Raludotatug Deruxtecan, a CDH6-Targeting Antibody–Drug Conjugate with a DNA Topoisomerase I Inhibitor DXd, Is Efficacious in Human Ovarian and Kidney Cancer Models

This article is featured in Selected Articles from This Issue, p. 255

Human Cadherin-6 (CDH6) is a single-transmembrane protein consisting of 790 amino acids classified into the type 2 Cadherin family. The human CDH6 gene was first cloned in 1995 (1). CDH6 is specifically expressed in the brain and kidneys during development and has been reported to play critical roles in the formation of central nervous system circuitry and expressed in the proximal renal tubule (2–6). CDH6 expression is increased specifically in several cancer types, such as serous-type ovarian cancer (OVC; ref. 7), renal cell carcinoma (RCC; refs. 6, 8), and thyroid cancer (9). CDH6 enhances epithelial–mesenchymal transition and promotes cell migration and invasion (10). CDH6 is also involved in lymph node metastasis in cancer patients and correlates with prognosis (11, 12).

Antibody–drug conjugate (ADC) technology is a promising treatment modality that takes advantage of the specific binding ability of monoclonal antibody (mAb) to deliver potent cytotoxic agents selectively to antigen-expressing tumor cells in order to increase the effectiveness and suppress off-target systemic toxicity. The mode of action of ADCs is generally considered to be as follows: after binding to the cell-surface antigen, the ADC-antigen complex is internalized into the lysosomal compartment where the ADC is cleaved to release a cytotoxic payload drug (13, 14). Several ADCs have been approved to date, and a large number of clinical studies for new ADCs are currently under way (13–17). DXd-ADC technology comprises an exatecan derivative (DXd) payload and a protease-cleavable maleimide Gly-Gly-Phe-Gly (GGFG) tetrapeptide-based linker (18–20). DXd has greater potency to inhibit DNA topoisomerase I (TOP1) than SN-38, which is an active metabolite of irinotecan (21). TOP1 inhibitors, such as exatecan and DXd, bind to TOP1–DNA cleavage complex and stabilize it, and then induce double-strand DNA breaks and apoptosis. Maleimide GGFG tetrapeptide-based linker is cleaved by lysosomal enzymes such as cathepsin, which is highly expressed in tumors, thereby limiting the free payload in plasma (22). These characteristics of DXd-ADC technology lead to high plasma stability, a short systemic half-life of the payload, and the ability to optimize the mean drug-to-antibody ratio (DAR) to as high as 8 for each target (18–20, 23, 24). The potent antitumor activity of ADCs applied with DXd-ADC technology has been demonstrated in preclinical studies and clinical settings, such as for trastuzumab deruxtecan (T-DXd), a HER2-targeting ADC (18–20, 22, 25–28); datopotamab deruxtecan (Dato-DXd), a TROP2-targeting ADC (29–31); patritumab deruxtecan (HER3-DXd), a HER3-targeting ADC (23, 32–34); and ifinatamab deruxtecan (I-DXd), a B7-H3-targeting ADC (35).

Raludotatug deruxtecan (R-DXd; DS-6000) is an ADC that is structurally composed of a humanized anti-CDH6 IgG1 mAb with DXd-ADC technology (18, 19). The target DAR is 8. R-DXd specifically binds to CDH6 on the surface of its target cells, which leads to the internalization of R-DXd into the lysosome. The DXd released from R-DXd in the target cells inhibits cell replication and induces apoptosis. Furthermore, we confirmed the membrane permeability of DXd, which might lead to cell death in tumors expressing CDH6 heterogeneously. In addition, R-DXd inhibited tumor growth and induced tumor regression in mice bearing CDH6-expressing tumors. Taken together, these findings suggest that R-DXd may be effective for treating patients with CDH6-expressing tumor. R-DXd can be efficiently internalized into cells, exhibits significant antitumor activity, and has an acceptable safety profile in monkeys, suggesting its potential for treating serous-type OVC and RCC patients.

DNA immunization with in vivo electroporation was conducted to generate anti-human CDH6 antibodies after treatment with hyaluronidase (Sigma-Aldrich Co. LLC) to the lower limbs of each rat. WKY/Izm rats (Japan SLC) were immunized biweekly five times by the intramuscular injection of a human CDH6-expressing vector, followed by in vivo electroporation using ECM830 Square Wave Electroporation System and BTX531 Two-Needle Array Electrodes (BTX). Rats with high antibody titers were boosted by the injection of 293T cells transiently expressing human CDH6 into muscle 3 days before harvesting cells from their lymph nodes. Hybridoma fusion with SP2/0-ag14 mouse myeloma cells was conducted by electrofusion using Hybrimune Hybridoma Production Systems (Cyto Pulse Sciences). Hybridoma supernatants that displayed selective reactivity with human CDH6 were screened by Cell ELISA using 293α/CDH6 cells. Clones that tested positive by flow cytometry were tested for binding to human CDH6 and cynomolgus monkey CDH6.

Rat anti-CDH6 antibodies were purified from the culture supernatant of each hybridoma. The cDNA encoding rat G019 mAb was sequenced and used to design a humanized IgG antibody, designated G019 H1L2. This G019 H1L2 antibody was produced in FreeStyle 293-F cells (Thermo Fisher Scientific Inc.) and purified by Protein-A chromatography.

R-DXd was synthesized in accordance with the published procedure (18), conjugating the DXd payload with the humanized anti-CDH6 antibody. The antibody solution (10 mg/mL in PBS, 5 mmol/L EDTA) was prepared to pH 7.0 by 1 mol/L K2HPO4. 10 mmol/L TCEP was added. The mixture was incubated for 2 hours at 37°C. Then, the mixture was cooled to 15 °C. Drug Linker (10 mmol/L in DMSO) were added. The mixture was incubated for 1 hour. N-acetyl-L-cysteine (100 mmol/L) was added. The reaction mixture was applied to equilibrated columns (NAP-25). The ADC was eluted by an appropriate volume of ABS buffer (10 mmol/L acetate buffer, 5% sorbitol, pH5.5). This purification process was repeated three times.

The drug distribution was analyzed by hydrophobic interaction chromatography, Column: TSKgel Butyl-NPR 4.6×100 mm, Tosoh Corporation, Flow: 0.8 mL/minute, mobile phase B% 5%–50% (0–20 minutes), mobile phase A: 1.5 mol/L ammonium sulfate, 25 mmol/L phosphate buffer, pH 7.0, mobile phase B: 25% Isopropanol, 25 mmol/L phosphate buffer, pH 7.0, detection: 280 nm.

