Prior flavivirus immunity skews the yellow fever vaccine response to cross-reactive antibodies with potential to enhance … – Nature.com

YF17D vaccination induces similar neutralizing antibody titers but expands a poor-neutralizing IgG response in TBEV-pre-vaccinated individuals

To investigate the effects of pre-existing cross-reactive immunity on the response to the live YF17D vaccine, we examined a longitudinal cohort of 250 YF17D healthy young vaccinees grouped based on their previous immunization with the inactivated TBEV vaccine (cohort-1). Given that TBEV vaccination is recommended in the region where this study was conducted, a representative fraction of participants self-reported ahistory of TBEV vaccination prior (>4 weeks) to study inclusion (n=162; 64.8%). TBEV pre-immunity was verified through positive results for TBEV neutralization by plaque reduction neutralization assay (PRNT) and positive detection of anti-TBEV-DIII IgG antibodies using enzyme-linked immunosorbent assay (ELISA). Individuals with discrepancies between self-reported vaccination status and serological assays were excluded from the analysis. The final study cohort-1 comprised 139 participants pre-vaccinated against TBEV and 56 TBEV-unvaccinated individuals. Different sexes, ages, and BMI values are equally distributed in both subgroups (Fig.1; TableS1). The exact strain, number of doses, and timing of TBEV vaccinations had not been documented. Consequently, TBEV-experienced donors showed heterogeneity in their TBEV neutralizing titers prior to YF17D vaccination (Fig.2B).

A Diagram representing the longitudinal PBMC, serum, and plasma sample collection of 250 participants. Samples were collected at baseline (immediately before vaccination) and on days 7, 14 and 28 post vaccination. Prepared with Biorender (www.biorender.com) B Flow chart of cohort members grouping according to TBEV pre-vaccination status. 139 individuals self-reported having received at least one TBEV-vaccine dose and contained neutralizing antibodies and anti-TBEV IgG at baseline. 56 TBEV-unvaccinated individuals were classified based on a self-reported negative vaccination history validated by the absence of detectable anti-TBEV IgG and neutralizing capacity at baseline C Histogram depicting the age distribution of the 139 TBEV-vaccinated participants (in orange) and the 56 TBEV-unvaccinated donors (in blue). D Table summarizing cohort-1 characteristics for all 250 individuals and separated according to TBEV-pre-vaccination status.

A YF17D 80% neutralization titer at day 28 post-vaccination for TBEV-vaccinated (n=137) and unvaccinated (n=56) individuals. B TBEV 80% neutralization titer before and 28 days after YF17D vaccination for TBEV-vaccinated individuals (n=137) and TBEV-unvaccinated individuals (n=56). C Longitudinal YF17D virion-specific IgM response in TBEV pre-vaccinated (baseline, n=132; day 7, n=128; day 14, n=128; day 28, n=129) and TBEV unvaccinated donors (baseline, n=54; day 7, n=54; day 14, n=54; day 28, n=53). D YF17D virion-specific IgM titer on days 14 and 28 in IgG-depleted serum for TBEV pre-vaccinated (n=56) and TBEV unvaccinated donors (n=28). E Longitudinal YF17D anti-E protein-specific IgG titers for TBEV-vaccinated (baseline, n=136; day 7, n=131; day 14, n=132; day 28, n=133) and TBEV-unvaccinated donors (baseline, n=52; day 7, n=52; day 14, n=52; day 28, n=52). F, G Plasmablasts and total sE-specific longitudinal B-cell response quantified by ELISpot and depicted as spot-forming units (SFU) in 100,000 PBMC (n=10 TBEV-vaccinated and n=9 TBEV-unvaccinated donors). Spot pictures are shown for a representative example of a TBEV-pre-vaccinated and unvaccinated individual. Significance compares B cell counts between groups on days 14 (F) and 28 (G). H, I Spearman correlation between the YF17D polyclonal neutralizing titer of sera on day 28 with the IgG titer (TBEV-vaccinated n=133 and TBEV-unvaccinated donors n=52) and IgM titer (TBEV-vaccinated n=129, TBEV-unvaccinated donors n=53). J Neutralization curves of undepleted polyclonal serum and IgG or IgM-depleted serum (in grey) for TBEV-pre-vaccinated (n=45 and 26, respectively) and unvaccinated individuals (n=19 and n=22, respectively). Quantification of the 80% neutralization cutoff before and after IgG (K) and IgM (L) depletions shown in J. TBEV-vaccinated participants are depicted in orange and TBEV-unvaccinated in blue. Boxplots show a horizontal line indicating the median and lower and upper hinges corresponding to the first and third quartiles. The lower and upper whiskers extend to 1.5x IQR (interquartile range) from the respective hinge. The curve fitting in J was done with local regression with a 0.95 confidence interval. Statistical significance between TBEV-vaccinated and unvaccinated individuals is shown above every comparison and was estimated with a two-sided Mann-Whitney test. Statistical significance between undepleted and depleted samples (K and I) and between timepoints (D) was calculated with a two-sided Wilcoxon signed-rank test. P values above 0.05 are considered non-significant (ns).

