Brain-machine interfaces (BMIs) have shown promise in rehabilitating peripheral sensory systems, such as restoring hearing or sight. Restoring vision, however, presents the most significant challenge due to the need for high-speed transmission (13 Hz) of complex spatial patterns. While epicortical or intracortical implants can achieve form perception, their lack of long-term sustainability drives the search for non-invasive, remote neuronal circuit activation methods. Optogenetics offers an alternative, demonstrated effectively in retinal applications, even reaching clinical trials. However, optical stimulation of the cortex faces limitations due to the dura mater, brain scattering, and light absorption, necessitating invasive light guides. Ultrasound (US) waves offer a potential solution for non-contact neuromodulation of cortical and subcortical brain areas. This requires craniotomy and high US frequencies for adequate spatial resolution. Most existing US neuromodulation strategies utilize low or mid-range frequencies, resulting in poor spatial resolution and/or prolonged responses. High-frequency (30 MHz) US has shown inhibitory neuromodulation, while other attempts at high-frequency stimulation have resulted in high acoustic energy levels, risking thermal heating and tissue damage. Sonogenetic therapy, using ectopic expression of US-sensitive proteins like TRP1, MscL, or prestin via AAV gene delivery, aims to generate neuronal mechanosensitivity. However, this approach has lacked the spatiotemporal resolution needed for vision restoration. This study aimed to investigate if MscL channels could enhance US sensitivity in vivo, target specific neurons via gene therapy, induce responses with millisecond precision sufficient for visual restoration, and achieve high spatial resolution using high-frequency US at low intensities to avoid adverse effects. The researchers used the retina, a readily accessible part of the central nervous system, as a model for initial testing.

The existing literature highlights the challenges and successes in visual restoration using BMIs. Multielectrode arrays (MEAs) have been successful in peripheral sensory system rehabilitation and restoring hearing and sight. However, vision restoration requires very high spatiotemporal precision to transmit complex spatial patterns at rates up to 13 Hz. While cortical implants can achieve form perception, long-term sustainability remains a major hurdle. Optogenetic techniques have provided an alternative approach, particularly successful in retinal applications, even demonstrating clinical efficacy. However, these methods encounter limitations when applied to the cortex due to the skull and its related challenges: the dura mater impedes light penetration; scattering and absorption of light within brain tissue necessitates the use of invasive light guides. Ultrasound (US) offers a potential non-invasive solution for neuromodulation of both cortical and subcortical areas; however, achieving the necessary spatiotemporal resolution remains a challenge. Most current US neuromodulation strategies employ low to mid-range frequencies, which limits spatial precision and often leads to slow or sustained responses. High-frequency US, while potentially offering improved resolution, may require higher energy levels that could induce thermal damage. Previous sonogenetic studies using US-sensitive proteins have shown promise in sensitizing neurons to US stimulation but have not yet reached the required spatiotemporal resolution for successful visual restoration.

The study utilized both in vitro and in vivo approaches. In vitro experiments were conducted on rat retinal ganglion cells (RGCs). An adeno-associated virus (AAV) encoding the *mscL* gene from *Escherichia coli* was delivered intravitreally to target RGCs. Two versions of the *mscL* gene were used: wild-type (WT) and a G22S mutant known to enhance pressure sensitivity. The researchers employed multi-electrode arrays (MEAs) to record RGC responses to ultrasound stimulation at varying frequencies and intensities. Different US parameters, such as frequency, pressure, and duration, were systematically tested to examine spatiotemporal resolution. In vivo experiments focused on the visual cortex of rats and mice. Similar AAV delivery methods were used, but with a CamKII promoter to target cortical neurons. High-frequency (15 MHz) focused ultrasound was used to stimulate the visual cortex, while electrophysiological recordings were made using penetrating microelectrode arrays and µEcoG electrode arrays. Behavioral assays, involving associative learning with water rewards, were used to assess whether the sonogenetic stimulation of the visual cortex could induce a perception in mice that could be associated with light perception.

The study demonstrated that ectopic expression of the MscL channel significantly increased the sensitivity of both retinal and cortical neurons to ultrasound stimulation. In retinal ganglion cells (RGCs), the G22S mutant of MscL showed enhanced sensitivity compared to the wild-type (WT) form. The responses to 15 MHz ultrasound stimulation were fast, with latencies as short as 5 milliseconds. The spatial resolution of the stimulation was remarkably high, at least 400 µm in the x-y plane. In vivo experiments showed that stimulation of the visual cortex with 15 MHz ultrasound elicited sustained responses in cortical neurons with latencies shorter than 10 ms. These responses were observed at various cortical depths, indicating efficient penetration of the ultrasound. Remarkably, the neurons were able to generate distinct responses to each US stimulus up to a repetition rate of 13 Hz, suggesting the potential for high-speed pattern presentation. Importantly, the study demonstrated that sonogenetic activation of the visual cortex in mice led to a behavioral response consistent with light perception. Mice trained to associate light stimulation of one eye with a water reward showed a similar response rate when the visual cortex was stimulated with ultrasound, indicating the perception of an equivalent stimulus. The acoustic intensities used in the study were well below the FDA safety limits for US imaging, mitigating concerns about potential tissue damage.

This study successfully addresses the long-standing challenge of achieving high spatiotemporal resolution in ultrasound-mediated neuronal stimulation. The use of the MscL channel, particularly the G22S mutant, and high-frequency ultrasound significantly enhanced the precision and speed of neuronal activation. The millisecond-scale responses and high spatial resolution demonstrated here are critical for applications such as visual restoration. The behavioral results further support the potential of this approach for restoring sensory function. The low acoustic intensities employed minimize the risk of tissue damage, paving the way for future clinical translation. This work surpasses previous sonogenetic studies which were limited by poor spatiotemporal resolution. The demonstrated 13 Hz stimulation frequency, while lower than the fastest opsins, might suffice for vision restoration. This technology significantly advances the possibilities for developing less invasive brain-machine interfaces for neurological treatments.

This research demonstrates that sonogenetics, using the G22S MscL channel and high-frequency ultrasound stimulation, enables millisecond-precise and spatially-confined activation of neurons in the retina and visual cortex. This technology shows great promise for the development of less invasive brain-machine interfaces for visual restoration and other neurological applications. Future research should focus on optimizing the gene delivery methods and further improving the spatiotemporal resolution to achieve even faster and more precise neuronal control.

While the study demonstrated high spatiotemporal resolution and behavioral effects, the animal models were rats and mice. Direct translation to humans might require further optimization of the parameters. The long-term effects of repeated ultrasound stimulation and gene expression need further investigation to ensure safety and efficacy. Additionally, while the behavioral tests suggested light perception, more sophisticated visual tasks could be performed to further confirm the nature of this perception.