You are on page 1of 13

GENE DELIVERY

Journal of Controlled Release 170 (2013) 401–413

Contents lists available at SciVerse ScienceDirect

Journal of Controlled Release


journal homepage: www.elsevier.com/locate/jconrel

Improving ultrasound gene transfection efficiency by controlling


ultrasound excitation of microbubbles
Z. Fan, D. Chen, C.X. Deng ⁎
Department of Biomedical Engineering, University of Michigan, 2200 Bonisteel Boulevard, Ann Arbor, 48109, USA

a r t i c l e i n f o a b s t r a c t

Article history: Ultrasound application in the presence of microbubbles has shown great potential for non-viral gene transfection
Received 24 January 2013 via transient disruption of cell membrane (sonoporation). However, improvement of its efficiency has largely
Accepted 30 May 2013 relied on empirical approaches without consistent and translatable results. The goal of this study is to develop
Available online 11 June 2013
a rational strategy based on new results obtained using novel experimental techniques and analysis to improve
sonoporation gene transfection. In this study, we conducted experiments using targeted microbubbles that were
Keywords:
Sonoporation
attached to cell membrane to facilitate sonoporation. We quantified the dynamic activities of microbubbles ex-
Ultrasound posed to pulsed ultrasound and the resulting sonoporation outcome, and identified distinct regimes of character-
Microbubbles istic microbubble behaviors: stable cavitation, coalescence and translation, and inertial cavitation. We found that
Intracellular delivery inertial cavitation generated the highest rate of membrane poration. By establishing direct correlation of
Gene transfection ultrasound-induced bubble activities with intracellular uptake and pore size, we designed a ramped pulse
High speed videomicroscopy exposure scheme for optimizing microbubble excitation to improve sonoporation gene transfection. We
implemented a novel sonoporation gene transfection system using an aqueous two phase system (ATPS) for
efficient use of reagents and high throughput operation. Using plasmids coding for the green fluorescence pro-
tein (GFP), we achieved a sonoporation transfection efficiency in rate aortic smooth muscle cells (RASMCs) of
6.9% ± 2.2% (n = 9), comparable with lipofection (7.5% ± 0.8%, n = 9). Our results reveal characteristic
microbubble behaviors responsible for sonoporation and demonstrated a rational strategy to improve
sonoporation gene transfection.
© 2013 Elsevier B.V. All rights reserved.

1. Introduction [12]. Effort to optimize ultrasound parameters to enhance delivery effi-


ciency [16–18] has been limited to empirical screening without detailed
Safe and efficient transport of therapeutic genetic materials across knowledge of the ultrasound induced microbubble activities and their
the cell plasma membrane is critical for successful gene therapy. roles in sonoporation. This is largely due to the complexity of ultrasound
Although viral vectors are efficient for gene transfection, the associated driven microbubble dynamics and the lack of appropriate measurement
inflammatory and adverse immunogenic responses post safety con- techniques.
cerns, motivating development of safer, non-viral delivery techniques Direct optical imaging has revealed the details of single microbubble
[1–3]. Ultrasound techniques have shown potential for non-viral gene dynamics including high speed fluid micro-jet that penetrates cell
delivery both in vitro and in vivo [4–7]. Sonoporation uses ultrasound membrane [8], expansion and contraction of a microbubble that can de-
to excite microbubbles to reversibly disrupt the cell membrane, thereby form and disrupt cell membrane [11,19]; and compression of bubble
transporting impermeant agents into cell cytoplasm without the use of against cell by acoustic radiation force that can rupture the membrane
viral vectors [8–11]. Since ultrasound application can non-invasively [20]. These single cell studies have focused on the dynamic behaviors
target a tissue volume with a superior safety profile, the technique is of single bubbles exposed to short ultrasound pulses, yet typical
particularly advantageous for treating various human diseases includ- sonoporation delivery approaches often involve a large number of
ing cardiovascular diseases, cancer, and thrombosis [12–15]. However, cells subjected to long ultrasound duration or multiple pulses in the
compared to viral or other delivery methods such as lipofection and presence of a population of microbubbles. For example, microbubbles
electroporation, sonoporation generally has relatively low efficiency and drugs are often co-administered intravenously in vivo while ultra-
sound is applied for a period of 10–60 s. For in vitro delivery, cells, either
adherent or in suspension, are mixed with microbubbles distributed in
the bulk solution of extracellular medium. In these cases, long ultra-
⁎ Corresponding author at: Department of Biomedical Engineering, University of
Michigan, 2200 Bonisteel Boulevard, Ann Arbor, 48109-2099, USA. Tel.: + 1 734 936
sound application or multiple pulses can induce highly dynamic activi-
2855; fax: + 1 734 936 1905. ties of microbubbles with inter-bubble interactions distinctly different
E-mail address: cxdeng@umich.edu (C.X. Deng). from single bubble exposed to short ultrasound pulses. Detailed and

0168-3659/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jconrel.2013.05.039
GENE DELIVERY
402 Z. Fan et al. / Journal of Controlled Release 170 (2013) 401–413

complete understanding of such highly dynamic systems relevant for Targestar™-SA microbubbles to bind with the specific antibody,
sonoporation has not been obtained. 8 μL microbubbles (1 × 109 bubble/mL) was mixed with 2 μL of an-
The goal of this study is to develop a rational strategy to improve tibody (0.5 mg/mL) and incubated for 20 min at room temperature
ultrasound-mediated gene transfection using targeted microbubbles. before the addition of 480 μL complete culture medium to reach
This study includes two parts. First, we used human umbilical vein 1.6 × 107 microbubbles/mL and 2 μg/mL for antibody.
endothelial cells (HUVECs) as a model system to investigate the gen- To conjugate the microbubbles with cells, the culture medium
eral processes of microbubble dynamic behaviors driven by pulsed from the cell-seeded dish was removed followed by immediate
ultrasound exposures and to identify experimental conditions for addition of 20 μL of the above solution containing the antibody/
sonoporation mediated intracellular uptake and cell viability. In ad- microbubble complexes. Then the petri dish was flipped over so
dition to transmembrane transport, sonoporation gene transfection, that the attached cells faced downward to facilitate attachment of
like other non-viral gene transfection approaches, faces the limita- the microbubbles with the cells via specific ligand–receptor binding.
tion of low diffusion in the crowded cytoplasm and nucleocytoplasmic When added to the dish immediately after removal of the culture
transport. Non-viral gene transfection is often cell type dependent and medium from the dish, the small amount, i.e. 20 μL, of solution
particularly difficult for non-dividing or primary cells. Thus in the sec- containing the antibody/microbubbles allowed the fluid to spread
ond part of this study, we used rat aortic smooth muscle cells (RASMCs), out on the monolayer to form a thin liquid layer covering all the
known to be hard to transfect, as a model system to demonstrate our ra- adherent cells on the dish bottom. The small volume of solution
tional strategy to improve sonoporation gene transfection. made it possible for the thin liquid layer to stay on the cells without
Targeted microbubbles incorporate ligands on their encapsulat- dripping when the dish was flipped upside down to allow binding of
ing shells to selectively bind to specific receptors expressed on the the microbubbles with cells. After 10 min, the dish was flipped back
cell membrane [21–24], enabling ultrasound molecular imaging by and gentle washing was performed with complete culture medium
recognizing molecular markers associated with diseases including in- to remove unbound microbubbles before experiments.
flammation, ischemia–reperfusion injury, angiogenesis, and thrombosis
[21,25–32]. With the potential to load therapeutic agents [23,24,33–36] 2.2. Experimental setup and ultrasound system
and to retain within specific tissue volume [37,38], targeted microbubbles
provide a unique opportunity for combined ultrasound imaging and As shown in Fig. 1 and described before [10], the 35 mm petri dish
targeted drug delivery with enhanced efficacy and reduced side effects with cells and targeted microbubbles was placed on the stage of an
[31,39,40]. We used high speed videomicroscopy to capture the complete inverted microscope (Nikon Eclipse Ti–U, Melville, NY) equipped with
dynamic process of the initially cell-bound microbubbles exposed to a 20× superfluo objective (Nikon MRF00200, Melville, NY; NA 0.75).
ultrasound pulses. We quantified the characteristic bubble activities A single element planar transducer (1.25 MHz, Advanced Devices,
responsible for intracellular uptake and cell survival. Based on detailed Wakefield, MA), driven by a waveform generator (Agilent Technologies
investigation of pore size and delivery efficiency in sonoporation affected 33250A, Palo Alto, CA) and a 75 W power amplifier (Amplifier Research
by ultrasound parameters at the individual cell level, we developed a 75A250, Souderton, PA), was used to generate ultrasound pulses with
ramped pulse scheme to improve sonoporation gene transfection. various acoustic pressure, duration of each pulse (pulse duration),
For typical in vitro transfection, the agents to be delivered are pulse repetition frequency (PRF) or pulse repetition period (PRP), and
generally mixed in the bulk medium solution with adherent or duty cycle (ratio of pulse duration over PRP). The transducer was
suspended cells. The large volume of medium required in such sys- pre-characterized in free field using a 40 μm calibrated needle hydro-
tems makes inefficient use of the often expensive reagents. In this phone (Precision Acoustics HPM04/1, UK). To avoid standing waves in
study, we implemented a novel sonoporation system using a polyethyl- the dish and to accommodate microscopic imaging, the transducer
ene glycol/dextran (PEG/DEX) aqueous two phase system (ATPS) was mounted at a 45° angle with its active surface submerged in
[41,42] to achieve efficient use of plasmid and high throughput opera- medium and ~ 7.5 × mm (Rayleigh distance) from the cells.
tion by confining GFP plasmid in the small volume (e.g. μL) of DEX
phase for gene transfection. 2.3. High-speed videomicroscopy and characterization of ultrasound-
induced bubble activities
2. Materials and methods
To capture the evolution of microbubble behaviors driven by ul-
2.1. Cell culture and targeted microbubbles trasound pulses, we used a high speed camera (Photron FASTCAM

