Sulfonic DJ-1 (Cys¹⁰⁶-SO₃H) Binds to and Colocalizes with the Intracellular Accumulation of Amyloid-Beta 42 (Aβ42) in Familial Alzheimer’s Disease PSEN1 E280A Cerebral Organoids Derived from Induced Pluripotent Stem Cells

1. Introduction

Familial Alzheimer’s disease (FAD) is a genetically induced neurodegenerative condition characterized by early-onset dementia (onset before age 65) and a family history of dementia. It is also associated with various non-cognitive neurological symptoms and signs, as well as a more aggressive course [1,2]. Similar to sporadic Alzheimer’s disease (SAD, [3]), FAD presents with abundant plaques composed of extracellular amyloid beta (eAβ), neurofibrillary tangles made of intracellular hyperphosphorylated tau protein (p-Tau), and loss of brain weight due to accelerated neuronal cell death. To date, at least 556 mutations in the presenilin (PSEN1 and PSEN2) and amyloid precursor protein (APP) genes have been identified in FAD (http://www.alzforum.org, accessed October 2025). These mutations affect a common pathogenic pathway in APP synthesis and proteolysis, leading to the excessive production of eAβ via a mechanism that is not yet fully understood [4]. PSEN1 is a key component of the aspartyl protease γ-secretase complex [5] and, together with β-secretase, preferentially cleaves APP (770 amino acids) at residues 713 and 671, respectively, producing an Aβ42 fragment [6]. Several hypotheses have been proposed to explain how eAβ42 induces AD and FAD. These include the amyloid cascade hypothesis [7], the cholinergic hypothesis [8], the oxidative stress hypothesis [9], the two-hit hypothesis [10], the mitochondrial hypothesis [11], the presenilin hypothesis [12], the inside-out amyloid hypothesis [13,14], the tau hypothesis [15], and the ApoE cascade hypothesis [16], among others [17,18]. However, none of these hypotheses have been proven conclusively, and some remain controversial (e.g., [19,20,21,22]. Therefore, the mechanism by which Aβ induces neurotoxicity and cell death is still open to validation.

Specifically, the intracellular amyloid hypothesis [23,24] posits that Aβ accumulation inside neurons, rather than just extracellular plaques, is an early driver of AD pathology, disrupting cellular functions like protein degradation, axonal transport, and cell survival. This accumulation can lead to tau hyperphosphorylation, neuronal dysfunction, and eventually cell death, making iAβ a promising therapeutic target. Consistent with this perspective, our research group has presented substantial evidence in support of the intracellular Aβ hypothesis. We have demonstrated that cholinergic-like neurons (ChLNs) with the E280A or the I416T mutation—which result from a substitution of aspartic acid (E) for alanine (A) at position 280 [25] or isoleucine (I) for threonine (T) at position 416 [26], respectively, —produce aberrant accumulation of iAβ, abnormal phosphorylation of tau, oxidative stress, mitochondrial depolarization, apoptosis, and calcium dysregulation [27,28]. Interestingly, we obtained PSEN1 E280A ChLNs derived from umbilical cord Wharton’s jelly mesenchymal stromal cells (WJ-MSCs) or menstrual mesenchymal stromal cells (MenSCs), which are tissue equivalents [29], using the Cholinergic-N-Run medium (Ch-N-Rm, [30], and observed that by day 7 of the transdifferentiation process, mutant ChLNs exhibited abnormal accumulation of iAβ42, oxidized DJ-1 (i.e., DJ-1Cys¹⁰⁶-SO₃), which is indicative of oxidative stress (OS), and aberrant accumulation of autophagosomes. However, there was no evidence of cell death [27,31]. By day 11, however, cholinergic mutant cells exhibited abnormal phosphorylation of the protein TAU (at Ser²⁰²/Thr²⁰⁵) and positive markers of apoptosis, such as tumor protein p53 (TP53), p-Ser⁶³/Ser⁷³ JUN, p53-upregulated modulator of apoptosis (PUMA), and cleaved caspase-3 (CC3). They also exhibited loss of mitochondrial membrane potential (ΔΨm) and dysfunctional acetylcholine (ACh)-induced Ca²⁺ ion influx [27]. These observations suggest that the accumulation of intraneuronal Aβ, oxidized DJ-1, and impairment of autophagy lysosomal pathway are early event in AD pathogenesis and precedes p-TAU and eAβ deposits [32]. Therefore, iA triggers signals that lead to neuronal dysfunction [33]. Because the earliest pathological detection in PSEN1 E280A ChLNs was the accumulation of iAβ and DJ-1Cys¹⁰⁶-SO₃ [27,31] and exposure to antioxidants, such as epigallocatechin-3-gallate and tramiprosate, simultaneously abolished the accumulation of iAβ and autophagosomes, and DJ-1 oxidation [31], this led us to wonder whether oxidized DJ-1 is essential to iAβ accumulation in mutant ChLN cells.

