Neutralizing antibody-independent immunity to SARS-CoV-2 in Syrian hamsters and human ACE-2 transgenic mice immunized with a RBD/Nucleocapsid fusion protein

The nucleocapsid (N) and the receptor binding domain (RBD) of the Spike (S) proteins elicit robust antibody and T cell responses either in vaccinated or COVID-19 convalescent individuals. We generated a chimeric protein that comprises the sequences of RBD from S and N antigens (SpiN). SpiN was highly immunogenic and elicited a strong IFN γ response from T cells and high levels of antibodies to the inactivated virus, but no neutralizing antibodies (nAb). Importantly, hamsters and the human Angiotensin Convertase Enzyme-2-transgenic mice immunized with SpiN were highly resistant to challenge with the wild type SARS-CoV-2, as indicated by viral load, clinical outcome, lung inflammation and lethality. This protective immunity was dependent of CD4 + T and CD8 + T cells, but not by transfer of antibody of vaccinated mice. Thus, our experiment provides an example T cell-mediated immunity and reinforce the concept that T cell target antigens other that the S protein maybe considered to improve SARS-CoV-2 vaccines, and eventually circumvent the immune scape by variants. The role of T cells on protective immunity to SARS-CoV-2 is poorly understood. In vitro , neutralizing antibodies (nAbs) bind to the Receptor Binding Domain (RBD) from Spike (S) protein and prevent the interaction of SARS-CoV-2 with the Angiotensin Convertase Enzyme-2 (ACE-2) and posterior host cell invasion 1 . While the levels of neutralizing antibodies (nAb) elicited by vaccination or infection are taken as the main predictors of protective immunity, a causal-effect relationship remains to be established 2-4 . Here, we report that immunized human-ACE-2 transgenic mice (K18-ACE-2) as well as hamsters, develop strong protective immunity to SARS-CoV-2 infection, even in the absence of nAbs. Hamsters and K18-ACE-2 mice immunized with a nucleocapsid (N) and RBD chimeric protein (SpiN), with no RBD conformational B cell epitopes, presented high levels of circulating anti-viral specific antibodies and T cells. Despite the high levels of antibodies specific for the N and RBD proteins, we were unable to detect nAbs in the sera of either rodent species vaccinated with SpiN. Regardless, immunization with SpiN induced a robust T cell-mediated protection to experimental challenge with the wild type (Wuhan strain) SARS-CoV-2, which was otherwise lethal to non-vaccinated mice. Importantly, depletion of either CD4 + T or CD8 + T cells from fully vaccinated mice impaired the protective immunity to SARS-CoV2 challenge. These unexpected findings of nAb-independent protective immunity elicited by N protein might be of value in improving current COVID-19 vaccines. amplify a 100bp amplicon from the E gene of SARS-CoV-2 49 . Cycling conditions were 45ºC for 15 and 95ºC for 3 min followed by 45 cycles at 95ºC for 15s and 58ºC for 60s, using Quantstudio 5 Real Time PCR system (Applied Biossystems, For viral load quantification, a standard curve based on plasmid containing the E gene sequence (SARS-CoV-2 Wuhan-Hu isolate sequence) was constructed. Serial 10-fold dilutions of plasmid that correspond to viral copies ranging × 10 were used as templates to prepare the standard Real-time PCR assays were carried out in triplicate and the resulting Ct values by plotted against copy number of viral genome.

The role of T cells on protective immunity to SARS-CoV-2 is poorly understood. In vitro, neutralizing antibodies (nAbs) bind to the Receptor Binding Domain (RBD) from Spike (S) protein and prevent the interaction of SARS-CoV-2 with the Angiotensin Convertase Enzyme-2 (ACE-2) and posterior host cell invasion 1 . While the levels of neutralizing antibodies (nAb) elicited by vaccination or infection are taken as the main predictors of protective immunity, a causal-effect relationship remains to be established [2][3][4] . Here, we report that immunized human-ACE-2 transgenic mice (K18-ACE-2) as well as hamsters, develop strong protective immunity to SARS-CoV-2 infection, even in the absence of nAbs. Hamsters and K18-ACE-2 mice immunized with a nucleocapsid (N) and RBD chimeric protein (SpiN), with no RBD conformational B cell epitopes, presented high levels of circulating anti-viral specific antibodies and T cells. Despite the high levels of antibodies specific for the N and RBD proteins, we were unable to detect nAbs in the sera of either rodent species vaccinated with SpiN. Regardless, immunization with SpiN induced a robust T cell-mediated protection to experimental challenge with the wild type (Wuhan strain) SARS-CoV-2, which was otherwise lethal to non-vaccinated mice. Importantly, depletion of either CD4 + T or CD8 + T cells from fully vaccinated mice impaired the protective immunity to SARS-CoV2 challenge. These unexpected findings of nAb-independent protective immunity elicited by N protein might be of value in improving current COVID-19 vaccines.
Since the end of 2019, over 200 million cases and four million deaths from COVID-19 have been reported worldwide. Except for the use of inactivated SARS-CoV-2, all other COVID-19 vaccines that are currently in use or in advanced stages of development, are based on the S protein 5,6 . However, positive selection of variants with non-synonymous mutations in the Spike gene has been observed 7, 8 . These SARS-CoV-2 variants have conformational changes in the RBD region of the S protein, augmenting their affinity to ACE-2 as well as the ability to escape from nAbs 9 . While maintaining fitness, these conformational changes in the Spike protein are associated with enhanced viral infectivity and spread in humans 10,11 . Indeed, the efficacy of the current vaccines targeting conformational RBD epitopes has been challenged by the emergence of new variants 10,[12][13][14] .
