Postmortem eyes and brains from human donors. Human eye and brain tissues collected from donor patients with premortem clinical diagnoses of MCI and AD dementia (and confirmed postmortem AD neuropathology), and age- and sex-matched NC controls (total n = 86 subjects) were primarily obtained from the Alzheimer’s Disease Research Center (ADRC) Neuropathology Core in the Department of Pathology (IRB protocol HS-042071) of Keck School of Medicine at the University of Southern California (USC, Los Angeles, CA). Additional eyes were obtained from the National Disease Research Interchange (NDRI, Philadelphia, PA) under approved Cedars-Sinai Medical Center IRB protocol Pro00019393. Subjects with macular degeneration, diabetes/diabetic retinopathy, and glaucoma were excluded. For a subset of patients and controls, we also obtained brain specimens from the ADRC Neuropathology Core at the University of California, Irvine (UCI [IRB protocol HS#2014 − 1526]). USC-ADRC, NDRI, and UCI-ADRC maintain human tissue collection protocols that are approved by their managerial committees and subject to oversight by the National Institutes of Health. Histological studies at Cedars-Sinai Medical Center were performed under IRB protocols Pro00053412 and Pro00019393. For histological examinations, 54 retinas were collected from deceased donors with confirmed AD (n = 24), MCI due to AD (n = 11), as well as from age- and sex-matched deceased donors with NC (n = 19). For a subset of patients, paired brain tissues were also analyzed (n = 39). For the biochemical assays (ELISA and mass spectrometry) of retinal proteins, eyes were collected from another deceased donor cohort (n = 14) comprised of clinically and neuropathologically confirmed AD patients (n = 7) and matched NC controls (n = 7). Demographic, clinical, and neuropathological information on human donors is detailed in Table 1; more data on individual human donors in Table S1. For mass spectrometry of brain proteins, fresh-frozen human brain tissue was obtained from additional donor cohort (n = 18) comprised of clinically and neuropathologically confirmed AD patients (n = 10) and matched NC controls (n = 8). Demographic, clinical, and neuropathological information on human donors is detailed in Table 2 and Table S2. Tissue allocation to histological and biochemical analyses is depicted in Fig. 1A. Patients’ identity was protected by de-identifying all tissue samples in a manner not allowing to be traced back to tissue donors.
Clinical and neuropathological assessments. ADRC provided the clinical and neuropathological reports on the patients’ neurological examinations, neuropsychological and cognitive tests, family history, and medication lists as collected in the ADRC system using the Unified Data Set (UDS)47. The NDRI provided the medical history of additional patients. Most cognitive evaluations had been performed annually and, in most cases, less than one year prior to death. Cognitive testing scores from evaluations made closest to the patient’s death were used for this analysis. Two global indicators of cognitive status were used for clinical assessment: the Clinical Dementia Rating (CDR scores: 0 = normal; 0.5 = very mild impairment; 1 = mild dementia; 2 = moderate dementia; or 3 = severe dementia)48 and the Mini-Mental State Examination (MMSE scores: normal cognition = 24–30; MCI = 20–23; moderate dementia = 10–19; or severe dementia ≤ 9)49. In this study, the composition of the clinical diagnostic group (AD, MCI, or NC) was determined by source clinicians based on findings of a comprehensive battery of tests including neurological examinations, neuropsychological evaluations, and the aforementioned cognitive tests. To obtain a final diagnosis based on the neuropathological reports, we used the modified Consortium to Establish a Registry for Alzheimer's Disease (CERAD)50–51, as outlined in the National Institute on Aging (NIA)/Regan protocols with revision by the NIA and Alzheimer’s Association52. The Aβ burden (measured as diffuse, immature, or mature plaques), amyloid angiopathy, neuritic plaques, neurofibrillary tangles (NFTs), neuropil threads (NTs), granulovacuolar degeneration, Lewy bodies, Hirano bodies, Pick bodies, balloon cells, neuronal loss, microvascular changes, and gliosis pathology were assessed in multiple brain areas, specifically in the hippocampus (particularly CA1, at the level of lateral geniculate body of the thalamus), entorhinal cortex, superior frontal gyrus of the frontal lobe, superior temporal gyrus of the temporal lobe, superior parietal lobule of the parietal lobe, primary visual cortex (Brodmann Area-17), and visual association (Area-18) of the occipital lobe. All cases underwent uniform brain sampling by a neuropathologist.
