Retinal pathological features and proteome signatures of Alzheimer’s disease

Alzheimer’s disease (AD) pathologies were discovered in the accessible neurosensory retina. However, their exact nature and topographical distribution, particularly in the early stages of functional impairment, and how they relate to disease progression in the brain remain largely unknown. To better understand the pathological features of AD in the retina, we conducted an extensive histopathological and biochemical investigation of postmortem retina and brain tissues from 86 human donors. Quantitative examination of superior and inferior temporal retinas from mild cognitive impairment (MCI) and AD patients compared to those with normal cognition (NC) revealed significant increases in amyloid β-protein (Aβ42) forms and novel intraneuronal Aβ oligomers (AβOi), which were closely associated with exacerbated retinal macrogliosis, microgliosis, and tissue atrophy. These pathologies were unevenly distributed across retinal layers and geometrical areas, with the inner layers and peripheral subregions exhibiting most pronounced accumulations in the MCI and AD versus NC retinas. While microgliosis was increased in the retina of these patients, the proportion of microglial cells engaging in Aβ uptake was reduced. Female AD patients exhibited higher levels of retinal microgliosis than males. Notably, retinal Aβ42, S100 calcium-binding protein B+ macrogliosis, and atrophy correlated with severity of brain Aβ pathology, tauopathy, and atrophy, and most retinal pathologies reflected Braak staging. All retinal biomarkers correlated with the cognitive scores, with retinal Aβ42, far-peripheral AβOi and microgliosis displaying the strongest correlations. Proteomic analysis of AD retinas revealed activation of specific inflammatory and neurodegenerative processes and inhibition of oxidative phosphorylation/mitochondrial, and photoreceptor-related pathways. This study identifies and maps retinopathy in MCI and AD patients, demonstrating the quantitative relationship with brain pathology and cognition, and may lead to reliable retinal biomarkers for noninvasive retinal screening and monitoring of AD. Supplementary Information The online version contains supplementary material available at 10.1007/s00401-023-02548-2.

. Spatial distribution of Ab42 burden in retinal subregions and correlations with various brain regions and cognition.
Stereological assessment of ST and IT retinal Aβ42 (12F4 + -IR area) in age-and sex-matched patients with NC (n=15), MCI (n = 7, 8), or AD (n = 17, 18) within predetermined geometric subregions: a. Quantitative 12F4 + -IR from central (C), mid-(M), and far peripheral (F) retinal subregions. b. Retinal 12F4 + -IR area in ST versus IT regions in a subset of donors with NC (n=11), MCI (n=4,5) or AD (n=6). c. Quantitative inner retinal Aβ42-IR area extending from the inner limiting membrane to the inner nuclear layer per C/M/F subregions. d. Quantitative outer retinal 12F4 + -IR area extending from the outer plexiform layer to the outer limiting membrane per C/M/F subregions in the same cohort. Pd-diagnostic groups; Pr-regions; Pi-interactions. e-g. Pearson's correlation coefficient (r) analysis of retinal 12F4 + -IR area plotted against brain Aβ plaque burden in (e) the superior temporal (Temp) and the primary visual (PV) cortices, (f) the superior parietal cortex and hippocampus, and (g) the superior frontal and visual association (VA) cortices. h. Correlation analyses of inner or outer retinal Aβ42 burden versus Aβ plaque in the entorhinal cortex (EC). i. Heatmaps summarizing Pearson's correlation coefficient (r) analyses between retinal Aβ42 and the cerebral Aβ neuritic plaque (NP) score in all analyzed regions of the brain and retina (n=25); pseudo-color grading illustrates the strength and significance of 'r' (red) and 'P' (blue) values, respectively. j-l. Pearson's correlation coefficient (r) analysis of (j) the retinal 12F4 + -IR area in the mid-and far peripheral subregions plotted against Aβ plaque burden in the EC, (k) the inner retinal 12F4 + -IR area plotted against Aβ plaque burden in the brain, and (l) the retinal 12F4 + -IR area plotted against the cerebral ABC score derived from a 4-point scale based on the following brain pathologies: amyloid plaque -A, NFT by Braak -B, and NP by CERAD -C. m. Retinal Aβ42 burden in patients stratified based on high (H) or low (L) brain ATN histopathology (severity scores) and color-labeled per diagnostic group (n=28); the extrapolated dotted-gray line marks the retinal Aβ42 level separating ATN H from ATN L individuals. Individual data points are presented with group means ± SEMs. n. Pearson's correlations between retinal Aβ42 burden per geometric subregion and MMSE cognitive scores. ST = superior-temporal, IT = inferior-temporal region of retina. Violin plots show median and lower and upper quartiles, and individual data points are presented. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, by one-or two-way ANOVA and Tukey's post-hoc multiple comparison test. Red and blue asterisks mark significance relative to NC and MCI, respectively. Filled and empty circles represent women and men, respectively.