Evidence of the importance of T cells in mediating immunity to SARS-CoV-2 have accumulated over the last year 15 . A coordinated response of CD4 + T cells, CD8 + T cells and nAbs seems to be ideal for a favorable outcome of disease 16 . Importantly, asymptomatic patients have low, often undetectable, levels of anti-SARS-CoV-2 antibodies and nAbs. In contrast, patients that develop moderate or severe disease present intermediate and high levels of circulating nAbs [17][18][19][20][21][22][23][24] . Furthermore, multiple T cell epitopes have been identified in SARS-CoV-2 proteins, some of which present homology to polypeptides from other coronavirus that circulate in the human populations that might explain the resistance of some seronegative individuals to symptomatic COVID-19 4,25,26 . Altogether, these studies suggest an important role of effector T cells in mediating resistance to primary infection with SARS-CoV-2. However, the vast majority of vaccines developed to protect against COVID-19 are based on conformational epitopes from the S antigen that elicit nAbs. Thus, the protective function of T-cells is difficult to dissociate from nAbs. Here, we provide a model that can be used to explore the role of effector CD4 + T and CD8 + T cell mediated immunity to SARS-CoV2.
With the aim of developing a vaccine that induces a strong T cell mediated immunity, we considered the N protein together with the canonical S antigen that is employed in most 4 COVID-19 recombinant vaccines. The N protein is the most abundant SARS-CoV-2 protein expressed in the host cell cytosol 27 . It is thus likely to be presented for cytotoxic CD8 + T cells via HLA-I through the endogenous pathway. The inclusion of the RBD region shall contribute with helper T cells for B lymphocytes and antibody response to the main SARS-CoV-2 antigens, including the nAbs . In addition, both N and S proteins have been shown highly immunogenic for CD4 + T and CD8 + T cells and B lymphocytes 4,16,26,28,29 .
First, we performed in silico analysis of the N and S proteins to identify the regions that are enriched on T cell epitopes 30,31 (Extended Data Fig. 1a-d, Fig. 1a). We found that in the S protein, the RBD segment has the greatest prevalence of potential T cell epitopes. It is also shown in Fig. 1a that the N protein is highly enriched on both CD4 + T and CD8 + T cell epitopes. Next, we analyzed the N and S protein sequences of 63 and 61 isolates, respectively, categorized as the three variants of concern (VOC; Alpha-B.1.1.7, Beta-B.1.351, Gama-P.1), and two variants under monitoring (Zeta-P.2, Epsilon-B.1.427/B.1.429) of SARS-CoV-2 distributed worldwide (Extended Data Fig. 2a,b) 32,33 . The peaks with circles indicate the position of most frequent amino acid changes in the N and S proteins, and the height of the peaks indicates the frequency that these changes occur. The black and red lines below the bars representing the N and S proteins indicate each of the putative CD8 + T and CD4 + T cell epitopes, respectively (Fig.  1a). Importantly, few of these epitopes overlapped with sites of amino acid mutations that are associated with the variants (peaks with purple circles). The vast majority of the putative T cell peptides from N protein and RBD domain were conserved (Fig. 1a).
Although no data are available, the N protein is considered a candidate for a COVID-19 vaccine 6,34,35 . Since the N protein is highly enriched for putative T cell epitopes, we used this structural protein as the basis of our vaccine. Importantly, N protein is the most abundant viral protein in infected cells, being highly expressed in their cytosol 27 , and thus readily available for processing and presentation to cytotoxic T cells via HLA-I. In contrast, the S protein is directed to the host cell membrane. In addition, we included in our chimeric protein the RBD region from the S antigen, the second most abundant SARS-CoV-2 protein in the host cells. The RBD is also highly immunogenic for helper T cells 16,26,27 that upon viral infection, could rapidly promote the production of anti-S nAbs. We hypothesize that a vaccine containing multiple linear T cell epitopes would be effective against SARS-CoV-2 variants that evade the protective nAbs based on a restricted number of nonsynonymous mutations in their genome 8,10,11,13,14 .
Polyacrylamide gels (Fig. 1b) and Western blots (Fig. 1c,d) show the highly purified N (~45 kDa) and RBD (25~ kDa) proteins expressed in Escherichia coli as well as S surface antigen (~98 kDa) obtained from eukaryotic cells. The SpiN chimeric protein that contains the N protein and the RBD region from S protein has an apparent molecular weight of 70 kDa (Fig. 1b-d). The proteins were purified either on a nickel column (RBD and N) or by ion exchange chromatography (SpiN). The immunoblots were generated with either rabbit polyclonal anti-N (Fig. 1c) or anti-RBD (Fig.  1d) sera and show that SpiN protein is recognized by both antibodies. To ensure that N or RBD proteins are recognized, both by antibodies and T cells from humans, we used samples from COVID-19 convalescent and individuals that have been vaccinated with an inactivated virus vaccine (CoronaVac). The levels of circulating antibodies in sera from convalescents were variable, whereas from vaccinated individuals were more uniform. This is likely to be due to the time of serum samples collection, which varied from 2 to 8 months after viral detection by RT-5 PCR in convalescents and 1 to 2 months after the second dose in vaccinated individuals. Convalescent individuals developed a high, but variable antibody response to S (Fig. 1e) and N proteins (Fig. 1f) and low response to the unfolded RBD expressed in bacteria (Fig. 1g). Interestingly, vaccinated individuals showed a low antibody response to both N and RBD proteins, while the response to N protein was higher in most convalescent individuals (Fig. 1f). Importantly, S (Fig. 1h), N (Fig. 1i) and RBD (Fig. 1j) proteins induced a strong T cell response in most patients, consistent with the high content of putative T cell epitopes. The vaccinated individuals, but not seronegative healthy donors (HD), also showed a robust IFNγ response to the N and RBD recombinant proteins.
