Rapid and stable mobilization of fully functional spike-specific CD8+ T cells preceding a mature humoral response after SARS-CoV-2 mRNA vaccination


 SARS-CoV-2 spike mRNA vaccines mediate protection from severe disease as early as 10 days post prime vaccination, when specific antibodies are hardly detectable and still lack neutralizing activity. Vaccine-induced T cells, especially CD8+ T cells, may thus be the main mediators of protection at this early stage. The details of antigen-specific CD8+ T cell induction after prime/boost vaccination, their comparison to naturally induced CD8+ T cell responses and their association with other arms of vaccine-induced adaptive immunity remain, however, incompletely understood. Here, we show on a single epitope level that both, a stable memory precursor pool of spike-specific CD8+ T cells and fully functional spike-specific effector CD8+ T cell populations, are vigorously mobilized as early as one week after prime vaccination when CD4+ T cell and spike-specific antibody responses are still weak and neutralizing antibodies are lacking. Boost vaccination after 3 weeks induced a full-fledged recall expansion generating highly differentiated CD8+ effector T cells, however, neither the functional capacity nor the memory precursor T cell pool was affected. Compared to natural infection, vaccine-induced early memory T cells exhibited similar frequencies and functional capacities but a different subset distribution dominated by effector memory T cells at the expense of self-renewing and multipotent central memory T cells. Our results indicate that spike-specific CD8+ T cells may represent the major correlate of early protection after SARS-CoV-2 mRNA/bnt162b2 prime vaccination that precede other effector arms of vaccine-induced adaptive immunity and are stably maintained after boost vaccination.

However, t-SNE analysis of the concatenated expression analysis revealed that spike 121 epitope-specific CD8+ Teff cells are qualitatively different at the peak expansion 122 following boost (obtained at 5-6 dpb) compared to prime (obtained at 9-12 dpp) with a 123 more consolidated cytotoxic effector cell phenotype (increased T-BET, TOX, CD39 Next to the emergence of an effective Teff response after bnt162b2 vaccination, we 136 also assessed the induction of a spike epitope-specific memory precursor CD8+ T cell 137 (TMP) pool. TMP cells are characterized by CD127, BCL-2 and TCF-1 expression and 138 are relevant for maintaining the CD8+ T cell response 13,14 . Roughly 30% of A*01/S865-139 and A*02/S269-specific CD8+ T cells expressed CD127 after prime followed by a To assess spike-specific CD8+ T cell function after prime/boost vaccination, we 152 analyzed expansion capacity, cytokine production and degranulation after peptide-153 specific expansion ( Fig. 2A/B). After two weeks of in vitro expansion, we detected 154 higher frequencies of A*01/S865 and A*02/S269-specific CD8+ T cells after boost 155 compared to prime vaccination (Fig. 2C). However, the expansion index, a measure 156 taking the input number of virus-specific CD8+ T cells into account was comparable 157 for spike epitope-specific CD8+ T cells after prime and boost vaccination (Fig. 2D). 158 Thus, the increased frequencies of spike epitope-specific CD8+ T cells after peptide-159 specific expansion most probably result from the increased ex vivo frequencies after 160 boost. We also assessed A*01/S865-and A*02/S269-specific IFN-γ and TNF production 161

Delayed appearance of circulating spike-specific CD4+ T cells, B cells and antibodies 171
Next, we compared kinetics and frequencies of spike-specific CD8+ T cells with spike 172 epitope-specific CD4+ T cells, B cells and antibodies with neutralizing activity. For this, 173 we first longitudinally assessed circulating spike-specific CD4+ T cells targeting 174 DRB1*15:01/S236 (Extended Data Fig. 5A) following prime and boost vaccination in 8 175 Table 1 Our comprehensive analysis of vaccine induced immunity revealed a rapid and strong 261 induction of CD8+ T cells targeting two spike epitopes as early as one week after prime 262 vaccination in all vaccinees at a time point when neutralizing antibodies are still lacking. 263

individuals (Supplementary
Importantly, these immediately mobilized spike-specific CD8+ T cells are fully 264 functional and able to mount a proficient recall response evoked e.g. by the boost 265 vaccination. These findings clearly suggest a major protective role of vaccine-elicited 266 CD8+ T cells early after first dose vaccination. In agreement with this notion, the 267 protective effect observed for mRNA vaccines starts at 10-12 dpp 2,3 and thus coincides 268 with the first peak of spike-specific CD8+ T cells. A major protective capacity of virus-269 specific CD8 T cells was also observed in natural SARS-CoV-2 infection, since e.g. 270

