Mapping human norovirus antigens during infection reveals the breadth of the humoral immune response

Phage display affinity selection and deep sequencing can identify multiple epitopes in a single sera sample

Previously, we constructed a GI.1 sheared genomic phage display library and utilized it to map the epitope of a human scFV antibody25 (Fig. S1). The library was also used to simultaneously map multiple epitopes from rabbit polyclonal antisera raised against GI.1 Virus-Like-Particles (VLPs)25. Greater than 450,000 clones were pooled to create this GI.1 genomic library, which was shown by NGS to be highly diverse with inserts displaying excellent coverage of the GI.1 open reading frames25.

In this study, we again utilized the GI.1 library to map the antigenic landscape for HuNoV infection. We verified the diversity of inserts in the library by repeating NGS of the insert region of the naive library, confirming the high diversity of the library (Fig. 1a, b). The cloning of sheared DNA into the phage display plasmid results in inserts in either the forward or the reverse orientation with respect to the GI.1 genome. Note that only the forward orientation inserts will encode in-frame HuNoV peptides. We counted a total of seven million forward strand inserts and nine million reverse-strand inserts present in the library. The inserts were aligned to the reference genome to obtain a per-nucleotide coverage score26,27. The coverage score was calculated as the number of occurrences of each nucleotide position in the pool of both forward and reverse inserts aligned to the reference genome (Fig. 1c)28. The results showed that for both forward and reverse strands, multiple inserts are present for every nucleotide position in the plasmid containing the HuNoV genome, further indicating excellent coverage of the genome by inserts (Fig. 1c). The size distribution of inserts was also calculated, showing that inserts encode peptides ranging from 10 to 170 amino acids in length, with the average insert encoding 47 amino acids (Fig. 1d). The wide range of peptide lengths as well as their continuous distribution suggests that both linear and conformational epitopes are encoded by inserts in the library.

Fig. 1: Deep sequencing analysis of the inserts present in the naive HuNoV Jun-Fos library.
figure 1

a Distribution of the forward-strand inserts (red) in the naive library mapped onto their positions in the pKS-NV68 KM plasmid. The position of the 5′ end of the insert in the pKS-NV68 KM plasmid is shown on the x-axis, while the position of the 3′ end of the insert is designated on the y-axis. The number of occurrences of the insert is defined as counts on the z-axis. b Distribution of reverse-strand inserts (black) in the naive library. c Coverage of the naive library defined by DNA sequence alignments. A total of 7,498,441 forward-strand inserts and 9,481,435 reverse-strand inserts were aligned to generate a per-nucleotide-position coverage score. The coverage score was defined as the number of occurrences of each nucleotide position in the set of DNA insert fragments aligned for the forward-strand inserts and reverse-strand inserts, respectively. The coverage scores for the forward-strand inserts of pKS-NV68 KM (red) are shown above the x-axis, while the coverage scores for the reverse-strand inserts (gray) are shown below the x-axis. d Distribution of peptide size in the naive library. The frequency of each unique peptide is shown on the y-axis while the peptide size is shown on the x-axis.

To profile the antibody response against HuNoVs during infection, serum samples of GI.1 infected persons were used for affinity selection of the GI.1 genomic phage display library. We screened sera from six persons who were infected with GI.1 HuNoV. These sera were collected as part of a previous study29,30. Affinity selections were performed on sera from four persons at 14 days after infection and two persons at 30 days after infection. Affinity selections were performed by first immobilizing antibodies in the sera using protein A/G magnetic beads. The naive phage library was added to the immobilized antibodies to allow for binding. After two rounds of affinity selection, the bound phages were eluted, and the insert region was PCR amplified. Deep sequencing was performed on the PCR products to assess the distribution of inserts when aligned to the plasmid sequence that includes the HuNoV ORFs (Fig. 1). The results for individual 715 are shown in Fig. 2 and results for all six individuals are in Fig. S2.

