Impact of SARS-CoV-2 exposure history on the T cell and IgG response

Summary Multiple severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) exposures, from infection or vaccination, can potently boost spike antibody responses. Less is known about the impact of repeated exposures on T cell responses. Here, we compare the prevalence and frequency of peripheral SARS-CoV-2-specific T cell and immunoglobulin G (IgG) responses in 190 individuals with complex SARS-CoV-2 exposure histories. As expected, an increasing number of SARS-CoV-2 spike exposures significantly enhances the magnitude of IgG responses, while repeated exposures improve the number of T cell responders but have less impact on SARS-CoV-2 spike-specific T cell frequencies in the circulation. Moreover, we find that the number and nature of exposures (rather than the order of infection and vaccination) shape the spike immune response, with spike-specific CD4 T cells displaying a greater polyfunctional potential following hybrid immunity compared with vaccination only. Characterizing adaptive immunity from an evolving viral and immunological landscape may inform vaccine strategies to elicit optimal immunity as the pandemic progress.

In brief Keeton et al. investigate the effect of multiple exposures to SARS-CoV-2 spike (through infection and/or vaccination) on T cell and IgG responses. They show that an increasing number of exposures enhances the magnitude of IgG responses but has no major impact on the magnitude of SARS-CoV-2 spike-specific T cells.

INTRODUCTION
It is now established that a coordinated and robust humoral and cellular response to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) plays a key role in regulating infection, transmission, and disease severity. 1,2 The COVID-19 pandemic over the past 2 and a half years has resulted in a complex virological and immunological landscape, with successive waves of distinct variants (from ancestral WU-1 to Omicron and its sub-lineages), 3 the introduction of different vaccines and booster regimens, 4 and subsequent emergence of breakthrough infections (BTIs). Hence, populations are now composed of heterogeneous groups, from rare immunologically naive individuals to those who have experienced multiple SARS-CoV-2 antigen exposures (through natural infection and/or vaccination), with diverse viral variants. It is thus essential to consider infection history when assessing the immune response to SARS-CoV-2. Recent publications assessing the impact of repeated SARS-CoV-2 exposures on antibody responses showed that an increased number of contacts with SARS-CoV-2 antigen(s) enhanced the quantity and quality of antibody responses, even against variants of concern. [5][6][7][8][9] It is now clear that, compared with vaccination only, infection leads to the generation of a broader T cell response targeting the structural, non-structural, and accessory proteins of SARS-CoV-2, which are detectable in blood and at the sites of infection. 10 These immunological properties could explain enhanced protection in the context of hybrid immunity when compared with vaccination alone. 11 However, less is known about the effect of recurrent exposures on the T cell immune response, in particular the quantity, quality, and crossreactivity of the response. In this study, to understand how the viral sequence of infection and/or vaccination, as well as Article ll repeated exposures, shape the immune response to SARS-CoV-2, we measured peripheral spike and nucleocapsid-specific T cell responses in individuals with a diverse SARS-CoV-2 exposure history, from a single exposure (i.e., one dose of Ad26. COV2.S vaccine or a COVID-19 episode) to four exposures (i.e., COVID-19 episode, prior to two Ad26.COV2.S doses, followed by a BTI), spanning four infection waves sequentially caused by the ancestral WU-1, Beta, Delta, and then Omicron variants.

