IntroductionThe mitochondrial cascade hypothesis of Alzheimer’s disease (AD) posits a link between mitochondrial dysfunction and neuropathology [1]. The causal nexus, however, is obscured by the complex reciprocal relationship between mitochondrial dysfunction and AD-associated pathological molecules such as amyloid beta (Aβ) [2,3,4,5,6]. As a result, the scientific community currently remains divided on the precise role of mitochondrial crisis in AD etiology. This contention is exemplified by an ongoing debate centered on whether mitochondrial complex IV (CIV) deficits actively contribute to the pathogenesis of AD, or are merely an adaptive response to amyloidosis and neurodegeneration in an AD-related context [5, 7,8,9,10,11]. The resulting ambiguity further fuels the debate on whether mitochondrial CIV deficits are detrimental or beneficial to the risk of this neurodegenerative disorder.Mitochondrial CIV, together with other mitochondrial complexes including complexes I (CI), II (CII), and III (CIII), constitute the enzymatic components of the mitochondrial electron transport chain (ETC), converting energy sources into a proton gradient for ATP production through oxidative phosphorylation (OXPHOS) [12, 13]. In addition to its importance to mitochondrial bioenergetics, CIV also modulates cellular activities through its regulation of reactive oxygen species (ROS) production and Ca2+ signaling pathways [14,15,16]. In past decades, CIV deficits have been repeatedly identified in patients as well as cell and animal models of AD [7, 8, 10, 17,18,19,20,21,22,23]. In addition to CIV vulnerability to Aβ toxicity, emerging evidence also implicates an impact of mitochondrial CIV deficits on the development of amyloidosis. Several studies showed that CIV inhibition promotes Aβ production in human brain organoids [10] and stressed mice [24]. Conversely, conflicting evidence also exists. A previous study found that CIV deficiency in a mouse model with human Aβ overproduction reduces brain Aβ load, thereby refuting CIV dysfunction as a contributor to AD-associated brain damage [5]. These divergent results may stem from the use of different experimental systems. However, it should also be noted that previous studies on this topic predominantly used CIV-disrupted approaches but lacked rescue experiments. This limitation restricts the capacity of those studies to establish a causal link or definitive conclusion regarding the role of a CIV deficit in the pathogenesis of this neurodegenerative disorder.To validate the contribution of defective CIV to neuronal damage in AD-relevant contexts, a rigorous experimental design requires a model where CIV function can be selectively restored. However, directly manipulating CIV is technically challenging due to its inherent structural complexity and the yet-poorly-defined pathways underlying CIV dysfunction in AD-related conditions [17, 25,26,27,28,29,30,31,32,33]. These technical and knowledge barriers therefore underscore the need for an indirect but practical approach to compensate for the loss of CIV function while circumventing the challenges of de novo structural reconstruction of the CIV complex. Alternative oxidase (AOX) is a single polypeptide expressed in many lower eukaryote lineages. It functions as a lifesaving means to cope with oxidative stress and mitochondrial crisis by bypassing the defective CIII-CIV segment of the respiratory chain to complete mitochondrial respiration [34,35,36]. Given the advantages of AOX, including its minimal impact on functional CIV [37, 38] and its full compatibility with higher eukaryotes [39,40,41,42,43,44], AOX has been tested in pathological conditions as a viable route for rescuing damages induced by CIV deficits [45], with one study reporting a protective effect of AOX in human Aβ-overexpressing Drosophila [46]. In this regard, we sought to employ AOX not only for our mechanistic study of defective CIV, but also to evaluate its therapeutic potential in pathological settings more relevant to human AD.In this study, we determined CIV dysfunction in synaptosomal mitochondria from the disease-sensitive temporal lobes of patients with AD and the association of CIV dysfunction with disease characteristics. To further investigate this link, we conducted experiments utilizing both mouse- and human-based