Feed-forward alpha particle radiotherapy ablates androgen receptor-addicted prostate cancer

Human kallikrein peptidase 2 (hK2) is a prostate specific enzyme whose expression is governed by the androgen receptor (AR). AR is the central oncogenic driver of prostate cancer (PCa) and is also a key regulator of DNA repair in cancer. We report an innovative therapeutic strategy that exploits the hormone-DNA repair circuit to enable molecularly-specific alpha particle irradiation of PCa. Alpha-particle irradiation of PCa is prompted by molecularly specific-targeting and internalization of the humanized monoclonal antibody hu11B6 targeting hK2 and further accelerated by inherent DNA-repair that up-regulate hK2 (KLK2) expression in vivo. hu11B6 demonstrates exquisite targeting specificity for KLK2. A single administration of actinium-225 labeled hu11B6 eradicates disease and significantly prolongs survival in animal models. DNA damage arising from alpha particle irradiation induces AR and subsequently KLK2, generating a unique feed-forward mechanism, which increases binding of hu11B6. Imaging data in nonhuman primates support the possibility of utilizing hu11B6 in man.

A lpha particles (α-particles) are potent therapeutic effectors that have entered clinical practice [1][2][3] . These charged helium nuclei emitted upon decay travel approximately 50-80 μm and have a high linear energy transfer (LET) of approximately 100 keV/µm with high relative biological effect. Thus, they are able to kill a target cell by depositing 5-8 MeV in a highly focused ionizing track that is only several cell diameters in length. Indeed, a single α-particle traversal through a cell can be cytotoxic 4,5 . Unlike the approved bone seeking calcium mimetic, Radium-223 dichloride (Xofigo, Bayer Healthcare) Actinium-225 ( 225 Ac (t 1/2 = 10 days)) is an α-particle emitting radionuclide that can be conjugated to targeting macromolecules such as antibodies for cancer cell specific therapy. We have deployed molecular targeting in our pharmacologic strategy to treat prostate cancer by exploiting upregulation of the androgen receptor (AR), a central oncogenic driver of the disease.
The overexpression of the AR, even in states of castrate resistant disease, is a hallmark of prostate cancer progression 6,7 . Recent insights into the mechanisms of DNA repair in prostate cancer cells have shown that AR is associated with a plethora of DNA repair genes and also that active AR-signaling increases repair rates 8,9 . The genotoxic consequences of α-particle irradiation to cancer cells result in significant, often lethal DNA damage.
In this study, we introduce a strategy that takes advantage of the oncoaddiction of prostate cancer to AR-and its upregulation following DNA damage-by radioimmunotherapy targeting of human kallikrein peptidase 2 (hK2). hK2 is a well-characterized protease, with 80% amino acid (a.a.) identity to prostate specific antigen (PSA), that is directly regulated by AR activity. A monoclonal antibody (hu11B6) has been designed to specifically address an epitope accessible only on the free, catalytically active form of human hK2 and has been developed as a diagnostic [10][11][12] . Importantly, when hu11B6 binds to active hK2, the immune complex is internalized by the cell and trafficked to lysosomal compartments in a process made possible by concomitant binding of the hu11B6 with neonatal Fc receptor (FcRn). Here we introduce a molecularly specific alpha particle emitting radiotherapeutic [ 225 Ac]hu11B6. Internalization of 225 Ac-radiolabeled hu11B6 drug by the prostate cancer cell increases the probability that the α-particle energy is deposited in the nucleus.
Our overall strategy seeks to take full advantage of the unique consequences of [ 225 Ac]hu11B6-tissue-specific targeting and high LET α-particle irradiation of prostate cancer. We recognized that the α-particle induced DNA damage and ensuing upregulation of AR and KLK2 would increase the prostate cancer targeting by [ 225 Ac]hu11B6, creating an amplification loop for cell-specific therapy. Using advanced small animal models of prostate cancer, we demonstrate the ability to effect disease control. Additionally, we establish a diagnostic imaging reporter for [ 225 Ac]hu11B6 evaluated in both murine models and nonhuman primates.
