AN3025 was generated through rabbit immunization with extracellular domain (ECD) of human TNFR2 and subsequent humanization by complementarity-determining regions (CDRs) grafting. Briefly, New Zealand white rabbits (specific pathogen-free) were immunized with extracellular domain of human TNFR2 (Acro Biosystem, TN2-H5227). Single-chain variable fragment phage library was constructed with rabbit VH and VL regions amplified from rabbit spleen cDNA. Monoclonal phage enzyme-linked immunosorbent assay (ELISA) was performed to identify binders against human TNFR2. Chimeric anti-human TNFR2 antibody was generated through fusion of selected rabbit VH to a human IgG₁ constant region and fusion of selected rabbit VL to a human IgG₁ kappa constant domain. The chimeric antibody was further humanized by inserting the rabbit CDRs into human germline frameworks.

Six- to ten-week-old humanized female B-hTNFR2 mice (Biocytogen) were housed in the specific-pathogen-free (SPF) barrier facility of the Animal Center of Beijing Biocytogen Co., Ltd in individual ventilated cage. The experimental animals were acclimatized for 7 days before being used in experiments. All experimental animal procedures were in accordance to the Institutional Animal Care and Use Committees (IACUC) guidance. Mice were euthanized with CO₂. All efforts were made to minimize animals’ distress and pain. Murine colon cancer MC38 cells were subcutaneously implanted into homozygous B-hTNFR2 mice. Mice were grouped when tumor volume reached approximately 100 mm³. Then mice were treated with PBS, anti-mouse PD-1 antibody (BioXcell, BP0146), AN3025 or combination of anti-mouse PD-1 antibody and AN3025 through i. p. injection. Anti-mouse CD4 (BioXcell, BP0003-1) and anti-mouse CD8 (BioXcell, BP0004-1) were used at 0.3mg per mouse per i.p. injection for depletion of CD4⁺T cells and CD8⁺ T cells. Animal well-being and behaviors were monitored during the experiment process. Animal body weight and tumor volume were measured twice a week. The anti-tumor efficacy is expressed as tumor growth inhibition in terms of tumor volume (TGI_TV). The TGI_TV in percent was calculated as below:

Where T_i= mean tumor volume of drug-treated group on the final day of the study, T₀= mean tumor volume of drug-treated group on first dosing day, C_i= mean tumor volume of control group on the final day of the study, C₀= mean tumor volume of control group on the first dosing day.

Human TNFR2 (Acro Biosystem, TN2-H5227), human TNFR1 (Acro Biosystem, TN1-H5222), cynomolgus TNFR2 (Sino Biological,90102-C08H), mouse TNFR2 (R&D systems, 426-R2-050/CF) or rat TNFR2 (R&D systems, 8348-R2-050) was coated on high binding polystyrene flat bottom micro-titer plates (Thermo Scientific, 3455). Plates were incubated at 4°C overnight. Plates were washed with PBST (wash buffer 0.1%) and blocked with 1% BSA (Fisher Scientific, BP1600-100) in PBS for 1 hour at 37°C. AN3025 or human IgG₁, κ (BioxCell, BE0297) was titrated, distributed in the wells and incubated at 37°C for 1 hour. Plates were washed with PBST (wash buffer 0.1%) and incubated with goat anti-human IgG Fc secondary antibody-horseradish peroxidase (HRP) conjugate (Invitrogen, A18817) for 1 hour at 37°C. Plates were washed and developed using TMB substrate solution (eBioscience,00-4201-56) and stopped with ELISA stop solution (Invitrogen, SS04). The level of bound antibodies was determined by reading absorbance at 450 nm. Mouse anti-human TNFR1 antibody (ThermoFisher Scientific, MA1-81005) and goat anti mouse IgG secondary antibody-HRP conjugate (Invitrogen, 62-5520) were used as positive control for human TNFR1 binding. Armenian hamster anti-mouse TNFR2 antibody (InVivoMab, BE0247) and HRP-goat anti-armenian hamster IgG (H+L) secondary antibody (Invitrogen, PA1-32045) were used as positive control for mouse TNFR2 binding. TNFR2 polyclonal antibody (Invitrogen, PA5-80159) and HRP conjugated Goat anti-Rabbit IgG Fc Secondary Antibody (Invitrogen, A16116) were used as positive control for rat TNFR2 binding.

