HEK293T, Vero E6, K562, and THP-1 cells were obtained from the American Type Culture Collection (ATCC, Rockville, MD, USA), and Expi293F cells were procured from Thermo Fisher Scientific (Waltham, MA, USA). Dulbecco’s modified Eagle medium (DMEM, Thermo Fisher Scientific) was used for HEK293T and VeroE6 cells, while Roswell Park Memorial Institute (RPMI, Thermo Fisher Scientific) media was used for K562 and THP-1 cells. Both media were supplemented with 10% (v/v) fetal bovine serum (FBS, Thermo Fisher Scientific) and 100 U/mL penicillin–streptomycin (Thermo Fisher Scientific). The cells were maintained at 37°C with 5% CO₂. Expi293F cells were cultured in Expi293™ Expression Media in shaking incubators at 37°C, 125 rpm, and 8% CO₂. Cell counting was performed using ADAM™ CellT (NanoEntek, Seoul, Republic of Korea).

The SARS-CoV-2 FP peptide (SFIEDLLFNKVTLADAGF) with a cysteine residue at the C-terminus was chemically synthesized and conjugated to bovine serum albumin (BSA) or ovalbumin (OVA) by Peptron (Daejeon, Republic of Korea). Reverse-phase high-performance liquid chromatography (HPLC) was performed to purify the peptide conjugates using a Vydac Everest C18 column (250 mm × 22 mm, 10 μm; Uvison, Sevenoaks, United Kingdom). The purity of all synthesized peptides was confirmed to be greater than 95%, and their molecular mass was verified by liquid chromatography–mass spectrometry using an Agilent HP1100 series HPLC system (Santa Clara, CA, United States).

Biopanning was performed from a human recombinant single-chain variable fragment (scFv) library using phage display technology to isolate SARS-CoV-2 FP-specific human mAbs. Four rounds of biopanning were carried out with M-270 epoxy Dynabeads™ (Invitrogen, Carlsbad, CA, USA), which were covalently coupled with 4 μg of either FP-BSA or FP-OVA. The first and third rounds of biopanning used FP-BSA-conjugated beads, and the second and fourth rounds used FP-OVA-conjugated beads to select FP-specific human scFv antibody clones. After completing the four rounds of biopanning, randomly selected 96 phage clones from the output colonies were then subjected to a phage enzyme-linked immunosorbent assay (ELISA) to test their reactivity to target antigens. Following DNA sequencing, scFv antibody clones with unique complementarity-determining region (CDR) sequences were chosen for further study.

To identify binders specific to SARS-CoV-2 FP, each colony from the 4^th round of biopanning was cultured in 1 mL of Super Broth (SB) media with 50 μg/mL of carbenicillin in 96-deep-well plates (Axygen, Union City, CA, USA) at 37°C overnight. Then, 10¹⁰ plaque-forming units (PFU) of helper phages (VCSM13; Agilent, Santa Clara, CA, USA) were added into each well, followed by the addition of kanamycin to the media with a final concentration of 70 μg/mL, and the mixture was incubated overnight at 37°C. The supernatant was obtained by centrifugation at 3,000 g and used for phage ELISA. High-binding 96-well microplates (Corning, Corning, NY, USA) were coated overnight with 0.1 μg of either FP-BSA or FP-OVA in phosphate-buffered saline (PBS) at 4°C. After blocking with 3% (w/v) BSA in PBS, the plates were incubated with 100 μL of phage supernatant at 37°C for 2 h. Subsequently, the plates were washed three times with PBS containing 0.05% (v/v) Tween 20 (PBST). HRP-conjugated anti-hemagglutinin antibody (1:3,000; Bethyl Laboratories, Montgomery, TX, US) was added and incubated at 37°C for 1 h. For colorimetric detection, 3,3’,5,5’-tetramethylbenzidine (TMB) substrate solution (Thermo Fisher Scientific) was added to the wells. The reactions were stopped by adding 1 M of H₂SO₄, and the absorbance was measured at 450 nm using a microtiter plate reader (Bio-Tek Instruments, Winooski, VT, USA).