DJ-1 is a 189-amino-acid protein that is expressed throughout the body and forms dimers under physiological conditions. It is encoded by the PARK7 gene, which was first associated with early-onset, familial forms of Parkinson’s disease (FPD) [34]. As a homodimer protein [35], DJ-1 protects against oxidative stress (OS) by operating as an antioxidant, neuroprotectant, and survival signaling molecule [36]. DJ-1 can sense OS through thiolate Cys¹⁰⁶-SH residue, which is highly susceptible to oxidation by reactive oxygen species (ROS), particularly H₂O₂ [37,38]. Depending on the strength of intracellular oxidation, ranging from moderate to high, the thiolate Cys¹⁰⁶-SH can be oxidized into sulfenic form of DJ-1 (Cys¹⁰⁶-SOH), which can be oxidized to the sulfinic acid form (Cys¹⁰⁶-SO₂H) and then to the sulfonic acid form (Cys¹⁰⁶-SO₃H). The sulfinic DJ-1 form is responsible for DJ-1’s neuroprotective actions [38,39,40]; whereas the sulfonic form is an unstable protein prone to aggregation and loss of function [41,42,43]. Interestingly, DJ-1 has been shown to undergo extensive and irreversible oxidation in the brains of patients with SAD [44,45]. Furthermore, Solti et al. [46] have shown that oxidized DJ-1 aggregates colocalize with pathological amyloid deposits in the postmortem brain tissue of human SAD patients. However, it is not yet known whether oxidized DJ-1 Cys¹⁰⁶-SO₃ colocalizes with iA in FAD postmortem brain tissue or in an in vitro model (e.g., organoids and 2D ChLNs culture).

To gain insight into these issues, we first sought to assess whether oxidized DJ-1, detected with rabbit recombinant monoclonal PARK7/DJ1 antibody, colocalizes with iAβ42 in the frontal and occipital cortex and hippocampus of PSEN1 E280A patients’ postmortem and control brains. Next, we evaluated whether oxidized DJ-1 drives intracellular Aβ42 aggregation using 2D (iPSCs-derived ChLNs) and 3D in vitro models (iPSCs-derived organoids), and in silico molecular docking analysis together with ELISA test and fibril kinetics analysis.

2. Materials and Methods

2.1. Induced Pluripotent Stem Cells Reprogramming

One vial containing 1x10⁶ fibroblast cells (3rd passage) was thawed in one well of a 6-well plate previously treated with Vitronectin (VTN-N, Thermo Fisher Scientific, cat# A14700, Waltham, MA, USA). Once cells became >90% confluent were detached and split in 4 wells of a 6-well plate (1:4 ratio) in fibroblast medium, which include High glucose DMEM, (Thermo Fisher Scientific, cat# 11965092, Waltham, MA, USA), 10% Fetal Bovine Serum (Thermo Fisher Scientific, cat# A5256701, Waltham, MA, USA; 1X NEAA (Thermo Fisher Scientific, cat#11140050, Waltham, MA, USA), and 1X -mercaptoethanol (Thermo Fisher Scientific, cat#21985023, Waltham, MA, USA). On day 0 cells were transduced using the CytoTune™ 2.0 Sendai reprogramming (Thermo Fisher Scientific, cat# A16517, Waltham, MA, USAby incubating overnight with 10 l KOS, 10 l of hc-Myc, and 7 l hKlf4. After 1 day, the medium was replaced with fresh complete fibroblast medium to remove the CytoTune™ 2.0 Sendai reprogramming vectors, then after we changed the medium every other day, and once cells became >90% confluent they were detached and split in 3 wells of a 6-well plate. On day 7, the medium was changed to complete Essential 8™ Medium (Thermo Fisher Scientific, cat# A1517001, Waltham, MA, USA). On days 9–28 the medium was replaced every day and monitor the culture vessels for the emergence of iPSC colonies.

2.2. Neural Precursor Cell Generation Protocol

Human iPSC cells were mechanically detached from VTN-N surface. Embryoid bodies (EBs) were generated by transferring iPSCs to non-adherent plates in E6 medium (Thermo Fisher Scientific, cat# A1516401, Waltham, MA, USA) at 37 °C in 5% CO2. After 7 days, EBs were transferred to a non-adherent plate and E6 medium was supplemented with 10ng/ml bFGF (Thermo Fisher Scientific, cat# 100-18B-50UG, Waltham, MA, USA), after 2 days, the floating structures were dissociated by trituration and transferred to an VTN-N-treated dish. For generation of neural precursor cells (NPCs), EBs were cultured in NPC medium (Neurobasal medium, 1% N2 supplement, 2% B27 supplement (Thermo Fisher Scientific, cat# 17504044, Waltham, MA, USA), 20 ng/ ml epidermal growth factor (Thermo Fisher Scientific, cat#PHG0311L, Waltham, MA, USA), 1g/ ml heparin sodium salt (Thermo Fisher Scientific, cat#A16198.MD, Waltham, MA, USA), 1ng/ ml bFGF, 1X -mercaptoethanol, and 1% penicillin/streptomycin (Thermo Fisher Scientific, cat#15140122, Waltham, MA, USA).

