Sonomechanobiology: Vibrational stimulation of cells and its therapeutic implications

Tissue, bone, and organs continuously grow and remodel as part of the natural process by which living organisms develop and adapt to their environment and to therapeutic intervention.¹ Central to these processes are cellular level responses to external physical, chemical, and biological cues. Although the role of mechanical forces constituting diverse physical cues that cells respond to in organogenesis has long been known,^1–3 it has only been of late that the mechanostimulatory effects on physiological and pathological conditions have been increasingly studied, given their therapeutic implications, such as for nanomedicine and regenerative medicine.^2,3

The mechanisms by which cells sense these mechanical forces and translate such cues into a cascade of cellular and molecular signaling events that ultimately determine their biological response and fate is, however, extremely complex.^2–4 The majority of investigations in this area have primarily employed the application of static forms of mechanical forcing, be it compressional, tensile, or shear forcing,^4,5 in which the applied loads were predominantly static, with the exception of a few “dynamic” studies,⁵ although these cases only involved mechanical forcing at very low repetition frequencies (⁠ $≤ 1$ Hz) such that the loads are essentially at constant force and the cells' mechanoresponse do not differ significantly from that observed for the steady, static cases.

Cells have nevertheless also been known to respond to vibration in which the applied force is time-dependent and essentially periodic, although this again has mostly been conducted at oscillation frequencies of several Hz—understandably given that this is the characteristic frequency associated with typical human or organism locomotion. More recently, though, it has been shown that more complex dynamic mechanical forcing in the form of acoustically driven vibrational stimuli, both at audible (20 Hz–20 kHz) and, curiously, even ultrasonic (20 kHz–1 GHz) and hypersonic (⁠ $> 1$ GHz), frequencies can also invoke a variety of cellular responses.

Existing reviews on mechanotransduction—the cellular process by which mechanical stimuli to the cell are transducted into biochemical signals^6–8—have not just largely focused on static mechanical forcing but have also been primarily confined to their effects, in particular, on sensory cells. While this is because such cells function specifically to detect mechanical cues and are hence ideal models to study these effects,⁸ they do not capture the myriad of effects observed when other types of cells are mechanostimulated. There also generally appears to be a patchwork of investigations on different types of mechanostimuli with little attempt to understand the similarities or differences between the cellular responses and the underlying mechanotransduction mechanisms between them.

We, therefore, attempt to address this by providing a systematic perspective on the various cellular mechanostimulation strategies employing sound wave vibration that have been examined to date, together with their associated mechanisms, phenomena, and applications (touched on briefly by Rufo et al.⁹) particularly since this is considerably more complex and relatively less understood compared to their conventional mechanostimulation counterparts involving steady, constant-force compressional, tensile, or shear forcing. We begin in Sec. II by briefly revisiting the principles that underpin mechanotransduction. We then proceed to summarize the different forms of mechanostimuli that have been employed, beginning first with a brief overview of static constant-force mechanical loading in Sec. III, followed by a more extensive treatise on dynamic (periodic) mechanostimulation driven by sound wave vibration over a wide range of frequencies and the signaling pathways they induce in Sec. IV. We subsequently discuss their various applications, particularly for, albeit not limited to, nanomedicine and regenerative medicine (namely, cell-free and cell-based therapies, and cell/tissue reorganization) in Sec. V and show in Sec. VI that it is possible to adopt a unified mechanistic framework centered around membrane-tension-driven second messenger calcium-induced signaling that holistically captures the way different cells perceive various forms of acoustically driven mechanostimuli irrespective of the applied frequency and intensity.

Mechanotransduction—in which cells respond to mechanical stimuli over millisecond timescales¹⁰ associated with the time it takes for the cues to propagate to the nucleus^11,12 where they are converted into autocrine, paracrine, or endocrine signals, which, in turn, trigger downstream local or global events that maintain homeostasis or direct cellular fate—is a particularly complex phenomenon. At a fundamental level, the mechanotransduction process can be viewed as comprising three distinct stages involving mechanosensing, mechanotransmission, and mechanoresponse steps (Fig. 1). In simpler models, these are considered to be in series, although more complex descriptions allow for dynamic coupling between the stages to account for regulation of the signaling pathways by the applied mechanical load.¹³ 

There are two categories of mechanosensors that facilitate cellular perception of a force applied to it through a change in either their molecular assembly or a conformational change in its protein structure: passive (outside–in), wherein the mechanosensing element detects an external force imparted on it and transmits the signal into the cell, or, active (inside–out), in which forces are generated by the mechanosensing element to detect changes in the extracellular matrix (ECM).^6,14 Various mechanosensory elements exist in the form of proteins embedded at the cell–ECM interface that are sensitive to membrane stiffness changes [e.g., focal adhesions (FAs), membrane receptor proteins such as G-protein coupled receptors, adherens junctions (AJs), and stress activated channels (SACs)], all of which are involved in either or both passive or active sensing.

FA protein complexes under the cell membrane, for example, play a role in both forms of mechanosensing by constituting the bridge between the ECM and the intracellular domain. This connectivity occurs through the recruitment of transmembrane reporters known as integrins, whose intracellular tails are connected to the actin cytoskeleton by resident adaptor proteins, such as vinculin and talin, and whose extracellular head domains bind to proteins in the ECM⁶—the latter being regulated by the mechanical stimuli, which induces a conformational change, i.e., unbending of integrins that result in the exposure of the binding sites.¹⁵ 

SACs or mechanosensitive ion channels, which facilitate gating of the ion transport in and out of the cell, constitute another dominant type of mechanosensor,¹⁶ especially given the role of calcium (Ca²⁺) as a key second messenger and signaling molecule that is critical to the dynamic triggering of downstream transcriptomic changes;¹⁷ intracellular Ca²⁺ concentrations in particular also play a central role in regulating a wide range of cellular processes.¹⁷ Two different models exist to describe the gating mechanism by which SACs operate (Fig. 2). In the stretch model [also known as the force-from-lipid (FFL) model¹⁸], tension imposed by any mechanical perturbation to the cell membrane leads to the reorganization of the lipids in the bilayer around the transmembrane SACs as well as a conformal change in its proteins that results in the opening of the channel.¹⁹ The tether, or force-from-filament (FFF), model, on the other hand, suggests that the SACs are physically connected directly to the ECM or indirectly through intracellular accessories such as the cytosekeleton.²⁰ These tethers then interact directly with channel or auxiliary proteins, indirectly imposing tension on the channel and allowing for long-range propagation of membrane tension, therefore facilitating long-distance sensing via the cytoskeletal structure.^20–22 

In eukaryotes, four main types of SACs have been identified: two-pore-domain (K2P) potassium channels, degenerin–epithelial sodium–acid-sensing ion channels (DEG–ENaC–ASIC), transient receptor potential (TRP) channels, and piezo channels.²³ The K2P family of ion channels consists of 15 members (KNCK genes), which can be distributed within six subfamilies: TWIK (tandem of pore domains in a weak inward rectifying K⁺ channel), TREK (TWIK related K⁺ channels), TASK (TWIK related acid-sensitive K⁺ channels), TALK (TWIK-related alkaline pH-activated K⁺ channels), THIK (tandem pore domain halothane inhibited K⁺ channels), and TRESK (TWIK related spinal cord K⁺ channel).²⁴ DEG–ENaC–ASIC channels in vertebrates comprise two subgroups: ENaC and ASIC; while ENaC is known to be sensitive to osmotic swelling, shrinking, and shear stress, the mechanosensitivity of ASIC has not as yet been fully explored.²⁵ The TRP channel family, on the other hand, consists of 28 channels that are grouped into 6 subfamilies, namely, canonical TRP (TRPC), vanilloid TRP (TRPV), melastatin-related TRP (TRPM), ankyrin TRP (TRPA), mucolipin TRP (TRPML), and polycystic TRP (TRPP).^26,27

The gating mechanisms involved in all of these ion channels, and, in particular, the DEG–ENaC–ASIC channels, are not completely understood.²⁸ However, it is likely that membrane tension plays a key role in controlling the opening and closing of these channels, although we note that the factors initiating the gating through membrane tension varies between each subfamily. For instance, TREK and TRAAK channels were found to be sensitive to shear stress, negative membrane pressure, and temperature,²⁹ whereas TRP channels are known to be insensitive to membrane tension induced by cell membrane stretching.³⁰ The TRPV, TRPM, and TRPA subfamilies, in contrast, are thermally activated, while exocytosis was found to activate the TRPC, TRPV, and TRPM channels.²⁷ 

Piezo channels—vertebrate mechanically activated nonselective cation channels—are a more recently discovered member of the SACs family,³¹ whose role has been widely implicated in mechanosensing. Whereas Piezo1 initiates Ca²⁺ signals predominantly in various non-excitable cells,³¹ Piezo2 mainly functions in sensory neurons and specialized cell types, such as tactile epithelial cells (e.g., Merkel cells³¹). Piezo channels can be activated by membrane tension (force-from-lipid), indicating that any physiological force that can cause indentation of the cell membrane on the apical surface-lumen facing side and hence alter the membrane tension, such as that produced with a blunt glass probe or suction to the membrane, can robustly activate the piezo channel.³² It has also been noted that mechanical perturbations imposed onto the basolateral surface of the membrane, for instance, those at the cell–cell and cell–matrix interfaces produced by stretching cells seeded on tensile substrates or on elastomeric pillar arrays, or by compressing the membrane using an atomic force microscopy (AFM) probe, can also activate these channels, thereby alluding to the key role that piezo channels play in sensing of the microenvironment.³³ 

Both localized and whole-cell responses can be mediated through Piezo1 regardless of the endogenous or exogenous origin of the mechanical stimulus.³⁴ As long-range propagation of membrane tension in intact cell membranes is limited, long-distance sensing is enabled through biochemical and functional tethering of the piezo channel to the actin cytoskeleton via the mechanotransduction complex comprising the AJ complex (cadherin and $β$-catenin) and cytoskeletal-associated proteins (vinculin), which allow complementary force-from-filament gating via the cytoskeletal network.³⁴ 

Upon their activation, piezo channels generate cationic nonselective currents; the dominant cation that arises is however dependent on the cell type.³⁵ For example, piezo channel activation in cells such as sensory neurons leads to membrane depolarization owing to Na⁺ influx, with Ca²⁺ playing a minor role,³⁶ whereas Ca²⁺-induced signaling cascades are elicited in endothelial cells (ECs)³⁷ and mesenchymal stem cells (MSCs).³⁸ As such, piezo channels have been recognized to possess different roles along diverse mechanotransduction pathways depending on the physiological environment that they are exposed to.

