Modulation of cardio-respiratory activity in mice via transcranial focused ultrasound

Abstract

Objective

The objective of this study was to investigate the effect of FUS on autonomic nervous system activity, including heart and respiratory rates, and to separate the thermal modulation from combined thermal and mechanical FUS effects.

Methods

The thalamus and hypothalamus of wild-type mice were sonicated with a continuous-wave, 2 MHz FUS transducer at pressures of 425 and 850 kPa for 60 seconds. Cardiac and respiratory rates were monitored as signs of autonomic nervous activity. FUS-induced changes in autonomic activity were compared to FUS targeted to a spatially-distant motor region and to laser-induced heating.

Results

FUS delivered to the primary target over the thalamus and hypothalamus at 850 kPa reversibly increased the respiratory rate by 6.5 ± 3.2 breaths per minute and decreased the heart rate by 3.2 ± 1.8 beats per minute. No significant changes occurred in this region at 425 kPa or when targeting the motor regions at 850 kPa. Laser heating with the same temperature rise profile produced by 850 kPa sonication resulted in cardiorespiratory modulation similar to that of FUS.

Conclusions

FUS is capable of reversibly and non-invasively modulating cardiorespiratory activity in mice. Localized changes in temperature may constitute the main cause for this activity, though further investigation is warranted into the distinct and complementary mechanisms of mechanically- and thermally-induced FUS neuromodulation. Close monitoring of vital signs during FUS neuromodulation may be warranted to monitor systemic responses to stimulation.

Introduction

Ultrasound neuromodulation is a field of non-invasive brain stimulation in which a focused or unfocused ultrasound stimulus is transmitted through the skull to generate localized changes in neural activity. Numerous studies employing various animal models, including rodents, rabbits, sheep, swine, and non-human primates (NHPs) have demonstrated the feasibility and safety of non-invasive transcranial focused ultrasound (FUS) for neuromodulation^1–7. FUS elicits electrophysiological and behavioral responses in both anesthetized and awake animals without causing tissue damage or deficits in behavioral and motor tasks^6–12. Responses to stimulation include elicitation of action potentials, motor-evoked potentials, observable motor activity (such as limb or tail movement), suppression of somatosensory-evoked potentials and visual-evoked potentials, modulation of antisaccade latencies, and changes in fMRI blood oxygenation level-dependent (BOLD) signals both during and after sonication^{2,3,5,7,10,13–16}. The published literature demonstrates that responses to FUS neuromodulation can be produced with spatial specificity as well as dependence on stimulus intensity and pulse parameters.

Due to the nature of ultrasound absorption in biological tissues, some degree of heat production is unavoidable in any FUS neuromodulation experiment. The extent to which FUS-induced temperature is a driver or confounding factor in modulation is a topic of ongoing debate⁹. Nevertheless, the careful design of sonication parameters allows one to maintain temperature below significant values¹⁷. Numerous ultrasound neuromodulation studies have attempted to effectively isolate mechanical effects of ultrasound from thermal effects by using short pulses and low duty cycles. Physiological responses to ultrasound have been reported with corresponding peak temperature increases of under 1 °C. This suggests that ultrasound neuromodulation can be achieved without significant thermal effects. However, some regions in the brain, as well as particular types of ion channels expressed in central neurons, do exhibit temperature-dependent changes in activity in certain mammals, reptiles, and songbirds (although not necessarily within the normothermic range)^18–23. Therefore, both the mechanical and thermal effects of ultrasound could affect neuromodulation depending on the stimulus parameters. Further, some studies have reported that sufficiently large temperature changes can, in fact play, a primary role in mediating neuromodulatory effects of ultrasound in the brain^9,24.

In this study, we employed continuous-wave FUS for non-invasive neuromodulation in mice while measuring concurrent changes in heart and breathing rate. We estimated the local change in brain temperature with thermocouple measurements and acoustic simulations. In addition to FUS stimulation, we conducted targeted laser heating via optical fiber of the same targets in order to separate the thermal only (laser) from combined thermal and mechanical (FUS) effects.

Materials and Methods

Animal subjects

In accordance with the National Institutes of Health Guidelines for animal research, all animal procedures for these experiments were reviewed and approved by the Institutional Animal Care and Use Committee at Columbia University prior to all studies presented herein. We performed this study on 24 male wild-type C57BL/6 mice weighing 20–25 grams.

