Passive immunization with influenza haemagglutinin specific monoclonal antibodies

ABSTRACT

The isolation of broadly neutralising antibodies against the influenza haemagglutinin has spurred investigation into their clinical potential, and has led to advances in influenza virus biology and universal influenza vaccine development. Studies in animal models have been invaluable for demonstrating the prophylactic and therapeutic efficacy of broadly neutralising antibodies, for comparisons with antiviral drugs used as the standard of care, and for defining their mechanism of action and potential role in providing protection from airborne infection.

Introduction & background

Influenza A viruses infect humans and a wide range of animals, including aquatic birds, poultry and pigs. During annual epidemics, seasonal influenza A virus infections cause an estimated 3 to 5 million cases of severe illness and 290,000 to 650,000 deaths globally.¹ Four influenza A virus pandemics have occurred over the past century, including the 1918 Spanish influenza (H1N1) that claimed at least 40 million lives, followed by the 1957 Asian influenza (H2N2), the 1968 Hong Kong influenza (H3N2) and the 2009 swine influenza (H1N1) pandemics.^2,11 Zoonotic infections occur sporadically; most notably highly pathogenic avian influenza (HPAI) H5 virus infections have been detected in humans in 16 countries with a reported case fatality rate that exceeds 60%,³ and 6 waves of H7N9 avian influenza virus infections reported in China since 2013.⁴

On the surface of the virion, the haemagglutinin (HA) protein is displayed as a trimeric glycoprotein with a globular head and a membrane proximal stem region. The globular head domain of the HA protein dictates the host specificity and tissue tropism of the virus by its capacity to interact with sialic acid receptors, though receptor distribution and oligosaccharide-linked specificity vary in different tissues and hosts.⁵ The HA stem region contains the fusion peptide, which mediates fusion of the cellular and virus membranes resulting in the release of viral nucleocapsid into the cytoplasm. The neuraminidase (NA) protein is required for the release of progeny virus particles from infected cells, thereby facilitating spread.

Influenza A viruses are subtyped based on antigenic differences in two major viral surface glycoproteins, the HA and NA, into 18 different HA subtypes (H1-H18) and 11 different NA subtypes (N1-N11). Phylogenetically, the 18 HA subtypes fall into two major groups, groups 1 and 2.⁵ Group 1 has been further divided into three clades, H1a (that includes the 1918 and 2009 pandemic and seasonal H1N1 viruses), H1b (that includes highly pathogenic avian influenza H5N1 viruses) and H9 (that includes avian influenza H9N2 viruses).^6,7 The two newest HA subtypes were identified by genetic methods in bats (H17 and H18) and are also in group 1.^8,9 Group 2 contains H3 (including currently circulating seasonal H3N2 viruses) and H7 (including H7N9 and highly pathogenic avian influenza H7N7) viruses, among others.

The HA protein is the primary target of the protective antibody response and influenza viruses evade virus-specific immunity by changing their antigenicity through two distinct mechanisms: antigenic drift and shift.¹⁰ Seasonal influenza epidemics are caused by antigenic drift variants in which minor antigenic changes in the HA and NA accumulate over time, resulting from mutations in the HA and NA genes due to an error-prone viral RNA polymerase and positive selection pressure driven by pre-existing immunity. This is an ongoing process. Antigenic shift, which occurs sporadically, is a process by which novel influenza A viruses, to which little or no immunity exists in the human population, emerge and become established in humans. Such novel influenza A viruses are derived from animal reservoirs by direct infection or by reassortment of gene segments encoding the surface glycoproteins (HA and/or NA) from an animal influenza virus with gene segments from previously circulating human influenza viruses. The 2009 pandemic H1N1 virus was a reassortant virus that contained gene segments from a human H3N2, classical swine H1N1, North American avian H1N1 and Eurasian avian-like swine H1N1 viruses.^5,11

