Potent Human Single-Domain Antibodies Specific for a Novel Prefusion Epitope of Respiratory Syncytial Virus F Glycoprotein

ABSTRACT

Respiratory syncytial virus (RSV) poses great health threats to humans. However, there are no licensed vaccines or therapeutic drugs to date. Only one humanized monoclonal antibody, palivizumab, is available on the market, but it is used prophylactically and is limited to infants with high risk. With advances in antibody engineering, it has been found that a single-domain antibody (sdAb) can be therapeutically administered by inhalation, which would be more efficient for respiratory diseases. Here, we identified two human sdAbs, m17 and m35, by phage display technology. They specifically bind to RSV fusion glycoprotein (F protein) in the prefusion state with subnanomolar affinity and potently neutralize both RSV subtypes A and B with 50% inhibitory concentration (IC₅₀) values ranging from pM to nM. Interestingly, these sdAbs recognize a novel epitope, termed VI, that is unique to the prefusion state. This epitope is located at the C terminus of the F1 subunit, close to the viral membrane, and might be sterically restricted. We further find that m17 and m35 neutralize RSV by preventing the prefusion F conformational arrangement, thus inhibiting membrane fusion. These two sdAbs have the potential to be further developed as therapeutic candidates and may also provide novel insight for developing other antiviral reagents against RSV.

IMPORTANCE Because respiratory syncytial virus (RSV) can cause serious respiratory disease in immunodeficient groups, including infants and seniors, the development of vaccines and therapeutic drugs, such as neutralizing antibodies, is urgently needed. Compared to the conventional full-length antibody, a single-domain antibody (sdAb) has been demonstrated to be efficient for respiratory diseases when administered by inhalation, thereby potentially introducing a kind of novel therapeutic agent in the market. Here, we discovered two potent neutralizing human sdAbs against RSV that recognized a novel prefusion epitope, termed VI, and prevented conformational arrangement during the fusion process. Our work provides not only therapeutic candidates but also novel targets for new drug and vaccine development.

INTRODUCTION

Respiratory syncytial virus (RSV) is the leading pathogen of severe pneumonia in children requiring hospital admission (1). Nearly all children are infected at least once before they are 2 years old and are reinfected throughout life due to immunity waning over time (2). Globally, it is estimated that 33.1 million episodes of acute lower respiratory infection (ALRI) are caused by RSV, resulting in approximately 120,000 deaths annually in children under 5 years old (3). Although it poses a moderate threat to healthy adults, RSV is also a major pathogen for elderly and high-risk patients with chronic heart or lung disease (4). To date, the humanized monoclonal antibody palivizumab (Synagis) is the only available agent approved by the FDA (5, 6). However, the application of palivizumab is limited to prophylactic use in high-risk infants due to its modest efficacy and high cost (7). Therefore, the development of vaccines and therapeutic agents, such as neutralizing antibodies, is urgently needed.

RSV belongs to the Orthopneumovirus genus of Pneumoviridae and can be divided into subtypes A and B (8). The genome encodes 11 proteins, including the attachment glycoprotein (G protein) and the fusion glycoprotein (F protein) on the surface of the viral particle (9). The G protein mediates attachment to cells but is not absolutely required for infection (10, 11). The F protein plays a crucial role in binding to the cell receptor and is the only protein responsible for membrane fusion, which is necessary for viral entry and spread by forming syncytia (12, 13). In addition, the F protein is more conserved (∼90%) than the G protein within subtypes (14, 15). Therefore, it has been considered the most ideal target for developing vaccines and drugs.

