The HLA-II immunopeptidome of SARS-CoV-2

SUMMARY

Targeted synthetic vaccines have the potential to transform our response to viral outbreaks, yet the design of these vaccines requires a comprehensive knowledge of viral immunogens. Here, we report severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) peptides that are naturally processed and loaded onto human leukocyte antigen-II (HLA-II) complexes in infected cells. We identify over 500 unique viral peptides from canonical proteins as well as from overlapping internal open reading frames. Most HLA-II peptides colocalize with known CD4⁺ T cell epitopes in coronavirus disease 2019 patients, including 2 reported immunodominant regions in the SARS-CoV-2 membrane protein. Overall, our analyses show that HLA-I and HLA-II pathways target distinct viral proteins, with the structural proteins accounting for most of the HLA-II peptidome and nonstructural and noncanonical proteins accounting for the majority of the HLA-I peptidome. These findings highlight the need for a vaccine design that incorporates multiple viral elements harboring CD4⁺ and CD8⁺ T cell epitopes to maximize vaccine effectiveness.

In brief

Weingarten-Gabbay et al. map the repertoire of SARS-CoV-2 peptides naturally presented on HLA-II. The authors uncover HLA-II peptides originating from noncanonical ORFs and highlight striking differences between viral proteins that are presented on class I and class II HLAs, resulting in distinct targets for killer and helper T cells.

Graphical Abstract

INTRODUCTION

Recent success with synthetic vaccines against viral diseases has demonstrated their promise for future use, but also highlighted the need for a deeper understanding of how the adaptive immune system recognizes and responds to viral infections.^1,2 In contrast to traditional attenuated vaccines that mimic the natural interaction between viruses and the immune system, rationally designed synthetic vaccines deliver only a small, selected portion of the viral genome, often a single protein, to stimulate an immune response. This allows rationally designed vaccines to be developed quickly, produced in large quantities with low-cost manufacturing, and administered safely.³ However, the development of rationally designed vaccines necessitates a comprehensive knowledge of viral immunology and the parts of the viral genome that are recognized by the different arms of the immune system.

Limitations of the coronavirus disease 2019 (COVID-19) vaccines developed during the course of the pandemic highlighted the importance of eliciting a more comprehensive immune response through T cells.⁴ Early vaccines mostly focused on eliciting humoral immune response by using the viral spike protein (S).^5–7 However, challenges with waning immunity led to a greater focus on providing durable immunity by eliciting CD8⁺ cytotoxic and CD4⁺ helper T cell responses. Together with the high frequency of mutations in S and the associated immune evasion of emerging variants, these findings motivated researchers to incorporate T cell targets, such as the nucleocapsid (N), in the next generation of COVID-19 vaccines.^8–12 CD8⁺ and CD4⁺ T cells recognize viral peptides that are endogenously processed and presented on the surface via human leukocyte antigen (HLA)-I and HLA-II complexes, respectively. These T cell epitopes originate from all viral proteins, in contrast to neutralizing antibodies that are mostly confined to external structural viral proteins. Although the presence of a large number of potential T cell epitopes in the viral genome offers a wide range of candidates, it can also present a challenge in identifying the most effective targets for T cell-inducing vaccines.

To understand the full range of T cell epitopes, it is important to implement nontargeted, comprehensive approaches for epitope discovery in addition to conventional T cell assays. Targeted T cell assays require researchers to decide a priori which peptides to screen in the experiment, based on assumptions about which viral proteins are expressed and processed for HLA presentation. Although T cell responses to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) have been extensively studied, many studies have focused on only a subset of viral proteins, particularly the structural proteins S, N, membrane (M), and envelope (E), rather than considering the full range of proteins in the SARS-CoV-2 proteome.^13–21 A smaller number of studies have looked at T cell responses to peptides from all annotated viral proteins and found significant responses to both structural and nonstructural proteins.^22–28 Although these studies provide a broader view, they have still generally included only peptides from annotated canonical proteins and have not considered peptides from noncanonical open reading frames (ORFs) that were identified through experimental translation measurements.²⁹

Taking an untargeted HLA immunopeptidome profiling approach, we recently revealed surprising insights about HLA-I presentation in SARS-CoV-2-infected cells. Immunopeptidome profiling uses mass spectrometry (MS) to detect peptides that are endogenously presented on the HLA complex in different disease contexts.^30–38 By applying it to SARS-CoV-2-infected cells, we uncovered HLA-I peptides from canonical proteins as well as overlapping ORFs in the coding region of N and S that were overlooked by previous T cell studies.³⁹ It is striking that some of the noncanonical peptides were more immunogenic in COVID-19 patients and humanized mice when compared to the strongest canonical protein-derived antigens reported to date. Two of the noncanonical peptides were further supported by an independent HLA-I peptidome study from another group.⁴⁰ We also uncovered that nonstructural proteins constitute a significant portion of the HLA-I peptidome and that the timing of viral protein expression is a key determinant of HLA-I presentation and immunogenicity. The list of SARS-CoV-2 HLA-I peptides resulting from our study was recently used to characterize a T cell–directed mRNA vaccine (BNT162b4).¹⁰ Of the 19 HLA-I peptides observed in cells expressing a multiepitope vaccine, 3 were identical to peptides that we detected in infected cells. HLA immunopeptidome profiling can identify new, highly potent T cell epitopes, inform vaccine design, and deepen our understanding of the determinants of viral antigen presentation.

Immunopeptidome profiling can be used to directly characterize another critical process in the antiviral immune response: HLA-II presentation to CD4⁺ helper T cells. In contrast to HLA-I presentation that samples endogenous cytosolic proteins, HLA-II can present peptides from proteins that have been taken up from outside the cell via endocytosis. Hence, different viral proteins may differentially access the HLA-I and HLA-II pathways. Understanding these differences can enable researchers to design vaccines that elicit both CD8⁺ and CD4⁺ T cell responses. Previous HLA-II peptidome studies of SARS-CoV-2 investigated HLA-II peptides derived only from the S protein, using a purified recombinant protein,^41,42 or from four viral proteins (N, M, E, and nonstructural protein 6 [nsp6]) using plasmid overexpression.⁴⁰ Thus, we have not yet achieved a systematic view of HLA-II peptides derived from the full SARS-CoV-2 genome in the context of authentic virus infection.

In this study, we set out to achieve a comprehensive map of SARS-CoV-2 peptides that are processed and presented by HLA-II complexes. Using these data, we dissected the source viral proteins and the processed regions within each viral protein that are presented by HLA-II. We contextualize these findings by using thousands of reported CD4⁺ T cell epitopes, inferring the contribution of antigen processing and presentation steps to T cell responses observed in COVID-19 patients. We then compared the immunopeptidome of HLA-I and HLA-II complexes, revealing important differences between SARS-CoV-2 presentation to CD4⁺ and CD8⁺ T cells. The newly identified CD4⁺ T cell targets and the insights that our study rendered about viral HLA-II presentation enables a more precise selection of peptides for the next generation of COVID-19 vaccines that aim to target multiple viral proteins. The concepts learned from this study can also be applied to the design of synthetic vaccines against other viral pathogens.