Immunodominance of strain-specific HA head binding antibodies

Natural infection induces antibodies that target one or more of five antibody binding sites on the HA head.^12,13 Immunisation with inactivated seasonal influenza vaccine primarily induces strain-specific HA antibodies that are directed at one or more of these epitopes and neutralize virus infectivity, while only a small fraction of cross-reactive antibodies are elicited.¹⁴ Neutralising, HA head-specific antibodies are detected in haemagglutination inhibition (HAI) assays, which measure antibodies that bind in close proximity to the receptor binding pocket of the HA thereby inhibiting virus attachment to red blood cells and their subsequent agglutination (haemagglutination). However, strain-specific antibodies provide incomplete protection against antigenic drift variant viruses of the same subtype and provide no protection against different HA subtypes. In 1993, Okuno et al characterised a murine monoclonal antibody (C179) that cross-reacted with H1 and H2 but not H3 subtypes;¹⁵ this antibody bound an epitope in the highly conserved HA stem. In the past decade, HA stem-reactive antibodies have been identified in humans.^16–18 Corti et al, reported the isolation of plasma cells from recipients of seasonal influenza vaccine that produced heterosubtypic antibodies that could neutralise avian influenza H5N1viruses.¹⁹ However, in other studies, cross-reactive antibodies to the H5N1 virus were only detected in a fraction of seasonal influenza vaccine recipients.²⁰ These findings suggest that while many individuals possess plasma cells capable of producing broadly neutralising antibodies (bNAb), the antibodies cannot be detected easily in the sera. The lack of cross-reactive serum antibody responses despite the presence of detectable plasma cells indicates that strain-specific head binding antibodies comprise the dominant response and this may compromise the development of sustained antibody responses to the stem.^14,20

Improved detection of broadly neutralising antibodies (bNAbs)

Two events paved the way for increased interest into identification, induction and application of bNAbs to protect against influenza. One was the development of methods for rapidly generating monoclonal antibodies from circulating B cells and the other was the 2009 influenza pandemic which boosted stem-reactive HA antibodies. In 2004, Traggiai et al reported the isolation of immortalised neutralising monoclonal antibody secreting cells using an improved method for Epstein-Barr virus transformation of human memory B cells isolated from patients recovering from severe virus infections.²¹ In 2008, Wrammert et al. developed a rapid method for generating high-affinity monoclonal antibodies from peripheral blood B cells of vaccinees, within a month of vaccination.²² Variable gene sequences for the heavy and light chains were amplified by RT-PCR, cloned into antibody expression vectors and co-transfected into mammalian cells for expression of the monoclonal antibody. These developments allowed researchers to study the binding specificity of broadly reactive monoclonal antibodies and to test their protective efficacy in prophylaxis and therapy.

Repeated exposure to seasonal influenza viruses via infection or vaccination focuses anti-HA antibody responses primarily towards immunodominant epitopes on the globular head of the HA protein; broadly cross-reactive antibody responses are rare.²⁰ When the 2009 pandemic H1N1 virus emerged, most of epitopes on the HA head were new, but conserved epitopes in the HA stem and head were boosted^18,23; therefore, HA stem antibodies were detected with much greater frequency following the 2009 pandemic than before. A similar sequence of events, with a boost in HA stem antibodies has led to success in generating HA stem antibodies from recipients of investigational pandemic influenza vaccines.^18,24–28

Broadly-reactive antibody responses following influenza virus infection

Plasma cells capable of producing broadly cross-reactive antibody responses have been detected following 2009 pandemic H1N1, HPAI H5N1 and experimental H3N2 virus infections.^3,14,23 After infection with the H1N1pdm09 virus, Wrammert et al reported two types of plasma cells producing neutralising antibody: (i) those that inhibited haemagglutination (HAI⁺), indicating binding specificity in proximity to the sialic acid receptor binding site (RBS) on the HA head, and (ii) those that did not inhibit haemagglutination (HAI⁻), indicating binding to sites other than the RBS.²³ The activity of the HAI⁻ antibodies was inhibited by previously characterised stem-binding monoclonal antibodies, suggesting that they bound the stem region of the HA. They reported that HA-specific plasma cells made up 30–50% of the H1N1pdm09 virus-specific plasma cell response, that half the neutralising and 10% of all antibodies bound a conserved region on the HA stem, and a majority (63%) of the virus-specific antibodies cross-reacted with seasonal influenza viruses that had circulated earlier. Importantly, they identified five stem-binding antibodies that displayed similar specificity to each other and bound with high affinity to representative H1 influenza viruses isolated over the previous 10 years, the 1918 pandemic virus and the H5 HPAI virus.²³ In a separate study, Hoa et al reported the detection of only low levels of stem-reactive bNAbs prior to infection with the H1N1pdm09 virus.²⁹ They observed an increase in the prevalence of stem-reactive bNAbs in about 67% of the infected and 44% of their household contacts who did not develop influenza-like illness (ILI). Stem-reactive antibodies develop over time, with the lowest titres seen in children and the highest levels observed in adults.^30,31 However, the levels of stem-reactive bNAbs are not sustained and titres dropped to pre-pandemic levels within 2 years.²⁹

In addition to stem reactive bNAbs, a few human monoclonal antibodies (hMAbs) that target highly conserved regions of HA head have also been described.^17,32–34 The properties of head and stem-specific HA antibodies are summarized in Table 1.

