Germline-encoded neutralization of a Staphylococcus aureus virulence factor by the human antibody repertoire

Abstract

Staphylococcus aureus is both an important pathogen and a human commensal. To explore this ambivalent relationship between host and microbe, we analysed the memory humoral response against IsdB, a protein involved in iron acquisition, in four healthy donors. Here we show that in all donors a heavily biased use of two immunoglobulin heavy chain germlines generated high affinity (pM) antibodies that neutralize the two IsdB NEAT domains, IGHV4-39 for NEAT1 and IGHV1-69 for NEAT2. In contrast to the typical antibody/antigen interactions, the binding is primarily driven by the germline-encoded hydrophobic CDRH-2 motifs of IGHV1-69 and IGHV4-39, with a binding mechanism nearly identical for each antibody derived from different donors. Our results suggest that IGHV1-69 and IGHV4-39, while part of the adaptive immune system, may have evolved under selection pressure to encode a binding motif innately capable of recognizing and neutralizing a structurally conserved protein domain involved in pathogen iron acquisition.

Figure 1. Characterization of anti-IsdB antibodies in the human memory B cell repertoire.

(a) A persistent population of IsdB+ memory B cells from the peripheral blood mononuclear cells (PBMC) of a donor (D3) was observed over a 15-month period. By FACS, ∼0.06% of the IgM− CD19+ CD27+ memory B cells in the total memory B cell repertoire of this donor bind IsdB. (b) Most of the cloned BCR transcripts of the IsdB+ memory B cells collected at month 1, 3 and 15 are clonally related. BCR sequences from single-cell cloning of IsdB+ memory B cells were clustered based on heavy chain V-gene usage and CDR-H3 sequences. In total, we identified 31 unique clusters from donor D3 over the three collection time points. The Venn diagram shows that sibling clones within a cluster can be isolated at multiple time points. (c) Two distinct sets (bin C and bin P) of function-blocking antibodies specifically target NEAT1 and NEAT2, respectively. Single-cell cloning was performed at three different time points for donor D3 and one time each for donors D1, D2, and D4. In total, 75 unique antibodies targeting IsdB were identified and characterized. Shown here are the results of a comprehensive epitope binning analysis of 67 antibodies. Each reformatted clone is shown as a box and coloured according to its VH germline usage. The height of the box indicates the number of clustered BCR transcripts represented for each reformatted clone. There are in total 9, 34, 327, and 68 anti-IsdB single-cell BCR transcripts for D1, D2, D3 and D4, respectively. Each column of clones represents an epitope bin and this is overlaid on top of a linear representation of the IsdB molecule with NEAT1 in orange, and NEAT2 in blue. Clones that are able to fully block haemoglobin binding are outlined with a red box.

Figure 2. Germline-encoded binding of IGHV1-69 to the NEAT2 domain of IsdB.

(a) Crystal structure of IGHV1-69-derived Fabs from two donors (D2-06-N2 and D4-30-N2) in complex with NEAT2. The Fabs of D2-06-N2 and D4-30-N2 show a near identical binding mechanism to NEAT2 as evidenced by the superimposed structures. To facilitate the crystallization process, a sandwiching Fab from an antibody (D3–13) that binds NEAT2 at a non-overlapping epitope was used. For clarity, the sandwiching Fab is removed from the figure, but is included in the Supplementary Data (Supplementary Fig. 10a,b). (b) Both IGHV1-69-derived antibodies use the conserved F54 on CDR-H2 to interact with the heme-binding pocket of NEAT2. The stem of the CDR-H2 loop also mediates major contacts with the β7-turn-β8 loop of NEAT2. (c) The heme pocket residues of NEAT2 which interact with the conserved F54 on CDR-H2 are highlighted in the complex with D2-06-N2. They are M362, M363 and F366 in α-helix 1, V435 on the β-strand 7, and Y440 and Y444 on the β-strand 8. (d) CDR-H2 dominates the interaction in terms of BSA in both structures. Structural analysis shows that 75–80% of the BSA is attributed to the heavy chain, and 20–25% to the light chain. In particular, the CDR-H2 contributes 41 and 49% of total BSA for the respective structures. (e) Mutational analysis confirms the structural data and demonstrates that all IGHV1-69-derived antibodies in this set bind NEAT2 with a similar mechanism. The K_D for all antibodies in this set and their respective F54A variants against IsdB NEAT2 were determined by SPR-based biosensor assays at 37 °C (K_D range, n≥2). Antibody binding to NEAT2 variants of V435R (heme-binding pocket) and D390A/K436A/T437A (β7-turn-β8 loop) was tested by ELISA (percentage binding relative to binding to wild type IsdB, one representative set of results out of three independent experiments is shown). Clones from each donor were reverted to VH germline sequence and tested for binding to NEAT2 by both ELISA and biosensor analysis, and ability to block haemoglobin binding. N.D. stands for not determined.

