Optimization of non-coding regions for a non-modified mRNA COVID-19 vaccine

The CVnCoV (CureVac) mRNA vaccine for severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) was recently evaluated in a phase 2b/3 efficacy trial in humans1. CV2CoV is a second-generation mRNA vaccine containing non-modified nucleosides but with optimized non-coding regions and enhanced antigen expression. Here we report the results of a head-to-head comparison of the immunogenicity and protective efficacy of CVnCoV and CV2CoV in non-human primates. We immunized 18 cynomolgus macaques with two doses of 12 μg lipid nanoparticle-formulated CVnCoV or CV2CoV or with sham (n = 6 per group). Compared with CVnCoV, CV2CoV induced substantially higher titres of binding and neutralizing antibodies, memory B cell responses and T cell responses as well as more potent neutralizing antibody responses against SARS-CoV-2 variants, including the Delta variant. Moreover, CV2CoV was found to be comparably immunogenic to the BNT162b2 (Pfizer) vaccine in macaques. Although CVnCoV provided partial protection against SARS-CoV-2 challenge, CV2CoV afforded more robust protection with markedly lower viral loads in the upper and lower respiratory tracts. Binding and neutralizing antibody titres were correlated with protective efficacy. These data demonstrate that optimization of non-coding regions can greatly improve the immunogenicity and protective efficacy of a non-modified mRNA SARS-CoV-2 vaccine in non-human primates. CV2CoV, a second-generation mRNA COVID-19 vaccine with non-modified nucleosides but optimized non-coding regions, is demonstrated to be effective against SARS-CoV-2 challenge when tested in non-human primates.

The CVnCoV (CureVac) mRNA vaccine for severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) was recently evaluated in a phase 2b/3 efficacy trial in humans 1 . CV2CoV is a second-generation mRNA vaccine containing non-modified nucleosides but with optimized non-coding regions and enhanced antigen expression. Here we report the results of a head-to-head comparison of the immunogenicity and protective efficacy of CVnCoV and CV2CoV in non-human primates. We immunized 18 cynomolgus macaques with two doses of 12 μg lipid nanoparticle-formulated CVnCoV or CV2CoV or with sham (n = 6 per group). Compared with CVnCoV, CV2CoV induced substantially higher titres of binding and neutralizing antibodies, memory B cell responses and T cell responses as well as more potent neutralizing antibody responses against SARS-CoV-2 variants, including the Delta variant. Moreover, CV2CoV was found to be comparably immunogenic to the BNT162b2 (Pfizer) vaccine in macaques. Although CVnCoV provided partial protection against SARS-CoV-2 challenge, CV2CoV afforded more robust protection with markedly lower viral loads in the upper and lower respiratory tracts. Binding and neutralizing antibody titres were correlated with protective efficacy. These data demonstrate that optimization of non-coding regions can greatly improve the immunogenicity and protective efficacy of a non-modified mRNA SARS-CoV-2 vaccine in non-human primates.
Efficacy results in humans have recently been reported for the CVnCoV (CureVac) mRNA vaccine in the phase 2b/3 HERALD trial in a population that included multiple viral variants. In this trial, the observed vaccine efficacy against symptomatic coronavirus disease 2019 (COVID- 19) was approximately 48% and 53% in the overall study population and in a subgroup of participants 18-60 years of age, respectively 1 . CV2CoV is a second-generation mRNA vaccine that incorporates modifications of non-coding regions that were selected by empiric screening for improved antigen expression 2,3 . Both CVnCoV and CV2CoV are based on RNActive technology [4][5][6][7] and consist of non-chemically modified sequence-engineered mRNA without pseudouridine [6][7][8][9][10][11][12] . Both vaccines encode the same full-length, pre-fusion stabilized severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) spike protein 13,14 and are encapsulated in lipid nanoparticles (LNPs) with identical composition. CV2CoV has been engineered with different non-coding regions flanking the open reading frame, which have previously been shown to improve transgene expression 3 and protection against SARS-CoV-2 in ACE2-transgenic mice 2 . Specifically, CV2CoV includes 5′ untranslated region (UTR) HSD17B4 and 3′ UTR PSMB3 elements followed by a histone stem-loop motif and a poly(A) sequence ( Fig. 1 and Methods). In the present study, we make a head-to-head comparison of the immunogenicity and protective efficacy of CVnCoV and CV2CoV against SARS-CoV-2 challenge in non-human primates.

