Emergence of immune escape at dominant SARS-CoV-2 killer T cell epitope

Abstract

We studied the prevalent cytotoxic CD8 T cell response mounted against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) Spike glycoprotein_269-277 epitope (sequence YLQPRTFLL) via the most frequent human leukocyte antigen (HLA) class I worldwide, HLA A^∗02. The Spike P272L mutation that has arisen in at least 112 different SARS-CoV-2 lineages to date, including in lineages classified as "variants of concern," was not recognized by the large CD8 T cell response seen across cohorts of HLA A^∗02⁺ convalescent patients and individuals vaccinated against SARS-CoV-2, despite these responses comprising of over 175 different individual T cell receptors. Viral escape at prevalent T cell epitopes restricted by high frequency HLAs may be particularly problematic when vaccine immunity is focused on a single protein such as SARS-CoV-2 Spike, providing a strong argument for inclusion of multiple viral proteins in next generation vaccines and highlighting the need for monitoring T cell escape in new SARS-CoV-2 variants.

Keywords: CD8 T cell; COVID-19; SARS-CoV-2; T cell; T cell receptors; immune escape; peptide-HLA; phylogenetic.

Graphical abstract

Figure 1

SARS-CoV-2 variation in the immunodominant YLQPRTFLL dominant HLA A^∗02-restricted CD8 T cell epitope (A) Cumulative frequency of all sequences; sequences in the B.1.177 lineage; and sequences possessing the P272L variant across the United Kingdom (left) and worldwide (right). See Figure S1 for the total number of P272L variants by nation, binned into periods of 16 epidemiological (EPI) weeks is shown. (B) Top 15 most frequently observed variants observed in worldwide Spike glycoprotein sequence data in lineage B.1.177 as a percentage of total sequences. Data for all lineages are shown in Table S1. (C) Phylogenetic tree showing 1,227 taxa with variants at Spike position 272, with colors indicating subtrees representing potential independent mutations, computed by ASR on a larger tree of 200,221 taxa (not shown). (D) Percentage of sequences possessing P272L variant in England and Wales per administrative region up to and including January 1^st, 2022. See also Figures S1 and S2.

Figure 2

SARS-CoV-2 viral dynamics of P272L Spike mutation in Western Europe Percentage of sequences containing the P272L variant by nation, binned into periods of 16 epidemiological weeks, from week 11 (beginning Sunday, March 8^th, 2020), the first time a sequence with P272L was observed in sequence data; up to and including November 28^th, 2021 (week 101). See Figure S1 for worldwide data. See also Figures S1 and S2.

Figure S1

Worldwide SARS-CoV-2 viral dynamics of P272L spike mutation, related to Figures 1 and 2 (A) Percentage of P272L variants by nation, binned into periods of 16 epidemiological weeks, from week 11 (beginning Sunday 8^th March 2020), the first time a sequence with P272L was observed in sequence data; up to and including November 28^th 2021 (week 101). (B) Top 15 most frequently observed variants observed in worldwide spike glycoprotein sequence data in all lineages up to January 31^st 2021. Those thought to be associated with enhanced viral transmission are shown in green and those listed as potentially escaping from antibody mediated-mediated immunity are shown in light pink (possible) and dark pink (high confidence) as assigned at http://sars2.cvr.gla.ac.uk/cog-uk/#shiny-tab-immunology. This colour-coding of substitutions is used throughout the figure. Variants of unknown function in grey. N501Y also escapes from antibodies (pink asterisk). (C and D) Mapping of mutants shown in A onto Spike prefusion structure (PDB 6VXX; (Walls et al., 2020)) and sequence map.

Figure S2

Transmission of P272L variant in B.1.177, B.1.1.7/Alpha and B.1.617.2/Delta variants, related to Figures 1 and 2 (A) Fraction of days per month where five or more P272L variants were sequenced in B.1.177 background. (B) Localised emergence of P272L in the B.1.1.7/Alpha lineage prior to dominance of B.1.617.2/Delta lineage. (C) Cumulative sequenced cases during 2021 in Sydney Australia of P272L mutant in the B.1.617.2/Delta lineage.

Figure 3

YLQPRTFLL T cell lines from SARS-CoV-2 positive donors do not activate with P272L variant peptide (A) Ex vivo PBMC Wuhan-YLQPRTFLL and P272L-YLQLRTFLL tetramer staining of a convalescent patient. HLA A^∗02:01-SLYNTVATL peptide from HIV used as an irrelevant tetramer (Cole et al., 2017). Percentage of CD8 T cells is shown. Ex vivo staining for three other convalescent donors is shown in Figure S3A. (B) Wuhan-YLQPRTFLL T cell lines enriched from nine CPs do not activate towards the P272L-YLQLRTFLL peptide (10⁻⁶ M used for both peptides). Percentage of reactive cells is displayed from single or duplicate conditions. M012 and F021 analysis performed on a different day. See Figure S3B for a graphical summary of the data. (C) TCR beta chain analysis of Wuhan-YLQPRTFLL tetramer sorted T cells from all patients in A (flow cytometry data in Figure S3E). Pie charts display the proportion (chart segments) and frequency (numbers in center) of public (blue) and private (grey) TCRs. Variable (V) (arc on the right) and joining (J) (arc on the left) gene rearrangements are shown below the pie charts, with the dominant clonotypes annotated. CDR3s sequences are listed in Figure S4A. T cell clones used in this study are indicated with asterisk and data shown in Figure 4. See also Figures S3 and S4.