Binding affinity was not affected by radiolabeling. The unmodified 11B6 antibody and its derivative radioimmunoconjugates demonstrated robust binding and dissociation curves with the purified hK2 on the protein A sensor chip. All the constructs that were analyzed yielded K D values in the lower nM range. The kinetic constants of [ 225 Ac]hu11B6 and [ 89 Zr]hu11B6 matched very well with those of the unmodified batch of 11B6 antibody that was used to prepare the radiolabeled constructsthus indicating minimal loss of immunoreactivity of the        3g) and longitudinally at four (Fig. 3h) and eight weeks ( Fig. 3i) after alpha particle therapy. Nonivasive monitoring and analysis of disease reduction vs. untreated controls in a GEM model reinforce the therapeutic potential of this internalizing alpha particle emitting antibody construct in vivo.
Alpha particles promote lethal feed-forward effect. The kinetic profile of [ 225 Ac]hu11B6 accumulation in LNCaP-AR tumors is shown in the curve in Fig. 4a and highlights the increasing tumor uptake of drug with time (slope is 0.037 ± 0.001 %IA/g/h and intercept is 11.1 ± 0.25 %IA/g; R 2 is 0.999). Therapeutic efficacy of [ 225 Ac]hu11B6 is evidenced in the tumor volume data shown in Fig. 4b that compares drug vs. vehicle-treated control. After 5 days, the trajectory of untreated tumor volumes increases while tumors irradiated with alpha particles delivered by the hu11B6 antibody significantly decrease in volume. These data correlate with the survival curve shown in Fig. 1c in which half the animals in the treatment arm had no tumor at 120 days. This response is due to the interaction of α-particles with tumor tissue that produces lethal DNA damage.  (Fig. 6). The tracer cleared from the blood with an effective halflife of 2.5 days and accumulated in the prostate tissue of these animals, as expected from the rodent studies. Quantification of the noninvasive imaging was made by computing the standardized uptake value of [ 89 Zr]hu11B6 activity distribution in organs of interest. The uptake of tracer in prostate peaked 5-8 days postadministration with the ratio of prostate SUV to muscle SUV reaching 23-fold on day 8.  15 have all demonstrated potent bioactivity with an ability to provide tumor control in man. When an alpha particle emitting radionuclide accumulates at the sites of malignancy, they locally deposit a high absorbed dose to the tumor target cells; normal healthy tissue is largely spared unless the delivery vehicle accumulates there in addition to tumor. The untoward off-target salivary and kidney uptake of [ 225 Ac]PSMA-617 may well limit the utility of this small molecule in treating prostate cancer as these healthy tissues also express PSMA 16 . This calls attention to the need for engineered biologic delivery molecules to truly disease tissue-specific targets. The interaction of alpha particles with cells significantly perturbs numerous biological pathways thus altering homeostatic cell function. The hormone-DNA repair circuit is engaged in direct response to alpha-induced DNA damage and the intricate pathway networks that sense genomic damage become activated in order to initiate repair mechanisms and maintain genomic integrity 8,9,17 . Until now alpha particle drug discovery efforts have focused solely on inherent cytotoxicity and neglected the subtle, yet profound radiobiological changes that accompany irradiation. We have factored the mechanistic advantages of AR pathway upregulation into our drug design in order to enhance alpha therapy.