The RSV F protein is a class I viral fusion protein that natively exists as a trimer (16, 17). Each monomer is initially synthesized as an inactive precursor. After cleavage by furin-like enzyme protease, it is activated and separated into the N-terminal F2 subunit and the C-terminal F1 subunit linked by a pair of disulfides (18). Subsequently, three monomers self-associate to form mature, trimeric prefusion F. Once triggered, the fusion peptide at the N terminus of F1 unfolds from the interior and inserts into the adjacent cell membrane. Moreover, the C terminus of F1 anchoring on the viral envelope refolds oppositely to form a six-helix bundle by combination with the N terminus, leading to membrane fusion. The conformation of prefusion F undergoes rearrangement to form a stable postfusion status (19). It has been demonstrated that prefusion F is responsible for the majority of RSV-neutralizing antibodies in convalescent-phase serum rather than postfusion (20). However, because recombinant wild-type F is inherently metastable, it has been stabilized in the prefusion state by introducing fibritin trimerization, intrachain disulfide and hydrophobic substitutions, termed DScav1 (21). This work accelerates the development of anti-RSV agents. Recently, the stability and immunogenicity of DScav1 have been further optimized to generate a series of prefusion F variations, such as DS2 (22).

To date, various antibodies against RSV F have been isolated and characterized. All these antigenic epitopes have also been defined at six different regions as follows: Ø, I, II, III, IV, and V regardless of quaternary epitopes spanning adjacent monomers (23). Generally, antibodies preferentially recognizing the prefusion epitopes have more potent neutralization activities than those targeting the shared epitopes in both pre- and postfusion F, while antibodies that only bind to the postfusion epitopes have weak or no neutralization (24).

Different from conventional full-length antibodies, single-domain antibodies (sdAbs), such as VHH derived from Camelidae (also termed nanobodies), contain only the antibody heavy chain variable domain. It is the smallest natural antigen-binding domain with a molecular weight of approximately 15 kDa, which is 1/10 that of full-length antibodies (e.g., immunoglobulin G [IgG]) (25). Although the lack of Fc fragment leads to abortion of Fc-mediated effector functions, sdAbs could directly neutralize virus by interfering with the viral entry process, and the small size allows it to bind to the narrow epitope that is usually inaccessible for full-length antibodies. In addition, sdAbs are highly stable for nebulization, which is suitable for respiratory diseases. An example sdAb is ALX-0171, which is a clinically evaluated nanobody for the treatment of RSV infection (26).

Here, we discovered two human sdAbs from a large phage display library based on the human antibody heavy chain variable domain (VH). These two sdAbs show high binding activity to prefusion F of RSV in the subnanomolar range but not to postfusion F and potently neutralize both RSV subtypes A and B. Interestingly, these sdAbs recognize a novel prefusion epitope VI at the bottom of prefusion F near the viral membrane, which is preferentially accessible to molecules with small size, such as sdAbs, due to steric hindrance.

RESULTS

Isolation and characterization of antibodies specific for prefusion F.

After four rounds of panning, the results of the polyclonal phage enzyme-linked immunosorbent assay (ELISA) showed obvious enrichment without nonspecificity. Then, the single clone specific for prefusion F was identified according to the monoclonal phage ELISA and sequencing results. A total of 13 different sdAb clones with positive binding signals were expressed, purified, and characterized by binding and neutralization assays at a concentration of 3 μM (Fig. 1A). The clones that showed poor binding to DS2 were excluded for further evaluation even though they seemed to neutralize RSV. Among the remaining four clones, H3 and G4 showed both high binding to DS2 and neutralization against RSV, whereas A10 and D9 only showed high binding. H3 and G4 inhibited RSV A2 infection with 50% inhibitory concentration (IC₅₀) values of approximately 21.9 μg/ml and 12.1 μg/ml, respectively (Fig. 1B).

FIG 1.

Evaluation of monoclones after panning. (A) Binding and neutralization assays of candidate sdAbs. The binding of the candidate clones to DS2 (red) and post-F (blue) are shown according to the left axis. The inhibition of the candidate clones against RSV A2 infection (green) is shown according to the right axis. scFv-mpe8 was used as a control. (B) H3 and G4 were measured for inhibiting RSV A2 infection with 2-fold dilution from 40 μg/ml. scFv-mpe8 was used as a control. Data are depicted as the mean ± standard deviation (SD) and are fitted by logistics.

Different binding to native F protein and its variants.

