Markedly Enhanced Skeletal Muscle Transfection Achieved by the Ultrasound-Targeted Delivery of Non-Viral Gene Nanocarriers with Microbubbles

Microbubble Fabrication

To prepare albumin MBs, a 1% solution of serum albumin in normal saline was placed in a flask with a blanket atmosphere of octafluoropropane gas above the aqueous phase. The solution was briefly sonicated (20 KHz, 30 sec) with an ultrasound disintegrator (XL2020, Misonix, Farmingdale, NY) equipped with an extended ½″ titanium probe. This formulation is somewhat similar to Optison (GE Healthcare), with a concentration range of 5.0–8.0×10⁸ MBs/mL.

Microbubble Sorting

MB size sorting was performed by differential centrifugal flotation in 3mL crimped vials, similar to a method previously described by Feshitan et al. [28]. The time required for various sizes of MBs to rise the distance of a 3mL vial at ~200 × g centrifugation was estimated. MBs were centrifuged for 40 seconds, and the eluent was isolated from the MB cake to obtain “small” MBs. This process was repeated two more times to separate out more small MBs, with subsequent MB centrifugations lasting 30 and 20 seconds. Next all “small” MBs were centrifuged once more for 60 seconds to remove the largest of the “small” MBs. Afterward all the “small” MBs were centrifuged for 6 min, after which the eluent was discarded, and only and the cake was isolated to remove the smallest of the “small” microbubbles. MBs were sized and counted with a Coulter Counter (Multisizer 3, Beckman Coulter, Fullerton, CA) with an orifice size of 50 μm and a lower size measurement limit of 1 μm.

Infusion Simulations to Test for MB Loss

Approximately 6.25 ×10⁶ MBs/mL were used for infusion simulations. A tail vein catheter, consisting of a 27 gauge ½″ needle connected to both 12 inches of PE20 polyethylene tubing and a second ⅝″ needle (25 gauge) was attached to a 1mL (BD; Franklin Lakes, NJ) syringe. Infusion (13.33 μl/min) was provided by a high precision Harvard Apparatus (Holliston, MA) syringe pump. The MB concentration and size distribution before and after infusion though the catheter were measured with a Coulter Counter (Multisizer 3, Beckman Coulter, Fullerton, CA) with a 50 μm aperture and a lower size limit of 1 μm.

Surface Modified Polyethyleneimine Nanoparticle Fabrication

To fabricate poly(ethylene glycol) (PEG) tethered poly(ethylenimine) (PEI) nanocomplexes, mPEG5k-N-hydroxysuccinimide (NHS) (Sigma-Aldrich, St. Louis, MO) was conjugated to 25 kDa branched PEI (Sigma) at a high PEG:PEI ratio (~50) to form PEGylated polymer. Gene carriers were formed upon the drop-wise addition of an aqueous luciferase (LUC) reporter plasmid (pBaL; Aldevron, Fargo, ND) solution to an aqueous polymer solution at the nitrogen to phosphate ratio (N/P) of 6. Specifically, 10 volumes of 0.2 mg/ml plasmid solution were added to 1 volume of polymer solution. After 30 min incubation at room temperature, the gene carrier solution was syringe-filtered to remove aggregates and subsequently ultracentrifuged multiple times with nuclease free water to remove free PEI and to concentrate the carrier solution. Following complex formation, the particle size and zeta potential were measured using a Zetasizer 3000HS (Malvern Instruments Inc., Southborough, MA).

Image Cytometry Analysis of PEG/PEI Nanocomplex-MB Interactions

For cytometry imaging, MBs were incubated with Cy3 (Mirus Bio LLC., Madison, WI) tagged PEG/PEI nanocomplexesor MBs were imaged alone. Imaging was performed on an ImageStream (Amnis Corporation, Seattle, WA) utilizing the 550nm wavelength laser. MB and PEG/PEI nanocomplex concentrations were matched to in vivo studies. Data were collected and then analyzed with IDEAS 3.0 software (Amnis Corporation). Focused single MBs were gated and potential colocalization with fluorescent PEG/PEI was assessed.

