Most malignant tumors are derived from epithelium, and claudin (CLDN)-3 and CLDN-4 are frequently overexpressed in such tumors. Although antibodies have potential in cancer diagnostics and therapy, development of antibodies against CLDNs has been difficult because the extracellular domains of CLDNs are too small and there is high homology among human, rat, and mouse sequences. Here, we created a monoclonal antibody that recognizes human CLDN-3 and CLDN-4 by immunizing rats with a plasmid vector encoding human CLDN-4. A hybridoma clone that produced a rat monoclonal antibody recognizing both CLDN-3 and -4 (clone 5A5) was obtained from a hybridoma screen by using CLDN-3– and -4–expressing cells; 5A5 did not bind to CLDN-1–, -2–, -5–, -6–, -7–, or -9–expressing cells. Fluorescence-conjugated 5A5 injected into xenograft mice bearing human cancer MKN74 or LoVo cells could visualize the tumor cells. The human-rat chimeric IgG1 monoclonal antibody (xi5A5) activated FcγRIIIa in the presence of CLDN-3– or -4–expressing cells, indicating that xi5A5 may exert antibody-dependent cellular cytotoxicity. Administration of xi5A5 attenuated tumor growth in xenograft mice bearing MKN74 or LoVo cells. These results suggest that 5A5 shows promise in the development of a diagnostic and therapeutic antibody for cancers.
Most malignant tumors are derived from epithelial tissues. A feature of normal epithelial cells is that they are polarized and spatially asymmetric (Nelson and Yeaman, 2001). In epithelial cells, the apical and basolateral membrane domains, which vary in protein and lipid content, are separated by specialized junctional complexes between adjacent cells, called tight junctions (TJs). TJs also act as a seal between adjacent cells to prevent large molecules from crossing the epithelial layer (Gumbiner, 1987). Malignant transformation of epithelium is accompanied by the loss of both cellular polarity and epithelial integrity leading to aberrant growth control, detachment of malignant cells from the primary tumor site, and formation of distant metastasis (Wodarz and Näthke, 2007). Dysregulation of these functions of TJ contributes to the initiation and progression of cancers (Martin and Jiang, 2009; Marchiando et al., 2010). However, the identification and characterization of novel targets for malignant tumors had been delayed due to our poor understanding of the biochemical structure of TJs.
Claudins (CLDNs) are tetratransmembrane proteins with molecular masses of approximately 23 kDa and are major components of TJ seals (Furuse et al., 1998). The CLDN multigene family consists of 27 members (Mineta et al., 2011). The expression profiles of the CLDNs vary among tissues, and CLDNs form homotypic and heterotypic strands on the lateral membrane. The CLDN strands between neighboring cells are associated with one another, forming TJ seals that prevent the free movement of lipids and proteins on the membrane between the apical and basolateral domains and the free movement of solutes across the epithelial cell sheets (Furuse and Tsukita, 2006; Anderson and Van Itallie, 2009).
A series of pathologic analyses revealed that numerous tumors, including gastric, colorectal, pancreatic, ovarian, breast, and prostate cancers, overexpress CLDN-3 and -4 (Tsukita et al., 2008; Turksen and Troy, 2011; Ding et al., 2013; Neesse et al., 2013). CLDN-3 and -4 are two of the known receptors for Clostridium perfringens enterotoxin (CPE), a food poison in humans (Sonoda et al., 1999; McClane and Chakrabarti, 2004). CPE has an N-terminal cytotoxic domain and a C-terminal receptor–binding domain (C-CPE) (Sonoda et al., 1999; McClane and Chakrabarti, 2004). CPE and C-CPE are the first identified CLDN-targeting toxin and CLDN binder, respectively. Intratumoral injection of CPE resulted in tumor suppression and necrosis in pancreatic or breast cancer cells without any observable adverse effects (Michl et al., 2001; Kominsky et al., 2004). Moreover, injection of fluorescence-dye conjugated C-CPE allowed visualization of pancreatic cancers in a xenograft model (Neesse et al., 2013). C-CPE also enhanced sensitivity to chemotherapy in ovarian cancers by modulating the TJ integrity in mice xenograft models of human cancer (Gao et al., 2011). Thus, CPE and C-CPE have contributed to the proof of concept for CLDN-3– and -4–targeted cancer diagnosis and therapy (Gao and McClane, 2012). However, CPE and C-CPE also bind to CLDN-6, -7, -8, and -14 (Fujita et al., 2000), and are immunogenic (Sugii, 1994; Suzuki et al., 2011). Overcoming these problems with the CPE technology is needed to develop CLDN-based cancer diagnosis and therapy.