Study cohort
To define the impact of repeated antigen exposure on SARS-CoV-2 immunity, we measured T cell and immunoglublin G (IgG) responses in 190 healthcare workers from an ongoing longitudinal study, 4,12 all of whom were offered the Janssen Ad26.COV2.S vaccine as part of the Sisonke vaccination study in South Africa. Participants were grouped according to their number of exposures to SARS-CoV-2 antigen through vaccination or a combination of infection and vaccination ( Figure 1A). Infection was ascertained by a positive viral PCR test or conversion to nucleocapsid (N) seropositivity or, in the case of a BTI with an existing nucleocapsid antibody response from previous infection, >2-fold increase in anti-nucleocapsid IgG optical density (OD). The ''1 exposure'' group (n = 73) was composed of individuals vaccinated with a single dose of Ad26.CoV2.S (n = 33) or unvaccinated patients who experienced one COVID-19 episode (n = 40). The ''2 exposures'' group (n = 67) consisted of 9 individuals who received two doses of Ad26.COV2.S approximately 6 months apart, 36 participants who experienced a COVID-19 episode prior to receiving one dose of Ad26. COV2.S, and 22 participants who had a BTI after one dose of Ad26.COV2.S. The ''3 exposures'' group (n = 40) encompassed three different exposure profiles: 20 participants who experienced a COVID-19 episode prior to two doses of Ad26.COV2.S, 10 participants who had a BTI after two doses of Ad26.COV2.S, and 10 participants who had COVID-19 prior to one dose of Ad26.COV2.S followed by a BTI. Finally, the ''4 exposures'' group included 10 individuals who had COVID-19 prior to two doses of Ad26.COV2.S followed by a BTI. However, some asymptomatic infections could have occurred that we did not detect by nucleocapsid serology; BTIs have been described where nucleocapsid-specific IgG did not increase despite a positive PCR test. 13 Thus, we cannot fully exclude the possibility that some participants may have experienced such an infection, where PCR testing was not performed, and nucleocapsid antibody levels did not change. Age and gender were comparable between each group. All COVID-19 cases were asymptomatic or mild and did not require hospitalization. The exact time since last exposure (infection or vaccination) was known for most participants, and the time range since the last negative sample was recorded for nucleocapsid seroconversions. The ''1 exposure'' and ''2 exposures'' groups had a median time since the last SARS-CoV-2 antigen exposure of 5.2 and 4.9 months, respectively. The majority of the ''3 exposures'' group were sampled <1 month since last exposure (0.81 months), and the ''4 exposures'' group all had samples collected <3 months after the last exposure, with 3/10 participants at a median 0.69 months.
Evolution of SARS-CoV-2 T cell and IgG response upon repeated exposures to SARS-CoV-2 antigens We first evaluated CD4 + and CD8 + T cell and IgG responses to ancestral SARS-CoV-2 spike according to the different number of exposures (SARS-CoV-2 infection and/or Ad26.COV2.S vaccination) ( Figure 1A). As expected, abundant spike-specific CD4 + and CD8 + T cell responses were detectable after one to four antigen exposures ( Figure 1B). The proportion of individuals exhibiting a detectable CD8 + T cell response to spike increased significantly between 1 and 3 exposures (p = 0.015), and the same trend was also observed for CD4 + T cell responders (p =0.055). Interestingly, no significant difference in the magnitude of spike-specific CD4 + or CD8 + T cells was detected regardless of the number of exposures. The evolution of the spike-specific T cell response upon repeated exposures is further illustrated in Figure 1C, showing that three exposures led to the highest proportion of concomitant CD4 + and CD8 + T cell responses. We then evaluated the polyfunctional potential of spike-specific T cells based on the number of exposures to SARS-CoV-2 antigen. With an increasing number of exposures, spike-specific CD4 + T cells were characterized by an increase in the proportion of cells producing interferon g (IFN-g) and interleukin-2 (IL-2) simultaneously, which was counterbalanced by a progressive decline in IL-2-and tumor necrosis factor a (TNF-a)-producing cells ( Figure S1A). In contrast, the polyfunctional profile of spike-specific CD8 + T cells was not affected by repeated SARS-CoV-2 antigen exposures, with cells producing mainly IFN-g alone ( Figure S1B).
In parallel, we measured spike-specific serum IgG by ELISA ( Figure 1D). The proportion of spike-specific IgG responders increased significantly between one and two exposures (64%-94%, respectively, p = 0.003), with a concomitant sharp increase in the magnitude of spike-specific IgG (median fold change: 3.8). A more modest but significant increase in magnitude was also observed between the second and third exposures (p = 0.038), with all participants seropositive by the third exposure. Of note, as previously reported, 14,15 the frequency of spike-specific CD4 + T cells was associated with the magnitude of spike IgG (p < 0.0001, r = 0.41; Figure S2A). We also determined whether the magnitude of antibody or T cell responses were related to time since last exposure and found no association for spike-specific T cells or IgG in any exposure group ( Figure S2B). Overall, these data show that spike-specific T cells and spike-specific IgG displayed distinct dynamics upon repeated SARS-CoV-2 antigen exposure and suggest that T cell and IgG responses plateau after three exposures.
