Theranostic Targeting of CUB Domain–Containing Protein 1 (CDCP1) in Multiple Subtypes of Bladder Cancer

Bladder cancer is the second most common malignancy of the genitourinary tract, leading to over 17,000 deaths in the US each year (1, 2). Despite advances in imaging, chemotherapy, and surgery, the survival of patients who develop metastatic bladder cancer (mBC) remains poor. Treatments for mBC were historically limited to platinum-based chemotherapy, until the approval of the first checkpoint inhibitor in 2016 (3). However, anti–programmed cell death protein 1 therapy is only effective in about 25% of patients (4). More recently, two antibody–drug conjugates (ADC), enfortumab vedotin (which targets the surface protein NECTIN4) and sacituzumab govitecan (which targets the surface protein TROP2) received full and accelerated FDA approval, respectively, based on trials conducted in heavily pretreated patients (5–7). With the milestone FDA approvals, these therapies are now being tested in earlier disease settings.

Muscle-invasive bladder cancer (MIBC) is a molecularly diverse disease and has been classified into six molecular subtypes (8), and it is therefore reasonable to anticipate that not all subtypes may express sufficient TROP2 or NECTIN4 to be vulnerable to the cognate ADCs (estimated to be ∼20,000 receptors per cell; refs. 9, 10). Consistent with this hypothesis, we recently profiled the six subtypes (11) and showed that NECTIN4 is highly enriched in the luminal subtypes of MIBC, which is critical for drug response, with comparatively lower expression in the other subtypes (9). These data begin to suggest a molecular diversity to MIBC that will likely necessitate a large repertoire of therapeutics targeting other antigens expressed in NECTIN4 or TROP2 null tumors.

Since its cloning in 2001 (12), the cell surface protein CUB domain–containing protein 1 (CDCP1) has emerged as a promising target for cancer therapy. The single pass transmembrane protein has been shown to be overexpressed in numerous solid tumor and hematologic malignancies at levels well above the threshold for therapeutic modalities like ADCs and CAR T-cell therapy (i.e., ∼10⁶ receptors/cell; refs. 13, 14). Moreover, expression in normal human tissues is generally low, with primarily cytosolic protein expression reported in colon endothelial cells and cell surface expression on CD34⁺ stem/progenitor cells in bone marrow (15). This restricted tissue expression has led several groups, including our own, to develop antibody-based therapeutics targeting the ectodomain of CDCP1 (16, 17). ADCs, antibody coated liposomes, and radiolabeled antibodies have displayed encouraging antitumor activity in mouse models of pancreatic, ovarian, prostate, colorectal, and breast cancer with limited evidence of host toxicity (13, 14, 18–21).

Despite the growing interest in CDCP1, virtually nothing is known about its expression in mBC excepting two anecdotes. First, Cantley and colleagues reported CDCP1 upregulation in bladder cancer primary tumor biopsies compared with matched normal bladder tissue, though the source of the tumor tissues and their corresponding subtypes were not described (22). Second, Ji and colleagues proposed a model for bladder cancer initiation in which N6 methyladenosine modification on CDCP1 mRNA by methyltransferase-like 3 causes CDCP1 overexpression. This knowledge gap as well as the potential upside of discovering a new cell surface target to treat bladder cancer led us to investigate in this study if (i) CDCP1 is overexpressed in bladder cancer and in which subtypes, (ii) CDCP1 overexpression occurs in TROP2 and/or NECTIN4 null bladder cancer tumors, and (iii) CDCP1 can be exploited for targeted radiotherapy (TRT) to treat bladder cancer.

