Extracellular Aβ42 Oligomers Induce ROCK2 Hyperactivation Through Dual Mediation by RhoA and GzmB: Significance of Moderate ROCK2 Activity in Neural Cells

Abstract

Highlights

What are the main findings?

• EAO induces ROCK2 hyperactivation via integrin/RhoA- and GzmB-associated processing.

• EAO-induced ROCK2 hyperactivation may be a key link between EAO and actin cytoskeletal dysfunction.

• Excessive inhibition of ROCK2 activity may result in excessive neurite outgrowth.

What are the implications of the main findings?

• An imbalance in ROCK2 activity may disrupt intrinsic neuronal networks.

• Modulating abnormal ROCK2 activity may provide a potential therapeutic strategy for AD.

Abstract

Alzheimer’s disease (AD) is characterized by neurite degeneration and neuronal death. Extracellular amyloid-β 1-42 (Aβ42) oligomers (EAO) not only disrupt the homeostasis and function of the extracellular matrix (ECM) but also damage neural cells through direct binding. Previous studies have demonstrated that EAO binding to membrane integrins reduces neuronal motility, adhesion, and neuritogenesis. To identify the key molecular switch(es) responsible for these actin cytoskeleton dysfunction-associated events, this study utilized neuronal and glial cell lines as well as AD model mice to investigate the cascade underlying EAO-induced actin cytoskeleton dysfunction. This study revealed that EAO induce the dual activation of ROCK2 through RhoA and granzyme B (GzmB) mediation, with GzmB-mediated ROCK2 activation constituting a significant component of this process. ROCK2 hyperactivation in response to EAO causes dynamic dysregulation of the actin cytoskeleton, defective neuritogenesis, and ultimately reduced cell survival, leading to disturbances in brain cell populations. However, the excessive inhibition of ROCK2 activity might cause excessive neurite outgrowth, which may disrupt intrinsic neuronal networks or normal neural transmission. Thus, the disruption of ROCK2 activity might lead to impaired neuritogenesis and disturbances in brain cell populations. The findings of this study may provide important insights into AD pathogenesis and feasible therapeutic targets.

1. Introduction

Alzheimer’s disease (AD) is the most commonly occurring neurodegenerative disorder and a leading cause of dementia. The aggregation and deposition of extracellular amyloid-β protein 1-42 (Aβ42) in brain tissue is considered the initiating event in the pathogenic cascade leading to the development and progression of AD [1]. In addition to directly disrupting the homeostasis and integrity of the extracellular matrix (ECM) through invasive physical, chemical, and biological processes [2,3], these metastable and heterogeneous Aβ42 aggregates, particularly the highly pathogenic soluble Aβ42 oligomers (Aβ42O) [4,5], also interact to varying degrees with specific membrane receptors or cell surface molecules, resulting in abnormal alterations in intracellular signaling pathways and metabolic dysfunction, including cell membrane damage [6,7,8], which may ultimately lead to neuronal damage or death.

Cell surface integrins are a major class of extracellular matrix receptors that directly mediate interactions between cells and the extracellular environment by altering their transmembrane conformation in response to recognition of sequences containing the RGD (S) (arginine–glycine–aspartate[-serine]) tripeptide/tetrapeptide present in the majority of matrix proteins [9,10]. The extracellular milieu of neurons in the brain consists of glial cells and the ECM, which support, buffer, and nourish neurons to maintain their proper morphology and function. Therefore, through integrin-mediated mechanisms, damage or pathological alterations in extracellular components (including extracellular factors, ligands, and the matrix) or glial cells greatly affect neuronal growth and morphogenesis, including neurite outgrowth and dendritic differentiation, as well as synapse formation and remodeling, which are essential for the establishment of neural/neuritic connections and communication within the brain [11]. However, the key processes and intracellular signaling pathways involved remain unclear.

Over the past 20 years, numerous studies have demonstrated a strong correlation between the onset and progression of AD and dysregulation of cell surface integrins [12,13,14,15]. Since the RHDS motif, corresponding to amino acid residues 5 to 8 of the Aβ42 chain within an EAO unit, shares similarities with the RGDS motif [16], our previous studies demonstrated that extracellular Aβ42O (EAO) functions as an unconventional and atypical ligand for integrins. Consequently, EAO binding to membrane integrins decreases or impairs neuronal motility, adhesion, and neurite outgrowth [17]. Because neurite outgrowth, migration, and adhesion are functional activities of brain cells that largely depend on dynamic alterations in F-actin [18,19,20], our findings, together with those of others, suggest the possibility of the dysregulation or dynamic imbalance of the actin cytoskeleton.

It is widely recognized that both intracellular regulation and the extracellular environment influence the dynamic assembly and remodeling of the actin cytoskeleton, enabling adaptation to the changes in extracellular homeostasis. The interaction between integrins and normal ECM components may be blocked or disrupted by the production and deposition of Aβ42O in the ECM or by EAO binding to membrane integrins. This suggests that the dynamic assembly and reorganization of the intracellular actin cytoskeleton may also be impaired or disrupted by EAO. Therefore, it is critical to understand how EAO influences the major downstream effectors of membrane integrins and how these effects contribute to actin cytoskeletal dysregulation in brain cells, thereby accelerating the development and progression of AD.

Currently, there is no proven method to stop or slow the progression of AD. Although anti-Aβ antibodies such as lecanemab and donanemab have attracted some attention, these therapies have not reversed existing neuronal damage or cured AD [21]. Therefore, the development of disease-modifying therapies is urgently needed. Given the critical role of Aβ42O pathology in the pathogenesis and progression of AD, as well as the occurrence of neurite impairment and actin cytoskeletal abnormalities in neural cells prior to cell death, this study systematically investigated the major intracellular effectors and regulatory pathways involved in the response to EAO-integrin binding. We used the neuronal cell line SH-SY5Y, oligodendrocyte cell line MO3.13, and AD model mice to examine the dynamic regulatory potential of EAO-associated cellular target(s) involved in these responses on the neuronal actin cytoskeleton under both in vitro and in vivo conditions. This study aims to clarify and define the nature of EAO-induced damage to the neuronal actin cytoskeleton and to contribute to the development of therapeutic approaches for AD.

2. Materials and Methods

2.1. Aβ42O and Other Protein/Peptide Agents

Human Aβ42 protein was purchased from Dalian Meilun Biological Co., Ltd. (MB10425, Dalian, China), with a purity of over 96%. Aβ42 was dissolved in dimethyl sulfoxide (DMSO) (Sangon Biotech, Shanghai, China) to prepare a solution of Aβ42 monomers (Aβ42Ms) at a concentration of 100–200 μM. This solution was stored at −80 °C until use. Based on the homogeneity characteristics of the Aβ42O prepared under different conditions in our preliminary experiments, the Aβ42O solution was prepared by incubating the Aβ42M solution at 37 °C for 3 h as described previously [22]. Each batch of Aβ42 oligomer preparations was subjected to electron microscopy and thioflavin T (ThT) validation, as previously described [22], to ensure consistency across batches. For ThT assay, the fluorescence intensity of the Aβ42O (5.0 μM) is typically 230 ± 20 corresponding to the Aβ42O state (spherical or ellipsoidal, with a diameter of approximately 3–12 nm) observed in electron microscopy and a molecular weight of 12–39 kD, as previously reported [22].

AF488-conjugated anti-Aβ42 antibody (Alexa Fluor^® 488) (green) and anti-ROCK2 (Rho-associated coiled-coil-containing protein kinase 2) antibody were purchased from Santa Cruz Biotechnology Co., Ltd. (sc-28365/D-11, Shanghai, China). Anti-human β1 integrin antibody, Cy5-/FITC-conjugated goat anti-mouse/rabbit IgG H&L (red)/(green), anti-granzyme B (GzmB) antibody, goat anti-mouse/rabbit IgG H&L/HRP, and mouse/rabbit IgG were purchased from Biosynthesis Biotechnology Co., Ltd. (bs-0486R/bs-0295G-Cy5/bs-0295G-FITC/bs-0296G-Cy5/bsm-60779R/bs-0296G-HRP/bs-0295G-HRP/bs-30102P/bsm-62714R, Beijing, China). Anti-RhoA antibody was purchased from Proteintech Group Co., Ltd. (10749-1-AP, Wuhan, China). AIIB2 (specific β1 integrin function-blocking antibody) was purchased from Developmental Studies Hybridoma Bank (AB_528306, Tianjin, China). Anti-Ezrin (pThr567)/-Radixin (pThr564)/-Moesin (pThr558) antibody, collectively referred to as anti-phosphorylated ERM (p-ERM) antibody, was purchased from Abcam Co., Ltd. (ab76247, Beijing, China). Anti-ERM antibody was purchased from Cell Signaling Technology Co., Ltd. (#3142, Shanghai, China). Trypsin was purchased from GIBCO Co., Ltd. (25200114, Shanghai, China). Protein A/G agarose beads was purchased from Beyotime Biotechnology Co., Ltd. (P2055, Shanghai, China). Prestained protein markers were purchased from Servicebio Technology Co., Ltd. (G2083/G2087, Wuhan, China) and Sikejie Biotechnology Co., Ltd. (EC1020, Shandong, China). TRITC-phalloidin and human serum albumin (HSA) were purchased from Solarbio Science & Technology Co., Ltd. (CA1610/A8230, Beijing, China).

2.2. Chemicals and Reagents

Dulbecco’s modified eagle medium (DMEM), fetal bovine serum (FBS), and penicillin–streptomycin solution were purchased from GIBCO (11965092/16140071/15140148, Shanghai, China). EGCG [(-)-Epigallocatechin gallate] were purchased from Shanghai Yuanye Biochemical Technology Co., Ltd. (B20106, Shanghai, China). DAPI (4′,6-diamidino-2-phenylindole) solution was purchased from Beijing Biosynthesis Biotechnology Co., Ltd. (C02-04002, Beijing, China). ROCK inhibitor Y-27632 [(+)-(R)-trans-4-(1-aminoethyl)-N-(4-pyridyl) cyclohexanecarboxamide dihydrochloride monohydrate] was purchased from Sikejie Biotechnology Co., Ltd. (SJ-MX1027A, Jinan, China). Paraformaldehyde solution (4%) were purchased from Solarbio Science & Technology Co., Ltd. (P1110, Beijing, China). Blocking buffer, BeyoECL Plus kit and methylthiazolyldiphenyl-tetrazolium bromide (MTT) were purchased from Beyotime Biotechnology Co., Ltd. (P0102/P0018M/ST316, Shanghai, China). GENMED cell ROCK2 kinase activity photometric quantitative assay kit was purchased from GENMED SCIENTIFICS Co., Ltd. (GMS50184.1, Shenzhen, China). Native lysis buffer, RIPA lysis buffer, protein phosphatase inhibitor, and protease inhibitor mixture was purchased from Solarbio Science & Technology Co., Ltd. (R0030/R0020/P1260/P6730, Beijing, China). Stripping Buffer was purchased from Dalian Meilun Biological Co., Ltd. (MA0189-2, Dalian, China). BCA protein quantitative assay kit was purchased from BioTeke Corporation (PP1001, Shanghai, China). All other chemicals were local products of analytical grade.

2.3. Cell Culture

Human neuroblastoma cell line SH-SY5Y (neuronal cell line) (100158, BeNa Culture Collection, Beijing, China) and human oligodendrocyte cell line MO3.13 (glial cell line) (BFN607200612, Qingqi Biotechnology Development Co., Ltd., Shanghai, China) provided by the Dalian Stem Cell and Precision Medicine Innovation Institute were used in this study as model cells. Unless otherwise stated, the cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) containing 10% fetal bovine serum (FBS) and 100 units/mL penicillin–100 µg/mL streptomycin in a humidified atmosphere with 5% CO₂ at 37 °C (standard conditions) to reach a specified confluence or for a specified period of time. All experiments were performed between the 3rd and 6th passages.

