Breadth of neutralization induced by vYF next generation yellow fever vaccine

A novel live-attenuated yellow fever (YF) vaccine candidate, vYF, is currently in development. Previous studies in non-human primates (NHPs) have evaluated its safety, immunogenicity, and efficacy. To further characterize the immune response, this study assessed the breadth and magnitude of neutralizing antibodies (NAbs) induced by vYF. Immunogenicity was evaluated in NHPs across multiple parameters, including NAb titers against the vaccinal virus strain from 14 to 258 days post-vaccination. The breadth of neutralization was assessed using a panel of four YF virus strains selected to represent diverse phylogeny, isolation dates, and geographic origins. Titers from vYF-vaccinated NHPs were compared with those from YF-VAX–vaccinated NHPs. Two additional control strains, Stamaril ® (current vaccine) and the wild-type parent strain Asibi, were included. Amino acid (AA) sequences of viral envelope residues associated with known neutralizing epitopes were analyzed to explore potential correlates with neutralization profiles. NAb titers from vYF-vaccinated NHPs were comparable to those obtained with the Asibi strain for two of the four wild-type strains, and 0.3 to 0.5 Log lower for the remaining two. All wild-type strains were neutralized with titers at least 3-fold higher than the established correlate of protection, indicating a broad neutralization profile. Neutralization patterns were similar between vYF- and YF-VAX–vaccinated NHP sera. This study adds to the evidence that the vYF may elicit the same breadth of neutralizing antibody responses in NHPs, than YF-VAX reference vaccine.

Yellow fever (YF) is caused by a virus transmitted by mosquitoes, which has historically evoked fear, societal stigma, illness, and mortality across human populations over centuries (1). Currently, the World Health Organization (WHO) designates a total of 42 countries spanning sub-Saharan Africa and the Americas (South and Central America) as regions to be at risk for YF transmission (2).

Yellow fever causes a broad spectrum of clinical severity in humans, ranging from asymptomatic infection and mild flu-like symptoms to severe manifestations, such as fever accompanied by jaundice, hemorrhage, and fatal outcomes (3). The case-fatality rates for severe YF illness vary between 20% and 60% across populations (4). Vaccination is an effective strategy for the prophylaxis of YF (5).

The YF-17D vaccine, developed in the 1930s, stands as one of the most effective and safe live-attenuated vaccines developed till date (6, 7). Recent studies showed that the YF-17D vaccine produces long-lasting immunity by activating all arms of the immune system—innate, cellular, and humoral. The humoral immunity mediated by neutralizing antibody is the correlate of protection induced by YF vaccines (8).

The immunological response to a vaccine should be assessed by measuring the neutralizing antibody titers. As per the WHO recommendation, YF neutralizing antibody titers in vaccine sera are conventionally determined on Vero cells using the YF-17D vaccine strain (9). Yellow fever neutralizing antibody epitopes are mainly conformational determinants of the envelope (E) protein. Quaternary epitope binding antibodies may constitute a substantial fraction of the neutralizing activity in human immune sera (8, 10). Most of the epitopes are conserved across the different viral strains. Nevertheless, some studies demonstrated a lower neutralization of recent South American isolates (11–14) specific neutralizing epitopes linked with lower neutralization sensitivity were mapped on the E protein (11), such as amino acid (AA) E67 and E83 (12) or E71/72 (15) in domain I; E153/155 in domain II (15); or the epitope from E269 to E272 in the domain I/II hinge (12). The composite polyclonal immune response to YF-17D vaccination is therefore capable of protecting against multiple strains of virus that might differ at one but not all epitopes. The multigenic basis of neutralizing domains in YF virus ensures recognition across diverse HLA (human leukocyte antigen) types in the human population. Probably, this antigenic profile is the reason behind the induction of breadth of neutralization after vaccination with a single YF strain.