Control ADC was synthesized using a humanized IgG1 isotype control antibody and the same drug linker as R-DXd, resulting in a comparable DAR.

The human ovarian cancer cell lines NIH:OVCAR-3 and PA-1, human renal cancer cell line 786-O, Chinese hamster ovary cell line CHO-K1, and human Burkitt lymphoma cell line Ramos were purchased from ATCC. The human ovarian cancer cell line JHOC-5 was purchased from Riken BioResource Research Center. All cell lines were cultured with appropriate media [RPMI-1640 medium for NIH:OVCAR-3, 786-O, and Ramos; EMEM for PA-1; McCoy's 5a medium for ES-2; DMEM/F12 supplemented with 0.1 mmol/L nonessential amino acids for JHOC-5; and Ham's F-12K (Kaighn's) medium for CHO-K1] supplemented with 10% FBS at 37°C in a 5% CO₂ atmosphere. All cell lines used in this study were routinely authenticated by comparing morphology and growth characteristics with the suppliers’ data. No Mycoplasma testing was performed on any of the cell lines. Cells were cultured in accordance with the manufacturers’ guidelines and working cell banks were created. The working cell banks were stored in liquid nitrogen or in a deep freezer set at −80°C and were allowed to propagate for no more than 3 months after thawing.

The targeted sequences were designed in exon 2 and intron 3 using CRISPRdirect (https://crispr.dbcls.jp/) based on the genomic sequence of human CDH6 (Gene ID: 1004). The targeted sequences were as follows: 5′-CTCACTACGC ACAGACTCGA CGG-3′ and 5′-ACACCCACAA CCATCCGCAT AGG-3′. Based on the designed sequences, GeneArt Precision gRNA Synthesis Kit was used to generate sgRNA. NIH:OVCAR-3 cells were introduced with Cas9 and sgRNA complexes by electroporation (Lonza Inc. Nucleofector:SE Buffer Prog. No. DS-120). Cell cloning by the limiting dilution method was performed from the introduced polyclonal cells. Genomic DNA was prepared from the obtained clones, and candidate cell lines in which the CDH6 gene had been knocked out by direct sequencing of amplified fragments were selected.

A DNA fragment encoding enhanced GFP-fusion human CDH6 (CDH6-eGFP) was chemically synthesized using the Gene Art Synthesis service by Thermo Fisher Scientific Inc. After the protein-coding DNA had been amplified by PCR (forward primer: 5′-GAGCGGCCGC GGATCCTTTG TACAAAAAAG CAGGCTTCGC CAC-3 and reverse primer: 5′-CGGTAGAATT GGATCCTTAT TTGTACAAGA AAGCTGGGTC CTTGTACAG-3′), the amplified fragment was inserted into the BamHI site of pLVSIN vectors (pLVSIN-CDH6-eGFP) using In-Fusion HD cloning kit (Clontech). Lentiviruses were prepared using 3rd Generation Packaging Systems Mix (Applied Biological Materials Inc.), in accordance with the manufacturer's instructions. Lipofectamine 2000 (Thermo Fisher Scientific Inc.) was used for transfection into LentiX-293T cells. The supernatants of the transfected cells were collected 72 hours later and the cell debris was cleared by centrifugation at 3,750 g for 5 minutes. The clear supernatants were filtered through a 0.45-μm filter and then used to infect CHO-K1 cells with RetroNectin (Recombinant Human Fibronectin Fragment; Takara Bio Inc.). For the selection of CHO-K1 cells having pLVSIN CDH6-eGFP, the infected cells were cultured in appropriate medium supplemented with 10 μg/mL puromycin.

A DNA fragment encoding the extracellular domain (Met1–Ala615) of human CDH6 (NP_004923.1) as soluble-form CDH6 (sCDH6) fused with a polyhistidine tag at the C-terminus was chemically synthesized using the Gene Art Synthesis service by Thermo Fisher Scientific Inc. After the protein-coding DNA had been amplified by PCR (forward primer: 5′-CCCCTTCACC GCTAGGCAGG CTTCGCCACC ATGC-3′ and reverse primer: 5′-AAACTCATTA GCTAGGCTGG GTCTTAGTGA TGGTGGTGG-3′), the amplified fragment was inserted into the Nhe1 site of pLenti6 (puro) vectors (pLenti6-sCDH6-His) using In-Fusion HD cloning kit. Lentiviruses were prepared using 2nd Generation Packaging Systems Mix (Applied Biological Materials Inc.), in accordance with the manufacturer's instructions. Lipofectamine 2000 was used for transfection into LentiX-293T cells. The supernatants of the transfected cells were collected 72 hours later and the cell debris was cleared by Lenti-X Concentrator (Takara Bio Inc.). The clear supernatants were used to infect PA-1 and JHOC-5 cells with polybrene. For the selection of the cells having pLenti6-sCDH6-His, the infected cells were cultured in an appropriate medium supplemented with 0.5 and 1 μg/mL puromycin for PA-1 and JHOC-5, respectively.

The release rate of DXd from 100 μg/mL R-DXd in mouse, rat, monkey, and human plasma and buffer containing 1% BSA was evaluated at 37°C for up to 21 days. Plasma and buffer samples were deproteinized with ethanol and then analyzed by LC-MS/MS (LC, Shimadzu; MS, AB SCIEX). The release rate of DXd was calculated using the mean concentration of DXd (N = 3).

The CDH recombinant proteins used in this study, containing a His-tag and/or human Fc-tag, are described below. Human Cadherin-6/CDH6 Protein (His-tag; hCDH6-his), Cynomolgus Cadherin-6/CDH6 Protein (His-tag; cCDH6-his), and Rat Cadherin-6/CDH6 Protein (His-tag; rCDH6-his) were purchased from Sino Biological Inc. Recombinant Mouse Cadherin-6/KCAD (mCDH6-his), Recombinant Human Cadherin-6/KCAD Fc Chimera (hCDH6-his-Fc), Recombinant Human Cadherin-7 (hCDH7-his), Recombinant Human Cadherin-9 Fc Chimera (hCDH9-Fc), Recombinant Human Pro-Cadherin-12 Fc Chimera (hCDH12-Fc), and Recombinant Human Cadherin-20 (hCDH20-his) were purchased from R&D Systems, Inc. Recombinant Human CDH10 Protein, Fc-tagged (hCDH10-Fc), was purchased from Creative BioMart. Recombinant Human CDH11 (hCDH11-his) was purchased from Novoprotein. Recombinant Human CDH18 Protein (hCDH18-his) was purchased from Abcam plc. Flat-bottomed 96-well immunoplates were coated with 1 or 2 μg/mL CDH recombinant proteins overnight at 4°C, washed, blocked with 1% BSA, and then incubated with R-DXd or Control ADC for 1 hour at room temperature. After washing with D-PBS(⁺; Thermo Fisher Scientific K.K.) supplemented with 0.05% Tween 20, horseradish peroxidase (HRP)-conjugated anti-human IgG F(ab′)₂ fragment-specific antibody (Jackson ImmunoResearch Laboratories, Inc.) was added and incubated for another hour. In the same manner, as described above, anti-His and anti-human IgG Fc fragment-specific antibodies were also used for His-tag and Fc-tag detection, respectively. Following washing and the addition of 3,3′,5,5′-tetramethylbenzidine soluble reagent (ScyTek Laboratories, Inc.), the absorbance at 450 nm was measured using a microplate reader, Envision 2105 (PerkinElmer Japan Co., Ltd.).

