Effects of Microbubble Size on Ultrasound-Mediated Gene Transfection in Auditory Cells

Gene therapy for sensorineural hearing loss has recently been used to insert genes encoding functional proteins to preserve, protect, or even regenerate hair cells in the inner ear. Our previous study demonstrated a microbubble- (MB-)facilitated ultrasound (US) technique for delivering therapeutic medication to the inner ear. The present study investigated whether MB-US techniques help to enhance the efficiency of gene transfection by means of cationic liposomes on HEI-OC1 auditory cells and whether MBs of different sizes affect such efficiency. Our results demonstrated that the size of MBs was proportional to the concentration of albumin or dextrose. At a constant US power density, using 0.66, 1.32, and 2.83 μm albumin-shelled MBs increased the transfection rate as compared to the control by 30.6%, 54.1%, and 84.7%, respectively; likewise, using 1.39, 2.12, and 3.47 μm albumin-dextrose-shelled MBs increased the transfection rates by 15.9%, 34.3%, and 82.7%, respectively. The results indicate that MB-US is an effective technique to facilitate gene transfer on auditory cells in vitro. Such size-dependent MB oscillation behavior in the presence of US plays a role in enhancing gene transfer, and by manipulating the concentration of albumin or dextrose, MBs of different sizes can be produced.


Introduction
Mammalian cochlear hair cells do not regenerate naturally, and hence damage to them can result in permanent hearing loss. This has prompted considerable attention to be paid to the regeneration of inner ear hair cells, which has led to demonstrations of the feasibility of gene therapy and stem cell transplantation for inner ear disease [1,2].
Gene therapeutic approaches to several forms of hearing disorders have been experimentally investigated using either viral [3,4] or nonviral [5] vectors. Delivery methods mainly utilize two surgical routes: direct injection or diffusion through the round window membrane (RWM) [5] or intracochlear infusion through a cochleostomy or canalostomy on the semicircular canal [4,6,7]. Although these deliveries achieve different transfection efficiency, the associated surgical trauma, inflammation, and possible damage in hearing deterioration suggest that a less invasive delivery method for future clinical application is imperative. Our previous study demonstrated the practical application of combining microbubbles (MBs) with ultrasound (US) to increase the RWM permeability for facilitating drug or medication delivery to the inner ear in an animal model [8]. Before extending this minimal invasive technique for application of inner ear gene therapy in vivo (unpublished data), the present study focuses on evaluating the impact of the MB-US technique for improving the efficacy of gene transfection on auditory cells-the HEI-OC1 cells-and the difference by manipulating different-sized MBs in transfection in vitro.
US-targeted MB destruction is a safe and noninvasive gene transfection technology. MBs are known to be contrast agents for ultrasonic imaging (their primary medical use). Recently, additional medical applications for MBs have focused on utilizing the interaction of MBs and US to produce microstreaming, known as cavitation. MBs are well suited as carriers of genes and drugs. Compared with viral vectors, MBs have greater capacity and can carry antisense oligonucleotides, DNA fragments, and even the entire chromosome [9]. The energy in the US field destroys the MBs, with cavitation or other mechanical effects increasing both the cell membrane permeability and the dimensions of the intercellular space between endothelial cells; these effects can lead to the drug or gene more easily reaching the tissue or cells via the rupture site and the widened interstitial space [10][11][12][13].
Microbubbles in various sizes have different effects on resonance frequency, expansion ratio, pressure thresholds for inertial cavitation and fragmentation, translational velocity, and lifetime of stable oscillation of MBs [14][15][16][17]. Although previous study has revealed an association between the focused US-induced blood brain barrier (BBB) opening and MB size [18], most of these investigations focused on lipidshelled MBs. In addition, dextrose was found to be able to change the characteristics of the MB shell, and increased dextrose concentration could produce MBs with thinner shells, good stability, and a wider range of resonance frequency, thus enhancing the efficiency of gene transfection [19]. However, the effects of albumin and dextrose concentrations on the MB size dependence of US-induced gene transfection remain unclear. Moreover, cell membranes in different cell types have different biophysical characteristics in terms of transfection and physical stability [20].
The present study aimed to evaluate the impact of the MB-US technique on gene transfection efficiency in auditory cells and to explore the ratio of glucose and albumin concentrations for producing albumin-shelled MBs in different sizes to improve the transfection rate in vitro.