Although existing YF-17D vaccines have proven highly effective, their egg-based production systems present inherent constraints that limit scalability and flexibility (16). These limitations pose challenges in meeting increasing global demand, particularly in the context of recent outbreaks and the anticipated expansion of YF-endemic regions due to climate and demographic changes (17). In addition, while serious adverse events such as vaccine-associated viscerotropic and neurotropic diseases (YF-VAD and YF-AND) remain rare, they highlight the importance of continued efforts to develop next-generation vaccine candidates with optimized safety and manufacturing profiles (18). Building on this need, the vYF vaccine candidate, cloned from a YF-VAX vaccine (YF-17D vaccine) and adapted for propagation in Vero cells cultured in serum-free media (16), was developed to maintain the well-established immuno-protective features of current vaccines while offering advantages such as enhanced production scalability, reduced dependency on egg-based manufacturing, and potentially improved safety (19).

We previously reported the development and selection process of vYF and showed it was less neurovirulent in mice than currently marketed YF-17D vaccines, Stamaril ® and YF-VAX ®, suggesting,an improved safety profile particularly regarding the risk of vaccine-associated neurotropic disease (YEL-AND). However, vYF exhibited similar viscerotropism, immunogenicity, and efficacy in small animal models (mice and hamsters) (16). The safety and immunogenicity profiles of vYF were also demonstrated in non-human primates (NHPs), described as “gold standard” model surrogates for humans, and results were comparable with that of 2 currently licensed YF-17D vaccines (20). These findings suggest that vYF has the potential to provide protection against YF virus infection in clinical practice, comparable with the outcomes achieved with currently marketed YF-17D vaccines. A first-in-human, phase 1 study demonstrated that vYF has a similar safety and immunogenicity profile to YF-VAX (19).

vYF is currently in clinical development stage (Phase II; ClinicalTrials.gov Identifier: NCT04942210 and NCT05011123).

In the previously published results of this NHP study, immunogenicity of vYF was evaluated through multiple parameters including the induction of high neutralizing antibody titers against the vaccinal virus strains from 14 to 258 days after NHPs were vaccinated with a single subcutaneous dose of vaccine (20). The main objective of this study was to evaluate the breadth of neutralization of the vYF-vaccinated NHP sera in comparison with the sera from NHPs immunized with reference vaccine YF-VAX against selected viral strains, representing the 4 YF virus sequence clusters. In addition, the neutralization results were analyzed regarding the strain genetic variability.

Study design

In an earlier investigation, NHPs underwent immunization with vYF or with YF-VAX vaccine. Pools of sera were prepared from NHP sera samples collected at Day (D) 27/D60/D221 post-vaccination (17). These serum samples were then exposed to the designated virus strains cultivated on Vero cells. In the next step, this serum/virus mixture was incubated in Vero cells followed by immunostaining to generate the neutralization assay results (Figure 1). These procedures are discussed in detail in the following sections.

Sera

Sera obtained from the vYF-vaccinated NHPs (20) were combined, and 2 pools of sera from 7 NHPs each, with “low” (lowest titers from 221 days post-vaccination sampling) and “high” (highest titers from 60 days post-vaccination sampling) neutralizing antibody titers were obtained. Additionally, a pool of sera was constituted with sera collected from 5 YF-VAX–vaccinated NHPs. The detailed compositions of the serum pools are described in Table 1.

Statistical analysis confirmed significant titer differences between the two serum pools using a neutralization assay with the Stamaril virus (p < 0.0001; mixed-effects model with strain and titer level as fixed factors, without adjustment for multiple comparisons). The fold-change between pools ranged from 1.9- to 3.6-fold (0.28 to 0.56 Log).

Viruses

Yellow fever viruses belong to a single serotype. The relatively high genetic stability of wild-type YF viruses compared to many other RNA viruses is critical to the effectiveness of a single virus strain (17D) as a vaccine (11). One hypothesis to explain the high conservation of YF virus over time and geographic location is the high fidelity of its RNA polymerase (21). In this regard, we have selected a limited number of strains based on geographic location, year of isolation, and position in a global phylogenetic tree.

To ascertain the minimal number of viral strains for comprehensive coverage of YF virus genetic diversity, full genome and protein sequences were retrieved from GenBank. Thereafter, a total of 705 full-length polyprotein sequences were aligned using the MAFFT (Multiple Alignment using Fast Fourier Transform) algorithm (22), followed by the construction of a phylogenetic tree using PhyML 3.0 (23). The resulting phylogenetic tree was then visualized using Dendroscope 3.0 (24), and 4 field isolates were selected corresponding to the 4 branches of the phylogenetic tree. Representative strains were selected to cover the major phylogenetic clades of YFV, ensuring that neutralization testing reflected the genetic diversity of circulating lineages. Selection was further limited to well-characterized isolates available in our laboratory to ensure assay feasibility and consistency. The details of the 4 selected strains are provided in Table 2.

