Characterizing Cellular and Molecular Variabilities of Peripheral Immune

Home , PREPARE, RNA vaccine

medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 Characterizing cellular and molecular variabilities of peripheral immune cells in healthy 2 inactivated SARS-CoV-2 vaccine recipients by single-cell RNA sequencing 3 4 Renyang Tong1#, Jianmei Zhong1#, Ronghong Li1, Yifan Chen1, Liuhua Hu1, Zheng Li1, 5 Jianfeng Shi1, Guanqiao Lin1, Yuyan Lyu1, Li Hu1, Xiao Guo1, Qi Liu1, Tian Shuang1, Chenjie 6 Zhang1, Ancai Yuan1, Minchao Zhang2, Wei Lin1*, Jun Pu1*

8 1 State Key Laboratory for Oncogenes and Related Genes, Division of Cardiology, School of 9 Medicine, Renji Hospital, Shanghai Cancer Institute, Shanghai Jiao Tong University, Shanghai, 10 China.

11 2 Colortech (Suzhou) Biotechnology, Suzhou, China.

12 13 * To whom correspondence should be addressed:

14 Jun Pu, MD, Ph.D.; State Key Laboratory for Oncogenes and Related Genes, Division of 15 Cardiology, School of Medicine, Renji Hospital, Shanghai Cancer Institute, Shanghai Jiao Tong 16 University, Shanghai, China. E-mail: [email protected].

17 Wei Lin, Ph.D.; State Key Laboratory for Oncogenes and Related Genes, Division of 18 Cardiology, School of Medicine, Renji Hospital, Shanghai Cancer Institute, Shanghai Jiao Tong 19 University, Shanghai, China. E-mail: [email protected].

20 21 #These authors contributed equally to this work

22 23 24 25 26 27 28 29 30 31 32 NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice. 33 medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 Abstract

2 We systematically investigated the transcriptomes of the peripheral immune cells from 6

3 inactivated vaccine, BBIBP-CorV recipients at 4 pivotal time points using single-cell RNA-seq

4 technique. First, the significant variation of the canonical immune-responsive signals of both

5 humoral and cellular immunity, as well as other possible symptom-driver signals were

6 evaluated in the specific cell types. Second, we described and compared the common and

7 distinct variation trends across COVID-19 vaccination, disease progression, and flu vaccination

8 to achieve in-depth understandings of the manifestation of immune response in peripheral blood

9 under different stimuli. Third, the expanded T cell and B cell clones were correlated to the

10 specific phenotypes which allowed us to characterize the antigen-specific ones much easier in

11 the future. At last, other than the coagulopathy, the immunogenicity of megakaryocytes in

12 vaccination were highlighted in this study. In brief, our study provided a rich data resource and

13 the related methodology to explore the details of the classical immunity scenarios.

14 15 Keywords: COVID-19; Inactivated SARS-CoV-2 Vaccines; Peripheral Immune Cells; Single 16 Cell Sequencing

17 medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 2 Introduction 3 As of May 2021, World Health Organization (WHO) has reported 148 million Coronavirus 4 disease 2019 (COVID-19) infections and 3.1 million deaths worldwide 5 (https://covid19.who.int/). Promotion and popularization of vaccine is vital to disease control, 6 with its safety and efficacy of the top priority. There are multiple SARS-CoV-2 vaccines 7 available and several vaccines under development, including the inactivated, genetically 8 engineered subunit, adenovirus and nucleic acid vaccine types. So far, a number of institutions 9 of different countries and regions have reported the results of safety and immunogenicity of 10 SARS-CoV-2 vaccine on phase III clinical trials or phase III interim analysis, such as, vaccines 11 BNT162b21 and mRNA-12732, adenovirus vaccines AZD12223, Ad5-nCoV4 and 12 inactivated vaccines BBIBP-CorV5 and CoronaVac6, 7, and protein subunit vaccine NVX-23738. 13 For example, Ugur Sahin et al.9 demonstrated that RNA vaccine (BNT162b2) not only elicited 14 adaptive humoral, but also stimulated the cellular immune response even at the low-dose 15 vaccine. The new vaccine types such as adenovirus-vectored vaccines and RNA vaccines have 16 emerged and fast developed with good efficacy in recent years, though severalserious adverse 17 events have been reported10, 11 12, 13. Their efficacy and long-term safety were yet to be reviewed 18 by a follow-up study. With the support of long clinical practice, traditional inactivated vaccines 19 play an irreplaceable role in the prevention and control of infectious disease. As a pattern of 20 inactivated SARS-CoV-2 vaccines, BBIBP-CorV, was manufactured by Sinopharm China 21 National Biotec Group (CNBG) and contained a SARS-CoV-2 strain inactivated inside Vero 22 Cells14, which was confirmed for its safety, tolerability, and immunogenicity in clinical trials of 23 healthy people in China5 and UAE, Bahrain, Egypt and Jordan (https://cdn.who.int/) . 24 Single-cell RNA-seq (scRNA-seq) technique has demonstrated its power in the research 25 of cell biology by deconvoluting the transcriptional signals in the multi-cellular samples of 26 heterogenous cellular compositions15,16. Other than characterizing the cell types in a 27 heterogeneous tissue, it enables us to detect the variabilities within a cell lineage. The recent 28 pandemic of COVID-19 has prompted the demand of single-cell studies in order to uncover the 29 characteristics immune responses of different immune cell types during the symptom 30 development of COVID-1917, 18, 19, 20. These published scRNA-seq datasets provide abundant 31 and detailed reference information of this disease, though, no results have been reported for any 32 COVID-19 vaccines yet. Moreover, a comprehensive comparison of the human immune 33 responsive signals across the transcriptome at single-cell resolution over the time becomes 34 practical and necessary. With no doubt, such detailed comparison facilitates the in-depth medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 understanding of the pathogenesis, symptom development and immunization of COVID-19 2 which guides us to the better vaccination strategy. Therefore, we aimed to look through a 'sight 3 glass' of the human immunity system, a.k.a., peripheral blood mononuclear cells (PBMCs) to 4 monitor the manifestation of vaccine response of the canonical immune cells. In the meantime, 5 we could take the advantage of the recent published scRNA-seq datasets of COVID-19 patient 6 samples21 and Flu vaccination samples22 to achieve such a goal. Besides, the roles of 7 megakaryocytes/platelets in COVID-19 recently have attracted some attentions due to recent 8 reports of blood-clot side effect by AZD122212, 23. Megakaryocytes were also found to 9 contribute to the complication of COVID-19 patients and be correlated to the organ failure and 10 mortality24, 25, 26. We were therefore also interested in looking into the signals related to immune- 11 responsive and coagulopathy in this non-canonical immune cell type using the scRNA-seq 12 dataset. 13 In this study, we extensively investigated the PBMC samples of BBIBP-CorV vaccine 14 recipients. Our scRNA-seq data delineated a high-resolution transcriptomic landscape of 15 peripheral immune cells during the vaccination of a classical vaccine type. Meanwhile, we 16 integrated the PBMC scRNA-seq datasets of COVID-19 patients and flu vaccination in order 17 to identify the common and distinct variation trends among our data set and the two other data 18 sets. All these efforts could provide a rich data resource and demonstrate a methodology to 19 explore the human immunity in disease development and vaccination at great details. Moreover, 20 this work could facilitate the in-depth understandings of the manifestation of immune response 21 in peripheral blood under different stimuli and provide many clues to design the new assay for 22 the assessment of the immune protection and adverse effects after the vaccination or infection 23 recovery.

