key: cord-289711-4ab3d00h
authors: Yarmarkovich, Mark; Warrington, John M.; Farrel, Alvin; Maris, John M.
title: Identification of SARS-CoV-2 Vaccine Epitopes Predicted to Induce Long-term Population-Scale Immunity
date: 2020-06-08
journal: Cell Rep Med
DOI: 10.1016/j.xcrm.2020.100036
sha: 
doc_id: 289711
cord_uid: 4ab3d00h

Summary Here we propose a SARS-CoV-2 vaccine design concept based on identification of highly conserved regions of the viral genome and newly acquired adaptations, both predicted to generate epitopes presented on MHC class I and II across the vast majority of the population. We further prioritize genomic regions that generate highly dissimilar peptides from the human proteome, and are also predicted to produce B cell epitopes. We propose sixty-five 33mer peptide sequences, a subset of which can be tested using DNA or mRNA delivery strategies. These include peptides that are contained within evolutionarily divergent regions of the spike protein reported to increase infectivity through increased binding to the ACE2 receptor and within a newly evolved furin cleavage site thought to increase membrane fusion. Validation and implementation of this vaccine concept could specifically target specific vulnerabilities of SARS-CoV-2 and should engage a robust adaptive immune response in the vast majority of the population.

The current SARS-CoV-2 pandemic has precipitated an urgent need for a safe 43

and effective vaccine to be developed and deployed in a highly accelerated timeframe 44 as compared to standard vaccine development processes 1 immunogenicity. We present a list of SARS-CoV-2 minigenes and propose their use in 57 multivalent vaccine constructs that should generate T and/or B cell epitopes that can be 58 delivered by scalable manufacturing techniques such as DNA or nucleoside mRNA. 59 SARS-CoV-2 is the third coronavirus in the past two decades to acquire 60 infectivity in humans and result in regional epidemics, and the first to cause a global 61 pandemic. The spike glycoprotein of coronaviruses mediates host cell entry and dictates 62 species tropism, with the SARS-CoV-2 spike protein reported to bind its target ACE2 63 with 10-20-fold higher affinity than SARS-CoV in humans 2, 3 . In addition, insertion of a 64 novel protease cleavage site 4 is predicted to confer increased virulence by facilitating 65 the cleavage necessary to expose the fusion peptide that initiates membrane fusion, 66 enabling a crucial step of viral entry into host cells 5, 6 . It is now clear that COVID -19 67 disease results when SARS-CoV-2 infects type II pneumocytes lining the pulmonary 68 alveoli that co-express ACE2 and the TMPRSS2 protease (Ziegler et al., 2020), likely 69 impairing release of surfactants that maintain surface tension. This impairment hinders 70 the ability to prevent accumulation of fluid and ultimately resulting in acute respiratory 71 distress syndrome 7, 8 . The immune response of convalescent COVID-19 patients 72

consists of antibody-secreting cells releasing IgG and IgM antibodies, increased 73 follicular helper T cells, and activated CD4 and CD8 T cells 9 , suggesting that a broad 74 humoral and T cell driven immune response mediates the clearance of infection and 75 that vaccination strategies directed at multiple arms of the immune response can be 76 effective. The large size of the SARS-CoV-2 (~30 kilobases) suggests that selection of 77 optimal epitopes and reduction of unnecessary antigenic load for vaccination may be 78 essential for safety and efficacy. 79

Rapid deployment of antibody-based vaccination against SARS-CoV-2 raises the 80 concern of accelerating infectivity through Antibody-Dependent Enhancement (ADE), 81 the facilitated viral entry into host cells mediated by subneutralizing antibodies (those 82 capable of binding viral particles, but not neutralizing them) 10 . ADE mechanisms have 83 been described with other members of the Coronaviridae family 11, 12 . It has already been 84 suggested that some of the heterogeneity in COVID-19 cases may be due to ADE from 85 prior infection from other viruses in the coronavirus family 13 . 86

While the immunogenicity map presented in this study can be used to inform 87 multiple modalities of vaccine development, we present peptide sequences that are 88 expected to be safe and immunogenic for use in T cell based vaccination, and highlight 89 B cell epitopes derived from peptides within the regions of the spike protein involved in 90 infectivity that we expect will minimize the risk of ADE. As it has been shown that T 91 helper (T H ) cell responses are essential in humoral immune memory response 14, 15 , we 92 anticipate that the T cell epitopes generated from the peptide sequences presented 93 here will aid the activation of CD4 T cells to drive memory B cell formation and somatic 94 hypermutation when paired with matched B cell epitopes. 95

