Preclinical immunogenicity of the LP.8.1-adapted BNT162b2 COVID-19 vaccine

Abstract

SARS-CoV-2 evolution toward antigenically distinct lineages drives escape from host immunity. JN.1 lineage derivatives have recently dominated the global epidemiologic landscape. In preclinical models, an LP.8.1-adapted BNT162b2 vaccine elicited higher serum neutralizing antibody responses against contemporary, circulating JN.1 sublineages, including the epidemiologically dominant XFG lineage, as compared to JN.1 and KP.2 vaccines. These findings supported the selection of an LP.8.1-adapted vaccine for the composition of the 2025-26 COVID-19 vaccine formula.

COVID-19 continues to cause substantial disease burden, with illness, hospitalization and mortality rates rivaling or exceeding those of influenza¹. The evolution of SARS-CoV-2, the cause of COVID-19, toward escape from host immunity necessitates continuous evaluation of vaccine antigens to ensure that they closely match circulating virus lineages. Since 2022, licensed COVID-19 vaccines have been periodically adapted to elicit protective immunity against contemporary SARS-CoV-2 lineages and have demonstrated consistent clinical and public health benefit across diverse populations in real-world settings^2–7. Updates are intended to keep pace with major antigenic changes in the SARS-CoV-2 Spike (S) protein^8–10. A major antigenic shift occurred in late 2023, with the emergence of the BA.2.86 lineage and its JN.1 sublineage¹¹. This change prompted regulatory authorities to recommend COVID-19 vaccine updates to a JN.1 composition in some parts of the world and to the JN.1 sublineage, KP.2, in other regions. JN.1 sublineages have caused most cases of COVID-19 over the last two seasons and remain globally dominant¹². In that time, the virus underwent extensive genetic and antigenic drift. Acquired mutations in the S protein conferred improved viral fitness, as reflected by increased transmissibility and escape from neutralization by COVID-19 vaccine anti-sera^13–15.

The accelerated evolution of SARS-CoV-2 has primarily manifested in the insertion, deletion, and substitution of amino acid residues in the receptor binding domain (RBD) and N-terminal domain (NTD) of the S protein. An interaction of the number and locations of residue changes in these domains generally has corresponded to the magnitude of immune escape. Two of the globally dominant JN.1 sublineages in early 2025 (Supplementary Fig. 1A)—XEC and LP.8.1—differ from one another by eight residues in the S protein and by five and nine residues compared to JN.1, respectively, when both substitutions and deletions are considered (Supplementary Fig. 1B) This exceeds the three-residue difference between JN.1 and KP.2¹⁶, the two recommended lineage compositions for 2024-2025 vaccine formula, but is far fewer than the 37 S residue changes between JN.1 and the prior season’s recommended vaccine lineage, XBB.1.5¹⁷. Surveillance of the S mutational landscape, together with preclinical immunogenicity and cross‑neutralization data, is an integral component of the total evidence used to support annual decisions for COVID-19 vaccine formula composition.

The epidemiologic trends in the first half of 2025 prompted us to evaluate the preclinical immunogenicity of both XEC- and LP.8.1-adapted BNT162b2 vaccines, as compared to the JN.1- and KP.2-adapted BNT162b2 vaccines. Over the course of the year, genetically drifted JN.1 sublineages emerged. Against this evolving landscape, we evaluated the immunologic breadth elicited by lineage-adapted vaccines against a panel of epidemiologically relevant SARS-CoV-2 lineages.

Neutralizing antibody responses elicited by LP.8.1- and XEC-adapted BNT162b2 vaccines were evaluated in female BALB/c mice in either a BNT162b2 vaccine-experienced or vaccine-naïve background. Vaccine-experienced mice received two doses of the original wild-type (WT) (Wuhan strain) BNT162b2 vaccine as a primary series and one dose of the bivalent BNT162b2 (WT + Omicron BA.4/5) vaccine one month later. Mice were then vaccinated, one month after the third dose, with either JN.1-, KP.2-, XEC- or LP.8.1-adapted BNT162b2 vaccines. Sera were collected prior to and after the final vaccination for assessment of neutralization of contemporary JN.1 sublineage pseudoviruses (JN.1, KP.2, XEC, LP.8.1, LP.8.1.1, LF.7, NB.1.8.1, XFG) and the BA.3.2 saltation lineage (Fig. 1a).

