Specific polyclonal F(ab’)₂ neutralize a large panel of highly pathogenic avian influenza A viruses (H5N1) and control infection in mice

Since 2003, highly pathogenic avian influenza (HPAI) A/H5N1 (H5N1) viruses have been associated with almost 650 reported infections to the WHO, with a high mortality rate (59%) [1]. Human infection with influenza H5N1 virus often occurs after direct handling or after close contact with infected poultry [2,3]. So far, very few cases of human-to-human transmission have been identified [3,4]. However, the continuing occurrence of human infections since 2003 provide the H5N1 virus opportunities for reassortment and/or mutations associated with better person-to-person transmission.

In humans, severe H5N1 influenza infection is usually associated with viral loads and hypercytokinemia [3]. In one study, infectious virus was recovered from the blood of 40% of severely ill patients [5]. Specific antibodies can be detected after approximately 2–3 weeks [6,7]. The combination of the direct damage caused by the virus and the intense inflammatory reaction of the patients is probably responsible for the high severity of H5N1 infection [8,9].

Since 2003, the fatality rate associated with H5N1 infection has not improved. Although several new antiviral drugs have been developed recently (e.g., RNA polymerase inhibitors such as Favipiravir, and inhibitors of the hemagglutinin-mediated viral entry) [10,11], the antiviral therapies currently marketed are limited to adamantanes (amantadine and rimantadine), which inhibit the M2 ion channel, and to the neuraminidase inhibitors (oseltamivir, peramivir, zanamivir and lanimavir). Owing to increased resistance to adamantanes in clade 1 viruses isolated in South-East Asian countries until 2012, the treatment mainly relies on neuraminidase inhibitors, such as oseltamivir. Reduced sensitivity of some H5N1 virus strains to neuraminidase inhibitors has been observed occasionally [12–14] and appearance of resistant mutations has also been reported in patients during the course of the treatment [15,16].

Passive immunotherapy could be an interesting alternative or an additional strategy when used in combination with antiviral drugs for the treatment of H5N1 infections. Some studies in mice have already demonstrated the efficacy of monoclonal [17–19] or polyclonal antibodies on H5N1 infections [20–22]. Human convalescent plasma was also reported to efficiently help in controlling the infection in severely ill patients [23,24].

H5N1 virus is rapidly evolving, which poses serious public health concerns [25]. This fast evolution makes prophylactic vaccine development complicated, and in this context the use of polyclonal immunoglobulins appears to have many advantages, limiting the risk of viral escape mutations and allowing rapid development and production of a novel efficient treatment. Moreover, such products are particularly well suited for emergency conditions, as is currently the case for rabies [26,27] or envenomation [28–30]. While in the past equine antisera were sometimes the cause of significant side effects due to anaphylactic shock, mainly due to the presence of Fc fragments, the new generation of processed and purified polyclonal immunoglobulins containing highly purified F(ab’)₂ fragments, are well tolerated [26].

Here we report the development and the evaluation in a mouse model of specific polyclonal anti-H5N1 immunoglobulins to investigate their therapeutic potential for the clinical management of humans exposed to HPAI H5N1 virus.

Material & methods

Equine hyperimmune sera

Four healthy French trotter horses were immunized with an anti-H5N1 split vaccine containing the seed-inactivated A/Vietnam/1194/2004 H5N1 virus. Horses had no detectable antibodies against H5N1 virus before immunization and were strictly controlled for several viruses. Blood samples were collected regularly up to 4 weeks after immunization. Plasma were prepared and stored at -20°C.

Preparation of equine F(ab’)₂ fragments

Pooled horse plasma was purified as described previously to obtain highly purified F(ab’)₂ fragments [31,32]. The main purification steps consist of: anion-exchange chromatography steps, which eliminate proteins including albumin; hydrolysis of whole immunoglobulins into F(ab’)₂ fragments in order to eliminate Fc fragments, which are horse specific; and pasteurization (heat treatment) at 60°C for 10 h for the viral safety of the product. The final bulk product (FBF001) is filtered at a level of 0.2 µm and then stored at +5°C ± 3°C until use.

