The list of materials, cell lines and animals used in this study is available in the Supplementary Materials in Supplementary Data Sheet .

Binding to PD-1: To assess the binding of JS207 to PD-1 and related proteins, 0.3 μg/mL of target protein (PD-1, BTLA, PD-L1, PD-L2 and ICOS, all from human) was coated onto 96‐well plates. After blocking with 2%BSA, test samples were then added. Bound antibody was detected using an HRP-goat anti‐human IgG (Fc‐specific) antibody. Absorbance was measured at 450 nm/620 nm. To assess the species cross-reactivity, PD-1 proteins from rat, mouse, Cynomolgus monkey and human were used ( Supplementary Data Sheet ).

Binding to VEGFA: To evaluate the binding of JS207 to VEGFA and related proteins, 0.3 μg/mL of target protein (VEGFA, VEGFB, VEGFC, VEGFD, VEGFE and PLGF, all from human) was coated onto 96‐well plates. After blocking with 2% BSA, test samples were added. Detection was carried out using an HRP-goat anti‐human IgG (Fc‐specific) antibody. To assess the species cross-reactivity, VEGFA proteins from rat, mouse, human and Cynomolgus monkey were used ( Supplementary Data Sheet ).

Blocking the interaction between human PD-1 with PD-L1/PD-L2: Inhibition of the interaction between human PD-1 and its ligands (PD-L1 and PD-L2) was assessed using blocking ELISAs. Human PD-1 protein (1.5 μg/mL) was coated on the plate and blocked with 2% BSA. Test samples were prepared in assay buffer containing biotin‐PD-L1 hFc (4.0 μg/mL). After detection with HRP-streptavidin, absorbance was read at 450 nm/620 nm ( Supplementary Data Sheet ).

Blocking the interaction between human VEGFA and VEGFR2: To assess JS207’s competitive inhibition of the VEGFA–VEGFR2 interaction, 0.5 μg/mL of human VEGFA was coated on 96‐well plates and blocked with 2% BSA. Test samples (JS207, VEGF-DotAb, AK112) were added in assay buffer containing biotin‐VEGFR2 (0.3 μg/mL). After detection with HRP-streptavidin, absorbance was read at 450 nm/620 nm ( Supplementary Data Sheet ).

Binding to human PD-1: The binding affinity of JS207 to human PD-1 was measured by Biacore T200. 40 µg/mL of anti-human Fc antibody was coated to CM5 chip, and then 2 μg/mL of JS207 were captured. Serially diluted human PD-1 was then applied, and the kinetic model was analyzed, and the binding affinity (K_D value) was calculated ( Supplementary Data Sheet ).

Binding to human VEGFA: The binding affinity of JS207 to human VEGFA was determined by the Biacore T200. 40 µg/mL of anti-human Fc antibody was coated to CM5 chip, and test samples were added. The kinetic model was analyzed, and the binding affinity (K_D value) was calculated ( Supplementary Data Sheet ).

Binding of JS207/VEGFA complex to human PD-1: The binding of pre-formed JS207/VEGFA complex to human PD-1 was assessed using the Octet RED96e system. 20 µg/mL of human PD-1-mFc was immobilized on an AMC biosensor, then incubated with JS207/VEGFA complex. Kinetic parameters were analyzed using Data Analysis 11.1 software ( Supplementary Data Sheet ).

Simultaneous binding of JS207 to human PD-1 and VEGFA: The ability of JS207 to simultaneously bind to human PD-1 and human VEGFA was measured using the Biacore T200. Two different experimental formats were used: (1) PD-1 first, then VEGFA, and (2) VEGFA first, then PD-1 ( Supplementary Data Sheet ).

PD-1 reporter gene assay: Jurkat/PD-1-NAFT-Luc (Jurkat/PD-1) cells and PD-L1 aAPC/CHO-K1 (CHO/PD-L1) cells were used in this assay. JS207 inhibits the binding of PD-1 in Jurkat/PD-1 cells to PD-L1 in CHO/PD-L1 cells leading to NFAT/luciferase reporter gene activation, and the anti-PD-1 potency was determined via bioluminescent measurement ( Supplementary Data Sheet ). The signal to noise (S/N) ratio was generated by dividing the Top response by Bottom response. EC₅₀ or IC₅₀ is the concentration of an antibody required to achieve 50% of its maximum biological effect.

