An Autophagy-Targeting Chimera Induces Degradation of Androgen Receptor Mutants and AR-v7 in Castration-Resistant Prostate Cancer

The characterization of an AUTOTAC-based degrader capable of inducing autophagic degradation of wild-type and mutated androgen receptors demonstrates the potential of this approach for targeting castration-resistant prostate cancer and overcoming drug resistance.

Abstract

Genetic alterations play a pivotal role in various human diseases, particularly cancer. The androgen receptor (AR) is a crucial transcription factor driving prostate cancer progression across all stages. Current AR-targeting therapies utilize competitive AR antagonists or pathway suppressors. However, therapy resistance often emerges due to AR mutations and AR splice variants, such as AR-v7. To overcome this, we developed ATC-324, an AR degrader using the innovative protein degradation technology platform AUTOphagy-TArgeting Chimera (AUTOTAC). ATC-324 was designed to comprise enzalutamide, an AR inhibitor, as a target-binding ligand and YT 6-2, a ligand of the autophagy receptor p62/SQSTM1, as an autophagy-targeting ligand. ATC-324 induces the formation of the AR/p62 complex, leading to autophagy–lysosomal degradation of AR. Importantly, ATC-324 effectively degrades AR mutants frequently detected in prostate cancer and codegrades AR-v7 as a heterodimer with full-length AR. ATC-324 reduces nuclear AR levels and downregulates the target gene expression of AR and AR-v7, leading to cytotoxicity in AR-positive prostate cancer cells. We also provide evidence of the therapeutic potential of ATC-324 in vivo as well as ex vivo bone organ culture. Moreover, ATC-324 remains potent in enzalutamide-resistant prostate cancer cells. These results demonstrate the potential of the AUTOTAC platform to target previously considered undruggable proteins and overcome certain drug resistance mechanisms.

Significance:

The characterization of an AUTOTAC-based degrader capable of inducing autophagic degradation of wild-type and mutated androgen receptors demonstrates the potential of this approach for targeting castration-resistant prostate cancer and overcoming drug resistance.

Graphical Abstract

Introduction

Traditional drug development often focuses on inhibiting the activity of disease-causing proteins using small molecules. However, the effectiveness of functional inhibition is often suboptimal, and their effects are temporary. As a result, protein degradation technology (PDT) has emerged as an innovative approach in drug development. Recently, we developed a groundbreaking PDT called the AUTOphagy-TArgeting Chimera (AUTOTAC) platform for targeted protein degradation. The bivalent AUTOTAC molecule comprises a target-binding ligand (TBL) linked to an autophagy-targeting ligand (ATL) of p62/SQSTM1 (1). AUTOTAC brings the target to the autophagic receptor p62 via its TBL and ATL, forming a p62–cargo complex that leads to autophagy–lysosomal degradation (2). Notably, AUTOTAC-mediated target degradation does not require interactions between the target protein and p62. This is in contrast to the most well-known PDT platform, PROteolysis TArgeting Chimera (PROTAC), which depends on the structure-function relationship between a specific E3 ligase and its target protein to facilitate target protein ubiquitination necessary for degradation (3). Based on this unique feature of the AUTOTAC platform, we hypothesize that AUTOTAC effectively degrades even mutated target proteins and/or their binding partners critical in their pathogenic activities.

The aim of the present study is to develop an androgen receptor (AR)–targeting AUTOTAC that effectively degrades various forms of AR proteins. A substantial body of evidence supports the critical role of AR in driving the progression of prostate cancer across all stages (4, 5). Accordingly, androgen deprivation therapy through surgical or chemical castration has been the standard of care for men with prostate cancer. Although approximately 90% of patients respond to initial androgen deprivation therapy, 10% to 20% of patients develop castration-resistant prostate cancer (CRPC) within a span of 5 years (6, 7). To target AR biology in CRPC, current treatment options include competitive AR antagonists, known as second-generation antiandrogens (SGAA), such as enzalutamide, apalutamide, and darolutamide as well as biochemical pathway suppressors like abiraterone that block testosterone production. Unfortunately, patients with advanced CRPC eventually develop therapy resistance to these treatments. The mechanisms underlying prostate cancer cell resistance to antiandrogens include gene amplification, increased expression, mutations, and the generation of splice variants of AR (8). Clinical studies showed that CRPC is often enriched with AR mutants that harbor mutations in the ligand-binding domain (LBD) and AR splice variants lacking the LBD, thereby being constitutively active in an androgen-independent manner (9, 10). These genetic alterations serve as significant drivers of resistance against AR antagonists (11–13). Furthermore, recent studies have shown that antiandrogen treatment can result in alterations of AR activity, thereby leading to the induction of noncanonical AR signaling with a gain-of-function effect (14). Experimentally, shRNA-mediated knockdown of AR results in regression of well-established CRPC in mice (15), indicating that depleting AR is more effective than antagonizing it through competitive inhibition.

In the present study, we have successfully developed ATC-324, an AR-targeting AUTOTAC, utilizing enzalutamide as a TBL and a small molecule ligand of p62 as an ATL. ATC-324 degrades not only full-length AR proteins but also the AR splice variant AR-v7. Our study indicates that AR-v7, lacking the LBD, can be cotargeted by ATC-324 when it forms a heterodimer with wild-type (WT) AR. RNA sequencing (RNA-seq) analysis further confirms that ATC-324 reverses the gene expression signatures associated with both WT AR and AR-v7. Furthermore, we demonstrate that ATC-324 exhibits promising potential in degrading AR mutants frequently detected in prostate carcinoma. Importantly, ATC-324-mediated AR degradation leads to cytotoxicity in castrate-resistant and enzalutamide-resistant prostate cancer cells. We also provide evidence of the therapeutic potential of ATC-324 in vivo as well as ex vivo bone organ culture.

Taken together, the present study successfully demonstrates the potential of the AUTOTAC platform to target previously “undruggable” proteins, such as AR mutants and AR-v7, and to overcome certain drug resistance mechanisms.

Materials and Methods

Chemical synthesis of AR-targeting AUTOTACs

LCMS was taken on a quadrupole mass spectrometer on Agilent 1260 HPLC and 6120MSD [Column: C18 (50 × 4.6 mm, 5 μm)] operating in ES (+) or (−) ionization mode; T = 30°C; flow rate = 1.5 mL/minutes; detected wavelength: 254 nm.

Scheme 1. (R)-N-(2-(2-(2-((3-(3,4-bis((4-fluorobenzyl)oxy)phenoxy)-2-hydroxypropyl)amino)ethoxy)ethoxy)ethyl)-4-(3-(4-cyano-3-(trifluoromethyl)phenyl)-5,5-dimethyl-4-oxo-2-thioxoimidazolidin-1-yl)-2-fluorobenzamide (ATC-323; Supplementary Fig. S1A).

• 1.1 Synthesis of 4-(3-(4-cyano-3-(trifluoromethyl)phenyl)-5,5-dimethyl-4-oxo-2-thioxoimidazolidin-1-yl)-2-fluorobenzoic acid (1-1).

To a solution of 1 (4-(3-(4-cyano-3-(trifluoromethyl)phenyl)-5,5-dimethyl-4-oxo-2-thioxoimidazolidin-1-yl)-2-fluoro-N-methylbenzamide, 15 g, 32.3 mmol) in dioxane (150 mL) was added concentrated HCl (150 mL), then the mixture was heated to reflux and stirred for 3 days. Water was added and the mixture was extracted with EtOAc. The organic phase was dried over Na₂SO₄ and concentrated. The residue was purified by column (PE:EA = 5:1–2:1) to afford 1-1 (4-(3-(4-cyano-3-(trifluoromethyl)phenyl)-5,5-dimethyl-4-oxo-2-thioxoimidazolidin-1-yl)-2-fluorobenzoic acid, 9.0 g, 20.0 mmol, 61.9% yield) as a pale yellow solid. (TLC: DCM/MeOH = 10/1, Rf = 0.2).

¹H-NMR (CDCl₃, 400 MHz) δ (ppm) 8.23 (t, 1H, J = 8 Hz), 8.01-7.95 (m, 2H), 7.83 (dd, 1H, 8.4 and 2 Hz), 7.27-7.20 (m, 2H), 1.64 (s, 6H).

• 1.2 Synthesis of 3,4-bis((4-fluorobenzyl)oxy)benzaldehyde (2-1)

To a solution of 2 (3,4-dihydroxybenzaldehyde, 100 g, 724.6 mmol) in acetonitrile (ACN; 1,000 mL) were added 1-(bromomethyl)-4-fluorobenzene (301.3 g, 1.59 mol) and K₂CO₃ (300 g, 2.17 mol). The mixture was stirred at 80°C for 16 hours. Then, the reaction was concentrated, the residue was purified by silica gel, eluted with EA/PE (1:15–1:8) to afford 2-1 (3,4-bis((4-fluorobenzyl)oxy)benzaldehyde, 187 g, 72.8% yield) as a white solid.

¹H-NMR (DMSO_d₆, 400 MHz) δ (ppm) 9.83 (s, 1H), 7.55-7.48 (m, 6H), 7.30-7.20 (m, 5H), 5.25 (s, 2H), 5.19 (s, 2H).

• 1.3 Synthesis of 3,4-bis((4-fluorobenzyl)oxy)phenol (2-2)

To a solution of 2-1 (3,4-bis((4-fluorobenzyl)oxy)benzaldehyde, 187 g, 526.7 mmol) in DCM (2,000 mL) was added m-CPBA (126 g, 730.4 mmol). The mixture was stirred at room temperature for 16 hours. Then, the reaction was washed with saturated sodium bicarbonate solution, concentrated under vacuum. Then, the crude product is added to methanol (1,500 mL) and water (200 mL) was added KOH (58.9 g, 1.05 mol). The mixture was stirred at room temperature for 3 hours. Then the reaction was filtered and the solid to dryness under vacuum. The crude compound was purified by silica gel, eluted with EA/PE (1:15–1:5) to afford 2-2 (3,4-bis((4-fluorobenzyl)oxy)phenol, 151 g, 83.7% yield) as an off-white solid.

