An HLA Association With COVID-19 Vaccine Reactogenicity Correlates With Fewer SARS-CoV-2 Infections and Monocyte Activation

An HLA Association With COVID-19 Vaccine Reactogenicity Correlates With Fewer SARS-CoV-2 Infections and Monocyte Activation | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Biological Sciences - Article An HLA Association With COVID-19 Vaccine Reactogenicity Correlates With Fewer SARS-CoV-2 Infections and Monocyte Activation Jill Hollenbach, Anshika Srivastava, Demetra Chatzileontiadou, and 32 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8282930/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Vaccination against SARS-CoV-2 has been the most effective tool in mitigating the COVID-19 pandemic. However, some individuals experience side effects that cause distress and interfere with daily activities, which can limit vaccine uptake, with important public health implications. Here, we considered the impact of HLA variation on the propensity for mild side effects with COVID-19 vaccination. We examined variation in HLA-A, -B, -C, -DRB1, and -DQB1 for association with self-reported side effects in a large cohort of U.S. Euro-ancestry vaccinated individuals (N = 50,535) and confirmed results in an independent replication cohort (N = 4,575). We found that HLA-A 03:01 was significantly associated with systemic side effects (OR = 1.36, CI = 1.31-1.41, p = 6.79×10-57) and fewer breakthrough infections, and that this phenomenon is specific to the COVID-19 vaccine. Surprisingly, we observed limited activation of CD8+ T cells in HLA-A 03:01+ samples to the Spike-derived peptides, excluding them as a likely source of the reported vaccine side effects. Rather, examination of immune cell subsets, prior and after vaccination, points to a central role for monocytes in the production of IL-6 and IL-8, which significantly correlates with the reported severity of side effects in HLA-A 03:01+ donors. Meanwhile, the large, mostly naïve, and low-affinity population of Spike-specific CD8+ T cells likely contribute to an inflammatory milieu in HLA-A 03:01 carriers through weak binding to antigen presenting cells. This work sheds light on the mechanisms underlying HLA-mediated COVID-19 vaccine reactogenicity and associated reduction in infections, providing important new insights that may support efforts to optimize vaccine efficacy and promote broader public involvement in vaccination programs. Biological sciences/Genetics Biological sciences/Immunology/Vaccines Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction COVID-19 vaccines are an important public health tool in preventing severe illness, hospitalization, and mortality due to infection with the virus 1–4 . In early studies examining the effectiveness of these vaccines, mRNA vaccines BNT162b2 (referred to hereafter by the brand name “Pfizer") and mRNA-1273 (referred to hereafter by the brand name “Moderna”) were shown to be 95% and 94.1% effective against symptomatic infection, respectively 5 . Adenovirus-based vaccines, such as ChAdOx1-S (brand name “Johnson & Johnson”, hereafter “J&J) and Ad26.COV2.S (brand name “AstraZeneca”), in contrast, showed 70% and 66% efficacy, respectively 6,7 . Thus, although mostly efficacious in preventing serious illness and hospitalization, ‘breakthrough’ infections (BTI) occur with all current COVID-19 vaccines. These infections may be attributed to primary vaccine failure, secondary vaccine failure, individuals’ age, immune evasion by novel viral variant, or waning vaccine efficacy over time 8–15 . Prior work has indicated that the immunogenicity of the COVID-19 vaccines can vary by individual and is correlated with their efficacy. For example, studies have reported different immunological responses to vaccines linked to age, sex, body mass index, nutritional status, and the composition of the gut microbiome 16,17 . Suggesting an immunogenetic feature of vaccine response, several studies have linked variation in the human leukocyte antigen ( HLA ) region with antibody levels and T-cell response after vaccination 18–21 . HLA is the most polymorphic region (6p21) of the human genome with thousands of known alleles. Variation in HLA is long recognized to play a role in viral illness 22 . Demonstrating the importance of HLA antigen presentation in the immune response to SARS-CoV-2, we have previously shown that HLA variation is associated with an asymptomatic disease course 23 and provided a functional and structural basis to explain the association. Likewise, numerous other HLA associations with the COVID-19 disease course have been identified 24 . While overwhelmingly safe, vaccination-induced immune activation can lead to side effects. Most adverse reactions are mild, including fever, muscle aches, headaches, and fatigue, as well as local reactions such as pain, redness, and swelling at the injection site 25 . These reactions typically appear within a few hours after vaccination and are short-lived, usually resolving within one to two days 26 . While bothersome, there is some evidence that side effects may be associated with improved vaccine efficacy. A study involving participants receiving the human papillomavirus vaccine, for example, found that the occurrence of inflammation-related adverse reactions is associated with concentrations of antibodies, suggesting that individuals who have vaccine-induced side effects have a more robust immune response 27,28 . Importantly, systemic side effects from COVID-19 vaccination, such as fever and fatigue, have also been associated with enhanced humoral and cellular immune responses 29 . Intriguingly, previous work has shown a specific HLA class I allele, HLA-A*03:01 to be associated with increased side effects after COVID-19 vaccination as well as increased antibody response 19,30,31 . Additionally, enhancement of T cell memory from COVID-19 mRNA booster doses was shown to be particularly pronounced in HLA-A*03:01 + COVID-19 recovered patients 20 . Nevertheless, despite strong evidence for the role of HLA in response to COVID-19 vaccination, there remains substantial knowledge gaps regarding the complex interplay between HLA genetic variation, HLA structural variation and antigen presentation, vaccine side effects, cellular and humoral immunity, and vaccine effectiveness. In the present study, we close these gaps by considering HLA variation and vaccine reactogenicity in a large cohort of over 50,000 vaccinated subjects and provide a global framework for our findings by examining differential gene expression, antibody response, T cell reactivity, and innate immunity. We demonstrate that HLA-A*03:01 -associated reactogenicity is associated with fewer total infections and is not driven by T cell activation, despite a large pool of mostly naïve Spike-specific T cells, but rather by monocyte-derived cytokine production, revealing an innate immune mechanism underlying HLA-linked COVID-19 vaccine side effects. Results Participant recruitment and baseline demographics Between January 2023 and January 2024, we collected responses to survey questions regarding respondents’ general health history, including experiences with COVID-19 and associated vaccinations, from potential bone marrow donors registered with the NMDP (formerly National Marrow Donor Program/Be The Match) and for whom high-resolution HLA genotyping data were available in the NMDP database. The specific survey questions related to COVID-19 and vaccinations, including side effects queried, are given in Supplementary Table 1 . Because this was a U.S.-based cohort, respondents received only vaccines approved for use in the country (Pfizer, Moderna, J&J). Among the 80,007 total respondents, 50,535 self-identified as White/European ancestry ( Supplementary Table 2a ) and reported having completed at a minimum the initial series of vaccination for SARS-CoV-2 (one dose for J&J, two for Pfizer/Moderna mRNA); these subjects constituted our discovery cohort. An additional 7403 respondents reported completion of the initial series self-identified with other ancestries ( Supplementary Table 2b ). For our replication cohort , we collected responsesbetween July 2020 and April 2022 via a mobile phone app with follow-up daily questions specific to vaccine side effects, as described in Augusto et al., 2023 23 . Among 10,595 respondents who reported completion of the initial vaccination series, 4,575 self-identified as White/European ancestry ( Supplementary Table 3 ). While these respondents are also NMDP donors with available high resolution HLA data, there was no overlap with individuals in the discovery cohort. Systemic side effects to COVID-19 vaccines co-occur and are associated with HLA-A*03:01 To understand the distribution of side effects to COVID-19 vaccines in our discovery cohort, we first calculated the covariation matrix for all reported side effects ( Supplementary Figure 1 ). Here, we only considered side effects reported after completion of the initial vaccination series. We found that systemic side effects (SSE) like fever or chills, muscle or body fatigue, or headaches, rather than localized side effects like runny nose ( Supplementary Table 4 ), showed substantial co-occurrence, with a median number of two SSE reported per individual. To evaluate whether HLA variation plays a role in increasing side effects in vaccination for SARS-CoV-2 and to capture cases with the greatest burden of symptoms, we first considered individuals who reported greater than the median number of SSE. Using a dominant model, we used multivariate logistic regression to test for association with the occurrence of three or four reported SSE for each HLA allele at each classical class I ( HLA-A, -B, -C ) and two HLA class II ( HLA-DRB1 and -DQB1 ) loci observed at a frequency >3% in our cohort, adjusting for sex and age ( Figure 1, Supplementary Table 5 ). This revealed as the top candidate a strong and significant association of HLA-A*03:01 with SSE (OR = 1.36, CI= 1.31 – 1.41, p = 6.79×10 -57 ). We did not observe any substantial dose effect for HLA-A*03:01 (homozygous, OR = 1.47, CI = 1.29 – 1.68, p = 1.65×10 -8 ; heterozygous, OR = 1.40, CI = 1.34 – 1.46, p = 3.15 ×10 -53 ), confirming the dominant model. Because HLA-A*03:01 belongs to the HLA-A3 supertype group 32 that also includes HLA-A*11:01, -A*31:01, -A*33:01, and - A*68:01 , and is characterized by shared peptide binding 33 , we hypothesized that this supertype might show shared associations with SSE across alleles. However, we found that among the HLA-A3 supertype alleles, only HLA-A*03:01 was significant for the association with increased SSE ( Supplementary Table 5 ). This association of increased SSE with HLA-A*03:01 clearly replicated in our additional cohort of 4,575 vaccinees with European ancestry, with a remarkably consistent effect size (OR = 1.46, CI = 1.21 – 1.77, p = 7.77 x 10 -5 ) relative to that in our discovery cohort ( Supplementary Table 6) . In addition, we found that HLA-A*03:01 was significantly associated with increased SSE reports in our self-identified Hispanic cohort (N = 4,287), again with extremely consistent effect size (OR = 1.45, CI = 1.25 – 1.69, p = 7.31× 10 -7 , Supplementary Table 7 ). While a similar trend was clearly observed, this association did not reach statistical significance in cohorts of other ancestries, where our sample sizes were much smaller ( Supplementary Tables 8 and 9 ). To better understand whether any specific reported side effect was driving this HLA association, we considered the association with HLA-A*03:01 for each side effect separately ( Supplementary Table 10 ). As expected, we observed a strong and highly significant negative association of HLA-A*03:01 with reporting “no vaccine side effects” (OR = 0.73, CI = 0.70 – 0.77, p = 5.82×10 -43 ), demonstrating that individuals with this allele are less likely to report having experienced no side effects after vaccination. Analysis of reports of specific side effects revealed the strongest association of HLA-A*03:01 with “fever or chills” (OR = 1.43, CI = 1.38 – 1.49, p = 1.37×10 -81 ), followed by “muscle or body aches” (OR = 1.305, CI = 1.25 – 1.35, p = 2.16 ×10 -45 ), “fatigue” (OR = 1.25, CI = 1.20 – 1.30, p = 5.57×10 -34 ), and “headaches” (OR = 1.23, CI = 1.19 – 1.29, p = 1.00×10 -24 ). We also found remarkably consistent results for our cohort of individuals who self-identify as Hispanic ( Supplementary Table 10 , Supplementary Figure 2 ).Only “fatigue” was found to have a significant association with HLA-A*03:01 in our African American cohort (OR = 1.49, CI = 1.07 – 2.09, p = 0.018); however, as noted previously, smaller sample sizes limited our power to detect associations in some cohorts ( Supplementary Table 8 and 9 ). Finally, we did not find a substantive difference in effect size when considering individuals who had reported infection prior to vaccination or breakthrough infection (BTI)( Supplementary Table 11 ). In summary, we find a highly significant, consistent association of HLA-A*03:01 with reported side effects post-COVID-19 vaccination. This association replicated across multiple independent cohorts and ancestries, supporting a role for HLA variation in driving vaccination side effect. Effect size of HLA-A*03:01 varies with vaccine brand While the median number of SSE reported across the entire discovery cohort was two, the proportion of individuals who experienced three or four SSE was higher in Moderna than in Pfizer or Johnson & Johnson vaccinated individuals ( Supplementary Figure 3 ). Because the frequency of SSE varied by vaccine manufacturer, we stratified our cohorts according to whether they received the Pfizer, Moderna, or J&J vaccine. To reduce confounding, we only considered individuals who received both doses of the initial series (in the case of mRNA vaccines) from the same manufacturer. We found that Pfizer-vaccinated individuals showed the largest effect size for all associations of HLA-A*03:01 with side effects relative to other vaccine brands, particularly fever or chills (OR = 1.71, CI = 1.61 – 1.79, p = 4.51×10 -88 ). Likewise, the chance of not experiencing any side effects is significantly lower in Pfizer (OR = 0.676, CI = 0.64 – 0.71, p = 2.83×10 -40 ) and Moderna recipients (OR = 0.782, CI = 0.72 – 0.85, p =1.59×10 -9 ) ( Supplementary Table 12 ). We did not observe significant associations of HLA-A*03:01 with side effects among J&J vaccine recipients except for fatigue (OR = 1.259, CI = 1.09 – 1.45, p = 1.40×10 -4 ), possibly owing to the small sample size for this brand (N = 3,312). We observed similar results in our replication cohort, where the association of HLA-A*03:01 showed similar patterns of effect size between brands, although many individual side effects did not reach statistical significance among Moderna recipients, likely due to more limited power in this cohort ( Supplementary Table 13 ). Thus, while we find that HLA-A*03:01 is associated with side effects across COVID-19 vaccine brands, this effect is much more pronounced and consistent among Pfizer recipients. Additional HLA class I alleles are associated with systemic side effects in COVID-19 vaccination In addition to HLA-A*03:01 , we observed numerous other HLA alleles that were significantly associated with reports of systemic side effects in our discovery cohort ( Figure 1 , Supplementary Table 5 ). Of these, HLA-A*29:02 (OR = 1.28, CI = 1.19 – 1.36, p = 2.10×10 -11 ), HLA-B*08:01 (OR = 0.76, CI = 0.73 – 0.80, p = 5.65×10 -32 ), HLA-C*07:01 (OR = 0.82, CI = 0.79 – 0.85, p = 1.10×10 -24 ), and HLA - DRB1*03:01 (OR = 0.85, CI = 0.82-0.90, p = 1.23×10 -12 ) replicated ( Supplementary Table 6 ). HLA-B*08:01, HLA-C*07:01, and HLA-DRB1*03:01 are components of the well-documented “ancestral 8.1 haplotype” (AH8.1) 34 , which is also evident from the high linkage disequilibrium (LD) values between these alleles ( Supplementary Table 14 ). Owing to the likelihood that these three alleles reflected a single primary association through LD, we performed conditional analyses to identify the primary associated allele in this haplotype. We found that HLA-B*08:01 had the strongest effect size (OR = 0.77, CI = 0.71 – 0.84, p = 2.71×10 -9 ) after controlling for HLA-C*07:01, and HLA-DRB1*03:01 , suggesting that this allele is responsible for the protective effect against SSE. Likely owing to the small sample size, we did not detect these associations in our cohorts with other ancestries ( Supplementary Tables 7 – 9 ). In summary, while HLA-A*03:01 demonstrated the strongest and most significant effect with respect to vaccine SSE, we find evidence for additional HLA class I involvement for both risk ( HLA-A*29:02 ) and protection ( HLA-B*08:01 ). HLA-A*03:01 carriage is associated with higher antibody levels after COVID-19 vaccination We next sought to confirm prior reports of association of HLA-A*03:01 with higher levels of antibody upon vaccination for SARS-CoV-2 19,35,36 in an independent cohort of 156 healthcare workers receiving two doses of the Pfizer vaccine. The cohort was stratified in three groups according to the levels of anti-Spike IgG antibodies after the second dose of the vaccine as follows, Group I: 1,000 – 4,000 AU/mL (N = 50), Group II: 4,001 – 20,000 AU/mL (N = 53), Group III: ≥ 20,000 AU/mL (N = 53). To investigate the association between HLA-A*03:01 and anti-Spike antibody levels, we performed logistic regression comparing Group III against Groups I and II. Our results confirmed that HLA-A*03:01 is associated with higher antibody levels post vaccination (OR = 3.25, CI = 1.31 – 8.33, p = 0.0117). We also observed that antibody positivity increased with HLA-A*03:01 allele dosage (0, 1, or 2 copies; Supplementary Figure 4 ). Thus, we confirm that in addition to its role in mediating vaccine SSE, individuals positive for HLA-A*03:01 display a higher anti-Spike antibody response post-vaccine, providing further evidence that vaccine side effects are reflective of a more robust immune response to vaccination and that this has an immunogenetic underpinning. HLA-A*03:01 carriers report fewer breakthrough infections, fewer total infections, and milder disease course Owing to the highly significant association of HLA-A*03:01 with vaccine side effects and increased antibody levels after vaccination, we next asked whether this allele might be associated with reduced breakthrough infection (BTI). Of the total vaccinated 57,938 individuals across ancestries, we excluded 10,227 individuals who did not have information for their 2 nd dose timing. Among the remaining participants who reported specific dates for both vaccination and infection, we found HLA-A*03:01 to be associated with a decreased risk of BTI, albeit with modest effect size (OR = 0.916, CI = 0.88 – 0.95, p = 1.61×10 -5, ); this association was also significant with stronger effect size in our cohort with Hispanic ancestry ( OR = 0.776, CI = 0.66 – 0.91, p = 1.44×10 -4 ), and while not reaching statistical significance, we observed similar effect sizes in other ancestries( Supplementary Table 15 ). We did not observe any other significant HLA associations with BTI in our cohort, including previously reported associations of HLA-DQB1*06 18,37 . The observed association of HLA-A*03:01 with reduction in BTI is also reflected in an overall decrease in total infections reported. We dichotomized participants according to those reporting having never been infected or having only a single infection, (about 86% of our discovery cohort, N= 43,527), and those reporting repeated reinfections (i.e. two or more infections, about 14% of our discovery cohort, N = 7008) between 2019 and 2023. We found that vaccinated individuals positive for HLA-A*03:01 were less likely to have experienced repeated SARS-CoV-2 infections (OR = 0.92, CI = 0.87 – 0.96, p = 0.00177). Finally, we asked whether among those who did report infection, HLA-A*03:01 is associated with milder disease course. We found that individuals positive for HLA-A*03:01 were less likely to report higher than the median number (three) of symptoms (considering all reported symptoms; OR = 0.90, CI = 0.88 – 0.94, p = 2.78 × 10 -7 ) suggesting a milder COVID-19 disease course. To understand whether the HLA-A*03:01 association with reduced BTI was secondary to its association with side effects, we considered the occurrence of vaccination side effects and the likelihood of BTI irrespective of HLA genotype. We found that each systemic side effect was independently inversely associated with the likelihood of a BTI ( Supplementary Table 16 ). Considering the combined measure of three or four SSE and BTI (and adjusting for sex, age, and vaccine brand), we found a highly significant protective effect with an effect size greater than that observed for HLA-A*03:01 (OR = 0.83, CI = 0.79 – 0.87, p = 6.03x10 -14 ). When we include HLA-A*03:01 as a covariate in our model, we observed a nearly identical association of SSE with BTI (OR = 0.82, CI = 0.79 – 0.87, p = 9.40×10 -14 ), while the HLA-A*03:01 (OR = 0.93, CI = 0.89 – 0.98, p = 5.81x10 -3 ) association was only borderline significant, suggesting an association independent from HLA . Finally, we stratified our cohort according to the carriage of HLA-A*03:01 and further observed a highly significant and consistent inverse association between the combined measure for three or four SSE and the risk of BTI in individuals without the HLA-A*03:01 allele (OR = 0.81 CI = 0.78 – 0.85, p = 4.32×10 -19 ). Thus, we conclude that the negative association of HLA-A*03:01 with BTI stems from its role in increasing vaccine reactogenicity, along with a global negative association of vaccine side effects with BTI. HLA-A*03:01 associated vaccine reactogenicity is specific to COVID-19 vaccines To fully contextualize our findings, we sought to compare responses to COVID-19 vaccines with those of other widely administered vaccines, such as the currently approved influenza vaccine. We administered a follow-up questionnaire to the same individuals included in our discovery cohort ( Supplementary Table 17 ). This survey specifically assessed the occurrence of influenza vaccine-associated side effects and the frequency with which participants received influenza vaccinations. As of April 2025, we received responses from a total of 13,499 participants from our discovery cohort, among whom 11,916 individuals reported having received the influenza vaccine at least once ( Supplementary Table 18 ). Of those receiving the vaccine, 3,417 individuals reported experiencing side effects. As for the COVID-19 vaccine, we performed multivariate logistic regression to assess associations between each HLA allele and the presence of at least one side effect ( Supplementary Table 19 ). None of the tested HLA alleles reached the Bonferroni-corrected threshold for statistical significance (p = 7.14 ×10 -4 ), including HLA-A*03:01 . Thus, the HLA association with vaccine reactogenicity appears to be specific to the COVID-19 vaccines. Despite the lack of HLA -specific associations, we observed an epidemiologically meaningful correlation: individuals experiencing side effects following influenza vaccination were more than three times (OR = 3.094, CI = 2.73 – 3.50, p = 2 ×10 -16 ) as likely to report side effects from the COVID-19 vaccine, after controlling with HLA-A*03:01. Thus, it appears that there are factors beyond HLA predisposing some individuals to experiencing vaccine reactogenicity. High frequency of low avidity Spike-specific T cell population is observed ex vivo The strong association between HLA-A*03:01 carriage and the presence of side effects after COVID-19 vaccination, coupled with more robust immunity to the virus (higher Spike-specific antibody and less breakthrough infection occurrence), led us to hypothesize that HLA-A*03:01 + individuals could mount a strong and perhaps disproportionate CD8 + T cell response, as previously suggested by others 30 . To assess this hypothesis, blood from HLA-A*03:01 + and HLA-A*03:01 - individuals was collected, most without SARS-CoV-2 infection, prior (V0) and post vaccination with either 1 dose (V1) or 2 doses (V2) of the Pfizer or AstraZeneca vaccine ( Supplementary Table 20 ). The peripheral blood mononuclear cells (PBMCs) and plasma were isolated and vaccination side-effects self-reported (converted to a side-effect severity score) ( Supplementary Table 21 ). To examine the population of Spike-specific CD8 + T cells present before and after vaccination, we considered the reported dominant HLA-A*03:01-restricted Spike-derived epitope 378 KCYGVSPTK 386 , hereafter named KCY 20,38 . It has been shown that KCY-specific memory T cells increase upon viral exposure or vaccination 20 . Therefore, we performed ex vivo tetramer-associated magnetic enrichment (TAME) to analyze the phenotype of the KCY + CD8 + T cells in samples collected before and after vaccination ( Figure 2A ). Surprisingly, even prior to vaccination (V0) a high frequency of KCY + T cells was present in all samples ( Figure 2A ). The frequency of KCY + T cells increased after vaccination ( Figure 2B-D ). Interestingly, after vaccination, a distinct population of KCY⁺ T cells with high mean fluorescence intensity (MFI) emerged ( Figure 2A , D ). The high MFI KCY⁺ T cell population expanded significantly after vaccination compared to baseline (V0 mean: 1.49 ± 0.89; V1 mean: 7.49 ± 2.01; V2 mean: 5.53 ± 0.65) ( Figure 2D ). In comparison, TAME with the HLA-A*02:01- restricted S 269-277 peptide 39–41 did not