Fig. 1. Pseudovirus neutralization titers elicited by JN.1-, KP.2-, XEC- and LP.8.1- adapted BNT162b2 vaccines administered to vaccine-experienced mice.

a Schematic for vaccination schedule including a fourth dose of the JN.1, KP.2, XEC, and LP.8.1 BNT162b2 vaccine in vaccine-experienced mice. Female BALB/c mice (10/group) that were previously vaccinated with two doses of monovalent original WT BNT162b2, and one subsequent dose of bivalent WT + Omicron BA.4/5 received a single intramuscular dose of one of these vaccine regimens: JN.1, KP.2, XEC or LP.8.1. All vaccine formulations contained a total dose of 0.5 µg. Sera were collected before the fourth dose and one month post-fourth dose. Spleens were harvested at one-month post-fourth dose from five mice. b Fifty-percent geometric mean serum neutralizing titers were assessed in a pseudovirus neutralization assay one month post-fourth dose against the WT reference strain Wuhan, the Omicron lineages and sublineages BA.4/5, JN.1, KP.2, XEC, LP.8.1, LP.8.1.1, LF.7, NB.1.8.1, XFG, and BA.3.2. 50% pseudovirus neutralization titers are shown as geometric mean titers (GMT) of 10 mice per vaccine group. Error bars represent the 95% confidence interval (CI). Each data point represents one animal. The limit of detection (LOD) is the lowest serum dilution, 1:20. The geometric mean ratio (GMR) is the GMT of individual pseudovirus responses of each vaccine group (JN.1, KP.2, XEC or LP.8.1) divided by the GMT of analogous pseudovirus responses of c JN.1 and d KP.2 vaccine groups. Error bars represent the 95% CI. One month following the fourth dose, splenocytes were harvested from mice (n = 5/group), and S-specific CD4⁺ and CD8⁺ T cells were characterized by a flow cytometry-based intracellular cytokine staining (ICS) assay. All samples were stimulated separately with S peptide pools from the WT reference strain Wuhan, BA.4/5, JN.1, KP.2, XEC or LP.8.1. Graphs show the mean ± SD frequency of e CD4⁺ CD154⁺ T cells and f CD8⁺ T cells expressing IFN-γ. Asterisks indicate statistical significance of pseudovirus GMR relative to the corresponding pseudovirus in the JN.1 vaccine group as determined by ANOVA using a Dunnett’s multiple comparisons test. *p < 0.05. a was created in BioRender.

In mice that received a fourth dose of the JN.1 vaccine, geometric mean neutralizing titers (GMTs) against XEC (GMT 2431) and LP.8.1 (GMT 1879) were respectively 2.8- and 3.7-fold lower than those elicited against the parental JN.1 lineage (GMT 6865) (Fig. 1b). KP.2 vaccine-elicited GMTs against XEC (GMT 1684) and LP.8.1 (GMT 3524) were 3.6- and 1.7-fold lower, respectively, than those elicited against JN.1 (GMT 6065) (Fig. 1b). The LP.8.1-adapted vaccine, by contrast, elicited the overall highest GMTs against most emergent lineages. Specifically, the LP.8.1 vaccine elicited significantly higher neutralizing responses against LP.8.1 (5.2-fold increase; p = 0.02), its derivative LP.8.1.1 (5.8-fold increase; p = 0.04), and XFG (3.4-fold increase; p = 0.04), compared to the JN.1 vaccine (Fig. 1c). The XEC vaccine elicited neutralizing responses that were lower than those elicited by the LP.8.1 vaccine but generally higher (>1) than those elicited by the JN.1 vaccine, though not to a threshold of statistical significance (Fig. 1c). As compared to the KP.2 vaccine, both XEC- and LP.8.1 vaccine-elicited neutralizing responses were generally higher but not statistically different (Fig. 1d).

T cell-mediated immunity elicited by the lineage-adapted vaccines was also assessed using peptide pools covering the full-length S protein from Wuhan, BA.4/5, JN.1, KP.2, XEC and LP.8.1 lineages. The LP.8.1 vaccine elicited frequencies of activated (CD40L⁺) IFNγ⁺ CD4⁺ (Fig. 1e) and IFNγ⁺ CD8⁺ T cells (Fig. 1f) that were comparable to the XEC, KP.2, and JN.1 vaccines. Evaluation of IL-2-, TNF-α-, and IL-4-expressing S-specific T cells revealed Th1-biased, cross-reactive responses to all six lineages across all vaccines tested (Supplementary Fig. 2a–e). The conserved, broadly cross-reactive T cell responses in this study trended similarly to those induced by the original and earlier lineage-adapted BNT162b2 formulas in previous reports^18,19.

