Dual role of cofilin in APP trafficking and amyloid-β clearance

Abstract

The accumulation of amyloid-β (Aβ) plays a pivotal early event in the pathogenesis of Alzheimer’s disease (AD). In the brain, neurons produce Aβ by the proteolytic processing of amyloid precursor protein (APP) through the endocytic pathway, whereas microglia mediate Aβ clearance also via endocytic mechanisms. Previous studies have shown the critical importance of cofilin, a filamentous actin–severing protein, in actin dynamics and pathogen-triggered endocytic processes. Moreover, the binding of Aβ42 oligomers to β1-integrin triggers the cofilin activation, and in turn, cofilin promotes the internalization of surface β1-integrin. However, a role for cofilin in APP processing and Aβ metabolism has not been investigated. In this study, we found that knockdown of cofilin in Chinese hamster ovary 7WD10 cells and primary neurons significantly reduces Aβ production by increasing surface APP (sAPP) levels. Expression of active (S3A) but not inactive (S3E) cofilin reduces sAPP levels by enhancing APP endocytosis. Accordingly, Aβ deposition in APP and presenilin 1 (PS1) transgenic mice is significantly reduced by genetic reduction of cofilin (APP/PS1;cofilin^+/−). However, the reduction of Aβ load in APP/PS1;cofilin^+/− mice is paradoxically associated with significantly increased ionized calcium-binding adaptor molecule 1–positive microglial activation surrounding Aβ deposits. Primary microglia isolated from cofilin^+/− mice demonstrate significantly enhanced state of activation and greater ability to uptake and clear Aβ42, which is reversed with the active (S3A) but not inactive (S3E) form of cofilin. These results taken together indicate a significant role for cofilin in Aβ accumulation via dual and opposing endocytic mechanisms of promoting Aβ production in neurons and inhibiting Aβ clearance in microglia.—Liu, T., Woo, J.-A. A., Yan, Y., LePochat, P., Bukhari, M. Z., Kang, D. E. Dual role of cofilin in APP trafficking and amyloid-β clearance.

Accumulation of amyloid-β (Aβ) is the defining pathologic hallmark of Alzheimer’s disease (AD), leading to neuronal dysfunction, synapse loss, and eventually neurofibrillary degeneration (1). Aβ is generated by the proteolytic processing of the amyloid precursor protein (APP) via β-site APP-cleaving enzyme (BACE1) and presenilin 1 (PS1) and presenilin 2 complex (2). With the induction of neuronal activity, APP is routed into BACE1-containing acidic organelles via clathrin-dependent endocytosis (3), where the vast majority of BACE1 cleavage of APP occurs (4). The nonamyloidogenic α-secretase processing of APP by A disintegrin and metalloprotease 10 and 17 occurs largely on the plasma membrane, and such processing is reduced by APP endocytosis (5, 6). The accumulation of Aβ during the aging process is regulated by the rate of production, aggregation, and clearance. Although a substantial portion of soluble Aβ is cleared from the brain via the blood-brain barrier by the low-density lipoprotein receptor–related protein (LRP) and other receptors (7–10), aggregated Aβ is cleared via endocytic mechanisms by brain-resident microglia (11) and infiltrating peripheral macrophages (12).

The actin cytoskeleton is critical to maintaining cell morphology and multiple cellular processes, including membrane trafficking and endocytosis (13). Filamentous actin (F-actin) is highly dynamic at the leading edge of cells, including membrane protrusions of phagocytic cells that can internalize solid particles by 2-dimensional protrusions of the cell surface homologous to lamellipodia and synaptic zones of neurons that can take up cell surface components by local membrane invagination, constriction of the bud neck, and vesicle scission (14). Cofilin is a key actin-binding protein, which has an essential role in actin dynamics via its F-actin severing, depolymerizing, and nucleating activities (15). Cofilin is inactivated by phosphorylation on Ser3 by LIM kinase 1 (LIMK1) (16) and activated by slingshot homolog 1 (SSH1)-mediated dephosphorylation of Ser3 phosphorylation (17). At regions of low cofilin:actin ratios, cofilin binds to ADP-actin and induces persistent F-actin severing to create new barbed and pointed ends, which contributes to actin remodeling at the synapse and elsewhere by enhancing filament depolymerization from pointed ends (−), filament growth from barbed ends (+), or both (18). However, at regions of high cofilin:actin ratios, activated cofilin can bind and stabilize F-actin in a twisted form, thereby promoting the nucleation of F-actin rather than severing. Oxidation-induced intermolecular disulfide bridging of activated cofilin together with high levels of ADP-actin promote the biogenesis of cofilin-actin rods or aggregates (19–21), which are significantly elevated in brains of patients with AD and APP transgenic mice (21–24). Indeed, we and others have shown that activated cofilin is significantly increased in brains of patients with AD and APP transgenic mice (25–27).