Detailed passage histories of the isolates were not available. However, all strains were isolated and subsequently amplified multiple times on Vero cells, including 2 passages in our laboratory prior to neutralization testing. Global sequence alignment showed minimal overall genetic variability among the strains. Each selected strain sequence was part of relatively large groups displaying almost identical polyprotein sequences, suggesting limited post-collection genetic evolution. The sequences of the isolates, following passages on Vero cells, remained representative of their respective groups. Therefore, there was minimal impact of virus passaging in this study.

Selected YF virus field isolates were received from UTMB (University of Texas Medical Branch), CDC (Centers for Disease Control and Prevention), or EVAg (European Virus Archive). Working stocks of these viruses were prepared by 2 passages on Vero cells. The resulting strains were characterized using the CCID 50 (cell culture infectious dose 50%) method for infectious titration on Vero cells and were used for genome sequencing.

In addition to field isolates, 2 control viruses were also included in this study: 1) the live-attenuated YF 17D-204 reference vaccine Stamaril ® (Sanofi); 2) the parent wild-type YF virus Asibi, obtained from an infectious Asibi clone kindly provided by Professor Stephen Higgs (World Reference Center for Emerging Viruses and Arboviruses at UTMB).

Virus production and characterization

Virus production

Vero LS11 cells were used for the production of viruses, titrations, and neutralization assays.

Vero LS11 cell line was derived from American Type Culture Collection (ATCC) CCL-81. Vero cells were seeded in F175 cm 2 flasks with approximately 8×10 6 cells per flask in Iscove’s Modified Dulbecco’s Medium (IMDM) supplemented with 4% fetal bovine serum (FBS) 3 days prior to the infection. Cells were infected at multiplicity of infection 0.001 (i.e., 1 virus/1000 cells) under a final volume of 5 mL/F175. Following 2 h of incubation at 37 °C, the inoculum was removed; subsequently, 25 mL of fresh IMDM culture medium with antibiotics was introduced into each flask. Flasks were then incubated at 37 °C with 5% CO 2. The infectious supernatant was harvested between D3 and D6 post-infection (depending on the cytopathic effect observed), was clarified by centrifugation at 300×g for 10 minutes at 4 °C and was stabilized by addition of 10% FBS. Subsequently, the infectious titer was determined for each virus by CCID50 titration in Vero cells and the viruses were used for genomic sequence analysis.

Infectious titration

The titration was performed using the CCID 50 method on Vero cells. Precisely, the viral suspensions were subjected to 4-fold serial dilution in IMDM + 4% fetal calf serum (FCS) starting from −0.6 Log to −4.2 Log in a 96-deep well plate. A control virus (Stamaril virus amplified once on Vero cells) was included in each test as a positive reference. One day before the titration of viral suspension, Vero cells were seeded in flat bottom 96-well plates (30,000 cells/well). On the day of titration, the viral suspension was diluted, and 100 μL of each virus dilution was transferred into 10 wells containing Vero cells. The samples were incubated for 6 days at 37 °C with 5% CO 2. In the next step, supernatants were discarded, and cells were fixed with 85% acetone before starting immunostaining. Plates were blocked with phosphate-buffered saline (PBS) + 0.05% Tween 20 + 2.5% skim milk and were incubated first with an antiflavivirus monoclonal antibody 4G2, followed by a goat anti-mouse IgG, a streptavidin-alkaline phosphatase conjugate. Finally, the plates were stained with the BCIP/NBT (5-bromo-4-chloro-3-indolyl-1-phosphate/nitro blue tetrazolium) chromogen. Positive wells that showed at least one viral antigen focus stained in black were counted, and the final titer was calculated using the least square regression method.

