Dissociation of the respiratory syncytial virus F protein-specific human IgG, IgA and IgM response

Human respiratory syncytial virus (RSV) is one of the most important causes of severe respiratory tract infections in early childhood. The only prophylactic protection is the neutralizing antibody, palivizumab, which targets a conformational epitope of the RSV fusion (F) protein. The F protein is generated as a F0 precursor containing two furin cleavage sites allowing excision of the P27 fragment and then gives rise to a fusion-competent version consisting of the N-terminal F2 subunit and the a C-terminal F1 subunits linked by two disulphide bonds. To investigate natural human F-specific antibody responses, F2 conferring the species-specificity of RSV, was expressed in Escherichia coli. Furthermore, the F0 protein, comprising both subunits F2 and F1, was expressed as palivizumab-reactive glycoprotein in baculovirus-infected insect cells. Six overlapping F2-derived peptides lacking secondary structure were synthesized. The analysis of IgG, IgA and IgM responses of adult subjects to native versions and denatured forms of F2 and F0 and to unfolded F2-derived peptides revealed that mainly non-conformational F epitopes, some of which represented cryptic epitopes which are not exposed on the proteins were recognized. Furthermore, we found a dissociation of IgG, IgA and IgM antibody responses to F epitopes with F2 being a major target for the F-specific IgM response. The scattered and dissociated immune response to F may explain why the natural RSV-specific antibody response is only partially protective underlining the need for vaccines focusing human antibody responses towards neutralizing RSV epitopes.


Scientific Reports
| (2021) 11:3551 | https://doi.org/10.1038/s41598-021-82893-y www.nature.com/scientificreports/ and several studies have elucidated the structure of the F protein in its pre-and post-fusion conformation [25][26][27][28][29][30] . The F protein belongs to the family of class I viral fusion proteins and is responsible for the fusion of the viral envelope and cell membrane during infection 31 . The RSV F protein is generated as a F0 precursor containing two furin cleavage sites that allow excision of the P27 fragment in the trans-Golgi network to give rise to a fusioncompetent version consisting of the N-terminal F2 subunit and the a C-terminal F1 subunits which are linked by two disulphide bonds. Much attention has been focused on the F1 subunit because it is responsible for fusion with the target cell membrane and harbours epitopes for several neutralizing monoclonal antibodies 27 . The F2 subunit has been less extensively studied although it contains epitopes which are recognized by neutralizing antibodies 30,32 . The F2 subunit is also of interest because it shows sequence variability among RSV strains from the two antigenic subgroups and especially varies between RSV strains infecting different species 33,34 . In this context it is of note that the F2 subunit has been identified as the major determinant of RSV host specificity 35 .
In this study the expression, purification and characterization of the F2 subunit containing P27 is reported. In addition, a series of overlapping peptides spanning the complete F2 subunit was produced. Furthermore, a recombinant palivizumab-reactive, soluble F protein with a mutated first furin cleavage site, deleted P27, linking F1 to F2 without transmembrane and cytoplasmic domains was expressed and purified to study the natural F-specific human antibody response regarding key isotypes (IgG, IgA and IgM). The results of our study reveal several novel aspects of the natural human antibody responses specific for the F protein of RSV.

Results
Strategy for the expression of recombinant F2-P27 and F0. For the expression of recombinant F2-P27 and F0 proteins we selected the A2 strain amino acid sequence (subgroup A: GenInfo Identifier: gi|138251 from https ://www.ncbi.nlm.nih.gov/). The deduced amino acid sequence of complete F0 protein of A2-RSV strain has been aligned with the sequences of the most variable strains from the two antigenic subgroups, A and B in Fig. 1a, b. An extended version of the F2 subunit, containing the P27 fragment and both furin cleavage sites, termed F2-P27, was expressed in E. coli. The recombinant protein which was expressed as non-glycosylated protein was designed to include amino acids 22-136 ( Fig. 1) and to contain a C-terminal hexahistidine tail to facilitate its purification by Nickel affinity chromatography.
