Development of a Novel Anti-CD44 Variant 8 Monoclonal Antibody C₄₄Mab-94 against Gastric Carcinomas

1. Introduction

Gastric cancer (GC) is the third leading cause of cancer-related deaths globally [1]. The GC incidence is higher in Eastern Asia than in Western countries [1]. The vast majority of GC are adenocarcinomas, which can be divided into intestinal-type gastric cancer (IGC), diffuse-type gastric cancer (DGC), and mixed histology according to the Lauren classification [2]. The World Health Organization classifies gastric adenocarcinomas into papillary, tubular, mucinous, and poorly cohesive carcinomas [3]. Furthermore, next-generation sequencing defined four molecular subtypes, including Epstein–Barr virus-positive, microsatellite instability, genomically stable, and chromosomally unstable types [4]. The analysis also revealed the alterations in the GC genome and provided treatment options with anti-human epidermal growth factor receptor 2 (HER2) therapy [5] or immune checkpoint inhibitor therapy [6]. However, the benefit of those therapies is limited to a small subset of patients. In patients with advanced GC, especially those with DGC, peritoneal metastasis and subsequent development of malignant ascites are the most frequent cause of death. Tanaka et al., therefore, performed a comprehensive multi-omic analysis of malignant ascitic samples and their corresponding tumor cell lines [7]. They identified the genomic alterations and significant amplification of known cancer driver genes, such as KRAS, FGFR2, MET, ERBB2, EGFR, MYC, CCND1, and CD44 in GC with peritoneal metastasis [7]. Among them, the cell surface antigens (FGFR2, MET, HER2, EGFR, and CD44) are potentially treatable with monoclonal antibody (mAb) therapy. Compared to the first four antigens, mAb therapy or diagnosis against CD44 has not been established.

CD44 plays important roles in the tumor progression and has various isoforms, which are generated by the alternative splicing of CD44 mRNA [8]. The mRNA of CD44 standard (CD44s) isoform is produced by constant region exons including the first five (1 to 5) and the last five (16 to 20) [9]. The mRNAs of CD44 variant (CD44v) isoform are produced by the assembling of variant exons (v1–v10) with the constant region exons of CD44s [10]. CD44s and CD44v receive the post-translational modifications, such as N-glycosylation and O-glycosylation. Therefore, the molecular weight of CD44s is reached 80~100 kDa, and CD44v has various molecular weights (100~250 kDa) due to a variety of glycosylation [11]. Both CD44s and CD44v (pan-CD44) can attach to hyaluronic acid (HA), which is important for cellular adhesion, homing, and motility [12].

CD44v is overexpressed in tumors [13] and promotes tumor malignant progression through the binding to growth factors, and the acquisition of invasiveness, stemness, and drug resistance [14,15,16]. These were mediated by the unique functions of the variant’s exon-encoded region. The v3-encoded region can recruit heparin-binding growth factors to their receptor, and promote the signal transduction [17,18]. The v6-encoded region forms a ternary complex with hepatocyte growth factor and its receptor MET, which is essential for the activation [19]. Furthermore, the v8–10-encoded region binds to and stabilizes a cystine–glutamate transporter (xCT), which enhances cystine uptake and glutathione synthesis [20]. The elevation of reduced glutathione (GSH) mediates the defense to reactive oxygen species (ROS) [20] and chemotherapeutic drugs [21]. The expression of CD44v8–10 is associated with the function of xCT and intracellular redox status, which is associated with the poor prognosis [22]. Therefore, the establishment of CD44v-specific mAbs are essential for CD44-targeting tumor diagnosis and therapy. However, the roles of the variant 8-encoded region in tumor development have not been fully elucidated.