A hierarchy of load-bearing cytoskeletal structures, such as F-actin, intermediate filaments, and microtubules, are critical for continuous propagation of the force across the entire length of the cell, either from within it, i.e., the nucleus, through the perinuclear space to the ECM, or vice versa. Such transmission occurs over time scales on the order of a tenth of a second¹⁰ and is regulated by a “molecular clutch” mechanism in which the size of the FAs along which the signals are transmitted are dependent on the applied force.³⁹ Key interactions implicated in the transmission of force across the cell include myosin-induced actin contraction and molecular bond (e.g., actin–adaptor proteins, adaptor protein–integrin, or integrin–ECM) formation and dissociation;¹³ the rate of the latter determining the efficiency of transmission.¹³ 

The final mechanotransduction stage, which involves processes such as calcium signaling, phosphorylation cascades, and DNA transcription factor binding encompasses the ability of the cell to transcribe the detected and transmitted mechanical cues into complex transcriptomic and signaling events. It is quite typical that the entire mechanotransduction process comprises a transcription feedback loop (Fig. 3) in which the cells' response is often transmitted back to the mechanosensitive structures that first produced the response.^13,40

Unlike mechanosensing and mechanotransmission, which occur rapidly on the order of milliseconds, mechanoresponses take place over significantly longer time scales: the activation of signaling pathways, for example, occurs immediately over minutes, whereas the induction of gene expression pathways is typically delayed and occurs over hours or days.⁴¹ Depending on the magnitude and duration of the imposed mechanical cue, these delayed responses can lead to transient changes in vesicle trafficking or cellular morphology, permanent compositional or structural alterations in adhesion complexes and cytoskeletal structures, and/or the triggering of secondary signaling cascades that facilitate cell migration or proliferation, cancer progression, angiogenesis, or differentiation.^3,42 Key mechanoresponses are depicted in Fig. 4.

As alluded to in Sec. II, cells have the ability to sense and respond to both exogeneous as well as endogenous forces. The literature to date has therefore found it convenient to broadly delineate these into extrinsic and intrinsic stimuli—the former constituting external forces physically exerted on the cell, and the latter being the cues “felt” by the cell from its surrounding environment, namely, the ECM. Both are known to regulate cell function. In reality, however, both extrinsic and intrinsic cues are intricately interrelated since mechanotransduction is essentially bidirectional, wherein nuclear mechanoresponses are transmitted back to the mechanosensors and cytoskeletal structures at the cell–ECM interface to give rise to a feedback loop, as discussed above. As such, the cell response to intrinsic stimuli cannot, therefore, be easily or simply decoupled from its extrinsic counterpart. Nevertheless, such a categorization provides a simple and convenient framework to understand the effects of different stimuli on cell fate.

Given our focus in this review on external mechanostimulatory cues, and, in particular, those arising through acoustic forcing, we refer the reader to the excellent articles that provide extensive commentary on intrinsic stimulations.^40,43–46 Briefly, intrinsic stimulation in the cell can arise due to changes to its shape and density, ECM, or substrate topology.⁴⁰ Altering the cell shape and density, both of which are interrelated, has, for example, long been known to affect cell spreading and proliferation: densely seeded cells tend to assume a rounder shape, whereas low seeding densities allow cells to spread and grow.^47,48 In stem cells, this was reported to promote osteogenesis over adipogenesis.⁴⁹ Restricting the cell shape through geometrical confinement or through the introduction of topology, for instance, through micropatterning of the substrate on which the cells are grown, on the other hand, can lead to apoptosis.^50,51 Cues arising from changes to the mechanical property of the ECM (e.g., stiffness and viscoelasticity), on the other hand, are detected through the FA adhesion complex that anchors the cell to ECM proteins;⁵² the accompanying conformation change in the integrin-binding domains (Sec. II A) is relayed to the cytoskeletal complex, triggering a signaling cascade involving the Rho family of small GTPases (RhoA, Rac and Cdc42) that are responsible for cytoskeletal organization. These, along with their downstream effectors, Rho-associated protein kinase (ROCK), myosin light chain (MLC) kinase and yes-associated protein (YAP), or transcriptional coactivator with PDZ binding motifs (TAZ), are known to induce cell migration and cell cycle progression apoptosis, or to direct stem cell differentiation.⁴⁰ 

For extrinsic mechanostimulation, the bulk of work carried out to date has primarily involved the application of static or steady forms of mechanical loading under compression, tension, or shear (Fig. 5). Periodic forcing, typically in the form of oscillatory vibrations or sound waves, has only been more recently explored and constitutes the subject of the present review. We nevertheless first provide here an overview of the existing work involving constant-force mechanical loading (Table I) to provide the reader with a contextual basis toward a better understanding of the more complex dynamic nature of their time-dependent (periodic forcing) counterpart to be discussed subsequently.

In studies to date, cells have generally been subjected to two different forms of compressive loading: hydrostatic compression/pressure and direct uniaxial compression.^95,96 In the former, compression is applied to the media, and hence, the applied pressure on the cells is isotropic, whereas directly compressing the cell-laden scaffold uniaxially in the latter maintains the applied force in a specific direction. Unlike direct compression, hydrostatic compression/pressure can initiate volumetric membrane deformation without shear induced deformation.⁹⁷ Yet, despite this difference, both forms of compression appear to induce similar mechanosensing and mechanotransmission mechanisms involving integrins, FAs, SACs, and Ca²⁺-signaling. In particular, compression induces distortion of cell surface receptors and integrins, through which the signal is transferred along the FA complex via the involvement of both vinculin and mature tight junctions.^53,98 Additionally, mechanosensing through SACs such as piezo and TRP channels has also been implicated in both forms of compression to drive an influx of Ca²⁺ in the cell, and, in some cases, activation of actin-related protein (ARP2/3) complexes, which, in turn, mediates actin remodeling within the cell.^54,55

Compression-induced downstream signaling is, however, not well understood to date,⁵⁷ although YAP and TAZ signaling are among suggested downstream cascades induced by compressive mechanostimulation,⁵⁷ which, in turn, affect the cells' ability to proliferate and undergo apoptosis, although these effects are largely dependent on the cell type.^97,99–103 Additionally, compressive loads along with biochemical factors (interleukin-6) have also been shown to promote ECM remodeling while mediating epithelial-to-mesenchymal transition.⁵⁶ In adipocytes, compression was observed to trigger Wnt/ $β$-catenin signaling that, in turn, induces dedifferentiation of the cells, resulting in a distinct transcriptome profile, long-term self-renewal, and serial clonogenicity (including osteogenesis, adipogenesis, myogenesis, chondrogenesis, and myofibrogenesis).¹⁰⁴ 

Numerous techniques have been developed to mechanically stretch a cell. These differ primarily based on the type of actuation scheme used to generate the tensile load.¹⁰⁵ Tensile stretching has been reported to influence cell polarizaion/migration, morphology, proliferation, lineage commitment, and differentiation.^106,107 The exact response that is observed, however, depends on both cell type¹⁰⁸ as well as the mode by which the stretch is generated, in addition to the properties of the ECM and the presence of soluble factors.^109–112 Major stretch-induced mechanotransduction signaling elements include integrins, force-sensitive kinases and proteins, cytoskeletal elements,⁶⁶ and SACs,^60–62 although the exact mechanism by which stretch-induced mechanotransduction occurs has not yet been clearly elucidated due to the complexity of the various processes and their dependence on a number of different parameters.^{58,59,63–65} In all cases though, the stretched-induced stimuli are relayed into the cells through a change in intracellular Ca²⁺ signaling. In particular, it was reported that mechanically stretched cells exhibited intracellular Ca²⁺ oscillations through SAC activation, which is manifested in F-actin assembly, actomyosin contractility, and associated Rho–ROCK signaling;⁶⁶ the downstream effector of these being the mitogen-activated protein kinase (MAPK) and extracellular signal-regulated kinase (ERK)-induced signaling cascades. Similarly, various stretch-induced signaling cascades have been found to initiate osteoblast differentiation and regulation of morphogenesis, angiogenesis, heart remodeling, and neuritogenesis.^113–115 

There have been a number of studies on the mechanotransductive effect of shear given its prevalence in physiological conditions where cells in the endothelium are constantly subjected to such stresses arising as a consequence of vasculatory flow. Studies employing flow channels to investigate the response of ECs under flow over a range of exposure durations from seconds to hours have now also been extended toward smaller scales to mimic conditions in the microvasculature with more recent advances in microfluidics.¹¹⁶ 

At the fundamental conceptual level, shear stress involves the activation of various intracellular and extracellular mechanosensors, such as transmembrane integrins,^62,67 ion channels and SACs (in particular, TRP channels),^62,68 G-proteins and their receptors,⁷⁰ tyrosine kinase receptors,^71,72 lipid rafts,^73,74 and primary cilia,⁷⁵ among others. More specifically, shear stresses and their gradients have been reported to upregulate $α$_v with $β$ integrin subunits^117–119 and their association with FAs, force–modulated through lipid raft diffusion,⁶⁹ and with the adaptor protein Shc.¹¹⁸ Shear flow is also known to drive upregulation in GJs, as well as the activation of TRP channels, to allow modulation of cations, particularly Ca²⁺,^76,77 which, in turn, results in the triggering of important downstream signaling pathways. These, depending on cell type, include, although are not limited to, the Rho–ROCK pathway in osteoblasts and osteocytes,^78,79 the ERK1/2 MAPK pathway in stem cells^80,81 and ECs,^120,121 or the Wnt/ $β$-catenin pathway in ECs and fibroblasts,^82,83 all of which play a role in modulating multiple cellular activities.^122–128 

Unlike the static forms of steady mechanical loading discussed in Sec. III, acoustic forcing provides an alternative and versatile means for cell stimulation. In addition to the ability of harnessing different forms of sound waves, the periodic nature of the vibrational forcing introduces further possibilities for dynamic stimulation that can be tuned over a broad range of applied frequencies, as will be discussed below.

Sound, in essence, is a mechanical wave generated by a vibrating surface or object. It mainly propagates through an elastic media as longitudinal (pressure; vibration displacement parallel to the direction of wave propagation) waves that involve the compression and rarefaction of the molecules in that medium. As it is possible to support vibrations in other directions in solids, transverse (shear) waves can also exist where the vibration displacement is transverse to the direction of wave propagation.