Human umbilical vein endothelial cells (HUVECs) were used for


experiments to investigate the details of microbubble dynamics and
sonoporation. The cells cultured in flasks in complete human endothe-
lial growth medium (Lonza CC-3124, Walkersville, MD) in a humidified
incubator at 37 °C and 5% CO2. Cells were plated into fibronectin coated
35 mm glass-bottom dishes (Biolegend, San Diego, CA) for one day to
reach 90–100% confluency for experiments. Rat aortic smooth muscle
cells (RASMCs) were used for gene transfection experiments. The cells
were maintained in flasks in complete growth medium consisting of
DMEM/F12 (Gibco, Grand Island, NY) supplemented with 10% fetal
bovine serum and 1% penicillin–streptomycin in a humidified incubator
at 3737 °C and 5% CO2. Cells were plated in fibronectin coated glass-
bottom 35 mm dish for one day to reach 70% confluency at the time
of experiments.
Targestar™-SA (Targeson, La Jolla, CA) microbubbles were used for
sonoporation. Biotinylated anti-human CD31 antibody (eBioscience,
Fig. 1. Experimental setup of ultrasound excitation of targeted microbubbles attached
San Diego, CA) was used to attach the microbubbles to HUVECs. to cells for sonoporation. Fluorescence imaging was used to detect intracellular delivery
Biotinylated anti-mouse CD51 antibody (BioLegend, San Diego, CA) and cell viability, while high speed videomicroscopy was used to monitor ultrasound-
was used to attach the microbubbles to RASMCs. To allow the induced bubble activities.
GENE DELIVERY
Z. Fan et al. / Journal of Controlled Release 170 (2013) 401–413 403

SA1, San Diego, CA) with a frame rate of 20 Kframes/s with 2.6. Size of pores generated by ultrasound excitation of microbubbles
512 × 512 pixels in each frame. To characterize ultrasound-induced
microbubble behaviors, we first calculate the “average rate of active As an ultrasound excited microbubble attached to a cell generates
size change” as 〈|ΔR|/TUS〉, where ΔR is the difference of bubble radius membrane disruption, i.e., pore, PI in the extracellular medium will
before and after an ultrasound pulse and TUS is the pulse duration. The diffuse into the cell cytoplasm and intercalate with DNA and/or RNA
average rate was computed as the mean of |ΔR|/TUS from all pulses. molecules, producing PI fluorescence signal. To estimate the size of
The “total displacement of bubble” was calculated as the sum of the a single pore, we recorded continuously the intracellular PI fluores-
absolute displacements of bubbles occurred during each pulse. cence intensity until 5 min post ultrasound. We calculated the pore
radius based on the temporal change of the total intracellular PI fluo-
rescence intensity after sonoporation, as described previously [46],
2.4. Gas diffusion model of microbubbles assuming the transport of PI into cells as a quasi-steady state diffusion
problem through a small hole in a plane.
The microbubbles used in this study are stabilized perfluorocarbon
gas bubbles encapsulated by a lipid shell, which prevents gas diffu- 2.7. Sonoporation system for gene transfection using aqueous two phase
sion and the effects of the surface tension. To assess the change of system (ATPS)
microbubble properties after an ultrasound pulse, we fit the radius–
time curve of microbubble during the “ultrasound-off” period using PmaxFP™-Green-C plasmid (Lonza, Walkersville, MD), encoding
the following gas diffusion model [43]: the GFP, was used in gene transfection experiments in adherent

0 1
dr L 1 þ 2σP shell −f
− ¼ r @ a r A ð1Þ
dt D þ Rshell 1 þ 3σ shell
ω 4P a r

where r(t) is the bubble radius, L is the Ostwald's coefficient (air, L =


0.02; PFC, L = 0.0002), and Dω is gas diffusivity. In 3.2% dextran solu-
tion, Dω = 1.46 × 10−5 cm2 s−1 based on oxygen diffusion [44] and
in perfluorocarbon (PFC) Dω = 0.7 × 10−5 cm2 s−1. Rshell is the re-
sistance of the shell to gas permeation. For free bubble without
shell, Rshell = 0; air bubble with lipid shell Rshell = 104 s m− 1; PFC
bubble with lipid shell, Rshell = 107 s m− 1. σshell is the surface ten-
sion of the shell. For free bubble without shell in 3.2% dextran solu-
tion [45], σshell = 72 mN m− 1; lipid shelled bubbles, σshell = 0. Pa
is the hydrostatic pressure outside the bubble (Pa = 101.3 kPa),
and f is the ratio of the gas concentration in the bulk medium over
that at saturation (degassed solution, f = 0). The goodness of the
∑ ðy −f Þ2
fitting was evaluated using R2 ¼ 1− i i i 2 , where yi is the ob-
∑i ðyi −y Þ
served value, fi the associated modeled value, and y the mean value.

2.5. Fluorescence imaging and sonoporation outcome

Propidium iodide (PI) is normally excluded from intact cells. We


added PI (100 μM) (Sigma Aldrich, St. Louis, MO) in the extracellular
medium and detected intracellular uptake of PI as fluorescent indica-
tor of cell membrane disruption and transmembrane transport.
We assessed cell viability using calcein AM, a cell-permeable, non-
fluorescent compound. After sonoporation, calcein AM was added
into the extracellular medium at a final concentration of 1 μM.
In live cell, the non-fluorescent calcein AM is converted into green-
fluorescent calcein, after acetoxymethyl ester hydrolysis by intracel-
lular esterases.
As described previously [10], for fluorescent imaging, a monochro-
mator (DeltaRAMX™ PTI, Birmingham, NJ) was used to excite the
samples at 538 nm for PI and 488 nm for calcein and GFP with appro-
priate dichroic filters to detect the emission signals at 610 nm (PI)
and 520 nm (calcein) simultaneously. The fluorescent images were
collected using a cooled CCD camera (Photometrics QuantEM, Tuscon,
AZ) and then pseudo-colored using Matlab.
The cells with attached bubbles were counted toward the total
number of cells. The percentage of calcein positive cells over the
total number of cells is defined as cell viability, the percentage of PI
and calcein positive cells over the total number of cells as the delivery Fig. 2. (A) A typical bright filed image of cells with attached bubbles. (B) Distribution of
rate, and the mean PI fluorescence intensity per cell as the delivery microbubbles per cell obtained from 5 typical experiments. (C) Distribution of bubble
efficiency. radii obtained from 5 typical experiments.
GENE DELIVERY
404 Z. Fan et al. / Journal of Controlled Release 170 (2013) 401–413

