Chemical traits of cerebral amyloid angiopathy in familial British‐, Danish‐, and non‐Alzheimerʼs dementias

Abstract Familial British dementia (FBD) and familial Danish dementia (FDD) are autosomal dominant forms of dementia caused by mutations in the integral membrane protein 2B (ITM2B, also known as BRI2) gene. Secretase processing of mutant BRI2 leads to secretion and deposition of BRI2‐derived amyloidogenic peptides, ABri and ADan that resemble APP/β‐amyloid (Aβ) pathology, which is characteristic of Alzheimer's disease (AD). Amyloid pathology in FBD/FDD manifests itself predominantly in the microvasculature by ABri/ADan containing cerebral amyloid angiopathy (CAA). While ABri and ADan peptide sequences differ only in a few C‐terminal amino acids, CAA in FDD is characterized by co‐aggregation of ADan with Aβ, while in contrast no Aβ deposition is observed in FBD. The fact that FDD patients display an earlier and more severe disease onset than FBD suggests a potential role of ADan and Aβ co‐aggregation that promotes a more rapid disease progression in FDD compared to FBD. It is therefore critical to delineate the chemical signatures of amyloid aggregation in these two vascular dementias. This in turn will increase the knowledge on the pathophysiology of these diseases and the pathogenic role of heterogenous amyloid peptide interactions and deposition, respectively. Herein, we used matrix‐assisted laser desorption/ionization mass spectrometry imaging (MALDI‐MSI) in combination with hyperspectral, confocal microscopy based on luminescent conjugated oligothiophene probes (LCO) to delineate the structural traits and associated amyloid peptide patterns of single CAA in postmortem brain tissue of patients with FBD, FDD as well as sporadic CAA without AD (CAA+) that show pronounced CAA without parenchymal plaques. The results show that CAA in both FBD and FDD consist of N‐terminally truncated‐ and pyroglutamate‐modified amyloid peptide species (ADan and ABri), but that ADan peptides in FDD are also extensively C‐terminally truncated as compared to ABri in FBD, which contributes to hydrophobicity of ADan species. Further, CAA in FDD showed co‐deposition with Aβ x‐42 and Aβ x‐40 species. CAA+ vessels were structurally more mature than FDD/FBD CAA and contained significant amounts of pyroglutamated Aβ. When compared with FDD, Aβ in CAA+ showed more C‐terminal and less N‐terminally truncations. In FDD, ADan showed spatial co‐localization with Aβ3pE‐40 and Aβ3‐40 but not with Aβx‐42 species. This suggests an increased aggregation propensity of Aβ in FDD that promotes co‐aggregation of both Aβ and ADan. Further, CAA maturity appears to be mainly governed by Aβ content based on the significantly higher 500/580 patterns observed in CAA+ than in FDD and FBD, respectively. Together this is the first study of its kind on comprehensive delineation of Bri2 and APP‐derived amyloid peptides in single vascular plaques in both FDD/FBD and sporadic CAA that provides new insight in non‐AD‐related vascular amyloid pathology. Cover Image for this issue: https://doi.org/10.1111/jnc.15424