2.4. Immunofluorescence (IF) Assay

Glass coverslips were placed in 48-well plates, and SH-SY5Y/MO3.13 cells were seeded at densities of approximately 3.0 × 10⁴/2.0 × 10⁵ cells per well in 200 µL of DMEM, followed by incubation under standard conditions (37 °C, 5% CO₂) for 24 h to allow cell attachment. Subsequently, the cells were cultured in several DMEM formulations as follows. After aspiration of the original medium, fresh DMEM containing Aβ42O (final concentration: 0.2 μM) was added to each well, with or without the optional addition of EGCG (final concentration: 50.0 μM), AIIB2 (final concentration: 0.1 μg/mL), or Y-27632 (final concentration: 10.0 μM). Groups cultured with fresh DMEM alone or fresh DMEM containing only EGCG, AIIB2, or Y-27632 served as controls for the Aβ42O-treated groups. Under the same atmospheric conditions, the cells were cultured for an additional 24 h. As required, the cells were fixed with 4% paraformaldehyde and selectively permeabilized with 0.1% Triton X-100 (200 μL/well) for 10 min. After three washes with phosphate-buffered saline (PBS), the cells were blocked for 30–60 min at room temperature (23 ± 2 °C) using blocking buffer (200 μL/well). Following PBS washing, the cells were incubated with the indicated primary antibodies, followed by Cy5- or FITC-conjugated goat anti-mouse/rabbit IgG secondary antibodies. F-actin was visualized using TRITC-conjugated phalloidin (red) in the relevant groups, and nuclei were counterstained with DAPI. All fluorescence-labeling procedures were performed in the dark. Finally, the coverslips were imaged using a laser-scanning confocal microscope (Zeiss LSM900 confocal microscope, Carl Zeiss Microscopy GmbH, Jena, Germany). Each experiment was repeated using three independent batches of Aβ42O and/or cells.

Colocalization of target proteins was defined as the spatial overlap of their fluorescence signals within the same pixel. To quantitatively assess colocalization, dual-channel confocal image stacks were obtained and analyzed using the Manders colocalization method, including calculation of the Manders overlap coefficient (MOC) and colocalization coefficients (M1 and M2), as previously described [17]. The MOC metric (range: 0–1.0) quantitatively reflects the degree of spatial overlap between the two fluorescent signals, whereas M1 and M2 indicate the percentage of colocalized pixels for each fluorescent channel relative to the total number of pixels, respectively. For experiments examining the subcellular localization of proteins of interest, fluorescence signals were analyzed to determine intracellular distribution within the cells. All data were obtained from quantitative analyses of six biological replicates (n = 6).

2.5. Co-Immunoprecipitation (Co-IP) Assay of ROCK2 and RhoA

The extent of RhoA binding to ROCK2 under different treatment conditions (EAO exposure with or without EGCG or AIIB2) was quantitatively analyzed using a conventional Co-IP assay with Protein A/G beads and anti-RhoA and anti-ROCK2 antibodies, following previously described procedures [23] and our prior optimization. Briefly, SH-SY5Y/MO3.13 cells were seeded at densities of approximately 1.2 × 10⁷/1.8 × 10⁷ per 100 mm dish (with pre-plated glass coverslips) in 8.0 mL DMEM and cultured under standard conditions (37 °C, 5% CO₂) for 24 h. The medium was then aspirated, and fresh DMEM-containing Aβ42O (final concentration: 0.2 μM) was added to each dish, with or without the optional addition of EGCG (final concentration: 50.0 μM) or AIIB2 (final concentration: 0.1 μg/mL). Cells cultured with fresh DMEM alone or fresh DMEM containing only EGCG or AIIB2 served as controls for the Aβ42O-treated groups. After 24 h of culture under standard conditions, the dishes were transferred to a cold room (4 °C) for subsequent processing. After removal of the culture medium, and the cells were washed twice with 8.0 mL ice-cold PBS (10 mM, pH 7.4). Each dish was then treated with 1.0 mL precooled native lysis buffer supplemented with protease and phosphatase inhibitors. The cells were gently mixed by repeated pipetting and left on ice for 10 min to facilitate efficient lysis while preserving intracellular protein interactions. The lysates were transferred into precooled 1.5 mL microcentrifuge tubes and clarified by centrifugation at 13,680× g for 10 min at 4 °C. The resulting supernatants were collected as whole-cell lysate (WCL) samples and used as input samples for subsequent IP and Co-IP assays. To verify the specificity of the anti-ROCK2 antibody and its binding efficiency to Protein A/G agarose beads, an IP assay targeting ROCK2 was initially performed to ensure the reliability of subsequent Co-IP analyses.

WCL samples from all groups (except the control IgG group) were divided into two equal portions: one for ROCK2 (bait protein) IP and the other for ROCK2-RhoA (prey protein) Co-IP assays. For the IP procedure, 500 μL of WCL was incubated with 3.0–4.0 μL anti-ROCK2 antibody (1.0 μg total antibody) or 1.0 μL control IgG in a precooled 1.5 mL tubes and continuously rotated at 4 °C for 12 h. Subsequently, Protein A/G beads obtained from 50 μL of Protein A/G slurry were added to the antigen–antibody complexes and incubated with gentle rotation at 4 °C for 4–6 h. Following immune complex formation, the samples were subjected to low-speed centrifugation (210× g, 1 min, 4 °C), and the precipitated beads were collected. Finally, the input, IP/Co-IP, and corresponding control samples (all derived from equal amounts of WCL) were resolved on 8%/10% SDS–PAGE gels. ROCK2 (bait protein) in IP samples and RhoA (prey protein) in Co-IP samples were subsequently detected by conventional Western blotting using anti-ROCK2 and anti-RhoA antibodies, respectively, with an unrelated mouse and rabbit IgG serving as negative controls for IP and Co-IP, respectively. Following incubation with horseradish peroxidase (HRP)-conjugated secondary antibodies, all immunoreactive bands were visualized using enhanced chemiluminescence (ECL). Experimental data were normalized against corresponding controls derived from untreated control cells.

The grayscale values of RhoA Western blot bands were semiquantitatively analyzed using ImageJ 1.52v software. Relative levels of RhoA bound to ROCK2 in each Co-IP group were expressed as the ratio of the 10-fold mean intensity of the RhoA band in the Co-IP samples to the mean intensity of the corresponding band in the input samples. All experiments were repeated using three independent batches of Aβ42O and/or cells.

2.6. Western Blot Assay for the Proteins of Interest

SH-SY5Y/MO3.13 cells were seeded in six-well plates at densities of 1.25 × 10⁶/4.0 × 10⁶ cells per well in 2.0 mL of DMEM containing 10% FBS and cultured under standard conditions (37 °C, 5% CO₂) for 24 h to allow cell attachment. After aspiration of the medium, fresh DMEM containing Aβ42O (final concentration: 0.2 μM) was added to each well, with or without the optional addition of EGCG (final concentration: 50.0 μM), AIIB2 (final concentration: 0.1 μg/mL), or Y-27632 (final concentration: 10.0 μM), followed by culture for 24 h. After washing with PBS (10 mM, pH 7.4), the cells were lysed using an appropriate volume of ice-cold RIPA lysis buffer supplemented with protease and phosphatase inhibitors, followed by incubation on ice for 5 min and centrifugation at 13,680× g for 10 min at 4 °C.

For each lysed sample, proteins of interest and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were analyzed by Western blotting. Briefly, lysed samples were separated on 8%/10% SDS–PAGE gels and analyzed through a conventional Western blot assay. Target proteins were probed using corresponding primary antibodies and HRP-conjugated secondary antibodies. Immunoreactive proteins were visualized using ECL. The grayscale values of target protein bands were quantified using ImageJ software.

To remove antibodies from used Western blot membranes, the ECL-developed membrane was gently shaken in 10.0–15.0 mL of PBST (PBS containing 0.1% Tween-20) washing buffer for 10 min. Subsequently, the membrane was incubated at room temperature (23 ± 2 °C) in 10.0–15.0 mL Western blot membrane regeneration solution for 20 min with gentle shaking to strip the antibodies from the membrane. The membrane was then washed again with PBST washing buffer for 5 min with gentle shaking. This washing step was repeated three times to complete antibody removal. The membrane was subsequently prepared for the next target protein Western blot analysis.

Relative protein levels in each sample were normalized against GAPDH. The percentage of the protein of interest in each group was calculated by dividing its mean relative level by the mean relative level in the control group and multiplying by 100%. All experiments were repeated using three independent batches of Aβ42O and/or cells.

2.7. ROCK2 Activity Assay

SH-SY5Y/MO3.13 cells (1.2 × 10⁷/1.8 × 10⁷ per dish) were cultured in DMEM containing Aβ42O (final concentration: 0.2 μM) with or without EGCG (final concentration: 50.0 μM), AIIB2 (final concentration: 0.1 μg/mL), or Y-27632 (final concentration: 10.0 μM) for 24 h, as described above (see Co-IP section), followed by measurement of intracellular ROCK2 activity using a ROCK2 activity assay kit according to the manufacturer’s instructions. Briefly, after washing with cleaning buffer (3.0 mL per dish), the cells were harvested by centrifugation (4 °C, 300× g, 5 min), resuspended in an appropriate volume of precooled lysis buffer, vortexed vigorously for 15 s, and incubated on ice for 30 min. Each lysate was transferred to a precooled 1.5 mL tube and clarified by centrifugation at 13,680× g for 5 min at 4 °C. The resulting supernatants were collected as WCL samples for subsequent protein concentration determination and ROCK2 activity assays.

The reaction solution was added to 96-well plates (190 μL per well) and incubated at 30 °C for 3 min. Each lysis supernatant (10 μL per well) was then added to the reaction solution, mixed thoroughly, and the absorbance at 340 nm was measured at 0 and 5 min using a microplate reader. The protein content of each sample was quantified using a bicinchoninic acid (BCA) protein assay kit according to the manufacturer’s instructions. All measurements were performed in triplicate and repeated using six independent batches of Aβ42O and/or cells.

2.8. Cell Viability Assay by MTT Assay

SH-SY5Y/MO3.13 cells (3.0 × 10⁴/9.0 × 10⁴ per well) were cultured in DMEM containing Aβ42O (final concentration: 0.2 μM) with or without EGCG (final concentration: 50.0 μM), AIIB2 (final concentration: 0.1 μg/mL), or Y-27632 (final concentration: 10.0 μM) for 24 h, as described above (see Western blot section), followed by measurement of cell viabilities in each group were determined using the conventional MTT method, as described previously [24]. All measurements were performed in triplicate and repeated using six independent batches of Aβ42O and/or cells.

2.9. Determination of p-ERM Levels by Indirect Enzyme-Linked Immunosorbent Assay (ELISA)

SH-SY5Y/MO3.13 cells were seeded into six-well plates at densities of approximately 1.25 × 10⁶/4.0 × 10⁶ cells per well in 2.0 mL of DMEM containing 10% FBS and cultured and lysed under different conditions as described above (see Western blot assay). Protein concentrations of each lysate supernatant were determined using the BCA method with HSA as the standard. Each lysate sample (100 µL, 20.0 µg/mL) and HSA standard were coated onto 96-well plates at 4 °C for 12 h, followed by centrifugation at 1000 rpm for 5 min, after which the coating solution was discarded. All coated samples were subsequently analyzed by conventional indirect ELISA at 37 °C for 1.5 h using anti-p-ERM antibody and goat anti-rabbit IgG H&L/HRP antibody. Finally, the absorbance of each well was measured at 450 nm using a microplate reader.

The relative p-ERM level of each experimental sample was calculated by dividing the mean absorbance value of the experimental sample-coated wells by that of the control sample-coated wells and multiplying by 100%. Measurements were performed in triplicate and repeated across six independent batches of Aβ42O and/or cells.

2.10. Molecular Simulation and Docking

The 3D structures of RhoA (PDB ID: 9AX5), ROCK2 (active site)-Y-27632 complex (PDB ID: 2H9V), ROCK1 (RBD)-RhoA complex (PDB ID: 1S1C), and ROCK2 (UniProt ID: 075116) were downloaded from the Protein Data Bank (http://www.rcsb.org, accessed on 25 September 2025) and UniProt Knowledgebase (https://www.uniprot.org/uniprotkb, accessed on 25 September 2025), respectively. Based on the crystal structures of the RhoA-iso1, the 3D model of RhoA_iso3 (NCBI ID: NP_001655.1) was built using the program AutoDock (Version No. 4.2.6) (https://autodock.scripps.edu/download-autodock4/ accessed on 25 September 2025), and optimized by energy minimization and molecular dynamics simulation as described previously [17]. Molecular docking of RhoA with ROCK2 (RBD) was performed as described previously [17], and compared with the model of ROCK1 (RBD)-RhoA complex model (PDB ID: 1S1C). The structural integrities of these molecular simulation and docking models was visually examined through Discovery studio Visualizer (Version No. 3.1) (http://accelrys.com/products/discoverystudio/visualization-download.php accessed on 25 September 2025), and finally these images were obtained.

