Molecular features of the serological IgG repertoire elicited by egg-based, cell-based, or recombinant haemagglutinin-based seasonal influenza vaccines: a comparative, prospective, observational cohort study

Abstract

Background: Egg-based inactivated quadrivalent seasonal influenza vaccine (eIIV4), cell culture-based inactivated quadrivalent seasonal influenza vaccine (ccIIV4), and recombinant haemagglutinin (HA)-based quadrivalent seasonal influenza vaccine (RIV4) have been licensed for use in the USA. In this study, we used antigen-specific serum proteomics analysis to assess how the molecular composition and qualities of the serological antibody repertoires differ after seasonal influenza immunisation by each of the three vaccines and how different vaccination platforms affect the HA binding affinity and breadth of the serum antibodies that comprise the polyclonal response.

Methods: In this comparative, prospective, observational cohort study, we included female US health-care personnel (mean age 47·6 years [SD 8]) who received a single dose of RIV4, eIIV4, or ccIIV4 during the 2018-19 influenza season at Baylor Scott & White Health (Temple, TX, USA). Eligible individuals were selected based on comparable day 28 serum microneutralisation titres and similar vaccination history. Laboratory investigators were blinded to assignment until testing was completed. The preplanned exploratory endpoints were assessed by deconvoluting the serological repertoire specific to A/Singapore/INFIMH-16-0019/2016 (H3N2) HA before (day 0) and after (day 28) immunisation using bottom-up liquid chromatography-mass spectrometry proteomics (referred to as Ig-Seq) and natively paired variable heavy chain-variable light chain high-throughput B-cell receptor sequencing (referred to as BCR-Seq). Features of the antigen-specific serological repertoire at day 0 and day 28 for the three vaccine groups were compared. Antibodies identified with high confidence in sera were recombinantly expressed and characterised in depth to determine the binding affinity and breadth to time-ordered H3 HA proteins.

Findings: During September and October of the 2018-19 influenza season, 15 individuals were recruited and assigned to receive RIV4 (n=5), eIIV4 (n=5), or ccIIV4 (n=5). For all three cohorts, the serum antibody repertoire was dominated by back-boosted antibody lineages (median 98% [95% CI 88-99]) that were present in the serum before vaccination. Although vaccine platform-dependent differences were not evident in the repertoire diversity, somatic hypermutation, or heavy chain complementarity determining region 3 biochemical features, antibodies boosted by RIV4 showed substantially higher binding affinity to the vaccine H3/HA (median half-maximal effective concentration [EC50] to A/Singapore/INFIMH-16-0019/2016 HA: 0·037 μg/mL [95% CI 0·012-0·12] for RIV4; 4·43 μg/mL [0·030-100·0] for eIIV4; and 18·50 μg/mL [0·99-100·0] μg/mL for ccIIV4) and also the HAs from contemporary H3N2 strains than did those elicited by eIIV4 or ccIIV4 (median EC50 to A/Texas/50/2012 HA: 0·037 μg/mL [0·017-0·32] for RIV4; 1·10 μg/mL [0·045-100] for eIIV4; and 12·6 μg/mL [1·8-100] for ccIIV4). Comparison of B-cell receptor sequencing repertoires on day 7 showed that eIIV4 increased the median frequency of canonical egg glycan-targeting B cells (0·20% [95% CI 0·067-0·37] for eIIV4; 0·058% [0·050-0·11] for RIV4; and 0·035% [0-0·062] for ccIIV4), whereas RIV4 vaccination decreased the median frequency of B-cell receptors displaying stereotypical features associated with membrane proximal anchor-targeting antibodies (0·062% [95% CI 0-0·084] for RIV4; 0·12% [0·066-0·16] for eIIV4; and 0·18% [0·016-0·20] for ccIIV4). In exploratory analysis, we characterised the structure of a highly abundant monoclonal antibody that binds to both group 1 and 2 HAs and recognises the HA trimer interface, despite its sequence resembling the stereotypical sequence motif found in membrane-proximal anchor binding antibodies.

