Background

Laryngeal carcinoma (LC) is a solid malignant tumor derived from the laryngeal mucosal epithelium with a high incidence and poor prognosis1 2; its main pathological type is laryngeal squamous cell carcinoma (LSCC). According to the Global Cancer Observatory website (https://gco.iarc.fr/survival/), the incidence of LSC and corresponding mortality are projected to increase by 58.6% and 66.2%, respectively, by 2040. This trend poses a significant threat to global medical and health resources as well as social and economic development.

The current standard treatments for LSCC comprise surgery, radiotherapy, and chemotherapy, all of which carry significant risks of limited efficacy and considerable side effects. Particularly concerning are tumors located in the highly functional larynx, as these treatment methods can severely impact the patient’s psychological well-being and overall quality of life,3 4 potentially leading to a substantial increase in suicide risk.5 Therefore, there is an urgent need to explore more targeted and precise noninvasive treatments to improve the survival rate and quality of life of patients with LSCC.

Antibody-drug conjugates (ADCs) combine monoclonal antibodies with chemotherapy agents, facilitating precise delivery to cancer cells and concurrently minimizing toxicity while enhancing therapeutic effectiveness.6 7 Compared with conventional ADCs, bispecific ADCs (BsADCs) exhibit superior specificity for targeting cells. They facilitate the internalization of the two targets and inhibit tumor cell growth signals by diminishing the expression of receptor proteins on the cell membrane, thereby enhancing safety and efficacy.8 Thus, developing BsADCs targeting therapeutic markers is of significant clinical importance.

Programmed cell death ligand 1 (PD-L1) is a transmembrane protein prominently expressed on the surface of various solid tumor cells, particularly those in head and neck cancers.9 10 Among them, its expression in LSCC is also highly significant,11 with 68.5% of LSCC expressing higher levels of PD-L1 transcripts.12 PD-L1 interaction with programmed cell death receptor 1 (PD-1) on T cells facilitates the transmission of immune inhibitory signals, leading to suppression of T-cell proliferation and aggregation. Consequently, tumor cells evade immune system elimination, representing a crucial mechanism for immune evasion during cancer progression.13 14 This mechanistic understanding has spurred the development of therapeutic approaches that use monoclonal antibodies targeting PD-1/PD-L1 to reinvigorate antitumor immunity.15–17 While numerous related drugs have shown promising outcomes in clinical trials and garnered widespread utilization,18–20 the overall clinical response rate of FDA-approved drugs is only approximately 13%–18%, with a significant proportion of patients exhibiting primary resistance.17 21 Consequently, exclusively targeting PD-L1 may not be the optimal choice.22

Similar to PD-L1, B7-H3 belongs to the B7 family and exhibits potential immune-regulatory functions. It is either not expressed or expressed at low levels in normal tissues but is highly expressed in the majority of malignant tumors, including head and neck cancers.23 24 Studies25 based on the Cancer Genome Atlas (TCGA) have shown that the expression of B7-H3 is significantly upregulated in LSC and closely related to poor prognosis. Although the immune regulatory function and specific mechanism of B7-H3 are not fully understood, numerous studies have reported the basis and clinical research targeting B7-H3 for cancer treatment.26–29 In murine models of hematological malignancies, ovarian cancers, melanomas, and colorectal cancers, the blockade of B7-H3 using monoclonal antibodies has been shown to augment the infiltration of CD8+T cells and natural killer cells within tumors, thereby leading to diminished tumor proliferation and prolonged survival.30–32 Additionally, coexpression of PD-L1 and B7-H3 has been identified in various solid tumors,33–36 suggesting that targeting these two molecules could potentially enhance tumor-binding specificity and induce tumor-killing efficacy by acting on both the innate and adaptive immune systems.

Monomethyl auristatin E (MMAE) is a commonly used warhead drug in ADCs,37 primarily acting by inhibiting the polymerization of microtubules.38 By disrupting cell mitosis and altering the cell’s cytoskeletal structure, it leads to cell death and exhibits strong cytotoxicity against rapidly dividing tumor cells.39 Due to the potent toxicity of MMAE as a stand-alone drug, its clinical application and therapeutic index are limited. Therefore, MMAE is frequently incorporated into targeted drugs.

In this study, we successfully developed a BsADC targeting PD-L1 and B7-H3 conjugated with the MMAE toxin (hereafter referred to as BsADC). Subsequently, we established an immunocompromised humanized mouse model and validated the efficacy of BsADC against LSCC, elucidating its potential mechanisms both in vivo and in vitro.

