Antiviral Activity of a Llama-Derived Single-Domain Antibody against Enterovirus A71

In the past few decades, enterovirus A71 (EVA71) has caused devastating outbreaks in the Asia-Pacific region, resulting in serious sequelae in infected young children. No preventive or therapeutic interventions are currently available for curing EVA71 infection, highlighting a great unmet medical need for this disease. Here, we showed that one novel single-domain antibody (sdAb), F1, isolated from an immunized llama, could alleviate EVA71 infection both in vitro and in vivo.

ABSTRACT

In the past few decades, enterovirus A71 (EVA71) has caused devastating outbreaks in the Asia-Pacific region, resulting in serious sequelae in infected young children. No preventive or therapeutic interventions are currently available for curing EVA71 infection, highlighting a great unmet medical need for this disease. Here, we showed that one novel single-domain antibody (sdAb), F1, isolated from an immunized llama, could alleviate EVA71 infection both in vitro and in vivo. We also confirmed that the sdAb clone F1 recognizes EVA71 through a novel conformational epitope comprising the highly conserved region of VP3 capsid protein by using competitive-binding and overlapping-peptide enzyme-linked immunosorbent assays (ELISAs). Because of the virion’s icosahedral structure, we reasoned that adjacent epitopes must be clustered within molecular ranges that may be simultaneously bound by an engineered antibody with multiple valency. Therefore, two single-domain binding modules (F1) were fused to generate an sdAb-in-tandem design so that the capture of viral antigens could be further increased by valency effects. We showed that the tetravalent construct F1×F1-hFc, containing two sdAb-in-tandem on a fragment crystallizable (Fc) scaffold, exhibits more potent neutralization activity against EVA71 than does the bivalent sdAb F1-hFc by at least 5.8-fold. We also demonstrated that, using a human scavenger receptor class B member 2 (hSCARB2) transgenic mouse model, a half dose of the F1×F1-hFc provided better protection against EVA71 infection than did the F1-hFc. Thus, our study furnishes important insights into multivalent sdAb engineering against viral infection and provides a novel strategic deployment approach for preparedness of emerging infectious diseases such as EVA71.

INTRODUCTION

Enterovirus A71 (EVA71) is a nonenveloped RNA virus that belongs to the Picornaviridae family. It is a major neurotropic pathogen that causes viral encephalitis and severe hand, foot, and mouth disease (HFMD) in young children worldwide. In 1998, EVA71 caused a large outbreak in Taiwan, with 405 severe cases in children, of which 78 were fatal (1, 2). A recent outbreak of severe EVA71-associated HFMD in Vietnam in 2018 also highlights the unmet need for infection control of the reemergence infection of EVA71 (3). Nevertheless, no effective drug is yet available for treating EVA71 infection.

At present, only two formalin-inactivated EVA71 vaccines, a Vero cell-based inactivated EVA71 C4 genotype vaccine (Sinovac Biotech, Beijing, China) and human diploid-cell line (KMB17 strain) inactivated EVA71 C4 genotype vaccine (Chinese Academy of Medical Sciences [CAMS]), have been approved by the National Medical Products Administration (the former China Food and Drug Administration) in China, while some vaccine candidates are in late-stage clinical trials in different countries (4–7). However, the currently approved vaccines may not function against most serotypes of HFMD-associated enteroviruses due to strain specificity-related issues (8), and none of them have been implemented outside China. To control the spread of EVA71, more antiviral vaccines or drugs will be necessary.

Since developing vaccines for new pathogens is difficult and time-consuming, recent rapid advances in monoclonal antibody (MAb) development could impact emerging infectious disease (EID) control (9). Thus far, more than 84 antibody-based products have been approved by the U.S. Food and Drug Administration (FDA) (10). Evidently, MAbs have also become a crucial tool for preparedness for and response to EID outbreaks (9). A number of MAbs are in the late stage of clinical development, including those for treating serious respiratory syncytial virus (RSV) (MEDI8897) and influenza virus (MEDI8852) infections (11–13). Several murine MAbs with anti-EVA71 activity and different mechanisms of action have been identified, but none have obtained FDA approval thus far (14–18).

The findings regarding antibodies devoid of light chains, also known as heavy-chain-only antibodies, obtained from camel, opened a novel landscape of antibody engineering (19). Single-domain antibodies (sdAbs) engineered from heavy-chain-only antibodies are favorable therapeutic entities because of their unique structural and functional properties (20). Each unique sdAb can be easily engineered to become a multivalent and/or multispecific fusion construct (21). Currently, more than 19 sdAb-based drug candidates targeting antiviral and cancer and autoimmune fields have been evaluated in different stages of clinical trials. Recently, the first llama-derived sdAb was approved to target von Willebrand factor for treatment of blood disorders (22). Several novel sdAbs were shown to be promising candidates for targeting different viruses; one of them, ALX-0171, had been evaluated in phase IIb clinical trials to study its efficacy on lower respiratory tract infection due to RSV (23–28).

In this study, we identified several novel sdAbs targeting EVA71. One of them could protect against EVA71 infection in vivo. Taken together, the identification of sdAb clone-F1 and its progressive refinements, as shown in this study, exhibits a strategic deployment of preventive interventions for preparedness of potential EID outbreaks.

RESULTS

EVA71-specific sdAb screening, identification, and expression.

