Explorations of high-intensity therapeutic ultrasound and microbubble-mediated gene delivery in mouse liver

Abstract

Ultrasound (US) combined with microbubbles (MBs) is a promising technology for non-viral gene delivery. Significant enhancements of gene expression have been obtained in our previous studies. To optimize and prepare for application to larger animal models, the luciferase reporter gene transfer efficacy of lipid-based Definity MBs of various concentrations, pressure amplitudes and a novel unfocused high-intensity therapeutic US (HITU) system were explored. Luciferase expression exhibited a dependence on MB dose over the range of 0–25 vol%, and a strong dependence on acoustic peak negative pressure at over the range of 0–3.2 MPa. Gene expression reached an apparent plateau at MB concentration ≥2.5 vol% or at negative pressures >1.8 MPa. Maximum gene expression in treated animals was 700-fold greater than in negative controls. Pulse train US exposure protocols produced an upward trend of gene expression with increasing quiescent time. The hyperbolic correlation of gene expression and transaminase levels suggested that an optimum gene delivery effect can be achieved by maximizing acoustic cavitation-induced enhancement of DNA uptake and minimizing unproductive tissue damage. This study validated the new HITU system equipped with an unfocused transducer with a larger footprint capable of scanning large tissue areas to effectively enhance gene transfer efficiencies.

Introduction

Ultrasound (US)-facilitated gene delivery of non-viral plasmid DNA has been developed in recent years to enhance in vitro^1–3 and in vivo^4–6 gene transfer efficiencies. We have previously achieved high levels of gene expression in a murine liver model.^7,8 Therapeutic US of low temporal average intensity associated with substantial acoustic negative pressures (that is, 2–3 MPa or more) were used to facilitate the delivery of the mixture of naked plasmid DNA and exogenous microbubbles (MBs; US contrast agents). As spatial peak, temporal peak (SPTP) pressure amplitudes of 2–3 MPa correspond roughly to SPTP acoustic intensities of ∼130–300W cm⁻², such acoustic exposures can be justifiably classified as high-intensity therapeutic US (HITU) treatments. Compared with ‘HIFU’ (high-intensity focused US), the term HITU is neither constrained to involve focusing to achieve high intensities, nor does it carry the implication of substantial thermal effects (although it may include the latter). Our data have been acquired under such low temporal average intensities that significant thermal effects are unlikely; rather, the majority of our data are consistent with a mechanism mediated by the induction of inertial acoustic cavitation in the targeted tissues by US during injection of pDNA and exogenous MBs.

In order for pDNA injected intravascularly to be expressed by extravascular target cells, several physical barriers must be overcome. The first is the blood vessel wall. The extravasation of macromolecules, dyes and erythrocytes by acoustic cavitation is well known; HITU and MBs presumably aid the gene delivery process first by either permeabilizing or physically penetrating the walls of microvessels, allowing large molecules access to the extravascular space. The second barrier is the plasma membrane of the target cells. Sonoporation of individual cells by acoustically activated MBs have also been shown in multiple studies. Among these, elegant in vitro studies of US and MB-induced poration of Xenopus oocytes have been conducted using voltage clamp methods.⁹ Neither MBs nor US alone produced pores, but in combination, pores formed at acoustic pressures of 0.2–0.6 MPa. Pore closure occurred on time scales of seconds. Sonoporation increased with increasing US duty cycle; at high duty cycles this was associated with cell death.¹⁰

Gene transfer of pDNA is faster via sonoporation than with liposome-based methods that depend on endocytosis,¹¹ suggesting that pDNA enters cells through transient pores. Sonoporation in vitro is inversely proportional to exogenous MB lifetime, suggesting a direct role for MB destruction,¹² and the resulting inertial cavitation activity.^13,14 However, even low-amplitude MB oscillations can rupture lipid membranes,¹⁵ and there is a direct correlation between cell surface deformations produced by MB oscillations and membrane permeabilization.¹⁶ US exposure of suspended cells can strip surface receptors from viable cells¹⁷ and can achieve low levels of reporter gene expression.¹⁸ High-speed photography of MB dynamical behavior at ultrasonic frequencies indicates that microjetting (asymmetrical MB collapse) has a role in sonoporation.¹⁹ Recently, very high-speed microphotographic observations of acoustically driven MB dynamics in viscoelastic small blood vessels²⁰ indicate that jetting fluid flows directed towards inertially collapsing MBs are often directed away from the adjacent vessel wall, in essence ‘pulling’ the vessel wall inward toward the bubble and vessel lumen, with wall velocities reaching ∼10ms⁻¹ in about 2 μs. It is tempting to suppose that very rapid, forced distortions of microvessels and wall defects so produced contribute to the mechanism by which acoustic cavitation promotes the extravasation of normally impermeable materials; indeed, perhaps it is the dominant mechanism. Regrettably this conjecture, however tempting, remains speculative for now.

