The characteristics of pre-existing humoral imprint determine efficacy of S. aureus vaccines and support alternative vaccine approaches

Abstract

The failure of the Staphylococcus aureus (SA) IsdB vaccine trial can be explained by the recall of non-protective immune imprints from prior SA exposure. Here, we investigate natural human SA humoral imprints to understand their broader impact on SA immunizations. We show that antibody responses against SA cell-wall-associated antigens (CWAs) are non-opsonic, while antibodies against SA toxins are neutralizing. Importantly, the protective characteristics of the antibody imprints accurately predict the failure of corresponding vaccines against CWAs and support vaccination against toxins. In passive immunization platforms, natural anti-SA human antibodies reduce the efficacy of the human monoclonal antibodies suvratoxumab and tefibazumab, consistent with the results of their respective clinical trials. Strikingly, in the absence of specific humoral memory responses, active immunizations are efficacious in both naive and SA-experienced mice. Overall, our study points to a practical and predictive approach to evaluate and develop SA vaccines based on pre-existing humoral imprint characteristics.

Keywords: S. aureus; antibody interference; immune imprinting; immunization; original antigenic sin; vaccine.

Graphical abstract

Figure 1

Quantitative and functional assessments of human and mouse anti-SA humoral imprints (A) Total and antigen-specific IgG corresponding to 10 proposed vaccine candidates from 9 healthy human donors. Ranking of human donors is based on total IgG per donor. Ranking of antigens is based on total quantity per antigen among all donors from lowest to highest. Value in each cell corresponds to titer in μg/mL serum. (B) Post-challenge bacterial burden in spleen of naive C57BL/6 mice adoptively transferred 25 μg total purified human IgG or PBS and then challenged i.p. with SA. (C) Skin lesion size 2 days post-infection (dpi) in CD1 mice adoptively transferred 25 μg total purified human IgG or PBS and then challenged i.d. with SA. (B and C) Each point represents an individual mouse. Line or bar corresponds to the median. (D) In vitro assessment of relative antigen-specific antibody function by OPK (anti-IsdB, anti-IsdA, anti-ClfA) and toxin neutralization (anti-Hla and anti-LukE). 10 μg/mL purified antibodies were used in OPK. Results are normalized to their respective control using normal mouse IgG (mIgG). The color of each cell corresponds to the mean of 8 technical replicates from two independent experiments. Statistically significant p values are noted within the cell. OPK made use of murine bone-marrow-derived neutrophils. (E) Mouse antigen-specific IgG corresponding to 10 proposed vaccine candidates over 7 weeks. Dashed line denotes time of SA exposure. Values in each cell correspond to the median level from n = 5 mice. Ranking of antigens is based on average titer among the time points from lowest to highest. (F) Post-challenge bacterial burden in spleen of naive C57BL/6 mice adoptively transferred 25 μg total purified mIgG from SA-infected or IsdB-vaccinated donor mice. The mice were subsequently challenged i.p. with SA. (G) Skin lesion size 2 dpi in CD1 mice adoptively transferred 25 μg total purified mIgG from SA-infected or IsdB-vaccinated donor mice. The mice were subsequently challenged i.d. with SA. (F and G) Each point represents an individual mouse. Line or bar corresponds to the median. (H and I) In vitro assessment of relative antigen-specific antibody function by OPK (anti-IsdB, IsdA, and ClfA) (H) or toxin neutralization (anti-Hla and LukE) (I) with antibodies purified from 3× SA or naive vaccination. 10 μg/mL purified antibodies were used in OPK or toxin-neutralization assay. Results are normalized to the respective mIgG control. OPK made use of murine bone-marrow-derived neutrophils. Unless otherwise stated, C57BL/6 mice were used. Each data point represents an individual mouse (B, C, F, and G); bar corresponds to the mean (B and C). Bar corresponds to the mean of 8 technical replicates from 2 independent experiments (H and I). n.s., not significant, ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001; Student’s t test (I) or one-way ANOVA (A–H).

Figure 2

Active immunizations targeting toxin antigens are protective in SA-experienced mice (A) Schematic of experimental design: naive mice are injected i.p. with PBS or 2.5 × 10⁷ CFU SA once every 7 days × 3 weeks. Seven days following the last injection, SA-infected (or PBS-injected) mice are immunized with Alum or Alum/IsdA i.p. weekly × 3. Seven days after the last immunization, the mice are challenged with 2.5 × 10⁷ CFU SA. Bacterial burden in spleen and kidneys are enumerated after 24 h. (B) IsdA-specific titers from n = 10 naive and SA-experienced mice 7 days after the last SA exposure. (C) Post-challenge bacterial burden in spleen of mock- or IsdA-immunized naive or SA-experienced mice. (D) In vitro assessment of relative protective serum function by OPK. 10 μg/mL purified antibodies were used in OPK. Results are normalized to mIgG control. Bar corresponds to the median. Each point represents serum from an individual mouse. (E) Post-challenge bacterial burden in spleen of naive C57BL/6 mice adoptively transferred 100 μL serum from mock- or IsdA-immunized naive or SA-experienced mice. (F and I) LukE and Hla-specific titers from n = 10 naive and SA-experienced mice 7 days after last SA exposure. (G and J) Post-i.p. challenge bacterial burden in mock- or (LukE- or Hla_(H35L)-)vaccinated naive or SA-experienced mice. (H) Post-challenge bacterial burden in spleen of naive mice adoptively transferred serum from mock- or LukE-vaccinated mice and then challenged i.p. with SA. (K) Skin lesion size 2 dpi in mock- or Hla-vaccinated naive or SA-infected mice challenged i.d. with SA. (L) Skin lesion size 2 dpi in mice adoptively transferred serum from mock- or Hla_(H35L)-vaccinated mice. Unless otherwise stated, C57BL/6 mice were used. Bar represents group median, and error bars represent means ± SD (B, F, and I). Each point represents an individual mouse (C, E, G, H, and J–L); line or bar corresponds to median (C, E, G, H, and J–L). Bar corresponds to the median (D). n.s., not significant, ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001; Student’s t test (B, F, and I) or one-way ANOVA (C–E, G, H, and J–L).