Prior to the fourth vaccination, all lineage-adapted vaccines elicited similar neutralizing responses against vaccine-matched lineages (Fig. 2a). The highest neutralizing responses across the four vaccine groups were against the Wuhan and BA.4/5 lineages, reflecting the influence of immune imprinting, as observed in prior COVID-19 vaccine-experienced animal studies^18,19 and a well-recognized phenomenon in humans^20–23. After the fourth dose, the LP.8.1-adapted vaccine elicited a 139-fold rise in GMTs (geometric mean fold rise (GMFR)) against the matched LP.8.1 lineage (Fig. 2b). This increase was more than triple than that observed for JN.1- (27-fold), KP.2- (27-fold), and XEC-adapted (45-fold) vaccines. While the induction of a novel, lineage-specific response was evident in these vaccine-experienced mice, it is likely that both novel and recall memory responses to conserved S epitopes help shape protective immunity against COVID-19 in human populations.

Fig. 2. Pseudovirus neutralization titers pre-fourth dose and rise at post-fourth dose with a BNT162b2-lineage adapted vaccine in vaccine-experienced mice.

Female BALB/c mice (5/group) that were previously vaccinated with two doses of monovalent original WT BNT162b2, and one subsequent dose of bivalent WT + Omicron BA.4/5 received a fourth intramuscular dose of one of these vaccine regimens: JN.1, KP.2, XEC, and LP.8.1. All vaccine formulations contained a total dose of 0.5 µg. a Serum neutralizing antibody responses were assessed in a pseudovirus neutralization assay for the pre-fourth dose against the WT reference strain Wuhan, and the BA.4/5, JN.1, KP.2, XEC and LP.8.1 lineages and sublineages. 50% pseudovirus neutralization titers are shown as geometric mean titers (GMT). Error bars represent the 95% CI. Each data point represents one animal. b The geometric mean titer fold-rise (GMFR) from pre-fourth dose (a) to one-month post-fourth dose is shown with 95% CI error bars. The limit of detection (LOD) is the lowest serum dilution, 1:20.

In contrast to a vaccine-experienced model, naïve mice typically generate more robust anamnestic responses to novel antigens, resulting in higher post-vaccination neutralizing responses. In this model, BALB/c mice received two doses of JN.1-, KP.2-, XEC-, or LP.8.1-adapted BNT162b2 vaccines at a three-week interval (Fig. 3a). As observed in prior studies^18,19, GMTs in immunologically naïve mice (Fig. 3b) were approximately an order of magnitude greater than those observed in vaccine-experienced mice (Fig. 1b). Neutralizing responses elicited by the LP.8.1-vaccine were 15.6-fold (p < 0.0001) higher against the homologous LP.8.1, and 5.4-fold (p = 0.003), 4.1-fold (p = 0.008), 3.7-fold (p = 0.018), and 3.5-fold (p = 0.025) higher against the JN.1 sublineages LP.8.1.1, LF.7, NB.1.8.1, and XFG, respectively, as compared to the JN.1 (Fig. 3c). As compared to the KP.2 vaccine, LP.8.1 vaccine-elicited responses were significantly higher against LP.8.1 (7.7-fold increase; p < 0.0001) (Fig. 3d). The XEC vaccine, in contrast, elicited a smaller magnitude of relative increase in neutralizing responses against most circulating JN.1 sublineages. Specifically, the XEC vaccine elicited a 3.3-fold higher (p = 0.043) response against the LP.8.1 pseudovirus, as compared to the JN.1 vaccine (Fig. 3c). Relative to the KP.2 vaccine, XEC vaccine-elicited neutralizing titers that were 0.9-to-2.2-fold higher; however, these differences were not statistically significant.