To explore why only H3 and G4 neutralized RSV among the four candidates (A10, D9, H3, and G4), we performed immunofluorescence (IF) staining to confirm whether these sdAbs could bind to native F protein on the membrane. The staining fluorescence of H3 and G4 was clear, indicating the recognition of native F protein, whereas no detectable fluorescence was observed in the case of A10 and D9 (Fig. 2A). In addition, the binding of H3 and G4 to engineered native-like F protein variants was also measured by ELISA. H3 and G4 bound to both the prefusion F variants, DS2 and DScav1, with different activities. In contrast, they did not bind to the postfusion F (Fig. 2B).

FIG 2.

Binding of the candidate antibodies to different F proteins. (A) Binding of sdAbs A10, D9, H3, and G4 to native F protein was detected by IF with green fluorescence. scFv-mpe8 was used as a control. (B) Binding of H3 and G4 to DS2, DScav1, and post-F was measured by ELISA. scFv-mpe8 was used as a control. (C) Binding of H3-Fc and G4-Fc to native F protein (subtype A2) was detected by IF with green fluorescence. IgG-mpe8 was used as a control. (D) Binding of H3-Fc and G4-Fc to DS2, DScav1, and post-F was measured by ELISA. IgG-mpe8 was used as a control. Data are depicted as the mean ± SD and are fitted by logistics.

To improve activities by increasing valency, H3 and G4 were fused to IgG1 Fc fragments. However, the binding of the fusion proteins H3-Fc and G4-Fc to membrane-associated native F protein dramatically decreased compared to that of H3 and G4 (Fig. 2C), but they could still bind to DS2 and DScav1, as measured by ELISA (Fig. 2D). This result showed that the binding epitopes recognized by H3 and G4 on membrane-associated native F protein might be more accessible for molecules with a smaller size.

Improved binding activities on both prefusion variants after affinity maturation.

Affinity maturation was performed by using DS2 as the antigen in the first two rounds and DScav1 in the next two rounds. A large number of candidate sdAb clones with improved binding were isolated and expressed. Finally, m17 from the H3 mutant library and m35 from the G4 mutant library were selected for further evaluation. The 50% effective concentration (EC₅₀) values of their binding to DS2 and DScav1 were 2.00 nM and 1.79 nM for m17 (Fig. 3A) and 1.53 nM and 2.35 nM for m35 (Fig. 3B), respectively. No obvious binding of these two clones to postfusion F was detected. Therefore, the specific binding activities increased significantly after affinity maturation.

FIG 3.

Binding of m17 and m35 to the F proteins. (A, B) Binding of m17 and m35 to DS2, DScav1, and post-F measured by ELISA. Data are depicted as the mean ± SD and are fitted by logistics. (C, D) The affinities of m17 and m35 to DS2 measured by BLI. Association and dissociation are divided by the black dashed line. Raw data are shown as full curves and then fitted as dashed curves.

The affinities of m17 and m35 to DS2 were measured by biolayer interferometry (BLI) assay. The K_D of m17 to DS2 was 0.544 nM, with a K_on of 4.93 × 10⁵ (1/ms) and a K_off of 2.68 × 10⁻⁴ (1/s) (Fig. 3C), and the K_D of m35 to DS2 was 0.386 nM, with a K_on of 5.86 × 10⁵ (1/ms) and a K_off of 2.26 × 10⁻⁴ (1/s) (Fig. 3D).

sdAbs potently neutralize both RSV subtypes.

After affinity maturation, the neutralization activities of m17 and m35 were also improved. m17 and m35 inhibited RSV A2 infection with IC₅₀ values of approximately 0.67 μg/ml and 0.70 μg/ml, respectively (Fig. 4A), which increased approximately 33-fold compared with H3 and 17-fold compared with G4. In another assay, m17 and m35 potently inhibited RSV A2 spread with IC₅₀ values of 0.017 μg/ml and 0.0095 μg/ml, respectively (Fig. 4B). Similar experiments were also performed with RSV B, which showed inhibition of virus infection with IC₅₀ values of approximately 1.0 μg/ml and 0.84 μg/ml for m17 and m35, respectively, and inhibition of virus spread with IC₅₀ values of approximately 0.0043 μg/ml and 0.0018 μg/ml for m17 and m35, respectively (Fig. 4C and D). Therefore, the antigenic epitopes of m17 and m35 are conserved within both subtypes A and B. In all neutralization assays, the sdAbs showed much stronger antiviral potency than palivizumab with a maximum increase of 700-fold.