Next, we evaluated the immunogenicity and whether immunization with RBD, N and SpiN recombinant proteins protects mice against SARS-CoV-2 challenge. Mice were immunized with either recombinant N, RBD or SpiN by giving two intramuscular doses scheduled 21 days apart (Fig. 2a). As immunological adjuvants we used the synthetic polyinosinic-polycytidylic acid (Poly I:C) mixed with the stabilizers carboxymethylcellulose and polylysine (Poly-ICLC, Hiltonol) or a MF59-like squalene-based oil-in-water nano-emulsion (Addavax, Vac-Adx-10 InvivoGen). We have chosen these adjuvants because they were shown to induce an effective immunity to influenza that also infects humans through the respiratory tract [36][37][38] . The Poly-IC derivatives are potent activator of Toll-Like Receptor 3 (TLR3) and other cytosolic innate immune receptors that recognize viral double stranded RNA and favors T cell-mediated immunity. It has also been used in multiple clinical trials for cancer therapy [39][40][41] . The MF59-like squalene-based oil-inwater nano-emulsion formulation lacks agonists for innate sensors, but is used in a commercially available vaccine and shows a high performance 38 .
The results demonstrate that either recombinant RBD or N protein associated to Poly ICLC are highly immunogenic inducing, respectively, high levels of anti-RBD or anti-N antibodies, both in the bronchioalveolar fluid (BALF) (Fig. 2b) and sera (Fig. 2c) of vaccinated mice. We also observed a strong IFNγ response (Fig. 2d), and less so for IL-10 ( Fig. 2e), in splenocytes stimulated with these recombinant proteins. The K18-ACE-2 mice are a model of severe disease 42 , and were used to evaluate the efficacy of immunization with either N or RBD recombinant proteins associated to Poly ICLC. Our results show that immunization with either protein resulted only in partial protection to SARS-CoV-2 challenge, as indicated by body weight loss and mortality (Fig. 2f, g).
Importantly, we report that immunization with adjuvanted SpiN fusion protein induced a robust viral-specific T cell and antibody responses, which are highly efficacious in protecting against experimental challenge with the SARS-CoV-2. Sera from mice immunized with SpiN associated to Poly ICLC showed very high titers of IgG antibodies specific for RBD (1:5,000) and N proteins (1:25,000) (Fig. 3a) as well as inactivated virus (1:5,000), but low titers of anti-S antibodies (Fig.  3b). The levels of anti-RBD and anti-N antibodies were equally higher in mice immunized with SpiN adjuvanted with Addavax (Extended Data Fig. 3a). The levels of IgG anti-N (1:400) (Fig. 3c) as well as anti-SARS-CoV-2 (1:200) (Fig. 3d) were higher in sera from COVID-19 convalescent individuals than from healthy controls, but relatively low when compared to immunized mice. In contrast to immunized mice, the titers of anti-RBD were low (Fig. 3c) and anti-S (1:200) high ( Fig. 3d) in sera of convalescent individuals. The results presented in Figures 3e and 3f show the increased levels of antibodies anti-N and anti-RBD in the BALF of mice vaccinated with SpiN, respectively.
Consistent with the high expression of the N protein 27 in infected cells, antibodies from mice immunized with either N or SpiN proteins, but not with the recombinant RBD, strongly recognized UV-inactivated SARS-CoV-2 in an ELISA (Fig. 3g). Relevant to this study, we found that immunization with neither N, RBD nor SpiN chimeric protein elicited nAbs to the wild type SARS-CoV-2, contrasting with the measurable levels in sera of COVID-19 convalescent patients (Fig. 3h). In agreement with the ELISA results (Fig. 3g), sera from mice immunized with the N protein strongly reacted with paraformaldehyde fixed SARS-CoV-2 infected cells showing a diffuse expression of N protein in the cytosol, as revealed by immunofluorescence (IFA) (Fig. 3i). In contrast, the sera from mice immunized with RBD anti-serum reacted with small vesicles in the cytosol, which might be consistent with Spike protein assembly in the endoplasmic reticulum-Golgi intermediate compartment (ERGIC) 43 .The reactivity of sera from SpiNimmunized mice showed a mixed pattern consistent with a diffuse expression in the cytosol and a punctate staining, as observed with anti-N and anti-RBD antisera, respectively (Fig. 3i).
Hamsters were used in our experiments as a model of moderate COVID-19 disease 44 . The results presented in Figure 4a and b show that SpiN associated with either Poly ICLC or Addavax induced high levels of total IgG anti-N and anti-RBD in immunized hamsters. The levels of anti-RBD were relatively low in hamsters immunized with SpiN associated with Poly ICLC (Fig. 4a,b). As in SpiN immunized mice, the levels of nAbs in vaccinated hamsters were undetectable (Fig.  4c). Nevertheless, the viral load detect by RT-PCR were lower in vaccinated, as compared to unvaccinated hamsters challenged with SARS-CoV-2 (Fig. 4d). Histopathological analysis from challenged animals at 4 days post-infection (dpi) show that the lungs in the PBS group challenged with the Wuhan strain of SARS-CoV-2 revealed an accentuated diffuse alveolar wall thickening, with moderate multifocal collapse (asterisk), associated with accentuated diffuse congestion (black arrow) and mixed inflammatory infiltrate (mononuclear and polymorphonuclear cells). In the bronchial space, inflammatory cells are noted with a predominance of neutrophils associated with cellular debris (arrowhead) (Fig. 4e, left panels). In comparison, hamsters immunized with SpiN adjuvanted with Poly ICLC had mild focal congestion (black arrow) with alveolar space preservation (Fig. 4e, mid panels). In the hamsters immunized with SpiN associated with Addavax moderate multifocal congestion is noted, associated with the presence of predominantly neutrophilic inflammatory infiltrate in the bronchi associated with cellular debris (arrowhead) (Fig. 4e,

right panels).
