Since the S trimer is described as the primary protein responsible for inducing a protective immune response against the SARS-CoV-2 virus, first we produced a secreted and soluble form of this protein self-assembled in the trimer using Chinese Hamster Ovary (CHO) cells as previously described (26). Briefly, the transmembrane domain and the C terminal intracellular tail were removed and replaced by a T4 foldon DNA sequence and an 8xHis tag. A signal peptide sequence was added. To stabilize the prefusion structure of the S trimer in our constructs, we deactivated the original RRA furin cleavage site R by changing it to RGSA. We introduced amino-acid mutations K986P/V987P ("2P") as suggested elsewhere (12). The construct used in this work had the D614G mutation shared by most of the SARS-CoV-2 variant of concern (B.1.1.7 - Alpha, B.1.351 - Beta, B.1.617.2 - Delta, B.1.1.529 - Omicron) widely spread during the 2020-2021 pandemic (27). This S protein construct was established to form trimers predominantly folded in the prefusion conformation (26). In addition, the RBD of the S protein (aa 319-541) was produced as a recombinant product (26) and a series of peptides covering the BIP sequence were synthesized and analyzed.

Peptide sequences corresponding to the full RBM_436-507 length (72 aa) as well as shorter fragments of 20-22 amino acids (P11-P16) described in Table 1 were synthesized and analyzed. Single cysteine residues in peptides P11, P12, and P13 (486-507, 476-495 and 466-485 of S protein, respectively) were replaced with serine to avoid unwanted spontaneous formation of disulfide dimers. Peptides were prepared by microwave-assisted solid-phase peptide synthesis (MW-SPPS), cleaved from the resin and, in the case of RBM_436-507 and P12, oxidized in solution with H₂O₂ at pH 9.0. (28) Purifications were performed by flash chromatography followed by semi-preparative HPLC to achieve purity >70% (RBM_436-507 and P16) or >87% (P11-P15). Final products were characterized by analytical UHPLC coupled with ESI single quadrupole mass spectrometry and/or MALDI-ToF analysis. Analytical data and details on the synthesis and purification procedures are available as Supplementary Information .

The CD spectrum of the RBM_436-507 peptide was recorded using quartz cells of 0.1 cm path length with a JASCO J-710 CD spectropolarimeter at 25 °C. The spectrum was measured in the 260−190 nm spectral range, 1 nm bandwidth, 64 accumulations, and 100 nm/min scanning speed. The peptide was dissolved in water to a concentration of 12 μM. The secondary structure content of the peptide was predicted using the online server for protein secondary structure analyses DichroWeb (29). Input and output units and the wavelength step were θ (mdeg) and 1.0 nm, respectively.

The mean residue molar ellipticity [Θ]MR (Y-axis label) was calculated, which is defined as:

where: Θ is ellipticity in mdeg, Cr is the mean residue molar concentration, l is the cell path in cm, and Cr = (n x 1000 x Cg)/Mr

where: n is the number of peptide bonds (residue), Cg is the macromolecule concentration (g/ml), Mr is the molecular weight of the peptide.The algorithm used was CDSSTR, and the reference database was set-7 (30).

The normalized root means square deviation (NRMSD) was 0.035.

A clinical protocol was developed, submitted to and approved by the local Ethical Committees (CEAVNO, Approval # 17522) in Italy and (CECIV, approval # 04-2020) in Colombia. Whole blood (10 mL) was collected from COVID-19 patients from both Italy and Colombia. Samples were collected by arm venipuncture using dry tubes after hospitalization, and upon the patient's written informed consent, socio-demographic data and clinical manifestations were recorded. SARS-CoV-2 infection was confirmed by RT-PCR. Blood was fractionated, and sera were collected and kept frozen at -20^oC until use for serology.

