Article Text
PDF
Abstract
Background Ovarian cancer remains a formidable challenge in oncology, necessitating innovative therapeutic approaches. Claudin-6 (CLDN6), a member of the tight junction molecule CLDN family, exhibits negligible expression in healthy tissues but displays aberrant upregulation in various malignancies, including ovarian cancer. Although several therapeutic modalities targeting CLDN6 are currently under investigation, there is still a need for more potent therapeutic options. While T-cell engagers (TCEs) hold substantial promise as potent immunotherapeutic agents, their current efficacy and safety in terms of target antigen selection and T-cell exhaustion due to only CD3 stimulation without co-stimulation must be improved, particularly against solid tumors. To provide an efficacious treatment option for ovarian cancer, we generated SAIL66, a tri-specific antibody against CLDN6/CD3/CD137.
Methods Using our proprietary next-generation TCE technology (Dual-Ig), SAIL66 was designed to bind to CLDN6 with one Fab and CD3/CD137 with the other, thereby activating T cells through CD3 activation and CD137 co-stimulation. The preclinical characterization of SAIL66 was performed in a series of in vitro and in vivo studies which included comparisons to a conventional TCE targeting CLDN6 and CD3.
Results Despite the high similarity between CLDN6 and other CLDN family members, SAIL66 demonstrated high specificity for CLDN6, reducing the risk of off-target toxicity. In an in vitro co-culture assay with CLDN6-positive cancer cells, we confirmed that SAIL66 strongly activated the CD137 signal in the Jurkat reporter system, and preferentially induced activation of both CD4+ and CD8+ T cells isolated from human peripheral blood mononuclear cells compared to conventional TCEs. In vivo studies demonstrated that SAIL66 led to a more pronounced increase in intratumor T-cell infiltration and a decrease in exhausted T cells compared with conventional CLDN6 TCE by contribution of CD137 co-stimulation, resulting in better antitumor efficacy in tumor-bearing mouse models.
Conclusion Our data demonstrate that SAIL66, designed to engage CLDN6, CD3, and CD137, has the potential to enhance antitumor activity and provide a potent therapeutic option for patients with ovarian and other solid tumors expressing CLDN6. Clinical trials are currently underway to evaluate the safety and efficacy of SAIL66.
- T cell
- Bispecific T cell engager - BiTE
- Immunotherapy
- Ovarian cancer
- Co-stimulatory molecules
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
While claudin-6 (CLDN6) has attracted attention as a promising target for cancer immunotherapy, there is a demand for the development of novel off-the-shelf immunotherapy using powerful modalities against CLDN6.
Conventional T-cell engagers (TCEs), which only activate T-cell receptor (TCR)/CD3 signals without co-stimulation, face challenges in terms of efficacious T-cell exhaustion can lead to loss of efficacy, particularly in the clinical treatment of solid tumors.
WHAT THIS STUDY ADDS
SAIL66 is a novel tri-specific TCE, designed to bind to CLDN6 with one Fab arm and to both CD3 and CD137 with the other.
In vitro characterization of SAIL66 demonstrates that it competitively binds to CD3 and CD137, with stronger binding to CD137, and specifically binds to CLDN6 without binding to CLDN3/4/9.
Unlike conventional CLDN6 TCE, SAIL66 activates T cells by both triggering CD3 signaling and co-stimulating CD137.
In vivo tumor models demonstrate that SAIL66 markedly increases intratumor T-cell infiltration and decreases the number of exhausted T cells, leading to enhanced antitumor efficacy compared with conventional TCE.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
The preclinical demonstration of superior antitumor activity and intratumor pharmacodynamics of SAIL66, compared with conventional CLDN6 TCE, supports its potential as a next-generation TCE that incorporates CD137 co-stimulation in addition to CD3 signaling.
These findings warrant further evaluation in ongoing clinical trials (NCT05735366).
Background
Epithelial ovarian cancer is a highly lethal gynecological malignancy with a poor 5-year survival rate of less than 25%,1 despite the current standard treatment involving surgery and chemotherapy. While immune checkpoint inhibitors (ICIs) exhibit efficacy across various cancers, their effectiveness is limited in cold tumors with few tumor-infiltrating lymphocytes,2 such as high-grade serous ovarian carcinoma where clinical studies of ICI have been disappointing.3 4 Thus, a more potent therapeutic option is still needed.
Claudins (CLDNs) are major components of tight junction molecules and play critical roles in regulating the permeability, barrier function, and cell polarity of epithelial layers. While most CLDN family members show broad expression in multiple adult tissues, Claudin-6 (CLDN6) is selectively expressed in tissues during embryonic development and undetectable in normal adult tissues.5 Moreover, the dysregulated expression of CLDN6 has been reported in multiple malignant tissues5–8 and was associated with poor prognosis.6 9 10 Due to its frequent expression in serous ovarian carcinomas, CLDN6 has appeared as an ideal target for tumor-targeting therapeutics for ovarian cancer. So far, several CLDN6 targeted therapies using different modalities, including T-cell engagers (TCEs),11 autologous chimeric antigen receptor T cells (CAR-T),12 antibody drug conjugates (ADCs),13 and bivalent antibodies which induce antibody-dependent cellular cytotoxicity,14 have been reported. Recently, BNT211, which comprises CAR-T targeting CLDN6 and CAR-T-amplifying RNA vaccine, demonstrated an antitumor response in testicular and patients with ovarian cancer in a phase I dose escalation study, suggesting that CLDN6 is a promising tumor-associated antigen when it is applied to powerful immunotherapeutic modalities.12 However, due to several challenges associated with CAR-T therapy, including toxicities, cost, and cumbersome and time-consuming manufacturing processes, the development of novel off-the-shelf immunotherapy remains an urgent need.
