Identification of Membrane-expressed CAPRIN-1 as a Novel and Universal Cancer Target, and Generation of a Therapeutic Anti-CAPRIN-1 Antibody TRK-950

Specific targets for cancer treatment are highly desirable, but still remain to be discovered. While previous reports suggested that CAPRIN-1 localizes in the cytoplasm, here we now show that part of this molecule is strongly expressed on the cell membrane surface in most solid cancers, but not normal tissues. Notably, the membrane expression of CAPRIN-1 extended to the subset of highly tumorigenic cancer stem cells and epithelial–mesenchymal transition (EMT)–induced metastatic cancer cells. In addition, we revealed that cancer cells with particularly high CAPRIN-1 surface expression exhibited enhanced tumorigenicity. We generated a therapeutic humanized anti-CAPRIN-1 antibody (TRK-950), which strongly and specifically binds to various cancer cells and shows antitumor effects via engagement of immune cells. TRK-950 was further developed as a new cancer drug and a series of preclinical studies demonstrates its therapeutic potency in tumor-bearing mouse models and safety in a relevant cynomolgus monkey model. Together, our data demonstrate that CAPRIN-1 is a novel and universal target for cancer therapies. A phase I clinical study of TRK-950 has been completed (NCT02990481) and a phase Ib study (combination with approved drugs) is currently underway (NCT03872947) in the United States and France. In parallel, a phase I study in Japan is in progress as well (NCT05423262). Significance: Antibody-based cancer therapies have been demonstrated to be effective, but are only approved for a limited number of targets, because the majority of these markers is shared with healthy tissue, which may result in adverse effects. Here, we have successfully identified CAPRIN-1 as a novel truly cancer-specific target, universally expressed on membranes of various cancer cells including cancer stem cells. Clinical studies are underway for the anti-CAPRIN-1 therapeutic antibody TRK-950.


Introduction
Cancer is one of the leading causes of mortality worldwide, and the incidence is still rising (1). In addition to expanding our ability for early detection, enlarging targets expressed in a broad range of cancers to develop cancer immunotherapies that are applicable to large populations of patients with solid cancers and having unmet medical needs.
For this purpose, we performed serologic analyses of expression cDNA libraries by the SEREX method (5) using both serum of tumor-bearing dogs and canine testis tissue, allowing us to clone cancer antigens eliciting high-titer antibodymediated immune responses. In this process, we identified CAPRIN-1 as a novel candidate target for cancer treatment. A previous report had shown CAPRIN-1 to be a cytoplasmic protein without a transmembrane domain (6). The literature has described several functions of cytoplasmic CAPRIN-1 as follows. CAPRIN-1 is expressed during cell activation from a resting state, during cell division, and is also required for cell growth (6,7). CAPRIN-1 binds to RNAbinding proteins, forming stress granules in the cells (8)(9)(10)(11)(12)(13)(14)(15)(16). However, to the best of our knowledge, no report to date has shown that CAPRIN-1 is expressed on the cell membrane surface. On the basis of our initial data using our newly generated anti-CAPRIN-1 antibodies, we reported the possibility that CAPRIN-1 is specifically expressed on cancer cell membranes and could be a target for antibody cancer treatment (17). After our report above, while several reports demonstrated that intracellular expression of CAPRIN-1 positively correlates with cancer progression and poor prognosis, suggesting that CAPRIN-1 could be a diagnostic biomarker and suitable cancer therapeutic target (18)(19)(20)(21)(22)(23)(24)(25), we continued to focus on performing a detailed analysis of CAPRIN-1 expression on cancer cell membranes. As a result, we can now reveal that CAPRIN-1 has a novel and distinct expression pattern. While CAPRIN-1 is not expressed on the cell membrane surface of nontransformed cells, it is strongly and reproducibly expressed on the cell membrane surface of many cancer types.
Importantly, CAPRIN-1 is also very highly expressed on the membrane surface of highly tumorigenic cancer stem cells. Consistently, cancer cells with particularly high CAPRIN-1 surface expression exhibited enhanced in vitro colony-forming activity and in vivo tumorigenicity. Together, these data suggest that CAPRIN-1 could be an attractive and widely applicable therapeutic target for cancer, for example, antibody-based treatment approaches. Indeed, the subsequently developed humanized and clinical-grade anti-CAPRIN-1 antibody TRK-950 efficiently eliminates various cancer cells and showed an excellent safety profile in primates. On the basis of these results, a phase I clinical study for TRK-950 has been completed (NCT02990481) and a phase Ib clinical study on the effects of TRK-950 in combination with approved anticancer treatments is currently ongoing (NCT03872947) in the United States and France. In parallel, a phase I clinical study in Japan is in progress as well (NCT05423262).

Primary Human Immune Cells
Human peripheral blood mononuclear cells (PBMC) were isolated from peripheral blood of healthy donors by centrifugation with Histopaque-1077 (Sigma-Aldrich, catalog no.10771). Human natural killer (NK) cells were isolated from PBMC using the Human NK Cell Isolation Kit (Miltenyi Biotec, catalog no.130-092-657). The purity of NK cells was confirmed to be ≥95% as assessed by flow cytometry using FITC-labeled human CD45 (BioLegend, catalog no. 304006, RRID:AB_314394) and Allophycocyanin (APC)-labeled human CD56 (BioLegend, catalog no. 318310, RRID:AB_604106) antibodies. Peripheral blood was collected and used at Toray Industries, Inc. in conformity with recognized ethical guidelines including the Declaration of Helsinki. Written informed consent was obtained from each donor. The study protocol was ap-proved by Ethics Review Committee on Research Involving Human Biological Subjects, Toray Industries, Inc. (HC2020-15, HC2021-15).

