Divergent Age-Dependent Conformational Rearrangement within Aβ Amyloid Deposits in APP23, APPPS1, and AppNL-F Mice

Amyloid plaques composed of fibrils of misfolded Aβ peptides are pathological hallmarks of Alzheimer’s disease (AD). Aβ fibrils are polymorphic in their tertiary and quaternary molecular structures. This structural polymorphism may carry different pathologic potencies and can putatively contribute to clinical phenotypes of AD. Therefore, mapping of structural polymorphism of Aβ fibrils and structural evolution over time is valuable to understanding disease mechanisms. Here, we investigated how Aβ fibril structures in situ differ in Aβ plaque of different mouse models expressing familial mutations in the AβPP gene. We imaged frozen brains with a combination of conformation-sensitive luminescent conjugated oligothiophene (LCO) ligands and Aβ-specific antibodies. LCO fluorescence mapping revealed that mouse models APP23, APPPS1, and AppNL-F have different fibril structures within Aβ-amyloid plaques depending on the AβPP-processing genotype. Co-staining with Aβ-specific antibodies showed that individual plaques from APP23 mice expressing AβPP Swedish mutation have two distinct fibril polymorph regions of core and corona. The plaque core is predominantly composed of compact Aβ40 fibrils, and the corona region is dominated by diffusely packed Aβ40 fibrils. Conversely, the AβPP knock-in mouse AppNL-F, expressing the AβPP Iberian mutation along with Swedish mutation has tiny, cored plaques consisting mainly of compact Aβ42 fibrils, vastly different from APP23 even at elevated age up to 21 months. Age-dependent polymorph rearrangement of plaque cores observed for APP23 and APPPS1 mice >12 months, appears strongly promoted by Aβ40 and was hence minuscule in AppNL-F. These structural studies of amyloid plaques in situ can map disease-relevant fibril polymorph distributions to guide the design of diagnostic and therapeutic molecules.

and App NL-F mouse before applying the rela ve filter se ngs to the images.(B) Overview of the same plaques a er applying the filter se ng that is subop mal for these genotypes i.e., APP23, APPPS1 and App NL-F .(C) Overview of the plaques with the best rela ve filter se ng at the intensity ra o of I 500 /I 540 nm encircled with the blue box.For APP23 and APPPS1 mice lower limit of filter se ng is μ, meanwhile for App NL-F mouse is μ+1σ.The upper limit is infinity for all the mouse models.(D) Overview of the plaques with the best rela ve filter se ng at the intensity ra o of I 500 /I 588 nm encircled with the green box.The filter se ng for all the genotypes is the same as at the intensity ra o I 500 /I 540 nm.(E) The violin plots represent the pixel distribu on ra ometric analysis of the different genotypes at intensi es I 500 /I 540 nm and I 500 /I 588 nm, but herein considering all the pixels from images instead of selected ROIs.Each violin plot comprises 10 filtered images from each genotype.The dots in each violin plot represent the mean value of the corresponding genotype.The straight lines from the bo om of the violin plot represent the 1 st quar le (25% data below this line), 2 nd quar le or median (50% data below this line) and 3 rd quar le (75% data below this line) respec vely.
Figure S1: (A) Fluorescence intensity ra ometric plot from the region of interest (ROI) from plaque cores ofApp NL-F mice at different age groups calculated at fluorescence intensity at 500 nm and 540 nm.Fluorescence intensity data from 3 mice at 9 months, 1 mouse at 12 months, 1 mouse at 15 months, 1 mouse at 18 months, and 3 mice at 21 months were analysed.The error bars represent SEM.An ordinary one-way ANOVA test was performed in GraphPad Prism for sta s cal analysis, ns = non-significant.The sta s cal analysis shows there is no significant change in fluorescence intensity ra o over me in this mouse genotype.(B) Fluorescence intensity ra ometric plot for APP23 mice at different age groups calculated from the region of interest (ROI) from plaque cores at fluorescence intensity at 500 nm and 540 nm.For the ra ometric plot, fluorescence intensity data from 5 mice at 6 months, 7 mice at 12 months, 5 mice at 18 months, 5 mice at 25 months and 5 mice at 30 months of age were analysed.An ordinary one-way ANOVA test shows significant differences in fluorescence intensity among different age groups, where ** = p < 0.01 and **** = p < 0.0001.

Figure S2 :
Figure S2: A schema c overview of image filtra on using a rela ve filter se ng in RStudio.(A) Overview of the raw image.(B) Text files at different wavelengths i.e., at 500 nm, 540 nm and 588 nm is extracted from the hyperspectral microscope and used to apply rela ve filter se ng to remove unwanted high and low intensi es from the raw image.(C) Ra o matrix is calculated using the filtered text files at desired wavelengths i.e., at I 500 / I540 nnm or I 500 /I 588 nm in RStudio.(D) A heatmap is generated from the ra o matrix as an output image without unwanted signal.

Figure S3 :
Figure S3: (A) Overview of hyperspectral images of qFTAA and hFTAA stained amyloid plaques from APP23, APPPS1and App NL-F mouse before applying the rela ve filter se ngs to the images.(B) Overview of the same plaques a er applying the filter se ng that is subop mal for these genotypes i.e., APP23, APPPS1 and App NL-F .(C) Overview of the plaques with the best rela ve filter se ng at the intensity ra o of I 500 /I 540 nm encircled with the blue box.For APP23 and APPPS1 mice lower limit of filter se ng is μ, meanwhile for App NL-F mouse is μ+1σ.The upper limit is infinity for all the mouse models.(D) Overview of the plaques with the best rela ve filter se ng at the intensity ra o of I 500 /I 588 nm encircled with the green box.The filter se ng for all the genotypes is the same as at the intensity ra o I 500 /I 540 nm.(E) The violin plots represent the pixel distribu on ra ometric analysis of the different genotypes at intensi es I 500 /I 540 nm and I 500 /I 588 nm, but herein considering all the pixels from images instead of selected ROIs.Each violin plot comprises 10 filtered images from each genotype.The dots in each violin plot represent the mean value of the corresponding genotype.The straight lines from the bo om of the violin plot represent the 1 st quar le (25% data below this line), 2 nd quar le or median (50% data below this line) and 3 rd quar le (75% data below this line) respec vely.

Figure S4 :
Figure S4: (A) Individual pixel density distribu on plots of App NL-F mice at different age groups: i) at 9 months with a mean intensity ra o vale 0.183 pooled from 3 different mice ii) at 12 months with a mean intensity ra o value of 0.152 obtained from a single individual mouse iii) at 15 months with a mean intensity ra o vale 0.119 obtained from a single individual mouse iv) at 18 months with a mean intensity ra o vale 0.163 obtained from a single individual mouse and v) at 21 months with a mean intensity ra o vale 0.163 again pooled from 3 individual mice.(B) Overlay of pixel density distribu on plots of APP23, APPPS1, and App NL-F aged mice (at 18-19 months) compiled from 10 images from a single mouse from each genotype.APP23 mouse shows mean intensity ra o value of 0.396, APPPS1 mouse has mean intensity ra o value of 0.255 and App NL-F mouse has mean intensity ra o value of 0.172.

Table S2 :
A guideline to choose the values for filter se ng for mouse ssue.