TCEs, which can bind CD3 on T cells together with an antigen abnormally expressed on cancer cells, have gained attention as potential cancer treatment agents.15–17 Their ability to recruit polyclonal-T cells into the tumor and direct their cytotoxic activity to cancer cells, independent of tumor immunogenicity and major histocompatibility complex class I-mediated antigen presentation, makes them a promising for low immunogenic cancers unresponsive to ICIs. Recently, promising early clinical activity has been reported for TCEs targeting several tumor antigens in solid tumors.18–20 Additionally, the approval of the tebentafusp, a gp100 peptide-HLA-directed TCE, for unresectable or metastatic uveal melanoma, known as a non-immunogenic tumor with low T-cell infiltration, suggests TCEs could overcome resistance to ICIs.21 22
Despite their potential benefits for cancer treatment, TCE therapeutic platforms still face challenges when applied to solid tumors. One fundamental problem is “on-target/off-tumor” toxicity, caused by the tumor antigen expression in normal tissues. As demonstrated by dose-limiting toxicity in clinical studies with anti-EpCAM TCEs, on-target toxicity can prevent reaching therapeutic doses.23 24 To minimize this, it is crucial to use tumor-specific antigens and ensure high antibody selectivity and specificity to prevent unwanted T-cell activation and systemic inflammation.
Another challenge is T-cell dysfunction caused by chronic exposure to TCE. A robust and long-term T-cell response requires TCR/CD3 signaling (signal 1) and co-stimulation signaling (signal 2). However, conventional TCEs only activate TCR/CD3 signals without co-stimulation, potentially leading to T-cell exhaustion and a loss of efficacy. Co-stimulation of CD137 (also known as 4-1BB) on T cells enhances interleukin (IL)-2 and interferon gamma (IFN-γ) production, T-cell proliferation, memory T-cell formation, and protection of T cells from apoptosis.25 The success of the second-generation CAR-T cell therapy using the intracellular co-stimulatory CD137 domain demonstrates the importance of CD137 co-stimulation in cancer immunotherapy.12 While many CD137 agonistic antibodies have been tested in clinical studies, systemic CD137 agonist activity induces hepatotoxicity, hindering their development.26 On the other hand, CD137 expression is induced by signal 1 mediated T-cell activation. Based on this, combining a tumor-targeting TCE and CD137 agonist is demonstrated to be an effective strategy. In fact, two such combination trials are being conducted clinically (NCT04826003 and NCT04077723).
In light of this, we generated SAIL66, a novel tri-specific TCE against CLDN6/CD3/CD137. A risk associated with the tri-specific targeting of tumor antigen/CD3/CD137 could potentially be systemic T-cell activation, caused by the simultaneous binding to both CD3 and CD137 on T cells outside of the tumor microenvironment. By applying our proprietary next-generation TCE technology (Dual-Ig), which is designed to competitively bind to either CD3 or CD137,27 SAIL66 activates CD3 and CD137-mediated immune activation exclusively within CLDN6-positive tumors and provides more potent antitumor activity. Our in vitro and in vivo studies demonstrate the potential of SAIL66 as a promising therapeutic option for CLDN6-positive cancers, including ovarian cancer.
Materials and methods
Reporter gene assay
For the evaluation of TCR/CD3 mediated-T cell activation, NFAT-luc2 Jurkat cells (Promega, J1601) were co-incubated with cancer cells at a ratio of 5:1. On the other hand, for the assessment of 4-1BB dependent-T cell activation, NF-kB-luc2 4-1BB Jurkat cells (Promega, J2332) were co-incubated with cancer cells at a ratio of 2:1. Following the addition of serially diluted antibodies to each well, the cells were further incubated for either 24 hours or 6 hours, respectively. Subsequently, the Bio-Glo Luciferase Assay Reagent (Promega, G7941) was added to each well and luciferase activity was measured using the EnVision multimode plate reader (PerkinElmer Japan). The resulting data were normalized to the values of untreated control samples and fitted to a logistic curve using the GraphPad Prism software.
In vitro cytotoxicity and T-cell activation assay
Cryopreserved human peripheral blood mononuclear cells (PBMCs) derived from healthy donors (Cellular Technology Limited, CTL-UP1) were incubated with the target cells in an Effector cell / Target cell (E/T) ratio of 10:1 in the presence of SAIL66. After that, cytokine measurements were performed 24 hours later, cytotoxicity assays were performed 48 hours later, and T-cell activation marker measurements were performed 72 hours after the reaction, respectively, as follows: To analyze the cytotoxicity, Lactate dehydrogenase (LDH) activity in the culture supernatant was measured using an LDH Cytotoxicity Detection Kit (Takara Bio, MK401) according to the manufacturer’s protocol. Spontaneous LDH release from target cells without treatment was used as a low control, and the maximum releasable LDH activity was determined by adding Triton X-100 to target cells. Cytokine levels in the culture supernatant were measured using the Human Th1/Th2 CBA Kit II (BD Biosciences, 551809) according to the manufacturer’s protocol. The values below the limit of quantification were handled as zero and the outliers (%CV>100) were removed from analysis. For flow cytometry analysis, the remaining PBMCs were collected and stained with fixable viability dye eFluor 780 (Thermo Fisher Scientific, 65-0865-14), and antibodies to human CD3, CD4, CD8, CD25, and CD69 (online supplemental tableS1). The data were obtained using FACSLyric or FACSVerse (BD Biosciences) and analyzed using FlowJo V.10 (BD Biosciences).
Supplemental material
In vitro real-time cell growth monitoring by xCELLigence
Colon38/hCLDN6 cells were cultured for 1 day, then 5-fold or 60-fold more mouse T cells derived from splenocytes by MACS separation (Miltenyi Biotec) were added in the presence of 1, 10 nmol/L concentrations of SAIL66 or CLDN6 TCE. Cell growth inhibition was monitored using an xCELLigence Real-Time Cell Analysis system (ACEA Biosciences). The percentage of cell growth inhibition was calculated as follows: Cell growth inhibition (%)=(100–(A/B)×100), where A was the cell index of each well with antibody, and B was the mean cell index of wells without antibody. Data represents the mean (n=2).
