Microfluidic manufacture of rt-PA -loaded echogenic liposomes

Madhuvanthi A. Kandadai1,2 & Prithviraj Mukherjee3 & Himanshu Shekhar4 &

George J. Shaw1& Ian Papautsky3 & Christy K. Holland4

Published online: 20 May 2016# Springer Science+Business Media New York 2016

Abstract Echogenic liposomes (ELIP), loaded with recombi-nant tissue-type plasminogen activator (rt-PA) andmicrobubbles that act as cavitation nuclei, are under develop-ment for ultrasound-mediated thrombolysis. Conventionalmanufacturing techniques produce a polydisperse rt-PA-loaded ELIP population with only a small percentage of par-ticles containing microbubbles. Further, a polydisperse popu-lation of rt-PA-loaded ELIP has a broadband frequency re-sponse with complex bubble dynamics when exposed topulsed ultrasound. In this work, a microfluidic flow-focusingdevice was used to generate monodisperse rt-PA-loaded ELIP(μtELIP) loaded with a perfluorocarbon gas. The rt-PA asso-ciated with the μtELIP was encapsulated within the lipid shellas well as intercalated within the lipid shell. The μtELIP had amean diameter of 5 μm, a resonance frequency of 2.2 MHz,and were found to be stable for at least 30 min in 0.5 % bovineserum albumin. Additionally, 35 % of μtELIP particles wereestimated to contain microbubbles, an order of magnitudehigher than that reported previously for batch-produced rt-PA-loaded ELIP. These findings emphasize the advantages

offered by microfluidic techniques for improving the encap-sulation efficiency of both rt-PA and perflurocarbonmicrobubbles within echogenic liposomes.

Keywords Echogenic liposomes . Recombinant tissue-typeplasminogen activator . Ultrasound-mediated thrombolysis .

Microfluidic flow-focusing . Stroke treatment

1 Introduction

Recombinant tissue-plasminogen activator (rt-PA) is the onlyFDA approved thrombolytic for the treatment of acute ische-mic stroke. However, the use of rt-PA is associated with anincreased risk of symptomatic intracranial hemorrhage, poorrecanalization efficiency, and high rates of reocclusion (Hackeet al. 2004; Alexandrov and Grotta 2002; Burgin et al. 2000).In vitro, animal, and human studies have shown that thethrombolytic efficacy of rt-PA is significantly enhanced withconcomitant exposure to ultrasound (Alexandrov et al. 2004;Braaten et al. 1997; Everbach and Francis 2000; Holland et al.2002; Laing et al. 2012; Lauer et al. 1992; Shaw et al. 2008;Suchkova et al. 1998). Ultrasound-mediated thrombolysis(UMT) is a promising strategy to help lower the dose of rt-PA and enhance therapeutic efficacy. One of the main mech-anisms responsible for UMT is acoustic cavitation, a processof sustained volumetric gentle oscillation (stable cavitation) orexpansion and rapid collapse (inertial cavitation) of gaseousand vapor bubbles in a liquid due to an acoustic pressurefield (Flynn 1982). Cavitation generates local velocityand pressure gradients, which can enhance the uptakeof therapeutics into the surrounding tissue (Sutton et al.2013). Previous in vitro studies have shown that sustainedstable cavitation is necessary for promoting clot lysis(Datta et al. 2006).

* Madhuvanthi A. Kandadaimadhu.kandadai@gmail.com

1 Department of Emergency Medicine, University of Cincinnati,231 Albert Sabin Way, Suite 1551, Cincinnati, OH 45267, USA

2 Department of Emergency Medicine, 231 Albert Sabin Way,CVC 3974, Cincinnati, OH 45267-0769, USA

3 Department of Electrical Engineering and Computing Systems,University of Cincinnati, 812 Rhodes Hall, Cincinnati, OH 45221,USA

4 Department of Internal Medicine, Division of Cardiovascular Healthand Diseases, University of Cincinnati, 231 Albert Sabin Way,Cincinnati, OH 45267, USA

Biomed Microdevices (2016) 18: 48DOI 10.1007/s10544-016-0072-0

Echogenic liposomes (ELIP) are phospholipid bilayer ves-icles that encapsulate microbubbles. The microbubbles act ascavitation nuclei that promote sustained cavitation activity inthe presence of an acoustic pressure field, and facilitate thedelivery of therapeutic agents (Datta et al. 2008, 2006;Klegerman et al. 2008; Ramachandran et al. 2006; Shawet al. 2009a, b). At an ultrasound frequency of 120 kHz, rt-PA-loaded ELIP demonstrated a significantly higher clot lysiscompared to unencapsulated rt-PAwith or without pulsed ul-trasound exposure, attributed to the synergistic effects of sta-ble cavitation, drug release, and fibrin targeting of rt-PA-ELIP(Datta et al. 2008; 2006; Holland et al. 2013; Laing et al. 2012;Shaw et al. 2009a, b; Smith et al. 2010; Sutton et al. 2013).