The sound wavelength $λ$ is related to the frequency f by the speed at which the sound wave propagates through the medium c, which is dependent on its density and elasticity. Given this relationship and the broad spectrum of sound frequencies—ranging from infrasound (⁠ $< 20$ Hz), audible (20 Hz–20 kHz), ultrasound (20 kHz–1 GHz) to hypersound (⁠ $> 1$ GHz)—together with the different configurations of sound wave generation devices, different wave modes—i.e., the different ways sound waves can propagate through the media—can arise.^129,130

Bulk acoustic waves [BAWs; Fig. 6(a)] do not only take the form of longitudinal (pressure) or transverse (waves). Thin solid sheets with thicknesses $h < λ$ (or $h / λ < 1$⁠), for example, can support plate waves that propagate parallel to its surface and through the thickness of the material. If the plate is infinitely wide, only a thickness mode exists, whereas a plate with finite width gives rise to symmetric (extensional) or asymmetric (flexural) Lamb waves that comprise both thickness and width modes. Surface acoustic waves [SAWs; Fig. 6(c)], which are also known as Rayleigh waves, in contrast, occur in piezoelectric substrates whose thicknesses are much greater than the acoustic wavelength, i.e., $h > λ$ or $h / λ$ > 1, and supports a combination of both longitudinal and transverse waves. A hybrid surface and bulk wave known as surface reflected bulk waves [SRBWs or pseudo-SAWs; Fig. 6(b)], on the other hand, can also exist in the intermediate transition regime where $h ≈ λ$ or $h / λ ≈$ 1.¹³¹ 

In practice, the coupling of sound waves to laboratory cell cultureware can be achieved with the use of a transducer. At low frequencies in the infrasound and audible range, longitudinal bulk acoustic waves can be induced in a culture chamber by coupling the piston-like vibration generated with a conventional sound transducer akin to that found in loudspeakers. To generate BAWs at ultrasound frequencies up to several MHz, piezoceramic transducers are typically used, on which electrode pads are patterned [alternatively, higher frequency (MHz) BAWs can also be generated on piezoelectric substrates, as illustrated in Fig. 6(a) ¹³²]. The thickness of the transducer or substrate then determines its resonant frequency, and hence, the frequency of the bulk waves that are produced. While it is possible to generate bulk vibration at higher frequencies beyond several MHz, this requires considerably thinner resonant structures or typically thin piezoelectric film resonators with thicknesses d that are equal to half the resonant wavelength $λ$ (i.e., the resonant frequency $f = c / 2 d$⁠, wherein c is the sound speed in the piezoelectric film) mounted atop either an air gap (thin film bulk acoustic resonators, or FBAR) or multiple reflection layers (solid mounted resonators, or SMR).¹³³ 

For SAWs and SRBWs, a chipscale piezoelectric substrate is employed, on which interdigitated transducers (IDTs)—electrodes patterned in an interleaved pattern—are photolithographically patterned, whose gap and spacing determines the resonant frequency of the device and hence the wavelength and frequency of the SAW or SRBW that propagates along the substrate [Figs. 6(b) and 6(c)]. Whether an SRBW [Fig. 6(b)] or SAW [Fig. 6(c)] is generated depends on this resonant frequency and hence wavelength relative to the substrate thickness.¹³¹ 

There has only been limited work to date on mechanostimulation using bulk waves at infrasound (⁠ $< 20$ Hz) and audible (20 Hz–20 kHz) frequencies. This is likely because of the relative ease by which it is possible to generate other mechanical stimuli at Hz-order frequencies (which are the frequencies typically experienced by cells in their local physiological environment during human motion, e.g., walking and running), for example, using oscillatory versions of mechanical forcing such as the laminar shear or stretching discussed in Sec. III. Similarly, it is relatively simple to excite cells at ultrasonic frequencies at tens to hundreds of kHz given decades of progress in ultrasound technology, borrowing from the extensive advances from imaging and diagnostic and imaging modalities, which will be discussed in Sec. IV B.

Most of the work on acoustic cell stimuli in the infrasound and audible ranges has primarily been limited to cell proliferation,^134–138 cell migration,^138,139 and even apoptosis.⁸⁵ Neuronal degeneration has also been reported in astrocytes, neurons, and cardiomyocytes exposed to 16 Hz infrasound,^85,140 whereas upregulation in neuronal differentiation was observed with 1 kHz mechanostimulated bone marrow-derived stem cells (BMD-SCs),¹⁴¹ although we note that the latter appeared to be dependent on the inclusion of specific biochemical additives, such as dexamethasone, ascorbate, $β$-glycerophosphate, bone morphogenetic protein (BMP), 3-isobutyl-1-methylxanthine, insulin, transferrin, valproic acid, forskolin, neural supplement and human transforming growth factor $β$1, among others.¹⁴² 

It should be noted though that even for acoustic stimulation in the same frequency range, the mechanoresponse varies considerably between cell types. NIH/3T3 mouse immortalized fibroblasts and NB2a mouse neuroblastoma cells, for example, were found to be insensitive to audible sound wave excitation over the range 55 Hz–4 kHz, whereas genes such as prostaglandin-endoperoxide synthase 2 (Ptgs2)—an osteoblastic differentiation gene, connective tissue growth factor (CTGF), and ECM protein tenascin-C (TNC) were observed in ST2 (mouse stromal) and C2C12 (mouse immortalized myoblast) cells, indicating that while normal cells may be responsive to acoustic mechanostimulation, cancer cells, and immortalized cells (depending on the tissue source) are resistant to bulk vibrational stimuli, at least in the audible range.¹⁴³ 

In any case, the majority of these studies do not explicate the mechanotransductive mechanisms responsible for these responses, although some have noted the involvement of intracellular Ca²⁺.^85,141 More specifically, upregulation in TRPV4 (transient receptor potential cation channel subfamily V member 4) expression has been observed, which led to an increase in intracellular Ca²⁺ concentration in cells exposed to infrasound at 16 Hz for more than 7 days.⁸⁵ Increases in intracellular Ca²⁺ concentration in astrocytes at this frequency have also been attributed to the opening of GJ proteins (connexin-43).¹³⁶ Such Ca²⁺-induced signal transduction has been noted to activate the calmodulin and protein kinase C pathways, resulting in the release of glutamate¹³⁶ or pro-inflammatory cytokines, both of which result in neuronal death.^140,144 Similarly, ryanodine receptor (RyR)-induced Ca²⁺ signaling has also been reported in non-auditory cells such as MSCs when stimulated by audible sound waves (1 kHz),¹⁴¹ leading to the alteration of the cytoskeletal structure that, in turn, initiated a downstream chain of mechanotransductive events,^139,143 primarily those related to intracellular Ca²⁺-induced signaling cascades, such as the ERK pathway.¹⁴¹ 

The main bulk of the studies on vibrationally induced mechanotransduction has been carried out using bulk sound wave excitation at tens to hundreds of kHz. For convenience, we have opted to use the terminology “low frequency ultrasound” to refer to sound wave excitation at frequencies in this range, specifically from 20 kHz to approximately 3 MHz, which is typically the upper limit at which BAWs at ultrasonic frequencies can be generated.¹⁴⁵ We have done so to distinguish this from the “high frequency ultrasound” mechanostimulation work carried out above this frequency range from several MHz up to 1 GHz that will be discussed in Sec. IV C given the considerable mechanistic differences that arise between them. Not least, high frequency ultrasonic excitation, in contrast, typically comprises surface or hybrid modes (SAWs or SRBWs), as introduced in the preamble of this section.

In the medical ultrasound imaging literature, low frequency ultrasonic bulk wave excitation has commonly been viewed to interact with biological matter through thermal and non-thermal effects.¹⁴⁶ At the cellular level, thermal effects, which arise as a consequence of the dissipation of sound energy as heat upon its absorption in tissue, can be extremely detrimental to cells, which typically denature above 40 °C cells due to hyperthermia.^147–150 Stem cells are even more sensitive to thermal effects and suffer from heat shock with temperature rises of merely 4 °C (even within the physiological range).¹⁵¹ Non-lethal thermal effects, on the other hand, include the release of heat shock proteins,¹⁵² and altered protein conformation, leading to misfolding or nonspecific aggregation of proteins, which result in cell cycle arrest or impaired growth rates.¹⁴⁷ Further, temperature variations can give rise to changes in the cell membrane tension, that, in turn, triggers the activation of ion channels such as TRP (particularly TRPV)²⁷ and K2P channels (TREK and TRAAK),²⁹ which alters the intracellular Ca²⁺ concentration and potentially instigates various downstream signaling cascades.¹⁵² As such, it is extremely important to control the intensity as well as the duration of the mechanostimulation to minimize these adverse effects on cells, which can also include poor reseeding viabilities and colony-forming ability,¹⁵³ although we acknowledge that a moderate increase in temperature has, in certain circumstances, been known to induce thermal mechanotransductive responses.¹⁵² 

Non-thermal effects, on the other hand, arise as a consequence of (1) the acoustic radiation pressure that is imposed on a cell when the sound waves encounter it along their propagation pathway; (2) the normal and shear stresses experienced by a cell due to the fluid flow that is generated by the sound wave propagation (known as “acoustic streaming”) through the cell media; and (3) the extreme pressures and temperatures that the cell is subject to during cavitation that typically accompanies low frequency ultrasonic excitation.¹⁴⁶ 

The acoustic radiation pressure is a steady, time-averaged normal force that is imparted on a cell as a consequence of its presence in an acoustic field. More specifically, it consists of the transfer of momentum from the sound wave to the cell when the propagating wave encounters it. The magnitude of the acoustic radiation pressure depends on the sound wave intensity, the cell dimension, the wave characteristics (wavelength, sound speed through the medium), and an acoustic contrast factor, which accounts for the relative densities and compressibilities between the medium and the cell. For the most part, cells, being of a dimension much smaller than typical sound wavelengths, at least in the low frequency ultrasound range below 1 MHz, tend to either be forced to translate under the influence of the acoustic radiation pressure in what is known as “acoustophoresis.”¹⁵⁴ In a standing wave field, which typically arises due to reflections and hence superpositioning of waves along the confines of the cell chamber, the cells then end up being trapped at the nodes or antinodes of the standing wave.¹⁵⁵ At high intensities, though, it is nevertheless possible for cells to fragment under the extreme mechanical stresses exerted on them by the acoustic radiation pressure.¹⁵⁶ 

Acoustic streaming, on the other hand, comprises the flow, or momentum flux, that arises from pressure and velocity fluctuations that are generated in the liquid media due to viscous dissipation of the energy of the sound wave upon its attenuation as it propagates and is absorbed within the liquid. Again, at low ultrasonic frequencies, acoustic streaming mainly results in the translation of the cells along flow streamlines as a consequence of the viscous drag imposed on the cells. Other effects of either or both acoustic radiation pressure and acoustic streaming on cells can typically arise at much higher frequencies when the sound wavelength becomes comparable to the cell dimension, which we discuss in Sec. IV C.