The neutralizing antibody titer against YF17D, commonly used as a correlate of vaccine-induced protection, was equally strong at day 28 post-vaccination (pv) in both flavivirus-nave and experienced individuals. This result indicates that TBEV-pre-immunity does not impair the neutralizing antibody response to YF17D (Fig.2A). Conversely, YF17D immunization did not alter the neutralizing activity against TBEV, indicating that the YF17D vaccine did not generate cross-neutralizing antibodies to TBEV (Fig.2B).

The presence of YF17D virion-specific IgM in serum was measured longitudinally by ELISA. The IgM titer reached a plateau between day 14 and 28 pv and was comparable in both groups of vaccinees as confirmed in IgG-depleted serum samples (Fig.2C, D). Unlike the IgM response, participants with a prior TBEV exposure had YF17D cross-reactive IgG antibodies already at baseline and, upon vaccination, the IgG titer was further boosted resulting in a 100-fold higher titer compared to the TBEV-unvaccinated group. The same dynamic was observed for E protein (Fig.2E) and full-virion specific IgG (Fig.S1B, C). The IgG titer continued to increase from day 14 to day 28 pv, at which timepoint all the study participants had seroconverted. While TBEV-pre-vaccinated donors showed an anti-E IgG response already at day 14 pv, only a fraction of the TBEV-unvaccinated participants generated detectable anti-E IgG levels at that timepoint, suggesting an earlier response to YF17D in individuals with immune experience (Fig.2E).

To assess the generation of vaccine-specific B cells, we implemented a soluble (s)E-specific ELISpot assay. The number of plasmablasts was quantified directly ex vivo and the total number of sE-specific B cells was measured following the differentiation of B cells into antibody-secreting cells. sE-specific IgG secreting plasmablasts peaked at day 14 pv in TBEV-pre-vaccinated participants but were below detection for unvaccinated donors (Fig.2F). The total amount of sE-specific B cells was in line with the IgG levels measured in serum with a significantly higher B cell number in flavivirus-experienced individuals. Likewise, sE-specific B cells were detected earlier, on day 14, in TBEV-pre-immunized individuals (Fig.2G).

Collectively, these results indicate that TBEV pre-immunization does not hinder the response to YF17D. TBEV-pre-immunized individuals have cross-reactive IgG antibodies to YF17D and experience an earlier and stronger IgG response.

