key: cord-0723358-cfkh0ypy
authors: Ogbe, A.; Kronsteiner, B.; Skelly, D. T.; Pace, M.; Brown, A.; Adland, E.; Adair, K.; Akhter, H. D.; Ali, M.; Ali, S.-E.; Angyal, A.; Ansari, M. A.; Arancibia-Carcamo, C. V.; Brown, H.; Chinnakannan, S.; Conlon, C. P.; de Lara, C.; de Silva, T.; Dold, C.; Dong, T. D.; Donnison, T.; Eyre, D. W.; Flaxman, A.; Fletcher, H. A.; Gardner, J.; Grist, J. T.; Hackstein, C.-P.; Jaruthamsophon, K.; Jeffrey, K.; Lambe, T.; Lee, L.; Li, W.; Lim, N.; Matthews, P. C.; Mentzer, A. J.; Moore, S. C.; Naisbitt, D. J.; Ogese, M.; Ogg, G.; Openshaw, P.; Pirmohamed, M.; Pollard, A. J.; Ramamurthy, N.; Rongkard, P.
title: T cell assays differentiate clinical and subclinical SARS-CoV-2 infections from cross-reactive antiviral responses
date: 2020-09-29
journal: nan
DOI: 10.1101/2020.09.28.20202929
sha: 1d991bd861661afedfde0c3ed8dcc6aa8ef9a8fd
doc_id: 723358
cord_uid: cfkh0ypy

A major issue in identification of protective T cell responses against SARS-CoV-2 lies in distinguishing people infected with SARS-CoV-2 from those with cross-reactive immunity generated by exposure to other coronaviruses. We characterised SARS-CoV-2 T cell immune responses in 168 PCR-confirmed SARS-CoV-2 infected subjects and 118 seronegative subjects without known SARS-CoV-2 exposure using a range of T cell assays that differentially capture immune cell function. Strong ex vivo ELISpot and proliferation responses to multiple antigens (including M, NP and ORF3) were found in those who had been infected by SARS-CoV-2 but were rare in pre-pandemic and unexposed seronegative subjects. However, seronegative doctors with high occupational exposure and recent COVID-19 compatible illness showed patterns of T cell responses characteristic of infection, indicating that these readouts are highly sensitive. By contrast, over 90% of convalescent or unexposed people showed proliferation and cellular lactate responses to spike subunits S1/S2, indicating pre-existing cross-reactive T cell populations. The detection of T cell responses to SARS-CoV-2 is therefore critically dependent on the choice of assay and antigen. Memory responses to specific non-spike proteins provides a method to distinguish recent infection from pre-existing immunity in exposed populations.

In late 2019 the new virus severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) emerged, causing the range of clinical diseases known as COVID-19 1, 2 . While the majority of SARS-CoV-2 infections are asymptomatic or result in mild disease, some individuals develop severe respiratory symptoms which may result in hospital admission and death leading to high global mortality 3, 4 , especially older adults and those with comorbidities 5 . Understanding the immune responses resulting from exposure to SARS-CoV-2, and distinguishing these from the responses made to seasonal coronaviruses, is a pre-requisite to defining immune correlates of infection and protection against subsequent SARS-CoV-2 disease. This in turn is centrally important in comparing with protective vaccine-induced immunity and may contribute to future public health policies including shielding advice.

Antibody responses to SARS-CoV-2 are important but remain complex. In a recent large-scale study of healthcare workers, PCR-confirmed SARS-CoV-2 infection resulted in measurable antibodies after 20 days in nearly all participants, with high specificity 6 . However, there is wide variability. Other studies have reported that antibodies may be absent early in disease, levels of neutralising antibodies are highly variable 7 , and antibody titres wane over time 8 . In contrast, studies of SARS-CoV infection indicate that T-cell responses may be more durable 9 . A number of studies have demonstrated the presence of T cell responses to the virus during acute disease and in recovery. Using in silico predicted HLA-class I and II peptide pools, CD4+ T cell responses to SARS-CoV-2 were demonstrated in all subjects who had recovered from COVID- 19 and CD8+ responses were demonstrated in 70% 10 . This study also found T cell reactivity to SARS-CoV-2 epitopes in 50% of archived samples from pre-pandemic (2015-2018) subjects using a 24-hour activation induced markers (AIM) assay. Additionally, a Swedish study demonstrated a highly activated cytotoxic phenotype in acute disease and vigorous polyfunctional T cell responses in convalescent subjects 11 . Interestingly, the latter study reported T cell responses to SARS-CoV-2 in seronegative household contacts, which may represent either seronegative infection or pre-existing cross-reactive immune memory to seasonal coronaviruses.