Fig. 2: Distribution of inserts after two rounds of affinity selection with sera of subject 715 mapped onto their positions in the reference pKS-NV68 KM plasmid.
figure 2

The x-axis denotes the 5′ end of the inserts, y-axis the 3′ end of the inserts, and z-axis the number of occurrences. The distribution of insert clusters defines the anti-GI.1 epitopes in a NS1/2 (p48), b NS3 (NTPase), c NS4 (p22), d NS5 (VPg), e NS6 (protease), and f NS7 (RdRp).

Epitopes were identified from the deep sequencing data based on the distribution of inserts expressed as count numbers mapped to the HuNoV genome with the premise that count numbers are proportional to the extent of enrichment by antibodies in the sera25,27. The distribution of inserts and associated coverage scores after two rounds of affinity selection with sera from the infected individuals showed a wide range of signals that represent epitopes located in the VP1 capsid protein as well as in several nonstructural proteins (Fig. S2).

The genomic phage display library was constructed by ligating randomly sheared genomic DNA into the phage display plasmid. Because the genomic inserts can be in either orientation (forward or reverse) and any of the three reading frames, the majority of inserts in the naive library are out of frame and will not produce a HuNoV peptide. If affinity selection is enriching for HuNoV peptides recognized by antibodies in sera, the frequency of inserts encoding in-frame peptides should increase with subsequent rounds of selection, while inserts encoding out-of-frame sequences should be eliminated. As seen in Fig. S3a, approximately 20% of the inserts in the naive library encode in-frame peptides, as expected for random inserts, while by round 2 of selection, 70–90% of the inserts were found to encode in-frame peptides while inserts encoding out-of-frame sequences were largely eliminated. These results indicate that the affinity selection enriched for HuNoV sequences that bind antibodies present in sera, i.e., epitopes. We also analyzed the fraction of in-frame inserts for each count number. Figure S3b shows that the fraction of the in-frame inserts in the naive library for each insert count group stayed consistently at 20%, which is the same as what we observed for the total fraction of in-frame inserts. Also, we observed that the frequency of in-frame inserts in the libraries after selection starts to increase with inserts with a count of five. This suggests that in-frame inserts that have a count of five or above encode for peptides enriched by affinity selection, i.e., epitopes. Therefore, in-frame inserts with a count of five or above in the libraries after selection were chosen to calculate the per-nucleotide-position coverage scores, as described above for the naive library.

To illustrate the depth of sequencing and how the sequences composing each epitope are repeatedly enriched by the antibodies in the sera, the coverage map for study participant 715 is shown as an example in Figs. 3 and S4. The coverage map indicates that the epitopes in ORF1 are located as in-frame amino acid sequences in NS1/2 (p48), NS3 (NTPase), NS4 (p22), NS5 (VPg), NS6 (protease), and NS7 (RNA-dependent RNA polymerase, or RdRp). The coverage map also indicates the presence of epitopes in the VP1 capsid protein (not labeled in Figs. 3 and S4). The epitopes revealed by the coverage scores can be resolved at higher resolution by aligning the sequences associated with regions of high coverage scores to accurately define epitopes at single amino acid resolution (Figs. 3 and S4). For example, alignment of the sequences associated with NS4 revealed a 100 amino acid epitope. Since linear epitopes are less than 20 amino acids long31, the NS4 epitope is likely conformational (Figs. 3, S4, Table S1). Alignments of sequences from regions of high coverage score from this individual similarly provided a high-resolution definition of other epitopes as seen in Fig. 3 and Table S1. Taken together, these results show that the affinity selection and deep sequencing approach can identify multiple epitopes simultaneously and at high resolution from a single polyclonal serum sample.