The size distribution of microbubbles is a key determinantof their acoustic response (Goertz et al. 2007). The majority ofmicrobubbles in commercial ultrasound contrast agents(UCA) range from 1 to 10 μm in diameter. Bubbles largerthan 10 μm are rapidly cleared by the capillaries in the lungs(de Jong et al. 1993). Nanobubbles (<1 μm in diameter) ex-hibit poor ultrasound scattering efficiency from 2 to 10 MHzand have low in vivo clearance rates (Palma and Bertolotto1999). Microbubbles excited at twice their resonance frequen-cy are known to undergo stable cavitation preferentially(Bader and Holland 2013). Therefore, the size range ofmicrobubbles can be engineered for a particular applicationby considering their resonant frequency and clearance rates(Feshitan et al. 2009; Shekhar et al. 2013). However, currentmanufacturing processes produce a polydisperse size distribu-tion of microbubbles. Moreover, studies have shown that only<20 % of ELIP manufactured using traditional liposomemanufacturing techniques possess a gas core that is responsiveto ultrasound (Raymond et al. 2014; Kodama et al. 2010).Current techniques for preparation of liposomes requirepost-processing steps, such as freeze thawing, which may ad-versely affect the enzymatic activity of protein drugs encap-sulated in lipids compared to native proteins (Pikal-Clelandet al. 2000). Microfluidic generation allows for fewer post-processing steps, which may preserve the enzymatic activityof a protein-drug payload. Therefore, liposomes preparedusing microfluidic techniques may improve the efficacy ofultrasound-mediated drug delivery. Microfluidic flow focus-ing has been previously reported to manufacture lipid-coatedmicrobubbles with a narrow size distribution for use as UCAs(Kodama et al. 2010;Hettiarachchi et al. 2007; Talu et al.2006;Dhanaliwala et al. 2013; Dixon et al. 2013), and to en-capsulate lipophilic drugs suspended in oil (Shih et al. 2013).However, loading of protein-drugs in phospholipid-basedmicrobubbles using microfluidic flow-focusing has not beenpreviously demonstrated. The objective of this work was todevelop rt-PA-loaded ELIP (μtELIP) with a perfluorocarbongas core using a microfluidic flow focusing technique, and tocharacterize their size distribution, acoustic attenuation, anddrug loading efficiency.

2 Materials and methods

2.1 Materials

A phospholipid mixture containing the phospholipids 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC) and 1,2-dis tearoyl-sn -glycero-3-phosphoethanolamine-N-[amino(polyethylene glycol)-2000] (ammonium salt)(DSPE-PEG2000)(Avanti Polar Lipids, Alabaster, AL,USA) in chloroform in the molar ratio 9:1 was dried at52 °C under argon gas for 1 h. The flask was further driedunder vacuum overnight, rehydrated in 0.9 % saline solution,and sonicated in an ice bath until a clear solution was obtained(~1 h). Propylene glycol and glycerol (Sigma, St. Louis, MO,USA) were added to the lipid solutions for a final concentra-tion of 10 % (v/v) concentration each. A 25 % Pluronic F-127(Sigma, St. Louis, MO, USA) surfactant solution was pre-pared by adding the appropriate amount to 0.9 % saline, leav-ing overnight at 4 °C, and further mixing in an ice-bath using astir-bar to achieve a homogenous distribution. The surfactantsolution was diluted in the lipid mixture to a final concentra-tion of 5 % (v/v). To prevent blockage of the microchannelswith impurities, the saline solution and lipid mixture werefiltered twice through a syringe filter (0.2 μm cellulose acetatefiltration media GE Lifesciences, Pittsburg, PA, USA) prior touse. A 0.5 % (w/v) bovine serum albumin (BSA, Sigma, St.Louis, MO, USA) solution in 0.9 % (w/v) saline was prepared.The drug rt-PA (Genentech, Inc., San Francisco, CA, USA)was suspended in 0.9 % saline to obtain a 1-mg/mL solution.Perfluorocyclobutane gas (C4F8, Specialty Gases of America,Inc., Toledo, Ohio, USA) was used directly from a pressurizedcylinder for microbubble entrainment.

2.2 Microfluidic device fabrication

Microfluidic chips were fabricated in polydimethylsiloxane(PDMS) using the standard soft lithography process. Briefly,a 25-μm high master was formed in SU-8 photoresist(Microchem Corp.). A mixture of PDMS base and curingagent (10:1 ratio) was poured onto the master, degassed for120 min under vacuum, and cured for 4 h on a 60 °C hotplate.Polydimethylsiloxane (PDMS) polymer mixed with a curingagent was poured on the prepared silicon wafer. Once cured,the PDMSwas separated from themaster and inlet/outlet portswere punched with a 14 gauge syringe needle. PDMS wasbonded to a standard glass slide using a hand-hold plasmasurface treater (BD-20 AC, Electro-Technic Products, Inc.).As shown in Fig. 1, the microfluidic devices contained a cen-tral gas channel, inner drug channels and outer lipid channels,which meet just upstream of the orifice. The channel dimen-sions were: 50 μm and 30 μm for the outer and inner liquidchannels respectively, 35 μm for the central gas channel, and10 μm at the orifice.

48 Page 2 of 10 Biomed Microdevices (2016) 18: 48

2.3 Generation of μtELIP

The lipid mixture and rt-PA solutions were loaded into a sy-ringe, placed in a syringe pump, and connected to themicrochip via 1/16″ o.d. tygon tubing. The lipid mixtureand rt-PA were allowed to flow in the outer and innerside channels, respectively. The total combined flow rates oflipid and drug channels ranged from 35 to 115 μL/min, whilea constant rt-PA flow-rate of 15 μL/min was maintained.Octafluorocyclobutane (C4F8) gas was delivered to the centralflow channel of the chip via a high precision pressure regula-tor (Swagelok, Cat. No. KLF1CFH412A20000, Cincinnati,OH, USA) using 1/16″ o.d. tygon tubing.

2.4 Size and number density of μtELIP

The extruded μtELIP were observed using an inverted micro-scope (IX-71, Olympus Inc.) with a 12-bit high-speed CCDcamera (Retiga EXi, QImaging). Particle size and numberdensity were measured using a Coulter counter (BeckmanCoulter Multisizer 4, Beckman Coulter Inc., Brea, CA,USA) fitted with a 30-μm aperture. For the size measurementsthe microbubbles were diluted in a 0.5 % bovine serum albu-min (BSA) solution by a factor of 100 (i.e., 100 μL of lipo-some solution per 10 mL of BSA). The cuvette was coveredwith a lid provided by the manufacturer between measure-ments to prevent evaporation, and gently inverting severaltimes before each measurement.