2.3. Generation of Cholinergic Neurons from Neural Precursor Cells (NPCs)

Neural precursor cells (NPC) were seeded at a density of 3×10⁴ cells/cm² in 24-well culture plates and maintained for 24 h in NPC culture medium under standard conditions. Following this period, the medium was replaced with a cholinergic differentiation medium (Cholinergic-N-Run, [30]), and cells were incubated at 37 °C for 7 days, as previously described [27]. After the induction phase, the differentiation medium was replaced with neural medium (NM) consisting of Neurobasal medium supplemented with 1×N2 (Thermo Fisher Scientific, cat#17502048, Waltham, MA, USA) and 1% penicillin/streptomycin.

2.4. Generation of Cerebral Organoids (COs)

Cerebral organoids (COs) were generated by differentiating wild-type (WT) and PSEN1 E280A mutant induced pluripotent stem cell (iPSC)-derived neural progenitor cells (NPCs), following the protocol described in ref. [29]. Briefly, WT and mutant NPCs were cultured in a novel medium formulation, Fast-N-Spheres V2 [47], supplemented with Corning Matrigel® (Cat# 356232, Thermo Fisher Scientific Inc., Santa Fe, NM, USA) and 1% fetal bovine serum (FBS) (Cat# CVFSVF00-01, Eurobio Scientific, Les Ulis, France). Cells were maintained under standard conditions until the spontaneous formation of neurospheres was observed. Subsequently, the spheres were transferred to ultra-low attachment culture dishes and continuously agitated at 60 rpm. The culture medium was refreshed every 3–4 days, and organoids were maintained for a total of 60 days.

2.5. Western Blot Analysis

Cells were incubated as described above, detached with 0.25% trypsin, and lysed in 50 mM Tris-HCl, pH 8.0, with 150 mM sodium chloride, 1.0% Igepal CA-630 (NP-40), and 0.1% sodium dodecyl sulfate and a protease inhibitor cocktail (cat#P8340, Sigma-Aldrich Co. LLC, (USA)). All lysates were quantified using the bicinchoninic acid assay (Thermo Scientific cat # 23225, Waltham, MA, USA). Extracted samples (40 μg of proteins) were heated at 95 °C for 5 min in 2 x SDS and 20x reducing agent (except for protein oxDJ-1) and loaded into 12% Bis/Tris gels at 120 V for 90 min, and the bands were transferred onto nitrocellulose membranes (Hybond-ECL, Amersham Biosciences) at 270 mA for 90 min using an electrophoretic transfer system (BIO-RAD). The membranes were incubated overnight at 4 °C amyloid β1–42 and ox(Cys106) DJ1 primary antibodies (1:5000). The anti-actin antibody (1:1000, cat #MAB1501, Millipore) was used as an expression control. Secondary infrared antibodies (goat anti-rabbit IRDye® 680RD, cat #926–68071; donkey anti-goat IRDye ® 680RD, cat # 926–68074; and goat anti mouse IRDye ® 800CW, cat #926–32270; LI-CORBiosciences, Lincoln, NE, USA) at 1:1000 were used for western blotting analysis, and data were acquired using Odyssey software. The assessment was repeated three times in independent experiments.

2.6. Immunofluorescence Analysis

For the analysis of neural-, Alzheimer’s disease-, oxidative stress- and cell death-related markers, the cells treated under different conditions were fixed with cold ethanol (-20 °C) for 20 min, followed by Triton X-100 (0.1%) permeabilization and 10% bovine serum albumin (BSA) blockage. Cells were incubated overnight with primary neural antibodies against OCT4 (1:500), SOX-2 (1:500), NANOG (1:500), and KLF4 (1:500), the neuronal marker Nestin (1:500; cat# MA1-5840, Invitrogen, Waltham, MA, USA); and glial fibrillary acidic protein (GFAP 1:200, cat# sc6170, Santa Cruz, Dallas, TX, USA), microtubule-associated protein 2 (MAP2, 1:250, cat MA1-25044, Invitrogen, Carlsbad, CA, USA), β-tubulin III (1:250, cat# G712 A, Promega, Madison, WI, USA) and choline-acetyltransferase (ChAT, 1:50, cat# AB144 P, Millipore, Burlington, MA, USA), amyloid β1–42 (1:500; clone 6E10, cat# 803014, Biolegend, San Diego, CA, USA), and primary antibodies against oxidized DJ-1 (1:500; ox(Cys¹⁰⁶)DJ1; spanning residue C¹⁰⁶ of human PARK7/DJ1; oxidized to produce cysteine sulfonic (SO₃) acid; cat #ab169520, Abcam, Cambridge, UK). After exhaustive rinsing, we incubated the cells with secondary fluorescent antibodies (DyLight 488 and 594 horse anti-rabbit, -goat and -mouse, cat DI 1094, DI 3088, and DI 2488, respectively) at 1:500. The nuclei were stained with 1 μM Hoechst 33342 (Life Technologies, Carlsbad, CA, USA), and images were acquired on a Floyd Cells Imaging Station microscope (Life Technologies, Carlsbad, CA, USA).