| INTRODUC TI ON
Familial British dementia (FDB) Holton et al., 2001;Lashley 2008) and familial Danish dementia (FDD) (Holton et al., 2002;Tomidokoro et al., 2005), are two forms of hereditary cerebral amyloid angiopathies (CAA) that lead to the deposition of amyloid peptides in small-to-medium-sized cerebral, leptomeningeal and parenchymal arteries. Collectively referred to as chromosome 13 dementias, these conditions are associated with furin-based processing of mutated integral membrane protein 2B (ITM2B, also known as BRI2) precursor protein, that results in the production of ABri or ADan amyloid peptides, respectively  ( Figure S1). In FBD a point mutation, while in FDD a decamer duplication insertion, abolishes the stop codon of the BRI2 gene resulting in C-terminally elongated precursor proteins (Ghiso et al., , 2006Rostagno et al., 2005Rostagno et al., , 2010Vidal et al., 2000). In vitro studies of synthetic homologs of full-length ABri and ADan both have been reported to have a higher tendency to aggregate than Aβx-42 (Peng et al., 2010). Just as for Aβ, both the ABri and ADan peptide display multiple proteoforms, with N-terminal pyroglutamate formation being a prominent feature (Revesz et al., 2009;Saul et al., 2013).
Interestingly, while the ABri peptide appears to be the sole component of amyloid lesions in FBD , the FDD amyloid deposits have been reported to display the presence of Aβ peptides in addition to the ADan proteoforms (Holton et al., 2002;Tomidokoro et al., 2005). Given that FDD shows an earlier disease onset than FBD, the question arises whether co-aggregation of Aβ and the distinct chemical properties of ADan deposition, respectively, are associated with an accelerated pathology.
The amino acid sequences of the ABri and ADan peptides do not differ for the first 22 amino acids, suggesting that it is the Cterminus that might be responsible for possible co-aggregation of the ADan and Aβ peptides in FDD, despite that there is no sequence homology between ADan and Aβ. Surprisingly, biochemical analysis of the Aβ peptides present in the FDD lesions displayed predominantly Aβx-42, rather than Aβx-40 as would be expected in Aβ cerebral amyloidosis (Gravina et al., 1995;Miller et al., 1993;Tomidokoro et al., 2005). The exact role of C-terminal Aβ truncation and its effect on ADan peptide processing, degradation and co-aggregation is thus unclear.
To further understand this interplay of FDD and Aβ it is critical to further elucidate the molecular architecture and chemical composition of CAA pathology in those diseases. Here, it was appropriate to compare CAA in FDD to cases with sporadic CAA pathology withoud AD (CAA+) that are similar to FDD and show pronounced CAA formation but are devoid of parenchymal plaques, which is significantly different from plaque pathology in AD (Greenberg et al., 2020;Kövari et al., 2013).
In classical neuropathological studies, immunohistochemistry (IHC) has been used to determine the localization of the amyloidogenic peptides in brain tissues. However, the reliability of the results highly depends on the pretreatments and quality of the antibodies, and the method cannot identify and distinguish different isoforms with confidence and there are limitations on simultaneous detection of the different isoforms.
Indeed, key developments in bioanalytical techniques, including chemical imaging, have increased our understanding of the molecular basis of single Aβ deposit formation at subcellular scales.
Particularly, mass spectrometry imaging (MSI) using matrix-assisted laser desorption ionization (MALDI) has been demonstrated to be a valuable approach for studying biochemical traits of Aβ pathology in both postmortem patient brains and AD mouse models warranted by the molecular specificity inherent to this technique (Michno et al., 2019(Michno et al., , 2020(Michno et al., , 2021. In the present study, we combined MALDI MS imaging of Aβ, ABri, and ADan across multiple brain regions with fluorescent, structural amyloid microscopy on the same section using structure sensitive, amyloid probes (luminescent conjugated oligothiophene, LCOs, [Nystrom et al., 2013]). This combination allowed us to delineate structural differences among CAA lesions in FBD, FDD, and CAA+ across different regions, and correlated them with ABri, ADan, and Aβ.

| Patientcharacteristics
Human postmortem brain tissue was obtained through the brain donation program at Queen Square Brain Bank for Neurological

Disorders (QSBB), Department of Clinical and Movement
Neurosciences, Institute of Neurology, University College London (UCL).
Standard diagnostic pathological criteria for CAA were used (Skrobot et al., 2016). The CAA+ exhibited moderate to high Aβ pathology but was not diagnosed as AD due to the restricted tau and Aβ plaque pathology (Braak & Braak, 1991;Montine et al., 2012;Thal et al., 2002). The demographic and neuropathological classifications are shown in

MALDI-MSI, LCO staining, and laser microdissection
Postmortem brain tissue was frozen on −80°C brass plates and stored at −80°C. For MALDI-MSI (and LCO analysis) 12μm-thick fresh frozen sections were cut on a cryostat microtome (Leica CM 1520, Leica Biosystems) at −18°C, thaw mounted on conductive indium tin oxide (ITO) glass slides. For LMPC, 12μm-thick fresh frozen sections were cut on and mounted on 0.17 PEN membrane slides. All tissues were stored at −80°C.