24 25 Results 26 Overview of immune responses and peripheral immune cells variations by scRNA-seq 27 To assess the immune cells content and the phenotypic alterations following the vaccination 28 with inactivated SARS-CoV-2 vaccine (BBIBP-CorV), six healthy young adults with 4 males 29 and 2 females from Pu Dong Cohort, China, were enrolled in this 2019-nCoV vaccine 30 observational study [NCT04871932] according to inclusion and exclusion criteria (Fig.1A). All 31 six volunteers received 2 doses of inactivated SARS-CoV-2 vaccine (BBIBP-CorV). Blood 32 samples were collected at 4 time points including the day before vaccination (BV), 7 days after 33 first dose (1V7), 7 days after second dose (2V7) and 14 days after second dose (2V14) (Fig. 34 1A). Then, serum and PBMCs were separated from all time points of samples within 2-hour of medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 collection. The 2019-nCoV Spike RBD antibodies and cytokines were measured in serum. After 2 isolating the PBMCs, we profiled V(D)J repertoires of T and B cells integrated with 5' Gene 3 Expression using droplet based single-cell sequencing method (10× Genomics) (Fig.1A). 4 Based on the integrated single-cell gene expression profiles, we recovered the 5 transcriptomes of ~120k high-quality cells in total from 24 samples and the total cells were 6 combined and projected in a 2-D space using the t-distributed stochastic neighbor embedding 7 (t-SNE). Unsupervised clustering analyses and marker mapping enabled us to uncover 11 major 8 types (Fig.1B). The markers for the 11 major immune cell lineages were shown on Fig.1C. As 9 the shown in Fig.1D, summary statistics of the 11 cell-type fractions showed plasma cells 10 increase significantly after the second dose, whereas the other 10 cell types pretty much 11 remained the same level. 12 Previously, Wang et al.14, Xia et al.5 have already demonstrated the satisfactory efficacy 13 by increasing seroprotective antibody titers in the animal experiments and human clinical trials 14 by this vaccine type. To further validate the efficacy in our participants, we measured 2019- 15 nCoV Spike RBD antibodies using indirect enzyme-linked immunosorbent assay (ELISA) kits. 16 Anti-S-RBD-specific IgG antibody level significantly increased after the second dose for total 17 6 participants (Fig.1E). The growth of plasma cells content and the observable production 18 neutralizing antibody by second dose suggests the robust humoral immune response. The 19 cellular immunity was also assessed. 10-cytokine serum assays were first performed on the total 20 24 samples on bead-based flow-cytometric assays on the BD LSRFortessa X-20 platform, 21 though, no obvious post-vaccination increase was observed (Fig.S1A). We then used an ultra- 22 sensitive bead-array-based immunoassay (single-molecule array technique) to validate the IL- 23 4 signal and found that the quantification was consistently lower (one order of magnitude) than 24 the readings of BD LSRFortessa X-20 platform (Fig.S1B), which suggested, 1. most of the 25 cytokines remained low level at these time points. 2. limit of quantification of bead-based flow- 26 cytometric assays likely above the real concentrations of these cytokines in these samples (Fig. 27 S1B). 28 29 Classical immune responsive signals of major immune cell types in vaccination. 30 After the successful characterization of the major cell types, we were interested in investigating 31 the gene expression patterns in the major immune cell types. The differentially expressed genes 32 (DEGs) were first identified in 11 cell types by the comparisons between the post-vaccination 33 samples vs. the pre-vaccination samples (BV). The DEGs list was shown in Supplementary 34 table 1. The unsupervised clustering of these DEGs exhibited a pattern of 3 dominant clusters medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 for the 24 samples in most of the cell types. We used the heatmap of B cells as an example to 2 demonstrate the sample clusters (Fig.2A). The red cluster included the samples mostly in the 3 2V7 or 2V14 time points, which reflected an active immune responsive state (R-state). The 4 green cluster included the samples mostly in early BV or 1V7 time points, which likely reflected 5 the baseline or minor-responsive state (B-state). The blue cluster covered broader time points, 6 which likely reflected a post-vaccination protection state (P-state) or recovery state. The 7 identical clustering patterns were observed when the regulons heatmap were calculated and 8 plotted using SCENIC27 (Fig.2B). Such unsupervised clustering provided an unbiased sample 9 grouping. More details of the signal variabilities by heatmaps for the major 6 cell types were 10 shown in Fig.S2A-2J. 11 The DEGs were re-calculated based on the updated sample clusters (B-R-P) and the 12 following 15 variable signaling keywords were identified based on these DEGs. The updated 13 DEGs list for this comparison was shown in Supplementary table 2 and showed very little 14 change. The coordinated gene expressions of a certain functional group or signaling pathway 15 suggested their involvement in the response to BBIBP-Corv vaccination. We used simple 16 schematic diagrams to illustrate the 5 dominant patterns of signal variation observed in the 17 heatmaps to delineate the variability of immune-responsive signals in B-R-P state as Fig.S2K. 18 As expected, we observed the elevation of the monocytes' type I interferon production 19 signal (Fig.2C, pattern 2), which was an essential signal of innate immunity for viral 20 suppression. We also observed significant elevation of the antigen-specific follicular T helper 21 cell (Tfh) activation (CD40LG, ICOS, SLAMF1, etc.) (Fig.2D, pattern 2) and the B cell 22 activation signal (TLR7, CD80/CD86, BCL6) (Fig.2E, pattern 2), that were associated with 23 neutralizing antibody production. In the meantime, we evaluated the expression of different 24 classes of constant gene of Immunoglobulin heavy chain. The IGHG went up and IGHM went 25 down in protection state (P-state), suggesting the occurrence of class switching of neutralizing 26 antibody molecules (Fig.2F). 27 TCR responsive signals (VAV3, AKT3, MALT1, MAPK1, etc.) (Fig.2G, pattern 1), as well 28 as cytotoxic signal were also enhanced (Fig.2H, pattern 3). These lymphocyte-related signals 29 contributed to antigen-specific immunity, a.k.a., adaptive immunity. 30 The immune regulatory signals such as TGF-β pathway (TGFB1, TGFBR1/2, SMAD2/3/4, 31 etc.) (Fig.2J pattern 5), and T cell exhaustion (LAG3, PDCD1) (Fig.2I pattern 4) were 32 suppressed in the high immune-responsive samples (R-state) until they went up to play their 33 roles in the protection state (P-state). 34 We also identified some changes contradictory to the expectation. For example, NF-κB medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 signal was expected to be upregulated in the most of the immunity situation (Fig.S2L). However, 2 NF-κB signal actually went down (pattern 4). Second, instead of enhancing the antigen 3 presenting, MHC I signal significantly dropped in most of the peripheral immune cell types in 4 R group (Fig.2L, pattern 4). Though, type II MHC genes mostly went up as expected (Fig.2K, 5 pattern 2). Later, when we examined COVID-19 dataset21, we didn't observe such changes in 6 COVID-19 patients. Why this change occurs and what outcomes will this change brings are 7 elusive. 8 Other variabilities of the classical immune-responsive signals including cell-cycles and 9 apoptosis (Fig.S2M-2N) were summarized in Supplementary table 3. High cell-cycle signal 10 (ORC2, CCNH, CDC7, CDK7, etc.) and low apoptosis signal (BAX, BAD, TP53, etc.) indicated 11 the high proliferative potential of the cells in R-state samples. 12 13 Characteristic transcriptions of the top TCR and BCR clones