VEGF reporter gene assay: H293/VEGFR2 cell that was engineered to express VEGFR2/NFAT-luciferase was used in this assay. When VEGFA binds to VEGFR2 to initiate the signaling pathway, the NFAT-luciferase reporter gene is activated. The anti-VEGFA potency of JS207 was determined via bioluminescent measurement ( Supplementary Data Sheet ).

VEGFA at 10 ng/mL and the test antibody solutions at 0.004 nM-10 nM were added to 96-well plate and incubated at 37°C for 30 minutes. Then, HUVECs at 3×10³ cells/well were seeded and incubated at 37°C for 96 hours. Cell counting-Lite luciferase assay reagent was added and chemiluminescence signals were measured ( Supplementary Data Sheet ).

Internalization assay using cell surface residual PD-1 quantification method: H293/PD-1 cells were seeded at 1×10⁵ cells/well in a 96-well plate. Serially diluted test samples were added in the absence or presence of VEGFA. After incubation at 4°C for 30-minutes, the cells were divided into two aliquots and incubated at 37°C and 4°C for 0.5, 1, 2, and 4 hours, respectively. All cells were stained with an anti-human IgG-PE antibody for 30 minutes at 4°C. The samples were then analyzed by flow cytometry. The mean fluorescence intensity (MFI) was determined, and the internalization index was calculated using the following formula ( Supplementary Data Sheet ):

Internalization assay using intracellular fluorescence method: JS207 and other test antibodies were conjugated with CypHer5E follow manufacture’s instruction (GE Healthecare). The conjugated antibodies were serially diluted and incubated with or without VEGFA, then incubated with Jurkat/PD-1 cells for 4 hours at either 4°C or 37°C. Cells were washed with cold medium and subsequently stained with a PE-labeled noncompetitive anti-PD-1 antibody (MIH4 PD-1 PE, BD Biosciences). Samples were then analyzed on a BD FACSCanto II flow cytometer ( Supplementary Data Sheet ).

Cell-based PD-1 binding: To assess the effect of VEGFA on cell-based PD-1 binding, H293/PD-1 cells were seeded at 1×10⁵ cells/well in a 96-well plate. Test samples were added in the absence or presence of VEGFA. The cells were incubated at 4°C for 30 minutes, stained with mouse anti-human Fc-PE antibody for 30 minutes at 4°C and then analyzed by flow cytometry. Data analysis was performed using FlowJo ( Supplementary Data Sheet ).

The effect of JS207 on IL-2 and IFN-γ release was evaluated using a mixed lymphocyte reaction (MLR) system. Peripheral blood mononuclear cells (PBMCs) resuspended in EasySep buffer and CD4⁺ T cells were isolated. Mature dendritic cells (mDCs) and purified CD4⁺ T cells were then seeded into 96-well plates at densities of 10,000 mDCs/well and 100,000 CD4⁺ T cells/well, respectively. Test samples were added with a final concentration ranging from 150 nM to 15 pM. The cells were incubated at 37 °C for 5 days. Supernatants were collected on days 3 and 5 for IL-2 and IFN-γ measurement ( Supplementary Data Sheet ).

Test samples were first diluted to 2 mg/mL in cell culture medium and then subjected to heat stress at 40°C for 0–96 hr, 55°C for 0 – 48hr, and 65°C for 0–4 hr to assess the impact of heat-stress on anti-PD-1 activity. To assess the impact of heat-stress on anti-VEGFA activity, samples were stressed at 40°C and 50°C for 0–6 days. The potency values of stressed samples was compared with the respective control samples (Time 0).

