Background
Colorectal cancer (CRC) is one of the most common malignancies, and the patient survival rate remains unacceptably low [
1]. Recently, monoclonal antibody (mAb)-based immune checkpoint inhibitors, particularly anti-programmed cell death-1 (PD-1)/programmed cell death ligand 1 (PD-L1) mAbs, have been added to CRC treatment regimens [
2]. However, only a fraction of patients benefits from the therapy [
3,
4].
4-1BB, also called CD137 or tumor necrosis factor receptor superfamily member 9 (TNFRSF9), is a costimulatory molecule expressed functionally on the surface of various types of leukocytes, such as T cells, natural killer (NK) cells and subsets of dendritic cells, and can be activated by its ligand 4-1BBL or activating anti-4-1BB antibodies to enhance tumor rejection; thus, it is regarded as a potential target for cancer immunotherapy [
5‐
7].
Several anti-4-1BB agonistic antibodies have advanced to clinical stages but have never been clinically successful because of the intolerable toxicity caused by systemic immune activation [
8]. Urelumab (BMS-663513), an IgG4 mAb, caused severe hepatotoxicity in more than 5% of patients enrolled in phase I and II clinical trials [
9]. In contrast, utomilumab (PF-05082566), an IgG2 mAb, showed fewer grade III–IV adverse effects and no dose-limiting toxicity up to the highest dose of 10 mg/kg, but it produced a much milder agonistic function than urelumab [
10,
11]. Therefore, new antibody drugs that effectively and safely target 4-1BB are urgently needed.
Here, we demonstrate that HuB6, a novel human recombinant anti-4-1BB mAb with high specificity, has a binding epitope distinct from those of other known anti-4-1BB mAbs and shows potent antitumor activity and immune memory induction in humanized mouse models bearing CRC tumors and no systemic toxicity in either humanized mice or cynomolgus monkeys.
Methods
Cell culture
CHO-K1 and HEK-293 cells were obtained from American Type Culture Collection (ATCC CCL-61 and CRL-1573). HEK-293/NFκB-Luci/4-1BB cells were genetically engineered and expressed human 4-1BB and a luciferase reporter driven by a response element sensitive to 4-1BB agonistic stimulation and cultured in DMEM supplemented with 1 μg/mL puromycin (Gibco, C11995500BT) and 800 μg/mL hygromycin B (Sangon Biotech, A600230-0001). CHO-K1/CD32A, CHO-K1/CD32B, CHO-K1/CD16 and CHO-K1/hu4-1BB cells were designed to express the human Fcγ receptors (FcγRIIA, FcγRIIB, FcγIRA) and 4-1BB on the cell membrane, respectively, and grown in DMEM/F12 (HyClone, SH30023.01) containing 1 mg/mL Geneticin (Gibco, 11811023). The murine and human CRC cell lines CT26 and Colo205 were obtained from the cell bank affiliated with the Shanghai Institute of Biochemistry and Cell Biology (SIBCB), and the murine CRC cell line MC38 was purchased from Cobioer Company (Nanjing, China), authenticated, tested for mycoplasma contamination and cultured in RPMI-1640 medium (HyClone, SH30809.01). All media were supplemented with 10% fetal bovine serum (Ausbian, VS500T) and a 1% penicillin–streptomycin solution (HyClone, SV30010), and cells were cultured at 37 °C in a humidified incubator with 5% CO2.
Protein expression and purification
The monomeric antigen (mono-Hu4-1BB) was produced by introducing a mutation (C121S) with His-tag at the C terminus. The sequences of urelumab and utomilumab were individually obtained from the patents US8137667B2 and US2012/0237498A1 and that of anti-CD3 antibody (clone: OKT3, IMGT/mAb-DB, ID: 92) was obtained from IMGT/mAb-DB (
http://www.imgt.org/mAb-DB). The sequences for 4-1BB of mouse and cynomolgus monkey and human 4-1BBL were obtained from UniProt (mouse 4-1BB: P20334, cynomolgus monkey 4-1BB: A9YYE7, human 4-1BBL: P41273). The antigens and 4-1BBL were generated by cloning DNA-encoding sequences with a mouse Fc tag sequence at the C terminus independently into the multiple cloning sites of the mammalian expression vector pcDNA 3.4 TOPO (Invitrogen, A14697). Transfection was conducted with Expi293F cells (Gibco, A14635) and the cell culture was collected in 96 h. The Antigens with mouse Fc tag and antibodies were purified through 1 mL MabSelect PrismA column (GE Healthcare) and the mono-Hu4-1BB was directly purified on a HisTrap excel nickel column (GE Healthcare) according to a previously published protocol [
10]. In addition, human IgG, used as an isotype control, was purchased from GenScript Biotech Corporation (Nanjing, China).
