Low-intensity transcranial ultrasound stimulation improves memory in vascular dementia by enhancing neuronal activity and promoting spine formation

Memory is closely associated with neuronal activity and dendritic spine formation. Low-intensity transcranial ultrasound stimulation (TUS) improves the memory of individuals with vascular dementia (VD). However, it is unclear whether neuronal activity and dendritic spine formation under ultrasound stimulation are involved in memory improvement in VD. In this study, we found that seven days of TUS improved memory in VD model while simultaneously increasing pyramidal neuron activity, promoting dendritic spine formation, and reducing dendritic spine elimination. These effects lasted for 7 days but disappeared on 14 d after TUS. Neuronal activity and dendritic spine formation strongly corresponded to improvements in memory behavior over time. In addition, we also found that the memory, neuronal activity and dendritic spine of VD mice cannot be restored again by TUS of 7 days after 28 d. Collectively, these findings suggest that TUS increases neuronal activity and promotes dendritic spine formation and is thus important for improving memory in patients with VD.


Introduction
Vascular dementia (VD) is a syndrome in which abnormal cerebral blood perfusion leads to memory, cognitive, and executive dysfunction, seriously affecting quality of life (O'Brien et al., 2003;O'Brien and Thomas, 2017).There are currently no available drugs specific for the prevention and treatment of VD.Recent research has shown that low-intensity transcranial ultrasound stimulation (TUS), which has the advantages of non-invasiveness, high penetration depth, and high spatial resolution (Cotero et al., 2019;Meng et al., 2021;Zhang et al., 2022), can increase the protein expression of brain-derived neurotrophic factor (Wang et al., 2022) and reduce pro-inflammatory cytokines (Huang et al., 2017) in VD model.The living environment of neurons and synapses improves learning and memory capabilities (Deng et al., 2010).Notably, the above studies did not provide real-time long-term monitoring of neuronal activity and dendritic spine formation/elimination in living samples of the VD model.Thus, they could not provide direct evidence of improvements in neuronal activity and synaptic plasticity.
Memory is closely related to neuronal activity and dendritic spine formation.Neuronal activity and formation/elimination of dendritic spines are important for learning and memory (Yang et al., 2009).For example, the increased activity of hippocampal neurons can significantly improve memory in AD mice (Hüttenrauch et al., 2016;Palop et al., 2007).Object discrimination learning induces the formation of persistent postsynaptic dendritic spines, and survival of these dendritic spines is closely related to memory performance after learning (Yang et al., 2009).We previously showed that TUS increases neuronal activity and promotes the formation of cortical dendritic spines (Zhao et al., 2023).However, it remains unclear whether and how neuronal activity and dendritic spine formation under ultrasound stimulation are involved in memory improvement in VD model.
Towards this goal, we used behavioral tests and two-photon in vivo imaging to study the memory ability of VD model under TUS, as well as the calcium activity of pyramidal neurons and the formation/elimination of dendritic spines in the barrel cortex, which strongly corresponded to behavior.

Animals and grouping
Male C57BL/6J-Tg (Thy1-GCaMP6f; Jackson Laboratory) (N = 25) and D2.Cg-Tg (Thy1-YFP) HJrs/SjJ (Jackson Laboratory) (N = 25) mice weighing 28-32 g were used.These transgenic mice were divided into five groups (sham, VD, VD+TUS (visual cortex + TUS and barrel cortex + TUS), VD+Trapezoidal pulse) with 10 mice per group (five GCaMP6f and five Thy1-YFP-H).All mice underwent cranial window surgery.In the VD, VD+TUS, and VD+Trapezoidal groups, an ischemic model was established 7 days after craniotomy.The VD+TUS, and VD+Trapezoidal pulse groups underwent TUS stimulation for 7 consecutive days after establishing the ischemia model.Ten mice per group were subjected to behavioral experiments that relied on whisker discrimination between novel objects and the T-maze.Among them, five GCaMP6f transgenic mice were used to observe Ca 2+ transient changes in layer 5 pyramidal neurons, and five Thy1-YFP-H Transgenic mice were used to observe changes in dendritic spines on layer 5 pyramidal neurons.The experimental timeline is shown in the right image of Fig. 1A.All mice were individually housed in standard temperature-and humidity-controlled mouse cages with free access to food and water and with a 12 h light/ dark cycle.All experiments were conducted during the light cycle.
All animal procedures were conducted in accordance with the guidelines of the Animal Ethics and Management Committee of the Yanshan University.All applicable international, national, and/or institutional guidelines for the care and use of animals were followed.

Time sequence
As shown in Fig. 1A, the time required to create the VD model was defined as the starting time (labeled 0 d).Skull windows were created in the sham, VD, VD+TUS, and VD+Trapezoidal pulse groups on 7 d before 0 d (marked as -7 d).We started performing ultrasound stimulation in the VD+TUS, and VD+Trapezoidal pulse groups on the 7th day after making the model (marked as 7 d) and continued until the 14th day (marked as 14 d).Memory behavior tests and two-photon imaging of pyramidal neuron activity and dendritic spines were performed in all groups at 7, 14, 21, and 28 days after establishing the VD model (7 d: start of TUS, 14 d: end of TUS, 21 d: 7 days after end of TUS, 28 d: 14 days after end of TUS).