Although inertial cavitation is implicated in most in vitro sonoporation studies, in others, the US exposures suggest that non-inertial MB effects (for example, microstreaming) may also have a role.²¹ Cell cultures exposed for tens of minutes to very low-amplitude 1 MHz US showed increased gene expression with little loss of cell viability.²² Pre-incubation of cells with pDNA or liposomes can promote in vitro US-enhanced transfection,^22,23 but in the in vitro environment, attack by serum nucleases and removal by phagocytes is not an issue, as it may be in vivo.²⁴ Partial internalization of shelled MBs by phagocytes has been shown.^25,26 Hence, in vivo, injection of a non-viral gene vector long before US treatment is unlikely to be effective.

Many recent studies have shown the potential for use of US for in vivo therapies. US-enhanced gene transfer in vivo or ex vivo has been explored in the cornea,²⁷ spinal cord,^28,29 brain tissues,^30–32 kidneys,³³ pancreas,³⁴ liver,^7,8 embryonic tissues,³⁵ intervertebral discs³⁶ and dental pulp.³⁷ US has been used to enhance drug delivery to synovial cells in vivo³⁸ to enhance cell transplantation,³⁹ and to transfect living skin equivalents with pDNA in vitro; when transplanted into mice, gene expression persisted for weeks.⁴⁰ Three areas have attracted the most attention with respect to US-enhanced drug or non-viral gene delivery: (1) tumor treatment;^41–44 (2) cardiovascular treatment^45–50 and (3) skeletal muscle treatment.^51–54 In many of these studies, the mechanism(s) involved are unclear.

We previously used a first-generation therapeutic US system to stimulate the transfection of a plasmid vector carrying the factor IX and luciferase genes in mouse livers.^7,8 Treatments were associated with transient liver damage, but this was not extensive and was repaired rapidly. In this study, we have further optimized US instruments, transduction protocols and the selection of MBs to achieve high-level gene expression in mouse livers. The successfully developed instruments and methods have high potential in facilitating efficient gene transfer in mice, and further in larger animal models.

Results

Definity MB dose-ranging experiments in mice

We have previously shown that Optison, an albumin-based MB agent, can significantly enhance the efficiency of US-mediated gene transfer, and that gene transfer is sensitive to the MB concentration in the injectate, which delivers the plasmid DNA.^7,8 Definity MBs contain the same core gas (octafluoropropane) as Optison, but are stabilized by a phospholipid shell, and is ∼20-fold more concentrated. To determine if an apparent optimum concentration could be identified, dose-ranging studies were undertaken with Definity over a concentration range of 0–25 vol% in the final pDNA mixture injected into the animals. At high concentrations, the lipid shell could conceivably affect pDNA gene transfer by surfactant effects; therefore, sham-insonified controls at Definity concentrations of 5, 15 or 25 vol% were included in the experimental design. The acoustic exposures were carried out using the small-footprint-focused transducer and coupling cone system (System I) under the standard acoustic condition optimized in our previous studies (frequency=1.17MHz; pulse length=20 cycles; pulse repetition frequency (PRF)=50Hz; peak negative acoustic pressure=4.5MPa).

The concentration of Definity MBs in the plasmid injectate had a strong influence on gene expression under otherwise comparable acoustic exposure conditions (Figure 1). With US exposure, there was little gene expression at Definity concentrations < 1% compared with the values of the groups without Definity (340 ± 79 relative light units per mg protein (RLU per mg protein)) or without US (380±91RLU per mg protein). At concentrations between ∼1 and ∼15 vol%, gene expression increased significantly with increasing Definity concentration, attaining an apparent maximum at a concentration of ∼15 vol% (8.3 ± 1.9×10⁴ RLU per mg protein), and then declining at higher concentrations. The highest level of gene expression was about 160-fold greater than those obtained from 0 vol% Definity controls, and significantly higher than all other groups. Expression levels in mice sham-exposed to US, but injected with plasmid and 5, 15 or 20 vol% Definity were at background levels, indicating no discernible surfactant effect.

Figure 1.

Dependence of luciferase gene expression on Definity concentration mediated by US exposure with simultaneous intraportal injection of pGL3 plasmid and Definity. Different concentrations of Definity (vol%: 0–25%) were used under what had been our standard positive control conditions for the focused HITU system (viz, 1.17MHz, 4.5MPa peak negative pressure, 20 cycle pulses, 50Hz PRF and 60s total US exposure, the first 30s of which were coincident with injection of pDNA and Definity). To present the Definity concentrations on a log scale (to better resolve the data at low Definity concentrations), the true 0 vol% Definity data are plotted at an arbitrarily small value (0.001 vol%) concentration as a graphic expedience. Solid black symbols represent reagent controls, that is, Definity+pDNA without ultrasound exposure. Error bars indicate s.e.m. (n=8–14).