The observed differences in the IgG titer did not correspond with the similar neutralization capacity between both groups of vaccinees. When groups were analyzed separately, the IgG titer at day 28 correlated weakly with neutralization (R=0.29, p 0.04 and R=0.34, p<0.0001), whereas there was a stronger association with the IgM levels (R=0.55, p<0.0001 and R=0.64, p<0.0001 for TBEV-unvaccinated and experienced individuals, respectively) (Fig.2H, I). We depleted serum of the IgM or IgG fractions in a subgroup of the study cohort-1 to precisely define the contribution of IgG and IgM antibodies to neutralization. We confirmed that depletions were successful, without loss of the alternative antibody fraction (Fig.S1D). We observed that the IgM fraction was the main mediator of virus neutralization, accounting for approximately 75% of the neutralizing titer on day 28 (Fig.2J, L). Consistently, IgG depletion led only to a 25% loss of neutralization capacity on days 14 and 28 (Fig.2J, K, and S1D,E). IgM-depleted sera of TBEV-pre-immunized individuals showed a significantly higher neutralizing capacity, indicating that the increased IgG titer contributed partly to virus neutralization (Fig.2L). There was no significant difference in the IgM antibody titers and IgM-mediated neutralization between groups (Fig.2D, J, K).

Thus, on day 28 the IgM fraction is primarily responsible for the neutralization capacity and is equally strong regardless of the pre-vaccination status. The IgG fraction can mediate neutralization, but the boosted response in TBEV-experienced individuals is predominantly directed towards poorly-neutralizing epitopes.

As reported by Chan et al. 2016, cross-reactive non-neutralizing or sub-neutralizing IgG antibodies can mediate FcR engagement and increase vaccine immunogenicity via ADE of YF17D virus infection20. Given that TBEV-pre-vaccinated individuals had YF17D cross-reactive antibodies at baseline, we measured whether they could facilitate YF17D infection of FcR expressing cell lines (THP-1 and K562). We observed an enhanced infection of a Venus-fluorescent YF17D virus in both cell lines in the presence of serum from TBEV-pre-immunized donors. ADE was IgG-dependent, as it was absent in IgG-depleted serum but not in IgM-depleted serum, and it was inhibited in the presence of FcR-blocking antibodies (Fig.3A and S2A, B, C). YF17D infection enhancement was observed exclusively with sera from TBEV-experienced individuals (Fig.3B) which showed the highest enhancing titer, calculated as area under the curve (AUC) (Fig.3C).

A Flow cytometric determination of venus-YF17D virus infection of THP-1 cells in the absence or presence of cross-reactive serum from a TBEV-vaccinated individual. The conditions tested include polyclonal serum alone or in combination with FcR-blocking antibodies and IgM or IgG-depleted serum. B ADE of YF17D mediated by study participants serum (n=132 TBEV-vaccinated, n=53 TBEV-unvaccinated individuals). Virus infection was quantified in combination with serially diluted serum and was normalized against the enhancement of a 1:20 diluted enhancing-serum control carried for all measurements. Thick dashed line indicates the mean of the different just-virus controls and dotted lines define +/ 1SD. The curve was fitted with local regression with a 0.95 confidence interval. C Quantification of the enhancing titer as AUC of the normalized virus infection across serial dilutions shown in B. D Comparison of anti-YF17D-DIII specific IgG titer at day 28 post-vaccination for both TBEV groups (n=36 TBEV-unvaccinated, n=117 TBEV-vaccinated individuals). Longitudinal quantification of anti-YF17D-DIII (E, n=97 donors) and anti-TBEV-DIII (F, n=114 donors) specific IgG in TBEV-experienced individuals. G Paired comparison of the DIII-specific IgG fold-change between day 28 and day 0 for TBEV and YF17D (n=76 pairs). H Spearman correlation in TBEV-experienced individuals of baseline anti-E IgG and enhancing titers versus YF17D vaccine-induced neutralizing antibody titer at day 28, anti-sE and anti-DIII IgG titers and baseline anti-YF17D-E and anti-YF17D-DIII IgG titers. Color intensity reflects the Spearman correlation coefficient. I Comparison of the first (n=34) and fourth (n=33) quartile group of YF17D-sE specific IgG at baseline and the anti-YFDIII IgG on baseline and day 28, the anti-YF17D-sE IgG titer on day 28 and the anti-YF17D neutralizing antibody titer on day 28 (high vs. low quartile n=24/23, 32/28, 33/32, 34/33 respectively). J Comparison of the first (n=33) and fourth (n=33) quartile group of the Enhancing titers at baseline and the anti-YFDIII IgG on baseline and day 28, the anti-YF17D-sE IgG titer at baseline and day 28 and the anti-YF17D neutralizing antibody titer on day 28 (high vs low quartile n=25/22, 28/30, 32/33, 32/33, 33/33, respectively). TBEV-vaccinated participants are depicted in orange and TBEV-unvaccinated in blue. Boxplots show a horizontal line indicating the median and lower and upper hinges corresponding to the first and third quartiles. The lower and upper whiskers extend to 1.5x IQR from the respective hinge. Statistical significance between TBEV-vaccinated and unvaccinated individuals (D) and between high and low quartiles (I, J) is shown above every comparison and was estimated with a two-sided Mann-Whitney test. Statistical significance in EG was calculated with a two-sided Wilcoxon signed-rank test. P values above 0.05 are considered non-significant (ns).

