Immunotherapy using OX40 agonist antibodies shows great preclinical efficacy in mouse tumor models. But in a clinical setting, OX40 agonist antibody alone or in combination with checkpoint blockade exhibits only modest efficacy due to lack of sufficient activation. We hypothesized that the limited antitumor activity in patients may due to insufficient clustering of OX40 antibody in the tumor. To test this hypothesis, we generated a tetravalent programmed death ligand-1 (PD-L1)/OX40 BsAb by fusing two PD-L1 VHH fragments to the C-terminus of a nonblocking agonistic anti-OX40 antibody. The resulting BsAb had intact function of each parental antibody, including efficiently blocking PD1/PD-L1 interaction and inducing OX40 activation. In addition, this BsAb showed significantly enhanced potency in activation of OX40-expressing T cells when PD-L1–expressing tumor cells or dendrite cells were present, through PD-L1–mediated cross-linking of OX40. Moreover, the BsAb exhibited superior antitumor activities over the parental monospecific antibodies alone or in combination in multiple in vivo tumor models. These results demonstrated a great potential for further clinical development of the potent immunostimulatory PD-L1/OX40 bispecific antibody.

Despite checkpoint-based immunotherapies having achieved unprecedented success in the clinical treatment of several malignancies (1), the majority of patients do not respond or acquire resistance after an early initial response (2). Checkpoint inhibitors targeting programmed death-1 (PD-1) and programmed death ligand-1 (PD-L1) restore the function of antigen-specific T cells, reinforce the antitumor immune response, and have shown limited, clinically validated efficacy. Antigen-specific T cells are regulated by inhibitory signals and costimulatory signals. Therefore, blocking inhibitory pathways in combination with providing costimulatory signals to boost T-cell function may enhance antitumor immunity (3).

OX40 (also known as CD134), a costimulatory molecule transiently expressed on activated human T cells, functions in T-cell activation, expansion, differentiation, generation, and maintenance of memory T cells (4, 5). OX40 belongs to the TNF receptor (TNFR) superfamily. The crystal structure of human OX40 and its ligand complex (OX40L) is a trimeric configuration composed of one homotrimeric OX40L molecule and three OX40 monomers (6). Members of the TNFR superfamily (TNFRSF) require high-order clustering to achieve sufficient downstream signal activation (7–9). In clinic, agonistic antibody monotherapies usually exhibit relatively modest efficacy due to the lack of sufficient activation. Even though anti-OX40 agonist as monotherapy showed very limited therapeutic efficacy (10), anti-OX40 in combination with anti-PD-1, anti-PD-L1, and anti-CTLA-4 revealed improved antitumor efficacy compared with a checkpoint blockade antibody alone in preclinical models (11–14).

Several agonistic anti-OX40 antibodies are currently being evaluated in phase I/II clinical trials either as monotherapies or in combination with other immunotherapies in patients with malignant tumors (15). To date, agonistic OX40 monotherapy has led to tumor regression or stable disease in patients with solid tumors (16) and early clinical data of the combination with a checkpoint blockade are still not available.

We hypothesized that designing a PD-L1/OX40 bispecific antibody, which could induce high-order OX40 clustering and achieve sufficient OX40 activation in a PD-L1–dependent manner, would augment the antitumor effect compared with an anti-PD-L1/anti-OX40 combination. The PD-L1/OX40 bispecific antibody will perform two functions: it will restore the antigen-specific T-cell antitumor effect by PD-L1 inhibition and further enhance the T-cell–mediated antitumor response by engaging OX40 activation. In addition, PD-L1–dependent OX40 activation could potentially reduce systemic toxicity and improve the safety profile compared with anti-OX40 monotherapies.

Generation of PD-L1/OX40 bispecific antibody

Human anti-OX40 antibodies were obtained from a fully human synthetic yeast display library (Adimab LLC; ref. 17). Anti-human PD-L1 single domain antibody (sdAb) was obtained from llamas immunized with PD-L1 antigen and screened via phage display. The heavy chain and light chain amino acid sequences of PD-L1/OX40 are depicted in WO2020/151761 (sequence IDs #1 and # 7). Pogalizumab was produced in-house from publicly available sequences (World Health Organization proposed INN list 114). PD-L1/OX40 was a recombinant IgG2 bispecific antibody constructed by linking PD-L1 sdAb to the C-terminal of an agonistic anti-OX40 antibody by a flexible (G4S)2 linker. Heavy chain- and light chain–expressing vectors were transiently cotransfected and expressed in Chinese hamster ovary (CHO, Lonza). The monospecific antibodies were purified by standard methods using protein A chromatography. Bispecific antibodies were purified using protein A chromatography followed by ion exchange chromatography. Subsequent size-exclusion chromatography (Supplementary Fig. S1A) and liquid chromatography-mass spectrometry analyses demonstrated a high purity and a lack of aggregate. Other biophysical properties, such as accelerated stability, were also tested (Supplementary Fig. S1B). hIgG control for the in vivo efficacy experiments was human IgG isolated from pooled serum or plasma (Equitech-Bio).

Binding affinity detection by surface plasmon resonance

The affinity of PD-L1/OX40 for its target proteins was measured using a Biacore T200 (GE Healthcare). For PD-L1 affinity detection, anti-human Fab IgG was immobilized onto a CM5 sensor chip and PD-L1/OX40 was captured onto the sensor for approximately 100 RU. Afterwards, serially 2-fold diluted human PD-L1 (0–20 nmol/L) was subsequently run over the sensor chip. For OX40 affinity detection, PD-L1/OX40 was captured via a human PD-L1 preimmobilized CM5 sensor for approximately 200 RU, 2-fold serial dilutions of human OX40 (0–200 nmol/L) were subsequently run over the sensor chip. Each target protein was injected with association time of 180 seconds and a pulse injection of 10 mmol/L glycine-HCl, pH 1.5, was used for surface regeneration. The binding affinities of antibodies with antigens were fitted with a 1:1 binding model using BiaEvaluation 3.1.

Cell lines

Raji, NCl-H292, and LoVo cell lines were obtained from the ATCC. CHO-S cells were purchased from Lonza and transfected to stably express human PD-L1 or human OX40 according to the manufacturer's instructions using the Freedom CHO-S Kit (Invitrogen). PD-L1 or OX40 coding cDNA was cloned into pCHO1.0 (Invitrogen). Human PD-L1–expressing Raji cells (Raji-PD-L1) were generated by stable transfection of PD-L1 into Raji cells. Human OX40-expressing NF-κB-Luc rep Jurkat reporter cells were generated by stable transfection of OX40 into a NF-κB-Luc rep Jurkat reporter cell line (Promega). Human PD-L1–expressing MC38 cell line was constructed by Nanjing Galaxy Biopharma Co, Ltd. Briefly, human PD-L1 encoding cDNA cloned into pCHO1.0 (Invitrogen) was stably transfected into MC38 (ATCC). Frozen human peripheral blood mononuclear cells (PBMC) were purchased from ALLCELLS.