Sequencing

Viral genomic RNAs were extracted using a QIAamp Viral RNA Mini Kit (Qiagen reference: 52904). The sequencing library was generated from 15 µL portion of extracted RNA on the automated Sciclone ® system (PerkinElmer) following the “Creating cDNA Libraries using Twist Library Preparation Kit for ssRNA Virus Detection - Revision 2.0, 25 August 2020” protocol from Twist Bioscience. This process converted RNA into cDNA libraries, including a fragmentation step of 13 min and a short amplification of 14 cycles by polymerase chain reaction (PCR) with unique dual indices. In the next step, the viral genome of interest was captured in the cDNA libraries using the “Twist Target Enrichment Protocol” (Twist Bioscience) by pooling 25 ng of each sample and an amplification by PCR during 14 PCR cycles. After quality and quantity assessments, final libraries were sequenced on a v3 flow cell on the MiSeq sequencer to generate paired end reads of 2×150 base pairs. The resulting run was demultiplexed with binary base call convert, and the data were uploaded in CLC platform for analysis.

Neutralizing antibody titration

Neutralizing antibodies in the serum were titrated using the SN50 (seroneutralization 50%) method on Vero cells. In summary, heat-inactivated sera obtained from the vaccinated NHPs underwent 4-fold serial dilution in IMDM (Gibco Life Technologies #31980-022) + 4% FCS (Hyclone CAT N° SH30084.04) + 1% penicillin/streptomycin starting from dilution of 1:10. Virus stocks were diluted to obtain a concentration of 2.5 Log CCID50/mL in IMDM and incubated for 1.25 h at 37 °C with diluted serum samples (v/v). The virus dose was also incubated with an equal volume of NHP serum without YF-specific antibody to determine the actual number foci forming units used in the assay.

For each serum dilution, the virus/serum mixture was then added in 6 wells of 96-well plates pre-seeded at 30,000 cells/100µL/well the day before. The plates were incubated for 6 days at 37 °C. After incubation, cells were fixed with 85% acetone before immunostaining. For immunostaining, non-specific sites were blocked with PBS + 0.05% Tween 20 + 2.5% skim milk and incubated first with an antiflavivirus monoclonal antibody 4G2 biotinylated (produced in vitro in house from hybridoma ATCC #D1-4G2-4-15 - HB-112) at 2 µg/mL and second with a streptavidin-alkaline phosphatase conjugate (Sigma, ref S2890, 1 mg). Finally, plates were stained with the BCIP/NBT chromogen (Sigma, cat# B5655). Uninfected wells, corresponding to wells without any immunostained foci, were counted manually. The final titer was calculated using the least square method and corresponded to the reciprocal of the dilution that neutralized 50% of the viral foci produced by the virus dose. The limit of detection (LOD) of the SN50 assay was about 20 SN50 when the first dilution of serum evaluated was 1:20. For the calculation of the mean value per group, an arbitrary value of half of the LOD was assigned to all samples below the LOD, i.e., 10 SN50. Each value represented the mean of 9 repeats. The neutralization assay was repeated 3 times by at least 2 different operators and each combination serum pool/virus was tested in triplicate in each assay run.

Statistical analysis

All statistical tests were 2-sided, and the nominal level of statistical significance was set to α = 5%. The analyses were performed on SEG SAS v9.4 ®. SN50 titers were analyzed in Log using the 2-way analysis of variance (ANOVA-2) with serum pool and viral strain as fixed factors and their interaction. P values <0.05 were deemed statistically significant. For the comparisons Tukey’s adjustment was performed. The model’s residuals were verified to evaluate the model’s validity (normality, extreme individuals, etc.).

Virus selection

To ascertain the minimal number of viral strains for comprehensive coverage of YF virus genetic diversity, 705 full genome and protein sequences were retrieved from GenBank, spanning 34 countries and ranging from year 1927 to year 2019. The resulting phylogenetic tree is shown in Figure 2. In the phylogenetic tree, 4 major clusters are clearly distinguished, illustrated by colors and named A, B, C and D. Detailed strain distribution is shown in Supplementary Table 1. Four field isolates available in the lab were chosen from each of the 4 branches of the phylogenetic tree.

Neutralization assay validity – virus titration

The infectivity titer of the virus doses used in neutralization assay was determined by titration in Vero cells concurrently during the performance of each neutralization assay. The targeted viral titer in neutralization assays was 2.5 Log CCID 50 /mL. The corresponding titers for each replicate are listed in Table 3.