Immunization with either SpiN associated to Poly ICLC or Addavax protected the K18-ACE-2 mice, a model of severe COVID-19, from weight loss (Fig. 5a,e) and other clinical signs of disease, such as affected motility, ruffle fur and hunching. Importantly, 100% of immunized mice survived until 11dpi, whereas all mice that received Poly ICLC or Addavax alone succumbed to infection (Fig. 5b,f). In addition, the viral RNA load both in the lungs and brain (Fig. 5c,d) were lower when comparing vaccinated (SpiN+Poly ICLC) versus non-vaccinated (Poly ICLC) controls, as indicated by RT-PCR. Similar results were obtained with mice immunized with SpiN adjuvanted with Addavax versus Addavax alone ( Fig. 5g-h). Histopathological analysis demonstrate that the lungs from Poly ICLC group, at 5 dpi, showed a diffuse interstitial pneumonia characterized by a mixed inflammatory infiltration (mononuclear and polimorphonuclear cells), accompanied by intense congestion (black arrows), intra-alveolar exudate (white arrows with black outline), hemorrhagic foci (white star) and a few areas of alveolar collapse (asterisks) (Fig 5i, top panels). In comparison, the immunized group (SpiN plus Poly ICLC) showed preservation of the pulmonary architecture, with the presence of mononuclear peribronchovascular inflammatory infiltrate (red arrows) (Fig. 5i, bottom panels). Similar results were obtained with mice immunized with SpiN adjuvanted with Addavax. Of note, the control group (Addavax only) presented a diffuse and accentuated alveolar wall thickening (red arrowhead), with congestion (black arrows) associated with predominantly mononuclear inflammatory infiltrate with the presence of neutrophils (Extended Data Fig. 3f, top panels). In contrast, the immunized mice (SpiN plus Addavax) showed preserved lung architecture (Extended Data Fig. 3f, bottom panels). Consistent with the intense inflammatory response, we found high levels of mRNA expression of chemokines (CCL2, CCL5, CXCL9 and CXCL10) as well as IL-6 and TNFα in the lungs of infected mice. In contrast, the level of type 2 cytokines (IL-4 and IL-5) was higher in the vaccinated mice (Figs. 5j,k). These results indicate that homeostatic cytokine response in the lung mucosal environment was preserved in protected mice, whereas a dramatic switch to an inflammatory reaction was observed in animals that developed severe disease.
In order to evaluate the immune response polarization induced by vaccination, we measured the levels of IgG1 and IgG2c subclasses specific for N, RBD, S proteins and inactivated SARS-CoV-2 ( Fig. 6a-d Fig. 3b,c). In mice immunized with SpiN associated to Poly ICLC, the levels of IgG2c antibodies to RBD are higher than IgG1, showing a trend to a type 1 immune response (Fig. 6a). In contrast, there is no difference in the levels of antigen specific IgG1 and IgG2c antibodies to N or inactivated SARS-CoV-2 ( Fig. 6,c). Similar levels of anti-RBD and anti-N of IgG1 and IgG2c in mice vaccinated with SpiN plus Addavax were observed (Extended Data Fig. 3). As shown for total IgG (Fig. 3b), the levels of antibodies to the S protein are very low (Fig. 6d). We also evaluated the recall T cell response in vaccinated mice, and despite the response to RBD was higher than the N, both proteins stimulated splenocytes to produce IFN-γ (Fig. 6e). In contrast, they produced low levels of IL-10 when stimulated with SARS-CoV-2 proteins (Fig. 6f). Likewise, splenocytes from mice immunized with SpiN plus Addavax produced high levels of IFNγ and low levels of IL-10 (Extended Data Fig. 3d,e). The ELISPOT results were consistent with ELISA and showed a larger number of IFN-γ-producing cells in response to RBD (Fig. 6g). In addition, flow cytometry shows that upon in vitro stimulation with proteins RBD and N, both CD4 + T and CD8 + T cells were highly activated as indicated by the expression of CD44 activation marker (Fig. 6h,i). When stimulated with soluble antigens, CD4 + T cells were shown to be the major source of IFN-γ (Fig. 6j,k). Altogether, these results indicate that Poly ICLC induced a Type I-biased, whereas the use of Addavax as adjuvant yielded a mixed Type I/II immune response.