Mouse colon cancer MC38 model in B-hPD-1 humanized mice: MC38 cells were implanted subcutaneously in the right flank of C57BL/6-Pdcd1^{tm1(PDCD1)Bcgen/Bcgen} mice (B-hPD-1 humanized mice) at 1×10⁵ cells/mouse. When the mean tumor volume reached approximately 115 mm³, mice were randomly assigned into seven groups (n = 8 per group): (1) saline, (2) toripalimab 0.6 mg/kg, (3) VEGF-DotAb 0.33 mg/kg, (4) toripalimab 0.6 mg/kg + VEGF-DotAb + 0.33 mg/kg, (5) JS207 0.75 mg/kg, (6) JS207 1.5 mg/kg, and (7) JS207 4.5 mg/kg. Treatments were administered intraperitoneally twice weekly for a total of 6 doses. Tumor volumes and animal body weights were recorded ( Supplementary Data Sheet ).

Malignant melanoma A375 model in NDG mice: A375 cells at 5×10⁶ cells/mouse were suspended mixed with Matrigel and implanted subcutaneously into NDG mice. When the average tumor volume reached approximately 137 mm³, 10×10⁶ PBMCs in 0.2 mL were injected intravenously. Two days after PBMC administration, the tumor-bearing mice were randomly assigned to five groups (n = 8 per group): (1) saline control group, (2) AK112 11.1 mg/kg, (3) JS207 1.0 mg/kg), (4) JS207 3.0 mg/kg, and (5) JS207 10.0 mg/kg. All the treatments were administered intraperitoneally twice a week for a total of 6 doses ( Supplementary Data Sheet ). Tumor volumes and animal body weights were recorded, and tumor growth inhibition rate (TGI_TV) was calculated as:

Where: Ti = mean tumor size of the treatment group on the i-th day of administration, T0 = mean tumor size of the treatment group on day 0 of administration; Vi = mean tumor size of the negative control group on the i-th day of administration, V0 = mean tumor size of the negative control group on day 0 of administration.

All animals of the above in vivo studies were housed in an SPF-grade facility of Suzhou Junmeng Biopharmaceuticals. The housing conditions were maintained at a temperature of 20–26 °C, relative humidity of 40%-70%, with 12-hour light/dark cycle. All protocols and procedures are approved by the local Institutional Animal Care and Use Committee (IACUC) under permit number YTSYDWIACUC202401.

Statistical significance was determined using Microsoft Excel and GraphPad Prism software. Comparisons between groups were performed using a two-tailed Student’s t-test. Data are presented as the mean ± standard deviation, and differences between groups were considered statistically significant when p < 0.05 (*p < 0.05; **p < 0.01, ***p < 0.001, ****p < 0.0001).

JS207 is a recombinant humanized bispecific antibody that targets both PD−1 and VEGFA and belongs to the IgG4 κ subtype. It was independently developed by Shanghai Junshi Biosciences Co., Ltd. for the treatment of advanced malignancies. JS207 comprises two identical light chains (LC) and two identical heavy chains (HC), which are linked by intra− and inter−chain disulfide bonds. The molecule incorporates the Fab, hinge, and Fc regions derived from an anti−PD−1 monoclonal antibody, with an anti−VEGFA nanobody fused to the hinge region of the heavy chain via flexible linkers ((G₄S)₃ and (G₄S)₁) ( Figure 1A ). The intact molecular weight of JS207 is approximately 180.5 kDa.

The binding activity of JS207 to human PD−1 and a panel of related immune proteins was examined by ELISA. As shown in Figures 1B, C , JS207 exhibited concentration−dependent binding to human PD−1. In contrast, it did not bind to human BTLA, CTLA−4, CD28, PD−L1, PD−L2, or ICOS. Comparative analysis using toripalimab (an in−house, commercially approved anti−PD−1 antibody, also known as JS001) and AK112 demonstrated that all three antibodies specifically bound to human PD−1. The calculated EC₅₀ values were 10.4 ng/mL (58.6 pM) for JS207, 8.9 ng/mL (60.5 pM) for toripalimab, and 20.1 ng/mL (100.0 pM) for AK112.