In vivo antitumor efficacy study
Human CD3 transgenic mice (hCD3 tgm) were generated by replacing the three components of the mouse CD3 complex, Cd3ε, Cd3δ, and Cd3γ, with their human counterparts.28 Human CD137 knock-in mice were generated by replacing mouse CD137 with its human counterpart.26 Human CD3 and CD137 transduced mice (hCD3/hCD137 KI) were generated by crossbreeding hCD3 tgm and hCD137 knock-in mice. The LLC1/hCLDN6 cells were inoculated subcutaneously into these mice (1×106 cells/mouse). After tumor establishment, animals were randomized and 1 mg/kg SAIL66 and CLDN6 TCE were intravenously administered. Tumor volume was measured twice weekly using the formula: (length×width2/2).
Humanized NOG (huNOG) mice were generated by NOG mice (In-Vivo Science) and intravenously injecting human hematopoietic stem cells from cord blood (Lonza) at 1×105 cells/mouse. The establishment of huNOG mice was confirmed by assessing the presence of human CD45-positive and CD3-positive cells in the blood. OV-90, NIH:OVCAR-3, or NCI-H1435 cancer cells were implanted subcutaneously with Matrigel Basement Membrane Matrix (Corning). After tumor establishment, SAIL66 or CLDN6 TCE were intravenously administered, and tumor volume, body weight, and human cytokine levels were measured. In premedication experiments, 33 mg/kg dexamethasone (Dexart, Fuji Pharma) was intraperitoneally administered 1 and 24 hours, or 1-hour prior to antibody injection. Cytokines and chemokines were measured using the Human Cytokine/Chemokine Magnetic Bead Panel (Merck, HCYTMAG-60K-PX38) and DropArray technology (Curiox Biosystems).
For the cancer-disseminated model, NOD-SCID mice (CLEA Japan) received intraperitoneal OV-90 cell implants (5×106 cells/mouse). Three days later, randomization was performed based on body weight. Human T cells were amplified from human PBMCs using Dynabead Human T-Activator CD3/CD28 (Life Technologies, DB11131) and intravenously injected (3×107 cells/mouse) as effector cells, followed by intravenous administration of 5 mg/kg SAIL66 or vehicle 3–5 hours later. Survival ratios were recorded based on general conditions after tumor inoculation.
Pharmacodynamic studies
For flow cytometry analysis, dissected tumors were dissociated using a Tumor Dissociation Kit, human (Miltenyi Biotec, 130-095-929) using a gentleMACS Octo Dissociator (Miltenyi Biotec). The cell suspensions were incubated with Zombie Aqua Fixable Viability Kit, and the antibodies to each antigen (online supplemental table S1) in the presence of FcR Blocking Reagent, mouse (Miltenyi Biotec, 130-092-575) and human (Miltenyi Biotec, 130-059-901). CountBright Absolute Counting Beads (Invitrogen, C36950) were used to calculate the number of tumor-infiltrating T cells. The data were obtained using LSRFortessa X-20 and analyzed using FlowJo V.10.
For nCounter analysis, tumors were dissected from mice and stored in RNAlater solution (Thermo Fisher Scientific, AM7021). RNA was extracted using the RNeasy Mini Kit (QIAGEN, 74106), according to the manufacturer’s protocol. RNA was subjected to gene expression analysis using an nCounter PanCancer Immune Profiling Panel (NanoString Technologies, XT-CSO-HIP1-12). The protocol was followed according to standard nCounter instructions. Sample processing and RNA counting were performed using the nCounter Prep Station and the nCounter digital analyzer (NanoString Technologies). Data were processed using nSolver software (NanoString Technologies), which included quality assessment of the runs. Z-scores for each gene were calculated using the mean and SD (Microsoft Excel). The Z-score of each gene was used to create the heatmap or graphs using GraphPad Prism V.8.4.3. The human gene set of canonical and non-canonical NF-kB pathway genes were obtained from the GSEA/MSigDB resource (https://gsea-msigdb.org).
Results
In vitro characterization of SAIL66
In the pursuit of an antibody capable of activating both CD137 and CD3 on CLDN6 binding, we combined the optimized CD3/CD137 dual-specific Fab29 and anti-CLDN6 Fab to create a CLDN6/CD3/CD137 tri-specific antibody, forming a humanized IgG1-structured antibody (figure 1A), which we named SAIL66.
SAIL66 is a potent CLDN6/CD3/CD137 tri-specific antibody. (A) SAIL66 has one CD3/CD137 dual-specific Fab and one anti-human CLDN6 Fab region. The Fc region is engineered to improve the efficiency of heavy chain heterodimerization whereas the binding to FcγRs and C1q is reduced. (B) The binding affinities of SAIL66 and conventional CLDN6 TCE to human and cynomolgus monkey CD3εγ, CD137, and CLDN6 were measured using surface plasmon resonance or kinetic exclusion assay. CLDN6, claudin-6; N.D., not determined; TCE, T-cell engager.
We introduced three amino acid substitutions into the Fc region of SAIL66 to prevent binding to FcγRs. SAIL66 demonstrated negligible binding to the eight types of human Fcγ receptors examined, unlike trastuzumab, which bound to all (online supplemental figure S1A). Meanwhile, SAIL66 binds to human FcRn similarly to trastuzumab (online supplemental figure S1B). This indicates that SAIL66 displays no FcγR-mediated effector function, while its pharmacokinetic (PK) characteristics remain unaffected. Additionally, SAIL66’s binding to human C1q was significantly less than that of rituximab, suggesting a reduced likelihood of undesired complement-dependent cellular cytotoxicity in humans (online supplemental figure S1C).
We assessed the binding affinities of SAIL66 and the bispecific CLDN6×CD3 antibody (CLDN6 TCE), used as a conventional TCE control, to human and cynomolgus monkey CD3, CD137, and CLDN6 using surface plasmon resonance (SPR) and a kinetic exclusion assay. SAIL66 bound to hCD3 with a KD value of 1.34 µmol/L and to hCD137 with 27.3 nmol/L, showing a 50-fold stronger affinity for CD137 than CD3 (figure 1B).
