Malignant tumors, which account for approximately 90% of human cancers, arise from epithelial tissues. Epithelial cells are connected via epithelial sheets that line organ cavities. Tight junctions (TJs) regulate permeability across epithelial sheets and play a role in the development and maintenance of cell polarity [1
]. Loss of TJ integrity leads to aberrant cell growth by increasing the influx of growth factor; it promotes the detachment of malignant cells from the primary tumor site by disrupting cell adhesion and polarity, resulting in the formation of distant metastasis [2
]. Tight junction components are attractive targets, since they are accessible in cancer cells compared to normal cells. In normal epithelia, cells grow parallel to the lumen maintaining their polarity, and the accessibility to the TJ components, located on the apical side of the epithelium, is limited. However, in the initial phase of epithelial tumorigenesis, the orientation of mitotic spindles becomes irregular, leading to out-of-plane division [4
]. The misorientation of the cell division axis can change the pattern of the cell–cell adhesion system and cause TJ components to be exposed externally by rotation of the spindle. Hence, TJ components are potential therapeutic targets for cancer treatment or delivery of anticancer agents, specifically into tumor tissues [5
Claudins (CLDNs) are four-transmembrane domain proteins and components of TJs of epithelial or endothelial cells; they function in paracellular permeability by forming TJ seals that prevent the free movement of lipids, proteins, and solutes between cells [9
]. Human claudin-3 (hCLDN3) has 220 amino acids (aa) and consist of cytoplasmic N- and C-termini, four transmembrane domains, and two extracellular loops, ECL1 (aa 27–80) and ECL2 (aa 144–159) from the N-terminus [11
]. Although the role of CLDN3 in carcinogenesis is unclear and controversial [12
], high expression of CLDN3 has been reported in various carcinomas including breast, colorectal, gastric, pancreatic, prostate, and ovarian cancer [17
]. High CLDN3 expression is correlated with poor prognosis and survival [21
], and CLDN3 is considered a pan-carcinoma biomarker.
Because of its expression in various carcinomas, CLDN3 has been investigated as a therapeutic target. Clostridium perfringens enterotoxin (CPE), which cause food poisoning, recognizes ECL2 of CLDN3 and CLDN4 [25
]. Clostridium perfringens enterotoxin, C-terminal fragment of CPE (C-CPE, a CLDN binding domain without cytotoxicity), and C-CPE fusion protein have been investigated for their therapeutic potential in cancer. Clostridium perfringens enterotoxin shows anti-tumor efficacy in prostate cancer [8
], breast cancer [27
], and ovarian cancer [28
]. C-terminal fragment of CPE increases the efficacy of chemotherapy in ovarian cancer [29
]. C-terminal fragment of CPE fusion proteins show cytotoxicity effects in tumors [30
] and have been investigated as possible drug carriers [32
]. Moreover, C-CPE has been used as an optical imaging agent and as a potential carrier to specifically deliver therapeutic drugs in ovarian cancer [33
]. These approaches using CPE provide proof of concept for cancer diagnosis and treatment targeting CLDN3 [35
]. However, CPE can also bind to CLDN5, CLDN6, CLDN7, CLDN9, and CLDN14 [26
]. C-terminal fragment of CPE induced an immune response in mice [37
], and C-CPE-fused toxin led to hepatic injury [38
]. Therefore, it is necessary to develop therapeutics targeting CLDN3 with high specificity, low immunogenicity, and low toxicity.
Antibodies are highly specific antigen-targeting molecules with low immunogenicity and toxicity [39
]. The bivalent property of antibody displays an avidity effect which is the accumulated strength of multiple affinities and is commonly referred to as a functional affinity. The avidity plays critical roles in penetration, catabolism, specificity, and efficacy [40
]. For these reasons, many therapeutic antibodies have been developed, and over 80 therapeutic antibodies have been approved by the FDA [41
]. Antibodies kill tumor cells directly through receptor blockade, antibody-dependent cellular cytotoxicity (ADCC) or complement-dependent cytotoxicity (CDC) [42
]. In addition, the precision targeting moiety of antibody allows specific and selective delivery of therapeutic agents including cytotoxic agents [44
], radioisotopes [45
], toxins [46
] or photosensitive agents [47
]. Diagnostic probes, such as fluorophores, radioisotopes, computed tomography (CT) contrast agents or paramagnetic particles, are also available for antibody-based imaging in a systemic and non-invasive manner for early cancer diagnosis, detecting cancer lesions, and monitoring the prognosis of treatment [48
It is difficult to produce antibodies against CLDNs because extracellular loops are so short that they show low immunogenicity; in addition, they have high sequence homology among humans, mice, and rats [50
]. Moreover, there are limitations associated with mimicking the four-transmembrane structure using recombinant proteins. In this study, we generated a human immunoglobulin G1 (IgG1) monoclonal antibody (h4G3) against CLDN3 from a human single-chain variable fragment (scFv) phage library [51
] using CLDN3-expressing CHO-K1 cells and CLDN3-embedded lipoparticles as antigens and confirmed its specific binding to human and mouse CLDN3 without affinity to other CLDN family members. We demonstrated the ADCC activity of h4G3 in many types of cancer cell lines according to CLDN3 expression levels. Finally, we visualized that h4G3 specifically recognized CLDN3-expressing tumors rather than normal organs in mouse xenograft models.