The binding of JS207 to human vascular endothelial growth factor (VEGF) family proteins, including VEGFA, VEGFB, VEGFC, VEGFD, VEGFE, and placental growth factor (PLGF), was evaluated by ELISA. As shown in Figures 1D, E , JS207 specifically bound to human VEGFA, with no detectable binding to human VEGFB, VEGFC, VEGFD, VEGFE, or PLGF.

The ability of JS207 to inhibit the binding of human PD-1 to its ligands (PD-L1 and PD-L2) and to block the interaction between human VEGFA and VEGFR2 was assessed using blocking ELISA. In these assays, toripalimab and VEGF-DotAb (an in-house anti-VEGFA Dotbody) served as positive controls, while an anti-KLH hIgG4 antibody was used as the negative control.

As presented in Figures 1F, G , JS207 effectively blocked the interaction between human PD-1 and PD-L1 with an IC₅₀ of 1149 ng/mL (6.37 nM) and between PD-1 and PD-L2 with an IC₅₀ of 776.6 ng/mL (4.3 nM). Moreover, JS207 inhibited the binding of human VEGFA to VEGFR2 with an IC₅₀ of 603.9 ng/mL (3.35 nM). In contrast, toripalimab only blocked the PD-1/PD-L1 and PD-1/PD-L2 interactions, with IC₅₀ values of 883.7 ng/mL (5.56 nM) and 587.7 ng/mL (3.92 nM), respectively, but did not block the VEGFA/VEGFR2 interaction. VEGF-DotAb specifically inhibited VEGFA binding to VEGFR2, with an IC₅₀ of 221.6 ng/mL (1.48 nM) ( Figure 1H ).

To evaluate the cross-species reactivity, the binding of JS207 to PD-1 proteins from different species was assessed by ELISA. As shown in Figures 2A–C , JS207 bound potently to human PD-1 and cynomolgus (cyno) PD-1 with EC₅₀ values of 8.2 ng/mL (45 pM) and 17.2 ng/mL (95 pM), respectively, while no binding was observed for rat or mouse PD-1. The negative control anti-KLH IgG4 exhibited no binding.

Similarly, the cross-species binding of JS207 to VEGFA was examined. As depicted in Figures 2D, E , JS207 bound to human and cyno VEGFA with an EC₅₀ of 4.2 ng/mL (23 pM) and also recognized rat and mouse VEGFA with EC₅₀ values of 5.3 ng/mL (29 pM) and 4.8 ng/mL (27 pM), respectively.

The binding affinity and kinetics of JS207 toward human PD-1 were characterized using surface plasmon resonance (SPR). Figure 3A shows the binding profile of JS207 to human PD-1, which is similar to that of toripalimab ( Figure 3C ), a finding that is consistent with the fact that the anti-PD-1 domain of JS207 is derived from toripalimab. In contrast, AK112 displayed a distinct profile ( Figure 3B ), with slower association and faster dissociation kinetics compared to JS207 and toripalimab. The equilibrium dissociation constant (K_D) for JS207 binding to human PD-1 was determined to be 4.60×10^-10 M, approximately 11-fold higher affinity than that of AK112 (5.05×10^-9 M). In addition, the PD-1 binding affinity of JS207 was comparable to that of toripalimab (5.55×10^-10 M) ( Figure 3G ).

The binding profile of JS207 to human VEGFA was similarly assessed by SPR. As shown in Figures 3D–F , the binding traces of JS207 were similar to those observed for AK112 and VEGF-DotAb. All three antibodies exhibited very high affinities for human VEGFA, with K_D values of 9.00×10^-12 M for JS207, <2.59×10^-11 M for AK112, and <7.14×10^-12 M for VEGF-DotAb ( Figure 3H ).