Table 1.

Properties of strain-specific and stem-specific HA antibodies.

Property	Strain-specific head Abs	Stem-specific bNAbs	References
Breadth	+	+++	17–19,23,37
Longevity	+++	+	29
Escape	+++	-	12,19,39,48,78
Somatic hypermutation	++	+++	23,24

- Indicates absence of this property.

+ to +++ indicates presence and strength of property.

Broadly-reactive antibody responses following influenza vaccine

The development of vaccines has resulted in significant reduction in the morbidity and mortality associated with many viral pathogens. Vaccination has resulted in the global eradication of smallpox and a drastic reduction in the global incidence of viral diseases such a poliomyelitis (only 37 cases reported globally in 2016).^35,36 However, the prevention of influenza is particularly challenging due to the changes resulting from antigenic shift and drift. Seasonal influenza vaccines are updated annually to keep pace with antigenic drift of circulating influenza viruses.

Several recent studies have highlighted the induction of plasma cells that can produce head and stem-specific broadly-reactive antibody responses.^17–19,24 Dreyfus et al reported on plasma cells that produce antibodies such as CR9114, with reactivity that spans selected HAs from group 1 and 2 (H1, H3, H5, H7 & H9) as well as influenza B viruses.¹⁷ In contrast, Corti et al. observed that seasonal influenza vaccination induced plasma cells that produced antibodies with reactivity that was primarily confined to group 1 influenza viruses, and did not extend to group 2 viruses.¹⁹ They suggested that the presence of an additional glycosylation site in the group 2 HA blocked binding of these antibodies to group 2 HAs. Li et al, reported increased rates of induction of broadly-reactive plasmablast responses following 2009 H1N1 pandemic vaccine administration, similar to their findings following 2009 H1N1 pandemic infection.¹⁸

As seen after influenza infections, three types of antibodies are observed following influenza vaccination: antibodies that bind the RBS (Neut⁺, HAI⁺) on the globular head, stem-binding antibodies that were either neutralizing antibodies (Neut⁺, HAI⁻ e.g. 70–1F02 and 41–5E04), or non-neutralizing antibodies (e.g. CR9114).^17,24 Kallewaard et al reported the development of MEDI18852, a neutralising antibody derived by optimisation of an hMAb that reacts with all 18 influenza A HA subtypes.^37,38

Preclinical evaluation

Binding and accessibility

Historically antibody binding sites were mapped by inducing escape mutations³⁹ but most are now mapped by solving the structure of the antibody when bound to the HA.^40–48 Classically, the stem-specificity of a newly derived MAb has been defined by its capacity to competitively inhibit previously known stem-binding antibodies, such as C179.^17,19,24,29 Despite the complex 3D structure and close packing of the HA trimers on the surface of the virion, the stem epitope on the HA was shown to be accessible to stem-binding bNAbs such as C179 by cryo-electron microscopy (cryo-EM).⁴⁹ In an effort to develop a universal influenza vaccine candidate, clinical trials are underway to test novel influenza viruses expressing chimeric HA molecules with the head domain derived from an avian influenza virus and stem from an H1 virus.⁵⁰ On cryo-EM the chimeric HA displayed a 60° rotation between the stem and the head domains leading to a more open structure and the recombinant virus expressing the chimeric HA displayed an increased density of HA molecules on the virion, relative to its native counterparts.

