Chronic basal forebrain activation improves spatial memory, boosts neurotrophin receptor expression, and lowers BACE1 and Aβ₄₂ levels in the cerebral cortex in mice

Abstract

The etiology of Alzheimer’s dementia has been hypothesized in terms of basal forebrain cholinergic decline, and in terms of reflecting beta-amyloid neuropathology. To study these different biological elements, we activated the basal forebrain in 5xFAD Alzheimer’s model mice and littermates. Mice received 5 months of 1 h per day intermittent stimulation of the basal forebrain, which includes cholinergic projections to the cortical mantle. Then, mice were behaviorally tested followed by tissue analysis. The 5xFAD mice performed worse in water-maze testing than littermates. Stimulated groups learned the water maze better than unstimulated groups. Stimulated groups had 2–3-fold increases in frontal cortex immunoblot measures of the neurotrophin receptors for nerve growth factor and brain-derived neurotrophic factor, and a more than 50% decrease in the expression of amyloid cleavage enzyme BACE1. Stimulation also led to lower Aβ₄₂ in 5xFAD mice. These data support a causal relationship between basal forebrain activation and both neurotrophin activation and reduced Aβ₄₂ generation and accumulation. The observation that basal forebrain activation suppresses Aβ₄₂ accumulation, combined with the known high-affinity antagonism of nicotinic receptors by Aβ₄₂, documents bidirectional antagonism between acetylcholine and Aβ₄₂.

Introduction

Alzheimer’s dementia was first identified by Alois Alzheimer in the early 1900s as a clear cognitive dementia (Möller and Graeber 1998). This dementia was later called Alzheimer’s disease. Studies in the 1970s began to show that deficits in the brain’s cholinergic systems were prominent in the disorder (Perry et al. 1978; Coyle et al. 1983). Projection neurons from the basal forebrain and hypothalamus target the entire cerebral sheet and hippocampus, and 90% of these projection neurons express choline acetyltransferase and likely release acetylcholine (Mesulam et al. 1983; Mesulam 2004; Zaborszky et al. 2012). These hypotheses have stood as the cholinergic hypothesis (Sarter and Bruno 1997; Bartus et al. 1982; Terry Jr and Buccafusco 2003), and the primary treatments for Alzheimer’s dementia, early in the disorder, are cholinesterase inhibitors (Birks 2006) that boost synaptic acetylcholine levels by suppressing its breakdown. Recent work has studied the complementary roles of the noncholinergic projection neurons (Semba 2000, 2004).

Another prominent hypothesis on Alzheimer’s dementia focuses on the neuropathological hallmarks, beta-amyloid plaques, and tau neurofibrillary tangles (Braak and Braak 1997; Braak et al. 2006). More recent evidence favors soluble beta-amyloid oligomers as being involved in the molecular mechanisms resulting in dementia (Balducci et al. 2010; Kayed and Lasagna-Reeves 2013). The oligomers have a number of neurobiological targets, and prominent among them are nicotinic acetylcholine receptors (Hernandez et al. 2010; Lilja et al. 2011; Inestrosa et al. 2013; Sadigh-Eteghad et al. 2014). We hypothesize that the interaction of beta-amyloid products with nicotinic receptors is central to some pathologies of Alzheimer’s dementia.

A well-documented finding from both postmortem Alzheimer’s dementia (AD) tissues and animal models of AD that is likely relevant to the hypotheses described above is the loss of neurotrophin support to the basal forebrain. Specifically, dysfunction of nerve growth factor (NGF) signaling via its high-affinity tropomyosin receptor kinase A (TrkA) receptor promotes the degeneration of basal forebrain cholinergic neurons and the progressive cognitive dysfunction of AD (Mufson et al. 2000; Counts et al. 2004; Ginsberg et al. 2006). There is also emerging evidence of a dynamic interplay between NGF/TrkA signaling and amyloid precursor protein (APP) metabolism within the context of AD neuropathology (Debeir et al. 1999; Bruno and Cuello 2006; Canu et al. 2017). Accordingly, novel therapeutic approaches designed to enhance both neurotrophin support in the brain and cholinergic activity could potentially have both pro-cognitive and disease modifying effects.

Our recent nonhuman primate work demonstrated that long-term activation of the basal forebrain resulted in cognitive improvements (Liu et al. 2017, 2018; Blake et al. 2017a; Qi et al. 2021). One hour of intermittent stimulation of the basal forebrain, for multiple days of the week over months, led to changes in working memory duration that were adequate to change a monkey’s rank percentile by 32–44% (Liu et al. 2017; Blake et al. 2017b). Although these effects were reliably induced, and acute impacts of stimulation on cognition were pharmacologically consistent with cholinergic mechanisms, underpinnings of change over those longer time scales were unclear.

We set out to test hypotheses on mechanisms of these improvements by performing long-term stimulation in mice. The mouse model enabled direct visualization of acetylcholine levels in the cortex from stimulation. We also chose mice to facilitate protein expression analysis. We chose the 5xFAD mouse for the Alzheimer’s model because it aggressively deposits Aβ₄₂ and has a genotype-dependent decline in cognition (Oblak et al. 2021). By applying deep brain stimulation of the basal forebrain, we evaluated potential behavioral improvements dependent on stimulation, and protein expression key to the cholinergic signaling, neurotrophin support, and beta-amyloid metabolic pathways.

Materials and methods

Test subjects

B6SJL-Tg (APPSwFlLon, PSEN1^*M146L^*L286V) 6799Vas/Mmjax or 5XFAD was used for study. Transgenic 5XFAD and wild-type littermates were identified by PCR using a sense primer (5′-CTA CAG CCC CTC TCC AAG GTT TAT AG-3′) and antisense primers (5′-AAG CTA GCT GCA GTA ACG CCA TTT-3′) and (5′-ACC TGC ATG TGA ACC CAG TAT TCT ATC-3′). The Institutional Animal Care and Use Committee of Augusta University approved of all protocols and procedures used in this study. Under the guidance of the 8th Edition of the National Institute of Health Guide for Care and Use of Laboratory Animals (2011), measures were taken to minimize pain and discomfort throughout the study.

Electrode fabrication

Stimulation electrodes were custom-made based on previously published specifications (McCairn and Turner 2009). Stainless steel 30 ga. hypodermic tubing (MicroGroup®, Medway, MA) was cut to a length of 2.8 mm using a Dremel® rotary saw. A 26G (Precision Glide®) needle was twisted into each end of the tubing to re-establish their patency and a tungsten wire reamed out any other debris. Teflon-coated platinum–iridium wire with an outer diameter of 139.7 μm (A-M Systems, Seattle, WA) was cut to 20 mm in length and threaded through the hypodermic tubing. Iris scissors grasped and unsheathed 0.5 mm of the Pt–ir wire to create the conductive tip. Polyimide tubing was cut to a 3.8 mm length and slide onto the hypodermic tubing and wire, so that the 1.0 mm excess polyimide tubing was left at the tail end of the electrode. The stimulation tip was adjusted in the hypodermic tubing exposing the 0.5 mm of uncoated wire. Using the included brush, cyanoacrylate (Krazy®) was applied to the excess polyimide opening until glue was drawn into the area creating a 4.8 mm long electrode from the stimulation tip to end of polyimide tubing. A butane torch uncoated about 3 mm of the tail-end wire.