To construct the sdAb library for phage display screening, one llama was immunized with inactivated EVA71. The serum antibody response to immunization was measured using ELISA (Fig. 1A). The antiserum responses remained similar between blood drawn after the fourth and sixth immunizations. The sdAb library size was estimated to be approximately 8.7 × 10⁸ CFUs on the basis of serial dilution and transformant counts. After one round of panning with inactivated EVA71 antigen (input, 2.0 × 10¹¹ PFU; output, 3.54 × 10⁵ PFU), the binders were enriched, and the percentage of EVA71-positive binders reached ∼66% (Fig. 1B). From a total of 62 positive clones obtained in a phage ELISA, the top 30 clones with high signal-to-background (S/N) ratios were selected for DNA sequencing and subsequent amino acid alignment analysis.

FIG 1.

Immune response against inactivated EVA71 and phage ELISA. (A) The serum titers for the test bleed were much higher than in the preimmune bleed. No significant difference was observed between 1st (after the fourth immunization) and 2nd (after the sixth immunization) bleed samples of different tests. NC, negative control. (B) Phage ELISA was performed to estimate the positive rate of the output phage against the target antigen. A total of 94 phage antibody clones were randomly picked from the 1^st-round panning output.

From these clones, four major lead sdAbs were successfully subcloned, which was followed by expression in CHO cells and purification through protein A chromatography. In these sdAbs, the sequence of complementarity-determining region 3 (CDR3) was more diverse than those of CDR1 or CDR2 (Fig. 2). The molecular weights of respective bivalent sdAb-hFc fusion proteins were in the range of 75 to 110 kDa under nonreducing conditions, according to SDS-PAGE results (Fig. 3A to D). The structural components of bivalent sdAb-hFc fusion proteins are depicted in Fig. 3F.

FIG 2.

Amino acid alignment of EVA71-specific sdAbs. Amino acid sequences of different sdAbs were aligned using the CLC Sequence Viewer (Qiagen). The CDRs were highly variable between each clone and are labeled in yellow. The dots represent conserved amino acids.

FIG 3.

Purification of bivalent and tetravalent sdAb-hFc fusion proteins. Bivalent and tetravalent sdAb-hFc fusion proteins were electrophoresed on SDS-PAGE under nonreducing (dithiothreitol [DTT]−) and reducing (DTT+) conditions. (A) Bivalent sdAb-hFc clone-D1 (D1-hFc). (B) Bivalent sdAb-hFc clone-F1 (F1-hFc). (C) Bivalent sdAb-hFc clone-F3 (F3-hFc). (D) Bivalent sdAb-hFc clone-F4 (F4-hFc). (E) Tetravalent sdAb-hFc clone-F1 (F1×F1-hFc). (F and G) Schematic of bivalent (sdAb-hFc) (F) and tetravalent (sdAb×sdAb-hFc) (G) sdAb-hFc fusion proteins.

Characteristics of anti-EVA71-specific sdAbs.

To examine the binding affinity of each sdAb-hFc fusion protein, we performed ELISA using formalin-inactivated EVA71 immobilized on solid supports. The binding affinity of each sdAb-hFc fusion protein to inactivated EVA71 ranged from 0.1 to 30 nM, as assessed using ELISA (Table 1). Enterovirus D68 (EV-D68), coxsackievirus B3 (CVB3), and influenza A/WSN/33 virus were used to test the specificity of the novel sdAb-hFc fusion proteins. None of the sdAb-hFc fusion proteins exhibited antiviral activities to the aforementioned viruses (Table 1). Of the four virion binders, only F1-hFc was effective in the cell-based EVA71 neutralization assay (50% effective concentration [EC₅₀], ≈406 nM; Table 1).

TABLE 1.

Binding affinity and neutralization activity of different sdAbs for EVA71 and other viruses

sdAb characteristic	Data for anti-EVA71 sdAb fusion protein:
	D1-hFc	F1-hFc	F3-hFc	F4-hFc	F1×F1-hFc
Mol wt of bivalent or tetravalent sdAb-hFc fusion protein (kDa)	∼100	∼95	∼75	∼80 kDa	∼120 kDa
Binding affinity to (nM):
EVA71 TW/73/12	∼0.1	∼46	∼0.3	∼20	∼26
A/WSN/33 influenza virus	>300	>300	>300	>300	>300
Neutralization activity to (EC50 [μg/ml]) (concn [nM]):
EVA71 TW/73/12 (mean ± SD)	>200	38.6 ± 21.8 (∼406.3)	>200	>200	8.4 ± 0.4 (∼70)
EV-D68	>200	>200	>200	>200	>200
CVB3	>200	>200	>200	>200	>200

Furthermore, none of sdAb-Fc fusion proteins could recognize linear epitopes from purified recombinant EVA71 viral capsid proteins coating either capsid proteins or sdAb-hFc fusion proteins via ELISA (Fig. 4A and 4B). MAb979, a commercially available anti-EVA71 antibody, was used as a positive control for recombinant viral VP2 protein (Fig. 4C). Additionally, F1-hFc also failed to react to formalin-inactivated EVA71 by Western blotting (Fig. 4D), indicating a conformational epitope for F1-hFc.

FIG 4.

