key: cord-0936065-v1aekhm7
authors: Corbett, Kizzmekia S.; Nason, Martha C.; Flach, Britta; Gagne, Matthew; O’ Connell, Sarah; Johnston, Timothy S.; Shah, Shruti N.; Edara, Venkata Viswanadh; Floyd, Katharine; Lai, Lilin; McDanal, Charlene; Francica, Joseph R.; Flynn, Barbara; Wu, Kai; Choi, Angela; Koch, Matthew; Abiona, Olubukola M.; Werner, Anne P.; Alvarado, Gabriela S.; Andrew, Shayne F.; Donaldson, Mitzi M.; Fintzi, Jonathan; Flebbe, Dillon R.; Lamb, Evan; Noe, Amy T.; Nurmukhambetova, Saule T.; Provost, Samantha J.; Cook, Anthony; Dodson, Alan; Faudree, Andrew; Greenhouse, Jack; Kar, Swagata; Pessaint, Laurent; Porto, Maciel; Steingrebe, Katelyn; Valentin, Daniel; Zouantcha, Serge; Bock, Kevin W.; Minai, Mahnaz; Nagata, Bianca M.; Moliva, Juan I.; van de Wetering, Renee; Boyoglu-Barnum, Seyhan; Leung, Kwanyee; Shi, Wei; Yang, Eun Sung; Zhang, Yi; Todd, John-Paul M.; Wang, Lingshu; Andersen, Hanne; Foulds, Kathryn E.; Edwards, Darin K.; Mascola, John R.; Moore, Ian N.; Lewis, Mark G.; Carfi, Andrea; Montefiori, David; Suthar, Mehul S.; McDermott, Adrian; Sullivan, Nancy J.; Roederer, Mario; Douek, Daniel C.; Graham, Barney S.; Seder, Robert A.
title: Immune Correlates of Protection by mRNA-1273 Immunization against SARS-CoV-2 Infection in Nonhuman Primates
date: 2021-04-23
journal: bioRxiv
DOI: 10.1101/2021.04.20.440647
sha: 5b44c4f529eddb610e49dbd60926b7a2710ccec0
doc_id: 936065
cord_uid: v1aekhm7

Immune correlates of protection can be used as surrogate endpoints for vaccine efficacy. The nonhuman primate (NHP) model of SARS-CoV-2 infection replicates key features of human infection and may be used to define immune correlates of protection following vaccination. Here, NHP received either no vaccine or doses ranging from 0.3 – 100 μg of mRNA-1273, a mRNA vaccine encoding the prefusion-stabilized SARS-CoV-2 spike (S-2P) protein encapsulated in a lipid nanoparticle. mRNA-1273 vaccination elicited robust circulating and mucosal antibody responses in a dose-dependent manner. Viral replication was significantly reduced in bronchoalveolar lavages and nasal swabs following SARS-CoV-2 challenge in vaccinated animals and was most strongly correlated with levels of anti-S antibody binding and neutralizing activity. Consistent with antibodies being a correlate of protection, passive transfer of vaccine-induced IgG to naïve hamsters was sufficient to mediate protection. Taken together, these data show that mRNA-1273 vaccine-induced humoral immune responses are a mechanistic correlate of protection against SARS-CoV-2 infection in NHP. One-Sentence Summary mRNA-1273 vaccine-induced antibody responses are a mechanistic correlate of protection against SARS-CoV-2 infection in NHP.

we extended the analysis to S-specific Tfh cells that express the surface marker CD40L or the 165 canonical cytokine IL-21. Most vaccinated animals had S-specific CD40L+ CD4 Tfh cell 166 responses -the magnitude of which was directly correlated with dose (p<0.001) (Fig. S5C) . A 167 direct correlation between dose and magnitude of S-specific IL-21 Tfh cell responses was also 168 observed (p=0.010). (Fig. S5D ). Consistent with previous results (13, 25, 26) , these data show 169 that mRNA-1273 induced Th1-and Tfh-skewed CD4 responses. 170