Article
We also compared the pseudovirus neutralizing antibody titres induced in macaques by two immunizations with 12 μg of CV2CoV to those induced by two immunizations with 30 μg of the Pfizer BNT162b2 clinical vaccine obtained as leftover product from pharmacies. At peak immunity at week 5, the neutralizing antibody responses induced by CV2CoV were comparable to those induced by BNT162b2 (Fig. 2e).
Most SARS-CoV-2 RBD-specific B cells reside within the memory B cell pool 19 . We used flow cytometry to assess memory B cell responses in the blood of non-human primates vaccinated with CVnCoV, CV2CoV or sham 20 . Higher numbers of RBD-specific and spike-specific memory B cells were detected in the CV2CoV-vaccinated animals as compared with those vaccinated with CVnCoV at week 6 (P = 0.022 and P = 0.0152, respectively) (Extended Data Fig. 3a, b). T cell responses were assessed by interferon γ (IFNγ) and interleukin (IL)-4 enzyme-linked immunosorbent spot (ELISPOT) assay using pooled spike peptides at week 6. IFNγ responses were detected in both groups but were higher in the CV2CoV group (P = 0.0065) (Extended Data Fig. 3c). IL-4 responses were not detectable, suggesting that CVnCoV and CV2CoV induce T helper type 1-biased responses (Extended Data Fig. 3d).

Protective efficacy
All animals were challenged at week 8 with 1.0 × 10 5 median tissue culture infectious doses (TCID 50 ) of the SARS-CoV-2 WA1/2020 strain via the intranasal and intratracheal routes. Viral loads were assessed in bronchoalveolar lavage (BAL) and nasal swab samples collected on days 1, 2, 4, 7 and 10 following challenge by quantitative PCR with reverse transcription (RT-PCR) specific for subgenomic RNA (sgRNA) 21 . The sgRNA levels in the BAL and nasal swab samples in the sham group peaked on day 2 and largely resolved by day 10. The sham controls had peak medians of 6.02 (range, 4.62-6.81) log 10 -transformed sgRNA copies per ml in the BAL and 7.35 (range, 5.84-8.09) log 10 -transformed sgRNA copies per swab in the nasal swab samples on day 2 (Fig. 3). The CVnCoV-immunized animals showed peak medians of 4.92 (range, 2.40-6.61) log 10 -transformed sgRNA copies per ml in the BAL and 6.42 (range, 4.46-7.81) log 10 -transformed sgRNA copies per swab in the nasal swab samples (Fig. 3). The CV2CoV-immunized animals exhibited peak medians of 2.90 (range, 1.70-4.64) log 10 -transformed sgRNA copies per ml in the BAL and 3.17 (range, 2.59-5.63) log 10 -transformed sgRNA copies per swab in the nasal swab samples (Fig. 3), with resolution of sgRNA levels in the BAL samples by day 2 in most animals and by day 4 in all animals. Overall, CV2CoV resulted in significantly lower peak viral loads than CVnCoV in both the BAL (P = 0.0411) and nasal swab (P = 0.0087) samples (Fig. 4a, b).
We next evaluated the immune correlates of protection. The log 10 -transformed ELISA and neutralizing antibody titres at week 6 were inversely correlated with the peak log 10 -transformed sgRNA copies per ml in the BAL samples (P = 0.0008, R = −0.7148 and P = 0.0015, R = −0.6912, respectively, by two-sided Spearman rank-correlation test) (Fig. 4c, e) and with the peak sgRNA copies per nasal swab in the nasal swab samples (P < 0.0001, R = −0.8346 and P < 0.0001, R = −0.8766, respectively, by two-sided Spearman rank-correlation test) (Fig. 4d, f). Consistent with prior observations from our laboratory and others 15,16,22 , these findings suggest that binding and neutralizing antibody titres are important correlates of protection for these SARS-CoV-2 vaccines in non-human primates. Similar correlates of protection were observed with viral loads assessed as area under the curve (Extended Data Fig. 4). Moreover, we assessed infectious virus titres by TCID 50 assay on day 2 after challenge, which showed no detectable virus in five of six animals in the CV2CoV group (Extended Data Fig. 5).
Following challenge, we observed anamnestic binding and neutralizing antibody responses in all CVnCoV-vaccinated animals and in a subset of the CV2CoV-vaccinated animals 16 (Extended Data Fig. 6). On day 10 after challenge, the animals were necropsied, and their lung tissues were evaluated by histopathology. Viral replication was largely resolved by day 10 in the animals vaccinated with CVnCoV and CV2CoV, and those with sham treatment had higher cumulative lung pathology scores 19 (CVnCoV animals compared with sham controls, P = 0.0368; CV2CoV animals compared with sham controls, P = 0.0022) (Extended Data Fig. 7a). Animals in the sham group also had more lung lobes affected (Extended Data Fig. 7b) and more extensive lung lesions, with a greater proportion of lung lobes showing evidence of interstitial inflammation, alveolar inflammatory infiltrates and type II pneumocyte hyperplasia (Extended Data Fig. 7c-h). No significant eosinophilia was observed. The pathological lesions in vaccinated animals were similar to those observed for animals in the sham group (Extended Data Fig. 7i-l) but were overall fewer in number and more focal in distribution.