FIG 4.

Neutralization of RSV by m17 and m35. (A) Inhibition of infection (RSV A2). (B) Inhibition of spread (RSV A2). (C) Inhibition of infection (RSV B). (D) Inhibition of spread (RSV B). Palivizumab and IgG-D25 were used as controls. Data are depicted as the mean ± SD and are fitted by logistics.

sdAbs target to a novel prefusion epitope.

Due to the trimeric conformation, there are quaternary epitopes spanning adjacent protomers of RSV F, which have been validated by the antibodies AM14 and VHH-4 (27, 28). Therefore, monomeric prefusion F (DS2-mono) was prepared without the T4 trimer motif and used for the binding test by ELISA (27). There was no obvious difference in binding between monomeric F (DS2-mono) and trimeric F (DS2) for either m17 or m35 (Fig. 5A). Therefore, the majority of their epitopes might focus on a single protomer. Next, the epitope range was narrowed by competitive ELISA using a panel of control antibodies targeting the well-defined epitopes in F protein (Fig. 5B and C). As mentioned above, m17 and m35 specifically recognized prefusion F but not postfusion F. However, no competition was observed with D25 or hRSV90, which target two prefusion-specific epitopes Ø and V as previously reported (23, 24). In contrast, both m17 and m35 could partially compete with 101F, which targets epitope IV, an epitope shared by both pre- and postfusion F (29). Thus, the epitopes recognized by m17 and m35 were different from epitope IV, and the competition might be caused by steric hindrance rather than overlapping. In addition, there was no obvious competition in the case of palivizumab recognizing epitope II and IgG-mpe8 recognizing epitope III. Together, these results suggested that m17 and m35 similarly recognized a new epitope, which is located near epitope IV and is specific for prefusion.

FIG 5.

Epitope mapping. (A) Antibodies binding to monomeric and trimeric DS2. scFv-mpe8 was used as a control. (B, C) Competitive ELISA for m17 and m35, respectively. The values are normalized as competitive rates related to the negative control. (D, E) Docking complexes of m17 and m35 with F protein, respectively. The two protomers of prefusion F trimer (PDB: 5K6I) are shown in surface representation with light and dark gray. The remaining one protomer is shown in ribbon representation, with various colors indicating different epitopes. The models of m17 and m35 docking to the protomer are also shown in ribbon representation and are colored to indicate the framework (yellow) and CDRs (hot pink).

To gain more detailed insight into this epitope in the context of the antibody/antigen complex, homology modeling was performed by SWISS-MODEL (30–34). The first one ordered by global model quality estimate (GMQE) was picked and docked to the crystal structure of the DS2 variant (PDB: 5K6I) (22, 35). Using the results from binding and competition ELISA as a reference, the structures of the complexes were constructed and presented here (Fig. 5D and E). m17 and m35 bind to an undefined epitope that is mostly composed of helixes α9 and α10 at the C terminus of F1. They overlap partially with epitopes I and IV, leading to competition with 101F due to steric hindrance. According to the order of epitopes that have been defined to date, the new epitope could be termed VI (23). Interestingly, because the interface is narrowed by the viral envelope together with the bottom of prefusion F, this epitope on the native virion seems hardly accessible for large molecules, such as H3-Fc and G4-Fc (Fig. 2C). Hence, these findings demonstrated the advantage of the smaller size of these two sdAbs.

