key: cord-0771942-b3llxrz4
authors: Szeto, Christopher; Chatzileontiadou, Demetra S.M.; Nguyen, Andrea T.; Sloane, Hannah; Lobos, Christian A.; Jayasinghe, Dhilshan; Halim, Hanim; Smith, Corey; Riboldi-Tunnicliffe, Alan; Grant, Emma J.; Gras, Stephanie
title: The presentation of SARS-CoV-2 peptides by the common HLA-A*02:01 molecule
date: 2021-01-22
journal: iScience
DOI: 10.1016/j.isci.2021.102096
sha: 689e527f530733fac8506ebc1885c94ed9f941b0
doc_id: 771942
cord_uid: b3llxrz4

CD8+ T cells are crucial for anti-viral immunity, however, understanding T cell responses requires the identification of epitopes presented by Human Leukocyte Antigens (HLA). To date, few SARS-CoV-2-specific CD8+ T cell epitopes have been described. Internal viral proteins are typically more conserved than surface proteins and are often the target of CD8+ T cells. Therefore, we have characterised eight peptides derived from the internal SARS-CoV-2 Nucleocapsid protein predicted to bind HLA-A*02:01, the most common HLA molecule in the global population. We determined not all peptides could form a complex with HLA-A*02:01, and the six crystal structures determined revealed that some peptides adopted a mobile conformation. We therefore provide a molecular understanding of SARS-CoV-2 CD8+ T cell epitopes. Furthermore, we show that there is limited pre-existing CD8+ T cell response towards these epitopes in unexposed individuals. Together, these data show that SARS-CoV-2 Nucleocapsid might not contain potent epitopes restricted to HLA-A*02:01.

Following the emergence of SARS-CoV-2 out of Wuhan, China in late 2019, there have been over 92 million infections and over 1.9 million deaths worldwide . This novel and highly infectious coronavirus has significantly encumbered the majority of the world, resulting in massive economic 'loss' in all global economies. Worldwide, significant efforts are being made towards the development of a SARS-CoV-2 vaccine (Amanat and Krammer, 2020) .

Studies have shown that SARS-CoV-2 specific antibodies might not be long-lived (Vabret, 2020 , Roltgen et al., 2020 , and it is unclear if they could provide long-term protective immunity. It is well established that cytotoxic CD8+ T cells, play a vital role in the control and clearance of viral pathogens. This is particularly well characterised in other viral respiratory infections, such as Influenza (McMichael et al., 1983) where pre-existing CD8+ T cells can decrease disease severity and overall symptom scores (Wells et al., 1981 , Bender et al., 1992 . Additionally, CD8+ T cell immunity can be long lasting, with longitudinal studies showing that CD8+ T cell immunity can be detected 10-50 years following vaccination (Miller et al., 2008, Ahmed and Akondy, 2011) , and at least 13 years following natural influenza infection (van de Sandt et al., 2015) . Furthermore, CD8+ T cells can be crossreactive, recognising variants within viral epitopes (Grant et al., 2018 , Valkenburg et al., 2010 , meaning that they can occasionally recognise distinct viral strains (Kreijtz et al., 2008 , van de Sandt et al., 2014 . As such, there is a great interest in understanding the CD8+ T cell response towards SARS-CoV-2 to determine if a CD8+ T cell mediated vaccine can provide broad and long-lasting protection. Indeed, there is emerging evidence that CD8+ T cells can be detected following infection with SARS-CoV-2 (Rydyznski Moderbacher et al., , Peng et al., 2020 , but their role in protective immunity is yet to be fully understood.

A significant challenge to investigating epitope-specific CD8+ T cell responses towards novel viruses is the lack of defined epitopes. CD8+ T cells recognise small peptides, derived from self or pathogenic proteins that are bound by Human Leukocyte Antigen (HLA) molecules. HLA molecules are themselves highly polymorphic, where the majority of their diversity lies within their antigen binding cleft, ensuring the presentation of a wide range of peptides from diverse pathogens. HLA molecules are divided into HLA class I and II, which J o u r n a l P r e -p r o o f are recognised by CD8+ and CD4+ T cells, respectively. HLA class I molecules bind small peptides (8-10 residues) in their antigen binding cleft through a series of pockets termed A-F.