Mechanism of action of broadly reactive antibodies

Classically, neutralisation has been regarded as the primary antiviral mechanism for virus-specific antibodies. However, influenza-specific antibodies mediate their antiviral activity through multiple mechanisms, including neutralisation of free virus particles (HA-specific antibodies), interference with virus release from cells (NA-specific antibodies) and complement-dependent lysis (CDL) of infected cells (HA stem-specific antibodies).^51–53

In vivo passive transfer studies have shown that both neutralising (e.g. MEDI18852, 70–1F02 and 41–5E04) and non-neutralising (e.g. CR9114) stem-reactive antibodies confer significant protection from lethal influenza virus challenge,^17,24,37,38 highlighting their potential for the development of effective therapeutics or universal vaccines against influenza.^19,24 Some stem-specific antibodies prevent pH-induced conformational changes required for membrane fusion.¹⁷ However, recent attention has focused on interactions between antibody Fc and Fc$\mathrm{\gamma }$R expressing cells in mediating in vivo protection.⁵⁴ HA stem-specific bNAbs have been shown to require Fc-Fc$\mathrm{\gamma }$R interaction to mediate in vivo protection, while strain-specific head-binding antibodies conferred in vivo protection even in the absence of Fc-Fc$\mathrm{\gamma }$R interaction.⁵⁵ Thus, Fc-Fc$\mathrm{\gamma }$R interaction may be necessary for in vivo protection mediated by the low titres of stem-specific bNAbs.^55,56 Some stem-specific bNAbs potently induce antibody-dependent cellular cytotoxicity (ADCC) by engaging and activating NK cells through Fc-Fc$\mathrm{\gamma }$R interaction. Non-neutralising HAI⁻ antibodies have also been identified that when passively transferred, confer in vivo protection through ADCC and CDL-independent mechanisms, possibly through phagocytosis of immune complexes.²⁴

Taken together, HAI⁺ antibodies that bind the HA head proximal to the RBS dominate the protective response following infection or vaccination, through neutralisation-based mechanisms.⁵⁷ When strain-specific responses wane or are absent, bNAbs may confer protection via Fc-Fc$\mathrm{\gamma }$R interaction mediated mechanisms including ADCC and phagocytosis.

Strategies to induce broadly cross-reactive antibodies

Currently licensed influenza vaccines are focussed on the induction of neutralising antibody responses, particularly strain-specific antibody responses directed at the globular head of the HA molecule that neutralise virus infectivity. However, seasonal influenza viruses are constantly evolving and novel zoonotic viruses emerge to which humans possess limited immunity. Thus, vaccines that induce broad cross-protection against antigenic variant viruses within the same subtype and ideally against other HA subtypes are of great interest. The observation that natural infection only elicits low titres of stem-specific antibodies,²⁹ coupled with the fact that influenza epidemics occur despite repeated exposure to antigenically diverse HAs through natural infection and seasonal influenza vaccination suggests that the induction of high titres of cross-reactive HA stem antibodies will require a novel approach.¹⁴

A number of strategies have been proposed to improve the induction of cross-reactive antibodies. Multiple successive exposures to chimeric HAs bearing antigenically distinct heads and conserved stems has been proposed to preferentially boost antibody responses to the conserved stem.^14,30,58,59 Studies in mice have shown that sequential administration of DNA vaccines encoding different H3 HAs or vaccination with only the stem portion of the HA were successful in inducing HA stem bNAbs.^60,61 However, the effectiveness of these strategies in human vaccinees is yet to be demonstrated.

Computationally designed small protein mimics of antibodies that specifically bind to the HA stem of group 1 viruses have been described as an alternative to development of bNAbs for therapy.^62,63 Despite the lack of an Fc region, small protein antibody mimics induced relatively weak cytokine responses and provided significant protection in both prophylactic and therapeutic studies in mice and would potentially be safe and effective even in immunodeficient individuals.

In vivo evaluation

Studies in animal models have been of great value in demonstrating efficacy for prophylaxis and treatment in comparison with antiviral drugs that represent the standard of care. As summarized in Table 2, these studies have been used to demonstrate a dose-response, breadth of activity against a range of influenza viruses, efficacy of treatment despite a delay in initiation of treatment, the mechanism of action of the antibody and the ability of monoclonal antibodies to protect from airborne infection.

Table 2.

Principles established through the use of animal models for the evaluation of MAbs against the influenza virus HA.

