The mRNA vaccine BNT162b2 demonstrates impaired TH1 immunogenicity in human elders in vitro and aged mice in vivo

mRNA vaccines have been key to addressing the SARS-CoV-2 pandemic but have impaired immunogenicity and durability in vulnerable older populations. We evaluated the mRNA vaccine BNT162b2 in human in vitro whole blood assays with supernatants from adult (18–50 years) and elder (≥60 years) participants measured by mass spectrometry and proximity extension assay proteomics. BNT162b2 induced increased expression of soluble proteins in adult blood (e.g., C1S, PSMC6, CPN1), but demonstrated reduced proteins in elder blood (e.g., TPM4, APOF, APOC2, CPN1, and PI16), including 30–85% lower induction of TH1-polarizing cytokines and chemokines (e.g., IFNγ, and CXCL10). Elder TH1 impairment was validated in mice in vivo and associated with impaired humoral and cellular immunogenicity. Our study demonstrates the utility of a human in vitro platform to model age-specific mRNA vaccine activity, highlights impaired TH1 immunogenicity in older adults, and provides rationale for developing enhanced mRNA vaccines with greater immunogenicity in vulnerable populations.


Abstract:
32 mRNA vaccines have been key to addressing the SARS-CoV-2 pandemic but have impaired 33 immunogenicity and durability in vulnerable older populations. We evaluated the mRNA vaccine 34 BNT162b2 in human in vitro whole blood assays with supernatants from adult (18-50 years) and 35 elder (≥60 years) participants measured by mass spectrometry and proximity extension assay 36 proteomics. BNT162b2 induced increased expression of soluble proteins in adult blood (e.g., C1S, 37 PSMC6, CPN1), but demonstrated reduced proteins in elder blood (e.g., TPM4, APOF, APOC2, 38 CPN1, and PI16), including 30-85% lower induction of TH1-polarizing cytokines and chemokines 39 (e.g., IFNγ, and CXCL10). Elder TH1 impairment was validated in mice in vivo and associated with 40 impaired humoral and cellular immunogenicity. Our study demonstrates the utility of a human in 41 vitro platform to model age-specific mRNA vaccine activity, highlights impaired TH1 42 immunogenicity in older adults, and provides rationale for developing enhanced mRNA vaccines 43 with greater immunogenicity in vulnerable populations. 44 45 (BNT162b2 stimulation slope -0.13, p = 0.04), and trends towards lower CXCL8 (BNT162b2 157 stimulation slope -0.1, p = 0.09), and CCL2 expression (BNT162b2 stimulation slope -0.07, p = 158 0.11) (Fig. 2G). Network representation of DEP pathway analyses indicated some similar 159 pathways induced in adult and elder participant samples ( Fig. 2H-I, e.g., "signaling by 160 interleukins"). Elder sample profiles had fewer proteins contributing to each pathway node and 161 an additional predominantly down regulated "immunoregulatory interactions between a 162 lymphoid and a non-lymphoid cell" node that was not observed in adult study participants. 163 Additionally, the "IL-4 and IL-13 signaling" enriched in adult samples were not observed in elder 164 samples. Overall, evaluation of human in vitro BNT162b2 stimulation identified broadly 165 dampened immune responses in elder participant samples across 2 proteomic platforms. 166 167 An additional evaluation by a targeted multiplex bead-based assay identified titratable 168 production of interleukin-6 (IL-6), CXCL8, tumor necrosis factor alpha (TNFa), and interferon 169 gamma (IFNg) in adult WBA samples (Fig. 3A). Other cytokines such as IL-17A were not induced. 