Next, to define whether immune responses to SARS-CoV-2 nucleocapsid present a similar profile, we compared nucleocapsid-specific T cell and IgG responses in uninfected participants and patients who had one or two COVID-19 episodes ( Figures 1E and 1F). T cells and IgG targeting nucleocapsid were undetectable in uninfected participants, as expected, and abundantly detectable in those who experienced infection. In convalescents, we did not detect any notable changes in the proportion of responders nor the magnitude of nucleocapsidspecific CD4 + and CD8 + T cells between individuals who experienced one or two COVID-19 episodes. The profile of the nucleocapsid-specific IgG response between one and two exposures was similar to that of spike-specific IgG, with a significant increase in both the proportion of responders (70%-100%, p = 0.008) and the magnitude of the response (median fold change: 2.7) (Figure 1F). Similarly, we found a positive association between the frequency of nucleocapsid-specific CD4 + T cells and nucleocapsid IgG (p < 0.0001, r = 0.52; Figure S2A).
While we did not detect any significant change in the overall magnitude of the spike-specific T cell response upon repeated exposures, it is essential to acknowledge that infection leads to the generation of a broad T cell repertoire targeting SARS-CoV-2 structural, nonstructural, and accessory proteins. Hence, in the context of infection or hybrid immunity, the overall magnitude of the T cell response to SARS-CoV-2 is expected to surpass that of individuals who have been vaccinated and have not been infected. To illustrate this, we compared the combined frequency of spike-and nucleocapsid-specific T cell responses upon different infection and/or vaccination exposures ( Figure 2). As expected, in the context of infection or hybrid immunity, the frequency of spike-plus nucleocapsid-specific CD4 T cells was significantly higher than vaccination alone (Figure 2A). This superior CD4 response magnitude persisted even when compared with two doses of vaccine. Interestingly, for the CD8 response, the combined frequency of spike-and nucleocapsid-specific specific CD8 T cells was not significantly improved after infection or hybrid immunity compared with vaccination alone, but the proportion of responders was augmented ( Figure 2B). Indeed, evaluating the proportion of spike and/or nucleocapsid responders after a SARS-CoV-2 infection showed that while two-thirds of the CD4 responders mounted a response to both spike and nucleocapsid, only a quarter of the CD8 responders exhibited a dual response and another quarter mounted a nucleocapsid response in the absence of a spike response ( Figure 2C). The proportion of spike and nucleocapsid T cell responders stratified by their exposure history is presented in Figure S3. Thus, it is important to acknowledge that the overall magnitude of SARS-CoV-2-specific T cell response was clearly underestimated in our study as responses to other SARS-CoV-2 structural and accessory proteins were not considered.   Table S1. ''3 exposures'' group). Clinical data for each sub-group are presented in Table S1. Where the date of infection was known or estimated within a narrow time window due to closely spaced study visits, we assigned these infections to the dominant circulating variant of that infection wave due to the highly virologically distinct infection waves that have occurred in South Africa.
A significant difference in the spikespecific CD4 + T cell profile was only observed within the ''1 exposure'' group where both the proportion of responders (70% versus 90%, p = 0.028) and the frequency of spike-specific CD4 + T cell responses (p = 0.0003) was significantly higher in the SARS-CoV-2-infected group compared with the Ad26.COV2.S-vaccinated individuals ( Figure 3A). On the contrary, spike-specific CD8 + T cell and IgG responses were comparable between these two sub-groups ( Figures 3B and 3C). For the ''2 exposures'' and ''3 exposures'' groups, there was no significant difference in the proportion of responders and the magnitude of spike-specific T cells and IgG regardless of the sequence of exposures (Figures 3A-3C). Moreover, for participants who experienced a COVID-19 episode, the infecting strain did not appear to impact the profile of spike-specific T cell responses. When comparing the polyfunctional profile of spike-specific CD4 + T cells according to the nature and sequence of exposures, our data show that in the context of a single exposure, CD4 responses to natural infection were enriched in polyfunctional cells (producing IFN-g, IL-2, and TNF-a) compared with those who received a single dose of Ad26.COV2.S ( Figure 4A) Figure 4B).