Surgery for imaging
To facilitate the long-term and stable two-photon imaging of the mouse barrel cortex (BC), we prepared a mouse window model using a previous method (Holtmaat et al., 2009).The specific operations were as follows: The hair above the mouse skull was removed, the scalp was cut to expose the skull, and tweezers were used to clean the hair and The position of the window was determined, and a skull drill was used to grind off the skull at the window position (3.0 mm to the right of the bregma, 1.5 mm to the back, with a diameter of about 2 mm) until the periphery of the skull became obviously loose.The tip of the tweezers was inserted into the loose part, and the skull was gently lifted.
A glass piece was fitted to open the windows.After ultraviolet disinfection, the windows were wiped with alcohol before the glass piece was fitted into the window opening.After wiping, the glass piece was placed at the window opening, and a positioner was used to press the glass piece tightly at the window opening.This was performed to ensure that the glass piece would not shift during the application of glue and dental cement and that the glue and dental cement would not enter the mouse brain from the window opening.
The glass piece was fixed, and a head-fixation device (two parallel metal rods) was attached.First, glue was applied around the periphery of the glass piece to solidify it.After the glue solidified, dental cement was applied to the entire head, and two parallel metal rods were fixed.One rod was placed between the eyes of the mouse, and the other between its ears.After the dental cement had solidified, the mice were removed and returned to their cages.During the operation, we injected dexamethasone sodium phosphate (0.02 mL at 4 mg mL − 1 ; ~2 μg g − 1 dexamethasone) intramuscularly into the legs of mice to reduce the cortical stress response during the operation and prevent brain damage.Mice were anesthetized with isoflurane (4 % for induction; 1.0-1.5 % for surgery with ~0.5 L min − 1 O 2 ) throughout the surgery.Two-photon imaging was initiated 14 days after preparation to eliminate the impact of surgery on the brain.

Mouse model of vascular dementia
We used bilateral common carotid artery occlusion to generate a mouse model of VD.The mice were fasted for 8 h before surgery, and water was not allowed.After the mice were anesthetized with an intraperitoneal injection of 2 % sodium pentobarbital (40 mg/kg), they were fixed in the supine position.A midline incision was made on the neck, and the bilateral common carotid arteries were isolated.The right common carotid artery was permanently ligated with No. 6 silk suture to block blood flow.Concurrently, an arterial clamp was used to block the blood flow of the contralateral common carotid artery for 30 min, followed by reperfusion for 10 min.This step was repeated three times.

Low-intensity ultrasound stimulation protocol
A signal generator (AFG3022C, Tektronix, USA) was used to continuously emit a modulated sine wave signal.The electrical signal was amplified by a power amplifier (E&I 240L, ENI Inc., USA) and then by a focused ultrasound probe (V301-SU, focus, diameter: 25.4 mm, radius of curvature: 40 mm, Olympus, USA) for ultrasound stimulation.A cone-shaped collimator (3D printed, resin material) filled with ultrasound coupling fluid was fixed between the output end of the ultrasound probe and the head of the mouse, allowing ultrasound to penetrate the brain of the mouse.The main parameters of the output ultrasonic wave were as follows: fundamental frequency = 1 MHz, number of fundamental cycles = 50,000, pulse repetition frequency = 1 Hz, duty cycle = 5 %, and stimulation time = 50 ms.The ultrasound pressure was 0.2 MPa, and the corresponding spatial peak pulse average intensity (I sppa ) and spatial peak time average (I spta ) values were 1.33 W/cm 2 and 0.066 W/cm 2 , respectively.Stimulation was performed at the same time every day, with breaks of 5 min, for a total of three groups of 25 min.The stimulation position included BC.COMSOL Multiphysics 5.6 finite element analysis software was used to simulate the ultrasonic sound field under normal experimental conditions.The ultrasonic sound field distribution used in the experiment is shown in Fig. 1B.

Whisker-dependent behavior test
The whisker-dependent new object discrimination behavioral experiment was performed with reference to previous experimental methods (Chen et al., 2017;Leger et al., 2013).The mice were trained for 2 days before the formal experiment.During training, the mice were placed in a test box (50 cm × 50 cm × 50 cm) with a bed at the bottom of the test box.The litter was filled to block the platform at the bottom of the sign, and the mice were allowed to move freely for 10 min.There were no signs at this time.Notably, a dedicated transport rat cage was used instead of a residential rat cage when transporting the mice, and alcohol was used to disinfect the test box before training and formal experiments to allow the mice to adapt to the odor of alcohol.The formal experiment flowchart is shown in Fig. 2A.During the experiment, the mice were first placed in an alcohol-sterilized test box and allowed to explore freely for 5 min (learning phase).At this time, the sign in the test box (4 cm × 0.5 cm × 15 cm cuboid with a sandpaper; the base was a cm × 4 cm × 0.5 cm transparent acrylic cuboid).
In both A and A', the sandpaper number was 80 (the left edge of the left sign was 10 cm away from the left side of the test box, and the distance between the right edge of the right sign was 10 cm on the right side of the test box, and 22 cm apart between the two signs).The mice were then removed and placed in a special transport cage for 10 min.Finally, the mice were placed in an alcohol-sterilized test box and allowed to explore freely for 5 min (test phase).The sandpaper mesh number of sign A'' in the test box was still 220, and the sandpaper number of the novel sign was 80 (randomly choose replacement label A or A').The total time t of the mouse approaching signs A and A' in the learning phase and the time t' of the mouse approaching the novel sign in the testing phase were counted.The total time t' occupied by time t' of the mouse approaching the novel sign was calculated as the total time t of the mouse approaching signs A and A'.The percentage of t was calculated as (t' / t) * 100 %.The entire experimental process was captured using a camera, and the mouse was defined to enter the sign when it was within 4 cm of the sign.