and Extended Data
A relevant finding of this study is the demonstration that mice immunized with SpiN are highly resistant to SARS-CoV-2, even in absence of circulating nAbs. An important question is regarding the levels of neutralizing antibodies in immunized rodents soon after the challenge with SARS-CoV-2. We found that at 4 dpi in hamsters (Fig. 4c) and 5 dpi in mice (Fig. 6l) the levels of neutralizing antibodies remained low, whereas the viral load was decreased in the lungs of immunized hamsters (Fig. 4d) and mice (Fig. 5c,g). These results further indicate a nAb-independent immunity in mice immunized with SpiN. At later time-points (8-11 days postinfection), we observed high levels of nAbs in sera of mice immunized with SpiN (Fig. 6l,m) and a drop in the viral load both in the lungs (Fig. 5c,g) and brain (Fig. 5d,h). These findings suggest that the priming of RBD-specific T cells during immunization may contribute to the secretion of nAbs triggered by the SARS-CoV-2 infection. We hypothesize that T cells act holding infection in the first days post-challenge and promote the production of nAbs at later stages of infection.
Because we observed a significant induction of cellular immunity by vaccination with SpiN, we decided to investigate whether CD4 + and CD8 + T cells are essential for protective immunity against SARS-CoV-2. For that, K18-ACE-2 mice were immunized with SpiN + Poly ICLC and on days -3, -2 and -1 before challenge they were treated with anti-CD4, anti-CD8 or anti-CD4 + CD8 antibodies. As depicted on Fig. 6n,o, 50% of mice that lacks CD4 + or CD8 + T cells showed weight loss and succumbed to infection, while 100% of animals deficient in both CD4 + and CD8 + lymphocytes lose weight and died up to 8 days post infection. In order to evaluate if solely antibodies are able to protect mice against SARS-CoV-2 infection, we transferred serum from control or SpiN-immunized mice to K18-ACE-2 mice at day -1 before challenge. All animals that received serum from vaccinated mice administered present body weight loss and succumbed to SARS-CoV2 infection (Fig. 6p,q).
Thus, our results indicate that in this model, immunity to earlier stages of SARS-CoV2 infection is solely mediated by T cells. However, vaccinated mice produced extremely high levels of anti-N antibodies and their role in mediating resistance to SARS-CoV-2 cannot be ruled out. It was clear that the anti-N antibodies were uncapable of blocking in vitro invasion of host cell by SARS-CoV-2 and did not mediated protective immunity when transferred to non-immune mice. Nevertheless, the N protein was essential in this vaccine formulation, since mice vaccinated with RBD alone, remained susceptible to infection. Importantly, the N protein is the most abundant viral protein in the host cells and is likely to be a major target of CD8 + killer cells. Since the N protein is associates with the viral RNA genome and not expressed the surface membrane, it is unlikely that anti-N antibodies act by mediating antibody dependent cell cytotoxicity (ADCC) or by promoting internalization of opsonized virus or infected host cells by macrophages. It is worth mentioning that infective SARS-CoV-2 does not grow well in macrophages, and we have no evidence that anti-N antibodies promote antibody-dependent enhancement (ADE) in vitro, or enhanced in vivo SARS-CoV-2 replication promoting pathology in vaccinated mice.
In conclusion, immunization with the N and RBD fusion protein adjuvanted with either Poly-ICLC or Addavax was highly efficient in protecting against SARS-CoV-2 challenge. Altogether our results support the hypothesis that this is primarily due to CD4 + T and CD8 + T cell-mediated immunity. Since T cell responses do not rely on a single epitope, it is unlikely that non-silent point mutations will severely undermine immune-mediated resistance induced by vaccination with the SpiN chimeric protein. Hence, while not denying the importance of nAbs, the N protein and more broadly the use of multiple T cell epitopes, should be considered to improve anti-COVID-19 vaccines and for the development of vaccines that overcome SARS-CoV-2 genetic plasticity.

Methods
Blood donors and ethics statement. Blood samples were collected from vaccinated individuals (n=33), convalescent patients (n=13), and healthy controls (n=9). All individuals were between 18 and 70 years old (36±11, female:male ratio=3.2) (Extended Table I). Vaccinated individuals received two doses of Coronavac (Sinovac, China) and were sampled 27-54 days after the second dose. Convalescent individuals reported having mild COVID-19 between 24-196 days before sampling, confirmed by PCR. All individuals were briefly interviewed before sampling and consent forms were signed. This study was performed under protocols reviewed and approved by the Ethical Committees on Human Experimentation from Fundação Hospitalar do estado de Minas Gerais (FHEMIG) -CAAE: 43335821.4.0000.5119. All patients were adults and were enrolled in the study after providing written informed consent. To build the Needle plot, a "homemade" script was implemented to count the number of divergences per alignment position considering each of the translated sequences in relation to the sequences of S or N proteins from lineage B. The number of amino acid divergences in each position was normalized by the number of N or S proteins initially selected for the analysis. The R package "mutsneedle" was used to construct the figure. The dendrograms was constructed using the iqtree tool 47 , employing the "Maximum likelihood" statistical approach and the JTTDCMut+I substitution model to obtain the correlations from the N or S amino acids sequences aligned. Lineage B was defined as "outgroup" and the R ggtree package 48 was used to the tree visualization.
Plasmid constructions and recombinant antigens production. Plasmids containing sequences encoding the full length N, the RBD of Spike and the chimeric SpiN protein with codons optimized for expression in E. coli were purchased from Genscript. Competent E. coli Star™ (DE3) were transformed with the pET24 vector with N or the RBD sequences and E. coli pRARE with the pET24_with SpiN. Transformed bacteria were grown in LB medium with kanamycin (50 µg/ml) at 37°C until OD600 0.6 was reached. At this point, protein expression was induced by adding IPTG to the culture at a final concentration of 0.5 mM. The induction of expression of the three proteins was done at 37°C for 3h for N and RBB and for 18h for SpiN. The N and RBD proteins contained a histidine tag and were purified through affinity chromatography step with the Histrap HP (GE HealthCare) column following the manufacturer's instructions. After bacterial lysis, N protein was purified from the soluble fraction and RBD protein from the insoluble fraction by adding 8M urea in the buffers for solubilization. The SpiN protein, expressed without histidine tag, was purified from the soluble and insoluble fractions of the bacteria cell lisate after addition of 8M urea and through two steps of chromatography. A cation exchange chromatography with the Hitrap SP HP column (GE HealthCare) was followed by molecular exclusion with the column HiPrep 26/60 Sephacryl S-100 HR (GE HealthCare), following the manufacturer's instructions. The S protein expressed in mammalian cells was kindly provided by Dr. Leda Castilho from Universidade Federal do Rio de Janeiro.