Animal preparation

Mice were shaved and depilated under 1.5 – 2.0% isoflurane anesthesia prior to each procedure. After animal preparation, isoflurane was removed and sodium pentobarbital (Nembutal) was injected intraperitonially (65 mg/kg) for anesthesia for the remainder of the procedure. Following injection, the mice were monitored for depth of anesthesia by pedal reflex and vital sign measurement. Once the pedal reflex was no longer present, the animal was placed in a stereotaxic frame (David Kopf Instruments, Tujunga, CA, USA) and secured with ear bars and a bite bar. Throughout all procedures, oxygen was supplied at 0.8–1.0 L/min. A graphical summary of the transducer parameters is displayed in Figure 1.

Figure 1:

Illustration of the experimental setup for transcranial FUS experiments. Left: Coupling of the FUS transducer to the mouse head. Right: Ventral view of a mouse showing placement of ECG electrodes and respiratory pressure sensor.

Ultrasound transducer

All sonications were performed using a single-element FUS transducer driven at 2 MHz (center frequency: 1.94 MHz; outer diameter: 70 mm; inner void diameter: 24 mm; focal depth: 60 mm; focal length 8.7 mm; focal width: 1 mm; Imasonic SAS, Voray-sur-l’Ognon, France). A graphical summary of the transducer parameters is displayed in Figure 2. The transducer was mounted to a 3-axis positioning system, the navigation of which was controlled programmatically in MATLAB (version R2020a, The MathWorks, Inc., Natick, MA, USA). The transducer output was driven by a function generator (33220A, Keysight Technologies Inc., Santa Rosa, CA, USA) through a 50-dB power amplifier (325LA, ENI, Inc., Rochester, NY, USA). The transducer pressure was calibrated using a hydrophone (HNP-0200, Onda Corp., Sunnyvale, CA, USA) in free water and through an ex vivo mouse skull. The latter was used to estimate the ultrasound attenuation through the in vivo mouse skull. A custom cone filled with degassed water was attached to the transducer to facilitate acoustic coupling of the transducer to the head.

Figure 2:

Graphic representation of transducer and focus dimensions with inset graphic of ultrasound waveform sequence..

Ultrasound targeting

After securing the mouse in the stereotaxic frame, we placed a custom, 3D-printed container filled with degassed water over the head of the mouse. An acoustically-transparent membrane (Tegaderm, 3M Company, St. Paul, MN, USA) over the opening of the container was coupled to the scalp by a thin layer of degassed ultrasound gel (Aquasonic Ultrasound Transmission Gel, Bio-Medical Instruments, Inc., Clinton, MI, USA). To assist in FUS targeting, a small metal cross centered over the lambda cranial landmark was located by a pulse-echo C-scan using a confocally-aligned single-element transducer (center frequency: 7.5 MHz, focal depth: 60 mm, diameter: 22.4 mm; model U8517133, Olympus NDT, Waltham, MA, USA). The cross was then removed and lambda was used as the origin of navigation during targeting.

Ultrasound neuromodulation

Two target locations were chosen for sonication in this study. Location I was located at +2 mm posterior and +2 mm lateral from lambda. When targeted over these coordinates, notable structures within the volume include the primary visual cortex (V1), hippocampus, thalamus, and hypothalamus. Although our intent was, specifically, modulation of activity in the thalamus and hypothalamus, the size of transducer focus, relative to the mouse brain, impedes adequately precise targeting in the axial direction. Location II was located at +5.2 mm posterior and +1.5 mm lateral to lambda. Location II was chosen as a spatially-distant and functionally-distinct region of the brain most significantly comprising the primary motor cortex (M1) and caudate putamen.

After positioning, the target location was sonicated for 60 seconds with 60 seconds between trials for up to 15 trials per mouse. Each experiment lasted for 15 trials or until the mouse showed signs of consciousness, such as sustained movement. A quantification of the magnitude of response versus the baseline rate immediately preceding stimulation (see Supplementary Figure S1 and S2) showed that the results were not biased by larger-magnitude responses in animals at lighter planes of anesthesia (as inferred from higher baseline cardiac and respiratory rate). Each mouse underwent one sonication session. The inset of Figure 2 contains a graphic representation of the sonication sequence. At Location I, three stimulation parameters were used: FUS at 425 kPa (derated values: 300 kPa PNP; 2.5 W/cm² I_SPPA), FUS at 850 kPa (derated values: 600 kPa PNP; 10 W/cm² I_SPPA), and laser heating at 38 mW. FUS was applied to Location II at 850 kPa.