According to the complexes analyzed by PDBePISA, several potential residues in the binding surface were identified and mutated to alanine to confirm the docking complexes (Fig. 6A; see also Tables S1 and S2 in the supplemental material) (36). Three DS2 mutants containing one mutation (F477A, N496A, and N500A) were constructed, while the other three mutants containing two consecutive mutations (F477A/Y478A, I492A/S493A, and I499A/N500A) were also prepared in case the single mutation had no obvious effect. The conformations of all six DS2 mutants were first validated by D25, which is specific for prefusion, and showed no obvious change in binding compared with DS2 (Fig. 6B). As expected, most mutations affected the binding to m17 and m35. There were also some differences; for example, residues N496A, I499A, and N500A on the middle of α10 are more crucial for m35 than m17, possibly because of differences in complementarity-determining region 3 (CDR3) (Fig. 6C and D). Taken together, these findings provided direct evidence supporting the identification of epitope VI.

FIG 6.

Epitope mapping by alanine mutants. (A) The mutated residues are labeled with numbers in one protomer of prefusion F (PDB: 5K6I) shown in ribbon representation. (B to D) Binding of D25, m17, and m35 to different DS2 mutants, as measured by ELISA. Data are depicted as the mean ± SD and are fitted by logistics.

sdAbs neutralize RSV by preventing rearrangement of the F1 C terminus.

During rearrangement, the C-terminal helixes α9, α10, and strand β22 pivot inversely to form an extended helix then associate with the N terminus to bring membranes together and promote fusion (Fig. 7A). The tertiary structure will be dramatically changed after fusion, which causes some epitopes, such as VI in prefusion F, to fail to be recognized again in postfusion. This phenomenon is similar to the reported epitopes Ø and V at the N terminus. Thus, we hypothesized that bound m17 or m35 immobilized prefusion F as an anchor at the C terminus of F1 and then prevented conformational rearrangement during the fusion process. We tested this hypothesis by fusion inhibition assay. When the antibody concentration was serially increased, the fluorescence spots staining the RSV-induced syncytia decreased, indicating inhibition of fusion (Fig. 7B). The IC₅₀ values were approximately 0.0029 μg/ml for m17 and 0.0017 μg/ml for m35 (Fig. 7C).

FIG 7.

Fusion inhibition. (A) Superimposition of one protomer between prefusion (PDB: 5K6I) and postfusion F (PDB: 3RRR) shown in ribbon representation. The residues unique to the prefusion are colored green, and the postfusion residues are colored orange. The residues shared by both pre- and postfusion F are colored gray. The dashed rings indicate the rearrangement of conformation during fusion. The identified epitope VI is boxed (red line), which undergoes dramatic conformational change to form an extended helix (orange). (B) Fusion inhibition recorded by a high content analysis system (HCA). RSV was stained by IF (green), and cell nuclei were stained with 4′,6-diamidino-2-phenylindol (DAPI) (blue). NC indicates the negative control without RSV infection. (C) Fusion inhibition rates measured by HCA. Data are depicted as the mean ± SD and are fitted by logistics.

DISCUSSION

With the progress of biotechnology, numerous novel vaccines and neutralizing antibodies against RSV have been developed in clinical trials (37, 38). However, most of these antibodies are conventional full-length IgGs and are administered by intramuscular injection for prophylactics, such as motavizumab, suptavumab, MK-1654, and the optimized D25 (nirsevimab) that has been proven to be highly efficacious in protecting infants from RSV infection recently (39–42). The trivalent VHH, ALX-0171, which has three identical sdAbs linked by flexible linkers to improve activity, has displayed several advantages, including low cost, multivalence, and inhalation for treatment (43). Although no related adverse event has been detected, ALX-0171 treatment induced antidrug antibodies in patients (https://www.ablynx.com/uploads/data/files/ablynx_alx-0171_first-in-infant%20study%20results_webcast%20presentation.pdf) (44). Here, we identified two human sdAbs, which may have lower immunogenicity.

