Differences in the ease with which mutant viruses escape from human monoclonal antibodies against the HA stem of influenza A virus

Abstract

Background

Broadly protective human monoclonal antibodies that recognize the conserved epitopes in the HA of influenza A virus are being developed as therapeutic agents. Emergence of resistant viruses must always be considered when developing therapeutic agents against influenza.

Objectives

We examined human hetero-reactive mAbs against the HA stem of influenza A virus for the ease with which escape mutant viruses emerged.

Study design

We attempted to generate the mutant viruses escaped from the hetero-reactive anti-HA stem antibodies. We also evaluated their protective efficacy, binding affinity, and epitopes.

Results

We obtained several human monoclonal antibodies (mAbs) that react with the HA of different HA subtypes of influenza A virus belonging to group 1. Upon attempting to generate escape mutant viruses, we found that the ease with which such viruses emerged differed among the mAbs; viruses barely escaped from two of the mAbs (clones S9-3-37 and F20C77), whereas escape from the third mAb (clone F5B7) occurred readily. Comparisons of the mAbs revealed that the HA stem epitopes, in vitro neutralization potency, binding affinity to H1-HA, and protective efficacy against lethal challenge with H1N1pdm09 virus were all comparable.

Conclusions

These results demonstrate the importance of determining the ease with which escape mutant viruses emerge when evaluating anti-HA stem antibodies as antiviral agents during preclinical testing.

Background

Hemagglutinin (HA) is one of the major protective antigens of influenza A virus. Phylogenetically, HA is classified into 18 subtypes (H1 through H18) that are separated into two groups: group 1 (H1, H2, H5, H6, H8, H9, H11, H12, H13, H16, H17, and H18) and group 2 (H3, H4, H7, H10, H14, and H15) ^{1, 2}. The trimeric HA protein consists of a head region and a stem region ³. The antigenicity of the HA head region is highly variable, but that of the HA stem is conserved among group members, meaning that most anti-HA stem antibodies recognize several HA subtypes. Such antibodies that possess anti-viral activity have potential as therapeutic agents to treat patients infected with influenza viruses including seasonal H1N1pdm09, H3N2, zoonotic H5N1, and H7N9 viruses. Accordingly, many broadly protective human monoclonal antibodies (mAbs) against the HA stem have been reported and evaluated as antiviral agents ^4–11. These antibodies inhibit viral growth in vitro by inhibiting both the conformational change in HA that is required for viral membrane fusion ¹² and virus particle release from virus-infected cells ⁸. For effective in vivo protection, some anti-HA stem antibodies require interactions between the Fc region of IgG and Fcγ receptors (FcγRs) to activate effector cells such as natural killer (NK) cells, macrophages, and neutrophils ^13–15. The activated NK cells, macrophages, and neutrophils suppress virus propagation via antibody-dependent cellular cytotoxicity, antibody-dependent cellular phagocytosis, and antibody-dependent neutrophil-mediated phagocytosis, respectively ^{16, 17}.

Novel classes of anti-influenza agents are needed because of concerns about the emergence of NA inhibitor-resistant viruses, including the highly pathogenic influenza H7N9 viruses ^18–20. The anti-HA stem mAbs that react with several subtypes of the HA represent one of the promising approaches in the development of novel anti-viral agents. However, the emergence of escape mutant viruses from such antibodies is also a major consideration. Of the many reports describing antibodies against the HA stem, some included attempts to obtain escape mutant viruses in vitro by passaging virus in the presence of each mAb. No mutant viruses escaped from F10, D8, A66 ²¹, FE43, or FB110 ²², after 1-3 passages, whereas mutant viruses escaped from CR6020 ²³, CR8020 ²⁴, CR8043 ²⁵, 045-051310-2B06, 042-100809-2F04, S6-B01 ²⁶, PN-SIA28 ²⁷, S9-1-10/5-1 ⁸, 1417infE21, or 1417infC10 ²⁸ within approximately 10 passages. Since these results were obtained under different experimental conditions, we still do not know whether mutant viruses can escape from anti-HA stem mAbs with ease or difficulty. Identifying whether there are differences in the ease with which mutant viruses escape from different neutralizing anti-HA stem antibodies would aid in the development of better therapeutic anti-HA stem mAbs.