RASMCs. We used a DEX/PEG ATPS to spatially confine and localize detected after 24 h incubation. Since the plasmid was only present
GFP plasmids within the DEX phase near the cells. As described pre- within the DEX phase, only the cells covered by the DEX droplets
viously [42], polyethylene glycol (PEG) (2.5% w/w) (35 kDa, Sigma will be transfected. The GFP positive cells relative to the total number
Aldrich, St. Louis, MO) and dextran (3.2% w/w) (500 kDa, Pharmacosmos, of cells was defined as gene transfection efficiency.
Copenhagen, Denmark) were used to form an ATPS. The aqueous DEX Lipofection transfection was performed using lipofectamine 2000
and PEG phases were prepared in culture medium separately before (Invitrogen, Carlsbad, CA). Lipofectamine 2000 and plasmid were added
use. The GFP plasmid was dissolved in DEX to reach a final concentration to dextran solution at the ratio of 1/1 (w/v) to allow transfection using
of 10 μg/mL. the same DEX/PEG ATPS as the sonoporation transfection experiments.
Before gene transfection, the culture medium was removed and After incubation for 4 h, the ATPS was removed and replaced with
3 mL PEG was added in the dish with the RASMCs and attached fresh complete culture medium. GFP expression efficiency was analyzed
microbubbles. Then multiple, discrete dextran droplets (0.4 μL) 20 h later. The percentage of GFP positive cells relative to the total num-
containing the GFP plasmid were added in the dish to form a sepa- ber of cells was defined as gene transfection efficiency.
rated phase at multiple treatment sites. After patterning the DEX
droplets, which spread over the cells, additional 5–7 mL PEG was 2.8. Kinetics of intracellular transport of GFP plasmid
added to the dish without disturbing the patterning to allow immer-
sion of the active element of the ultrasound transducer. Ten minutes We labeled the GFP plasmid with nucleic acid stain fluorophore
after ultrasound exposure, the ATPS was removed from the dish and BOBO™-3 iodide (Invitrogen, Carlsbad, CA) by mixing BOBO-3 with
replaced with fresh complete culture medium. GFP expression was plasmid (molecule ratio of 1 dye molecule per 30 base pairs) for

Fig. 3. Characteristic bubble dynamics induced by pulsed ultrasound exposures. (A) Selective images showing stable cavitation of microbubbles with minimal translational movement.
The cell was outlined in yellow. Acoustic pressure was 0.06 MPa, duty cycle 20%, and PRF 20 Hz. A schematic illustration of the pulsed ultrasound exposure was plotted above the images.
No PI uptake was observed, and cell viability indicated by calcein retention. (B) Coalescence and translation of microbubbles. Acoustic pressure 0.4 MPa, duty cycle 20%, and PRF 20 Hz.
PI fluorescence indicates cell membrane disruption. The absence of calcein indicated cell death. (C) Inertial cavitation with minimal displacement showing shrinkage of microbubbles after
each ultrasound pulse before eventually disappeared. Acoustic pressure was 0.4 MPa, duty cycle 0.016%, and PRF 20 Hz. PI uptake was observed and calcein AM assay showed cell
survived. (D) Characterization of bubble dynamics by the average rate of active bubble size change and total displacement of microbubbles. Data include 257 microbubbles. (E) Cell
viability. (F) PI delivery rate (n = 6 for each group). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
GENE DELIVERY
Z. Fan et al. / Journal of Controlled Release 170 (2013) 401–413 405

30 min at the room temperature in the dark. The labeled plasmid contraction of microbubbles typical of stable cavitation. Most cells
was then used in gene transfection experiments. After sonoporation remained viable and the delivery rate induced by this type of bubble
or lipofection, the cells were washed to remove the BOBO-labeled activities, as measured by PI uptake, is low (Fig. 3A, E, and F).
plasmid in the extracellular medium. Live cell imaging was then If a higher acoustic pressure was used (e.g. 0.43 MPa, the same duty
performed to monitor the internalized BOBO–plasmid complexes cycle 20%), a distinctively different type of dynamic behaviors emerged
using a Carl Zeiss Axio Observer Z1 microscope with an AxioCam which was dominated by microbubble coalescence/aggregation and
camera (Carl Zeiss MicroImaging, Thornwood, NY) and Pecon XL S1 significant translational movement (Fig. 3B, Supplemental video 2).
live-cell incubator system (Pecon GmbH, Erbach, Germany), which In these cases, the initially cell-bound microbubbles detached from
is a humidified thermostatic chamber at 37 °C supplied with 5% the cells almost immediately (e.g. b0.5 ms) after ultrasound application
CO2. The time-lapse bright field images and fluorescent images were and subsequently moved rapidly toward each other to form aggregates
captured every 30 min over 24 h. and coalesce. This is due to the strong attractive forces among bubbles
(the secondary acoustic radiation force or Bjerknes force [47,48]),
3. Results which overcame the receptor–ligand binding between the bubbles
and cells. Aggregated and coalesced bubbles continued to move rapidly
The use of targeted microbubbles in this study enabled a con- across cells in their path, leading to death of many cells (Fig. 3B, E and F).
trolled starting point for ultrasound excitation of microbubbles in Although the changes in microbubble sizes were significant during the
relation to cells. We ensured consistent experimental conditions first ultrasound pulses, translation and coalescence of bubbles in these
by controlling the number of microbubbles attached to each cell. cases were very short lived (b 0.1–0.2 s), resulting in low overall values
Fig. 2A shows a typical bright field image of cells with stably attached of 〈|ΔR/TUS|〉 (Fig. 3D).
microbubbles before ultrasound application. In typical experiments, The third type of bubble behaviors was characterized by large rate
about 7% of cells had no bubbles, 60% of cells had one bubble, of active size change and limited translation movement of microbubbles
22% of cells two bubbles, 8% with three bubbles, and 3% with more (Fig. 3C and Supplemental video 3). In these cases, significant reduction
bubbles (Fig. 2B). The majority of the microbubbles had a radius of of bubble size was observed after each pulse, indicating the occurrence
2.2–3.1 μm (Fig. 2C). of inertial cavitation accompanied by large amplitude expansion/
contraction driven by high acoustic pressures (e.g. 0.43 MPa). Limited
3.1. Characteristic behaviors of microbubbles induced by translational movement of microbubbles was due to the short pulse
pulsed ultrasound duration (e.g. 8 μs), during which the bubbles were not able to gain
enough momentum to detach from the cells. Compared to the other
We first examined the responses of a population of initially two types of bubble dynamics, this type of bubble behaviors signified
cell-bound microbubbles to pulsed ultrasound exposures with vari- by inertial cavitation generated the highest intracellular delivery
ous combinations of ultrasound parameters (e.g. acoustic pressure while maintaining good level of cell viability (Fig. 3E and F). Fig. 3D
from 0.06 to 0.8 MPa, pulse duration from 8 μs to 10 ms, PRF from shows the clustering of 〈|ΔR/TUS|〉 and total displacement of 257 bub-
20 Hz to 1 kHz, and total application duration of 1 s). While the bles exposed to pulsed ultrasound exposures obtained from 36 ex-
microbubbles appeared to exhibit a broad range of complex dynamic periments where acoustic pressure ranged from 0.06 to 0.4 MPa,
activities, characteristic behaviors emerged when 〈|ΔR/TUS|〉 and the duty cycle 0.016–20%, PRF 20 Hz, and total duration 1 s. Correlation
total displacement of microbubbles was considered. Despite the of these characteristic parameters with sonoporation outcome sug-
large parameter space of ultrasound exposures, we observed that gests that inertial cavitation generated efficient membrane disrup-
the seemly widely varied microbubble activities can be characterized tion and delivery while significant displacement of aggregated/
by three basic types of behaviors with key common features, as coalesced bubbles caused cell death (Fig. 3E and F).
shown by the examples in Fig. 3A–C and Supplemental videos 1–3.
The first characteristic type of bubble behaviors was signified by 3.2. Ultrasound induced microbubble shell disruption and gas diffusion
small or no change in size and location of microbubbles (Fig. 3A and
Supplemental video 1). Here, microbubbles typically exhibited limited We examined how microbubbles change after an ultrasound pulse
translational movement (b 1–2 μm) with no or slight reduction of radi- to assess whether ultrasound application disrupted the encapsulating
us after each pulse (Fig. 3D), suggesting small amplitude expansion and shell and affect microbubble stability.

