Thermal Sensitive Liposomes Improve Delivery of Boronated Agents for Boron Neutron Capture Therapy

Abstract

Boron neutron capture therapy (BNCT) has the potential to become a viable cancer treatment modality, but its clinical translation requires sufficient tumor boron delivery while minimizing nonspecific accumulation. Thermal sensitive liposomes (TSLs) are designed to have a stable drug payload at physiological temperatures but engineered to have high permeability under mild hyperthermia. We found that TSLs improved the tumor-specific delivery of boronophenylalanine (BPA) and boronated 2-nitroimidazole derivative B-381 in D54 glioma cells. Uniquely, the 2-nitroimidazole moiety extended the tumor retention of boron content compared to BPA. This is the first study to show the delivery of boronated compounds using TSLs for BNCT, and these results will provide the basis of future clinical trials using TSLs for BNCT.

Boron neutron capture therapy (BNCT) is an emerging cancer treatment modality with distinct advantages over radiation or chemotherapy. In contrast to chemotherapeutics, traditional BNCT agents have minimal systemic cytotoxicity. Unlike traditional radiation therapies, BNCT utilizes a non-ionizing neutron beam. When a boron-10 (¹⁰B) atom captures a neutron, the subsequent nuclear fission reaction generates an alpha particle and lithium ion (1–4). These high energy particles create a localized cytotoxic effect due to the short path length (5–10 microns) of the aforementioned particles (5). In contrast to x-ray, gamma-ray and charged-particle based therapies, the neutron beam is non-ionizing and thereby mitigates off-target cytotoxicity to surrounding normal tissues. Only tissues containing ¹⁰B will have significant cytotoxicity following neutron irradiation. However, selective boron tumor delivery remains one of the biggest hurdles for translation of BNCT into clinical practice.

The most extensively evaluated agent in BNCT clinical trials is 4-borono-L-phenylalanine (BPA), commonly administered intravenously as a BPA fructose adduct (BPA-f). This boronated phenylalanine derivative has minimal systemic cytotoxicity but has limited ability to accumulate preferentially in a tumor (usually a tumor/normal tissue ratio < 3) (6–11). The hurdle of optimal tumor drug delivery has long been recognized in the BNCT field and numerous strategies to improve delivery such as nanoparticles, liposomes, and monoclonal antibodies have shown promise (6). Our laboratory has recently published a metabolic approach to target the hypoxic (and often therapy resistant) glioma tumor microenvironment with a boronated 2-nitroimidazole derivative (B-381) (12). In a hypoxic microenvironment (pO₂ of 1–2%) 2-nitroimidazole derivatives form intracellular protein conjugates leading to their preferential accumulation and retention (13, 14). B-381 represents a new class of BNCT agents in which their tumor selectivity arises from the unique hypoxic metabolism in tumors. Once 2-nitroimidazole derivatives enter a hypoxic microenvironment, they are reduced and subsequently form covalently bonded intracellular protein conjugates, leading to their specific tumor accumulation and longer retention compared to phenylalanine derivatives like BPA (12).

Despite the development of liposomal formulations for BNCT (15), to date, the efficacy of thermal sensitive liposomes (TSLs) have not been evaluated in the BNCT field. TSLs are designed to have a stable drug payload at physiological temperature (37 °C) but engineered to have high permeability under mild hyperthermia (41–43°C). Administration of a TSL will enhance drug delivery to the hyperthermic region due to increased blood flow, improved vascular permeability and localized drug release at the tumor site (16). Mild hyperthermia is often achieved in the clinical setting using either high-intensity focused ultrasound (HIFU) or radiofrequency energy (17, 18). ThermoDox (Celsion), a TSL loaded with doxorubicin is one of the more thoroughly studied systems and is currently in Phase III clinical trials for non-resectable hepatocellular carcinoma (19, 20).

We hypothesized that a TSL containing BPA-f will have superior drug delivery compared to the free drug. Furthermore, we hypothesized that delivery of previously reported B-381 would have superior long-term tumor retention compared to BPA-f due to the unique intracellular protein conjugates 2-nitroimidazole derivatives form in a hypoxic microenvironment.

To determine the efficacy of TSLs as a novel delivery vehicle for BNCT drugs, both TSL and non-TSLs were developed. The TSL permeability and subsequent drug release is highest under mild hyperthermia (41–43°C) because the major component of the liposome formulation DPPC has a phase transition temperature (T_m) of 41.4 °C (15). Lipids with higher phase T_m such as DSPC (T_m = 54.9 °C) are added to minimize baseline release at physiological temperature from TSLs, while DSPE-PEG₂₀₀₀ increases blood circulation time of TSLs by decreasing reticuloendothelial system uptake (15).