Thus, the recombinant F2 protein was made to contain the heptad repeat region (HR3, K75-M97) (Fig. 1c) which is supposed to play an important role for the stability and conformation of the F protein 36 . Furthermore, our recombinant F2 protein contains the entire P27 fragment (F2-P27). As illustrated in the sequence alignment in Fig. 1, the recombinant F2-P27 is the most divergent segment regarding its amino acid sequence among the antigenic groups of human RSV strains 33 . Furthermore, recombinant F2-P27 contains the portion of the fusion protein which shows the lowest sequence conservation among RSV strains which are pathogenic in human and animals (e.g., cows) and accordingly was found to be responsible for host cell specificity of RSV 34,35 .
The construct for expression of recombinant F0 protein in insect cells was designed to contain a mutation of the first furin cleavage site I (RARR to QAQR) to prevent cleavage by furin proteases (Fig. 1). Furthermore, the P27 fragment with the second furin cleavage site II (amino acids 110-136, KKRKRR), the hydrophobic transmembrane anchor and the hydrophilic cytoplasmic tail (amino acids 525-574) were deleted. Recombinant F0 with the modification described above, comprising amino acids Q 26 to N 524 was expressed with a C-terminal hexahistidine tag as glycosylated protein in insect cells (Fig. 1c).
Characterization of recombinant F2-P27. Recombinant F2-P27 was soluble in physiologic buffers and migrated at 14 kDa in SDS-PAGE (Fig. 2a). Matrix-assisted laser desorption/ionization time-of-flight analysis (MALDI-TOF) shows a main peak of 14,147.9 Da for F2-P27 which is in agreement with the predicted mass of 14,150 Da by ProtParam software for the recombinant F2-P27 protein with methionine and hexahistidine-tag (Fig. 2b).
The analysis by far-UV CD showed that the spectrum of recombinant F2-P27 protein was characterized by minima at 208-210 nm and at 218-220 nm consistent with a protein resembling a mix of alpha-helical and betasheet secondary structure. We also analyzed recombinant F2-P27 at different temperatures (Fig. 2c) and found that the protein exhibited a high thermal stability. Even at a temperature of 95 °C the protein seemed to preserve its secondary structure (Fig. 2c, lower part). However the addition of 15 mM of the reducing reagent TCEP, 0.46% SDS and pre-heating (95 °C) of the protein sample caused a structural transition to a predominant random coil structure with a negative peak starting at ~ 190 nm and a broad negative band at 220-230 nm (Fig. 3a).
F2-derived synthetic peptides lack secondary structure. For the mapping of F2-specific antibody responses we synthesized six overlapping peptides of 25 amino acids length which spanned the F2-P27 sequence (Figs. 1a, 4). Figure 3b shows the CD spectra of the six overlapping peptides (P1-P6). They were characterized by minima around 195-202 nm consistent with the spectra of unfolded molecules. According to Kyte-Doolittle hydrophobicity and surface accessibility calculated by ExPASy Protscale, peptides P2 and P4 seemed to be hydrophobic and buried whereas peptides P3 and P6 displayed high surface accessibility (data not shown).

Expression, purification and characterization of palivizumab-reactive recombinant F0. Recombinant soluble F0 protein could be expressed in baculovirus infected insect cells and purified by
Nickel-affinity chromatography. Figure 4a shows the Coomassie-stained SDS-PAGE of recombinant F0 which appeared in the form of two bands of approximately 57 and 55 kDa corresponding to differently glycosylated versions of the protein. Both bands were found to be reactive with a monoclonal anti-His-tag antibody (data not shown) confirming their identity. The therapeutic human monoclonal antibody palivizumab which is derived from the mouse monoclonal antibody mAb 1129 requires the intact helix-loop-helix secondary structure of  (Fig. 4b). We found that palivizumab reacted strongly with recombinant F0 and that denaturation of recombinant F0 almost completely abolished binding of palvizumab ( Fig. 4b). No binding of palivizumab to recombinant F2-P27, F2-derived peptides or to HSA was observed (negative controls, Fig. 4b,c). Palivizumab also did not bind to synthetic peptides (P11, P12, P13) comprising the amino acids which are part of the conformational epitope of palivizumab on F0 (Fig. 4b,c).