The role of prior exposure to human seasonal coronaviruses including alpha coronaviruses (HCoV-NL63 and HCoV-229E), and beta coronaviruses (HCoV-HKU1 and HCoV-OC43), that may generate SARS-CoV-2 cross-reactive T cell immune responses, is of substantial interest. Whilst prior exposure to the original SARS-CoV and to MERS-CoV is rare and restricted to outbreaks, exposure to the seasonal human coronaviruses is widespread. Population sero-surveys have shown that detectable baseline levels of IgG against at least one of the four known HCoV is near universal 12, 13, 14 , but there is evidence that re-infection with the same virus can occur 15, 16 . T cell immunity to other coronaviruses is less well studied prior to the 2020 pandemic, but a recent study from Singapore demonstrated the presence of reactive responses to SARS-CoV-2 in people who had recovered from the SARS-CoV epidemic 17 years earlier, which are likely to represent cross-reactive memory 9 . Such cross-reactive . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. . https://doi.org/10.1101/2020.09. 28.20202929 doi: medRxiv preprint responses to HCoV may be protective against SARS-CoV-2, be irrelevant, or could in theory contribute to immunopathology. The role of pre-existing cross-reactive T cell responses in immunity has been studied for other viruses including flaviviruses. In one study where such responses were fine-mapped, we observed that pre-existing cross-reactive responses to dengue virus were linked to disease protection from Japanese Encephalitis, while symptomatic disease was linked to the emergence of strain-specific T cells 17 .

Divergent data regarding SARS-CoV-2 T cell cross-reactivity have emerged so far: recent studies of T cell immunity to SARS-CoV-2 have reported levels of cross-reactive immunity to HCoV in SARS-CoV-2 unexposed populations of between 0-50% 9, 10, 11, 18, 19, 20, 21 using a variety of immune assays. One such study from our centre 20 did not find significant ex vivo IFN-γ ELISpot responses to SARS-CoV-2 in uninfected, seronegative subjects. The differences between these results might reflect the use of different assays employing a range of antigenic targets, peptide concentrations and proliferation times.

Here we set out to address two questions using a panel of T cell assays. First, do COVID-19 patients and seronegative controls show different levels of responsiveness in distinct assays of T cell function?

Second, can T-cell responses distinguish persons previously infected by SARS-CoV-2 from those previously infected by seasonal coronaviruses? We find -in a large cohort of subjects with a range of viral exposures -that cross-reactive memory responses to spike protein are almost universally detected using more sensitive assays, but that increasing viral exposure leads to an increase in magnitude and breadth of both effector and memory responses. These data have implications for our understanding of T cell cross-protection and for future studies of memory following the pandemic.

We first examined the T cell response to SARS-CoV-2 in fresh PBMC using an ex vivo IFN-γ ELISpot assay from 168 subjects with PCR-confirmed SARS-CoV-2 infection, and 111 negative controls without evidence of SARS-CoV-2 infection (Supplementary Table S1 ). IgG antibody responses to spike measured by ELISA are shown in Figure 1a and neutralising antibodies measured by a pseudoparticle assay are shown in Supplementary Figure 1a . Firstly, we evaluated the magnitude of the T cell response to SARS-CoV-2 to assess the effector T cell response following stimulation of peripheral blood mononuclear cells (PBMCs) with pools of overlapping peptides spanning all SARS-CoV-2 proteins except the non-structural ORF1 (Figure 1b and Supplementary Table S2 ). We found responses to summed pools covering SARS-CoV-2 spike protein (12 mini-pools of 15-mers overlapping by 10 peptides referred to as P1 -P12) (Figures 1b, 1c) , and the structural and accessory proteins ( is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. . https://doi.org/10.1101/2020.09.28.20202929 doi: medRxiv preprint 1b, 1d). Median ELISpot responses to spike peptides were lower than those induced by immunisation with the candidate ChAdOx1 22 , although the vaccine trial ELISpot assay used a higher concentration of peptides. We also screened PBMCs with pools containing predicted optimal peptides targeting MHC Class II epitopes on the SARS-CoV-2 spike protein (CD4S), and other viral proteins (CD4All), and predicted Class I binding peptides split into CD8A and CD8B described in Grifoni et al, 2020 13 (Figure   1b, Supplementary Figure 1c) .

