Treatment of Prostate Cancer with CD46-targeted ²²⁵Ac Alpha Particle Radioimmunotherapy

Prostate cancer is the most diagnosed noncutaneous malignancy and second leading cause of death in men worldwide in the year 2021 (1). While early-stage tumors may be effectively treated with surgery, radiation, or androgen deprivation therapy, a significant fraction of patients progress to metastatic, castration-resistant prostate cancer (mCRPC), which is refractory to these treatments. Therapeutic options for patients with mCRPC include androgen signaling inhibitors, chemotherapy, immunotherapy, and radiopharmaceutical therapies including ¹⁷⁷Lu-PSMA-617 (Lu-177 vipivotide tetraxetan) or ²²³Ra (2, 3). Those treatments are not curative, and patients eventually progress. As such, there is a need for preclinical and translational development of new treatments.

In prostate cancer, prostate-specific membrane antigen (PSMA)-targeted radiopharmaceutical therapies have overwhelmingly attracted the most attention. PSMA is a type II glycoprotein overexpressed on prostate cancer cells. Currently, PSMA PET imaging using ⁶⁸Ga-PSMA-11 or ¹⁸F-DCFPyL is used in routine clinical care to localize and stage patients, and ¹⁷⁷Lu-PSMA-617 may be used for treatment (4). However, a significant fraction of patients with mCRPC are PSMA negative at diagnosis or develop PSMA negativity on imaging over the course of treatment (5–8). Intrapatient heterogeneity of PSMA expression is one of the reasons for resistance to the PSMA-directed therapies (9). In addition, the prolonged treatment with androgen deprivation therapy could also result in treatment-emergent neuroendocrine/small cell prostate cancer (t-SCNC) with low PSMA expression (10). Therefore, there is a need to identify and evaluate other targets in mCRPC.

We have previously used a non–gene expression–based approach to identify novel tumor cell surface targets, including conformational and posttranslationally modified epitopes. We selected, by laser capture microdissection, billion-member phage human antibody display libraries on prostate cancer patient specimens following counter selection on normal cells and tissues, and identified a panel of human antibodies binding to tumor selectively in vitro and in vivo (11–13). We identified one of the target antigens as CD46 (14). CD46 is best known for being a negative regulator of the complement activation process (15). We showed that CD46 maintains a high level of expression in prostate cancer across differentiation patterns (14). Compared with PSMA, CD46 is more homogeneously expressed, especially in patients who become resistant to androgen blockade, including t-SCNC (14). As YS5 binds to cancer specific epitope, little binding is observed in healthy tissues, with the exception of placental trophoblasts, and to a lesser extent the prostate epithelium (14). We developed and optimized a CD46-targeting human antibody, YS5, which binds to a tumor-selective conformational epitope, enabling therapeutic targeting of CD46. We have developed an antibody–drug conjugate (ADC) based on YS5, demonstrated its preclinical efficacy in xenograft models, and shown that it is well tolerated in non-human primates (14). The ADC (FOR46) is being tested in patients with mCRPC, as a single agent (phase I, NCT03575819) and a combination treatment (phase I/II, NCT05011188), with promising early results reported previously (16). To enable patient stratification, we have also developed an immunoPET agent based on YS5, ([⁸⁹Zr]DFO-YS5), and shown that it detects adenocarcinoma, neuroendocrine prostate cancer (NEPC), and PSMA-positive or -negative tumor xenografts in preclinical studies (17). A first-in-human study of this PET agent is ongoing in patients with mCRPC (NCT05245006).

Radioimmunotherapy (RIT) exploits target-specific affinity agents to deliver therapeutic radioisotopes specifically to the target tissue. By using an antibody-based targeting approach, many RIT agents have been produced for clinical applications. Alpha-particle therapies demonstrate both potential advantages as well as limitations when compared with other radiopharmaceutical therapy methods, including greater efficacy, but also at the cost of potentially increased toxicity. Actinium-225 (²²⁵Ac), an alpha-emitting radionuclide, is being explored for prostate cancer and other malignancies (18). The advantage of using ²²⁵Ac is that it has a long half-life of 9.9 days, and it decays by emitting net four alpha particles generated from ²²¹Fr (6.3 MeV), ²¹⁷At (7.1 MeV), ²¹³Bi (5.8 MeV), and ²¹³Po (8.4 MeV). Early clinical studies in mCRPC show that ²²⁵Ac-PSMA-617 has efficacy, although treatment is associated with xerostomia, and the daughter isotopes are believed to accumulate in kidneys, causing renal toxicity (19). A number of investigators are evaluating ²²⁵Ac-labeled small molecules or antibodies for prostate cancer in both preclinical and clinical studies (20, 21). These studies reveal potential for ²²⁵Ac-based radiopharmaceutical therapy in prostate cancer, although none of the agents have yet emerged as clinical standard of care. Herein, we report the development and preclinical evaluation of an alpha-particle therapy against CD46 by labeling the anti-CD46 antibody (YS5) with ²²⁵Ac ([²²⁵Ac]DOTA-YS5).

[²²⁵Ac]DOTA-YS5 binding as well as internalized fractions were determined using 22Rv1 and DU145 cells. The detailed experimental procedure has been given in Supplementary Data.

All the animal experiments were carried out in compliance with Institutional Animal Care and Use Committee established guidelines at Laboratory Animal Resource Center, University of California, San Francisco, CA. All the mice used in our studies were male mice. Methods for animal studies including toxicity, subcutaneous xenograft preparation, patient-derived xenograft (PDX) inoculation, and microPET scan (for ⁶⁸Ga-PSMA11, [⁸⁹Zr]DFO-YS5 imaging and biodistribution) were described in the Supplementary Data.