170 Adult and elder responses were fold change-normalized with stimulated over paired RPMI 171 vehicle control (Fig. 3B). Evaluation of fold-induction of mRNA-stimulated over RPMI control 172 identified induction of multiple cytokines in both age groups, including CXCL10, IL-1RA, and 173 IFNg. Nevertheless, across multiple stimulation doses, elder samples had 30-59% lower IFNg, 174 42-85% lower CXCL10, and 54-85% lower IL-1RA FC induction compared to adults. Multiplex-175 quantified analytes were grouped by function (per Table S2) as TH1, TH2, TH17, or T regulatory 176 (Treg) polarizing, chemokine, hematopoiesis-supporting, and trained immunity-associated. A 177 linear modelling analysis, GEEGLM, evaluated if age interacted with each function. TH1 support 178 was significantly impaired (p = 0.027) in elders compared to adults, by an average of 7.4% less 179 in each analyte involved ( Fig. 4A-C). The other functions evaluated were not significantly 180 different (Extended data Fig. 1) indicating a predominant impairment in inducing analytes 181 capable of inducing TH1 polarization. 182 183 In vivo murine intramuscular BNT162b2 vaccination (Extended data Fig. 2A) validated 184 age-specific observations. As observed in humans 25,26 , elder mice sera displayed significantly 185 lower total immunoglobulin G (IgG), IgG2a, and IgG1 Ab immunogenicity, with lower anti-186 receptor binding domain (RBD) Ab titers than adult mice (Fig. 5A). Adult and elder mice 187 displayed waning immunity between Days (D) 42 and 210 post-prime immunization, at various 188 immunization doses (Extended data Fig. 2B-G). Mirroring human elder observations 36 , greater 189 waning of immunity was observed in 1 µg-immunized elder mice, with 63-75% more waning 190 immunity across IgG, IgG2a, and IgG1 based on the median fold change of D210 over D42 191 between age groups (Extended data Fig. 2H). A trend of 30-83% increased waning was observed 192 at other immunization doses. Ab isotypes IgG2a and IgG1, respective markers of TH1 and TH2 193 polarized immunity 57 , were induced over nonvaccinated controls (Fig. 5A). The IgG2a/IgG1 ratio 194 inferring TH1 (>1) or TH2 (<1) polarization identified a transient impairment of TH1 on D28 post-195 prime in elder mice (Fig. 5B). Ab function was inferred via sera inhibition of RBD binding to 196 recombinant human angiotensin-converting enzyme 2 (hACE2) in a surrogate virus 197 neutralization assay (sVNT), as a correlate of protection 58,59 . Elder mice had lower sVNT than 198 adult mice at multiple immunization doses (Fig. 5C). A non-linear positive correlation of D42 199 total IgG Ab with sVNT was observed in both ages, with similar model fitting (rho) (Fig. 5D). 200 Murine Ab neutralization of live Washington-1 (WA-1) SARS-CoV-2 in vitro demonstrated an 201 impaired elder response compared to adult mice (Fig. 5E). Spike peptide splenocyte stimulation 202 induced CD4 + T cell IFNg, IL-2, TNF, and dual stained IL-4 and -5 positivity, alongside CD8 + TNF 203 (Extended data Fig. 3, Table S3). Baseline population differences in CD4 + T cell populations were 204 accounted for by dividing mouse BNT162b2-immunized responses by the average of age-205 matched vehicle control immunized mice. Elder mice had significantly less fold-induction of 206 CD4 + T cell IFNg and TNF cell positivity compared to adult mice (59% and 43% lower median fold 207 induction, respectively, Fig. 5F). IL-2 was unchanged, while IL-4/5 was non-significantly trending 208 lower in elder mice (54% lower median elder FC induction). Similarly, CD8 + TNF + T cell fold 209 induction was significantly impaired in elder vs. adult mice (45% less median elder FC, Fig. 5G). 210 In vivo murine evaluation mirrored human results with age-associated impaired Ab production, 211 Ab function, class switching, and CD4 + and CD8 + CMI. 212 Discussion: 213 Herein, we have demonstrated for the first time that (a) human in vitro modeling of 214 proteomic responses to mRNA vaccines is feasible, (b) such modeling demonstrates marked 215 age-dependent differences; and (c) results in vitro can be verified in aged mice in vivo. While 216 mRNA vaccines have been crucial in combatting the SARS-CoV-2 pandemic, much remains to be 217 learned regarding mRNA vaccine-induced immune activation, and mechanisms contributing to 218 age-associated decreased immunogenicity 26,30-32 . We hypothesized that elder participants 219 would have differential immune activation post-BNT162b2 WBA stimulation skewing immune 220 responses towards distinct