PEG/PEI Nanocomplex Delivery to Gracilis Muscle

On the day of treatment, C57BL/6 mice were anesthetized with an intraperitoneal (i.p.) combination injection of ketamine hydrochloride (1.56 mL/kg body weight), xylazine (0.52 mL/kg body weight) and sterilized water (3.12 mL/kg body weight). For US+MB gene delivery studies, the tail vein was cannulated for intravenous (i.v.) administration of a co-injection of “naked” LUC plasmids (40μg DNA) or LUC plasmids condensed in PEG/PEI nanocomplexes (40μg DNA) and either small or unsorted MBs (4× 10⁴ MBs/gram body weight). A water based US gel (Aquasonic 100; Parker Laboratories, Inc., Fairfield, NJ) was applied to the skin above the hindlimb, and a 0.75″ diameter 1 MHz unfocused transducer (A314S; Panametrics, Waltham, MA) was ultrasonically coupled to the skin. Injection duration and rate were controlled by an infusion pump (Harvard Apparatus PHD 2000; Harvard Apparatus, Holliston, MA). The dead space of the tail vein catheter was cleared prior to initiating US pulsing. Each pulse consisted of 5 sets of 300 consecutive 1 MHz sinusoids of 1V or 0.4V peak-to-peak amplitude from a waveform generator (AFG-310; Tektronix, Inc., Beaverton, OR), separated by 100ms and applied every 10 sec for 12 mins. The ultrasound pulsing protocol is graphically illustrated in Figure 1. The waveform signal was amplified by a 55 dB RF power amplifier (ENI 3100LA; Electronic Navigation Industries, Richardson, TX). Peak negative pressure was measured with a needle hydrophone (Specialty Engineering Associates, Model PVDF-Z44–0400) at a distance from the transducer face that corresponded to the distance between the transducer face and the deep adductor muscles during in vivo experiments. At this distance, peak-negative pressures were 0.8MPa and 0.6MPa for 1.0V and 0.4V peak-to-peak amplitudes, respectively. For IM gene delivery, 40μg DNA (1mg DNA/mL saline) or 6μg (66 μg DNA/mL saline) condensed in PEG/PEI nanocomplexes was injected directly into the gracilis muscle at 5 evenly spaced sites spanning the length and width of the proximal adductor muscle, using a 24-gauge needle.

Bioluminescence imaging

Luciferase activity induced by luciferase gene expression was determined with an in vivo real-time bioluminescence imaging system (IVIS 100; Caliper Life Sciences, Alameda, CA, USA). Prior to imaging, D-luciferin (150 mg/kg) was i.p. injected into mice. Light emission was measured serially for 8 mins, starting 6 mins after D-luciferin injection, using Living Image Software. The resulting pseudocolor images represent the spatial distribution of photon counts within the animal. Bioluminescence imaging was performed at 3, 5, 7 and 10 days after treatment.

Statistical analysis

Data in Figures 2–5 were analyzed by Two-Way analysis of variance (ANOVA) followed by pairwise comparisons with the Holm-Sidak method. Significance was assessed at P<0.05. All statistical analyses were performed using SigmaStat 3.5 (Systat Software, San Jose, Ca).

Characterization of PEG/PEI Nanocomplexes and Microbubbles

PEG/PEI nanocomplexes had a mean diameter of 42.9 ± 5.6 nm and a zeta potential of +7.6 ± 2.3 mV. Image cytometry was performed to determine whether PEG/PEI nanocomplexes adhered to MBs when combined in solution before being co-injected into mice (Figure 2). In the scatterplot (Figure 2A), each point represents one or more particles passing through the cytometer. MBs are much larger than PEG/PEI nanocomplexes, but they are not fluorescent; therefore, they exhibit high brightfield intensity and low fluorescence intensity. In contrast, PEG/PEI nanocomplexes exhibit low brightfield intensity and high fluorescence intensity due to the presence of Cy3. Brightfield and fluorescence images were acquired for each data point in Figure 2A, and representative images, with brightfield signals false-colored to magenta and fluorescence signals false-colored to orange, are shown in Figures 2B and 2C. Two MB populations, P1 and P2, were selected from the data based on low and high fluorescence intensity, respectively. The low fluorescence-intensity P1 population (87% of total) consisted only of MBs (Figure 2B). The P2 population (11% of total) contained points where both a MB and one or more PEG/PEI nanocomplexes were present, indicating the potential for adhesion (Figure 2C). However, of the images from the P2 population, only a small percentage (~10%) exhibited potential co-localization (arrow in Figure 2C). Therefore, we estimate that fewer than ~1% of all MBs may have had one or more PEG/PEI nanocomplexes attached to their shells.