14 As our scRNA-seq study also included TCR and BCR V(D)J sequencing results, it allowed us 15 to track down the TCR/ BCR clonal expansion in these samples. The results showed the 16 diversities of the T cell and B cell clones wend down after vaccination due to the expansion of 17 the antigen-specific clones (Fig.3A, 3D and 3G). The top 20 abundant clones of both T cells 18 and B cells were pulled out to investigate their characteristics in comparison with the low 19 abundance clones as the counterpart.

20 First of all, it was found that top BCR clones were enriched in the samples after second 21 dose, indicating the observable clone expansion occurred after the second dose. Second, it was 22 found that the top 20 clones were highly correlated to low MHC II signal and high B cell 23 activation signal (Fig.3B and 3C). This was expected, though, it was still encouraging to 24 confirm that the possible antigen-specific B cells could display transcriptional signature in the 25 peripheral blood, which made it more specific to access the antigen-specific BCR sequence 26 using single-cell sequencing technique.

27 Top TCR clones didn't necessarily occur in the post-vaccination time, which made it less 28 straightforward to confirm their vaccine-specificity. We were still able to find the correlation 29 between the top T cell clones and the immune-responsive signals. In CD4+ T cells, it was found 30 that the top clones were associated with high follicular T helper (Tfh) and low NF-κB signal 31 (Fig. 3E and 3F). In CD8+ T cells, it was found that the top clones were associated with high 32 cytotoxic but low NF-κB signal (Fig.3H and 3I).

33 medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 Non-trivial roles of megakaryocytes in vaccine immunity

2 Megakaryocytes were mostly recognized for its role in the hemostasis and wound healing. Little 3 has been discussed for its contribution in the disease or vaccine immunity. After examining this 4 atypical immune cell type, we observed the gene expression variation of megakaryocytes in 5 vaccine immunity.

6 First, the variation of platelet aggregation signal (GP9, ITGB3, TBXA2R, FLNA, VASP, 7 etc.) in megakaryocytes (Fig.4A and 4B, pattern 4) were identified. These genes were 8 significantly down-regulated in the immune-responsive samples (R-state), which suggested low 9 risk of thrombosis during the period of BBIBP-CorV vaccination. Later, when we inspected the 10 same set of genes in the COVID-19 patients21 and influenza vaccination22 samples at the 11 corresponding immune-responsive time point, instead of getting down-regulated, we found the 12 platelet aggregation signal significantly went up (dashed red box in Fig.4E and Fig.S3D). That 13 means, the variation of platelet aggregation signal could be distinct by different vaccines.

14 As a myeloid-lineage cell type, it was found that megakaryocytes actually contributed 15 more inflammatory signal than we thought, such as interleukin-1 beta (IL-1β) production (TLR4, 16 IFI16, CASP1, CASP8, NLRP1, NLRP2, etc.) and neutrophil activation (SA0018, S100A12, 17 FCER1G, VNN1, etc.) (Fig.4A, 4C and 4D). The expression levels of the related genes were 18 actually as high as the monocyte population (Fig. S2O and S2P), which was considered the 19 dominant cell type of innate immunity in PBMCs. In flu vaccination, we observed the elevation 20 of IL-1β production and neutrophil activation signal as well at 28-day sample, the putative high 21 response time point (Fig. 4F and 4G). All these results suggested the roles of megakaryocytes 22 in pathogenesis and vaccination should not be overlooked.