functional states. guided PEA-based proteomic assay validated results (Fig. 1E, F), observing a similar age-254 dependent pattern of BNT162b2-induced adult up and elder down regulation as the LC/MS 255 proteomics, but with distinct analytes. The unsupervised heatmap clustering of 4/5 adult study 256 participants (Fig. 1G) support consistent adult immune activation patterns, while lack of elder 257 clustering further demonstrating broad age-related differences (Fig. 1H). The marked 258 differences in early proteomic responses between adult and elder participants may contribute 259 to differences in BNT162b2 immunogenicity. 260 261 Analysis of the proteome in supernatants derived from BNT162b2-stimulated adult vs. 262 elder WBAs via LC/MS (Fig. 2B-D) and PEA ( Fig. 2E-F) analyses demonstrated marked age-263 dependent differences. Multiplex analysis revealed that elders had impaired BNT162b2-264 stimulated chemokine CCL4 production (Fig. 2G), though functional categorization of multiplex-265 quantified chemokines did not identify broad differences in chemokine induction (Extended 266 data Fig. 1). Significance of the single CCL4 chemokine, but not the chemokine functionally 267 grouped analysis, could indicate specific immunoregulation but requires further validation and 268 evaluation. CCL4 induction has been negatively correlated with age 13 , potentially impacting 269 monocyte and antigen presenting cell (APC) chemotaxis to injection site and lymph nodes, 270 respectively 84-87 . Network analysis of DEPs ( Fig. 2H pathway not observed in adults, "immunoregulatory interactions between a lymphoid and non-276 lymphoid cell" with predominantly impaired induction of analytes. One of the inferred genes, 277 Cytotoxic and regulatory T cell molecule (CRTAM), can support CD4 + and CD8 + T cell 278 differentiation 89 , such that downregulation in elders may reduce CMI. 279 280 A targeted multiplex assay quantifying cytokines and chemokines enabled analyses of 281 functionally grouped analytes, yielding further insight. In the in vitro WBA, BNT162b2 induced 282 concentration-dependent production of IL-6, CXCL8, TNF-α, and IFNγ ( Fig. 3A) BNT162b2-induced WBA cytokine and chemokine induction was age-dependent with 293 impaired production of IL-1RA and the TH1 polarizing 93-99 CXCL10, and IFNg in elders (Fig. 3B).

294
Expression of these analytes was consistent across each of the 3 LC/MS proteomics, PEA 295 proteomics, and multiplex platforms increasing confidence of true positives. Impaired IFNg 296 production may be critical, as blocking IFNg has impaired BNT162b2 responses in adult mice 100 . 297 Adult humans have TH1-polarized immunity after BNT162b2 and mRNA-1273 immunization 101-298 103 . This adult TH1 polarization has been partly attributed to cationic lipids, like DOTAP (1,2-299 dioleyl-3-trimethylammonium-propane chloride salt) used in cancer RNA-LNP therapies 101 . This 300 lipid differs from BNT162b2's ALC-0315 ((4-hydroxybutyl)azanediyl)bis(hexane-6,1-diyl)bis (2-301 hexyldecanoate)) 8 , and BNT162b2 benefits from the reduced cytotoxicity from ionizable lipids 8 . 302 This could impact self-adjuvanticity between formulations, and requires further investigation, 303 as the head groups of cationic lipids can be toxic 104 . 304 305 Impaired TH1 polarization has been observed following BNT162b2 immunization of the 306 elderly 29 , and could reflect impaired production of polarizing cytokines, impaired naïve T cell 307 frequency, reduced sensitivity to cytokines during differentiation, and/or impaired chemotaxis 308 of APCs. We hypothesized functionally grouping analytes could identify age-dependent 309 impairment or dysregulation of TH1, TH2, TH17, or Treg polarizing, chemokine, hematopoiesis-310 supporting, or trained immunity associated analytes. Interpreting individual cytokines from a 311 multiplex is difficult for polyfunctional analytes and can cause overinterpretation of results. 