The Mechanosensitive Ion Channel Piezo1 Significantly Mediates In Vitro Ultrasonic Stimulation of Neurons.

Ultrasound brain stimulation is a promising modality for probing brain function and treating brain disease non-invasively and with high spatiotemporal resolution. However, the mechanism underlying its effects remains unclear. Here, we examine the role that the mouse piezo-type mechanosensitive ion channel component 1 (Piezo1) plays in mediating the in vitro effects of ultrasound in mouse primary cortical neurons and a neuronal cell line. We show that ultrasound alone could activate heterologous and endogenous Piezo1, initiating calcium influx and increased nuclear c-Fos expression in primary neurons but not when pre-treated with a Piezo1 inhibitor. We also found that ultrasound significantly increased the expression of the important proteins phospho-CaMKII, phospho-CREB, and c-Fos in a neuronal cell line, but Piezo1 knockdown significantly reduced this effect. Our findings demonstrate that the activity of mechanosensitive ion channels such as Piezo1 stimulated by ultrasound is an important contributor to its ability to stimulate cells in vitro.

Introduction

Neuromodulation technologies developed recently, such as magnetic, electric, and optogenetic stimulation methods, have provided unprecedented ways to study brain function and treat its diseases and conditions. Ultrasound is another modality being studied for its neuro-stimulation abilities. Ultrasound beams can be focused on millimeter-sized areas in sub-cortical regions through the intact skull (King et al., 2013, Legon et al., 2014, Tufail et al., 2010). It has been demonstrated to be capable of generating neural responses from the human visual cortex (Lee et al., 2016), somatosensory cortex (Lee et al., 2015, Legon et al., 2014), motor cortex (Legon et al., 2018b), and even the thalamus (Legon et al., 2018a) without obvious side effects. These features have spurred research into ultrasonic neuromodulation, with attempts being made to use it therapeutically for conditions such as Parkinson disease, epilepsy, and depression (Leinenga et al., 2016).

The observed effects of ultrasound, a physical stimulus, in the brain points to central nervous system cells possessing mechanosensitivity by various mechanisms (reviewed by Tyler (2012)). At the low intensities required to apply ultrasound in the brain through the intact skull safely and successfully (Duck, 2007, Fry and Goss, 1980, Tyler et al., 2008), the thermal and cavitation effects of ultrasound are likely to be minor (Dinno et al., 1989, Kim et al., 2014, Tyler et al., 2018). Another possible mechanism of ultrasound is through its action on mechanoresponsive components of cellular machinery, which sense physical forces and initiate cellular signaling. Given the observed timescale of ultrasound's stimulation effects, a plausible explanation is offered by the activity of mechanosensitive ion channels (Bystritsky et al., 2011, Fomenko et al., 2018, Tyler et al., 2018). Such channels are implicated in sensing a wide variety of physical stimuli, including sound, membrane stretch, and shear forces (Martinac, 2012). Some mechanosensitive ion channels have been shown to be activated by ultrasound, but these studies lack the context of brain stimulation, studying either non-mammalian channels (Ibsen et al., 2015, Kubanek et al., 2018, Ye et al., 2018) or mammalian channels expressed in non-neuronal cells (Kubanek et al., 2016).

Among the possible ultrasound-responsive mechanosensitive ion channels, members of the Piezo family are eminent candidates (Coste et al., 2010). The Piezos are very large, evolutionarily conserved transmembrane proteins, and Piezo1 is among the channels most sensitive to physical force (Cox et al., 2017). It can activate and deactivate in the range of milliseconds (Coste et al., 2010) to forces estimated to be as low as 10 pN (Wu et al., 2016). Although it allows cations to permeate cells in general, it is reported to exhibit a preference for calcium ions (Ca2+ (Coste et al., 2010). Heterologously expressed Piezo1 could be activated by ultrasound in non-neuronal cells (Gao et al., 2017, Pan et al., 2018, Prieto et al., 2018) but only with the use of very high frequencies or with microbubbles. It remains to be shown whether endogenous Piezo1 in neurons can be activated by ultrasound alone and what role this interaction could play in neuro-stimulation by ultrasound.