Antibodies are potential therapeutics for CLDN-targeted cancer therapy because they have high antigen specificity, high stability, and moderate (or low) immunogenicity (Stockwin and Holmes, 2003; Espiritu et al., 2014). However, the extracellular loop domains of CLDNs are very small (first loop, approximately 50 amino acids; second loop, approximately 15 amino acids), the loop domains show high similarity among the human, mouse, and rat sequences (approximately 90% homology), and it is difficult to prepare recombinant proteins of CLDNs (currently, only CLDN-4 can be prepared as a recombinant protein) (Mitic et al., 2003; Evans et al., 2007). Therefore, attempts to produce anti-CLDN antibodies have had little success. In this study, we created a monoclonal antibody (mAb) to CLDN-3 and -4 by immunizing rats with a plasmid vector encoding CLDN-4; we then characterized this mAb.
Materials and Methods
BALB/c male mice (6 weeks old) and BALB/c Slc-nu/nu female mice (6 weeks old) were obtained from Shimizu Laboratory Supplies Co., Ltd. (Kyoto, Japan), and were housed in an environmentally controlled room at 23°C ± 1.5°C with a 12-hour light/dark cycle. The mice had free access to water and commercial chow (Type MF; Oriental Yeast, Tokyo, Japan). Experimental protocols involving mice were performed according to the ethics guidelines of the Osaka University Graduate School of Pharmaceutical Sciences.
Mouse fibroblasts stably expressing murine CLDN-3 or -4 (mCLDN-3/L or mCLDN-4/L cells, respectively) were kindly provided by Dr. S. Tsukita (Kyoto University, Kyoto, Japan). Jurkat/FcγRIIIa/NFAT-Luc cells, in which the luciferase gene, transcriptionally controlled by the activation of FcγRIIIa, was used to evaluate the activation of FcγRIIIa (Tada et al., 2014). Human gastric cancer MKN74 cells and human colorectal cancer LoVo cells were obtained from American Type Culture Collection (Manassas, VA). The mCLDN-3/L or mCLDN-4/L cells were maintained in modified Eagle’s medium supplemented with 10% fetal bovine serum (FBS) and 500 μg/ml G418 [(2R,3S,4R,5R,6S)-5-amino-6-[(1R,2S,3S,4R,6S)-4,6-diamino-3-[(2R,3R,4R,5R)-3,5-dihydroxy-5-methyl-4-methylaminooxan-2-yl]oxy-2-hydroxycyclohexyl]oxy-2-(1-hydroxyethyl)oxane-3,4-diol; Nacalai Tesque, Kyoto, Japan]. Jurkat/FcγRIIIa/NFAT-Luc cells were maintained in RPMI 1640 medium supplemented with 10% FBS, 500 μg/ml G418, and 500 μg/ml hygromycin B (Nacalai Tesque). MKN74 cells were maintained in RPMI 1640 supplemented with 10% FBS. LoVo cells were maintained in modified Ham’s F12 medium supplemented with 20% FBS. All cells were incubated in a 5% CO2 atmosphere at 37°C.
Establishment of Stable CLDN Transfectants.
Human CLDN-1, CLDN-2, CLDN-3, CLDN-4, CLDN-5, CLDN-6, CLDN-7, and CLDN-9 cDNA was amplified by polymerase chain reaction (PCR), and the resultant cDNAs were cloned into pcDNA3.1(−) (Invitrogen, Carlsbad, CA). The CLDN-expression vectors were then transfected into HT1080 cells by using the X-tremeGENE HP DNA transfection reagent (Roche Diagnostics, Basel, Switzerland), and G418-resistant clones were isolated, resulting in the isolation of stable transfectants of each CLDN. CLDNs/HT1080 cells were maintained in Dulbecco’s modified Eagle’s medium containing 10% FBS.