¹H-NMR (DMSO_d₆, 400 MHz) δ (ppm) 7.50-7.41 (m, 4H), 7.24-7.15 (m, 4H), 6.82 (d, 1H, J = 8.4 Hz), 6.49 (d, 1H, J = 2.4 Hz), 6.25 (dd, 1H, J = 8.8, 2.8 Hz), 5.04 (s, 2H), 4.95 (s, 2H).

• 1.4 Synthesis of (R)-2-((3,4-bis((4-fluorobenzyl)oxy)phenoxy)methyl)oxirane (2-3)

To a solution of 2-2 (3,4-bis((4-fluorobenzyl)oxy)phenol, 45.8 g, 134 mmol) in EtOH (500 mL) were added water (25 mL) and KOH (17.2 g, 307 mmol). Then, (R)-2-(chloromethyl)oxirane (37 g, 400 mmol) was added to the reaction. The resulting mixture was stirred at room temperature for 16 hours. Then the reaction was quenched by addition water, extracted with EA. The organic layer was washed with brine, dried over Na₂SO₄, filtered and concentrated. The residue was purified by silica gel, eluted with EA/PE (1:15–1:10) to afford 2-3 ((R)-2-((3,4-bis((4-fluorobenzyl)oxy)phenoxy)methyl)oxirane, 26 g, 48.7% yield) as a white solid.

¹H-NMR (DMSO_d₆, 400 MHz) δ (ppm) 7.51-7.43 (m, 4H), 7.25-7.16 (m, 4H), 6.94 (d, 1H, J = 9.2 Hz), 6.73 (d, 1H, J = 2.8 Hz), 6.45 (dd, 1H, J = 8.8, 2.8 Hz), 5.10 (s, 2H), 5.00 (s, 2H), 4.24 (dd, 1H, J = 11.2, 2.8 Hz), 3.75 (dd, 1H, J = 11.2, 6.4 Hz), 3.30-3.28 (m, 1H), 2.83 (t, J = 5.2 Hz, 1H), 2.68 (dd, 1H, J = 5.2, 2.8 Hz).

• 1.5 Synthesis of (R)-1-((2-(2-(2-aminoethoxy)ethoxy)ethyl)amino)-3-(3,4-bis((4-fluorobenzyl)oxy)phenoxy)propan-2-ol (2-4).

To a solution of 2-3 ((R)-2-((3,4-bis((4-fluorobenzyl)oxy)phenoxy)methyl)oxirane 200 mg, 0.502 mmol) in ACN (10 mL) was added 2,2′-(ethane-1,2-diylbis(oxy))diethanamine (111 mg, 0.753 mmol). The solution was stirred at 70°C for 16 hours. Then the solution was concentrated. The residue was purified by pre-HPLC to afford 2-4 ((R)-1-((2-(2-(2-aminoethoxy)ethoxy)ethyl)amino)-3-(3,4-bis((4-fluorobenzyl)oxy)phenoxy)propan-2-ol, 60 mg, 21.8% yield) as a yellow solid.

¹H-NMR (DMSO_d₆, 400 MHz) δ (ppm) 7.48-7.42 (m, 4H), 7.24-7.18 (m, 4H), 6.92 (d, 1H, J = 8.8 Hz), 6.67 (d, 1H, J = 2.4 Hz), 6.45-6.43 (m, 1H), 5.09 (s, 2H), 4.99 (s, 2H), 3.85-8.83 (m, 3H), 3.50 (s, 4H), 3.37-3.35 (m, 3H), 2.67-2.62 (m, 6H).

• 1.6 Synthesis of (R)-N-(2-(2-(2-((3-(3,4-bis((4-fluorobenzyl)oxy)phenoxy) hydroxypropyl)amino)ethoxy)ethoxy)ethyl)-4-(3-(4-cyano-3-(trifluoromethyl)phenyl)-5,5-dimethyl-4-oxo-2-thioxoimidazolidin-1-yl)-2-fluorobenzamide (ATC-323).

To a solution of 2-4 ((R)-1-((2-(2-(2-aminoethoxy)ethoxy)ethyl)amino)-3-(3,4-bis((4-fluorobenzyl)oxy)phenoxy)propan-2-ol, 60 mg, 0.1 mmol) in DCM (10 mL) was added 1-hydroxybenzotriazole (16.3 mg, 0.12 mmol), 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (23 mg, 0.12 mmol), and Et₃N (50.5 mg, 0.5 mmol). The solution of 4-phenylbutanoic acid (45.2 mg, 0.1 mmol) in DCM (5 mL) was added at 0°C. The solution was concentrated and the residue was purified by pre-HPLC to give ATC-323 ((R)-N-(2-(2-(2-((3-(3,4-bis((4-fluorobenzyl)oxy)phenoxy)-2-hydroxypropyl)amino)ethoxy)ethoxy)ethyl)-4-(3-(4-cyano-3-(trifluoromethyl)phenyl)-5,5-dimethyl-4-oxo-2-thioxoimidazolidin-1-yl)-2-fluorobenzamide, 17 mg, 17.4% yield) as white solid.

¹H-NMR (DMSO-d₆, 400 MHz) δ (ppm) 8.38 (d, J = 8.4 Hz, 1H), 8.28 (d, J = 6 Hz, 2H), 8.07 (d, J = 8 Hz, 1H), 7.77 (t, J = 8 Hz, 1H), 7.49-7.42 (m, 5H), 7.33 (dd, J = 8.4 Hz, 1H), 7.23-7.15 (m, 4H), 6.93 (d, J = 8.8 Hz, 1H), 6.66 (d, J = 2.8 Hz, 1H), 6.43 (dd, J = 8.8 and 2.8 Hz, 1H), 5.08 (s, 2H), 4.99 (s, 2H), 3.85-3.82 (m, 3H), 3.45-3.42 (m, 10H), 2.83-2.79 (m, 3H), 2.72-2.69 (m, 1H), 1.53 (s, 6H).

ESI-MS Calcd m/z for C₄₉H₄₇F₆N₅O₈S: 979.99, found 980.8.

Scheme 2. (R)-N-(18-(3,4-bis((4-fluorobenzyl)oxy)phenoxy)-17-hydroxy-3,6,9,12-tetraoxa-15-azaoctadecyl)-4-(3-(4-cyano-3-(trifluoromethyl)phenyl)-5,5-dimethyl-4-oxo-2-thioxoimidazolidin-1-yl)-2-fluorobenzamide (ATC-324; Supplementary Fig. S1B).

• 2.1 Synthesis of N-(14-amino-3,6,9,12-tetraoxatetradecyl)-4-(3-(4-cyano-3-(trifluoromethyl)phenyl)-5,5-dimethyl-4-oxo-2-thioxoimidazolidin-1-yl)-2-fluorobenzamide (3).

To a solution of tert-butyl (14-amino-3,6,9,12-tetraoxatetradecyl)carbamate (4.0 g, 11.9 mmol) in DCM (100 mL) were added 1-1 (4-(3-(4-cyano-3-(trifluoromethyl)phenyl)-5,5-dimethyl-4-oxo-2-thioxoimidazolidin-1-yl)-2-fluorobenzoic acid, 5.37 g, 11.9 mmol), 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (3.41 g, 17.8 mmol), 1-hydroxybenzotriazole (2.40 g, 17.8 mmol), and trimethylamine (2.40 g, 23.8 mmol), then the mixture was stirred at room temperature overnight. The mixture was washed with water and brine. The organic phase was dried over Na₂SO₄ and concentrated. The residue was purified by column (DCM:MeOH = 50:1–20:1) to afford tert-butyl (1-(4-(3-(4-cyano-3-(trifluoromethyl)phenyl)-5,5-dimethyl-4-oxo-2-thioxoimidazolidin-1-yl)-2-fluorophenyl)-1-oxo-5,8,11,14-tetraoxa-2-azahexadecan-16-yl)carbamate (7.5 g, 9.75 mmol, 82.0% yield ) as colorless oil.

To a solution of tert-butyl (1-(4-(3-(4-cyano-3-(trifluoromethyl)phenyl)-5,5-dimethyl-4-oxo-2-thioxoimidazolidin-1-yl)-2-fluorophenyl)-1-oxo-5,8,11,14-tetraoxa-2-azahexadecan-16-yl)carbamate (7.5 g, 9.75 mmol) in DCM (75 mL) was added HCl/MTBE (75 mL, 8 mol/L), then the mixture was stirred at room temperature for 4 hours. The mixture was concentrated and dissolved in DCM, then washed with saturated aqueous Na₂CO₃ and brine. The organic phase was dried over Na₂SO₄ and concentrated to afford 3 (N-(14-amino-3,6,9,12-tetraoxatetradecyl)-4-(3-(4-cyano-3-(trifluoromethyl)phenyl)-5,5-dimethyl-4-oxo-2-thioxoimidazolidin-1-yl)-2-fluorobenzamide, 6.0 g, 8.97 mmol, 92.0% yield) as yellow oil.

¹H-NMR (CDCl₃, 400 MHz) δ (ppm) 8.23-8.20 (m, 1H), 8.02-7.97 (m, 2H), 7.86-7.84 (m, 1H), 7.48 (brs, 1H), 7.28-7.15 (m, 2H), 3.79-3.62 (m, 16H), 3.52-3.50 (m, 2H), 2.83 (t, 2H), 1.63 (s, 6H).

• 2.2 Synthesis of ((R)-N-(18-(3,4-bis((4-fluorobenzyl)oxy)phenoxy)-17-hydroxy-3,6,9,12-tetraoxa-15-azaoctadecyl)-4-(3-(4-cyano-3-(trifluoromethyl)phenyl)-5,5-dimethyl-4-oxo-2-thioxoimidazolidin-1-yl)-2-fluorobenzamide (ATC-324)

To a solution of 3 (N-(14-amino-3,6,9,12-tetraoxatetradecyl)-4-(3-(4-cyano-3-(trifluoromethyl)phenyl)-5,5-dimethyl-4-oxo-2-thioxoimidazolidin-1-yl)-2-fluorobenzamide, 3.0 g, 4.48 mmol) in MeOH (6 mL) and ACN (6 mL) was added 2-3 ((R)-2-((3,4-bis(4-fluorobenzyloxy)phenoxy)methyl)oxirane, 1.78 g, 4.48 mmol), then the mixture was stirred at 65°C overnight. The mixture was concentrated and purified by column (DCM:MeOH = 50:1–20:1) to afford ATC-324 ((R)-N-(18-(3,4-bis((4-fluorobenzyl)oxy)phenoxy)-17-hydroxy-3,6,9,12-tetraoxa-15-azaoctadecyl)-4-(3-(4-cyano-3-(trifluoromethyl)phenyl)-5,5-dimethyl-4-oxo-2-thioxoimidazolidin-1-yl)-2-fluorobenzamide, 1.1 g, 1.03 mmol, 23.0% yield, HPLC purity at 254 nm : 98.2%) as a pale yellow solid.