Objectives

We examined three human hetero-reactive mAbs against the HA stem of influenza A virus for the ease with which escape mutant viruses emerged; we also evaluated their protective efficacy, binding affinity, and epitopes.

Study design

Ethics and biosafety statements.

Human blood was collected according to protocols that were approved by the Research Ethics Review Committee of the Institute of Medical Science, the University of Tokyo. Written informed consent was obtained from all participants.

All experiments with mice were performed in accordance with the University of Tokyo’s Regulations for Animal Care and Use and were approved by the Animal Experiment Committee of the Institute of Medical Science, the University of Tokyo.

Cells.

Madin-Darby canine kidney (MDCK) cells were maintained in Eagle’s minimal essential medium (MEM) containing 5% newborn calf serum (NCS). Human embryonic kidney 293T cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% FCS. These cells were incubated at 37 °C under 5% CO₂. Expi293F (Thermo Fisher Scientific) or ExpiCHO cells (Thermo Fisher Scientific), maintained in Expi293 or ExpiCHO expression medium (Thermo Fisher Scientific), respectively, were incubated on an orbital shaker platform rotating at 125 rpm at 37 °C under 8% CO₂.

Hybridomas.

The hybridomas used in this study were obtained previously ⁸ by fusion of peripheral blood mononuclear cells (PBMCs), which were isolated from healthy volunteers vaccinated with the H5N1 pre-pandemic vaccine, with SPYMEG cells (MBL) ²⁹.

Expression of monoclonal human IgG.

Plasmids expressing human IgG consisted of each VH and VL sequence of the hybridoma clone, and the constant gamma heavy (IgG₁) and kappa or lambda light chain coding sequences were constructed using a Mammalian PowerExpress System (TOYOBO). The determined nucleotide sequences of the variable region were analyzed and compared to sequences in the NCBI database by using the IgBlast software (http://www.ncbi.nlm.nih.gov/igblast/). Each plasmid encoding heavy and light chains was transfected into Expi293F or ExpiCHO cells by using ExpiFectamine 293 or ExpiFectamine CHO (Thermo Fisher Scientific) according to the manufacturer’s protocol. At 4–12 days post-transfection, the human antibody in the culture medium was purified by using a HiScreen MabSelect SuRe LX column (GE Healthcare) and the automated chromatography system ÄKTA pure 25 (GE Healthcare). The concentration of the purified antibodies was measured by using a Pierce BCA Protein Assay Kit (Thermo Fisher Scientific).

Reactivity of human mAbs.

293T cells were transfected with each plasmid expressing the H1- to H18-HA ⁸ by use of Trans-IT 293 (Takara bio). At 48 h post-transfection, the cells were stained with the indicated human mAb, as described previously²⁸. The broadly reactive mAb FI6v3 ⁷ served as a positive control.

Viruses.

A/California/04/2009 (CA04; H1N1pdm09) and its mouse-adapted strain (MA-CA04) ³⁰ were propagated and titrated in MDCK cells.

Virus neutralization assay.

Purified antibody (50 µg/ml) in quadruplicate was serially two-fold diluted with MEM containing 0.3% bovine serum albumin (BSA-MEM) prior to being mixed with 100 TCID₅₀ (50% tissue culture infectious doses) of the indicated viruses at 37 °C for 30 min. The mixtures were inoculated into MDCK cells and incubated for 1 h at 37 °C. BSA-MEM containing TPCK-treated trypsin was added to each well, and the cells were incubated for 3 days at 37 °C. The cytopathic effect (CPE) was examined, and antibody titers required to reduce virus replication by 50% (IC₅₀) were determined by using the Spearman-Kärber formula.

Selection of escape mutants.

Escape mutants were selected by culturing CA04 in the presence of S9-3-37, F20C77, or F5B7 as described previously ⁸. Selections (n=3) were performed for 4–30 cycles to obtain escape mutants. The open reading frame of the HA of the escape mutants was directly sequenced.

Virus rescue.

