PLGA Nanoparticle-based Anti-TLR2 scFv Gene Delivery for the Treatment of Alzheimer’s Disease

Abstract

In Alzheimer’s disease (AD), persistent microglial neuroinflammation and the poor brain exposure and durability of current therapies underscore the need for new, long-acting treatments. We developed a non-viral gene therapy that suppresses microglial Toll-like receptor 2 (TLR2) signaling using poly(lactic-co-glycolic acid) (PLGA) nanoparticles (NPs) loaded with a plasmid encoding the anti-TLR2 single-chain variable fragment (scFv33). Following intra–cisterna magna delivery, PLGA NPs exhibited microglia-biased uptake and enabled brain-wide transgene expression in mice. In 5xFAD mice, a single administration of scFv33 NPs improved recognition memory in the novel object recognition (NOR) assay, outperforming 8 weeks of weekly recombinant scFv33-Fc dosing. Histology showed selective reduction of small hippocampal Aβ plaques and a shift toward a ramified microglial morphology, indicative of reduced activation. In primary neuron–microglia co-culture, scFv33 reduced microglial hypertrophy, restored process complexity, and enhanced Aβ phagocytosis. Together, these data indicate that sustained, local expression of an anti-TLR2 scFv via a clinically translatable PLGA platform recalibrates microglial state and preferentially limits early-stage plaque accumulation, yielding cognitive benefit after a single dose.

INTRODUCTION

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder marked by cognitive decline, extracellular β-amyloid (Aβ) deposition, and tau pathology [1, 2]. Despite its global burden, current pharmacotherapies—cholinesterase inhibitors and the NMDA receptor antagonist memantine—provide only symptomatic relief and do not alter disease progression [3-5]. Moreover, adverse effects—including gastrointestinal disturbances, dizziness, and bradycardia—further limit long-term use. Recently approved anti-amyloid antibodies such as lecanemab and donanemab are limited by amyloid-related imaging abnormalities (ARIA), including cerebral edema and microhemorrhages, and by the need for frequent intravenous administration, which reduces treatment accessibility and adherence [6-8]. Across active and passive Aβ immunotherapies, substantial amyloid lowering has not consistently translated into meaningful clinical benefit, suggesting that Aβ clearance alone may be insufficient as a central therapeutic strategy in AD [9-11]. Accordingly, novel therapies that target core disease mechanisms with improved safety, durability, and accessibility are urgently needed.

Against this backdrop, attention has increasingly shifted to neuroinflammation as a disease-driving process and therapeutic target. Chronic neuroinflammation is a defining feature of AD. Acute inflammatory responses are protective, but when sustained and excessive these become pathogenic, thereby driving neuronal dysfunction and neuronal loss [12-14]. Extensive evidence indicates that the dysregulated secretion of pro-inflammatory cytokines—including IL-1β, IL-6, TNF-α—is induced by mid-life and remains high thereafter, driving sustained neuroinflammation and accelerating neuronal injury [15-17].

Microglia, the resident immune cells of the CNS, play a central role in initiating and sustaining neuroinflammation [18]. Following tissue stress or protein aggregation, microglia initiate immune responses by releasing pro-inflammatory cytokines, chemokines, and reactive oxygen and nitrogen species (ROS/RNS). Morphologically, activation is accompanied by stereotyped remodeling: homeostatic microglia display small somata and highly ramified, thin processes, whereas activated cells exhibit enlarged somata, retracted/thickened processes, and amoeboid profiles [19]. Such morphological remodeling signals a maladaptive shift to a pro-inflammatory state: process retraction narrows parenchymal surveillance and synaptic support, while soma enlargement with lysosomal upregulation accompanies heightened cytokine/ROS output, aberrant synapse engulfment, and—when persistent—impaired Aβ-handling efficiency [20-24]. Many previous studies support a causal role for microglia activation and neuroinflammation in AD: microglial activation fosters Aβ aggregation, IL-1β drives tau phosphorylation, and elevated baseline indices of microglial activation predict tau propagation and cognitive deterioration [25-27].

A critical component of microglial activation in immune responses involves pattern-recognition receptors, such as Toll-like receptors (TLRs) [28, 29]. These receptors detect pathogen-associated molecular patterns (PAMPs) and damage-associated molecular patterns (DAMPs). Among the TLR family, TLR2 is prominently expressed on microglia [30, 31]. Upon ligand binding, TLR2 activates MyD88–NF-κB/MAPK signaling, inducing pro-inflammatory cytokines (e.g., TNF-α, IL-1β, IL-6). In transgenic AD models, genetic or pharmacologic attenuation of TLR2 signaling reduces glial reactivity, improves Aβ handling, and ameliorates memory deficits, whereas sustained TLR2 activation promotes a maladaptive microglial phenotype characterized by excessive inflammation and impaired phagocytosis [32-34]. Collectively, these observations nominate microglial TLR2 as a rational therapeutic target in AD.

