Magnetic Microbubble Mediated Chemo-Sonodynamic Therapy using a Combined Magnetic-Acoustic Device

: Recent pre-clinical studies have demonstrated the potential of combining chemotherapy and sonodynamic therapy for the treatment of pancreatic cancer. Oxygen-loaded magnetic microbubbles have been explored as a targeted delivery vehicle for this application. Despite preliminary positive results, a previous study identified a significant practical challenge regarding the co-alignment of the magnetic and ultrasound fields. The aim of this study was to determine whether this challenge could be addressed through the use of a magnetic-acoustic device (MAD) combining a magnetic array and ultrasound transducer in a single unit, to simultaneously concentrate and activate the microbubbles at the target site. in vitro experiments were performed in tissue phantoms and followed by in vivo treatment of xenograft pancreatic cancer (BxPC-3) tumours in a murine model. In vitro , a 1.4-fold (p<0.01) increase in the deposition of a model therapeutic payload within the phantom was achieved using the MAD compared to separate magnetic and ultrasound devices. In vivo, tumours treated with the MAD had a 9% smaller mean volume 8 days after treatment, while tumours treated with separate devices or microbubbles alone were respectively 45% and 112% larger. This substantial and sustained decrease in tumour volume suggests that the proposed drug delivery approach has the potential to be an effective neo-adjuvant therapy for pancreatic cancer patients.


Introduction
Microbubbles (MB) and ultrasound (US) are in routine clinical use for diagnostic imaging, and are being actively investigated for a range of therapeutic applications including in recent clinical trials for cancer treatment [1,2]. MBs consist of gas cavities 1-in diameter stabilised by a surfactant, lipid and/or polymer coating [3]. Due to the compressibility of the gas core, MBs undergo volumetric oscillations when exposed to a US field, and the resulting acoustic scattering may be used to enhance the contrast between blood vessels and the surrounding tissue in US images [4]. The use of sufficiently high acoustic pressures can lead to MB fragmentation and the dispersion of their coating material [5]. As the MB coating can be loaded with drugs through the addition of single therapeutic molecules, drug-loaded liposomes or other nanoparticles, the ability to non-invasively trigger the release of therapeutics using US makes MBs an attractive drug delivery platform [6 8]. Additionally, US targeted MB destruction (UTMD) has been associated with increased payload distribution [9] and sonoporation [10] in tissue. Recent reviews of UTMD have summarised the use of this method for the delivery of chemotherapies [11,12], genes and thrombolytic drugs [13].
Several complementary targeting techniques have been developed to increase treatment localisation and so reduce potential side effects resulting from systemic administration. For example, MBs have been functionalised by attachment of antibodies enabling their binding to target tissues [14]. The short half-life of MBs (<5 minutes in circulation) has presented a challenge 3 for this method, and so acoustic radiation force has been investigated as a way to concentrate MBs and increase their binding to target sites [14,15]. Another approach has been to incorporate magnetic material into MBs and accumulate them in a target region using an external magnetic field [16]. This method was used in a recent study by the authors in which magnetically responsive oxygen MBs (MagO2MBs) were used to deliver a combination of an antimetabolite drug (5fluouracil) and sonodynamic therapy (SDT) to pancreatic tumours [17]. While the combination of magnetic and US fields demonstrated an improved tumour growth delay and increased apoptotic cell signalling compared to the treatment with US only, the simultaneous application and alignment of magnetic and US fields produced by separate devices represented a significant practical challenge in vivo. This problem is particularly acute in small animal models due to space constraints and may greatly limit the potential synergistic benefits of magnetic-acoustic targeting.
In the present study, this was addressed by using a prototype probe enabling co-aligned US and magnetic fields to be applied simultaneously [18]. The aim of the experiments described in the next section was to determine the effect upon drug delivery first in vitro in a tissue mimicking phantom and subsequently investigate the effect on treatment efficacy in vivo in a murine pancreatic cancer model.

Materials and methods
This section details the materials and suppliers used, followed by a description of the Polar Lipids (Alabaster, Alabama, USA). All reagents and equipment used for magnetic oxygen MB production were as previously described in [17] with the exception of the superparamagnetic iron oxide nanoparticles (IONPs) and therapeutic agent (i.e. 8). The IONPs (50 nm hydrodynamic diameter) were custom-conjugated by Ocean NanoTech (San Diego, CA, USA).
The design and calibration of the magnetic-acoustic device (MAD) having co-aligned acoustic and magnetic fields is described in [18].