Our group previously established an anti-pan-CD44 mAb, C₄₄Mab-5 (IgG₁, kappa) [23] using the Cell-Based Immunization and Screening (CBIS) method. Moreover, another anti-pan-CD44 mAb, C₄₄Mab-46 [24] was developed by immunizing mice with CD44v3–10 ectodomain. Both C₄₄Mab-5 and C₄₄Mab-46 have epitopes within the standard exon 2 and 5-encoding regions, respectively [25,26,27]. We further showed that both C₄₄Mab-5 and C₄₄Mab-46 are applicable to flow cytometry and immunohistochemical analyses in oral squamous cell carcinomas (OSCC) [23] and esophageal SCC [24]. Furthermore, we produced a class-switched and a defucosylated version of C₄₄Mab-5 (5-mG_2a-f) using fucosyltransferase 8-deficient ExpiCHO-S cells and evaluated the antitumor effects of 5-mG_2a-f in OSCC xenograft bearing mice [28]. We have developed various anti-CD44v mAbs, including anti-CD44v4 (C₄₄Mab-108) [29], anti-CD44v5 (C₄₄Mab-3) [30], anti-CD44v6 (C₄₄Mab-9) [31], anti-CD44v7/8 (C₄₄Mab-34) [32], and anti-CD44v9 (C₄₄Mab-1) [33].

In this study, we established a novel anti-CD44v8 mAb, C₄₄Mab-94 (IgG₁, kappa) using the CBIS method and evaluated its applications.

2. Materials and Methods

2.1. Cell Lines

The human OSCC cell line (HSC-3) and the human gastric cancer cell lines (MKN45 and NUGC-4) were obtained from the Japanese Collection of Research Bioresources (Osaka, Japan). The human pancreatic cancer cell line (PANC-1) was obtained from the Cell Resource Center for Biomedical Research Institute of Development, Aging, and Cancer at Tohoku University (Sendai, Japan). Chinese hamster ovary (CHO)-K1 and P3X63Ag8U.1 (P3U1; a mouse multiple myeloma) cell lines were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). HSC-3 was cultured in DMEM medium (Nacalai Tesque, Inc., Kyoto, Japan), supplemented with 100 μg/mL streptomycin, 100 U/mL penicillin, and 0.25 μg/mL amphotericin B (Nacalai Tesque, Inc.), and 10% (v/v) heat-inactivated fetal bovine serum (FBS; Thermo Fisher Scientific, Inc., Waltham, MA, USA). The cell lines (MKN45, NUGC-4, PANC-1, CHO-K1, and P3U1) were cultured in RPMI-1640 medium (Nacalai Tesque, Inc.), supplemented as indicated above. All cells were cultured using a humidified incubator at 37°C, in an atmosphere of 5% CO₂ and 95% air.

2.2. Construction of plasmid DNA and Establishment of Stable Transfectants

The cDNAs of CD44s and CD44v3–10 were obtained as described previously [23]. The cDNAs were cloned into pCAG-ssPA16 vectors with a signal sequence and N-terminal PA16 tag (GLEGGVAMPGAEDDVV) [23,34,35,36,37], which can be detected by NZ-1 mAb, (anti-human podoplanin [PDPN] mAb） [38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53]. N-terminal PA16-tagged CD44v3–10 deletion mutants (dN224, dN266, dN304, dN343, dN386, dN430, dN464, dN494, and dN562) were amplified using a HotStar HiFidelity Polymerase Kit (Qiagen Inc., Hilden, Germany), and subcloned into the pCAG-ssPA16 vector. The pCAG- ssPA16-CD44s, pCAG-ssPA16-CD44v3–10, and pCAG-ssPA16-CD44v3–10 deletion mutant vectors were transfected into CHO-K1 cells. The pCAG-ssPA16-CD44v3–10 vector was transfected into PANC-1 cells. The transfection was performed using a Neon transfection system (Thermo Fisher Scientific, Inc.). By the limiting dilution method, stable transfectants PANC-1/CD44v3–10, CHO/CD44s, CHO/CD44v3–10, and several deletion mutants of CHO/CD44v3–10 (dN224, dN266, dN304, dN343, dN386, dN430, dN464, dN494, and dN562) were finally established.

2.3. Production of Hybridomas

The 6-week-old female BALB/c mice (CLEA Japan, Tokyo, Japan) were intraperitoneally immunized with PANC-1/CD44v3–10 (1 × 10⁸ cells) and Imject Alum (Thermo Fisher Scientific Inc.). Additional immunizations of PANC-1/CD44v3–10 (1 × 10⁸ cells, three times) and a booster injection of PANC-1/CD44v3–10 (1 × 10⁸ cells) 2 days before the sacrifice were performed. Hybridomas were produced as described previously [30]. The supernatants, which are positive for CHO/CD44v3–10 cells and negative for CHO-K1 cells, were selected by flow cytometry, SA3800 Cell Analyzers (Sony Corp. Tokyo, Japan).

