|Home | About | Journals | Submit | Contact Us | Français|
Norwalk virus, a member of the family Caliciviridae, is an important cause of acute epidemic nonbacterial gastroenteritis. Norwalk and related viruses are classified in a separate genus of Caliciviridae called Norovirus, which is comprised of at least three genogroups based on sequence differences. Many of the currently available immunologic reagents used to study these viruses are type specific, which limits the identification of antigenically distinct viruses in detection assays. Identification of type-specific and cross-reactive epitopes is essential for designing broadly cross-reactive diagnostic assays and dissecting the immune response to calicivirus infection. To address this, we have mapped the epitopes on the norovirus capsid protein for both a genogroup I-cross-reactive monoclonal antibody and a genogroup II-cross-reactive monoclonal antibody by use of norovirus deletion and point mutants. The epitopes for both monoclonal antibodies mapped to the C-terminal P1 subdomain of the capsid protein. Although the genogroup I-cross-reactive monoclonal antibody was previously believed to recognize a linear epitope, our results indicate that a conformational component of the epitope explains the monoclonal antibody's genogroup specificity. Identification of the epitopes for these monoclonal antibodies is of significance, as they are components in a commercially available norovirus-diagnostic enzyme-linked immunosorbent assay.
Norwalk virus (NV) is the prototype member of the genus Norovirus in the virus family Caliciviridae. These viruses are responsible for 98% of all nonbacterial acute epidemic outbreaks of gastroenteritis in the United States, resulting in an estimated 23 million cases per year (4). Although it has been 30 years since NV was first identified, the study of this virus is still hampered by the lack of a cell-culture system or an animal model. However, expression of the 3′ end of the genome in a baculovirus expression system results in the production of recombinant NV virus-like particles (rNV VLPs) that are morphologically and antigenically similar to native NV virions (7, 15, 28). The availability of these rNV VLPs (as well as those of other noroviruses) has also enabled the generation of reagents, monoclonal antibodies in particular, for the study of the serological and antigenic properties of these viruses.
The Norwalk virus has icosahedral symmetry and is composed of 180 molecules of a single major capsid protein, VP1, organized into 90 dimers (29). The virus has a surface structure characteristic of animal and human caliciviruses, in which archlike structures protrude from the surfaces surrounding cuplike depressions at the three- and fivefold axes of symmetry.
The capsid protein (530 amino acids [aa]) itself folds into two domains. The amino-terminal shell (S) domain is highly conserved among animal caliciviruses (3, 6, 28). The sequence of the C-terminal protruding (P) domain, which forms protruding arches on the capsid, is more diverse, with the most variation seen in the P2 subdomain at the outer surface of the virion (3, 10).
Norwalk and related viruses are classified as noroviruses in a separate genus of the family Caliciviridae, which is comprised of at least three genogroups based on sequence similarities in the polymerase and capsid regions of the genome (36). Furthermore, these viruses have been classified into multiple antigenic groups based on results from immune electron microscopy and cross-protection studies done in volunteers (19-23), but it is unclear how these antigenic and genetic characterizations relate to each other. Information about the location of the type-specific and cross-reactive epitopes on the virus capsid is limited. Specific identification of these epitopes is essential for designing broadly reactive diagnostic assays and for helping to dissect the immune response to calicivirus infection and may be useful in identifying potential targets for antivirals.
We previously generated a panel of monoclonal antibodies (MAbs) to the rNV VLPs (9, 10, 16). Several of these MAbs are cross-reactive between viruses within distinct genogroups (9, 10, 16). Therefore, mapping the residues these MAbs recognize on the NV capsid will provide information on the determinants of cross-reactivity for caliciviruses. Because generation of neutralization escape mutants of the virus is unavailable for the noroviruses, we approached the problem using biochemical methods such as those involving deletion mutants and site-directed mutagenesis. We describe here the identification of the epitopes for the genogroup I (GI)- cross-reactive MAbs NV3901 and NV3912 and the genogroup II (GII)-cross-reactive MAb NS14. MAbs NV3901 and NV3912 map to the same epitope by competition enzyme-linked immunosorbent assay (ELISA) and recognize a common epitope shared by GI viruses, and MAb NV3901 is capable of detecting a high proportion of GI viruses in fecal samples (9). MAb NS14 reacts with multiple GII VLPs in both ELISAs and Western blots and reacts weakly with GI VLPs in ELISAs (16). Identification of these epitopes is of significance because MAbs NV3912 and NS14 are currently being used as capture antibodies in a commercially available NLV diagnostic ELISA kit, SRSV (II) AD (Denka Seiken, Tokyo, Japan) (1).