In the present study we confirm endogenous Piezo1 expression in mouse primary neurons, and in a neuronal cell line, and show that these channels play a role in neuronal activation in vitro. We demonstrate that low-intensity low-frequency ultrasound alone can activate heterologously expressed HEK293T cells, as well as endogenous Piezo1 channels, initiating Ca2+ influx and increasing levels of c-Fos. We also show ultrasound is capable of significantly affecting the levels of downstream Ca2+ signaling proteins crucially involved in neuronal function and that knocking down Piezo1 significantly decreased this effect. Thus, we show that Piezo1 activation plays a major role in the mechanotransduction of ultrasound and that ultrasound alone is capable of significantly affecting the function and activation of neurons in vitro.

Results

Customized In Vitro Ultrasound Stimulation Setup

We developed a customized in vitro ultrasound stimulation system incorporated with a calcium imaging system (Figure 1A). Briefly, the calcium imaging system consisted of a modified upright epifluorescence microscope. The excitation light was generated by a dual-color LED, filtered and delivered to the sample to illuminate the calcium sensor. Signals from the cells were collected by a water immersion objective, passed through a dual-filter wheel, and captured by a camera. To minimize phototoxic effects, the LEDs were triggered at 1 Hz and synchronized with sCMOS time-lapse imaging. Ultrasound was delivered through a triangle waveguide attached to the transducer placed below the culture dish at a 45-degree angle. The waveguide was also attached to an acoustic absorber to minimize acoustic reverberation. This design helped us to minimize the generation of standing waves and created a strong water-air interface on the bottom of the dish. This results in a controllable ultrasound field, as measured by an acoustic pressure mapping system consisting of a needle hydrophone and high-precision 3D motor (Figure 1B). The obtained acoustic pressure map is displayed in Figure 1C, using an acoustic pressure of 0.3 MPa for illustration. Each stimulus was composed of 200 tone burst pulses at a center frequency of 500 kHz with a duty cycle of 40% at a pulse repetition frequency (PRF) of 1 kHz, at low acoustic intensities (Figure 1D). These parameters corresponded to very short bursts of ultrasound stimulation, helping to minimize any resultant thermal effects (Kubanek et al., 2018).

Figure 1

The Mouse Piezo1 Channel Transfected into HEK293T Cells Is Activated by Ultrasound, Using Our Ultrasound Stimulation System

(A) A schematic illustration of our combined calcium imaging and ultrasound system.

(B) A schematic illustration of how the acoustic map of our customized ultrasound system was generated.

(C) Acoustic mapping profile of the controllable ultrasound field generated using the customized setup.

(D) A schematic illustration of the ultrasound stimulation parameters used in our experiments, with a peak-to-peak acoustic pressure of 0.3 MPa used as an example.

(E) The expression of functional mouse Piezo1 in Piezo1-transfected cells (“Piezo1”), compared with a vector control (“Ctrl”), was verified in three ways. Left: qRT-PCR was performed for mouse Piezo1, normalized to β-actin and expressed as a fold change. Bar charts represent the mean ± SEM of three independent experiments. n = 3, ***p < 0.0001, unpaired two-tailed t test. A representative Western blot of the two groups is also shown. Piezo1 functionality was verified by stimulating cells with 1 μM Piezo1 agonist Yoda1 and performing calcium imaging with Cal-590. Middle: a representative time course of Yoda1 stimulation on the two groups is shown. Right: bar chart shows the mean ± SEM of three independent Ca2+ imaging experiments. n = 10, ***p < 0.0001, unpaired two-tailed t test.

(F) Left: a representative time course of Ca2+ imaging comparing ultrasound stimulation at different intensities (numbers represent MPa) of Piezo1-transfected cells compared with the control. Right: the bar chart shows the mean ± SEM of three independent Ca2+ imaging experiments. n = 9, ***p < 0.0001, two-tailed unpaired t test with Holm-Sidak correction. All statistically significant differences are shown.

See also Figure S1.