In addition, we quantified the IgG fraction targeting DIII. DIII is often used for serological diagnosis35 and although cross-reactive epitopes have been described36,37, responses to DIII are generally virus-specific38. As observed for anti-sE IgG antibodies, TBEV-pre-vaccinated individuals had a stronger IgG response against YF17D-DIII (Fig.3D). We then compared the fold change in anti-YF17D-DIII and anti-TBEV-DIII IgG titers between baseline and day 28 pv. Interestingly, the strong expansion of anti-YF17D-DIII-reactive IgG (10-100-fold) contrasted with the moderate increase in TBEV-DIII reactive IgG (<10-fold) (Fig.3EG). Since the IgG fraction targeting TBEV-DIII was not expanded to the same extent, we conclude that YF17D is not only boosting cross-reactive TBEV-induced memory responses but also triggers antibodies towards previously unseen, non-cross-reactive epitopes in DIII, potentially via enhanced immunogenicity.

The correlation between the baseline levels of sE-specific IgG and the baseline enhancing titer with the post-immunization IgG response to sE and DIII, as well as the post-vaccination neutralizing titers, suggests that ADE of YF17D infection could mediate an increase in vaccine immunogenicity (Fig.3H, S2D, E, F). To gain insight into the size effects of ADE of YF17D infection on vaccine immunogenicity, we grouped the vaccinees into quartiles based on their enhancing and baseline IgG titers. This approach removed individuals with intermediate TBEV-vaccine-induced IgG levels from the analysis and therefore improved the precise identification of true differences between the high and low vaccine enhancers. Participants in the highest quartile of the enhancing titer had significantly higher neutralizing titer and IgG levels against sE and DIII (Fig.3I) than the participants in the lowest quartile. The same associations were observed with baseline IgG quartiles (Fig.3J). Thus, within the TBEV-pre-vaccinated group, ADE of YF17D infection was associated with increased vaccine immunogenicity.

Altogether, in addition to an anamnestic response, these results suggest that TBEV-pre-immunity may increase the magnitude and breadth of the humoral response to YF17D vaccine facilitated by ADE of YF17D infection.

The IgG response in both groups was investigated for cross-reactivity with other members of the Flaviviridae family. Using an indirect immunofluorescence test, the IgG and IgM binding to ZIKV, JEV, WNV, TBEV, YFV, and all serotypes of DENV was measured in a cohort subset (Fig.4, S3). At baseline, TBEV pre-vaccination induced an IgG response cross-reactive with all flaviviruses at similar magnitude. This pan-flavivirus cross-reactive IgG signature was boosted upon vaccination with the YF17D vaccine. In contrast, the YF17D vaccine induced an IgG response that targeted uniquely YFV in unvaccinated individuals (Fig.4A). Consistent with previous studies, the IgM signature was YFV-specific and could not be detected at baseline (Fig.4B).

IgG (A) and IgM (B) subtype cross-reactivity evaluation before and after YF17D vaccination using an indirect immunofluorescent test for a panel of nine human-pathogenic flaviviruses: DENV 14, ZIKV, WNV, JEV, YFV and TBEV. A subgroup of n=39 individuals was tested (Fig.S3), out of which n=15 were TBEV-unvaccinated and n=24 TBEV-experienced. Bars indicate titer mean and dots reflect the antibody amounts as serum dilution end-point titers. TBEV-vaccinated participants are depicted in orange and TBEV-unvaccinated in blue.