The glutathione S-transferase (GST) expression vector used to express the deletion mutants of VP1 was pGEX-2TK (Amersham Pharmacia Biotech, Piscataway, NJ). The NV primers used for this study are shown in Table Table1.1. Primers for a genogroup II.4 virus called the Houston virus (HOV) used for this study are shown in Table Table2.2. The HOV virus is related to Lordsdale virus (GenBank accession number X86557) and has 94% amino acid similarity and 91% amino acid identity therewith in the VP1 region. Each primer contains an engineered restriction site indicated in the name of the primer to facilitate cloning. The numbering of the sense primers indicates the N-terminal (first) norovirus residue contained within a construct generated with a particular primer. The numbering of the antisense primers indicates the C-terminal (last) norovirus residue contained within a construct generated with a particular primer.
PCR-amplified fragments of the NV or HOV ORF2 were generated using the primer pairs indicated by the names of the constructs. The template used for the PCR was the plasmid pG4145 (15) containing the complete NV ORF2. PCR fragments were either directly cloned into the pGEX-2TK vector or first cloned into the pCR4 Blunt-TOPO vector (Invitrogen Life Technologies, Carlsbad, CA) and then subcloned into pGEX-2TK.
Purified DNA for each clone was prepared and used to transform BL21 cells (Novagen, Madison, WI). Positive transformants were identified by PCR. To express the GST-NV fusion proteins, overnight cultures of BL21 cells transformed with each plasmid were diluted to a ratio of 1:10 in fresh LB broth supplemented with 100 μg/ml ampicillin. Cells were grown at 37°C until a density was reached where the A600 was 0.6 to 0.7. Expression was induced by addition of 1.0 mM isopropyl-β-d-thiogalactopyranoside (IPTG) (Invitrogen Life Technologies), and cultures were grown for an additional 3 h. Cells were pelleted by centrifugation for 15 min at 13,800 × g at 4°C. Supernatant was removed, and the cell pellet was suspended in 1/10 volume lysis buffer (50 mM Tris, pH 8, 120 mM NaCl, 50 mM EDTA, 3 mg/ml lysozyme) and incubated on ice for 15 min. Following incubation, Triton X-100 and 2-mercaptoethanol were added to concentrations of 1% and 10 mM, respectively, as were the following protease inhibitors: leupeptin, pepstatin, and phenylmethylsulfonyl fluoride. The suspension was subjected to two freeze-thaw cycles, and the insoluble fraction was removed by centrifugation for 15 min at 19,800 × g at 4°C. The supernatant was reserved, and the insoluble pellet was suspended in 1/10 volume solubilization buffer containing 1.5% (wt/vol) sarkosyl, 25 mM triethanolamine, and 1 mM EDTA (pH 8) and incubated on ice for 10 min. Following incubation, the sample was separated by centrifugation at 19,800 × g at 4°C. The supernatant was pooled with the reserved supernatant from the previous centrifugation adjusted to 1% Triton X-100 and 1 mM CaCl2, aliquoted into 1-ml fractions, and stored at −20°C.
Analysis of proteins by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was done according to the method of Laemmli with modifications (18). Polyacrylamide 12% resolving gels were used with a 4% acrylamide stacking gel. Prewashed glutathione-Sepharose 4B beads (Amersham Pharmacia Biotech) were added to an aliquot of supernatant and incubated with agitation for 1.5 h at 4°C. Beads were pelleted by centrifugation for 5 min at 500 × g at 4°C. The supernatant was discarded, and the beads were washed with 10 bed volumes of ice-cold phosphate-buffered saline (PBS), and then with PBS with 500 mM NaCl, and then again with PBS. Following the final wash, the beads were suspended in sample buffer containing 1% SDS, 10% 2-mercaptoethanol, 0.05 M Tris-HCl (pH 6.8), 10% glycerol, and 0.0025% phenol red and boiled for 5 min.