24 Transcriptomic similarity and distinction of COVID-19 and vaccination immunity in 25 PBMC 26 It was postulated that the dysregulated immune response could be reflected in differential 27 signals between the COVID-19 infection patients and the vaccine recipients. We were 28 particularly interested in looking for the signatures that were associated to the symptom 29 development and the later immune protection. By examining the similarity and the distinction 30 of the developmental trend of these signals, we aimed to gain more insights of both disease 31 development and vaccine protection. 32 We used the keywords in Supplementary table 3 to assess the COVID-19 samples. There 33 were three natural sample groups in previous COVID-19 study, i.e., healthy control, progression medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 samples of mild and severe symptoms and the convalescent samples of both patient types21. 2 Other than the resemblance between COVID-healthy samples and the vaccine-baseline samples, 3 it was intuitively believed that the COVID-convalescence samples were similar to the vaccine- 4 protection samples (P-state). What do these immune signals vary in the progression samples are 5 of our interest. 6 Many of the abovementioned signals only showed an obvious turning point at 7 convalescence group, including type I interferon productions, Tfh and B cell activation, TGF- 8 β, cell-cycle, apoptosis and so forth (Fig.5A. and Fig.S3A-S3B). We were able to observe 9 obvious signal boundary line between healthy/progression samples and convalescence samples. 10 That means, unlike what happened in vaccination, some of the critical signals of immunity were 11 not initiated but suppressed until the recovery.

12 In general, there were many immune responsive signals in certain cell types in the 13 progression samples whose trends were similar to those in R-state of vaccination. For examples, 14 cytotoxic signal in T cells and NK cells significantly increased in the patients’ progression 17, 28 15 samples consistent with Zhang et al. and Wilk et al. study (Fig.S3C).

16 We also revealed the distinctions between disease immunity and vaccination immunity 17 using both datasets, especially for type I and type II MHC. In those antigen presenting cells like 18 B cells and monocytes, type I MHC signal significantly decreased in vaccine response group 19 (R group) whereas it tended to increase in the progression patients (Fig.5B). Type II MHC signal 20 was suppressed in progression in both mild or severe patients (Fig.5B), whereas it was up- 21 regulated during the vaccination. Another distinction was that T cells exhaustion signal went 22 down in vaccination but went up in progression patients (Fig.S3C). 23 In summary, we identified the distinct signal variation patterns between the COVID-19 24 progression and vaccination samples, such as 1. type I and II MHC signals 2. T cells exhaustion 25 signal. 26 27 Potential effects of the identified variable classical immune signals in flu vaccination 28 To further evaluate the responsive role of the variable classical immune signals we had 29 identified in inactivated SARS-CoV-2 vaccination, a specific comparison was adopted with a 30 public scRNA-seq data set on the lymph nodes and PBMC samples of the quadrivalent 31 influenza vaccination22. This data set was designed for a single-dose vaccine and included 5 32 time points of PBMC samples with a long post-vaccination follow-up time (2 months). 33 Using the 15 variable signaling keywords identified in BBIBP-Corv vaccination and the 34 involved differential genes, we calculated their modular scores for each cell in flu dataset. The medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 signal curves showed that 28-day sample gave the most obvious variation in response to the flu 2 vaccine compared to other time points. We were able to detect high B cell activation signal in 3 B cells (Fig.5C), high cytotoxic signal in CD8+T cells (Fig.5D), and low TGF-β signal (Fig.5E) 4 in most of the immune cells at 28-day time point. 5 Elevated B cell activation, elevated cytotoxic signals and suppressed TGF-β signal at 28 6 days suggested this sample was mostly corresponding to the COVID-19-vaccine samples at 7 immune R-state. We were not able to detect the variation of other signature signals, possibly 8 because their sampling missed the pivotal time point on this volunteer. The megakaryocytes 9 also had shown the significant variation at the same time point, which was mentioned earlier 10 (Fig.4E). 11

12 Discussion 13 In this study, we extensively investigated the variability of the single-cell transcriptomes of the 14 peripheral immune cells in the process of vaccination by a classical inactivated viral vaccine. 15 Single-cell sequencing technology enabled us to deconvolute the immune-responsive signals 16 into specific cell types. 17 Unsupervised clustering on the cell types from these samples defined three dominant 18 clusters of the immune cells that were putatively corresponding to different immune-responsive 19 states. Such an unbiased method overcame the challenge of sample grouping by the distinct 20 immune-background, vaccine susceptibility or variable pace of immune responses by different 21 participants. 22 The coordinated up-/down- regulation of the gene expressions of a certain signaling 23 pathway suggested their involvement of the immune response during the vaccination. Tracking 24 down the variation of these signals in a typical vaccination process allowed us to uncover their 25 possible roles in the immune protection. Using this method, we observed the activation of both 26 humoral immunity and cellular immunity by BBIBP-CorV, including cytotoxicity, follicular 27 helper T mediated immunity as well as B cell activation et al. Meanwhile, our results provided 28 a good reference for the future study of other vaccine types. 29 Surprisingly, we observed the down-regulation of NF-κB signal in the post- BBIBP-CorV 30 vaccination PBMC samples. Previous study suggested that the suppression of NF-κB signal 31 likely alleviated the vaccine adverse reaction29, 30, which might partially explain much less 32 adverse reaction cases was reported by the vaccine recipients of BBIBP-CorV. Even though, 33 the antibody productions of our participants were satisfactory. 34 The correlation between the top TCR/BCR clones and the modular scores of the keywords medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 reveals what phenotypes have been associated with the clones of expansion. Some of the 2 signature transcriptions identified in this work enabled us to capture the target B cells for full- 3 length sequencing and molecular cloning of the antigen-specific antibodies. 4 The platelet aggregation signal and inflammatory signals were also inspected in the 5 megakaryocytes due to the rising concern on the blood-clot side-effects reported on several 6 vaccine types10,11, 12. This cell type was usually overlooked in immunity, though, in response to 7 vaccine, the inflammatory gene expressions in megakaryocytes not only commensurate with 8 the monocytes but also in a coordinated fashion. In addition, unlike severe COVID-19 samples 9 and influenza vaccine samples, our result showed a down-regulation of blood-clot-related gene 10 expression in our samples under active response (R-state), which likely explained the much 11 fewer reports of thrombus-related adverse events by BBIBP-CorV vaccine5.These findings 12 necessitate a more extensive investigation on the roles of the megakaryocytes in the human 13 immunity in the future. 14 The comparison of the immune-responsive signals between the COVID-19-progression 15 and the vaccination allowed us to identify the discrepancy of immune responsive signals and 16 made it possible to locate the missing puzzle pieces of the unsuccessful or incomplete immunity 17 that contributed the symptoms in patients. These signaling pathways could be used to assess the 18 immunity or as the target of the intervention for the control of COVID-19 symptoms or the 19 vaccine adverse reaction. Though, more validation experiments will need to be done in the 20 future. 21 We also inspected these signals using the dataset generated from a single-dose flu vaccine 22 recipient22. The results showed a significant fluctuation, especially, the B cell activation signal 23 displayed a peak at 28-day. This was concordant to the detection of high content of 24 CD20hiCD38int B cells from lymph nodes at 28-day in the original paper, though, a single case 25 study might limit the power of this study. 26 Our study has some limitations. Due to the currently high cost of single-cell sequencing 27 and inaccessibility of other vaccine type at the moment, the coverage and the size of sampling 28 of this study are relatively limited, which limits the power of this study. Due to the urgent need 29 of COVID-19 related research, the follow-up time couldn’t be long enough, either. 30 Despite of these limitations, we were able to observe the protective immunity induced by 31 BBIBP-CorV at the single-cell level and identified a set of genes that not only featured the 32 immune-responsive variation of the peripheral immune cells but also exhibited many 33 interpretable signaling pathways. The signature expression described in here might facilitate to 34 design a new assay to assess the immune protection and adverse effects after the vaccination or medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 infection recovery. 2