Analysis of binding of sdAb-hFc to EVA71 capsid proteins by indirect ELISA and immunoblotting. (A) An ELISA was performed by coating wells with recombinant viral capsid proteins or formalin-inactivated EVA71. Various concentrations of target proteins were added in duplicate and were detected with an HRP assay. The amount of bound target proteins is presented as the average ± standard deviation (SD) OD₄₅₀. Anti-EVA71 VP1-specific antibody (D5 IgG) was used as a positive control and E18 IgG can recognize conformational epitopes on the surface of EVA71 virions but not individual recombinant EVA71 capsid proteins by ELISA. (B) An ELISA was performed by coating wells with recombinant sdAb-hFc, EVA71 D5, and E18 IgG. Target protein (2 μg/ml) was added in duplicate, and bound proteins were detected with an HRP assay. The amount of bound target protein is presented as the average ± SD OD₄₅₀. (C) An ELISA was performed by coating wells with recombinant viral proteins. MAB979 antibodies were added in duplicate, and bound antibody was detected with an HRP assay. The amount of bound MAB979 is presented as the average ± SD OD₄₅₀. Anti-EVA71 VP2-specific antibody (MAB979) was used as a positive control for VP2 adsorption. (D) Formalin-inactivated EVA71 was separated by SDS-PAGE and transferred onto a membrane. The membrane was probed with F1-hFc and MAB979 antibody.

Several studies have indicated that viral neutralization by an antibody can be substantially enhanced through the augmentation of binding avidity (29–31). Based on this rationale, a tetravalent sdAb×sdAb-hFc fusion protein was designed (Fig. 3G). In brief, the sdAb×sdAb-hFc fusion protein comprises two tandem sdAbs via a GGGGSGGGS (G₄SG₃S) linker fused to the N terminus of the human IgG1 hinge, followed by an Fc fragment. The molecular weight of the tetravalent sdAb-hFc clone-F1 (F1×F1-hFc) is approximately 120 kDa and exhibits 5.8-fold higher antiviral activity than that with the bivalent F1-hFc (Table 1). Additionally, the tetravalent sdAb-hFc clone-F1 not only could neutralize an EVA71 genotype C strain, but it also could combat an EVA71 genotype B strain in cell-based neutralization assay (Table 2).

TABLE 2.

Antiviral activity of the tetravalent sdAb-hFc construct F1×F1-Fc against various viruses

Virus	Genotype	EC50 in (mean ± SD):
		μg/ml	nM
EVA71 TW/13-50144	B5	7.5 ± 0.3	62.5
EVA71 TW/10-96050	C4	8.2 ± 0.2	68.3
EVA71 TW/07-72043	C5	7.8 ± 0.4	65.0
CVA6		>200	>1,666.6
CVA16		>200	>1,666.6

To evaluate the possible binding conformational epitope of F1×F1-hFc for EVA71, an ELISA-based competitive binding assay was applied using binding epitope-confirmed EVA71-specific antibodies such as D5, E18, and E19 IgGs. D5 IgG mainly recognizes a conserved epitope located at the VP1 GH loop of EVA71 and E18 IgG against EVA71 target to VP4-VP2-VP3-VP1 protomers (15, 16). In contrast to E18 IgG, E19 IgG binds wholly within a single protomer, and its Fab fragment binds primarily to the knob region of EVA71 VP3 (16). Neither E18 IgG nor D5 IgG could be influenced by an F1×F1-hFc competitor in an ELISA-based competitive binding assay, and consequently, these antibodies do not recognize the same epitope (Fig. 5A). Strikingly, we showed that F1×F1-hFc could interfere with E19 IgG, illustrating that F1×F1-hFc might recognize the same epitope or that it confers the conformation change of the E19-binding region so as to weaken the binding of E19 to an EVA71 virion (Fig. 5A). Furthermore, both bivalent F1-hFc and tetravalent F1×F1-hFc could impair the binding of E19 IgG to EVA71 in a dose-dependent manner (Fig. 5B and 5C). Practically, 54% reduction of interaction between EVA71 and E19 IgG, at 0.2 μM, was achieved by a half dose (0.1 μM) of F1xF1-hFc.

FIG 5.

Epitope mapping study by ELISA-based competition binding assay and VP3 overlapping peptides-ELISA. (A) An ELISA-based competitive binding assay was performed by coating wells with formalin-inactivated EVA71. The tested antibodies (1 μM) were competitively bound with or without F1×F1-hFc (0.25 or 0.5 μM), and bound antibodies were detected by an HRP assay. (B) E19 IgG (0.5 μM) was competitively bound with or without F1-hFc at concentrations ranging from 0.25 to 2 μM, and bound E19 IgG was detected by an HRP assay. (C) E19 IgG (0.2 μM) was competitively bound with or without F1×F1-hFc at concentrations ranging from 0.1 to 2 μM, and bound E19 IgG was detected by an HRP assay. (D) Different concentrations of antibodies were incubated with a 20 μg/ml EVA71 VP3 (58 to 74 aa) synthetic peptide-coated plate, and bound antibodies were detected by an HRP assay. Anti-EVA71 VP1 specific antibody (D5 IgG) was used as a negative control. (E) A panel of an EVA71 VP3 peptide library (47 × 15-mer peptides) was used in an overlapping peptide-ELISA for epitope mapping against F1×F1-hFc (200 nM), and bound antibody was detected by an HRP assay.