To evaluate the protective efficacy of mRNA-1273 vaccination, all animals in experiment VRC-172 20-857.4 (Fig. S1C) were challenged 4 weeks post-boost with a total dose of 8x10 5 PFU of a 173 highly pathogenic stock of SARS-CoV-2 (USA-WA1/2020) by combined intranasal and 174 intratracheal routes for upper and lower airway infection, respectively. This challenge dose was 175 chosen to induce viral loads similar to or higher than those detected in nasal secretions of humans following SARS-CoV-2 infection (27). The primary efficacy endpoint analysis used 177 subgenomic RNA (sgRNA) qRT-PCR for the nucleocapsid (N) gene (Fig. 3) . N sgRNA is the 178 most highly expressed sgRNA species as a result of discontinuous transcription and thus 179 provides greater sensitivity than the envelope (E) gene (Fig. S6 ) (28), which is most commonly 180 used in other NHP SARS-CoV-2 vaccine studies (13) to quantify replicating virus. 181

We observed a vaccine dose effect for protection against viral replication in the upper and lower 182 airway. On days 2 and 4 post challenge, there were ~2 and 5 log10 reductions in sgRNA_N in 183 BAL compared to control animals at doses of 1 µg and 30 µg, respectively (Fig. 3A) . Moreover, 184 by day 4 post-challenge, the majority of animals vaccinated with 1 µg or higher had low to 185 undetectable sgRNA_E in BAL (Fig. S6A ). By contrast, the reduction in sgRNA in nasal swabs 186 was primarily limited to animals receiving 30 µg of mRNA-1273 as compared to control animals 187 ( Fig. 3B, Fig. S5B ). These data highlight differences in immune responses required for reduction 188 in viral replication for upper and lower airway protection. Post-challenge, there was a strong 189 correlation between sgRNA in the upper and lower airways; however, the virus was more rapidly 190 cleared from the BAL compared to the nasal swab samples. Thus, there was a time-dependent 191 loss of concordance in the correlations with upper and lower airways samples ( Fig. 3C-E) , 192 suggesting distinct mechanisms for viral clearance in the two compartments. 193

Animals in each of the dose groups were assessed for detection of virus in the lung and 195 histopathology 7-or 8-days post SARS-CoV-2 challenge. In the control animals, SARS-CoV-2 infection caused moderate to severe inflammation that often involved the small airways and the 197 adjacent alveolar interstitium consistent with previous reports (29-31). Alveolar air spaces 198 occasionally contained inflammatory cell infiltrates, alveolar capillary septa were moderately 199 thickened, and moderate and diffuse type II pneumocyte hyperplasia was observed. Multiple 200 pneumocytes in the lung sections from the control group were positive for SARS-CoV-2 viral 201 antigen by immunohistochemistry (IHC) (Fig. S7 , Table S1 ). Viral antigen was detected in both 202 control animals but only sporadically across vaccinated animals in various dose groups (Table  203   S1 ). These observations show that NHP develop mild inflammation in the lung over 1 week 204 following SARS-CoV-2 infection and that vaccination limits or completely prevents 205 inflammation or detection of viral antigen in the lung tissue. 206

Following SARS-CoV-2 challenge, we assessed antibody responses in blood, BAL, and nasal 208 washes for up to 28 days to determine if there were anamnestic or primary responses to S or N 209 proteins, respectively (Fig. S8) . This analysis provides a functional immune assessment of 210 whether the virus detected in the upper and lower airways by PCR following challenge is 211 sufficient to boost vaccine-induced S-specific antibody responses or elicit primary N responses. 212

In sera, there was no post-challenge increase in S-specific (Fig. S8A ), RBD-specific (Fig. S8B) , 213 or neutralizing antibodies (Fig. S8C ) in the 3, 10, or 30 µg dose groups. In contrast, at doses 214 below 1 µg, there were increased primary S-specific (Fig. S8A ), RBD-specific (Fig. S8B) , and 215 neutralizing antibody responses (Fig. S8C ) at day 28 post-challenge compared to pre-challenge. 216

Similar primary S-specific antibody response trends were also apparent with BAL and nasal 217 wash IgG and IgA responses (Fig. S9 ). Of note, in comparing pre-challenge N-specific IgG 218 responses to those post-challenge, we only observed seroconversion in the control animals and 219 animals immunized with <3 µg of mRNA-1273 (Fig. S8D) . 220