show tetramer + T cells before vaccination, and only a high MFI population of tetramer + T cells was observed after vaccination( Supplementary Figure 6A ). We also confirmed that the large population of KCY tetramer + T cells observed ex vivo was specific to the KCY peptide and not binding to an HLA-A*03:01-restricted influenza-derived peptide that we characterized previously, called Flu-NP 265 42 ( Figure 2E ). Given the large frequency of KCY + T cells ex vivo , we wondered if this population would be expanded after activation in vitro . To test this, we generated T cell lines using PBMCs cultured with either the KCY peptide or an HLA-A*03:01-restricted influenza-derived peptide, called Flu-NP 265 that we previously characterized 42 , to be tested in the same samples as the KCY peptide. The KCY T cell lines showed specificity for the KCY peptide only, and the Flu-NP 265 T cell lines were specific only to the Flu-NP 265 peptide ( Supplementary Figure 5B ). The frequency of KCY + CD8 + T cells in vitro was low (average of 0.71 ± 0.39 %, n = 3) ( Supplementary Figure 5B ), while it was 8-times higher for the Flu-NP 265 + T cells (average of 5.71 ± 1.67 %, n = 3) ( Supplementary Figure 5B ). This is in contrast with the high frequency observed ex vivo ( Figure 2A ); however, the ex vivo data were obtained after TAME. Therefore, we compared tetramer staining in ex vivo and in vitro samples, without tetramer magnetic enrichment. We observed that there was a 16- and 44-fold lower frequency of KCY + T cells in vitro compared to ex vivo in vacSG82-V2 and vacSG88-V2 samples, respectively ( Supplementary Figure 5C ). In addition, while the majority of ex vivo KCY + T cells had low MFI, the in vitro cells were largely high MFI ( Supplementary Figure 5C ). Overall, even prior to vaccination, an unusually large number of low avidity KCY-specific T cell populations was present in HLA-A*03:01 + samples ex vivo that were not expanded upon KCY presentation. In addition, vaccination led to an increase in high avidity KCY + T cell population, although these cells remained present at low frequencies relative to the low avidity population. KCY + T cells mostly exhibit a naïve phenotype We next assessed the phenotype of KCY + T cells ex vivo to determine if there was a difference between the high and the low avidity population ( Figure 2F-G , Supplementary Figure 6A ). The majority of the low MFI KCY + T cell population exhibited a naïve phenotype (T N : CCR7 + /CD45RA + ) independent of the vaccine status of the donors (average of 69.95, 56.35 and 63.9 % for V0, V1 and V2, respectively), and some stem cell memory T cells (T SCM : CCR7 + /CD45RA + /CD95 + ) present especially in one donor (V1) ( Figure 2F , Supplementary Figure 6A ). The low MFI effector memory T cell (T EM : CCR7 - /CD45RA - ) frequency was ~16%, and the terminally differentiated effector T cell (T EMRA : CCR7 - /CD45RA + ) frequency was around ~10% independent of vaccination status. The comparison of high and low MFI tetramer + CD8 + T cell phenotype showed that while the proportion of low MFI naïve T cells remains the same before and after vaccination ( Figure 2F, Supplementary Figure 6A ), the high MFI naïve T cell proportion decreased after vaccination while high MFI effector and central memory T cell proportion increased ( Figure 2G, Supplementary Figure 6A ). Strikingly, the phenotype of the KCY + T cells observed in HLA-A*03:01 + samples was different from the one observed for the S 269 peptide in HLA-A*02:01 + samples ( Supplementary Figure 6B ). We did not observe any S 269 + T cells in V0 samples, and S 269 + T cells were only observed in 1 out of 3 donors after V1 with a T CM and naïve phenotype ( Supplementary Figure 6B-C ). In V2 samples T EM, but no T EMRA cells, were observed. This is in contrast with the presence of both KCY + T EM and T EMRA cells, even prior to vaccination ( Figure 2F-G , Supplementary Figure 6C ). In two samples collected > 400 days after the 3 rd vaccine dose (vacSG64-V5 and vacSG86-V5), we could observe 36% and 8.85% of KCY + T EMRA cells, respectively, and a large population of naïve cells (32% and 79.7%, respectively) ( Supplementary Figure 6D ). Overall, despite the naïve phenotype of the large proportion of low avidity KCY-specific T cells, the cells are peptide specific. In addition, the high avidity effector memory cells are present before vaccination, increased after vaccination and persist over time. Nevertheless, the vast majority of KCY-specific T cells remain low avidity, naïve, and in high numbers. Spike-specific T cell activation in HLA-A*03:01 + donors is relatively weak To understand whether the T cell response was likely responsible for the observed vaccine reactogenicity in HLA-A*03:01 + donors, T cell lines were generated with Spike-derived peptide pools covering the whole length of the Spike protein and restimulated with either the peptide pools or the KCY peptide. The response before vaccination (V0) against the Spike-derived peptide pools was low for both CD8 + and CD4 + T cells ( Figure 2H , Supplementary Figures 7 – 10) . A trend upward of IFNγ producing CD8 + T cells was observed after vaccination (V1; mean ± SD; 0.09 ± 0.08 %, and V2; 0.39 ± 0.48 %) ( Figure 2H ), but not for CD4 + T cells (V1: 0.02 ± 0.04 %, and V2: 0.02 ± 0.07 %) ( Supplementary Figures 10A ). We also assessed the KCY peptide specific response. We observed limited IFNγ production in 50% of the samples before vaccination (n=4/8, 0.05 ± 0.06 %), and a larger IFNγ production in 85% (n=6/7, average of 0.16 ± 0.19 %) and 57% (n=4/7, average of 0.15 ± 0.19 %) after the first and second vaccine dose, respectively ( Figure 2I , Supplementary Figures 7-8 ). Despite the increase of IFNγproducing KCY-specific CD8 + T cells after vaccination, the response was overall weak compared with other well characterized Spike-derived epitopes such as HLA-A*02:01-restricted S 269-277 (average of 5.85% IFNγ + CD8 + T cells 39 or HLA-B*15:01-restricted S 919-927 (average of 0.36% IFNγ + CD8 + T cells in pre-pandemic samples 23 ). In addition, we previously characterized an HLA-A*03:01-restricted influenza-derived peptide, called Flu-NP 265 42 that was also tested here in the same samples side by side with the KCY peptide. The response to the Flu-NP 265 peptide was substantially stronger than that to the KCY peptide, with IFNγ production being ~13-fold higher (Flu-NP 265 : 3.32 ± 2.36 %; KCY: 0.27 ± 0.19 %) ( Figure 2J , Supplementary Figure 8 , 10D ), TNF production ~25-fold higher (Flu-NP 265 : 3.31 ± 2.28 %; KCY: 0.13 ± 0.06 %), and CD107a ~10-fold higher than KCY (Flu-NP 265 : 3.32 ± 2.36 %; KCY: 0.34 ± 0.33 %) ( Supplementary Figure 10E ). The response towards the Flu-NP 265 peptide demonstrated that HLA-A*03:01 + CD8 + T cells can produce high level of cytokines. However, the CD8 + T cell response towards the Spike-derived peptides, even for the dominant KCY peptide, was overall weak even after vaccination, and therefore unlikely to underpin the vaccine side effect correlation observed in HLA-A*03:01 + individuals. Full-length Spike protein stimulated high expression of IL-6 and IL-8 by monocytes Despite the high number of KCY + T cells present ( Figure 2A ), the overall T cell response did not show the high levels of cytokine production that could explain vaccine reactogenicity associated with HLA-A*03:01 carriage ( Figure 2H-I ). Therefore, we asked whether other immune cells could lead to inflammation that would underpin vaccine side effects. To address this, we used the different vaccine components to stimulate PBMCs; the empty lipid nanoparticle (LNP) using the Pfizer vaccine formula 43 the soluble full-length Spike protein (HexaPro) 44 , or the Spike-derived peptide pools (S1 and S2), as well as positive and negative controls. Among the 20 HLA-A*03:01⁺ and 14 HLA-A*03:01 - PBMC samples tested ( Supplementary Table 20 ), detectable cytokine production was only observed in samples stimulated with the full-length soluble Spike protein ( Figure 3 , Supplementary Figure 11 ). Neither the Spike-derived peptide pools nor the LNPs induced measurable cytokine responses in either group, except for Monocyte Chemoattractant Protein-1 (MCP-1) ( Figure 3 ). Although MCP-1 levels showed a modest increase in HLA-A*03:01⁺ samples following Spike stimulation (120.7 ± 340.9 pg/mL) compared with HLA-A*03:01 - samples (0.0 ± 0.0 pg/mL), the response remained within the expected baseline variation observed in healthy individuals (~250 pg/mL), indicating no biologically meaningful induction of MCP-1 by Spike stimulation. Similar response was observed for eLNP and Spike Pool stimulations (253.4 ± 432.1 pg/mL and 333.3 ± 523.5 pg/mL, respectively). Cytokines induced by whole Spike stimulation, included Interleukin (IL)-6, IL-8, IL-1a, IL-1b, IL-10, macrophage inflammatory protein-1a (MIP1a), granulocyte-macrophage colony-stimulating factor (GM-CSF), RANTES, MCP-1, tumor necrosis factor (TNF), inducible protein 10 kDa (IP-10), and Interferon-γ (IFNγ) ( Figure 3 , Supplementary Figure 11 ). The increase was observed in both HLA-A*03:01 + and HLA-A*03:01 - samples, and no significant difference in the level of cytokine was observed between the groups. However, IL-6 and IL-8 were modestly elevated in HLA-A*03:01⁺ samples after Spike stimulation (12,186 ± 13,659 pg/mL and 29,875 ± 30,682 pg/mL, respectively), representing ~1.1- and ~1.3-fold increase compared with IL-6 (11,234 ± 11,177 pg/mL) and IL-8 (22,436 ± 20,784 pg/mL) levels in HLA-A*03:01 - samples ( Figure 3 ). IL-6 and IL-8 cytokines were also expressed at the highest concentration compared to other cytokines. We next asked which cell subset was responsible for the production of IL-6 and IL-8 upon Spike stimulation. To address this, cytokine production was assessed in Spike-stimulated PBMCs using flow cytometry using multiple cell surface markers. IL-6 was produced predominantly by monocytes (CD14 + ) despite the low number of CD14 + cells in blood ( Supplementary Figure 12A - B ), and at lower levels by Natural Killer (NK) cells (CD56 + ), suggesting a primary role of the innate response in vaccine reactogenicity ( Figure 4A ). In HLA-A*03:01⁺ samples, IL-6⁺ NK cells averaged 0.23 ± 0.23 % (n = 9/9 positive) and monocytes 7.7 ± 11.9 % (n = 3/9), whereas in HLA-A*03:01 - samples comparable NK responses (0.45 ± 0.56 %, n = 7/9) but broader monocyte positivity (9.5 ± 12.4 %, n = 5/9) were observed ( Figure 4A ). In contrast, IL-6⁺ T and B cells were rare, with only traceable responses (< 0.03 %, n ≤ 6/9 donors). IL-8 responses were even more restricted. Only a minority of monocyte-positive donors showed detectable IL-8, averaging 2.2 ± 6.7 % (n = 1/9) in HLA-A*03:01⁺ and 3.3 ± 3.8 % (n = 5/9) in HLA-A*03:01 - samples. Other subsets (NK, T, and B cells) produced negligible IL-8 (< 0.02 %, n ≤ 5/9 donors) ( Figure 4B ). Compared to the IL-6 and IL-8 levels secreted ( Figure 3A ), the levels of IL-6⁺ and IL-8⁺ cells detected were at low frequencies in PBMCs, therefore, we examined the Spike uptake capacity by the PBMCs. Phagocytic scores (engulfment of Spike-coated microbeads; Figure 4C , Supplementary Figure 12C ) confirmed that Spike uptake was mainly driven by monocytes and at lower level by NK cells. Overall, the high level of IL-6 and IL-8 production was observed only in the presence of the full Spike protein and was primarily produced by monocytes. Expression of IL-6 and IL-8 correlates with side effect severity in HLA-A*03:01 donors While our results suggest a role for IL-6 and IL-8 in vaccine reactogenicity, the differences observed between grouped HLA-A*03:01⁺ and HLA-A*03:01 - donor samples were not sufficient to explain the observed differences in response to vaccination. However, as with most complex phenotypes, the association of HLA-A*03:01⁺with vaccine reactogenicity is incompletely penetrant. While more individuals carrying this allele report SSE with vaccination, there was a range of severity reported among our PBMC donors; thus, we sought to determine whether levels of IL-6 and IL-8 were correlated with reported side effect severity in these donors. Strikingly, in HLA-A*03:01⁺ samples, both IL-8 and IL-6 levels showed strong and significant positive correlations with severity score (IL-8: r = 0.70, p = 0.02; IL-6: r = 0.75, p = 0.01; Figure 4D-E ), indicating that higher cytokine production was associated with increased vaccine side effect severity. In contrast, no significant correlation was observed in HLA-A*03:01 - individuals ( Figure 4F-G ). To ensure that the correlation observed in HLA-A*03:01 + samples was due to the transient presence of Spike protein, we checked the baseline levels of IL-6 and IL-8 cytokines in serum of HLA-A*03:01 + collected prior to vaccination, or two-weeks post first and second dose of vaccine, alongside TNF and IFNγ as control ( Supplementary Figure 13 ). Overall, the level of cytokines was low (< 100 pg/mL) or moderate, and no significant increase was observed before or after vaccination. This demonstrates that the cytokine production upon Spike presentation is likely transient. Together, these data establish that HLA-A*03:01⁺ samples display a distinct pro-inflammatory signature, with IL-6 and IL-8 production strongly linked to vaccine side-effect severity. The early and transient nature of this cytokine production primarily by monocytes and NK cells strongly suggests an innate immune response underlying HLA-A*03:01 vaccine reactogenicity. HLA-A*03:01 is an eQTL for IRF4 driving monocyte differentiation Finally, to better understand the relationship between HLA-A*03:01 carriage and the observed role of innate immune cells in vaccine reactogenicity , we examined patterns of differential gene expression driven by HLA-A*03:01 . Limiting our analysis to genes on chromosome 6 (723 tests), differential gene analysis for 21 donors showed that IRF4, which is involved in monocyte differentiation to dendritic cells (DCs) and homing to lymph nodes 45 , is significantly upregulated in HLA-A*03:01 donors (log fold 1.863, p-value = 9.56 × 10 -6 , p adj = 0.0069) ( Supplementary Table 22 ). Thus, it appears that the HLA-A*03:01 association with SARS-CoV-2 vaccine reactogenicity may be in part due to its role as an eQTL, resulting in increased differentiation of monocytes to DCs, increasing antigen-presenting cells and homing them to the lymph nodes. Discussion Vaccination has been a crucial public health intervention for decades, significantly contributing to the prevention and control of infectious diseases worldwide. However, concerns about vaccine safety and efficacy, and negative public perceptions regarding side effects, have emerged as a significant challenge in maintaining high vaccination coverage. Because the binding between HLA and peptide antigen is highly specific and a fundamental component in initiating the adaptive immune system, understanding the role of HLA variation in vaccine response can be crucial in determining factors that underlie the effectiveness of vaccination. Variation in HLA has previously been reported as associated with SARS-CoV-2 vaccine reactogenicity 30,31,37 . Here, in a much larger cohort than previously examined, we sought to refine and understand the relationship between variation at all HLA loci and reports of side effects associated with the vaccine. We leveraged a large, registry-based cohort of more than 50,000 individuals to provide the necessary statistical power and diversity to reliably identify genetic factors that influence individual susceptibility to systemic reactions following vaccination. Variation in the HLA region has previously been associated with interindividual differences in humoral immune responses after vaccination. For example, HLA variation has been linked to either increased or decreased immune responses to influenza 46 , measles 47 , rubella 48 , and hepatitis B vaccination 49 , respectively. In COVID-19 vaccination, previous reports have shown a suggested association of HLA variation with specific systemic mild side effects such as fever and fatigue, including HLA-A*03:01 30 . Here, we showed significant associations of HLA-A*03:01 with increased side effects such as fever, chills, and muscle pain, particularly with higher effect sizes in individuals who received the Pfizer vaccine. Evidence from this and prior studies 19,50 demonstrates that HLA-A*03:01 is also significantly associated with high serum levels of anti-SARS-CoV-2-Spike antibodies. This supports the notion that this allomorph promotes an especially robust immune response to vaccination for SARS-CoV-2, including both cell-mediated and humoral immunity. Additionally, we demonstrated that a negative correlation of HLA-A*03:01 with BTI is likely mediated by the systemic inflammatory response that causes vaccination adverse effects. To fully contextualize the findings, we compared responses to COVID-19 vaccines with those of influenza vaccine. Our finding of no significant association of HLA-A*03:01 with side effects from influenza vaccine (or any other HLA allele) demonstrates that the observed HLA-A*03:01 associated vaccine reactogenicity is specific to the COVID-19 vaccine. Owing to the crucial role of HLA class I molecules in antigen presentation, the role of CD8 + T cells in HLA-A*03:01-mediated vaccine reactogenicity presented a clear initial line of inquiry into the mechanisms underlying reported side effects. We observed a high frequency of CD8 + T cells able to recognize the immunodominant Spike-derived peptide KCY specifically 20 , even prior to vaccination or infection. Interestingly, both prior to and subsequent to vaccination, the majority of the CD8 + T cells exhibited low MFI (Mean Florescence Intensity), suggesting that the cells are low avidity; in addition, a large proportion of those peptide-specific cells had a naïve phenotype. This contrasts with our previous study on a dominant Influenza-derived peptide, Flu-NP 265 42 for which we observed low and high MFI cells ex vivo , consistent with the fact that prior viral exposure and/or vaccination leads to the presence of high MFI T cells. Post-COVID-19 vaccination we did observe the expansion of high MFI KCY-specific CD8 + T cells and an increased proportion of memory phenotype. However, in contrast with HLA-A*02:01 + samples, even after multiple doses of the COVID-19 vaccine, the frequency of KCY + naïve CD8 + T cells remained high, likely reflecting the large proportion of naïve T cells able to specifically bind the KCY peptide independent of vaccine status. Most strikingly, even post-vaccination, the T cell response to KCY was muted relative to responses to Influenza-derived peptide, such that it is unlikely that the observed COVID-19 vaccine reactogenicity can be attributed to a robust T cell activation. Monocytes thus emerged as a central population of interest: they were the predominant source of IL-6 and IL-8 following Spike stimulation, driving the pro-inflammatory response that correlated with side-effect severity in HLA-A*03:01 ⁺ donors. Strikingly, this cytokine induction was accompanied by a reduction of the already low frequency of CD14⁺ monocytes in blood, suggesting a shift in lineage fate rather than simple activation, consistent with our RNA-sequencing results. Our findings reveal a distinct gene expression signature associated with HLA-A*03:01 , suggesting upregulation of the monocyte to DC differentiation pathway. Increased IRF4 expression in HLA-A*03:01 + individuals suggests increased DC priming and more efficient migration to lymph nodes 45 . Given that CD14 downregulation is a hallmark of monocyte-to-dendritic cell differentiation, our results suggest that HLA-A*03:01 ⁺ individuals likely have higher frequencies of dendritic cells, highly efficient antigen presenters, homed to the lymph node than HLA-A*03:01 - individuals. Despite what appears to be the central role of the monocyte-DC lineage in cytokine production associated with vaccine reactogenicity, the unusually high numbers of low-avidity, naïve, Spike-specific CD8 + T cells observed in HLA-A*03:01 + donors likely provide an inflammatory milieu in the lymph node after vaccination. While these T cells do not appear to become activated themselves, their binding to peptide-HLA presented by DCs at high frequency may constitute a signal of immune activity for DCs 51 ; this likely contributes to activation of these DCs, resulting in increased production of IL-6 and other cytokines 52,53 . Thus, we postulate that in HLA -A*03:01 + individuals, a high number of low-avidity, naïve T cells is available to bind to already primed DCs stimulated by Spike, resulting in an amplified immune response after COVID-19 vaccination. This study is intrinsically constrained by its dependence on self-reported data to assess transient mild vaccine side effects in both the discovery and replication cohorts, which potentially can lead to some imprecision in association results. Additionally, sample size limitations restrict some significant findings to individuals who self-identify as White. Likewise, our cohort was predominantly female (78.4%), which may limit the generalizability to broader populations. Moreover, the low number of monocyte cells in the blood limited the ability to show significant differences, in addition to the incomplete penetrance of the observed genetic effect, which is typical for complex traits. The transient features of SSE also constrained these observations. Despite these limitations, our findings regarding the role of HLA- mediated COVID-19 vaccine reactogenicity and the associated evidence for protection from subsequent infection provide important and novel insights regarding these responses, which may inform efforts toward improved vaccine efficacy and increased public participation in vaccination programs. Methods Discovery cohort We recruited our study population via email to all potential volunteer bone marrow donors registered in the NMDP database with available email addresses and high-resolution HLA genotyping information available. The email contained a custom link directing them to a consent page for a health history survey. Subject recruitment, consent process, and survey administration were conducted using both email outreach and a web interface to ensure effective data collection. Upon consenting to provide responses and allow linking with their HLA genotype data, participants spent ten to fifteen minutes completing a detailed survey to gather baseline information and health history. As of January 19th, 2024, a total of 80,016 eligible donors completed the survey. Among these respondents, 667 individuals (0.83%) were excluded for filling out the survey multiple times, while 1,301 participants were removed due to incomplete HLA variation data. After excluding these cases and individuals that were also participated in the study that formed our replication cohort, below (N = 836), there remained a total of 77,212 participants in the study. Of these individuals, 56,938 people who self-identified as White, Hispanic, African American, or Asian Pacific Islanders, had completed their initial series of vaccinations. We excluded participants who identified as multiple ancestry, unknown ancestry or Native American due to small sample sizes. Replication cohort Our replication cohort consisted of participants who participated in a prior study with NMDP tracking experiences with COVID-19 through a mobile app, described in detail in Augusto et al. 2023 23 . Once enrolled, the participants are asked to complete an initial 10 to 15-minute survey about baseline demographics, their health history, and daily habits. Follow-up daily questions specific to vaccine side effects are delivered by push notification or text message on an ongoing basis and require 5 to 15-minute per week. All the participants provided written informed consent agreeing to the research and publication of research results. We restricted our analysis to individuals who had self-identified as ‘White’ (which we use as a proxy for European ancestry) due to insufficient numbers for analysis in the other groups, allowing an analysis of 10,595 individuals reporting vaccination for SARS-CoV-2. Of those, 4,575 individuals completed their initial series of vaccinations. Symptoms are self-reported at the baseline and in daily surveys. Within the baseline survey, the respondents were asked to report whether they had any of a list of symptoms ( Supplementary Table 23 ) for 3 days or longer at any time after their complete dose of vaccination. Serum antibody levels in vaccinated subjects The population examined consisted of 156 healthcare workers from “Evangelismos” General Hospital in Athens, Greece, including doctors, nurses, pharmacists, biologists, dentists, technicians and administrative staff. Enrollment was open to all hospital personnel scheduled for vaccination and not restricted by any pre-specified criteria. All individuals received two doses of the mRNA Pfizer-BioNTech vaccine. Data on prior SARS-CoV-2 infection and symptoms experienced after each dose were collected for all participants. Antibody concentrations were assessed at two time points: 21 ± 1 days after the first dose and 24 ± 2 days after the second dose. Levels of circulating SARS-CoV-2 anti-Spike IgG (S) and anti-nucleocapsid IgG (N) antibodies were quantified using the Abbott Diagnostics SARS-CoV-2 IgG chemiluminescent microparticle immunoassay (Abbott Diagnostics, Abbott Park, Illinois) on an Abbott Diagnostics Architect i2000 SR