All materials and chemicals were purchased from commercial vendors and used without further processing/purification. HT-1376, HT-1197, 5637, TCCSUP, and 639V cells were obtained from the UCSF Cell Culture Facility. UMUC3, T24, 253JBV, and UMUC9 cells were gifts from Bradley Stohr (UCSF) and David McConkey (Pathology Core, Bladder Cancer SPORE, MD Anderson Cancer Center). Cellular identity was confirmed with short tandem repeat analysis. Bladder cancer lines were grown in standard MEM media (Life Technologies) supplemented with 10% FBS (Seradigm). Cellular identity was authenticated by visually inspecting morphology and probing for signature expression markers on immunoblot. Mycoplasma was tested after thawing cryostocks with the MycoAlert Kit (Lonza). Bladder cancer patient-derived xenografts (PDX) were generated at UCSF in collaboration with the Preclinical Therapeutics Core. Sotorasib and trametinib were purchased from MedChemExpress and used without further purification. The monoclonal antibody 4A06 was expressed and purified in the IgG1 format as previously described (18). p-SCN-Bn-Deferoxamine (B-705) and p-SCN-Bn-DOTA (B-205) were purchased from Macrocyclics (Plano, TX) and directly used for conjugation to antibody. ⁸⁹Zr-oxalate was obtained from 3D Imaging, LLC (Maumelle, AR). ¹⁷⁷LuCl₃ was obtained from Oak Ridge National Laboratory. Iodine-125 was obtained from Perkin Elmer.

Four retrospective publicly available cohorts of patients with MIBC were analyzed as previously described (9, 10). Briefly, the Sjödahl 2012 (n = 93) dataset [obtained from NCBI Gene Expression Omnibus (GEO), accession number GSE32894] had undergone batch effect correction, quantile normalization, log2-transformation, and gene centering by the authors (23). The Sjödahl 2017 (n = 243) dataset (obtained from GEO, accession number GSE83586) had been preprocessed including RMA normalization and gene centering by the authors (24). The Cancer Genome Atlas (TCGA; n = 406) dataset (25) was obtained from cBioPortal and had been normalized by RSEM and was further transformed by log₂(RSEM+1), and was renormalized to transcripts per million (TPM) when undergoing comparisons with healthy patient data (26, 27). The Seiler 2017 (n = 305) dataset (obtained from GEO, accession number GSE87304) had been SCAN-normalized by the authors (28). The IMvigor210 (n = 316 patients, 348 samples) dataset [9] was obtained from http://research-pub.gene.com/IMvigor210CoreBiologies/ and was TPM-normalized and further transformed by log₂(TPM+1). For Supplementary Fig. S4, cancer spread to the lymph nodes only was classified as locally advanced, while liver or visceral spread was classified as metastatic. These data consisted of de-identified and anonymized gene expression profiles and were deposited into the public domain. All cohorts underwent consensus molecular cluster subtyping as previously described (11). A healthy cohort from the GTEx Consortium was used to investigate normal tissue expression of CDCP1 (29). All further analyses were made in the R statistical environment v4.1.1.

A bladder cancer tissue microarray (TMA), which contains 80 formalin-fixed, paraffin-embedded (FFPE) biopsy specimens in duplicate, was obtained from the University of British Columbia (30). All studies involving human subjects were performed with approval from an Institutional Review Board, and all human studies were conducted in compliance with the Declaration of Helsinki. Written informed consent was obtained prior to collecting the biopsy samples. Sections of FFPE tissue were placed into the Ventana Discovery Ultra automated slide stainer. Antigen retrieval was performed using heat-inactivated antigen retrieval buffer (Tris-EDTA) according to the manufacturer instructions (Roche) and then stained with a rabbit polyclonal CDCP1 primary antibody (Cell Signaling Technology, #4115S, 1:50). Secondary antibodies (Anti-Rabbit HQ and HQ-HRP, from Ventana) were incubated for 12 minutes each, and DAB was used for detection for single stains. Slides were counterstained with hematoxylin per standard protocol. H-scores for CDCP1 staining were assigned by two independent pathologists (1+, 2+, and 3+ multiplied by the percentage), and the average of the H-scores was calculated.