2.11. Mouse Experiments

Nine-month-old male APP/PS1 transgenic mice (AD model mice, positive control mice) (SPF grade, 30.0 ± 2.5 g) and age-matched wild-type male C57BL/6J mice (negative control mice) (SPF grade, 30.0 ± 2.5 g) (SPF Biotechnology Co., Ltd., Beijing, China) were used in this study. All mice were housed and experimentally manipulated as previously described [25]. All animal experimental procedures were conducted in strict accordance with the relevant regulations governing animal experimentation and were approved by the Institutional Animal Care and Use Committees (IACUC) of SPF Biotechnology Co., Ltd. and Servicebio Technology Co., Ltd. (Wuhan, China) (approval numbers: AWE2024102101 and 2025238).

AD mice were randomly divided into four groups: EGCG-treated (n = 6), AIIB2-treated (n = 3), Y-27632-treated (n = 3), and positive control (n = 6) groups. Mice in the negative control group (n = 6) and positive AD control group received oral administration of normal saline, whereas mice in the EGCG-, AIIB2-, and Y-27632-treated groups received oral administration or intraperitoneal injections of EGCG, AIIB2, or Y-27632 at doses of 1.0, 0.1, and 10.0 mg/kg body weight, respectively. Treatments were administered once daily for EGCG or every other day for AIIB2 and Y-27632 over 14 consecutive days at the doses described above. Twenty-four hours after the final administration, the mice were euthanized by carbon dioxide asphyxiation, and brain tissues were collected.

(1) Measurement of GzmB levels and ROCK2 activities in AD mice

Brains obtained from EGCG-treated AD mice, as well as positive and negative control mice (n = 3 per group), were divided into two equal parts and placed in 4.0 mL ice-cold RIPA lysis buffer containing protease and phosphatase inhibitors. The samples were homogenized using an ST-M50 tissue homogenizer with steel beads and then incubated on ice for 1 h. Brain tissue lysates were centrifuged at 13,680× g for 10 min, and the resulting supernatants were collected. GzmB levels and ROCK2 activities in each brain sample were measured by Western blotting and using the ROCK2 activity assay kit, respectively, as described above (see Section 2.6 and Section 2.7, respectively). Protein concentrations in each brain sample were determined using the BCA method with HSA as the standard.

(2) Hematoxylin–eosin (HE) and immunohistochemistry (IHC) staining of brain sections from AD and control mice

Brains obtained from EGCG-, AIIB2-, and Y-27632-treated AD mice, as well as positive and negative control mice (n = 3 per group), were fixed in 4% (v/v) paraformaldehyde for 48 h and cryoprotected in graded sucrose solutions (15–30%). Subsequently, brain sections were analyzed using conventional HE and IHC staining, as previously described [24]. Microscopic images of hippocampal sections after staining were acquired using an imaging system (Nikon Eclipse E100, Shanghai, China). For quantitative evaluation, bilateral hippocampal areas from each section were selected for analysis of cell morphology and Aβ deposition spots/plaque. The Aβ-positive area was quantified using Image-Pro Plus 6.0 software.

The Aβ deposition spots/plaque load within the selected hippocampal area was calculated by dividing the mean Aβ-positive area of the selected region by the total area of that region. The relative Aβ plaque load in the positive control group was defined as 100%, and the relative Aβ plaque load in each experimental group was calculated by dividing its Aβ plaque load by that of the positive control group and multiplying by 100%.

2.12. Statistical Analysis

For cell culture data, data were obtained from at least three independent batches of Aβ42O preparations and/or cells. Data obtained from IF, Co-IP, and Western blot assays were presented as means ± SD (standard deviation), while data obtained from the absorbance measurement were expressed as the mean ± SEM (standard error of the mean).

In mouse experiments, at least three mice per group were used to measure GzmB levels and ROCK2 activities, as well as for HE and IHC staining; furthermore, both of the latter two assays included samples from both the left and right halves. Cellular measurements in each experimental group were performed in a random order. During image capture and analysis, the investigators were blinded to experimental group assignments. Normality and homogeneity of variance tests were performed using the GraphPad Prism 11 statistical analysis software. For data that met normality of distribution and homogeneity of variance, intergroup comparisons performed using one-way analysis of variance (ANOVA), followed by Šídák’s multiple comparisons test; p < 0.05 was considered statistically significant.

3. Results

3.1. Binding of EAO to Membrane Integrins in Both Neuronal and Glial Cell Cultures

To examine the effects of EAO binding to integrins on integrin-associated intracellular effectors in neuronal and glial cells, the colocalization of EAO and membrane integrins in SH-SY5Y and MO3.13 cells in the presence of EAO with or without EGCG or AIIB2 was assessed by double-label IF staining using specific anti-Aβ42 (green) and anti-integrin (red) antibodies, respectively. EGCG is a naturally occurring amphiphilic polyphenol (Both EGCG and Aβ42 are chemically amphiphilic molecules). Previous studies have reported that EGCG can bind to EAO particles and render them inert by blocking or masking the exposed hydrophobic clusters and N-terminal polar fragments on EAO particles, thereby preventing EAO from targeting and binding to neural cells [24]. AIIB2 is an integrin function-blocking antibody; therefore, its presence in the ECM blocks integrin signaling.

As shown in Figure 1, in both cell systems, the extent of EAO-integrin colocalization (yellow arrows), regarded as the level of EAO bound to integrins, was significantly greater in the EAO-treated group (second column from the left in Figure 1) than in the groups co-treated with EAO and EGCG or AIIB2. These results demonstrated that the binding of EGCG to EAO or AIIB2 to integrins considerably decreased the capacity of EAO to approach and bind to integrins. Simultaneously, EGCG also reduced the binding of cell-associated but integrin-unbound EAO to the cells (green arrows), consistent with previous findings, although the experimental conditions varied slightly [24]. In contrast, AIIB2 had no effect on EAO binding to plasma membranes (Figure 1, first column from the right). These results suggest that, in addition to its presence in the ECM, EAO has the potential to interact with neural cells through multiple mechanisms, a substantial proportion of which involve binding to membrane integrins. This might be related to the primary structure of the Aβ42 chain, particularly the RHDS motif, and the integrated spatial conformation of the Aβ42O unit [5,17]. More importantly, EAO binding to integrins might disrupt integrin function, as EAO and AIIB2 appeared to compete for integrin binding, as evidenced by the significant difference in ratio of yellow to green arrows between the first and third columns from the right in Figure 1. This result suggests that, as an unconventional ligand for integrins, EAO binding to integrins might trigger abnormal integrin signaling and further implies that EAO might compete with physiological integrin ligands of integrins for binding to integrins because previous molecular docking analyses have shown that EAO can bind to the extracellular domain of integrins and occupy the binding site of their physiological ligands [17]. Furthermore, it was noted that while levels of unbound EAO were comparable between the two cell systems, levels of both integrin-bound and cell-associated EAO were lower in MO3.13 cells than in SH-SY5Y cells (Figure 1); although there are currently no reports on membrane integrin levels in these two cell lines, this suggests that EAO might interact more extensively with neurons than with glial cells, indicating potential differences in the binding capacity of EAO particles between these neural cell types. Nevertheless, under physiological conditions, affected glial cells may subsequently exert adverse effects on neurons.

Figure 1.

Analyses of EAO interaction or binding to membrane integrins in SH-SY5Y and MO3.13 cell systems in the presence of EAO with or without EGCG/AIIB2 for 24 h. Representative images of double-label immunofluorescence (IF) staining for Aβ42 (green) and integrins (red) using AF488-labeled anti-Aβ42 antibody (green) and anti-human β1 integrin antibody with Cy5-conjugated secondary antibody (red). Green arrows: cell-associated but integrin-unbound Aβ42O spots, including membrane-bound but integrin-unbound EAO, as well as EAO that may already be internalized; yellow arrows: integrin-bound Aβ42O spots; white arrows: cell-unassociated Aβ42O spots. The number of each type of arrow was shown at the bottom of each image. The corresponding count for each type of arrow was shown at the bottom of each merged image. EAO-only images included Aβ42O (green) and nuclei (blue). Merged images included integrin (red), Aβ42O (green), and nuclei (blue).

As shown in Table 1, the degree of EAO-integrin colocalization in each group in Figure 1 was further quantified using the MOC and the colocalization coefficients M1 (for integrins) and M2 (for Aβ42). In both cell systems, only the EAO-treated group exhibited MOC values greater than 0.6, and the corresponding M1 and M2 values were also significantly higher than those of the other groups (b and f in Table 1), indicating significant colocalization between EAO and membrane integrins only in the EAO-treated group. These results indicate a strong interaction between EAO and integrins, which could be blocked or inhibited by EGCG or AIIB2 through binding to EAO or integrins, respectively. Furthermore, in both cell systems, the M2 values of the groups co-treated with EAO and AIIB2 were lower (d and h in Table 1) than those of the groups co-treated with EAO and EGCG (c and g in Table 1), suggesting that AIIB2 inhibited or blocked EAO–integrin interaction more effectively than EGCG.

Table 1.

Values * of MOC, M1, and M2 parameters for Figure 1.

Cell	Group		MOC	M1 (Integrin)	M2 (EAO)
SH-SY5Y	(a)	Control	0.056 ± 0.008	0.010 ± 0.002	0.020 ± 0.004
	(b)	EAO	0.692 ± 0.010	0.256 ± 0.009	0.486 ± 0.011
	(c)	EAO&EGCG	0.211 ± 0.007	0.092 ± 0.009	0.141 ± 0.005
	(d)	EAO&AIIB2	0.140 ± 0.005	0.088 ± 0.004	0.126 ± 0.007
MO3.13	(e)	Control	0.200 ± 0.009	0.011 ± 0.004	0.036 ± 0.005
	(f)	EAO	0.685 ± 0.016	0.264 ± 0.009	0.474 ± 0.006
	(g)	EAO&EGCG	0.141 ± 0.010	0.012 ± 0.004	0.106 ± 0.007
	(h)	EAO&AIIB2	0.117 ± 0.011	0.011 ± 0.006	0.088 ± 0.008

Note: * These data were calculated based on the fluorescence signals from the corresponding merged images of double-labelling IF for the colocalization between EAO and membrane integrins shown in Figure 1 The MOC values for merged images range from 0 to 1.0, with the values between 0 and 0.6 indicating very low or no colocalization and values between 0.6 and 1.0 indicating substantial colocalization. M1: fraction of integrins overlapping EAO; M2: fraction of EAO overlapping integrin. The data were calculated based on all images obtained from at least three repeats with different batches of cells. All data were the mean ± SD.

3.2. EAO-Induced an Increase in the Interaction Between RhoA and ROCK2

Distinct conformational changes in integrins can induce different intracellular responses, among which the response of RhoA, a major member of the Rho family of small GTPases with a molecular weight of 22 kD, is directly involved in regulation of actin cytoskeleton reorganization [26]. Since RhoA functions as an activator of ROCK2 (EC 2.7.11.1) in brain cells, EAO binding to integrins would be expected to affect ROCK2 activity. ROCK2 is a protein kinase that plays a particularly important role in regulating neurite outgrowth and synapse formation. In brain cells, ROCK2 can be activated through two mechanisms: reversible activation induced by RhoA binding and irreversible activation resulting from granzyme B (GzmB)-mediated cleavage of the ROCK2 molecule [27].

The large ROCK2 molecule (molecular weight of approximately 160 kD) consists of an N-terminal kinase domain, a coiled-coil region containing a specific RhoA-binding domain (RBD), and a C-terminal domain containing the GzmB catalytic site (a and b in Figure 2A) [28]. To illustrate the interaction between ROCK2 and RhoA, molecular docking analysis of the ROCK2’s RBD and RhoA was performed (c in Figure 2A) and compared with the three-dimensional (3D) structure of the ROCK1 RBD-RhoA reference complex (PDB ID: 1S1C) (d in Figure 2A). As shown in Figure 2A, the two structures were highly similar (c and d in Figure 2A).

Figure 2.