Heterologously Expressed Piezo1 Can Be Activated by Ultrasound and Induce Ca2+ Influx in HEK293T Cells

To confirm that ultrasound could activate Piezo1 using our customized setup, we overexpressed mouse Piezo1 heterologously in HEK293T cells, known to show minimal response to mechanical stimulation (Coste et al., 2010, Syeda et al., 2015). We transfected cells with a pcDNA3.1-mPiezo1-IRES-GFP plasmid (described in Coste et al. (2010)) or a vector control (described in Schaefer et al. (2008)) and treated them with a Piezo1-specific agonist, Yoda1 (Syeda et al., 2015). We verified the overexpression using qRT-PCT and Western blotting for Piezo1, and calcium imaging revealed that 1 μM Yoda1 induced significantly more Ca2+ influx in the Piezo1-transfected cells, compared with the control (Figure 1E). We thus confirmed the expression and functionality of transfected mouse Piezo1 channels in HEK293T cells.

We proceeded to perform Ca2+ using ultrasound at acoustic pressures corresponding to a range previously reported to have elicited responses (Tufail et al., 2010). We found that increasing ultrasound pressure on Piezo1-transfected cells resulted in increasing Ca2+ influx into cells, with 0.3 MPa ultrasound showing a significant increase, whereas the control remained largely unchanged (Figure 1E). To exclude the possibility that the fluorescence changes were due to cells being out-of-focus due to ultrasound perturbation, we also performed ratiometric Ca2+ imaging. We observed that the fluorescence ratio (340/380) of control cells showed no response, whereas the Piezo1-expressing cells showed a robust and large response to 0.1 MPa ultrasound (Figure S1). Pre-treating the Piezo1-expressing cells with a Piezo1-specific inhibitor, GsMTx-4 (Bae et al., 2011), at 40 μM reduced the cells' response to the same level as the control cells (Figure S1). This finding confirmed that the elevated cytoplasmic Ca2+ levels seen upon ultrasound stimulation were not experimental artifacts. Thus, our ultrasound stimulation setup could successfully activate Piezo1 to allow Ca2+ influx at low, physiologically relevant acoustic pressures.

Mouse Primary Neurons Are Activated by Ultrasound in a Piezo1 Activity-Dependent Manner

We next tested the effects of ultrasound on primary cortical neurons harvested from embryonic mice to test the feasibility of stimulating live neurons with ultrasound. We first probed whether primary neurons cultured using our protocol could express any functional Piezo1. Treating primary neurons at DIV 10 with 10 μM Yoda1 stimulated Ca2+ influx, whereas pre-treatment with 40 μM GsMTx-4 significantly reduced it (Figure 2A). The inhibitory effect of GsMTx-4 upon Yoda1-induced Ca2+ influx was consistent with the known mechanism of its action as a Piezo1 gating modifier (Bae et al., 2011, Gnanasambandam et al., 2017). Immunofluorescent staining revealed that primary neurons at day in vitro (DIV) 10 expressed Piezo1 endogenously (Figure 2B). Thus, primary neurons at DIV 10 expressed some functional Piezo1, and we proceeded to evaluate the effects of ultrasound on these cells. We also acutely isolated mouse cortical neurons from mouse pups at P3, to evaluate in vivo expression of Piezo1 in neurons of approximately equal maturity. These neurons influxed Ca2+ when stimulated with Yoda1, but this effect was almost completely abrogated in presence of the mechanosensitive ion channel antagonist Ruthenium red (Figure S2). We thus found that mature neurons, both in vitro and in vivo, expressed functional Piezo1 with detectable effects.

Figure 2

Ultrasound Induces Ca2+ Influx and c-Fos Expression in Primary Neurons in a Piezo1 Activity-Dependent Manner

(A) Functionality of Piezo1 in primary neurons was examined by Ca2+ imaging of cells stimulated with Piezo1 agonist Yoda1, including cells pre-treated with Piezo1 blocker GsMTx-4. Top: representative Ca2+ imaging time course for cells treated with either 10 μM Yoda1 alone or pre-treated with 40 μM GsMTx-4, respectively. Bar chart shows the mean ± SEM of three independent experiments. n = 15, ***p < 0.001, unpaired two-tailed t test.