3 Materials and Methods

4 Participants and ethics

5 All human samples used in this study were processed under Institutional Review Board 6 approved protocols at Shanghai Jiao Tong University. All study participants were recruited after 7 providing informed consent and with approval by the Ethics Committee of Ren Ji Hospital 8 (KY2021-046), School of Medicine, Shanghai Jiao Tong University, and the study was 9 conducted according to the criteria set by the Declaration of Helsinki31 (2013). This study was 10 registered with ClinicalTrials.gov (NCT04871932). A written informed consent letters were 11 routinely obtained from all participants in the study and all relevant ethical regulations 12 regarding human research participants were complied with by the investigators. Six healthy 13 non-frail individuals who had not received any vaccine in the past one year were recruited in 14 the Pu Dong Cohort, China, including 4 males and 2 females, aged 30-40 years old. All 15 participants were confirmed negative result of SARS-CoV-2 infection by RT-qPCR tests and 16 negative result for serum SARS-CoV-2-specific IgM/IgG-antibody ELISA tests. The 17 participants had no history of epidemiological contact of COVID-19 patients. They were free 18 of clinical symptoms including fever, cough, headache, sore throat, malaise, loss of smell, runny 19 nose, abdominal pain, diarrhea, joint pain, wheezing or dyspnea within 28 days before 20 vaccination; They were also free of any steroid usage, chronic diseases, pregnancy, lactation, 21 and obvious allergy to any known ingredients contained in the inactivated vaccine.

23 Vaccine and Vaccination protocol

24 The inactivated SARS-CoV-2 vaccine (Vero cells), a single-dose schedule of 4μg/0.5ml, were

25 supported by the Beijing Institute of Biological Products (Beijing, China). Vaccine recipients

26 received BBIBP-CorV twice according to drug instruction via intra muscular with 2 to 4 weeks

27 between each injection.

29 Sample collection, preparation and storage

30 Peripheral venous blood samples were collected at 4 time points including the day before

31 vaccination (BV), 7 days after first dose (1V7), 7 days after second dose (2V7) and 14 days

32 after second dose. The blood samples were processed within 2 hours of collection. The medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 peripheral blood mononuclear cells (PBMCs) were isolated using Vacutainer CPT tubes (BD)

2 according to the manufacture protocols32, and cells were immediately used 10× single-cell

3 RNA-seq or cryopreserved in 10% dimethylsulfoxide in FBS and stored at -80℃ freezer. The

4 viability and quantity of PBMCs in single-cell suspensions were determined using Trypan Blue.

5 The cell viability of total samples is >90%. Blood samples in the coagulation promoting tubes

6 were set upright on the tube stand for 20 minutes at 4 °C and were centrifuged at 1000g for 10

7 minutes at 4 °C. The supernatant serum samples were extracted and stored at -80 °C for future

8 assays.

10 Antibody and Serum cytokine detection 11 The levels of anti-SARS-CoV-2 (2019-nCoV) Spike RBD IgG in serum were tested using Sino 12 Biological® SARS-CoV-2 (2019-nCoV) Spike RBD Antibody Titer Assay Kit (Catalog 13 Number: KIT002) following the manufacturer’s instructions. Absorbance was read at 450 nm 14 with a subtraction at 570 nm by MultiSkan FC reader (51119000, Thermo Fisher Scientific Inc., 15 Waltham, MA, USA). 16 The protein levels of IFN-γ, IFN-α, interleukin (IL)-1β, IL-2, IL-4, IL-5, IL-6, IL-10, IL- 17 17A and Tumor necrosis factor (TNF) -α in serum were tested using Aimplex kit (Hang zhou 18 CEGER P010041), a multiplex beads-based flow cytometric assay, following the 19 manufacturer’s instructions. The concentrations of IL-4 in the serum samples were further 20 assayed using the ultra-sensitive single molecule array immunoassay on Medivh SX-L platform 21 (ColorTech Biotechnology, Suzhou, China). Samples were incubated with anti-IL-4 capture 22 antibody coated magnetic beads and biotinylated detection antibody to form the immuno- 23 complex. Then beads were loaded into femto-sized array for detection. 24

25 Single-Cell RNA-seq library construction and sequencing 26 Using a Single Cell 5’ Library and Gel Bead kit (10× Genomics, 1000006) and Chromium 27 Single Cell A Chip kit (10× Genomics, 120236), the cell suspensions (600–1000 living cells 28 per ul as determined by Count Star) were loaded onto a Chromium single cell controller (10× 29 Genomics) to generate single-cell gel beads in the emulsion (GEMs) following the 30 manufacturer’s protocol. Approximately 12,000 cells were added to each channel and 31 approximately 7,000 target cells were recovered. Single-cell RNA libraries were prepared using 32 the Chromium Single Cell 5’ v2Reagent (10× Genomics, PN-1000263) was used to prepare medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 single-cell RNA libraries according to the manufacturer’s instructions. Each sequencing library 2 was generated with a unique sample index. The libraries were sequenced using Illumina 3 platform with a sequencing depth of at least 100,000 reads per cell and 150 bp (PE150) paired- 4 end reads (performed by GENE SHINE, Shanghai). 5