| Preparationoffixedtissue,and immunohistochemical (IHC) staining
One brain hemisphere was fixed in formalin and embedded in paraffin, according to QSBB standard procedures. For IHC validation of ABri and ADan peptide expression in CAA plaques, 8μm-thick sections were deparaffinized and rehydrated using xylene and graded ethanol, respectively, as described previously .
Endogenous peroxidase activity was blocked by the addition of 0.3% H 2 O 2 in methanol for 10 min. Tissue sections were pre-treated in 100% formic acid (FA) for 10 min, washed, and further treated in citrate buffer (pH 6.0) for 10 min in a pressure cooker. Non-specific binding was blocked with 10% dried milk solution. Incubation with the primary antibody (anti-Aβ, epitope amino acids 8-17, DAKO, 6E10 antibody (1:500, ABri and ADan)) was performed for 1 h at TA B L E 1 Patient chart summarizing the demographics and diagnostic scores of the familial British dementia (FBD), familial Danish dementia (FDD), and two sporadic cerebral amyloid angiopathy (CAA) used in the study

| HyperspectralimagingofLCO-stainedvessels
The hyperspectral imaging of double LCO-stained tissues was performed using an inverted laser scanning confocal microscope (LSM780, Zeiss), equipped with a 32-Channel GaAsP spectral detector, in parallel spectral detection design, enabling simultaneous 32-channel spectral readout in lambda mode. The acquisition was performed using a 35nW, 458 nm Argon-laser, and a Plan-Apochromat 20x /0.8 objective lens. The continuous emission was acquired in the range from 405 to 750 nm. First, an overview image of the entire tissue was acquired at low resolution (900nm).
Thereafter, areas containing individual vessels were sequentially acquired at larger spatial resolution (300nm).

| Extractionofspectralsignaturesfrom hyperspectral light microscopy datasets
To automatically extract the spectra from the recorded images, we implemented our own image analysis algorithm in MATLAB using MALDI-MSI acquisition was performed at 10 μm spatial resolution in a high-speed MALDI-TOF/TOF instrument (rapifleX, Bruker Daltonics). The MALDI source is equipped with a scanning Smartbeam 3D laser featuring a laser beam diameter of 5 μm.
Spectra were acquired using custom laser settings with a resulting field size of 10 μm. The measurements were performed with the laser operating at a frequency of 10 000 Hz, and 100 shots per pixel.
Acquisition and subsequent processing were performed using the instrument software FlexImaging 5.0 (Bruker Daltonics).

| MALDI-MSIdataprocessingand statistical analysis
MSI data analysis was performed in SciLS Lab (v. 2019, Bruker Daltonic). The MALDI imaging data were total ion current (TIC) normalized and cluster analysis-based spatial segmentation (bisecting kmeans) were used to identify characteristic peptide distributions and for region of interest (ROI) annotation. CAA ROIs were exported as *.csv. This was followed by binning analysis for data reduction. Here, all ROI data were imported into Origin (v. 8.1 OriginLab) and peaks and peak widths were detected on the average spectra of each ROI using the implemented peak analyzer function. Bin borders for peak integration were exported as tab-delimited text files and were used for area under curve peak integration within each bin (peak-bin) of all individual ROI average spectra using an in-house developed R script.

| Tandemmassspectrometry-basedAβ, Adan,andAbrisequenceverification
The identity of the ABri, ADan, and Aβ peptides in the LMPC isolated extracts from FBD, FDD, and CAA+ subjects was verified through LC-MS/MS analysis as previously described with few modifications (Michno et al., 2021). Briefly, the analysis with an alkaline mobile phase was carried out using a Q Exactive quadrupole-Orbitrap hybrid mass spectrometer equipped with a heated electrospray ionization source (HESI-II) (Thermo Scientific) and UltiMate 3000 binary pump, column oven, and autosampler (Thermo Scientific). The Q Exactive was operated in data-dependent mode. The resolution settings were 70.000 and target values were 1 × 10 6 both for MS and MS/MS acquisitions. Acquisitions were performed with 1 μscan/acquisition. Precursor isolation width was 3 m/z units, and ions were fragmented using higher-energy collision-induced dissociation at a normalized collision energy of 25.

