HEK293T cells (ATCC CRL-3216) were cultured in DMEM (Gibco) supplemented with 10% heat-inactivated FCS (Bovogen), 1% of 10,000 U/mL penicillin and streptomycin (Gibco) and 1mM sodium pyruvate at 37 °C in a humidified incubator with 5% CO₂. BHK-21 cells (ATCC CCL-10) were cultured in DMEM (Gibco) supplemented with 5% heat-inactivated foetal calf serum (Bovogen), 1% of 10,000 U/mL penicillin and streptomycin (Gibco) at 37 °C in a humidified incubator with 5% CO₂. ExpiCHO (Thermofisher Scientific) were cultured in ExpiCHO expression medium (Thermofisher Scientific) at 37 °C in a humidified incubator with 7.5% CO₂.

To generate recombinant bat antibodies against henipavirus fusion protein, the variable region of heavy (VH) and light chain (VL) of murine 5B3 or C05 were cloned into bat IgG1 framework region (GenBank: GQ427152.1, ELK10654.1) with human (Hsp) or bat signal peptide (Bsp) sequence in pNBF expression plasmids (National Biologics Facility) (33, 54–56). All constructs were sequence verified by Australian Genome Research Facility (AGRF). Detailed information on construct and signal peptides can be found in Figure 1; Supplementary Figure S1; Supplementary Table 1. Cloning was performed using In-Fusion cloning and stellar bacterial cells (TakaraBio) as per the manufacturer’s recommendation. In brief, the PCR amplicon consisting of variable region (5B3 or C05) and KpnI digested pNBF plasmids were mixed at a 3:1 molar ratio and incubated at 50 °C for 15 minutes. Expression plasmids containing 15μg of light and 10μg heavy chains were added to ExpiCHO cells at a density of 6x10⁶ cells per mL. Seven days post-transfection, cell culture supernatant was harvested by sterile filtration and centrifuging at 4800xg for 30mins. Cell culture supernatant was purified with AKTA Start or Pure (Cytiva) using HiTrap Protein A or Protein G antibody purification columns (Cytiva). Protein A column was washed with buffer containing 25mM Tris Base, 25mM NaCl, pH7.4 and eluted with 100mM sodium acetate, 150mM NaCl, pH3. Protein G column was washed with buffer containing 20mM sodium pyruvate, 150mM NaCl, pH7.4 and eluted with 100mM glycine, pH2.7. Antibodies were then sterile filtered, concentrated and buffer exchanged into PBS pH7.4, and concentration quantified using Nanodrop One (ThermoFisher). For antigens, NiV F and Influenza HA (A/Brisbane/59/2007) Clamps were produced as previously described (57–59).

Using SDS-PAGE (Bio-Rad), molecular weight and purity of proteins were assessed under reducing (100mM dithiothreitol) by loading 5μg of boiled sample onto stacking (4%) and resolving (12.5%) polyacrylamide gel. After gel electrophoresis, the gel was stained with R-250 Coomassie brilliant blue for 1 hour, destained overnight with 40% methanol and rinsed with Milli-Q water. Using analytical SEC, purified antigens and antibodies were further evaluated for aggregates and oligomeric states. 50-100μg of sample in PBS were manually loaded onto a 500μL loop before passing it through Superose 6 Increase 10/300 GL column and AKTA Pure (Cytiva). Samples corresponding to retention volume with absorbance peaks were collected in 1mL fractions in 96-well DeepWell plates and data we normalised to the highest absorbance observed per run as relative mAU. The molecular weight of proteins was established with reference to three standard proteins.

Nunc MaxiSorp flat-bottom plates (ThermoFisher) were coated with 50μL PBS containing antigens at a concentration of 2μg/mL and incubated overnight at 4 °C. Nonspecific binding was blocked by incubating all wells on plate with 150μL of blocking agent (5% Milk Diluent (SeraCare) in PBS with 0.1% TWEEN-20 for one hour at room temperature. After blocking, antigens were probed with 50μL of serially diluted sera from immunised bats (starting at 1 in 10 dilution) or primary antibodies (Human: h5B3, hC05, anti-Clamp; Bat: b5B3Hsp, b5B3Bsp, bC05Hsp, bC05Bsp; Goat: goat anti-bat IgG conjugated to HRP (Novus Biologicals)) and incubated at 37 °C for one hour. Plates were then washed thrice in water before adding 50μL of secondary antibodies conjugated with HRP (goat anti-bat (Novus Biologicals, Alpha Diagnostic and Bethyl) or goat anti-human (Sigma Aldrich) were used at a concentration of 0.334μg/mL and 0.167μg/mL, respectively). Plates were then incubated at 37 °C for one hour and were subsequently washed thrice in tap water before drying on a paper towel. After drying, 50μL of warmed TMB (ThermoFisher) was added to plates and developed at room temperature for 5 minutes. 25μL of 1M H₂SO₄ was added to stop the enzymatic reaction before reading the plates at 450nm on Varioskan LUX (ThermoFisher). The background signal was determined by PBS-only wells and subtracted from all readings.

