Immunopeptidome profiling of human coronavirus OC43-infected cells identifies CD4 T cell epitopes specific to seasonal coronaviruses or cross-reactive with SARS-CoV-2

Seasonal “common-cold” human coronaviruses are widely spread throughout the world and are mainly associated with mild upper respiratory tract infections. The emergence of highly pathogenic coronaviruses MERS-CoV, SARS-CoV, and most recently SARS-CoV-2 has prompted increased attention to coronavirus biology and immunopathology, but identification and characterization of the T cell response to seasonal human coronaviruses remain largely uncharacterized. Here we report the repertoire of viral peptides that are naturally processed and presented upon infection of a model cell line with seasonal human coronavirus OC43. We identified MHC-I and MHC-II bound peptides derived from the viral spike, nucleocapsid, hemagglutinin-esterase, 3C-like proteinase, and envelope proteins. Only three MHC-I bound OC43-derived peptides were observed, possibly due to the potent MHC-I downregulation induced by OC43 infection. By contrast, 80 MHC-II bound peptides corresponding to 14 distinct OC43-derived epitopes were identified, including many at very high abundance within the overall MHC-II peptidome. These peptides elicited low-abundance recall T cell responses in most donors tested. In vitro assays confirmed that the peptides were recognized by CD4+ T cells and identified the presenting HLA alleles. T cell responses cross-reactive between OC43, SARS-CoV-2, and the other seasonal coronaviruses were confirmed in samples of peripheral blood and peptide-expanded T cell lines. Among the validated epitopes, S903–917 presented by DPA1*01:03/DPB1*04:01 and S1085–1099 presented by DRB1*15:01 shared substantial homology to other human coronaviruses, including SARS-CoV-2, and were targeted by cross-reactive CD4 T cells. N54–68 and HE128–142 presented by DRB1*15:01 and HE259–273 presented by DPA1*01:03/DPB1*04:01 are immunodominant epitopes with low coronavirus homology that are not cross-reactive with SARS-CoV-2. Overall, the set of naturally processed and presented OC43 epitopes comprise both OC43-specific and human coronavirus cross-reactive epitopes, which can be used to follow T cell cross-reactivity after infection or vaccination and could aid in the selection of epitopes for inclusion in pan-coronavirus vaccines.

). The levels of HLA-DQ were ~20-fold lower than HLA-DR and HLA-DP. Thus, we 152 restricted immunopeptidome analysis to HLA-ABC, HLA-DR, and HLA-DP. NetMHCpan [50,51], as shown in Figure S1. The characteristic C*07:02 motif [50,52] was not 192 observed in the clustering analysis. We observed 2 motifs for HLA-DR, representing 73 and 24%  Table S2d). Peptide P17 (Fig 2C), derived from the 214 spike protein, was assigned to HLA-A2 by motif analysis, with predicted binding in the top 0.5% 215 (Table S2d). Peptides P15 and P16 ( Fig 2C) were derived from the 3C-like proteinase of the ORF 216 1ab polyprotein and were assigned to HLA-B7 and HLA-A2 respectively, based on predicted 217 binding within the top 0.5% for these alleles, although weak binding of peptide P16 to HLA-C7 218 was also predicted (1.5%-tile) ( Table S2d). The low abundance of virus-derived peptides within 219 the overall MHC-I peptidome might be a result of MHC-I immune-evasion mechanisms, similar to 220 those reported for 55,56].

221
Eighty MHC-II-binding viral peptides were identified, derived from nucleoprotein, spike, 222 hemagglutinin esterase (HE), and envelope proteins ( Table S2d). Some of these were among the  The nested sets of peptides characteristic of MHC-II peptidomes are comprised of length variants 240 surrounding a 9-residue core epitope that includes the major sites of MHC-peptide interaction.