Surgical procedures

Stimulation electrode and headcap implantation

Surgery was performed on a sterile field under 1.5% isoflurane anesthesia. The mouse was placed in a dual arm, rodent stereotaxic instrument (Stoelting®, Wood Dale, IL) fitted with an anesthesia mask and heating pad set to 37 °C. Stereotaxic measurements were made in reference to the bregma landmark (−0.6 mm AP or caudal, 1.7 mm ML or lateral) to target sites for electrode implantation. A 0.5-mm hole was drilled at target sites using a carbide bur (HM1-005-FG, Meisinger, Centennial, CO) and dental drill (Volvere GX, Brasseler, Savannah, GA) until the cranium was breached. An electrode was loaded into a guide tube positioned above the target site and pushed into the cerebrum with a stylus until the back of the electrode became flush with the cortical surface to ensure achievement of 4.8 mm depth. Cyanoacrylate was placed at the target site to stabilize the electrode and seal off exposed tissue. The procedure was repeated for implantation of the second electrode in the contralateral hemisphere. All protein expression and behavioral testing animals that received stimulation were stimulated equally bilaterally. With the same burr, a hole was made at each temporal and frontal bone and stainless steel screws (Antrin, Fallbrook, CA) were hand screwed two rotations. Before implanting the final bone screw in the left frontal bone, a 20-mm length of pt–ir wire with 3 mm of coating removed at each end was procured. One end was bent with forceps and then slid into the bone screw hole so that the tip resided between the top of the cortex and bottom of the skull. A layer of cyanoacrylate was applied to all screws and electrode sites and left until dry. Dental cement was applied to the top of the skull with the periosteal elevators. A type B micro-USB receptacle (part no. 84X0185, Newark®, Chicago, IL), with the 2nd and 4th connector pins removed, was fixed to one of the stereotaxic arms and positioned over the skull. The exposed wire from the bone screw was soldered to the grounding pin of the micro-USB port while the electrode wires were soldered to the remaining 2 pins. A thin coat of dental cement was applied to the soldered pins to ensure firm connection. The micro-USB receptacle was dismounted from the stereotactic arm and positioned just above the top of the bone screws. A larger stainless steel screw (MX-M025, Small Parts Inc., Logansport, IN) was adhered to the side of the micro-USB receptacle that would serve as a handle when connecting mice to the stimulating computer. Dental cement was applied to fill in all areas between the base of the skull and base of the micro-USB receptacle to create the stimulating headcap. A micro-USB receptacle was placed into the top of the headcap to prevent cage-mates from chewing on the internal connector. The mouse was returned to the cage and monitored until conscious and consuming water and food. Control mice underwent identical procedures; however, the electrode and grounding wires were clipped instead of attaching to the micro-USB receptacle pins.

Lesions were created in pilot animals developing implantation approaches until targeting was consistent. Lesions were created with a 30 μA DC current, electrode negative, for 10 s, which generated a clearly visible mark on perfused and unstained sections (Fig. 2A).

Fig. 2.

Electrode position and behavioral impact of stimulation. (A) Electrolytic lesion mapping. Coronal wet specimen of mouse brain after delivery of anodal current pulses to induce an electrolytic lesion at site of exposed microelectrode tip implanted −0.6 mm AP, 1.7 mm ML, and 4.8 mm DV to bregma. Black arrow: cortical site of microelectrode penetration. Black circle: electrolytic lesion. Black region on right: substantia innominata region of intended stimulation from Allen mouse brain reference atlas. (B) Effects of stimulation on water maze escape latency by group are plotted on graphs. Sample desaturated heatmap tracks from each group are shown on the right. (C) Visible platform test. Mean escape latency to visible platform for stimulated and unstimulated 5xFAD and wild-type mouse groups. (D) Accelerating rotarod performance. Mean time spent balancing on rotating rod for each mouse group. (E) Water maze probe test. Mean number of mouse crossings over the removed escape platform per mouse group. Control wild-type mice: N = 6 males and 6 females. Stimulated wild-type mice: 7 males and 7 females. Control 5xFAD mice: N = 3 males and 4 females. Stimulated 5xFAD mice: N = 9 males and 2 females. ^** indicates P < 0.01.

Viral transduction: Surgery was performed on a sterile table under 1.5% isoflurane anesthesia. A small hole was carefully made (−2.0 mm AP, 2.0 mm ML from bregma) over the somatosensory cortex. A micropipette was inserted to a depth of 700 μm. Non-transgenic littermates were injected with 400 nL of (AAV9-hsyn-ACh4.3, Vigene) at 100 nL/min using a micropump (WPI). The ACh4.3 was later renamed GACh3.0, and for reduced confusion we refer to it as GACh3.0 in Section Results. After a 5-min wait period, the micropipette was slowly raised. The viral vector incubated for 3 weeks. To confirm vector expression, mice were anesthetized and prepped as stated above. The scalp was re-incised to expose the injection site, and the presence of fluorescence was checked using confocal epifluorescence microscopy fitted with a fluorescein isothiocyanate (FITC) filter (emission of 513-556 nm, excitation of 467–498 nm). Mice demonstrating positive fluorescence and thus GACh3.0 expressions were then implanted with an optical window and single microelectrode respectively detailed below.

Optical window and electrode implantation

Implantation of a cranial window was performed as described previously (Mostany et al. 2008). Briefly, after mice were anesthetized with isoflurane (4% induction, 1.2–1.5% maintenance), mice were shaved and then skin and connective tissue attached to the skull were removed. To form a ~3 x 3 mm round cranial window, cranial bone over the somatosensory cortex was carefully removed using a 1/4 round carbide dental bur (Midwest). The brain was then irrigated with sterile cortex buffer containing the following (in mM: 135 NaCl, 5.4 KCl, 1 MgCl2, 1.8 CaCl2, and 10 HEPES, pH 7.3), and a 5-mm-diameter #1 glass coverslip (Electron Microscopy Sciences) was placed over the window and sealed to a custom-made metal ring using cyanoacrylate and dental cement. A single 6.8 mm microelectrode was fabricated and inserted as described previously but at −5.4 mm AP and 1.7 mm ML to bregma and implanted at a 45° angle to the horizontal plane to stimulate the basal forebrain without obstructing the optical window. As in stimulation electrode placement shown in Fig. 2A, a series of experiments using electrolytic lesions were performed in pilot animals to ensure the electrode position was precisely the same in the imaging and behavioral experiments.

