Ancestral SARS-CoV-2-speci � c T cells cross-recognize Omicron ( B . 1 . 1 . 529 )


 The emergence of the SARS-CoV-2 variant-of-concern Omicron (B.1.1.529) has destabilized global efforts to control the impact of COVID-19. Recent data have suggested that B.1.1.529 can readily infect people with naturally acquired or vaccine-induced immunity, facilitated in some cases by viral escape from antibodies that neutralize ancestral SARS-CoV-2. However, severe disease appears to be relatively uncommon in such individuals, highlighting a potential role for other components of the adaptive immune system. We report here that SARS-CoV-2 spike-specific CD4+ and CD8+ T cells induced by prior infection and, more extensively, by mRNA vaccination provide comprehensive heterologous immune reactivity against B.1.1.529. Pairwise comparisons across groups further revealed that SARS-CoV-2 spike-reactive CD4+ and CD8+ T cells exhibited similar functional attributes, memory distributions, and phenotypic traits in response to the ancestral strain or B.1.1.529. Our data indicate that established SARS-CoV-2 spike-specific CD4+ and CD8+ T cell responses, especially after mRNA vaccination, remain largely intact against B.1.1.529.

To address this question, we collected peripheral blood mononuclear cells from mRNA-vaccinated individuals 6 months after the second dose of P zer/BioNTech BNT162b2 (n = 40), individuals in the convalescent phase 9 months after mild or severe COVID-19 (n = 48), and seronegative individuals (n = 48). Cells were stimulated in parallel with overlapping peptide pools spanning the entire spike protein sequences of the Wuhan reference strain (wildtype) or B.1.1.529. Activation-induced marker assays were used to quantify spike-speci c CD4 + T cell responses via the upregulation of CD69 and CD40L (CD154) and spike-speci c CD8 + T cell responses via the upregulation of CD69 and 4-1BB (CD137) (Extended Data Fig. 1a).
The overall magnitude of the SARS-CoV-2 spike-speci c CD4 + T cell response against B.1.1.529 showed a median reduction of 9% in mRNA-vaccinated individuals and a median reduction of 16% in convalescent individuals relative to the wildtype response (Fig. 1a, b). The corresponding response frequencies were also slightly lower for B.1.1.529 (Fig. 1c). Individual comparisons further revealed maximum reductions in magnitude of 58% in mRNA-vaccinated individuals, 56% in convalescent individuals, and 75% in seronegative individuals, comparing SARS-CoV-2 spike-speci c CD4 + T cell responses against B.1.1.529 versus wildtype (Fig. 1d). These ndings were validated using independently synthesized peptide pools spanning the entire spike protein (Extended Data Fig. 1b).
The overall magnitude of the SARS-CoV-2 spike-speci c CD8 + T cell response against B.1.1.529 showed a median reduction of 8% in mRNA-vaccinated individuals and a median reduction of 30% in convalescent individuals relative to the wildtype response (Fig. 2a, b). These differences were mirrored in the corresponding response frequencies for B.1.1.529 versus wildtype (Fig. 2c). Individual comparisons further revealed maximum reductions in magnitude of 55% in mRNA-vaccinated individuals, 63% in convalescent individuals, and 60% in seronegative individuals, comparing SARS-CoV-2 spike-speci c CD8 + T cell responses against B.1.1.529 versus wildtype (Fig. 2d). These ndings were again validated using independently synthesized peptide pools spanning the entire spike protein (Extended Data Fig. 1c).
In further experiments, we investigated the phenotypic characteristics of SARS-CoV-2 spike-speci c CD8 + T cells that cross-recognized B.1.1.529, focusing on classic markers of memory differentiation (CCR7, CD45RA). Late effector memory T (T EMRA ) cells predominated among SARS-CoV-2 spike-speci c CD8 + T cells in mRNA-vaccinated, convalescent, and seronegative individuals, but no signi cant differences in subset composition were detected across intragroup comparisons of SARS-CoV-2 spike-speci c CD8 + T cell responses against B.1.1.529 versus wildtype (Fig. 2e). We also assessed the functionality of SARS-CoV-2 spike-speci c CD8 + T cells in mRNA-vaccinated individuals, measuring the intracellular expression of granzyme B, IFN-γ, TNF, and IL-2 alongside CD69 and CD137. Akin to the corresponding analyses of SARS-CoV-2 spike-speci c CD4 + T cells, no signi cant differences in the ability of SARS-CoV-2 spikespeci c CD8 + T cells to deploy multiple functions were apparent in response to stimulation with peptides representing B.1.1.529 versus wildtype (Fig. 2f). In a nal set of analyses, we merged all SARS-CoV-2 spike-speci c CD4 + and CD8 + T cell data points together and found a signi cantly lower total T cell response against B.1.1.529 versus wildtype in convalescent, but not vaccinated subjects (Extended Data Fig. 1d). These analyses suggest that SARS-CoV-2 spike-speci c T cell responses remain less intact against B.1.1.529 in convalescent individuals.
The current global pandemic has been destabilized by the recent emergence of B.1.1.529, which continues to spread rapidly and supersede other VOCs. Our collective data indicate that SARS-CoV-2 spike-speci c CD4 + and CD8 + T cells elicited by mRNA vaccination or prior infection remain largely intact against B.1.1.529. Alongside intrinsic viral factors, such as altered tropism and decreased replication in the lower respiratory tract 21 , such heterologous immune reactivity may explain why severe disease appears to be relatively uncommon after infection with this particular VOC. Moreover, the degree of crossreactivity varied to some extent among individuals, most likely as a consequence of genetically encoded differences in antigen presentation, which could further modulate clinical outcomes associated with B.1.1.529. It should be noted that our evaluations were con ned to peripheral blood samples, which do not necessarily re ect the entirety of the cellular immune response against SARS-CoV-2 22 . In addition, SARS-CoV-2 spike-speci c CD4 + and CD8 + T cells cross-recognized B.1.1.529 less comprehensively in convalescent versus mRNA-vaccinated individuals, suggesting that booster immunization may provide bene ts that extend beyond the induction of broadly neutralizing antibodies to enhance natural protection against recurrent episodes of COVID-19 2 .  Figure 1). Convalescent individuals after previous COVID-19 were sampled 9 months after RT-PCR veri ed SARS-CoV-2 infection leading to mild (i.e. non-hospitalized, n=26) or severe (i.e. hospitalized, n=22) disease during the rst wave of the pandemic in March-April of 2020, before the emergence of the Alpha, Beta and Delta variants. The cohort of convalescent individuals after mild COVID-19 consisted of 8 women (31%)/18 men (69%), and the median age (range) was 54 (44-68) years.