2. Materials and Methods
2.1. Cell Lines and Cell Culture
Human breast cancer (T47D, MCF-7, and HCC202), human ovarian cancer (OVCAR-3, Caov-3, and TOV-112D), human colon cancer (T84, NCI-H508, and SW1116), human gastric cancer (SNU-216 and Hs746T), human liver cancer (NCI-H684), and human pancreatic cancer (AsPC-1) cell lines were used as target cancer cells. The CHO-K1 (Chinese hamster ovary), L cells (mouse fibroblast), HEK293 (human embryonal kidney), and NK-92MI (human lymphoblast) cells were used for preparing the transfectants. All cell lines were purchased from Korean Cell Line Bank (KCLB; Seoul, Korea), except TOV-112D, T84, CHO-K1, L cells, and NK-92MI cells, which were purchased from the American Type Culture Collection (ATCC; Manassas, VA, USA). The Caov-3, Hs578T, and NCI-H684 cells were cultured in DMEM (HyClone, Logan, UT, USA) and supplemented with 10% fetal bovine serum (FBS; HyClone), 100 units/mL penicillin, and 100 μg/mL streptomycin. The TOV-112D cells were maintained in a 1:1 mixture of Media199/MCDB medium (HyClone) containing 15% FBS, 100 units/mL penicillin, and 100 μg/mL streptomycin. The T84 cells were cultured in DMEM/F12 (Gibco, Carlsbad, CA, USA) containing 5% FBS, 100 units/mL penicillin, and 100 μg/mL streptomycin. The NK-92MI cells were maintained in MEM alpha (Gibco) containing 20% FBS, 100 units/mL penicillin, and 100 μg/mL streptomycin. All other cell lines were cultured in RPMI-1640 (HyClone) supplemented with 10% FBS, 100 units/mL penicillin, and 100 μg/mL streptomycin. All cells were incubated at 37 °C in humidified 5% CO2 atmosphere.
2.2. Establishment of Stable CLDN Transfectants
The cDNAs for CLDN1, 3, 4, 5, 6, 8, 9, and 17 were cloned into pcDNA3.1(+) (Invitrogen, Carlsbad, CA, USA). The CLDN3 expression construct was then transfected into CHO-K1, L cells, HEK293, and TOV-112D cells using FuGENE HD transfection reagent (Promega, Madison, WI, USA); other CLDN expression constructs were transfected into HEK293 cells. After transfection, G418-resistant cells were selected and isolated using a clonal cylinder. To generate chimeric ECL fusion CLDN1/CLDN3 transfectants, cDNA fragments encoding aa 1~104 of CLDN1 fused with aa 104~220 of CLDN3 and aa 1~103 of CLDN3 fused with aa 105~211 of CLDN1 were transfected into HEK293 cells and subjected to the procedure described above.
2.3. Establishment of NK-92MI-CD16a Cell Line
A high affinity variant (158V) of FcγRIIIa (CD16a) cDNA was cloned into pcDNA3.1(+) (Invitrogen). The NK-92MI cells were transfected with the CD16a plasmid by electroporation and selected with G418. After selection, CD16a-expressing cells were isolated using BD FACSAriaTM III (BD Biosciences, San Jose, CA, USA) at high-expression populations.
