Gold Coated Lanthanide Phosphate Nanoparticles for Targeted Alpha Generator Radiotherapy

Targeted radiotherapies maximize cytotoxicty to cancer cells. In vivo α-generator targeted radiotherapies can deliver multiple α particles to a receptor site dramatically amplifying the radiation dose delivered to the target. The major challenge with α-generator radiotherapies is that traditional chelating moieties are unable to sequester the radioactive daughters in the bioconjugate which is critical to minimize toxicity to healthy, non-target tissue. The recoil energy of the 225Ac daughters following α decay will sever any metal-ligand bond used to form the bioconjugate. This work demonstrates that an engineered multilayered nanoparticle-antibody conjugate can deliver multiple α radiations and contain the decay daughters of 225Ac while targeting biologically relevant receptors in a female BALB/c mouse model. These multi-shell nanoparticles combine the radiation resistance of lanthanide phosphate to contain 225Ac and its radioactive decay daughters, the magnetic properties of gadolinium phosphate for easy separation, and established gold chemistry for attachment of targeting moieties.


Introduction
Beta emitting radionuclides have found widespread use in cancer therapy. A major advance in nuclear medicine was the development of targeted endo-radiotherapies with two targeted radiotherapy agents approved for clinical use. BEXXARH, labeled with 131 I, is used to treat follicular lymphoma while ZevalinH, containing 90 Y, is used for treatment of B cell non-Hodgkins lymphoma [1][2]. Other targeted radiotherapy agents labeled with b 2 emitters 131 I, 90 Y, 177 Lu, and 188 Re are showing promise in ongoing clinical trials [3][4]. One of the challenges associated with b 2 emitting targeted radionuclide therapies is, however, the inherent toxicity from the death of normal, healthy cells resulting from the crossfire radiation damage from the relatively long ranges of the b 2 particles in tissue [5]. For example, b 2 particles from 177 Lu (b max = 0.5 MeV) have a range of 1.5 mm in tissue and b 2 particles from 90 Y (b max = 2.3 MeV) deposit their energy over a range of 12 mm. Targeted radiotherapies based on a particles are a promising alternative to b 2 particles because the a particles deposit all of their energy within a few cell diameters (50-100 mm). Because of their much shorter range, targeted a-radiotherapy agents have great potential for application to small, disseminated tumors and micro metastases and treatment of hematological malignancies consisting of individual, circulating neoplastic cells [6]. Compared with b 2 particles, a particles provide a very high relative biological effectiveness, killing more cells with less radioactivity. The high linear energy transfer of a particles induces significantly more DNA double strand breaks than b 2 particles [7]. Also, the biological effectiveness of a particles does not depend upon hypoxia or cell cycle considerations [8][9]. Most a emitters also have a relatively low c-ray component in their decay allowing for outpatient treatments and lower radiation doses to nuclear medicine personnel [10].
A number of targeted alpha therapy (TAT) agents based on the single alpha emitting radionuclides 211 At (t 1/2 = 7.2 h), 213 Bi (t 1/ 2 = 46 m), 212 Pb (t 1/2 = 10.6 h), and 212 Bi (t 1/2 = 61 m) have been developed and are showing promise in pre-clinical and clinical trials [11]. The radiotherapeutic efficacy of TAT could, however, be further enhanced by use of in vivo a-generator radionuclides like 225 Ac, which emits four a particles in its decay chain ( Figure 1). The median lethal dose of 225 Ac constructs is one to two orders of magnitude lower than the LD 50 values for the corresponding single a emitting 213 Bi labeled antibodies in vitro with a number of cancer cell types [12]. Moreover, the longer half-life of 225 Ac (t 1/2 = 10 d) reduces activity loss during radiopharmaceutical synthesis and allows greater time for localization of antibodies to receptor sites.