Primary Cells from Patients
With Institutional Review Board approval, cellular material was collected in conformity with internationally recognized ethical guidelines, including the Declaration of Helsinki. Primary human cancer cells were collected upon patients' written informed consent. Primary gastric cancer (GC-125, GC-187) and colorectal cancer cells (CRC-515, CRC-1278) were collected at the Candiolo The following human cell lines were obtained from European Collection of Authenticated Cell Culture: human uterine cancer Ishikawa (catalog no. 99040201, RRID:CVCL_2529), human normal thyroid Nthy-ori 3-1 (catalog no. 90011609, RRID:CVCL_2659) and mouse myeloma SP2/0 (catalog no. 85072401, RRID:CVCL_2199). 293FT cells (catalog no. R70007, RRID:CVCL_6911) were obtained from Thermo Fisher Scientific. Human ovarian cancer OVCAR3 (catalog no. RCB2135, RRID:CVCL_0465) cells were obtained from RIKEN BRC. All cell lines were cultured according to the cell bank's instructions.
Cells were cryopreserved at early passages (<5) and used from fresh stock vial for each experiment and never continued the culture for more than 2 months. These cells were not further authenticated by our laboratory; however, routine confirmation of in vitro growth properties, morphology provided evidence of correct cell identity.
Mycoplasma testing was performed using TaKaRa PCR Mycoplasma Detection Set (TaKaRa, catalog no. 6601) at the time of stock vial preparation.

SEREX
The healthy canine testis tissue and the serum of tumor-bearing beagle dogs were obtained from the Veterinary ME Research Center, Japan. Messenger RNA was purified from canine testis tissue using the Oligotex dT30 mRNA Purification Kit (Takara, catalog no. 9086), and a cDNA phage library was gen-erated using the ZAP-cDNA Gigapack III Gold Cloning Kit (Agilent, catalog no. 200450). XL1-Blue MRF' competent cells (Agilent, catalog no. 200230) were infected with the phages in the cDNA library and seeded on NZY agar plates (Quality Biological, catalog no. 340-124-231). After incubation for 4 hours at 42°C, we placed a nitrocellulose membrane Hybond-C Extra (Cytiva, catalog no. RPN82E) containing 20 mmol/L Isopropyl β D-thiogalactoside (Nacalai Tesque, catalog no.19742-94) on the plate for 4 hours at 37°C to transfer the expressed proteins to the membrane. After the membrane was blocked with 0.5% skim milk in TBS for 1 hour at room temperature, the serum of a malignant mammary tumor of a beagle dog was incubated for 3 hours at room temperature. The membrane was washed four times with TBS-T (TBS with 0.1% Tween 20) and incubated with Goat anti-Dog IgG antibody (BETHYL Laboratories, catalog no. A40-123, RRID:AB_66900) diluted 3,000 times for 1 hour at room temperature. Positive plaques were detected by NBT/BCIP solution (Roche, catalog no. 11-681-451-001) and collected. We isolated phage plasmids from positive plaques using the ZAP-cDNA Gigapack III Gold Cloning Kit and conducted sequence analysis of these plasmids using T3 and T7 primers.

Rabbit Polyclonal Antibody
The rabbit anti-CAPRIN-1 polyclonal antibody was generated by Biologica Co.
Recombinant human CAPRIN-1 protein (0.5 mg) was mixed with same amount of Freund's Incomplete Adjuvant solution and administered to a rabbit subcutaneously. Immunization was performed once every 2 weeks for four times in total. After the end of immunization, whole blood was collected to obtain serum, and anti-CAPRIN-1 polyclonal antibody (pAb-1) was purified using rProtein A Sepharose Fast Flow (Cytiva, catalog no. 17127902). The control antibody was purified from peripheral blood of rabbits to which the antigen was not administered.

Rabbit Monoclonal Antibody (mAb)
The generation of rabbit anti-CAPRIN-1 mAb was conducted at Epitomics, Inc.

Mouse mAb
We generated mouse anti-CAPRIN-1 mAbs by the following procedure.

Chicken/Human Chimera Antibody
The generation of chicken anti-CAPRIN-1 mAb and chicken/human chimera antibody were conducted at Hiroshima Bio-Medical Co., Ltd. Recombinant human CAPRIN-1 protein (0.1 mg) was mixed with Freund's Complete Adjuvant and administered to the peritoneal cavity of chicken. Immunization was performed once every 3 weeks and three times in total. After a confirmed increase in titer by ELISA, we performed the final immunization, extracted the spleen, and fused the splenocytes with chicken B cells to generate chicken hybridoma cells. Screening of reactivity to CAPRIN-1 was conducted as indicated above.
mAbs were purified and obtained from the supernatants of the cultured hybridoma cells. On the basis of the chicken antibody, chicken/human chimera antibody (mAb-4) was generated by genetic recombination.