These results highlight the capacity of the vaccine strain of YFV to prevent a cross-reactive response when administered to flavivirus-nave individuals.

Depending on previous flavivirus exposure, YF17D induces a differential cross-reactivity pattern while eliciting a comparable neutralizing response. We hypothesized that the neutralizing capacity is predominantly driven by antibodies targeting quaternary dimeric epitopes, equivalent to EDE (EDE-like), and the FL-proximal region, whereas cross-reactive antibodies target the immunodominant FLE. To better understand this change in the immunodominance, we designed a set of recombinant sE protein mutants for the study of the IgG response to different epitopes (Fig.5A). Besides the monomeric sE protein containing all three ectodomains and variants consisting of either only DI-II or only DIII, we designed constructs displaying quaternary dimeric epitopes to reproduce the epitope landscape of YF17D more realistically. The substitution S253C in DII allows the formation of an inter-protomer disulphide bond across the two sE protomers, generating a covalently bound dimer39. This construct (hereinafter referred to as breathing-dimer) retains the ability to oscillate and exposes EDE-like dimeric epitopes, FLE, and the FL-proximal region. Furthermore, a W101H substitution was introduced in the breathing-dimer setting to disrupt the FLE (breathing-dimerW101H) (Fig.5A). In addition, we produced a locked-dimer E protein by introducing a double substitution L107C and T311C following the strategy used by Rouvinski et al (2017) and Slon-Campos et al (2019) with DENV and ZIKV. This construct displays quaternary epitopes on a pre-fusion dimeric structure that is bridged with two disulfide bonds between DI and DIII of opposing protomer units (Fig.S4A, B). Correct folding and epitope exposure of the recombinant proteins were verified by binding with specific antibodies using ELISA and SEC analysis (Fig.S6). Both the FLE KO construct (breathing-dimerW101H) and the locked dimer failed to bind antibodies recognizing the FLE but were recognized by an antibody binding DII outside of the FL. Both sE monomer and the breathing-dimer were recognized by fusion loop specific antibodies (Fig.S6).

A Recombinant proteins for the dissection and functional analysis of different antibody specificities. The illustration depicts the envelope protein ectodomains (sE), DI-II, and DIII produced separately as well as recombinantly produced dimeric structures. The table summarizes the epitopes displayed by the protein antigens used. B IgG endpoint titer quantification for breathing-dimer and breathing-dimerW101H specificities by ELISA at baseline (n=23 TBEV-vaccinated donors) and day 28 (n=24 TBEV-vaccinated, n=20 TBEV-unvaccinated individuals). C Antibody binding competition to sE of participants serum with the FL-mab 4G2 at baseline (n=55 TBEV-vaccinated) and day 28 (n=55 TBEV-vaccinated and n=43 TBEV-unvaccinated donors). The percentage of remaining binding is calculated by comparing the binding signal with and without 4G2 as competitor D Spearman correlation between antibody binding loss estimated with the 4G2 competition assay (C) and with the breathing-dimerW101H (B) (n=23 pairs of baseline samples, n=23 pairs of TBEV-vaccinated and n=20 TBEV-unvaccinated of day 28 samples) E Longitudinal quantification of IgG-producing B-cells specific for breathing-dimer and breathing-dimerW101H (n=10 TBEV-vaccinated and n=9 TBEV-unvaccinated donors). Units represent spot-forming units per 100.000 PBMC. F Table describing the antigens used for antigen-specific IgG depletions and the expected specificities of the remaining undepleted fraction used for YF17D neutralization assays. G YF17D neutralization titers (50% cutoff) of IgM-depleted and antigen-specific-depleted sera as explained in F (n=9 TBEV-vaccinated and n=9 TBEV-unvaccinated participants). TBEV-vaccinated participants are depicted in orange and TBEV-unvaccinated in blue. Envelope structure accession number (PDB: 6IW4) was edited using Pymol. Individual selection is shown in Supplementary Fig.3 (SF3). Boxplots display a horizontal line indicating the median and lower and upper hinges corresponding to the first and third quartiles. The lower and upper whiskers extend to 1.5x IQR from the respective hinge. Barplots in C and G indicate the mean and the error bars the standard error of the mean. Statistical significance between TBEV-vaccinated and unvaccinated individuals (C) was estimated with a two-sided Mann-Whitney test. Statistical significance in B, E and G was calculated with a two-sided Wilcoxon signed-rank test. P values above 0.05 are considered non-significant (ns).