241
This is believed to result from variable trimming of MHC-bound peptides by endosomal proteases,

242
leaving different numbers of residues flanking the core regions. As expected, for each of the 243 nested sets of peptides, the predicted core epitope (underlined in Fig 2D) was found in the center 244 of the overlapping set. Core epitopes for the eluted peptides were among the top-ranked predicted 245 binders for each protein (Fig S3A-C), helping to explain why these particular peptides were 246 selected for presentation. For instance, the top-ranked predicted peptides for nucleoprotein, 247 spike, and envelope contain the binding core from the HLA-DP-eluted peptides P2, P5, and P14, 248 respectively ( Fig S3A). Similarly, the top-ranked predicted peptides for nucleoprotein, spike, and 249 HE contain the binding core from the HLA-DR-eluted peptides P10, P11, and P8, respectively

251
For HLA-DR, peptides were tentatively assigned to DR2a or DR2b by motif analysis. In some 252 cases, one allele was clearly preferred, with predicted binding in the top 5th percentile to DR2b 253 but not DR2a as for P8, P9, P10, and P11 peptides (

278
Positive responses were observed in most donors tested (6/9 for DP and 8/9 for DR). Responding

279
T cells were present at low frequencies, which varied considerably between donors (0.007-280 0.057% for DP; 0.001-0.011% for DR). Note that in this assay other HLA alleles are present in 281 the donors besides the HEK293 alleles used for the elution studies, but with very few exceptions 282 these alleles are the best predicted binders among the HLA-DR, HLA-DP, and HLA-DQ alleles 283 present in each donor (Table S5).

284
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted December 1, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 13 To increase the frequency of OC43-responding cells for detailed assessment of the responses to 285 individual peptides, we expanded peptide-specific T cells in vitro. Using the expanded T cell 286 populations, we measured IFN-γ production in response to re-stimulation with the same peptides, frequently observed (p=0.006) in DR15 donors (80-100%) than in DP4 donors (60-88%), while 295 responses were slightly stronger for DP peptides (3.7 ± 2.1x10 3 SFU/10 6 cells) than DR peptides 296 (2.2 ± 1.8x10 3 SFU/10 6 cells) when tested at 1 µg/mL peptide concentration, although this 297 difference is not significant (Fig 3D). There was a weak but significant correlation between the 298 eluted peptide abundance (sum of precursor ion intensities by nested set) and the observed T 299 cell response (r= 0.64, p= 0.009, Spearman). No correlation was observed between binding 300 (predicted or experimental) and T cell responses, nor between binding and peptide abundance.

301
To explore the overall sensitivity of the different peptide-expanded T cells, dose-response 302 experiments were performed, and the minimal activating peptide concentrations were determined 303 ( Fig 3E-F). In general, a wide range of minimal concentrations was observed. For instance, for 304 P2 and P3 the minimal concentrations were 10 -6 µg/mL and 10 -7 µg/mL, respectively for expanded 305 cells from donor 61, while for P11 (donor 07) and P9 (donor 40), the minimal concentration was 306 1 µg/mL. This indicates that T cells responding to P2 and P3 in donor 61 were more sensitive to 14 heterogeneous population that responds to different antigens with different efficiencies. In some 310 cases, different donors showed similar sensitivity to a particular peptide, as is the case of P10 in 311 donors 18, 22, and 40, which all responded at 10 -5 µg/mL. However, in other cases, there was 312 heterogeneity in the responses to a given peptide. For instance, for P4 the minimal concentration 313 varied between 0.1 and 10 -5 µg/mL in 4 donors. All these results may reflect the different history 314 of exposure to OC43 and other coronaviruses and the evolution of the responding T cell repertoire 315 in each individual, which translates to a lack of a clear hierarchy of functional avidity and 316 immunodominance for most of the eluted peptides.

317
To characterize the T cells producing these responses, we performed intracellular cytokine 318 staining (ICS) assays using the single-peptide-expanded T cell lines. As in the ELISpot assays, peptides, we were able to measure CD107a mobilization along with IFN-γ production (Fig S2B), 323 and production of low levels of TNF-a was observed for 1 peptide (Fig S2C). No IL-2 or IL-10 324 production was observed for any peptide (not shown). This suggests that the CD4 T cells 325 responding to the eluted OC43 peptides could be polyfunctional and have cytotoxic potential.