Despite these advantages, there is a distinct challenge associated with targeted in vivo a-generator radiotherapy. If the a-emitting daughter products in the 225 Ac decay chain are not sequestered at the target site, they can migrate and deliver a potentially toxic dose to non-target tissue [11]. The recoil energy of the 225 Ac daughters following alpha decay (.100 keV) will sever any metal-ligand bond used to form the bioconjugate, releasing the daughter radionuclides from the targeting agent. Renal toxicity is currently the dose-limiting factor in clinical use of 225 Ac. In the recent work of Schwartz et al., almost 80% of the absorbed dose to the renal medulla was delivered by free 213 Bi when using a metal-ligand bioconjugate to deliver 225 Ac in a mouse model [13]. Metal-ligand bioconjugates fail to sequester the daughter products of 225 Ac ( 221 Fr, 217 At, and 213 Bi) and the released 213 Bi accumulates in the kidney [14][15]. An alternative strategy to this challenge, incorporating 225 Ac in engineered liposomes, was found to retain less than 10% of the 213 Bi activity from the decay of 225 Ac in vitro [16].
The in vivo a generator 223 Ra, which also emits four alpha particles in its decay chain, is an effective treatment for metastatic bone cancer [17]. Radium-223 chloride has been granted Fast Track designation by the U.S. Food and Drug Administration for the treatment of hormone-refractory prostate cancer in patients with bone metastases [18]. It is effective in this case because radium is a calcium mimic with a high affinity for bone tissue and the daughter products either have short half-lives or have a high affinity for bone ( 211 Pb, t 1/2 = 36 m). Translation of in vivo a generators to anything besides metastatic bone cancer with a calcium mimic will require a different mechanism for both delivering and retaining the radioactive daughters in the target tissue.
In this work, we demonstrate that a multilayered nanoparticle (NP) can contain the recoiling daughters of the in vivo a generator, and when coupled to a targeting antibody, can bind to biologically relevant receptors and deliver multiple a particles to each receptor site. While the range of the recoiling daughters is less than the diameter of the layered NPs, the a-particles lose less than 0.2% of their energy as they exit from the center of the NPs. The layered NPs consist of 225 Ac-doped {La 0.5 Gd 0.5 }PO 4 @GdPO 4 @Au. A schematic of the layered NP design is illustrated in Figure 2. These multi-shell particles combine the radiation resistance of crystalline lanthanide phosphate [19], [20][21] to encapsulate and contain atoms of the therapeutic radionuclide 225 Ac and its radioactive daughters, the magnetic properties of gadolinium phosphate NPs, and established surface chemistry of gold NPs for attachment of targeting agents [22].
Nanoparticles offer a number of favorable properties with regard to drug delivery. The ability to hold multiple copies of a therapeutic or imaging moiety provides the ability to generate efficacious results even against targets with low receptor numbers in vivo [23]. Multi-functional, layered NPs allow for synergistic combinations of properties exhibited by the individual NPs including containment, purification and conjugation. Moreover, in the system described in this work, GdPO 4 could function as a magnetic resonance imaging (MRI) contrast agent [24] and Au can be doped with the single photon emission computed tomography (SPECT) radionuclide 199 Au for c-ray imaging [25].

Results and Discussion
{La 0.5 Gd 0.5 }( 225 Ac)PO 4 core particles were synthesized by hydrolysis of sodium tripolyphosphate (Na-TPP) in the presence of equimolar mixtures of La and Gd salts with 225 Ac present at the tracer level. Nanoparticles of diameter ca. 4 nm precipitated out of solution after heating for 3 hours at 90uC. In order to improve retention of the decay daughters, four additional shells of pure GdPO 4 were added, each by dispersing the precipitated NPs in a solution of Gd 3+ and Na-TPP and heating for an additional 3 hours. Gold was then added to the NPs by reduction of NaAuCl 4 with sodium citrate.
Rietveld refinement of x-ray diffraction (XRD) patterns indicated that LaPO 4 NPs exhibited the rhabdophane phase consistent with the description of Buissette et al. [26]. However, the {La 0.5 Gd 0.5 }PO 4 , {La 0.25 Gd 0.75 }PO 4 , and GdPO 4 systems crystallized in the anhydrous monazite phase [27]. The monazite phase for LaPO 4 NPs was previously observed for crystalline  Addition of GdPO 4 shells to the core {La 0.5 Gd 0.5 }PO 4 NP causes epitaxial growth of the particle. Mean diameters increase sequentially with each shell addition ( Table 1). Addition of four GdPO 4 shells to the core {La 0.5 Gd 0.5 }PO 4 produces 22 nm diameter NPs and addition of an outer gold layer increases the particle diameter to 27 nm. Electron energy loss spectroscopy (EELS)-TEM images of the NPs are shown in Figure 4.