Materials and methodsCell lines and culture

The SNU46 and SNU899 cell lines are both human LSC carcinoma cell lines established by Seoul National University. They originate from the supraglottic and glottic larynx, respectively, of middle-aged Asian males. SNU46, SNU899 and Hela cells were cultured in RPMI 1640 medium (Gibco) supplemented with 10% FBS (HyClone) and 1% penicillin/streptomycin (P/S, HyClone). FADU, SCC15 and MC38 cells were cultured in DMEM (Gibco) supplemented with 10% FBS and 1% P/S. All cells were maintained at 37°C in a humidified incubator with 5% carbon dioxide.

Using the CRISPR-Cas9 system, PD-L1 and/or B7-H3 gene knockout SNU46 and SNU899 cell lines were generated. We designed guide RNAs (gRNAs) targeting the exons of PD-L1 and B7-H3 genes using an online server and subcloned them into the lentiCRISPR V2 vector (Addgene plasmid #52961). The PD-L1 and B7-H3 KO gRNA sequence were AGCTACTATGCTGAACCTTC, ATGCGTTGCCCTGTGCCAGC, respectively. When the cells reached 30%–50% confluency, they were transduced with lentivirus and selected with puromycin. The knockout efficiency was validated by sequencing and immunofluorescence (IF) staining. The PD-L1 or B7-H3 overexpressed Hela cell lines were established by virus transduction and puromycin selection using lentivirus vectors.

Drug preparationConstruction and expression of humanized antibodies

We selected independently screened completely humanized PD-L1 and B7-H3 monoclonal antibodies from our own laboratory (patent application numbers: CN202211366304.0 and CN202410690763.7) using traditional hybridoma technology, and validated using IF staining, with detailed screening procedures described in previous studies.40 Subsequently, we constructed a PD-L1 complete antibody, a B7-H3 complete antibody, and a bispecific antibody (BsAb, plotted in online supplemental figure 2C) containing the PD-L1 complete antibody and B7-H3 scFv through pcDNA3.1 plasmid modification. Besides, hIgG1 (isotype hIgG1, #HG1K, Sino biological) was connected as Fc in all the BsAb and ADCs.

Preparation of the conjugate

The antibody solvent was replaced with ADC conjugation buffer using a 30 kDa ultrafiltration tube, and the antibody concentration was measured and adjusted to 5 mg/mL. Subsequently, under sterile conditions, the interchain disulfide bonds of the antibody were reduced and conjugated with MMAE. Tris (2-carboxyethyl) phosphine (TCEP) was added to the antibody in a 1:3 molar ratio and reacted at 37°C, 120 rpm for 3 hours to reduce the antibody. Following this, the drug-linker (HY-15575; MedChemExpress) dissolved in DMSO was slowly added to the TCEP-treated antibody reaction solution at a molar ratio of 1:6, while gradually supplementing DMSO to 10% of the total reaction volume. The solution was then reacted in the dark at 25°C, 40 rpm for 3 hours. After centrifugation (4000 rpm, 5 min), the supernatant containing the conjugated drug was collected. The solvent was exchanged for ADC storage buffer using a protein ultrafiltration tube (at least three times to remove unbound small molecules). The protein concentration was determined, followed by filtration, aliquoting, snap-freezing in liquid nitrogen, and storage at −80°C (Detailed solvent ratios can be found in online supplemental materials).

Grouping

Subsequent studies were divided into five groups: control (Ctrl), BsAb, PD-L1 ADC, B7-H3 ADC, and BsADC. Specifically, equivalent hIgG1 (which was used in the construction of BsAb and ADCs) conjugated with MMAE was used as the Ctrl group in the in vitro efficacy experiments. In vivo, the Ctrl group (Vehicle group) received equivalent amount of saline. Additionally, in the target expression validation experiments for cell lines, the Ctrl group was a blank control.

Characterization of ADCMass spectrometry

The ADC samples were reconstituted in a solution of 33% acetonitrile and subsequently analyzed using liquid chromatography coupled with mass spectrometry, employing online solid-phase extraction (XILING LAB Pharmaceutical). To establish a quantitative relationship for the measurement of the released drug in the samples of interest, the peak area corresponding to each standard drug was divided by that of the internal standard. The resulting peak area ratios were then plotted against the standard concentrations while resulting data points were fitted to a curve using linear regression. Using this derived equation, the peak area ratios obtained for the liberated drug relative to the internal standard in the experimental samples were converted into corresponding drug concentrations.

Avidity detecting

The avidities of the antibodies and ADCs toward the recombinant protein B7-H3 (B7-H3-Fc) or PD-L1 (PD-L1-Fc) were evaluated using a Biacore 8K instrument equipped with a CM5 sensor chip (GE Healthcare), following a previously published method.41 To facilitate the immobilization process, the CM5 chip was affixed to a rabbit anti-mouse IgG antibody (mouse antibody capture kit, GE Healthcare) via amine coupling (amine coupling kit, GE Healthcare). Subsequently, B7-H3-Fc or PD-L1-Fc proteins were indirectly captured on the chip’s surface. The antibodies and ADCs were then injected across the chip at various concentrations in a dilution series. The equilibrium dissociation constant was determined using the BIAevaluation software 8K.

