Ras Inhibitor Lonafarnib Rescues Structural and Functional Impairments of Synapses of Aβ_1-42 Mice via α7nAChR-Dependent BDNF Upregulation

Abstract

Alzheimer's disease (AD) is characterized pathologically by the structural and functional impairments of synapses in the hippocampus, inducing the learning and memory deficiencies. Ras GTPase is closely related to the synaptic function and memory. This study was to investigate the effects of farnesyl transferase inhibitor lonafarnib on the synaptic structure and function in AD male mice and explore the potential mechanism. Our results showed 50 mg/kg lonafarnib (intraperitoneal) rescued the impaired spatial memory and improved the damaged synaptic transmission and plasticity of Aβ_1-42 mice. In addition, lonafarnib ameliorated the morphology of synaptic dendrites and spines in Aβ_1-42 mice. Furthermore, lonafarnib enhanced α7nAChR cell surface expression and phosphorylation of downstream Akt and CaMKII in Aβ_1-42 mice, which were inhibited by α7nAChR antagonist methyl lycaconitine (MLA), and increased the phosphorylation of CREB in a CaMKII- but not ERK-dependent way. Lonafarnib enhanced hippocampal brain-derived neurotrophic factor (BDNF) concentration in Aβ_1-42 mice, which was sensitive to MLA and KN93 (an inhibitor of CaMKII), but not related to ERK and Akt pathways. H-Ras, but not Rhes, was related to the lonafarnib induced improvement of α7nAChR cell surface expression and BDNF content. Interestingly, lonafarnib induced improvement of synaptic transmission, plasticity and spatial cognition in Aβ_1-42 mice was abolished by BDNF deprivation with TrkB/Fc chimera protein. Our results indicate that lonafarnib can rescue the structural and functional impairments of synapses in the Aβ_1-42 mice, which may be related to the improvement of BDNF content through the H-Ras-α7nAChR-dependent CaMKII-CREB pathway, leading to the improvement of spatial cognition.

SIGNIFICANCE STATEMENT Alzheimer's disease (AD) is characterized pathologically by the structural and functional impairments of synapses in the hippocampus, inducing the learning and memory deficiencies. However, no effective drugs have not been developed for the treatment of AD synaptic. This study for the first time reported the beneficial effects of Ras inhibitor lonafarnib on the synaptic structure and function in AD mice, providing an alternative way for the treatment of “synaptic disease” in AD patients.

Introduction

Alzheimer's disease (AD) is the most common cause of dementia, with an estimated prevalence of over 55 million people globally, which will increase to 139 million by 2050 (reports from Alzheimer's Disease International). However, as a complex multifactorial disorder, the effective therapeutic drugs for AD are still poorly understood.

As an neurodegenerative disease, AD is characterized by the learning and memory deficiencies, which is closely associated with the structural and functional impairments of synapses in the hippocampus (Spires-Jones and Hyman, 2014), such as the loss of synapses (Scheff et al., 2007), changes in the morphology of neurons (de Pins et al., 2019) and synaptic proteins (Counts et al., 2014), and consequent destroyed transmission and plasticity of synapses (Oddo et al., 2003). Thus, AD has also been conceptualized as a “synaptic disease” (Spires-Jones and Hyman, 2014).

Ras GTPase overexpression has been reported in the brain of AD patients (Kirouac et al., 2017). Ras signaling pathway is closely related to the synaptic function and memory (Ye and Carew, 2010). Ras overexpression in the neurofibroma mice significantly impairs the cognitive memory and synaptic function (Costa et al., 2002), while Ras activity inhibition significantly improves the cognitive function of these mice. The deficits in long-term potential (LTP) induction and learning have been reported in the mice with heterozygous knock-out of Nf1, a gene encoding a protein related to Ras inactivation, which can be reversed by inhibiting Ras pathway (Li et al., 2005). However, whether Ras inhibition can rescue the functional and/or structural impairments of synapses in the AD mice is still unclear.

Brain-derived neurotrophic factor (BDNF) is one of the most distributed and explored neurotrophins in the brain and also a major factor regulating the function and structure of synapses between neurons (Zagrebelsky and Korte, 2014). BDNF is secreted during LTP induction and also functionally necessary for the signaling cascades in the LTP (Park and Poo, 2013; Lu et al., 2014). In vivo studies have established a crucial role of BDNF signaling (Reichardt, 2006) in the modulation of dendritic growth and branching, as well as dendritic spine density (Yamada et al., 2001; Hu et al., 2011), which are both important for the induction and maintenance of LTP (Kovalchuk et al., 2002; Zagrebelsky and Korte, 2014). A decline of BDNF content may result in the synaptic deficiency of AD mice (von Bohlen Und Halbach and von Bohlen Und Halbach, 2018), which is known as one of the significant pathologic alterations in the frontal cortex and hippocampus of AD patients (Ferrer et al., 1999; Holsinger et al., 2000) and early-stage AD mice (Kaminari et al., 2017). There is evidence showing that the rescue of BDNF loss in the astrocytes can significantly improve the amplitude of LTP in the 5xFAD mice (de Pins et al., 2019).

The BDNF expression is closely related to α7nAChRs (J. Kim et al., 2019; Moriguchi et al., 2020). Blocking α7nAChRs reduces the BDNF expression in the hippocampus (Freedman et al., 1993). On the other hand, α7nAChR agonist can enhance BDNF expression in the aged 3xTg-AD mice with robust plaques and tangles (Medeiros et al., 2014). The activation of α7nAChR also triggers BDNF secretion and recruits TrkB receptors onto cell surface in the SH-SY5Y neuroblastoma cells (Serres and Carney, 2006). Our previous study revealed that Ras inhibition by statins or farnesyltransferase inhibitors (FTI, FTI-277, or lonafarnib) could increase the activity and function of α7nAChR in control mice (T. Chen et al., 2016b, 2018, 2020). However, the effects of Ras inhibition on the BDNF expression, the structural and functional properties of synapses and the involvement of α7nAChR are still unclear in case of AD.

This study was to investigate the effects of Ras inhibitor lonafarnib on the impaired spatial cognition, hippocampal CA1 synaptic plasticity, neuronal morphology (dendritic spine and dendrite), and synapse protein expression in the Aβ_1-42 mice, and explore their associations with α7nAChR dependent BDNF content improvement.

Materials and Methods

Experimental animals

The study protocol was approved by Experimental Animal Care and Ethical Committee of Nantong University. All the procedures followed the Guidelines of Institute for Laboratory Animal Research of the Nantong University. The procedures involving animals and their care were conducted in conformity with the ARRIVE Guidelines of Laboratory Animal Care (Kilkenny et al., 2012). Male C57BL/6J mice aged two months (SLAC Laboratory Animal Co, Ltd.) were maintained in a constant environment (23 ± 2°C; 55 ± 5% humidity; 12/12 h light/dark cycle) in the Animal Center of Nantong University. Animals were given ad libitum access to food and water.

Establishment of AD model

The AD model was established according to the procedures reported in our previous study (Jin et al., 2016). The Aβ_1-42 (Sigma) was prepared as previously described (Bouter et al., 2013). Briefly, Aβ_1-42 was dissolved in the 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP; Sigma-Aldrich), flash-frozen in the liquid nitrogen, and then lyophilized to completely remove the solvent. The lyophilized Aβ_1-42 peptides were then dissolved in 100 mm NaOH at 6 mg/ml, aliquoted, flash-frozen in the liquid nitrogen, and stored at −80°C until use.

For intracerebroventricular injection of soluble Aβ_1-42, mice were intraperitoneally anesthetized with isoflurane, and then placed in a stereotactic apparatus (Motorized Stereotaxic Stereo Drive; Neurostar). Freshly prepared Aβ_1-42 (0.3 nmol/2 µl in 0.1 m PBS) was injected into the bilateral ventricles (0.3 mm posterior, 1.0 mm lateral, and 2.5 mm ventral to the bregma) using a stepper-motorized micro-syringe at 0.2 µl/min. The aggregated Aβ_1-42 in the hippocampus has been confirmed by immuno-staining with the Aβ-specific antibody (Richardson et al., 2002). The mice injected with vehicle of the same volume served as controls.

Drug administration

As a precursor protein, Ras is first farnesylated by FPP and then anchored to the inner leaflet of the cell membrane (Shields et al., 2000). Considering the pleiotropic effect of statins, FTase inhibitor (FTI) lonafarnib was used to specifically inhibit Ras activation by blocking its farnesylation (Mans et al., 2012). Lonafarnib (SCH-66336) was purchased from MedChem Express (MCE). For in vivo experiment, lonafarnib was dissolved in dimethylsulfoxide (DMSO), which then diluted in saline containing 20% (2-hydroxypropyl)-β-cyclodextrin (Chaponis et al., 2011). Lonafarnib was intraperitoneally injected at different doses of 10, 30, 50, and 80 mg/kg (T. Chen et al., 2016a) starting at 4 h after Aβ_1-42 injection for 14 d. Control mice were intraperitoneally treated with an equal volume of vehicle.

The α7nAChR antagonist methyl lycaconitine (MLA), PI3K inhibitor LY294002, MEK inhibitor U0126, and CaMKII inhibitor KN93 were purchased from Sigma. MLA, LY294002, U0126, and KN93 were dissolved in 0.5% DMSO, at 30 min before lonafarnib administration, MLA (0.1 nmol/mouse), U0126 (0.3 nmol/mouse), Ly294002 (0.3 nmol/mouse), or KN93 (1 µg/mouse) was injected into the right ventricle. For repeated intracerebroventricular injection, a 28-G stainless steel guide cannula (Plastic One) was implanted into the right lateral ventricle (0.3 mm posterior, 1.0 mm lateral, and 2.5 mm ventral to the bregma) and anchored to the skull with three stainless steel screws and dental cement (C. Wang et al., 2015). The mice injected with vehicle (0.5% DMSO) of the same volume served as controls.

Recombinant mouse TrkB/Fc chimera protein, which can sequester endogenous BDNF, was purchased from R&D Systems. It was dissolved in sterile PBS and 0.9% sterile physiological saline and injected intracerebroventricularly (10 ng/mouse; Yajima et al., 2005; Renn et al., 2011) immediately before Aβ injection once daily for 14 consecutive days at 30 min before lonafarnib administration.