Beam profile simulation

We performed 3D acoustic simulations in k-wave, a MATLAB toolbox for acoustic simulations in a tissue-realistic environment²⁵. Simulations in free water were used to determine the simulated transducer geometry that best matched the hydrophone-measured beam profile. Density and speed of sound for the simulated mouse head were derived from a mouse micro-CT scan with 0.08 mm isotropic resolution. Attenuation values for skull and brain were adjusted to best match the ex vivo skull attenuation measured by a hydrophone as well as temperatures measured by a thermocouple.

Pulse and respiration rate measurement

Electrocardiograms (ECG) and respiration rates were recorded using a Biopac data acquisition system (MP150, Biopac Systems, Inc., Goleta, CA, USA). ECG signals were obtained using a bipolar surface electrode arrangement, with the positive and negative electrodes (placed on the left and right fore paws, respectively) measured relative to a ground electrode (placed on the left hind paw). Signals were acquired through an ECG-specific amplifier module (ECG100C, Biopac Systems, Inc., Goleta, CA, USA). Respiration rates were obtained using a differential pressure transducer (TSD160, Biopac Systems, Inc., Goleta, CA, USA) placed on the underside of the mouse and acquired through a general transducer signal amplifier (DA100C, Biopac Systems, Inc., Goleta, CA, USA). All signals were acquired at a 2 kHz sampling frequency and processed offline in MATLAB.

Pulse and respiration rate processing and analysis

All post-processing of ECG and respiration signals was performed in MATLAB. Raw ECG and respiration signals were band-pass filtered within the physiologically appropriate frequency band; the pass-band for the ECG signal was 300 – 800 Hz and for the respiratory signal was 30 – 180 Hz. Cardiac and respiration rates were then extracted from the band-passed signals by peak detection with the MATLAB function findpeaks. For the length of each signal, the instantaneous frequency corresponded to each beat-to-beat interval. The cardiac or respiratory time series was then smoothed using a moving median filter centered over a 5-second window. Individual sonication events were separated into 105-second windows (15 seconds pre-sonication, 60 seconds sonication duration, and 30 seconds post-sonication).

To avoid bias, data was processed blindly, and sections of data or entire sonications were excluded for which the extraction of cardiac or breathing rates was unsuccessful, often due to poor quality of the raw data or motion-related artifacts. This allowed for the majority of the data from a given mouse to be used even if a small number of sonications were ultimately discarded. A table of the total number of non-corrupted sonications used per mouse can be found in Supplementary Table I. For each sonication, the change in cardiac or respiration rate was calculated as the mean rate during the 15 second baseline subtracted from the mean rate during the 60 second sonication. Each sonication had its own baseline level due to variations in cardiac rate and respiration rate over time and between mice that can be related to anesthesia, individual variation, and the time course of the experiment. Subtracting from these varied baselines from the trial measurements was justified to investigate the transient changes linked to thermal stimulation apart from longitudinal changes over the course of an experiment. Finally, the average change in cardiac or respiration rate across all sonications was then calculated for each animal.

Thermocouple measurements

To measure the temperature increase caused by the application of ultrasound, we inserted a thermocouple (TC) (Models: T-29X/T-36X T-Type thermocouples, ThermoWorks, American Fork, UT, USA) into the animal’s brain near the sonication region. We drilled a small burr hole in the skull of the animal at a site contralateral to the target location. We then threaded a thin-needle T-type (FUS trials) or thin-wire T-type (laser trials) thermocouple through the burr hole using a micropositioner. The thermocouple was lowered approximately 3–4 mm below the skull. The thin-needle type probe was used during FUS trials, as the rigid needle was both easier to position and less affected by viscous heating artifacts (discussed below). The thin-wire type probe used for laser heating, on the other hand, could be threaded around and easily inserted along with optical fiber. During sonication, minute motions of the thermocouple relative to the surrounding tissue create a ”viscous heating artifact” such that the thermocouple registers an artificially high temperature. This artifact must be accounted for to obtain accurate measurements. Thermocouple measurements reported in this article have been corrected for this artifact using the method described by Kamimura et al. (2020)²⁶. In brief, the study found that the viscous heating artifact during FUS contributes to an overestimation of temperature of approximately 13% for thin-needle thermocouples (referring to the model used in the present article). The thermocouple data for FUS heating presented in Fig. 6 therefore include this 13% correction. No correction was applied to the laser heating measurement data, as no viscous heating artifact would be present.

Figure 6:

In vivo thermocouple measurements of heating due to FUS and laser pulses. Thermocouple measurements for FUS heating are corrected to account for approximately 10% overestimation of temperature due to viscous heating artifacts, as detailed in²⁶ All measurements reach 90% of their maxima within 15 s and return to baseline (within the ±0.1°C) thermocouple error range) within 60 s of stimulus offset.