Cytotoxicity assaysIC50

Tumor cells were seeded at a density of 5×103 cells/well in a 96-well plate. The next day, BsADC was administered at concentrations of 0 nM, 31.25 nM, 62.5 nM, 125 nM, 250, nM and 500 nM, with each concentration having three replicate wells. After 72 hours of incubation in the cell culture incubator, 10 µL of MTT was added to each well and incubated at 37°C temperature for 4 hours. The solution in each well was then aspirated, and 100 µL of DMSO was added for dissolution. After shaking in the dark for 2 min, the plate was incubated at 37°C temperature for an additional 10 min, also protected from dark. Bioluminescence was measured at a detection wavelength of 492 nm using a multifunctional microplate reader. GraphPad Prism V.9 software was used to fit the in vitro cytotoxicity curve and calculate IC50 values.

Immunofluorescence

Cells were added at a density of 5×104 to each well of a 24-well confocal plate. The next day, cells were incubated with RPMI 1640 or BsADC (100 nM, the approximate IC50 dose) at 37°C for 4 hours. The cells were washed with PBS, fixed with 4% paraformaldehyde, and permeabilized with 0.5% Triton X-100 (Sigma-Aldrich, Cat# T8787) in PBS. They were then incubated overnight at 4°C with anti-tubulin antibody (Abcam, ab6160) diluted in 3% BSA/PBS. After washing with PBS the next day, the cells were stained with diluted Alexa Fluor 488 Fluorescent anti-tubulin antibody (ab197737, Abcam) and cell nuclei were counterstained with DAPI (Solarbio, Cat#C0060). Images were captured using a confocal microscope (Olympus IXplore Spin SR).

Internalization assay

ADC internalization was detected using two methods: flow cytometry and confocal microscopy. For flow cytometry, five groups (Ctrl, BsAb, PD-L1 ADC, B7-H3 ADC, and BsADC) were prepared, and each group was further divided into 0, 1, 2, 4, and 6 hours time points. The cells were seeded in a 12-well plate at a density of 2×105 cells/well. The next day, the cells were treated with 100 nM at 4°C for 1 hour, washed with PBS, and then incubated at 37°C for 0, 1, 2, 4, or 6 hours, separately. Subsequently, cells were washed, acquired, and fixed, followed by incubation with secondary antibodies (ab6854; Abcam) at 4°C for 40 min.

For confocal microscopy, cells were seeded and grown in a 24-well confocal plate at a density of 5×104 cells per well. The next day, cells were treated with RPMI 1640 or BsADC (100 nM) at 4°C for 1 hour to facilitate drug-cell binding. Subsequently, cells were transferred to a 37°C incubator for 1 or 4 hours, washed with PBS, fixed with 4% paraformaldehyde, and permeabilized with 0.5% Triton X-100 in PBS. The cells were then stained with goat anti-human IgG (Abcam, ab6854) and the cell nuclei were stained with DAPI (Solarbio, Cat# C0060). Images were captured using a confocal microscope (Olympus IXplore Spin SR).

Cell proliferation assay

Tumor cells (2×103 cells/well) were seeded in a 96-well plate and cultured overnight with five plates for each cell type, six groups per plate, and three replicate wells per group. The next day, the cells were treated with 100 nM, and bioluminescence was detected using the MTT assay, as described in the IC50 detection method. Subsequently, one plate of each cell type was monitored daily for cell proliferation for a total of 5 days. The in vitro cell antiproliferation curves were fitted using the GraphPad Prism V.9 software.

Cell apoptosis and cell cycle

Cells were seeded at a density of 1×105 cells/well in six-well plates and cultured overnight. Subsequently, the culture medium was replaced with RPMI 1640 containing 100 nM of each drug. After 48 hours of incubation, cells were harvested using 0.25% trypsin (Gibco) and centrifuged at 800 rpm for 3 min. The cells were then washed with cold PBS. For the apoptosis analysis, cells were stained with Annexin V-FITC for 15 min, followed by PI staining for 5 min. To analyze the cell cycle, cells were fixed with precooled 70% ethanol and stored at 4°C for 2 hours. Finally, PI staining was performed on the fixed samples, and RNA was degraded with RNase A at 37°C for 30 min.

Flow cytometry

The cells were washed twice with PBS and incubated at 37°C for 5 min with 0.25% trypsin (Gibco), followed by centrifugation and removal of the supernatant. After washing with PBS, the cells were blocked in a solution containing 1% BSA in PBS for 20 min in the dark. The cells were washed twice, incubated in the dark for 30 min with antibodies from our own laboratory (patent application numbers: CN202211366304.0 and CN202410690763.7) or commercially available primary antibody (B7-H3: #331606, BioLegend; PD-L1: #374514, BioLegend) according to the manufacturer’s instructions, and then detected using a NovoExpress flow cytometer (Agilent, USA). Flow cytometry analysis was performed using NovoExpress software (V.1.6.2) or FlowJo (V.10.8.1).