The comparison between the breathing-dimer and breathing-dimerW101H-specific IgG titers serves to measure the fraction targeting theFLE. For TBEV-experienced individuals, the antibody fraction targeting the breathing-dimer was significantly reduced in baseline samples and at day 28 by the W101H mutation (45 and 64% reduction respectively) whereas TBEV-nave individuals showed no significant difference (Fig.5B). Additionally, the sE-specific IgG titer was quantified in binding competition assays with 4G2 (pan-flavivirus FLE-specific mAb) and 2D12 (YFV-neutralizing, non-cross-reactive anti-E mAb) (Fig.5C, S5A). As anticipated, TBEV-pre-exposed vaccinees lost over 80% of the sE-IgG binding fraction at day 28 in competition with 4G2, while flavivirus-nave individuals ranged from 0 to 60% binding loss, demonstrating that FLE is a dominant binding site for the antibody response in TBEV-experienced individuals, but not in TBEV-unvaccinated individuals. Likewise, baseline antibodies also competed with 4G2 for binding (Fig.5C). Consistently, binding loss caused by the W101H mutation and competition with 4G2 correlated with each other (R=0.65, p=0.0012), serving as cross-validation of these assays to quantify FLE antibodies in serum (Fig.5D).

Additionally, we performed an ELISpot assay to quantify the number of epitope-specific circulating B cells. We observed that the number of breathing-dimer-specific B cells was larger for TBEV-pre-immunized compared to TBEV-unvaccinated participants. Approximately 50% of the specific-B cells in TBEV-pre-immunized donors (100 cells/100.000 lymphocytes) required the unmutated FL for binding (Fig.5E). Consistently with serum antibody levels, TBEV-unvaccinated individuals had lower numbers of breathing-dimer specific B cells, although a relevant fraction produced antibodies requiring FL for binding. These B cells secrete antibodies that may be binding dimeric structures or FL-proximal regions whose binding site includes amino acids located in the FL (Fig.5A, E). The locked-dimer-specific IgG and B cell response was in line with the findings showing increased responses in TBEV-experienced individuals (Fig.S4C, D).

Taken together, these results show that the IgG fraction targeting the FLE is dominant in TBEV-pre-immunized but not in TBEV-unvaccinated participants. However, the fusion loop region is a binding site for antibodies elicited in both groups.

To assess the neutralizing capacity of antibodies with different specificities, we performed antigen-specific IgG depletion from serum samples that had been pre-depleted of IgM antibodies (Methods and Fig.5F, G). Depleting the IgM fraction allowed for a more precise dissection of the main neutralizing sites targeted by the long-term, durable, IgG response. As expected, IgM removal greatly reduced the neutralizing capacity of the sera (Fig.2G). Further depletion with the breathing-dimer protein resulted in a remarkable loss of neutralizing capacity in both TBEV-experienced and nave individuals. The neutralizing titers of sera depleted with the breathing-dimerW101H antigen remained high, demonstrating that neutralizing epitopes include the FL as a binding site (Fig.5G, left panels). In fact, the main chain of the FL is part of the EDE epitope in DENV10. Depletions performed with the locked-dimer construct resulted in only a slight decrease in neutralizing capacity (Fig.S4E). Despite the occlusion of the FL in the locked-dimer, we thought this construct would deplete antibodies with dimeric specificities and would reduce greatly the serum neutralization activity. However, the direct modification of the FLE by the L107C mutation of this construct also had an impact on the integrity of the dimer epitope, affecting the ability to deplete antibodies targeting dimeric specificities. A comparable construct for dengue40, although able to bind most of the dengue EDE antibodies, showed a reduction in binding for selected EDE. Similarly, the YF17D E locked-dimer construct may have failed to deplete the principal antibody fraction responsible for the virus neutralization (Fig.S4).