Cell-based binding

Human CD4+ T cells were isolated from PBMCs (ALLCELLS) with the EasySep human CD4+ T-cell Enrichment Kit (Stemcell Technologies). Cell suspension (50 μL) at a density of 1 × 105 cells per well of human PD-L1–expressing CHO-S (CHO-S/hPD-L1), human OX40-expressing CHO-S (CHO-S/hOX40), NCl-H292, or human primary CD4+ T cells were seeded into 96-well U-bottom Plates (Corning) and incubated with 50 μL of serially diluted antibodies for 30 minutes at 4°C. Then, cells were centrifuged at 400 × g for 5 minutes and washed once in 150 μL PBS. Cells were centrifuged once again at 400 × g for 5 minutes and the supernatants were discarded. The cells were resuspended in 100 μL of PE-conjugated goat anti-human IgG Fc secondary antibodies (Southern Biotech) and incubated on ice for approximately 30 minutes in the dark. Cells were washed in 150 μL of PBS three times and resuspended in 50 μL of PBS buffer. Fluorescence values were measured with an Accuri C6 plus Flow Cytometry (BD Biosciences).

Blocking assays

Human PD-L1–expressing or human OX40-expressing CHO-S cells at a density of 2 × 105 cells per well were seeded into 96-well U-bottom plates and incubated with serially diluted antibodies containing biotinylated PD-1/PDCD1 (Acro Biosystems) or biotinylated OX40L (Acro Biosystems) for 30 minutes at 4°C. The cells were centrifuged at 400 × g for 5 minutes and washed in PBS three times. After the final wash, cells were resuspended in Streptavidin-R-Phycoerythrin (SAPE, Thermo Fisher Scientific) and incubated at 4°C for 30 minutes. Cells were washed in PBS twice, and then reconstituted in PBS. Fluorescence values were measured with an Accuri C6 plus Flow Cytometry (BD Biosciences).

PD-1/PD-L1 blockade bioassay

Human PD-L1–expressing CHO-K1 cells and PD-1 effector cells, which express human PD-1 and a luciferase reporter driven by a NFAT response element, were provided by the PD-1/PD-L1 Blockade Bioassay (Promega). PD-L1/CHO-K1 cells at a density of 4 × 104 cells per well were coincubated with PD-1 effector cells at a density of 4 × 104 cells per well in the presence of serially diluted antibodies for 6 hours. Assays were performed according to the manufacturer's instructions (Promega).

Cell–cell bridging assay

CHO-S/hPD-L1 and CHO-S/hOX40 cells were differentially labeled with CellTracker Deep Red (CTDR, Thermo Fisher Scientific) and Cell Trace CFSE (Invitrogen), respectively. The CTDR-labeled CHO-S/hPD-L1 cells, at a density of 3 × 105 cells per well, were incubated with PD-L1/OX40 or control antibodies (100 nmol/L) at 4°C for 30 minutes. The cells were then washed in PBS four times. CFSE-labeled CHO-S/hOX40 cells, at a density of 2 × 105 cells per well, were added into each well and incubated at room temperature for 1 hour followed by flow cytometry analysis. FCS files were exported and analyzed in FlowJo Software.

OX40-NF-κB-Luc Rep Jurkat reporter cell assay

NCl-H292 cells, at a density of 4 × 104 cells per well, and OX40-expressing NF-κB-Luc rep Jurkat reporter cells, at a density of 2 × 104 cells per well, were added into 96-well flat white-bottom plates (Nunclon) and incubated with serially diluted antibodies at 37°C for 12 hours. Then, samples were equilibrated to room temperature, the Bio-Glo Luciferase Assay Reagent (Promega) was added, and relative luminescence units were measured using Spectra Max I3 Multi-mode Microplate Reader (Molecular Devices).

Mixed leukocyte reaction assay

Monocyte-derived immature dendrite cells (DC) isolated from human PBMCs (ALLCELLS) were maintained in AIM V medium CTS medium supplemented with GM-CSF (10 ng/mL, R&D Systems) and IL4 (20 ng/mL, R&D Systems) at 37°C and 5% CO2 for 5 days. DCs were supplemented with TNFα (1,000 U/mL, 10 ng/mL, R&D Systems), IL1β (5 ng/mL, R&D Systems), IL6 (10 ng/mL, R&D Systems), and 1 mmol/L PGE 2 (Tocris) and incubated at 37°C and 5% CO2 for another 2 days. Human CD4+ T cells isolated from PBMCs (ALLCELLS) by the EasySep Human CD4+ T-cell Enrichment Kit (Stemcell Technologies) were preactivated by Dynabeads CD3/CD28 (Gibco) for 3 days. Preactivated CD4+ T cells, at a density of 1 × 105 cells per well, and allogeneic monocyte-derived DCs, at a density of 1 × 104 cells per well, were seeded into 96-well round-bottom plates (Corning) and incubated with serially diluted antibodies. Three days later, supernatants were harvested and subjected to IL2 and IFNγ detection by Human IL2 Kit (CISBIO) and Human IFNγ Kit (CISBIO).

Primary T-cell activation

Human CD4+ T cells isolated from PBMCs by the EasySep Human CD4+ T-cell Enrichment Kit (Stemcell Technologies) were prestimulated with Dynabeads CD3/CD28 (Gibco). Three days later, prestimulated T cells, at a density of 1 × 105 cells per well, and Raji-PD-L1 cells, at a density of 2 × 104 cells per well, were cocultured with serially diluted antibodies and Staphylococcus Enterotoxin E (1 ng/mL, SEE, Toxin Technology). Three days later, IL2 and IFNγ in supernatants were detected using the methods described for mixed leukocyte reaction (MLR) assay.

In vivo mouse models

All animals were maintained under pathogen-free conditions in the Experimental Animal Center of Innovent Biologics Co., Ltd. All animal-related experiments were approved by the Animal Use and Care Committee of Innovent Biologics. For all in vivo studies, the tumor volume in mm3 was calculated using the following formula: (width)2 × length/2. Animals were euthanized when tumor volume exceeded 2,000 mm3 or when the percentage of body weight loss exceeded 20%.