Based on previous studies, the conformational epitopes of MAb 10D3 and MAb 5H7 were mapped to the highly conserved knob region on EVA71 VP3 capsid protein using escape mutants, including those with the P59L, A62D, E67D, or S74L mutation (14, 32). Therefore, a synthetic peptide, VP3^58–74, harboring amino acids 58 to 74 of the VP3 sequence, was used for binding epitope screening. Compared with E19 and D5 IgG, a higher concentration of F1×F1-hFc was necessary to create a slight reaction with the coated VP3^58–74 synthetic peptide (Fig. 5D), indicating that the conformational epitope of F1×F1-hFc might be located around this region. Additionally, our data not only confirmed that E19 IgG had a limited effect on binding to the linear peptide of knob region of EVA71 VP3 but also reflected the conformationally dependent binding mode of E19 IgG.

Furthermore, an EVA71 VP3 peptide library containing 47 peptides, with 10-mer overlapping between adjacent peptides, was applied for epitope mapping. F1×F1-hFc only reacted with two major groups of peptides with an optical density (OD) of more than 1.5, corresponding to residues 46 to 60 (46-LCQVETILEVNNVPT-60 [VP3^46–60]) and 106 to 130 (106-GYYTQWSGSLEVTFMFTGSFMATGK-130 [VP3^106–130]) within the EVA71 VP3 region (Fig. 5E). Although both of these regions are not located in the central sites of the highly conserved knob region of VP3, they all are 100% identical sequences among EVA71 strains of different genotypes, including (genotype, GenBank accession no.) EVA71 BrCr (A, AAB39968.1), EVA71 Fuyang.Anhui.P.R.C/17.08/2 (B1, ACD63040.1), MS/7423/87 (B2, AAB39969.1), 26M/AUS/4/99/GuaR1 (B3, AFJ40582.1), 5865/sin/000009 (B4, AAK13008.1), NUH0083/SIN/08 (B5, ACK37353.1), C1/FY08-C30-P11 (C1, ADQ54215.1), NUH0075/SIN/08 (C2, ACH87535.1), C3-KOR-EV71-01 (C3, AFN20478.1), SH12-276 (C4, AGK07542.1), and C5-575-TW-07 (C5, AFN20481.1) (Fig. 6). Currently, several anti-EVA71 VP3 antibodies, 10D3 and 5H7, have been identified, but none of their epitopes are the same as that of F1×F1-hFc, indicating that llama-derived sdAb clone F1 might target a new conformational epitope on VP3 (14, 32–34).

FIG 6.

Sequence comparison of VP3 region of different genotype EVA71. VP3 sequences of different genotype EVA71 were aligned using the CLC Sequence Viewer (Qiagen). Dots indicate positions that have a single, fully conserved residue. Strain diversity is labeled in bright green. The discontinuous epitopes of EVA71 VP3 protein recognized by sdAb clone F1 are highlighted in yellow in the aligned sequences. The GenBank accession numbers are AAB39968.1 for EVA71 genotype A, ACD63040.1 for EVA71 genotype B1, AAB39969.1 for EVA71 genotype B2, AFJ40582.1 for EVA71 genotype B3, AAK13008.1 for EVA71 genotype B4, ACK37353.1 for EVA71 genotype B5, ADQ54215.1 for EVA71 genotype C1, ACH87535.1 for genotype C2, AFN20478.1 for EVA71 genotype C3, AGK07542.1 for genotype C4, AFN20481.1 for EVA71 genotype C5.

Taken together, our results demonstrate that the novel EVA71-specific F1-hFc can alleviate EVA71 infection in vitro. The antiviral activity of F1-hFc could be further enhanced using F1×F1-hFc, a multivalent antibody based on the bivalent F1-hFc. Compared with a traditional antibody with heteromeric heavy and light chains, this single gene that encoded the tetravalent sdAb×sdAb-hFc fusion protein can greatly simplify the chemistry, manufacture, and control of the preclinical development of the novel anti-EVA71 tetra-sdAbs.

Prophylactic efficacy of EVA71-specific sdAbs in hSCARB2 transgenic mice.

To evaluate the in vivo neutralizing effect of the bivalent or tetravalent sdAb based on the clone F1, 3-week-old hSCARB2 transgenic mice were challenged with a lethal dose of EVA71 (10⁵ PFU/mouse), followed by intraperitoneal administration of 200 μl of solution (200 μg protein/mouse) premixed with F1-hFc or F1×F1-hFc and EVA71 at 37°C for 1 h. Mice were monitored on a daily basis for the development of clinical signs. On day 9 after infection, the survival rate was 10% in the phosphate-buffered saline (PBS)-treated group, whereas it was 50% and 70% in the bivalent F1-hFc and tetravalent F1×F1-hFc–treated groups, respectively (Fig. 7A). Moreover, the trend of disease scores was consistent with the survival rates in all tested groups (Fig. 7B). Only mice receiving tetravalent F1×F1-hFc treatment, and not those receiving F1-hFc, were free from paralysis.

FIG 7.

Tetravalent F1×F1-hFc is more protective than bivalent F1-hFc in vivo. (A and B) A Kaplan-Meier survival curve (A) and disease scores (B) of 3-week-old hSCARB2 transgenic mice intraperitoneally challenged with sdAbs-pretreated EVA71 (EVA71 TW/73/12) are shown. The total number (n) of mice per group from two independent experiments is shown. A log rank test was used to analyze the statistic difference of survival rate. One-way analysis of variance (ANOVA) with Kruskal-Wallis test was used to analyze the statistical differences of the individual groups with disease score. The symbols *, **, and *** are used to indicate P values of <0.05, <0.01, and <0.001, respectively.

