The study was approved by the IRB of Johns Hopkins University. Written informed consent was obtained from all participants prior to their inclusion in the study. The study population consisted of 5 participants who received either the Pfizer-BioNTech or Moderna SARS-CoV-2 ancestral spike/BA.5 spike bivalent vaccines and 6 participants who received the Pfizer-BioNTech or Moderna SARS-CoV-2 XBB.1.5 monovalent booster vaccines. 1 participant, HD58, was included in both studies. There were 5 men and 6 women with a median age of 33 (range 23-58 years of age). Details about the study participants are provided in Supplementary Table 1 .

An interferon-gamma ELISpot assay (Mabtech Science) was conducted using 181 overlapping ancestral spike peptides, varying in length from 12 to 17 amino acids with a 10 amino acid overlap, which were obtained from Biodefense and Emerging Infections (BEI) Research Resources. 125,000 to 200,000 PBMCs in 200 uL RPMI media with 10% fetal bovine serum were added to each well, and the plates were incubated for 20 hours with peptides at a concentration of 10 ug/ml. Two replicates per peptide were performed. ELISpot plates were analyzed by an independent investigator using an AID iSpot Spectrum and vendor-provided software to quantify IFN-y spot-forming units (SFUs) per well. SFU/million cells was calculated by adjusting spots per well with the appropriate dilution factor. The stimulation index (fold-change over untreated controls) for each donor was calculated by dividing the SFU of the peptide condition by the SFU of the untreated. A positive response was defined as a mean SFU of greater than or equal to 30 with a stimulation index of 3. All of the raw data for all the ELISpot assays are shown in Supplementary Table 2 .

We used the functional expansion of specific T cells (FEST) assay to identify mono- and cross-reactive TCRs. The FEST assay, which has been extensively reported, is a quantitative, reproducible assay that sequences the CDR3 region of the beta chain of the T cell receptor (TCR) of cells that have been cultured with peptide antigens, and therefore can identify expanded antigen-specific clones (25). Although it was originally developed to study cancer neoantigens (25–27), it has recently been adapted to robustly identify virus-specific T cell responses (24, 28–30, 41). The antigen specificity of TCRs identified via FEST has been previously validated by cloning the receptors into Jurkat cell lines (28), thus supporting the high accuracy of this assay in identifying bona fide antigen-specific TCRs. We have also used CMV, EBV and influenza peptides (JPT Peptide Technologies) as a control to confirm the specificity of the TCR response to SARS-CoV-2 spike peptides (28). For the present study, the ancestral, Omicron BA.5, and BA.2.86, spike peptide pools (JPT Peptide technologies) were utilized to activate CD8+ T cell-depleted PBMCs sourced from 6 participants. Initially, 2 million PBMCs were cultured in a medium comprising IMDM, 5% human AB serum, 10 IU/mL IL-2, and 50 μg/mL gentamicin, along with 1 μg/mL of either ancestral, BA.5, BA.2.86, or XBB.1.5 peptides. Each experimental condition underwent triplicate testing. Media replenishment occurred on days 3 and 7 by replacing half of it with fresh culture media. Harvesting of cells and subsequent DNA extraction from CD8+ T cell-depleted PBMCs utilized the QIAmp micro-DNA kit following the manufacturer’s guidelines (QIAGEN) on day 10. Subsequently, TCR-Seq analysis of DNA extracted from cultured CD8+ T cell-depleted PBMCs was conducted by the Johns Hopkins FEST and TCR Immunogenomics Core Facility using the AmpliSeq for Illumina TCR beta-SR panel. Unique molecular identifiers were used in the PCR amplification step to reduce amplification bias. Sequencing was performed on an Illumina MiSeq platform with unique dual indexes after pooling the samples. Preprocessing steps were implemented to remove nonproductive TCR sequences, align, and trim nucleotide sequences to retain only the CDR3 region. Sequences not conforming to specific criteria, such as those not starting with C or ending with F or W, and having fewer than 7 amino acids in the CDR3 were excluded. For CD8+ T cell-depleted PBMCs stimulated with BA.2.86 spike peptides, the median total number of receptors analyzed per person was 2,209 T cell receptors, ranging from 1,165 to 3,127 receptors. For CD8+ T cell-depleted PBMCs stimulated with XBB.1.5 spike peptide, the median total number of receptors analyzed per person was 2,148 T cell receptors and ranged from 1,593 to 3,070 receptors. Processed data files were then uploaded to our publicly available MANAFEST analysis web application (http://www.stat-apps.onc.jhmi.edu/FEST/) to bioinformatically identify antigen-specific T cell clonotypes. A response was considered positive based on specific criteria, including a mean frequency threshold of greater than 0.1% for each of the three replicates, with at least two replicates having a frequency greater than 0.1%, and the mean frequency being at least 5-fold greater than the mean frequency of wells containing DMSO alone. A mono-reactive response to ancestral, BA.5, BA.2.86, or XBB.1.5 spike was identified if all three conditions were met, and the mean frequency of the three replicates was also 5-fold higher than the response to the other spike protein. In the analysis of spike specific TCRs pre- and post-XBB.1.5 booster, we included TCRs that were identified as mono-reactive or cross-reactive TCRs pre-booster and were present post-booster, regardless of whether they expanded in the FEST assay post-booster. Additionally, we included spike specific TCRs that were identified post-booster and were also present pre-booster, irrespective of their expansion in the FEST assay pre-booster. All of the individual receptors analyzed in all the FEST assays are shown in Supplementary Table 2 .

