key: cord-0747345-txnpgf08
authors: Mudd, Philip A.; Minervina, Anastasia A.; Pogorelyy, Mikhail V.; Turner, Jackson S.; Kim, Wooseob; Kalaidina, Elizaveta; Petersen, Jan; Schmitz, Aaron J.; Lei, Tingting; Haile, Alem; Kirk, Allison M.; Mettelman, Robert C.; Crawford, Jeremy Chase; Nguyen, Thi H.O.; Rowntree, Louise C.; Rosati, Elisa; Richards, Katherine A.; Sant, Andrea J.; Klebert, Michael K.; Suessen, Teresa; Middleton, William D.; Wolf, Joshua; Teefey, Sharlene A.; O’Halloran, Jane A.; Presti, Rachel M.; Kedzierska, Katherine; Rossjohn, Jamie; Thomas, Paul G.; Ellebedy, Ali H.
title: SARS-CoV-2 mRNA vaccination elicits a robust and persistent T follicular helper cell response in humans
date: 2021-12-23
journal: Cell
DOI: 10.1016/j.cell.2021.12.026
sha: 70a640663a7d02f83ba25664463f955691c15d20
doc_id: 747345
cord_uid: txnpgf08

SARS-CoV-2 mRNA vaccines induce robust anti-spike (S) antibody and CD4+ T cell responses. It is not yet clear whether vaccine-induced follicular helper CD4+ T (TFH) cell responses contribute to this outstanding immunogenicity. Using fine needle aspiration of draining axillary lymph nodes from individuals who received the BNT162b2 mRNA vaccine, we evaluated the T cell receptor sequences and phenotype of lymph node TFH. Mining of the responding TFH T cell receptor repertoire revealed a strikingly immunodominant HLA-DPB1∗04-restricted response to S167-180 in individuals with this allele, which is among the most common HLA alleles in humans. Paired blood and lymph node specimens show that while circulating S-specific TFH cells peak one week after the second immunization, S-specific TFH persist at nearly constant frequencies for at least six months. Collectively, our results underscore the key role that robust TFH cell responses play in establishing long-term immunity by this efficacious human vaccine.

The COVID-19 pandemic necessitated rapid late-stage clinical trials of mRNA vaccine technology (Anderson et al., 2020; Baden et al., 2021; Jackson et al., 2020; Polack et al., 2020; Verbeke et al., 2021; Walsh et al., 2020; Widge et al., 2021) that resulted in the first FDAapproved vaccine using this technology platform. The two mRNA vaccines developed by Pfizer/BioNTech (BNT162b2) (Polack et al., 2020) and Moderna (mRNA-1273) (Baden et al., 2021) have proven instrumental in the initiation of widespread vaccination campaigns in the United States and around the world. Both vaccines engender high-titer circulating anti-SARS-CoV-2 Spike (S) protein-specific antibodies that can neutralize the originally circulating SARS-CoV-2 strain (Jackson et al., 2020; Walsh et al., 2020) as well as other variants that have emerged since the vaccine design phase (Chen et al., 2021; Wang et al., 2021a Wang et al., , 2021b Wu et al., 2021) . Neutralizing antibodies induced by mRNA vaccines appear to be the key correlate of protection from COVID-19 in animal models (Corbett et al., 2021) and in humans (Khoury et al., 2021) . COVID-19 mRNA vaccines exhibit the highest efficacy in phase 3 studies among widely utilized COVID-19 vaccines worldwide (Al Kaabi et al., 2021; Baden et al., 2021; Logunov et al., 2021; Polack et al., 2020; Sadoff et al., 2021; Voysey et al., 2021) . Understanding exactly how mRNA vaccines elicit such robust and protective immune responses in humans is necessary for extending the application of this novel platform to vaccines against other important human pathogens.

Germinal center (GC) reactions that occur in draining lymph nodes following infection or vaccination are critical for developing long-lasting, high-affinity antibody responses (Ripperger and Bhattacharya, 2021; Victora and Nussenzweig, 2012) . T follicular helper (TFH) cell responses in the lymph node are necessary for forming and sustaining GC reactions and for the J o u r n a l P r e -p r o o f development of both long-lived plasma cells and memory B cells (Crotty, 2011; Qi, 2016; Ueno et al., 2015) . Detailed analysis of the specificity and dynamics of vaccination induced GC reactions in humans is increasingly being explored through sampling draining lymph nodes using serial fine needle aspiration (FNA) following intramuscular immunization (Turner et al., 2020 (Turner et al., , 2021 Kim et al., 2021) . Importantly, it appears that the GC reaction in humans persists over a longer period of time (Turner et al., 2020 (Turner et al., , 2021 Kim et al., 2021) than what was anticipated from studies in preclinical animal models (Good-Jacobson et al., 2014; Weisel et al., 2016) .

Determining the epitope targets and dynamics of SARS-CoV-2-specific TFH cells induced in human draining lymph nodes during an active immune response is critical for understanding the role of TFH in the development of long-lived plasma cells and memory B cells following vaccination.