F2-and F0-specific IgG antibodies from adult RSV-exposed subjects recognize non-conformational epitopes. Sera from adult subjects were collected at the same time of the year (i.e., October-April) to reduce the possibility of variations of RSV-specific antibody responses due to seasonal differences in RSV exposure. Figure 5, shows the IgG reactivity of 16 subjects to recombinant native F2-P27, denatured F2-P27, recombinant native F0 and denatured F0 as well as to 6 unfolded peptides spanning the sequence of F2-P27. We found that there were no significant differences in the IgG responses to native and denatured F2-P27 and between native F0 and denatured F0 (Fig. 5). IgG antibody responses to F0 were significantly higher than to F2-P27 which represents a fragment comprising only approximately 25% of F0 (Fig. 5). For the 16 patients tested, peptides P2, P3 and P5 were frequently recognized by human IgG antibodies whereas peptides P1, P4 and P6 were less frequently reactive and IgG levels were lower (Fig. 5). In this context it should be noted that a large portion of P5 was part of the P27 region which upon intracellular cleavage of F0 is removed before it appears on the surface of infected cells. P3 contains several amino acids (i.e., SNIKENKC) which are reported to be part of the antigenic site Ø 30 . Since denatured F2-P27 and F0 showed similar IgG reactivity as native F2-P27 and F0 and unfolded F2-derived peptides were frequently recognized, it seems that IgG antibodies of adult subjects react mainly with non-conformational epitopes of F0.
F2-P27 is a frequent target for IgM antibodies and contains cryptic epitopes. Next, we analyzed the IgG, IgA and IgM response of the adult subjects to native F2-P27, F0 and F2-derived peptides in more detail using sera from all 23 subjects. Regarding the IgG response we found again that the F0-specific IgG response was significantly higher than the F2-P27-specific IgG response (Fig. 6a). We noted that certain peptides such as P3, P5 and for one subject also P4 showed much higher IgG reactivity than complete recombinant F2-P27 (Fig. 6a). For example, subject 16 and 21 showed much stronger IgG reactivity to P3 than to F2-P27, subject 23 www.nature.com/scientificreports/ reacted stronger to P4 than to F2-P27 and subject 22 reacted stronger to P5 than to F2 (Supplemental Fig. 1) indicating the presence of cryptic contiguous epitopes in F2. The IgA response of the subjects was lower than the IgG response, in particular to F0. Interestingly, F2-P27 despite representing only 25% of F0 showed significantly stronger IgA reactivity than F0 (Fig. 6b). This surprising finding was even more pronounced for the IgM response. F2-P27 showed a significantly higher IgM reactivity than F0 and IgM reactivity to unfolded F2-derived peptides P2, P3, P5 and P6 was especially pronounced and significantly higher than IgM response to P1 (Fig. 6c). The IgM response was directed against P5 and P6 which overlap with the P27 region and P3 which contains amino acids of antigenic site Ø (Fig. 6c) 30 .
In order to localize peptides P1-P6 in the three-dimensional structure of the pre-fusion and post-fusion form of the F protein, the peptides were highlighted in the corresponding three-dimensional structures (Fig. 7a,b). Both forms exist as trimers with the three-fold axis aligned with the vertical axis. The atomic coordinates for P5 and P6 are missing completely in the post-fusion F protein structure and only ten residues of P6 have been determined in the pre-fusion F protein structure (brown α-helix in panel A). P1, P3 and P4 display several surface exposed regions, whereas P2 is mainly buried.