Fig. 3: Coverage scores of the inserts present in the HuNoV Jun-Fos library and alignment of enriched peptides in NS6 (protease) after two rounds of affinity selection with sera of subject 715.
figure 3

The coverage of the inserts after two rounds of affinity selection with sera from subject 715 was determined by DNA sequence alignment to the plasmid containing GI.1 ORFs and is shown on the left. The per-nucleotide coverage score is shown on the y-axis while the position on the plasmid sequence is shown on the x-axis. The positions of HuNoV ORFs 1 to 3 are shown below the x-axis as a reference. Only the forward-strand inserts are shown. The alignment of enriched peptides in NS6 (protease) is shown on the right. The numbering at the top of the alignment indicates the amino acid residue positions in the nonstructural proteins. The Jalview Zappo color scheme is used, in which aliphatic/hydrophobic residues (I, L, V, A, and M) are peach, aromatic residues (F, W, and Y) are gold, positively charged residues (K, R, and H) are blue, negatively charged residues (D and E) are red, hydrophilic residues (S, T, N, and Q) are green, conformationally special residues (G and P) are magenta, and cysteine is yellow. The aligned peptide families from left to right in the coverage map defines the anti-GI.1 epitopes in NS1/2 (p48), NS3 (NTPase), NS4 (p22), NS5 (VPg), NS6 (protease), and NS7 (RdRp).

Antibody responses recognize similar sets of GI.1 nonstructural and structural proteins in different individuals

We next compared the distribution of epitopes for antibodies elicited by HuNoV infection among the different individuals. This was accomplished by comparing the coverage score maps to detect the shared and unique epitopes among the cohort of six individuals. Epitopes were further verified by comparing the fraction of the in-frame inserts in a window that contains a potential epitope to the fraction of in-frame inserts in the naive library. If the in-frame insert frequency of the libraries after affinity selection is higher than that of the naive library in a given window, the presence of an epitope is confirmed (Fig. 4).

Fig. 4: Coverage scores of the inserts present in the HuNoV Jun-Fos library versus post-infection sera before and after affinity selection.
figure 4

a The coverage of the inserts in the naive library (top) as well as the inserts after two rounds of affinity selection with sera from subjects 715, 720, 723, 731, 732, and 750 was determined by DNA sequence alignment to the plasmid containing GI.1 ORFs. The per-nucleotide coverage score is shown on the y-axis while the position on the plasmid sequence is shown on the x-axis. The positions of HuNoV ORFs 1 to 3 are shown below the x-axis as a reference. Only the forward-strand inserts are shown. b The fraction of in-frame inserts in ORF2 in the naive library versus in-frame inserts after two rounds of affinity selection with sera from subjects 715, 720, 723, 731, 732, and 750. An eight amino acid sliding window along GI.1 ORF2 is shown on the x-axis while the fraction of the in-frame inserts of the naive library (black bars) and libraries after two rounds of affinity selection (red bars) are shown on the y-axis.

We identified 13 epitopes in the nonstructural proteins. There are three prominent epitopes in NS1/2 found in multiple individuals. The alignments indicate that the epitopes map to amino acid positions 131–190, 248–282, and 315–355 of ORF1, and were found in one, two, and one individual, respectively (Figs. 4a, 5, Table S1). Another 3 prominent epitopes are in NS3, located at positions 406–448, 498–554, each found in one individual, and 697–743, found in two individuals. The 7th epitope is 100 amino acids long covering positions 759–858 of ORF1 in the N-terminal domain (NTD) of NS4. As noted, the length of this sequence suggests it is a conformational epitope. This epitope was found in five individuals. The 8th epitope is in NS5 at positions 987–1037 and was found in all six individuals. The 9th epitope is in NS6. This epitope, found in five individuals, spans 86 residues from positions 1095 to 1180. Four epitopes were identified in NS7 and are located at positions 1311–1353, 1488–1539, 1577–1653, and 1709–1780 of ORF1 with lengths of 43, 52, 77, and 72 amino acids, respectively. The first two NS7 epitopes were found in two individuals while the third epitope was found in one individual, and the last epitope was found in four individuals (Figs. 4a, 5, Table S1). The length of the sequences suggests that they are conformational epitopes.