IFN-γ responses to spike (S) pools were seen in PBMC from 34/75 (45%) of convalescent subjects tested (Figure 1c) with notably high and frequent responses to the P2 (up to 313 SFC/10 6 PBMC) and P8 minipools (up to 353 SFC/10 6 PBMC). We identified IFN-γ responses to the structural and accessory proteins in 65/103 (63%) of convalescent subjects, with especially high-magnitude responses to the membrane (M) and nucleocapsid (NP) proteins (Figure 1d) . Combined, there was variation in the breadth and magnitude of SARS-CoV-2-specific responses (Supplementary figure 1b) , with an apparent peak of responses 28 -32 days post onset of symptoms before declining, but longitudinal studies underway will better define the time course. IFN-γ responses were also seen in 24/29 (83%) of convalescent subjects following stimulation with the four pools of predicted epitopes. Interestingly, we found especially high-frequency responses to the CD8A pool which comprises predicted epitopes predominantly from the large ORF1 13, highlighting the need for further exploration of immune responses to this region (Supplementary figure 1c) .

There was a correlation between summed responses to spike and non-spike structural proteins (Spearman R = 0.579, P < 0.0001, Supplementary figure 1d), as well as the structural and accessory proteins and the predicted pools (data not shown), indicating that when an individual mounted a T cell response to one part of the proteome they were likely to respond to another part, and responses declined with time from symptoms (Supplementary figure 1e and 1f) . IFN-γ responses to either M or NP were correlates of the global response to spike, structural and accessory proteins (Figure 2a and   2b ), indicating that an assay to measure responses to M or NP could reflect the global effector T cell response.

We found a significant difference (P=0.0013) in the magnitude of the IFN-γ response measured by ELISpot assay to spike and to the structural and accessory proteins depending on the presence or absence of symptoms (Figure 2c and Supplementary Table S3) . It was not possible to explore the difference for the predicted optimal peptide pools due to insufficient numbers of asymptomatic subjects tested. Comparison between symptomatic and asymptomatic subjects may be confounded by differences in time from the start of the SARS-CoV-2 infection, which could be any time from 2 days to 6 weeks or more for the asymptomatic subjects 25 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. . https://doi.org/10.1101/2020.09.28.20202929 doi: medRxiv preprint and 1f, time from infection impacts on the IFN-γ response. We therefore next examined whether a selfreport of fever during COVID-19 in the symptomatic subjects was associated with a more vigorous T cell response by ex vivo IFN-γ ELISpot assay (P=0.0289, Figure 2d and Supplementary Table S4) , and observed a greater magnitude of the IFN-γ response to spike in HCWs who reported fever compared to those who did not. There was no relationship for anti-spike IgG or neutralising antibody levels for the presence or absence of either symptoms or fever (data not shown), although another UK study has reported lower antibodies in asymptomatic and mild cases compared to more severe disease 8 . These results suggest that SARS-CoV-2 infections with a higher symptom burden such as fever induce a higher magnitude of T cell immunity than milder or asymptomatic infections. We did not find a significant difference between IFN-γ ELISpot response and either age or sex ( Supplementary   Figures 1g and 1h) , but larger studies including older adults are needed for further exploration.