Enzyme linked immunosorbent assay (ELISA)
The indirect ELISA method was used, and plates were coated with 4-1BB-ECD (extracellular domain)-mFc (6 nM 4-1BB-ECD-mFc, 1 μg/mL mouse 4-1BB-ECD-mFc or 350 ng/mL cynomolgus 4-1BB-mFc) in carbonate buffer at 4 °C overnight. After blocking with 1% BSA at 37 °C for 2 h, serially diluted test antibodies were added, incubated at room temperature for 2 h and then subjected to detection with HRP-conjugated secondary antibodies (goat anti-human Fc, Jackson ImmunoResearch Laboratories, 146460). After washing with a PBST solution three times, TMB (Invitrogen, 002023) was added as a substrate, and the absorbance was detected at 405 nm. Meanwhile, this ELISA protocol was also used to analyze the binding of HuB6 to other human TNF receptor superfamily (TNFRSF) proteins (CD40: CD4-HM140; OX40: P43489, CD27: CD2-HM127) purchased from Kactus Biosystems (Shanghai, China).
Antibody affinity determination
The affinity of HuB6, utomilumab or urelumab for human 4-1BB was determined at 25 °C by surface plasmon resonance (SPR) using a Biacore T200 carried out in single-cycle mode with a protein A biosensor chip (GE Healthcare) and biolayer interferometry (BLI) using Octet Red96 system (Pall ForteBio Analytics) according to the manufacturers’ manuals. Briefly, for SPR measurement, human 4-1BB ECD was coupled on a Series S Sensor Chip CM5 (29104988, GE Healthcare) to 300 RU using an Amine Coupling Kit (BR100050, GE Healthcare). The tested antibody and control (80 μg/mL) with two-fold serial dilutions were injected across the immobilized human 4-1BB surface at a flow rate of 30 μL/min (association and dissociation time, 300 s, stabilization time, 120 s). The chip surface was regenerated by an injection of 50 mM NaOH at a flow rate of 30 μL/min for 60 s at the end of each cycle. Binding of antibodies to human 4-1BB was analyzed using a 1:1 Langmuir model. The kinetic rate constants, association rate constant (Kon), dissociation rate constant (Koff) and equilibrium dissociation constant (KD), were calculated using the evaluation software (version 3.1, GE Healthcare). For the BLI method, the antibody was prepared at 5, 10 and 20 μg/mL in 1 × PBS running buffer and dispensed into a 96-well tilted-bottom microplate and a second 96-well microplate contained human 4-1BB (Hu4-1BB-His, Acro biosystems) at the seven titrated concentrations (200–12.5 nM with two-fold dilutions). Antibodies were loaded onto AHC biosensor for a 200-s loading step. After a 60-s baseline dip in 1 × PBS buffer, the binding kinetics were measured by dipping the antibody-coated sensors into the wells containing human 4-1BB. The binding interactions were monitored over a 500-s association period followed by a 30-min dissociation period in new wells containing fresh 1 × PBS buffer. Dissociation constants were calculated from raw data with analysis software (version 6.3, ForteBio).
Competitive protein binding assay
CHO-K1-Hu4-1BB cells were cultured to 80% confluence, digested with trypsin, centrifuged at 1000 rpm for 5 min and collected in EP tubes. HuB6, utomilumab and urelumab were labeled with biotin to create the corresponding bio-antibodies and diluted to 1.5 μg/mL with PBS. The working concentration of human 4-1BBL ranged from 90 to 0.35 μg/mL with a fourfold gradient dilution. Each Bio-antibody was mixed with 4-1BBL and then incubated with CHO-K1-Hu4-1BB cells for 30 min. After washing twice with a PBS buffer solution, the cells were incubated with a streptavidin-FITC secondary antibody (BioLegend, 405202) and incubated for 30 min in the dark. Finally, the prepared cells were suspended in 500 μL PBS and detected by flow cytometry (Beckman Coulter, CytoFLEX). In addition, the flow cytometry analysis protocol was used to evaluate the selectivity of HuB6 between 4-1BB and the other TNFRSF proteins including OX40, CD40 and CD27, which were transiently overexpressed on the surface of HEK293F cells. The goat anti-human Fc-PE (14–4998-82, Invitrogen) was used as the secondary antibody.