Accordingly, we adopted a PLGA nanoparticle-mediated anti-TLR2 antibody gene-delivery strategy in the 5xFAD mouse model. For the anti-TLR2 antibody, we used a recombinant single-chain variable fragment (scFv), developed and validated in our previous study [35]. By genetically linking VH and VL domains via a flexible peptide and lacking an Fc region, scFvs exhibit enhanced functional selectivity and facilitate cost-effective engineering and production [36, 37]. To enable practical CNS delivery, we encapsulated the anti-TLR2 scFv expression cassette in poly(lactic-co-glycolic acid) (PLGA) nanoparticles for non-viral intrathecal administration. The clinical safety profile of PLGA is supported by multiple FDA-approved depot formulations and its widespread use as a biocompatible, biodegradable polymer [38-41]. In our prior work, PLGA nanoparticles were efficiently internalized by CNS microglia and delivered genetic material, including the anti-TLR2 scFv expression cassette [42-44].

Here, we validated the therapeutic potential of a PLGA NP-mediated anti-TLR2 scFv gene therapy for AD. We evaluated PLGA NP-mediated gene delivery and expression in the mouse brain. We then demonstrated that sustained scFv expression in the brain attenuated microglial activation, enhanced microglial Aβ phagocytosis, reduced hippocampal Aβ pathology, and improved cognition in 5xFAD mice. Together, our findings suggest that delivery of the anti-TLR2 scFv gene via a clinically translatable, non-viral PLGA carrier can recalibrate microglial signaling to achieve durable immunomodulation, thereby addressing key limitations of conventional therapies in AD.

MATERIALS AND METHODS

Immunocytochemistry (ICC)

Cells were rinsed with sterile 0.1 M PBS and fixed in 2% paraformaldehyde for 10~15 min on ice. After fixation, cells were washed with 0.1 M PBS, permeabilized with 0.1% Triton X-100 in 0.1 M PBS, and then blocked for 1 h in 0.1 M PBS containing 5% normal donkey serum, 2% BSA, and 0.1% Triton X-100. After blocking, cells were incubated overnight at 4°C with the following primary antibodies: mouse anti-His-Tag (Santa Cruz Biotechnology, Dallas, TX, USA), rabbit anti-Iba1 (1:1,000; Wako, Japan), mouse anti-β-Amyloid, 1-16 (1:1,000; Biolegends, San Diego, CA, USA), rat anti-CD68 (1:1,000; Bio-Rad Laboratories, Herucles, CA, USA), and FITC conjugated goat anti-human IgG FcR (1:200; Jackson ImmunoResearch, West Grove, PA, USA). The following day, samples were washed three times in PBS and incubated for 2 h at room temperature with secondary antibodies labeled with FITC, Cy3, Alexa Fluor 647 (1:200; Jackson ImmunoResearch). After five additional PBS washes, fluorescent images were obtained using a fluorescence microscope (DP72 / Olympus BX51; Olympus, Tokyo, Japan) or a confocal microscope (LSM800; Carl Zeiss). Quantification of microglial activation and phagocytosis activity was performed using IMARIS (Version 9.8.0, Oxford Instruments, Abingdon, UK).

Immunohistochemistry (IHC)

Mice were transcardially perfused with 0.1 M phosphate buffer (pH 7.4), followed by 4% paraformaldehyde. The brain was dissected and post-fixed overnight in the same fixative at 4°C. Tissues were then cryoprotected in 30% sucrose for at least 48 h, embedded, and sectioned coronally at 40 μm thickness using a cryostat (CM1860; Leica, Wetzlar, Germany). Tissue sections were blocked for 1 h at room temperature in a PBS containing 5% normal donkey serum, 2% BSA, and 0.2% Triton X-100. Samples were then incubated overnight at 4°C with the following primary antibodies: rabbit anti-Iba1 (1:1,000; FUJIFILM Wako, Osaka, Japan), mouse anti-β-Amyloid, 1-16 (1:1,000; BioLegend). Following three washes in 0.1 M PBS, sections were incubated for 2 h at room temperature with secondary antibodies conjugated to FITC and Cy3 (1:200; Jackson ImmunoResearch). Following five washes in 0.1 M PBS, stained sections were mounted on glass slides using Vectashield mounting medium (Vector Laboratories, Burlingame, CA, USA). Fluorescent images were obtained using a confocal microscope (LSM800; Carl Zeiss). Image analysis for Aβ plaque counting was conducted by imageJ. Glial cell activation was analyzed using IMARIS (Version 9.8.0, Oxford Instruments).

Western blot

For cell lysate preparation, transfected cells were lysed with RIPA buffer containing protease inhibitors (1 mM), sodium orthovanadate (1 mM), and sodium fluoride (1 mM). After incubation on ice for 10 min, the cell suspension was collected, centrifuged at 12,000 rpm for 10 min at 4°C, and the supernatant was transferred to a fresh tube. For analysis of secreted protein, 400 μl of culture medium was mixed with 1.6 ml of ice-cold acetone (final 4:1 ratio), vortexed briefly, and incubated at -20°C for 1 h. Samples were then centrifuged at 12,000 rpm for 15 min at 4°C. The supernatant was discarded, and the protein pellet was air-dried for 10 min. The pellet was resuspended in 200 μl of lysis buffer, vortexed, and briefly sonicated. After an additional centrifugation step (12,000 rpm, 15 min, 4°C), the final supernatant was collected, avoiding any insoluble debris. Equal amounts of protein samples were mixed with 5× SDS sample buffer in lysis buffer and boiled. Proteins were separated on 12% SDS-PAGE gels and transferred onto 0.22 μm nitrocellulose membranes (GE Healthcare, Chicago, IL, USA) at 300 mA for 1 h. Membranes were blocked in 5% skim milk in TBS with 0.1% Tween-20 for 1 h at room temperature, followed by overnight incubation at 4°C with mouse anti-His antibody (1:1,000; Santa Cruz Biotechnology). The next day, membranes were washed, incubated with goat anti-mouse IgG-HRP (Abfrontier, Seoul, Republic of Korea), and developed using WESTSAVE^TM Gold ECL Solution (Abfrontier).