Humanized LoVo NOG tumor model

Human colon cancer LoVo cells were cultured according to the ATCC instructions (F-12K Medium, ATCC). LoVo cells (2.5 × 106 cells) and human PBMCs (0.625 × 106 cells) suspended in 0.2 mL PBS were injected subcutaneously into the right dorsal flank of female NOG mice (Beijing Vital River Laboratory Animal Technology Co.). Three days later, when tumor volume reached approximately 50 mm3, mice were randomly assigned to groups (N = 7). Mice were intraperitoneally injected with test antibodies on days 3, 7, 10, and 14. Tumor dimensions and body weights were measured twice a week.

PD-L1–expressing MC38 model in hPD-L1/hOX40 transgenic mice

Human OX40 transgenic mice (hOX40tg) were developed in a C57BL/6 background by replacing the murine OX40 extracellular domain coding sequences with human OX40 coding sequences, keeping the mouse transmembrane and cytoplasmic domains (Shanghai Model Organisms Center, Inc.). Human PD-L1 transgenic mice (hPD-L1tg) were developed by Shanghai Model Organisms Center. Mice expressing the human PD-L1 and human OX40 (hPD-L1/hOX40tg) were generated by breeding hOX40tg together with hPD-L1tg (Shanghai Model Organisms Center, Inc.). The PD-L1–expressing MC38 cells (2 × 106 cells) suspended in 0.2 mL PBS were injected subcutaneously into the right dorsal flank of hPD-L1/hOX40tg female mice. Five days later, mice were randomly assigned to groups (n = 7) on the basis of tumor volume (range, 47.02–112.67 mm3). Mice were administrated test antibodies intraperitoneally on days 5, 9, 12, and 16. Tumor dimensions and body weights were measured twice a week.

Humanized NCl-H292 NOG tumor model

Human PBMCs (2.5 × 106 cells) suspended in 0.2 mL PBS were injected into the orbital vein of female NOG mice (Beijing Vital River Laboratory Animal Technology Co.). Human lung cancer NCl-H292 cells were cultured according to the ATCC instructions (RPMI1640 Medium, ATCC). Five days after PBMC injection, NCl-H292 cells (2.5 × 107 cells) suspended in 0.2 mL PBS were injected subcutaneously into the right dorsal flank of NOG mice. On day 6, mice were randomly assigned to treatment groups (n = 8). Mice were administrated with test antibodies intraperitoneally on days 6, 10, 14, and 17. Tumor dimensions and body weights were measured twice a week.

Pharmacokinetics in mice

Single dose of PD-L1/OX40 at 10 mg/kg was injected via tail vein into female Balb/C mice. Serum concentration of PD-L1/OX40 at various timepoints was determined using a sandwich ELISA. Noncompartmental analysis was conducted with PK Sover 2.0.

Statistical analysis

Statistical analyses of in vitro data were performed using GraphPad Prism (version 6.01). For specific information on the statistical tests performed, please refer to the figure legends. Values of P < 0.05 were statistically significant (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001).

Characterization of the binding capability of PD-L1/OX40

The tetravalent bispecific PD-L1/OX40 was constructed by fusing two anti-PD-L1 sdAbs to the C-terminal of a weak agonistic anti-OX40 IgG2 antibody (Fig. 1A). IgG2 constant region was adopted to reduce the risk of antibody-dependent cellular cytotoxicity against OX40-positive T cells or PD-L1–independent activation of OX40-positive cells via Fcγ receptor (FcγR) mediated cross-linking. The physical binding kinetics and affinity of PD-L1/OX40 to PD-L1 or OX40 proteins was analyzed by surface plasmon resonance (SPR; Fig. 1B). The binding affinity of PD-L1/OX40 to recombinant OX40 was 99.76 nmol/L, which was nearly 10-fold weaker than that of PD-L1/OX40 to PD-L1.

Figure 1.

The molecular format and binding profiles of PD-L1/OX40. A, Schematic of PD-L1/OX40 bispecific format. Heavy chains and light chains of OX40 are shown in dark blue and in light blue, respectively. Anti-PD-L1 sdAbs (in red) were fused to the C-terminal of OX-40 antibody by (GGGGS)2 linkers. B, Binding affinity of PD-L1/OX40 to soluble PD-L1 or soluble OX40 was determined by SPR. The binding curve, binding affinity, and kinetic constants of PD-L1/OX40 to soluble PD-L1 antigens are shown on the left. The binding curve, binding affinity, and kinetic constants of PD-L1/OX40 to soluble OX40 are shown on the right. KD, binding dissociation equilibrium constants; Ka, kinetic association rate; Kd, kinetic dissociation rate. C, Binding of PD-L1/OX40 to PD-L1–expressing CHO-S and OX40-expressing CHO-S, respectively. All the data are representative of at least three independent experiments. mean ± SD.

Figure 1.

The molecular format and binding profiles of PD-L1/OX40. A, Schematic of PD-L1/OX40 bispecific format. Heavy chains and light chains of OX40 are shown in dark blue and in light blue, respectively. Anti-PD-L1 sdAbs (in red) were fused to the C-terminal of OX-40 antibody by (GGGGS)2 linkers. B, Binding affinity of PD-L1/OX40 to soluble PD-L1 or soluble OX40 was determined by SPR. The binding curve, binding affinity, and kinetic constants of PD-L1/OX40 to soluble PD-L1 antigens are shown on the left. The binding curve, binding affinity, and kinetic constants of PD-L1/OX40 to soluble OX40 are shown on the right. KD, binding dissociation equilibrium constants; Ka, kinetic association rate; Kd, kinetic dissociation rate. C, Binding of PD-L1/OX40 to PD-L1–expressing CHO-S and OX40-expressing CHO-S, respectively. All the data are representative of at least three independent experiments. mean ± SD.