The potential of epitope-based vaccines to induce a cytolytic T cell response and 96 drive memory B cell formation is complicated by the diversity of HLA alleles across the 97 human population. The HLA locus is the most polymorphic region of the human 98 genome, resulting in differential presentation of antigens to the immune system in each 99 individual. Therefore, individual epitopes may be presented in a mutually exclusive 100 manner across individuals, confounding the ability to immunize a population with 101 broadly presented antigens. While T cell receptors (TCRs) recognize linearized peptides 102 anchored in the MHC groove, B cell receptors (BCRs) can recognize both linear and 103 conformational epitopes, and are therefore difficult to predict without prior knowledge of 104 a protein structure. Here we describe an approach for prioritizing viral epitopes derived 105 from a prioritized list of 33mer peptides predicted to safely target the vulnerabilities of 106 SARS-CoV-2, generate highly immunogenic epitopes on both MHC class I and II in the 107 vast majority of the population, and maximize the likelihood that these peptides will drive 108 an adaptive memory response. 109

We applied our recently published methods for scoring population-scale HLA 112 presentation of all 9mer peptides along the length of individual oncoproteins in human 113 cancer to analyze the population-scale HLA presentation of peptides derived from all 10 114 SARS-CoV-2 genes across 84 class I HLA alleles 16 Table S1 ). 121

We next tested various peptide sequence lengths to maximize HLA presentation 122 on multiple alleles within a single k-mer, finding that 33 amino acids generated maximal 123 population-scale HLA presentation. We show that 99.7% of all 9,303 possible 33mers 124 are predicted to generate at least one HLA class I epitope, and propose that expression 125 and presentation of these 33mers in dendritic cells is expected to induce an immune 126 response across a significant proportion of the population 18, 19 . We identified viral 127 regions predicted to generate epitopes that would presented across the majority of the 128 population, highlighting a single 33mer ISNSWLMWLIINLVQMAPISAMVRMYIFFASFY 129 containing multiple epitopes predicted to bind 82 of the 84 HLAs alleles, suggesting that 130 this single 33mer can potentially induce an immune response in up to 99.4% of the 131 population given proper antigen processing (Table S1) predicted to be presented on HLA class I and II across the majority of the population. As 140 HLA frequencies vary significantly by population, the frequency of individual HLA alleles 141 can be adjusted based on specific populations using the SARs-CoV-2 immunogenicity 142 map presented here, such as to customize vaccine design for groups with distinct HLA 143 allele distributions (Table S1) . 144

Next, we sought to identify the most highly conserved regions of the SARS-CoV-145 2 virus, positing that conserved regions are essential to viral replication and maintaining 146 structural integrity, while non-conserved regions can tolerate mutations and result in 147 antigens prone to immune evasion. To do this, we compared the amino acid sequence 148 of SARS-CoV-2 to 14 closely related mammalian alpha and beta coronaviruses (human, 149 bat, pig, and camel) from the Coronaviridae family (Table S2) , scoring each amino acid 150 for conservation across the viral strains. Additionally, we scored the conservation across 151 the 727 SARS-CoV-2 genes sequences available at the time of this analysis (Table S2) , 152 equally weighing contributions from cross-species and interhuman variation (scores 153 normalized to 0-1, with entirely conserved regions scoring 1). As expected, evolutionary 154 divergence was greatest in the tropism-determining spike protein and lowest in ORF1ab 155 which contains 16 proteins involved in viral replication (Figure 1B, bottom) . 156

We then compared predicted viral MHC-presented epitopes to self-peptides 157 presented in normal tissue on 84 HLA alleles across the entire human proteome as 158 listed in the UniProt database, prioritizing antigens that are most dissimilar from self-159 peptides based on: 1) higher predicted safety based on decreased likelihood of inducing 160 autoimmunity due to cross-reactivity with similar self-peptides presented on MHC; and Table S1 ). We find regions of the viral 172 proteome that are identical or highly similar to portions of the normal human proteome 173 predicted to be presented on MHC, suggesting that an immune response mounted 174 against these viral epitopes could result in an autoimmune response, while other high-175 scoring regions are highly dissimilar from self and expected to generate antigens with 176 minimal likelihood of cross-reactivity (Table S1) . 177

