The human ethics committee of the University of New South Wales approved this tissue study. Following institutional approvals, 10 μm formalin-fixed paraffin-embedded temporal cortex sections from 17 cases with different severities of AD pathology (female: male = 13:4, age = 87 ± 9.5 (years old), post mortem delay = 15 ± 11 (h), low severity, Braak stages 0–2 and not demented: intermediate severity and Braak stages 3–4: high severity, Braak stages 5–6 and dementia = 5:3:9) were obtained from the Sydney Brain Bank which collects brain tissue with informed consent from longitudinally followed research participants. Subjects were characterized according to recent criteria for the pathological diagnosis of AD. One of the cases with an intermediate severity had a diagnosis of AD for 3 years, while all those with a high probability of AD had dementia for between 5 and 17 years. Exclusion criteria were alternate neurodegenerative disorders or a dominant family history of neurodegenerative disorder.

For immunofluorescence, the 10 μm formalin-fixed, paraffin-embedded sections underwent heat-induced antigen retrieval with Citrate Buffer pH 6 for 3 min, followed by formic acid pre-treatment for 3 min. Tissue sections were double-labeled with rabbit anti-ROCK1 (1:50, Abcam) and mouse monoclonal anti-human Aβ1–42 (1:200, M0872, Dako, Glostrup, Denmark), and visualized with goat anti-rabbit Alexa Fluor 488 (1:500, A-11008, Life Technologies) and goat anti-mouse Alexa Fluor^® 594 (1:500, A-11005, Life Technologies), respectively. Sections were coverslipped with Vectashield HardSet Antifade Mounting Medium with DAPI (H-1500, Vector Laboratories, Burlingame, CA, USA) and examined on confocal microscope (Nikon Eclipse E400).

For quantification, 10 images were made for each case where plaques were present. All diffuse and cored plaques on each image was identified using the Aβ1–42 immunofluorescence, and the proportion that had ROCK1 co-localization recorded separately for both diffuse and core plaques. In total, an average of 40 ± 17 plaques were examined per case, with 66% on average being diffuse and 36% being cored. Multivariate statistics were performed in SPSS (IBM, v23) to assess differences between the percentages of plaques colocalizing ROCK1 immunoreactivity and the severity of cortical plaques observed, as assessed by the CERAD score (0 = none, 1 = mild or 1–5/image, 2 = moderate or 6–15/image and 3 = severe or more than 15/image), as well as to case diagnosis based on pathological load of different plaque types and neurofibrillary tangles (control, preclinical AD and AD).

HEK293 cells stably transfected with human APP695 harboring the “Swedish” mutation (HEK293 APP695sw) and SH-SY5Y human neuroblastoma were maintained in Eagle’s minimal essential media or DMEM (Gibco), respectively, with 10% fetal bovine serum, and 1% penicillin/streptomycin. Cells were plated at 100,000 cells/cm² density in 6-well dishes that were coated with 100 μg/ml poly-lysine. On day 2 post plating, cells were transduced with indicated lentivirus with a multiplicity of infection of 1. For all studies, cells were treated with drugs at 72 h post-transduction in conditioned media for 16 h. For RNAi knockdown of ROCK1, cells were harvested 96 h post-transduction. For transductions or transfections, equivalent amounts of cells were plated, and transfections were performed using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions. At 72 h post-transfection, cells were treated as indicated and then processed for western blot analysis or fluorescence imaging.

Cell proliferation was measured by the CCK-8 Kit (Beyotime) according to the manufacturer’s instructions. Cells in 100 μl media were treated with 10 μl CCK-8 reagent at 37°C for 2 h, followed by measuring the absorbance at 450 nm on a microplate reader.