The second and last residues of the peptide (P2 and PΩ, respectively) bind within the B and F pockets, respectively. These residues are often referred to as anchor residues, due to their critical role in anchoring the peptide into the HLA cleft. Each HLA molecule binds peptides with characteristic motifs that comprise particular anchor residues (P2 and PΩ), that are well adapted to the chemical properties of the HLA B and F pockets, respectively. For example, in HLA-A*02:01 restricted peptides, a Leucine or Methionine is often preferred at P2, and a smaller hydrophobic residue such as Valine or Leucine being characteristic of the PΩ residue (Sette and Sidney, 1999) . To date, only a few SARS-CoV-2 epitopes have been identified, and this limited knowledge represents a major roadblock to studying T cell immunity towards this new virus. There are numerous technical challenges to overcome, including limited access to patient samples, limited sample volume (e.g. blood), the need for HLA typing, and confirming the HLA restriction for each peptide. An approach to identify potential epitopes to a novel virus is to use known epitopes from a closely related virus or the use of peptide prediction algorithms.

In this study, we selected eight SARS-CoV-2 peptides that were likely to bind to the HLA-A*02:01 molecule, an allele common to ~40% of the global population. These peptides are a mix of predicted SARS-CoV-2 peptides and previously described SARS peptides (Cheung et al., 2007 . We have focused on the nucleocapsid (N) protein, which is typically conserved between closely related viruses, due to its important function. This makes the N protein an ideal target for CD8+ T cells and therefore, a CD8+ T cell mediated vaccine.

We firstly determined the ability of each peptide to form a stable complex with the HLA-A*02:01 molecule, showing that one was unable to form a complex with HLA-A*02:01, and three form a poorly stable complex. We then determined the crystal structure of six HLA-A*02:01-SARS-CoV-2 complexes, providing the first description of CD8+ T cell SARS-CoV-2 epitopes at an atomic level. Interestingly, three of our selected peptides have since been shown to be immunogenic (N 219-227 (Isabel Schulien, 2020 ), N 222-230 (Andrew P Ferretti, 2020 , Isabel Schulien, 2020 ), and N 316-324 (Jennifer R Habel, 2020 , Isabel Schulien, 2020 ) in COVID-19 recovered individuals. As expected, these immunogenic peptides were able to form a stable complex with the HLA-A*02:01 molecule. Furthermore, we compared the sequences of our selected peptides to other circulating common cold coronaviruses and found that some of our peptides were highly conserved. As such, we assessed the functional CD8+ J o u r n a l P r e -p r o o f T cell response towards these peptides in healthy HLA-A*02:01 individuals with no known exposure to SARS-CoV-2. We found that, despite some conservation, there was limited preexisting CD8+ T cell immunity towards these peptides.

Overall, we demonstrate that not all selected peptides were able to form stable complexes with HLA-A*02:01, which was a consequence of unfavourable P2 and/or PΩ residues, and an important factor for immunogenicity. In addition, we saw limited preexisting CD8+ T cell response for these peptides in unexposed donors, while peptides that have subsequently been shown to be immunogenic in COVID-19 recovered patients were stable and adopted a canonical conformation in the cleft of HLA-A*02:01. Altogether, our data provides molecular insight into CD8+ T cell epitopes from SARS-CoV-2.

Viral nucleocapsid (N) proteins are typically highly conserved due to their important functions, making them ideal targets for vaccine design. The N protein of SARS-CoV-2 is particularly important for RNA packaging during the release of the virion. As internal proteins are more conserved than surface proteins, the N protein is an excellent target for the adaptive immune system, and particularly cytotoxic CD8+ T cells.