¹H-NMR (CD₃OD, 400 MHz) δ (ppm) 8.22 (t, J = 8.4 Hz, 1H), 7.97-8.03 (m, 2H), 7.84 (dd, J1 = 8.0 Hz, J2 = 2.0 Hz, 1H), 7.30-7.42 (m, 5H), 7.14-7.25 (m, 2H), 7.01-7.08 (m, 4H), 6.85 (d, J = 8.8 Hz, 1H), 6.60 (d, J = 2.8 Hz, 1H), 6.41 (dd, J1 = 8.8 Hz, J2 = 2.8 Hz, 1H), 5.07 (s, 2H), 5.02 (s, 2H), 4.05-4.08 (m, 1H), 3.90 (d, J = 9.2 Hz, 2H), 3.62-3.71 (m, 18 H), 2.76-2.93 (m, 5H), 1.62 (s, 6H).

ESI-MS Calcd m/z for C₅₃H₅₅F₆N₅O₁₀S: 1,068.10, found 1,069.9.

Scheme 3. (R)-N-(24-(3,4-bis((4-fluorobenzyl)oxy)phenoxy)-23-hydroxy-3,6,9,12,15,18-hexaoxa-21-azatetracosyl)-4-(3-(4-cyano-3-(trifluoromethyl)phenyl)-5,5-dimethyl-4-oxo-2-thioxoimidazolidin-1-yl)-2-fluorobenzamide (ATC-325; Supplementary Fig. S1C).

ATC-325 (R)-N-(24-(3,4-bis((4-fluorobenzyl)oxy)phenoxy)-23-hydroxy-3,6,9,12,15,18-hexaoxa-21-azatetracosyl)-4-(3-(4-cyano-3-(trifluoromethyl)phenyl)-5,5-dimethyl-4-oxo-2-thioxoimidazolidin-1-yl)-2-fluorobenzamide was synthesized as white solid (24.6% yield, HPLC purity at 254 nm: 98.2%) in the same manner as scheme 2 by using N-(20-amino-3,6,9,12,15,18-hexaoxaicosyl)-4-(3-(4-cyano-3-(trifluoromethyl)phenyl)-5,5-dimethyl-4-oxo-2-thioxoimidazolidin-1-yl)-2-fluorobenzamide (4) instead of 3.

¹H-NMR (CD₃OD, 400 MHz) δ (ppm) 8.24 (t, J = 8.4 Hz, 1H), 7.98-8.02 (m, 2H), 7.85 (dd, J = 8.4, 2.0 Hz, 1H), 7.36-7.43 (m, 4H), 7.16-7.26 (m, 3H), 7.01-7.09 (m, 4H), 6.85 (d, J = 8.8 Hz, 1H), 6.60 (d, J = 2.8 Hz, 1H), 6.41 (dd, J = 8.8, 2.8 Hz, 1H), 5.07 (s, 2H), 5.02 (s, 2H), 4.09-4.13 (m, 1H), 3.91-3.94 (m, 2H), 3.65-3.73 (m, 25 H), 2.81-2.99 (m, 5H), 1.63 (s, 6H).

ESI-MS Calcd m/z for C₅₇H₆₃F₆N₅O₁₂S: 1,156.20, found 1,157.9.

Establishment of xenograft mouse models and Bone-In-Culture Array

NOD.Cg-Prkdc^scidIl2rg^tm1Wjl/Szj mice (RRID: IMSR_JAX:005557) ages 6 weeks were purchased from the The Jackson Laboratory. All maintenance and mice experiments were approved and conducted in accordance with the guidelines of Institutional Animal Care and Use Committee, Seoul National University (IACUC number SNU-190618-3-3) and the Korean Food and Drug Administration. Animal procedures were also approved by and conducted within the guidelines of the Wayne State University Institutional Animal Care and Use Committee (Protocol number IACUC-21-05-3542). To establish a xenograft model, 8-week-old male NOD.Cg-Prkdc^scidIl2rg^tm1Wjl/Szj mice were injected with 2 × 10⁶ LNCaP cells or 4 × 10⁶ 22Rv1 cells mixed with Matrigel (Corning, 254262) in the lower dorsal region. After the tumor volume reached approximately 100 mm³, mice were randomly assigned to treatment groups and intraperitoneally injected with 20 mg/kg of ATC-324 (5% DMSO + 10% Solutol + 85% PBS) five times a week. After euthanasia with CO₂ asphyxiation, tumors were harvested and stored for further analysis. The graphical representation of the xenograft experiment was created using BioRender.com (RRID: SCR_018361).

The Bone-In-Culture Array (BICA) was adapted from Wang and colleagues (16, 17). Briefly, 2.5 × 10⁶ luciferase-expressing 22Rv1 cells were injected into the external iliac artery of C57BL/6 mice (RRID: IMSR_JAX:005557). After about 30 minutes, the femur and tibia of the injected leg were dissected, and the distal epiphysis and metaphysis of the femur and the proximal epiphysis and metaphysis of the tibia were cut and fragmented. Bone fragments containing cancer cells were grown on 96-well ultralow attachment plates (Corning) in DMEM/F12 media supplemented with 0.5% FBS (Gibco, Thermo Fisher Scientific) and were randomly assigned to treatment groups. On day 9, treatment was started. Media change and bioluminescence imaging were performed twice weekly (Bruker In-Vivo Xtreme) at the Microscopy, Imaging and Cytometry Resources Core at Wayne State University, School of Medicine. Samples with a bioluminescence count of at least 4,000 on day 9 of growth (day 0 of treatment) were selected for pairwise analysis.

Cell culture

HeLa (RRID: CVCL_0030), HEK293T (RRID: CVCL_0063), C4-2B (RRID: CVCL_4784), 22Rv1 (RRID: CVCL_1045), PC3 (RRID: CVCL_0035), and DU145 (RRID: CVCL_0105) cells were obtained from ATCC. Luciferase-expressing 22Rv1 cells were generated through transfecting 22Rv1 cells with pGL4.5-Luc2 plasmid and selected with hygromycin B. p62^+/+ and p62^−/− mouse embryonic fibroblasts (MEF) were kind gifts from Keiji Tanaka (Tokyo Metropolitan Research Institute) with Tetsuro Ishii’s permission. Parental and enzalutamide-resistant LNCaP and LAPC4 cells were kind gifts from Dr. Steven Kregel (Loyola University Chicago, Chicago, IL). HeLa, HEK293T, and MEF cells were grown in DMEM medium with 10% FBS and antibiotics (100 U/mL penicillin and 100 mg/mL streptomycin). C4-2B, LNCaP, enzalutamide-resistant LNCaP, and 22Rv1 cells were maintained in RPMI-1640 medium with 10% FBS and antibiotics (100 U/mL penicillin and 100 mg/mL streptomycin). LAPC4 cells were cultured in Iscove's Modified Dulbecco's Medium supplemented with 10% FBS, glutamine, and antibiotics (100 U/mL penicillin and 100 mg/mL streptomycin). Enzalutamide-resistant LNCaP and LAPC4 cells were maintained in 20 µmol/L enzalutamide. All cells were tested for Mycoplasma contamination upon receipt and maintained up to passage 30. All cells were cultured in a 5% CO₂, 37°C incubators

Plasmids, antibodies, and reagents

pGL4.5-Luc2 plasmid was purchased form Promega (E131A). pEGFP-C1-AR (RRID: Addgene_28235) and pEGFP-C1-AR v7 (RRID: Addgene_86856) were gifts from Marco Marcelli and Michael Mancini (Baylor College of Medicine). Flag-tagged AR plasmids with LBD mutations were constructed utilizing Flag-M4-AR (RRID: Addgene plasmid_171240), a gift from Steven Balk (Harvard Medical School), via site-directed mutagenesis. The sequences of primers used in site-directed mutagenesis are listed in Supplementary Table S1.

The following primary antibodies were used in this study: Antibodies against cleaved caspase-3 (#9664, RRID: AB_2070042), Sp1 (#5931, RRID: AB_10621245), and AR (#5153, RRID: AB_10691711) were purchased from Cell Signaling Technology; histone H1 (#MABE71, RRID: AB_10845941), CK8 (#MABT329, RRID: AB_2891089), and TsfR (#MABS1981) from Millipore; LC3 (#L7543, RRID: AB_796155) and FLAG M2 (#F1804, RRID: AB_262044) from Sigma; AR (#sc-7305, RRID: AB_626671) and GFP (#sc-9996, RRID: AB_627695) from Santa Cruz; p62 (#ab56416, RRID: AB_945626) and AR-v7 (#ab198394, RRID: AB_2861275) from Abcam; β-actin (#AP0060, RRID: AB_2797445) from Bioworld; HSP90 (#H38220) from BD Transduction. The following secondary antibodies were used in this study: Alexa Fluor 488 goat anti-rabbit IgG (#A11034, RRID: AB_2576217), Alexa Fluor 488 goat anti-mouse IgG (#A11029, RRID: AB_2534088), Alexa Fluor 555 goat anti-rabbit IgG (#A32732, RRID: AB_2633281), and Alexa Fluor 555 goat anti-mouse IgG (#A32727, RRID: AB_2633276) were purchased from Invitrogen; anti-rabbit IgG-HRP (#7074, RRID: AB_2099233) and anti-mouse IgG-HRP (#7076, RRID: AB_330924) from Cell Signaling Technology. Enzalutamide (S1250) was purchased from Selleck Chemicals and bafilomycin A1 (B1793), E64D (E8640), and pepstatin A (P5318) were purchased from Sigma.