Proteins separated by SDS-PAGE were transferred onto a nitrocellulose membrane (Hybond-C; Amersham Pharmacia Biotech) as described previously (37). The proteins were detected using a polyclonal goat anti-GST antibody (Amersham Pharmacia) at a dilution of 1:8,000, a mouse hyperimmune anti-rNV VLP serum at a dilution of 1:5,000, or a rabbit hyperimmune anti-rHOV VLP serum at a dilution of 1:5,000 in 0.5% Blotto (Carnation nonfat dry milk in 0.01 M PBS). MAb NV3901, MAb NV3912 (10), and MAb NS14 (16) ascites were used for detection at dilutions of 1:1,000. All secondary antibodies used were conjugated to horseradish peroxidase (Sigma, St. Louis, MO). Membranes were developed by chemiluminescence using Western lightning detection reagent (Perkin-Elmer Life Sciences, Inc., Boston, MA) following the manufacturer's protocol.
Specific residues in the GST-NV 454-520 construct were altered using the Quickchange XL site-directed mutagenesis kit (Stratagene, La Jolla, CA) according to the manufacturer's instructions. Briefly, the indicated primer pairs (Table (Table3)3) containing the desired mutation were annealed to the complementary regions of the parental plasmid template and extended using Pfu polymerase to generate a mutated plasmid containing staggered nicks. The methylated parental DNA was digested by treatment with DpnI, and the newly synthesized nicked vector DNA was used to transform competent bacterial cells. Mutant clones were confirmed by sequencing.
Synthetic peptides (Sigma-Genosys, The Woodlands, TX) were used as competitors for MAb binding to rHOV VLPs in an antigen detection ELISA. The sequence of the HOV peptide is ALLRFVNPDTGRVLFECKLHKS. MAb NS14 was purified using a protein G column (Pierce, Rockford, IL) according to the manufacturer's instructions. Optimized concentrations of coating rVLP and detector MAb were determined by checkerboard serial dilution. rHOV VLPs prepared in 0.05 M carbonate bicarbonate buffer, pH 9.6, were used to coat flat-bottomed polyvinylchloride microtiter plates (Dynatech Laboratories, Inc., Alexandria, VA) overnight at 4°C. In separate tubes, constant concentrations of MAb NS14 (to give an optical density at 450 nm of between 0.5 and 0.6) were added to increasing concentrations of competitor peptide (1, 5, 10, and 15 μM) in 0.5% Blotto and incubated overnight at 4°C. A control MAb in 0.5% Blotto without peptide was included in each plate. The VLP-coated microtiter plates were washed three times with PBS containing 0.05% Tween 20 and blocked with 5% Blotto for 1 h at 37°C.
Following six washes with PBS-Tween, each of the MAb-peptide mixtures was added to duplicate wells, and the plates were incubated for 2 h at 37°C. After washing six times, a 1:5,000 dilution of goat anti-rabbit immunoglobulin G conjugated to horseradish peroxidase (Sigma) in 0.5% Blotto was added to each well, and the plates were incubated for 1 h at 37°C. To develop the ELISA, 100 μl of 3,3′,5,5′-tetramethylbenzidine (Kirkegaard & Perry Laboratories, Inc., Gaithersburg, MD) was added to each well, and the color reaction was stopped by the addition of 100 μl of 1 M phosphoric acid. The optical density at 450 nm was read, and the average values for duplicate wells were calculated.
Three initial GST-NV VP1 capsid fusion proteins were constructed based on structural information available for NV. These constructs corresponded to the full-length protruding domain of the capsid protein (aa 225 to 530), the P2 subdomain (aa 278 to 406), and the C-terminal 252 amino acids comprising the P2 and the far C-terminal portion of the P1 subdomains (aa 278 to 530) (Fig. (Fig.1C).1C). These proteins were expressed as GST fusions in bacteria and purified using glutathione-Sepharose beads. Purified proteins were analyzed by Western blotting for reactivity with anti-GST- and anti-NV-specific antisera. Each of the fusion proteins was detected using the anti-GST serum, and NV-specific proteins but not GST alone were detected in lanes corresponding to the fusion proteins using the NV anti-VLP antiserum (Fig. (Fig.1A,1A, top and middle panels).
