Xenoepitope Substitution Avoids Deceptive Imprinting and Broadens the Immune Response to Foot-and-Mouth Disease Virus

Abstract

Many RNA viruses encode error-prone polymerases which introduce mutations into B and T cell epitopes, providing a mechanism for immunological escape. When regions of hypervariability are found within immunodominant epitopes with no known function, they are referred to as “decoy epitopes,” which often deceptively imprint the host's immune response. In this work, a decoy epitope was identified in the foot-and-mouth disease virus (FMDV) serotype O VP1 G-H loop after multiple sequence alignment of 118 isolates. A series of chimeric cyclic peptides resembling the type O G-H loop were prepared, each bearing a defined “B cell xenoepitope” from another virus in place of the native decoy epitope. These sequences were derived from porcine respiratory and reproductive syndrome virus (PRRSV), from HIV, or from a presumptively tolerogenic sequence from murine albumin and were subsequently used as immunogens in BALB/c mice. Cross-reactive antibody responses against all peptides were compared to a wild-type peptide and ovalbumin (OVA). A broadened antibody response was generated in animals inoculated with the PRRSV chimeric peptide, in which virus binding of serum antibodies was also observed. A B cell epitope mapping experiment did not reveal recognition of any contiguous linear epitopes, raising the possibility that the refocused response was directed to a conformational epitope. Taken together, these results indicate that xenoepitope substitution is a novel method for immune refocusing against decoy epitopes of RNA viruses such as FMDV as part of the rational design of next-generation vaccines.

INTRODUCTION

Antigenic variation is a common technique employed by pathogens to evade host adaptive immune responses. This can be accomplished by phase variation, in which antigenic proteins are turned on and off at the genetic level (a method often used by bacteria), or through random mutation within epitopes, as is common with error-prone RNA viruses (9, 27). Immunodominant protein sequences that frequently undergo antigenic variation and do not have any apparent structural or functional purpose are often referred to as decoy epitopes. These hypervariable sequences appear to confer an advantage to RNA viruses as they keep the host's immune system one step behind the pathogen in the evolutionary arms race by inducing a vigorous immune response to a dispensable epitope, ostensibly attracting less attention to more highly conserved, yet functionally vital, epitopes (28). The identification of decoy epitopes in RNA viruses is becoming more common, and vaccinologists have struggled to deal with them since their discovery. Foot-and-mouth disease virus (FMDV) is an RNA virus that may encode one such epitope as the flexible G-H loop of VP1 is considered the immunodominant determinant of several of the known serotypes of the virus, and sequence variation within the loop is well documented (4, 35, 7). However, the RGD sequence motif found in the central region of the G-H loop is highly conserved among isolates and is responsible for binding to integrin receptors (15). Mutations in the loop appear to have little effect on its structure, but amino acid divergence here appears to contribute to serological differences found among different serotypes and subtypes (33, 6).

The quasispecies nature of RNA viruses is a driving mechanism for antigenic variability as it allows for subdominant virions to emerge as predominant within a viral population in the host. This is often facilitated by antibody-mediated neutralization of virions encoding the predominant quasispecies sequence at major epitopes, thereby allowing the outgrowth of variant viruses. Such virions are referred to as neutralization escape variants (NEVs), and antibody escape can be a particular problem in vaccine design when the major neutralizing sequences are also decoy epitopes. The evolution of NEVs has resulted in a vaccine breakthrough outbreak of FMDV in India. Numerous mutations were observed in the field isolate associated with this outbreak in the G-H loop and antigenic site C of VP1 compared to the vaccine strain that was in use, indicating that amino acid changes in these regions of the viral capsid can lead to antibody neutralization escape (38). In fact, observations of potential positive selection for NEVs with mutations in VP1 have been made in the naturally infected epithelium of FMDV-infected animals (8). Viral variants appear to be abundant and arise within lesions during the last round of viral replication, before purifying selective pressures eliminate unfit virions (40).