AD modeling systems. These studies, which included the ectopic expression of AOX within the models, reinforced the finding that defective CIV segment of mitochondrial electron transport contributes to mitochondrial dysfunction and neuronal stress, and notably, that CIV dysfunction enhanced Aβ production by promoting amyloidogenic processing of amyloid precursor protein (APP). Additionally, a proteome-wide association study (PWAS) using genetic data from large-scale AD cohorts identified an association between downregulation of COX11 and AD risk. Collectively, our data suggests mitochondrial CIV dysfunction as a bona fide AD pathology and a promising therapeutic target.Materials and methodsHuman postmortem tissuePostmortem temporal lobe tissue samples were obtained from the University of Kansas Alzheimer’s Disease Research Center (KUADRC) through support from the National Institutes of Health (NIH) Grant P30 AG072973. Associated demographic and neuropathological data (age, sex, MMSE, Braak stage, CERAD score) were provided by the KUADRC. All procedures were conducted in accordance with the Declaration of Helsinki. Informed consent was obtained from all subjects. MMSE scores included in the analysis were derived from the final assessment conducted before death. Postmortem interval (PMI) of all human cases used in the study are within 24 h of death. Due to the small sample size (n = 14) and sex imbalance within the cohort (12 males, 2 females), sex was excluded as a covariate in statistical analyses.MouseAll animal procedures were approved by the University of Kansas Institutional Animal Care and Use Committee (IACUC) and conducted in compliance with the NIH guidelines for the care and use of laboratory animals. B6N(Cg)-Apptm1.1Aduci/J mice (humanized amyloid-β knock-in, hAβ KI; Jackson Laboratory strain #032013) and nontransgenic (nonTg) C57BL/6N controls (Jackson Laboratory strain #005304) were obtained from Jackson Laboratory and maintained in the KU Animal Care Unit under standard housing conditions (12 h light/dark cycle, controlled temperature and humidity). Breeding was performed in-house to generate experimental cohorts. Genotyping was carried out on tail biopsies using polymerase chain reaction (PCR) with allele-specific primers. Animals with body condition score 2 or lower were excluded from the study. No formal randomization procedures were used. Animals were assigned to experimental groups based on genotype and study design requirements. No formal blinding procedures were implemented during animal allocation because experimental groups were defined by genotype and treatment conditions. Sample size were determined based on power calculation with effect size 0.4, significance level 0.05 and statistical power 0.8.Synaptosome isolationPostmortem temporal lobe tissues from AD patients and nonAD controls, as well as cerebral organoids were used for the preparation of synaptosome-rich fractions as previously described [47]. Dissected brain tissues were homogenized in mitochondria isolation buffer (IB buffer, 225 mM mannitol, 75 mM sucrose, 2 mM K2HPO4, and 5 mM HEPES; pH 7.3). The homogenate was centrifuged at 1300 × g for 5 min at 4 °C, and the resulting supernatant was layered onto 15 and 23% Percoll gradients (GE Healthcare, 17–0891-01). Gradients were centrifuged at 34,000 × g for 13 min at 4 °C, yielding two distinct layers; synaptosomes were located at the interface between the 15 and 23% Percoll solutions. This fraction was collected and washed twice in isolation buffer at 4 °C. The final pellet was retained for subsequent analyses.Crude mitochondria isolation and ATP-generating capacity measurementCrude mitochondrial from mouse primary neuronal cultures and hiPSC-derived forebrain neurons were isolated as described below. Neurons were harvested in ice-cold isolation buffer (IB, 225 mM mannitol, 75 mM sucrose, 2 mM K₂PO₄, 0.1% BSA, 5 mM HEPES, and 1 mM EGTA, pH 7.2) and homogenized by passing the cell suspension through a 21-gauge needle 20 times. The homogenate was centrifuged at 1300 × g for 4 min at 4 °C to remove cellular debris. The resulting supernatant was then centrifuged at 10,000 × g for 15 min at 4 °C, and the crude mitochondrial pellet was collected. For the measurement of ATP-generating capacity in mitochondria isolated from hiPSC-derived