Fig. 4. Bubble dissolution after stable and inertial cavitation. (A) Selected time-lapse images of a bubble undergoing stable cavitation. (B) Radius–time curve of the bubble in (A) after the 1st and
2nd ultrasound pulse and fitting using the diffusion model. Acoustic pressure was 0.06 MPa, duty cycle 20%, PRF 20 Hz. (C) Selected time-lapse images of a bubble undergoing inertial cavitation.
(D) Radius–time curve of the bubble shown in (C) after 1st and 2nd ultrasound pulses and fitting with diffusion model. Acoustic pressure 0.4 MPa, duty cycle 0.016%, PRF 20 Hz.
GENE DELIVERY
406 Z. Fan et al. / Journal of Controlled Release 170 (2013) 401–413

For microbubbles exhibiting stable cavitation, their size stayed decreased delivery rate due to high rate of cell death. We conducted
unchanged or decreased slightly after an ultrasound pulse and during experiments to test whether increasing the acoustic pressure of single
the following ultrasound-off period, as shown by the example in Fig. 4A ultrasound can improve sonoporation outcome. Using acoustic pres-
and B. Fitting of the time-dependent radii during the ultrasound-off sure of 0.4 MPa, 0.6 MPa, 0.8 MPa, and 1.6 MPa respectively, we
period with the gas diffusion model of microbubble in Eq. (1) show found that higher acoustic pressure killed more cells (Fig. 6A) and
that these bubbles behaved like PFC gas bubbles with an intact shell. that the delivery rate (percentage of PI positive cells) increased
In contrast, microbubbles after inertial cavitation not only sustained when pressure increased to 0.6 MPa but decreased with further in-
a significant size decrease after an ultrasound pulse, as shown by the ex- crease of pressure (Fig. 6B). However, the delivery efficiency (PI uptake
ample in Fig. 4C and D (arrows at t = 0 and 50 ms), they also continued per cell) continued to increase and was markedly higher at 0.8 MPa and
to decrease substantially in size during the “ultrasound-off” periods 1.6 MPa (Fig. 6C). More cells had higher PI uptake at higher acoustic
between pulses, indicating diffusion of gas and disruption of the en- pressures (Fig. 6D, F, H, and J). While the overall size distribution of
capsulating shell. Fitting of the radius–time curves during the 1st microbubbles was similar for all experiments, grouping of the bubbles
ultrasound-off period indicates that the bubbles were without shell into different cohorts based on their impacts on cells shows that higher
and composed of 45% ± 5% PFC and (55 ± 5) % air (n = 10, R2 =
0.9 ± 0.05) due to leakage of PFC gas and infusion of air into the bubbles.
Fitting the radius–time curve after the 2nd pulse shows that the bubbles
consisted of 12% ± 8% PFC and (88 ± 8)% air without shell (n = 10,
R2 = 0.83 ± 0.07). The faster size reduction is attributed to a more
dominate surface tension and higher content of the more diffusive air.
These results show that inertial cavitation significantly reduced
microbubble radius during and after an ultrasound pulse. Such size re-
duction will influence how the bubbles respond to additional ultrasound
pulses, due to higher inertial cavitation threshold for smaller bubbles.

3.3. Parameters affecting inertial cavitation and sonoporation outcome

The high efficiency of inertial cavitation in membrane poration


is likely due to robust ultrasound excitation of microbubbles. Here
we report our experimental results to demonstrate optimization of
sonoporation parameters based on their impact on a population of
microbubbles and correlated impacts on cells.

3.3.1. Number of ultrasound pulses


We conducted three sets of experiments using 20, 2, or 1 ultra-
sound pulse(s) respectively to examine the detailed correlation of
bubble activities with delivery. Each ultrasound pulse was 8 μs in du-
ration. Acoustic pressure of 0.43 MPa was used to generate inertial
cavitation of microbubbles with minimal translational movement.
PRF was 20 Hz for all experiments. Under these conditions, applica-
tion of 20 pulses induced most cell death (viability 52% ± 10%,
n = 5), while the viability increased to 72% ± 11% (n = 5) and
95% ± 3% (n = 6) for two pulse and one pulse application respec-
tively (Fig. 5A). Interestingly, delivery rate increased from 31% ± 5%
to 44% ± 7% and 59% ± 5% with fewer pulses (Fig. 5B).
We grouped the microbubbles into different cohorts based on the
status of the cells to which the bubbles were attached: 1) bubbles
associated with non-viable cells, 2) bubbles associated with viable
cells with PI uptake, and 3) bubbles associated with viable cells
without PI uptake. We only analyzed the cells that had one attached
microbubble so that the delivery outcome can be directly related to
the bubble activities. Fig. 5A–C shows that the three cohorts of
microbubbles are clearly separated. Besides the group of bubbles
that generated membrane disruption and PI uptake (blue curves in
Fig. 5A–C), bubbles with larger initial radius tend to exist longer and
cause cell death (red curves in Fig. 5A–C), while the bubbles that be-
came too small quickly failed to induce PI uptake for all experiments
(green curves in Fig. 5A–C). The bubbles in experiments using 20
pulses had the longest duration to impact the cells (although most
of the bubbles were gone after 5 pulses) and caused most cell death
(Fig. 5A–E). While it is possible that these bubbles were effective in Fig. 5. Effects of number of pulses on sonoporation delivery. (A) Microbubbles in
disrupting the cell membrane, the longer impact time eventually experiments using 20 ultrasound pulses (n = 6). Grouping of microbubbles into
caused cell death and significantly decreased the delivery rate. cohorts based on their effect on cells: cell death (red curves, 32 bubbles), PI delivery
(blue curves, 20 bubbles), and no effect (green curves, 18 bubbles). (B) Microbubbles
in experiments using 2 pulses (n = 6). (C) Microbubbles in experiments using 1 pulse
3.3.2. Effect of acoustic pressure on sonoporation delivery (n = 5). Acoustic pressure 0.4 MPa, pulse duration 8 μs, and PRF 20 Hz. (For interpretation
Our results show that application of one ultrasound pulse can ef- of the references to color in this figure legend, the reader is referred to the web version
fectively induce membrane disruption and application of more pulses of this article.)
GENE DELIVERY
Z. Fan et al. / Journal of Controlled Release 170 (2013) 401–413 407

pressures excited a wider range of bubbles (more smaller bubbles) to 3.3.3. Effects of number of bubbles per cell on sonoporation
deliver PI (blue bars in Fig. 6E, G, I, and K). More bubbles were excited As shown by Fig. 2B, a cell may have different number of bubbles
and killed cells (red bars in Fig. 6E, G, I, and K). attached to it, although we used microbubble concentration to ensure

Fig. 6. Effects of acoustic pressure amplitude of single ultrasound pulse (pulse duration 8 μs) application on sonoporation delivery. (A) Viability, (B) delivery rate, (C) Delivery
efficiency (average PI intensity per cell). n = 5 for each acoustic pressure. (D, F, H, and J) Distribution of intracellular PI intensity. (E, G, I and K) Size distribution of different cohorts
of microbubbles grouped based on their roles in sonoporation: viable cells with PI uptake (blue bars), non-viable cell (red), viable cell without PI uptake (green). (For interpretation
of the references to color in this figure legend, the reader is referred to the web version of this article.).
GENE DELIVERY
408 Z. Fan et al. / Journal of Controlled Release 170 (2013) 401–413

most of the cells had only one bubble attached to them, and the num- bubbles decreased in size significantly after inertial cavitation (Fig. 4D),
ber of bubbles attached to a cell may impact the sonoporation out- we hypothesize that a ramped pulse scheme is more effective in exciting
come for that cell. In Fig. 7, we show the results obtained from 5 the microbubbles to induce membrane disruption. Here we report
independent experiments where 1 pulse ultrasound exposure of du- results of pore size distribution and sonoporation outcome from
ration of 8 μs at 0.6 MPa (Fig. 7A) or 1.6 MPa (Fig. 7B) was used. experiments using three different ultrasound exposure schemes:
The percentages of cells were represented in three categories, viable 1) one pulse, 2) two pulses with equal amplitude, and 3) two pulses
without PI uptake (green), viable with PI uptake (blue), and non-viable with ramped amplitude where the 2nd pulse had higher pressure
cells (red). As the number of bubble/cell increased, the percentage of than the 1st pulse.
viable cells with PI uptake increased slightly until there were 4 We estimated the size of pores from the measured time-dependent
bubbles/cell (blue curve in Fig. 7A), and then decreased when there PI fluorescence intensity in individual cells. As shown by the example
were more bubbles attached to a cell. The large variations were observed in Fig. 8A, intracellular transport of PI induced through a pore by an
because there were fewer cells with more bubbles attached to them attached microbubble increased the total PI fluorescence intensity,
(Fig. 2B). The decrease of viable cells with PI uptake is likely due to which eventually reached a stable level (Fig. 8B) with resealing of the
that the cells were killed due to the presence of more bubbles. These pore. Larger pores were generated by application of two pulses, espe-
trends were also observed when higher acoustic pressure of 1.6 MPa cially with amplitude ramping, than a single pulse (Fig. 8C). For exam-
was used (Fig. 7B), although the decrease of the percentage of cells ple, the percentage of pores with diameters > 25 nm increased from
with PI uptake occurred with fewer (3 vs. 4 bubbles/cell). The overall 5% (7 out of 140) with one pulse application to 16.4% with two pulses
percentage was less than that in experiments using lower pressure, of equal amplitude, and reached to 29.3% with two pulses with ramped
as expected. amplitude, corresponding to the fact that more cells had higher amount
of PI uptake (Fig. 8D), suggesting that the 2nd pulse with higher pres-
3.4. Size of pores in sonoporation and improvement of delivery efficiency sure generated more robust response of the microbubbles (Fig. 8E).
Although ramped pulses decreased cell viability (Fig. 8F) and delivery
The size of membrane disruptions, or pores, can be a limiting rate (Fig. 8G), the ability to generate more larger pores is desirable for
factor for delivery of macromolecules such as genes into cells. Since delivering macromolecules such as plasmid.