312 Pairing individual analysis with functionally grouped analytes can generalize changes to 313 describe broader differences and assist with analyte redundancies 105 . The 41 multiplex-314 quantified analytes were categorized by function following targeted literature searches (Table  315 S2). Additional analytes were not categorized in Table S2, as they were not measured by the 316 bead-based multiplex. Functional assignment required evidence of being a polarizing molecule, 317 not from being produced by a polarized cell. GEEGLM analysis identified a significant reduction 318 (average 7.4% across multiple analytes, p = 0.027) of TH1 polarizing analyte induction in elder 319 WBA responses (Fig. 4). To reduce the risk of overinterpretation, we employed a conservative 320 GEEGLM analysis that averaged the functionally grouped analytes, including those that were 321 not individually induced, thereby biasing towards no difference. Significantly impaired TH1 322 polarizing capacity persisted in elder samples. A lack of significance in the other 6 functions 323 could indicate no impairment for these endpoints, or sample size limitations. Analyte source 324 was not evaluated in this study, but various cells could contribute to immunosenescence. 325 Ontogeny affects dendritic cells, monocytes, natural killer, and T cells 106-108 , and additional 326 investigation is needed to identify which specific cell types contribute to elder impairment in 327 the context of mRNA vaccines. We present a decreased production of multiple analytes, 328 particularly a decrease in those polarizing towards induction of TH1 polarizing IFNg and CXCL10 (Fig. 3B), and a functional analysis identifying broad 340 impairment of TH1-polarizing analyte responses (Fig. 4)  human in vitro modeling with animal models represents a powerful approach to provide both 350 species-specific and in vivo assessments. Increased age is associated with impaired human 351 humoral immunity following BNT162b2 or mRNA-1273 vaccination 2,26,36,120 . Aged mice 352 demonstrated impaired Ab induction at all immunization doses (Fig. 5A), and waning immunity 353 was more rapid in aged mice (Extended data Fig. 2), mirroring human observations 36,121 . Indirect 354 evaluation of TH1/TH2 polarization was observed via obtaining the ratio of IgG2a and IgG1 Ab 355 isotypes, markers of TH1 and TH2 polarization, respectively 57 . The relative ratio following first 356 immunization of IgG2a/IgG1 was moderately TH2-shifted and not different between both age 357 groups, while the post-booster was TH1-shifted in adult, but not aged, mice (Fig. 5B). The timing 358 of elder mouse impaired TH1 polarization post-booster but not post-prime suggests potential 359 booster specific impairments that could impact future booster campaigns. Ab efficacy was 360 evaluated with sera neutralization capacity, an important correlate of protection 58,59,122,123 . Elder 361 mice had impaired neutralization in both sVNT (Fig. 5C) and live-virus WA-1 SARS-CoV-2 assays 362 (Fig. 5E), highlighting the vulnerability of this population in a controlled environment, and 363 mirroring age-dependent human observations of neutralization 124 . The murine model results of 364 lower elder anti-RBD Ab titer, reduced markers of TH1 polarization, and impaired neutralization 365 capacity had each matched age-specific human observations. These results validate the age 366 specificity of the murine model. sVNT significantly correlated with anti-spike IgG immunity in 367 both age groups, suggesting that Ab quantity was the important neutralization factor (Fig. 5D).

368
CMI can support humoral immunity development and induce cytotoxic immunity against infected 369 cells. Impaired TH1-polarizing analytes and Ab isotypes enable inference of polarization, but 370 directly observing T cells can provide additional insight. Stimulation of splenocytes with spike-371 specific peptide induced IFNg and TNF in CD4 + T cells, markers of TH1 polarization 125 , with 43-59% 372 less median induction of cell positivity in elder mice compared to adult mice (Fig. 5F).