2.4. Production of Human Monoclonal Antibody (mAb)
The Freedom pCHO 1.0 vector (Gibco), including heavy and light chains of h4G3, was transfected into Freedom CHO-S cells (Gibco) in accordance with the manufacturer’s instructions to establish stable h4G3-expressing CHO-S cells. This transfectant was fed 4 g/L glucose on day 3 and 5, and 6 g/L glucose on day 7, and incubated for up to 2 weeks. The culture supernatant was loaded onto MabSelect SuRe Protein A resin (GE Healthcare, Piscataway, NJ, USA). The resin was washed with five column volumes (CVs) of 35 mM sodium phosphate and 500 mM NaCl (pH 7.2). The bound antibody was eluted with five CVs of 0.1 M sodium citrate (pH 3.6) and neutralized by 1 M Tris-HCl (pH 8.0). The buffer exchange and concentration were achieved using an Amicon Ultra-15 (Merck Millipore, Billerica, MA, USA). The purified antibody was quantified using the Cedex Bio Analyzer (Roche, Indianapolis, IN, USA), and the intact form of the antibody was analyzed using SDS-PAGE under reducing and non-reducing conditions.
2.5. Flow Cytometry Analysis
To analyze h4G3 binding to CLDN, cells were detached with enzyme-free, PBS-based cell dissociation buffer (Gibco). A total of 2.5 × 105 cells were incubated with human IgG (10 μg/mL) (Jackson Immunoresearch Laboratories, West Grove, PA, USA) or h4G3 (10 μg/mL) in PBS containing 1% FBS for 1 h on ice. The cells were then washed three times with PBS containing 1% FBS and incubated with goat anti-human IgG-FITC (Jackson Immunoresearch Laboratories) (1:100 dilution) for 1 h on ice. Stained cells were washed three times and analyzed using a BD FACSCalibur system equipped with the Cell Quest Pro software (BD Biosciences).
2.6. Western Blotting and Immunoprecipitation
For Western blotting, cells were lysed in RIPA lysis buffer (150 mM sodium chloride, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, 50 mM Tric-HCl, pH 7.4, 2 mM EDTA) including protease inhibitor cocktail (Roche). The supernatant was collected by centrifugation at 14,000 rpm for 15 min at 4 °C, and protein concentration was determined using the BCA protein assay kit (Thermo Fisher Scientific, Waltham, MA, USA). Lysed proteins were resolved by 15% SDS-PAGE and transferred to a polyvinyl difluoride (PVDF) membrane (Bio-Rad Laboratories, Hercules, CA, USA). The membrane was then blotted with anti-CLDN1 (Santa Cruz Biotechnology, Santa Cruz, CA, USA), anti-CLDN3 antibody (Abcam, Cambridge, UK) or β-actin (Santa Cruz Biotechnology). For immunoprecipitation under non-denaturing conditions, cells were resuspended in 600 μL PBS containing a protease inhibitor cocktail (Roche) and sonicated for 10 cycles of 2 s pulses on and 5 s pulses off using a Branson Digital Sonifier 450 (Branson Ultrasonics, Danbury, CT, USA). After quantifying the lysed proteins using the BCA protein assay kit, 1 mg of proteins was incubated with 1 μg human IgG (Jackson Immunoresearch Laboratories) or 1 μg h4G3 for 1 h at 4 °C with rotation. The antibody-mixed lysate was incubated with 50 μL of Protein A-Agarose (Roche) for 1 h at 4 °C with rotation. The precipitated beads were washed three times with PBS, and the precipitated proteins were analyzed by Western blotting.
For immunofluorescence, hCLDN3/TOV-112D, TOV-112D, OVCAR-3, and Caov-3 cells were seeded on a 4 well cell culture slide, grown to 80% confluency and treated with 5 μg/mL of human IgG (Jackson Immunoresearch Laboratories) or h4G3 for 1 h at 4 °C. The cell culture slide was washed with PBS, and cells were fixed by 4% formaldehyde for 15 min. Cells were washed with PBS, blocked with PBS containing 5% BSA for 1 h, and incubated with goat anti-human IgG-FITC (Jackson Immunoresearch Laboratories) (1:200 dilution) for 1 h. Cells were washed with PBS, stained with Hoechst 33342 (Invitrogen) for staining nuclei, and mounted using Fluoromount Aqueous Mounting Medium (Sigma–Aldrich, St Louis, MO, USA). Images were taken with the LSM 700 ZEISS laser scanning confocal microscope (Carl Zeiss, Jena, Germany). Data were processed using ZEN confocal software (Carl Zeiss).