Two photon imaging: The Nikon A1R MP multiphoton system mounted on the FN1 upright microscope was used to collect images with a 25x/1.1 NA water-immersion objective. The Spectra-Physics Mai-Tai EHP laser tuned to 950 nm was used for 2-photon excitation, and emission light was collected by a GaAsP detector using a bandpass filter (500–550 nm). The time-lapse images consisting of a single section were taken at 1 frame per second for 5 min across 78 × 78 μm imaging field within layer I of the sensorimotor cortex.

Brain stimulation

Electrode impedance measurement

The individual impedances of each electrode had to be calculated so that the stimulation voltages could be tailored to deliver 100 μA of current to each electrode tip. Because previous pilot studies demonstrated a consistent daily impedance, impedances were measured at the beginning of each week. Impedance was measured using custom programmed software that triggered a stimulus isolator (A385 World Precision Instruments, Sarasota, FL) to produce a known 100 μA current square wave and then use a connected oscilloscope (TDS 2012C Tektronix, Beaverton, OR) to measure the voltage change. The same method was applied to the left and right electrodes of each mouse head cap. Impedances averaged 8–14 kΩ. Variable resistors were added to the stimulation cord circuit path when needed to stimulate multiple mouse electrodes at the same time with a uniform voltage output. Typically, stimulation was delivered at close to 1 V peak per phase across an impedance of 10 kΩ.

Mouse stimulation: Mice were brought to the stimulation room and allowed 30-min of acclimation before beginning stimulation. Remaining in their home cages, the large screw adhered to the side of the micro-USB receptacle was secured with hemostats while a micro-USB cable was plugged into each mouse headcap. The mouse cage lids were returned ajar so that mice had continued access to food and water. All micro-USB charging cables were threaded through a pulley system equipped with a counter balance and suspended from the ceiling to remove any upward or downward forces on the headcaps during stimulation and allowing for better mouse mobility. Tantalum capacitors (10 μF 25v, Kemet, Fort Lauderdale, FL) were soldered to each stimulating cable end to block net charge transfer. These capacitors and the cable’s grounding wires were connected to a multiple functional I/O device (USB-6211, National Instruments, Austin, TX) that served as a voltage generator creating the voltage-dependent stimulation pulses signaled by custom programmed software. Stimulation was delivered with biphasic, negative first, unipolar 100 μA pulses with 100 μS per phase. 60 pulses were delivered per second for 20 s followed by 40 s of no stimulation. Mice received 1 h of bilateral stimulation 5 days each week. We estimate the activating function radius for this stimulation as 315 μm using prior work as our guide (Stoney et al. 1968; Murasugi et al. 1993; McIntyre et al. 2004). The stimulation typically evoked a brief voltage change of 1 volt (Rozman et al. 2000), safely within limits for platinum for brief pulses (Rose and Robblee 1990), and had charge densities well below limits for tissue damage (Shannon 1992).

An observation during stimulation sessions was a universal augmentation in mouse activity during the 20 s intervals of stimulation. This activity primarily manifested as hand-to-mouth self-grooming behaviors and occasionally digging. These behaviors would immediately cease during the 40 s rest interval, and this phenomenon continued for about 6 weeks. It is difficult to discern if this observation was an increase in wakeful activity or a biproduct of tongue or hand paresthesias, which have been reported in the deep brain stimulation of human Parkinson’s disease subjects due to activation of ventral intermediate nucleus of the thalamus or the internal capsule with afferent fibers proximal to the subpallidal basal forebrain (Ondo et al. 1998; Kim et al. 2021). These changes upon stimulation beginning were prominent and reliable enough that electrode impedance would be immediately checked if the behavioral changes were not observed, in the first 6 weeks. Thereafter, stimulation caused behaviors ceased.

Tissue processing

Brain extraction

Mice were placed in a bell jar with 0.5 mL of isoflurane until loss of righting reflex and unresponsive to toe pinch. Grasping the base of the skull and tail, manual cervical dislocation was performed to euthanize the mouse. Surgical scissors transected through cervical tissue and spinal cord in order to isolate the head. A periosteal elevator removed the brain from the cranium and transferred it face down onto a sterile petri dish on a bed of crushed ice. A #11 scalpel made a coronal incision at the origin of the cranial nerve II white matter tracts. The cortical portion of the anterior section was removed with forceps and a scalpel. The cortical portion of the posterior brain section was reflected from the midline using a blunt probe to expose both hemispheres of the hippocampus, which were resected with forceps. All frontal cortex and hippocampus tissues were separated by hemisphere and transferred to labeled 1.7 mL microcentrifuge tubes on ice and then stored at −80 °C.

Tissue homogenization

About 100 μl of 4 °C radioimmunoprecipitation buffer (Sigma-Aldrich, St. Louis, MO) with 10 μl/ml of protease inhibitor cocktail (AEBSF at 104 mM, Aprotinin at 80 μM, Bestatin at 4 mM, E-64 at 1.4 mM, Leupeptin at 2 mM and Pepstatin A at 1.5 mM, Sigma-Aldrich, P8340) were added to each frozen brain section on ice. Samples received a 5-s pulse from an ultrasonic homogenizer (Qsonica®, Newtown, CT) while sitting in crushed ice. Homogenized samples underwent centrifugation at 15,000 g for 15-min at 4 °C. Supernatants were transferred to pre-chilled microcentrifuge tubes and stored at −80 °C until used for protein analysis.

Protein analysis

Western blot analysis

Protein concentrations of each sample were determined using the bicinchoninic acid (BCA) protein assay kit (Thermo Fisher Scientific, Waltham, MA), which used bovine serum albumin standards to create a best fit linear regression line based on protein concentrations using the 562 nm absorbance mode of a microplate reader and its Gen5 Data Analysis program (Biotek, Winooski, VT). Equal amounts of protein were resolved in SDS-polyacrylamide gels and electrophoretically transferred to nitrocellulose membranes (Bio-Rad, Hercules, CA) at 4 °C. Membranes were blocked for 1-h in Tris-buffered saline containing Tween 20 (TBST; 10 mM Tris–HCl, pH 8.0, 138 mM NaCl, 2.7 mM KCl, and 0.05% Tween 20) and 5% non-fat milk.

Primary antibodies used were anti-brain-derived neurotrophic factor (BDNF; 1:1,000, GTX134514, GeneTex®, Irvine, CA), anti-NGF, (1:1,000, AN-240, Alomone Labs), anti-TrkA (1:1,000, GTX132966, GeneTex®, Irvine, CA), anti-pTrkA-Y490 (1:1,000, product no. 9141S, Cell Signaling Technology®, Danvers MA), anti-TrkB (1:1,000, GTX54857, GeneTex®, Irvine, CA), anti-pTrkB-Y705 (1:1,000, ab229908, Abcam, Cambridge, MA), anti-BACE1 (1:1,000, GTX103757, GeneTex®, Irvine, CA), and anti-ADAM10 (1:1,000, GTX104940, GeneTex®, Irvine, CA).