Humanized Antibody (TRK-950)
The generation of humanized anti-CAPRIN-1 antibody was conducted at Epitomics, Inc. Total mRNA was extracted from the selected rabbit anti-CAPRIN-1 mAb (mAb-1) hybridoma to generate cDNA via reverse transcription reaction. The variable regions of heavy chain and light chain were cloned from cDNA by PCR. Parts of the frame region in the variable region were replaced with human IgG1 sequence. This sequence was fused with the Fc portion of human IgG1a to generate an expression plasmid (27). This plasmid was transfected into HEK293 cells, and humanized CAPRIN-1 antibody (TRK-950) was purified from the supernatant using HiTrap Protein G HP (Cytiva, catalog no. 17040403).

Competitive Inhibition Assay
The anti-CAPRIN-1 antibodies were incubated at 10 μg/mL with 250 μg/mL of rCAPRIN-1 protein or BSA for 30 minutes at room temperature. Then, we added 1 × 10 5 cells to them and incubation of 1 hour at 4°C. The samples were washed with FACS buffer, the second antibody was added and analyzed by Flow Cytometry.

CAPRIN-1 Stably Expressing Cells
The generation of vector pRP (Exp)-EGFP/Puro-CMV (RRID:Addgene_41841) which coded the full length of human CAPRIN-1 and control vector was conducted by Vector Builder. These vectors were transfected to 293FT cell using Lipofectamine 3000 (Invitrogen, catalog no. 3000001). The transfected cells were added 3 μg/mL of Puromycin (Gibco, catalog no. A11138-03) for selection. After 4 weeks of culture, the expression of CAPRIN-1 was evaluated by Western blot analysis and flow cytometry. We used parent vector transfected cells as mock.

Tissue and Cell Immunostaining
We used formalin-fixed paraffin-embedded cancer and normal tissues on human tissue microarrays (US Biomax, Inc., Shanghai Outdo Biotech Co., Ltd., BioChain Inc., and Diagnostic Oncology CRO, LLC). For cell immunostaining, we fixed cells with formalin, agar, or paraffin embedded them, prior to prepare 2-μm-thick sections. The sections were deparaffinized with xylene, rehydrated with ethanol, and pressure cooked with 0.01 mol/L citrate buffer (pH 6.0) to retrieve the antigen. Intrinsic peroxidase activity was blocked by 3% H 2 O 2 .
To avoid nonspecific staining, samples were treated with 20% normal goat serum/PBS for 20 minutes prior to be incubated with 2 μg/mL mAb-1 antibody overnight at 4°C. We used the MACH4 Universal HRP-Polymer Detection System (BIOCARE MEDICAL, catalog no. BRR4012H) for detection. The proportion of the cells that positively stained for CAPRIN-1 in the cytosol and cell membrane was categorized as follows: no reactivity (0), faint/barely perceptible reactivity (1+), weak to moderate reactivity (2+), and strong reactivity (3+).

Detection of Cancer Stem Cells
Cells were resuspended in FACS buffer and incubated with mAb-1 or TRK-950.

Preparation of Macrophage Conditioned Medium
Monocyte-derived human macrophage cultures were established from PBMCs collected from healthy donors. PBMCs were cultured in Iscove's modified Dulbecco's medium (Gibco, catalog no. 12440061) supplemented with 10% human AB serum (Sigma-Aldrich, catalog no. H4522). To generate "M2-like tumor -associated" macrophages, adherent PBMC-derived monocytes were polarized the day after initial seeding with 0.5 ng/mL of MCSF (PeproTech, catalog no.

Induction of EMT and Immunofluorescent Staining
SiC-003 cells in 100 μL RPMI1640 were seeded in 12-well chamber slides (Ibidi, catalog no. 81201) and cultured overnight. Medium was replaced with 200 μL MCM for induction of EMT or the same volume of standard RMPI1640 as control and cultured for 48 hours (28). Then, cells were washed in PBS, and immediately fixed in 4% paraformaldehyde for 10 minutes at room temperature. Cells were again washed with PBS three times for 10 minutes at room temperature and permeabilized with 0.1% Triton X-100 in PBS for 10

Cell Sorting
HEC-1-A, SK-MEL-5, and Malme-3M cells were sorted into two populations with high and low expression levels of CAPRIN-1 using mAb-1 by flow cytometer (FACS Aria Fusion, BD Biosciences). Sorted cells were washed twice with culture media and cultured at 37°C.

Soft Agar Colony Formation Assay
Anchorage-independent growth of sorted cells was evaluated by soft agar colony formation assay. The sorted cells were passaged within 2 weeks before processing. The cells (1 × 10 4 cells) were incubated in a top layer of 0.4% noble agar (Difco, catalog no. 214220) in DMEM with 10% FBS. The suspension was overlaid on a bottom layer of 0.6% noble agar in DMEM with 10% FBS in a 6well plate. DMEM was added to each well and cultured at 37°C for 2 to 4 weeks. Each well was scanned automatically using the BZ-X810 system (Keyence), and the whole well was reconstructed from the scanned images using the BZ-X Analysis software (Keyence). Colonies larger than 20,000 μm 2 were counted using the Hybrid Cell Count software (Keyence). The experiments were conducted three times independently.

Cell Proliferation Assay
Proliferation of sorted cells was analyzed using CellTiter-Glo Luminescent Cell Viability Assay (Promega, catalog no. G7570) according to the manufacturer's instructions. The luminescence was detected using SpectraMac iD3 (Molecular Devices).
The proliferation of siRNA-transfected cells and TRK-950-treated cells were evaluated using 3H-thymidine (PerkinElmer, catalog no. NET027Z). The detached cells were seeded at 5 × 10 3 cells/well in 96-well plates and incubated at 37°C with CO 2 . After 3 and 6 days, we added 37 Bq of 3H-thymidine per well. After 18 hours, cells were harvested using a FilterMate Cell Harvester (Perkin Elmer) and Unifilter GF/B (Perkin Elmer, catalog no. 6005177), and 3H-thymidine was detected by TopCount NXT (Perkin Elmer).

