|Home | About | Journals | Submit | Contact Us | Français|
Identification of broadly neutralizing antibodies against influenza A viruses has raised hopes for the development of monoclonal antibody-based immunotherapy and ‘universal’ vaccines for influenza. However, a significant part of the annual flu burden is caused by two cocirculating, antigenically distinct lineages of influenza B viruses. Here we report human monoclonal antibodies, CR8033, CR8071 and CR9114, which protect mice against lethal challenge from both lineages. Antibodies CR8033 and CR8071 recognize distinct conserved epitopes in the head region of the influenza B hemagglutinin (HA), whereas CR9114 binds a conserved epitope in the HA stem and protects against lethal challenge with influenza A and B viruses. These antibodies may inform on development of monoclonal antibody-based treatments and a universal flu vaccine for all influenza A and B viruses.
Influenza viruses continue to cause significant morbidity and mortality worldwide (1). Because current vaccines are typically only effective against the specific viral strains used for vaccination and closely related viruses (2), and increasing resistance reduces the effectiveness of the available antiviral drugs (3), an urgent need remains for innovative new treatments, both prophylactic and therapeutic (4). To this end, we and others have previously described human monoclonal antibodies (mAbs) that neutralize a wide spectrum of influenza A viruses by binding to highly conserved epitopes in the stem region of hemagglutinin (HA), the major viral surface glycoprotein (5-9). To date, influenza B viruses have received less attention, as they are largely restricted to humans and thus lack the large animal reservoirs that are key to the emergence of pandemic influenza A viruses (10).
Although the morbidity and mortality rates attributable to influenza B are lower than H3N2 viruses, they are higher than H1N1 viruses (11). While influenza B viruses are classified as a single influenza type, two antigenically and genetically distinct lineages co-circulate (12), represented by the prototype viruses B/Victoria/2/1987 (Victoria lineage) and B/Yamagata/16/1988 (Yamagata lineage) (13). Vaccine manufacturers have therefore recently initiated clinical evaluation of quadrivalent vaccines that include strains from each influenza B lineage, H1N1 and H3N2 (14). Given that influenza B viruses are the major cause of seasonal influenza epidemics every two to four years leading to substantial absenteeism, hospitalization, and death (11), mAbs with broad neutralizing activity (bnAbs) against influenza B viruses have significant clinical potential.
Combinatorial display libraries, constructed from human B cells of volunteers recently vaccinated with the seasonal influenza vaccine (9), were panned using soluble recombinant HA from various influenza A and B viruses, and phages were subsequently screened for binding to HAs of both influenza B lineages (15). We recovered three immunoglobulins (IgGs) that bound HAs from both lineages, CR8033 (VH3-9, Vκ3-20) and CR8071 (VH1-18, Vλ1-47) (Fig. 1, A and B), as well as CR9114, a VH1-69 antibody, which additionally binds influenza A viruses from both group 1 and group 2 (Fig. 1A and fig. S1). Importantly, CR8033 and CR8071 neutralized representative viruses from either lineage (Fig. 1C), whereas polyclonal sheep sera did not (table S1). CR8033 showed hemagglutination-inhibition (HI) activity against the Yamagata lineage, but not against the Victoria lineage. Thus, while CR8033 likely neutralizes Yamagata strains by blocking receptor binding, it appears to neutralize Victoria strains by another mechanism. In contrast, CR8071 showed no HI activity against either lineage. Although CR9114 neutralized all influenza A viruses tested, it did not show in vitro neutralizing activity against influenza B viruses at the tested concentrations (Fig. 1C). Since recent work indicated that the protective efficacy of broadly neutralizing influenza antibody FI6 is substantially dependent on antibody effector functions (5), we evaluated the protective efficacy of all three mAbs against B/Florida/4/2006 (Yamagata) and B/Malaysia/2506/2004 (Victoria) infections in mice.
Doses as low as 0.6 mg/kg and 0.2 mg/kg of CR8033 fully protected mice from lethality upon challenge with B/Florida/4/2006 and B/Malaysia/2506/2004, respectively, and lower doses still resulted in increased survival and reduced weight loss (Fig. 2A and fig. S2A). Although CR8071 is somewhat less potent in vivo than CR8033 (Fig. 2B and fig. S2B), the difference is less marked than expected based on the microneutralization assay, indicating that in vitro neutralization is not fully predictive of in vivo potency. Despite the apparent lack of in vitro neutralizing activity, 15 mg/kg and ≥5 mg/kg of CR9114 fully protected mice from lethality following challenge with B/Florida/4/2006 and B/Malaysia/2506/2004, respectively, with significant protection against the latter virus with 1.7 and 0.6 mg/kg (Fig. 2C and fig. S2C). Similarly, 1.7 and 5 mg/kg CR9114 protected mice against challenge with lethal doses of influenza A H1N1 and H3N2 viruses, respectively (Fig. 2D and fig. S2D).