Human TNFα (Gibco, PHC3011) was biotinylated using an EZ-Link™ Micro Sulfo-NHS-Biotinylation Kit (ThermoFisher Scientific, 21925) according to manufacturer’s instructions. Human TNFR2 (Acro Biosystem, TN2-H5227) was coated on high binding polystyrene flat bottom micro-titer plates (Thermo Scientific, 3455). Plates were incubated at 4°C overnight. Plates were washed with PBST (wash buffer 0.1%) and blocked with 1% BSA (Fisher Scientific, BP1600-100) in PBS for 1 hour at 37°C. AN3025 or human IgG₁, κ (BioxCell, BE0297) was titrated, distributed 50μL per well and incubated for 1 hour at 37°C. 50μL biotinylated human TNFα was added per well to achieve the final TNFα concentration at 100ng/ml. After another 1-hour incubation, plates were washed with PBST (wash buffer 0.1%). Streptavidin-HRP conjugate (Thermo Scientific, N504) was distributed to detect bound biotinylated human TNFα. After 1-hour incubation of secondary antibody at 37°C, plates were washed with PBST (wash buffer 0.1%). Plates were developed using TMB substrate solution (eBioscience,00-4201-56) and stopped with ELISA stop solution (Invitrogen, SS04). The level of bound biotinylated human TNFα was determined by reading absorbance at 450 nm.

Full length human TNFR2 was cloned into pLVX-EF1a-IRES-Puro vector (Takarabio, 631253). Lentivirus production and target cell transduction were performed according to manufacturer’s instructions. Briefly, human TNFR2-lentivector was mixed with ViraPower™ Packaging Mix (Invitrogen, K4975-00) and transfected 293T cell line (Sigma-Aldrich, 12022001) using Lipofectamine 2000 Reagent (Invitrogen, 11668-019) to produce a lentiviral stock. Jurkat E6-1 cells (ATCC, TIB-152) were transduced with human TNFR2-lentivirus in the presence of 8µg/mL polybrene (EMD, Millipore, TR-1003-G) and selected with 1µg/mL puromycin (Gibco, A11138-03). The pool of surviving cells was expanded. Single clones were generated from the hTNFR2 expressing pooled cells by single cell plate sorting on cell sorter (Sony, SH800). The parental Jurkat cells and single cell clones were blocked with human TruStain FcX™ (Biolegend, 422302) for 10 minutes at room temperature according to manufacturer’s instructions. Then cells were stained with phycoerythrin (PE) conjugate anti-human TNFR2 antibody (ThermoFisher Scientific, TNFR7504) for 30 minutes at 4°C. Cells were analyzed by flowcytometry (Sony, SA3800). Data was analyzed with Flowjo software.

Parental Jurkat E6-1 cells or TNFR2 overexpressing Jurkat single clone cells were plated in opaque 96 well plates with 10⁴ cells in 100μL complete RPMI medium per well. Gradient concentration of human TNFα (Gibco, PHC3011) was added. The cell viability after 24 hours culture was measured with Promega CellTiter-Glo^® Luminescent Cell Viability Assay (Promega, G7570) according to manufacturer’s instructions. AN3025 or human IgG₁, κ (BioxCell, BE0297) was added to the hTNFR2 overexpressing Jurkat cells by dose titration along with constant human TNFα (0.5 ng/ml). The cell viability after 24 hours culture was measured with Promega CellTiter-Glo^® Luminescent Cell Viability Assay (Promega, G7570).

