Humoral profiles of toddlers and young children following SARS-CoV-2 mRNA vaccination

Although young children generally experience mild symptoms following infection with SARS-CoV-2, severe acute and long-term complications can occur. SARS-CoV-2 mRNA vaccines elicit robust immunoglobulin profiles in children ages 5 years and older, and in adults, corresponding with substantial protection against hospitalizations and severe disease. Whether similar immune responses and humoral protection can be observed in vaccinated infants and young children, who have a developing and vulnerable immune system, remains poorly understood. To study the impact of mRNA vaccination on the humoral immunity of infant, we used a system serology approach to comprehensively profile antibody responses in a cohort of children ages 6 months to 5 years who were vaccinated with the mRNA-1273 COVID-19 vaccine (25 μg). Responses were compared with vaccinated adults (100 μg), in addition to naturally infected toddlers and young children. Despite their lower vaccine dose, vaccinated toddlers elicited a stronger functional antibody response than adults, including against variant of concerns (VOCs), without report of side effects. Moreover, mRNA vaccination was associated with a higher IgG3-dependent humoral profile against SARS-CoV-2 compared to natural infection, supporting that mRNA vaccination is effective at eliciting a robust antibody response in toddlers and young children.


44
Despite the early misconception that children were spared from COVID-19, children continue to 45 account for approximately twenty percent of all documented cases of COVID-19 infection in the 46 United States, with infants and children under 5 years of age disproportionately affected by high 47 In order to characterize the activation of humoral immunity in young children after SARS-CoV-67 2-specific mRNA vaccination, we used an unbiased system serology approach to analyze antibody 68 levels and Fc-mediated functions in individuals ages 6 months through 5 years. We 69 comprehensively profiled their antibody response following vaccination with the mRNA-1273 70 COVID-19 vaccine (25 μg) and compared it with antibody profiles of vaccinated adults (100 μg), 71 as well as children infected with SARS-CoV-2. Our results reveal a strong activation of humoral 72 immunity post-vaccination in these young children, with a highly functional and cross-reactive 73 humoral immunity in comparison to adults and naturally infected infants. 74 75

mRNA-vaccinated infants and toddlers generate robust Immunoglobulin G (IgG) responses 77
Our first objective was to profile vaccine-induced humoral immunity in infants and toddlers ages 78 6 months through 5 years (n = 18) after completion of the two doses of the pediatric mRNA-1273 79 vaccination series (vaccine dose: 25mcg mRNA-1273) and compare these serologic responses to 80 those generated by fully vaccinated adults (n = 13; vaccine dose: 100mcg mRNA-1273). 81 Demographics of participants are included in Table 1; mean age of vaccinated pediatric 82 participants was 2.2 years (range 7 months-4.5 years). None of the vaccinated adults or children 83 reported SARS-CoV-2 infections prior to or during their vaccine series, which was supported by 84 absence of elevated nucleocapsid responses (Fig. S1). 85

86
Our results demonstrate that despite their young age and receipt of only one quarter of the adult 87 dose, total anti-Spike and anti-RBD IgG levels and IgG subclass in young children were similar to 88 adults (Fig. 1A, Fig. S2). Interestingly, in contrast to IgG, this young population displayed lower 89 levels of vaccine-induced anti-Spike and anti-RBD IgM and IgA1, which shows the distinct 90 isotype selection between adults and children (Fig. 1A). We then compared the binding of spike 91 and RBD-specific antibodies to Fc receptors (FcR), as well as antibody effector functions, 92 including antibody-dependent cellular (monocyte) phagocytosis (ADCP), antibody-dependent 93 neutrophil phagocytosis (ADNP) and antibody-dependent complement deposition or activation 94 (ADCD) in young children and adults. We saw that infants less than 5 years old were able to 95 produce antibodies with strong FcγR2A, FcγR2B, FcγR3A, and FcγR3B binding at similar levels 96 as adults, and remarkably, anti-RBD antibodies exhibited stronger ADCP and ADNP effector 97 functions in young children than in adults ( Fig. 1B and 1C, Fig. S1). 98