Preparation of chemo/sonodynamic therapy complex (biotin-RB-Gem)
In our previous work, we investigated the combination of 5-fluouracil (chemotherapy) and Rose-Bengal (sonodynamic therapy). As gemcitabine has been reported as the antimetabolite therapy of choice for pancreatic cancer, superseding treatments with 5-fluouracil [20], recent work by the authors has also presented microbubbles loaded separately with gemcitabine and Rose Bengal [21] and this was the combination selected for the present study.

6
To enable the loading of both the RB and Gem on the MB surface, a novel therapeutic was formulated with a single biotin anchor connected to both drugs. The methods used to prepare this complex are described in this section following Scheme 1.

In vitro treatment of BxPC-3 and Mia-PaCa-2 cells with biotin-RB-Gem and Gemcitabine
The human primary pancreatic adenocarcinoma cell line BxPC-3 was maintained in RPMI 1640 medium which was supplemented with 100 U/mL penicillin, 100 mg/mL streptomycin, and
Compound 8 or 9 (1 mL, 5.2 mM in PBS with 0.5% (v/v) DMSO) was loaded onto the MagMBs following the method reported in [17]. All microbubbles evaluated in this manuscript were washed three times by centrifugation to remove excess material from the suspension. In vitro, MagMB characterisation was completed using 9 to prevent wastage of the more difficult to synthesise 8.
Drug-loaded MagMBs were kept in reduced light conditions and on ice prior to use and are referred to as MagO2MB-RB-Gem ( Fig. 1) or MagO2MB-RB depending on which drug product was used. If the samples were not sparged with oxygen, they are referred to as MagMB-RB-Gem and MagMB-RB.

Characterisation of MB size and concentration
The MBs were characterised for their size and concentration following analysis of optical microscope images using a custom MATLAB script [23]. diluted (1:20 v/v) sample in PBS was loaded onto a haemocytometer and imaged 30 times using an optical microscope fitted with a 40X objective, leading to approximately 1800 microbubbles examined per microbubble batch.

Characterisation of MB drug loading
The drug loading of MBs was investigated for both 8 and 9 using UV-Vis spectroscopy. As the ratio of RB to Gem in each molecule of 8 is 1:1, the concentration of both Gem and RB attached to MagO2MB-RB-Gem can be determined from the RB absorbance at 559 nm, using a previously constructed calibration graph. Similarly, for 9, the RB concentration of MagO2MB-RB was determined based on the RB absorbance at 559 nm. For each MB batch prepared, a 50 of MBs was sonicated (40 kHz ultrasound bath) for 5 seconds before diluting it (1:100 v/v) and recording the absorbance at 559 nm using a plate-reader.

Characterisation of MB iron loading
As DBPC-IONPs were incorporated within the MB coating, the functionalisation of MBs with drug products on its outer surface is unlikely to affect the iron loading of the MBs. The iron content of MBs was therefore measured without the addition of drugs to prevent wastage of synthesised ligands, and was determined by ICP-OES measurements of samples diluted in 2% nitric acid at a wavelength of 238 nm. 12 2.8 Production of singlet oxygen from MagO2MB-RB exposed to US The production of singlet oxygen ( 1 O2) from activated RB exposed to US was determined using SOSG. A sample of 9 mL degassed PBS, ± 5x10 7 MB/mL, ± 541 -RB, and 1.25 SOSG was exposed to 1.17 MHz, 0.70 MPa peak negative pressure, 30% duty cycle (DC), 100 Hz pulse repetition frequency (PRF) US for 3.5 minutes. Sample exposure was undertaken using The MAD was designed as described by Barnsley et al. and assembled as shown in Fig. 2 [18]. Briefly, the magnetic body consisted of N52 grade NdFeB permanent magnet material whose geometry was optimized to have a maximum magnetic field of 0.2 T at a distance of 10 mm from integrated ultrasonic element with a focal distance also of 10 mm provided a pressure field that spatially overlapped with the magnetic field peak, with sufficient amplitude to cause inertial cavitation of MBs used in this study. An aluminium-bodied copy of the MAD ( aMAD ) was produced to provide an US-only control for in vitro and in vivo experimentation. In order to span the gap between the US element and the delivery site of interest in the present work, a coupling cone (Fig. 2) was cast from paraffin wax and secured with US gel. The cone material was chosen for its ease of casting and minimal transmission loss in the 1 MHz frequency range as determined by through-transmission measurements. Since the acoustic boundary 14 conditions for this configuration were different from those used in the initial characterization [18], both the MAD and the aluminium copy were recalibrated (Fig. S4).