The YS5 human IgG1 antibody was produced in HEK293 cells (ATCC, RRID:CVCL_0045) following transient transfection, and produced by protein affinity chromatography followed by ion exchange chromatography as described previously (14). The conjugation of p-SCN-Bn-DOTA on YS5 was carried out according to our previously published method with slight modifications (17). Briefly, YS5 (5 mg) in HEPES buffer was exchanged with 0.1 mol/L Na₂CO₃/NaHCO₃ buffer (pH = 9.0) using YM30K centrifugal filter unit (Amicon ultra-0.5 mL, regenerated cellulose, Merck Millipore Ltd.). It was further diluted with 0.1 mol/L Na₂CO₃/NaHCO₃ buffer (pH = 9.0) to adjust the final volume to 1 mL. A total of 20 eq. of p-SCN-Bn-DOTA in DMSO (4.5 mg, 20 μL) was added to the solution containing 5 mg of YS5 followed by incubation at 37°C for 1 hour. The reaction mixture was purified using a PD10 gel filtration column by eluting with 0.25 mol/L NaOAc buffer, pH = 6. The DOTA-conjugated YS5 was stored in −20°C until the radiolabeling reactions.

For radiolabeling, 1 mCi of ²²⁵Ac(NO₃)₃ was received from Oak Ridge National Laboratory by the ²²⁹Th decay pathway in solid form, and it was dissolved in 0.2 mol/L HCl (100 μL). The radiolabeling was performed by incubating the 50 μCi (6 μL) of ²²⁵Ac(NO₃)₃, 2 mol/L NH₄OAc (50 μL, pH = 5.8), L-Ascorbic acid (20 μL, 150 mg/mL), and YS5-DOTA (200 μg, 15.64 μL, 12.785 mg/mL) at 40°C for 2 hours. The radiolabeling progress was monitored by radio thin layer chromatography (TLC) by eluting with a mobile phase 10 mmol/L EDTA (pH = 5.5) on iTLC-SG (Gelman Science Inc.). The radiolabeled antibody was diluted with 300 μL of 0.9% saline, transferred to a YM30K centrifugal filtration unit (Millipore), and centrifuged at 10,640 × g for 10 minutes. Then, 200 μL of 0.9% saline was added to the centrifugal filtration unit, followed by centrifugation at 10,640 × g for 5 minutes. The [²²⁵Ac]DOTA-YS5 (20 μCi, 40%) conjugate was isolated, and the purity was assessed by radio-TLC [iTLC-SG 10 mmol/L EDTA (pH = 5)] immediately after purification.

To confirm the stability and aggregation of the radiolabeled antibody, size exclusion chromatography (SEC) was performed. For this, a Merck Hitachi LaChrom Elite system comprised of an L-2130 Pump, L-2200 autosampler, L-2450 DAD detector, L-2400 UV detector, and Carrol and Ramsey Associates model 105S radioactivity detector was utilized. 1X PBS buffer was used as a mobile phase using a column BioSep 5 μmol/L SEC-s3000 290 Å (Phenomenex, Inc.) with a flow rate of 1 mL per minute. One-mL fractions were collected and then counted on Hidex automatic gamma counter when secular equilibrium was reached at 24 hours. The chromatogram was plotted as counts per minute (CPM) versus time. The stability of the purified radioimmunoconjugate [²²⁵Ac]DOTA-YS5 (0.05 mL, 7.43 μCi) was verified by incubating with human serum (0.45 mL; Sigma-Aldrich) and 0.9% saline (0.45 mL, Medline) at 37°C at various timepoints (0 hour, 1 day, 2 days, 3 days, 4 days, 5 days, 7 days, and 14 days). At each timepoint, an aliquot of 8 μL was spotted in duplicate on iTLC-SG and eluted with 10 mmol/L EDTA (pH = 5.5). The decomplexation was monitored by radio TLC and allowed for 24 hours to reach secular equilibrium before counting on a BioScan Ar 2000 radio-TLC Imaging scanner.

To study the biodistribution of [²²⁵Ac]DOTA-YS5, 22Rv1 xenograft-bearing mice were used. These mice with tumor volumes ranging from 200 to 300 mm³ received a 0.5 μCi activity of [²²⁵Ac]DOTA-YS5 (specific activity 0.1 μCi/μg) via tail vein injections. Following this, the blood, tumor, and selected organs were collected after 24 hours (day 1), 48 hours (day 2), 96 hours (day 4), 168 hours (day 7), 264 hours (day 11), 336 hours (day 14), and 408 hours (day 17) to study activity present in the tissue. Following this, the presence of [²²⁵Ac]DOTA-YS5 in each organ was counted (gamma energy window: 25 to 2,000 keV) with the gamma counter (HIDEX). The standards of injected dose activity were measured in a similar fashion to calculate the %ID/g for each organ without reaching secular equilibrium.

Methodology for histology and immunofluorescence staining of the mice organs as well as tumor tissue has been included in the Supplementary Data.

To study the distribution of [²²⁵Ac]DOTA-YS5 within the tissue, iQID (ionizing radiation quantum imaging detector) camera imaging was performed. For this, tumor, liver, kidneys, spleen, heart, and lung tissues from the mice injected with a 0.5 μCi activity of [²²⁵Ac]DOTA-YS5 were collected in an Optimum Cutting Temperature (OCT) medium. The tissues in OCT were sliced (20-μm thick) and mounted with a scintillator on the iQID camera (QScint Imaging Solutions, LLC) and acquisition of alpha particles was done for 24 hours at 1.5 V.