373
Additionally, elder mice had significant impairments in BNT162b2-induced TNF + CD8 + T cells ( Our study features multiple strengths, including (a) use of a human WBA in vitro that is 393 replete with age-specific cellular and soluble factors that could reflect or predict vaccine 394 responses in vivo 39,132 , (b) use of three complementary proteomic approaches (mass 395 spectrometry, PEA and multiplex assay) to gain a comprehensive view of the impact of 396 BNT162b2 on the WBA proteome, and (c) validation of findings using aged vs adult mice in vivo. 397 The use of human in vitro assays is notable as such approaches enable species-and age-specific 398 modeling with individuals serving as both control and test conditions permitting paired analyses 399 of new and established/licensed vaccine formulations, thereby accelerating and de-risking 400 vaccine discovery and development 39,40 . 401 402 As with any research effort, our study has multiple limitations, including (a) grouping 403 individuals into adult (18-50Y) and elder (≥60Y), which is common 110,133-136 , but does not 404 capture the gradient of immune impairments seen in >80Y 26,135 , or >100Y 137 , (b) age-dependent 405 differences in vaccine uptake resulted in differential vaccination coverage between groups that 406 could not be controlled for, (c) the in vitro WBA assay lacks fluid flow and interaction with other 407 tissues (e.g., muscle) and may not completely reflect immune responses in vivo, (d) the WBA 408 did not define the cellular origin of the TH1 polarizing cytokine and chemokines measured, and 409 (e) immune proteins have redundancies 105 that are progressively lost during 410 immunosenescence 138 . We reduced the risk of over-interpreting the biologic significance of 411 single analytes by evaluating impaired elder induction of both individual TH1 polarizing analytes, 412 IFNγ and CXCL10 93-98 (Fig. 3B), and the analysis of functionally grouped responses (Fig. 4). We 413 further validated the observed in vitro responses in an in vivo aged mouse model (Fig. 5).

415
In summary, supernatants from adult and elder WBA (Extended data and reduced CD4 + T cell TH1 polarization (Extended data Fig. 4G). Our study has demonstrated 427 the value of a human in vitro platform to model age-specific responses to the mRNA vaccine 428 BNT162b2. mRNA vaccines are expected to continue to be essential for combatting the on-429 going coronavirus pandemic and are additionally being evaluated for a range of infectious 430 diseases (influenza, RSV, and HIV) and oncology 5,139 . Identifying critically divergent functions 431 between age groups is an essential step for optimizing the next generation of precision mRNA 432 vaccines to overcome immunosenescence.
Online Methods: Human participant consent and sample processing 3 Samples were collected from younger and middle-aged adult (18-50Y), and elderly, older adults 4 (≥60Y) study participants following informed consent, under a protocol reviewed and approved 5 by Boston Children's Hospital (BCH) Institutional Review Board (protocol # X07-05-0223). Per 6 Table S1, study participants were 41.7% (5/12) and 35.7% (5/14) female in adult and aged 7 groups, respectively. Self-reported SARS-CoV-2 infection history, at time of sampling, was <10%. 8 Age groups differed in the percent prime and boost-immunized at time of sampling: 33% of 9 adults had 1-2 vaccinations and 78.6% of elders were twice vaccinated. Peripheral blood was 10 obtained from healthy adults (18-50 years old) and elders (>60 years old). Participants were 11 excluded if they had symptoms of an infection within the last 7 days, had anti-inflammatory 12 medication, were less than 110 pounds, had donated blood or been immunized within the past 13 week, were known to be pregnant or were known to be HIV infected. Blood was drawn with 14 pyrogen-free heparin anti-coagulation used at 20 units/mL (American Pharmaceutical Partners 15 Inc. Selection of individual significantly induced proteins increases the risk of interpretation bias 100 from analyte polyfunctionality and the potential of false positives. We augmented the classical 101 approach of individual analyte interpretations from multiplex assays by additionally analyzing 102 based on functional categorization to evaluate if age significantly interacted with each function. 