The SARS-CoV-2 N protein is composed of 419 residues and divided into two main domains, the N-terminal (NTD) and the C-terminal domains (CTD) that are connected via a Ser-Arg rich linker (LKR) and flanked by a N-arm and C-tail loops, similar to the N protein of SARS (Chang et al., 2014) (Figure 1 ). Alignment of the N proteins from SARS-CoV-2 with SARS and four other coronavirus strains responsible for the common cold (NL63, OC43, HKU1, and 229E (Gagneur et al., 2002) ) revealed that SARS-CoV-2 had a higher sequence identity with SARS (90%) than with the remaining four coronaviruses (23 -29% sequence identity). The NTD and LKR domains exhibited the highest sequence homology between the 6 viruses ( Figure 1) . We selected a total of eight N-derived peptides; six from the conserved NTD and LKR domains and two from the CTD section ( Table 1) . Out of the eight peptides predicted to bind the HLA-A*02:01 (Campbell et al., 2020) that are conserved in SARS (Cheung et al., 2007) , six of the peptides are known SARS CD8+ T cell epitopes (all except N 351-358 and N 351-359 ) . We then compared the sequence conservation within each of the peptides. These conserved SARS-CoV-2 peptides (exceptions J o u r n a l P r e -p r o o f of N 222-230 and N 226-234 ) had 22 -44 % sequence identity or homology with OC43, HKU1, or 229E virus, and none with NL63 virus (Table S1) .

We established if the selected peptide sequences are representative of circulating SARS-CoV-2 strains, and determined their conservation within N protein sequences derived from SARS-CoV-2 viruses circulating within Oceania (4049 sequences), Asia (1,152 sequences), Europe (390 sequences) and North America (10,044 sequences). All peptides were shown to be highly conserved, and our selected peptides are represented in 96-99% of all circulating strains (Table S2) , making them ideal for potential vaccine candidates. Interestingly, the immunogenicity of 5 of our selected peptides has since been characterised in COVID-19 recovered individuals. Two peptides, N 138-146 and N 159-167 , were not immunogenic in COVID-19 recovered individuals (Isabel Schulien, 2020) , while three peptides, namely N 219-227 (Isabel Schulien, 2020) , N 222-230 (Isabel Schulien, 2020) and N 316-324 (Isabel Schulien, 2020 , Jennifer R Habel, 2020 were immunogenic in COVID-19 recovered individuals, further highlighting the importance of further investigation into these epitopes.

Overall, the selected N-derived SARS-CoV-2 peptides are representative of the currently circulating strains of the virus, and share some similarity with coronaviruses responsible for the common cold.

To determine if the 8 N-derived peptides ( Table 1) were able to form a complex with HLA-A*02:01, we firstly refolded each peptide separately with the HLA-A*02:01 heavy chain and β2-microglobulin (β2m). Seven of our 8 peptides were successfully refolded with the HLA-A*02:01 molecule, however the N 351-358 peptide (p) failed to stabilise HLA-A*02:01, as no refolded protein was obtained following purification. This result is not completely surprising, as the N 351-358 peptide does not have the favoured PΩ-Val/Leu typical of HLA-A*02:01 binding (Sette and Sidney, 1999) , but instead has a charged Aspartic acid residue.

Interestingly, the overlapping N 351-359 peptide, which has a small hydrophobic Alanine residue at PΩ was able to form a complex with HLA-A*02:01 (details below).