For all treatment studies, the body weights and tumor measurements for each mouse were measured until the mice reached a humane endpoint including body condition score below 2, weight loss by 20%, or tumor volume of 2,000 mm³. The tumor measurements were carried out twice a week, and tumor volumes were calculated by formula V = [length × (width)²]/2.

Animals were randomized in groups, and [²²⁵Ac]DOTA-YS5 injections were performed when the tumor reached 100 mm³. The 22Rv1 and DU145 xenograft-bearing mice were injected with saline, 0.25 μCi, or 0.5 μCi of [²²⁵Ac]DOTA-YS5 (specific activity 0.1 μCi/μg). In addition, a separate group received a 0.5 μCi dose of [²²⁵Ac]DOTA-IgG. For fractionated dose study, the treatment group was administered with three doses of 0.125 μCi of [²²⁵Ac]DOTA-YS5 on day 1, day 37, and day 49.

The LTL-545–bearing mice were injected with saline and [²²⁵Ac]DOTA-YS5 (0.03, 0.06, 0.12, and 0.25 μCi) with specific activity of 0.11 μCi/μg. For Fc blocking, native IgG (0.5 mg/mouse) was used. Similarly, the LTL-484 mice were injected intravenously with 0.06, 0.12, and 0.25 μCi doses of [²²⁵Ac]DOTA-YS5 activity when tumors reached to 100 mm³.

The results reported are presented as mean ± SD and plotted using Graphpad Prism software. To compare the blood cells and liver and kidney function tests between saline control and treated groups, one-way ANOVA was used with Dunnett multiple comparisons test. The significance of median survival from saline control and treatment groups was determined by log-rank (Mantel–Cox) test. Students t test was used to compare the number of phospho-γ-H2AX foci in sections from saline-treated and [²²⁵Ac]DOTA-YS5–treated mice groups.

The data presented in the current study are available within the article and Supplementary Data or from the corresponding author on reasonable request.

Western blot analysis of the series of PDXs was carried out to study the expression of CD46 and PSMA. For this study, we utilized well-characterized PDXs from the Living Tumor Laboratory (22–24). As shown in Fig. 1A, PSMA expression was restricted to the adenocarcinoma subtypes we evaluated, while CD46 expression was found in not only adenocarcinoma subtypes but also some neuroendocrine prostate cancer models. IHC staining on PDX tissues for PSMA (Supplementary Fig. S1) showed a similar expression pattern to that observed in Western blots. Our previous study reported that PSMA-positive, androgen receptor (AR)-positive (LTL-484, LTL-331) and PSMA-negative, AR-negative (LTL-545, LTL-331R) xenografts could be readily detected with a CD46 PET scan using [⁸⁹Zr]DFO-YS5 probe, including the LTL-331, LTL-331R, and LTL-545 models (17). LTL-545 and LTL-331R tumor models used in our experiments are neuroendocrine tumors, while LTL-484 and LTL-331 are adenocarcinoma tumors. Herein, we further demonstrate high uptake in the LTL-484 model, calculated at 23.7% ± 4.7%ID/g based on region of interest analysis of microPET/CT images (Fig. 1B). Taken together with our prior report, these data demonstrate that CD46 is highly expressed across a panel of clinically relevant PDX, and a high amount of [⁸⁹Zr]DFO-YS5 can be delivered and detected using PET imaging.

To compare the results of CD46 targeting [⁸⁹Zr]DFO-YS5 with a clinically used PSMA targeting probe, we performed microPET/CT scans and biodistribution (Fig. 1C and D; Supplementary Table S1) using ⁶⁸Ga-PSMA11 in LTL-545, LTL-484, LTL-331, and LTL-331R PDXs. Tumor uptake of ⁶⁸Ga-PSMA11 was low to moderate; %ID/g for LTL-545, LTL-484, LTL-331, and LTL-331R was found to be 3.32 ± 0.40, 4.41 ± 1.11, 2.29 ± 0.82, and 0.43 ± 0.13, respectively (Fig. 1D). The uptake seen in the LTL-545 model, which is unexpected because of low PSMA expression by Western blotting analysis, may be attributable to differences in vascularity, or to non–PSMA-mediated binding as described recently (22). These data suggest that CD46 is an actionable target in prostate cancer for imaging and therapy using the YS5 antibody, including in PSMA-low and -negative tumors.

We adopted a standard approach for antibody radiolabeling utilizing DOTA chelator and ²²⁵Ac (Fig. 2A; ref. 25). The characterization of the conjugated antibody activity was done utilizing a recently developed immunoreactivity assay (26). First, the conjugation of DOTA to the YS5 antibody was performed in 0.1 mol/L Na₂CO₃/NaHCO₃ buffer, followed by SEC purification. The successful conjugation and number of DOTA chelator on YS5 was confirmed with matrix-assisted laser deposition/ionization–time of flight (MALDI-TOF) mass spectroscopy (Supplementary Fig. S2A), which revealed that an average of 8.7 ± 0.2 (n = 2) equivalents of DOTA were conjugated to each YS5. The DOTA-YS5 was stored at −20°C until radiolabeling reactions were carried out. The isolated yield of [²²⁵Ac]DOTA-YS5 from [²²⁵Ac]Ac(NO₃) was 39.4% ± 4.4% (n = 6) with a specific activity of 0.10 ± 0.01 μCi per μg of antibody (n = 6). As shown in iTLC in Fig. 2B, purity of radiolabeled antibody was 94.8% ± 8.1% (n = 3). In addition, SEC analysis (Supplementary Fig. S2B) showed no aggregation of the [²²⁵Ac]DOTA-YS5 after the labeling and purification process. In vitro stability of the radioimmunoconjugate, [²²⁵Ac]DOTA-YS5 was evaluated in human serum and saline buffers at 37°C for 14 days. As shown in Supplementary Fig. S2C, >85% of radioimmunoconjugate was intact at all the timepoints except for the 14th day measurement in the saline buffer (∼55.24% is intact). The beads-based binding assay (26) results shown in Fig. 2C confirm that binding of [²²⁵Ac]DOTA-YS5 was 86.4% ± 1.5%, which was reduced to 17.4% ± 0.5% in the blocking sample where 30-fold excess of cold (unlabeled) YS5 was added, and 14.3% ± 1.0% in non–antigen-coated beads. These measurements were performed without waiting for secular equilibrium of the isotope, and similar results were also found after secular equilibrium (Supplementary Fig. S2D). Overall, these results demonstrate that [²²⁵Ac]DOTA-YS5 was efficiently radiolabeled with retention of immunoreactivity.