Interpretation: Although all three licensed seasonal influenza vaccines elicit serological antibody repertoires with indistinguishable features shaped by heavy imprinting, the RIV4 vaccine selectively boosts higher affinity monoclonal antibodies to contemporary strains and elicits greater serum binding potency and breadth, possibly as a consequence of the multivalent structural features of the HA immunogen in this vaccine formulation. Collectively, our findings show advantages of RIV4 vaccines and more generally highlight the benefits of multivalent HA immunogens in promoting higher affinity serum antibody responses.

Funding: Centers for Disease Control and Prevention, National Institutes of Health, and Bill & Melinda Gates Foundation.

Figure 1:. Strategy for proteomic profiling of serum IgG repertoires specific to A/Singapore/INFIMH-16–0019/2016 H3/HA

BCR-Seq=B-cell receptor repertoire sequencing. ccIIV4=cell culture-based inactivated quadrivalent seasonal influenza vaccine. eIIV4=egg-based inactivated quadrivalent seasonal influenza vaccine. HA=haemagglutinin. Iq-Seq=high-resolution proteomics analysis of antigen-specific serum immunoglobulins repertoires. LC-MS/MS=liquid chromatography–tandem mass spectrometry. MS=mass spectrometry. PBMC=peripheral blood mononuclear cells. RIV4=recombinant haemagglutinin-based quadrivalent seasonal influenza vaccine.

Figure 2:. Proteomic analysis of serum IgG repertoires elicited by RIV4, standard-dose eIIV4, and ccIIV4 2018–19 quadrivalent vaccination

(A) Day 28 serum microneutralisation titres against cell-grown A/Singapore/INFIMH-16–0019/2016 H3 vaccine viruses. (B) Serum ELISA binding titres to recombinant A/Singapore/INFIMH-16–0019/2016 H3/HA on day 0 or day 28 for each vaccine cohort. The dotted line between the two timepoints connects the same individual. Statistical analysis was performed using the two-tailed Wilcoxon matched-pairs signed rank test. (C) Clonal composition and relative abundance of serum IgG repertoires before and after immunisation. Each row in the repertoire heatmap represents a unique clonotype ID detected at day 0 or day 28 for each individual. The heatmap includes all clonotypes detected at >0·5% of the total amount (integrated extracted ion chromatogram peak areas). (D) Serum abundance of pre-existing clonotypes in the post-vaccination repertoire (the green box in panel C) was compared between different vaccine cohorts. (E) Fold change in MFI to nucleoprotein at day 28 relative to day 0 for the two atypical individuals (ID A5 and C4) and typical individuals (n=13). Statistical analysis was done using the two-tailed Mann–Whitney test. The horizontal line indicates the median, and the error bar indicates the 95% CI (A, B, D, and E). Each point represents an individual (A, B, D, and E). For multiple comparisons across the three vaccine cohorts, Kruskal–Wallis tests followed by Dunn’s post-hoc tests were performed (A, B, and D). ccIIV4=cell culture-based inactivated quadrivalent seasonal influenza vaccine. eIIV4=egg-based inactivated quadrivalent seasonal influenza vaccine. HA=haemagglutinin. MFI=mean fluorescence intensity. RIV4=recombinant haemagglutinin-based quadrivalent seasonal influenza vaccine.

Figure 3:. Comparative analysis of molecular features in the serological IgG repertoire to A/Singapore/INFIMH-16–0019/2016 (H3N2) HA