Fig. 3. Pseudovirus neutralization titers elicited by JN.1-, KP.2-, XEC- and LP.8.1- adapted BNT162b2 vaccines administered as a primary series in naïve mice.

a Female BALB/c mice (10/group) were vaccinated with two doses of either JN.1-, KP.2, XEC- or LP.8.1-adapted vaccine, at a 21-day interval. All vaccine formulations contained a total dose of 0.5 µg. Sera and spleens were collected one month after the second dose. b Fifty-percent geometric mean serum neutralizing titers were assessed in a pseudovirus neutralization assay one month after the second dose against JN.1, KP.2, XEC, LP.8.1, LP.8.1.1, LF.7, NB.1.8.1, XFG, and BA.3.2 lineages. 50% pseudovirus neutralization titers are shown as geometric mean titers (GMT) of 10 mice per vaccine group. Error bars represent the 95% CI. Each data point represents one animal. The limit of detection (LOD) is the lowest serum dilution, 1:20. The geometric mean ratio (GMR) is the GMT of individual pseudovirus responses of each vaccine group (JN.1, KP.2, XEC or LP.8.1) divided by the GMT of analogous pseudovirus responses of the (c) JN.1 and d KP.2 vaccine group. Error bars represent the 95% CI. e The antigenic map visualizes cross-reactivity across a panel of SARS-CoV-2 lineages using sera from vaccine-naïve mice. SARS-CoV-2 lineages are shown as circles and sera are indicated as squares. Each square corresponds to sera of one individual BNT162b2-immunized mouse: BNT162b2 JN.1 (dark purple), KP.2 (yellow), XEC (green), or LP.8.1 (dark pink) vaccine. Antigenic distance is represented in both horizontal and vertical axes. Each square in the matrix represents 1 antigenic unit, which reflects a two-fold difference in neutralization titer. The points that are closer together reflect higher cross-neutralization and are, therefore, antigenically more similar. One month following the second dose (completion of primary series), splenocytes were harvested from mice (n = 5/group), and S-specific CD4⁺ and CD8⁺ T cells were characterized by a flow cytometry-based ICS assay. All samples were stimulated separately with S peptide pools from the WT reference strain Wuhan, BA.4/5, JN.1, KP.2, XEC, or LP.8.1. Bar graphs show the mean ± SD frequency of f CD4⁺ CD154 + T cells and g CD8⁺ T cells expressing IFN-γ. Asterisks indicate statistical significance of pseudovirus GMR relative to the corresponding pseudovirus in the JN.1 vaccine group as determined by ANOVA using a Dunnett’s multiple comparisons test. ****p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05. a was created in BioRender.

As a primary series, all lineage-adapted vaccines, including the LP.8.1 vaccine, elicited comparable frequencies of activated (CD40L⁺) IFNγ-expressing CD4⁺ (Fig. 3f) and CD8⁺ T cells (Fig. 3g) against Wuhan, BA.4/5, JN.1, KP.2, XEC, and LP.8.1 S peptide pools, Frequencies of CD40L⁺ IFNγ⁺ CD4⁺ cells in naïve mice were generally lower than those observed in vaccine-experienced mice (Fig. 1e), where there was likely greater effector and central memory T cell recall to conserved S epitopes. The LP.8.1, XEC, JN.1, and KP.2 vaccines also induced similar frequencies of IL-2-, TNF-α-, and IL-4-expressing T cell responses (Supplementary Fig. 3a–e).

Although our assessment of breadth focused on dominant and subdominant JN.1 sublineages, the emergence of the saltation variant BA.3.2 in 2024 raised concerns, as it harbors 78 amino acid differences in the S protein compared to LP.8.1. Published reports have shown that these residue changes confer neutralization escape from COVID-19 mRNA vaccine anti-sera^13,24. In our vaccine-experienced mouse model, we observed vaccine-elicited neutralizing responses against BA.3.2 that were 34-, 23-, 25-, and 46-fold lower than against vaccine-matched lineages in the respective JN.1, KP.2, XEC and LP.8.1 vaccine groups (Fig. 1b). In naïve mice, neutralizing titers against BA.3.2 were at or near the assay limit of detection, indicating immune escape from all four lineage-adapted vaccines in the absence of prior SARS-CoV-2 immunity (Fig. 3b). These findings, supported by neutralization data for BA.3.2 from LP.8.1 BNT162b2-vaccinated individuals^25,26, suggest that BA.3.2 exhibits substantial antigenic divergence from JN.1 sublineages and reduced susceptibility to neutralization by LP.8.1 vaccine immune sera. However, additional epidemiologic and clinical data are needed to assess any implications at the population level.