RESULTS

Immunopeptidome profiling of HLA-II peptides in SARS-CoV-2-infected cells

To interrogate the HLA-II immunopeptidome of SARS-CoV-2, we infected A549 and HEK293T cells with the virus after inducing the HLA-II presentation pathway, immunoprecipitated (IP) HLA-II-peptide complexes, and identified bound peptides by liquid chromatography-tandem MS (LC-MS/MS) (Figure 1A). Both cell lines were engineered to stably express angiotensin-converting enzyme 2 (ACE2) and transmembrane serine protease 2 (TMPRSS2), two important SARS-CoV-2 entry factors. To induce the HLA-II pathway, we transduced the cells with the major histocompatibility complex (MHC) class II transactivator (CIITA) using a lentiviral vector. Overexpression of CIITA, a master transcriptional regulator, facilitates the peptide-loading machinery and cell surface expression of HLA-II complexes and has been previously used to interrogate the HLA-II immunopeptidome of virus-infected cells^43,44 and tumors.^45,46 In addition, some lung epithelial cells express HLA-II,^47,48 and thus, studying the HLA-II immunopeptidome of infected cells mimics HLA-II presentation that can occur in vivo during the course of infection with a respiratory virus.

Figure 1. HLA-II immunopeptidome profiling of SARS-CoV-2-infected cells.

(A) Schematic of the experimental workflow. (B) HLA-II expression measured by flow cytometry using a fluorescein isothiocyanate (FITC)-conjugated anti HLA-DR/DP/DQ antibody. Gating was based on unstained cells.

(C) SARS-CoV-2 infection levels in A549/AT and HEK293T/AT cells with and without CIITA transduction. The infection was quantified using immunofluorescence staining for the nucleocapsid protein. Bar height is equal to the mean percentage of infected cells counted in 12 different fields; error bars correspond to mean ± SD.

(D) Length distribution of the eluted peptides in SARS-CoV-2-infected cells shown for peptides that were derived from human proteins (black), SARS-CoV-2 proteins (blue), and BSA (gray).

(E) GibbsCluster⁴⁹ deconvoluted motifs of the eluted peptides.

(F) The percentage of peptides originating from bovine proteins detected in the HLA-I and HLA-II immunopeptidomes. HLA-I data are from Weingarten-Gabbay et al.³⁹

(G) Length distribution of human and bovine peptides in the A549 and HEK293T HLA-I and the HLA-II immunopeptidomes. See also Figure S1 and Table S1.

We characterized the CIITA-transduced cells to ensure the proper induction of proteins in the HLA-II pathway. We compared the cell surface levels of HLA-II in A549 cells with those in a positive control human melanoma A375 cell line that endogenously expresses HLA-II.⁵⁰ The cell surface flow cytometry revealed strong induction (~80-fold) of HLA-II expression in A549/ACE2/TMPRSS2 (A549/AT) cells upon CIITA transduction, with a fluorescence intensity similar to that in A375 cells (Figure 1B). To monitor the expression of additional proteins in the HLA-II pathway, we examined the whole proteome of A549/AT and HEK293T/AT cells by analyzing lysates after the HLA-II IP using LC-MS/MS. We observed the expected increase in CIITA-induced proteins that are localized in the MHC class II region of the MHC locus, including HLA-DM, HLA-DO, and TAP1 (Figure S1A).

We also ensured that CIITA overexpression did not affect the cell susceptibility to SARS-CoV-2 infection, because a recent study reported that CIITA can restrict SARS-like coronaviruses in U2OS osteosarcoma cells.⁵¹ We quantified SARS-CoV-2 infection by immunofluorescence staining of cells with an anti-N antibody at 24 h postinfection (hpi) (Figures 1C and S1B). When infected at different MOIs, A549/AT cells exhibited similar infection levels, regardless of CIITA expression. In contrast, HEK293T/AT cells showed reduced infection upon CIITA expression, although a substantial number of cells were still positive, with ~50% infected cells at an MOI of 3 (the MOI we used for our HLA-II IP experiment).

To gauge the technical performance of our experimental system, we examined whether the peptides detected by LC-MS/MS match the known characteristics of HLA-II peptides. We performed HLA-II IP of noninfected and infected cells in 2 biological replicates at 24 hpi using a mixture of antibodies targeting the 3 HLA-II loci (HLA-DR, HLA-DP, and HLA-DQ). We recovered 21,541 and 29,600 unique HLA-II peptides from A549/ACE2/TMPRSS2/CIITA (A549/ATC) and HEK293T/ACE2/TMPRSS2/CIITA (HEK293T/ATC) cells, respectively (Table S1B). Of those, 0.5% (n = 119) of A549 and 1.3% (n = 389) of HEK293T peptides were derived from SARS-CoV-2 proteins (Tables S1C and S1D). This result aligns with the overall low percentage of viral peptides reported in previous studies that investigated the HLA-II immunopeptidome of EBV-B cells infected with measles virus⁵² and monocyte-derived dendritic cells (moDCs) pulsed with the recombinant H1-HA protein of influenza virus.⁵³ Furthermore, the recovered HLA-II peptides exhibited the expected 12–25 amino acid length distribution (Figure 1D). We examined whether detected peptide sequences conform to the binding preferences of the HLA-II alleles expressed in A549 and HEK293T cells. The unbiased identification of binding motifs, deconvoluted by GibbsCluster,⁴⁹ agreed with the known preferences of HLA-DR heterodimers⁵⁴ expressed in these cell lines (Figure 1E; Table S1A). As expected, the deconvoluted motifs matched more to HLA-DR heterodimers, and less so to HLA-DP and HLA-DQ heterodimers, in both cell lines, as HLA-DR is often expressed at higher levels when compared to HLA-DP and HLA-DQ.⁵⁵ Notably, some of the HLA-II alleles expressed by the two cell lines and confirmed by motif deconvolution are highly prevalent in the European (EUR) and US populations, including DRB1*07:01 (~14% EUR, ~12% US), expressed by A549 and DRB1*15:01 (~14% EUR, ~11% US), and DRB5*01:01 (~16% EUR), expressed by HEK293T (Table S1A).

Finally, we confirmed that CIITA-transduced cells presented peptides derived from extracellular proteins, as expected for HLA-II presentation. To this end, we quantified the fraction of HLA-II peptides derived from BSA, a nonhuman protein present in the cell growth medium, as done previously.⁴⁵ As a negative control, we examined the HLA-I immunopeptidome of the same cells.³⁹ Because HLA-I peptides are processed mostly from endogenous proteins, we expected low representation of the exogenous BSA protein in these data. Indeed, we observed 5.2- and 4.6-fold more BSA-derived peptides in the HLA-II peptidome than the HLA-I peptidome in A549/ATC and HEK293T/ATC cells, respectively (Figure 1F). Moreover, BSA-derived HLA-II peptides across both A549/ATC and HEK293T/ATC experiments had the expected lengths of 12–25 amino acids (~87% of peptides), matching the distribution of human-derived HLA-II peptides in the same samples (Figures 1D and 1G). In contrast, the BSA-derived peptides in HLA-I samples had longer lengths than canonical HLA-I peptides, suggesting that these peptides may have arisen from exogenous peptidase trimming and binding to empty surface HLA-I (Figure 1G). Finally, unsupervised analysis of binding motifs by GibbsCluster yielded concordant solutions for BSA-derived and human-derived peptides (Figure S1C), and sample-specific allele binding predictions by NetMHCpanII revealed proportionate allele distributions (Figure S1D). Altogether, these data indicate that CIITA-transduced A549/AT and HEK293T/AT cells can present HLA-II peptides generated from endocytosed proteins.