The reduction of viral replication as determined by sgRNA coupled with limited pathology in the 221 lung and no detectable anamnestic S responses or induction of primary responses to N provide 222 three distinct measures suggesting that vaccine-elicited immune responses, particularly at high 223 doses, were protective. To understand this further, and to establish immune correlates of 224 protective immunity, we explored relationships between immune parameters and viral load. 225

Prior to conducting study VRC-20-857.4 (Fig. S1C) , we pre-specified that our analysis of a 227 potential correlate would focus primarily on the relationship between S-specific binding 228 antibodies and sgRNA levels in NS. Correlations with sgRNA levels in BAL served as an 229 important secondary analysis. The pre-defined primary hypothesis of the study was that S-230 specific IgG at 4 weeks post-boost (pre-challenge) would inversely correlate with viral 231 replication in the NS at day 2 post-challenge and that vaccine dose may not be predictive of viral 232 replication after adjustment for S-specific IgG. The hypotheses were analogous for the 233 relationship between S-specific IgG at 4 weeks post-boost and day 2 BAL sgRNA. 234 S-specific IgG at week 8 correlated strongly with sgRNA in both the NS (p=0.001) (Fig. 4G , 235 Table S2 ) and BAL (p<0.001) (Fig. 4A , Table S2 ) at day 2. As shown in Table S2 , a 1 log10 236 change in S-specific IgG corresponds to a 1 log10 change in sgRNA at day 2 in the NS, and a 0.9 237 log10 change in the sgRNA in the BAL at day 2. Once the S-specific IgG was included in a linear 238 model predicting sgRNA, including dose in the model did not substantially increase the adjusted 239 R 2 , nor was the coefficient significant (p=0.115 for NS and p=0.214 for BAL). This suggests that 240 the effect of dose on day 2 sgRNA in NS and BAL is fully captured by the adjustment for S-241 specific IgG and that, in this model, S-specific IgG meets our pre-specified criteria to be 242 considered as a correlate of sgRNA levels in NS and BAL. 243

As RBD-specific IgG, ACE2 binding inhibition, pseudovirus neutralization, and live virus 244 neutralization correlated with S-specific IgG (Fig. 2) , analyses of these as potential correlates of 245 sgRNA were also planned. All six antibody measurements were highly correlated with each 246 other ( Fig. 2) , with vaccine dose ( Table S2A , for all six antibody measurements, dose was not significantly 248 predictive of sgRNA in the BAL after adjusting for antibody levels; for NS, dose remained 249 significantly predictive after adjusting for VSV-based pseudovirus neutralization and marginally 250 significant after adjusting for live virus neutralization. This suggests that in addition to S-specific 251 IgG, RBD-specific IgG, ACE2 binding inhibition, and lentiviral-based pseudovirus 252 neutralization meet our criteria for potential correlates of protection. Furthermore, lower and 253 upper airway S-specific antibodies in the BAL and NS negatively correlated with BAL ( To assess the robustness of these findings, these analyses were repeated using logistic regression 256 to model the probability that the sgRNA was below a threshold, defined as 10,000 sgRNA copies 257 for BAL and 100,000 sgRNA copies for NS. These thresholds were chosen to be below all of the sgRNA values in the control animals and within the range of the values for the mRNA-1273-259 vaccinated animals. The results of these analyses were similar to the primary analyses done on 260 the (log) linear models. In these data, no animal with S-specific IgG >336 IU/mL had BAL 261 sgRNA >10,000 copies/mL (Fig. 4A) , and no animal with S-specific IgG >645 IU/mL had NS 262 sgRNA >100,000 copies/swab (Fig. 4G) . Last, no animals with a S-binding binding titer of >488 263 IU/mL had higher N-specific primary antibody responses post-challenge above the background 264 value at the time of challenge; consistent with that, there was a strong negative correlation 265 between pre-challenge S-specific antibodies and post-challenge N-specific antibodies (Fig. 4M ). 266