Isolation of Anti–CLDN-4 mAbs.
Six-week-old male Wistar rats were immunized with a eukaryotic expression vector encoding human CLDN-4 every 2 weeks for 2 months according to proprietary GENOVAC technology (GENOVAC GmbH, Freiburg, Germany). Lymphocytes were removed 7 days after the last immunization and fused with P3-UI cells in the presence of polyethylene glycol 1000, generating hybridoma cells. Hybridoma cells producing anti–CLDN-3 and anti–CLDN-4 mAbs were initially screened for the ability of their conditioned medium to bind to human CLDN-4 transiently transfected 293T cells but not to wild-type 293T cells; the hybridoma cells producing anti–CLDN-4 mAb were then subsequently selected mAbs that interacted with human CLDN-3/HT1080 cells, resulting in the isolation of a hybridoma that produced a mAb to both human CLDN-3 and -4 (clone 5A5). The immunoglobulin class and subclass of 5A5 was determined by using a rat immunoglobulin isotyping enzyme-linked immunosorbent assay kit (BD Biosciences, Franklin Lakes, NJ).
Surface Plasmon Resonance Analysis.
Human CLDN-4 recombinant proteins were purified by using a Sf-9 cell expression system as previously described (Uchida et al., 2010). To determine the binding kinetics of 5A5 to human CLDN-4 protein, we performed a surface plasmon resonance analysis by using a Biacore T200 instrument (GE Healthcare, Little Chalfont, UK) as previously described (Uchida et al., 2010). We used 10 mM HEPES, 150 mM NaCl, 3 mM EDTA, and 0.05% Tween 20 (pH 7.4) as the running buffer. Amine-coupling chemistry was used to immobilize the anti-rat mAb on a CM5 sensor chip (GE Healthcare) surface. The carboxymethyl surface of the CM5 chip was activated with a 1:1 ratio of 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride/N-hydroxysuccinimide. Anti-rat mAb was diluted to 20 μg/ml in 10 mM acetate (pH 4.5) and injected over the surface. Excess activated groups were blocked by 1 M ethanolamine (pH 8.5). Approximately 5000 RU of anti-rat mAb was immobilized. Then, the culture supernatant of the mAb-producing hybridoma cells was injected for 2 minutes. Human CLDN-4 protein was serially diluted to 10, 100, 200, 300, and 500 nM in the running buffer (10 mM HEPES, 150 mM NaCl, 3 mM EDTA, and 0.05% Tween 20, pH 7.4). Within a single binding cycle, human CLDN-4 protein was injected sequentially in order of increasing concentration over both the ligand and the reference surfaces. The reference surface, an unmodified flow cell, was used to correct for systematic noise and instrumental drift. The sensorgrams were globally fitted by using a 1:1 binding model to determine ka, kd, and KD values with Biacore T200 Evaluation Software.
Purification of the mAb.
Six-week-old female BALB/c Slc-nu/nu mice were intraperitoneally injected with the adjuvant pristane and 1 × 107 hybridoma cells producing mAb (clone 5A5). Ascites were collected and the mAb was purified by using Ab-Capcher ExTra (ProteNova, Kagawa, Japan) and a protein G column.
Flow Cytometric Analysis.
To analyze the CLDN binding of the mAb, cells were detached and incubated with the mAb (5 μg/ml) and then treated with secondary fluorescein-conjugated goat anti-rat IgG or goat anti-human IgG for 5A5 or human-rat chimeric 5A5, respectively. mAb-bound cells were analyzed by using a FACSCalibur flow cytometer (BD Biosciences).
Cells were lysed in lysis buffer [20 mM Tris, pH 7.4, 150 mM NaCl, 1% Triton X-100; protease inhibitor cocktail (Sigma-Aldrich, St. Louis, MO)], and cell lysates (10 μg protein) were subjected to 15% SDS-PAGE. The separated proteins were transferred onto a polyvinylidene difluoride membrane, and then immunoblotted with anti–CLDN-1, anti–CLDN-2, anti–CLDN-3, anti–CLDN-4, and anti–CLDN-5 antibody (Invitrogen), or with an anti–β-actin antibody (Sigma-Aldrich). The immunoreactive band was visualized by using ECL Western Blotting Detection Reagents (GE Healthcare).