We also compared the dependence of gene transfer efficiency upon peak negative pressure in the presence of Definity or Optison. During the course of this study, we have improved our methods for evaluation of luciferase gene expression levels. With a more optimized luciferase assay, lower luciferase activities were obtained from samples isolated from control mice treated with either plasmid only or plasmid+Definity only. We found that a maximum of 700-fold enhancement of gene expression levels was achieved using Definity compared with a maximum of 85-fold enhancement with Optison MBs under the same US exposure (Supplementary Figure 1).

Effect of co-agitation of plasmid DNA with Definity precursor solution on US-mediated gene delivery

When examined under the microscope, fluorescently labeled DNA mixed with Definity MBs showed a slight adherence after mixing, although there is no electrostatic attraction between DNA and Definity MBs (data not shown). On the basis of our optical observations, it was hypothesized that physical association between DNA and MBs may be increased if pGL3 plasmid was co-agitated with the Definity MB precursor solution to generate MB particles. The gene expression of pGL3 co-agitated or simply mixed with 15% Definity MBs was explored. Co-agitation was carried out by dissolving re-precipitated plasmid DNA in the Definity solution, refilling the mixing vial with the perfluorocarbon gas and then shaking for 45 s. The co-agitated or simple mixtures were injected via the portal vein for 30 s with simultaneous 3 MPa US exposure for 60 s.

Both the co-agitated mixtures and the simple mixtures of pDNA and Definity MBs significantly enhanced US-mediated gene expression relative to negative controls (pGL3 and Definity simple mixtures without US treatment); however, the effects were equivalent (Figure 2). Thus, the hypothesis was not supported by the data, and hoped-for technical improvement was not achieved by this acoustic exposure protocol.

Figure 2.

Effect of co-agitated plasmid DNA and Definity on US-mediated gene delivery. In total, 50 μg pGL3 with or without co-agitation with Definity was delivered through intraportal injection for 30s into the mouse liver, which was simultaneously exposed to 3 MPa US for 60s. Control mice were injected with plasmid and Definity without US exposure. *P<0.05, compared with negative control group without US exposure and co-agitation. Error bars indicate s.e.m. (n=7, 16, 9, 8).

Effect of pulse train US exposure on gene delivery

The objective of this experiment was to attempt to increase the treatment efficacy through the use of interleafed periods of exposure to repetitive US pulses of sufficient amplitude to destroy contrast agent MBs and quiescent periods between US exposures to enable reperfusion of the liver tissues with MBs. It was hypothesized that longer quiescent periods between bursts of short pulses would thus increase treatment efficacy. Under each condition, the gated ‘on’ periods in which US was emitted at 50Hz PRF was 2 s per event; the ‘off’ period between events was 0 (control; uninterrupted delivery of pulses at 50 Hz PRF), 1 or 2 s (Figure 3a). In all cases, the SPTP-negative acoustic pressure was 3 MPa. System I was used in these studies. The pulse length was adjusted slightly to maintain total ‘on’ time a constant. We used inter-burst quiescent periods of seconds on the basis of macroscopic observation of liver perfusion rates using injected dyes. An external pulse generator (Model DG 535; Stanford Instruments, Sunnyvale, CA, USA) was used to provide precise gating signals. All acoustic exposures were 60 s in total duration, the first 30 s of which were coincident with intraportal injection of pGL3 and Definity MBs. The acoustic parameters are shown in Table 1.

Figure 3.

Effect of discontinuous acoustic exposure on gene delivery. US exposure was performed using either uninterrupted bursts of 20 cycle pulses repeated at 50 Hz PRF for 60 or 2 s exposures to bursts of pulsed US, interrupted by quiescent periods of 1 or 2 s. All groups were treated by US exposures for 60 s simultaneous 30 s injection of pDNA and MBs via the portal vein. (a) Schematic diagram of treatment with US and intraportal injection. The upper sketch represents routine pattern of continuous 30 s injection and 60 s US exposure. The lower sketch represents the pattern of continuous injection and discontinuous US exposure used in this experiment. (b) Trend lines of specific activities plotted against off time between pulse trains. KEY (all regarding specific activity values). Open squares: maximum values; open triangles: minimum values; open circles: median values; and filled circles: mean values. Error bars indicate s.e.m. (n=8, 7, 5). Trend lines determined by linear regression are provided for each data set, as are the associated Pearson's correlation coefficients. Gene expression tends to improve with 2 s on 1 s off or 2 s on 2 s off, although the differences are not statistically significant with P>0.05.

Table 1. Acoustic parameters used in pulsed train exposure experiments, in which 2-s periods of exposure to pulsed US were interleaved with quiescent periods of 0 s (positive control, that is, 60 s of on-going exposure to pulses repeated at 50 Hz), 2 s ‘on’:1 s ‘off’, or 2 s ‘on’:2 s ‘off’.

