Ultrasound‐mediated therapies for the treatment of biofilms in chronic wounds: a review of present knowledge

Summary Bacterial biofilms are an ever‐growing concern for public health, featuring both inherited genetic resistance and a conferred innate tolerance to traditional antibiotic therapies. Consequently, there is a growing interest in novel methods of drug delivery, in order to increase the efficacy of antimicrobial agents. One such method is the use of acoustically activated microbubbles, which undergo volumetric oscillations and collapse upon exposure to an ultrasound field. This facilitates physical perturbation of the biofilm and provides the means to control drug delivery both temporally and spatially. In line with current literature in this area, this review offers a rounded argument for why ultrasound‐responsive agents could be an integral part of advancing wound care. To achieve this, we will outline the development and clinical significance of biofilms in the context of chronic infections. We will then discuss current practices used in combating biofilms in chronic wounds and then critically evaluate the use of acoustically activated gas microbubbles as an emerging treatment modality. Moreover, we will introduce the novel concept of microbubbles carrying biologically active gases that may facilitate biofilm dispersal.

Bacterial biofilms are an ever-growing concern for public health, featuring both inherited genetic resistance and a conferred innate tolerance to traditional antibiotic therapies. Consequently, there is a growing interest in novel methods of drug delivery, in order to increase the efficacy of antimicrobial agents. One such method is the use of acoustically activated microbubbles, which undergo volumetric oscillations and collapse upon exposure to an ultrasound field. This facilitates physical perturbation of the biofilm and provides the means to control drug delivery both temporally and spatially. In line with current literature in this area, this review offers a rounded argument for why ultrasound-responsive agents could be an integral part of advancing wound care. To achieve this, we will outline the development and clinical significance of biofilms in the context of chronic infections. We will then discuss current practices used in combating biofilms in chronic wounds and then critically evaluate the use of acoustically activated gas microbubbles as an emerging treatment modality. Moreover, we will introduce the novel concept of microbubbles carrying biologically active gases that may facilitate biofilm dispersal.

The bacterial biofilm: development and aetiology
Though the microbial world is vastly diverse, the development of a biofilm remains perhaps the most ubiquitous means by which microbial cells can thrive within their given environment (Wu et al., 2015;Flemming et al., 2016). A biofilm can be described as a localized aggregation of microorganisms in a heterogeneous, sessile community, embedded in a dynamic matrix of extracellular polymeric substances (EPS) (Singh et al., 2017). From biogeochemical cycling in the ecosystem and the human microbiome to biofouling and disease, biofilms are simultaneously an essential part of life and a prominent concern for industry and public health (Donlan, 2002;Flemming et al., 2016;Kuliasha et al., 2017). The gross architecture of the biofilm is complex; proteomic investigation has shown that at least in Pseudomonas aeruginosa, biofilm development is regimented and sequential ( Fig. 1) (Hall-Stoodley et al., 2004). Though the specific stages of biofilm development have not been characterized for each prokaryotic organism individually, it should be noted that the development archetype is markedly conserved in vitro (Figueiredo et al., 2017;Lohse et al., 2017;Bartell et al., 2019).
The initial stages of biofilm formation are characterized by an impermanent association of planktonic cells with a surface, which is superseded by a robust adhesion and the production of EPS (Hall-Stoodley et al., 2004;Holt et al., 2017). Although proteins, carbohydrates, lipids, DNA, RNA and water are stable constituents of the EPS (Taglialegna et al., 2016), there is a notable variability in the quantity of each macromolecule present between biofilms (Flemming et al., 2007). The mucoid phenotype of P. aeruginosa is found extensively in the cystic fibrosis lung and usually arises from the overproduction of the exopolysaccharide alginate (Schurr, 2013). This serves as a good example of the individual differences seen in the EPS of biofilms, exemplifying that variation can be both species-specific and influenced by the immediate environment (Maleki et al., 2016). Adherent cells will then begin to propagate and assemble into microcolonies; the initiation of this process is attributed to monomeric adhesins (Jamal et al., 2015). Aggregation and microcolony formation in P. aeruginosa are attributed to the Type IV pili surface adhesin, whereas in Staphylococcus aureus, it is thought to be induced by polysaccharide intercellular adhesin (Persat et al., 2015;Maleki et al., 2016). Maturation of a biofilm into its distinctive three-dimensional structure is achieved through sophisticated cell-cell communication termed quorum sensing (Laganenka and Sourjik, 2017). Quorum sensing is facilitated by the production of auto-inducer signalling molecules, which allow microbial cells to determine cell density and collectively adjust gene expression in response (Rutherford and Bassler, 2012). This ensures the formation of an encapsulating extracellular matrix, with water-filled channels for the transport and exchange of nutrients and waste products within the biofilm (Parsek and Singh, 2003). Programmed dispersal of microbial cells is the final stage of biofilm development, which is achieved by either the release of newly formed cells from biofilm aggregates or detachment of constituent peripheral cells by species-specific saccharolytic enzymes (Marsh and Zaura, 2017). Erosion and sloughing may also occur due to mechanical (i.e. shear) stress upon the biofilm, which causes peripheral cells to disengage indiscriminately from the biofilm and enter the local environment (Rmaile et al., 2014;Jamal et al., 2018). Though dispersed cells regain motility, they remain physiologically unique from cells in the planktonic and biofilm phase; these dispersed cells are highly virulent in nature towards macrophages, which is a useful attribute given their main purpose is the colonization of new sites (Chua et al., 2014). Dispersion represents one of the most virulent stages in the biofilm life cycle, but perhaps also one of the easiest to target and thus potentiate killing of microbial cells (Hall and Mah, 2017).