2.4. Enzyme-Linked Immunosorbent Assay (ELISA)

Four peptides, covering from v7, v8, and v9 regions of CD44v3–10, were obtained from Sigma-Aldrich Corp. (St. Louis, MO, USA).

The peptide sequences were as follows.

CD44p421–440 (GHQAGRRMDMDSSHSTTLQP); v7/v8,

CD44p431–450 (DSSHSTTLQPTANPNTGLVE); v8,

CD44p441–460 (TANPNTGLVEDLDRTGPLSM); v8,

CD44p451–470 (DLDRTGPLSMTTQQSNSQSF); v8/v9.

The peptides (10 µg/mL) were immobilized on 96-well immunoplates (Nunc Maxisorp; Thermo Fisher Scientific Inc). The blocking was performed with 1% (w/v) bovine serum albumin (BSA) in PBST. C₄₄Mab-94 (10 µg/mL) or blocking buffer was added to the peptides-coated wells. The detection was performed as described previously [30].

2.5. Flow Cytometry

In the CBIS screening and epitope mapping, the hybridoma supernatants were treated with CHO-K1, CHO/CD44v3–10, or CHO/CD44v3–10 deletion mutants. In the dose-dependent assay, CHO-K1, CHO/CD44s, CHO/CD44v3–10, MKN45, and NUGC-4 were incubated with C₄₄Mab-94, C₄₄Mab-46, or control blocking buffer (0.1% BSA in PBS). Then, the cells were treated with anti-mouse IgG conjugated with Alexa Fluor 488 (1:2000; Cell Signaling Technology, Inc., Danvers, MA, USA). The data were analyzed using the EC800 Cell Analyzer or the SA3800 Cell Analyzer (Sony Corp.).

2.6. Determination of Dissociation Constant (K_D) by Flow Cytometry

The diluted C₄₄Mab-94 (from 1300 to 0.08 nM) was suspended with CHO/CD44v3–10, MKN45, and NUGC-4 cells. Then, the cells were treated with anti-mouse IgG conjugated with Alexa Fluor 488 (1:200). Fluorescence data were analyzed and the apparent K_D was determined by the fitting binding isotherms to built-in one-site binding models of GraphPad Prism 8 (GraphPad Software, Inc., La Jolla, CA, USA).

2.7. Western Blot Analysis

Cell lysates were prepared as described previously [33], and were denatured in sodium dodecyl sulfate (SDS) sample buffer (Nacalai Tesque, Inc.). The 10 μg of proteins were subjected to electrophoresis using polyacrylamide gels (5–20%; FUJIFILM Wako Pure Chemical Corporation, Osaka, Japan) and transferred onto polyvinylidene difluoride membranes (Merck KGaA, Darmstadt, Germany). The membranes were blocked with 4% skim milk (Nacalai Tesque, Inc.) in PBST, and were incubated with 10 μg/mL of C₄₄Mab-94, 10 μg/mL of C₄₄Mab-46, or 0.5 μg/mL of an anti-β-actin mAb (clone AC-15; Sigma-Aldrich Corp.). The detection was performed as described previously [31].

2.8. Immunohistochemical Analysis

Formalin-fixed paraffin-embedded (FFPE) gastric carcinoma (BS01012e and BS01011b) and OSCC (OR601c) tissue arrays were obtained from US Biomax Inc. (Rockville, MD, USA). The tissue arrays were autoclaved in citrate buffer (pH 6.0; Nichirei Biosciences, Inc., Tokyo, Japan) for 20 min. The blocking was performed using SuperBlock T20 (Thermo Fisher Scientific, Inc.). The sections were incubated with C₄₄Mab-94 (5 μg/mL) and C₄₄Mab-46 (5 μg/mL). The detection was performed as described previously [30].