2.7. Flow Cytometry Analysis

For flow cytometry analyses, cells were detached using trypsin and centrifuged for 10 min at 2000 rpm. Then, cells were fixed using cold ethanol at -20 °C overnight. Then, cell suspensions were washed with PBS and incubated with 0.2% Triton X-100 plus 1.5% bovine serum albumin (BSA) for 30 min. After, cells were incubated with primary (see above). After exhaustive rinsing, we incubated the cells with secondary fluorescent antibodies (DyLight 488 and 594 horse anti-rabbit, -goat and -mouse, cat DI 1094, DI 3088, and DI 2488, respectively, Thermo Fisher Scientific, Waltham, MA, USA) at 1:500. Fluorescence analysis was performed on a BD LSRFortessa II flow cytometer (BD Biosciences, Becton, Dickinson and Company, BD Biosciences, 2350 Qume Dr, San Jose, CA 95131-1812, USA). Cells without primary antibodies served as a negative control. For assessment, 10,000 events and quantitative data and figures were obtained using FlowJo 7.6.2 Data Analysis Software (TIBCO® Data Science, Palo Alto, Ca, USA). Events analysis was performed by determining the cell population (Forward Scatter analysis, Y axis) that exceeded the basal fluorescence (488 nm or 594 nm, X axis) of the negative control. Accordingly, contour diagrams were created from event analysis, and the cells located in the box (quadrants labeled as + or (+)) represent the cell population exceeding the basal fluorescence.

2.8. Molecular Docking Analysis

To enable the 3D structure of DJ-1, the PDB database was used to access the PDB format of DJ-1 proteins under different oxidation status (Sulfenic, PDB:4p34; Sulfinic, PDB: 1soa; Sulfonic, PDB: 3bwe (aggregated)). Moreover, AlphaFold2.ipynb program (https://colab.research.google.com/github/sokrypton/ColabFold/blob/main/AlphaFold2.ipynb, accessed in July 2025) was loaded with the 42-aminoacid amyloid beta protein (Structured with AlphaFold2). The blind molecular docking was performed with HDOCK Server. For analysis, we selected the docking poses with the strongest Vina score. The generated PDB files of the molecular docking of each protein were visualized with the PDB viewer interphase.

2.9. Real-Time Quaking Induced Conversion (RT-QuIC)

Amyloid aggregation measurement was performed following minor modifications to a previously optimized protocol [48]. Purified DJ-1 protein was prepared at a concentration of 100 µg/mL in PBS, and 1 µg of protein was subsequently diluted in seeding buffer (PBS, pH 7.4). Synthetic Aβ42 peptide (100 µg/ml, Cat# ab120301, Abcam, Cambridge, UK,) was dissolved in DMSO (Cat#34869, Sigma-Aldrich Co. LLC, (USA)) and sonicated for 5 minutes immediately before the reaction. The peptide was then added to either untreated or H₂O₂ (100 µM) treated recombinant DJ-1 protein (Cat#P219-31H, Sino Biological, Beijing, China) in the presence of Congo Red (CR, 10 µM) in PBS. The final reaction volume for each mixture was 100 µL. Multiple technical replicates of each condition were incubated simultaneously in a Multiskan SkyHigh Plate Reader (Thermo Fisher Scientific, cat#A51119600DPC, Waltham, MA, USA) for 48 hours under intermittent shaking (600 rpm for 1 minute every 60 minutes) at 37 °C. Absorbance measurements were recorded every 60 minutes at 420 nm and 540 nm and used for subsequent analysis.

2.10. ELISA Test (Modified)

To determine the levels of oxidized DJ-1 (oxDJ-1) bound to Aβ₄₂ peptides, we designed a modified version of the Human Aβ₄₂ solid-phase sandwich ELISA (Cat. No. KHB3441, Invitrogen, Waltham, MA, USA). Briefly, recombinant DJ-1 protein (1 µg) was either left untreated or oxidized with H₂O₂ for 5 min. Each protein sample was then incubated with recombinant Aβ₄₂ peptide (1 µg) for 10 min. The resulting protein mixtures were subsequently used as substrates in the ELISA procedure. During the initial incubation step (in plate wells), the oxidized form of DJ-1 was detected using a specific anti-PARK7/DJ-1 antibody (oxidized cysteine sulfonic acid, Cat. No. ab169520, Abcam, Cambridge, UK) as the capture antibody. The remaining steps were performed according to the manufacturer’s instructions. Binding levels were determined as the absorbance values after subtraction of the blank. This assay was conducted in triplicate in three independent experiments, with the experimenter blinded to sample identity.