Significance of biofilms in chronic wound infections
The causality between pathogenic microorganisms and infection has been understood for over a century; yet, most research into the pathogenesis of microorganisms has focused solely on acute infection by planktonic cells. Over the last decade, this focus has shifted; greater emphasis is now placed on the role of multidrug-resistant (MDR) organisms and biofilms, which mediate over 90% of chronic wound infections (Attinger and Wolcott, 2012;Bjarnsholt, 2013). The augmented persistence of biofilms can be attributed in part to inherited genetic antibiotic resistance traits, which actively reduce the efficacy of an administered antimicrobial agent. This (1) Initial attachment can be transient, but this association can become robust and lead to the aggregation of cells (2). This stabilized attachment leads to the production of extracellular polymeric substances (EPS), which encapsulate aggregated cells forming microcolonies (3). The maturation of the biofilm structure is achieved through intercellular signalling (4); a mature biofilm commonly features a concentration gradient of oxygen and nutrient availability. Oxygen is consumed by biofilm cells at a faster rate than it can diffuse in, which coincides with the gradient of nutrient availability. Consequently, the majority of metabolically active cells are located at the periphery of the biofilm, whilst persister, dormant or dead cells are found at the biofilm-substratum interface. In its final stages, the biofilm undergoes programmed dispersal of cells, which includes cellular mass and EPS sloughing off (5). commonly includes the use of membrane-associated efflux pumps, which prevent antimicrobial agents reaching lethal intracellular concentrations, and antibiotic degradation enzymes such as beta-lactamase, which alter the pharmacokinetic properties of beta-lactam antibiotics (Høiby et al., 2011;Geisinger and Isberg, 2017). However, of particular concern is that when compared to their planktonic counterparts, it has been conclusively shown that the biofilm phenotype confers an innate physical tolerance to antimicrobial agents (Hengzhuang et al., 2012;Algburi et al., 2017). It is also noted that the extracellular polysaccharides within the dynamic EPS matrix potentiate this effect by acting as a shield, compromising the ability of the host immune system to detect the biofilm infection (Limoli et al., 2015;Kumar et al., 2017). Moreover, the production of extracellular toxins and lytic enzymes facilitates destruction of local immune cells, which provides a source of cellular components that can be utilized by microbial cells (Cooper et al., 2014).
It is evident that current treatment options available for biofilms are limited in both availability and effectiveness; thus, seeking to resolve a chronic infection by eradicating a competent biofilm formed in a wound bed is a multifactorial challenge. In response to a wound derived from acute trauma, tissue will undergo a sequential process of reparation that results in the reinstitution of anatomical integrity (Clark, 1993). Under normal physiological conditions, the progression of wound healing can be broadly categorized into four phases: haemostasis, inflammation, proliferation and remodelling (Guo and DiPietro, 2010). The phases within the process of wound repair are not mutually exclusive, but involve dynamic integration of cellular processes that overlap temporally (Robson, 2004). Diabetic foot ulcers (DFU) are a prevalent example of a chronic wound; commonly arising from comparatively minor trauma to the foot, they have been strongly associated with substantial morbidity and mortality (Walsh et al., 2016). The key pathophysiological difference between a common acute wound and a chronic wound is that the latter is typically associated with recalcitrant infection, ischaemia of the tissue and a prolonged or arrested inflammatory phase . One of the hallmarks of a chronic wound is high microbial burden and diversity, which is routinely attributed to the formation of poly-microbial drugresistant biofilms in the wound bed (Banu et al., 2015). Given the pathogenicity and associated virulence factors of biofilms, there is credible evidence that they are implicit in preventing normal mechanisms of wound healing (Malone et al., 2017). It is reported that every 30 s worldwide, there is a lower-limb amputation as a direct result of DFU (Yazdanpanah et al., 2015). This is undoubtedly accompanied by significant physical and emotional stress, as well as an increase in mortality rate (Costa et al., 2017).
In addition, there is an undeniable economic burden associated with the cost of health care, from disease management to major intervention (Walsh et al., 2016).

Current treatment strategies for biofilms in chronic wounds
The most established treatment for the removal of necrotic tissue and biofilms from chronic wounds is sharp debridement, but this mechanical method of biofilm disruption lacks both efficiency and effectiveness (Cooper et al., 2014;Yazdanpanah et al., 2015). Although the debridement of chronic wounds in clinical trials is largely concurrent with a reduction in the surface area of a wound, the period over which intervention is required is typically in the order of weeks to months and does not significantly correlate with complete wound closure (Williams et al., 2005;Rhoads et al., 2008;Wolcott et al., 2009;Yazdanpanah et al., 2015). Studies have shown that debridement can expedite wound healing by stimulating re-epithelialization of the tissue; however, complete healing is typically observed in less than 20% of patients (Cardinal et al., 2009;Kim et al., 2018). The presence of persister cells allows the regeneration of the biofilm within the wound bed, which means that debridement is by no means a complete or permanent solution (Lebeaux et al., 2014). The efficacy of debridement can be improved by chemical and biological adjuvants, such as hydrogen peroxide and enzymes respectively (Watters et al., 2016). By causing the EPS matrix of the biofilm to degrade, and thus removing its principal means of protection and nutrition, the rate of wound healing is significantly increased (Kim et al., 2018). The physical perturbation of the biofilm caused by debriding has also been shown to temporarily restore antibiotic sensitivity; as the biofilm begins to regenerate, key antibiotic targets such as cell wall synthesis (glycopeptides) and protein synthesis (aminoglycosides) become viable (Wolcott et al., 2009;Hall and Mah, 2017).
