Ultrasound with microbubbles enhances gene expression of plasmid DNA in the liver via intraportal delivery

Abstract

Current ultrasound (US)-mediated gene delivery methods are inefficient due, in part, to a lack of US optimization. We systematically explored the use of microbubbles (MBs), US parameters and plasmid delivery routes to improve gene transfer into the mouse liver. Co-presentation of plasmid DNA (pDNA), 10% Optison MBs and pulsed 1-MHz US at a peak negative pressure of 4.3 MPa significantly increased luciferase gene expression with pDNA delivered by intrahepatic injection to the left liver lobe. Intraportal injection delivered pDNA and MBs to the whole liver; with insonation, all lobes expressed the transgene, thus increasing total gene expression. Gene expression was also dependent on acoustic pressure over the range of 0–4.3 MPa, with a peak effect at 3 MPa. An average of 85-fold enhancement in gene delivery was achieved. No enhancement was observed below 0.25 MPa. Increasing pulse length while decreasing pulse repetition frequency and exposure time to maintain a constant total energy during exposure did not further improve transfection efficiency, nor did extend the US exposure pre- or postinjection of pDNA. The results indicate that coupled with MBs, US can more efficiently and dose-dependently enhance gene expression from pDNA delivered via portal vein injection by an acoustic mechanism of inertial cavitation.

Introduction

Nonviral gene transfer represents a safer and less costly gene delivery method compared with viral gene transfer. The potential for therapeutic ultrasound (US) to enhance the efficiency of drug or nonviral gene transfer has recently received much attention. Many of the desired characteristics of gene therapy can be found in the combination of ultrasound-targeted microbubble destruction (UTMD) and plasmid vectors, including low toxicity, low immunogenicity, the potential for repeated application, organ specificity and broad applicability to acoustically accessible organs.¹ We have recently obtained significant enhancement of human factor IX (hFIX) gene delivery into a murine liver model.² However, the efficiency of UTMD to transfect plasmid DNA (pDNA) was insufficient for clinical applications.

US exposures, particularly those involving acoustically excited microbubbles (MBs), can create transient pores in sonoporated cell membranes and can produce petechial hemorrhages in vivo;^3,4 the latter permits extravasation of injected materials from the microvasculature into the interstitial spaces of surrounding tissues, and thus overcomes one of the barriers that impede pDNA incorporation into nonvascular tissues. The potential for sonoporation to increase pDNA loading of cells was recognized years ago,⁵ and a large literature has grown since. Of special interest here is the use of cavitation-related mechanical bioeffects to enable or enhance pDNA gene transfection. A key feature of US-enhanced gene delivery is that DNA delivery can be targeted.

Sonoporation in vitro is usually attended by high cell mortality, with a small fraction of surviving cells showing probe molecule uptake.⁶ Sonoporation is associated with inertial cavitation (IC) and is increased by the addition of MBs.^7–9 For example, the enhancement of transfection of Chinese Hamster Ovary (CHO) cells on exposure to 2.2 MHz US was critically dependent on exogenous MBs.⁶ Under appropriate US conditions, MB-dependent, pDNA-based transfection could occur by ‘spontaneous’ acoustic generation of cavitation nuclei.¹⁰ The involvement of sonochemicals produced by IC in sonoporation is not settled.^11,12

The potential for the use of UTMD for in vivo therapies has been the subject of many recent review articles.^13–15 Three areas have attracted the most attention with respect to US-enhanced drug or nonviral gene delivery: (1) tumor treatment,^16–19 (2) cardiovascular treatment^20–23 and (3) skeletal muscle treatment.^24–26 UTMD-enhanced gene transfer in vivo or ex vivo has also been explored in the cornea,²⁷ spinal cord,^28,29 brain tissues,^30,31 kidneys,³² pancreas,³³ liver,² embryonic tissues,³⁴ intervertebral discs³⁵ and dental pulp.³⁶ Although acoustic cavitation is generally associated with sonoporation, in many of these studies the specific mechanism responsible for increasing the gene delivery efficiency is uncertain. In some cases, the acoustic pressures used suggest that IC was involved, whereas in others, the pressure amplitudes were lower than that would be expected to produce IC, suggesting the involvement of stable cavitation of gas bodies generated by the acoustic disruption of stabilizing MB ‘shells.’ To date, this method has not been translated to clinical therapy in animals or humans.³⁷