2.10. Data Analysis

In this experimental design, two codes of iPSCs were cultured (WT PSEN1 and PSEN1 E280A) and the cell suspension was pipetted at a standardized cellular density of 2 × 10⁴ cells/cm² into different wells of a 24- or 6-well plate. Cells (i.e., the biological and observational units) [49] were randomized to wells by simple randomization (sampling without replacement method), and then wells (i.e., the experimental units) were randomized to treatments by a similar method. Experiments were performed on three independent occasions (n = 3) blind to the experimenter and/or flow cytometer and/or microscopy analyst. The data from the three repetitions, i.e., independent experiments, were averaged, and representative flow cytometry density or histogram plots from the three independent experiments were selected for illustrative purposes, whereas the bars in the quantification figures represent the mean ± SD and the three black dots show the data point of each experimental repetition. Based on the assumptions that the experimental unit (i.e., the well) data comply with the independence of observations, the dependent variable is normally distributed in each treatment group (Shapiro–Wilk test), and there is homogeneity of variances (Levene’s test), where the statistical significance is determined by a one-way analysis of variance (ANOVA) followed by Tukey’s post hoc comparison calculated with GraphPad Prism 5.0 software. Differences between groups were only deemed significant with a p-value of 0.05 (*), 0.01 (**), and 0.001 (***). All data are presented as the mean ± S.D.

3. Results

3.1. A 42 and Sulfonic DJ-1 (Cys¹⁰⁶-SO₃H) Aggregates Colocalize in Familial Alzheimer’s Disease (FAD) PSEN1 E280A Brain Samples

Because oxidized DJ-1 aggregates colocalize with pathological amyloid deposits in the postmortem brain tissue of human sporadic Alzheimer’s disease (SAD) patients [46], we examined three regions of the brain: the hippocampus, the frontal and the occipital cortices. These regions are specifically involved in FAD caused by the PSEN1 E280A mutation [50], as well as in control cases. Immunohistochemical analysis revealed no reactivity of specific monoclonal antibodies (e.g., E610) against Aβ42 or sulfonic DJ-1 (SO₃H) in the wild-type (WT) PSEN1 hippocampus (Figure 1A–C and 1A′–C′), frontal cortex (Figure 1H–J and 1H′–J′), and occipital cortex (Figure 1O–Q and 1O′–Q′). However, evident oxDJ-1 (Cys¹⁰⁶) and Aβ42 fluorescent aggregates were detected in the PSEN1 E280A hippocampus (Figure 1D–F), frontal cortex (Figure 1K–M), and occipital cortex (Figure 1R–T). A close examination revealed that oxDJ-1 (Cys¹⁰⁶) and Aβ42 aggregates colocalize in PSEN1 E280A samples (Figure 1D’-F’, 1K’-M’, and 1R’-T’). Overall, PSEN1 E280A exhibited higher intracellular colocalization levels than control samples in the hippocampus (Figure 1G), frontal cortex (Figure 1N), and occipital cortex (Figure 1U).

3.2. Intracellularly, Aggregated Sulfonic DJ-1 (Cys¹⁰⁶-SO₃H) Colocalized with Aggregated Aβ42 in PSEN1 E280A Cerebral Organoids (COs).

Next, we investigated whether sulfonic DJ-1 (Cys¹⁰⁶-SO₃H) and intracellular Aβ42 could colocalize in cerebral organoids, which serve as a model of familial Alzheimer’s disease (FAD). To achieve this aim, we first generated induced pluripotent stem cell (iPSC)-derived cerebral organoids (COs). Figure 2 shows time-lapse microphotographs of iPSC transformation into COs from wild-type (WT) (Figure 2A–D) and PSEN1 E280A (Figure 2E–H) cells from day 0 to 20. Light microscopy analysis shows no significant morphological alterations in WT or mutant COs. Therefore, we selected day 20 to evaluate the expression of the neuronal marker MAP2, the cholinergic lineage marker ChAT, and the astrocyte lineage marker GFAP (Figure 2I and 2J). We observed that the three markers were readily expressed in both WT (Figure 2I’-2III’’’) and mutant (Figure 2J’-2J’’’) COs. Overall, there was no statistical difference in the expression of the MAP2 (Figure 2K), ChAT (Figure 2L), and GFAP (Figure 2M) markers between WT and PSEN1 E280A COs. The cultures were then left to progress until day 50. Time-lapse microphotographs from days 25 to 50 show no morphological changes (Figure 2N–U) or statistically significant differences in diameter length between WT and PSEN1 E280A COs (Figure 2V). Therefore, we inferred that the PSEN1 E280A mutation does not alter the normal development of iPSC-derived COs (Figure 2A–V).