The stability of the pHLA is an important factor as it influences the half-life of the complex, in turn impacting the likelihood of a peptide being present for long enough at the cell surface to then be recognised by T cells (Blaha et al., , Harndahl et al., 2012 . We therefore assessed the stability of the remaining 7 pHLA complexes, and compared them to the well characterised HLA-A*02:01-restricted immunogenic influenza M1 peptide (GILGFVFTL) J o u r n a l P r e -p r o o f (Valkenburg et al., 2016) using differential scanning fluorimetry (DSF). All 7 HLA-A*02:01-SARS-CoV-2 complexes had a lower thermal shift temperature (Tm) than the highly stable HLA-A*02:01-M1 which exhibited a Tm of ~60 °C consistent with previously published reports (Valkenburg et al., 2016) (Table 1, Figure S1 ). The most stable pHLA complexes were the one with the N 222-230 (Tm of 55 °C) and N 316-324 (Tm of 49 °C) peptides, followed by the one with N 219-227 , N 226-234 and N 351-359 peptides with a Tm of ~40 °C (Table 1, Figure   S1 ). Surprisingly, the Tm was ~35 °C for complexes with the NTD-derived peptides N [138] [139] [140] [141] [142] [143] [144] [145] [146] and N 159-167 , an extremely low Tm value for pHLA-A*02:01 complexes (Valkenburg et al., 2016 , Blaha et al., 2019 , Khan et al., 2000 .

The three peptides that resulted in a pHLA complex with a Tm above 40 °C, were recently described as immunogenic in COVID-19 recovered patients (N 219-227 (Isabel Schulien, 2020) , , Isabel Schulien, 2020 , and N 316-324 (Jennifer R Habel, , Isabel Schulien, 2020 ), while the two peptides leading to a pHLA with a Tm below 40 °C were described as non-immunogenic (N 138-146 and N 159-167 (Isabel Schulien, 2020) ) ( Table   1 ). This suggests, that indeed pHLA complex stability may play a role in peptide immunogenicity.

Together, our data shows that HLA-A*02:01 was poorly stable in complex with some of the predicted N-derived SARS-CoV-2 peptides, which will impact T cell response.

To gain a better understanding of the presentation of SARS-CoV-2 peptide by HLA-A*02:01, and why some of these complexes displayed unusually low Tm values, we solved the structures of 6 HLA-A*02:01-SARS-CoV-2 complexes. Unfortunately, the HLA-A*02:01-N 219-227 structure could not be determined due to low yields of protein required for crystallization. All structures were solved to a high resolution of 1.3 -2.15Å (Table 2) ,

showing unbiased electron density for the peptides ( Figure S2) . Overall, the pHLA structures were similar, with a root mean-square deviation (r.m.s.d.) on the antigen binding cleft of 0.14 -0.22Å between the 6 structures (Figure 2A (Table 1) .

Conversely, the N 138-146 peptide has additional anchor residues relative to the other structures, adopting a constrained conformation within the cleft (Theodossis et al., 2010) . The P2-Leu and P9-Ile act as conventional primary anchor residues, and three residues acting as secondary anchors, namely P3-Asn, P5-Pro and P7-Asp ( Figure 2D) . As a result, the backbone of the peptide adopts a zig-zag conformation, with residues pointing up and down spanning P3 to P9. The unusual P5-Pro secondary anchor is located within a pocket that is positively charged, which is not a favourable environment for a hydrophobic Proline residue.

This constrained conformation of the N 138-146 peptide and the presence of multiple secondary anchor residues (P5-Pro and P7-Asp) might explain the low Tm value of the pHLA complex ( Table 1) .

The N 226-234 peptide has a P9-Met, which is not anchored into the small hydrophobic F pocket in HLA-A*02:01 within this particular crystal structure. However, two self and one synthetic peptides have been crystallised with a P9-Met binding to the F pocket (Mohammed et al., 2008 , Hassan et al., 2015 , Riley et al., 2018 , therefore other residues within the N 226-234 peptide seems to destabilise the anchoring of the P9-Met. As a result, the Tm of HLA-A*02:01-N 226-234 complex was low (39.2 °C, Table 1 ) and despite solving the structure of the pHLA at high resolution (1.9Å, Table 2), the peptide was poorly resolved. We could only observe density for P1, P2 and P3 residues in the Fo-Fc map. The generation of composite omit maps (Afonine et al., 2012) did not reveal any additional residual density for the peptide. The only residual density observed was shown at the location of carboxylic moiety of the PΩ residue, however after refinement it was clear that a Methionine could not fit in this density. Instead, an acetate ion was placed mimicking the carboxylic group for PΩ ( Figure 2E) . While the SARS-CoV-2 N 226-234 peptide was only able to bind to HLA-A*02:01 by its N-terminal region ( Figure 2E ) due to an unfavourable PΩ-Met, the homologous peptide from SARS virus possesses a PΩ-Val (Table S1) , which is favoured within the F pocket, able to stabilise HLA-A*02:01 and stimulate CD8+ T cells (Tsao et al., 2006) .