(B) Piezo1 expression in primary cortical neurons was examined by immunocytochemical staining. Representative images of Piezo1 and MAP2 immunocytochemical staining of mouse primary cortical neurons at DIV 10.

(C) Ca2+ imaging of primary neurons stimulated with ultrasound, including cells pre-treated with GsMTx-4. Top: representative Ca2+ imaging time course of primary neurons treated with 0.45 MPa ultrasound or pre-treated with 40 μM GsMTx-4 and then with ultrasound. Bottom: bar chart represents mean ± SEM of three independent experiments treating primary neurons with varying intensities of ultrasound and GsMTx-4. n = 9, *p < 0.05, **p < 0.01, ***p < 0.001, one-way ANOVA with post-hoc Tukey test. All statistically significant differences are shown. See also Figure S2.

(D) An illustration of the ultrasound setup used to treat cells placed inside a cell culture incubator for immunofluorescence or Western blots.

(E) Left: representative IF images of c-Fos and MAP2 staining, in cells that were untreated, treated with 0.3 MPa ultrasound, or with 20 μM GsMTx-4 followed by ultrasound. Right: bar chart represents the mean ± SEM of three independent experiments. n = 3, *p < 0.05, **p < 0.01, one-way ANOVA with post-hoc Tukey test.

Ultrasound stimulation resulted in dose-dependent Ca2+ influx into the neurons, and this influx was abrogated when the cells were pre-treated with 40 μM GsMTx-4 (Figure 2C). Ultrasound at 0.3 MPa and above was able to induce significant Ca2+ influx, so we evaluated the effects of ultrasound on neuron activation by immunofluorescent staining of c-Fos, a well-established molecular marker of neuronal activation that is responsive to Ca2+ influx (Chaudhuri et al., 2000, Ghosh et al., 1994, Sheng and Greenberg, 1990). Untreated cells, cells treated with 0.3 MPa ultrasound for 20 min and cells pre-treated with 20 μM GsMTx-4 before ultrasound inside a standard cell culture incubator (setup illustrated in Figure 2D), were compared. We found that c-Fos expression in the nuclei of neurons (identified by MAP2 staining) significantly increased upon US treatment compared with the untreated control and reduced significantly with GsMTx-4 pre-treatment (Figure 2E). Hence, low-intensity ultrasound could activate primary neurons in vitro by opening the Piezo1 channel to allow Ca2+ influx.

Ultrasound Requires Piezo1 to Induce Calcium-Dependent Downstream Signaling in a Neuronal Cell Line

We were interested in exploring the signaling implications of Piezo1-mediated ultrasound effects on neurons in greater depth, to help elucidate the possible downstream effects of Ca2+ influx through ultrasound-activated Piezo1 channels. We chose the mouse hippocampal cell line mHippoE-18 (CLU199) as an in vitro representation of normal (non-cancerous) neuronal cells to enable better quantification of these effects. Expression of Piezo1 in CLU199 cells was confirmed through semi-quantitative RT-PCR, with HeLa cells as a positive control (McHugh et al., 2010) and Western blot (Figure 3A). To evaluate the treatment's effects on cell signaling downstream of the observed Ca2+ influx, we investigated levels of the phosphorylated (activated) forms of the Ca2+/calmodulin-dependent protein kinase type II (p-CaMKII) and the transcription factor CREB (p-CREB). Both these proteins are crucial players in neuronal Ca2+ signaling. CaMKII is directly regulated by calmodulin, a good indicator of the level of Ca2+ inside the cell, and both CaMKII and CREB are involved in critical neuronal functions such as neurotransmitter secretion, plasticity, transcription regulation, learning, and memory (Hardingham et al., 2001, Yamauchi, 2005). We observed the levels of these activated proteins, along with c-Fos, by Western blot to gauge whether ultrasound could affect aspects of neuronal function and activation downstream of Ca2+ influx. Ultrasound treatment inside an incubator for 20 min increased the levels of p-CaMKII, p-CREB, and c-Fos in a dose-dependent manner, with 0.3 and 0.5 MPa inducing significant increases (Figure 3B). We then evaluated Piezo1's contribution to these effects by siRNA knockdown. We were able to achieve over 50% knockdown of Piezo1 and found that this significantly reduced the 1 μM Yoda1-induced Ca2+ influx compared with cells treated with non-targeting siRNA (“Ctrl”) (Figure 3C). CLU199 cells with Piezo1 knockdown also displayed no significant upregulation of p-CaMKII, p-CREB, and c-Fos than the control when treated with 0.3 MPa ultrasound compared with the control (Figure 3D). Thus, we determined that ultrasound stimulation significantly affects the levels of important proteins in CLU199 neuronal cells, and these effects were significantly mediated by the Piezo1 channel.