Cerebral amyloid plaques and neurofibrillary tangles were evaluated using anti–β-amyloid mAb clone 4G8, Thioflavin-S (ThioS), and Gallyas silver stain in formalin-fixed, paraffin-embedded tissues. Two neuropathologists provided scores based on independent observations of β-amyloid, NFT burden, and/or neuropil threads (0 = none; 1 = sparse 0–5; 3 = moderate 6–20; 5 = abundant/frequent 21–30 or greater; or N/A = not applicable), and an average of 2 readings was assigned to each individual patient.
A final diagnosis included AD neuropathological change using an “ABC” score derived from 3 separate 4-point scales. We used the modified Aβ plaque Thal score (A0 = no Aβ or amyloid plaques; A1 = Thal phase 1 or 2; A2 = Thal phase 3; or A3 = Thal phase 4 or 5)53. For the NFT stage, the modified Braak staging for silver-based histochemistry or p-tau IHC was used (B0 = no NFTs; B1 = Braak stage I or II; B2 = Braak stage III or IV; or B3 = Braak stage V or VI)54. For the neuritic plaques, we used the modified CERAD score (C0 = no neuritic plaques; C1 = CERAD score sparse; C2 = CERAD score moderate; or C3 = CERAD score frequent)50.
Neuronal loss, gliosis, granulovacuolar degeneration, Hirano bodies, Lewy bodies, Pick bodies, and balloon cells were all evaluated (0 = absent or 1 = present) in multiple brain areas by staining tissues with hematoxylin and eosin (H&E). Brain atrophy was evaluated (0 = none; 1 = mild; 3 = moderate; 5 = severe; or 9 = not applicable). We classified the brain ATN histopathology scores based on high (H) and low (L) severity of brain Aβ deposition (A; cutoff score = 2.0), tauopathy (T; cutoff score = 1.1), and neurodegeneration (N; cutoff score = 0.4), which was modified from the latest NIA-AA’s in vivo ATN biomarker classification scheme for AD55. Our ATNHigh and ATNlow were determined according to a combined cutoff score of 3.5; the terms were used to refer to humans with high and low histopathology levels of brain amyloid, tauopathy, and neuronal loss, respectively.
Processing of eye and brain tissues. Donor eyes were collected within an average of 7.5 hours after time of death and were 1) preserved in Optisol-GS media (Bausch & Lomb, 50006-OPT) and stored at 4°C for less than 24 hours; 2) fresh frozen (snap frozen; stored at -80°C); or 3) punctured once and fixed in 10% neutral buffered formalin (NBF) or 4% paraformaldehyde (PFA) and stored at 4°C. In addition, fresh brain tissues (hippocampus, occipital lobe–primary visual cortex [Brodmann area 17], and frontal cortex [area 9]) from the same donors were snap frozen and stored at -80°C. Portions of fresh-frozen brain tissues were fixed in 4% PFA for 16 hours following dehydration in 30% sucrose/PBS. Brain tissues were sectioned (30-µm thick) on a cryostat and placed in phosphate-buffered saline (PBS) with 0.01% sodium azide (Sigma-Aldrich) at 4°C. Regardless of the source of the human donor eye (USC-ADRC or NDRI), the same tissue collection and processing methods were applied.