326
Altogether, these results present clear evidence of CD4+ T cells that recognize and respond to   CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted December 1, 2022. ; https://doi.org/10.1101/2022.12.01.518643 doi: bioRxiv preprint

333
The substantial sequence homology between OC43 and the other HCoVs ( Fig S4A) raises the 334 question of whether responding T cells could cross-react between the different orthologs.

335
Sequence alignments of the naturally processed OC43 peptides with homologous sequences 336 from other HCoVs are shown in Fig S4B, and a heatmap of conservation indices is shown in Fig   337   S4C. Overall, the highest conservation is between OC43-and HKU1-derived peptides, with less 338 for the other beta-coronaviruses MERS-CoV, SARS-CoV, and SARS-CoV-2, and even less for

344
To evaluate experimentally the potential for cross-reactivity we initially focused on OC43 and 345 SARS-CoV-2. We measured responses to the eluted OC43 peptides and their SARS-CoV-2 346 homologs, using T cell populations expanded with individual OC43 peptides from PBMC samples 347 banked pre-pandemic before the outbreak of SARS-CoV-2 into the human population. Peptides 348 with no homolog in SARS-CoV-2 (P1, P8, P9), or with no response in our donor pool (P8, P13, 349 P14) were excluded. We measured T cell responses in single-peptide-expanded T cell lines using 350 IFN-γ ELISpot assays, using partial-HLA-matched donors as before. Only the P4 and P11 SARS-

351
CoV-2 homologs induced cross-reactive T cell responses in the single-peptide expanded lines 352 ( Fig 4A). Across a larger set of donors, similar cross-reactive responses were observed, with 353 somewhat lower responses to the heterologous SARS-CoV-2 homologs than the OC43 peptides 354 used for expansion (average 2-fold, p=0.044 for P4 and average 3.5-fold, p=0.011 for P11; paired 355 t-test; Fig 4B). This indicates that a substantial proportion of T cells responding to the OC43-P4 356 and OC43-P11 peptides can cross-react with their SARS-CoV-2 homologs. To evaluate the 357 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted December 1, 2022. ; https://doi.org/10.1101/2022.12.01.518643 doi: bioRxiv preprint sensitivity of these T cell lines to cross-reactive stimulation, we measured the dose-response to 358 cognate and heterologous peptides. Robust cross-reactivity to heterologous stimulation was 359 observed across the dose-response range for both P4 and P11 homologs in all donors tested, 360 including pre-pandemic ( Fig 4C) and those with recent COVID-19 infection (Fig 4D), with minimal 361 stimulatory peptide concentrations in a wide range but similar for OC43 and SARS-CoV-2 362 homologs ( Fig 4E).

363
To explore factors that could have resulted in the observed pattern of OC43 and SARS-CoV-2 364 cross-reactive responses, we measured MHC binding of the SARS-CoV-2 homologs and 365 compared them to the OC43 peptides ( Fig 4F). We found weaker binding for most of the SARS-

366
CoV-2 homologs, with the exception of P4, for which DP4.1 binding was 10-fold greater for the 367 SARS-CoV-2 homolog. In addition to altering MHC binding affinity, amino acid substitutions can 368 cause shifting of the preferred binding register, which would interfere with T cell recognition of 369 homologous peptides. Of the nine peptides tested, only P3, P4, and P11 retain the predicted 370 binding register in the SARS-CoV-2 homologs (Fig 4G), and only for P4 and P11 are the predicted 371 T cell contacts completely or mostly conserved (shaded in Fig 4G).

