Microbubble-enhanced ultrasound-mediated gene transfer – Towards the development of targeted gene therapy for cancer

Abstract

Ultrasound-mediated gene transfer is emerging as a possible alternative to viral gene transfer, and pre-clinical data suggest that it may play a significant role in gene therapy-based approaches to the treatment of disease. As an extracorporeal stimulus, ultrasound can non-invasively and transiently compromise cell membrane permeability (sonoporation), thereby offering the promise of delivering either genes or oligonucleotide-based therapeutics to cells and tissues in a site-specific manner. The membrane-permeabilising effects of ultrasound can be greatly enhanced using microbubble preparations, many of which have, in the past, found application as ultrasound contrast agents. Because these ultrasound-responsive agents are highly amenable to surface modification it has been suggested that they may be exploited as ultrasound-responsive nucleic acid delivery vehicles. In this article we seek to explore the potential role ultrasound, in combination with microbubble-based agents, may play in providing site-specific gene therapy-based approaches for the treatment of cancer.

• Microbubble
• gene
• cancer
• sonoporation
• ultrasound

Introduction – Gene transfer, ultrasound and cancer

The concept of using genes to facilitate therapeutic effects is not a novel one and offers very significant potential in the treatment of a wide range of disorders including cancer. Originally, the concept involved the insertion of a functional intact gene to replace a non-functional endogenous gene or the use of an exogenous intact gene to provide a therapeutic output. With increasing technological developments and our expanding knowledge relating to control of gene expression, however, the concept currently includes the suggested use of therapeutic oligonucleotides, siRNA, shRNA and RNAi species to either up or down regulate gene expression for therapeutic purposes. Irrespective of the approach, it is a pre-requisite that the therapeutic nucleic acid is taken up by cells at the relevant target site in order to elicit the required therapeutic effect. Although gene therapy offers exciting therapeutic potential, the level of translation of pre-clinical approaches to direct clinical application has been somewhat disappointing. The latter has resulted, at least in part, from the inability to deliver the therapeutic nucleic acid to the required tissue and/or into specific target cells within the tissue to be treated. Over the past decade the method of choice for achieving efficient gene transfer in vivo for therapeutic purposes has involved the use of viral vectors, primarily because these have the ability to transfer nucleic acid into cells in an extremely efficient manner [1]. However, identified challenges associated with the use of such vector systems have included immunotoxicity [1] and insertional mutagenesis where insertion of viral sequences/promoters has resulted in the activation of proto-oncogenes [2], [3]. In addition, viruses are somewhat non-specific with respect to target tissues, and this lack of specificity complicates therapeutic approaches that might benefit from a degree of site-directed targeting. Recognition of these deficits has prompted the development of a variety of non-viral approaches to facilitate gene delivery, and one of the alternative delivery systems that have gained most momentum is electroporation, or the use of electric fields to induce transient compromise to cell membrane permeability [4]. Indeed, this approach has been translated to the clinic, and a number of trials using electroporation for gene therapy-based treatment of cancer are currently ongoing [5–7]. Because of the requirement for precise electrode placement around the target tissue for delivery of the requisite electric field, the procedure is considered minimally invasive and as such is more amenable to topical application in the treatment of more accessible forms of cancer such as melanoma. In addition, because of variations in tumour architecture, composition, tissue electrical impedance, resistance and conductivity, control of electric field delivery is difficult to predict and great care must be taken to incorporate all of the target tissues in order to elicit a homogeneous desired response (in this case, gene transfer). Although such aspects are being addressed with the emergence and application of novel analytical technologies, the above aspects continue to challenge generalised clinical application of the approach, and access to deep-seated tissue targets also remains a significant challenge [8].

