Combined sensory gamma stimulation may help the brain clear amyloid waste

All animal experiments were conducted in accordance with National Institutes of Health (NIH) guidelines and were overseen by and adherent to the rules set forth by the Massachusetts Institute of Technology Institutional Animal Care and Use Committee. All of the animal holding rooms were maintained within temperature (18–26 °C) and humidity ranges (30–70%) described in the ILAR Guide for the Care and Use of Laboratory Animals (1996). Mice were housed in groups no larger than five on a standard 12 h–12 h light–dark cycle (lights on at 07:00; all experiments were performed during the light cycle). All efforts were made to keep animal usage to a minimum, and male and female mice were used. 5XFAD (Tg 6799) breeding pairs were acquired from the Mutant Mouse Resource and Research Center (MMRRC) (Jax 034848) and crossed with C57BL/6 J mice to generate offspring for this study. For experiments involving genetic manipulations of VIP⁺ interneurons, Vip^tm1(cre)Zjh/J (Jax 010908) were crossed to 5XFAD to generate VIP-Cre 5XFAD heterozygous mice. To determine the specificity and fidelity of the Cre expression in VIP-Cre mice, we used B6.Cg-Gt(ROSA)26Sor^{tm9(CAG-tdTomato)Hze}/J (Jax 007909) to indelibly label VIP interneurons with tdTomato and to generate Ai9/VIP-Cre 5XFAD triple transgenic mice. Since circadian rhythms and brain state are known to regulate glymphatic flux and AQP4 polarization²¹, all experimental groups were evaluated at consistent levels in the circadian cycle (~2–6 h after lights on).

Multisensory stimulation was performed as described previously⁶. In brief, mice were moved from the vivarium and held in a quiet room. Following 1 h of habituation to the room, individual mice were placed in separate chambers. The chamber was illuminated by a light-emitting diode programmed to either 8 Hz (125 ms light on, 125 ms light off), 40 Hz (12.5 ms light on, 12.5 ms light off, 60 W), or 80 Hz. Speakers (AYL, AC-48073) were placed above the chambers and programmed to present a 10 kHz tone that was 1 ms in duration and delivered at 60 decibels tones at 8 Hz or tones at 40 Hz. The LED and speakers were programmed via a microcontroller (Teensy) such that the sensory input was delivered simultaneously (that is, stimulus pulses of each modality were aligned to the onset of each pulse).

Following sensory stimulation or control, mice were given a lethal dose of anaesthetic (isoflurane overdose) then transcardially perfused with PBS (pH 7.4) with heparin (10 U ml⁻¹, Sigma H3149) followed by PBS with 4% paraformaldehyde (PFA) (Electron Microscopy Sciences 15710). Whole mounts of the dural meninges were prepared as described^11,24. Following perfusion, skull caps were removed, then placed in 4% PFA at 4 °C for 12 h. The dural meninges (dura mater and arachnoid) were peeled from the skull cap under a dissecting microscope using Dumont forceps (Fine Science Tools) then placed in a 24-well plate (VWR 10861-558) with PBS for immunohistochemistry. Deep cervical lymph nodes were dissected, fixed in 4% PFA for 16 h, then gently cleaned under a dissecting microscope to gently remove non-lymph node surrounding tissue. Lymph nodes were then dehydrated in 30% sucrose until the lymph nodes sank, embedded in OCT (Tissue-Tek), then frozen at −80 °C, then cut at 40 µm in a cryostat and mounted on SuperFrost slides. Immunohistochemistry was then conducted on slide-mounted tissue sections. Lymph node sections mounted on slides and treated for immunohistochemistry (described below). Brains were kept in 4% PFA for 18–24 h, then washed in PBS, the cut using a vibratome into 40 μm thick sections. Coronal brain sections were kept in PBS at 4 °C until preparation for immunohistochemistry.

Lymph nodes, coronal brain sections, and meninges were treated for immunohistochemistry using the following protocol. First, tissue was washed with PBS for 10 min, permeabilized with 0.3% Triton X-100 in PBS for 10 min, underwent blocking (5% normal donkey serum and 0.3% Triton X-100 in PBS) for 1 h at room temperature, and immunostained with the primary antibodies in blocking solution overnight. Following three 5-min washes with blocking buffer, we added secondary antibodies in blocking buffer for 2 h at room temperature, then washed with PBS five times for 5 min each. A list of antibodies are provided in Supplementary Table 1 (Supplementary Information). On the penultimate wash we used 1:1,000 Hoechst (Thermo Fisher Scientific, H3570). Tissue was mounted on SuperFrost slides and sealed with Prolong Gold mounting medium (Thermo Fisher Scientific, P36930↗).