The N 351-359 peptide overlaps the N 351-358 peptide ( Table 1) (Figure 2F ). In addition, the P5-His acts as a secondary anchor, but unfavourably binding into a positively charged C pocket (Arg97, His70). Altogether, these sub-optimal primary and secondary anchors give rise to a low Tm (42.9 °C, Table 1 ) and leads to poorly defined density around the central part of the peptide that indicates flexibility ( Figure S2J) .

The low stability of the overall pHLA complexes, and the unusual conformations of the peptides, may explain the lack of immunogenicity of N 138-146 and N 159-167 , due to a short half-life on the cell surface that would compromise T cell interactions.

The N 222-230 and N 316-324 peptides have been shown to be weakly immunogenic in a few HLA-A*02:01+ COVID-19 recovered patients (Jennifer R Habel, 2020 , Isabel Schulien, 2020 , Andrew P Ferretti, 2020 . They both share the preferred P2-Met/Leu and PΩ-Val/Leu characteristics of HLA-A*02:01 restricted peptides (Sette and Sidney, 1999) . As a result, both peptides were able to form a stable pHLA complex, which correlate to the high Tm values observed ( Table 1) .

The HLA-A*02:01-N 222-230 complex exhibited the highest Tm value (54.7 °C, Table 1) among the HLA-A*02:01-N-SARS-CoV-2 complexes tested here, and is similar to the HLA-A*02:01-M1 complex (~60 °C (Valkenburg et al., 2016) ). In addition, the N 222-230 peptide showed a well-defined electron density ( Figure S2F and L) , suggesting a rigid peptide conformation. The N 316-324 peptide adopts a rather flat conformation in the cleft, where P5-Ile and P7-Met are only partially buried between the peptide backbone and the HLA α2-helix ( Figure 2G) . The N 316-324 peptide P4-Arg and P8-Glu are exposed to the solvent, and represent potential contact points for CD8+ T cells. While the N 316-324 peptide only has two prominent residues exposing their side chains for potential contact with a TCR (Figure 2G) , the N 222-230 has four large solvent exposed residues (P4-Asp, P5-Arg, P7-Asn and P8-Gln, Figure 2H ). The P5, P7 and P8 side chains are flexible and allowed to sample their molecular surrounding of the pHLA complex ( Figure 2H) . The abundance of solventexposed residues in the N 222-230 peptide offers a variety of potential contact points for interaction with TCRs, and therefore recognition by CD8+ T cells.

Altogether, the stability and crystal structures of the N 316-324 and N 222-230 SARS-CoV-2 peptides in complex with HLA-A*02:01 show that favoured HLA primary anchor residues, promote well-defined and stable peptides in HLA, that in this case underlie immunogenicity.