High-performance liquid chromatography

High-performance liquid chromatography (HPLC) using a TSK G3000 SW column packed with silica support (length 60 cm; internal diameter [ID] 7.5 mm) was employed to determine the molecular size distribution of anti-H5N1 immunoglobulin fragments at different production stages. A total of 120 µg of protein contained in 30 µl was injected, and the separation was performed at a flow rate of 0.5 ml/min.

Viruses & animals

All influenza A H5N1 strains used in this study were amplified using Madin-Darby canine kidney (MDCK) cells. The infectious titer was determined by limit-dilution in MDCK cells cultures, and expressed as the 50% tissue culture infectious dose (TCID₅₀). The H5N1 strains used in this study were selected for their representativeness of H5N1 virus circulation in the world since 2003 and to represent the last 9 years of virus evolution in Cambodia (with viruses from clades 1, 1.1.1 and 1.1.2) (Figure 1).

All in vitro experiments involving infectious virus were performed in a biosafety level 3 laboratory. In vivo experiments, in 8–10-week-old BALB/c mice (Charles River Laboratories, MA, USA) were performed in a biosafety level 4 laboratory. Mice were housed and manipulated according to the guidelines of Directive 2010/63/UE.

Mice were anesthetized with 5% isoflurane (1-chloro-2,2,2-trifluoroethyl difluoromethyl ether) in an induction chamber before viral infection and treatment with FBF001.

The 50% lethal dose (LD₅₀) for A/Vietnam/1194/2004 virus was determined in BALB/c mice by mortality analysis of six groups of five mice infected with six different dilutions of infectious virus (mock or 10² to 10⁶ × TCID₅₀). The titer was calculated by the Reed-Muench method [33]. A/Vietnam/1194/2004 virus was diluted in D-PBS to obtain solutions titrated at respectively 1, 10 and 100 × LD₅₀ (corresponding to 10³, 10⁴ and 10⁵ × TCID₅₀, respectively). Mice were inoculated through intranasal route with 16 µl of one of the infectious virus doses.

Microneutralization & hemagglutination inhibition assays

For standard microneutralization (MN) assay, 100 × TCID₅₀ of each virus strain tested (Table 1) were incubated with two different serial dilutions of FBF001 product (solution concentration: 29.4 g/l; from 1/20 to 1/18,000) for 1 h at room temperature prior to addition to MDCK cells. Cell monolayers were incubated for a further 3 days and examined for cytopathic effects. Determination of the end point neutralizing antibody titers was performed in four wells per dilution. The neutralizing titer was defined as the reciprocal of the highest dilution of serum at which the infectivity of 100 × TCID₅₀ of H5N1 virus for MDCK cells was completely neutralized in 50% of the wells. The titer was calculated by the Reed-Muench method [33]. Hemagglutination inhibition (HI) testing was performed with human group O red blood cells according to the method described by Rowe [34]. All tests were performed in triplicate.

Protective efficacy of polyclonal equine F(ab’)₂ against A/Vietnam/1194/2004 in mice

Mixtures of FBF001 and virus were prepared in D-PBS to obtain doses corresponding to 5, 0.5, 0.1, 0.05, 0.025, 0.01 and 0.005 µg FBF001/animal. Each dose also contained 10 × LD₅₀ of virus. After 1 h of incubation at 34°C, mice received an intraperitoneal (i.p.) injection (300 µl) of one of the mixtures.

In vivo prophylactic & therapeutic efficacy of polyclonal equine F(ab’)₂ against A/Vietnam/1194/2004 in mice

8–10 week-old BALB/c mice were inoculated intranasally (i.n.) with 1, 10 or 100 × LD₅₀ of virus at day 0. 200 µl of FBF001 solutions were injected i.p. to obtain final doses of 2.5, 5, 10, 20 or 40 mg/kg, 1 day after infection or during 5 consecutive days after infection. As negative control, non-infected mice were i.p. inoculated with FBF001 with the highest concentration of FBF001. As positive control, mice were i.p. inoculated with 200 µl of buffer.

