Acute infectious mononucleosis generates persistent, functional EBNA-1 antibodies with high cross-reactivity to alpha-crystalline beta

Abstract

We investigate the magnitude, specificity, and functional properties of Epstein-Barr virus (EBV) nuclear antigen 1 (EBNA-1)-specific antibodies in young adults over the course of primary infection. EBNA-1-specific binding antibodies, as well as antibodies capable of antibody-dependent cellular phagocytosis (ADCP) and antibody-dependent complement deposition (ADCD), are detected. These antibodies primarily target a region of EBNA-1 known to elicit cross-reactive antibodies to several self-peptides. Higher EBNA-1 binding and ADCD antibodies are observed in individuals with at least one HLA-DRB1^∗15:01 allele. Alpha-crystallin beta (CRYAB) binding and complement-fixing antibodies are detected at 6 months and 1 year following infectious mononucleosis, and CRYAB antibodies are resistant to denaturation, consistent with an affinity-matured response. Blocking experiments show that CRYAB antibodies are cross-reactive with EBNA-1. Altogether, high levels of functional EBNA-1 antibodies are generated in primary EBV infection, some of which are cross-reactive with CRYAB. Further investigation is warranted to determine whether these responses contribute to autoimmunity.

Keywords: CP: Immunology; CP: Microbiology; EBNA-1; EBV; HLA-DRB1(∗)15:01; alpha-crystalline beta; complement deposition; infectious mononucleosis; molecular mimicry; multiple sclerosis.

Figure 1.. IgG1 and IgG3 binding to EBNA-1 C-terminal peptides

(A) Violin and boxplot showing IgG1 responses toward EBNA-1 C-terminal domain peptides at acute presentation (n = 97), 6 weeks (n = 67), 6 months (n = 30), and 1 year (n = 67) post-IM diagnosis. EBV-seropositive (SP-noHx [seropositive without a history of IM], n = 30; and SP-Hx [seropositive with a history of IM], n = 20) and EBV-seronegative (SN; n = 10) controls are also included. The y axis units are IgG1-binding levels quantified through median fluorescence intensity (MFI) as arbitrary units (A.U.). (B) Overall IgG1-binding heatmap to regions of EBNA-1. Shown on the right-hand side are the peptides used. Influenza hemagglutinin (HA) is used as a positive control, and Ebolavirus glycoprotein (GP) is used as a negative control. At the top are the time points analyzed for the IM cohort, as well as SP and SN individuals as controls. Binding is shown as a fraction of maximum row binding, with the heatmap legend shown on the far right. (C) Same as (A) but for IgG3 responses to EBNA-1 C-terminal domain peptides. (D) Same as (B) but for IgG3-binding heatmap to regions of EBNA-1. For all comparisons, statistical significance is indicated as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 using a Wilcoxon test followed by Bonferroni correction for multiple comparisons. Statistical comparisons were conducted across all time points; however, as acute and 6 weeks post-infection did not significantly differ, only acute comparisons and p values were retained to avoid figure crowding; non-significant comparisons are not shown.

Figure 2.. The presence of the HLA DRB1*15:01 allele enhances EBNA-1 IgG1 responses

(A) Violin and boxplot showing IgG1 responses toward EBNA-1 C-terminal domain peptides (aa 365–420, 377–459, and 393–448) at acute, 6 months, and 1 year post-IM. Groups were distinguished based on being positive (blue) or negative (red) for at least 1 DRB1*15:01 allele. Shown for reference are SP individuals and SP controls in individuals stratified by HLA DRB1*15:01 status. (B) Fold changes in IgG1 levels at 1 year post-IM relative to acute to EBNA-1 C-terminal domain peptides (aa 365–420, 377–459, and 393–448) were plotted using box and whisker plots for individuals who were positive (blue) or negative (red) for at least one DRB1*15:01 allele. This comparison was made to control potential baseline/acute differences between the groups. The numbers of individuals in the IM cohort with at least one HLA-DRB1*15:01 allele: 19 of 97 participants (20%) at the acute visit, 8 of 30 (27%) participants at 6 months, and 11 of 67 (16%) participants at 1 year. In the SP individuals, there are 11 HLA-DRB1*15:01-positive and 38 HLA-DRB1*15:01-negative participants. For all comparisons, statistical significance is indicated as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 using a Wilcoxon test followed by Bonferroni correction for multiple comparisons. Non-significant comparisons are not shown.

Figure 3.. FcγRIIA- and FcγRIIIA-binding antibodies targeting EBNA-1 peptides persist up to 1 year post-IM

(A) Violin and boxplot showing FcγRIIA-binding antibody responses toward EBNA-1 C-terminal domain peptides at acute presentation (n = 97), 6 weeks (n = 67), 6 months (n = 30), and 1 year (n = 67) post-IM diagnosis. EBV-seropositive (SP-noHx, n = 30 and SP-Hx, n = 20) and EBV-seronegative (SN; n = 10) controls are also included as controls. The y axis units are FcγRIIA-binding antibody levels quantified through MFI as arbitrary units (A.U.). (B) Overall FcγRIIA-binding antibody heatmap to regions of EBNA-1. Shown on the right-hand side are the peptides used. Influenza HA is used as positive control, and Ebolavirus GP is used as a negative control. On top are the time points analyzed for the IM cohort, as well as SP and SN individuals as controls. Binding is shown as a fraction of maximum row binding, with the heatmap legend shown on the far right. (C) Same as (A) but for FcγRIIIA-binding antibody responses toward EBNA-1 C-terminal domain peptides. (D) Same as (C) but for FcγRIIIA-binding antibody heatmap to regions of EBNA-1. For all comparisons, statistical significance is indicated as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 using a Wilcoxon test followed by Bonferroni correction for multiple comparisons. Statistical comparisons were performed across all time points; however, as no significant differences were found between acute and 6 weeks post-infection, only acute comparisons and p values were retained to avoid figure crowding; non-significant results are not shown.