CD8+ T cells compensate for impaired humoral immunity in COVID-19 patients with 271
hematologic cancer 17 . 272 The approved SARS-CoV-2 mRNA vaccine design includes a boost after 3-6 weeks 2,3 . 273 Our data demonstrate that the boost after 3 weeks hits the spike-specific CD8+ T cells 274 at the beginning of the contraction phase of the primary response and results in a 275 robust recall expansion of spike-specific CD8+ T cells. The generation of a fortified Teff 276 response, however, neither leads to an improved functional capacity nor to an 277 increased frequency of memory precursor T cells relevant for maintaining the CD8+ T 278 cell response. These observations may indicate that a robust spike-specific CD8+ T 279 cell response is already elicited after prime with only transient effects of boosting after 280 3 weeks. 281 In contrast to CD8+ T cells, efficient mobilization of a humoral immune response 282 comprising neutralizing antibodies and antigen-specific memory B cells to the 283 periphery was first detectable after boost. This is in line with previous reports 4,11,12,18 284 and most probably represents maturation of the response that is initially localized in 285 secondary lymphoid organs (SLO) 19 and subsequently released to the circulation. 286 However, how the boost affects this SLO response requires further investigation. After 287 boost, highly cross-neutralizing antibodies are present in the sera clearly adding a 288 major correlate of protection on top of the early-mobilized spike-specific CD8+ T cell 289

response. 290
A recent study has highlighted the important role of spike-reactive CD4+ T cells in 291 coordinating the humoral and CD8+ T cell response 8 . Similar to CD8+ T cells, spike-292 reactive CD4+ T cells are also mobilized early after prime 8,18 . We also observed an 293 early but moderate appearance of spike-specific CD4+ T cells with a predominance of 294 a TH1 over a TFH phenotype in the periphery. While a dominant TH1 polarization has 295 also been reported by others 4 , spike-reactive TFH cells may also be crucial for 296 coordinating humoral and cellular immune responses upon mRNA vaccination 8 . 297 Further studies need to clarify whether the parallel TH1-and TFH-directed coordination 298 is restricted by different CD4+ T cell epitopes or clonotypes, by pre-existing cross-299 reactive CD4+ T cells or by CD4+ T cell localization (periphery versus SLO). Unlike 300 spike-specific CD8+ T cells, only a limited recall expansion was detectable for 301 circulating spike-specific CD4+ T cells, supporting their coordinating role after mRNA 302