Cancer cells were seeded onto a 96-well plate at 1 × 104 cells/well. After 24 hours of culture, the cells were treated with the mAb for 48 hours at the indicated concentrations. Cytotoxicity was assessed by using a WST-8 kit according to the manufacturer’s instructions (Nacalai Tesque).
Preparation of CF750-Conjugated mAbs.
Rat IgG and 5A5 were labeled with the fluorescent dye CF750 by using a XenoLight CF750 rapid antibody-labeling kit (Caliper Life Sciences, Inc., Hopkinton, MA), in accordance with the manufacturer’s instructions. The concentrations of conjugated mAbs were calculated by using the following equation provided in the manufacturer’s protocol:
5A5 Distribution in a Nude Mouse Xenograft Tumor Model.
Six-week-old female BALB/c Slc-nu/nu mice were injected subcutaneously into the right flank with cancer cells (1 × 107 cells) in 100 μl phosphate-buffered saline (PBS). After 5 weeks, mAbs labeled with CF750 were intraperitoneally injected into the mice at 20 μg/100 μl PBS per mouse. After 6, 24, 48, 72, or 96 hours, the mice were anesthetized with thiamylal sodium and set under a Maestro EX in vivo imaging system (version 2.10.0; Cambridge Research & Instrumentation Inc., Woburn, MA) to visualize CF750-conjugated mAbs. The imaging system was equipped with an excitation filter (wavelength 229–684 nm). Fluorescence was detected by use of a charge-coupled device camera equipped with a C-mount lens and a long-pass emission filter (745 nm). Spectral data “cubes” were created by the acquisition of a series of images obtained by using different wavelengths. The mice were then euthanized under anesthesia with thiamylal sodium and their lungs, liver, kidney, thyroid, and tumors were excised. The organs from each mouse were placed side by side and imaged at a 500-millisecond exposure time. In the spectral data cubes, each pixel is associated with a spectrum. Maestro software can be used to analyze these data; any autofluorescence can be identified, separated from the CF750 fluorescence, and removed. The signal-detected intensity was set at 5 × 10−5, and the resulting signals (counts) from each tissue were used to evaluate the mAb distributions. The levels of mAbs in each tissue, as percentages of the injected doses per gram of tissue, were then calculated.
Preparation of the Human-Rat Chimeric IgG1 mAb.
The cDNA encoding the heavy-chain and light-chain variable domains of 5A5 were amplified by PCR, and the PCR products were subcloned into the pFUSE-CHIg-hG1 and pFUSE2-CLIg-hk vectors (InVivoGen, San Diego, CA), respectively. Human-rat chimeric 5A5 (xi5A5) was prepared by using the FreeStyle MAX CHO Expression System (Life Technologies, Carlsbad, CA). Briefly, CHO-S cells were cotransfected with pFUSE-CHIg and pFUSE-CLIg of 5A5 by using the FreeStyle MAX Reagent, and then the transfected cells were cultured for 6 days in FreeStyle CHO Expression Medium. The conditioned medium was recovered and applied to a protein G column. The column was washed with 20 mM sodium phosphate buffer (pH 6.8), and the mAb was eluted by using 0.1 M glycine-HCl (pH 3.0). The eluted fraction containing the mAb was neutralized with 1 M Tris-HCl (pH 8.0), and then desalted by using a PD-10 column (GE Healthcare) with PBS as the exchange buffer. The concentration of the purified antibody was determined by measuring the absorbance at 280 nm.
Measurement of FcγRIIIa Activation.
Cells were seeded onto a 96-well plate at 1 × 104 cells/well. After 24 hours of culture, Jurkat/FcγRIIIa/NFAT-Luc cells (1 × 105 cells/well) suspended in Opti-MEM I Reduced Serum Medium were added to the 96-well plate in the presence of the mAbs. After a 5-hour incubation at 37°C in a 5% CO2 atmosphere, the luciferase activities were measured by using a commercially available luciferase assay kit as previously described (Tada et al., 2014).
In Vivo Antitumor Activity.