We employed antigenic cartography to map neutralizing titers, represented as antigenic distance, elicited by lineage-adapted vaccines in naïve mice. In this two-dimensional representation, JN.1-sublineages were located within two antigenic units of each other, corresponding to a 4-fold difference in GMT (Fig. 3e). This indicates an incremental, but progressive, antigenic drift of more contemporary JN.1 sublineages from their parental lineage. BA.3.2, however, is approximately 8 antigenic units away from the LP.8.1 lineage, likely representing a major antigenic shift, as it has a similar distance from earlier Wuhan, BA.1, BA.4/5, XBB.1.5, and JN.1 lineages. As of March 2026, there have been more than 2,200 cases of sequence-confirmed BA.3.2 (Supplementary Fig. 4) in GISAID²⁷. Although BA.3.2 has a reported lower infectivity and slower epidemiologic growth rate than other circulating lineages, its recently rising incidence in multiple global regions justifies its close monitoring¹³.

The progressive antigenic drift of SARS-CoV-2 presents an ongoing challenge to maintaining vaccine-induced protective immunity. Regular updates to the lineage composition of vaccines have helped mitigate this challenge and are consistently supported by comprehensive preclinical data packages. Vaccine-experiencing regimens remain highly relevant, and the best available in vivo model, for approximating BNT162b2 responses in the general population of varied SARS-CoV-2 immune backgrounds. Although our experienced mouse model mimics prior vaccine exposure history, it cannot replicate the complex exposure history of most humans, which includes heterogeneous aging populations and hybrid immunity acquired through repeated vaccinations and infections²⁸. For this reason, preclinical responses are inferential and not conclusive of all possible immune responses in the general human population.

Recent JN.1 sublineages have drifted far enough antigenically to evade neutralization elicited by JN.1- or KP.2-adapted vaccines^13,15. The LP.8.1-adapted BNT162b2 vaccine, in mice with varying immune backgrounds, elicited higher neutralization titers against the homologous LP.8.1 sublineage, and genetically related sublineages XEC, NB.1.8.1, and XFG, as compared to JN.1- and KP.2- vaccines. These findings are supported by real-world outcomes, as LP.8.1-adapted BNT162b2 vaccine effectiveness was demonstrated when XFG dominated during the early 2025-26 respiratory virus season²⁹. Together, the evidence supports that closer matching of vaccine antigens to circulating lineages closely associates with improved preclinical^18,19 and clinical immunogenicity of BNT162b2 vaccines^30–35, and extends these findings to settings where circulating lineages have genetically drifted, but not shifted, from the lineage composition of the vaccine of interest. We have shown previously how small differences in the amino acid sequences in the S proteins of two closely related lineages, such as KP.2 and JN.1, can result in structural changes that potentially underpin differences in immunogenicity¹⁸. Recently published reports for more contemporary lineages discussed herein have recapitulated this association between structure and functional immune responses, wherein a few amino acid residue changes can influence the conformation of the RBD and consequent exposure of neutralizing epitopes, resulting in antibody evasion^36,37. The evidence to date, from our prior published work and concordant findings from other groups, provides a potential mechanistic basis for differences observed between the lineage-adapted vaccines evaluated.

Overall, these preclinical data demonstrate that the LP.8.1-adapted BNT162b2 vaccine maintains potent cross-neutralization of currently and recently predominant SARS-CoV-2 lineages, particularly as compared to JN.1- and XEC-adapted vaccines, which have been recommended and approved as an alternative vaccine composition in some regions³⁸. The ability of the LP.8.1 BNT162b2 vaccine to induce significant improvements in neutralization responses against related lineages, as compared to the prior season’s JN.1 and KP.2 vaccines, helps inform future strategies for vaccine composition updates, particularly during periods of extended viral genetic and antigenic drift.

Methods

Animal ethics

All mouse experiments were performed at Pfizer, Inc. (Pearl River, NY, USA), which is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC). All procedures performed on animals, as outlined in the animal use protocol (AUP-23-37), were in accordance with regulations and established guidelines reviewed and approved by the Pfizer Pearl River Institutional Animal Care and Use Committee (IACUC). Animals were observed after all procedures and injection sites were monitored following each vaccination. Per IACUC guidelines, prior to euthanasia, animals were induced to a surgical plane of anesthesia via inhalation of ~2–4% isoflurane. Exsanguination was then performed via cardiac puncture, with cervical dislocation as a confirmatory method.