2.5 Fluorescence imaging

Nile red (Sigma, St. Louis, MO, USA), a highly lipophilicdye, was added to μtELIP to stain the phospholipid shell se-lectively. Nile red was purchased as a solid powder, and dis-solved in acetone at a concentration of 0.5 μM. The dyecalcein (Sigma, St. Louis, Missouri, USA) at a concentrationof 0.1 mM was added to the rt-PA prior to μtELIPmanufacturing. A 10 mM cobalt chloride solution wasused to quench any fluorescence from unencapsulatedrt-PA. The μtELIP were stabilized against motion in a 3Dchemotaxis slide. Fluorescence emissions from calcein andNile red were observed using 495/515 nm and 485/525 nmfilters, respectively.

2.6 Attenuation spectrum measurement

Broadband attenuation of US through a μtELIP samplesuspended in 0.5 % BSA was measured using a through-transmission acoustic spectroscopy system (Raymond et al.2014; Kopechek et al. 2011; Raymond et al. 2013). Briefly,an ultrasound pulser-receiver (Panametrics 5077PR, OlympusNDT, Waltham, MA, USA) was used to generate the excita-tion pulse and amplify the received ultrasound signal over afrequency range of 2 to 25 MHz. The peak rarefactional pres-sure of the acoustic pulse was 31kPA (Raymond et al. 2014).Test samples of μtELIP (n = 3) were placed in an unmodifiedcell-culture cassette (CLINIcell, Mabio, Tourcoing, France)

Fig. 1 a Microfluidic devicedesign showing the inner drugchannels, outer lipid channels,orifice, and the outlet reservoir (b)zoomed-in view of the extrusionorifice, and c the microfluidicdevice

Biomed Microdevices (2016) 18: 48 Page 3 of 10 48

with luer-lock ports to introduce the sample suspension. Theattenuation was calculated as the ratio of received signalstrength with and without liposomes in BSA.

For the attenuation measurements, the liposomes were di-luted in a 0.5 % BSA solution by a factor of 12.5 (i.e., 400 μLof liposome solution in 5 mL of BSA). The cassette wasclosed between measurements to prevent evaporation, andgently inverted several times before each measurement.Measurements with only the diluent, 0.5 % BSA, were per-formed, and subtracted from the attenuation spectrum of theliposome suspensions. All measurements were performed atroom temperature (22.5 ± 0.5 °C).

2.7 Chromogenic assay

A spectrophotometric assay was employed to measure the en-zymatic activity of μtELIP. This assay exploits the reactionbetween a chromogenic substrate (S-2288, Chromogenix,DiaPharma Group, Inc., Westchester, OH, USA) and rt-PA(Smith et al. 2010). Specifically, the chromogenic substrate ishydrolyzed by rt-PA, which results in the production of thechromophore para-nitroaniline (pNA). The enzymatic activitycan be inferred by measuring the change of absorbance in so-lution over time at 405 nm. In this study, enzymatic activity wasreported with respect to that of commercially available rt-PA(Activase®, Genetch, San Francisco, CA, USA). Commercialrt-PA was obtained from the manufacturer in the form of alyophilized powder. The rt-PAwas reconstituted in sterile waterto a concentration of 1 mg/mL and stored at –80 °C until use. Ithas been demonstrated that this procedure preserves the enzy-matic activity of rt-PA for at least 7 years (Shaw et al., 2009a,b). Prior to spectrophotometric measurement, the rt-PA wasthawed and diluted to concentrations of 0.32, 1.58 and3.15 μg/mL in 1 % BSA and stored in disposable cuvettes.Measurements were performed over 5 min at 37 °C using aspectrophotometer (UV-1700, Shimadzu, Japan) equippedwitha temperature controller (TCC-240A, Shimadzu, Japan). Usinglinear regression, an rt-PA standard curve was generated. Next,spectrophotometric measurements were performed for a mix-ture of rt-PA and lipid to assess the effect of the μtELIPmanufacturing process on the activity of rt-PA. Subsequently,μtELIP were diluted to a concentration of 3.25 μg/mL and thert-PA activity assessed.

2.8 Estimation of shell parameters

The shell properties of μtELIP were estimated using a linear-ized acoustic model proposed by de Jong et al. in which theUCA shell is characterized by ad hoc parameters for shellelasticity (Sp in N/m) and shell friction (Sf in kg/s) (de Jongand Hoff 1993). The details of this model have been describedpreviously (Raymond et al. 2014; Goertz et al. 2007). Briefly,this model assumes that the acoustic attenuation of a

population of microbubbles is the sum of the contribution ofindividual microbubbles. Estimates of shell parameters wereobtained by minimizing the sum-squared error difference be-tween the estimated and measured attenuation coefficients.Suspensions of μtELIP may also contain non-echogenic par-ticles that do not contribute to the acoustic attenuation. Thesize distribution of echogenic μtELIP was assumed to be pro-portional to the number of particles determined by Coultercounter measurements. The number density of μtELIP thatcontributed to the attenuation (Nfit) was determined empirical-ly by minimizing the difference between the cumulativedeviations of model predictions from the mean of themeasured response. The coefficient of determination(R2) was calculated to indicate the goodness of fit. The valuesof the shell parameters that lead to the doubling of the errorfunction value were selected as the limits for the confidenceinterval (Raymond et al. 2014).