Discussion
RSV infections are a major cause of respiratory illness in young children and infant mortality 1, 2 . The RSV F protein is important for the infectivity of the virus because it mediates both binding to the receptor on host cells and membrane fusion 24,25 . Importantly, the F protein is recognized by several virus-neutralizing antibodies, of which palivizumab has been registered for passive vaccination of high risk children 21,38 . We have prepared a set of defined recombinant proteins and synthetic peptides for studying the natural human isotype (IgG, IgA, IgM) response against the F protein. Our study is the first to report the expression and purification of a F2-P27 subunit assuming secondary structure as detected by CD. F2 is important because it contains binding sites of virusneutralizing antibodies and is responsible for the species-specificity of RSV 30,35,38 . According to CD analysis, the recombinant F2-P27 protein represented a protein consisting of mixed alpha helical and beta-sheet secondary structure. However, it needs to be born in mind and is a possible limitation of our study that the recombinant F2-P27 was expressed in E. coli and therefore lacks post-translational modifications which may play a role in antibody binding. Furthermore, recombinant F2-P27 produced by us contains the P27 portion which is removed by proteolytic cleavage of the F0 precursor from the F2 subunit before it assembles with the F1 subunit in F.
In addition, we succeeded to express a recombinant F0 in insect cells which resembled the epitope recognized by palivizumab. Furthermore, a set of synthetic overlapping peptides, each containing 25 amino acids, spanning F2 and three peptides derived from the binding site of palivizumab were prepared. The first set of experiments performed with palivizumab demonstrated that the recombinant F0 contained the epitope recognized by palivizumab because denaturation of F0 results in an almost complete loss of its binding (Fig. 4c). Palivizumab also did not react with unfolded synthetic peptide P12 containing the contiguous (i.e., sequential) epitope comprising amino acids 254-277 which is part of the palivizumab epitope 22,39 . This finding is of note, because viral-like particles presenting this peptide were recognized by palivizumab and antibodies induced with the particles inhibited palivizumab and exhibited virus-neutralizing activity 40 .
When we analyzed natural IgG, IgA and IgM response of adult subjects against F2-P27, F2-derived peptides, F0 and the three F1-derived peptides, several novel findings were made. First of all, our results indicated that natural IgG antibodies were directed mainly against sequential, non-conformational epitopes of F. In fact, we found that there was no significant difference in IgG reactivity to native F2-P27 versus denatured F2-P27 and likewise, palivizumab-reactive F0 showed comparable IgG reactivity as denatured F0 which had lost palivizumab reactivity. In addition, we found that several unfolded F2-derived peptides (e.g., P2, P3, P5) were frequently recognized by human IgG antibodies. One of these peptides, P3 contains several amino acids which are part of the antigenic site Ø recognized by a RSV-neutralizing antibody 30 . However, it is quite likely that the P3-specific antibodies are not virus-neutralizing because a recent study demonstrated that the inclusion of this peptide in lipid core vaccine candidates induced a specific immune response but the induced antibodies had no virusneutralizing effects 41 . Another interesting result was that P5 was frequently recognized by human IgG antibodies although it is part of the P27 region which is removed from the F0 protein upon its maturation in the host cell before it appears on cell surface and in the infectious virus. One possible explanation for this could be that F0 is released from damaged virus-infected cells and becomes immunogenic in its immature preform. This possibility is supported by another finding: in fact, several subjects showed much stronger IgG responses to F2-derived unfolded peptides containing only sequential epitopes than to the complete F2-P27 protein. This indicates that these peptides represent cryptic epitopes which only become immunogenic for example after proteolysis or after destruction of the intact protein/virus. Especially peptide P2, which was frequently recognized by IgM > IgG antibodies seems to represent a cryptic epitope because it was largely buried in the structure of the pre-fusion   www.nature.com/scientificreports/ and post-fusion F structure (Fig. 7). This may indicate that RSV liberates fragments and/or particles of F which may act as a decoy to trap antibody responses as suggested also for the G protein 42 .