372
We extended this analysis to the other seasonal human coronaviruses, using the T cell lines 373 expanded in vitro with P4 and P11 peptides from pre-pandemic and COVID-19 donors. The P4 374 and P11 homologs from the seasonal coronaviruses mostly retained binding to DP4.1 (for P4) 375 and DR2a/DR2b (for P11) (Fig 4H), and we measured the cross-reactive T response to these 376 peptides. In general, all the P4-and P11-expanded T cell lines recognized each of the homologs, 377 with the exception of P11 from 229E, which was recognized poorly by T cell lines expanded with 378 SARS-CoV-2 or OC-43 homologs (Fig 4I). CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted December 1, 2022. ; https://doi.org/10.1101/2022.12.01.518643 doi: bioRxiv preprint

381
The immune response to seasonal human coronaviruses is largely understudied and few T cell

392
although T responses to the two highly-cross-reactive epitopes P4 and P11 have been reported 393 previously in studies characterizing seasonal coronavirus cross-reactivity to identified SARS-

395
We identified only a few OC43-derived peptides presented by MHC-I molecules, and these were 396 present at very low abundance within the overall MHC-I peptidome. One peptide from the spike 397 protein and one from the 3C-like proteinase encoded by the ORF1ab polyprotein, both likely 398 presented by HLA-A2, and a second 3C-like proteinase peptide likely presented by HLA-B7, were

404
regulation has not been previously reported for OC43, but is a common feature of many viruses 405 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted December 1, 2022.

415
By contrast, eighty OC43-derived peptides presented by MHC-II molecules were found at high 416 abundance within the overall MHC-II peptidome. Indeed, three of the top four most intense ions 417 in the HLA-DR peptidome mass spectrum, and the third and fourth most intense ions in the HLA-418 DP peptidome mass spectrum, correspond to OC43-derived peptides. Most of the OC43-derived 419 MHC-II-bound peptides were from spike and nucleoprotein, the major coronavirus structural 420 proteins, consistent with the over-representation of these proteins we observed in the whole-cell 421 proteome of infected cells. Several peptides derived from the hemagglutinin-esterase protein, 422 which is believed to be required for cleavage of sialic acid residues to promote the release of 423 progeny virus from infected cells, similarly to hemagglutinin-esterase proteins from influenza C 424 and certain toroviruses and orthomyxoviruses [64]. Finally, one set of low-abundance peptides is 425 derived from the small envelope protein. All the OC43-derived MHC-II-bound peptides were found 426 as nested sets, except for three very low abundance peptides found as singletons. In each case, 427 the nested sets surrounded the predicted nine-residue core epitope, with 1-9 residue extensions, 428 consistent with endosomal protease trimming of MHC-bound peptides as expected in the MHC-II 429 antigen-presentation pathway. We selected one representative peptide from each nested set to 430 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted December 1, 2022. ; https://doi.org/10.1101/2022.12.01.518643 doi: bioRxiv preprint confirm binding to MHC-II, and to assign presenting MHC allotypes to the HLA-DR peptides, 431 which could derive from either DR2a (DRB5*01:01) or DR2b (DRB1*05:01), both of which are 432 expressed by HEK293 cells and co-purified with the LB3.1 antibody that we used for 433 immunoaffinity. Each of the eight representative HLA-DP eluted peptides bound to DP4.1, 434 although with varying affinity not entirely predicted by NetMHCIIpan4.1. Of the six representative 435 HLA-DR peptides, one (P12) bound exclusively to DR2a, four exclusively to DR2b, and one to 436 both allotypes (P9). As previously observed in another study of naturally processed MHC-II 437 peptides in virus-infected cells [34], the eluted peptides generally were among the top predicted 438 binders for each viral protein, one exception being P1 from the hemagglutinin-esterase protein.