SARS-CoV-2 HLA-II peptides from canonical and out-of-frame overlapping ORFs

We next analyzed the HLA-II presented peptides derived from SARS-CoV-2 proteins, detecting 469 unique peptides from canonical viral proteins: N, S, M, ORF3a, ORF6, nsp3 and nsp4 (Figure 2A; Tables S1C and S1D). Mapping HLA-II peptides to SARS-CoV-2 proteins, we observed the expected clustering of peptides into nested sets, with shared core sequences but different N- or C-terminal ends (Figures 2B and S2). This pattern is a hallmark of the HLA-II pathway and differentiates it from the class I presentation. Although MHC class I molecules structurally constrain the length of loaded peptides, the open structure of the binding groove of MHC class II molecules allows interaction with peptides of variable length, with parts of the peptides protruding out of the binding groove.⁵⁶ Interestingly, although A549/ATC and HEK293T/ATC cells express different HLA-II alleles (Table S1A) with distinct binding preferences, some clusters contained ≥2 peptides from both cell lines (Figures 2B and S2B; Table S1F). This observation suggests that viral antigen processing steps upstream of HLA-II peptide loading play a key role in shaping the HLA-II immunopeptidome.

Figure 2. SARS-CoV-2 HLA-II immunopeptidome.

(A) Summary of viral proteins presented on HLA-II complexes in infected A549/ATC and HEK293T/ATC cells.

(B) The location of HLA-II peptides in canonical structural proteins M, N, and S. Peptides detected in A549/ATC and HEK293T/AT/C cells are depicted in black and gray, respectively.

(C) The location of HLA-II peptides in 2 noncanonical internal ORFs: ORF9b and ORF3c. See also Figures S2 and S3.

The untargeted nature of our approach allowed us to search for peptides originating from noncanonical ORFs,²⁹ in addition to the canonical SARS-CoV-2 proteins. Overall, we detected 11 peptides in HEK293T cells from 2 internal overlapping ORFs: ORF9b (overlapping with N, also called N.iORF1) and ORF3c (overlapping with ORF3a, also called 3a.iORF1). ORF9b gave rise to 10 peptides, all of which clustered into 2 nested sets in the first half of the protein (Figure 2C; Tables S1C, S1D, and S1F). From each nested set, at least 1 peptide was predicted to bind one of the HLA-II alleles expressed in HEK293T cells: DRB5*01:01 and DQA1*01:02/DQB1*06:02 in the first and second sets, respectively. Interestingly, HLA-II peptides arose from a different region of the ORF9b protein compared to HLA-I peptides, which originated from the C-terminal region.³⁹ We identified 1 HLA-II peptide (ALHFLLFFRALPKS) from ORF3c. Although it was only 1 peptide, it was observed in the same fraction (fxn 3) in 2 biological replicates and was a high scoring peptide, which supports its authenticity. This observation provides experimental evidence for ORF3c expression at the protein level because it was originally detected using ribosome profiling²⁹ and computational predictions.^57–59

To confirm the amino acid sequences of peptides derived from the 2 noncanonical ORFs, we compared the tandem mass spectra of 3 synthetic peptides with the experimental spectra obtained from the HLA-II immunopeptidome. We observed high correlation between fragment ions and retention times for all 3 peptides: ALHFLLFFRALPKS from ORF3c and VGPKVYPIILRLGSPL and VGPKVYPIILRLGSPLS from ORF9b (Figure S3).

SARS-CoV-2 HLA-II peptides colocalize with epitopes that elicit CD4⁺ T cell responses in COVID-19 patients

To evaluate whether the HLA-II peptides that we detected by LC-MS/MS contribute to T cell responses in COVID-19 patients, we compared our HLA-II viral peptides to reported CD4⁺ epitopes derived from SARS-CoV-2 proteins. We used a curated dataset of T cell epitopes reported by Grifoni et al.⁶⁰ This dataset combines 9 studies that tested CD4⁺ T cell responses using various assays, including ELISpot, intracellular cytokine staining, and activation-induced markers (Figure 3A).^{14,15,18,25,26,28,61–63}

Figure 3. Systematic comparison between the SARS-CoV-2 HLA-II peptides and reported CD4⁺ T cell epitopes.

(A) A list of 9 studies that identified CD4⁺ T cells epitopes in COVID-19 patients as described in Table 1 of Grifoni et al.⁶⁰ Studies that included peptides from all canonical SARS-CoV-2 proteins are highlighted with an asterisk.

(B) Comparing the immunogenicity of SARS-CoV-2 proteins that were detected on the HLA-II complex to proteins that were not. For each protein (each dot), we plotted the fraction of CD4⁺ T cell epitopes from all of the detected epitopes in 4 genome-wide studies (denoted with asterisk in A). The box shows the quartiles, the bar indicates median, and the whiskers show the distribution. Wilcoxon rank-sum p value is shown.

(C–E) Comparing the location of HLA-II peptides that were detected by LC-MS/MS in infected A549/AT/CIITA and HEK293T/AT/CIITA cells to previously reported CD4⁺ T cell epitopes. (Left) Heatmap showing the density of HLA-II peptides and the reported CD4⁺ T cell epitopes across individual viral proteins. Rows were normalized according to the maximal density in each row. (Right) Venn diagram showing the number of amino acids that are covered by HLA-II peptides, CD4⁺ T cell epitopes, and both. To reduce background levels, we counted amino acids (aa) with greater coverage than the mean of each row. Hypergeometric p value is shown for each viral protein.

We checked the overlap between viral proteins that generated HLA-II peptides and those known to contain CD4⁺ T cell epitopes. We computed the fraction of total CD4⁺ epitopes that were derived from each of the SARS-CoV-2 proteins. To avoid biases stemming from overrepresentation of highly characterized viral proteins, such as S, N, and M, we limited our analysis to 4 studies that surveyed the entire canonical SARS-CoV-2 proteome (highlighted by the asterisk in Figure 3A).^25,26,28,63 As expected, viral proteins for which we observed HLA-II peptides elicited T cell responses significantly more frequently than those with no detectable HLA-II peptides (Wilcoxon rank-sum p < 10⁻³; Figure 3B).

We then considered individual viral proteins and tested whether the HLA-II peptides colocalize with regions that elicited CD4⁺ T cell responses in COVID-19 patients. We focused on the 3 structural proteins that gave rise to the majority of the detected HLA-II peptides: N (n = 281), S (n = 143), and M (n = 56). These 3 proteins were also the most abundant source of epitopes in the compiled T cell data, accounting for 54% of the total CD4⁺ epitopes detected in the canonical SARS-CoV-2 proteome. To determine whether HLA-II peptides were derived from regions that were more immunogenic in patients, we counted the number of amino acids that were covered by HLA-II peptides, CD4⁺ T cell epitopes, and both, and computed a hypergeometric p value that estimates the overlap between these 2 groups. Because we compared peptide localization within individual proteins, we accounted for epitopes reported in all 9 CD4⁺ T cell studies listed in Figure 3A, including studies that examined only a few SARS-CoV-2 proteins. We found a statistically significant enrichment of HLA-II peptides in regions from which CD4⁺ T cell epitopes were derived with p < 10⁻²⁹, p < 10⁻⁵, and p < 10⁻⁵ for M, N, and S, respectively (Figures 3C–3E).