Further analysis was performed to detect the presence of iAβ42 and oxidized DJ-1 in COs on day 20. As expected, wild-type (WT) COs showed almost no iAβ42 or oxidized DJ-1 (Figure 3A and the inset), whereas PSEN1 E280A COs conspicuously expressed iAβ42 (Figure 3B’ and the inset) and oxidized DJ-1 (Figure 3B’’). Interestingly, oxDJ-1 colocalized with Aβ42 in mutant COs, showing a significant increase in oxDJ-1 (Figure 3C), iAβ42 (Figure 3D), and a higher colocalization ratio in PSEN1 COs (Figure 3E).

To confirm that the colocalization of A42 and OxDJ-1 aggregates was intracellular and not due to Aβ plaques and oxidized DJ-1 [46], we stained COs or postmortem tissue samples (as controls) with BTA-1, which is a probe for β-amyloid aggregates or A plaques. Figure 4 shows that neither WT (Figure 4A) nor PSEN1 E280A COs (Figure 4B) stain positive for BTA-1 (Figure 4C). In contrast, while post-mortem negative control brain sample showed none BTA-1 stain reactivity (Figure 4D), PSEN1 E280A showed abundant A plaques as stained in pink fluorescence (Figure 4E and 4F).

3.3. Intracellularly, Aggregated Sulfonic DJ-1 (Cys¹⁰⁶-SO₃H) Colocalized with Aggregated Aβ42 in PSEN1 E280A Cholinergic Neurons (ChNs).

To further refine our investigation of intracellular Aβ42-DJ-1 colocalization at the cellular level, we obtained cholinergic neurons (ChNs) from induced pluripotent stem cell (iPSC)-derived neural precursor cells (NPCs). Figure 5A–F illustrate the progressive development of fibroblast-derived iPSCs in wild-type (WT) and PSEN1 E280A cells over the course of three time points: day 1 (Figure 5A and 5B), day 7 (Figure 5C and 5D), and day 28 (Figure 5E and 5F). The iPSC colonies were noticeable by day 7 in the wild-type (WT) cells (Figure 5C, indicated by a blue broken circle) and in the PSEN1 E280A mutant cells (Figure 5D, indicated by a red broken circle). Additional analysis of pluripotency markers using the immunofluorescence technique revealed that both the WT and PSEN1 E280A fibroblast-derived iPSCs readily expressed OCT4 and SOX2 (Figure 5G and 5I), as well as NANOG and KLF4 (Figure 5H and 5J), respectively. There was no statistical difference in pluripotency marker expression between WT and PSEN1 E280A iPSCs (Figure 5K). Similar results were obtained using flow cytometry analysis (Figure 5L–N). The induced pluripotent stem cell (iPSC)-derived wild-type (WT) (Figure 5O) and PSEN1 E280A (Figure 5P) embryoid bodies (EBs) and cells obtained by dissociation (Figure 5Q and 5R) expressed Nestin, a widely used marker for neural stem and progenitor cells (NSPCs), and SOX2, a neuronal lineage marker, in both WT (Figure 5S) and PSEN1 E280A (Figure 5T). There was no statistical difference in Nestin or SOX2 expression between WT and PSEN1 E280A iPSCs (Figure 5K). Similar results were obtained via flow cytometry analysis (Figure 5V and 5W).

The NPC-derived wild-type (WT) (Figure 6A) and NPC-derived PSEN1 E280A (Figure 6E) cholinergic neurons (ChNs) expressed β-III tubulin, the earliest marker of neuronal differentiation (Figure 6B and 6F), as well as ChAT, a specific cholinergic marker (Figure 6C and 6G). However, they expressed low levels of the astrocyte marker GFAP (Figure 6D and 6H). No statistical difference was observed in β-III tubulin, ChAT, and GFAP expression between NPC-derived WT and PSEN1 E280A ChNs (Figure 6I). Similar results were obtained via flow cytometry analysis (Figure 6J, 6K, and 6L). Previous studies have shown that the earliest pathological markers found in ChLNs are iAβ42 and oxidized DJ-1 (Cys¹⁰⁶-SO₃) on day 0 (day 7 after transdifferentiation) [27,31]. Here, we extend this observation to include days 2 and 4 (days 9 and 11 after transdifferentiation). Western blot analysis revealed that WT ChNs showed no detectable iAβ42 or ox-DJ-1 at any evaluated time point (Figure 6M). However, PSEN1 E280A ChNs expressed iAβ42 and ox-DJ-1 in a time-dependent manner (Figure 6N). Moreover, both iAβ42 or ox-DJ-1 aggregates significantly increase in PSEN1 E280A ChNs compared to WT ChNs (Figure 6O and 6P). Similar results were obtained by immunofluorescence microscopy analysis (Figure 6Q-Y).