Behavioral assessment (Morris water maze; MWM)

MWM test was conducted for eight consecutive days according to previously reported in our study (T. Chen et al., 2016b), to assess the spatial cognition of mice (Tong et al., 2012). A circular pool made of black plastic (120 cm in diameter) is artificially divided into four quadrants and marked on the wall with entry points for each quadrant. The water temperature was maintained at 23 ± 2°C, and an appropriate amount of white food additives was added into the water. The swim paths were analyzed using a computer system with a video camera (AXIS-90 Target/2; Neuroscience). The test is divided into two parts, place navigation, including visible platform test (days 1–2) and hidden platform test (training days 3–7), and spatial probe test (day 8). In the first 2 d of training, a cylindrical dark-colored platform (7 cm in diameter) was placed 0.5 cm above the surface of water. During training days 3–7, the acquisition-testing phase, the platform was submerged 1 cm below the surface of water. Mice were allowed to search the platform within 90 s in the pool. The latency to reaching the visible or hidden platform and the swimming distance were measured. During the training, each mouse was placed at one of four quadrants randomly, with its head toward the wall. If the mouse failed to find the platform within 90 s, it was guided to the platform and the test was terminated. Four trials were conducted every day with an interval of 30 min. On day 8, the retention of spatial reference memory was recorded by a probe trial with the platform being removed from the pool, and the percent time spent in each quadrant was calculated.

Electrophysiological analysis

Slice preparations

Mice were decapitated and the brains were rapidly removed. The coronal brain slices (400 µm) were cut using a vibrating microtome (Leica VT1200s) in ice-cold oxygenated (95% O₂/5% CO₂) modified artificial CSF (mACSF) composed of (in mm) 126 NaCl, 1 CaCl₂, 2.5 KCl, 1 MgCl₂, 26 NaHCO₃, 1.25 KH₂PO₄, and 20 D-glucose. After 1 h recovery, the slices were transferred to a recording chamber and perfused continually with the oxygenated mACSF at 30 ± 1°C. A glass microelectrode with the resistance of 4–5 MΩ filled with 2 m NaCl was inserted into the radiatum layer of CA1 region.

Field potential recording

EPSP was generated by stimulating the Schaffer collateral/commissural pathway using a stimulator (SEN-3301, Nihon Kohden). Stimulus pulses (0.1-ms duration) were produced every 15 s. Signals were obtained using an Axoclamp 2B amplifier (Molecular Devices), sampled at 20 kHz, and filtered at 10 kHz, and the output was digitized with a Digidata 1200 converter (Molecular Devices). The stability of baseline recordings was established by delivering single pulses (0.05 Hz, 0.1-ms pulse width) for 15 min before the data collection. (1) Input/output (I/O) curve: EPSP slopes were evoked by testing stimulation of various intensities (0.1–1.1 mA). (2) Paired-pulse facilitation (PPF) was induced by double stimuli with interpulse interval (IPI) of 25–100 ms. The value of paired-pulse ratio (PPR) is expressed by the second EPSP slope relative to the first EPSP slope. (3) LTP was induced by high-frequency stimuli (HFSs; 100 Hz, 100 pulse). After delivering HFS, the EPSP slopes were recorded for a further period of 60 min. In 55–60 min after delivering HFS, if the EPSP slopes were 20% larger than baseline, LTP was determined (Y. Wang et al., 2018).

Slice biotinylation and cell surface protein extraction

The procedure of biotinylation of slices was previously reported in our study (T. Chen et al., 2018). Hippocampal slices were placed on a six-well plate and washed with cold ACSF for 5 min. Then, the slices were incubated with ACSF containing EZ-link Sulfo-NHS-SS-Biotin (0.5 mg/ml, Pierce) for 25 min at 4°C. These slices were washed with ACSF containing 50 mm NH₄Cl thrice (5 min for each) at 4°C to remove excess biotin. After biotinylation, the hippocampus CA1 region was isolated and homogenized with lysis buffer [50 mm Tris–HCl (pH 7.4), 150 mm NaCl, 1.5 mm MgCl₂, 1 mm EGTA, 0.5 mm DTT, 50 mm NaF, 2 mm sodium pyruvate, 25% glycerol, 1% Triton X-100, 0.5% sodium deoxycholate, and 1% protease inhibitor cocktail; Sigma]. The supernatant was centrifuged at 20,000 × g for 20 min at 4°C, the resultant supernatant was collected, and the protein concentration was determined by Bradford protein assay. Biotinylated proteins (50 mg) were incubated with streptavidin-coated magnetic beads (30 ml) for 45 min at room temperature, and then the beads containing biotinylated proteins were washed thrice with lysis buffer containing 0.1% SDS, and separated with a magnet. The biotinylated proteins were eluted in sample buffer (62.5 mm Tris-HCl, 2% SDS, 5% glycerol, 5% 2-mercaptoethanol) at 100°C for 5 min. The protein lysates were denatured with the same method. Then, the protein lysates (cytoplasmic proteins) and biotinylated proteins (cell surface proteins) were stored at −20°C until assay.

Western blotting

Animals were anesthetized, and the brain was harvested, followed by the separation of hippocampus. The hippocampal tissues or brain slices were homogenized in the lysis buffer containing 50 mm Tris-HCl (pH 7.5), 150 mm NaCl, 5 mm EDTA, 10 mm NaF, 1 mm sodium orthovanadate, 1% Triton X-100, 0.5% sodium deoxycholate, 1 mm phenylmethylsulfonyl fluoride and protease inhibitor cocktail (Complete; Roche), followed by incubation for 30 min at 4°C. After sonication, the samples were centrifuged at 12,000 rpm for 15 min at 4°C, and the supernatant was harvested. The protein concentration was determined with BCA Protein Assay kit (Pierce Biotechnology Inc.). Then, proteins of equal amount were mixed with loading buffer and boiled for 5 min.

Proteins of equal amount (20 µg) were separated by SDS–PAGE and then transferred onto polyvinylidene fluoride (PVDF) membrane, which was subsequently incubated with blocking solution (5% nonfat milk) for 60 min at room temperature. After washing thrice, the membrane was incubated overnight at 4°C with primary antibodies as follows: mouse anti-Aβ (6E10; 1:1000; BioLegend, catalog #803014, RRID: AB_2728527), anti-Flag (1:1000; Sigma, catalog #F1804, RRID: AB_262044), rabbit polyclonal anti-α7nAChR (1:1000; Abcam, catalog #ab23832, RRID: AB_776765), anti-phospho-CREB (Cell Signaling Technology, catalog #9198, RRID:AB_2561044), and anti-phospho-ERK (1:1000; Cell Signaling Technology, catalog #9101, RRID: AB_331646) and anti-phospho-CaMKII (1:1000; Cell Signaling Technology, catalog #12716, RRID:AB_2713889), anti-phospho-Akt (1:1000; Cell Signaling Technology, catalog #4056, RRID: AB_331163), rabbit polyclonal anti-CREB (1:1000; Thermo Fisher Scientific, catalog #PA1-850, RRID: AB_2086016), anti-CaMKII (1:1000; Abcam, catalog #ab126789, RRID: AB_11131025), Akt (1:1000; Cell Signaling Technology, catalog #9272, RRID: AB_329827), and ERK1/2 (1:1000; Cell Signaling Technology, catalog #4695, RRID: AB_390779). GAPDH (1:2000; Cell Signaling Technology, catalog #5174, RRID: AB_10622025) or β-actin (1:2000; Cell Signaling Technology, catalog #3700, RRID: AB_2242334) served as internal control. The biotinylated membrane surface α7nAChR protein was normalized by a marker of cytoplasmic membrane, mouse anti-pan-cadherin (H. Wang et al., 2008; 1:1000; Sigma, catalog #SAB4200731, RRID: AB_2904558), which was not affected in AD brains (Zhou et al., 2013) and not reported be related to Ras GTPase. Appropriate HRP linked secondary antibodies were used for detection by enhanced chemiluminescence (Pierce). ImageJ (NIH Image) was used to determine the protein expression which was normalized to the expression of internal control.

Golgi–Cox staining

Golgi–Cox staining was used to examine subtle morphologic alterations in the neuronal dendrites and dendritic spines (Luo et al., 2014; Zhu et al., 2014). Golgi–Cox staining was performed as described previously (Zhu et al., 2014, 2018). Fresh, nonperfused brains were used for Golgi–Cox staining with FD Rapid GolgiStain kit (catalog #PK401; FD NeuroTechnologies) according to the manufacturer's instructions, and the brains were cut into 100-µm coronal sections using a vibratome (VT1000s; Leica). To calculate the spine density of Golgi-stained neurons in the hippocampal CA1 region, dendrites were traced from images taken with a Nikon Eclipse Ti-E confocal microscope, images were analyzed using ImageJ, the exact length of the dendritic segment was calculated and the number of spines along the dendritic segment was counted. Sholl analysis was performed with images of traced dendritic arbors using an ImageJ plugin. Sampling step size was set at 10 µm (range, 0–200 µm).

Tissue fixation and immunofluorescence

Animals were anesthetized with isoflurane and transcardially perfused with 4% paraformaldehyde. Brains were removed and postfixed overnight in the same solution, Brains were transferred into 20% and 30% sucrose in sequence. After gradient dehydration, serial coronal sections of the hippocampus (40 µm) were obtained by using the freezing microtome (Leica). Each mouse brain produced 50 slices and every fifth section throughout the hippocampus was processed for cell counting. In order to obtain a homogenous representation of the hippocampus, no more than two sections were lost during sectioning of a single brain. The sections were treated with 3% normal goat serum, and then incubated overnight at 4°C with antibodies for anti-PSD-95 (1:500, Millipore, catalog #MAB1596, RRID: AB_2092365), anti-synaptophysin (1:500, Synaptic Systems, catalog #101-008, RRID: AB_2864779), anti-MAP2 (1:500, Sigma, catalog #HPA012828, RRID: AB_1853946). After PBS rinses, the sections were incubated with the subtype-specific fluorescent secondary Alexa Fluor 488 anti-rabbit (1:250, Thermo Fisher Scientific, catalog #A-11 008, RRID: AB_143165) or anti-mouse 555 (1:250, Thermo Fisher Scientific, catalog #A-21422, RRID: AB_2535844) for 2 h at room temperature. No signal was detectable in the control sections incubated in the absence of primary antibody. We coded all slides from the experiments before quantitative analysis. PSD-95/synaptophysin-positive puncta number and area fraction were counted by another experimenter who was unaware of the experimental conditions of each sample. The sections were observed under a microscope (Leica TCS SP8) at a magnification of 146×. The labeled PSD-95/synaptophysin-positive puncta number and area fraction were quantified with NIH ImageJ freeware (Wayne Rasband, NIH).