Quorum sensing is an important regulator of biofilm development; it is the principal means by which microbial cells communicate within a given environment (Miller and Bassler, 2001;Rutherford and Bassler, 2012). The ability of microbial cells to carry out such sophisticated communication is a potent advantage; therefore, quorum sensing can also be considered as a valuable therapeutic target (Singh et al., 2017). By incapacitating this signalling mechanism, the regulation of gene expression, essential metabolic processes and virulence can be irreparably altered (Khmel, 2006). Natural and synthetic inhibitors of quorum sensing, such as furanones and Manuka honey, work by downregulating four major quorum sensing genes, which in turn has downstream consequences for genes associated with the biofilm phenotype (Wang et al., 2012;Jakobsen et al., 2018). However, it is important to note this is principally a method to render a biofilm more inert; it does not offer an immediate solution to detachment or physical eradication (LaSarre and Federle, 2013). An additional means of controlling biofilms in chronic wounds is impeding their attachment to surfaces; one way in which this has been achieved is with the ironchelating glycoprotein, lactoferrin (Garc ıa-Montoya et al., 2012). Lactoferrin is an important endogenous antimicrobial component of the innate immune system; it is principally found in tears, saliva, mucous secretions and breast milk of mammals (Cooper et al., 2014). Its most potent properties include sequestering iron essential for bacterial motility, and direct interaction with Gram-negative bacterial cell walls to induce lysis at the site of infection (Valenti et al., 2015). This consequently means that its bacteriostatic effect can prevent biofilm development, but also disrupt cells which have already become adherent. Lactoferrin has the additional benefit of exhibiting anti-inflammatory properties, which may play an important role in mitigating chronic inflammation associated with delayed wound healing (Valenti et al., 2017). Allison et al. (2015) showed that as a component of breast milk, lactoferrin at a concentration of 3 mg ml À1 significantly decreased Streptococcus mutans biofilm formation in vitro. The artificial sweetener xylitol has been shown to bind to Gram-positive organisms, preventing the organism from adhering to other cells or surfaces (Ferreira et al., 2015). The co-administration of lactoferrin and xylitol has shown great efficacy in eliminating Pseudomonas aeruginosa biofilms in vitro (Ammons et al., 2011a(Ammons et al., ,2011b. The acquisition of iron by lactoferrin causes membrane disruption, whilst xylitol prevents P. aeruginosa biofilms successfully responding to the environmental change . This treatment modality has been successfully implemented for the treatment of chronic wounds; a lactoferrin and xylitol hydrogel, in conjunction with a silver wound dressing, were shown to achieve a significant log reduction in Staphylococcus aureus and P. aeruginosa biofilm viability of 6.5 AE 2.4 and 4.9 AE 0.9 respectively (Ammons et al., 2011a(Ammons et al., ,2011b. However, this method is comparably only as efficacious as sharp debridement and still requires treatment over a number of weeks to months, but may provide a means to prevent recalcitrant and exacerbated infection (Kim et al., 2018). By avoiding mechanisms of microbial resistance, quorum sensing inhibition, bacteriostasis and adhesion prevention are pertinent alternatives to traditional antimicrobial therapies. To meet the multifactorial challenge presented by biofilms in chronic wounds, this review proposes that gas-filled microbubbles (MBs) can be the versatile biomedical tool required.

Gas microbubbles: a method of controlled drug delivery
The architecture of gas-filled microbubbles is variable according to their intended application, though they commonly consist of a surfactant, polymer, protein or phospholipid shell, which encapsulates a gaseous core ( Fig. 2) (Sirsi and Borden, 2009;Owen et al., 2018). The composition of the MB shell is integral to conferring mechanical stability, preventing coalescence and determining its acoustic response to stimulation by ultrasound (US) (Borden et al., 2005;Stride, 2008;Carugo et al., 2017). Characterizing MB size is an important step for determining not only its acoustic response and drug loading capacity, but also its longevity in circulation and thereby its safety for in vivo applications (Lee et al., 2015b). MBs are typically manufactured with a diameter distribution in the range between 1 and 10 lm, and the mean MB diameter during storage increases over time (Ferrara et al., 2007). Notably, there exists a pressure difference between the inside of a MB and the outer environment (known as Laplace pressure), which is caused by the surface tension of the curved gas-liquid interface. For a given MB shell formulation, the Laplace pressure is inversely proportional to MB radius. Therefore, gas diffuses from the smaller bubbles to the larger ones, leading to dissolution and disappearance of the smaller bubbles. It is well documented that factors such as shell composition, fabrication method used, the relative chemical environment and temperature can substantially influence MB size and its temporal evolution (Mulvana et al., 2010;Sun et al., 2014;Lee et al., 2015b;Taylor et al., 2017).