As a result of its ability to penetrate through tissues and because it can be focused to a precise point in space, ultrasound has attracted significant attention from those interested in achieving site-specific therapeutic effects. The latter is of particular interest in the treatment of cancer because the primary disease exists as a tissue focus and such foci are amenable to targeted localised treatment. Currently, the use of ultrasound to facilitate non-invasive and site-directed therapeutic effects in the clinic is relatively well established in cancer therapeutics, with the emergence of approaches exploiting high intensity focused ultrasound (HIFU) as a means of inducing hyperthermic tissue ablation [9–11]. Ultrasound has proven much more effective than alternative methods for inducing hyperthermia (e.g. radio frequency irradiation) because it may be guided to the target by coupling with magnetic resonance imaging (MRI) technology and can be delivered from an extracorporeal source to a precise point in space within the body. Indeed, the development of HIFU as a hyperthermic ablative treatment for cancer originally necessitated developments in device engineering that enabled very precise deposition of energy at a single point in three dimensions, and this technology is now considered to be widely available. While developments in this area were ongoing, others had been examining ultrasound-induced phenomena that were elicited at ultrasound power densities somewhat lower than those employed for HIFU-based ablative treatments. Much of this work centred on observations that ultrasound could be employed to facilitate transient compromise to cell membrane permeability (sonoporation), and it became increasingly recognised that the phenomenon could be exploited in the context of macromolecular drug and gene intracellular delivery. With the emergence of sophisticated ultrasound targeting devices (i.e. integrated MRI and focused ultrasound delivery devices) and an increasing awareness of the clinical potential offered by non-invasively modifying or controlling uptake of macromolecular substances by cells and tissues using an extracorporeal stimulus, it became increasingly recognised that such a combination could be exploited in site-directed therapeutics. Here we will examine some of the approaches that have been taken in attempting to exploit ultrasound as a means of facilitating gene or nucleic acid delivery/cellular uptake and explore some of the pre-clinical evidence that indicates a significant role for this technology in clinical gene therapy-based approaches to the treatment of cancer.

Ultrasound and microbubbles

If ultrasound is to be employed to facilitate gene/nucleic acid delivery for effective therapeutic purposes in the treatment of cancer, any ideal therapeutic approach will necessitate; (1) non-invasive and precise delivery of ultrasound to a pre-determined target site, (2) targeted delivery of the therapeutic nucleic acid to the lesion and (3) delivery of the therapeutic nucleic acid into cells at the relevant target site. Studies on the development of HIFU-based treatments have addressed many of the issues associated with non-invasive and precise delivery of ultrasound to target sites, and even many of the challenges associated with transcranial targeted delivery of ultrasound have been, or are currently being, addressed [9], [10], [12]. However, delivery of nucleic acid to target tissues in a site-specific manner and promoting entry of the delivered nucleic acid into target cells in those tissues represent two of the most significant challenges in enabling effective ultrasound-mediated gene therapy. So how might microbubbles provide a solution to the above challenges?

Originally, microbubble preparations were developed and exploited as contrast agents in ultrasound imaging, essentially because of their ability to provide enhanced echogenic effects. Such effects could be exploited in areas such as non-invasively examining cardiovascular blood flow properties/dynamics [13] and in assessing vascular integrity [14]. Indeed, more recently, microbubble preparations have been employed to enhance ultrasound-based diagnostic imaging of tumour vasculature with a view towards monitoring tumour angiogenesis [14–16]. Over the years a number of commercially available microbubble reagents have been launched onto the market as diagnostic aids, and some of these are shown in Table 1. These stabilised microbubbles contain different core gases and the microbubble shell may consist of denatured albumin, phospholipids, polymers, or indeed, combinations of these. Although microbubbles have historically been designed and used for ultrasound-based diagnostic imaging applications, it was quickly realised that these reagents could also find application in the design of drug/gene delivery systems. Such suggestions were primarily based on observations of microbubbles in ultrasonic fields. It has been demonstrated that microbubbles oscillate in an ultrasonic field and the intensity of those oscillations is influenced primarily by the ultrasound intensity (power density) and resulting acoustic pressure at any given ultrasound frequency. At relatively lower ultrasound intensities, microbubble oscillation may be stable and this is referred to as stable cavitation. When the ultrasound intensity at any given frequency is increased above a characteristic threshold the microbubble becomes unstable, eventually undergoing catastrophic collapse as a result of inertial inward movement of the surrounding medium, and this is referred to as inertial cavitation [17], [18]. In the former, it has been suggested that the physical movement resulting from stable cavitation can result in micro-environmental disturbances such as microstreaming and/or shear stress and this can influence cell membrane permeability without actually physically rupturing the cell membrane (19). In the latter, the catastrophic or inertial collapse of the microbubble results in more significant effects such as micro-jetting and if these events occur in the micro-environment of cell membranes, the structural integrity of those membranes becomes compromised so that discrete pores are formed. This results in significant compromise to membrane permeability. Although it had previously been demonstrated that ultrasound alone could be employed to facilitate transient ‘poration’ of cell membranes [20], the addition of exogenous microbubble-based agents was shown to reduce the acoustic pressure required to bring about this effect [21]. It has further been demonstrated that this ultrasound-induced compromise to cell membrane permeability or sonoporation, can either result in cell death or be transient in nature and it is the latter that is most relevant to our discussions here. It should be noted that although the cavitational effects leading to sonoporation events are mechanical in nature, subtle effects on cell membranes resulting from mild heating during exposure to ultrasound cannot be ruled out. It should further be noted that induction of inertial cavitation of microbubbles is also a frequency-dependent phenomenon where decreasing the ultrasound frequency increases microbubble wall velocity and subsequent microbubble collapse [22]. The recognition that an external stimulus could be used to transiently influence cell membrane permeability to either low molecular weight drugs (e.g. cancer chemotherapeutics) or macromolecules (e.g. nucleic acids) suggested a potential therapeutic role for this phenomenon.