As our ELISpots assays were performed on total PBMCs, discrimination between distinct T cell lineages inducing the response was not possible. Moreover, the sensitivity of the ELISpot did not allow detection of responses in all COVID-19 recovered subjects. We therefore used a sensitive and functional flow cytometer-based assay capable of distinguishing the CD4+and CD8+ T cell responses. For this, we used a T cell proliferation assay to gain further insights into the contribution and relative proficiency of the CD4+ or CD8+ T cell compartments to drive a proliferative anti-SARS-CoV-2 immune response in our convalescent HCW cohort. We first validated our assays on a small cohort of healthy control subjects recruited for a hepatitis C virus (HCV) vaccine clinical trial pre-COVID19 23 . We showed that HCV seronegative control subjects made strong proliferative responses to pools of optimal peptides covering Influenza, EBV, CMV and Tetanus (FEC-T) but as expected, not to peptides covering HCV NS3 or core proteins (Supplementary Figure S2a -S2e) . We then evaluated the ability of CD4+ and CD8+ T cells from the COVID-19 convalescent HCW cohort to proliferate in response to peptide pools spanning key proteins from SARS-CoV-2. Live lymphocytes were separated into CD4+ or CD8+ T cells by gating strategy and the frequency of proliferating cells analysed following a 7-day stimulation (Supplementary Figure S2a) . We found a high frequency of proliferating cells and broad targeting of SARS-CoV-2-specific CD4+ and CD8+ T cells (Figure 3a and 3b) (Figure 3c and   3d ). This represents higher sensitivity to detect antigen-specific T cell responses than in the ex vivo ELISpot assay. Although we observed a trend for the overall magnitude of the proliferating CD4+ T cell response to SARS-CoV-2 peptide pools to be higher than that of the CD8+ T cell driven response, this did not reach significance for the peptide pools tested with the exception of M (P=0.0012) Figure S3a) . Also of note, we did not find any difference in the magnitude of responding CD4+ or CD8+ T cells in individuals who had asymptomatic disease (detected on HCW is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. We then performed ICS experiments using M, NP, S1, and S2 pools on an additional 26 SARS-CoV-2 PCR positive individuals to compare the immune responses among these peptide pools (Supplementary Figure S5) . M, S1, and S2 pools all trended towards higher levels of IL-2 expression by CD4+ T cells compared to CD8+ T cells (Figure S5a Figure S5) . The reduced polyfunctionality seen in this patient cohort is likely due to the lack of enrichment of ELISpot positive individuals, although potential loss in polyfunctionality with time may be a possible contribution as these donors were farther from symptom onset as a group. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. . https://doi.org/10.1101/2020.09.28.20202929 doi: medRxiv preprint Seronegative control subjects show strong CD4+ and CD8+ T cell memory responses to the S1 and S2 subunits of the SARS-CoV-2 spike protein We studied SARS-CoV-2 seronegative controls (Figure 1a ) for whom we also evaluated T cell responses to SARS-CoV-2 peptides using IFN-γ ELISpot, ICS and proliferation assay. In contrast to convalescent HCWs, SARS-CoV-2-specific IFN-γ responses were scarcely seen in any of the SARS-CoV-2 peptide pools as measured by ex vivo ELISpot assays in 23 seronegative healthy control subjects ( Figure 1b ) and ICS (data not shown). Responsiveness to common antigens (CEF-T) in these control subjects indicated that there were no inherent defects in the ability of PBMCs from these donors to mount an antigen driven immune response. This finding of a lack of response to SARS-CoV-2 peptides in seronegative control subjects by an 18-hour ex vivo IFN-γ ELISpot assay was confirmed in 13 subjects by an independent laboratory in Sheffield, UK (Figure 5a) . We also evaluated cryopreserved PBMC from pre-pandemic healthy subject archives, and found minimal responses to spike, structural and accessory proteins in 12 subjects in Oxford (Figure 5b ) and in the predicted epitope pools 10 in 48 subjects in Liverpool, UK (Figure 5c ).

However, using cellular proliferation assays on 20 seronegative subjects, we show high frequency of proliferating CD4+ and CD8+ T cells responding to the S1 and S2 subunit of the spike protein with a CD4+ T cell response detected in 17/20 (85%) and a CD8+ T cell response in 10/20 (50%) (Figure 6a and 6b). In contrast, we observed weak or no CD4+ and CD8+ T cell proliferative responses to the structural and accessory proteins studied (M, NP, ORF3, ORF6, ORF7 and ORF8 (Figure 6a and 6b) .

As the 20 seronegative participants were sampled in early 2020, we also analysed cryopreserved samples from 2008-2019 (pre-UK COVID19 pandemic) to exclude the possibility of asymptomatic and undetected prior infection. Similar to the pandemic seronegative controls, we found no or low effector T cell responses by ELISpot assay to any of the spike, structural or accessory proteins (Figure 5b ), but as for the pandemic seronegative controls we detected robust T cell responses by proliferation assay to spike proteins S1 and S2 which was of greater breadth in the CD4+ T cells compared to their CD8+ T cell counterparts (Figure 6c and 6d) . The responses show a CD4+ skew with 15/15 showing a CD4+ T cell response and only 8/15 showing a CD8+ T cell response above background level. Most importantly, there was very limited cross-reactivity to the structural and accessory proteins as measured by the proliferation assay. As with the convalescent HCW cohort, we also performed a cellular lactate assay using supernatants obtained after 4 days of stimulation on 8 of these subjects. We confirm crossreactive responses to spike S1 and S2 subunits, and non-existent or minimal responses in supernatants obtained from M, NP and accessory protein-stimulated PBMCs (Figure 6e) . We compared the magnitude of the proliferative responses to the different SARS-COV-2 peptide pools in seronegative controls from 2020, symptomatic and asymptomatic SARS-COV-2 PCR+ subjects (Figure 6f and g) .

We found no difference in the spike -S1 and S2 -responses but higher magnitude of proliferative responses to M and NP in both CD4+ and CD8+ T cells and ORF3 and ORF8 in CD8+ T cells alone in subjects who had tested positive to SARS-COV-2 (Figure 6f and g) . For confirmation, we also compared the magnitude of proliferative T cell responses in SARS-COV-2 seronegative controls from 2020 with the cryopreserved pre-pandemic seronegative controls and found the magnitude of is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. . https://doi.org/10.1101/2020.09.28.20202929 doi: medRxiv preprint proliferative cells in these two seronegative groups to be similar (Supplementary Figure S6a and S6b). These results, in addition to our earlier results from subjects who did not generate effector T cell responses to spike peptides in the IFN-γ ELISpot assay (Figure 1b and Figures 5a -5c) , demonstrate the existence of central memory T cell immunity to spike protein in the pre-existing T cell repertoire of subjects naïve to SARS-CoV-2.