3.4. Sulfonic DJ-1 (Cys¹⁰⁶-SO₃H) Protein Binds A42 More Efficiently than Sulfenic DJ-1 (Cys¹⁰⁶-SOH) or than Sulfinic DJ-1 (Cys¹⁰⁶-SO₂H) In Vitro

The above observations prompted us to evaluate the molecular interactions between the Aβ42 peptide and the DJ-1 protein. In silico molecular docking analysis showed that DJ-1 binding to monomeric Aβ42 depends on the oxidative state of the DJ-1 protein. Theoretical calculations predict an increase in binding affinity from sulfenic DJ-1 (DS: -195.81Figure 7A) to sulfinic DJ-1 (DS: -203.03, Figure 7B) to aggregated sulfonic DJ-1 (DS: -228.94, Figure 7C) in the presence of monomeric Aβ42 (Table 1). Sulfonic DJ-1 (aggregated) displayed more receptor-ligand interface residue pairs with Aβ42 (e.g., 75) than sulfenic (e.g., 43) or sulfinic DJ-1 (e.g., 45). Overall, sulfonic DJ-1 (aggregated) efficiently forms a protein complex with Aβ42 with 70% affinity folds (bonding residues) compared to sulfenic or sulfinic DJ-1. To verify the binding of sulfonic DJ-1 with Aβ42, we performed an ELISA assay (Figure 7D and 7E) and real-time quaking-induced conversion (RT-QuIC) experiments. In the absence of H₂O₂ (Figure 7D and 7F), the ratio of sulfonic DJ-1 binding to A2 was extremely low, whereas in the presence of H₂O₂, the binding ratio increased 28-folds (Figure 7E and 7F). In the absence of H₂O₂ (Figure 7G), the RT-QuIC analysis revealed that synthetic Aβ42 underwent fibrillar elongation, which was characterized by in a lag phase (0–5 h, 0-20% CR relative absorbance). This phase represents the time at which sub-detectable growth of or initial growth of Aβ seeds occurs. The analysis also revealed an exponential growth phase (5–9 h, 20-90% CR), which represents the time at which detectable Aβ42 fibril growth occurs and generation of new seeding surfaces by fibril fragmentation and/or secondary nucleation. Finally, the analysis revealed a plateau phase (9–48 h, 70-90% CR), which reflects the exhaustion of available Aβ42. Under H₂O₂ exposure (Figure 7H), the CR50% was almost unaltered for both conditions; however, sulfonic DJ-1 significantly reduced typical Aβ₄₂ fibrillar growth, primarily during the plateau phase (Figure 7H). This reduction occurred from an initial CR relative absorbance of almost 90% (A42 as control) to ~70% by 9-48 h, representing a -22% decrease in Aβ42 CR relative absorbance.

4. Discussion

In the present study, we demonstrate that sulfonated DJ-1 (DJ-1 Cys¹⁰⁶-SO₃H) colocalizes with intracellular Aβ₄₂ in cells from the hippocampi and frontal and occipital cortices of brains of patients with FAD caused by the PSEN1 E280A mutation. These findings contradict those of Solti et al. [46]. While those authors suggest that oxidized DJ-1 aggregates colocalize with amyloid deposits in the frontal cortex of human SAD patients, we found intracellular accumulation of Aβ42 and DJ-1 in brain cells (see Figure 5). One possible explanation for this discrepancy is the different antibodies used to detect Aβ. The Solti’s group used a polyclonal rabbit antibody (Abcam, Cat# ab2539) raised against human APP, whereas we used a monoclonal anti-β-amyloid antibody (BioLegend, Clone 6E10, Cat# 803014, RRID: AB_2728527). Interestingly, we detected the colocalization of DJ-1 and Aβ42 in blood vessels for the first time in FAD brain samples (Figure 1T). This observation suggests that not only do DJ-1 and Aβ42 proteins aggregate in neuronal cells, but endothelial cells are also capable of accumulating and aggregating intracellular Aβ42 [51] and oxidized DJ-1. Consequently, the positive identification of sulfonic DJ-1 and Aβ42 should be considered a marker of cerebral amyloid angiopathy (CAA) in FAD brain pathology. However, it remains unclear whether PSEN1 E280A neurons accumulate and aggregate intracellular Aβ by a similar mechanism as endothelial cells [52], a question that deserves further investigation. We also present evidence of the concurrent accumulation of iAβ and oxidized DJ-1 in PSEN1 E280A iPSC-derived cerebral organoids neurons (COs) and NPC-derived cholinergic neurons (ChNs). Similar proteinaceous pathology was also observed in MSCs/MenSCs PSEN1 E280A-derived ChLNs and cerebral spheroids [31,53,54] or PSEN1 I416T ChLNs derived from MenSCs [28]. These observations suggest that the colocalization of DJ-1 with Aβ42 occurs independently of the cells’ pluripotent or stromal origin. Furthermore, both proteins appear to colocalize simultaneously in the early stages of cholinergic lineage development (e.g., [27,31]). However, the nature of their colocalization remains unclear i.e., are they accidental encounters or partners? Our findings support the notion that A42 and oxidized DJ-1 are associates proteins.