Transfection

Plasmids were transfected into PC3 cells using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions. Predesigned sip62 RNA (8878-1, Bioneer) was transfected into LNCaP and 22Rv1 cells using RNAimax (13778075, Invitrogen) according to the manufacturer’s instructions.

Infection and luciferase assay

22Rv1 cells were infected with prepackaged lentiviral particles carrying the AR response element luciferase reporter gene (#LTLR032, GnP Biosciences) following the manufacturer’s instructions. A stable 22Rv1-LTV-ARRE-Luc cell line was then established through puromycin selection. Luciferase assay was performed using the Luciferase Assay System (#E1501, Promega) following the manufacturer’s instructions. A matching WST1 assay was performed to normalize for cytotoxicity.

Immunoblotting

Cell pellets were lysed in a RIPA buffer supplemented with protease inhibitor cocktails (#P8340, Sigma). Protein concentrations were quantified using the BCA Protein Assay Kit (#23225, Thermo Fisher Scientific). Protein lysates were prepared using 4× Laemmli sample buffer (#1610747, Bio-Rad Laboratories) with 10% β-mercaptoethanol and boiled. An equal amount of protein lysate was separated using SDS-PAGE and detected by diluted primary and horseradish peroxidase–conjugated secondary antibodies, sequentially. Densitometry of Western blot bands was measured and analyzed with image J_Fiji (RRID: SCR_002285; ver.1.54f; NIH).

IHC

Paraffin-embedded tissue samples were sectioned and subsequently deparaffinized with Neo-clear (109843, Merck), followed by gradual hydration with decreasing concentrations of ethanol. Next, the slides were boiled in sodium citrate buffer at 98°C for antigen retrieval and incubated with 3% H₂O₂ diluted with PBS to inactivate endogenous peroxidase. Subsequently, the slides were blocked with the blocking solution (2% normal goat serum, 1% BSA, 0.1% Triton X-100, and 0.05% Tween-20 in PBS). After blocking, the slides were incubated with primary antibodies diluted in blocking solution at 4°C for overnight, followed by host-specific HRP-conjugated secondary antibodies. Signal detection and nucleus counter-staining were performed using the DAB substrate kit (#SK-4105, Vector Laboratory) and hematoxylin QS (#H-3404-100, Vector Laboratory), respectively. The stained slides were then covered with cover glass using mounting solution (#H-5000-60, Vector Laboratory). Tissue images were analyzed using ZEISS Axio Scan Z1.

In vitro p62 oligomerization assay

HEK293T and PC3 cells were treated with AR-targeting AUTOTACs for 24 hours and subsequently harvested. The cells were lysed through repeated cycles of freezing/thawing. After centrifugation, the supernatants were collected, and protein concentration was quantified using a BCA protein assay kit. Equal amounts of protein lysates were mixed with nonreducing 4× LDS sample buffer (#NP0007, Thermo Fisher Scientific) and subsequently boiled. Using SDS-PAGE, the lysates were separated, and the oligomerized p62 was detected through immunoblotting.

Immunocytochemistry

Cells were seeded on coverslips coated with poly L-lysine (P8920, Sigma). After the indicated treatments, the cells were fixed with 4% paraformaldehyde in PBS (pH 7.4) and permeabilized with 0.1% Triton X-100 in PBS. The cells were blocked with 1% BSA in PBS, followed by overnight incubation with diluted primary antibodies. After overnight incubation, Alexa Fluor–conjugated secondary antibodies corresponding to species were added to the coverslips. Using a mounting medium with DAPI (H-1200-10, Vector laboratories), the coverslips were mounted on the glass slides. Confocal images were taken by laser scanning confocal microscope 510 Meta (ZEISS) and analyzed by ZEISS LSM Image Browser (ver.4.2.0.121). The number of puncta and the Pearson correlation coefficient were analyzed using EzColocalization plugin (18) from ImageJ.

qRT-PCR

Cells were lysed with TRIzol reagent (#15596018, Invitrogen) to extract total RNA according to the manufacturer’s instructions. Using PrimeScript First Strand cDNA Synthesis Kit (#6110A, Takara), an equal amount of RNA was reverse-transcribed into cDNA. Next, cDNAs were quantitatively analyzed with iCycler detection system (Bio-Rad) using TB Green Premix Ex Taq II (#RR820, Takara). The sequences of primers used in qRT-PCR are listed in Supplementary Table S2.

RNA-seq

Total RNA concentration was calculated by Quant-IT RiboGreen (R11490, Invitrogen). To assess the integrity of the total RNA, samples are run on the TapeStation RNA ScreenTape (5067-5576, Agilent Technologies). Only high-quality RNA preparations, with RIN greater than 7.0, were used for RNA library construction. A library was independently prepared with 1 μg of total RNA for each sample by Illumina TruSeq Stranded mRNA Sample Prep Kit (20020595, Illumina). The poly-A–containing mRNA molecules were purified with poly‐T–attached magnetic beads. Following purification, the mRNA is fragmented into small pieces using divalent cations under elevated temperatures. The cleaved RNA fragments are copied into first-strand cDNA and followed by second-strand cDNA synthesis. These cDNA fragments go through an end-repair process and are enriched with PCR to create the final cDNA library. The libraries were quantified using KAPA Library Quantification kits for Illumina Sequencing platforms according to the qPCR Quantification Protocol Guide (KK4854, KAPA Biosystems) and qualified using the TapeStation D1000 ScreenTape (5067-5582, Agilent Technologies), followed by sequencing using the NovaSeq6000 platform (Illumina). The gene set enrichment analysis (GSEA) was performed using the statistical software package R (version 4.2.1). The RNA-seq data generated in this study are publicly available in Gene Expression Omnibus at GSE248895.

Cell viability assay

Cells were plated to reach 60% to 70% confluency on the day of drug treatment. Medium-only wells were used as controls to measure background absorbance. The following day, cells were treated with compounds without medium change for 1 or 5 days. After adding the WST-1 reagent (11644807001, Millipore) and incubating for 2 hours, the absorbance of each well was read at 450 nm.

Clonogenic cell survival assay

A total of 2000 cells of 22Rv1 were seeded in six-well plates and treated with compounds at various concentrations. Cells were maintained in the compound-containing medium, which was changed every 3 days. On the 12th day, cells were fixed with 70% ethanol and stained with 1% crystal violet. The colonies were imaged and counted using GelCount (Oxford Optronix Ltd.)

Subcellular fractionation

The subcellular fractionation of 22Rv1 was conducted using the Subcellular Protein Fractionation Kit for Tissues (78840, Thermo Fisher Scientific) according to the manufacturer’s instructions. Briefly, homogenized cell pellets using ice-cold cytoplasmic extraction buffer were centrifuged at 500 × g for 5 minutes. The supernatant was transferred to another tube and labeled as the cytoplasmic fraction. The pellets were vortexed with different buffers (MEB, NEB, chromatin-bound extract buffer, and PEB buffer) and centrifuged repeatedly to isolate the membrane, soluble nucleus, chromatin-bound nucleus, and cytoskeletal fractions. Fractions were analyzed by immunoblotting.

Proximity ligation assay

The proximity ligation assay (PLA) was performed using the Duolink In Situ Red Starter Kit Mouse/Rabbit (DUO92101, Sigma-Aldrich) according to the manufacturer’s instructions. Briefly, after fixation and permeabilization, cells were blocked with blocking solution. Then, the cells were incubated with diluted primary antibodies, followed by PLA probes incubation in 37°C. After incubation, ligation and amplification were performed by adding diluted ligase solution and diluted amplification solution, respectively. Using mounting medium, the coverslips were mounted on the glass slides. Confocal images were taken by laser scanning confocal microscope 510 Meta (ZEISS) and analyzed by ZEISS LSM Image Browser (ver.4.2.0.121). The number of puncta was analyzed with image J_Fiji.

Statistical analysis

Data were statistically analyzed using GraphPad Prism 9.0 (GraphPad software; RRID: SCR_002798) and the statistical software package R (version 4.2.1). Data were expressed as the mean ± SEM. Statistical comparisons were performed using an unpaired t test for two groups or one/two-way ANOVA, followed by Dunnett’s post hoc procedures, for three or more groups. Linear mixed-effects models were employed for analyzing time-series longitudinal data, and, when necessary, a Bonferroni correction for multiplicity was applied. Replicates among experiments were both biologically and technically independent, carried out at different time points and by different individuals (where possible) to maximize reproducibility. For immunocytochemistry assays, 50 cells were counted from three independent experiments. For immunoblotting assays and densitometry analyses, at least three independent experiments were performed and analyzed. For clonogenic assays, three biologically independent measurements were taken. For WST-based cell viability assays, four biologically independent measurements were taken. For RNA-seq and qRT-PCR, three biologically and technically independent samples were collected. Mice xenograft experiments were performed with three to four mice in each group (LNcaP xenograft; vehicle: n = 3, ATC-324-injected: n = 3/22Rv1 xenograft; vehicle: n = 3, ATC-324-injected: n = 4). Where possible, sample sizes were chosen based on established protocols in previous publications and/or generally accepted criteria in the scientific community. A P value less than 0.05 was considered statistically significant.

Data availability

The raw data of RNA seq have been deposited to NCBI Gene Expression Omnibus (GSE248895). All other raw data generated in this study are available upon request from the corresponding author.

Results

AR-targeting AUTOTACs activate p62-dependent autophagy

To achieve targeted autophagic degradation of AR, we developed AR-targeting AUTOTACs (AR-AUTOTAC). The candidate TBLs were selected through in silico screening of 4,244 potential AR antagonists from the ChemDiv library, and prioritized based on their binding affinities to AR, lack of cross-reactivity with other steroid receptors, and safety considerations. MDV3100 (enzalutamide, ENZ), an FDA-approved drug with a good safety profile and higher binding affinity compared with first-generation antiandrogen inhibitors like bicalutamide, was chosen as the TBL. Enzalutamide was then conjugated with the previously reported p62 ZZ domain-targeting ligand (YT 6-2), which induces sequestration of p62 and autophagosome biogenesis (19), to serve as an ATL. PEG linkers were introduced to achieve higher water solubility and cell permeability (20). A series of AR-AUTOTACs were initially synthesized with various linker lengths, ranging from 3 to 5, named ATC323, ATC324, and ATC325, respectively (Fig. 1A).