99
To determine cross-reactivity of the vaccine-induced humoral response against variants of 100 concerns (VOCs), we quantified antibody levels and FcR binding against Spike and RBD for six 101 different SARS-CoV-2 VOCs including wild type (WT), Alpha, Beta, Gamma, Delta, and 102 Omicron. While IgM, IgA1, and the FcR for IgA1 (FcaR) were higher in adults across the different 103 SARS-CoV-2 variants, IgG response was essentially indistinguishable between young children 104 and adults. In fact, the only exceptions were total IgG against RBD Omicron and IgG4 against 105 Spike Gamma, RBD Alpha, RBD Delta, and RBD Omicron, which were significantly increased 106 in young children (Fig. 1D). 107 108 To further characterize the capacity of the pediatric population to generate a broad SARS-CoV-2-109 specific humoral response following mRNA-1273 vaccination, we calculated a Spike and RBD 110 protein breadth score. The breadth score highlights that children less than 5 years old are able to 111 induce a humoral response as robust as adults, with a strong recognition of different VOCs while 112 IgM-and IgA-specific immunity is higher in adults (Fig. 1E). Taken together, these results show 113 specificities regarding isotypes selection between young children and adults, with overall similar 114 to enhanced antibody functionality against SARS-CoV-2 proteins in infants and toddlers less than 115 5 years old compared to adults. 116 117 When looking more broadly at antibody responses against common respiratory infections, 118 including non-SARS-CoV-2 human coronavirus (HCoV) HKU1 Spike (HKU1), HCoV-OC43 119 Spike (OC43), and Influenza haemagglutinin (HA), we see a strong age-related difference. In 120 contrast to the robust SARS-CoV-2 vaccine-induced humoral immunity across the age spectrum, 121 young children have significantly lower antibodies titers against HKU1, OC43, and HA. 122 Multivariate analysis highlights a clear separation between the two age categories distributions, as 123 attested by the Partial least squares discriminant analysis (PLS-DA) (Fig. S3A). The LASSO-124 selected features that were used to build the PLS-DA model revealed an enrichment of antibody 125 levels and FcγR binding against HKU1, OC43, and HA in adults (Fig. S3B). Co-correlates analysis 126 showed strong connections between isotypes and FcγR features against non-SARS-CoV-2 127 antigens ( Fig. S1C), all of which being enriched in older individuals. These antibody profiles in 128 adults reflect prior exposure to these respiratory viruses over their lifetime, while these young 129 children may remain naïve. Alternatively, the lower antibodies could reflect lower total antigen-130 specific humoral responses to prior infection or the non-mRNA influenza vaccine, or more rapidly 131 waning immunity in these young children.

mRNA-1273 vaccination induces lasting, cross-reactive immunity in young children 136
In order to evaluate the impact of mRNA-1273 vaccination on the evolution of humoral immunity 137 in young children, we measured antibody levels and Fc functionality prior to vaccination (V0), 138 one month after the first dose (V1), one month after the second dose (V2), six months after 139 vaccination (V6), in addition to one month after boosting (Post-boost) (Fig. 2, 3). After just one 140 vaccine dose, strong production of IgG, IgM, and IgA against Spike WT could be observed in 141 these infants ( Fig. 2A), with robust antibody binding to FcγR (Fig. 2B). Similarly, antibody 142 effector function, characterized by ADCP, ADNP and ADCD, was significantly increased at V1 143 compared to V0 (Fig. 2C). Peak antibody responses were generally observed 2 months after the 144 first dose of vaccination ( Fig. 2), as attested by the high antibody levels and functionality ( Fig.  145 2C). Of note, IgM levels started to wane after V1 ( Fig. 2A); In the setting of rising IgG and IgA1 146 titers, this supports antibody maturation with efficient class switching. Although the number of 147 individuals that was included for the post-boost analysis was low, our results highlighted a strong 148 activation of the immune system one month after boosting, especially for FcγR binding and 149 antibody-induced neutrophil activation (Fig. 2B, C). 150

151
To evaluate the ability of mRNA-1273 vaccination to elicit broadly cross-reactive antibody 152 responses and their durability over time in young children, we compared the antibody responses 153 to Spike antigens from wild type through Omicron variants at each time point (Fig. 3). Total IgG 154 responses to full-length Spike were similar across all variants. However, IgG responses to the 155 Omicron RBD were consistently lower post-vaccination for all subclasses and FcγRs (Fig. 3A, 156 3B), which is not unexpected given the large number of mutations in the RBD of Omicron in 157 comparison to other variants and is consistent with cross-reactivity seen in older individuals 13,14 . 158 To determine the breadth of antibody responses over time, breadth scores were calculated for each 159 IgG subclass and each FcγR over the time as described above (Fig. 3C). Breadth was highest for 160 IgG3, although this response did wane prior to boost. IgG2 and IgG3 responses both expanded 161 with boosting, with minimal change in IgG1 responses. FcγR binding showed similar breadth for 162 each FcγR tested, with a robust initial response, some waning in response at 6 months after 163 vaccination, and increased breadth after boosting. Again, the breadth scores highlighted a broad 164 anti-SARS-CoV-2 antibody response shortly after vaccination that wanes over 6 months, but then 165 appears to re-expand to peak levels post-boost. 166 167