Drug delivery comparison in agar between devices
Drug delivery was quantified in vitro by flowing MagO2MB-RB through a tissue mimicking agar phantom (Fig. 3). The phantom body was formed within a Delrin frame filled with 1.25% agar gel. For acoustic treatments (MAD or aMAD), the device was held in place with a lab clamp, and the cone tip was coupled to the phantom face with water. Acoustic drive pulses (3000 cycles of 1.17 MHz, 30% duty cycle) were provided so that the peak negative pressure 10 mm in front of the device would be 0.7 MPa. This level matched the spatial peak value from a separate device that had been used in previous SDT experiments [17,19].
Ultrasonic emissions from the channel were recorded using a single PCD. Signal conditioning and post-processing procedures were performed using the same procedures and instrumentation as in Section 2.8. Initial positioning of the agar flow channel and the acoustic instrumentation was guided by a crosshair laser to ensure proper alignment. Data were collected for six groups as indicated in Table 1, with three separate phantoms tested per group. The MAD was used for: (1) its co-

Treatment of xenograft ectopic BxPC-3 tumours in SCID mice
All animals employed in this study were treated in accordance with the licenced procedures

Synthesis of biotin-RB-Gem (8) and its efficacy in pancreatic cancer cells
To enable both the SDT sensitiser RB and antimetabolite Gem to be conjugated to the MB surface, MBs were surface functionalised with avidin and a tripodal ligand was designed to have a single biotin anchor connected to both RB and Gem (8). To synthesise 8, the Nhydroxysuccinimide ester of biotin (1) was first reacted with tris(2-aminoethyl)amine (2) in a 1:1 molar reaction to encourage only one of the primary amines on 2 to form an amide bond with 1.
The resulting product 3, was then reacted with disuccinimidyl suberate (4) in a 1:2 molar ratio forming amide bonds with the remaining two primary amine residues of 3, yielding compound 5 that also contained two pendant active esters. The active esters of 5 were reacted in turn with gemcitabine (6) and amine derivatised Rose Bengal (7), generating ester and amide linkages respectively with 5 to form target compound 8. The structure of 8 was characterised using 1 H and 13 C NMR spectroscopy and positive electrospray mass spectroscopy (Figs S1-S3) of Gem to enable its activation by deoxycytidine kinase mediated phosphorylation, rapid cleavage of RB from 8 is less important, as the mechanism of action for SDT does not require the sensitiser to bind to a receptor or be metabolised for reactive oxygen species to be generated. 7.5±4.0 x 10 8 MB/mL (p<0.01) and the mean diameter to 1.9±0.4 (Fig. 5). A comparison of the size distributions from before and after conjugation with 9 suggests that the loading and subsequent washing processes removes the smallest MBs (Fig. 5). However, the 1 -8 m diameter MBs appear to stable during this process and the mean hydrodynamic diameter of the population was not significantly affected by the washing procedure.
Detailed characterisation of the MB stability and their magnetic and acoustic properties prior to drug loading was reported in [24]. The iron content measured in the present study was 0.07 pg iron per microbubble after three centrifuge steps. This is higher than the previously reported values of 0.025 pg iron / MB [24] but was associated with manipulation variations in the resuspension of microbubbles during cleaning. This loading was calculated to equate to approximately 10% coverage of the microbubbles (supporting information) and their response to a magnetic field was confirmed visually (Fig. S5).
Analysis of the power of acoustic emissions over time from the MBs indicated that the surface addition of drugs on microbubbles significantly lengthened the time over which an increase in acoustic emissions was recorded compared to magnetic microbubbles without drug (Fig. S6).
These  MPa peak negative pressure, 30% duty cycle, 100 Hz pulse repetition frequency for 3.5 minutes.
The performance of the MAD compared to the use of two separate devices and the contributions of US and of the magnetic field individually were assessed in an agar phantom containing a cylindrical flow channel (Fig. 3). The concentration of RB delivered was determined of MagMB-RB through it while exposed to one of the device configurations. Fig. 7 shows the concentration of RB delivered for the different groups considered, and when US was used, the associated cavitation activity is provided. A significantly higher quantity of RB was delivered (p < 0.01) using the MAD compared to all other groups; more specifically a 1. 6  could be associated with residual RB in the suspension diffusing across pores in the agar gel [27] and was found to be significant compared to the untreated group (p < 0.01). Individually, US and Mag fields enhanced delivery to a similar degree, in agreement with previous results from Stride et al. [28].