PLGA nanoparticle (NP) synthesis

PLGA NPs encapsulating pDNA were synthesized based on a previously described double emulsion solvent evaporation method [44, 45]. Briefly, 100 μg of pDNA dissolved in 200 μl of TE buffer (pH 8.0) was added dropwise into 1 ml of dichloromethane (DCM; Samchun Chemicals, Seoul, Republic of Korea) containing 20 mg of PLGA (Resomer^® RG 502H; Sigma-Aldrich, St. Louis, MO, USA). The mixture was emulsified by probe sonication (Vibra-Cell^TM VCX 130; Sonics, Newtown, CT, USA) at 40 W for 1 min to generate a primary water-in-oil (W1/O) emulsion. This primary emulsion was then transferred into 2 ml of 1% polyvinyl alcohol (Alfa Aesar, Ward Hill, MA, USA) and sonicated for 2 min to form a water-in-oil-in-water (W1/O/W2) double emulsion. The resulting mixture was diluted with 6 ml of 1% polyvinyl alcohol solution containing 12.5 mM CaCl₂ and stirred at room temperature for 3 h in a fume hood to allow DCM evaporation. Synthesized NPs were collected by centrifugation at 15,000×g for 15 min at 4°C (Hanil Scientific Inc., Daejeon, Korea), washed twice with deionized water, and lyophilized for storage. The PLGA-FITC NPs were synthesized using ITC-conjugated PLGA (Poly(lactide-co-glycolide)-fluorescein, Sigma-Aldrich) following same protocol without DNA loading.

Nanoparticle characterization

The hydrodynamic diameter and zeta potential of the NPs were determined by dynamic light scattering (DLS) using a Zetasizer Nano-ZS instrument (Malvern Instruments, Malvern, UK). Encapsulation efficiency (EE) was calculated by (DNA_total-DNA_supernatant-DNA_wash1-DNA_wash2)/DNA_initial. DNA_initial represents the total amount of pDNA initially added during NP preparation. DNA_total represents the encapsulated pDNA within PLGA NPs and was obtained by dissolving synthesized NPs in 0.1 N NaOH containing 0.5% SDS. After the NP synthesis, the particles were pelleted by centrifugation following the solvent evaporation step, and the unencapsulated DNA remaining in the post-synthesis supernatant (DNA_supernatant) was quantified. During the subsequent two washing steps, NPs were again pelleted by centrifugation, and DNA present in the supernatants from each wash (DNA_wash1 and DNA_wash2) was similarly measured.

Surface plasmon resonance (SPR)

For TLR2 protein coating on a gold sensor chip (IMSPR-ProX, ICluebio, Seoul, Korea), the surface was first stabilized and a baseline was established by flowing the running buffer (PBST, 0.05% Tween-20) for 2 min. The COOH-Au chip (Icluebio, Ansan, Republic of Korea) was then activated using the Amine Coupling Kit (Icluebio), which contains NHS solution, EDC, and ethanolamine-HCl quenching buffer. After pre-injection of borate buffer (30 s at 50 μl/min; Polysciences, Warrington, PA, USA), a freshly prepared 1:1 mixture of EDC (0.4 M, prepared in acetate buffer provided in the kit) and NHS solution (0.1 M, supplied in the kit) was injected (30 s at 50 μl/min) to activate the carboxyl groups on the chip surface. Recombinant human TLR2 protein was diluted in acetate buffer (pH 4.0; 15 μg in 300 μl) and injected to covalently immobilize the ligand onto the sensor surface (20 min at 10 μl/min). The surface was subsequently blocked with 100 μg/ml BSA in acetate buffer (pH 4.0) and quenched with ethanolamine-HCl quenching buffer (5 min at 50 μl/min). Unbound materials between each step were removed by running buffer (30 s at 50 μl/min). Chip regeneration was performed using glycine-HCl buffer (pH 2.5) between analyte injections. Antibody samples were prepared in PBST (0.05% Tween-20) and injected at concentrations ranging from 0 to 500 nM in two-fold serial dilutions (60 s at 50 μl/min), followed by a dissociation phase for 5 min and chip regeneration for 3 min. Sensorgrams (CH1–CH2) were collected in real-time and analyzed using integrated kinetic analysis software. Kinetic parameters, including the association rate constant (Ka), dissociation rate constant (Kd), and equilibrium dissociation constant (K_D), were calculated using a 1:1 Langmuir binding model.

Animals

All animal procedures were conducted with approval from the Institutional Animal Care and Use Committee (IACUC) of Seoul National University. Male C57BL/6J mice (8~10 weeks old) were purchased from Daehan Biolink (DBL, Eumsung, Korea). 5xFAD transgenic mice on a C57BL/6 background [B6SJL-Tg(APPSwFlLon,PSEN1M146LL286V)6799Vas/Mmjax; RRID:MMRRC_034840-JAX] were provided by the Korea Research Institute of Bioscience and Biotechnology (KRIBB, Daejeon, Republic of Korea). They were housed under specific pathogen-free (SPF) conditions at 22~24°C and 55% relative humidity, with a 12-h light/dark cycle. Food and water were provided ad libitum. All experiments were carried out in accordance with the guidelines of the International Association for the Study of Pain (IASP).