Preparation of retinal strips. Eyes that were fixed in 10% NBF or 4% PFA were dissected to create eyecups. The complete neurosensory retinas were isolated, detached from the choroid and sclera, flat mounts were prepared, and vitreous humor thoroughly removed manually, as previously described18. Alternatively, fresh-frozen eyes and eyes preserved in Optisol-GS were dissected, with the anterior chambers removed to create eyecups. Vitreous humor liquid was allowed to flow out and the complete neurosensory retina was isolated. Next, the vitreous gel was further thoroughly removed. Fresh-frozen retina was isolated in cold PBS with 1⋅ Protease Inhibitor cocktail set I (Calbiochem 539131). For all flat mount retinas, the 4 topographical quadrants were defined by identifying the macula, optic disc (OD), and blood vessels30. Flat mount strips (~ 2mm wide) were prepared from 4 predefined geometric subregions: superior-temporal (ST), inferior-temporal (IT), inferior-nasal (IN), and superior-nasal (SN) retina, spanning diagonally from the OD to the ora serrata (Fig. 1A). Fixed retinal strips were processed for cross-sectioning. Fresh retinal strips (~ 5mm wide) were prepared and stored at -80°C for protein analysis. In a subset of freshly isolated donor eyes, additional ~ 2mm-wide strips were dissected and fixed in 4% PFA for processing of retinal cross-sections. Each strip measured approximately 2-2.5cm from the optic disc to the ora serrata, and the central (C), mid-(M), and far (F) peripheral subregions were further defined based on their radial distance from the OD (Fig. 1A). This sample preparation technique allowed for extensive and consistent access to retinal quadrants, layers, and pathological subregions.
Paraffin-embedded retinal cross-sections. Flat mount-derived strips were initially paraffinized using the standard techniques. Next, strips were embedded in paraffin after flip-rotating 90° horizontally. The retinal strips were sectioned (7–10µm thick) and placed on microscope slides treated with 3-aminopropyltriethoxysilane (APES, Sigma A3648). Before immunohistochemistry, the sections were deparaffinized with 100% xylene twice (10 min each), rehydrated with decreasing concentrations of ethanol (100–70%), and washed with distilled water followed by PBS.
Immunohistochemistry. Following deparaffinization, brain sections and retinal cross-sections were treated with target retrieval solution (pH 6.1; S1699, DAKO) at 99°C for 1 hour and washed with PBS. The brain sections and retinal cross-sections were then treated with formic acid 70% (ACROS) for 10 min at room temperature (RT) before staining for Aβ burden. Peroxidase-based and fluorescence-based immunostaining was performed. A list of antibodies and their working dilutions are shown in Table S3.
Prior to peroxidase-based immunostaining, tissues were treated with 3% H2O2 for 12 min, and two peroxidase-based staining protocols were followed. First, we used a Vectastain Elite ABC HRP kit (Vector, PK-6102, Peroxidase Mouse IgG) according to the manufacturer’s instructions; second, we used a Dako reagents protocol. Following treatment with formic acid, the tissues were washed with wash buffer (Dako S3006) and adding 0.2% Triton X-100 (Sigma, T8787) for 1 hour and then treated with H2O2 and rinsed with wash buffer. Primary antibody (Ab) was diluted with background-reducing components (Dako S3022) and incubated with the tissues for 1 hour at 37°C for JRF/cAβ42/26 #8151 (Aβ42) or JRF/Aβtot/17 Pur 117–120 (N-terminal region of Aβ) antibodies, or overnight at 4°C for 12F4 (Aβ42) mAb. Tissues were rinsed twice with wash buffer on a shaker, incubated for 30 minutes at 37°C with secondary Ab (goat anti-mouse Ab, HRP conjugated, DAKO Envision K4000), and rinsed again with wash buffer. For both protocols, 3,3-diaminobenzidine (DAB) substrate was used (DAKO K3468). Hematoxylin counterstaining was performed followed by mounting with Paramount aqueous mounting medium (Dako, S3025). Routine controls were processed using identical protocols while omitting the primary antibody to assess nonspecific labeling. Fluorescence-based immunostaining was performed by using blocking solution (DAKO X0909) and adding 0.2% Triton X-100 (Sigma, T8787). Primary Abs (list in Table S3) were incubated overnight at 4°C, followed by secondary Abs applied for 1.5 hours at RT.