A total of 30 male and female, 6-8 weeks old BALB/c mice of 20 ± 5 g of body weight were randomly selected and distributed in three groups (A, B and C) of 10 animals each. Each group was further divided into experimental (Exp) and control (Ctrl) sub-groups of five mice each and were further immunized with SARS-CoV-19 S (group A) or RDB (group B) recombinant proteins as well as with the synthetic RBM_436-507 peptide (group C). Each group of mice was immunized subcutaneously (s.c.) at the base of the tail on days 0, 20 and 40 with 20μg of each antigen diluted in 50 μL PBS and emulsified in Montanide ISA-51 (Seppic Inc., Paris, France) according to the manufacturer's recommendations. Mice were bled from submandibular veins on days 1-2 before the first and third immunizations, 20 days after the third dose and every 60 days until day 140. Whole blood (~100 μL) was collected, and sera were separated by centrifugation and stored frozen at -20°C until use for serological analyses. Animal studies were carried out at the Caucaseco Research Center in Cali (Colombia) and approved by the Animal Ethics Committee of MVDC in Colombia. Animal care, housing, and handling were performed according to institutional guidelines and following the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Nunc Maxisorp polystyrene plates were coated with Spike Trimer (Excellgene, Monthey, Switzerland) or RBD (Excellgene, Monthey, Switzerland) at 1 μg/ml in PBS pH 7.4 (50 μl/well) overnight at 4°C; peptide RBM_436-507 coating was at 2 μg/ml in Carbonate buffer, pH 9.6; 20-mers P11-P16 at 10 μg/ml in PBS, pH 7.4. After blocking for 1 hr at room temperature (RT) with PBS pH 7.4, BSA 3% (A4503 - Merck KGaA, Darmstadt, Germany), sera diluted 1/100 in PBS pH 7.4, BSA 1%, Tween-20 0.05% were incubated on the plate (50 μl/well) for 2 hours at RT. After 3 washings with PBS Tween-20 0.05% (150 μl/well), goat anti-human IgG HRP (A0293 - Merck) diluted 1:5000 in PBS BSA 1% Tween-20 0.05% was added to the plates at 50 μl/well and incubated for 2 hours. For IgM and IgA determination, goat anti-human IgM HRP conjugate (A0420 – Merck) or goat anti-human IgA HRP conjugate (A0295 - Merck) diluted 1:20,000 in PBS, BSA 1%, Tween 0.05% were added to the plates. After three washings with PBS Tween-20, 0.05%, enzymatic activity was measured at 450 nm after TMB addition (T4444 - Merck) and blocked by H₂SO₄ 1M.

The ability of anti-RBM_436-507 antibodies to inhibit the binding of ACE2 to RBD was evaluated using a modification of the SPIA commercial kit (Diametra Srl, Spello, Pg - Italy, ImmunoDiagnostic System Group). Anti-RBD antibodies were used as a positive control. Anti-N1 (20-mer linear peptide of SARS-CoV-2 nucleocapsid, aa 366-388) and anti-TT (tetanus toxoid) antibodies were used as virus-related and -unrelated negative controls. Specific antibodies were eluted from four sera with high anti-COVID-19 antibody titers using polystyrene plates coated with RBD, RBM_436-507, N1 and TT. Briefly, the plates were blocked with PBS BSA 3%, and COVID-19 sera diluted 1/50 in PBS BSA 1% Tween-20 0.05% and incubated for 2 hours at RT. Plates were washed three times with PBS Tween-20 0.05%, and bound antibodies were eluted with 200μl PBS pH 3.0 and immediately neutralized at pH 7.4 with basic phosphate buffer. The concentration of eluted antibodies was evaluated by A₂₈₀ absorbance measurement with Nanodrop, and binding to the respective antigen was confirmed by indirect ELISA. For ACE inhibition assay, anti-RBD, anti- RBM_436-507, anti-N1 and anti-TT eluted antibodies were incubated onto Diametra SPIA plates coated with recombinant RBD. Calibrator and controls were loaded as per the manufacturer's instructions. Ready-to-use ACE2 conjugated with horseradish peroxidase was then added to the wells, and plates were incubated for 90 minutes at 37°C. After washings, plates were incubated with TMB for 15 minutes and acid stop solution was added before reading the absorbance at 450 nm. Results were expressed as percentage inhibition according to the manufacturer's instruction.

Antibody titers were compared between mouse groups. A descriptive analysis was performed to evaluate differences in humoral immune responses within each group of mice. Kruskal-Wallis was performed to compare the antibody response to each protein, followed by Dunn's multiple comparison test. Results of anti-S, anti-RBD and anti- RBM_436-507 antibodies were expressed as Odd Ratio (OR) of a positive internal control set at 1.0. A p-value < 0.05 was considered statistically significant. Data were analyzed and plotted using GraphPad Prism software (version 5.01; GraphPad Software Inc, San Diego, California, USA).