Binding Affinity
The binding affinity analysis of TRK-950 against CAPRIN-1 protein using

Antibody-dependent Cellular Cytotoxicity Activity
The The ADCC activity of TRK-950 via human and monkey PBMCs was conducted at Translational Drug Development (TD2). We used human PBMCs (AllCells, catalog no. PB003F) and Cynomolgus Monkey PBMCs (Zen-Bio) as effector cells.

Antibody-dependent Cellular Phagocytosis Activity
We evaluated the antibody-dependent cellular phagocytosis (ADCP) activity of TRK-950 by phagocytosis of cancer cells through cancer cell lines, human monocyte-derived macrophages, mouse bone marrow (BM)-derived macrophages, tumor-infiltrating macrophage.
To obtain human macrophages were isolated from human PBMC and cultured in 50 ng/mL human GMCSF (PeproTech, catalog no. 300-03) at 37°C in a CO 2 incubator. Medium was replaced 2 days later, and the cells were harvested after a total of 5 more days. The induction of human macrophages was confirmed by flow cytometry using PE anti-human CD11b antibody (BioLegend, catalog no. 101208, RRID:AB_312791) and PE anti-human CD14 antibody (BioLegend, catalog no. 301806, RRID:AB_314188).
To obtain mouse BM macrophages, BM cells were collected from femora of NOD-SCID mice and cultured for 7 days in RPMI1640 medium with 10% FBS and 1% PS to which 30% conditioned medium from L-929 cells was added.

Xenograft Model
The tumorigenicity of high and low CAPRIN-1-expressing cells was evaluated in xenograft models. The sorted cells were allowed to grow for 4 days before injection. HEC

Cell Fixation and Permeabilization
The cells were harvested using 2.5 mmol/L EDTA, washed with PBS, fixed with 4% formaldehyde for 20 minutes at room temperature, and permeabilized with 0.1% TritonX-100 in PBS (v/v) for 15 minutes at room temperature.

Toxicity Study
The

Data Availability Statement
The data generated in this study are available upon reasonable request from the corresponding author.

CAPRIN-1 is Expressed on the Surface of the Cancer Cell Membrane
A result of the SEREX screening using both serum of mammary gland tumorbearing dog and canine testis tissue, we identified CAPRIN-1 as a novel candidate target for cancer treatment. We generated a rabbit anti-CAPRIN-1 polyclonal antibody (pAb-1) and tested its binding to the surface membrane of four human breast cancer cell lines, using flow cytometry. Although CAPRIN-1 is reportedly an intracellular protein, antibody pAb-1 unexpectedly binds to the surface of all tested cancer cells (Fig. 1A), suggesting that CAPRIN-1 may be expressed on the surface of breast cancer cells. To further explore whether CAPRIN-1 is indeed expressed on cancer cell membranes, we next generated a large variety of mAbs binding to CAPRIN-1 protein from rabbits, mice, and chicken. Most of the generated antibodies, such as mAb-7, mAb-8, and mAb-9, bound to recombinant CAPRIN-1 protein but did not bind to the surface of cancer cell membranes ( Fig. 1B and C; Supplementary Fig. S1A and S1B). Moreover, commercially available anti-CAPRIN-1 antibodies, such as mAb-10 and pAb-2, also did not bind to the surface of cancer cell membranes. However, other antibodies, such as the anti-CAPRIN-1 antibodies mAb-1 to 6, did indeed bind to both CAPRIN-1 protein and the membrane surface of cancer cells such as BT-474 (breast cancer) and SW780 (bladder cancer; Fig. 1B and C).
To further confirm whether antibodies that bound to the surface of cancer cell membranes specifically recognize CAPRIN-1, we conducted Western blot analysis with cell lysates and a single band was detected at approximately 120 kDa (Fig. 1D). Moreover, we investigated the specificity of antibody binding by performing competition assays. Recombinant CAPRIN-1 protein inhibited the binding of anti-CAPRIN-1 antibodies to the surface of cancer cell membranes ( Fig. 1E; Supplementary Fig. S1C). These results strongly indicated that the generated antibodies specifically recognize CAPRIN-1 expressed on the surface of cancer cell membranes.
To further confirm that CAPRIN-1 is indeed expressed on cancer cell membranes, we separated BT-474 cells into cytosolic and membrane fractions for evaluation by Western blotting. The results indicated that CAPRIN-1 was detected both in cytosol and cell membranes ( Fig. 2A). Both mAb-1, which binds to the surface of cancer cell membranes, as well as mAb-7, which does not bind to the membrane surface, showed equal binding patterns once the cells were fixed and permeabilized ( Supplementary Fig. S1D). Furthermore, mAb-1 immunostaining of BT-474, SW780, KKU-213 (cholangiocarcinoma), and OVCAR3 (ovarian cancer) cells revealed intense and consistent staining of cytoplasm and cell membranes. Upon knockdown of CAPRIN-, the staining of each component of the cell was strongly diminished or even abrogated (Fig. 2B). Consistently, CAPRIN- knockdown reduced the binding of mAb-1 to OVCAR3 cells as evaluated via flow cytometry (Fig. 2C). As a result of the knockdown of CAPRIN-1 in cancer cells, cell growth was inhibited, which is consistent with previous reports (Supplementary Fig. S2A; refs. 6, 7), in addition, dead cells increased in some cell lines ( Supplementary Fig. S2B). In contrast, when we introduced the CAPRIN-1 expression vector in 293FT cells to generate cell clones stably expressing CAPRIN-1, we observed increased binding of mAb-1 to the membrane surface (Fig. 2D). These results indicated that CAPRIN-1 is expressed not only in the cytoplasm but also on cell membranes of cancer cells and that some of our antibodies, for example, mAb-1, are capable of specifically detecting both cytoplasmic and membrane CAPRIN-1.