Property/Principle	Animal model	Reference
Efficacy of prophylaxis	Mice, ferrets	3,19,68,38
Efficacy of treatment in comparison with oseltamivir	Ferrets	38
Dose-response	Mice, ferrets	3,19,68
Breadth of activity against a range of influenza viruses	Mice, ferrets	19
Effectiveness despite delay in initiation of treatment	Mice, ferrets	3,38
Mechanism of action	Mice	68
Protection from airborne transmission	Ferrets	38

In prophylaxis studies, the MAb is administered prior to influenza infection while in treatment studies, the MAb is administered after infection, generally with a delay of 48 or 72 hours. The efficacy of treatment with MAbs can be compared with antiviral drug treatment (e.g. oseltamivir), or a combination of the two in experimental animal models.³⁸

The first step of in vivo testing is usually performed in inbred mice. BALB/c mice are commonly used, though DBA mice are used by some investigators because they manifest more severe disease with influenza virus infection.⁶⁴ The antibody is administered by intraperitoneal injection at a single or a range of concentrations; on rare occasions the antibody is administered intranasally.^19,65,79 A control group of animals receive saline or an irrelevant antibody. The animals are then challenged with intranasally administered influenza virus. Weight loss is monitored daily as an objective measure of morbidity; animal ethics committees usually require that animals be humanely euthanized when they lose a previously specified percent of body weight. If the challenge virus is lethal for mice, the proportion of the animals that survive is recorded over the course of 14 days. The ability of passively transferred antibody to prevent weight loss and/or death, compared to a mock-treated group, demonstrates the efficacy of the MAb and the breadth of activity of the MAb is demonstrated against group 1 and 2 challenge viruses.¹⁹

Generally, the next step is an evaluation of protective efficacy in ferrets. Ferrets are a favoured model for influenza virus research because they develop clinical signs of disease that resembles human influenza, including sneezing, elevated body temperature and rhinorrhoea. Some influenza viruses cause significant weight loss, neurologic signs such as hind limb paralysis and ataxia in ferrets that necessitate euthanasia. However, passively transferred human MAbs have a very short half-life in ferrets,^66,67 in contrast to mice, in which a single dose of hMAb can protect mice from death on day 8–12 post-infection.^3,38,68 Some investigators administer additional doses of hMAbs to ferrets if survival from lethal infection is the desired endpoint.

Passive transfer of strain-specific head (7B2, PY102 & 4C04), broadly reactive head (4G05 & 1F05) and broadly reactive stem (F16 & 2G02) binding antibodies have been shown to confer significant protection from lethal influenza virus infections in mice.^55,56 In both prophylaxis and treatment studies, administration of a MAb with neutralising and haemagglutination inhibition activity conferred better protection compared to MAbs lacking neutralising and haemagglutination inhibition activity.²⁴ We recently reported an evaluation of MEDI8852, a novel optimised antibody developed from a broadly reactive antibody derived from B cells of a donor, that had exceptional breadth of reactivity to all 18 HA subtypes, including group 1 and 2 viruses with pandemic potential.^37,38 We evaluated whether this MAb was effective for prophylaxis and therapy against representative group 1 (H5N1) and group 2 (H7N9) influenza viruses in mice and ferrets, compared to an irrelevant control hMAb and oseltamivir. The hMAb was administered to mice and ferrets by intraperitoneal injection at varying doses, 24 hours prior to intranasal infection with H5N1 and H7N9 viruses for prophylaxis, and 24, 48, and 72 hours post-infection for treatment. We found that MEDI8852 was effective for prophylaxis and treatment of H7N9 and H5N1 infection in mice, with a clear dose-dependent response and treatment with MEDI8852 initiated 1, 2, or 3 days post-infection was superior to oseltamivir for H5N1 virus infection. Furthermore, MEDI8852 alone was effective treatment for lethal H5N1 infection in ferrets compared to oseltamivir and the irrelevant hMAb. Notably, the combination of MEDI8852 with oseltamivir was better than oseltamivir alone. MEDI8852 or oseltamivir alone early in infection were equally effective for H7N9 infection in ferrets while the combination yielded similar protection when treatment was delayed. Thus, MEDI8852, alone or with oseltamivir, shows promise for prophylaxis or therapy of group I and II influenza A viruses with pandemic potential.³⁸

Because treatment strategies capable of interrupting the spread of influenza through sustained human-to-human transmission would limit the public health burden, we asked whether MEDI8852 could be used to protect ferrets from infection by influenza H1N1pdm09 through airborne transmission. For this study, naive ferrets received MEDI8852 or an irrelevant hMAb prior to respiratory contact exposure to ferrets that were experimentally infected with an H1N1pdm09 virus. We found that administration of MEDI8852 to naïve contact ferrets protected them from airborne transmission of the H1N1pdm09 virus, a unique finding among influenza-specific MAbs.³⁸