BDNF detection by ELISA

BDNF content was detected on day 14 after Aβ_1-42 treatment by ELISA (Promega; C. Wang et al., 2015). Hippocampus was homogenized by sonication with homogenization buffer (50 mm Tris pH 7.5, 300 mm NaCl, 0.1% Triton X-100, 10 mg/ml aprotinin, 0.1 mm benzethonium chloride, 1 mm benzamidine, and 0.1 mm phenylmethylsulfonyl fluoride), followed by centrifugation for 15 min at 12,000 × g at 4°C. To detect the content of free mature BDNF (not total free BDNF), the samples were not treated with 1 N hydrochloric acid (instructions for use of products G7610 and G7611). Briefly, titer plates were coated with anti-BDNF monoclonal antibody overnight at 4°C. The plates were incubated with BDNF polyclonal antibody at room temperature. Horseradish peroxidase-conjugated anti-IgY rabbit antibody was added, followed by incubation at room temperature. The absorbance was measured at 450 nm with an automated microplate reader. BDNF content was expressed as pg/mg total soluble protein. All assays were performed in triplicate.

Virus injections

To produce the (adeno-associated viral vectors) AAV-hSyn-H-Ras-EGFP and AAV-hSyn-Rhes-EGFP, the coding regions of H-Ras and Rhes were amplified from C57BL/6J mice cDNA by PCR, standard cloning procedures were used to subclone the EGFP (Enhanced Green Fluorescent Protein) cassettes into the backone of AAV-hSyn-MCS-3FLAG expression plasmid. Following DNA sequencing screening, the AAV plasmids were packaged into AAV serotype 9 virus from GeneChem CO., Ltd., with the titer at 1 × 10¹³ virus particles per milliliter.

Mice were anaesthetized by isoflurane, and then placed in a stereotactic apparatus. AAV-hSyn-H-Ras-EGFP (AAV-H-Ras) virus, AAV-hSyn-Rhes-EGFP (AAV-Rhes) virus or AAV-hSyn-EGFP control (AAV-Con) virus was bilaterally microinjected into the hippocampus according to the following coordinates: 2.0 mm behind the bregma and ±1.5 mm lateral from the sagittal midline at a depth of 2 mm below the skull surface (Huang et al., 2020). Virus was delivered with a 10-µl Hamilton syringe (1 µl per site) at a rate of 0.05 µl/min. The needle was remained in the brain for an additional 10 min to prevent the backflow of virus suspension, guaranteeing infection efficiency. The incision was sutured, and the mice were allowed to recover for 3 d. Green GFP fluorescence was detected to confirm the infection position, and objective protein overexpression was confirmed by Western blotting of FLAG expression.

Experimental design and statistical analyses

Experimental design

All mice were randomly assigned to different experimental groups: (1) behavioral tests (n = 8 or 12 mice per group)→Western blotting/ELISA (n = 8 mice per group, unilateral hippocampal tissue per mice was used for Western blotting or ELISA) were sequentially performed in the same cohorts; (2) electrophysiological analysis was performed in the separate cohorts (n = 8 mice per group); (3) Golgi–Cox staining was performed in the separate cohorts (n = 6 mice per group); (4) immunofluorescence staining was done in the separate cohorts (n = 8 mice per group); (5) Western blotting and ELISA for groups treated with different inhibitors or AAV were examined in the same cohorts (unilateral hippocampal tissue per mice was used for Western blotting or ELISA, n = 8 mice per group). Each experiment was performed by two experimenters blind to the grouping. The behavioral tests were done at days 7–14 after the first lonafarnib treatment (4 h after Aβ_1-42 injection); mice in electrophysiological examination were killed and examined at days 13 and 14 after the first lonafarnib treatment; mice in Western blotting, ELISA, Golgi–Cox staining and immunofluorescence staining were killed at 24 h after the last lonafarnib treatment.

Statistical analyses

For each experiment, descriptions of critical variables (e.g., number of animals, neurons, and other samples evaluated) as well as statistical design can be found in Results. All statistical analyses were performed with GraphPad Prism 8. Data are presented as mean ± SEM unless mentioned otherwise. The sample size was predetermined by analyzing the data of pilot study with PASS 157 (power analysis and sample size) software. Unpaired Student's t test was used to analyze statistical significance between two groups. One-way ANOVA was used for comparisons with one variable among group followed by Tukey's post hoc tests. Two-way ANOVA or two-way repeated-measure ANOVA was used for comparisons with two variables among groups, with Tukey's, Sidak's, or Dunnett's test for post hoc test. Three-way ANOVA was used for comparisons with three variables among groups, with Tukey's or Sidak's test for post hoc test. The p and F values are given in the results. Statistical significance was set at *p < 0.05, **p < 0.01. A value of p < 0.05 was considered statistically significant.

Data availability statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Results

Lonafarnib rescues Aβ_1-42-induced spatial memory impairment in a dose-dependent manner

Aβ is a major component of senile plaques that are commonly found in the brain of AD patients (Glabe, 2008). Aβ aggregation (such as oligomer, pro-fibril, and filamentous fibril formations) and accumulation in the brain play a major role in the pathogenesis of AD (Roychaudhuri et al., 2009). To determine the pathologic characteristics of Aβ_1-42 mice in the present study, the Aβ oligomers in the hippocampus were examined after injection of exogenous Aβ. Aβ oligomerization was assessed based on the presence of high-molecular weight (HMW) oligomers (>23 kDa). Results showed that the amount of Aβ oligomerization increased in the Aβ_1-42 mice in a time-dependent manner at 3, 7, and 14 d postinjection (F_(3,28) = 55.13, p < 0.0001, one-way ANOVA, followed by Tukey's test; Fig. 1A). Notably, the level of Aβ oligomerization was significantly higher at 3 d in the Aβ_1-42 mice than in the control mice treated with NS for 14 d (p < 0.0001), and it in the Aβ_1-42 mice further increased at 7 d (p = 0.0041) and 14 d (p < 0.0001) as compared with that at 3 d, which was consistent to the findings reported by Tao et al. (2020). Based on the pathologic characteristics, the spatial cognition was assessed by MWM at days 7–14 after Aβ_1-42 injection. In the MWM test, the latency to finding the visible-platform can reflect the searching behavior or visual acuity, and the latency to finding the hidden-platform is used to judge the spatial learning and memory. As shown in Figure 1B, upper, the latency to finding the visible-platform was comparable between control mice and Aβ_1-42 mice (interaction factor: F_(1,14) = 0.082, p = 0.7788; time factor: F_(1,14) = 10.55, p = 0.0058; model factor: F_(1,14) = 0.007, p= 0.9319, repeated-measure ANOVA), the latency to reaching the hidden-platform progressively decreased with the training days in the control mice and Aβ_1-42 mice (time factor: F_(4,56) = 30.59, p < 0.0001, repeated-measure ANOVA), which was significantly longer in the Aβ_1-42 mice than in the control mice, especially at days 5–7 (interaction factor: F_(4,56) = 2.275, p = 0.0725; model factor: F_(1,14) = 15.5, p = 0.0015; post hoc test: day 5: p = 0.019, day 6: p = 0.0051, day 7: p = 0.0060, n = 8 mice per group; repeated-measure ANOVA, followed by Tukey's test; Fig. 1B, upper). There was no significant difference in the swimming speed during the training days between control mice and Aβ_1-42 mice (interaction factor: F_(6,84) = 0.4736, p = 0.8262; time factor: F_(6,84) = 0.3884, p = 0.8846; model factor: F_(1,14) = 0.01,922, p = 0.8917, repeated-measure ANOVA; Fig. 1B, bottom). A probe trial was performed at 24 h after the hidden platform test, in which the swimming time spent in four quadrants (platform, opposite, right and left adjacent quadrants) was measured to estimate the memory trace strength, especially in the platform quadrant. Compared with control mice, Aβ_1-42 mice had less swimming time in the platform quadrant (t = 2.287, df = 14, p = 0.0383, unpaired Student's t test; Fig. 1C).

Figure 1.

Lonafarnib treatment improves spatial learning memory in a dose-dependent manner in the Aβ_1-42 mice. A, The amount of Aβ oligomerization (HMW: high-molecular weight) in mice injected with vehicle (intracerebroventricular; control mice) for 14 d, and in mice treated with Aβ_1-42 (intracerebroventricular; Aβ_1-42 mice) at 3 d, 7 d and 14 d postinjection; **p < 0.01 versus control mice, ##p < 0.01 versus Aβ_1-42 mice. B, Latency to findings the visible-platform and hidden-platform in the MWM test of Aβ_1-42 mice and control mice. Representative images of swimming paths in the hidden platform test (days 3–7) from different groups at day 6, which is representative and can show the difference in search time and tracks (upper panels). Black circles, position of platform; *p < 0.05, **p < 0.01 versus control mice. C, Percentage of swimming time (%) in the platform quadrant in the MWM test; *p < 0.05 versus swimming time in the PQ of control mice. Representative images of swimming paths in the spatial probe test (day 8, platform was removed) from two groups, which is representative and can show the difference in search time and tracks (upper panels). D, Latency to finding the visible-platform and hidden-platform in the MWM test of the Aβ_1-42 mice and Aβ_1-42 mice treated with lonafarnib (intraperitoneal) at 10 mg/kg [AL10: Aβ_1-42/+lonafarnib (10 mg/kg)], 30 mg/kg [AL30: Aβ_1-42/+lonafarnib (30 mg/kg)], 50 mg/kg [AL50: Aβ_1-42/+lonafarnib (50 mg/kg)], and 80 mg/kg [AL80: Aβ_1-42/+lonafarnib (80 mg/kg)]. Black circles, The position of platform; *p < 0.05, **p < 0.01, Aβ_1-42/+lonafarnib (50 mg/kg) mice versus Aβ_1-42 mice, ##p < 0.01, Aβ_1-42/+lonafarnib (80 mg/kg) mice versus Aβ_1-42 mice. Representative images of swimming paths in the hidden platform test (days 3–7) from different groups at day 6, which is representative and can show the difference in search time and tracks (upper panels). E, Percentage of swimming time (%) in the PQ in the MWM test. Representative images of swimming paths in the spatial probe test (day 8, platform was removed) from different groups, which is representative and can show the difference in search time and tracks (right panels); **p < 0.01 versus swimming time in the PQ of Aβ_1-42 mice.

