|Home | About | Journals | Submit | Contact Us | Français|
The conserved influenza virus hemagglutinin (HA) stem domain elicits cross-reactive antibodies, but epitopes in the globular head typically elicit strain-specific responses because of the hypervariability of this region. We isolated human monoclonal antibody 5J8, which neutralized a broad spectrum of 20th century H1N1 viruses and the 2009 pandemic H1N1 virus. Fine mapping of the interaction unexpectedly revealed a novel epitope between the receptor-binding pocket and the Ca2 antigenic site on HA. This antibody exposes a new mechanism underlying broad immunity against H1N1 influenza viruses and identifies a conserved epitope that might be incorporated into engineered H1 virus vaccines.
H1N1 influenza viruses circulated in humans from the 1918 pandemic until 1957 and again from 1977 to the present (14). A novel H1N1 virus exhibiting some antigenic similarity to the 1918 virus was reintroduced into humans from animal reservoirs, causing a pandemic in 2009 (1, 7). Elderly people had preexisting antibodies against the 2009 virus (2, 9, 10). We and others have shown that the conservation of 1918-like sequences in the Sa antigenic site in the globular head domain of 2009 virus hemagglutinin (HA) is a likely structural correlate for this cross-reactivity (11, 13, 19, 22). Sequence conservation, including the absence of glycosylation in early-20th-century strains and again in 2009 H1N1 virus HA, make this area on the HA protein surface an important target for human humoral immunity (19, 22). A second class of recently identified antibodies, many of which are encoded by the VH1-69 germ line antibody variable-gene segment, are directed to the conserved H1N1 HA stem region (6, 8, 17, 18, 21). There may be other cross-reactive epitopes on H1N1 viruses (21). Here, we describe monoclonal antibody (MAb) 5J8, which was isolated from a healthy middle-aged woman by hybridoma technology. We found that MAb 5J8 inhibited a broad spectrum of 20th-century H1N1 strains and the pandemic 2009 H1N1 virus. The epitope of this antibody revealed a novel conserved H1 epitope adjacent to the receptor binding site domain (RBD) on the HA globular head.
Peripheral blood mononuclear cells were isolated from a healthy 47-year-old human subject and Epstein-Barr virus (EBV) transformed in 384-well plates (Nunc) in the presence of 2.5 μg/ml of CpG oligodeoxynucleotides (ODNs) 2006 (Invitrogen), 10 μM Chk2 inhibitor II (Sigma C3742), and 1 μg/ml of cyclosporine A (Sigma) essentially as previously described (24, 25). The supernatant was screened by enzyme-linked immunosorbent assay (ELISA) against a panel of recombinant soluble HA proteins. B cells were fused with HMMA2.5 myeloma cells, cultured in selection medium, and cloned by limiting dilution. The antibody genes were cloned molecularly from mRNA isolated from the cloned hybridoma cell line using previously described primer sets (15) into a pGEM-T Easy vector (Promega) and eventually into pEE12.4/pEE6.4 mammalian expression vectors (Lonza), from which they were expressed (22). They were purified by fast protein liquid chromatography (FPLC) on a protein G column (for IgG1) or via CaptureSelect λ resin (for Fab; BAC B.V.). Analysis with the international ImMunoGeneTics information system (IMGT) (12) identified MAb 5J8 as an antibody encoded by the IGHV4-b*01, J4*02, D3-3*02, IGLV3-21*02 or *03, J2*01, or J3*01 variable gene segment. The recombinant antibody was used for all of the following studies.
Expression plasmids encoding HA protein molecules were coexpressed with neuraminidase to produce virus-like particles (VLPs) in 293T cells (5, 25). Hemagglutination inhibition (HAI) assays were performed as described previously (20) using VLPs (for 1918 influenza) or live virus (for all other strains). MAb 5J8 inhibited all tested H1N1 influenza strains from 1918 to 1977 and the pandemic 2009 virus but not the seasonal H1N1 strains from 1999 or 2007 (Tables 1, ,2,2, and and3).3). HAI activity was the most potent against 1918 VLPs and the 1930 virus at 40 ng/ml.
Different dilutions of antibody were incubated with 5 log10 50% tissue culture infective doses (TCID50) of each virus for 1 h. The mixture was used to infect MDCK cells in triplicate for an hour at 37°. The plate was harvested 3 days later and read in an HA assay. The endpoint was the lowest concentration that gave no HA activity. The microneutralization assay generally agreed closely with the HA assay (Table 1).
The binding affinities of 5J8 Fab to recombinant trimeric His-tagged HA proteins were measured using anti-penta-His tips on an Octet Red instrument (FortéBio); 1918 HA variants were created with a QuikChange II XL mutagenesis kit (Agilent). The extracellular domain of full-length 1918 HA cDNA was cloned into a pcDNA3.1(+) construct that contained an SGR linker, a thrombin recognition site, a fibritin trimerization domain, and a 6×His tag. The protein was expressed in 293F cells and purified from the supernatant on an ÄKTA FPLC instrument using Ni2+ columns (GE). The binding affinity of 5J8 Fab to 2009 H1 HA protein was 2.6 × 10−8 M (Table 4).