We used a Zeiss confocal 710, 880 or 900 for confocal microscopy. The same microscope was used for each imaging experiment, and identical imaging settings were used for all settings acquired by the blinded investigator. For quantification of amyloid in lymph node, we imaged regions of lymph node in draining regions based on CD31/LYVE1 staining and imaged at 425.10 µm² (1.204 pixels per µm), at 11 µm z-stacks at 2-µm step sizes. For quantification of amyloid in the prefrontal cortex, the region to be imaged was selected based on Hoechst reference and comparison with the mouse brain atlas, then we imaged a region of 319.45 µm² (3.2055 pixels per µm) using a 30 µm z-stack imaged at 1-µm step sizes. To ensure consistency and unbiased imaging by the blinded investigator, we used the Hoechst channel to set the upper and lower boundaries of each z-stack. Zeiss ZEN Blue (v3.3.89) (Carl Zeiss Microscopy) was used for image acquisition. For data analysis, Fiji image processing software (v1.54) (NIH) and Imaris (v9.1) (Oxford Instruments) were used.

To modulate AQP4 function in mice, we used TGN020 (TargetMol T5102), administered 30 min prior to sensory stimulation (100 mg kg⁻¹, intraperitoneal injection). We used this dose based on prior literature suggesting a modulation of CSF distribution¹³. For experiments involving VIP, we used HSDAVFTDNYTRLRKQMAVKKYLNSILN (19113, Bachem); for VIP receptor agonists, we used acetyl-(d-Phe2,Lys15,Arg16,Leu27)-VIP (1–7)-GRF (8–27) (202463-00-1, Bachem), [Lys15, Arg16, Leu27]-VIP (1–7)-GRF (8–27) (064-24, Phoenix Pharmaceuticals); for VIP receptor antagonists, we used [d-p-Cl-Phe6,Leu17]-VIP (3054, Tocris). Peptides were aliquoted and stored at −20 °C.

To selectively reduce AQP4 in astrocytes, we synthesized AAV delivering eGFP followed by miR30-based shRNA²⁶ targeting mouse Aqp4 under the astrocyte-specific GFAP promoter (AAV-EGFP-shAqp4). To broadly reduce AQP4 levels, we designed three target sequences for AQP4 knockdown. As a control, we designed AAV carrying eGFP with lacZ shRNA (AAV-EGFP-shLacZ). Oligonucleotides containing shAqp4 or shLacZ within miR30 backbone were synthesized (IDT), annealed, and cloned into pAAV.GFAP.eGFP.WPRE.hGH (Addgene plasmid #105549) using the NheI site. All constructs were assembled using standard cloning methods and confirmed by DNA sequencing. Plasmids expressing miR30-based shAqp4 or shLacZ was packaged into AAV5 (Janelia Viral Core). The sequence of oligonucleotides can be found in Supplementary Table 4.

Anaesthesia was induced using isoflurane (induction, 3%; maintenance, 1–2%), ophthalmic ointment (Puralube Vet Ointment, Dechra) was applied to the eyes to prevent corneal drying, and metacam (1 mg kg⁻¹ intraperitoneal injection) and buprenorphine (0.05 mg kg⁻¹, subcutaneous injection) were administered as analgesics. Mice were placed in a stereotactic frame (Kopf Instruments) and a heating pad was used to maintain body temperature. Scalp fur was trimmed and treated with three alternating swabs of betadine and 70% ethanol. A small circular section of skin (~1 cm in diameter) was excised using surgical scissors (Fine Science Tools). The periosteum was bluntly dissected away and bupivacaine (0.05 ml, 5 mg ml⁻¹) was topically applied as a topical analgesic. A circular titanium headplate was attached to the skull using dental cement (C&B Metabond, Parkell), centred around prefrontal cortex (1.7 mm anterior to bregma, centred over the midline). Under a continuous gentle flow of PBS (137 mM NaCl, 27 mM KCl, 10 mM phosphate buffer), a ~ 4-mm circular section of the skull, slightly larger than the window, was removed using a 0.5-mm burr (Fine Science Tools) and a high-speed hand dental drill, taking great care not to compress brain tissue or damage dural tissue. Sugi swabs (John Weiss & Son) were used to absorb trace bleeding. A 3-mm glass coverslip (Warner Instruments) was gently placed over the brain. Veterinary adhesive (Vetbond, Fisher Scientific) was used to form a seal between the coverslip and the skull. A layer of Metabond was then applied for added durability. The mouse was then placed in a cage, half-on and half-off of a 37 °C heating pad, until it regained sternal recumbency. Metacam (1 mg kg⁻¹ intraperitoneal injection) was administered as an analgesic 24 h after surgery, and as needed thereafter. Mice were allowed 3–4 weeks of recovery before imaging.