Among these effects, however, cavitation—the rapid growth, oscillation, and implosion of a bubble or an ensemble of bubbles beyond a critical sound wave intensity (Fig. 7)—by far, constitutes the central feature of low frequency ultrasonic mechanostimulation. Not only does the violent event associated with the rapid collapse of the bubble(s) lead to the formation of shock waves¹⁵⁷ and high velocity microjets that induce intense local shear within the vicinity,¹⁵⁸ extreme pressures (⁠ $> 1$ GPa), and temperatures (⁠ $> 10 4$ K) are also generated which can be extremely detrimental to cells.^147,159,160 Additionally, cells are known to suffer from post-cavitation DNA damage when exposed to the free radicals that are typically produced during cavitation that lead to the production of reactive oxygen species (ROS).¹⁶¹ 

If moderate cell viabilities can be maintained at lower sound intensities, particularly at the higher end of the low frequency ultrasound range, whatever sufficient deformation experienced by the cell due to any of the aforementioned effects can potentially give rise to various mechanically activated cell signaling pathways.^149,150 While the exact mechanism by which such mechanotransduction arises is as yet unclear, the deformation of the cell membrane likely plays a key role in the mechanosensory process. Pore formation in the cell membrane (sonoporation), which is commonly observed in low frequency ultrasonic excitation particularly as a consequence of cavitation,^162,163 and which constitutes the dominant mechanism that underpins ultrasound-induced intracellular delivery,¹⁶⁴ is expected to alter the membrane curvature, which, in turn, can activate SACs, such as the TRP and piezo channels.^165–167 Additionally, transmembrane receptors such as integrins (⁠ $α$ and $β$⁠) in many cells (including fibroblasts, osteoblasts, and chondrocytes)^168–170 and FAs^169,171 have been reported to alter with low frequency acoustic stimulation.^89,172

These effects, and, in some cases, even cavitation-induced ROS production,^162,173,174 have been claimed to alter intracellular Ca²⁺ concentration, leading to cytoskeletal reorganization both with or without FAK activation,^175,176 while integrin activation is suggested to trigger the PI3K/Akt (phosphatidylinositol-3-kinase/serine–threonine kinase) pathway.¹⁶⁹ In addition, low frequency ultrasonic mechanostimulation has also been reported to regulate the ERK/MAPK,^168,170,177 nuclear factor kappa B (NF- $κ$B),^178,179 Rho–ROCK/ERK,¹⁷⁷ and Wnt/ $β$-catenin pathways,^91,180 in addition to calcineurin–NFAT (nuclear factor for activated T cells) activation¹⁷⁹ and YAP nuclear translocation in MSCs.¹⁸¹ In ECs, regulation of expression of vascular endothelial growth factor (VEGF) and interleukin-8 (IL-8) has been reported.^182,183

The translation of these signaling pathways can be observed in various downstream effects and cell fate in response to the low frequency ultrasonic mechanostimulation, such as MSC proliferation and upregulation of many cyclins,^180,184 along with improvement in cell migration and an increase in adhesion molecules such as CXCR4, integrin‐1 $β$⁠, and CCR2.^172,185 Additionally, upregulation in Nanog expression has also been observed with such stimuli to thereby maintain the stemness of mouse primary osteoblasts and mesenchymal stromal stem cells.¹⁸⁶ Depending on the culture conditions (e.g., the presence of biochemicals), low frequency ultrasound stimulation of multipotent stem cells can promote osteogenesis,¹⁸⁷ chondrogenesis,¹⁸⁸ neurogenesis,^189,190 and adipogenesis.¹⁹¹ The commitment to osteogenic lineage in MSCs in the absence of chemical differentiation factors under low frequency ultrasound (stimulated for 28 days) has also been reported.^{167,192–195} In ECs, low frequency ultrasonic stimulation has been shown to improve proliferation and adhesion,^183,196 in addition to stimulating angiogenesis^182,183 to promote tube formation in human umbilical vein endothelial cells (HUVECs).¹⁹⁷ 

A drawback, nevertheless, of low-frequency ultrasonic mechanostimulation is the long times (typically days) that are commonly required for the cells to be exposed to the irradiation to evoke more complex cellular responses.^90,192 Such prolonged exposure often evokes deleterious effects on the cells, which include a considerable reduction in cell viability as well as a greater chance of contamination.

High frequency ultrasonic waves have yielded completely distinct and often nonlinear phenomena when coupled into fluids and materials,¹⁴⁵ stemming from the smaller sound wavelengths that approach the characteristic system length scale—in this case, that of the cell dimension or the confinement they are present within. Moreover, cavitation, which tends to be the dominant mechanism responsible for a host of phenomena driven by low frequency ultrasonic excitation, such as membrane poration, which we have seen in Sec. IV B to trigger a variety of mechanotransduction cascades, becomes increasingly non-existent at higher frequencies since the threshold sound intensity required to generate cavitation scales as the inverse of frequency.^145,198 In addition to therefore involving an alternative mechanosensory mechanism for the excitation of mechanotransductive responses in the cell, the absence of cavitation allows for considerably greater retention of cell viabilities (typically $>$90% for high frequency ultrasound compared to around 60%–80% for low frequency ultrasound has generally been reported).¹⁹⁹ 

Moreover, only short-duration exposure (typically up to 10 min per day for several days) of 10 MHz order surface (SAW) and hybrid (SRBW) wave excitation, for example, was generally required to evoke similar cell mechanoresponses^84,92 that would have otherwise necessitated considerably longer exposure times with low frequency ultrasound (typically continuous excitation over several days). In these instances, perturbation of the cell membrane as a result of the SAW or SRBW excitation was reported to lead to activation of the SACs, and, in particular, the piezo channels,^84,200 to drive an influx of extracellular Ca²⁺ into the cytoplasm. This, in turn, resulted in a transient decrease in the Ca²⁺ concentration in the immediate vicinity between the intercellular junctions to faciliate rearrangement of the AJ (e.g., cadherin) structure.⁸⁷ Such cytoskeletal rearrangement is then responsible for triggering a series of downstream signaling cascades. In MSCs, MAPK signaling, for example, was seen to be activated when chemical factors were absent, whereas the Wnt/ $β$-catenin pathway appeared to dominate in the presence of chemical cues.⁸⁴ Concurrently, the transient increase in the Ca²⁺ concentration in the cytoplasm promotes storage of intracellular Ca²⁺ in the endoplasmic reticulum (ER) by activating the sarcoplasmic reticulum Ca²⁺-ATPase (SERCA) pump,⁹² which subsequently led to activation of IP3R (inositol 1,4,5-trisphosphate receptor) and RyR to trigger signaling by the secondary second messenger cyclic adenosine monophosphate (cAMP). Downstream effects that transpire then include Rho–ROCK, ERK1/2 MAPK, and BMP2 signaling, as well as activation of the ESCRT (endosomal sorting complexes required for transport) and Epac1 (cAMP-activated guanine nucleotide exchange protein 1)–Rap1 (Ras-related protein 1) pathways to drive a variety of cellular activities depending on the type of cell that was stimulated^84,92 (to be discussed in Sec. V).

A few studies have also found similar increases in intracellular Ca²⁺ in various cells, including HUVECs,²⁰¹ Chinese hamster ovary (CHO) cells, and both invasive and noninvasive cancer cells,²⁰² under high-frequency bulk waves, albeit with longer exposure durations over several hours. This was attributed to SACs gating, such as that of piezo channels, when both CHO and epithelial human breast cancer (MDA-MB-231) cells were stimulated at approximately 40–50 MHz at intensities of approximately 90 W/cm².²⁰² The involvement of TRP channels in cancer cells²⁰³ or connexin-43 in MSCs⁸⁸ has also been reported. In the latter, the opening of connexin-43 is accompanied by the release of adenosine triphosphate (ATP) into the extracellular space, which, in turn, results in the activation of phospholipase C followed by Ca²⁺ release from the ER.⁸⁸ As with other mechanostimulation studies using different frequencies and types of waves, it is expected that a myriad of downstream signaling cascades can be activated, although most investigations involving high-frequency bulk wave excitation (with the exception of Ref. 88) have focused primarily on demonstrating their effect on cell proliferation (see, for example, Refs. 204 and 205) and therefore its potential utility, for example, for the development of cancer therapeutics.²⁰⁶ 

Given the diverse mechanoresponses of cells to mechanostimulation driven by acoustic excitation in Sec. IV, sound waves, particularly at ultrasonic frequencies, have been employed to modulate a range of cellular activities that include cell growth, adhesion, proliferation, migration, death, and lysis.^{149,207–209} Additionally, there is a large body of literature on acoustically mediated intracellular delivery, enabled primarily by poration of the membrane by the acoustic forcing (i.e., sonoporation) that arises predominantly through cavitation (see, for example, the reviews by Coussios and Roy,²¹⁰ Sitta and Howard,^164,211 and Rich et al.²¹²) although poration-free acoustic membrane permeabilization methods for intracellular delivery using high frequency (>10 MHz) excitation that are equally as efficient but also facilitate significantly higher retention in cellular viability have also been recently reported.^{199,213–216}

In this section, we discuss a number of exemplary applications where acoustically driven mechanostimulation of different cell types can be deployed. These have primarily been focused on applications in regenerative therapeutics given the considerable work that has been carried out to date on the application of mechanical forces to control in vitro cell behavior for tissue remodeling and regeneration,^217–219 although we also dedicate a brief discussion at the end of this section to other applications. As the scope of the present work is limited to acoustic modulation of internal cellular processes, we have excluded applications of ultrasonic technology for the physical manipulation of cells, such as cell trapping and sorting, or rotation (e.g., for spheroid generation^220–222), for which we refer the reader to the various reviews available on this subject (see, for example, Refs. 9, 220, 223, and 206).

Given their multipotency, trophicity, immunosuppressive properties, and capacity for self-renewal, stem cells, and, in particular, MSCs, are attractive candidates for regenerative therapy. Depending on the targeted disease/application, the required stem cell dose can range from $0.2 × 10 6$ to $1.2 × 10 9$ cells, and often involves multiple treatment courses.²²⁴ As such, the isolation and processing of these required numbers of cells constitutes the current bottleneck in practical stem cell therapies. Effective techniques for high-throughput in vitro stem cell expansion that can also address challenges associated with donor-related heterogeneity are therefore required.²²⁵ As constituents within the expansion media itself can potentially alter the characteristics of the expanded cells,²²⁶ efforts have predominantly been focused on developing mechanical intervention methods as an alternative strategy to chemically augmented cell proliferation.

Acoustic stimulation ranging from tens of Hz to approximately 150 MHz, as short as 5 min/day for one day through to continuous treatment over 12 days, has been noted to enhance the proliferation rate in various types of cells, irrespective of species or tissue source (Table II). While the propensity for cells to proliferate typically increases with vibration-induced stimuli, it is difficult to generally conclude how cell proliferation is affected by specific system parameters (frequency, intensity or exposure duration) given that these studies were conducted on widely varying cell types and conditions, often with markedly distinct trends observed. For example, while the proliferation rate of primary ECs was seen to be generally enhanced irrespective of applied frequency or intensity, that of nonmalignant cells tended to increase under low intensities on the order of 10 W/cm², while malignant cells appeared to be more responsive to high intensity stimulation on the order 100–1000 W/cm² (Table II). As such, cell type is likely to be a key determinant in the cell proliferation response to acoustic stimulation.