3 Results

3.1 Particle size and stability

As shown in Fig. 2, the size of the extruded μtELIP decreasedwith increasing lipid flow rates (40–63.5 μL/min) for aconstant drug flow rate of 15 μL/min, and a constantgas pressure of 9–10 psi. Using lipid concentrationsbetween 0.1 and 2 mg/mL, flow rates between 30 and45 μL/min, and gas pressures between 3 and 4 psi weresufficient to produce particles in a size range appropriate foruse in ultrasound-mediated drug-delivery (1–10 μm).However, these particles either aggregated to form larger par-ticles (Fig. 3a) or rapidly contracted and disappeared withinseconds. In contrast, a higher (10 mg/mL) lipid concentrationusing the same flow rates and gas pressures resulted in stablemicrobubbles (Fig. 3b). In the absence of the Pluronic F-127surfactant, the particles demonstrated poor stability lasting onthe order of 10–300 s. The size distribution of μtELIP isshown over a 30-min time interval in Fig. 4. Both number-weighted and volume-weighted distributions peak at a particlediameter of around 5 ± 0.5 μm. The number-weighted sizedistribution also shows the presence of particles smaller than3 μm.

3.2 Broadband attenuation

Figure 5 shows the attenuation spectrum of the μtELIP as afunction of frequency over a 30-min period. The μtELIP spec-trum peaks at 2.2 MHz, corresponding to the particles in the5 ± 0.5 μm size range. Negligible attenuation was observed athigher frequencies, which suggests that the smaller particlesobserved in the size measurements did not contribute signifi-cantly to the acoustic response of μtELIP. The attenuation

48 Page 4 of 10 Biomed Microdevices (2016) 18: 48

remained stable for the first 15 min of the experiment, follow-ed by a decrease in attenuation below 4 MHz after 30 min.

3.3 Estimation of shell parameters

Figure 6 shows the measured attenuation coefficient ofμtELIP as a function of frequency and the theoreticalfit based on the estimated shell parameters. The theoret-ical fit is in good agreement with the measured attenu-ation coefficients (R2 = 0.94). The estimated shell pa-rameters are shown in Table 1 with the estimated shellparameters of Definity® and batch–produced rt-PA-loaded ELIP reported by Raymond et al. (Raymondet al. 2014) Though the shell elasticity parameter (SP)of μtELIP is comparable to Definity®, the shell frictionparameter (SF) is approximately five times higher thanDefinity®. The percentage of echogenic particles in theμtELIP suspension (NFIT/NMEAS*100) was found to beapproximately 35 %.

3.4 Encapsulation of rt-PA

Extruded μtELIP liposomes with the calcein-stained rt-PA(green) and the Nile Red-stained phospholipids (red) areshown in Fig. 7. In the absence of a quenching agent, fluores-cence from calcein was visible both on the μtELIP surface andin the core (Fig. 7a–b). The addition of cobalt chloride, aquenching agent for calcein, resulted in the loss of fluores-cence from calcein on the surface of the μtELIP particles.However, fluorescence from calcein associated with the rt-PA encapsulated within the μtELIP particles was still visible(Fig. 7c).

Figure 8 shows the results of the chromogenic assay per-formed on a mixture of rt-PA and phospholipids extrudedthrough the microfluidic device and the μtELIP particles.The rt-PA and phospholipid mixture were extruded throughthe microfluidic device at the same flow rates as that used forμtELIP generation. A comparison with the rt-PA standardcurve showed good agreement between the expected and mea-sured rate of change of absorbance for the rt-PA in the extrud-ed mixture, indicating that extrusion through the microfluidicdevice itself did not cause degradation of rt-PA (data notshown). The amount of rt-PA unencapsulated in suspen-sion and associated with the shell of μtELIP was mea-sured to be (64.30 ± 3.80 μg/mL), and that associated withthe rt-PA and phospholipid mixture was found to be(148.70 ± 30.80 μg/mL).

(b)

(50 μm)(200 μm)

(a)Fig. 3 Microscopic images ofparticles obtained using (a) a lowlipid concentration (2 mg/mL),and b higher lipid concentration(10 mg/mL)

Fig. 2 Microbubble formation downstream of the orifice showing theeffect of increasing flow rates on particle sizes. The flow rates inimages (a–c) are 40, 50, and 63.5 μL/min respectively. Scale bar: 50 μm

Biomed Microdevices (2016) 18: 48 Page 5 of 10 48

4 Discussion

The use of microfluidic flow focusing to manufacture mono-disperse lipid-based ultrasound contrast agents has been dem-onstrated previously (Hettiarachchi et al. 2007, 2009; Jahn et al.2004; Shih et al. 2013; Talu et al. 2007, 2006). However, thestability of the particles in these studies was variable and rangedfrom 10 min to several hours. In this work, we found thatemploying higher lipid concentrations improved particle stabil-ity. Specifically, a lipid concentration of 10 mg/mL DSPC wasnecessary to achieve particle stability of at least 30 min consis-tently. Although, this concentration is higher than lipid concen-trations used for the microfluidic manufacturing of liposomesreported in literature (0.3–3.5 mg/mL DSPC), this higher lipidconcentration is similar to the concentration used in conven-tional batch manufacturing of ELIP with or without drug load-ing (Huang et al. 2001, 2008; Swanson et al. 2010; Thomsonet al. 2014). It has been demonstrated previously that increasingthe acyl chain length increases the stiffness of microbubbles,

resulting in higher resonance frequency and in vivo stability(Borden et al. 2005). However, increasing the acyl chain lengthmay also result in a higher threshold for microbubble ruptureand for cavitation emissions from microbubbles. Further, it hasbeen shown that changing the main lipid component ofmicrobubbles can alter ligand distribution, binding area, andbound microbubble shape (Kooiman et al. 2014). Our futurestudies will investigate the impact of changing acyl chainlength and lipid components in μtELIP.