Next, we analyzed the binding competition of the Dual-Fab arm to CD3 and CD137 using SPR (online supplemental figure S2). The results revealed a rapid dissociation of CD3 from the Dual-Fab arm when only CD3 was injected. Interestingly, even with the addition of CD137 to the CD3-Dual-arm complex, no additional binding response of CD137 was observed. Instead, the bound CD3 appeared to be quickly replaced by CD137, resulting in a decrease in binding response due to CD137’s lower molecular weight compared with CD3. This replacement was further corroborated by the slow dissociation in the dissociation phase. These results suggest that the Dual-Fab arm has the desired property of shifting from CD3 to CD137 binding, thereby potentially enabling the CD3/CD137 dual-specific Fab arm to induce T-cell activation with suitable kinetics.
CD3 and CD137 signal activation by SAIL66 depending on the CLDN6
For assessing CD3-mediated T-cell activation, we co-cultured the human ovarian cancer cell line NIH:OVCAR-3, which endogenously expresses CLDN6, with Jurkat cells possessing an NFAT luciferase reporter gene (figure 2A and online supplemental figure S3). Both SAIL66 and CLDN6 TCE significantly enhanced NFAT signaling, whereas a control tri-specific Ab (isotype control) comprising anti-keyhole limpet hemocyanin (KLH) and a CD3/CD137 dual-specific Fab arm, had no effect. We further evaluated CLDN6-dependent CD137 and CD3 stimulation by SAIL66 and CLDN6 TCE. SAIL66 more potently induced the robust activation of NF-κB reporter luminescence than CLDN6 TCE. No luminescence activation was observed with the CLDN6-negative 5637 cell line (figure 2A and online supplemental figure S3). These results suggest that SAIL66 activates both CD3 and CD137 signaling in a CLDN6-dependent manner.
SAIL66 binds CLDN6 specifically and induces CD3 and CD137 signaling in T cells. (A) Jurkat-NFAT-luc reporter cells or CD137-expressing Jurkat-NF-κB-luc reporter cells were co-cultured with the CLDN6-expressing NIH:OVCAR-3 cell line or CLDN6-negative 5637 cell line in the presence of SAIL66 or control antibodies to assess activation of CD3 and CD137 pathways, respectively. Data are presented as mean+SD of fold changes in luciferase activity (n=3). (B) Outline figure of sequence differences in the extracellular domain between CLDN6 and CLDN9. (C) CLDN6-specific binding of SAIL66 was evaluated by flow cytometry using various CLDN-expressing Ba/F3 cell lines (n=1). CLDN, claudin; TCE, T-cell engager.
Binding specificity of SAIL66 to CLDN6
Next, we evaluated the binding specificity of SAIL66 to human CLDN6 using flow cytometry on Ba/F3 cell lines overexpressing several human CLDNs (online supplemental figure S4). SAIL66 showed specific binding to human and cynomolgus monkey CLDN6 with no cross-reactivity to human CLDN3, 4, and 9 (figure 2C). We also confirmed the binding of SAIL66 to I143V-mutated CLDN6, a major single nucleotide polymorphism (SNP) of human CLDN6 (CLDN6 (SNP)).30 The substitution of glutamine at position 156 on loop 2 of CLDN6 with leucine, as observed in CLDN9 (figure 2B), resulted in a complete loss of SAIL66 binding (as indicated by Q156L in figure 2C). This suggests that this amino acid residue is vital for the binding specificity of SAIL66 to CLDN6.
In vitro cytotoxic activity and T-cell activation ability of SAIL66
We evaluated the in vitro pharmacological profiles of SAIL66 using several cancer cell lines including OV-90, NIH:OVCAR-3 (human ovarian cancer cell lines), and PA-1 (human teratocarcinoma cell line), which endogenously express CLDN6 (online supplemental figure S3).11 The cytotoxic activity of SAIL66 was evaluated by a T cell-dependent cellular cytotoxicity (TDCC) assay, where cancer cells were co-cultured with human PBMCs from three different donors and a different concentration of SAIL66. SAIL66 induced strong cell killing against all CLDN6-positive cancer cell lines with a picomolar range (figure 3A, and online supplemental figure S5A). The cytotoxic activity was accompanied by dose-dependent T-cell activation, as evidenced by the induction of CD69 expression on CD8+ and CD4+ T cells (figure 3B, and online supplemental figure S5B) and the production of the cytokine IL-2 in the culture supernatant (figure 3C, and online supplemental figure S5C). These results indicate that SAIL66 stimulates T cells and directs TDCC to CLDN6-positive tumors.
SAIL66 activates T cells. Human peripheral blood mononuclear cells from three healthy donors were co-cultured with the OV-90 cell line in the presence of SAIL66 at indicated concentrations. (A) Cytotoxicity was evaluated by LDH release assay (B) CD69 expression on CD8+ T cells and CD4+ T cells were analyzed by flow cytometry (C) IL-2 concentration in culture supernatant was measured. The data is represented in different colors for each donor and mean+SD are shown (n=3). LDH, lactate dehydrogenase; IL, interleukin.
Next, we compared the T-cell activation capabilities of SAIL66 and CLDN6 TCE. CD8+ or CD4+ T cells were isolated from PBMCs and co-cultured with PA-1 cells in the presence of SAIL66 or CLDN6 TCE. After 72 hours, gene expression in T cells were quantified by real-time quantitative PCR. The results demonstrated that stimulation with SAIL66 resulted in a more pronounced increase in the expression of IFN-γ, TNF-α, CXCL9, and GZMB, compared with CLDN6 TCE in both CD8+ and CD4+ T-cell subsets (online supplemental figure S6A-D). Flow cytometry analysis also showed a significant increase in the protein expression levels of GZMB in both CD8+ and CD4+ T-cell subsets compared with CLDN6 TCE (online supplemental figure S6E).