One method by which antigenically variable viruses avoid host clearance is via deceptive imprinting (DI) (23). DI can occur when B cells recognize and respond to the predominant amino acid sequence of an immunodominant decoy epitope within a viral population, undergo somatic hypermutation and affinity maturation to elicit highly focused antibodies directed against the decoy epitope, and essentially ignore conserved subdominant epitopes and mutated variants. DI becomes a particular problem when the imprinted epitope is both immunodominant and neutralizing as this can lead to the evolution of NEVs and increases the duration of infection. The issue of DI is exacerbated when recombinant DNA or peptide-based vaccines are utilized as they may focus immune responses to a restricted subset of viral epitopes. Such immunofocusing can actually suppress multivalent responses, which may include mutated epitopes, as has been observed with some recombinant HIV vaccines (36). Several picornaviruses are known to undergo DI via rapid mutation within immunodominant epitopes and persist through the generation of NEVs. Human rhinovirus 17 (HRV-17) is a clear example of one such pathogen. Not only does immune pressure from incubation of HRV-17 with homologous antiserum lead to the generation of NEVs, but also incubation of virus with antiserum raised against HRV-17 NEVs results in new NEVs that are not recognized by antiserum directed against the original parental strain (31). Genome comparisons of 34 isolates of HRV revealed that these viruses undergo rapid mutation, and there appears to be a diversifying selective pressure on surface residues of the viral capsid proteins that is not observed with nonstructural proteins (21). Such selective pressures likely account for the DI and generation of NEVs observed with some isolates of HRV. A similar phenomenon has also been studied with the related picornavirus FMDV. As previously mentioned, the VP1 G-H loop is considered the immunodominant antigen of the virus, and hypervariation in this region results in the generation of NEVs (37). The plethora of NEVs found within a single lesion of an animal infected with FMDV (40), along with past vaccine breakthroughs, shows that this virus deceptively imprints its host to avoid clearance in vivo.

Attempts have been made to rectify the problem of DI in vaccine design through the technique of immune refocusing. As the name implies, immune refocusing involves redirecting the immune response away from DI epitopes toward more conserved, subdominant epitopes. One approach to immune refocusing involves deleting or hiding decoy epitopes in vaccine constructs. Such a strategy was originally employed with HIV vaccine constructs, in which the response to the immunodominant gp120 V3 loop was diminished through carbohydrate addition to actively dampen immune responses to this decoy epitope (17). Genetic alteration of decoy epitopes is another approach for immune refocusing in vaccine design. DNA vaccines encoding the FMDV capsid proteins (and essential 3C protease) with mutations and/or deletions in the highly divergent immunodominant VP1 G-H loop have been tested in mice. Animals receiving a construct in which the G-H loop was either partially or entirely replaced by glycine residues produced antibody responses that were weaker against homologous virus than constructs encoding wild-type (WT) capsids or mutants bearing only point mutations. However, G-H loop replacement also appears to have generated a humoral response that has increased cross-reactivity with different serotypes (and many escape variants) as well. Complete loss of the G-H loop was not protective when animals were challenged with a lethal dose of homologous virus; however, partial deletion of the G-H loop was protective from challenge (16). These results indicate that rational immune refocusing away from decoy epitopes can protect animals from homologous viral challenge and still broaden the immune response to heterologous serotypes.

DI caused by RNA viruses has led to the need for vaccinologists to develop new methods to refocus immune responses. Such techniques should take into consideration not only the utilization of decoy epitopes by viruses but also the potential need for such regions of viral capsids for the overall immunogenicity of the virus. Here, we describe the identification of a contiguous decoy epitope within the G-H loop of type O of FMDV through multiple sequence alignments of VP1 proteins. We utilized this information to generate cyclic peptide immunogens resembling the G-H loop of type O₁ BFS (fused to a VP4 T helper cell epitope), some bearing substitutions of the entire decoy epitope with immunogenic epitopes from other viruses (HIV or porcine respiratory and reproductive syndrome virus [PRRSV]) or a presumably tolerogenic epitope from murine albumin (here, MT). Substituted xenoepitopes were selected based on their immunogenicity, length, and secondary structure, using the wild-type G-H loop as a template. These chimeric peptides were used to inoculate BALB/c mice, and cross-reactivity of each mouse's serum to each peptide was determined by enzyme-linked immunosorbent assay (ELISA). Additionally, virus binding of the murine antiserum was ascertained through the use of a virus capture ELISA. Potential contiguous epitopes in responding animals were then mapped using overlapping 11-mer linear peptides based on the wild-type sequence. The implications of using xenoepitope substitution for immune refocusing in vaccine design are discussed.