Cells were seeded in 96-well clear-bottomed white plates at 2,000 cells/well. After overnight culture, each diluted test substance was added. After treatment for 6 days, cell viability was measured by the CellTiter-Glo 2.0 assay (Promega K.K.), using ARVO X3 multilabel counter (PerkinElmer Japan Co., Ltd.).

Cells were seeded in 96-well Transwell plates (upper chamber, Corning Inc.) and clear-bottomed white plates (lower chamber, Corning Inc.). The lower chamber was seeded with 1,000 cells/well and the upper chamber with 8,000 cells/well. The day after seeding, 10 ng/mL R-DXd or 10 nmol/L DXd was added. After treatment for 6, 8, or 10 days, the upper chamber was removed and the viability of cells in the lower chamber was measured by the CellTiter-Glo 2.0 assay, using Envision 2105.

Cells were lysed in RIPA buffer supplemented with protease and phosphatase inhibitors (Thermo Fisher Scientific Inc.). The lysates were collected after centrifugation at 15,000 rpm for 10 minutes at 4°C, and then their protein concentration was quantified using BCA Protein Assay (Thermo Fisher Scientific Inc.). Western blotting for cleaved caspase-3 (cCASP3), Chk1 phosphorylated at Ser317 (pChk1), total Chk1, and β-actin was performed using the Simple Western system (Protein Simple), in accordance with the manufacturer's instructions. Briefly, four volumes of lysate was mixed with one volume of fluorescent 5× Master Mix containing 200 mmol/L dithiothreitol (Protein Simple) and denatured at 95°C for 5 minutes. Primary antibodies against cCASP3, pChk1, total Chk1, and β-actin were diluted in antibody diluent 2 (Protein Simple). The prepared samples, the biotinylated ladder, the primary antibodies, the secondary antibodies (supplied by the manufacturer), the chemiluminescent substrate, and wash buffer were dispensed into designated wells in an assay plate. The prepared assay plate was placed into the Simple Western system (Wes; Protein Simple). All subsequent separation, immunodetection, and analysis steps were performed automatically by the system. Primary antibodies for cCASP3, pChk1, and β-actin were purchased from Cell Signaling Technology, Inc., and that for total Chk1 was from Santa Cruz Biotechnology, Inc.

NIH:OVCAR-3 cells were seeded in 96-well clear-bottomed black plates at 8 × 10⁴ cells/well. After overnight culture, cells were treated with 10 nmol/L Alexa Fluor 488-conjugated R-DXd (R-DXd-Alexa488) for 15 minutes at 4°C. After washing with D-PBS(+), FITC fluorescence intensity was measured as “F₀” using Envision 2105. After measurement, samples were incubated for 0, 0.5, 1, 3, 6, 12, 24, and 48 hours at 37°C. After incubation, FITC fluorescence intensity was measured as “Fx.” To detect internalized R-DXd-Alexa488, cell-surface-bound R-DXd-Alexa488 was quenched. Then, 10 μg/mL anti-Alexa 488 antibody was added and incubated on ice for 30 minutes. After incubation, FITC fluorescence intensity was measured as “Qx.” Quench rate (QR) was calculated as follows: (F₀ – Q₀)/F₀. Internalization rate was calculated as follows: [Fx − (Fx − Qx)/QR]/F₀.

NIH:OVCAR-3 cells were seeded in 96-well clear-bottomed black plates at 8 × 10⁴ cells/well. After overnight culture, cells were treated with 10 nmol/L R-DXd for 15 minutes at 4°C. The “With wash prior to incubation” group was washed with D-PBS(+) to remove R-DXd from the supernatant, and the other group was not washed. Samples were incubated for 0, 0.5, 1, 3, 6, 12, 24, and 48 hours at 37°C. After incubation, cells were stained with Alexa Fluor 488-conjugated anti-human CDH6 antibody (R&D Systems) on ice for 30 minutes in the dark. FITC fluorescence intensity was measured using Envision 2105.

NIH:OVCAR-3 cells were seeded in a 96-well plate (CellCarrier-96 Ultra, PerkinElmer) at 2 × 10⁴ cells/well. After overnight culture, cells were treated with Hoechst 33342 (100 ng/mL) for 30 minutes at 37°C, and then treated with pHrodo-labeled R-DXd or Control ADC (1.5 μg/mL). After 24 hours of treatment, fluorescence images were obtained using a 403 confocal Opera Phenix high-content screening system (PerkinElmer).

Enhanced GFP-fusion human CDH6 (CDH6-eGFP)-expressing CHO-K1 cells were seeded in a 96-well plate (CellCarrier-96 Ultra) at 5,000 cells/well. After 3 days of culture, cells were treated with Hoechst 33342 (1 μg/mL) and LysoTracker Deep Red (2 nmol/L) for 30 minutes at 37°C, and then treated with Alexa Fluor 594-conjugated R-DXd (10 μg/mL) for 15 minutes at 4°C. After the medium had been replaced with fresh culture medium, time-lapse fluorescence images were obtained for up to 24 hours, using Opera Phenix.

Three to six mice were housed together in sterilized cages and maintained under specific pathogen-free conditions. The mice were euthanized with CO₂ gas when they reached the endpoints (tumor volume exceeding 3,000 mm³, 25% reduction of body weight, or clinical signs, indicating that they should be euthanized for ethical reasons). Female nude mice (CAnN.Cg-Foxn1^nu/CrlCrlj), ages 4 to 5 weeks old, were purchased from Charles River Laboratories Japan, Inc., for cell line-derived xenograft (CDX) studies; and female nude mice (Hsd: Athymic Nude-Foxn1^nu), ages 4 to 8 weeks old, were purchased from Envigo for patient-derived xenograft (PDX) studies.