Gold coated NPs with four epitaxially added GdPO 4 shells were further characterized by dynamic light scattering. Hydrodynamic diameters and zeta potentials are shown in Table 2. An increase of the hydrodynamic diameter on addition of polyethylene glycol (PEG) and antibody is common for NPs. The highly negative zeta potentials should lead to stability in water which was confirmed by monitoring changes in the UV-Vis spectrum of the particles over a 1 month period in both 18 MV water and saline solution. No shift was observed in the plasmon resonance over this time period.
Nanoparticles with GdPO 4 shells followed by Au coating dramatically increased radioactive daughter retention in vitro compared with previously published results for core-only lanthanum phosphate NPs [28]. Adding 2 shells increased retention of the 221 Fr daughter from 50% for the LaPO 4 core to 70%. With four shells of GdPO 4 , the initial retention of the 221 Fr daughter was 98%. Daughter retention decreased by roughly 2% per day over the course of a week, and stabilized at 88%. Further, the presence of the Au/4 GdPO 4 shells increased the retention of the 225 Ac parent itself by roughly an order of magnitude. Over the course of 3 weeks, the multi-layered particles retained greater than 99.99% of the 225 Ac parent radionuclide. Particles with more than 4 shells of GdPO 4 settled out of solution rapidly and were difficult to manipulate. Monitoring the plasmon resonance indicated that the multi-layered particles remained stable towards aggregation in PBS over the course of one month.
For in vivo biodistribution testing, the NPs were conjugated to the mAb 201b monoclonal antibody via a lipoic acid-PEG 12 -COOH linker [29]. MAb 201b targets thrombomodulin receptors which are highly expressed in lung endothelium. The antibody quickly localizes to its vascular target and clears from circulation with a half-life of 40 hours [30]. 3-sulfo-N-hydroxysuccinimide (sulfo-NHS) and 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide (EDC) activated the carboxylate of the PEG for coupling to amine groups on the antibody, leading to the formation of an amide bond. The reaction was quenched with glycine and conjugates were purified by centrifugation. The conjugated NPs were redispersed in phosphate buffered saline (PBS) containing bovine serum albumin (BSA). The antibody conjugation process is summarized in Figure 5.   In vivo biodistribution experiments of the 225 Ac containing NPs (ca. 2 mCi/animal) demonstrated that the antibody-targeted NPs localized in the lung consistent with the binding properties of mAb 201b. The NPs exhibit high lung uptake with the antibody conjugate after 1 hour (151%ID/g). This high lung uptake dropped to 16.8%ID/g when the antibody conjugated NPs were competed with unconjugated antibody (Figure 6). These results demonstrate that the antibody retained its binding affinity and specificity even after conjugation to the NPs and that the NPs localized in the lung through antibody binding. While the antibody-labeled NPs cleared rapidly from the lungs in these proof-of-principle experiments (after 24 hours, 225 Ac activity was predominantly present in the liver and spleen), previous strategies used to reduce reticuloendothelial functioning such as treatment with clodronate liposomes could be applied to mitigate the rapid clearance [31], [32][33].
Retention of 213 Bi, from the decay of 225 Ac in the a-generator NPs, was 69% 63% in lung tissue after 1 hour and increased to 84% 63% after 24 hours. Similar 213 Bi retention values were observed in liver (1 h, 81% 64%; 24 h, 92% 61%) and spleen tissue (1 h, 72% 63%; 24 h, 82% 616%). Despite the widespread renal toxicity concerns associated with 213 Bi relocation to the kidney from 225 Ac a-generator therapies, only 2.8% of the 213 Bi from the injected dose migrated to kidney tissues after 1 hour. After 24 hours, this number further decreased to 1.5%.