CR8033, CR8071, and CR9114 do not compete with each other for HA binding, suggesting they recognize different epitopes (fig. S3). However, CR9114 competes with CR6261 and CR8020, (fig. S4), suggesting it binds an HA stem epitope. To identify their epitopes and understand how they achieve broad neutralization, we obtained x-ray and EM structures for CR8033, CR8071 and CR9114 (Fig. 3) in complex with representative HAs of influenza A and B viruses.
We generated a 3D reconstruction of the HA ectodomain of B/Florida/4/2006 in complex with Fab CR8033 (Fig. 3B) using negative stain electron microscopy (EM). Crystal structures of Fab CR8033 at 1.9 Å and influenza B/Brisbane/60/2008 HA ectodomain at 3.45 Å provided atomic models for fitting into the EM reconstruction (fig. S5 and tables S2-S5). Three CR8033 Fabs bind the HA trimer on an epitope overlapping the receptor binding pocket and surrounding antigenic sites. The EM model suggests that almost all contacts are made by VH using all three HCDRs loops and framework 3 (fig. S6) (16). While the receptor binding pocket of influenza B HA is well conserved, variability in surrounding residues (fig. S7) may account for differences in CR8033 binding and neutralization of the two influenza B lineages.
EM reconstructions were also performed on Fab CR8071 in complex with both B/Florida/4/2006 (Yamagata) and B/Malaysia/2506/2004 HAs (Victoria) (Fig. 3B and fig. S8D), along with crystal structures of Fab CR8071 with a B/Florida/4/2006 HA1 construct (2.7 Å) and CR8059, of which CR8071 is a stabilized variant (see SOM), with B/Brisbane/60/2008 HA (5.65 Å) (fig. S8A-C). CR8071 binds the vestigial esterase domain at the base of the HA head distant from the receptor-binding site and in an orientation (Fig. 3B) consistent with the observed lack of HI. The epitope is highly conserved in influenza B HA with 17 of 19 residues >98 % conserved.
Crystal structures were determined for Fab CR9114 with HAs from a highly pathogenic H5N1 virus (A/Vietnam/1203/2004 (H5N1); Viet04/H5, 1.7 Å) (Fig. 3C), as well as H3 (5.25 Å) and H7 (5.75 Å) (fig. S9). EM studies further illustrated the CR9114 cross reactivity with influenza A H1, H3, H7, and H9 subtypes, and influenza B (Fig. 3E). CR9114 binds the HA stem (fig. S10), recognizing an epitope nearly identical to that of CR6261 (17), using the same HCDR loops (1, 2, 3) and FR3 with no light-chain contacts. Recently, human antibody FI6 was identified that broadly neutralizes influenza A viruses by targeting approximately the same epitope (5). However, FI6 uses a different V-gene (VH3-30) and its light chain (fig. S11), and is rotated by ~90°compared to VH1-69 antibodies CR6261 and CR9114. Thus, FI6 represents an alternative solution to cross-group neutralization of influenza A, but not influenza B viruses.
The CR9114 epitope is highly conserved in essentially all influenza A subtypes and influenza B (Fig. 3D, tables S6 and S7). At least three obstacles must be overcome for an antibody to bind to a similar epitope across influenza A and B. First, most group 1 HAs have HA2 Thr49, whereas most group 2 HAs have a larger Asn (fig. S10 and table S6) (6, 18) that can be accommodated by CR9114. Second, polymorphisms at HA2 position 111, His (group 1), Thr/Ala (group 2) or Glu (influenza B) result in subtly different conformations of HA2 Trp21, which affect its ability to make favorable interactions with hot spot residue HCDR2 Phe54 in CR6261 (fig. S10) (19). The more favorable orientation and approach of Phe54 and apparent plasticity of the CR9114 combining site (20) (fig. S12) may allow CR9114 to make high affinity interactions (including a hydrogen bond) with the various conformations of Trp21 in group 1 and group 2 HAs. Third, the predominant conformation of a conserved glycan at HA1 Asn38 in H3, H7, H10, and H15 in group 2, and at HA1 Asn332 in influenza B, would obscure the epitope surface (figs. S10 and S11) (21-23). Thus, this glycan and Asn side chain must be able to adopt a permissive alternative conformation as observed in the CR9114-H3 crystal structure (24). Thus, displacement of the HA1 38 glycan appears essential for CR9114 binding to several influenza A subtypes, and a similarly positioned glycan at HA1 332 in influenza B must also be avoided (fig. S11). Such glycan flexibility at HA1 38 was also observed in the FI6 structure with H3 HA but, in this case, the glycan at 332 in flu B may be more difficult to avoid (25).