In a previous study (25), we showed that Aβ42 oligomers (oAβ42) promote cofilin dephosphorylation and activation and that genetic reduction of cofilin abolishes oAβ42–induced loss of F-actin and F-actin–associated synaptic proteins as well as deficits in long-term potentiation. Specifically, oAβ42 binds to β1-integrin conformers and induces the activation of cofilin, an event that subsequently triggers the internalization of β1-integrin from the cell surface. Such loss of surface β1-integrin is completely abolished by small interfering RNA (siRNA)-mediated silencing of cofilin, suggesting that cofilin activation is required for oAβ42–induced β1-integrin internalization (25, 28). In contrast to oAβ42, the Reelin ligand binds to β1-integrin in physical association with APP, which promotes the nonamyloidogenic α-secretase processing of APP in an APP- and integrin-dependent manner (29). This enhanced α-secretase processing of APP and reduced Aβ production by Reelin is attributable to increased surface APP (sAPP) by blocking its endocytosis (29). Interestingly, the activity of Reelin is associated with cofilin phosphorylation and inactivation (30, 31), suggesting a role for cofilin in APP processing and trafficking, in addition to that of β1-integrin.

An intriguing role of cofilin in controlling actin-dependent endocytic mechanisms is found in the way pathogens (i.e., viruses, bacteria, fungi) attempt to infect host cells by regulating the activation state of cofilin. Multiple studies have shown that inactivation followed by activation cycle of cofilin is critical for entry of different types of viruses into host cells (32). Another role of cofilin in phagocytosis of pathogens by macrophages has also been explored. In the case of Candida albicans, an opportunistic fungal pathogen, cofilin inactivation and F-actin assembly are associated with increased phagocytosis of the pathogen by alveolar macrophages (33). Hence, the regulation of cofilin activity and state of the F-actin network appear to be critical to both endocytic processing of surface proteins and cellular entry of pathogens alike. At present, the role of cofilin in APP processing, trafficking, and Aβ clearance has not been investigated. In this study, we assessed the role of cofilin in APP processing and Aβ clearance via endocytic mechanisms in cultured cells, primary neurons, microglia, and in vivo. Our findings indicate the surprising dual and opposing role of cofilin in APP endocytic processing and Aβ production in neurons as well as inhibition of Aβ clearance by microglia, both of which contribute to Aβ deposition in APP/PS1 transgenic mice.

MATERIALS AND METHODS

Ethics approval

The Institutional Animal Care and Use Committee at University of South Florida (USF) has approved the all the experimental methods and protocols involving mice in this study. The Institutional Animal Care and Use Committee and Institutional Biosafety Committees at USF have also approved that all the methods used in this study were performed in accordance with the relevant guidelines and regulations.

Mice

Wild-type (WT), APP/PS1 (34), APP/PS1;cofilin^+/−, and cofilin^+/− mice were bred in the C57BL6 background as previously described in Woo et al. (25). APP/PS1 mice express the transgenes APPswe and PS1ΔE9 driven by the prion protein promoter (34). Mice were housed together until the time they were euthanized at 7 mo of age. Water and food were supplied ad libitum with 12-h light/dark cycle under standard vivarium conditions.

Cell culture

Chinese hamster ovary (CHO) 7WD10 cells were cultured in DMEM (Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin (P/S). Mouse cortical primary neurons from postnatal day (P)0) pups were cultured in neurobasal medium supplemented with 1× B-27 supplement and 1× l-glutamine (Thermo Fisher Scientific) All the cells were cultured in a humidified atmosphere (5% CO₂) at 37°C. To culture primary microglia, mixed glial cultures were derived from postnatal day (P)0 pups and grown in culture flasks in 50:50 DMEM-F12 with 10% FBS, 1% P/S, and 1% GlutaMax. Microglia were purified on 14 d in vitro (DIV) by shaking as previously described in Tamashiro et al. (35) and transferred to poly-d-lysine–coated 12-mm round coverslips. Cells were allowed to adhere to coverslips for at least 24 h before experiments. BV2 cells were maintained in Opti-MEM containing 10% FBS, 1% P/S, and 1% GlutaMax. BV2 cell lines were obtained from Kevin Nash (Byrd Neuroscience Institute, USF).

Antibodies and reagents

Mouse anti-APP (6E10) recognizing amino acid residues 1–16 of human Aβ sequence in APP and β-actin monoclonal antibody were obtained from MilliporeSigma (Burlington, MA, USA). Anti-cofilin (D3F9) and anti–Aβ (D54D2) primary antibodies were purchased from Cell Signaling Technology (Danvers, MA, USA). The pAb CT15 detecting the 15-aa residue C-terminal fragment (CTF) of APP was previously described in refs. 36 and 37. FITC–Aβ1–42 was obtained from GenicBio (Shanghai, China). Aβ1–42 peptide was purchased from Thermo Fisher Scientific. To prepare oligomeric Aβ1–42, the powder of Aβ1–42 was dissolved in hexafluoroisopropanol at 1 mM, and then the Aβ1–42 in hexafluoroisopropanol was evaporated in a fume hood for overnight and subjected to speed vacuum for 1 h. Then Aβ1–42 film was dissolved in DMSO (5 mM), and PBS was added to a final concentration of 100 μM Aβ1–42 and incubated at 4°C (for Aβ oligomers) or 37°C (for Aβ fibrils) for 24 h, as we previously described in Woo et al. (25). Anti–sAPP-α and anti–sAPP-β were obtained from IBL America (Minneapolis, MN, USA). Anti-CD68 (FA-11) was obtained from Bio-Rad (Hercules, CA, USA). Anti–ionized calcium-binding molecule 1 (Iba-1) was obtained from Wako BioProducts (Richmond, VA, USA).