Quantitative real-time RT-PCR (qRT-PCR)

RNA was extracted using TRIzol^TM Reagent (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA), and cDNA was synthesized according to the manufacturer’s protocol. Quantitative real-time RT-PCR was performed using 40 ng of cDNA per reaction with SYBR^TM Green PCR Master Mix (Thermo Fisher Scientific) on a StepOnePlus^TM Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). The amplification program was as follows: initial denaturation at 95°C for 15 min; two cycles of 94°C for 15 sec and 49°C for 15 sec; followed by 32 cycles of 94°C for 15 sec, 62°C for 15 sec, and 74°C for 15 sec with signal acquisition during the extension step; and a final step at 84°C for 10 sec and 88°C for 15 sec. The following PCR primer sequences were used: GAPDH forward, 5′-AGT ATG ACT CCA CTC ACG GCA A-3′; GAPDH reverse, 5′-TCT CGC TCC TGG AAG ATG GT-3′; TNF-α forward, 5′-GGC TCT TCT GGA TCT TGG TG-3′; TNF-α reverse, 5′-TTT CAT GGC TGC TGT GAG TC-3′; IL-1β forward, 5′-TTG TGG CTG TGG AGA AGC TGT-3′; IL-1β reverse, 5′-AAC GTC ACA CAC CAG CAG GTT-3′; IL-6 forward, 5′-TCC ATC CAG TTG CCT TCT TGG-3′; IL-6 reverse, 5′-CCA CGA TTT CCC AGA GAA CAT G-3′. Gene expression levels were normalized to mouse GAPDH and expressed as fold changes relative to the control group, calculated using the 2^-ΔΔCt method, as previously described [46].

Intra-cisterna magna injection

Before injection, mice were anesthetized with isoflurane in an O₂ carrier (induction 2% and maintenance 1.5%). A total volume of 10 μl was administered using a 50 μl Hamilton syringe (Hamilton Company, Reno, NV, USA) with a 30-gauge needle. The needle was inserted percutaneously into the cisterna magna with the head fixed in a flexed position by hand. A total volume of 10 μl was slowly injected over 30 sec. The needle was left in place for additional 10 sec before removal, and then direct pressure was applied to the puncture site.

Flow cytometry

To collect the sample for flow cytometry, the brain tissue from cerebellum, hippocampus, striatum, and prefrontal cortex was separately removed and homogenized into a single cell suspension. Cells were washed with ice-cold PBS containing 2% FBS and incubated with Fc BlockerTM (BD Bioscience, San Jose, CA, USA) for 10 min at 4°C prior to staining. Then, cells were stained with the following antibodies; CD11b-APC (Biolegend), ACSA-2-PE (Miltenyi Biotec, Bergisch Gladbach, Germany), Thy-1.2 violet (Biolegend). Cellular uptake of PLGA-FITC NPs was analyzed using a BD FACSVERSE flow cytometer (BD Bioscience). Data were analyzed using BD FACSuite v1.2 (BD Biosciences).

Novel object recognition (NOR) test

NOR test was conducted as previously described [47], with minor modification. Briefly, mice were habituated for 10 min to a 30 cm×30 cm open-field arena made of opaque acrylic. On the following day, animals were exposed to two identical objects for 10 min during the familiarization phase, followed by a 3-h retention interval. In the test phase, one of the familiar objects was replaced with a novel object, and exploration time was recorded. A discrimination index was calculated as the difference in time spent exploring the novel versus familiar object, divided by the total exploration time. The exploration time for each object was automatically recorded using an automated video tracking system (SMART 3.0; Panlab, Harvard Apparatus, Holliston, MA, USA).

Primary neuron and microglia co-culture

Primary hippocampal neurons were isolated from embryonic day 17~18 mouse embryos. The dissected hippocampi were treated with 0.125% trypsin for 15 min at 37°C, followed by gentle trituration to obtain a single-cell suspension. The cell suspension was filtered through a 70 μm cell strainer, centrifuged at 500×g for 5 min, and resuspended in Neurobasal medium supplemented with 2% B27, 2 mM L-glutamine, and 0.5× penicillin–streptomycin. Neurons were seeded onto poly-L-ornithine (PLO)-coated glass-bottom dishes at a density of 1~5×10⁵ cells per well and maintained at 37°C in a humidified atmosphere with 5% CO₂. To limit glial overgrowth, cytosine β-D-arabinofuranoside (AraC; 0.5 μM) was added 24 h after plating. Primary mixed glial cultures were prepared from the whole brains of postnatal day 1~2 neonatal mice. Brain tissues were dissociated and cultured in high-glucose DMEM (4.5 g/L) supplemented with 10% fetal bovine serum (FBS), 10 mM HEPES, 2 mM L-glutamine, 1× penicillin–streptomycin, and 1× non-essential amino acids. Glial cultures were incubated at 37°C with 5% CO₂, and the culture medium was refreshed every 5 days. After 15 days in vitro, microglial cells were detached using 0.05% Trypsin–EDTA and seeded onto neuron- containing dishes at a microglia-to-neuron ratio of 1.5:1. Co-cultures were stabilized for 2 days prior to further experimental manipulation.