Our finding that much of the F-specific IgG response is directed sequential epitopes, some of which are cryptic, would also fit with results obtained in a study demonstrating the presence of IgG antibodies in humans which can discriminate between the pre-and post-fusion conformation of F in humans 43 . In fact, sequential epitopes, such as portions of antigenic site Ø (e.g., P3) may be accessible only in the pre-fusion but not in the post-fusion form of F or vice versa. However, it is a limitation of our study that we cannot say if the F protein produced by us, although reactive with palivizumab, assumes a correct fold because we could not compare it with the recombinant pre-fusion and post-fusion F proteins for which the three-dimensional structures have been determined [28][29][30] . In fact, the F protein produced by us is different from the crystallized post-fusion F protein which consists of a www.nature.com/scientificreports/ cleaved F2 and F1 portion subsequently assembled in mammalian cells 28 whereas in our protein the F2 and F1 portion are connected within one protein and were expressed in insect cells. Furthermore, our F protein was different from the crystallized pre-fusion F protein which was obtained by co-expression with a monoclonal antibody that stabilizes it in the pre-fusion form 30 . We also did not have antibodies available for testing our F protein which would discriminate between the pre-fusion and post-fusion F protein 30 . Accordingly, we may have missed F protein-specific conformational epitopes which are present only on the reported pre-fusion and post-fusion F proteins 45 but this does not affect our finding of sequential and cryptic F epitopes. The comparison of F-specific IgG, IgA and IgM responses provided some more interesting results. While the IgG response to F0 was significantly higher than to F2-P27, the opposite was observed for IgM. In fact, F2-P27 reacted significantly stronger with human IgM than F0 and as observed for IgG, IgM antibodies frequently reacted with unfolded peptide epitopes. An in depth analysis of the F2-P27-specific IgM response in children with different course of disease (mild versus severe) in future studies will be interesting to find out if high levels of F2-P27-specific IgM may be associated with more severe disease because of possible complement activation 44 .
In agreement with an earlier study, which reported low RSV-specific IgA responses in individuals which were experimentally infected with RSV we also found a low F-specific IgA response both against the recombinant proteins and peptides 45 . Using serum samples obtained from children who had experienced virus-triggered exacerbations of wheeze and were PCR positive for RSV we found increases of F0 and especially G-specific IgG responses approximately ten weeks after the wheezing exacerbation suggesting infection by RSV as a possible trigger factor 46 .
Finally, we analyzed the correlation of F and F-derived peptide-specific isotype responses. We found that IgG, IgA and IgM responses were dissociated in terms of magnitude and epitope recognition. Thus, it seems that the F-specific isotype responses is scattered which is different from certain other viral infections. For example, the natural rhinovirus (RV)-specific IgG, IgA and IgM was highly synchronized and directed against a N-terminal peptide of the VP1 coat protein 47 . This finding actually allowed to establish serological tests based on microarrayed VP1-derived peptides from different RV-strains for robust monitoring of RV-specific antibody responses for the diagnosis of RV-induced asthma exacerbations [48][49][50] .
Likewise, it was found that the natural HIV-specific antibody response was directed against distinct peptides of gp120 and gp41 which seem to be conserved among different HIV clades 51,52 .
In contrast, the RSV-specific isotype response was scattered similar as the allergen-specific IgE and IgG response in allergic subjects where most of the allergen-specific IgG antibodies seem to be directed against epitopes which are different from those recognized by IgE 53 . Accordingly, new types of allergy vaccines used for allergen-specific immunotherapy are designed to focus allergen-specific IgG responses against IgE epitopecontaining regions to induce a protective immune response [54][55][56] .  -helix in a). P1, P3 and P4 display several surface exposed regions, whereas P2 is mainly buried. www.nature.com/scientificreports/ Therefore, our data obtained for RSV support currently pursued strategies which suggest obtaining novel generations of improved RSV vaccines by engineering immunogens so that they can focus antibodies against neutralizing epitopes on RSV 18,19,26 .