For intracellular Aβ oligomer (AβOi) staining, scFvA13 was used as primary antibody56–58, which required intermediate anti-V5 tag antibody, and followed by a secondary antibody. The scFvA13 is a conformation-sensitive and sequence-specific antibody in a format of single chain Fv fragment (scFv), which selectively recognizes AD-relevant AβOs. To eliminate background autofluorescence, we treated the sections with 0.3% (w/v) Sudan Black B (199664, Sigma-Aldrich) in 70% ethanol (v/v) for 10 minutes at RT. Next, we mounted the samples using ProLongTM Gold antifade mounting media with DAPI (Thermo Fisher; #P36935). Routine controls were processed using identical protocols, omitting the primary antibody to assess nonspecific labeling.
Biochemical determination of Aβ 1−42 levels by sandwich ELISA. The neurosensory retinas were extracted from freshly isolated eye globes, dissected, and homogenized, and subsequently frozen for further analysis. In detail, fresh-frozen human retinal flat mount-derived strips from the temporal hemisphere (ST, IT) were weighed and placed in a tube (mg tissue with 10 µl buffer) with cold homogenizing buffer (100 mM TEA Bromide [Sigma, 241059], 1% sodium deoxycholate [SDC; Sigma, D6750] and 1x Protease Inhibitor cocktail set I [Calbiochem, 539131]), then homogenized by sonication (Qsonica Sonicator with an M-Tip probe, amplitude 4, 6 W, for 90 sec; the sonication pulse was stopped every 15 sec to allow the cell suspension to cool down for 10 sec). The ultrasonic probe positioned inside the sample tube was placed in ice water. After determination of protein concentration (Thermo Fisher Scientific), the amount of retinal Aβ1−42 was determined using an anti-human Aβ1−42 end-specific sandwich ELISA kit (Thermo Fisher, KHB3441).
Brain Aβ burden. ADRC neuropathological reports provided comprehensive data on cerebral Aβ plaques in different brain areas. Brain Aβ plaque severity scores were calculated from neuropathological reports based on the Aβ burden assigned to each individual, as described earlier. Brain Aβ plaque severity for each patient was calculated based on grades that were obtained after staining with Gallyas silver stain and anti-β-amyloid mAb 4G8 or ThioS. Averages of burden scores were calculated for each brain area/region separately, for total brain regions, and for neuritic plaques, NFTs, and neuropil threads.
Mapping retinal pathology. Retinal Aβ42 burden, intracellular oligomeric Aβ, gliosis, microgliosis, and retinal atrophy were determined by examining four radial cross-sections (for each experiment) from the temporal hemisphere (strips from the ST and IT regions). Following specific staining, 10 images were captured at 20⋅ objective from each retinal strip, representing the neuroretina from the optic disc to the ora serrata, including central, mid-, and far retinal areas (C, M, and F subregions, respectively). To assess retinal layer distribution, retinal pathology burden was analyzed in the inner and outer retina separately. The inner retina was anatomically defined as extending from the inner limiting membrane (ILM) to and including the inner nuclear layer (INL). The outer retina was analyzed from the outer plexiform layer (OPL) to the outer limiting membrane (OLM). Given that retinal thickness differs greatly throughout the C, M, and F subregions – thickest near the optic nerve head and thinnest toward the ora serrata in the F subregion (~ 260–300 µm and ~ 80–120 µm in elderly healthy persons, respectively) – we also normalized IR area to retinal thickness; the width was measured at three defined tissue areas per each image. Images were captured at 20⋅ or 40⋅ objective at a respective resolution of 0.5 or 0.25 µm. Images were exported to ImageJ (version 1.52o; NIH) to calculate the total area of Aβ42 burden, intracellular oligomeric Aβ, gliosis, and microgliosis.