and an Alinityi Analyzer, according to the manufacturer’s instructions. Results were expressed in AU/mL and were interpreted as positive if ≥ 50 AU/mL 54 . Informed consent was obtained from all participants, and the study was approved by the Institutional Review Board of “Evangelismos” Hospital (PN 9/21-01-21). High resolution HLA class I and II genotyping was performed as described 55 . Peripheral blood mononuclear cells (PBMCs) HLA-A*03:01 + and HLA-A*03:01 - donors vaccinated with either the Comirnaty BNT162b2 COVID-19 mRNA vaccine (Pfizer) or the Oxford–AstraZeneca COVID‑19 vaccine (AstraZeneca), and most of them naïve for SARS-CoV-2 infection, were recruited ( Supplementary Table 20 ). PBMCs were separated from whole blood or buffy coats using density-gradient centrifugation. PBMCs were used fresh or were cryogenically stored until use. All individuals consented to research and publication of research results and had been previously HLA genotyped. Ethics approval to undertake the research was obtained from the La Trobe University Human Research Ethics Committee (HEC21097). The HLA genotyping was performed by AlloSeq Tx17 (CareDx Pty) using AllType NGS high-resolution genotyping on the IonTorrent NGS platform or by the Department of Clinical Immunology and PathWest at Fiona Stanley Hospital, Murdoch, Australia. Tetramer-associated magnetic enrichment (TAME) Peptide-loaded HLA-A*03:01 tetramers were generated using Streptavidin conjugated to phycoerythrin (PE). Tetramer-stained cells were enriched using anti-PE antibody-coated immunomagnetic beads on LS columns (Miltenyi Biotech) according to manufacturer instructions. After enrichment, cells were stained with an antibody panel including anti-CD3-BV480 (dilution 1:100), anti-CD8-PerCP-Cy5.5 (1:50), anti-CD4-FITC (1:100), anti-CD14-APCH7 (1:200), anti-CD19-APCH7 (1:100), anti-CD45RA-BUV395 (1:100), anti-CD27-APC (1:100), anti-CCR7-PE-Cy7 (1:50), anti-CD95-BV421 (1:50), anti-PD1-BV605 (1:100), anti-CXCR5-BV650 (1:100) (all BD Biosciences) and Live/Dead Fixable Near-IR Dead Cell Stain (1:1,000) (Life Technologies). Cells were resuspended in MACS buffer (PBS, 0.5% BSA, 2 mM EDTA) and were analysed using the BD FACSymphony A3 system. For the tetramer staining experiments, the TAME cells were stained for 1 hour at room temperature with the APC-conjugated Flu-NP 265 HLA-A*03:01-restricted peptide tetramer, followed by surface staining using the same antibody panel as above, excluding anti-CD27-APC. Gating strategy shown on Supplementary Figure 14 . Generation of peptide-specific CD8 + T cell lines CD8 + T cell lines were generated as previously described 56,57 . In brief, PBMCs were incubated with 1 μM of individual SARS-CoV-2 Spike-derived peptide or 10μg/mL of Spike-derived peptide Pool 1 (25 μg/peptide, 15mers, 1 – 126) and Pool 2 (25 μg/peptide, 15mers, 127 – 253) (Mimotopes B#33200); and cultured for 10 – 14 days in RPMI-1640 supplemented with 2 mM MEM non-essential amino acid solution (Sigma-Aldrich), 100 mM HEPES (Sigma-Aldrich), 2 mM l-glutamine (Sigma-Aldrich), penicillin–streptomycin (Life Technologies), 50 mM 2-ME (Sigma-Aldrich) and 10% heat-inactivated fetal bovine serum (Bovogen). The cultures were supplemented with 10 IU IL-2 2 – 3 times weekly. CD8 + T cell lines were used fresh for subsequent analysis. For the tetramer staining experiments 0.5 × 10 6 cells from the CD8 + T cell lines were stained with a PE-conjugated tetramer (HLA-A*03:01-KCY) or double-stained with two tetramers (PE-conjugated KCY and APC-conjugated Flu-NP 265 HLA-A*03:01-restricted peptide tetramer) for 1 h at room temperature. Cells were washed and surface-stained with anti-CD3-BV480 (dilution 1:100), anti-CD8-PerCP-Cy5.5 (1:50), anti-CD4-BV650 or -FITC (1:100), anti-CD14-APCH7 (1:200) and anti-CD19-APCH7 (1:100) antibodies (all BD Biosciences) and Live/Dead Fixable Near-IR Dead Cell Stain (Life Technologies). Cells were analysed using the BD FACSymphony A3 system. Intracellular cytokine assay CD8 + T cell lines were stimulated with 1 μM of individual peptide or 2μg/mL of the SARS-CoV-2 Spike-derived peptide Pool 1 (25 μg/peptide, 15mers, 1 – 126) and Pool 2 (25 μg/peptide, 15mers, 127 – 253) (Mimotopes B#33200) and were incubated for 4 – 5 hour in the presence of GolgiPlug, GolgiStop and anti-CD107a-FITC (dilution 1:100) (all BD Biosciences). After stimulation, cells were surface stained for 30 min with anti-CD3-BV480 (1:100), anti-CD8-PerCP-Cy5.5 (1:50) and anti-CD4-BV650 (1:100) antibodies (all BD Biosciences) and Live/Dead Fixable Near-IR Dead Cell Stain (Life Technologies). Cells were fixed and permeabilized using BD Cytofix/Cytoperm solution (BD Biosciences) and then intracellularly stained with anti-IFN-γ-BV421 (1:100), anti-TNF-PE-Cy7 (1:100), anti-IL2-PE (1:100) and anti-MIP-1β-APC (1:100) antibodies (all BD Biosciences) for a further 30 min. Cells were acquired on the BD FACSymphony A3 system using the FACSDiva software (v.9.0.). Post-acquisition analysis was performed using FlowJo software (v.10). Cytokine detection levels identified in the no-peptide control condition were subtracted from the corresponding test conditions in all summary graphs to account for non-specific, spontaneous cytokine production. Gating strategy shown on Supplementary Figure 14 . PBMCs short-term stimulation 1 x 10^6 PBMCs were stimulated with either 15 μg/mL of empty Pfizer-BioNTech lipid nanoparticle (LNP) 43 or 15 μg/mL custom made SARS-CoV-2 Spike protein (Wuhan strain); or 7.5 μg/mL of the SARS-CoV-2 Spike-derived peptide Pool 1 (25 μg/peptide, 15mers, 1 – 126) and 7.5 μg/mL Pool 2 (25 μg/peptide, 15mers, 127 – 253) (MIMOTOPES B#33200); or Cell Stimulation Cocktail (500X) (eBioscience™); or nothing (Negative Control) for 15 hours in RPMI-1640 supplemented with 2 mM MEM non-essential amino acid solution (Sigma-Aldrich), 100 mM HEPES (Sigma-Aldrich), 2 mM l-glutamine (Sigma-Aldrich), penicillin–streptomycin (Life Technologies), 50 mM 2-ME (Sigma-Aldrich) and 10% heat-inactivated fetal bovine serum (Bovogen). Following the 15-hour stimulation, the cells were restimulated for a further 2-hour using the same conditions and the supernatant was collected for the BD Cytometric Bead Array (CBA). The same conditions were also used in the presence of GolgiPlug and GolgiStop (BD Biosciences). Following the 15-hour stimulation and the 2-hour restimulation, the cells were surface-stained for 30 minutes with anti-CD3-BV480 (1:100), anti-CD14-PerCP-Cy5.5 (1:100), anti-CD16-BV421 (1:50), anti-CD19-APC (1:50), anti-CD56-PECy7 (1:50) antibodies (all BD Biosciences) and Live/Dead Fixable Near-IR Dead Cell Stain (Life Technologies). Cells were fixed and permeabilized using BD Cytofix/Cytoperm solution (BD Biosciences) and then intracellularly stained with anti-IL6-FITC (1:50) and anti-IL8-PE (1:50) antibodies (both BD Biosciences) for a further 30 minutes. Cells were acquired on the BD FACSymphony A3 system using the FACSDiva software (v.9.0.). Gating strategy shown on Supplementary Figure 15 . BD Cytometric Bead Array (CBA) Following the PBMC short-term stimulation, the supernatant level of Interleukin (IL)-1α, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-9, IL-10, IL-12p70, IL-13, Interferon (IFN)-α, IFN-γ, IFN-γ inducible protein 10 kDa (IP-10), granulocyte-macrophage colony-stimulating factor (GM-CSF), Lymphotoxin-alpha (LT-α), Eotaxin, Monocyte Chemoattractant Protein-1 (MCP-1), macrophage inflammatory protein-1 alpha (MIP-1α), RANTES, tumor necrosis factor (TNF) and were measured using the BD Cytometric Bead Array (CBA, BD Biosciences) following the manufacturer’s instructions. Samples were acquired in a BD FACSymphony A3 system using the FACSDiva software (v.9.0.). The analysis was performed by using the FCAP Array Software v3.0. Cytokine detection levels identified in the Negative Control condition were subtracted from the corresponding test conditions in all summary graphs to account for non-specific, spontaneous cytokine production. SARS-CoV-2 specific phagocytosis assays After PBMCs stimulation with full length Spike protein for 15-hour, PBMCs (1 × 10 5 cells in 50µL) were added onto 60 µL of the donor plasma opsonised microbeads (10µL plasma and 50µL microbeads) in a 1.5mL Eppendorf tube, mixed by gentle tapping, adjusted to 600 µL using RPMI 1640 containing 0.1% Human serum and 0.1 M HEPES pH 7.4 and transferred into 37ºC, 5 % CO 2 incubator. After 2-hour of incubation, cells were washed once with 1mL of cold PBS containing 0.5 % FBS and 0.005 % sodium azide and gentle centrifugation at 335 x g for 5 minutes at 4ºC, fixed in 400 µL of 1 % paraformaldehyde, and kept at 4ºC in the dark until the acquisition of data using BD FACSCaliburTM Flow cytometer. A total of 2 x10 4 events per tube were acquired from each donor conditions. Relevant assay controls included the acquisition of 2 x10 4 events per tube from cells incubated with no beads, Spike-coated nonopsonized beads. The proportions of cells that phagocytosed the beads (% of cells that took up the beads) and their fluorescent intensities (amounts of beads taken up per cell) were analyzed using BD FlowJo version 10.5.0 software. Phagocytic scores (p-score) were then calculated based on the proportion of cells that took up the opsonized beads denoting the number of positive cells and mean fluorescence intensity (MFI) representing the average bead uptake by the positive cells as described. A positive p-score was defined as three standard deviations above the background mean phagocytic score of healthy donors as described previously 58 . In selected experiments, the intracellular uptake of the opsonized microbeads by effector cells was confirmed by confocal microscopy as described 58 . In brief, PBMC after (1 × 10 4 cells) after phagocytosis assay were washed twice with cold PBS containing 0.5% FBS and 0.005 % of sodium azide, fixed with 1% paraformaldehyde for 5 minutes at room temperature, and rinsed twice with PBS. The fixed cells were blocked with 1% BSA in PBS, incubated with 1:1000 dilution of Alexa-555-conjugated Phalloidin (Sigma, USA) for 30 minutes at room temperature, mounted in DAPI nuclear stain-containing media (Molecular Probes, USA), and imaged using ZEISS LSM 880 confocal microscope (Carl Zeiss AG, Germany), using 63X/1.4 Plan-Apochromat Oil Immersion objective, with Diode 405 nm (DAPI), Argon ion 488 nm (Alexa-488) and DPSS 561 nm (Alexa-555 phalloidin) laser excitation sources, emitted light was filtered using a combination of emission filters and imaged onto Airy detector array producing an effective lateral resolution of ~100 nm. All the images were Airyscan processed with Zen Black Edition (Zeiss Software). Assessment of vaccine side effects To standardize the evaluation of vaccine-associated side effects, we developed a composite scoring framework termed the Side Effect Severity Score (SESS) adapted from the vaccine side effect guideline described elsewhere 59 . Each reported symptom was first categorized by type and severity (local, systemic, or severe) according to previously established criteria for vaccine reactogenicity. For each donor and vaccine dose, symptoms were graded as: 0 = none; 1 = mild (e.g., injection site pain, mild fatigue, or a single local symptom); 2 = moderate (systemic but not severe, e.g., fever, chills, headache, myalgia, or multiple mild symptoms); or 3 = severe (multiple systemic symptoms, prolonged recovery >2 days, swelling requiring medical review, dose-limiting reaction, or hospitalization). To account for compounded burden, additional multipliers were applied: +1 if multiple symptoms occurred at the same dose, +1 if symptoms recurred across multiple doses, and +1 if symptom duration exceeded 2 days (if reported). Scores were then summed to generate an overall SESS per participant, which was categorized as: 0 (no side effects), 1 – 2 (mild), 3 – 4 (moderate), and ≥ 5 (severe). The scores are summarized in Supplementary Table 21 . RNA Sequencing We performed bulk RNAseq in a total of 21 samples of PBMCs (n = 10 HLA-A*03:01 + , n = 11 HLA-A*03:01 - ) and compared the two groups based on HLA genotype. Total RNA libraries were prepared and sequenced by Novogene Corporation using Illumina NovaSeq platforms following standard protocols 60 . RNA quantity and integrity were assessed to ensure a minimum RNA Integrity Number (RIN) of > 3.0. The raw sequence data generated by Novogene met strict quality criteria as described by the provider 60,61 . RNAseq data processing and analysis We analyzed all 21 donors from an independent cohort which included individuals of European and non-European ancestries. Samples were sequenced in two batches, with raw RNA-seq data merged and then normalized for batch effects ( Supplementary Figure 16 ) using Combat-seq 62 These sequences were then analyzed using the nf-core/rnaseq pipeline (version 3.19.0), executed through Nextflow 63 . Initial quality control was performed with FastQC, followed by adapter and low-quality base trimming using Trim Galore. Reads were then aligned to the reference genome [Human Genome Assembly GRCh38.p14] using STAR. Transcript quantification was accomplished using Salmon, as defined by the pipeline configuration. The resulting gene-level count matrices were imported into R for normalization and differential expression analysis using the DESeq2 package 64 . Additional downstream analyses, including gene set enrichment analysis, were conducted using standard R workflows. Statistical analysis HLA associations: In our discovery cohort, we examined the association of five HLA loci ( HLA-A, -B, -C, -DRB1, -DQB1 ). Data analysis included the first two fields of the allele name as described in the HLA nomenclature, representing the complete molecule at polypeptide sequence resolution. We calculated allele frequencies for all the HLA loci, haplotype frequencies using Haplostats R package 65 and R2 for Linkage Disequilibrium (from our in-house script) between all the pair of loci ( Supplementary Table 14 ). We employed a generalized logistic regression model using ‘glm’ in the R (V 4.3) base package to consider relevant covariates, including sex and age. For the replication cohort, we tested only the allele of interest, using the generalized logistic regression model framework as described. We utilized our in-house Python script to construct forest plots. We conducted an analysis of variance (ANOVA) to assess the statistical significance of differences in antibody positivity rates across groups with different counts of HLA-A*03:01 allele(0,1 or 2). All additional analyses were performed in GraphPad Prism v10.1. Data are shown as mean ± SD, with symbols representing individual donors. Paired comparisons across time points were assessed using two-tailed Wilcoxon matched-pairs tests, and unpaired comparisons between HLA-A*03:01⁺ and HLA-A*03:01⁻ groups by two-tailed Mann–Whitney U tests. For the CBA, cytokine values were background-subtracted from unstimulated or controls; analytes below detection limits were excluded. Correlations between cytokine levels and side-effect severity were evaluated using Spearman’s r. Frequencies of cytokine-producing or tetramer-positive T cells and phagocytic scores were compared using non-parametric tests. All tests were two-tailed; p < 0.05 was considered significant. Declarations Acknowledgements This work was supported by NIH R01 AI159260 and NIH R01 AI158861 to JAH. National Health and Medical Research Council (NHMRC, GNT2014002 and GNT1161832) and Australian Research Council (ARC); National Collaborative Research Infrastructure Strategy (NCRIS) Therapeutic Innovation Australia (TIA) to T.R.M.; University of Queensland to T.R.M. SG is supported by an NHMRC Leadership Investigator Grant (#2034677). PJN was supported by NIH R01 AI158410. We wish to thank the volunteer donors registered with NMDP who participated in this study. References Rahmani, K. et al. The effectiveness of COVID-19 vaccines in reducing the incidence, hospitalization, and mortality from COVID-19: A systematic review and meta-analysis. Front Public Health 10 , (2022). Wu, N. et al. Long-term effectiveness of COVID-19 vaccines against infections, hospitalisations, and mortality in adults: findings from a rapid living systematic evidence synthesis and meta-analysis up to December, 2022. Lancet Respir Med 11 , 439–452 (2023). DeCuir, J. et al. Interim Effectiveness of Updated 2023–2024 (Monovalent XBB.1.5) COVID-19 Vaccines Against COVID-19–Associated Emergency Department and Urgent Care Encounters and Hospitalization Among Immunocompetent Adults Aged ≥18 Years — VISION and IVY Networks, September 2023–January 2024. MMWR Morb Mortal Wkly Rep 73 , 180–188 (2024). Payne, A. B. et al. Effectiveness of Bivalent mRNA COVID-19 Vaccines in Preventing COVID-19–Related Thromboembolic Events Among Medicare Enrollees Aged ≥65 Years and Those with End Stage Renal Disease — United States, September 2022–March 2023. MMWR Morb Mortal Wkly Rep 73 , 16–23 (2024). Polack, F. P. et al. Safety and Efficacy of the BNT162b2 mRNA Covid-19 Vaccine. New England Journal of Medicine 383 , 2603–2615 (2020). Mascellino, M. T., Di Timoteo, F., De Angelis, M. & Oliva, A. Overview of the Main Anti-SARS-CoV-2 Vaccines: Mechanism of Action, Efficacy and Safety. Infect Drug Resist Volume 14 , 3459–3476 (2021). Sadoff, J. et al. Final Analysis of Efficacy and Safety of Single-Dose Ad26.COV2.S. New England Journal of Medicine 386 , 847–860 (2022). Alcendor, D. J. et al. Breakthrough COVID-19 Infections in the US: Implications for Prolonging the Pandemic. Vaccines (Basel) 10 , 755 (2022). Suntronwong, N. et al. COVID-19 Breakthrough Infection after Inactivated Vaccine Induced Robust Antibody Responses and Cross-Neutralization of SARS-CoV-2 Variants, but Less Immunity against Omicron. Vaccines (Basel) 10 , 391 (2022). Gopinath, S. et al. Characteristics of COVID-19 Breakthrough Infections among Vaccinated Individuals and Associated Risk Factors: A Systematic Review. Trop Med Infect Dis 7 , 81 (2022). Suleyman, G. et al. Risk Factors Associated With Hospitalization and Death in COVID-19 Breakthrough Infections. Open Forum Infect Dis 9 , (2022). Lopera, T. J. et al. Humoral Response to BNT162b2 Vaccine Against SARS-CoV-2 Variants Decays After Six Months. Front Immunol 13 , (2022). Wiedermann, U., Garner-Spitzer, E. & Wagner, A. Primary vaccine failure to routine vaccines: Why and what to do? Hum Vaccin Immunother 12 , 239–243 (2016). Chen, J., Wang, R., Gilby, N. B. & Wei, G.-W. Omicron Variant (B.1.1.529): Infectivity, Vaccine Breakthrough, and Antibody Resistance. J Chem Inf Model 62 , 412–422 (2022). Notarte, K. I. et al. Characterization of the significant decline in humoral immune response six months post‐SARS‐CoV‐2 mRNA vaccination: A systematic review. J Med Virol 94 , 2939–2961 (2022). Falahi, S. & Kenarkoohi, A. Host factors and vaccine efficacy: Implications for COVID‐19 vaccines. J Med Virol 94 , 1330–1335 (2022). Lynn, D. J., Benson, S. C., Lynn, M. A. & Pulendran, B. Modulation of immune responses to vaccination by the microbiota: implications and potential mechanisms. Nat Rev Immunol 22 , 33–46 (2022). Mentzer, A. J. et al. Human leukocyte antigen alleles associate with COVID-19 vaccine immunogenicity and risk of breakthrough infection. Nat Med 29 , 147–157 (2023). Esposito, M. et al. Human leukocyte antigen variants associate with BNT162b2 mRNA vaccine response. Communications Medicine 4 , 63 (2024). Mayer-Blackwell, K. et al. mRNA vaccination boosts S-specific T cell memory and promotes expansion of CD45RAint TEMRA-like CD8+ T cells in COVID-19 recovered individuals. Cell Rep Med 4 , 101149 (2023). Bian, S. et al. Genetic determinants of IgG antibody response to COVID-19 vaccination. The American Journal of Human Genetics 111 , 181–199 (2024). Blackwell, J. M., Jamieson, S. E. & Burgner, D. HLA and Infectious Diseases. Clin Microbiol Rev 22 , 370–385 (2009). Augusto, D. G. et al. A common allele of HLA is associated with asymptomatic SARS-CoV-2 infection. Nature 620 , 128–136 (2023). Srivastava, A. & Hollenbach, J. A. The immunogenetics of COVID-19. Immunogenetics 75 , 309–320 (2023). Hervé, C., Laupèze, B., Del Giudice, G., Didierlaurent, A. M. & Tavares Da Silva, F. The how’s and what’s of vaccine reactogenicity. NPJ Vaccines 4 , 39 (2019). Shimabukuro, T. T., Cole, M. & Su, J. R. Reports of Anaphylaxis After Receipt of mRNA COVID-19 Vaccines in the US—December 14, 2020-January 18, 2021. JAMA 325 , 1101 (2021). Zhuang, C.-L. et al. Inflammation-related adverse reactions following vaccination potentially indicate a stronger immune response. Emerg Microbes Infect 10 , 365–375 (2021). Hermann, E. A. et al. Association of Symptoms After COVID-19 Vaccination With Anti–SARS-CoV-2 Antibody Response in the Framingham Heart Study. JAMA Netw Open 5 , e2237908 (2022). Yoshida, M. et al. Association of systemic adverse reaction patterns with long-term dynamics of humoral and cellular immunity after coronavirus disease 2019 third vaccination. Sci Rep 13 , 9264 (2023). Bolze, A. et al. HLA-A∗03:01 is associated with increased risk of fever, chills, and stronger side effects from Pfizer-BioNTech COVID-19 vaccination. Human Genetics and Genomics Advances 3 , 100084 (2022). Magri, C. et al. Genome‐wide association studies of response and side effects to the BNT162b2 vaccine in Italian healthcare workers: Increased antibody levels and side effects in carriers of the HLA‐A*03:01 allele. HLA 102 , 707–719 (2023). Sidney, J. et al. Definition of an HLA-A3-like supermotif demonstrates the overlapping peptide-binding repertoires of common HLA molecules. Hum Immunol 45 , 79–93 (1996). Nguyen, A. T., Szeto, C. & Gras, S. The pockets guide to HLA class I molecules. Biochem Soc Trans 49 , 2319–2331 (2021). Horton, R. et al. Variation analysis and gene annotation of eight MHC haplotypes: The MHC Haplotype Project. Immunogenetics 60 , 1–18 (2008). Bertinetto, F. E. et al. The humoral and cellular response to mRNA SARS‐CoV ‐2 vaccine is influenced by HLA polymorphisms. HLA 102 , 301–315 (2023). Crocchiolo, R. et al. Strong humoral response after Covid‐19 vaccination correlates with the common HLA allele A*03:01 and protection from breakthrough infection. HLA 103 , (2024). Xie, J. et al. Relationship between HLA genetic variations, COVID-19 vaccine antibody response, and risk of breakthrough outcomes. Nat Commun 15 , 4031 (2024). Meyer, S. et al. Prevalent and immunodominant CD8 T cell epitopes are conserved in SARS-CoV-2 variants. Cell Rep 42 , 111995 (2023). Szeto, C. et al. Molecular basis of a dominant SARS-CoV-2 spike-derived epitope presented by HLA-A*02:01 recognised by a public TCR. Cells 10 , 2646 (2021). Kared, H. et al. SARS-CoV-2–specific CD8+ T cell responses in convalescent COVID-19 individuals. J Clin Invest 131 , (2021). Shomuradova, A. S. et al. SARS-CoV-2 Epitopes Are Recognized by a Public and Diverse Repertoire of Human T Cell Receptors. Immunity 53 , 1245-1257.e5 (2020). Nguyen, A. T. et al. Homologous peptides derived from influenza A, B and C viruses induce variable CD8+ T cell responses with cross-reactive potential. Clin Transl Immunology 11 , e1422 (2022). Leighton, L. J. et al. The design, manufacture and LNP formulation of mRNA for research use. Nature Protocols 2025 1–30 (2025) doi:10.1038/s41596-025-01174-4. Lu, M. et al. SARS-CoV-2 prefusion spike protein stabilized by six rather than two prolines is more potent for inducing antibodies that neutralize viral variants of concern. Proc Natl Acad Sci U S A 119 , e2110105119 (2022). Bajaña, S., Roach, K., Turner, S., Paul, J. & Kovats, S. IRF4 Promotes Cutaneous Dendritic Cell Migration to Lymph Nodes during Homeostasis and Inflammation. The Journal of Immunology 189 , 3368–3377 (2012). Zhong, S. et al. Single Nucleotide Polymorphisms in the Human Leukocyte Antigen Region Are Associated With Hemagglutination Inhibition Antibody Response to Influenza Vaccine. Front Genet 13 , (2022). Hayney, M. S., Poland, G. A., Jacobson, R. M., Schaid, D. J. & Lipsky, J. J. The influence of the HLA-DRB1*13 allele on measles vaccine response. J Investig Med 44 , 261–3 (1996). Lambert, N. D. et al. Polymorphisms in HLA-DPB1 Are Associated With Differences in Rubella Virus-Specific Humoral Immunity After Vaccination. Journal of Infectious Diseases 211 , 898–905 (2015). Li, Z.-K., Nie, J.-J., Li, J. & Zhuang, H. The effect of HLA on immunological response to hepatitis B vaccine in healthy people: A meta-analysis. Vaccine 31 , 4355–4361 (2013). Santos-Rebouças, C. B. et al. Immune response stability to the SARS-CoV-2 mRNA vaccine booster is influenced by differential splicing of HLA genes. Sci Rep 14 , 8982 (2024). Ozga, A. J. et al. pMHC affinity controls duration of CD8+ T cell-DC interactions and imprints timing of effector differentiation versus expansion. Journal of Experimental Medicine 213 , 2811–2829 (2016). Kranzer, K. et al. Induction of maturation and cytokine release of human dendritic cells by Helicobacter pylori. Infect Immun 72 , 4416–4423 (2004). Tada, Y. et al. Differential effects of LPS and TGF-β on the production of IL-6 and IL-12 by Langerhans cells, splenic dendritic cells, and macrophages. Cytokine 25 , 155–161 (2004). Bryan, A. et al. Performance Characteristics of the Abbott Architect SARS-CoV-2 IgG Assay and Seroprevalence in Boise, Idaho. J Clin Microbiol 58 , (2020). Norman, P. J. et al. Defining KIR and HLA Class I Genotypes at Highest Resolution via High-Throughput Sequencing. Am J Hum Genet 99 , 375–391 (2016). Lineburg, K. E. et al. CD8+ T cells specific for an immunodominant SARS-CoV-2 nucleocapsid epitope cross-react with selective seasonal coronaviruses. Immunity 54 , 1055-1065.e5 (2021). Grant, E. J. & Gras, S. Protocol for generation of human peptide-specific primary CD8+ T cell lines. STAR Protoc 3 , 101590 (2022). Adhikari, A. et al. Longitudinal Characterization of Phagocytic and Neutralization Functions of Anti-Spike Antibodies in Plasma of Patients after Severe Acute Respiratory Syndrome Coronavirus 2 Infection. J Immunol 209 , 1499–1512 (2022). Causality assessment of an adverse event following immunization (AEFI) User manual for the revised WHO classification. Novogene - USA Based Lab Guaranteed Data Security. https://www.novogene.com/us-en/. A basic guide to RNA-sequencing - Novogene. https://www.novogene.com/us-en/resources/blog/a-basic-guide-to-rna-sequencing/. Zhang, Y., Parmigiani, G. & Johnson, W. E. ComBat-seq: batch effect adjustment for RNA-seq count data. NAR Genom Bioinform 2 , (2020). Ewels, P. A. et al. The nf-core framework for community-curated bioinformatics pipelines. Nature Biotechnology 2020 38:3 38 , 276–278 (2020). Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 15 , 1–21 (2014). haplo.stats. Additional Declarations There is NO Competing Interest. Supplementary Files Supp2FiguresA3FIN.pdf Supplementary figures file 2 SupplementarytablesA3FIN.xlsx Supplementary Table 1–23 Supp1FiguresA3FIN.pdf Supplementary figures file 1 Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8282930","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Biological Sciences - Article","associatedPublications":[],"authors":[{"id":561312835,"identity":"23787185-7d06-4675-af9a-6c44eafa1f69","order_by":0,"name":"Jill Hollenbach","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA8UlEQVRIiWNgGAWjYBAC9mbmxgOMbQwyfAwMbED+ATmQ6AcGBmacWhibGRtAWnjYoFqMQYIz8GppQNOS2EBQSztYiw0PG3vvsQc/au6kbzh+/GEDQ4U1SC8eh/1L42HjOZdu2HPsWe6GMzmGDQxn0gloaTvMwyaRYybB23A4d8MNHvYHQBEitMi/MZP823A43eAG+8MGxn+4tQgibOExkwbakmBwg8EQGCa4tUgzA7UktoH8kmNuLHPssOFMkF8SjqUb49LCx3/44IOPbTZy/OxnzB6+qTkszwcKsQ811rK4tIBBAhEio2AUjIJRMApIAQCKUF43bwYbPQAAAABJRU5ErkJggg==","orcid":"https://orcid.org/0000-0001-6572-7962","institution":"University of California San Francisco","correspondingAuthor":true,"prefix":"","firstName":"Jill","middleName":"","lastName":"Hollenbach","suffix":""},{"id":561312836,"identity":"882f4d96-4283-44cb-9f10-f3f50c0ecb73","order_by":1,"name":"Anshika Srivastava","email":"","orcid":"","institution":"University of California San Francisco","correspondingAuthor":false,"prefix":"","firstName":"Anshika","middleName":"","lastName":"Srivastava","suffix":""},{"id":561312837,"identity":"3b2da2a5-b56e-4bb3-843d-0655d013d2e7","order_by":2,"name":"Demetra Chatzileontiadou","email":"","orcid":"https://orcid.org/0000-0003-1132-8952","institution":"Monash University","correspondingAuthor":false,"prefix":"","firstName":"Demetra","middleName":"","lastName":"Chatzileontiadou","suffix":""},{"id":561312838,"identity":"3aa13b3c-ca1d-4989-82a3-238cd725869f","order_by":3,"name":"Anurag Adhikari","email":"","orcid":"","institution":"Immunity and Infection Program, La Trobe Institute for Molecular Science (LIMS), La Trobe University","correspondingAuthor":false,"prefix":"","firstName":"Anurag","middleName":"","lastName":"Adhikari","suffix":""},{"id":561312839,"identity":"941ae474-1d49-497e-a511-5d8dc5002041","order_by":4,"name":"Rayo Suseno","email":"","orcid":"","institution":"University of California San Francisco","correspondingAuthor":false,"prefix":"","firstName":"Rayo","middleName":"","lastName":"Suseno","suffix":""},{"id":561312840,"identity":"ed5f4de3-41b0-4923-bf67-b218b75465eb","order_by":5,"name":"Sean Lin","email":"","orcid":"","institution":"University of California San Francisco","correspondingAuthor":false,"prefix":"","firstName":"Sean","middleName":"","lastName":"Lin","suffix":""},{"id":561312841,"identity":"16b37f84-cda2-47f3-b72c-3d5602305324","order_by":6,"name":"Juliano Boquett","email":"","orcid":"","institution":"University of California San Francisco","correspondingAuthor":false,"prefix":"","firstName":"Juliano","middleName":"","lastName":"Boquett","suffix":""},{"id":561312842,"identity":"4abd19b5-5a7f-476d-86a6-74b73ebf4a88","order_by":7,"name":"Jamie Tuibeo","email":"","orcid":"","institution":"Immunity and Infection Program, La Trobe Institute for Molecular Science (LIMS), La Trobe University,","correspondingAuthor":false,"prefix":"","firstName":"Jamie","middleName":"","lastName":"Tuibeo","suffix":""},{"id":561312843,"identity":"8ec5c8f7-5b87-4f9b-9308-666899aaf778","order_by":8,"name":"Tasneem Yusufali","email":"","orcid":"","institution":"University of California San Francisco","correspondingAuthor":false,"prefix":"","firstName":"Tasneem","middleName":"","lastName":"Yusufali","suffix":""},{"id":561312844,"identity":"2b8bd94f-0122-4de5-ac5f-a245e82ca9af","order_by":9,"name":"Noah Peyser","email":"","orcid":"https://orcid.org/0000-0003-4529-8325","institution":"University of California San Francisco","correspondingAuthor":false,"prefix":"","firstName":"Noah","middleName":"","lastName":"Peyser","suffix":""},{"id":561312845,"identity":"3f9770c7-9211-46da-b159-78ee1c593a0f","order_by":10,"name":"Ticiana Farias","email":"","orcid":"","institution":"University of Colorado School of Medicine,","correspondingAuthor":false,"prefix":"","firstName":"Ticiana","middleName":"","lastName":"Farias","suffix":""},{"id":561312846,"identity":"b2d782b0-2c1d-4cce-94ac-4714da4c1143","order_by":11,"name":"Katherine Kichula","email":"","orcid":"","institution":"University of North Carolina at Charlotte","correspondingAuthor":false,"prefix":"","firstName":"Katherine","middleName":"","lastName":"Kichula","suffix":""},{"id":561312847,"identity":"fecab710-8fc7-42bc-9c97-1102b8b7d663","order_by":12,"name":"Andrea Nguyen","email":"","orcid":"","institution":"2.\tImmunity and Infection Program, La Trobe Institute for Molecular Science (LIMS), La Trobe University,","correspondingAuthor":false,"prefix":"","firstName":"Andrea","middleName":"","lastName":"Nguyen","suffix":""},{"id":561312848,"identity":"84ba4df8-6cac-4c9a-aab5-7f62c2f33836","order_by":13,"name":"Irvin Jose","email":"","orcid":"","institution":"Immunity and Infection Program, La Trobe Institute for Molecular Science (LIMS), La Trobe University,","correspondingAuthor":false,"prefix":"","firstName":"Irvin","middleName":"","lastName":"Jose","suffix":""},{"id":561312849,"identity":"f69bfc80-de67-42fa-a507-9706c251949e","order_by":14,"name":"Dhilshan Jayasinghe","email":"","orcid":"","institution":"Immunity and Infection Program, La Trobe Institute for Molecular Science (LIMS), La Trobe University,","correspondingAuthor":false,"prefix":"","firstName":"Dhilshan","middleName":"","lastName":"Jayasinghe","suffix":""},{"id":561312850,"identity":"ed1420f1-57c4-41a4-97da-720707b4eb05","order_by":15,"name":"Katerina Tarassi","email":"","orcid":"","institution":"Blood Bank Department, “Evangelismos” General Hospital,","correspondingAuthor":false,"prefix":"","firstName":"Katerina","middleName":"","lastName":"Tarassi","suffix":""},{"id":561312851,"identity":"07171ece-1ac6-4ba5-98e1-238ca38e5791","order_by":16,"name":"Elissavet Kontou","email":"","orcid":"","institution":"Immunology-Histocompatibility Department, “Evangelismos” General Hospital","correspondingAuthor":false,"prefix":"","firstName":"Elissavet","middleName":"","lastName":"Kontou","suffix":""},{"id":561312852,"identity":"7bc98ed0-c73f-4237-b3d3-67e635549255","order_by":17,"name":"Janesha Maddumage","email":"","orcid":"","institution":"Immunity and Infection Program, La Trobe Institute for Molecular Science (LIMS), La Trobe University,","correspondingAuthor":false,"prefix":"","firstName":"Janesha","middleName":"","lastName":"Maddumage","suffix":""},{"id":561312853,"identity":"2a283165-e731-4132-9448-b6dd58e4c75d","order_by":18,"name":"Kleio Ampelakiotou","email":"","orcid":"","institution":"Blood Bank Department, “Evangelismos” General Hospital,","correspondingAuthor":false,"prefix":"","firstName":"Kleio","middleName":"","lastName":"Ampelakiotou","suffix":""},{"id":561312854,"identity":"4bb50b05-93b5-4642-a770-e55241cad316","order_by":19,"name":"Alexandra Tsirogianni","email":"","orcid":"","institution":"Blood Bank Department, “Evangelismos” General Hospital,","correspondingAuthor":false,"prefix":"","firstName":"Alexandra","middleName":"","lastName":"Tsirogianni","suffix":""},{"id":561312855,"identity":"a84ae552-95bf-4cca-8a4f-6cd31ba0e6bb","order_by":20,"name":"Michael Dewar-Oldis","email":"","orcid":"","institution":"Immunity and Infection Program, La Trobe Institute for Molecular Science (LIMS), La Trobe University,","correspondingAuthor":false,"prefix":"","firstName":"Michael","middleName":"","lastName":"Dewar-Oldis","suffix":""},{"id":561312856,"identity":"cc71ffd7-4c95-4465-9fc7-55f88ac3e0f8","order_by":21,"name":"Peter Barnard","email":"","orcid":"https://orcid.org/0000-0003-1755-7991","institution":"La Trobe university","correspondingAuthor":false,"prefix":"","firstName":"Peter","middleName":"","lastName":"Barnard","suffix":""},{"id":561312857,"identity":"225f9cea-687b-431b-93d8-8a0607f4e93e","order_by":22,"name":"Joe Sabatino","email":"","orcid":"","institution":"Department of Neurology, University of California,","correspondingAuthor":false,"prefix":"","firstName":"Joe","middleName":"","lastName":"Sabatino","suffix":""},{"id":561312858,"identity":"b91e5d8d-147f-4c13-a4dd-225cd3efbf93","order_by":23,"name":"Dimitrios Zoulas","email":"","orcid":"","institution":"Immunity and Infection Program, La Trobe Institute for Molecular Science (LIMS), La Trobe University,","correspondingAuthor":false,"prefix":"","firstName":"Dimitrios","middleName":"","lastName":"Zoulas","suffix":""},{"id":561312859,"identity":"adc6fc3e-15cb-4656-b43e-8c1bf013a5bc","order_by":24,"name":"Emily Ariens","email":"","orcid":"","institution":"BASE mRNA Facility, Australian Institute for Bioengineering and Nanotechnology (AIBN), University of Queensland,","correspondingAuthor":false,"prefix":"","firstName":"Emily","middleName":"","lastName":"Ariens","suffix":""},{"id":561312860,"identity":"470b5987-0ecd-4071-98bf-6281d0a8f281","order_by":25,"name":"Timothy Mercer","email":"","orcid":"https://orcid.org/0000-0001-8780-894X","institution":"University of Queensland","correspondingAuthor":false,"prefix":"","firstName":"Timothy","middleName":"","lastName":"Mercer","suffix":""},{"id":561312861,"identity":"375c2790-3675-4212-97f6-4939e41fe273","order_by":26,"name":"Emma Grant","email":"","orcid":"","institution":"Immunity and Infection Program, La Trobe Institute for Molecular Science (LIMS), La Trobe University,","correspondingAuthor":false,"prefix":"","firstName":"Emma","middleName":"","lastName":"Grant","suffix":""},{"id":561312862,"identity":"571ce52a-92de-459a-8a63-bc87798a80f8","order_by":27,"name":"Lloyd D'Orsogna","email":"","orcid":"https://orcid.org/0000-0002-6824-4189","institution":"University of Western Australia","correspondingAuthor":false,"prefix":"","firstName":"Lloyd","middleName":"","lastName":"D'Orsogna","suffix":""},{"id":561312863,"identity":"c003868f-5986-4abe-b3bc-8dadeca02a57","order_by":28,"name":"Corey Smith","email":"","orcid":"https://orcid.org/0000-0002-7550-9595","institution":"QIMR Berghofer Medical Research Institute","correspondingAuthor":false,"prefix":"","firstName":"Corey","middleName":"","lastName":"Smith","suffix":""},{"id":561312864,"identity":"f63618e3-1e0d-43d8-a1f6-98369068e5d5","order_by":29,"name":"Paul Norman","email":"","orcid":"","institution":"University of Colorado, Denver","correspondingAuthor":false,"prefix":"","firstName":"Paul","middleName":"","lastName":"Norman","suffix":""},{"id":561312865,"identity":"7dbfd128-59ef-4910-8e4a-d3f9d0670536","order_by":30,"name":"Gregory Marcus","email":"","orcid":"https://orcid.org/0000-0001-5197-7696","institution":"Division of Cardiology, University of California, San Francisco","correspondingAuthor":false,"prefix":"","firstName":"Gregory","middleName":"","lastName":"Marcus","suffix":""},{"id":561312866,"identity":"2608be93-de61-40bd-b472-48d7a33d77f4","order_by":31,"name":"Jeffrey Olgin","email":"","orcid":"","institution":"UCSF","correspondingAuthor":false,"prefix":"","firstName":"Jeffrey","middleName":"","lastName":"Olgin","suffix":""},{"id":561312867,"identity":"8f6a574f-254b-4e05-8927-6e4ae2631919","order_by":32,"name":"Mark J. Pletcher","email":"","orcid":"https://orcid.org/0000-0002-6966-1312","institution":"UCSF","correspondingAuthor":false,"prefix":"","firstName":"Mark","middleName":"J.","lastName":"Pletcher","suffix":""},{"id":561312868,"identity":"897c6617-43e8-4485-9f73-7003205d262d","order_by":33,"name":"Martin Maiers","email":"","orcid":"","institution":"National Marrow Donor Program","correspondingAuthor":false,"prefix":"","firstName":"Martin","middleName":"","lastName":"Maiers","suffix":""},{"id":561312869,"identity":"cb1231c6-3129-468e-841d-18163f0f0fe6","order_by":34,"name":"Stephanie Gras","email":"","orcid":"https://orcid.org/0000-0001-7416-038X","institution":"La Trobe University","correspondingAuthor":false,"prefix":"","firstName":"Stephanie","middleName":"","lastName":"Gras","suffix":""}],"badges":[],"createdAt":"2025-12-04 23:40:27","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8282930/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8282930/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":98390751,"identity":"362d56e8-d46b-409d-9d23-879f0c731b81","added_by":"auto","created_at":"2025-12-17 09:27:33","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":247207,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAssociation of \u003c/strong\u003e\u003cem\u003e\u003cstrong\u003eHLA\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e alleles with three or more systemic side effects in the discovery cohort (European ancestry, N = 50,535).\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eOdds ratio is shown on the x-axis and p-value is shown on the y-axis. Alleles that are significant in both discovery and replication cohort are shown as solid-colored circles.\u003c/p\u003e","description":"","filename":"FiguresSrivastavaFIN1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8282930/v1/6f4128cc99b97862febe50a6.jpg"},{"id":98390752,"identity":"fda4162d-d1cb-4376-8764-e5293886b0fd","added_by":"auto","created_at":"2025-12-17 09:27:33","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":775301,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSpike-specific CD8⁺ T cells are present before vaccination and expand into a high-MFI effector population after vaccination.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eA\u003c/strong\u003e) Tetramer-associated magnetic enrichment (TAME) plots of HLA-A*03:01-KCY-specific CD8⁺ T cells across before (V0) and after vaccination (V1, V2). (\u003cstrong\u003eB\u003c/strong\u003e) Quantification of KCY⁺ CD8⁺ T cell frequencies by TAME. (\u003cstrong\u003eC-D\u003c/strong\u003e) Mean fluorescence intensity (MFI) analysis of HLA-A*03:01-KCY-specific CD8⁺ T cells with (\u003cstrong\u003eC\u003c/strong\u003e) low- and (\u003cstrong\u003eD\u003c/strong\u003e) high-MFI at the three time point V0, V1 and V2. (\u003cstrong\u003eE\u003c/strong\u003e) Double-tetramer staining of HLA-A*03:01-KCY-specific CD8\u003csup\u003e+\u003c/sup\u003e T cells with HLA-A*03:01-Flu-NP\u003csub\u003e265\u003c/sub\u003e. (\u003cstrong\u003eF-G\u003c/strong\u003e) Phenotypic composition of (\u003cstrong\u003eF\u003c/strong\u003e) low- and (\u003cstrong\u003eG\u003c/strong\u003e) high-MFI HLA-A*03:01-KCY-specific CD8⁺ T cells showing the proportion of naïve (red), effector memory (T\u003csub\u003eEM\u003c/sub\u003e, orange), central memory (T\u003csub\u003eCM\u003c/sub\u003e, black), stem like memory cells (T\u003csub\u003eSCM\u003c/sub\u003e, grey), and terminally exhausted memory re-expressing RA cells (T\u003csub\u003eEMRA\u003c/sub\u003e, blue) phenotype. (\u003cstrong\u003eH-I\u003c/strong\u003e) IFNγ⁺ CD8⁺ T cell responses with cell lines generated with the Spike-derived peptide pool and restimulated with either the Spike-derived peptide pool (\u003cstrong\u003eH\u003c/strong\u003e) or with the HLA-A*03:01-restricted KCY peptide (\u003cstrong\u003eI\u003c/strong\u003e) before vaccination (V0) and after the first (V1) and second (V2) vaccine doses. (\u003cstrong\u003eJ\u003c/strong\u003e) Paired comparison of KCY- and influenza NP\u003csub\u003e265\u003c/sub\u003e-specific responses showing IFNγ production by CD8⁺ T cells. Bars represent mean ± SD; symbols denote individual donors. V0, pre-vaccination; V1, post-dose 1; V2, post-dose 2; V5, \u0026gt; 400 days post-dose 3. *p \u0026lt; 0.05; **p \u0026lt; 0.01.\u003c/p\u003e","description":"","filename":"FiguresSrivastavaFIN2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8282930/v1/b23cf50f27e61762c183c142.jpg"},{"id":98441528,"identity":"8ceef538-721d-4304-9602-fae15cdac07b","added_by":"auto","created_at":"2025-12-17 17:05:34","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":564882,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSpike stimulation activates a large IL-6 and IL-8 production\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCytokine levels measured in PBMCs from HLA-A*03:01⁺ (n = 19, red dots) and HLA-A*03:01\u003csup\u003e-\u003c/sup\u003e (n = 16, black dots) samples after stimulation with full-length soluble Spike protein, Spike peptide pool, or lipid nanoparticle (LNP) control. Only full-length Spike induced broad cytokine responses, including IL-6, IL-8, IP-10, GM-CSF, TNF, MCP-1, IL-1β, IL-10, MIP-1α, RANTES, IL-1α and IFNγ. Bars represent mean ± SD; symbols denote individual donors.\u003c/p\u003e","description":"","filename":"FiguresSrivastavaFIN3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8282930/v1/9940f85d8e58750fd745e5fc.jpg"},{"id":98441209,"identity":"c89faf0c-ed77-409f-b261-544a1f7cc3da","added_by":"auto","created_at":"2025-12-17 17:05:04","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":402759,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eIL-6 and IL-8 cytokines are mainly produced by monocytes and correlates with vaccine side-effect severity for HLA-A*03:01⁺ donors\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eA-B\u003c/strong\u003e) Intracellular cytokine staining (ICS) of Spike-stimulated PBMCs showing IL-6⁺ (\u003cstrong\u003eA\u003c/strong\u003e) and IL-8⁺ (\u003cstrong\u003eB\u003c/strong\u003e) cells among CD3⁺ T cells, CD19⁺ B cells, CD56⁺ NK cells, and CD14⁺ monocytes from HLA-A*03:01⁺\u003cem\u003e \u003c/em\u003e(red dots) and HLA-A*03:01\u003csup\u003e-\u003c/sup\u003e samples (black dots). (\u003cstrong\u003eC\u003c/strong\u003e) Phagocytic score (P score) representing the uptake of Spike-coated microbeads by monocytes, NK cells and T cells in HLA-A*03:01⁺\u003cem\u003e \u003c/em\u003e(red dots) and HLA-A*03:01\u003csup\u003e-\u003c/sup\u003e samples (black dots). Bars represent mean ± SD; symbols denote individual donors. (\u003cstrong\u003eD-G\u003c/strong\u003e) Correlation of IL-8 (\u003cstrong\u003eD\u003c/strong\u003e,\u003cstrong\u003e F\u003c/strong\u003e) and IL-6 (\u003cstrong\u003eE\u003c/strong\u003e,\u003cstrong\u003e G\u003c/strong\u003e) concentrations with self-reported vaccine side-effect severity scores from HLA-A*03:01⁺ (red dots) and HLA-A*03:01\u003csup\u003e-\u003c/sup\u003e donors (black dots).\u003c/p\u003e","description":"","filename":"FiguresSrivastavaFIN4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8282930/v1/197223665c6ada3b64c81213.jpg"},{"id":104400179,"identity":"e52ec924-2b68-4596-b791-099251e1bdc4","added_by":"auto","created_at":"2026-03-11 12:09:07","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":4639759,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8282930/v1/b9e9345f-3a24-4b3b-ad8a-6e3c28e9f31f.pdf"},{"id":98390755,"identity":"30450e69-0dbd-4b22-a431-15bcaa8885df","added_by":"auto","created_at":"2025-12-17 09:27:34","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":728110,"visible":true,"origin":"","legend":"Supplementary figures file 2","description":"","filename":"Supp2FiguresA3FIN.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8282930/v1/108c4402645436e34c15a9b2.pdf"},{"id":98390754,"identity":"c2a8c096-36b6-44cd-a785-7cbf94b1e184","added_by":"auto","created_at":"2025-12-17 09:27:34","extension":"xlsx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":709445,"visible":true,"origin":"","legend":"Supplementary Table 1\u0026#x2013;23","description":"","filename":"SupplementarytablesA3FIN.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-8282930/v1/2f04f0bec6cb7f7108525fe2.xlsx"},{"id":98390757,"identity":"016df88b-5071-4387-a2cc-c9967fc7dec8","added_by":"auto","created_at":"2025-12-17 09:27:34","extension":"pdf","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":2411702,"visible":true,"origin":"","legend":"Supplementary figures file 1","description":"","filename":"Supp1FiguresA3FIN.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8282930/v1/73d7c20e6880a2b80e202ba2.pdf"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"An HLA Association With COVID-19 Vaccine Reactogenicity Correlates With \r\nFewer SARS-CoV-2 Infections and Monocyte Activation","fulltext":[{"header":"Introduction","content":"\u003cp\u003eCOVID-19 vaccines are an important public health tool in preventing severe illness, hospitalization, and mortality due to infection with the virus\u003csup\u003e1–4\u003c/sup\u003e. In early studies examining the effectiveness of these vaccines, mRNA vaccines BNT162b2 (referred to hereafter by the brand name “Pfizer\") and mRNA-1273 (referred to hereafter by the brand name “Moderna”) were shown to be 95% and 94.1% effective against symptomatic infection, respectively\u003csup\u003e5\u003c/sup\u003e. Adenovirus-based vaccines, such as\u0026nbsp;ChAdOx1-S\u0026nbsp;(brand name “Johnson \u0026amp; Johnson”, hereafter “J\u0026amp;J) and\u0026nbsp;Ad26.COV2.S\u0026nbsp;(brand name “AstraZeneca”), in contrast, showed 70% and 66% efficacy, respectively\u003csup\u003e6,7\u003c/sup\u003e.\u0026nbsp;Thus, although mostly efficacious in preventing serious illness and hospitalization, ‘breakthrough’ infections (BTI) occur with all current COVID-19 vaccines. These infections may be attributed to primary vaccine failure, secondary vaccine failure, individuals’ age, immune evasion by novel viral variant, or waning vaccine efficacy over time\u003csup\u003e8–15\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003ePrior work has indicated that the immunogenicity of the COVID-19 vaccines can vary by individual and is correlated with their efficacy. For example, studies have reported different immunological responses to vaccines linked to age, sex, body mass index, nutritional status, and the composition of the gut microbiome\u003csup\u003e16,17\u003c/sup\u003e.\u0026nbsp;Suggesting an immunogenetic feature of vaccine response, several studies have linked variation in the human leukocyte antigen (\u003cem\u003eHLA\u003c/em\u003e) region with antibody levels and T-cell response after vaccination\u003csup\u003e18–21\u003c/sup\u003e. \u003cem\u003eHLA\u003c/em\u003e is the most polymorphic region (6p21) of the human genome with thousands of known alleles. Variation in \u003cem\u003eHLA\u003c/em\u003e is long recognized to play a role in viral illness\u003csup\u003e22\u003c/sup\u003e. Demonstrating the importance of HLA antigen presentation in the immune response to SARS-CoV-2, we have previously shown that \u003cem\u003eHLA\u003c/em\u003e variation is associated with an asymptomatic disease course\u003csup\u003e23\u003c/sup\u003e and provided a functional and structural basis to explain the association. Likewise, numerous other \u003cem\u003eHLA\u003c/em\u003e associations with the COVID-19 disease course have been identified\u003csup\u003e24\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWhile overwhelmingly safe, vaccination-induced immune activation can lead to side effects. Most adverse reactions are mild, including fever, muscle aches, headaches, and fatigue, as well as local reactions such as pain, redness, and swelling at the injection site\u003csup\u003e25\u003c/sup\u003e. These reactions typically appear within a few hours after vaccination and are short-lived, usually resolving within one to two days\u003csup\u003e26\u003c/sup\u003e. While bothersome, there is some evidence that side effects may be associated with improved vaccine efficacy. A study involving participants receiving the human papillomavirus vaccine, for example, found that the occurrence of inflammation-related adverse reactions is associated with concentrations of antibodies, suggesting that individuals who have vaccine-induced side effects have a more robust immune response\u003csup\u003e27,28\u003c/sup\u003e. Importantly, systemic side effects from COVID-19\u0026nbsp;vaccination, such as fever and fatigue, have also been associated with enhanced humoral and cellular immune responses\u003csup\u003e29\u003c/sup\u003e. Intriguingly, previous work has shown a specific \u003cem\u003eHLA\u0026nbsp;\u003c/em\u003eclass I allele, \u003cem\u003eHLA-A*03:01\u003c/em\u003e to be associated with increased side effects after COVID-19 vaccination as well as increased antibody response\u003csup\u003e19,30,31\u003c/sup\u003e. Additionally, enhancement of T cell memory from COVID-19 mRNA booster doses was shown to be particularly pronounced in \u003cem\u003eHLA-A*03:01\u003csup\u003e+\u003c/sup\u003e\u003c/em\u003e COVID-19 recovered patients\u003csup\u003e20\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eNevertheless, despite strong evidence for the role of HLA in response to COVID-19 vaccination, there remains substantial knowledge gaps regarding the complex interplay between \u003cem\u003eHLA\u003c/em\u003e genetic variation, HLA structural variation and antigen presentation, vaccine side effects, cellular and humoral immunity, and vaccine effectiveness.