ROCK1 plasmid (pCAG-myc-ROCK1^myc-727 Δ3, ROCK1 CA) was constructed as previously reported (Ishizaki et al., 1997). The siRNA sequence against ROCK1 (sense 5′-GGCAGAGGAAGAAUAUAAATT-3′; antisense 5′-UUUAUAUUCUUCCUCUGCCTT-3′) and scrambled siRNA was synthesized by GenePharma (Shanghai, China). Cells were transfected with ROCK1 plasmid, ROCK1 siRNA or scrambled siRNA using Lipofectamine 2000 reagent according to the manufacturer’s instructions. At 72 h post-transfection, cells and supernatant were harvested and then processed for western blot analysis.

For all cells, conditioned media were used for 48 h, and then the cells and media collected separately for biochemical analyses. Aβ40 was detected using a sandwich ELISA for human Aβ40 following the manufacturer’s instructions. Plates were read at 450 nm on a Synergy MX plate reader (BioTeck, Winooski, VT, USA).

Cells were lysed with 1% RIPA Lysis Buffer (Beyotime, China). Cell lysate was subjected to a 13,000 rpm spin to remove nuclei and debris. Cleared lysate was used for the indicated biochemical assay. Protein concentrations were determined using the Enhanced BCA Protein Assay Reagent (Beyotime). Equal amounts of cell lysate were loaded onto SDS-PAGE gels and then transferred to PVDF membranes. Membranes were blocked with 5% fat-free dry milk in Tris buffered saline (TBS), containing 0.05% Tween-20, and incubated with primary antibodies. Protein bands were detected by horseradish peroxidase-conjugated species-specific secondary antibodies. Actin was used as loading control. Images were captured using an Odyssey Image Station (LI-COR), and band intensities were quantified using Scion Image.

In order to evaluate the effects of ROCK1 plasmid on GFP–LC3B puncta formation, SH-SY5Y cells were plated on poly-lysine-coated 96-well plates and transduced with adenovirus expressing mCherry-GFP-LC3B fusion protein. After 24 h and 48 h, cell images were obtained by EVOS fl Auto (Life, USA).

To determine whether Aβ associates with ROCK1 in the brains of patients with AD, co-localization studies were performed using specific antibodies in healthy aged controls (n = 5 aged 89 ± 4 years, Braak neuritic stages 0–2), preclinical AD (n = 4 aged 85 ± 3 years, Braak neuritic stages 3–4) and clinical end-stage AD (n = 8 aged 86 ± 4 years, Braak neuritic stages 5–6). Consistent with this finding, ROCK1 immunoreactivity was identified in more cored than diffuse plaques (Figure 1). Quantitation showed that the majority of cored plaques contained ROCK1 immunoreactivity, and that colocalization increased with increasing severity of cortical Aβ deposition as assessed by CERAD plaque score (Figure 1J, Wilks’ Lambda p < 0.0001 covarying for age and brain size, post hoc Bonferroni p = < 0.002). Nearly half of all Aβ plaques in preclinical AD contained ROCK1 immunoreactivity, with significantly higher numbers of ROCK1-immunopositive Aβ plaques found in AD patients (Figure 1K, Wilks’ Lambda p = 0.03 covarying for age and brain size, post hoc Bonferroni p = 0.003 for AD diagnosis). These results demonstrate a close relationship between ROCK1 localization and pathological Aβ plaque in both preclinical and clinical AD.

To further explore the effects of ROCK1 cleavage on the production of Aβ, we constructed a plasmid expressing this constitutively active ROCK1 fragment (ROCK1 CA), as previously reported (Ishizaki et al., 1997). In HEK 293T cells stably transfected with the APP695 Swedish mutation (APPsw), ROCK1 levels were manipulated by transfection with ROCK1 CA plasmid, ROCK1 siRNA or scrambled siRNA. Forty eight hours after transfections, the levels of secreted Aβ40 in the cell culture medium were measured by ELISA. As shown in Figure 2, secreted Aβ40 levels were increased by 80% compared to the control after depletion of ROCK1 (Figures 2A,B), which is consistent with a recent study. In contrast, secreted Aβ40 levels were decreased following ROCK1 CA over-expression in HEK293T APPsw cells (Figures 2C,D). The Aβ42 level was too low to be detected by ELISA. ROCK1 CA plasmid transfection did not change cell proliferation (Figure 2E). These results confirm that ROCK1 influences Aβ secretion.