Human peripheral blood mononuclear cells (PBMC) were isolated from normal healthy donor blood (New York Blood Center) with Ficoll (GE HealthCare, 17-1440-03) according to manufacturer’s instructions. Human CD4⁺ T cells were isolated from PBMC with human CD4⁺ T cell isolation kit (StemCell Technologies, 17952). Tregs were induced with human Tregs differentiation kit (R&D Systems, CDK006) according to manufacturer’s instructions. Autologous CD4⁺ T cells were label with carboxyfluorescein succinimidyl ester (CFSE, Invitrogen, C34554). Tregs and autologous CD4⁺T cells were co-cultured with human T-activator CD3/CD28 Dynabeads (Gibco,11131D) for 4 days in the presence of AN3025 or human IgG₁, κ (BioxCell, BE0297). Proliferation of CD4⁺ T effector cells was evaluated with CFSE^lo population by flowcytometry (Sony, SA3800). Interferon gamma (IFNγ) production in the supernatant was detected by ELISA kit (R&D Systems, DY285B).

MC38 tumor tissues were collected at the end of MC38 tumor model study and minced into pieces. Single cell suspension was obtained by digesting minced tumor tissues with tumor dissociation kit (Miltenyi Biotec,130-096-730) in GentleMACS Octo dissociator. Cells were blocked with purified anti-mouse CD16/32 (Biolegend,101302) and stained with fluorescent antibodies against CD45 (Biolegend, 103116), CD3 (Biolegend, 100328), CD4 (Biolegend, 116004), Foxp3 (eBioscience, 25-5773-82) and fixable viability dye eFluor ™506 (eBioscience, 65-0866-14). Intracellular staining was performed with Foxp3/transcription factor staining buffer set (eBioscience, 00-5523-00) according to manufacturer’s instructions. The tumor infiltrating lymphocytes were analyzed with flowcytometry.

To perform the TILs analysis by immunohistochemistry (IHC), MC38 tumors were collected from B-hTNFR2 mice and fixed with formalin and embedded in paraffin. Tissues sections (3-4μm thick) were taken on 3-aminopropyltriethoxysilane (APTS) coated glass slides then de-paraffinized in xylene followed by hydration in graded ethanol. Specimens were heated at 100°C for 20 min in 0.01M citrate buffer (pH 6.0) using an EZ antigen retriever system (Biogenex, USA) to retrieve antigen. Endogenous peroxidase was blocked by treatment with 0.3% hydrogen peroxide. Non--specific binding sites were blocked with a protein block. Sections were covered with primary antibodies against mouse CD4 (Cell signaling, 25229S, Rabbit IgG1) or CD8 (Cell signaling, 98941, Rabbit IgG1) and incubated in moist chamber overnight at 4°C. After labeling with primary antibody, slides were washed with TBS (pH 7.4), blocked with post primary block at room temperature. Sections were then washed twice in TBS followed by incubation with Novolink polymer at room temperature. After washing with TBS for three times, sections were treated with DAB chromogen (3, 3’-diaminobenzidine tetrahydrochloride) in the dark. Sections were counterstained with hematoxylin, dehydrated with ethanol and xylene, and mounted permanently with Di-n-butylPhthalate in Xylene (DPX). Negative control slides omitting the primary antibody were included in all batches. Scoring of immune stained positive TILs was done independently by at least two pathologists. CD4⁺ and CD8⁺ TILs were counted in five randomly selected fields at 40X magnification and the counts were averaged.

All data are shown as Mean ± SEM. Experimental data were analyzed with the student’s t-test to calculate the P values. Statistical analysis of the tumor volumes on the final day of experiments was performed via t-test. Statistical analysis of tumor growth was also performed via two-way ANOVA and the corresponding P values were listed in Supplementary Table 1 . Values of P < 0.05 were considered statistically significant. *P<0.05; ** P<0.01; ***P<0.001.

AN3025 was generated through rabbit immunization with extracellular domain of human TNFR2 and subsequent humanization by CDRs grafting. ELISA assay was performed to confirm and characterize its binding affinity and selectivity. AN3025 binds to human TNFR2 on plate with sub nanomolar affinity (EC₅₀ = 0.052nM, Figure 1A ) but not to human TNFR1 ( Figure 1B ). Species cross reactivity was also evaluated. AN3025 exhibited sub nanomolar affinity to cynomolgus TNFR2 ( Figure 1C , EC₅₀ = 0.053nM), but didn’t recognize mouse ( Figure 1D ) or rat TNFR2 ( Figure 1E ).