6 Single-Cell Data Processing 7 The matrices of unique molecular identifier (UMI) were generated for each sample by the Cell 8 Ranger (10× Genomics, Version 4.0.0) Pipeline coupled with human reference version GRCh38 9 (10× Genomics, Version 3.0.0.) using STAR (version 2.5.1b). Then the expression matrix was 10 analyzed by R software (v.3.6.0) with the Seurat package (v.3.0.0) and DoubletFinder (version 11 2.0.2) for filtering, data normalization, dimensionality reduction, clustering (Butler et al., 2018). 12 The filtered scRNA-seq data were analyzed with the following steps: 1. To remove the cells 13 with low quality, cells with fewer than 500 genes detected and a mitochondrial gene ratio of 14 greater than 10% were excluded. And genes with at least one feature count in more than three 15 cells were used for the following analysis. 2. To remove potential doublets, cells with UMI 16 counts above mean ± 2SD are filtered out. Additionally, we applied DoubletFinder package 17 (version 2.0.2) to identify potential doublets. 3. For each cell, we normalized the gene 18 expression profiles with SCT-transform in Seurat package. 4. Variable genes were selected 19 using the ‘FindVariableFeatures’ function with the default parameters. 5. Variable genes were 20 projected onto a low-dimensional subspace using canonical correlation analysis (CCA) across 21 samples to correct batch effects by using the ‘RunMultiCCA’ functions. 6. A shared nearest 22 neighbor graph was constructed based on the Euclidean distance in the low-dimensional 23 subspace spanned by the selected significant principal components. Cells were clustered using 24 the ‘FindClusters’ function at an appropriate resolution. 7. Cells were visualized using a 2- 25 dimensional t-distributed stochastic Neighbor Embedding (tSNE) algorithm with the 26 ‘RunTSNE’ function. Using the above pipeline, we processed the scRNA-seq data of ~120k 27 high-quality cells from 24 samples.

29 Reference Single-Cell Datasets in comparison

30 Other than the scRNA-seq dataset generated in this study on the participants of BBIBP-CorV

31 vaccination, we also used the published PBMC scRNA-seq dataset to make meaningful

32 comparison over the immune-responsive signals. The dataset on COVID-19 patients came from medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 GEO accession GSE15805521 and the dataset on flu vaccine came from GEO accession

2 GSE14863322.

4 Cell-type annotation and cluster marker identification 5 After nonlinear dimensional reduction and projection of all cells into two-dimensional space by 6 tSNE, cells were clustered together according to the similar features. ‘FindClusters’ function 7 was used to define the unsupervised clusters. “FindAllMarkers” function in Seurat was used to 8 find markers for each cluster. Clusters were then preliminarily classified and annotated based 9 on expressions of canonical markers and signature genes of particular cell types. In total, we 10 identified 11 major cell types including CD4+ T and CD8+ T cells (CD3D, CD3E, CD3G, 11 CD40LG, CD4, CD8A, CD8B), γδ T cells (TRDV2, TRGV9), B cells (MS4A1, CD79A, CD79B), 12 plasma cells (JCHAIN, MZB1), NK cells (NCAM1, GNLY, NKG7), Monocyte and Dendritic 13 cells (CD14, FCGR3A, CST3, LYZ, FCER1A, CD1D, LILRA4) and Megakaryocytes (Mega) 14 (PPBP, PF4). We also used SuperCT33 and 31 reference cell types [https://github.com/weilin- 15 genomics/rSuperCT_models] to predict the cell types of the total 120k cells. SuperCT 16 predictions were used to for cross-validation. We assigned the final cell types based on the entire 17 cluster rather than single cells. The dominant prediction for the cluster will correct the 18 discrepancy in the same cluster. Additional details of prediction on top of SuperCT 31-type, 19 such as γδT cells, mDC, pDC were assigned by the enriched markers and detailed cell-type 20 boundaries were determined by the Seurat ‘FindClusters’ function with default parameters.

22 Identification of differentially expressed genes (DEGs) and function enrichment 23 The gene expression signals of a certain cell group (by cell type and by sample) were 24 represented by copy number per million UMIs (CPM). See details in the following equation.

∑ 푒 푗 푖,푗,푘 6 25 퐶푃푀푖,푘 = ∙ 10 ∑푖 ∑푗 푒푖,푗,푘

26 Let 푒 be number of unique molecular identifiers (UMIs) of i gene determined by unique 27 molecular barcode in a certain cell j in group k. 28 The CPMs of each gene were compared using student tests by pairwise or non-pairwise 29 groups. P value < 0.01 was used as a threshold to define the significant differential expression 30 genes (DEGs). 31 When we compared the differential genes between the post-vaccination time points and 32 the base line samples, we could use pairwise t-test by each individual. When we compare the medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 differential genes across B-R-P sample groups, we could only use independent t-test. 2 GO and KEGG pathway enrichment analyses on these DEGs identify the biological 3 keywords, that allowed us to summarize the keywords of classical immune responsive signal. 4

5 Transcriptional factor analysis 6 To identify the potential regulatory transcription factors that dictate the vaccine response, the 7 regulons in major cell types were assessed using SCENIC27 with scaled expression matrix as 8 input. The regulons expression profile cluster plots were visualized using the ComplexHeatmap 9 package of R. 10

11 Assessment of the overall signal score of the keywords of immunity 12 We used cellular modular scores to assess the immune state after infection or vaccination. We 13 selected differentially expressed genes enriched in MSigDB 34 to calculate the modular scores. 14 The involved genes for 12 immune-responsive signals were shown in supplemental table3. The 15 modular score of a certain keyword was calculated with ‘AddModuleScore’ function in Seurat 16 with default settings. The keywords signal score of different immune state were analyzed by 17 two-sided wilcox test and all differences with p < 0.05. 18