To assign an overall score for putative T cell antigens, we normalized each of our 178 four scoring parameters (represented in Figure 1A and 1B) between 0-1 and summed 179 each metric to obtain a final 33mer peptide score, highlighting the local maxima of 180 potentially generated epitopes scoring in the 90 th percentile (55 top scoring T cell 181 peptides) across 10 SARS-CoV-2 genes as peptide sequences for vaccination ( Figure  182 1C; Table S3) . 183

Finally, we sought to characterize B cell epitopes, assessing linear epitopes in 184 spike (S), matrix (M), and envelope (E) proteins that are exposed and expected to be 185 accessible to antibodies; we also characterized conformational epitopes in the spike 186 protein for which structural data are available using We combined T cell epitope scores calculated above with available B cell epitope 194 scores derived from the S, M, and E genes, providing a list of 65 peptides predicted to 195 stimulate both humoral and cellular adaptive immunity ( Figure 1F ; Table S5 ). 196

To estimate the accuracy of our predictions, we compared the 65 unique 33mer 197 peptides presented in Table S5 to 92 epitopes derived from the first SARS virus CoV) in the Immune Epitope DataBase (IEDB: https://www.iedb.org/home_v3.php) 199

shown to elicit T cell responses. We found a significant enrichment in immunogenic Table 1) , demonstrating that epitopes selected using this analysis epitopes are more 207 likely to be processed and immunogenic based on previous studies with SARS-CoV, 208

and supporting the hypothesis that a single 33mer is capable of generating multiple 209 unique epitopes presented by multiple HLA alleles. We also found that a significant 210

proportion of the peptides present within prioritized 33mer have been predicted to bind 211 MHC based on structural predictions 22 . bound to the spike protein from SARS-CoV-2 as compared to SARS-CoV 23 , we 230 searched for 33mers containing the five acquired residues that increase spike binding to 231 ACE2, identifying KPFERDISTEIYQAGSTPCNGVEGFNCYFPLQS as the highest 232 ranked peptide sequence containing each of these residues (hotspots underlined; Table  233 1). Additionally, a D614G mutation in the spike protein has been reported as a 234

potentially dominant strain with increased pathogenicity 24, 25 . We thus suggest including 235 the highest scoring 33mer (NTSNQVAVLYQGVNCTEVPVAIHADQLTPTWRV) 236 predicted to present this mutant epitope in a vaccine construct. Finally, it is known that 237 mRNA transcripts proximal to the 3' end of the Coronaviridae family genome show 238 higher abundance consistent with the viral replication process, with S, E, M, and N 239 genes shown to have significantly higher translational efficiency compared to the 5' 240 transcripts, with the highest expression in the N gene, and consistent with the high 241 degree of MHC presentation as described above five immunogenic peptides derived 242 from a single N protein 33mer [26] [27] [28] . We therefore posit that viral epitopes derived from 3' 243 terminus including the S, E, M, and N genes will have a higher representation on MHC 244 and suggest their prioritization in a vaccine construct. Table S5 lists the highest priority 245 viral peptides we suggest should be considered for inclusion in vaccine constructs. 246

Here we present a comprehensive immunogenicity map of the SARS-CoV-2 248 virus (Table S1) , and propose sixty-five 33mer peptide sequences predicted to generate 249 B and T cell epitopes from a diverse sampling of viral domains across all 10 SARS-250

CoV-2 genes (Tables 1 and S5 ). Based on our computational algorithms, we expect 251 that the highest scoring peptides will result in safe and immunogenic T cell epitopes, 252

and that B cell epitopes should be evaluated for safety and efficacy using previously 253 reported methods with validated subsets of these 65 epitopes 11 . DNA and mRNA 254 vaccines have been shown to be safe and effective in preclinical studies, and can be 255 rapidly and efficiently manufactured at scale 29, 30 . Nucleoside modification of RNA has 256 been shown to improve efficacy, which has been attributed to a reduction of RNA-257 induced immunogenicity 31 . We suggest that multivalent constructs composed of the 258 SARS-CoV-2 minigenes encoding subsets of the B and/or T cell epitopes proposed 259 here (Tables 1, S3, S4, and S5) can be used in a DNA on mRNA vaccine for 260 expression in antigen-presenting cells. 261

These epitopes can be used in tandem with a TLR agonist such as tetanus 262 toxoid or PADRE [32] [33] [34] [35] to drive activation of signals 1 and 2 in antigen presenting cells. 263

