key: cord-312414-g5px0b65
authors: Takagi, Akira; Matsui, Masanori
title: An immunodominance hierarchy exists in CD8+ T cell responses to HLA-A*02:01-restricted epitopes identified from the non-structural polyprotein 1a of SARS-CoV-2
date: 2020-09-19
journal: bioRxiv
DOI: 10.1101/2020.09.18.304493
sha: 
doc_id: 312414
cord_uid: g5px0b65

COVID-19 vaccines are being rapidly developed and human trials are underway. Almost all of these vaccines have been designed to induce antibodies targeting spike protein of SARS-CoV-2 in expectation of neutralizing activities. However, non-neutralizing antibodies are at risk of causing antibody-dependent enhancement. Further, the longevity of SARS-CoV-2-specific antibodies is very short. Therefore, in addition to antibody-induced vaccines, novel vaccines on the basis of SARS-CoV-2-specific cytotoxic T lymphocytes (CTLs) should be considered in the vaccine development. Here, we attempted to identify HLA-A*02:01-restricted CTL epitopes derived from the non-structural polyprotein 1a of SARS-CoV-2. Eighty-two peptides were firstly predicted as epitope candidates on bioinformatics. Fifty-four in 82 peptides showed high or medium binding affinities to HLA-A*02:01. HLA-A*02:01 transgenic mice were then immunized with each of the 54 peptides encapsulated into liposomes. The intracellular cytokine staining assay revealed that 18 out of 54 peptides were CTL epitopes because of the induction of IFN-γ-producing CD8+ T cells. In the 18 peptides, 10 peptides were chosen for the following analyses because of their high responses. To identify dominant CTL epitopes, mice were immunized with liposomes containing the mixture of the 10 peptides. Some peptides were shown to be statistically predominant over the other peptides. Surprisingly, all mice immunized with the liposomal 10 peptide mixture did not show the same reaction pattern to the 10 peptides. There were three pattern types that varied sequentially, suggesting the existence of an immunodominance hierarchy, which may provide us more variations in the epitope selection for designing CTL-based COVID-19 vaccines. Importance For the development of vaccines based on SARS-CoV-2-specific cytotoxic T lymphocytes (CTLs), we attempted to identify HLA-A*02:01-restricted CTL epitopes derived from the non-structural polyprotein 1a of SARS-CoV-2. Out of 82 peptides predicted on bioinformatics, 54 peptides showed good binding affinities to HLA-A*02:01. Using HLA-A*02:01 transgenic mice, 18 in 54 peptides were found to be CTL epitopes in the intracellular cytokine staining assay. Out of 18 peptides, 10 peptides were chosen for the following analyses because of their high responses. To identify dominant epitopes, mice were immunized with liposomes containing the mixture of the 10 peptides. Some peptides were shown to be statistically predominant. Surprisingly, all immunized mice did not show the same reaction pattern to the 10 peptides. There were three pattern types that varied sequentially, suggesting the existence of an immunodominance hierarchy, which may provide us more variations in the epitope selection for designing CTL-based COVID-19 vaccines.

In December 2019, the coronavirus disease 2019 caused by the severe 64 acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was firstly identified in Wuhan, 65

Hubei province, China. Since then, its subsequent spread of global infection has still 66 continued to gain momentum. As of September 16th, 2020, the COVID-19 pandemic has 67 infected more than 29.4 million people around the world and caused more than 931,000 68 deaths. Although the clinical symptom is varied from asymptomatic or mild self-limited 69 infection to severe life-threating respiratory disease, the mechanism of disease outcome 70 remains unclear. Many nations are struggling to find appropriate preventive and control 71 strategies. However, there are no vaccines or antiviral drugs available for the treatment of 72 this infectious disease. 73

There are seven coronaviruses that infect humans. In addition to SARS-CoV-2, 74 SARS-CoV and middle-east respiratory syndrome coronavirus (MERS-CoV) cause 75 severe pneumonia, whereas the other four human coronaviruses including HCoV-229E, 76 -NL63, -OC43 and -HKU1 cause common cold (1). Like other coronaviruses, 77 SARS-CoV-2 possesses a large single-stranded positive sense RNA genome (2). As 78 shown in Fig. 1 , the 5'-terminal two-thirds of the genome are composed of ORF1a and 79