CDX studies: All models were established by subcutaneously injecting a cell suspension into female mice. NIH:OVCAR-3, 786-O, PA-1, soluble CDH6 (sCDH6)-PA-1, JHOC-5, and sCDH6-JHOC-5 models were established by injecting 4 × 10⁶ (786-O), 5 × 10⁶ (PA-1, sCDH6-PA-1, and sCDH6-JHOC-5), or 1 × 10⁷ cells/head (NIH:OVCAR-3 and JHOC-5) in Matrigel. The ES-2 model was established by injecting 1 × 10⁶ cells/head in saline.

PDX studies: CTG-0258/0703/0711(OVC) and CTG-1366/1370(RCC) studies were outsourced to Champions Oncology, Inc. Models were established by the subcutaneous inoculation of tumor fragments derived from OVC and RCC, which were maintained in host mice.

Animals were randomized into treatment and vehicle control groups, and dosing was started (day 0) when the tumor volume reached approximately 100–300 mm³, except in sCDH6-related experiments. In the sCDH6-related experiments, it was started when the tumor size reached approximately 500 mm³ in sCDH6-PA-1- and parental PA-1-bearing mice, and approximately 700 mm³ in sCDH6-JHOC-5- and parental JHOC-5-bearing mice, when plasma sCDH6 was detected. Tumor volume defined as 1/2 × length × width² (CDX models) or 0.52 × length × width² (PDX models) was measured twice a week. Body weight was also measured at the same time. ABS buffer [10 mmol/L acetate buffer (pH 5.5) containing 5% sorbitol] or formulation buffer [10 mmol/L histidine buffer (pH 5.4) containing 0.03% polysorbate 80 and 9% sucrose] was used for the vehicle control group. All animal experiments performed in this study were approved by the Institutional Animal Care and Use Committee at Daiichi Sankyo Co., Ltd.

Blood was collected from the jugular vein of mice for sCDH6 measurement. The collected blood was placed in a tube containing lithium heparin (Becton, Dickinson and Company) and centrifuged to collect plasma samples.

Excised xenograft tumors were formalin-fixed and paraffin-embedded, and 4-μm tissue sections were used for IHC. IHC staining was carried out using the Ventana Benchmark ULTRA (Roche Diagnostics K.K.) staining platform. All of the primary antibodies (Rabbit Anti-CDH6 Monoclonal Antibody S129 and Rabbit Monoclonal Antibody IgG XP Isotype Control DA1E) were diluted with Ventana Antibody Diluent. Antigen retrieval was conducted with ULTRA CC1. Slides were incubated with the primary antibody diluted to 4 μg/mL with antibody diluent. Primary antibody was detected using the OptiView DAB IHC Detection Kit. Enzymatic detection of the primary antibody was accomplished with anti-rabbit IgG conjugated to HQ (DISCOVERY anti-Rabbit HQ), followed by anti-HQ-labeled HRP. Chromogen was deposited by a reaction with hydrogen peroxide in the presence of DAB and copper sulfate. The anti-Rabbit HQ, HRP multimer, and all chromogen reagents were applied at the instrument's default times. BenchMark Ultra LCS was added as a coverslip as appropriate during the staining process. Dehydration, permeation, and coverslipping were conducted using Tissue-Tek Prisma Plus and Tissue-Tek Glas g2 (Sakura Finetek Japan Co., Ltd.). The quality of CDH6 IHC staining was judged with a quality control slide for CDH6 IHC. Whole-slide images were scanned by NanoZoomer-XR (Hamamatsu Photonics K.K.) and used for image analysis. Membranous CDH6 staining intensities in tumor cells were scored with the Membrane IHC module (version 1.4) on HALO image analysis software (version 2.0, Indica Labs). The staining intensity of each tumor cell was classified as negative or positive with three levels of intensity (strong, moderate, and weak). Based on the percentage of tumor cells at each staining intensity level, an H score was calculated using the following formula:

H score = [3 × (percentage of strongly positive tumor cells)] + [2 × (percentage of moderately positive tumor cells)] + [1 × (percentage of weakly positive tumor cells)].

R-DXd dissolved in vehicle (formulation buffer) was intravenously administered to female 786-O xenografted nude mice (N = 3) at doses of 0.25, 0.5, 1, 2, 4, and 8 mg/kg. Plasma samples were collected 1, 6, 24, 72, 168, 336, and 504 hours later. The plasma concentrations of R-DXd were measured using Gyrolab xP workstation (Gyros Protein Technologies AB) to perform ligand-binding assay using biotinylated anti-DXd IgG mAb as a capture reagent and Alexa Fluor 647-labeled anti-R-DXd idiotype IgG mAb as a detection reagent. The lower limit of quantitation (LLOQ) was 0.01 μg/mL. PK parameters were calculated based on a noncompartmental analysis technique using WinNonlin 6.4 (Certara).

R-DXd dissolved in vehicle (formulation buffer) was intravenously administered to male cynomolgus monkeys (N = 3) at doses of 0.3, 1, 3, and 10 mg/kg. Plasma samples were collected 0.083, 1, 7, 24, 72, 168, 336, 504, and 672 hours later. The plasma concentrations of R-DXd were measured in the same way as for the plasma samples from mice, and those of total Ab were quantified by Gyrolab xP workstation using Goat Anti-Human IgG, Monkey ads-BIOT (Southern Biotech) as a capture reagent, and Goat Anti-Human IgG, Monkey ads-AF647 (Southern Biotech) as a detection reagent. The LLOQ was 0.2 μg/mL. PK parameters were calculated based on a noncompartmental analysis technique using WinNonlin 6.4.

R-DXd was intravenously administered at 0 (vehicle control), 10, and 30 mg/kg at 3-week intervals over a 6-week period to cynomolgus monkeys (three animals of each sex/dose), up to a total of three doses. Necropsy was performed on the day after the last administration. Additional animals (two animals of each sex/dose) treated at 10 and 30 mg/kg were used to assess the reversibility after a 9-week recovery period. The 80 mg/kg group consisted of five animals of each sex. The animals in the 80 mg/kg group were administered once only and necropsied at the end of the first cycle of dosing (21 days after dosing) because one male was found dead on day 7 (6 days after the first administration). Clinical signs, body weight, food consumption, and clinical pathology were monitored throughout the study.