To study the interaction between S and ACE2, we focused on the surface of the RBD involved in the ACE2 receptor binding, which should represent the target of the neutralizing antibodies. Our analysis of the 3D structure of the RBD-ACE2 complex showed that the large part of the RBD interacting surface, the Receptor Binding Motif (RBM), is composed of a 436-507 aa segment ( Figures 1A–C ). Since peptide synthesis technology has several advantages compared to recombinant proteins (31–34), we selected this RBM region for peptide synthesis and subsequent experimental studies. The central part of RBM_436-507 should mimic well the native-like conformation due to a disulfide bond. The peptide flanking parts should be unstructured and highly flexible both in peptides as well as within the 3D structure of the S-protein. In addition to the critical surface localization of the RBM_436-507 in the S protein, its amino acid sequence is specific to the SARS-CoV-2 and contains several predicted T-cell epitopes (33). The sequence of RBM_436-507 ( Table 1 ) was N-terminal acetylated and C-terminal amidated to avoid including terminal charged groups not present in the native protein.

The conformation of RBM_436-507 in water at pH 7 was explored by CD spectrometry ( Figure 1D ). We then evaluated the antigenic properties of this peptide. The absence of a defined minimum around 200 nm, diagnostic of random coil conformation, is compatible with a certain degree of structuration of the peptide. The secondary structure content was predicted based on the CD spectrum using the online server for protein secondary structure analyses, DichroWeb (29): 2% helix, 30% β-strand, 19% β-turn, and 49% random coil. The relatively high percentage of β-strand conformation suggests the intriguing hypothesis that RBM_436-507 peptide can partially preserve the extended conformation displayed along most of its sequence within the folded Spike protein (pdb code 6VXX) (8, 21).

As shown in Figure 2 , sera from all immunized animals tested by ELISA at 1:100 dilution, in response to the S, RBD and RBM_436-507 antigens, indicated specific IgG seroconversion after the first immunization dose. Furthermore, most of them displayed a boosting response after the second immunization dose, with the highest levels against the three proteins observed on day 40. However, while animals immunized with RBM_436-507 and RBD developed similar high level antibody profiles (3.0 to 3.5 OD), mice immunized with the S protein displayed significantly lower responses (1.0 to 2.0 OD). For RBM_436-507 and RBD, antibodies remained at high levels (>2.0 OD) after day 140, whereas antibodies against the S protein notably decreased (< 0.5 0D) during the same period. None of the control mice immunized with adjuvant alone seroconverted. The antibody titration (three-fold dilutions) using sera collected on day 140 indicated titers of 1:24,300, 1:72,900 to RBM_436-507 and RBD respectively, and 1:900 to S ( Supplementary Material , Figure S1 ).

The analysis of the homologous and cross recognition of the S, RBD and RBM_436-507 antigens by antibodies elicited upon mice immunization is shown in Figure 3 . ELISA results showed a high homologous sera reactivity but different reactivity with the other proteins/domains. Reactivity of sera diluted at 1:100 showed OD values ranging from 1.2 to 2.0 against the full-length S antigen, 3.2-3.5 to the RBD and 3.0-3-5 to the RBM_436-507 fragment. The titration of this homologous reactivity indicated that final reactivity (OD 0,2) at 1:10⁴ dilution to the S protein ( Figure 3A ), whereas at the final dilution tested (1:10⁴) the OD values were higher for RBD (OD= 2.5-3.0) and RBM (0.5-1.7) ( Figures 3B, C , respectively).

Regarding the analysis of the cross reactivity, anti-S antibodies displayed similar recognition of RBD and RBM_436-507 ( Figure 3 ), and the anti-RBD antibodies high recognition of both the S- and -RBM_436-507 proteins, although the S-protein was better recognized. In contrast, for the anti-RBM_436-507 antibodies, only two mice presented cross reactivity with end point of 1:10⁴ whereas the remaining animals of the group presented only weak reactivity at 1:100 dilution. Notably, these antibodies did not cross react with the S-protein ( Figures 3D–I ).

The final reactivity titer of the anti-S antibodies was 1:10⁴ against the S protein, and 1.8x10³ against RBD and RBM_436-507. In the case of RBD, mouse immunization elicited a vigorous antibody response ( Figure 3 ) with high optical densities even at 1:10⁴ dilution. Although reactivity to the S protein and the RBM_436-507 peptide were lower, recognition remained significant even at dilutions of 1:10⁴ and 5:10³, respectively. Sera from mice immunized with RBM_436-507 peptide also displayed high reactivity with the homologous peptide and the RBD protein; however, these sera did not react with the S protein ( Figure 3 ). We further analyzed reactivity of anti-RBM_436-507 antibodies upon solid-phase capture on ELISA plates followed by glycine elution with its homologous peptide, the RBD and the S proteins. As shown in Figure 4 , while there was significant reactivity of eluted antibodies with RBM_436-507, no recognition of the motif on the RBD and S proteins was observed.