The viral titers were similar to the expected titers, considering the variability of the CCID 50 titration method (95% CI of ± 0.3 Log). However, isolate #4 in the neutralization assay 3 replicate had an antibody titer of 3.22 Log CCID 50 /mL, which was 0.72 Log higher than expected. In the context of neutralization assay, high viral titers can lead to the observation of lower neutralization titers for this specific viral strain. Despite this risk, it was decided to proceed with the analysis and finalize the results without conducting a retest.

Neutralization assay results

Neutralizing antibody titers for the vYF low and high pools of sera are presented in Figure 3A and Figure 3B, respectively, and for the YF-VAX pool in Figure 3C. Along with pool of sera from YF-VAX–vaccinated NHPs, vYF high and low pools of sera neutralized all the wild-type isolates with titers at least 3-fold above the correlate of protection threshold, i.e., detectable neutralizing antibodies at 10 PRNT 50 and by homology 10 SN50.

In each neutralization assay, the previously mentioned 2 control viruses were included: the live-attenuated YF 17D-204 reference vaccine Stamaril ® (Sanofi) and the parent wild-type YF virus Asibi. For the YF-VAX pool, the mean neutralizing antibody titers for YF 17D-204 reference vaccine Stamaril ® (Sanofi) and the parent wild-type YF virus Asibi were 2.66 Log SN50 and 2.20 Log SN50, respectively. For the vYF low pool, the corresponding titers were 1.91 Log SN50 and 2.06 Log SN50, respectively, and for the high pool about 2.33 Log SN50 and 2.00 Log SN50, respectively.

The titers detected for the Asibi parent strain were considered as reference, and all the neutralizing antibody titers measured for the virus field isolates were statistically compared to these titers.

SN50 titers were analyzed in Log using ANOVA-2 with serum pool and viral strain as fixed factors and their interaction. The comparisons were adjusted using Tukey’s method, as described in Supplementary Table 2.

For the YF-VAX pool of sera, the neutralizing antibody titers calculated for isolates #1 and #3 were not significantly different from the reference titer (all p -values ≥0.931). In the case of the Stamaril strain, the titer value was 0.46 Log SN50 higher than the reference titer (p -value <0.001). For isolates #2 and #4, the measured neutralizing antibody titers were 0.40 Log SN50 and 0.34 Log SN50 lower respectively than the reference titer measured for the Asibi strain (all p -values <0.001).

As for the vYF high pool of sera, the neutralizing antibody titers measured for isolates #1 and #3 were not significantly different from the reference titer (all p -values ≥0.999). The neutralizing antibody titer measured for the Stamaril strain was 0.33 Log SN50 higher than the reference titer (p -value <0.001). For isolates #2 and #4, the titers were 0.49 Log SN50 and 0.28 Log SN50 lower respectively than the reference titer (all p -values ≤0.004).

Finally, for the vYF low pool of sera, the neutralizing antibody titers measured for isolates #1, #3, and #4 and for Stamaril were not significantly different from the reference titer (all p -values ≥0.372).

The neutralizing antibody titer measured for isolate #2 was 0.32 Log SN50 lower than the reference titer measured for the Asibi strain (p -value =0.004).

Sequence analysis

The correlation between reduced neutralization titers identified in certain strains and their corresponding virus sequences was analyzed, with a focus on 3 epitopes (8, 10) that had been previously associated either with neutralization-resistant mutants (12, 15) or the highly potent neutralizing epitope (25, 26). The neutralization-resistant mutants were described in 2 previous studies. Haslwanter et al. (12) reported a decreased activity of YF-17D-vaccine induced antibody against South American strains linked to 2 epitopes located in domain II (AA E67 and E83) of the E protein and in the domain I/domain II hinge (AA 269-272). The second epitope (AA E269-272) overlaps with an N-glycosylation site at position E269. Daffis et al. (15) identified the AA 71 in domain II as the most critical residue within an epitope also involving AA 153-155, crucial for neutralization by a human monoclonal antibody obtained from naturally infected individuals.