Supplementary Fig. 5. Extended data on the intracellular scFvA13 + -Ab oligomers (AbOi) distribution in the retinas of MCI and AD patients compared to retinas of NC controls and correlations with cerebral pathology and cognition. a-b. Pearson's correlation coefficient (r) analysis of retinal scFvA13 + (AβOi)-IR versus (a) age and (b) PMI. c. Bar graph displays the retinal scFvA13-burden separated by sex in AD (n=12) and MCI (n=8) patients vs. NC controls (n=13). Pd-diagnostic groups; Ps-females vs. males (sex groups); Pi-interactions. df. Mapping of retinal scFvA13 + in NC, MCI, and AD patient cohorts (n = 31-33) for (d) total ST/IT retina and (e) total IT, per C/ M/ F subregions, and (f) ST versus IT regions. Pd-diagnostic groups; Pr-regions; Pi-interactions. g-k. Pearson's correlation coefficient (r) analyses of retinal AβOi plotted against (g) cerebral Aβ plaque burden for all groups, (h) cerebral neurofibrillary tangle (NFT) pathology for all groups, (i) cerebral neuropil threads (NT) for all groups, (j) brain atrophy for all groups, and (k) NC and MCI groups. l. Analysis of the scFvA13 + -AβOi area in human retinas stratified based on high (H) or low (L) brain ATN histopathology and color-labeled per diagnostic group (n=22); extrapolated dotted-gray line marks the retinal AβOi level separating ATN H from ATN L individuals. m. Pearson's correlation between retinal AβOi per C/M/F subregions and MMSE cognitive scores. Median and lower and upper quartiles are indicated on each box plot, and individual data points are presented. *P < 0.05, **P < 0.01, by one-or two-way ANOVA and Tukey's post-hoc multiple comparison test. Red asterisks mark significance relative to NC. Filled and empty circles represent women and men, respectively. AbOi: intraneuronal Ab oligomers. Supplementary Fig. 6. Extended data on retinal S100b + and GFAP + macrogliosis distribution in MCI and AD patients and correlations to cerebral pathology and cognition. a, b. Pearson's correlation coefficient (r) analyses of the retinal gliosis S100β + -IR area vs. (a) age and (b) PMI. c. Quantitative retinal S100β + -IR area in males compared to females in NC (n = 5) vs. MCI/AD groups (n = 10-12). d, e Pearson's correlation coefficient (r) analyses of the retinal gliosis GFAP + -IR area vs. (d) age and (e) PMI. f. Quantitative retinal GFAP + -IR area in males compared to females in NC (n = 16), MCI (n = 9) and AD (n = 17) groups. g-i. Mapping of the retinal GFAP + gliosis-IR area in (g) the inner retinal layer from C/M/F subregions, (h) the outer retinal layer from C/M/F subregions, and (i) the ST vs. IT retina in the same cohort. Pd-diagnostic groups; Pr-regions; Pi-interactions. j-l. Pearson's correlation coefficient (r) analyses in a subset of NC, MCI, and AD patients between (j) reactive gliosis S100β + against GFAP + , reactive gliosis (S100β + ; orange or GFAP + ; green) area against (k) brain NT, and (l) brain atrophy score. m, n. Retinal (m) S100β + -IR area and (n) GFAP + -IR area data in human donors stratified according to high (H) and low (L) brain ATN histopathology (severity scores) and color-labeled per diagnostic group (S100β + , n=12; GFAP + , n=28); extrapolated dotted-gray line signifies the level of retinal gliosis separating ATN H and ATN L individuals. o, p. Pearson's correlation coefficient (r) analyses between the retinal (o) S100β + area or (p) GFAP + area per C/M/F regions vs. MMSE score. Violin plots show median and lower and upper quartiles, and individual data points are presented. *P < 0.05, **P < 0.01 by one-or two-way ANOVA and Tukey's post-hoc multiple comparison test. Red asterisks mark significance relative to NC.