In silico molecular docking analysis suggest that sulfonic DJ-1 physically interacts with iAβ42. Such theoretical prediction was confirmed by ELISA test and RT-QuIC. According to ELISA test, sulfonic DJ-1, but not reduced DJ-1, binds to A42. Likewise, RT-QuIC experiments show that sulfonic DJ-1, but not reduced DJ-1, was able to bind to monomeric iAβ42. In fact, in the RT-QuIC test, sulfonic DJ-1 significantly reduced Aβ42 fibril formation during the plateau phase, likely due to a shortage of available monomeric Aβ. Since sulfonic DJ-1 did not interfere with the growth phase of Aβ fibrils, these results suggest that DJ-1 binds to iAβ42 in a noncovalent manner. Interestingly, sulfonic DJ-1 has been shown to form aggregates not only with Aβ but also with α-synuclein (α-Syn) and p-Thr205 tau, intraneuronal pathogenic proteins involved in PD and AD, respectively [46]. Unexpectedly, these DJ-1-protein complexes (Aβ, α-synuclein, and p-Tau) may overwhelm the autophagy-lysosomal pathway, an efficient protein degradation system most impaired in AD [31,55]. Further investigation is needed to determine if sulfonic DJ-1 physically interacts with α-synuclein and p-Tau, as demonstrated in the present study with Aβ42.

Our findings support the intracellular amyloid hypothesis [23,24]. Indeed, we (e.g., [27,28] and others [56,57,58,59,60,61,62,63] have provided evidence for the intracellular accumulation of Aβ within neurons, including studies on postmortem AD brains, transgenic mouse brains, and in vitro AD models. Recently, we have demonstrated that, during the transdifferentiation of mesenchymal stromal cells (MenSCs) into the PSEN1 E280A cholinergic lineage, the production and accumulation of the iAβ42 fragment increased after seven days of transdifferentiation of the PSEN1 E280A ChLNs compared to day zero of transdifferentiation [31]. In parallel, we observed a significant increase in DJ-1C¹⁰⁶-SO₃ at day 7 and an increase in autophagosome accumulation at day 5 of transdifferentiation, though there were no detected cell death markers. These observations suggest that the earliest pathological events in PSEN1 E280A ChLNs are the simultaneous accumulation of iAβ42, sulfonic DJ-1 (Cys¹⁰⁶-SO₃), and autophagosomes [31]. Since antioxidant agents (e.g., EGCG and tramiprosate) abolished these molecular events in mutant ChLNs, we suspected a link between them. However, the mechanism by which iA, DJ-1 Cys¹⁰⁶-SO₃, and autophagosomes interact has not been clearly delineated. One possible explanation is that PSEN1 E280A-overproduced iA42 generates H₂O₂ through direct or indirect inhibition of mitochondrial complex I [64,65]. Since DJ-1 is an atypical peroxiredoxin-like peroxidase that scavenges H₂O₂ through oxidation of Cys¹⁰⁶-SH [66], this molecule can turn sulfenic acid (Cys¹⁰⁶-SOH) into sulfonic acid groups (Cys¹⁰⁶-SO₃H) in DJ-1 protein [37,67], leading to loss of function and protein aggregation [42]. Interestingly, sulfonic acid DJ-1 (Cys¹⁰⁶-SO₃) can form a protein complex with iA42 (this work), which possibly overcharged the autophagy-lysosomal pathway, wherein the autophagosomes accumulate and fail to mature into functional autolysosomes [68], leading to a blockage in the degradation process in mutant ChLNs. Therefore, early suppression of H₂O₂ generation (e.g., by antioxidants) might block the cascade of events, such iA42> mitochondria Complex I > H₂O₂> DJ-1 (Cys¹⁰⁶-SO₃)>> iA42-sulfonic DJ-1 complex, leading to dysregulation of autophagy. Our present work and previous one (e.g., [27,31]) supports such a scenario (Figure 8).

5. Conclusions

We have consistently observed the simultaneous presence of Aβ42 and oxidized DJ-1 in PSEN1 E280A postmortem brain tissue, including the frontal and occipital cortices and the hippocampus, as well as in cerebral organoids derived from PSEN1 E280A induced pluripotent stem cells (iPSCs) and neuronal precursor cell-derived cholinergic neurons. An in silico molecular analysis predicted that sulfonic DJ-1 might bind to Aβ42. The ELISA test and RT-QuIC findings corroborated this assumption. Taken together, these observations suggest that the colocalization of Aβ42 and DJ-1 in pathological tissue, organoids, or cells may be due to a physical interaction between the two proteins. Based on previous findings by our group [27,31] and this study, Figure 8 illustrates a potential mechanism by which oxidized DJ-1 and Aβ42 protein complexes may impair the autophagy-lysosome system and cause neuronal death.