DNA construct, siRNA, and adenovirus

Monomeric red fluorescent protein (mRFP), cofilin-mRFP, cofilin-S3A-mRFP, and cofilin-S3E-mRFP constructs were obtained from Dr. James Bamburg (Colorado State University, Fort Collins, CO, USA). The phluorin-APP construct was obtained from Dr. Roy Subhojit (University of California–San Diego, La Jolla, CA, USA). The siRNA duplexes (19 nt) targeting cofilin (5′-GGAGGACCUGGUGUUCAUC-3′) were obtained from Dharmacon (Lafayette, CO, USA). Adenoviruses expressing mRFP, cofilin-S3A-mRFP, and cofilin-S3E-mRFP were kind gifts from Dr. James Bamburg.

DNA transfections and adenoviral transductions

DNA plasmids were transiently transfected in CHO-7WD10 cells and BV2 cells using Fugene HD (Promega, Madison, WI, USA) and Opti-MEM I (Thermo Fisher Scientific) according to the manufacturer’s instructions. Lipofectamine 2000 (Thermo Fisher Scientific) and Opti-MEM I were applied for siRNA transfections, and siRNA was transfected twice every 24 h. After 4–6 h transfections, the medium was replaced with new complete medium. Generally, cells were incubated 24–48 h for plasmid transfections and 72 h for siRNA transfections prior to harvest. Adenoviruses of mRFP and cofilin-mRFP variants were transduced to cells for 5 d prior to assessment.

Immunoblotting

Cell lysate or medium was lysed with RIPA lysis buffer (50 mM Tris pH 7.4, 150 mM NaCl, 2 mM EDTA, 1% NP-40, 0.1% sodium dodecyl sulfate). Total protein concentrations were quantified by a colorimetric detection assay (BCA Protein Assay; Pierce, Rockford, IL, USA). Equal amounts of protein lysates were separated by SDS-PAGE and transferred to Immobilon-P membranes (MilliporeSigma). Proteins of interest were probed by primary antibodies and corresponding peroxidase-conjugated secondary antibodies, followed by detection by ECL (Merck, Darmstadt, Germany) and capture using the LAS-4000 imager (GE Healthcare Biosciences, Waukesha, WA, USA).

Cell-free β-secretase assay

In vitro generation of β-secretase cleavage products was performed in a similar manner to that described in Weggen et al. (38) with modifications. Briefly, postnuclear supernatants were prepared from CHO-7WD10 cells, and the membranes were isolated by centrifugation at 20,000 g for 45 min at 4°C, after which membranes were washed and respun. The membranes corresponding to cells from half of a 10-cm dish were then resuspended in 25 μl of assay buffer (150 mM sodium citrate, pH 4.5) for β-secretase activity assay at 37°C for the indicated times.

Aβ uptake assay

Primary microglial cultures or BV2 cells were incubated with FITC–Aβ1–42 oligomers or fibrils for the indicated times and washed 3 times with PBS on ice, and fresh medium was added. FITC–Aβ1–42 fluorescence signal was quenched with trypan blue for 2 min before fixing the cells.

Immunocytochemistry and immunohistochemistry

For immunocytochemistry, cells were washed with PBS and fixed at room temperature for 15 min with 4% paraformaldehyde. After washing with PBS, fixed cells were incubated with blocking solution containing 0.2% Triton X-100 and 3% normal goat serum for 1 h, followed by overnight incubation at 4°C with related primary antibodies. After 3 washes with PBS-Tween (PBS-T), cells were incubated for 1 h with Alexa 488– or Alexa 594–conjugated secondary IgG antibodies (Vector Laboratories, Burlingame, CA, USA). Slides were then washed 3 times with PBST and mounted with fluorochrome mounting solution (Vector Laboratories). For immunohistochemistry, mice were perfused with PBS, and half of the brains were immediately stored at −80°C for biochemical analysis, and the other half were fixed with 4% paraformaldehyde at 4°C for 24 h followed by cryoprotection in 30% sucrose. Sections (30 μm) were blocked using normal goat serum for 1 h and subjected to primary antibodies at 4°C for overnight, followed by secondary antibody (Alexa 594 and 488) incubation for 1 h at room temperature prior to mounting. Images were captured with the Olympus FV10i confocal microscope (Olympus, Tokyo, Japan), and the immunoreactivities were quantified from hippocampus region using ImageJ software (National Institutes of Health, Bethesda, MD, USA). Immunoreactivities were quantitated from every 12th serial section through an entire hippocampus. In immunocytochemistry and immunohistochemistry experiments, all comparison images were acquired with identical laser intensity, exposure time, and filter. Adjustments to the brightness and contrast were applied equally to all comparison images. Regions of interest were chosen randomly, and investigators were blinded to genotypes of mice and experimental conditions during image acquisition and quantification.

Statistical analysis

All graphs were analyzed and made using Prism v.6.0 software (GraphPad Software, La Jolla, CA, USA) using a Student’s t test or 1-way ANOVA, followed by Tukey’s post hoc test. Differences were deemed significant when P < 0.05. All data in graphs were expressed as means ± sem (error bars).