Plasmid-based reverse genetics for virus generation using eight RNA polymerase I plasmids encoding the CA04 segments was performed as previously described ³¹. Mutations in the HA gene of CA04 virus were generated by PCR amplification of the respective RNA polymerase I plasmid ³⁰ with primers possessing the desired mutations (primer sequences available upon request). Each rescued virus was propagated in MDCK cells and stored as a stock virus. The titers of the stock viruses were determined in MDCK cells. All constructs, and the HA gene of all rescued viruses, were sequenced to confirm the absence of unwanted mutations.

K_D determination.

K_D was determined by bio-layer interferometry (BLI) using an Octet Red 96 instrument (ForteBio), as reported previously ^{8, 28}. Briefly, recombinant HAs of A/California/07/2009 (H1N1pdm09) (Sino Biological) were loaded onto Ni-NTA biosensors (ForteBio) and incubated with three concentrations of F5B7, S9-3-37, or F20C77. The ratio of k_on to k_off determined the K_D reported here.

Competitive binding assay.

Competitive binding was assessed by using an Octet Red 96 instrument as reported previously⁸. After binding of the recombinant HAs of A/California/07/2009 (H1N1pdm09) to the Ni-NTA biosensors, the sensors were incubated with S9-3-37, F20C77, F5B7, anti-HA stem antibody CR9114 ⁴, or anti-HA head antibody 4-6-19/6 ⁸ (10 μg/ml) for 600 seconds. They were then incubated with a second antibody [S9-3-37, F20C77, F5B7, CR9114, 4-6-19/6, or anti-B-HA 1429C6/3-3 ⁸; 10 μg/ml] for 600 seconds.

Evaluation of the in vivo protective efficacy of the mAbs in mice.

Baseline body weights of 6-week-old female BALB/c mice (Japan SLC) were measured. Four mice per group were intraperitoneally injected with the indicated antibodies at a concentration of 0.6, 3, or 15 mg/kg. One day later, the mice were anaesthetized and inoculated with 10 mouse lethal dose 50 (MLD₅₀) (50 µl) of the MA-CA04 virus. Body weight and survival were monitored daily for 14 days. Mice with body weight loss of more than 25% of their pre-infection values were humanely euthanized.

Statistical analysis.

A non-parametric Kruskal-Wallis test was performed using GraphPad Prism software. P values < 0.01 were considered significantly different. No samples were excluded from the analysis.

Results

Reactivity of nine human monoclonal antibodies to the 18 HA subtypes.

We previously generated and cloned hybridomas derived from human volunteers who had been vaccinated with the influenza H5N1 pre-pandemic vaccine ⁸. As a result of previous screening by using an ELISA with recombinant H1-, H3-, H5-, H7-, and B-HA, we obtained nine clones (S9-3-37, F20C77, S9-4-4/4, 10-5-64/6, F5B7, 4-5-26, S9-1-17/2-4, 4-4-7, and S9-4-31), which recognized both H1- and H5-HA but not H3-, H7-, or B-HA. The nucleotide sequences of the VH and VL regions for these human monoclonal antibodies (mAbs) were analyzed, and the most probable inferred germline progenitors were determined by using IgBlast. For the heavy chain, F20C77 and S9-4-4/4 used the IGHV1-69 germline gene, and S9-1-17/2-4 and 4-4-7 used the IGHV4-39 germline gene (Table 1). The other clones used different germline genes. For the light chain, F5B7 and 4-4-7 used the lambda chain, whereas the other seven clones used the kappa chain (Table 1).

Table 1.