Luciferase assay
The activity of the mAb HuB6 was detected using the luciferase reporter gene method in vitro. The same number (3 × 104)of HEK-293/NFκB-Luci/4-1BB cells and CHO-K1 cells expressing different FcγRs per well were seeded in 96-well plates and incubated with serially diluted antibodies overnight in a CO2 incubator at 37 °C.Then, both firefly luciferase activity and Renilla luciferase activity were measured with a Glomax multidetection system luminometer using a Dual-Luciferase Reporter Assay System (Promega). Firefly luciferase activity was normalized against Renilla luciferase activity to measure antibody activity.
Mutant antigen binding detection
Extracellular amino acid sites (M101, I32, and N42) in the human 4-1BB antigen were independently mutated to synthesize different target genes, which were inserted into the pcDNA3.4 vector to obtain DNA plasmids carrying the different 4-1BB antigens. Expi293F cells expressing mutant or wild-type (WT) 4-1BB antigens on the surface were obtained by transient transfection and incubated with utomilumab, urelumab or HuB6 at a concentration of 10 μg/mL and subsequent threefold gradient dilution for 30 min. After washing twice with PBS buffer, the cells were incubated with a FITC-conjugated goat anti-human IgG (H + L) secondary antibody (Invitrogen, H10301) for 30 min, washed and resuspended in PBS, and then subjected to an antibody binding assay evaluated by flow cytometry (Beckman Coulter, CytoFLEX).
Lymphocyte isolation and T cell-activation assay
Blood leukopaks were obtained from healthy people at the Shanghai Zhaxin Hospital of Integrated Traditional Chinese & Western Medicine under institutional review board-approved protocols. Peripheral blood mononuclear cells (PBMCs) were isolated according to the manufacturer’s instructions (Ficoll 400, F8636, Sigma–Aldrich). Human CD4 + and CD8 + T cells were purified using BD IMag anti-human CD4 (No. 557767) and anti-human CD8 beads (No. 557766), CD3-CD56 + NK cells were prepared by separation with magnetic beads (NK purification kit, Miltenyi Biotec), and the purities were confirmed with flow cytometry (Beckman Coulter, CytoFLEX). For a cell proliferation assay, CD8 + T cells were labeled with 10 μM carboxyfluorescein succinimidyl ester (CFSE, Invitrogen) according to the manufacturer’s protocol and assayed by flow cytometry. To determine the IFNγ secretion activity of the lymphocytes, 96-well cell culture plates (Corning) were pretreated with an anti-CD3 antibody (clone: OKT3, 0.4 μg/mL) for CD4 + T and CD8 + T cells and 100 U/mL recombinant human IL-2 (rhIL-2, PeproTech) for NK cells. After washing twice with PBS, 1 × 105 CD4 + T, CD8 + T or NK cells in 200 μL of complete RPMI-1640 medium were added and treated with HuB6, urelumab or utomilumab across a range of doses (0.04 μg/mL, 0.4 μg/mL and 4 μg/mL). After 3 days of culture in a CO2 incubator at 37 °C, the IFN-γ content in the cell culture supernatant was determined with an ELISA kit according to the manufacturer’s manual (BioLegend).
CD8 + T cell assay with FcγR crosslinking
To test 4-1BB agonist activity dependent on FcγR, CD8 + T cells were adjusted to 2 × 104 cells/well and cocultured with CHO-K1 cells expressing different FcγRs at 1.0 × 104 cells/well in a 96-well microplate bounded with 0.4 µg/mL anti-CD3 antibody (OKT3). Then, serially diluted antibodies were incubated with cocultured cells for 3 days in a CO2 incubator at 37 °C. After incubation, the secreted IFN-γ and IL-2 levels in the cell supernatants were measured by ELISA.