In vivo antitumor effect in syngeneic mouse models
We examined the in vivo antitumor efficacy of SAIL66 using syngeneic mouse models bearing LLC1 overexpressing hCLDN6 (LLC1/hCLDN6: online supplemental figure S3). As SAIL66 does not bind to murine CD3 and CD137, we used two genetically modified mouse models: human CD3εδγ-expressing mice (hCD3 transgenic mice, or hCD3 tgm), and both human CD3εδγ and human CD137-expressing mice, which were generated by crossbreeding hCD3 tgm and hCD137 knock-in mice (hCD3/hCD137 KI mice).26 28 SAIL66 and CLDN6 TCE showed comparable antitumor efficacy against LLC1/hCLDN6 tumors established in hCD3 tgm (figure 4A). However, SAIL66 demonstrated significant antitumor efficacy compared with CLDN6 TCE in hCD3/hCD137 KI mice (figure 4A).
SAIL66 inhibits in vivo tumor growth by promoting intratumoral T-cell infiltration and suppression of exhausted T-cell population in a syngeneic mouse model. (A) Antitumor efficacy of a single administration of SAIL66 and conventional CLDN6 TCE at 1 mg/kg was measured in hCD3 tgm and hCD3/hCD137 KI mice bearing LLC1/hCLDN6 cells. Mean+SD of tumor volume is shown (n=5–7). P values were determined using the Wilcoxon rank-sum test. *p<0.05, **p<0.01. (B) CD8+ T-cell number and TOX expression as an exhaustion marker on CD8+ T cells in LLC1/hCLDN6 tumors were evaluated by flow cytometry at day 7 after single antibody administration (n=3–5). P values were determined by the Wilcoxon rank-sum test. *p<0.05. (C) Cytotoxic activity improvement of SAIL66 at 1 nmol/L by CD137 co-stimulation against Colon38/hCLDN6 cell lines was evaluated with T cells derived from hCD3 tgm and hCD3/hCD137 KI mice using xCELLigence. Cell growth inhibition ratio was calculated at 48 hours after SAIL66 and T cells addition. (D) Antitumor efficacy of SAIL66 and anti-PD-L1 antibody was evaluated in hCD3/hCD137 KI mice bearing LLC1/hCLDN6 cells. SAIL66 (0.2 mg/kg) was administered at day 6 and anti-PD-L1 antibody (10 mg/kg) was administered at day 6, 8, 10, 12, and 14 after tumor inoculation. Mean+SD of tumor volume is shown (n=5). P values were determined by the Wilcoxon rank-sum test. *p<0.05, **p<0.01. CLDN6, claudin 6; PD-L1, programmed cell death 1 ligand 1; TCE, T-cell engager.
Flow cytometric analysis revealed that SAIL66 treatment led to a significant increase in CD8+ T cells compared with CLDN6 TCE (figure 4B). Furthermore, the proportion of TOX-high T cells, indicative of T-cell exhaustion, was significantly lower in the SAIL66-treated group compared with CLDN6 TCE-treated group (figure 4B).
We conducted an in vitro TDCC assay using splenic immune cells from both types of genetically engineered mice (online supplemental figure S7). No difference was observed in the in vitro TDCC activity between SAIL66 and CLDN6 TCE when using splenic immune cells from hCD3 tgm. However, SAIL66 demonstrated superior in vitro TDCC activity compared with CLDN6 TCE when using splenic immune cells from hCD3/hCD137 KI mice, reflecting the in vivo study results. Next, we evaluated TDCC in the presence of a bispecific antibody for KLH and CD137 which blocks SAIL66 from binding to CD137. We found that KLH/CD137 BsAb decreased the TDCC of SAIL66 in a concentration-dependent manner, suggesting that CD137 signaling contributes to the stronger TDCC of SAIL66 (figure 4C).
We evaluated the therapeutic potential of SAIL66 in combination with a checkpoint inhibitor using an LLC1/hCLDN6 syngeneic tumor model. While SAIL66 monotherapy significantly reduced tumor volume, an anti-programmed cell death 1 ligand 1 (PD-L1) antibody alone failed to inhibit tumor growth, suggesting resistance to ICI treatment in this model, as previously reported.31 However, combining SAIL66 with the anti-PD-L1 antibody enhanced therapeutic outcomes in tumor growth inhibition, indicating the potential effectiveness of SAIL66 monotherapy and combination therapy with programmed cell death protein-1 (PD-1)/PD-L1 blockade in ICI-resistant tumors (figure 4D).
In vivo antitumor efficacy in xenograft mouse model
To further evaluate the in vivo antitumor efficacy of SAIL66 against CLDN6-positive human ovarian cancer, we used a humanized mouse model (huNOG) reconstituted with human CD34+ stem cells.32 Single administration of SAIL66 significantly inhibited tumor growth in the OV-90 subcutaneous xenograft model in a dose-dependent manner, with tumor shrinkage observed at doses as low as 0.2 mg/kg (figure 5A). SAIL66 also effectively suppressed tumor growth in the CLDN6-positive non-small cell lung cancer NCI-H1435 cell line, which exhibits lower CLDN6 expression (online supplemental figure S3 and S8).
SAIL66 inhibits in vivo subcutaneous and intraperitoneal tumor growth in humanized mouse model. (A) Antitumor efficacy of multiple dosages of SAIL66 was measured in humanized mice bearing OV-90 human ovarian cancer cells. Humanized mice were generated by human hematopoietic stem cell injection for NOG mice. SAIL66 was administered one time after OV-90 tumor establishment in humanized mice subcutaneously at indicated concentrations. Mean+SD of tumor volume is shown (n=6–7). P values were determined by Dunnett’s multiple comparison test. **p<0.01. (B) Antitumor efficacy of single administration of SAIL66 was measured against the OV-90 disseminated model. Human T cells which were derived and expanded from healthy donor peripheral blood mononuclear cells and SAIL66 at 5 mg/kg were injected intravenously 3 days after tumor intraperitoneal cavity implantation (n=9). P values were determined by log-rank test.
We evaluated SAIL66 in a peritoneal disseminated model of ovarian cancer, as peritoneal dissemination represents a specific form of malignant progression in ovarian cancer.33 NOD-SCID mice with intraperitoneally implanted OV-90 cells received intravenous injections of human T cells and either SAIL66 or vehicle 3 days post-implantation. Survival significantly increased in the SAIL66 group, ranging between 57 and 94 days, compared with the control group (figure 5B), supporting its potential as a treatment for patients with ovarian cancer with peritoneal metastases.