Discussion
CV2CoV elicited substantially greater humoral and cellular immune responses and provided significantly improved protective efficacy against SARS-CoV-2 challenge as compared with CVnCoV in macaques. These data suggest that optimization of non-coding elements of the mRNA backbone can substantially improve the immunogenicity and protective efficacy of mRNA vaccines. Both CVnCoV and CV2CoV contain only non-modified nucleosides with no pseudouridine or derivates, and CV2CoV has previously been shown to lead to higher antigen expression than CVnCoV in cell culture 3 . The neutralizing antibody titres induced by CV2CoV were comparable in macaques to those induced by the clinical BNT162b2 vaccine, which incorporates pseudouridine. These results suggest that strategies other than nucleoside modification can also markedly improve mRNA potency. Previous studies with rodents and non-human primates have demonstrated protection by CVnCoV 2,23,24 . However, protection in macaques was primarily observed in the lower respiratory tract 23,24 . In the present study, CVnCoV provided only modest viral load reductions in BAL and nasal swab samples compared with sham controls. In contrast to CVn-CoV, CV2CoV induced >10-fold-higher neutralizing antibody responses against multiple viral variants and provided >3 log reductions in sgRNA copies per ml in BAL and >4 log reductions in sgRNA copies per swab in nasal swab samples compared with sham controls.
Previous mRNA vaccine clinical trials have demonstrated onset of protective efficacy after the first dose with improved protection after the boost immunization 25,26 . In the present study, the prime immunization with CV2CoV induced binding and neutralizing antibodies in all macaques by week 2, and these responses had increased substantially by 1 week after the boost immunization. The neutralizing antibody titres induced by CV2CoV in this study also appear to be similar to those reported for other mRNA vaccines in macaques 27,28 . Moreover, the neutralizing antibody titres induced by BNT162b2 in our study (Fig. 2e) were comparable to those reported for BNT162b2 in a prior study 28 .
CV2CoV induced both antigen-specific memory B cell responses and T cell responses. Although the correlates of protection in this study were binding and neutralizing antibody titres 34,35 , it is likely that CD8 + T cells contribute to viral clearance in tissues 36,37 . We previously reported that depletion of CD8 + T cells partially abrogated protective efficacy against SARS-CoV-2 re-challenge in convalescent macaques 22  . c-f, Antibody correlates of protection for binding antibodies (c, d) and neutralizing antibodies (e, f). Statistical analysis was performed using the two-tailed non-parametric Mann-Whitney test, and correlation was analysed by two-sided Spearman rank-correlation test. The bars indicate median values.
Article although this study was not specifically designed as a safety study, it is worth noting that we did not observe any adverse effects following CVnCoV or CV2CoV vaccination, nor did we observe unexpected or enhanced pathology in the vaccinated animals at necropsy 40 .
In summary, our data show that optimization of non-coding regions in a SARS-CoV-2 mRNA vaccine can substantially improve its immunogenicity against multiple viral variants and can enhance its protective efficacy against SARS-CoV-2 challenge in macaques. The improved characteristics of CV2CoV over those of CVnCoV might translate into increased efficacy in humans; accordingly, clinical trials of CV2CoV are planned.

Online content
Any methods, additional references, Nature Research reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41586-021-04231-6.