To further investigate the prophylactic efficacy, hSCARB2 transgenic mice were administered with bivalent F1-hFc or tetravalent F1×F1-hFc 1 day before lethal-dose EVA71 infection. On day 9 infection, the survival rates were 0%, 40%, and 80% in the PBS-treated, bivalent F1-hFc-treated, and tetravalent F1×F1-hFc-treated groups, respectively (Fig. 8A). Similar to the present study (Fig. 7B), only mice receiving tetravalent F1×F1-hFc, but not bivalent F1-hFc, treatment were free from paralysis (Fig. 8B). From the data in Fig. 8C and D, it is apparent that the antibodies demonstrated an in vivo neutralization ability in a dose-dependent manner. Notably, a half dose of the tetravalent F1×F1-hFc provided a better protective effect than did a half dose of the bivalent F1-hFc in vivo (70% versus 40% survival rate, respectively) (Fig. 8C and D).

FIG 8.

Single treatment using tetravalent F1×F1-hFc or bivalent F1-hFc can alleviate EVA71 in vivo. (A and B) Kaplan-Meier survival curve (A) and disease scores (B) of 3-week-old hSCARB2 transgenic mice challenged with lethal-dose EVA71 (EVA71 TW/73/12) 1 day after administration of sdAb fusion proteins. (C and D) Kaplan-Meier survival curves for different doses of bivalent F1-hFc (C) and tetravalent F1×F1-hFc (D) administered to EVA71-challenged 3-week-old hSCARB2 transgenic mice. The total numbers (n) of mice per group from two independent experiments are shown. A log rank test was used to analyze the statistical differences in survival rate. One-way ANOVA with Kruskal-Wallis test was used to analyze the statistical differences of the individual groups with disease score. The symbols ** and *** are used to indicate P values of <0.01 and <0.001, respectively.

Therapeutic efficacy of EVA71-specific sdAbs in hSCARB2 transgenic mice.

To further investigate the therapeutic efficacy of F1×F1-hFc, hSCARB2 transgenic mice were administered 200 μg/mouse tetravalent F1×F1-hFc 1 day after infection with a lethal dose (1 × 10⁶ PFU) of EVA71 mouse-adapted strain MP4. On day 4 after infection, the survival rates were 0% and 66.67% in the PBS-treated group and the tetravalent F1×F1-hFc treated groups, respectively (Fig. 9A) (n = 6, P < 0.001). We also conducted new animal experiments using minocycline as a positive control because of a recent paper describing minocycline’s antiviral activity in EVA71 infection. Based on the report by Liao et al., the survival rate of infected suckling mice with double-dose minocycline (5 mg/kg of body weight) treatment was significantly different from that in the PBS-treated group (80.0% versus 38.5%, respectively; P < 0.05) (35). Therefore, 3-week-old hSCARB2 transgenic mice were intraperitoneally injected with 100 μg F1×F1-hFc or 25 mg/kg minocycline 4 h, 24 h, and 48 h after EVA71 infection (2 × 10⁵ PFU of MP4). Neither triple-dose treatment of 100 μg F1×F1-hFc nor 25 mg/kg minocycline was functional for blocking MP4 infection (Fig. 9B). Therefore, we can conclude that the mouse-adapted MP4 strain was more fatal than was the nonadapted strain in hSCARB2 transgenic mouse, and that the medication minocycline (25 mg/kg) could not serve as an available positive control for this severe challenge.

FIG 9.

Survival days of F1×F1-hFc or minocycline-treated hSCARB2 transgenic mice preinfected with EVA71 mouse-adapted strain MP4. (A) Three-week-old mice preinfected with 1 × 10⁶ PFU of EVA71 were given 200 μg/mouse tetravalent F1×F1-hFc via intraperitoneal injection at 24 h postinfection (n = 6). The symbol *** was used to indicate a P value of <0.001. (B) Three-week-old mice preinfected with 2 × 10⁵ PFU of EVA71 were given 100 μg/mouse tetravalent F1×F1-hFc or 25 mg/kg minocycline via intraperitoneal injection at 4, 24, and 48 h postinfection (n = 7 or 8).

DISCUSSION

EVA71 is an important neurotropic picornavirus that can result in polio-like paralysis even in countries where polio has been eradicated. To date, no efficacious anti-EVA71 vaccine or drug has been available around the world (36). Therefore, the unmet need is particularly urgent for anti-EVA71 vaccines or therapeutics to control imminent and threatening outbreaks of EVA71 infection. Just as palivizumab is a proven RSV prophylaxis in at-risk infants (9), the tetravalent F1×F1-hFc antibody developed in this study also holds promise for EVA71 prophylaxis to protect high-risk individuals or to interrupt transmission in the population, according to the in vitro and in vivo results.

In this study, we discovered a novel sdAb, F1, isolated from an immunized llama, followed by phage display selection for anti-EVA71 activity. F1 showed in vitro and in vivo antiviral activity, and the binding epitopes of F1 were identified to target new conformational epitopes in two VP3 regions, VP3^46–60 and VP3^106–130. Subsequently, we demonstrated an effective strategy to engineer a tetravalent sdAb with much improved therapeutic efficacy against EVA71. The sdAbs with different valency were assessed for antiviral efficacy in EVA71 in vivo challenge models. Both the bivalent F1-hFc (200 μg/mouse) and the tetravalent F1×F1-hFc (100 μg/mouse) reduced the mortality and the disease scores in EVA71-infected mice (Fig. 7 and 8). These results are in congruence with those from previous reports indicating that tetravalent or multivalent designs would give rise to superior refinements (37, 38). Further crystallography or structural studies to elucidate the mechanisms involved in this phenomenon are needed.