| LC-MS/MSdataprocessingandanalysis
Processing of raw LC-MS/MS data obtained for Aβ peptide verification, was performed using Xcalibur 2.2 Quanbrowser (Thermo Scientific). Spectra were deconvoluted using Mascot Distiller before submission to database search using the Mascot search engine (both Matrix Science) as described previously (Pannee et al., 2016). The MS/MS spectra were searched toward the SwissProt database containing the mutant human APP sequences using the following search parameters: taxonomy; Homo sapiens, precursor mass ± 15 ppm; fragment mass ± 0.05 Da; no enzyme; no fixed modifications; variable modifications including deamidated (NQ), Glu-> pyro-Glu (Nterm E), oxidation (M); disulfide bonds (C-C) instrument default. Only peptides with ion score of around 100 were considered.

| CAAacrossdifferentdementiasexhibits structural differences in amyloid aggregation
Higher order amyloid assemblies, such as Aβ in AD, display a high degree of conformational variation, both in smaller assemblies such as Aβ oligomers, all the way to Aβ fibrils or plaques (Fändrich et al., 2018;Rasmussen et al., 2017;Tycko, 2015). Only recently, have these polymorphic features gained attention and gradually been identified to originate from the differences in folding of individual peptides and intermediate assembly structures (e.g., oligomers or protofibrils) into fibrils and later plaques (Tywoniuk et al., 2018). One of the means for characterization of such conformational polymorphism, the electrooptically active luminescent conjugated oligothiophene probes (LCO) have been used for instance to demonstrate age-dependent changes in conformational polymorphism within individual plaques (Nystrom et al., 2013), conformation-specific properties of prions (Sigurdson et al., 2007), as well as variability in Aβ-amyloid aggregate structures between plaques of AD subtypes (e.g., fAD, sAD [Rasmussen et al., 2017]). Recently, the application of LCO together with mass spectrometry analysis has allowed the identification of peptide differences between diffuse and cored Aβ plaques sporadic AD, as well as cognitively unimpaired amyloid-positive subjects (Michno et al., 2019).
To understand whether and how amyloid polymorphism (Aβ, Adan, Abri) might differ between CAA in FBD, FDD, and CAA+, we investigated the gross structural features of the vascular amyloids in two brain regions (frontal cortex (FC) and occipital cortex (OC)).
All the cases displayed the highest grade of the CAA pathology with prominent vascular amyloid deposits as demonstrated in the IHC staining (Figure 1a-d). We, therefore, proceeded with LCO double staining using q-FTAA and h-FTAA fluorescent probes that have previously been reported to recognize more aggregated (q-FTAA) or less high-order, immature amyloid aggregates (h-FTAA).
In order to facilitate unbiased vessel selection and precise spectra extraction, we developed an inhouse spectra extraction tool based on maximum intensity projection of hyperspectral data (see Methods). This allowed us to annotate individual vessels in an automatic fashion (Figure 1e-h), end extract average emission spectra from these vessels (we analyzed 50-100 vessels per area for each of the patients).
Visual analysis of the average LCO emission spectra from each subject and brain region suggested the presence of distinct local maxi at 500 and 580 nm, respectively, in these spectra ( Figure 1g).
We therefore quantified the relative ratio of the 500/580 nm signal for individual patients. Indeed, this analysis revealed a higher 500/580 ratio in the sporadic CAA cases as compared to the FBD and FDD subjects ( Figure 1h). To our surprise, we did not observe any larger spectral differences between the two forms of BRI2 mutations.
Thus, our results suggest that the vascular amyloid pathology in sporadic CAA cases consists of potentially more aggregated fibrillary structures. These are characterized by the formation of more mature fibrils that are preferentially stained by q-FTAA.