MATERIALS AND METHODS

Computational identification of the VP1 decoy epitope.

A total of 118 FMDV type O₁ VP1 sequences were downloaded from the FAO World Reference Laboratory for FMD (http://www.wrlfmd.org/fmdv_seqs/FMDV-O_seq.aspx) and aligned using ClustalW2 (24). Divergent amino acids were tabulated in Microsoft Excel. Amino acids that are divergent in four or more isolates were considered variable, and motifs of four or more contiguous variable amino acids were considered to comprise a hypervariable locus.

Cyclic peptide design and synthesis.

Using solid-phase synthesis with N-terminal protection (ThinkPeptides, Bradenton, FL), 46-mer cyclic peptides based on the FMDV O₁ BFS VP1 G-H loop were constructed. Cyclic peptides were designed to include a previously described VP4 Th epitope (3) at the N terminus and also replaced the wild-type sequence at amino acids 21 to 28 (RYSRNAVP) with xenoepitopes in all but the WT peptide. The WT G-H loop secondary structure was analyzed by NNPREDICT (22) and PSIPRED (20), and no notable secondary structure was identified at amino acids 21 to 28 (in keeping with the type O G-H loop crystal structure) (25). Xenoepitopes from the HIV gp120 V3 loop (HIGPGRAF), the envelope glycoprotein of PRRSV (HFQSIYNL), and albumin from Mus musculus (RDELPADL) were selected based on length, antigenicity, and a lack of secondary structure as predicted by the above programs. Cyclization of peptides was performed using thallium(111) trifluoroacetate on the resin with dry dimethylformamide (DMF), which induced a disulfide bond between a naturally occurring cysteine at amino acid 20 to a substituted C-terminal cysteine (replacing the natural proline at amino acid 46).

Animals and immunization.

Specific-pathogen-free (SPF) BALB/c mice (2 months old; female) were purchased from Charles River Laboratories (Worcester, MA) and acclimated for 1 week (five mice per cage) before being subjected to any procedures, in keeping with an approved IACUC protocol. Vaporized isoflurane (2.5%, vol/vol) was used to sedate animals prior to handling. Mice were bled via the mandibular vein using sterile 5-mm Goldenrod animal lancets (MEDIpoint, Inc., Mineola, NY) until a volume of approximately 100 μl was collected in heparin-coated tubes (Termumo, Elkton, MD). Bleeding was conducted on a biweekly basis starting on week 0 and continuing until termination of the experiment. Serum was obtained from whole-blood samples after centrifugation of blood collection tubes at 2,000 × g for 10 min at room temperature (RT). Lyophilized cyclic peptide immunogens were reconstituted in a volume of double-distilled H₂O (ddH₂O) with 20% (vol/vol) dimethyl sulfoxide (DMSO) to a final concentration of 1 mg/ml. Sigma Adjuvant System (monophosphoryl lipid A and trehalose dicorynomycolate [MPL-TDM] oil-in-water emulsion; Sigma-Aldrich, St. Louis, MO) was reconstituted in 700 μl of sterile phosphate-buffered saline (PBS) and then mixed with reconstituted peptides at a 1:1 ratio. Mice (n = 5 per group) were inoculated immediately after being bled (while still sedated) by the intramuscular (i.m.) route via injection of 100 μl of peptide/adjuvant into each hindlimb (precleaned with an alcohol wipe). Animals were primed on week 0 and boosted biweekly over 10 weeks, at which point terminal bleeds were conducted, and animals were humanely euthanized by cervical dislocation.

Anti-cyclic peptide ELISA.