Figure S3

YLQPRTFLL specific T cells from SARS-CoV-2 convalescent patients do not stain or activate with P272L-YLQLRTFLL, yet the variant epitope binds HLA A^∗02:01 similarly to the Wuhan peptide, related to Figures 3, 4, 5 and 6 (A) PBMCs from three pre-vaccine convalescent patients were stained with HLA A^∗02:01-Wuhan and P272L tetramers. Preproinsulin (PPI) (ALWGPDPAAA) irrelevant tetramer. (B) PBMCs from convalescent patients were enriched with Wuhan tetramers to create T cell lines, which were used in a 4 h T107a assay with 10⁻⁶ M Wuhan or P272L peptides. Error bars depict SD of duplicate condititons. Some patient lines performed as a single condition. (C) HLA A^∗02:01 staining from a T2 cell binding assay using Wuhan and P272L exogenous peptides. HLA B^∗35:08 binding HPVGEADYFEY was used as an irrelevant control. Error bars depict SD of duplicate conditions. (D) Refolded Wuhan and P272L HLA A^∗02:01 monomers shown by SDS-PAGE gel used to assemble tetramers for staining T cells. (E) Wuhan tetramer staining of patient T cell lines used for flow cytometry sorting and TCR sequencing. HIV (Gag, SLYNTVATL) tetramers used as a negative control.

Figure S5

TCRs and peptide reactivity of Wuhan-YLQPRTFLL tetramer enriched T cell lines from six vaccinated donors, related to Figures 5 and 6 (A) TCR usage and CDR3 sequences of Wuhan-YLQPRTFLL specific T cells from vaccinated donors. TCRs that appear in multiple individuals highlighted as public, according to the key. (B) CD107a upregulation of the T cell lines with Wuhan and P272L-YLQLRTFLL peptides. Peptides (10⁻⁶ M) were pulsed on to T2 antigen-presenting cells prior to assay. CD3/CD28 Dynabeads used as a positive control. Performed in duplicate. Gating of positive cells shown for vaccinee 201a.

Figure 4

YLQPRTFLL-specific T cell clones from SARS-CoV-2 convalescent patients are unable to recognize P272L YLQPRTFLL peptide reactive CD8 clones were grown from convalescent patients indicated in the brackets. α and β TCR chain usage and CDR3s of each clone are shown and the public or private status of each indicated according to the key. Upper: peptide sensitivity assay with Wuhan-YLQPRTFLL and P272L-YLQLRTFLL peptides. MIP-1β ELISA and error bars depict SD of duplicates. Middle: peptide-HLA tetramer staining. HLA A^∗02:01 SLYNTVATL (SLY) peptide from HIV used as an irrelevant tetramer (Cole et al., 2017). Lower: A549 cells expressing HLA A^∗02:01 (A549-A2) and full-length P272L Spike were not recognized, whereas endogenously expressed Wuhan Spike was recognized. MIP-1β ELISA and error bars depict SD of duplicates. See also Figure S4B for HLA A2 and Spike staining of A549 cells. See also Figures S3 and S4.

Figure S4

TCR sequencing of Wuhan-YLQPRTFLL tetramer sorted T cells from convalescent patients and Spike transduced A549 cells used to test the reactivity of CD8 T cell clones, related to Figures 3 and 4 (A) TCR sequences of the Wuhan-YLQPRTFLL specific T cells from COVID-19 patients. M012 and F021 analysis performed on a different day. CD8 T cell clones grown from some of the donors annotated according to the key. (B) HLA A^∗02:01 (A2) and spike protein expression of the Wuhan and P272L spike transduced A549-A2 cells. HLA A^∗02 staining with BB7.2 Ab clone. Unconjugated isotype and anti-spike antibodies used in conjunction with a PE conjugated secondary antibody on Y axis.

Figure 5

YLQPRTFLL-specific T cells from vaccinees fail to stain with P272L variant tetramers Ex vivo tetramer staining of PBMC from 7 vaccinees with HLA A^∗02:01-YLQPRTFLL and P272L tetramers. Vaccinee gender is indicated next to each donor code. Vaccinees received the AstraZeneca (AZ) ChAdOx1 nCov-19 vaccine or the BNT162b2 (BNT) vaccine as indicated. The number of days post vaccine when blood was taken is indicated for each vaccinee, with data for 0439 shown after the second dose of the vaccine. Tetramer enriched T cell lines were successfully created from 6 of the 7 vaccinees and used for further analysis (Figure 6). See also Figures S3 and S5.