The surfactant, Pluronic F-127, combined with a 10% pro-pylene glycol and glycerol water (PG water) added to the lipidmixture was essential to the generation of stable particles.Higher solution bulk viscosity has been shown to decreasebubble coalescence due to a decreased film drainage rate

(a) (b)

(c) (d)

(e) (f)

Fig. 4 Multisizer measurements showing number-weighted (left) and volume-weighted (right) μtELIP size distributions at (a–b) 10 min, c–d 16 min,and e–f 32 min

Fig. 5 Average US attenuation of μtELIP suspended in 0.5 % BSA at22.5 ± 0.5 °C measured after 30 min (N = 3). Error bars representstandard deviations

Fig. 6 Measured attenuation coefficients as a function of frequency(dashed lines) and theoretical fits (solid lines) based on the estimatedshell parameters of the μtELIP. The gray band denotes the 95 %confidence interval for the fit

48 Page 6 of 10 Biomed Microdevices (2016) 18: 48

(Sanadaa et al. 2005; Talu et al. 2006). The higher bulk solu-tion viscosity achieved with the combined use of PG waterand a high concentration lipid solution likely contributed tothe enhanced liposome stability. Addition of the surfactant at a5 % (v/v) concentration effectively reduced particle aggrega-tion post-production and significantly enhanced the numberdensity during particle generation. Similar observations havebeen made in previous studies where different pluronic sur-factants were used for the microfluidic generation of non-lipidor lipid shelled microbubbles (Angilè et al. 2014; Shih et al.2013). The decreased coalescence of particles was attributedto a decrease in the surface tension caused by rapid adsorptionof surfactant molecules on the liposome surface. The additionof a surfactant has also been shown to protect protein confor-mation and functionality (Akers and Defelippis 2000).Moreover, the surfactant pluronic F-127 is a thermoreversiblegel, and may contribute to the greater stability of the μtELIP atphysiological temperatures (Escobar-Chavez et al. 2006).

The number-weighted size distributions of μtELIP overtime show a large number of microparticles with diameterslower than 3μm.However, the volume-weighted distributionsreveal that particles with diameters between 5 ± 0.5 μm ac-count for the majority of the volume fraction of μtELIP.Therefore, liposomes with diameters between 4.5 μm -5.5 μm are expected to be important for therapeutic delivery.These liposomes will also be the dominant contributors to theacoustic backscatter from μtELIP (Gorce et al. 2000). Theresonance frequency of the μtELIP was close to 2 MHz,which may be suitable for transcranial sonothrombolysis ap-plications (Alexandrov et al. 2004). It is noteworthy that thisfrequency may not be able to penetrate the skull for nearly15 % of patients (Wijnhoud et al. 2008). For such patients,sub-megahertz frequencies may be employed (Bouchouxet al. 2012, 2014) at which microbubbles in the 20–50 μmrange are resonant. However, microbubbles larger 7 μm arefiltered by the lungs and thus for cavitation nucleation forenhancement of rt-PA thrombolysis. Resonant bubbles of20–50 μm could be formed by coalescence of smaller bubblesliberated from μtELIP experiencing Bjerkness forces (Baderet al. 2015a, b). Therefore, μtELIP could be tested forsonothrombolysis at sub-megahertz frequencies.

As shown in Fig. 7a–b, fluorescence from calcein inμtELIP particles diluted 50 times in 0.9 % saline indicates astrong association between rt-PA and the lipid shell. The rt-PAattached to the surface of ELIP can potentially enable selectivetherapy through the fibrin-targeting sites on rt-PA, whichcould enhance localized cavitation activity and ultrasound-mediated thrombolysis (Sutton et al. 2013). Cobalt chloridewas added to μtELIP to quench fluorescence from calceinassociated with the surface of the μtELIP or that associatedwith excess unencapsulated rt-PA in solution. As shown inFig. 7c, fluorescence was still observed from the μtELIP in-dicating that rt-PA had been encapsulated within the μtELIP.The amount of rt-PA encapsulated within μtELIP was foundto be about 57 % of the total free rt-PA added to the system.Previous studies have reported a total loading efficiency ofapproximately 50 % for batch-produced rt-PA loaded ELIP,of which about 30 % is associated with the shell surface andthe remaining 20 % with the rt-PA encapsulated within theELIP (Smith et al. 2010). In this work, the chromogenic assaywas performed on suspensions of μtELIP with an intact shelland represents measurement of the rt-PA loaded onto the out-side surface of the μtELIP as well as unencapsulated rt-PA.When free rt-PAwas passed through the microfluidic device,no degradation of the enzymatic activity was observed.Therefore, the amount of rt-PA encapsulated in μtELIP wasestimated from the difference between the concentration of rt-PA injected into the microfluidic device and the concentrationof rt-PA either loaded on the μtELIP or unencapsulated in thesuspension.

Previous studies have shown that compared to a polydis-perse microbubble population, monodisperse microbubblesproduce a sharper peak in the frequency dependent ultrasoundattenuation (Gong et al. 2010) and can be detectedwith greatersensitivity when imaged close to their resonance frequency(Talu et al. 2007). Compared with the attenuation spectra ofvarious commercial ultrasound contrast agents and rt-PA-loaded ELIP, the monodisperse μtELIP show a sharp peakand a narrower bandwidth in the attenuation spectrum(Raymond et al. 2014). No change in the magnitude or thepeak position of the attenuation spectrum is observed for thefirst 15 min of the experiment, indicative of good stability of

Table 1 Estimated numberdensity and shell parameters ofμtELIP compared to Definity®and batch-produced rt-PA-loadedELIP*

Number/mL

Nmeas/109 Nfit/10

9 Sp (N/m) Sf/10−6 (kg/s) R2

μtELIP* 0.02 ± 0.00 0.007 ± 0.00 1.75 ± 0.68 2.71 ± 0.99 0.94

Definity** 9.64 ± 0.90 8.58 ± 1.78 1.76 ± 0.16 0.47 ± 0.05 0.93

rt-PA ELIP** 33.20 ± 8.00 0.41 ± 0.20 3.69 ± 0.76 1.88 ± 0.23 0.90

*Values for μtELIP after correcting for dilution (1:1000)

**Values for Definity® and the batch-produced rt-PA-loaded ELIP at 25 °C computed from Raymond et al.(Raymond et al. 2014) after correcting for dilution (1: 2000 for Definity® and 1:200 for rt-PA-loaded ELIP)

Biomed Microdevices (2016) 18: 48 Page 7 of 10 48

the encapsulated gas. The decrease in the magnitude ofthe attenuation spectrum and the shift in the peak atten-uation from 2.2 MHz to 3.2 MHz after 30 min, is mostlikely due to passive diffusion of the gas from resonant-sizedmicrobubbles.