T-maze behavior test
Before the formal T-maze experiment, all involved mice underwent water deprivation training wherein each mouse was fed 1 mL water at a fixed time every day, with no food restriction, 7 days after the cranial window was opened until the end of the experiment.The T-maze consisted of a starting arm (50 × 10 × 0.5 cm) and two target arms (30 × × 0.5 cm).The starting end of the starting arm served as the starting point for the mouse, and the target arms on both sides were selected for the mouse.During the adaptation phase (habituation), the inside of the T-maze was wiped with alcohol, and a small container with 1 mL water was placed at the end of the left and right target arms of the T-maze.The mice were placed in the maze and allowed to explore freely for 15 min.After adapting for 2 d, a formal experiment was conducted.The formal experiment was divided into two stages: a pretest (sample) and a test (test).During the pre-test, one arm of the maze was randomly selected and closed with a baffle so that the mice could only be forced to enter the open arm (with water at the end).After the mouse drank water in the open arm, it was put into the mouse cage, and the inside of the maze was quickly wiped with alcohol.The baffle of the opposite arm was removed, and the mouse was placed in the starting arm after 30 s.The starting point was defined as the beginning of the testing phase.If the mouse's limbs entered any arm, the selection was considered valid, and no further retreat was allowed.If the mouse chose the contralateral arm in the pre-test stage during the test stage, it was counted as one correct choice.Each mouse was tested seven times at intervals of 15 min (Deacon and Rawlins, 2006).The experimental schematic of the T-maze is shown in Fig. 3A.

Two-photon calcium imaging in vivo
Two-photon imaging was performed using a confocal laser microscope (LSM880; ZEISS, Germany) equipped with a Ti:Sapphire laser (MaiTai DeepSee; Spectra Physics, USA).The imaging platform was equipped with a heat-preservation system to maintain the mouse body temperature at 35-36 • C. GCaMP6f transgenic mice were used for in vivo two-photon calcium ion imaging experiments to observe the activity of pyramidal neurons in layer 5 of the BC cortex.The ZEN 2 (black edition) software was used to control the LSM880 system and capture time series pictures (× 20 water lens, 1.0 numerical aperture, × 2 zoom magnification, time series mode, resolution 512 × 512, laser wavelength 920 nm).Each frame took 460 ms, and shooting a total of 600 frames took 276 s.During the image collection process, the mice were kept awake at all times.We use the integrated software "Mesmerize" of CaImAn and Mesmerize on the VirtualBox platform to analyze the collected pictures.The analysis included rigid motion correction of the pictures, denoising, and automatic identification and processing of neurons.Finally, we obtained the ΔF/F0 curve of the target neuron.The peak amplitude of ΔF/F0 was calculated to determine the maximum value of Ca 2+ activity and the number of calcium transients per unit time (1 min) to determine the frequency of lateral nerve activity (transients/min).The sum of ΔF/ F 0 (area under the curve) was used as a comprehensive index to evaluate Ca 2+ activity.The calcium transient threshold was defined as three times the standard deviation within the first 10 % (27.6 s) of the total duration of ΔF/F 0 .

Dendritic spine imaging
The imaging platform was LSM880 system.Thy1-YFP-H transgenic mice were used in vivo two-photon dendritic spine imaging experiments to observe the dendritic spines of pyramidal neurons in layer 5 of the BC cortex.During the image collection process, the mice were kept awake at all times.The bottom of the shooting platform was equipped with a heat preservation system to maintain the body temperature at 35-36 • C and ensure that the mice were protected from hypothermia.The ZEN 2 (black edition) software was used to control the LSM880 system and collect 2D images by layer along the z-axis (× 20 water mirror, 1.0 numerical aperture, × 4 zoom magnification, Z-stack mode, layer spacing 0.62 μm, resolution 1024 × 1024, laser wavelength 900 nm).The total number of layers for imaging was not fixed.Before imaging, the blood vessel distribution on the surface layer was recorded so that the same position could be repeatedly located.MATLAB and ImageJ software version 1.52a (National Institute of Health, Bethesda, MD, USA) were used to denoise the images and count the number of dendritic spines.Multiple dendrites were selected as target dendrites, and multiple overlapping upper and lower projections of each target dendrite were taken as the final image.During data analysis, the BM3D algorithm (Lebrun, 2012) was first used to denoise the image, and the number of dendritic spines at 7 d (that is, the first time the dendritic spines were photographed) was selected as the base number.Finally, the rate of change in new spines, the rate of change in the death of original spines, and the rate of change in the total number of spines relative to the base were calculated.