Based on previously described TSL formulations for doxorubicin, our initial TSL formulation contained an 80:15:5 molar ratio of DPPC:DSPC:DSPE-PEG₂₀₀₀ loaded with B-381 (19). We found this formulation had a high (>90%) drug release at 42°C but an undesirable 40% baseline release at 37°C (Supplementary Figure 1A). In order to improve liposome stability and decrease the baseline B-381 release at 37°C, we incorporated increasing amounts of cholesterol (10, 20 and 30 mol%) into the DPPC:DSPC:DSPE-PEG₂₀₀₀ liposome formulation. The different TSL formulations were then loaded with B-381 and the drug release at 37°C or 42°C was determined by ICP-OES (Figure Supplementary 1A). We found that all formulations with 10, 20 or 30 mol% cholesterol had a decreased baseline release at 37 °C, but the higher cholesterol ratios (20 and 30 mol%) had diminished thermal sensitivity and demonstrated decreased release at 42°C. The optimal TSL formulation contained 10 mol% cholesterol which stabilized release to 20% at 37°C, while maintaining thermal sensitivity at 42°C (releasing ~90% of the B-381 payload) (Supplementary Figure 1A). Similarly, we loaded BPA-f into the 10 mol% cholesterol TSL formulation and evaluated its release at 37 °C or 42°C; we found that this formulation had minimal BPA-f release at 37°C (19%) but released 77% of the BPA-f payload at 42°C. We further validated the thermal sensitivity of the 10 mol% cholesterol TSL formulation by loading it with doxorubicin and determining its release at 37 °C or 42°C using spectrophotometry. We found that the TSL with 10 mol% had significant doxorubicin release at 42°C, but minimal release at 37°C. Moreover, we developed a control non-thermal-sensitive liposome (non-TSL) formulation containing only DSPC, DSPE-PEG₂₀₀₀, and 10 mol% cholesterol. By removing the thermal sensitive phospholipid DPPC from the formulation, the non-TSL lost its temperature dependent drug release and had minimal doxorubicin release both at 37°C and 42°C (Supplementary Figure 1C). Therefore, the TSL and non-TSL formulations containing 10 mol% cholesterol were chosen for subsequent in vitro and in vivo studies.

The optimized TSL formulation contained a 71:14:5:10 molar ratio of DPPC/DSPC/DSPE-PEG₂₀₀₀/cholesterol, while the non-TSL formulation consisted of a 85:5:10 molar ratio of DSPC/DSPE-PEG₂₀₀₀/cholesterol, with the omission of DPPC (Figure 1A). For drug loading, the lipid films were rehydrated with aqueous solutions of BPA-f or B-381, which were passively entrapped in the aqueous core (Figure 1B). Unbound drug was removed by centrifugation, and BPA-f TSL, BPA-f Non-TSL and B-381 TSL formulations were characterized by DLS. All formulations had a narrow size range of 133.6 to 143.9 nm, with a low polydispersity index between 0.034 and 0.052, indicating a homogenous size-population (Figure 1C). The zeta potential seen for all three liposome formulations are slightly negative as shown in Figure 1C. The DPPC and DSPC lipids are zwitterionic where the choline and phosphate groups are positive and negative, respectively. The ionic strength of the liposome buffer dictates zeta potential (21). In this case, the phosphate-buffered saline (PBS) used, which had a low ionic strength, enabled the phosphate group in the lipid head to orient itself above the choline group. Encapsulation efficiencies were < 5% which is anticipated for a passive loading method. Using ICP-OES, the final boron content in the liposome stocks ranged from 1317–1801 ppb, which was used to insure equimolar boron dosing for in vivo studies. Additionally, the thermal effect of heating on liposome size (Figure 1D) and polydispersity (Figure 1E) was evaluated by incubating the liposomes at 21, 37, 42, and 60°C for different time periods. Liposome particle size and polydispersity remained stable even when heated to 60°C for 1 hour. This data suggests that hyperthermia does not cause aggregation, degradation, or change in the particle sizes of the liposomes, but rather causes a transient increase in membrane permeability thereby facilitating drug release; this is due to the T_m only affecting the fluidity of the lipid membrane (22). These results are in agreement with a previous finding by Grit and Crommelin who incubated liposomes at 60°C for 700 hours and did not observe a change in size (23).