Discussion
The androgen receptor pathway has an integral role in prostate cancer biology and manipulating AR expression is a viable strategy to eradicate this disease. Our data support a unique feedforward mechanism that exploits the role of AR in prostate cancer resistance to therapy to become lethally addicted to [ 225 Ac] hu11B6 (Fig. 7). Herein, we demonstrate that cell specific alpha particle radiation has the potential to not only kill tumor cells, but also to induce DNA damage and upregulate AR and KLK2. We have constructed a system wherein [ 225 Ac]hu11B6 initiates a potent feed-forward loop that ultimately addicts the cancer to the lethal alpha particle emitting drug. Alternative strategies using small molecule AR inhibitors have become an important treatment option for late stage and castration-resistant prostate cancer 18 . Unfortunately, the bioactivity of these AR-inhibitors eventually fails as the tumor escapes through heterogenous mechanisms of acquired resistance 19     provides robust correlative pharmacokinetic information that is used to report the behavior of the alpha emitting version of this drug in vivo. Two human xenograft models of prostate cancer (LNCaP-AR and VCaP) are used to demonstrate targeting specificity and efficacy. VCaP expresses more hK2 than LNCaP-AR and demonstrates greater accumulation of drug as predicted. The transgenic cancer susceptible mouse with prostate specific expression of human hK2 (Hi-Myc x pb_KLK2) recapitulates the biology of human prostate cancer and was engineered to investigate the radiobiological mechanisms that are the design foundation for [ 225 Ac]hu11B6. This GEM model shows higher KLK2 expression in the ventral and dorsal lateral prostate lobes than the anterior, which correlates to the commonly observed rate of spontaneous adenocarcinoma development and intraepithelial neoplasia in the respective lobes of a Hi-Myc GEM. Our data ( Fig. 3a) confirm this hK2 tissue distribution and neoplastic behavior as evidenced by drug accumulation and is confirmed by MRI (Fig. 3f-h). The alpha emitting drug accumulates in the cancer present in these lobes and not in the nearby ureter and seminal vesicles (Fig. 3c) thus sparing these tissues due to the short alpha particle range. Specificity and efficacy of our hu11B6mediated alpha therapy is apparent from the distinct lobe volume reduction when compared to control as a function of time (Fig. 3d-e).
The binding and internalization of hu11B6 has been engineered to perform several important tasks. First, targeting and binding hK2, an epitope expressed exclusively on prostate tissue and cancer in vivo; second, internalizing and transporting radionuclide cargo inside the cell to optimize the geometry of parent and progeny alpha decay. The hu11B6 binds to active, produces an alpha nanogenerator drug with progeny that can yield 4 net alpha emissions 4,14 . Cellular internalization of the hK2:[ 225 Ac]hu11B6 immune complex significantly improves the probability that the emitted alpha particles (of parent and progeny) will deposit energy inside the cancer cell. Indeed, in small animal models of osseous disease of bone-tropic prostate cancer, we are able to visualize the distribution of the [ 225 Ac]hu11B6 conjugate throughout the lesion (Fig. 2a). This is in contrast to the labeling of active bone surfaces, some of which are apposite the bone lesion, with the calcium mimetic 223 Ra (Fig. 2b). As noted above, the short path length of alpha particles limits the radiobiological effect to only several cell diameters, however with high biological effectiveness due to high LET. Molecularly specific radiotherapy through the antibodyconjugate approach in combination with internalization ensures that the alpha particle cascade inflicts significant DNA damage and prompts increased AR expression. This hormone-DNA circuit increases hK2 expression which establishes a feedforward process leading the cancer to bind more [ 225 Ac] hu11B6. This mechanism is evidenced in the increasing tumor accumulation of [ 225 Ac]hu11B6 with time (Fig. 4a) despite the reduction in tumor burden (Fig. 4b). The predicted upregulation of AR is quantified in these tumors using the BLI readout that correlates with AR expression (Figs. 4c,d). The downstream effects of increased AR on KLK2 and KLK3 were measured (Fig. 4d). FOLH1 which controls PSMA expression is notably not upregulated in these samples (Fig. 4d) and suggests that alpha particle therapeutics targeting PSMA expressed on prostate cancer (e.g., J591 or PSMA-617) will not engage a similar feed-forward mechanism.