Application of T-cell assays reveal responses in seronegative highly exposed healthcare workers Finally, to explore the use of these T cells assays to identify people potentially exposed to SARS CoV-2, we recruited a group of 10 highly exposed doctors working in acute medicine who had experienced symptoms compatible with COVID-19 but had not received PCR testing at the time of symptoms, or tested negative, and were subsequently seronegative. 3/10 of these subjects showed effector T cell responses by ELISpot assay to S1, S2, M or NP ( Figure 7a ) whilst 8/8 of those tested showed M and / or NP-specific T cell responses in the proliferation assay compatible with prior infection (Figure 7b and 7c). For one donor where cells were available, a fresh ICS assay confirmed this ELISpot response (data not shown). Analysis of the breadth of SARS-CoV-2 antigen targeted by the responding CD4+ and CD8+ T cells shows that in the highly exposed doctors the CD4+ and CD8+ T cell response is directed to a broader number of structural (M and NP) and accessory (ORFs 3, 6, 7 and 8) SARS-CoV-2 peptide pools (Figure 7d and e). This reached statistical significance for both CD4+ and CD8+ T cells compared to seronegative control groups (Figure 7d and e) . Lastly, we compared the magnitude of the T cell response to SARS-CoV-2 structural and accessory proteins in the three groups -the highly exposed doctors, seronegative controls from 2020 and pre-pandemic seronegative controls (combined into one group). We found significantly higher magnitude of CD4+ but not CD8+ T cells proliferating in response to the M, N, ORF3, 6, 7 and 8 in the highly exposed doctors (Figure 7f and g).

As the global COVID-19 pandemic continues, it is important to define which immune responses are important for protection. In this study we have used distinct T cell assay platforms across the same individuals to identify the differences between T cell responses associated with recent SARS-CoV-2 infection and long-term cross-reactive memory T cell responses in unexposed populations. The effector T cell response as measured by our 18-hour ex vivo IFN-γ ELISpot assay showed a remarkable absence of SARS-CoV-2 specific responses in most of the healthy seronegative subjects. The ELISpot assay therefore represents potential as a specific assay for identification of recent infection with SARS-CoV-2, although the longevity of such responses requires further analysis in longitudinal studies. We already noted a significant inverse correlation with magnitude by ELISpot assay over time in the short follow up performed here, and ongoing work with the current convalescent HCW cohort will define the durability of these T cell responses induced by SARS-CoV-2 infection.

In contrast, the same healthy subjects showed responses to the S1 and S2 subunits of spike protein in a 7-day CTV proliferation assay, confirmed by analysis of lactate production. The most likely . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. . https://doi.org/10.1101/2020.09.28.20202929 doi: medRxiv preprint explanation for this is that people retained cross-reactive central memory responses to the spike protein of seasonal coronaviruses that circulate in the UK, although cross-reactivity from other human microorganisms is also possible, as has been described for HIV, influenza and Ebola epitopes in naïve subjects 24, 25 . These cross-reactive responses may have been underestimated in previous reports using more ex vivo type assays with limited sensitivity.

Individuals in convalescence from SARS-CoV-2 infection showed strong and broad effector CD4+ and CD8+ polyfunctional T cell responses to peptides spanning the SARS-CoV-2 genome as previously reported 20 . ELISpot responses to the M and NP proteins were especially frequent and high, and each correlated with the summed response to spike, structural and accessory peptides, indicating their suitability as antigens for screening individuals and populations for evidence of T cell immunity following exposure to SARS-CoV-2. Additionally, central memory responses to M and NP were frequent and strong in the proliferation assay for subjects in convalescence from SARS-CoV-2 but significantly less so in the seronegative control subjects, further supporting the use of these antigens as markers of T cell responsiveness more closely linked with SARS-CoV-2 exposure. Further mapping studies could identify peptides with the highest sensitivity and specificity for SARS-CoV-2 infection, with potential for use in defining T cell immunity at an individual and population level.