Tissue-atrophy morphometric analysis. For morphometric analysis, 10 images were obtained (20⋅; three images from the far periphery, four from the mid-periphery, and three from the central area of the retina). Thickness measurements (µm) were manually performed using Axiovision Rel. 4.8 software. Retinal measurements were taken from the ILM to the OLM. Moreover, measurements were taken from six subregions: the nerve fiber layer (NFL), ganglion cell layer (GCL), inner plexiform layer (IPL), inner nuclear layer (INL), outer plexiform layer (OPL), and outer nuclear layer (ONL). NFL thickness was defined as extending from the ILM to the inner border of the GCL. GCL thickness was defined from the outer border of the NFL to the inner border of the IPL. IPL thickness was defined from the outer border of the GCL to the inner border of the INL. INL thickness was defined from the outer IPL to the inner border of the OPL. OPL thickness was defined from the outer INL to the inner border of the ONL. ONL thickness was defined from the outer OPL to the inner border of the photoreceptor layer, the OLM. Subregion borders are illustrated in Fig. 4A.
Severity score analysis of retinal atrophy
To better assess the magnitude of retinal tissue loss in disease, we devised a scoring system for retinal atrophy that is similar to severity scores assigned to brain atrophy in postmortem neuropathological reports. For this purpose, cross-sectional thicknesses of central, mid-, and far subregions were used to calculate the mean retinal thickness for each patient. Retinal atrophy scores were subsequently assigned according to a severity range of 0–5, separated by 0.5 intervals, with 0 assigned to the most thick and intact retina (155 µm) and 5 to the thinnest and most atrophied retina (110 µm).
Microscopy. Fluorescence and bright field images were acquired using a Carl Zeiss Axio Imager Z1 fluorescence microscope with ZEN 2.6 blue edition software (Carl Zeiss MicroImaging, Inc.) equipped with ApoTome, AxioCam MRm, and AxioCam HRc cameras. Multi-channel image acquisition was used to create images with multiple channels. Tiling mode and post-acquisition stitching were used to capture and analyze large areas. Images were repeatedly captured at the same focal planes with the same exposure time. Images were captured at 20⋅, 40⋅, 63⋅, and 100⋅ objectives for different purposes.
Quantitative immunohistochemistry. For the analyses, images were captured using the same exposure time. We randomly acquired three images from the far periphery, four from the mid-periphery, and three from the central retina for analytical purposes (as shown in Fig. 1A). Images were exported to ImageJ (version 1.52o; NIH) to analyze parameters of interest. Throughout the analysis process, the researchers were blinded to the patients' diagnosis. The fluorescence of specific signals was captured, using the same setting and exposure time for each image and human donor, with an Axio Imager Z1 microscope (with motorized Z-drive) and an AxioCam MRm monochrome camera ver. 3.0 (at a resolution of 1388 ⋅ 1040 pixels, 6.45 µm ⋅ 6.45 µm pixel size, and dynamic range of > 1:2200, which delivers low-noise images due to a Peltier-cooled sensor). Images were captured at either 20⋅ or 40⋅ objective at a respective resolution of 0.5 or 0.25 µm. Acquired images were converted to grayscale and standardized to baseline by using a histogram-based threshold in ImageJ. The images were then subjected to ImageJ particle analysis for each biomarker to determine total area of immunoreactivity (IR). For each biomarker, the IR area was determined using the same threshold percentage from the baseline in ImageJ with the same percentage threshold setting for all diagnostic groups.
Transmission electron microscopy (TEM) analysis. Fixed retinal flat mounts were washed in PBS, treated with target retrieval solution at 98°C for 1 hour (pH 6.1; Dako, #S1699), and washed with PBS. The tissues were treated with 3% H2O2 for 12 min and washed again in PBS. Peroxidase-based immunostaining was performed with primary mouse anti-human Aβ42 mAb (12F4), as described previously. To ensure that the component detected in the tissue was not derived from exogenous normal serum, the blocking step was omitted for a subset of retinal tissues. In addition, routine controls were processed using identical protocols while omitting the primary antibody to assess nonspecific labeling. Stained tissues were prepared and microdissected for electron microscopic analysis. The samples were dehydrated in a graded series of ethanol and then infiltrated in Eponate 12 resin (Ted Pella, Inc. Redding, CA, USA) prior to embedding between 2 acetate sheets. Ultrathin sections of retina were cut (vertical and en face) at a thickness of 70 nm, collected on copper grids, and lightly contrasted using Reynolds’ lead stain. The sections on grids were analyzed using a JEOL JEM-2100 LaB6 TEM at 80 kV (JEOL USA). Images were captured using the Orius SC1000B CCD camera (Gatan) and processed and colorized using Adobe Photoshop CS4 (Adobe Inc.).