Table 1.  A selection of microbubble-based agents and their reported application.

Microbubble	Diameter (µm)	Shell	Gas	Application
Acusphere	2.2	Poly(L-lactide-co-glycolide) (negative)	Perfluorocarbon	US contrast agent, drug delivery
Albunex (Molecular Biosystems)	4.5	Albumin	Air	US contrast agent, gene delivery
AS-0100 (Artison)	2.4	Stabilised lipid	Perfluorocarbon	US induced cell death, free radical formation
BG6356A (Bracco)	1.8	Triglyceride	Nitrogen	US induced cell death, free radical formation
BG-6437 (Bracco Research)	3	Stabilised lipid	Perfluorobutane/nitrogen	Imaging
BG-6438 (Bracco Research)	3	Stabilised lipid/streptavidin	Perfluorobutane/nitrogen	Targeted imaging
biSphere (Point Biomedical)	3	Polylactide/albumin (slightly negative)	Nitrogen	Drug delivery, lymphatic imaging
BR14 (Bracco)	3	Stabilised lipid (negative)	Perfluorobutane	US contrast agent, thermal ablation
BY963 (Nycomed)		Lipid	Air
Cardiosphere (Point Biomedical)		Polylactid/albumin	Nitrogen	Infusion
Definity (Bristol-Myers Squibb Medical Imaging)	1.5	Lipid/surfactant (negative)	Perfluoropentane	US contrast agent, BBB opening
Echogen (Sonus)		Surfactant	Dodecafluoropentane
Echovist-200® (Schering)	8–12	No shell (galactose matrix)	Air	US contrast agent
Imagent (IMCOR Pharmaceuticals)	6	Lipid: DMPC (neutral)	Perfluorohexane/nitrogen	US contrast agent
Imagify (Acusphere)		Poly(L-lactide-co-glycolide) (negative)	Decafluorobutane	Myocardial perfusion
Levovist (Schering)	2–4	Galactose with palmitic acid (negative)	Air	US contrast agent, gene delivery in vitro
MP1950 (Mallinckrodt)		Phospholipid	Perfluorobutane
Myomap (Quadrant)		Albumin	Air	Myocardial perfusion
Optison (Mallinckrodt/Amersham)	3–5	Albumin (slightly negative)	Octafluoropropane	Gene/drug delivery
PESDA	4.7	Dextrose-albumin	Decafluorobutane	US contrast agent, gene transfer in vitro
Quantison (Andaris)		Albumin	Air	US contrast agent
SDM201 (Sonidel)	1.7	Stabilised lipid	Perfluorobutane	Gene transfer
SDM202 (Sonidel)	1.7	Stabilised lipid (cationic)	Perfluorobutane	Gene transfer
SDM302 (Sonidel)	1.7	Biotinylated stabilised lipid (cationic)	Perfluorobutane	Gene transfer
Sonavist or Sonovist (Schering)	–	Cyanoacrylate (polymer shell)	Sulphur hexafluoride	US contrast agent
Sonazoid (GE Healthcare)	2.6	Stabilised lipid	Perfluorocarbon	US contrast agent, anticancer treatment
SonoGen (Sonus Pharmaceuticals)		Surfactant (negative)	Dodecafluoropentane	US contrast agent
SonoVue (Bracco)	2.5	Phospholipid (negative)	Sulphur hexafluoride	HIFU ablation, gene/drug delivery
Targestar-B (Targeson)	2.5	Biotinylated stabilised lipid	Perfluorocarbon	US contrast agent
Targestar-P (Targeson)	2.5	PEGylated lipid	Perfluorocarbon	US contrast agent

US, ultrasound; HIFU, high intensity focused ultrasound.