The selected scFv clones were cloned into an IgG4-encoding pCEP4 mammalian expression vector (Thermo Fisher Scientific) to generate scFv-Fc antibodies. Here, IgG4 was engineered to contain an S228P mutation to prevent IgG4 Fab-arm exchange. Following biochemical and functional characterization, an scFv clone with strong affinity and a high production yield was designated as K107.1 and used as the parental antibody for bsAb generation.

We designed three types of IgG4-(scFv)₂ fusion proteins by fusing the scFv region of K107.1 with K102.1, an anti-SARS-CoV-2 RBD-specific IgG4 with an S228P mutation, to create bsAbs. The scFv was linked to the C-terminus of the heavy chain of K102.1 (type A), the C-terminus of the light chain of K102.1 (type B), or the N-terminus of the light chain of K102.1 (type C). Each recombinant DNA was transiently expressed using the Expi293 Expression System (Thermo Fisher Scientific) following the manufacturer’s recommendations. The overproduced antibodies were purified from the culture media using affinity column chromatography with Protein A-Sepharose (Repligen, Waltham, MA, USA) as previously described (27).

The real-time interaction between antibodies and antigens was assessed using surface plasmon resonance (SPR) with a Biacore T200 instrument (Cytiva, Marlborough, MA, USA). The CM5 sensor chip (Cytiva) was prepared by activating its carboxyl groups using a mixture of 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide and N-hydroxysuccinimide. Recombinant wild-type SARS-CoV-2 RBD or FP-BSA was then covalently immobilized on the sensor chip surface for the immobilization of 250 response units through standard amine coupling. To evaluate the binding interaction, FP-specific antibodies (K107.1, K107.2, and K107.3) were injected onto the sensor chip with immobilized FP-BSA, and increasing concentrations of the antibodies (8, 16, 32, 64, and 128 nM) were used at a flow rate of 50 µL/min. Similarly, the bsAb was injected onto the sensor chip containing either immobilized wild-type SARS-CoV-2 RBD or FP-BSA at a flow rate of 50 µl/min. After each binding cycle, the sensor chips were regenerated by injecting 10 mM glycine-HCl (pH 3.0), which effectively removed any bound antibodies from the sensor chip surface. The equilibrium dissociation constants (K_D) were determined using Biacore T200 Control Software Version 3.2 (Cytiva).

To further verify the properties of K203.A bsAb and its independent binding to different target sites, additional real-time binding experiments were conducted. In this experiment, FP-BSA or RBD was first immobilized on the CM5 sensor chip surface. Then, 1 μM of K203.A was introduced onto the sensor chip surface to saturate the binding sites corresponding to the FP. Next, purified SARS-CoV-2 RBD or FP-BSA was introduced onto the same sensor chip with the K203.A-saturated surface. The generated binding curves were analyzed using Prism Software 8.0 (GraphPad Software, La Jolla, CA, USA).

A protein thermal shift assay was conducted to investigate the thermal stability of the K203.A antibody. In each well of a MicroAmp^® Optical 8-tube strip (Applied Biosystems, Foster City, CA, USA), 5 μg of K203.A antibody was added. Additionally, 2.5 μL of Protein Thermal Shift Dye (8×, Applied Biosystems, Foster City, CA, USA) was added. For the negative control, PBS was mixed with Protein Thermal Shift Dye. The thermal shift measurements were performed using a real-time PCR instrument (QuantStudio™ 3 Real-Time PCR, Applied Biosystems), according to the manufacturer’s instructions. Furthermore, all experiments were performed in duplicate to ensure reliability and accuracy.