Figure 3

Pre-existing human anti-Hla antibodies blunt the efficacy of the anti-Hla monoclonal antibody suvratoxumab (A) Neutralization of toxin-mediated THP-1 cytolysis using 0.05–100 μg purified anti-Hla human antibodies or suvratoxumab. Each point represents the mean of 4 technical replicates. (B) Anti-Hla antibodies binding to recombinant Hla in the presence or absence of 4 M urea treatment. Bar corresponds to the mean of 6 technical replicates from two independent experiments. (C) Left: retention of purified human anti-Hla antibody (10 μg/mL) binding to rHla after washing and addition of competing suvratoxumab (30 μg/mL) or PBS; right: retention of suvratoxumab (10 μg/mL) binding to rHla after washing and addition of competing purified human antibodies (30 μg/mL) or PBS. Bar corresponds to the mean of 3 technical replicates. (D) Skin lesion size 2 dpi in CD-1 mice passively immunized with anti-Hla antibodies (50 μg/mouse), or PBS 1 h prior to infection, followed by suvratoxumab treatment (10 μg/mouse) or PBS 1 h post-infection. Each point represents an individual mouse. Bar corresponds to the median. Error bars represent means ± SD; n.s., not significant, ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001; one-way ANOVA (B–D).

Figure 4

Subdominant SA CWAs induce protective immunity in SA-experienced mice (A and E) ClfA-, SdrE-, and EsxAB-specific titers from n = 10 naive and SA-experienced mice 7 days after the last SA exposure. Bar corresponds to the median. Error bar corresponds to the range. (B and F) Schematic of experimental design: naive mice were injected i.p. with PBS or 2.5 × 10⁷ CFU SA once every 7 days × 3 weeks. Seven days after the last injection, SA-infected (or PBS-injected) mice were immunized with Alum or Alum/ClfA i.p. weekly × 3. Seven days after the last immunization, the mice challenged with SA or sera were harvested and injected i.v. into naive recipient mice, followed by challenge with 2.5 × 10⁷ CFU SA. Bacterial burdens in spleen and kidneys were enumerated after 24 h. Post-challenge bacterial burden in spleen of mock- or ClfA-, SdrE-, or EsxAB-vaccinated naive or SA-experienced mice is shown. Bar corresponds to the median. (C) Post-challenge bacterial burden in spleen of mice adoptively transferred serum from mock- or ClfA-vaccinated naive or SA-experienced mice. (D and G) In vitro assessment of relative protective serum anti-ClfA, anti-SdrE, and anti-EsxAB antibody function by OPK. Results are normalized to mock (mIgG) control. C57BL/6 mice were used. Bar represents group median; error bars represent means ± SD (A–E). Each point represents an individual mouse (B, C, and F); bar corresponds to the median, and dashed lines indicate the limit of detection (LOD) (B, C, and F). Bar represents the mean (D and G). n.s., not significant, ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001; Student’s t test (A and E) or one-way ANOVA (B–D, F, and G).

Figure 5

Naturally occurring human anti-ClfA antibody levels variably impact the anti-ClfA monoclonal antibody tefibazumab (A) Anti-ClfA IgG titer before and after depletion through a ClfA-antibody-adsorbing column. (B) Tefibazumab (1 μm/mL) binding to recombinant ClfA in the presence or absence of competing purified anti-ClfA human antibodies (10 μg/mL). Bar corresponds to the mean of 3 technical replicates. (C and D) Schematic of experiments. Naive mice were injected with 100 μL human sera or human sera depleted of ClfA antibodies i.v., administered 300 μg tefibazumab i.p. after 17 h, and then infected with SA 1 h after tefibazumab injection. Experiment using whole human sera (C) or human sera depleted of ClfA antibodies (D). (E) Composite analysis of association between median antibody titer and vaccine-mediated protection in SA-experienced mice. The x axis represents the median antibody titer (μg/mL) 7 days after the last infection. The y axis represents relative protection (%) in vaccinated SA-experienced mice calculated as the ratio between vaccine efficacy in SA-experienced mice and vaccine efficacy in naive mice. C57BL/6 mice were used. Bar represents group median; error bars represent means ± SD (B). Each point represents an individual mouse (C and D); bar corresponds to the median (C and D). n.s., not significant, ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001; Student’s t test (C and D) or one-way ANOVA (B, C, and D).