Statistics and illustrations

All statistical analyses were performed using GraphPad Prism version 7.0 for Windows (GraphPad Software Inc., La Jolla, CA, USA). Some of the figures in this study were created using BioRender.com, a web-based science illustration tool. Plasmid map was generated using SnapGene (from Insightful Science; available at snapgene.com).

RESULTS

Validation of scFv33 formats for protein and gene therapy

Alzheimer’s disease (AD) is characterized by chronic and sustained neuroinflammation and neurotoxicity, necessitating interventions with durable activity. We therefore adopted a PLGA NP–mediated gene therapy to drive sustained in vivo expression of our anti-TLR2 antibody, scFv33 [35]. To compare the gene therapy with recombinant protein administration, we engineered an Fc-fused format (scFv33-Fc) to prolong in vivo half-life [48]. Recombinant scFv33 and scFv33-Fc produced in E. coli and purified by Ni-NTA migrated at the expected sizes on SDS–PAGE (Fig. 1a). Surface plasmon resonance confirmed preserved TLR2 binding, with scFv33-Fc exhibiting higher affinity than scFv33 (K_D≈7.2 nM vs 72 nM; Fig. 1b). In THP-1, a TLR2-expressing human monocytic cell line, immunocytochemistry detected membrane-associated anti-IgG signal consistent with cell-surface binding of scFv33-Fc (Fig. 1c). Functionally, pretreatment with either scFv33 or scFv33-Fc significantly reduced Pam3CSK4-induced TNF-α, IL-1β, and IL-6 mRNA expression in BV2 microglia (Fig. 1d).

Fig. 1.

Validation of scFv33-Fc and comparison with scFv33. (a) SDS–PAGE of recombinant scFv33 and scFv33-Fc expressed in E. coli with IPTG induction and purified using a Ni-NTA column. (b) Surface plasmon resonance analysis of binding to TLR2. Equilibrium dissociation constants (K_D) are indicated. (c) Immunocytochemistry showing cell-surface binding of scFv33-Fc (10 µg/ml, 2 h, RT) detected by anti-IgG staining in THP-1 cells. (d) qRT-PCR of TNF-α, IL-1β, and IL-6 mRNA expression in BV2 cells pre-treated with scFv33 or scFv33-Fc (10 µg/ml, 1 h) followed by Pam3CSK4 (50 ng/ml, 3 h); the control group was treated with PBS for both pre- and post-treatment. Data represent mean±SEM from 2 technical replicates. Statistical analysis was performed using one-way ANOVA with Dunnett’s post hoc test (vs. PBS-treated control).

Using our previously validated mammalian scFv33 expression vector [35] for in vivo delivery (Fig. 2a), we confirmed transgene expression and secretion in HEK293T cells. For comparison, we included a control plasmid (CTL) that expresses only the reporter EGFP. EGFP fluorescence confirmed successful transfection, while 6His-tag immunostaining revealed strong intracellular expression of scFv33 (Fig. 2b). Western blotting detected scFv33 in both cell lysates and culture supernatants, indicating efficient expression and secretion (Fig. 2c). To assess cell-intrinsic activity of endogenously produced scFv33, we generated a BV2 cell line stably expressing scFv33 (Fig. 2d, e). Following Pam3CSK4 stimulation, qRT-PCR showed a significant reduction in TNF-α mRNA induction in scFv33-expressing cells compared with CTL (Fig. 2f). Taken together, these data verify that the validated scFv33 vector drives robust mammalian expression and secretion and that transgene expression attenuates TLR2-dependent cytokine induction, supporting subsequent in vivo studies.

Fig. 2.

Construction and functional validation of the scFv33 expression system. (a) Schematic representation of the control plasmid (CTL) and scFv33 plasmid used for in vivo delivery. (b) Immunocytochemical analysis of HEK293T cells transfected with CTL or scFv33 constructs, 48 h post-transfection. (c) Western blot analysis of cell lysates and culture supernatants from HEK293T cells transfected with CTL or scFv33 plasmids. (d) Immunocytochemical analysis of a BV2 stable cell line expressing scFv33. (e) qRT-PCR validation of scFv33 mRNA expression in the scFv33 BV2 stable cell line relative to CTL. (f) Relative mRNA expression of pro-inflammatory cytokines TNF-α and IL-1β in CTL and scFv33 BV2 stable cell lines after Pam3CSK4 (50 ng/ml, 3 h) stimulation, analyzed by qRT-PCR. Data represent mean±SEM from 3 biological replicates. Statistical analysis was performed using one-way ANOVA with Dunnett’s post hoc test (**p<0.01 vs. PBS-treated group).