We followed previous reports in order to perform intracisterna magna cannulation⁵³. Mice were anaesthetized with isoflurane (3% induction, 1% maintenance), ophthalmic ointment (Puralube Vet Ointment, Dechra) was applied to the eyes, and the head and neck were shaved and sterilized with povidone-iodine (Dynarex) and 70% ethanol. 1 mg ml⁻¹ of bupivacaine was injected subcutaneously at the incision site and buprenorphine (0.05 mg kg⁻¹, subcutaneous injection) was administered for preemptive analgesics. The mouse was fixed in the stereotaxic frame (Knopf) by the zygomatic arch, and the head was titled to form a 120° angle with the body. The occipital crest was identified, the overlying skin (~1 cm) cut, and sterile forceps were used to pull apart the superficial connective tissue and neck muscles in an anterior-to-posterior direction to expose the cisterna magna, where the cerebellum and medulla were visible behind the translucent dural membrane. A cotton swab (Sugi) was used to dry the dural membrane and a 30 G needle prepared prior to surgery fixed with PE10 tubing (Polyethylene Tubing 0.024” OD x 0.011” ID, BD Intramedic) and filled with fresh artificial CSF (ACSF) (126 mM NaCl, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 2 mM MgSO₄, 2 mM CaCl₂, 10 mM glucose, 26 mM NaHCO₃) was carefully inserted through the dural membrane, carefully avoiding damage to the cerebellum and medulla. Trace CSF leak was dried using sterile cotton swabs (Sugi), and cyanoacrylate glue (Loctite) was used to secure the cannula into the dural membrane and glue accelerator was applied to cure the glue. The needle was then secured in place using dental cement (Parkell) and a handheld cauterizer (Fine Science Tools) was used to seal the tubing. The mouse was then placed in a cage, half-on and half-off of a 37° heating pad, until it regained sternal recumbency. Following recovery from cannulation, CSF tracer infusion and awake stimulation was conducted.

Mice were head-fixed to a custom titanium head fork using no. 0-80 screws. The head-fixed mouse was positioned over a 3D printed running wheel covered in waterproof neoprene foam. Mice quickly learned to run or quietly rest (motionlessly) while in a head-fixed position. We habituated mice to gentle handling and this head-fixed position for 3 days prior to imaging experiments to avoid motion artefacts during the experiment. Prior to imaging, and while the mouse was head-fixed, the cranial window was gently cleaned using a cotton-tipped swab and a small ~1 ml dollop of Aquasonic Clear Ultrasound Transmission Gel (Parker) was placed over the cranial window. Two-photon microscopy images were acquired using an Olympus FVMPE-RS microscope. A low magnification image was acquired to facilitate returning to the same imaging site over time, and high-resolution, high numerical aperture imaging was used to acquire experimental data.

Mice had received a cranial window (3 weeks prior) and intracisterna magna implant (~3 h prior) and habituated to awake head-fixed imaging. We prepared fluorescent CSF tracer (fluorescein-conjugated dextran, 3 kD, Invitrogen D3306), formulated to a 0.5% solution in ACSF (126 mM NaCl, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 2 mM MgSO₄, 2 mM CaCl₂, 10 mM glucose, 26 mM NaHCO₃). We infused 10 µl of tracer via a cisterna cannula into awake mice at a rate of 1 μl min⁻¹ for 10 min with a syringe pump (WPI), a rate we chose based on prior reports^11,22, sealed the tube using a handheld cauterizer (Fine Science Tools, 18010-00), and placed mice in a chamber for 1 h of noninvasive multisensory stimulation or control. Following 1 h of stimulation, mice were head-fixed and positioned under the objective. To visualize tracer movement from the cisternal compartments into the brain parenchyma, we used a Spectra-Physics InsightX3 DeepSee laser tuned to 920 nm to visualize CSF tracer (labelled by fluorescein dextran) and blood vessels (labelled via retroorbital injection of Texas Red–dextran 70 kD injected prior to the experiment). Fluorescence was collected using a 25×, 1.05 numerical aperture water immersion objective with a 2-mm working distance (Olympus), and signal was detected through gallium arsenide phosphide photomultiplier tubes using the Fluoview acquisition software (Olympus). We simultaneously acquired images in the red channel (bandpass filter 575–645 nm) to visualize vascular arbors and in the green channel (bandpass filter 495–540 nm) for CSF tracer. We imaged z-stacks using a galvano scanner (z-stacks were 200 µm from the cortical surface, imaged at 2-µm step sizes; the imaging rate was set to 2.0 µs per pixel for the 512 × 512 pixel region, covering ~509.117 µm²). Three areas were imaged per mouse. Tracer influx was quantified by a blinded investigator using ImageJ and Imaris, and an average fluorescence intensity was calculated between z-stacks and normalized to non-treated mice.