ORF1b. ORF1a encodes the polyprotein 1a (pp1a) containing non-structural proteins 80

with peptide-encapsulated liposomes. 166

Each of 54 peptides selected were encapsulated into liposomes as described in the 167 materials and methods. HLA-A*02:01 transgenic (HHD) mice (37) were then 168 subcutaneously (s.c.) immunized twice at a one-week interval with each of 169 peptide-encapsulated liposomes together with CpG adjuvant. One week later, spleen cells 170 of immunized mice were prepared, stimulated in vitro with a relevant peptide for 5 hours, 171 and stained for their expression of cell-surface CD8 and intracellular interferon-gamma 172 (IFN-. As shown in Fig. 2 , the intracellular cytokine staining (ICS) assay showed that 173 significant numbers of IFN--producing CD8+ T cells were elicited in mice immunized 174 with 18 liposomal peptides including pp1a-38, -52, -84, -103, -445, -597, -641, -1675, 175 -2785, -2884, -3083, -3403, -3467, -3583, -3662, -3710, -3732, and -3886, revealing that 176 these 18 peptides are HLA-A*02:01-restricted CTL epitopes derived from SARS-CoV-2 177 pp1a. As indicated in Table 2 , multiple epitopes are located in small proteins such as nsp1 178 (180 aa) and nsp6 (290 aa), whereas only one epitope is seen in the large nsp3 composed 179 of 1945 amino acids. On the other hand, the remaining 36 peptides out of 54 peptides in 180 liposomes were not able to stimulate peptide-specific CTLs in mice (data not shown), 181 demonstrating the necessity to generate data through wet-lab experiments. Interestingly, 182 four epitopes including pp1a-103, -2884, -3403, and -3467 are located in the amino acid 183 sequence of SARS-CoV pp1a as well (Table 2) . pp1a-3467 was previously reported by us 184 in the identification of SARS-CoV-derived CTL epitopes (30). However, any of 18 185 epitopes are not found in the amino acid sequence of either MERS-CoV or the four 186 common cold human coronaviruses involving HCoV-OC43, In the 18 positive peptides, 10 peptides including pp1a-38, -84, -641, -1675, -2884, 189 -3467, -3583, -3662, -3710, and -3732 were selected for the following analyses because 190 of the high ratios of IFN- + cells in CD8 + T cells (Fig. 2) . 191 192

To confirm that the 10 peptides are effective epitopes for peptide-specific CTL 194 responses, we examined whether peptide-specific killing activities were elicited in mice 195 with each of 10 peptides in liposomes. HHD mice were immunized s.c. twice with each 196 of peptide-encapsulated liposomes and CpG adjuvant. One week later, equal numbers of 197 peptide-pulsed and -unpulsed target cells were transferred into immunized mice via i.v. 198 injection, and peptide-specific lysis was analyzed by flow cytometry. In support of the 199 data of ICS (Fig. 2) , peptide-specific killing was observed in mice immunized with any of 200 10 liposomal peptides (Fig. 3A) . 201

We next attempted to identify dominant CTL epitopes among the 10 CTL epitopes. 202

The same amounts of the 10 peptide solutions at an equal concentration were mixed 203 together and encapsulated into liposomes. Seventeen mice were immunized once with the 204 liposomes containing the mixture of 10 peptides. One week later, spleen cells were 205 incubated with each of the 10 peptides for 5 hours, and the ICS assay was performed. As 206 shown in Fig. 3B & C, pp1a-38 and -84 were statistically predominant over almost all 207 other peptides in the induction of peptide-specific IFN- + CD8 + T cells. Furthermore, 208 pp1a-641 and pp1a-3732 were significantly superior to pp1a-1675/-3583 and 209 pp1a-1675/-3583/-3662 in the stimulation of IFN--producing CD8 + T cells, respectively 210 ( Fig. 3B) . 211