103 Immunosenescence could be driven by differential production of analytes polarizing towards 104 CD4 + T helper cell (TH) 1, TH2, TH17, and Treg differentiation, and those supporting chemotaxis, 105 hematopoiesis, and/or associated with secondary effects of vaccine (e.g., trained immunity, 106 nonspecific effects). TH1 polarized immune responses can trigger effective intracellular 107 pathogen responses 4 , including CD8 + T cell-mediated immunity 5 , B cell class switching 6,7 and 108 induction of TFH-like activity for effective B cell responses in the absence of TFH 8 . TH2 responses 109 can support Ab production but can bias towards IgE Ab class switching with potential age-110 dependent differences 6,9 . TH17 has been associated with B cell differentiation and class 111 switching to IgA 10-12 , with increased mucosal immunity 13 . Chemokine responses are critical for 112 mounting an effective immune response 14 , through both initial recruitment of monocytes to 113 the vaccination site, and subsequent chemotaxis of mature antigen presenting cells (APC) to the 114 draining lymph node 15-18 . Treg can restrain germinal center reactions 19,20 . Hematopoiesis-115 inducing compounds could be important immunoregulators, as impaired hematopoiesis has 116 been associated with reduced vaccine responsiveness in the elderly 21,22 . mRNA vaccines may 117 also induce trained immunity 23,24 . Impact of age on each function was evaluated through a 118 targeted multiplex cytokine and chemokine assay measuring 41 predominantly polyfunctional 119 analytes. mice, which were all female. Ear clipping identification allowed mouse tracking and a guided 146 treatment randomization to balance treatments across cages, thereby reducing variability 147 between treatment groups. 148 149 Murine intramuscular injection 150 Mice were injected with mRNA vaccine (BNT162b2) in 50 µL inoculum to the mouse's right hind 151 limb via intramuscular (IM) administration in either conscious or isoflurane-anesthetized mice. 152 A prime-boost schedule was followed, separated by 14 days. 153 154 Murine SARS-CoV-2 specific antibody evaluation 155 At 14, 28, 42, and 210 days post-prime immunization animals were anesthetized under 3% 156 isoflurane and had 100-200 µL of blood collected by retroorbital bleed into non-heparinized 157 glass capillary tubes (Drummond Cat. 1-000-1000). Prompt expelling of blood into 158 microcentrifuge tubes was followed by allowing samples to clot. Blood was centrifuged within 2 159 hours (1500g, 7.5 minutes), transferred to new microcentrifuge tubes, recentrifuged, and 160 serum was aliquoted for storage at -80 o C. 161 162 Anti-spike and anti-RBD titers were evaluated by ELISA utilizing a previously described 163 protocol 19 . In brief, flat-bottomed high-binding 96-well Corning plates (NY, catalogue 9018) 164 were coated with 25 ng per well of SARS-CoV-2 wildtype sequence of recombinant RBD 165 (GenBank MN975262.1, amino acids R319-K529) or 50 ng per well of recombinant spike 166 (GenBank MN90894, amino acids M1-Q1208) glycoprotein. These proteins were produced with 167 constructs consisting of a TwinStrepTag, an HRV3C cleavage site, and an 8XHisTag C-terminal 168 modification from Aaron G. Schmidt from the Ragon Institute, and Barney S. Graham from the 169 NIH Vaccine Research Center, respectively. Overnight incubation at 4 o C was followed by 0.05% 170 Tween 20 in PBS-wash of plates, with subsequent 1% bovine serum albumin (BSA) blocking for 171 1 hr at RT. Serum samples were initially diluted 1:100 then 4-fold serially diluted to a dilution 172 factor of 1.05E8, followed by incubation in the pre-coated plate for 2 hr at RT. Following 3 173 washes a 1 hr RT incubation with horse radish peroxidase (HRP)-conjugated goat anti-mouse 174 IgG, IgG2a, or IgG1 (purchased from Southern Biotech respective cat. 1036-05, 1081-05, 1071-175 05) was performed. Following 5 x 0.05% Tween 20 in PBS washes, RT tetramethylbenzidine 176 (TMB, BD OptEIA substrate solution from BD Biosciences) was added for 5 minutes, then 177 stopped with sulfuric acid, 2N H2SO4. Optical density (OD) was determined at 450 nm in a 178 SpectraMax iD3 microplate reader (Molecular Devices). Assignment of antibody titer was 179 calculated from the final dilution where TMB was over 3x background. Anything that did not 180 register 3x background was assigned half the initial serum dilution of 100. 181 182 Murine surrogate virus neutralization titre (sVNT) evaluation 183 Murine sera were evaluated using a previously described protocol 19 . reduce the amount of time that the spleen was within the mouse without active circulation. 220 Downstream splenocyte processing was batched with no more than 3 mice at a time to reduce 221 the amount of time that cells were without circulatory support and off of ice. Aseptic dissection 222 included care to dissect away pancreatic tissue, which otherwise can impact cell viability.  Code availability 331 Data analysis in this manuscript was performed in R and R studio, as summarized above. 332 Authors will provide specific codes for analyses and graphing to those who make a request of 333 the corresponding author.