The inhibitory potency of selected antibodies on the interaction between the purified SARS-CoV-2 RBD and hACE2 was evaluated using ELISA. Fc-tagged hACE2 (hACE2-Fc; R&D Systems, Minneapolis, MN, USA) was immobilized on 96-well plates at a concentration of 50 ng per well. After blocking with a blocking buffer (BPS Bioscience), 25 nM of purified His-tagged SARS-CoV-2 wild-type or Delta variant RBD (Sino Biological) was preincubated with either K102.1 mAb or K203.A bsAb at concentrations of 1 nM or 10 nM for 1 h at 25°C. The preincubated mixtures of RBD and antibodies were then added to the wells containing immobilized hACE2-Fc and incubated for an additional 1 h. An HRP-conjugated anti-His secondary antibody (BPS Bioscience) was used for detection. Furthermore, the neutralizing activity of the selected antibodies against the RBD-hACE2 interaction was assessed by measuring chemiluminescence intensity with a Synergy H1 microplate reader (Bio-Tek Instruments).

Replication-deficient Moloney murine leukemia virus (MLV) particles were used to express variants of the SARS-CoV-2 S protein and a firefly luciferase reporter gene. The MLV particles were obtained from eEnzyme (Gaithersburg, MD, USA). HEK293T/hACE2 cells (1 × 10⁴ cells in 50 μL culture medium) were seeded in 96-well tissue culture plates and allowed to adhere overnight. The following day, serially diluted K102.1 mAb, K203.A bsAb, or K107.1 mAb at a 200 nM concentration were each preincubated with pseudovirus (1 × 10⁷ PFU/mL) for 10 min at 25°C to allow interaction with the virus. After the preincubation, the antibody–pseudovirus mixtures were added to the cells and incubated for 24 h. To assess viral infection, the presence of the firefly luciferase reporter gene was assessed using ONE-Glo™ luciferase substrate (Promega, Madison, WI, USA). In addition, luminescence signals were measured with a Synergy H1 microplate reader (Bio-Tek Instruments), and the half-maximal inhibitory concentration (IC₅₀) values were determined using Prism Software 8.0 (GraphPad Software).

SARS-CoV-2 wild-type (BetaCoV/Korea/KCDC03/2020, NCCP no. 43326) and Delta variant (hCoV-19/Korea/KDCA119861/2021, NCCP no. 43390) were obtained from the Korea Disease Control and Prevention Agency (KDCA, Osong, Republic of Korea). These viruses were propagated in Vero E6 cells and titrated using a plaque assay. Subsequently, they were stored at −80°C for subsequent in vitro infection assessments. For the in vitro infection assessment, VeroE6 cells (1.5 × 10⁴ cells in 100 μL of culture medium) were seeded on 96-well tissue culture plates and allowed to adhere overnight. Serial dilutions of the antibodies were preincubated with SARS-CoV-2 (at a concentration of 4 × 10² TCID₅₀/mL) for 1 h at 25°C to allow the interaction of antibodies with the virus. After the preincubation period, the virus/antibody mixture was added to the Vero E6 cells. The cells were then incubated with the virus/antibody mixture for three days. For the determination of neutralization potency, reverse transcription-quantitative polymerase chain reaction (RT-qPCR) was performed for viral RNA quantification using the QuantiTect SYBR^® Green RT-PCR kit (Qiagen, Hilden, Germany).