Fluorescent CSF tracer (OVA-647; 45 kDa; O34784↗, Invitrogen) was formulated to a 0.5% solution in ACSF (126 mM NaCl, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 2 mM MgSO₄, 2 mM CaCl₂, 10 mM glucose, 26 mM NaHCO₃). We infused 10 µl of tracer via a cisterna cannula into awake mice at a rate of 1 μl min⁻¹ for 10 min with a syringe pump (WPI), a rate we chose based on prior reports suggesting that this method only maintains intracranial pressure following infusion^11,22. To maintain intracranial pressure following infusion, we sealed the tube using a handheld cauterizer (Fine Science Tools, 18010-00). We then placed mice in a chamber for 1 h of noninvasive multisensory stimulation or control and 1 h of recovery. Since death is associated with the collapse of paravascular space and non-physiological influx of CSF²³, we sought to avoid this potential confound entirely, so mice were euthanized within 60 s of the end of the experiment via isoflurane overdose, decapitated, and brain was fixed overnight by immersion in 4% paraformaldehyde in PBS at 4 °C with gentle rotation. To visualize tracer movement from the cisternal compartments into the brain parenchyma, we sliced brain sections at 100 µm using a vibratome (Leica) and imaged fluorescence on a Zeiss 880 confocal microscope (425.1 µm² imaging region; 1.2044 pixels per µm). Tracer influx was quantified by a blinded investigator using ImageJ. The cerebral cortex in each slice was manually outlined, and the mean fluorescence intensity within the cortical regions of interest was measured. An average of fluorescence intensity was calculated between six slices for a single mouse, resulting in a single biological replicate. Equivalent coronal brain slices were used for all biological replicates.

To image arterial pulsation, we labelled vasculature using Texas Red–dextran 70 kD via retroorbital injection prior to the experiment. Mice previously fixed with a cranial window had been habituated to head fixation under the two-photon imaging apparatus for awake imaging. We used a Spectra-Physics InsightX3 DeepSee laser tuned to 920 nm. Fluorescence was collected using a 25×, 1.05 numerical aperture water immersion objective with a 2-mm working distance (Olympus), and signal was detected through gallium arsenide phosphide photomultiplier tubes using the Fluoview acquisition software (Olympus). We acquired images in the red channel (bandpass filter 575–645 nm) for blood plasma. In a subset of experiments, we also acquired images in the green channel (bandpass filter 495–540 nm) for microglia, and movement in the green channel was used for motion artefact detection and were easily detected. We used a resonance scanner to acquire time series of arterial pulsatility in awake mice. A single recording was 328.90 s and covered an area of 160.7 µm² at a rate of 0.067 ms per pixel and 0.127 ms per line; in total, 5,000 frames were recorded at an imaging rate of 65.779 ms per frame. We validated the absence of motion artefact in our analysis based on the absence of vessel change in venous segments obtained in the same imaging areas as the arterial segments, as well as by using the soma of microglia in CX3CR1 5XFAD mice. To avoid subtle xy changes in motion, we used the phase correlation rigid registration method implemented in suite2p, using the microglia channel to align the vascular channel. To quantify arterial pulsatility, we used a perpendicular segment of the artery binarized using ImageJ, and the diameter segment was quantified using Python: first, a savgol filter (window size 7, polynomial order 5) was applied to the vasomotion trace, and peaks were identified using find_peaks.

To image ISF efflux by laser ablation, we recorded vascular segments spanning an area of 169.706 µm². For baseline imaging, we imaged at a rate of 65.779 ms per frame, 0.067 ms per pixel, 0.127 ms per line for 5,000 frames at a rate of 65.779 ms per frame. We imaged vascular beds using Spectra-Physics InsightX3 DeepSee laser tuned to 920 nm (IR laser power set at 2.22 W, and imaged using ~3.5–4.5% transmissivity). To induce ablation, we used a second two-photon laser (Mai Tai DeeSee) laser tuned to 800 nm (IR laser power at 2.79 W and transmissivity at 20–30%). Next, we induced an ellipsis region of interest for stimulation, drawn along a vascular segment approximately 3 µm in diameter. We induced stimulation using the following settings: 80 μs per pixel, 3.20 μs per line, for a total of 100 ms. Following successful ablation, a bolus of dextran was removed, and we used the InsightX3 to continue imaging to monitor the efflux and diffusion of the extravasted dextran (imaging for 328.90 s, covering an area of 160.7 µm² at 65.779 ms per frame for 5,000 frames). In pilot experiments to validate the reperfusion of blood vessels following focal ablation, we used line scans of blood vessels perfused and volumetric scans of the surrounding vascular area, using single line scans in the central lumen of along 15 µm for a capillary segment. Space-time scans were acquired using one-way galvano scanning, and the line speed was 1.989 μm per pixel for 5.7 s (5,000 frames). We performed this assay in three areas per mouse following gamma stimulation, and quantified the rate of efflux by quantifying the ratio of the extravasted dextran signal intensity at the peak of the extravasation and the end of the diffusion period, using identical distances between vascular segments between both treatment groups.