Cells were trypsinized and washed, and then incubated with either a PE-anti-CDCP1 antibody (BioLegend, #324017, 1:100) or 4A06 for 30 minutes on ice. For cells incubated with unlabeled 4A06, after the 30 minutes incubation, the cells were washed with PBS multiple times and placed in a fluorophore conjugated secondary antibody solution for 30 minutes on ice (1:1,000, 109–546–097, Jackson ImmunoResearch). Cells were collected and washed before being placed in PBS and passed through a cell strainer. Cells were analyzed using an Attune NxT Flow and data were analyzed using FlowJo software.

Cell lines and PDX samples were lysed in Pierce RIPA lysis buffer (89900, Thermo Fisher Scientific) with protease and phosphatase inhibitors (1861281, Thermo Fisher Scientific). The samples were homogenized using a probe sonicator (Omni TH-01) and then centrifuged for 15 minutes at 15,000 × g. After calculating protein concentration using a Bradford Assay kit, 15 μg of lysate was resolved via electrophoresis using a 4% to 12% Bis-Tris gel (NW04120BOX, Invitrogen). Gels were transferred onto an Immobilon-P membrane (IPVH00010, Millipore). Membranes were blocked in 5% milk in TBST for 30 minutes at room temperature before being placed in a primary antibody solution. The primary antibodies used were anti-CDCP1 (4115, Cell Signaling Technology, 1:1,000), TROP2 (214488, Abcam, 1:1,000), NECTIN4 (17402S, Cell Signaling Technology, 1:1,000), and anti-beta actin (A5441, Sigma-Aldrich, 1:5,000), or GAPHDH (Cell Signaling Technology, #5174). The antibodies were incubated for 1 hour at room temperature or overnight at 4°C. Membranes were then washed with TBST and incubated with a secondary antibody solution for 30 minutes at room temperature. Secondary antibodies used were goat anti-rabbit (65–6120, Invitrogen, 1:5,000), goat anti-rat (62–6520, Invitrogen, 1:5,000), or HRP-anti-rabbit (Cell Signaling Technology, #7074). Proteins were detected using West Pico Chemiluminescent Substrate for 20 seconds (34578, Thermo Fisher Scientific) and then exposed to film (30–507, Blue Devil). Each immunoblot was reproduced at least once with freshly harvested protein samples.

¹²⁵I-4A06 was prepared using our previously reported protocol (14). The final yield of purified ¹²⁵I-4A06 was ∼ 60% (specific activity ∼ 1 μCi/μg) and purity was > 99%. UMUC3, TCCSUP, 5637, and UMUC9 cells (0.6×10⁶ cells/well) were seeded on 12-well plates using DMEM (10% FBS). The cells were washed with PBS for the saturation binding assay. Total binding of ¹²⁵I-4A06 was determined by adding it to cell suspensions at seven concentrations from 0.025 nmol/L to 10 nmol/L. Nonspecific binding was determined by adding 1,000x cold 4A06 to ¹²⁵I-4A06/cell mixtures at three concentrations. Cells were incubated with ∼ 0.5 μCi ¹²⁵I-4A06 at room temperature for 1 hour, washed with PBS, and lysed by adding 1.0 mol/L NaOH. The bound and unbound radioactive fractions were collected and measured on a Hidex Gamma counter (Turku, FI). Bmax was calculated using Prism v8.0.

All animal studies were conducted in compliance with Institutional Animal Care and Use Committee at UCSF. For tumor imaging or treatment studies with bladder cancer xenografts from cell line implants, 6- to 8-week-old intact male athymic nu/nu mice (Charles River) were used. Mice were inoculated subcutaneously (∼1.5 × 10⁶ cells) in the flank with a slurry of cells in 1:1 mixture (v/v) of media and Matrigel (Corning). Xenografts were generally palpable within 3 to 4 weeks after injection. 4A06 was functionalized with desferrioxamine and subsequently radiolabeled with Zr-89 as previously described (14). Tumor-bearing mice received ∼300 μCi of ⁸⁹Zr-4A06 in 100 μL saline solution volume intravenously using a custom mouse tail vein catheter with a 28-gauge needle and a 100 to 150 mm long polyethylene microtubing. 4A06 was functionalized with DOTA and radiolabeled with Lu-177 as previously described (14). Mice bearing subcutaneous UMUC3 tumors received ¹⁷⁷Lu-4A06 (400 μCi) or vehicle (saline) via tail vein at day 0 and day 7 of the study period. Mice bearing subcutaneous HT-1376 tumors received ¹⁷⁷Lu-4A06 (400 μCi) or vehicle (saline) via tail vein at day 0 and day 7 of the study period. Mice were arranged in treatment arms using a simple randomization approach (31). Animals were weighed at the time of injection, and 3 times per week until the completion of the study. Tumor volume measurements were calculated at the same time points with calipers. The study endpoints were death due to tumor volume > 2,000 mm³ or ≥ 20% loss in mouse body weight. The researcher performing the tumor volume and body weight measurements was blinded to the treatment arms.