Phospholipids are perhaps the most common excipients of the MB shell; key examples include distearoylphosphatidylcholine (DSPC) and dipalmitoylphosphatidylcholine (DPPC) (Swanson et al., 2010). The lipid molecules assemble in a monolayer at the hydrophilichydrophobic interface to stabilize the gas core, exposing their hydrophilic polar head to the aqueous environment and their hydrophobic tails towards the gaseous core (Fig. 2). The acyl chain length of a lipid dictates the permeability of the MB shell to gases (Borden, 2016); longer hydrophobic acyl chains provide greater cohesion (or packing density) between adjacent lipids (Hosny et al., 2013), consequently reducing MB shell permeability to gases and increasing its stability during storage and upon administration (Zhuang et al., 2016). Surfactants can be integrated into the MB shell, which is particularly important for biomedical applications of MBs such as drug delivery (Abou-Saleh et al., 2014). The addition of the surfactant polyethylene glycol (PEG) to a MB shell simultaneously provides the means to functionalize the MB shell with biological components or assemblies such as targeting ligands, antibodies and liposomes (Abou-Saleh et al., 2014), whilst mitigating immunogenicity to lipid antigens and preventing phagocytosis of the MB by macrophages (Paolino et al., 2017). The content of the MB gaseous core is arguably just as important as the composition of the encapsulating shell, as it dictates some of the MB properties (Vohra and Jasuja, 2016). High molecular weight perfluorocarbon compounds and sulfur hexafluoride are commonly used as the gaseous core for MBs in vivo; the poor water solubility and low diffusion rate of fluorinated gases prolong MB longevity by enhancing stability (Casini et al., 2016;Carugo et al., 2017). The drug loading capacity of a MB is strongly linked to the efficacy of the treatment, as it directly impacts the amount of a given therapeutic agent that can be delivered to a target site (Tzu-Yin et al., 2013). The use of electrostatic force to bond drugs to the MB surface (Rychak and Klibanov, 2014), insertion into the MB shell (Lentacker et al., 2009), loading the drug into the gaseous void and placing a layer of oil at the interface between the gaseous core and MB shell (Tinkov et al., 2009) are all considered low-capacity methods of drug loading (Sirsi and Borden, 2009). To yield a greater MB drug loading capacity, it is common for nanoparticles or liposomes to be conjugated to MBs covalently or with the use of biotin-avidin bridges (Lentacker et al., 2010;Liang et al., 2018). However, methods of achieving a greater drug loading capacity also directly impact upon the stability and acoustic properties of the MB, due to their altered shell thickness and composition (Tzu-Yin et al., 2013;Kooiman et al., 2014). Drug loading modalities, methods of microbubble fabrication and their biophysical effects have been extensively reviewed elsewhere (Unger et al., 2002;Stride and Edirisinghe, 2008;Fix et al., 2015;Lee et al., 2015b).

Ultrasound-mediated physical effects of gas microbubbles
Manipulation of the MB fabrication method and shell composition dictates their physico-chemical properties and size, allowing them to elicit different modes of action in response to acoustic stimulation. These ultrasoundmediated behaviours of the MBs can be further regulated by adjusting the parameters of the US delivered (e.g. frequency, acoustic pressure, duration, pulse repetition frequency); this makes MBs perhaps one of the most versatile tools available in biomedicine. In biomedical applications of MBs, their efficacy for a given task is correlated to their acoustic response (Datta et al., 2006;Choi et al., 2014), which is typically categorized into either stable or inertial cavitation (Lentacker et al., 2014). In response to the pressure changes of ultrasound waves, MBs experience alternating volumetric compression and expansion (or rarefaction) (De Jong et al., 2002). Upon exposure to low-intensity US, the nature of these oscillations is typically repetitive over several US cycles; this behaviour is referred to as stable cavitation (Stride and Coussios, 2009). Above a critical US intensity, the periodicity of this oscillation is lost; MBs expand rapidly, and the inertia of the surrounding fluid during contraction causes them to collapse violently (Wu and Nyborg, 2008;Stride and Coussios, 2009). This process is known as inertial or transient cavitation and often leads to MB fragmentation into smaller bubbles. It should be noted that whilst there is a simple relationship between US intensity and pressure for a plane travelling wave, it is more complex in 2D fields and standing wave fields. The propensity for fragmentation of lipid-shelled MBs is inversely correlated to the phospholipid alkyl chain length and also depends upon the type of emulsifier used (Borden et al., 2005). Notably, the MB resonant frequency and amplitude of oscillation, the transition pressure from stable to inertial cavitation, and MB fragmentation diameter all strongly depend upon the initial MB size (for a given shell formulation and suspension medium) (Povey et al., 1998;Borden et al., 2005). Therefore, the selection of a specific US frequency to elicit a desired acoustic response should take into account the typically broad size distribution of a MB suspension and variations in MB size during sample storage and/or handling.