vaccination. 303
Hallmark of an efficient vaccine is the induction of lasting immune memory. In parallel 304 to the early appearance of circulating spike-specific CD8+ T cells, a stable memory 305 precursor pool is established that develops after boost into a fully functional early 306 memory CD8+ T cell response. Of note, the functional capacity of spike-specific early 307 memory CD8+ T cells is similar after vaccination and natural infection up to three 308 months post boost/symptom onset. Compared to natural infection, however, the early 309 memory pool of spike-specific CD8+ T cells after vaccination contains a higher 310 proportion of effector memory 1 (TEM1) T cells at the expense of central memory 311 14 (TCM) and early differentiated (TED) T cells. Of note, the latter maintain long-term 312 memory e.g. after yellow fever virus (YFV) vaccination 20 . In line with this, expression 313 of CCR7, BCL-2 and TCF-1 is lower after vaccination compared to infection. These 314 observations may hint towards a restricted self-renewal and maintenance of spike-315 specific CD8+ T cells after vaccination compared to infection. This difference may be 316 caused by differential duration and location of antigen contact after vaccination versus 317 infection 21,22 . Indeed, we observed a lower CD38 expression on early memory spike-318 specific CD8+ T cells after vaccination possibly indicating limited antigen recognition 319 compared to natural infection 4,23 . So far, we have followed our cohort for only 3 months 320 after boost; follow-up studies including larger cohorts of vaccinees and SARS-CoV-2 321 convalescent donors should comparatively assess longevity of CD8+ T cell immunity. 322 In addition, our study was limited to circulating spike-specific adaptive immunity, not 323 addressing local immunity at the viral entry site, the respiratory tract. 324 In sum, our data clearly highlight that a robust, stable and fully functional spike-specific 325 CD8+ T cell response is rapidly mobilized already after prime vaccination and is able 326 to patrol the periphery for SARS-CoV-2 at least within the first months. These 327 observations do not only provide insides into the protective mechanisms underlying 328 bnt162b2 vaccination, but are also of potential relevance for the development of novel 329 vaccination strategies against emerging pathogens and cancer.   In vitro expansion of spike-specific CD8+ T cells and assessment of effector function 380 1.5*10 6 PBMCs were stimulated with A*01/ S865 or A*02/ S269 spike-specific peptides 381 (5 µM) and anti-CD28 mAb (0.5 µg/mL, BD) and expanded for 14 days in complete 382 RPMI culture medium containing rIL2 (20 IU/mL, Stemcell Technologies). Intracellular 383 cytokine production and degranulation was assessed with spike-specific peptides (15 384 μM) in the presence of anti-CD107a (H4A3, 1:100) (BD Bioscience, Germany) for 1 h 385 at 37 °C. Afterwards, brefeldin A (GolgiPlug, 0.5 μL/mL) and monensin (GolgiStop, 0.5 386 μL/mL) (all BD Biosciences, Germany) were added for additional 5 h, followed by 387 surface and intracellular staining. The expansion capacity was calculated based on 388 peptide-loaded HLA class I tetramer staining as described before 25 . 389 390

Magnetic bead-based enrichment of spike-specific CD8+ T cells 391
Spike-specific CD8+ T cells were enriched as previously described 26 . Briefly, 1-2*10 7 392 PBMCs (with an average of 15.7% CD8+ T cells) were labelled with PE-coupled 393 peptide-loaded HLA class I tetramers for 30 min. Enrichment was then performed using 394 anti-PE beads with MACS technology (Miltenyi Biotec, Germany) according to the 395 manufacturer's instructions. Subsequently, enriched spike-specific CD8+ T cells were 396 analyzed by multiparametric flow cytometry and frequencies of spike-specific CD8+ T 397 cells were calculated as described before 26 . Only enriched samples with ≥ 5 spike-398 specific CD8 T cells were included in further analyses, resulting in a detection limit of 399 5*10 -6 . 400 401

Magnetic bead-based enrichment of spike-specific CD4+ T cells 402
Enrichment of spike-specific CD4+ T cells was adapted from the method described 403

PBMCs of vaccinated subjects and patients with a history of SARS-CoV-2 infection 488
were plated at 0.5*10 6 cells/ml and polyclonally stimulated for 9 days with thiol-modified 489 CpG (0.25 µM, TCGTCGTTTTGTCGTTTTGTCGTT) and hIL-2 (100 ng/ml, 490 Immunotools). At day 9, the supernatants of the in vitro culture were cleared from 491 debris by centrifugation and used to determine the presence of SARS-CoV-2 Spike-492 specific IgG antibodies (Anti-SARS-CoV-2-QuantiVac-ELISA (IgG), Euroimmun) 493 according to the manufacturer's instructions. To detect S1 specific IgM, supernatant of 494 the in vitro culture and serum of vaccinated subjects was incubated on a S1 pre-coated

Data availability statement 531
Raw data is available upon reasonable request. Further supporting data are available 532 from the corresponding authors upon reasonable request. All requests for raw and 533 analyzed data and materials will be reviewed by the corresponding authors to verify if 534 the request is subject to any intellectual property or confidentiality obligations. Patient-535 related data not included in the paper were generated as part of clinical examination 536 and may be subject to patient confidentiality. Any data and materials that can be shared 537 will be released via a Material Transfer Agreement. Source data are provided with this 538 paper.