2.8. Cell-Based Affinity Kinetics
The binding kinetics against CLDN3 on the cell surface were measured using LigandTracer Green (Ridgeview Instruments AB, Vänge, Sweden). The hCLDN3/HEK293, hCLDN3/TOV-112D, and mCLDN3/HEK293 cells, as positive cells, and HEK293 and TOV-112D cells, as negative cells, were seeded on a limited area of 100 mm culture dish at a density of 3 × 105 cells/mL in 500 μL culture medium, and after 6 h, 10 mL growth medium was added to culture dish. Cells were incubated overnight, and 3 mL of the medium was changed before the experiment. The h4G3 was labeled with DyLight dye 488 using DyLight Antibody Labeling Kits (Thermo Fisher Scientific) following the manufacturer’s instructions. The cell culture dish was clamped onto the device and the fluorescence baseline was recorded. Each time the respective fluorescence reached equilibrium, Dylight dye 488-labeled h4G3 was added stepwise to a final concentration of 3 nM and 9 nM for hCLDN3 cell lines, and 30 nM and 90 nM for mCLDN3 cell lines. In the dissociation phase, the remaining medium was removed, and 3 mL fresh medium was added to the culture dish. All measurements were performed using a 15 s detection time and 4 s detection delay. Recorded data were analyzed by TraceDrawer (Ridgeview Instruments AB).
2.9. In Vitro Antibody-Dependent Cellular Cytotoxicity (ADCC)
The ADCC assay was performed in various cell lines. Target cells were seeded on 96 well plates at a density of 2 × 104
cells/well and incubated overnight. In the following steps, RPMI 1640 with 5% FBS was used for preparing materials. The NK-92MI-CD16a cells, which stably express CD16a, were used as effector cells. The target cells were incubated with human IgG (Jackson Immunoresearch Laboratories) or h4G3 at a final concentration of up to 10 μg/mL and 8 × 104
effector cells in a CO2
incubator for 4 h at 37 °C (target cells/effector cells ratio was 1/4). Target cell lysis was measured by detecting the release of lactate dehydrogenase (LDH) using CytoTox 96®
Non-Radioactive Cytotoxicity Assay according to the manufacturer’s instructions. The absorbance of the plates was analyzed on a SparkTM
10M microplate reader (Tecan, Männedorf, Switzerland) at 490 nm. For data analysis, the percentage of specific ADCC was calculated as follows:
The dose–response curve and EC50 values were estimated using GraphPad Prism 7 (GraphPad Software, San Diego, CA, USA).
2.10. Biodistribution in a Nude Mouse Xenograft Tumor Model
Human IgG (Jackson Immunoresearch Laboratories) and h4G3 were labeled with fluorescence dye CF750 using VivoBrite Rapid Antibody Labeling Kit (Biotium Inc., Hayward, CA, USA) in accordance with manufacturer’s instructions. To generate xenograft, 5 × 106 OVCAR-3 cells in 100 μL PBS were injected subcutaneously into the right flank of 6 week old female athymic nude mice (Orient Bio, Seongnam, Gyeonggi, Korea). For the T47D xenograft model, 1 × 107 T47D cells in 100 μL PBS were injected subcutaneously into athymic nude mice planted with 17β-estradiol pellets (Innovative Research of America, Sarasota, FL, USA). Tumor-bearing mice were treated intravenously with CF750-labeled antibodies as 100 μg/100 μL PBS. After 6, 24, 48, 72, and 96 h, the mice were anesthetized with TerrellTM isoflurane (Piramal Critical Care Inc., Bethlehem, PA, USA) and placed in the IVIS Spectrum CT (Perkin Elmer, Waltham, MA, USA) to visualize CF750-labeled antibodies. The fluorescence was detected using an excitation filter (710 nm) and emission filter (780 nm). At the final time point, the mice were sacrificed and the liver, kidney, lung, spleen, intestine, and tumor were excised. The organs from each mouse were placed in 100 mm Petri dishes and imaged using IVIS Spectrum CT (Perkin Elmer). The intensity of fluorescence was analyzed using Live Imaging software (Perkin Elmer), and the average fluorescence intensity from organs was calculated by creating a region of interest over each organ. All animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of Seoul National University (SNU-190216-1).
2.11. Statistical Analysis
The significance of differences among groups was evaluated using two-way analysis of variance (ANOVA). Data were analyzed with GraphPad Prism 7 (GraphPad Software) and p < 0.05 was considered statistically significant.