Statistical analysis
All data are presented as violin plots.Statistical analyses, except for the unpaired t-test used to analyze the data in Figs.S4 and S5, were performed using conventional one-way ANOVA.Figs.S1-S3 adopt polynomial second-order (Y = A + B*X + C*X 2 ) least squares fitting, where the best fitting value (A,B,C) of the fitting curve and the value of R 2 are given in Table S4.All the statistical details of the experiments are shown in Figs.2-5.Figs.3A-C and 4A-C show images of neuronal somata or dendritic spines from individual animals.Each experiment was repeated independently and quantified in the corresponding figures.All statistical analyses were performed using GraphPad Prism version 8 (San Diego, CA, USA).P value < 0.05 was considered significant.Confidence intervals for all data are 95 %.

Experimental sequence and ultrasonic parameter characterization
To verify the modulatory effect of TUS on memory, neuronal activity, and dendritic spine formation in VD mice, we designed the experiment according to the following sequence (Fig. 1).Based on the sound field distribution and the mouse brain atlas, we adjusted the position of the ultrasound transducer such that the maximum sound intensity was located in the mouse barrel (visual) cortex.

Ultrasound stimulation improves memory behavior in VD mice
The new-object discrimination experiment evaluated the memory ability of animals by observing the time spent exploring new and old objects.It is often used to study memory mechanisms in animals (Chen et al., 2017;Leger et al., 2013).We first studied the memory abilities of the different groups of mice using a novel object discrimination experiment.Behavioral tests were conducted in each group on 7 d, 14 d, 21 d, and 28 d.The results showed that memory ability in the VD+TUS and VD groups were significantly decreased on 7 d.Compared to the sham group, the VD+TUS and VD groups explored new textures and spent less time during the learning and testing stages (Fig. 2B; F (2, 27) = 43.59,P < 0.001, ordinary one-way ANOVA).On 14 d, memory ability in the VD+TUS group was significantly better than that in the VD group.Specifically, the time spent exploring new textures was significantly higher in the VD+TUS group than in the VD group.This result indicated that 7 days of TUS improved the memory ability of VD mice (Fig. 2C; F (2, 27) = 81.76,P < 0.001, ordinary one-way ANOVA).On 21 d, we observed experimental results similar to those on 14 d.The memory ability of mice in the VD+TUS group was still significantly better compared to that of mice in the VD group (Fig. 2D; F (2, 27) = 105.2,P < 0.001, ordinary one-way ANOVA).This shows that 7 days of TUS maintained the memory ability of mice for 7 days after the end of TUS, indicating that ultrasound stimulation has a sustained effect.Next, we further evaluated the sustained effect on behavioral improvement and found that the time spent exploring new textures in the VD+TUS group was similar to that in the VD group, and both were significantly lower than that in the sham group on the 28 d (Fig. 2E; F (2, 27) = 91.67,P < 0.001, ordinary one-way ANOVA).This indicated that the persistence of the enhanced memory ability of VD mice after 7 days of ultrasound stimulation was not maintained until 14 days after the end of TUS.The above results support that after 7 days of TUS, the memory ability of VD mice was significantly improved, and this effect lasted for 7 days after the end of TUS.
At 2 h after the new object discrimination experiment, we continued to conduct the T-maze test experiments, one of the most commonly used methods for evaluating spatial memory ability in animals (Deacon and Rawlins, 2006).The results showed that the memory abilities in the VD+TUS and VD groups were significantly decreased after 7 days.Specifically, compared with the sham group, the VD+TUS and VD groups made fewer correct choices during the test phase (Fig. 2G; F (2, 27) = 5.586, P < 0.05, ordinary one-way ANOVA), close to the results of the novel object recognition experiment.On 14 d and 21 d, the memory ability of the mice in the VD+TUS group was significantly better than that of the mice in the VD group.Specifically, there was no significant difference in the number of times the mice in the VD+TUS group chose the correct arm compared with that in the sham group.The sham and VD+TUS groups were more accurate in selecting the correct arm than the VD group (Fig. 2H, I; F (2, 27) = 21.62,F (2, 27) = 18.48,P < 0.001, ordinary one-way ANOVA).On 28 d, memory ability in the VD+TUS group returned to a level similar to that in the VD group.Compared to the sham group, both groups showed significantly lower number of correct choices (Fig. 2J; F (2, 27) = 12.08, P < 0.01, P < 0.001, ordinary one-way ANOVA).These results are consistent with the results of the new object discrimination experiment, further demonstrating the effectiveness and sustainability of 7 days of ultrasound stimulation in improving memory.In summary, both memory-related behaviors indicated that the memory ability of mice declined significantly after VD.However, memory behavior was significantly improved after 7 days of TUS, and the effect lasted for 7 days after stimulation.