Treatment	Pk neg press (MPa)	Cycles per pulse	PRF while ‘on’ (Hz)	Pulse duration (μs)	Tot. exp. duration (s)	Tot. ‘on’ duration (μs)
2 s ‘on’:0 s ‘off’	3.0	20	50	17.1	60	51 300
2 s ‘on’:1 s ‘off’	3.0	30	50	25.6	60	51 300
2 s ‘on’:2 s ‘off’	3.0	40	50	34.2	60	51 300

Abbreviations: Pk neg press, peak negative pressure; PRF, pulse repetition frequency; US, ultrasound.

The results of these explorations are illustrated in Figure 3b. Although the mean values obtained for gene expression were not increased significantly relative to the positive control for either of the experimental pulsed train exposures, four measures of gene expression (maximum, minimum, median and mean) all showed a positive correlation between increasing ‘off’ time between bursts and gene expression, suggesting that modest gains in expression can be obtained if the US exposure is correctly timed. However, these data are preliminary and limited in scope, and the hypothesis needs to be tested further.

Efficacy testing of the H-158A unfocused transducer system in the murine model

The H-158A unfocused transducer HITU system (System II) was devised to enable larger tissue volumes to be treated more rapidly at supra-threshold acoustic pressures than would be possible with our focused US system. Initial tests of the efficacy of System II were conducted in mice, comparing treatment outcomes produced by the System I focused transducer (−6 dB beam diameter=2.7mm; ‘footprint’ area ∼0.06cm²) with outcomes produced using the System II unfocused transducer system applied at the near-field–far–field (NF–FF) transition (−6 dB beam diameter ∼8.5mm; ‘footprint’ area ∼0.6cm²), or applied at the near-field configuration (−6dB beam diameter ∼12mm; ‘footprint’ area ∼1.1 cm²).

In our first biological testing of System II, the efficacy of treatment with Systems I or II (far-field exposure) were compared directly, using pGL3 plasmid to provide the reporter gene (luciferase). With the exception of beam width, the HITU exposure conditions were matched as closely as achievable. With System I, the conditions were as follows: frequency=1.17MHz; pulse length=20 cycles; PRF=50Hz; and peak negative acoustic pressure=3.0MPa. With System II, the acoustic frequency was slightly lower (1.1 MHz), the peak negative acoustic pressure nominally lower (2.96 MPa) and all else the same. Mice injected with plasmid/Definity mixtures in the absence of US exposure were used as negative controls.

In subsequent biological testing of System II, we compared the efficacy of the H-158A transducer operated in the near-field condition with exposures conducted at the NF–FF transition zone. When comparing results across experiments in this paper, it is important to note that a more efficient plasmid DNA pGL4 was used in this experiment. The near-field acoustic exposure conditions were as follows: acoustic frequency=1.1MHz and 3.15MPa peak negative pressure. Pulse lengths remained at 20 cycles as before, but the PRF was reduced from 50 Hz (as in earlier experiments) to 13.9 Hz PRF, because to produce acoustic pressures of such amplitudes at the face of the transducer required sufficiently high-power levels from the power amplifier to exceed allowable settings at higher pulse repetition frequencies or duty cycles. The PRF used was thus the maximum attainable using 20 cycle pulse lengths.

As shown in Figure 4, exposure of the murine liver to US from either the focused 1.17 MHz transducer or that of the unfocused H-158A transducer applied at the NF–FF transition zone promoted luciferase expression to nearly identical extents. The enhancements produced by both systems were significantly greater than in negative controls, but were statistically indistinguishable between exposure treatments. The result indicates that with System II, the gene delivery efficiency can reach a level that was previously obtained with System I. However, exposure to the larger beam of the H-158A transducer far field did not increase gene expression relative to that produced with the focused HITU (System I), suggesting that adequate—and ‘saturating’—exposure had been attained in the small liver of the mouse with either system.

Figure 4.

Luciferase expression mediated by ultrasound using US Systems I or II with far-field transducer configuration. In total, 50 μg pGL3 mixed with 15% Definity was injected via the portal vein into the mouse liver for 30 s with simultaneous acoustic exposure for 60 s using two different US Systems (I and II) as described in the Materials and Methods section. Gene expression from mice injected with plasmid and Definity without US treatment was used as the negative control. Equivalent expression results were obtained with both systems. *P<0.05, compared with negative control group without acoustic exposure. Error bars indicate s.e.m. (n=5, 5, 7).

Liver damages in mice treated by Systems I or II were also evaluated by transaminase assay at 24 h after treatment. Table 2 shows the alanine transaminase (ALT) and aspartate transaminase (AST) values of plasma from retro-orbital bleeding. Compared with either ALT or AST levels of normal mice (∼50IUl⁻¹) and negative control mice (∼100IUl⁻¹), the mice treated by Systems I or II had similarly increased values (∼1000IUl⁻¹), indicating that some liver damage was produced by US exposure using either system, and that the extent of damage as assessed by liver enzymes was about the same with either system.