Mice were imaged on a small animal PET/CT scanner (Inveon, Siemens Healthcare, Malvern, PA). Animals were typically scanned for 30 minutes for PET, and the CT acquisition was performed for 10 minutes. The co-registration between PET and CT images was obtained using the rigid transformation matrix generated prior to the imaging data acquisition because the geometry between PET and CT remained constant for each of PET/CT scans using the combined PET/CT scanner. For microPET/CT data, PET images were reconstructed using the ordered subsets expectation maximization algorithm (OSEM) provided by the scanner manufacturer. The parameters for OSEM were 16 subsets and 4 iterations, and the resulting reconstructed image volume was in a matrix of 128×128×159 with a voxel size of 0.0776 mm × 0.0775 mm × 0.0796 mm. CT images for attenuation correction were reconstructed using a conebeam Feldkamp algorithm provided by the scanner manufacturer. The data were acquired using x-ray tube voltage of 80 kVp and current of 0.5 mA for 120 angular steps over 220 degrees, and 175 ms of exposure at each angular step. The reconstructed CT volume was in a matrix of 512×512×700 with a voxel size of 0.195 mm × 0.195 mm × 0.195 mm. The precalibrated scaling was used to convert the CT images to attenuation maps for correction in PET reconstruction.

For SUV computation, we used freeware software, Amide (amide.sourceforge.net), and used its automated SUV calculation tool by entering decay-corrected injected activity and the animal weight. For each volume of interest, a spherical VOI (2–3 mm diameter) was drawn and SUV was calculated by VOI statistics.

At a dedicated time after radiotracer injection, animals were euthanized by cervical dislocation. Blood was harvested via cardiac puncture. Tissues were removed, weighed, and counted on a Hidex automatic gamma counter (Turku, Finland). The activity of the injected radiotracer was calculated and used to determine the total number of counts per minute by comparison with a standard of known activity. The data were background- and decay-corrected and expressed as the percentage of the injected dose/weight of the biospecimen in grams (%ID/g).

For the transcriptomic analysis, correlation was calculated by Spearman's rank correlation. ANOVA and Kruskal–Wallis tests were used to test for differences when there were more than two groups, and the Wilcoxon rank-sum test was used to test for differences between two groups unless otherwise stated. Binary comparisons between two treatment arms were made with an unpaired, two-tailed Student t test. Differences at the 95% confidence level (P < 0.05) were considered statistically significant. Unless otherwise stated, all data were expressed as mean ± SD. Kaplan–Meier survival analyses using log-rank testing were performed on quartiles of CDCP1 mRNA expression with overall survival as the endpoint.

All data are available upon request from the corresponding authors of this study.