Materials and methods
All experiments were performed in accordance with relevant guidelines and regulations and all experimental protocols were approved. (https ://www.medun iwien .ac.at/web/en/recht liche s/good-scien tific -pract ice/).   (Fig. 1). The cDNA was codon-optimized for the expression in E. coli, synthesized with the addition of a DNA coding for a hexahistidine tag at the 3′ end (Genscript, Piscataway, NJ, USA) and inserted into the 5′ NdeI and 3′ XhoI restriction sites of the pET-27b ( +) vector (ATG: biosynthetics, Merzhausen, Germany). Correct insertion of the F2 construct was confirmed by NdeI/XhoI restriction analysis of midi-prep plasmid DNA (Promega, Madison, WI, USA) followed by DNA sequencing (Eurofins MWG Operon, Ebersberg, Germany). Recombinant F2-P27 was expressed in BL21-Gold (DE3) E. coli cells (Agilent Technologies, USA) at 37 °C at 180 rpm overnight in the presence of 1 mM IPTG (isopropyl-b-d-thiogalactopyranoside). Recombinant F2-P27 protein was purified by affinity chromatography using Hi60 Ni-NTA resin (Clontech Laboratories, Inc.; Takara Biotechnology; Fitchburg, Wisconsin, US) under denaturing conditions (Qiagen, Hilden, Germany). Recombinant F2-P27 was enriched in inclusion bodies and purified as described by Gallerano et al. 57 . Pure protein fractions were pooled together and dialysed in a Slide-A-Lyzer dialysis cassette, cut-off of 3.5 kDa (Thermo Scientific, Pierce Protein Biology Products, Rockford, USA) against the storage buffer (10 mM acetic acid/natrium acetate, 0.15 M NaCl, pH 5 or ddH 2 O). His-tagged F2-P27 yielded 2.5 mg/L culture, of a more than 95% pure recombinant protein, which was soluble in physiologic buffers and migrated at 14 kDa in SDS-PAGE (Fig. 2a). Peptide synthesis and purification. Overlapping peptides with a length of 25 amino acids and an overlap of 5 aa were produced by solid phase synthesis, using the same methods, resins and instruments as described by Borochova et al. 2020. Purification and confirmation of correct peptide identity was also performed as described previously 46 .

Expression in baculovirus-infected insect cells and purification of recombinant F0. The amino
acid sequence of the complete fusion envelope protein was retrieved from UniProtKB/Swiss-Prot-Expasy: full fusion glycoprotein F0 (accession number: P03420.1). The cDNA coding for the extracellular domain of F0 protein (amino acids Q 26 to N 524) was modified as follows: A mutation was introduced in the first furin cleavage site I (RARR-QAQR) to prevent cleavage by intracellular proteases. Furthermore, the P27 fragment with the second furin cleavage site II (amino acids 110-136, KKRKRR), the hydrophobic transmembrane anchor and the hydrophilic cytoplasmatic tail (amino acids 525-574) were deleted whereas a DNA coding for a C-terminal hexahistidine tag was added. This DNA construct was codon-optimized for insect cell expression, was synthesized (ATG: biosynthetics, Merzhausen, Germany) and cloned into the BamHI/SmaI sites of the pTM1 vector, harbouring the baculoviral polyhedron promoter sequence, similar to the protocol described by Pahr et al. 60  Recombinant F0 protein was purified under denaturing conditions from insect cells as described for recombinant F2-P27. After Nickel-affinity chromatography, fractions containing pure F0 were pooled and dialysed in Spectra/Por dialysis membrane tubes, cut-off of 12-14 kDa against the storage buffer (0.2 M NaCl, 10 mM Tris, 0.1 M NaH 2 PO 4 , pH 6.5). Identity and purity of recombinant F0 and F2-P27 were evaluated by 12% Coomassie-stained SDS-PAGE followed by immunoblotting using 1:1000 diluted monoclonal mouse anti-His-epitope tag antibody (Dianova, Hamburg, Germany). Bound