Since some level of conservation was seen within these 8 peptides between SARS-CoV-2 and the coronaviruses that cause the common cold (OC43, HKU1, 229E, Table S1), we next asked whether HLA-A*02:01+ individuals have some pre-existing immunity towards the SARS-CoV-2 peptides. To test this, we stimulated peripheral blood mononuclear cells (PBMCs) derived from HLA-A*02:01+ individuals without any known infection or exposure to SARS-CoV-2, with a pool of the 8 peptides (n=5 donors) or the immunodominant influenza-derived peptide M1 as a control (n=3 donors). Functional responses indicative of pre-existing immunity were then assessed using an Intracellular Cytokine Staining (ICS) assay (Figure 3) . Limited CD8+ T cell responses were observed towards the pool of peptides in all 5 donors (Figure 3A and 3D) , contrasting to the robust CD8+ T cell responses towards the control influenza-derived M1 peptide (Figure 3B and 3C) consistent with previous reports (Valkenburg et al., 2016) . Responses were similarly minimal even when assayed against higher concentrations of the SARS-CoV-2 derived peptides individually ( Figure S3) . Therefore, even though there was up to 44% sequence identity for N 138-146 and N 159-167 , between SARS/SARS-CoV-2 and some common cold coronaviruses (Table S1), the remaining differences within the peptides might hinder the cross-reactive potential of CD8+ T cells.

Therefore, despite the conservation of the selected peptides with other commonly circulating coronaviruses (Table S1) , there is limited pre-existing CD8+ T cell response towards these SARS-CoV-2 derived peptides in HLA-A*02:01+ individuals.

Discovering immunogenic CD8+ T cell epitopes of the SARS-CoV-2 virus is undoubtedly important to help us understand the magnitude and strength of the immune response in COVID-19 patients. However, we currently have limited knowledge on SARS-CoV-2 epitopes and their HLA restriction. Epitope identification is a significant bottleneck when trying to characterise novel viruses. Multiple approaches can be utilised, such as mass-J o u r n a l P r e -p r o o f spectrometry (Koutsakos et al., 2019) or overlapping peptide screening (Grant et al., 2013) , with each method having its own pros and cons. Mass-spectrometry can be utilised to identify peptides presented by a selected HLA allele. Although this is beneficial, it is time consuming, expensive, and without additional screening, does not provide any information on the immunogenicity of the identified peptides. Conversely, overlapping peptide screening identifies only peptides that are immunogenic, and additional screening of HLA restriction is required. This method requires large sample sizes and is not efficient if trying to identify peptides for a particular HLA allele, such as HLA-A*02:01, the most prevalent HLA molecule in the global population.

An alternate option is to predict which peptide(s) can bind to a HLA molecule based on its preferred anchor residue for the HLA allele of interest (Sette and Sidney, 1999) . In this study, we have characterised 8 SARS-CoV-2 peptides predicted to bind to the highly prevalent HLA-A*02:01 molecule (Cheung et al., 2007) , 6 of which are known SARS epitopes. One was unable to form a complex with HLA-A*02:01, and 3 formed poorly stable complexes.

This highlights that, although efficient, peptide prediction algorithms are not always accurate.

Indeed, immunogenicity studies by recent groups (Isabel Schulien, 2020 , Andrew P Ferretti, 2020 , Jennifer R Habel, 2020 , along with our previous work (Grant et al., 2013) , have shown that predictive peptides are not always indicative of immunogenic epitopes. However, when used in combination with functional assays, this method can be a fast and ideal way to identify a range of peptides worthy of further characterisation particularly when faced with multiple technical challenges such as access to patient samples and limited sample volume (e.g. blood).

A particularly important factor to consider when looking at the T cell response towards viral peptides, is their ability to form a stable complex with the HLA, ensuring their presence on cell surfaces for long enough to interact with circulating T cells. We have demonstrated that 7 of our 8 peptides are able to form a complex with HLA-A*02:01, however the stability of the complexes ranged from 35 -54 °C. Interestingly, the immunogenicity of 5 of our selected peptides has since been investigated, and we noticed a trend between the ability of peptides to form a stable complex with HLA-A*02:01 and their ability to stimulate CD8+ T cells in some COVID-19 recovered individuals (Isabel Schulien, 2020 , Andrew P Ferretti, 2020 , Jennifer R Habel, 2020 . Indeed, the N 219-227 , N 222-230 and N 316-324 peptides have a Tm value above 40 °C and have been described as immunogenic in COVID-19 recovered patients (Isabel Schulien, 2020 , Andrew P Ferretti, 2020 , Jennifer R Habel, 2020 while N 138-146 and J o u r n a l P r e -p r o o f N 159-167 with a Tm values below 40 °C are described as not immunogenic (Isabel Schulien, 2020) . This correlation between the stability of the pHLA complex and immunogenicity is not surprising, and has been observed in previous research (Blaha et al., , Harndahl et al., 2012 , suggesting that Tm should be considered, and may even be predictive, when assessing the immunogenic potential of vaccine peptide candidates.