Cytokine release analysis
For the cytokine release assay, PBMCs from 5 healthy donors (2 × 105 cells/well) in RPMI-1640 medium with 10% FBS in 96-well flat-bottom plates were treated with 10 µg/mL of the tested antibodies. HuB6 was compared with an isotype control human IgG4 as well as positive control antibody (OKT3). After 48 h of incubation, the levels of the cytokines IFN-γ, TNF-α, IL-10, IL-2, IL-6, IL-4 and IL-17A in the culture medium were measured by cytometric bead array assay (C60021, QuantoBio) according to the manufacturer’s instructions. Fluorescence signals were measured by a CytoFlex system (Beckman).
Model mouse
To establish two types of CRC tumor-grafted mouse models, 8-week-old human 4-1BB/4-1BBL double knock-in C57BL/6 and B-NDG B2m KO plus mice, in which human PBMCs were transplanted to reconstitute human immune cells, were purchased from Biocytogen Corporation (Beijing, China) and used according to the appropriate experimental protocol. Briefly, 2 × 106 MC38 or CT26 cells mixed with Corning Matrigel in a 1:1 volume ratio were inoculated subcutaneously into human 4-1BB/4-1BBL double knock-in C57BL/6 mice. Similarly, 2 × 106 Colo205 cells mixed with 1:1 Matrigel were inoculated subcutaneously into B-NDG B2m KO plus mice. After palpable tumors were established, the mice were randomized on the basis of tumor volume and body weight. Subsequently, treatment with a mAb or an isotype control was performed twice a week for up to 3 weeks by intraperitoneal injection. Tumor growth was monitored twice a week by measuring tumor length and width. Tumor volume was calculated according to the following equation: 0.5 × length × width × width.
Toxicology study
HuB6 toxicity studies were conducted with humanized model mice and purpose-bred cynomolgus monkeys. Humanized 4-1BB mice were intraperitoneally injected with a low dose (3 mg/kg) or a high dose (30 mg/kg) of HuB6, utomilumab, urelumab or an isotype antibody once every 3 days for 6 total injections. Cynomolgus monkeys were administered repeated intravenous doses of 3, 10, and 30 mg/kg/week for 5 weeks or single doses of 60 and 180 mg/kg for toxicity studies. Two male and 2 female cynomolgus monkeys were randomly assigned to each group, and the antibodies were administered via intravenous infusion at a dose of 5 mL/kg administered at a rate of 1 mL/min. Mouse necropsies were performed according to a standard protocol, and the major organs were collected for histological evaluation. All the tissues were fixed in 10% neutral-buffered formalin, routinely processed, embedded in paraffin, sectioned, stained with hematoxylin and eosin (HE) and analyzed by a professional pathologist. In addition, blood and serum were collected for clinical hematological and chemical analyses using a Sysmex BX4000.
Statistical analysis
Statistical analyses were performed using GraphPad Prism 8.0 (GraphPad Software), and one-way or two-way ANOVA was used to compare intergroup differences. A p value of < 0.05 was considered significant.
Discussion
In this study, we demonstrated that the humanized anti-4-1BB mAb Hub6 induced T cell proliferation and activity and had potent efficacy in tumor inhibition and immune memory induction without systemic toxicity; thus, it was regarded as a potential candidate for cancer immunotherapy. As 4-1BB is one of the costimulatory receptors of immune cells, 4-1BB-targeting mAbs have shown promising antitumor effects in preclinical models [
20]. However, the clinical development of two leading molecules, utomilumab and urelumab, is facing serious challenges due to low efficacy or severe systemic toxicity [
8,
11].
Recently, several bispecific tumor antigen-targeted 4-1BB agonists have been developed [
21‐
24]. However, their therapeutic efficacy relies fully on the expression of tumor antigens, limiting their clinical application to patients with antigen overexpression. In addition, tumor antigen, such as EGFR, is widely expressed in normal, non-neoplastic tissues, and its use as a target antigen can lead to severe on-target, off-tumor immunotoxicity [
21,
24]. It follows that the selected tumor antigens should be highly tumor-specific and their high expression is often limited to only specific types of cancer [
8]. In contrast, the success of immune checkpoint inhibitors, such as anti–PD-1/PD-L1 antibodies, is partially attributed to their broad applicability in a variety of cancers regardless of the antigen expression status. Therefore, a novel humanized anti-4-1BB agonistic antibody that has strong agonistic activity, a high safety profile and broad applicability without depending on tumor antigen expression is urgently needed.