Microbubbles and gene delivery

The ability of microbubbles to respond to an ultrasonic field in the above manner (microbubble oscillation and/or collapse with collateral transient compromise to cell membrane permeability) suggested that these reagents could be employed to deliver therapeutic agents to a relevant target site and also promote intracellular delivery of the relevant therapeutic agent. Indeed these attributes, together with the ability to non-invasively deliver the stimulus (ultrasound) to a specific target site appear to fulfil all the above-mentioned requisite components for an ideal gene delivery system. In some of the earlier studies associated with the use of microbubbles to serve as nucleic acid delivery vehicles it was demonstrated that microbubbles, prepared by sonication of dextrose albumin in the presence of perfluorocarbon gas and a synthetic antisense oligonucleotide, resulted in a microbubble preparation that had oligonucleotide bound to its shell [23]. It was further demonstrated in this study that the oligonucleotide could be released with ultrasound both in vitro and following intravenous (IV) administration in vivo. In other studies it was demonstrated that plasmid DNA could be bound to a phospholipid-stabilised microbubble preparation, particularly when cationic lipids were incorporated into the shell, and this could serve to deliver plasmid DNA [24]. These initial data provided very clear evidence that microbubbles could also serve as nucleic acid carriers. Since it was established that ultrasound could be exploited to effect transfer of plasmid DNA encoding chloramphenicol acyltransferase into human tumour cell lines [24], the number of reports describing ultrasound-mediated, microbubble-enhanced gene transfer into a wide range of tumour cell lines has been extensive [25]. In many of those studies, the underlying principle that microbubbles enhance the ability of ultrasound to promote uptake of plasmid DNA by target cells in vitro is clearly demonstrated, and ultrasound power densities required to accomplish this are relatively low. One other significant conclusion that may be deduced from those in vitro studies is that the efficiency of gene transfer is inversely proportional to cell viability and this is similar to other physical methods employed for gene transfer including electroporation. It has been suggested that reduced cell viability following exposure of cells to ultrasound and microbubbles results from the catastrophic effects that may derive from ultrasound-induced inertial cavitation, and this is supported by demonstrating reduced cell viability with increasing concentration of exogenously added microbubbles in vitro [26]. It would appear, therefore, that achieving efficient gene transfer and expression using ultrasound in combination with microbubbles in vitro, necessitates achieving a careful balance between microbubble concentration and ultrasound power density if high cell viability and gene expression are to be facilitated [26].

Although in vitro studies have provided valuable insights into the mechanism(s) by which ultrasound and microbubbles facilitate gene transfer, it is generally recognised that extrapolation from in vitro to in vivo is notoriously difficult in this field. This usually results from differences in the transmission characteristics of ultrasound through target systems in vitro and through living tissues in vivo. It can also result from the instability of nucleic acids in vivo and from the variable response of microbubble preparations to ultrasound in vivo [27], [28]. Regardless of these issues, it is clear from the literature that ultrasound can be employed to facilitate gene delivery to a wide variety of tissues in vivo and a number of examples are shown in Table 2 [26], [29–36]. The chosen examples clearly demonstrate that ultrasound exposure can facilitate gene transfer in a wide variety of tissues and as such clearly supports the suggestion that ultrasound might play a role in facilitating gene transfer in diseases such as cancer where lesion targets exhibit ubiquity with respect to tissue source and siting. It should also be noted that in most of the chosen examples the ultrasound frequency employed for gene transfer into those tissues is in the region of 1 MHz and this frequency is primarily chosen to maximise tissue penetration. This ultrasound frequency, which is lower than those frequencies normally employed for diagnostic imaging (2.5–10 MHz) is also chosen to induce microbubble collapse which, in turn, induces sonoporation [22]. As mentioned above, it has been suggested that, in addition to enhancing gene transfer, microbubbles might also provide advantage as a nucleic acid carrier in vivo. This is adequately illustrated by the use of microbubbles with cationic shells that electrostatically bind the therapeutic nucleic acid [26]. In addition to stabilising the therapeutic nucleic acid in vivo, direct attachment to a microbubble preparation would preclude separate administration of nucleic acid and the microbubbles, and this would also have advantages from a clinical perspective [26]. Indeed, the latter point would be particularly relevant if one was considering the use of therapeutic RNA species which tend to be extremely unstable. That microbubbles can be used to enhance ultrasound-mediated gene transfer is no longer in doubt, and success in this area is further evidenced by the emergence of commercially available microbubble preparations specifically designed for such purposes (Table 1). The initial observations and wealth of information accumulated thus far demonstrating that ultrasound, in combination with microbubbles, can be used to effect site-directed delivery and enable trans-membrane transfer of nucleic acid in various organs and tissues in vivo, have led to the suggestion that the approach may have a significant role to play in gene therapy-based approaches for the treatment of cancer.

Table 2.  Ultrasound-mediated gene transfer into various tissues in vivo.