Fig. 5: Dendrogram of GI.1 Norwalk epitope profiles.
figure 5

The dendrogram compares the epitope profiles shared among the six individuals. Pink blocks indicate epitopes for each individual in the specified nonstructural or structural protein domain.

To identify epitopes in VP1, we divided ORF2 into 67 windows that are 24 nucleotides, or eight amino acid residues long. Similar to how we defined epitopes in ORF1, we compared the frequencies of the in-frame inserts in the libraries after selection and in the naive library. If the frequency of the in-frame inserts of the selected libraries is higher than that in the naive library in a given 8-mer window, we aligned the inserts within the window to identify an epitope. Overall, we identified 35 epitopes in VP1 with 19 in the S domain, 12 in the P1 subdomain, and four in the P2 subdomain. Similar to the epitopes we found in the nonstructural proteins, most VP1 epitopes were shared among more than two individuals, with five epitopes being present in all six individuals, suggesting these epitopes are highly immunogenic (Figs. 4b, 5, Table S1).

There are 19 epitopes in the S domain, ranging in length from 12 to 35 residues. We also mapped three epitopes that were recognized by the HJT-R3-A9 scFv antibody. Based on length, the majority of these sequences appear to be linear epitopes. We identified 16 epitopes in the P domain, ranging from 8 to 38 residues in length. The majority of these sequences are fewer than 20 residues long, suggesting they are linear epitopes (Figs. 4b, 5, Table S1).

Although the majority of epitopes were found in at least two individuals, the constellation of epitopes in each individual is unique. That is, no two individuals have exactly the same set of epitopes. These results are best visualized in a dendrogram that shows similarities between individuals among the sets of recognized epitopes (Fig. 5). For example, the dendrogram shows that individuals 731, 750 and 732 have a very similar set of epitopes and thus a similar antibody response, while individuals 720 and 750 have a more divergent antibody response to infection. It is also interesting to note that there are more epitopes observed in VP1 for individuals 731 and 732, whose sera was collected at 30 days post infection, compared to the other three individuals (715, 720, and 723) who had their sera collected at 14 days post infection (Figs. 4, 5).

Taken together, the results showed that the antigenic landscape includes all of the nonstructural and structural proteins, with each person containing antibodies in their polyclonal sera to a unique collection of linear and conformational epitopes that nevertheless are widely shared among the set of individuals.

Pre-existing epitopes and de novo epitopes associated with an adaptive response persist at 180 days post-infection

Previous studies have estimated that protective immunity against HuNoV infections ranges from 4.1 to 8.7 years32,33,34. However, the antigenic landscape associated with humoral immunity throughout an infection is unclear. Therefore, we performed a longitudinal study of the epitope landscape associated with the antibody response over the course of HuNoV infections in multiple individuals. For this purpose, we surveyed sera collected over a span of 180 days, including five different timepoints for study subjects 731, 732, and 750. The five timepoints included pre-infection, 7, 14, 30, and 180 days post-infection. The insert count distribution and coverage scores obtained after two rounds of affinity selection for each timepoint revealed epitopes along the HuNoV ORFs. For clarity, only the epitopes in the nonstructural proteins were labeled. Each of the study participants had antibodies recognizing defined epitopes for HuNoV in the pre-infection sera, indicating they had previously been infected with HuNoV (Figs. 6, S5). These pre-existing epitopes were verified by confirming that the fraction of in-frame inserts was greater in the selected libraries than that in the naive library. All the epitopes identified in individual 731 were pre-existing and the majority of them persisted throughout the entire course of infection. There were three epitopes that emerged post-infection in subject 732, which are the first epitopes in NS1/2, NS5, and NS6, and one epitope that appeared post-infection in subject 750, which is the first epitope in NS7 (Figs. 6, S5). The remainder of the epitopes in subjects 732 and 750 were pre-existing.