Constructs can be designed to contain a combination of optimal B and/or T cell 264 epitopes, or deployed as a construct consisting of the top scoring T cell epitopes to be 265 used in combination with the vaccines currently being developed targeting spike protein 266 in order to drive the adaptive memory response. DNA vaccine sequences can also be 267 codon optimized to increase CpG islands such as to increase TLR9 activation 36 . 268

With the third epidemic in the past two decades underway, all originating from the 269 coronavirus family, these viruses will continue to threaten the human population, and Table S1 can be used to design 281 customize multi-valent vaccines based on the HLA frequencies of specific populations. 282

Although we suggest the use of 33mers based on optimal MHC presentation across the 283 population, these methods can be generalized and applied to the evaluation k-mers of 284 various sizes depending on desired application. Since antigens may arise from the 285 junctions between epitopes, the analyses presented here can also be used to evaluate 286 epitope generation at the junction of specific vaccine constructs, such as to engineer 287 linker regions that reduce the potential immunodominant epitopes elicited from 288 irrelevant sequences. We also find up to five peptides reported by Grifoni within a single 33mer, and up to 12 302 peptides reported by Ahmed contained in the 33mer 303 AQFAPSASAFFGMSRIGMEVTPSGTWLTYTGAI described above. Taken together, 304 these comparisons show a significant convergence on a subset of epitopes using 305 agnostic analyses, while also reporting unique epitopes in each study. The finding that 306 up to 12 epitopes from previous analyses are represented in a single 33mer from our 307 agnostic analysis further supports the our prediction that cocktails of 33mer epitopes 308 can be used for population-scale vaccination. 309

By narrowing the pool of peptides selected for downstream screening, we expect 310 that the analyses presented here will contribute to maximizing the efficiency of vaccine 311 development. Antigenic burden from epitopes that do not contribute to viral protection 312 can cause autoimmune reactions, reactogenicity, detracting from the efficacy of the 313 vaccine, or result in ADE. We found that the vast majority of the SARS-CoV-2 virus is 314 immunogenically silent on MHC class I and II and suggest these regions should be 315 excluded from vaccine development. Though empirical testing is necessary to evaluate 316 ADE, we suggest that antibodies directed at the receptor binding domain and furin 317 cleavage sequences may mitigate ADE by blocking the processes needed to achieve 318 membrane fusion. To avoid potential T cell cross-reactivities a priori, we selected 319 maximally immunogenic epitopes with the highest degree of dissimilarity to the self-320 proteome with minimal potential of cross-reactivity that can lead to adverse reaction or 321 weaken the efficacy of vaccination. In addition to the predicted safety of these epitopes 322 (stemming from lack of potentially cross-reactive normal proteins), we expect that a 323 greater repertoire of viral antigen-specific T cells will be present due to the absence of and symptoms of anosmia 42, 43 . While CD8 vaccines targeting conserved antigens in 339 influenza did not completely block infection upon challenge with virus, they effectively 340 reduced viral replication, morbidity, and mortality 44, 45 . Taken together, these findings 341 suggest that CD8-based immunity can be a viable strategy in quelling SARS-CoV-2.

Studies demonstrating protection against multiple influenza strains imply that CD8-343 mediated vaccination may act more broadly than antibody responses in protecting 344 against multiple virus family members through targeting of conserved non-structural 345 proteins critical in the viral life-cycle. 346

Currently, targeting CD8 epitopes has been complicated by HLA restriction of 347 peptides and antigenic drift resulting from viral regions in which mutation is tolerable. concordance with predicted population-scale presentation 16 . Although we expect that a 366 significant fraction of predicted antigens to be presented on MHC, binding predictions 367 alone do not determine which antigens will elicit an immunodominant response. While 368 the dissimilarity scoring predicts that TCRs specific for these antigens are more likely to 369 exists (because these TCRs far less likely to have undergone negative thymic 370 selection), these predictions are confounded by the TCR repertoire of a given individual 371 and the intrinsic immunogenicity of a particular peptide which cannot be predicted 372 without empirical testing. Since MHC binding is a prerequisite for antigen 373 immunogenicity, we expect that immunodominant antigens will be contained within our 374 highest scoring epitopes. However, experimental validation will be necessary to 375 determine the contribution of individual antigens to immunity. As a best approximation 376 for our predictions, we show a significant enrichment of peptides previously reported in 377 IEDB to be immunogenic in the SARS-CoV virus, contained within the 65 prioritized 378 epitopes that we present, supporting the concept that multiple antigens derived from 379 33mers can be presented across multiple HLA alleles. 380