CAPRIN-1 is Expressed on Various Types of Cancer Tissues but not on Normal Tissues
We analyzed CAPRIN-1 expression on various types of cancer cells using mAb-1 for flow cytometry. We found that CAPRIN-1 was expressed on the membrane surface of most cancer cell lines and some cell lines, such as TOV-21G (ovarian cancer) and A-375 (melanoma), were negative, although there were not so many negative cancer cell lines (Fig. 3A). Conversely, we did not find the cell surface expression of CAPRIN-1 on normal cells ARPE-19, Nthy-ori 3-1 and WI-38 (nontransformed cell lines derived from normal tissues). And we did not detect CAPRIN-1 surface expression in a comprehensive set of established hematologic cancer cell lines, although we detected CAPRIN-1 in the cell lysates of both normal and hematologic cancer cells by Western blot analysis ( Fig. 3B; Supplementary Fig. S3). Furthermore, the expression of CAPRIN-1 in cancer tissue and normal tissue arrays was analyzed via IHC using mAb-1. As shown in Fig. 3C (representative IHC images of tissues), CAPRIN-1 was expressed on the membrane of various cancer tissues but hardly expressed on the cell membranes of normal tissues. Weak cytoplasmic CAPRIN-1 expression could be detected by mAb-1 in several normal tissues (Fig. 3C, bottom). Positive ratio of CAPRIN-1 on cell membrane was remarkably high in various cancer tissues (Table 1).

CAPRIN-1 is Strongly Expressed on Primary Human Cancer (Stem) Cells
We next evaluated CAPRIN-1 surface expression via flow cytometry on various primary human cancer cells. The analysis of cells derived from patients with gastric and colorectal cancer demonstrated that CAPRIN-1 was strongly expressed on virtually all cell membrane surfaces ( Fig. 4A and B).
Expression of CAPRIN-1 could also be detected on cancer cells derived from a patient with breast cancer and PDAC cells cultured in sphere conditions ( Fig. 4C and D), suggesting that CAPRIN-1 is also expressed in cells enriched for highly tumorigenic cancer stem cells (29,30). Therefore, we next investigated the expression of CAPRIN-1 on the contained subset of cancer stem cells identified by the surface marker protein CD133 (29), which revealed strong and consistent expression of CAPRIN-1 on these particularly aggressive and treatment-resistant cells (Fig. 4E). These data were further corroborated by a detailed investigation of CAPRIN-1 expression on cancer stem cells identified by their inherent feature of autofluorescence (31). Analysis of the auto-fluorescent subset of cancer stem cells in four different primary cultures derived from patients with advanced-stage PDAC revealed that CAPRIN-1 was expressed on the membrane surface of virtually all cells, with the exception of one patient for which we found 84.7% of cancer stem cells to be positive for CAPRIN-1 (Fig. 4F).
To demonstrate that CAPRIN-1 is not downregulated during EMT (28), we analyzed its surface expression in PDAC cells before and after the induction of EMT using MCM (32). As expected, E-Cadherin was downregulated and Vimentin was upregulated in PDAC cells that had undergone EMT in response to exposure to MCM (Supplementary Fig. S4). Notably, our data show that the strong CAPRIN-1 surface expression remained unchanged, irrespective of the state of the cells related to EMT and stemness (Fig. 4G).
Importantly, CAPRIN-1 could not be detected on the membrane surface of white blood cells isolated from healthy volunteers, normal human fibroblasts, and cancer-associated fibroblasts from patients with PDAC (Fig. 4H). These AACRJournals.org Cancer Res Commun; 3(4) April 2023   NOTE: The number of tissues which CAPRIN-1 was expressed on cancer cell membrane and total number of tested and positive percentage in IHC using mAb-1.
results indicated that CAPRIN-1 is strongly and consistently expressed on primary human cancer cells, including the subset of cancer stem cells. Lack of expression on nontransformed cells suggests that CAPRIN-1 represents a promising target for cancer immunotherapies with a large therapeutic window due to low risk for adverse reactions related to non-cancer targeting.