Figure 1.

AR-targeting AUTOTACs activate p62-mediated autophagy. A, Chemical structures of AR-targeting AUTOTACs, ATC-323, ATC-324, and ATC-325. The p62-binding and AR-binding moieties are labeled with pink and blue boxes, respectively. B, Immunoblot analysis in WT and p62^−/− MEF cells treated with DMSO (vehicle), YT 6-2 (5 µmol/L), enzalutamide (10 µmol/L), and ATC-323, -324, -325 (5 µmol/L) for 24 hours. The levels of β-actin were examined as a loading control. C, Cell-based p62 oligomerization assay in HEK293T cells treated with ATC-324 and 325 (5 µmol/L, 24 hours). D, Immunofluorescence images of p62 and LC3 colocalization in HeLa cells treated with ATC-324 (2 µmol/L, 24 hours) in the presence or absence of bafilomycin A1 (100 nmol/L, 18 hours) under confocal microscopy. E–G, Quantification of the number of p62 (E) or LC3 puncta (F) and p62⁺LC3⁺puncta (G) in each cell. n = 50 cells. H and I, Immunoblot analysis of LC3 level in LNCaP cells treated with ATC-324 at indicated concentrations for 24 hours (H) and densitometry of H (I). The protein levels were normalized to β-actin. J and K, Immunoblot analysis of LC3 level in LNCaP cells treated with ATC-324 at indicated duration at 5 µmol/L (J) and densitometry of J (K). The protein levels were normalized to β-actin. Data are shown as means ± SEM (n = 3 independent experiments), and statistical significance was analyzed using two-way ANOVA, followed by Tukey post hoc procedures.

Nuclear magnetic resonance and mass spectrometry data confirmed the intended structures of these compounds (Supplementary Figs. S2 and S3). To assess the effectiveness of the newly synthesized AR-AUTOTACs in autophagy activation and their reliance on p62, we examined the conversion of LC3-I to LC3-II in WT and p62^−/− MEFs following treatments with AR-AUTOTACs. AR-AUTOTACs with an intermediate-length linker (ATC-324) and a long linker (ATC-325) exhibited high potency in activating autophagy in MEFs, similar to the p62 ligand YT 6-2. In contrast, no changes in the levels of LC3-II were observed in p62-deficient cells, suggesting that AR-AUTOTACs induce p62-mediated autophagy (Fig. 1B). Given that oligomerization of p62 is a crucial step for promoting autophagosome biogenesis and cargo collection leading to autophagic degradation (21, 22), we assessed the effectiveness of AR-AUTOTACs in inducing p62 polymerization. Notably, ATC-324 exhibited slightly higher efficacy in promoting p62 polymerization compared with ATC-325 (Fig. 1C). Accordingly, subsequent studies primarily utilized ATC-324. The p62 oligomerization ability of ATC-324 is derived from its p62-targeting component, YT 6-2, rather than from enzalutamide (Supplementary Fig. S4A). The increased levels of p62 oligomers destined for autophagic degradation were validated with the cotreatment of ATC-324 and bafilomycin A1 (Supplementary Fig. S4B). The confocal microscopic analysis provided further confirmation of ATC-324’s ability to induce autophagy in the prostate cancer cell line LNCaP. This was evident by the colocalization of p62 and LC3, along with the formation of distinct p62 and LC3 puncta. The accumulation of p62⁺LC3⁺ puncta was significantly enhanced when autophagy–lysosomal degradation was inhibited by bafilomycin A1, an inhibitor of autophagosome-lysosome fusion (Fig. 1D–G). Kinetics studies revealed that ATC-324–induced autophagy is both concentration- and time-dependent (Fig. 1H–K). These findings provide validation that our newly developed AR-AUTOTAC, ATC-324, exhibits high potency for activating autophagy in a p62-dependent manner.

ATC-324 induces degradation of full-length ARs, resulting in cytotoxicity in prostate cancer cells

The degradation capability of AR-AUTOTACs on AR proteins was evaluated in LNCaP cells. ATC-324 and ATC-325, both capable of inducing autophagic flux, effectively degraded AR proteins (Fig. 2A). In contrast, ATC-323, which is ineffective in inducing autophagy (Fig. 1B), had minimal impact on AR degradation (Fig. 2A). In control experiments, enzalutamide (TBL) or YT 6-2 (ATL) treatments had a minimal effect on AR level (Fig. 2A). The half-maximal degradation concentration (DC₅₀) of ATC-324 in LNCaP was determined to be 2.05 µmol/L, with more than 50% of AR being degraded within 6 to12 hours of treatment (Fig. 2B–E). ATC-324 also effectively degrades AR in C4-2B cells, which is a castration-resistant cell line derived from LNCaP (Supplementary Fig. S5A and S5B). It should be noted that autophagic degradation involves temporary sequestration of cargos by autophagic membranes, which may lead to functional inhibition even before degradation (23). Notably, ATC-324 demonstrated sustained and progressive degradation of AR proteins over 24 hours. Confocal microscopic analysis indicates that ATC-324 recruits AR to p62 via its TBL and ATL as anticipated, resulting in the sequestration of AR puncta in the cytoplasm. In contrast, in control cells, AR proteins are predominantly localized in the nucleus (Fig. 2F and G). Next, we investigated whether AR degradation by ATC-324 occurs through p62-mediated autophagy. Upon p62 knockdown, ATC-324 was ineffective in degrading AR proteins (Fig. 2H). In the absence of p62, AR remains in the nucleus even with ATC-324 treatment (Supplementary Fig. S5C). Consistent with AR degradation by ATC-324 through autophagy, the presence of E64D and pepstatin A, which inhibit autophagic degradation, resulted in the accumulation of LC3-II and AR (Fig. 2I), reversing the cytotoxicity associated with AR degradation (Supplementary Fig. S5D). These findings provide evidence that ATC-324 induces the formation of an oligomeric complex comprising AR/enzalutamide/YT 6-2/p62, resulting in the autophagic degradation of AR.

Figure 2.

AUTOTAC ATC-324 induces selective autophagic degradation of AR. A, Immunoblot analysis of AR level in LNCaP cells treated with YT 6-2 (5 µmol/L, 24 hours), enzalutamide (10 µmol/L, 24 hours), and ATC-323, 324, 325 (5 µmol/L, 24 hours). The levels of β-actin were examined as a loading control. B and C, Immunoblot analysis of AR level in LNCaP cells treated with ATC-324 at indicated concentrations for 24 hours (B) and densitometry of B (C). The protein levels were normalized to β-actin. D and E, Immunoblot analysis of AR level in LNCaP cells treated with ATC-324 at indicated duration at 5 µmol/L (D) and densitometry of D (E). The protein levels were normalized to β-actin. F and G, Immunofluorescence images of p62 and AR in LNCaP cells treated with enzalutamide (10 µmol/L, 12 hours) and ATC-324 (5 µmol/L, 12 hours) under confocal microscopy (G) and the colocalization was analyzed using the Pearson correlation coefficient (total n = 25 cells). H, Immunoblot analysis of AR in control and p62 knockdown LNCaP cells treated with ATC-324 (5 µmol/L, 6 hours). I, Immunoblot analysis of AR in LNCaP cells treated with ATC-324 (5 µmol/L, 24 hours) in the presence or absence of E64D and pepstatin A. The levels of β-actin were examined as a loading control. Data are shown as means ± SEM (n = 3 independent experiments), and statistical significance was analyzed using one-way ANOVA, followed by Tukey post hoc procedures.

ATC-324 induces cytotoxicity involving apoptosis in AR-positive prostate cancer cells

To evaluate the therapeutic potential of ATC-324, we treated various prostate cancer cell lines with increasing concentrations of ATC-324 and determined cell survival. WST-1 assay revealed that ATC-324 induces cytotoxicity in AR-expressing LNCaP and 22Rv1 while demonstrating minimal cytotoxicity in AR-null PC3 and DU145 cells (Fig. 3A). Particularly noteworthy is that ATC-324 exhibits enhanced effectiveness in inducing cell death in castrate-resistant 22Rv1 cells that express high levels of the full-length AR along with several AR splice variants including AR-v7, as compared with hormone-sensitive LNCaP cells that express full-length AR proteins. Control experiments demonstrated no cytotoxicity with equivalent concentrations of YT 6-2 or enzalutamide treatments (Fig. 3A). The growth inhibitory effect of ATC-324 on 22Rv1 cells was further confirmed using clonogenic cell survival assay, which also showed no cytotoxicity of YT 6-2 or enzalutamide (Fig. 3B and C), consistent with the WST-1 assay results. Because AR signaling is known to inhibit apoptotic cell death in prostate cancer cells (24, 25), we investigated caspase activation in prostate cancer cells following treatments with ATC-324. Interestingly, cleaved caspase-3 was readily detected in 22Rv1 cells and to a lesser degree in LNCaP cells, whereas no active caspase-3 was observed in PC3 or DU145 cells upon ATC-324 treatments (Fig. 3D). These findings indicate that ATC-324-induced cytotoxicity is associated with apoptosis induction.

Figure 3.

ATC-324 inhibits proliferation and induces apoptosis in AR-positive prostate cancer cells. A, WST-1 assay of PC3, DU145, LNCaP, and 22Rv1 cells in four replicates, treated at the indicated concentrations of vehicle, YT 6-2, enzalutamide, and ATC-324 for 24 hours. n.s., not significant; PCa, prostate cancer. B and C, Clonogenic cell survival assay of 22Rv1 cells in triplicates, exposed to vehicle, YT 6-2, enzalutamide, and ATC-324 at indicated concentrations (B) and the quantification of B (C). D, Immunoblot analysis of cleaved caspase-3 in PC3, DU145, LNCaP, and 22Rv1 cells treated with ATC-324 at indicated concentrations for 24 hours. The levels of β-actin were examined as a loading control. Data are shown as means ± SEM (n = 3 independent experiments), and statistical significance was analyzed using one-way ANOVA, followed by Tukey post hoc procedures.