Immunogold labeling transmission electron microscopy. Retinal paraffin blocks were sectioned at 7-µm thickness on a microtome and floated in a 40°C distilled water bath. The sections were transferred onto a Superfrost Plus slide, allowed to dry overnight, and stored at RT until use. Prior to staining, the slides were baked at 55°C and deparaffinized with xylene and alcohol. The slides were placed into a Coplin jar containing citrate antigen retrieval buffer and heated in a microwave until boiling. Then they were allowed to cool down to RT, rinsed in PBS, and incubated in blocking solution (2% BSA in PBS) for 30 min. Next, the sections were incubated in a moist chamber overnight at 4°C with primary mouse anti-human Aβ42 mAb (12F4) that was diluted in the 2% BSA PBS blocking solution. After rinsing with PBS, the sections were incubated with goat anti-mouse 10-nm gold conjugate secondary antibody (Ted Pella, Inc.) for one hour at 37°C before being rinsed in PBS again. Samples were fixed in ½ Karnovsky’s fix for 15 min, rinsed in 0.1 M cacodylate buffer pH 7.2, postfixed with 2% osmium for 20 min, rinsed again in cacodylate buffer pH 7.2, and finally rinsed in sodium acetate. The slides were then en bloc stained with 1% uranyl acetate for one hour and rinsed in sodium acetate. Samples were passed through a dehydration process involving 50% EtOH; 70%, 85%, 95%, and 100% EtOH; 1:1 EtOH/propylene oxide (PO) mix; 1:2 EtOH/PO mix; and pure PO; followed by PO/Eponate resin, and 100% Epon. Following this, the samples were embedded with Eponate using a beam capsule filled with Eponate and inverted onto the section. The beam capsules were removed from the slides using liquid nitrogen, sectioned ultra-thin at 70 nm, and placed onto grids. The sections were analyzed in a JEOL JEM-2100 LaB6 TEM at 80 kV (JEOL USA). Images were captured using an Orius SC1000B CCD camera (Gatan).
Proteome Analysis by Mass Spectrometry (MS):
Preparation of retinal and brain samples from NC and AD individuals. Frozen brains (from the neuropathology core of the ADRC at the University of California, Irvine) and retinas (from the ADRC Neuropathology Core in the Department of Pathology at the University of Southern California, Los Angeles, CA) were processed for MS by the University of Queensland in accordance with approval granted by the institution’s Human Research Ethics Committee (#2017000490). Frozen brain aliquots from the hippocampus, medial temporal gyrus, and cerebellum were used for the brain analysis study. Frozen tissues were transferred into a Precellys homogenization tube (Bertin Technologies), homogenized in liquid nitrogen, and lysed in ice-cold T-PER extraction buffer (Thermo Scientific) containing protease and phosphatase inhibitors. Next, tissue lysates were cleared of any debris by ultracentrifugation at 100,000g for 60 min at 4°C. Retinal tissues were carefully extracted from donor eye tissues without the vitreous and retinal pigment epithelium, and retinal temporal hemisphere (ST, IT) tissues were homogenized (100 mM TEA Bromide [Sigma, 241059], 1% SDC [Sigma, D6750], and 1 ⋅ protease inhibitor cocktail set I [Calbiochem 539131]) by sonication (Qsonica Sonicator with M-Tip probe, amplitude 4, 6 W, for 90 seconds; the sonication pulse was stopped every 15 sec to allow the cell suspension to cool down for 10 sec). Insoluble materials were removed by centrifugation at 15,000 g for 10 min at 4°C. Protein concentrations of brain and retinal lysates were determined via the Bradford assay (Bio-Rad Laboratories). Extracted brain and retinal proteins were reduced using 5-mM DTT alkylation with 10 mM iodoacetamide. Protein concentration was determined using a BCA assay kit (Pierce). Dual digestion was carried out on 150 µg protein, initially using Lys-C (Wako, Japan) at a 1:100 enzyme:protein ratio overnight at RT followed by trypsin (Promega) at a 1:100 enzyme:protein ratio overnight at 37°C.