We conducted multiplex cytokine analysis using MSD assays, utilizing the V-Plex SARS-CoV-2 Panel 2 from MesoScale Diagnostics. This allowed us to measure TNF-alpha, IFN-gamma, and IL4 levels following stimulation with various spike peptides (ancestral, BA.5, BA.2.86, and XBB.1.5). Samples were collected from the cell culture medium on day 3 post-stimulation. Our assays followed the manufacturer’s protocol for accurate cytokine concentration measurement. Validation was performed by ensuring samples fell within the established detection range via standard curve and positive controls. Data analysis was conducted using the MSD discovery workbench software.

Data were analyzed using GraphPad Prism (version 10.4.1). All statistical tests were two-tailed, and a p-value of < 0.05 was considered statistically significant.

For TCR and cytokine analysis where multiple conditions were compared across the same set of participants at a single time point ( Figure 1C , Supplementary Figure 2 ), One-Way ANOVA with Tukey Multiple Comparison Testing was performed.

The paired t-test was used to compare cytokines and the frequency of XBB.1.5 mono-reactive TCRs before and after the XBB.1.5 monovalent vaccine ( Figures 2 , 3 ).

For the comparison of epitopes in study participants who did and did not receive the XBB.1.5 booster ( Figure 4C ), the Mann-Whitney U Test was performed.

We first tested the individual response to 181 overlapping peptides spanning the ancestral spike protein using IFNγ ELISpot (24) in 6 healthy donors who had received the ancestral and BA.5 spike mRNA bivalent vaccine. A median of 32 epitopes per donor was identified (range 5 to 90). Of these epitopes, a median of 20.8%, 29.9%, and 40% contained amino acid mutations in the BA.5 (24), XBB.1.5 and BA.2.86 subvariants respectively ( Figure 1A , Supplementary Figure 1 ).

We implemented a multiplex electrochemiluminescent assay to measure the secretion of IFN-gamma, TNF-alpha, and IL-4 following stimulation with different spike peptides. There were no significant differences in the secretion of any of the cytokines measured in response to stimulation with ancestral peptides versus BA.5 and BA.286 peptides ( Supplementary Figure 2 ).

Using the FEST assay, we assessed the proportion of CD4+ T cells demonstrating cross-recognition of ancestral BA.5 or BA.2.86 spike proteins in 6 donors. We also identified the percentage of clonotypes that were mono-reactive for each spike protein. After culturing CD8+ T cell-depleted PBMCs with ancestral, BA.5, or BA.2.86 spike peptide pools, we sequenced T cell receptor (TCR) Vβ CDR3 repertoires on day 10. We then used a bioinformatic pipeline to identify mono- and cross-reactive TCR clonotypic expansions (26). Across our participant cohort, we observed 5 broad categories of spike-specific TCRs: those mono-reactive to ancestral, BA.5, or BA.2.86 peptides; those cross-reactive for ancestral and BA.5 peptides and those cross-reactive for ancestral and BA.2.86 spike proteins. Examples of cross-reactive and mono-reactive clonotypes are shown in Figure 1B . For each of the six participants analyzed, we determined the absolute number of TCRs falling into each category and then calculated the percentage of spike-specific TCRs exhibiting mono-reactivity versus cross-reactivity.