antibodies were detected with 1:1000 diluted alkaline phosphatase-coupled rat anti-mouse IgG antibodies (BD Bioscience/Pharmingen, Heidelberg, Germany). For determination of correct assembly of the palivizumab epitope in F0, palivizumab (Synagis, Abbvie, Maidenhead, UK) was used in ELISA experiments using a 1:5000 dilution of HRP anti-human-IgG antibodies (BD Pharmingen, San Diego, CA, USA). Protein concentrations were determined in micro-plate format by BCA Protein Assay Kit (Pierce, Rockford, IL, US). The pre-stained protein molecular weight marker (PageRuler prestained Protein Ladder Plus, Fermentas, St Leon-Rot, Germany) was used as a standard in SDS-PAGE. Circular dichroism spectroscopy. Circular dichroism (CD) spectroscopy was recorded at 25 °C using a Jasco J-180 spectropolarimeter (Japan Spectroscopic Co., Tokyo, Japan) using a high transparency 1 mm light path quartz cuvette. The CD spectra were measured at a F2-P27 protein concentration of 0.1 mg/ml and peptide www.nature.com/scientificreports/ concentrations from 0.05-0.5 mg/ml (P1 0.05 mg/ml, P2 0.3 mg/ml, P3 0.5 mg/ml, P4 0.5 mg/ml, P5 0.3 mg/ml, P6 0.5 mg/ml) in ddH 2 O. Each spectrum represented an average of 5 scans collected from 190 to 260 nm, with a resolution of 0.5 nm and at a rate of 50 nm/min. The final spectra were noise-reduced using the instrument algorithm and baseline corrected by subtracting the spectrum of a buffer blank (ddH 2 O) obtained under identical conditions. Results were expressed as the mean residue ellipticity [θ] (Deg cm 2 /dg), at a given wavelength. Temperature scans of recombinant F2-P27 subunit were performed according to a step-scan procedure, where the sample was heated stepwise at 2 °C/min from 25 to 95 °C and finally cooled down to a starting point at the same rate. Three spectra were recorded every 5 °C, and averaged. The conformational state of the F2-P27 protein was calculated using the secondary structure estimation algorithm CDSSTR with reference set number 4, optimized for the wavelength range 190-260 nm as well as algorithms SELCON3 and CONTIN using DichroWeb (http://dichr oweb.cryst .bbk.ac.uk/html/home.shtml ). Results from the CD measurements were also compared to the sequence-based predictions of the protein secondary structure performed by PsiPred (bioinf.cs.ucl.ac.uk/ psipred). CD spectrometry of the F2-P27 subunit under denaturing conditions was performed under the same instrumental settings. Native F2-P27 protein sample (10 mM acetic acid/natrium acetate, 0.15 M NaCl, pH 5) was measured and baseline corrected as described above and compared with a thermally denatured and reduced F2-P27 sample (10 mM acetic acid/natrium Acetate, 0.15 M NaCl, pH 5, 0.46% sodium dodecyl sulfate (SDS), 15 mM tris(2-carboxyethyl) phosphine hydrochloride (TCEP)) which had been heated up to 95 °C for 5 min before measurement. Baseline correction was carried out using the respective buffers used for the proteins. Data were analysed and plotted using Microsoft Excel.
Enzyme-Linked Immunosorbent Assay (ELISA) experiments. Stored residual serum samples, which had been collected during a period of the year when RSV infections are common (i.e., October to April) were pseudo-anonymized and analyzed with approval from the ethics committee of the Medical University of Vienna (EK1641-2014). Informed written consent was obtained from all subjects. Samples were from healthy adult subjects (n = 23; males/females: 7/16 age range: 21-58; mean age: 32.8 years). Please note that for the experiments, described in Fig. 5, only 16 subjects were available and analyzed. PCR testing for RSV or other respiratory viruses was not performed in the subjects when serum samples were obtained.