For some studies, MBs were separated into a small diameter sub-population by differential centrifugation. To account for any changes in the MB size and/or concentration caused by flotation within the syringe and/or infusion through the syringe and catheter, these parameters were measured via Coulter counter after the MBs had passed through the injection system. MB diameter distributions for the small and unsorted MBs are shown in Figure 2D. Mean MB diameters were relatively similar - 1.5 μm and 1.3μm for the unsorted and small MB populations, respectively. Nonetheless, the distributions were considerably different. For example, within the small MB population, virtually no MBs were larger than 1.5 μm, yet >35% of the unsorted MBs were >1.5 μm in diameter.

US-Targeted Skeletal Muscle Transfection with Unsorted MBs Is Markedly Enhanced by PEG/PEI Nanocomplexes

Bioluminescence imaging (IVIS^™ Imaging System, Caliper Life Sciences, CA) was used to analyze transgene expression in skeletal muscle at 3, 5, 7, and 10 days after US-MB mediated gene delivery. Representative IVIS images acquired 7 days after insonation of unsorted MBs co-administered with “naked” plasmid or PEG/PEI nanocomplexes are shown in Figure 3A. Quantitative results are displayed in Figure 3B. The effect of US application on transfection is illustrated by comparing the US treated hindlimb to the contralateral “no-US” control. For the PEG/PEI nanocomplex group insonated at a PNP of 0.6MPa, signal intensity was 180-fold greater on the US-treated side when compared to the contralateral “no US” side (Figure 3A, Left). The effect of gene carrier on US-MB mediated transfection is illustrated by comparing the unsorted MBs + PEG/PEI nanocomplex groups to the unsorted MBs + plasmid groups (Figure 3B). Here, PEG/PEI nanocomplexes caused a significant increase in gene expression overall, with pairwise comparisons revealing 71-, 63-, and 71- fold increases in luciferase activity in PEG/PEI nanocomplex treated muscles over muscles treated with naked plasmid DNA at, respectively, 5, 7, and 10 days after insonation at 0.6MPa (Figure 3B).

US-MB-Targeted Skeletal Muscle Transfection Is Dependent on Peak Negative Pressure

PEG/PEI nanocomplex and MB concentrations were identical within unsorted and small MB groups. Therefore, we can determine whether PNPs influenced transfection by comparing treatments at 0.8MPa and 0.6MPa within size-sorted MB groups (Figure 4). Here, significant overall increases in transfection using PEG/PEI nanocomplexes were observed for both small and unsorted MBs at 0.6MPa, as compared 0.8MPa (Figure 4A & 4B). Of particular note, at 7 days after US application, gene expression was 20-fold higher following treatment at 0.6MPa with small MBs as compared to treatment at 0.8MPa.

US-MB-Targeted PEG/PEI Nanocomplex Delivery Yields Substantially Higher Gene Expression When Compared to Direct Intramuscular Injection

To compare the efficacy of US-MB-targeted PEG/PEI nanocomplex gene delivery to direct intramuscular (IM) injection, 2 PEG/PEI nanocomplex concentrations (100% and 15% of the I.V. dose) were IM injected. Based on previous studies targeting NP delivery to skeletal muscle with US and MBs [29], we estimated that, in an absolute best case scenario, 15% of the injected PEG/PEI nanocomplexes would be delivered to the hindlimb. Representative IVIS images acquired 7 days after IM injection are shown on the right in Figure 5A. In comparison, IVIS images after treatment at 0.6 MPa US with small and unsorted MBs are shown in the left and middle panels, respectively, of Figure 5A. IVIS scans were quantitatively analyzed to determine the transgene expression at 5, 7, and 10 days after treatment (Figure 5B). US treatment with intravenously administered small MBs at a PNP of 0.6MPa resulted in 10- and 130-fold increases in transfection compared with high and low dose IM injection, respectively, at day 7 after treatment (Figure 5B).