Figure 6

YLQPRTFLL-specific T cell clonotypes from vaccinated donors do not activate with P272L-YLQLRTFLL peptide (A) Wuhan-YLQPRTFLL tetramer enriched T cell lines from 6 of the 7 vaccinees in Figure 5. HLA A^∗02:01-restricted SLYNTVATL peptide from HIV used as an irrelevant tetramer (Cole et al., 2017). Percentage of CD3⁺ cells is shown. (B) The lines were re-sorted with Wuhan-YLQPRTFLL tetramer for TCR analysis. Pie charts display the proportion (chart segments) and frequency (numbers in center) of public (blue) and private (grey) TCRs. Public or private status based on data from convalescent patient studies. Variable (V) (arc on the right) and joining (J) (arc on the left) gene rearrangements are annotated with dominant and public chains with CDR3s of interest. CDR3 sequences are listed in Figure S5A. (C) CD107a upregulation of the T cell lines with Wuhan-YLQPRTFLL or P272L-YLQLRTFLL peptides. Peptides (10⁻⁶ M) were pulsed on to T2 antigen-presenting cells prior to assay. CD3/CD28 Dynabeads used as a positive control. Error bars depict SD of duplicates. Flow cytometry data shown in Figure S5B. See also Figures S3 and S5.

Figure 7

3D structure of antigens and YLQ36 TCR HLA A^∗02:01-YLQPRTFLL complex (A) Comparison of HLA A^∗02:01-YLQPRTFLL (blue sticks) and HLA A^∗02:01-YLQLRTFLL (red sticks). HLA A^∗02:01 shown as grey cartoon. Intrapeptide bonds present in HLA A^∗02:01-YLQPRTFLL are shown as blue dashes. (B) Comparison of unbound HLA A^∗02:01-YLQPRTFLL (grey sticks) and TCR-bound HLA A^∗02:01-YLQPRTFLL (blue sticks) peptide presentation. HLA A^∗02:01 shown as grey cartoon. Intrapeptide bonds present in unbound HLA A^∗02:01-YLQPRTFLL and TCR-bound HLA A^∗02:01-YLQPRTFLL are shown as black and blue dashes, respectively. See Figure S6 for the sequence of the YLQ36 TCR. (C) Heat map of YLQ36 TCR contacts with the YLQPRTFLL peptide. (D) YLQPRTFLL peptide residue Arg5 shown as blue sticks. Important YLQ36 TCR residues are labeled. Black dotted lines indicate van der Waals interactions. Red dotted lines indicate hydrogen bonds. Yellow dotted lines indicate salt bridges. (E) YLQPRTFLL peptide residue Thr6 shown as blue sticks. Important YLQ36 TCR residues are labeled. Black dotted lines indicate van der Waals interactions. Red dotted lines indicate hydrogen bonds. (F) YLQPRTFLL and YLQLRTFLL P4 residues shown as blue and red sticks, respectively. Important HLA A^∗02:01 residues shown as grey sticks. YLQ36 CDR3α loop shown as cyan sticks. Interactions involving the YLQ36 CDR3α loop are shown as black dashes. Also see Figure S7. .

Figure S6

Sequence of the HLA A^∗02:01 restricted Wuhan-YLQPRTFLL specific YLQ36 TCR, related to Figure 7 and STAR methods (A) Native nucleotide sequence of TCRα and TCRβ genes showing V(D)J assignment. (B) Bacterially expressed protein sequences manufactured in E.coli and refolded to make soluble YLQ36 TCR for biophysical and structural studies. Sequences include non-native cysteine residues as indicated in bold underlined red text to form non-native disulphide bonding between the TCR α and β constant domains. Two other substitutions in the TCRβ chain that aid refolding are indicated in bold text. The Cys to Ala substitution was included to remove the possibility of incorrect disulphide bind formation.

Figure S7

Structural analysis of pHLA and TCRs bound to HLA A^∗02:01-YLQPRTFLL, related to Figure 7 (A) Omit maps obtained after solving each pHLA structure with PHASER using a model not including the peptide. The density is displayed in light blue from the observed map, green for the positive difference map, and red for the negative difference map. (B) Overview of the YLQ36:HLA A^∗02:01-YLQPRTFLL 3D structure. (C) TCR footprints: top-down view of YLQ36, YLQ7/YLQ-SG3 (YLQ7 used, PDB:7N1F) and NR1C (PDB:7N6E) TCRs bound to HLA A^∗02:01-YLQPRTFLL. TCR CDR loops shown as colored cartoon, with the peptide shown as blue sticks. Pie charts beneath each footprint indicate the percentage contribution of each TCR chain, individual CDR loop and framework (FW) residue to the overall contacts within the complexes.