To assess CDCP1 mRNA expression across the molecular subtypes of MIBC, we analyzed four cohorts of patients with localized MIBCs (23–25, 28). The clinical characteristics of these cohorts were previously described (9). Using the consensus classifier subtypes (11), we found that median CDCP1 expression was elevated in the basal/squamous subtype of MIBC relative to the luminal (LumP, LumNS, and LumU), stromal-rich and neuroendocrine (NE)-like subtypes (Fig. 1A–D; Supplementary Table S1). To determine whether CDCP1 is enriched in patients with bladder cancer, we also compared median expression of CDCP1 in healthy patients to expression of CDCP1 in a similarly normalized dataset of patients with MIBC (25). We validated our findings that CDCP1 protein was enriched in basal/squamous tumors by IHC using a previously described TMA (Fig. 1E and F; ref. 30). We found that CDCP1 mRNA expression and CDCP1 protein expression were broadly observed across clinical stages and pathologic subtypes. There was no enrichment of CDCP1 expression in any single subtype or clinical stage (Supplementary Figs. S1–S3). In addition, CDCP1 mRNA was mutually expressed in patients with locally advanced versus metastatic disease, and did not vary significantly between the primary bladder or metastatic biopsy site (Supplementary Figs. S4 and S5). Finally, no significant associations between quartiles of CDCP1 expression and overall survival were seen in cohorts for which retrospective clinical data were available (Supplementary Fig. S6).

We next evaluated CDCP1 protein expression in human bladder cancer cell lines representing luminal (5637, HT-1197, HT-1376, UMUC9) and basal subtypes (639V, T24, 253JBV, UMUC3, TCCSUP). Cell surface protein expression was observed in all cell lines by flow cytometry using 4A06, a monoclonal antibody we previously developed to target the ectodomain of CDCP1 (Fig. 2A; ref. 18). UMUC3 and TCCSUP had the highest expression of CDCP1, while UMUC9 and 253JBV had the lowest relative level. We further quantified cell surface receptor numbers using ¹²⁵I-labeled 4A06. Saturation binding assays showed that UMUC3 and TCCSUP had ∼1×10⁶ receptors per cell, while UMUC9, HT-1376 and 5637 had ∼1×10⁵ receptors per cell (Table 1; Supplementary Fig. S7).

Full length CDCP1 (∼140 kDa) can be proteolytically cleaved on the cell surface to generate a truncated form (∼90 kDa), and some recent studies have suggested either form promotes cancer aggressiveness through discrete mechanisms (19, 32). On this basis, we next evaluated CDCP1 expression on immunoblot. Full length and cleaved CDCP1 were detected in cell lines (Fig. 2B). In general, full length CDCP1 was equivalently or more abundantly expressed than cleaved CDCP1. In one cell line, 5637, cleaved CDCP1 was more abundant than full length. Interestingly, we found that CDCP1 was robustly expressed in UMUC3, T24, 639V, and TCCSUP lines, which are all negative for NECTIN4 and TROP2; our previous work demonstrated that these NECTIN4^NEG and TROP2^NEG cells lines are not effectively targeted by enfortumab vedotin or sacituzumab govitecan (9, 10). In addition, CDCP1 was detected in three of five bladder PDX samples (Fig. 2C). To understand the relationship between CDCP1, NECTIN4, and TROP2 more broadly, we analyzed the expression correlations between these genes. Across all four datasets, we found that CDCP1 expression was either not correlated or inversely correlated with both NECTIN4 and TROP2 expression (Fig. 2D; Supplementary Fig. S8).

To understand if CDCP1 can be exploited for bladder cancer therapy, we next assessed the tumoral uptake of 4A06, a recombinant human monoclonal antibody we previously developed against an extracellular epitope contained on both full length and cleaved CDCP1 (18). 4A06 was functionalized with desferrioxamine and radiolabeled with Zr-89 as previously described (14, 18). PET and biodistribution studies were conducted at 72 hours post injection of ⁸⁹Zr-4A06, as we have previously shown peak tumor to background values at this time point (14). We assessed radiotracer uptake in five human xenograft models representing higher (UMUC3, T24, HT-1376) and relatively lower (UMUC9, 5637) CDCP1 expression in vitro. Both on PET and ex vivo biodistribution, relative levels of ⁸⁹Zr-4A06 uptake in tumors aligned with the relative CDCP1 expression in vitro (Fig. 3A and B). The highest tumoral uptake of ⁸⁹Zr-4A06 was observed in UMUC3 tumors at ∼20% ID/g. In all cases, the tumor to background ratio was > 1 (Fig. 3C).