For in vivo efficacy studies, 8-week-old male B6.Cg-Tg(K18-ACE2)2Prlmn/J mice (The Jackson Laboratory, CA, USA) were used. These mice were transgenic for hACE2 and served as a model for studying SARS-CoV-2 infection. The mice were housed in a certified animal biosafety level 3 (ABSL3) facility at the Ji Seok-Yeong Biomedical Research Institute of Seoul National University Bundang Hospital in Seongnam, Republic of Korea. All experimental procedures were ethically reviewed and approved by the Institutional Animal Care and Use Committee (IACUC; Approval No. BA-2108-325-078). Additionally, the Institutional Biosafety Committee of Seoul National University Bundang Hospital approved biosafety experimental protocols (Approval No. IBC-2105-A-008) to ensure the safety of researchers and animals during the study. The hACE2-transgenic (TG) mice (n = 4) were intranasally inoculated with 50 μL of Delta variant virus (1 × 10⁴ PFU) under anesthesia. After three hours of infection, either PBS, 5 mg/kg or 30 mg/kg of K203.A bsAb was intravenously injected. Six days after the SARS-CoV-2 Delta variant infection, lung tissues were harvested from the hACE2-TG mice. Furthermore, the viral RNA present in the lung tissues was quantified using RT-qPCR.

The mice were euthanized by CO₂ as per approved animal welfare guidelines. The lung tissues were then collected and fixed in 10% neutral buffered formalin (Sigma, St. Louis, MO, USA) for 24 h. Furthermore, the lung tissues were processed for paraffin embedding. In this procedure, the tissues were dehydrated using a series of alcohol solutions and then infiltrated with paraffin wax. The paraffin-embedded tissues were then sectioned at a 4 μm thickness using a microtome. The sections were subjected to hematoxylin and eosin staining. Histopathological examination was performed using light microscopy (Olympus, Tokyo, Japan).

For the in vivo toxicity and serum pharmacokinetic analysis, 8-week-old female Institute of Cancer Research (ICR) mice (Orient Bio Inc., Seongnam, Republic of Korea) were used. The study received approval from the Institutional Animal Care and Use Committee (IACUC) of the National Cancer Center, Republic of Korea (Approval No. NCC-21-693). A total of three mice were included in the study. They received an intravenous injection of either 5 mg/kg or 30 mg/kg of K203.A bsAb. The body weight of each mouse was monitored regularly for 21 days post-injection to evaluate the potential systemic toxicity of K203.A. At the end of the 21-day period, blood and serum samples were obtained from the mice. The levels of several biochemical markers, including glutamic oxaloacetic transaminase (GOT), glutamic pyruvic transaminase (GPT), total bilirubin (TBIL), creatinine (CRE), and blood urea nitrogen (BUN), were quantified from these serum samples using a Fuji Dri-Chem 3500 Biochemistry Analyzer (Fujifilm, Tokyo, Japan).

To determine the pharmacokinetic profile of K203.A, 50 µL of blood samples were collected from each mouse at specific time points: 4, 8, 24, 72, 120, 168, 264, 384, and 504 h post-injection. The blood samples were centrifuged at 5,000 × g for 20 minutes at 4°C for the collection of the serum samples. The K203.A concentration in the serum samples was measured at each time point using a human IgG ELISA kit (Abcam) based on the manufacturer’s instructions. Optical density values were measured at 450 nm using a Synergy H1 microplate reader (Bio-Tek Instruments).

To investigate antibody-dependent enhancement (ADE), HEK293T, HEK293T/hACE2, K562, or THP-1 cells (1 × 10⁴ cells/well) were seeded in 96-well plates. Different concentrations of K203.A bsAb (0.044 to 100 nM) were preincubated with each variant of the SARS-CoV-2 pseudovirus (wild-type, D614G, Alpha, Beta, Gamma, Delta, and Kappa; 1 × 10⁷ PFU/mL) for 30 min at 25°C. The mixtures of bsAb and pseudoviruses were added to the respective cells in the 96-well plates, which were then cultured for 24 h. To assess the level of viral infection in the cells, luciferase activity was measured as described in Section 2.8.

The statistical analysis of the data was performed using Prism Software 8.0 (GraphPad Software). A two-tailed Student’s t-test was used when comparing data between two groups, and a one-way analysis of variance (ANOVA) with Bonferroni’s correction was used for multiple comparisons between more than two groups. Results are expressed as the mean ± standard deviation (SD). Statistical significance was considered at P < 0.05 (*), P < 0.01 (**), and P < 0.001 (***).