We conducted a prospective observational study to follow vaccine-induced immune responses in a cohort of 41 healthy adults who received the BNT162b2 mRNA vaccine (Turner et al., 2021) . Demographics of the full cohort have been previously reported (Turner et al., 2021) . Fifteen members of the cohort underwent axillary lymph node FNA. All subjects were vaccinated with two 30 µg doses of BNT162b2 approximately twenty-one days apart. Blood and/or FNA samples were obtained at day 0 (prior to the first vaccine dose), day 21 (immediately prior to the second vaccine dose), day 28, day 35, day 60, day 110 and day 200 according to the schedule listed in Figure 1A . This manuscript reports exclusively on the 15 subjects who underwent lymph node FNA. Demographics of the included individuals are listed in Table 1 .

None of the included subjects reported previous infection with SARS-CoV-2.

We first evaluated the size of the human TFH population in relation to the size of the GC B cell population in the lymph node. We analyzed the frequency of the GC B cell response (defined as CD19 + IgD low Bcl-6 + CD38 int B cells) among all lymph node resident B cells and the frequency of total lymph node-resident CD4 + T cells that exhibited a TFH cell phenotype (Bcl-6 + CXCR5 + PD1 + FoxP3 -) in 95 separate lymph node samples taken from each of the 15 individuals over the course of the study ( Figure S1A , Table S1 ). These FNA samples were obtained between 21 and 200 days following primary vaccination. Six of the fifteen subjects underwent repeated sampling of two separate axillary lymph nodes (Table S1). We found a significant correlation between the size of the GC B cell population in the lymph node and the total TFH cell population frequency following mRNA vaccination ( Figure S1B) . We also noted a significant correlation between the size of the SARS-CoV-2 Spike-specific GC B cell population in the lymph node and the total lymph node TFH cell population frequency ( Figure S1C ).

We next sought to illuminate the antigen-specificity of the lymph node TFH population.

To do this, we sorted total TFH from FNA samples obtained on day 60 from four separate subjects ( Figure 1B) and reconstructed their T cell receptor (TCR) repertoires using unpaired sequencing of the TCR α and TCR β chains ( Figure 1C) . Surprisingly, clonally expanded TCRs formed a prominent α chain cluster that was shared among all 4 donors ( Figure 1C ), corresponding to 0.9-7.7% of the total lymph node TFH cells in each donor. We did not observe a similar shared cluster in the TCR β chain repertoires. We observed the same α motif in a previously published paper (Minervina et al., 2021a) , where it was the largest signal and corresponded to 0.2% of total CD4 + T cells and 16.3% of estimated SARS-CoV-2-responding J o u r n a l P r e -p r o o f 7 CD4 + T cells in the blood at the peak of the acute response. Large clusters of TCRs with sequence similarity are an indication of convergent selection of similar receptors to the same antigen (Dash et al., 2017; Glanville et al., 2017; Pogorelyy et al., 2019) . As this motif was present among expanded clones in many donors, it likely recognizes an immunodominant epitope from SARS-CoV-2 presented in the context of a common HLA class II allele.

In order to decode the specificity of the heterodimer αβTCR, we first needed to determine what β chains pair with the TCRα chain motif that we identified (Figure 2A) . To do this, we queried publicly available CD4 + paired TCR datasets. We used two datasets that have paired αβTCR sequences from CD4 + T cells after antigen-reactive T cell enrichment following stimulation with SARS-CoV-2 peptides Meckiff et al., 2020) . We searched for our CDR3α motif ("CA[G/A/V]XNYGGSQGNLIF") in these datasets and found 1329 out of 44256 unique TCRs in Bacher et al., but only 53 out of 43745 in Meckiff et al. with the matched CDR3α motif. We next used the identified β chains to look for overlap in the MIRA dataset (Nolan et al., 2020 )a large dataset produced by Adaptive Biotech linking TCR sequences to SARS-CoV-2 epitopes. We identified 64 TCRs from Bacher et al. highly similar (up to one amino acid mismatch in CDR3, identical CDR1 and CDR2) to MIRA TCRs reactive to the overlapping peptide pool from SARS-CoV-2 Spike protein 160-218 positions ( Figure 2B) .

Interestingly, this part of the spike protein was not used for stimulation in Meckiff et al., explaining why we found only a small number of TCRs of interest in this dataset and indirectly supporting the predicted identification of the peptide region from the MIRA dataset.

Five of six subjects recognizing this peptide pool in the MIRA database had available HLA-typing. These five shared the DPB1*04:(01/02) and DQB1*06:(02/03) alleles. To establish HLA-restriction of the response of interest and to narrow the search to a single peptide, we next J o u r n a l P r e -p r o o f used NetMHCII2.3 (Jensen et al., 2018) to look for predicted epitopes from the S160-218 peptide pool that are presented by one or both of these shared alleles. We found that peptides containing the core sequence YVSQPFLMD were predicted to strongly bind the DPB1*04:01 and DPB1*04:02 alleles, while no strong binders were identified for the DQB1*06:(02/03) alleles.

Interestingly, TCR epitopes with this core sequence (YVSQPFLMD, S170-178) have been previously described in prominent epitope discovery studies (Peng et al., 2020; Tarke et al., 2021) , where the response was identified in multiple donors. However, this response has not previously been reported to be HLA-DPB1*04-restricted.

As an initial investigation of this possible HLA-restriction, we obtained post-vaccination peripheral blood from participants in the ongoing SJTRC study (SJTRC, NCT04362995).