We next tested if CDCP1-directed TRT could be applied to treat bladder cancer tumors. We chose to prioritize TRT for several reasons. First, TRT has a straightforward mechanism of action, requiring only antigen overexpression. Second, TRT is agnostic to protein function, which is important as a functional role for CDCP1 overexpression in bladder cancer has yet to be determined (33). Third, TRT is undergoing a clinical and commercial renaissance for cancer therapy, and the recent FDA approvals of Lutathera (¹⁷⁷Lu-DOTATATE) for NE tumors, Azedra (¹³¹I-MIBG) for pediatric malignancies, and Pluvicto (¹⁷⁷Lu-PSMA 617) for patients with prostate cancer fully underscore that TRT can be effective against biologically diverse metastatic solid tumor types (34–36). Finally, in contrast to other antibody-based therapeutic modalities, TRT using alpha or beta emitting radioisotopes has the competitive advantage of irradiation tumor cells not directly engaged by drug (the so-called “cross-fire effect”). We hypothesize this feature of TRT will be particularly beneficial for treating heterogeneous tumor types like bladder cancer.

To test the antitumor effects of RLT, 4A06 IgG1 was first conjugated to the chelator 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA) as described before (14). DOTA-4A06 was radiolabeled with Lu-177 using our previously reported protocol. ¹⁷⁷Lu-4A06 was administered intravenously in two doses (400 μCi/dose) on day 0 and day 7 of the study period to mice bearing subcutaneous UMUC3 tumors. Treatment with ¹⁷⁷Lu-4A06 significantly suppressed tumor growth compared with vehicle controls (Fig. 4A; Supplementary Fig. S9). Notably, four of nine tumors in the treated arm underwent a complete response with no tumor outgrowth even at 120 days post injection. Assessing the entire cohort, ¹⁷⁷Lu-4A06 significantly extended survival compared with the control arm (Fig. 4B). Mice receiving vehicle had a median survival of 13 days, while mice receiving ¹⁷⁷Lu-4A06 had a median survival of 24 days (HR = 4; P < 0.001). The therapy was also well tolerated and no mice in the treatment arm experienced unsafe weight losses (Fig. 4C). Finally, to evaluate antitumor effects in a bladder cancer tumor model with at least 10-fold lower CDCP1 expression, mice with subcutaneous HT-1376 tumors were treated with vehicle or ¹⁷⁷Lu-4A06. Mice received two doses (400 μCi/dose) on day 0 and day 7 of the study period. ¹⁷⁷Lu-4A06 treatment significantly reduced tumor growth compared with vehicle by day 28 (Fig. 4D; Supplementary Fig. S10).

Finally, we carried out an exploratory study to begin probing the mechanisms driving CDCP1 overexpression in bladder cancer. RAS/Raf/MEK signaling has been previously shown to induce expression of CDCP1 in other cancers (Fig. 5A; ref. 37). Hyperactive RAS signaling is a relatively common feature of bladder cancer due to mutations in RAS isoforms, upstream RAS inducers, and downstream effectors (Fig. 5B); therefore, we tested if this pathway regulates CDPC1 in bladder cancer. Treatment of UMUC3 cells (which harbor a KRAS G12C mutation) with G12C inhibitor sotorasib for 48 hours suppressed CDCP1 expression in a dose-dependent fashion on immunoblot and flow cytometry (Fig. 5C and D). Treatment of UMUC3 cells with the MEK inhibitor trametinib for 48 hours also suppressed CDCP1 expression in a dose-dependent fashion on immunoblot and flow cytometry (Fig. 5E and F). These data, while preliminary, underscore a potential role for CDCP1-directed therapeutics in mutant RAS bladder cancer.