Imaging the Ac-225 radionuclide has not been possible at the activities projected for clinical use. [ 89 Zr]hu11B6 is a positron emitting construct that we have evaluated as a surrogate reporter for [ 225 Ac]hu11B6. The biodistribution data obtained by tissue harvest and counted at 225 Ac secular equilibrium is shown in Fig. 5. This study clearly shows that molecular hu11B6 directs the pharmacokinetic profile as evidenced by the strong parallel in 89 Zr and 225 Ac tissue and blood distribution and clearance values. The DFO and DOTA chelates stably contain 89 Zr and 225 Ac, respectively and are integral to construct design. With this validated noninvasive surrogate reporter platform for [ 225 Ac]hu11B6 in hand, we proceeded to investigate pharmacokinetics and safety in a nonhuman primate. The crab-eating macaque (Macaca fascicularis) also expresses hK2 in the prostate, albeit at 1000-fold lower levels than humans 21,22 , and it cross-reacts with the hu11B6 antibody. Our quantitative [ 89 Zr]hu11B6 PET data in adult male subjects (Fig. 6) showed that the tracer cleared from the blood with an effective half-life of 2.5 days and accumulated in their prostate tissue. Tracer uptake in prostate peaked 5-8 days post-administration and on day 8 the prostate-to-muscle SUV ratio was 23. This study established the hK2-expressing prostate targeting ability of hu11B6 antibody in a healthy nonhuman primate model. Accumulation of [ 89 Zr]hu11B6 even at this extremely low-level of hK2 in the macaque indicates that robust targeting in humans of the therapeutic construct should have the opportunity to provide significant therapeutic benefit.
Our overall strategy has been designed to take full advantage of the unique consequences of [ 225 Ac]hu11B6 tissue-specific targeting and high LET α-particle irradiation of prostate cancer. We recognized that the alpha particle induced DNA damage and relied on ensuing upregulation of AR and KLK2 to addict the prostate cancer to the therapeutic [ 225 Ac]hu11B6 agent. Radiobiological addiction favorably drives the cancer to bind more of the cytotoxic drug, resulting in ablation of the disease.

Methods
Cell lines. The VCaP cell line was purchased from American Type Culture Collection (Manassas, VA) and cultured according to the manufacturer's instructions. LNCaP-AR-luc (LNCaP cell line with over-expression of wild-type AR expressing luciferase under the control of ARR2-Pb) was a kind gift from the laboratory of Dr. Charles Sawyers, which previously developed and reported the cell line 23 . These lines are routinely surveyed for contamination by scrutinizing morphology, cell kinetics and testing for mycoplasma.
Transgenic KLK2 mouse models. A transgenic mouse model expressing prostate tissue specific activated hK2 was used. Here, site-directed mutagenesis of APLILSR to APLRTKR at positions 4, −3, and −2 the zymogen sequence of KLK2 was performed using a Quick Change Lightning Mutagenesis Kit (Stratagene). This enabled furin, a ubiquitously expressed protease in rodent prostate tissue, to efficiently cleave the short activation peptide at the cleavage site (−1 Arg/ + 1 Ile), resulting in enzymatically active hK2. Sequencing was performed to verify the genotype using the following primers: 5′-TTC TCT AGG CGC CGG AAT TA-3′ (forward), 3′-CCC GGT AGA ATT CGT TAA CCT-3′ (reverse). This construct was cloned into a SV40 T-antigen cassette downstream of the short rat probasin (pb) promoter and microinjected into fertilized mouse embryos (C57BL/6) and implanted into pseudopregnant female mice. A cancer-susceptible transgenic mouse model with prostate specific hK2 expression was created by crossing the pb_KLK2 transgenic model with the Hi-Myc model (ARR2PB-Flag-Myc-PAI transgene). The Hi-Myc x pb_KLK2 GEM model used in the studies has previously been described in detail 10 . Integration of genes into the genome of the offspring was confirmed by Southern blot analysis and PCR. Mice were monitored closely in accordance with IACUC-established guidelines and RARC animal protocol (# 04-01-002).  Step 4 Step 1 Step 3 Step 2 DNA damage response Fig. 7 The mechanism of action in alpha particle-promoted feed-forward oncoaddiction. The interaction of α-particles with tumor tissue produces DNA damage. The ensuing transcriptional response prompts the hormone-DNA circuit to upregulate AR expression. KLK2 expression subsequently upregulates due to AR and produces more hk2 target epitope for [ 225 Ac] hu11B6 developed after 3-7 weeks. Osseous tumors were established in the tibia of the right hindlimb of BALB/c nude mice using LNCaP-AR, as previously reported 24 , and monitored by bioluminescent imaging.