CR9114 and CR6261 both utilize the VH1-69 germline Vgene and have relatively small numbers of somatic mutations compared to other bnAbs (26, 27). However, only five mutations are shared (fig. S13) and, hence, they are not simple variants of one another. Aside from the heavy-chain only binding, both antibodies are fairly conventional, in contrast to many bnAbs to other viruses, such as HIV-1, that have unusual features, such as extensive hypermutation, long indels in one or more CDRs, tyrosine sulfation, or domain swapping (26, 27). Consequently, antibodies like CR9114 are likely to be readily generated and may be present in the repertoire of many individuals, suggesting that an appropriate vaccination strategy may be able to selectively amplify such cross-reactive clones and trigger a bnAb response (4, 28, 29).
Despite CR8033 and CR8071 binding to the more variable HA globular head, but consistent with the apparent high level of epitope conservation, escape variants could only be generated after extensive passaging. For B/Florida/4/2006 virus, no CR8033 escape variants were generated in 20 passages, while 15 passages were required to generate Lys38Glu and Tyr40His mutants with reduced CR8071 susceptibility. For B/Malaysia/2506/2004 virus, 15 passages were required to generate a Pro161Gln reduced susceptibility mutant for CR8033, while 20 rounds of passaging did not result in any CR8071 escape variants. Lys38 and Pro161 are highly conserved (99.4% and 100% respectively) in all 494 (unique) full-length influenza B HA sequences in the NCBI database. Although Tyr40 is only 29.1% conserved, all other strains have His40, including B/Harbin/7/94, B/Malaysia/2506/04 and B/Brisbane/60/08, which are effectively neutralized by CR8071 (Fig. 1C). In contrast, the Lys38Glu mutation has not been observed in any influenza B isolates. Because CR9114 does not neutralize influenza B viruses in vitro, no influenza B escape variants could be generated (30).
The identification of conserved neutralizing epitopes in the influenza B HA globular head parallels the recent identification of such epitopes in influenza A viruses (31, 32) and establishes that both the HA stem and head regions contains broadly protective epitopes. Whereas the stem-binding CR9114 blocks the HA pH-induced conformational changes associated with membrane fusion (Fig. 4, A and B; and figs. S14 and S15) (33), this mechanism is not applicable to CR8033 and CR8071 that bind the head region (Fig. 4A). Indeed, whereas CR8033 has HI activity and blocks viral infection when pre-incubated with B/Florida/4/2006 virus (Fig. 1C), it had no such effect on B/Malaysia/2506/2004 virus (Fig. 4B). Furthermore, CR8071 did not prevent infection of cells by either virus (Fig. 4B) (34).
Interestingly, these antibodies appear to prevent propagation of viruses without preventing entry and genome replication (Figs. 1C and and4B).4B). Thus, we assessed the effect of CR8033 and CR8071 on formation of viral progeny from infected cells. Whereas HA was readily detected in both lysates and supernatants of MDCK cell cultures that were first infected and subsequently incubated with a non-binding control antibody or CR9114, no HA could be detected in the supernatants upon incubation with CR8033 or CR8071, despite abundant HA in the lysate (Fig. 4C). The effects of CR8033 and CR8071 then resemble that of neuraminidase (NA) inhibitor zanamivir, which interferes with release of progeny virions from infected cells (35). Indeed, the dense aggregation of virions on the surface of infected cells, particularly in the presence of antibody CR8033, closely resembles that observed in zanamivir-treated cells as examined by scanning electron microscopy (Fig. 4D).
