Antigenic drift of the pandemic 2009 A(H1N1) influenza virus in A ferret model

Abstract

Surveillance data indicate that most circulating A(H1N1)pdm09 influenza viruses have remained antigenically similar since they emerged in humans in 2009. However, antigenic drift is likely to occur in the future in response to increasing population immunity induced by infection or vaccination. In this study, sequential passaging of A(H1N1)pdm09 virus by contact transmission through two independent series of suboptimally vaccinated ferrets resulted in selection of variant viruses with an amino acid substitution (N156K, H1 numbering without signal peptide; N159K, H3 numbering without signal peptide; N173K, H1 numbering from first methionine) in a known antigenic site of the viral HA. The N156K HA variant replicated and transmitted efficiently between naïve ferrets and outgrew wildtype virus in vivo in ferrets in the presence and absence of immune pressure. In vitro, in a range of cell culture systems, the N156K variant rapidly adapted, acquiring additional mutations in the viral HA that also potentially affected antigenic properties. The N156K escape mutant was antigenically distinct from wildtype virus as shown by binding of HA-specific antibodies. Glycan binding assays demonstrated the N156K escape mutant had altered receptor binding preferences compared to wildtype virus, which was supported by computational modeling predictions. The N156K substitution, and culture adaptations, have been detected in human A(H1N1)pdm09 viruses with N156K preferentially reported in sequences from original clinical samples rather than cultured isolates. This study demonstrates the ability of the A(H1N1)pdm09 virus to undergo rapid antigenic change to evade a low level vaccine response, while remaining fit in a ferret transmission model of immunization and infection. Furthermore, the potential changes in receptor binding properties that accompany antigenic changes highlight the importance of routine characterization of clinical samples in human A(H1N1)pdm09 influenza surveillance.

Figure 1. Time course of influenza transmission through passage lines.

Naïve donor 0 (D0) ferrets were directly infected with A/Tasmania/2007/2009 virus and co-housed with naïve donor 1 (D1) ferrets to begin the contact transmission model. D0 and D1 ferrets were not included in any experimental group. Once D1 was infected, D0 was removed and a ferret from one of the four experimental groups was added (R0). Transmission then proceeded through R0–R7 ferrets for each group. Line A of each group is shown. The same D0 and D1 ferrets were used to establish the passage line A of the naïve and MIV experimental groups. The same D0 ferret infected two different D1 ferrets to establish the passage line A of the PBS+IFA and MIV+IFA experimental groups. Nasal washes were collected daily from ferrets and virus load was measured by real time RT-PCR. During the experiment, both the rapid test result (PBS+IFA, MIV+IFA day 0–26) and the raw Ct value (naïve, MIV, MIV+IFA day 27 onwards) were used as a marker of infection and transmission. The data points at which transmission of virus to recipient ferrets was deemed to have occurred are identified as red symbols. Direct intranasal inoculation indicated by arrow.

Figure 2. Virus replication and transmission kinetics in ferret passage lines and emergence and persistence of the N156K mutant.

(A–C) Nasal washes were collected daily from ferrets from passage lines (R0–R7) and assayed for viral RNA by real time RT-PCR as a measure of infection and transmission. (A) The peak viral load detected in the nasal wash from each ferret, (B) the viral growth rate ((peak virus load – first day virus load)/days to reach peak) and (C) the serial interval of virus transmission between ferrets (number of days between detection of infected and infecting animals) were determined for each immunization line. Ferrets artificially infected by intranasal virus inoculation are not included. Ferrets that failed to be naturally infected by contact transmission are indicated in white circles. Statistics do not include ferrets that failed to be naturally infected. (D–E) The proportion of N156 wildtype (black) and N156K (white) viruses in peak day nasal wash samples from all ferrets from the MIV+IFA (D) and N156K naïve (E) passage lines quantified by pyrosequencing. (F–G) Mixtures of (F) N156 wildtype and N156K or (G) N156K and K142N+N156K viruses were passaged by contact transmission through naïve ferrets. The proportion of N156 wildtype (black), N156K (white) or K142N+N156K (striped) in peak day nasal wash samples was quantified by pyrosequencing.

Figure 3. Growth, detection and adaptation of N156K virus in cell cultures.

(A) Virus underwent two passages in MDCK-SIAT1 cells (P1 and P2). Supernatant was collected during both passages, at 24, 48 and 72 h, and virus load was quantified by real time RT-PCR and hemagglutination (HAU). Infected cells were also harvested at 48 h and surface HA and M protein expression measured by flow cytometry. Cell culture adaptations detected by sequencing in the HA protein are shown in each graph. No adaptations in NA were detected. (B) Virus underwent two passages (P1 and P2) in the indicated cell lines. Virus load was quantified in supernatant as above, and cell culture adaptation was detected by sequencing the HA protein as indicated. Limit of detection is 10^3.8 copies. (C) Location of HA mutations in antigenic regions identified in this study. Visualization of mutation positions relative to classical antigenic sites are shown as colored bubbles (Sa – red; Sb – green; Ca1 – cyan; Ca2 – blue; Cb – yellow). Pink balls indicate a bound host receptor ligand (α-2,6-linked); white balls indicate sugars on glycosylation sites. (D) Location of antigenic sites in HA trimer, top and side view, respectively.

Figure 4. Analysis of antigenicity and glycan binding to HA on influenza virus-infected cells.

MDCK-SIAT1 cells were inoculated with A/Tasmania/2004/2009 egg-grown inoculum, ferret nasal wash (indicated by ∧) or rescued reverse genetics ferret-adapted viruses (indicated by ‘rg’) or a control A(H3N2) egg-grown virus. After 48 h, cells were analysed for binding of (A) anti-HA mAbs, (B) post-immunization ferret antisera (MIV or MIV+IFA) or post-infection ferret antisera (wildtype A/Tasmania/2004/2009, N156K or K142N+N156K virus) or (C) synthetic glycans. (B) To control for differences in infection rate between wildtype and N156K viruses, antibody binding is expressed as the proportion relative to K142N+N156K antiserum (% positive for test antiserum/% positive K142N+N156K antiserum). Mean+standard deviation of six paired experiments are shown for all antisera except MIV+IFA which is calculated from three paired experiments. (C) Glycan binding is expressed as a proportion to itself ((mean fluorescent intensity (MFI) with glycan-MFI without glycan)/MFI with glycan×100), for influenza A matrix positive cells only. Individual experiments are shown by each symbol.

Figure 5. 3-D modeling of HA containing N156K and cell culture adaptations.

(A–D) Comparison of electrostatic surface potential of wildtype and single and pair mutations in the HA head domain, calculated with the Particle Mesh Ewald method implemented in YASARA. Blue indicates positive and red indicates negative charge potential. A host receptor analogue is shown in magenta. (E,F) Structural modeling of single and pair mutations in HA with bound α-2,6- or α-2,3-linked host receptor ligands. (E) Model of N156 wildtype (green HA/yellow ligand), N156K (cyan HA/red ligand) and N156E (blue HA/purple ligand) shown with α-2,3 host receptor analogue. (F) Comparison of N156 wildtype (green HA/yellow ligand) and N156K (cyan HA/red ligand) with double mutant K153E+N156K (light blue HA/orange ligand).