We tested 5J8 for therapeutic efficacy in a lethal animal model of 1918 virus infection (25). Female BALB/c mice (8 weeks old, weighing approximately 20 g) were inoculated intranasally with five times the 50% lethal dose (LD50) in a 50-μl volume of the virulent reconstituted virus. At 24 h after inoculation, mice were administered 200, 20, or 2 μg of 5J8 or an equal amount of human IgG (Sigma), each by the intraperitoneal route, in groups of 10 mice. Mice were observed for weight loss for 14 days (Fig. 1B). Subsets of four animals treated with the MAbs were euthanized on day four after infection, and whole lungs were homogenized in 1 ml of sterile phosphate-buffered saline (PBS). Virus titers in lung tissue homogenates were determined by plaque titration in MDCK cell monolayer cultures. MAb 5J8 protected all animals at the high and medium dose from lethal challenge (Fig. 1A) and reduced lung virus titers compared to the IgG control by 2.6 log10 PFU/ml at the 200-μg-dose level, by 2.0 log10 PFU/ml at the 20-μg-dose level, and by 0.7 log10 PFU/ml at the 2-μg-dose level (Table 5).
We selected and sequenced the HA gene of the new antibody escape mutant viruses (4, 23). Following MAb 5J8 selection, HA mutations were identified in positions 133A, 137, 199, and 222 (based on H3 numbering ) in some of these strains (Table 3). We then introduced these naturally occurring mutations into the soluble 1918 HA protein for in vitro binding studies to validate the effect of the putative escape mutations. 133A K→I, A137T, and K222Q each eliminated binding of MAb 5J8 to the mutant protein (Table 4). However, binding of MAb 5J8 to the D199H mutant (which we detected only in concert with a 133A K→Q mutant) was retained (Table 4). We were technically unable to create the 133A K→Q mutant construct, but this residue would likely mediate escape, based on the effect of the validated 133A K→I escape mutant. Of note, all of these escape mutations are outside conventionally defined antigenic sites (3, 4) and are located between the receptor-binding pocket and the Ca2 antigenic site (Fig. 2).
We then queried the Influenza Research Database (IRD; www.fludb.org) for all naturally occurring, nonredundant human H1N1 viruses between 1918 and 2008 to identify the frequency of sequence variation in the residues mediating escape from MAb 5J8. Of note, residues 137 (alanine) and 222 (lysine) were each found in greater than 98% of all the sequences (Table 2). There was greater sequence variability at the position of the critical residue 133A: the lysine from the 1918 virus was replaced by arginine (a conservative mutation to a similarly positively charged residue) as early as in the 1930 virus without a subsequent change in HAI activity. A lysine was seen again in the 2009 H1N1 virus, but residue 133A was deleted in previous seasonal strains. When built back into 1918 HA, this deletion by itself did lead to loss of binding (Table 4). In fact, over 80% of all strains in the data set showed this deletion, but this high percentage is biased because most sequences in the IRD are recent. The 133A deletion was rarely observed prior to 1997 but has been dominant since then.
Thus, MAb 5J8 recognizes a novel, conserved epitope on the globular head of H1N1 HA that is characterized by residues 133A, 137, and 222 in close proximity to the RBD, which illuminates the molecular basis for the breadth of neutralization of MAb 5J8. There likely are very strong structural constraints on this epitope. It is certainly possible that MAb 5J8 makes contacts deeper within this pocket that H1N1 cannot change without loss of replicative capacity and that would not be revealed by escape mutations.
In conclusion, MAb 5J8 helps paint a more detailed picture of why older human subjects possessed 2009 H1N1 cross-reactive HA antibodies prior to exposure to the 2009 H1N1 virus (2, 9, 10), since epitopes other than the Sa site (11, 13, 19, 22) or stem epitopes (6, 8, 17, 18, 21) contributed to this cross-protective effect. The relative importance of these epitopes for the antiviral humoral response may vary from person to person.
The fact that it took decades for H1N1 viruses to escape MAb 5J8-like antibodies may reflect the fact that such antibodies are relatively rare and thus pose little evolutionary pressure on H1N1. We propose that presenting the RBD epitope in a more exposed way without surrounding hypervariable loops might make it more immunogenic and thus might contribute to a universal influenza virus vaccine design strategy.
Antibody nucleotide sequences have been deposited in GenBank under accession numbers JF791168 and JF791169.
We thank the anonymous blood donor; Jose A. Archuleta, Jr., and Cheryl Kinnard of the Vanderbilt Clinical Trials Center; M. Posner and L. Cavacini for the HMMA2.5 cell line; Patricia B. Smith for technical assistance; Kimberly S. Crimin for statistical help; and Cyrille Dreyfus, Damian C. Ekiert, Ian A. Wilson, and Scott A. Smith for helpful discussions.
J.E.C. is a Burroughs Wellcome Fund Clinical Scientist in Translational Research. This work was supported by NIH grants P01 AI058113 and R21 AI085306, by NIH contract HHSN272200900047C, and by DOD grant HDTRA1-08-10-BRCWMD-BAA.
The findings and conclusions in this report are those of the authors and do not necessarily reflect the views of the funding agency.
Vanderbilt University submitted a patent covering the diagnostic and therapeutic use of the antibody described in this paper. J.E.C. is the founder of Corbeau Biotech LLC.
Published ahead of print on 17 August 2011.
After this manuscript was accepted for publication, data on a similar H1N1 antibody, MAb CH65, were published (J. R. Whittle et al., Proc. Natl. Acad. Sci. U.S.A. 108:14216–14221, 2011). MAb 5J8 did not inhibit HA with deletions in position 133A, MAb CH65 needs a deletion in position 133A of HA for inhibition of virus. Strikingly, these two antibodies complement each other in their breadth, and a cocktail of CH65 and 5J8 might inhibit most human H1N1 strains. Furthermore, the fact that CH65 makes contacts deeply within the receptor-binding pocket would suggest that 5J8 has a similar mode of binding.