Target tissue/cells	Animal	Gene product	Reference
Skeletal muscle	Mouse	Luciferase	[26]
		rhBMP-9	[29]
Peritoneal macrophages	Mouse	GFP	[30]
Salivary gland	Rat	siRNA - GADPH	[31]
Ocular ciliary muscle	Rat	Luciferase	[32]
Myocardium	Mouse	GFP-Luciferase fusion	[33]
Skin	Mouse	Minicircle-VEGF	[34]
Carotid artery	Rat	Luciferase/p53	[35]
Embryonic limb bud mesenchyme/ectoderm	Chick	GFP/β-gal	[36]

rhBMP-9, recombinant human bone morphogenetic protein-9; GFP, green fluorescent protein, GADPH, glyceraldehyde-3-phosphate dehydrogenase; VEGF, vascular endothelial growth factor; β-gal, β-galactosidase.

Ultrasound-mediated targeting of gene expression and cancer

In considering gene therapy-based approaches for the treatment of cancer, an increasing number of potential therapeutic targets are being presented on an almost daily basis, and this primarily results from our ever-expanding knowledge of the molecular mechanisms that drive processes such as carcinogenesis, tissue invasion, tumour angiogenesis, metastasis and associated processes. Awareness of the molecular genetic mechanisms by which cancer progresses has led to the evolution of a number of gene therapy-based strategies that, in pre-clinical models, have proven successful, and these have included the replacement of defective tumour suppressor genes [37], down regulation of oncogenes [38], introduction of exogenous genes encoding enzymes capable of prodrug activation [39], introduction of endogenous genes encoding immunological mediators [40] and introduction of genes capable of preventing tumour angiogenesis [41], to mention but a few. However, as mentioned previously, the negative effects associated with an over-dependence on viral vectors to facilitate gene delivery have hindered translation of many of those approaches to the clinic. The use of ultrasound, however, together with the appropriate microbubble preparation could provide a solution to the challenges presented by viral vectors. Indeed, it has been demonstrated that ultrasound together with microbubble preparations has been used to facilitate gene transfer into a variety of cancerous tissues in preclinical models in a site-directed manner, and a selected number of relatively recent examples are shown in Table 3 [42–49]. In the selected examples, the ultrasound frequency chosen for the purposes of stimulating gene transfer ranged from 1–1.3 MHz with the exception of one study that employed 3 MHz [48]. Again, the lower frequencies are chosen both for enhanced tissue penetration and enhanced induction of inertial cavitation. A number of aspects are noteworthy about this compilation of reports. (1) All of these studies demonstrated that it is possible to employ ultrasound, together with microbubbles, to facilitate expression of a plasmid-encoded transgene in a wide range of tumour tissues in vivo using either preclinical syngeneic mouse tumour models or preclinical models based on the use of human tumour xenografts in mice. (2) In four of those studies the microbubble/plasmid preparation was administered via intratumoural injection prior to the application of ultrasound [42–45], while four of those studies reported systemic administration of the microbubble/plasmid dose prior to treatment with ultrasound [46–49]. Essentially, the former studies demonstrated ultrasound-mediated uptake of plasmid DNA by tumour cells at the ultrasound-treated site, whereas the latter studies demonstrated ultrasound-mediated, site-specific targeting of plasmid DNA to the tumour site and concomitant uptake of that DNA by cells in the targeted tumour. The latter also demonstrated that microbubble preparations, administered systemically and travelling through the tumour vasculature, could respond to the ultrasound stimulus ensuring both nucleic acid deposition and uptake at the tumour site. From applied therapeutic and regulatory perspectives this latter aspect is extremely important. In an ideal therapeutic scenario, it would be advantageous to exploit the microbubble preparation both as a carrier and as a means of facilitating targeted uptake of the nucleic acid payload rather than administering the nucleic acid and microbubble preparation as separate entities. Indeed, in two of these latter studies the microbubble preparations were specifically designed to serve as nucleic acid carriers [46], [49]. Interestingly, it was found that when microbubbles were employed in this manner, expression of the transgene was detected both in the vascular endothelium and in the tumour cells, suggesting that the nucleic acid was capable of escaping from the tumour microvasculature [46]. Given the atypical and leaky nature of tumour vasculature, this is perhaps not altogether surprising, and taken in combination with the observation that ultrasound can be used to modify vascular permeability [50], it is likely that at least some proportion of the nucleic acid payload would leak into the extravascular tissues in a solid tumour. In another of those studies quoted in Table 3 where the nucleic acid was bound to the microbubble as a lipoplex, administered systemically and targeted to tumours using ultrasound, it was suggested that such an ‘on demand’ system could preclude exploiting the ‘enhanced permeability and retention’ effect exhibited by solid tumours and exploited in other drug or gene delivery modalities [49]. This is an interesting perspective and will be discussed later in the context of novel ultrasound responsive preparations. (3) The final, and perhaps most relevant aspect that can be noted from the selected examples shown in Table 3, is that where a therapeutic plasmid/nucleic acid was delivered to tumour models in vivo, a therapeutic effect was observed whether or not the approach involved intralesional or systemic administration of the microbubble-nucleic acid combination. Since it has been suggested that interferon β (INF-β) can induce apoptosis by enhancing expression of the tumour suppressor p53 in tumour