Electroencephalogram (EEG) and electromyography (EMG) implants were installed in 6-month-old 5XFAD mice under isoflurane anaesthesia as described⁵⁴. For analysis of sleep architecture based on EEG and EMG recordings, all mice were included. All mice implanted for electrophysiological recordings were housed individually in open cages before surgery and in individually ventilated cages during a recovery period of about 1 week after surgery. For sleep recordings, mice were transferred to separate custom-made Plexiglas cages (20.3 × 32 × 35 cm), which were placed in sound-attenuated and light-controlled Faraday chambers (Campden Instruments), with each chamber fitting two cages. Mice were allowed free access to food pellets and water at all times and underwent daily health inspection. After an acclimatization period of at least 3 days, during which mice were habituated to the tethered recording conditions, a period of continuous recording starting at light onset was performed on a designated baseline day. On the subsequent day, all mice received either no stimulation, 40 Hz noninvasive multisensory stimulation, or 8 Hz noninvasive multisensory stimulation conditions (see ‘Noninvasive multisensory stimulation’) and were recorded for the entire 1-h stimulation period and the entire 1 h of post-stimulation. Recordings between groups were conducted at the same time of day because circadian rhythms affect glymphatic clearance²¹. EMG and EEG data were acquired using Synapse (Tucker–Davis Technologies) and continuously recorded, filtered between 0.1 and 100 Hz, and stored at a sampling rate of 305 Hz. EEG and EMG signals were resampled at a sampling rate of 256 Hz using custom code in MATLAB (MathWorks, v2017a). Sirenia Sleep Pro (v2.2.1, Pinacle Technology) was used for sleep scoring. EEG and EMG recordings were partitioned into epochs of 4 s. Vigilance states were assigned manually to each recording epoch based on visual inspection of the frontal and occipital EEG derivations in conjunction with the EMG. Epochs with recording artefacts due to gross movements, chewing or external electrostatic noise were assigned to the respective vigilance state but not included in the electrophysiological analysis. Overall, 18.8% ± 3.5% of wake, 0.7% ± 0.4% of NREM and 0.9% ± 0.4% of REM epochs contained artefactual EEG signals across all mice included in the EEG spectral analysis, with no significant difference between stimulation conditions. EEG and LFP power spectra were computed using a fast Fourier transform routine (Hanning window) with a 0.25-Hz resolution.

The novel object recognition task consisted of a habituation phase followed by training and testing, as used in our lab previously⁶. Mice were habituated in an open field testing box for 10 min on 3 consecutive days. On the fourth day, 2 identical wooden blocks (Premium wooden building blocks set, Cubbie Lee) were placed in the chamber, and mice were allowed to explore the objects for 10 min, then the mice were returned to their home cage. Twenty-four hours later in the test phase, one of the wooden blocks was switched to a novel wooden block with a different shape, and the time spent exploring the familiar and new objects was measured for 10 min. Discrimination index was calculated as time spent to explore the new object divided by the sum of time spent to explore both old and new object by a recognition index. EthoVision (XT 14) (Noldus) was used for behaviour tracking.

Forty-micrometre coronal brain sections fixed in 4% PFA were expanded according to protein expansion protocols. In brief, after immunolabelling with anti-AQP4 and anti-eNOS, samples were treated with AcX overnight, gelled for 2 h at 37 °C, and digested with proteinase K overnight. After expansion, samples were imaged using a glass bottom plate (Cellvis, P06-1.5H-N) and imaged using an inverted Zeiss LSM 710 confocal microscope.

Perfused brains were in 4% PFA in PBS and post fixed in 4% PFA in PB overnight at 4 °C. Sections were then washed in 0.02 M glycine for 15 min. Brains were cut at 40 µm using a vibratome, then permeabilized in 0.1% Triton X-100, blocked in 1% BSA, and incubated with rabbit-anti-AQP4 overnight at 4 °C. Preparation was completed at the Harvard Electron Microscopy Core. For epon embedding, 0.5% osmium was added for 30 min, washed in water, then dehydrated using ethanol. Propyleneoxide was used and infiltrated in propyleneoxide and TAAB Epon overnight. Sections were flat embedded between two sheets of Aclar in fresh TAAB Epon, then polymerized at 60 °C for 48 h. Ultrathin sections (~60 nm) were cut on a Reichert Ultracut-S microtome, picked up on to copper grids stained with 0.2% lead citrate and examined in a JEOL 1200EX Transmission electron microscope. Images were recorded with an AMT 2k CCD camera.