Concerning the biophysical effects of MBs, it has been postulated that the systematic expansion and compression of MBs in stable cavitation create localized pushing and pulling forces, which in turn cause disruption to the integrity of cell membranes located in their vicinity (Fig. 3B) (Lee et al., 2015a). It is also possible for MB oscillation to drive a steady flow of the surrounding fluid (also known as cavitation microstreaming). The volumetric oscillation of MBs generates flows that are purely divergent (i.e. radial), whilst interaction with a dissimilar surface (e.g. a target tissue) can generate a circulatory flow that enhances shear stress over nearby cells, potentially causing transmembrane pores to form (Fig. 3A) (Ferrara et al., 2007). Furthermore, the streaming flow field can drive shedding of shell constituentssuch as therapeutic compoundsaway from a MB (Luan et al., 2014), which in turn can be exploited as a mechanism to deposit (or 'print') therapeutic material over the membrane of target cells (De Cock et al., 2016). Microjet formation and shockwaves are more transient physical effects attributed to the collapse of MBs in inertial cavitation, which respectively puncture proximal membranes and increase membrane permeability through mechanical stress ( Fig. 3C and D) (Collis et al., 2010). Although inertial cavitation can release energy in the form of heat, this is rapidly dissipated in the surrounding fluid, which has a significantly greater total volume than the volume occupied by MBs (Ye et al., 2013). In comparison with the mechanical stresses imparted by inertial cavitation, it could be inferred that stable cavitation is a comparatively gentle means of facilitating drug uptake.

The use of ultrasound alone as an anti-biofilm treatment
The ability of low-frequency US to improve the efficacy of antibiotics was shown as early as 1994; the minimum inhibitory concentration (MIC) of aminoglycoside and macrolide antibiotics was consistently reduced by up to 50%, in planktonic cultures of P. aeruginosa and S. aureus exposed to continuous US at 67 kHz and 0.3 W cm À2 intensity (Pitt et al., 1994). This was later supported in a study by Runyan et al. (2006), who demonstrated that US both potentiates antibiotics against planktonic cultures and biofilms by facilitating the transport of antibiotics into biofilms. The prevailing theory is that US increases the permeability of cell membranes to systemically available antibiotics, without physically disrupting or dispersing the biofilm (Mohammad et al., 2015). The efficacy of US as an adjuvant to antibiotics can be supported by a study conducted by Pitt and Ross (2003), which showed that in the absence of antibiotic low-frequency US (70 kHz) of < 2 W cm À2 acoustic intensity, increased the growth rate of planktonic Escherichia coli and P. aeruginosa. It can be reliably deduced that by the same means, US potentiates antibiotics by promoting transport across cell membranes; in the absence of antibiotics, US may increase the rate at which waste products are removed and metabolites are transported to cells, consequently enhancing their growth.
Investigations conducted in vivo on the effect of adjuvant US on biofilms implanted subcutaneously in rabbits showed that after 24 h of continuous US exposure (28.48-kHz, 100 and 300 mW cm À2 ), recovered E. coli colony-forming units (CFU) were significantly reduced, whereas there was no observable effect on P. aeruginosa (Rediske et al., 1999). The literature in this area is clear that US alone has no effect on cell viability, though they do lack congruency in the US frequency, intensity, and pulse length or repetition frequency used (Jiang et al., 2016;Cai et al., 2017). The principal limitation of this approach is that it currently lacks translation to a viable clinical application; the most efficacious low-frequency US (28.48 kHz) and intensity (100 and 300 mW cm À2 ) used in vitro were shown to induce tissue damage when applied in vivo (Rediske et al., 1999;Jiang et al., 2016). It should also be noted that the majority of studies have continuous treatment times of 24-48 h, which is unlikely to be considered economically or socially practical. Moreover, this method of utilizing the synergistic relationship between US and antibiotic efficacy does not allow for targeted or controlled delivery of the antibiotic to a localized region, since it relies on the traditional oral or intravenous administration of antibiotics of sufficient dose to ensure an effective serum concentration. The limitation of this method is that any antibiotic administered in this manner would still have a non-specific impact on unintended systemic targets, which would include dysbiosis of the host commensal microbiota (Carding et al., 2015). Due to the localized release of antibiotics conjugated with microbubbles, the dosage required to achieve an effective local antibiotic concentration is significantly lower than orally or intravenously administered antibiotic (Horsley et al., 2019). In quantitative terms, the typical recommended dosage for systemic administration of aminoglycoside and b-lactam antibiotics ranges from 2 to 16 g day À1 (Taccone et al., 2011), whilst the concentration associated with microbubble administration is typically of MIC (i.e. lg ml À1 ) (Pitt et al., 2004;Zhu et al., 2014). Therefore, the residual serum concentration of antibiotics delivered by microbubbles is much lower, likely reducing or removing non-specific systemic targets.
The application of low-frequency (20-60 kHz) ultrasound for chronic wound debridement has shown great potential, with recent advancements extensively reviewed elsewhere (Chang et al., 2017;Liu et al., 2017). To summarize, ultrasonic debridement has been shown to potentiate antibiotics against bacteria within the chronic wound, emulsify dead cells within a localized area and stimulate peripheral healthy cells to promote the healing process. The efficacy of this treatment modality has been assessed in a number of clinical trials, which generally concur that ultrasonic debridement is a valuable wound care adjuvant (Amini et al., 2013;Murphy et al., 2018). It is important to note that although these trials have demonstrated that ultrasonic debridement improves short-term treatment outcomes, the frequency and total duration of the treatment are comparable to non-surgical sharp debridement with no significant difference in healing rate between the modalities after 6 months of treatment (Amini et al., 2013;Michailidis et al., 2018).