PEG/PEI Nanocomplexes Do Not Appear to Damage Muscle

One week after PEG/PEI nanocomplex delivery using either US and MBs (unsorted MBs, 0.6MPa) or direct IM injection, cross-sections of gracilis muscle were H&E stained and observed microscopically. As shown in Figure 6, PEG/PEI nanocomplex-treated muscles appeared normal, with no obvious signs of muscle fiber necrosis, muscle regeneration, and/or leukocyte infiltration into the tissue (i.eno tissue inflammation).

Enhanced Transfection by US-MB Delivery of PEG/PEI Nanocomplexes

Sonoporation via US-MB interactions increases gene delivery; however, genetic material is often degraded in route to the target and/or does not efficiently overcome a variety of biological barriers to target cell nuclei. In this study, we showed that condensing DNA into small PEG/PEI nanoparticles prior to intravenous infusion significantly increased US-MB mediated gene transfer when compared to “naked” plasmid DNA (Figure 3). We hypothesize that increased transfection was achieved because: 1) PEI protects DNA against degradation and facilitates DNA transport to the nucleus, and 2) the tethering of PEG to the PEI/DNA nanocomplex increases circulation time [31] and diffusion in tissue following US-MB mediated delivery. Consistent with our findings, other investigators have shown that the insonation of MBs co-administered with PEI and DNA yields greater transgene expression levels as compared to acoustic activation of MBs in combination with “naked” plasmid [32], and a recent study has shown that coupling a reporter plasmid DNA to PEI MBs may be used to transfect ultrasound-targeted tumors implanted in the mouse kidney [15]. However, in contrast to our findings, it has also been reported that US only improves gene transfer if PEGylated lipoplexes are bound to MBs [33, 34]. In our study, PEG/PEI nanocomplexes did not interact with albumin MBs (Figure 2). The discrepancies between these studies could be due to any number of factors, including secondary carrier formulation (i.e. core material, charge, size, PEG chain length, PEG coating density), MB and secondary carrier concentrations, internalization mechanisms, acoustic parameters and/or fundamental differences between in vitro and in vivo conditions. Without surface area restrictions imposed by the MBs, ~10⁷ PEG/PEI nanocomplexes per MB were administered per animal in our study, and we estimate the number of secondary carriers per MB was at least four orders of magnitude greater in our study compared to in vitro [33, 34] studies. Furthermore, prolonged nanocomplex persistence in circulation in vivo due to PEGylation likely served to increase particle delivery to the region of insonation in our study.

In certain formulations, the excess of positive charge of PEI nanocomplexes can cause cytotoxicity, aggregation with serum proteins, and non-specific transfection [35]. Here, we observed no local toxic effects of PEI in the muscle 7 days after the IM injection of PEG/PEI nanocomplexes or the acoustic activation of MBs co-administered with PEG/PEI (Figure 7). Toxicity was likely reduced by tethering PEG to nanocomplexes [36]. Additionally, PEG shielding decreases cell surface interactions [31], which may be responsible for the limited gene expression observed in regions that were not targeted with ultrasound.

Influence of Peak-Negative Pressure on US-MB-Targeted Gene Expression

In this study, we varied both mean MB diameter and PNP in order to investigate their influence on gene expression. As shown in Figure 4, insonation at lower PNPs clearly resulted in increased transfection, especially with small diameter MBs. The mechanism(s) of enhanced transfection with low PNPs is not completely understood; however, the most likely explanation is that sonoporation is favored with small MBs and reduced PNPs, resulting in enhanced nanocomplex delivery. PNP is a key factor for predicting MB behavior in an acoustic field and may be a significant predictor of nanocomplex delivery. For PNPs below the cavitation threshold, MBs oscillate. For PNPs at or above the cavitation threshold, MBs are destroyed though inertial cavitation. The minimum shell rupture threshold for unconstrained unsorted (2.0 – 4.5μm) albumin shelled MBs insonated at 0.9–1.1 MHz has been determined to be between 0.13–0.53MPa [37, 38]. While we did not characterize MB rupture here, MB-microvessel interactions [39, 40], MB size distribution, and tissue attenuation of US power may have increased the cavitation threshold in this study. Below the cavitation threshold and/or prior to cavitation, MB oscillation reversibly opens pores in cell membranes, a process referred to as sonoporation. Sonoporation permits the transfer of nanocomplexes to the endothelium. Conversely, at or above the cavitation threshold, MB cavitation creates mechanical effects and chemical agents that may contribute to irreversible capillary permeabilization [41]. Following capillary permeabilization, convective forces enhance NP delivery to the interstitium [42, 43]. However, once destroyed, MBs are no longer present to contribute to endothelial sonoporation and/or capillary wall permeabilization. In this study, it is likely that insonation at both 0.6MPa and 0.8MPa resulted in simultaneous MB oscillation and cavitation. While both sonoporation and irreversible capillary wall permeabilization may have resulted, it is possible that insonation at a lower PNP (0.6MPa) increased the duration of MB oscillation and the relative extent of sonoporation. As compared to immediate MB destruction at high PNPs, enhanced sonoporation through prolonged MB oscillation may be more effective for US-MB mediated gene transfer [44].