Thermal simulations

To estimate the temperature throughout the brain during FUS, we performed thermal simulations of Pennes’ bioheat equation in MATLAB using the k-wave toolbox²⁷. Density values were defined in the same manner as the acoustic simulations. The values for thermal conductivity and heat capacity in the skull or brain were randomly varied within a standard deviation of the mean value for the tissue to approximate small variations in biological tissue. All properties for calculating heat diffusion were defined using values from the Foundation for Research on Information Technologies in Society (IT’IS) database of tissue properties²⁸.

To estimate heating due to laser stimulation, we used a MATLAB software package designed to simulate the propagation of light through the mouse brain during optogenetic stimulation delivered via optical fiber²⁹. The MATLAB program MonteCarloLight uses scattering and absorption characteristics of murine brain tissue, as well as the laser stimulation parameters (pulse duration, output power, wavelength) and fiber characteristics (diameter, numerical aperture (NA)), to predict light propagation through the tissue using a Monte Carlo method. It then applies the intensity output to a second program, HeatDIffusionLight, to determine the resulting temperature field by solving Pennes’ bioheat equation.

Laser heating

Using a 50 mW, 650 nm laser (Visual Fault Locator, J-Deal TL532), coupled to a fiber optic cable, we heated the target locally, following a procedure inspired by Darrow et al.²⁴. These experiments were conducted to compare purely thermal modulation to the combined thermal and mechanical influence of FUS. Similarly to our thermocouple placement, we drilled a small burr hole in the mouse skull, through which we inserted a 200μm, 0.22 NA multi-mode fiber optic cable (ThorLabs, Inc., Newton, NJ, USA). Starting immediately below the skull, we lowered the fiber into the brain using a stereotaxic micro-positioner (David Kopf Instruments, Tujunga, CA, USA). At a depth of 4–5 mm, the laser was turned on for 60 s, followed by a cool-down period of 60 s, with 5 repetitions. A smaller number of repetitions was used for the laser heating compared to the FUS sonication due to the additional preparation time needed for the fiber optic insertion.

To determine the laser power output required to achieve a temperature increase comparable to FUS heating at 850 kPa, we directly measured the in vivo temperature increase via thermocouple over 60 s of illumination at different power levels. The laser power was adjusted by varying the pulse duty cycle of the Arduino microcontroller used to control the laser output (Arduino Uno Rev3, Arduino LLC, Boston, MA, USA). Once the temperature increase matched FUS heating, we measured the corresponding power level with a fiber optic power meter (Model PM20C, ThorLabs, Inc., Newton, NJ, USA). We then verified that the power output matched the calibrated value before and after each of the in vivo laser heating trials. The measured power output for laser heating was approximately 38 mW. Laser heating trials followed the same time course as the ultrasound trials (60 s on/60 s off) and physiological signals were recorded and processed in the same manner.

Statistical analysis

Statistical testing was performed in Prism 6 (GraphPad, San Diego, CA). For both cardiac and respiration rates, statistical testing was performed on the total average change in rate per mouse. One-way analysis of variance (ANOVA) was conducted between the stimulation groups. Multiple comparisons testing was then conducted to find differences between individual groups using Tukey’s method to correct for multiple comparisons.

Results

Ultrasound pressure field characterization

Based on k-Wave simulations, the large impedance mismatch at the water/skull boundary attenuates the ultrasound beam by approximately 30%. This matched ultrasound attenuation measurements made through an ex vivo skull, detailed above in Methods. Figure 3 shows a simulated cross-section of the mouse head at Location 1 overlaid with an anatomical cross-section though the plane of the FUS focus. The region of the skull over which the transducer was situated is relatively flat, so the focus is not significantly refracted away from its initial axis and retains its shape after passing through the skull. An overlay of the approximate FUS focus that shows the incident orientation is shown in blue in Fig. 3. The slight curvature of the skull results in an angled reflection of the US beam away from the skull at the water/skull boundary, despite the normal incidence angle of the transducer with respect to the horizontal.

Figure 3:

Simulation of the pressure profile within the mouse head. The simulations performed in k-Wave demonstrate that significant standing waves are present within the ultrasound focus but that large pressure nodes (<6 dB) are not present in the brain outside of the ultrasound focus.

Due to the size of the ultrasound wavelength (0.75 mm) relative to the size of the mouse skull (10 mm or 13.3 wavelengths), continuous-wave excitation results in the formation of standing waves within the mouse skull cavity. This leads to multiple small pressure peaks located throughout the brain. However, few pressure peaks outside of the focal volume exceed the −3dB threshold. Corresponding peaks in temperature are notably absent in thermal simulations, as thermal diffusion dominates over the small pressure variations.