TMT Labeling. To accommodate 14 retinal samples (7 AD and 7 NC) and 54 brain samples (10 AD and 8 NC) for each corresponding tissue (hippocampus, cerebellum, and frontal cortex), 8 separate TMT10plex experiments were performed. As described previously, 50-µg peptides from each sample were labeled with 0.8 mg TMT reagent. Labeling was carried out at RT for one hour with continuous vortexing. To quench any remaining TMT reagent and reverse the tyrosine labelling, 8 µl of 5% hydroxylamine was added to each tube, followed by vortexing and incubation for 15 min at RT. Combined samples from each TMT experiment were subjected to HpH fractionation using an Agilent 1260 HPLC system equipped with a quaternary pump, a degasser, and a multi-wavelength detector (set at 210-, 214- and 280-nm wavelengths). Peptides were separated on a 55-min linear gradient from 3 to 30% acetonitrile in 5 mM ammonia solution (pH 10.5) at a flow rate of 0.3 ml/minute on an Agilent 300 Extend C18 column (3.5-µm particles, 2.1-mm inner diameter, 150-mm length). The 96 fractions were finally consolidated into 17 fractions. Each peptide fraction was dried by vacuum centrifugation, resuspended in 1% formic acid, and desalted again using SDB-RPS (3M Empore) stage tips.
Nanoflow Liquid Chromatography Electrospray Ionization Tandem Mass Spectrometry (nano LC-ESI-MS/MS). Cleaned peptides from each fraction were analyzed using a Q Exactive Orbitrap mass spectrometer (MS; Thermo Scientific) coupled to an EASY-nLC1000 nanoflow HPLC system (Thermo Scientific). Reversed-phase chromatographic separation was performed on an in-house packed reverse-phase column (75 µm ⋅ 10 cm Halo 2.7-µm 160 Å ES-C18, Advanced Materials Technology). Labeled peptides were separated for two hours using a gradient of 1–30% solvent B (99.9% acetonitrile/0.1% formic acid) and Solvent A (97.9% water/2% acetonitrile/0.1% formic acid). The Q Exactive MS was operated in the data-dependent acquisition mode to automatically switch between full MS and MS/MS acquisition. Following the full MS scan from m/z 350–1850, MS/MS spectra were acquired at a resolution of 70,000 at m/z 400 and an automatic gain control target value of 10^6 ions. The top 10 most abundant ions were selected with a precursor isolation width of 0.7 m/z for higher-energy collisional dissociation (HCD) fragmentation. HCD-normalized collision energy was set to 35%, and fragmentation ions were detected in the Orbitrap at a resolution of 70,000. Target ions that had been selected for MS/MS were dynamically excluded for 90 sec.
Database Searching, Peptide Quantification, and Statistical Analysis. Raw data files were processed with Proteome Discoverer V2.1 software (Thermo Scientific) using Mascot (Matrix Science, UK). Data were matched against the reviewed SwissProt Homo sapiens protein database. The MS1 tolerance was set to ± 10 ppm and the MS/MS tolerance to 0.02 Da. Carbamidomethyl (C) was set as a static modification, while TMT10-plex (N-term, K), oxidation (M), deamidation (N, Q), Glu->pyro-Glu (N-term E), Gln->pyro-Glu (N-term Q), and acetylation (Protein N-Terminus) were set as dynamic modifications. The percolator algorithm was used to discriminate correct from incorrect peptide-spectrum matches and to calculate statistics including q value (FDR) and posterior error probabilities. Search results were further filtered to retain protein with an FDR of < 1%, and only master proteins assigned via the protein grouping algorithm were retained. Proteins were further analyzed using the TMTPrepPro analysis pipeline. TMTPrepPro scripts are implemented in the R programming language and are available as an R package, which was accessed through a graphic user interface provided by a local Gene Pattern server. In pairwise comparison tests, the relative quantitation of protein abundance was derived from the ratio of the TMT label S/N detected in each condition (AD vs. NC), and differentially expressed proteins (DEPs) were identified based on Student’s t-tests between AD and NC group ratios (log-transformed). The overall fold changes were calculated as geometric means of the respective ratios. Differential expression required the proteins to meet both a ratio fold change (> 1.2 for upregulated or < 0.80 for downregulated expression) and a p value cutoff (t-test p < 0.05). Information for down- and upregulation of DEPs in the human retina and brain (temporal cortex, hippocampus, and cerebellum) is listed in Tables S4-S11.