PBMC cultures and IFNγ measurements.
Peripheral blood mononuclear cells (PBMCs) were isolated from heparinized blood by ficoll gradient. Briefly, blood layered on Ficoll-Paque plus (GE-Healthcare) were centrifuged 410 x g, 40 minutes, RT. One-million cells per well were distributed in 96-well flat-bottom plates and incubated in complete media (RPMI 1640, 10% FBS, 100 mg/ml streptomycin, 100 U/ml penicillin) with 5 µg/mL from either N, RBD and S recombinant antigen or anti-CD3 (1µg/mL) and anti-CD28 (0.5µg/mL) as positive controls. Unstimulated cells were used to assess the background production of cytokines. Culture supernatants were harvested after 72h and frozen at -80 o C until analysis. Levels of IFNγ were measured by ELISA following the manufacturer's protocol (BD, OptEIA™ Human IFNγ).

Detection of human, mouse and hamster antigen-specific antibodies.
Plates were coated overnight with 0.4 µg/well of either N, RBD, S recombinant proteins, or alternatively 10 4 PFU/well of UV-inactivated SARS-CoV-2, and blocked for 2 hours with PBS containing 2% bovine serum albumin (PBS-2% BSA) at 37 o C. Serum were serially diluted and the bronchoalveolar lavage (BALF) samples tested at 1:1 dilution in PBS-2% BSA and incubated for 1 hour at 37 o C, and then incubated with anti-human IgG-HRP antibody (Fapon), anti-hamster IgG-HRP or antimouse total IgG, IgG1, IgG2c conjugated with streptavidin-HRP (Southern Biotech). After 5 washes, plates were revealed with 1-Step ultra TMB substrate solution (Biolegend) for 15 minutes in the dark, and reaction was stopped by adding 2N H 2 SO 4 (Sigma). Plates were read in 450nm and results were expressed as raw optical density (OD). The antibody titer was determined by the sera dilution that yielded 50% of the maximum antibody reactivity to the antigen in the ELISA.

Reduction Neutralization Assay.
One day prior infection, 10 5 Vero E6 cells were seeded in Dulbecco's modified eagle media (DMEM) (Vitrocell, Brazil) with 10% Fetal Bovine Serum (FBS) to each well of a 48 wells plate. On the next day, sera samples from mice or human were heat inactivated by incubation at 56ºC during 1h on a warm bath. Samples were two-fold serially diluted (1:10 to 1:320 (v/v)) in DMEM and mixed with 100 PFUs of SARS-CoV-2 viral stock. Media only was used instead of sera sample for positive control. The mixture was incubated during 1h ta 37ºC to allow antibody binding to viral particle. Next, Vero E6 cell culture supernatant was removed and the cells were inoculated with 50µl/well of sera-virus mixture, incubated for 1h at room temperature under gently rocking to allow viral-biding to cells. Then, 1ml of pre-warmed DMEM with 2% FBS and 2% of carboxymethylcellulose (CMC) was gently added to each well and the plates were incubated at 37ºC and 5% CO 2 atmosphere during 4 days to allow viral plaque formation. Then, the cells were fixed with 4% formaldehyde solution diluted in PBS during 2 hours and stained with 1% Naphtol blue black (Sigma, USA) solution for 1 hour for plaque visualization. Neutralization activity were determined by plaque numbers reduction compared to the positive control.

Viral Quantitative Reverse Transcription Polymerase Chain Reaction.
Total RNA was extracted from homogenized mice tissues using the RNA Mini Kit (Qiagen, USA), according to protocols provided by the manufacturers. qRT-PCR was performed in 12 µL reactions using GoTaq Probe 1-step RT-qPCR System (Promega, US) according to the manufacturer's instructions, using 75 ng of total RNA per reaction. Primers and fluorescente probes were designed based on previous described diagnostic qRT-PCR protocol specific for SARS-CoV-2, which amplify a 100bp amplicon from the E gene of SARS-CoV-2 49 . Cycling conditions were 45ºC for 15 min and 95ºC for 3 min followed by 45 cycles at 95ºC for 15s and 58ºC for 60s, using Quantstudio 5 Real Time PCR system (Applied Biossystems, USA). For viral load quantification, a standard curve based on plasmid containing the E gene sequence (SARS-CoV-2 Wuhan-Hu isolate sequence) was constructed. Serial 10-fold dilutions of plasmid DNA that correspond to viral copies ranging from 2 to 2 × 10 5 were used as templates to prepare the standard curves. Real-time PCR assays were carried out in triplicate and the resulting Ct values by plotted against copy number of viral genome.