Close modal

To confirm the binding ability of PD-L1/OX40 to cell surface PD-L1 and OX40 proteins, flow cytometry was performed in target-overexpressing cell lines, primary cells, and tumor cell lines. PD-L1/OX40 showed dose-dependent binding to human PD-L1–expressing CHO-S cells, OX40-expressing CHO-S cells (Fig. 1C and D), PD-L1–positive NCl-H292 cell lines (Supplementary Fig. S2A), and human primary T cells (Supplementary Fig. S2B). Taken together, these results confirmed that PD-L1/OX40 is capable of binding to cell surface PD-L1 and OX40. These results also demonstrated the binding of PD-L1/OX40 to the double targets was slightly weaker than that of the corresponding parental anti-PD-L1 and anti-OX40. The binding between human PD-1 and PD-L1–expressing CHO-S was competitively inhibited by PD-L1/OX40 (Fig. 2A). The PD-L1 blocking ability of PD-L1/OX40 was also analyzed in a PD-1/PD-L1 blockade bioassay by measuring luminance of the downstream NFAT reporter. As expected, either PD-L1/OX40 or anti-PD-L1 or the combination of anti-PD-L1 and anti-OX40 could induce the NFAT activation by blocking the interaction between PD-1 and PD-L1 (Fig. 2B). The binding between OX40L and OX40-expressing CHO-S cells was not competitively blocked by PD-L1/OX40, whereas the anti-OX40 blocking antibody, pogalizumab, could block the binding of OX40L in a dose-dependent manner (Fig. 2C). The OX40-expressing NF-κB-Luc rep Jurkat luciferase reporter cells were coincubated with serially diluted antibodies containing human OX40L (20 μg/mL) for 12 hours. As expected, neither anti-PD-L1 nor IgG2 (isotype control) affected the OX40 ligand–induced activation of the downstream NF-κB signaling, whereas either the PD-L1/OX40 or the parental anti-OX40 further enhanced the NF-κB signaling in a dose-dependent manner (Fig. 2D). In contrast, pogalizumab could block OX40L-induced NF-κB in a dose-dependent manner (Fig. 2D).

Figure 2.

The blocking properties of PD-L1/OX40. A, The binding curves of biotinylated PD-1 to PD-L1–expressing CHO-S in the presence of different concentrations of PD-L1/OX40 or anti-PD-L1. IgG2 is the isotype control. B, The PD-L1 blocking ability of PD-L1/OX40 or the parental monospecific antibodies or in combination was analyzed in PD-1/PD-L1 blockade bioassay by measuring luminance of the downstream NFAT reporter. C, OX40/OX40L blocking assay was conducted for PD-L1/OX40, anti-OX40, anti-PD-L1, and IgG2 with biotinylated OX40L. After incubation and washing, biotinylated OX40L was detected by SAPE conjugate. D, Jurkat OX40 NF-κB reporter cells were incubated with OX40L and PD-L1/OX40, the parental mAb alone or in combination, followed by luciferase activity determination. All the data are representative of at least three independent experiments. mean ± SD.

Figure 2.

The blocking properties of PD-L1/OX40. A, The binding curves of biotinylated PD-1 to PD-L1–expressing CHO-S in the presence of different concentrations of PD-L1/OX40 or anti-PD-L1. IgG2 is the isotype control. B, The PD-L1 blocking ability of PD-L1/OX40 or the parental monospecific antibodies or in combination was analyzed in PD-1/PD-L1 blockade bioassay by measuring luminance of the downstream NFAT reporter. C, OX40/OX40L blocking assay was conducted for PD-L1/OX40, anti-OX40, anti-PD-L1, and IgG2 with biotinylated OX40L. After incubation and washing, biotinylated OX40L was detected by SAPE conjugate. D, Jurkat OX40 NF-κB reporter cells were incubated with OX40L and PD-L1/OX40, the parental mAb alone or in combination, followed by luciferase activity determination. All the data are representative of at least three independent experiments. mean ± SD.

Close modal

PD-L1/OX40 bridges PD-L1–positive cells and OX40-positive cells

To determine the simultaneous binding of PD-L1/OX40 to both PD-L1–positive cells and OX40-positive T cells, different fluorescence dye labeled PD-L1–expressing CHO-S and OX40-expressing CHO-S cells were mixed in ratio of 1:1 and coincubated with bispecific antibody and respective monospecific antibodies alone or in combination, followed by double positive event detection by flow cytometry. In the presence of PD-L1/OX40, there was a significant increase in the proportion of double positive cell complex, indicating the bispecific antibody engaged targets on two different cells simultaneously (Fig. 3A). Beyond engineered cell lines, similar results were observed by using OX40-expressing NF-κB-Luc rep Jurkat reporter cells as effector cells and PD-L1–positive human lung cancer NCl-H292 cells as cancer cells to evaluate the cross-linking ability of PD-L1/OX40 (Supplementary Fig. S3A).

Figure 3.

PD-L1/OX40 demonstrated PD-L1–dependent OX40 activation. A, CHO-S/hPD-L1 cells and CHO-S/hOX40 cells were labeled with CFSE or CTDR, respectively, and then were mixed at 1:1 ratio and incubated with antibodies for 1 hour. CFSE and CTDR double positive events were detected by flow cytometry. Mean percentages of double positive events are shown for each treatment group (left). n = 3, mean ± SD. Representative flow plot showing the percentage of double positive cell complexes (right). Numbers represent the percentage of cells within the gates. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001; one-way ANOVA with Tukey post hoc test. B, PD-L1 relative expression levels were evaluated in NCI-H292 and Raji-PD-L1 cells by flow cytometry. Jurkat OX40 NF-κB reporter cells were incubated with PD-L1/OX40, the parental mAb alone or in combination, and cocultured with Raji (C), NCI-H292 (D), or Raji-PD-L1 (E), followed by luciferase activity determination. n = 3, mean ± SD, two-tailed t test.

Figure 3.

PD-L1/OX40 demonstrated PD-L1–dependent OX40 activation. A, CHO-S/hPD-L1 cells and CHO-S/hOX40 cells were labeled with CFSE or CTDR, respectively, and then were mixed at 1:1 ratio and incubated with antibodies for 1 hour. CFSE and CTDR double positive events were detected by flow cytometry. Mean percentages of double positive events are shown for each treatment group (left). n = 3, mean ± SD. Representative flow plot showing the percentage of double positive cell complexes (right). Numbers represent the percentage of cells within the gates. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001; one-way ANOVA with Tukey post hoc test. B, PD-L1 relative expression levels were evaluated in NCI-H292 and Raji-PD-L1 cells by flow cytometry. Jurkat OX40 NF-κB reporter cells were incubated with PD-L1/OX40, the parental mAb alone or in combination, and cocultured with Raji (C), NCI-H292 (D), or Raji-PD-L1 (E), followed by luciferase activity determination. n = 3, mean ± SD, two-tailed t test.

Close modal

PD-L1/OX40 actives OX40 downstream signaling in a PD-L1–dependent manner

We hypothesized that the binding affinity difference between the two arms of PD-L1/OX40 and their antigens would drive the bispecific molecules more toward the PD-L1–positive cells. Thus, the proximity of PD-L1 molecules on tumor cell surface would assist OX40 clustering on T cells when the bispecific molecules bridge two cell types.