Ultrasound stimulation modulates neuronal activity in VD mice
Increased neuronal activity in the barrel cortex helps to improve memory performance in mice, as manifested by better information retention and processing capabilities, allowing mice to perform tasks that require temporary storage and manipulation of information.More active barrel cortex neurons may improve the ability of mice to cope with interference and distracting factors, thereby improving interference-resistant memory (Chen et al., 2017;Feng et al., 2017;Stüttgen and Schwarz, 2018;Vitali and Jabaudon, 2014).Despite the above results, we found that ultrasound stimulation significantly improved the memory behavior of VD mice, and this effect lasted until 7 days after TUS.However, it was unclear whether the improvement in memory caused by ultrasound stimulation corresponded to changes in neuronal activity.As such, two-photon fluorescence imaging was used to image layer 5 pyramidal neurons in the mouse barrel cortex 2 h after each T-maze behavioral test.Neuronal calcium signaling parameters, including firing frequency, peak value, and intensity, were used to characterize neuronal activity.Then, the integrated software "Mesmerize" of CaImAn (Giovannucci et al., 2019) and Mesmerize (Kolar et al., 2021) was used to analyze the collected images.The transient frequency of calcium ions per minute and peak amplitude of calcium ion activity (ΔF/F) and the total signal intensity of calcium activity (ΔF/F) were also obtained.Compared with the sham group, the VD and VD + TUS groups showed significantly lower calcium signal indicators above the neurons (Fig. 3D, H, L; F (2, 3168) = 8.253, F (2, 3168) = 9.139, F (2, 3168) = 5.865, P < 0.05, P < 0.01, P < 0.001, ordinary one-way ANOVA) on 7 d, meaning that cortical neuron activity was reduced after VD.Notably, this corresponded to the behavioral results described previously.Next, changes in neuronal function after TUS were analyzed.The results showed that the neuronal activity of the mice in the VD+TUS group was significantly better than that of the mice in the VD group on 14 d and 21 d.Specifically, the neuronal activity in the VD+TUS group was significantly enhanced by TUS.The three indicators of calcium ions in the neurons were significantly higher in the VD+TUS group than in the VD group.This result shows that 7 days of TUS enhanced the neuronal activity of VD mice (14 d: Fig. 3E, I, M Next, we further evaluated this sustained effect and found that neuronal activity in the VD+TUS group was similar to that in the VD group and was significantly lower than that in the sham group on 28 d (Fig. 3G, K, 3O; F (2, 3578) = 21.19,F (2, 3578) = 12.61, F (2, 3578) = 9.052, P < 0.001, ordinary one-way ANOVA).This supported that the enhanced neuronal activity after 7 days of TUS did not persist beyond 14 days.These results indicate that the neuronal activity of VD mice is significantly enhanced by 7 days of TUS and that this effect can last until 7 days after the end of TUS.Importantly, when the results of neuronal activity were matched to the above behavioral results, the changes in neuronal activity had a positive correlation with behavioral improvement (Supplementary Material Fig. S1 and Table S1).These findings provide important evidence that ultrasound stimulation improves memory behavior in VD mice by increasing neuronal activity.

Ultrasound stimulation promotes the formation of dendritic spines in the barrel cortex of VD mice
The above experiments show that TUS functionally enhances neuronal activity in VD mice and that changes in neuronal activity correlate with memory.In addition, we know that the dendritic spine structure of neurons plays a key role in information processing and storage.The formation or elimination of dendritic spines in the same region affects neuronal activity (Xu et al., 2023;Zhou et al., 2020).Increasing the number of dendritic spines may make it easier for neurons to establish new synaptic connections, thereby improving their ability to encode and store information and memory (Albarran et al., 2021;Stepanyants et al., 2002).However, it remains unclear whether improvements in memory following ultrasound stimulation correspond to changes in dendritic spine formation or elimination.We used Thy1-YFP-H mice in neuronal dendritic spine experiments to observe changes in dendritic spines in layer 5 of the mouse barrel cortex, using two-photon imaging in vivo (Fig. 4A-C).The apical and basal dendritic trees of pyramidal cells have different experience-related plasticity (Kolb and Teskey, 2012).We referred to previous literature (Yang et al., 2014(Yang et al., , 2009;;Zhou et al., 2020) and analyzed the apical dendrites.We monitored a total of 224 fragments.We judged whether they came from the same cell based on the direction of the dendritic spines to ensure that these fragments came from different cells.The number of dendritic spines on 7 d was used as the baseline number to determine the relative changes in the number of dendritic spines in the different groups on days 14 d, 21 d and 28 d.The results showed that the number of neuronal dendritic spines in the VD+TUS group was significantly higher compared with that in the VD group on 14 d and 21 d.This ratio was also significantly higher than that in the VD group (14 d: Fig. 4D, F (2, 12) = 149.0,P < 0.001; 21 d: Fig. 4E, F (2, 12) = 79.13,P < 0.001, ordinary one-way ANOVA).The formation rate of neuronal dendritic spines was significantly higher in the VD+TUS group than in the VD group, whereas the elimination rate of dendritic spines was significantly lower than that in the VD group on 14 d and 21 d (14 d: 4 G, 4 J; F (2, 12) = 36.72,F (2, 12) = 60.08,P < 0.001; 21 d: Fig. 4H, K; F (2, 12) = 62.23,F (2, 12) = 127.7,P < 0.05, P < 0.001, ordinary one-way ANOVA).In addition, the changes in neuronal dendritic spines on 28 d in the VD+TUS group were similar to those in the VD group, and the number was significantly lower than that in the sham group (Fig. 4F, I, L; F (2, 12) = 28.08,F (2, 12) = 59.00,F (2, 12) = 0.8170, P < 0.001, ordinary one-way ANOVA).These results indicate that 7 days of TUS promoted the formation of dendritic spines and inhibited their elimination.This stimulation effect strongly corresponded to the improvement of behavior and enhancement of neuronal activity over time (Figs.S2, S3, Tables S2, and S3).Notably, these findings provide important evidence that ultrasound stimulation improves memory behavior in VD mice by promoting dendritic spine formation.