It has been reported that antibodies specific for the prefusion F conformation, such as epitopes Ø and V, are more potent in neutralizing RSV than those targeting shared epitopes (24, 45). We first panned the library with prefusion F DS2, an engineered native-like F variant, using postfusion F for depletion, which resulted in two candidate clones H3 and G4 (22). However, these clones showed relatively weak binding to DScav1, which is another engineered native-like F variant (21). Therefore, DScav1 was used as the second antigen for sequential panning during affinity maturation, which could avoid the bias in selection of candidate clones that preferentially recognize the artificial “epitope” caused by introduction of mutations. As expected, the binding activities toward each variant were increased to a similar level. Both m17 and m35 efficiently neutralized RSV subtypes A and B in different inhibition assays, indicating that they are ideal for further development as clinical candidates.

m17 and m35 have a common CDR1, CDR2, and framework before affinity maturation, but their CDR3s are different. We performed competitive ELISA, molecular docking, and alanine scanning to determine the binding sites and found that they similarly recognized an unidentified prefusion-specific epitope, termed VI here. This epitope is localized at the C terminus of F1, which undergoes dramatic rearrangement during the fusion process according to the resolved structures (19). The bound sdAbs could immobilize the prefusion conformation to potently inhibit membrane fusion and infection. Hence, epitope VI is a promising target for drug development. Intriguingly, epitope VI is located at the bottom of the F protein very close to the viral membrane, which may result in steric restriction to large molecules, and it prefers molecules with smaller sizes, such as sdAbs m17 and m35, thereby explaining why it was difficult to be discovered before. More importantly, the other two prefusion-specific epitopes, Ø and V, are located at the apex of the F protein and are responsible for the major neutralizing antibodies in vivo, which may lead to higher immune pressure and lower conservation (16, 46, 47). In comparison, epitope VI is relatively conserved within subtypes due to lower exposure to the immune system (Fig. 8) (48, 49). In addition, the bound antibody of epitope VI would not block epitopes Ø and V to induce an immune response against RSV infection, which is crucial for immature infants (50). These findings will also provide useful information on vaccine design.

FIG 8.

Amino acid variation of RSV F from 472 to 513. All 1,182 complete amino acid (AA) sequences of RSV F were obtained by searching “human respiratory syncytial virus” and “fusion glycoprotein” from UniProt. After alignment to the A2 strain, the AA variation was calculated.

In summary, we discovered two human sdAbs against RSV. These sdAbs recognize a novel prefusion-specific epitope VI located at the C terminus of F1. This study provides novel insight into treating RSV infection in the future.

MATERIALS AND METHODS

Cells and virus.

Hep2 cells (GDC004, CCTCC) were cultured in Dulbecco’s modified Eagle medium (DMEM; Gibco) supplemented with 10% fetal bovine serum (FBS; Gibco) and penicillin and streptomycin (Gibco) at 37°C with 5% carbon dioxide (CO₂). The 293F cell line (Thermo) was cultured in FreeStyle 293 medium (Gibco) as a suspension at 37°C with 8% CO₂ in an orbital shaker. RSV subtypes A (A2; National Virus Resource Center [NVRC]) and B [CH93(18)-18; ZeptoMetrix] were propagated and quantified in Hep2 cells by a modified plaque assay. In brief, Hep2 cells were seeded in 96-well culture plates and cultured overnight until confluent. RSV was diluted in DMEM and added to cells for infection. After 1 h, cells were overlaid with 0.75% methylcellulose dissolved in DMEM containing 2% FBS and incubated for 72 h. Cells were then fixed and stained with the IgG-mpe8 monoclonal antibody (see below) specific for RSV F followed by fluorescein isothiocyanate (FITC)-anti-human Fc (Abcam). Plaques were visualized and counted using fluorescence microscopy.

Construction, expression, and purification of RSV (strain A2) F proteins.