Antigen-specific indirect ELISA was performed on antiserum to assess the response against each chimeric FMDV peptide, essentially as previously described (13). Briefly, flat-bottomed 96-well Immulon IV HBX microtiter plates (Thermo Labsystems, Franklin, MA) were coated with each chimeric peptide (from the same 1 mg/ml stock solutions described above) following dilution to a concentration of 10 μg/ml in 0.05 M sodium carbonate buffer (pH 9.6) and then applied to microtiter wells at a volume of 50 μl/well and incubated overnight at RT. Background control wells were similarly coated using bovine serum albumin (BSA). Plates were washed three times with 125 μl of PTA-BSA wash buffer (PBS with [vol/vol] 0.05% Tween 20, 0.02% sodium azide, 0.01% BSA) with shaking using a Bio-Tek EL-403 plate washer (Bio-Tek Instruments, Inc., Winooski, VT). Week 0 and 10 sera (and week 8 serum from one of the animals from the PRRSV peptide group, which died before completion of the study) were diluted 1:1,000 in PTA-BSA and then added to plates after they were washed at a volume of 50 μl/well and incubated at RT for 1 h. Plates were then washed again, an alkaline phosphatase-conjugated goat anti-mouse IgG (Fc) secondary antibody (Sigma-Aldrich) was added at a dilution of 1:5,000 (50 μl/well), and plates were incubated overnight at 4°C. Plates were then washed, and p-nitrophenyl phosphate substrate (1 mg/ml in 0.05 M Na₂CO₃ and 1 mM MgCl₂ buffer, pH 9.8) was added at a volume of 50 μl/well. Plates were allowed to develop for 100 min, and product formation was determined by measuring the optical density (OD) of each well at an absorbance wavelength (λ) of 405 nm with an EL-311 Bio-Tek plate reader. Results are reported as the Δ-Δ OD (week 10 OD value minus week 0 OD value, with background BSA well OD value subtracted from both) with comparisons between all peptides relative to serum from a single group (with significance determined in comparison to homologous peptide or ovalbumin [OVA]).

Virus capture ELISA.

Flat-bottomed 96-well Immulon IV HBX microtiter plates were coated overnight with 50 μl of anti-FMDV O₁ Manisa antiserum raised in rabbits (Institute for Animal Health, Pirbright, United Kingdom) at a dilution of 1:4,000 in 0.05 M sodium carbonate buffer (pH 9.6). Plates were then washed five times in PBS-T (PBS–0.05% Tween 20), and blocking was performed using 200 μl of blocking buffer (PBS-T with 10% horse serum; Gibco,) for 1 h at 37°C in a shaking incubator (these wash and incubation protocols were used throughout the rest of the assay). FMDV O₁ Campos was captured upon addition of 50 μl of supernatants from previously infected BHK-21 cells (uninfected cell supernatants were used in negative-control wells). After virus capture and washing, 50 μl of week 0 or week 10 mouse serum (diluted 1:50 in blocking buffer) was added to each well. After incubation and a washing step, 50 μl of horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG (whole molecule; Sigma-Aldrich) was added at a dilution of 1:1,000. After a final incubation and wash step, 50 μl of 3,3′,5,5′-tetramethylbenzidine (TMB) Blue HRP substrate (Millipore, Billerica, MA) was added to each well. The reaction was allowed to proceed for 30 min at RT and stopped upon addition of 50 μl/well of 2N H₂SO₄. Product formation was determined by measuring the optical density (OD) of each well at an absorbance λ of 450 nm with an EL-311 Bio-Tek plate reader. Results are reported as the Δ-Δ OD (week 10 OD value minus week 0 OD value, with the background BSA well OD value subtracted from both); each xenoepitope-harboring peptide was compared to controls (WT control [WT-C] and OVA). Virus capture ELISAs were performed in a biosafety level 3 (BSL-3) laboratory at the USDA Plum Island Animal Disease Center (Greenport, NY).

Linear B cell epitope mapping.