3.5. Rationally designed strategy to improve gene transfection

For gene transfection experiments, we used a DEX/PEG ATPS for


efficient use of GFP plasmid by confining the plasmid within the
DEX phase. As shown by the example in Fig. 9A, BOBO-labeled GFP
plasmids were confined within DEX droplets covering regions of
RASMCs in the same monolayer. This system not only allowed effi-
cient use of plasmid, but also enabled patterned delivery/transfection
as well as improved throughput in sonoporation delivery.
We successfully transfected the RASMCs with GFP plasmid by
ultrasound excitation of targeted microbubbles (Fig. 9B and C)
achieving transfection efficiency of 1.5 ± 0.8% (n = 9) using one
pulse (0.43 MPa, duration 8 μs) or 3.0 ± 1.1% (n = 6) using two
pulses with equal amplitude (0.43 MPa, duration 8 μs) and an inter-
val of 0.05 s (Fig. 9D). Compared with the delivery rate of PI in
HUVECs (43–60%, Fig. 8G) generated with the same ultrasound pa-
rameters, the low gene transfection efficiency may be due, at least
in part, to the small percentage of large pores permitting intracellu-
lar transport of the GFP plasmid. This is consistent with the increased
gene transfection efficiency (6.9% ± 2.2%, n = 9) achieved by appli-
cation of two pulses with ramped amplitude (0.43 MPa for the 1st
pulse followed by a pulse at 0.6 MPa) (Fig. 9D), which are shown to
be able to generate more large pores (Fig. 8C and D). This transfec-
tion efficiency is comparable with lipofection (7.5 ± 0.8%, n = 9).
Interestingly, the benefit of the 2nd pulse disappeared when interval
of 0.5 s was used, resulting in a gene transfection efficiency of 2.4 ±
1.3% (n = 6) (Fig. 9D). In these cases, the bubbles shrank to smaller
sizes after the longer delay (Fig. 9E).

3.6. Intracellular transport of plasmid in gene transfection

To gain insight of plasmid internalization and intracellular traf-


ficking, we used live cell imaging to examine the kinetics of BOBO-
labeled plasmid in the cells after sonoporation and lipofection. As
shown by Fig. 10A, not all cells with BOBO/plasmid uptake after
sonoporation expressed GFP in the end. In the cell that did express
Fig. 7. Effects of number of bubbles per cell on sonoporation delivery in terms of GFP, diminishing of the BOBO signal corresponded with the appearance
percentage of cells that were non-viable (red curves), or viable without PI uptake
(green curves), or viable with PI uptake (blue curves). (A) Experiments (n = 5) using 1
of GFP signal around 4–8 h. The BOBO signal in other non-expressing
pulse ultrasound exposure with duration 8 μs, acoustic pressure 0.6 MPa. (B) Experiments cells disappeared well before 24 h (Fig. 10A and C), indicating degrada-
(n = 5) using 1 pulse ultrasound exposure with duration 8 μs, acoustic pressure 1.6 MPa. tion of the plasmid. In contrast, BOBO signal after lipofection decreased
GENE DELIVERY
Z. Fan et al. / Journal of Controlled Release 170 (2013) 401–413 409

Fig. 8. (A) Intracellular transport of PI. (B) Total PI fluorescence after sonoporation. (C) Sizes of pores generated in sonoporation using different ultrasound conditions: 1 pulse
(0.4 MPa, 8 μs), 2 pulses with equal amplitude (0.4 MPa, 8 μs, pulse interval 0.05 s), and 2 pulses with ramped amplitude (0.4 MPa and 0.6 MPa, pulse duration 8 μs, pulse interval
0.05 s). n = 140 for each condition. (D) Distribution of intracellular PI uptake for the three ultrasound conditions. (E) Representative radius–time curve of the bubbles (n = 10 for
each condition). (F) Cell viability. (G) Delivery rate. n = 6 for each group.

slower and was present 24 h after transfection (Fig. 10B and D) with a exhibited very complex dynamic behaviors. Coalescence of bubbles
later GFP expression (~12 h) (Fig. 10B). and large/rapid translation movement of bubbles overwhelming gener-
ate high rate of cell death, negatively affecting sonoporation outcome.
4. Discussion Therefore we focused on investigating the responses of a population
of microbubbles under the influence of pulsed ultrasound exposures
4.1. Large parameter space for sonoporation their effects on sonoporation outcome.
Since the resonant frequency is mostly determined by microbubble
The response of microbubbles to ultrasound excitation depends on a diameter, shell properties, and gas content within the bubble, physical
multitude of parameters including the center frequency of ultrasound binding of bubbles to cell membrane is not expected to have signifi-
pulses, acoustic pressure amplitude, pulse repetition frequency, dura- cant effect on bubble response in the context of ultrasound frequency,
tion of each pulse, and total application duration etc. It is a large param- although the presence of the cells in the vicinity of bubbles might in-
eter space considering variation of all these parameters. In this study, fluence the bubble oscillation and collapse behaviors. In addition,
we chose the center frequency to be 1.25 MHz without varying it targeting of the microbubbles to cells increases the likelihood of im-
because this frequency is close to the resonance frequency (i.e. the pact of ultrasound-induced microbubble activities to the cells.
Minnaert resonant frequency) of the microbubbles used in this study,
which had an average radius of about 2.5 μm. The use of frequency 4.2. High speed videomicroscopy to investigate ultrasound induced
close to the bubbles' resonant frequency is expected to generate robust microbubble activities
response of individual microbubbles. However, the response of multiple
or a population of microbubbles to pulsed ultrasound exposures was Sonoporation often uses pulsed ultrasound exposures with a cen-
more complicated, particularly by the interaction of microbubbles. ter frequency of 1–3 MHz, pulse duration of μs–ms, PRF 10–100 Hz,
Microbubble interactions are greatly affected by the combined influ- and total application 1–100 s. The dynamic responses of microbubbles
ence of acoustic pressure, PRF, and duration of each pulse. As shown to such ultrasound exposures can exhibit features of multiple time
by our results, a population of microbubbles subjected to pulsed ultra- scales spanning orders of magnitude. Imaging of these bubble dy-
sound exposures, which is the typical setting in sonoporation studies, namic activities is challenging.
GENE DELIVERY
410 Z. Fan et al. / Journal of Controlled Release 170 (2013) 401–413

Fig. 9. (A) Bobo/plasmid confined in patterned DEX droplets using a DEX/PEG ATPS. (B) GFP transfection in RASMCs. (C) Superimposed image of bright filed image with images of GFP
expression. (D) Gene transfection efficiency. For experiments using 1 pulse (n = 6), acoustic pressure was 0.4 MPa. For 2 pulse exposure (n = 6) with equal amplitude, acoustic pressure
was 0.4 MPa and pulse interval 0.05 s. For 2 pulses with ramped amplitude, the acoustic pressure was 0.4 MPa for the 1st pulse and 0.6 MPa for the 2nd pulse. The time interval was either
0.5 s (n = 6) or 0.05 s (n = 9). Pulse duration was 8 μs for all pulses. (E) The radius–time curves for bubbles exposed to 2 ramped pulses with 0.5 s and 0.05 s time interval. n = 10.