\u0026nbsp;In the present study, we close these gaps by considering \u003cem\u003eHLA\u003c/em\u003e variation and vaccine reactogenicity in a large cohort of over 50,000 vaccinated subjects and provide a global framework for our findings by examining differential gene expression, antibody response, T cell reactivity, and innate immunity. We demonstrate that \u003cem\u003eHLA-A*03:01\u003c/em\u003e-associated reactogenicity is associated with fewer total infections and is not driven by T cell activation, despite a large pool of mostly naïve Spike-specific T cells, but rather by monocyte-derived cytokine production, revealing an innate immune mechanism underlying HLA-linked COVID-19 vaccine side effects.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003e\u003cem\u003eParticipant recruitment and baseline demographics\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBetween January 2023 and January 2024, we collected responses to survey questions regarding respondents’ general health history, including experiences with COVID-19 and associated vaccinations, from potential bone marrow donors registered with the NMDP (formerly National Marrow Donor Program/Be The Match) \u0026nbsp;and for whom high-resolution \u003cem\u003eHLA\u003c/em\u003e genotyping data were available in the NMDP database. The specific survey questions related to COVID-19 and vaccinations, including side effects queried, are given in \u003cstrong\u003eSupplementary Table 1\u003c/strong\u003e. Because this was a U.S.-based cohort, respondents received only vaccines approved for use in the country (Pfizer, Moderna, J\u0026amp;J). Among the 80,007 total respondents, 50,535 self-identified as White/European ancestry (\u003cstrong\u003eSupplementary Table 2a\u003c/strong\u003e) and reported having completed at a minimum the initial series of vaccination for SARS-CoV-2 (one dose for J\u0026amp;J, two for Pfizer/Moderna mRNA); these subjects constituted our discovery cohort. An additional 7403 respondents reported completion of the initial series self-identified with other ancestries (\u003cstrong\u003eSupplementary Table 2b\u003c/strong\u003e).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFor our replication cohort\u003cstrong\u003e,\u0026nbsp;\u003c/strong\u003ewe collected responsesbetween July 2020 and April 2022 via a mobile phone app with follow-up daily questions specific to vaccine side effects, as described in Augusto et al., 2023\u003csup\u003e23\u003c/sup\u003e. Among 10,595 respondents who reported completion of the initial vaccination series, 4,575 self-identified as White/European ancestry (\u003cstrong\u003eSupplementary\u003c/strong\u003e \u003cstrong\u003eTable 3\u003c/strong\u003e). While these respondents are also NMDP donors with available high resolution \u003cem\u003eHLA\u003c/em\u003e data, there was no overlap with individuals in the discovery cohort.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eSystemic side effects to COVID-19 vaccines co-occur and are associated with HLA-A*03:01\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo understand the distribution of side effects to COVID-19 vaccines in our discovery cohort, we first calculated the covariation matrix for all reported side effects (\u003cstrong\u003eSupplementary\u003c/strong\u003e \u003cstrong\u003eFigure 1\u003c/strong\u003e). Here, we only considered side effects reported after completion of the initial vaccination series. We found that systemic side effects (SSE) like fever or chills, muscle or body fatigue, or headaches, rather than localized side effects like runny nose (\u003cstrong\u003eSupplementary Table 4\u003c/strong\u003e), showed substantial co-occurrence, with a median number of two SSE reported per individual.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo evaluate whether HLA variation plays a role in increasing side effects in vaccination for SARS-CoV-2 and to capture cases with the greatest burden of symptoms, we first considered individuals who reported greater than the median number of SSE.\u0026nbsp;Using a dominant model, we used multivariate logistic regression to test for association with the occurrence of three or four reported SSE for each \u003cem\u003eHLA\u003c/em\u003e allele at each classical class I (\u003cem\u003eHLA-A, -B, -C\u003c/em\u003e) and two \u003cem\u003eHLA\u003c/em\u003e class II (\u003cem\u003eHLA-DRB1\u003c/em\u003e and \u003cem\u003e-DQB1\u003c/em\u003e) loci observed at a frequency \u0026gt;3% in our cohort, adjusting for sex and age (\u003cstrong\u003eFigure 1,\u003c/strong\u003e \u003cstrong\u003eSupplementary Table 5\u003c/strong\u003e). This revealed as the top candidate a strong and significant association of \u003cem\u003eHLA-A*03:01\u003c/em\u003e with SSE (OR = 1.36, CI= 1.31 – 1.41, p = 6.79×10\u003csup\u003e-57\u003c/sup\u003e). We did not observe any substantial dose effect for \u003cem\u003eHLA-A*03:01\u003c/em\u003e (homozygous, OR = 1.47, CI = 1.29 – 1.68, p = 1.65×10\u003csup\u003e-8\u003c/sup\u003e; heterozygous, OR = 1.40, CI = 1.34 – 1.46, p = 3.15 ×10\u003csup\u003e-53\u003c/sup\u003e), confirming the dominant model. Because \u003cem\u003eHLA-A*03:01\u003c/em\u003e belongs to the HLA-A3 supertype group\u003csup\u003e32\u003c/sup\u003e that also includes \u003cem\u003eHLA-A*11:01, -A*31:01, -A*33:01,\u003c/em\u003e and -\u003cem\u003eA*68:01\u003c/em\u003e, and is characterized by shared peptide binding\u003csup\u003e33\u003c/sup\u003e, we hypothesized that this supertype might show shared associations with SSE across alleles. However, we found that among the HLA-A3 supertype alleles, only \u003cem\u003eHLA-A*03:01\u003c/em\u003e was significant for the association with increased SSE (\u003cstrong\u003eSupplementary Table 5\u003c/strong\u003e). This association of increased SSE with \u003cem\u003eHLA-A*03:01\u003c/em\u003e clearly replicated in our additional cohort of 4,575 vaccinees with European ancestry, with a remarkably consistent effect size (OR = 1.46, CI = 1.21 – 1.77, p = 7.77 x 10\u003csup\u003e-5\u003c/sup\u003e) relative to that in our discovery cohort (\u003cstrong\u003eSupplementary Table 6)\u003c/strong\u003e. In addition, we found that \u003cem\u003eHLA-A*03:01\u003c/em\u003e was significantly associated with increased SSE reports in our self-identified Hispanic cohort (N = 4,287), again with extremely consistent effect size (OR = 1.45, CI = 1.25 – 1.69, p = 7.31× 10\u003csup\u003e-7\u003c/sup\u003e, \u003cstrong\u003eSupplementary Table 7\u003c/strong\u003e). While a similar trend was clearly observed, this association did not reach statistical significance in cohorts of other ancestries, where our sample sizes were much smaller (\u003cstrong\u003eSupplementary Tables 8\u0026nbsp;\u003c/strong\u003eand\u003cstrong\u003e\u0026nbsp;9\u003c/strong\u003e).\u003c/p\u003e\n\u003cp\u003eTo better understand whether any specific reported side effect was driving this HLA association, we considered the association with\u0026nbsp;\u003cem\u003eHLA-A*03:01\u003c/em\u003e for each side effect separately (\u003cstrong\u003eSupplementary Table 10\u003c/strong\u003e).\u0026nbsp;As expected, we observed a strong and highly significant negative association of \u003cem\u003eHLA-A*03:01\u003c/em\u003e with reporting “no vaccine side effects” (OR = 0.73, CI = 0.70 – 0.77, p = 5.82×10\u003csup\u003e-43\u003c/sup\u003e), demonstrating that individuals with this allele are less likely to report having experienced no side effects after vaccination. Analysis of reports of specific side effects revealed the strongest association of \u003cem\u003eHLA-A*03:01\u003c/em\u003e with “fever or chills” (OR = 1.43, CI = 1.38 – 1.49, p = 1.37×10\u003csup\u003e-81\u003c/sup\u003e), followed by “muscle or body aches” (OR = 1.305, CI = 1.25 – 1.35, p = 2.16 ×10\u003csup\u003e-45\u003c/sup\u003e), “fatigue” (OR = 1.25, CI = 1.20 – 1.30, p = 5.57×10\u003csup\u003e-34\u003c/sup\u003e), and “headaches” (OR = 1.23, CI = 1.19 – 1.29, p = 1.00×10\u003csup\u003e-24\u003c/sup\u003e).\u0026nbsp;We also found remarkably consistent results for our cohort of individuals who self-identify as Hispanic (\u003cstrong\u003eSupplementary Table 10\u003c/strong\u003e,\u003cstrong\u003e\u0026nbsp;Supplementary Figure 2\u003c/strong\u003e).Only “fatigue” was found to have a significant association with\u0026nbsp;\u003cem\u003eHLA-A*03:01\u003c/em\u003e in our African American cohort\u0026nbsp;(OR = 1.49, CI = 1.07 – 2.09, p = 0.018); however, as noted previously, smaller sample sizes limited our power to detect associations in some cohorts (\u003cstrong\u003eSupplementary Table 8\u0026nbsp;\u003c/strong\u003eand\u003cstrong\u003e\u0026nbsp;9\u003c/strong\u003e).\u0026nbsp;Finally, we did not find a substantive difference in effect size when considering individuals who had reported infection prior to vaccination or breakthrough infection (BTI)(\u003cstrong\u003eSupplementary Table 11\u003c/strong\u003e).\u003c/p\u003e\n\u003cp\u003eIn summary, we find a highly significant, consistent association of \u003cem\u003eHLA-A*03:01\u003c/em\u003e with reported side effects post-COVID-19 vaccination. This association replicated across multiple independent cohorts and ancestries, supporting a role for \u003cem\u003eHLA\u003c/em\u003e variation in driving vaccination side effect.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eEffect size of HLA-A*03:01 varies with vaccine brand\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWhile the median number of SSE reported across the entire discovery cohort was two, the proportion of individuals who experienced three or four SSE was higher in Moderna than in Pfizer or Johnson \u0026amp; Johnson vaccinated individuals (\u003cstrong\u003eSupplementary Figure 3\u003c/strong\u003e). Because the frequency of SSE varied by vaccine manufacturer, we stratified our cohorts according to whether they received the Pfizer, Moderna, or J\u0026amp;J vaccine. To reduce confounding, we only considered individuals who received both doses of the initial series (in the case of mRNA vaccines) from the same manufacturer.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe found that Pfizer-vaccinated individuals showed the largest effect size for all associations of \u003cem\u003eHLA-A*03:01\u003c/em\u003e with side effects relative to other vaccine brands, particularly fever or chills (OR = 1.71, CI = 1.61 – 1.79, p = 4.51×10\u003csup\u003e-88\u003c/sup\u003e). Likewise, the chance of not experiencing any side effects is significantly lower in Pfizer (OR = 0.676, CI = 0.64 – 0.71, p = 2.83×10\u003csup\u003e-40\u003c/sup\u003e) and Moderna recipients (OR = 0.782, CI = 0.72 – 0.85, p =1.59×10\u003csup\u003e-9\u003c/sup\u003e) (\u003cstrong\u003eSupplementary\u003c/strong\u003e \u003cstrong\u003eTable 12\u003c/strong\u003e). We did not observe significant associations of \u003cem\u003eHLA-A*03:01\u003c/em\u003e with side effects among J\u0026amp;J vaccine recipients except for fatigue (OR = 1.259, CI = 1.09 – 1.45, p = 1.40×10\u003csup\u003e-4\u003c/sup\u003e), possibly owing to the small sample size for this brand (N = 3,312). We observed similar results in our replication cohort, where the association of \u003cem\u003eHLA-A*03:01\u003c/em\u003e showed similar patterns of effect size between brands, although many individual side effects did not reach statistical significance among Moderna recipients, likely due to more limited power in this cohort\u0026nbsp;(\u003cstrong\u003eSupplementary Table 13\u003c/strong\u003e).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThus, while we find that\u0026nbsp;\u003cem\u003eHLA-A*03:01\u003c/em\u003e is associated with side effects across COVID-19 vaccine brands, this effect is much more pronounced and consistent among Pfizer recipients.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eAdditional HLA class I alleles are associated with systemic side effects in COVID-19 vaccination\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn addition to \u003cem\u003eHLA-A*03:01\u003c/em\u003e, we observed numerous other \u003cem\u003eHLA\u003c/em\u003e alleles that were significantly associated with reports of systemic side effects in our discovery cohort (\u003cstrong\u003eFigure 1\u003c/strong\u003e,\u003cstrong\u003e\u0026nbsp;Supplementary Table 5\u003c/strong\u003e). Of these, \u003cem\u003eHLA-A*29:02\u003c/em\u003e (OR = 1.28, CI = 1.19 – 1.36, p = 2.10×10\u003csup\u003e-11\u003c/sup\u003e), \u003cem\u003eHLA-B*08:01\u003c/em\u003e (OR = 0.76, CI = 0.73 – 0.80, p = 5.65×10\u003csup\u003e-32\u003c/sup\u003e), \u003cem\u003eHLA-C*07:01\u0026nbsp;\u003c/em\u003e(OR = 0.82, CI = 0.79 – 0.85, p = 1.10×10\u003csup\u003e-24\u003c/sup\u003e), and \u003cem\u003eHLA\u003c/em\u003e-\u003cem\u003eDRB1*03:01\u0026nbsp;\u003c/em\u003e(OR = 0.85, CI = 0.82-0.90, p = 1.23×10\u003csup\u003e-12\u003c/sup\u003e) replicated (\u003cstrong\u003eSupplementary Table 6\u003c/strong\u003e). \u003cem\u003eHLA-B*08:01, HLA-C*07:01,\u003c/em\u003e and \u003cem\u003eHLA-DRB1*03:01\u003c/em\u003e are components of the well-documented “ancestral 8.1 haplotype” (AH8.1)\u003csup\u003e34\u003c/sup\u003e, which is also evident from the high linkage disequilibrium (LD) values between these alleles (\u003cstrong\u003eSupplementary Table\u003c/strong\u003e \u003cstrong\u003e14\u003c/strong\u003e). Owing to the likelihood that these three alleles reflected a single primary association through LD, we performed conditional analyses to identify the primary associated allele in this haplotype. We found that \u003cem\u003eHLA-B*08:01\u003c/em\u003e had the strongest effect size (OR = 0.77, CI = 0.71 – 0.84, p = 2.71×10\u003csup\u003e-9\u003c/sup\u003e) after controlling for \u003cem\u003eHLA-C*07:01,\u003c/em\u003e and \u003cem\u003eHLA-DRB1*03:01\u003c/em\u003e, suggesting that this allele is responsible for the protective effect against SSE. Likely owing to the small sample size, we did not detect these associations in our cohorts with other ancestries (\u003cstrong\u003eSupplementary Tables 7 – 9\u003c/strong\u003e).\u003c/p\u003e\n\u003cp\u003eIn summary, while\u0026nbsp;\u003cem\u003eHLA-A*03:01\u0026nbsp;\u003c/em\u003edemonstrated the strongest and most significant effect with respect to vaccine SSE, we find evidence for additional HLA class I involvement for both risk (\u003cem\u003eHLA-A*29:02\u003c/em\u003e) and protection (\u003cem\u003eHLA-B*08:01\u003c/em\u003e).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eHLA-A*03:01 carriage is associated with higher antibody levels after COVID-19 vaccination\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe next sought to confirm prior reports of association of\u0026nbsp;\u003cem\u003eHLA-A*03:01\u003c/em\u003e with\u0026nbsp;higher levels of antibody upon vaccination for SARS-CoV-2\u003csup\u003e19,35,36\u0026nbsp;\u003c/sup\u003ein an independent cohort of 156 healthcare workers receiving two doses of the Pfizer vaccine. The cohort was stratified in three groups according to the levels of anti-Spike IgG antibodies after the second dose of the vaccine as follows, Group I: 1,000 – 4,000 AU/mL (N = 50), Group II: 4,001 – 20,000 AU/mL (N = 53), Group III: ≥ 20,000 AU/mL (N = 53). To investigate the association between \u003cem\u003eHLA-A*03:01\u0026nbsp;\u003c/em\u003eand anti-Spike antibody levels, we performed logistic regression comparing Group III against Groups I and II. Our results confirmed that \u003cem\u003eHLA-A*03:01\u003c/em\u003e is associated with higher antibody levels post vaccination (OR = 3.25, CI = 1.31 – 8.33, p = 0.0117). We also observed that antibody positivity increased with \u003cem\u003eHLA-A*03:01\u003c/em\u003e allele dosage (0, 1, or 2 copies; \u003cstrong\u003eSupplementary Figure 4\u003c/strong\u003e). Thus, we confirm that in addition to its role in mediating vaccine SSE, individuals positive for \u003cem\u003eHLA-A*03:01\u0026nbsp;\u003c/em\u003edisplay a higher anti-Spike antibody response post-vaccine, providing further evidence that vaccine side effects are reflective of a more robust immune response to vaccination and that this has an immunogenetic underpinning.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eHLA-A*03:01 carriers report fewer breakthrough infections, fewer total infections, and milder disease course\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eOwing to the highly significant association of \u003cem\u003eHLA-A*03:01\u003c/em\u003e with vaccine side effects and increased antibody levels after vaccination, we next asked whether this allele might be associated with reduced breakthrough infection (BTI). Of the total vaccinated 57,938 individuals across ancestries, we excluded 10,227 individuals who did not have information for their 2\u003csup\u003end\u003c/sup\u003e dose timing. Among the remaining participants who reported specific dates for both vaccination and infection, we found \u003cem\u003eHLA-A*03:01\u003c/em\u003e to be associated with a decreased risk of BTI, albeit with modest effect size (OR = 0.916, CI = 0.88 – 0.95, p = 1.61×10\u003csup\u003e-5,\u003c/sup\u003e); this association was also significant with \u0026nbsp;stronger effect size in our cohort with Hispanic ancestry\u003cstrong\u003e\u0026nbsp;(\u003c/strong\u003eOR = 0.776, CI = 0.66 – 0.91, p = \u0026nbsp;1.44×10\u003csup\u003e-4\u003c/sup\u003e), and while not reaching statistical significance, we observed similar effect sizes in other ancestries(\u003cstrong\u003eSupplementary Table 15\u003c/strong\u003e). \u0026nbsp;We did not observe any other significant HLA associations with BTI in our cohort, including previously reported associations of \u003cem\u003eHLA-DQB1*06\u003c/em\u003e\u003csup\u003e18,37\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe observed association of \u003cem\u003eHLA-A*03:01\u003c/em\u003e with reduction in BTI is also reflected in an overall decrease in total infections reported. We dichotomized participants according to those reporting having never been infected or having only a single infection, (about 86% of our discovery cohort, N= 43,527), and those reporting repeated reinfections (i.e. two or more infections, about 14% of our discovery cohort, N = 7008) between 2019 and 2023. We found that vaccinated individuals positive for \u003cem\u003eHLA-A*03:01\u003c/em\u003e were less likely to have experienced repeated SARS-CoV-2 infections\u0026nbsp;(OR = 0.92, CI = 0.87 – 0.96, p = 0.00177). Finally, we asked whether among those who did report infection, \u003cem\u003eHLA-A*03:01\u003c/em\u003e is associated with milder disease course. We found that individuals positive for \u003cem\u003eHLA-A*03:01\u003c/em\u003e were less likely to report higher than the median number (three) of symptoms (considering all reported symptoms;\u0026nbsp;OR = 0.90, CI = 0.88 – 0.94, p = 2.78\u0026nbsp;× 10\u003csup\u003e-7\u003c/sup\u003e)\u0026nbsp;suggesting a milder COVID-19 disease course.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo understand whether the \u003cem\u003eHLA-A*03:01\u003c/em\u003e association with reduced BTI was secondary to its association with side effects, we considered the occurrence of vaccination side effects and the likelihood of BTI irrespective of \u003cem\u003eHLA\u003c/em\u003e genotype. We found that\u0026nbsp;each systemic side effect was independently inversely associated with the likelihood of a BTI (\u003cstrong\u003eSupplementary Table 16\u003c/strong\u003e). Considering the combined measure of three or four SSE and BTI (and adjusting for sex, age, and vaccine brand), we found a highly significant protective effect with an effect size greater than that observed for \u003cem\u003eHLA-A*03:01\u0026nbsp;\u003c/em\u003e(OR = 0.83, CI = 0.79 – 0.87, p = 6.03x10\u003csup\u003e-14\u003c/sup\u003e). When we include \u003cem\u003eHLA-A*03:01\u0026nbsp;\u003c/em\u003eas a covariate in our model, we observed a nearly identical association of SSE with BTI (OR = 0.82, CI = 0.79 – 0.87, \u0026nbsp;p = 9.40×10\u003csup\u003e-14\u0026nbsp;\u003c/sup\u003e), while the \u003cem\u003eHLA-A*03:01\u0026nbsp;\u003c/em\u003e(OR = 0.93, CI = 0.89 – 0.98, p = 5.81x10\u003csup\u003e-3\u003c/sup\u003e) association was only borderline significant, suggesting an association independent from \u003cem\u003eHLA\u003c/em\u003e. Finally, we stratified our cohort according to the carriage of \u003cem\u003eHLA-A*03:01\u003c/em\u003e and further observed a highly significant and consistent inverse association between the combined measure for three or four SSE and the risk of BTI in individuals without the \u003cem\u003eHLA-A*03:01\u0026nbsp;\u003c/em\u003eallele (OR = 0.81 CI = 0.78 – 0.85, p = 4.32×10\u003csup\u003e-19\u003c/sup\u003e). Thus, we conclude that the negative association of \u003cem\u003eHLA-A*03:01\u003c/em\u003e with BTI stems from its role in increasing vaccine reactogenicity, along with a global negative association of vaccine side effects with BTI.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eHLA-A*03:01 associated vaccine reactogenicity is specific to COVID-19 vaccines\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo fully contextualize our findings, we sought to compare responses to COVID-19 vaccines with those of other widely administered vaccines, such as the currently approved influenza vaccine. We administered a follow-up questionnaire to the same individuals included in our discovery cohort (\u003cstrong\u003eSupplementary Table 17\u003c/strong\u003e). This survey specifically assessed the occurrence of influenza vaccine-associated side effects and the frequency with which participants received influenza vaccinations. As of April 2025, we received responses from a total of 13,499 participants from our discovery cohort, among whom 11,916 individuals reported having received the influenza vaccine at least once (\u003cstrong\u003eSupplementary Table 18\u003c/strong\u003e). Of those receiving the vaccine, 3,417 individuals reported experiencing side effects. As for the COVID-19 vaccine, we performed multivariate logistic regression to assess associations between each \u003cem\u003eHLA\u003c/em\u003e allele and the presence of at least one side effect (\u003cstrong\u003eSupplementary Table 19\u003c/strong\u003e). None of the tested \u003cem\u003eHLA\u003c/em\u003e alleles reached the Bonferroni-corrected threshold for statistical significance (p = 7.14 ×10\u003csup\u003e-4\u003c/sup\u003e), including \u003cem\u003eHLA-A*03:01\u003c/em\u003e. Thus, the \u003cem\u003eHLA\u003c/em\u003e association with vaccine reactogenicity appears to be specific to the COVID-19 vaccines.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eDespite the lack of \u003cem\u003eHLA\u003c/em\u003e-specific associations, we observed an epidemiologically meaningful correlation: individuals experiencing side effects following influenza vaccination were more than three times (OR = 3.094, CI = 2.73 – 3.50, p = 2 ×10\u003csup\u003e-16\u003c/sup\u003e) as likely to report side effects from the COVID-19 vaccine, after controlling with \u003cem\u003eHLA-A*03:01.\u0026nbsp;\u003c/em\u003eThus, it appears that there are factors beyond \u003cem\u003eHLA\u003c/em\u003e predisposing some individuals to experiencing vaccine reactogenicity.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eHigh frequency of low avidity Spike-specific T cell population is observed ex vivo\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe strong association between \u003cem\u003eHLA-A*03:01\u003c/em\u003e carriage and the presence of side effects after COVID-19 vaccination, coupled with more robust immunity to the virus (higher Spike-specific antibody and less breakthrough infection occurrence), led us to hypothesize that \u003cem\u003eHLA-A*03:01\u003csup\u003e+\u003c/sup\u003e\u003c/em\u003e individuals could mount a strong and perhaps disproportionate CD8\u003csup\u003e+\u003c/sup\u003e T cell response, as previously suggested by others\u003csup\u003e30\u003c/sup\u003e. To assess this hypothesis, blood from \u003cem\u003eHLA-A*03:01\u003csup\u003e+\u003c/sup\u003e\u003c/em\u003e and \u003cem\u003eHLA-A*03:01\u003csup\u003e-\u003c/sup\u003e\u003c/em\u003e individuals was collected, most without SARS-CoV-2 infection, prior (V0) and post vaccination with either 1 dose (V1) or 2 doses (V2) of the Pfizer or AstraZeneca vaccine (\u003cstrong\u003eSupplementary Table 20\u003c/strong\u003e). The peripheral blood mononuclear cells (PBMCs) and plasma were isolated and vaccination side-effects self-reported (converted to a side-effect severity score) (\u003cstrong\u003eSupplementary Table 21\u003c/strong\u003e).