In order to test the dose-dependent effects of lonafarnib on the Aβ_1-42-induced spatial cognition impairment, lonafarnib (10, 30, 50, and 80 mg/kg) was administered once daily since 4 h after Aβ_1-42 injection for consecutive 14 d. As shown in Figure 1D, lonafarnib affected the prolonged latency in the Aβ_1-42 mice (interaction factor: F_(16,220) = 1.688, p = 0.0503; time factor: F_(4,220) = 71.70, p < 0.001; treatment factor: F_(4,55) = 10.01, p < 0.001, repeated-measure ANOVA). As compared with vehicle-treated Aβ_1-42 mice, lonafarnib at 50 and 80 mg/kg significantly reduced the escape latency [post hoc test: Aβ_1-42 vs Aβ_1-42/+lonafarnib (50 mg/kg): p = 0.0006, Dunnett's test, especially at day 5: p = 0.0022, day 6: p = 0.0106, day 7: p = 0.0070, n = 12 mice per group, Tukey's test; vs Aβ_1-42/+lonafarnib (80 mg/kg): p = 0.0043, Dunnett's test, especially, day 5: p = 0.0038, day 6: p = 0.0018, n = 12 mice per group; Tukey's test; Fig. 1D], but not at 10 or 30 mg/kg [post hoc test: Aβ_1-42 vs Aβ_1-42/+lonafarnib (10 mg/kg): p = 0.9909, n = 12 mice per group, vs Aβ_1-42/+lonafarnib (30 mg/kg): p = 0.0789, n = 12 mice per group, Dunnett's test; Fig. 1D]. There was no marked difference between Aβ_1-42 mice treated with lonafarnib at 50 and 80 mg/kg (p = 0.9895, n = 12). Lonafarnib at any dose failed to alter the latency to finding the visible-platform (interaction factor: F_(4,55) = 0.1514, p = 0.9616; time factor: F_(1,55) = 13.49, p = 0.0005; treatment factor: F_(4,55) = 0.6560, p = 0.6251, repeated-measure ANOVA; Fig. 1D) and the swimming speed (interaction factor: F_(24,330) = 0.2744, p = 0.9998; time factor: F_(6,330) = 0.4156, p = 0.8686; treatment factor: F_(4,55) = 2.284, p = 0.9213; Fig. 1D, bottom, repeated-measure ANOVA). In the probe trial test, lonafarnib at 50 or 80 mg/kg, but not 10 or 30 mg/kg, increased the swimming time spent in the platform quadrant in the Aβ_1-42 mice (F_(4,55) = 11.05, p < 0.0001; post hoc test: lonafarnib of 0 vs 50 mg/kg: p = 0.0004, vs 80 mg/kg: p = 0.0001, vs 10 mg/kg: p = 0.8987, vs 30 mg/kg: p = 0.9991, n = 12 mice per group, one-way ANOVA, followed by Tukey's test; Fig. 1E), and there was no significant difference between 50 mg/kg lonafarnib group and 80 mg/kg lonafarnib group (p = 0.9935). These results indicated the Aβ_1-42-treated mice developed spatial memory impairment, and lonafarnib improved the cognitive function in a dose-dependent manner.

Lonafarnib rescues Aβ_1-42-induced impairments of hippocampal CA1 synaptic transmission and plasticity

To further explore the mechanisms of lonafarnib-rescued spatial cognition, the basal synaptic transmission and plasticity (LTP) were examined in the Aβ_1-42 mice treated with vehicle or lonafarnib (50 mg/kg) for consecutive 14 d using the field potential recording. The Schaffer collateral-CA1 synaptic property was analyzed by plotting fractional changes in the EPSP slopes against stimulating intensities from 0.1 to 1.1 mA. As shown in the I/O curve (Fig. 2A), lonafarnib treatment enhanced the slopes of EPSPs induced by stimulant intensities, which reduced in the Aβ_1-42 mice, however, lonafarnib did not affect the I/O curve in control mice (interaction factor: F_(15,140) = 1.264, p = 0.2327; stimulating intensities factor: F_(5,140) = 87.62, p < 0.0001; treatment factor: F_(3,28) = 9.989, p = 0.0001; post hoc test: Aβ_1-42 vs control, p = 0.0031, n = 8 mice per group, 0.7 mA: p = 0.0292, 0.9 mA: p = 0.0062 and 1.1 mA: p = 0.0021; Aβ_1-42 vs Aβ_1-42/+lonafarnib: p = 0.0030, n = 8 mice per group, 0.7 mA: p = 0.0200; 0.9 mA: p = 0.0241; 1.1 mA: p = 0.0128; control vs control+ lonafarnib: p > 0.9999, n = 8 mice per group; repeated-measure ANOVA, followed by Tukey's test). Then, the PPF of EPSP slopes evoked by paired-pulse stimulation with a 25- to 100-ms IPI was recorded to estimate the property of presynaptic neurotransmitter release. The PPR in the Aβ_1–42 mice increased as compared with the control mice, which indicated a reduction of presynaptic transmitter release (T. Chen et al., 2016b), but lonafarnib treatment reduced the PPR. Similarly, lonafarnib treatment had no influence on the PPR in the control mice (interaction factor: F_(9,84) = 0.7969, p = 0.6200; IPI factor: F_(3,84) = 23.72, p < 0.0001; treatment factor: F_(3,28) = 3.012, p = 0.0467; post hoc test: Aβ_1-42 vs control, p = 0.0188, n = 8 mice per group, 50-ms IPI: p = 0.011; Aβ_1-42 vs Aβ_1-42/+lonafarnib: p = 0.0261, n = 8 mice per group; 50-ms IPI: p = 0.0426; control vs control+lonafarnib: p = 0.9702, n = 8 mice per group; repeated-measure ANOVA, followed by Tukey's test; Fig. 2B). Then, the synaptic plasticity LTP induction, a cellular model of learning and memory (Bliss and Collingridge, 1993), was further examined by applying a HFS (100 Hz, 100 pulses). As shown in Figure 2C, the HFS induced increases in the EPSP slopes were different among four groups. The amplitudes of post-tetanus potentiation (PTP; at 1–5 min after HFS; Fig. 2D) and L-LTP (at 56–60 min after HFS; Fig. 2E) were further plotted to investigate this alteration in the EPSP slopes. PTP, a form of presynaptic plasticity, reflects the change in the probability of Ca²⁺-dependent vesicle release and is helpful for the stabilization and modulation of synaptic strength (Wojtowicz et al., 1994; Zucker and Regehr, 2002; Fioravante and Regehr, 2011). L-LTP requires gene transcription and de novo protein synthesis, reflects the efficiency of synaptic plasticity (Davies et al., 1989; Frey et al., 1993). As shown in Figure 2D, the amplitudes of PTP induced by HFS consistently decreased in the Aβ_1-42 mice as compared with control mice, which was rescued by lonafarnib treatment (interaction factor: F_(1,28) = 3.905, p = 0.0489; model factor: F_(1,28) = 35.39, p < 0.0001; treatment factor: F_(1,28) = 17.86, p = 0.0002; simple effect: Aβ_1-42 vs control: p < 0.0001, Aβ_1-42 vs Aβ_1-42/+lonafarnib: p = 0.0008, n = 8 mice per group, two-way ANOVA, followed by Sidak's test; Fig. 2D). However, lonafarnib had no effect on the amplitude of PTP in the control mice (p = 0.5858). Lonafarnib treatment also affected the L-LTP amplitude (interaction factor: F_(1,28) = 0.8424, p = 0.3666; model factor: F_(1,28) = 27.15, p < 0.0001; treatment factor: F_(1,28) = 23.15, p < 0.0001; post hoc test: control vs Aβ_1-42: p = 0.0003; control vs control +lonafarnib: p = 0.0204; Aβ_1-42 vs Aβ_1-42/+lonafarnib: p = 0.0007, n = 8 mice per group; two-way ANOVA, followed by Sidak's test; Fig. 2E). These results showed that, in the Aβ_1-42 mice, lonafarnib treatment enhanced Ca²⁺ influx and presynaptic transmitter release, as well as the synaptic plasticity. In the control mice, lonafarnib enhanced the synaptic plasticity, but not the presynaptic release. These findings indicated that lonafarnib rescued the deficiencies in the synaptic transmission and plasticity of Aβ_1-42 mice, improved the synaptic plasticity of control mice, but had no influence on the synaptic transmission of control mice.

Figure 2.

Lonafarnib enhances synaptic function of Aβ_1-42 mice and control mice. A, I/O curve (input/output curve). EPSP (excitatory postsynaptic potential) slopes were plotted against stimuli ranging from 0.1 to 1.1 mA in the control mice, control mice treated with lonafarnib (50 mg/kg; control+lonafarnib), Aβ_1-42 mice, and Aβ_1-42 mice treated with lonafarnib (50 mg/kg; Aβ_1-42/+lonafarnib). Typical traces evoked by test stimulation (70 μA); *p < 0.05, **p < 0.01 versus control mice, #p < 0.05, ##p < 0.01 versus Aβ_1-42 mice. CO: control mice, CL: control+lonafarnib mice, Aβ: Aβ_1-42 mice, AL: Aβ_1-42/+lonafarnib mice. B, PPF (Paired-pulse facilitation) of EPSPs. PPRs (Paired-pulse ratio, %) of EPSPs were plotted against various IPIs ranging from 25 to 100 ms. Typical traces evoked with IPI of 50 ms; *p < 0.05 versus control mice, #p < 0.05 versus Aβ_1-42 mice. C, Induction of LTP by HFS (high-frequency stimuli, 100 Hz–100 p) in slices. Each point represents the mean value (±SEM) of EPSP slopes expressed as percentage of baseline. A solid arrow, HFS was given. The PTP (post-tetanus potential, 1–5 min after HFS) and L-LTP (late-phase long-term potential 55–60 min after HFS) were marked by the box. D, E, The amplitude of PTP and L-LTP in four groups is expressed as the percentage of their baseline; **p < 0.01 versus control mice, ##p < 0.01 versus Aβ_1-42 mice.