Moreover, the exact mechanisms by which cell proliferation is altered by acoustic stimulation are still not well understood. As heat is generated due to viscous absorption of the sound wave as it is transmitted through the media, most studies involving acoustic mechanostimulation incorporate some attempt to control fluctuations in temperature, as these can trigger thermally dependent mechanotransductive processes.^227,228 Even then, the various mechanisms associated with non-thermal mechanical effects remain complex. Given the ubiquitous involvement of Ca²⁺ upregulation and Rho–ROCK activation across all forms of acoustic stimulation, as discussed in Sec. IV, and the corresponding changes to the cytoskeletal structure and FAs,^{168,229–231} it is probable that Ca²⁺-induced signaling triggers upregulation of YAP that, in turn, enhances actin nucleation and stability, cytokinesis, and cell cycle progression¹⁸¹ along with improved expression of growth factor receptors and cyclins, such as cyclin D1, E1, A2, and B1.^94,184,232 Other possible signals triggered by acoustic stimulation that have also been implicated in mediating cell proliferation include the ERK1/2, Akt serine/threonine kinase,^94,233 and glycogen synthase kinase-3 $β$ (GSK- $3 β$⁠)/ $3 β$-catenin²³⁴ pathways. However, we caution that careful consideration of the extent of stimulation applied is necessary given that prolonged exposure is known to inhibit cell growth or to induce apoptosis.²³³ 

Cell migration is a complex process involving changes to tension in the cell membrane in addition to rearrangement of the cytoskeletal structure²⁴¹ that is critical to various cellular activities, including homeostasis, morphogenesis, immune cell trafficking, cancer metastasis, and wound healing.²⁴² In stem cells, mobilization (migration and homing of cells) to defective regions is critical for successful regeneration.²⁴³ In the bid to improve regenerative therapy, various mechanical loads, including acoustic forcing at frequencies ranging from a few Hz to MHz with intensities between 20 and 250 mW/cm² (Table III) together with the addition of biochemical stimuli, have been explored to regulate cell migration.^242–245 Based on the available data, higher frequencies, particularly that at MHz order, appeared to facilitate more cell migration over lower frequencies. A lone study nevertheless found that 20 kHz acoustic stimulation, in contrast, reduced migration of smooth muscle cells, with the baseline migration only being restored after 24 h of post-exposure incubation.²⁴⁶ 

In addition to promoting migration, continuous acoustic stimuli have been reported to orientate cells along the direction of the acoustic wave propagation, at least in fibroblasts,^201,230,247 although this observation was not always universal. For example, while osteosarcoma exposed to 159 MHz at an intensity of 300 mW/cm² displayed improved migration, there was no preference for its orientation.²⁴⁰ 

The fundamental mechanisms that govern such random or directed collective migration under acoustic stimulation are unclear. A complication is that migration effects are typically accompanied by cell proliferation and is often unclear if the improvement in migration is due to the increased proliferation rate or the actual movement of the cell. At least in one study, it was seen that acoustic stimulation at 45 kHz and 1.5 MHz frequencies under intensities between 25 and 250 mW/cm² led to reductions in the migration rate of odontoblast and osteoblast cells in the presence of mitomycin C (MMC)—a cell proliferation inhibitor, suggesting that the proliferative capacity of the cells under acoustic stimuli dominates their propensity to migrate.^248,249

Similarly, little is understood about the mechanotransductive processes that govern the migration of cells under acoustic stimuli. What is known to date is the involvement of the ECM, FA and cytoskeletal structures.²⁵⁰ In particular, acoustic stimuli have been reported to activate FA proteins, vinculin and tubulin, that, in turn, give rise to rearrangement of the cytoskeletal network.^246,251 Additionally, both FAK and Ras-related protein 5 (Rab5) have been found to regulate cell migration driven by the acoustic stimulation,⁸⁶ whereas mechanosensitive molecules such as Ca²⁺ and ROS (reactive oxygen species) are known to regulate the activation of PI3K/Akt and JNK (c-Jun N-terminal kinase) pathways, in addition to Rac1 (Rac Family Small GTPase 1) signaling,^201,251,252 which increase the formation of lamellipodia and membrane ruffling to result in increased cell spreading and the formation of actin filopodia to drive cellular migration.¹³⁹ 

Acoustically induced mechanical loading has been shown to influence angiogenesis^{183,196,229,259}—the in situ formation of new vasculature through a complex process that involves the proliferation, migration, and differentiation of vascular ECs—to thereby facilitate tissue regeneration and wound healing. Exposure of these cells to 0.5–1 MHz acoustic stimulation at intensities between 0.3 and 2.2 W/cm² for 5–30 min, for example, displayed enhanced propensity for proliferation and migration, that, in turn, promoted sprouting of new capillaries.^183,196,229 It has been claimed that the microstreaming from cavitation-induced microbubbles generated as a consequence of acoustic stimulation at low ultrasonic frequencies (1 MHz) also leads to improved cell proliferation and migration, both of which are crucial for angiogenesis, although it is unclear if this was a direct consequence of either the vibrational stimuli on the cells and the membrane deformation and corresponding SAC activation that is induced, or an indirect consequence of the shear stress imposed by the microstreaming.¹⁵⁰ 

In any case, the exact fundamental mechanism by which angiogenesis is triggered by mechanostimuli, in general, is not well understood, although the role of mechanosensitive molecules, such as Ca²⁺, nitric oxide (NO) integrins, FAKs, and caveolin, have been implicated in the process.²⁶⁰ In particular, acoustic stimulation has been shown to increase intracellular Ca²⁺ and NO to effect cytoskeletal rearrangement in ECs.^261,262 It is then possible that these induce nuclear translocation of YAP and activation of the Akt signaling cascade to result in improved vascular endothelial growth factor (VEGF) expression that accompanies angiogenesis.^93,259 However, due care has to be taken when exposing the ECs to these low frequency ultrasonic waves to avoid inducing apoptosis in the cells.

Mechanical stimuli involving laminar shear or cyclic stretching/pressure have long been adopted to guide the differentiation fate of stem and progenitor cells, particularly toward osteogenic lineages,^126,263,264 although these have typically been administered in the presence of at least one biochemical stimulant. Bulk acoustic stimuli at frequencies ranging from Hz to MHz have also been used to instigate various differentiation fates, particularly osteogenesis, in stem cell or progenitor cells, albeit typically in the presence of biochemical factors as well (Table IV). The focus of the majority of these studies has, nevertheless, been primarily rooted in the demonstration of the mechanostimuli for stem cell differentiation, rather than explicating the fundamental mechanisms by which the acoustic stimulation drives the process. That the mechanostimuli have been applied in the presence of the biochemical differentiation factors, however, has led to the prevalent thought that the observed changes in the protein profile can primarily be attributed to the chemical rather than the acoustic stimuli.²⁶⁵ 

An exception to the aforementioned studies is the work by Dalby and co-workers, who showed the possibility of directing MSCs toward osteogenic lineage using nanoscale-amplitude acoustic stimuli at kHz-order (50 Hz–10 kHz) frequencies without requiring any biochemical factors. In their studies, a modest improvement in the expression of the osteogenic marker RUNX2 (Runt-related transcription factor 2) was reported, particularly with 1 kHz mechanostimulation continuously for 7 days.¹⁹³ This was ameliorated by co-culturing stem cells with osteoprogenitors, osteoblasts, and osteocytes, and also in collagen-based soft gel (0.26 Pa) scaffolds and on nanotopographically manipulated surfaces.^{167,192,195,266} The acoustic stimulation, delivered continuously for 27 days, resulted in upregulation of various osteogenic markers, such as RUNX2, bone morphogenic protein (BMP-2), osteocalcin (OCN), osteopontin (OPN), osteonectin (ONN), and osterix.

Of particular note is the finding in the aforementioned studies of little upregulation in osteogenic markers beyond 1 kHz, and hence the assertion that there is little benefit in operating at higher frequencies.¹⁹⁵ Contrary to those findings, however, it has recently been shown that 10 MHz mechanostimulation involving SRBWs not just results in significant osteogenic marker upregulation in human MSCs from different tissue sources and donors without requiring chemical/biochemical stimuli, but that it also does so considerably earlier at day 3 following commencement of the stimulation, and with just 10 min of stimulation per day for 5 days.⁸⁴ Despite such short stimulation periods, the cells are nevertheless observed to retain their long-term osteogenic potential, even after trypsinization and reseeding after 5 days. As such, there are appreciable advantages in using these higher frequencies since the early osteogenic commitment lowers the dedifferentiation risk in the stem cells.²⁶⁷ Moreover, the considerably shorter, non-continuous stimulation not only increases cell viability and proliferability but also offers a simpler and more cost-effective solution for clinical translation. In addition, the early, yet persistent, differentiation of SRBW-stimulated cells facilitates easier and more effective incorporation of committed cells into scaffolds for implantation^84,268 or delivery direct to the injury site.²⁶⁹ This is because such early commitment allows the embedded or delivered cells to grow and adapt within the naturally evolving niche in real time, therefore circumventing the cumbersome need for designing and developing multicomponent scaffolds that mimic the natural niche during ex vivo or in vitro differentiation,^270–273 in addition to facilitating the adaptation of the cells to different physiological needs for the development of personalized therapeutics.

The exact mechanostimulation-induced signaling cascade involved in osteogenesis is complex due to the involvement of multiple mechanosensors and pathways, and their interconnectivity. With mechanical loading, integrins,^{59,172,274,275} SACs such as TRP^124,167,276 and piezo channels,^84,167,277 cadherins, and connexins have been implicated in the process. The integrins trigger mechanotransductive processes that lead to activation of kinase pathways^169,278 involving FAs such as paxillin and vinculin, whereas activation of SACs and the opening of connexin hemichannels result in increases in intracellular Ca²⁺ that initiates the MAPK pathway.^279,280 On the other hand, mechanostimulation is also known to dissociate cadherins from auxiliary proteins such as $β$-catenin, whose concentration increase in the intracellular domains initiates the Wnt/ $β$-catenin pathway.^281,282

It was suggested that the low frequency (50 Hz–10 kHz) acoustic stimulation, in particular, directed activation of TRPV channels (which, being insensitive to membrane stretch,³⁰ are gated through changes in temperature that affects the membrane phospholipids to drive changes in the membrane tension^26,27,283) that, in turn, triggered the Wnt/ $β$-catenin pathway. In addition to such temperature-induced ion channel gating, it is also probable that the osteogenic differentiation in these studies could also be induced through the influence of the surface nanotopography of the substrates employed, or due to the presence of collagen, even though the scaffold used was considerably softer than those known to promote osteogenesis, since both the ECM and patterned substrates are known to participate in osteogenic differentiation through Wnt/ $β$-catenin and MAPK signaling.^284–287 A different signaling transduction cascade, however, has been explicated for the SRBW mechanostimulation in the absence of any biochemicals, wherein the transient membrane aberrations that are induced by the excitation are suggested to activate piezo channels that, in turn, lead to Rho–ROCK or calpain activation to trigger bone morphogenetic protein-2 (BMP-2) production.⁸⁴ Regardless, all of these pathways typically upregulate RUNX2, which commits the cells toward osteogenesis (Fig. 8).^288,289