19 TCR and BCR V(D)J sequencing and analysis

20 TCR and BCR clonal types were determined using CellRanger V(D)J pipeline（10× Genomics,

21 Version 4.0.0）. TCR/BCR diversity was measured as Shannon’s entropy using ‘H’ function in

22 philentropy package:

30 H = − ∑ xp(x) ∗ log2[p(x)]

23 in which the p(x) represents the frequency of a given TCR/BCR clone among all T/B cells

24 with TCR/BCR identified. The TCR/BCR diversity among different times was analyzed by

25 two-sided Student’s t-test and all differences with p < 0.05. We loaded the clonal type

26 information from the CellRanger V(D)J output to the meta data of the seurat objects based on

27 the barcode mapping. The cells belonging to the top 20 TCR and BCR clones were labeled as

28 ‘top-clones’. The gene expression signal and the immune responsive signals were compared

29 between the top-clones and their counterparts using Seurat ‘VlnPlot’ function. 31 medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 Acknowledgements

2 We thank the volunteers from Pu Dong Cohort and Professor Ling Ni at Tsinghua University

3 for their contribution to the study. This work was supported by grants from the National Science

4 Fund for Distinguished Young Scholars (81625002), the National Natural Science Foundation

5 of China (81930007, 81470389, 81500221 and 81800307), the Shanghai Outstanding

6 Academic Leaders Program (18XD1402400), Innovative research team of high-level local

7 universities in Shanghai (SSMU-ZDCX20180200) and Shanghai Municipal Education

8 Commission Gaofeng Clinical Medicine Grant Support (20152209).

10 Author contributions

11 P.J and L.W conceived the need for the article. L.Z, L.Y.Y, H.L, G.X, L.Q, S.T, Z.C.J and Y.A.C

12 recruited volunteers. T.R.Y, C.Y.F, S.J.F, and L.G.Q collected the samples from donors. T.R.Y,

13 H.L.H, C.Y.F, Y.A.C and Z.M.C completed the experiments. L.W, T.R.Y, Z.J.M and L.R.H

14 finished the data analysis. L.W, T.R.Y and C.Y.F drafted the initial version. C.Y.F and T.R.Y

15 provided the production of the flowchart. P.J and L.W put forward many constructive comments

16 for the final version. All the authors read and approved the manuscript.

18 Competing interests

19 The authors declare no competing interests. 20 21 References

22 1. Polack FP, et al. Safety and Efficacy of the BNT162b2 mRNA Covid-19 Vaccine. The New England

23 journal of medicine 383, 2603-2615 (2020).

24 2. Baden LR, et al. Efficacy and Safety of the mRNA-1273 SARS-CoV-2 Vaccine. The New England

25 journal of medicine 384, 403-416 (2021).

26 3. Voysey M, et al. Safety and efficacy of the ChAdOx1 nCoV-19 vaccine (AZD1222) against SARS-

27 CoV-2: an interim analysis of four randomised controlled trials in Brazil, South Africa, and the UK.

28 Lancet (London, England) 397, 99-111 (2021).

29 4. Logunov DY, et al. Safety and efficacy of an rAd26 and rAd5 vector-based heterologous prime-

30 boost COVID-19 vaccine: an interim analysis of a randomised controlled phase 3 trial in Russia. medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 Lancet (London, England) 397, 671-681 (2021).

2 5. Xia S, et al. Safety and immunogenicity of an inactivated SARS-CoV-2 vaccine, BBIBP-CorV: a

3 randomised, double-blind, placebo-controlled, phase 1/2 trial. Lancet Infect Dis 21, 39-51 (2021).

4 6. Wu Z, et al. Safety, tolerability, and immunogenicity of an inactivated SARS-CoV-2 vaccine

5 (CoronaVac) in healthy adults aged 60 years and older: a randomised, double-blind, placebo-

6 controlled, phase 1/2 clinical trial. Lancet Infect Dis, (2021).

7 7. Zhang Y, et al. Safety, tolerability, and immunogenicity of an inactivated SARS-CoV-2 vaccine in

8 healthy adults aged 18-59 years: a randomised, double-blind, placebo-controlled, phase 1/2 clinical

9 trial. Lancet Infect Dis 21, 181-192 (2021).

10 8. Keech C, et al. Phase 1-2 Trial of a SARS-CoV-2 Recombinant Spike Protein Nanoparticle Vaccine.

11 The New England journal of medicine 383, 2320-2332 (2020).

12 9. Sahin U, et al. COVID-19 vaccine BNT162b1 elicits human antibody and TH1 T cell responses.

13 Nature 586, 594-599 (2020).

14 10. Wise J. Covid-19: European countries suspend use of Oxford-AstraZeneca vaccine after reports of

15 blood clots. BMJ 372, n699 (2021).

16 11. Schultz NH, et al. Thrombosis and Thrombocytopenia after ChAdOx1 nCoV-19 Vaccination. The

17 New England journal of medicine, (2021).

18 12. Greinacher A, Thiele T, Warkentin TE, Weisser K, Kyrle PA, Eichinger S. Thrombotic

19 Thrombocytopenia after ChAdOx1 nCov-19 Vaccination. The New England journal of medicine,

20 (2021).

21 13. Blumenthal KG, et al. Acute Allergic Reactions to mRNA COVID-19 Vaccines. Jama 325, 1562-

22 1565 (2021).

23 14. Wang H, et al. Development of an Inactivated Vaccine Candidate, BBIBP-CorV, with Potent

24 Protection against SARS-CoV-2. Cell 182, 713-721 e719 (2020).

25 15. Stuart T, Satija R. Integrative single-cell analysis. Nature reviews Genetics 20, 257-272 (2019).

26 16. Yifan C, Fan Y, Jun P. Visualization of cardiovascular development, physiology and disease at the

27 single-cell level: Opportunities and future challenges. Journal of molecular and cellular cardiology

28 142, 80-92 (2020).

29 17. Wilk AJ, et al. A single-cell atlas of the peripheral immune response in patients with severe COVID-

30 19. Nature medicine 26, 1070-1076 (2020). medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 18. Liao M, et al. Single-cell landscape of bronchoalveolar immune cells in patients with COVID-19.