It is evident that additional research is required to determine which peptides from SARS-CoV-2 are able to activate CD8+ T cells and identify which HLA molecules they are restricted to. These findings will guide our understanding of the immune response, as well as elucidate the potential for a protective and long-lived immune response. We might uncover whether some HLA molecules are better equipped to bind particularly immunogenic viral epitopes capable of stimulating a potent T cell response that could be the target of future vaccines. Indeed, there are numerous studies that describe a strong link between the expression of certain HLA molecules and improved disease outcome (Altfeld et al., 2006 , van de Sandt et al., 2019 , Valkenburg et al., 2016 . For example, in the context of Human Immunodeficiency virus (HIV) infection, certain individuals are naturally able to control the virus and limit the progression to Acquired Immunodeficiency syndrome (AIDs), this ability has been linked with the expression of protective HLA alleles such as HLA-B*57:01 and HLA-B*27:05 (Altfeld et al., 2006) , while HLA-B*35:01 is detrimental, and allows rapid progression to AIDs (Altfeld et al., 2006) . Similarly, during influenza infection, HLA-A*02:01 seems to provide some protection (Valkenburg et al., 2016) , while HLA-A*68:01 is associated with poor clinical outcomes (van de Sandt et al., 2019) .

Our data provides some rationale for the weak or lack of immunogenicity observed in HLA-A*02:01+ individuals recovered from COVID-19 towards N-derived SARS-CoV-2 peptides, due to unconventional anchor residues and poor stability of pHLA complexes. It is unclear whether HLA-A*02:01 can present immunogenic peptides from SARS-CoV-2 that results in a strong CD8+ T cells response, or whether these peptides derived from N protein are not strongly immunogenic for CD8+ T cells. Further work is required to uncover the drivers of protective CD8+ T cells response to SARS-CoV-2 infection.

Our study has focused on a set of 8 peptides derived from the N protein of SARS-CoV-2 virus, for their conservation with SARS virus as well as their predicted binding to HLA-A*02:01. It is possible that additional N-derived peptides will be able to stably bind HLA-A*02:01, and therefore extend the epitopes repertoire from this SARS-CoV-2 protein. In addition, some additional work will be required to define if the N 226-234 and N 351-359 peptides are immunogenic in COVID-19 recovered patients, as well as fully characterise the CD8+ T cells that respond towards these peptides.

Further information and requests for resources and materials should be directed to the Lead Contact, Prof. Stephanie Gras (S.Gras@latrobe.edu.au)

Materials are available upon reasonable request. Table 1 . SARS-CoV-2 potential HLA-A*02:01-restricted CD8+ T cell epitopes.

Sequence Tm (°C) Immunogenicity N 138-146 ALNTPKDHI 35.7 ± 0.6 Null (0/11, (Isabel Schulien, 2020) ) N 159-167 LQLPQGTTL 35.8 ± 1.5 Null (0/11, (Isabel Schulien, 2020) ) N 219-227 LALLLLDRL 41.5 ± 0.5 Weak (2/11, (Isabel Schulien, 2020) ) N 222-230 LLLDRLNQL 54.7 ± 0.4 Weak (3/11, (Isabel Schulien, 2020) ) N 226-234 RLNQLESKM 39.2 ± 1.0 ND N 316-324 GMSRIGMEV 49.0 ± 0.1 Weak (2/11, (Isabel Schulien, 2020) ), (Jennifer R Habel, 2020) 7KGT 7KGP 7KGO a R p.i.m = Σ hkl [1/(N-1)] 1/2 Σ i | I hkl, i -<I hkl > | / Σ hkl <I hkl >. b R factor = Σ hkl | | F o | -| F c | | / Σ hkl | F o | for all data except ≈ 5% which were used for R free