Because JS207 exhibits high affinity for human VEGFA, intravenous administration is expected to result in the rapid formation of a JS207/VEGFA complex in circulation prior to its engagement with PD-1 within the tumor microenvironment. To determine whether this preformed complex retains its ability to bind human PD-1, we employed an Octet-based assay ( Figure 3I ). In this experiment, human PD-1 conjugated with a mouse Fc fragment was immobilized on an AMC biosensor. Next, JS207, the preformed JS207/VEGFA complex, AK112, and the AK112/VEGFA complex were injected ( Figures 3I–N ). As shown in Figure 3K , the JS207/VEGFA complex produced a strong binding signal to human PD-1, with a K_D value of <1.0×10^-12 M, comparable to JS207 alone ( Figures 3J–L ). Similarly, the AK112/VEGFA complex bound to PD-1 with an affinity similar to that of AK112 alone ( Figures 3L–N ). These results indicate that both JS207/VEGFA and AK112/VEGFA complexes maintain robust binding activity to human PD-1.

To assess whether JS207 can bind human PD-1 and human VEGFA simultaneously, we designed an SPR assay. In this experiment, AK112 was used as a comparator, while toripalimab and VEGF-DotAb served as positive controls for PD-1 and VEGFA binding, respectively. In one format, human PD-1 was first captured; subsequent injection of JS207, followed by VEGFA, produced an additional binding response ( Figure 3O ), demonstrating that JS207 can engage both antigens simultaneously. A similar dual-binding profile was observed for AK112 ( Figure 3P ). In contrast, toripalimab bound exclusively to PD-1 ( Figure 3S ), and no binding was detected in the buffer control ( Figure 3T ). In a complementary approach, when human VEGFA was immobilized first, JS207 was subsequently able to bind human PD-1, further confirming JS207’s dual-binding capability ( Figures 3Q, R ). AK112 exhibited a similar binding pattern as JS207 ( Figures 3P, R ). As expected, toripalimab bound only to PD-1, while VEGF-DotAb bound exclusively to VEGFA ( Figure 3U ); no binding signal was observed in the corresponding buffer control ( Figure 3V ).

The anti−PD−1 potency of JS207 was evaluated using a PD−1/PD−L1 reporter gene assay (RGA). Jurkat effector cells stably overexpressing human PD−1 and an NFAT−driven luciferase reporter were co−cultured with CHO target cells expressing human PD−L1. JS207 effectively blocked the PD−1/PD−L1 interaction, thereby promoting T cell activation. The EC₅₀ value for JS207 was 2.89 nM with a signal-to-noise (S/N) ratio of 4.95. In parallel, the EC₅₀ values and S/N ratios for AK112 and toripalimab were 8.34 nM, 4.47 and 3.49 nM, 6.0, respectively. Hence, the anti−PD−1 potency of JS207 is comparable to that of toripalimab and superior to that of AK112 ( Figure 4A ).

The anti−VEGFA activity of JS207 was assessed using a VEGFA/VEGFR2 RGA in H293 cells engineered to overexpress VEGFR2. As shown in Figure 4B , JS207, AK112, and VEGF−DotAb effectively inhibited the binding of VEGF to VEGFR2. The IC₅₀ value for JS207 was 0.773 nM with an S/N ratio of 4.49; for AK112, the corresponding values were 1.131 nM and 4.79; and for VEGF−DotAb, 0.479 nM and 4.83, respectively. Thus, JS207 exhibited anti−VEGF activity comparable to or slightly better than that of AK112 but was marginally less potent than VEGF−DotAb ( Figure 4B ).

The ability of JS207 to inhibit human umbilical vein endothelial cell (HUVEC) proliferation was examined in vitro. HUVECs were cultured in the presence of 10 ng/mL VEGFA along with serial dilutions of JS207, AK112, and VEGF−DotAb (0.005–10.0 nM). The IC₅₀ values and S/N ratios were determined to be 0.296 nM, 2.76 for JS207, 0.328 nM, 2.41 for AK112, and 0.243 nM, 3.22 for VEGF−DotAb. These data indicate that the inhibitory effect of JS207 on HUVEC proliferation is comparable to or marginally better than that of AK112, albeit somewhat less potent than VEGF−DotAb ( Figure 4C ).