The relative PD-L1 expression levels in NCl-H292 cells and engineered PD-L1–expressing Raji cells were determined by flow cytometry (Fig. 3B). To assess the ability of PD-L1/OX40 to simultaneously bind to PD-L1–positive cells and T cells and induce PD-L1–dependent OX40 costimulation activity, OX-40–expressing NF-κB-Luc rep Jurkat luciferase reporter assay cells were coincubated with PD-L1–negative Raji cells (Fig. 3C), NCl-H292 cells (Fig. 3D), or PD-L1–expressing Raji cells (Fig. 3E) with serially diluted bispecific antibodies or parental monospecific antibodies. In the presence of PD-L1–positive cells, but not PD-L1–negative Raji cells, PD-L1/OX40 induced activation of the OX40 downstream NF-κB signaling pathway in a dose-dependent manner, whereas the respective monotherapies or the combination therapy did not (Fig. 3C–E). Furthermore, the maximal activation of OX40 downstream signaling was positively correlated with PD-L1 expression levels. PD-L1/OX40-induced OX40 activation signaling was lower in primary NCl-H292 cells, which expressed lower levels of PD-L1 compared with the engineered PD-L1–expressing Raji cells (Fig. 3B).

PD-L1/OX40 activates primary T cells more efficiently

The ability of PD-L1/OX40 to activate primary T cells was evaluated in MLR assays. T-cell activation was evaluated by the secretion of IL2 and IFNγ. When prestimulated CD4+ T cells were incubated with matured DCs expressing PD-L1, T-cell activation was blocked by the interaction between PD-L1 on DCs and PD-1 on T cells. Anti-OX40 alone did not show any effect on T-cell activation, as expected (Fig. 4A and B). Anti-PD-L1 restored T-cell activation by blocking the interaction between PD-L1 and PD-1(Fig. 4A and B). In contrast, anti-PD-L1/OX40 or the combination of anti-PD-L1 with anti-OX40 resulted in increased secretion of IL2 and IFNγ in a dose-dependent manner compared with anti-PD-L1 alone (Fig. 4A and B). Furthermore, PD-L1/OX40-induced T-cell activation via PD-L1 cross-linking was evaluated. In the presence of PD-L1–expressing Raji cells, PD-L1/OX40 or combination therapy demonstrated increased IL2 and IFNγ secretion compared with the isotype control, whereas anti-PD-L1 or agonistic anti-OX40 alone showed only a slight increase over the IgG2 isotype control (Fig. 4C and D).

Figure 4.

PD-L1/OX40 displayed superior bioactivity than combination in primary human assays. The T-cell–activating ability of PD-L1/OX40 was tested in MLR assay and T-cell activation assay. A and B, In an MLR assay, matured DCs and anti-CD3/CD28 dynabeads stimulated CD4+ T cells were cocultured in different concentration of antibodies for 72 hours. IL2 (A) and IFNγ (B) were assessed by ELISA. C and D, T-cell activation assay. PD-L1–expressing Raji cells and anti-CD3/CD28 dynabeads stimulated CD4+ T cells were cocultured in different concentration of antibodies for 72 hours. Secretion of IL2 (C) and IFNγ (D) was analyzed by ELISA. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001; one-way ANOVA with Tukey post hoc test.

Figure 4.

PD-L1/OX40 displayed superior bioactivity than combination in primary human assays. The T-cell–activating ability of PD-L1/OX40 was tested in MLR assay and T-cell activation assay. A and B, In an MLR assay, matured DCs and anti-CD3/CD28 dynabeads stimulated CD4+ T cells were cocultured in different concentration of antibodies for 72 hours. IL2 (A) and IFNγ (B) were assessed by ELISA. C and D, T-cell activation assay. PD-L1–expressing Raji cells and anti-CD3/CD28 dynabeads stimulated CD4+ T cells were cocultured in different concentration of antibodies for 72 hours. Secretion of IL2 (C) and IFNγ (D) was analyzed by ELISA. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001; one-way ANOVA with Tukey post hoc test.

Close modal

PD-L1/OX40 displays potent antitumor efficacy in mouse models

To assess the antitumor efficacy of PD-L1/OX40 against PD-L1–positive cancer cells, multiple mouse tumor models were established. Tumor cell lines used in the mouse models were subjected to evaluate the relative expression levels of PD-L1 by flow cytometry (Fig. 5A).

Figure 5.

The antitumor activity of PD-L1/OX40 in multiple in vivo mouse tumor models. The antitumor effect of PD-L1/OX40 in LoVo (colon), NCI-H292 (lung), and PD-L1–expressing MC38 cell lines was tested. A, PD-L1 relative expression level was evaluated in LoVo, PD-L1–expressing MC38, and NCI-H292 cell lines by flow cytometry. B, The antitumor effects of PD-L1/OX40, anti-OX40, anti-PD-L1, or hIgG were evaluated in LoVo humanized model. LoVo and human PBMCs were coseeded in NPG mice subcutaneously, PD-L1/OX40, anti-OX40, anti-PD-L1, or hIgG were administrated intraperitoneally. Tumor volume was recorded twice a week. n = 7. C, The antitumor effects of PD-L1/OX40, anti-OX40, anti-PD-L1, or hIgG were evaluated in hPD-L1/hOX40tg mice seeded by PD-L1–expressing MC38 cells. Tumor volume was recorded twice a week. n = 7. D, The antitumor effects of PD-L1/OX40, anti-OX40, anti-PD-L1, or hIgG were evaluated in NCI-H292 humanized model. NCl-H292 and human PBMCs were coseeded in NOG mice subcutaneously. PD-L1/OX40, anti-OX40, anti-PD-L1, or hIgG were administrated intraperitoneally with low, middle, and high dose. Tumor volume was recorded twice a week. n = 8. Bars represent mean ± SEM. Significances between PD-L/OX40 and other groups are marked with *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001; two-way ANOVA with Tukey post hoc test.

Figure 5.

The antitumor activity of PD-L1/OX40 in multiple in vivo mouse tumor models. The antitumor effect of PD-L1/OX40 in LoVo (colon), NCI-H292 (lung), and PD-L1–expressing MC38 cell lines was tested. A, PD-L1 relative expression level was evaluated in LoVo, PD-L1–expressing MC38, and NCI-H292 cell lines by flow cytometry. B, The antitumor effects of PD-L1/OX40, anti-OX40, anti-PD-L1, or hIgG were evaluated in LoVo humanized model. LoVo and human PBMCs were coseeded in NPG mice subcutaneously, PD-L1/OX40, anti-OX40, anti-PD-L1, or hIgG were administrated intraperitoneally. Tumor volume was recorded twice a week. n = 7. C, The antitumor effects of PD-L1/OX40, anti-OX40, anti-PD-L1, or hIgG were evaluated in hPD-L1/hOX40tg mice seeded by PD-L1–expressing MC38 cells. Tumor volume was recorded twice a week. n = 7. D, The antitumor effects of PD-L1/OX40, anti-OX40, anti-PD-L1, or hIgG were evaluated in NCI-H292 humanized model. NCl-H292 and human PBMCs were coseeded in NOG mice subcutaneously. PD-L1/OX40, anti-OX40, anti-PD-L1, or hIgG were administrated intraperitoneally with low, middle, and high dose. Tumor volume was recorded twice a week. n = 8. Bars represent mean ± SEM. Significances between PD-L/OX40 and other groups are marked with *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001; two-way ANOVA with Tukey post hoc test.