Cisterna magna delivery of scFv33 PLGA nanoparticles achieves brain expression

To enable plasmid delivery, we formulated PLGA NPs encapsulating either the CTL or the scFv33 expression cassette. Scanning electron microscopy images revealed uniformly spherical particles with smooth surfaces (Fig. 3a). Encapsulation efficiencies were high (CTL, ~75.9%; scFv33, ~72.5%), indicating efficient plasmid incorporation (Fig. 3b). Dynamic light scattering measurements showed hydrodynamic diameters of ~311.2±12.1 nm (CTL) and ~308.43±1.59 nm (scFv33), with mildly negative zeta potentials (−5.04±0.48 and −7.26±0.25 mV, respectively) (Fig. 3b). Functionally, scFv33 NPs mediated in vitro gene transfer in HEK293T cells, yielding EGFP fluorescence and co-localized 6His immunostaining 48 h post-transfection (Fig. 3c). Together, these data establish the PLGA formulation as a competent vehicle for delivering the scFv33 expression vector in vitro, with suitable size/charge and high loading efficiency.

Fig. 3.

PLGA nanoparticle formulation and in vitro gene delivery of scFv33. (a) Scanning electron microscopy images of PLGA nanoparticles (NPs) loaded with control plasmids (CTL) or scFv33 plasmids. (b) Physicochemical characterization and encapsulation efficiency (EE) of CTL NP and scFv33 NPs. (c) Immunocytochemistry of HEK293T cells 48 h after transfection with CTL or scFv33 NPs.

For in vivo CNS delivery, we administered PLGA NPs via intra–cisterna magna injection (Fig. 4a). At 24 h after injection of FITC-labeled PLGA NPs, flow cytometry revealed preferential uptake by CD11b⁺ microglia across brain regions, with minimal uptake by ACSA-2⁺ astrocytes or Thy-1⁺ neurons (Fig. 4b). Mice receiving scFv33 NPs showed robust transgene expression by qRT-PCR 5 days post-injection, detectable in the prefrontal cortex (PFC), striatum (ST), hippocampus (HPC), and most prominently in the cerebellum (CB) relative to saline controls (Fig. 4c). Together, these data demonstrate that cisterna magna delivery enables brain-wide access of PLGA NPs with microglia-biased cellular uptake and supports regional expression of the scFv33 transgene in vivo.

Fig. 4.

In vivo gene delivery and cellular uptake of PLGA nanoparticles following intra-cisterna magna injection. (a) Schematic illustration of cisterna magna administration of PLGA NPs in mice. (b) Flow cytometric analysis of PLGA–FITC NP uptake in each brain region 24 h post-injection, showing the percentage of FITC⁺ cells among CD11b⁺ microglia, ACSA-2⁺ astrocytes, and Thy-1⁺ neurons. (c) qRT-PCR analysis of scFv33 mRNA expression in different brain regions—prefrontal cortex (PFC), striatum (ST), hippocampus (HPC), and cerebellum (CB)—5 days post-injection of scFv33 NPs compared with saline-treated controls.

Single-dose scFv33 nanoparticles enhance recognition memory compared with repeated protein dosing

We next tested whether scFv33 improves recognition memory in 5xFAD mice and whether PLGA NP–mediated gene delivery offers advantages over protein administration. Recognition memory was assessed using the novel object recognition (NOR) assay (Fig. 5a). Mice received intra–cisterna magna injection of scFv33-Fc (2 µg) once weekly for 8 weeks (Fig. 5b) or a single injection of scFv33 NPs (200 µg) (Fig. 5c). scFv33-Fc injection induced a trend toward improvement in the discrimination index versus saline-treated 5xFAD mice (Fig. 5b). In contrast, a single scFv33 NP injection significantly increased the discrimination index relative to CTL NP–treated 5xFAD mice, recovering to the wild-type levels (Fig. 5c). These results underscore the advantage of NP-mediated gene delivery, achieving superior cognitive rescue with a single intracisternal dose compared with repeated protein administration over 8 weeks.

Fig. 5.

Cognitive improvement by scFv33-Fc and scFv33 nanoparticles in 5xFAD mice. (a) Schematic illustration of the novel object recognition (NOR) test paradigm. (b, c) Experimental timeline for cisterna magna injection of (b) scFv33-Fc (2 µg) or (c) scFv33 NPs (200 µg) in 5xFAD mice and subsequent NOR assessment 2 months post-injection. Each dot represents an individual mouse. Data represent mean±SEM. Statistical analysis was performed using one-way ANOVA with Šidák post hoc test for multiple comparisons (*p<0.05).

scFv33 improves cognition via microglial remodeling and selective reduction of small hippocampal plaques

To probe mechanisms underlying the cognitive benefit, we quantified Aβ plaque burden in the cortex and hippocampus 3 months after intra–cisterna magna delivery of scFv33 NPs in the same 5xFAD cohorts used for the NOR assay (Fig. 6). Plaque counts showed no significant change in the cortex, whereas hippocampal plaques in the smaller-size bin (20~100 µm²) were significantly reduced in scFv33 NP–treated mice relative to CTL NPs (Fig. 6b). According to previous studies, Aβ plaques mature from diffuse to fibrillar and ultimately to compact dense-core forms, accompanied by increasing size [49, 50]. Notably, as plaques become denser, microglial phagocytic efficiency markedly declines, with reduced phagocytic clearance and a tendency to encapsulate dense-core plaques [51, 52]. Consistent with this model, the selective reduction of small hippocampal plaques after scFv33 NP treatment may reflect enhanced microglial clearance at earlier stages of plaque development.

Fig. 6.