To accompany systemic signs of disease, the titre of challenge virus in the respiratory tract is an obvious choice for a virologic endpoint that is commonly used in vaccine studies.^{3,30,37,62,68} However, even when there is a remarkable difference in morbidity and mortality between MAb-treated and control animals, the difference in lung virus titres can be very modest.^3,68 Using immunohistochemical staining for influenza viral antigen in lung tissues, we have demonstrated that viral antigen is highly localised to large airways in MAb-treated animals and the lung parenchyma is spared, compared to the widespread distribution of viral antigen in animals that receive an irrelevant control MAb or saline.³ The absence of viral antigen in the parenchyma correlates well with reduced morbidity and mortality in MAb-treated animals and these findings are much more striking than the modest reduction in virus titres in the lungs. We infer that this discrepancy occurs because the partitioning of virus noted by immunohistochemistry to cells lining the large airways versus the parenchyma is lost when lung tissue is homogenized. Virus titres in lung homogenates reflect the amount of virus present in the airways as well as the parenchyma, but morbidity and mortality are associated with pneumonia rather than bronchial infection. Thus, assessment of lung virus titres should be accompanied by histologic and immunohistochemical staining to get a complete picture of the extent of pulmonary infection.

Virus titres in the lungs of mice are ideally assessed in subsets of mice that are sacrificed at the peak of viral replication and also at a later time point when virus clearance occurs. The practical consequence is that the course of viral infection cannot be followed in individual animals. In vivo imaging of animals infected with a bioluminescent reporter virus allows sequential assessment of infection in the same mouse rather than in subsets of mice from the treatment groups.⁶⁹ This approach was used to assess the effect of prophylaxis with a head-specific or a stem-specific MAb, or treatment with the head-specific MAb in mice that were challenged with a reverse-genetics derived bioluminescent reporter virus (H1N1pdm09-NLuc). As was previously demonstrated in prophylaxis studies, administration of the MAbs prior to infection did not completely prevent infection but bioluminescence was significantly lower by day 4 post-infection and remained lower in mice that received MAb prophylaxis than in mice that received an irrelevant antibody.⁶⁹ Similarly, in the immunotherapy group, protection was observed regardless of the timing of MAb administration (24 h or 72 h post-infection). Chest bioluminescence peaked at day 4 post-infection, but a day later, a statistically significant reduction in bioluminescence was evident in the MAb-treated mice compared to mice that received the irrelevant antibody. Furthermore, the in vivo imaging technology demonstrated distinct patterns of bioluminescent signal kinetics in animals from the control, prophylaxis and treatment groups.⁶⁹

Investigating the mechanism of action of MAbs

The in vivo mechanism of action of bNAbs can be evaluated in mouse models as we recently reported using two bNAbs CR6261 and CR9114.⁶⁸ CR6261 was isolated from a phage display library derived from seasonal influenza vaccinees. This antibody showed broad in vitro neutralizing activity against H1, H2, H5, H6, H8, and H9 influenza subtypes. However, within the H2 subtype, CR6261 showed reduced or limited neutralising activity against human isolates compared to avian viruses.⁷⁰ CR9114¹⁷ displayed binding to representative viruses from 14 influenza A subtypes and influenza B viruses, but had reduced binding to human H2N2 viruses.¹⁷ The crystal structures of CR6261 bound to H1 and H5 HA and CR9114 bound to H3, H5, and H7 HA have been solved and revealed that both antibodies bind the HA stem.^17,71 Importantly, while CR6261 displayed reduced in vitro activity and CR9114 had limited binding to human H2 viruses,^70,72,73 neither antibody had been evaluated for in vivo activity against H2 influenza viruses. Thus, we investigated the prophylactic efficacy and protective mechanisms of CR6261 and CR9114 against representative human- and animal-origin H2 viruses in mice.⁶⁸