Temperature field characterization

Using the output pressure map from the acoustic simulation, we simulated the temperature changes resulting from the absorption of acoustic energy using the k-Wave function kWaveDiffusion to solve Pennes’ bioheat equation in 2-D. The thermal simulation in Figure 4 shows that, over the 60s stimulation period, the intensity map of the total temperature change in the brain is defined by the FUS focal region, outlined in gray. The peak brain temperature of 2.1 °C is located just beneath the skull, where absorption of ultrasound is highest. The maximum temperature within the regions of the thalamus and hypothalamus are approximately 1.8 °C and 1.75 °C, respectively. The simulation also shows an estimation of the global increase in temperature throughout the brain as heat diffuses away from the focus to regions of lower temperature.

Figure 4:

Simulation of the temperature profile within the mouse brain due to FUS at 850 kPa. Both the axial and lateral heating profile correspond approximately to the FUS focus (−6 dB pressure threshold outlined in gray). Heat diffuses throughout the rest of the brain during sonication, resulting in a global temperature increase ranging from 0.5 − 1.5 °C.

The result from simulating the 60s laser heating experiment is shown in Figure 5. The temperature profile for laser heating is more confined to the target region than that of FUS heating, as the fiber tip is inserted directly into the brain and thus the light does not propagate through the same volume of skull and tissue as FUS. The extent of heating with respect to the source is also limited due to the higher absorption of light by brain tissue than that of the 2 MhHz ultrasound wave. The peak temperature at the tip of the fiber is 1.97 °C. At a depth of 1 mm below the fiber tip, located within the hypothalamus, the temperature decreases to 1.48 °C.

Figure 5:

Simulation of the temperature profile due to laser heating within the mouse brain. Laser heating simulations demonstrate focal heating near the fiber tip (fiber indicated by a thin white bar extending down from the top of the graph at Δx = −2 mm.

In addition to estimating temperature through simulations, we directly measured the temperature increase at the primary target by thin needle or thin wire thermocouple. Figure 6 shows how the temperature measured at the target changes over time for FUS at 425 kPa or 850 kPa, as well as laser heating. Each line represents the average of five FUS/laser pulses each from two mice, with the shaded region around each line representing the maximum and minimum values recorded. The peak recorded temperatures for FUS at 425 kPa, FUS at 850 kPa, and laser heating were 0.30°C, 1.95°C, and 1.93°C, respectively. The rates of heating and heat diffusion were consistent between trials and between mice. For each parameter, temperature increased rapidly and remained elevated for the duration of the trial. Temperatures for FUS at 425 kPa, FUS at 850 kPa, and laser heating reached 90% of their peak values within 8.3 s, 15.0 s, and 14.5 s, respectively. At stimulus offset, all temperatures sharply dropped toward baseline values. In the case of FUS at 425 kPa, the temperature reached its baseline value approximately 30 s after stimulus offset. It continued to decline to a minimum of −0.06 °C below baseline. While temperatures for FUS at 850 kPa and laser did not, on average, reach baseline within 60s of stimulus offset, the average minimum temperature above baseline was 0.07 °C and 0.02 °C, respectively. The differences in baseline temperature at 60 s following stimulus offset for FUS at 850 kPa and laser heating are small relative to peak temperatures (3% and 1% of peak values, resp.), and all final temperature measurements are within the thermocouple error range of the baseline (minimum error: ±0.1 °C).

Focused ultrasound modulates cardiac and respiratory rate

We measured cardiac and respiratory rates during ultrasound delivery to the in vivo murine brain in order to assess FUS-induced changes in arousal and autonomic nervous activity. Arousal and autonomic functions are closely linked to changes in environmental, cerebral, and core body temperature, so we hypothesized that a localized, FUS-induced increase in temperature would affect neuronal activity in a temperature-sensitive brain region, which could be observed by monitoring changes in physiological activity associated with that region. For these reasons, the target coordinates were chosen to encompass both the thalamus and hypothalamus. It is important to note, however, that the length of the focal volume contained structures outside our target region of interest, most notably the visual cortex and hippocampus.