The effect of VEGFA on the anti−PD−1 activity of JS207 was investigated using the PD−1 RGA. As indicated in Figure 4D , the presence of 20.8 nM VEGFA markedly enhanced the anti−PD−1 activity of JS207 compared with JS207 alone. Further increasing the VEGFA concentration to 333.3 nM elevated the maximum response (upper asymptote) of JS207’s anti−PD−1 potency. VEGFA at 1000 nM also enhanced the anti−PD−1 potency of AK112 with similar extent as JS207, whereas toripalimab remained unaffected ( Figure 4E ).

Cell−based binding studies using H293/PD−1 cells demonstrated that VEGFA enhanced the binding of both JS207 and AK112 to PD−1, while toripalimab showed no such effect ( Figure 4F ). Similar results were observed in PD-1 expressing Jurkat/PD−1 cells ( Supplementary Figure S1 in Supplementary Data Sheet ).

The mixed lymphocyte reaction (MLR) system was used to assess JS207’s effect on T cell activation. In this assay, in vitro–induced mature dendritic cells and CD4⁺ T cells from four donors were co−incubated with JS207 and control antibodies over a concentration range of 15 pM to 150 nM. JS207 significantly promoted the release of IL−2 and IFN−γ in a dose−dependent manner. The cytokine responses induced by JS207 were comparable to those observed for AK112, toripalimab, and the combination of toripalimab plus VEGF−DotAb ( Figures 4G, H ). Notably, VEGF−DotAb alone did not induce IL−2 or IFN−γ release relative to the anti−KLH hIgG4 control.

To investigate PD−1 internalization, two complementary methods were employed: the cell surface residual PD−1 assay and the intracellular fluorescence assay.

Cell Surface Residual PD-1 Method: Human PD−1–expressing H293 cells were incubated at 4°C with 30 nM of JS207, AK112, or toripalimab for 30 minutes in the presence or absence of VEGFA (60 nM). After washing to remove unbound antibodies, cells were stained with PE−labeled anti−PD−1 human IgG to quantify the residual cell surface–bound PD−1. In the absence of VEGFA, approximately 16–20% of surface PD−1 was internalized within 30 minutes, increasing to 21–32% after 60–240 minutes of incubation for all three antibodies. However, in the presence of VEGFA, JS207− and AK112−treated cells exhibited a marked increase in PD-1 internalization, reaching 46–50% within 30 minutes and up to 65% after 4 hours, while toripalimab was unaffected ( Figures 5A–C ).

Intracellular Fluorescence Method: JS207, AK112, toripalimab, and JS501 (an anti−VEGFA mAb) were conjugated with CypHer5E, a pH−sensitive cyanine dye that exhibits enhanced fluorescence in acidic intracellular compartments. Jurkat/PD−1 cells were incubated with serial dilutions of these CypHer5E−conjugated antibodies at 37°C for 4 hours. At the end of incubation, a noncompetitive anti−PD−1 antibody (MIH4 PD−1−PE) was added to measure the remaining cell surface PD−1. As shown in Figures 5D–F , the intracellular fluorescence intensity of CypHer5E−labeled JS207 and AK112 increased in a dose−dependent manner, while the cell surface PD−1 signal detected by MIH4 PD−1−PE concomitantly decreased, providing further evidence of PD−1 internalization. In the presence of VEGFA, both fluorescence intensity and PD−1 internalization were further enhanced for JS207 and AK112. In contrast, toripalimab–induced internalization was not affected by VEGFA ( Figure 5F ). Only minimal internalization was observed when cells were incubated at 4°C ( Figure 5G ).

In the absence of VEGFA, JS207 and toripalimab exhibited comparable internalization activities with EC₅₀ values of 62.3 ng/mL (0.35 nM) and 58.4 ng/mL (0.39 nM), respectively, both of which were more potent than AK112 (EC₅₀ = 322.7 ng/mL, or 1.61 nM). To assess the impact of varying VEGFA concentrations on PD−1 internalization, a PD−1 RGA was performed using a fixed concentration of JS207 (25 nM) and AK112 (25 nM). As shown in Figure 5I , when VEGFA concentrations were low (0.017–1.37 nM), JS207 exhibited higher anti-PD−1 potency than AK112. However, at high VEGFA concentrations (111.1 nM and 333.3 nM), both antibodies demonstrated comparable anti-PD−1 potencies, with EC₅₀ values of 11.52 nM for JS207 and 11.62 nM for AK112 ( Figure 5I ).