Fig. 6: Coverage scores of the inserts present in the HuNoV library after affinity selection.
figure 6

The coverage of the inserts after two rounds of affinity selection with sera collected at different timepoints from subjects 731, 732, and 750 was determined by DNA sequence alignment to the plasmid containing GI.1 ORFs. The per-nucleotide coverage score is shown on the y-axis while the position on the plasmid sequence is shown on the x-axis. The positions of HuNoV ORFs 1 to 3 are shown below the x-axis as a reference. Only the forward-strand inserts are shown. a The coverage of inserts after two rounds of affinity selection with subject 731 sera before challenge, and 7, 14, 30, and 180 days after challenge. b The coverage of inserts after two rounds of affinity selection with subject 732 sera before challenge, and 7, 14, 30, and 180 days after challenge. c The coverage of inserts after two rounds of affinity selection with subject 750 sera before challenge, and 7, 14, 30, and 180 days after challenge. Dotted lines represent pre-existing epitopes that persist from pre-infection sera through 180 days after infection while solid lines represent epitopes that emerge post-infection.

Among the epitopes that arose post-infection, the epitope in NS5 showed increased enrichment for all three individuals and reached the highest coverage scores at 14 and 30 days for subjects 731 and 750 and at 14 days for subject 732. The epitopes in NS4 and NS6 for subject 732 also showed higher enrichment at 14 days until 180 days post-infection. This indicates that immune responses targeting combinations of the epitopes in NS4, NS5, and NS6 were induced at the onset of the infection for different individuals. Finally, by 180 days after infection, the insert counts and coverage scores have reverted to the baseline epitope landscape observed before HuNoV challenge for the majority of epitopes. Specifically, antibodies that were induced post-infection as part of the adaptive response also remained at day 180 (i.e., NS4 and NS6 for subject 732). Overall, these results show that the epitope landscape differs between individuals and that the antibodies present during a HuNoV infection are a combination of pre-existing antibodies presumably arising from an earlier infection as well as new antibodies induced by the HuNoV challenge. It is noteworthy that these pre-existing, persistent antibodies apparently did not provide protection from infection in that all of these individuals were infected in the presence of the antibodies.

Affinity selections versus sera from GII.4 infected individuals reveal epitopes common to both GI.1 Norwalk and GII.4 HOV genotypes

Noroviruses are genetically diverse and are classified into 10 different genogroups broadly based on sequence identity and are further subdivided into 49 genotypes4. Ideally, a norovirus vaccine would target epitopes conserved across genogroups and genotypes. GII.4 HuNoVs have been responsible for the majority of outbreaks over the past 15 years worldwide35,36,37. They also induce more severe symptoms that require intense medical care in children38. To understand more about the cross-reactivity of the immune response against HuNoVs as well as the antigenic landscape of GII.4 norovirus, we generated a GII.4 HOV sheared genomic phage display library, using the same methods as described for the GI.1 library. Deep sequencing of the GII.4 HOV naive library revealed a dense distribution of both the forward and reverse inserts along the plasmid sequence with the GII.4 HOV ORFs (Fig. 7a, b). Approximately eight million forward inserts and five million reverse inserts were aligned to obtain a per-nucleotide position coverage score (Fig. 7c). The results revealed excellent coverage at every nucleotide position of the plasmid containing the GII.4 HOV genome that was used to construct the library. The distribution of peptide sizes encoded by the inserts range from approximately 10 to 180 amino acids with an average size of 58 amino acids (Fig. 7d). Similar to the GI.1 library, the GII.4 HOV library had a continuous distribution of peptide sizes, indicating that the insert sizes are very diverse.