We expect that the comprehensive immunogenicity map presented here can be 381 used by the scientific community to inform the design of various vaccination modalities. 382

We are presently designing a set of vaccine vectors and validation reagents based on 383 these analyses that we plan to make available to the research community for testing. 384

The 65 epitopes presented here out of the 9,303 possible 33mers derived from SARS-385

CoV-2 can significantly narrow the focus of vaccine development (Table S5) ; these 386 epitopes can be expressed as a single <7kb construct, or more likely tested in various 387 combinations delivered as a cocktail of RNA constructs encoding individual 33mers. 388

These vaccine constructs can be rapidly and efficiently tested for the neutralizing 389 potential of antibodies using SARS-CoV-2 pseudo-virus 47 , the formation of memory B 390 cells, and for induction of T cell activation using methods that we have recently 391 developed for interrogating antigen specificities in a highly multiplexed manner 48 . As 392 SARS-CoV-2 has precipitated the need to rapidly develop and deploy vaccines in 393 pandemic situations 49 , we suggest that this comprehensive analysis can be 394 incorporated into a process that can be rapidly implemented when future novel viral 395 pathogens emerge. 396

The in silico analysis of the SARS-CoV-2 genome reported here has yet to be 398 experimentally validated. While it is reassuring that we demonstrate enrichment of 399 predicted epitopes from the original SARS virus previously reported in IEDB that have 400 been shown to be immunogenic, rigorous experimental validation of our findings is 401 required. Computational pMHC binding predictions do not consider critical variables in 402 antigen presentation such as proteasomal degradation and peptide processing. In 403 addition, it is unclear whether the 33mers designed to elicit a B cell will properly fold into 404 conformations resembling the native spike protein such as to elicit a protective antibody 405 response. We have designed multiple DNA and mRNA constructs containing 406 combinations of 33mers proposed here to test hypotheses that these vaccines can elicit 407 memory and/or cytolytic T cell response and/or protective antibodies against a SARS-408

Spike-GFP pseudovirus 47 in HLA-A2 transgenic mice 50 the population predicted to have at least one of these HLAs, normalized dissimilarity scores, normalized conservation 466 scores, across the 33mer, total T cell score, B cell score, and combined B and T cell percentile for 33mers. Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, John M. Maris (maris@email.chop.edu).

Vaccine constructs and testing reagents are available from the Lead Contact, John M. Maris with a completed Materials Transfer Agreement. Please email maris@chop.edu.

All raw data has been reported in paper and models are described in methods.

We identified potential SARS-CoV-2 epitopes by applying our recently published algorithm for scoring population-scale HLA presentation of tumor driver gene, to the SARS-CoV-2 genome (GenBank Acc#: MN908947.3) 51 . All possible 33mer amino acid sequences covering every 9mer peptide from the 10 SARS-CoV-2 genes were generated and we employed netMHC-4.0 to predict the binding affinities of each viral 9mer peptide across 84 HLA class I alleles 52 . We considered 9mer peptides with binding affinities <500nM putative epitopes. MHC class II binding affinities were predicted as described above across 36 HLA class II alleles population using netMHCII 2.3 53 . All 9mers present in a 33mer contribute to the score. 33mer scores calculated by infering population scale hla presentation of all predicted peptides within 9mer on class I and ii.

The frequencies of HLA class I alleles -A/B/C and HLA class II alleles -DRB1/3/4/5 were obtained from Be the Match bone marrow registry 17 .HLA class II alleles -DQA1/DQB1 and -DPA1/DPB1 were obtained from 54 and 55 , respectively.

We obtained all 727 unique protein sequences categorized by each of the 10 SARS-CoV-2 genes available from the NCBI as of 25 March 2020. All sequences were aligned using Clustal Omega 56 and each position summed for homology. In addition to human sequences, we scored each amino acid position for homology across 15 species of related coronavirus found in bats, pigs, camels, mice, and humans (SARS-CoV, SARS-CoV-2, and MERS). Each amino acid was scored up to 100% conservation. 33mer peptides were then scored in Equation 1:

Where C is the 33mer conservation score, A is the conservation percentage of an amino acid position, Y is the minimum 33mer conservation percentage sum, and Z is the maximum 33mer conservation percentage sum. In the same way, we ranked the conservation across 274 SARS-CoV-2 amino acid sequences available at the time of this study. A final conservation score was generated by averaging the conservation scores from cross-species and interhuman variation and 33mer peptides with the highest score were considered the most conserved.