Overexpression of CLDN3 has been reported in various malignant carcinomas, and CLDN3 has emerged as a potential cancer biomarker and therapeutic target because of its surface expression pattern associated with tumorigenesis [4
]. Antibody screening using recombinant proteins is not suitable for antibody development against antigens with complex structures such as four-transmembrane domain proteins including CLDNs. In a structural binding study between C-CPE and ECL2 of CLDN4, various GST fusion recombinant proteins containing different domains of CLDN4 reveal that ECL2 with the downstream transmembrane domain of CLDN4 shows optimal binding activity to C-CPE, suggesting that ECL2 in the context of neighboring domains displays better conformation than a separated peptide such as only the ECL2 domain [56
]. Therefore, it is crucial to choose a proper antigen for developing antibodies capable of recognizing an antigen on the cell membrane.
In this study, we generated a human mAb (h4G3) against CLDN3 through the phage display technique using native antigens, stably CLDN3 overexpressing cell lines, and CLDN3-embedded lipoparticles. Unlike other murine or chimeric antibodies against CLDN3 [57
], h4G3 is a human mAb, which minimizes the immune response and improves safety and efficacy. The h4G3 could bind CLDN3 on the cell membrane, whereas it did not detect CLDN3 in an immunoblot or formaldehyde-fixed samples or recombinant CLDN3 protein (data not shown), indicating that h4G3 recognizes the native conformation of CLDN3.
The CLDNs have fairly conserved extracellular loops [11
] which regulate paracellular tightness and selective ion permeability [10
]. To avoid an off-target effect in targeting CLDN3, antibody specificity is important. The h4G3 demonstrated high specificity for CLDN3 among phylogenetically close members. In addition, h4G3 recognized mCLDN3, in which ECL2, the h4G3 binding region, differs from the human sequence by only 1 aa. The binding affinity of h4G3 for mCLDN3 was lower than hCLDN3. Especially, the association rate constant during the initial step (ka1
; A + B ↔ AB) and formation rate constant during the conformational change step (ka2
; AB ↔ AB*) for mCLDN3 were lower than hCLDN3 in the “one-to-one two-state” fitting. In the “one-to-two” fitting, the second association rate constant (ka2
) for mCLDN3 was also lower than hCLDN3. The region that differs in ECL2 between hCLDN3 (150V) and mCLDN3 (150L) has a strong hydrophobic area at the turn region among helices [60
]. In another structural study of platelet endothelial cell adhesion molecule-1 (PECAM-1) polymorphisms, the difference in side-chain size and orientation between valine and leucine affects the conformational structure of neighboring residues and leads to change in domain direction [61
]. Thus, the different sequences in ECL2 may alter the binding affinity of h4G3 between hCLDN3 and mCLDN3. Nevertheless, the affinity of h4G3 for mCLDN3 showed a sub-nanomolar affinity in the “one-to-two” model by the avidity effect. All these suggest that the efficacy and safety of h4G3 can be evaluated in a mouse model to assess the adverse effects of targeting CLDN3.
The CLDN3 is overexpressed in various carcinomas including breast, colorectal, gastric, pancreatic, prostate, and ovarian cancer [17
]. The ADCC activity of h4G3 according to CLDN3 expression levels was examined in representative cancer cell lines using NK-92MI-CD16a cells. Although ADCC was dependent on CLDN3 expression, ADCC was lower in the human pancreatic cell line AsPC-1 than in the gastric cancer cell line SNU-216 which has lower CLDN3 expression. Jochems et al. [62
] reported that AsPC-1 cells are more resistant to lysis from the natural killing of NK-92 cells expressing CD16a than other cell lines. Potential mechanisms underlying resistance to ADCC, such as low sensitivity to the perforin–granzyme system or impaired perforin binding to the target cell surface [63
], need to be examined to overcome resistance in cancer treatment.
The CLDN3 is also expressed in human normal tissues. For example, high levels of CLDN3 are observed in colon, rectum, thyroid, and salivary, and some levels of CLDN3 express in pancreas, prostate, liver, and kidney [21
]. Therefore, selective targeting of tumor tissues is necessary for the safety of CLDN3-targeted therapy in clinical application. Both CPE- and C-CPE-based cancer therapies have shown promising anti-tumor efficacy and provide proof of concept for CLDN3-targeted therapy [27
]. In intact TJs, which form paired TJ strands, CLDNs are sterically inaccessible to CPE because the CPE binding residues of CLDNs are blocked [67
]. The molecular weight of CPE is 35 kDa, and antibodies with a molecular weight of 150 kDa cannot easily access CLDNs in intact TJ strands. Among other antibodies targeting cell adhesion molecules, those targeting carcinoembryonic antigen (CEA) show good distinction between tumor and normal tissues, and radioactively labeled cytotoxic antibodies are well tolerated [68
]. An antibody targeting epithelial cell adhesion molecule (EpCAM) shows limited in vivo accessibility in normal tissues [71
]. Because CLDN3 is a component of TJs, its accessibility to antibodies may be much lower than that of CEA and EpCAM. The h4G3 conjugated to a fluorescent dye showed a lower distribution in normal tissues than in the tumor site in mice bearing xenograft tumors despite its ability to recognize mouse CLDN3. To assess the tissue cross-reactivity of h4G3, the immunohistochemistry of human normal tissues demonstrated that h4G3 mainly detected CLDN3 in glandular epithelium of gastrointestinal tract rather than smooth muscles, connective tissues or other organs (data not shown). These findings suggest that h4G3 could be applied to targeting CLDN3-positive cancer.