SEC purified proteins (10μg/mL) were coated onto glow-discharged carbon-coated grids (EMS) and incubated for 2 minutes. Grids were then rinsed with three drops of Milli-Q water and stained with 2% uranyl acetate. Micrographs of the sample were collected with HITACHI HT7700 operated at 120 keV with 40,000x magnification.

Ten micrograms of purified antibodies were boiled for 10 minutes and chilled on ice. Samples were then reduced with 1μL NP-40, and N-linked oligosaccharides were removed by adding 1μL of PNGase F enzyme (New England Biolabs) before 1 hour incubation at 37 °C. The glycosylation state of samples was then analysed with PAGE as described in SDS-PAGE protocol above.

Pseudovirus neutralisation assays were performed using lentivirus-based pseudotypes as previously described (58, 60–62). Briefly, HEK293T cells were transfected with p8.91 (encoding for HIV-1 gag-pol), CSFLW (lentivirus backbone expressing a firefly luciferase reporter gene) and viral glycoproteins (NiV F and G) using LTX transfection reagent. Supernatants containing pseudotyped virus were harvested at 48- and 72-hours post-transfection, pooled and centrifuged at 1,300xg for 10 minutes at 4 °C to remove cellular debris. For the neutralisation test, HEK293T cells were then seeded overnight at a density of 2 ×10⁴ in 100µL and incubated at 37 °C, 5% CO₂. The antibodies were diluted in serum-free media in triplicate titrated 5-fold and incubated with pseudo-particles added at a dilution equivalent to 10⁶ signal luciferase units in DMEM-10% to final volume of 100µL. The complex was incubated for 1 hour at 37 °C, 5% CO₂. Firefly luciferase activity was then measured with BrightGlo luciferase reagent and a Glomax-Multi⁺ Detection System (Promega) after 48 hours post infection. Pseudotyped virus neutralisation titres were calculated by interpolating the point at which there was 50% reduction in luciferase activity, relative to control antibody.

Immunisations were performed at Colorado State University using a closed, specific pathogen free colony of Jamaican fruit bats. All animal procedures were approved by the Colorado State University (CSU) Institutional Animal Care and Use Committee (protocol 1085) and were in compliance with U.S. Animal Welfare Act. CSU has a captive colony of Jamaican fruit bats, a neotropical fruit bat indigenous to much of South America, Central America and the Caribbean. Colony bats are kept in a free flight room measuring 19’w x 10’l x 9’h. Roosting baskets are hung from the ceiling throughout the room, and drapes of different cloth materials are positioned for hanging and roosting. Ambient temperature is maintained between 20 °C and 25 °C, humidity between 50% and 70%, and a 12-hour light/12-hour dark-light cycle via a computer-controlled system. Diets consist of a combination of fruits (Shamrock Foods, Fort Collins, CO), Tekald primate diet (Envigo, Huntington, UK), molasses, nonfat dry milk and cherry gelatin that are placed in multiple feeding trays around the room once a day. Fresh water is provided. In addition, fruit is hung around the room to stimulate foraging behaviour and serve as enrichment. Bats were immunised twice with NiV F proteins adjuvanted with Addavax (InvivoGen) before blood collection 14 days apart. Blood was collected 14 days after the booster. A maximum blood volume between 1 and 1.5mL is collected in a syringe and transferred to a red top tube (RTT). RTTs sat at room temperature for one hour to allow a clot to form and then centrifuged at 1000xg for 10 min at room temperature. Serum was removed from the clot, placed in a new microcentrifuge tube and stored at -20 °C.