Effect of scFv33 nanoparticles on amyloid-β plaque burden in 5xFAD mice. (a) Representative immunofluorescence images of Aβ staining in brain sections from 5xFAD mice treated with control plasmid-loaded nanoparticles (CTL NPs) or scFv33 plasmid-loaded nanoparticles (scFv33 NPs), 3 months post-injection. (b) Number of Aβ plaques in the cortex and hippocampus, categorized by plaque size. Each dot represents an individual mouse. Data represent mean±SEM. Statistical analysis was performed using two-way ANOVA with Šidák post hoc test (*p<0.05).

To determine whether the cognitive benefit and reduced hippocampal plaque burden were associated with altered microglial activation, we quantified Iba-1⁺ microglial morphology (Fig. 7). In the cortex, soma area, total process length, and the number of process segments per cell were not significantly different between groups (Fig. 7b). By contrast, in the hippocampus, scFv33 NP–treated mice exhibited a more ramified, less activated morphology: total process length and the number of process segments were significantly increased (Fig. 7c). These data indicate that NP-mediated scFv33 expression attenuates microglial activation in the hippocampus, consistent with reduction of hippocampal small Aβ plaques.

Fig. 7.

Microglial activation after scFv33 nanoparticle treatment in 5xFAD mice. (a) Representative immunofluorescence images of Iba-1 staining in brain sections from 5xFAD mice treated with control plasmid-loaded nanoparticles (CTL NPs) or scFv33 plasmid-loaded nanoparticles (scFv33 NPs), 3 months post-injection. (b, c) Morphological analysis of individual Iba-1⁺ cells in the (b) cortex and (c) hippocampus, including soma area, total process length, and number of process segments per cell. Each dot represents an individual cell. Data represent mean±SEM. Statistical analysis was performed using unpaired two-tailed t-tests (**p<0.01, ***p<0.001).

To test whether scFv33 directly modulates microglial morphology and Aβ handling, we used a primary neuron–microglia co-culture (Fig. 8). Cells were exposed to Aβ and treated with the scFv33 antibody either 1 h before (pre) or 1 h after (post) Aβ administration. Cells were fixed 48 h later for immunofluorescence analysis of microglial morphology and Aβ uptake. Aβ alone did not significantly change cell area or volume relative to saline but markedly decreased ramification (Fig. 8b, c), as reflected by shorter processes and fewer branch points (Fig. 8d, e). We speculate that the enlarged cell area and volume observed in saline-treated cultures reflects culture–induced stress. Notably, both scFv33 pre- and post-treatment significantly reduced cell area and volume compared to saline- and Aβ-treated groups, and pre-treatment restored process length and branch points to saline levels, indicating a more ramified morphology. Functionally, scFv33 significantly increased microglial Aβ uptake (Fig. 8f). These results indicate that scFv33 promotes a ramified microglial state that enhances Aβ phagocytosis, consistent with the hippocampal findings in 5xFAD mice where NP-mediated scFv33 expression increased process metrics.

Fig. 8.

scFv33 reduces microglial activation and enhances Aβ phagocytosis in a primary neuron-microglia co-culture. (a) Representative immunofluorescence images of primary neuron–microglia co-cultures stained for Iba-1, CD68, and Aβ. Cells were treated with scFv33 (30 μg/ml) 1 h before (pre) or 1 h after (post) Aβ exposure (10 μM) and fixed 48 h later. (b~e) Morphological analysis of Iba-1⁺ microglia, including (b) cell area, (c) cell volume, (d) process length, and (e) number of branch points. (f) Quantification of Aβ phagocytosis by microglia, determined by measuring the volume of Iba1⁺CD68⁺Aβ⁺ colocalized signal. (g) Representative immunofluorescence images of Iba-1, CD68, and Aβ staining in the brain hippocampus from 5xFAD mice treated with CTL or scFv33 NPs, 3 months post-injection. White boxes indicate magnified regions shown in the insets. (h) Quantification of microglial lysosomal engagement, presented as the proportion of CD68⁺ area relative to the total hippocampal Iba-1⁺ area. Each dot represents an individual mouse. (i) Quantification of Aβ phagocytosis in hippocampal microglia, determined by measuring the area of Iba-1⁺CD68⁺Aβ⁺ colocalized signal. Data represent mean±SEM. Statistical analysis was performed by one-way ANOVA with Tukey’s post hoc test (*p<0.05, ****p<0.0001).

To test whether the increased microglial lysosomal handling of Aβ observed in vitro contributes to Aβ clearance in vivo, we examined microglial lysosomal engagement and Aβ phagocytosis in the hippocampus of 5xFAD mice treated with CTL or scFv33 NPs. Three months after intra–cisterna magna injection, scFv33 NP–treated mice exhibited more CD68-enriched microglia clustered around hippocampal Aβ deposits (Fig. 8g). Quantitatively, the proportion of CD68⁺ signal within the total Iba1⁺ microglial area showed a modest increase in scFv33 NP–treated mice compared with CTL NP–treated mice (Fig. 8h). Nevertheless, scFv33 NPs markedly increased the area of Iba1⁺CD68⁺Aβ⁺ colocalized signal (Fig. 8i), demonstrating enhanced lysosomal processing of Aβ in vivo. These findings parallel the increased Aβ uptake observed in neuron–microglia co-cultures, supporting a model in which scFv33 enhances phagolysosomal engagement to promote Aβ clearance.