Overall, these data show that (1) in the case of a single exposure, SARS-CoV-2 infection induces a more robust spikespecific CD4 + T cell response compared with a single dose of Ad26.COV2.S; (2) the number of SARS-CoV-2 exposures, rather than the sequence in which vaccination or infection occur, shapes the SARS-CoV-2 immune response; and (3) exposure to viral variants preserves T cell responses to ancestral spike.
Profile and cross-reactivity of spike-specific T cells after BTIs For 18 participants who had a BTI with Delta or Omicron variants, we had access to longitudinal samples (pre-and post-BTI). The characteristics of each of these participants are presented in Table S2. Pre-BTI samples were obtained approximately 1 month after their last Ad26.COV2.S vaccination (median: 0.98 months, interquartile range [IQR]: 0.7-3.8), and post-BTI samples were (B) Comparison of the polyfunctional profile of spike-specific CD4 + T cells after two exposures, namely two vaccinations (dark gray circles), infection followed by a single vaccination (blue circles), or single vaccination followed by infection (purple circles). The median proportion and IQR are shown. Each response pattern (i.e., any possible combination of IFN-g, IL-2, or TNF-a expression) is color coded, and data are summarized in the pie charts. Statistical comparisons were performed using a permutation test for the pie charts and a Wilcoxon unpaired t test for each response pattern. The number of participants included in each graph is indicated on top of the pie charts. taken a median of 4.2 months (IQR: 1.7-6) later. The exact date at which the BTI occurred was known for 4 participants who recorded a SARS-CoV-2-positive PCR test, while the remainder (n = 14) were characterized by nucleocapsid seroconversion or >2-fold increase in nucleocapsid IgG OD compared with the previous sample. Overall, in individuals experiencing a BTI, the median frequency of CD4 + or CD8 + T cells to ancestral spike was comparable preand post-BTI ( Figure 5A), while spikespecific IgG increased significantly after BTI (p = 0.009; Figure 5B). These data are consistent with the results obtained from the cross-sectional cohort (Figures 1B and 1D). The evolution of T cell and IgG responses pre-and post-BTI for each individual participant is presented in Figure S4 according to their exposure history.
Next, we assessed the cross-reactive potential of spike-specific CD4 + and CD8 + T cells in samples collected after a Delta or Omicron BTI in an expanded set of participants with post-BTI samples available (n = 10 classified as Delta BTI and n = 17 as Omicron BA.1 BTI). CD4 + T cell frequencies to the BTI variant spike were significantly lower compared with ancestral spike (p = 0.035 for Delta and 0.012 for Omicron; Figure 5C), resulting in a median decrease in CD4 responses of 16% toward the BTI variant as demonstrated by fold change ( Figure 5D). These data confirm the ability of CD4 + T cells to effectively cross-recognize SARS-CoV-2 variants 16-18 even in highly mutated variants such as Omicron. For the CD8 response, the frequency of CD8 + T cells recognizing spike from the BTI variant was reduced by 50% or more in two-thirds of the participants (3/6 for Delta BTI and 8/12 for Omicron BTI) compared with the ancestral spike CD8 response ( Figure 5D). This suggests that in some individuals who experienced a BTI, Delta or Omicron mutations may escape from specific HLArestricted T cell responses induced by prior infection and vaccination.