forebrain neurons, freshly isolated crude mitochondria were energized in IB buffer supplemented with 5 mM glutamate and 5 mM malate. Energized mitochondria were subsequently incubated with the ATP detection reagent (Abcam), and ATP production was initiated by the addition of 200 μM ADP. ATP-generating capacity was determined by monitoring the kinetic increase in ATP-dependent luminescence using a BioTek Synergy Neo2 microplate reader.Enzymatic activity measurementThe activities of mitochondrial oxidative phosphorylation (OXPHOS) for complexes I to V in temporal lobe tissue homogenates, synaptosomes, and crude mitochondria from neuron cultures were measured and calculated as previously described [22, 47, 48]. Tissue homogenate was prepared in phosphate-buffered saline (PBS; Gibco, 10010-031) with protease inhibitor cocktail (SIGMA, P8340). Human temporal lobe synaptosomes and crude mitochondria from neuron cultures were prepared as described in separate sections. The key reagents used for enzymatic activity assays include β-nicotinamide adenine dinucleotide reduced dipotassium salt (SIGMA, N4505), coenzyme Q10 (ACROS, 457950010), decyl ubiquinone (SIGMA, D7911), 2,6-dichlorophenolindophenol sodium salt hydrate (Fluka, 33125-5g-r), cytochrome c from equine heart (SIGMA, C7752), phosphoenolpyruvic acid (SIGMA, P7127), adenosine 5’-triphosphate magnesium salt (SIGMA, A9187), potassium cyanide (KCN) (Fisher Scientific, AAL1327322), disodium succinate (SIGMA, W327700), and oxaloacetic acid (SIGMA, O4126).Blue native polyacrylamide gel electrophoresis (BN-PAGE)Synaptosomes isolated from the temporal lobe were prepared in 4 × NativePAGE™ sample buffer supplemented with 1% NativePAGE™ G-250 additive (Thermo Scientific, BN2008) for BN-PAGE. Samples were resolved by electrophoresis on NativePAGE™ Novex 4–16% Bis-Tris gels and subsequently transferred to PVDF membranes (Bio-Rad) for immunodetection.Mouse primary neuron culture and treatmentMouse primary cortical neuron culture was performed as previously described [49]. Cortices from nonTg or hAβ KI mice were isolated from postnatal day 0 (P0) mouse pups and immediately placed in ice-cold Dulbecco’s Modified Eagle Medium (DMEM; Corning, 10-013-CV). Tissue was enzymatically dissociated in trypsin (Gibco, 25200072) prepared in Hanks’ Balanced Salt Solution (HBSS; Corning, 21-021-CV) at 37 °C for 15 min. Tissue was gently triturated using sterile serological pipettes to obtain a single-cell suspension. Cells were pelleted by centrifugation at 300 × g for 5 min at 4 °C, then resuspended in Neurobasal-A medium (Gibco, 10888022) supplemented with 2% B-27 (Gibco, 17504044) and 2 mM GlutaMAX (Gibco, 35050079). Cell viability was assessed by trypan blue exclusion (0.4%; Gibco, 15250061) using a hemocytometer. Neurons were plated at the appropriate density onto poly-D-lysine-coated culture plates (Sigma-Aldrich, P7405) and maintained under standard culture conditions until use.KCN treatment: At 14–15 days in vitro (DIV), nonTg neurons were treated with 100 μM potassium cyanide (KCN, Sigma-Aldrich, BP310-100) or vehicle for 24 h then proceeded to experiment.AOX expression: Lentiviral packed AOX-HA (pLV-UBC>cin_LOC108950628[XM_018816634.1](co)*/HA, VectorBuilder) or vehicle lentivirus (VectorBuilder) were added to nonTg and hAβ KI neurons. Neurons were maintained until 14-15 DIV to examine AOX expression and neuronal function.ImmunoblottingMouse primary neuron cultures or hiPSC-derived neuron were collected in 1 × bolt LDS loading buffer (Novex, B0007) supplemented with 25 mM DTT. Cerebral organoids were homogenized in in urea buffer (8 M urea, 10% glycerol, 1% SDS, 5 mM DTT, and Tris-HCl, pH 6.8). Proteins were separated in either 10% Bis-Tris Gels (Invitrogen) or 10–20% Tris-Glycine Gels (Novex) and then transferred to PVDF membranes. Membranes were then blocked with 5% non-fat milk at room temperature for 1 h and incubated with the following primary antibodies overnight at 4 °C: mouse anti-APP C-terminal fragment antibody (1:2000, Biolegend, 802803), rabbit recombinant anti-BACE1 antibody (1:1000, Abcam, ab183612), rabbit anti-TOMM40 antibody (1:3000, Proteintech, 18409-1-AP), mouse anti-beta Actin antibody (1:5000, Santa Cruz Biotechnology, sc-47778), rabbit anti-HA (1:1000, CST, 3724 P), rabbit anti-AOX1/2 (1:1000, AGRISERRA, AS04054), rabbit