The identification and characterization of monoclonal antibodies with broad neutralizing activity against influenza B viruses, together with previously described broadly neutralizing antibodies against group 1 and group 2 influenza A viruses (6, 7, 9) brings a universal therapy a step closer, and may serve as guides for design of broadly protective vaccines. In particular, a vaccine that elicits antibodies targeting the CR9114 epitope may provide the ultimate goal of protection against all influenza A and influenza B viruses.
We thank J. Juraszek, B. Brandenburg, M. Koldijk, K. Hegmans, J. Meijer, N. Hafkemeier, A. Apetri, J. Dulos, L. Dekking, E. Vietsch, H. Tien, D. Marciano, the staff of the APS GM/CA-CAT 23ID-B and 23ID-D, SSRL BL11-1, SSRL BL7-1 and ALS, X. Dai, R. Stanfield, W. Yu, J.P. Julien, and R.N. Kirchdoerfer, and D. Brown of Ottawa University, Canada for the mouse-adapted A/Hong Kong/1/68 strain, Crystallization experiments were carried out on the Rigaku Crystalmation system at the JCSG, which is supported by the NIGMS Protein Structure Initiative U54 GM094586. The electron microscopy and image analysis was conducted by R.K., J.H.L., and Z.M. at the National Resource for Automated Molecular Microscopy, which is supported by NIH through the National Center for Research Resources’ P41 program (RR017573). This project has been funded in part by the Area of Excellence Scheme of the University Grants Committee, Hong Kong (grant AoE/M-12/06); a predoctoral fellowship from the Achievement Rewards for College Scientists Foundation (D.C.E.); grant GM080209 from the NIH Molecular Evolution Training Program (D.C.E.); a Saper Aude Postdoc grant from The Danish Council for Independent Research, Natural Sciences (N.S.L) and the Skaggs Institute (I.A.W.). Portions of this research were carried out at the Stanford Synchrotron Radiation Lightsource, a national user facility operated by Stanford University on behalf of the U.S. Department of Energy (DOE), Office of Basic Energy Sciences. The Stanford Synchrotron Radiation Lightsource (SSRL) Structural Molecular Biology Program is supported by the DOE Office of Biological and Environmental Research and by NIH, National Center for Research Resources, Biomedical Technology Program (P41RR001209), and the National Institute of General Medical Sciences. The GM/CA CAT 23-ID-B beamline has been funded in whole or in part with federal funds from National Cancer Institute (Y1-CO-1020) and NIGMS (Y1-GM-1104). Use of the Advanced Photon Source (APS) was supported by the U.S. Department of Energy, Basic Energy Sciences, Office of Science, under contract no. DE-AC02-06CH11357. The content is solely the responsibility of the authors and does not necessarily represent the official views of NIGMS or the NIH. The Advanced Light Source is supported by the Director, Office of Science, Office of Basic Energy Sciences, of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231. Coordinates and structure factors are deposited in the Protein Data Bank (PDB code 4FQH for CR9114 Fab, 4FQI for CR9114 Fab-A/H5 HA, 4FQJ for CR8071 Fab-B/Florida HA1, 4FQK for CR8059 Fab-B/Brisbane HA, 4FQL for CR8033 Fab, 4FQM for B/Brisbane HA, 4FQY for CR9114 Fab-A/H3 HA and 4FQV for CR9114 Fab-A/H7 HA). Image reconstructions have been deposited at the EMDB under accession numbers EMD-2143 (CR8033/Florida), EMD-2144 (CR8071/Florida), EMD-2145 (CR8071/Malaysia), EMD-2146 (CR9114/Florida), EMD-2147 (CR9114/H1), EMD-2148 (CR9114/H3), EMD-2149 (CR9114/H9), EMD-2150 (CR9114/H7). Nucleotide sequences for the CR8033, CR8059 (CR8071) and CR9114 variable regions have been deposited in GenBank (accession numbers JX213635 (CR8033, VH), JX213636 (CR8033, Vκ), JX213637 (CR8059, VH), JX213638 (CR8059, Vλ), JX213639 (CR9114, VH) and JX213640 (CR9114, Vλ)). Patent applications relating to antibodies CR8033, CR8059, CR8071 and CR9114 have been filed (CR8033, CR8059 and CR8071 are claimed in EP 12158525.1 and US 61/608,414. CR9114 is claimed in EP 11173953.8 and US 61/572,417). Sharing will be subject to standard material transfer agreements.
This is publication 21741 from The Scripps Research Institute.
Materials and Methods