IgG, IgM and IgA reactivity of human sera were determined. Antigens (i.e., synthetic peptides, native F2-P27, heat-denatured and reduced F2-P27, native and heat-denatured-reduced F0) were coated at a concentration of 2 µg/ml overnight at 4°C onto ELISA plates (Microplate, 96 well, PS, half area, clear and binding; Greiner bio-one, Germany). The concentration of 2 µg/ml was established in pilot experiment to ensure excess of antigen. Coating of denatured and reduced proteins was carried out by using coating buffer (0.05 M carbonate-bicarbonate, pH 9.6, 0.46% w/v sodium dodecyl sulfate (SDS), 15 mM tris (2-carboxyethyl) phosphine hydrochloride (TCEP)) overnight (ON). Denatured protein samples were pre-heated 95 °C for 10 min before coating. For native protein samples and peptides, a coating buffer without SDS and TCEP was used (0.05 M sodium-bicarbonate, pH 9.6). After incubation, plates were washed two times with washing buffer (PBS + 0.05% Tween 20) and blocked with 2% w/v BSA (bovine serum albumin) solution (PBS + 0.05% Tween 20 and 2% w/v BSA) for five hours at room temperature (RT). Following the removal of blocking solution, a 1:50 serum dilution, in dilution buffer (PBS + 0.05% Tween 20 and 0.5% w/v BSA) was added or dilution buffer only (control wells). Plates were incubated ON at 4°C and on the next day washed five times with washing buffer and incubated at RT with appropriate antibodydilution. Bound IgG antibodies were detected with horseradish peroxidase (HRP)-conjugated anti-human-IgG antibodies, diluted in dilution buffer 1:5000 (HRP anti-human IgG; cat: 55788; BD Pharmingen, San Diego, CA, USA), after one hour incubation time at RT and a five times washing step.
IgA and IgM binding was determined with monoclonal mouse anti-human IgA or IgM antibodies with a dilution factor of 1:1000 (IgA antibody cat.: 555886, clone: G18-1; IgM antibody cat: 555856, clone: JDC-15; BD Pharmingen, San Diego, CA, USA). IgA or IgM antibody dilutions were incubated on plates for two hours at RT. After washing, the plates five times with washing buffer a secondary incubation with the HRP-conjugated sheep anti-mouse IgG antibody (NA931; GE Healthcare UK Limited, Buckinghamshire, UK), diluted 1:2000 in dilution buffer, was carried out for one hour at RT. Following the final five times washing step, the colour reaction was started by adding 50 µl/well of substrate solution as described previously 46 . The optical density (OD) values relate to the levels of antigen-specific antibodies and were measured at 405 and 492 nm on Perkin Elmer EnSpire 2300 ELISA reader device 46 . Buffer controls, without addition of serum were also measured on each plate to calculate cut-off levels. All determinations were performed in duplicates, with a variation of less than 5%, and results were expressed as normalized mean values. Plate to plate normalization was accomplished by including a control serum with established antigen-specific antibody levels on each of the plates 61 .
Localization of peptides on the three-dimensional structures of pre-and post-fusion F. Cartoon and surface depictions were created with PyMOL (PyMOL Molecular Graphics System, Version 2.0 Schrödinger, LLC). The PDB structures used were 6APD for the pre-fusion form and 6RRR for the post-fusion form.
Statistical analysis and sequence comparison. Statistical analysis was performed using Graph-Pad Prism 6 Software. Antibody responses towards the six peptides were analysed by One-way ANOVA and Kruskal-Wallis test. For comparison of antibody responses to the two proteins (F2-P27, F0), Mann-Whitney-U-test was performed. A p value < 0.05 was considered statistically significant (p < 0.05 = *). The correlations between antigen-specific IgG, IgA and IgM levels were calculated by XY Correlation option using the nonparametric, two tailed Spearman correlation method. p values below 0.05 were considered as significant. Difference Scientific Reports | (2021) 11:3551 | https://doi.org/10.1038/s41598-021-82893-y www.nature.com/scientificreports/ in palivizumab binding between denatured and native recombinant F0 protein was determined by Mann-Whitney-U-test. p value below 0.05 was considered as significant, p < 0.0001 = ****). For sequence alignment Clustal Omega-multiple sequence alignment program was used (https ://www.ebi.ac.uk/Tools /msa/clust alo/).

Data availability
The dataset generated during and/or analyzed during the current study are available from the corresponding author on reasonable request. License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.