Many of the therapeutic anti-EVA71 antibody-related studies employed the immunodeficient AG129 mice or suckling mouse model (18, 39, 40). The identification of EVA71 viral receptors and the subsequent development of hSCARB2 transgenic mice have facilitated anti-EVA71 studies and therapeutic development (14, 15, 41, 42). To our knowledge, the current study provides the first evidence that sdAbs can prevent EVA71 infection in weaning-age hSCARB2 transgenic mice.

Importantly, the current study results also indicate that one dose of prophylactic treatment is sufficient to protect mice from EVA71-mediated diseases. Just as a half-life extension antibody has become a potential RSV prophylaxis for infants during the RSV season (43), additional efforts concerning further engineering of the tetravalent and highly potent anti-EVA71 sdAb to bear optimized properties as potential EVA71 prophylaxis for high-risk populations during HFMD epidemics are also warranted.

The current study examined the prophylactic efficacy of F1×F1-hFc in mice, similar to previous studies (15, 32, 44). Based on our present studies, the EC₅₀s of F1×F1-hFc for EVA71 genotypes C4, C5, and B5 are all less than 9 μg/ml (Table 2). Furthermore, only EVA71 genotype C4 vaccines have been approved for marketing in China (45). Moreover, EVA71 strains of genotype C have been the major outbreak strains in the Asia-Pacific region, including Taiwan, China, Japan, the Philippines, and South Korea, since 2010 (46). Therefore, the therapeutic effect of F1×F1-hFc was evaluated in genotype C EVA71 (MP4)-infected hSCARB2 transgenic mice. Notably, single higher-dose treatment of F1×F1-hFc (200 μg/mouse, postinfection 24 h) could extend the survival days of mouse-adapted EVA71-infected hSCARB2 transgenic mice with a significant difference from the PBS-treated group (n = 6, P < 0.001) (Fig. 9A). We also conducted new animal experiments using minocycline as a positive control because of a recent paper describing minocycline’s antiviral activity in EVA71 infection (35). Neither triple-dose treatment of 100 μg F1×F1-hFc nor 25 mg/kg minocycline was functional for blocking MP4 infection (Fig. 9B). Therefore, we conclude that mouse-adapted MP4 strain was more fatal than was the nonadapted strain in hSCARB2 transgenic mouse since medication with minocycline (25 mg/kg) was ineffective in this severe challenge. Future studies are needed to evaluate if the administration of F1×F1-hFc, with an optimal dosing regimen, after EVA71 infection can reduce disease scores and rescue mice from a lethal challenge.

In summary, more effective therapeutic strategies for EVA71 or related pathogens are needed, and the results from this study demonstrated a progressive refinement strategy for effectively preventing or treating EVA71 and other emergent viruses. Further optimization on antibody developability based on the newly identified sdAb clone F1 and its tetravalent counterpart F1×F1-hFc may result in more effective antivirals for EVA71.

MATERIALS AND METHODS

EVA71-specific sdAb selection, expression, and purification.

Llama immunization and antibody library construction were performed by GenScript Co., Ltd. (Nanjing, China). The EVA71 TW/73/12 (genotype C4) used as an antigen was isolated from Chang Gung Memorial Hospital in Taiwan in 2012. EVA71 purification was performed using sucrose gradient ultracentrifugation, according to methods previously described (47). Formalin-inactivated EVA71 was mixed with an adjuvant or phosphate-buffered saline (PBS) and injected into a llama. The animal was immunized seven times over the course of the project. The peripheral blood samples were collected at the preimmunization stage and after the fourth and sixth immunizations for immune response evaluation. The immune response was evaluated through enzyme-linked immunosorbent assay (ELISA) by using the sera against the immobilized immunogen (4 μg/ml formalin-inactivated EVA71).

Total RNA was extracted from peripheral blood lymphocytes, converted into complementary DNA, and cloned in a phagemid vector to generate an sdAb phage display library after the final injection. Bacteriophages carrying EVA71-specific sdAbs were selected by panning on an immobilized antigen (20 μg/ml formalin-inactivated EVA71). An ELISA-positive reaction was determined at a high signal-to-background (S/N) ratio (i.e., ratio of the optical density at 450 nm [OD₄₅₀] of the sample to that of the blank) of >15. Thirty positive phage clones were selected for the DNA sequencing and alignment process to generate four lead EVA71-specific binders.

To generate recombinant bivalent sdAb-hFc, synthetic CHO codon-optimized sdAbs from major binders were cloned into a mammalian cell expression vector (modified from pSecTag2 mammalian expression vector [Thermo Fisher Scientific, Inc.]) containing the hinge and fragment crystallizable (Fc) domains of human IgG1 (GenBank accession no. AEV43323.1) and purified from the culture supernatant fraction via protein A-derived affinity medium (GE Healthcare). An IgG1 backbone was fused with optimized sdAbs to produce bivalent sdAb through an easy purification process without additional tags for fusion. Recombinant tetravalent sdAb×sdAb-hFc was then generated through genetic fusion of two sdAb repeats to the hinge and Fc domains of human IgG1.

Virus preparation and cell-based neutralization assay.