Figure 3

Ultrasound's Ability to Initiate Calcium-Dependent Downstream Signaling in CLU199 Cells Is Dependent on Piezo1

(A) Levels of Piezo1 in a mouse neuronal cell line, CLU199, were evaluated in two ways. Left: PCR results of Piezo1 expression in multiple samples of CLU199 cells, with HeLa cells for comparison. Right: Western blot of Piezo1 expression in multiple samples of CLU199.

(B) Western blot for expression levels of p-CaMKII, p-CREB, and c-Fos, in CLU199 cells treated with varying ultrasound intensities. Left: representative Western blot images. Right: bar chart shows the levels of each protein as a fold change compared with the untreated control. Results are mean ± SEM of three independent experiments. n = 3, *p < 0.05, unpaired two-tailed t test.

(C) Piezo1 was knocked down in CLU199 cell using non-targeting (‘Ctrl’) or Piezo1 siRNA (“Piezo1 KD”). Left: qRT-PCR was performed for Piezo1, normalized to β-actin, and expressed as a fold change. Bar chart represents mean ± SEM of three independent experiments. n = 3, **p < 0.01, unpaired two-tailed t test. Also shown are representative IF images of Piezo1 staining in CLU199 cells. Middle: representative Ca2+ imaging time course for Ctrl and Piezo1 KD cells treated with Yoda1. Right: bar chart shows the mean ± SEM of three independent Ca2+ imaging experiments. n = 13, ***p < 0.001, unpaired two-tailed t test with Holm-Sidak correction.

(D) Western blot for levels of p-CaMKII, p-CREB, and c-Fos in CLU199 cells treated with siRNA and ultrasound. Left: representative Western blot images. Right: bar chart shows the levels of each protein as a fold change compared with the untreated control. Results are mean ± SEM of three independent experiments. n = 3, *p < 0.05, unpaired two-way ANOVA with post-hoc Tukey test.

Discussion

In the present study, we show that Piezo1 expressed in neurons is an important mediator of ultrasound's in vitro effects. We observed that heterologous Piezo1 expression in cells could enable them to respond to ultrasound (HEK293T). We showed that mouse neuronal cells express Piezo1 endogenously and that ultrasound alone can open these channels, allowing Ca2+ influx into them. We also demonstrated that activation of primary neurons by ultrasound in vitro can be significantly reduced by inhibiting Piezo1's activity. Our experiments used low-intensity low-frequency ultrasound, without microbubbles, allowing us to minimize the thermal and cavitation effects of ultrasound. We manipulated endogenous Piezo1 on a functional level, using an inhibitor drug, and at the expression level, using siRNA, and found that the effects of ultrasound on proteins crucially involved in neuronal activation and calcium signaling were significantly reduced in both cases. Blocking the activity of heterologous Piezo1 could also suppress ultrasound-induced Ca2+ influx. Thus, we see an important role played by Piezo1 activation in enabling neurons to respond to ultrasound stimulation.