3. Results

3.1. Establishment of an Anti-CD44v8 mAb, C₄₄Mab-94

We previously used CHO/CD44v3–10 cells as an immunogen and generated anti-CD44 mAbs, including C₄₄Mab-5 (pan-CD44), C₄₄Mab-3 (v5) [30], C₄₄Mab-9 (v6) [31], and C₄₄Mab-1 (v9) [33]. In this study, we newly established a stable transfectant (PANC-1/CD44v3–10 cells) as another immunogen (Figure 1A). Mice were immunized with PANC-1/CD44v3–10 cells (Figure 1B), and hybridomas were seeded in 96-well plates (Figure 1C). The supernatants, which are positive for CHO/CD44v3–10 cells and negative for CHO-K1 cells, were selected using flow cytometry (Figure 1D). After cloning, anti-CD44-mAb-producing clones were finally established. Among established clones, we focused on C₄₄Mab-94 (IgG₁, kappa), and the epitopes were determined by flow cytometry and/or ELISA (Figure 1E).

To determine the C₄₄Mab-94 epitope, we examined the reactivity to CHO/CD44v3–10 and the N-terminal CD44v3–10 deletion mutants (dN224, dN266, dN304, dN343, dN386, dN430, dN464, dN494, and dN562)-expressed CHO-K1 cells by flow cytometry (Figure 2A). As shown in Figure 2B, C₄₄Mab-94 reacted with dN224, dN266, dN304, dN343, dN386, dN430, and CD44v3–10. In contrast, the reactivity completely disappeared in dN464, dN494, and dN562. Because CD44v3–10 and the deletion mutants possess PA16 tag at the N-terminus, we could confirm all expression on the cell surface by anti-PA16 tag mAb, NZ-1 (Figure 2C). These results suggest that C₄₄Mab-94 recognizes the v8-encoding sequence.

To further assess the C₄₄Mab-94 epitope, we performed ELISA using synthetic peptides from the v7- to v9-encoded sequences. As shown in Figure 3, C₄₄Mab-94 reacted with CD44p431–450 (DSSHSTTLQPTANPNTGLVE, v8 region), but not another v8 region (CD44p441–460), v7/v8 region (CD44p421–440), or v8/v9 region (CD44p451–470). These results indicated that C₄₄Mab-94 recognizes the CD44 variant-8-encoded sequence, but not the border sequence between v7 and v8, or v8 and v9.

3.2. Flow Cytometric Analysis of C₄₄Mab-94 against CD44-Expressing Cells

We next examined the reactivity of C₄₄Mab-94 against CHO/CD44v3–10 and CHO/CD44s cells by flow cytometry. C₄₄Mab-94 recognized CHO/CD44v3–10 cells in a dose-dependent manner (Figure 4A). In contrast, C₄₄Mab-94 recognized neither CHO/CD44s (Figure 4B) nor CHO-K1 (Figure 4C) cells. We confirmed that a pan-CD44 mAb, C₄₄Mab-46 [24], recognized both CHO/CD44s and CHO/CD44v3–10 cells (Supplementary Figure S1A and B, respectively), but not CHO-K1 cells (Supplementary Figure S1C). Furthermore, C₄₄Mab-94 also recognized the OSCC cell line (HSC-3) and GC cell lines (MKN45, and NUGC-4) in a dose-dependent manner (Figure 4D, E, and F, respectively).

We measured the apparent binding affinity of C₄₄Mab-94 to CHO/CD44v3–10, MKN45, and NUGC-4 cells by flow cytometry. The dissociation constant (K_D) of C₄₄Mab-94 for CHO/CD44v3–10, MKN45, and NUGC-4 was 2.8 × 10⁻⁷ M, 3.0 × 10⁻⁷ M, and 5.7 × 10⁻⁷ M, respectively.

3.4. Western Blot Analysis

We next conducted western blot analysis to assess the sensitivity of C₄₄Mab-94. Total cell lysates from CHO-K1, CHO/CD44s, CHO/CD44v3–10, HSC-3, MKN45, and NUGC-4 were examined. As shown in Figure 5A, C₄₄Mab-94 detected CD44v3–10 at more than 180-kDa and ~75-kDa bands mainly. Furthermore, C₄₄Mab-94 detected endogenous CD44v8 at more than 100-kDa bands in HSC-3, MKN45, and NUGC-4 cells. An anti-pan-CD44 mAb, C₄₄Mab-46, recognized the lysates from both CHO/CD44s (~75 kDa) and CHO/CD44v3–10 (>180 kDa) (Figure 5B). Although C₄₄Mab-46 strongly recognized the lysates from NUGC-4, the reactivity to the HSC-3 and MKN45 lysates was weak. These results indicated that C₄₄Mab-94 specifically detects exogenous CD44v3–10 and endogenous CD44v8.