Genetic hallmarks of the nine human mAbs.

mAb	Heavy chain		Light chain
	VH	CDR3	VL	CDR3
S9-3-37	IGHV1-18*01	AREFRTQIVLGYFDWLEGNAFDM	IGKV2-24*01	TQASQFPRT
F20C77	IGHV1-69*01	ARRVIVTEPGTFDS	IGKV3-15*01	QQYNDWPPRYT
S9-4-4/4	IGHV1-69*04	ARAETNTGWYGLLDY	IGKV3-20*01	QQYSRSPT
10-5-64/6	IGHV3-23*01	AKDRETRQWLTGSMDY	IGKV3-11*01	QQRSNWPPIT
F5B7	IGHV4-30-2*01	ARGRLEFGRLSTPYYHGMDV	IGLV2-14*01	SSYTSSALV
4-5-26	IGHV4-31*03	ARDASFWGGKNAFDI	IGKV1-39*01	QQSYDTPQT
S9-1-17/2-4	IGHV4-39*01	ARHVAPGRANVIVVYYFDY	IGKV3-11*01	QQRTNRGGIT
4-4-7	IGHV4-39*02	ARVAWFGGQNWFDP	IGLV2-11*01	CSYAGSYSWV
S9-4-31	IGHV4-59*08	ARLRGGTMLVVLGTTHVFDI	IGKV1-12*01	QQGNSFPWT

We attempted to determine the breadth of reactivity of the mAbs by immunostaining 293T cells that transiently expressed H1- through H18-HA. All nine of the human mAbs tested recognized both H1- and H5-HA as well as several other group 1 HA subtypes (Table 2); however, they failed to bind all group 2 HAs (Table 2). FI6v3, which was previously reported to be broadly reactive to both group 1 and 2 HAs⁷, bound to 16 subtypes of HA other than H11 and H13 under these experimental conditions. These results indicate that all of the nine mAbs primarily recognize HA subtypes that belong to group 1.

Table 2.

Reactivity of the human mAbs with HA expressed in cells

mAb	Group 1												Group 2
	H1	H2	H5	H6	H8	H9	H11	H12	H13	H16	H17	H18	H3	H4	H7	H10	H14	H15
S9-3-37	++a	−	++	++	++	++	−	++	−	++	++	++	−	−	−	−	−	−
F20C77	++	++	++	+	++	+	−	−	−	+	++	++	−	−	−	−	−	−
S9-4-4/4	++	+	++	−	++	−	−	−	−	−	−	++	−	−	−	−	−	−
10-5-64/6	++	++	++	−	−	−	−	−	+	−	−	−	−	−	−	−	−	−
F5B7	++	−	++	−	−	−	−	+	−	−	+	++	−	−	−	−	−	−
4-5-26	++	++	++	++	++	−	++	−	++	++	++	+	−	−	−	−	−	−
S9-1-17/2-4	+	−	++	−	−	−	−	−	−	−	+	+	−	−	−	−	−	−
4-4-7	+	−	+	−	−	−	−	−	−	−	−	+	−	−	−	−	−	−
S9-4-31	++	−	+	−	++	−	−	−	−	−	++	+	−	−	−	−	−	−
FI6v3	+	++	++	+	++	++	−	+	−	++	++	++	+	+	+	+	+	+

^aReactivity of each mAb (1 μg/ml) was stratified according to optical density at 450 nm, ++ (>0.5), + (0.1–0.5), and − (< 0.1).

Neutralization activity of the nine human mAbs.

We next examined whether these nine human mAbs possessed neutralization activity against H1N1pdm09 virus in vitro. Briefly, 100 TCID₅₀ of A/California/04/2009 (H1N1pdm09; CA04) was incubated with serially two-fold-diluted mAb before infection of MDCK cells. The presence of CPE was examined 3 days after infection. 4-5-26 and S9-1-17/2-4 failed to suppress the replication of CA04 at the highest concentration tested (50 μg/ml) (Table 3). The other seven mAbs inhibited CA04 propagation at similar 50% inhibitory concentration (IC₅₀) values, which ranged from 5.3–10.5 μg/ml (Table 3). We then compared the ease with which mutant viruses escaped from each of the seven neutralizing mAbs by passaging CA04 in the presence of various concentrations of mAb. We deemed an escape mutant virus to have emerged when a CPE appeared in the cells inoculated with viruses that were pre-treated with neutralizing mAbs at a concentration of 250 μg/ml before inoculation. Mutant viruses escaped from F5B7 appeared quickly after 4–5 passages, whereas more than 11 passages were required to obtain mutant viruses that escaped from S9-3-37 and F20C77 (p=0.06 and p<0.01; non-parametric Kruskal-Wallis test) (Table 3). Between 5 and 28 passages were required to obtain escape mutants from the other four clones. From these results, we conclude that it is relatively hard for H1N1pdm viruses to escape from S9-3-37 and F20C77 but easy for them to escape from F5B7. Therefore, we further characterized S9-3-37 and F20C77 as mAbs that viruses find difficult to escape from and F5B7 as a mAb from which escape mutants are easily generated.