Statistical analyses

• • in vitro data: all statistical analyses were performed using Stata/SE version 12.0 (StataCorp, TX, USA). Significance was assigned at p = 0.05 for all parameters and the standard deviation (SD) was calculated for mean. Non-parametric test Wilcoxon rank-sum test were used for continuous variables;

• • In vivo data: all statistical analyses were performed using GraphPad Prism 6 (GraphPad Software, CA, USA). Survival data were analyzed by Gehan-Breslow-Wilcoxon test (treated group vs non-treated group).

Results

Preparation of highly purified F(ab’)₂ fragments

The production of highly purified F(ab’)₂ fragments was based on hydrolysis and purification of hyperimmune equine plasma, produced by immunization of horses with inactivated A/Vietnam/1194/2004 H5N1 virus. Equine hyperimmune plasma was processed, following a validated process described previously [31,32]. SDS-PAGE showed that digestion with pepsin produces fragments of approximately 100 kDa, corresponding to F(ab’)₂ fragments (data not shown). Smaller proteins, including albumin, α- and β-globulins and Fc fragments, which are responsible for the high reactogenicity of the crude preparations are eliminated by the process. At the end of the purification process, HPLC analysis of the bulk solution of immunoglobulins (FBF001) showed the presence of approximately 90% of F(ab’)2 fragments in the product (Figure 2). The proportion of F(ab’)₂ plus Fab’ in the product was almost 99%, showing the very high purity level of this polyclonal immunoglobulin fragments solution.

in vitro activity of FBF001 anti-H5N1 F(ab’)₂

FBF001 was tested against the homologous strain A/Vietnam/1194/2004 and titers of 1/2560 and 1/4000 were obtained by HI and MN tests, respectively.

Cross-neutralization activity of FBF001

A study was designed to confirm the neutralizing activity of FBF001 on 21 different H5N1 strains isolated over a 9-year period in different parts of the world, and representative of 11 different clades/sub-clades that emerged worldwide following H5N1 virus natural evolution (Figure 1). in vitro neutralization titers ranged from 1/1000 to 1/8000 for all clades tested (Table 1). These neutralization results were confirmed by HI assay with titers ranging from 1/960 to 1/2560 (Table 1). The mean MN and HI titers were respectively 1/4416 and 1/1600 for clade 1 viruses, 1/3666 and 1/1600 for clade 2 viruses, and 1/8000 and 1/1920 for clade 4, 7 and 9 viruses. None of the differences of MN and HI titers between clade 1, 2, 4, 7 and 9 viruses were significant.

Protective efficacy of FBF001 in vivo

To confirm the neutralizing activity of FBF001 in vivo, specific experiments were performed using a BALB/c mice model. This mouse model has already been shown susceptible to non-adapted H5N1 strains [35], with evidence of virus replication in the lungs.

As shown in Figure 3, 0.1 µg of FBF001 was sufficient to prevent the lethal effects of 10 × LD₅₀ of H5N1 viruses i.n. injected in mice and 5 ng protected 60% of animals against death by avian influenza infection.

In vivo proof-of-concept of efficacy of FBF001 to treat H5N1 infection in mice

The BALB/c i.n. mouse model was also used to assess the potential therapeutic efficacy of FBF001 against H5N1 virus, using various bodyweight doses of FBF001. One or multiple doses of FBF001 at 40 mg/kg were i.p. injected in mice inoculated with 10 × LD₅₀ of A/Vietnam/1194/2004 H5N1 influenza virus. Mice received FBF001 24 h after virus challenge, or at days +1, +2, +3, +4 and +5 after challenge.