Immunization, challenge and histopathology. Hamsters and C57BL/6 or hACE2 mice received two administrations at 21 days apart, containing 10 µg of RBD, N or SpiN adjuvanted with 50 µg of Hiltonol® (Poly ICLC, supplied by Oncovir, Washington, D.C.) 39,40 , or with AddaVax TM (oil-inwater emulsion similar to MF59) in a volume of 1:1 AddaVax TM :Antigen 50 . The solution was inoculated intramuscularly in a final volume of 50 µL into each tibial muscle. Thirty days post immunization, animals were challenged intranasally with 5 x 10 4 PFU of SARS-CoV-2 (for hACE2 mice) and 10 5 PFU for hamsters. The body weight, clinical signs and survival were evaluated for 11 days post infection. For the histopathology analyses, harvested tissues were fixed in phosphate-buffered 10% formalin for seven days, embedded in paraffin, processed using a Tissue processor PT05 TS (LUPETEC, UK) tissue processor and embedded in histological paraffin (Histosec®, Sigma-Aldrich). The 4 μm thick sections were stained with hematoxylin and eosin.

CD8 + and CD4 + T cells depletion and antibody passive transfer.
K18-hACE2 mice immunized with SpiN + Poly ICLC were treated via i.p. with either 0.5mg/mouse of rat anti-mouse CD8a or CD4 mAbs (BioXCell, clone 2.43 and GK1.5, respectively), or both. Control groups received 0.2 mg/mouse of isotype control rat anti-KLH IgG (clone LTF-2, BioXCell). The treatment was carried out on days -3, -2 and -1 before challenge. Depletion was confirmed by flow cytometry analysis of whole blood before infection. For antibody passive transfer, non-immunized K18-hACE2 mice were administered with 200µL of sera from mice immunized with SpiN + Poly ICLC or with Poly ICLC only. The inoculation was performed via i.p. one day prior challenge. Antibodies' titer anti-N and RBD was confirmed by ELISA.
Measurements of mouse cytokines. Mouse splenocytes were isolated by macerating the spleen through a 100 μm pore cell strainer (Cell Strainer, BD Falcon) followed by treatment with ACK buffer for erythrocytes lysis. The number of cells was adjusted to 10 6 cells per well and then stimulated with 10 μg/mL of RBD or N. Concanavalin A (Sigma, 5 μg/mL) was used as positive control. The supernatants were collected 72 hours post-stimulation and the levels of IFNγ and IL-10 determined by and ELISA (R&D Systems). IFNγ production was also assessed by ELISPOT, in which plates (MAHAS4510 -Millipore) were coated with capture antibody anti-IFNγ (R4-6A2 -BD) and incubated overnight at 4°C. The plates were blocked and 10 6 cells were added per well, together with 10 μg/mL of RBD or N stimuli plus rIL-2 (100 U/mL) and anti-CD28 (1 μg/mL). After 20 hours of culture, the plates were washed and incubated with biotinylated antibody anti-IFN-γ (XMG1.2 -BD). The reaction occurred after the addition of streptavidin-HRP for 1 hour followed by incubation with 3,3'-Diaminobenzidine (DAB).

Cytokines and chemokines measurements by qRT-PCR.
The RNA samples isolated from the lungs of immunized and challenged mice at 5DPI were treated with DNase (Promega), and then converted into cDNA using High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems), according to manufacturer's instructions. qPCRs reactions were performed with Sybr Green PCR Master Mix (Applied Biosystems) in an ABI7500 Real Time PCR System (Applied Biosystems), under standard conditions. Primer sequences are presented in Extended Table 2. qRT-PCR data were presented as 1/ΔCT. Immunofluorescence assays. A 16-well chamber slide (ThermoFisher) was coated with 10 4 Vero E6 cells/well and incubated overnight with SARS-CoV-2 in a multiplicity of infection (M.O.I) of 10. Then, the cells were treated with Brefeldin A for 4 hours, fixed with paraformaldehyde 4% and permeabilized with PBS-P (PBS 0,5% BSA + 0,5% saponin). Later, the wells were blocked with BSA 1% and incubated with sera from mice immunized with RBD, N or SpiN. The secondary antibody Alexa Fluor 594 anti-mouse IgG (ThermoFisher) was added and the nucleus was stained with DAPI (ThermoFisher). The slides were analyzed in the confocal microscope LSM 780 Carl Zeiss AxioObserver, objective 63x oil NA 1.4. Images were processed with the software ImageJ version 2.1 for Mac.
Statistical analysis. Statistical analysis was conducted using GraphPad Prism 6.0 for Mac (GraphPad Inc, USA). First, outliers were detected with Grubbs's test and then D'Agostino-Pearson was run to verify data normality. The tests used on each data analysis are explained on figure legends. In general, comparison between the groups were performed through unpaired t-test or Mann-Whitney U test, according to data distribution. Weight measurement data were analyzed by Two-Way ANOVA, followed by Sidak's multiple comparisons test. For survival analysis, the log-rank test was used. Statistical differences were considered significant when p values ≤ 0.05.