Six-week-old female BALB/c Slc-nu/nu mice were injected subcutaneously with cancer cells (1 × 107) in 100 μl PBS into the right flank on day 0. After the inoculation of the cells, the mAb was intraperitoneally injected into the xenograft tumor model at 1 mg/kg body weight twice a week. Body weight and tumor size were measured prior to each injection. The tumor volume was calculated by using the following equation: tumor volume = length × width2/2.
Data were analyzed by the t test or the one-way analysis of variance followed by post hoc pairwise comparison. The statistical significance for all comparisons was set at P < 0.05.
Creation of a mAb to CLDN-3 and CLDN-4.
To create a mAb that recognizes CLDN-3 and -4, rats were immunized with a plasmid vector encoding human CLDN-4. B cells isolated from immunized rats were fused with mouse myeloma cells. The resultant hybridomas were screened by using CLDN-3– or -4–expressing cells, resulting in the identification of a hybridoma that produces a mAb (clone 5A5) that recognizes both human CLDN-3 and -4. Clone 5A5 bound to CLDN-3 and CLDN-4 but not to CLDN-1, -2, -5, -6, -7, or -9 exogenously expressed in HT1080 cells (Fig. 1). The subclass of 5A5 was IgG2a. To our knowledge, of the CLDN family, only recombinant CLDN-4 protein can be prepared. Therefore, we carried out a surface plasmon resonance analysis to determine the kinetics of 5A5 to only CLDN-4. The binding kinetics of 5A5 to CLDN-4 were as follows: ka, 1.33 (1/Ms, × 104); kd, 0.58 (1/s, × 10−4); and KD, 4.35 nM.
To evaluate the diagnostic and therapeutic potential of 5A5 for cancer therapy, we selected human gastric cancer MKN74 cells and human colonic cancer LoVo cells because these cells are reported to express CLDN-3 and CLDN-4 mRNA and they have been used effectively in mice xenograft tumor models (http://www.lsbm.org/site_e/index.html). CLDN-1, -4, and -5 were detected in MKN74 cells, and CLDN-1, -2, -4, and -5 (faint signal) were observed in LoVo cells (Fig. 2A). However, CLDN-3 protein was not detected in these cells. Flow cytometric analysis showed that 5A5 bound to both MKN74 cells and LoVo cells (Fig. 2B). Therefore, we used MKN74 cells and LoVo cells to further characterize the antitumor activity of 5A5.
In Vivo Imaging in Xenograft Models.
To investigate whether 5A5 could be used as an in vivo targeting molecule for cancer, a fluorescent dye (XenoLight CF750) was conjugated with 5A5 via its reactive amine group because the fluorescent signal of CF750 can be noninvasively detected under an in vivo imaging system. CF750-conjugated 5A5 retained the ability to bind to CLDN-4–expressing cells (Supplemental Fig. 1). MKN74 cells were subcutaneously inoculated into BALB/c-nu/nu mice. Five weeks later, the conjugated 5A5 was injected into the xenograft tumor models at 20 μg per body. 5A5 distributed through the tumor tissue in a time-dependent manner, reaching a plateau at 72 hours after injection (Fig. 3); 5A5 was similarly distributed through the tumor tissue in mice bearing LoVo cells (Fig. 4).
Preparation of Human Chimeric IgG1.
Most therapeutic mAbs for cancer therapy are human IgG1 because human IgG1 has effector activities such as antibody-dependent cellular cytotoxicity (ADCC) (Houot et al., 2011; Shuptrine et al., 2012). Therefore, we next prepared a human-rat chimeric IgG1 mAb (xi5A5) by grafting the variable regions of the heavy and light chains of 5A5 (Fig. 5A). xi5A5 retained the CLDN specificity of the parental mAb (Supplemental Fig. 2). Activation of FcγRIIIa is involved in the activation of ADCC (Houot et al., 2011). Accordingly, we investigated the effects of xi5A5 on the activation of FcγRIIIa by using Jurkat/FcγRIIIa/NFAT-Luc reporter cells, in which luciferase expression was accompanied by activation of FcγRIIIa (Tada et al., 2014). Jurkat/FcγRIIIa/NFAT-Luc reporter cells were not activated by xi5A5 when cocultured with CLDN-1/HT1080 cells (Fig. 5B); however, the reporter cells were activated when cocultured with CLDN-3/HT1080 or CLDN-4/HT1080 cells (Fig. 5, C and D). The activation of the reporter cells by xi5A5 were also detected when cocultured with MKN74 or LoVo cells (Fig. 6), suggesting that xi5A5 activate FcγRIIIa in an antigen binding–dependent manner, and may exert ADCC activity against CLDN-3/CLDN-4–expressing tumor cells.