Analysis of the interaction between RhoA and ROCK2 in SH-SY5Y and MO3.13 cells in SH-SY5Y and MO3.13 cells in the presence of EAO with or without EGCG/AIIB2 for 24 h. (A) Three-dimensional (3D) structures of ROCK2 and RhoA and molecular docking models of RhoA binding to the RhoA-binding domain (RBD) of ROCK2. (a): 3D structure of ROCK2 molecule (UniProt ID: 075116). (b): ROCK2 kinase domain and its active site, showing the bound inhibitor Y-27632 (black) (PDB ID: 2H9V). (c): Docking model of ROCK2 (residues 1010–1045, RBD) (blue) and the RhoA molecule (yellow). (d): 3D structure of the ROCK1 (residues 945–1014, RBD) (blue)–RhoA (yellow) complex (PDB ID: 1S1C). Double-headed arrows indicate the interaction between ROCK2 and RhoA. (B) Representative images of double-label immunofluorescence (IF) staining for RhoA (green) and ROCK2 (red) in SH-SY5Y(a) and MO3.13 (b) cells using anti-RhoA antibody with FITC-conjugated secondary antibody (green) and anti-ROCK2 antibody with Cy5-conjugated secondary antibody (red). Merged images include RhoA (green), ROCK2 (red), and colocalized RhoA and ROCK2 (yellow). Corresponding nuclei (blue) are shown in separate images above. Ctrl: control. Scale bar: 20 μm. (C) Representative Western blot results of ROCK2 and RhoA Co-IP samples prepared from SH-SY5Y and MO3.13 cell lysates. (a,b): SH-SY5Y cells; (c,d): MO3.13 cells. (a,c): Detection of ROCK2 (bait) from ROCK2 IP using an anti-ROCK2 antibody; (b,d): Detection of RhoA (prey) from the Co-IP of ROCK2 and RhoA using anti-RhoA antibody. Input: input samples; IP/Co-IP: IP/Co-IP beads. IgG: unrelated IgG serving as negative controls. Ctrl: control. kD: kilodaltons. (D) Relative levels of RhoA bound to ROCK2 based on normalized ratios of 10 × Co-IP to input, after semiquantitative grayscale analysis of RhoA western blot bands obtained from the ROCK2-RhoA Co-IP assays shown in Figure 2C. The ratio of the 10-fold intensity of the Co-IP bands to the intensity of the corresponding input bands was used as a measure of the relative levels of RhoA bound to ROCK2. Ctrl: control. Between-group comparisons were performed using one-way ANOVA, followed by Šídák’s multiple comparisons test. All data are presented as mean ± SD from three independent experiments. Symbols *, #, and § indicate significant differences between the experimental and control groups, between EAO-alone and EAO plus EGCG/AIIB2 groups, and between the EAO plus EGCG/AIIB2 and EGCG/AIIB2-alone groups, respectively. */# p < 0.05, ** p < 0.01, §§§/### p < 0.001, and ****/§§§§ p < 0.0001.

ROCK2 functions as a downstream effector of RhoA. The small molecule RhoA activates ROCK2 by competitively binding to its RBD and thereby removing ROCK2’s self-inhibition [28]. RhoA-mediated activation of ROCK2 represents a form of allosteric activation that depends on successful interaction between RhoA and ROCK2. To determine the effects of EAO binding to integrins on the interaction between intracellular RhoA and ROCK2, subsequent experiments investigated the colocalization of RhoA and ROCK2 in SH-SY5Y and MO3.13 cells using double-label IF staining and quantitative analysis.

RhoA and ROCK2 were fluorescently labeled using specific anti-RhoA (green) and anti-ROCK2 (red) antibodies, respectively, and images of SH-SY5Y and MO3.13 cells are shown in Figure 2B. The extent of RhoA-ROCK2 colocalization in both cell types was significantly greater in the groups treated with EAO alone than in all other groups, including those co-treated with EAO and EGCG or AIIB2, whereas the extent of colocalization in the EGCG-alone group was nearly equivalent to that in the control group. These differences were quantitatively confirmed by the MOC values the colocalization coefficients M1 (for ROCK2) and M2 (for RhoA), as shown in Table 2.

Table 2.

Values * of MOC, M1, and M2 parameters for the colocalization of RhoA and ROCK2 in Figure 2B.

Cell	Group		MOC	M1(ROCK2) (Red)	M2(RhoA) (Green)
SH-SY5Y	(a)	Control	0.746 ± 0.009	0.648 ± 0.007	0.626 ± 0.005
	(b)	EAO	0.854 ± 0.009	0.808 ± 0.010	0.662 ± 0.007
	(c)	EAO&EGCG	0.759 ± 0.004	0.766 ± 0.003	0.645 ± 0.008
	(d)	EGCG	0.750 ± 0.005	0.608 ± 0.008	0.618 ± 0.007
	(e)	EAO&AIIB2	0.758 ± 0.009	0.751 ± 0.004	0.683 ± 0.004
	(f)	AIIB2	0.811 ± 0.009	0.702 ± 0.005	0.653 ± 0.007
MO3.13	(g)	Control	0.700 ± 0.006	0.602 ± 0.010	0.593 ± 0.008
	(h)	EAO	0.804 ± 0.009	0.858 ± 0.010	0.667 ± 0.005
	(i)	EAO&EGCG	0.725 ± 0.007	0.713 ± 0.010	0.617 ± 0.006
	(j)	EGCG	0.700 ± 0.008	0.676 ± 0.009	0.600 ± 0.009
	(k)	EAO&AIIB2	0.745 ± 0.004	0.802 ± 0.004	0.641 ± 0.003
	(l)	AIIB2	0.746 ± 0.006	0.700 ± 0.007	0.532 ± 0.002

Note: * These data were calculated based on the fluorescence signals from the corresponding merged images of double-labeling IF staining for RhoA and ROCK2 shown in Figure 2B. The MOC values for merged images range from 0 to 1.0, with the values between 0 and 0.6 indicating very low or no colocalization and values between 0.6 and 1.0 indicating substantial colocalization. M1: fraction of ROCK2 overlapping RhoA; M2: fraction of RhoA overlapping ROCK2. For the M1 values, the differences between all EAO-treated groups and their control groups were statistically significant (p < 0.001 or p < 0.0001). The data were calculated based on all images obtained from at least three repeats with different batches of cells. All data were the mean ± SD.

The MOC values for all groups were greater than 0.6 (Table 2), indicating substantial RhoA-ROCK2 colocalization in all groups (a–l in Table 2). Higher M1 values in the EAO-treated groups (b, c, e, h, i, and k in Table 2), particularly in the EAO-alone groups (b and h in Table 2), confirmed increased RhoA-ROCK2 colocalization under these conditions (for their M1 values, p < 0.001 or p < 0.0001, vs. controls). In contrast, the M1 values in the EGCG-alone groups (d and j in Table 2) were similar to or lower than those in the control groups (a and g in Table 2), indicating that RhoA-ROCK2 colocalization in these groups was comparable to or slightly lower than that in controls. In addition, comprehensive analysis of the M1 and M2 values in the AIIB2-alone groups (f and l in Table 2) suggested that the extent of RhoA-ROCK2 colocalization in these groups was comparable to that in controls. Finally, combined analysis of the M1 and M2 values in Table 1 and Table 2 indicated that EAO-induced increase in intracellular RhoA-ROCK2 interactions was directly associated with EAO binding to membrane integrins. Accordingly, inhibition of EAO or integrins by EGCG or AIIB2 attenuated the EAO-induced increase in RhoA-ROCK2 interaction (Table 2).

To quantitatively evaluate the extent of direct RhoA binding to ROCK2, the relative amount of RhoA bound to ROCK2 was measured by Co-IP in the presence of EAO with or without EGCG or AIIB2, using ROCK2 and RhoA as the bait and prey proteins, respectively. Figure 2C demonstrates that ROCK2 and RhoA signals were detected only in samples incubated with anti-ROCK2 and anti-RhoA antibodies, whereas no signal was observed with control IgG (IgG bands in groups co-treated with EAO and EGCG/AIIB2 are not shown). Notably, ROCK2 IP showed two bands: a major band at 160 kD corresponding to full-length ROCK2 and a smaller band at 130 kD corresponding to truncated ROCK2 (a and c in Figure 2C) [29]. These results indicate that both mechanisms of ROCK2 activation described above were present in these neural cells (for cleavage-mediated ROCK2 activation, see the GzmB section below).

Quantitative analysis shown in Figure 2D demonstrated that relative changes in the amount of RhoA bound to ROCK2 were highly similar between the two cell systems. Compared to the control group, all EAO-treated groups showed increased levels of RhoA bound to ROCK2, whereas the EGCG-alone groups showed significantly reduced levels and the AIIB2-alone groups showed slightly reduced levels (* p < 0.05, ** p < 0.01, or **** p < 0.0001, vs. controls). These data clearly reflect the extent of direct interaction between RhoA and ROCK2.

In addition, the pattern of increased RhoA–ROCK2 binding in all groups treated with EAO or EAO plus EGCG/AIIB2 was highly similar to the pattern of increased EAO-integrin colocalizations (Table 1 and Figure 2D). These results suggest that EAO binding to integrins induced increased interactions between RhoA and ROCK2, which would consequently enhance RhoA-mediated ROCK2 activation. Furthermore, ROCK2–RhoA binding levels in the EGCG-alone groups decreased by approximately 50% compared with controls (Figure 2D). This result was inconsistent with the extent of RhoA-ROCK2 colocalization indicated by the M1 and M2 values in Table 2. This discrepancy might be related to the neuroprotective effects of EGCG, which might stabilize intracellular physiological conditions and the functional state of RhoA, thereby markedly reducing direct RhoA to ROCK2 interactions to levels lower than those observed in control groups (Figure 2D). These results suggest that although ROCK2–RhoA binding affinity is directly associated with RhoA levels, it is also influenced by the intracellular microenvironment and the functional state of RhoA. In contrast, colocalization between RhoA and ROCK2 did not necessarily indicate direct interaction between the two proteins. Furthermore, the data shown in Figure 2D indicate that levels of RhoA bound to ROCK2 were generally lower in MO3.13 cells than in SH-SY5Y cells. This suggests that the metabolic basis of RhoA-mediated ROCK2 activation might differ between these two neural cell types; however, the resulting physiological changes associated with this process might still comparable in response to the same extracellular stimulus, such as EAO. Finally, although AIIB2 blocked integrins signaling to some extent, it did not cause an increase in RhoA–ROCK2 binding (Figure 2D). This result indicates that EAO-mediated integrin binding and AIIB2-mediated integrin blockade induced distinct intracellular physiological responses.

Taken together, the results of double-label IF staining and Co-IP analyses suggested that EAO can induce a direct interaction between RhoA and ROCK2, which might enhance reversible ROCK2 activation by RhoA in neural cells.

3.3. EAO Increased GzmB Levels in Neuronal and Glial Cell Cultures and in the Brain of AD Mice

GzmB is a serine protease with a molecular weight of 32 kD that is expressed in both immune and non-immune cell types. GzmB expression is commonly associated with chronic inflammatory disorders [30,31]. GzmB has been detected in the cerebral cortex, hippocampus, and diencephalon of the mouse brain [32]. In the human brain, it is primarily detected in the hypothalamus, although lower levels have also been reported in several other regions [33] (https://www.proteinatlas.org/, accessed on 17 June 2025). Both brain-derived GzmB and exogenous GzmB capable of crossing the blood–brain barrier (BBB) may significantly contribute to local neuroinflammatory responses and the pathogenic mechanisms underlying neurodegenerative diseases [34]. In addition, the human body expresses two GzmB isoforms: isoform 1 (NP_001332940) and isoform 2 (NP_004122.2). Both isoforms possess the same active enzymatic domain; however, GzmB-isoform 2 is considered constitutively expressed because it lacks the predicted signal peptide present in GzmB-isoform 1. Numerous neurological disorders have been reported to positively correlate with elevated intracellular or extracellular GzmB levels in both cell cultures and brain tissues [30,35].

ROCK2 in neural cells has been identified as a specific substrate of GzmB [29]. GzmB catalyzes the cleavage of the ROCK2 molecule after Asp1131, resulting in release of the C-terminal domain. This cleavage induces a substantial conformational change in truncated ROCK2, exposing its active site and converting it into an active form of ROCK2. Therefore, GzmB-mediated activation of ROCK2 is irreversible and resembles proenzyme activation. The truncated ROCK2 molecule, with a molecular weight of approximately 130 kD, represents another active form of ROCK2 and exhibits constitutive kinase activity. This truncated form remains susceptible to inhibition by the ROCK2 inhibitor Y-27632. In addition, the 130-kD ROCK2 molecule can still interact with RhoA, although the binding affinity between the two proteins is markedly reduced [29].