One of the main limitations of batch-produced ELIP is thatonly a small percentage of particles encapsulate a gas-core.The percentage of μtELIP particles contributing to attenuation(NFIT/NMEAS*100) estimated using the model was found to besignificantly higher (approximately 35 %) than that estimatedfor batch produced rt-PA-loaded ELIP (1.2 %), 22.5 ± 0.5 C(Table 1). Perfluorocarbon gases are known to improve thelifetime of echo contrast agents due to low solubility in aque-ous media and low diffusivity through lipid shells, and areused in commercial contrast agents such as Definity®(Raymond et al. 2014). Studies have shown that the cavitationactivity from rt-PA loaded ELIP with an air core is lower thanDefinity®, suggesting that incorporating perfluorocarbon gaswithin ELIP instead of air may enhance their cavitation activ-ity and thrombolytic efficacy (Bader et al. 2015a, b). The shellfriction parameter (SF) for μtELIP is significantly higher thanDefinity® or batch-produced rt-PA-loaded ELIP (Table 1).The enhanced rigidity of μtELIP prepared using a longer acylchain phospholipid (DSPC), compared to Definity® or batch-produced rt-PA-loaded ELIP, which contain shorter acylchains phospholipids, (DPPC, DPPA, and DPPG) may be re-sponsible for the higher SF parameter of μtELIP. Another keydifference between the μtELIP and Definity® formulations isthe use of PEG molecules of different molecular weights. Theuse of lower molecular weight PEGmolecules combined witha higher phospholipid concentration may cause a tighter pack-ing of the molecules in the μtELIP shell compared toDefinity®, contributing to the higher SF value. The enhancedbulk viscosity due to the use of a significantly higher phos-pholipid concentration (10 mg/mL) in the μtELIP formula-tion, compared to Definity® (0.75 mg/mL) may also contrib-ute to the higher SF parameter. A higher shell friction param-eter will produce increased damping, resulting in lower back-scatter intensity and an increased threshold for nonlinear os-cillations and microbubble rupture compared to Definity®.Although many models of microbubble oscillation have been

Fig. 7 a μtELIP diluted in 0.9 % saline without cobalt chloridequenching demonstrating the presence of adsorbed rt-PA on the μtELIPexterior surface (b) Fluorescent staining of a μtELIP showing the lipidlayer in red and the calcein associated with encapsulated rt-PA in greenwithout cobalt chloride (c) and after addition of cobalt chloride to quenchfluorescence from rt-PA on the exterior of the liposomes

Fig. 8 The amount of rt-PA associated with the rt-PA and lipid mixtureextruded through the microfluidic device (148.70 ± 30.80 μg/mL)comparedwith that associated with the μtELIP shell (64.30 ± 3.80 μg /mL)

48 Page 8 of 10 Biomed Microdevices (2016) 18: 48

reported (Faez et al. 2013), the model from de Jong et al. (deJong andHoff 1993) was chosen.When linear approximationsare employed at low acoustic pressures, the results of thesemodels are equivalent (Kumar and Sarkar 2015).

A significant limitation of microfluidic manufacturing ofliposomes is the low rate of particle generation compared tobatch processes. In this work, the number density of μtELIPgenerated was about 2.4 × 108 liposomes/mL, significantlylower than that reported for batch produced rt-PA-loadedELIP or Definity® (1010 bubbles/mL). Awide range of parti-cle concentrations has been reported in literature for studies onsonothrombolysis. Our group has demonstrated thrombolysisin vitro for approximately 1 × 106 microbubbles/ml (Baderet al. 2015a, b), two orders of magnitude lower than the lipo-some concentration achieved in this work. Shih et al. havereported microbubble generation rates as high as 3 × 106 pers (Shih et al. 2013). The number density of the micbubblespost-production was not investigated over time by these in-vestigators. Device multiplexing is a possible solution to en-hance the limited μtELIP generation rate (Fabiilli et al. 2014).Another limitation of this work is that the measurements wereperformed at room temperature. Additional experiments todetermine the physical and acoustic properties of the μtELIPat physiological conditions are required. In this work, the par-ticles were not separated from unencapsulated rt-PA post-pro-duction. The effect of separation steps such as centrifugationor dialysis on the physical properties of the particles alsoneeds to be studied further. In addition, the effect of bloodviscosity on the threshold of stable cavitation nucleation fromμtELIP should be assessed (Helfield et al. 2016).

5 Conclusions

In this work, microfluidic devices were used to generate mono-disperse rt-PA loaded microbubbles in a clinically-relevantsize-range. The gas-fraction of the μtELIP was significantlyhigher than that reported previously for batch-produced rt-PA-loaded ELIP. The temporal stability of the particles wasat least 30 min, with the use of a high lipid concentration of10 mg/mL, 5(v/v) % surfactant pluronic F-127, and 10 % PGwater solution in 0.9 % saline. Continued investigation of thephysicochemical and acoustomechanical properties, efficacy,safety and stability of the μtELIP over time, for ultrasound-mediated thrombolysis may help in the development of anultrasound-mediated thrombolytic therapy.