We employed biopanning, a powerful approach based on phage display technology (27), to isolate human antibodies that target the SARS-CoV-2 FP. Using this iterative technique, we subjected a well-established human recombinant scFv library containing SARS-CoV-2 FP-specific scFvs to a series of selection rounds ( Figure 1A ). During each round, we exposed the scFv library to immobilized FP-BSA and FP-OVA, allowing the scFvs with binding affinity to the target FP to interact with and attach to the solid support. Unbound scFvs were subsequently washed away to eliminate non-specific binders, while the specifically bound scFvs were eluted and amplified using host bacteria for subsequent rounds of selection.

A phage ELISA assay showed that we isolated a variety of scFv clones with strong binding affinity to FP-BSA and FP-OVA following numerous rounds of biopanning ( Figure 1B ). Importantly, these scFv clones did not bind to BSA, which was used as a negative control, thus confirming their specificity for the SARS-CoV-2 FP. The biopanning process served as an effective and robust method for enriching the human antibody library with scFvs that specifically recognize the SARS-CoV-2 FP.

To further refine our selection, we chose three scFv clones with distinct complementarity-determining region sequences, all of which exhibited specific binding and strong affinity to the SARS-CoV-2 FP. Subsequently, we transformed these scFv clones into human scFv-Fc antibodies by fusing the scFvs to the Fc region of human immunoglobulin G (IgG). These recombinant human scFv-Fc antibodies were produced using a suspension-adapted HEK293 cell line (Expi293F), an expression system optimized for large-scale antibody production. Furthermore, we purified antibodies using Protein A-Sepharose affinity chromatography. The purity of these antibodies exceeded 90%, as shown in Supplementary Figure 1 , and the production yields were approximately 50 mg/L, 30 mg/L, and 80 mg/L, respectively, as illustrated in Supplementary Figure 2 .

To evaluate the binding kinetics between the chosen antibodies and SARS-CoV-2 FP, we conducted real-time kinetic analysis and determined the dissociation constant (K_D) values for K107.1, K107.2, and K107.3 against SARS-CoV-2 FP as 7.18 nM, 17.29 nM, and 34.57 nM, respectively ( Figures 1C–E and Table 1 ). Among these antibodies, K107.1 exhibited the most remarkable binding affinity towards the SARS-CoV-2 FP, suggesting a strong and specific interaction. Based on these results, we chose K107.1 as the parental antibody specific to the FP.

Three distinct bsAbs utilizing the IgG4-(scFv)₂ format were designed to target the RBD and FP of SARS-CoV-2. These were generated based on two key components: the previously identified SARS-CoV-2 RBD-specific neutralizing mAb, K102.1, and the newly discovered FP-specific antibody, K107.1 (27). The three types of bsAbs, designated as K203.A, K203.B, and K203.C, were engineered to optimize their binding properties ( Figures 2A–C ). K203.A was designed with scFvs attached to the C-terminus of each IgG4 heavy chain. In K203.B, the scFvs were linked to the C-terminus of the IgG4 light chains, while K203.C was constructed with scFvs connected to the N-terminus of each IgG4 light chain.

We incorporated the S228P mutation into the IgG4 region of each bsAb to ensure their stability and homogeneity. This mutation was strategically introduced to prevent unwanted Fab arm exchange, which could lead to heterogeneous antibody mixtures due to half-molecule exchange with endogenous IgG4 (28). By adopting the IgG4-(scFv)₂ format and incorporating the S228P mutation, we aimed to create bsAbs with improved stability and specific binding capacity to both the RBD and FP of SARS-CoV-2.

Upon successful overproduction and purification, the final production yields of purified K203.A, K203.B, and K203.C were approximately 80, 26, and 8 mg/L, respectively ( Figure 2D ). The purity of the resulting bsAbs was over 90%, verified through sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and Coomassie Brilliant Blue staining ( Supplementary Figure 3 ). Given the importance of production yield in antibody development, we selected K203.A, which had a superior production yield, as the prime bsAb for additional studies.