Furthermore, our analysis showed that HLA-II peptides greatly overlap the 2 immunodominant regions reported in the M protein. These regions, M:144–163 and M:173–192, were recently identified as hotspots of CD4-restricted epitopes that elicited T cell responses in 8 and 6 COVID-19 convalescent samples, respectively.¹⁴ The HLA-II peptides that we detected by LC-MS/MS were also confined to the same region of M, with the highest density in the 2 reported hot spots: M:121–171 (n = 34 peptides) and M:172–192 (n = 16 peptides), both detected in HEK293T/ATC and A549/ATC cells (Figures 2B and 3C; Tables S1C, S1D, and S1F). Moreover, the predicted HLA restriction of the 2 T cell epitopes from M:144–163 described by Keller et al.¹⁴ matches 2 of the HLA-II alleles expressed in A549: DRB1*11:04 and DRB4*01:01. Altogether, these results indicate that the HLA-II immunopeptidome of SARS-CoV-2 colocalizes with reported CD4⁺ T cell epitopes and defines the immunogenic regions in the M protein.

Distinct subsets of viral proteins are presented on the HLA-I and HLA-II complexes

In the context of vaccine design, it is important to decipher both CD4⁺ and CD8⁺ T cell epitopes because effective vaccines require potent induction of both arms of the adaptive cellular response. Many of the current vaccine strategies rely on the delivery of a single viral protein to induce both CD8⁺ and CD4⁺ T cell responses and mainly focus on the structural proteins S or N.^8,64–69 However, HLA-I and HLA-II presentation, which facilitate CD8⁺ and CD4⁺ T cell responses, respectively, have distinct antigen-processing steps, raising the likelihood that each of these pathways samples a different subset of viral proteins.

To compare viral proteins presented on HLA-I versus HLA-II complexes, we examined the HLA-I³⁹ and HLA-II immunopeptidomes of SARS-CoV-2-infected HEK293T/ATC and A549/ATC cells. We computed the number of peptides observed from each source protein and assessed the representation of proteins in 4 groups: nonstructural proteins, structural proteins, accessory proteins, and noncanonical ORFs. HLA-II presentation was dominated by the structural proteins N, S, and M, accounting for 95.2% of the detected peptides in HEK293T/ATC and A549/ATC cells, with negligible contribution of the nonstructural (1.2%), accessory (1.4%) and noncanonical proteins (2.2%) (Figure 4A). In contrast, only 27% of the detected HLA-I peptides were derived from structural proteins, with a large fraction of peptides from nonstructural proteins (45.9%) and noncanonical ORFs (24.3%) (Figure 4B). Together, these results point to a different presentation profile of SARS-CoV-2 on HLA-I and HLA-II complexes and suggest that CD8⁺ and CD4⁺ T cells engage with different parts of the virus.

Figure 4. SARS-CoV-2 protein representation on the HLA-I and HLA-II complexes.

(A) A bar chart showing the number of HLA-II peptides detected in SARS-CoV-2-infected A549/AT/CIITA and HEK293T/AT/CIITA cells for each viral protein. Inside the frame is a pie chart showing the relative abundance of peptides derived from structural proteins, nonstructural proteins, accessory proteins, and noncanonical ORFs.

(B) Similar to (A) for HLA-I peptides reported in Weingarten-Gabbay et al.³⁹.

(C) The source proteins for the HLA-II processing and presentation pathway. Viruses are endocytosed by an APC. The viral structural proteins are cleaved within the endosomal-lysosomal compartment and loaded onto HLA-II complexes.

(D) The source proteins for the HLA-I processing and presentation pathway. Viral proteins are produced from the translation of genomic and subgenomic viral RNAs in infected cells. These proteins are cleaved and loaded onto HLA-I complexes.

DISCUSSION

We provide a systematic view of HLA-II peptides that are naturally processed and presented from the SARS-CoV-2 proteome. This genome-wide view allowed us to comprehensively compare the HLA-II immunopeptidome to the T cell epitopes in COVID-19 patients and to the HLA-I immunopeptidome of SARS-CoV-2 and to uncover insights into antigen presentation.

Our work adds to a growing list of studies that use the overexpression of the CIITA master regulator to infer the HLA-II immunopeptidome of cancer cells and viruses.^43–46 Although the cells we profiled were not professional antigen-presenting cells (APCs), we believe that our measurements revealed genuine HLA-II peptides. The peptides that we detected were in the expected length range and contained sequence motifs that matched the HLA-II alleles of the respective cell lines. In addition, we detected the presentation of exogenous proteins derived from extracellular bovine serum. Most important, our immunopeptidome data successfully captured the SARS-CoV-2 CD4⁺ T cell epitopes that were detected in a wide range of independent studies using conventional targeted T cell assays. The source proteins of the HLA-II peptides in our study were also found to be the most immunogenic proteins in convalescent COVID-19 patients. Moreover, the HLA-II peptides colocalized with regions in the SARS-CoV-2 proteins that are known to elicit strong CD4⁺ T cell responses.

In the context of highly pathogenic viruses, CIITA induction in HLA-II null cell lines provides an efficient platform to probe the entire viral genome within the restrictions of high-containment facilities. To date, HLA-II immunopeptidome studies of SARS-CoV-2 were limited to a single protein, using pulse experiments with a recombinant S protein,^41,42 or 4 exogenously expressed proteins using plasmid overexpression of N, M, E, and nsp6.⁴⁰ By inducing the HLA-II pathway in SARS-CoV-2-infected cells, we could detect HLA-II peptides from the entire viral genome. Although HLA-II peptides are predominantly presented by APCs, such as DCs and macrophages, they can also be presented by nonimmune cells upon induction of HLA-II expression, such as lung epithelial cells exposed to interferon-γ (IFN-γ).^47,48 Thus, the CIITA induction system can also recapitulate naturally occurring HLA-II presentation events of infected cells in vivo. Overexpressing CIITA in the same cell lines in which we profiled the HLA-I immunopeptidome allowed us to use the same reagents and inactivation protocol that we optimized for Biosafety Level 3 (BSL3) laboratory.⁷⁰ Thus, we believe that this approach can be readily extended to additional high-containment viruses.

Our analysis uncovers striking differences in the subset of viral proteins that are presented on HLA-I versus HLA-II complexes. We hypothesize that the observed differences stem from the different stages of the viral life cycle at which viral proteins are processed and loaded onto HLA molecules (Figures 4C and 4D). The HLA class II pathway can present peptides from exogenous sources (e.g., mature virions, infected cell debris) that are cleaved within endosomal-lysosomal compartments and loaded onto HLA-II complexes (Figure 4C). Hence, the repertoire of HLA-II peptides reflects the viral proteins that are part of mature virus particles. In contrast, the HLA-I pathway samples viral proteins that are actively translated in the cytoplasm of infected cells. Thus, the HLA-I immunopeptidome mirrors the translatome of the virus, including nonstructural proteins, accessory proteins, and noncanonical proteins, which are not necessarily incorporated into mature virions (Figure 4D). These differences may extend to other viruses as well, many of which encode proteins that are required for viral replication in infected cells but are not packaged into mature virions. The distinct repertoire of viral peptides that are presented on HLA-I versus HLA-II complexes suggests that CD8⁺ and CD4⁺ T cells recognize different parts of the viral genome. This observation warrants revision of the current one-protein-fits-all approach that forms the basis of most synthetic vaccines, including the broadly administered COVID-19 mRNA vaccines. Instead, incorporating targets from different classes of viral proteins has the potential to activate both CD4⁺ and CD8⁺ T cells, with each class reacting to viral components naturally presented to them during infection (e.g., including structural proteins to elicit CD4⁺ responses and nonstructural/noncanonical proteins to elicit CD8⁺ responses).