| FamilialBritishdementiaCAAisdominated by pyroglutamated, full-length ABri1pE-34
To retrieve novel chemical information on the Aβ, ABri, and ADan composition of these vascular deposits, we used unbiased high dimensionality peptide analysis using MALDI MSI on adjacent sections. In order to facilitate detailed characterization of the identified peptide, we performed laser microdissection pressure catapulting non-homogenous distribution of the peptides across the vessels, that was however largely similar between each of the ABri peptides ( Figure 2cIII,IV, Figure S2c).
To compare the peptide isoforms between the brain regions, we extracted and quantified the signal intensity of each of the peptides from the individual vessels (50-100 vessels per area). We expressed the data relatively to the full-length ABri1-34 peptide. This allowed subsequent comparison of relative N-terminal and C-terminal processing of the peptide between brain regions and more generally Bri2 derived peptides across the different disease cases (i.e. FBD vs. FDD). Indeed, as observed from the average MALDI-MSI spectra, the vascular lesions were dominated by ABri1pE-34 and ABri1pE-29 followed by ABri3-34 (Figure 2d).
To gain a holistic perspective on the ABri peptide subtypes, we divided and grouped the peptide subtypes into four subgroups: full-length, C-terminally truncated, N-terminally truncated, and pyroglutamate-modified (Figure 2e). The comparison of the average ABri signal from the occipital and frontal cortex displayed minor overall differences (1-4% depending on subgroup) ( Figure S2e).
Overall, this grouping showed us that the majority of the ABri peptides in the FBD case were N-terminally truncated, and largely pyroglutamate-modified.
Together, our results demonstrate a prominent pyroglutamate modification of the ABri peptides in the FBD subject and only a limited diversity in the ABri peptide subtypes. Consistent with previous studies, we did not observe any Aβ peptides in the FBD case F I G U R E 2 MALDI MSI of familial British dementia (FBD). (a) Average mass spectra of vessels from occipital cortex(blue) and frontal cortex(red) in FBD patient. (b) List of peptides identified through LC-MS-based analysis of laser micro-dissected vessels. FBD patient displayed prominent N-terminal pyroglutamate formation (orange letters), as well as the presence of disulfide bridges between the two cysteine residues in the ABri sequence (red letters). (c.I) spatial segmentation by k-means clustering allowed for the identification of amyloidpositive vessel areas which corresponded well with vessels as (c.II) identified by LCO staining. (c.III-VI) single ion images of ABri peptides that contributed to the clustering. (d) Bar plot representing signal intensity of each of the peptides, relatively to the full-length, unmodified ABri1-34 peptide, extracted from 50 to 100 individual vessels in both brain regions. Error bars indicate SD. (e) Fractional content of fulllength, C-terminally truncated, N-terminally truncated, and pyroglutamate modified isoforms among detected peptides in FBD patient analyzed. We further provide novel insight into ABri pathology by demonstrating the presence and location of disulfide bridges disulfide bridges in the ABri peptide sequences.

| FDDischaracterizedbythepresenceof ADan and Aβ peptides
MALDI-MSI of the FDD subject revealed much more complex peptide profile than that of the FBD. Just as for the FBD the visual inspection of the overview spectra ( Figure 3a) suggested a no bigger difference between the brain regions ( Figure S3a Figure S3e).
Here, we considered the peptides based on the different precursor proteins (APP vs. BRI2) and compared the overall amyloid (all amyloid), Aβ and ADan separately. Again, we did observe only minor overall differences between occipital and frontal cortex (between 1-5% depending on subgroup). Interestingly, the overall average truncation patterns were different between Aβ and ADan.
Because of the differences in localization and area coverage of Overall, our results confirm the presence of Aβ and ADan within the amyloid-affected vasculature in the FDD patient. Importantly, the relative amount of Aβ as compared to ADan is small. The spatial distribution of Aβ and ADan does differ, and there is only a limited spatial correlation between these amyloidogenic peptides. The ADan peptides appear to a larger extent pyroglutamate modified as compared to the Aβ peptides.