\u003c/p\u003e\n\u003cp\u003eTo examine the population of Spike-specific CD8\u003csup\u003e+\u003c/sup\u003e T cells present before and after vaccination, we considered the reported dominant HLA-A*03:01-restricted Spike-derived epitope \u003csub\u003e378\u003c/sub\u003eKCYGVSPTK\u003csub\u003e386\u003c/sub\u003e, hereafter named KCY\u003csup\u003e20,38\u003c/sup\u003e. \u0026nbsp; It has been shown that KCY-specific memory T cells increase upon viral exposure or vaccination\u003csup\u003e20\u003c/sup\u003e. Therefore, we performed \u003cem\u003eex vivo\u003c/em\u003e tetramer-associated magnetic enrichment (TAME) to analyze the phenotype of the KCY\u003csup\u003e+\u003c/sup\u003e CD8\u003csup\u003e+\u003c/sup\u003e T cells in samples collected before and after vaccination (\u003cstrong\u003eFigure 2A\u003c/strong\u003e). Surprisingly, even prior to vaccination (V0) a high frequency of KCY\u003csup\u003e+\u003c/sup\u003e T cells was present in all samples (\u003cstrong\u003eFigure 2A\u003c/strong\u003e). The frequency of KCY\u003csup\u003e+\u003c/sup\u003e T cells increased after vaccination (\u003cstrong\u003eFigure 2B-D\u003c/strong\u003e). Interestingly, after vaccination, a distinct population of KCY⁺ T cells with high mean fluorescence intensity (MFI) emerged (\u003cstrong\u003eFigure 2A\u003c/strong\u003e,\u003cstrong\u003e\u0026nbsp;D\u003c/strong\u003e). The high MFI KCY⁺ T cell population expanded significantly after vaccination compared to baseline (V0 mean: 1.49 ± 0.89; V1 mean: 7.49 ± 2.01; V2 mean: 5.53 ± 0.65) (\u003cstrong\u003eFigure 2D\u003c/strong\u003e). In comparison, TAME with the HLA-A*02:01- restricted S\u003csub\u003e269-277\u003c/sub\u003e peptide\u003csup\u003e39–41\u003c/sup\u003e did not show tetramer\u003csup\u003e+\u003c/sup\u003e T cells before vaccination, and only a high MFI population of tetramer\u003csup\u003e+\u003c/sup\u003e T cells was observed after vaccination(\u003cstrong\u003eSupplementary Figure 6A\u003c/strong\u003e). We also confirmed that the large population of KCY tetramer\u003csup\u003e+\u003c/sup\u003e T cells observed \u003cem\u003eex vivo\u003c/em\u003e was specific to the KCY peptide and not binding to an HLA-A*03:01-restricted influenza-derived peptide that we characterized previously, called Flu-NP\u003csub\u003e265\u003c/sub\u003e\u003csup\u003e42\u003c/sup\u003e (\u003cstrong\u003eFigure 2E\u003c/strong\u003e).\u003c/p\u003e\n\u003cp\u003eGiven the large frequency of KCY\u003csup\u003e+\u003c/sup\u003e T cells \u003cem\u003eex vivo\u003c/em\u003e, we wondered if this population would be expanded after activation \u003cem\u003ein vitro\u003c/em\u003e. To test this, we generated T cell lines using PBMCs cultured with either the KCY peptide or an HLA-A*03:01-restricted influenza-derived peptide, called Flu-NP\u003csub\u003e265\u003c/sub\u003e that we previously characterized\u003csup\u003e42\u003c/sup\u003e, to be tested in the same samples as the KCY peptide. The KCY T cell lines showed specificity for the KCY peptide only, and the Flu-NP\u003csub\u003e265\u003c/sub\u003e T cell lines were specific only to the Flu-NP\u003csub\u003e265\u003c/sub\u003e peptide (\u003cstrong\u003eSupplementary Figure 5B\u003c/strong\u003e). The frequency of KCY\u003csup\u003e+\u003c/sup\u003e CD8\u003csup\u003e+\u003c/sup\u003e T cells \u003cem\u003ein vitro\u003c/em\u003e was low (average of 0.71 ± 0.39 %, n = 3) (\u003cstrong\u003eSupplementary Figure 5B\u003c/strong\u003e), while it was 8-times higher for the Flu-NP\u003csub\u003e265\u003c/sub\u003e\u003csup\u003e+\u003c/sup\u003e T cells (average of 5.71 ± 1.67 %, n = 3) (\u003cstrong\u003eSupplementary Figure 5B\u003c/strong\u003e). This is in contrast with the high frequency observed \u003cem\u003eex vivo\u003c/em\u003e (\u003cstrong\u003eFigure 2A\u003c/strong\u003e); however, the \u003cem\u003eex vivo\u003c/em\u003e data were obtained after TAME. Therefore, we compared tetramer staining in \u003cem\u003eex vivo\u003c/em\u003e and \u003cem\u003ein vitro\u003c/em\u003e samples, without tetramer magnetic enrichment. We observed that there was a 16- and 44-fold lower frequency of KCY\u003csup\u003e+\u003c/sup\u003e T cells \u003cem\u003ein vitro\u003c/em\u003e compared to \u003cem\u003eex vivo\u003c/em\u003e in vacSG82-V2 and vacSG88-V2 samples, respectively (\u003cstrong\u003eSupplementary Figure 5C\u003c/strong\u003e). In addition, while the majority of \u003cem\u003eex vivo\u003c/em\u003e KCY\u003csup\u003e+\u003c/sup\u003e T cells had low MFI, the \u003cem\u003ein vitro\u003c/em\u003e cells were largely high MFI (\u003cstrong\u003eSupplementary Figure 5C\u003c/strong\u003e).\u003c/p\u003e\n\u003cp\u003eOverall, even prior to vaccination, an unusually large number of low avidity KCY-specific T cell populations was present in \u003cem\u003eHLA-A*03:01\u003csup\u003e+\u003c/sup\u003e\u003c/em\u003e samples\u003cem\u003e\u0026nbsp;ex vivo\u003c/em\u003e that were not expanded upon KCY presentation. In addition, vaccination led to an increase in high avidity KCY\u003csup\u003e+\u003c/sup\u003e T cell population, although these cells remained present at low frequencies relative to the low avidity population.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eKCY\u003csup\u003e+\u003c/sup\u003e T cells mostly exhibit a naïve phenotype\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe next assessed the phenotype of KCY\u003csup\u003e+\u003c/sup\u003e T cells \u003cem\u003eex vivo\u003c/em\u003e to determine if there was a difference between the high and the low avidity population (\u003cstrong\u003eFigure 2F-G\u003c/strong\u003e,\u003cstrong\u003e\u0026nbsp;Supplementary Figure 6A\u003c/strong\u003e). The majority of the low MFI KCY\u003csup\u003e+\u003c/sup\u003e T cell population exhibited a naïve phenotype (T\u003csub\u003eN\u003c/sub\u003e: CCR7\u003csup\u003e+\u003c/sup\u003e/CD45RA\u003csup\u003e+\u003c/sup\u003e) independent of the vaccine status of the donors (average of 69.95, 56.35 and 63.9 % for V0, V1 and V2, respectively), and some stem cell memory T cells (T\u003csub\u003eSCM\u003c/sub\u003e: CCR7\u003csup\u003e+\u003c/sup\u003e/CD45RA\u003csup\u003e+\u003c/sup\u003e/CD95\u003csup\u003e+\u003c/sup\u003e) present especially in one donor (V1) (\u003cstrong\u003eFigure 2F\u003c/strong\u003e, \u003cstrong\u003eSupplementary Figure 6A\u003c/strong\u003e). The low MFI effector memory T cell (T\u003csub\u003eEM\u003c/sub\u003e: CCR7\u003csup\u003e-\u003c/sup\u003e/CD45RA\u003csup\u003e-\u003c/sup\u003e) frequency was ~16%, and the terminally differentiated effector T cell (T\u003csub\u003eEMRA\u003c/sub\u003e: CCR7\u003csup\u003e-\u003c/sup\u003e/CD45RA\u003csup\u003e+\u003c/sup\u003e) frequency was around ~10% independent of vaccination status.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe comparison of high and low MFI tetramer\u003csup\u003e+\u003c/sup\u003e CD8\u003csup\u003e+\u003c/sup\u003e T cell phenotype showed that while the proportion of low MFI naïve T cells remains the same before and after vaccination (\u003cstrong\u003eFigure 2F, Supplementary Figure 6A\u003c/strong\u003e), the high MFI naïve T cell proportion decreased after vaccination while high MFI effector and central memory T cell proportion increased (\u003cstrong\u003eFigure\u003c/strong\u003e \u003cstrong\u003e2G, Supplementary Figure 6A\u003c/strong\u003e). Strikingly, the phenotype of the KCY\u003csup\u003e+\u003c/sup\u003e T cells observed in \u003cem\u003eHLA-A*03:01\u003csup\u003e+\u003c/sup\u003e\u003c/em\u003e samples was different from the one observed for the S\u003csub\u003e269\u003c/sub\u003e peptide in \u003cem\u003eHLA-A*02:01\u003csup\u003e+\u003c/sup\u003e\u003c/em\u003e samples (\u003cstrong\u003eSupplementary Figure 6B\u003c/strong\u003e). We did not observe any S\u003csub\u003e269\u003c/sub\u003e\u003csup\u003e+\u003c/sup\u003e T cells in V0 samples, and S\u003csub\u003e269\u003c/sub\u003e\u003csup\u003e+\u003c/sup\u003e T cells were only observed in 1 out of 3 donors after V1 with a T\u003csub\u003eCM\u003c/sub\u003e and naïve phenotype (\u003cstrong\u003eSupplementary Figure 6B-C\u003c/strong\u003e). In V2 samples T\u003csub\u003eEM,\u0026nbsp;\u003c/sub\u003ebut no T\u003csub\u003eEMRA\u003c/sub\u003e cells, were observed. This is in contrast with the presence of both KCY\u003csup\u003e+\u003c/sup\u003e T\u003csub\u003eEM\u003c/sub\u003e and T\u003csub\u003eEMRA\u003c/sub\u003e cells, even prior to vaccination (\u003cstrong\u003eFigure 2F-G\u003c/strong\u003e, \u003cstrong\u003eSupplementary Figure 6C\u003c/strong\u003e). In two samples collected \u0026gt; 400 days after the 3\u003csup\u003erd\u003c/sup\u003e vaccine dose (vacSG64-V5 and vacSG86-V5), we could observe 36% and 8.85% of KCY\u003csup\u003e+\u003c/sup\u003e T\u003csub\u003eEMRA\u003c/sub\u003e cells, respectively, and a large population of naïve cells (32% and 79.7%, respectively) (\u003cstrong\u003eSupplementary Figure 6D\u003c/strong\u003e).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eOverall, despite the naïve phenotype of the large proportion of low avidity KCY-specific T cells, the cells are peptide specific. In addition, the high avidity effector memory cells are present before vaccination, increased after vaccination and persist over time. Nevertheless, the vast majority of KCY-specific T cells remain low avidity, naïve, and in high numbers.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eSpike-specific T cell activation in HLA-A*03:01\u003csup\u003e+\u003c/sup\u003e donors is relatively weak\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo understand whether the T cell response was likely responsible for the observed vaccine reactogenicity in \u003cem\u003eHLA-A*03:01\u003c/em\u003e\u003csup\u003e+\u003c/sup\u003e donors, T cell lines were generated with Spike-derived peptide pools covering the whole length of the Spike protein and restimulated with either the peptide pools or the KCY peptide. The response before vaccination (V0) against the Spike-derived peptide pools was low for both CD8\u003csup\u003e+\u003c/sup\u003e and CD4\u003csup\u003e+\u003c/sup\u003e T cells (\u003cstrong\u003eFigure 2H\u003c/strong\u003e,\u003cstrong\u003e\u0026nbsp;Supplementary Figures 7 – 10)\u003c/strong\u003e. A trend upward of IFNγ producing CD8\u003csup\u003e+\u003c/sup\u003e T cells was observed after vaccination (V1; mean ± SD; 0.09 ± 0.08 %, and V2; 0.39 ± 0.48 %) (\u003cstrong\u003eFigure 2H\u003c/strong\u003e), but not for CD4\u003csup\u003e+\u003c/sup\u003e T cells (V1: 0.02 ± 0.04 %, and V2: 0.02 ± 0.07 %) (\u003cstrong\u003eSupplementary Figures 10A\u003c/strong\u003e).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe also assessed the KCY peptide specific response. We observed limited IFNγ production in 50% of the samples before vaccination (n=4/8, 0.05 ± 0.06 %), and a larger IFNγ production in 85% (n=6/7, average of 0.16 ± 0.19 %) and 57% (n=4/7, average of 0.15 ± 0.19 %) after the first and second vaccine dose, respectively (\u003cstrong\u003eFigure 2I\u003c/strong\u003e,\u003cstrong\u003e\u0026nbsp;Supplementary Figures 7-8\u003c/strong\u003e). Despite the increase of IFNγproducing KCY-specific CD8\u003csup\u003e+\u003c/sup\u003e T cells after vaccination, the response was overall weak compared with other well characterized Spike-derived epitopes such as HLA-A*02:01-restricted S\u003csub\u003e269-277\u003c/sub\u003e (average of 5.85% IFNγ\u003csup\u003e+\u003c/sup\u003e CD8\u003csup\u003e+\u003c/sup\u003e T cells\u003csup\u003e39\u003c/sup\u003e or HLA-B*15:01-restricted S\u003csub\u003e919-927\u003c/sub\u003e (average of 0.36% IFNγ\u003csup\u003e+\u003c/sup\u003e CD8\u003csup\u003e+\u003c/sup\u003e T cells in pre-pandemic samples\u003csup\u003e23\u003c/sup\u003e). In addition, we previously characterized an HLA-A*03:01-restricted influenza-derived peptide, called Flu-NP\u003csub\u003e265\u003c/sub\u003e\u003csup\u003e42\u003c/sup\u003e that was also tested here in the same samples side by side with the KCY peptide. The response to the Flu-NP\u003csub\u003e265\u003c/sub\u003e peptide was substantially stronger than that to the KCY peptide, with IFNγ production being ~13-fold higher (Flu-NP\u003csub\u003e265\u003c/sub\u003e: 3.32 ± 2.36 %; KCY: 0.27 ± 0.19 %) (\u003cstrong\u003eFigure 2J\u003c/strong\u003e,\u003cstrong\u003e\u0026nbsp;Supplementary Figure 8\u003c/strong\u003e,\u003cstrong\u003e\u0026nbsp;10D\u003c/strong\u003e), TNF production ~25-fold higher (Flu-NP\u003csub\u003e265\u003c/sub\u003e: 3.31 ± 2.28 %; KCY: 0.13 ± 0.06 %), and CD107a ~10-fold higher than KCY (Flu-NP\u003csub\u003e265\u003c/sub\u003e: 3.32 ± 2.36 %; KCY: 0.34 ± 0.33 %) (\u003cstrong\u003eSupplementary Figure 10E\u003c/strong\u003e).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe response towards the Flu-NP\u003csub\u003e265\u003c/sub\u003e peptide demonstrated that HLA-A*03:01\u003csup\u003e+\u003c/sup\u003e CD8\u003csup\u003e+\u0026nbsp;\u003c/sup\u003eT cells can produce high level of cytokines. However, the CD8\u003csup\u003e+\u003c/sup\u003e T cell response towards the Spike-derived peptides, even for the dominant KCY peptide, was overall weak even after vaccination, and therefore unlikely to underpin the vaccine side effect correlation observed in \u003cem\u003eHLA-A*03:01\u003csup\u003e+\u003c/sup\u003e\u003c/em\u003e individuals. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eFull-length Spike protein stimulated high expression of IL-6 and IL-8 by monocytes\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDespite the high number of KCY\u003csup\u003e+\u003c/sup\u003e T cells present (\u003cstrong\u003eFigure 2A\u003c/strong\u003e), the overall T cell response did not show the high levels of cytokine production that could explain vaccine reactogenicity associated with \u003cem\u003eHLA-A*03:01\u003c/em\u003e carriage (\u003cstrong\u003eFigure 2H-I\u003c/strong\u003e). Therefore, we asked whether other immune cells could lead to inflammation that would underpin vaccine side effects. To address this, we used the different vaccine components to stimulate PBMCs; the empty lipid nanoparticle (LNP) using the Pfizer vaccine formula\u003csup\u003e43\u003c/sup\u003e the soluble full-length Spike protein (HexaPro)\u003csup\u003e44\u003c/sup\u003e, or the Spike-derived peptide pools (S1 and S2), as well as positive and negative controls.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAmong the 20 \u003cem\u003eHLA-A*03:01⁺\u003c/em\u003e and 14 \u003cem\u003eHLA-A*03:01\u003csup\u003e-\u003c/sup\u003e\u003c/em\u003e PBMC samples tested (\u003cstrong\u003eSupplementary Table 20\u003c/strong\u003e), detectable cytokine production was only observed in samples stimulated with the full-length soluble Spike protein (\u003cstrong\u003eFigure 3\u003c/strong\u003e, \u003cstrong\u003eSupplementary Figure 11\u003c/strong\u003e). Neither the Spike-derived peptide pools nor the LNPs induced measurable cytokine responses in either group, except for Monocyte Chemoattractant Protein-1 (MCP-1) (\u003cstrong\u003eFigure 3\u003c/strong\u003e). Although MCP-1 levels showed a modest increase \u003cem\u003ein HLA-A*03:01⁺\u003c/em\u003e samples following Spike stimulation (120.7 ± 340.9 pg/mL) compared with \u003cem\u003eHLA-A*03:01\u003csup\u003e-\u003c/sup\u003e\u003c/em\u003e samples (0.0 ± 0.0 pg/mL), the response remained within the expected baseline variation observed in healthy individuals (~250 pg/mL), indicating no biologically meaningful induction of MCP-1 by Spike stimulation. Similar response was observed for eLNP and Spike Pool stimulations (253.4 ± 432.1 pg/mL and 333.3 ± 523.5 pg/mL, respectively). Cytokines induced by whole Spike stimulation, included Interleukin (IL)-6, IL-8, IL-1a, IL-1b, IL-10, macrophage inflammatory protein-1a\u0026nbsp;(MIP1a), granulocyte-macrophage colony-stimulating factor (GM-CSF), RANTES, MCP-1, tumor necrosis factor (TNF), inducible protein 10 kDa (IP-10), and Interferon-γ (IFNγ) (\u003cstrong\u003eFigure 3\u003c/strong\u003e,\u003cstrong\u003e\u0026nbsp;Supplementary Figure 11\u003c/strong\u003e). The increase was observed in both HLA-A*03:01\u003csup\u003e+\u003c/sup\u003e and HLA-A*03:01\u003cem\u003e\u003csup\u003e-\u003c/sup\u003e\u003c/em\u003e samples, and no significant difference in the level of cytokine was observed between the groups. However, IL-6 and IL-8 were modestly elevated in \u003cem\u003eHLA-A*03:01⁺\u0026nbsp;\u003c/em\u003esamples after Spike stimulation (12,186 ± 13,659 pg/mL and 29,875 ± 30,682 pg/mL, respectively), representing ~1.1- and ~1.3-fold increase compared with IL-6 (11,234 ± 11,177 pg/mL) and IL-8 (22,436 ± 20,784 pg/mL) levels in HLA-A*03:01\u003csup\u003e-\u003c/sup\u003e samples (\u003cstrong\u003eFigure 3\u003c/strong\u003e). IL-6 and IL-8 cytokines were also expressed at the highest concentration compared to other cytokines.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe next asked which cell subset was responsible for the production of IL-6 and IL-8 upon Spike stimulation. To address this, cytokine production was assessed in Spike-stimulated PBMCs using flow cytometry using multiple cell surface markers. IL-6 was produced predominantly by monocytes (CD14\u003csup\u003e+\u003c/sup\u003e) despite the low number of CD14\u003csup\u003e+\u003c/sup\u003e cells in blood (\u003cstrong\u003eSupplementary Figure 12A\u003c/strong\u003e-\u003cstrong\u003eB\u003c/strong\u003e), and at lower levels by Natural Killer (NK) cells (CD56\u003csup\u003e+\u003c/sup\u003e), suggesting a primary role of the innate response in vaccine reactogenicity (\u003cstrong\u003eFigure 4A\u003c/strong\u003e). In HLA-A*03:01⁺ samples, IL-6⁺ NK cells averaged 0.23 ± 0.23 % (n = 9/9 positive) and monocytes 7.7 ± 11.9 % (n = 3/9), whereas in HLA-A*03:01\u003csup\u003e-\u003c/sup\u003e samples comparable NK responses (0.45 ± 0.56 %, n = 7/9) but broader monocyte positivity (9.5 ± 12.4 %, n = 5/9) were observed (\u003cstrong\u003eFigure 4A\u003c/strong\u003e). In contrast, IL-6⁺ T and B cells were rare, with only traceable responses (\u0026lt; 0.03 %, n ≤ 6/9 donors). IL-8 responses were even more restricted. Only a minority of monocyte-positive donors showed detectable IL-8, averaging 2.2 ± 6.7 % (n = 1/9) in HLA-A*03:01⁺ and 3.3 ± 3.8 % (n = 5/9) in HLA-A*03:01\u003csup\u003e-\u003c/sup\u003e samples. Other subsets (NK, T, and B cells) produced negligible IL-8 (\u0026lt; 0.02 %, n ≤ 5/9 donors) (\u003cstrong\u003eFigure 4B\u003c/strong\u003e). Compared to the IL-6 and IL-8 levels secreted (\u003cstrong\u003eFigure 3A\u003c/strong\u003e), the levels of IL-6⁺ and IL-8⁺ cells detected were at low frequencies in PBMCs, therefore, we examined the Spike uptake capacity by the PBMCs. Phagocytic scores (engulfment of Spike-coated microbeads; \u003cstrong\u003eFigure 4C\u003c/strong\u003e,\u003cstrong\u003e\u0026nbsp;Supplementary Figure 12C\u003c/strong\u003e) confirmed that Spike uptake was mainly driven by monocytes and at lower level by NK cells.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eOverall, the high level of IL-6 and IL-8 production was observed only in the presence of the full Spike protein and was primarily produced by monocytes.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eExpression of IL-6 and IL-8 correlates with side effect severity in HLA-A*03:01 donors\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWhile our results suggest a role for IL-6 and IL-8 in vaccine reactogenicity, the differences observed between grouped HLA-A*03:01⁺ and HLA-A*03:01\u003csup\u003e-\u003c/sup\u003e donor samples were not sufficient to explain the observed differences in response to vaccination. However, as with most complex phenotypes, the association of HLA-A*03:01⁺with vaccine reactogenicity is incompletely penetrant. While more individuals carrying this allele report SSE with vaccination, there was a range of severity reported among our PBMC donors; thus, we sought to determine whether levels of IL-6 and IL-8 were correlated with reported side effect severity in these donors. Strikingly, in HLA-A*03:01⁺ samples, both IL-8 and IL-6 levels showed strong and significant positive correlations with severity score (IL-8: r = 0.70, p = 0.02; IL-6: r = 0.75, p = 0.01; \u003cstrong\u003eFigure 4D-E\u003c/strong\u003e), indicating that higher cytokine production was associated with increased vaccine side effect severity. In contrast, no significant correlation was observed in HLA-A*03:01\u003csup\u003e-\u003c/sup\u003e individuals (\u003cstrong\u003eFigure 4F-G\u003c/strong\u003e). To ensure that the correlation observed in HLA-A*03:01\u003csup\u003e+\u003c/sup\u003e samples was due to the transient presence of Spike protein, we checked the baseline levels of IL-6 and IL-8 cytokines in serum of HLA-A*03:01\u003csup\u003e+\u003c/sup\u003e collected prior to vaccination, or two-weeks post first and second dose of vaccine, alongside TNF and IFNγ as control (\u003cstrong\u003eSupplementary Figure 13\u003c/strong\u003e). Overall, the level of cytokines was low (\u0026lt; 100 pg/mL) or moderate, and no significant increase was observed before or after vaccination. This demonstrates that the cytokine production upon Spike presentation is likely transient.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTogether, these data establish that HLA-A*03:01⁺ samples display a distinct pro-inflammatory signature, with IL-6 and IL-8 production strongly linked to vaccine side-effect severity. The early and transient nature of this cytokine production primarily by monocytes and NK cells strongly suggests an innate immune response underlying \u003cem\u003eHLA-A*03:01\u003c/em\u003e vaccine reactogenicity.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eHLA-A*03:01 is an eQTL for IRF4 driving monocyte differentiation\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFinally, to better understand the relationship between \u003cem\u003eHLA-A*03:01\u003c/em\u003e carriage and the observed role of innate immune cells in vaccine reactogenicity\u003cem\u003e,\u0026nbsp;\u003c/em\u003ewe examined patterns of differential gene expression driven by \u003cem\u003eHLA-A*03:01\u003c/em\u003e. Limiting our analysis to genes on chromosome 6 (723 tests), differential gene analysis for 21 donors showed that \u003cem\u003eIRF4,\u0026nbsp;\u003c/em\u003ewhich is involved in monocyte differentiation to dendritic cells (DCs) and homing to lymph nodes\u003csup\u003e45\u003c/sup\u003e, is significantly upregulated in \u003cem\u003eHLA-A*03:01\u003c/em\u003e donors (log fold 1.863, p-value = 9.56 × 10\u003csup\u003e-6\u003c/sup\u003e, p\u003csub\u003eadj\u0026nbsp;\u003c/sub\u003e= 0.0069) (\u003cstrong\u003eSupplementary Table 22\u003c/strong\u003e). Thus, it appears that the \u003cem\u003eHLA-A*03:01\u0026nbsp;\u003c/em\u003eassociation with SARS-CoV-2 vaccine reactogenicity may be in part due to its role as an eQTL, resulting in increased differentiation of monocytes to DCs, increasing antigen-presenting cells and homing them to the lymph nodes.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eVaccination has been a crucial public health intervention for decades, significantly contributing to the prevention and control of infectious diseases worldwide. However, concerns about vaccine safety and efficacy, and negative public perceptions regarding side effects, have emerged as a significant challenge in maintaining high vaccination coverage.\u0026nbsp;Because the binding between HLA and peptide antigen is highly specific and a fundamental component in initiating the adaptive immune system, understanding the role of \u003cem\u003eHLA\u003c/em\u003e variation in vaccine response can be crucial in determining factors that underlie the effectiveness of vaccination. Variation in HLA has previously been reported as associated with SARS-CoV-2 vaccine reactogenicity\u003csup\u003e30,31,37\u003c/sup\u003e. Here, in a much larger cohort than previously examined, we sought to refine and understand the relationship between variation at all \u003cem\u003eHLA\u003c/em\u003e loci and reports of side effects associated with the vaccine.\u003c/p\u003e\n\u003cp\u003eWe leveraged a large, registry-based cohort of more than 50,000 individuals to provide the necessary statistical power and diversity to reliably identify genetic factors that influence individual susceptibility to systemic reactions following vaccination.\u0026nbsp;Variation in the HLA region has previously been associated with interindividual differences in humoral immune responses after vaccination. For example, \u003cem\u003eHLA\u003c/em\u003e variation has been linked to either increased or decreased immune responses to influenza\u003csup\u003e46\u003c/sup\u003e, measles\u003csup\u003e47\u003c/sup\u003e, rubella\u003csup\u003e48\u003c/sup\u003e, and hepatitis B vaccination\u003csup\u003e49\u003c/sup\u003e, respectively. In COVID-19 vaccination, previous reports have shown a suggested association of HLA variation with specific systemic mild side effects such as fever and fatigue, including \u003cem\u003eHLA-A*03:01\u003c/em\u003e\u003csup\u003e30\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eHere, we showed significant associations of \u003cem\u003eHLA-A*03:01\u0026nbsp;\u003c/em\u003ewith increased side effects such as fever, chills, and muscle pain, particularly with higher effect sizes in individuals who received the Pfizer vaccine. Evidence from this and prior studies\u003csup\u003e19,50\u003c/sup\u003e demonstrates that \u003cem\u003eHLA-A*03:01\u0026nbsp;\u003c/em\u003eis also significantly associated with high serum levels of anti-SARS-CoV-2-Spike antibodies. This supports the notion that this allomorph promotes an especially robust immune response to vaccination for SARS-CoV-2, including both cell-mediated and humoral immunity.\u0026nbsp;Additionally, we demonstrated that a negative correlation of \u003cem\u003eHLA-A*03:01\u003c/em\u003e with BTI is likely mediated by the systemic inflammatory response that causes vaccination adverse effects.