Fig. 7: Deep sequencing analysis of the inserts present in the naive GII.4 HOV Jun-Fos library.
figure 7

a Distribution of the forward-strand inserts (red) in the naive library mapped onto their positions in the plasmid containing the GII.4 HOV ORFs. The position of the 5′ end of the insert in the plasmid is shown on the x-axis, while the position of the 3′ end of the insert is designated on the y-axis. The number of occurrences of the insert is defined as counts on the z-axis. b Distribution of reverse-strand inserts (black) in the naive library. c Coverage of the naive library defined by DNA sequence alignments. A total of 8,364,766 forward-strand inserts and 5,535,252 reverse-strand inserts were aligned to generate a per-nucleotide-position coverage score. The coverage score was defined as the number of occurrences of each nucleotide position in the set of DNA insert fragments aligned for the forward-strand inserts and reverse-strand inserts, respectively. The coverage scores for the forward-strand inserts of the plasmid (red) are shown above the x-axis, while the coverage scores for the reverse-strand inserts (gray) are shown below the x-axis. d Distribution of peptide size in the naive library. The frequency of each unique peptide is shown on the y-axis while the peptide size is shown on the x-axis.

We performed affinity selections using the GII.4 HOV library against sera from three individuals with high serum anti-GII.4 Sydney 2012 antibody levels, one of whom (BCM16-1) had a recent GII.4 Sydney 2012 virus infection. We additionally performed affinity selection using the GII.4 HOV library against the scFv HJT-R3-A9 antibody and rabbit anti-GI.1 VLP antisera, both of which bind GI.1 VP1. The frequency of inserts encoding in-frame HuNoV peptides in the GII.4 HOV naive library and after each round of affinity selection were assessed. Figure S6a shows that, similar to the progression of affinity selections from rounds one to two using the GI.1 library, affinity selection enriched for HuNoV sequences that bind to serum antibodies. The fraction of in-frame inserts for each count number, as shown in Figure S6b, also began to increase at inserts with a count of five, though not as apparent as the results for GI.1. The inserts with a count of five or above were used to calculate the per-nucleotide-position coverage scores.

The insert count and coverage maps generated after deep sequencing of the GII.4 HOV library selections revealed epitopes that are located in NS1/2 (p48), NS3 (NTPase), NS4 (p22), NS5 (VPg), NS6 (protease), and the NS7 (RdRp) (Figs. 8, S7, Table S2). Epitopes mapping to NS1/2 from sera of all three individuals share an epitope from residues 68 to 113 of ORF1. The epitope in NS3 was also shared with all three persons and is 81 residues long (395–475), suggesting it is a conformational epitope. The epitope in NS4 also appeared in all three individuals with a sequence of 78 residues (702–779). The epitope in NS5 has two residues that overlap the C-terminus of NS4 and appeared in all three individuals. This epitope has a common sequence of 69 residues (872–942), suggesting it is a conformational epitope. Finally, the epitope in NS6 overlaps NS5 and NS6 is also shared among all three individuals and is located from position 982 to 1029 of ORF1. The last epitope was only present in subject BCM16-1 and is in NS7. This epitope is 64 residues long (1220–1283), again suggesting a conformational epitope (Figs. 8, S7, Table S2).

Fig. 8: Coverage of the inserts present in the GII.4 HOV Jun-Fos library versus GII.4 antisera before and after affinity selection.
figure 8

a The coverage of the inserts in the naive library (top) as well as the inserts after two rounds of affinity selection with sera from subjects BCM16-1 sera, BCM13-1 sera, BCM16-2 sera, anti-NV rabbit polyclonal antibodies, and HJT-R3-A9 scFv antibody determined by DNA sequence alignment to the plasmid containing the GII.4 HOV ORFs. The per-nucleotide coverage score is shown on the y-axis while the position on the plasmid is shown on the x-axis. The positions of GII.4 HOV ORFs 1 to 3 are shown below the x-axis as a reference. Only the forward-strand inserts are shown. b The fraction of in-frame inserts in ORF2 in the naive library versus in-frame inserts after two rounds of affinity selection with sera from subjects BCM16-1, BCM13-1, BCM16-2, anti-NV rabbit polyclonal antibodies, and HJT-R3-A9 scFv antibody in GII.4 HOV ORF2. An eight amino acid sliding window along GI.1 ORF2 is shown on the x-axis while the fraction of the in-frame inserts of the naive library (black bars) and libraries after two rounds of affinity selection (red bars) are shown on the y-axis.