PBMCs from these participants were stimulated with purified S166-180 peptide (CTFEYVSQPFLMDLE) and the responses were measured by intracellular cytokine staining and flow cytometry. We determined that participants with the HLA-DPB1*04 allele had increased cell counts per million PBMCs of monofunctional CD4 + CD69 + T cells producing IL-2, TNFα, or IFN compared to participants without this allele ( Figure 2C , Figure S2B -C).

Further, we noted that each DPB1*04 + donor had activated polyfunctional T cells producing two or three cytokines in response to peptide stimulation, in both vaccinated naive and SARS-CoV-2 convalescent individuals (Figure S2D-E) .

We then moved forward with more rigorous experimental validation of our paired TCR, peptide epitope, and restricting HLA combination (Figure 2A) . To do this, we selected two paired TCRs from Bacher et al. that included the same TCRα, but distinct TCRβ chains that we designated TCR4.1 and TCR6.3. We transduced these each into separate Jurkat TCR-negative cell lines that also express an endogenous NFAT-GFP reporter to allow for tracking intracellular J o u r n a l P r e -p r o o f signaling downstream of the transduced paired TCR following TCR engagement. The TCRtransduced Jurkat cell lines were co-cultured with PBMCs from an HLA-DPB1*04 + donor and pulsed with S166-180 peptide to evaluate TCR activation. Consistent with our prediction, we observed strong NFAT activation from the CTFEYVSQPFLMDLE-stimulated cells expressing either TCR pairing ( Figure 2D, Figure S3 , Figure S4 ). Further, we performed additional stimulation experiments employing a mutant version of the S166-180 peptide found in the GISAID database (CTFEYSQPFFMDLE), as well as a set of overlapping peptides ( Figure S4 ) to determine the core peptide required for TCR engagement. Both TCR lines recognized the mutated epitope as well as the overlapping peptides containing the YVSQPFLM amino acid stretch suggesting this core sequence is crucial to TCR engagement. Interestingly, this core is truncated at P8 in comparison to the core predicted by NetMHC (YVSQPFLMD). In contrast, the N-terminus part of the core (YVSQPFLMD) did not tolerate any truncations, highlighting the importance of P1 and providing a clear specificity control for the peptide stimulation experiment.

In a canonical orientation of the TCR binding to HLA-DPB1*04, the TCR α-chain can be expected to reside above the N-terminal portion of the peptide, whereas the β-chain should reside above the C-terminal portion of the peptide. It is reasonable to assume that preferential TRAV35 selection is driven by some strong interactions between the TCR α-chain and a feature in the Nterminal portion of the peptide. Thus, a TRAV biased TCR may be particularly sensitive to a truncation of the peptide N-terminus. The selection of multiple TRBV gene segments suggests that interactions between peptide and TCR β-chain are less critical, which may explain why the P8 truncation is tolerated.

We next generated an HLA class II tetramer to probe the antigen-specific T cell response we discovered. We tested our HLA-DPB1*04 S167-180 tetramer using the two transduced TCR4.1 J o u r n a l P r e -p r o o f and TCR6.3 Jurkat cell lines and showed high sensitivity and low-background staining ( Figure   2E ). We then used the S167-180 tetramer to look for antigen-specific CD4 + T cells in PBMC from three HLA-DPB1*04 + SARS-CoV-2 convalescent donors and a control HLA-DPB1*04 + SARS-CoV-2 naïve donor. We found a small number of tetramer-specific cells predominantly in the naïve subpopulation (CCR7 + CD45RA + ) in the naïve donor, and a much larger number of tetramer-specific cells that were primarily effector memory (CCR7 -CD45RA -) in the SARS-CoV-2 convalescent donors ( Figure 2F ). The frequency of tetramer-positive cells was comparable to the frequencies of the total Spike-specific cells observed using a separate AIM assay with overlapping Spike-peptides ( Figure S5 ), suggesting that tetramer staining provided a higher sensitivity to detect epitope-specific responses. We then sequenced tetramer-specific TCRs from these convalescent donors using our previously described scTCRseq approach (Wang et al., 2012) . The majority (64%) of sequenced cells had the same TRAV35-CA[G/A/V]XNYGGSQGNLIF TCRα motif that we initially identified, and >80% of all sequences included TRAV35, suggesting that the discovered TCRα motif is the most frequent mode of recognition for this epitope (Table S2) . We also found the TCR4.1β (exactly matching amino acid sequence) and TCR6.3β (one mismatch) in the single cell TCR sequencing of tetramer specific T cells from convalescent individuals. This is a further independent validation that the αβTCRs selected for Jurkat cell line generation are S167-180 specific and occur in multiple patients.

