Neutralizing antibodies provide immediate treatment options for influenza pandemic emergency, particularly, for acutely exposed people while more time-consuming developments of vaccines and new drugs are ongoing 
. Despite numerous epidemics and two major pandemics, the ectodomain of M2 (M2e) protein has shown remarkable conservation; namely, it remains essentially unchanged since the first influenza strains were isolated in 1918 
. The M2 protein of influenza A virus has thus become a target for both vaccine and antibody development to achieve broad protection from infection of influenza A variants 
. However, current influenza vaccines have not achieved a significant anti-M2 humoral response due to low immunogenicity of M2e and fewer number of M2 molecules presented on influenza virus particles 
. A multivalent M2e vaccine is in clinical trial which may show great promise 
, but perhaps not be effective for poorly responding populations such as the elderly, very young children and immuno-compromised individuals 
. Passive immunization with monoclonal antibodies would complement such a vaccine, allowing the treatment of disadvantaged people 
. Indeed, a number of M2e-specific antibodies have been generated in recent years that showed anti-influenza A virus activities both in prophylactic and therapeutic settings. The mechanisms of action by these antibodies are mostly by ADCC or CDC, targeting infected cells but not directly neutralizing the viruses, which could limit their efficacies in eliminating infections 
Single-domain antibody (VHH) fragments are emerging as new versatile reagents for the diagnosis and also the therapy of infectious diseases such as RV-induced diarrhea, HIV, and foot-and-mouth disease 
. In comparison to conventional antibodies, one of the unique features of VHH is that it is particularly suitable for binding to the pocket or cleft of targeted antigen owing to its small size and long CDR3 
. Therefore, VHHs were chosen to target tetrameric M2 ion channel. Synthetic VHH phage display libraries were constructed using universal framework cAbBCII10, which is expressed well, stable in bacteria, and has the plasticity allowing transfer of donor antigen binding sequences without compromising their binding capabilities 
. VHH libraries with variable CDR3 length (9 to 20 amino acids) were independently constructed and mixed for panning against recombinant full length M2 protein. VHH phages after four rounds of panning were further selected based on their binding to M2 protein on the cell membrane of influenza infected MDCK cells. The candidate VHHs were then evaluated by plaque inhibition assay. This screening/selection approach was designed to isolate VHHs that bind native M2, the tetramer structure essential for its ion channel activity.
M2-7A, one of the six VHH candidates, showed strong affinity not only for the recombinant full length M2 protein but also the native M2 protein on the virion (). However, it failed to bind a 23-amino acid synthetic M2e peptide (). Flow cytometry and immunofluorescence staining showed M2-7A also recognized M2 expressed on the cell surface ( and ). Our results demonstrated that M2-7A specifically recognized native M2. We further observed that M2-7A inhibited replication of both A/Hong Kong/8/68 (H3N2, amantadine-sensitive) and A/PR/8/34 (H1N1, amantadine-resistant) viruses (). The finding is, to our knowledge, the first report for an antibody that is capable of targeting both wild-type and amantadine-resistant influenza A viruses in vitro
. In a mice challenge model, M2-7A was able to protect mice from a lethal dose of A/PR/8/34 when given 1 or 2 days post-infection. The two-day consecutive M2-7A treatment was more efficacious than a single dose treatment (), in support of findings by others 
. Different from many conventional anti-M2e antibodies whose antiviral activities in vivo
are mediated through ADCC or CDC, the protection by M2-7A lack of Fc fragment is likely through blockage of M2 ion channel on the virion and influenza-infected cell surface. The in vivo
efficacy of M2-7 led to the reasonable prediction that M2-7A-Fc would be more potent than traditional anti-M2 antibodies, owing to the combination of neutralization capability with ADCC or CDC activity.
Other factors such as binding affinity and half-life could also affect in vivo
efficacy of M2-7A, which has a moderate binding strength (Kd
39.5 nM) for M2 protein. Further affinity maturation based on M2-7A CDRs can be done as described previously 
. Alternatively, multivalent M2-7A can be engineered, for example, a peptide linker can be placed between two monomers to generate a bivalent VHH with high affinity or avidity and half-life, similar to an anti-vWF nanobody (camel VHH, Ablynx) on phase II clinical trial.
M2-7A was generated with the framework that has high degree of human antibody sequence homology, thus expected low immunogenicity when used in humans. Due to the small size, M2-7A can be easily humanized to lower or avoid immunogenicity. No B- or T-cell responses have been detected in mice and baboons treated with camel VHHs. In addition, camel VHHs against RANKL and vWF have already successfully passed phase I clinical trial (Ablynx NV), indicating that they were not immunogenic to human.
Influenza A virus M2 protein is a 97-residue single-pass membrane protein with its amino termini towards the outside and carboxyl termini inside the virion. It is a homotetramer in its native state with four transmembrane helices forming a channel for proton conductance. The opening of M2 ion channel is essential for viral uncoating inside the host cell. The detailed mechanism of M2 ion channel opening and closing remains to be resolved, despite a general agreement on its channel pore structure and function 
. Transmembrane residues, His37 and Trp41, are critical for channel activity serving respectively as a pH sensor and gate. Amantadine, an antiviral drug against influenza A virus, affects the opening of M2 ion channel, and the resistance to amantadine occurs with high frequencies. Although amantadine-resistant mutations are mainly in the transmembrane pore, amantadine binding was surprisingly found at the channel's lipid-exposed outer surface 
. A possible explanation is that binding by amantadine causes conformational change of distant M2 transmembrane pore, affecting the open and close of the channel, whereas mutations suppress the drug action. Our study has demonstrated that M2-7A protected M2-expressing cells from pH shock-induced cell death, and the protection is equally effective for both M2wt- and M2mu (amantadine-resistant)-expressing cells (), strongly indicating at least a partial blockage of proton influx by M2-7A. Although the exact mechanism of the blockage remains to be investigated, we hypothesized that the binding of the extracellular portion by M2-7A causes a conformational change of the M2 ion channel transmembrane helices, thereby, preventing the opening of the channel for proton influx. The binding epitope of M2-7A might be conformational since our selection strategy against native M2 protein was favorable in generating conformation-dependent VHHs. This notion was supported by the ELISA data showing the weak binding of M2-7A to M2e peptide conjugate. Moreover, long CDR3 of M2-7A could form the structure that fits the pocket or cleft formed by extracellular regions of M2 tetramer. Interestingly, CDR3 of M2-7A has 18 amino acids (IH
TKQNRTTY) with multiple histidine residues and showed no homology to other five anti-M2 VHHs. The roles of these residues in binding to and blocking M2 ion channel protein requires further investigation.
In conclusion, this study demonstrated the feasibility of generating a novel class of antiviral drugs using synthetic VHH libraries and utility of function-based screening/selection approach for neutralizing antibodies. VHH M2-7A, showed preferred binding to native M2, and potent neutralizing activities for both wild-type and amantadine-resistant influenza A viruses, likely through direct interference with M2 ion channel function. In vivo efficacy of M2-7A warranted its further study as a clinical candidate for broad protection against influenza infection in human.