ATC-324 is capable of degrading AR-LBD mutants as well as AR-v7 heterodimerized with full-length AR

Various forms of AR mutants have been identified in patients with CRPC. Particularly noteworthy are the AR mutants that harbor mutations in the LBD as well as AR splice variants lacking the LBD like AR-v7 (Supplementary Fig. S6A). These genetic alterations serve as significant drivers of resistance against AR antagonists (11–13). Recognizing the clinical significance of these mutants, we investigated the potential of ATC-324 in degrading AR-LBD mutants or AR-v7. For this purpose, AR-negative PC3 cells were transiently transfected with AR-LBD mutant constructs containing an N-terminal FLAG tag and subsequently subjected to treatments with ATC-324, enzalutamide (TBL only), or YT 6-2 (ATL only). The mutant constructs employed in this study included F877L and T878A, as prior research indicated their particular significance in enzalutamide responses, demonstrating their ability to utilize enzalutamide as an agonist (26). Remarkably, ATC-324 exhibited significant efficacy in degrading all tested AR mutants such as L702H, H874Y, F877L, T878A, and M896V, whereas enzalutamide or YT 6-2 displayed minimal impact on their expression levels (Fig. 4A). A similar degradation pattern was also observed in HEK293T cells transfected with the AR-LBD mutant constructs (Supplementary Fig. S6B).

Figure 4.

ATC-324 degrades AR-LBD mutants and AR-v7/AR-FL heterodimer. A, Immunoblot analysis of AR with or without LBD mutation (L702H, F877L, T878A, M896V, and H874Y) in PC3 cells transfected with the indicated plasmid and treated with ATC-324 (5 µmol/L, 24 hours). The levels of β-actin were examined as a loading control. B and C, Immunoblot analysis of AR-FL and AR-v7 in 22Rv1 cells treated with ATC-324 at the indicated concentrations for 24 hours (B) and densitometry of B (C). The protein levels were normalized with β-actin. D and E, Fluorescence images of the PLA of FLAG and p62 in PC3 cells (D) and quantification of the number of red puncta (E). The cells were transfected with untagged AR-FL and FLAG-AR-v7 plasmids and treated with indicated chemicals (5 µmol/L, 12 hours). n >30 cells. F, Immunoblot analysis of FLAG in PC3 cells transfected with indicated plasmids and treated with or without ATC-324 (5 µmol/L, 24 hours). The protein levels were normalized to β-actin and the densitometry values of FLAG corresponding to AR-FL and AR-v7 were labeled below the panels. G, Immunoblot analysis of FLAG in PC3 cells transfected with indicated plasmids and treated with or without ATC-324 (5 µmol/L, 24 hours). The protein levels were normalized to β-actin and the densitometry values of FLAG corresponding to AR-FL and AR-v7 were labeled below the panels. Data are shown as means ± SEM (n = 3 independent experiments), and statistical significance was analyzed using one-way ANOVA, followed by Tukey post hoc procedures.

Enzalutamide, used as a TBL, is a nonsteroidal AR antagonist that competitively inhibits androgen binding to AR. Thus, ATC-324 was anticipated to primarily degrade full-length AR (AR-FL), while not affecting AR-v7. To our surprise, ATC-324 effectively induced cytotoxicity in 22Rv1 cells characterized by high levels of AR-v7 expression (Fig. 3). This finding was unexpected, considering a previous report indicating that 22Rv1 cells continue to proliferate when only AR-FL is downregulated (27). Intriguingly, ATC-324 treatment resulted in the codegradation of both AR-FL and AR-v7 in 22Rv1 cells (Fig. 4B and C). The degradation of AR-v7 was not induced by YT 6-2 treatment alone (Supplementary Fig. S6C). Furthermore, the knockdown of p62 or autophagy inhibition blocked the degradation of both AR-FL and AR-v7 by ATC-324 (Supplementary Fig. S6D and S6E). AR-v7 has been demonstrated to form homodimers or heterodimers with AR-FL, leading to transcriptional activation of tumor growth-related genes even in the absence of androgen (28). Thus, we hypothesized that ATC-324 degrades AR-v7, which lacks the LBD, by binding to AR-FL/AR-v7 heterodimers. To address this hypothesis, we conducted PLA targeting FLAG and p62 in PC3 cells transfected with untagged AR-FL and FLAG-tagged AR-v7 expression vectors, followed by treatment with ATC-324, enzalutamide (TBL) or YT6-2 (ATL). ATC-324 resulted in a significant increase in red puncta, indicating ATC-324–induced interaction between FLAG-tagged AR-v7 and p62 when AR-v7 is coexpressed with AR-FL (Fig. 4D and E). Importantly, this interaction was absent in PC3 cells transfected with FLAG-tagged AR-v7 in the absence of AR-FL (Supplementary Fig. S6F). Additionally, we transfected PC3 cells with FLAG-tagged AR-FL and AR-v7, either separately or together, and treated them with ATC-324. As expected, ATC-324 effectively degraded AR-FL but not AR-v7 when transfected separately. However, when AR-v7 was coexpressed with AR-FL, ATC-324 effectively degraded both AR-FL and AR-v7 (Fig. 4F). To further explore whether the degradation of AR-v7 by ATC-324 resulted from the degradation of AR-FL/v7 heterodimer, we generated the plasmid encoding AR-v7 with the mutated FXXLF motif, critical for AR dimerization (28). We found that in the presence of AR-FL, the mutated AR-v7 (²³AQNAA²⁷) was not degraded whereas WT AR-v7 (²³FQNLF²⁷) was efficiently degraded by ATC-324 treatment (Fig. 4G). These results highlight the promising therapeutic potential of ATC-324 in targeting a wide range of AR mutants and splice variants relevant to prostate cancer.

ATC-324 reduces the levels of nuclear AR and inhibits the transcriptional activity of both AR-FL and AR-v7

We aimed to determine whether the downregulation of AR-FL and AR-v7 translates into inhibiting their activity as transcription factors. Initially, we investigated the subcellular localization of AR proteins with and without ATC-324 treatment. Notably, AR-v7 proteins were predominantly localized in the nucleus of 22Rv1 cells. Following ATC-324 treatment, we observed downregulation of both AR-FL and AR-v7 in both the cytoplasm and nucleus (Fig. 5A). Upon longer exposure of the AR immunoblot, we also observed a reduction of chromatin-bound AR following ATC-324 treatment (Fig. 5B). To investigate the impact of ATC-324 on the expression of AR target genes, we performed gene expression profiling through RNA-seq in 22Rv1 cells treated with vehicle (DMSO), YT 6-2, enzalutamide, or ATC-324. For the RNA-seq analysis, AR signature (29), neuroendocrine prostate cancer (NEPC) signature (30), AR-v7-upregulated (31) and downregulated gene sets (32) were curated from previous studies. GSEA revealed a significant decrease in the expression of AR target genes upon treatment with ATC-324 [normal enrichment score (NES) = −2.032, P = 0.001 and FDR = 0.002). Consistent with the degradation of AR-v7 by ATC-324, a gene set known to be upregulated by AR-v7 was also significantly downregulated (NES = −2.829, P < 0.001 and FDR < 0.001), whereas the induction of the AR-v7-downregulated gene set was statistically insignificant. Previous studies have shown that AR-targeting drugs are often associated with the induction of AR-negative, NEPC-like phenotype, or enrichment of preexisting NEPC (33). However, when we examined the effect of ATC-324 treatment on the NEPC signature gene set, we only observed minimal and insignificant effects (Fig. 5C–E).

Figure 5.

ATC-324 decreases the levels of both cytoplasmic and nuclear AR, including chromatin-bound AR, and suppresses the transcriptional activity of AR-FL and AR-v7. A, Immunoblot analysis of AR-FL and AR-v7 in subcellular fractions of 22Rv1 cells treated with ATC-324 for 24 hours at the indicated concentrations. B, Immunoblot analysis of AR-FL and AR-v7 in the chromatin-bound fraction of 22Rv1 cells treated with ATC-324 for 24 hours at the indicated concentrations. Subcellular fractionation was validated through the presence of HSP-90, transferrin receptor (TsfR), SP1, histone H1, and CK8. C, GSEA plots with the NES of the AR signature (29), NEPC signature (30), AR-v7-upregulated (31) and downregulated gene sets (32) in 22Rv1 cells treated with ATC-324 (5 µmol/L, 24 hours), compared with vehicle groups. D, qRT-PCR of AR-target genes (PSA, ATAD, KLK2, and TMPRRS2) and AR-v7–specific upregulated genes (NUP210, EDN2, UBE2C, and BUB1) in 22Rv1 cells treated with vehicle, YT 6-2 (5 µmol/L, 24 hours), enzalutamide (5 µmol/L, 24 hours), and ATC-324 (5 µmol/L, 24 hours). n = 3. E, Heatmap of the expression levels of indicated gene sets among vehicle-, YT 6-2 (5 µmol/L, 24 hours)-, enzalutamide (5 µmol/L, 24 hours)-, and ATC-324 (5 µmol/L, 24 hours)-treated 22Rv1 cells (n = 3). Expression levels of each gene were log₂-transformed, and the signature score was generated from GSEA using gene set variation analysis that is a nonparametric, unsupervised method for estimating variation of gene set enrichment through the samples of an expression data set. Data are shown as means ± SEM (n = 3), and statistical significance was analyzed using two-way ANOVA, followed by Tukey post hoc procedures. n.s., not significant.

To validate the findings from the RNA-seq analysis, we performed qRT-PCR analysis. The results demonstrated that treatment with ATC-324 led to the downregulation of AR target genes such as PSA, TMPRSS2, KLK2, and ATAD2 in comparison to both the control solvent or enzalutamide treatment. Additionally, the expression of genes specifically regulated by AR-v7 including NUP210, EDN2, UBE2C, and BUB1 exhibited a reduction following ATC-324 treatment, whereas enzalutamide treatment did not affect these gene expressions (Fig. 5D). Importantly, the expression of non-AR–induced genes such as HK1 and TNFα remained unchanged (Supplementary Fig. S7A). These findings provide strong evidence for the effective and specific targeting of AR by ATC-324.