We also examined the peptide-specific induction of CD107a + CD8 + T cells and CD69 + 212 CD8 + T cells. CD107a and CD69 are markers of degranulation and early activation on 213 CD8 + T lymphocytes, respectively. Nine mice were immunized once with the liposomes 214 encapsulating the mixture of the 10 peptides. After one week, spleen cells were 215 stimulated with each peptide, and stained for their expression of CD107a or CD69 of 216 CD8 + T cells. At first glance, the graphs of CD107a ( Taken together, 10 peptides differed significantly in their ability to induce 226 SARS-CoV-2 pp1a-specific CTLs when mice were immunized with the mixture of 10 227 peptides in liposomes. Thus, some peptides were found to be dominant CTL epitopes 228 although each peptide alone of the 10 peptides has the capability to efficiently activate 229 peptide-specific CTL (Figs. 2 & 3A) . 230 231 Existence of an immunodominance hierarchy. 232

The data in Fig. 5 indicate reactivity to the 10 peptides in each of 15 mice immunized 233 with the mixture of the 10 peptides in liposomes. Each graph represents reactivity of an 234 individual mouse (Fig. 5) . Unexpectedly, all mice did not show the same reaction pattern 235 against the 10 peptides. It looks like there were roughly three types that varied 236 sequentially in terms of the reaction pattern to the 10 peptides. Type A is a group of mice 237 in which pp1a-38, -84, -641-specific IFN- + CD8 + T cells were predominantly elicited, 238 whereas the remaining 7 peptides were not able to activate peptide-specific IFN- + CD8 + 239 T cells. In the case of type B, pp1a-3732 stimulated peptide-specific IFN- + CD8 + T cells 240 as well as pp1a-38, and -84. In addition to these three peptides, several other peptides also 241 induced IFN- + CD8 + T cells in Type C. These data suggest that there seems to be an 242 immunodominance hierarchy composed of three stages in CD8 + T cell responses to the 10 243 peptides. The immunodominance hierarchy may provide us more variations for designing 244 CTL-based COVID-19 vaccines. 245 246

After the epidemics of SARS and MERS, scientists have not succeeded yet in 248 developing preventive or therapeutic vaccines available for re-emergence of them. In the 249 SARS and MERS vaccine development, the full-length S protein or its S1 subunit have 250 frequently been used as an antigen to produce anti-RBD neutralizing antibodies. 251

However, these vaccine candidates provided partial protection against virus challenge in 252 animal models accompanied by safety concerns such as ADE (1). Furthermore, antibody 253 responses to coronaviruses rapidly wane following infection or immunization (23, 24, 254 26). Considering the above, it should be necessary to consider CTL-based vaccine against 255 SARS-CoV-2 to provide robust long-lived T cell memory although neutralizing antibody 256 responses are a primary vaccine target. 257

In the current study, we aimed to find HLA-A*02:01-resctricted CTL dominant 258 epitopes derived from SARS-CoV-2. Dominant epitopes induce strong immune response 259 to eliminate a certain pathogen fast and effectively, and also contribute to make the 260 memory T cell pool. We focused on pp1a of SARS-CoV-2 to find out CTL epitopes 261 because pp1a is a largest and conserved polyprotein among the constituent proteins. 262

Further, pp1a is produced earlier than structural proteins, suggesting that pp1a-specific 263

CTLs can eliminate infected cells before the formation of mature virions. To predict CTL 264 epitopes, we utilized bioinformatics to select 82 peptides with high scores in four kinds of 265 computer-based programs (Table 1 ). In the evaluation of peptide binding, 54 peptides 266 showed high or medium binding affinities to HLA-A*02:01 molecules, whereas the 267 14 remaining 28 peptides displayed low binding affinities or no binding (Table 1) . Out of 268 them, only eighteen peptides were found to be CTL epitopes. Hence, we have to keep in 269 mind that currently available algorithms have a limited ability to accurately predict CTL 270 epitopes although the bioinformatics approach is very useful to quickly predict a number 271 of epitopes in a large protein (38, 39) . 272