3.5. Immunohistochemical Analysis Using C₄₄Mab-94 against Tumor Tissues

We next investigated whether C₄₄Mab-94 could be applied to immunohistochemical analysis using FFPE sections. We first examined the reactivity of C₄₄Mab-94 in the OSCC tissue array because the type was revealed as the second highest CD44-positive cancer type in the Pan-Cancer Atlas [55]. As shown in Supplementary Figure S2A and B, the membranous staining in OSCC was observed by C₄₄Mab-94 and C₄₄Mab-46. In a stromal invaded OSCC section, C₄₄Mab-94 strongly stained invaded OSCC and could clearly distinguish tumor cells from stromal tissues (Supplementary Figure S2C). In contrast, C₄₄Mab-46 stained both invaded OSCC and surrounding stroma cells (Supplementary Figure S2D). Supplementary Table S1 summarized the result of OSCC tissue staining.

We next stained the GC tissue array (BS01011b) using C₄₄Mab-94 and C₄₄Mab-46. C₄₄Mab-94 exhibited membranous staining in IGC (Figure 6A). C₄₄Mab-46 also stained the same type of cancer cells (Figure 6B). Furthermore, membranous and cytoplasmic staining by C₄₄Mab-94 and C₄₄Mab-46 was observed in stromal-invaded tumor cells (Figure 6C,D). In DGC (Figure 6E,F), diffusely spread tumor cells were strongly stained by both C₄₄Mab-94 and C₄₄Mab-46. In contrast, both C₄₄Mab-94 and C₄₄Mab-46 did not stain the ductal epithelial structure of IGC (Figure 6G,H). Additionally, stromal staining by C₄₄Mab-46 was observed in the tissue (Figure 6H).

We summarized the immunohistochemical analysis of GC in Table 1; C₄₄Mab-94 stained 28 out of 72 cases (39%) of GC. A similar staining was also observed in another tissue array (BS01012e, Supplementary Figure S3). We summarized the data of immunohistochemical analysis in Supplementary Table S2. These results indicated that C₄₄Mab-94 is useful to detect CD44v8 in immunohistochemical analysis of FFPE tumor sections.

4. Discussion

The VFF series anti-CD44v mAbs were previously established by the immunization of bacterial-expressed CD44v3–10 and glutathione S-transferase fusion protein [56,57]. The clones, VFF-8 (v5), VFF-18 (v6), VFF-9 (v7), VFF-18(v7/8), and VFF14 (v10) have been used for various applications [58]. Furthermore, VFF18 was humanized as BIWA-4 [59], and developed to bivatuzumab-mertansine, an antibody-drug conjugate (ADC), for clinical trials [60,61]. An anti-CD44v3 mAb (clone 3G5) [62] and an anti-CD44v9 mAb (clone RV3) [20] were also developed and widely used for researches. However, a CD44v8-specific mAb has not been developed.

In this study, we developed a novel anti-CD44v8 mAb, C₄₄Mab-94 using the CBIS method (Figure 1). We determined the epitope as a v8-encoded region using deletion mutants of CD44 (Figure 2), and synthetic peptides (Figure 3). We have established anti-CD44 mAbs using CHO/CD44v3–10 [23,30,31,33], PANC-1/CD44v3–10 (in this study), or purified CD44v3–10 ectodomain [24,32] as antigens. We listed them in our original "Antibody Bank" (see Supplementary Materials). However, clones which recognize the v8-encoded region were rare, suggesting that the region has low antigenicity and/or locates the inside of CD44v3–10 protein. Although the affinity of C₄₄Mab-94 is low against target cells, C₄₄Mab-94 can be applied to various applications, including flow cytometry (Figure 4), western blotting (Figure 5), and immunohistochemistry (Figure 6).

We confirmed that C₄₄Mab-94 recognizes a synthetic peptide of v8-encoded region (DSSHSTTLQPTANPNTGLVE), but not border regions (v7/v8 and v8/v9) by ELISA (Figure 3). The epitope region possesses multiple confirmed and predicted O-glycosylation sites [63]. C₄₄Mab-94 recognized a ~75-kDa band in the lysate of CHO/CD44v3–10 (Figure 5A), which is similar to the predicted molecular size from the amino acids of CD44v3–10. Therefore, C₄₄Mab-94 could recognize CD44v3–10 regardless of the glycosylation. The detailed epitope analysis and the influence of glycosylation on C₄₄Mab-94 recognition are requied in future studies.