In this study, our aim was to begin to optimize the efficiency of US-facilitated gene delivery by using a luciferase reporter gene. The following five specific hypotheses were addressed: (1) If the gene transfection enhancement effect is mediated by acoustic cavitation, the effect should be strongly influenced by the presence or absence of exogenous cavitation nuclei. (2) The use of longer acoustic pulses and lower pulse repetition frequencies will be more effective than shorter pulses and higher pulse repetition frequencies by promoting sustained cavitation activity during the longer pulses and by allowing more time between pulses for perfusion to replace MBs destroyed during the preceding acoustic pulse. (3) Intraportal injection will be more effective than direct intrahepatic injection by allowing plasmids and MBs direct access to the hepatic sinusoids, allowing transfection of whole liver instead of single liver lobe. (4) Pretreatment or extended exposure of the liver with US will increase gene expression by ‘sensitizing’ the tissues prior to the principal treatment or by activating cavitation nuclei that had not been destroyed by the principal acoustic treatment. (5) If the gene enhancement effect is mediated by an IC mechanism, then the acoustic peak negative pressure amplitude dependence of the effect should show characteristics of IC dose evolution. These hypotheses were tested, and our study demonstrated that robust and consistent enhancement of gene expression was achieved in the mouse liver via intraportal injection of pDNA with the best US protocol, and strongly support the overarching hypothesis that the effect is mediated by an IC mechanism.

Results

Effect of US contrast agent (microbubbles) and US exposure on gene delivery via intrahepatic injection

To test the hypothesis that acoustic cavitation is the dominant mechanism by which the gene delivery enhancement is achieved with US exposure, we performed experiments using pGL3, in the presence or absence of exogenous MBs and with or without US. Four treatments were compared: plasmid only, plasmid+MB, plasmid+US or plasmid+MB+US.

In the absence of US, pDNA alone produced a background luciferase activity (1.89±1.29×10³ Relative Light Unit (RLU) per mg protein, Figure 1). The addition of Optison MB to the plasmid solution did not enhance luciferase gene expression without US (1.10±0.34×10³ RLU per mg protein, P =0.98 vs pDNA only). Without Optison MBs, US produced a nominal but insignificant increase in luciferase expression (3.45±1.07×10³ RLU per mg protein, P =0.92 vs pDNA only). It was only when Optison MBs were added that US significantly enhanced luciferase gene expression (1.07±0.4 ×10⁵ RLU per mg protein, P =0.0005 vs pDNA+Optison MB, which was a 97-fold increase vs pDNA+Optison). These results indicate that US can significantly improve plasmid gene delivery and that MBs are essential to the process in our model, but MBs alone were without effect on gene delivery.

Figure 1.

Gene expression enhanced by ultrasound (US) with microbubbles (MBs). A 50 μg portion of pGL3 mixed with or without 10% Optison MBs was directly injected into the left liver lobe of the mice in 30 s, with or without 60 s US exposure. **P<0.01 vs pGL3 alone or negative control. N =3, 4, 4, 3, respectively.

Effect of varying US pulse length, pulse repetition frequency or total exposure time for enhancing gene expression via intrahepatic injection

It was hypothesized that longer pulses at lower pulse repetition frequency (PRF) would increase the effect by allowing longer quiescent periods between pulses during which new pDNA and MBs could enter the tissue and replenish those MBs destroyed by the preceding acoustic pulse (via lower PRF) and promote sustained IC activity (via longer pulse durations). To test this, we increased US pulse length while decreasing PRF and exposure time to maintain constant total US energy exposure (see Table 1). Three modifications of US parameters were effective in enhancing gene delivery (vs negative control, P<0.01); however, no further enhancement was detected compared with the positive control (Figure 2). Thus modifications in pulse length and PRF appear to have little importance for enhancing the rate of transfection in this model using intrahepatic injection.

Table 1.