Although considerable attention has recently been directed toward neuroinflammation in AD [36,37,38,39], the role of GzmB plays, particularly the relationship between intracellular GzmB and EAO, remains unclear. Although this relationship has not previously been reported, the 130-kD band observed in ROCK2 IP assays (a and c in Figure 2C) indicated the presence of GzmB activity in SH-SY5Y and MO3.13 cells. To investigate the relationship between EAO and GzmB and to evaluate the role of GzmB in cleavage-mediated ROCK2 activation in neural cells, GzmB levels in both cell systems were measured by Western blotting in the presence of EAO with or without EGCG/AIIB2/Y-27632.

The corresponding GzmB levels were quantified as shown in Figure 3C,D following analysis of immunoblot densities for each group of cultured cells and AD mice shown in Figure 3A and Figure 3B, respectively. The results shown in Figure 3C demonstrated that 0.2 μM EAO induced a significant increase in intracellular GzmB levels in both cell systems within 24 h. GzmB levels in the different EAO-treated groups were generally within a similar range, except for the groups co-treated with EAO and EGCG. Specifically, GzmB levels in groups treated with EAO-alone or EAO plus Y-27632/AIIB2 increased by approximately 40–49% relative to controls (**** p < 0.0001 vs. control), whereas those in groups co-treated with EAO and EGCG increased by approximately 20–26% (***p < 0.001 or ** p < 0.01 vs. control). In contrast, GzmB levels in groups treated with EGCG, AIIB2, or Y-27632 alone were comparable to those in control groups. These results indicate that EAO induced a significant increase in intracellular GzmB levels in neuronal and glial cells through an extracellular-to-intracellular cascade that appeared to be independent of integrin signaling, although both integrins and GzmB are associated with EAO. Furthermore, the significant differences in GzmB levels between the EAO-alone and EAO plus EGCG groups (# p < 0.05 or #### p < 0.0001) might be associated with both the binding of EGCG to EAO and the antioxidant and anti-neuroinflammatory properties of EGCG, which attenuated the cascade of events induced by EAO. It should also be noted that GzmB-mediated ROCK2 activation appeared to be unidirectional without feedback regulation, as GzmB levels in the EAO-alone and EAO plus Y-27632 groups were similar, whereas GzmB levels in the Y-27632-alone group were comparable to those in controls. These results are consistent with a proenzyme activation mechanism.

Figure 3.

Measurement of GzmB levels in SH-SY5Y and MO3.13 cells and AD mouse brains by Western blot analysis. (A,B) Representative Western blot images obtained from SH-SY5Y and MO3.13 cells (A) and AD mouse brains (B). kD: kilodaltons. (C,D) Relative GzmB levels in each group shown in Western blot images obtained from SH-SY5Y and MO3.13 cells (A) and AD mouse brains, determined by semiquantitative grayscale analysis, normalized to the corresponding to internal reference, GAPDH, and then expressed as a percentage relative to the normalized control group (Ctrl) or WT-mouse (negative control, nCtrl) groups. The GzmB level in the Ctrl or nCtrl group was defined as 100%. All data are presented as means ± SD from three independent experiments. Between-group comparisons were performed using one-way ANOVA, followed by Šídák’s multiple comparisons test. All data are presented as means ± SD from three independent experiments. Symbols *, #, and § indicate significant differences between the experimental and control groups, between the EAO-alone and EAO plus EGCG/AIIB2 groups, and between EAO plus EGCG/AIIB2 and EGCG/AIIB2-alone groups, respectively. # p < 0.05, **/§§ p < 0.01, ***/§§§ p < 0.001, and ****/####/§§§§ p < 0.0001.

In addition, the results shown in Figure 3C demonstrated that MO3.13 cells exhibited more pronounced changes in GzmB levels, whether increases or decreases over the 24 h treatment period than SH-SY5Y cells across all EAO-, EGCG-, and AIIB2-treated groups (a–j in Figure 3C). This phenomenon became more apparent with prolonged treatment duration. These findings suggest that GzmB expression may be more sensitive to modulation by extracellular factors in MO3.13 cells than in SH-SY5Y cells.

Because GzmB is present both intracellularly and extracellularly in brain tissue [30,35], extracellular GzmB may enter target cells through direct diffusion across the plasma membrane [40]. To further evaluate the effects of EAO on GzmB levels in the brain, GzmB levels in the brains from WT mice, AD mice, and EGCG-treated AD mice following 14 consecutive days of administration were quantitatively analyzed, as shown in Figure 3D. GzmB levels in brains from AD mice and EGCG-treated AD mice were both significantly higher than those in WT mice (approximately 152% and 128% of the WT levels, respectively) (** p < 0.01 and *** p < 0.001 vs. WT control), whereas GzmB levels in EGCG-treated AD mice were significantly lower than those in untreated AD mice (# p < 0.05 vs. AD control). Previous studies have demonstrated that EGCG possesses antioxidant and anti-neuroinflammatory properties capable of attenuating the neurotoxic effects of Aβ42 aggregates in vivo [24,41,42]. Therefore, these results were likely associated with the presence of Aβ42 aggregates in the brain and the neuroinflammatory responses they induce. Elevated GzmB levels in AD brains, whether brain-derived or blood-derived, supported the results shown in Figure 3C and were consistent with previous reports of increased GzmB levels in the central nervous system during other immune responses [43]. Overall, the effects of EAO or EGCG on GzmB levels were consistent between in vitro and in vivo experiments (Figure 3C,D).

Taken together, these results indicate that the presence of Aβ42O in the extracellular matrix increased GzmB levels under both in vitro and in vivo conditions, whereas EGCG attenuated this effect. However, the EAO-induced increase in intracellular GzmB levels appeared to occur independently of integrin signaling.

3.4. EAO Induced ROCK2 Hyperactivation in Neuronal and Glial Cell Cultures and in the Brain of AD Mice

ROCK2 is a downstream target of both RhoA and GzmB, and these two factors act on ROCK2 independently. Since ROCK2 is the predominant ROCK subtype in the central nervous system and is closely associated with the pathogenesis of AD, ROCK2 activity was considered to represent total ROCK activity in this study, although ROCK1 activity was not entirely ruled out. Given that the results above demonstrated that EAO induced increases in RhoA–ROCK2 bindings and GzmB levels, determining the relationship between EAO and ROCK2 is important for understanding both EAO-targeted pharmacology and the pathophysiology associated with EAO. Therefore, in the presence of EAO with or without EGCG, AIIB2, or Y-27632, the levels of the two forms of ROCK2 in SH-SY5Y and MO3.13 cells were quantified by Western blotting, and the ROCK2 activity in both cell types and AD mouse brains was evaluated using a ROCK2 activity assay kit. The results are shown in Figure 4.

Figure 4.

Measurement of ROCK2 protein levels and ROCK2 activity in SH-SY5Y and MO3.13 cells and AD mouse brains in the presence of EAO with or without EGCG/AIIB2 for 24 h by Western blot analysis and ROCK2 activity assay. (A) Representative Western blot images obtained from SH-SY5Y and MO3.13 cells. kD: kilodaltons. (B,C) Relative levels of the 130 kD (gray) and 160 kD (black) forms of ROCK2 in each group shown in (A), determined by semiquantitative grayscale analysis and normalized to the corresponding control groups. The ROCK2 level in each control group was defined as 100%. Ctrl: control. Between-group comparisons were performed using one-way ANOVA, followed by Šídák’s multiple comparisons test. All data are presented as means ± SD from three independent experiments. (D,E) Relative ROCK2 specific activity in SH-SY5Y and MO3.13 cells (D) and mouse brains (E), normalized to the corresponding control (Ctrl) or WT-mouse (negative control, nCtrl) groups. The specific activity of ROCK2 in the Ctrl or nCtrl groups was defined as 100%. Between-group comparisons were performed using one-way ANOVA. All data are presented as means ± SEM from three or six independent experiments. WCP: whole cell protein. BTP: brain tissue protein. Symbols *, #, and § indicate significant differences between the experimental and control groups, between the EAO-alone and EAO plus EGCG/AIIB2/Y-27632 groups, and between the EAO plus EGCG/AIIB2/Y-27632 and EGCG/AIIB2/Y-27632 alone groups, respectively. */# p < 0.05, **/§§ p < 0.01, ***/###/§§§ p < 0.001, and ****/####/§§§§ p < 0.0001.

As shown in Figure 4A, two ROCK2 bands with molecular weights of 130 kD and 160 kD, as shown were detected in all groups. The smaller band represented the truncated form of ROCK2 generated by GzmB-mediated proteolytic cleavage, corresponding to the constitutively active form of ROCK2; whereas the larger band represented full-length ROCK2, corresponding to the fraction of total ROCK2 protein whose activity is reversibly regulated by RhoA binding and dissociation. The relative levels of these two forms were subsequently quantified by grayscale analysis, with the results shown in Figure 4B,C for SH-SY5Y and MO3.13 cells, respectively. Overall, corresponding groups in the two cell types exhibited similar trends in the relative levels and alterations in the two ROCK2 forms. The levels of full-length ROCK2 (gray bars in Figure 4B,C) in all experimental groups did not differ significantly from those in the control groups within 24 h (p > 0.05 vs. controls). In contrast, levels of truncated ROCK2 (black bars in Figure 4B,C) were significantly elevated in these groups, ranging from approximately 112% to 120% of control levels (** p < 0.01 or *** p < 0.001 vs. controls). By comparison, the relative levels of both ROCK2 forms in the EAO plus EGCG groups remained largely unchanged over the 24 h period (p > 0.05 vs. controls), with the exception of truncated ROCK2 levels in SH-SY5Y cells (* p < 0.05 vs. controls). These findings, together with the results shown in Figure 3C, suggest that the increase in truncated ROCK2 resulted from elevated GzmB levels, although the amount of the corresponding full-length ROCK2 protein decreased only slightly (with no statistically significant difference compared to their respective controls). Total ROCK2 levels measured in EAO-treated groups were nearly identical to those in the control groups, suggesting that total intracellular ROCK2 levels remained relatively constant in both cell types. More importantly, the results shown in Figure 4B,C indicate that EAO induced cleavage of a portion of full-length ROCK2 into truncated ROCK2, which would inevitably alter ROCK2 activity. Furthermore, the stable levels of full-length ROCK2 and the relatively low levels of truncated ROCK2 observed in the EGCG-treated groups were consistent with the stable GzmB levels observed in these groups (Figure 3C), suggesting that both EAO binding inhibition and the anti-neuroinflammatory properties of EGCG attenuated the effects of EAO on GzmB and ROCK2.

As shown in Figure 4A, truncated ROCK2 accounted for only a small fraction of total ROCK2 in control SH-SY5Y and MO3.13 cells. This might reflect the physiological levels of truncated ROCK2 in neural cells in vivo. Whereas truncated ROCK2 remains constitutively active until degraded, full-length ROCK2 includes both self-inhibited (RhoA-unbound) and active (RhoA-bound) forms. Therefore, under both in vitro and in vivo conditions, total ROCK2 activity should represent the combined activity of these two active forms.

Figure 4D shows ROCK2 activity in each experimental group. ROCK2 activity in the groups treated with EAO-alone or EAO plus AIIB2 was significantly higher than that in all other groups (approximately 186% or 168% of control levels in SH-SY5Y cells and 203% or 184% of control levels in MO3.13 cells, respectively) (*** p < 0.001 or **** p < 0.0001 vs. controls). ROCK2 activity in these groups was also significantly higher than that in the EAO plus EGCG groups (#### p < 0.0001 vs. EAO-treated groups), in which ROCK2 activity was approximately 120% of control levels. In contrast, ROCK2 activity remained unchanged in groups treated with EGCG or AIIB2 alone (p > 0.05 vs. controls). These results indicate that ROCK2 activity levels (Figure 4D) did not directly correspond to ROCK2-bound RhoA levels (Figure 2D) or GzmB levels (Figure 3C), but instead reflected the combined effects of dual activation by RhoA and GzmB. These results further suggest that the marked increase in ROCK2 activity induced by EAO was associated with both the initial interaction between EAO and integrins and the GzmB-associated neuroinflammatory response induced by EAO. The ability of EGCG to reduce EAO binding to integrins (Table 1) and attenuate ROCK2 hyperactivation (Figure 4D) further supports these observations.