Fig. 8 shows the mean change in cardiac (a) and respiratory rate (b) during the sonication window for each group. Significant differences were found between all groups in both cardiac rate (F (4, 18) = 6.37, p < 0.01) and respiration rate (F (4, 18) = 4.60, p < 0.01). As can be seen by Fig. 8, changes in cardiac and respiratory rates varied both by stimulus intensity and targeting coordinates. Fig. 7a and Fig. 7b show examples of the changes in cardiac and respiratory rates, respectively, that are associated with sonication at 850 kPa in Location 1. Short-term changes in heart and breathing rates occurred during each 60-s sonication window (Figs. 7a & 7b, gray boxes), followed by an immediate correction back toward pre-stimulus values.

Figure 8:

Average difference in value before and during sonication of (a) cardiac rate and (b) respiratory rates. Values are the mean and errors are in standard deviations. Statistical significance shows results from post-hoc testing with Tukey’s correction for multiple comparisons. (*: < 0.05, **: < 0.01). N-number per group is noted in the legend.

Figure 7:

Example plots of changes in vital rates during sonication. Dark gray bars indicate the 15 second baseline prior to each sonication whereas light gray bars indicate the 60 second sonication window. For all sonications, changes in vital rates were taken as the average value in the dark gray bar subtracted from the average value in the light gray bar. (top) Respiratory rate plot of the first seven sonications of one mouse sonicated at 850 kPa, targeted to Location 1. (bottom) Heart rate plot for the same mouse.

Focused ultrasound modulation of cardiac and respiratory rates varies with pressure

Fig. 8 illustrates differences in the magnitude of FUS-induced cardiac and respiratory rate changes across FUS pressure at the same sonication location. In the sham condition, changes in cardiac and respiratory rate of −0.26 ± 0.38 bpm and 0.51 ± 0.31 brpm, respectively, were found. At 425 kPa, corresponding to a temperature increase of approximately 0.3 °C, the changes in cardiac and respiratory rate were small (respectively, 2.5 ± 0.74 bpm and −2.7 ± 3.3 brpm) and not significantly different from the sham group (Cardiac rate: q(5, 18) = 2.56, p = 0.40; Respiratory rate: q(5, 18) = 1.48, p = 0.83). At a greater sonication pressure of 850 kPa in Location I, however, a mean decrease in the cardiac rate of 3.2 ± 1.8 bpm and an increase in the respiratory rate of 6.5 ± 3.2 brpm occurred, both of which are larger in magnitude than the rate changes observed at 425 kPa. The increase in respiratory rate at 850 kPa at Location I was also significantly different from sham (q(5, 18) = 5.04, p < .05).

Focused ultrasound modulation of pulse and respiration rate differs by target

Differences were apparent with respect to targeting between Location I and Location II (Fig. 8). At Location I, sonication at 850 kPa resulted in a significant increase in respiratory rate compared to sham and a non-significant decrease in heart rate. The same pressure at Location II resulted in an increase in respiratory rate (3.0 ± 3.3 brpm), but the result was not significantly different from sham (q(5, 18) = 1.85, p = 0.69). In contrast to all sonication and laser heating in Location I, 850 kPa in Location II resulted in an increase in heart rate (1.3 ± 1.1 bpm). This heart rate increase was not significantly different from sham (q(5, 18) = 1.65, p = 0.77), but was significantly different from the breathing rate decrease at the same pressure at Location I (q(5, 18) = 5.34, p < .05).

Laser heating modulates pulse and respiration rate similarly to FUS

During the FUS experiments, both mechanical and thermal effects are present. Due to the nature of acoustic wave propagation in tissue, these effects are inextricably linked. In order to investigate a purely thermal mode of stimulation, we assessed the effect of localized heating on heart and respiration rates using an optical fiber coupled to a 50 mW laser (operating at an effective output power of 38 mW) and inserted into the brain at Location I. When an increase in temperature comparable to that of FUS at 850 kPa was achieved using laser stimulation, the effect on cardiac and respiratory rate was comparable (Fig. 8). Laser stimulation lowered the heart rate, similar to FUS at 850 kPa, although to a greater extent. The −4.5 ± 1.8 bpm decrease with laser heating was significantly differs from sham (q(5, 18) = 4.42, p < .05) and from the increase in heart rate found at 850 kPa at Location II (q(5, 18) = 6.08, p < .001). Similarly, the change in breathing rate (6.6 ± 3.2 brpm) significantly different from the sham group (q(5, 18) = 4.55, p < .05). This laser-induced increase in breathing rate of 6.6 ± 3.2 brpm was highly comparable to the 6.5 ± 3.2 brpm FUS at 850 kPa.