(A) Fold change in the D80 diversity index of the anti-H3/HA serological repertoire at day 28 compared with day 0 (n=13 typical donors). Each dot represents each individual.The horizontal line and error bar represent the median and 95% CI. Comparison of percentage VH somatic hypermutation (B), CDRH3 hydrophobicity index (GRAVY index; C), and CDRH3 amino acid length (D) for day 28 serum antibody clonotypes specific to antigen. The violin and box-whisker plot are overlaid (B and C). The smooth distribution curve was calculated using kernel density estimation and was superimposed on the histogram (D). (E) Frequency of VH genes used in the anti-H3/HA serum repertoire after vaccination. Colours indicate different VH genes. Each dot represents all HA-specific day 28 IgG clonotypes (B, C, and E). Statistical analysis was performed using ordinary one-way ANOVA followed by Tukey’s multiple comparison tests (A–D). ccIIV4=cell culture-based inactivated quadrivalent seasonal influenza vaccine. CDRH3=complementarity determining region 3. eIIV4=egg-based inactivated quadrivalent seasonal influenza vaccine. HA=haemagglutinin. RIV4=recombinant haemagglutinin-based quadrivalent seasonal influenza vaccine. VH=variable heavy chain.

Figure 4:. Different qualities of monoclonal antibodies elicited by distinct vaccine platforms

The binding affinity to recombinant A/Singapore/INFIMH-16–0019/2016 (A) or A/Texas/50/2012 (B) H3/HAs for representative monoclonal antibodies identified at high abundance in serum was compared in the three vaccine cohorts. The relative abundance of monoclonal antibodies in the serum normalised by their binding potency (1/EC50) to either A/Singapore/INFIMH-16–0019/2016 (C) or A/Texas/50/2012 (D) HAs was compared across different vaccine cohorts. Each dot represents monoclonal antibodies selected from an individual immunised with either RIV4, eIIV4, or ccIIV4 (A–D). Comparison of the binding landscape of top-most abundant monoclonal antibodies with the binding landscape of complete serum in individuals A4 (E) and B5 (F). (G) The complete serum binding landscape to H3/HAs before and after vaccination for each individual immunised with RIV4, eIIV4, or ccIIV4. MFI was measured using monoclonal antibodies (E and F) or serum on day 0 or day 28 (G) via multiplexed Luminex assay. The line plot connects the mean fluorescence intensity across different HA antigens (G). The dotted lines indicate the limit of quantification (A, B, and G). Statistical analysis was performed using Kruskal–Wallis tests followed by Dunn’s post-hoc tests (A–D). AU=arbitrary unit. ccIIV4=cell culture-based inactivated quadrivalent seasonal influenza vaccine. EC50=half-maximal effective concentration. eIIV4=egg-based inactivated quadrivalent seasonal influenza vaccine. HA=haemagglutinin. MFI=Mean fluorescence intensity. RIV4=recombinant haemagglutinin-based quadrivalent seasonal influenza vaccine.

Figure 4:. Different qualities of monoclonal antibodies elicited by distinct vaccine platforms

The binding affinity to recombinant A/Singapore/INFIMH-16–0019/2016 (A) or A/Texas/50/2012 (B) H3/HAs for representative monoclonal antibodies identified at high abundance in serum was compared in the three vaccine cohorts. The relative abundance of monoclonal antibodies in the serum normalised by their binding potency (1/EC50) to either A/Singapore/INFIMH-16–0019/2016 (C) or A/Texas/50/2012 (D) HAs was compared across different vaccine cohorts. Each dot represents monoclonal antibodies selected from an individual immunised with either RIV4, eIIV4, or ccIIV4 (A–D). Comparison of the binding landscape of top-most abundant monoclonal antibodies with the binding landscape of complete serum in individuals A4 (E) and B5 (F). (G) The complete serum binding landscape to H3/HAs before and after vaccination for each individual immunised with RIV4, eIIV4, or ccIIV4. MFI was measured using monoclonal antibodies (E and F) or serum on day 0 or day 28 (G) via multiplexed Luminex assay. The line plot connects the mean fluorescence intensity across different HA antigens (G). The dotted lines indicate the limit of quantification (A, B, and G). Statistical analysis was performed using Kruskal–Wallis tests followed by Dunn’s post-hoc tests (A–D). AU=arbitrary unit. ccIIV4=cell culture-based inactivated quadrivalent seasonal influenza vaccine. EC50=half-maximal effective concentration. eIIV4=egg-based inactivated quadrivalent seasonal influenza vaccine. HA=haemagglutinin. MFI=Mean fluorescence intensity. RIV4=recombinant haemagglutinin-based quadrivalent seasonal influenza vaccine.