cells, Yamaguchi et al. [42] injected microbubbles and plasmid encoding INF-β directly into human melanoma xenografts established in nude mice. Following treatment with ultrasound, tumour growth, production of INF-β and the onset of apoptosis were monitored. In addition to demonstrating expression of the INF-β in tumour tissues following treatment, the authors also demonstrated a very significant reduction in tumour growth in the group of animals treated with microbubbles, plasmid-encoding INF-β and ultrasound. In addition they also demonstrated induction of caspase 3, a marker of the onset of apoptosis [42]. Tsai et al. [43] administered microbubbles plus plasmid DNA encoding the angiogenic suppressor endostatin directly to a preclinical hepatocellular carcinoma model followed by treatment with ultrasound, and found that with repeat treatment a 74% inhibition in tumour growth was observed. In an approach that used interleukin-12 (IL-12) to promote immuno-modulatory anti-tumour effects and to exploit its ability to promote anti-angiogenesis, Suzuki et al. [45] treated a murine ovarian carcinoma with ultrasound following intratumoural injection of bubble liposomes together with plasmid DNA encoding IL-12. In this case also, tumour growth was significantly reduced, with complete regression in 80% of the animals treated [45]. Using an approach that involved the use of herpes simplex virus-1 thymidine kinase (HSV-tk) activation of the pro-drug ganciclovir, Carson et al. [46] systemically administered microbubbles with attached plasmid DNA incorporating the HSV-tk gene to treat a mouse squamous cell carcinoma. Three days after treatment of tumours with ultrasound to facilitate targeted delivery and uptake of the therapeutic transgene by the tumour, animals were treated with the pro-drug ganciclovir and a small, but significant increase in tumour doubling time was noted. Although the author suggested limitations associated with their study, it is very significant because in addition to demonstrating a therapeutic effect in a pre-clinical model, it demonstrated the use of ultrasound to effect targeted delivery and uptake of the therapeutic gene following systemic administration of an ultrasound-responsive nucleic acid carrier (i.e. the microbubble). Indeed, in terms of its use for this particular form of gene therapy, with optimisation the approach could provide significant advantage because the previously employed clinical approach necessitated surgical intervention to ensure sufficient distribution of the transgene throughout the target tissues [51]. In an approach that involved down regulation or gene silencing of the apoptotic inhibitor survivin, using short hairpin RNA (shRNA) interference, Chen et al. [48] employed a plasmid encoding shRNA targeting the human survivin gene to induce apoptosis in human cervical cancer xenografts in nude mice. In this approach the authors used a combination of microbubbles together with a polyethylenimine/plasmid DNA (PEI/DNA) complex and the mixture (of separated entities) was introduced to the animals via intravenous injection (tail vein). Following treatment with ultrasound, the authors were able to demonstrate tissue-specific targeting of gene expression (with reporter gene signals some 1000-fold higher in tumour tissues than in liver), down regulation of survivin in the targeted tumours and consequential induction of apoptosis in those treated tissues [48]. Under normal circumstances, IV administered PEI/DNA complexes would result in enhanced uptake by the liver/kidney/lung (depending on design) and more or less random distribution in other tissues [52], [53]. Although the ultrasound-mediated targeting system described by Chen et al. [48] did not involve preparing microbubbles that served as a direct carrier for the PEI/DNA complexes, the approach of systemic co-administration followed by treatment with ultrasound did demonstrate ultrasound-promoted tumour-specific expression of the therapeutic nucleic acid and induction of apoptosis in the targeted tumour tissues. However, in this report no data relating to the effect of using such a combination on tumour growth were presented. In a more recent report Sirsi et al. [49] described the synthesis of microbubble-PEI hybrids where the PEI was chemically linked to the microbubble. This was then mixed with DNA prior to IV administration to mice harbouring human Ewing's sarcoma xenografts so that essentially the microbubble served as a carrier for the PEI/DNA complexes. Following treatment of the tumour site with ultrasound, site-specific expression of a luciferase reporter gene was observed, suggesting that the approach could be employed for targeting therapeutic genes to tumours. Although this approach is relatively sophisticated in terms of facilitating carriage of nucleic acids, others including our own group have described direct binding of nucleic acid to cationic microbubbles and the highlighted the advantages associated with their use for ultrasound-mediated gene delivery [26], [54], [55]. Cationic microbubbles have been particularly indicated for gene therapeutics such as oligonucleotides (ODNs) or unstable therapeutic RNA species, and it has been shown that cationic microbubbles could be used to deliver anti-androgen receptor ODNs to prostate tumours following systemic delivery and ultrasound treatment [55]. Although no down regulation of androgen receptor or a therapeutic effect was demonstrated in the treated prostate tumours in this study, ultrasound-mediated targeting of the ODN was clearly demonstrated using the cationic microbubbles. Regardless of the mechanism by which the therapeutic nucleic acid is delivered to the target or indeed the underlying mechanism by which the therapeutic nucleic acid exerts its effect, it is clear from the above that ultrasound is emerging as a significant player in terms of facilitating targeted nucleic acid delivery in gene-based therapies for the treatment of cancer.