| VascularlesionsinsporadicCAAare dominated by Aβ x-40
Probing the two sporadic CAA subjects with MALDI-MSI revealed an Aβx-40-dominated pathology. Again, visual inspection of average spectra from these cases did not suggest any bigger difference between the brain regions in either of the cases (Figure 4a, Figure S4a Figure S4b). This allowed to confirm that Aβ1-40 is indeed the most prominent peptide in both sporadic CAA+ cases.
Here, we observed even smaller differences between different brain regions for each of the peptides, as compared to FBD and FDD. Interestingly, while one of the subjects (CAA1) displayed a prominent signal of both Aβ3pE-40 and Aβ4-40, the second subject (CAA2) had much lower amount of Aβ4-40 present in their vascular amyloid lesions. The first CAA+ case (CAA1) displayed also much larger overall signal of other peptide species as compared to the second subject, with the exception of the Aβ1-37 peptide which was much higher in the CAA2 case ( Figure 4E, Figure S4f).
Given the relative differences in peptide patterns between the two sporadic CAA cases (CAA+), we again grouped the quantified data from individual vessels into the four peptide subgroups: Fulllength, C-terminally truncated, N-terminally truncated, and pyroglutamate modified (Figure 4f). This allowed us to gain a holistic view of the data and revealed that only 1-2% of all the peptides in either of the subjects were full-length Aβ (i.e., Aβ1-42). On the other hand, over 90% of all detected Aβ species in each CAA+ subjects were Overall, we observed a variety of differentially truncated amyloid peptides in single CAA across different brain regions. We further observed co-aggregation of Bri2-derived amyloid peptides (ADan) with Aβ peptides in FDD but not FBD, which agrees with previous studies that are based on IHC and brain extract analysis (Holton et al., 2002;Saul et al., 2013;Tomidokoro et al., 2005). Further, compared to previous studies that relied on either IHC, western or approximate mass spectral annotation, our analysis also demonstrated the presence of disulfide bridges between the two-cysteine amino acids in the ABri/ ADan sequences. Further, we were also able to observe mainly two ABri1-29 (~37.9%)), the opposite is the case in terms of ADan peptides (hydrophobicity of ADan1-34 (~41.2%), ADan1-33 (~42.2%), ADan1-29 (~48.3%), ADan1-28 (~50%)). Even without C-terminal processing, full-length ADan is estimated to be more hydrophobic than the full-length ABri peptide.
These differences in the hydrophobicity, known to be crucial components driving the aggregation of these peptides, could be the source of co-aggregation of Aβ as observed in FDD, and the lack of such co-aggregation in FBD. If this was the case, one would expect a large degree of co-localization of the Aβ peptides with the ADan peptides in the FDD subject. Indeed, single pixel analysis of the co-localization between Aβ and ADan peptides showed a limited correlation between these two peptide types with the strongest correlation for Aβ3-40 and Aβ3pE-40 but not Aβx-42. This is of interest as Aβx-42 species have previously been implicated in FDD amyloid pathology rather than Aβx-40 (Holton, 2002. Moreover, 3pE-40 deposition suggests the Aβx-40 derived species to be a potential interaction partner in ADan aggregation and CAA formation in FDD. Of note, pyroglutamation of Aβ 3-x species has previously been shown to promote Aβ aggregation because of the hydrophobic functionalization of the Aβ species (Michno et al., 2019).
Both C-terminal truncation and thereby increased hydrophobicity of ADan peptide as well as N-terminal functionalization of Aβ x-40 (Aβ 3-40 and Aβ 3pE-40) could hence indicate a rationale for ADan/ Aβ co-deposition, Mechanistic implication of these data to elucidate whether either truncation is necessary for co-aggregation, by acting as a 'seed', is an interesting objective for in vitro co-aggregation experiments I follow-up studies.
In general, both in FDD and in CAA+, we observed an interesting pathological phenomenon, where the Aβx-40 and Aβx-42 peptides showed different localization from one another. Previous studies have reported a large degree of Aβx-40 deposition in the cerebral vasculature of CAA+ patients, but less focus has been given to the Aβx-42 peptides. Given that it is well accepted that Aβ1-40 is less aggregation prone and is less hydrophobic in nature (hydrophobicity of Aβ1-40 (~42.5%) and of Aβ1-42 (~45.2%)), it is rather natural to speculate that Aβx-42 is necessary to act as a seed for this process.