In addition to osteogenesis, acoustic stimuli have also been shown to direct stem cells toward other cellular fates, such as neurogenesis, adipogenesis, and chondrogenesis (Table V), although this always necessitated the inclusion of the specific biochemical differentiation factor. Interestingly, the incubation period between consecutive acoustic exposure at 90 Hz on embryonic MSCs was shown to have an inhibitory effect on adipogenesis,²⁹⁰ although another work at the same frequency on BMD-SCs, in the absence of such incubation periods between stimuli but again with the addition of adipogenic differentiation media, resulted in commitment to an adipogenic lineage.¹⁹¹ Most of these studies, however, do not examine the mechanotransduction pathways that are involved in the process, although it has been suggested that the ERK/MAPK pathway is activated in acoustically triggered neurogenesis in BMD-SCs.¹⁸⁹ More broadly, it is likely from various studies that MAPK signaling can induce either osteogenesis, neurogenesis, chondrogenesis, or adipogenesis in BMD-SCs or chondrocytes depending on the biochemical factor that is added during the mechanostimulation.^{141,191,290,291}

The endothelial barrier, which partitions the intravascular and interstitial tissue compartments, plays a vital role in controlling the passage of solute and nutrient molecules in and out of organs. To enable such transport, its permeability is regulated by interactions between neighboring ECs that facilitate dynamic remodeling of the cytoskeletal structure through the modification of the associated junctional proteins. Given we have seen that cytoskeletal perturbation constitutes a dominant cellular response to various mechanostimuli, it then stands to reason that acoustic stimuli can potentially constitute a key driver of cytoskeletal remodeling and is hence a potent modulator of endothelial barrier function, which is crucial for tissure remodeling.

Indeed, 1 MHz-order bulk acoustic waves, either through direct mechanostimulation of ECs, or through the generation of cavitational microbubbles, have been exploited for enhancing the permeability of the endothelial barrier in both in vitro and in vivo conditions to facilitate transport of therapeutic molecules across the barrier.^{162,314–316} In either case, the acoustic stimuli, through the modification of the associated junctional proteins, were observed to drive influx of Ca²⁺ into the cell to induce the formation of actin stress fibers across the cell that, in turn, alters its contractility, leading to a change in the intracellular gap and hence the permeability of the barrier.^314–317 We note the effect to be considerably larger in the case of bubble generation given their propensity to porate the cell membrane. In both cases, though, only slight increases in intracellular Ca²⁺ within the cell were observed, given that higher Ca²⁺ concentrations (⁠ $< 10 μ$M) were associated with the inability for cells to recover from the imposed disturbance.³¹⁸ 

More recently, 10 MHz SRBW mechanostimulation of endothelial monolayers has also been observed to facilitate modulation of endothelial barrier integrity: up to around fourfold increase in transendothelial electrical resistance (TEER), corresponding to a decrease in paracellular permeability by 5% that was sustained for over 4 h, was observed with just 8 min of SRBW excitation of ECs from different tissue sources (human umbilical vein, human saphenous vein, human coronary artery, and human aorta cells). Moreover, the effect was able to be sustained over longer durations through repeated treatments every 6 h.⁸⁷ 

The same increase in intracellular Ca²⁺ influx into the cell as a result of transient membrane permeabilization and piezo channel activation under SRBW mechanostimulation in Sec. V D for stem cells, was observed to induce a local depreciation of Ca²⁺ near the junctions between the ECs, causing a reversible invagination of the immature zipper-like vascular endothelial (VE)-cadherin structure that forms the backbone of the AJs anchoring together the actin cytoskeletons of the ECs to form its barrier (termed the “sonochallenge” phase). This leads to the formation of mature linear VE-cadherin structures as a consequence of activation of the Ca²⁺-induced cAMP-activated guanine nucleotide exchange protein 1 (Epac1)–Ras-related protein 1 (Rap1) pathway (termed the “sonotransformation” phase). This, together with Ca²⁺ remodeling of the actin cytoskeletal structure, in which actin stress fibers across the cytosol were progressively replaced by circumferential actin bundles lining the cell periphery near its junctions, then resulted in a reduction in barrier permeability.⁸⁷ 

The fourfold improvement in TEER with SRBW mechanostimulation is considerably higher than the 1.1–1.2-fold enhancement reported for chemical stimulation, therefore alluding to the SRBW as a potent means for improving endothelial barrier capacity in the development of in vitro models that more faithfully recapitulate in vivo barrier function—a current limitation of endothelial tissue models such as the blood–brain barrier, blood–retinal barrier or pulmonary air–liquid interface, and hence a bottleneck in the development and testing of therapeutic interventions and delivery strategies.⁸⁷ 

The mechanotransduction process involved in the SRBW endothelial barrier modulation, illustrated in Fig. 9, clearly shows how acoustic stimulation is able to invoke the activation of multiple signaling cascades. Upon SRBW stimulation, permeabilization of the membrane and SAC activation leads to a rapid influx of Ca²⁺ into the cell. Such an increase in intracellular Ca²⁺ results in immediate Rho–ROCK activation given that Rho signaling simply involves phosphorylation/dephosphorylation, which triggers the formation of actin stress fibers in the initial sonochallenge phase. The disruption of the AJ structure into zipper-like patterns that ensues then leads to contraction of the cell and a corresponding increase in the intercellular gap that results in permeabilization of the endothelial barrier. This response is however transient, since the increase in intracellular Ca²⁺ also activates the SERCA pump to facilitate storage of surplus Ca²⁺ within the ER to maintain cytoplasmic Ca²⁺ homeostasis, as described in Sec. IV C. Upon relaxation of the SRBW excitation, the stored Ca²⁺ is subsequently released from the ER back into the extracellular milieu over time primarily through RyR activation. This is accompanied by activation of the Epac1–Rap1 pathway to result in formation of circumferential actin bundles and stable linear AJs in the sonotransformation phase, consequently resulting in an increase in the integrity of the endothelial barrier.⁸⁷ 

In a manner akin to chemical intervention (e.g., subjecting cells to chemical additives such as calcium ionophores or calcium phosphate) or physical insults (e.g., inducing hypoxia or exposing cells to radiation), all of which serve to increase Ca²⁺ levels in the cell, the increase in influx of Ca²⁺ into cancer cells as a consequence of SRBW mechanostimulation due to aberrations generated on the cell membrane and piezo channel activation has been observed to activate various signaling pathways associated with cellular homeostasis. More specifically, the elevation of Ca²⁺ in the cytoplasmic matrix under the SRBW was observed to trigger sequestration of the accessory protein ALIX [ALG-2 (apoptosis-linked gene 2)-interacting protein X]³¹⁹ at the site of the membrane aberrations to initiate exosome biogenesis,⁹² in which the membrane is remodeled to facilitate protein cargo sorting, plasma and nuclear envelope membrane repair, and the formation of multivesicular bodies (MVBs). In particular, membrane invagination resulted in the generation of intraluminal vesicles within these MVBs, which upon fusion with the plasma membrane, are released into the ECM as exosomes.^320,321

As such, the SRBW-facilitated stimulation of exosome production can be viewed from the perspective of the cells' attempt to maintain homeostasis by repairing the transient membrane aberrations induced by the mechanical stimulation. In this process, the cell membrane acts as the mechanosensor with the intracellular Ca²⁺ as the mechanotransmission element and the ESCRT pathway as the mechanoresponse (Fig. 10). As with the endothelial and stem cell treatments in Secs. V D–V F, we note again that only short exposure to the SRBW (8 min) is sufficient to drive longer term effects of exosome release within a half hour post-exposure period. Given similarly high post-exposure cell viabilities (90+%), it is then possible to envisage cycling the SRBW irradiation to achieve tenfold enrichment in the exosome concentration to a homogeneous yield of around 1.7–2.1 fold/h, thereby constituting a facile but effective method for circumventing the perennial limitation of insufficient homogeneous exosome yield that constitutes the bottleneck in exosome diagnostics and therapeutics.⁹² 

While the majority of applications involving acoustic mechanostimulation have predominantly been confined to those within the regenerative therapeutics space, there are also a number of other applications in which the utility of such stimulation has proven beneficial for the manipulation of a diverse range of cells, including immune cells [e.g., natural killer (NK) cells], neurons, oocytes, embryos, and sperm.^322–327 Acoustically stimulating oocytes, for example, at 20 Hz and between 1 and 3 MHz with intensities of approximately 2 W/cm², have been observed to maintain higher numbers of cells in the blastocyst (suggesting that the stimulation prevented cellular damage during oocyte activation), in addition to facilitating nuclear activation, which is dependent on the extracellular Ca²⁺ concentration,^328,329 noting that 0.1–0.5 mM is optimal for oocyte activation.³²³ In fact, oocyte activation deficiency is the primary reason for total fertilization failure (TFF), with abnormal Ca²⁺ signaling being a key characteristic of the condition.³³⁰ Interestingly, similar increases in the rate at which embryos develop as a consequence of their activation have been observed when fused embryos are exposed to 1 MHz mechanostimulation at an intensity of 2 W/cm² for just 30 s per day.³²² 

Short duration (seconds) exposure of sperms to 19.28, 48.5, and 100 MHz SAWs with intensities ranging between 250 mW and 2 W, on the other hand, has been shown to influence their motility without compromising their structure or function.³²⁵ The influence of the acoustic stimulation on sperm motility was suggested to arise through the ability of the acoustic stimulation to mediate transport of Na⁺, H⁺, Ca²⁺, and/or HCO³⁻ into the cells by activating various ion channels, although the authors were compelled to attribute a lesser role to Ca²⁺ transport given their observations that sperm motility remained elevated even under Ca²⁺-free conditions. We note from their results though that the motility did appear to decrease from 34% to 19% in the absence of Ca²⁺, therefore implying that Ca²⁺ transport still has an effect on sperm motility, which would be consistent with prior studies reporting Ca²⁺ regulation of intracellular ATP and cAMP, both of which play a central role in sperm motility.^331,332 While a large influx of Ca²⁺ is known to produce hypermotile sperm,³²⁵ we note Ca²⁺ influx to be a key driver in the activation of the cAMP/PKA pathway that regulates sperm motility.^325,331 Given that it is the ratio of intracellular to extracellular Ca²⁺ that regulates the energy (ATP) required for sperms to swim,³³³ it is therefore likely that the magnitude of Ca²⁺ influx constitutes a key parameter in regulating sperm motility.