The NV-GST fusion proteins were then tested for reactivity with MAbs NV3901 and NV3912. These monoclonal antibodies have been previously demonstrated to recognize a GI-specific epitope in the P domain of the NV VP1 capsid protein that was thought to be continuous based on its reactivity with denatured NV VLPs in Western blots (9, 10). Previous work also suggested that the antibodies bound to a region between aa 457 and 530. MAb NV3901 recognized the constructs corresponding to the full-length protruding domain and the C-terminal 252 amino acids (aa 225 to 530 and aa 278 to 530) but not the one corresponding to the P2 subdomain alone (Fig. (Fig.1A,1A, bottom panel). Identical results were found with MAb NV3912 (data not shown). These data suggested that the epitope for MAbs NV3901 and NV3912 was contained within the region between aa 406 and 530 (Fig. (Fig.1C1C).
To further define the epitope for MAb NV3901, additional deletion mutants were constructed from the existing fusion proteins. These constructs contained various amounts of the C-terminal 152 amino acids. Western blot analysis of these new deletion mutants with MAb NV3901 (Fig. (Fig.1B)1B) defined the minimal epitope-containing region for MAb NV3901 as being between aa 454 and 520 (Fig. (Fig.1C).1C). Further deletion of the C terminus of this region (NV 278-514) resulted in a reduction of binding, suggesting that the amino acids between 514 and 520, while not critical for recognition by MAb NV3901, enhanced binding. Truncated proteins with a shortened N terminus of the minimal binding region (NV 466-520) failed to be recognized by MAb NV3901. Identical results were obtained with MAb NV3912 (data not shown).
The epitope for MAb NV3901 and NV3912 was previously defined as continuous (10) based on the ability of the monoclonal antibody to recognize, by Western blotting, capsid protein that had been boiled and subjected to SDS-PAGE. However, the size of a region containing a continuous (nonconformational) epitope would be expected to be much smaller than the minimal binding region of aa 454 to 540 described above. To eliminate the possibility that the relatively large size of the GST tag might be obscuring the NV sequence and therefore masking antibody binding, a series of constructs (NV 454-530, NV 466-530, and NV 454-520) were expressed as fusions to six-histidine tags. Identical results were obtained with these new fusions (data not shown), suggesting that the GST tag did not obstruct the binding of the antibodies. A series of overlapping peptides spanning the minimal binding region also failed to react with the MAbs in a peptide ELISA (data not shown).
To further define the epitope for MAbs NV3901 and NV3912, specific residues that might be important for the genogroup specificity of the antibody were examined. Four genogroup-conserved residues were identified within the minimal binding region (Fig. (Fig.2A)2A) using data from an evolutionary trace phylogenetic analysis of noroviruses (2). These residues vary between genogroups but are absolutely conserved among GI viruses.
Three of the four residues were changed in the NV 454-520 deletion mutant to the corresponding GII residues using site-directed mutagenesis (Fig. (Fig.2A).2A). The valine at position 500 was changed to a tyrosine, and alanine was substituted for the lysine at position 514. Although the serine at position 520 is invariant in GI viruses, it is substituted by either glycine or asparagine at the corresponding position in GII viruses, so mutants containing both of these substitutions were tested. The alanine at position 519 was not changed to its corresponding GII residue (glycine) due to the conservative nature of the change. Each of the point mutants was tested in a Western blot with MAb NV3901 (Fig. (Fig.2B).2B). The V500Y, S520G, and S520N mutations did not affect MAb NV3901 binding; however, mutation of the lysine at position 514 to alanine resulted in a dramatic loss of MAb NV3901 binding, suggesting this residue is critical for genogroup-specific recognition by MAb NV3901. Identical results were obtained with MAb NV3912 (data not shown).