Towards automated crystallographic structure refinement with phenix.refine

Insights into human CD8(+) T-cell memory using the yellow fever and smallpox vaccines

HLA Alleles Associated with Delayed Progression to AIDS Contribute Strongly to the Initial CD8(+) T Cell Response against HIV-1

SARS-CoV-2 Vaccines: Status Report. Immunity

COVID-19 Patients Form Memory CD8+ T Cells that Recognize a Small Set of Shared Immunodominant Epitopes in SARS-CoV-2

Transgenic mice lacking class I major histocompatibility complex-restricted T cells have delayed viral clearance and increased mortality after influenza virus challenge

Prediction of SARS-CoV-2 epitopes across 9360 HLA class I alleles

The SARS coronavirus nucleocapsid protein--forms and functions

Induction of Tcell response by a DNA vaccine encoding a novel HLA-A*0201 severe acute respiratory syndrome coronavirus epitope

An interactive web-based dashboard to track COVID-19 in real time

Coronavirus-related nosocomial viral respiratory infections in a neonatal and paediatric intensive care unit: a prospective study

Nucleoprotein of influenza A virus is a major target of immunodominant CD8+ T-cell responses

A Sequence Homology and Bioinformatic Approach Can Predict Candidate Targets for Immune Responses to SARS-CoV-2

Peptide-MHC class I stability is a better predictor than peptide affinity of CTL immunogenicity

Naturally processed non-canonical HLA-A*02:01 presented peptides

Ex vivo detection of SARS-CoV-2-specific CD8+ T cells: rapid induction, prolonged contraction, and formation of functional memory

Suboptimal SARS-CoV-2-specific CD8+ T-cell response associated with the prominent

The structure and stability of an HLA-A*0201/octameric tax peptide complex with an empty conserved peptide-N-terminal binding site

Human CD8(+) T cell cross-reactivity across influenza A, B and C viruses

Cross-recognition of avian H5N1 influenza virus by human cytotoxic T-lymphocyte populations directed to human influenza A virus

Cytotoxic Tcell immunity to influenza

Human effector and memory CD8+ T cell responses to smallpox and yellow fever vaccines

Phosphorylation-dependent interaction between antigenic peptides and MHC class I: a molecular basis for the presentation of transformed self

Structural features impose tight peptide binding specificity in the nonclassical MHC molecule HLA-E

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

T cell receptor cross-reactivity expanded by dramatic peptide-MHC adaptability

SARS-CoV-2 Antibody Responses Correlate with Resolution of RNAemia But Are Short-Lived in Patients with Mild Illness. medRxiv

Nine major HLA class I supertypes account for the vast preponderance of HLA-A and -B polymorphism

Constraints within major histocompatibility complex class I restricted peptides: presentation and consequences for T-cell recognition

HLA-A*0201 T-cell epitopes in severe acute respiratory syndrome (SARS) coronavirus nucleocapsid and spike proteins

Protective efficacy of cross-reactive CD8+ T cells recognising mutant viral epitopes depends on peptide-MHC-I structural interactions and T cell activation threshold

Molecular basis for universal HLA-A*0201-restricted CD8+ T-cell immunity against influenza viruses

Challenging immunodominance of influenza-specific CD8(+) T cell responses restricted by the risk-associated HLA-A*68:01 allomorph

Human cytotoxic T lymphocytes directed to seasonal influenza A viruses cross-react with the newly emerging H7N9 virus

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

We would like to thank all of our healthy volunteers for donating blood, the Australian Red Cross Lifeblood (ARCLB) for access to Buffy coats, and Dr David Sayer from CareDx for his collaboration in HLA typing some of our samples. We would like to thank the staff at the All authors have read and edited the manuscript.

The authors declare no competing interests.