Acoustically activated gas microbubbles for the treatment of bacterial biofilms
The success of low-frequency US (20-100 kHz) in facilitating the uptake of systemic antibiotics by biofilms has been variable; therefore, the use of US-responsive MBs in the light of their controllable physical response may be viewed as an evolution in antimicrobial drug delivery. Alteration in the permeability of biofilms to macromolecules, such as antimicrobial compounds, has been routinely demonstrated with the use of the red-fluorescent nucleic acid stain, propidium iodide (PI). Dong et al. (2017) demonstrated that acoustically activated MBs were capable of enhancing the permeability of Staphylococcus epidermidis biofilms grown in OptiCell TM chambers in vitro. The fluorescent signal emitted increases proportionally with the quantity of PI that has intercalated with DNA; therefore, it can be utilized as a direct means of assessing how MBs facilitate uptake of macromolecules (Stiefel et al., 2015). It could be therefore hypothesized that US-activated MBs facilitate the uptake of antibiotics by biofilms, principally through increasing membrane permeability of bacterial cells and via heterogeneous alterations to the biofilm architecture, which can include the development of pores in the EPS matrix Jang et al., 2017;Hu et al., 2018). Fluid shear stress has also been shown to significantly affect biofilm morphology; at shear stresses under 1 Pa, biofilms are shown to grow in a laboratory-typical mushroom-like shape with interstitial channels and voids (Salta et al., 2013). However, when the shear stress overcomes the biofilm adhesion strength (> 2 Pa), erosion and sloughing of the biofilm are more pronounced, which leads to low-density biofilm aggregates with large interspaces (Rmaile et al., 2014). Shear stress in terms of human physiology is variable according to location; arteries are typically subject to average wall shear stress of 1-3 Pa, whilst in arterioles and capillaries it is commonly of 2-6 Pa (Sheikh et al., 2003;Shaik et al., 2009). The effect of fluid shear stress on the detachment of biofilms has shown to be compounded in the presence of microbubbles, clearing up to 70% of biofilm in a speciesdependent manner in vitro (Sharma et al., 2005). It has been shown that the in vitro application of 0.08 MHz, 1.0 W cm À2 , 50% duty cycle and 10 min duration US, MBs and vancomycin can significantly decrease the number of viable cells (7.17 log 10 CFU ml À1 ) from S. epidermidis biofilms compared to an untreated control (10.51 log 10 CFU ml À1 ) (He et al., 2011). It is important to note, however, that this study could not demonstrate a significant difference between these groups using an in vivo rabbit model, in which S. epidermidis biofilms grown on polyethylene discs were subcutaneously implanted bilateral to the vertebral column. The work carried out by He et al. (2011) demonstrated that US-activated sulfur hexafluoride MBs with a mean diameter of 2.5 lm and vancomycin create micropores within the biofilm architecture, which does support the hypothesis of facilitated uptake mediated by membrane/EPS disruption. An interesting point is that not only does the membrane disruption facilitate antibiotic uptake, but the influx of nutrients may induce a phase of active growing in deeper layers of the biofilm, potentiating the efficacy of the antibiotic . In addition, studies have consistently shown that the combination of US and MBs is capable of halving the MIC of the administered antibiotic (Kasimanickam et al., 2013). This has been supported in recent work by Horsley et al. (2019) in which gentamicin-loaded liposomes bound to ultrasoundresponsive microbubbles were utilized to significantly enhance direct antibiotic delivery to intracellular uropathogenic bacteria. The ultrasound-mediated delivery of gentamicin in concentrations of 0.53-1.32 lg ml À1 with a 20 s exposure time showed an equivalent efficacy in bacterial killing to a 2 h exposure to free gentamicin at the significantly higher clinically approved dosage of 200 lg ml À1 (Horsley et al., 2019). Moreover, the ultrasound-mediated therapy achieved a 75% greater reduction in bacterial bioburden than free gentamicin alone, with no evidence of cellular damage (Horsley et al., 2019). This effect is perhaps indicative of the mechanical action of the oscillating microbubble, aiding the physical detachment of bacteria from proximal surfaces. This work has exemplified the utility of ultrasound-mediated intracellular delivery of antimicrobial agents, as a viable alternative to the use of orally administered antibiotics. It should not be overlooked that to date, research on enhancing efficacy of US-activated microbubbles has only been performed on naive single-species biofilm models. Furthermore, although the biophysical effect of acoustically activated MBs is evident, the underlying mechanisms of interaction between the bubbles and the biofilm have not been elucidated yet.