Given that there is an established relationship between ROCK1 and Aβ secretion, we next investigated the effects of ROCK1 on APP cleavage. After transfection of ROCK1-targeted siRNA or ROCK1 CA plasmid to HEK293T APPsw cells, cells were harvested 48 h later and lysates were analyzed. The holo-APP and sAPPα were examined by Western Blot with the mAb 6E10, specific for residues 1–17 of the Aβ sequence (Figure 3A). Electrophoresis revealed no significant difference of full-length APP (Figure 3C), however the amount of sAPPα was higher after ROCK1 depletion (Figure 3D). In cells with increased ROCK1 CA levels, sAPPα was significantly decreased (Figure 3D). These results confirm that ROCK1 modulates the metabolism of APP. To further explore the interaction between ROCK1 and APP/Aβ, we performed normal HEK293T cell culture with Aβ40 treatment or APP695 plasmid transfection. Addition of Aβ40 or APP695 over-expression promoted endogenous ROCK1 expression in normal HEK293T cells (Figures 3B,E).

In further experiments, we explored the mechanism of ROCK1 as an enhancer of Beclin 1 dependent autophagy. When ROCK1 was depleted, the expression of Beclin1 was decreased, and when ROCK1 CA was over-expressed by plasmid transfection, the level of Beclin1 was also increased. To determine whether autophagy is involved in the mechanism by which ROCK1 depletion increases Aβ secretion, SH-SY5Y cells were co-transfected with mCherry-LC3B adenovirus and ROCK1 CA plasmid and the patterns of LC3 puncta were analyzed by immunoflouresence. Figure 4A shows an increase in LC3 puncta after ROCK1 CA plasmid transfection. LC3 was also assessed in HEK293T cells and, as the results show, the LC3II/I ratio was increased in cells when ROCK1 CA was over-expressed and decreased following ROCK1 depletion (Figures 4E,G). We also explored whether ROCK1 interacted with Beclin1 during autophagy (Figures 4B,C). HEK293T cells transfected with myc-tagged human ROCK1 were lysed and the proteins subjected to co-immunoprecipitation with anti-ROCK1 and anti-myc antibody. Our results showed that endogenous ROCK1 and Beclin1 interacted with each other. Moreover, when ROCK1 CA was over-expressed, this interaction was enhanced (Figures 4D,F). These results showed that an interaction between ROCK1 and Beclin1 triggered the autophagic build-up of intracellular Aβ and a decrease in Aβ secretion.

To further elucidate ROCK1 influence on autophagy, 3-MA was used as an inhibitor of autophagy. After HEK293T APPsw cells were treated with 3-MA for 6 h, they were transfected with ROCK1 CA plasmid for 48 h. To test whether ROCK1 CA over-expression rescued 3-MA-induced autophagy inhibition, the cells were lysed and proteins analyzed by ELISA and Western blot. Inhibiting autophagy increased intracellular Aβ40 levels (Figure 4J), and over-expression of ROCK1 by transfection with ROCK1 CA plasmid reversed this effect even though 3-MA treatment decreased Beclin1 levels (Figures 4H,I).

After confirming the relationship between ROCK1 and autophagosome formation in vitro, we next assessed whether ROCK1 and autophagosome proteins are upregulated with Aβ in vivo by conducting immunofluorescence and IHC staining of brain sections from 10 months old APP/PS1 transgenic C57BL/6 mice. Our data showed that ROCK1 was increased significantly with neuronal autophagosomes accumulation (Figures 5A,C) and that Beclin1 was decreased in the AD mouse model (Figures 5B,D). These observations suggested that autophagy was impaired in neurons and increased ROCK1 inhibited Aβ secretion, resulting in intracellular Aβ accumulation and neurotoxicity. Taken together, these data demonstrated the same associations between ROCK1, intracellular Aβ and autophagosomes accumulation as we identified in our in vitro experiments.