To assess the binding affinity of K203.A to both SARS-CoV-2 RBD and FP, we conducted real-time SPR analysis using wild-type SARS-CoV-2 RBD and FP-BSA as binding partners. The obtained K_D values for K203.A against SARS-CoV-2 RBD and FP were 1.65 nM and 8.76 nM, respectively ( Figures 2E, F , and Table 2 ). Moreover, we evaluated the thermal stability of K203.A through a protein thermal shift (PTS) assay, which measures the melting temperature (T_m) of the antibody. K203.A exhibited robust thermal stability, with a T_m value of 72.03°C ( Supplementary Figure 4 ).

To corroborate and further solidify the binding affinity of K203.A towards both SARS-CoV-2 RBD and FP, we performed additional experiments using SPR. We utilized sensor chips immobilized with SARS-CoV-2 FP-BSA or SARS-CoV-2 RBD, and first saturated them with K203.A. Subsequently, we introduced additional SARS-CoV-2 RBD or FP to each sensor chip. The results demonstrated the subsequent binding of each applied antigen on both K203.A-saturated sensor chips ( Figures 2G, H ), confirming the dual-targeting potential of K203.A to simultaneously engage with both SARS-CoV-2 RBD and FP.

To determine the inhibitory activity of K203.A bsAb on the interaction between SARS-CoV-2 RBD and hACE2, we carried out interaction-inhibition assays using ELISA with recombinant hACE2 and RBDs from both the wild-type SARS-CoV-2 and the Delta variant. The assays were conducted in the absence and presence of the parental RBD-specific mAb K102.1 and K203.A. The results demonstrated that K203.A exerted a comparable inhibitory effect on the hACE2–RBD interactions as K102.1 ( Supplementary Figure 5 ).

We conducted live virus neutralization assays with the wild-type and the Delta variant of SARS-CoV-2 using the Vero E6 cell line to comprehensively assess the neutralizing activity of K203.A against live viral infections. To assess the neutralization efficacy of K203.A, we systematically increased the concentration of both K102.1 and K203.A. The neutralization efficacy of K203.A was quantified by measuring the expression of the viral open reading frame 1a (ORF1a) gene, a common target in RNA-based quantification methods for SARS-CoV-2 due to its proven effectiveness in COVID-19 diagnosis (29). To perform this quantification, we employed RT-qPCR, a sensitive and widely used technique for precisely measuring the levels of RNA transcripts in a sample. The results revealed that K203.A exhibited markedly enhanced inhibitory efficacy on the expression of the viral gene in both the wild-type SARS-CoV-2 and the Delta variant, thus surpassing the neutralization efficacy of the parental mAb K102.1 ( Figures 3A, B ).

To evaluate the neutralization efficacy of K203.A against the SARS-CoV-2 variants, pseudovirus neutralization assays were performed utilizing a HEK293T cell line stably overexpressing hACE2 (HEK293T/hACE2). The assays encompassed several prominent variants, including D614G, Alpha, Beta, Gamma, and Kappa. Throughout the experiments, both K102.1 and K203.A antibodies were present, so their impact on viral infection could be compared. The results demonstrated that K203.A exhibited significantly higher inhibitory effects on infection by all examined pseudovirus types compared to its parental mAb, K102.1. The IC₅₀ values for K203.A were in the sub- to low-nanomolar range against all tested variants ( Figures 4A–E and Table 3 ). On the other hand, the FP-specific mAb K107.1 demonstrated no statistically significant neutralizing efficacy against wild-type SARS-CoV-2 or the D614G, Alpha, Beta, Gamma, Delta, or Kappa variants at 200 nM ( Supplementary Figure 6 ). These findings suggest that the development of a bsAb targeting both the SARS-CoV-2 RBD and FP, as exemplified by K203.A, has the potential to achieve enhanced neutralization against diverse SARS-CoV-2 variants.