Pharmacokinetic tissue distribution. Biodistribution studies were conducted to evaluate the uptake and pharmacokinetic distribution of Zirconium-89 labeled hu11B6 ([ 89 Zr]hu11B6) in human prostate cancer xenograft and GEM models. Mice received a single 5.55 MBq dose of [ 89 Zr]hu11B6 (0.150 mCi of activity on 5-50 μg of protein) or a single 11.1 kBq dose of [ 225 Ac]hu11B6 (300 nCi on 5 μg antibody) for injection via intravenous tail-vein injection (t = 0 h). Animals (n = 4-5 per group) were euthanized by CO 2 asphyxiation at 4, 48, 120 and 360 h postinjection of [ 225 Ac]hu11B6; or at 120 and 340 h post-injection of reporter [ 89 Zr] hu11B6. Blood was immediately harvested by cardiac puncture. Tissues (including the tumor) were removed, rinsed in water, dried on paper, weighed, and counted on a gamma-counter using a 370-510 KeV window at secular equilibrium. Aliquots (0.020 mL) of the injected activities were used as decay correction standards and background signal was subtracted from each sample. The percentage of injected activit per gram of tissue weight (%IA/g) was calculated for each animal and data plotted as mean ± SD. Statistical analysis of data was performed using Prism software (Graphpad Software Inc, La Jolla, CA).
Bioluminescence imaging. In vivo AR-activity was recorded by measuring luciferase activity of LNCaP-AR/luc s.c. bilateral xenografts in BALB/c nude mice. These tumors co-expressed exogenous AR and the AR-dependent luciferase reporter vector ARR2-Pb-Luc. Mice were placed under anesthesia with isoflurane prior to retro-orbital injection of 10 μL D-Luciferin (30 mg/mL, dissolved in PBS). A sequence of images with a range of acquisition times was acquired immediately after injection. Radiance (photons/s) was recorded (using Living Image® 4.5.2) from each individual tumor and divided by its respective volume measured by caliper (mm 3 ).
Therapy studies. Study I examined a single 300 nCi activity of [ 225 Ac]hu11B6, [ 225 Ac]hu11B6H435A or 225 Ac-labelled non-specific huIgG 1 injected in three groups of ten male LNCaP-AR s.c. mice. Length (l) and width (w) of the tumors were measured by caliper and the volume for a rotated ellipsoid (V = ½ w2l) was calculated. Weight loss of 20% or a tumor diameter exceeding 15 mm was set as an endpoint. Study II examined two groups of 15-16 week old male Hi-Myc x pb_KLK2 GEM that were randomly selected for specific treatment with 300 nCi of [ 225 Ac]hu11B6 (n = 6) or vehicle (n = 6) at week 40. Tumor progression in each animal in both treatment arms was followed by longitudinal MRI. Information on treatment was blinded to the readers analyzing the imaging data. Volumetric MRI measurement of the three prostate lobes was collected over time and compared to tumor growth progression between [ 225 Ac]hu11B6 and non-treated Hi-Myc x pb_KLK2 animals.