Next, we studied binding and cytotoxicity of [²²⁵Ac]DOTA-YS5 to prostate cancer cells in vitro. For binding, 22Rv1 cells incubated with [²²⁵Ac]DOTA-YS5 at 7 nmol/L showed 19.25% ± 0.70% binding (Supplementary Fig. S3), which was reduced to 4.46% ± 0.17% by the addition of 2-fold cold YS5. The in vitro uptake of [²²⁵Ac]DOTA-YS5 was studied on 22Rv1 and DU145 cells. Results in Fig. 2D and E show number of cell-associated (membrane-bound, internalized, and total) molecules for the cells incubated with [²²⁵Ac]DOTA-YS5. Interestingly, the two cell lines used in this study show different binding and internalization patterns. In case of 22Rv1 cells (Fig. 2D), most of the [²²⁵Ac]DOTA-YS5 was internalized or bound in the first 30 minutes of incubation, whereas the membrane-bound activity increased up to 24 hours. Total cell-associated molecules (membrane-bound + internalized) for 22Rv1 cells at 24 hours were 9.7 × 10¹¹ ± 4.5 × 10¹⁰, which suggests that a greater number of [²²⁵Ac]DOTA-YS5 molecules were interacting with CD46 on cell surface over time. In contrast, the total number of [²²⁵Ac]DOTA-YS5 antibody molecules bound to DU145 cells (Fig. 2E) increased from 8.0 × 10¹⁰ ± 7.6 × 10⁹ at 30 minutes to 4.5 × 10¹¹ ± 1.9 × 10⁹ molecules at 24 hours.

The cell-killing activity of [²²⁵Ac]DOTA-YS5 was studied with cell viability and clonogenic assays. A dose-dependent reduction in cell viability was observed by the treatment of [²²⁵Ac]DOTA-YS5 on 22Rv1 cells (Fig. 2F). The IC₅₀ of [²²⁵Ac]DOTA-YS5 and control nonspecific [²²⁵Ac]DOTA-IgG were found to be 80.1 ± 23.5 pCi/mL, and 7.34 ±0.9 nCi/mL, respectively. In the clonogenic assay where tumor cells grow into large colonies, IC₅₀ of [²²⁵Ac]DOTA-YS5 was determined to be 10.09 ± 3.61 nCi/mL (Fig. 2G). These data suggest that [²²⁵Ac]DOTA-YS5 has high specific binding and potent cytotoxicity against prostate cancer cells.

Because imaging of [²²⁵Ac]DOTA-YS5 is challenging owing to low emission rate of imageable photons, we utilized ex vivo biodistribution studies to assay the tissue distribution of the radiopharmaceutical. Mice bearing 22Rv1 xenografts (PSMA positive, AR positive) were administered 0.5 μCi activity of [²²⁵Ac]DOTA-YS5 via tail vein. As shown in Fig. 3A and Supplementary Table S2, tumor uptake of [²²⁵Ac]DOTA-YS5 at 24 hours (day 1) post injections (p.i.) was 11.64% ± 1.37%ID/g, which kept increasing up to 96 hours (day 4) to 28.58% ± 10.88%ID/g. The average %ID/g for 168 hours (day 7), 264 hours (day 11), and 336 hours (day 14) were found to be 29.35 ± 7.76, 18.10 ± 8.12, and 21.12 ± 12.30, respectively (Fig. 3A). Similarly, the last observation on 408 hours p.i (day 17) showed uptake of 31.78% ± 5.89%ID/g. The %ID/g for tumors was higher than nontargeted organs for all timepoints, except 24 hours, where a high uptake was seen in kidneys. Tumor to blood, tumor to kidney, and tumor to muscle ratios (Fig. 3B; Supplementary Table S3) increased from day 1 to day 17, suggesting the clearance of the [²²⁵Ac]DOTA-YS5 in most of the organs including muscle and kidneys, with simultaneous accumulation in tumor tissue. As seen from gamma energy spectra in Fig. 3C and Supplementary Fig. S4, the higher uptake seen in kidneys at 24 hours could be due to the accumulation of ²¹³Bi (from ²²⁵Ac decay) in the kidneys, a phenomenon that has been reported by others (27, 28). This was not observed in other tissues, including the bones (Supplementary Fig. S4). The reduction in activity of blood from 24 hours (11.64 ± 1.37%ID/g) to 408 hours (1.71 ± 0.47%ID/g) indicates gradual clearance via liver and kidney (Fig. 3A).

Supplementary Figure S5 shows dosimetry analysis from the biodistribution studies. Dosimetric analysis was performed by considering ²²⁵Ac as a dominant emitter of alpha particles and primary contributor for therapeutic effect; the daughter isotopes were not included in the calculations because there is an uncertainty that all daughters stay at the same site and the dose contributions are minor compared with that from alpha emission of ²²⁵Ac. The tumor with an average weight of 0.5 g received an average 37 Sv of equivalent dose with 0.5 μCi injections of [²²⁵Ac]DOTA-YS5. The highest amount was delivered to tumor tissue, suggesting the potential of antitumor response from [²²⁵Ac]DOTA-YS5. In addition, normal healthy organs, including bone, kidneys, stomach, and heart, are exposed to high doses (roughly 5 Sv or higher; Supplementary Fig. S5).