Functional Network and Computational Analysis. Differentially expressed proteins were classified according to KEGG pathways and biological processes using the Cytoscape stringApp plugin (http://apps.cytoscape.org/apps/stringapp). Significantly changed proteins were loaded into Cystoscape, and the Homo sapiens protein database in the StringDB was selected to reveal protein interactions in the context of enriched pathways. Detectable protein hierarchies, displayed as heatmaps from R. Venn diagram, were created using Venny 2.1 (https://bioinfogp.cnb.csic.es/tools/venny/). Volcano plots were created using Prism9 (GraphPad). DAVID pathway analysis was performed using the David Bioinformatic Database (https://david.ncifcrf.gov/). The pie chart of PANTHER functional classification analysis was created using http://pantherdb.org/geneListAnalysis.do. Biological functions analysis was performed using Ingenuity Pathway Analysis (IPA) by Qiagen.
For the comparison of the retinal data against the literature brain cortex proteome data59–63, proteins were regarded as DEPs if the t-test p-value was less than 0.05, up or down regulated based on the log fold change AD/control (> 0 up-regulated, < 0 down-regulated). Proteins P-values less than 0.05 were regarded as unchanged (Tables S12&S13). R Shiny tool (https://insightstats.shinyapps.io/meta-checker/) was developed and used to visualize the overlap.
Statistical Analysis. Analyses were performed using GraphPad Prism version 9.3.1. One- or two-way analysis of variance (ANOVA) were applied for comparisons between three or more groups followed by Tukey’s post-hoc multiplicity adjustment. In two-way ANOVA analyses, the Pd (diagnosis), PL (IR/OR layers), Pr (C/M/F regions), PS (sex), and/or Pi (interactions) were presented. Two-sample Student’s t-tests were used for two-group comparisons. The statistical association between 2 or more Gaussian-distributed variables was determined by Pearson’s correlation coefficient (r) test (GraphPad Prism), with Holm-Bonferroni correction for multiple analyses as required. Scatterplot graphs present the null hypothesis of pair-wise Pearson’s r with the unadjusted P-values that indicate direction and strength of the linear relationship between two variables. For multiple comparisons in groups of six retinal markers, seven brain regions, or thirty-six retinal markers and brain/cognitive parameters, Pearson’s correlation coefficient (r) determined associations between variables with Holm-Bonferroni adjusted P-values using SAS version 9.4 (SAS Institute). Multiple linear regression models were made with MMSE score as the outcome and two explanatory variables, one a marker of retinal gliosis and the other retinal amyloidosis controlled for markers of brain pathology (atrophy, Aβ plaques, or NFT-tau). The reported slopes and P-values are for the markers of retinal gliosis or retinal amyloidosis, which were the association of interest, controlling for brain pathologies. Model fit was assessed using the R-squared as all models had two explanatory variables. All tests were two sided with < 0.05 significance level. Results are expressed as the mean ± standard deviation (SD) or standard error of the mean (SEM). Degrees of significance between groups is represented as follows: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. A P-value less than 0.05 was considered significant. Data analysis was conducted with coded identifiers and analysts remained blinded to the diagnostic group until after completion of all analyses.