Additionally, there was limited to no lung pathology or viral antigen detection in animals with 267 <10,000 sgRNA copies/mL in BAL, providing additional evidence that mRNA-1273-vaccinated 268 animals were protected from lower airway disease. 269

We also examined the correlations between T cell responses and sgRNA and found that CD40L+ 270

Tfh cells and any Th1 response were each univariately associated with reduced sgRNA in both 271 BAL and NS. After adjustment for S-specific IgG, none of these remained significantly 272 associated with sgRNA levels in the BAL, suggesting that these T cell measures do not predict 273 sgRNA independently of the binding antibody measured in BAL. However, IL-21+ Tfh, 274 CD40L+ Tfh, and any Th1 response remained significantly predictive of sgRNA levels in NS 275 (Table S2B ) further confirming that clearance of virus from BAL and NS have distinct 276 immunological requirements (Fig. 3C-E) . 277

High titer antibody responses in blood and upper and lower airways associated with the rapid 279 control of viral load and lower airway pathology in the lung suggested that antibody was the 280 primary immunological mechanism of protection. To directly address whether vaccine-induced 281 antibody was sufficient to mediate protection, mRNA-immune NHP IgG was purified from 282 pooled sera 2 weeks post-boost of 100 µg of mRNA-1273 (13) (Fig. 5A ) and passively 283 transferred to hamsters (Fig. 5B) . Two or 10 mg of total mRNA-1273-immune NHP IgG or 10 284 mg of pre-immune NHP IgG (control) was administered to 8 Syrian hamsters/group, and 285 immediately before challenge, humoral S-specific IgG ( confidence interval, respectively. In (M), red dotted horizontal line represents 6, the maximum of all pre-challenge values across all groups, and the red dotted vertical line represents a reciprocal S-specific IgG titer of 500, above which none of the animals had day 28 N Binding titers above 6. 'r' represents Spearman's correlation coefficient, and 'p' the corresponding p-value. (B) mRNA-1273 immune NHP IgG (2 mg, yellow or 10 mg, orange) or pre-immune NHP IgG (10 mg, gray) was passively transferred to Syrian hamsters (n = 8/group) 24 hours prior to SARS-CoV-2 challenge. Twenty-three hours post-immunization, hamsters were bled to quantify circulating S-specific IgG (C) and SARS-CoV-2 pseudovirus neutralizing antibodies (D). 

An interactive web-based dashboard to track COVID-19 in 384 real time

Immunogenicity and structures of a rationally designed prefusion 386 MERS-CoV spike antigen

Cryo-EM structure of the 2019-nCoV spike in the prefusion 389 conformation

Safety and Efficacy of the BNT162b2 mRNA Covid-19 Vaccine

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

Safety and efficacy of the ChAdOx1 nCoV-19 vaccine (AZD1222) 395 against SARS-CoV-2: an interim analysis of four randomised controlled trials in Brazil

Safety and efficacy of an rAd26 and rAd5 vector-based 398 heterologous prime-boost COVID-19 vaccine: an interim analysis of a randomised 399 controlled phase 3 trial in Russia

COVID-19 vaccines: where we stand and challenges ahead

Nomenclature for immune correlates of protection after 403 vaccination

Immunogenicity of clinically relevant SARS-CoV-405 2 vaccines in nonhuman primates and humans

DNA vaccine protection against SARS-CoV-2 in rhesus macaques

Single-shot Ad26 vaccine protects against SARS-CoV-2 in rhesus 409 macaques

Evaluation of the mRNA-1273 Vaccine against SARS-CoV-2 in 411 Nonhuman Primates

Respiratory disease in rhesus macaques inoculated with SARS-CoV-413 2

Correlates of protection against SARS-CoV-2 in rhesus macaques. 415

Serum Neutralizing Activity Elicited by mRNA-1273 Vaccine

Neutralization of SARS-CoV-2 Variants B.1.429 and B.1.351

Viridot: An automated virus plaque (immunofocus) counter for 502 the measurement of serological neutralizing responses with application to dengue virus

Surrogate endpoints in clinical trials: definition and operational criteria

Mixture models for single-cell assays with applications to vaccine studies

and term infants. J Pediatr 120, 686-689 (1992