DISCUSSION
Understanding the effects of repeated antigen exposures, through infection and/or vaccination, on the development of SARS-CoV-2 memory T cell and antibody responses is essential for determining susceptibility to subsequent infections and informing booster vaccination strategies. We compared SARS-CoV-2-specific T cell and IgG responses in individuals who experienced between one and four SARS-CoV-2 antigen exposures. When measuring spike-specific IgG, we found that the proportion of responders and magnitude of spike-specific IgG progressively increased upon repeated exposures. In contrast, the impact on spike-specific T cell responses appeared more modest, with robust responses at stable frequencies over repeated exposures, concomitant with an increased number of individuals mounting a T cell response. (D) Fold change in the frequency of CD4 + and CD8 + T cells between ancestral and Delta spike responses (teal circles) and ancestral and Omicron responses (orange circles). Bars represent median fold change of responders. Delta BTIs are depicted by teal circles, Omicron BTI by an orange circle, and unknown variants (Delta or Omicron) by gray circles. The number of participants included in each graph is indicated, and the median value is indicated at the top of the graph. No significant differences were observed between CD4 or CD8 fold change using a Wilcoxon unpaired t test. For (A)-(C), a two-tailed Wilcoxon signed-rank test was used to assess statistical differences between paired samples. See also Table S2 and Figure S4.
Our results are in accordance with several publications showing that hybrid immunity and mRNA vaccination boosters are beneficial for the antibody response, enhancing its magnitude and broadening neutralizing potency even against highly divergent SARS-CoV-2 variants. [5][6][7][8][9]19 However, humoral responses, generated upon natural infection or vaccination, can substantially and rapidly wane, [20][21][22] emphasizing the importance of vaccination boosters. 23 Less is known about the impact of repeated antigen exposures on T cell immunity toward SARS-CoV-2. First, primary SARS-CoV-2 infection or vaccination induces robust T cell immunity, with CD4 T cell responses being more prevalent than CD8 responses. 1,12,24,25 Moreover, unlike antibodies, T cell responses are retained up to 6 to 12 months post-infection. [26][27][28] In this study, we show that while a third exposure significantly increased the proportion of individuals exhibiting a T cell response to SARS-CoV-2 spike, the magnitude of those responses was not affected by repeated exposures. Since antigen-specific T cells were measured in the memory phase, it is possible that vaccinationor infection-induced T cell responses had reached an immunologic plateau with limited evolution even upon repeated stimulation. However, quantity may be contrasted with qualitative changes that occurred; we observed that increased exposures affected the polyfunctional profile of CD4 + T cells, resulting in a progressive increase in the proportion of IFN-g + IL-2 + cells upon repeated exposures. This may arise through autocrine production of IL-2, which is thought to support memory T cell development by providing survival signals. 29 Others have also described Cell Reports Medicine 4, 100898, January 17, 2023 7 Article ll OPEN ACCESS phenotypic characteristics changing after repeated exposures. 30 Together, these observations demonstrate dynamic changes to SARS-CoV-2-specific T cells upon repeated exposures underlying stable response frequencies. However, while we did not observe any significant change in the overall frequency of peripheral T cells against spike upon repeated exposures, testing full-length spike could have obscured the generation of T cell specificities targeting additional spike epitopes, while other T cell specificities contracted. This could be done by assessing the T cell response against distinct sections of the spike protein, selectively covering the N-terminal domain, S1 chain, S2 chain, or the receptor-binding domain. 31,32 Importantly, our study was limited to the analysis of spike-and nucleocapsid-specific T cell responses in the blood. There is now clear evidence that infection also stimulates mucosal immunity. 10 SARS-CoV-2-specific resident T cells have been described in the nose and lungs of convalescent COVID-19 patients, and such responses were absent in individuals who had only been vaccinated. 33,34 Moreover, a detailed comparison of the specificity of SARS-CoV-2 T cell responses between the nasal mucosa and peripheral blood showed that circulating SARS-CoV-2-specific T cell responses were dominated by spike-specific cells, while in the nasal mucosa, responses against nucleocapsid or NSP-12 were more prevalent. 33 These data clearly indicate that, unlike vaccination, infection induces a broad T cell response targeting multiple SARS-CoV-2 proteins and promotes the establishment of SARS-CoV-2-specific tissue-resident T cells. It remains to be established whether vaccination after an infection has an impact on spike-specific T cell responses at sites of infection.