3,524 viral epitopes were compared against the normal human proteome on each of their MHC binding partners, testing a total of 12, 383 peptide/MHC pairs against the entire human proteome (85,915,364 normal peptides across HLAs), assigning a similarity score for each peptide. Residues in the same position of the viral and human peptides with a perfect match, similar amino acid classification, or different polarity, were assigned scores of five, two, or negative two respectively. Similarity scores were calculated based on amino acid classification and hydrophobicity were determined using non-anchor residues on MHC ( Figure S1A) . The canonical TCR-interaction hotspots (residues four through six) were double weighted [57] [58] [59] . The similarity scores generated for each viral peptide were converted to Z-scores and peptides with a p <0.0001 were selected for comparison to viral epitopes ( Figure S1B ). The overall dissimilarity score for the viral peptide was then calculated using Equation 2:

where is the overall dissimilarity score for the viral peptide, is the highest possible Z-score given a perfect sequence match to the viral peptide, is the highest Z-score from the human proteome, % & is the number of statistically significant peptides from the human proteome, and & is the mean Z-score from the statistically significant peptides given a p < 0.001.

We used BepiPred 2.0 and DiscoTope 2.0 20,21 to score individual amino acid residues, assessing linear epitopes in Matrix, Envelope, and Spike proteins, and conformational epitopes for Spike protein, based on published structure (PDB 6VYB). To we summed and normalized linear and conformational, using separate normalizations for proteins in which only linear predictions were available. Table Legends   Table S1 . Immunogenicity map of SARS-CoV-2, related to Figure 1 and Table 1 .

Combined analysis of SARS-CoV-2 for HLA class I & II binding, dissimilarity scoring, and conservation scoring for peptides beginning at each position in the viral proteome.

HLA binding reported as percent rank of epitopes and 33mers beginning at each position, with a <=0.5 threshold for listed binders (strong binders). Dissimilarity, conservation and B cell (linear and conformational) scores listed for 33mers starting at each peptide. Table S2 . Sequences used for conservation scoring, related to Figure 1 and Table   1 . Sequences for 14 alpha and beta coronavirus genes and 727 SARS-CoV-2 genes. Table S3 . Top Scoring SARS-CoV-2 T Cell 33mers, related to Figure 1 and Table 1 .

Fifty-five highest scoring 33mer peptides based on a combination of HLA class I, class II, dissimilarity, and homology scoring (contains peptides found to be B cell epitopes as well). Table S5 . Prioritized list of 65 33mer peptide sequences enriched for population scale immunity, related to Figure 1 and Table 1 .

Yarmarkovich et al. report SARS-CoV-2 peptides for use in multi-epitope vaccines. These peptides are predicted to activate CD4 and CD8 T cells, are highly dissimilar from the self-proteome, and are conserved across 15 related coronaviruses. Presented epitopes are expected to drive long-term immunity in the majority of the population.

• Selecting optimal epitopes is essential for vaccine safety and efficacy

• We report 65 vaccine peptides predicted to drive long-term immunity in most people

• Epitopes contain domains conserved in 15 coronaviruses and newly evolved SARS2 regions

• Epitopes can be used to generate B and/or T cell vaccines (RNA and DNA)

Middle East respiratory syndrome: obstacles and prospects for vaccine development

Cryo-EM structure of the 2019-nCoV spike in the prefusion conformation

Structure, Function, and Antigenicity of the SARS-CoV-2 Spike Glycoprotein

The spike glycoprotein of the new coronavirus 2019-nCoV contains a furin-like cleavage site absent in CoV of the same clade

Structure of the Hemagglutinin Precursor Cleavage Site, a Determinant of Influenza Pathogenicity and the Origin of the Labile Conformation

Activation of the SARS coronavirus spike protein via sequential proteolytic cleavage at two distinct sites

Histopathologic Changes and SARS-CoV-2 Immunostaining in the Lung of a Patient With COVID-19

High expression of ACE2 receptor of 2019-nCoV on the epithelial cells of oral mucosa

Breadth of concomitant immune responses prior to patient recovery: a case report of non-severe COVID-19

Dengue virus sero-cross-reactivity drives antibody-dependent enhancement of infection with zika virus

Immunodominant SARS Coronavirus Epitopes in Humans Elicited both Enhancing and Neutralizing Effects on Infection in Non-human Primates

Molecular Mechanism for Antibody-Dependent Enhancement of Coronavirus Entry

Is COVID-19 receiving ADE from other coronaviruses?