Patient sera were first screened by ELISA using S and RBD proteins and compared to a group of pre-pandemic normal sera. IgG antibody levels higher than the 97.5^th percentile of normal sera were detected in 45% (29/64) of patient sera on S and in 53% (34/64) on RBD ( Figures 5A, B ). A strong positive correlation (p<0,0001) was observed between antibody levels for the two recombinant proteins ( Figure 5C ).

It has been shown that low pH affects spike structure, favoring a closed conformation of the trimer (34), affecting epitope exposure (16). We thus performed the ELISA assay at acidic pH, obtaining a similar level of antibodies in patient sera ( Supplementary Figures 2A, B ). Sera from COVID-19 patients and normal subjects were tested by ELISA using RBM immobilized on polystyrene plates (see Materials and Methods for details). IgG anti-RBM_436-507 higher than the 97.5^th percentile of the healthy population was detected in 21/60 (35%) of the COVID-19 patients. IgG antibody levels were significantly higher in patients than in controls (p<0.05) ( Figure 6A ) and were correlated with anti-S and anti-RBD antibody levels (p < 0.01) ( Figures 6D, E ). Anti RBM_436-507 of IgM and IgA isotype were also evaluated, with IgM anti- RBM_436-507 detected in 7/60 (11.6%) and IgA in 6/60 (10%) ( Figures 6B, C ). IgM and IgA antibody levels were not significantly different in COVID-19 patients and controls. There was coexpression of anti-RBM_436-507 Ig isotypes in COVID-19 samples ( Figure 6F ).

The analysis of the neutralizing activity of antibodies elicited by mouse immunization showed that, for mice immunized with S and RBD, the neutralization was significantly boosted after the second and third doses. Both S and RBD sera induced total neutralization after the third dose and remained high until the last test on day 115. In contrast, antibodies to RBM_436-507 reached 40% neutralization, which remained at that level until day 115 ( Figures 7A–C ).

To determine whether RBM_436-507 represents a target of neutralizing antibodies in natural conditions, we first carried out an extensive ELISA analysis of sera from both COVID-19 patients and immunized mice, and second, we compared the ACE2-RBD binding neutralization by antibodies to the whole RBD and to RBM_436-507. In the ELISA analysis of human sera (n= 100) from COVID-19 patients 35 (35%) reacted with the RBM_436-507 indicating a lower reactivity than the same sera with the S and RBD. Positive samples displayed distinct reactivity with different regions of RBM_436-507, more frequently with the N-terminal portion (P15-P16). Neutralizing activity of anti-RBM_436-507 antibodies has been evaluated by inhibition of RBD binding to ACE2, an assay considered a SARS-CoV-2 surrogate virus neutralization test (35–37). Neutralizing antibodies may bind to sequences exposed both in the closed and the open conformation of the S protein or only in the open one; most of these sequences are comprised in RBM_436-507. In contrast to human patients, mice immunized with RBM_436-507 presented good recognition of RBM_436-507 and RBD but no reactivity with S.

Since neutralizing antibodies mostly specific for RBD but also to several targeted epitopes are produced during natural infection (21, 22), in the ACE2-RBD binding neutralization assay, antibodies to the whole RBD and to RBM_436-507 were compared. In the case of humans with confirmed COVID-19 infection, sera positive to RBM_436-507 were tested using the 20-mer overlapping peptides covering the entire RBM sequence ( Table 1 ). As shown in Figure 8 , immune response mainly targets the N terminal domain (P15-P16) rather than the C-terminal part (P11-P12). To evaluate the ability of antibodies to RBD or RBM_436-507 sequences to block ACE2 binding to RBD, specific anti-RBD and anti-RBM_436-507 antibodies were eluted from COVID-19 positive sera using antigen-coated wells and incubated with labeled ACE2 on solid-phase RBD. Anti-RBD antibodies eluted from 4 COVID-19 sera inhibited the binding of labeled ACE2 to solid-phase RBD ( Figure 9 ). Anti-RBM_436-507 antibodies from 2 out of 4 sera displayed some inhibition, higher than anti-N1 and anti-TT control antibodies.