Tracking S167-180 antigen-specific CD4 + T cell responses in blood and draining lymph nodes following BNT162b2 vaccination

With the discovery of an immunodominant SARS-CoV-2-S epitope restricted by the HLA-DPB1*04:01 allele that is found at high frequency (>40%) in many populations around the J o u r n a l P r e -p r o o f world (allelefrequencies.net), we used the S167-180 HLA class II tetramer to evaluate 14 of the mRNA vaccine study subjects with available blood and lymph node samples to empirically determine which individuals were HLA-DPB1*04:01 + and thus made the S167-180-specific CD4 + T cell response. Nine of 14 subjects made a detectable S167-180-specific response in peripheral blood following boost vaccination. We next tracked and characterized this response over time in frozen PBMC (N=8 subjects) and frozen lymph node FNA samples (N=6 unique lymph nodes from 5 subjects) from a convenience sample of the subjects with sufficient samples remaining for analysis. The S167-180-specific CD4 + T cell response peaked in peripheral blood 28 days after primary vaccination, 7 days after vaccine boost, and remained present in the blood at detectable frequencies through the entire study interval (Figure 3A and 3B). Most S167-180-specific CD4 + T cells circulating in peripheral blood exhibited a CD45RO + CCR7effector memory surface phenotype similar to what we observed in SARS-CoV-2 convalescent donors ( Figure 3C) . A subset of tetramer-positive CD4 + T cells in the first 35 days following primary vaccination exhibited an activated surface phenotype characterized by upregulation of both CD38 and HLA-DR ( Figure 3D ). This activated CD4 + T cell phenotype disappeared by day 60 post-primary vaccination. Most circulating S167-180-specific CD4 + T cells expressed both PD1 and ICOS at high levels on days 21 and 28 following primary vaccination with a gradual decrease in the mean fluorescent intensity of PD1 and ICOS throughout the remaining study interval to a level more consistent with that found on the majority of circulating CD4 + T cells in line with resolution of T cell activation ( Figure 3E) . A subset of S167-180-specific CD4 + T cells accounting for approximately 5-15% of the total number of circulating S167-180-specific CD4 + T cells exhibited the CXCR5 + PD1 + circulating TFH phenotype ( Figure 3F ). These circulating S167-180-specific TFH cells peaked 28 days after primary vaccination, 7 days after vaccine boost and then decreased J o u r n a l P r e -p r o o f over time, becoming difficult to detect in the blood of some subjects by the final study time-point ( Figure 3G) . We evaluated the expression of the Th1-associated chemokine receptor CXCR3 on the surface of S167-180-specific CD4 + T cells from a single subject with available sample and noted that most of the S167-180-specific cells expressed CXCR3 but not CXCR5 at days 21 and 28 following primary vaccination ( Figure S6) . Collectively, these results demonstrate that the circulating S167-180-specific CD4 + T cell population exhibits a dynamic surface phenotype over time with a general bias towards surface phenotypes that do not include circulating TFH.

In contrast to circulating populations of TFH cells, the frequency of S167-180-specific CD4 + TFH cells remained high in the draining axillary lymph node through at least day 60 following primary vaccination and persisted at high frequency in three of the five study subjects through day 200 following primary vaccinationmore than 170 days following vaccine boost ( Figure   4 ). The prolonged persistence of S-specific TFH that we report here in the draining axillary lymph nodes corresponds well with the long-lived germinal center B cell responses recently reported in the same cohort of subjects (Turner et al., 2021) . The vast majority of S167-180-specific CD4 + T cells in lymph node FNA samples co-expressed CXCR5 and PD1, surface markers of TFH cells, throughout the study interval ( Figure 4A) . Furthermore, the frequency of S167-180-specific CD4 + T cells in the FNA samples remained consistently high or even increased as the frequency of S167-180-specific CD4 + T cells in the peripheral blood contracted. These lymph node TFH responses remained high frequency until the conclusion of the GC response in 2 of the 5 subjects at day 200 ( Figure 4B ).

We next examined the frequency of S167-180-specific CD4 + T cells in the total CXCR5 + PD1 + TFH population in both the blood and the lymph nodes over time. We found that this population rapidly expanded in the bloodpeaking at day 28 after primary vaccination, 7 J o u r n a l P r e -p r o o f days after vaccine boostand then became challenging to detect by days 110 and 200 ( Figure   4C ) as we had previously noted when examining this population as a proportion of total CD4 + T cells in Figure 3G . In contrast, the frequency of the S167-180-specific TFH population remained consistently elevated within the total TFH population over time in the lymph nodeuntil the resolution of the lymph node GC response at day 200 in 2 of the 5 subjects ( Figure 4C , Table   S1 ). Together, these results demonstrate that a small subset of antigen-specific CD4 + T cells circulating in peripheral blood following vaccination develop a surface phenotype consistent with circulating TFH cells. This coincides with the development of TFH cells in the draining lymph node with the same antigen-specificity. Furthermore, while this population nearly disappears from circulating blood 110 days after vaccination, the response remains constant in the lymph node in the presence of an ongoing GC reaction. Overall, our findings are consistent with the development of diverse lineages of effector CD4 + T cellsthose that express a surface phenotype consistent with TFH and those that do not -from a single population of naïve CD4 + T cells that share a common TCR alpha chain motif. This is consistent with observations in mouse models where the specificity and duration of the T cell receptor / peptide / MHC class II interaction correlated with the overall balance between Th1 and TFH cell frequency (Tubo et al., 2013) .

We next quantified the contribution of the S167-180 TFH population to the broader clonotypic diversity of the TFH population found in the lymph node from four of the subjects by further analyzing the TCR sequencing data from sorted TFH cells generated for Figure 1C . The clonotypes that compose the S167-180 response made up the largest percentage of total clonotypes present in the lymph node for three of the four subjects and composed the second highest J o u r n a l P r e -p r o o f percentage of clonotypes in the fourth subject ( Figure 5A ). This underscores the importance of the immunodominant HLA-DPB1*04-restricted S167-180 response in the total SARS-CoV-2specific TFH response of HLA-DPB1*04 + vaccinees, who make up approximately 40-50% of the world's population.