sE monomer cannot be used to deplete exclusively monomer-specific IgG antibodies as dimer-specific antibodies may assemble sE monomers together into dimers and therefore, this construct would also deplete antibodies with dimer-specificities41. To ensure the removal of antibodies with monomeric but not dimeric specificities, we then performed subsequent depletions with DI-II and DIII. Even though the depletions resulted in a clear loss of neutralizing capacity, especially in TBEV-experienced individuals, monomeric specificities only made up a minor fraction of the polyclonal neutralizing antibody response when compared to the breathing-dimer depleted sera (Fig.5G right panels).

Altogether, these results highlight the importance of dimer epitopes as the main YFV neutralizing sites and reveal that the FL is a critical component of the binding site for potent neutralizing dimeric antibodies.

Given that in TBEV-experienced individuals YF17D boosts a pan-flavivirus cross-reactive IgG response, we examined whether sera from these individuals mediate ADE of DENV and ZIKV infection using viral reporter replicon particles (VRP)42.

Interestingly, the antibody response induced by TBEV immunization was sufficient to enhance DENV and ZIKV infection. The enhancing capacity was further increased after YF17D vaccination. As expected, flavivirus-nave individuals did not facilitate DENV and ZIKV infection in vitro at baseline and the YF17D vaccine did not induce antibodies with enhancing potential. This is consistent with the absence of cross-reactive antibodies in these individuals (Fig.6A).

A Antibody-dependent enhancement of infection with DENV-2 (16681) VRPs at baseline and day 28 post-YF17D vaccination (n=23 TBEV-vaccinated, n=15 TBEV-unvaccinated individuals) B Antibody-dependent enhancement of infection with ZIKV (MR-766 African strain) VRPs at baseline and day 28 post-YF17D vaccination (n=16 TBEV-vaccinated, n=8 TBEV-unvaccinated individuals). C Dengue ADE for TBEV-vaccinated and unvaccinated individuals driven by: undepleted serum (n=22 per group), IgM-depleted serum (n=22 per group), and, as detailed in Fig.5F, IgM & antigen-specific-IgG-depleted serum (n=9 per depletion group). Relative infectivity is estimated as the normalized fold-increase of infection to an internal control carried for all the assays. Curves were fitted with local regression. In A and BTBEV-vaccinated participants are depicted in orange and TBEV-unvaccinated in blue.

To explain the antibody specificities mediating ADE, we first removed the IgM fraction and then measured the enhancing capacity after antigen-specific depletions of the remaining IgG fraction. ADE to DENV was lost in serum depleted of antibodies binding to the breathing-dimer, sE monomer (DI-II and III) or DI-II. In contrast, samples depleted with breathing-dimerW101H or locked-dimer constructs retained antibodies mediating ADE (Fig.6B, S5F). These results point to FLE-antibodies as responsible for cross-reactivity and ADE.

In conclusion, the YF17D vaccine expands FLE-antibodies with the potential to mediate ADE of DENV and ZIKV infection in vitro in TBEV-pre-immunized individuals, but not in the flavivirus-nave population.

The B cell response to the YF17D vaccine continues to mature for 6 to 9 months after vaccination29. Likewise, as the immune response advances, the antibody response undergoes diversification in terms of epitope recognition and binding affinity43. To extend our findings beyond day 28 post-vaccination, we analyzed serum samples from an independent cohort of 20 individuals collected one year after YF17D vaccination (cohort-2). Although baseline samples were unavailable, a review of the vaccination records identified that 16 individuals had received at least one TBEV vaccine dose before YF17D vaccination, while 4 were TBEV-nave. Additionally, two individuals who received the YF17D vaccine 9 and 11 years before sample collection, with no record of TBEV vaccination, were part of this cohort (Fig.7A, B).