Figure 1. Liposome Synthesis and Characterization:

A) Optimized phospholipid molar ratios for thermal (TSL) and non-thermal sensitive liposome (non-TSL) formulations; B) Illustration of a TSL containing boronated drug in aqueous core C) Liposome size, polydispersity index (PDI), zeta potential, encapsulation efficiency and boron content of liposomes. D) Change in TSL average diameter as a function of heating time at 21, 37, 42 or 60 °C. E) Change in TSL polydispersity index as a function of heating time at 21, 37, 42 or 60 °C.

Next, we evaluated the specific boron tumor delivery in vivo using the BPA-f TSL and BPA-f non-TSL formulations. We developed an animal model of athymic nude mice containing bilateral D54 glioma tumors on their right and left flank (2 tumors/mouse). Only the left tumor was heated against a hyperthermia window attached to a water circulator (42°C) designed in house (Figure 2A). After 5 minutes of pre-hyperthermia treatment, mice were tail vein injected with an equimolar boron content of BPA-f free drug, BPA-f TSL, or BPA-f non-TSL. The left side tumor was exposed to hyperthermia (42°C) for an additional 30 minutes post-injection while the right side tumor served as a 37°C control. Mice were then sacrificed, and both tumors, blood, liver, spleen, kidney, heart and lung were collected. All samples were digested and the boron content of each organ was analyzed by ICP-OES.

Figure 2. In Vivo Comparison of BPA-f, BPA-f TSL and BPA-f Non-TSL delivery with or without hyperthermia and long-term retention of B-381 compared to BPA-f:

A) Experimental setup for achieving local tumor hyperthermia: B) Tumor boron content after 30 minute hyperthermia (42°C) treatment following i.v. injection of BPA-f free drug, BPA-f TSL or BPA-f non-TSL compared to contralateral control tumor (37°C) C) Biodistribution of BPA-f free drug, BPA-f TSL or BPA-f non-TSL after 30 minute hyperthermia treatment. D) Average tumor boron content 1 or 24 h post-hyperthermia treatment with BPA-f TSL. E) Average tumor boron content 1 or 24 h post-hyperthermia treatment with B-381 TSL, *p<0.05.

A similar boron content was observed in the hyperthermic (216±70 ppb) and physiological (341 ± 88 ppb) treated tumors after delivery of BPA-f as a free drug (Figure 2B). In contrast, BPA-f TSL showed an 8.4-fold higher boron accumulation (692 ± 99 ppb) in the hyperthermic treated tumor compared to the low accumulation in the physiologic tumor (82 ± 20 ppb). The non-TSL BPA-f administration resulted in no detectable signal at physiologic temperature and a low signal of only 104 ± 41 ppb in the hyperthermic tumor.

These results indicate that BPA-f is rapidly and selectively released from the TSL formulation in the hyperthermia treated tumor, resulting in high boron accumulation. The TSL drug release is localized to the hyperthermia treated side as evident by the significantly lower uptake observed in the contralateral tumor at 37°C. In contrast, administration of BPA-f as a free drug or as a non-TSL formulation had lower and non-specific accumulation of boron in the hyperthermic and physiologic tumors. Additionally, when comparing hyperthermia treated tumors, BPA-f TSL had a 6.7-fold higher accumulation compared to the non-TSL formulation. This data supports that the increased boron uptake observed is not just solely an enhanced permeability and retention effect but represents actual localized drug release from a TSL. Even at 42°C, the non-TSL does not have significant drug release because the liposome composition is not thermal sensitive.

In addition to tumor boron levels, the biodistribution of all 3 drug vehicles was also evaluated (Figure 2C). While free BPA-f is rapidly cleared from the circulation 30 minutes post-administration, both BPA-f TSL and BPA-f non-TSL have prolonged circulation. This data further supports that the TSL formulation locally released BPA-f whereas the non-TSL does not since both liposome formulations have equivocal blood concentrations of ~1500 ppb at t = 30 minutes. No significant difference in accumulation was observed in other organs for the different formulations. However, there is a considerable increase in BPA accumulation in the spleen following injection of the TSL or non-TSL formulations. This is not expected to cause any side effects due to the non-toxic profile of BPA by itself. BPA only becomes cytotoxic when irradiated with neutrons during BNCT (4). Since the spleen and other organs are not going to be irradiated with neutrons in the clinic, we do not expect to have toxic effects in these organs due to boron accumulation.

We have previously shown that boronated 2-nitroimidazole derivative B-381 had a more specific and long-term tumor accumulation compared to BPA (12). Therefore, we compared the initial delivery and tumor retention of a B-381 TSL formulation to the BPA-f TSL formulation in vivo. Mice with bilateral D54 glioma tumors on their right and left flank were utilized. Once again, the left tumor was exposed to hyperthermia (42°C) while the right tumor served as a 37°C control. Mice were treated with BPA-f TSL or B-381 TSL, sacrificed at 1 or 24 hours post-injection, followed by tumor extraction and boron content analysis by ICP-OES.