To elucidate the clonal composition of the TFH response over time, we sequenced samples from two time-points that were available from these individuals. Three subjects were sequenced at day 60 and day 110 post primary vaccination and one subject was sequenced at day 28 and day 60 following primary vaccination ( Table S3) . Three of the subjects exhibited evidence of ongoing antigen-specific TFH responses associated with germinal center responses at all tested time-points by flow cytometry (Figure 4) , while subject #22 did not have sufficient remaining samples to be analyzed by flow cytometry. Supporting our observations in the flow cytometry analysis of the S167-180 population, we found a positive correlation between the frequency of a large number of the TCR α clonotype sequences at the two time points (Figure 5B ), including the known S167-180-specific TCR clonotypes ( Figure 5B , red data points). This was especially true of the clonotypes found at the highest frequency in each FNA sample which are those that are most likely to represent antigen-specific clonotypes due to their increased presence in the lymph node following vaccination. This positive correlation means that many of these clonotypes were found at similar frequency at both tested time-points. Therefore, the maintenance of consistently high frequency antigen-specific TFH responses over time during an ongoing antigen-specific GC B cell response (Turner et al., 2021) that we observed in the context of the S167-180-specific CD4 + TFH response ( Figure 4C ) is generalizable to other clonally-related and presumably antigen-specific TFH populations in the human lymph node following BNT162b2 vaccination. Our data support a model whereby the antigen-specific human germinal center TFH J o u r n a l P r e -p r o o f response is maintained at a relatively consistent and high frequency in the setting of an active and ongoing germinal center reaction, rather than a response that peaks or dynamically changes in frequency over time.

In this report, we show that the BNT162b2 COVID-19 mRNA vaccine induces robust and persistent TFH responses in the draining lymph nodes of vaccinated individuals. Indirect evidence has existed for some time that robust CD4 + T cell responses are required for the generation of high-titer neutralizing antibody responses following COVID-19 infection or mRNA vaccination. This includes data showing a lack of seroconversion in individuals with uncontrolled HIV and extremely low CD4 + T cell counts during vaccination (Touizer et al., 2021) as well as several reports that have demonstrated a lack of seroconversion to the standard two-dose BNT162b2 regimen in individuals subjected to T cell-focused immunosuppressive regimens following solid organ transplantation (Kamar et al., 2021) . Our current results provide strong and direct evidence that a high-magnitude, antigen-specific CD4 + T cell response in the draining lymph nodes is present during the development of high-titer neutralizing antibody responses in the setting of COVID-19 mRNA vaccination.

The temporal relationship we observe between the early appearance and then disappearance of S167-180-specific CD4 + T cells exhibiting a circulating TFH phenotype in the blood at the same time that we observe TFH cells in the draining lymph node suggests a complex relationship between these two populations of cells. Our present data support a model of human TFH cell development whereby phenotypically heterogeneous, or even plastic, antigen-specific CD4 + T cell populations induced by primary vaccination are activated and expand in the lymph node and circulating compartments prior to the development and migration of more specialized J o u r n a l P r e -p r o o f subpopulations that co-express CXCR5 and PD1 to the lymph node GC (Crotty, 2018) . In our S167-180 tetramer data, most S167-180-specific CD4 + T cells in the blood do not exhibit a circulating TFH phenotype even at the day 28 post-primary vaccination peak of circulating S167-180-specific TFH. Very few, if any, maintained S167-180-specific memory CD4 + T cells in the blood express both CXCR5 and PD1. Nevertheless, S167-180-specific TFH cells compose the largest or second largest S-specific TFH population in the lymph node of all evaluated subjects despite the near absence of these cells in the circulating blood at the same late time-points. Together, our data support a model whereby clonal populations of circulating CD4 + T cells develop into many different lineages, including the TFH cell lineage. Furthermore, we were unable to find a strong direct relationship between the cells known as circulating TFH (circulating antigen-specific CD4 + CXCR5 + PD1 + cells) and the presence of large populations of clonally-matched antigenspecific TFH participating in an ongoing GC in the lymph node. This is in contrast to data from a study of matched tonsil and blood samples in subjects who were not recently vaccinated or infected where they found substantial clonal overlap between tonsil TFH populations and circulating TFH populations but little overlap between tonsil TFH populations and circulating non-TFH populations (Brenna et al., 2020) . Further studies are required to determine the relationship between populations of circulating and lymph node resident TFH cells in both the steady state and following vaccination as these systems are quite distinct.

The discovered DPB1*04-restricted S167-180 response is notable for the extraordinarily constrained TCRα sequence diversity. This single TCRα motif is immediately obvious with even cursory inspection of bulk CD4 + TCR sequences from vaccinated or infected individuals.