Correlation of CAPRIN-1 Expression Levels on the Cell Surface with Tumor Cell Growth In Vitro and In Vivo
As aforementioned, we evaluated the expression of CAPRIN-1 on the membrane surface in numerous cancer cells and found that some cells such as HEC-1-A (uterine), SK-MEL-5 (melanoma), and Malme-3M (melanoma) contain subpopulations with varying expression levels for CAPRIN-1 (Fig. 5A).
To study the potential functional implications of this heterogeneity, cells were sorted into two populations with high and low expression levels of CAPRIN-1. After culturing the cells again for at least 1 week after sorting, the expression levels of CAPRIN-1 remained unchanged ( Fig. 5A; Supplementary Fig. S5A), suggesting a stable phenotype. The expression level was analyzed using several anti-CAPRIN-1 antibodies. Indeed, CAPRIN-1 high (high CAPRIN-1 membrane expression) cells showed enhanced anchorage-independent growth in soft agar colony-forming assay (Fig. 5B), whereas no obvious difference was observed for their growth rate in regular adherent conditions ( Supplementary Fig. S5C). When identical numbers of CAPRIN-1 high and CAPRIN-1 low (low CAPRIN-1 membrane expression) cells were implanted into NOD-SCID (NOD.CB17-Prkdcscid/NcrCrl) mice, CAPRIN-1 high cells resulted in much faster tumor growth ( Fig. 5C; Supplementary Fig. S5B). These results indicated that the cells with a high expression of CAPRIN-1 on the membrane surface bear higher in vivo tumorigenicity. To rule out contamination in the cancer cells used, multiple genes known to be expressed in each cancer cell were selected from information in the database and expression was evaluated. As a result, expression profiling of the original unsorted cells versus the cells sorted for CAPRIN-1 levels showed consistent expression levels for various genes, thereby reassuring that the cell populations were derived from the same parental clones ( Supplementary Fig. S5D).
Next, to investigate whether the expression levels of CAPRIN-1 on the membrane surface of cells are affected by a hypoxic state, which is one of the general features of the intratumoral environment (33), the aforementioned cell lines with high and low expression fractions of CAPRIN-1 on the membrane surface of cells were cultured under hypoxic conditions. The results showed that the proportion of CAPRIN-1 high cells increased and the proportion of CAPRIN-1 low cells decreased in hypoxic cultures compared with normal cultures (Supplementary Fig. S5E). Of note, SK-MEL-5 caused cell death under hypoxic conditions and could not be used for the evaluation.

CAPRIN-1 Expression on the Cell Membrane Surface Increases Due to Stressful Stimuli to Cancer Cells
Knowing that cytoplasmic CAPRIN-1 migrates and accumulates in stress granules in response to stressful stimuli (12), we exposed cancer cells to the Endoplasmic Reticulum (ER) stress inducer Tunicamycin and evaluated whether the expression level of CAPRIN-1 on the cancer cell membranes changes. Indeed, we found that Tunicamycin increased surface CAPRIN-1 expression in cancer cells with virtually no CAPRIN-1 expression at baseline. The stressinduced surface expression of CAPRIN-1 was abrogated following CAPRIN-1 knockdown, confirming that the increased expression was due to CAPRIN-1 ( Supplementary Fig. S6).

Generation of Humanized Anti-CAPRIN-1 Antibody TRK-950
On the basis of the rabbit mAb (mAb-1) that reliably identifies CAPRIN-1 on the surface of cancer cell membranes, the humanized anti-CAPRIN-1 antibody (TRK-950) was generated. TRK-950 maintained the specificity of the original antibody and consistently binds to both CAPRIN-1 protein and CAPRIN-1 on the surface of cancer cell membranes ( Fig. 6A and B). The binding affinity of TRK-950 to recombinant CAPRIN-1 protein was evaluated using Biacore3000 ( Supplementary Fig. S7A), which demonstrated an affinity (K D ) of 3.98 × 10 −13 (mol/L).

Mechanism of Action of TRK-950
Because TRK-950 did not directly affect cancer cell growth ( Supplementary  Fig. S7B), we evaluated the indirect antitumor effects of TRK-950 via immune cells including human NK cells (for ADCC assay) and human monocytederived macrophages (for ADCP assay). Importantly, TRK-950 exhibited cytotoxicity and phagocytosis to CAPRIN-1 + cancer cells in a dose-dependent fashion ( Fig. 6C and E). Notably, TRK-950 did not exhibit cytotoxicity using normal cells (Fig. 6D). We conducted these experiments using NK cells and monocytederived macrophages from several donors and obtained similar results.

Antitumor Effect and Cancer-specific Accumulation of TRK-950 Using In Vivo Murine Models
To evaluate the antitumor effect of TRK-950, TRK-950 was administered into BT474-bearing NOD-SCID mice at 100, 30, 10, 5, 3, 1, or 0.3 mg/kg dose once a week. We observed statistically significant inhibition of tumor growth at day 49 compared with the control group. Tumor growth inhibition was also observed in several dose groups in twice a week and every other week administration studies (Fig. 6F). In addition, TRK-950 administration showed tumor growth inhibition in COLO 205 (colon cancer) bearing mouse model ( Supplementary   Fig. S8A). During these studies, no significant adverse reactions in mice were observed for the TRK-950 groups.
AACRJournals.org Cancer Res Commun; 3(4) April 2023  We studied ADCC and ADCP activities using murine effector cells from NOD-SCID mice. We did not detect any ADCC activity of TRK-950 mediated through mouse NK cells (Supplementary Fig. S8B). Conversely, we noticed strong ADCP activity via murine J774.1 cells, mouse BM-derived macrophages ( Supplementary Fig. S8C and S8D). Macrophages were the most abundant tumor-infiltrating CD45 + cells in the BT-474-bearing mice (Supplementary Fig. S8E). In addition, TRK-950 showed specific ADCP activity via macrophages separated from tumor mass of the mice (Supplementary Fig. S8F). From these data, the action mechanism of the TRK-950 antitumor effect in the xenograft model is considered as an ADCP activity via macrophages.
To confirm the cancer specificity of TRK-950 in vivo, IR800Dye-conjugated TRK-950 was administered to BALB/c mice transplanted with CAPRIN-1expressing mouse 4T1 cells and the in vivo localization of the conjugate via imaging was analyzed. One day after administration, the conjugate accumulated in the tumor and remained visible until day 6, suggesting that TRK-950 is recruited to the tumor in vivo and shows tumor-specific accumulation ( Supplementary Fig. S8G).