Acoustic parameters used in the US-mediated gene transfer experiments

Treatment group	Acoustic parameters
	Frequency (MHz)	Peak negative pressure (MPa)	Pulse length (cycles)	Pulse duration (μs)	Pulse repetition frequency (Hz)	Duty factor	US time (s)	Total acoustic exposure (ms)
Negative	0	0	0	0	0	0	60	0
Positive	1.17	4.3	20	17	50	0.0009	60	51.3
Model 1	1.17	4.3	40	34	50	0.0017	30	51.3
Model 2	1.17	4.3	333	285	3	0.0009	60	51.2
Model 3	1.17	4.3	667	571	3	0.0017	30	51.3

Abbreviation: US, ultrasound.

Figure 2.

Gene expression mediated by modified ultrasound (US) with pGL3 delivered via intrahepatic injection. US pulse length, pulse duration and US time were increased or decreased (modifications 1, 2 or 3) without changing total acoustic energy exposure (see Table 1 for details) and compared with negative or positive controls. pGL3 with Optison microbubbles was delivered by intrahepatic injection to the left liver lobe. N =6, 7, 4, 5, 4, respectively.**P≤0.001 vs negative, ^#P =0.028 vs positive.

Comparison of specific and total luciferase activities after intrahepatic and intraportal injection

Although intrahepatic direct injection is straightforward, there are several disadvantages. Firstly, plasmid delivery is limited to only one liver lobe at any given time. Secondly, results are highly variable due to uncertain injection sites. Thirdly, the needle can easily penetrate completely through the liver or cause solution leakage and therefore increase failure rates. These technical issues may explain why we had a high variability in hFIX expression between animals subjected to nominally identical treatments using the intrahepatic injection method.² To circumvent these problems, we explored a better-defined delivery route via the portal vein, which ensures pDNA/MB delivery to the sinusoids of the whole liver.

As intrahepatic injection was performed on the left liver lobe, we first compared US-mediated gene expression on the left liver lobe achieved by intraportal or intrahepatic injection. The same conditions were used as in the previous positive control treatment except for the route of delivery; and with intrahepatic injection, the US exposure was targeted only to the left liver lobe, whereas with intraportal injection, the US exposure covered the whole liver to the extent possible. To evaluate the distribution of gene expression, the left, median and right lobes (the latter including the right and caudate lobes) of the liver were harvested separately. As shown in Figure 3a, US-mediated gene expression on the left liver lobe via intraportal injection was comparable or higher to that obtained via intrahepatic injection (9.58±2.48 ×10⁴ vs 7.75±2.12 ×10⁴ RLU per mg protein, P =0.58, n =8 and 7, respectively). We also observed an average of 1.34-fold increase (range from 1.2- to 2.5-fold) of total luciferase activity from the whole liver extract compared with the left liver lobe extract when the US probe focused mostly on the left lobe (8.01±1.51 ×10⁶ vs 5.98±1.39 ×10⁶ RLU, N =8, P =0.011, Figure 3b), indicating that there are contributions from other lobes to total gene expression. In addition, fewer leakage problems occurred during intraportal injection, which led to more consistent results.

Figure 3.

Ultrasound (US)-mediated gene delivery: intrahepatic vs intraportal injection. (a) Comparison of gene expression in left lobes alone when plasmid was delivered through either intrahepatic or intraportal injection. Optison microbubbles (10%) was added to 50 μg plasmid DNA pGL3. N =6, 7, 8, 8, respectively. Conditions: for each injection route, the negative control treatment comprised of pGL3+MB injection, without US. The positive control comprised of US+MB+pGL3 treatment. (b) Total gene expression increased with intraportal injection. Total RLU calculated from gene expression of all lobes was compared to that of the left lobe alone. N =8, 8, respectively, **P =0.011.

Effect of varying US pulse length, PRF, total exposure time or injection time for enhancing gene expression via intraportal injection

With the better route of delivery (intraportal injection), we first repeated the previous experiments (Figure 2) to examine the effect of varying pulse length, PRF or total exposure time on gene expression (Table 1). As in the intrahepatic injection experiment, none of the modified US conditions further improved gene expression compared with the positive control (Figure 4a).

Figure 4.