Immunohistochemistry

Commercially available tumor tissue microarrays (Bioaitech, HN049La01 and XN04801) and tumor slices from in vivo experiments were stained using commercial monoclonal antibodies against PD-L1 (#13684, Cell Signaling Technology, CST), B7-H3 (#14058, CST), Ki67 (#9027, CST), and CD8 (A0663, ABclonal), according to the manufacturer’s standard protocols. Fluorescence microscopy (DP80, Olympus) was used for detection and image capture. Three random fields were selected, and quantification of immunohistochemistry (IHC)-positive staining was performed using Fiji,42 wherein the resulting percentage was considered as the staining score (The specific quantification methods are supported in online supplemental text 1).

Western blotting

Cell lysis was performed using RIPA buffer (Sigma), followed by separation using sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transfer to nitrocellulose membranes (GE Healthcare). After incubating with the primary antibody (Caspase-3: A11021, ABclonal; pHH3: AP1431, ABclonal) in tris-buffered saline-0.1% Tween-20 (TBST) containing 5% non-fat dry milk at 25°C temperature for 1 hour, the membranes were incubated overnight at 4°C. The membranes were then washed thrice with TBST and incubated with the secondary antibody (anti-rabbit: AS014, ABclonal) in TBST-BSA (TBST containing 5% BSA) at 25°C temperature for 1 hour, followed by three additional washes with TBST. Detection was performed using ChemiDoc MP Imaging System.

Coculture with human peripheral blood mononuclear cellPeripheral blood mononuclear cell extraction and culture

Lymphocyte separation solution (3 mL) was added to a 10 mL lymphocyte separation tube and centrifuged at 800 rpm for 1 min. Then, every 10 mL of fresh blood from the donor was collected, diluted with PBS in a 1:1 ratio and added in batches to the lymphocyte separation tube. The tube was then centrifuged at 1500 rpm for 20 min (acceleration, 5; deceleration, 2). After centrifugation, the layers in the separation tube, from top to bottom, were serum, peripheral blood mononuclear cell (PBMC), lymphocyte separation solution, and RBC precipitate. PBMCs were carefully aspirated, and cell counting was performed. After extracting PBMCs, we used IL-15, IL-2 and IL-7 to activate T cells and other effector cells.

Co-culture study

Cells were seeded at a density of 5×103 cells/well in a 96-well plate. The following day, the cells were treated with 100 nM and incubated for 24 hours. After co-culturing with 1×105 PBMCs for 48 hours, the cells were collected for flow cytometry analysis.

Immunogenic cell death induction detection

Flow cytometry was used to analyze cell surface calreticulin (CRT), following procedures similar to those used for apoptosis detection. The CRT antibody (ab92516, Abcam) was used for incubation, followed by staining with goat anti-rabbit IgG (FITC) (ab6717, Abcam). HMGB1 was determined by IF, following the manufacturer’s instructions (10 829-1-AP, Proteintech). ATP levels in culture supernatants were measured by luminescence using the ENLITEN ATP assay system bioluminescence detection kit (Promega) for ATP Measurement, according to the manufacturer’s instructions. Additionally, p-EIF2α (AP0692, ABclonal) and EIF2α (A21221, ABclonal) were detected by western blotting (WB). The IFN-γ content of blood samples was detected by ELISA.

In vivo experimentsAnimals

Six-week-old immunodeficient female NCG mice were obtained from Gempharmatech and housed in a specific pathogen-free environment at West China Hospital of Sichuan University. All efforts were made to minimize animal suffering.

NCG mouse humanization of PBMCs

The method for extracting PBMCs was conducted as previously described, with all blood samples obtained from the same donor, a healthy adult female, resulting in a total collection of 160 mL of fresh blood. Each NCG mouse was injected with 1×107 freshly extracted PBMCs (100 µL cell suspension) via tail vein injection. Blood samples were collected at 2 and 3 weeks post-injection for flow cytometry analysis to assess the status of immune reconstitution.