The DNA encoding postfusion F (post-F) was synthesized and subcloned into pSectag2A (Invitrogen) according to a previous publication (51). The plasmids encoding prefusion F were constructed by PCR based on the post-F. The first prefusion F DScav1 was consistent with the previous publication (21). The second prefusion F was consistent with one clone of the optimized DS2 immunogens, sc9-10-A149C-Y458C-S46G-K465Q-S215P-E92D-N67I, which was termed DS2 in this study (22). The monomeric version of prefusion F was constructed by deleting the T4 trimerization motif of DS2 (27). To produce proteins, these plasmids were transiently transfected into 293F cells with polyethylenimine (Polysciences). Approximately 5 days later, the supernatant was harvested and purified with a 6×His tag according to the instructions of Ni Sepharose Excel (GE Healthcare). The purification tags of DS2 and DScav1 were then removed by thrombin (Sigma) digestion and size exclusion chromatography.

Construction, expression, and purification of RSV F positive-control antibodies.

According to previous publications, DNA encoding the control antibodies (D25, palivizumab, mpe8, 101F, and hRSV90) was synthesized and cloned into pVitro2-neo-mcs (Invitrogen) (5, 16, 52–54). These antibodies were expressed transiently in 293F cells as described above and purified by protein A resin (GE Healthcare) according to the manufacturer’s instructions. In addition, a single-chain variable fragment (scFv) format of mpe8 (scFv-mpe8) with His and Flag tags was cloned to a modified pComb3XSS vector for expression (55).

Panning and identification of sdAbs specific for prefusion F.

The human antibody VH library was constructed as previously reported (55, 56). The DS2 antigen was then labeled with sulfo-NHS-LC-biotin (Pierce) and coupled to streptavidin magnetic beads (Invitrogen) for the first two rounds of panning, according to a previously reported protocol (57). In the remaining two rounds, nonlabeled DS2 antigen was coated on plates at 10 μg/ml. Panning was performed using postfusion F for depletion according to another protocol (58). After four rounds, polyclonal phage ELISA and monoclonal phage ELISA were performed to evaluate the panning results and identify positive candidate clones. The enriched positive clones were further sequenced for analysis.

Expression and purification of sdAbs.

The plasmids of positive sdAb clones were transformed into Escherichia coli HB2151 for expression. After induction by isopropyl-β-d-thiogalactopyranoside (IPTG) at 37°C overnight, the bacterial cells were harvested and lysed by polymyxin B (Sangon) in phosphate-buffered saline (PBS). The clarified supernatant was then collected by centrifugation and further purified by nickel-nitrilotriacetic acid (Ni-NTA) agarose (Qiagen) according to the instructions. Additionally, H3 and G4 were fused with the human IgG1 Fc fragment by PCR. The recombinant DNA was cloned into the pcDNA3.1(+) vector (Invitrogen) for expression in 293F cells. The methods of transfection and purification were the same as the preparation of control antibodies described above.

ELISA.

Half-area plates were coated with 4 μg/ml of RSV F protein at 4°C overnight and blocked with 3% milk in PBS for approximately 1 h at 37°C. Subsequently, serially diluted antibodies were added and incubated for 1.5 h at 37°C. The horseradish peroxidase (HRP)-anti-Flag monoclonal antibody (Sigma) or HRP-anti-human Fc monoclonal antibody (Sigma) was used as the secondary antibody. Color was developed with 2,2′-azinobis(3-ethylbenzthiazolinesulfonic acid) (ABTS) substrate (Life Technologies), and absorbance was measured at 405 nm.

RSV neutralization assay.

Neutralization was performed using a modified microneutralization assay and divided into inhibiting infection and inhibiting spread according to the time of antibody incubation (43). On the day before the assay, Hep2 cells were seeded and cultured in 96-well plates at a subconfluent density. To inhibit infection, the diluted antibodies were mixed with 100 PFU of RSV in a total volume of 50 μl and incubated at 37°C for 1 h. Next, the mixtures were used to infect Hep2 cells in each well for approximately 1 h at 37°C. Cells were then washed with PBS, and the culture medium was changed to DMEM containing 2% FBS. To inhibit spread, Hep2 cells were first infected with 100 PFU of RSV for approximately 2 to 3 h. The antibodies were diluted in DMEM containing 2% FBS at different concentrations. After infection, cells were washed with PBS and overlaid with the prepared culture medium containing antibodies. All infected cells were cultured continuously for 3 to 4 days at 37°C. Finally, cells were fixed, and the virus was measured with a method similar to the ELISA described above. The IgG-mpe8 monoclonal antibody was used as the primary antibody, and HRP-anti-human Fc (Sigma) was used as the secondary antibody.