Gene expression mediated by modified US with pGL3 delivered in 30 or 60 s via intraportal injection. The same modified US parameters as in Figure 2 were used (see Table 1 for details). (a) pGL3 was delivered through intraportal injection in 30 s. N =5, 4, 3, 5, 5, respectively. **P<0.01, *P<0.05 vs corresponding lobe of the positive control. P<0.01 for all US treated lobes vs corresponding lobes of the negative control. (b) pGL3 was delivered through intraportal injection in 60 s. N =4, 5, 5, respectively. P<0.01 for US treated lobes vs the corresponding lobes of the negative controls. P =NS for Modification 2 vs positive control in corresponding lobe with 60 s injection of panel b. *P<0.05 vs positive control with 30 s injection in panel a.

We then tested whether prolonged plasmid and MB solution injection time (60 s vs the usual 30 s used in Figure 4a) without changing US parameters (Table 1) could further enhance gene expression. Although gene expression enhancement was achieved with both the positive control and ‘Modification 2’ US treatments using 60 s injection time when data were compared with those of the negative control (P≤0.01 by lobe, Figure 4b), no enhancement was observed compared with the positive control using 30 s injection time. Taking the studies of both intrahepatic (Figure 2) or intraportal (Figure 4) injection together, it appears that increasing pulse length while decreasing PRF without changing the total acoustic energy within the ranges tested does nothing to improve US-mediated gene delivery. The same is also true with prolonged injection time.

Effect of extended US exposure pre- or post-pDNA administration through intraportal injection

In the positive control described above, we had delivered the pDNA in 30 s and applied US for 60 s, that is, there were 30 s of coincident pDNA injection and US exposure, followed by an additional 30 s of US exposure after pDNA injection ceased. To test whether US exposure prior to the delivery of pDNA could better prepare the tissue for gene delivery, or longer US exposure time is needed after pDNA delivery, we compared gene expression with an additional 60 or 90 s US exposure pre- or postinjection. Neither pre- nor postinjection additional US exposure further improved US-facilitated gene delivery compared to the positive control (Figure 5; only left lobe data shown). Thus, the enhancement of gene delivery occurred only during the period of coincident US+MB+plasmid injection, further supporting a cavitation-induced effect.

Figure 5.

Gene expression mediated by extended pre- or post-injection US exposure. Data were obtained by intraportal injection. Only left lobe data are shown. N =3, 4, 4, 3, respectively. **P<0.01 vs negative control. P =NS for both pre-, post-US treatments vs positive control.

Effect of US peak negative pressure on gene delivery efficiency via intraportal injection of pDNA

All livers treated with pDNA+Optison, using intraportal injection and an acoustic pressure of 4.3 MPa, showed macroscopic evidence of tissue damage immediately after US and when harvested 24 h after treatment. Such livers showed hemorrhage immediately after US treatment and subsequently appeared to lose their ruddy color, becoming grey in areas at the time of tissue harvest, suggesting ischemia secondary to cavitation-induced thrombosis.³⁸ On the basis of these observations, we hypothesized that 4.3 MPa P⁻ treatments were supraoptimal, that is, the use of a somewhat lower pressure might reduce tissue damage while producing higher levels of pDNA reporter gene expression because the amount of cavitation-induced cell killing would be lesser. As our prior data on gene transfection as a function of acoustic pressure: (1) had been performed with intrahepatic rather than intraportal injection; (2) had used fairly coarse increments in acoustic pressure, and so might have failed to resolve a local ‘peak’; and (3) did not produce the type of damage observed at the same pressure but with intrahepatic injection, we were motivated to examine the relationship between acoustic pressure and pDNA expression in finer detail, also extending the range of tested pressures to include a number of points below 1 MPa. Our expectation was that the livers would be more sensitive (that is, respond at lower acoustic pressures) when direct vascular access was used to administer the MBs and plasmids than when intrahepatic injection was used. However, as mentioned above, the pressure increments used in the intrahepatic injection study were too coarse to permit a direct comparison of results. In the intraportal injection study, luciferase gene expression of the whole liver showed strong dependence on acoustic peak negative pressure (P<0.0001, Figure 6). Although there was an upward trend of gene expression enhancement with acoustic pressures over the range of 0.25–2 MPa, statistically significant enhancement of luciferase gene expression relative to the negative control was observed only when P⁻≥2 MPa. A sharp inflection occurred between 2 and 3 MPa, resulting in a peak at 3 MPa, at which the mean gene expression was 85-fold greater than that in the negative controls. No further increase was observed when P⁻ was increased from 3 to 4.3 MPa, the ‘default’ pressure we had used in earlier experiments. These results indicate that it is possible to achieve maximum gene expression at more modest acoustic pressures than we had used previously; such treatments resulted in less macroscopic damage to the livers.