As controls, we also analyzed the effects of YT 6-2 (ATL only) or enzalutamide (TBL only) treatment on the expression levels of AR and AR-v7 target genes. Interestingly, although YT 6-2 exhibited minimal effects on AR signature genes, we observed a significant reduction in the AR-v7 upregulated gene set (NES = −2.429, P < 0.001, FDR < 0.001; Supplementary Fig. S7B). When comparing the effects of ATC324 and YT 6-2 treatments, we found a more pronounced downregulation of AR-v7 upregulated genes by ATC-324 (NES = −2.959, P < 0.001, FDR < 0.001; Supplementary Fig. S7C). This finding is consistent with our observation that ATC-324 leads to the downregulation of AR-v7 by activating p62-mediated autophagy even though the TBL (enzalutamide) does not interact with AR-v7 (Fig. 4F and G).

Enzalutamide treatment downregulated AR signature genes (NES = −2.148, P < 0.001, FDR < 0.001; Supplementary Fig. S7D). However, it had a minimal effect on AR-v7–regulated gene expression. This is consistent with its known function as an AR inhibitor by its binding to the LBD of AR. In contrast, ATC-324 treatment showed a remarkable and significant downregulation of AR-v7 upregulated gene expression when compared with enzalutamide treatment (NES = −2.949, P < 0.001, FDR < 0.001; Supplementary Fig. S7E). We further validated that AR transcriptional activity was impaired by ATC-324 in a concentration-dependent manner using 22Rv1 cells stably expressing an AR-responsive luciferase reporter. This finding supported that the disrupted AR activity correlates with AR degradation by ATC-324 (Supplementary Fig. S8A). Furthermore, for a comprehensive exploration of GSEA, we expanded our investigation to include AR and NEPC signatures curated by different studies (34, 35). This approach consistently yielded comparable findings (Supplementary Fig. S8B–S8E).

In vivo and ex vivo efficacies of ATC-324 in inhibiting tumor growth

To evaluate the therapeutic potential in vivo, we established 22Rv1 and LNCaP xenograft mouse models. The administration of ATC-324 at a dose of 20 mg/kg through ii.p. injection effectively suppressed tumor growth in both 22Rv1 and LNCaP models (Fig. 6A–C). However, control treatments with enzalutamide or YT 6-2 did not significantly affect the growth rate of 22Rv1 tumors (Supplementary Fig. S9A). Throughout the injection period and at the time of harvest, there were no noticeable adverse effects in mice treated with ATC-324, and body weight remained stable (Fig. 6D and E; Supplementary Fig. S9B). Hematoxylin and eosin staining of tumor tissues showed that ATC-324 induced cytotoxicity in both 22Rv1 and LNCaP tumors, resulting in microcystic spaces and discohesion (Fig. 6F and G). Consistent with in vitro findings, IHC analysis showed decreased levels of AR expression in ATC-324–treated tumors compared with control tumors (Fig. 6H and I). Further IHC analysis also revealed decreased Ki67 expression, a marker for cell proliferation, and increased expression of cleaved caspase-3, a marker of cell apoptosis, in ATC-324-treated tumors from 22Rv1 xenograft (Fig. 6J and K; Supplementary Fig. S9C and S9D). These findings underscore ATC-324’s ability to effectively inhibit cell proliferation and induce cell death in tumors of 22Rv1 xenografts.

Figure 6.

ATC-324 inhibits the growth of prostate cancer in xenograft models. A, Schematic of establishing LNCaP and 22Rv1 xenografts in NSG mice and the treatment schedule with ATC-324 (20 mg/kg). B and C, Body weight of LNCaP xenograft mice (B) and 22Rv1 xenograft mice (C) from the vehicle and ATC-324–injected groups measured at the end of the experiment. D, Tumor volume of LNCaP xenograft mice measured on the indicated day. E and F, Hematoxylin and eosin (H&E) staining (E) and IHC analysis of AR (F) of LNCaP xenograft tumor from vehicle and ATC-324–injected groups. G, Tumor volume of 22Rv1 xenograft mice measured on the indicated day. H–K, Hematoxylin and eosin staining (H) and IHC analysis of AR-FL and AR-v7 (I), Ki67 (J) of 22Rv1 xenograft tumor from vehicle and ATC-324-injected groups. K, Quantification of J. L and M Luminescence of bone pieces derived from luciferase-expressing 22Rv1 cells in the BICA assay (L) and the quantification of luminescence (M). The number of bone pieces ≥15. Data are shown as means ± SEM and statistical significance was analyzed using unpaired t tests and linear mixed-effects models, followed by multiple comparison correction. (LNCaP xenograft; Veh., ATC-324: n = 3/22Rv1 xenograft; Veh.: n = 3, ATC-324: n = 4). n.s., not significant. (A, Created with BioRender.com.)

The tumor microenvironment is a critical determinant of cancer cell response to therapeutics (36). In up to 90% of patients with advanced CRPC, prostate cancer progression leads to skeletal metastasis (37). Bone metastasis is currently incurable and associated with greater morbidity and mortality, making it a major clinical and life-threatening complication in men with advanced prostate cancer. Although next-generation antiandrogens such as enzalutamide and abiraterone are initially effective, they fail in the majority of prostate cancer bone metastases. Thus, it is of significance to determine whether ATC-324 induces cytotoxicity in prostate cancer cells in the bone microenvironment. To address this, we utilized the BICA platform, developed to model the bone colonization of cancer cells and intraosseous tumor growth in ex vivo cultures (16). BICA maintains the in vivo features of the bone microenvironmental niche up to 6 weeks (16), thus serving as a powerful platform to rapidly screen for drug efficacy in the bone microenvironment. Luciferase-tagged 22Rv1 or LNCaP cells were delivered into the hindlimbs of mice through intra-iliac artery injection. BICA was then established by extracting and fragmenting the femur and tibia containing the cancer cells, followed by ex vivo cultures. Only ATC-324 effectively suppressed the growth of bone-resident 22Rv1 cells in a dose-dependent manner (Supplementary Fig. S9E and S9F). At a concentration of 2 µmol/L, ATC-324 nearly eradicated intraosseous 22Rv1 cells, suggesting that its inhibition of prostate tumor growth is equally, if not more, potent in the bone microenvironment compared with the in vitro culture condition. In contrast, even at the highest concentration tested (8 µmol/L), enzalutamide treatment failed to induce cytotoxicity in intraosseous 22Rv1 cells. Interestingly, instead of inhibiting growth, enzalutamide treatment promoted the growth of intraosseous CRPC 22Rv1 cells (Fig. 6L and M).

In the LNCaP BICA assay, only 9.5% of BICA bone fragments showed detectable tumor growth, and at much slower rates, in contrast to the exponential tumor growth observed in 37.5% of 22Rv1 bone fragments. This finding aligns with our previous study, which demonstrated that LNCaP fails to form intraosseous tumors in an intratibial injection mode (38). In the LNCaP BICA assay, ATC-324 treated group showed less tumor growth compared with other groups (Supplementary Fig. S9G). Although enzalutamide promoted growth of bone-resident 22Rv1 cells expressing AR mutants (Fig. 6M), it slightly reduced the growth of bone-resident LNCaP cells with WT AR. Due to insufficient growth of LNCaP cells in the BICA model, the data were statistically insignificant.

ATC-324 remains potent in enzalutamide-resistant prostate cancer cells

Patients with prostate cancer who have stopped responding to hormone therapy are most commonly treated with antiandrogens such as enzalutamide. Unfortunately, a majority of patients eventually develop enzalutamide-resistant phenotypes. Therapy resistance is thought to involve the activation of noncanonical AR pathways (14). To investigate whether ATC-324-mediated degradation of AR proteins can overcome enzalutamide resistance, we evaluated its potency using two pairs of control and enzalutamide-resistant prostate cancer cell lines. The generation and characterization of enzalutamide-resistant cell line derivatives, namely LNCaP^EnzR and LAPC-4^EnzR, have been previously reported (39). Briefly, these resistant cell lines were generated by subjecting their parental cell lines to long-term cultures (>6 months) in the presence of 10 µmol/L enzalutamide. To generate control cell lines with similar passage numbers, parental LNCaP and LAPC4 cells were grown alongside the Enz-treated cell lines. We noted that the high-passage LNCaP cells (control and Enz-resistant) tend to form clusters, which slightly increases their resistance to ATC-324 compared with low-passage LNCaP cells shown in Fig. 3A. Nonetheless, as shown in Fig. 7, ATC-324 resulted in AR degradation and induced cytotoxicity in enzalutamide-resistant prostate cancer cell lines almost as effectively as in the corresponding control cell lines.

Figure 7.

ATC-324 degrades AR and exhibits cytotoxicity in enzalutamide-resistant prostate cancer cells. A, WST-1 assay of control and enzalutamide-resistant LNCaP and LAPC4 cells in four replicates, treated with at indicated concentrations of vehicle, YT 6-2, enzalutamide, and ATC-324 for 5 days. B, Immunoblot analysis of AR in control and enzalutamide-resistant LAPC4 and LNCaP cells treated with increasing concentrations of YT 6-2, ATC-324, and enzalutamide. The protein levels were normalized to β-actin. Data are shown as means ± SEM and statistical significance was analyzed using two-way ANOVA, followed by Tukey post hoc procedures.

Discussion

In this study, we successfully developed ATC-324, a novel compound capable of degrading various forms of AR, including WT AR, AR-v7, L702H-, F877L-, T878A-, H874Y-, and M896V-mutated AR proteins through the autophagy–lysosomal degradation pathway. Although bulk autophagy is a nonselective process that indiscriminately targets cytosolic components, selective autophagy specifically recognizes and degrades substrates through receptor proteins, such as p62/SQSTM1, NBR1 and OPTN. Previously, we discovered that N-terminally arginylated proteins are recognized by the ZZ domain of p62, resulting in oligomerization of p62, formation of autophagosome, and subsequent degradation through autophagy (22). Based on this discovery, we have developed small molecule ligands targeting the ZZ domain of p62 and activating p62-dependent autophagy (21). Using these synthetic ligands, we recently developed the AUTOTAC platform for targeted protein degradation (1). Because a bivalent AUTOTAC molecule brings the target protein to the autophagic receptor p62, AUTOTAC-mediated target protein degradation is independent of ubiquitination. In contrast, PROTAC, the most well-known TPD platform, relies on the structure–function relationship between a specific E3 ligase and its target protein to facilitate ubiquitination. This key distinction underscores the versatility and robustness of the AUTOTAC platform, particularly in degrading mutated, structurally altered target proteins, possibly alongside their binding partners critical in disease progression.