The effects of EAO on ROCK2 activity in the brain were further evaluated in WT mice, AD mice, and EGCG-treated AD mice. As shown in Figure 4E, EGCG-treated AD mice exhibited significantly lower ROCK2 activity in the brain than untreated AD mice (### p < 0.001 vs. AD controls). However, ROCK2 activity in both groups remained significantly higher than that in WT mice (approximately 150% and 117% of WT levels, respectively) (* p < 0.05 and **** p < 0.0001 vs. WT controls). These results indicate that ROCK2 was hyperactivated in AD brains and that 14 days of EGCG administration reduced ROCK2 hyperactivation. The neuroprotective effects of EGCG against EAO [24] and its ability to reduce the RhoA–ROCK2 binding levels (Figure 2D) and to maintain physiological levels of GzmB (Figure 3C) appeared to be closely associated with its attenuating effects on ROCK2 activity. Overall, EAO or EGCG exhibited similar effects on ROCK2 under both in vitro and in vivo conditions. Taken together, these results indicate that EAO significantly increased ROCK2 activity while only modestly altering the relative levels of the two ROCK2 forms, whereas EGCG attenuated the effects of EAO on ROCK2 activity both in vitro and in vivo.

3.5. EAO-Induced ROCK2 Hyperactivation Led to a Significant Increase in Phosphorylated ERM (p-ERM) and Rigid F-Actin Bundles/Cables on the Inner Surface of the Cytoplasmic Membrane

The highly homologous ezrin/radixin/moesin (ERM) proteins are downstream effectors and direct substrates of ROCK. Within cells, ERM proteins function as scaffold proteins that provide a dynamic linkage between the plasma membrane and the underlying actin cytoskeleton [44]. In neural cells, ERM proteins also play crucial roles in coordinating the formation and development of cell surface structures, including neurites and synapses, by linking the cytoplasmic actin cytoskeleton to the inner surface of the plasma membrane [45].

ERM proteins consist of three distinct regions: (1) an N-terminal domain that binds membrane lipids or the cytoplasmic domains of transmembrane proteins; (2) an intermediate α-helical region; and (3) a C-terminal domain that interacts with F-actin. Because of the strong interaction between the N-terminal and C-terminal domains, dephosphorylated ERM (inactive form) adopts a closed conformation and is unable to bind membrane lipids, membrane proteins, or F-actin filaments, thereby preventing it from functioning as a scaffold protein. In contrast, phosphorylation of a conserved Thr residue within the C-terminal domain (Thr567, Thr564, and Thr558 in human ezrin, radixin and moesin, respectively) by ROCK2 induces charge-related conformational changes, causing p-ERM (active form) to adopt an open conformation that exposes actin-binding sites and enables scaffold function [46]. Therefore, ERM function depends on reversible phosphorylation, and maintenance of the balance between phosphorylation and dephosphorylation is essential in neural cells.

Although several enzymes can phosphorylate ERM proteins, ROCK2 is considered the predominant kinase involved [47]. Given the EAO-induced increase in ROCK2 activity (Figure 4D), the extent to which ROCK2 hyperactivation disrupted p-ERM levels and F-actin distribution in neuronal and glial cells was investigated. Total ERM (T-ERM) and p-ERM levels in SH-SY5Y and MO3.13 cells treated with EAO with or without EGCG, AIIB2, or Y-27632 were first examined by Western blotting. Their relative levels were subsequently quantified by indirect ELISA, and colocalization of p-ERM and F-actin was evaluated by double-label IF staining using anti-p-ERM antibody and TRITC-conjugated phalloidin. Figure 5A shows the Western blot bands for T-ERM and p-ERM in each group. Semiquantitative grayscale analysis demonstrated that T-ERM levels remained nearly identical across all experimental and control groups, whereas p-ERM levels varied substantially. These results suggest that EAO and EGCG/AIIB2/Y-27632 affected ERM phosphorylation rather than ERM expression in both cell systems.

Figure 5.

Determination of p-ERM and F-actin in SH-SY5Y and MO3.13 cells in the presence of EAO with or without EGCG/AIIB2/Y-27632 for 24 h. (A) Western blot analysis of total ERM (T-ERM) and phosphorylated ERM (p-ERM) in SH-SY5Y and MO3.13 cells. kD: kilodaltons. (B) Relative levels of p-ERM in each group shown in (A), determined by indirect ELISA and normalized to the corresponding control groups. The p-ERM levels in the control groups were defined as 100%. Between-group comparisons were performed using one-way ANOVA, followed by Šídák’s multiple comparisons test. Ctrl: control. All data are presented as mean ± SEM from six independent experiments. Symbols *, #, and § indicate significant differences between the experimental and control groups, between EAO-alone and EAO plus EGCG/AIIB2/Y-27632 groups, and between EAO plus EGCG/AIIB2/Y-27632 and EGCG/AIIB2/Y-27632-alone groups, respectively. # p < 0.05, **/§§ p < 0.01, *** p < 0.001, and ****/####/§§§§ p < 0.0001. (C) Representative images of double-label immunofluorescence (IF) staining for p-ERM (green) and F-actin (red) in SH-SY5Y and MO3.13 cells using anti-p-ERM antibody/FITC-conjugated secondary antibody (green), and TRITC-conjugated phalloidin (red). The merged images include p-ERM (green), F-actin (red), nucleus (blue), and colocalized p-ERM and F-actin (yellow). Ctrl: control.

Figure 5B shows the quantified p-ERM levels in each group. Overall, in both cell types, p-ERM levels were positively correlated with ROCK2 activity levels (Figure 4D and Figure 5B), except in the Y-27632-treated groups (k to n in Figure 5B). Specifically, p-ERM levels in the EGCG- and AIIB2- alone groups were comparable to those in the control groups (p > 0.05 vs. controls), whereas the p-ERM levels in the groups treated with EAO-alone or EAO plus AIIB2 were significantly elevated particularly in MO3.13 cells, reaching approximately 141% to 150% (a and b in Figure 5B) or 131% to 140% (g and h in Figure 5B) of those in the control groups, respectively (**** p < 0.0001 vs. controls). In contrast, the p-ERM levels in the EAO plus EGCG groups were only moderately elevated, ranging from approximately 116% to 128% of the control levels (c and d in Figure 5B). In addition, in SH-SY5Y cells only, the p-ERM levels were slightly increased in the groups co-treated with EAO and Y-27632 (k in Figure 5B), reaching about 108% of that in the control group.

These results indicate that ERM phosphorylation was predominantly regulated by ROCK2 and that EAO significantly increased p-ERM levels through ROCK2 hyperactivation. This conclusion was further supported by the observation that p-ERM levels in Y-27632-treated groups were significantly lower than those in the other groups and, in some cases, even lower than those in the control groups, particularly in MO3.13 cells. These results also suggest that although other kinases may contribute to ERM phosphorylation in cells, their contribution in neural cells appeared to be limited. Notably, changes in p-ERM levels, whether increases or decreases, appeared more pronounced in MO3.13 cells than in SH-SY5Y cells across all EAO-, EGCG-, and AIIB2-treated groups (a–j in Figure 5B). This pattern was consistent with changes in GzmB levels shown in Figure 3C, further suggesting that glial cells might be more sensitive to these extracellular factors. In addition, the p-ERM levels in all Y-27632-treated groups (k–n in Figure 5B) were significantly lower in MO3.13 cells than in SH-SY5Y cells, suggesting that ERM phosphorylation mediated by kinases other than ROCK2 may occur to a lesser extent in MO3.13 cells.

The subcellular distribution of p-ERM and F-actin filaments in SH-SY5Y and MO3.13 cells was subsequently examined by double-label IF staining using anti-p-ERM antibody (green) and TRITC-conjugated phalloidin (red). As shown in Figure 5C, colocalization of F-actin and p-ERM was primarily restricted to cell-surface structures or localized beneath the plasma membrane where actin filaments were attached. However, in the EAO-alone groups, F-actin and p-ERM markedly increased colocalization (yellow) on the inner surface of the plasma membrane, and the F-actin filaments appeared to be stiffened into rigid bundles/cables. These results suggest that aggregation and stiffening of F-actin were associated with increased recruitment or anchoring of p-ERM at the inner membrane layer. In addition, cell rounding and a markedly impaired neurite outgrowth, both in neurite number and length, were observed in these groups (second column from the left in Figure 5C). In contrast, the co-colocalization of F-actin and p-ERM adjacent to the plasma membrane was markedly reduced in the groups co-treated with EAO and EGCG/AIIB2 groups, particularly in MO3.13 cells (fourth column from the right in Figure 5C), although the extent of F-actin accumulation beneath the plasma membrane remained greater than that in the control groups. Furthermore, the p-ERM and F-actin colocalization and overall cell morphology in the EGCG/AllB2 alone groups (fifth and third columns from the right in Figure 5C) were similar to those observed in control cells.

Conversely, in the groups treated with EAO plus Y-27632 or Y-27632 alone (first and second columns from the right in Figure 5C), accumulation of p-ERM and F-actin near the plasma membrane was rarely observed, and rigid F-actin bundles/cables were absent. Instead, characteristic neural cell morphology and elongated neurites were observed, particularly in SH-SY5Y cells. Notably, these elongated neurites formed under conditions associated with low p-ERM levels (Figure 5B). These results suggest that reduced p-ERM levels promoted neurite outgrowth by weakening the anchoring role of p-ERM. However, neurite outgrowth in these Y-27632-treated groups appeared to exceed normal physiological levels primarily in neurite length rather than neurite number; it is worth noting, however, that this may potentially affect synapse formation between neighboring neurons.

Taken together, these results indicate that Aβ42O-induced ROCK2 hyperactivation in both cell types caused substantial increases in p-ERM levels and promoted formation of rigid F-actin bundles/cables beneath the plasma membrane, ultimately restricting or impairing neurite outgrowth (neuritogenesis). Conversely, excessively low p-ERM levels induced by ROCK2 inhibitor Y-27632 were also associated with excessive neurite outgrowth; however, it could be speculated that this may adversely affect synapse formation between adjacent neurons.

In addition to ERM, ROCK2 also indirectly promotes the phosphorylation of cofilin-1 at Ser3 through prior phosphorylation of LIMK. Dephosphorylated cofilin-1 (active form) functions as a major actin-depolymerizing factor that cleaves F-actin filaments thereby promoting actin filament depolymerization and release of G-actin subunits. This process maintains cytoskeletal dynamics and plasticity by facilitating the dynamic interconversion, between F-actin and G-actin [48]. In neural cells, neurite outgrowth and structural remodeling depend on dynamic assembly and the remodeling of actin cytoskeleton near the plasma membrane, especially at the growth cone, synaptic terminals, and dendritic spines. These processes are essential for synapse formation and plasticity. Therefore, cofilin-1 activity is critical for normal actin cytoskeleton dynamics. Because phosphorylation alters its charge properties, phosphorylated cofilin-1 (p-cofilin-1) loses its ability to sever or depolymerize F-actin and no longer interacts effectively with F-actin filaments. Consequently, it cannot promote F-actin filament disassembly or conversion of F-actin to G-actin, thereby severely restricting dynamic assembly and remodeling of the actin cytoskeleton. Thus, cofilin-1 phosphorylation is also closely associated with abnormalities in neurite outgrowth and synapse formation in neural cells.

Given the relationship between ROCK2 activation and cofilin-1 phosphorylation, Western blotting and double-label IF staining with anti-p-cofilin-1 antibody and TRITC-conjugated phalloidin were performed to evaluate p-cofilin-1 levels and its subcellular distribution relative to F-actin in SH-SY5Y and MO3.13 cells. Figure S1 showed the Western blot bands for total cofilin-1 (T-cofilin-1) and p-cofilin-1 in each group. According to semiquantitative grayscale analysis, T-cofilin-1 levels, including both monomers and aggregates forms, remained comparable across all experimental and control groups (p > 0.05 vs. control). In contrast, p-cofilin-1 levels exhibited marked alterations, with a pattern closely resembling that of p-ERM (Figure 5B), as shown in Figure S1B. The results shown in Figure S1B demonstrated that relative changes in p-cofilin-1 levels in almost all experimental groups were comparable to the changes in p-ERM (Figure 5B), with the exception in the EAO-alone group (a and b in Figure S1B). These results indicate that the EAO-induced ROCK2 hyperactivation markedly increased cofilin-1 phosphorylation and that EGCG, AIIB2, and Y-27632 exerted effects similar to those observed for p-ERM regulation. However, relative levels of p-cofilin-1 levels in EAO-treated groups were substantially greater than those observed for p-ERM, reaching approximately 192% of the control levels in MO3.13 cells and 179% in SH-SY5Y cells. These results suggest that activated LIMK may contribute to amplification of the phosphorylation cascade associated with cofilin-1 and that the extent of this amplification depends on LIMK activation levels.