Primary antibodies were added to 10 mL of 5% non-fat milk and TBST solution and applied to the blocked membrane to incubate overnight on an orbital shaker at 4 °C. The following day, membranes were washed 3 times with TBST for 5-min and then incubated for 1-h with horseradish peroxidase-conjugated goat anti-rabbit IgG (ab6721, Abcam, Cambridge, MA). Membranes were washed 3 more times with TBST for 5 min.

Blots were developed using Pierce™ ECL Western Blotting Substrate (ThermoFisher, USA, Cat# 32106) at a sufficient volume to ensure that the blot was completely wet with the substrate and the blot did not become dry (0.1 mL/cm²). Membrane incubated with the substrate working solution for 1 min. Blot was removed from ECL solution and placed in a plastic sheet protector or clear plastic wrap. An absorbent tissue removed excess liquid and carefully pressed out any bubbles from between the blot and the membrane protector. Then the blot was analyzed using The ChemiDoc XRS+ Gel Imaging System (BioRad, USA). The images were analyzed using Image Lab image acquisition and analysis software (BioRad, USA).

To avoid saturation and ensure linearity, 30 μg of each sample were loaded in the gel. Study has shown that GAPDH showed linearity up to 30 μg of protein loaded. Along with this, gels were stained with Coomassie stain to ensure equal amount of protein loaded (Eaton et al. 2013; Fosang and Colbran 2015).

For the different molecular weights, membranes were cut into strips (horizontally using protein ladder/marker), then probed individually with different antibodies. However, for very close molecular weights and/or for the GAPDH, membrane blots were stripped using western blot stripping buffer (Thermo Fisher, USA, cat# 21059). For stripping, blots were washed 5 min x 3 with TBST to remove chemiluminescent substrate, and then incubated in western blot stripping buffer for 5–5 min at RT. Then they were again washed 5 min x 3 with TBST, and then performed next immunoblot experiment. The original blots are provided in the supplementary file.

Enzyme-linked immunosorbent assay: Following the manufacturer’s instructions for the mouse Aβ₄₂ enzyme-linked immunosorbent assay (ELISA) kit (Invitrogen®, Waltham, MA), a solution of guanidine-HCl in 50 mM Tris (pH 8.0) was added to homogenized brain samples to achieve 5 M guanidine-HCl and mixed for 4-h on orbital shaker at room temperature. Samples were quantified with the BCA kit and diluted with the included diluent buffer to achieve kit recommended concentrations. Samples were then added to the well plate, coated with monoclonal antibody against the NH₂ terminus of Aβ₄₂, covered with an adhesive plate cover, and incubated for 2 h at room temperature. 100 μl of Ms detection antibody solution was added to each well. The plate was re-covered and incubated at room temperature for 1 h. Wells were washed 4 times using a squeeze-type wash bottle with the provided wash buffer. 100 μl of Anti-Rabbit IgG HRP solution was added to each well and incubated for 30 min. Wells were rewashed as previously stated. 100 μl of stabilized chromogen was added to each well and incubated for 30 min before adding 100 μl of the provided stop solution. Aβ₄₂ standards were included in this protocol to create a best-fit standard curve and sample absorbance was detected at 450 nm using the microplate reader. Aβ₄₂ sample concentrations were calculated by dividing the Aβ₄₂ concentration from the ELISA kit by the total protein concentration from the BCA kit.

Water maze

Test apparatus

The water maze consisted of a circular tank (1.80-m diameter and 0.76-m height) made of black plastic and filled to a depth of 35 cm of water. An escape platform (20-cm diameter) was placed 1 cm beneath the surface of the water and centered in 1 of the 4 tank quadrants. Water (maintained at 22 ± 1.0 °C) was made opaque with nontoxic white paint. The maze was surrounded by visual cues including a green watering can, blue lab coat hanging from a standing hanger, and black curtains, on which white paper geometric images were affixed to provide visual cues. These curtains also hid the experimenter and resting test subjects from view. Swimming activity was monitored using an overhead mounted camera that relayed mouse swimming location and trajectory as well as latency to platform measurements to a video tracking system (Noldus EthoVision® Pro 3.1).

Hidden platform task

For each test day, mice were transferred to new testing cages and then given 30 min to acclimate to the water-maze room surroundings before starting the first trial of the day. For each daily test session, mice were given 90-s per trial to locate and climb onto the escape platform. A trial was initiated by placing each mouse into the water directly facing the center of the maze pool at 1 of 7 locations around the pool wall. Starting locations were pseudorandomized so that starting locations were equally distributed among the 4 quadrants for every mouse over the 5-day testing period. Mice that found the platform successfully were given a 30-s rest period on the platform before being returned to their testing cage. Mice that did not find the platform within the 90-s trial were guided to it using a curved plastic handle and then given a 30-s rest period before returning to the testing cage. Mice were given a 30-min rest in their training cage before undergoing the subsequent trial. Each day consisted of 3 trials per day for 5 consecutive days.

Probe trials

After 24 h from the last hidden platform task, mice were returned to the water-maze room with a 30-min acclimation period. The hidden platform was removed from its quadrant in the pool. For this single trial, each mouse was placed at the same start position in the quadrant opposite of the previously hidden platform quadrant. For the observer on the tracking monitor, the video tracking system projected a 30 cm radius circle over the previous location of the escape platform. Each mouse was given 90 s in the pool, and the observer counted how many times each mouse enters and then re-enters the probe test zone.

Visible platform task

To confirm general visual acuity in test animals, a screwdriver was placed face down through one of the escape platform holes so that the red handle was raised 12 cm above the water surface. After a 30-min period after completing the probe trial, each mouse was placed in the same starting location and time to swim to the platform was measured.

Y-maze spontaneous alternation test

Spatial working memory was evaluated using Y-maze spontaneous alternation test. The Y-shaped maze consisted of 3 identical arms (30 x 6 x 15 cm) positioned at 120° to each other. The test was conducted during the late light phase of the light dark cycle under dim illumination. Briefly, mice were placed into one of the arms, faced to the center of the maze and allowed to freely explore the maze for 10 min. Rodents typically show a natural tendency to choose to enter a new arm of the maze rather than returning to one that was recently visited; this choice is called alternation. The sequence of arm choices and the total number of arm entries were recorded. The percentage of spontaneous alternation was calculated as [actual number of alternations/(total number of arm entries – 2)] × 100. While the alternation percentage was used as a measure of spatial working memory, the total number of arm entries during the test sessions reflected spontaneous activity.

Accelerating rotarod

Test apparatus

Motor coordination and balance were tested using an accelerating rotorod (Rotor-Rod System®, San Diego Instruments).