We first demonstrated that prophylaxis with CR6261 and CR9114 protected mice against lethal H2 influenza virus challenge, irrespective of the in vitro neutralising activity of the antibodies. To further explore the mechanism of in vivo protection in the absence of in vitro neutralisation of human H2N2 viruses by CR9114, we performed studies with engineered variants of CR9114 with mutations to reduce complement binding (KA) or complement binding and Fc receptor (FcR) binding in human and monkey cells (LALA). All antibody-treated mice survived challenge, while less than 40% of vehicle-treated animals survived. However, weight loss in the CR9114- and CR9114-KA-treated groups was minimal, while the CR9114-LALA-treated mice showed greater and prolonged weight loss of 12 to 15%, with recovery starting on day 8. These findings suggested that FcR-dependent mechanisms contributed to the in vivo efficacy of CR9114 against human H2N2 virus challenge. As both CR9114-KA and CR9114-LALA reduced complement binding, we sought to verify that complement did not contribute to protection in vivo and performed prophylaxis studies in complement-deficient DBA mice, which are highly sensitive to influenza challenge due to a deficiency in complement factor C’5 of the complement pathway.^64,74 All DBA mice treated with the irrelevant antibody or vehicle rapidly lost weight and succumbed to infection, while all of the mice treated with CR9114 or CR6261 survived challenge with only transient weight loss. These findings demonstrated that the protection mediated by CR6261 and CR9114 was not dependent on the complement pathway.⁶⁸

Because FcR-induced antibody-mediated cell cytotoxicity (ADCC) and antibody-mediated cell phagocytosis (ADCP) are effector mechanisms of bNAbs^55,56,75,76 and our experiments with CR9114-LALA suggested that reduced FcR binding resulted in enhanced morbidity⁶⁷, we proceeded to evaluate CR9114 and CR6261 in Fcer1g/ mice, that lack FcRI, FcRIII, and FcRIV, but express the inhibitory receptor FcRII.^55,75 Following virus challenge, all of the CR9114- and isotype control-treated mice rapidly lost weight and succumbed to infection by day 7. All of the CR6261-treated mice lost weight rapidly; however, 2 of 8 animals recovered from weight loss and survived. These experiments demonstrated that the in vivo protection conferred by CR6261 and CR9114 against human H2N2 virus challenge was mediated by FcR interactions and importantly, demonstrated the importance of in vivo studies to demonstrate the efficacy of HA stem-binding antibodies.

Conclusions

Influenza viruses are pathogens that cause annual infections and are associated with significant morbidity and mortality, with significant global economic and social impact. The emergence of novel influenza viruses resulting in pandemics and spill-over of zoonotic influenza viruses has happened in the past with devastating outcomes and the potential for such future events is of particular concern. There is therefore an urgent need to develop strategies to protect individuals and populations from such threats.

Vaccination is a successful and cost-effective strategy to control many viral and bacterial pathogens. However, the antigenic plasticity of influenza viruses due to antigenic shift and drift requires annual update of seasonal influenza vaccines. Current influenza vaccines are effective in reducing morbidity and mortality caused by viruses that are antigenically closely related to the strains included in the vaccine. In contrast, a universal influenza vaccine would induce protective immune responses against most, if not all influenza viruses. A number of strategies to develop a universal influenza vaccine are being pursued including targeting cellular immune responses to conserved epitopes, sequential immunisation with antigens expressing distinct head and conserved stem domains and computationally designed HAs to protect against past, present and potentially future influenza virus strains.⁷⁷

There is limited use of passively administered influenza virus-specific antibodies for prophylaxis or therapy. Concerns about the use of antibodies include cost, storage, shelf-life and the development of anti-drug antibodies. However, antibody-based strategies would be beneficial in specific scenarios where influenza immunisation may be inadequate, such as when there is insufficient time for the vaccine to induce a protective immune response (post-exposure prophylaxis in contacts of infected patients), in the event of a pandemic where a vaccine is not yet available, when influenza vaccine cannot be administered (e.g. to very young infants), and to prevent transmission in localised outbreaks in high-risk settings (aged-care facilities and hospitals). Several broadly cross-reactive monoclonal antibodies are in clinical development. In animal models antibody treatment has been shown to confer significantly better protection than oseltamivir alone and has been shown to augment the protective efficacy of oseltamivir, when used in combination.³⁸ If these antibodies are safe and effective in clinical trials, they could be stockpiled by hospitals and governments to serve as a first line of defence during outbreaks.

Funding Statement

This work at the Melbourne WHO Collaborating Centre for Reference and Research on Influenza was supported by the Australian Government Department of Health.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.