Evaluation of Re-TUS for improvement of memory in VD mice
In order to evaluate the improvement effect of applying TUS again on the memory ability of VD mice after 28 days, we continued to apply TUS for 7 days on ten VD mice (five Thy1-GCaMP6f mice, five Thy1-YFP-H mice, the ultrasound parameters were consistent with those of the previous experiments).And then the memory behavior, neuronal activity, and dendritic spine number and density were measured on days 35 and 42, respectively.In the new object discrimination experiment, we found that the time in the VD+TUS group exploring new textures on 35d and 42d was similar to that on 28d.In the T maze test, there was no significant difference of the memory ability between 35d, 42d and 28d (Fig. 5A, F (5, 54) = 6.784,P < 0.05, P < 0.01; Fig. 5B, F (5, 54) = 5.465, P < 0.05, ordinary one-way ANOVA).Next, we also found that the peak amplitude of calcium ion activity (ΔF/F) and the total signal intensity of calcium ion activity (ΔF/F) on the 35d and 42d are close to those on the 28d (Fig. 5C, F (5, 2454) =9.955, P < 0.05, P < 0.01, P < 0.001; Fig. 5D, F (5,2520) =5.193, P < 0.05, P < 0.01, P < 0.001, ordinary one-way ANOVA).Finally, we noticed that the changes in neuronal dendritic spines of mice in the VD+TUS group on the days 35d and 42d were similar to those on 28d (Fig. 5E, F (4,20) =85.96,P < 0.001; Fig. 5F, F (4,20) =8.864, P < 0.05, P < 0.01; ordinary one-way ANOVA).The above results of behavioral, neuronal activity and dendritic spine changes indicated that memory and functional neuronal status of VD mice cannot be restored again by TUS of 7 days after 28d.