An intriguing question in T cell immunology is what determines the immunodominance of a specific protein or a region within a protein. Observing T cell reactivity in convalescent patients to immunodominant epitopes represents the final outcome in a chain of events that determine which peptides will elicit a T cell response. Some factors defining the immunodominance of a given epitope include source protein expression levels, accessibility to proteolytic cleavage and antigen processing, loading onto the HLA complex, the presence of a matching T cell receptor (TCR) in the repertoire of naive T cells, the binding affinity of an HLA-peptide complex with its matching TCR, and elimination of T cells reacting to self epitopes. Because the immunopeptidome represents antigen processing and presentation steps occurring before interaction with TCRs, it distinguishes between epitope selectivity at the level of presentation versus T cell recognition. A recent study identified that the HLA-II peptidome of influenza virus defines H1-HA immunodominant regions targeted by memory CD4⁺ T cells.⁵³ Similar to this observation, we found that the 2 immunodominant hotspots in the M protein¹⁴ greatly overlap the HLA-II immunopeptidome of infected cells, suggesting that antigen processing and presentation steps define the immunodominant regions of M. Together, these observations demonstrate the robustness of immunopeptidome analysis for identifying relevant protein regions for designing T cell assays and selecting candidates for vaccines.

Our work uncovers HLA-II peptides from 2 noncanonical overlapping ORFs: ORF9b and ORF3c. In contrast to HLA-I presentation, in which noncanonical peptides were enriched on the HLA-I complex, noncanonical peptides represented only a small fraction of the HLA-II immunopeptidome. However, the detection of peptides from noncanonical ORFs in both the HLA-I and the HLA-II immunopeptidomes emphasizes the importance of incorporating noncanonical ORFs into T cell studies to achieve a complete picture of the antiviral immune profile. The discovery of a peptide from ORF3c highlights another important aspect of immunopeptidome studies and its potential impact on our understanding of noncanonical ORFs. Although noncanonical ORFs have numerous roles in the viral life cycle, including regulating viral gene expression and modulating virus infection, their detection in tryptic proteome experiments is often challenging due to their small size, shorter half-life, and, in some cases, the lack of observable tryptic peptides. Of the 23 noncanonical ORFs that were identified in the SARS-CoV-2 genome,²⁹ only one (ORF9b) has thus far been detected in global tryptic proteomic experiments.³⁹ The longer half-life of the HLA peptide complex compared to the noncanonical ORF translation product in the cell may increase the probability of detection by LC-MS/MS.⁷¹ As an example, although we and others identified 3 peptides from S.iORF1 (an internal overlapping ORF in S) on the HLA-I complex,^39,40 this protein was not detected in whole tryptic proteome analysis of the same infected cell lysates. ORF3c was recently shown to inhibit innate immunity by restricting IFN-β production, exposing an important mechanism of SARS-CoV-2 immune evasion.⁷² Our study provides evidence that this important noncanonical protein is expressed in cells that are infected with SARS-CoV-2. Thus, in addition to enhancing our understanding of viral antigen presentation, immunopeptidome studies contribute to our basic understanding of viruses by illuminating the complete set of canonical and non-canonical viral proteins.

Limitations of the study

Although CIITA overexpression activated the HLA-II pathway in the cell lines used in our study, allowing the investigation of SARS-CoV-2 HLA-II immunopeptidome, these cells may not capture the unique biology of how APC subtypes uptake and process mature virions and virus-infected cells in vivo. In contrast to the cells profiled in this study, APCs are not naturally infected by SARS-CoV-2. It is possible that productive infection affects the antigen processing and presentation steps and provides an internal source of viral proteins, in addition to the endocytosed particles, for production of HLA-II peptides. For instance, Ghosh and colleagues⁷³ showed that β-coronaviruses can traffic to lysosomes and egress by Arl8b-dependent lysosomal exocytosis. In this context, lysosomes are deacidified, which can inactivate proteolytic enzymes in infected cells and impair antigen presentation. Viral infection can alter the expression of endogenous host proteins, which, in turn, can affect the HLA-II peptidome derived from human proteins. Although we did not analyze the host immunopeptidome in our study, we have made these data available to readers, and they should be aware of this limitation. Finally, our study only uncovered the peptides that are presented by the HLA-II alleles expressed in A549 and HEK293T cells. As such, we may have missed SARS-CoV-2 HLA-II peptides that were incompatible with A549 and HEK293T cells. Nonetheless, our study identifies HLA-II peptides that can be presented on virally infected cells and provides valuable insights into which viral proteins are more likely to be presented by SARS-CoV-2-infected cells.

STAR★METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to the lead contact and corresponding author, Shira Weingarten-Gabbay (shirawg@broadinstitute.org).

Materials availability

Cell lines transduced with ACE2, TMPRSS2, and CIITA are available upon request.

Data and code availability

1. The original mass spectra, peptide spectrum matches, and the protein sequence database used for searches have been deposited in the public proteomics repository MassIVE (http://massive.ucsd.edu) under the identifier MSV000091943 and are accessible at ftp://massive.ucsd.edu/MSV000091943/.

2. This paper does not report original code

3. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Biocontainment

All experiments involving authentic SARS-CoV-2 were performed in a state-of-the-art biosafety level 3 (BSL3) facility at the National Emerging Infectious Diseases Laboratories (NEIDL) of Boston University using biosafety protocols approved by the Institutional Biosafety Committee (IBC), comprising scientists, biosafety and compliance experts as well as local community members. The biosafety protocols were further approved by the Boston Public Health Commission. All personnel received rigorous biosafety, biosecurity, and BSL3 training before participating in experiments. Special personal protective equipment, including scrubs, disposable overalls, shoe covers, double-layered gloves, and powered air-purifying respirators, was used during BSL3 work.