The actual expansion/contraction and/or collapse of a microbubble aggregation/coalescence and growth via rectified diffusion) driven by
driven by MHz ultrasound can only be resolved using ultrafast imag- the acoustic radiation force.
ing with frame rate above several Mframes/s [8,11,49–53]. However, Reflective of the fundamental aspects of microbubble responses
ultrafast imaging is not only high cost, but also limited by the small to ultrasound application, these characteristic factors quantitatively
number of images (e.g. 10–30) or short recording time (e.g. several μs) categorized microbubble behaviors into three distinct types. Robust
restricting their utility in studying bubble dynamics when multiple collapse of microbubble (inertial cavitation) was most efficient in
pulses or longer ultrasound application is often used. sonoporation.
We used high speed videomicroscopy with frame rate of 20–
200 Kframes/s in this study. Capable of imaging for longer period of 4.4. Targeted microbubbles
time (seconds), high speed imaging at frame rate of 20 Kframes/s
also provides sufficient temporal resolution for capturing dynamic We found that inertial cavitation of targeted bubbles generated by
behaviors at a time scale of the pulse duration, typically several μs ultrasound pulses with short pulse duration and sufficient pressure is
to tens of ms. The dynamic features associated with pulsed ultra- most efficient for sonoporation. The intimate contact of targeted
sound exposures has a time scale of the pulse repetition period microbubbles with cells (b100 nm [54]) ensured effective impact of
(or 1/PRF), i.e., 10–100 ms. Since the features of microbubble activi- microbubbles on cells and also increased cell vulnerability. We ob-
ties during the total duration represent repetitive behaviors, we fo- served that the acoustic radiation forces associated with longer
cused our experiments with ultrasound duration of 1 s. We show pulse duration can detach the initially cell-bound bubbles and result
that high speed imaging can capture the key features of the dynamic in aggregation/coalescence of microbubbles. Translation of the aggre-
behaviors of a population of microbubbles at time scales relevant to gated bubbles across the cell monolayer caused cell death.
pulsed ultrasound exposures.
4.5. Short term sonoporation outcome

4.3. Characteristic microbubble behaviors and their impact on To characterize ultrasound-induced microbubble activities and
intracellular delivery their effects on cells, we used PI as markers to assess the delivery
rate and the delivery efficiency via the intracellular PI fluorescence
The large parameter space (e.g. acoustic pressure, pulse duration, intensity during or immediately after sonoporation. Although the
PRF, etc.) associated with pulsed ultrasound exposures can generate delivery rate may be underestimated if small amount of PI uptake
microbubble activities that are highly dynamic and varied, making was not detected in some cells due to the sensitivity of the imaging
quantitative analysis difficult. system, use of PI enabled efficient investigation and correlation of
In this study, by focusing analysis of bubble activities at the time the dynamic process of sonoporation with microbubble activities
scales corresponding to the pulse duration and pulse repetition peri- induced by different ultrasound parameters at the individual cell
od, we identified two characteristic factors that capture the key level. For example, we estimated the size of pores generated in
features of the bubble behaviors in sonoporation. The rate of active sonoporation based on the measurements of the time-dependent in-
size change of microbubbles, which is calculated at the end of each tracellular transport of PI. The amount of PI uptake in a cell was used
pulse, captures the acute change incurred during the ultrasound to assess delivery related to the pore size and duration.
pulse. Our previous study using ultrafast imaging has shown that Calcein assay used in this study may overestimate the cell viability
such observed size reduction immediately after an ultrasound pulse because indication of live cells was based on esterase activity, which
is directly related to the amplitude of bubble expansion/contraction may still be present if the cells were in the early stages of apoptosis.
in cavitation [46]. The total displacement of microbubbles emphasizes We recently showed that targeted microbubbles excited by pulsed
the significant translational movement of large bubbles (from bubble ultrasound can induce significant cell apoptosis exhibiting annexinV
GENE DELIVERY
Z. Fan et al. / Journal of Controlled Release 170 (2013) 401–413 411

Fig. 10. (A) Time lapse images of intracellular delivery of BOBO–plasmid (middle panel) and GFP expression after sonoporation (bottom). (B) Live cell imaging of BOBO–plasmid
and expression after lipofection. (C) 24 h after sonoporation. (D) 24 h after lipofection. The left panels in (C and D) are fluorescent images showing GFP expression. The middle
panels show intracellular BOBO–plasmid complexes. The right panels are phase contrast images superimposed with fluorescent images of GFP and BOBO–plasmid.

labeling, loss of mitochondrial membrane potential, caspase activa- of the small DEX droplets over the cells in the monolayer retained the
tion and changes in nuclear morphology [55]. plasmid within the DEX close to the cells as the interfacial energy at
the PEG–DEX interface prevents the plasmid from diffusing from DEX
4.6. ATPS in sonoporation gene transfection into the PEG phase. This eliminated the need to use large amount of
plasmid in the entire volume of medium in the cell culture dish and
We used ATPS to confine GFP plasmid close to the adherent cells by reduced the amount of plasmid by as much as 20,000-fold because plas-
stably partitioning the plasmid in DEX phase. Small droplets of DEX mid was confined in 0.4 μL DEX droplets instead of 8–10 mL solution in
(i.e. 0.5–2.0 μL) was deposited onto the monolayer cells (with attached typical sonoporation experiments in a dish or similar container.
microbubbles) in PEG. DEX has a slightly greater density than PEG, caus- Both the DEX and the PEG used in our system are biocompatible
ing the DEX droplets to sink and contact the cell monolayer. Spreading and noncytotoxic, as demonstrated in lipofection transfection and
GENE DELIVERY
412 Z. Fan et al. / Journal of Controlled Release 170 (2013) 401–413