Figure 5:. Molecular, biochemical, and structural characterisation of UT14 elicited by ccIIV4

(A) Stereotypical sequence features matching known membrane-proximal anchor antibodies (red) for UT14. CDRH3 and CDRK3 tryptic peptides were detected in the eluate of H3 HA affinity chromatography and absent in the flow-through on day 28, consistent with the presence of UT14 in serum and its binding to H3/HA. Amino acid sequences were defined as follows: AKERDRDGYNEGIYDYW=AlaLysGluArgAspArgAspGlyTyrAsnGluGlyIleTyrAspTyrTrp; GQGTLVTVSS=GlyGlnGlyThrLeuValThrValSerSer; CQQRYNWPITF=CysGlnGlnArgTyrAsnTrpProIleThrPhe; GQGTRLEIK=GlyGlnGlyThrArgLeuGluIleLys; ERDRDGYNEGIYDYWGQGTLVTVSSASTK=GluArgAspArgAspGlyTyrAsnGluGlyIleTyrAspTyrTrpGlyGlnGlyThrLeu.ValThrValSerSerAlaSerThrLys; DRDGYNEGIYDYWGQGTLVTVSSASTK=AspArgAspGlyTyrAsnGluGlyIleTyrAspTyrTrpGlyGlnGlyThrLeuValThr;ValSerSerAlaSerThrLys; DGYNEGIYDYWGQGTLVTVSSASTK=AspGlyTyrAsnGluGlyIleTyrAspTyrTrpGlyGlnGlyThrLeuValThrValSerSerAlaSerThrLys; and YNWPITFGQGTR=TyrAsnTrpProIleThrPheGlyGlnGlyThrArg. (B) Binding breadth of UT14 tested by multiplexed Luminex assay against A/group 1, A/group 2, and B/HAs. The dotted line indicates the limit of quantification. (C) Biolayer interferometry competition of UT14 with known trimer interface (D1 H1–3/H3–3, FluA-20, and H2214), membrane-proximal anchor region (047–09 4F04), or central stalk region (FI6v3 and CR9114) monoclonal antibodies for binding to A/California/07/2009 H1 HA. (D) Biolayer interferometry binding kinetics of UT14 Fab to monomer (top) and trimeric (bottom) forms of A/Singapore/INFIMH-16–0019/2016 H3/HA. Raw sensorgram data, black. Curve fit, red. Mean (SD) of the dissociation constant (K_D) was shown. BCR-Seq=B-cell receptor repertoire sequencing. ccIIV4=cell culture-based inactivated quadrivalent seasonal influenza vaccine. CDRH3=complementarity determining region 3 of the heavy chain. CDRK3=complementarity determining region 3 of the kappa light chain. HA=haemagglutinin. Ig-Seq=bottom-up liquid chromatography–tandem mass spectrometry proteomics analysis of antigen-specific serum immunoglobulins. MFI=mean fluorescence intensity. NGS=next-generation sequencing. PSM=peptide spectra match. XIC=extracted ion chromatogram.

Figure 6:. Cryo-EM structural analysis of UT14 binding to the trimer interface epitope of H3 HA head

(A) Cryo-electron microscopy three dimensional reconstruction of a side view of UT14 variable region bound to A/Singapore/INFIMH-16–0019/2016 H3/HA head. (B) Comparison of the angles of approach among UT14, FluA-20 (PDB identifier: 6OCB), H2214 (PDB identifier: 6E56), S5V2–29 (PDB identifier: 6E4X), and FL-1066 (PDB identifier: 6N5E) monoclonal antibodies in complex with H3/HA heads (purple). HA regions with poor or missing electron microscopy density were excluded from the UT14-H3 model. PDB=Protein Data Bank. VH=variable heavy chain. VL=variable light chain.