Close modal

Human LoVo xenograft tumors were established in NOG mice inoculated with human PBMCs. A suboptimal dose of the parental anti-PD-L1 (75 kDa) at 0.01 mg/kg showed moderate tumor growth inhibition, the parental agonistic, anti-OX40 (150 kDa), at 0.02 mg/kg did not show any antitumor effect, and the combination of anti-OX40 and anti-PD-L1 did not enhance the antitumor growth activity of anti-PD-L1 monotherapy. However, an equimolar dose of anti-PD-L1/OX40 (177 kDa) at 0.023 mg/kg resulted in an early onset of tumor reduction and more profound tumor regression (Fig. 5B) without any signs of body weight reduction, which suggests less concerns in systemic toxicity (Supplementary Fig. S4A).

We also evaluated the antitumor effects of PD-L1/OX40 in a human PD-L1–expressing MC38 cells syngeneic model established in hPD-L1/hOX40tg mice. The parental anti-PD-L1, at a dose of 2.5 mg/kg, or anti-OX40, at a dose of 5.0 mg/kg, showed moderate tumor growth inhibition and the combination further enhanced the antitumor growth activity compared with anti-OX40 monotherapy. Notably, an equimolar dose of anti-PD-L1/OX40 resulted in the most pronounced tumor regression (Fig. 5C) without body weight reduction (Supplementary Fig. S4B). Dose titration experiments in a NCl-H292 humanized mouse model clearly showed that PD-L1/OX40 inhibited tumor growth in a dose-dependent manner (Fig. 5D). Furthermore, compared with either monotherapies or the combinatory therapy, the BsAb exhibited significantly enhanced antitumor effect, especially in higher dosage groups (Fig. 5D; Supplementary Fig. S5). In summary, PD-L1/OX40 demonstrated more potent antitumor activity than the combination of anti-PD-L1 and anti-OX40 in different mouse models of various cancer types.

Pharmacokinetics in mice

Pharmacokinetics of PD-L1/OX40 were evaluated in female Balb/C mice following a single-dose administration via tail vein injection at 10 mg/kg. A summary of the pharmacokinetics parameters is shown in Table 1. The plasma concentrations were found to decline in a biphasic manner (Supplementary Fig. S6A and S6B). The elimination half-life was more than 12 days in mouse (Table 1). In conclusion, the pharmacokinetics parameters of PD-L1/OX40 demonstrate that it behaves similar to a conventional mAb.

Table 1.

Pharmacokinetic profile of PD-L1/OX40 in mice.

PK summary of PD-L1/OX40 in Balb/C mouse following a single 10 mg/kg i.v. administration
CompoundCmax (μg/mL)AUC0–t (μg*hour/mL)AUC0–inf_obs (μg*hour/mL)MRT0-inf_obs (hour)T½ (hour)CL (μg)/(μg/mL)/hourVss (μg)/(μg/mL)
PD-L1/OX40 Coated with PD-L1 199 29,168 44,658 448 295 0.004 2.007 
PD-L1/OX40 Coated with OX40 249 38,181 66,030 558 386 0.003 1.716 
PK summary of PD-L1/OX40 in Balb/C mouse following a single 10 mg/kg i.v. administration
CompoundCmax (μg/mL)AUC0–t (μg*hour/mL)AUC0–inf_obs (μg*hour/mL)MRT0-inf_obs (hour)T½ (hour)CL (μg)/(μg/mL)/hourVss (μg)/(μg/mL)
PD-L1/OX40 Coated with PD-L1 199 29,168 44,658 448 295 0.004 2.007 
PD-L1/OX40 Coated with OX40 249 38,181 66,030 558 386 0.003 1.716 

Abbreviations: CL, clearance rate; Cmax, maximum plasma concentration; i.v., intravenous; MRT, mean residence time; PK, pharmacokinetics; T1/2, drug half-life; Vss, volume of distribution at steady state.

Immunotherapies involving immune checkpoint inhibitors, such as anti-PD-1/PD-L1 antibodies, have achieved remarkable clinical benefits especially in certain types of cancers, including Hodgkin lymphoma and melanoma (1). Unfortunately, in the majority of patients with advanced cancers, the overall response rate is low and the response is even poorer in cancers with inhibitory immune signatures or insufficient immune cell infiltrations, such as colorectal and prostate cancers (2). It is well known that T-cell activity is sophisticatedly regulated by multiple costimulatory/inhibitory signals; thus, the combined targeting of different signaling pathways is being actively explored in ongoing clinical trials, and some have shown promising clinical benefits over monotherapies (20, 21). OX40 stimulation with agonistic anti-OX40 antibodies has demonstrated augmented effects on T-cell activities by elevating levels of cytokines/prosurvival molecules. In addition, OX40 agonists have shown synergistic antitumor effects with PD-1/PD-L1 blockers in preclinical studies, and the combined therapy is being investigated in several clinical trials (20, 22). A novel CTLA4/OX40 bispecific antibody in an IgG1 format, targeting a checkpoint inhibitor and a T-cell costimulatory receptor simultaneously, has demonstrated tumor-specific and long-term immune memory in preclinical study. Currently the bispecific molecule is being tested in a phase I clinical trial (23).

Here, we describe a novel PD-L1/OX40 bispecific antibody with a desirable cotargeting strategy. On the basis of design rationales and our preclinical characterization, this molecule is speculated to elicit its distinctive antitumor profile through at least four mechanisms of action: first, PD-L1 binding blocks the PD-1/PD-L1 interaction, alleviating T-cell exhaustion; second, the agonistic stimulation of OX40 augments T-cell activation signaling; third, simultaneous binding to both OX40 and PD-L1 physically bridges activated T cells to tumor cells; and finally, OX40 clustering and activation is conditionally and efficiently mediated by the PD-L1–expressed tumor microenvironment. These mechanisms collectively explain the superior antitumor efficacy of the PD-L1/OX40 bispecific antibody compared with anti-PD-L1 and anti-OX40 monotherapies or their combination in our in vitro functional studies and in vivo models (Figs. 4 and 5).