Human TNFα-TNFR2 signaling axis has been reported to induce cell death of TNFR2 overexpressing Jurkat cells but didn’t affect the viability of wildtype Jurkat cells lacking TNFR2 expression (30, 31). We firstly detected the expression of human TNFR2 on wildtype Jurkat cells and stimulated Jurkat cells with titrated human TNFα up to 1000ng/ml. Consistent with previous reports (30, 31), Jurkat cells didn’t express human TNFR2 ( Figure 2A ). Meanwhile, treatment with human TNFα didn’t show much effect on the Jurkat cell viability ( Figure 2B ). A hTNFR2 overexpressing Jurkat single cell clone was established by transducing Jurkat cells with hTNFR2-lentivirus. The overexpression of hTNFR2 on the single cell clone was confirmed with flowcytometry ( Figure 2C ). TNFR2 overexpressing Jurkat cells are highly sensitive to TNFα-induced cell death with 99% cell killing achieved at TNFα concentration as low as 2 ng/ml ( Figure 2D ).

Taken together, these results demonstrated the activation of TNFα-TNFR2 signaling pathway induced cell death of TNFR2 overexpressing Jurkat cells. This assay could be used to evaluate the antagonistic activity of AN3025 against the TNFα-TNFR2 signaling pathway. TNFR2 overexpressing Jurkat cells were treated simultaneously with constant concentration of 0.5ng/ml human TNFα and titrated human IgG₁, κ control or AN3025. AN3025 rescued TNFR2 overexpressing Jurkat cells from TNFα-induced cell death through a dose-dependent way ( Figure 2E ), indicating AN3025 was capable to block the TNFα-TNFR2 signaling pathways.

To further investigate the mechanism by which AN3025 blocked the TNFα-TNFR2 signaling pathway, we next examined if AN3025 compete with human TNFα for receptor binding of hTNFR2. Human TNFR2 was coated on plate and incubated with human IgG₁, κ control or AN3025. Biotinylated human TNFα was added. Bound human TNFα was detected with HRP-conjugated streptavidin. Pre-incubation with AN3025 decreased the TNFα binding to hTNFR2 compared with the IgG₁, κ control ( Figure 2F ). These data suggest AN3025 blocked the TNFα-hTNFR2 signaling by competing with human TNFα for binding of TNFR2.

During the chronic development of cancer, induced Tregs (iTregs) arise from naive CD4⁺ cells in the periphery, infiltrate into the tumor microenvironment and suppress the effector T cells in aspects of cell proliferation and cytokines production (12). In addition, TNFR2 was reportedly to be highly expressed on Tregs (19), therefore we sought to evaluate whether AN3025 could affect the function of iTregs. To this end, iTregs were produced and subsequently co-cultured with CFSE stained autologous CD4⁺ T effector cells in the presence of AN3025 or human IgG₁, κ control. AN3025 addition increased CD4⁺ T effector proliferation ( Figure 2G , P=0.0544) and significantly elevated IFNγ production in the supernatant ( Figure 2H ). Thus, AN3025 is capable to inhibit iTregs mediated immune suppression, leading to enhanced immune activation of T effector cells.

ELISA assay has demonstrated AN3025 lacks cross reactivity to mouse TNFR2 ( Figure 1D ). To determine the in-vivo anti-tumor activity of AN3025, TNFR2 humanized mice (B-hTNFR2) from Biocytogen were used for murine tumor model study. Leukocyte subpopulations, including T/NK cells, Tregs, B cells, DC, granulocytes, and monocytes/macrophages, were comparable in B-hTNFR2 mice to those in wild-type mice both in spleen and in lymph node (32). In addition, human TNFR2 instead of murine TNFR2 expression was confirmed on the spleen Tregs in B-hTNFR2 mice (32). Interestingly, one previous study (33) implied that human TNFα couldn’t function through mouse TNFR2, we therefore sought to test whether mouse TNFα could instead interact with human TNFR2, given the hTNFR2 humanized mice used in the in-vivo studies. Our data showed mouse TNFα induced a dose-dependent cell death of human TNFR2 overexpressing Jurkat cells to a similar extend as for human TNFα, indicating mouse TNFα can function through human TNFR2 ( Supplementary Figure 1 ).