Supplementary Fig. 7. Extended data on spatial distribution of retinal IBA1 + microgliosis and correlations with retina and brain pathology and cognition.
a. Representative immunofluorescence images of retinal cross-sections from AD patients showing IBA1 + -microgliosis (red), 12F4 + -Aβ42 or 4G8 + -Aβ42 deposits (green), and nuclei (DAPI, blue). Enlarged image depicts retinal microglia engulfing Aβ42 deposits (white arrows). b, c. Pearson's correlation coefficient (r) analyses between Iba1 + -microgliosis plotted against (b) age and (c) PMI in cohorts of donors with AD, MCI,. d. Mapping of the retinal IBA1 + -IR area in the ST compared to the IT in NC controls (n=10) and MCI (n=5) and AD (n=6) patients. Pd-diagnostic groups; Pr-regions; Pi-interactions. e-h. Pearson's correlation coefficient (r) analyses of retinal IBA1 + -microgliosis in ST/IT regions plotted against (e) retinal AβOi, (f) retinal S100β + gliosis, (g) retinal brain neuropil threads (NT), and (h) brain atrophy scores. i. Analysis of the retinal IBA1 + -microgliosis area in individuals stratified according to high (H) or low (L) brain ATN histopathology and color-labeled per diagnostic group (n = 24); extrapolated dotted-gray line marks the level of retinal Iba1 +microgliosis separating ATN H and ATN L individuals. j. Pearson's correlation between retinal IBA1 microgliosis per C/M/F subregions and MMSE cognitive score. Median and lower and upper quartiles are indicated on each violin plot, and individual data points are presented. *P < 0.05, **P < 0.01 by one-or twoway ANOVA and Tukey's post-hoc multiple comparison test. Red asterisks mark significance relative to NC group. Filled and empty circles represent women and men, respectively. age, (f) PMI, (g) retinal IBA1 + -IR, (h) cerebral Aβ plaque burden per C/M/F subregions, (i) brain neurofibrillary tangles (NFTs), (j) brain NFTs per C/M/F subregions, (k) brain neuropil threads, (l) brain neuropil threads per C/M/F subregions, (m) brain atrophy and (n) brain atrophy per C/M/F subregions. o. Analysis of retinal thickness stratified by high (H) or low (L) brain ATN histopathology and color-labeled per diagnostic group (n = 21); extrapolated dotted-gray line marks the level of retinal thickness (µm) separating ATN H and ATN L individuals. p. Correlations of retinal thickness in the 3 subregions vs. MMSE cognitive scores. Violin plots show median, lower and upper quartiles, and individual data points are presented. **P < 0.01, ***P < 0.001, ****P < 0.0001 by one-or two-way ANOVA and Tukey's post-hoc multiple comparison test. Individual data points are presented with group means ± SEMs. Filled and empty circles represent women and men, respectively. Supplementary Fig. 9. Extended global proteomics data in retinas and brains from AD patients and NC controls.