Competitive binding of [ 225 Ac]hu11B6 and [ 89 Zr]hu11B6. Biotinylated 11B6 (100 μL; 2 mg/L) was added to streptavidin-coated microtiter plates, followed by 1 h of incubation with shaking. The plate was washed, after which 20, 100, 200, 400 or 1000 µg of compound (antibody) in 100 μL of DELFIA Assay Buffer was added to the wells, in duplicates, to compete with the capture antibody. Samples containing 0.34 ng/mL or 3.4 ng/mL in 100 μL of DELFIA Assay Buffer were then added to the wells. After 2 h incubation with shaking, the plate was washed, and the Eu 3+ labeled tracer antibody 6H10 was added (200 μL; 0.5 mg/L). The plate was incubated for 1 h with shaking and then washed. DELFIA Enhancement Solution (200 μL) was added and the time-resolved fluorescence was measured 5 min later.
RNA isolation and quantitative PCR. Tissue samples were flash-frozen and immediately used for RNA extraction. Approximately 20 mg of tumor sample was used for RNA extraction. Samples were homogenized using a bullet blender in 600 µL of buffer RLT (RNeasy mini kit) with β-mercaptoethanol. The lysates were centrifuged for 3 min at full speed and the supernatant was used for RNA extraction using RNeasy mini kit (QIAGEN). On column DNA digestion was done using RNase-Free DNase set (QIAGEN). RNA quality and quantity was determined using a spectrophotometer at 260 and 280 nm (Nanodrop-2000, Thermo Scientific). cDNA was generated using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems; Life Technologies). Quantitative-PCR was done using RT2 SYBR Green Fluor qPCR Mastermix and RT2 qPCR primers (Qiagen) on a CFX96 Touch Real-Time PCR Detection System (Bio Rad). KLK2 (hK2), KLK3 (PSA), AR (androgen receptor), FOLH1 (PSMA) expression was quantified relative to ACTB (beta actin) using the comparative CT method.
Tissue histology and autoradiography. After mice were euthanized a tissue package containing prostate lobes, seminal vesicles and prostatic urethra was surgically excised and incubated in Tissue-Tek optimal cutting temperature compound (Sakura Finetek USA, Inc) on ice for 45 min, and then snap-frozen on dry ice in a cryomold. Sets of contiguous 15 µm thick tissue sections were cut using a CM1950 cryostat microtome (Leica Microsystems Inc) and arrayed onto Super-frostPlus glass microscope slides (Thermo Scientific). The slides were exposed for 144 h on phosphor plates and read by a Fujifilm BAS-1800II bio-imaging analyzer (Fuji Photo Film Co.) generating digital images with 50 μm pixel dimensions. Digital images were obtained with an Olympus BX60 System Microscope (Olympus America, Inc.) equipped with a motorized stage (Prior Scientific, Inc.).
Magnetic resonance imaging. We used MRI to assess treatment response in our prostate GEM models. Mouse prostate MR images were acquired on a Bruker 4.7 T Biospec scanner operating at 200 MHz and equipped with a 400 mT/m ID 12 cm gradient coil (Bruker Biospin MRI GmbH, Ettlingen, Germany). A custom-built quadrature birdcage resonator with ID of 32 mm was used for RF excitation and acquisition (Stark Contrast MRI Coils Research Inc., Erlangen, Germany). Mice were anesthetized with oxygen and 1% isoflurane gas. Animal breathing was monitored using a small animal physiological monitoring system (SA Instruments, Inc., Stony Brook, New York). T2 weighted scout images along three orthogonal orientations were first acquired for animal positioning. The T2-weighted fast spinecho RARE sequence (rapid acquisition with relaxation enhancement) was used to acquire axial mouse pelvic images with a slice thickness of 0.8 mm, FOV 40 mm × 35 mm with a spatial resolution of 117 mm × 133 mm using the following acquisition parameters: TR = 3824.4 s, TE = 48 ms, RARE factor 8 for an average of eight scans. Following on-line reconstruction, data were exported to FIJI (NIH) the area of each individual lobe was assessed. Cystic dilations were not included in our volumetric measurements.