We investigated the effect of corticosteroid premedication on SAIL66 efficacy and the suppression of treatment-mediated cytokine release syndrome (CRS), a common clinical issue with TCE use often managed with steroids.34 An increase in serum cytokine levels was observed immediately after the administration of SAIL66, but this increase was suppressed by premedication with dexamethasone (DEX) (figure 6A). Conversely, DEX had minimal impact on the antitumor efficacy of SAIL66 (figure 6B). Gene expression analysis in tumors after SAIL66 administration using nCounter revealed a significant increase in immune-related genes within the tumor, but no reduction was observed even with DEX premedication (figure 6C). These results confirm that DEX premedication attenuates cytokine production without suppressing the antitumor activity and intratumor immune activation induced by SAIL66.
The use of dexamethasone eliminates CRS, but does not limit the antitumor efficacy of SAIL66. (A) The effect of dexamethasone (DEX) premedication on the plasma cytokine expression was evaluated at indicated time points after SAIL66 administration. DEX (33 mg/kg) was intraperitoneally administered 1 and 24 hours (DEX ×2), or 1 hour (DEX ×1) prior to SAIL66 (0.2 mg/kg) administration in humanized mice bearing OV-90 cells. (B) The antitumor efficacy of SAIL66 and DEX were indicated. Mean+SD are shown (n=6). P values were determined by the Wilcoxon rank-sum test. **p<0.01. (C) RNA profiling was conducted on RNA extracted from OV-90 tumors, which were dissected on day 7 following a single administration of SAIL66. This was done using the NanoString nCounter Human PanCancer Immune Profiling Panel. Data analysis was performed using the nSolver Analysis Software V.4.0, and the nCounter Advanced Analysis modules were used for Gene Set Analysis. This analysis compared differential expression in the vehicle versus the drug treatment group. The extent of differential expression in each gene set was summarized using a “global significance score”. The mean score of each group is displayed (n=3). CRS, cytokine release syndrome; IFN, interferon; IL, interleukin.
Immune profiling of SAIL66 in tumor-bearing huNOG mouse model
To understand the mechanism of action of SAIL66 and compare the difference in potency between SAIL66 and CLDN6 TCE, we treated NIH:OVCAR-3 bearing xenograft huNOG mice with SAIL66, CLDN6 TCE, and a control bispecific antibody (isotype control). SAIL66 demonstrated relatively higher antitumor efficacy than CLDN6 TCE (figure 7A) and significantly reduced CLDN6 messenger RNA expression in tumor tissue on day 14 (figure 7B). Flow cytometry analysis of tumor tissues revealed that both antibodies promoted CD3+ T-cell infiltration into the tumor tissue following antibody treatment (figure 7C). Notably, tumors treated with SAIL66 maintained a significant number of infiltrating CD3+ T cells within the tumor even 14 days post-administration. Within the T cell subsets, CD8+ T cells were found to infiltrate the tumor at a higher frequency compared with CD4+ T cells (Online supplemental figure S9). An analysis of the exhausted T-cell population on days 2, 7, and 14 revealed that the percentages of PD-1, Lymphocyte activation gene 3 protein (LAG3), and T-cell immunoglobulin mucin receptor 3 (Tim-3) triple-positive CD3+ T cells were lower in the SAIL66 group than in the CLDN6 TCE group (figure 7C).
SAIL66 achieves superior efficacy over conventional TCE by enhancing T-cell infiltration and suppressing exhausted T-cell population in humanized mice model. (A) Antitumor efficacy of SAIL66 and conventional CLDN6 TCE was compared in humanized mice bearing NIH:OVCAR-3 cells. Humanized mice were generated by human hematopoietic stem cell injection for NOG mice. Antibodies were administered one time after NIH:OVCAR-3 tumor establishment in humanized mice subcutaneously at 1 mg/kg. Mean+SD of tumor volume is shown (n=5). (B) CLDN6 messenger RNA expression in tumor tissue was quantified at days 2, 7, and 14 after single administration (n=5). (C) Flow cytometry analysis of T-cell density and T-cell phenotype in tumor tissue were evaluated on days 2, 7, and 14 after single administration (n=5). (D) RNA expression of immune-related genes in NIH:OVCAR-3 tumors was measured using nCounter Analysis System. The expression levels of the indicated genes in tumors were quantified at 2-day, 7-day, and 14-day post-treatment and are shown as Z-scores. (E) Plasma IL-2 and CXCL10 concentrations were measured at the indicated time points after single administration (n=5). (F) Pathological analysis of tumor tissue by H&E staining and IHC from the tissues of OV-90 bearing huNOG model at day 2, 7, and 14 after SAIL66 injection at 1 mg/kg, and at day 2 after vehicle injection. P values in A, B, and C were determined by the Wilcoxon rank-sum test. P values of gene expression shown in D between the CLDN6 TCE-treatment and SAIL66-treatment group (n=5) were determined using Student’s t-test. *p<0.05, **p<0.01, ***p<0.001. CLDN6, claudin-6; IHC, immunohistochemistry; IL, interleukin; LAG3, lymphocyte activation gene 3 protein; PD-1, programmed cell death protein-1; TCE, T-cell engager; Tim-3, T-cell immunoglobulin mucin receptor 3.
We also compared transcriptomic profiles in tumor tissues using the NanoString nCounter PanCancer Immune Profiling Panel. The heat map plot revealed that tumors treated with SAIL66 exhibited higher expression of a gene set encompassing T/B-cell function, cytotoxicity, antigen processing, and natural killer cell function (online supplemental figure S10). Furthermore, SAIL66 administration induced a more potent increase in the expression of canonical and non-canonical NF-κB pathway gene set in the tumor on day 2 post-treatment compared with CLDN6 TCE (online supplemental figure S11). These results suggest that SAIL66 activated the CD137 signal more strongly than CLDN6 TCE in the tumor. Focusing on the temporal changes of several representative genes, the expression of IFN-γ (as T-cell activation marker), MYD88 (as NF-κB pathway), and CXCL9/10 (as T-cell recruiting chemokines) were transiently induced on day 2 (figure 7D). By day 7, these genes were no longer upregulated. Instead, the expression of CD3/CXCR3 (as T-cell marker), and GZMA/GZMB/perforin (as cytotoxicity marker) increased (figure 7D). These gene expression changes were more significant in the SAIL66-treated group compared with the CLDN6 TCE-treated group.