anti-COXIV (1:1000, Proteintech, 11242-1-AP). After probing with goat anti-mouse IgG HRP conjugated or goat anti-rabbit IgG HRP conjugated secondary antibodies (1: 2000–8000, Thermo Fisher Scientific, 31430 and 31460), blots were developed using SuperSignal West Atto Ultimate Sensitivity Substrate (Thermo Scientific, A38556). Images were captured and analyzed with a ChemiDoc XRS+ Gel Imaging System (Bio-Rad).Mitochondrial membrane potentialMitochondrial membrane potential was examined through TMRM (tetramethylrhodamine methyl ester, Invitrogen) staining. Mouse primary neurons were labeled with 200 nM TMRM and 50 nM MitoTracker Deep Red FM (Invitrogen, M22426). Live cell images were captured with a Nikon Ti2 confocal microscope with a Tokai incubator (37 °C, 5% CO2). Images were then quantified using Nikon elements software. Cerebral organoid mitochondria were labeled with 500 nM TMRM and then proceeded to signal detection using Miltenyi MACSQuant Analyzer 10 Flow Cytometer. Data was then analyzed with FlowJo software.Mitochondrial ROS measurementMitochondrial ROS was examined through MitoSOX Red (Invitrogen) staining. Mouse primary neurons were labeled with 50 nM MitoSOX Red and 50 nM MitoTracker Deep Red FM (Invitrogen). Live cell images were captured with Nikon Ti2 confocal microscope with a Tokai incubator (37 °C, 5% CO2). Images were then quantified using Nikon NIS elements software. Cerebral organoid mitochondria were labeled with 200 nM MitoSOX Red then proceeded to signal detection using Miltenyi MACSQuant Analyzer 10 Flow Cytometer. Data was then analyzed with FlowJo software.ATP content measurementATP content in the human temporal lobe, mouse primary neuron cultures, and hiPSC-derived cerebral organoids was examined using an ATP detection assay kit (Abcam, ab113849) following the user manual. All the values were normalized to the amount of protein used for the assay.ImmunocytochemistryMouse primary neuron cultures were fixed in 4% paraformaldehyde (PFA) for immunocytochemistry. Cerebral organoids were prepared according to the protocol from Stemcell Technologies. Briefly, organoids were fixed in 4% PFA then embedded in freshly prepared 7.5% gelatin and then sliced using a Leica CM3050 S cryostat. Neuron and cerebral organoid slices were blocked with blocking buffer (5% GOAT or donkey serum, 0.3% Triton-X 100 in PBS) for 1 h in room temperature followed by overnight room temperature incubation with the following primary antibodies in combination as needed: mouse anti-PSD95 antibody (1:400, Cell Signaling Technology, 36233S), guinea pig anti-vGlut1 antibody (1:400, Synaptic Systems, 135304), mouse anti-MAP2 (1:800, Invitrogen, ma5-12826), rabbit anti-AOX1/2 (1:1000, AGRISERRA, AS04054), mouse anti-VDAC1 (1:300, Abcam, ab154856). After 4 times 5 min washes with PBS, slides were incubated with Alex Fluor secondary antibodies conjugated with fluorophores (1:400, Alexa Fluor 488/594/647, Thermo Fisher) at room temperature for 1 h. Cell nuclei were labeled with DAPI (Invitrogen, D1306). Images were captured with a Nikon Ti2 confocal microscope.Amyloid beta ELISAThe amount of hAβ40 and hAβ42 in mouse primary neuron culture and hiPSC-derived cerebral organoids were measured by commercial kits: human Aβ40 (Invitrogen, KHB3481) and human Aβ42 ELISA kits (Invitrogen, KHB3441). Neurons and cerebral organoids were homogenized with 5 M guanidine HCl/50 mM Tris HCl (pH 8.0) and incubated overnight at room temperature with shaking. The homogenates were then centrifuged at 12,000 × g for 5 min to remove the insoluble material. The supernatants were utilized for Aβ quantification. All the values were normalized with the amount of protein used for the assay. The signals were read on a BioTek Neo2 microplate reader and normalized to total protein content in the samples.Oxygen consumption and mitochondrial respiratory control ratio measurementOxygen consumption and mitochondria respiratory control ratio were measured using a Clark type electrode as previously described (Hansatech) [47]. Synaptosomes isolated from cerebral organoids were energized in IB buffer containing 5 mM glutamate and 5 mM malate. State III respiration, reflecting ADP-stimulated oxygen consumption, was initiated by the addition of 200 µM ADP. Following ADP depletion, oxygen consumption attributable