A Prospector PEPscreen custom peptide library (ThinkPeptides) of overlapping linear 11-mer peptides (offset by two amino acids, each with N-terminal biotinylation), based on the previously described TCA peptide (consisting of VP4 T helper cell epitope and VP1 site C and VP1 site A epitopes) (2), was used to map linear B cell epitopes. Peptides used were as follows (peptide numbers are listed in order of the WT cyclic peptide amino acid sequence): 1, SIINNYYMQQY; 2, INNYYMQQYQN; 3, NYYMQQYQNSM; 4, YMQQYQNSMDG; 8, TLPPSVPNLRG; 9, PPSVPNLRGDL; 10, SVPNLRGDLQV; 11, PNLRGDLQVLA; 12, LRGDLQVLAQK; 13, GDLQVLAQKVA; and 14, LQVLAQKVART. Three additional peptides were synthesized to cover the region of the G-H loop not found in the TCA peptide, as follows: 5, YNGECRYSRNA; 6, GECRYSRNAVP; and 7, CRYSRNAVPNL. Lyophilized peptides were reconstituted to a concentration of 10 mg/ml in sterile ddH₂O with 80% DMSO. A total of 200 μl/well (diluted to 5 μg/ml in sterile PBS) of each peptide was coated in SigmaScreen Streptavidin high-capacity coated plates (Sigma-Aldrich) overnight at RT. Plates were washed with 300 μl of PBS-T and blocked with 225 μl of blocking buffer (10× stock diluted to 1× in PBS-T; Sigma-Aldrich) for 1 h at 37°C. Samples and the secondary antibody were diluted using PBS-T; otherwise, all other conditions were the same as those described in the paragraph “Anti-cyclic peptide ELISA.”

Statistical analyses.

Data distributions were determined to be non-Gaussian; therefore, a nonparametric Kruskal-Wallis analysis of variance (ANOVA) on ranks test was performed with multiple pairwise comparisons conducted using a Student-Newman-Keuls posthoc test (α = 0.05).

RESULTS

Type O VP1 sequence analysis and identification of a hypervariable decoy epitope in the G-H loop.

The VP1 amino acid sequences of 118 different isolates of FMDV type O were compared by multiple sequence alignment to identify variable amino acids and hypervariable contiguous motifs. A low degree of variability was observed throughout VP1 with the greatest divergence detected within the integrin binding G-H loop (amino acids 131 to 160). Nine amino acids were determined to be variable within this region (Fig. 1), including six contiguous residues at positions 137 to 142, and are defined as hypervariable). Mutations within these amino acids appeared to be highly random, with most structural classes (i.e., polar, hydrophobic, or charged) represented at each position (data not shown). Secondary-structure prediction programs determined that the G-H loop has little secondary structure, with only a potential alpha-helix predicted downstream of RGD (in keeping with the published crystal structure) (25). Previous studies have determined that peptides spanning residues 137 to 142 cross-react with, and even may induce, virus neutralizing antibodies (41, 5), thereby indicating that these amino acids form a functional epitope. Hence, the hypervariable epitope at positions 137 to 142 of the immunodominant G-H loop does not appear to serve any known function (e.g., receptor binding), supporting the notion that it is a decoy epitope.

Fig 1.

Identification of variable amino acids within the G-H loop and cyclic peptide design. The G-H loop sequence (top) is listed with variable amino acids in bold and underlined. The RGD integrin binding motif is denoted by italics and a surrounding box. A representative structure of the cyclic peptides (bottom) is shown with the xenoepitope substitutions underlined. aa, amino acid.

Cross-reactivity of sera from BALB/c mice inoculated with chimeric cyclic peptides harboring xenoepitope substitutions.