Surprisingly, no prominent TCRβ motif is observed in any of our sequencing of this response, emphasizing the importance of the alpha chain in certain instances of specific epitope J o u r n a l P r e -p r o o f recognition (Dash et al., 2017; Minervina et al., 2020; Shomuradova et al., 2020) . The high prevalence of DPB1*04 in worldwide populations means that this response is likely immunodominant across multiple populations and contributes significantly to the measured responses in many studies, though its restriction has not been previously assigned. Thus far, none of the prevalent variant SARS-CoV-2 strains including the delta and omicron variants have acquired stable mutations in this peptide sequence.

In conclusion, we find that mRNA vaccine technology has an exceptional ability to induce high-frequency antigen-specific B cell (Turner et al., 2021) and antigen-specific CD4 + TFH cell responses in the human lymph node following prime-boost administration. These characteristics underlie the development of high titer neutralizing antibodies and protection from infection in vaccinated individuals. The selective enhancement of lymph node TFH responses induced by vaccine regimens represents a broad strategy for improving future vaccines.

Our study has several limitations, including the small number of included subjects, the relatively young age of the included participants, and the lack of comprehensive epitope mapping beyond the immunodominant response we identified. Furthermore, the complex nature of both the vaccination rollout during the ongoing pandemic and the FNA sampling procedure itself eliminated our ability to sample lymph nodes prior to vaccination and at earlier time-points following the primary vaccination. Furthermore, although we repeatedly sampled some axillary lymph nodes until the apparent conclusion of the GC response in those nodes, we were unable to sample non-draining control lymph nodes at distal sites. Limitations in the small number of available cells from the FNA procedure precluded total spike-specific T cell response measurement in the LN samples using assays such as AIM or ICS. In addition, limitations to the J o u r n a l P r e -p r o o f convalescent patient sample study precluded longitudinal analysis of these responses in the previously infected patient cohort. There are several questions that we did not address that will be useful topics for future studies, including the extent of clonal overlap between the blood and lymph node CD4 + T cell compartments, and the transcriptional profiles of the lymph node TFH response over the long period of clonal stability. We worked to ensure gender balance in the recruitment of human subjects. We worked to ensure ethnic or other types of diversity in the recruitment of human subjects. Meckiff et al., 2020) to identify potential partner TCRβ chains and then matched to the large MIRA dataset that used TCRβ sequencing (Nolan et al., 2020) to predict HLA-restriction and cognate epitope. To validate our prediction, we generated a T cell line expressing the putative αβTCR and we also generated HLA class II tetramers. The majority of S167-180 tetramer + cells retain an "effector memory" (CD45RO + CCR7 -) surface phenotype following vaccination. (D) A subset of S167-180 tetramer + cells undertake an "activated" surface phenotype (HLA-DR + CD38 + ) in the two weeks following vaccination. (E) ICOS and PD-1 are upregulated on the majority of S167-180 tetramer + cells prior to and 7 days following boost vaccination. (F) A small subset of S167-180 tetramer + cells undertake a "circulating TFH" surface phenotype (CXCR5 + PD1 + ) following boost vaccination, but the majority of circulating S167-180 tetramer + cells do not exhibit this phenotype. (G) S167-180 tetramer + CXCR5 + PD1 + cells as a percentage of total live CD3 + CD4 + T cells over time. Representative flow cytometry plots of subject 20 demonstrating the frequency of S167-180 tetramer + cells expressed as a percentage of total CD4 + T cells in the lymph node FNA sample (top row). The bottom row demonstrates CXCR5 and PD1 surface expression on the gated S167-180 tetramer + cells from the row above. (B) The percentage of total CD4 + T cells that are S167-180 tetramer + in blood (red lines) and FNA (blue lines) in matched samples taken at the same timepoints from subjects with available sample. (C) The percentage of CXCR5 + PD1 + T cells that are S167-180 tetramer + over time in both the blood (red lines) and FNA (blue lines). J o u r n a l P r e -p r o o f Figure S1 . Human lymph node TFH population frequency correlates with the GC B cell population frequency. Related to Figure 1 , Table 1 and Table S1 . (A) Gating strategy for the lymph node TFH (CD3 + CD4 + CXCR5 + PD1 + Bcl-6 + FoxP3 -) and GC B cell (CD19 + IgD low Bcl- (Dykema et al., 2021; Loyal et al., 2021) ) as well as stimulation of line expressing irrelevant TCR (specific to NQKLIANQF epitope from the spike protein of SARS-CoV-2, described in (Minervina et al., 2021b) with CTFEYVSQPFLMDLE peptide were used as negative controls. 

6 + CD38

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Ali H. Ellebedy (ellebedy@wustl.edu).

The HLA-DPB1*04 S167-180 HLA class II tetramer has been submitted to the NIH tetramer core facility (tetramer.yerkes.emory.edu). No other new unique reagents were generated in this study.

 Processed TCR sequencing data have been submitted to the GEO database under the accession number GSE183393, and the raw sequencing data have been linked to that dataset with the accession number SRP335569. All sequencing data is publicly available as of the date of publication. Any raw flow cytometry data not available in the supplemental tables will be shared by the lead contact upon request.

 This paper does not report original code.