The neutralizing-protective double-lock supersite corresponding to the epitopes E65-74:81:84:87:99-104:149:240-244:267 was recognized by monoclonal antibodies 5A, YD6, and YD73 and described by Lu et al. and Li et al. (25, 26). This epitope is named 5A in the subsequent descriptions. Table 4 displays the status of these epitopes on the 4 selected strains in comparison with the major consensus sequence found in databases and with the sequences previously described as linked with neutralization sensitive and resistant strains. No obvious link was identified between the neutralization titers and the epitopes described as involved in neutralization resistance (12, 15) or the 5A epitope (25, 26). Indeed, isolate 1 and 3 display mutations in three and one out of the four known neutralizing epitopes, respectively, without impact on seroneutralization. Conversely, isolate 2 shows lower seroneutralization that can be attributed to mutations in these epitopes, whereas isolate 4 that displays no mutation in the documented neutralizing epitopes shows lower seronjeutralization.

The aim of this study was to evaluate the breadth of serum neutralizing antibody induced by the new live-attenuated YF vaccine candidate (vYF) in comparison with sera from YF-VAX–vaccinated NHPs. Assessment of neutralizing antibodies in vaccinated human sera is the most widely utilized biomarker for YF vaccine immunogenicity, since it is a key indicator of vaccine protection (12), and NHPs offer the best proxy response to YF vaccines as compared the immune response of humans (27). The breadth of protection induced by vYF vaccine in NHPs was evaluated by neutralization assay against 4 strains. Pool of sera sampled from the vYF-vaccinated NHPs showed the potential to neutralize all the wild-type isolates with titers at least 3-fold above the correlate of protection threshold, suggesting evidence of the induction of the breadth of protection. Two strains (isolates #1 and #3) resulted titers similar to the Asibi parental strain, and 2 strains (isolates #2 and #4) resulted significantly lower titers (from −0.3 Log to −0.5 Log). However, the differences are closed to the variability of the assay (± 0.3 Log) and all the titers remained at least 3-fold above the threshold of neutralization.

The primary limitation of this study was the restricted availability of non-human primate (NHP) serum samples, which necessitated pooled sample analysis and limited strain selection. Strains were selected based on sequence analysis, with one representative strain per sequence cluster chosen from available isolated live virus stocks. Serum pools were stratified according to neutralization titers against the YF-17D strain to create high-titer and low-titer groups. Given the homogeneous nature of neutralizing antibody titers in NHPs, achieving pools with distinct neutralization levels required sampling across different time points: 7 samples with the highest titers were selected from day 60, while 7 samples with the lowest titers were obtained from day 221. This temporal sampling strategy may introduce bias, as both the magnitude and breadth of immune responses can vary over time. Due to these potential confounding factors, strain-specific responses were analyzed separately for each serum pool and cannot be directly compared between pools.

The present study confirms previously published findings demonstrating reduced neutralization of the Brazilian BeH622205 strain. Conversely, our results revealed diminished neutralization of the Cachatra Guinea-Bissau strain when tested with both high-titer sera from vYF-vaccinated NHPs and pooled sera from YF-VAX-vaccinated NHPs. However, in pooled sera with low titers after vaccination with vYF this difference was not significant and previous studies did not see significant differences between Asibi and the Cachatra strain (13). In this study, analyses were performed in triplicate across three independent assays conducted on separate days by different operators using different cell and reagent batches. The lower neutralization of Isolate #4 (Cachatra) was consistently observed across all 9 replicates when tested with both high-titer sera from vYF-vaccinated NHPs and pooled sera from YF-VAX-vaccinated NHPs. We identified no evidence to suggest these results are artifactual.

The clinical impact of these observations in vaccinees requires further investigation, as some studies have identified cases of vaccine escape in Brazil (28, 29) that may be linked to the circulating strain. Additional in vitro studies are also necessary to better understand the infection mechanisms of these particular strains compared to the highly adapted Asibi virus. However, the current study design does not permit us to address these crucial questions. Despite these limitations, our results confirmed that the immune response to the vYF vaccine candidate is similar to that of the reference vaccine YF-VAX.