RESULTS

Knockdown of cofilin reduces Aβ production and BACE1-mediated APP processing in CHO-7WD10 cells

We previously showed that cofilin is required to mediate oAβ42–induced loss of surface β1-integrin and downstream neurotoxicity (25), and other studies have demonstrated a link between cofilin activation status and various endocytic processes (32, 39, 40). Because APP endocytosis is critical to BACE1 and presenilin-dependent processing of APP to produce Aβ (6, 41, 42), we reasoned that cofilin might be involved in APP processing and Aβ generation. Thus, we first examined CHO cells stably expressing APP751, so-called CHO-7WD10 cells (6). Transfection of cofilin siRNA in CHO-7WD10 cells significantly reduced the secretion of Aβ by >60%, which was accompanied by a reduction in sAPP-β and increase in sAPP-α, indicative of reduced BACE1-mediated processing of APP. Biotinylation of cell surface proteins showed a significant increase in cell sAPP levels upon cofilin siRNA transfection (Fig. 1A, B), consistent with increased secretion of sAPP-α. To confirm changes in sAPP in a different way, we stained for sAPP ectodomain (6E10) without cell membrane permeabilization followed by cell permeabilization and phalloidin (Alexa Fluor 647) staining to label F-actin. Indeed, we confirmed the significant increase in sAPP intensity upon cofilin knockdown (Fig. 1C, D), which was accompanied by a significant increase in the intensity of F-actin (Fig. 1C, E), consistent with the known role of cofilin in F-actin severing (18). Staining for the intracellular CTF of APP (CT15 antibody) or actin failed to detect specific signals under the same conditions (Supplemental Fig. S1).

Figure 1.

Knockdown of cofilin mitigates Aβ secretion and increases sAPP in CHO-7WD10 cells. CHO-7WD10 cells transfected with control (Cont) or cofilin (Cof) siRNA. A, B) Proteins from medium and cell lysates were subjected to immunoblotting for Aβ, sAPP-α, sAPP-β, total APP, and cofilin (n = 4–6 each). C–E) Cells were fixed and subjected in APP staining using anti-APP (6E10) antibody without cell permeabilization, followed by membrane permeabilization and phalloidin (Alexa Fluor 647) staining to label F-actin. The intensity of green (sAPP) and blue (F-actin) were quantified (n = 3 repeats). Scale bar, 10 μm. F, G) Membranes isolated for cell-free β-secretase activity assay. APP and CTF-β blots after reaction (Rxn) (n = 3 repeats). **P < 0.01, ^#P < 0.001 (Student’s t test).

To further assess the role of cofilin in APP processing, we performed cell-free β-secretase activity assays (pH 4.5) using isolated membranes derived from control and cofilin siRNA–transfected cells. Incubation of control membranes at 37°C for 1–2 h demonstrated a robust appearance of the BACE1-mediated cleavage product of APP, CTF-β (Fig. 1F). However, membranes derived from cofilin siRNA–transfected cells significantly reduced the appearance of CTF-β (Fig. 1F, G), suggesting the loss of APP proximity to BACE1.

Constitutively active cofilin promotes the loss of sAPP, but only WT cofilin promotes Aβ secretion

To examine which activation state of cofilin regulates sAPP levels, we utilized cofilin-mRFP variants representing the constitutively activated cofilin (S3A) and the phosphomimetic inactive form of cofilin (S3E). The S3A mutant constitutively severs F-actin, whereas the S3E mutant is unable to sever F-actin and frequently functions in a dominant-negative manner (43, 44). CHO-7WD10 cells transfected with cofilin-S3A-mRFP showed significantly reduced sAPP and F-actin compared with cells transfected with mRFP control, as detected by APP ectodomain staining without cell permeabilization (Fig. 2A, B) and subsequent permeabilization and phalloidin staining (Fig. 2A, C). In contrast, cells transfected with cofilin-S3E-mRFP consistently exhibited significantly increased sAPP (Fig. 2A, B) and F-actin staining (Fig. 2A, C) compared with S3A-mRFP and control mRFP, indicating that the active form of cofilin promotes the loss of sAPP together with F-actin destabilization. To determine whether the effect of cofilin is specific to APP, we also examined endogenous surface LRP. Like APP, cofilin-S3A-mRFP decreased surface LRP, and cofilin-S3E-mRFP increased surface LRP (Fig. 2D, E), suggesting that the effects of cofilin are generalizable to multiple surface proteins.

Figure 2.

Constitutively activated cofilin reduces sAPP but only WT cofilin enhances Aβ secretion. A, D) CHO-7WD10 cells were transfected with mRFP, S3A-mRFP, or S3E-mRFP for 48 h and were subjected to sAPP (A) or LRP (D) staining without permeabilization, followed by membrane permeabilization and phalloidin (Alexa Fluor 647) staining to label F-actin (A). B, C, E) The green intensity for sAPP (B), blue intensity for F-actin (C), and green intensity for surface LRP (E) were quantified (ANOVA, and Tukey’s post hoc test). ^#P < 0.001 (n = 4 repeats). F) CHO-7WD10 cells were transfected with mRFP control or cofilin (Cof)-mRFP, S3A-mRFP, and S3E-mRFP and blotted from media and lysates for Aβ, sAPP-α, sAPP-β, total APP, and actin using specific antibodies. G) Aβ, sAPP-α, and sAPP-β levels were quantified. Scale bars, 10 μm. (ANOVA and Tukey’s post hoc text). *P < 0.05 (n = 4 repeats).

Given the increase in APP internalization with the constitutively activated form of cofilin, we assessed Aβ and sAPP secretion upon transfection of cofilin-mRFP variants in CHO-7WD10 cells. Transfection of WT cofilin-mRFP significantly increased Aβ secretion by ∼50%, with a corresponding increase in sAPP-β. However, neither cofilin-S3A-mRFP nor cofilin-S3E-mRFP consistently altered Aβ secretion compared with control mRFP (Fig. 2F, G), suggesting that the cofilin activation-inactivation cycle (i.e., increased F-actin dynamics), rather than its constitutive activation or inactivation, is critical for Aβ secretion.