Reporting summary. Further information on research design is available in the Nature Research
Reporting Summary linked to this paper. The blue and purple circles indicate common mutations and those observed on variants of concern, respectively. The sum of the amino acid changes for each segment (S1, RBD and S2) of the S and N proteins is also shown. The vertical black and red lines below the bars illustrating the N and S polypeptides indicate each of the putative CD4 + T and CD8 + T cell epitopes identified by in silico epitope prediction. b, SDS-PAGE of purified SpiN, RBD, N and S proteins. c, d, Western blots of purified SpiN, RBD and N proteins using rabbit polyclonal anti-N and a mouse monoclonal anti-RBD antibodies are shown in panels. e-g, The levels of IgG antibodies specific for S, N and RBD proteins found in sera from healthy controls (HC), vaccinated and convalescents, as indicated. h-j, The IFNγ response, respectively, to the S, N and RBD proteins by the PBMCs from HC, vaccinated and convalescents. The lines link the levels of IFNγ produced by PBMCs cultured in the absence and presence of antigens, as indicated. IgG antibodies measurements were analyzed through Two-way ANOVA followed by Dunn's multiple comparisons test. Statistical analysis of IFNγ production was performed using Wilcoxonmatched pairs signed rank. NS indicate that difference is not statistically significant. ** P < 0.01 and **** P < 0.0001.  Figures 2 to 6 to analyze the immune response as well as protection against SARS-CoV-2 infection. Antigen-specific IgG antibodies measured in the bronchoalveolar lavage (BALF) at 1:1 (b) and serially diluted sera from immunized mice (c). The levels of IFNγ (d) and IL-10 (e) were measured on culture supernatant of splenocytes stimulated with RBD or N antigens. f, g, Body weight and survival of mice immunized with either RBD or N associated with Poly ICLC and challenged with the Wuhan strain of SARS-CoV-2. b-g, Data are representative of two independent experiments. b-e, n=3 mice/group. f,g, n=4 mice/group. Statistical analysis of IgG measured in BALF was performed using two-tailed unpaired t-test, t=86.39 df=4. Cytokine measurements were analyzed through Two-way ANOVA followed by Tukey's multiple test, df=30 for IFNγ, df=21 for IL-10. * P < 0.05, *** P < 0.001 and **** P < 0.0001.  23 independent experiments, n=5 hamsters/group. d, Pooled data from two independent experiments, n=8, 7 and 9 for PBS, SpiN + Poly ICLC and SpiN + Addavax, respectively. Statistical analysis of PRNT 50 was performed using Kruskal-Wallis followed by Dunn's multiple comparisons test. Data of viral load quantification was analyzed through One-way ANOVA followed by Bonferroni's multiple comparisons test, df=21 * P < 0.05 and ** P < 0.01. 24 weight (a, e) survival (b, f) and viral load in the lungs (c, g) and brains (d, h) from K18-hACE2 Tg mice immunized with SpiN associated with Poly ICLC or Addavax, and respective controls that received Poly ICLC or Addavax alone, as indicated. Viral load was measured by RT-PCR. i, Histopathological sections of lungs control and vaccinated hamsters at 5 and 7 days postchallenge with SARS-CoV-2. The tissue sections were stained with either hematoxylin and analyzed for congestion, inflammatory infiltrates, hemorrhagic foci, intra-alveolar exudate and alveolar collapse at 2,5x, 5x and 20x magnification. j, k, RNA was extracted from the lungs of control (Poly ICLC) and vaccinated (SpiN+Poly ICLC) mice at 5 dpi, reversely transcribed and the expression of cytokines (j) and chemokines (k) mRNAs was quantified by qRT- PCR. a, b, e, f Pooled data from two independent experiments. a,b n=4 for Poly ICLC and n=7 for SpiN + Poly ICLC. e, f, n=4 for Addavax, n=5 for SpiN + Addavax. c, d, g-k Data are representative of two independent experiments. c, d, n=3 mice/group. g, h, n=3 for Addavax and n=4 for SpiN + Addavax. j-k, n=3 mice/group. Statistical analysis of weight measurements was performed using Two-way ANOVA, df=89 (a), df=49 (e). Survival analysis was performed with Log-rank test. qRT-PCR data was analyzed with unpaired two-tailed t tests. Data of viral quantification was analyzed using Two-way ANOVA followed by Sidak's multiple comparisons test, df=9 (c, d), df=11 (g), df=10 (h). RT-PCR analysis was performed with multiple t tests. * P < 0.05, ** P < 0.01, *** P < 0.001 and **** P < 0.0001.  (b), SARS-CoV-2 (c) and S (d) proteins in sera from mice that received adjuvant alone (black symbols in the bottom) or were immunized with SpiN protein associated with Poly ICLC. e, f The levels of IFNγ (e) or IL-10 (f) produced by splenocytes from vaccinated mice cultured in the absence (RPMI) or presence of RBD or N. Concanavalin A (ConA) was used as positive control. g, ELISPOT was used to quantify the frequency of IFNγ-producing T cells upon stimulation with either RBD or N protein. h-k, The frequency of activated CD4 + (h) and CD8 + (i) T lymphocytes was evaluated by measuring the expression of cell surface CD44, and intracellular expression of IFNγ (j,k) by flow cytometry. l,m, nAbs on 5 or 8-11 DPI are shown as PRNT 50 (l) and as percent of neutralization on 8-11 DPI (m). n,o, K18-hACE2 mice immunized with SpiN + Poly ICLC were treated with anti-CD4, anti-CD8, or anti-CD4 + anti-CD8 and then challenged with SARS-CoV-2. Body weight (n) and mortality (o) were evaluated for eleven days. p,q, Sera from SpiNimmunized mice were transferred to non-immunized K18-hACE2 mice one day before challenge with SARS-CoV-2. Body weight (p) and mortality (q) were evaluated for eleven days. a-q, Data are representative of two independent experiments, n=3-4 mice/group. Statistical analysis of cytokine measurements, flow cytometry and PRNT 50 were performed using Two-way ANOVA followed by Sidak's multiple comparisons test. e, df=24. f, df=36. g, df=18. h, i df=20. j, k df=12. * P < 0.05, ** P < 0.01, *** P < 0.001 and **** P < 0.0001.