Among 18 epitopes which we have identified in the current study, 4 epitopes including 273 pp1a-103, -2884, -3403, and -3467 are present in the amino acid sequence of SARS-CoV 274 pp1a (100% identity) (Table 2) . Therefore, CTLs induced by these four epitopes could 275 work fine for the clearance of SARS-CoV as well. In support of this data, Le Bert et al., 276 reported that long lasting memory T cells in SARS-recovered individuals cross-reacted to 277 N protein of SARS-CoV-2 (40). Recently, several studies found that 278 SARS-CoV-2-reactive CD4 + and CD8 + T cells were detected in a substantial proportion 279 of healthy donors who have never infected with SARS-CoV-2 or SARS-CoV (40-44). It 280 is most likely that these individuals were previously infected with one of the four human 281 coronaviruses (HCoV-229E, -NL63, -OC43 and -HKU1) that cause seasonal common 282 cold. Nelde et al. demonstrated evidence that the amino acid sequences of several 283 SARS-CoV-2 T cell epitopes recognized by unexposed individuals are similar to some 284 amino acid sequences in the four seasonal common cold human coronaviruses with 285 identities ranging from 10% to 89% (not 100% identity) (44). However, anyone has not 286 shown evidence that people with this cross-reactivity are less susceptible to It may be also possible to assume that pre-existing T cell immunity might be detrimental 288 through mechanisms such as original antigenic sin or ADE (45). In the current data, any 289 of the 18 epitopes was not found in the amino acid sequences of the four human 290 coronaviruses, suggesting that effective common CTL epitopes derived from pp1a, if any, 291 might be very few. 292

Here, we focused on CTL epitopes restricted by HLA-A*02:01 which is the most 293 common HLA class I allele in the world, and used highly reactive HLA-A*02:01 294 transgenic mice, termed HHD mice (37). Although we can use lymphocytes of 295 SARS-CoV-2-infected individuals to identify CTL epitopes, there are mainly two 296 advantages to using HHD mice. First, a large amount of blood of patients is required for 297 examine many candidates of CTL epitopes, but any number of mice can be prepared for 298 this purpose. Second, when using patients' lymphocytes, we are only testing whether the 299 peptide candidates are recognized by memory CTLs. When using naïve mice, however, 300 we can find whether the epitope candidates are able to prime peptide-specific CTLs, 301 which may be a better criterion to judge them as vaccine antigens. It is supposed that the 302 efficient epitope for CTL recognition is not always efficient for CTL priming. However, 303

we should take into account that the immunogenic variation in HLA class I transgenic 304 mice may not be identical to that in humans because the antigen processing and 305 presentation differ between them. 306

In the previous studies, we used peptide-linked liposomes as an immunogen (30). The 307

surface-linked liposomal peptides were effective for peptide-specific CTL induction in 308 mice. However, attaching peptides to the surface of liposomes followed by purifying 309 16 them through the column is a fairly complicated process and time-consuming. In the 310 current study, peptide-encapsulated liposomes were used as an immunogen. In contrast to 311 the peptide-linked liposomes, the peptide-encapsulated liposomes are prepared by just 312 mixing liposomes and the peptide. In addition, the peptide-encapsulated liposomes are 313 able to prime peptide-specific CTLs in mice as efficiently as the peptide-linked 314 liposomes. Liposome itself consisting of lipid bilayers is a very safe material for humans. 315 Therefore, the peptide-encapsulated liposomes are considered to be promising as a 316 CTL-based vaccine candidate. 317

Understanding the mechanism of immunodominance is obviously important for the 318 development of effective vaccines. When mice were immunized with liposomes 319 containing the mixture of 10 peptides, it was found that some peptides induced 320 peptide-specific CTLs stronger than other peptides (Figs. 3 & 4) . As shown in Figs Table I , the peptide affinity 326 of pp1a-84 to HLA-A*02:01 is very high (BL 50 = 6.8 M), while pp1a-38 is a medium 327 binder (BL 50 = 76.8 M). Interestingly, the peptide affinity of pp1a-38 is lowest among 328 the 10 peptides selected (Table I) . These data suggest that the affinity of TCR to the 329 peptide-MHC-I complex is critical for CTL immunodominance. In the selection of 330 antigenic epitopes for the CTL-based vaccine against SARS-CoV-2, dominant epitopes 331 such as pp1a-38 and -84 should be chosen because they produce strong CTL response to 332 eliminate virus-infected cells effectively. However, the immunological pressure exerted 333 by dominant epitopes may allow the epitope sequences of SARS-CoV-2 to be mutated, 334 and therefore, a vaccine containing multiple antigenic epitopes should be recommended 335 for a successful COVID-19 vaccine. 336