A 96-well plate was coated with 1 μg/mL anti-human CDH6 antibody (R&D Systems) overnight at 4°C, blocked with 1% BSA, and then incubated with human CDH6 recombinant protein (Sino Biological Inc.) or samples for 2 hours at room temperature. After washing with D-PBS(+) supplemented with 0.05% Tween 20, 0.1 μg/mL rat anti-CDH6 antibody clone G019 (parental antibody of R-DXd, Daiichi Sankyo) or Naked CDH6 Ab was added and incubated for 1 hour. After washing again, HRP-conjugated mouse monoclonal [KT98] anti-Rat IgG2b H&L (Abcam plc.) or Peroxidase AffiniPure Goat Anti-Human IgG, F(ab')₂ fragment-specific (Jackson ImmunoResearch Laboratories Inc.) was added and incubated for another hour. Following further washing and the addition of 3,3′,5,5′-tetramethylbenzidine soluble reagent, the absorbance at 450 nm was measured using Envision 2105. Heparin-treated plasma from healthy controls and from cancer (OVC, RCC) patients was purchased from Tissue Solutions Limited.

The detached NIH:OVCAR-3, PA-1, and ES-2 cells were stained with PE-conjugated anti-human CDH6 antibody (R&D Systems) or PE-conjugated mouse IgG1 (R&D Systems) on ice for 30 minutes in the dark. Washed cells were analyzed using the flow cytometer BD LSRFortessa X-20 (Becton, Dickinson and Company).

NIH:OVCAR-3 cells and KO-OVCAR-3 cells were subjected to quantitative analysis of cell-surface antigen using QIFI kit (Dako) and flow cytometry. Cells were stained with anti-human CDH6 antibody (R&D Systems) or isotype control IgG (R&D Systems) and FITC-labeled anti-mouse IgG (Dako). Cells were analyzed using a flow cytometer (CytoFLEX S, Beckman Coulter).

Antibody-dependent cellular cytotoxicity (ADCC) activities were evaluated using human or cynomolgus monkey peripheral blood mononuclear cells derived from a donor as effector cells and NIH:OVCAR-3 cells as target cells. The effector cells (2 × 10⁵ cells) and the calcein-labeled target cells (1 × 10⁴ cells) were incubated with each substance, and the indicated effector:target (E:T) ratio was 20:1. After 4 hours of incubation, ADCC activity was measured by determining fluorescence intensity in the culture supernatant. Trastuzumab (Chugai Pharmaceutical Co., Ltd.) and Ultra-LEAF Purified Human IgG1 Isotype Control Recombinant Antibody (hIgG1, BioLegend Inc.) were used as positive and negative controls, respectively.

Complement-dependent cytotoxicity (CDC) activities were evaluated using human complement serum (Sigma-Aldrich Co. LLC). NIH:OVCAR-3 cells or Ramos cells were incubated with each substance and human complement serum. After 1 hour of incubation, CDC activity was measured by detecting ATP using a CellTiter-Glo 2.0 Assay. Rituximab (Zenyaku Kogyo Co., Ltd.) was used as a positive control.

All statistical analyses except for in the studies using monkeys were performed using SAS System Release 9.2 (SAS Institute Inc.). Statistical analyses in the studies using monkeys were performed using the MiTOX System (Mitsui Zosen Systems Research Inc.). All EC₅₀ values were determined by a sigmoid E_max model. Dunnett multiple comparison tests were used to compare the vehicle control group and the treatment groups in cell-line-derived xenograft model studies, and an unpaired t test was used to compare the vehicle group and the treatment group in PDX model studies.

The data generated in this study are available upon request from the corresponding author.

Anti-human CDH6 mAb (Naked CDH6 Ab) was generated by a DNA immunization method in rats. It had a humanized design to reduce immunogenicity. R-DXd was bound to Naked CDH6 Ab and TOP1 inhibitor DXd via a GGFG tetrapeptide-based cleavable linker by cysteine conjugation (Fig. 1A). Hydrophobic interaction chromatography showed a homogeneous drug distribution and the DAR of R-DXd was observed to be approximately 8 (Fig. 1B). The release rates of DXd from R-DXd ranged from 1.8% to 5.5% on day 21 in mouse, rat, monkey, and human plasma (Fig.1C), which were comparable to or rather lower than those of other ADCs, such as T-DXd, T-DM1 (trastuzumab emtansine), and SGN-35 (brentuximab vedotin; refs. 26, 36, 37). R-DXd showed binding activity to human and cynomolgus monkey CDH6, bound weakly to mouse CDH6, but not to rat CDH6 (Supplementary Fig. S1A). Defined as the Kd values, the EC₅₀ values of R-DXd were as follows: human CDH6, 64.7 ng/mL; cynomolgus monkey CDH6, 70.4 ng/mL; and mouse CDH6, 228 ng/mL. Among human CDH family proteins, R-DXd bound to human CDH6; however, no binding was detected to other human CDH family proteins, including CDH9 and CDH10, the human proteins most closely related to CDH6 (Fig. 1D). The immobilization of each CDH protein was confirmed to be appropriate (Supplementary Fig. S1B and S1C). Furthermore, the in vitro cell growth-inhibitory activity of R-DXd was evaluated using the CDH6-positive human cancer NIH:OVCAR-3 and PA-1 cells, and the CDH6-negative human cancer ES-2 cells (Supplementary Fig. S2A). R-DXd showed cell growth-inhibitory activity against CDH6-positive NIH:OVCAR-3 and PA-1 cells; however, it did not exhibit such inhibitory activity against CDH6-negative ES-2 cells (Fig. 1E). Naked CDH6 Ab and Control ADC did not show cell growth-inhibitory activity against any of the cells. All of the cell lines showed sensitivity to DXd (Supplementary Fig. S2B). The amount of DXd in the culture supernatant 24 hours after the addition of R-DXd and Control ADC was measured, and the DXd was detected in an antigen expression-dependent manner (Supplementary Table S1). To further explore the CDH6 dependence, we established a CDH6-knockout NIH:OVCAR-3 cell line (KO-OVCAR-3; Supplementary Fig. S3A). We confirmed that the DXd sensitivity of KO-OVCAR-3 cells did not differ significantly from that of the parental lines (Supplementary Fig. S3B); R-DXd activity was weakened in KO-OVCAR-3 (Fig. 1F). These findings suggest that the cytotoxic efficacy of R-DXd is dependent on the CDH6 expression.