MHC-II neoantigens shape tumour immunity and response to immunotherapy

Molecular programming of B cell memory

Immunogenicity and Immune Silence in Human Cancer

Six-locus high resolution HLA haplotype frequencies derived from mixed-resolution DNA typing for the entire US donor registry

Quantitative and qualitative analyses of the immune responses induced by a multivalent minigene DNA vaccine

Efficient identification of mutated cancer antigens recognized by T cells associated with durable tumor regressions

BepiPred-2.0: improving sequence-based B-cell epitope prediction using conformational epitopes

Reliable B Cell Epitope Predictions: Impacts of Method Development and Improved Benchmarking

Structure-based modeling of SARS-CoV-2 peptide/HLA

Structural basis of receptor recognition by SARS-CoV-2

SARS-CoV-2 viral spike G614 mutation exhibits higher case fatality rate

Spike mutation pipeline reveals the emergence of a more transmissible form of SARS-CoV-2

High-Resolution Analysis of Coronavirus Gene Expression by RNA Sequencing and Ribosome Profiling

Severe Acute Respiratory Syndrome Coronavirus as an Agent of Emerging and Reemerging Infection

Quantification of individual subgenomic mRNA species during replication of the coronavirus transmissible gastroenteritis virus

Rapid development of a DNA vaccine for Zika virus

Modified mRNA Vaccines Protect against Zika Virus Infection

Zika virus protection by a single low-dose nucleosidemodified mRNA vaccination

The optimization of helper T lymphocyte (HTL) function in vaccine development

DNA Vaccines Encoding Ii-PADRE Generates Potent PADRE-specific CD4+ T-Cell Immune Responses and Enhances Vaccine Potency

C4 DNA vaccine with fragment C of Tetanus toxin increases protective immunity against CEA-expressing tumor

Clinical evaluation of safety and immunogenicity of PADRE-cytomegalovirus (CMV) and tetanus-CMV fusion peptide vaccines with or without PF03512676 adjuvant

Toll-like receptor 9 (TLR9) agonists in the treatment of cancer

Preliminary Identification of Potential Vaccine Targets for the COVID-19 Coronavirus (SARS-CoV-2) Based on SARS-CoV Immunological Studies

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

Mechanism of Protective Immunity Against Influenza Virus Infection in Mice Without Antibodies

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

Reduction of influenza virus transmission from mice immunized against conserved viral antigens is influenced by route of immunization and choice of vaccine antigen

Olfactory Bulb Magnetic Resonance Imaging in SARS-CoV-2-Induced Anosmia: The First Report

Smell and taste dysfunction in patients with COVID-19. The Lancet Infectious Diseases

Reduction of influenza virus transmission from mice immunized against conserved viral antigens is influenced by route of immunization and choice of vaccine antigen

Nsp3 of coronaviruses: Structures and functions of a large multi-domain protein

Characterization of spike glycoprotein of SARS-CoV-2 on virus entry and its immune cross-reactivity with SARS-CoV

High throughput pMHC-I tetramer library production using chaperone-mediated peptide exchange

Developing Covid-19 Vaccines at Pandemic Speed

A multivalent vaccination strategy for the prevention of Old World arenavirus infection in humans

Immunogenicity and Immune Silence in Human Cancer

Gapped sequence alignment using artificial neural networks: application to the MHC class I system

Improved methods for predicting peptide binding affinity to MHC class II molecules

Divergent motifs but overlapping binding repertoires of six HLA-DQ molecules frequently expressed in the worldwide human population

Balancing selection and heterogeneity across the classical human leukocyte antigen loci: a meta-analytic review of 497 population studies

Fast, scalable generation of high-quality protein multiple sequence alignments using Clustal Omega

Structural Bases for the Affinity-Driven Selection of a Public TCR against a Dominant Human Cytomegalovirus Epitope

The Structural Dynamics and Energetics of an Immunodominant T Cell Receptor Are Programmed by Its V&#x3b2; Domain

Unraveling a Hotspot for TCR Recognition on HLA-A2: Evidence Against the Existence of Peptide-independent TCR Binding Determinants