Treatment with BPA-f TSL resulted in high boron content at 1 hour post-injection in the hyperthermic tumor only, which eventually was cleared at 24 hours post-injection (Figure 2D). In contrast, treatment with B-381 TSL resulted in significant boron content in the tissue 1 hour post-injection which was retained at 24 hours post-injection (Figure 2E). These results may be attributed to the fact that BPA is an amino acid analog and does not have a mechanism for long-term retention, while 2-nitroimidazole derivatives such as B-381 can form intracellular covalent protein conjugates in hypoxic cells.

In conclusion, two novel TSL formulations have been prepared and show promise for future BNCT studies. Both B-381 and BPA-f TSL formulations exhibited a hyperthermia-dependent release of its payload and showed specific tumor delivery of boron in vivo. Additionally, delivery of B-381 TSLs demonstrated a long-term retention at 24 hours compared to BPA-f TSLs. This is the first study to show delivery of boronated compounds using TSLs, and these results will provide the basis of future clinical trials using TSLs loaded with boronated compounds for boron neutron capture therapy.

Methods:

Reagents:

All chemical reagents were purchased from Sigma Aldrich (St. Louis, MO) unless explicitly stated otherwise. 2-nitroimidazole derivative B-381 was synthesized in house (12). Phospholipids 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC), 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC), 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[amino(polyethylene glycol)-2000] (ammonium salt) (DSPE-PEG₂₀₀₀) were purchased from Avanti Polar Lipids (Alabaster, AL). BPA fructose adduct (BPA-f) was prepared fresh according to a previously published protocol before liposome loading (24).

Liposome Synthesis and Characterization:

Liposomes were prepared using thin-film hydration followed by microextrusion at 60°C (Avanti Mini Extruder). A cholesterol-free (0 mol %) TSL was prepared according to previously reported molar ratio of 80:15:5 DPPC/DSPC/DSPE-PEG₂₀₀₀ respectively (19). Additionally, 10, 20 and 30 mol % cholesterol formulations were also evaluated with DPPC/DSPC/DSPE-PEG₂₀₀₀/cholesterol molar ratios of 71:14:5:10, 64:12:4:20, 56:10:4:30 respectively. A non-TSL was synthesized with a 0:85:5:10 DPPC/DSPC/DSPE-PEG₂₀₀₀/cholesterol molar ratio to serve as a negative control.

To prepare lipid films, the desired phospholipids were dissolved in 5 mL chloroform and concentrated on a rotatory evaporator. The lipid film was rehydrated for 1 hour at 280 rpm (60°C) with 1 mL of the following solutions: 30 mg/mL BPA-f (pH = 7) or 40 mg/mL B-381 in 300 mM sodium citrate (pH=4). Liposome mixture was then microextruded at 60 °C using 200 nm and 100 nm polycarbonate membranes sequentially. Liposome suspensions were ultracentrifuged at 39,000 rpm for 2 hours and resuspended in HEPES buffer (pH = 7). Encapsulation efficiency was determined using the volume normalized boron content in the resuspended liposomes compared to the removed supernatant boron level. Average particle size, polydispersity index and zeta potential were determined by dynamic light scattering measurements (Malvern Zetasizer Nano ZS).

Doxorubicin release from TSL and non-TSL formulation:

Doxorubicin was loaded into a previously reported TSL (80:15:5 DPPC/DSPC/DSPE-PEG₂₀₀₀ molar ratio) and compared to our non-TSL (95:5 DSPC/DSPE-PEG₂₀₀₀ molar ratio) using a sodium citrate loading method (encapsulation efficiency of 67%). TSL and non-TSL liposome aliquots (500 μL) were transferred to 14 kDa dialysis membranes and incubated at 37°C or 42°C. Dialysate was sampled over 0.25, 1, 3 and 19 h and doxorubicin fluorescence intensity was determined using a plate reader (λ_Ex: 482 nm; λ_Em: 594 nm).