We used established methods to quantify AQP4 polarization^12,21. We found that AQP4 labelled astrocytic endfeet that ensheathed blood vessels as well as surrounding parenchyma. AQP4 segments were selected on confocal z-stack projections, then marked cross-sectionally using the line plot tool in ImageJ to include AQP4 signal from vascular endfeet and from the surrounding parenchyma. The ratio of AQP4 signal from endfeet to parenchyma fluorescence intensity ratio was used as a measure of AQP4 polarization.

The protocol for the isolation of nuclei from frozen post-mortem brain tissue was adapted from a previous study⁵⁵. All procedures were carried out on ice. Following 1 h of gamma stimulation or control and 1 h of rest, cortices were dissected and snap frozen in liquid nitrogen and stored at −80 °C. Then, 3 mouse cortices were pooled per sample (4 samples per condition) and homogenized in 1 ml homogenization buffer (320 mM sucrose, 5 mM CaCl₂, 3 mM Mg(CH₃COO)₂, 10 mM Tris HCl pH 7.8, 0.1 mM EDTA pH 8.0, 0.1% IGEPAL CA-630, 1 mM β-mercaptoethanol, and 0.4 U µl⁻¹ recombinant RNase inhibitor (Clontech)) using a Wheaton Dounce tissue grinder (15 strokes with the tight pestle). The homogenized tissue was filtered through a 40-μm cell strainer, mixed with an equal volume of working solution (50% OptiPrep density gradient medium (Sigma-Aldrich), 5 mM CaCl₂, 3 mM Mg(CH₃COO)₂, 10 mM Tris HCl pH 7.8, 0.1 mM EDTA pH 8.0, and 1 mM β-mercaptoethanol) and loaded on top of an OptiPrep density gradient (29% OptiPrep solution (29% OptiPrep density gradient medium,134 mM sucrose, 5 mM CaCl₂, 3 mM Mg(CH₃COO)₂, 10 mM Tris HCl pH 7.8, 0.1 mM EDTA pH 8.0, 1 mM β-mercaptoethanol, 0.04% IGEPAL CA-630, and 0.17 U µl⁻¹ recombinant RNase inhibitor)) on top of 35% OptiPrep solution (35% OptiPrep density gradient medium, 96 mM sucrose, 5 mM CaCl₂, 3 mM Mg(CH₃COO)₂, 10 mM Tris HCl pH 7.8, 0.1 mM EDTA pH 8.0, 1 mM β-mercaptoethanol, 0.03% IGEPAL CA-630, and 0.12 U µl⁻¹ recombinant RNase inhibitor). The nuclei were separated by ultracentrifugation using an SW32 rotor (5 min, 10,000g, 4 °C). Nuclei were collected from the 29%–35% interphase, washed with PBS containing 0.04% BSA, centrifuged at 300g for 3 min (4 °C) and washed with 1 ml of PBS containing 1% BSA. The nuclei were counted and diluted to a concentration of 1,000 nuclei per μl in PBS containing 1% BSA. Libraries were prepared using the Chromium Single Cell 3′ Reagent Kits v.3.1 (Dual Index) according to the manufacturer’s protocol (10X Genomics). The generated scRNA-seq libraries were sequenced using NextSeq 500/550 High Output (150 cycles).

Raw reads were aligned to the mouse genome and the gene counts were estimated by CellRanger software (v3.0) (10X Genomics)⁵⁶. Seurat (v4.0.3) was used for downstream analysis⁵⁷. Cells with more than 500 protein-coding genes with detected unique molecular identifiers from protein-coding genes were selected for further analysis. We also use the ratio of mitochondrial genes to measure the quality of cells (cells with higher than 5% were removed). We used DoubletFinder to remove the potential doublets from snRNA-seq data. The top 2,000 highly variable genes were used for principal component analysis. The first 30 principal components were used for non-linear dimensionality reduction (UMAP) for visualization. FindMarkers function in Seurat was used to identify marker genes for each cluster and each cell type, and DEGs between mice receiving gamma stimulation or no stimulation control. For DEG analysis, the cut-off used in the function FindMarkers in Seurat was: min.pct: 0.25, only test genes that are detected in a minimum fraction of min.pct cells in either of the two populations; logfc.threshold: 0.25. Enrichr was used to perform the Gene Ontology enrichment analysis⁵⁸ with P value < 0.05 as a cut-off. Negative log₁₀-transformed P value was used for visualization by heat map with the selected representative terms based on the diverse functional categories. A list of DEGs are available in Supplementary Table 3 (Supplementary Information)

Following 1 h of gamma stimulation or control and 1 h of rest, cortices were dissected and snap frozen in liquid nitrogen and stored at −80 °C. Total RNA was extracted using TRIzol (Invitrogen) according to the manufacturer’s instructions. Reverse transcription of total RNA was carried out using RNA to cDNA EcoDry Premix (Clontech) according to the manufacturer’s protocol. qPCR was performed using a Bio-Rad CFX-96 quantitative thermocycler and SsoFast EvaGreen Supermix (Bio-Rad). Relative changes in gene expression were determined using the 2^−ΔΔCt method. Primer sequences used for qPCR can be found in Supplementary Table 2.