In vivo results
To test the utility of the MAD as a platform for the delivery of combined magnetic and US fields in vivo, SCID mice were implanted with ectopic human pancreatic BxPC-3 tumours and randomly distributed into 4 groups for treatment as described in Table 2. The results in Fig. 8 indicate a 37% reduction in tumour volume relative to the pre-treatment volume 4 days after the initial treatment for animals treated with MagO2MB-RB-Gem and the MAD, compared to a 9% reduction when combined magnetic and US fields were simultaneously delivered using separate probes (U + M). This difference was maintained for four days, and 12 days following the initial treatment, tumours treated with MagO2MB-RB-Gem and the MAD were still 9% smaller than their pretreatment volume and 54% smaller than tumours treated with MagO2MB-RB-Gem and separate probes. Beyond day 12, all groups showed tumour growth. The MAD group showed the least, but there was no statistically significant difference from the separate probe group. This indicates that further investigation of the treatment schedule with larger groups is warranted.
The observed differences between treatments performed with the MAD and physically separate but simultaneous magnetic and ultrasonic field generating devices illustrate the importance of field alignment. When co-aligned as in the MAD, the tendency of bubbles to be pushed away from the US focus by radiation force is counteracted by the pulling force of the magnet. This effect, which helps maintain a population of bubbles in the US beam, is weaker and potentially becomes detrimental to bubble availability in the US focus when the ultrasonic and magnetic fields are separated by a large angle, as with the separate device tests.
also well tolerated, as the animals remained healthy and exhibited no weight loss over the duration of the study (Fig. S10). Scaling of the MAD to human length scales is demonstrated in [18].

Limitations
Although the results shown in Figs. 8 and S10 are encouraging, there are several limitations of the work that need to be discussed. First, the MAD used in this study was designed for small animal experiments but, as discussed in [18], a compromise had to be made between peak magnetic force and acoustic pressure at the focus. This will be overcome in future development of the device, which will also include treatment monitoring capability. Second, it was not possible with the equipment available to assess tumour vascularity prior to treatment. This is, however, likely to be an important predictor of therapeutic response as it determines the quantity of microbubbles entering the target volume. Whilst ultrasound and microbubbles are able to improve the distance to which therapeutic material is delivered from the nearest blood vessel, they cannot compensate entirely for poor perfusion. This then contributes to the variance in tumour volumes which was notably higher in the results for the groups without magnetic field (Fig. 8, S7, S9).
Finally, residual RB and Gem in the microbubble suspension, suggested by the in vitro results, could also explain the decreased tumour size observed in subjects receiving microbubbles only, compared to no treatment. Further development of the therapeutic microbubbles will focus on product loading and cleaning protocols to minimise off-target accumulation and undesired sideeffects.

Conclusions
A novel therapeutic incorporating the antimetabolite Gemcitabine, the sensitiser Rose Bengal and the attachment ligand biotin was synthesised and enabled the simultaneous loading of both drugs onto oxygen-filled magnetic microbubbles. The co-aligned application of magnetic and ultrasound fields to the target region using the MAD produced an increase in drug deposition in vitro and tumour response in vivo compared to the application of both fields using two separate devices. These results indicate the importance of co-alignment of the magnetic and ultrasound fields and provide further supporting evidence for the potential use this approach to downstage pancreatic tumours in cancer patients with borderline resectable lesions, enabling them to undergo surgery.