Genetic reduction of cofilin mitigates Aβ secretion and is restored by WT cofilin

To confirm the above results in neuronal cells, we isolated and cultured primary cortical neurons from APP/PS1 and littermate APP/PS1;cofilin^+/− P0 pups. APP/PS1 mice express the APPswe and PS1ΔE9 transgenes driven by the prion protein promoter (34). APP/PS1;cofilin^+/− neurons exhibited a ∼70% reduction in Aβ secretion on DIV20 (Fig. 3A, B), which was associated with a corresponding decrease in secretion of sAPP-β (Fig. 3A, B). Transduction of WT cofilin-mRFP adenovirus but not mRFP control adenovirus restored Aβ and sAPP-β secretion to levels comparable to APP/PS1 neurons, demonstrating a complete rescue of APP processing by WT cofilin.

Figure 3.

Cofilin regulates Aβ secretion, sAPP, and internalization of APP in primary neurons. A, B) Cortical primary neurons derived from APP/PS1 and littermate APP/PS1;cofilin^+/− pups were transduced with cofilin (Cof)-mRFP adenovirus on DIV15 for 5 d. Endo-Cof, endogenous coflin. Proteins from medium and lysates were subjected to immunoblotting for Aβ, sAPP-α, sAPP-β, total APP, actin, and cofilin using specific antibodies (ANOVA and Tukey’s post hoc text; n = 3). C, D) WT cortical primary neurons were cotransfected with the same amount of phluorin-APP + mRFP, S3A-mRFP, or S3E-mRFP DNA for 36 h on DIV6, and cells were fixed after the treatment of the glycine (200 μM) for 0, 10, and 30 min. Scale bar, 10 μm. The mean intensity of phulorin-APP in dendrites was quantified from 30 neurons per condition in 3 independent experiments, and fold changes were normalized to the 0-min condition (repeated measures ANOVA and Bonferroni post hoc test). *P < 0.05, **P < 0.01.

We next assessed whether cofilin activation status alters APP internalization in WT primary neurons. For this purpose, we utilized the phluorin-APP construct, a pH-sensitive indicator previously used to measure APP internalization. Upon stimulation of neurons with glycine, phluorin-APP undergoes rapid endocytosis and localizes to endosomes where the pH is low (pH ∼4.5), thereby quenching phluorin-APP fluorescence (3, 45). In WT primary neurons cotransfected with cofilin-S3A-mRFP and phluorin-APP on DIV6 for 36 h, glycine treatment reduced the mean phluorin-APP fluorescence significantly faster over 10–30 min than neurons transfected with phluorin-APP and mRFP control or cofilin-S3E-mRFP (Fig. 3C, D), indicating that the active form of cofilin promotes APP endocytosis from the cell surface. Taken together, these results show that whereas constitutively activated cofilin promotes APP endocytosis, WT cofilin, which is capable of both activation and inactivation, enhances Aβ secretion.

Genetic cofilin reduction mitigates Aβ deposition and enhances microglial activation in vivo

Our results in CHO-7WD10 cells and APP/PS1;cofilin^+/− primary neurons indicated that cofilin reduction mitigates Aβ production. Hence, we examined APP/PS1 and APP/PS1;cofilin^+/− brains for Aβ deposition at 7 mo of age. Staining brains for Aβ (4G8) showed a ∼60% reduction in Aβ deposition both in the hippocampus and cortex (Fig. 4A, B). Despite the dramatic reduction in Aβ burden in APP/PS1;cofilin^+/− brains, a paradoxical 2-fold increase in Iba-1 immunoreactivity was observed, indicative of increased microglial activation (Fig. 4C, D). Careful examination of z-stacked confocal images showed significantly increased colocalization of Aβ with Iba-1⁺ microglia surrounding Aβ deposits in APP/PS1;cofilin^+/− vs. APP/PS1 mice (Fig. 4E, F), suggesting that the reduction in Aβ deposition in APP/PS1;cofilin^+/− brains may arise in part from increased microglial phagocytosis of Aβ.

Figure 4.

Decreased Aβ burden, increased microglial activation, and increased Iba-1–Aβ colocalization by genetic reduction of cofilin in APP/PS1 mice. A, B) Seven-month-old APP/PS1 and APP/PS1;cofilin^+/− mice stained for Aβ in the hippocampus (HIPPO) and frontal cortex (CTX) and area covered by Aβ (4G8) quantified (n = 4 each). Scale bar, 300 μm. C, D) Seven-month-old APP/PS1 and APP/PS1;cofilin^+/− mice stained for Aβ and Iba-1. Note the generally increased Iba-1⁺ microglia, especially surrounding Aβ deposits in APP/PS1;cofilin^+/− hippocampus (n = 4 each). Scale bar, 200 μm. E) Z-stacked confocal images of Aβ deposits showing increased colocalization of Aβ and Iba-1 in APP/PS1;cofilin^+/− mice. Middle panels amplified from dashed lines and z-stacked right panels from solid lines rotated forward 45°. E, F) Quantification of Aβ and Iba-1 colocalization from z-stacked images (white arrows) of Aβ deposits (n = 4 mice/genotype). Cof, cofilin; DG, dentate gyrus. **P < 0.01, ^#P < 0.001 (Student’s t test).

