Three-dimensional cell agglomerates are broadly useful in tissue engineering and drug testing. We report a well-free method to form large (1.4-mm) multicellular clusters using 100-MHz surface acoustic waves (SAW) without direct contact with the media or cells. A fluid couplant is used to transform the SAW into acoustic streaming in the cell-laden media held in a petri dish. The couplant transmits longitudinal sound waves, forming a Lamb wave in the petri dish that, in turn, produces longitudinal sound in the media. Due to recirculation, human embryonic kidney (HEK293) cells in the dish are carried to the center of the coupling location, forming a cluster in less than 10 min. A few minutes later, these clusters may then be translated and merged to form large agglomerations, and even repeatedly folded to produce a roughly spherical shape of over 1.4 mm in diameter for incubation-without damaging the existing intercellular bonds. Calcium ion signaling through these clusters and confocal images of multiprotein junctional complexes suggest a continuous tissue construct: intercellular communication. They may be formed at will, and the method is feasibly useful for formation of numerous agglomerates in a single petri dish.

Ultrasound has been used to non-invasively manipulate neuronal functions in humans and other animals. However, this approach is limited as it has been challenging to target specific cells within the brain or body. Here, we identify human Transient Receptor Potential A1 (hsTRPA1) as a candidate that confers ultrasound sensitivity to mammalian cells. Ultrasound-evoked gating of hsTRPA1 specifically requires its N-terminal tip region and cholesterol interactions; and target cells with an intact actin cytoskeleton, revealing elements of the sonogenetic mechanism. Next, we use calcium imaging and electrophysiology to show that hsTRPA1 potentiates ultrasound-evoked responses in primary neurons. Furthermore, unilateral expression of hsTRPA1 in mouse layer V motor cortical neurons leads to c-fos expression and contralateral limb responses in response to ultrasound delivered through an intact skull. Collectively, we demonstrate that hsTRPA1-based sonogenetics can effectively manipulate neurons within the intact mammalian brain, a method that could be used across species.

Ultrasound has been used to manipulate cells in both humans and animal models. While intramembrane cavitation and lipid clustering have been suggested as likely mechanisms, they lack experimental evidence. Here, high-speed digital holographic microscopy (kiloHertz order) is used to visualize the cellular membrane dynamics. It is shown that neuronal and fibroblast membranes deflect about 150 nm upon ultrasound stimulation. Next, a biomechanical model that predicts changes in membrane voltage after ultrasound exposure is developed. Finally, the model predictions are validated using whole-cell patch clamp electrophysiology on primary neurons. Collectively, it is shown that ultrasound stimulation directly defects the neuronal membrane leading to a change in membrane voltage and subsequent depolarization. The model is consistent with existing data and provides a mechanism for both ultrasound-evoked neurostimulation and sonogenetic control.

To study cerebellar activity during learning, we made whole-cell recordings from larval zebrafish Purkinje cells while monitoring fictive swimming during associative conditioning. Fish learned to swim in response to visual stimulation preceding tactile stimulation of the tail. Learning was abolished by cerebellar ablation. All Purkinje cells showed task-related activity. Based on how many complex spikes emerged during learned swimming, they were classified as multiple, single, or zero complex spike (MCS, SCS, ZCS) cells. With learning, MCS and ZCS cells developed increased climbing fiber (MCS) or parallel fiber (ZCS) input during visual stimulation; SCS cells fired complex spikes associated with learned swimming episodes. The categories correlated with location. Optogenetically suppressing simple spikes only during visual stimulation demonstrated that simple spikes are required for acquisition and early stages of expression of learned responses, but not their maintenance, consistent with a transient, instructive role for simple spikes during cerebellar learning in larval zebrafish.

Ultrasound has been shown to affect the function of both neurons and non-neuronal cells, but, the underlying molecular machinery has been poorly understood. Here, we show that at least two mechanosensitive proteins act together to generate C. elegans behavioral responses to ultrasound stimuli. We first show that these animals generate reversals in response to a single 10 msec pulse from a 2.25 MHz ultrasound transducer. Next, we show that the pore-forming subunit of the mechanosensitive channel TRP-4, and a DEG/ENaC/ASIC ion channel MEC-4, are both required for this ultrasound-evoked reversal response. Further, the trp-4;mec-4 double mutant shows a stronger behavioral deficit compared to either single mutant. Finally, overexpressing TRP-4 in specific chemosensory neurons can rescue the ultrasound-triggered behavioral deficit in the mec-4 null mutant, suggesting that both TRP-4 and MEC-4 act together in affecting behavior. Together, we demonstrate that multiple mechanosensitive proteins likely cooperate to transform ultrasound stimuli into behavioral changes.