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

Human subjects

Human subjects who elected to receive the BNT162b2 mRNA vaccine were recruited into this prospective observational study. Written informed consent was obtained from each subject. The study was approved by the Washington University in St. Louis Institutional Review Board (approval # 2020-12-081). Details of the entire study cohort have been previously reported (Turner et al., 2021) . The age and sex of the subjects included in the present study are listed in 

Cell sorting and flow cytometry for a total of 30 minutes on ice. Cells were then washed twice with P2 and live, singlet, CD4 + CD19 -CXCR5 + PD1 + cells were sorted on a FACSAria II into Trizol before being immediately frozen on dry ice.

To analyze antigen-specific B cell populations, we generated labeled recombinant soluble SARS-CoV-2 spike protein as previously described (Stadlbauer et al., 2020) . Following expression for 3 days at 27 ˚C, cell supernatant was concentrated and buffer exchanged in a Tangential Flow Filtration system into 500 mM NaCl, 10 mM Tris-HCl pH8 and subsequently purified via immobilised metal affinity chromatography and Superdex S200 gel permeation chromatography (GPC) in 150 mM NaCl, 10 mM Tris-HCl pH8. The linked CLIP peptide was cleaved with factor Xa for 6 h at 21°C prior to peptide exchange, and factor Xa cleaved HLA-DP4 was subsequently incubated in the presence of a 10-fold molar excess of peptide and a 1/5 molar ratio of HLA-DM for 16h at 37°C in 100 mM sodium citrate pH 5.4.

HLA-DP4 loaded with S167-180 peptide was buffer exchanged into 50mM NaCl, 20 mM Tris-HCl pH8, purified via Hi-Trap Q ion exchange chromatography and biotinylated using BirA biotin J o u r n a l P r e -p r o o f ligase. Following a final Superdex S200 GPC step in PBS, biotinylated HLA-DP4-S167-180 monomer was concentrated to approx. 1mg/ml and stored at -80 ˚C.

Biotinylated HLA-DP4-monomers loaded with TFEYVSQPFLMDLE peptide (S167-180)

were tetramerized using PE-Streptavidin (Biolegend). One volume PE-conjugated streptavidin was added to one volume of HLA-DP4-monomer (1 mg/mL). The volume of PE-streptavidin Flow cytometry data were analyzed using FlowJo software (BD Biosciences). The quality of the S167-180 HLA class II tetramer was judged by staining of the relevant T cell line and low background in irrelevant Jurkats and naive PBMCs. for subsequent scTCR sequencing. scTCR library preparation and sequencing was performed as previously described (Wang et al., 2012) . In brief, cDNA underwent two rounds of nested multiplex PCR amplification with a forward primer mix specific for V-segments and reverse primers for C-segments of TCRalpha and TCRbeta and sequenced on Illumina MiSeq platform (2x150 read length). TCR sequences with undefined alpha-chain were excluded from the analysis. Resulting TCR sequences can be found in Table S2 .

TCRalpha and TCRbeta bulk repertoires were generated with the 5'RACE protocol described in (Egorov et al., 2015) . RNA was isolated using Trizol reagent ( (Nolan et al., 2020) . Processed single cell paired chain TCR datasets from ARTE assays after 6 and 24 hour stimulation with SARS-CoV-2 peptides were used as supplied by authors in original publications: Table S3 from , Table S4A from (Meckiff et al., 2020) .

Bulk TCR repertoire data was demultiplexed and assembled into the UMI consensuses with migec (v. 1.2.7; with collision filter and force-overseq parameters set to 1) (Shugay et al., 2014) . V and J-segment alignment, CDR3 identification and assembly of reads into clonotypes were performed with MiXCR (v. 3.0.3) with default parameters . Resulting processed repertoire datasets and reference to raw TCR repertoire sequencing data are available at GEO database (acc. GSE183393). Analysis of bulk repertoire data was performed using R language for statistical computing, with merging and subsetting of data performed using data.table package. stringdist and igraph R packages were used to build TCR similarity network, gephi software was used for TCR similarity networks layout and visualization and ggplot2 library for other visualizations.

Descriptive and comparative statistics were employed in the manuscript as described in the figure legends with the number of replicates indicated.

J o u r n a l P r e -p r o o f Table S1 . Flow cytometry raw data. Related to Figure 1 , Figure S1 , Figure 3 and Figure 4 . The LN B cell and TFH tab includes raw data from the 95 lymph node samples evaluated in Figure S1 along with the number of CD4 + T cell events captured during the flow cytometry experiment.

The S 167-180 tetramer tab lists all raw data for Figure 3 and Figure 4 . Table S2 . Paired T cell receptor sequences. Related to Figure 2F . Sequences were obtained from sorted S167-180 tetramer + CD4 + T cells obtained from convalescent subject samples. Table S3 . T cell receptor bulk sequencing experiment quality metrics. Related to Figure 1C and Figure 5 . Sample cell counts along with sequencing quality metrics for analysis performed in 

Effect of 2 Inactivated SARS-CoV-2 Vaccines on Symptomatic COVID-19 Infection in Adults: A Randomized Clinical Trial

Safety and Immunogenicity of SARS-CoV-2 mRNA-1273 Vaccine in Older Adults

Low-Avidity CD4+ T Cell Responses to SARS-CoV-2 in Unexposed Individuals and Humans with Severe COVID-19

Efficacy and Safety of the mRNA-1273 SARS-CoV-2 Vaccine

MiXCR: software for comprehensive adaptive immunity profiling

CD4+ T Follicular Helper Cells in Human Tonsils and Blood Are Clonally Convergent but Divergent from Non-Tfh CD4+ Cells