Figure 4.. Antibody-dependent cellular phagocytosis is activated post-IM to the EBNA-1 C-terminal domain

Violin and boxplots showing antibody-dependent cellular phagocytosis by monocytes (ADCP) quantified through phagoscores (see STAR Methods). Phagoscores were quantified against EBNA-1 peptides (aa 377–459, 365–420, and 393–448) across different EBV infection stages: acute, 6 weeks, 6 months, and 1 year post-IM. EBV-seropositive (SP-noHx, n = 30 and SP-Hx, n = 20) and EBV-seronegative (SN) individuals are included as controls. For all comparisons, statistical significance is indicated as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 using a Wilcoxon test followed by Bonferroni correction for multiple comparisons. Statistical comparisons were conducted across all time points; however, as acute and 6 weeks post-infection did not significantly differ, only acute comparisons and p values were retained to avoid figure crowding; non-significant comparisons are not shown.

Figure 5.. Antibody-dependent complement deposition to EBNA-1 C-terminal peptides increases over time and is enhanced in DRB1*15:01-positive individuals

(A) Violin and boxplots showing antibody-dependent complement deposition (ADCD) quantified through C3 deposition units for EBNA-1 peptides (aa 377–459, 365–420, and 393–448) across different EBV infection stages: acute, 6 weeks, 6 months, and 1 year post-IM. EBV-seropositive (SP-noHx, n = 30 and SP-Hx, n = 20) and EBV-seronegative (SN) individuals are included as controls. For all comparisons, statistical significance is indicated as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 using a Wilcoxon test followed by Bonferroni correction for multiple comparisons. Statistical comparisons were conducted across all time points; however, as acute and 6 weeks post-infection did not significantly differ, only acute comparisons and p values were retained to avoid figure crowding; non-significant comparisons are not shown. (B) Overall ADCD to regions of EBNA-1. Shown on the right-hand side are the peptides used. Influenza HA is used as positive control, and Ebolavirus GP is used as negative control. At the top are the time points analyzed for the IM cohort, as well as SP and SN individuals as controls. The scale is shown as a fraction of the maximum row ADCD, with the heatmap legend shown on the far right. (C) Line graphs showing ADCD over time to EBNA-1 peptides (aa 377–459, 365–420, and 393–448) for individuals who were positive (blue) or negative (red) for at least one DRB1*15:01 allele. Statistical analysis was performed using a linear mixed model with interaction terms for EBV infection stage and DRB1*15:01 status. Significant increases in ADCD were observed at 1 year compared to acute, 6 weeks, and 6 months for all peptides (p < 0.001). Interaction effects at the 1-year time point revealed that DRB1*15:01-positive individuals showed a more pronounced ADCD response compared to those negative for the peptides aa 377–459 (p = 0.000737) and 365–420 (p = 0.000875). No significant interaction effects were observed at earlier time points.

Figure 6.. EBNA-1 antibodies are cross-reactive to CRYAB and retain avidity and ADCD to the self-peptide

(A) Violin and boxplots showing IgG binding to CRYAB across different EBV infection stages: acute, 6 weeks, 6 months, and 1 year post-IM. EBV-seropositive (SP-noHx, n = 30 and SP-Hx, n = 20) and EBV-seronegative (SN) individuals are included as controls. (B) Violin and boxplots showing IgG binding to CRYAB in the presence (green) or absence (brown) of 3 M urea across different EBV infection stages: acute, 6 weeks, 6 months, and 1 year post-IM. EBV-SP and EBV-SN individuals are included as controls. (C) Violin and boxplots showing ADCD to CRYAB across different EBV infection stages: acute, 6 weeks, 6 months, and 1 year post-IM. EBV-SP and EBV-SN individuals are included as controls. See also Figure S6, where we show data for GlialCAM and Ano2. For all comparisons, statistical significance is indicated as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 using a Wilcoxon test followed by Bonferroni correction for multiple comparisons. Comparisons were made for all time points, but as there were no significant differences between acute and 6 weeks post-infection, only acute comparisons and p values are shown to avoid figure crowding; non-significant comparisons are not shown.

Figure 7.. EBNA-1 specific antibodies from post-IM individuals cross-react with CRYAB

(A) Schematic of the blocking assay: EBNA-1 peptides were used to deplete Igs targeting the viral protein. If these antibodies are cross-reactive with CRYAB, then the EBNA-1 immunodepletion should result in a loss of CRYAB binding. As a control, the same serum was blocked with flu-HA peptides to control non-specific immunodepletion. Lastly, responses to VCA were quantified for both EBNA-1 and flu-HA immunodepletions, as this is not cross-reactive with either or should not result in any binding loss. (B) Violin and boxplots depicting IgG antibody reactivity to CRYAB, flu-HA, and EBV VCA across different treatments (EBNA-1 peptides or flu-HA peptides). Samples included pooled 6 months–1 year (n = 55), EBV seropositive (SP; n = 25), and EBV seronegative (SN; n = 10). For all comparisons, statistical significance is indicated as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 using a Wilcoxon test followed by Bonferroni correction for multiple comparisons. All time points were statistically compared; however, given the lack of significant differences between acute and 6 weeks post-infection, only acute comparisons and p values were shown to avoid figure crowding; non-significant comparisons are not shown.