Indeed, in case of parenchymal plaque pathology, this has been repeatedly the case (and has recently also been demonstrated with help of heavy isotope spatio-temporal tracing of both Aβ1-42 and shorter peptides (Michno et al., 2021)). However, in case of CAA plaques, when considering the distinct localization or Aβx-40 and Aβx-42 in both FDD and CAA+ cases, there could exist an independent aggregation mechanism for these two peptide isoforms, possibly specific to the vasculature. One possible source of these differences could be the origin of Aβx-40 and Aβx-42. The former is more freely circulating and depositing gradually on the endothelial cell wall in the vasculature. The latter is aggregating directly in the extracellular space. Here, the rather uniform nature of the Aβ composition between brain regions as well as in relation to ADan in FDD, further support that vascular Aβx40 and Aβx-42 aggregation are independent of each other. Further, CAA is associated with vascular Aβ clearance (Weller, 1998;Weller et al., 1998), which can further explain prefered deposition of more freely. circulating Aβx-40 species along with passive and independent deposition of Aβx-42 species.
Noteworthy is also the distribution pattern of the N-terminal truncations of Aβx-40 and Aβx-42. In detail, N-terminally truncated isoforms of the respective full-length peptide (Aβ1-40 and Aβ1-42) showed a similar pattern of deposition as their full-length equivalents. This further supports the notion that these isoforms arise from separate origins and not from sequential cleavage of full-length Aβ1-42. This further suggests that the N-terminal variation do originate from processing that takes place a at the site of deposition.
Lastly, we analyzed the broad conformational state of the aggregates present in the FBD, FDD, and CAA+ cases using LCO microscopy. Although the quantification was performed on the average signal from individual vessels, rather than single pixels, the analysis still revealed relatively a higher aggregation state of the Aβ vessels in CAA+ as compared to vessels with ABri or ADan peptide deposition.
Such observations were expected as recent structural studies based on cryo-EM and solid-state NMR studies have demonstrated differences in folding polymorphism both in between aggregates consisting of different peptides (e.g. Aβ1-40 vs. Aβ1-42), but also between aggregates consiting of the same peptide (Aβ1-42) (Tycko, 2015).
Although the hydrophobicity of individual ADan peptides might be higher than that of Aβ peptides, the relation between peptide hydrophobicity and aggregation state/maturity of larger aggregates are likely not directly connected. Indeed, previous studies have demonstrated that the Aβ1-40 fibrils are over 50 times less elastic than the Aβ1-42 fibrils (Dong et al. 2016). This can be attributed to different β-sheet organization within each fibrillary layer of mature Aβ fibrils. Therefore, just as for Aβ1-40 and Aβ1-42 (with Aβ1-40 being less hydrophobic but the fibrils being denser), the relatively hydrophobicity of ADan, as compared to Aβ peptides does not directly render it a denser conformational organization of larger aggregates. We would like to emphasize again, that a limitation of the current study is the small number of patients included, making it a descriptive study in nature. This is, however attributed to the rarity of cases available.
Further, we chose to compare FDD to CAA+ cases and not AD, Finally, a potential source of variation includes differences in ionization/desorption of diverse amyloid peptides could be a possible concern influencing the comparison of the amyloid truncations. This issue is to some extent accounted for by performing relative quantification compared to the largest peptide peak. Further, the trends observed in our data are in agreement with previous literature for CAA (Gkanatsiou et al., 2019) and in addition, a linear behavior in MALDI-TOF analysis has been previously described for the main Aβ peptides (Michno et al., 2018).

| CON CLUS IONS
In summary, in this work, we developed novel tools for vascular amyloid analysis in postmortem human brain tissue. In difference to previous approaches based on brain extracts or indirect antibody staining, the use of mass spectrometry imaging enabled us to visualize a large diversity in ABri, ADan, and Aβ peptide trunca-

ACK N OWLED G EM ENTS
We thank Prof Per Hammarström and Prof Peter Nilsson at Linköping University for kindly providing the LCOs. We thank the staff at the Centre for Cellular Imaging (CCI), Core Facilities, The Sahlgrenska Academy, University of Gothenburg, and the National Microscopy Infrastructure, NMI (VR-RFI 2019-00022), for help with development of imaging paradigm and microscopy expertise.

DATAAVA I L A B I L I T YS TAT E M E N T
All data will be provided upon request.