Lonafarnib alters the morphology of dendritic spines in the Aβ_1-42 mice

The alterations in the morphology of synaptic dendrites and spines serve as a common factor in the pathogenesis of neurodegenerative disorders (Herms and Dorostkar, 2016; Batool et al., 2019), and may affect the synaptic functions. In addition, the pathologic process of AD is always accompanied by the structural changes in the neurons (Zagrebelsky and Korte, 2014). Thus, whether the phenotypic improvements after lonafarnib treatment in the Aβ_1-42 mice were related to the synaptic changes was further explored. Golgi staining was performed in the hippocampal CA1 region to examine the dendritic spine density and the morphology of pyramidal neurons in the Aβ_1-42 mice. As shown in Figure 3A, as compared with control mice, the dendritic length reduced in the Aβ_1-42 mice, which was reversed by lonafarnib treatment (interaction factor: F_(1,20) = 10.38, p = 0.0043; model factor: F_(1,20) = 5.487, p = 0.0296; treatment factor: F_(1,20) = 4.539, p = 0.0457; simple effect: Aβ_1-42 vs control: p = 0.0042, Aβ_1-42 vs Aβ_1-42/+lonafarnib: p = 0.0059, n = 6 mice per group; two-way ANOVA, followed by Tukey's test; Fig. 3A). Similarly, the branch number also decreased in the Aβ_1-42 mice, which was reversed by lonafarnib treatment (interaction factor: F_(1,20) = 7.207, p = 0.0142; model factor: F_(1,20) = 6.054, p = 0.0231; treatment factor: F_(1,20) = 4.859, p = 0.0394; simple effect: Aβ_1-42 vs control: p = 0.0082, Aβ_1-42 vs Aβ_1-42/+lonafarnib: p = 0.0122, n = 6 mice per group; two-way ANOVA, followed by Tukey's test; Fig. 3B). Sholl analysis of neurite tracings revealed a decrease in the dendritic complexity at different distances from the soma in the hippocampal neurons of Aβ_1-42 mice by analyzing the number of intersections, and this reduction was improved by lonafarnib treatment (interaction factor: F_(36,240) = 2.275, p < 0.0001; distance factor: F_(12,240) = 86.84, p < 0.0001; treatment factor: F_(3,20) = 4.994, p = 0.0096; simple effect: Aβ_1-42 vs control: 20 µm: p = 0.0010; 30 µm: p = 0.0441; 40 µm: p = 0.0137; 50 µm: p = 0.0291, n = 6 mice per group; Aβ_1-42 vs Aβ_1-42/+lonafarnib: 20 µm: p = 0.0124; 30 µm: p = 0.0119; 50 µm: p = 0.0339; 60 µm: p = 0.0012; 70 µm: p = 0.0139, n = 6 mice per group; two-way repeated-measure ANOVA, followed by Tukey's test; Fig. 3C). The number of spines in the apical dendrites from the pyramidal neurons also significantly decreased in the Aβ_1-42 mice as compared with the control mice, which was rescued by lonafarnib treatment (interaction factor: F_(1,20) = 5.686, p = 0.0271; model factor: F_(1,20) = 4.784, p = 0.0408; treatment factor: F_(1,28) = 4.835, p = 0.0398; simple effect: Aβ_1-42 vs control: p = 0.02, Aβ_1-42 vs Aβ_1-42/+lonafarnib: p = 0.0197, n = 6 mice per group; two-way ANOVA, followed by Tukey's test; Fig. 3D). These results suggested that lonafarnib treatment improved the impaired morphology of synaptic dendrites and spines in the Aβ_1-42 mice, but had no effect on it in the control mice.

Figure 3.

Lonafarnib improves the morphology of dendrites and spines in the Aβ_1-42 mice. A, B, Quantification of total dendritic length (A) and number of dendritic branches (B) of hippocampal CA1 pyramidal neurons in the control mice and Aβ_1-42 mice treated with lonafarnib or vehicle. C, Sholl plot comparing the dendritic arbors of pyramidal neurons of four groups. CO: control mice, CL: control+lonafarnib mice, Aβ: Aβ_1-42 mice, AL: Aβ_1-42/+lonafarnib mice. The x-axis represents the distance from the cell body, while the y-axis represents the number of intersections of a tracing with a particular Sholl shell. The insets show representative tracings of neurons superimposed on circles (0–200 μm, a circle per 20 μm) denoting Sholl shells. D, Quantification of dendritic spine density of pyramidal neurons of four groups; *p < 0.05, **p < 0.01 versus control mice, #p < 0.05, ##p < 0.01 versus Aβ_1-42 mice. The black rectangular frames are the selected areas with high magnification in the image below.

Four lonafarnib increases the number of synapses in the Aβ_1-42 mice

The expressions of presynaptic and postsynaptic markers, Syn and PSD-95, were further detected aiming to explore the changes in the excitatory synapses which are associated with dendrite and spine alterations. Both Syn and PSD-95 serve as vital markers of synaptogenesis and play an essential role in the synaptic plasticity (S. Hong et al., 2020). The reduced expressions of Syn and PSD-95 have been reported in the brain of AD patients (Sze et al., 1997; D.H. Kim et al., 2018), both Syn and PSD-95 are the targets of BDNF-TrkB intracellular signaling pathway (Cao et al., 2013; X. Chen et al., 2017). Immunofluorescence staining was conducted for the presynaptic marker Syn (red), postsynaptic marker PSD-95 (Red), and dendritic marker MAP2 (green) in the hippocampal CA1 stratum radiatum (Fig. 4). Results showed that the PSD-95-positive (red) area fraction (Fig. 4C) and PSD-95-positive puncta (Fig. 4E) decreased in the CA1 stratum radiatum of Aβ_1-42 mice as compared with the control mice, and lonafarnib rescued the reduction of synaptic proteins in the Aβ_1-42 mice [Fig. 4C, PSD-95-positive (red) area fraction: interaction factor: F_(1,28) = 4.318, p = 0.0470; model factor: F_(1,28) = 9.016, p = 0.0056; treatment factor: F_(1,28) = 2.77, p = 0.1071; simple effect: Aβ_1-42 vs control: p = 0.0012, Aβ_1-42 vs Aβ_1-42/+lonafarnib: p = 0.0132, n = 8 mice per group; two-way ANOVA, followed by Tukey's test; Fig. 4E, PSD95-positive puncta: interaction factor: F_(1,28) = 12.63, p = 0.0014; model factor: F_(1,28) = 36.11, p < 0.0001; treatment factor: F_(1,28) = 23.70, p < 0.0001; simple effect: Aβ_1-42 vs control: p < 0.0001, Aβ_1-42 vs Aβ_1-42/+lonafarnib: p < 0.0001, n = 8 mice per group, two-way ANOVA, followed by Tukey's test]. Similar tendency was observed in the Syn-positive (red) area fraction (interaction factor: F_(1,28) = 11.22, p = 0.0023; model factor: F_(1,28) = 12.74, p = 0.0013; treatment factor: F_(1,28) = 5.824, p = 0.0226; simple effect: Aβ_1-42 vs control: p < 0.0001, Aβ_1-42 vs Aβ_1-42+lonafarnib: p = 0.0007, n = 8 mice per group; two-way ANOVA, followed by Sidak's test; Fig. 4D), and Syn-positive puncta (interaction factor: F_(1,28) = 20.68, p < 0.0001; model factor: F_(1,28) = 29.93, p < 0.0001; treatment factor: F_(1,28) = 14.01, p = 0.0008; simple effect: Aβ_1-42 vs control: p < 0.0001, Aβ_1-42 vs Aβ_1-42/+lonafarnib: p < 0.0001, n = 8 mice per group; two-way ANOVA, followed by Tukey's test; Fig. 4F). Taken together, these results indicated that the improvements of synaptic plasticity and morphology after lonafarnib treatment in the Aβ_1-42 mice were accompanied by an improvement of synaptic density, as shown by the clusters of presynaptic and postsynaptic markers.

Figure 4.

lonafarnib improves presynaptic and postsynaptic markers in the CA1 pyramidal neurons of Aβ_1-42 mice. Representative images of CA1 pyramidal neurons stained for (A) PSD-95 (red) and MAP2 (green), (B) Syn (red) and MAP2 (green) in the control mice and Aβ_1-42 mice with or without lonafarnib treatment (left), and representative images of PSD-95-positvie and Syn-positive puncta along the dendrites of CA1 pyramidal neurons (right). CO: control mice, CL: control+lonafarnib mice, Aβ: Aβ_1-42 mice, AL: Aβ_1-42/+lonafarnib mice. Scale bar: A, B, 30 μm (left) and 20 μm (right). The white boxes are the selected areas with high magnification in the right image. Bar graph shows densities of PSD-95 and Syn-positive area fraction (C, D) and puncta along the dendrites of neurons in four groups (E, F); **p < 0.01 versus control mice, #p < 0.05, ##p < 0.01 versus Aβ_1-42 mice.

Lonafarnib increases hippocampal BDNF content in the Aβ_1-42 mice via α7nAChR

BDNF is related to synaptic plasticity and neuroprotection in animal models of neurodegenerative diseases or brain injury (Lu et al., 2014; Scharfman and MacLusky, 2014). Furthermore, BDNF has also been reported to modulate dendritic spine density (Tyler and Pozzo-Miller, 2001; Ji et al., 2010). It has been reported that α7nAChR activation can directly increase the calcium influx (Kabbani and Nichols, 2018) and regulate the PI3K/Akt and ERK pathways (Apati et al., 2003; Danciu et al., 2003). PI3K/Akt pathway mediates the increased release of BDNF in the brain (Ishrat et al., 2012; Duris et al., 2017). The phosphorylated ERK may subsequently activate downstream CREB, leading to its phosphorylation. Then, the phosphorylated CREB binds to the cAMP-response element (CRE) in the promoter region at exon IV of BDNF gene, regulating the transcription of BDNF (Benarroch, 2015). Chronic nicotine treatment significantly rescues the decreased autophosphorylation of CaMKIIα in null mice (Moriguchi et al., 2020). Moriguchi et al. (2015) reported that CaMKIIα autophosphorylation regulated BDNF transcription in the CaMKIV null mice via CREB (Ser-133) phosphorylation.