In this report, we provide evidence in support of targeting the cell surface antigen CDCP1 to treat bladder cancer. Profiling mRNA and protein expression in bladder cancer biopsies showed that CDCP1 is expressed in all of the consensus subtypes, with the highest levels found in the aggressive basal/squamous subtype and lowest levels in the NE-like subtype. Flow cytometry and saturation binding studies confirmed that CDCP1 is expressed on the cell surface of both basal and luminal human bladder cancer cell lines. Saturation binding studies in cell line models showed a range of overexpression from 10⁵ to 10⁶ receptors per cell. Immunoblot of human bladder cancer cell lines and PDX samples demonstrated that both full length and cleaved CDCP1 are expressed.

Interestingly, we found robust CDCP1 expression in several NECTIN4 and TROP2-negative bladder cancer lines, suggesting that CDCP1-directed therapies may benefit a different molecular subset of patients, given that antigen expression is one of the key determinants of therapeutic response to currently approved ADCs (38, 39). PET/CT studies with ⁸⁹Zr-4A06 detected tumor autonomous expression of CDCP1 in four human bladder cancer models of either basal or luminal histology. An antitumor assessment study showed that ¹⁷⁷Lu-4A06 potently suppressed the growth of UMUC3 tumors compared with vehicle controls and extended overall survival. Of note, we observed durable complete responses in 44% of the mice (four of nine) in the ¹⁷⁷Lu-4A06 treated cohort. Furthermore, we showed that a similar dosing schema for ¹⁷⁷Lu-4A06 suppressed the growth of HT-1376 tumors, a model with 10-fold lower expression of CDPC1 compared with UMUC3. Perhaps unsurprisingly, the antitumor effects were less dramatic than what was observed in the UMUC3 cohort, and no mice were cured. More generally, these data add to an ongoing narrative spanning several antitumor assessment studies in mouse models of prostate and pancreatic cancer showing that ∼10⁵ receptors per cell are required for durable responses to ¹⁷⁷Lu-4A06 IgG (13, 14, 19).

While our data demonstrate that CDCP1 overexpression can be a unifying feature of otherwise molecularly diverse bladder cancer subtypes, the mechanisms driving this overexpression remains to be elucidated. Our preliminary mechanism studies reaffirmed a role for RAS hyperactivity in promoting CDCP1 overexpression, which likely accounts for the high expression found in RAS-mutant cell lines like UMUC3, T24 (HRAS G12V), 639V (KRAS G12D) and HT-1197 (NRAS Q61R). However, we also observed high CDCP1 expression in several cell lines that lack RAS hyperactivity, for example TCCSUP and HT-1376. In these cases, other drivers of CDCP1 may be relevant. More generally, CDCP1 can be induced by numerous oncogenic signaling events relevant to bladder cancer including HER2, EGFR, and PDGFR, and tumor microenvironment effects like hypoxia (40). We are currently working through these candidates to better define the mechanisms that promote CDCP1 overexpression in bladder cancer.

Our study has several important limitations. Firstly, our gene expression data in > 1,000 clinical samples may not necessarily reflect CDCP1 surface protein expression (although we have validated the surface protein expression by IHC using a TMA and in cell line models with flow cytometry). Secondly, due to data availability, the samples analyzed in these cohorts were obtained from primary tumor tissue and may not necessarily represent expression levels in metastatic tumors. Additional study of biopsies harvested from metastatic sites is needed to better characterize the similarities and differences in CDCP1 expression between primary tumors and metastases. Also, there may be intra-patient heterogeneity of expression amongst different metastatic lesions that is underappreciated when profiling single biopsies. Finally, we have performed two antitumor assessment studies using models with high and moderate CDCP1 expression. While the data show significant inhibition and even ablation of a highly proliferative tumor model, the study in its current form does not present additional tumor replicates and does not address the minimum level of CDCP1 expression required for CDCP1-directed TRT to be effective.