The applications of nitric oxide for the treatment of bacterial biofilms
Nitric oxide (NO) has been utilized to facilitate healing of chronic wounds such as DFU for a number of years; this is motivated by the role of NO as an important biological signalling molecule (Witte and Barbul, 2002). Cellular proliferation, angiogenesis and remodelling are key biological and physiological processes mediated by NO, which have been principally applied to wounds in the form of inducible NO synthase (Dhall et al., 2014). It is important to note that traditionally NO has only been applied to chronic wounds in the context of tissue repair, and not with the specific intention to treat the underlying biofilm (Saidkhani et al., 2016;Han and Ceilley, 2017). The administration of NO in a therapeutic capacity has historically been difficult; at high concentrations, NO is bactericidal and cytotoxic, which significantly impairs the progression of the normal healing process (Schulz and Stechmiller, 2006). Studies of the biofilm life cycle have elucidated endogenous mechanisms which can be exploited as therapeutic targets, these principally involve the use of NO in the sub-micromolar range to induce the biofilm dispersal phase (Barraud et al., 2015). By inducing dispersal of the biofilm with NO, the physical barrier imposed by the EPS matrix can be negated entirely.
Research has shown that the dispersed cells are considerably more susceptible to antimicrobial treatments; it can therefore be inferred that adjuvant NO potentiates antibiotics against biofilms (Howlin et al., 2017). Utilizing the spontaneous NO donor sodium nitroprusside (SNP), Howlin et al. (2017) successfully showed that NO disrupted P. aeruginosa biofilms from cystic fibrosis sputum samples in vitro. The same study also highlighted the importance of dispersal as a means of therapy, since the administration of the antibiotic tobramycin alone caused a significant increase in biomass and biofilm thickness compared to untreated controls. Xu et al. (2017) have proposed and tested the implementation of NO-releasing agents for indwelling medical device surfaces, and they were able to successfully prevent biofilm formation on a functionalized polyurethane surface. Utilizing NO to prevent bacterial growth and adhesion on implanted surfaces in conjunction with antimicrobial therapy may result in better clinical outcomes for patients and significant cost savings for health service providers. There is relatively little published data either in vivo or in vitro, on the successful implementation of exogenous NO in a gaseous form for biofilm dispersal. The inhalation of NO has been an FDA-approved therapeutic agent for nearly two decades, and clinical trials have shown that NO gas at 5-10 ppm can achieve a 3.5 log reduction in respiratory P. aeruginosa biofilm aggregates (Howlin et al., 2017). The major problem with the delivery of NO in a gaseous form is that it is highly reactive, which translates to a half-life of only seconds (Thomas et al., 2001). A solution to this challenge is to utilize an US-responsive agent such as the MB, which could provide the means to successfully control the delivery and release of NO.
The mechanism of action for nitric oxide-induced dispersal of biofilms The control of dispersal events in the biofilm life cycle is linked to the intracellular second messenger molecule cyclic-di-guanosine monophosphate (c-di-GMP), which is regulated by cellular phosphodiesterase (Reinders et al., 2016). Recent studies have shown that the activity of cellular phosphodiesterase is increased in the presence of NO at concentrations in the pico-and nanomolar range, which results in the degradation of c-di-GMP and subsequent dispersal of the biofilm (Algburi et al., 2017;Howlin et al., 2017). The dispersed cells return to a motile state and become susceptible to antimicrobial treatment; this effect is shown to be conserved across species such as P. aeruginosa and Staphylococcus aureus and in both single and complex multispecies biofilms (Barraud et al., 2009). Dispersal has been achieved with NO concentrations as low as 450 pM, reaching peak efficacy at 450 nM (Howlin et al., 2017).

Biologically active nitric oxide gas microbubbles and their applications
Though high molecular weight gases are most commonly used as the MB core constituent, it is also possible to generate MBs that possess a biologically active gaseous core. Recent research has already highlighted the advantages of using MBs with an oxygen core, to increase the therapeutic efficacy of chemo-sonodynamic therapy (SDT) in the targeted treatment of solid malignant tumours (McEwan et al., 2015;Nesbitt et al., 2018). Bioactive gases such as NO have significant therapeutic potential, as they mediate a number of important biological signal pathways (Basudhar et al., 2016). However, the molecule is highly reactive with both exogenous molecules, such as oxygen, and endogenous scavengers such as haemoglobin (Azarov et al., 2005). This can be mitigated by utilizing the gas as the core of acoustically active MBs; protected by the MB shell, the release of NO can then be both spatially and temporally controlled by US (Fig. 4) (Fix et al., 2015). The stable expansion and contraction of the MB are an important attribute of its associated biophysical effects; however, the effects of cavitation-enhanced gas exchange have been less investigated. During MB compression, there is an efflux of core gas into the local environment, followed by a net influx of gas upon expansion (Crum, 1984;Lentacker et al., 2014;Yusof et al., 2016). This is particularly important in regard to the use of nitric oxide microbubbles (NOMBs), since it shows that the MB has the capacity to deliver a locally concentrated therapeutic NO payload.
There have been few studies to date, which investigated the use of microbubbles for the delivery of NO. Tong et al. (2013) and Wang et al. (2013) generated nitric oxide microbubbles (NOMBs) with a mean diameter of 3.85 lm by continuous sonication of a lipid suspension at 100 W. The organophosphorus compound 1,2bis(diphenylphosphino)ethane (DPPE), PEG 2000 and phosphatidylcholine were used to encapsulate NO, which was administered at a constant stream of 4 ml min À1 for 5 min under anoxic conditions. In contrast, Grishenkov et al. (2015) and Cavalieri et al. (2008), used biocompatible polyvinyl alcohol (PVA) and high shear stirring to fabricate air-filled MBs, which were then freeze-dried with liquid nitrogen. The subsequent 4 lm diameter PVA capsules were enclosed in a pressure chamber purged with nitrogen, before loading with NO and re-suspend in water. All four studies focused on the intravascular application of NOMBs; thus, their research is linked by some common themes.