In vivo therapeutic efficacy

Each immunocompromised NCG mouse (≈20 g) was subcutaneously inoculated with 100 µL cell suspension (5×106 SNU899 cells in 100 µL RPMI 1640 medium without FBS) near the right thigh. When the tumor volume reached ≈50–60 mm3, mice were randomly divided into six groups (n=5) and received intravenous injections of saline (Vehicle group), BsAb, PD-L1 ADC, B7-H3 ADC, and BsADC at a dose of 5 mg/kg (100 µL). The first treatment cycle was completed after 2 weeks and was subsequently repeated. The end point was the fourth week after treatment initiation. Tumor volume and body weight were measured every 4 days during the two treatment cycles. At the end of the study, the mice were euthanized, and tumors, as well as major organs, were dissected to assess any potential side effects of the treatment. Tumor cell necrosis, apoptosis, and proliferation were observed using HE, TUNEL, and Ki67 staining, respectively. Post-treatment conditions were examined using B7-H3 and PD-L1 staining, and immune status was assessed using CD8 staining. To further investigate the changes in gene expression post-treatment, three randomly selected tumor samples from each group were subjected to RNA sequencing.

RNA sequencing analysis

Shanghai OE Biotech conducted transcriptomic sequencing and subsequent analyses. The RNA-seq data provided in the FASTQ files were subjected to quality control and visualization using FastQC. HISAT2 was used to map the reads to the human genome assembly, GRCh38. The raw gene expression counts were extracted using these features.

In the differential analysis, data from the BsADC group were compared with the Vehicle group to identify specifically upregulated Differentially Expressed Genes (DEGs) in the BsADC group. The Deseq2 R package was used for DEG identification, selecting those with a log2-fold change (FC)>1 and an adjusted p value (p<0.01). Upregulated DEGs were identified using a significance threshold of p<0.01.

To gain insights into the biological processes, GO enrichment analysis was performed using the ClusterProfiler R package. Statistical significance was determined using Benjamini-Hochberg-adjusted p values of <0.01. The results of the enrichment analysis were visualized using a dot plot function, specifically focusing on the enrichment of biological processes related to graphene oxide.

Statistics

Statistical analyses were performed by using GraphPad Prism V.9 (V.9.5.1). An unpaired two-tailed Student’s t-test was used to compare differences between two independent samples, while ordinary one-way analysis of variance was used to compare differences among multiple groups. The significance levels were set at *p<0.05, **p<0.01, ***p<0.001, and ****p<0.0001. Error bars represent mean±SD.

ResultsPD-L1 and B7-H3 exhibited high expression in head and neck squamous cell carcinoma with limited expression in normal tissues

A comprehensive analysis of 519 cases of head and neck squamous cell carcinoma from TCGA database revealed that both PD-L1 and B7-H3 expression levels were elevated in head and neck squamous cell carcinoma compared with normal tissues, with B7-H3 showing statistically significant differences (figure 1A). While a trend toward lower overall survival was observed in patients with high PD-L1 expression, this was not statistically significant. In contrast, patients with high B7-H3 expression demonstrated a significantly lower overall survival rate than those with low expression (p=0.0055) (figure 1B).

Figure 1

Expression of PD-L1 and B7-H3 in head and neck cancers. (A) Analysis of PD-L1 and B7-H3 expression in head and neck cancer and normal tissue samples using the TCGA database (*p<0.05). (B) Relationship between the expression levels of PD-L1 and B7-H3 and the overall survival rate of head and neck cancer patients. (C) IHC staining of PD-L1 and B7-H3 expression in human LSCC tissues and normal laryngeal tissues (commercial antibodies). Scale bar, 50 µm. (D) IHC staining of PD-L1 and B7-H3 expression in normal human tissues (heart, liver, lung, kidney) (commercial antibodies). Scale bar, 50 µm. (E) Quantitative analysis of PD-L1 and B7-H3 expression in human LSCC tissues and normal laryngeal tissues (Commercial antibodies). Data are presented as mean±SD, and statistical significance was determined by unpaired two-tailed Student’s t-test. p<0.05, p<0.01, p<0.001, and p<0.0001 were considered statistically significant and marked with *, **, ***, and ****, respectively. (F) Flow cytometry showed high expression of PD-L1 and B7-H3 in head and neck cancer cell lines (FADU, SCC15, SNU46, SNU899) (commercial antibodies). IHC, immunohistochemistry; LSCC, laryngeal squamous cell carcinoma; TCGA, the Cancer Genome Atlas.

Furthermore, IHC results demonstrated a significant elevation of PD-L1 and B7-H3 expression in LSCC tissues, contrasting with significantly lower expression in normal laryngeal mucosal epithelium (figure 1C,E and online supplemental figure 1B,D) and other normal tissues, such as the heart, liver, lung, and kidney (figure 1D and online supplemental figure 1E). Moreover, restricted expression of PD-L1 and B7-H3 was observed in other normal human tissues, as illustrated in online supplemental figure 1A (including the brain, cervix, colon, breast, ovary, pancreas, prostate, and skin).