In Vivo Antitumor Activity in a Xenograft Model.
Finally, we investigated the antitumor activity of xi5A5 in mice bearing xenograft MKN74 or LoVo cells. After subcutaneous inoculation of mice with MKN74 or LoVo cells, xi5A5 was intraperitoneally injected twice a week at 1 mg/kg. xi5A5 suppressed tumor growth in both MKN74- and LoVo-bearing mice on days 7 and 14 postinoculation, respectively (Fig. 7, A and C). Administration of xi5A5 caused no apparent adverse effects or weight loss (Fig. 7, B and D). On the other hand, xi5A5 showed no cytotoxicity in MKN74 or LoVo cells in vitro even at 10 μg/ml (Supplemental Fig. 3). The antitumor activity of xi5A5, therefore, may be due to effector activity such as ADCC.
Recent advances in our understanding of the biochemical structures of TJs have provided us with new insights into targets for therapeutic intervention for cancer: the CLDNs (Morin, 2005; Tsukita et al., 2008; Lal-Nag and Morin, 2009; Turksen and Troy, 2011; Ding et al., 2013; Neesse et al., 2013). In this study, we created a mAb (clone 5A5) that recognizes human CLDN-3 and -4. Although 5A5 bound to CLDN-3– and -4–expressing cells, it could not detect CLDN-3 and -4 in an immunoblot (data not shown), indicating that 5A5 recognizes the intact forms of human CLDN-3 and -4. We also found that 5A5 could serve as a diagnostic probe for tumor tissues and that a human chimeric IgG1 mAb (xi5A5) had in vivo antitumor activity in mice bearing MKN74 or LoVo xenograft tumors.
The CLDNs constitute an integral protein family of 27 members, which share approximately 50% homology among their extracellular domains (Morita et al., 1999; Kato-Nakano et al., 2010; Mineta et al., 2011). They are ubiquitously expressed in various normal tissues, exhibiting barrier and fence functions in epithelia (Furuse and Tsukita, 2006; Anderson and Van Itallie, 2009). Therefore, to avoid adverse effects caused by cancer therapies targeting CLDNs, the specificity of the CLDN member and ligands for the type of cancer being treated are critical. CLDN-3 and -4 are receptors for CPE, but CPE also binds to CLDN-6, -7, -8, and -14 (Fujita et al., 2000). Although CPE- and C-CPE–based cancer therapy has established the proof of concept for CLDN-targeted cancer treatment, therapeutic interventions have been limited because of CLDN-member specificity and immunogenicity (Gao and McClane, 2012). Clone 5A5 showed binding specificity to CLDN-3 and -4. Thus, 5A5 may be superior to C-CPE as a CLDN-targeted ligand in terms of specificity and immunogenicity.
CLDN-3 and -4 are both highly expressed in lung, intestine, pancreas, and kidney; CLDN-3 is also highly expressed in liver (Morita et al., 1999; Rahner et al., 2001). The safety of CLDN-3– and -4–based cancer therapy is very important for its clinical application. We previously found that the liver and kidney are two major sites of C-CPE distribution and that intravenous injection of a C-CPE–fused toxin led to hepatic but not renal injury (Li et al., 2014). The C-CPE mutant, which lacks the CLDN-binding domain, distributed through the kidney with similar kinetics to those of C-CPE. CLDN-3 and -4 are expressed in the lateral membrane of the epithelium in Henle’s loop, the distal tubule, and the collecting duct (Balkovetz, 2009). The CLDN-3– and -4–based TJs regulate the paracellular transport of ions (Hou et al., 2010; Milatz et al., 2010), and the size of the CLDN-based pores in the TJs has been estimated to be 0.8 nm (Van Itallie and Anderson, 2011). Therefore, a CLDN-targeted toxin with a molecular mass of approximately 60 kDa, and also 5A5 with a molecular mass of 150 kDa, could not easily access the CLDNs embedded in the renal TJs. Indeed, intratumoral or intraperitoneal injection of CPE (35 kDa) attenuated tumor growth without apparent adverse effects on the xenograft tumor models (Michl et al., 2001; Kominsky et al., 2004; Santin et al., 2005).