Table 2. ALT and AST of plasma collected from mice treated with different transducers and US systems.

Treatment	Normal	pGL4+MB	pGL4+MB+System I	pGL4+MB+System II (far field)	pGL4+MB+System II (near field)
ALT (IUl−1)	39 ± 15	139 ± 57	1528 ± 97	1418 ± 390	998 ± 156
AST (IUl−1)	50 ± 18	136 ± 35	845 ± 92	900 ± 273	514 ± 151

Abbreviations: ALT, alanine transaminase; AST, aspartate transaminase; MB, microbubbles; US, ultrasound.

No significant difference in luciferase gene expression was observed between the mice treated under near-field or NF–FF transition zone exposure conditions; expressions levels were about 200 000 and 300 000 RLU per mg protein, respectively, and did not differ from one another. Both were significantly elevated relative to the negative controls, and were about 700-fold greater than negative controls without US exposure (Figure 5). As for the study in liver damage, either ALT or AST of mice treated by near-field exposure showed lower values than ones treated by NF–FF exposure (Table 2). In addition, during harvest the treated liver of both groups extensively appeared congestion-like dark-red, but the livers treated by NF–FF exposure showed more focused white necrosis-like dots on the liver surface. It is possibly of significance, in the conventional sense of the word, that the transaminase levels and macroscopic appearance of the livers indicated that the near-field exposures generated less liver damage than did exposures at the NF–FF transition zone.

Figure 5.

Luciferase expression of US-mediated gene delivery using near-and far-field exposures produced by US System II. pGL4 plasmid and Definity microbubble were mixed and then injected intraportally into mice for 30 s with simultaneous US treatment for 60 s. Gene expression from the mice injected with plasmid and Definity, but without US treatment, was used as the negative control. *P<0.05, compared with negative control group without US exposure. Error bars indicate s.e.m. (n=5).

Effect of peak negative pressure generated by System II (near field) on gene delivery

The experiments were performed with System II (1.1MHz, 20 cycle pulses, 13.9 Hz PRF) using near-field acoustic exposure of 0–3.2 MPa spatial average, temporal peak-negative acoustic pressure amplitude across the near-field beam. Plasmid DNA pGL4 was used in this experiment.

As shown in Figure 6a, luciferase expression increased gradually from 0 to 1.8 MPa, and then increased slowly with further increments in acoustic pressure up to ∼3.2 MPa. It appears that a limiting value is approached as acoustic pressure amplitudes attain or exceed ∼2MPa.

Figure 6.

The effect of US peak negative pressure on gene delivery efficiency and its association with liver damage via intraportal injection of the pDNA+MB mixture. Experimental mice were treated by ultrasound System II with the transducer applied in the near-field configuration. Gene expression was measured on the whole liver. The experiments were performed with modified conditions upon acoustic exposures of 0–3.2 MPa peak negative pressure (1.1 MHz, 20 cycle pulses, 13.9Hz PRF and 60 s total exposure, the first 30 s of which was co-incident with injection of pGL4 plasmid and Definity). Definity concentration was 13.6%. (a) Luciferase gene expression upon exposure at various US peak negative pressures. The inflection point was observed on 1.8MPa, followed by a plateau from ∼2 to 3.3 MPa. (b) Correlation between gene expression and liver damage produced by exposure to acoustic pressures of 0–3.2MPa (negative). Liver damage was assessed by ALT and AST levels with plasma collected at 24 h after treatment. Numbers adjacent to data points are the negative acoustic pressures associated with the gene expression and liver damage levels. Error bars indicate s.e.m. (n=5).

In the hope of understanding the mechanism that underlies the apparent limiting value of gene expression produced by US exposures of various acoustic pressures with intravascular MBs present, we explored the correlation of enhanced gene expression and liver damages potentially induced by acoustically-activated microbubbles. The transaminase levels (ALT and AST) 24 h after treatment both increased with increasing acoustic pressure, and correlated nearly 1:1 with each other (data not shown), indicating that either enzyme is equivalently useful as a spot estimator of liver damage for present purposes. In Figure 6b, gene expression is plotted as a function of liver damage as assessed by ALT levels 24 h after treatment. It shows that over the range from 100 to 400 IUl⁻¹ ALT, corresponding to 0–1.8 MPa, gene expression increased very rapidly (from ∼300 to ∼116 000 RLU per mg protein) with modest increases in liver damage (from ∼140 in surgical controls to ∼420IUl⁻¹), that is, gene expression increased about 400-fold with a threefold increase in liver damage. However, over the range of ALT levels from 400 to 1000IUl⁻¹, corresponding to a range of acoustic pressures of ∼2– 3.2 MPa, there was only <2-fold increment in gene expression. This result suggests that the acoustic pressure treatment at which the crossover between productive and non-productive liver damage occurs is around 2 MPa, and that treatment at higher pressures may produce more damage, but may not produce more of the desired effect.