Digital autoradiographic imaging (iQID camera) was performed to study tissue distribution of [²²⁵Ac]DOTA-YS5 in tumor and mouse organs. Figure 3D shows iQID camera digital autoradiographs for organs collected on day 1, day 2, day 4, and day 7 after injections of [²²⁵Ac]DOTA-YS5. Signal from [²²⁵Ac]DOTA-YS5 was not uniform in tumor tissues (Fig. 3D). For the remaining organs, distribution of [²²⁵Ac]DOTA-YS5 was homogenous.

As shown in Fig. 4, the 22Rv1 xenografts treated with 0.5 μCi activity of [²²⁵Ac]DOTA-YS5 were stained with hematoxylin and eosin (H&E) and a DNA damage marker protein (phosphorylated γ-H2AX). Figure 4A and B shows that treatment of 0.5 μCi activity dose of the [²²⁵Ac]DOTA-YS5 for day 7 or day 14 increases the γ-H2AX foci in the cell nucleus confirming DNA double-strand breaks. The number of γ-H2AX foci per cell in 7 days treatment of 0.5 μCi activity doses were found to be 4.46 ± 0.75, significantly higher than saline (1.37 ± 0.10 foci per cell, P value, 0.02; Fig. 4A; Supplementary Fig. S6). Similarly, as shown in Fig. 4B and Supplementary Fig. S6, a higher number of the γ-H2AX foci (4.90 ± 0.55 per cell) were also observed for 14 days p.i. of [²²⁵Ac]DOTA-YS5 (P value, 0.02). In addition, DAPI and H&E staining for the tumors from [²²⁵Ac]DOTA-YS5 14 days after treatment (Fig. 4B), showed deformed and pycnotic nuclei confirming high nuclear damage. To further investigate the [²²⁵Ac]DOTA-YS5 uptake and DNA damage response, H&E staining as well as CD46 and γ-H2AX staining of the tumor was correlated with the digital autoradiographs. The results for tumor tissues stained on day 1, day 2, day 4, and day 7 after treatment (Supplementary Fig. S7) show that CD46 was homogeneously expressed, with the exception of some areas of central necrosis visualized on H&E stain. These areas of necrosis also demonstrated reduced uptake of [²²⁵Ac]DOTA-YS5 on the autoradiography. In general, γ-H2AX foci colocalized with CD46 expression.

Initial acute toxicity studies were carried out for 15 days in male nude mice (without tumor) injected with saline, 0.06, 0.125, 0.25, or 0.5 μCi of [²²⁵Ac]DOTA-YS5. Bodyweight measurements shown in Supplementary Fig. S8A indicate no significant weight loss in [²²⁵Ac]DOTA-YS5–treated mice for 15 days. Following the 15 days of observation, mice in these groups were euthanized for further evaluation. The liver and kidney function test results shown in Supplementary Fig. S8B and Supplementary Table S4 demonstrate no significant difference in saline versus treated groups. Along with the liver and kidney function test, complete blood count measurements also showed no significant changes in the blood cell counts for saline versus treated groups (Supplementary Fig. S8C; Supplementary Table S5). Histologic evaluation of the organs collected after 15 days showed no significant changes in the saline versus treated groups.

Following the acute toxicity studies, chronic toxicity studies were performed (Fig. 5; Supplementary Tables S6 and S7). The male nude mice were injected with saline, 0.25, 0.5, or 1 μCi of [²²⁵Ac]DOTA-YS5 and were observed for 117 days for any sign of illness, pain, and body weight loss. All the mice in 1 μCi treatment group were euthanized on 14th day p.i. due to significant toxicity and body weight loss (Fig. 5A). Out of 4 animals in the 0.5 μCi group, a significant body weight loss was observed, and 1 mouse was euthanized on day 103 due to body weight loss by 20%, while the saline and 0.25 μCi groups demonstrated no detectable body weight loss (Fig. 5A and B). On laboratory analysis, creatine and blood urea nitrogen levels were elevated at 0.5 μCi activity level (Fig. 5C; Supplementary Table S6), consistent with renal toxicity. Alkaline phosphatase levels (for liver or bone damage) were also higher in 0.5 μCi treatment (Fig. 5C), while the liver function tests for alanine transaminase and aspartate transaminase were unchanged. There were no significant changes in blood counts for saline versus treated groups (Fig. 5D; Supplementary Table S7).

Histologic evaluation of the organs collected in this chronic toxicity study revealed that 0.5 μCi resulted in significant renal damage. As shown in Fig. 5E–G, for the 0.5 μCi activity level, glomeruli demonstrated marked fibrin deposition in the capillary loops and glomerulosclerosis (Fig. 5E and F). In addition, the renal tubules showed dilatation and flattening with scattered cells containing enlarged nuclei with anisonucleosis and necrotic luminal epithelial cells (Fig. 5G). Other organs including spleen, bone and bone marrow, lungs, heart, and liver showed no histologic evidence of damage (Supplementary Fig. S9).