439
We tested representative eluted peptides for recognition by T cells from HLA-matched donors.

440
Of fourteen peptides tested, we observed robust T cell responses to eleven. In other systems, 441 characterization of naturally-processed, MHC-bound peptides by mass spectrometry of infected 442 cells has proven to be an efficient route for T cell epitope discovery [31,32,[32][33][34][35][36][37][38][39]65,66]. We 443 observed a correlation between the observed T cell response and epitope abundance in the 444 overall immunopeptidome, whereas a significant correlation was not observed for the predicted 445 or even observed peptide binding affinity. Thus, characterization of naturally-processed peptides 446 from virus-infected cells can be a highly efficient epitope discovery approach, particularly 447 compared to screening comprehensive overlapping peptides libraries or large sets of predicted 448 MHC binders, where typically T cell responses are observed to only a small fraction of the 449 candidate epitopes. A similar trend relating T cell response to epitope abundance has been 450 observed in some [39] but not all [32,65,66] previous studies, although it should be noted that all 451 of these previous studies involved CD8 T cell responses. Three eluted peptides (P8, P14, and 452 P13) were not recognized by T cells from HLA-matched donors. These peptides were present at 453 relatively low abundance in the peptidomes, although in some cases (P6, P7, P9) peptides with 454 even lower abundance were recognized. We examined whether these peptides might not be 455 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted December 1, 2022. ; https://doi.org/10.1101/2022.12.01.518643 doi: bioRxiv preprint 20 immunogenic because of homology to self-peptides [67]. The peptides that were not recognized 456 had similar homology scores to the closest matching self-peptides as did peptides that were 457 recognized, although the number of exact matches in the core epitope region was somewhat 458 larger for peptides that were not recognized (mean 6.3 vs 4.6, p=0.016).

459
Among human and animal coronaviruses, the approach of characterizing naturally-processed 460 peptides presented by MHC proteins in infected cells to date has only been applied to SARS-

478
Several previous studies of the T cell response to SARS-CoV-2 in pre-pandemic donors have 479 identified T cell responses that are cross-reactive with homologous epitopes from seasonal 480 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted December 1, 2022. ; https://doi.org/10.1101/2022.12.01.518643 doi: bioRxiv preprint 21 coronaviruses including OC43 [16,21,26,27,59,[72][73][74][75]. However, there is still not a consensus on

503
For studies of the differential response to SARS-CoV-2 and seasonal coronaviruses, epitopes 504 specific to the seasonal coronaviruses are required. Among the OC43-eluted peptides for which 505 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted December 1, 2022. ; https://doi.org/10.1101/2022.12.01.518643 doi: bioRxiv preprint 22 cross-reactive T cell responses to SARS-CoV-2 homologs were not observed, P10 N54-68 elicited 506 recall responses in all donors tested. Responding CD4 T cells showed a high sensitivity, with 507 minimal peptide concentrations of about 10 pg/mL. This epitope is not strongly conserved among 508 the HCoVs (Suppl Fig S4) and may be a good candidate to study and follow OC43-specific

518
There are some limitations to this study. The HEK cells used for immunopeptidome 519 characterization were manipulated to ensure stable expression of MHC-II proteins by introducing 520 the CIITA gene, which may favor the processing and presentation in the MHC-II compartment. In 521 addition, these cells may not be representative of the natural targets of infection in the respiratory 522 tract. Also, we assumed that the pre-pandemic donors would have been exposed to OC43. We 523 did not consider T cell responses restricted by the mismatched MHC molecules. Finally, T cell 524 responses not associated with IFN-γ, not able to expand with peptide stimulation in vitro, or below 525 our detection level would have been missed by our approach.

526
In summary, we characterized the spectrum of naturally-processed viral peptides presented by 527 MHC molecules in HEK293.CIITA cells infected with the human seasonal coronavirus OC43.

528
MHC-II presented peptides dominated the OC43-derived viral immunopeptidome, possibly due to 529 the potent down-regulation of MHC-I molecules in infected cells. The spike protein is the major 530 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted December 1, 2022. ; https://doi.org/10.1101/2022.12.01.518643 doi: bioRxiv preprint 23 source of OC43-derived epitopes, with contributions from nucleoprotein and hemagglutinin-531 esterase. Most of the naturally-processed peptides are recognized by T cells from HLA-matched 532 donors. Three seasonal-coronavirus-specific CD4 T cell epitopes and two SARS-CoV-2-cross-533 reactive CD4 epitopes were identified. These epitopes provide a basis for studies of the cellular 534 immune response to OC43, and for evaluating the role of pre-existing seasonal coronavirus 535 immunity in SARS-CoV-2 infection and vaccination. 536 537 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

558
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

605
A linear gradient was developed from 5% solvent A to 35% solvent B in 60 min. Ions were 606 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted December 1, 2022. (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted December 1, 2022. ; https://doi.org/10.1101/2022.12.01.518643 doi: bioRxiv preprint 28 not match the HLA-DP binding motif and the single-ion spectrum was poor, so this sequence was 630 not considered further.