Ultrasound-Microubble-Targeted PEG/PEI Nanocomplex Delivery Compared to Intramuscular Injection

One particularly significant result was that delivering PEG/PEI nanocomplexes with 0.6 MPa US and small MBs led to substantially enhanced transfection when compared to the IM injection of either 40 μg of PEG/PEI (i.e. the total amount of PEG/PEI administered intravenously in US-MB-mediated delivery experiments) or 6 μg of PEG/PEI (i.e. an amount estimated from a previous study [29] as a best case scenario for PEG/PEI delivery with US and MBs) (Figure 5). These results suggest that, in addition to facilitating PEG/PEI nanocomplex delivery to muscle, US-MB interactions may also enhance transfection through other unknown mechanisms, such as improved PEG/PEI entry into cells via membrane pore formation and/or the activation of cellular uptake. Alternatively, US-MB interactions could reduce intracellular degradation of PEG/PEI through the degradative ensolysosomal route and/or amplify PEG/PEI trafficking to the nucleus [potentially transiently opening the nuclear envelope]. Given the apparent significance of these secondary effects, future study in this area is warranted.

In apparent contrast to our findings in Figure 5, IM injection has been reported to lead to greater overall skeletal muscle transfection when compared to an US-MB approach wherein plasmid was directly attached to microbubbles [14]. While there are several potential reasons for why these results conflict with ours, including significant differences in microbubble properties, ultrasound pulsing parameters, and MB-plasmid coupling, perhaps the most significant difference is that PEG/PEI nanocomplexes were used to enhance gene delivery in our study, suggesting that nanoparticle complexation may be important in the further development of non-viral gene delivery approaches that utilize US-MB-targeting.

It is also interesting to note that the use of small MBs (1.3 μm) resulted in enhanced gene transfer, as compared to the unsorted MB population (Figure 5). However, these results should be interpreted carefully. This is primarily because infusion simulations revealed that MB administration reduced injected MB concentration by 55% for unsorted MBs, but only 14% for small MBs. This effect, which was likely due to size dependent MB floatation differences in the syringe, implies that the injected concentration of small MBs was significantly higher than unsorted MBs, and this likely contributed to differential transfection levels. In addition, once present in the bloodstream, MB elimination through lung filtration, phagocytosis, and/or dissolution is known to be dependent on MB diameter [45]. Thus, even if the injected MB concentrations were similar for the small and unsorted MB groups, our data would not necessarily indicate that smaller MBs have a more favorable acoustic response for delivery, as has been previously suggested [39, 40].

Another consideration is that MB behavior may be affected by the presence of nearby MBs, in turn effecting acoustic response and transfection efficacy. Although possible, at the MB concentrations used in this study, we assert that MB-MB interactions were minimal. In support, assuming bolus injection, 14% MB loss during administration (i.e. small MBs), and a blood volume of 0.77 mL, ~1.82 × 10⁴ MBs/μl were in circulation at the time of treatment. At this MB dose, a 40μm length vessel with an inner diameter of 40 μm and containing a blood volume of 5 × 10⁻⁵ μl would be expected to contain ~0.3 MBs [40]. Here, circulating MB concentration was even less because we used a steady infusion; therefore, MB-MB interactions in the microcirculation were highly unlikely to occur.