The bystander antitumor effect of DXd-ADC was reported (27), which involves the killing of neighboring cells by DXd passing through the cell membrane. To evaluate the bystander antitumor effect of R-DXd, the Transwell assay was performed using KO-OVCAR-3 and CDH6-expressing parental NIH:OVCAR-3 cell line (Fig. 1G). The cell survival in the lower well was measured at each time point after drug treatment. When CDH6-positive cells were placed in the upper well and CDH6-negative cells in the lower well, cell death after ADC treatment was observed with a delay of several days compared with when CDH6-positive cells were placed in the lower well. On the other hand, cell death occurred in the same manner regardless of the cell combination after the addition of DXd (Supplementary Fig. S3C). These results confirmed that, after R-DXd had been taken up by the CDH6-positive cells in the upper chamber, the payload attacked the lower CDH6-lacking cells by the in vitro bystander effect. TOP1 inhibitors are known to induce double-strand DNA breaks and apoptosis in tumor cells. Therefore, we evaluated DNA damage and apoptosis induced by treatment with R-DXd. The results showed that R-DXd and DXd induced the phosphorylation of Chk1, a DNA damage marker, and the cleavage of caspase-3, an apoptosis marker (Fig. 1H). Other mechanisms that might contribute to the antitumor activity of antibody-based therapy are ADCC and CDC. However, R-DXd exhibited neither ADCC nor CDC activity (Supplementary Fig. S4). These results indicate that R-DXd binds to CDH6 and induces apoptosis with the bystander effect of DXd.

An ADC is supposed to be internalized into cancer cells with the cell-surface target antigen and then be cleaved by lysosomal enzymes and release its payload. Once R-DXd bound to the cell surface, its internalization rate increased in a time-dependent manner and reached almost 100% 24 hours after treatment, suggesting that R-DXd has high internalization activity (Fig. 2A). Lysosomal trafficking of R-DXd was confirmed using ADC labeled with a pH-sensitive dye, pHrodo. After 24 hours of treatment, R-DXd was localized to the lysosome, which was traced by monitoring pHrodo probe label-derived orange dots, compared with the findings for Control ADC (Fig. 2B). Furthermore, not only the internalization rate of the ADC but also the recovery rate of the cell-surface antigen is important for the effectiveness of the ADC. The CDH6 expression on the cell surface decreased within a few hours after R-DXd treatment (Fig. 2C). This decreased expression of CDH6 was maintained at 48 hours after R-DXd treatment when R-DXd was not removed from the medium. On the other hand, when R-DXd was washed out from the medium after R-DXd binding, the amount of cell-surface CDH6 transiently decreased but recovered to around 80% within 24 hours (Fig. 2C). Combined with the results of Fig. 2A, these results indicate that the cell-surface expression of CDH6 was restored even during the internalization by R-DXd. These results show the high intrinsic potential of R-DXd and the rapid resilience of CDH6 membrane expression.

We next performed fluorescence live-cell imaging using enhanced GFP-fused CDH6-expressing cells. CDH6-eGFP was localized at the cell–cell boundary before R-DXd treatment (Fig. 2D). Alexa Fluor 594-conjugated R-DXd was colocalized with CDH6-eGFP just after the treatment. Subsequently, when the area of CDH6 was decreased, R-DXd was localized in a dot-like pattern (Fig. 2D and E). The dots of R-DXd were confirmed to localize at lysosomes (Fig. 2F). These results indicate that R-DXd could bind to CDH6 localized at the cell–cell boundary, followed by internalization.

To confirm that the antitumor activity of R-DXd depended on the expression level of CDH6, we next compared the antitumor activity of R-DXd with those in the control ADC, naked CDH6 Ab, and vehicle-treated groups, using xenografts from cells with a range of CDH6 expression levels, including ES-2 (CDH6-negative), 786-O (CDH6-positive), and NIH:OVCAR-3 (CDH6-positive; Fig. 3A). Tumor growth was significantly suppressed in the R-DXd-treated group for 786-O (P < 0.001) and NIH:OVCAR-3 (P < 0.001; Fig. 3B). In contrast, tumor growth did not differ significantly between the R-DXd group and the vehicle group for the ES-2 model, and the Naked CDH6 Ab did not show any antitumor effect as well. R-DXd also showed dose-dependent antitumor activity (Fig. 3C). These results reveal that the antitumor activity of R-DXd was dependent on the CDH6 expression level and that the payload-derived cytotoxic activity is the main contributor to this activity. R-DXd showed a favorable PK profile with low distribution volume (V_ss: 55.6 to 84.6 mL/kg) and clearance (CL: 25.0 to 50.7 mL/day/kg) across doses in mice (Fig. 3D; Supplementary Table S2A). The systemic exposure level (AUC_0–21d) increased in a dose-dependent manner.

To evaluate the efficacy of R-DXd against tumors subjected to intensive carboplatin and paclitaxel treatment, which is the standard-of-care for OVC in the first-line setting, we prepared an OVC xenograft model undergoing long-term treatment with the combination of carboplatin and paclitaxel. When the average tumor volume reached approximately 250 mm³, NIH:OVCAR-3 xenograft was repeatedly dosed with carboplatin and paclitaxel in this study. In the first dosing, the tumor regressed and then regrowth occurred within 6 weeks in all cases. However, with repeated administration, tumor regression was attenuated. After the ninth dosing, five mice, whose tumor volumes were within the range of 150 mm³ to 500 mm³, were assigned to R-DXd treatment. R-DXd also showed potent antitumor activity in these mice (Fig. 3E). There was no serious body weight loss in the experiment (Fig. 3F). The results suggested that R-DXd is also effective against tumors with decreased sensitivity to the combination therapy of carboplatin and paclitaxel.

To examine the therapeutic potential of R-DXd in more clinically relevant models, we administered it at a dose of 10 mg/kg every 3 weeks to human OVC and RCC PDX models. CDH6 expression was measured by IHC (Fig. 4A). R-DXd showed significant tumor growth-inhibitory activity against CDH6-positive PDX models (all P < 0.01), whereas it did not show a significant antitumor effect against CTG-0258, a CDH6-negative PDX model (Fig. 4B). These results suggest that R-DXd has therapeutic potential against various CDH6-expressing solid tumors.

The pharmacokinetic profile of R-DXd resembled that of the total Ab across doses in monkeys (Supplementary Fig. S5A and S5B; Supplementary Table S2B). This suggests that the drug linker was stable in plasma, which was similarly observed for T-DXd in a previous study (26). The CL value of R-DXd ranged from 8.62 mL/day/kg to 15.8 mL/day/kg. The V_ss value of R-DXd ranged from 47.7 mL/kg to 84.9 mL/kg, which was similar to the plasma volume seen in cynomolgus monkeys. The systemic exposure level (AUC_0–28d) increased in a dose-dependent manner.