The subcellular distribution of p-cofilin-1 relative to F-actin in SH-SY5Y and MO3.13 cells reflected the relationship between rigid F-actin filaments and elevated levels of cofilin-1 phosphorylation. As shown in Figure S1C, in both cell types exhibited F-actin distribution patterns and morphological characteristics similar to those observed in Figure 5C, including rigid F-actin bundles/cables beneath the plasma membrane and cell rounding in the EAO-treated group and elongated neurites in the Y-27632-treated groups. In contrast to the localization pattern of p-ERM, p-cofilin-1 was distributed diffusely throughout the cytoplasm or around the nucleus rather than concentrated at the cell periphery. These results indicate that stiffening and abnormal organization of F-actin were also closely associated with elevated p-cofilin-1 levels induced by excessive ROCK2 activation. Consequently, the results shown in Figure S1B,C indicate that abnormal increases in p-cofilin-1 levels impaired the dynamic F-actin assembly and neurite outgrowth in neural cells.

Taken together, these results indicate that EAO-induced ROCK2 hyperactivation in neural cells resulted in the excessive phosphorylation of ERM and cofilin-1, leading to stiffening and retention of F-actin near the plasma membrane. These alterations restricted plasma membrane expansion and impaired neurite outgrowth and cytoskeletal plasticity.

3.6. Analysis of SH-SY5Y and MO3.13 Cell Viability in the Presence of EAO with or Without EGCG/AIIB2/Y-27632 by MTT Assay, and Analysis of Cell Population Morphology and Aβ Deposit in AD Brain by HE and IHC Staining

The stiffening of F-actin and impaired neurite outgrowth associated with elevated p-ERM and p-cofilin-1 levels inevitably affect neuronal metabolism and survival. Our recent studies demonstrated that EAO impairs neuronal viability and survival [17,24], whereas EGCG protects neural cells against EAO-induced neurotoxicity and improves the AD brain microenvironment [24,49]. To determine whether the effects of EAO and co-administered EGCG on neural cell viability and survival were associated with ROCK2 hyperactivation induced through the integrin-RhoA-ROCK2 and GzmB-ROCK2 signaling pathways, the viability of SH-SY5Y and MO3.13 cells in the presence of EAO with or without AIIB2 or Y-27632 was measured and compared to previously obtained results for EAO and EGCG treatment. In addition, following 14 days of administration of EGCG, AIIB2, and Y-27632, brain cell populations and Aβ deposition spots/plaques in the hippocampal region of AD mice were examined.

The results shown in Figure 6A demonstrated that cell viability in the EAO-alone groups was significantly lower than that in the groups co-treated with EAO and EGCG/AIIB2/Y-27632 in both cell systems (### p < 0.001 or #### p < 0.0001 vs. EAO-alone). However, the viability of all treatment groups remained lower than that of the control groups, reaching approximately 60–62%, 93–95%, 76–80%, and 86–89% of control levels, respectively (* p < 0.05 to **** p < 0.0001). These results indicate that EAO-induced ROCK2 hyperactivation not only impaired F-actin assembly and neurite outgrowth (Figure 5C and Figure S1C) but also ultimately reduced neural cell viability and survival. In addition to the protective effects of EGCG [24,49], blockade of membrane integrins and/or inhibition of ROCK2 activity effectively attenuated EAO-induced reductions in neural cell viability and survival. These results were consistent with the results shown in Figure 5C and Figure S1C. However, it should be noted that cell viability was slightly reduced in the AIIB2-alone groups (approximately 89–92% of control levels), whereas viability remained largely unchanged in the Y-27632-alone groups with viability values reaching approximately 103% of control levels (Figure 6A). These results suggest that inhibition of ROCK2 activity may be safer than integrins blockade for reducing or suppressing EAO-induced damage to neural cell survival, at least under in vitro conditions.

Figure 6.

Effects of EGCG, AIIB2, and Y-27632 against EAO on the viability of SH-SY5Y and MO3.13 cells and on Aβ deposition in the brains of AD model mice. (A) Viability of SH-SY5Y and MO3.13 cells in the presence of EAO with or without EGCG, AIIB2, or Y-27632, respectively, at 37 °C for 24 h, as measured by the MTT assay. The relative viability of the control group (Ctrl) was defined as 100%. Symbols *, #, and § indicate significant differences between the experimental and control groups, between the EAO-alone and EAO plus EGCG/AIIB2/Y-27632 groups, and between the EAO plus EGCG/AIIB2/Y-27632 and EGCG/AIIB2/Y-27632-alone groups, respectively. */§ p < 0.05, ** p < 0.01, ***/###/§§§ p < 0.001, and ****/#### p < 0.0001. (B) Representative images of the hippocampal region from the negative control (nCtrl) (a), positive control pCtrl (b), EGCG-administered (c), AIIB2-administered (d), and Y-27632-administered (e) groups, as assessed by HE staining (first row) and Aβ-reactive IHC staining with anti-Aβ42 antibody (second row). Arrows indicate Aβ plaques. Scan bar: 200 μm. (C,D) Quantitative analysis of the mean Aβ-positive areas in the selected region in (B) and the mean number of Aβ deposit spots/plaques per square millimeter in (B), respectively. The mean Aβ-positive area of the pCtrl group was defined as 100%. Samples vs. pCtrl: ** p < 0.01, *** p < 0.001, and **** p < 0.0001.

Given the bioavailability of AIIB2 [50] and Y-27632 [51], as well as the ability of Y-27632 to penetrate the BBB [52], the overall morphology of brain cell populations and Aβ deposition spots/plaques in the hippocampal region of AD brains was further examined via HE and IHC staining. The results were compared with those obtained from EGCG-treated AD mouse brains [24]. HE staining images in Figure 6B (first row) showed that brain cell populations and hippocampal cell bands in the negative control (nCtrl) group (a in Figure 6B) exhibited homogeneous morphology, high cellular density, and orderly organization, whereas those in the positive control (pCtrl) group (b in Figure 6B) showed heterogeneous morphology, loose arrangement, and marked disorganization. In contrast, as previously reported [24], the morphology of cell populations and hippocampal cell bands in the EGCG-treated group more closely resembled that of the nCtrl group, whereas the morphology observed in the AIIB2-treated group appeared more similar to that of the pCtrl group. The overall morphological characteristics observed in the Y-27632-treated group appeared intermediate between those of the nCtrl and pCtrl groups. These results indicate that, in addition to EGCG, Y-27632 significantly improved the overall condition of hippocampal cell populations after crossing the BBB, suggesting partial reversal of degenerative alterations in AD brain tissue. These observations further support the in vitro results that EAO-induced ROCK2 hyperactivation reduced neuronal survival (Figure 6A) in addition to impairing neurite outgrowth (Figure 5C and Figure S1C). Therefore, EAO-induced ROCK2 hyperactivation appears to represent one of the major mechanisms underlying EAO neurotoxicity in the brain.

Although AIIB2 was likely absorbed into the circulation following intraperitoneal injection, it was probably unable to effectively cross the BBB and enter the brain to block EAO-mediated disruption of membrane integrins, as intact antibody molecules generally exhibit limited BBB permeability. Accordingly, no obvious improvement in overall brain cell population morphology was observed in the AIIB2-treated group (d in Figure 6B). These results further support the proposed mechanism through which EAO exerts neurotoxic effects via integrins.

Previous studies have demonstrated that amyloid precursor protein (APP) processing in vivo is associated with ROCK2 activity [53], and that increased ROCK2 activity promotes the production and secretion of Aβ42 in the brain [54]. Because monomeric Aβ42 may exert beneficial effects in neural cells [23], it remains unclear whether increased ROCK2 activity directly promotes Aβ deposition in the brain or whether ROCK2 inhibition can reduce Aβ deposition in AD brains. As shown in Figure 6B (second row) and Figure 6C,D, Aβ deposition spots and plaques in each group were further examined by IHC, and their levels were quantified.

As shown in Figure 6B (second row), compared with the pCtrl group (b in Figure 6B), the groups treated with EGCG, AIIB2, or Y-27632 (c, d, and e in Figure 6B) appeared to exhibit fewer and less intensely stained Aβ deposition spots/plaques. The results in the EGCG-treated group were consistent with those reported previously [24]. Quantitative analysis of Aβ-positive areas and numbers of observed Aβ spots/plaques (Figure 6C,D) demonstrated that the average number of Aβ spots/plaques was similarly reduced in the EGCG-, AIIB2-, and Y-27632-treated groups (Figure 6D), whereas the mean Aβ-positive area was reduced by approximately 24%, 11%, and 32%, respectively, compared with the pCtrl group (100%) (** p < 0.01 or *** p < 0.001 vs. controls) (Figure 6C). These results indicate that, similar to EGCG, AIIB2 and Y-27632 reduced Aβ deposition in AD brains.

In AD mouse brains, the results from the Y-27632-treated groups showed that inhibition of increased ROCK2 activity reduced Aβ deposition and/or promoted clearance of Aβ deposits, indicating improvement in the brain microenvironment and/or attenuation of EAO-induced damage to brain cells. These results were consistent with the morphological improvements in brain cell populations observed by HE staining in this group (e in Figure 6B). In contrast, results observed in the AIIB2-treated groups might suggest that AIIB2 enhanced peripheral transport of brain-derived Aβ42, indirectly contributing to reduced Aβ deposition in the brain (see Section 4).

Taken together, these results indicate that reduced neuronal and glial cell viability and survival under both in vitro and in vivo conditions were directly associated with ROCK2 hyperactivation in the presence of EAO. Treatment with AIIB2 or Y-27632 attenuated EAO-induced reductions in neural cell survival, reduced Aβ deposition in AD mouse brains, and improved the morphology of brain cell populations. However, cellular analyses also suggested that AIIB2 might exert certain adverse effects on neural cell survival.

4. Discussion

Aβ42 molecules are metabolic products in the brain. Although Aβ42 monomers (Aβ42M) have been reported to possess neurotrophic properties [23], progressive alterations in the physiological environment and metabolic state of the brain gradually change the composition and characteristics of the extracellular environment, particularly the microenvironment that supports the neurogenic niche and neural cell survival. These changes promote the gradual formation and deposition of metastable and heterogeneous Aβ42 aggregates in the ECM. Compared with larger Aβ42 aggregates, such as Aβ42 fibrils, the smaller and more soluble Aβ42 oligomers (Aβ42O) can more readily access neural cells. Consequently, owing to the unique amphiphilic properties of Aβ42O units [5], they are more likely to disrupt or damage specific structures and components on the surface of neural cells, including membrane receptors, membrane proteins, and membrane lipids, ultimately resulting in neuronal damage and AD progression.

Although anti-Aβ monoclonal antibody, such as lecanemab and donanemab, have shown some disease-modifying effects by reducing amyloid burden and modestly slowing cognitive decline in early AD patients, their benefits remain limited. These therapies neither reverse existing neuronal damage, nor cure AD [21]. Currently, no effective approaches is available to halt or slow the progression of AD. Part of this challenge may be attributed to the incomplete understanding of the specificity and safety of the mechanisms underlying the disease. Our previous studies [17], demonstrated that EAO can bind membrane integrins, thereby reducing neuronal motility and adhesion and impairing neurite outgrowth. Integrins are ECM receptors that are abundantly distributed on the surface of neural cells with a large surface area. Because integrins interact with a wide range of ligands, direct blockade of integrins may attenuate EAO-induced neural cell damage but may also disrupt normal cellular metabolism and physiological functions, potentially leading to adverse effects. Therefore, identifying key downstream targets within the integrin signaling pathway that are primarily responsible for EAO-induced neurotoxicity is particularly important. This study focused on the specific disruption of actin cytoskeleton in neural cells by EAO and systematically investigated the major effectors and signaling events associated with EAO-induced impairment of neurite outgrowth (or neuritogenesis) and neuronal survival. By clarifying their roles in EAO-mediated neurotoxicity, this study provides an experimental and theoretical basis for the development of safe and effective strategies to reduce or prevent EAO-induced neuronal damage and may contribute to the development of therapeutic approaches for AD.