Table 3.

Properties of human mAbs against H1N1pdm virus

mAb	IC50 value (μg/ml)	Passage number until appearance of escape mutants in triplicate
		Clone 1	Clone 2	Clone 3
S9-3-37	5.3	22	11	12
F20C77	8.8	12	24	30
S9-4-4/4	6.3	8	6	15
10-5-64/6	8.8	5	22	28
F5B7	8.8	5	4	4
4-5-26	>50	–a	–	–
S9-1-17/2-4	>50	–	–	–
4-4-7	10.5	18	8	6
S9-4-31	8.8	20	5	18

^aNot tested

Comparisons of in vivo protective efficacy.

We next asked whether the ease with which escape mutant viruses emerge affect the in vivo protective efficacy of the mAbs. To evaluate the protective efficacy of the three mAbs, 4 mice per group were intraperitoneally administered 0.6, 3, or 15 mg/kg of S9-3-37, F20C77, or F5B7 or with 15 mg/kg of the anti-B-HA 1429C6/3-3 as a negative control. At 1 day-post administration, these mice were intranasally inoculated with 10 MLD₅₀ (50% mouse lethal dose) of mouse-adapted CA04. The body weight changes of mice treated with each mAb at each concentration were similar (Fig. 1). All (4/4), 75% (3/4), or 50% (2/4) of the mice that received 15, 3, or 0.6 mg/kg of S9-3-37 survived for the 2-week observation period after the challenge infection (Fig. 1). One of the 4 mice that received 3 mg/kg of F20C77 died. Three of the four mice (75%) that received 15, 3, or 0.6 mg/kg of F5B7 survived for the 2-week observation period. We then analyzed the nucleotide sequence of the HA detected in the lungs of these seven dead mice (3, 1, and 3 mice from the S9-3-37-, F20C77-, and F5B7-treated groups, respectively) and found no amino acid substitutions in the HA. All mice that received 1429C6/3-3 died within 5 days of challenge. These results demonstrate that S9-3-37, F20C77, and F5B7 show similar protective efficacy against lethal challenge infection with H1N1pdm virus, indicating that the ease of emergence of escape mutant viruses does not affect in vivo protective efficacy.

Figure 1. In vivo protective efficacy in mice.

Four mice per group were intraperitoneally injected with the indicated antibodies at 0.6, 3, or 15 mg/kg. One day later, the mice were challenged with 10 MLD₅₀ of MA-CA04 (H1N1pdm09). Body weight and survival were monitored daily for 14 days. 1429C6/3-3 at 15 mg/kg, which recognizes influenza B virus HA, served as a negative control. Mouse body weights are expressed as mean ± SD.

Comparisons of binding affinity.

To examine why the number of passages required for the emergence of escape mutant viruses from S9-3-37, F20C77, and F5B7 differed, we compared the binding affinities of the three mAbs to recombinant H1-HA. We found that all three mAbs showed similar k_on values (range = 4.46–7.89 × 10⁵), whereas he k_off values of all three mAbs were less than 1 × 10⁻⁷, resulting in identical K_D values (Table 4). These results indicate that the binding affinities of S9-3-37, F20C77, and F5B7 to H1-HA are comparable and play no role in the ease of emergence of escape mutant viruses.

Table 4.

Binding kinetics of human mAbs with recombinant HAs.

mAb	kon (1/Ms)	koff (1/s)	KD (M)
S9-3-37	7.89 × 105	< 1 × 10−7	< 1 × 10−12
F20C77	5.75 × 105	< 1 × 10−7	< 1 × 10−12
F5B7	4.46 × 105	< 1 × 10−7	< 1 × 10−12

Epitopes of S9-3-37, F20C77, and F5B7.