653
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted December 1, 2022. ; https://doi.org/10.1101/2022.12.01.518643 doi: bioRxiv preprint 29 Assignments were made using a 10-ppm mass tolerance for the precursor and 0.05 Da mass 654 tolerance for the fragments. All nonfiltered search results were processed by Scaffold (version 655 4.4.4, Proteome Software, Inc.) utilizing the Trans-Proteomic Pipeline (Institute for Systems 656 Biology) with a 1% false-discovery rate. The data was processed using MaxQuant as well which 657 uses Andromeda search engine and search parameters were kept the same as Mascot Server.

658
The search was performed against a concatenated target-decoy database with modified reversing

674
For each sample, we selected the cluster that included the largest number of peptides analyzed.

675
For HLA-DR and HLA-DP peptides, a preference for hydrophobic residue at P1 was used to align 676 the motifs at the P1 position. For HLA-ABC peptides, MHC-I ligands of length 8-13 residues 677 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made   tested. Secreted IFN-γ was detected following the manufacturer's protocol. Plates were analyzed 723 using the CTL ImmunoSpot Image Analyzer (ImmunoSpot, Cleveland, OH) and ImmunoSpot 7 724 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made   (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

748
Peptides for these studies were obtained from 21 st Century Biochemicals (Marlborough, MA) and 749 BEI Resources (Manassas, VA). Peptide sequences using in the assays are shown in Table S6.

752
Sequence logo of predicted motifs obtained using Motif Viewer in NetMHCpan or NetMHCIIpan.

753
The Immune Epitope Database IEDB [19] was used to search for T cell responses to seasonal 754 and pandemic coronavirus epitopes.

801
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted December 1, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022      sequence is shown (first and last residues indicated); the predicted core epitope in each sequence 844 is underlined. Nested sets of eluted peptides comprising length variants with the same core 845 epitope are shown by lines below the sequence. The peptide sequence highlighted in red was 846 used for biochemical and immunological assays (see Table 1). In C and D, each eluted sequence 847 or nested set was identified by "P" followed by a number. E. Label-free quantification of proteins    CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted December 1, 2022. ; https://doi.org/10.1101/2022.12.01.518643 doi: bioRxiv preprint 41 filled symbols and negative responses as empty symbols. In B and E, statistical analysis was 906 done by unpaired t-test. * p<0.05). 907 908 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted December 1, 2022.   (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted December 1, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 50 100% identity to OC43 peptide as 1, and no conservation as 0. An average per peptide is shown 1218 at the bottom of the heatmap. NA indicates no homolog protein between OC43 and the 1219 corresponding virus. 1220 Table S1: Cellular proteomics analysis on the OC43 infected and uninfected HEK293.CIITA cells.

1240
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted December 1, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 51

1243
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted December 1, 2022. ;