Acoustic stimulation has also been shown to offer a nondestructive, reversible, and adjustable means of manipulating neurons,³³⁴ giving rise to an entire field of sonogenetic neuromodulation for the regulation of the central, peripheral, or autonomic nervous system.^{207,324,326,335–338} Sciatic nerves and cortical neurons, excited by 100–800 MPa acoustic pressure at 0.8 and 27 MHz, for example, can be activated by acoustic stimulation,^339,340 leading to an increase in their conduction velocity and action potential, thereby altering their excitability. Similar effects were noted in acoustically stimulated sciatic nerves at 2–7 MHz and 100–800 W/cm².³⁴¹ Meanwhile, a reduction in the action potential of sciatic nerves due to thermal effects was reported with acoustic stimulation at 0.661 and 1.986 MHz with an intensity of 44 W/cm².³⁴² 

Increased neural activity of cells in the superficial layer of salamander retina with 43 MHz bulk ultrasound has, in addition, been reported, although the effect failed to be reproduced in retinal ganglion cells.³⁴³ Although the precise mechanism by which neuron excitability is modulated through acoustic stimulation was not explicated in this study, it is possible that their observations can be attributed to piezo channel activation, which was reported in HEK293 cells exposed to bulk ultrasound excitation at similar frequencies,²⁰² and since SACs such as piezo channels are endogenously expressed in neurons.³¹ Moreover, the excitation of mouse neurons under 500 kHz bulk mechanostimulation at intensities of 0.1–0.5 MPa for 20 min has been reported, wherein Ca²⁺ influx upon activation of the piezo channel as a result of the stimuli was seen to promote upregulation in calcium/calmodulin-dependent protein kinase II (CaMKII), cAMP Response Element-Binding protein (CREB) and c-Fos,¹⁶⁶ the latter having been implicated in the induction of neuron excitability.^341,344

Recently, it was as well noted that exposing natural killer (NK-92) cells to 96.7 MHz SAWs at an applied voltage of 3.6 V_pp for 30 min with a 5% duty cycle led to an increase in intracellular Ca²⁺. While the increase, in itself, was insufficient to activate the NK cells alone, which is known to occur through calcium signaling, it was shown that the SAW mechanostimulation led to an approximate 1.5-fold enhancement in the expression of lysosomal-associated membrane protein 1 (LAMP-1)—a marker of NK cell degranulation and hence activation—when applied in conjunction with the addition of phorbol 12-myristate 13-acetate (PMA) that is commonly employed to trigger protein kinase C (PKC) signaling for initiating T-cell activation.³²⁷ 

Despite (1) the myriad of mechanosensors that can be activated, (2) the multiple signaling pathways that arise, and (3) the variety of downstream mechanoresponses observed—all as a consequence of the diverse forms of the application of acoustically driven mechanostimulation modes on different cells across a range of frequencies that have been discussed in Sec. IV F, we make the case here for the possibility of a mechanotransduction mechanism that is universal across all such acoustically driven mechanostimulation, at least for both low and high ultrasonic frequencies.

The cell membrane, which comprises an asymmetric bilayer containing glycerophospholipids, sphingolipids, cholesterol, and carbohydrates, in addition to high amounts of transmembrane or peripheral proteins,^345,346 compartmentalizes the cell from its surrounding matrix. As the interface between the microenvironment and the cellular metabolism, it therefore plays an important role in the transduction of mechanical cues into the cell. Specifically, the cell membrane couples directly with effector molecules. These include transmembrane proteins, such as SACs, AJ proteins, GJ proteins,^347–349 and, mechanosensitive proteins, such as FA proteins.³⁵⁰ Any perturbation imposed on the fluidic cell membrane is therefore expected to alter these transmembrane proteins and associated cytoskeletal structures.

These mechanosensory structures on the cell membrane constantly interact with its surrounding microenvironment, allowing the cell to adapt to mechanical stimulation. This is facilitated by tension within the membrane that is generated as a response to the application of an external force on the membrane surface. Membrane tension develops within lipid bilayers and constitutes the basis that defines the structure of all biological membranes. Unlike in pure lipid vesicles, the membrane tension of a cell can vary locally owing to peripheral protein binding, the presence of transmembrane proteins, and interactions with the underlying actomyosin cortex.^351,352 These interactions provide additional resistance to changes in the membrane area and can constrain such changes to a localized area of the cell. Moreover, they also facilitate mechanotransduction and therefore have the ability to modulate various cellular processes such as endocytosis, exocytosis, cell migration, cytokinesis, mitosis, and intracellular trafficking. For example, changes in membrane tension have been suggested to induce ligand-independent integrin signaling that arises during cell migration.³⁵³ 

When a cell is mechanically stimulated in a way that changes its membrane tension and shape, the primary response of the cell is to counteract the effect and attempt to restore the membrane tension in order for it to maintain homeostasis, particularly to avoid apoptosis. The actual response depends on the nature and level of the mechanical stimulation since different stimuli alter the membrane tension and shape differently. Mechanical stimuli that constricts the cell leads to membrane compression [Fig. 11(ii)], which reduces the cell size by forming membrane folds of different shapes and sizes with lipid packaging defects in highly curved areas and phosphoinositide clusters.^354–356 Such changes in membrane curvature are often accompanied by an increase in endocytosis owing to a decrease in membrane tension. To counteract the reduction in membrane tension, vacuole/vesicle-like dilations (VLDs) can also be formed. The formation of these endocytic pits and/or VLDs, in turn, can activate SACs.^54,357

In contrast, any form of mechanical stimuli which increases the cell area (e.g., stretching the cell) causes both direct activation of SACs through the force-from-lipid model, as well as the flattening of membrane folds, to accommodate the change in tension [Fig. 11(iii)]. The corresponding increase in membrane area involves the use of lipid reserves that, in turn, increases the membrane tension and order, in addition to increasing the propensity for vesicles to fuse to the membrane to increase its area. Both effects result in transient membrane aberrations or perturbations, which also lead to the activation of SACs through the force-from-filament model along with actomyosin contraction.^34,357–359

In short, all forms of mechanical stimulation—be it static stretch (as a result of tension or shear), compression, or a periodic combination of both effects to constitute more complex dynamical forms of mechanical stimuli such as those due to cavitational-driven poration at low ultrasonic frequencies, or membrane oscillation under SAWs or SRBWs that possess both in-plane and out-of-plane vibrational displacement—lead to some form of disruption to the cell membrane, whether it be to reduce or increase its tension. ^{84,87,92,199,360,361} [While some studies have proposed that mechanical stimuli may induce intra-membrane cavitation, which, in turn, can modulate transmembrane proteins (SACs) and the cytoskeletal structure, they do not explicitly rule out the involvement of the membrane itself since the periodic expansion and contraction of the intra-membrane spaces as a result of such cavitation can itself also lead to stresses on the membrane (Refs. 362 and 363).] Whichever the case, the resultant change in membrane tension drives activation of the SACs, leading to an increase in intracellular Ca²⁺ and triggers downstream effects associated with various Ca²⁺-induced signaling cascades that constitute the cell's mechanoresponse. We note that such mechanosensitivity also applies conversely when the membrane correspondingly relaxes upon cessation of the mechanostimulation.

For it to invoke a wide range of biological effects, a metal ion must interact with biological ligands such as proteins, lipids, and carbohydrates by binding to commonly available reaction centers in these biomolecules. Unlike Mg²⁺ and Zn²⁺, which bind with greater affinity to nitrogen ligands, Ca²⁺ has the most affinity for carboxylate oxygen, and, as such, is able bind to these groups in aspartic and glutamic acid, and to the oxygen atoms in Asn (asparagine), Gln (glutamine), Ser (serine), Thr (threonine), and Tyr (tyrosine).³⁶⁴ 

Another relevant parameter that is critical for an efficient second messenger is its binding kinetics. Ca²⁺ binds to and dissociates from a protein a hundred or so times faster than Mg²⁺.³⁶⁵ Moreover, its high coordination number (6–8) and often irregular (i.e., protein-induced) coordination geometry allow it to be accommodated into a larger number of proteins.³⁶⁴ Ca²⁺-mediated cross-linking is moreover reversible and thus responsive to a change of conditions,³⁶⁶ thus allowing cells to easily regulate their concentrations of free and sequestered Ca²⁺ both in time and space by utilizing their calcium stores and calcium-specific ion channels.³⁶⁷ Furthermore, Ca²⁺ can either operate within small cellular compartments or act on the entire cytoplasm, with signaling that can last from microseconds to hours, occurring either transiently or in a pulsatile manner.

The underlying principle of Ca²⁺ signaling is the cellular response to a change in its internal Ca²⁺ concentration. In general, the stimulation of a cell leads to an acute increase in its internal Ca²⁺ concentration above the basal level. Upon removal of the stimulus, this concentration returns back to the resting state. The intensity and duration of the stimulus determines the extent of the calcium dynamics. Increases in internal Ca²⁺ concentration can either originate from Ca²⁺ release from intracellular stores and/or Ca²⁺ influx from the extracellular space, assisted by ion channels or transporters on the cell membrane or the ER within the cell.

As large and sustained elevations in internal Ca²⁺ concentration can lead to a variety of deleterious cellular effects, the cell naturally attempts to remove extra Ca²⁺ from the cytoplasm in an effort to return the Ca²⁺ concentration back to its basal level to maintain homeostasis.³⁶⁸ This is regulated by the simultaneous interplay of multiple counteracting processes known as Ca²⁺ “on” and “off” mechanisms. During the “on” mechanism, ion channels located on the cell membrane regulate the supply of Ca²⁺ from the ECM, and ion channels on the ER from the intracellular storage space. In the “off” mechanism, cells remove Ca²⁺ from the cytoplasm either by storing extra Ca²⁺ in its intracellular stores such as the ER or by effluxing Ca²⁺ back into the extracellular milieu through Ca²⁺ATPases on the cell membrane and the ER through activation of its ion channel receptors, such as inositol 1,4,5-trisphosphate (IP3R) and ryanodine (RyR), along with ion exchangers that utilize gradients of other ions to provide the energy to transport Ca²⁺ out of the cell.¹⁷ 

At highly elevated Ca²⁺ levels, quick restoration of the cytosolic Ca²⁺ concentration to its basal level occurs through the movement of intracellular Ca²⁺ via the SERCA pump that mediates its redistribution within the ER (Fig. 12).³⁶⁹ As such, the influx of Ca²⁺ from the extracellular milieu can initiate Ca²⁺-induced Ca²⁺ release (CICR) and RyR activation when the stores' Ca²⁺ content exceeds normal physiological values (i.e., “store overload”),³⁷⁰ releasing Ca²⁺ primarily into the cytosol where it may be released back into the extracellular milieu.³⁶⁹ The RyR-mediated Ca²⁺ release, in turn, activates IP3R-mediated Ca²⁺ release from the ER.³⁷¹ Such Ca²⁺ release from the ER could then enhance the Ca²⁺ efflux activity of sodium/calcium exchangers (NCX) and plasma membrane calcium ATPase (PMCA) pumps.³⁷² This process by which a cell maintains its internal Ca²⁺ concentration spatiotemporally through various receptors and ion channels is known as Ca²⁺ mobilization.