BNT162b2 mRNA XEC and LP.8.1 vaccine modification and formulation

The BNT162b2-adapted vaccines encode the S(P2) of JN.1 (GISAID EPI_ISL_18374006), KP.2 (GISAID EPI_ISL_18912216), XEC (GISAID EPI_ISL_19283891), and LP.8.1 (GISAID EPI_ISL_19467828) on the BNT162b2 RNA backbone. Purified nucleoside-modified RNA was formulated into lipid nanoparticles as previously described by Maier et al.³⁹.

Immunogenicity in BNT162b2-experienced mice

Female BALB/c mice (10 per group, age 6–8 weeks; Jackson Laboratory) were vaccinated and bled as shown in Fig. 1a. In brief, mice were vaccinated intramuscularly with a two-dose series (days 0 and 21) of a 0.5 µg dose level of original BNT162b2 WT vaccine, followed by a third dose booster (day 49) of bivalent WT + Omicron BA.4/5 vaccine, and a fourth dose booster (day 162) of either JN.1-, KP.2- XEC- or LP.8.1-adapted vaccines. Formulations contained a total dose of 0.5 µg. A control group of ten mice received saline injections in place of active vaccines. A total volume of 50 µL of vaccine or saline was administered intramuscularly to the upper outer hind leg for each animal. Sera were collected for evaluation of pseudovirus neutralizing antibody responses prior to the fourth dose (day 162) and at one-month post-vaccination timepoint (day 189). Spleens were collected on day 189 to evaluate cell-mediated immune responses.

Immunogenicity in naïve mice

Female BALB/c mice (10 per group, age 6–8 weeks; Jackson Laboratory) were vaccinated and bled as shown in Fig. 3a. Mice were vaccinated intramuscularly on days 0 and 21 with either JN.1-, KP.2-, XEC- or LP.8.1 -adapted vaccines. The control group received saline injections in place of the active vaccine groups. Sera and spleens were collected 28 days after the second dose (day 49) for evaluation of pseudovirus neutralizing antibody responses and cell-mediated immune responses, respectively.

Pseudovirus generation

Pseudovirus stocks were generated in HEK-293T cells (ATCC, CRL-3216) using SARS-CoV-2 S plasmid DNA and vesicular stomatitis virus (VSV; VSVΔG(G)-GFP virus: Kerafast, EH1019-PM), as described previously¹⁸. VSV-based pseudoviruses used in the assay expressed the S protein from the following SARS-CoV-2 variants: WT (Wuhan-Hu-1, ancestral strain), BA.4/5, JN.1, KP.2, XEC, LP.8.1, LF.7, LF.7.2.1, NB.1.8.1, XFG, and BA.3.2. Transfected cells were transduced with VSVΔG(G)-GFP for 24 h, after which cells were pelleted and virus-containing supernatant was collected. Before collection, residual VSVΔG was neutralized using anti-VSV-G antibody (clone 8G5F11; Kerafast, EB0010), and cells were assessed for cytopathic effects due to infection.

Pseudovirus neutralization assay

Serial dilutions of heat-inactivated mouse immune sera (3-fold) were incubated with pseudovirus (VSVΔG(G)-GFP expressing SARS-CoV-2 S protein) for 1 h at 37 °C before inoculating confluent Vero (ATCC, CCL81.2) cell monolayers in 96-well plates. Fluorescent virus-infected foci were detected 19–21 h after inoculation with an anti–VSV polyclonal antibody (Imanis Life Sciences, REA005) and Alexa488-conjugated secondary antibody (Invitrogen, A-11008) and enumerated using a CTL Immunospot Analyzer (Cellular Technology Limited). A 50% neutralization titer (NT₅₀) was calculated as the last reciprocal serum dilution at which 50% of the virus is neutralized compared to wells containing virus only. Each serum sample dilution was tested in duplicate. Any serum sample that yielded a titer >43,470 was prediluted and repeated to extend the upper titer limit; sera that failed to neutralize at the lowest serum dilution (1:20) were reported to have a neutralizing titer of 20 (lower limit of detection, LLOD).