Toxicity Study of TRK-950 in Monkeys
The toxicity of TRK-950 was evaluated using cynomolgus monkeys. CAPRIN-1 is a protein with extremely high interspecies homology, with at least 95% homology (of the amino acid sequence) between humans and mice, and as shown in the above results ( Supplementary Fig. S8G), the anti-human CAPRIN-1 antibody (TRK-950) binds to the cell membrane surface of mouse cancer cells (4T1), and therefore it also has binding affinity with mouse CAPRIN-1. CAPRIN-1 of cynomolgus monkeys and that of human share amino acid sequence with a homology of 100% in the main variant and TRK-950 binds to both human and monkey CAPRIN-1, and no concern about cross-reactivity to the target was observed ( Supplementary Fig. S9A). In addition, ADCC and ADCP activities of TRK-950 to cancer cells were comparable with those of human and monkey effector cells ( Supplementary Fig. S9B  decrease in the weight of the thyroid and parathyroid were observed, but the dose-response trend was not observed in these changes. The weight of these organs returned to baseline following the 8-week recovery period. In the pharmacokinetic analysis, a minor ADA formation that did not have a significant impact on exposure was observed. At any administered dose, TRK-950 was sufficiently exposed. The exposure of the maximum dose of 200 mg/kg was 396,000 ± 48,000 μgh/mL in male (AUC, 0-168 hours) and 351,000 ± 39,000 μgh/mL in female (AUC, 0-168 hours). The half-life was 4.6 to 7.9 days at 200 mg/kg on day 15. Throughout the test period, TRK-950 was sufficiently exposed, and no significant difference was observed between sexes. When TRK-950 was administered in the dose range as discussed above (3-200 mg/kg), only minor changes were observed and all these changes were reversible, indicating the high tolerability and safety of TRK-950.