Although h4G3 accumulated in the tumor in mouse xenograft models, in vivo anti-tumor efficacy was not observed (Figure S9
). The ADCC depends on several parameters such as antibody-FcγR binding affinity, density and stability of the antigen on the surface of the target cell, and antibody–antigen affinity. The hIgG1 is the most commonly used isotype as a therapeutic antibody because of its excellent efficacy, such as ADCC, by binding to the FcγRs on effector cells such as NK cells, macrophages, and dendritic cells. The hIgG1 can bind to all activating mFcγRs and induces ADCC in mouse NK cells and mouse macrophages. Although the activation of cellular immune effector function by hIgG1 is strong, the in vitro ADCC activity of an anti-EGFR hIgG1 is three times lower than that of mIgG1 using mNK cells, and hIgG1 is not as effective as mIgG2a in an A431 xenograft model [72
]. Therefore, the function of hIgG1 might be underestimated in mouse models compared with humans. Regarding the antigen, CLDN3 is becoming accessible during epithelial tumorigenesis through misorientation of the cell division axis [7
]. However, not all CLDN3 become accessible and some still maintain tight junctions. Thus, the density of the antigen could be too low to cause the effector function. Because of the complicated factors involved in the immune reaction, the efficacy of antibodies targeting CLDN3 may be diminished in mouse xenograft models. To date, previously known anti-CLDN3 antibodies have not shown in vivo anti-cancer efficacy [57
]. The dual-targeting antibodies against CLDN3 and CLDN4 have shown anti-cancer efficacy in a preventive model in which the antibody was injected on day 0 after the inoculation of cancer cells [58
]. However, in patients, cancer has already been formed and has a different microenvironment. Therefore, a definite therapeutic efficacy is needed.
Recently, immunoconjugates have received attention as promising targeted therapies for cancer treatment. Antibodies that deliver drugs, such as pharmacologic agents, radioisotopes or photosensitive agents, to tumor cells can improve tumor-to-normal tissue selectivity and specificity in cancer treatment [47
]. In nuclear medicine, antibodies are especially applied to theranostics which provide both diagnosis and therapy [75
]. Furthermore, tumor immunotherapy by introducing a chimeric antigen receptor (CAR) into T cells or NK cells through genetic modification has been demonstrated as a promising strategy for cancer treatment [55
]. The CAR-engineered immune cell therapy has shown good responses in hematological malignancies and achieved successful trials in solid tumors. Especially, CLDN18.2-specific CAR-T cells generated using scFv from a developed antibody show promising therapeutic efficacy in gastric cancer [79
]. The h4G3-activated NK-92MI-CD16a cells function through a similar but indirect mechanism to that of CAR activation, indicating their potential application in CAR-NK cells. Altogether, the ability of h4G3 to specifically recognize CLDN3-positive tumor cells suggests that it could be applied to immunoconjugates and CAR immunotherapy (Figure 6
). For these reasons, we are currently testing the efficacy of h4G3 conjugated to radioisotopes such as 111
In or 188
Re as well as nanoparticles containing polydopamine (PDA) which generate heat when irradiated with near-infrared laser. In addition, we plan to develop CAR-NK cells using the 4G3 scFv sequence against CLDN3-positive solid tumors.
In summary, scFv phage display using CLDN3-expressing CHO-K1 cells and CLDN3-embedded lipoparticles as antigens was used to generate the anti-CLDN3 human mAb (h4G3). We showed that h4G3 was highly specific for CLDN3 in vitro and in vivo. Furthermore, h4G3 may be suitable for cancer imaging, therapeutic agent conjugation, and CAR immunotherapy for CLDN3-positive cancer.