We also examined the IL-2 and CXCL10 concentration in plasma. The release of IL-2 into circulation peaked at 6 hours post-antibody administration for both SAIL66 and CLDN6 TCE. On the other hand, consistent with the result of gene expression by nCounter, both SAIL66 and CLDN6 TCE induced plasma concentration of CXCL10 at 6 hours post-administration, and it remained detectable up to 72 hours in the SAIL66-treatment group, but was undetectable in the CLDN6 TCE-treatment group by that time (figure 7E).
Tumor tissues from OV-90 cell xenografted huNOG mice, sampled on days 2, 7, and 14 after administration of vehicle or SAIL66, were subjected to H&E staining, CD3 and CLDN6 immunohistochemistry (IHC) (figure 7F). In the vehicle-treated group, CLDN6-positive cancer cells were frequently observed, while CD3+ T cells were rarely present. Over time in the SAIL66-treated group, granulocyte infiltration and tumor cell degeneration/necrosis increased. Additionally, IHC demonstrated that CLDN6-positive cancer cells disappeared on SAIL66 treatment compared with the vehicle-treated group, with a marked rise in CD3+ T cells on days 7 and 14 (figure 7F). These findings clearly demonstrated that SAIL66 reduces CLDN6-positive cells through immune cell infiltration in the tumor microenvironment.
Non-human primate study revealed linear pharmacokinetics of SAIL66
Given that SAIL66 cross-reacts with monkey-derived antigens, we conducted the PK evaluation using cynomolgus monkeys as a model. PK parameters were calculated using data from single intravenous dosing of SAIL66 at 1, 10, and 100 µg/kg in male cynomolgus monkeys. We also assessed anti-drug antibodies (ADA). SAIL66 showed biphasic plasma concentration-time profiles typical of antibodies (online supplemental figure S12). For animals where ADA-positive responses were detected from day 7 onwards, PK parameters were calculated prior to the initial ADA-positive response, due to the impact of ADA formation on PK. The mean Cmax and the area under the curve (AUC)0–7 d increased dose-proportionally from 1 to 100 µg/kg, with mean t1/2 values ranging from 5.85 to 9.74 days for all groups (online supplemental table S2). In the 4-week toxicity study, transient changes in normal tissues were observed without serious pathological effects, indicating a manageable safety profile for SAIL66.
Discussion
In this study, we introduced SAIL66, a novel tri-specific antibody targeting CLDN6, CD3 and CD137. A luciferase reporter Jurkat cell assay confirmed SAIL66 activated both CD3 and CD137 signals in the presence of CLDN6-expressing cells, unlike conventional CLDN6 TCE activating only CD3 signal. In vitro, SAIL66 more potently induced IFN-γ, TNF-α, CXCL9, and GZMB expression in both CD4+ and CD8+ T cells compared with CLDN6 TCE. Furthermore, SAIL66 demonstrated robust antitumor activity in three distinct mouse models: syngeneic mice engineered to express human CD3 and CD137 (hCD3/hCD137 KI mouse), humanized NOG (huNOG) mice model, and a murine model implanted with human T cells.
While higher affinity to CD3 augments cytotoxicity of TCE, it also escalates cytokine production, thereby elevating the risk of CRS.34 Essentially, T-cell activation, proliferation, memory T-cell differentiation, and survival require not only “signal 1” from CD3 but also co-stimulation via “signal 2”. CD3-based TCEs that solely activate “‘signal 1” without co-stimulation may trigger T-cell activation followed by T-cell dysfunction, ultimately undermining sustained antitumor activity. SAIL66, with its reduced CD3 affinity, activates both signals 1 and 2 through CD137 binding. Despite its 3.6-fold weaker affinity for CD3 (1.34 µmol/L) compared with CLDN6 TCE (375 nmol/L), SAIL66 demonstrates comparable or superior antitumor activity in tumor-bearing mouse models, implying that the activation of the CD137 signal by SAIL66 contributes to its potent antitumor activity. The importance of CD137 co-stimulation has been proven by the effectiveness of second-generation/third-generation CD137/CD3ζ CAR-T cells in clinical settings.12 Additionally, CD137 agonist antibodies in combination with tumor antigen targeting TCEs have been shown to enhance tumor-infiltrating T-cell expansion and potentiate antitumor activity.35 Along with these therapeutic strategies, SAIL66 provides a groundbreaking off-the-shelf agent that activates both signals 1 and 2 in a CLDN6-dependent manner, exhibiting exceptional antitumor activity.
The enhancement of in vivo antitumor activity through CD137 signal activation by SAIL66 can be attributed to various mechanisms. While we confirmed the superiority of SAIL66 in activating T cells compared with conventional TCE in in vitro assays, we also observed stronger induction of genes encoding IFN-γ and the NF-κB pathway in the tumors treated with SAIL66 compared with conventional TCEs at an early time point on day 2 post-administration. These findings suggest that SAIL66 may potently activate T cells via the CD137 signal shortly after treatment. Notably, the early induction of CXCL9/10 expression, crucial chemokines for T-cell infiltration into tumor tissue,36 implies that CD137 signal activation by SAIL66 promotes T-cell activation and chemokines induction, eventually enhancing tumor-infiltrating lymphocyte accumulation and thereby augmenting its antitumor activity.