Declarations
The cohort of convalescent individuals after severe COVID-19 consisted of 3 women (14%)/19 men (86%), and the median age (range) was 58 (33-66) years. Fourteen of the patients with severe COVID-19 (64%) had received intensive care at the time of COVID-19. None of the convalescent individuals had received any COVID-19 vaccination at the time of sample collections. Seronegative samples were identi ed by exclusion using a LIVE/DEAD Fixable Aqua Dead Cell Stain Kit (Thermo Fisher Scienti c). Cells were then washed in FACS buffer and xed/permeabilized using a FoxP3 Transcription Factor Staining Buffer Set (Thermo Fisher Scienti c). Intracellular stains were performed for 30 min at room temperature. Stained cells were washed in FACS buffer, xed in PBS containing 1% PFA (Biotium), and acquired using a FACSymphony A5 (BD Biosciences). All ow cytometry reagents are detailed in Supplementary Table 1.

Data analysis and statistics
Flow cytometry data were analyzed using FlowJo version 10.8.0 (FlowJo LLC). Stimulation indices were calculated as fold change in frequency relative to the negative control (equivalent DMSO). Positive responses were identi ed using a threshold stimulation index >2. Statistical analyses were performed using Prism version 9 (GraphPad). Signi cance between two paired groups was assessed using the Wilcoxon signed rank test. Functional pro les were deconvoluted using Boolean gating in FlowJo version 10.8.0 (FlowJo LLC) followed by downstream analyses in SPICE version 6.1 (https://niaid.github.io/spice/).

Data availability
Requests for data should be addressed to the corresponding author and will be made available within the bounds of con dentiality and data protection obligations.