Determination of BPA-f and B-381 release from liposomes:

In general, 500 μL aliquots of liposome stock solution were transferred inside of a 14 kDa Dialysis Membrane (6.4 mm x 10 mm, Ward’s Science, ON, Canada). The dialysis bag contents were placed in 50 mL of PBS buffer at either 37°C or 42°C and 100 uL aliquots were collected at various time points. The dialysate aliquots were resuspended to a final volume of 3 mL in 5% nitric acid and analyzed for boron content by ICP-OES. The dialysis bag contents were mixed into the bulk volume at the end of the experiment and used as a 100% control to determine % release. Boronated drug release from thermal and non-thermal liposome formulations was analyzed using Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES, Optima 7300 V series, Perkin Elmer, Waltham, MA). Samples were analyzed for boron content (λ = 249.677 nm) against a calibration curve of boron standards in 5% nitric acid between 0 and 250 parts per billion (ppb) prepared from a 10 parts per million boron standard solution (Inorganic Ventures, Christiansburg, VA). For in vivo experiments, tumor and organs were resected, weighed, and digested using a MARS 6 Microwave Digestion System (CEM Corporation, Matthews, NC). All samples were then brought to a normalized final volume of 4 mL containing 5% nitric acid (v/v). Control organs were utilized to establish baseline signal for each organ type, and final reported boron levels were normalized to organ mass.

In Vivo Tumor Implantation:

Approval for all animal studies was obtained from the Ethical Committee for Animal Experiments at Washington University in St. Louis Medical School. D54 Glioma cell line was a kind gift from Dr. Dinesh Thotala (Department of Radiation Oncology, Cancer Biology Division, Washington University in Saint Louis School of Medicine). Prior to in vivo implantation, cells were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM, Corning CellGro, Mediatech, Manassas, VA) supplemented with 20% fetal bovine serum (FBS, Gibco, Life Technologies, Grand Island, NY), 2 mmol/L of L-glutamine, 100 U/mL Penicillin and 100 μg/mL Streptomycin (CellGro, Mediatech, Manassas, VA).

Prior to in vivo implantation, cells were washed with phosphate-buffered saline (PBS, Corning CellGro, Mediatech, Manassas, VA), trypsinized with 0.05% Trypsin-EDTA 1x (Gibco, Life Technologies, Grand Island, NY), spun for 5 minutes (1000 RPM) and resuspended in fresh DMEM media and Matrigel (Corning) according to manufacturer protocol. Athymic Nude-Foxn1^nu mice (N=10, female, 6 weeks old) obtained from Envigo (Indianapolis, IN) were anesthetized with ketamine/xylazine and bilaterally injected with Matrigel solution under the skin of each hindlimb (two injections per mouse, 1.5 × 10⁶ D54 cells per tumor). Tumors were grown to an average tumor size of 300 mg and were clearly palpable to facilitate isolated heating of just 1 tumor per mouse.

In Vivo Comparison of BPA-f, BPA-f TSL and BPA-f Non-TSL with or without hyperthermia:

The mice were split into control (N=1) or 3 treatment groups: BPA-f free drug (N=3), BPA-f TSL (N=3) or BPA-f Non-TSL (N=3). Mice were anesthetized and positioned under tygon tubing (Inner Diameter = 0.375 inches) connected to a water bath circulator maintained at 41–43°C with a high flow rate. The tygon tubing contained a series of three 1 cm² openings sealed with a single layer of Tegaderm to facilitate heat transfer while preventing leaks. This facilitated heating 3 mice simultaneously. Mice were positioned carefully to only heat their left tumor while the right tumor served as control. The variability between mice was minimized since each mouse had a hyperthermia-treated (42°C) and control (37°C) tumor. A thermometer placed against the hyperthermia window remained stable at 41–43°C over the treatment period to validate stable heating and prevent reaching thermoablative temperatures > 45 °C. Mice were exposed to a 5 minute pre-hyperthermia heating period prior to tail vein injections of either BPA-f free drug, BPA-f TSL, and BPA-f Non-TSL (normalized to have equimolar boron content). After 30 minutes of hyperthermia exposure, blood was collected and mice were immediately sacrificed for organ collection. All excised tissues were weighed, digested with microwave digestion system diluted with deionized water to a final acid concentration of 5% (v/v) for ICP-OES analysis.

In Vivo Retention of BPA-f and B-381 TSL Formulations:

Twelve (N=12) Athymic Nude-Foxn1^nu mice containing bilateral D54 glioma flank tumors were split into 2 treatment groups receiving equimolar boron injections of either BPA-f TSL (N=6) or B-381 TSL (N=6). All mice were subjected to 5 minutes of hyperthermia, followed by tail vein injection of liposomes and 30 minutes of hyperthermia exposure. For each treatment group, 3 mice were sacrificed immediately after hyperthermia treatment (t = 30 minutes) while 3 mice were sacrificed at 24 hours post-treatment. Tumors were resected, weighed and analyzed by ICP-OES for boron content.

Statistical analysis:

Measurements were made in triplicate for each group. The data were expressed as means ± standard deviation. Results were analyzed using student t-test for statistical significance and were considered significantly different for p value less than 0.05.