previous studies that used cell microinjection of fluorophore-conjugated microbubble dynamics and impact on cells can be obtained to im-
polymers such as dextran for cell tracing. The ATPS medium imposes no prove sonoporation gene transfection and delivery.
acoustic discontinuity at the interface of the two phases while confining Supplementary data to this article can be found online at http://
the reagents within a fluid. Capable of placing multiple, discrete treat- dx.doi.org/10.1016/j.jconrel.2013.05.039.
ment sites within the same monolayer, our method enables not only
efficient use of reagents but also high throughput operation.
Acknowledgment
4.7. Rational design of ultrasound exposures to improve sonoporation
gene transfection Funding from the Beyster Foundation and NIH (CA116592) is
acknowledged. We thank Dr. John Frampton for his assistance on
Unlike conventional sonoporation studies, where optimization is ATPS and Dr. Jianping Fu for the use of live cell imaging system.
based on statistical association of ultrasound parameters with cell
status without any knowledge of the microbubble activities during
References
ultrasound application [17,56–58], our current study developed a
rational scheme to improve sonoporation based on detailed analysis [1] R. Waehler, S.J. Russell, D.T. Curiel, Engineering targeted viral vectors for gene
of microbubble behaviors and their correlation with delivery outcome therapy, Nat. Rev. Genet. 8 (2007) 573–587.
[2] Z.C. Hartman, D.M. Appledorn, A. Amalfitano, Adenovirus vector induced innate
at the individual cell level. Specifically, PI uptake and cell viability of immune responses: impact upon efficacy and toxicity in gene therapy and vaccine
an individual cell were correlated with the dynamic activities of the applications, Virus Res. 132 (2008) 1–14.
specific microbubble attached to that cell. We found that 1) inertial [3] S. Nayak, R.W. Herzog, Progress and prospects: immune responses to viral vectors,
Gene Ther. 17 (2010) 295–304.
cavitation was most effective in generating membrane disruption; [4] Y.S. Li, E. Davidson, C.N. Reid, A.P. McHale, Optimising ultrasound-mediated gene
2) the initial size and active life time of a bubble determined its role transfer (sonoporation) in vitro and prolonged expression of a transgene in vivo:
in sonoporation; and 3) ultrasound pulses with ramped amplitude potential applications for gene therapy of cancer, Cancer Lett. 273 (2009) 62–69.
[5] S. Tsunoda, O. Mazda, Y. Oda, Y. Iida, S. Akabame, T. Kishida, M. Shin-Ya, H. Asada,
efficiently generated larger pores for gene transfection. S. Gojo, J. Imanishi, H. Matsubara, T. Yoshikawa, Sonoporation using microbubble
BR14 promotes pDNA/siRNA transduction to murine heart, Biochem. Biophys.
4.8. Mechanism and processes involved in sonoporation-mediated Res. Commun. 336 (2005) 118–127.
[6] S. Chen, J.H. Ding, R. Bekeredjian, B.Z. Yang, R.V. Shohet, S.A. Johnston, H.E.
gene transfection
Hohmeier, C.B. Newgard, P.A. Grayburn, Efficient gene delivery to pancreatic
islets with ultrasonic microbubble destruction technology, Proc. Natl. Acad. Sci.
In this study, we achieved a sonoporation transfection efficiency of U. S. A. 103 (2006) 8469–8474.
[7] A. Delalande, M.F. Bureau, P. Midoux, A. Bouakaz, C. Pichon, Ultrasound-assisted
6.9% ± 2.2% in RASMCs, which is comparable with lipofection (7.5 ±
microbubbles gene transfer in tendons for gene therapy, Ultrasonics 50 (2010)
0.8%) and much higher than reported ultrasound transfection in 269–272.
RASMCs [59]. RASMCs are known to be difficult to transfect and the [8] P. Prentice, A. Cuschieri, K. Dholakia, M. Prausnitz, P. Campbell, Membrane
gene transfection efficiency we obtained is much lower than PI deliv- disruption by optically controlled microbubble cavitation, Nat. Phys. 1 (2005)
107.
ery rate (59% ± 5%) in HUVECs using the same ultrasound parame- [9] R.K. Schlicher, H. Radhakrishna, T.P. Tolentino, R.P. Apkarian, V. Zarnitsyn, M.R.
ters. This may underline the importance of the size of the pores for Prausnitz, Mechanism of intracellular delivery by acoustic cavitation, Ultrasound
different agents to be delivered (PI is 668 Da and the GFP plasmid Med. Biol. 32 (2006) 915–924.
[10] Z. Fan, R.E. Kumon, J. Park, C.X. Deng, Intracellular delivery and calcium transients
~ 3000 kDa). As shown in Fig. 8C, the size of the pores generated in generated in sonoporation facilitated by microbubbles, J. Control. Release 142
sonoporation ranged from several nm to 60 nm, but most of the (2010) 31–39.
pores are smaller than 25 nm, limiting intracellular transport of plas- [11] A. van Wamel, K. Kooiman, M. Harteveld, M. Emmer, F.J. ten Cate, M. Versluis, N.
de Jong, Vibrating microbubbles poking individual cells: drug transfer into cells
mid. Increased number of larger pores resulted in higher gene trans- via sonoporation, J. Control. Release 112 (2006) 149–155.
fection efficiency when ramped pulse application was used (Figs. 8C [12] R. Bekeredjian, P.A. Grayburn, R.V. Shohet, Use of ultrasound contrast agents for
and 9D). gene or drug delivery in cardiovascular medicine, J. Am. Coll. Cardiol. 45 (2005)
329–335.
After sonoporation, migration of plasmid toward nucleus and
[13] Y. Taniyama, J. Azuma, H. Rakugi, R. Morishita, Plasmid DNA-based gene transfer
across the nucleic envelope may also affect successful gene expres- with ultrasound and microbubbles, Curr. Gene Ther. 11 (2011) 485–490.
sion [33,60–62]. For example, naked DNA in the cytoplasm may be [14] F. Xie, E.C. Everbach, S. Gao, L.K. Drvol, W.T. Shi, F. Vignon, J.E. Powers, J. Lof, T.R.
Porter, Effects of attenuation and thrombus age on the success of ultrasound and
degraded by resident cytosolic DNases before expression takes place
microbubble-mediated thrombus dissolution, Ultrasound Med. Biol. 37 (2011)
[60]. Molecules smaller than 40 kDa can diffuse through nuclear 280–288.
pores, but macromolecule such as plasmid may not across the nuclear [15] N.Y. Rapoport, A.M. Kennedy, J.E. Shea, C.L. Scaife, K.H. Nam, Controlled and targeted
envelop efficiently by passive diffusion [62,63]. Additional studies are tumor chemotherapy by ultrasound-activated nanoemulsions/microbubbles,
J. Control. Release 138 (2009) 268–276.
needed to investigate the mechanisms of how kinetics of plasmid in- [16] V. Zarnitsyn, C.A. Rostad, M.R. Prausnitz, Modeling transmembrane transport
side the cells affects gene expression to further improve sonoporation through cell membrane wounds created by acoustic cavitation, Biophys. J. 95
gene transfection efficiency. (2008) 4124–4138.
[17] B.D. Meijering, R.H. Henning, W.H. Van Gilst, I. Gavrilovic, A. Van Wamel, L.E.
Deelman, Optimization of ultrasound and microbubbles targeted gene delivery
5. Conclusion to cultured primary endothelial cells, J. Drug Target. 15 (2007) 664–671.
[18] L.B. Feril Jr., R. Ogawa, K. Tachibana, T. Kondo, Optimized ultrasound-mediated
gene transfection in cancer cells, Cancer Sci. 97 (2006) 1111–1114.
We developed a rational strategy to improve sonoporation gene [19] P. Marmottant, S. Hilgenfeldt, Controlled vesicle deformation and lysis by single
transfection by optimizing ultrasound parameters to induce con- oscillating bubbles, Nature 423 (2003) 153.
trolled and robust microbubble activities. By characterizing and [20] Y. Zhou, K. Yang, J. Cui, J.Y. Ye, C.X. Deng, Controlled permeation of cell membrane
by single bubble acoustic cavitation, J. Control. Release 157 (2012) 103–111.
correlating microbubble activities with sonoporation outcome, we [21] C.R. Anderson, X. Hu, H. Zhang, J. Tlaxca, A.E. Decleves, R. Houghtaling, K. Sharma,
identified three characteristic bubble behaviors and found that inertial M. Lawrence, K.W. Ferrara, J.J. Rychak, Ultrasound molecular imaging of tumor
cavitation of microbubbles generated the highest rate of membrane dis- angiogenesis with an integrin targeted microbubble contrast agent, Invest. Radiol.
46 (2011) 215–224.
ruption and intracellular delivery. Based on how ultrasound parameters
[22] B.A. Kaufmann, J.R. Lindner, Molecular imaging with targeted contrast ultrasound,
affected microbubble activities and sizes of the pores generated, we Curr. Opin. Biotechnol. 18 (2007) 11–16.
designed a ramped pulse scheme to improve gene transfection. We [23] K.W. Ferrara, M.A. Borden, H. Zhang, Lipid-shelled vehicles: engineering for
implemented a novel sonoporation system using ATPS that enabled ultrasound molecular imaging and drug delivery, Acc. Chem. Res. 42 (2009)
881–892.
efficient use of reagents and high throughput operation. Our results [24] A.L. Klibanov, Preparation of targeted microbubbles: ultrasound contrast agents
show that rational design of ultrasound parameters based on for molecular imaging, Med. Biol. Eng. Comput. 47 (2009) 875–882.
GENE DELIVERY
Z. Fan et al. / Journal of Controlled Release 170 (2013) 401–413 413