MC38 (5E5) murine colon cancer cells were inoculated into B-hTNFR2 mice, and mice bearing MC38 tumor were later treated with AN3025 antibodies separately at 1 mg/kg, 3mg/kg or 10mg/kg. AN3025 treatments at dosage of 10mg/kg and 3mg/kg achieved significant tumor growth inhibition (TGI) at 67.2% and 60.0% respectively ( Figure 3A ). Meanwhile, treatment at 1mg/kg didn’t show tumor growth inhibition effect ( Figure 3A ). None of the three dosages induced significant body weight change in the experimental animals compared with the control group ( Figure 3B ).

To further investigate the anti-tumor mechanism of AN3025, we analyzed the Tregs frequency in the MC38 tumors after AN3025 treatments. Both treatments at 10mg/kg and 3mg/kg significantly decreased Tregs frequency in the tumors ( Figure 3C and Supplementary Figure 2 ). This data demonstrated that AN3025 might inhibit either the proliferation or infiltration of Tregs in the tumor microenvironment, thus consequently activating the anti-tumor T cell response to exert anti-tumor activity. Consistently, RNA seq analysis ( Supplementary Figure 3 ) of MC38 tumor tissues showed AN3025 treatment elevated expression of immune activation genes such as ifn (gene for IFNγ) and gzmk (gene for Granzyme K). In order to further test the immune dependency of AN3025 in controlling tumors, antibodies were used to deplete CD4⁺ T and CD8⁺ T cells simultaneously with AN3025 treatment. The anti-tumor activity of AN3025 was abolished by CD4⁺ and CD8⁺ T cells depletion ( Figure 3D ). No body weight change was observed with the indicated treatments ( Figure 3E ). Therefore, the anti-tumor activity of AN3025 is dependent on the existence of CD4⁺ and CD8⁺ effector T cells.

Besides the murine MC38 tumors, the anti-tumor activity of AN3025 as a monotherapy was also tested in melanoma B16F10 tumors with TNFR2 humanized mice. AN3025 (10mg/kg) achieved tumor growth inhibition at 46.6% in B16F10 tumor model without body weigh change in TNFR2 humanized mice ( Supplementary Figure 4 ).

We also compared the anti-tumor effect of AN3025 with anti-PD-1 antibody. We found the anti-tumor effect of AN3025 at 10mg/kg is comparable to that of anti-mouse PD-1 (mPD-1) antibody (10mg/kg) in the MC38 tumor bearing B-hTNFR2 mice in the same study ( Figure 4A ). No body weight change was observed with the indicated treatments ( Figure 4B ). Tumor infiltrating CD8⁺ T cells and CD4⁺ T cells in MC38 tumors from B-hTNFR2 mice were detected with by immunohistochemistry (IHC). AN3025 significantly increased tumor infiltrating CD8⁺ T cells ( Figures 4C, D ) and CD4⁺ T cells ( Figures 4E, F ), while anti-mouse PD-1 antibody showed less effects on tumor infiltrating CD8⁺ and CD4⁺ T cell in the same study ( Figures 4C–F ). This observation might imply different underlying anti-tumor mechanisms for AN3025 and anti-PD-1 antibody, suggesting a combination treatment of AN3025 and anti-PD-1 antibody might achieve better anti-tumor efficacy.

To explore the potential therapeutic strategy, we then treated MC38 tumor bearing B-hTNFR2 mice with the combination of AN3025 and anti-mPD-1 antibody. AN3025 in combination with anti-mouse PD-1 antibody demonstrated stronger in-vivo anti-tumor activity than either monotherapy ( Figure 4G ). Neither monotherapies nor the combination therapy caused changes of mice body weight compared with the control group ( Figure 4H ).