Our results showing potent inhibition of an aggressive bladder cancer xenograft model support revisiting TRT-directed therapies for localized and advanced bladder cancer. Indeed, TRT has been sporadically investigated as a treatment option for bladder cancer over the past 30 years. As early as the 1990s, several groups demonstrated that radiolabeled antibodies targeting the cell surface proteins TROP1 (also known as HMFG2) and MUC1 (as known as AUA1 antigen) accumulated in superficial primary bladder tumors at high levels above background (41–43). More recently, an anti-EGFR antibody was coupled to the alpha emitter Bi-213 and safely administered by intravesical infusion to 12 patients with carcinoma in situ (44). Notably, 11 of 12 patients had radiographic responses or stable disease 8 weeks after instillation. We are optimistic that our data, combined with these prior clinical experiences, will provide a compelling rationale to advance TRTs into clinical trials for patients with both localized and metastatic disease.

K. Trepka reports grants from NIH during the conduct of the study. K. Leung reports grants from BMS during the conduct of the study. F.Y. Feng reports personal fees from Bluestar Genomics, Astellas, Foundation Medicine, Exact Sciences, Tempus, POINT Biopharma, Janssen, Bayer, Myovant, Roivant, Bristol Myers Squibb, and Novartis; personal fees and other support from Serimmune; and other support from Artera outside the submitted work. J.A. Wells reports grants from Bristol Myers Squibb during the conduct of the study; in addition, J.A. Wells has a patent for Patent Disclosure to UCSF pending. J. Chou reports grants from NIH/NCI during the conduct of the study. J. Chou also reports personal fees from Exai Bio, as well as grants from Department of Defense Prostate Cancer Program and Prostate Cancer Foundation outside the submitted work; in addition, J. Chou has a patent for Radiolabeled CDCP1 Antibodies and Their Use as Cancer Detection and Treatment Agents pending. M.J. Evans reports grants from Congressionally Directed Medical Research Programs during the conduct of the study; M.J. Evans also reports other support from Hap10 Inc. and ORIC Pharmaceuticals Inc., as well as grants from NIH, V Foundation, American Cancer Society, and LAM Foundation outside the submitted work. In addition, M.J. Evans is a scientific cofounder and equity holder in SUBA Therapeutics, Inc; SUBA is licensing patents from the University of California that relate to CDCP1-based therapeutics. No disclosures were reported by the other authors.

The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

S. Chopra: Data curation, investigation, writing–original draft, writing–review and editing. K. Trepka: Data curation, formal analysis, investigation. S. Sakhamuri: Data curation, investigation. A. Carretero-González: Investigation. J. Zhu: Investigation. E. Egusa: Investigation. J. Zhou: Resources, investigation. K. Leung: Resources, investigation. N. Zhao: Investigation. N. Hooshdaran: Investigation. F.Y. Feng: Resources, writing–original draft. J.A. Wells: Resources, data curation, investigation, writing–original draft, writing–review and editing. J. Chou: Conceptualization, data curation, formal analysis, supervision, investigation, writing–original draft, project administration, writing–review and editing. M.J. Evans: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, investigation, writing–original draft, project administration, writing–review and editing.

We gratefully acknowledge Drs. Youngho Seo and Ryan Tang for technical assistance with small animal PET/CT studies and Dr. Henry VanBrocklin and Mr. Joseph Blecha for assistance with radiochemistry. We thank Dr. Peter Black (University of British Columbia) for sharing the bladder cancer TMA, Dr. Bradley Stohr (UCSF) and Dr. David McConkey (Johns Hopkins) for sharing cell lines, Dr. Jennifer Bolen for histology assistance, and Drs. Sima Porten, Veronica Steri, and the UCSF Preclinical Therapeutics Core for sharing the PDX samples.

M.J. Evans acknowledges funding from the Congressionally Directed Medical Research Programs (W81XWH-21–1-0498), American Cancer Society (130635-RSG-17–005–01-CCE), and the Precision Imaging of Cancer and Therapy program at UCSF. K. Trepka is supported by the UCSF Medical Scientist Training Program. A. Carretero-González acknowledges funding and support from Fundación Alfonso Martín Escudero. J. Chou acknowledges a research gift from Eric and Laurel Speier and grant funding from the NCI (K08CA273514). Research reported in this publication was supported by the NCI of the NIH, and in part by the University of California, San Francisco Histology & Biomarker Core (HBC) under Award Number P30CA082103.

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