Affinity selection of the GII.4 HOV library against these targets also showed enrichment of sequences from VP1. As with the GI.1 experiment, we divided ORF2 into 68 windows that are 24 nucleotides, or eight residues long, and compared the frequency of the in-frame inserts of the selected libraries and the naive library in each window. The in-frame inserts with a higher frequency after selection than that in the naive library were aligned to identify epitopes. We identified a total of 18 epitopes in VP1 with 10 in the S domain, four in the P1 subdomain, and four in the P2 subdomain. Six individuals had a unique epitope. Among the 30 individuals that had more than two epitopes, five epitopes in the NTD of S domain were present in all six individuals, suggesting these five epitopes are highly immunogenic. (Fig. 8, Table S2). The epitope profiles of the sera from the three individuals as well as the A9 antibody, and anti-NV rabbit sera are shown in the dendrogram, which shows that the immune profiles between the A9 antibody and the anti-NV sera are the most similar while the profiles among the sera are more closely related, with BCM16-1 and BCM16-2 having the most similar epitope profile (Fig. S8).

Overall, fewer epitopes were identified using the GII.4 HOV library against the sera from GII.4 Sydney infected individuals than that observed in the selections of the GI.1 library against the sera of GI.1 infected individuals. This is likely due to the use of the GII.4 HOV library against sera from individuals infected with a different GII.4 genotype. In addition, the GII.4 sera were from natural infections and the timeline between infection and sera collection is unclear. The timing of sera collection is expected to impact the epitopes observed.

Since the individuals were not infected with GII.4 HOV 2002 and the A9 antibody and anti-rabbit VLP sera were shown to identify GI.1 and GII.4 epitopes, the epitopes identified from affinity selection of the GII.4 HOV library versus these targets would be for cross-reactive antibodies that recognize both GI.1 and GII.4. To identify cross-reactive epitopes and determine the level of sequence conservation within the epitopes, we aligned the sequences of the GI.1 and GII.4 HOV epitopes (Fig. S9a) and calculated the percent identity (PI) between the GI.1 and GII.4 HOV epitope sequences. The alignments revealed that NS7 epitope has the highest PI value, 81.4%, among all the epitopes (Fig. S9a), indicating that this epitope is the most conserved between genogroups. The epitope with the lowest PI value (27.5%) is that in NS1/2. Within VP1, the epitope with the highest PI value (79.5%) is in the S domain, or the 11th epitope in Fig. S9a. The epitope with the lowest PI value (31.6%) is in the NTD of the S domain, or the 7th epitope in S9a. The high levels of conservation for most of the cross-reactive epitopes in the nonstructural and structural proteins suggest possible integration into a broadly protective vaccine if they exhibit HBGA-blocking properties. In addition, they could find use as broadly reactive diagnostic tools to detect norovirus infections (Fig. S9). Finally, we searched the sequences of the norovirus epitopes identified here using protein BLAST and did not identify high sequence similarity matches to viruses other than noroviruses. Thus, the antibodies binding the epitopes we identified are likely specific for noroviruses.

Taken together, the results for the GI.1 and GII.4 HOV post-infection sera, anti-NV sera, and scFv HJT-R3-A9 antibody showed that phage display affinity selections coupled with deep sequencing can be used on complex polyclonal human sera to simultaneously identify multiple epitopes. Our data identified 48 GI.1 and 24 GII.4 HOV epitopes in the nonstructural and structural proteins. Similar epitope profiles were identified among infected individuals, although unique epitopes were also present. We also observed that epitopes change over the course of infections, illustrating the dynamic nature of the adaptive immune response. Lastly, we observed antibodies that are cross-reactive across GI.1 and GII.4 HOV. The 10 epitopes recognized by the cross-reactive antibodies have the potential to be incorporated into a vaccine or diagnostic tool for the detection of norovirus infections.

Read more here: Source link