As membrane receptors, to carry out their physiological functions in neuronal cells, nAChRs have to travel from the endoplasmic reticulum (ER) to the plasma membrane through the secretory pathway, and then reach specific domains of plasma membrane such as the presynaptic axonal compartments or postsynaptic dendritic compartments (Colombo et al., 2013). Therefore, the expression of α7nAChR on cell surface which represents the functional activated receptor was further detected by testing the amount of biotinylation. Results showed lonafarnib treatment affected the cell surface expression of α7nAChR, which reduced in the Aβ_1-42 mice (interaction factor: F_(1,28) = 0.0388, p = 0.8451; model factor: F_(1,28) = 21.19, p < 0.0001; treatment factor: F_(1,28) = 18.83, p = 0.0002; post hoc test: control vs Aβ_1-42: p = 0.0041, control vs control+lonafarnib: p = 0.0133, Aβ_1-42 vs Aβ_1-42/+lonafarnib: p= 0.0067, n = 8 mice per group; two-way ANOVA, followed by Sidak's test; Fig. 5A). The phosphorylation of Akt was reduced in the Aβ_1-42 mice, which were enhanced by lonafarnib, and the α7nAChR antagonist MLA reduced Akt phosphorylation both in lonafarnib-treated Aβ_1-42 mice and control mice [interaction factor (model × lonafarnib × MLA treatment): F_(1,28) = 4.286, p = 0.0474; interaction factor (model × lonafarnib treatment): F_(1,28) = 5.732, p = 0.0409; model factor: F_(1,28) = 33.01, p < 0.0001; lonafarnib treatment factor: F_(1,28) = 35.4, p < 0.0001; MLA treatment factor: F_(1,28) = 38.29, p < 0.0001; simple effect: Aβ_1-42 vs control: p = 0.0157, Aβ_1-42 vs Aβ_1-42/+lonafarnib: p = 0.0433, control vs control+lonafarnib: p = 0.0310, Aβ_1-42/+lonafarnib vs Aβ_1-42/+lonafarnib+MLA: p = 0.017, control+lonafarnib vs control+lonafarnib+MLA: p = 0.0184, n = 8 mice per group; three-way ANOVA, followed by Tukey's test; Fig. 5B]. Different from Akt, the phosphorylation of ERK was not downregulated in the Aβ_1-42 mice, and lonafarnib also had no effect on its phosphorylation, however, MLA reduced ERK phosphorylation both in lonafarnib-treated Aβ_1-42 mice and control mice [interaction factor (model × lonafarnib × MLA treatment): F_(1,28) = 9.78, p = 0.0041; model factor: F_(1,28) = 1.685, p = 0.2048; lonafarnib factor: F_(1,28) = 1.674, p = 0.2064; MLA treatment factor: F_(1,28) = 72.06, p < 0.0001; simple effect: Aβ_1-42/+lonafarnib vs Aβ_1-42/+lonafarnib+MLA: p = 0.0002, control+lonafarnib vs control+lonafarnib+ MLA: p = 0.0005, n = 8 mice per group, three-way ANOVA, followed by Tukey's test; Fig. 5C]. The phosphorylation of CaMKII reduced in the Aβ_1-42 mice, which was rescued by lonafarnib treatment, and MLA treatment reversed the enhancement of the CaMKII phosphorylation in lonafarnib-treated Aβ_1-42 mice [interaction factor (model × lonafarnib × MLA treatment): F_(1,28) = 11.21, p = 0.0023; interaction factor (model × lonafarnib treatment): F_(1,28) = 11.21, p = 0.0023; model factor: F_(1,28) = 145.2, p < 0.0001; lonafarnib treatment factor: F_(1,28) = 50.02, p < 0.0001; MLA treatment factor: F_(1,28) = 2.486, p = 0.1261; simple effect: Aβ_1-42 vs control: p = 0.0048, Aβ_1-42 vs Aβ_1-42/+lonafarnib: p = 0.0146, Aβ_1-42/+lonafarnib vs Aβ_1-42/+lonafarnib+MLA: p = 0.0388, n = 8 mice per group; three-way ANOVA, followed by Tukey's test; Fig. 5D]. The phosphorylation of ERK remained unchanged in the lonafarnib treated control mice (control vs control+lonafarnib: p > 0.9999), but the Akt and CaMKII phosphorylation was upregulated (control vs control+lonafarnib: Akt: p = 0.0335; CaMKII: p = 0.0034). In addition, the CREB phosphorylation reduced in the Aβ_1-42 mice, and was rescued by lonafarnib (interaction factor: F_(1,28) = 15.37, p = 0.0005; model factor: F_(1,28) = 12.92, p = 0.0012; treatment factor: F_(1,28) = 10.21, p = 0.0034; simple effect: Aβ_1-42 vs control: p < 0.0001, Aβ_1-42 vs Aβ_1-42/+lonafarnib: p = 0.001, n = 8 mice per group; two-way ANOVA, followed by Tukey's test; Fig. 5E). However, lonafarnib had no effect on the phosphorylation of CREB in the control mice (p = 0.8192). Then, MEK inhibitor U0126 and CaMKII inhibitor KN93 were used to explore the pathway involved in lonafarnib-induced phosphorylation of CREB in the Aβ_1-42 mice. As shown in Figure 5F, U0126 reversed the upregulation of CREB phosphorylation by lonafarnib (interaction factor: F_(2,42) = 13.1, p < 0.0001; lonafarnib treatment: F_(1,42) = 13.25, p = 0.0007; inhibitor treatment: F_(2,42) = 12.89, p < 0.0001; simple effect: Aβ_1-42/+lonafarnib vs Aβ_1-42/+lonafarnib+U0126: p < 0.0001, n = 8 mice per group; Aβ_1-42/+lonafarnib vs Aβ_1-42/+lonafarnib+KN93: p < 0.0001, two-way ANOVA, followed by Tukey's test; Fig. 5F). Furthermore, the content of endogenous mature BDNF was detected by ELISA, and results showed that the hippocampal BDNF content significantly reduced in the Aβ_1-42 mice, which was reversed by lonafarnib treatment (interaction factor: F_(1,28) = 10.47, p = 0.0031; model factor: F_(1,28) = 17.00, p = 0.0003; treatment factor: F_(1,28) = 10.71, p = 0.0028; simple effect: Aβ_1-42 vs control: p < 0.0001, Aβ_1-42 vs Aβ_1-42/+lonafarnib: p = 0.0005, n = 8 mice per group; two-way ANOVA, followed by Sidak's test; Fig. 5G). In addition, the lonafarnib induced increase of BDNF content in the Aβ_1-42 mice was blocked by MLA and KN93, but not by U0126 or PI3K inhibitor Ly294002 (interaction factor: F_(4,70) = 10.45, p < 0.0001; lonafarnib treatment: F_(1,70) = 53.76, p < 0.0001; inhibitor treatment: F_(4,70) = 8.843, p < 0.0001; simple effect: Aβ_1-42/+lonafarnib vs Aβ_1-42/+lonafarnib+MLA: p < 0.0001, Aβ_1-42/+lonafarnib vs Aβ_1-42/+lonafarnib+KN93: p < 0.0001, Aβ_1-42/+lonafarnib vs Aβ_1-42/+lonafarnib+ U0126: p = 0.9997, Aβ_1-42/+lonafarnib vs Aβ_1-42/+lonafarnib+ Ly294002: p > 0.9999, n = 8 mice per group, two-way ANOVA, followed by Tukey's test; Fig. 5H). These results indicated that, in the Aβ_1-42 mice, lonafarnib improved the BDNF content in the hippocampus, through α7nAChR dependent CaMKII-CREB pathway.

Figure 5.

Lonafarnib increases the hippocampal BDNF content in the Aβ_1-42 mice in an α7nAChR dependent pathway. A, Expression of biotinylated α7nAChR (membrane) in the hippocampus of control mice and Aβ_1-42 mice treated with vehicle or lonafarnib. Surface α7nAChR was normalized to the surface Pan-Cadherin protein, which was again normalized by vehicle-treated control group; *p < 0.05, **p < 0.01 versus control mice, ##p < 0.01 versus Aβ_1-42 mice. B–D, Expressions of phospho-Akt, phospho-ERK, and phospho-CaMKII in the hippocampus of control mice and Aβ_1-42 mice treated with vehicle, MLA, lonafarnib, or lonafarnib/MLA; *p < 0.05, **p < 0.01 versus control mice; #p < 0.05, ##p < 0.01 versus lonafarnib treated control mice, +p < 0.05 versus Aβ_1-42 mice; $p < 0.05, $$p < 0.01 versus lonafarnib-treated Aβ_1-42 mice. CO: control mice, CM: control+MLA mice, CL: control+lonafarnib mice, CLM: control+lonafarnib+MLA mice, Aβ: Aβ_1-42 mice, AM: Aβ_1-42/+ MLA mice, AL: Aβ_1-42/+lonafarnib mice, ALM: Aβ_1-42/+lonafarnib+MLA mice. E, Expressions of phospho-CREB in the control mice and Aβ_1-42 mice treated with lonafarnib, **p < 0.01 versus control mice; ##p < 0.01 versus Aβ_1-42 mice. F, Expressions of phospho-CREB in the Aβ_1-42 mice or Aβ_1-42/+lonafarnib mice treated with U0126 or KN93; **p < 0.01 versus Aβ_1-42 mice; ##p < 0.01 versus Aβ_1-42/+lonafarnib mice. G, Bar graphs indicate the hippocampal BDNF content (%) in the control mice and Aβ_1-42 mice treated with lonafarnib, **p < 0.01 versus control mice, ##p < 0.01 versus Aβ_1-42 mice. H, The hippocampal BDNF content (%) in the Aβ_1-42 mice treated with MLA, U0126, Ly294002, KN93, lonafarnib, lonafarnib/MLA, lonafarnib/U0126, lonafarnib/Ly294002 or lonafarnib/KN93; **p < 0.01 versus Aβ_1-42 mice; ##p < 0.01 versus lonafarnib-treated Aβ_1-42 mice.

H-Ras but not Rhes is involved in the effects of lonafarnib on the α7nAChR and BDNF in the Aβ_1-42 mice

As described by previous study, lonafarnib induced lysosomal-mediated tau degradation and prevented pathology in a tau mouse via targeting Rhes (Hernandez et al., 2019). Study has reported that, in the Ras GTPase superfamily (e.g., H-Ras, N-Ras, K-Ras), SCH-66336 effectively inhibits the isoprenylation of H-Ras but not K- or N-Ras as these proteins can be alternatively lipidated by GGTase I (Ashar et al., 2001; Desrosiers et al., 2005). In our study, we speculated that lonafarnib improved the synaptic disease of Aβ_1-42 mice through Ras-mediated α7nAChR dependent BDNF upregulation. To investigate the effect of lonafarnib on the α7nAChR or BDNF in the Aβ_1-42 mice is related to H-Ras or Rhes, AAV-H-Ras, and AAV-Rhes were used to construct the H-Ras and Rhes overexpression mice. As shown in Figure 6A,B, the adeno-associated viral vectors (AAV-H-Ras and AAV-Rhes) effectively infected the hippocampus, produced considerable FLAG expressions (Fig. 6A,B). The mice with H-Ras (but not Rhes) overexpression showed reduced cell surface expression of α7nAChR, and lonafarnib affected the α7nAChR surface expression of all groups (interaction factor: F_(2,42) = 0.01,457, p = 0.9855; model factor: F_(2,42) = 10.54, p = 0.0002; treatment factor: F_(1,42) = 21.69, p < 0.0001; post hoc test: AAV-H-Ras vs AAV-Con: p = 0.0119; AAV-Rhes vs AAV-Con: p = 0.9444; AAV-H-Ras vs AAV-H-Ras+lonafarnib: p = 0.0258; AAV-Con vs AAV-Con+lonafarnib: p = 0.043; AAV-Rhes vs AAV-Rhes+lonafarnib: p = 0.0251, n = 8 mice per group; two-way ANOVA, followed by Sidak's test; Fig. 6C). H-Ras (but not Rhes) overexpression mice also showed reduced BDNF content, which was rescued by lonafarnib treatment, but lonafarnib had no effect on the BDNF content of control mice or Rhes overexpression mice (interaction factor: F_(2,42) = 9.599, p = 0.0004; model factor: F_(2,42) = 9.332, p = 0.0004; treatment factor: F_(1,42) = 9.059, p = 0.0044; simple effect: AAV-H-Ras vs AAV-Con: p < 0.0001; AAV-Rhes vs AAV-Con: p = 0.9879; AAV-H-Ras vs AAV-H-Ras+lonafarnib: p < 0.0001; AAV-Con vs AAV-Con+lonafarnib: p = 0.9973; AAV-Rhes vs AAV-Rhes+lonafarnib: p = 0.9988, n = 8 mice per group; two-way ANOVA, followed by Sidak's test; Fig. 6D). AAV-H-Ras, but not AAV-Rhes, could mimic the reduction of α7nAChR cell surface expression and the decrease of BDNF content in the control mice, which were both enhanced by lonafarnib treatment. However, lonafarnib also enhanced the surface expression of α7nAChR in the AAV-Rhes mice, which might be caused by lonafarnib inhibited H-Ras levels in the Rhes-overexpressed mice, similar with the effect of lonafarnib on the control mice, regardless the overexpression of Ras (T. Chen et al., 2020).