It was surprising that all of the genetically identical mice did not show the same 337 reactive pattern against the 10 peptides when they were immunized with liposomes 338 containing the mixture of 10 peptides (Fig. 5 ). There were roughly three pattern types, 339

A-C, that varied sequentially, suggesting the existence of an immunodominance 340 hierarchy composed of three stages in CD8 + T cell responses to the 10 peptides (Fig. 5) . 341

The differences among the three types might be explained by the timing of CTL 342 expansion. In the type A, dominant peptides, pp1a-38, -84, and -641 presumably 343 activated T cells more efficiently than the other peptides, and hence, dominant 344 peptide-specific CTLs proliferate faster and curtail the expansion of CTLs specific for the 345 other peptides. In the type B, it is considered that the expansion of dominant CTLs 346 specific for pp1a-38, and -84 was delayed for some reason compared to that in type A, 347 and thereby subdominant CTLs specific for pp1a-3732 could afford to expand. It is also 348 thought that even non-dominant CTLs proliferated because the expansion of both 349 dominant CTLs and subdominant CTLs in the type C was later than that in the type B. In vivo CTL assay was carried out as described (46). In brief, spleen cells from naive 450 HHD mice were equally split into two populations. One population was pulsed with 50 451 M of a relevant peptide and labeled with a high concentration (2.5 M) of 452 carboxyfluorescein diacetate succinimidyl ester (CFSE) (Molecular Probes, Eugene, OR). 453

The other population was unpulsed and labeled with a lower concentration (0.25 M) of 454

CFSE. An equal number (1 × 10 7 ) of cells from each population was mixed together and 455 adoptively transferred i.v. into mice that had been immunized once with a liposomal 456 peptide. Sixteen hours later, spleen cells were prepared and analyzed by flow cytometry. 457

To calculate specific lysis, the following formula was used: % specific lysis = 458 immunized twice with either each of the 10 liposomal peptides (pp1a-38, -84, -641, 705 -1675, -2884, -3467, -3583, -3662, 3710, and -3732) or liposomes alone together with 706 CpG. One week later, in vivo peptide-specific killing activities were measured. Three to 707 five mice per group were used, and the data of % specific lysis are shown as the mean ± 708 

Lessons for COVID-19 immunity from other 479 coronavirus infections

Structural genomics of SARS-CoV-2 indicates evolutionary 482 conserved functional regions of viral proteins

Structure, function, and antigenicity of the SARS-CoV-2 spike glycoprotein

SARS-CoV-2 cell entry depends on ACE2 and TMPRSS2 and is blocked by a 490 clinically proven protease inhibitor

Breadth of concomitant immune responses prior to patient recovery: a case report of 494 non-severe COVID-19

Development of an inactivated vaccine candidate for 499 SARS-CoV-2

NVX-CoV2373 vaccine protects cynomolgus macaque upper and lower 503 airways against SARS-CoV-2 challenge

vaccine BNT162b2 selected for a pivotal efficacy study

518 exhaustion of T cells in patients with coronavirus disease 2019 (COVID-19). Front 603 immunol 11:1729

Longitudinal analyses 610 reveal immunological misfiring in severe COVID-19

Origin and evolution of pathogenic coronaviruses

Efficient induction of 615 cytotoxic T lymphocytes specific for severe acute respiratory syndrome 616 (SARS)-associated coronavirus by immunization with surface-linked liposomal 617 peptides derived from a non-structural polyprotein 1a