Figure 6.

The effect of ultrasound (US) peak negative pressure on gene delivery efficiency via intraportal injection of the pGL3+microbubble (MB) mixture. A 50 μg portion of pGL3 with 10% Optison MBs was delivered by intraportal injection in 30 s and the liver was exposed to US of 0–4.3 MPa for 60 s (30 s simultaneous). Gene expression was measured on the whole liver. The acoustic pulse length, PRF and total treatment time were fixed at 20 cycles, 50 Hz and 60 s, respectively. N =5, 4, 5, 5, 5, 4, 5, respectively. **P<0.001, *P<0.05 vs negative control, ^§§P≦0.0016 vs others. PRF, pulse repetition frequency.

Discussion

Viral gene therapy has shown efficacy in animals and some clinical trials, but adverse events have been related to enhancer-mediated mutagenesis of genomic DNA^39,40 or immunological responses to viral proteins.⁴¹ Before these permanent or long-term side effects are fully understood and resolved, the safety of using these vectors is questionable. In contrast, delivery of naked pDNA does not transport toxic or immunogenic viral protein or polymer particles in vivo. The proof of principle study of hydrodynamics-based delivery of naked plasmid^42,43 showed that long-term therapeutic effect can be achieved with an optimal, tissue-specific gene therapy vector. Nevertheless, the hydrodynamic procedure is clinically unsuitable in its current form, and injection of pDNA alone without hydrodynamic pressure was ineffective. We thus explored US-mediated gene transfer to achieve efficient gene transfer of pDNA. We previously demonstrated US-enhanced naked FIX plasmid transduction in mouse livers.² However, the efficiency is still much lower than that obtained via hydrodynamic gene delivery. To develop clinically feasible US treatments to facilitate pDNA delivery, we have performed systematic optimization studies of administration routes of the plasmids and US parameters.

In these experiments, we initially used intraheptic injection (directly injecting the liver with 300 μl pDNA solution in 30 s) to ensure that US was applied to cells exposed to plasmid solution. However, we have demonstrated in this study that plasmids can be successfully delivered into hepatocytes via portal vein injection of plasmids and MBs during US treatments. We believe that this is achieved by overcoming two major barriers. The first barrier is the blood vessel or sinusoid wall. On the basis of the reports published by others,^4,44 it is reasonable to speculate that at high acoustic pressures, the US activation of injected MBs promoted the extravasation of pDNA solution, blood cells and perhaps ‘surviving’ MBs or derivative gas bodies through the endothelial wall and into the extrahepatic space, thus increasing the opportunity for pDNA to contact hepatic parenchymal cells. This is consistent with macroscopically visible subcapsular hemorrhage that we observed in many of the treated livers after treatment. However, at present, we do not have histological evidence of extravasation. The second barrier is the plasma membrane of the targeted hepatocytes. On the basis of an abundant literature derived from in vitro studies, we further speculate that the US treatment facilitated the transfer of pDNA across cellular membranes via transient pore formation. Nonetheless, bubble dynamical behavior in the constrained in vivo environment may differ substantially from that in bulk media,⁴⁵ and we have no direct evidence for sonoporation of individual parenchymal cells. The indirect evidence is that gene transfer efficiency is significantly enhanced by US and MB treatment. In addition, from our studies using a plasmid encoding green fluorescent protein, most cells expressing green fluorescent protein were hepatocytes (data not shown). A component of the total transgene expression may involve Kupffer cells, which are known to phagocytose shelled MBs and, presumably, also to internalize pDNA. However, in the absence of insonation, transgene expression is very low in livers injected with pDNA and MBs. Kupffer cells appear to play a minor role, if any, in our model system.

The success in using intravascular delivery of pDNA in US-mediated gene transfer protocols into the liver opens the possibility of using different delivery routes of pDNA for potential clinical applications. For example, relative to open surgical exposure of the liver and portal vein as used in the present small animal studies, it would be vastly superior to use a much less invasive peripheral vein injection method or the routine procedure of jugular vein incision, catheter-guided portal vein injection method to deliver pDNA to the liver in conjunction with transcutaneous US exposure. We are presently working on this problem, but significant obstacles to efficient transfection using peripheral vein injection seem to exist, for example, substantial systemic dilution of the injectate and rapid clearance of both plasmids and MBs from the systemic blood pool.