CD46-expressing cell line–derived prostate cancer xenograft models were used to study therapeutic efficacy of [²²⁵Ac]DOTA-YS5. Cohorts of n = 7 22Rv1 xenograft (PSMA-positive)-bearing mice were randomized to injection with saline, 0.25 or 0.5 μCi activity levels of [²²⁵Ac]DOTA-YS5, or 0.5 μCi of [²²⁵Ac]DOTA-IgG. Nontargeting [²²⁵Ac]DOTA-IgG was kept as a negative control test to assay for nonspecific treatment effects. As shown in Fig. 6A, treatment with 0.25 and 0.5 μCi of [²²⁵Ac]DOTA-YS5 significantly inhibited tumor growth in an activity-dependent manner, while mice in control cohorts (saline and [²²⁵Ac]DOTA-IgG) showed rapid tumor growth. The survival plot (Fig. 6A) indicates that median survival of the mice treated with saline and [²²⁵Ac]DOTA-IgG were 37 and 51 days (P = 0.0012), respectively. In contrast, the median survival of 0.25 and 0.5 μCi groups of [²²⁵Ac]DOTA-YS5 was significantly improved to 103 (P = 0.0006) and 131 days (P = 0.0003), respectively. The body weight measurements of the mice in these groups are shown in Fig. 6A; results indicate that a 0.5 μCi dose of [²²⁵Ac]DOTA-YS5 results in a gradual decrease in body weights over the period, consistent with the toxicity studies. Figure 6A shows that 0.5 μCi dose of [²²⁵Ac]DOTA-YS5 has the highest antitumor response but also greater toxicity. Overall, these data demonstrate that 0.25 and 0.5 μCi doses of [²²⁵Ac]DOTA-YS5 are effective for treatment of 22Rv1 xenografts, with toxicity observed at the 0.5 μCi activity level.

We also tested a fractionated therapy regimen to further evaluate the treatment efficacy of [²²⁵Ac]DOTA-YS5. We administered fractionated dose (three doses) of 0.125 μCi of [²²⁵Ac]DOTA-YS5 to the mice with 22Rv1 xenografts on day 1, day 37, and day 49. As shown in Supplementary Fig. S10, the three fractionated administrations of 0.125 μCi of [²²⁵Ac]DOTA-YS5 delayed tumor growth significantly compared with the saline control group. The median survival of the saline group (17 days) was significantly lower than the study group treated with three doses of 0.125 μCi (71.5 days, P = 0.0002) of [²²⁵Ac]DOTA-YS5 (Supplementary Fig. S10). Out of the 10 mice in the fractionated dose group (0.125 μCi), no mouse was euthanized because of body weight loss, while only 3 mice showed a transient 15% body weight loss (Supplementary Fig. S10), and all mice regained their body weights. These data suggest that a fractionated therapy approach represents a potential method for treatment utilizing [²²⁵Ac]DOTA-YS5.

In addition, a CD46-positive, PSMA-negative, AR-negative prostate cancer cell line–derived xenograft model (DU145) was also employed to study the effect of [²²⁵Ac]DOTA-YS5. As shown in Supplementary Fig. S11, 0.25 and 0.5 μCi dose of [²²⁵Ac]DOTA-YS5 had an antitumor effect that delayed the tumor growth. Saline-treated mice showed 5-fold tumor growth in 44 days, while similar tumor growth was seen in 124 days for 0.125 μCi activity treatment. In fact, the 0.5 μCi dose showed tumor growth inhibition as soon as day 5 of the injections, and all mice from this group showed complete response. However, due to the indolent growth of the tumors in the control group in this study, an overall survival benefit was not observed (Supplementary Fig. S11).

We next evaluated antitumor potential of the [²²⁵Ac]DOTA-YS5 in PDX models. CD46-overexpressing PDXs were used for studying the therapeutic efficiency of the [²²⁵Ac]DOTA-YS5. The PSMA-negative neuroendocrine prostate cancer PDX, LTL-545–bearing mice were injected with saline, 0.03, 0.06, 0.125, and 0.25 μCi dose of [²²⁵Ac]DOTA-YS5 (Fig. 6B). As shown in tumor volume plot (Fig. 6B), all the mice from 0.125 and 0.25 μCi groups demonstrated complete resolution of the tumor, whereas 0.06 μCi delayed the tumor growth as compared with saline control. Median survival of the animals treated with saline, 0.03, 0.06, 0.125, and 0.25 μCi were found to be 29, 44 (not statistically significant), 84 (P = 0.0025), 198 (P = 0.0025), and 140 days (P = 0.0025), respectively (Fig. 6B). During the first 15 days of treatment (Fig. 6B), the body weights of LTL-545–bearing mice treated with [²²⁵Ac]DOTA-YS5 decreased, but these mice gradually regained their body weights.

NSG mice bearing LTL-484 PDX (CD46 and PSMA positive) were randomized into four arms for saline and [²²⁵Ac]DOTA-YS5 injections (0.06, 0.125, and 0.25 μCi; Fig. 6C). The 0.25 μCi dose injected mice demonstrated acute toxicity manifested by hunched posture and reduced locomotion on 11th day p.i. On the other hand, we found that that 0.06 (median survival = 71, P = 0.335) and 0.125 μCi (median survival = 83, P = 0.169) dose delayed the tumor growth with a nonstatistically significant trend toward increase in the median survival as compared with saline control (median survival = 30). The NSG mice used in these experiments were more sensitive to radiation dose compared with nude mice used for the cell line–derived xenograft studies, with a constant decrease in body weight seen over the period of treatment (Fig. 6C; ref. 29).