The existence of substantial T cell cross-reactivity to SARS-CoV-2 from prior HCoV exposure has been demonstrated in non-SARS-CoV-2 infected populations from a range of geographical locations 9, 10, 11, 18, 19, 21 . Here, we demonstrate use of the ELISpot assay to identify SARS-CoV-2 specific responses, and our finding of absent T cell responses in unexposed subjects was confirmed by similar results in our three independent laboratories (Universities of Oxford, Liverpool and Sheffield). T cell assays vary in their sensitivity, influenced by cell number, incubation time, antigen choice and concentration and markers of T cell activity measured. Our ELISpot assay does not detect the T cell responses in unexposed populations to spike and other SARS-CoV-2 proteins reported elsewhere. This may be due to the relatively low cell number used in our assay (200,000 per well) but most likely the focus on IFNγ release rather than detection of cell activation markers. In this data we see a greater magnitude of ELISpot responses in convalescent symptomatic subjects who reported fever during their illness, compared with symptomatic subjects who did not report fever. We also saw higher responses in symptomatic people compared with asymptomatic people. Febrile symptomatic disease represents a greater systemic response and such individuals appear to mount a more vigorous T cell response (akin to a higher vaccine dose). This could represent failure of early/innate immune control necessitating a larger adaptive response and that hypothesis is consistent with the correlation we saw between T cell ELISpot and ELISA antibody levels. This is seen in other settings -for example, higher antigen-specific CD4+ T cell responses in more severe cases of H1N1/09 influenza A 26 .

Most convalescent subjects in the study made antibodies, as detected by IgG ELISA and pseudoparticle neutralisation assay. Emerging literature suggests that SARS-CoV-2 IgG titres meeting the threshold for positivity may be relatively short-lived 8, 27 . The current study represents a cross-sectional "snapshot" is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. . https://doi.org/10.1101/2020.09.28.20202929 doi: medRxiv preprint in time of human T cell responses to SARS-CoV-2 after infection, and the data suggests a peak in magnitude of the effector T cell response around 4 weeks after the onset of symptoms. Ongoing followup studies of this cohort and surveillance 28 for re-infection aligned to the UK SIREN study 29 will allow further delineation of the time course of T cell responses in parallel with humoral responses, and the timing of any assay must be taken into account in defining its utility. While an association is seen between antibody and ELISpot in the PCR-positive cohort, a disjunct exists between the antibodies and memory responses, since strong spike responses can be seen in the PCR-negative / unexposed and pre-pandemic groups. We need to assess in future whether any relationship exists between the levels of these responses and levels of seroreactivity to HCoVs.

Our study of the large ORF1 was restricted to use of the in silico predicted pool CD8A 10 , where high magnitude of responses were seen. Further work will characterise the time course of T cell responses observed in this cohort, evaluate the ability of our assays to correctly distinguish individuals with confirmed SARS-CoV-2 infection from unexposed controls, and prospectively seek to identify the relationship between measurable T cell immunity to the SARS-CoV-2 proteome and subsequent primary or secondary infection with SARS-CoV-2.

Overall, we have shown that assessments of T cell immunity using different assays but with the same antigens give very different results. Our ELISpot measure of ex vivo IFN-g release is valuable in defining the potential role of T cell immunity in recently infected donors without cross-reactivity in unexposed subjects. In contrast, our proliferation assay allows dissecting out pre-existing vs SARS-CoV- is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. . https://doi.org/10.1101/2020.09.28.20202929 doi: medRxiv preprint 

Healthcare workers at Oxford University Hospitals NHS Foundation Trust who tested positive for SARS-CoV-2 following either presentation to the hospital's Occupational Health Department with symptoms or had a positive PCR test on the staff screening programme 28 were asked to indicate whether they were willing to be contacted by researchers. Individuals who agreed to be contacted received an email invitation to participate in the study. Subjects recruited from the staff screening programme were classified as asymptomatic if they did not report any symptoms of COVID-19 (including fever, shortness of breath, cough, loss of taste or smell, sore throat, coryza or diarrhoea), either prior to staff screening or in the seven days following testing positive. In total 126 symptomatic and 33 asymptomatic subjects were recruited for this study. In addition, 9 hospitalised patients with WHO severe or critical COVID-19 were studied. (iii) Highly exposed seronegative individuals (highly exposed) 10 acute medicine doctors, who worked in patient facing services during the pandemic and experienced symptoms compatible with COVID-19, but did not receive PCR testing at the time of symptoms or tested negative, and were anti-spike IgG negative two months after the pandemic peak, were recruited as highly exposed seronegative participants.

Peripheral blood mononuclear cells were isolated by density gradient centrifugation using Lymphoprep TM (p=1.077 g/ml, Stem Cell Technologies) as previously described 30 . Plasma was . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. . https://doi.org/10.1101/2020.09.28.20202929 doi: medRxiv preprint collected and spun at 2000g for 10 minutes to remove platelets before freezing at -80⁰C for later use. 