Our previous studies showed that HER2-targeted antibodies with different binding epitopes exhibited different antitumor properties [
25,
26]. Here, we screened twelve humanized 4-1BB-targeted IgG4 subtype scFvs and then generated a novel anti-4-1BB mAb, HuB6, with an antigen epitope distinct from those of other known antibodies, such as utomilumab and urelumab; thus, HuB6 has unique antitumor efficacy and a high safety profile. As shown in Fig.
1d, the binding epitope of HuB6 is between CRD1 and CRD2 of the 4-1BB protein, while urelumab binds to CRD1, and utomilumab binds between CRD3 and CRD4. The binding site of 4-1BBL overlaps with those of HuB6 and utomilumab but not with that of urelumab; thus, both HuB6 and utomilumab are ligand-blocking antibodies, while urelumab is a non-ligand-blocking antibody, which was confirmed by the 4-1BBL competitive binding result and thus could allow HuB6 to activate T cells mimicking 4-1BBL. As a 4-1BB agonist, HuB6 increased the proliferation of CD8 + T cells and the production of the antitumor cytokine IFN-γ, inhibited tumor growth in all the mouse models tested, induced potent antitumor immune memory and exerted an enhanced tumor-inhibiting effect in combination with an anti-PDL1 mAb, similar to utomilumab, urelumab and several other potential therapeutics [
8,
27‐
30]
.
However, obvious differences in agonistic activity and toxicity are also easily found among these antibodies. First, the affinity of HuB6 for human 4-1BB was similar to that of utomilumab, but the tumor inhibitory efficacy in several mouse models seemed to be greater than that of utomilumab although the difference was not significant. Second, Hub6 was constructed as a recombinant human IgG4 mAb, which had more agonistic activity than its IgG2 counterpart (Fig.
3d). Moreover, Hub6 showed high safety in vitro and in vivo (Figs.
6,
7). However, as the same IgG4 mAb, urelumab caused severe liver toxicity, consistent with a previous study [
9]. The above differences could be interpreted with epitope accessibility theory [
10]. Namely, the binding site on the very N-terminus of CRD1 targeted by urelumab orients the antibody Fc domain to be optimally exposed for interaction with FcγR, which potentially enhances ADCC and CDC. The utomilumab epitope, which is closer to the cell surface, orients the antibody parallel to the membrane, where engagement of FcγR may be more restricted, and the HuB6 binding epitope makes the interaction with FcγR mild, potentially between the interactions of urelumab and utomilumab. Furthermore, IgG4 is known to engage FcγRIA and FcγRIIB more than IgG2, although both isotypes are generally characterized by relatively low FcγR interactions [
16]. Our results also showed that FcγRs, especially FcγRIA, should be critical for the activity of HuB6, which also ensures its high safety. In contrast, urelumab is currently the strongest agonist and can even induce 4-1BB stimulation in the absence of FcγRs, and the IgG4 isotype likely boosts its in vivo activity with FcγR interaction, which may explain why it has severe hepatotoxicity. As an IgG2 mAb, Utomilumab has weak binding to FcγRIIA and FcγRIIB and does not even activate T cells in the presence of only FcγRIA, which may result in its weak activity. These data suggested that balancing agonistic activity with the affinities of FcγRs should be a strategy to screen 4-1BB agonistic mAb, and provided new evidence in understanding how therapeutic mAb works and new insight for the design of mAb candidate therapeutics.
Several important limitations should be considered in our study. First, the antitumor efficacy of HuB6 is mainly studied on CRC, which is also one of routinely selected types for cancer immunotherapy [
31,
32], but more cancer types should be tested to explore the indication of HuB6 therapy. In addition, the capacity of HuB6, such as lymphocyte recruitment and activation in tumor microenvironment, needs to be determined in the future. Currently, HuB6 has been approved for evaluation in clinical trials based on the preclinical evidence, and multiple types of solid cancers including CRC will be investigated.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.