Cross-reactive memory T cells associate with protection against SARS-CoV-2 infection in COVID-19 contacts

Despite widespread vaccination, correlates of protection against SARS-CoV-2 infection remain uncertain. Not all exposures lead to infection, raising the possibility that pre-existing T cells primed by endemic human coronaviruses (hCoVs) provide protection in SARS-CoV-2–naïve individuals. Prior studies documented SARS-CoV-2–reactive T cells in unexposed individuals and in patients with COVID-19, but an association with real-world infectious exposure had not been established. This study investigates whether the frequency of pre-existing cross-reactive T cells at the time of exposure correlates with protection from infection in household contacts of newly diagnosed COVID-19 cases. Using a dual-cytokine FLIspot assay targeting epitopes from spike (S), nucleocapsid (N), membrane (M), envelope (E) and ORF1 proteins, the authors quantify IFN-γ and IL-2 responses shortly after exposure and relate baseline T cell frequencies to subsequent PCR outcomes.

Previous work has shown SARS-CoV-2–specific or cross-reactive T cells in pre-pandemic cohorts and in COVID-19 patients (e.g., Mateus et al., Grifoni et al., Le Bert et al., Bacher et al.). Studies using limited, well-defined epitopes often reported higher detection of cross-reactive responses than studies using large protein-spanning pools. Cross-reactivity is particularly noted in S protein, but significant homology exists across SARS-CoV-2 and endemic hCoVs in multiple proteins, including ORF1, N, M, and E. These observations motivated the design of an epitope pool leveraging predicted HLA-binding motifs and conserved regions shared by SARS-CoV-2 and hCoVs (notably OC43 and HKU1), supplemented with previously confirmed cross-reactive epitopes from the literature.

Study design and participants: The INSTINCT prospective household contact study enrolled 52 exposed contacts of newly diagnosed COVID-19 cases. Baseline sampling occurred 1–6 days after index case symptom onset, with follow-up at days 7 and 28. Nasal swabs at baseline, day 4, and day 7 were tested by RT-PCR (targeting E gene and ORF1a). Blood was collected at all time points; PBMCs were isolated at baseline and cryopreserved. Ethical approval: North West – Greater Manchester East Research Ethics Committee (IRAS: 282820; REC 20/NW/031). Epitope prediction and peptide pools: Consensus SARS-CoV-2 protein sequences (S, N, M, E, ORF1) were used to predict HLA class I and II epitopes via SYFPEITHI/IEDB/NetMHCpan 3.1. Cross-reactive candidates were identified by aligning SARS-CoV-2 and hCoV (OC43, HKU1) sequences to find conserved or homologous motifs predicted to bind the same HLA alleles. The cross-reactive pool combined 28 in silico–predicted epitopes with 14 in vitro–confirmed cross-reactive epitopes from prior work. Protein-spanning overlapping peptide pools for S, N, M, and E were also used. Dual cytokine FLIspot assay: Cryopreserved PBMCs were rested overnight and stimulated with peptide pools (1 µg/mL) to enumerate IFN-γ and IL-2 spot-forming cells (SFC) per 1×10^5 PBMCs. Controls included a CMV/EBV/influenza (CEFT/CEB) peptide pool and anti-CD28 stimulation. Responses were measured at baseline and at follow-up time points for longitudinal analyses. Short-term T cell lines: Antigen-specific T cell lines were generated from PBMCs with peptide stimulation plus IL-2, expanded over ~12 days, and re-stimulated for functional confirmation via ELISPOT. Serology: RBD-specific total antibodies were quantified using a double-antigen bridging assay (dABA), reporting binding ratios (BR), with BR ≥ 1 considered positive. Exposure and hCoV serology: An exposure score was defined based on relationship/exposure to the index case. Baseline sera were assayed for antibodies to seasonal hCoVs (NL63, 229E, OC43, HKU1). Statistical analyses: Group comparisons used Welch’s t-test, Mann–Whitney U test, Fisher’s exact test, Pearson/Spearman correlations, binary regression for odds ratios, and mixed-effects models for longitudinal changes. Outcomes (PCR-positive vs PCR-negative) were related to baseline T cell frequencies while assessing potential confounders (exposure score, timing since index symptom onset, age).