Positron emission tomography. PET/CT imaging of male crab-eating macaque (Charles River) was performed with a Siemens Biograph64 mCT PET/CT system. PET images were acquired at 1, 24, 72, 144, 216, and 384 h after intravenous injection (IV) of 10 mg of 2.50 mCi (92.5 MBq) [ 89 Zr]hu11B6. Whole body acquisitions consisting of six bed positions covering from the crown of the animal's head to mid-tibia. The head and feet were scanned with 4 min per bed, while the beds over the pelvic region were scanned with 14 min per bed, for a total imaging time (including gaps between scans) of about 44 min. Animals were maintained under 2% isoflurane/oxygen anesthesia during the scanning. Data was exported in raw format and the rigid body (3 degrees of freedom) co-registration between PET and CT data was performed in Amira 5.3.3 (FEI). Amira and FIJI were used to produce the majority of the figures in the manuscript. Siemens Syngo Multi-Modality Workplace software (version VE40A) was used to analyze and quantify [ 89 Zr]hu11B6 uptake on PET images.
[ 223 Ra]RaCl 2 and [ 225 Ac]hu11B6 bone metastasis localization. Using an intratibial inoculation model of metastasis, we evaluated the acute microdistribution of [ 225 Ac]hu11B6 compared to [ 223 Ra]RaCl 2 in intratibial lesions of LNCaP-AR, which is a mixed osteolytic and osteoblastic bone lesion that we have previously evaluated and reported 10,24 . Tumor growth was monitored by bioluminescence and x-ray computed tomography (CT) using the IVIS SpectrumCT (Perkin Elmer; Waltham, MA). The tumor bearing hindlimb of the mice were cryosectioned 24 and 120 h following injection by [ 223 Ra]RaCl 2 (150 nCi; 5.5 kBq) and [ 225 Ac]hu11B6 (300 nCi; 11.1 kBq), respectively, and scanned by α-camera, as previously described 26 . Briefly, no fixation or decalcification chemicals were applied to mouse tissues, which were either stored (−80 o C) or flash-frozen in liquid nitrogen. Autoradiography was performed on fresh-frozen sections cut on a custom-modified (cryo-cooled) Leica 1860 cryostat at 10 µm. Alpha camera images consist of undecalcified tissue sections that were placed in direct contact with silver activated zinc sulfide scintillant (Eljen Technologies) and imaged using a cryocooled EMCCD (Photometrics CascadeII). The adjacent section was then processed for Safranin-O development for bone mineral and proteoglycan localization.
Radiochemistry. 225 Ac (ORNL, Oak Ridge, TN) was conjugated to the hu11B6 antibody or isotype-matched control antibody and purified using a 2-step labeling procedure 4,20 . Activity was measured at secular equilibrium with a Squibb CRC-17 Radioisotope Calibrator (E.R. Squibb and Sons, Inc., Princeton, NJ) set at 775 and multiplying the displayed activity value by 5. The radiochemical purity of the final product, [ 225 Ac]hu11B6, was determined using instant thin-layer chromatography (ITLC) with a stationary phase of silica gel impregnated paper (Gelman Science Inc., Ann Arbor, MI) and two different mobile phases. Mobile phase I is 10 mM ethylenediaminetetraacetic acid and II is 9% sodium chloride/10 mM sodium hydroxide. The strips were counted in a Packard Cobra γ-counter (Packard Instrument Co., Inc., Meriden, CT) using a 370-510 KeV window. The purified radioimmunoconstruct was formulated in a solution of 1% human serum albumin (HSA, Swiss Red Cross, Bern, Switzerland) and 0.9% sodium chloride (Normal Saline Solution, Abbott Laboratories, North Chicago, IL) for intravenous injection. Radiochemically pure 223 Ra was eluted from an Actinium-227 source, as described 27 and formulated in 0.03 M citrate in saline. An empirically derived