Acknowledgments The authors thank Jason Raymond, JonathanKopachek, Kevin Haworth, Xiao Wang, and the past and present mem-bers of the Image-guided Ultrasound Therapeutics Laboratories forhelpful discussions. Funding for this work was provided by theNational Institutes of Health (K02-NS052653, R01-NS047603 andP50-NS044283) and the Interdisciplinary Faculty Grant sponsored by theUniversity Research Council at the University of Cincinnati.

References

M. J. Akers and M. R. Defelippis, in Pharmaceutical FormulationDevelopment of Peptides and Proteins, edited by S. Frokjaer andL. Hovgaard (Taylor and Francis, Philadelphia, 2000), pp. 145–177.

A. V. Alexandrov, J. C. Grotta, Neurology 59(6), 862–867 (2002)A. V. Alexandrov, A. W. Wojner, J. C. Grotta, J. Neuroimaging 14(2),

108–112 (2004)F. E. Angilè, K. B. Vargo, C. M. Sehgal, D. A. Hammer, D. Lee,

Langmuir 30(42), 12610–12618 (2014)K. B. Bader, C. K. Holland, Phys. Med. Biol. 58(1), 127–144 (2013)K. B. Bader, G. Bouchoux, T. Peng, M. E. Klegerman, D. D. McPherson,

C. K. Holland, J. Thromb. Thrombolysis 40(2), 144–155 (2015a)K. B. Bader, M. J. Gruber, C. K. Holland, Ultrasound Med. Biol. 41(1),

187–196 (2015b)M. A. Borden, D. E. Kruse, C. F. Caskey, S. Zhao, P. A. Dayton, K. W.

Ferrara, IEEE Trans. Ultrason. Ferroelectr. Freq. Control 52(11),1992–2002 (2005)

G. Bouchoux, K. B. Bader, J. J. Korfhagen, J. L. Raymond, R.Shivashankar, T. A. Abruzzo, C. K. Holland, Phys. Med. Biol.57(23), 8005–8022 (2012)

G. Bouchoux, R. Shivashankar, T. A. Abruzzo, C. K. Holland,Ultrasound Med. Biol. 40(6), 1154–1166 (2014)

J. V. Braaten, R. A. Goss, C.W. Francis, Thromb. Haemost. 78(3), 1063–1068 (1997)

W. S. Burgin, M. Malkoff, R. A. Felberg, A. M. Demchuk, I. Christou, J.C. Grotta, A. V. Alexandrov, Stroke 31(5), 1128–1132 (2000)

S. Datta, C.-C. Coussios, L. E. McAdory, J. Tan, T. Porter, G. DeCourten-Myers, C. K. Holland, Ultrasound Med. Biol. 32(8),1257–1267 (2006)

S. Datta, C.-C. Coussios, A. Y. Ammi, T. D.Mast, G.M. de Courten-Myers,C. K. Holland, Ultrasound Med. Biol. 34(9), 1421–1433 (2008)

N. de Jong, L. Hoff, Ultrasonics 31(3), 175–181 (1993)N. de Jong, F. J. ten Cate, W. B. Vletter, J. R. Roelandt, Ultrasound Med.

Biol. 19(4), 279–288 (1993)A. H. Dhanaliwala, J. L. Chen, S. Wang, J. A. Hossack, Microfluid.

Nanofluid. 14(3–4), 457–467 (2013)A. J. Dixon, A. H. Dhanaliwala, J. L. Chen, J. A. Hossack, Ultrasound

Med. Biol. 39(7), 1267–1276 (2013)J. J. Escobar-Chavez, M. Lopez-Cervantes, A. Naik, Y. N. Kalia, D.

Quintanar-Guerrero, and A. Ganem-Quintanar, J. Pharm Pharm.Sci.:Publ. Can. Soc. Pharm. Sci., Societe canadienne des sciencespharmaceutiques 9 (3), 339–358 (2006).

E. C. Everbach, C.W. Francis, UltrasoundMed. Biol. 26, 1153–1160 (2000)M. L. Fabiilli, J. Silpe, C. Rush, D. Lemmerhirt, E. Tang, G. Vasey, and

O. D. Kripfgans, IEEE Int. Ultrason Symp. Proc. (2014).T. Faez, M. Emmer, K. Kooiman, M. Versluis, A. van der Steen, N. de

Jong, IEEE Trans. Ultrason. Ferroelectr. Freq. Control 60(1), 7–20(2013)

J. A. Feshitan, C. C. Chen, J. J. Kwan, M. A. Borden, J. Colloid InterfaceSci. 329(2), 316–324 (2009)

H. G. Flynn, J. Acoust. Soc. Am. 72(6), 1926–1932 (1982)D. E. Goertz, N. de Jong, A. F. van der Steen, Ultrasound Med. Biol.

33(9), 1376–1388 (2007)Y. Gong,M. Cabodi, T. Porter, Bubble Sci. Eng. Technol. 2(2), 41–47 (2010)J. M. Gorce, M. Arditi, M. Schneider, Investig. Radiol. 35(11), 661–671

(2000)W. Hacke, G. Donnan, C. Fieschi, M. Kaste, R. von Kummer, J. P.

