation is constant throughout life, while Pyroglu. Aβ 3pE-x deposits can be associated with aging (Moro et al., 2018). Further studies should assess if truncated Aβ species play a larger role in CAA in PS1 FAD. Besides, vascular impact of Pyroglu. Aβ 3pE-x aggregation can be of relevance given the possible use of pyroglutamylated Aβ detection in plasma as a biomarker for plaque pathology in AD (Wang et al., 2020).
We have previously reported neuropathological differences between PS1 E280A FAD and SAD cases, including differential pTau pathology , distinct APP processing patterns (Pera et al., 2013), and mitochondrial damage and decreased Ca 2+ channels in the cerebellum (Sepulveda-Falla et al., 2014). In this study, we describe yet another difference between PS1 E280A FAD and SAD, regarding Aβ pathology and possibly linked to the underlying γ secretase dysfunction in PS1 FAD cases. We propose the usage of specific Aβ 1-38, 1-40, 1-42, and 1-43 antibodies and their immunosignal values and ratios as a viable option for better assessment of FAD cases and the identification of SAD cases with similarly, altered Aβ production to be studied further.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article can be made available by the corresponding author upon reasonable request.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Comité de Bioética, Facultad de Medicina, Universidad de Antioquia. Acta número 011 de 2016. The patients/participants provided their written informed consent to participate in this study.