Previous studies have indicated that therapeutic antibodies sharing overlapped binding epitopes with OX40L on OX40 may lead to a profound decrease in the differentiation of activated B cells into highly Ig-secreting plasma cells (24). Another study showed that OX40L-blocking antibody could significantly inhibit thymic stromal lymphopoietin-induced inflammation in the lung and skin (25). In contrast, the OX40 binding moiety in our molecule does not block the OX40L–OX40 interaction, hence this design strategy could allow our molecule to have a more favorable safety profile compared with OX40L-blocking agonistic OX40 monospecific antibodies.

OX40 is a member of the TNFRSF and forms a homotrimeric complex upon binding to its natural ligand, OX40L. In addition, TNFRSF members usually require clustering of multiple such homotrimers to trigger sufficient signaling (26). In this respect, bivalent agonistic antibodies heavily depend on Fc-mediated binding to FcγR for high-hierarchy cross-linking and clustering of OX40. However, with the presence of 100- to 1,000-fold endogenous IgG molecules, the FcγR-dependent clustering in vivo is less efficient (27). In addition, while most adverse effects induced by anti-OX40 monotherapy were grade 1 or 2, severe lymphopenias of grade 3 or 4 still existed because of off-tumor FcγR-dependent activation of OX40 (16). To overcome this problem, we designed our tetravalent molecule by coupling OX40 cross-linking to PD-L1 binding. In contrast to conventional bivalent OX40 antibody, our PD-L1/OX40 is speculated to induce OX40 clustering and activation at sites of PD-L1 expression, such as tumor microenvironment. Thus, our molecule may elicit more efficient and tumor-specific PD-L1–dependent activation of OX40 and, theoretically, with reduced concerns in systemic toxicity. Additional studies in cynomolgus monkeys to further characterize the safety profile of the bispecific antibody will be performed in future. Clinical outcome of anti-PD-1/PD-L1 therapy was positively correlated with PD-L1 expression level (28). A large group of patients who had relatively low PD-L1 expression, as evaluated by IHC of tumor samples, tended to show weak or no response to a PD1/PD-L1 checkpoint blockade (29, 30). Unlike common tumor-associated antigen-like Her2 or EGFR, the expression of PD-L1 is inducibly expressed on tumor tissue by inflammatory cytokines like INFγ (31), and PD-L1 expression density varies in different tumor foci and different individuals (32). To elucidate the PD-L1 level required for OX40 activation by our molecule, we tested different tumor models where PD-L1 expression was low (LoVo), moderate (MC38-PD-L1), or high (NCI-H292), and found that PD-L1/OX40 consistently exhibited superior efficacy over combined therapy, indicating a broad range of permissive PD-L1 levels. In conclusion, the preclinical results support clinical investigation of the PD-L1/OX40 bispecific antibody as an immunotherapy in advanced solid tumors.

Z. Kuang reports other from Innovent Biologics (employment, gets paid) outside the submitted work, as well as has a patent for WO2020/151761 issued. P. Pu reports other from Innovent Biologics (employment, get paid) during the conduct of the study. M. Wu reports other from Innovent Biologics (employment, gets paid) and Innovent Biologics (employment, company stocks) outside the submitted work. Z. Wu reports other from Innovent Biologics (employment, gets paid) outside the submitted work. L. Wang reports other from Innovent Biologics (employment, gets paid) outside the submitted work. Y. Li reports other from Innovent Biologics (employment, gets paid) and Innovent Biologics (employment, holds stocks) outside the submitted work. S. Zhang reports other from Innovent Biologics (employment, gets paid) during the conduct of the study. H. Jing reports other from Innovent Biologics (employment, gets paid) and Innovent Biologics (employment, holds company stocks) outside the submitted work. W. Wu reports other from Innovent Bio. Inc (employment, gets paid, holds company stocks) outside the submitted work. B. Chen reports other from Innovent Biologics (employment, gets paid) and Innovent Biologics (employment, holds company stocks) outside the submitted work, as well as has a patent for WO2020/151761 issued (public). J. Liu reports other from Innovent Biologics (employment, takes salary from Innovent, and holds company stocks) outside the submitted work, as well as has a patent for WO2020/151761 issued. No other disclosures were reported.

Z. Kuang: Conceptualization, data curation, formal analysis, validation, methodology, project administration. P. Pu: Conceptualization, writing-original draft, writing-review and editing. M. Wu: Data curation, formal analysis, methodology. Z. Wu: Data curation, formal analysis, methodology. L. Wang: Resources, formal analysis, methodology. Y. Li: Writing-original draft. S. Zhang: Writing-review and editing. H. Jing: Data curation, formal analysis, methodology. W. Wu: Data curation, formal analysis. B. Chen: Data curation, formal analysis, supervision, validation, writing-review and editing. J. Liu: Conceptualization, supervision, project administration, writing-review and editing.