The thermal stability is an important factor to be considered during antibody CMC manufacturing and storage. In general, due to multiple domain composition, BsAbs tend to have lower thermal stability compared to conventional monoclonal antibodies that add additional challenge during BsAb development (34, 35). The thermal stability of JS207 was evaluated using two potency assays: anti−PD−1 RGA and anti−VEGFA RGA.

At 40°C, neither JS207 nor AK112 exhibited significant changes in anti−PD−1 potency after 24, 48, or 96 hours of heat stress ( Figures 6A, B ). Similarly, at 40°C, no significant changes in anti−VEGFA potency were observed for JS207 and AK112 after 1, 4, and 6 days of incubation ( Figures 6E, F ). Under extended periods of high temperature (50°C, up to 6 days), JS207 showed a time-dependent increase in IC50 values (potency decrease) but retained an unchanged S/N ratio for anti-VEGFA potency ( Figure 6G ). Under the same stress condition (50°C, up to 6 days), AK112 exhibited increases in IC50 values (potency decrease) and a decrease in the S/N ratio for anti-VEGFA potency ( Figure 6H ).

At 55°C, AK112 lost nearly all anti-PD−1 activity by 24 hours and completely lost anti-PD-1 activity by 48 hours. In contrast, JS207 maintained relatively strong anti−PD−1 activity, retaining 65% potency at 24 hours and 31% at 48 hours compared to time 0. Furthermore, even after 4 hours at 65°C, JS207 retained 15.5% anti-PD-1 activity while AK112 had no activity at all ( Supplementary Figure S2 in Supplementary Data Sheet ). These results demonstrate that JS207 possesses enhanced thermal stability in anti-PD-1 potency. These results support previous finding that Fab format (anti-PD-1 domain of JS207) is more stable than scFv format (anti-PD-1 domain of AK112) (36).

To evaluate the in vivo anti−tumor activity of JS207, mouse colon cancer MC38 cells were subcutaneously implanted into C57BL/6-Pdcd1^{tm1(PDCD1)Bcgen}/Bcgen humanized mice (abbreviated as B-hPD-1 mice). As shown in Figures 7A, C , JS207 significantly inhibited tumor growth in a dose−dependent manner when administered at 0.75, 1.5, and 4.5 mg/kg, achieving tumor growth inhibition (TGI) rates of 76.1%, 78.0%, and 84.4%, respectively, at day 20 post−treatment. At equivalent molar doses, JS207 (0.75 mg/kg) exhibited superior anti−tumor activity compared to toripalimab monotherapy (0.6 mg/kg) or toripalimab plus VEGF−DotAb combination therapy (0.6 mg/kg + 0.33 mg/kg). Notably, none of the treatment groups showed significant body weight loss or other overt side effects, indicating that JS207 was well tolerated ( Figure 7E ).

The anti−tumor efficacy of JS207 was further examined in a human malignant melanoma A375 model using human PBMC transplanted NDG mice. In the saline−treated group, the mean tumor volume reached 585 ± 83 mm³ at day 21. In contrast, mice treated with JS207 at doses of 1, 3, and 10 mg/kg exhibited mean tumor volumes of 362 ± 38 mm³, 344 ± 32 mm³, and 262 ± 25 mm³, corresponding to TGI rates of 49.6%, 53.7%, and 72.0%, respectively ( Figures 7B, D ). Additionally, mice treated with AK112 at 11.1 mg/kg displayed a mean tumor volume of 349 ± 62 mm³ with a TGI of 52.7%. At equivalent molar doses (AK112 at 11.1 mg/kg versus JS207 at 10 mg/kg), JS207 demonstrated superior anti−tumor efficacy compared with AK112. Importantly, the body weights of animals in the treatment groups did not significantly differ from those in the saline group, underscoring the favorable tolerability of both JS207 and AK112 ( Figure 7F ).