A larger dose (ca. 80 mCi/animal) of 225 Ac NPs was imaged using CT/SPECT of the 221 Fr c-ray (218 keV, 11.6%). Mice injected with this larger dose were sacrificed 1 hour post-injection and imaged 3 hours post-sacrifice to allow the daughter products of 225 Ac to reach their equilibrium activities. The CT/SPECT images (Figure 7) clearly show large uptake in the lung for the {La 0.5 Gd 0.5 }( 225 Ac)PO 4 @GdPO 4 @Au-mAb-201b NPs which is in agreement with the biodistribution data. When competed with unconjugated mAb 201b antibody, the images showed high uptake in the liver. If the antibody conjugated NPs cannot bind their in vivo target, they are cleared from circulation via the reticuloendothelial system. Finally, PEG coated NPs without antibody also show high uptake in the reticuloendothelial system (Figure 7), further indicating that the lung uptake is not due to particulate trapping in the small capillary system. {La 0.5 Gd 0.5 }( 225 Ac)PO 4 @GdPO 4 @Au NPs represent a novel system for targeted a radiotherapy. Adding a Au surface onto a LnPO 4 core (Ln = La, Gd) allows for facile, reproducible surface functionalization. The addition of Gd into the particles creates a magnetic moment which is sufficient to separate the gold NPs containing Gd from any gold NPs produced in the gold coating step. This separation ensures that gold NPs without a radioactive core will not compete with the TAT conjugate for receptor sites.
Compared with single a-emitting therapies, the use of in vivo a generators holds the potential to deliver a much larger biologically effective dose to target tissues. Effective design of in vivo TAT agents with isotopes like 225 Ac requires two major components. First, the therapeutic agent must be able to deliver the generator radionuclide specifically to target tissue at a cytotoxic dose. The high, receptor-mediated uptake of particles in the lung endothelium demonstrates the ability of {La 0.5 Gd 0.5 }( 225 Ac)PO 4 @Gd-PO 4 @Au NPs to deliver 225 Ac to a tissue target that is present in the vascular space. Second, the TAT must be able to retain the daughter products of the generator in the target tissue. Migration of daughter products to non-target tissue will severely limit the administered therapeutic dose. Retention of the decay daughters can be achieved in a number of ways. First, the radionuclide may be selected so that the daughter half-lives are sufficiently short that they will not have time to migrate throughout the body. Alternatively, the radionuclide can be chosen so that the daughter products exhibit similar in vivo behavior and remain in the target tissue. This is the principle behind the recent successes using 223 RaCl 2 for treatment of bone metastases [34]. The 223 Ra daughter products either have short half-lives or have a high affinity for bone ( 211 Pb, t 1/2 = 36 m). While effective in this case, translation of this in vivo a generator to other tumor types would require a different mechanism of retaining the 211 Pb and 211 Bi daughters in the target tissue. A third solution to the daughter retention problem involves internalization of the parent radionuclide in the target cell itself [12]. This approach utilizes the internal milieu of the cell to contain the daughter decay products. Tumor targets for internalization occur largely in the extravascular space,  which is difficult to access with larger constructs that promote endocytosis. Attempts to reduce 213 Bi toxicity through targeted, metal-chelate based internalizing antibodies have shown only moderate success [35]. The NP construct described in this work improves 225 Ac daughter retention relative to both chelate approaches and previous NP constructs. {La 0.5 Gd 0.5 }( 225 Ac)PO 4 @GdPO 4 @Au NPs contain 88% of the 221 Fr daughter in vitro, compared with 50% retention observed with La( 225 Ac)PO 4 NPs [28]. Additionally, the in vivo a-generator delivery agent has a negligible effect on the energies of the emitted a particles. A 6 MeV a-particle loses less than 0.2% of its energy in the layered NP whereas the range of the 100 keV recoiling daughters is ,20 nm in bulk LnPO 4 . Moreover, a portion of the kinetic energy of the daughter particle may be transferred to the entire particle. If a portion of the recoil energy is distributed throughout the highly structured crystalline lattice, the recoiling range of the daughter radionuclides will be significantly decreased [36].