Cofilin regulates microglial activation and uptake of Aβ42

Because total Aβ accumulation in brain is due not only to production but also to clearance and removal, we next assessed whether cofilin alters the activation of microglia and their Aβ clearance activity. Thus, we isolated primary microglia from WT and cofilin^+/− P0 pups. Treatment of microglia with a low dose of LPS (2 ng/ml), Aβ42 fibrils (fAβ42, 1 μM), or oAβ42 (1 μM) for 18 h showed a dramatic increase (2.5–3-fold) in Iba-1 intensity in cofilin^+/− microglia compared with WT microglia (Fig. 5A, B), indicating that cofilin^+/− microglia are hypersensitive to activation. Moreover, upon 2 h exposure to FITC–oAβ42 (1 μM), cofilin^+/− microglia contained significantly elevated CD68 immunoreactivity, a marker of microglial phagosomes (Fig. 5C, D). To determine potential changes in Aβ42 uptake and clearance, we treated WT and cofilin^+/− microglia with FITC–oAβ42 (1 μM) for 2 h, changed medium, and observed its clearance for 2 and 4 h (Fig. 5E). Indeed, 2 h exposure to FITC–oAβ42 showed that cofilin^+/− microglia take up a significant ∼2-fold more FITC–oAβ42 than WT microglia (Fig. 5C, D). During the clearance period after changing to fresh medium, cofilin^+/− microglia also demonstrated significantly faster clearance of FITC–oAβ42 after both 2 and 4 h (Fig. 5E–G).

Figure 5.

Reduction of cofilin enhances microglial activation and Aβ42 clearance. A, B) Primary microglia derived from WT or cofilin^+/− mice treated with fAβ42, oAβ42, or LPS. Quantification of Iba-1 intensity after treatment with LPS (2 ng/ml), oAβ42 (1 μM), and fAβ42 (1 μM) for 18 h, normalized to WT (n = 5 repeats). C, D) Primary microglia derived from WT and cofilin^+/− mice treated with 1 μM FITC–oAβ42 for 2 h, medium changed, and then subjected to staining for CD68 (n = 4 repeats). E–G) Primary microglia derived from WT and cofilin^+/− mice treated with 1 μM FITC–oAβ42 for 2 h, medium changed, and chased for 2 or 4 h. E) Representative images of FITC–oAβ42 uptake (2 h) and removal (2 and 4 h). F, G) Quantification of FITC–oAβ42 uptake and removal (n = 5 repeats). Removal of FITC–oAβ42 is expressed as percent remaining of initial 2-h uptake. Cof, cofilin. Scale bars, 10 μm. ^#P < 0.001 (Student’s t test).

To determine whether the enhanced uptake of Aβ42 in cofilin^+/− microglia is indeed due to cofilin, we transduced cofilin^+/− microglia with cofilin-mRFP adenovirus. In cofilin^+/− (cofilin-mRFP) microglia treated with FITC–oAβ42 (1 μM) for 2 h, the amount of cellular FITC–oAβ42 uptake was significantly reduced compared with mRFP-only–transduced cofilin^+/− microglia (Fig. 6A, B). Moreover, transduction of cofilin-S3A-mRFP but not cofilin-S3E-mRFP adenovirus significantly reduced FITC–oAβ42 uptake in cofilin^+/− microglia (Fig. 6C, D). This was accompanied by significantly reduced F-actin (phalloidin) intensity by cofilin-S3A-mRFP but not cofilin-S3E-mRFP expression (Fig. 6C, E), indicating that activated cofilin inhibits F-actin stability and Aβ uptake in microglia. Similarly, adenoviral transduction of cofilin-mRFP significantly reduced the uptake of FITC–oAβ42 in BV2 microglial cells (Supplemental Fig. S2A, B). To assess the role of cofilin activation status on the clearance of Aβ42, we transduced BV2 microglia with mRFP, cofilin-S3A-mRFP, or cofilin-S3E-mRFP. We then treated transduced BV2 cells with FITC–oAβ42 for 2 h, changed to fresh medium, and monitored FITC–oAβ42 clearance for 2 and 4 h (Fig. 6F, G). Indeed, cofilin-S3A-mRFP, but not cofilin-S3E-mRFP expression not only reduced the uptake of FITC–oAβ42 but also significantly delayed its clearance compared with mRFP control (Fig. 6F–H), indicating that that activated cofilin inhibits both the uptake and removal of Aβ in microglia. Taken together, our results indicate that cofilin regulates both Aβ production in neurons and Aβ clearance in microglia, both of which likely contribute to the reduced Aβ deposition seen in the APP/PS1;cofilin^+/− brains.

Figure 6.

Activation of cofilin regulates microglial activation and phagocytosis in cofilin^+/− microglia. A, B) Primary microglia derived from cofilin^+/− mice transduced with mRFP or cofilin (Cof)-mRFP adenovirus for 3 d and treated with FITC–oAβ42 (1 μM) for 2 h (Student’s t test). **P < 0.01 (n = 5 repeats). C–E) Primary microglia derived from cofilin^+/− mice transduced with mRFP, S3A-mRFP, or S3E-mRFP adenovirus and treated with FITC–oAβ42 (1 μM) for 2 h. Then cells were fixed, permeabilized, and subjected to phalloidin (Alexa Fluor 647) staining for F-actin (ANOVA, Tukey’s post hoc test), ^#P < 0.001 (n = 5 repeats). F–H) BV2 microglial cells transduced with mRFP, S3A-mRFP, or S3E-mRFP adenovirus and treated with FITC–oAβ42 (1 μM) for 2 h, followed by changing to fresh medium and monitory clearance for 2 and 4 h (ANOVA and Tukey’s post hoc test). **P < 0.01, ^#P < 0.001 (n = 5 repeats). Removal of FITC–oAβ42 is expressed as percent remaining of initial 2 h uptake. Scale bars, 10 μm.