Table 3.  Pre-clinical studies relating to ultrasound-mediated delivery of transgenes to in vivo.

Target tumour/cells	Host	Gene product	Reference
Melanoma	Mouse	IFN-β	[42]
Hepatocellular carcinoma	Mouse	EGFP/endostatin	[43]
Fibrosarcoma	Mouse	EGFP/luciferase	[44]
Ovarian carcinoma	Mouse	IL-12/luciferase	[45]
Squamous cell carcinoma	Mouse	HSV-tk/luciferase	[46]
Sarcoma	Mouse	Luciferase	[47]
HeLa xenografts	Mouse	shRNA – survivin/luciferase	[48]
Ewing's sarcoma	Mouse	Luciferase	[49]

IFN-β, interferon β; EGFP, enhanced green fluorescent protein; IL-12, interleukin 12; HSV-tk, Herpes simplex virus thymidine kinase; shRNA-survivin, short hairpin RNA interference targeting survivin.

Future challenges

In the previous section we have sought to highlight the potential offered by the use of ultrasound to facilitate site-directed gene-based therapies for cancer treatment. However, we also feel that those examples may serve to highlight some of the remaining challenges facing the translation of this technology to full clinical application. In addition to using small animal models, many of the above examples employed relatively unsophisticated ultrasound generating systems to provide the stimulus for nucleic acid delivery to the tumour. Translation to the clinic would involve the use of more sophisticated ultrasound generating devices incorporating a focusing capability and aiming systems to treat and target deep-seated tumours with larger tumour masses. Although these aspects represent challenges to clinical translation of ultrasound-mediated gene transfer, they are somewhat ameliorated by progress achieved in engineering device configurations for use in the area of HIFU treatment as mentioned above.