To estimate the epitopes of S9-3-37, F20C77, and F5B7, we performed a competitive binding assay using recombinant H1-HA. Briefly, H1-HA immobilized to a biosensor was saturated with the anti-HA stem mAb CR9114 or the anti-HA head mAb 4-6-19/6, S9-3-37, F20C77, or F5B7. Each biosensor was additionally exposed to the anti-B-HA mAb 1429C6/3-3 as a negative control, CR9114, 4-6-19/6, S9-3-37, F20C77, or F5B7. S9-3-37, F20C77, and F5B7 competed with CR9114 but not with 4-6-19/6 for binding to H1-HA, and vice versa (Fig. 2). 1429C6/3-3 did not bind to H1-HA. These results show that S9-3-37, F20C77, and F5B7 recognize the HA stem.

Figure 2. Competitive binding assay.

Competitive binding was performed using the Octet system. The H1-HA immobilized to the biosensor was saturated with anti-HA stem CR9114, or anti-HA head 4-6-19, S9-3-37, F20C77, or F5B7 (from 0 s to 600 s). Additional binding of the indicated human mAbs was measured (from 600 s to 1200 s). When the antibody bound to the H1-HA, the shift value was increased.

To narrow down the epitopes of S9-3-37, F20C77, or F5B7 by identification of escape mutations, we analyzed the nucleotide sequence of the HA segment of each escape mutant obtained in Table 3 by means of direct sequencing. We found that the mutant viruses that escaped from S9-3-37 possessed the N11S, G155E, and S190R, the N11Y, K119N, and L425M, or the N11Y and K119N mutations in their HA, whereas the mutant viruses that escaped from F20C77 possessed the N11Y and N125D, the T13A, I96T, and N156D, or the N11S and K119N mutations (Table 5). All three mutants that escaped from F5B7 acquired the N11S and K119N mutations (Table 5). The N11S, N11Y, and T13A mutations abolished the N-linked glycosylation site at positions 11–13, whereas the K119N mutation formed a novel potential N-linked glycosylation site. We mapped these mutations onto the 3D structure of the H1-HA trimer (PDB; 3LZG) by using the molecular graphics system PyMOL. The mapping revealed that the N11S, N11Y, and T13A mutations were located on the HA stem, and the K119N, N125D, N156D, G155E, and S190R mutations were located on the HA head (Fig. 3). The I96T and L425M mutations were not exposed on the protein surface and were located on a trimer interface. Taken together with the results from competitive binding assay, our findings suggested that the N11S, N11Y, and T13A mutations were likely to be key mutations for escape. Therefore, we prepared mutant viruses possessing the N11S (HA-N11S), N11Y (HA-N11Y), or T13A (HA-T13A) substitution in their HA by using reverse genetics and tested the neutralizing activity of S9-3-37, F20C77, and F5B7. The HA-N11Y and -N11S viruses were not neutralized by S9-3-37, F20C77, or F5B7 at the highest concentration tested (50 μg/ml) (Table 6). The HA-T13A virus was neutralized by S9-3-37 with an IC₅₀ value of 2.6 μg/ml but not by F20C77 or F5B7 at 50 μg/ml (Table 6). These results indicate that these three mAbs recognize a similar epitope that includes asparagine at position 11 in the HA stem.

Table 5.

Amino acid substitutions in the HA of viruses passaged in the presence of mAbs.

Passaged in the presence of	Clonea	Passage number	Amino acid substitutions in HA
S9-3-37	1	22 b	N11S, G155E, S190Rc
	2	11	N11Y, K119N, L425M
	3	12	N11Y, K119N
F20C77	1	12	N11Y, N125D
	2	24	T13A, I96T, N156D
	3	30	N11S, K119N
F5B7	1	5	N11S, K119N
	2	4	N11S, K119N
	3	4	N11S, K119N

^aClones 1–3 were independently obtained.

^bThe passage number at which escape mutants were obtained.

^cH1 numbering.

Figure 3. Mapping of mutations found in escape mutants.