Discussion

US has been shown to be a highly promising method to enhance the efficiency of non-viral gene transfer both in vitro and in vivo. In particular, the presence of MBs is critical in US-mediated gene transfer.^1,55 Various types of MBs with different materials, sizes, concentrations, and so on might produce different effects under certain mechanisms.^56–58

In our previous study, gene expression of plasmid DNA co-injected with Optison MBs into the liver via the portal vein was enhanced efficiently by acoustic exposure. An inertial cavitation mechanism was indicated. In this study, we tested Definity in our US-mediated gene delivery experiments in place of Optison. Optison is an albumin-coated MB, whereas Definity is a lipid-encapsulated MB; both contain the same perfluoropropane gas. The flexible mono layer lipid shell causes Definity to be more stable to acoustic disruption than Optison, although the threshold is nonetheless on the order of ∼0.5MPa. In addition, Definity suspensions contain much higher MB concentrations (maximum 2×10¹⁰ bubbles per ml) than those of Optison (5∼8×10⁸ bubbles per ml). All else being equal, more cavitation nuclei are expected to produce more cavitation activity. In addition, at high concentrations, the lipid shell may affect pDNA gene transfer via surfactant effects. Our results showed the relative gene expression enhancement with the use of Definity (700-fold) was more than that with Optison (85-fold) under the same standard US exposure. In addition, gene expression showed a dependence of Definity MB concentrations up to 15 vol%, consistent with the cavitation mechanism. Although luciferase expression levels did not vary significantly at the bubble concentrations of 2.5–25%, luciferase expression showed a downward trend at MB concentrations over 15 vol%. A similar trend in transgene expression vs MB concentrations was reported in an in vitro study by Rahim et al .,⁵⁹ which was attributed to the progressive decrease in cell viability. Thus, the gene expression probably reached a peak level at a Definity concentration where ‘constructive’ or ‘productive’ cavitation-related damage (that is, damage that promotes gene transfer) still outweighed purely destructive tissue damage.

Cationic lipids can greatly enhance gene delivery by combining with DNA to construct lipid/DNA complexes. A significant increase of gene expression in tumors was obtained using US in cationic lipid-mediated gene transfer.⁶⁰ Subsequent studies using cationic MBs constructed to carry genes as well as bear targeting ligands resulted in encouraging enhancements.^61,62 The Definity MBs used in our study is a lipid-coated MB without surface charge. Nonetheless, we observed by microscopy that pDNA can attach subtly to Definity. Therefore, we co-agitated Definity suspension with DNA to attempt to enhance gene expression by increasing the association between Definity and pDNA. However, co-agitation did not produce further enhancement.

We have also explored pulsed train US exposure protocols in a preliminary experiment. It was hypothesized that intermittent exposures would increase the gene delivery efficiency by providing quiescent periods between groups or series of acoustic pulses in which new MBs might replenish those destroyed by the preceding acoustic pulses. Previously, we found that use of relatively short inter-burst quiescent periods (up to 300 ms) did not demonstrably improve gene transfer in the mouse. In the experiments described here, we used longer inter-burst quiescent periods (1 or 2 s), mimicking the time required for injectate to distribute throughout the liver as observed by injecting dye into the portal vein, using the same protocols as when injecting plasmids and MBs. It is encouraging that a positive correlation was observed between the inter-burst time and gene expression. Better enhancement may be achieved by systematic investigation of varying the timing, PRF and other parameters.

In the studies on gene delivery with US System II, to optimize parameters for gene transfer efficiency, the transducer was applied in either near or far-field configuration. Although the efficacy of US System II with either far- or near-field transducer configurations is similar to that obtained using US System I, the livers treated with System II using the near-field exposure configuration showed lower transaminase levels and macroscopic damages. This may be due to the lower PRF used with near-field exposures with System II (∼14 Hz); a PRF of 50 Hz was used for both the focused System I and the unfocused System II as applied in the far-field configuration. Furthermore, our new US system equipped with an unfocused transducer allowed unfocused acoustic exposure with more homogeneous average intensity compared with a highly focused acoustic exposure using the old HIFU system. This study tested a novel design of the combination of a new US system with a higher power source and an unfocused transducers with a larger footprint and validated the feasibility of the new system to effectively treat mouse livers to enhance gene delivery efficiencies. This new system equipped with the unfocused transducers with a larger footprint will allow the scanning and treatment of larger tissue areas in larger animals in a relatively short period of time, which cannot be achieved by an HIFU system with very small focal area. This is critical as the plasmid solution was infused into a dynamic system and can only be retained in the target tissue for short time during in vivo gene delivery. This study represents an important developmental step towards US-mediated gene delivery into larger animals and clinical trials.