The authors would like to thank Dr. Jia Zou for his careful review of this article.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Sengupta
R
,
Honey
K
. 
AACR cancer progress report 2018: harnessing research discoveries for patient benefit
.
Clin Cancer Res
2018
;
24
:
4351
.
2.
Jenkins
RW
,
Barbie
DA
,
Flaherty
KT
. 
Mechanisms of resistance to immune checkpoint inhibitors
.
Br J Cancer
2018
;
118
:
9
16
.
3.
Abdin
SM
,
Zaher
DM
,
Arafa
EA
,
Omar
HA
. 
Tackling cancer resistance by immunotherapy: updated clinical impact and safety of PD-1/PD-L1 inhibitors
.
Cancers
2018
;
10
:
32
.
4.
Watts
TH
. 
TNF/TNFR family members in costimulation of T cell responses
.
Annu Rev Immunol
2005
;
23
:
23
68
.
5.
Croft
M
. 
Control of immunity by the TNFR-related molecule OX40 (CD134)
.
Annu Rev Immunol
2010
;
28
:
57
78
.
6.
Compaan
DM
,
Hymowitz
SG
. 
The crystal structure of the costimulatory OX40-OX40L complex
.
Structure
2006
;
14
:
1321
30
.
7.
Vanamee
ES
,
Faustman
DL
. 
Structural principles of tumor necrosis factor superfamily signaling
.
Sci Signal
2018
;
11
:
eaao4910
.
8.
Mayes
PA
,
Hance
KW
,
Hoos
A
. 
The promise and challenges of immune agonist antibody development in cancer
.
Nat Rev Drug Discov
2018
;
17
:
509
27
.
9.
Zhang
D
,
Armstrong
AA
,
Tam
SH
,
McCarthy
SG
,
Luo
J
,
Gilliland
GL
, et al
Functional optimization of agonistic antibodies to OX40 receptor with novel Fc mutations to promote antibody multimerization
.
MAbs
2017
;
9
:
1129
42
.
10.
Sugamura
K
,
Ishii
N
,
Weinberg
AD
. 
Therapeutic targeting of the effector T-cell co-stimulatory molecule OX40
.
Nat Rev Immunol
2004
;
4
:
420
31
.
11.
Polesso
F
,
Weinberg
AD
,
Moran
AE
. 
Late-stage tumor regression after PD-L1 blockade plus a concurrent OX40 agonist
.
Cancer Immunol Res
2019
;
7
:
269
81
.
12.
Jacobi
FJ
,
Wild
K
,
Smits
M
,
Zoldan
K
,
Csernalabics
B
,
Flecken
T
, et al
OX40 stimulation and PD-L1 blockade synergistically augment HBV-specific CD4 T cells in patients with HBeAg-negative infection
.
J Hepatol
2019
;
70
:
1103
13
.
13.
Shrimali
RK
,
Ahmad
S
,
Verma
V
,
Zeng
P
,
Ananth
S
,
Gaur
P
, et al
Concurrent PD-1 blockade negates the effects of OX40 agonist antibody in combination immunotherapy through inducing T-cell apoptosis
.
Cancer Immunol Res
2017
;
5
:
755
66
.
14.
Redmond
WL
,
Linch
SN
,
Kasiewicz
MJ
. 
Combined targeting of costimulatory (OX40) and coinhibitory (CTLA-4) pathways elicits potent effector T cells capable of driving robust antitumor immunity
.
Cancer Immunol Res
2014
;
2
:
142
53
.
15.
Han
X
,
Vesely
MD
. 
Stimulating T cells against cancer with agonist immunostimulatory monoclonal antibodies
.
Int Rev Cell Mol Biol
2019
;
342
:
1
25
.
16.
Curti
BD
,
Kovacsovics-Bankowski
M
,
Morris
N
,
Walker
E
,
Chisholm
L
,
Floyd
K
, et al
OX40 is a potent immune-stimulating target in late-stage cancer patients
.
Cancer Res
2013
;
73
:
7189
98
.
17.
Kuang
Z
,
Jing
H
,
Wu
Z
,
Wang
J
,
Li
Y
,
Ni
H
, et al
Development and characterization of a novel anti-OX40 antibody for potent immune activation
.
Cancer Immunol Immunother
2020
;
69
:
939
50
.
18.
Nimmerjahn
F
,
Ravetch
JV
. 
Fcγ receptors as regulators of immune responses
.
Nat Rev Immunol
2008
;
8
:
34
47
.
19.
Vidarsson
G
,
Dekkers
G
,
Rispens
T
. 
IgG subclasses and allotypes: from structure to effector functions
.
Front Immunol
2014
;
5
:
520
.
20.
Omar
HA
,
Tolba
MF
. 
Tackling molecular targets beyond PD-1/PD-L1: novel approaches to boost patients' response to cancer immunotherapy
.
Crit Rev Oncol Hematol
2019
;
135
:
21
9
.
21.
Long
GV
,
Atkinson
V
,
Cebon
JS
,
Jameson
MB
,
Fitzharris
BM
,
McNeil
CM
, et al
Standard-dose pembrolizumab in combination with reduced-dose ipilimumab for patients with advanced melanoma (KEYNOTE-029): an open-label, phase 1b trial
.
Lancet Oncol
2017
;
18
:
1202
10
.
22.
Guo
Z
,
Wang
X
,
Cheng
D
,
Xia
Z
,
Luan
M
,
Zhang
S
. 
PD-1 blockade and OX40 triggering synergistically protects against tumor growth in a murine model of ovarian cancer
.
PLoS One
2014
;
9
:
e89350
.
23.
Kvarnhammar
AM
,
Veitonmaki
N
,
Hagerbrand
K
,
Dahlman
A
,
Smith
KE
,
Fritzell
S
, et al
The CTLA-4 x OX40 bispecific antibody ATOR-1015 induces anti-tumor effects through tumor-directed immune activation
.
J Immunother Cancer
2019
;
7
:
103
.
24.
Stuber
E
,
Strober
W
. 
The T cell-B cell interaction via OX40-OX40L is necessary for the T cell-dependent humoral immune response
.
J Exp Med
1996
;
183
:
979
89
.
25.
Seshasayee
D
,
Lee
WP
,
Zhou
M
,
Shu
J
,
Suto
E
,
Zhang
J
, et al
In vivo blockade of OX40 ligand inhibits thymic stromal lymphopoietin driven atopic inflammation
.
J Clin Invest
2007
;
117
:
3868
78
.
26.
Wajant
H
. 
Principles of antibody-mediated TNF receptor activation
.
Cell Death Differ
2015
;
22
:
1727
41
.
27.
Gieffers
C
,
Kluge
M
,
Merz
C
,
Sykora
J
,
Thiemann
M
,
Schaal
R
, et al
APG350 induces superior clustering of TRAIL receptors and shows therapeutic antitumor efficacy independent of cross-linking via Fcγ receptors
.
Mol Cancer Ther
2013
;
12
:
2735
47
.
28.
Herbst
RS
,
Soria
JC
,
Kowanetz
M
,
Fine
GD
,
Hamid
O
,
Gordon
MS
, et al
Predictive correlates of response to the anti-PD-L1 antibody MPDL3280A in cancer patients
.
Nature
2014
;
515
:
563
7
.
29.
Aguiar
PN
 Jr
,
De Mello
RA
,
Hall
P
,
Tadokoro
H
,
Lima Lopes
G
. 
PD-L1 expression as a predictive biomarker in advanced non-small-cell lung cancer: updated survival data
.
Immunotherapy
2017
;
9
:
499
506
.
30.
Incorvaia
L
,
Fanale
D
,
Badalamenti
G
,
Barraco
N
,
Bono
M
,
Corsini
LR
, et al
Programmed death ligand 1 (PD-L1) as a predictive biomarker for pembrolizumab therapy in patients with advanced non-small-cell lung cancer (NSCLC)
.
Adv Ther
2019
;
36
:
2600
17
.
31.
Shi
Y
. 
Regulatory mechanisms of PD-L1 expression in cancer cells
.
Cancer Immunol Immunother
2018
;
67
:
1481
9
.
32.
Ilie
M
,
Hofman
V
,
Dietel
M
,
Soria
JC
,
Hofman
P
. 
Assessment of the PD-L1 status by immunohistochemistry: challenges and perspectives for therapeutic strategies in lung cancer patients
.
Virchows Arch
2016
;
468
:
511
25
.