Amino acid substitutions found in the HA of three independent mutant clones that escaped from S9-3-37, F20C77, or F5B7 were mapped onto the 3D structure of the H1-HA trimer (PDB; 3LZG) by using the molecular graphics system PyMOL. Yellow indicates amino acids involved in receptor binding. Mutations are shown with H1 numbering.

Table 6.

IC₅₀ values (μg/ml) of the human mAbs against mutant viruses

mAb	Wild-type	HA-N11Y	HA-N11S	HA-T13A
S9-3-37	3.7	>50	>50	2.6
F20C77	4.4	>50	>50	>50
F5B7	8.8	>50	>50	>50

Discussion

Because the development of therapeutic antibodies is expensive and time-consuming, such antibodies should ideally maintain their effectiveness during treatment. Here we obtained several hetero-reactive mAbs against the HA of influenza A virus belonging to group 1. Using the mAbs that possessed neutralization activity, we generated escape mutant viruses in vitro by passaging viruses under the selective pressure of the neutralizing mAb. This experiment revealed that the ease with which escape mutant viruses emerged differed among the human anti-HA stem mAbs: viruses found it difficult to escape from S9-3-37 and F20C77, but comparatively easy to escape from F5B7. This finding suggests that escape mutant viruses would shortly emerge if patients were treated with an anti-HA stem mAb such as F5B7, thereby demonstrating its lack of suitability as a treatment option. One reason why we see a difference in the ease of emergence of escape mutants among mAbs may be as follows: If a single amino acid change disrupts an epitope of an antibody and does not affect the functions of HA negatively, an antigenic escape mutant against the antibody would be readily selected. However, if a mutation that abolishes the binding of an antibody is also detrimental to the functions of HA, an antigenic escape mutant would either not selected or additional mutations that compensate for the detrimental mutation would need to be introduced, thus requiring more passages to allow for the accumulation of such compensatory mutations. The potential compensatory mutations N125D and N156D have been found in circulating H1N1pdm09 viruses, although viruses possessing these mutations did not become dominant until the 2017–2018 influenza season ^32–34. The presence of these substitutions in a virus might influence the ease of generation of escape mutant viruses. To develop therapeutic antibodies for clinical use, we need to select anti-HA stem mAbs with the lowest propensity for the generation of escape mutant viruses.

When developing therapeutic antibodies for infectious diseases, the emergence of antibody-resistant pathogens is an important consideration. The method we used to examine the ease of emergence of antigenic escape mutants from antibodies is not quantitative, but it does suggest tendencies. It will be important to establish quantitative assays to measure such properties as therapeutic antibodies continue to be developed.

Among the hetero-reactive human monoclonal antibodies reported here, mutant viruses that escaped from S9-3-37 and F20C77 rarely emerged, whereas those that escaped from F5B7 readily appeared. Since the generation of escape mutant viruses is laborious, which may be why most reports do not include such experiments, we searched for a surrogate marker to select the mAb with the lowest propensity to generate escape mutant viruses. However, both kinds of mAbs recognized a similar epitope in the HA stem, possessed similar neutralization potency in vitro, showed similar binding affinity to H1-HA, and similarly protected mice from lethal challenge infection with H1N1pdm09 virus, demonstrating that these factors are not affected by the ease with which escape mutant viruses emerge. Thus, there may be no appropriate surrogate for generating escape mutant viruses when attempting to select mAbs with the lowest likelihood of yielding escape mutant viruses. However, since co-crystal structures of S9-3-37, F20C77, or F5B7 with HA have not been elucidated, epitope mapping at the atomic level may yet find a difference between two kinds of mAbs.

HA-N11S/Y and/or HA-T13A, which eliminate the N-glycosylation site at positions 11–13 (NST), abolished the neutralization activity of three mAbs tested (Table 6). Removal of the N-glycosylation site at positions 11–13 decreased the pH stability of the virus, triggering membrane fusion at a higher pH ^{35, 36}. Thus a structural alteration due to the loss of glycosylation and possibly HA instability affected antibody binding to HA. The HA stem is a promising target for antivirals and vaccine antigens. Further studies of escape mechanisms are essential to the development of well-designed anti-influenza strategies.