Furthermore, syngeneic tumor models showed a smaller population of TOX-high exhausted T cells in the tumors treated with SAIL66 compared with CLDN6 TCE. Given that CD137 signaling is known to prevent T-cell exhaustion,37 it is plausible that SAIL66 not only activates T cells but also concurrently prevents their exhaustion. Additionally, in vitro assays using T cells isolated from hCD3/hCD137 KI mice have confirmed higher cell cytotoxicity with SAIL66 compared with CLDN6 TCE. Overall, SAIL66 may enhance antitumor activity by increasing T-cell infiltration via chemokines, preventing T-cell exhaustion, and increasing T-cell cytotoxicity, mediated through CD137 co-stimulation. However, more detailed analyses are still needed to fully understand how SAIL66 differentially impact the immune system over CLDN6 TCE.
TCEs offer a powerful therapeutic modality using T-cell cytotoxicity, but off-target reactions remain a critical safety concern. CLDN6, highly expressed in many solid tumors but minimally expressed in healthy adult tissues, represents an ideal target for cancer immunotherapy.38 However, generating CLDN6-selective monoclonal antibodies (mAbs) is challenging due to the high sequence similarity of extracellular loops of CLDN3, 4, and 9, which are ubiquitously expressed in normal tissues.39–41 To avoid potential off-target reactions from cross-reactivity with the broadly expressed CLDN3, 4, and 9, therapeutic mAbs must bind CLDN6 with high specificity, excluding closely related CLDNs. We discovered that the Q156 residue in the second loop, differing between CLDN6 and CLDN9, is pivotal for CLDN6 specificity. SAIL66 demonstrated selective binding to CLDN6 with no cross-reactivity to CLDN3, CLDN4, and CLDN9, making it a promising clinical candidate due to its high CLDN6 specificity, offering both efficacy and safety.
In our syngeneic mouse model, SAIL66 demonstrated potent therapeutic effects even in a PD-L1 antibody-unresponsive model. Combining SAIL66 with a PD-L1 antibody enhanced its antitumor activity, suggesting SAIL66 could potentially overcome resistance to anti-PD-1/PD-L1 therapies, including ovarian cancer. Although further investigation is needed, SAIL66 may convert the tumor microenvironment from immunologically “cold” to “hot” by expanding functional intratumoral T cells, thereby increasing the efficacy of PD-L1 antibodies. Additionally, intravenous SAIL66 demonstrated efficacy in a peritoneal ovarian cancer dissemination model. As around two-thirds of advanced ovarian cancers undergo peritoneal dissemination associated with poor prognosis,42 SAIL66 represents a promising therapeutic option for advanced ovarian cancer with peritoneal dissemination.
In conclusion, our engineered next-generation TCE, SAIL66, designed to engage CLDN6, CD3, and CD137, has shown significant potential in enhancing antitumor activity. The activation of both CD3 and CD137 signaling pathways by SAIL66 offers an innovative approach to augment T-cell responses, resulting in increased tumor infiltration and reduced exhaustion. The specific targeting of CLDN6 underscores the potential clinical impact of SAIL66. Furthermore, its synergy with ICIs provides a promising avenue to address the limitations of current immunotherapies, particularly in cancers resistant to PD-1/PD-L1 blockade. However, murine models have inherent limitations in their ability to accurately recapitulate the immune response and antitumor effects induced by SAIL66 in humans. While phase I trials to evaluate the safety and efficacy of SAIL66 have been initiated (NCT05735366), future translational studies will be needed to comprehensively elucidate the long-term efficacy, safety, biomarkers, and mechanisms of action associated with SAIL66.
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information.
Ethics statements
Patient consent for publication
Ethics approval
All in vivo mouse experimental procedures were approved by the Institutional Animal Care and Use Committee of Chugai Pharmaceutical (Approval No. IACUC20-200, 22-232, 21-195, and 21-007). Chugai Pharmaceutical is accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care (AAALAC) International. The study using cynomolgus monkeys was performed at Shin Nippon Biomedical Laboratories, Ltd., Drug Safety Research Laboratories. The study was approved by the Institutional Animal Care and Use Committee (Approval No. IACUC036-647) and performed in accordance with the animal welfare by-laws of Shin Nippon Biomedical Laboratories, Ltd., Drug Safety Research Laboratories, which are also accredited by AAALAC International.
Acknowledgments
We acknowledge the technical and scientific support of our colleagues at Chugai Pharmaceutical, Chugai Research Institute for Medical Science, and Chugai Pharmabody Research: Hirofumi Sakumoto, Shigeto Kawai, Naoko Wada, Otoya Ueda, Akira Hayasaka, Daiki Kashiwagi, Shiho Ohtsu, Sun Silvia, Shogo Kamikawaji, Jinki Hadano, Etsuko Fujii, Asako Harada, Nozomi Fujisawa, Keiichi Morita, Genki Nakamura, Samantha Ho, Shun Shimizu, Naoka Hironiwa, Mari Kinoshita, Toshiaki Tsukaguchi, Ryo Nakamura, and we are grateful to Dr Kunihiro Nishimura and Dr Shuichi Tsusumi at The University of Tokyo for construction of the human exosome database and collaboration on the target identification.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors Conceptualization: NK, TKa, SI, MM, TKo, and TI. Methodology: TKa, NK, SI, MM, KT, MY, RU, MM-O, CK, JS, SA, and SN. Resources: SI, MM, CL, GG, and SF. Investigation: TKa, NK, SI, MM, MY, KT, RU, MM-O, CK, JS, SA, SN, NT, NN, CL, GG, and SF. Formal analysis: TKa, NK, SI, MM, KT, MY, MM-O, CK, JS, SA, SN, NT, and NN. Visualization: TKa, NK, MM, KT, MY, CK, and SN. Supervision: MS, MK-S, HA, TK, and TI. Project administration: NK and MM. Writing—original draft: NK, TKa, SI, MM, KT, RU, MY, CK, and SA. Writing—review and editing: NK. All authors discussed the results and commented on the manuscript. TI is the guarantor.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests All authors except for HA are current employees of Chugai Pharmaceutical and Chugai Pharmabody Research. HA received a collaborative research grant from Chugai Pharmaceutical. SI, NK and TKa are inventors on the patent application (WO2021200939), TKa, SI, NK, and TKo are inventors on the patent application (WO2023053282 and WO2023054421) submitted by Chugai Pharmaceutical, which covers SAIL66.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.