The work by Cavalieri et al. (2008) was the first in vitro study of its kind, utilizing NOMBs for localized delivery of NO for the prevention of clot formation. Similarly, the NOMBs developed by Grishenkov et al. were implemented in a theranostic capacity for myocardial ischaemia, showing that they are a highly effective tool for the localized and targeted delivery of NO. Unlike the work by Tong et al. (2013), this study sought to use NOMBs as a prophylaxis for patients at risk of thromboembolism. In a rodent model of myocardial infarction, 60 s of continuous US (1 MHz, 1 W cm À2 ) applied to NOMBs in conjunction with mesenchymal stem cells was successfully used to promote angiogenesis (Tong et al., 2013). The applications of this as a successful therapy are promising, with the scope to prevent heart failure by restoring adequate blood flow to damaged cardiac tissue (Cochain et al., 2013). The use of NOMBs without ultrasound for the resolution of deep vein thrombosis was investigated by Wang et al. (2013), who successfully showed a reduction in thrombus size by 40% and mitigated development of chronic inflammation.
The research conducted by Grishenkov et al. (2015) was the only study to assess the final NOMB gas content and dissolution rate in solution, with and without the application of US (1-15 MHz, < 100 kPa). High-performance liquid chromatography (HPLC) was used to measure nitrite and nitrate in both aerated and degassed saline, which are oxidized products of NO. Passive release of NO from degassed saline occurred in 17 min, and this is increased to 55 min in aerated saline; the application of US reduced the exponential time constant to 10 min and 4 min respectively (Grishenkov et al., 2015). The assessment of nitrate and nitrite in solution is concordant with expected parameters of diffusion; NO enters the degassed solution at a faster rate than an aerated solution.
Multifunctional agents for the delivery of nitric oxide to biofilms: present and future perspectives To the best of the authors' knowledge, only one previous study has investigated the use of a NO-releasing particle for the treatment of biofilms. The research carried out by Hetrick et al. (2009) investigated the use of NO-releasing silica nanoparticles; however, it focused only on the bactericidal efficacy of the NO and not on biofilm dispersal. Consequently, there are a number of unexplored and novel aspects in this area, pertaining specifically to the application of NOMBs for the treatment of bacterial biofilms. There are currently no acoustically stimulated NOMBs used for the treatment of biofilms in vitro or in vivo; consequently, there is currently no evidence on the efficacy of NOMBs or nanoparticle-induced dispersal of biofilms. Furthermore, though its effects have been observed, the specific interaction between MBs and the biofilm has not been fully elucidated. Previous research has shown that lipid transfer occurs between MBs and biological membranes, which undoubtedly impacts upon cellular integrity, permeability and signalling (Carugo et al., 2017). This effect could be successfully employed to exert a priming effect upon biofilms, to stimulate disruption of the biofilm prior to the administration of US.
Bacterial infection and subsequent development of biofilms in open wounds pose a significant risk to human health. Due to the increased tolerance of biofilms to traditional antimicrobial therapies and non-specific drug delivery, interdisciplinary techniques are being explored as novel treatment methods. Ultrasound-responsive drug delivery agents provide a dynamic means of delivering therapeutic compounds, with high temporal and spatial specificity. Ongoing research has shown that ultrasoundresponsive agents can facilitate drug delivery, utilizing both bioactive components and mechanical stimulation to eliminate biofilms. There is no clinically viable translation of this treatment modality for chronic wounds at Fig. 4. Gas microbubbles undergoing stable cavitation in response to ultrasound have been shown to enhance membrane permeability; this can increase the local intracellular concentration of a target drug. By encapsulating biologically active nitric oxide as the gaseous core of a shelled microbubble, the combined mechanical action of the oscillating microbubble and biological effect of nitric oxide may induce targeted dispersal and elimination of biofilms from a surface. The co-administration of antibiotic and microbubbles may provide a novel means of combating biofilm-associated antibiotic tolerance. present, perhaps due to the complexity of ensuring consistent and efficacious implementation with minimal training or expertise. Essential parameters such as the consistent production and administration of MBs, controlled transmission of US and handling of biohazardous residuum would require a robust system with clearly defined instructions for use in lay terms. Moreover, due to the lack of congruity in size, shape and depth of wounds, it is likely a successful clinical translation would be primarily targeted at early stages of DFU development for patients presenting with small lesions. This would then have the potential to either prevent or limit biofilm development, in addition to dispersing and treating any adherent cells already present in the wound bed. The transmission of US via a fluid stream to biotic and abiotic surfaces for the purpose of biological decontamination has recently demonstrated that clinical translation and utility in this field are achievable (Birkin et al., 2015(Birkin et al., , 2016. In order to achieve fundamental change in healthcare practices such as the treatment of chronic wounds, we believe this review exemplifies the need for collaborative and interdisciplinary research to potentiate existing therapies and develop novel treatment modalities.