Rotorod training

Mice were placed into testing cages and then brought to the testing room to acclimate for 30 min. Each mouse was then placed on the station rod for 10 s before starting the rotation motor. The rod was then gradually increased from 0 to 10 rpm over a 1-min period and maintained at 10 rpm for 5 min. During the first trial, mice that fell off of the rod were placed back onto it until the end of the 5-min duration. Mice performed 3 trials total with an intertrial interval of 20-min and latency to fall was recorded for each. Trials where mice hang onto the rod and make 2 continuous rotations without any further footsteps would be considered a fall and the latency also recorded. Mice were then returned to their home cages and room.

Rotorod test

After 24 h from training, mice were placed into training cages and brought to the rotorod room for a 30-min acclimation period. Each mouse was given a single training trial as described above to refresh them of the task and returned to their testing cages. After a 20-min rest, mice were returned to the rod that would now increase from 0 to 23 rpm over a 2-min period during the 5-min trial. Mice underwent 3 trials with 20-min rest in between each trial and latency to fall was recorded for each mouse. Trials where mice made 2 consecutive rotations without taking a step were recorded as a fall.

Statistical analysis

Statistical analyses were performed using GraphPad Prism 9 software.

Results

Cortical acetylcholine release caused by stimulation

A fluorescent reporter system and mesoscale imaging were used to validate release of acetylcholine was caused by the applied stimulation. Groups of wild-type mice had the vector GACh3.0 infused into the cerebral cortex (Jing et al. 2018). After verification of expression, animals were implanted with both an optical window in somatosensory cortex and an electrode positioned in the floor of the globus pallidus (basal forebrain), and mesoscale fluorescent imaging was conducted. Somatosensory cortex was chosen for convenience in imaging, as the basal forebrain cholinergic projections release acetylcholine and cause blood flow increases across the entire cortical mantle except the hippocampus and medial temporal lobe (Sato and Sato 1995). The same electrode tip location was used in all experiments reported in this manuscript. Expression of the receptor is shown in Fig. 1A. This receptor expresses throughout the plasma membrane of expressing neurons (Jing et al. 2018). Responses to 20 s pulse train stimulation are shown in Fig. 1D. Each trace in Fig. 1D and F is a raw frame intensity average from 1 trial. The average fluorescent intensity per frame increased 15% within 4 s of the onset of stimulation, and returned to baseline 30 s after stimulation ended. Continuous pulse-train stimulation was applied, and the induced fluorescent intensity changes are shown in Fig. 1E. The continuous stimulation induced a similar response to the 20 s pulse train in the first 20 s. Nonintuitively, after the first 20 s the fluorescent intensity began to decline, and reached baseline, or lower, within 3 min. Intermittent stimulation was applied, and fluorescent intensity changes were observed for each 20 s pulse train separated by 40 s of rest. Fig. 1B and C shows the induced fluorescent intensity pixel by pixel for one of the intermittent stimulation examples near peak fluorescence and near the low point to show the spatial pattern of fluorescence remains the same throughout the cycle.

Fig. 1.

GACh3.0 responses to stimulation. (A) Cortical expression of the GACh3.0 reporter focused in layer 1, 20 microns below the surface in a C57bl/6j mouse. (B) Single frame color-coded fluorescence intensity from a sample image at the nadir of intensity in (F). (C) As in panel (B), color-coded fluorescence intensity at the peak. (D) Phasic response to 20 s of 60 Hz pulse stimulation in mouse one. (E) Overlaid mesoscale average-pixel fluorescence intensity traces from mouse 2 and 3 showing responses to continuous 60 pulse per second stimulation. (F) As in panel (E), for 60 pulse per second stimulation that cycles on for 20 s and off for 40 s. (D–F) Single trial mesoscale frame-average fluorescence intensity data. Gray bars in D, E, and F indicate times of stimulation.

Behavioral results

Groups of wild-type and 5xFAD mice were implanted bilaterally with stimulating electrodes in the basal forebrain. Targeting of the basal forebrain was performed in test animals by electrolytic lesioning. Stereotaxic coordinates were practiced until macrohistological reconstructions demonstrated consistent targeting. One example is shown in Fig. 2A. Stimulation was applied with intermittent stimulation for 20 s of each minute, and 40 s of rest each minute, for 1 h per day, 5 days per week. Each pulse in the 60 pulse per second trains was monopolar and biphasic, negative first, with 100 μA current, and 100 μs per phase. Animals were stimulated in standard housing boxes in the lab, and were under no behavioral control. Animals were stimulated from month four of life until month 9, at which point they were entered in behavioral testing. The 5xFAD mouse begins depositing beta-amyloid plaques at 2 months of age (Oakley et al. 2006), begins to show cognitive deficits at 4 months of age (Girard et al. 2014), and begins to show locomotor deficits at 9 months of age, with more clear motor deficits at 12 months of age (O’Leary et al. 2020). Genotype-dependent water-maze deficits have been documented in the 7–9 month age range (Urano and Tohda 2010).

In water-maze testing, stimulated groups displayed supranormal spatial memory formation, and 5xFAD mice performed worse than wild-type littermates. Behavior was tested in the Morris water maze (Morris 1984) using 3 randomized starting points each day, with a 90 s timeout. Swim times are displayed in Fig. 2B. The control group contained 12 mice, the stimulated wild-type 14 mice, the control 5xFAD 7 mice, and the stimulated 5xFAD 11 mice. Animals were genotyped after behavioral testing, but testing was not blind to stimulation condition. A 3-way ANOVA found significant effects of genotype (F(1,115) = 11.06, P = 0.0012), day (F(4,115) = 18.5, P < 0.0001), and stimulation (F(1,85) = 35.35, P < 0.0001). A post-hoc Fishers test found that the wild-type stimulated group had lower swim times on days 4 and 5 (wt t(120) = 3.21, P < 0.0086; t(120) = 2.94, P < 0.0199) than the wild-type unstimulated, and in the 5xFAD groups on days 3–5 (t(120) = 3.38, P < 0.0057; t(120) = 3.51, P < 0.0038; t(120) = 3.62, P < 0.0027). In the 5 days of water-maze testing, the wild-type littermate groups had faster swim times than the 5xFAD groups on every data point except for day 5 in the stimulated groups. On the water-maze probe test (Fig. 2E), the stimulated 5xFAD group crossed the platform location significantly more often than unstimulated groups (wt t = 4.13, P < 0.002, 5xFAD t = 4.07, P < 0.002).

Control behaviors were used to assess the likelihood of differences based in motor ability rather than memory. Visible platform swim results are presented in Fig. 2C. A 1-way ANOVA did not find significant effects of stimulation or genotype on swim times (F(40) = 0.32, P < 0.81). Time to fall on the rotorod was tested, and results are shown in Fig. 2D. A 1-way ANOVA did not find significant effects of group (F(40) = 1.40, P < 0.26).

A 10-minute Y-maze spontaneous alternation test was also performed. Two-way ANOVAs on percentage of alternation and number of arm entries did not show statistical differences based on stimulation treatment (F(1,35) = 0.25 P = 0.62 and F(1,35) = 0.55 P = 0.46, respectively). Transgenic animals made fewer arm entries (F(1,35) = 7.6, P < 0.0009). On each day of behavioral testing, animals were stimulated at least 2 h after behavior testing ended.