When treatment was administered for 5 consecutive days after viral challenge (days +1 to +5; group 2 [Figure 4]) 100% protection against H5N1 challenge was observed. One dose administered 24 h after viral challenge resulted in a 48-h delay in mortality, but ultimately no significant effect on mouse survival rate was observed in comparison with the control group.

Determination of the optimal therapeutic dose in mice

The same model was used to perform a dose-ranging study with multiple conditions of experimental infection, designed to evaluate the optimal dose of FBF001 to be used in a Day+1 to Day+5 protocol. Doses varying from 2.5 to 10 mg/kg were tested in animals infected with 1 × LD₅₀ of A/Vietnam/1194/2004 virus H5N1 virus (Figure 5A).

A mortality rate of 80% was observed after 8 days in the control group (Figure 5A; group 5). A better survival rate was observed with doses of 5 mg/kg or more (groups 2 and 3). Treatment with doses at 10 mg/kg resulted in a 60% decrease in mortality compared with untreated animals (Figure 5A, group 3).

When mice were challenge with 10 × LD₅₀ of A/Vietnam/1194/2004 virus H5N1 virus (Figure 5B), the survival rate did significantly increase in comparison with untreated animals (group 5) at 14 days after virus challenge when a 5-day protocol designed at 20 mg/kg dose (group 2) was used. A 10% increase in survival was observed in animals treated with 10, 5 or 2.5 mg/kg of FBF001 and mortality was delayed in all treated groups (Figure 5B).

In the animal groups challenged with 100 × LD₅₀ of virus (Figure 5C), a 48-h delay in occurrence of the first death associated with a 10% mortality rate was obtained in groups that received 2.5 and 5 mg/kg. An increase in 30% of the survival rate combined with a 48-h delay in first death occurrence was obtained with a dose of 10 mg/kg (group 2).

Discussion

The HPAI H5N1 virus, which was first detected in 1997 in Asia, remains a major source of public health concern. Although the number of cases is limited on a global scale, H5N1 infection is consistently associated with a very high fatality rate of almost 59%. Only a few mutations in the receptor-binding site or a reassortment between H5N1 and a seasonal influenza virus could lead to the emergence of a new influenza strain with pandemic potential.

Usually, the disease develops after a short incubation period and a peak of viremia is reached in only a few days [8]. The presence of virus in extrapulmonary tissues of some patients suggests that HPAI H5N1 virus has, as in animals, the potential to disseminate to other organs and tissues in humans [7]. Therefore, the use of post-exposure vaccination does not seem a suitable therapeutic option in exposed individuals. Additionally, current antiviral drugs such as oseltamivir have only a limited impact on mortality, even at twice the recommended dose, particularly if they are not administered during the very early phase of the disease [3]. It is therefore prudent to explore other therapeutic options.

In clinical emergency situations, drugs must provide a broad spectrum of action and instant activity to stop, as quickly as possible, the progression of the infection. By its mechanism of action, passive immunotherapy has the advantage over the orally administrated drugs, thanks to an immediate effect on the virus. Some authors have proposed the development of monoclonal human or humanized antibodies against HPAI H5N1 [17,19,36]. However, despite encouraging results, this strategy presents major limitations, such as the possibility of emergence of escape mutants, the long research and development phase and specific process development that are incompatible with a clinical emergency, and the high cost of such treatment.

By contrast, polyclonal immunoglobulins may offer many advantages in terms of cross-reactivity, rapidity and cost of development. Unlike antiviral drugs, they could provide specific and instant protection and may eventually reduce the risk of transmission through droplets if the correct epitopes are neutralized [37].

Convalescent human plasma already demonstrated promising results in the treatment of H5N1 infections, most probably by decreasing the viral load [24]. A meta-analysis on the use of convalescent plasma during the 1918 influenza pandemic has shown that this strategy was effective in reducing mortality if administered early [38].