To generate ReBAs, we de novo synthesised and constructed mammalian expression plasmids containing constant heavy chain (Hc) or kappa (κ) light chain (Lc) derived from IgG1 of black flying foxes (Figures 1A, B). A KpnI restriction site was introduced at the N-terminal of Hc or Lc plasmids for subsequent downstream cloning applications. To validate the ReBAs system, we selected two well-characterised antibodies as prototypes. These include anti-henipavirus F protein antibody, 5B3 (54), and the anti-influenza A hemagglutinin (HA) antibody, C05, that targets H1, H2 and H3 subtypes (55). Both Hc and Lc variable domains from these antibodies were cloned in-frame with corresponding constant domains. Sequence alignment of the bat and human Hc constant domain shows that the bat constant region contains five disulphide bonds whereas the human constant region contains six disulphide bonds (Supplementary Figure S1A). Specifically, the missing disulphide bond in the bat Hc region could be attributed to the deletion of the first cysteine at the hinge region (Figure 1C; Supplementary Figure S1). Bat Lc is similar to the human IgG1 Lcκ constant region; both have two interchain disulphide bonds and one C-terminal cysteine that enables intra-chain disulphide formation with the HC (Supplementary Figure S1B). The glycan site position in the bat Hc region is similar to human IgG1 (Figure 1D; Supplementary Figure S1). In addition, to assess the impact of the signal peptide on the recombinant bat antibodies expression, two N-terminal signal peptides, human signal peptide (Hsp) and bat signal peptide (Bsp) derived from black flying fox were incorporated into the antibody Hc. Sequence alignment shows 8 differences between Hsp and Bsp (Figure 1B). These features are summarised in Figures 1A, E, illustrating the envisaged configuration of black flying fox IgG1.

To characterise ReBAs or human IgG, we purified the cell culture supernatant using protein A (pA) or protein G (pG) resin columns. Figures 2A, B depict the SDS-PAGE analysis of the purified products under reducing conditions. Human IgG and recombinant bat antibodies exhibited similar band patterns, characterised by two prominent bands observed at approximately 50 and 25 kDa (Figures 2A, B). These bands correspond to typical mammalian antibody Hc (50kDa) and Lc (25kDa). Nevertheless, we note a small but discernible distinction between the molecular weights of the bat and human antibodies compared to human 5B3 (h5B3) and human C05 (hC05) with ReBAs (Figures 2A, B). The bat Lc consistently showed a smaller size than h5B3 and hC05 Lc (Figure 2A, lanes 2, 3, 5 and 6), whereas the bat Hc (Hsp and Bsp) of ReBAs appears marginally larger than that of h5B3 and hC05 (Figure 2A, lanes 2, 3, 5 and 6). This apparent shift in molecular weight may be attributed to differences in the isoelectric point (pI) between the proteins. Use of either human or bat signal peptides demonstrated no discernible differences, with bat 5B3Hsp (b5B3Hsp) and b5B3Bsp (Figure 2A, lanes 2 and 3) displaying similar band profiles and molecular weights. This result was consistent and also observed for bat C05Hsp (bC05Hsp) and bC05Bsp (Figure 2A, lanes 5 & 6). Similar profiles were observed with pG purified ReBAs (Figure 2B, lane 2, 3, 5 & 6). Still, some samples displayed faint bands outside the expected sizes of 50 and 25kDa (Figures 2A, B), suggesting the presence of impurities. Overall, the results obtained from SDS-PAGE analysis indicate that the recombinant bat antibodies – ReBAs IgG1, share a molecular weight profile similar to human IgG1 antibodies. This finding also shows that both pG and pA purification methods can effectively isolate ReBAs.

To investigate the putative glycan sites within bat and human Hc (Figure 1D), we treated ReBAs with PNGase F, an enzyme that selectively removes N-linked oligosaccharides from proteins. For comparative analysis, h5B3 and bat b5B3 antibodies were examined under reducing conditions, with an untreated sample as the control. Upon treatment with PNGase F, the bands corresponding to Hc positioned at approximately 50kDa exhibited a downward shift compared to those of untreated samples (Figure 2E). This shift demonstrates that glycans were removed from the Hc, indicating that ReBAs are modified by complex N-linked glycosylation similar to human antibodies.

To further evaluate the oligomeric state of bat antibodies, b5B3Hsp and b5B3Bsp antibodies were analysed by SEC on Superose 6 Increase 10/300 GL gel filtration column at physiological pH7.4. For both pA and pG purified antibodies, b5B3Hsp and b5B3Bsp yielded a peak at a retention volume of approximately 17.8mL (Figures 2C, D). Similarly, h5B3 eluted at a retention volume of about 17.8mL (Figures 2C, D). Substitution into the regression equation established using protein standards yielded a predicted molecular weight of approximately ~145 kDa (Supplementary Figure S2), which closely resembles the molecular weight of human IgG1 (63). For pG purified antibodies, both b5B3Hsp and b5B3Bsp also eluted as a single peak at a retention volume of approximately 17.8mL (Figure 2D). These results show that the oligomeric state of ReBAs are similar to human IgG1. In addition, we also visualised the elution peak of SEC purified samples of b5B3Bsp with negative stain transmission electron microscopy (negTEM). The collected micrograph depicts particles of ~10nm in size with distinct Y-shaped morphology (Figure 2F). These features are characteristic of a typical mammalian IgG and demonstrate that our construct yielded recombinant bat antibodies that resemble human IgG1.