To determine the in vivo neutralization efficacy of K203.A against the SARS-CoV-2 Delta variant, we conducted experiments using K18-hACE2-transgenic (TG) mice that were infected with the virus intranasally. Following a three-hour incubation period, the infected mice were administered two different doses of K203.A through intravenous injection: a low dose of 5 mg/kg and a high dose of 30 mg/kg ( Figure 5A ). At 6 days post-infection (dpi), lung samples were excised from all sacrificed mice, and the viral titer was determined using RT-qPCR. The results revealed a significant reduction in viral titers in both the low-dose and high-dose K203.A-treated groups compared with the PBS-treated groups ( Figure 5B ), highlighting K203.A’s ability to neutralize the virus in the lung tissues of infected mice.

We also conducted a histopathological examination at 6 dpi of the lungs of the mice that were infected with the SARS-CoV-2 Delta variant. The mice that received PBS treatment displayed severe lung damage from the Delta variant infection, with evident pulmonary lesions characterized by pronounced pulmonary edema and alveolar hemorrhage. Conversely, each of the K203.A-treated groups exhibited a marked reduction in pulmonary edema and alveolar hemorrhage, indicating a notable attenuation of lung tissue damage. The histopathological analysis revealed a significant protective effect of K203.A against the deleterious effects of the Delta variant on the lungs ( Figure 5C ). These findings provide strong evidence of the neutralization efficacy of K203.A in mitigating lung pathology caused by the Delta variant infection, highlighting the critical role of bsAbs like K203.A in addressing the challenges posed by evolving SARS-CoV-2 variants.

An in-depth in vivo toxicity study was performed on ICR mice to assess the safety of K203.A. Two different doses of the antibody (5 mg/kg and 30 mg/kg) were administered to the mice through intravenous injections. Throughout the study, we carefully monitored liver and kidney functions, as well as any changes in body weight, to evaluate potential adverse effects. To evaluate liver function, we measured serum concentrations of GOT, GPT, and TBIL. To assess kidney function, we measured BUN and CRE concentrations. The results showed no significant changes in these parameters in the K203.A-treated groups. Thus, the lack of significant alterations in liver and kidney functions, as well as in body weight, suggests the absence of in vivo toxicity associated with the administration of K203.A ( Figures 6A–F ).

Following the in vivo toxicity assessment, we proceeded with an in vivo pharmacokinetic analysis of K203.A to improve our understanding of its behavior in the bloodstream. A 5 mg/kg dose of K203.A was administered to ICR mice, followed by periodic blood sample collection for further examination. The concentration of K203.A in the blood was quantified using ELISA to determine its in vivo half-life. In the pharmacokinetic analysis, K203.A showed an in vivo half-life of 2.09 days in mice ( Figure 6G ). This pharmacokinetic profile suggests a favorable duration of K203.A’s presence in the bloodstream, which may contribute to its sustained neutralization effect.

To further investigate potential adverse effects, we examined the impact of K203.A on ADE. For this investigation, we employed permissive cells expressing hACE2 (HEK293T/hACE2) and cells expressing the Fc gamma receptor (HEK293T, K562, and THP-1), which were infected with multiple SARS-CoV-2 pseudotyped variants, including wild-type, Alpha, Beta, Gamma, Delta, and Kappa. Notably, no significant changes in pseudovirus infection were observed, indicating that K203.A may not induce ADE ( Supplementary Figure 7 ).

Together, these results demonstrate the favorable safety profile of K203.A, as evidenced by the absence of in vivo toxicity, its extended half-life, and the lack of antibody-dependent enhancement effects. These findings enhance the potential of the dual-targeting bsAb strategy as a promising and safe therapeutic option against fast-evolving SARS-CoV-2 variants.