Discussion
Ultrasound stimulation is valuable in the treatment and rehabilitation of neurological and psychiatric diseases owing to its advantages of high spatial resolution, high penetration depth, and non-invasiveness.Ultrasound stimulation can improve memory behavior in memory disorders, such as VD (Huang et al., 2017;Wang et al., 2022) and Alzheimer's disease (Beisteiner et al., 2020).The current study found new evidence that TUS improves memory behavior in mice with VD.Specifically, the results showed that 7 days of ultrasound stimulation significantly improved the memory behavior of VD mice and significantly enhanced neural activity and dendritic spine formation in memory-related brain areas, which had important effects on learning, memory, and neural processes.The modulation of neuronal activity and dendritic spine elimination had significant effects within 1 week.In addition, the memory, neuronal activity and dendritic spine of VD mice cannot be restored again by TUS of 7 days after 28 d.This study provides a robust theoretical basis for the use of TUS for the treatment of VD.
Physical therapy modalities for VD include transcranial magnetic stimulation (TMS) (Zhang et al., 2015), transcranial direct current stimulation (tDCS) (Guo et al., 2020), and optogenetic stimulation (Zhou et al., 2022).However, owing to the limitations of their physical characteristics, TMS and tDCS cannot modulate deep brain tissue.Although optogenetic stimulation can be used to implant optical fibers into any brain area for neuromodulation, the physical modulation principle currently limits its application in patients.Compared to TMS and tDCS, ultrasound stimulation has a higher stimulation depth and can stimulate deep brain areas (e.g., human thalamus) (Monti et al., 2016).Unlike optogenetic stimulation, ultrasound can noninvasively penetrate the skull to accurately stimulate the target area (Bystritsky et al., 2011).Our study demonstrated that ultrasound stimulation can improve memory in VD mice.The inherent physical advantages of ultrasound, along with our experimental results, prove that it has great potential for clinical rehabilitation and treatment of VD patients.
The current study showed that 7 days of TUS can improve memory in VD mice while simultaneously increasing pyramidal neuronal activity, promoting dendritic spine formation, and reducing dendritic spine elimination.However, the causal relationship among behavior, neuronal activity, and dendritic spine formation and elimination remains unclear.Previous studies have shown that increased cortical neuronal activity improves memory storage and information retention in mice (Kamigaki and Dan, 2017;Li et al., 1999).In addition, the formation and disappearance of dendritic spines affect neuronal activity in the same area (Xu et al., 2023;Zhou et al., 2020).Combining our results with those of previous reports, we speculate that ultrasound stimulation may further modulate neuronal activity by promoting the formation of dendritic spines, ultimately improving the memory behavior of VD mice.In addition to the above speculations, we cannot rule out other possible causal links, such as dynamic changes in neuronal activity and dendritic spine structure as a result of behavioral changes.In future studies, we will focus on their causal relationship by inhibiting neuronal activity and dendritic spine formation.
We studied the sustained effects of ultrasound stimulation and found that 7 days of ultrasound stimulation improved VD memory behavior for 7 days after stimulation; however, it returned to the pre-stimulation level 14 days after the end of TUS.We noticed that more new dendritic spines were formed and fewer dendritic spines disappeared in the VD+TUS group on days 14 and 21, which may correspond to an improvement in memory behavior.On 28 d (14 days after the end of TUS), most of the newly formed dendritic spines after ultrasound stimulation were eliminated; therefore, the memory behavior of the VD mice returned to the level before TUS.Previous research mentioned that new spine activity is more task-specific, while original/existing spines are more active in the same and different tasks (Qiao et al., 2022).This indirectly shows that new spine is more related to related tasks, while original/existing spines are involved in more tasks (Qiao et al., 2022).The theory of synaptic homeostasis hypothesis such as Giulio Tononi also believes that changes in synaptic strength are the main regulatory mechanism of learning and memory (Bushey et al., 2011;Tononi and Cirelli, 2014).To avoid saturation or disappearance of neural signals and memory traces, synaptic potentiation and depression must be balanced.It is generally believed that overall synaptic strength is modulated during learning.Due to energy and signaling requirements, learning should occur primarily through synaptic potentiation during wakefulness, resulting in an overall increase in synaptic strength (de Vivo et al., 2017).The strengthening and weakening of synapses mostly refer to changes in the size of original/existing spines.The above studies can also indirectly show that original/existing spines participate in more experiences.In addition, most spines in mice are very stable (especially the adult turnover rate is relatively low), and most spines can exist throughout the mouse's life cycle (Grutzendler et al., 2002;Pan and Gan, 2008).Therefore, it can be inferred that original/existing spines must be involved in various experiences/tasks.Therefore, the original/existing spines have existed longer than the new spine, will have experienced more changes and participated in more experiences.This affects the sustained effect of ultrasound stimulation on the memory behavior of the VD mice.Guo et al. (2018) and Sato et al. (2018) reported that ultrasound stimulation activates cortical neurons via non-specific auditory responses.However, these results contradict those of Mohammadjavadi et al. (2019).In our study, ultrasound stimulation enhanced the activity of cortical pyramidal neurons.To evaluate whether the modulation was affected by auditory effects, ten VD mice were stimulated with ultrasound of specific waveforms (trapezoidal pulses) based on a previous study (Mohammadjavadi et al., 2019).The results (Fig. S4) were similar for the rectangular and trapezoidal pulses.These results indicated that the modulatory effect of TUS in VD mice was not due to auditory effects.
In our previous study (Zhao et al., 2023), we found that TUS can significantly improve the learning and memory ability of healthy mice, increase neuronal firing activity and promote the growth rate of dendritic spines in the barrel cortex.It provided a potential research basis for the application of TUS in the treatment of learning and memory-related diseases.Based on previous study (Zhao et al., 2023), we carried out further research on the disease model in this work.We used ultrasound to stimulate VD mice and found that TUS can improve the memory behavior of VD mice and simultaneously increase neuronal activity and promote the growth of dendritic spines, and these changes strictly correspond to the improvement of behavior in time.We further explore and understand the relationship between ultrasound stimulation improving memory behavior, modulating neuronal activity and dendritic spine structure in VD, especially the relationship between these modulation effects and memory formation and maintenance in VD mice.It proves that TUS can have an effective modulation effect on VD, and provides a research basis for TUS on the clinical treatment of VD.Ultrasound stimulation has the potential to be used in the clinical treatment of VD.In addition, we also found that the memory and functional neuronal status of VD mice cannot be restored again by TUS of 7 days after 28d, which indicates the importance of timely TUS intervention.In summary, this work is a more in-depth and detailed study to our previous research (Zhao et al., 2023).
Some limitations in our study: (1) Due to limitations of experimental conditions, we were unable to use two-photon fluorescence imaging to observe neuronal activity in the area below the barrel cortex.In order to prove whether TUS modulates neural activity below the barrel cortex, we performed experiments under the barrel cortex.A microwire electrode (single channel, tip electrode wire diameter 35 μm, length 3 mm) was implanted in the brain tissue under the barrel cortex (anteroposterior (AP) = 1.6 mm, mediolateral (ML) = 2.8 mm, dorsoventral (DV) = − 1.1 mm).The local field potential was recorded before and after TUS.Fig. S5A (4-100 Hz, 1-4 Hz is removed to eliminate the impact of vibration on the local field potential (Wang et al., 2020)) showed that the amplitude of the local field potential increases after TUS.The results also showed that the absolute power intensity of the local field potential at 1 s after TUS is significantly higher than that at 1 s before TUS.(Fig. S5B, n = 5, Mean±SEM, ***P < 0.001, unpaired t-test).It means that TUS can induce neural activity in the brain tissue under barrel cortex.In our experiment, we combined the ultrasound sound field distribution map and the mouse brain atlas to adjust the position of the ultrasound transducer to ensure that the maximum ultrasound intensity acts on the barrel cortex and minimize the impact on other brain areas.(2) Our existing two-photon fluorescence imaging equipment cannot examine the structure and density of dendritic spines below the barrel cortex.Because we observed that ultrasound stimulation elicited neuronal activity in the brain tissue under barrel cortex, we hypothesized that the structure and density of dendritic spines may also be altered.The results are shown in Fig. S5.(3) In addition to the barrel cortex, we tried to stimulate the cortex adjacent to the barrel cortex to evaluate the effects on the memory behavior of VD mice as well as the neural activity and synaptic plasticity of the barrel cortex.We performed TUS on the visual cortex of ten VD mice (five Thy1-GCaMP6f mice, five Thy1-YFP-H mice) with the same parameters to experiment in the barrel cortex.We found that 7 days of TUS on the visual cortex did not affect the T maze and object discrimination behavior, nor did not affect neuronal activity and dendritic morphology in the barrel cortex.The results are shown in Fig. S6.
In conclusion, 7 days of TUS improved memory in VD mice by enhancing neuronal activity and promoting spine formation.Ultrasound stimulation has the potential to be used in the treatment of VD patients in clinic.