Alpha-particle therapies hold great potential for treating metastatic prostate cancer. The PSMA-targeted alpha-particle therapy (²²⁵Ac-PSMA-617) showed potential antitumor response but with considerable toxicities (19, 30). Recently, several studies have acknowledged the issue of less or no expression of PSMA in advanced prostate tumors which could result in resistance to PSMA-directed therapies (31). Our prior report taken together with new data (Fig. 1) showed that anti-CD46 antibody (YS5)-based radiotracer probe ([⁸⁹Zr]DFO-YS5) successfully targeted the overexpression of CD46 on prostate cancer cell surface for PET imaging, including in PSMA-negative models (17). Inspired by these results as well as prior studies demonstrating efficacy of the YS5 antibody as an antibody–drug conjugate, we hypothesized that the YS5 antibody would serve as an ideal targeting vector for delivery of therapeutic alpha particles (14).

We report the CD46-targeted alpha-particle therapy with [²²⁵Ac]DOTA-YS5 for PSMA-positive and PSMA-negative tumors. We developed a reproducible radiosynthesis and found that [²²⁵Ac]DOTA-YS5 was able to target protein CD46 in vitro and in vivo. In a biodistribution study, high tumor uptake of [²²⁵Ac]DOTA-YS5 was observed over 17 days, along with clearance from healthy organs. We noted all the healthy organs except liver and bones demonstrated gradual clearance of [²²⁵Ac]DOTA-YS5. The steady uptake in liver corroborates with previous studies reported for ²²⁵Ac-labeled PSMA-targeting antibodies (32). Similar results for liver were also noted in earlier studies with [⁸⁹Zr]DFO-YS5 over the period of 168 hours (17). Favorable tumor to blood and tumor to muscle ratios were found, suggesting feasibility for therapeutic studies.

We conducted a detailed toxicologic analysis following treatment, with the key finding of dose-limiting nephrotoxicity. Our data suggest that this is due at least in part to non–target-mediated redistribution of the daughter isotope, ²¹³Bi. Though an antibody-chelate may retain the ²²⁵Ac, as the time passes ²²⁵Ac decays into daughter isotopes ²¹³Bi and ²²¹Fr which are ejected from the chelator due to the recoil effect. These daughter isotopes are reported to accumulate in kidneys, resulting in off-target toxicity (28). Our observations from the gamma-ray spectra of ²¹³Bi and ²²¹Fr confirm that there is a nonhomogenous distribution of daughter nuclei in mouse organs. Specifically, comparison of gamma spectra between organs in Supplementary Fig. S4 shows increased ²¹³Bi peak intensity and therefore ²¹³Bi accumulation in kidneys. In addition, histopathology analysis reveals that ²¹³Bi accumulation results in renal parenchymal damage, fibrin thrombi formation, and glomerulosclerosis. We conducted a dosimetry estimation, neglecting the daughter ²¹³Bi, which did not demonstrate disproportionate renal radiation dose, further supporting our hypothesis that redistribution plays a key role (Supplementary Fig. S5). This substantiates the previous finding of ²¹³Bi accumulation in kidneys resulting in kidney toxicity. Unexpectedly, we have also observed uptake of [²²⁵Ac]DOTA-YS5 in bone, but no damage to bone marrow or hematologic toxicity was seen, possibly due to the short range and low penetrability of alpha particles. Given that the YS5 antibody does not bind to mouse CD46 protein, the toxicity study results should be interpreted with caution, as we may underestimate the toxicity profile in humans. In addition, the toxicity of [²²⁵Ac]DOTA-YS5 may not be comparable with the ²²⁵Ac-PSMA-617, owing to its high molecular weight, different clearance route, and increased blood half-life.

Evaluation of therapeutic efficiency revealed that [²²⁵Ac]DOTA-YS5 shows a significant antitumor response across a panel of prostate cancer models, including in 22Rv1 (weakly PSMA-positive) and DU145 (PSMA-negative), as well as the LTL-545 (PSMA-negative neuroendocrine) and LTL-484 (PSMA-positive adenocarcinoma) PDX models. Marked increases in overall survival were observed in the 22Rv1 and LTL-545 models, with the latter demonstrating complete eradication of the tumor at the higher dose levels with no evidence of recovery even at delayed timepoints. Considering the long period of therapeutic efficiency studies, 0.25 μCi was the MTD in the nu/nu model, with a prominent antitumor response. In the highly radiosensitive NSG model (required to grow the PDX models), the MTD is at a lower dose (0.125 μCi activity) with significant antitumor effects and an increase in survival.

Apart from PSMA, potential targets for prostate cancer for radiopharmaceutical therapy include prostate stem cell antigen (PSCA), delta-like ligand 3 (DLL3), kallikrein 2 (hK2), and CUB domain-containing protein 1 (CDCP1; refs. 20, 33–35). When evaluated in preclinical imaging and therapy studies, these agents show promising results. Our results demonstrate that CD46 targeting with [²²⁵Ac]DOTA-YS5 could deliver the radiopharmaceutical to a broader range of prostate cancers compared with PSMA targeting. This could be further evaluated in a detailed study of CD46 expression across a panel of PDXs and potentially patient biopsy samples.

Feasibility of translation of [²²⁵Ac]DOTA-YS5 to the clinic is high. We discovered that CD46 is an excellent target for mCRPC across differentiation patterns, developed a human antibody YS5 that binds to a tumor-selective conformational epitope, and constructed a YS5-based ADC (14), which is in multiple clinical trials (NCT03575819 and NCT05011188). In addition, we recently developed a YS5-based PET agent (17), which is now in a first-in-human study in patients with mCRPC (NCT05245006). Thus, there is an established path for YS5-based agents to move from the bench to the clinic.