JN.1, KP.2, XEC and LP.8.1 antigenic cartography

Antigenic maps were constructed using the antigenic cartography toolkit Racmacs v1.2.9 package (https://acorg.github.io/Racmacs/index.html) in R, as previously described¹⁸. Antigenic distance was visually estimated using the antigenic unit grid of the map. The spacing between grid lines is one unit of antigenic distance, corresponding to a two-fold dilution of anti-serum in the neutralization assay. Distance was calculated using Euclidean distance between two coordinates in the optimized antigenic map space, utilizing the “mapDistances(), agCoords(), srCoords()” function in Racmacs.

Global prevalence analysis of JN.1 sublineages and BA.3.2

SARS-CoV-2 virus sequences were downloaded from GISAID²⁷ in November 2025, and XEC, LP.8.1, LF.7, NB.1.8.1, XFG, and BA.3.2 lineages were assigned using Nextclade⁴⁰. Each major lineage group includes this specific lineage name as labeled and its sublineages. Isolates were filtered by collecting data starting from October 2024 or January 2025. Global prevalence trends were plotted using R.

T-cell response assay

Spleens were harvested from vaccinated mice, processed into a single cell suspension, and stimulated ex vivo with overlapping peptides representing the full S protein from Wuhan (WT), BA.4/5, and JN.1 lineages and KP.2, XEC, and LP.8.1 sublineages. Unstimulated splenocytes were treated with DMSO only. Six individual, custom peptide pools (15aa, 11aa overlap, 1 to 2 µg/mL/peptide) represented the full-length S sequence of the WT strain (JPT), BA.4/5 (JPT), JN.1 (JPT), KP.2 (Mimotopes), XEC (Mimotopes), and LP.8.1 (Mimotopes) lineages. Following stimulation, splenocytes were stained for CD154- (CD40L), CD69-, IFN-γ-, TNF-α-, IL-2- and IL4-positive CD4⁺ and CD8⁺ T cells, as previously described^18,19. Data acquisition was performed on a 5-laser Cytek® Aurora system using SpectroFlo® software (v 3.1.2). The instrument was subject to daily quality control procedures using SpectroFlo® QC Beads per manufacturer recommendations. Acquired data files were analyzed using OMIQ® software. The gating strategy is shown in Supplementary Fig. 5.

Statistical analysis

Immunogenicity titers from mouse sera were log₁₀-transformed for analysis. Vaccine group means at the last post-vaccination time point were compared using pairwise contrasts resulting from a one-way ANOVA. Dunnett’s procedure was used on the contrasts to control the familywise significance level at 0.05 within each pseudovirus, using either the JN.1 or KP.2 vaccine as reference. Means and mean differences (and their 95% CI) on the log₁₀ scale were inverse-transformed and reported as GMT and GMR, respectively. All analyses were performed with SAS v9.4 software.

Author contributions

C.K. wrote the original draft. K.M. and P.S.A. contributed to the writing of the original draft, data interpretation, investigation, and supervision of the work. K.M. provided conceptualization of the work and designed all studies. C.K., L.T.M., S.R., and W.L. contributed to data visualization. C.K., W.C., W.L., K.R.T., L.T.M., S.R., E.B., A.Y., S.L., S. Shrivastava, S.O., S. Saha, H.Y., L.H., R.M.M., T.C., and P.M. contributed to investigation, methodology, data generation, and interpretation. C.I.C.C. and A.M. contributed to data interpretation. W.S., K.A.S., A.S.A., and U.Ş. contributed to supervising work. All authors contributed to the review and development of the manuscript and have read and agreed to the published version.

Data availability

The datasets generated and/or analyzed during the current study are not publicly available due to sponsor data-sharing policies. Primary datasets are available from the corresponding author upon reasonable request and with permission from Pfizer Inc. All relevant summary data supporting the conclusions of this study are included in the article and/or supplementary materials.

Competing interests

All authors are current or former employees of Pfizer or BioNTech and may, therefore, be respective shareholders. Pfizer and BioNTech participated in the design, analysis, and interpretation of the data as well as the writing of this report and the decision to publish. W.L., U.Ş., K.A.S., and K.M. are inventors on patents related to COVID-19 vaccines. P.S.A. and U.Ş. are inventors on patents related to RNA technology.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41541-026-01515-8.