Cells

All cell lines were incubated at 37°C and 5% CO2 in a humidified incubator. Human embryonic kidney HEK293T cells (female; ATCC; CRL-3216), human lung adenocarcinoma A549 cells (male; ATCC; CCL-185), human melanoma A375 cells (female; ATCC; CRL-1619), and African green monkey kidney Vero E6 cells (ATCC; CRL-1586) were maintained in DMEM (Gibco; #11995–065) containing 10% FBS and 1X non-essential amino acids. Mycoplasma negative status of all cell lines was confirmed. Detailed information regarding all the cell lines used was provided in the key resources table.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
BD Pharmingen™ FITC Mouse anti-human HLA-DR, DP, DQ	BD Biosciences	Cat#562008 RRID:AB_10897011
Sars Nucleocapsid Protein Antibody	Rockland	Cat#200–401-A50 RRID:AB_828403
Goat anti-Rabbit IgG (H + L) Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 568	Thermo Fisher Scientific	Cat#A-11011 RRID:AB_143157
Recombinant Anti-HLA-DR antibody [TAL 1B5]	abcam	Cat#ab20181 RRID:AB_445401
Recombinant Anti-HLA-DPB1 antibody [EPR11226]	abcam	Cat#ab157210 RRID:AB_2827533
Anti-HLA-DQ antibody [B-K27] - BSA and Azide free	abcam	Cat#ab47342 RRID:AB_742182
Bacterial and virus strains
SARS-CoV-2 USA-WA1/2020	BEI Resources	Cat#NR-52281
Chemicals, peptides, and recombinant proteins
DMEM, high glucose, pyruvate	Gibco	Cat#11995065
Puromycin dihydrochloride from Streptomyces alboniger	Millipore Sigma	Cat#P8833
Blasticidine S hydrochloride	Millipore Sigma	Cat#3513–03-9
Hygromycin B	ThermoFisher Scientific	Cat#10687010
DMEM	ATCC	Cat#30–2002
Trypsin-EDTA 0.25%	Gibco	Cat#25200056
PBS, pH 7.4	Gibco	Cat#10010023
DPBS, no calcium, no magnesium	Gibco	Cat#14190144
Paraformaldehyde, 8% w/v aq. soln., methanol free	Thermo Scientific	Cat#047347.9M
Triton X-	Sigma Aldrich	Cat#T9284
DAPI (4’,6-diamidino-2-phenylindole, dihydrochloride)	Thermo Scientific	Cat#62247
1 M Tris, pH 8.0	Invitrogen	Cat#AM9855G
EDTA	Sigma Aldrich	Cat#7789
Sodium chloride	Sigma Aldrich	Cat#71376
Magnesium chloride	Sigma Aldrich	Cat#7786–30-3
Octyl b-d-glucopyranoside	Sigma Aldrich	Cat#08001
Iodoacetamide	Sigma Aldrich	Cat#A3221
complete™, Mini, EDTA-free Protease Inhibitor Cocktail	Roche	Cat#11836170001
Phenylmethanesulfonyl fluoride	Sigma Aldrich	Cat#78830
Benzonase	Thomas Scientific	Cat#E1014–25KU
Gammabind Plus Sepharose beads	Millipore Sigma	Cat#17–0886-01
Methanol	Honeywell	Cat#34966
Acetonitrile	Honeywell	Cat#34967
Formic Acid	Sigma Aldrich	Cat#F0507
Acetic acid, glacial	Sigma Aldrich	Cat#AX0073
Ammonium hydroxide solution, 28% (wt/vol) in H2O	Sigma Aldrich	Cat#338818
UltraPure™ SDS Solution, 10%	Thermo Fisher Scientific	Cat#15553027
Dithiothretiol, No-Weigh Format	Fisher Scientific	Cat#20291
Phosphoric acid	Millipore Sigma	Cat#7664–38-2
Tetraethylammonium bromide	Millipore Sigma	Cat#71–91-0
Trypsin, Mass Spec Grade	Promega	Cat#V511X
Lysyl endopeptidase	Wako Chemicals	Cat#129–02541
Ammonium formate	Millipore Sigma	Cat#540–69-2
Critical commercial assays
Pierce™ BCA Protein Assay Kit	Thermo Scientific	Cat#23225
Deposited data
Original mass spectra, peptide spectrum matches and databases	MassIVE	MSV000091943
Experimental models: Cell lines
HEK293T	ATCC	Cat#CRL-3216; RRID:CVCL_0063
A549	ATCC	Cat#CCL-185; RRID:CVCL_0023
VERO E6	ATCC	Cat#CRL-1586; RRID:CVCL_0574
A375	ATCC	Cat#CRL-1619; BRID: CVCL_0132
Recombinant DNA
pLOC_hACE2_PuroR	Chen et al., 2023	N/A
pLOC_hTMPRSS2_BlastR	Chen et al., 2023	N/A
pcDNA3 myc CIITA	Addgene	Cat#14650
pTRIP-SFFV-Hygro-2A	Gentili et al., 2022	N/A
pTRIP-SFFV-Hygro-2A-myc-CIITA	This study	N/A
Software and algorithms
Orbitrap Exploris 480 Xcalibur	Thermo Scientific	3.1–3.1.231.6/3.1.279.9
Spectrum Mill (SM)	Broad Institute	https://proteomics.broadinstitute.org, Version 7.08
Qual Browser Thermo Xcalibur	Thermo Scientific	3.0.63
UCSC Table Browser	University of California, Santa Cruz	https://genome.ucsc.edu/cgi-bin/hgTables

Plasmids

Lentiviral vectors pLOC_hACE2_PuroR and pLOC_hTMPRSS2_BlastR, harboring human ACE2 and TMPRSS2, respectively, have been described.⁷⁴ To generate a lentiviral vector containing human CIITA, we amplified the CIITA cDNA from pcDNA3 myc CIITA (Addgene #14650) and cloned it into pTRIP-SFFV-Hygro-2A (previously described⁷⁵) via Gibson assembly. The resultant plasmid was named pTRIP-SFFV-Hygro-2A-myc-CIITA.

METHOD DETAILS

Generation and characterization of stable cell lines

To generate HEK293T and A549 cells overexpressing human ACE2, TMPRSS2, and CIITA, we transduced these cells with lentiviral vectors pLOC_hACE2_PuroR, pLOC_hTMPRSS2_BlastR, and pTRIP-SFFV-Hygro-2A-myc-CIITA, and selected for the triple-transduced cells in culture medium supplemented with 1 μg/mL each of puromycin and blasticidin and 320 μg/mL of hygromycin.

Flow cytometry

1.5*10^6 A549 or A375 cells were incubated with 2ul FITC anti HLA-DR, DP, DQ antibody (BD Pharmingen #562008) in 100ul PBS at 4°C for 45 min. Cells were washed three times with PBS, resuspended in 400ul PBS and analyzed using a CytoFLEX flow cytometer.

SARS-CoV-2 virus stock preparation and titration

The SARS-CoV-2 USA-WA1/2020 isolate (NCBI accession number: MT246667) was deposited by the Centers for Disease Control and Prevention and obtained through BEI Resources, NIAID, NIH (NR-52281). We then passaged the virus twice onto Vero E6 cells to obtain the P2 stock, as previously described (Chen et al., JVI). The virus titration was performed on Vero E6 cells. All experiments in this study utilized the P2 stock.

Quantification of virus infectivity using immunofluorescence

A549 and 293T cells stably overexpressing ACE2, TMPRSS2, and CIITA were infected with SARS-CoV-2 at an MOI of 0.5, 1, or 3 for 12, 18, 24, 36, or 48 h. At indicated times, the culture medium was removed, and the cells were fixed with 4% paraformaldehyde for 60 min at room temperature. The cells were then permeabilized with 0.1% of Triton X-100 in PBS for 10 min and hybridized with anti-SARS-CoV nucleocapsid (rabbit polyclonal) antibody (1:2000, Rockland, #200–401-A50) at 4°C overnight. Alexa Fluor 568 goat anti-rabbit antibody (Invitrogen, #A11011) was used as the secondary antibody. Finally, DAPI was used to stain cell nuclei. Images were captured with an EVOS microscope using a 10x lens, and the percentage of infected cells was calculated with ImageJ.

Immunoprecipitation of HLA-II complexes from cells

Cells from 3 × 15cm dishes were scraped into 2.5mL/dish of cold lysis buffer (20mM Tris, pH 8.0, 100mM NaCl, 6mM MgCl2, 1mM EDTA, 60mM Octyl β-d-glucopyranoside, 0.2mM Iodoacetamide, 1.5% Triton X-100, 50xC0mplete Protease Inhibitor Tablet-EDTA free and PMSF) obtaining a total of ~9 mL lysate. This lysate was split into 6 eppendorf tubes, with each tube receiving 1.5 mL volume, and incubated on ice for 15 min with 1ul of Benzonase (Thomas Scientific, E1014–25KU) to degrade nucleic acid. The lysates were then centrifuged at 4,000 rpm for 22 min at 4°C and the supernatants were transferred to another set of 6 eppendorf tubes containing a mixture of pre-washed beads (Millipore Sigma, GE17–0886-01) and 12.5 μL (12.5 μg) of MHC class II antibodies in a 3:1:1 mixture of TAL-1B5 (Abcam, ab20181), EPR11226 (Abcam, ab157210) and B-K27 (Abcam, ab47342). The immune complexes were captured on the beads by incubating on a rotor at 4°C for 3hr in the BSL3 lab. Virus inactivation was confirmed before subsequent samples processing outside the BSL3 using plaque assay 39,70. In total, nine washing steps were performed; one wash with 1mL of cold lysis wash buffer (20mM Tris, pH 8.0, 100mM NaCl, 6mM MgCl2, 1mM EDTA, 60mM Octyl β-d-glucopyranoside, 0.2mM Iodoacetamide, 1.5% Triton X-100), four washes with 1mL of cold complete wash buffer (20mM Tris, pH 8.0, 100mM NaCl, 1mM EDTA, 60mM Octyl β-d-glucopyranoside, 0.2mM Iodoacetamide), and four washes with 20mM Tris pH 8.0 buffer. Dry beads were stored at −80°C until mass spectrometry analysis was performed.