Although T cell responses generated upon vaccination or prior infection are highly cross-reactive with SARS-CoV-2 variants, [16][17][18][35][36][37] we did not observe significant boosting of the T cell response after BTI. This suggests that BTIs may not lead to the generation of de novo T cells targeting mutated spike epitopes in variants. Indeed, when we measured cross-reactivity to Delta or Omicron spike after BTIs with these variants, CD4 responses were well preserved, but CD8 responses were less cross-reactive, as previously described for vaccinees and convalescent patients. 16,18,36,37 We found that participants retained only 50% CD8 cross-reactivity to Omicron spike after BTIs during the Omicron surge. This may reflect antigenic imprinting, where the secondary response focuses epitope recognition on conserved spike epitopes. There is evidence for imprinting of the response for both T cell and B cell SARS-CoV-2 responses. 38,39 For example, initial Beta and Omicron BA.4 infection led to greater cross-reactivity to multiple variants compared with Delta and Omicron BA.1 infections. 40 However, our study was not designed to comprehensively assess the cross-reactive potential of SARS-CoV-2-specific T cells or identify the potential emergence of de novo SARS-CoV-2 T cell responses. To do so, it will be necessary to (1) measure the cross-reactivity of SARS-CoV-2 T cell responses to all SARS-CoV-2 variants of concerns both pre-and post-BTI and (2) probe T cell responses specifically targeting mutated epitopes from the breakthrough SARS-CoV-2 variant.
Whether exposure to specific variants shapes the T cell response to a similar extent remains to be fully defined for CD8 + T cell responses. So, too, it remains to be determined whether exposure history can be ''rewritten'' and expanded through heterologous vaccination 32 or the use of adjuvants. It should be noted that BTI readily induces new non-spike T cell responses, 30 which would supplement the cache of T cells available to engage and protect against severe disease upon viral re-encounter.
While we describe that the overall T cell response is stable (plateauing early after exposure through vaccination or infection), there is considerable inter-individual heterogeneity spanning 1.5 log in T cell frequencies to spike. Moreover, while all participants mounted spike-specific CD4 + T cell responses after four antigen exposures, approximately 20%-30% of participants remained refractory to the induction of spike-specific CD8 + T cell responses. Multiple studies demonstrate this response deficit for CD8 + T cell responses during vaccination and infection. 41,42 Specific class I and II alleles have emerged as associating with increased or decreased spike or nucleocapsid-specific T cell responses. 43,44 Which alleles are linked to CD8 hypo-responsiveness remains to be determined. There is evidence that the CD8 response is more narrowly directed than the considerable breadth of CD4 epitopes that are targeted (on average, 26-29 CD4 epitopes in spike 45 ) and that the CD8 response is more affected by mutations in variants. 46 These observations demonstrate the need to investigate SARS-CoV-2-specific CD8 + T cell responses in greater detail.
Overall, our study describes the quantity, quality, and dynamics of SARS-CoV-2-specific T cells in an increasingly complex immunological landscape. Multiple exposures result in limited immunological gains with respect to the magnitude of already robust virus-specific T cells in the circulation, but how boosting shapes T cell breadth, durability, and tissue homing remains an important question for further study. Differential exposure to distinct variants in different populations with diverse immunogenetics may affect future recognition of variants, and monitoring of vaccine efficacy and population immunity must continue as the pandemic rolls on.
Limitations of the study Our study had several limitations that should be considered. We did not measure neutralizing antibody responses. However, a host of studies have demonstrated that repeated exposures, including BTIs, improve neutralization titers and breadth against SARS-CoV-2 variants. [5][6][7][8][9]19 While we show that the magnitude of the T cell response reaches a plateau after repeated SARS-CoV-2 exposures, it remains to be determined whether the breadth of the response is expanded by multiple exposures, which could be performed by epitope mapping studies and T cell receptor (TCR) clonotyping. We did not address durability of the T cell response since we measured responses only approximately 1 month after the third or fourth SARS-CoV-2 antigen exposure. It is plausible that multiple exposures may improve long term T cell memory 28 and/or antigen-specific tissue resident T cells. In the context of hybrid immunity (particularly BTIs), SARS-CoV-2-specific T cells may migrate to and persist in the airways, 10 leading to an underestimation of total T cell frequencies induced upon multiple exposures.