Rhabdomyosarcoma (RD) cells (American Type Culture Collection [ATCC] accession no. CCL-136) were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Gibco) supplemented with 10% fetal bovine serum (FBS). EVA71 TW/73/12 (genotype C4), EVA71 TW/13-50144 (genotype B5), EVA71 TW/10-96050 (genotype C4), EVA71 TW/07-72043 (genotype C5), enterovirus D68 (EV-D68) TW/02795/14, coxsackievirus B3 (CVB3), coxsackievirus A6 (CVA6), and coxsackievirus A16 (CVA16) were isolated from clinical specimens from the Clinical Virology Laboratory of Chang Gung Memorial Hospital (Linkou, Taiwan) (48). EVA71 TW/4643/MP4 (MP4, genotype C2) (the fourth passage of mouse-adapted EVA71 strain from EVA71 TW/4643/98) with increased virulence in mice was kindly provided by Jen-Ren Wang, National Cheng Kung University, Tainan, Taiwan (49). EVA71, EV-D68, CVB3, CVA6, and CVA16 were propagated in RD cells. After the RD cells exhibited cytopathic effect, we collected the supernatant to centrifuge the EVA71, EV-D68, CVB3, CVA6, and CVA16 viruses and then stored it at −80°C. The influenza A/WSN/33 virus was purchased from the ATCC (Manassas, VA) and grown on Madin-Darby canine kidney (MDCK) cells (ATCC accession no. CCL-34), using standard methods.

The neutralization assay was used to measure the ability of a test antibody to inhibit the cytopathic effects induced by viruses. In brief, 96-well tissue culture plates were seeded with 3 × 10⁴ cells/well in DMEM with 10% FBS and incubated at 37°C for 24 h. The sdAb fusion proteins at various concentrations were incubated with 100 50% tissue culture infective doses (TCID₅₀s) of EVA71 for 1 h at 37°C. After incubation, RD cells were infected with the mixture of sdAb fusion proteins and virus. After adsorption, the infected cells were covered with medium containing 2% FBS. The infected cells were further incubated at 37°C for 64 h.

The plates were fixed with 0.5% formaldehyde and then stained with 0.1% crystal violet. The density of the well at 570 nm was measured. Each experiment was performed in triplicate and repeated at least two times. The 50% effective concentration (EC₅₀) was calculated using the formula [(Y − B)/(A − B) × (H − L) + L], where Y represents half of the mean OD at 570 nm (OD₅₇₀) of the cell control without the compound, B represents the mean OD₅₇₀ of wells with the compound dilution nearest to and less than Y, A represents the mean OD₅₇₀ of wells with the compound dilution nearest to and more than Y, and L and H represent the compound concentrations at B and A, respectively.

Binding of sdAb fusion proteins to EVA71.

The binding affinity of EVA71-specific sdAbs was determined using ELISA with immobilized virions (20 μg/ml formalin-inactivated EVA71). Binding of sdAb fusion proteins was revealed with a goat anti-human IgG-Fc fragment cross-adsorbed antibody (Bethyl Laboratories, Inc.) conjugated with horseradish peroxidase (HRP) and HRP substrate (United States Biological). Data were analyzed using GraphPad Prism.

Preparation of recombinant EVA71-specific antibodies.

Expression and purification of recombinant VP1-specific D5 antibody (D5 IgG), E18 antibody (E18 IgG), and E19 antibody (E19 IgG) against conformational epitopes for EVA71 were also assessed using the same strategy based on CHO cell expression system and protein A-derived affinity medium, as previously described in "EVA71-specific sdAb screening, identification, and expression" above. In brief, CHO cell codon-optimized nucleotide sequences were synthesized and then cloned into expression vectors such as pFUSE2ss-CLIg-hK and pFUSEss-CHIg-hG1 (InvivoGen). Heavy- and light-chain amino acid sequences of D5 antibody were obtained from the Protein Data Bank (PDB) (VH sequence identifier [ID] 3JAU_H and VL sequence ID 3JAU_L) (15). Heavy- and light-chain amino acid sequences of E18 antibody were obtained from the National Center for Biotechnology Information (NCBI) data bank (VH, accession no. AZK29196.1; VL, accession no. AZK29195.1), and those of the E19 antibody were obtained from the NCBI data bank (VH, accession no. AZK29199.1; VL, accession no. AZK29197.1) (16).

Preparation of recombinant EVA71 viral capsid proteins.

The pET expression vectors harboring EVA71 viral capsid proteins (VP1, VP2, and VP3) modified from pTBSG1-his vector were kindly provided by Yen-Hung Chow and Chia-Chyi Liu (National Health Research Institutes [NHRI], Taiwan) (50). E. coli-produced recombinant EVA71 viral capsid proteins were purified by nickel-nitrilotriacetic acid (Ni-NTA) chromatography (Qiagen).

Immunoblot and ELISA analysis.

Formalin-inactivated EVA71 (6 μg/well) was separated on a SDS-PAGE, transferred to a polyvinylidene difluoride (PVDF) membrane, and probed with the EVA71-specific MAB979 antibody (Merck KGaA) or F1-hFc. The antibodies that bound to the membrane were detected by incubation with the HRP-conjugated anti-mouse IgG secondary antibody (Thermo Fisher Scientific) specified for MAB979 or the anti-human IgG-Fc fragment cross-adsorbed antibody (Bethyl Laboratories, Inc.) specified for sdAb fusion proteins.