In vivo, the increase of retention of 213 Bi in the target tissue over time results from a combination of the ability of the layered NPs to retain the daughter products and endocytosis of the TAT NP. In this work, 213 Bi daughter retention in vivo with the layered NP showed improvement over the LaPO 4 core NP [28]. The 213 Bi retention is lower than the 221 Fr retention because prior decays of 225 Ac, 221 Fr, and 217 At can move the remaining a-emitting nuclides towards the surface of the NP. From this position nearer the surface, subsequent a decays are likely to release the daughter nuclide from the NP. The amount of 213 Bi which relocates to the kidney from other tissues shows marked improvement with the {La 0.5 Gd 0.5 }( 225 Ac)PO 4 @GdPO 4 @Au-mAb-201b system compared with the La( 225 Ac)PO 4 -mAb-201b system. 27 Only 2.8% of the injected dose migrated to the kidney as 213 Bi after 1 hour and 1.5% after 24 hours in the layered NPs while 10% of the ID relocated to the kidney after 1 hour and 5% after 24 hours with the core only lanthanum phosphate NPs.
These experiments demonstrate that multi-functional, layered NPs can be used to deliver and retain 225 Ac and its daughter radioisotopes at a target site thereby reducing the absorbed dose to non-target organs. TAT experiments in a model tumor system are in progress to directly assess the efficacy of the constructs.

Materials and Methods
All chemicals were used as received from Sigma-Aldrich and were at least ACS grade unless otherwise noted. Water originated from an in house 18 MV MilliQ system. Radioactivity measurements were performed with c-ray spectroscopy employing a calibrated high purity germanium detector employing a PC-based multichannel analyzer (Canberra Industries) windowed on 221 Fr (212 keV) and 213 Bi (440 keV). 225 AcCl 3 was prepared as previously described from a 229 Th cow [28]. A Spectra/Por 10 kDa molecular weight cutoff (MWCO) regenerated cellulose dialysis membrane was used to separate NPs from solutions. Dialysis membranes were washed of preservatives before use against 18 MV water.  total Ln:Na-TPP ratio of 1:2 resulting in a clear, colorless solution. If the solution remained turbid after addition of Na-TPP, it was vortexed with small (10 mL) additions of Na-TPP until the solution appeared clear. The resulting solution was then capped and heated at 90uC for 3 hours giving a turbid, white suspension of particles. Particles were purified via dialysis overnight. This preparation produced monodisperse particles of ,4 nm diameter which were characterized by transmission electron microscopy (TEM, JEOL 1400), neutron activation analysis (NAA) and x-ray diffraction (XRD, Scintag X2).

Layering of Particles
Core particles described above were centrifuged at 3,000 g for 3 minutes and the supernatant was removed. The particles were redispersed in a solution consisting of 200 mL of 0.05 M GdCl 3 and 400 mL 0.05 M Na-TPP. The resulting mixture was vortexed briefly then sonicated using a bath sonicator for 10 minutes before heating at 90uC for three hours. This process was repeated for up to four shell additions, at which point the solution becomes a thick milky white. Particles were purified by dialysis as above before gold coating.
The dialyzed particles (12 mg) were collected and split evenly between three 5-mL V-bottom vials. 300 mL of 0.1 M tribasic sodium citrate was added to each vial along with 1.5 mL of 18 MV water. Next, 2.5 mL of 1 mM NaAuCl 4 2 was added dropwise to the solution slowly at the rate of 1 mL every 10 minutes. After the final addition, the solution was kept at 90 0 C for 30-45 minutes. A large NdFeB magnet (surface field = 0.4 T) was placed next to the V-bottom vial for 16 hours to separate the particles from solution. The supernatant was decanted to isolate the magnetically active particles. In the radiotracer labeling experiments, the separation efficiency was determined by c-ray spectrometry of the removed supernatant and magnetically collected particles [28]. Non-radioactive analogs of the particles were characterized by EELS-TEM (Zeiss Libra 120) and NAA.

In Vitro Testing of 221 Fr Retention
In order to test retention of the 225 Ac decay products in vitro, the 225 Ac-NPs were loaded into a dialysis membrane and dialyzed against 400 mL of 18 MV water. The dialysis tube was stirred for a sufficient time for daughter equilibrium to be established (.3 hours), then a 5 mL aliquot was taken for c-ray spectrometry analysis. Each sample was re-analyzed at a later time to determine the level of 225 Ac in the removed dialysate fraction. The measured activities were corrected for decay and dialysate loss from prior aliquot removals. The 213 Bi activity in the dialysate was used as a measure of the 221 Fr that was released from the NP, as 213 Bi which escaped from the particles did not move across the dialysis membrane [28].