We used RNAscope for fluorescence in situ hybridization following the manufacturer’s protocol. The probes we used are listed in the appropriate figure legends. Tissue was prepared as in the section above describing tissue preparation with the following deviation. Following overnight fixation at 4 °C in 4% PFA in PBS, brains were cryopreserved using 30% sucrose and cut at 40 µm using a cryostat (Leica). Coronal brain sections were preserved at −80 °C until the RNAscope experiment was conducted.

We used a sequence analogous to another G-protein-coupled-receptor-based sensor⁴¹. We replaced the third intracellular loop of the VPAC1 module with a cpGFP module from the genetically encoded calcium indicator GCaMP6 using linker sequences (LSSLI-cpGFP-NHDQL). The linker sequences to the VIP sensor were designed using SnapGene. To generate AAV, we used Janelia Virus Core. For imaging VIP sensor in HeLa cells and mouse neuronal culture, we used wide-field fluorescence imaging using epifluorescence inverted microscope (Eclipse Ti-E, Nikon) equipped with a Photometrics QuantEM 512SC camera and a 75 W Nikon xenon lamp or a Zyla5.5 sCMOS camera (Andor) and a SPECTRA X light engine (Lumencor). NIS-Elements Advanced Research (Nikon) was used for automated microscope and camera control. Cells were imaged with a 60× NA1.49 oil or 20× NA0.75 air objective lenses (Nikon) at room temperature. For dual-colour imaging with miRFP, NIR (650/60 nm excitation and 720/50 nm emission) and green (490/15 nm excitation and 525/50 nm emission) filter sets were rotated into the emission light path. The GRABVIP1.0 sensor was provided by Y. Li. HEK293T cells (Invitrogen) cultured in Dulbecco’s modified Eagle’s medium (Gibco) with 10% fetal bovine serum (FBS, YEASEN Biotech) were seeded on 15 mm cover glasses (Wuxi NEST Biotech) coated with Matrigel (Millipore) and incubated at 37 °C with 5% CO₂ for 24 h before transfection. Cells were transfected with liposomal methods according to the manufacturer’s protocol (Hieff Trans, YEASEN Biotech). HEK293T cells were imaged 24 h post-transfection by an inverted wide-field Nikon Eclipse Ti2 microscope equipped with a SPECTRA III light engine (Lumencor) and an Orca Flash4.0v3 camera (Hamamatsu), controlled by NIS-Elements AR software and using a 20× 0.75 NA objective lens. Cells were imaged in the Tyrode buffer (150 mM NaCl, 4 mM KCl, 2 mM MgCl₂, 2 mM CaCl₂, 10 mM glucose and 10 mM HEPES at pH 7.35). The stock solutions of neuropeptides including CCK-4s (lot no. ab141328, Abcam), SST-14 (lot no. SP-50401-1, Alpha Diagnostic), SST-28 (lot no. SP-52221-1, Alpha Diagnostic), NPY (lot no. ab120208-500 µg, Abcam), PACAP (lot no. HY-P0176A, MedChemExpress), VIP (lot no. B6079-1, Tocris) were dissolved in water, except for CCK-8s (lot no. ab120208-1 mg, Abcam) dissolving in 0.1% NH₄OH. The working concentration of corresponding neuropeptides was 1 µM in Tyrode buffer. These neuropeptides were administrated to transfected cells via manual addition or replacing the medium with the diluted buffer using custom build perfusion system. Hippocampal neurons were prepared from postnatal day 0–1 C57BL/6 J mouse pups as described. In brief, the hippocampi were dissected in HBSS and digested with 0.25% Trypsin (Yeason) at 37 °C for 12 min. After digestion, the hippocampi were washed three times with plating culture medium (90% advanced MEM + 10% FBS) and then aspirated to dissociate the neurons. The dissociated neurons were plated at a density of 80,000 per 12-mm glass coverslip coated with Matrigel (Corning 356234) in 24-well plate. The next day, the culture medium was half replaced with NeuroBasal Medium supplemented with 1% GlutaMAX and 2% B27. AraC (0.002 mM, Sigma) was added when glia density reached 50–70% confluence. At DIV5-6, neurons were transfected with pAAV-Syn-GRABVIP1.0 or plasmid (1 µg per well) using a commercially available calcium phosphate transfection kit (Life Technologies). At DIV 12–15, fluorescence imaging was performed on an inverted wide-field Nikon Eclipse Ti2 microscope equipped with a SPECTRA III light engine and Orca Flash4.0v3 camera (Hamamatsu), using a 20×, 0.75 NA objective lens. Neurons were incubated in the extracellular solution containing: 150 mM NaCl, 4 mM KCl, 2 mM MgCl₂, 2 mM CaCl₂, 10 mM glucose and 10 mM HEPES at pH 7.35. VIP stock solution were diluted with extracellular solution and applied manually using pipette.