Safety is a critical issue that needs to be carefully investigated. One potential limitation of our study is selection of only two timepoints for toxicity analysis. Future studies will investigate multiple timepoints for acute and chronic toxicity. While ²²⁵Ac-antibodies have demonstrated great efficacy, prior studies have also revealed dose-dependent toxicity (21, 36). Potential mechanisms for toxicity include redistribution of the daughter isotopes away from the original site, due to the “recoil” effect (37, 38). We observed this effect in higher uptake of ²¹³Bi in the kidney compared with the other organs. Our results are consistent with prior literature which demonstrates both remarkable efficacy but also toxicity for treating cancer with ²²⁵Ac radiopharmaceutical therapy. Moreover, dosimetric analysis of the biodistribution data revealed higher apparent radiation doses to some tissues including bone, kidneys, stomach, and heart. Despite this finding, the only detectable toxicity in our study was to the kidneys. There are several potential routes to mitigate this toxicity, including dose reduction or fractionation, chelation therapy, linker optimization, pretargeting, or the use of alternative isotopes such as ¹⁷⁷Lu or ²¹²Pb (39–42).

In conclusion, we have prepared and developed [²²⁵Ac]DOTA-YS5 as a RIT agent for the treatment of prostate cancer, which shows efficacy for both PMSA-deficient and PSMA-positive tumors. Because of the CD46-targeting nature of [²²⁵Ac]DOTA-YS5, we were able to deliver a high amount of radiation dose to tumor tissue, resulting in a potent antitumor effect. The data presented here provide strong evidence of therapeutic efficacy of [²²⁵Ac]DOTA-YS5 in preclinical models, and support future clinical translation of CD46-targeted radioligand therapy.

D. Beckford-Vera reports other support from Actinium Pharmaceuticals outside the submitted work. Y. Su reports a patent for US20170362330A1 issued, licensed, and with royalties paid from Fortis Therapeutics. J. He reports grants from NCI during the conduct of the study; in addition, J. He holds equity in Molecular Imaging and Therapeutics, Inc., which was converted to equity of Fortis Therapeutics that licensed intellectual properties from the University of California. R. Aggarwal reports grants from Fortis Therapeutics during the conduct of the study. R. Aggarwal also reports grants from Janssen; grants and personal fees from Merck, AstraZeneca, and Novartis; and personal fees from Bioexcel Therapeutics, EcoR1, Dendreon, Tersara, Targeted Oncology, OncLive, Lumanity, Exelixis, PCCTC, Jubliant Therapeutics, Modra Therapeutics, SNMMI, and Bayer outside the submitted work. H.F. VanBrocklin reports grants from CDMRP Department of Defense and UCSF PICT program funded by NIH P30 during the conduct of the study. Y. Seo is equity holder of Molecular Imaging and Therapeutics, Inc., which was converted to equity of Fortis Therapeutics that licensed intellectual properties from the University of California. J. Chou reports grants from Department of Defense Prostate Cancer Research Program, NIH/NCI K08CA273514, and Prostate Cancer Foundation, as well as personal fees from Exai Bio outside the submitted work. B. Liu reports grants from NIH during the conduct of the study, and is a founder/board member of Fortis Therapeutics. In addition, B. Liu has a patent for US20140271685A1 issued, licensed, and with royalties paid from Fortis Therapeutics; a patent for US20170362330A1 issued, licensed, and with royalties paid from Fortis Therapeutics; and a patent for US8865873B2 issued, licensed, and with royalties paid from Fortis Therapeutics. R.R. Flavell reports grants from U.S. Department of Defense and NIH during the conduct of the study; in addition, R.R. Flavell has a patent on the development of new chelators for alpha-radiotherapy, not related directly to this work. No disclosures were reported by the other authors.

A.P. Bidkar: Data curation, software, formal analysis, investigation, visualization, methodology, writing–original draft, writing–review and editing. S. Wang: Data curation, formal analysis, investigation, methodology. K.N. Bobba: Data curation, investigation, methodology. E. Chan: Resources, data curation, methodology. S. Bidlingmaier: Resources, methodology. E.A. Egusa: Data curation, visualization. R. Peter: Software, formal analysis. U. Ali: Resources, data curation. N. Meher: Data curation, software, formal analysis, methodology. A. Wadhwa: Investigation, visualization. S. Dhrona: Validation. C. Dasari: Visualization. D. Beckford-Vera: Formal analysis, methodology. Y. Su: Formal analysis, methodology. R. Tang: Formal analysis, investigation. L. Zhang: Data curation, formal analysis. J. He: Formal analysis, supervision, validation, methodology. D.M. Wilson: Conceptualization, resources, data curation, supervision. R. Aggarwal: Conceptualization, resources, formal analysis, supervision, methodology. H.F. VanBrocklin: Conceptualization, resources, data curation, supervision, methodology. Y. Seo: Conceptualization, resources, supervision, investigation, methodology. J. Chou: Conceptualization, resources, data curation, formal analysis, methodology. B. Liu: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, methodology, writing–original draft, project administration, writing–review and editing. R.R. Flavell: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, methodology, writing–original draft, project administration, writing–review and editing.

R.R. Flavell acknowledges funding from Translational Science awards from the Prostate Cancer Research Program of the Congressionally Directed Medical Research Program of the U.S. Department of Defense (W81XWH-21-1-0792, W81XWH-20-1-0292), pilot funding through the UCSF Precision Imaging of Cancer and Therapy Program, and pilot funding through the UCSF Cancer Imaging Resource and Preclinical Therapeutics Core through P30CA082103. ²²⁵Ac was supplied by the U.S. Department of Energy Isotope Program, managed by the Office of Isotope R&D and Production. MALDI-MS data were provided by the Mass Spectrometry Facility, Department of Chemistry, University of Alberta (Edmonton, Alberta, Canada). Blood cell counts and organ function tests were performed by Comparative Pathology Laboratory, UC Davis School of Veterinary Medicine. Histology staining procedures were carried out by Histology and Biomarker core, University of California San Francisco.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.