to proton leak across the mitochondrial inner membrane was measured as State IV respiration. The respiratory control ratio (RCR) was calculated as the ratio of State III to State IV oxygen consumption.Cytochrome c oxidase (CcO)-mediated oxygen consumption was assessed by adding 5 mM ascorbic acid and 5 mM artificial CcO substrate N,N,N’,N’-tetramethyl-p-phenylenediamine (TMPD, Sigma Aldrich, T3134-5G) to synaptosome suspensions. Baseline respiration, representing CcO-independent oxygen consumption, was determined following the addition of the CcO-specific inhibitor 1 mM KCN. To evaluate the contribution of AOX, 100 μM AOX inhibitor n-propyl gallate (nPG, Fisher Scientific, MP210274780) was applied for analysis of its effect on the mitochondrial respiratory control ratio.4-HNE measurementCerebral organoid 4-HNE levels were examined using OxiSelect HNE Adduct Competitive ELISA Kit (Cell Biolabs, Inc, STA-838) following the manufacturer’s instructions. All the values were normalized with the amount of protein used for the assay.hiPSC-derived forebrain neuron culture and treatmentHuman induced pluripotent stem cells (iPSCs) were obtained from the Jackson Laboratory (REV/WT: JIPSC001212, SNV/WT: JIPSC001200), the Corriel institute (CU, AG27611), and Wicell (SAD, UCSD231i-SAD1-3). Mycoplasma contamination was tested through RT-PCR every 2 months. Cells were plated on Matrigel-coated culture plates (Corning, 08-774-552) and maintained in mTeSR Plus medium (STEMCELL Technologies, 100-0276) at 37 °C in a humidified incubator with 5% CO₂. iPSCs were routinely passaged for the generation of forebrain neurons.Differentiation into neural progenitor cells (NPCs) and forebrain neurons was performed using the STEMdiff™ Forebrain Neuron Differentiation Kit (STEMCELL Technologies) according to the manufacturer’s instructions. Briefly, hiPSCs were dissociated into single cells using Accutase and plated onto Matrigel-coated culture vessels in mTeSR™ Plus medium supplemented with 10 μM Y-27632 ROCK inhibitor. Upon reaching the appropriate confluency, neural induction was initiated using STEMdiff™ neural progenitor cell Medium. Cells were maintained with regular medium changes according to the manufacturer’s protocol. At approximately days 10–12 of differentiation, NPCs were expanded in Forebrain Neuron Differentiation Medium and passaged as required for further differentiation. To generate forebrain neurons, NPCs were dissociated using Accutase and replated onto Matrigel-coated culture vessels at the density recommended by the manufacturer. Cells were subsequently cultured in STEMdiff™ Forebrain Neuron Maturation Medium with routine medium replacement to promote neuronal maturation. Differentiated neurons were maintained for 30–45 days before experimental analyses. During maturation, cultures developed extensive neurite networks and exhibited typical neuronal morphology. For AOX expression studies, lenti-AOX-HA or vehicle control were introduced at the forebrain maturation stage (day 3 after changing to forebrain maturation medium), and cultures were maintained throughout neuronal maturation. Differentiated forebrain neurons were subsequently used for immunocytochemistry, immunoblotting, mitochondrial functional assays, and measurements of oxidative stress.hiPSC-derived cerebral organoid culture and treatmentREV/WT and SNV/WT hiPSCs were maintained as described in the previous section and were routinely passaged for the generation of cerebral organoids.Cerebral organoids were generated using a commercial kit (STEMCELL Technologies, 08570) according to the manufacturer’s protocol. Briefly, iPSCs were seeded into ultra-low attachment round-bottom 96-well plates (Corning, 7007) at a density of 9000 cells per well in organoid formation medium. On day 5, aggregates were transferred to ultra-low attachment 24-well plates (Corning, 3473) and cultured in induction medium. On day 7, organoids were embedded in Matrigel and moved to ultra-low attachment 6-well plates (STEMCELL Technologies, 100-0083) for expansion over three days. Subsequently, organoids were maintained in maturation medium on an orbital shaker at 37 °C with 5% CO₂. Lentiviral packaged AOX-HA (pLV-UBC>cin_LOC108950628[XM_018816634.1](co)*/HA, VectorBuilder) or vehicle lentivirus (VectorBuilder) were added to day 21 cerebral