HLA-II peptidome desalting and LC-MS/MS data generation

HLA peptides were eluted and desalted from beads as follows: wells of the tC18 40mg Sep-Pak desalting plate (Waters, Milford, MA) were activated with 2× 1 mL of methanol (MeOH) and 500 μL of 99.9% acetonitrile (ACN)/0.1% formic acid (FA), then washed with 4× 1 mL of 1% FA. A 10μmPE fritted filter plate containing the HLA-IP beads was placed on top of the Sep-Pak plate. To dissociate peptides from HLA molecules and facilitate peptides binding to the tC18 solid phase, 200 μL of 3% ACN/5% FA was added to the beads in the filter plate. 100 fmol internal retention time (iRT) standards (Biognosys SKU: Ki-3002–2) was spiked into each sample as a loading control and pushed through both the filter plate and 40 mg Sep-Pak plate. Following sample loading there was one wash with 400 μL of 1% FA. Beads were then incubated with 500 μL of 10% acetic acid (AcOH) three times for 5 min to further dissociate bound peptides from the HLA molecules. The beads were rinsed once with 1 mL 1% FA and the filter plate was removed. The Sep-Pak desalt plate was rinsed with 1 mL 1% FA an additional three times. The peptides were eluted from the Sep-Pak desalt plate using 250 μL of 15% ACN/1% FA and 2× 250 mL of 50% ACN/1% FA. HLA peptides were eluted into 1.5 mL micro tubes (Sarstedt, Nümbrecht, Germany), frozen, and dried down via vacuum centrifugation. Dried peptides were stored at −80°C until microscaled basic reverse phase separation.

Briefly, peptides were loaded on Stage-tips with 2 punches of SDB-XC material (Empore 3M). HLA-II peptides were eluted in three fractions with increasing concentrations of ACN (5%, 15%, and 40% in 0.1% NH₄OH, pH 10). Peptides were reconstituted in 3% ACN/5% FA prior to loading onto an analytical column (35 cm, 1.9 μm C18 (Dr. Maisch HPLC GmbH), packed in-house PicoFrit 75 μm inner diameter, 10 μm emitter (New Objective)). Peptides were eluted with a linear gradient (EasyNanoLC 1200, Thermo Fisher Scientific) ranging from 6 to 30% Solvent B (0.1% FA in 90% ACN) over 84 min, 30–90% B over 9 min and held at 90% B for 5 min at 200 nL/min. MS/MS data were acquired on a Thermo Orbitrap Exploris 480 equipped with (HLA-I) and without (HLA-II) FAIMS (Thermo Fisher Scientific) in data-dependent acquisition. FAIMS compensation voltages (CVs) were set to −50 and −70 with a cycle time of 1.5 s per FAIMS experiment. MS2 fill time was set to 100 ms; collision energy was 30, 34 or 36 CE see Table S2 for file mappings and LC-MS/MS acquisition variations.

Proteome analysis from HLA enrichment flow-through

Flow-throughs of the HLA-II IP that were stored as flash-frozen native protein lysates were briefly thawed on ice for ~15 min. Once thawed, 10% SDS was added for a final concentration of 5% SDS to denature the lysate, resulting in a final volume of ~1.5 mL lysate which was prepared for S-Trap digestion.⁷⁶

Protein concentration was estimated using a BCA assay for scaling of digestion enzymes. Disulfide bonds were reduced in 5 mM DTT for 30 min at 25°C and 1000 rpm shaking and cysteine residues were alkylated in 10 mM IAA in the dark for 45 min at 25°C and 1000 rpm shaking. Lysates were then transferred to a 15 mL conical tube to prepare for protein precipitation. 27% phosphoric acid was added at a 1:10 ratio of lysate volume to acidify and proteins were precipitated with 6x sample volume of ice-cold S-Trap buffer (90% methanol, 100 mM TEAB). The precipitate was transferred in successive loads of 3 mL to an S-Trap Midi (Protifi) and loaded with 1 min centrifugation at 4000 × g, mixing the remaining precipitate thoroughly between transfers. The precipitated proteins were washed 4x with 3 mL S-Trap buffer at 4000 × g for 1 min. To digest the deposited protein material, 350 μL digestion buffer (50 mM TEAB) containing both trypsin and endopeptidase C (LysC), each at 1:50 enzyme:substrate, was passed through each S-Trap column with 1 min centrifugation at 4000 × g. The digestion buffer was then added back atop the S-Trap and the cartridges were left capped overnight at 25°C.

Peptide digests were eluted from the S-Trap, first with 500 μL 50 mM TEAB and next with 500 μL 0.1% FA, each for 30 s at 1000 × g. The final elution of 500 μL 50% ACN/0.1% FA was centrifuged for 1 min at 4000 × g to clear the cartridge. Peptide concentration of the pooled elutions was estimated with a BCA assay, and 10 μg peptide was used for stagetip fractionation. Each 25ug proteome sample was reconstituted in 4.5 mM ammonium formate (pH 10) in 2% (v/v) acetonitrile and separated into four fractions using basic reversed phase fractionation on a C-18 Stage-tip. Fractions were eluted at 5%, 12.5%, 15%, and 50% ACN/4.5 mM ammonium formate buffer (pH 10) and dried. Fractions were reconstituted in 3%ACN/5%FA, and 1 μg was used for LC-MS/MS analysis.

Data-dependent acquisition was performed using a Thermo Orbitrap Exploris 480 in positive ion mode at a spray voltage of 1.9 kV MS1 spectra were measured with a resolution of 60,000, a normalized AGC target of 300% for, a maximum injection time of 10 ms, and a mass range from 350 to 1800 m/z. The data-dependent mode cycle was set to trigger MS/MS on up to the top 20 most abundant precursors per cycle at an MS2 resolution of 45,000, an AGC target of 30%, an isolation window of 0.7 m/z, a maximum injection time of 105 ms for proteome, and an HCD collision energy of 34%. Peptides that triggered MS/MS scans were dynamically excluded from further MS/MS scans for 20 s for the tryptic proteome, with a ±10 ppm mass tolerance. Theoretical precursor envelope fit filter was enabled with a fit threshold of 50% and window of 1.2 m/z. Monoisotopic peak determination was set to peptide and charge state screening was enabled to only include precursor charge states 2–6 with an intensity threshold of 5.0e3. Advanced peak determination (APD) was enabled. “Perform dependent scan on single charge state per precursor only” was disabled.