Figure 6.

H-Ras but not Rhes is involved in the effects of lonafarnib on the α7nAChR and BDNF in Aβ_1-42 mice. A, B, The hippocampus infected by adeno-associated viral vectors (AAV-H-Ras and AAV-Rhes; left), produced considerable FLAG expression (right). The white boxes are the selected areas with high magnification in the image below. C, D, Expression of biotinylated α7nAChR (C) and BDNF content (%; D) in the hippocampus of AAV-Con mice, AAV-H-Ras mice, and AAV-Rhes-mice treated with vehicle or lonafarnib; *p < 0.05, **p < 0.01 versus AAV-Con mice, #p < 0.05, ##p < 0.01 versus AAV-H-Ras mice, +p < 0.05 versus AAV-Rhes mice. CO: control mice+AAV-EGFP, CL: control+AAV-EGFP+lonafarnib mice, H-Ras: control+AAV-HRas mice, HL: control+AAV-HRas+lonafarnib mice, Rhes: control+AAV-Rhes mice, RL: control+AAV-Rhes+lonafarnib mice. E, F, The hippocampus infected by adeno-associated viral vectors (AAV-H-Ras and AAV-Rhes; left). G, H, Expression of biotinylated α7nAChR (G) and BDNF content (%; H) in the hippocampus of Aβ_1-42 mice and Aβ_1-42/+lonafarnib mice, injected with AAV-Con, AAV-H-Ras, and AAV-Rhes after lonafarnib treatment; **p < 0.01 versus Aβ_1-42+AAV-Con mice, ##p < 0.01 versus Aβ_1-42/+lonafarnib+AAV-Con mice. Aβ: Aβ_1-42+AAV-EGFP mice, AH: Aβ_1-42+AAV-H-Ras mice, AR: Aβ_1-42+AAV-Rhes mice, AL: Aβ_1-42+lonafarnib+AAV-EGFP mice, ALH: Aβ_1-42+lonafarnib+AAV-H-Ras mice, ALR: Aβ_1-42+lonafarnib+AAV-Rhes mice.

The AAV-H-Ras and AAV-Rhes were also used in the Aβ_1-42 mice after lonafarnib treatment for 14 d, to induce the H-Ras and Rhes overexpression, aiming to explore whether the supplement of H-Ras or Rhes reverses the effects of lonafarnib in the Aβ_1-42 mice. As shown in Figure 6E,F, AAV-H-Ras, and AAV-Rhes effectively infected the hippocampus. Interestingly, lonafarnib enhanced cell surface expression of α7nAChR and increased BDNF content in the Aβ_1-42 mice, both of which were significantly revered by H-Ras overexpression, but not by Rhes overexpression (α7nAChR: interaction factor: F_(2,42) = 9.066, p = 0.0005; model factor: F_(2,42) = 9.543, p = 0.0004; treatment factor: F_(1,42) = 26.05, p < 0.0005; simple effect: Aβ_1-42+AAV-Con vs Aβ_1-42/+lonafarnib+AAV-Con: p < 0.0001; Aβ_1-42/+lonafarnib+AAV-Con vs Aβ_1-42/+lonafarnib+AAV-H-Ras: p < 0.0001; Aβ_1-42/+lonafarnib+AAV-Con vs Aβ_1-42/+lonafarnib+AAV-Rhes: p > 0.9999, n = 8 mice per group; two-way ANOVA, followed by Tukey's test; Fig. 6G; BDNF: interaction factor: F_(2,42) = 25.95, p < 0.0001; model factor: F_(2,42) = 38.34, p < 0.0001; treatment factor: F_(1,42) = 111.2, p < 0.0001; simple effect: Aβ_1-42+AAV-Con vs Aβ_1-42/+lonafarnib+AAV-Con: p < 0.0001; Aβ_1-42/+lonafarnib+ AAV-Con vs Aβ_1-42/+lonafarnib+AAV-H-Ras: p < 0.0001; Aβ_1-42/+lonafarnib+AAV-Con vs Aβ_1-42/+lonafarnib+ AAV-Rhes: p = 0.7393, n = 8 mice per group; two-way ANOVA, followed by Tukey's test; Fig. 6H). Overexpression of H-Ras in the hippocampus of Aβ mice did not result in lower α7 cell surface expression (Aβ_1-42 vs Aβ_1-42/+AAV-H-Ras: p = 0.9472) and BDNF secretion (Aβ_1-42 vs Aβ_1-42/+AAV-H-Ras: p = 0.6196) than in the Aβ mice, possibly because of floor effect.

All these results indicate that H-Ras, but not Rhes, is related to the improvement of α7nAChR cell surface expression, BDNF content and BDNF-related synaptic function of Aβ_1-42 mice after lonafarnib treatment.

BDNF deprivation reverses the beneficial effects of lonafarnib on the cognitive function in the Aβ_1-42 mice

To explore the relationship of increased BDNF content with the improved cognitive function in lonafarnib-treated Aβ_1-42 mice, mice were pretreated with TrkB/Fc chimera protein. Results showed pretreatment with TrkB/Fc chimera protein significantly abolished the lonafarnib induced improvement of EPSP slopes (interaction factor: F_(1,28) = 5.207, p = 0.0303; lonafarnib treatment factor: F_(1,28) = 7.083, p = 0.0127; Fc treatment factor: F_(1,28) = 10.08, p = 0.0036; simple effect: Aβ_1-42/+lonafarnib vs Aβ_1-42/+lonafarnib+Fc: p = 0.0037, n = 8 mice per group; two-way ANOVA, followed by Sidak's test; Fig. 7A). TrkB/Fc chimera also affected the PPR (interaction factor: F_(1,28) = 2.763, p = 0.1076; lonafarnib treatment factor: F_(1,28) = 7.575, p = 0.0103; Fc treatment factor: F_(1,28) = 8.136, p = 0.0081; post hoc test: Aβ_1-42/+lonafarnib vs Aβ_1-42/+lonafarnib+Fc: p = 0.0069, n = 8 mice per group; two-way ANOVA, followed by Sidak's test; Fig. 7B). In addition, the PTP in the Aβ_1-42 mice treated with lonafarnib reduced in the presence of TrkB/Fc chimera protein pretreatment (interaction factor: F_(1,28) = 4.882, p = 0.0355; lonafarnib treatment factor: F_(1,28) = 7.352, p = 0.0113; Fc treatment factor: F_(1,28) = 8.104, p = 0.0082; simple effect: Aβ_1-42/+lonafarnib vs Aβ_1-42/+lonafarnib+Fc: p = 0.0077, n = 8 mice per group; two-way ANOVA, followed by Sidak's test; Fig. 7C). TrkB/Fc chimera protein also reduced the L-LTP amplitude (interaction factor [model × lonafarnib × Fc treatment]: F_(1,28) = 0.0014, p = 0.9702; interaction factor [lonafarnib × Fc treatment]: F_(1,28) = 17.14, p = 0.0003; model factor: F_(1,28) = 42.13, p < 0.0001; lonafarnib treatment factor: F_(1,28) = 7.108, p = 0.0126; Fc treatment: F_(1,28) = 8.418, p = 0.0072; simple effect: Aβ_1-42/+lonafarnib vs Aβ_1-42/+lonafarnib+Fc: p = 0.0262, control+lonafarnib vs control+lonafarnib+Fc: p = 0.0301, n = 8 mice per group, three-way ANOVA, followed by Tukey's test; Fig. 7D). In addition, BDNF deprivation compromised the improved spatial memory, manifested as the longer time to finding the hidden platform [interaction factor: (time × lonafarnib × Fc treatment): F_(4,70) = 1.675, p = 0.1655, (lonafarnib × Fc): F_(1,70) = 20.95, p < 0.0001; model factor: F_(4,70) = 22.85, p < 0.0001; lonafarnib treatment factor: F_(1,70) = 10.68, p = 0.0017; Fc treatment factor: F_(1,70) = 12.58, p = 0.0007; simple effect: Aβ_1-42/+lonafarnib vs Aβ_1-42/+lonafarnib+Fc: day 5: p = 0.0106, day 6: p = 0.0372, day 7: p = 0.0443; repeated-measure three-way ANOVA, followed by Tukey's test; Fig. 7E] and reduced the swimming time spent in the target quadrant, as compared with lonafarnib-treated Aβ_1-42 mice (interaction factor: F_(1,28) = 8.553, p = 0.0068; lonafarnib treatment: F_(1,28) = 4.531, p = 0.0422; Fc treatment: F_(1,28) = 5.831, p = 0.0225; simple effect: Aβ_1-42/+lonafarnib vs Aβ_1-42/+lonafarnib+Fc: p = 0.0040, n = 8 mice per group; two-way ANOVA, followed by Tukey's test; Fig. 7F).

Figure 7.

BDNF deprivation reverses the beneficial effect of lonafarnib on the spatial memory in the Aβ_1-42 mice. A–D, EPSP slopes recorded at 70 μA (A), PPR (B), PTP amplitude (C) in the Aβ_1-42 mice treated with vehicle, TrkB/Fc chimera protein, lonafarnib or lonafarnib/TrkB/Fc chimera protein; *p < 0.05 and **p < 0.01 versus Aβ_1-42 mice, ##p < 0.01 versus Aβ_1-42 mice treated with lonafarnib. D, L-LTP amplitude in the control and Aβ_1-42 mice treated with vehicle, TrkB/Fc chimera protein, lonafarnib or lonafarnib/TrkB/Fc chimera protein; *p < 0.05 versus control mice; +p < 0.05 versus Aβ_1-42 mice; #p < 0.05 versus lonafarnib treated control mice; $p < 0.05 versus lonafarnib-treated Aβ_1-42 mice. E, Latency (s) to finding visible-platform and hidden-platform in the MWM test of the Aβ_1-42 mice treated with vehicle, TrkB/Fc chimera protein, lonafarnib or lonafarnib/TrkB/Fc chimera protein. Tracings of typical swimming patterns in the hidden-platform task (upper panels). Black circles, position of platform; *p < 0.05 versus Aβ_1-42 mice, #p < 0.05 versus Aβ_1-42 mice treated with lonafarnib. Representative images of swimming paths in the hidden platform test (days 3–7) from different groups at day 6, which is representative and can show the difference in search time and tracks (right panels). Aβ: Aβ_1-42 mice, Aβ/FC: Aβ_1-42/+FC mice, AL: Aβ_1-42/+lonafarnib mice, AL/FC: Aβ_1-42/+lonafarnib+FC mice. F, Percentage of swimming time (%) in PQ in the MWM test of Aβ_1-42 mice treated with vehicle, TrkB/Fc chimera protein, lonafarnib or lonafarnib/TrkB/Fc chimera protein. Tracings of typical swimming patterns in the probe task (upper panels); **p < 0.01 versus swimming time in PQ of Aβ_1-42 mice, ##p < 0.05 versus swimming time in PQ of Aβ_1-42 mice treated with lonafarnib. Representative images of swimming paths in the spatial probe test (day 8, platform was removed) from different groups, which is representative and can show the difference in search time and tracks (right panels).