Such a unique ability of the cellular machinery to spatiotemporally modulate intracellular Ca²⁺ concentration renders Ca²⁺ an ideal messenger for mechanotransduction. The versatility of Ca²⁺ signaling, which is evidenced from the fact that Ca²⁺ controls almost every aspect of cellular activity from its infancy to death, and its diversity in directing different activities depending on the cell type stemming from non-transcriptional events (phosphorylation/dephosphorylation events to activate proteins) or transcriptional effects,³⁷³ arises from its potential to interact with a wide range of molecules, including cAMP, nitric oxide, phosphatidylinositol-3-OH kinase, and MAPK. As seen in Fig. 4, these interactions trigger a multitude of signaling cascades, such as the phosphatidyl inositol (PI), phospholipase C (PLC), inositol triphosphate (IP3), protein kinase C (PKC), and calmodulin (CaM) or calmodulin kinase II pathways, and can crosstalk with G protein-coupled receptor (GPCR) signaling [e.g., protein kinase A (PKA) and receptor tyrosine kinase (RTK)] pathways to control almost every cellular activity, from short-term effects (e.g., gene transcription, and, cell contraction and secretion) to longer-term effects regulating cell development and regeneration (e.g., cell proliferation, migration, differentiation, apoptosis, and necrosis).

Interestingly, downstream effectors of Ca²⁺ signaling can also, in turn, regulate Ca²⁺ mobilization. For instance, RyR-mediated Ca²⁺ release from the ER is accompanied by activation of adenylyl cyclase (AC) and an increase in cAMP, which then initiates the Epac-mediated signaling cascade. RyR activation can also trigger Epac directly that initiates PLC signaling to result in the formation of IP3 that initiates IP3R-triggered Ca²⁺ release from the calcium store.³⁷⁴ Essentially, any signaling cascade triggering the formation of IP3 is able to activate IP3R-triggered Ca²⁺ release from the ER, whereas increases in intracellular Ca²⁺ alone triggers RyR activation.³⁷¹ Whichever the case, Ca²⁺ release from the ER enhances the Ca²⁺ efflux activity of NCX³⁷² while Ca²⁺ efflux through plasma membrane calcium ATPase (PMCA) is regulated by calmodulin, PKC, and PKA (products of the cAMP–Epac pathway).³⁷⁵ In short, intracellular Ca²⁺ not only regulates cellular activities but also its own mobilization, and, as such, initiates multiple signaling cascades at any stage of the mobilization.

A possible explanation for this exquisite behavior could be the plethora of Ca²⁺-binding proteins (which regulate Ca²⁺ homeostasis to activate different Ca²⁺ signaling cascades that determine downstream cell response and fate) that exist, and the wide range of affinities they possess (which vary by 10⁶⁺-fold), depending on their location and function.³⁷⁶ Based on a study conducted on effectors of calcium-binding proteins such as Ca²⁺/calmodulin-dependent protein kinase II (CAMKII) and PKC, different cellular activities were observed to be determined by the magnitude and frequency of the Ca²⁺ oscillation:^377–380 brief Ca²⁺ oscillations/spikes evoke simpler cellular activities such as cytoskeletal rearrangement, vesicle trafficking (including neurotransmitter release), and contraction,¹⁷ whereas complete Ca²⁺ mobilization has the propensity to trigger more complex processes such as differentiation^381–383 and angiogenesis,^384–386 or even oocyte activation.^387,388

All forms of acoustic stimulation, irrespective of frequency, are able to modulate intracellular Ca²⁺ to drive its mobilization within the cell in one way or another by altering the tension of the cell membrane, as described in Sec. VI A and illustrated in Fig. 13. At lower (Hz- to kHz-order) frequencies where cavitation is prevalent, this could occur either directly through the formation of membrane pores or via activation of SACs such as the TRPV channels. At higher (MHz-order) frequencies where cavitation is suppressed, the influx of Ca²⁺ is primarily driven through transient formation of VLDs or the activation of piezo channels (the involvement of voltage-gated ion channels has also been implicated at GHz frequencies³⁸⁹) with Ca²⁺ efflux from the ER back into the extracellular milieu being dependent on, or, in some instances,³⁹⁰ independent of, the extracellular Ca²⁺ concentration.

Differences in Ca²⁺ mobilization between the frequency regimes can, nevertheless, provide an explanation for why prolonged exposure times are required with low frequency acoustic stimulation whereas significantly shorter durations are sufficient to invoke similar cellular responses at high frequencies, particularly that with SRBW mechanostimulation. Moreover, given the different stages of Ca²⁺ mobilization that exist, a plethora of signaling cascades can be evoked, which can explain the distinct mechanoresponses observed for the various forms of acoustic stimuli on different cell types, or even that observed with the same acoustic stimuli but under different excitation parameters. As such, a deeper understanding of the Ca²⁺ mobilization induced for each specific case is required to design and develop more effective and targeted therapeutic strategies.

Sound wave excitation offers a considerably broader means of mechanically stimulating cells. Unlike steady forms of mechanical loading (e.g., compression, tension, or shear) applied on the cells under constant loads, either statically or quasi-statically at frequencies of several Hz, dynamic vibrational stimuli can be excited in distinct forms of bulk, surface, or hybrid acoustic waves across a broad amplitude and frequency spectrum to not only evoke far richer, albeit more complex, cellular responses, but to also allow tunability for a diverse range of applications.

Low frequency (Hz to kHz) excitation, for example, has longer characteristic length scales over which the acoustic energy decays, thus allowing greater penetration, particularly through skin and tissue, and thereby enabling their use for in vivo applications. A corollary of the use of low frequency mechanostimulation, however, is the existence of cavitational effects, which, besides exerting considerable stress on cells, can also induce poration along the cell membrane. The larger intensities and prolonged exposure durations that are typically required with low frequency acoustic excitation can also introduce considerable thermal effects that could potentially induce the cells to produce heat shock proteins.

These effects can be circumvented by operating at higher frequencies (MHz order and beyond), whose wavelengths are more closely matched to the characteristic cell dimension and hence allows for localization of the mechanostimulation signal, thereby reducing the intensity required to elicit the same effect. This reduction in power and hence heating as a consequence of absorption of the acoustic energy, together with the diminished propensity for cavitation, allows significantly greater retention in cell viability. The considerably shorter penetration depths associated with high frequency excitation however precludes their use for in vivo applications unless implanted at the local site to be targeted. Nevertheless, the much shorter exposure durations that have been demonstrated with the use of surface and hybrid waves—which have been shown to facilitate both efficient poration-free cellular uptake as well as exosome release in addition to stem cell differentiation and endothelial barrier modeling—together with the possibility for parallelization of the chipscale technology can be appealing for large-scale cellular engineering.

In addition to providing an overview of the advances of acoustically driven cell mechanostimulation across these frequencies to date, and their utility for various applications, particularly in, but not limited to, regenerative therapeutics and nanomedicine, we postulate that the mechanoresponses stemming from such stimuli, particularly those driven at ultrasonic frequencies across 10 kHz–1 GHz, arise from a universal mechanotransductive mechanism regardless of the considerable differences between the underlying mechanostimulation modalities (e.g., constant force, static or quasi-static stimuli such as compression, tension or shear, or, more complex dynamic stimuli involving time-dependent vibrational forcing in the form of bulk, surface or hybrid acoustic waves) across a wide range of frequencies.

Broadly, the cellular response to acoustic forcing, and, mechanostimulation, in general, arises as an outcome of the cells' attempt to maintain both membrane and Ca²⁺ homeostasis. We posit that the cell membrane plays a central mechanosensory role and that the mechanoresponses that are observed can be attributed to the effect of the mechanical cue, whatever its form, imparted on the cell membrane. Be it static compression or tension (stretching) of the membrane, or a dynamic combination of these constituents as a consequence of the time-dependent, i.e., periodic, vibrational forcing associated with the acoustic excitation, the resultant membrane aberrations (e.g., VLDs) or pores that arise lead to a change in the membrane tension that then drives the activation of SACs. It is primarily these SACs, together with auxiliary gating mechanisms such as other transmembrane proteins (e.g., integrins) and cell junctional proteins (e.g., connexins and cadherins), that facilitate the transmission of the imparted mechanical force through the rest of the cell via its cytoskeletal network to trigger a cascade of downstream molecular signaling events.

More specifically, we observe the second messenger Ca²⁺ to play a central role in maintaining cellular homeostasis in response to the mechanical insult. In the first Ca²⁺ mobilization stage, Ca²⁺ influx into the cell facilitated by the SAC and cell junctional proteins trigger the activation of a series of signaling cascades that include Rho–ROCK signaling and ESCRT pathways. In the second stage of Ca²⁺ mobilization, the aforementioned influx of Ca²⁺ into the cell as a consequence of the mechanical load is regulated by channeling it into internal stores primarily through SERCA pumps or returning it to the extracellular milieu through calcium transporters on the membrane. Upon Ca²⁺ overload in the internal store, Ca²⁺ is further released through RyR activation, which results in the production of other second messengers such as cAMP and IP3. These second messengers further initiate other signaling cascades (e.g., the Epac1-Rap1 pathway) to facilitate activation of transcription factors, including c-Jun, c-Fos, or ERK1/2, which regulate a diverse set of cellular activities such as cell proliferation, migration, and exosome generation, in addition to other cell-specific processes such as stem cell differentiation, endothelial barrier modulation, neuromodulation, and oocyte activation.

The translation of mechanostimulation to practical living systems for clinical therapeutic applications, however, remains a challenge. Unlike their in vitro counterparts, in vivo systems are considerably more complex due to the lack of appropriate models and means to study the collective behavior arising from the number of different cell types that are simultaneously stimulated, and the multitude of mechanoresponses over widely varying response timescales that can arise.³⁹¹ As such, there is, at present, insufficient conclusive evidence for coordinated mechanotransduction in vivo, and a thorough understanding of the underlying mechanisms that are involved is still elusive.^392,393 In addition, the vast number of systems and protocols associated with multicellular bodies make it significantly more difficult to compare and interpret the data required to upscale these platforms for practical applications. Nevertheless, a better understanding of the in vitro mechanotransduction mechanisms stemming from that discussed in this review would hopefully constitute a start in advancing the technology toward clinical translation.

L.Y.Y. is grateful for funding through Australian Research Council Discovery Project DP210101720. Some images were created using BioRender.

The authors have no conflicts to disclose.

Lizebona August Ambattu: Conceptualization (equal); Data curation (lead); Writing – original draft (lead). Leslie Yeo: Conceptualization (equal); Funding acquisition (lead); Methodology (equal); Project administration (lead); Resources (lead); Supervision (lead); Validation (lead); Writing – review & editing (lead).

Data sharing is not applicable to this article as no new data were created or analyzed in this study.