A Diagram representing the serum sample collection of 22 participants. Prepared with Biorender (www.biorender.com). B Table summarizing cohort-2 characteristics and TBEV-pre-vaccination status. The table indicates the age at the time of vaccination. C YF17D anti-E protein-specific IgG titers in relative units (RU). D Neutralization curves of undepleted polyclonal serum and IgG- or IgM-depleted serum (in grey) for TBEV-pre-vaccinated or unvaccinated individuals. E Quantification of the 80% neutralization cutoff before and after IgM depletion. F Antibody binding competition to sE of participants serum with the FL-mAb 4G2. Percentage of remaining binding is calculated by comparing the binding signal with and without 4G2 as competitor. G Antibody-dependent enhancement of infection with DENV-2 VRPs. TBEV-vaccinated participants (n=16) are depicted in orange and unvaccinated (n=6) in blue. One-year pv samples are represented with circles and over 9 years pv with triangles. All individuals are included in every assay except for D and E (n=15 TBEV-vaccinated donors). Curve fitting in D and G was calculated with local regression. Boxplots display a horizontal line indicating the median and lower and upper hinges corresponding to the first and third quartiles. The lower and upper whiskers extend to 1.5x IQR from the respective hinge. Barplots in F indicate the mean and the error bars the standard error of the mean. Statistical significance between TBEV-vaccinated and unvaccinated individuals (C, E and F) was estimated with a two-sided Mann-Whitney test. Statistical significance between undepleted and IgM-depleted serum samples in E was calculated with a two-sided Wilcoxon signed-rank test. P values above 0.05 are considered non-significant (ns).

Similar to cohort 1, TBEV-pre-vaccinated individuals had significantly higher IgG antibody titers against the YF17D sE protein than TBEV-unvaccinated participants (Fig.7C). Of note, the difference in titers was approximately one order of magnitude, lower than what we observed on day 28. Moreover, flavivirus-nave individuals exhibited increased IgG levels compared to those at day 28 (Fig.7C). Both subgroups efficiently neutralized YF17D one-year pv, with the IgM fraction retaining significant neutralizing potential in participants sera. Surprisingly, we observed a trend of improved neutralizing titers in TBEV-unexperienced individuals compared to TBEV-pre-vaccinated individuals for both undepleted and IgM-depleted serum (Fig.7D, E). While the TBEV-unvaccinated sample size is limited, these results suggest that TBEV-pre-vaccinated individuals expanded a high titer of sE-specific antibodies with limited neutralizing potential, while flavivirus-nave individuals generated a more efficient IgG response for virus neutralization (Fig.7D, E).

In serum samples from TBEV-pre-vaccinated individuals one-year pv, the sE-IgG binding fraction competed with 4G2 mAb by over 80%. In contrast, flavivirus-nave individuals retained between 50% and 100% of sE-IgG binding in the presence of 4G2. This result, consistent with the analysis of cohort-1, confirms that the FLE is still a dominant target of the IgG response in TBEV-pre-vaccinated individuals but not in TBEV-unvaccinated donors after one-year pv (Fig.7F). Moreover, one year after YF17D vaccination, serum from TBEV-experienced donors was still capable of mediating ADE of DENV infection in vitro, while flavivirus-nave individuals did not elicit an IgG response with enhancing potential (Fig.7G).

Collectively, these results confirm our observations at day 28 post-vaccination in cohort 1 in an independent cohort sampled one year after YF17D vaccination. In conclusion, TBEV-pre-exposed donors develop a cross-reactive FLE-directed IgG response capable of mediating ADE of DENV infection. However, in flavivirus-nave individuals, YF17D primes for a non-cross-reactive yet effective neutralizing antibody response.

Original post:

Prior flavivirus immunity skews the yellow fever vaccine response to cross-reactive antibodies with potential to enhance ... - Nature.com