The heterogeneity of tumour architecture represents one of the biggest challenges facing clinical translation of any novel cancer treatment modality and indeed it is perhaps more so for a technology that involves the use of ultrasound and microbubbles to facilitate gene transfer. Many of the therapeutic molecules used in gene-based medicine (e.g. plasmid DNA, oligodeoxynucleotides, siRNA, shRNA, etc.) are much larger than conventional cancer chemotherapeutic drugs, and even delivery of the latter is often complicated by the heterogeneity of tumour architecture. It is generally recognised that the rapid initial growth of tumour tissues tends to out-pace the development of the tumour vasculature and this gives rise to a number of vascular structural and flow dynamic deficits [56]. Tumour blood vessels are atypical with an abnormal basement membrane if that basement membrane is present and abnormal vascular branching patterns with loops, dead ends and arteriolar-venous shunts. This can result in a reduced pressure differential between arterioles and venules and blood flow resistance which can, in turn, lead to a reduced tumour blood supply. In addition the structural integrity of the endothelium is compromised, leading to a leaky vasculature and compromised lymphatic drainage from the tumour mass. It has been suggested that structurally compromised tumour endothelium together with poor tumour drainage can lead to enhanced permeability and retention (EPR), and this can lead to accumulation of macromolecules or small nanoparticulates in tumour tissues. On the surface it would appear that this would provide advantage in terms of delivering nucleic acids to tumour tissues. However, it should also be noted that delivery of a microbubble-based vehicle for nucleic acid would be hindered by aspects such as microbubble size and the high interstitial fluid pressure (IFP) that can build up in tumour masses as a result of poor lymphatic drainage. The IFP represents a fluid dynamic ‘back pressure’ that reduces vascular flow through the tumour mass, and if this flow is reduced, then the number of microbubbles travelling through the vasculature will be reduced. In addition, when the microbubble is ‘activated’ by ultrasound at the tumour site, one would expect microbubble disruption and subsequent extravasation of the macromolecular nucleic acid. The leaky vasculature would obviously be expected to provide benefit here since it would lead to trapping of the nucleic acid/microbubble debris at the target site by the EPR effect. However, even this perceived benefit is countered by the observation that increased permeability is not homogenous throughout the tumour mass, and in fact this is why necrotic foci develop within many tumours. One obvious way of addressing these challenges, particularly the latter aspect of heterogeneous permeability, would be to use a gene therapy approach that would provide a significant ‘bystander’ effect. Using a therapeutic gene encoding HSV-tk and subsequent administration of ganciclovir would be a prime example of such an approach where the transgene can activate the pro-drug and the activated pro-drug can exert a ‘bystander’ therapeutic effect by subsequent diffusion away from the cell expressing the transgene [46]. In addition, it has been known for some time that exposure of tissues to ultrasound in combination with microbubbles can enhance vascular permeability, thereby promoting extravastation at the target site. Bekeredjian et al. [57] exploited this phenomenon to demonstrate a five-fold increase in deposition of the vascular impermeable dye Evan's blue in rat hepatomas. Characterising this phenomenon further, Bohmer et al. [58] demonstrated a similar effect using a mouse adenocarcinoma model, but suggested that extravasation of the dye was restricted to the more vascularised areas of the tumour with little or no dye in the necrotic areas of the tumour. Along the same lines our own group has demonstrated that ultrasound, at acoustic intensities suitable for gene transfer in vivo and in the absence of microbubbles, is capable of facilitating diffusion of low molecular weight substances through relatively poorly vascularised tumour tissues [59]. This is a particularly interesting phenomenon that could address issues relating to the lack of diffusion of agents through highly inaccessible anoxic tissues as demonstrated by Bohmer et al. [58]. The issue relating to the possible excessive size of microbubbles and the limited perfusion of either sonoporative effects or deposited nucleic acid beyond the immediate vicinity of the tumour vasculature is also being addressed by the development of nano-sized bubbles. Reducing the bubble size would permit exploitation of the EPR effect and facilitate extravasation into less accessible areas. In a very recent study Horie et al. [60] demonstrated that nanobubbles with sizes in the region of 200 nm could be used together with ultrasound to enable gene transfer of a tumour necrosis factor-α (TNF-α) encoding gene into a murine breast carcinoma in vivo. Although the preparation was administered by intratumoural injection, transgene expression and consequential retardation of tumour growth was demonstrated. The authors suggested that with this approach, the EPR could be exploited, at least in part, if the preparation was administered systemically. They also suggested benefit in terms of possible enhanced dispersion throughout tumour tissues. In an innovative approach, Rapoport et al. [61] used ultrasound-responsive, paclitaxil-loaded nanoemulsions to demonstrate accumulation of the nanodroplets at the tumour site that delivered a therapeutic effect following exposure to ultrasound. The nanodroplets comprised a liquid perfluorocarbon core that could be converted to the gaseous state on exposure to ultrasound. Although this study was not concerned with gene transfer, it did demonstrate the concept of exploiting the EPR for tumour uptake of the ultrasound-responsive particles from systemic circulation. It also demonstrated ultrasound-mediated conversion of nanodroplets to microbubbles in a site-directed manner. Exploiting this conversion to subsequently or simultaneously use the resulting microbubble as a sonoporative stimulus for the enhanced uptake of a therapeutic nucleic acid offers exciting possibilities in terms of addressing many of the issues discussed above.

In conclusion, although a number of important issues remain to be resolved in exploiting ultrasound for site-directed gene therapy-based clinical treatments for cancer, we believe that the literature provides very significant ‘proof of concept’, indicating therapeutic benefit in a wide range of pre-clinical tumour models. Recent advances in the use of ultrasound-responsive microbubbles as direct nucleic acid carriers and the development of ultrasound-responsive nanobubble preparations offer significant advantage in terms of addressing specific issues relating to delivery of nucleic acids to more inaccessible tumour tissues. With device requirements (in terms of ultrasound delivery) currently in place, we believe that ultrasound targeted gene-based therapy could provide significant patient benefit, particularly in the treatment of more recalcitrant or less accessible lesions.

Declaration of interest: Nikolitsa Nomikou is Research and Development Manager at Sonidel Ltd. During the preparation of this article Nikolitsa Nomikou was partially funded by a grant under the EuroStars Programme administered by Enterprise Ireland. The authors alone are responsible for the content and writing of the paper.