Simultaneously, flow cytometry results indicated positive membrane expression of both PD-L1 and B7-H3 in all four head and neck cell lines (FADU, SCC15, SNU46, and SNU899), emphasizing the widespread expression of these two targets in head and neck cancers (figure 1F and online supplemental figure 1C). Furthermore, knockout SNU46 and SNU899 cell lines using CRISPR/Cas9 technology were included. The results showed a significant decrease in membrane expression at the corresponding target of the knockout cell lines (online supplemental figure 2A). Additionally, we overexpressed PD-L1 and B7-H3 in Hela cell lines, which normally do not express PD-L1 and only express limited B7-H3. Similarly, the results verified the target specificity in other aspect (online supplemental figure 2B).

Construction and characterization of ADCs

The BsADC was constructed by conjugating the complete PD-L1 antibody and B7-H3 scFv with MMAE, with human IgG1 (which was used in the construction of BsAb and ADCs) also incorporated into the BsADC. Each scFv comprised a corresponding light chain (VL) and a heavy chain (VH), connected by three 5-amino acid (G4S) linkers. Figure 2A illustrates the structure of BsADC and its potential mechanism for antitumor activity.

Figure 2

Construction and characterization of ADCs. (A) Schematic diagram illustrating the structure of BsADC and its antitumor mechanism. BsADC targets and binds to corresponding antigens on the tumor cell surface through anti-PD-L1 or B7-H3 scFv, internalizes, and releases MMAE, initiating tumor cell apoptosis. On the other hand, BsADC can competitively bind tumor cell surface PD-L1 through the PD-L1 antibody with PD-1 on activated T cells, preventing immune suppression. Similarly, B7-H3 is also believed to have the same function, though the competitor is not yet clearly defined. Additionally, due to MMAE, when tumor cells die, they transition from non-immunogenic to immunogenic, mediating an antitumor immune response by releasing DAMPs (such as CRT, ATP, HMGB1), activating the adaptive immune response and promoting the death of tumor cells not directly targeted by BsADC. (B) Determination of the avidity of ADCs using Biacore. (C) (UP) Flow cytometry comparing the internalization ability of BsAb, PD-L1 ADC, B7-H3 ADC, and BsADC in SNU46 and SNU899 cell lines. (Below) Internalization quantification, measured by the ratio of FITC intensity at different time points to the 0 hour FITC intensity for each group. (D) IF assay showed the internalization of BsADC over time in SNU46 and SNU899 cells. ADC, antibody-drug conjugate; BsADC, bispecific ADC; IF, immunofluorescence; MMAE, monomethyl auristatin E.

After the completion of ADC conjugation, we initially commenced by corroborating the modification in molecular weight ensuing the conjugation, employing the method of SDS-PAGE gel electrophoresis (online supplemental figure 3A). Subsequently, a preliminary characterization of the BsADC properties was conducted. Mass spectrometry analysis was employed to detect the ADC coupling rate, yielding results showed the average Drug-to-Antibody Ratio (DAR) determination for BsADC was 1.73 (online supplemental figure 2D). Furthermore, the results of SPR avidity detection (figure 2B) revealed that the BsAbs demonstrated robust and comparable avidities for PD-L1 and B7-H3, with KD values of 8.19E-11 and 4.48E-10, respectively. After conjugation with MMAE, BsADC consistently maintained high avidity, with KD values of 1.04E-10 and 8.51E-10 for PD-L1 and B7-H3, respectively, underscoring the sustained binding effectiveness of BsADC.

Figure 3

In vitro antitumor activity of ADCs. (A) SNU46, SNU899, PD-L1 and B7-H3 knock-out SNU46 and SNU899 cells were treated with different doses of BsADC for 72 hours, and IC50 was determined by the MTT assay. (B) Cell proliferation curves for SNU46 and SNU899 cells treated with different drugs over 5 days, assessed by the MTT assay. (C) Quantitative analysis of apoptosis in SNU46 and SNU899 cells treated with different drugs (n=3). (D) Quantitative analysis of cell cycle distribution in SNU46 and SNU899 cells treated with different drugs (n=3). Data are presented as mean±SD, and statistical significance was determined by unpaired two-tailed Student’s t-test. p<0.05, p<0.01, p<0.001, and p<0.0001 were considered statistically significant and marked with *, **, ***, and ****, respectively. (E) Western blot analysis of microtubule damage and apoptosis-related proteins in SNU46 and SNU899 cells treated with specified concentrations of drugs in vitro. ADCs, antibody-drug conjugates; BsADC, bispecific ADC.

Since ADC primarily exerts its efficacy by releasing targeted small-molecule drugs into cancer cells, various methods have been employed to investigate and compare the drug internalization efficiency of BsADC and other groups. Figure 2C illustrates the surface expression over time in the SNU46 and SNU899 cell lines for different groups at 0, 1, 2, 4, and 6 hours after drug incubation. The expression of BsADC on the cell surface decreased over time, reflecting its internalization. Notably, BsADC exhibited comparable internalization efficiency as the other two ADC groups. Additionally, the drug was gradually transferred from the cell membrane to the cytoplasm of both SNU46 and SNU899 cell lines, as observed under a confocal microscopy (100×) over time (figure 2D). And the controls, including PD-L1/B7-H3 knockout SNU46 and SNU899 cell lines, showed no signs of internalization (online supplemental figure 3B).