CLDN-3–deficient mice are not available, but CLDN-4 knockout mice have been developed. Fujita et al. (2012) showed that CLDN-4–deficient mice develop normally until about 12 months of age, but then begin to exhibit urothelial hyperplasia and lethal hydronephrosis. CLDN-4 deficiency did not affect epithelial TJ integrity in the nephrons or urothelium. An increase in the proliferation of the pelvic and ureteral urothelial cells was thought to be associated with the urothelial hyperplasia and development of hydronephrosis. CLDN-4 heterodeficient mice showed none of these abnormal phenotypes. Although the underlying mechanism for renal dysfunction in CLDN-4 homodeficient mice is unclear, therapeutic and adverse effects might trade off in CLDN-3– and -4–targeted cancer therapy at least when using mAbs because of their size advantage.
Currently, conjugation of mAbs with cytotoxic drugs or radionuclides is receiving much attention as a new option for therapeutic mAbs (Lambert, 2005; Wu and Senter, 2005; Senter, 2009). It is critical for the conjugation that chemical modification of mAbs does not affect binding of mAbs to antigen and intracellular uptake of the drug-conjugated mAbs. Here, we chemically conjugated 5A5 with a fluorescent dye by reacting the dye with an amine group such as that of the lysine residue of 5A5. The conjugated 5A5 showed similar CLDN-binding tropism to that of the parental 5A5. Moreover, when MKN74 cells were treated with 5A5, 5A5 may appear to enter into the cells (Supplemental Fig. 4). These findings suggest that 5A5 could be developed as an antibody–drug conjugate because conjugation of drugs to a mAb takes place at a reactive amine group, such as that of lysine (Panowksi et al., 2014).
In conclusion, we developed a dual-specific mAb (clone 5A5) to human CLDN-3 and CLDN-4. Although the druggability of 5A5 requires the improvement through humanization and the safety evaluations in monkeys, 5A5 appears to be a promising lead mAb in the development of CLDN-based agents for the diagnosis and treatment of epithelium-derived malignant tumors.
The authors thank Dr. K. Endo and all of the members of their laboratory for their technical support, instruction, and useful comments.
Participated in research design: Li, Iida, Tada, Kondoh.
Conducted experiments: Li, Iida, Tada, Kawahigashi, Kimura.
Contributed new reagents or analytic tools: Tada, Watari, Yamashita, Ishii-Watabe, Uno, Fukasawa.
Performed data analysis: Li, Iida, Tada, Ishii-Watabe, Kuniyasu, Yagi, Kondoh.
Wrote or contributed to the writing of the manuscript: Li, Tada, Ishii-Watabe, Kondoh.
- Received May 23, 2014.
- Accepted August 8, 2014.
This research was supported by the Ministry of Health, Labour and Welfare of Japan [Health and Labour Sciences Research Grant]; the Ministry of Education, Culture, Sports, Science and Technology of Japan [Grant-in-Aid for Scientific Research 24390042]; and funds from the Adaptable and Seamless Technology Transfer Program through Target-Driven R&D, Japan Science and Technology Agency; Platform for Drug Discovery, Informatics, and Structural Life Science from the Ministry of Education, Culture, Sports, Science and Technology, Japan; the Takeda Science Foundation; the Nakatomi Foundation; and the Advanced Research for Medical Products Mining Programme of the National Institute of Biomedical Innovation.
- antibody-dependent cellular cytotoxicity
- C-terminal fragment of CPE
- Clostridium perfringens enterotoxin
- fetal bovine serum
- monoclonal antibody
- phosphate-buffered saline
- polymerase chain reaction
- tight junction
- Copyright © 2014 by The American Society for Pharmacology and Experimental Therapeutics