IFNγ ELISPOT assay
ELISPOT plates were coated with mouse anti-human IFNγ monoclonal antibody (BD Pharmingen) at a concentration of 5 μg per well overnight at 4 °C. The plates were washed with DPBS containing 0.25% Tween-20 and were blocked with R10 medium (RPMI with 11% FBS and 1.1% penicillin-streptomycin) for 1 h at 37 °C. The S1 and S2 peptide pools (custom made, JPT Peptide Technologies) used in the assay contained peptides of 15 amino acids in length, overlapping by 11 amino acids, that spanned the protein sequence and reflect the N-terminal and C-terminal halves of the protein, respectively. The S1 and S2 peptide pools were prepared at a concentration of 2 μg per well, and 200,000 cells per well were added. The peptides and cells were incubated for 18-24 h at 37 °C. All steps following this incubation were performed at room temperature. The plates were washed with ELISPOT wash buffer and were incubated for 2 h with 1 μg ml −1 rabbit polyclonal anti-human IFNγ biotin obtained from U-Cytech. The plates were washed a second time and were then incubated for 2 h with 1 μg ml −1 streptavidin-alkaline phosphatase antibody obtained from Southern Biotech. The final wash was followed by the addition of nitro-blue tetrazolium chloride/5-bromo-4-chloro 3′ indolyl phosphate p-toludine salt (NBT/BCIP chromagen) substrate solution (ThermoFisher Scientific) for 7 min. The chromogen was discarded, and the plates were washed with water and were dried in a dim location for 24 h. The plates were then scanned and counted using an ELISPOT analyser (Immunospot).

IL-4 ELISPOT assay
ELISPOT plates precoated with monoclonal antibody against IL-4 (Mabtech) were washed and blocked. The assay was then performed as described above except that the development time with NBT/BCIP chromagen substrate solution was 12 min.
Subgenomic RT-PCR assay SARS-CoV-2 E gene sgRNA was assessed by RT-PCR using primers and probes as previously described 15,17 . A standard was generated by first synthesizing a gene fragment of the subgenomic E gene. The gene fragment was subsequently cloned into the pcDNA3.1+ expression plasmid using restriction site cloning (Integrated DNA Technologies). The insert was transcribed in vitro to RNA using the AmpliCap-Max T7 High Yield Message Maker kit (CellScript). Log dilutions of the standard were prepared for RT-PCR assays, ranging from 1 ×10 10 copies to 1 ×10 −1 copies. The viral loads were quantified from BAL fluid and nasal swab samples. RNA extraction was performed on a QIAcube HT using the IndiSpin QIAcube HT Pathogen kit according to the manufacturer's specifications (Qiagen). The standard dilutions and extracted RNA samples were reverse-transcribed using SuperScript VILO Master Mix (Invitrogen) following the cycling conditions described by the manufacturer. A Taqman custom gene expression assay (ThermoFisher Scientific) was designed using the sequences targeting the E gene sgRNA. The sequences for the custom assay were as follows: forward primer, sgLeadCoV2.Fwd: 5′-CGATCTCTTGTAGATCTGTTCTC-3′; E_ Sarbeco_R: 5′-ATATTGCAGCAGTACGCACACA-3′; E_Sarbeco_P1 (probe): 5′-VIC-ACACTAGCCATCCTTACTGCGCTTCG-MGBNFQ-3′. Reactions were carried out in duplicate for samples and standards on QuantStudio 6 and 7 Flex Real-Time PCR systems (Applied Biosystems) with the following thermal cycling conditions: initial denaturation at 95 °C for 20 s followed by 45 cycles of 95 °C for 1 s and 60 °C for 20 s. Standard curves were used to calculate the sgRNA copies per millilitre or per swab. The quantitative assay sensitivity was determined as 50 copies per millilitre or per swab.

TCID 50 assay
Vero TMPRSS2 cells (obtained from A. Creanga, NIH) were plated at 25,000 cells per well in DMEM with 10% FBS and gentamicin, and the cultures were incubated at 37 °C, 5.0% CO 2 . Medium was aspirated and replaced with 180 μl of DMEM with 2% FBS and gentamicin. Serial dilution of samples as well as positive (virus stock of known infectious titre) and negative (medium only) controls were included in each assay. The plates were incubated at 37 °C, 5.0% CO 2 , for 4 d, and the cell monolayers were visually inspected for cytopathic effects. TCID 50 was calculated using the Read-Muench formula.

Histopathology
At the time of fixation, lungs were suffused with 10% formalin to expand the alveoli. All tissues were fixed in 10% formalin and block-sectioned at 5 μm. The slides were baked for 30-60 min at 65 °C, deparaffinized in xylene, rehydrated through a series of graded ethanol to distilled water and then stained with haematoxylin and eosin. Blinded histopathological evaluation was performed by a board-certified veterinary pathologist (A.J.M.).

Statistical analyses
Statistical analyses were performed using GraphPad Prism (version 9.0) software (GraphPad Software), and comparisons between groups were performed using a two-tailed non-parametric Mann-Whitney U test. P values of less than 0.05 were considered as significant. Correlations were assessed by applying two-sided Spearman rank-correlation tests.

Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this paper.

Data availability
All data are available in the manuscript and its Supplementary Information. Source data are provided with this paper.