Preparation of Albumin-Shelled
MBs. Albumin MBs were prepared according to the procedure used in our previous studies [8,21,22]. Figure 1 illustrates the TEM image of albumin MBs. The shell thickness of the albumin MBs was about 30-60 nm [23]. For albumin/dextrose MB preparation, dextrose (D(+)-glucose; Acros Organics, Fair Lawn, NJ, USA) was purchased to prepare stock solutions of 5%, 10%, 15%, 20%, 30%, 40%, and 45% (weight (w)/volume (v)) dextrose in physiological saline (pH 7.4, 0.9% sodium chloride). Human serum albumin (HSA) was purchased as a sterile 20% solution (Octapharma, Vienna, Austria), which was diluted with physiological saline to make stock solutions containing 0.66%, 1.32%, 2%, 3.5%, or 5% (w/v) HSA. Briefly, albumin/dextrose MBs were generated by sonication in 10 mL of solution by mixing the albumin and dextrose (Table 1)     The concentration change was calculated as the percentage decrease in the optical density as follows: Δ Optical density at 530 nm (%) where OD pre and OD post are the optical densities of the MBs before and after US treatment, respectively.  Shortly before transfection, the serum-containing medium was removed from the cell culture plate and replaced with 4 × 10 7 /200 L MBs. As shown in Figure 2, the plate was sonicated by US from the bottom to the top for 2 min before being rinsed with PBS three times. The MBs were then replaced by 400 L of 10% serum-containing medium. The pEGFP-C1-Lipofectamine complex was added, and the sample was placed in the incubator for 24 h. The mode of US was set as follows: voltage of 30 V, transducer with a center frequency of 3.185 MHz, duty cycle of 50%, burst rate of 2 Hz, acoustic intensity of 0.46 W/cm 2 , and 2 min duration. The medium was exchanged for DMEM, and the cells were incubated for another 24 h. The transfection efficiency was determined with the aid of fluorescence microscopy (CKX41; Olympus, Tokyo, Japan). Cell lysate buffer (radioimmunoprecipitation assay, RIPA) was then added for cell lysis. The fluorescence intensity of green fluorescent protein was measured using a hybrid multimode microplate reader. The normalized fluorescence increase was calculated as follows:

Plasmid
where FI US+MB and FI MB are the fluorescence intensities in groups UM and M, respectively.

Statistical Analysis.
The obtained data were analyzed statistically using Student's -test and results are expressed as means ± standard deviation (SD). Differences were considered significant at < 0.05.

Production of Albumin-Dextrose-Shelled MBs of Different
Sizes. The concentration of the produced MBs was not proportional to the albumin concentration (Figure 3(a)) or the dextrose concentration (Figure 3(b)), but their size was influenced by changing the composition ratio of albumin alone or of dextrose in the albumin (Figures 3(c) and 3(d)).
When the compositions of dextrose in the 1.32% albumin saline solution were 5%, 10%, 15%, 20%, 30%, 40%, and 45%, the produced MBs had diameters of  (Figure 4(b)). Results of the destruction efficiency of MBs of different sizes suggest that MBs smaller than 2 m were easier to destroy for a constant US power density ( Table 2).

Transfection Efficiency for MBs of Different Sizes.
The transfection efficiency of HEI-OC1 cells with plasmid DNA by using a combination of US and MBs was investigated ( Figure 5). The fluorescence intensities (in arbitrary units) in groups of P, M, U, and UM were 796.3 ± 211.9, 774.3 ± 125.1, 978.3±101.0, and 1270.3±146.7, respectively; the transfection efficiency was highest in group UM and was significantly higher in groups UM ( < 0.01) and U ( < 0.05) than in control group P. There was no significant difference between groups P and M, indicating that MBs would not influence the transfection efficiency in the short time that they soak up the cells without the US treatment. The transfection efficiency when using albumin MBs of three different sizes was also evaluated ( Figure 6). The normalized percentage increases in fluorescence for US and plasmid combined with 0.66, 1.32, and 2.83 m MBs were 30.58 ± 12.53%, 54.10 ± 33.95%, and 84.74 ± 29.37%, respectively; the transfection efficiency showed an MB sizedependent tendency and was highest for US combined with 2.83 m MBs and plasmid. In comparison, the transfection efficiency of 2.83 m MBs was significantly higher than that of 0.66 m MBs ( < 0.05), whereas there was no significant difference between 0.66 and 1.32 m MBs. Taken together, the results indicate that the transfection efficiency could be improved by 54.16% when the size of the albumin MBs is increased 3.3-fold.
Transfection results for HEI-OC1 cells with albumin/dextrose MBs of three different sizes are shown in Figure 7       and plasmid combined with 1.39, 2.12, and 3.47 m MBs were 15.86 ± 13.56%, 34.34 ± 19.98%, and 82.67 ± 24.47%, respectively. The transfection efficiency differed significantly between these three groups, with improving efficiency as the size of the albumin/dextrose MBs increased.

Cell Viability Analysis.
To investigate whether the application of US and MBs would cause cell damage, cell viability following MBs and US treatments with different power densities was tested. There were no statistically significant differences among the various groups ( > 0.05) by quantifying the 24 and 48 h cell survival (Figure 8), confirming that using US at power densities from 0.2 to 0.84 W/cm 2 did not affect the survival of HEI-OC1 cells even when combined with MBs.