The safety profile of R-DXd was characterized in intermittent intravenous dosing studies in cynomolgus monkeys, which is a species in which R-DXd cross-reactivity occurs. R-DXd was administered once every 3 weeks for up to 6 weeks (three times in total), followed by a 9-week recovery period at 10 and 30 mg/kg, and only once at 80 mg/kg. Target organs of toxicity are summarized in Table 1. Major toxicity was found in the intestines, kidneys, liver, lungs, bone marrow, and skin. Notably, a decrease in platelet count was also observed. As for liver toxicities, increases in aspartate aminotransferase (AST) and alanine aminotransferase (ALT) were noted in monkeys dosed with 30 mg/kg or above. At 80 mg/kg, slight degenerative/necrotic changes in the liver were also observed. Lung lesions observed at 30 mg/kg or above included slight interstitial inflammation and alveolar edema. In the kidney, minimal tubular injuries were observed at 30 mg/kg or above. As for the skin, abnormal skin color was observed at 30 mg/kg or above, and inflammatory changes with regenerative responses and pigmentation in the epidermis were observed at 80 mg/kg. In animals given 10 or 30 mg/kg, all of these findings were minimal or mild. After the 9-week recovery period, slight tubular injuries in the kidney and epidermal pigmentation in the skin were observed. One male died at the highest dose of 80 mg/kg on day 7 (6 days after the first administration). In this animal, vomiting, soft stool, diarrhea, and decreased food consumption were also noted. Histopathologically, lesions in the gastrointestinal epithelia or mucosa were observed as unique changes for this animal in addition to the above changes. Therefore, the cause of death was assumed to be a deteriorated physical condition resulting from diarrhea, soft stool, decreased food consumption, and cellular injury of the intestine. The other animals in the 80 mg/kg group were dosed once and then necropsied on day 22.

On the basis of the toxicological findings, the highest non-severely toxic dose (HNSTD) for monkeys was considered to be 30 mg/kg (Table 1), corresponding to the human equivalent dose (HED) of 9.6 mg/kg (38).

A previous study reported the presence of sCDH6 in the blood of OVC patients (39). To confirm that sCDH6 is present in human blood in practice, we measured heparin-treated plasma samples from five patients each with ovarian or renal cancer and detected sCDH6 at 0.26 and 0.38 nmol/L, respectively (Fig. 5A). Next, to determine whether sCDH6 inhibits the efficacy of R-DXd, we established cell lines in which sCDH6 was exogenously expressed in PA-1 (sCDH6-PA-1) and JHOC-5 (sCDH6-JHOC-5). These two parental cell lines are CDH6-positive and sensitive to R-DXd in vivo (Fig. 5B, dotted lines), while having lower levels of sCDH6 than sCDH6-overexpressing cells (Supplementary Table S3). sCDH6-overexpressing cells produced a high level of sCDH6 and the amounts of sCDH6 in the plasma of sCDH6-PA-1 and sCDH6-JHOC-5 were 0.94 and 1.00 nmol/L, respectively (Fig. 5C). Therefore, sCDH6 levels in these models are higher than that in the patient samples. Meanwhile, R-DXd showed antitumor efficacy irrespective of whether sCDH6 was present in the plasma (Fig. 5B, solid red lines). These findings indicate that the antitumor efficacy of R-DXd is not affected by sCDH6.

Ovarian cancer ranks fifth as a cause of cancer-related death among women, accounting for more deaths than any other cancer of the female reproductive system. Epithelial tumors account for almost 90% of ovarian cancers and are subclassified into the histologic subtypes of serous (75%), endometrioid (10%), clear cell (10%), and mucinous (3%; ref. 40). Renal cancer is also one of the 10 most common tumors, with 403,000 new cases diagnosed worldwide and 175,000 deaths in 2018 (41). It has been reported that human CDH6 is highly expressed in serous-type OVC and renal cancer, but expressed at a low level in normal tissues (6–8). Therefore, CDH6 has been considered as an attractive target for treating these cancers. However, no CDH6-targeting therapeutic agents have been approved to date. A phase I study of HKT288, a CDH6-targeting DM4-ADC (42), was terminated due to unexpected neurologic adverse events (43).

Here, we report on R-DXd, a novel CDH6-targeting ADC with the membrane-permeable and potent TOP1 inhibitor payload DXd. R-DXd is composed of the same linker-payload technology as T-DXd, an approved HER2 ADC. In vitro evaluation of the stability of R-DXd showed that the payload release in plasma was very low, similar to that of T-DXd (26). Furthermore, in toxicity studies in monkeys, the HNSTD was determined to be 30 mg/kg, which is the same as for T-DXd (26).

Several potential hypotheses suggested that R-DXd may be able to overcome toxicities observed in HKT288 clinical development. First, T-DXd, an ADC with the same payload and linker as R-DXd has demonstrated significant clinically meaningful efficacy in phase III DESTINY-Breast03 study, without neurotoxicity (25). In addition, many clinical studies of ADCs with DXd-ADC technology are ongoing and such neurotoxicity has not been reported. On the other hand, the safety profile of mirvetuximab soravtansine, which shares the same linker/payload as HKT288 with a different target (folate receptor alpha), has not observed neurologic adverse events with the exception of peripheral neuropathy (44). Therefore, it is unlikely that the linker/payload alone is solely responsible for the observed neurotoxicity in the HKT288 clinical trial. The sensitivity against ADC is supposed to be determined by the payload sensitivity of the cells and the amount of the payload delivered to the cells. Neurons, which have axons, are known to rely on tubulin for their activity, and it is supposed that they are highly sensitive to tubulin inhibitors such as maytansine-derived cytotoxic payload used in HKT288. Although it has been reported that CDH6 plays a role in CNS development, its expression in the adult nervous system has not been reported. However, if there is a possibility of minimal expression of CDH6 in the adult nervous system, there is a potential risk that a CDH6-directed tubulin inhibitor such as the form of HKT288 selectively induces neurotoxicity. In addition, HKT288 has a narrow therapeutic window. Although the preclinical mouse studies demonstrated efficacy at 2.5 mg/kg or 5 mg/kg, the HNSTD in monkeys was determined to be 2 mg/kg (43). Based on these nonclinical safety data, the phase I trial tested doses of 0.3 mg/kg and 0.75 mg/kg every 3 weeks, which were relatively low compared with the mouse model. Therefore, the narrow therapeutic window might have been one of the contributing factors for the termination of this trial, in addition to the unexpected neurotoxicity they reported.