LC-MS/MS data interpretation

Peptide sequences were interpreted from MS/MS spectra using Spectrum Mill (SM) v 7.08 (proteomics.broadinstitute.org) against a RefSeq-based sequence database containing 41,457 proteins mapped to the human reference genome (hg38) obtained via the UCSC Table Browser (https://genome.ucsc.edu/cgi-bin/hgTables) on June 29, 2018, with the addition of 13 proteins encoded in the human mitochondrial genome, 264 common laboratory contaminant proteins, 553 human non-canonical small open reading frames, 28 SARS-CoV2 proteins obtained from RefSeq derived from the original Wuhan-Hu-1 China isolate NC_045512.2 (https://www.ncbi.nlm.nih.gov/nuccore/1798174254;⁷⁷), and 23 unannotated virus ORFs whose translation is supported by Ribo-seq²⁹ for a total of 42,337 proteins. Among the 28 annotated SARS-CoV2 proteins we opted to omit the full-length polyproteins ORF1a and ORF1ab, to simplify peptide-to-protein assignment, and instead represented ORF1ab as the mature 16 individual non-structural proteins that result from proteolytic processing of the 1a and 1ab polyproteins. We added the D614G variant of the SARS-Cov2 Spike protein that is commonly observed in European and American virus isolates, and also added 2036 entries from 6-frame translation of the SARS-Cov2 genome for all possible ORFs longer than 6 amino acids.

For immunopeptidome data, MS/MS spectra were excluded from searching if they did not have a precursor MH+ in the range of 600–4,000, had a precursor charge > 5, or had a minimum of < 5 detected peaks. Parameters for the SM MS/MS search module for HLA-II immunopeptidomes included: no enzyme specificity; precursor and product mass tolerance of ±10 ppm; minimum matched peak intensity of 30%; ESI-QEXACTIVE-HCD-HLA-v3 scoring; fixed modification: carbamidomethylation of cysteine; variable modifications: cysteinylation of cysteine, oxidation of methionine, deamidation of asparagine, acetylation of protein N-termini, and pyroglutamic acid at peptide N-terminal glutamine; and precursor mass shift range of −18 to 81 Da. For tryptic proteomes, parameters included: “trypsin allow P” enzyme specificity with up to 4 missed cleavages, precursor and product mass tolerance of ±20 ppm, and 30% minimum matched peak intensity. Scoring parameters were ESI-QEXACTIVE- HCD-v2. Allowed fixed modifications included carbamidomethylation of cysteine and selenocysteine.Allowed variable modifications for whole proteome data were acetylation of protein N-termini, oxidized methionine, deamidation of asparagine, hydroxylation of proline in PG motifs, pyro-glutamic acid at peptide N-terminal glutamine, and pyro-carbamidomethylation at peptide N-terminal cysteine with a precursor MH + shift range of −18 to 97 Da. Using the SM Autovalidation module, peptide-spectrum matches (PSMs) for individual spectra were confidently assigned by applying target-decoy based FDR estimation to achieve <1.0% FDR at the PSM, peptide levels for the HLA-II immunopeptidome. PSM-level thresholding was done with a minimum peptide length of 7, minimum backbone cleavage score of 5, and <1.0% FDR across all three fractions. Allowed precursor charges were +2 to +6. For whole proteome data, MS/MS spectra were excluded from searching if they did not have a precursor MH+ in the range of 600–6,000, had a precursor charge > 5, had a minimum of < 5 detected peaks, or failed the spectral quality filter with a sequence tag length > 0 (i.e., minimum of 2 masses separated by the in-chain masses of 1 amino acid) based on ESI-QEXACTIVE-HCD-v4–30-20 peak detection. Similar spectra with the same precursor m/z acquired in the same chromatographic peak were merged. MS/MS search parameters included: ESI-QEXACTIVE-HCD-v4–30-20 scoring parameters; Trypsin allow P specificity with a maximum of 4 missed cleavages; fixed modification: carbamidomethylation of cysteine and seleno-cysteine; variable modifications: oxidation of methionine, deamidation of asparagine, acetylation of protein N-termini, pyroglutamic acid at peptide N-terminal glutamine, and pyro-carbamidomethylation at peptide N-terminal cysteine; precursor mass shift range of −18 to 64 Da; precursor mass tolerance of ± 20 ppm; product mass tolerance of ± 20 ppm, and a minimum matched peak intensity of 30%. Peptide spectrum matches (PSMs) for individual spectra were automatically designated as confidently assigned using the Spectrum Mill auto-validation module to apply target-decoy based FDR estimation at the PSM level of < 1.0% FDR. Protein level data was summarized by top uses shared (SGT) peptide grouping and non-human contaminants were removed. HLA-II peptides were further filtered to remove non-human contaminants (Bovine retained for only Figure 1 F, G) and peptides that match peptides identified in blank bead negative control IPs.^30,33

LC-MS/MS synthetic peptide analysis of noncanonical HLA-II peptides

Synthetic peptides were purchased from GenScript for the MS/MS spectra comparisons shown in Figure S3. Synthetic peptides were analyzed at 1, 10, 50, and 100 fmol without background. The synthetic peptide data were collected on a Thermo Orbitrap Exploris 480 mass spectrometer equipped with a NanoSpray Flex NG ion source.

Data-dependent acquisition of the peptides was performed using Orbitrap Exploris 480 in positive ion mode at a spray voltage of 1.9 kV MS1 spectra were measured with a resolution of 60,000, a normalized AGC target of 100% for, a maximum injection time of 50 ms, and a mass range from 350 to 1700 m/z. The data-dependent mode cycle was set to trigger MS/MS on up to the top 15 most abundant precursors per cycle at an MS2 resolution of 15,000, an AGC target of 50%, an isolation window of 1.1 m/z, a maximum injection time of 100 ms, and an HCD collision energy of 34%. Peptides that triggered MS/MS scans were dynamically excluded from further MS/MS scans for 10 s, with a ±10 ppm mass tolerance. Theoretical precursor envelope fit filter was enabled with a fit threshold of 50% and window of 1.4 m/z. Monoisotopic peak determination was set to peptide and charge state screening was enabled to only include precursor charge states 2–5 with an intensity threshold of 1.0e3. Advanced peak determination (APD) was enabled.

Parameters for the SM MS/MS search module for HLA-II immunopeptidomes were the same as above.

QUANTIFICATION AND STATISTICAL ANALYSIS

Co-localization of HLA-I peptides and CD4⁺ T cell epitopes

CD4⁺ T cell epitopes that were detected in COVID-19 patients and their position within each viral protein were obtained from Table 1 of Grifoni et al.⁶⁰ For each amino acid (aa) in each viral protein, we computed the number HLA-II peptides (this study) and the number of T cell epitopes (Table 1 of Grifoni et al.) covering this position. We then divided each array (row) by the maximal value among all the amino acids to compute relative densities of HLA-II peptides and T cell epitope. To evaluate co-localization between HLA-II peptides and CD4⁺ T cell epitopes, we computed a Hypergeometric p value using python function p = hypergeom.sf(k-1, M, n, N) where M is the number of aa in the protein; n is the number of aa with greater HLA-II density than the mean (aa covered by HLA-II peptides); N is the is the number of aa with greater CD4⁺ T cell epitopes density than the mean (aa covered by T cell epitopes); and k is the number of aa with greater densities than the mean of both HLA-II peptides and CD4⁺ T cell epitopes (covered by both).

Highlights.

• Immunopeptidome analysis of SARS-CoV-2 peptides naturally presented on HLA class II

• Some HLA-II peptides originate from noncanonical SARS-CoV-2 proteins ORF9b and ORF3c

• Class I and class II HLA complexes present different subsets of viral proteins