Nerve growth factor and related receptors

The release of acetylcholine in the cortical mantle has been reported to increase NGF signaling through its activation of nicotinic acetylcholine receptors (Jonnala et al. 2002; Hotta et al. 2009). NGF, in turn, is expected to bind its receptor, TrkA, leading to its phosphorylation, a process that completes in hours after activation (Kaplan et al. 1991). To assess whether these pathways were altered by the 5 months of stimulation, animals were left unstimulated for over 24 h prior to tissue harvest to allow short-term processing to finish. Frontal cortex was probed using antibodies for mature NGF (mNGF), precursor NGF (proNGF), the TrkA receptor, and the activated TrkA receptor as indicated by its phosphorylation (pTrkA). Sample blots are shown in Fig. 3A. All blots are included in Supplementary Data. Statistical tests pooled animals by genotype to test the null hypothesis that basal forebrain stimulation did not change the expression of the protein inspected in this pooled sample. Tissue recovery prior to protein analysis reduced the cross-genotype numbers to 3 animals per groups that we felt was insufficient to analyze genotypic differences, but enabled this combined genotype analysis of stimulation.

Fig. 3.

Concentrations of NGF pathway markers after stimulation. Frontal cortex tissue was homogenized from WT and 5xFAD mice and immunoblotted using antibodies described in Section Materials and methods section. (A) Representative immunoblots. (B–E) Optical density values normalized against GAPDH and expressed as a percentage of the control group. X symbol marks transgenic animals, and O marks wildtypes. N = 3 animals per group. (B) Immunoblot of pro-NGF at 27 kDa. No significant effects. (C) Immunoblot of mNGF at 13.5 kDa. (D) Immunoblot of NGF’s phosphorylated receptor pTrkA. (E) Immunoblot of the NGF receptor TrkA. Stimulated mice had significantly more TrkA.

TrkA was predicted to increase, as application of nicotine in different preparations leads to increases in receptor levels (Jonnala et al. 2002; Hernandez and Terry 2005). Fig. 3E shows significant increases in TrkA in both stimulated mice compared with unstimulated (t = 3.36, P = 0.007).

Fig. 3A–C shows the changes in mNGF and proNGF. Both trended higher but changes were not significant (proNGF t = 0.700, P = 0.50, mNGF t = 1.28, P = 0.23). Fig. 3D shows increases in pTrkA caused by stimulation that again trended higher in stimulated animals but were insignificant (t = 1.17, P = 0.27).

Brain-derived neurotropic factor and related receptors

The brains of Alzheimer’s patients have lower levels of TrkB than age-matched controls (Allen et al. 1999). TrkB, the BDNF receptor, responds in some ways similarly to the NGF receptor. Its activation leads to phosphorylation and endocytosis, a process that lasts for up to 6 h after activation (Soppet et al. 1991). We queried whether frontal cortex mature BDNF, precursor BDNF (proBDNF), TrkB, and phosphorylated TrkB (pTrkB) protein levels were altered by basal forebrain stimulation. Sample blots are shown in Fig. 4A. All blots are include in Supplementary Data. Statistical tests pooled animals by genotype to test the null hypothesis that basal forebrain stimulation did not change the expression of the protein inspected in this pooled sample. Tissue recovery prior to protein analysis reduced the cross-genotype numbers to 3 animals per groups, which we felt was insufficient to analyze genotypic differences, but enabled this combined genotype analysis of stimulation.

Fig. 4.

Concentrations of the BDNF pathway markers after stimulation. Frontal cortex tissues were homogenized from WT and 5xFAD mice and immunoblotted using antibodies described in Section Materials and methods section. (A) Representative immunoblots. (B–E) Optical density values normalized against GAPDH and expressed as a percentage of the control group. Data are represented as mean n = 3 animals per group. 5xFAD mice are marked with an X symbol, and wildtype marked with an O (B) proBDNF (C) mBDNF. (D) pTrkB at 140 kDa. (E) TrkB.

Stimulation caused nearly 3-fold increases in TrkB levels compared with unstimulated mice, shown in Figure 4E, and this effect was significant (t = 4.20, P = 0.002). Nonsignificant results were obtained for the precursor BDNF (proBDNF, t = 1.20, P = 0.26), mature BDNF (mBDNF, t = 0.961, P = 0.36), and the phosphorylated TrkB receptor (t = 0.367, P = 0.72) as shown in Fig. 4.

Amyloid precursor protein processing

The transgenic genotype used was the 5xFAD genotype (Oakley et al. 2006), which has 3 APP mutations and 2 presenilin mutations to result in beta-amyloid production and accumulation. To assess the impact of stimulation on the presence of APP pathway elements, we probed frontal cortex tissue for the proteins ADAM10, BACE1, and for Aβ₄₂. Again, statistical tests for ADAM10 and BACE1 pooled animals by genotype to test the null hypothesis that basal forebrain stimulation did not change the expression of the protein inspected in this pooled sample.

Western blotting for ADAM10, the alpha cleavage enzyme, shown in Fig. 5A did not find significant changes caused by stimulation (t = −0.209, P = 0.83). BACE1, the beta-amyloid cleavage enzyme was significantly lower in stimulated mice (t = 2.61, P = 0.026). The trend for BACE1 and TrkA (Fig. 3) to show related changes was investigated with a correlation, resulting in Pearson’s r = −0.70, P < 0.005. A similar but weaker correlation was found between TrkB (Fig. 4) and BACE1 (r = −0.56, P < 0.05). TrkA and TrkB were not significantly correlated (r = 0.48). TrkA and TrkB expressions are thus each significantly anticorrelated with BACE1 expression. An ELISA measured concentration of Aβ₄₂ found levels in the tissue was reduced significantly, a reduction of 43% (Fig. 5C, t = 2.50, P = 0.04.). Only 5xFAD animals were used in the examination of Aβ₄₂, as wild-type mice do not create this isoform of Aβ.

Fig. 5.

Concentrations of the APP metabolic pathway markers after stimulation. Proteins were homogenized from frontal cortex of WT and 5xFAD mice and immunoblotted using antibodies described in Materials and methods section. The upper panel represents the immunoblot of the listed biomarker protein. The lower panel represents optical density values normalized against GAPDH and expressed as a percentage of the control group. Data are represented as mean per group n = 6 animals per group. 5xFAD mice are marked with an X symbol, and an O symbol marks wild-type mice. (A) Immunoblot of ADAM10 at 60 kDa. (B) Immunoblot of BACE1 at 56 kDa. (C) ELISA-measured concentration of Aβ₄₂ showed significant decrease in stimulated 5xFAD mice frontal cortex compared with unstimulated controls. N = 4–5. ELISAs were only run on 5xFAD mice.

