Neutralization and spike stability of JN.1-derived LB.1, KP.2.3, KP.3, and KP.3.1.1 subvariants

Abstract

During the summer of 2024, coronavirus disease 2019 (COVID-19) cases surged globally, driven by variants derived from JN.1 subvariants of severe acute respiratory syndrome coronavirus 2 that feature new mutations, particularly in the N-terminal domain (NTD) of the spike protein. In this study, we report on the neutralizing antibody (nAb) escape, infectivity, fusion, and spike stability of these subvariants-LB.1, KP.2.3, KP.3, and KP.3.1.1. Our findings demonstrate that all of these subvariants are highly evasive of nAbs elicited by the bivalent mRNA vaccine, the XBB.1.5 monovalent mumps virus-based vaccine, or from infections during the BA.2.86/JN.1 wave. This reduction in nAb titers is primarily driven by a single serine deletion (DelS31) in the NTD of the spike, leading to a distinct antigenic profile compared to the parental JN.1 and other variants. We also found that the DelS31 mutation decreases pseudovirus infectivity in CaLu-3 cells, which correlates with impaired cell-cell fusion. Additionally, the spike protein of DelS31 variants appears more conformationally stable, as indicated by reduced S1 shedding both with and without stimulation by soluble ACE2 and increased resistance to elevated temperatures. Molecular modeling suggests that DelS31 enhances the NTD-receptor-binding domain (RBD) interaction, favoring the RBD down conformation and reducing accessibility to ACE2 and specific nAbs. Moreover, DelS31 introduces an N-linked glycan at N30, shielding the NTD from antibody recognition. These findings underscore the role of NTD mutations in immune evasion, spike stability, and viral infectivity, highlighting the need to consider DelS31-containing antigens in updated COVID-19 vaccines.IMPORTANCEThe emergence of novel severe acute respiratory syndrome coronavirus 2 variants continues to pose challenges for global public health, particularly in the context of immune evasion and viral stability. This study identifies a key N-terminal domain (NTD) mutation, DelS31, in JN.1-derived subvariants that enhances neutralizing antibody escape while reducing infectivity and cell-cell fusion. The DelS31 mutation stabilizes the spike protein conformation, limits S1 shedding, and increases thermal resistance, which possibly contribute to prolonged viral persistence. Structural analyses reveal that DelS31 enhances NTD-receptor-binding domain interactions by introducing glycan shielding, thus decreasing antibody and ACE2 accessibility. These findings emphasize the critical role of NTD mutations in shaping viral evolution and immune evasion, underscoring the urgent need for updated coronavirus disease 2019 vaccines that account for these adaptive changes.

Keywords: COVID-19 vaccine; DelS31 mutation; JN.1 subvariants; SARS-CoV-2; cell-cell fusion; immune evasion; spike protein stability.

Fig 1

Infectivity of JN.1 subvariants in 293T-ACE2 and CaLu-3 cells. (a) Mutations that characterize JN.1-derived subvariants FLiRT, KP.2, KP.3, LB.1, KP.2.3, and KP.3.1.1. Recorded incidences of infection by relevant variants in (b) the USA and (c) globally based on data collected by the Centers for Disease Control and Prevention and Global Initiative of Sharing All Influenza Data. Other JN.1*: JN.1 subvariants. Infectivity of pseudotyped lentiviral vectors bearing variant spikes of interest was determined in (d) 293T-ACE2 cells and (e) CaLu-3 cells. Bars in panels d and e represent means and standard deviation from six independent infections (n = 6). Significance was determined and displayed relative to the parental FLiRT, KP.2 and KP.3 variants in each group; stars represent **P < 0.01 and ***P < 0.001; ns, P > 0.05.

Fig 2

Neutralization of JN.1 variants by antibodies in bivalent-vaccinated HCWs, XBB.1.5-vaccinated hamsters, BA.2.86/JN.1-infected people, and monoclonal antibody S309. NAb titers were determined against JN.1-derived variants of interest in the sera of (a and b) HCWs that received at least two doses of monovalent mRNA vaccine and a dose of bivalent (WT + BA.4/5) mRNA vaccine (n = 10), (c and d) individuals that were infected during the BA.2.86/JN.1 wave of infection in Columbus, Ohio (n = 10), and (e and f) golden Syrian hamsters that were vaccinated with two doses of a mumps virus-based monovalent XBB.1.5 spike vaccine (n = 10). Plots in panels a, c, and e represent geometric mean nAb titers at 50% with standard errors. Geometric mean antibody titers are depicted at the top of the plots with fold changes relative to JN.1 above them. (b, d, and f) Heatmaps that depict the corresponding nAb values for each cohort listed by individual samples. Significance was determined and displayed relative to JN.1 using log10 transformed values, unless otherwise indicated; stars represent *P < 0.05, **P < 0.01, and ****P < 0.0001; ns, P > 0.05. (g) Neutralization by class 3 monoclonal antibody S309 was determined by serial dilution of S309, followed by incubation of the S309 with pseudotyped viruses. Representative plots were displayed. The calculated IC₅₀ values are shown in Fig. S2.