Indirect immunofluorescence assay.

Hep2 cells were seeded in 96-well culture plates and cultured overnight until reaching confluence. Cells were then infected with RSV A2. When the cytopathic effect was clear approximately 2 days after infection, cells were fixed and incubated with the candidate antibodies at 1 μM to bind the native RSV F protein on the membrane. sdAbs, including H3, G4, A10, and D9 as well as scFv-mpe8, were stained with FITC-anti-Flag (Abcam) as the secondary antibody. H3-Fc, G4-Fc, and IgG-mpe8 were stained with FITC-anti-human Fc (Abcam) as the secondary antibody.

Affinity maturation.

Affinity maturation was performed by random mutagenesis on the candidate sdAbs with the phage display technique (59). Error-prone PCR was performed with the GeneMorph II random mutagenesis kit (Agilent) to introduce random mutations. DS2 was used as the antigen for the first two rounds of panning, and DScav1 was used for the last two rounds. All rounds were performed on plates and depleted with bovine serum albumin (BSA) as described above. The clones from the fourth round were identified by ELISA using expressed supernatant. Briefly, TG1 monoclones were seeded in 96-well plates separately. After culture and induction by IPTG overnight, the supernatant containing leaky expression was collected and diluted with milk as the primary antibody. The remaining steps were the same as the ELISA above.

BLI assay.

To determine binding kinetics, a BLI assay was performed on an Octet QK system (ForteBio) (60). The DS2 antigen was labeled with sulfo-NHS-LC-biotin (Pierce) for immobilization and diluted at 40 μg/ml in PBS containing 0.01% Tween 20 and 0.01% BSA. The antibodies were diluted serially from 200 nM to 3.125 nM in the same buffer. Data were fitted and analyzed by data analysis software.

Mapping the epitope by competitive ELISA.

Plates were coated with DS2 antigen at 4 μg/ml and blocked with milk. The control antibodies targeting different epitopes were diluted serially and mixed with an equal volume of m17 or m35 at 1.5 nM. The mixtures were used as the primary antibodies, and the HRP-anti-Flag monoclonal antibody (Sigma) was used as the secondary antibody.

Molecular modeling and docking.

SWISS-MODEL was used for homology modeling according to the primary amino acid sequences of m17 and m35 (30–34). After evaluation, the best models ordered by GQME were selected for docking. Docking was performed on the ZDOCK server with a combination of the crystal structure of the DS2 variant (PDB: 5K6I) (22, 35). The top 10 predictions were downloaded, and the complexes were selected after comparison with each other. Finally, the interfaces were analyzed with PDBePISA and PyMOL (36).

Alanine scanning.

DS2 mutants, including F477A, F477A/Y478A, I492A/S493A, N496A, I499A/N500A, and N500A, for mapping the epitope were constructed mainly according to the predicted interface between m35 and DS2. These mutants were then expressed and purified as described above. Finally, they were coated on plates and tested with IgG-D25, m17, and m35 by ELISA.

Fusion inhibition assay.

The fusion inhibition assay was performed according to a general protocol (53). Hep2 cells were seeded in 96-well culture plates and cultured overnight until reaching confluence. Plates were first precooled for 1 h at 4°C. After washing with chilled PBS, cells were incubated with 100 PFU of RSV A2 at 4°C for attachment, avoiding fusion. After 1 h, unbound viruses were washed off with chilled PBS, and the antibodies were then serially diluted in DMEM containing 2% FBS and added to each well. Plates were continuously incubated for 1 h at 4°C, which allowed the antibodies to bind to the virus prior to fusion. Cells were cultured at 37°C for 72 h. Finally, virus was detected by IF as described above. Images were acquired using an Operetta CLS (PerkinElmer), and the sum of fluorescence intensity was measured simultaneously.