Mice were anaesthetized by intraperitoneal injection with ketamine-xylazine. The virus was diluted in 100 μl sterile saline and administered in the sinus behind the eye. Following the injection, Puralube was administered and mice were kept at 37 °C until they regained sternal recumbency. Virus was allowed to express for at least 3 weeks.

Six-month-old VIP-Cre 5XFAD mice previously injected with PHPeB-AAV-Syn-DIO-hM4Di-mCherry were deeply anaesthetized with sodium pentobarbital (200 mg kg⁻¹, intraperitoneal injection) and then were decapitated. Brains were quickly removed and placed in an oxygenated ice-cold cutting solution containing (in mM): 2.5 KCl, 1.25 NaH₂PO₄•H₂O, 20 HEPES, 2 thiourea, 5 sodium ascorbate, 3 sodium pyruvate, 92 N-methyl-d-glucamine, 30 NaHCO₃, 25 d-glucose, 0.5 CaCl₂•2H₂O and 10 MgSO₄•7H₂O. Brain slices (180 μm, coronal section) were made using a Leica VT1000S vibratome (Leica Biosystems). Brain slices were incubated in oxygenated cutting solution at 34 °C for 20 min to recover. After recovery, slices were transferred into oxygenated ACSF at room temperature (24 °C) for recording. ACSF solution contains (in mM): 125 NaCl, 2.5 KCl, 1.2 NaH₂PO₄, 1.2 MgCl₂.6H₂O, 2.4 CaCl₂•2H₂O, 26 NaHCO₃ and 11 d-glucose. A single slice was transferred into a recording chamber and continually superfused with oxygenated ACSF. Cells were visualized using infrared differential interference contrast (IR-DIC) imaging on an Olympus BX-50WI microscope. Action potentials were recorded at 32 °C using the whole-cell current clamp configuration of a patch-clamp amplifier (Multiclamp 700B; Molecular Devices). Action potentials were obtained by a gap-free acquisition mode using Clampex software (Molecular Devices). Signals were filtered at 1 kHz using the amplifier’s four-pole, low-pass Bessel filter, digitized at 10 kHz with a Digidata 1550B interface (Molecular Devices) and stored on a personal computer. Pipette solution contained (in mM) 121 KCl, 4 MgCl₂•6H₂O, 11 EGTA, 1 CaCl₂•2H₂O, 10 HEPES, 0.2 GTP, and 4 ATP. CNO was applied via bath perfusion.

The in vitro blood–brain barrier (iBBB) cultures were created and maintained as described⁵⁹. The iBBB consisted of a co-culture of human astrocytes, endothelial cells, and pericytes co-encapsulated in hydrogel and cultured for two weeks prior to analysis (additional details are available in Supplementary Information) Following iBBB differentiation and culture, VIP receptor agonist was added and 24 h later, cultures were fixed using 4% PFA and imaged using immunohistochemistry using antibodies against human CD31 (also known as PECAM-1) (sheep, R&D systems, AF806) and AQP4 (rabbit, Thermo Fisher, PA5-53234).

The following software was used to collect the data in this study: Olympus Fluoview (FV31-S, 2.3.1.163) (Olympus); Zeiss ZEN Blue (v3.3.89) (Carl Zeiss Microscopy); EthoVision (XT 14) (Noldus). The following software was used to analyse the data in this study: Fiji image processing software (v1.54) (NIH); Prism (v9.2) (Graph Pad); Python (v3.9); CellRanger (v3.0) (10X Genomics); Seurat (v4.0.3); Imaris (v9.1) (Oxford Instruments).

Further information on research design is available in thelinked to this article. Nature Portfolio Reporting Summary

Nature thanks the anonymous reviewer(s) for their contribution to the peer review of this work.

Sequencing data is available at the Gene Expression Omnibus (GEO) under accession number GSE249644↗. All data necessary for the conclusions of this study are presented in the manuscript. Source data are provided with this paper.

The code used for the vasomotion analysis can be found at 10.5281/zenodo.10476268. The code used for the snRNA-seq analysis can be found at 10.5281/zenodo.10480275.