BsADC showed strong antitumor efficacy in vitro

The in vitro antitumor efficacy of BsADC was assessed using the SNU46 and SNU899 cell lines. BsADC exhibited in vitro cytotoxicity against SNU46 (IC50=115.4 nM) and SNU899 (IC50=73.56 nM) tumor cells (figure 3A). Therefore, we compromised and used 100 nM as the drug dose for subsequent in vitro experiments. Besides, PD-L1 and B7-H3 knockout SNU46 and SNU899 cell lines were also used as controls in BsADC cytotoxicity analysis (shown in figure 3A).

Furthermore, antiproliferative curves plotted after drug administration revealed the inhibition of proliferation in both cell lines, with the BsADC group exhibiting the most significant suppression (figure 3B). Subsequently, the rate of apoptosis was assessed (figure 3C and online supplemental figure 3D). The average apoptosis rates in the BsADC group for the SNU46 and SNU899 cell lines were 42.01% and 8.42%, respectively, which were significantly higher than those in the other two ADC groups (PD-L1 ADC vs B7-H3 ADC, SNU46, 33.04% vs 25.41%, SNU899, 6.06% vs 7.32%).

MMAE blocked the cell cycle in the G2/M phase by inhibiting the polymerization of microtubules43 (verification experiments for the MMAE microtubule protein damage mechanism is detailed in figure 3E and online supplemental figure 3C). Our results demonstrated that all three ADCs impeded the cell cycle progression of LSCCs (online supplemental figure 3E), which was consistent with the previously mentioned induction of apoptosis. Changes in the cell cycle distribution after treatment in the different groups are shown in figure 3D. Comparative results indicated a decrease in the proportion of cells in the G0/G1 phase and a stable S phase, along with a varying degree of increase in the G2/M phase for all ADC groups that targeted tumor cells to deliver MMAE, compared with the Ctrl group. This suggested that the cell cycle was arrested in G2/M phase at the end of the cell cycle. Importantly, BsADC significantly enhanced this effect in both SNU46 (G2/M phase, BsADC vs PD-L1 ADC vs B7-H3 ADC, 33.22% vs 29.54% vs 26.19%) and SNU899 cells (G2/M phase, BsADC vs PD-L1 ADC vs B7-H3 ADC, 56.11% vs 44.59% vs 45.62%).

To delve deeper into the molecular mechanisms underlying the cell cycle arrest and apoptosis induced by BsADC, WB analysis was used to assess the protein expression levels of phosphorylated histone H3 (pHH3) and Caspase-3. The results (figure 3E) revealed that after 48 hours of treatment with BsADC, pHH3 protein levels increased, confirming its role in initiating cell-cycle arrest during the mitotic phase. Additionally, there was an increase in the levels of the apoptotic pathway protein, Caspase-3, indicating the inhibition of DNA repair and initiation of DNA degradation. These findings further supported the hypothesis that BsADC inhibits cell division and promotes apoptosis.

In summary, BsADC demonstrated the best antitumor efficacy in vitro among all groups, effectively inhibiting cell proliferation and division while promoting apoptosis.

BsADC demonstrated robust antitumor efficacy and a favorable safety profile in vivo

We evaluated the therapeutic efficacy and primary safety of BsADCs in vivo. The experimental protocol is shown in figure 4A. Given the utilization of humanized antibody sequences, NCG mice were humanized with human PBMCs via tail injection. Subsequently, the proportion of human immune cells in mouse circulation was measured at 2 and 3 weeks postinjection. The results indicated that at 2 weeks, approximately 15% of the circulating cells in the NCG mice were CD45+T cells, and at 3 weeks, this proportion increased to nearly 40% (figure 4B). Moreover, in both measurements, CD3+T cells accounted for approximately 90%–100% of CD45+T cells, aligning with the immune system profile of NCG mice after human PBMC injection (figure 4C).

Figure 4

In vivo antitumor activity of ADCs. (A) Schematic representation of the experimental procedure. (B, C) Flow cytometry analysis of CD45 and CD3 expression in NCG mice injected with human PBMCs at 2 and 3 weeks. (D, E) Tumor growth curves and body weight curves of SNU899 xenograft NCG mice treated with BsADC and control formulations (n=5). (F, G) Histological images and quantitative analysis of HE, Ki67, TUNEL, PD-L1, and B7-H3 staining in tumor sections from mice at the end of the 28-day treatment. Scale bar represents 50 µm. Data are presented as mean±SD, and statistical significance was determined by unpaired two-tailed Student’s t-test. p<0.05, p<0.01, p<0.001, and p<0.0001 were considered statistically significant and marked with *, **, ***, and ****, respectively. ADCs, antibody-drug conjugates; BsADC, bispecific ADC; PBMCs, peripheral blood mononuclear cells.