In this study, the observed neutralizing antibody titers to Stamaril were lower in comparison with the titers reported in a prior study by Piras-Douce et al. (20). This variation is mainly attributed to methodological variations in the assay type, quantity of virus used, virus counting method, and statistical method for titer calculation. In this study, we employed the SN50 assay based on endpoint dilution to determine neutralizing antibody titers, while Piras-Douce et al. (20) used the µPRNT 50 assay method, based on plaque counting. Plaque counting is considered as a standard assay procedure. Despite a sensitivity loss with SN50, it did not impact the results as titers were at least 3-fold above the LOD. We counted the positive wells, unlike the previous study (20) which counted individual viral plaques. Furthermore, we calculated the final titers employing the least square regression method, whereas in the earlier study (16), a direct viral plaque counting method was utilized for titer determination. We used low passage clinical isolates in this study to limit the risk of virus adaptation to the culture in Vero cells; as a result, some isolates did not form plaques. Consequently, it necessitated the exploration and implementation of an alternative methodology to ensure the generation of dependable neutralizing antibody titers.

It is generally accepted that strains of YF virus constitute a single serotype (30). Phylogenetic studies have previously defined 7 genotypes, including 5 African and 2 South American (31); in these studies, genotypes were defined based on the sequence of precursor membrane and E genes only. In the last decade, the development of high throughput sequencing (32) allowed drastically the number of full genome sequence analyses to increase. To select our panel of strains, we ran a new phylogenetic analysis based on the 705 sequences available in GenBank in September 2019. Based on the full polyprotein sequence analysis, we observed 4 sequence clusters that seemed to be related to the geographic origin and selected one strain from each cluster.

In this study, limited to the analysis of 4 viral isolates, we were not able to evidence a consistent relationship between neutralization titers and the geographical origin, year of isolation, or genetic clustering of the yellow fever virus strains tested. Further sequence analysis of E-protein residues previously associated with known neutralizing epitopes also failed to reveal any consistent link to neutralizing antibody titers, contrary to earlier reports (12, 15). Notably, strains sharing identical residue patterns at these key positions exhibited divergent neutralization profiles, suggesting that these markers alone cannot account for the variability observed across strains. Likewise, comparison of established neutralizing epitope motifs between the vaccine candidate and selected representative strains showed no association between sequence differences and seroneutralization levels. Together, these results indicate that the vaccine-induced neutralizing response is not confined to a limited set of known epitopes and that vaccine efficacy does not rely solely on these specific motifs. Instead, the broad seroneutralization observed in our study likely reflects the contribution of additional, yet undefined epitopes targeted by the polyclonal immune response. Further analysis would be useful with various strains presenting individual mutations to complete the analysis and conclude on individual mutation impact on neutralization titers.

Differences in neutralization can also be driven by other parameters than directly linked to the virus sequence. Among these phenomena, virus fitness, ratio of non-infectious viruses, neutralization cell type, presence of different viral quasi-species can be underlined. Recently, Bibby et al. (14) also demonstrated that a single mutation, as the attenuation-linked mutation E380 impacts the viral structure and a correlation of the virus structure with the neutralization sensitivity (14). They described that bumpy viruses are more sensitive to neutralization by mAbs that target cryptic epitopes, much more accessible on the surface of bumpy viruses in comparison to mature, smooth and compact virions. To be able to document the impact of structure variations on the polyclonal immune response to Yellow Fever, Electron Microscopy Polyclonal Epitope Mapping (EMPEM) study would be instrumental (33).

In summary, we demonstrated the neutralization antibody profiles elicited by the vYF vaccine in NHPs were similar to the antibody response elicited in NHPs by the reference vaccine, YF-VAX. Although slight differences were observed, it is noteworthy that the neutralizing antibody titers consistently remained above the established correlate of protection threshold, regardless of the diverse strains of YF virus tested against antibody elicited in NHPs by the vYF virus and existing approved YF vaccine viruses. These findings have significant implications for global YF prevention strategies, as they highlight the potential of vYF to provide similar broad protection against diverse strains of YF virus than the current vaccines while offering manufacturing advantages that have the potential to address global YF vaccine shortages. The Vero cell-based production platform for vYF could support more rapid scale-up during outbreaks compared to traditional egg-based manufacturing, potentially improving vaccine availability in endemic regions. Therefore, these findings further demonstrated that the novel live-attenuated vYF vaccine is a promising candidate for the prevention of human disease caused by YF virus, with potential advantages in manufacturing scalability.