In all of our in vivo experience with US- and MB-enhanced non-viral gene delivery, exposures that are effective always produce some level of mechanical damage to the tissues, enabling extravasation of the plasmid and perhaps creating transient pores in hepatocytes. This straightforward observation gave rise to the questions: Is there a correlation between gene expression and tissue damage? Is there a point at which US- and MB-promoted damage ceases to have beneficial effects and becomes merely destructive? The answer to both questions appears to be ‘yes’. In the study of gene transfer with US exposure under various pressures, we investigated not only the relationship between luciferase expression and peak negative pressure, but also the relation of luciferase expression to liver damages. The efficiency of gene delivery mediated by US and Definity showed a dependence on acoustic peak negative pressure, similar to our previous study with Optison, with characteristics indicating that an inertial cavitation mechanism is involved. Meanwhile, the correlation of gene expression and transaminase levels presented a crossover point of ‘productive’ damage and ‘over-treated damage’, which could promote the gene transfer effect to a limiting level, which seems to represent a treatment regimen that produces the maximal desired effect (gene transduction) for the least treatment cost (tissue damage). An upper bound on gene transfer efficiency that can be achieved by increasing MB concentration may arise by a similar mechanism, although other physical mechanisms may come into action when MB concentration is very high. Thus, decreasing acoustic pressure or MB concentrations within the range at which gene expression levels reaching the plateau would be a promising strategy to minimize the liver damage for future clinical application.

In this study, we have further optimized US instruments, transduction protocols and the selection of MBs to achieve high-level gene expression in mouse livers. Treatment using US and lipid-based Definity MBs was shown to dramatically enhance delivery of plasmid DNA to the murine liver. Unfocused, broad ‘footprint’ US beams of high temporal peak pressure amplitudes were shown to be an effective means to treat the murine liver without the need for significant scanning, which has implications for scaling up to treatment of significantly larger organs. Specifically, with a broad footprint and scanning, it should be possible to treat large tissue volumes rapidly and thus expeditiously expose injected MBs and plasmids during their brief transit through the target organ. The study of pulse train US exposures suggested a possible strategy to more evenly distribute pDNA/MB mixtures into the liver to generate more effective gene delivery. The dependence of gene expression on MB dosages or US pressures provided a range of the two parameters, in which the gene transfer efficiency approaches a plateau. Furthermore, the correlation between gene expression and liver damages produced by US treatment implied that an ‘optimal’ treatment, that is, one which produces the most gene transduction with the least tissue damage, can be achieved in practice by systematic optimization studies. Our present data provide early indications of what these conditions might be. At present best, we achieved an approximately 700-fold enhancement of gene expression in treated animals relative to controls. To our knowledge, this is the best enhancement achieved using US-facilitated gene delivery. These experiments were essential in preparation to scale up gene delivery in vivo from mouse models to larger animal models before human clinical trials.

Materials and Methods

Preparation of plasmids

The luciferase reporter plasmids pGL3 and pGL4 were used in gene transfection. pGL3 (Promega, Madison, WI, USA) was amplified in Escherichia coli and purified using Endo-free plasmid mega or giga kit (Qiagen Inc., Valencia, CA, USA), according to the manufacturer's protocol. Plasmid concentration was measured by the absorbance at 260 nm with NanoDrop ND-1000 UV–Vis Spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). pGL4.13 [luc2/SV40] (Promega) was prepared by GenScript Inc. (Piscataway, NJ, USA).

US systems

Two US systems were employed in this study. System I, as described previously,⁸ was comprised by an Agilent Model 33120A arbitrary function generator whose signal was amplified by an ENI AP400B amplifier (Electronic Navigation Systems, Rochester, NY, USA), driving a home-made transducer having an air-backed, 35mm diameter PZT (Lead Zirconate Titanate, PbZrTi) element and an aluminum lens of F-number 1. The lens was fitted with a water-filled polycarbonate cone whose truncated tip was located at the focus of the acoustic beam; the cone was sealed with a polyurethane membrane. Transducer output was calibrated using a membrane hydrophone (Model MHA 200; NTR Systems, Seattle, WA, USA) to measure acoustic pressures. The full-width, half-maximum pressure beam diameter at the focus was 2.7 mm, corresponding to a treatment ‘footprint’ of ∼0.06 cm². Transducers were driven at their resonance frequencies (1.10–1.17 MHz).