2 Nature medicine 26, 842-844 (2020).

3 19. Zhu L, et al. Single-Cell Sequencing of Peripheral Mononuclear Cells Reveals Distinct Immune

4 Response Landscapes of COVID-19 and Influenza Patients. Immunity 53, 685-696 e683 (2020).

5 20. Shaath H, Vishnubalaji R, Elkord E, Alajez NM. Single-Cell Transcriptome Analysis Highlights a

6 Role for Neutrophils and Inflammatory Macrophages in the Pathogenesis of Severe COVID-19.

7 Cells 9, (2020).

8 21. Ren X, et al. COVID-19 immune features revealed by a large-scale single-cell transcriptome atlas.

9 Cell 184, 1895-1913 e1819 (2021).

10 22. Turner JS, et al. Human germinal centres engage memory and naive B cells after influenza

11 vaccination. Nature 586, 127-132 (2020).

12 23. Sørvoll IH, et al. An observational study to identify the prevalence of thrombocytopenia and anti-

13 PF4/polyanion antibodies in Norwegian health care workers after COVID-19 vaccination. Journal

14 of thrombosis and haemostasis : JTH, (2021).

15 24. Goshua G, et al. Endotheliopathy in COVID-19-associated coagulopathy: evidence from a single-

16 centre, cross-sectional study. The Lancet Haematology 7, e575-e582 (2020).

17 25. Hottz ED, et al. Platelet activation and platelet-monocyte aggregate formation trigger tissue factor

18 expression in patients with severe COVID-19. Blood 136, 1330-1341 (2020).

19 26. Manne BK, et al. Platelet gene expression and function in patients with COVID-19. Blood 136,

20 1317-1329 (2020).

21 27. Aibar S, et al. SCENIC: single-cell regulatory network inference and clustering. Nature methods 14,

22 1083-1086 (2017).

23 28. Zhang JY, et al. Single-cell landscape of immunological responses in patients with COVID-19.

24 Nature immunology 21, 1107-1118 (2020).

25 29. Moser BA, et al. Increased vaccine tolerability and protection via NF-κB modulation. Sci Adv 6,

26 (2020).

27 30. Moser BA, et al. Small Molecule NF-κB Inhibitors as Immune Potentiators for Enhancement of

28 Vaccine Adjuvants. Front Immunol 11, 511513 (2020).

29 31. World Medical A. World Medical Association Declaration of Helsinki: ethical principles for medical

30 research involving human subjects. Jama 310, 2191-2194 (2013). medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 32. Corkum CP, Ings DP, Burgess C, Karwowska S, Kroll W, Michalak TI. Immune cell subsets and

2 their gene expression profiles from human PBMC isolated by Vacutainer Cell Preparation Tube (CPT)

3 and standard density gradient. BMC Immunol 16, 48 (2015).

4 33. Xie P, et al. SuperCT: a supervised-learning framework for enhanced characterization of single-cell

5 transcriptomic profiles. Nucleic acids research 47, e48 (2019).

6 34. Liberzon A, Birger C, Thorvaldsdottir H, Ghandi M, Mesirov JP, Tamayo P. The Molecular

7 Signatures Database (MSigDB) hallmark gene set collection. Cell Syst 1, 417-425 (2015). 8 9 10 11 12 13 14 15 16 medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 Figure Legends

2 3 Figure 1. Overview of vaccination efficacy, scRNA-seq data quality, cell compositions. medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 A. Workflow of vaccination, sampling and scRNA-seq; B. tSNE charts of ~120 k single cells 2 by color coded by putative cell types and time points; C. Markers for the putative cell types 3 using violin plots in scRNA-seq data; D. Box charts showing variabilities of the contents of 4 each cell type by time point. All differences with p < 0.05 and analyzed by two-sided Student’s 5 t-test; E. Serological responses of 6 healthy recipients to recombinant S-RBD at 4 time-points. 6 Dilution of 1:55 for IgG. 7 medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 2 3 4 Figure 2. Classical immune responsive signals of major immune cell types in vaccination. medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 A. Heatmap of DEGs detected in B cells; B. Heatmap of regulons in B cells; C. Modular scores 2 of type I interferon production related signals; D. Tfh signal of CD4+ T cells; E. B cell activation 3 signal of B cells; F. Relative UMI percentages of different classes of antibody genes; G. TCR 4 signal of CD4+ T cells; H. Cytotoxicity signal of CD8+ T cells; I. Exhaustion signal of CD8+ T 5 cells; J. TGF-β signal of CD8+ T cells; K. Type II MHC signal of Monocytes; L. Type I MHC 6 signal of Monocytes. All differences with p < 0.05 and analyzed by two-sided wilcox test. 7

8 Fig. 3. Characteristic transcriptions of top TCR and BCR clones. 9 A. Diversity of BCR repertoire at different vaccination time points; B and C. variable immune- 10 responsive signals in top BCR clones; D and G. Diversity of TCR repertoires of CD4+ T and medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 CD8+T cells at different vaccination time points; E and F. Variable immune-responsive signals 2 in top CD4+ T cells TCR clones; H and I. Variable immune-responsive signals in top CD8+ T 3 cells TCR clones. All differences with p < 0.05 and analyzed by two-sided Student’s t-test.

4 5 Figure 4. Non-trivial roles of megakaryocytes in vaccine immunity. 6 A. Heatmap of DEGs detected in megakaryocytes; B. Platelet aggregation signal of 7 megakaryocytes; C. IL-1β production signal of megakaryocytes; D. Neutrophil activation 8 signal of megakaryocytes; E-G. Platelet aggregation signal, IL-1β production signal and 9 Neutrophil activation signal of megakaryocytes at 28-day of flu vaccination. All differences 10 with p < 0.05 and analyzed by two-sided wilcox test. 11 medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 2 Figure 5. The immune-responsive signals in COVID-19 patients and Flu vaccination. 3 A. Turning point of immune-responsive signals at convalescence; B. Variability of type I and 4 type II MHC signals at COVID-19 patients. C-E. Variable immune-responsive signals at 28- medRxiv preprint doi: https://doi.org/10.1101/2021.05.06.21256781; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 day of flu vaccination. 2 3