[25] A.L. Klibanov, Ultrasound molecular imaging with targeted microbubble contrast [44] C.S. Ho, L.K. Ju, C.T. Ho, Measuring oxygen diffusion coefficients with polaro-
agents, J. Nucl. Cardiol. 14 (2007) 876–884. graphic oxygen electrodes. II. Fermentation Media, Biotechnol. Bioeng. 28
[26] B.A. Kaufmann, J.M. Sanders, C. Davis, A. Xie, P. Aldred, I.J. Sarembock, J.R. Lindner, (1986) 1086–1092.
Molecular imaging of inflammation in atherosclerosis with targeted ultrasound [45] A. Ludwig, M. Van Ooteghem, Influence of the surface tension of eye drops on the
detection of vascular cell adhesion molecule-1, Circulation 116 (2007) 276–284. retention of a tracer in the precorneal area of human eyes, J. Pharm. Belg. 43
[27] H. Leong-Poi, J. Christiansen, P. Heppner, C.W. Lewis, A.L. Klibanov, S. Kaul, J.R. (1988) 157–162.
Lindner, Assessment of endogenous and therapeutic arteriogenesis by contrast ultra- [46] Z. Fan, H. Liu, M. Mayer, C.X. Deng, Spatiotemporally controlled single cell
sound molecular imaging of integrin expression, Circulation 111 (2005) 3248–3254. sonoporation, Proc. Natl. Acad. Sci. U. S. A. 109 (2012) 16486–16491.
[28] P.M. Winter, A.M. Morawski, S.D. Caruthers, R.W. Fuhrhop, H. Zhang, T.A. [47] P.A. Dayton, K.E. Morgan, A.L. Klibanov, G. Brandenburger, K.R. Nightingale, K.W.
Williams, J.S. Allen, E.K. Lacy, J.D. Robertson, G.M. Lanza, S.A. Wickline, Molecular Ferrara, A preliminary evaluation of the effects of primary and secondary radiation
imaging of angiogenesis in early-stage atherosclerosis with alpha(v)beta3- forces on acoustic contrast agents, IEEE Trans. Ultrason. Ferroelectr. Freq. Control
integrin-targeted nanoparticles, Circulation 108 (2003) 2270–2274. 44 (1997) 1264–1277.
[29] H. Leong-Poi, Molecular imaging using contrast-enhanced ultrasound: evaluation [48] P. Palanchon, P. Tortoli, A. Bouakaz, M. Versluis, N. de Jong, Optical observations of
of angiogenesis and cell therapy, Cardiovasc. Res. 84 (2009) 190–200. acoustical radiation force effects on individual air bubbles, IEEE Trans. Ultrason.
[30] S. Tinkov, C. Coester, S. Serba, N.A. Geis, H.A. Katus, G. Winter, R. Bekeredjian, Ferroelectr. Freq. Control 52 (2005) 104–110.
New doxorubicin-loaded phospholipid microbubbles for targeted tumor therapy: [49] C.D. Ohl, M. Arora, R. Ikink, N. de Jong, M. Versluis, M. Delius, D. Lohse,
in-vivo characterization, J. Control. Release 148 (2010) 368–372. Sonoporation from jetting cavitation bubbles, Biophys. J. 91 (2006) 4285–4295.
[31] X. Xiong, F. Zhao, M. Shi, H. Yang, Y. Liu, Polymeric microbubbles for ultrasonic [50] M. Postema, P. Marmottant, C.T. Lancee, S. Hilgenfeldt, N. de Jong, Ultrasound-
molecular imaging and targeted therapeutics, J. Biomater. Sci. Polym. Ed. 22 induced microbubble coalescence, Ultrasound Med. Biol. 30 (2004) 1337–1344.
(2011) 417–428. [51] J.E. Chomas, P. Dayton, D. May, K. Ferrara, Threshold of fragmentation for ultra-
[32] F. Cavalieri, M. Zhou, M. Ashokkumar, The design of multifunctional microbubbles for sonic contrast agents, J. Biomed. Opt. 6 (2001) 141–150.
ultrasound image-guided cancer therapy, Curr. Top. Med. Chem. 10 (2010) 1198–1210. [52] T. Faez, I. Skachkov, M. Versluis, K. Kooiman, N. de Jong, In vivo characterization
[33] J.L. Tlaxca, C.R. Anderson, A.L. Klibanov, B. Lowrey, J.A. Hossack, J.S. Alexander, of ultrasound contrast agents: microbubble spectroscopy in a chicken embryo,
M.B. Lawrence, J.J. Rychak, Analysis of in vitro transfection by sonoporation Ultrasound Med. Biol. 38 (2012) 1608–1617.
using cationic and neutral microbubbles, Ultrasound Med. Biol. 36 (2010) [53] M. Postema, A. van Wamel, C.T. Lancee, N. de Jong, Ultrasound-induced encapsu-
1907–1918. lated microbubble phenomena, Ultrasound Med. Biol. 30 (2004) 827–840.
[34] S.M. Janib, A.S. Moses, J.A. MacKay, Imaging and drug delivery using theranostic [54] V. Garbin, M. Overvelde, B. Dollet, N. de Jong, D. Lohse, M. Versluis, Unbinding of
nanoparticles, Adv. Drug Deliv. Rev. 62 (2010) 1052–1063. targeted ultrasound contrast agent microbubbles by secondary acoustic forces,
[35] S.R. Sirsi, S.L. Hernandez, L. Zielinski, H. Blomback, A. Koubaa, M. Synder, S. Phys. Med. Biol. 56 (2011) 6161–6177.
Homma, J.J. Kandel, D.J. Yamashiro, M.A. Borden, Polyplex-microbubble hybrids [55] J.P. Frampton, Z. Fan, A. Simon, D. Chen, C.X. Deng, S. Takayama, Aqueous
for ultrasound-guided plasmid DNA delivery to solid tumors, J. Control. Release two-phase system patterning of microbubbles: localized induction of apoptosis
157 (2012) 224–234. in sonoporated cells, Adv. Funct. Mater. (2013), http://dx.doi.org/10.1002/adfm.
[36] X. Feng, F. Lv, L. Liu, H. Tang, C. Xing, Q. Yang, S. Wang, Conjugated polymer 201203321. (in press).
nanoparticles for drug delivery and imaging, ACS Appl. Mater. Interfaces 2 (2010) [56] V.G. Zarnitsyn, M.R. Prausnitz, Physical parameters influencing optimization
2429–2435. of ultrasound-mediated DNA transfection, Ultrasound Med. Biol. 30 (2004) 527–538.
[37] A.L. Klibanov, J.J. Rychak, W.C. Yang, S. Alikhani, B. Li, S. Acton, J.R. Lindner, K. Ley, [57] A. Rahim, S.L. Taylor, N.L. Bush, G.R. ter Haar, J.C. Bamber, C.D. Porter, Physical
S. Kaul, Targeted ultrasound contrast agent for molecular imaging of inflamma- parameters affecting ultrasound/microbubble-mediated gene delivery efficiency
tion in high-shear flow, Contrast Media Mol. Imaging 1 (2006) 259–266. in vitro, Ultrasound Med. Biol. 32 (2006) 1269–1279.
[38] J.J. Rychak, Molecular Imaging of Carotid Plaque with Targeted Ultrasound Contrast, [58] R. Karshafian, P.D. Bevan, R. Williams, S. Samac, P.N. Burns, Sonoporation by
Springer, 2011, pp. 153–161. ultrasound-activated microbubble contrast agents: effect of acoustic exposure
[39] K. Kooiman, M. Foppen-Harteveld, A.F. van der Steen, N. de Jong, Sonoporation parameters on cell membrane permeability and cell viability, Ultrasound Med.
of endothelial cells by vibrating targeted microbubbles, J. Control. Release 154 Biol. 35 (2009) 847–860.
(2011) 35–41. [59] L.C. Phillips, A.L. Klibanov, B.R. Wamhoff, J.A. Hossack, Targeted gene transfection
[40] A.V. Patil, J.J. Rychak, A.L. Klibanov, J.A. Hossack, Real-time technique for improv- from microbubbles into vascular smooth muscle cells using focused, ultrasound-
ing molecular imaging and guiding drug delivery in large blood vessels: in vitro mediated delivery, Ultrasound Med. Biol. 36 (2010) 1470–1480.
and ex vivo results, Mol. Imaging 10 (2011) 238–247. [60] S. Mehier-Humbert, T. Bettinger, F. Yan, R.H. Guy, Ultrasound-mediated gene
[41] H. Tavana, A. Jovic, B. Mosadegh, Q.Y. Lee, X. Liu, K.E. Luker, G.D. Luker, S.J. Weiss, delivery: kinetics of plasmid internalization and gene expression, J. Control.
S. Takayama, Nanolitre liquid patterning in aqueous environments for spatially Release 104 (2005) 203–211.
defined reagent delivery to mammalian cells, Nat. Mater. 8 (2009) 736–741. [61] M.S. Al-Dosari, X. Gao, Nonviral gene delivery: principle, limitations, and recent
[42] H. Tavana, B. Mosadegh, S. Takayama, Polymeric aqueous biphasic systems for progress, AAPS J. 11 (2009) 671–681.
non-contact cell printing on cells: engineering heterocellular embryonic stem [62] J.M. Escoffre, J. Teissie, M.P. Rols, Gene transfer: how can the biological barriers be
cell niches, Adv. Mater. 22 (2010) 2628–2631. overcome? J. Membr. Biol. 236 (2010) 61–74.
[43] K. Ferrara, R. Pollard, M. Borden, Ultrasound microbubble contrast agents: [63] H. Kamiya, Y. Fujimura, I. Matsuoka, H. Harashima, Visualization of intracellular
fundamentals and application to gene and drug delivery, Annu. Rev. Biomed. trafficking of exogenous DNA delivered by cationic liposomes, Biochem. Biophys.
Eng. 9 (2007) 415–447. Res. Commun. 298 (2002) 591–597.

You might also like