As shown in Figure 1C , the median percentage of total spike-specific T cells demonstrating cross-reactivity to ancestral and BA.5 spike protein was 40.9% (range: 34.3%-46.0%). The median percentage of T cells that cross-recognized ancestral and BA.2.86 spike peptides was comparable at 37.8% (range: 24.7%-45.5%). 6.6% of the spike-specific T cells were mono-reactive for the ancestral spike protein (range: 2.8%-10.4%), while 8% (range: 2.8%-26.8%) and 3.8% (range: 0.0%-8.7%), of spike-specific T cells were mono-reactive for the BA.5 and BA.2.86 subvariants respectively.

We measured T cell responses in 5 individuals before and after the XBB.1.5 booster. IFN-gamma, TNF-alpha, and IL4 concentrations were measured in response to stimulation of CD8+ T cell-depleted PBMCs with different spike peptides pre- and post-XBB.1.5 booster administration. The median concentrations of IFN-gamma and IL4 were not significantly different in response to stimulation with the different spike peptides pre and post the XBB.1.5 booster. In contrast, there was a significant increase in TNF-alpha secretion after stimulation with all 3 spike proteins after the booster was given ( Figure 2 ).

Mono-reactive TCRs have previously been shown to have higher functional avidity than cross-reactive TCRs (28, 42). Thus, we used the FEST assay to assess the proportion of CD4+ T cells that were cross-reactive for ancestral and BA.5, or ancestral and XBB.1.5 spike proteins compared to those that were mono-reactive for each protein before and after the XBB.1.5 booster. As shown in Figure 3A , the median percentage of total spike-specific T cells that were cross-reactive for ancestral and BA.5 spike protein was 36.6% (range: 33.0%-44.1%) before the XBB.1.5 booster, and 35.1% (range: 33.1%-43.3%) after the booster. The percentage of TCRs that were cross-reactive for ancestral and XBB.1.5 spike protein was 39.0% (range: 30.9%-46.4%) pre-booster and 38.1% (range: 30.6%-46.0%) after the booster. The percentage of cells that were mono-reactive for the ancestral spike protein was 6.4% (range: 5.0%-9.8%) pre-booster and 7.1% (range: 4.5%-12.3%) after the booster. A median of 4.9% of the cells were mono-reactive for BA.5 spike protein (range: 0.9%-11.7%) pre-booster, with a median of 8.1% (range: 1.3%-11.2%) after the booster. Most importantly, a median of 7.6% of the cells were mono-reactive for XBB.1.5 spike (range: 3.6%-19.1%) before the booster, compared to a median of 12.3% (range: 4.1%-17.3%) after the booster. There were also similar numbers of individual clonotypes mono-reactive for XBB.1.5 peptides before and after the booster ( Figure 3B ).

We next asked whether the XBB.1.5 booster would lead to an increased number of targeted peptides that would be affected by XBB.1.5 mutations. PBMCs from 2 study participants (HD58 and HD75) were stimulated with each of 181 overlapping ancestral spike peptides individually in an IFN-gamma ELISpot. In these participants, 23 and 43 individual peptides respectively were identified as targets before the booster. Of these targeted peptides, 34.8% and 20.9% would be affected bymutations in the XBB.1.5 subvariant. After the booster, 18 and 61 individual peptides were targeted ( Figure 4A ). The percentage of these peptides that have mutations in the XBB.1.5 subvariant was not markedly different (27.8% and 19.7% respectively, Figure 4A ). We also analyzed another 3 individuals only after the XBB.1.5 booster because PBMCs before the booster were not available. These individuals targeted 21.7%, 24.4%, and 33.3% of peptides respectively that would be affected by XBB.1.5 mutations ( Figure 4B ). In all, the 5 recipients of the XBB.1.5 booster did not target significantly more of the ancestral spike peptides with potential XBB.1.5 mutations than 6 participants who either did not receive this booster or were tested before the booster was given ( Figure 4C ).