Vaccination produces greater IgG3 than natural infection in young children 168
To evaluate whether natural infection induces equivalent immunity compared to vaccination, we 169 compared anti-Spike and anti-RBD titers, and Fc binding and effector function in serum collected 170 from a group of 8 children ( Table 1: mean age, 3.7 years; range: 1-5 years) one-month following 171 acute SARS-CoV-2 infection, defined as symptomatic COVID-19 confirmed by PCR or rapid 172 antigen test at the time of illness, and a second group of children one month after completion of 173 their first dose of vaccine (V1). We did not detect a significant difference in total IgG levels ( In addition to this strong and functional antibody response in young children two months after the 230 first dose of vaccination, our results showed that this pediatric population was able to maintain 231 functional humoral immunity for at least 6 months. We observed signs of efficient antibody class 232 switching 27, 28 , as IgM levels rapidly decreased 1 month after vaccination, when IgG and IgA 233 continued to be produced, in addition to increasing FcR binding and Fc-mediated functionality. 234 Moreover, the analysis of vaccine-induced humoral immunity against VOC highlighted a strong 235 and sustained antibody response over time with Alpha, Beta, Gamma and Delta, while Omicron-236 specific immunity tended to be slightly lower, as reported previously 13 . It has been hypothesized 237 that the naïve pediatric immune system facilitates the evolution and adaptation of immune response 238 to allow broader immunity against future viral exposures 29, 30 , which might explain the more robust 239 VOCs-specific antibody response in infants compared to adults. Of the IgG subclasses, though, 240 IgG3 declined the most by 6 months but responded well to boosting, highlighting the importance 241 of boosters in maintaining effective protection against SARS-CoV-2 over time. Collectively, these 242 data suggest that the vaccine can provide long-term immunoprotection against COVID-19 in 243 young children, with likely efficacy against emerging VOCs. 244 245 Interestingly, our analysis suggests that mRNA vaccine series provides superior protection than 246 viral exposure, as attested by the higher IgG3 response against different VOCs in the vaccinated 247 group. With IgG3 being the most functional IgG subclass 16,17,18 , these data show that vaccination 248 in this young population elicits a stronger and more functional humoral immune response 249 compared to natural infection. We also observed that the antibody response against Delta, which 250 is the strain that was circulating at the time of sample collection, was higher in the infected group. 251 This suggests that adapting vaccine strategies to incorporate genetic variations that appear in 252 emerging respiratory viruses will be an important strategy to maintain vaccine efficacy.

Sample collection 283
Blood was collected prior to vaccination (Pre-vaccine), one month following the first vaccination 284 (V1), one month following the second vaccination, (V2), and six months following the second 285 vaccination (V6). If a booster dose was received, blood was collected prior to receipt of the booster 286 dose (if greater than six months from first vaccination), and one month following the booster (post-287 boost). Participants could opt out of providing blood at any of the time points. Blood was collected 288 by venipuncture or by capillary microneedle device, processed for plasma, and stored at -80°C. 289 approved study were previously published [21]. 294

VOC breadth score 312
Spike and RBD protein breadth score were calculated by categorizing each antigen response as 313 positive or negative and calculating the percentage of Spike and RBD variant antigen responses 314 for each secondary (isotype or FcR) at each timepoint. We defined a positive response as six 315 standard deviations above the mean of the SARS-CoV-2-unexposed controls for the same antigen 316 and isotype or Fc receptor. 317

Statistical analysis 319
GraphPad Prism (v.9.2.0) and RStudio (v.1.3 and R v.4.0) were used to perform data analyses. We 320 calculated breadth score by categorizing each antigen response as positive or negative and 321 calculating the percentage of positive Spike and RBD variant antigen responses for each antibody 322 feature at each timepoint. We defined a positive response as six standard deviations above the 323 mean of the COVID-unexposed controls for the same antigen and isotype or Fc receptor. 324 325 Multivariate analyses to compare vaccinated adults and children were built as described 326 previously 35, 37 . Data were normalized using z-scoring, then a least absolute shrinkage and 327 selection operator (LASSO) approach was used for feature selection. For classification and 328 visualization, partial least square discriminant analysis (PLS-DA) models were performed using 329 LASSO-selected features, followed by a ten-fold cross-validation to assess model accuracy. Co-330 correlates of LASSO selected features were represented in a network format and identified using 331   corresponds to the group that has the highest antibody response. (E) Breadth score was calculated 485 by categorizing each antigen response as positive or negative, with positive response defined as 6 486 standard deviations above the mean of the COVID-unexposed controls, then calculating the 487 percentage of positive Spike and RBD variant antigen responses for each secondary. Non-488 parametric Mann-Whitney U-test was used to calculate statistical significance, followed by 489 Benjamini-Hochberg correction for multiple testing. *p < 0.05, **p < 0.01, ***p < 0.001, ****p 490 the second dose, 6 months (V6) after the first mRNA-1273 vaccination in addition to 5 month after 497 the 2 doses, as well as 1 month after boosting (post-boost). Connecting lines represent identical 498 individuals that were followed over time, and statistical differences were calculated between 2 499 consecutive timepoints. Wilcoxon signed rank test was used to calculate differences between 500 timepoints for paired data. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.    Whitney U-test was used to calculate statistical significance, followed by Benjamini-Hochberg 518 correction for multiple testing. *p < 0.05, **p < 0.01, ***p < 0.001.   (50) Race, number (%)