organoids for AOX expression. Medium was changed every 3-4 days until organoids reached maturity.RT-PCRReal-time polymerase chain reaction (RT-PCR) was performed to examine expression of AOX in hiPSC-derived cerebral organoids. mRNA from the organoids were prepared using RNeasy kits for RNA purification (Qiagen, 74104). mRNA were then converted to cDNA using PrimeScript RT Master Mix (TaKaRa, RR036B). The following primers were used for RT-PCR: AOX1 forward: 5’- GAG AAC TTC CCC CAC TTC CG -3’, reverse: 5’- TCT GCC GGT ATA GGT TCC CA -3’. AOX2 forward: 5’- ACA GGA ACT GGA AAG CGT GA -3’, reverse: 5’- CTG CAC GGA GTA GTA AGC GA -3’. GAPDH forward: 5’-ACA ACT TTG GTA TCG TGG AAG G-3’, reverse: 5’- GCC ATC ACG CCA CAG TTT C-3’. PowerUp™ SYBR™ Green Master Mix (Applied Biosystems, A25742) was used to for target amplification. RT-PCR was performed and analyzed using QuantStudio 5 Real-Time PCR Systems from Applied Biosystems.BACE1 activity measurementBACE1 activity was measured using BACE1 activity assay kit (BPS Biosciences, 71656) following manufacturer’s instructions. All the values were normalized to the amount of protein used for the assay.RNA sequencing and pathway analysisRNA sequencing was performed by the KU genomic sequencing core. mRNA from the organoids were prepared using RNeasy kits for RNA purification (Qiagen, 74104). RNA quality was examined using Agilent Tapestation 4000. The following kits were used for RNA library preparation: NEBNext® Ultra™ II Directional RNA Library Prep with Sample Purification Beads(E7765L), NEBNext® Poly(A) mRNA Magnetic Isolation Module (E7490L), NEBNext® Multiplex Oligos for Illumina® (96 Unique Dual Index Primer Pairs Set 1) (E6440S), NEBNext® Multiplex Oligos for Illumina® (96 Unique Dual Index Primer Pairs Set 2) (E6442S). RNA sequencing was performed with NextSeq 2000 P3 Reagents (100 Cycles) (20040559). Alignment of RNA sequencing raw data was performed using STAR followed by normalization and differential gene expression (DEG) using DEseq2. The raw sequencing data have been deposited in the NIH Sequence Read Archive (SRA) under accession number PRJNA1548129. Mitochondrial genes were filtered using HumanMitoCarta 3.0. Heatmap for RNA sequencing results were generated using R package pheatmap. QIAGEN Ingenuity pathway analysis (IPA) was used to perform mitochondrial canonical pathway and comparative analysis.PWAS and meta analysisProteome-wide association analysis (PWAS) was performed through Mendelian randomization (MR) based on previous publication [50]. Cis-pQTLs (Protein Quantitative Trait Locus) from ROSMAP was used to perform linkage disequilibrium (LD) clumping. AD GWAS summary statistics from three cohorts were used to perform mendelian randomization (Cohort 1: nonAD = 3013, AD = 199; Cohort 2: nonAD = 383,375, AD = 71,880; Cohort 3: nonAD = 3289, AD = 4120) [51,52,53]. Multiple-testing correction was performed using the Bonferroni method to control the family-wise error rate (FWER). The significance threshold was defined as 0.05 divided by the number of proteins tested. The meta-analysis of PWAS statistics from the three cohorts was performed with the use of the weighted Fisher’s method (wFisher) in metapro R package [54]. Mitochondrial genes were filtered using Human MitoCarta 3.0.StatisticsAll statistical analyses were performed using GraphPad Prism 9 (GraphPad Software, San Diego, CA). Depending on the dataset, either unpaired Student’s t-tests or one-way ANOVA followed by Bonferroni post hoc correction were applied. Correlations were assessed using Pearson’s correlation coefficient. Area under the curve (AUC) values were calculated in GraphPad Prism 9.Data are presented as interleaved box-and-whisker plots, where the median is indicated by a line within the box, the interquartile range (IQR) by the box, and 95% confidence intervals (CI) by whiskers; outliers (95% CI) are shown as individual points. Sample sizes were determined based on the availability of human postmortem specimens and on group sizes routinely used in previous studies employing similar biochemical, cellular, and organoid-based approaches. Statistical significance was assessed using predefined statistical tests, and the observed effect sizes were sufficient to detect significant differences in key outcomes. Statistical significance was defined as P