Our results firmly established that MBs were essential in US-facilitated gene delivery, but plasmid with MBs alone did not produce any enhancement, consistent with a mechanism mediated by the induction of cavitation in the targeted tissues by US during injection of pDNA and exogenous MBs. Furthermore, it was shown by our previous study² and the present study that gene delivery efficiency by US-mediated method is strongly dependent on acoustic peak negative acoustic pressure, which also implicates the involvement of IC. As shown in Figure 6, the ‘operational’ threshold pressure—as discernable from our data—was 2 MPa. The same pressure dependence was observed for the evolution of total cavitation dose or hemolysis in vitro,^46,47 which similarly shows a modest, monotonic increase with increasing pressure over the range of ~0.5–2 MPa, and then sharply inflect. For a free bubble comparable in size to an Optison MB, it is expected that violent inertial bubble collapse will have a threshold of ~0.3–0.5 MPa at 1 MHz.⁴⁸ Thus the evidence is consistent with an IC-mediated phenomenon. However, unlike in vitro bioeffects or IC dose evolution, the data in Figure 6 suggest that there is an ‘optimum,’ or local peak response at 3 MPa. At 4.3 MPa, the luciferase expression level was about one-half that at 3 MPa, suggesting that high negative pressure may cause more liver damage and lower the number of viable transduced cells. Our previous histology results obtained from intrahepatic delivery indeed showed that more severe damages occurred at higher US-negative pressures (4.3≫3>2 Mpa). There appears to be a crossover point at which US-induced damage becomes counterproductive. Our limited data indicate that the crossover point was shifted to lower negative pressure with intraportal delivery compared to intrahepatic delivery. Nevertheless, other than the presence of exogenous MBs and acoustic pressure, acoustic conditions that were expected to further promote cavitation activity (for example, use of longer pulses, lower pulse repetition frequencies and extended exposure after injection) did not produce demonstrable improvements in gene delivery relative to our positive control conditions.

Recent reports indicate that prolonged US exposure (30 min) might further increase gene delivery efficiency in PC2 cells.^49,50 However, this observation should be interpreted conservatively, because increasing exposure from 10 to 30 min resulted in only a nominal increase by luciferase assay in vitro⁴⁹ and not at all in vivo,⁵⁰ and produced no difference in green fluorescent protein expression in vitro.⁴⁹ In our model system, there was no significant effect produced by US without MBs (Figure 1), or by extending US exposure pre- or post-pDNA injection. Even without acoustically driven Optison MB destruction, the agent clears rapidly from the circulation (mean half-time of 1.3 min (Optison package insert)). In our studies, in which the MBs were insonated using relatively high acoustic pressures directly in the organ into which the MBs were injected, bubble lifetime is expected to be still much shorter. Therefore, US exposure beyond 1–2 min is not expected to produce much additional effect.

In summary, we have made progress toward optimizing US parameters and methods for US-mediated gene delivery in an in vivo murine liver model. Most in vivo US-mediated protocols have only increased gene delivery efficiency by 2- to 10-fold. Our current protocol, which produced up to 85-fold enhancement, is therefore one of the most effective demonstrations of using US and MBs to increase in vivo gene delivery efficiency. We found that (1) MBs were essential to produce the US-facilitated gene delivery effect in our system, but MBs without US did not promote gene expression; (2) modification of pulse length, PRF or total exposure duration did not further improve gene delivery; (3) intraportal injection overcame the shortcomings of intrahepatic injection (such as, leakage) and achieved not only similar level of gene delivery in the left lobe, but also resulted in transduction of the other lobes. This increased the total level of gene expression relative to injecting and insonating only the left lobe; (4) neither US pre-exposure nor extended post-injection exposure improved gene delivery; (5) the acoustic pressures necessary to significantly increase gene delivery in our system were generally higher than those associated with diagnostic US, with an apparent peak effect at 3 MPa peak negative pressure at ~1 MHz. Furthermore, relative to our previous study,² we have established a much more reproducible methodology using delivery via intraportal route to generate a higher success rate and more consistent results. We identified a US protocol that more efficiently enhanced gene delivery while causing less tissue damage. The overall efficiency via intraportal delivery was improved up to 2- to 6-fold compared with the previous intrahepatic delivery protocol. The injection volume was also significantly reduced, a necessary consideration for later translation into larger animal studies and human clinical applications.