In a GC cell line, the major transcripts of CD44v, including CD44v3, 8–10, CD44v6–10, CD44v8–10, and CD44v3, 8 were identified [54] (Figure 1A). C₄₄Mab-94 can cover all products of the transcripts, and detect the broad CD44v-expressing GC. Since CD44 v8–10 plays critical roles in the regulation of ROS defense and GC progression [20], an anti-CD44v9 mAb (clone RV3) was mainly used in immunohistochemistry to date. Several studies revealed that CD44v9 is a predictive marker for the recurrence of GC [64] and a biomarker for GC patient selection and efficacy of xCT inhibitors, sulfasalazine [65]. Further investigations are required to reveal the relationship between CD44v8 expression and clinical factors using C₄₄Mab-94. Additionally, C₄₄Mab-94 recognized both IGC (Figure 6A) and DGC (Figure 6E) in immunohistochemistry. It is worthwhile to investigate whether CD44v8 is expressed in a specific molecular subtype of GC [4] in the future study.

A comprehensive analysis of malignant ascites identified the amplifications of cancer driver genes including CD44 [7]. Although the expression pattern of CD44v is not identified, CD44v8 is thought to be an important target for mAb therapy due to the commonly included region in GC [54]. We have shown the antitumor activity using class-switched and defucosylated IgG_2a recombinant mAbs [28,66,67,68,69,70,71,72]. The defucosylated IgG_2a mAbs can be produced by CHO-K1 lacking fucosyltransferases 8, and exhibited potent ADCC activity in vitro, and suppressed the growth of xenograft [28,66,67,68,69,70,71,72]. Therefore, the production of defucosylated C₄₄Mab-94 is one of the strategies to evaluate the antitumor effect on GC with peritoneal metastasis in the preclinical model.

Clinical applications of a humanized anti-CD44v6 mAb (BIWA-4) bivatuzumab−mertansine drug conjugate to solid tumors failed because of the skin toxicities [60,61]. The accumulation of mertansine drug was thought to be a cause of the toxicity [60,61]. Human acute myeloid leukemia (AML) cells also express high levels of CD44 mRNA due to suppression of CpG islands methylation in the promoter [73]. Furthermore, higher expression of CD44v6 was observed in AML patients with FLT3 or DNMT3A mutations. Therefore, a mutated version of BIWA-4, called BIWA-8 was engineered to develop chimeric antigen receptors (CARs) for AML. The CD44v6 CAR-T cells exhibited potent anti-leukemic effects [73], indicating that CD44v6 is a rational target of CAR-T therapy for AML harboring FLT3 or DNMT3A mutations. Additionally, the CD44v6 CAR-T also showed an antitumor effect in lung and ovarian cancer xenograft models [74], which is expected for a wider development toward solid tumors.

Because CD44 mRNA is elevated in AML, other CD44 variants could be transcribed in AML. Furthermore, CD44v8–10 was elevated during chronic myeloid leukemia (CML) progression from chronic phase to blast crisis in a humanized mouse model, which is required for the maintenance of stemness of CML [75]. Therefore, we will investigate the reactivity of C₄₄Mab-94 against hematopoietic malignancy. Further studies are required to investigate the selective expression of CD44v8 in leukemia cells, but not in hematopoietic stem cells to certify its safety as a CAR-T antigen.

In this study, we used tumor cell-expressed CD44v3–10 as an immunogen. This strategy is important for the establishment of cancer-specific mAbs (CasMabs). We previously developed PDPN-targeting CasMabs [76,77,78,79] and podocalyxin-targeting CasMabs [80], which recognize cancer-type aberrant glycosylation of the targets [81]. Anti-PDPN-CasMabs are currently applied to CAR-T therapy in preclinical models [48,82,83]. For CasMab development, we should do further screening of our established anti-CD44 mAbs by comparing the reactivity against normal cells. Anti-CD44 CasMabs could be applicable for designing the modalities including ADCs and CAR-T.