Polyclonal immunoglobulins have been used for decades in the management of other clinical emergencies, such as snakebite envenomation or severe poisoning (e.g., tetanus toxin, botulin toxin, digoxin) [39]. Severe infectious diseases, such as rabies, are also treated by intramuscular injection of specific polyclonal immunoglobulins, in combination with classical anti-rabies vaccines [27].

Equine rabies polyglobulin production methods were followed to produce these equine anti-H5N1 polyclonal immunoglobulins that consist of highly purified F(ab’)₂, free of the Fc fragment, which holds the species determinant. While historically equine serum could be responsible for significant side effects, there is now an extensive history of use of purified equine polyglobulin fragments (F(ab’)₂) demonstrating their efficacy and safety in humans, with no serious adverse reaction or serum sickness reported [26,40].

Our data surprisingly demonstrate that the equine anti-H5N1 preparation, FBF001, not only has a high neutralization activity (titer of 1/6000) against the homologous A/Vietnam/1194/04 strain (Table 1), but also a very broad spectrum of neutralizing activity against heterologous strains from the same or different clades, while a previous study demonstrated only slight cross-neutralization activity of equine F(ab’)₂ on heterologous strains [22]. This suggests that FBF001 may recognize a wide range of epitopes, including conserved epitopes, such as those located in regions like the stalk domain of the HA protein [41]. The protective efficacy of FBF001 was confirmed in vivo(Figure 3) even at concentrations lower than 0.1 µg (still 60% protection with 5 ng of FBF001), suggesting a potential indication for use as prophylactic treatment in persons recently exposed to the virus.

Previous studies have already demonstrated the efficacy of equine anti-H5N1 F(ab’)₂ in reducing viremia and mortality in mice [20,22]. The experiments we performed with FBF001 are the first proof of concept for efficacy of the product in mice that also allowed the design of a therapeutic protocol that could be eventually used in humans (Figures 4 & 5). In a post-exposure protocol a single injection 24 h after exposure was not sufficient to prevent the development of infection, and only led to delay by 48 h of the first case of death. This study demonstrates the necessity of a multiple injection protocol to manage HPAI H5N1 virus in infected mice. Previously published pharmacokinetic data indicated a half-life of 13.7 h after intravenous administration of 10 mg/kg of equine F(ab’)₂ in mice [42]. Since the peak of viremia is reached approximately 5 days after H5N1 infection in humans [8], a 5-day protocol is justified to maintain the availability of adequate concentrations of neutralizing F(ab’)₂ just before and during the peak of viremia in order to control the virus replication and reduce the risks of systemic spread. The use of a repeated dose protocol may not impact the efficacy and safety of the product, such as previously described for anti-SARS-CoV equine F(ab’)₂ tested in macaques in a 4-week daily injection protocol [43].

Two key observations in the mouse model may suggest a potential efficacy of FBF001 in humans: an increase in the survival rate observed in animals by day 14 after infection; and the induction of a delay in infection-related death (in patients a delay of a few days could give an additional time window for other treatments such as antiviral drugs to reach the tissues with optimal concentrations and for the patient to develop its own immune response).

Additionally, although a regimen of 10 mg/kg of FBF001 administered daily for 5 consecutive days was sufficient to obtain an optimal efficacy (Figure 5), a good laboratory practice toxicity test performed in mice using seven repeated 100 mg/kg dose injections did not show any toxicity (data not shown), confirming the good safety potential of such product.

Conclusion

These results demonstrate a real potential of FBF001 (proposed trade name Fabenflu^®, Fab’entech, Lyon, France) for the prophylaxis of individuals exposed to a wide range of HPAI H5N1 viruses, as well as for the treatment of infected patients, especially in association with antiviral drugs, a combined therapy that remains to be evaluated in the mouse model. The optimal therapeutic protocol developed in mice and consisting of five consecutive injections of 10 mg/kg or more of FBF001 on days 1–5 after viral challenge could easily be transposed into treatment for humans after clinical development of this product.