Discussion
Here we demonstrate for the first time that CAPRIN-1 is strongly expressed on the cell membrane surface in most solid cancers, but not normal tissues. When the mRNA expression of CAPRIN-1 was compared between cancer tissues (The Cancer Genome Atlas data) and normal tissues (GTex data) at the GEPIA site, the expression tended to be high in cancer tissues. On the other hand, tissue staining data from Protein Atlas showed the expression of CAPRIN-1 in both normal tissues and cancer tissues with a high score of almost 100%. These do not distinguish between CAPRIN-1 within cells and CAPRIN-1 on the cell membrane. On the other hand, in our data (Fig. 3), mAb1 recognizing CAPRIN-1 on the cell membrane was used and the evaluation was performed under the condition of detecting the expression of CAPRIN-1 on the cell membrane, and a tendency different from the above was observed. CAPRIN-1 was originally identified as a cell membrane protein with a GPI-anchored domain (34). However, it was later determined that the originally published nucleotide sequence was partially wrong, and it was concluded that CAPRIN-1 is a strictly intracellular protein (6). Therefore, it came as a great surprise when we detected CAPRIN-1 on the cell surface and later could indeed validate CAPRIN-1 as a cancer-specific cell surface antigen. The specificity of cancer cell membrane binding anti-CAPRIN-1 antibodies we generated was confirmed by (i) competition assays using soluble recombinant CAPRIN-1 and (ii) abolished binding to CAPRIN-1-deficient cancer cells following CAPRIN- knockdown. As shown in Fig. 2B, CAPRIN-1 proteins in both the cell membrane and cytoplasm were suppressed in CAPRIN- knockdown cells. As a result of the knockdown of CAPRIN- in cancer cells, dead cells increased, suggesting the possibility that CAPRIN-1 is indispensable for cancer cell survival. In flow cytometry analysis, CAPRIN-1 expression on the cell membrane surface was decreased but not abrogated completely in CAPRIN- knocked down cells. We think that the cell membrane surface has a longer turnover time than and may take longer to inhibit expression than CAPRIN-1 in the cytoplasm. Details regarding this can be explained by elucidating the mechanism of membrane expression of CAPRIN-1 in the future.
We confirmed that CAPRIN-1 is present in the cytoplasm, which is consistent with previous reports, and simultaneously determined that a part of the total CAPRIN-1 is actually exposed on the cell membrane surface, a phenomenon that occurs exclusively in cancer cells. Interestingly, analyzing the amino acid sequence using the TMHMM server v.2.0 and SOSUI system we could not predict the formation of transmembrane helices. While these results seem counterintuitive, it is important to note that surface expression of intracellular molecules on cancer cells has been reported previously for various proteins, including heat shock chaperones, cytokeratin, Vimentin, etc. (35). Mechanistically, cell surface exposure can be caused by (i) differential splicing, or (ii) posttranscriptional modifications such as cancer specific glycosylation allowing protein access to the cancer cell surface. Because we did not find differential splicing for CAPRIN-1, we reason that CAPRIN-1 is best classified as a moonlighting protein, which refers to a growing group of intracellular proteins with various nongenetic functions inside and outside the cell (36).
Cancer cells with enhanced stemness features, also referred to as cancer stem cells, are defined by their ability to self-renew and give rise to more differentiated progenies. Moreover, cancer stem cells are highly metastatic, more resistant to anticancer drugs and radiation, thereby driving relapse of disease, and are correlated with poor prognosis (37,38). While targeting cancer stem cells is a promising therapeutic strategy, current treatments mostly spare this highly aggressive subset of cancer cells, resulting in their relative enrichment and subsequently more aggressive disease during relapse (39). Therefore, we were particularly intrigued to find that CD133 + cancer stem cells derived from patients with pancreatic cancer also strongly express CAPRIN-1 on the cell surface. We further corroborated high level expression of CAPRIN-1 on pancreatic cancer stem cells using autofluorescence as a distinct method for detecting cancer stem cells with high accuracy in primary cancer cells (31). Notably, cancer cell surface expression of CAPRIN-1 was not affected by the EMT process, which is an important initial step of the metastatic process. While EMT induction resulted in a marked downregulation of well-known epithelial cancer markers such as E-Cadherin, rendering such epithelial surface markers rather inappropriate for targeting of metastatic disease, there was still high surface expression of CAPRIN-1. Future study is needed to clarify the relationship between stemness or EMT status and CAPRIN-1 membrane surface expression.
In addition, CAPRIN-1 high cancer cells have higher in vitro colony-forming activity and in vivo tumorigenicity compared with CAPRIN-1 low cancer cells.
The sorted CAPRIN-1 low Malme-3M cells tended to have a higher percentage of floating cells than the CAPRIN-1 high cells ( Supplementary Fig. S5A). This result suggests that the CAPRIN-1 on the cell membrane surface may function in cell adhesion, and this function may be involved in the tumorigenic potential. A hypoxic state is one of the common features of the intratumoral environment (33). Thus, to investigate whether hypoxic conditions affect the expression levels of CAPRIN-1 on the membrane surface of cells, cell lines with high and low expression fractions of CAPRIN-1 on the membrane surface of cells were cultured under hypoxic conditions. The results showed that the proportion of CAPRIN-1 high cells increased and the proportion of CAPRIN-1 low cells decreased in hypoxic cultures compared with normal cultures. These results suggest that tumor-specific environmental influences, such as hypoxia and other stress stimuli, may cause cancer cells to adapt to the environment and thereby cause CAPRIN-1 to be expressed on the membrane of cells, thereby contributing to malignant transformation to cancer including cancerous proliferation (enhanced tumorigenicity).
On the basis of the capacity of CAPRIN-1 to relocate within the cytoplasm in response to stressful stimuli (12), we hypothesized a potential dynamic translocation of CAPRIN-1 to the cell membrane surface. To explore this concept, we tested various stress stimuli, and found that CAPRIN-1 surface expression was indeed slightly increased in response to ER stress.
Further investigations to gain insights into the mechanisms of how CAPRIN-1 is recruited to the cancer cell membrane and its potential molecular interactions with other surface proteins are warranted and will be the subject of future basic research. Despite these open questions, we felt confident that CAPRIN-1 is a promising target for antibody drugs in solid cancers, and also represents an exciting opportunity for ADCs and CAR-T cells. Therefore, we prepared a humanized anti-CAPRIN-1 mAb namely TRK-950, advanced to the preclinical stage for the treatment of solid cancers.
Preclinical testing was conducted using mouse xenograft models transplanted with human cancer cells, which demonstrated a strong antitumor effect. As expected, in this model ADCC activity via NK cells was greatly reduced because of the functional deficiency of NK cells in NOD-SCID mice. To investigate the antitumor mechanism in this model, we analyzed tumor-infiltrating immune cells and revealed that the main cells were macrophages which showed strong ADCP activity with TRK-950. As it was revealed that TRK-950 showed strong ADCC and ADCP activities against cancer cells via human immune cells in vitro, we expect that TRK-950 exerts a high therapeutic effect on patients with cancer by synergistic action of both activities.
Prior to clinical testing we had to ensure safety for use in humans. No overt toxicity up to the highest dose was recorded, and a no-observed-adverse-effectlevel (NOAEL) of ≥200 mg/kg/dose was determined. In contrast, previous toxicity studies in non-tumor-bearing monkeys for antibody drugs targeting cancer cell surface antigens such as HER2 (trastuzumab, Herceptin; NOAEL: 59 mg/kg/dose) and CD20 (rituximab, Rituxan; NOAEL ≥20 mg/kg/dose) had indeed shown toxicity (interview form of Herceptin, Chugai Pharmaceutical Co. Ltd.; interview form of Rituxan, Chugai Pharmaceutical Co., Ltd.). Most likely, the toxicity observed in these nonclinical studies was due to shared cancer antigen expression in normal tissues. Therefore, these data indicate that TRK-950 for targeting cell surface CAPRIN-1 provides an excellent safety profile with very low toxicity in cynomolgus monkeys.
Collectively, these data showed that TRK-950 represents a novel and promising cancer drug for clinical applications, a notion that is further supported by the fact that we have already successfully completed a phase I clinical study, and are currently conducting a phase Ib clinical study on TRK-950 as a new cancer drug candidate in combination with conventional cancer drugs.