DISCUSSION

This study demonstrates that cisterna magna delivery of PLGA NPs loaded with the anti-TLR2 scFv33 pDNA durably recalibrates microglial state and mitigates amyloid-associated pathology in 5xFAD mice. A single NP dose improved recognition memory to a greater extent than 8 weeks of weekly scFv33-Fc protein dosing, coinciding with hippocampal microglial ramification and a reduction of small Aβ plaques (Figs. 5~7). In primary neuron–microglia co-culture, scFv33 directly promoted a more ramified morphology and increased microglial Aβ uptake (Fig. 8), supporting a causal link between TLR2 blockade, microglial remodeling, and enhanced clearance of early-stage plaques.

Microglia integrate damage- and pattern-associated cues through TLR pathways to mount NF-κB/MAPK-dependent responses and undergo a characteristic shift toward an amoeboid morphology [20-24]. Our data indicate that suppressing TLR2 signaling is sufficient to shift microglia away from an amoeboid toward a ramified, surveillance-competent phenotype. In vivo, scFv33 NPs increased hippocampal process length and branching without altering cortical morphology (Fig. 7), while in vitro scFv33 restored processes and reduced cell hypertrophy in the context of Aβ exposure and boosted phagocytic uptake (Fig. 8). Together with the selective reduction of small hippocampal plaques (Fig. 6), these findings are consistent with a model in which TLR2 inhibition preserves or restores microglial capacity to handle early, less compact Aβ species that are more readily cleared [51, 52], thereby interrupting feed-forward inflammatory loops that otherwise promote plaque maturation [32-34, 50, 51].

Although an Fc-fused scFv improved biophysical properties and retained anti-inflammatory activity (Fig. 1), repeated protein dosing produced only a trend toward behavioral improvement, whereas a single NP dose yielded significant rescue in NOR (Fig. 5). Gene delivery likely achieved more sustained intraparenchymal exposure, continuous local secretion, and better engagement of microglial TLR2 within relevant niches, while avoiding the pharmacokinetic peaks and subtherapeutic troughs of intermittent protein administration. These data support NP-mediated gene delivery as a practical route to long-acting innate-immune modulation in the CNS.

Transgene expression after cisterna magna injection was brain-wide but regionally heterogeneous, with higher levels in subcortical regions than in the cortex and clear functional/morphological effects in the hippocampus (Figs. 4, 6, 7). The hippocampus is an early-vulnerable region in 5xFAD and showed the greatest concordance across endpoints: elevated scFv33 mRNA, microglial ramification, reduced small-plaque burden, and improved recognition memory. Although only a small fraction of hippocampal microglia internalized PLGA NP (Fig. 4b), even limited modulation of this subset may influence surrounding cells through non–cell-autonomous signaling, as microglia readily propagate changes in cytokine tone and activation state to neighboring populations. In addition, because scFv33 is a secreted antibody fragment, NP-positive microglia likely act as localized producer cells, enabling paracrine suppression of TLR2 signaling in nearby microglia that did not internalize nanoparticles. These mechanisms together provide a plausible explanation for how partial microglial targeting can yield the robust functional and behavioral improvements observed in the hippocampus.

By contrast, cortical plaque counts and microglial morphology were unchanged at the same endpoint. These dissociations likely reflect differential transgene exposure. Practically, they highlight that optimizing the efficiency and spatial distribution of PLGA NPs throughout the brain will be essential to broaden efficacy across cognitive domains.

This non-viral, biodegradable delivery platform builds on the established clinical use of PLGA and the design flexibility of scFv constructs. Several features increase translational plausibility: CSF delivery, microglia-biased uptake that limits off-target exposure, and durable secretion after a single administration. At the same time, delivery refinements could expand the therapeutic footprint: PEGylation or ligand decoration to enhance parenchymal spread, tuning particle size/charge, and cassette-level improvements (codon optimization, promoter selection) to boost cortical expression without sacrificing cell-type bias [53-55]. Because scFv33 lacks an Fc, off-target Fc-mediated effects are minimized; however, the trade-off is shorter systemic persistence, and this limitation is mitigated by continuous local expression after gene delivery.

Several points warrant further work. First, the present mechanistic readouts focused on morphology and Aβ handling; deeper transcriptomic and phosphoproteomic profiling would map how TLR2 blockade rebalances NF-κB/MAPK modules, lysosomal programs, and metabolic states that support phagocytosis. Second, tau-inclusive models should be tested to assess generalizability beyond amyloidopathy, particularly given reports that TLR2 also contributes to tau-driven inflammation in AD [56-59]. Third, expanding cortical delivery may unlock benefits in additional cognitive domains. Finally, long-term safety requires formal evaluation in aged/immunocompromised settings, including infection susceptibility under sustained TLR2 suppression and tolerance of PLGA metabolite load; reversible control systems for gene therapy (e.g., doxycycline-inducible) could provide dose- and time-controlled expression if needed [60, 61].

A single intrathecal administration of PLGA NPs encoding an anti-TLR2 scFv reprograms microglia, preferentially limits early-stage hippocampal plaque accumulation, and rescues recognition memory in 5xFAD mice, outperforming prolonged protein dosing. These findings position NP-mediated anti-TLR2 gene therapy as a feasible route to durable, regionally effective immunomodulation in AD and motivate delivery optimization and tau-inclusive validation as next steps toward translational development.