Discussion

To our knowledge, the present study for the first time reported the beneficial effects of Ras inhibitor lonafarnib on the synaptic structure and function in AD mice, providing a new insight and an effective way to treat the “synaptic disease” in AD patients.

Lonafarnib enhances the secretion of BDNF in an α7nAChR-dependent way in Aβ_1-42 mice, which is dependent on H-Ras but not Rhes

BDNF is initially transcribed and synthesized in the ER (Lessmann and Brigadski, 2009) as a form of precursor protein. Pro-BDNF is then proteolytically processed by endopeptidases (Benarroch, 2015; Kojima et al., 2020) and the cleaved BDNF is sorted by the secretion vesicles and secretory granules. The subsequent exocytosis is triggered by the α7nAChRs activation after nicotine binding, resulting in the release of a mixture of pro- and mature-BDNF (mBDNF; Machaalani and Chen, 2018). As a modulator of BDNF secretion, the activated α7nAChR has been reported to upregulate the release of BDNF (Machaalani and Chen, 2018; J. Kim et al., 2019). Our study showed that chronic lonafarnib treatment markedly enhanced α7nAChR cell surface expression and BDNF content (mature) in the hippocampus of Aβ_1-42 mice, the elevation of BDNF content was significantly inhibited by the selective α7nAChR antagonist MLA. This indicates the involvement of α7nAChR in the lonafarnib induced upregulation of BDNF content in the Aβ_1-42 mice. The α7nAChR activation induces PI3K/Akt phosphorylation, mediating the release of BDNF in the brain, and also induces ERK phosphorylation and Ca²⁺ release, which then regulates BDNF release or transcription via activating CREB (Ishrat et al., 2012; Duris et al., 2017).

In the present study, lonafarnib upregulates the α7nAChR-dependent activation of PI3K/Akt pathway in the Aβ_1-42 mice, but lonafarnib enhanced BDNF content in Aβ_1-42 mice was not affected by LY294002. PI3K/Akt is one of three downstream classical pathways of BDNF-TrkB (Reichardt, 2006). We speculate that, in the Aβ_1-42 mice, lonafarnib increases the PI3K/Akt phosphorylation, which is more modulated by α7nAChR or BDNF-TrkB pathway, but there is no great feedback to the BDNF expression. Ras activation is known to active Raf/ERK pathway (Shields et al., 2000). Lonafarnib has been reported to reduce the Ras-mediated ERK activation (Kloog et al., 1999). In our study, lonafarnib had no influence on the ERK phosphorylation in the Aβ_1-42 mice, and the ERK phosphorylation in lonafarnib treated control mice or Aβ1-42 mice reduced by MLA. This result indicates that ERK activation in the lonafarnib treated mice is mediated by Ras/Raf pathway and also modulated by the enhanced α7nAChR. This may be attributed to the enhanced activation of α7nAChR and BDNF-TrkB signaling (Gupta et al., 2013), both of which can activate ERK pathway. Interestingly, in the Aβ_1-42 mice, lonafarnib increased the CaMKII phosphorylation, which was inhibited by MLA, lonafarnib enhanced CREB phosphorylation was sensitive to the CaMKII inhibitor KN93, lonafarnib enhanced BDNF content was reduced by KN93.

As shown in Figure 6C,D, H-Ras (but not Rhes) overexpression in healthy control mice leaded the reduction of α7nAChR cell surface expression and decrease of BDNF content. BDNF content is affected by the activation of α7nAChR. The results mean that H-Ras is directly related to the α7nAChR cell surface expression. Our previous study indicated that Ras inhibition by simvastatin enhanced the Ca²⁺ influx of NMDAr, leading to an increase in the p-CaMKII (T. Chen et al., 2016a). Consistent with it, as shown in Figure 5D, an increase in the p-CaMKII was observed in the lonafarnib treated control mice. A large body of evidence indicates that the farnesylation of small GTPases may alter their interactions with intracellular molecules to regulate downstream effectors, including PKC, PKA, and CaMKII (McTaggart, 2006; T. Chen et al., 2018). In the control mice, downregulated H-Ras enhanced the α7nAChR cell surface expression through enhancing the p-CaMKII (T. Chen et al., 2018). CaMKII activating enhances α7nAChRs through stimulating α7nAChR trafficking or increasing receptor localization in the membrane (Kanno et al., 2012). As shown in Figure 5A, in the Aβ_1-42 mice, the Ras activation increased (Y. Wang et al., 2020), p-CaMKII reduced (Fig. 5D), and the α7nAChR cell surface expression was downregulated. In this way, we speculate that, as shown in Figure 6C, H-Ras overexpression might result in the downregulation of p-CaMKII, leading to a reduction of α7nAChR cell surface expression, consequently, leading to the downregulation of BDNF. However, lonafarnib also enhanced the α7nAChR surface expression in the AAV-Rhes mice, which might which might be caused by lonafarnib inhibited H-Ras levels in the Rhes-overexpressed mice, similar with the effect of lonafarnib on the control mice (T. Chen et al., 2020).

In the Aβ_1-42 mice, lonafarnib enhanced α7nAChR cell surface expression and increased BDNF content in the hippocampus, both of which were significantly compromised by H-Ras supplement, but not by Rhes supplement, indicating that the effects of lonafarnib on the α7nAChR and BDNF are related to H-Ras but not Rhes. Overall, we speculate that lonafarnib treatment upregulates hippocampal BDNF of Aβ_1-42 mice through targeting H-Ras, which is related to the α7nAChR-cascading CaMKII-CREB pathway.

Lonafarnib rescues the structural deficit of synapses to enhance synaptic plasticity

Our study indicated the alterations in the structural properties of synapses, prominent dendrites, spine pathology and loss of synaptic markers (PSD-95 and Syn) in the Aβ_1-42 mice as compared with control mice. These structural alterations of synapses were rescued by lonafarnib treatment, which was accompanied by the increase of BDNF content.

BDNF has been implicated in numerous processes of functional and structural synaptic plasticity (Gottmann et al., 2009; Park and Poo, 2013; Zagrebelsky and Korte, 2014), both of which are deficient in the AD (Spires-Jones and Hyman, 2014). BDNF-induced increase in the spine density depends on the membrane insertion of transient receptor potential canonical subfamily 3 (TRPC3) channel, which is related to the activation of TrkB-PLCγ pathway (Amaral and Pozzo-Miller, 2007). In addition, the activation of MAPK/ERK1/2 (Alonso et al., 2004; Ji et al., 2010) and Akt (Luikart et al., 2008), downstream factors of the TrkB, is required for the modulation of dendrite morphology on BDNF treatment in the hippocampal pyramidal neurons. In addition, BDNF produced by astrocytes plays a beneficial role in the dendrite maturation in the 5xFAD mice and cultured neurons, which is related to the BDNF-TrkB-PLCγ/ERK1/2 pathway (de Pins et al., 2019). Kellner et al. (2014) reported that BDNF-TrkB affected the F-actin content within the spine heads through the Rho GTPase Rac1, which was directly responsible for the changes in the spine density and morphology. BDNF-TrkB-PLC-γ and BDNF-TrkB-ERK pathways are also reported to be important for the correct PSD-95 location and expression, respectively (Parsons and Raymond, 2014; Yoshii and Constantine-Paton, 2014). BDNF downstream signaling can also regulate the Syn expression (Tartaglia et al., 2001; Zhang et al., 2017) and function/location (Bamji et al., 2006) to affect the number of synapses.

Therefore, we speculate that lonafarnib improves the dendrite and spine pathology in the hippocampus of Aβ_1-42 mice directly through upregulating BDNF and activating its downstream pathways; in addition, lonafarnib enhances the number of hippocampal synapses in the Aβ_1-42 mice through upregulating BDNF and its downstream pathways, which protects the dendrite and spine morphology. The exact pathways involved in the alteration of synapse structure are needed to be further studied.

Lonafarnib enhances synaptic function of Aβ_1-42 mice through increasing BDNF content

BDNF plays a crucial role in the neuroprotection and modulation of short-lasting and long-lasting synaptic interactions (Gonzalez et al., 2016; Sasi et al., 2017), and synaptic plasticity (Lynch et al., 2008). In our study, lonafarnib treatment in the Aβ_1-42 mice increased BDNF content significantly, enhanced presynaptic transmitter release, synaptic transmission and synaptic plasticity (LTP). PLC-γ pathway, as well as downstream ERK1/2 and Akt pathways (Abraham and Tate, 1997) of BDNF-TrkB, plays dominant roles in the hippocampal LTP induction (Minichiello, 2009). BDNF has the capability to change neurotransmitter release (presynaptic effect) via the TrkB- PLC-γ or ERK1/2 pathway (Alder et al., 2005; Gottmann et al., 2009) as well as postsynaptic receptor/channel properties (Gärtner et al., 2006). Akt and ERK signaling pathways play a crucial role in the LTP induction (Perkinton et al., 2002). Thus, lonafarnib may increase the BDNF content directly through these downstream pathways, improving the synaptic function of Aβ_1-42 mice.

PSD-95 and Syn are crucial for the synaptic functions (Shilpa et al., 2017; M. Hong et al., 2020). Syn has been identified as an integral membrane protein of presynaptic vesicles, and can regulate the synaptic transmission (Wiedenmann and Franke, 1985; Alder et al., 1995) and exocytosis (Valtorta et al., 2004). PSD-95 is an essential factor in the synaptic plasticity and dendritic spine formation at excitatory synapses, and it is also involved in the recruitment, trafficking and stabilization of NMDA receptors (Kim and Sheng, 2004; X. Chen et al., 2015). The beneficial effects of lonafarnib on the Syn and PSD-95 directly improve the synaptic transmission and plasticity in the Aβ_1-42 mice.

Taken together, our findings indicate Ras inhibitor lonafarnib can rescue the impaired dendritic spine and dendrite morphology and synaptic makers (Syn and PSD-95), consequently, improve the synaptic transmission and plasticity in the Aβ_1-42 mice via upregulating BDNF content through the H-Ras-α7nAChR-dependent CaMKII-CREB pathway.