For functional assays, PBMC were stimulated with three groups of peptide pool for SARS-CoV-2: (1) Spike: 15-mers overlapping by 10 amino acid residues for spike (S), divided into 12 "minipools" P1-P12 (Proimmune) 22 , and grouped into pools S1 (P1-6) and S2 (P7-12) for some assays (2) is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. . https://doi.org/10.1101/2020.09.28.20202929 doi: medRxiv preprint Intracellular cytokine stimulation (ICS) assay PBMC resuspended in R10 were plated at 1x10 6 live cells/well into 96 well round bottom plates and stimulated with SARS-CoV-2 peptide pools (2ug/ml per peptide) as indicated in the figure legends.

Media containing DMSO (0.1%, Sigma) was used as negative control and PMA (0.05ug/mL) with ionomycin (0.5ug/mL, Sigma) as a positive control. CD107a BV421 (BD Biosciences) and Brefeldin A (MP Biomedicals) were added to cultures at a final concentration of 0.04ug/mL and 10ug/mL respectively and cells were incubated for 6 hours at 37⁰C, 5% CO2, 95% humidity. Plates were placed at 4⁰C overnight and subjected to flow cytometry staining as described below. In addition to the three cytokines, CD107a was examined as was CD154 in CD4+ T cells. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. . https://doi.org/10.1101/2020.09.28.20202929 doi: medRxiv preprint (PFA, Sigma) and stored at 4⁰C in the dark until data acquisition. Cells from ICS assays were fixed with fixation/permeabilization solution (BD Biosciences) for 20 min at 4⁰C, washed with permeabilization buffer (BD Biosciences) followed by incubation with fluorochrome-conjugated human-specific antibodies. After washing with permeabilization buffer, the samples were resuspended in 1xPBS and stored at 4⁰C in the dark until data acquisition. Data was acquired on an LSRII flow cytometer (BD Biosciences) and analysis was performed with FlowJo Version 10 (BD Biosciences). Specific gating strategies can be found in the Supporting information (MIFlowCyt File).

Supernatants from the proliferation assay were analysed using a previously published assay 33 . Briefly, colorimetric L-lactate assay kits (Abcam, Cambridge, UK) were used as per manufacturer's instructions.

A standard concentration curve was defined, and the lactate concentration in each day 4 supernatant from the proliferation assay was calculated using a 96-well plate reader.

The lactate proliferation index was calculated on a per-well basis using the following equation 1:

(1) Proliferation (%) = 100 x (TStim -mean(TDMSO) )/ TStim Where TStim is the concentration of lactate for a given well with either PHA or SARS-CoV-2 peptides, and mean(TDMSO) is the average background lactate production from negative control wells. A significant proliferative response to a given peptide was greater than 0, as determined by equation 2:

(2) Significance = mean(TStim) -3xstd(TStim)

Where mean(TStim) is the mean % proliferative response of a specific participant to a stimulus, and std(TStim) is the standard deviation of the participant to a given stimulus.

Total anti-SARS CoV-2 spike antibodies were determined using an indirect ELISA as described previously 22 , which is based on the Krammer assay 34 using a standard curve derived from a pool of SARS-COV-2 convalescent plasma samples on every plate. Standardised EUs were determined from a single dilution of each sample against the standard curve which was plotted using the 4-Parameter logistic model (Gen5 v3.09, BioTek). Each assay plate consisted of samples and controls plated in triplicate, with ten standard points in duplicate and four blank wells.

Frozen plasma samples were thawed, heat-inactivated at 56C for 30 minutes, and assayed for neutralisation of a lentivirus-based viral particle carrying a luciferase reporter and pseudotyped with full- is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. . https://doi.org/10.1101/2020.09.28.20202929 doi: medRxiv preprint

Statistical analysis was performed with IBM SPSS Statistics 25 and figures were made with GraphPad Prism 8. Chi-square was used to compare ratio difference between two groups. After testing for normality using Kolmogorov-Smirnov test, Independent-samples t test or Mann-Whitney U test was employed to compare variables between two groups, and Kruskal-Wallis-ANOVA with Dunn's multiple comparisons test was performed to compare variables between three or more groups with a nonparametric distribution. Correlation was performed via Spearman's rank correlation coefficient. For polyfunctionality analyses, data was prepared using PESTEL v2.0 for formatting and baseline subtraction, followed by export of data to SPICE v6.0 for analysis. Statistical significance was set at P<0.05 and all tests were 2-tailed. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. . https://doi.org/10.1101/2020.09.28.20202929 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. . https://doi.org/10.1101/2020.09.28.20202929 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. . https://doi.org/10.1101/2020.09.28.20202929 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. . https://doi.org/10.1101/2020.09.28.20202929 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. . is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. . https://doi.org/10.1101/2020.09.28.20202929 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. . https://doi.org/10.1101/2020.09.28.20202929 doi: medRxiv preprint are shown for CD4+ (h) and CD8+ T cells in (i). Polyfunctionality analysis was also performed with S2 pools and are shown for CD4+ (k) and CD8+ T cells (l). is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. . https://doi.org/10.1101/2020.09.28.20202929 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. . https://doi.org/10.1101/2020.09.28.20202929 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. . https://doi.org/10.1101/2020.09.28.20202929 doi: medRxiv preprint