High levels of transgene expression were obtained in this study relative to the results reported by other investigators. Although the gene delivery efficiency achieved by our current US-mediated protocol is still two orders of magnitude lower than the hydrodynamics-based and some viral vector-mediated gene delivery systems, the current study is a stepping stone for further development of US-mediated gene delivery aiming at clinical applications for therapeutic treatment.

Materials and methods

Plasmids

pGL3 is a luciferase reporter plasmid driven by a SV40 promoter (Promega, Madison, WI, USA). The plasmid was amplified in Escherichia coli, isolated and purified using Endo-free plasmid mega or giga kit (Qiagen Inc., Valencia, CA, USA), according to the manufacturer’s manual. Plasmid concentration was measured by the absorption at 260 nm with NanoDrop ND-1000 UV-Vis Spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA).

Ultrasound

The default US condition was the best condition in our previous study with human factor IX,⁵¹ namely 1.17 MHz frequency, an acoustic peak negative pressure amplitude of 4.3 MPa, 20 cycle pulses, 50 Hz PRF and 60 s exposure (30 s with simultaneous pDNA injection and 30 s after injection).

The source transducer and impedance matching network were laboratory-made. The transducer comprised a single, round, air-backed, 1.375 inch (35 mm) diameter PZT element cemented to an aluminum lens, which provided static geometric focusing at ~7.5 cm focal distance (F number 2.1). The transducer was fitted with a truncated, water-filled polycarbonate cone whose poly-urethane water retention membrane was centered at the beam focus. This membrane was placed directly on the surface of the surgically exposed murine liver; thus the acoustic transmission path to the liver comprised entirely of ambient temperature water, with the exception of the polyurethane membrane and a small amount of US gel used to couple the transducer to the liver. The measured focal length of the transducer assembly was 75 mm, at which distance the truncated cone tip was placed. At this location, the transaxial beam profile lacked significant side lobes. The measured −6 dB beam diameter at the focus was 2.7 mm. Because the axial beam profile of the assembled water cone transducer could not be measured directly, a numerical model was used; the model indicated that the −3 dB depth of focus was ~25 mm. As there was no intervening tissue, specified acoustic pressures are not de-rated, that is, the acoustic pressures at the cone tip were the same as those applied to the liver surface.

The transducer was driven at its resonance frequency (1.17 MHz) in pulse mode by an ENI AP400B power amplifier (ENI Inc., Rochester, NY, USA), which received sine wave input signals from an Agilent 33120A signal generator, by which variable pulse parameters were set. The transducer was calibrated regularly (at ~3 month intervals) using a membrane hydrophone (Model MHA 200; NTR Systems, Seattle, WA, USA). An unavoidable systematic calibration error was spatial averaging over the 200 μm hydrophone aperture. A random calibration error was small variations in hydrophone and transducer alignment from one calibration to the next. However, the calibrations were generally reproducible to within ~1 dB. When in use, the transducer was scanned manually over the liver surface at a rate on the order of 1 cm s⁻¹; as much as possible of the liver was exposed, using a roughly circular scanning motion with a repeat time of on the order of 1–2 s. However, because the surface of the truncated transducer cone was relatively large in comparison with the murine liver, much of the liver was visually obscured by the transducer tip during scanning, making precise visual guidance difficult. We assume that reasonably uniform exposure of different areas of the liver was achieved stochastically. Observations of trypan blue dye injected into the portal veins of a few mice dedicated to exploring the speed at which the dye distributed visibly in the livers indicated that the livers were filled with the injected dye in about 2 s. Therefore, while it is not certain that injected MBs were not destroyed acoustically near the point of entry into the liver, the scanning protocol was, to first order, of a timescale consistent with the rate at which materials injected into the portal vein might be expected to be distributed throughout the liver. Because the murine liver is quite thin and the transducer’s depth of focus was relatively large, underlying tissues were exposed to US to greater or lesser extents, depending on factors such as whether air had become trapped below the liver during manipulations, a factor over which we had no control in the small animal model. That underlying structures were sometimes damaged by the acoustic exposures was evidenced by intestinal hemorrhage in some animals.