The tumor growth curves are illustrated in figure 4D (Tumor resection images are shown in online supplemental figure 4A). Remarkably, tumor growth in the BsADC group exhibited significantly slower progression than that in the other treatment groups, indicating the successful inhibition of LSCC growth. Importantly, at the endpoint, the tumor volume in the BsADC group was significantly smaller than that in the other two ADC groups, highlighting the tumor inhibitory effect of BsADC. Furthermore, the HE, TUNEL, and Ki67 staining results (figure 4F,G) also supported the significant therapeutic effect of BsADCs. In this group, cell necrosis and apoptosis were most significant, whereas cell proliferation experienced the most robust inhibition, consistent with the in vitro studies. Simultaneously, B7-H3 and PD-L1 staining of tumor slices was also performed. The results indicated a significant reduction in the fraction of PD-L1 positive staining %Area in the BsADC group. Although B7-H3 staining showed a significant difference between the BsADC group and the PD-L1 group, there was no statistical difference between the BsADC group and the B7-H3 group (figure 4F,G).

Body weight was measured every 4 d after the initial dose, remaining stable in all groups throughout the treatment period (figure 4E). In addition, no apparent pathological changes were observed in major organs, including the heart, liver, pancreas, lungs, and kidneys (online supplemental figure 4D). Besides, murine PD-L1 ADC toxicity was also detected, and no significant abnormalities were found (online supplemental figure 4E–G, online supplemental text 4.1). In summary, BsADC demonstrated superior antitumor efficacy with no obvious signs of toxicity in vivo.

BsADC significantly stimulated the immune system, enhancing the immune cytotoxicity against LSCC

The aforementioned experiments revealed that BsADC demonstrated a remarkably superior tumor-killing capacity compared with that of all other groups, especially in vivo. Given the immunological relevance of both PD-L1 and B7-H3 as targets and considering previous studies demonstrating that MMAE can induce immunogenic cell death (ICD),44 we further explored the activation of tumor immunity by BsADC.

We performed CD8 staining of the dissected tumors at the endpoint of the in vivo treatment (figure 5A). These findings indicate a significantly higher proportion of CD8+T cell infiltration in the BsADC group than in the other groups, suggesting that BsADC possesses a superior immune activation effect compared with that of the other two ADCs and BsAbs. Besides, humanized tumor-bearing mice showed a significantly better response to BsADC treatment compared with non-humanized NCG mice (online supplemental figure 4B).

Figure 5

ADCs activate immune cytotoxicity. (A) Histological images and quantitative analysis of CD8 staining in tumor sections from mice at the end of the 28-day treatment. Scale bar represents 50 µm. (B, C) In vitro activation of CD69+T cells in co-cultures of SNU46 and SNU899 with human PBMCs, with different treatment groups, and quantitative analysis. (D, E). In vitro activation of monocytes in co-cultures of SNU46 and SNU899 with human PBMCs, with different treatment groups, and quantitative analysis. Data are presented as mean±SD, and statistical significance was determined by unpaired two-tailed Student’s t-test. p<0.05, p<0.01, p<0.001, and p<0.0001 were considered statistically significant and marked with *, **, ***, and ****, respectively. ADCs, antibody-drug conjugates; BsADC, bispecific ADC; PBMCs, peripheral blood mononuclear cells.

To further validate the immune-killing effects of BsADC, we co-cultured SNU46 and SNU899 cells with human PBMCs in vitro. The results indicated a significantly higher proportion of CD69+cells among CD3+T cells in the BsADC group than that in the other groups, with 29.50% and 21.76% in the SNU46 and SNU899 cell lines, respectively (figure 5B,C).

Additionally, we assessed the expression of the monocyte costimulatory molecule CD86. Flow cytometry results demonstrated that compared with other groups, tumor cells treated with BsADC induced a notable increase in CD86 expression in monocytes, particularly in the SNU899 cell line (figure 5D,E). These findings highlight the ability of BsADCs to significantly activate immune cells, thereby enhancing their immune cytotoxicity against tumors.

Moreover, we investigated DAMPs mediating ICD effects, including CRT, ATP, HMGB1, and EIF2α. Initially, the expression of CRT on the surface of tumor cells (figure 6A) revealed a significant increase in the level of CRT on the membrane of tumor cells compared with that in the other treatment and control groups. Furthermore, compared with the Ctrl group, the BsADC group exhibited a significantly more pronounced occurrence of EIF2α phosphorylation, which is the prerequisites for CRT translocation (figure 6D).