SYFPEITHI: database for MHC ligands and peptide motifs

A hybrid approach for predicting promiscuous 622 MHC class I restricted T cell epitopes

ProPred1: Prediction of promiscuous MHC class-I 624 binding sites

The immune epitope database (IEDB)

Cytotoxic T 630 lymphocytes recognize a fragment of influenza virus matrix protein in association 631 with HLA-A2

An optimal viral peptide recognized by 633

CD8 + T cells binds very tightly to the restricting class I major histocompatibility 634 complex protein on intact cells but not to the purified class I protein

1-restricted education and cytolytic activity of CD8 + T lymphocytes from 638 beta2 microglobulin (beta2m) HLA-A2.1 monochain transgenic H-2D b beta2m 639 double knockout mice

A 641 sequence homology and bioinformatics approach can predict candidate targets for 642 immune responses to SARS-CoV-2

COVID-19 coronavirus vaccine design using 644 reverse vaccinology and machine learning

SARS-CoV-2-specific T cell 649 immunity in cases of COVID-19 and SARS, and uninfected controls

Selective 655 and cross-reactive SARS-CoV-2 T cell epitopes in unexposed humans

Sette A. 660 2020. Targets of T cell responses to SARS-CoV-2 coronavirus in humans with

COVID-19 disease and unexposed individuals

SARS-CoV-2-reactive T cells in healthy donors and patients with COVID-19

SARS-CoV-2-epitopes define heterologous and COVID-19-induced T-cell 676 recognition

Pre-existing immunity to SARS-CoV-2: the knowns and 678 unknowns

Introduction of a point 680 mutation into an HLA class I single-chain trimer induces enhancement of CTL 681 priming and antitumor immunity

Partial purification and some properties of 684

A cytotoxic monoclonal antibody with specificity for HLA-A2 and a variant 685 of HLA-A28

B & C) Comparison of the 10 peptides in the induction of IFN- + CD8 + T cells

Seventeen HHD mice were immunized once with the mixture of 10 peptides involving

After one week, spleen cells were stimulated with or without each of the 10

CD8+ T cells was stained. (C) Y-axis indicates the 713 relative percentages of IFN- + cells in CD8 + T cells which were calculated by subtracting

the % of IFN- + cells in CD8 + T cells without a peptide from the % of IFN- + cells in

Each blue circle represents an individual mouse

Data are shown as the mean (horizontal bars) ± SD. (B) Statistical comparisons of the 717

relative % values of IFN- + CD8 + T cells among the 10 peptides in Fig. 3C were made by 718 one-way ANOVA followed by post-hoc tests. *, P < 0

Comparison of the 10 peptides in the induction of CD107a + CD8 + T cells (A) and 722

After one week, spleen cells were stimulated with or 725 without each of the 10 peptides

T cells was analyzed. Data indicates the relative percentages of CD107a + (A) and CD69 + 727 (B) cells in CD8 + T cells which were obtained by subtracting the % of CD107a + and 728

CD69 + cells in CD8 + T cells without a peptide from the % of CD107a + and CD69 + cells

Each red (A) or green (B) circle represents 730 an individual mouse. Data are shown as the mean (horizontal bars) ± SD. Statistical 731 analyses of the data among the 10 peptides in Fig. 4A and Fig. 4B were performed by 732 one-way ANOVA followed by post-hoc tests in Fig

Three types of reactivity in mice immunized with the mixture of the 10 peptides

Fifteen mice were immunized once with the mixture of 10 peptides including pp1a-38

After one week, spleen cells were stimulated with or without each of the 10 peptides, and 739 intracellular IFN- in CD8+ T cells was stained

peptides, fifteen mice were divided into three types, A-C. Each graph represents 741 reactivity of an individual mouse. Y-axis indicates the relative percentage of IFN- + cells 742

CD8 + T cells which was calculated by subtracting the % of IFN- + cells in CD8 + T 743 cells without a peptide from the % of IFN- + cells in CD8 + T cells with a relevant peptide

Statistical analyses of the relative % values to 10 peptides in each type were performed by 745 one-way ANOVA followed by post-hoc tests