Acoustic System II had as its core a purpose-designed single element, nominally 16mm diameter transducer (Model H158A; Sonic Concepts, Bothell, WA, USA) capable of dissipating large temporal peak electrical powers and of generating high-pressure amplitudes across most of the transducer face. The transducer was driven by a combination pulse generator and high-power radio frequency amplifier capable of generating up to 1000W into a 50Ω load (Model RFG-1000; JJ&A Instruments, Duvall, WA, USA) and which was under the control of a laptop PC. Transaxial acoustic pressure output measurements were made by membrane hydrophone at two planes: (1) as close as possible to the face of the transducer (at 6.2mm distance; near field), and (2) at the tip of the truncated cone (NF–FF transition). The full-width, half-maximum pressure beam diameter at the NF–FF transition was approximately 8mm. For this configuration, the SPTP pressure amplitude was taken as the value representative of exposure. In the near-field, the spatial average rarefaction pressure over the full-width, half-maximum beam diameter was taken as the pressure metric.

The acoustic treatment to the liver was performed by applying the transducer or the tip of the coupling cone directly to the surface of the surgically exposed liver, using a small amount of commercial diagnostic US gel used to couple the transducer to the liver. Where transducers with cones were applied, a roughly circular scanning motion was performed manually. Where the H-158A transducer was applied using near-field exposures, mechanical clearance issues did not allow for much transducer movement in the mouse model; under these conditions, the exposed liver can be considered to have been exposed as if to the beam of a stationary ‘spotlight’.

Microbubbles

Definity (Lantheus Medical Imaging, Billerica, MA, USA) is a US contrast agent of lipid microspheres containing octafluoropropane gas. According to the manufacturer, the maximum concentration of Definity is 1.2×10¹⁰ bubbles per ml, with most MBs in the diameter range of 1.1–3.3 μm. In most of the experiments described here, aliquots of pre-agitated Definity MBs were mixed with plasmids in phosphate-buffered saline immediately before injection into mice. In other experiments in which plasmids and Definity were co-agitated, an unagitated vial of Definity was opened by removing the septum, which resulted in the loss of the octafluoropropane gas originally present in the vial head space. A small aliquot of highly concentrated plasmid was added to the vial contents, and the septum was then replaced. An 18 G needle was inserted through the septum into the vial headspace to serve as a vent. A second 18 G needle was used to reintroduce octafluoropropane gas into the vial, using many headspace turnover volumes as measured using a flow meter calibrated for octafluoropropane. Both needles were then withdrawn, the vial sealed with a paraffin laboratory film and the vial agitated immediately before use as would normally be carried out to generate Definity MBs.

Animal procedures

Eight-week-old C57/BL6 male mice were purchased from the Jackson Lab (Bar Harbor, ME, USA) and maintained at a specific pathogen-free vivarium. All animal experiments were carried out in accordance with the guidelines for animal care of both National Institutes of Health and Seattle Children's Research Institute. Mice were anesthetized by continuous inhalation of isofluorane during treatment and recovered within 2 h following treatment. A midline incision was made to expose the liver and portal vein in mice.

In most experiments, simple mixtures of pre-agitated Definity and plasmid DNA were used. In those experiments, 50 μg plasmid DNA per mouse was dissolved in phosphate-buffered saline containing 5% glucose, and then mixed with Definity MBs at desired concentrations just before injection into the animals. For each mouse, about 400 μl total volume of plasmid and MB mixture was injected into the liver via the portal vein using a 30-G needle. The duration of the injection was 30 s. US treatment of the mouse liver started simultaneously to injection, and continued for 30 s after injection concluded, that is, the total exposure duration was 60 s. With immediate hemostasis and suturing, the mice recovered from anesthesia and lived with low apparent morbidity and very low mortality until the second day, when the animals were killed for liver harvest. Blood samples were collected by retro-orbital bleeding immediately before killing the animal, and were used to assess liver damage by measuring ALT (or SGPT) and AST (or SGOT) levels (see below).

Evaluation of luciferase gene expression

To analyze luciferase gene expression in the liver, all liver lobes were harvested at 24 h after gene delivery and homogenized in reporter lysis 1×buffer (Promega) at a ratio of 3mlg⁻¹ liver. The lysis was performed with three cycles of freezing and thawing of the homogenate to ensure complete release of luciferase. After vortexing and centrifuging at 18 000 g, the supernatant was transferred and stored at −80 °C until measurement. Luciferase activity was measured by using Luciferase Assay System (Promega) and a luminometer (Victor 3; Perkin-Elmer, Wellesley, MA, USA). Luciferase activity was normalized to the total protein content of the tissue samples, and is expressed using the dimensions of RLU per mg protein.

Transaminase assay

Blood samples were collected from treated mice 24 h after gene delivery for the transaminase assay. ALT and AST levels were determined by using commercial Teco Diagnostics assay regents (Anaheim, CA, USA). Normal, untreated mice were used as controls.

Statistical analysis

All data are shown as mean±s.e.m. Student's t-test was used to determine statistical significance for independent samples. Data were considered significant when P-value <0.05.