Discussion
Although previous investigators have depicted protocols for producing uniformly sized MBs [24], methods that often involve several preparation procedures and parameter settings, in this study we produced different-sized MBs through a simple laboratory method by adjusting the concentration of albumin and dextrose. We found that the MB would increase in size as the proportion of albumin or dextrose in the composition increased. In the albumin and dextrose mixed condition, a larger MB was produced under two conditions: (1) a higher dextrose concentration and a lower albumin concentration and (2) a lower dextrose concentration (or without adding glucose) and a higher albumin concentration. This report describes the first successful demonstration of  the correlation of albumin, dextrose, and albumin-dextrose compositions in producing MBs of different sizes. The efficiency of gene transfer was found to increase with increasing MB diameter. Although the mechanism for MB size-dependent sonoporation increase is not yet known, the duration of the MB's interaction with the vessel wall has been found to increase with increasing MB diameter [14]. Using focused US in conjunction with MBs, the exchange rate between the blood plasma and the brain tissue has been shown to be proportional to the MB size [26]. The opening volume of the BBB in the mouse brain using small bubbles was found to be significantly lower than for the larger MBs [27], which is in agreement with our findings that gene transfer efficiency is dependent on the MB size.
MBs produced in this study had sizes ranging from 0.5 to 3.5 m. The efficiency of MB destruction appears to be conversely proportional to the MB size; larger MBs were more resistant to destruction, whereas in gene transfer, larger MBs exhibited more enhanced transfection efficiency because MB behavior depends not only on the US parameters but also on the MB size and physicochemical properties. Various MB dynamic activities generated by US exposure include local fluid microstreaming, shear stress, and high-speed fluid microjet, leading to different sonoporation effects [28][29][30]. In this experiment, MBs smaller than 2 m were easier to destroy for a constant US power density, implying that oscillation of small-size MBs may become unstable under US activation and may fragment prior to interacting with cells. Such effects have also been noted in previous study [27]. In contrast, stable cavitation within the lifetime of the larger MB may play a role in increasing sonoporation of gene transfer on cultivated HEI-OC1 cells. In addition, an MB's motion and linear oscillation when in close contact with the cell membrane can cause local deformation and transient porosity in the cell membrane without rupturing it [31]. Therefore, the efficiency of MB destruction may not faithfully reflect the efficacy of gene transfer, as shown in our data.
Our experiments also demonstrated that gene transfer assisted by 2.83 m albumin MBs-mediated US was able to achieve a gene transfection efficiency of around 85%, which was significantly higher than those in the other treatment groups and was shown without adverse effect on cellular viability. This transfection efficiency is higher than the previously reported value of 70% when using 2.89 or 2.98 m lipid MBs [32], suggesting that different MB shell compositions, such as surfactants, lipids, proteins, polymers, or a combination of these materials, may consist of different physicochemical properties for a wide variety of biomedical applications. For gene transfer applications, albumin-based MBs can potentially incorporate large amounts of plasmid DNA within the thick protein shell, whereas the drugcarrying capacity of lipid-based MBs is relatively low [33]. Due to covalent cross-linking, albumin MBs form relatively rigid shells by disulfide bridging of proteins as compared with other compositions of MBs [34]; this stabilizes the shell and prevents gas dissolution, resulting in a greater yield and increasing their acoustic durability.
Compared to conventional liposome-mediated transfection, our results demonstrated that MB-US-mediated gene transfer appears to be advantageous for increasing the cell membrane permeability of auditory cells and allowing plasmid DNA to enter the cells. Additionally, the proximity between the MB, nucleic acid, and target cells would be expected to enhance cell "poration" effects, thus improving nucleic acid transfer to the target cell [35][36][37]. To facilitate the efficiency of gene transfer, some previous studies proposed using cationic MBs to augment interactions between MBs and cells and reduce the separation between MBs and plasmid DNA [32,37]. Cationic MBs (zeta potential of approximately 4-5 mV relative to the surrounding environment) were found to be attracted to negatively charged plasmid, while the neutral MBs (−0.7 mV) did not attract plasmid. Those studies demonstrated the influence of MB surface modifications on their interaction with plasmid DNA and target cells and the functional consequences of those interactions in terms of US-mediated gene transfer [37]. In the present study, the albumin-shelled MBs were negatively charged (−21.4 mV) [19,38], with this negative charge increasing the binding of positively charged liposomes to the cells and influencing the transfection efficiency.

Conclusion
This study explored the impact of MB size-dependent gene transfer in vivo. Larger MBs exhibited an increased resistance to ultrasonic destruction and enhanced the transfection efficiency of auditory hair cells for a constant US power density. Production of MBs in different sizes can be manipulated by adjusting the concentration of albumin or dextrose alone or the combined albumin and dextrose mixture. This study provides a promising strategy for auditory cell gene transfer in vivo by using MB application in the presence of US. The recommended safe power densities would be from 0.2 to 0.84 W/cm 2 . Further research is needed to clarify the feasibility of using MB-US as a tool to improve inner ear gene transfer in vivo.