Thompson, C., Grayson, N., Paton, R., Bolton, J., Lourenço, J., Penman, B., . . . Isaric C Investigators. Detection of neutralising antibodies to SARS coronavirus 2 to determine population exposure in Scottish blood donors between March and May 2020. medRxiv; 2020.

. CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 29, 2020. . https://doi.org/10.1101/2020.09.28.20202929 doi: medRxiv preprint 

Coronaviridae Study Group of the International Committee on Taxonomy of Viruses. The species Severe acute respiratory syndrome-related coronavirus: classifying 2019-nCoV and naming it SARS-CoV-2

Genomic characterisation and epidemiology of 2019 novel coronavirus: implications for virus origins and receptor binding

China Medical Treatment Expert Group for Covid. Clinical Characteristics of Coronavirus Disease 2019 in China

Characteristics of and Important Lessons From the Coronavirus Disease 2019 (COVID-19) Outbreak in China: Summary of a Report of 72 314 Cases From the Chinese Center for Disease Control and Prevention

Features of 20 133 UK patients in hospital with covid-19 using the ISARIC WHO Clinical Characterisation Protocol: prospective observational cohort study

Performance characteristics of five immunoassays for SARS-CoV-2: a head-to-head benchmark comparison. The Lancet Infectious Diseases

Evaluating the Association of Clinical Characteristics With Neutralizing Antibody Levels in Patients Who Have Recovered From Mild COVID-19 in

Doores, K. Longitudinal evaluation and decline of antibody responses in SARS-CoV-2 infection. medRxiv

Bertoletti, A. SARS-CoV-2-specific T cell immunity in cases of COVID-19 and SARS, and uninfected controls

Targets of T cell responses to SARS-CoV-2 coronavirus in humans with COVID-19 disease and unexposed individuals

Robust T cell immunity in convalescent individuals with asymptomatic or mild COVID-19

First infection by all four non-severe acute respiratory syndrome human coronaviruses takes place during childhood

Development of a Nucleocapsid-Based Human Coronavirus Immunoassay and Estimates of Individuals Exposed to Coronavirus in a

Prevalence of Antibodies to Four Human Coronaviruses Is Lower in Nasal Secretions than in Serum

Direct observation of repeated infections with endemic coronaviruses

Human Coronavirus NL63 Molecular Epidemiology and Evolutionary Patterns in Rural Coastal Kenya

Solomon, T. Human T cell responses to Japanese encephalitis virus in health and disease

Phenotype and kinetics of SARS-CoV-2-specific T cells in COVID-19 patients with acute respiratory distress syndrome

Thiel, A. SARS-CoV-2-reactive T cells in healthy donors and patients with COVID-19

Dong, T. Broad and strong memory CD4(+) and CD8(+) T cells induced by SARS-CoV-2 in UK convalescent individuals following COVID-19

Weiskopf, D. Selective and cross-reactive SARS-CoV-2 T cell epitopes in unexposed humans

Safety and immunogenicity of the ChAdOx1 nCoV-19 vaccine against SARS-CoV-2: a preliminary report of a phase 1/2, single-blind, randomised controlled trial

Barnes, E. A human vaccine strategy based on chimpanzee adenoviral and MVA vectors that primes, boosts, and sustains functional HCV-specific T cell memory

Proteome-wide analysis of HIV-specific naive and memory CD4(+) T cells in unexposed blood donors

Virus-Specific CD4<sup>+</sup> Memory-Phenotype T Cells Are Abundant in Unexposed Adults

High levels of virus-specific CD4+ T cells predict severe pandemic influenza A virus infection

Antibody testing for COVID-19: A report from the National COVID Scientific Advisory Panel

Differential occupational risks to healthcare workers from SARS-CoV-2 observed during a prospective observational study

The impact of detectable anti SARS-COV2 antibody on the incidence of COVID-19 in healthcare workers

Guidelines for the use of flow cytometry and cell sorting in immunological studies

MIFlowCyt: the minimum information about a Flow Cytometry Experiment

Extracellular Lactate: A Novel Measure of T Cell Proliferation

A serological assay to detect SARS-CoV-2 seroconversion in humans