The findings of this study suggest that EAO can induce the hyperactivation of intracellular ROCK2, a process that involves at least a dual-mediation by the integrin/RhoA and GzmB signaling pathways, as this effect was significantly associated with EAO-induced increases in both GzmB and RhoA–ROCK2 binding levels. Therefore, our data support the possibility that EAO-induced ROCK2 hyperactivation represents the combined consequence of RhoA- and GzmB-mediated ROCK2 activation, in response to EAO. RhoA-mediated ROCK2 activation was integrin-dependent, whereas GzmB-mediated ROCK2 activation was integrin-independent but might have been associated with EAO-induced neuroinflammation. The EAO-induced increases in the levels and activities of these proteins and enzymes resulted in dysregulation of the actin cytoskeleton, suppression of neuritogenesis, reduced neural cell survival, and abnormalities in brain cell populations. Parallel inhibition experiments performed using EGCG, the integrin inhibitor AIIB2, and the ROCK inhibitor Y-27632 further supported the detrimental effects of EAO on neural cells by causing hyperactivation of ROCK2. It is worth noting that, in addition to integrin receptors, EAO can also bind to other receptors [6,55], such as N-methyl-D-aspartate receptor (NMDAR), p75 neurotrophic factor receptor (p75NTR), metabotropic glutamate receptors (mGluR5) and others, thereby inducing various forms of neurotoxicity, such as mitochondrial dysfunction, dysregulation of Ca²⁺ signaling, the formation of reactive oxygen species (ROS), and neuroinflammation, ultimately leading to synaptic dysfunction and neurodegeneration. These pathological processes have all been reported to be involved in the pathology of AD, and both EAO/integrin/RhoA/ROCK2 and EAO/GzmB–ROCK2 mechanisms identified in this study and these reported mechanisms may synergistically contribute to dynamic dysregulation of the actin cytoskeleton, growth cone collapse, and neurite defects in AD.

Previous studies demonstrated that EGCG binding to EAO reduced EAO binding to neural cells [23,24]. The present study further demonstrated that EGCG inhibited the targeted binding of EAO to integrins. EGCG is a naturally occurring polyphenolic compound with antioxidant and anti-neuroinflammatory properties, which may explain its ability to reduce EAO-induced increases in GzmB levels both in vitro and in vivo. However, the antioxidant properties of EGCG also render it susceptible to oxidative metabolism in vivo. Consequently, although EGCG exhibited marked neuroprotective effects in vitro, its neuroprotective efficacy and capacity to enhance EAO metabolism in vivo may be more limited. Despite extensive evidence indicating that natural polyphenols play important roles in attenuating EAO-induced oxidative stress and neuroinflammation [49], maintaining effective in vivo concentrations of these compounds remains a challenge requiring further investigation [56].

AIIB2 is an anti-β1 integrin antibody. The present study demonstrated that AIIB2 prevented EAO binding to membrane integrins and attenuated neural cell damage associated with EAO-induced ROCK2 hyperactivation, while also reducing Aβ deposition in the brains of AD model mice. However, cellular-level analyses suggested that AIIB2 might exert certain adverse effects on neural cells, particularly with respect to neural cell survival. Blockade of integrin signaling may have induced intracellular metabolic disturbances. Therefore, further evaluation of the safety of AIIB2 is required. In addition, although previous studies using other anti-integrin antibodies reported similar reductions in brain Aβ load [57], the mechanisms underlying the in vivo effects of anti-integrin antibodies may be complex because intact antibodies generally have limited ability to cross the BBB. Nevertheless, based on reported neuroprotective effects of nattokinase and anti-oligomeric Aβ42 antibodies [25], it may be speculated that once absorbed into the circulation, AIIB2 may enhance transport of Aβ42 from the brain to peripheral tissues by preventing blood-derived Aβ42, and from damaging vascular endothelial cells, thereby indirectly reducing Aβ deposition in the brain.

Compared with blockade of integrins, reduction in EAO-induced ROCK2 hyperactivation appeared to exert fewer adverse effects on neuronal metabolism and survival, as shown in Figure 6A. The beneficial effects of the ROCK2 (or ROCK1) inhibitor Y-27632 on neural cells, mediated through suppression of excessive ROCK2 activation both in vivo and in vitro, are also supported by previous genetic studies [58], although no significant differences in ROCK2 protein levels were observed 24 h after EAO treatment. Therefore, the findings of the present study further support the association between ROCK2 hyperactivation and EAO-induced neurotoxicity in neurodegenerative diseases. This study indicates that synergistic activation by RhoA and GzmB within the cytoplasm might contribute to the ultimate hyperactivation of intracellular ROCK2. Notably, although full-length ROCK2 represented the majority of total ROCK2 molecules and truncated ROCK2 generated through GzmB-mediated cleavage constituted only a relatively small fraction of total ROCK2 (owing to low basal intracellular GzmB levels), the contribution of truncated ROCK2 to overall ROCK2 activity should not be underestimated (Figure 4A). This is because: (1) RhoA-mediated activation of ROCK2 is reversible, meaning that full-length ROCK2 exists in both self-inhibited and active states simultaneously; (2) RhoA activity is regulated by multiple intracellular and extracellular factors, and only active GTP-bound RhoA can bind and activate ROCK2; and (3) truncated ROCK2 represents a constitutively active enzyme whose activity persists until the protein is degraded.

Furthermore, the findings of this study indicates that excessive inhibition of ROCK2 activity (significantly below normal ROCK2 activity levels) may lead to excessive neurite outgrowth, primarily reflected by increased neurite length rather than neurite number. However, excessively elongated neurites may extend beyond the optimal range required for appropriate synapse formation, thereby potentially disrupting synaptic development, intrinsic neural network organization, and neural transmission. Although this study did not examine these possible consequences, its findings suggest that a physiological threshold of ROCK2 activity may exist in neural cells. If ROCK2 activity substantially exceeds this threshold, as observed under EAO-induced conditions, F-actin aggregation and cytoskeletal dysregulation may occur, which could result in impaired neurite development and synapse formation. Conversely, if ROCK2 activity falls substantially below this threshold, neurite outgrowth may become excessive and similarly disrupt synapse formation. Taken together, these findings suggest that maintenance of moderate ROCK2 activity is essential for normal neural cell function.

5. Conclusions

In summary, EGCG, AIIB2, and Y-27632 may drive the reduction in Aβ deposition in the brains of AD mice through different physiological mechanisms, despite their differing molecular natures, chemical properties, and targets of action. These mechanisms primarily included: (1) a reduction in Aβ42 production, which is directly related to ROCK2 activity [53,54]. The results from the EGCG and Y-27632 groups correlated positively with this mechanism; (2) reduced aggregation of Aβ42, which is related not only to decreased Aβ42 production but also to the microenvironment, as the aggregation and deposition of Aβ42 peptide chains are directly related to physiological environmental conditions, including ionic strength, pH, and interacting factors such as EGCG [25]; and (3) peripheral transport and metabolism of Aβ42 aggregates are related to improvements in the intracerebral physiological environment and changes in the BBB. Aβ levels in the blood of AD mice are significantly elevated [25]; AIIB2 uptake may inhibit EAO’s interference with the integrin signaling pathway in vascular endothelial cells and improve the BBB. A schematic summary of these findings is presented in Figure 7. Table S1 summarizes the baseline effects of EGCG, AIIB2, and Y-27632 in this study.

Figure 7.

Schematic summary of EAO-induced hyperactivation of intracellular ROCK2 through a dual-mediation involving RhoA and GzmB, and the resulting F-actin aggregation and stiffening at the inner membrane layer, cell rounding, and impaired neurite outgrowth. Aβ42O: Aβ42 oligomer; ROCK2: Rho-associated coiled coil-containing protein kinase 2; EGCG: epigallocatechin gallate; ERM: ezrin/radixin/moesin; FAK: focal adhesion kinase; GzmB: granzyme B; LIMK: LIM-domain kinase; RhoGEF: Rho guanine nucleotide exchange factors; RhoGAP: Rho GTPase-activating proteins.

This study has several limitations. SH-SY5Y and MO3.13 are neuronal and oligodendrocyte cell lines, respectively. Oligodendrocytes are the most abundant type of glial cell type [59] and play essential roles in the development of myelin sheaths surrounding neurons within the central nervous system. Accordingly, SH-SY5Y and MO3.13 cells are widely employed in studies of neurodegenerative diseases. Because β1 integrins are the predominant integrin subtype expressed in oligodendrocytes [60,61], these findings related to oligodendrocytes and glial cells in this study primarily reflect the effects of EAO on β1 integrins. Additional studies are therefore required to determine whether these findings also apply to other integrin subtypes. In addition, although the findings of this study demonstrated that EAO induces a dual increase in the roles of both RhoA and GzmB in activating ROCK2, this conclusion was not validated through loss-of-function experiments. In future studies, we will consider incorporating this into our investigation aimed at elucidating the details of the EAO-ROCK2 pathological mechanism.

In conclusion, EAO induced ROCK2 hyperactivation, which may be attributed to dual mediation by RhoA and GzmB, leading to dysregulation of the actin cytoskeleton and impaired neurite formation in neuronal and glial cells. These mechanisms may represent one aspect of EAO-induced neurotoxicity. Conversely, excessive suppression of ROCK2 activity may also result in abnormal neurite outgrowth and may impair synapse formation. Therefore, irrespective of EAO presence, disruption of ROCK2 activity may reduce neuronal and glial cell viability and survival and contribute to abnormalities in brain cell populations. Given the broad functional roles of integrins and RhoA, development of approaches targeting ROCK2 or the GzmB/ROCK2 pathway, in addition to strategies aimed at eliminating EAO, may provide promising directions for delaying the progression of AD. Despite extensive advances in the neurophysiology of dementia, the pathogenesis of AD remains incompletely understood, and safe and effective therapeutic strategies remain urgently needed. The findings the study highlights the importance of maintaining balanced ROCK2 activity in neural cells and suggests are a highly relevant of ROCK2 activation to current neuroscience and neurodegeneration research. The findings of study will provide important insights for better understanding AD pathophysiology and identifying productive strategies for the development of disease-modifying drugs.

Abbreviations

Aβ42	β-amyloid protein 1-42	CNS	Central nervous system
AD	Alzheimer disease	ELISA	Enzyme-linked immunosorbent assay
APP	Amyloid precursor protein	PBS	phosphate-buffered saline
ECM	Extracellular matrix	Co-IP	Co-immunoprecipitation
EAO	Extracellular Aβ42 oligomers	MOC	Manders’ overlap coefficients
EGCG	Epigallocatechin gallate	M1/M2	Manders’ colocalization coefficients
GAPDH	Glyceraldehyde-3-phosphate dehydrogenase	BBB	Blood–brain barrier
ROCK2	Rho-associated coiled coil-containing protein kinase 2	DMEM	Dulbecco’s modified eagle’s medium
RBD	Rho-binding domain	FBS	Fetal bovine serum
GzmB	Granzyme B	SD	Standard deviation
LIMK	LIM domain kinase 1	SEM	Standard error of the mean
NMDAR	N-methyl-D-aspartate receptor	p75NTR	p75 neurotrophic factor receptor
mGluR5	metabotropic glutamate receptors	ROS	reactive oxygen species

Supplementary Materials

The following supporting information can be downloaded at https://www.mdpi.com/article/10.3390/cells15151379/s1, Figure S1: Analyses of cofilin-1 and p-cofilin-1 in SH-SY5Y and MO3.13 cells in the presence of EAO with or without EGCG/AIIB2 for 24 h; Table S1: Baseline effects of EGCG, AIIB2 and Y 27632 alone in this study.

Author Contributions

All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by C.Z. and H.L. Molecular docking was performed by C.Z., K.W., and T.Z. The project was led by Y.Z. The first draft of the manuscript was written by C.Z. and all authors commented on previous versions of the manuscript. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

All animal experiments were performed in accordance with the guidelines for the care and use of experimental animals and were approved by the Institutional Animal Care and Use Committees (IACUC) of SPF (Beijing) Biotechnology Co., Ltd. (Approval Code: AWE2024102101; date of approval: 21 October 2024) and Wuhan Secvicebio Technology Co., Ltd. (Approval Code: 2025238; date of approval: 9 September 2025).

Informed Consent Statement

Not applicable.

Data Availability Statement

The datasets and materials used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Conflicts of Interest

There are no conflicts of interest to declare.

Funding Statement

This work was supported by the grant from the National Natural Science Foundation of China (Grant No. 31970883).