Employing a small molecule ligand of p62 to target AR in prostate carcinoma may offer several advantages. Recent studies have suggested that p62 serves as a prognostic marker in human cancers (40, 41). Thus, ATC-324 employing the p62 ligand may preferentially target malignant cancer cells. In addition, p62 contains nuclear localization sequences and nuclear export signal, allowing it to shuttle between the nucleus and the cytoplasm (42). Consequently, ATC-324 may effectively target both cytoplasmic and nuclear AR proteins. Subcellular fractionation of 22Rv1 cell lysates followed by immunoblot analysis confirmed that ATC-324 reduces the levels of both cytoplasmic and nuclear AR proteins (Fig. 5). ATC-324–mediated nuclear AR/p62 complex likely exits the nucleus as evidenced by AR⁺/p62⁺ puncta being predominantly in the cytoplasm (Fig. 2F). By comparing the cell viability and AR degradation efficiency results across various AR-positive prostate cancer cell lines (Figs. 3A, 7A and B; Supplementary Fig. S5A and S5B), we estimate that approximately 60% to 80% of AR was degraded at concentrations near the IC₅₀ of ATC-324 (Supplementary Table S3). To further examine the relationship between AR protein degradation and its transcription activity, we established a 22Rv1 cell line engineered to express the AR response element luciferase reporter (LTV-ARRE-Luc). We found that ATC-324 inhibition of AR activity (Supplementary Fig. S7A) correlates with ATC-324–mediated AR degradation (Fig. 4B). We surmise that the sequestered ARs are rendered nonfunctional even before they are hydrolyzed in the lysosome. Our data show that as long as AR mutants retain the enzalutamide binding motif, ATC-324 can degrade them regardless of other structural alterations. Moreover, ATC-324 is capable of degrading AR splice variants without the LBD domain, provided they heterodimerize with WT AR.

Currently, most available selective AR modulators, such as SGAAs, target the LBD of AR. SGAA resistance is largely attributed to AR mutations in the LBD or AR splice variants lacking the LBD (12, 13). Because the DNA-binding domain (DBD) of AR shares high homology with other nuclear receptors and the N-terminal domain (NTD) of AR exhibits intrinsic disorder, developing small molecule inhibitors that target the DBD or NTD faces challenges (43). Therefore, targeted protein degradation (TPD), exemplified by the PROTAC platform, has emerged as a promising therapeutic approach. TPD is binding event-driven and is independent of functional inhibition. As a result, TPD compounds often exhibit their efficacies at significantly lower concentrations compared with traditional small molecule inhibitors (3). However, initial PROTAC compounds have only shown promise in treating patients with CRPC with specific AR mutations excluding AR-v7, which is a significant driver of resistance to the current antiandrogen therapies (13, 44, 45). Recently, AR variant-targeting PROTACs using DBD or NTD-targeting ligands or peptides, such as MTX-23, ITRI-90, and Au-AR pep-PROTAC, were developed (46–48). Despite these advancements, several limitations remain. PROTACs rely on ubiquitination via E3 ligases to degrade target proteins, necessitating precise ternary complex formation between the PROTAC, the target protein, and the E3 ligase (3, 20). In the case of AR-v7, the structural differences between AR-FL and AR-v7, particularly the absence of the LBD, make forming an adequate ternary complex challenging, as evidenced by the significant gap between DC₅₀ of AR-FL and AR-v7. Off-target issues are another limitation for AR-v7 PROTACs, as they utilize DBD- or NTD-targeting ligands, previously known to have off-target effects, to target AR-v7. In contrast, the AUTOTAC platform does not rely on ubiquitination and ternary complex formation for its mode of action. Consequently, AR-AUTOTAC shows comparable DC₅₀ values for both AR-FL and AR-v7 and utilizes enzalutamide as the AR-binding moiety, which is validated for its safety and specificity. These advantages of AR-AUTOTAC offer a new therapeutic approach as clinically applicable compounds for patients with SGAA-resistant CRPC.

Cancer cells inevitably develop acquired resistance to most, if not all, prolonged therapies (49). We acknowledge that AR degradation alone is unlikely to eradicate all prostate cancer cells. In anticipation of resistance, we are also exploring additional therapeutics that can synergize with ATC-324. One promising pathway is ferroptosis, which our RNA-seq analysis of 22Rv1 cells treated with ATC-324 shows to be significantly impacted. Our preliminary study demonstrates that erastin, an inducer of ferroptosis, can synergize with ATC-324 to induce cell death in various cancer cell lines. We are currently investigating this combination of treatments to mitigate potential resistance mechanisms. In a separate experiment, we have also examined whether ATC-324 synergizes with first-line taxane chemotherapy. Our preliminary data suggest that ATC-324 and docetaxel treatments exhibit synergistic cytotoxicity in castrate-resistant 22Rv1 cells.

The TPD platform is not exempt from potential resistance development. Upon long-term treatments with cereblon (CRBN)- and von Hippel–Lindau (VHL)–mediated PROTACs, cancer cells manage to silence components of the E3 ligase complex involved in CRBN- and VHL-mediated protein degradation, resulting in resistance to those PROTAC treatments (50, 51). The human genome encodes more than 600 E3 ligases, and VHL and CRBN are dispensable components of the ubiquitin–proteasome system (UPS) required to maintain homeostasis. Currently, there is controversy with regard to the essential components of the autophagic machinery, and it remains to be investigated whether prolonged AUTOTAC treatment also leads to acquired resistance. Addressing the potential of acquired resistance and understanding the underlying mechanism will be crucial for optimizing the effectiveness of the AUTOTAC platform in the future. We intend to characterize prostate cancer cells following long-term treatment with ATC-324 and address critical questions, including the potential induction of an AR-negative, NEPC-like phenotype, in our follow-up studies.

Despite being preliminary, our animal study and BICA assay demonstrate ATC-324’s therapeutic potential. The AUTOTAC platform may be particularly well-suited for protein degradation in vivo. The UPS utilized by PROTAC is maximally active in rapidly proliferating cells to regulate the half-life of regulatory proteins involved in cell cycle, replication, and cell signaling as well as quality control of misfolded and damaged proteins (52, 53). UPS-based degraders exhibit excellent degradation efficacy (DC₅₀) ranging from sub nmol/L to low nmol/L in cultured cells (54–56). However, the minimal efficient doses in vivo are surprisingly high, ranging from 5 to 100 mg/kg (54–56). Moreover, the first-in-class AR-targeting PROTAC, ARV-110, requires a high dose of 420 mg, which exceeds the doses typically used for traditional small molecule inhibitor therapies (44). In contrast to the UPS, the primary functions of autophagy include the elimination of misfolded proteins and their aggregates. Autophagy is constitutively active in both cultured cells and mammalian tissues, and the autophagy–lysosome machinery has the capacity to degrade large protein complexes independent of proliferative status. Consistently, AUTOTAC degraders exhibit narrower differences between efficient concentrations in vitro and in vivo. ATC-324 showed DC₅₀ of ∼2 to 5 µmol/L in cell cultures and an efficient dose of 15 to 20 mg/kg in mice by i.p. injection. Undoubtedly, further research is required to enhance the efficacy of AR-targeting AUTOTAC through medicinal chemistry and to develop optimal AR-AUTOTAC drug delivery systems.

Supplementary Material

Supplementary Information

Supplementary legends Supplementary Tables 1-3 Supplementary Figures 1-8

Authors’ Disclosures

T.H. Bae reports grants from the Ministry of Science and ICT, Korea and the Ministry of Education, Korea, during the conduct of the study, as well as nonfinancial support from AUTOTAC Bio. Inc. outside the submitted work, as well as a patent for PCT/KR2022/017075 pending to Seoul National University, AUTOTAC Bio. Inc. K.W. Sung reports a patent for PCT/KR2022 /017075 pending to Seoul National University and AUTOTAC Bio. Inc. S.R. Mun reports grants from National Research Foundation of Korea during the conduct of the study, as well as grants from National Research Foundation of Korea outside the submitted work. H.K. Kwon reports a patent for PCT/KR2022/017075 pending to Seoul National University, AUTOTAC Bio. Inc. E.I. Heath reports other support from Astellas, AstraZeneca, Bayer, EMD Serono, Gilead, Novartis, Sanofi, Janssen, Astellas, Caris, Seattle Genetics, Arvinas, Bio X Cell, Bristol Myers Squibb, Calibr, Calithera, Corcept, Corvis, Daiichi Sankyo, Eisai, Exelixis, Five Prime, Fortis, GlaxoSmithKline, Gilead Sciences, Harpoon, Hoffman-La Roche, Infinity, iTeos, Merck Sharp & Dohme, Merck, Mirati, Modra, Oncolys, Peloton, Pfizer, Pharmacyclics, POINT Biopharma, and Seattle Genetics outside the submitted work. M.L. Cher reports other support from Dept. of Urology research fund during the conduct of the study. Y.T. Kwon reports a patent for PCT/KR2022/017075 pending to Seoul National University, AUTOTAC Bio. Inc. H.-R.C. Kim reports grants from NIH R01CA123362, NIH R01 CA282040, and NIH P50CA186786 during the conduct of the study. No disclosures were reported by the other authors.

Authors’ Contributions

T.H. Bae: Conceptualization, resources, data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. K.W. Sung: Conceptualization, resources, investigation. T.M. Pham: Conceptualization, resources, data curation, formal analysis, validation, investigation, writing–original draft, writing–review and editing. A.J. Najy: Data curation, formal analysis, validation, investigation, writing–review and editing. A. Zamiri: Resources, data curation, validation, investigation. H. Jang: Data curation, formal analysis. S.R. Mun: Resources, data curation, investigation. S. Kim: Data curation, formal analysis, supervision, methodology, writing–review and editing. H.K. Kwon: Resources, investigation. Y.S. Son: Methodology. D. Shi: Formal analysis, validation. S. Kregel: Resources. E.I. Heath: Formal analysis, validation. M.L. Cher: Formal analysis, validation. Y. Kwon: Conceptualization, supervision, funding acquisition, project administration, writing–review and editing. H.-R.C. Kim: Conceptualization, data curation, supervision, project administration, writing–review and editing.