Resistance of SARS-CoV-2 variants to neutralization by monoclonal and serum-derived polyclonal antibodies

Immune Correlates of Protection by mRNA-1273

Immunization against SARS-CoV-2 Infection in Nonhuman Primates (Immunology)

Follicular Helper CD4 T Cells (T FH )

Do Memory CD4 T Cells Keep Their Cell-Type Programming: Plasticity versus Fate Commitment

Quantifiable predictive features define epitope-specific T cell receptor repertoires

Functional characterization of CD4+ T cell receptors crossreactive for SARS-CoV-2 and endemic coronaviruses

Quantitative Profiling of Immune Repertoires for Minor Lymphocyte Counts Using Unique Molecular Identifiers

Identifying specificity groups in the T cell receptor repertoire

Regulation of germinal center responses and B-cell memory by the chromatin modifier MOZ

An mRNA Vaccine against SARS-CoV-2 -Preliminary Report

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

Three Doses of an mRNA Covid-19 Vaccine in Solid-Organ Transplant Recipients

Neutralizing antibody levels are highly predictive of immune protection from symptomatic SARS-CoV-2 infection

Germinal centre-driven maturation of B cell response to SARS-CoV-2 vaccination

Safety and efficacy of an rAd26 and rAd5 vector-based heterologous prime-boost COVID-19 vaccine: an interim analysis of a randomised controlled phase 3 trial in Russia

Cross-reactive CD4 + T cells enhance SARS-CoV-2 immune responses upon infection and vaccination

Imbalance of Regulatory and Cytotoxic J o u r n a l P r e -p r o o f SARS-CoV-2-Reactive CD4+ T Cells in COVID-19

Primary and secondary anti-viral response captured by the dynamics and phenotype of individual T cell clones

Longitudinal high-throughput TCR repertoire profiling reveals the dynamics of T-cell memory formation after mild COVID-19 infection

Convergent epitope-specific T cell responses after SARS-CoV-2 infection and vaccination

A subset of HLA-DP molecules serve as ligands for the natural cytotoxicity receptor NKp44

A large-scale database of T-cell receptor beta (TCRβ) sequences and binding associations from natural and synthetic exposure to SARS-CoV-2

Rapid induction of antigen-specific CD4+ cellular immunity to SARS-CoV-2 mRNA vaccination

Broad and strong memory CD4+ and CD8+ T cells induced by SARS-CoV-2 in UK convalescent individuals following COVID-19

Detecting T cell receptors involved in immune responses from single repertoire snapshots

Safety and Efficacy of the BNT162b2 mRNA Covid-19 Vaccine

T follicular helper cells in space-time

Transcriptional and Metabolic Control of Memory B Cells and Plasma Cells

Synchronous infection of SIV and HIV in vitro for virology, immunology and vaccine-related studies

Safety and Efficacy of Single-Dose Ad26.COV2.S Vaccine against Covid-19

A molecular basis for the association of the HLA-DRB1 locus, citrullination, and rheumatoid arthritis

Epitopes Are Recognized by a Public and Diverse Repertoire of Human T Cell Receptors

Towards errorfree profiling of immune repertoires

SARS-CoV-2 seroconversion in humans: a detailed protocol for a serological assay, antigen production, and test setup

Comprehensive analysis of T cell immunodominance and immunoprevalence of SARS-CoV-2 epitopes in COVID-19 cases

Failure to seroconvert after two doses of BNT162b2 SARS-CoV-2 vaccine in a patient with uncontrolled HIV

Single Naive CD4+ T Cells from a Diverse Repertoire Produce Different Effector Cell Types during Infection

Human germinal centres engage memory and naive B cells after influenza vaccination

SARS-CoV-2 mRNA vaccines induce persistent human germinal centre responses

Pathophysiology of T follicular helper cells in humans and mice

The dawn of mRNA vaccines: The COVID-19 case

Germinal Centers

Safety and efficacy of the ChAdOx1 nCoV-19 vaccine (AZD1222) against SARS-CoV-2: an interim analysis of four randomised controlled trials in Brazil, South Africa, and the UK

Safety and Immunogenicity of Two RNA

Based Covid-19 Vaccine Candidates

Antibody resistance of SARS-CoV-2 variants B.1.351 and B.1.1.7

mRNA vaccine-elicited antibodies to SARS-CoV-2 and circulating variants

A Temporal Switch in the Germinal Center Determines Differential Output of Memory B and Plasma Cells

Durability of Responses after SARS-CoV-2 mRNA-1273 Vaccination

Serum Neutralizing Activity Elicited by mRNA-1273 Vaccine

CoV-2 vaccines induce robust human TFH responses in draining lymph nodes  HLA-DPB1*04 restricts the immunodominant SARS-CoV-2 S167-180 epitope  S167-180 is recognized by T cell receptors with a public alpha chain motif  S-specific TFH are maintained in draining lymph nodes 6 months after vaccination Analysis of draining lymph nodes of individuals vaccinated BNT162b2 mRNA vaccine against SARS-CoV-2 identifies viral spike-specific follicular helper CD4+ T cells that persist for months

J o u r n a l P r e -p r o o f