After each use on an animal, the continued functioning of the transducer was verified by a qualitative ‘acoustic fountain’ test. Modeling studies indicated that no significant tissue heating effects were likely.

US microbubbles

Optison (Amersham; GE Health Care, Princeton, NJ, USA) is a MB US contrast agent comprised of octafluoropropane gas bodies surrounded by albumin shells. A 10% (v/v) concentration of Optison or equivalent PBS was mixed with the pDNA solutions immediately prior to injection. The total Optison injected was either 0 or 30 μl per animal. No special measures were taken to degas the pDNA solutions.

Animal experiments

Six- to eight-week-old C57/BL6 male mice were purchased from the Jackson Lab (Bar Harbor, ME, USA) and maintained at a specific pathogen-free (SPF) vivarium. Mice were kept in the vivarium for at least 3 days before an experiment. Animals were kept according to the guidelines for animal care of both National Institutes of Health and the University of Washington.

During the gene delivery procedure, mice were anesthetized by intraperitoneal injection of Avertin (250 mg/kg; Tribromoethanol, Sigma-Aldrich, St Louis, MO, USA). The liver or the portal vein was exposed by midline section. A 50 μg portion of pGL3 plasmids was prepared in 300 μl PBS containing 5% glucose. An ‘extra’ volume of 100 μl was prepared to compensate for the loss to the syringe. A 10% (v/v) concentration of Optison MBs or equal volume PBS (no MB control) was added to the plasmid just before injection. The injection was carried out in 30 or 60 s with a 30 G needle inserted directly into the left liver lobe of mice or through the portal vein. US was applied simultaneously with or without extended pre- or postinjection exposure over the left liver lobe or the whole liver, but with the left liver lobe receiving much of the exposure. Bleeding from the injection site was stopped with cotton swabs soaked with thrombin solution (100 U ml⁻¹; Sigma-Aldrich). Mice recovered from anesthesia in about 2 h, and had recovered well from the surgery on the second day. In some experiments, the whole liver was harvested 24 h after injection, frozen immediately on dry ice and stored at −80°C until luciferase assay. In other experiments, individual liver lobes were harvested separately as left, median and right lobes; the latter included the right and the caudate lobes for convenience. Post-harvest processing was as described for whole livers.

As reduction of the injection volume of the plasmid solution will facilitate eventual translation into human clinical trials, we conducted experiments in which we compared the use of either 500 μl (used in our earlier study²) or 300 μl injection volume of the plasmid solution containing 50 μg of pGL3. With either the intrahepatic or the intraportal injection pathway, the control background level of gene expression was reduced when the volume was reduced from 500 to 300 μl, probably due to less hydrodynamic effect of plasmid injection. However, the enhanced gene expression from US-mediated gene delivery remained at the same level (data not shown). We therefore chose to use 300 μl injection volume for this study.

Luciferase assay

Luciferase gene expression was measured by a firefly luciferase assay system (Promega, Madison, WI, USA). Briefly, the frozen whole liver or each of the liver lobes were weighed, homogenized in 3 ml per g liver of 1 × reporter lysis buffer (RLB) and frozen on dry ice. The homogenate was then thawed in a 37 °C water bath to completely release luciferase from lysed cells. The liver lysate was then vortexed and centrifuged for 2 min at 18 000 g at room temperature and the supernatant was collected. The firefly luciferase activity was measured for 10 s after 100 μl luciferase substrate was added to 20 μl supernatant by an autoinjector and shaken for 2 s using a luminometer (Victor 3; Perkin Elmer, Wellesley, MA, USA). Opaque wall 96-well plates were used to prevent well-to-well crosstalk. Luciferase activity was normalized by the protein content of the same supernatant as measured by the Bradford method (Bio-Rad, Hercules, CA, USA), with bovine serum albumin as the standard.

Statistical analysis

Data are presented as means±s.e.m. Two sample means were compared by Student’s t-tests. Multiple group studies were carried out with one-way ANOVA (analysis of variance), and when there were differences, least significant difference (LSD) multiple comparisons were made.