DISCUSSION

Previous studies have shown the significant roles of cofilin in actin dynamics (15, 46), pathogen-triggered endocytic processes (32), mitochondrial dysfunction (25, 47), Aβ–induced neurotoxicity (25, 44), and tauopathy (48). However, a role of cofilin in APP processing and Aβ metabolism has not been investigated. We initiated this study based on our prior observation that cofilin was essential for oAβ42–induced loss of surface β1-integrin (25, 28). In this study, we made a series of novel observations, utilizing cell lines, primary neurons, primary microglia, and genetically modified mice, implicating the dual role of cofilin in Aβ generation and clearance relevant for AD pathogenesis. Specifically, we found that knockdown of endogenous cofilin in CHO-7WD10 cells and primary neurons significantly reduces Aβ production by increasing sAPP levels. Expression of active (S3A) but not inactive (S3E) cofilin reduced sAPP levels coincident with F-actin destabilization and enhanced APP endocytosis. In agreement with results in cultured cells, Aβ deposition in APP/PS1 transgenic mice was significantly reduced by ∼60% with genetic reduction of cofilin (APP/PS1;cofilin^+/−). However, the reduction of Aβ load in APP/PS1;cofilin^+/− mice was paradoxically associated with significantly increased microglial activation (Iba-1⁺) surrounding Aβ deposits, suggesting a role for cofilin^+/− microglia in the clearance of Aβ. Three-dimensional reconstruction of z-stacked confocal images demonstrated significantly increased amount of Aβ colocalized within CD68⁺ microglia in APP/PS1;cofilin^+/− brains. Accordingly, primary microglia isolated from cofilin^+/− mice demonstrated significantly enhanced state of activation and greater ability to uptake and clear Aβ42 associated with increased F-actin stabilization, which was reversed with the active (S3A) but not inactive (S3E) form of cofilin. These results taken together indicate a critical role for cofilin in Aβ accumulation, in addition to its known role in Aβ-induced neurotoxicity, mitochondrial dysfunction, and tauopathy.

The role of actin dynamics in plasma membrane bending and endocytosis has long been known. APP endocytosis occurs via clathrin-coated pits (49, 50). Polymerization of F-actin against the plasma membrane is thought to provide force for membrane bending during clathrin-mediated endocytosis (13). The depolymerization of the endocytic F-actin network then coincides with vesicle uncoating and endocytic scission (51). Although multiple actin-binding proteins control this process, cofilin is a key mediator of F-actin severing and depolymerization, and careful studies in budding yeast have shown that defects in cofilin function reduce actin flux and decrease endocytic internalization at the plasma membrane (39). In mammalian cells, both actin polymerizing and depolymerizing agents similarly slow the recruitment of dynamin 2 needed for membrane scission and clathrin-coated endocytosis (52), indicating a coordinated role for actin dynamics in this process. Our observation that cofilin reduction increases sAPP levels, whereas the constitutively activated cofilin (S3A) reduces sAPP and promotes its endocytosis, is consistent with the notion that cofilin-mediated F-actin depolymerization (or slowed polymerization) and actin monomer recycling may be a rate-limiting step in the endocytosis of APP in neurons and CHO cells (Fig. 7, schematic model). Such a role for cofilin in APP internalization and trafficking is consistent with the observed role of cofilin in actin-mediated endocytic trafficking in Saccharomyces cerevisiae (39). Interestingly, slower local actin polymerization (or faster depolymerization) favors membrane invagination (13), a critical step in clathrin-mediated endocytosis. This model is consistent with increased APP internalization and reduced F-actin level with activated or dephosphorylated cofilin (S3A) (Fig. 7, schematic model). However, we also observed that WT cofilin but not cofilin-S3A or cofilin-S3E could enhance Aβ secretion, indicating that both the activation and inactivation phases of cofilin cycling contribute to Aβ secretion. F-actin dynamics play important roles not only in endocytosis but also in other components of membrane trafficking, including vesicle exocytosis and recycling (51), both of which also affect APP processing (50). Intriguingly, interaction of viruses with their cognate receptors on the cell surface also triggers the biphasic regulation of cofilin, initially by inactivation (phosphorylation) followed by activation (dephosphorylation), which coincides with the assembly, then disassembly, of the F-actin network and subsequent entry into host cells (32).

Figure 7.

Schematic model of cofilin in APP internalization in neurons and Aβ uptake and clearance in microglia. A) In neurons and other nonphagocytic cells, destabilization of F-actin and faster APP endocytosis by activatedor dephosphorylated cofilin (S3A) is consistent with the model in which slower local actin polymerization (or faster depolymerization) favors membrane invagination (13), a critical step in clathrin-mediated endocytosis. B) In phagocytic cells such as microglia, stabilization of F-actin and increased efficiency of Aβ42 uptake and clearance by the loss of cofilin or expression of inactive phosphorylated cofilin (S3E) are consistent with the model in which faster local actin polymerization (or increased F-actin) favors membrane protrusion rather than invagination (13), which is necessary for formation of phagocytic cups and macropinocytic protrusions to mediate phagocytosis and micropinocytosis, respectively.