The mode of action of ADC involves several steps: binding to the target, antibody internalization, intracellular trafficking, and payload release (14, 45). Cadherin family proteins form adherens junctions and are involved in cell adhesion (46). In addition to flow cytometry and protein ELISA, we confirmed the binding of R-DXd to CDH6 by live-cell imaging, which obtains findings that more closely resemble the biological environment. After binding to cells, R-DXd was internalized within a few hours and localized to lysosomes. The internalization rate of R-DXd reached over 50% at 3 hours and, surprisingly, almost all ADC molecules were taken up into cells within 24 hours. Because the internalization rate of trastuzumab is reported to be approximately 20% at 3 hours (47), it can be said that R-DXd has high internalization efficiency. On the other hand, Naked CDH6 Ab having the same binding and internalization ability as R-DXd did not show any antitumor effect, so it is considered that the effect on the tumor by the antibody part of R-DXd is limited. Furthermore, although the amount of CDH6 on the cell surface initially decreased after the R-DXd treatment, it then recovered within a few hours under the condition that the ADC was removed from the supernatant. These findings suggest that CDH6 is a promising target of ADC with high internalization efficiency and strong ability to restore membrane expression. We also confirmed the payload release from CDH6-positive cells treated with R-DXd. In addition, R-DXd showed a bystander antitumor effect like T-DXd, and can be expected to be effective against tumors with heterogeneous expression of CDH6 (27). It was reported that soluble-form CDH6 was detected in the blood of ovarian cancer patients (39). However, there was no obvious change in the efficacy of R-DXd for cell lines in which sCDH6 was forcibly expressed. This indicates that the level of ADC was more than enough for the amount of sCDH6 in the blood. The sCDH6 level in the plasma of these models is higher than the clinically confirmed level, which suggests that the effect of sCDH6 is limited even in cancer patients with a high level of sCDH6.

TOP1 inhibitor causes DNA cleavage and induces cell death through the activation of caspase-3 (48). DXd is a more potent TOP1 inhibitor than SN-38, an active metabolite of irinotecan (21). This study showed that R-DXd and DXd induce the activation of caspase-3 and cause CDH6 expression-dependent cell death in vitro. R-DXd exhibited potent antitumor effects against the high-CDH6 ovarian and renal CDX models. In these in vivo models, Control ADC also showed weak nonspecific antitumor effects, but they were much weaker than those by R-DXd at the same dose levels, indicating that R-DXd showed CDH6-dependent antitumor effects. Continued carboplatin and paclitaxel treatment, which is standard chemotherapy against OVC, for approximately 200 days gradually attenuated the efficacy, but R-DXd still showed tumor regression in the model after long-term combination treatment. This indicates that switching to treatment with R-DXd may be a valuable approach when there is a decrease in sensitivity to carboplatin and paclitaxel. Furthermore, we utilized PDX models as well as CDX mouse models to evaluate in vivo antitumor activity. R-DXd significantly inhibited tumor growth and even led to tumor regression in CDH6-positive models, but not in CDH6-negative ones. These results from in vitro and in vivo studies indicate that CDH6 expression is crucial for the antitumor activity of R-DXd.

R-DXd showed an acceptable safety profile in the toxicity study in cynomolgus monkeys, which is a species exhibiting cross-reactivity for R-DXd. The HNSTD in monkeys was suggested to be 30 mg/kg, because no severe toxicity was observed up to this dose. This value is nearly equivalent to those of T-DXd and other DXd-ADCs (26). The lung lesions of monkeys treated with R-DXd were generally similar to those in monkeys treated with T-DXd (26). One of the causes of death in a male monkey given 80 mg/kg R-DXd was assumed to be the toxicity to the gastrointestinal tract. On the basis of the Human Protein Atlas (https://www.proteinatlas.org), the expression of CDH6 in the gastrointestinal epithelia or mucosa is not evident, so GI injury is likely to be off-target toxicity. Toxicity was observed in the kidney among the CDH6-positive organs, although it was minimal and unclear whether the target-dependent distribution of R-DXd was involved in the lesion. These nonclinical toxicological findings up to 30 mg/kg suggest an acceptable safety profile of R-DXd, which supports its entry into human trials with a starting dose of 1.6 mg/kg (one-sixth of the HED of HNSTD in monkey).

The first-in-human phase I study of R-DXd in patients with RCC and OVC is ongoing (NCT04707248), and the interim data from the dose-escalation part showed acceptable tolerability with early signs of efficacy (49).

A. Takatsuka reports a patent for anti-CDH6 antibody and anti-CDH6 antibody–drug conjugate issued. K. Nakamura reports a patent for anti-CDH6 antibody and anti-CDH6 antibody–drug conjugate issued. No disclosures were reported by the other authors.

H. Suzuki: Conceptualization, resources, data curation, formal analysis, supervision, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. S. Nagase: Conceptualization, software, formal analysis, investigation, visualization, writing–review and editing. C. Saito: Data curation, validation, investigation, writing–review and editing. A. Takatsuka: Validation, investigation, writing–review and editing. M. Nagata: Data curation, formal analysis, supervision, validation, writing–review and editing. K. Honda: Data curation, validation, investigation, writing–review and editing. Y. Kaneda: Data curation, investigation, visualization, writing–review and editing. Y. Nishiya: Data curation, formal analysis, supervision, writing–review and editing. T. Honda: Software, formal analysis, validation, investigation, writing–review and editing. T. Ishizaka: Data curation, formal analysis, validation, investigation, writing–review and editing. K. Nakamura: Data curation, investigation, writing–review and editing. T. Nakada: Formal analysis, validation, investigation, writing–review and editing. Y. Abe: Supervision, writing–review and editing. T. Agatsuma: Resources, supervision, writing–review and editing.

The authors thank Hayato Tsuji, Naomi Kasanuki, Hirokazu Ishikawa, Satoshi Fujii, and Kimihisa Ichikawa (Daiichi Sankyo RD Novare Co., Ltd.) for the generation of KO-OVCAR-3, CDH6-eGFP-expressing CHO-K1, and sCDH6-expressing PA-1 and JHOC-5 cell lines. The authors also thank Yoshihiro Tani, Masanori Funabashi, Jun Harada, Riki Goto, Ichiro Hayakawa, Chigusa Yoshimura, Yoko Nakano, Takanori Aoki, Masato Amano, and Daigo Asano (Daiichi Sankyo Co., Ltd.) for their assistance. Anti-CDH6 Ab for IHC was provided by Roche Tissue Diagnostics. In vivo experiments in PDX models were performed by Champions Oncology, Inc.