Examination of the crystal structure of the NV capsid protein in this region (Fig. (Fig.3)3) shows that the lysine at position 514 can interact with the glutamic acid at position 472, potentially forming a salt bridge. The glutamic acid at position 472 is conserved in both GI and GII viruses. A series of point mutations at these two positions was generated to test the importance of this interaction for MAb NV3901 and NV3912 binding (Fig. (Fig.4).4). An alanine substitution at position 472 abolished MAb NV3901 binding. The substitution of a conservative arginine at position 514 was sufficient to restore binding; however, even a conservative substitution (E472D) at position 472 was not tolerated, and binding was lost. Identical results were obtained with MAb NV3912 (data not shown).
Additional residues within the minimal binding region were tested to confirm the structural requirement for MAb NV3901 and NV3912 binding. The histidine at position 460 is in close proximity to, and thus also has the potential to interact with, the glutamic acid at position 472 (Fig. (Fig.3).3). The proline at residue 515 (Fig. (Fig.3)3) may also have an important contribution to the structure of the minimal binding region, particularly since it is part of the region that enhances MAb NV3901 recognition. Although mutation of either residue did not abrogate MAb NV3901 binding, both mutants showed a reduction in binding compared to that seen with the wild-type fragment (Fig. (Fig.5).5). This is similar to the reduction seen with the construct lacking aa 515 to 520 (Fig. (Fig.1),1), suggesting that these residues, while not critical to MAb NV3901 recognition, enhance binding. Identical results were obtained with MAb NV3912 (data not shown).
A second MAb, NS14, reacts with multiple GII viruses (16). To identify the binding site for MAb NS14, a series of deletion mutants was generated using a genogroup II.4 Houston virus (HOV). Western blot analysis of the deletion mutants (Fig. (Fig.6A)6A) defined the epitope for MAb NS14 between aa 473 and 494 (Fig. (Fig.6B),6B), although sequences between aa 453 and 473 may enhance binding. Sequence alignments of the region support the presence of a conserved epitope with a high level of conserved residues (95% aa similarity) within GII viruses. Additionally, there is a short stretch within the region, aa 473 to 484, which contains a high degree of GI/GII conservation (11 of 12 amino acids). This finding may explain the slight cross-reactivity of MAb NS14 to GI viruses.
The specificity of NS14 binding to aa 473 to 494 was demonstrated using a peptide competition ELISA. Increasing concentrations of either the HOV peptide or an unrelated peptide (RV VP4) were incubated with MAb NS14. These mixtures were then used as the detection antibody in an ELISA to detect HOV VLPs bound to microtiter plates. HOV 473-494 was able to compete for MAb NS14 binding in a dose-dependent manner, while the unrelated peptide was not (Fig. (Fig.7).7). A series of NV-specific peptides flanking the region also failed to compete for MAb NS14 binding, and the HOV peptide was unable to compete for MAb NV3901 binding (data not shown).
The production of genogroup-cross-reactive monoclonal antibodies has enabled the development of first-generation diagnostic ELISA kits to detect noroviruses in clinical samples. These ELISAs have been shown to be useful for investigating outbreaks of gastroenteritis (1, 31). However, some strains are not detected in the currently available assays, and knowledge of the locations of cross-reactive and type-specific epitopes on the norovirus capsid should help develop better second-generation assays.
MAbs NV3901 and NV3912 were found to recognize a conformational epitope located between aa 454 and 520. This region of the genome has a high degree of amino acid similarity (91%) among GI noroviruses. The large size of the MAb NV3901 epitope was unexpected, suggesting the epitope might be conformational. The conformational nature of the epitope was also unexpected, because these MAbs react with denatured capsid protein by Western blotting and so were initially called continuous epitopes (10, 16). It is possible that the minimal binding region partially renatures during Western blotting (12), allowing the discontinuous epitope to be correctly presented. Alternatively, the MAb NV3901 epitope may consist of both continuous and discontinuous elements that allow it to be recognized under the denaturing conditions of a Western blot. An antibody with similar characteristics has been described for an epitope of the pseudorabies virus glycoprotein B (42). The above alternative hypothesis is supported by the results from experiments with the deletion mutants in which the presence of residues 515 to 520 is not required for binding MAb NV3901 but significantly enhances binding when present. The mutagenesis results also suggest that the genogroup-specific nature of this antibody may be due to the interaction of the conserved residues that contribute to the conformation of the epitope.