Histology

No evidence of cell damage was found within or adjacent to the FUS focus in any of the three mice used for histology. Figure 9 shows four regions of a hematoxylin and eosin-stained brain section. The section is a coronal slice through the center of the FUS focus in a single mouse (approximately 2 mm posterior to lambda), and the images in Fig. 9 are, clockwise from top left, 1) the skull/brain interface, 2) cerebral cortex, 3) hippocampus, and 4) thalamus. In one mouse, red blood cells were found, infrequently, in regions far outside the FUS focus and/or still contained within blood vessels, and not in the parenchyma. While FUS-induced damage is possible outside of the focal area, for example due to the presence of standing waves of sufficient magnitude, we do not have reason to believe that this is the case here. Our simulations suggest that standing waves were present but at significantly lower amplitude than at the FUS focus. High-resolution scans of the histology results are included in Supplementary Material.

Figure 9:

Histological (H&E) evaluation of brain regions within the FUS focus, including 1) the skull/brain interface, 2) the cerebral cortex, 3) the hippocampus, and 4) the thalamus. No evidence of cell damage could be found in or near the focus.

Discussion

In this study, we have demonstrated reversible, non-invasive neuromodulation of cardiorespiratory activity in anesthetized mice using focused ultrasound. Continuous-wave mode operation of the transducer was chosen to allow that thermal effects would dominate over mechanical effects, compared to pulsed operation. By our choice of parameters, we sought to investigate the effect of FUS-induced localized heating in a brain region containing both temperature-sensitive neurons and structures central to arousal and autonomic regulation (i.e., the thalamus and hypothalamus). Changes in thermoregulation, states of arousal, and autonomic activity are reflected, in part, by changes in cardiorespiratory activity. We observed that sonication in this region produced consistent and reversible changes in cardiac and respiratory rates in a pressure-dependent and target-dependent manner. During a 60-s sonication at 850 kPa targeted to the thalamus and hypothalamus, a moderate increase in local temperature (i.e., 2 °C) resulted in a decrease in heart rate and a significant increase in breathing rate. Either a lower increase in temperature (i.e. < 0.5 °C, FUS at 425 kPa) in the same region or a moderate temperature increase (2 °C, FUS at 850 kPa) targeted to a distant motor-related region were insufficient to significantly modulate cardiorespiratory activity compared to sham. The FUS at 850 kPa in the motor-related region did, however, cause the only increase in heart rate, compared to the heart rate decreases found at both FUS pressures and with laser heating at the thalamus and hypothalamus.

The target dependence of the observed effect due to ultrasound indicates the differences in response to stimulation could be linked to regional differences in brain function. We observed large differences in the magnitude of respiratory rate increase (an approximately two-fold difference between Location II and Location I) and differing direction of the change in cardiac rate (decrease of 3.2 bpm at Location I to increase of 1.3 bpm at Location II). We, therefore, conclude that the inclusion of regions of the brain that are susceptible to thermal stimulation, such as the hypothalamus, is a key finding in this study. This also indicates that the response is unlikely due to global changes in cerebral blood flow to maintain homeostatic temperatures or a sympathetic stress response to pain, auditory response, or skull vibrations, as these would not significantly vary between targets. The changes are thus unlikely due to an inadequate depth of anesthesia that could induce a startle or stress response. The same positive responses were observed in deeply anesthetized mice that were unresponsive to toe pinches, and mice in lighter planes of anesthesia, as inferred from elevated cardiac and respiratory rates, were not associated with increased responsiveness to stimulation (see Supplementary Figures S1 and S2). This study does not address the possibility that the effect of the temperature increase on neuronal activity is indirect (e.g. increase in activity subsequent to temperature-induced changes in cerebral blood flow, augmented cellular metabolism, etc.) as opposed to direct (e.g. direct activation of temperature-sensitive ion channels). The spatial localization of responses to an area of the brain that regulates, and is regulated by, environmental temperature, and the similarity in responses between a simultaneous thermal/mechanical stimulus (FUS) and a purely thermal stimulus (laser), indicate that temperature changes are, however, central to these results. We do not have evidence to draw any conclusions as to why the effect on heart rate of the laser stimulation and FUS stimulation is different. It is possible that the difference is due to the invasive nature of the laser stimulation experiments or an additional influence of the mechanical stimulus at that pressure.

Deviations in the simulated temperature profile (Fig. 4) from that measured in vivo via thermocouple (Fig. 6) could be attributed to the influence of a number of variables for which it is difficult to account, including the effect of anesthesia on brain temperature and cerebral blood flow, as well as differences in the estimated versus actual location of the thermocouple with respect to the simulated temperature profile.