WT hla levels proteomics
Average intensity HLA-DR and HLA-DP eluted viral peptides. A schematic representation of each source protein with the location of each eluted sequence is shown (first and last residues indicated); the predicted core epitope in each sequence is underlined. Nested sets of eluted peptides comprising length variants with the same core epitope are shown by lines below the sequence. The peptide sequence highlighted in red was used for biochemical and immunological assays (see Table 1). In C and D, each eluted sequence or nested set was identified by "P" followed by a number. E. Label-free quantification of proteins present in infected cells; proteins were ranked from most to least abundant, with viral proteins highlighted in color. F. Relationship between viral protein abundance and eluted peptide abundance. For each source protein, the sum of intensities of all eluted peptides derived from it was used to calculate the peptide abundance.   Ex vivo T cell responses to OC43 eluted peptides (pooled by HLA allele) in pre-pandemic PBMC samples from donors with a partial HLA match to HEK293 cells. The plot shows IFN-γ production measured by ELISpot (SFU/10 6 cells); pie graphs show the percentage of donors responding to the pool. B-C. Responding T cells from partially HLA-matched pre-pandemic donors were expanded in vitro by stimulation with each of the eluted peptides presented by a single allele antigen-presenting cells (APC). IFN-γ responses by expanded T cell populations from the same set of donors are shown in (B) for the HLA-DP peptides presented by DPA1*0301/DPB1*0401(DP4.1) and in (C) for the DR peptides presented by DRB1*1501 (DR2b) or DRB5*0101 (DR2a); pie graphs show the percentage of donors responding to the peptide. D. Summary of responses of single-peptide in vitro expanded T cells to the peptides, grouped by allele. E-F. Lowest peptide dose (10 -10 -7 µg/mL) eliciting a positive response to each eluted peptide, in experiments where the single-peptide in-vitro expanded T cells were tested for IFN-γ response to HLA-DP (E) or HLA-DR (F) eluted peptides presented by single allele APC (as in B-C). Each symbol represents a different donor. G. Response of single-peptide in-vitro expanded T cells to peptide stimulation followed in IFN-γ intracellular cytokine secretion (ICS) assay. Dot blots show CD4 expression (xaxis) and IFN-γ production (y-axis). DMSO, negative control. Responses > 3-fold background (DMSO) were considered positive. The gating strategy is presented in Figure S2. H-I. Summary of IFN-γ producing cell percentages in ICS assays for multiple donors for HLA-DP (H) and HLA-DR (I) peptides; only positive responses are shown. In A-C, statistical analysis to determine positive ELISpot responses was done by distribution-free resampling (DFR) method [89]; the size of the filled symbols indicates positive responses by DFR2x or DFR1x, while negative responses are shown as empty symbols. In A and D, statistical analysis was done by unpaired t-test (ns: not significant).

Figure 4: Epitope-specific T cell cross-reactivity between OC43 and other human coronaviruses. A.
Screening of cross-reactive T cell responses in partially HLA-matched pre-pandemic donors. IFN-γ responses (SFU/10 6 cells) to OC43 (green) or SARS-CoV-2 (blue) peptides using T cell lines expanded in vitro by stimulation with the eluted OC43 peptides and single allele APC. Pies show the fraction of responding donors to each peptide. B. For the two cross-reactive peptides (P4, P11), the screening was extended to more donors. C. Dose-response assay for the two cross-reactive peptides (P4, P11) in pre-pandemic donors. T cells were expanded in vitro with the OC43 peptide (TCL vs OC43, top row) or SARS-CoV-2 peptide (TCL vs CoV2, bottom row) and IFN-γ responses of each line to the OC43 peptide (green) or SARS-CoV-2 peptide (blue) were tested using single allele APC as before. D. Same as C but for COVID-19 convalescent donors. E. Lowest observed dose for a positive response for the cross-reactive peptides (tested in panels C and D). Pre-pandemic donors shown as circles and COVID-19 donors as triangles. F. Experimental binding of OC43 peptides (green) and the SARS-CoV-2 homologs (blue) to the relevant alleles. Half-maximal inhibitory concentration (IC50) values are shown. G. Sequence alignment of OC43 peptides and their SARS-CoV-2 homologs. OC43 sequences shown on top, with predicted core epitope shown in magenta and flanking regions in green; SARS-CoV-2 sequences on bottom, with residues different from OC43 shown and dots indicating identical residues. Predicted SARS-CoV-2 core epitope highlighted in turquoise with flanking regions shown in blue. Positions within the 9mer core epitope are indicated by numbers shown below the sequences; major T cell contacts are enclosed in circles. Arrowheads indicated gaps in the aligned sequences. If OC43 and SARS-CoV