Hippocampal markers

The hippocampi of mice were also probed (Fig. 6A). The projections from the basal forebrain densely target the cortical mantle, but only sparsely target the hippocampus, which receives analogous innervation from the medial septum, which was not stimulated. TrkA increases and BACE1 decreases occurred but were insignificant (TrkA, t = −0.69, P = 0.50. For BACE1, t = 1.41, P = 0.19). Again, TrkA and BACE1 expressions were anticorrelated significantly (Pearson’s r = −0.72, P < 0.005). Three animals in each of the 4 groups were used.

Fig. 6.

Neurotrophic and APP metabolism markers in the hippocampus after stimulation. Proteins were immunoblotted using antibodies described in Section Materials and methods section. (A) The upper panel represents the immunoblot of the listed biomarker protein. The lower panel represents optical density values normalized against GAPDH and expressed as a percentage of the control group. Data are represented as mean per group. N = 6 animals per group. 5xFAD data are marked with an X symbol, and wildtype with an O. (B) Immunoblot of NGF’s receptor TrkA and BACE1.

Discussion

The present study was conducted to assess the effects of intermittent stimulation of the basal forebrain, which contains cholinergic projection neurons, on brain pathways relevant in Alzheimer’s dementia. To accomplish this, we stimulated mice that produce a high Aβ₄₂ burden and display cognitive deficits. We used the Morris water maze to test stimulation impacts on visuospatial memory. Western blots and ELISAs allowed us to quantify the expression of neurotrophic and β-amyloidogenic markers of interest between treated and control groups. We found improved visuospatial memory, and increased expression of the BDNF and NGF receptors TrkB and TrkA, a reduction in overall Aβ₄₂ accumulation, and a significant relationship between increases in neurotrophin receptor expression and decreases in BACE1.

The cognitive enhancement findings in this study are consistent with previous NGF and water maze related studies (Janis et al. 1997; Pham et al. 1999; De Rosa et al. 2005; Conner et al. 2009). We predicted NGF would be released during stimulation based on prior work (Hotta et al. 2009). Arguably, our experiment used endogenous acetylcholine pathways (Zaborszky et al. 2012) to trigger the creation of NGF in cortical receiving zones. Although the basal forebrain projects to the cortical mantle while the water maze more classically tests hippocampal function, it should be noted that immunotoxic lesioning of cholinergic neurons in the basal forebrain severely inhibits Morris water-maze performance, while lesioning the hippocampal-projecting medial septum neurons has much less of a detrimental effect on performance (Mandel and Thal 1988; Berger-Sweeney et al. 1994). Additionally, the basal forebrain is more of a complex continuum than completely segregated neuronal populations, so a hippocampal intensive task could still be relevant to a basal forebrain stimulation (Carlsen et al. 1985; Brashear et al. 1986; Záborszky et al. 1986; Zaborszky et al. 1991). Therefore, we believe the water maze to be an appropriate behavior for testing basal forebrain-induced cognitive enhancement particularly because it contains a strong reinforcement incentive, removal from cold water. Animals were stimulated on days in which water-maze testing occurred, but this stimulation occurred 2 h after the end of testing. We cannot refute the possibility that this stimulation, as compared with the 5 month history of 1 stimulation hour per day, had a positive contribution to water-maze differences.

Comparable studies with ours are fairly limited due to the advanced age and long-term treatment periods of our mouse groups as well as water-maze variability in pool size and average latency times (Schneider et al. 2014; Gee et al. 2020). Because our testing facility water maze is intended to also work with rats, its pool surface area is almost double the size of mouse-only studies, which had the expected impact of drawing out the learning curve. To date, there are 2 similar stimulation studies worth comparing. 3xTg AD mice were stimulated in the entorhinal cortex with tungsten wire electrodes, which has been shown to increase neurogenesis in the dentate gyrus (Mann et al. 2018). Their stimulation pattern lasted for 7-h per day over the course of 25-days at a current half the amperage of this study and twice the pulse frequency. These mice demonstrated improvement in the probe test, but not in the 3-day acquisition period of their water maze (Mann et al. 2018). Aged APP/PS1 AD mice have been implanted with stainless steel electrodes in the basal forebrain and stimulating for various lengths of time and frequencies at a 10-fold stronger current than our study (Huang et al. 2019). They expectedly demonstrated that escape latencies in the water maze were best improved in the earliest implanted group at 4 months old. Because the radius of the zone activated by stimulation increases proportionally to the square root of the current (Stoney et al. 1968; McIntyre et al. 2004), the much higher 1 mA current used would triple the stimulated tissue radius at the site of implantation. In other work, neither unstimulated 5xFAD nor wild-type littermates demonstrate learning curves after the 5th day of the water maze (Schneider et al. 2014), similar to both of our non-treated mouse groups.

Our primary planned neurotrophic biomarker was TrkA. The expression of Trka, of the mNGF receptor, is reduced with losses in cognition, interacts with the amyloid precursor protein, and covaries with the integrity of the forebrain cholinergic projections (Boissiere et al. 1997; Matrone et al. 2011). We show here that our stimulation increases TrkA in mice in the cortex. The classic cholinergic subregions in the basal forebrain (Ch1–Ch4) are not entirely segregated (Zaborszky et al. 2012). For example, some medial septum efferent fibers travel to the basal forebrain (Richardson and DeLong 1988; Fujishiro et al. 2006), and sparse basal forebrain efferents project to the hippocampus. The stimulated region could also include fibers of passage traveling through or nearby the basal forebrain (Price and Amaral 1981) and on their way to the hippocampus or medial septum provided they passed within our activating function radius of roughly 300 microns. Our demonstration of changes in cholinergic release caused by stimulation may be combined with prior work showing increased NGF levels caused by stimulation are dependent on nicotinic receptor activation (Hotta et al. 2009). Analogous work using cholinesterase inhibitors to increase acetylcholine levels has been linked to neurotrophin activation (Hernandez et al. 2006; Autio et al. 2011) and chronic treatment with cholinesterase inhibitors improves cognition in mouse models of beta-amyloid pathology (Van Dam et al. 2008; Guo et al. 2015).

The magnitude of the behavioral impacts of stimulation, and TrkA protein expression upregulation, motivated our look at brain-derived neurotrophic pathway measures because it is also low in Alzheimer’s dementia post-mortem brains (Allen et al. 1999; Ferrer et al. 1999). Our finding, that TrkB is also upregulated, suggests the stimulation may use an intermediary messenger, a serine protease. This cleavage step is shared in common to convert precursor forms of NGF and BDNF into mature forms (Bruno and Cuello 2006; Gray and Ellis 2008).