\u0026nbsp;To fully contextualize the findings, we compared responses to\u0026nbsp;COVID-19\u0026nbsp;vaccines with those of influenza vaccine. Our finding of no significant association of\u0026nbsp;\u003cem\u003eHLA-A*03:01\u0026nbsp;\u003c/em\u003ewith side effects from influenza vaccine (or any other \u003cem\u003eHLA\u003c/em\u003e allele) demonstrates that the observed \u003cem\u003eHLA-A*03:01\u0026nbsp;\u003c/em\u003eassociated vaccine reactogenicity is specific to the COVID-19 vaccine.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eOwing to the crucial role of HLA class I molecules in antigen presentation, the role of CD8\u003csup\u003e+\u003c/sup\u003e T cells in HLA-A*03:01-mediated vaccine reactogenicity presented a clear initial line of inquiry into the mechanisms underlying reported side effects. We observed a high frequency of CD8\u003csup\u003e+\u003c/sup\u003e T cells able to recognize the immunodominant Spike-derived peptide KCY specifically\u003csup\u003e20\u003c/sup\u003e, even prior to vaccination or infection. Interestingly, both prior to and subsequent to vaccination, the majority of the CD8\u003csup\u003e+\u003c/sup\u003e T cells exhibited low MFI (Mean Florescence Intensity), suggesting that the cells are low avidity; in addition, a large proportion of those peptide-specific cells had a naïve phenotype. This contrasts with our previous study on a dominant Influenza-derived peptide, Flu-NP\u003csub\u003e265\u003c/sub\u003e\u003csup\u003e42\u003c/sup\u003e for which we observed low and high MFI cells \u003cem\u003eex vivo\u003c/em\u003e, consistent with the fact that prior viral exposure and/or vaccination leads to the presence of high MFI T cells. Post-COVID-19 vaccination we did observe the expansion of high MFI KCY-specific CD8\u003csup\u003e+\u003c/sup\u003e T cells and an increased proportion of memory phenotype. However, in contrast with \u003cem\u003eHLA-A*02:01\u003csup\u003e+\u003c/sup\u003e\u003c/em\u003e samples, even after multiple doses of the COVID-19 vaccine, the frequency of KCY\u003csup\u003e+\u003c/sup\u003e naïve CD8\u003csup\u003e+\u003c/sup\u003e T cells remained high, likely reflecting the large proportion of naïve T cells able to specifically bind the KCY peptide independent of vaccine status. Most strikingly, even post-vaccination, the T cell response to KCY was muted relative to responses to Influenza-derived peptide, such that it is unlikely that the observed COVID-19 vaccine reactogenicity can be attributed to a robust T cell activation.\u003c/p\u003e\n\u003cp\u003eMonocytes thus emerged as a central population of interest: they were the predominant source of IL-6 and IL-8 following Spike stimulation, driving the pro-inflammatory response that correlated with side-effect severity in \u003cem\u003eHLA-A*03:01\u003c/em\u003e⁺ donors. Strikingly, this cytokine induction was accompanied by a reduction of the already low frequency of CD14⁺ monocytes in blood, suggesting a shift in lineage fate rather than simple activation, consistent with our RNA-sequencing results.\u0026nbsp;Our findings reveal a distinct gene expression signature associated with \u003cem\u003eHLA-A*03:01\u003c/em\u003e, suggesting upregulation of the monocyte to DC differentiation pathway. Increased \u003cem\u003eIRF4\u003c/em\u003e expression in \u003cem\u003eHLA-A*03:01\u003csup\u003e+\u003c/sup\u003e\u003c/em\u003e individuals suggests increased DC priming and more efficient migration to lymph nodes\u003csup\u003e45\u003c/sup\u003e. Given that CD14 downregulation is a hallmark of monocyte-to-dendritic cell differentiation, our results suggest that \u003cem\u003eHLA-A*03:01\u003c/em\u003e⁺ individuals likely have higher frequencies of dendritic cells, highly efficient antigen presenters, homed to the lymph node than \u003cem\u003eHLA-A*03:01\u003c/em\u003e\u003csup\u003e-\u003c/sup\u003e individuals.\u003c/p\u003e\n\u003cp\u003eDespite what appears to be the central role of the monocyte-DC lineage in cytokine production associated with vaccine reactogenicity, the unusually high numbers of low-avidity, naïve, Spike-specific CD8\u003csup\u003e+\u003c/sup\u003e T cells observed in\u0026nbsp;\u003cem\u003eHLA-A*03:01\u003csup\u003e+\u003c/sup\u003e\u0026nbsp;\u003c/em\u003edonors likely provide an inflammatory milieu in the lymph node after vaccination. While these T cells\u0026nbsp;do not appear to become activated themselves, their binding to peptide-HLA presented by DCs at high frequency may constitute a signal of immune activity for DCs\u003csup\u003e51\u003c/sup\u003e; \u0026nbsp;this likely contributes to activation of these DCs, resulting in increased production of IL-6 and other cytokines\u003csup\u003e52,53\u003c/sup\u003e. \u0026nbsp;Thus, we postulate that in \u003cem\u003eHLA\u003c/em\u003e\u003cem\u003e-A*03:01\u003csup\u003e+\u003c/sup\u003e\u003c/em\u003e individuals, a high number of low-avidity, naïve T cells is available to bind to already primed DCs stimulated by Spike, resulting in an amplified immune response after COVID-19 vaccination.\u003c/p\u003e\n\u003cp\u003eThis study is intrinsically constrained by its dependence on self-reported data to assess transient mild vaccine side effects in both the discovery and replication cohorts, which potentially can lead to some imprecision in association results. Additionally, sample size limitations restrict some significant findings to individuals who self-identify as White. Likewise, our cohort was predominantly female (78.4%), which may limit the generalizability to broader populations. Moreover, the low number of monocyte cells in the blood limited the ability to show significant differences, in addition to the incomplete penetrance of the observed genetic effect, which is typical for complex traits. The transient features of SSE also constrained these observations. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eDespite these limitations, our\u0026nbsp;findings regarding the\u0026nbsp;role of \u003cem\u003eHLA-\u003c/em\u003emediated COVID-19 vaccine reactogenicity and the associated evidence for protection from subsequent infection provide important and novel insights regarding these responses, which may inform efforts toward improved vaccine efficacy and increased public participation in vaccination programs.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003e\u003cstrong\u003e\u003cem\u003eDiscovery cohort\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe recruited our study population via email to all potential volunteer bone marrow donors registered in the NMDP database with available email addresses and high-resolution HLA genotyping information available. The email contained a custom link directing them to a consent page for a health history survey. Subject recruitment, consent process, and survey administration were conducted using both email outreach and a web interface to ensure effective data collection. Upon consenting to provide responses and allow linking with their HLA genotype data, participants spent ten to fifteen minutes completing a detailed survey to gather baseline information and health history.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAs of January 19th, 2024, a total of 80,016 eligible donors completed the survey. Among these respondents, 667 individuals (0.83%) were excluded for filling out the survey multiple times, while 1,301 participants were removed due to incomplete HLA variation data. After excluding these cases and individuals that were also participated in the study that formed our replication cohort, below (N = 836), there remained a total of 77,212 participants in the study. \u0026nbsp; Of these individuals, 56,938 people who self-identified as White, Hispanic, African American, or Asian Pacific Islanders, had completed their initial series of vaccinations. We excluded participants who identified as multiple ancestry, unknown ancestry or Native American due to small sample sizes.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eReplication cohort\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eOur replication cohort consisted of participants who participated in a prior study with NMDP tracking experiences with COVID-19 through a mobile app, described in detail in Augusto et al. 2023\u003csup\u003e23\u003c/sup\u003e. Once enrolled, the participants are asked to complete an initial 10 to 15-minute survey about baseline demographics, their health history, and daily habits. Follow-up daily questions specific to vaccine side effects are delivered by push notification or text message on an ongoing basis and require 5 to 15-minute per week. All the participants provided written informed consent agreeing to the research and publication of research results.\u003c/p\u003e\n\u003cp\u003eWe restricted our analysis to individuals who had self-identified as ‘White’ (which we use as a proxy for European ancestry) due to insufficient numbers for analysis in the other groups, allowing an analysis of 10,595 individuals reporting vaccination for SARS-CoV-2. Of those, 4,575 individuals completed their initial series of vaccinations. Symptoms are self-reported at the baseline and in daily surveys. Within the baseline survey, the respondents were asked to report whether they had any of a list of symptoms (\u003cstrong\u003eSupplementary Table\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e23\u003c/strong\u003e) for 3 days or longer at any time after their complete dose of vaccination.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eSerum antibody levels in vaccinated subjects\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe population examined consisted of 156 healthcare workers from “Evangelismos” General Hospital in Athens, Greece, including doctors, nurses, pharmacists, biologists, dentists, technicians and administrative staff. Enrollment was open to all hospital personnel scheduled for vaccination and not restricted by any pre-specified criteria. \u0026nbsp;All individuals received two doses of the mRNA Pfizer-BioNTech vaccine. Data on prior SARS-CoV-2 infection and symptoms experienced after each dose were collected for all participants. Antibody concentrations were assessed at two time points: 21 ± 1 days after the first dose and 24 ± 2 days after the second dose. Levels of circulating SARS-CoV-2 anti-Spike IgG (S) and anti-nucleocapsid IgG (N) antibodies were quantified using the Abbott Diagnostics SARS-CoV-2 IgG chemiluminescent microparticle immunoassay (Abbott Diagnostics, Abbott Park, Illinois) on an Abbott Diagnostics Architect i2000 SR and an Alinityi Analyzer, according to the manufacturer’s instructions. Results were expressed in AU/mL and were interpreted as positive if ≥ 50 AU/mL\u003csup\u003e54\u003c/sup\u003e. \u0026nbsp;Informed consent was obtained from all participants, and the study was approved by the Institutional Review Board of “Evangelismos” Hospital (PN 9/21-01-21). High resolution HLA class I and II genotyping was performed as described\u003csup\u003e55\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003ePeripheral blood mononuclear cells (PBMCs)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eHLA-A*03:01\u003csup\u003e+\u003c/sup\u003e\u003c/em\u003e and \u003cem\u003eHLA-A*03:01\u003csup\u003e-\u003c/sup\u003e\u003c/em\u003e donors vaccinated with either the Comirnaty BNT162b2 COVID-19 mRNA vaccine (Pfizer) or the Oxford–AstraZeneca COVID‑19 vaccine (AstraZeneca), and most of them naïve for SARS-CoV-2 infection, were recruited (\u003cstrong\u003eSupplementary Table 20\u003c/strong\u003e). PBMCs were separated from whole blood or buffy coats using density-gradient centrifugation. PBMCs were used fresh or were cryogenically stored until use. All individuals consented to research and publication of research results and had been previously HLA genotyped. Ethics approval to undertake the research was obtained from the La Trobe University Human Research Ethics Committee (HEC21097). The HLA genotyping was performed by AlloSeq Tx17 (CareDx Pty) using AllType NGS high-resolution genotyping on the IonTorrent NGS platform or by the Department of Clinical Immunology and PathWest at Fiona Stanley Hospital, Murdoch, Australia.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eTetramer-associated magnetic enrichment (TAME)\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePeptide-loaded HLA-A*03:01 tetramers were generated using Streptavidin conjugated to phycoerythrin (PE). Tetramer-stained cells were enriched using anti-PE antibody-coated immunomagnetic beads on LS columns (Miltenyi Biotech) according to manufacturer instructions. After enrichment, cells were stained with an antibody panel including anti-CD3-BV480 (dilution 1:100), anti-CD8-PerCP-Cy5.5 (1:50), anti-CD4-FITC (1:100), anti-CD14-APCH7 (1:200), anti-CD19-APCH7 (1:100), anti-CD45RA-BUV395 (1:100), anti-CD27-APC (1:100), anti-CCR7-PE-Cy7 (1:50), anti-CD95-BV421 (1:50), anti-PD1-BV605 (1:100), anti-CXCR5-BV650 (1:100) (all BD Biosciences) and Live/Dead Fixable Near-IR Dead Cell Stain (1:1,000) (Life Technologies). Cells were resuspended in MACS buffer (PBS, 0.5% BSA, 2 mM EDTA) and were analysed using the BD FACSymphony A3 system.\u003c/p\u003e\n\u003cp\u003eFor the tetramer staining experiments, the TAME cells were stained for 1 hour at room temperature with the APC-conjugated Flu-NP\u003csub\u003e265\u003c/sub\u003e HLA-A*03:01-restricted peptide tetramer, followed by surface staining using the same antibody panel as above, excluding anti-CD27-APC. Gating strategy shown on \u003cstrong\u003eSupplementary Figure 14\u003c/strong\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eGeneration of peptide-specific CD8\u003csup\u003e+\u003c/sup\u003e T cell lines\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCD8\u003csup\u003e+\u003c/sup\u003e T cell lines were generated as previously described\u003csup\u003e56,57\u003c/sup\u003e. In brief, PBMCs were incubated with 1 μM of individual SARS-CoV-2 Spike-derived peptide or 10μg/mL of Spike-derived peptide Pool 1 (25 μg/peptide, 15mers, 1 – 126) and Pool 2 (25 μg/peptide, 15mers, 127 – 253) (Mimotopes B#33200); and cultured for 10 – 14 days in RPMI-1640 supplemented with 2 mM MEM non-essential amino acid solution (Sigma-Aldrich), 100 mM HEPES (Sigma-Aldrich), 2 mM\u0026nbsp;l-glutamine (Sigma-Aldrich), penicillin–streptomycin (Life Technologies), 50 mM 2-ME (Sigma-Aldrich) and 10% heat-inactivated fetal bovine serum (Bovogen). The cultures were supplemented with 10 IU IL-2 2 – 3 times weekly. CD8\u003csup\u003e+\u003c/sup\u003e T cell lines were used fresh for subsequent analysis.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFor the tetramer staining experiments 0.5 × 10\u003csup\u003e6\u003c/sup\u003e cells from the CD8\u003csup\u003e+\u003c/sup\u003e T cell lines were stained with a PE-conjugated tetramer (HLA-A*03:01-KCY) or double-stained with two tetramers (PE-conjugated KCY and APC-conjugated Flu-NP\u003csub\u003e265\u003c/sub\u003e HLA-A*03:01-restricted peptide tetramer) for 1 h at room temperature. Cells were washed and surface-stained with anti-CD3-BV480 (dilution 1:100), anti-CD8-PerCP-Cy5.5 (1:50), anti-CD4-BV650 or -FITC (1:100), anti-CD14-APCH7 (1:200) and anti-CD19-APCH7 (1:100) antibodies (all BD Biosciences) and Live/Dead Fixable Near-IR Dead Cell Stain (Life Technologies). Cells were analysed using the BD FACSymphony A3 system.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eIntracellular cytokine assay\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCD8\u003csup\u003e+\u003c/sup\u003e T cell lines were stimulated with 1 μM of individual peptide or 2μg/mL of the SARS-CoV-2 Spike-derived peptide Pool 1 (25 μg/peptide, 15mers, 1 – 126) and Pool 2 (25 μg/peptide, 15mers, 127 – 253) (Mimotopes B#33200) and were incubated for 4 – 5 hour in the presence of GolgiPlug, GolgiStop and anti-CD107a-FITC (dilution 1:100) (all BD Biosciences). After stimulation, cells were surface stained for 30 min with anti-CD3-BV480 (1:100), anti-CD8-PerCP-Cy5.5 (1:50) and anti-CD4-BV650 (1:100) antibodies (all BD Biosciences) and Live/Dead Fixable Near-IR Dead Cell Stain (Life Technologies). Cells were fixed and permeabilized using BD Cytofix/Cytoperm solution (BD Biosciences) and then intracellularly stained with anti-IFN-γ-BV421 (1:100), anti-TNF-PE-Cy7 (1:100), anti-IL2-PE (1:100) and anti-MIP-1β-APC (1:100) antibodies (all BD Biosciences) for a further 30 min. Cells were acquired on the BD FACSymphony A3 system using the FACSDiva software (v.9.0.). Post-acquisition analysis was performed using FlowJo software (v.10). Cytokine detection levels identified in the no-peptide control condition were subtracted from the corresponding test conditions in all summary graphs to account for non-specific, spontaneous cytokine production. Gating strategy shown on \u003cstrong\u003eSupplementary Figure 14\u003c/strong\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003ePBMCs short-term stimulation\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e1 x 10^6 PBMCs were stimulated with either 15 μg/mL of empty Pfizer-BioNTech lipid nanoparticle (LNP)\u003csup\u003e43\u003c/sup\u003e or 15\u0026nbsp;μg/mL custom made SARS-CoV-2 Spike protein (Wuhan strain); or 7.5\u0026nbsp;μg/mL of the SARS-CoV-2 Spike-derived peptide Pool 1 (25 μg/peptide, 15mers, 1 – 126) and\u0026nbsp;7.5\u0026nbsp;μg/mL Pool 2 (25 μg/peptide, 15mers, 127 – 253) (MIMOTOPES B#33200); or\u0026nbsp;Cell Stimulation Cocktail (500X) (eBioscience™); or nothing (Negative Control) for 15 hours in\u0026nbsp;RPMI-1640 supplemented with 2 mM MEM non-essential amino acid solution (Sigma-Aldrich), 100 mM HEPES (Sigma-Aldrich), 2 mM l-glutamine (Sigma-Aldrich), penicillin–streptomycin (Life Technologies), 50 mM 2-ME (Sigma-Aldrich) and 10% heat-inactivated fetal bovine serum (Bovogen).\u0026nbsp;Following the 15-hour stimulation, the cells were restimulated for a further 2-hour using the same conditions and the supernatant was collected for the BD Cytometric Bead Array (CBA).\u0026nbsp;The same conditions were also used in the presence of GolgiPlug and GolgiStop (BD Biosciences).\u0026nbsp;Following the 15-hour stimulation and the 2-hour restimulation, the cells were surface-stained\u0026nbsp;for 30 minutes with anti-CD3-BV480 (1:100), anti-CD14-PerCP-Cy5.5 (1:100), anti-CD16-BV421 (1:50), anti-CD19-APC (1:50), anti-CD56-PECy7 (1:50) antibodies (all BD Biosciences) and Live/Dead Fixable Near-IR Dead Cell Stain (Life Technologies). Cells were fixed and permeabilized using BD Cytofix/Cytoperm solution (BD Biosciences) and then intracellularly stained with anti-IL6-FITC (1:50) and anti-IL8-PE (1:50) antibodies (both BD Biosciences) for a further 30 minutes. Cells were acquired on the BD FACSymphony A3 system using the FACSDiva software (v.9.0.). Gating strategy shown on \u003cstrong\u003eSupplementary Figure 15\u003c/strong\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eBD Cytometric Bead Array (CBA)\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFollowing the PBMC short-term stimulation, the supernatant level of Interleukin (IL)-1α, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-9, IL-10, IL-12p70, IL-13, Interferon (IFN)-α, IFN-γ, IFN-γ inducible protein 10 kDa (IP-10), granulocyte-macrophage colony-stimulating factor (GM-CSF), Lymphotoxin-alpha (LT-α), Eotaxin, Monocyte Chemoattractant Protein-1 (MCP-1), macrophage inflammatory protein-1 alpha (MIP-1α), RANTES, tumor necrosis factor (TNF) and were measured using the BD Cytometric Bead Array (CBA, BD Biosciences) following the manufacturer’s instructions. Samples were acquired in a\u0026nbsp;BD FACSymphony A3 system using the FACSDiva software (v.9.0.).\u0026nbsp;The analysis was performed by using the FCAP Array Software v3.0.\u0026nbsp;\u0026nbsp;Cytokine detection levels identified in the Negative Control condition were subtracted from the corresponding test conditions in all summary graphs to account for non-specific, spontaneous cytokine production.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eSARS-CoV-2 specific phagocytosis assays\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAfter PBMCs stimulation with full length Spike protein for 15-hour, PBMCs (1 × 10\u003csup\u003e5\u0026nbsp;\u003c/sup\u003ecells in 50µL) were added onto 60 µL of the donor plasma opsonised microbeads (10µL plasma and 50µL microbeads) in a 1.5mL Eppendorf tube, mixed by gentle tapping, adjusted to 600 µL using RPMI 1640 containing 0.1% Human serum and 0.1 M HEPES pH 7.4 and transferred into 37ºC, 5 % CO\u003csub\u003e2\u003c/sub\u003e incubator. After 2-hour of incubation, cells were washed once with 1mL of cold PBS containing 0.5 % FBS and 0.005 % sodium azide and gentle centrifugation at 335 x g for 5 minutes at 4ºC, fixed in 400 µL of 1 % paraformaldehyde, and kept at 4ºC in the dark until the acquisition of data using BD FACSCaliburTM Flow cytometer. A total of 2 x10\u003csup\u003e4\u003c/sup\u003e events per tube were acquired from each donor conditions. Relevant assay controls included the acquisition of 2 x10\u003csup\u003e4\u003c/sup\u003e events per tube from cells incubated with no beads, Spike-coated nonopsonized beads. The proportions of cells that phagocytosed the beads (% of cells that took up the beads) and their fluorescent intensities (amounts of beads taken up per cell) were analyzed using BD FlowJo version 10.5.0 software. Phagocytic scores (p-score) were then calculated based on the proportion of cells that took up the opsonized beads denoting the number of positive cells and mean fluorescence intensity (MFI) representing the average bead uptake by the positive cells as described. A positive p-score was defined as three standard deviations above the background mean phagocytic score of healthy donors as described previously\u003csup\u003e58\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eIn selected experiments, the intracellular uptake of the opsonized microbeads by effector cells was confirmed by confocal microscopy as described\u003csup\u003e58\u003c/sup\u003e. In brief, PBMC after (1 × 10\u003csup\u003e4\u003c/sup\u003e cells) after phagocytosis assay were washed twice with cold PBS containing 0.5% FBS and 0.005 % of sodium azide, fixed with 1% paraformaldehyde for 5 minutes at room temperature, and rinsed twice with PBS. The fixed cells were blocked with 1% BSA in PBS, incubated with 1:1000 dilution of Alexa-555-conjugated Phalloidin (Sigma, USA) for 30 minutes at room temperature, mounted in DAPI nuclear stain-containing media (Molecular Probes, USA), and imaged using ZEISS LSM 880 confocal microscope (Carl Zeiss AG, Germany), using 63X/1.4 Plan-Apochromat Oil Immersion objective, with Diode 405 nm (DAPI), Argon ion 488 nm (Alexa-488) and DPSS 561 nm (Alexa-555 phalloidin) laser excitation sources, emitted light was filtered using a combination of emission filters and imaged onto Airy detector array producing an effective lateral resolution of ~100 nm. All the images were Airyscan processed with Zen Black Edition (Zeiss Software).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eAssessment of vaccine side effects\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo standardize the evaluation of vaccine-associated side effects, we developed a composite scoring framework termed the Side Effect Severity Score (SESS) adapted from the vaccine side effect guideline described elsewhere\u003csup\u003e59\u003c/sup\u003e. Each reported symptom was first categorized by type and severity (local, systemic, or severe) according to previously established criteria for vaccine reactogenicity. For each donor and vaccine dose, symptoms were graded as: 0 = none; 1 = mild (e.g., injection site pain, mild fatigue, or a single local symptom); 2 = moderate (systemic but not severe, e.g., fever, chills, headache, myalgia, or multiple mild symptoms); or 3 = severe (multiple systemic symptoms, prolonged recovery \u0026gt;2 days, swelling requiring medical review, dose-limiting reaction, or hospitalization). To account for compounded burden, additional multipliers were applied: +1 if multiple symptoms occurred at the same dose, +1 if symptoms recurred across multiple doses, and +1 if symptom duration exceeded 2 days (if reported). Scores were then summed to generate an overall SESS per participant, which was categorized as: 0 (no side effects), 1 – 2 (mild), 3 – 4 (moderate), and ≥ 5 (severe). The scores are summarized in \u003cstrong\u003eSupplementary Table 21\u003c/strong\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eRNA Sequencing\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe performed bulk RNAseq in a total of 21 samples of PBMCs (n = 10 \u003cem\u003eHLA-A*03:01\u003csup\u003e+\u003c/sup\u003e\u003c/em\u003e, n = 11 \u003cem\u003eHLA-A*03:01\u003csup\u003e-\u003c/sup\u003e\u003c/em\u003e) and compared the two groups based on \u003cem\u003eHLA\u003c/em\u003e genotype. Total RNA libraries were prepared and sequenced by Novogene Corporation using Illumina NovaSeq platforms following standard protocols\u003csup\u003e60\u003c/sup\u003e. RNA quantity and integrity were assessed to ensure a minimum RNA Integrity Number (RIN) of \u0026gt; 3.0. The raw sequence data generated by Novogene met strict quality criteria as described by the provider\u003csup\u003e60,61\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eRNAseq data processing and analysis\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe analyzed all 21 donors from an independent cohort which included individuals of European and non-European ancestries. Samples were sequenced in two batches, with raw RNA-seq data merged and then normalized for batch effects (\u003cstrong\u003eSupplementary Figure 16\u003c/strong\u003e) using Combat-seq\u003csup\u003e62\u003c/sup\u003e These sequences were then analyzed using the nf-core/rnaseq pipeline (version 3.19.0), executed through Nextflow\u003csup\u003e63\u003c/sup\u003e. Initial quality control was performed with FastQC, followed by adapter and low-quality base trimming using Trim Galore. Reads were then aligned to the reference genome [Human Genome Assembly GRCh38.p14] using STAR. Transcript quantification was accomplished using Salmon, as defined by the pipeline configuration. The resulting gene-level count matrices were imported into R for normalization and differential expression analysis using the DESeq2 package\u003csup\u003e64\u003c/sup\u003e. Additional downstream analyses, including gene set enrichment analysis, were conducted using standard R workflows.