To determine if the antigen-binding functions of the ReBAs were preserved, we performed an indirect enzyme-linked immunosorbent assay (ELISA) against immobilised bat-borne NiV F protein and influenza virus HA protein with our panel of ReBAs. These antigens are stabilised by the Molecular Clamp (first-generation) and are described extensively elsewhere (57, 61). Due to the limited data available for anti-bat IgG secondary antibodies, we also tested a small panel of commercially available HRP conjugated anti-bat IgG secondary antibodies from Alpha Diagnostic, Bethyl and Novus Biologicals. After initial testing, we selected goat anti-bat antibodies from Novus Biologics (Cat: NB7238), that was raised against bat IgG Hc and Lc, for its favourable binding (lowest K_D) to ReBAs (Supplementary Figure S3). To further scrutinise this reagent, we supplemented our findings on ReBAs with an investigation into the reactivity of immunised bat sera to NiV F Clamp. Supplementary Figure S4 shows that 2 out of 3 bats seroconverted 14 days after booster immunisation with NiV F appended with foldon trimerisation domain, and NiV F specific antibodies were detected via ELISA with absorbance readings ranging from 1.2-1.3 absorbance units (AU). Conversely, those from control groups (N = 3) have a lower absorbance (<0.5AU at the highest concentration) (Supplementary Figure S4). Importantly, this indicates that the selected goat anti-bat secondary antibody could detect IgG from captive Jamaican fruit bats (Artibeus jamaicensis) and also ReBAs, which incorporates IgG constant regions of the black flying fox.

Next, we examined the binding kinetics of the henipavirus F-specific antibody, h5B3 and its bat derivatives that were purified with pA column. The pA purified b5B3Hsp and b5B3Bsp had apparent affinities of 0.55 and 0.48nM, respectively, which is within the nanomolar range of h5B3 with an apparent affinity (K_D) of 0.24nM (Figures 3A, B). This analysis demonstrated that the apparent affinities of h5B3, b5B3Hsp and b5B3Bsp were comparable (Figure 3B). Interestingly, maximum specific binding (B_max) values of h5B3 was higher at 2.98AU, while the absorbance units of b5B3 were similar across Hsp and Bsp (1.82 – 2.04AU) (Figure 3B). pG purified b5B3 also exhibits a similar magnitude of binding (B_max) but binding affinities were approximately two-fold lower compared to its pA purified human counterpart (Supplementary Figure S5). As negative control, 5B3 ReBAs were used against HA antigens, and C05 ReBAs were tested on NiV F antigens to determine assay background levels. A trimerisation domain specific antibody, anti-Clamp1 (HIV1281), was used throughout as a positive control (64). Overall, this shows that h5B3 and its bat derivative generally shares the same binding profile but subtle differences were observed in magnitude of binding (B_max).

As a comparison, we also examined influenza HA specific hC05 and its bat derivatives against influenza (A/Brisbane/59/2007) HA proteins (Supplementary Figures S6A, B). The apparent K_D values for pA purified hC05 was 0.17nM, while K_D values for bC05Hsp and bC05Bsp were 0.25 and 0.39nM respectively (Supplementary Figure S6B). Interestingly, the binding affinity of hC05 to HA antigen was similar to bC05Hsp but is approximately two-fold higher than bC05Bsp. B_max values of hC05 pA was highest at 3.94AU, while readings for bC05Hsp and bC05Bsp were analogous at 3.15AU and 3.12AU (Supplementary Figure S6B). Intriguingly, these results suggest that h5B3 and hC05 may work marginally better than their bat derivatives. However, the comparison between human IgG and ReBAs should be interpreted cautiously due to the different secondary antibodies. Overall, this result confirms that ReBAs containing paratope retained their antigen-specific binding capacity.

To further investigate the neutralising capabilities of ReBAs, we used a NiV pseudovirus (NiV-pps) assay (58, 61). As illustrated in Figures 4A, B, b5B3Hsp and b5B3Bsp could neutralise NiV-pps well at a half-maximal inhibitory concentration (IC₅₀) of 0.046nM and 0.094nM, while h5B3 has an IC₅₀ of 0.064nM. Here, C05 on the bat and human constant regions were included as a negative control; it had minimal activity against NiV-pps, reinforcing that ReBAs’ binding activity is antigen-specific and depends on its paratope.