The evidence for direct activation of the neurotrophin receptors TrkA and TrkB was weaker in this study than the evidence receptor levels were high at the end of study. The decision to wait over 24 h after the last stimulation session before euthanasia and tissue analysis was made specifically to avoid acute impacts of stimulation, which fade within a few hours of activation (Kaplan et al. 1991; Soppet et al. 1991). One day later, normal receptor phosphorylation and neurotrophin levels may be expected. We hypothesize that stimulation induced higher levels of mNGF and mBDNF, and led to phosphorylation of both their receptors, but those effects were largely absent by the 24 h time point. A study directly comparing these same markers within an hour of finishing a stimulation session, and 24 h later, could address this concern. Application of nicotine to PC12 cell cultures does result in upregulation of TrkA expression dependent on activation of nicotinic receptors (Jonnala et al. 2002). This nicotine-induced expression has also been demonstrated in vivo (Hernandez and Terry 2005; Formaggio et al. 2010). For these reasons, we think the changes in TrkA expression are caused by activation of the cholinergic pathways, which activate nicotinic receptors, which induce an NGF response that leads to upregulation of TrkA. Similar relations are noted with nicotine and BDNF (Machaalani and Chen 2018).

Our primary β-amyloidogenic marker was BACE1, the rate limiting step in forming Aβ₄₂ (Lahiri et al. 2014). Cellular mechanisms have been found to regulate expression and activity of BACE1. Mice with a p75^NTR −/− genotype demonstrate a half-fold reduction in both BACE1 expression as well as overall Aβ₄₂ burden that is striking given that the p75^NTR containing neurons are only the cholinergic neurons, a significant a minority of the neuropil, which otherwise has fairly ubiquitous BACE1 expression (Costantini et al. 2005). We believe the decreased BACE1 signaling induced by basal forebrain stimulation is likely due to TrkA and TrkB-mediated PI3K/PKC signaling cascades that inhibit nSMase ceramide production, a testable hypothesis. The correlations observed between TrkA and BACE1, and TrkB and BACE1, support this hypothesis. Prior work has generated support for TrkA activation leading to alterations in beta-amyloid processing (De Rosa et al. 2005; Capsoni et al. 2010).

We conclude that basal forebrain activation improves visuospatial memory, increases the expression of both the NGF receptor TrkA and the BDNF receptor TrkB, reduces expression of the beta-amyloid cleavage enzyme BACE1, and reduces Aβ₄₂ accumulation. This work strengthens and elaborates knowledge of these pathways. Basal forebrain stimulation releases acetylcholine in cortex (Kurosawa et al. 1989; Sato and Sato 1995). This release of acetylcholine causes an increase in neurotrophins dependent on nicotinic activation (Hotta et al. 2007, 2009). NGF, or nicotinic activation, results in an increase in TrkA expression (Kaplan et al. 1991;Jonnala et al. 2002 ; Hernandez and Terry 2005). Modulation of TrkA activation can alter beta-amyloid accumulation and BACE1 expression (Capsoni et al. 2000, 2010; De Rosa et al. 2005). Our study connects each of these links to show that basal forebrain stimulation increases neurotrophin receptor expression and reduces BACE1 expression and Aβ₄₂ accumulation. There was perhaps less prior evidence linking basal forebrain elicited acetylcholine and the BDNF receptor pathway (Parikh et al. 2016; Machaalani and Chen 2018; Khalifeh et al. 2020), and our work also strengthens the evidence linking neurotrophin receptor expression and suppression of beta-amyloid cleavage.

Our work demonstrates that basal forebrain activation releases acetylcholine, which promotes neurotrophin receptor expression and reduces BACE1 expression and beta-amyloid accumulation. Although the basal forebrain contains projection neurons that are not cholinergic (Semba 2000, 2004), we favor the hypothesis that these effects depend on acetylcholine, as they are consistent with pharmacologically driven increases in TrkA using cholinergic receptor nicotinic agonists (Jonnala et al. 2002; Hernandez and Terry 2005). The relationship between nicotinic receptors and cognition is well established (Rezvani and Levin 2001; Newhouse et al. 2012). Nicotinic receptors are also impacted, in a dose-dependent manner, by beta-amyloid peptides (Sadigh-Eteghad et al. 2014). Lower doses are agonistic, but higher doses antagonize. Impacts on cognition are observed with picomolar levels of beta-amyloid peptide (Puzzo et al. 2008). Impairments caused by beta amyloid in animal models are preventable by the activation of specific subtypes of nicotinic receptors e.g. α7 and α4β2 (Inestrosa et al. 2013; Sun et al. 2019). Thus, basal forebrain activation including cholinergic release suppresses accumulation of beta amyloid, while beta amyloid antagonizes nicotinic acetylcholine receptors in a dose-dependent manner. There appear to be bidirectional antagonistic links between Aβ₄₂ and cholinergic activation acting in the Alzheimer’s dementia neurodegenerative processes, which include directly modulating neurotrophin activation.

Authors’ contributions

DTB and JK and AP and SK designed all studies. JK performed all surgeries and all brain stimulation for behavioral and pharmacological experiments. JS and JK performed surgeries for GACh3.0 experiments. JK performed water maze tests under supervision from PC. ATerry designed all pharmacological dosing. YL and XYL performed and analyzed alternating Y maze tests. JK and ATripathi performed all biochemical analyses in AP lab. JS and JK performed GACh experiments in SK lab. DTB and JK drafted the manuscript with feedback from all authors.

CRediT author statement

Jacob Kumro (Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Writing—review & editing), Ashutosh Tripathi (Data curation, Formal analysis, Methodology, Supervision), Yun Lei (Data curation, Formal analysis, Investigation, Methodology), Jeremy Sword (Data curation, Formal analysis, Investigation, Methodology, Supervision), Patrick Callahan (Methodology, Supervision), Alvin Terry (Methodology, Resources, Supervision, Writing—review & editing), Xin-yun Lu (Investigation, Methodology, Supervision, Writing—review & editing), Sergei Kirov (Conceptualization, Methodology, Resources, Supervision, Writing—review & editing), Anilkumar Pillai (Conceptualization, Methodology, Resources, Supervision, Writing—review & editing), David T. Blake (Conceptualization, Data curation, Formal analysis, Funding acquisition, Methodology, Project administration, Resources, Software, Supervision, Visualization, Writing—original draft, Writing—review & editing)

Funding

National Institute of Health (NIH, RF1-AG060754 to DTB) and US National Institute of Health/National Institute of Mental Health (NIMH) (grants MH120876 and MH121959), and the Merit Review Award (BX004758) from the Department of Veterans Affairs, Veterans Health Administration, Office of Research and Development, Biomedical Laboratory Research and Development to AP, US National Institutes of Health/National Institute of Neurological Disorders and Stroke NS083858 to SAK. The contents do not represent the views of the Department of Veterans Affairs or the United States Government. AP acknowledges the funding support from Louis A Faillace Endowed Chair in Psychiatry.

Data availability

All original Western blots are included in the Supplement. Behavioral data, and data for Figure 1, are available upon request.

Conflict of interest statement: AP received pre-clinical research support from ACADIA Pharmaceuticals.