Fig. 1 .
Fig. 1.Ultrasound stimulation and VD model characterization.(A) Development of VD model, ultrasonic stimulation, behavioral testing, and two-photon imaging timing.(B) Ultrasonic sound field distribution diagram simulated by finite element analysis software and normalized profile at the white dotted line.The diameter of the sound spot in this plane is ~2 mm (Y-X plane) and ~4 mm (Y-Z plane).The sound field simulation diagram in the Y-X plane is shown in the left.The diameter of the sound spot in this plane is approximately 2 mm.The sound field simulation diagram in the Y-Z plane is shown in the right.The diameter of the sound spot in this plane is approximately 4 mm.

Fig. 2 .
Fig. 2. Ultrasound stimulation improves memory behavior in VD mice.(A) Schematic diagram of the experimental process of novel object discrimination.(B-E) Inter-group comparison of the preference for new textures on 7 d, 14 d, 21 d and 28 d in the sham, VD, and VD+TUS groups.(F) Schematic diagram of the T-maze experimental process.(G-J) Inter-group comparison of the number of times the mice chose the correct arm on 7 d, 14 d, 21 d, and 28 d among the sham, VD, and VD+TUS groups.N = 10 per group, *P < 0.05, **P < 0.01, and ***P < 0.001.Conventional one-way analysis of variance (ANOVA) followed by Tukey's post hoc test is used.

Fig. 3 .
Fig. 3. Ultrasound stimulation increases neuronal activity of VD mice.(A-C) Two-photon imaging of pyramidal neurons in layer 5 of the barrel cortex of a typical mouse in the sham group, VD group, and VD+TUS group and Ca 2+ fluorescence change curves of five randomly selected neurons in the (A) sham group, (B) VD group, and (C) VD+TUS group.(D-G) Calcium firing frequency of layer 5 pyramidal neurons in the barrel cortex in the sham group, VD group, and VD+TUS group at 7 d, 14 d, 21 d, and 28 d.(H-K) Peak amplitude of calcium ion activity (ΔF/F) of layer 5 pyramidal neurons in the barrel cortex of mice in the sham group, VD group, and VD+TUS group at 7 d, 14 d, 21 d, and 28 d.(L-O) Total signal intensity of calcium ion activity (ΔF/F) of layer 5 pyramidal neurons in the barrel cortex of mice in the sham group, VD group, and VD+TUS group at 7 d, 14 d, 21 d, and 28 d N = 5 per group.*P < 0.05, **P < 0.01, and ***P < 0.001, using conventional one-way analysis of variance (ANOVA) followed by Tukey's post hoc test.

Fig. 4 .
Fig. 4. Ultrasound stimulation promotes the formation of barrel cortical dendritic spines in VD mice.(A-C) Repeated two-photon imaging of dendritic spines in layer 5 of the barrel cortex of a typical mouse on 7 d, 14 d, 21 d, and 28 d in the sham group, VD group, and VD+TUS group.Pink and blue arrows indicate eliminated and formed dendritic spines, respectively.(D-F) Change rate of the total number of dendritic spines in layer 5 of the barrel cortex of mice in the sham group, VD group, and VD+TUS group on 14 d, 21 d, and 28 d.(G-I) Elimination rate of dendritic spines in layer 5 of the barrel cortex of mice in the Sham group, VD group, and VD+TUS group on 14 d, 21 d, and 28 d.(J-L) The formation rate of layer 5 dendritic spines in the barrel cortex on 14 d, 21 d, and 28 d in the sham group, VD group, and VD+TUS group.N = 5 in each group.*P < 0.05, **P < 0.01, and ***P < 0.001, using conventional one-way analysis of variance (ANOVA) followed by Tukey's post hoc test.

Fig. 5 .
Fig. 5. Evaluation of Re-TUS for improvement of memory in VD mice.(A) Intra-group comparison of VD+TUS group preference for new textures at 7 d, 14 d, 21 d, 28 d, 35 d and 42 d.(B) Intra-group comparison of VD+TUS group of the number of times the mice chose the correct arm at 7 d, 14 d, 21 d, 28 d, 35 d and 42 d.N = 10 per group, *P < 0.05, **P < 0.01, and ***P < 0.001.Conventional one-way analysis of variance (ANOVA) followed by Tukey's post hoc test is used.(C) Peak amplitude of calcium ion activity (ΔF/F) of layer 5 pyramidal neurons in the barrel cortex of mice in the VD+TUS group at 7 d, 14 d, 21 d, 28 d, 35 d and 42 d.(D) Total signal intensity of calcium ion activity (ΔF/F) of layer 5 pyramidal neurons in the barrel cortex of mice in the VD+TUS group at 7 d, 14 d, 21 d, 28 d, 35 d and 42 d.(E) Change rate of the total number of dendritic spines in layer 5 of the barrel cortex of mice in the VD+TUS group on 14 d, 21 d, 28 d, 35 d and 42 d in the VD+TUS group.(F) The formation rate of layer 5 dendritic spines in the barrel cortex on 14 d, 21 d, 28 d, 35 d and 42 d in the VD+TUS group.N = 5 per group, *P < 0.05, **P < 0.01, and ***P < 0.001.Conventional one-way analysis of variance (ANOVA) followed by Tukey's post hoc test is used.