Figure 3. Allelic specificity of IGHV1-69-derived NEAT2 binders.

(a) Amino acid differences among functional alleles of IGHV1-69. (b) The VH of clone D2-06-N2 was germline-reverted to all alleles with amino acid differences, and tested for binding to IsdB by ELISA. All alleles with a G50R substitution lost binding. ELISA data is an average of three independent experiments. Error bars are defined as s.d. (c.) Three individual variants (G50R, F54L and T56I) of D2-06-N2 (IGHV1-69*01 germline-reverted) were generated and their binding to IsdB was tested. Only variant G50R showed significant loss of binding. ELISA data is an average of three independent experiments. Error bars are defined as s.d. (d) Analysis of the structure illustrates how a change from G to R (most frequent rotamer) at position 50 is expected to cause a steric clash in the binding to NEAT2.

Figure 4. Germline-encoded binding of IGHV4-39 to the NEAT1 domain of IsdB.

(a) Crystal structure of an IGHV4-39-derived Fab (D4-10-N1) in complex with NEAT1. The two aromatic resides (Y52 and F53) in CDR-H2 interact with the α-helix1 of NEAT1 which is normally involved in binding haemoglobin. IGHV4-39 CDR-H2 F53 of Fab D4-10-N1 protrudes into a hydrophobic pocket of NEAT1, which is structurally homologous to the heme binding pocket of NEAT2. Crystallization was facilitated by the use of a sandwiching Fab from an antibody (D3–19) that binds NEAT1 at a non-overlapping epitope (bin H). For clarity, the sandwiching Fab is removed from the figure but is included in the Supplementary Data (Supplementary Fig. 10c). (b) CDR-H2 dominates the interaction in terms of BSA. Structural analysis shows that 79% of the BSA is attributed to the heavy chain, and 21% to the light chain. The CDR-H2 contributes about 45% of total BSA. (c) Mutational analysis confirms the structural data and demonstrates that all IGHV derived antibodies in this set bind NEAT1 with a similar mechanism. The K_D for all antibodies in this set was determined by SPR-based biosensor binding analysis to recombinant full-length IsdB at 37 °C (K_D range, n≥2). The binding of antibody variants at positions 52 and 53 of CDR-H2 to wild type IsdB and the binding of antibodies to NEAT1 variant Y165R (α-helix 1) were evaluated by ELISA (percentage binding relative to binding between original isolated antibodies and wild type IsdB, one representative set of results out of three independent experiments is shown). Every clone was reverted to VH germline sequence and tested for binding to NEAT1 by biosensor analysis and ability to block haemoglobin binding. ND stands for not determined.

Figure 5. Germline and allelic specificity of IGHV4-39-derived NEAT1 binders.

(a) Clone D4-10-N1 was reverted to all allelic variants with amino acid differences relative to IGHV4-39*01. No differences in IsdB binding were observed. Results shown are an average of three independent experiments. Error bars are defined as s.d. (b) IGHV4-39*01 has high sequence homology to IGHV4-30*04 and IGHV4-61*01 (they only differ by six amino acids in the variable region), and they all have the critical Y52 and Y53 residues in the CDR-H2. (c) The VH of four clones, one from each donor, were reverted to both IGHV4-30*04 and IGHV4-61*01, and their binding to IsdB was tested by ELISA. All IGHV4-30*01-derived variants were unable to bind IsdB, while most of the IGHV4-61*01-derived variants exhibited significantly loss of binding. ELISA data is an average of three independent experiments. Error bars are defined as s.d.

Figure 6. Naïve IGHV4-39-derived antibodies from naïve B cells.

(a) Schematics of naïve IGHV4-39 antibody phage library generation. CD19+ CD27− IgM+ naïve B cells were isolated individually by FACS from 36 donors. Total Ig RNA was converted into cDNA using an IgM specific reverse primers, and then uniquely barcoded IGHV4-39 primers for each donor were used to selectively amplify the IGHV4-39 VH gene from the cDNA. The amplified VH genes were then pooled together and paired with the light chain variable genes from IGKV families 1–4 amplified from the same set of donors to generate the single-chain Fv library. Antibody libraries were then displayed on phage and 4 rounds of panning against recombinant IsdB NEAT1 were performed. (b) Binding characterization of IGHV4-39 encoded naïve NEAT1-binding antibodies from seven different donors. Heavy and light chain germlines usage, CDR-H3 sequence identities and number of variable heavy chain framework nucleotide mutation of the seven NEAT1 binders are shown. The observed single framework mutation in selected clones may have been introduced by the amplification process during library generation. The binding of the parental antibodies and their Y52A/Y53A variants to full-length IsdB was determined by SPR-based biosensor binding analysis at 37 °C. The binding of the parental antibodies to the full-length IsdB Y165R variant was determined by ELISA (n=2).