Fig 3

Antigenic mapping of neutralization data against JN.1 variants. The Racmacs program was used to plot relative antigenic distances between each spike antigen (circles) and sera sample (squares) for (a) the bivalent-vaccinated HCWs, (b) the BA.2.86/JN.1 wave-infected people, and (c) the XBB.1.5-vaccinated hamsters. The scale bar represents 1 AU which is equivalent to about a twofold difference in nAb titer. (d) The antigenic distances of each variant relative to JN.1 from three groups of cohorts (n = 3) were averaged and plotted. The scale bar represents 1 AU. The detailed antigenic distances of each subvariant relative to D614G or JN.1 in three different cohorts were presented in Fig. S3a through c.

Fig 4

Cell-cell fusion of JN.1-derived spikes. Fusion triggered between membranes by the spike proteins of interest was determined between 293T cells expressing the spike and 293T-ACE2 cells overexpressing ACE2 (and b) or CaLu-3 cells expressing an endogenous level of ACE2 (c and d). Representative images of fusion are depicted for (a) 293T-ACE2 and (c) CaLu-3, and quantification of total areas of fusion across three images is represented for (b) 293T-ACE2 and (d) CaLu-3. Scale bars represent 750 µM. Bars represent means with standard deviation, significance was determined relative to the ancestral variants FLiRT, KP.2 and KP.3 in each group as indicated, and stars represent *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001; ns, P > 0.05.

Fig 5

Surface expression and processing of JN.1-derived spikes. (a and b) The surface of 293T cells used to produce pseudotyped vectors was probed with anti-S1 antibody to compare surface expression between spikes of interest. (a) Representative histograms depicting surface expression and (b) geometric mean fluorescence intensities (MFIs) of surface S1 are depicted (n = 3). (c) Processing of spikes into S1/S2 subunits by furin was determined by lysing 293T cells used to produce pseudotyped viruses and probed by using anti-S1, anti-S2, anti-p24, and anti-glyceraldehyde-3-phosphate dehydrogenase (anti-GAPDH) antibodies. Relative ratios of S2/S or S1/S were quantified using NIH ImageJ, calculated by comparing to D614G, and are displayed under corresponding blots. The plot in panel b represents geometric means with standard deviation, and significance was determined relative to parental FLiRT, KP.2, or KP.3 variant as indicated; stars represent **P < 0.01, ***P < 0.001, and ****P < 0.0001; ns, P > 0.05.

Fig 6

Stability of JN.1 variant spikes and pseudotyped viral particles. (a) Lentiviral pseudovirions were purified (without serum) and incubated at indicated temperatures (37 to 43°C) for 1 h, and viral infectivity was determined by infecting 293T-ACE2 cells. Relative percent of infection is plotted by comparing the titer at 4°C, which was set to 100%. For each variant, the temperature at which the viral infectivity was lost by 50% (T_1/2) was determined and displayed by a dashed line in panel a and listed in panel b. (c) HEK293T cells were transfected with spike constructs of interest and treated with or without soluble ACE2 (sACE2) (10 µg/mL) for 4 h. Cell culture media and lysates were collected, with shed S1 proteins being immunoprecipitated with an anti-S1 antibody. Relative signals were quantified by NIH ImageJ by setting the value of JN.1 to 1.0. (d) HEK293T cells transfected with spike constructs of interest, without being treated with sACE2, were harvested and stained with anti-S1 antibody; geometric mean fluorescence intensities (MFIs) of anti-S1 signal were collected and depicted (n = 3). The value represents means with standard deviation, and significance was determined relative to parental FLiRT, KP.2, or KP.3 variant as indicated. ***P < 0.001; ****P < 0.0001. (e) Cell lysates were blotted with anti-S2, anti-S1, and anti-GAPDH antibodies, respectively.

Fig 7

Structural modeling of key NTD mutations in LB.1, KP.2.3, and KP.3.1.1. (a) Structural comparisons of NTD between KP.3.1.1 and JN.1 spike proteins. The DelS31 mutation shifts F32, thus altering its side chain direction to form hydrophobic interactions with surrounding NTD core residues, including T29, R34, V62, L56, Y91, and F216, while introducing glycosylation at N30. (b) The DelS31 mutation stabilizes the NTD, reshaping its conformation and enhancing its interaction with the RBD to favor the RBD down conformation. (c) The down conformation reduces RBD accessibility to the ACE2 receptor (yellow surface). (d) This down conformation restricts the accessibility of class 1 and 4 antibodies, but not class 2 and 3 antibodies. Antibodies are depicted as semi-transparent surfaces. (e and f) Mutations at residues (shown as sticks) H146 and Q183 disrupt the epitopes of certain NTD-targeting antibodies, such as 4A8 and C1520. (g) Glycosylation (shown as sticks) at N30 interferes with the recognition of some NTD-targeting antibodies, such as C1717.