Broderick, T. Brott, M. Frankel, J. C. Grotta, E. C. Haley Jr., T.Kwiatkowski, S. R. Levine, C. Lewandowski, M. Lu, P. Lyden, J.R.Marler, S. Patel, B. C. Tilley, G. Albers, E. Bluhmki, M.Wilhelm,S. Hamilton, Lancet 363(9411), 768–774 (2004)

B. Helfield, J. J. Black, B. Qin, J. Pacella, X. Chen, F. S. Villanueva,Ultrasound Med. Biol. 42(3), 782–794 (2016)

Biomed Microdevices (2016) 18: 48 Page 9 of 10 48

K. Hettiarachchi, E. Talu,M. L. Longo, P. A. Dayton, A. P. Lee, Lab Chip7(4), 463–468 (2007)

K. Hettiarachchi, S. Zhang, S. Feingold, A. P. Lee, P. A. Dayton,Biotechnol. Prog. 25(4), 938–945 (2009)

C. K. Holland, S. S. Vaidya, C.-C. Coussios, G. J. Shaw, J. Acoust. Soc.Am. 112(5), 2370 (2002)

C. K. Holland, J. T. Sutton, N. M. Ivancevich, S. R. Perrin Jr., D. C. Vela,J. Ultrasound Med. 32(S7) (2013)

S.-L. Huang, A. J. Hamilton, A. Nagaraj, S. D. Tiukinhoy, M. E.Klegerman, D. D. McPherson, R. C. Macdonald, J. Pharm. Sci.90(12), 1917–1926 (2001)

S.-L. Huang, D. D.McPherson, R. C.MacDonald, UltrasoundMed. Biol.34(8), 1272–1280 (2008)

A. Jahn, W. N. Vreeland, M. Gaitan, L. E. Locascio, J. Am. Chem. Soc.126(9), 2674–2675 (2004)

M. E. Klegerman, Y. Zou, D. D.Mcpherson, J. Liposome Res. 18(2), 95–112 (2008)

T. Kodama, N. Tomita, S. Horie, N. Sax, H. Iwasaki, R. Suzuki, K.Maruyama, S. Mori, F. Manabu, J. Electron Microsc. 59(3), 187–196 (2010)

K. Kooiman, T. J. A. Kokhuis, T. van Rooij, I. Skachkov, A. Nigg, J. G.Bosch, A. F. W. van der Steen, W. A. van Cappellen, N. de Jong,Eur. J. Lipid Sci. Technol. 116(9), 1217–1227 (2014)

J. A. Kopechek, K. J. Haworth, J. L. Raymond, T. Douglas Mast, S. R.Perrin, M. E. Klegerman, S. Huang, T. M. Porter, D. D. McPherson,C. K. Holland, J. Acoust. Soc. Am. 130(5), 3472–3481 (2011)

K. N. Kumar, K. Sarkar, J. Acoust. Soc. Am. 138(2), 624–634 (2015)S. T. Laing, M. R. Moody, H. Kim, B. Smulevitz, S.-L. Huang, C. K.

Holland, D. D. McPherson, M. E. Klegerman, Thromb. Res. 130(4),629–635 (2012)

C. G. Lauer, R. Burge, D. B. Tang, B. G. Bass, E. R. Gomez, B. M.Alving, Circulation 86(4), 1257–1264 (1992)

L. D. Palma and M. Bertolotto, Eur. Radiol. 9 (0), S338-S342 (1999).K. A. Pikal-Cleland, N. Rodriguez-Hornedo, G. L. Amidon, J. F.

Carpenter, Arch. Biochem. Biophys. 384(2), 398–406 (2000)S. Ramachandran, A. P. Quist, S. Kumar, R. Lal, Langmuir 22(19), 8156–

8162 (2006)

J. L. Raymond, K. J. Haworth, K. B. Bader, K. Radhakrishnan, S.-L.Huang, D. D. McPherson, and C. K. Holland, presented at the14th World Congress of Ultrasound in Medicine and Biology, SaoPaulo, Brasil, (2013) (unpublished)

J. L. Raymond, K. J. Haworth, K. B. Bader, K. Radhakrishnan , J. K.Griffin, S.-L. Huang, D. D. McPherson, and C. K. Holland,Ultrasound Med. Biol. 40 (2), 410–421 (2014).

T. Sanadaa, M. Watanabea, T. Fukanob, Chem. Eng. Sci. 60(19), 5372–5384 (2005)

G. J. Shaw, J.M.Meunier, C. J. Lindsell, C. K. Holland, UltrasoundMed.Biol. 34(11), 1783–1792 (2008)

G. J. Shaw, J. M. Meunier, S.-L. Huang, C. J. Lindsell, D. D. McPherson,C. K. Holland, Thromb. Res. 124(3), 306–310 (2009a)

G. J. Shaw, M. Sperling, and J. M. Meunier, BMC research notes 2, 117(2009b)

H. Shekhar, J. J. Rychak, M. M. Doyley, Med. Phys. 40(8), 082903(2013)

R. Shih, D. Bardin, T. D. Martz, P. S. Sheeran, P. A. Dayton, A. P. Lee,Lab Chip 13(24), 4816–4826 (2013)

D. A. B. Smith, S. S. Vaidya, J. A. Kopechek, S.-L. Huang, M. E.Klegerman, D. D. McPherson, C. K. Holland, Ultrasound Med.Biol. 36(1), 145–157 (2010)

V. Suchkova, F. N. Siddiqi, E. L. Carstensen, D. Dalecki, S. Child, C. W.Francis, Circulation 98(10), 1030–1035 (1998)

J. T. Sutton, K. J. Haworth, G. Pyne-Geithman, C. K. Holland, ExpertOpin. Drug Deliv. 10(5), 573–592 (2013)

E. J. Swanson, V. Mohan, J. Kheir, B. M. A., Langmuir 26(20), 15726–15729 (2010)

E. Talu, M. M. Lozano, R. L. Powell, P. A. Dayton, M. L. Longo,Langmuir 22(23), 9487–9490 (2006)

E. Talu, K. Hettiarachchi, S. Zhao, R. L. Powell, A. P. Lee, M. L. Longo,P. A. Dayton, Mol. Imaging 6(6), 384–392 (2007)

L. M. Thomson, B. D. Polizzotti, F. X. McGowan, and J. N. Kheir, J. VisExp: JoVE (87) (2014)

A. D. Wijnhoud, M. Franckena, A. van der Lugt, P. J. Koudstaal, E. D.Dippel, Ultrasound Med. Biol. 34(6), 923–929 (2008)

48 Page 10 of 10 Biomed Microdevices (2016) 18: 48