Modulation of cardiorespiratory activity has been achieved by stimulation of structures that fall within the Location I focal volume of our study, including the hypothalamus, thalamus, hippocampus, globus pallidus, and subthalamic nucleus^18,30–35. Specifically, hypothalamic electrical stimulation in rabbits has produced similar effects at high-frequency stimulation as those seen in this study when stimulating at Location I (which encompasses the hypothalamus), namely bradycardia with concurrent tachypnea³³. Rabbits given electrical stimulation to the hypothalamus experienced a transient decrease in cardiac rate lasting throughout stimulation. At lower stimulation strength, the authors acknowledged that the cardiac rate changes were likely related to the baroreceptor reflex to maintain blood pressure. When the rabbits were stimulated during artificial ventilation to maintain a constant respiratory rate, the cardiovascular effects remained. These results were also accompanied by increases in blood pressure (BP), which were not measured in the present study. In humans, electrical stimulation of the hypothalamus, thalamus, and subthalamic nucleus have also been shown to have effects on cardiorespiratory activity³⁴.

Localized warming of the preoptic area of the hypothalamus has been used to induce increased respiration in urethane anesthetized rats¹⁸. In that study, the increase in respiratory rate was similar to that observed during comparable whole-body warming. The preoptic area is also known to contain temperature-sensitive neurons and receives input from peripheral thermoreceptors. Signals from the hypothalamus to the brainstem go on to coordinate the appropriate physiological response in the respiratory network. The aforementioned study did not, however, report a concomitant change in heart rate.

The results presented herein are of potential relevance to clinical applications of FUS, where vital sign measurement is critical to monitoring the physiological response of the patient during FUS. Knowledge of the expected impacts of stimulation on heart and breathing rates could assist clinicians in confirmation of targeting. Additionally, treatment intensity and duration can be adjusted to account for anticipated effects on heart and breathing rate to ensure patient safety.

Due to the lateral extent of the ultrasound focus, it is possible that some degree of warming targeted to Location II overlaps with more caudal structures in the hypothalamus, including the lateral preoptic nucleus. This could explain the presence of a small effect on respiratory rate when targeting Location II and highlights the difficulty of activating isolated functional regions when the ultrasound focus is large relative to the animal’s brain.

In addition to the cardiorespiratory depressive effects of the anesthesia used, the room temperature water bath used for coupling the FUS transducer to the mouse head is a significant source of brain cooling, which in turn has a potentially large effect on brain activity and depth of anesthesia. In particular, we have noted that cardiac and respiratory rates can drop dramatically upon placement of the water bath and application of ultrasound gel. Thus, warming of the brain after it has cooled due to the effects of anesthesia and the water bath may serve to counteract some of the depressive forces on cardiorespiratory activity that are already acting.

A shortcoming of this study is the lack of confirmation as to the targeting and spatial extent of laser-induced heating, confirmation of changes in neuronal activity in the targeted regions, confirmation of modulation of blood flow in the target regions, and the spatial extent of the ultrasound beam encompassing broad regions of the brain that are not limited to the intended targets, resulting in difficulty in separating the effects of distinct brain regions in terms of the effect that their stimulation causes.

Conclusions

We have demonstrated that neuromodulation via transcranial focused ultrasound can evoke changes in the cardiac and respiratory rates in mice. Continuous-wave FUS, targeted to the thalamus and hypothalamus, suppressed cardiac rate and increased respiratory rate for the 60-s duration of sonication. Upon stimulus offset, both rates reverted toward their respective pre-stimulus trends. The impacts of the FUS stimulus were target- and intensity-dependent. Low-pressure sonication failed to significantly alter either measure, indicating the presence of an activation threshold. Further, sonication of distant motor-related targets did not significantly affect cardiac or respiratory rate. This suggests that the effect is dependent on the modulation of brain structures involved in regulating cardiorespiratory activity. Localized heating by a 650 nm laser was capable of producing similar significant changes in cardiac and respiratory rate, which indicates that the effect is mediated partially by a thermal mechanism, as opposed to a purely mechanical mechanism or auditory confounds. Our findings are akin to previously published results on electrical stimulation and localized warming of the hypothalamus in rabbits and rats. Ongoing investigations include monitoring of vital signs as a means of assessing systemic impacts of FUS neuromodulation and as a possible indicator of temperature-related effects. In conclusion, our study demonstrates that localized brain warming via transcranial focused ultrasound can produce transient changes in cardiorespiratory activity in mice and warrants further research to elucidate the temperature-related mechanisms of FUS neuromodulation that are both distinct and complementary to its mechanical effects.

Data Availability

Any data presented herein can be made available on demand.