calibrator setting of #277, using a CRC-127R dose calibrator (Capintec Inc), was used to dose 223 Ra. 89 Zr (MSKCC Radiochemistry and Imaging Probes (RMIP) Core Facility or 3D Imaging, Little Rock AR) was conjugated to a hu11B6-DFO antibody as previously described 10,24 . Activity was measured with a Squibb CRC-17 Radioisotope Calibrator set at #465. Briefly, 8.5 mCi of 89 Zr[Zr]oxalate was neutralized with 0.025 mL of 1 M Na 2 CO 3 to pH 7.0. Any colloidal precipitate was removed with centrifugal filtration (0.22 µm Corning Spin-X filter at 1000×g) and 0.25 mL of 11B6-DFO (8.8 g/L; Fuji Film; lot # NBS0131-6-2 (Prost B3298) in 25 mM sodium acetate buffer (pH 5.5) was added to the clear neutralized 89 Zr[Zr] filtrate. The reaction mixture was purified after 75 min at ambient temperature by size exclusion chromatography using a 10DG column (BioRad) mobile phase and a 1% HSA mobile phase. The radiochemical purity of the final product, [ 89 Zr]hu11B6, was assayed using ITLC (silica gel impregnated paper and Mobile Phases I and II) as described above.
Binding affinity of labeled and unlabeled hu11B6. The binding kinetics for the association (k a ), dissociation (k d ) and affinity constant (K D ) of native hu11B6 vs. radioimmunoconjugates were used as ligands captured on a Protein A sensor chip (29127558, GE Healthcare). Antibody capture was accomplished by diluting the unmodified hu11B6 antibody and the various immunoconjugates to a concentration of 1 μg/mL in HBS-EP buffer (BR100188, GE Healthcare), and flowing the solution over a protein A sensor chip for 1 min at a flow rate of 5 μL/ min. Recombinant human hK2 (Turku University, Finland) 28 was used as the analyte that was flowed over the protein A sensor chip having the captured hu11B6 antibody or derivative. The binding kinetics of the antibody (ligand) captured on the sensor chip was evaluated across a concentration series of the hK2 antigen (analyte) at 0, 3.125, 6.25, 12.5, 25, 50, 100, and 200 nM in HBS-EP buffer. Each concentration of analyte was injected for 5 min at a flow rate of 5 µL/min to allow it to bind to the antibody captured on the protein A sensor chip. Next, the binding buffer (HBS-EP) was allowed to flow over the sensor chip for 15 min (5 µL/min) to allow dissociation of the antigen from the 11B6 antibody-hK2 antigen immunocomplex on the chip. Finally, regeneration buffer (10 mM glycine/HCl, pH 2.5) was passed over the chip surface for 1 min (5 µL/min) to effect complete dissociation of captured antibody. HBS-EP buffer was flown (5 µL/min) over the chip for 2 min to stabilize the protein A chip surface prior to the injection of the next sample in the concentration series of the analyte as described above. Biacore control software 3.2 was used to analyze the kinetic data and the 1:1 binding with mass transfer fit was used to derive kinetic constants for the interaction between the various 11B6 immunoconjugates and purified hK2 antigen.
Statistical analysis. Statistical analysis of data was performed using Prism 6 software (Graphpad Software Inc, La Jolla, CA). Data are displayed as means with their standard errors (SEM). P-values for comparisons between treatment groups were obtained in GraphPad Prism using Student's t-test for statistical significance (unpaired, two-tailed t-test). p-Values <0.05 were considered statistically significant. Log-rank (Mantel-Cox) and Mantel-Haenszel tests were used in Prism software to compare survival outcomes.
Data availability. All data generated or analyzed during this study are available within the article and Supplementary Files, or available from the authors upon request.