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eStatistical analysis\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eHLA\u003c/em\u003e associations: In our discovery cohort, we examined the association of five \u003cem\u003eHLA\u003c/em\u003e loci (\u003cem\u003eHLA-A, -B, -C, -DRB1, -DQB1\u003c/em\u003e). Data analysis included the first two fields of the allele name as described in the HLA nomenclature, representing the complete molecule at polypeptide sequence resolution. We calculated allele frequencies for all the \u003cem\u003eHLA\u003c/em\u003e loci, haplotype frequencies using Haplostats R package\u003csup\u003e65\u003c/sup\u003e and R2 for Linkage Disequilibrium (from our in-house script) between all the pair of loci (\u003cstrong\u003eSupplementary Table 14\u003c/strong\u003e). We employed a generalized logistic regression model using ‘glm’ in the R (V 4.3) base package to consider relevant covariates, including sex and age. For the replication cohort, we tested only the allele of interest, using the generalized logistic regression model framework as described. We utilized our in-house Python script to construct forest plots. We conducted an analysis of variance (ANOVA) to assess the statistical significance of differences in antibody positivity rates across groups with different counts of \u003cem\u003eHLA-A*03:01\u0026nbsp;\u003c/em\u003eallele(0,1 or 2).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAll additional analyses were performed in GraphPad Prism v10.1. Data are shown as mean ± SD, with symbols representing individual donors. Paired comparisons across time points were assessed using two-tailed Wilcoxon matched-pairs tests, and unpaired comparisons between HLA-A*03:01⁺ and HLA-A*03:01⁻ groups by two-tailed Mann–Whitney U tests. For the CBA, cytokine values were background-subtracted from unstimulated or controls; analytes below detection limits were excluded. Correlations between cytokine levels and side-effect severity were evaluated using Spearman’s r. Frequencies of cytokine-producing or tetramer-positive T cells and phagocytic scores were compared using non-parametric tests. All tests were two-tailed; p \u0026lt; 0.05 was considered significant.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by NIH R01 AI159260 and NIH R01 AI158861 to JAH. National Health and Medical Research Council (NHMRC, GNT2014002 and GNT1161832) and Australian Research Council (ARC); National Collaborative Research Infrastructure Strategy (NCRIS) Therapeutic Innovation Australia (TIA) to T.R.M.; University of Queensland to T.R.M. SG is supported by an NHMRC Leadership Investigator Grant (#2034677). PJN was supported by NIH R01 AI158410. We wish to thank the volunteer donors registered with NMDP who participated in this study.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eRahmani, K. \u003cem\u003eet al.\u003c/em\u003e The effectiveness of COVID-19 vaccines in reducing the incidence, hospitalization, and mortality from COVID-19: A systematic review and meta-analysis. \u003cem\u003eFront Public Health\u003c/em\u003e\u003cstrong\u003e10\u003c/strong\u003e, (2022).\u003c/li\u003e\n\u003cli\u003eWu, N. \u003cem\u003eet al.\u003c/em\u003e Long-term effectiveness of COVID-19 vaccines against infections, hospitalisations, and mortality in adults: findings from a rapid living systematic evidence synthesis and meta-analysis up to December, 2022. \u003cem\u003eLancet Respir Med\u003c/em\u003e\u003cstrong\u003e11\u003c/strong\u003e, 439\u0026ndash;452 (2023).\u003c/li\u003e\n\u003cli\u003eDeCuir, J. \u003cem\u003eet al.\u003c/em\u003e Interim Effectiveness of Updated 2023\u0026ndash;2024 (Monovalent XBB.1.5) COVID-19 Vaccines Against COVID-19\u0026ndash;Associated Emergency Department and Urgent Care Encounters and Hospitalization Among Immunocompetent Adults Aged \u0026ge;18 Years \u0026mdash; VISION and IVY Networks, September 2023\u0026ndash;January 2024. \u003cem\u003eMMWR Morb Mortal Wkly Rep\u003c/em\u003e\u003cstrong\u003e73\u003c/strong\u003e, 180\u0026ndash;188 (2024).\u003c/li\u003e\n\u003cli\u003ePayne, A. B. \u003cem\u003eet al.\u003c/em\u003e Effectiveness of Bivalent mRNA COVID-19 Vaccines in Preventing COVID-19\u0026ndash;Related Thromboembolic Events Among Medicare Enrollees Aged \u0026ge;65 Years and Those with End Stage Renal Disease \u0026mdash; United States, September 2022\u0026ndash;March 2023. \u003cem\u003eMMWR Morb Mortal Wkly Rep\u003c/em\u003e\u003cstrong\u003e73\u003c/strong\u003e, 16\u0026ndash;23 (2024).\u003c/li\u003e\n\u003cli\u003ePolack, F. P. \u003cem\u003eet al.\u003c/em\u003e Safety and Efficacy of the BNT162b2 mRNA Covid-19 Vaccine. \u003cem\u003eNew England Journal of Medicine\u003c/em\u003e\u003cstrong\u003e383\u003c/strong\u003e, 2603\u0026ndash;2615 (2020).\u003c/li\u003e\n\u003cli\u003eMascellino, M. T., Di Timoteo, F., De Angelis, M. \u0026amp; Oliva, A. Overview of the Main Anti-SARS-CoV-2 Vaccines: Mechanism of Action, Efficacy and Safety. \u003cem\u003eInfect Drug Resist\u003c/em\u003e\u003cstrong\u003eVolume 14\u003c/strong\u003e, 3459\u0026ndash;3476 (2021).\u003c/li\u003e\n\u003cli\u003eSadoff, J. \u003cem\u003eet al.\u003c/em\u003e Final Analysis of Efficacy and Safety of Single-Dose Ad26.COV2.S. \u003cem\u003eNew England Journal of Medicine\u003c/em\u003e\u003cstrong\u003e386\u003c/strong\u003e, 847\u0026ndash;860 (2022).\u003c/li\u003e\n\u003cli\u003eAlcendor, D. J. \u003cem\u003eet al.\u003c/em\u003e Breakthrough COVID-19 Infections in the US: Implications for Prolonging the Pandemic. \u003cem\u003eVaccines (Basel)\u003c/em\u003e\u003cstrong\u003e10\u003c/strong\u003e, 755 (2022).\u003c/li\u003e\n\u003cli\u003eSuntronwong, N. \u003cem\u003eet al.\u003c/em\u003e COVID-19 Breakthrough Infection after Inactivated Vaccine Induced Robust Antibody Responses and Cross-Neutralization of SARS-CoV-2 Variants, but Less Immunity against Omicron. \u003cem\u003eVaccines (Basel)\u003c/em\u003e\u003cstrong\u003e10\u003c/strong\u003e, 391 (2022).\u003c/li\u003e\n\u003cli\u003eGopinath, S. \u003cem\u003eet al.\u003c/em\u003e Characteristics of COVID-19 Breakthrough Infections among Vaccinated Individuals and Associated Risk Factors: A Systematic Review. \u003cem\u003eTrop Med Infect Dis\u003c/em\u003e\u003cstrong\u003e7\u003c/strong\u003e, 81 (2022).\u003c/li\u003e\n\u003cli\u003eSuleyman, G. \u003cem\u003eet al.\u003c/em\u003e Risk Factors Associated With Hospitalization and Death in COVID-19 Breakthrough Infections. \u003cem\u003eOpen Forum Infect Dis\u003c/em\u003e\u003cstrong\u003e9\u003c/strong\u003e, (2022).\u003c/li\u003e\n\u003cli\u003eLopera, T. J. \u003cem\u003eet al.\u003c/em\u003e Humoral Response to BNT162b2 Vaccine Against SARS-CoV-2 Variants Decays After Six Months. \u003cem\u003eFront Immunol\u003c/em\u003e\u003cstrong\u003e13\u003c/strong\u003e, (2022).\u003c/li\u003e\n\u003cli\u003eWiedermann, U., Garner-Spitzer, E. \u0026amp; Wagner, A. Primary vaccine failure to routine vaccines: Why and what to do? \u003cem\u003eHum Vaccin Immunother\u003c/em\u003e\u003cstrong\u003e12\u003c/strong\u003e, 239\u0026ndash;243 (2016).\u003c/li\u003e\n\u003cli\u003eChen, J., Wang, R., Gilby, N. B. \u0026amp; Wei, G.-W. Omicron Variant (B.1.1.529): Infectivity, Vaccine Breakthrough, and Antibody Resistance. \u003cem\u003eJ Chem Inf Model\u003c/em\u003e\u003cstrong\u003e62\u003c/strong\u003e, 412\u0026ndash;422 (2022).\u003c/li\u003e\n\u003cli\u003eNotarte, K. I. \u003cem\u003eet al.\u003c/em\u003e Characterization of the significant decline in humoral immune response six months post‐SARS‐CoV‐2 mRNA vaccination: A systematic review. \u003cem\u003eJ Med Virol\u003c/em\u003e\u003cstrong\u003e94\u003c/strong\u003e, 2939\u0026ndash;2961 (2022).\u003c/li\u003e\n\u003cli\u003eFalahi, S. \u0026amp; Kenarkoohi, A. Host factors and vaccine efficacy: Implications for COVID‐19 vaccines. \u003cem\u003eJ Med Virol\u003c/em\u003e\u003cstrong\u003e94\u003c/strong\u003e, 1330\u0026ndash;1335 (2022).\u003c/li\u003e\n\u003cli\u003eLynn, D. J., Benson, S. C., Lynn, M. A. \u0026amp; Pulendran, B. Modulation of immune responses to vaccination by the microbiota: implications and potential mechanisms. \u003cem\u003eNat Rev Immunol\u003c/em\u003e\u003cstrong\u003e22\u003c/strong\u003e, 33\u0026ndash;46 (2022).\u003c/li\u003e\n\u003cli\u003eMentzer, A. J. \u003cem\u003eet al.\u003c/em\u003e Human leukocyte antigen alleles associate with COVID-19 vaccine immunogenicity and risk of breakthrough infection. \u003cem\u003eNat Med\u003c/em\u003e\u003cstrong\u003e29\u003c/strong\u003e, 147\u0026ndash;157 (2023).\u003c/li\u003e\n\u003cli\u003eEsposito, M. \u003cem\u003eet al.\u003c/em\u003e Human leukocyte antigen variants associate with BNT162b2 mRNA vaccine response. \u003cem\u003eCommunications Medicine\u003c/em\u003e\u003cstrong\u003e4\u003c/strong\u003e, 63 (2024).\u003c/li\u003e\n\u003cli\u003eMayer-Blackwell, K. \u003cem\u003eet al.\u003c/em\u003e mRNA vaccination boosts S-specific T\u0026nbsp;cell memory and promotes expansion of CD45RAint TEMRA-like CD8+ T\u0026nbsp;cells in COVID-19 recovered individuals. \u003cem\u003eCell Rep Med\u003c/em\u003e\u003cstrong\u003e4\u003c/strong\u003e, 101149 (2023).\u003c/li\u003e\n\u003cli\u003eBian, S. \u003cem\u003eet al.\u003c/em\u003e Genetic determinants of IgG antibody response to COVID-19 vaccination. \u003cem\u003eThe American Journal of Human Genetics\u003c/em\u003e\u003cstrong\u003e111\u003c/strong\u003e, 181\u0026ndash;199 (2024).\u003c/li\u003e\n\u003cli\u003eBlackwell, J. M., Jamieson, S. E. \u0026amp; Burgner, D. HLA and Infectious Diseases. \u003cem\u003eClin Microbiol Rev\u003c/em\u003e\u003cstrong\u003e22\u003c/strong\u003e, 370\u0026ndash;385 (2009).\u003c/li\u003e\n\u003cli\u003eAugusto, D. G. \u003cem\u003eet al.\u003c/em\u003e A common allele of HLA is associated with asymptomatic SARS-CoV-2 infection. \u003cem\u003eNature\u003c/em\u003e\u003cstrong\u003e620\u003c/strong\u003e, 128\u0026ndash;136 (2023).\u003c/li\u003e\n\u003cli\u003eSrivastava, A. \u0026amp; Hollenbach, J. A. The immunogenetics of COVID-19. \u003cem\u003eImmunogenetics\u003c/em\u003e\u003cstrong\u003e75\u003c/strong\u003e, 309\u0026ndash;320 (2023).\u003c/li\u003e\n\u003cli\u003eHerv\u0026eacute;, C., Laup\u0026egrave;ze, B., Del Giudice, G., Didierlaurent, A. M. \u0026amp; Tavares Da Silva, F. The how\u0026rsquo;s and what\u0026rsquo;s of vaccine reactogenicity. \u003cem\u003eNPJ Vaccines\u003c/em\u003e\u003cstrong\u003e4\u003c/strong\u003e, 39 (2019).\u003c/li\u003e\n\u003cli\u003eShimabukuro, T. T., Cole, M. \u0026amp; Su, J. R. Reports of Anaphylaxis After Receipt of mRNA COVID-19 Vaccines in the US\u0026mdash;December 14, 2020-January 18, 2021. \u003cem\u003eJAMA\u003c/em\u003e\u003cstrong\u003e325\u003c/strong\u003e, 1101 (2021).\u003c/li\u003e\n\u003cli\u003eZhuang, C.-L. \u003cem\u003eet al.\u003c/em\u003e Inflammation-related adverse reactions following vaccination potentially indicate a stronger immune response. \u003cem\u003eEmerg Microbes Infect\u003c/em\u003e\u003cstrong\u003e10\u003c/strong\u003e, 365\u0026ndash;375 (2021).\u003c/li\u003e\n\u003cli\u003eHermann, E. A. \u003cem\u003eet al.\u003c/em\u003e Association of Symptoms After COVID-19 Vaccination With Anti\u0026ndash;SARS-CoV-2 Antibody Response in the Framingham Heart Study. \u003cem\u003eJAMA Netw Open\u003c/em\u003e\u003cstrong\u003e5\u003c/strong\u003e, e2237908 (2022).\u003c/li\u003e\n\u003cli\u003eYoshida, M. \u003cem\u003eet al.\u003c/em\u003e Association of systemic adverse reaction patterns with long-term dynamics of humoral and cellular immunity after coronavirus disease 2019 third vaccination. \u003cem\u003eSci Rep\u003c/em\u003e\u003cstrong\u003e13\u003c/strong\u003e, 9264 (2023).\u003c/li\u003e\n\u003cli\u003eBolze, A. \u003cem\u003eet al.\u003c/em\u003e HLA-A\u0026lowast;03:01 is associated with increased risk of fever, chills, and stronger side effects from Pfizer-BioNTech COVID-19 vaccination. \u003cem\u003eHuman Genetics and Genomics Advances\u003c/em\u003e\u003cstrong\u003e3\u003c/strong\u003e, 100084 (2022).\u003c/li\u003e\n\u003cli\u003eMagri, C. \u003cem\u003eet al.\u003c/em\u003e Genome‐wide association studies of response and side effects to the \u0026lt;scp\u0026gt;BNT162b2\u0026lt;/scp\u0026gt; vaccine in Italian healthcare workers: Increased antibody levels and side effects in carriers of the \u003cem\u003eHLA‐A*03:01\u003c/em\u003e allele. \u003cem\u003eHLA\u003c/em\u003e\u003cstrong\u003e102\u003c/strong\u003e, 707\u0026ndash;719 (2023).\u003c/li\u003e\n\u003cli\u003eSidney, J. \u003cem\u003eet al.\u003c/em\u003e Definition of an HLA-A3-like supermotif demonstrates the overlapping peptide-binding repertoires of common HLA molecules. \u003cem\u003eHum Immunol\u003c/em\u003e\u003cstrong\u003e45\u003c/strong\u003e, 79\u0026ndash;93 (1996).\u003c/li\u003e\n\u003cli\u003eNguyen, A. T., Szeto, C. \u0026amp; Gras, S. The pockets guide to HLA class I molecules. \u003cem\u003eBiochem Soc Trans\u003c/em\u003e\u003cstrong\u003e49\u003c/strong\u003e, 2319\u0026ndash;2331 (2021).\u003c/li\u003e\n\u003cli\u003eHorton, R. \u003cem\u003eet al.\u003c/em\u003e Variation analysis and gene annotation of eight MHC haplotypes: The MHC Haplotype Project. \u003cem\u003eImmunogenetics\u003c/em\u003e\u003cstrong\u003e60\u003c/strong\u003e, 1\u0026ndash;18 (2008).\u003c/li\u003e\n\u003cli\u003eBertinetto, F. E. \u003cem\u003eet al.\u003c/em\u003e The humoral and cellular response to \u0026lt;scp\u0026gt;mRNA SARS‐CoV\u0026lt;/scp\u0026gt; ‐2 vaccine is influenced by \u0026lt;scp\u0026gt;HLA\u0026lt;/scp\u0026gt; polymorphisms. \u003cem\u003eHLA\u003c/em\u003e\u003cstrong\u003e102\u003c/strong\u003e, 301\u0026ndash;315 (2023).\u003c/li\u003e\n\u003cli\u003eCrocchiolo, R. \u003cem\u003eet al.\u003c/em\u003e Strong humoral response after Covid‐19 vaccination correlates with the common \u0026lt;scp\u0026gt;HLA\u0026lt;/scp\u0026gt; allele \u003cem\u003eA*03:01\u003c/em\u003e and protection from breakthrough infection. \u003cem\u003eHLA\u003c/em\u003e\u003cstrong\u003e103\u003c/strong\u003e, (2024).\u003c/li\u003e\n\u003cli\u003eXie, J. \u003cem\u003eet al.\u003c/em\u003e Relationship between HLA genetic variations, COVID-19 vaccine antibody response, and risk of breakthrough outcomes. \u003cem\u003eNat Commun\u003c/em\u003e\u003cstrong\u003e15\u003c/strong\u003e, 4031 (2024).\u003c/li\u003e\n\u003cli\u003eMeyer, S. \u003cem\u003eet al.\u003c/em\u003e Prevalent and immunodominant CD8 T\u0026nbsp;cell epitopes are conserved in SARS-CoV-2 variants. \u003cem\u003eCell Rep\u003c/em\u003e\u003cstrong\u003e42\u003c/strong\u003e, 111995 (2023).\u003c/li\u003e\n\u003cli\u003eSzeto, C. \u003cem\u003eet al.\u003c/em\u003e Molecular basis of a dominant SARS-CoV-2 spike-derived epitope presented by HLA-A*02:01 recognised by a public TCR. \u003cem\u003eCells\u003c/em\u003e\u003cstrong\u003e10\u003c/strong\u003e, 2646 (2021).\u003c/li\u003e\n\u003cli\u003eKared, H. \u003cem\u003eet al.\u003c/em\u003e SARS-CoV-2\u0026ndash;specific CD8+ T cell responses in convalescent COVID-19 individuals. \u003cem\u003eJ Clin Invest\u003c/em\u003e\u003cstrong\u003e131\u003c/strong\u003e, (2021).\u003c/li\u003e\n\u003cli\u003eShomuradova, A. S. \u003cem\u003eet al.\u003c/em\u003e SARS-CoV-2 Epitopes Are Recognized by a Public and Diverse Repertoire of Human T Cell Receptors. \u003cem\u003eImmunity\u003c/em\u003e\u003cstrong\u003e53\u003c/strong\u003e, 1245-1257.e5 (2020).\u003c/li\u003e\n\u003cli\u003eNguyen, A. T. \u003cem\u003eet al.\u003c/em\u003e Homologous peptides derived from influenza A, B and C viruses induce variable CD8+ T cell responses with cross-reactive potential. \u003cem\u003eClin Transl Immunology\u003c/em\u003e\u003cstrong\u003e11\u003c/strong\u003e, e1422 (2022).\u003c/li\u003e\n\u003cli\u003eLeighton, L. J. \u003cem\u003eet al.\u003c/em\u003e The design, manufacture and LNP formulation of mRNA for research use. \u003cem\u003eNature Protocols 2025\u003c/em\u003e 1\u0026ndash;30 (2025) doi:10.1038/s41596-025-01174-4.\u003c/li\u003e\n\u003cli\u003eLu, M. \u003cem\u003eet al.\u003c/em\u003e SARS-CoV-2 prefusion spike protein stabilized by six rather than two prolines is more potent for inducing antibodies that neutralize viral variants of concern. \u003cem\u003eProc Natl Acad Sci U S A\u003c/em\u003e\u003cstrong\u003e119\u003c/strong\u003e, e2110105119 (2022).\u003c/li\u003e\n\u003cli\u003eBaja\u0026ntilde;a, S., Roach, K., Turner, S., Paul, J. \u0026amp; Kovats, S. IRF4 Promotes Cutaneous Dendritic Cell Migration to Lymph Nodes during Homeostasis and Inflammation. \u003cem\u003eThe Journal of Immunology\u003c/em\u003e\u003cstrong\u003e189\u003c/strong\u003e, 3368\u0026ndash;3377 (2012).\u003c/li\u003e\n\u003cli\u003eZhong, S. \u003cem\u003eet al.\u003c/em\u003e Single Nucleotide Polymorphisms in the Human Leukocyte Antigen Region Are Associated With Hemagglutination Inhibition Antibody Response to Influenza Vaccine. \u003cem\u003eFront Genet\u003c/em\u003e\u003cstrong\u003e13\u003c/strong\u003e, (2022).\u003c/li\u003e\n\u003cli\u003eHayney, M. S., Poland, G. A., Jacobson, R. M., Schaid, D. J. \u0026amp; Lipsky, J. J. The influence of the HLA-DRB1*13 allele on measles vaccine response. \u003cem\u003eJ Investig Med\u003c/em\u003e\u003cstrong\u003e44\u003c/strong\u003e, 261\u0026ndash;3 (1996).\u003c/li\u003e\n\u003cli\u003eLambert, N. D. \u003cem\u003eet al.\u003c/em\u003e Polymorphisms in HLA-DPB1 Are Associated With Differences in Rubella Virus-Specific Humoral Immunity After Vaccination. \u003cem\u003eJournal of Infectious Diseases\u003c/em\u003e\u003cstrong\u003e211\u003c/strong\u003e, 898\u0026ndash;905 (2015).\u003c/li\u003e\n\u003cli\u003eLi, Z.-K., Nie, J.-J., Li, J. \u0026amp; Zhuang, H. The effect of HLA on immunological response to hepatitis B vaccine in healthy people: A meta-analysis. \u003cem\u003eVaccine\u003c/em\u003e\u003cstrong\u003e31\u003c/strong\u003e, 4355\u0026ndash;4361 (2013).\u003c/li\u003e\n\u003cli\u003eSantos-Rebou\u0026ccedil;as, C. B. \u003cem\u003eet al.\u003c/em\u003e Immune response stability to the SARS-CoV-2 mRNA vaccine booster is influenced by differential splicing of HLA genes. \u003cem\u003eSci Rep\u003c/em\u003e\u003cstrong\u003e14\u003c/strong\u003e, 8982 (2024).\u003c/li\u003e\n\u003cli\u003eOzga, A. J. \u003cem\u003eet al.\u003c/em\u003e pMHC affinity controls duration of CD8+ T cell-DC interactions and imprints timing of effector differentiation versus expansion. \u003cem\u003eJournal of Experimental Medicine\u003c/em\u003e\u003cstrong\u003e213\u003c/strong\u003e, 2811\u0026ndash;2829 (2016).\u003c/li\u003e\n\u003cli\u003eKranzer, K. \u003cem\u003eet al.\u003c/em\u003e Induction of maturation and cytokine release of human dendritic cells by Helicobacter pylori. \u003cem\u003eInfect Immun\u003c/em\u003e\u003cstrong\u003e72\u003c/strong\u003e, 4416\u0026ndash;4423 (2004).\u003c/li\u003e\n\u003cli\u003eTada, Y. \u003cem\u003eet al.\u003c/em\u003e Differential effects of LPS and TGF-\u0026beta; on the production of IL-6 and IL-12 by Langerhans cells, splenic dendritic cells, and macrophages. \u003cem\u003eCytokine\u003c/em\u003e\u003cstrong\u003e25\u003c/strong\u003e, 155\u0026ndash;161 (2004).\u003c/li\u003e\n\u003cli\u003eBryan, A. \u003cem\u003eet al.\u003c/em\u003e Performance Characteristics of the Abbott Architect SARS-CoV-2 IgG Assay and Seroprevalence in Boise, Idaho. \u003cem\u003eJ Clin Microbiol\u003c/em\u003e\u003cstrong\u003e58\u003c/strong\u003e, (2020).\u003c/li\u003e\n\u003cli\u003eNorman, P. J. \u003cem\u003eet al.\u003c/em\u003e Defining KIR and HLA Class I Genotypes at Highest Resolution via High-Throughput Sequencing. \u003cem\u003eAm J Hum Genet\u003c/em\u003e\u003cstrong\u003e99\u003c/strong\u003e, 375\u0026ndash;391 (2016).\u003c/li\u003e\n\u003cli\u003eLineburg, K. E. \u003cem\u003eet al.\u003c/em\u003e CD8+ T cells specific for an immunodominant SARS-CoV-2 nucleocapsid epitope cross-react with selective seasonal coronaviruses. \u003cem\u003eImmunity\u003c/em\u003e\u003cstrong\u003e54\u003c/strong\u003e, 1055-1065.e5 (2021).\u003c/li\u003e\n\u003cli\u003eGrant, E. J. \u0026amp; Gras, S. Protocol for generation of human peptide-specific primary CD8+ T cell lines. \u003cem\u003eSTAR Protoc\u003c/em\u003e\u003cstrong\u003e3\u003c/strong\u003e, 101590 (2022).\u003c/li\u003e\n\u003cli\u003eAdhikari, A. \u003cem\u003eet al.\u003c/em\u003e Longitudinal Characterization of Phagocytic and Neutralization Functions of Anti-Spike Antibodies in Plasma of Patients after Severe Acute Respiratory Syndrome Coronavirus 2 Infection. \u003cem\u003eJ Immunol\u003c/em\u003e\u003cstrong\u003e209\u003c/strong\u003e, 1499\u0026ndash;1512 (2022).\u003c/li\u003e\n\u003cli\u003eCausality assessment of an adverse event following immunization (AEFI) User manual for the revised WHO classification.\u003c/li\u003e\n\u003cli\u003eNovogene - USA Based Lab Guaranteed Data Security. https://www.novogene.com/us-en/.\u003c/li\u003e\n\u003cli\u003eA basic guide to RNA-sequencing - Novogene. https://www.novogene.com/us-en/resources/blog/a-basic-guide-to-rna-sequencing/.\u003c/li\u003e\n\u003cli\u003eZhang, Y., Parmigiani, G. \u0026amp; Johnson, W. E. ComBat-seq: batch effect adjustment for RNA-seq count data. \u003cem\u003eNAR Genom Bioinform\u003c/em\u003e\u003cstrong\u003e2\u003c/strong\u003e, (2020).\u003c/li\u003e\n\u003cli\u003eEwels, P. A. \u003cem\u003eet al.\u003c/em\u003e The nf-core framework for community-curated bioinformatics pipelines. \u003cem\u003eNature Biotechnology 2020 38:3\u003c/em\u003e\u003cstrong\u003e38\u003c/strong\u003e, 276\u0026ndash;278 (2020).\u003c/li\u003e\n\u003cli\u003eLove, M. I., Huber, W. \u0026amp; Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. \u003cem\u003eGenome Biol\u003c/em\u003e\u003cstrong\u003e15\u003c/strong\u003e, 1\u0026ndash;21 (2014).\u003c/li\u003e\n\u003cli\u003ehaplo.stats.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-8282930/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8282930/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Vaccination against SARS-CoV-2 has been the most effective tool in mitigating the COVID-19 pandemic. However, some individuals experience side effects that cause distress and interfere with daily activities, which can limit vaccine uptake, with important public health implications. Here, we considered the impact of HLA variation on the propensity for mild side effects with COVID-19 vaccination. We examined variation in HLA-A, -B, -C, -DRB1, and -DQB1 for association with self-reported side effects in a large cohort of U.S. Euro-ancestry vaccinated individuals (N = 50,535) and confirmed results in an independent replication cohort (N = 4,575). We found that HLA-A*03:01 was significantly associated with systemic side effects (OR = 1.36, CI = 1.31-1.41, p = 6.79×10-57) and fewer breakthrough infections, and that this phenomenon is specific to the COVID-19 vaccine. Surprisingly, we observed limited activation of CD8+ T cells in HLA-A*03:01+ samples to the Spike-derived peptides, excluding them as a likely source of the reported vaccine side effects. Rather, examination of immune cell subsets, prior and after vaccination, points to a central role for monocytes in the production of IL-6 and IL-8, which significantly correlates with the reported severity of side effects in HLA-A*03:01+ donors. Meanwhile, the large, mostly naïve, and low-affinity population of Spike-specific CD8+ T cells likely contribute to an inflammatory milieu in HLA-A*03:01 carriers through weak binding to antigen presenting cells. This work sheds light on the mechanisms underlying HLA-mediated COVID-19 vaccine reactogenicity and associated reduction in infections, providing important new insights that may support efforts to optimize vaccine efficacy and promote broader public involvement in vaccination programs.","manuscriptTitle":"An HLA Association With COVID-19 Vaccine Reactogenicity Correlates With \nFewer SARS-CoV-2 Infections and Monocyte Activation","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-12-17 09:27:27","doi":"10.21203/rs.3.rs-8282930/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"e522eb48-7a18-4ce8-8a6c-8bb286f0ddc7","owner":[],"postedDate":"December 17th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":59766757,"name":"Biological sciences/Genetics"},{"id":59766758,"name":"Biological sciences/Immunology/Vaccines"}],"tags":[],"updatedAt":"2026-03-02T10:35:24+00:00","versionOfRecord":[],"versionCreatedAt":"2025-12-17 09:27:27","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8282930","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8282930","identity":"rs-8282930","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}