An ELISA was performed by coating Nunc MaxiSorp ELISA plates with 2 or 3 μg/ml recombinant viral capsid proteins or 3 μg/ml recombinant sdAb-hFc fusion proteins. EVA71 VP1-specific D5 antibody and EVA71 VP2-specific MAB979 antibody were used as positive controls. E18 antibody failed to react with recombinant EVA71 viral capsid proteins as a conformational epitope property control. Bound antibody was detected by incubation with the HRP-conjugated goat anti-human IgG-Fc fragment cross-adsorbed antibody (Bethyl Laboratories, Inc.) or the HRP-conjugated F(ab′)₂ goat anti-mouse IgG-F(ab′)₂ fragment cross-adsorbed antibody (Bethyl Laboratories, Inc.). Bound recombinant EVA71 viral capsid protein was reacted by the rabbit anti-His tag polyclonal antibody (GenScript) and then detected by incubation with the HRP-conjugated F(ab′)₂ goat anti-rabbit IgG-F(ab′)₂ fragment cross-adsorbed antibody (Bethyl Laboratories, Inc.).

An ELISA-based competitive binding assay was performed by coating Nunc MaxiSorp ELISA plates with 10 μg/ml formalin-inactivated EVA71. The antibodies E18, E19, and D5 (1 μM) were prepared with or without different concentrations of F1×F1-hFc or F1-hFc for the competitive binding assay. Bound antibody was detected by incubation with the HRP-conjugated goat anti-human Igκ chain antibody (Merck KGaA). The percentage difference in competitive inhibition was calculated using the formula {100 – [(B/A) × 100]}, where A represents the OD at 450 nm (OD₄₅₀) of the IgG control group without the F1×F1-hFc or F1-hFc, and B represents the OD₄₅₀ of the experimental group with the F1×F1-hFc or F1-hFc, respectively.

An ELISA-based epitope sequence binding assay was performed by coating Nunc MaxiSorp ELISA plates with 20 μg/ml synthetic peptide. A peptide (V58-S74 [VP3^58–74]) from the knob region of EVA71 VP3 (VPTNATSLMERLRFPVS) was synthesized by the CRO Company (GenScript). The purity of the synthesized peptide exceeded 85%. The synthetic peptide was resuspended in double-distilled water (ddH₂O) at a concentration of 10 mg/ml stock solution at –80°C. Bound antibodies were detected by incubation with the HRP-conjugated goat anti-human IgG-Fc fragment cross-adsorbed antibody (Bethyl Laboratories, Inc.).

Overlapping-peptide ELISA-based epitope screening experiments were performed by coating Nunc MaxiSorp ELISA plates with 20 μg/ml synthetic peptides. A library of 47 synthetic 15-mer peptides covering the VP3 region of EVA71 (TW/2086/98) was kindly provided by Yen-Hung Chow and Chia-Chyi Liu (NHRI, Taiwan) (51). This 15-mer peptide library with a 10-amino-acid (aa) overlap between adjacent peptides were applied for epitope mapping. Bound antibodies were detected by incubation with the HRP-conjugated goat anti-human IgG-Fc fragment cross-adsorbed antibody (Bethyl Laboratories, Inc.).

Prophylactic efficacy of sdAb fusion proteins in hSCARB2 transgenic mice.

The hSCARB2 transgenic mice were bred and maintained as described previously (52). Groups of 10 hSCARB2 transgenic mice aged 3 weeks were intraperitoneally injected with virus (EVA71 TW/73/12) at the indicated dose of 10⁵ PFU/mouse and were observed for clinical signs. The severity of central nervous system syndromes was scored from 0 to 4, using the following criteria for scoring clinical signs: 0, normal movement; 1, jerky movement; 2, paralysis of one hind leg; 3, paralysis of both hind legs; and 4, death. To evaluate the neutralizing effect of EVA71-specific sdAb fusion proteins, hSCARB2 transgenic mice were injected intraperitoneally with 200 μl of the solution containing premixed sdAb fusion proteins (200 μg/mouse) and EVA71 at 37°C for 1 h. To evaluate the prophylactic efficacy of EVA71-specific sdAb fusion proteins, hSCARB2 transgenic mice were intraperitoneally injected with 200 or 100 μg sdAb fusion proteins 1 day before EVA71 infection.

Therapeutic efficacy of sdAb fusion proteins in hSCARB2 transgenic mice.

To evaluate the therapeutic effect of EVA71-specific sdAb fusion proteins, 3-week-old hSCARB2 transgenic mice were intraperitoneally injected with 200 μg sdAb fusion proteins (F1×F1-hFc) 1 day after EVA71 infection (1 × 10⁶ PFU of MP4). To evaluate the therapeutic effect of triple-dose treatment of F1×F1-hFc or minocycline, 3-week-old hSCARB2 transgenic mice were intraperitoneally injected with 100 μg F1×F1-hFc or 25 mg/kg minocycline 4 h, 24 h, and 48 h after EVA71 infection (2 × 10⁵ PFU of MP4). Minocycline (minoline [100 mg/capsule]) was dissolved in PBS solution, and this solution was prepared fresh daily.

Ethics statement.

All animal experiments were conducted in accordance with the policies and procedures set forth in the Guide for the Care and Use of Laboratory Animals (53). All procedures were approved by the Institutional Animal Care and Use Committee of Chang Gung University, Taiwan (IACUC approval no. CGU15-064).