In addition to increased c-Fos expression in primary neurons, our results in a neuronal cell line also showed that the expression of the activated forms of CaMKII and CREB can be increased by ultrasound stimulation, along with increasing c-Fos expression. These proteins are involved in complex neuronal functions such as neuronal plasticity, learning, and memory. Although results from a cell line must not be overinterpreted, previous studies of ultrasound have also shown increases of calmodulin-dependent kinases in primary neurons (Liu et al., 2017) and increases in the crucial neuroprotective protein BDNF in mouse and rat brains treated with ultrasound (Chen et al., 2018, Su et al., 2017, Tufail et al., 2010). The observed Piezo1-dependence of these effects in our experiments also helps to pinpoint their source to the ultrasound stimulation. Such results at the cellular level also provide evidence of the basis for non-auditory effects of ultrasound. Our results help to bolster the idea that ultrasound treatment can have therapeutic effects in the brain and could contribute to identifying which conditions could benefit most from such a treatment.

The obvious next step would be to demonstrate Piezo1-dependence of ultrasound effects in vivo, by upregulating or downregulating it in rodent brains. It will likely be challenging to upregulate Piezo1 in vivo, due to its size and complexity (Coste et al., 2010, Coste et al., 2015), but also to downregulate it, as knocking it out totally in mice embryos can be lethal (Li et al., 2015, Ranade et al., 2014). However, delivering a Piezo1 inhibitor or Piezo1 knockdown reagents in rodents is feasible and is a promising approach for further understanding the effects of ultrasound. Transgenic mice with conditional knockout of Piezo1 in the brain are also a viable option. Concurrently, research about the Piezo1 molecule itself could contribute to both evaluating and enhancing the role of brain-based Piezo1 in mediating ultrasound's effects. Studies comparing the different segments show some domains to be more mechanosensitive than other (Wu et al., 2016). This raises the possibility that the effects of ultrasound on such domains could be studied instead of the entire molecule, making it easier to tailor ultrasound parameters to activate Piezo1 more efficiently. Knowledge of this kind could also have significance in other fields, as Piezo1 plays crucial roles in other systems, such as the development of blood vessels (Li et al., 2015, Ranade et al., 2014) or T-cell activation (Liu et al., 2018). Thus, understanding ultrasound's mechanism of action on Piezo1 could have implications for usefully modulating neuronal activity, as well as that of a host of other cell types.

Limitations of the Study

Although we aimed to demonstrate that ultrasound opens the Piezo1 channel, we could not use the standard method of patch clamping to determine this. This is because the recording pipette in our system was found to be incompatible with low-frequency ultrasound, a phenomenon also reported elsewhere (Prieto et al., 2018). We thus used intracellular ratiometric Ca2+ imaging to measure the ion influx resulting from ultrasound opening Piezo1, which helped us work around the vibration issue. It is also important to note that the cells used in this study were cultured on lysine-coated glass, a hard substrate, as this can complicate data interpretation. Ultrasound could cause vibrations in the base of the dish, amplifying the observed effects (Kubanek et al., 2016), leading to an underestimation of the acoustic pressure required to activate Piezo1. Secondly, neurons develop divergently depending upon the stiffness of their surroundings (Chang et al., 2017, Koser et al., 2016), and substrate stiffness is also reported to affect Piezo1's activity (Pathak et al., 2014). A possible solution would be to coat plates with a soft gel that mimics the mechanical properties of brain tissue. However, there is no standard method to achieve this, which is why we opted to maintain a simple culture protocol to avoid unnecessary complications. An aim of future studies ought to be to solve such issues in vitro, to gain more physiologically relevant data.

Finally, we did not examine the role of other mechanosensitive ion channels that may be involved in mediating the cellular response to ultrasound, because we were primarily interested in whether low-intensity ultrasound stimulation could activate our chosen Piezo1 channel. For all three issues, in the context of the present study, we based our conclusions upon comparison to controls, with lower Piezo1 expression or with a Piezo1 blocker, which were also exposed to ultrasound but showed significantly lower (but not abrogated) responses. We thus argue that Piezo1 mediated ultrasound's effects in our cells to a considerable degree, a finding that could be adapted and applied to in vivo conditions as appropriate. However, future studies with in vivo Piezo1 knockout/knockdown in brains and using sensitive genetically encoded Ca2+ or voltage sensors for real-time tracking of ultrasound's effects will be crucial to truly establish the role of channels as Piezo1 in this context.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.