- Among 52 exposed household contacts, 26 remained PCR-negative and 26 became PCR-positive. Baseline PBMCs (1–6 days post-index symptom onset) showed detectable SARS-CoV-2–reactive T cells in both groups. - Cross-reactive IL-2-secreting T cells were significantly more frequent at baseline in contacts who remained PCR-negative: Welch’s t-test p = 0.013 (reported as p = 0.0139 in the abstract). In adjusted binary regression, higher cross-reactive IL-2 T cell frequency predicted PCR negativity (OR 10.95; 95% CI 1.011–12.1; p = 0.0295). - Nucleocapsid (N)-specific IL-2-secreting T cells were also higher in PCR-negative contacts (p = 0.0355). - No significant differences were observed for spike-specific T cell responses between PCR-negative and PCR-positive contacts, suggesting limited protective effect from spike-cross-reactive T cells in this cohort. - There were no significant differences in total responses to S, M, E, N pools (IFN-γ or IL-2) between groups overall (e.g., p = 0.098 for cumulative responses). - Exposure metrics were similar between groups: relationship score (p = 0.5004) and days since index symptom onset at baseline (p = 0.2935). - PCR-positive contacts had significantly lower lymphocyte counts than PCR-negative contacts (median 1.3 vs 1.7 ×10^9/L; p = 0.0039), yet their antigen-specific T cell frequencies to pools were not reduced. - Following exposure/infection, 91% (n = 22 with follow-up) of PCR-positive contacts developed RBD-specific antibodies, confirming infection. These individuals also showed strong induction of IFN-γ and IL-2 T cells to conserved SARS-CoV-2 epitopes. - In PCR-negative contacts with baseline cross-reactive responses, frequencies of cross-reactive IL-2-secreting T cells changed dynamically post-exposure, consistent with peripheral depletion, potentially due to migration to respiratory mucosa (mixed-effects p = 0.0039). Control viral peptide responses (CMV, EBV, influenza) did not show such changes. - No association was found between baseline cross-reactive SARS-CoV-2–specific T cell frequency and age (Spearman r = -0.174; p = 0.216), nor between PCR status and age (p = 0.5863); the cohort had limited age range with few children.

Baseline, pre-existing IL-2-secreting cross-reactive T cells, likely primed by prior exposure to endemic hCoVs, were associated with protection against acquiring SARS-CoV-2 infection after household exposure. The lack of a spike-specific differential suggests that non-spike targets (e.g., N, ORF1) may be more relevant for protective cross-reactive T cell immunity. The kinetics in PCR-negative contacts—declines in circulating cross-reactive T cells after exposure—imply biological engagement, possibly trafficking to the respiratory mucosa. In PCR-positive contacts, seroconversion (RBD antibodies) and robust induction of both IFN-γ and IL-2 T cells to conserved epitopes validate the peptide pools and confirm infection. These findings provide evidence that pre-existing, non-spike cross-reactive memory T cells can contribute to sterilizing protection in SARS-CoV-2–naïve individuals, supporting T cell-focused vaccine strategies that include conserved non-spike antigens to mitigate immune escape by spike variants.

The study demonstrates that higher baseline frequencies of non-spike, cross-reactive IL-2–secreting memory T cells correlate with protection from SARS-CoV-2 infection in exposed household contacts, whereas spike-specific responses did not differ between infected and uninfected groups. This supports incorporating conserved non-spike antigens (e.g., N, ORF1) into second-generation vaccines to elicit protective T cell immunity resilient to spike antigenic drift. Future research should validate these findings in larger, diverse cohorts, delineate the phenotype and tissue homing of protective T cells (including mucosal compartments), define precise protective thresholds, and assess vaccine designs that broaden T cell targets.

- Modest sample size (n = 52) limits statistical power and generalizability. - Limited age range with few children; no association with age detected, but age effects cannot be fully assessed. - Inability to adjust associations for symptom severity or peak viral load because baseline cross-reactive T cells were largely absent in PCR-positive contacts. - Potential measurement constraints: peripheral blood sampling may not capture mucosal T cell responses; dynamic trafficking could influence circulating frequencies. - Some methodological details (e.g., epitope pool composition, assay parameters) rely on in silico predictions and previously validated epitopes, which may not capture all relevant cross-reactive determinants.