Surface Modification
Surfaces were modified using a lipoamide-dPEG 12 -acid linker (Quanta Biodesign). Two mg of dPEG were added, followed by 6 mg of Tris(2-carboxyethyl)phosphine reducing agent to cleave the disulfide bond. The pH of the solution was adjusted to 7 using 0.1 M NaOH and the reaction mixture was stirred for 4 hours. Connection of the linker was determined by a shift in the plasmon resonance near 530 nm as monitored by UV-Vis spectroscopy before and after the addition of the linker.

Antibody Conjugation
To attach antibodies to the linker, the carboxylate group on the PEG linker was first activated using 8 mL of 10 mg/mL aqueous 3sulfo-N-hydroxysuccinimide (sulfo-NHS) and 80 mL of 10 mg/mL 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide (EDC). 4 mg of EDC and 0.4 mg of sulfo-NHS were used per mg of NPs. After 15 minutes, the solutions were centrifuged to remove excess EDC/ NHS, the supernatant removed, and the particles redispersed in phosphate buffered saline (0.01 M NaPO 4 pH 7.6 in 0.15 M NaCl, PBS). MAb 201b was added (, 1 mg of mAb/mg of NP), and the mixture was mixed by rotation overnight. The reaction was then quenched with 1 mg of glycine, which was allowed to react for 15 minutes. The particles were then centrifuged to remove excess antibody. The supernatant was removed and the particles were redispersed in PBS containing 5 mg/mL bovine serum albumin (BSA/PBS). The particles were sonicated with a Branson microprobe for 10 sec and vortexed prior to injection. The final product of the mAb conjugated NP was ,3 mg/mL NP with 400 mCi of 225 Ac and ,1 mg mAb 201b.

Biodistribution Studies
All experiments involving mice were performed according to the Institutional Animal Care and Use Committee of the University of Tennessee approved protocol 1502. Female BALB/c mice (body mass ,20 g) were used for all biodistribution and imaging experiments. Biodistribution and daughter retention assays were done on three groups, consisting of three mice per group, were each injected intravenously (tail vein). Groups 1 and 2 were injected with Au/GdPO 4 /{La 0.5 Gd 0.5 ( 225 Ac)}PO 4 -mAb-201b outer shell/inner shell/core conjugates, while group 3 was treated with Au/GdPO 4 /{La 0.5 Gd 0.5 ( 225 Ac)}PO 4 -PEG NPs as a control. Group 1 mice received 14.6 mg of NP with 1.95 mCi of Ac-225 and , 5 mg of attached mAb 201b (this value was estimated from data in a parallel experiment wherein about 30% of added radioiodinated mAb was incorporated in NP under similar conditions). Group 2 received the same amount of targeted NP but with the addition of 750 mg of free mAb 201b as competitor. Group 3 received the same amount of NP and Ac-225, but with no targeting agent conjugated. Mice were housed with food and water ad libitum in a light/dark cycle environment before sacrificing at 1 and 24 h post-injection for biodistribution and in vivo retention studies. Biodistribution studies were performed on lungs, liver, spleen, and kidneys to evaluate the amount of both 221 Fr and 213 Bi in target organs by measuring weighed tissue samples in a c-ray scintillation counter at a specific time postsacrifice and again after the radioisotopes had achieved decay equilibrium (.3 h). Quantities of 221 Fr and 213 Bi present at the time of animal sacrifice were determined by appropriate crossover and decay corrections as previously described [28].
MicroSPECT/CT Imaging Small animal imaging was performed using a microCAT II SPECT dual modality platform (Siemens Preclinical Imaging, Knoxville, TN). Mice were injected with approximately 40 times more NP than were the mice for biodistribution studies. Thus the mice for competition with cold mAb 201b did not have the same ratio of cold competitor to radiolabeled NP and competition was not as complete.Animals were sacrificed by overdose of isoflurane at 1 h postinjection and imaged via microSPECT/CT 3 h later when the 225 Ac and its daughters had reached equilibrium. SPECT data for the final images were acquired as previously described by Woodward et al. [28].