Alignment of the binding sites for MAb NV3901 and MAb NV3912 and that of MAb NS14 show that while the GI- and GII-specific antibodies have distinct binding properties, the regions containing the epitopes overlap (Fig. (Fig.8).8). The epitope for MAb NS14 overlaps the N-terminal region of the minimal binding domain of MAb NV3901 and contains the important conserved glutamic acid residue at position 472. However, the MAb NS14 epitope does not cover the remaining 42 residues of the MAb NV3901 minimal binding region, including the critical lysine at position 514. The slight GI cross-reactivity of NS14 suggests that the region this monoclonal antibody recognizes could serve as a genogroup I- and genogroup II-cross-reactive region.
Both monoclonal antibodies described in this work recognize an epitope contained within the C-terminal P1 subdomain of the norovirus capsid. Identification of a common region identified by both GI- and GII-cross-reactive monoclonal antibodies suggests that this region may be dominant immunologically in the mouse. This is supported by previous work by Yoda et al. (40), who identified a GI-cross-reactive monoclonal antibody (8C7) that mapped to a region in the C-terminal P1 subdomain and overlaps the binding sites described in this work (Fig. (Fig.8);8); however the majority of the antibodies produced by Yoda's group have binding sites within the first 70 aa of the capsid protein (38, 40, 41). This includes two genogroup I- and II-cross-reactive monoclonal antibodies (1B4 and 1F6) which both mapped to an 11-amino-acid stretch at the far N terminus of the capsid protein (Fig. (Fig.8)8) (38, 40). These data led the authors to speculate that the S domain of the capsid protein may contain the majority of the antigenic epitopes.
These previous results contrast with the data presented here as well as with previous studies of monoclonal antibodies generated after immunization with VLPs (10, 16). These disparate results may be due to differences in the immunogens used to develop the monoclonal antibodies (native VLPs versus soluble bacterially expressed protein) or to differences in the routes of immunization. A comparison of the route of immunization used in the development of antibodies to noroviruses indicates that the production of monoclonal antibodies from mice immunized subcutaneously and subsequently subjected to intraperitoneal boost or direct immunization of spleen cells, regardless of the type of antigen used, resulted in antibodies which predominantly recognized the N terminus of the capsid protein (14, 34, 39-41). However, when mice were immunized intraperitoneally or orally, antibodies to the C-terminal domain of the capsid protein dominated (10, 16). This dichotomy of responses is also seen in the animal caliciviruses. Neutralizing and nonneutralizing antibodies to feline calicivirus (5, 26, 30, 33) and canine calicivirus (25) map to a hypervariable region in the C-terminal half of the capsid protein similar to the P2 domain of noroviruses, while antibodies to rabbit hemorrhagic disease virus predominantly map to the N terminus of the capsid protein (24, 32, 35). The reason for this polarization in localization of antibody epitopes requires further study to better predict how to produce antibodies that will be useful in diagnostic assays or possibly protective in volunteers. While the S domain is the most highly conserved region within the capsid, the P domain is highly exposed in intact particles and is also present in the soluble 32,000-molecular-weight trypsin cleavage product found in high concentrations in stool (8, 11), making it a potentially more accessible target for monoclonal antibodies that would be used in diagnostic assays.
Norovirus infections are highly prevalent in the population. In addition to being the cause of a majority of nonbacterial acute epidemic outbreaks of gastroenteritis in many countries (4, 13, 17), norovirus infection also may be the most prevalent cause of gastroenteritis outbreaks among infants (27) and norovirus has been classified as a category B pathogen according to the National Institute of Allergy and Infectious Diseases classification of pathogens important for biodefense. Improved, rapid, and broadly reactive diagnostics to detect noroviruses are necessary in order to accurately diagnose and track outbreaks and to increase understanding of virus epidemiology.
This work was supported by Public Health Service grants RO1 AI38036, P30 DK056338, and PO1 AI57788 and training grants T32 A107471 (T.D.P.) and T32 DK07664 (A.M.H.).
We thank Robert Atmar and Sue Crawford for critically reviewing the manuscript.