In alveolar macrophages, phagocytosis of the C. albicans fungal pathogen is positively regulated by secretion of leukotrienes, which inactivates cofilin, promotes F-actin assembly, and enhances phagocytic activity (33). On the other hand, generation of prostaglandin E2, which promotes cofilin activation, inhibits the ingestion and phagocytosis of C. albicans by alveolar macrophages (33). Moreover, inactivation of cofilin also contributes to the phagocytosis of Cryptococcus neoformans, another common fungal pathogen (53) These examples are akin to our observation that cofilin reduction or expression of the inactive (S3E) but not the active (S3A) cofilin promotes activation of microglia as well as the uptake and clearance of Aβ42 associated with increased F-actin levels. Although this study did not conduct a detailed investigation regarding the mode of Aβ42 clearance (i.e., phagocytosis, micropinocytosis, or both), LPS, fAβ42, and oAβ42 all similarly induced enhanced activation of cofilin^+/− microglia. Whether via phagocytosis or macropinocytosis, it has been shown that oligomeric and fibrillar Aβ are internalized largely via receptor-mediated mechanisms (54). In phagocytic cells such as macrophages and microglia, we postulate that the F-actin assembly driven by cofilin inactivation is a rate-limiting step in the endocytic process, because receptor clustering at the attachment site generates signals that in turn leads to actin polymerization associated with membrane protrusions. This rich F-actin network provides the force needed for membrane bending or the formation of phagocytic cups or macropinocytic protrusions. Indeed, models of membrane-bending forces by actin illustrate that faster local actin polymerization (hence increased F-actin) favors membrane protrusion rather than invagination (13), which is necessary for formation of phagocytic cups and macropinocytic protrusions to mediate phagocytosis or micropinocytosis (Fig. 7, schematic model). Hence, our data, which show that increased F-actin (by the loss of cofilin or S3E expression) is associated with enhanced Aβ42 uptake and clearance in microglia, are consistent with this model of increased local actin polymerization favoring membrane protrusion rather than invagination (Fig. 7). Our observation that cofilin^+/− microglia are highly sensitive to activation by oAβ42 or fAβ42 suggests that increased preservation or polymerization of the F-actin network in these cells also readily avails receptor clustering for microglial activation and protrusive membrane bending at sites of Aβ42 engagement. However, it is important to point out that cofilin^+/− microglia still contain endogenous cofilin activity, and overall actin dynamics is critical to Aβ endocytic activity, because complete pharmacological blockade of either actin assembly or disassembly can inhibit Aβ uptake (55).

Previous studies have shown that brains of patients with AD and APP transgenic mice contain significantly increased levels of activated cofilin (25–27). Our findings indicate that reduction of cofilin level and activation are associated with ∼60% reduced Aβ deposition in brain (i.e., APP/PS1;cofilin^+/−). Cofilin not only increases Aβ generation in neurons and inhibits microglia-mediated Aβ clearance (this study) but also mediates Aβ-induced neurotoxicity (25, 44), because cofilin reduction mitigates Aβ42–induced mitochondrial and synaptic dysfunction (25) as well as tauopathy (48). Intriguingly, cofilin activation appears to be a common link mediating both Aβ accumulation and tauopathy, the latter driven by displacement of τ from microtubules (48). Moreover, activation of cofilin is a necessary step in the formation of cofilin-actin pathology (23, 46), which is significantly increased in brains of patients with AD and APP/PS1 transgenic mice (21–24). Hence, partially reducing cofilin levels or activity could be a therapeutic strategy to slow multiple AD pathologies and preserve the synaptic actin network. It is well established that cofilin is activated by SSH1 or chronophin (17, 56) and deactivated by LIMK1 (16). Although complete blockade of cofilin activation is unwarranted, partially reducing cofilin levels, enhancing LIMK1 activity, and blocking the cofilin-activating enzymes SSH1 and chronophin may be viable strategies to mitigate multiple facets of AD pathogenesis.

Glossary

Aβ

amyloid-β

AD

Alzheimer’s disease

APP

amyloid precursor protein

BACE1

β-site APP-cleaving enzyme

CHO

Chinese hamster ovary

CTF

C-terminal fragment

DIV

days in vitro

F-actin

filamentous actin

fAβ42

Aβ42 fibril

FBS

fetal bovine serum

Iba-1

ionized calcium-binding molecule 1

LIMK1

LIM kinase 1

LRP

low-density lipoprotein receptor–related protein

mRFP

monomeric red fluorescent protein

oAβ42

Aβ42 oligomer

P/S

penicillin-streptomycin

PS1

presenilin 1

sAPP

surface APP

siRNA

small interfering RNA

SSH1

slingshot homolog 1

USF

University of South Florida

WT

wild type

AUTHOR CONTRIBUTIONS

T. Liu designed the research, performed experiments and data analysis, and wrote the manuscript; J.-A. A. Woo contributed to critical experiments, research design, data interpretation, and data analysis; Y. Yan, P. LePochat, and M. Z. Bukhari performed experiments; D. E. Kang supervised the whole study, assisted in the design of all experiments and interpretation of data, and wrote the manuscript; and all authors read and approved the final manuscript.

Supplementary Material

This article includes supplemental data. Please visit http://www.fasebj.org to obtain this information.