Using information garnered from the computational analysis of the VP1 G-H loop of FMDV type O, peptides were synthesized that retained a cyclic structure and contained an N-terminal fusion to a known Th epitope (Fig. 1). The decoy epitope at residues 137 to 142 was replaced with known epitopes from the HIV V3 loop, PRRSV glycoprotein, or a presumed tolerogenic epitope from murine albumin (referred to as MT). A cyclic peptide based on the WT sequence was used for comparative purposes, and OVA was used as a negative control. Mice were immunized with one of the peptides on a biweekly basis for 10 weeks, and serum was collected for analysis of cross-reactivity to each peptide by ELISA (Fig. 2). Animals inoculated with the WT peptide had a far more robust type-specific reaction to the WT than to any other peptide although reactivity was higher to all peptides than to OVA. These results indicate that the decoy epitope is one of several B cell epitopes found within the G-H loop (as has been previously described), but antibody responses directed to this region appear to be stronger than to all other areas combined. Animals inoculated with peptides containing the HIV epitope demonstrated only type-specific immunity, indicating that these animals produced a focused antibody response. Conversely, mice immunized with peptides bearing the PRRSV xenoepitope substitution elicited broadly cross-reactive antibodies that recognized each FMDV peptide equally, indicating that these animals refocused their immune responses to a cross-reactive epitope elsewhere in the G-H loop. Results from animals inoculated with the MT chimeric peptide were surprising as there was a stronger type-specific response observed in these mice than that directed toward heterologous peptides. Therefore, the MT epitope (in the context of the FMDV G-H loop, a potent adjuvant system, and repeated booster doses) was able to break tolerance and induce immunity to this epitope. Taken together, these results appear to demonstrate that the resolution of immunity (focused/type specific, broad/refocused, or mixed) of animals inoculated with chimeric cyclic peptides is highly dependent on the amino acid sequence of the xenoepitope substitutions.

Fig 2.

Cross-reactivity of sera from mice inoculated with chimeric cyclic peptides. Sera from mice immunized with chimeric cyclic peptides were probed against each cyclic peptide coated to a 96-well plate by ELISA. Mean group responses plus 1 standard deviation are represented. Significant differences were determined by ANOVA on ranks with multiple pairwise comparisons conducted using a Student-Newman-Keuls posthoc test (α = 0.05). +, different from OVA; *, different from the homologous peptide; MT, murine tolerogen.

Mapping of linear B cell epitopes in WT and PRRSV xenoepitope substitution cyclic peptides.

In order to identify cross-reactive B cell epitopes contained within the FMDV portion of the PRRSV chimeric peptide, sera from immunized animals was probed against a library of overlapping linear peptides by ELISA (Fig. 3). A comparison was made to mice inoculated with the WT cyclic peptide to assess the level of cross-reactivity. Strong and fairly equal responses were found against the WT cyclic peptide with linear peptides 5, 6, and 7, each of which harbors six to eight (out of eight) amino acids of the decoy epitope. The response to linear peptides 8, 9, and 10 was even higher before waning throughout the upstream region and the RGD integrin binding motif. A third region of strong reactivity against the WT peptide was found at the C terminus of the G-H loop (linear peptide 14), which has been previously identified as a B cell epitope (5, 41). Curiously, no obvious linear epitopes (based on the WT sequence) could be identified using sera from animals immunized with the cyclic peptide containing the PRRSV xenoepitope substitution. This epitope mapping approach is limited in two ways; first, it does not take into consideration the presence of conformational epitopes and second, neo-epitopes found in the chimeric peptides were also not assessed. However, given the broad cross-reactivity to all cyclic peptides induced by the PRRSV chimeric peptide, it is unlikely that a neo-epitope is present in this antigen. Thus, it appears likely that the PRRSV xenoepitope substitution refocused the immune response of inoculated mice to a conserved conformational epitope found within the G-H loop, such as one previously identified in the G-H loop of type O FMDV (30). This putative conformational epitope also exists on the virus capsids as the same antisera recognized native virus based on results from the virus capture ELISAs (as described below).

Fig 3.

B cell epitope mapping relative to the wild-type G-H loop. Sera from mice inoculated with WT or PRRSV chimeric cyclic peptides were probed against a library of short linear peptides (numbered 1 to 14) based on the wild-type sequence of type O1 BFS (shown just below the graph) by ELISA. Some peptides were used in a previous study and contain divergent amino acids (underlined). Mean group responses to each peptide (top) in the WT group (black bars) and PRRSV chimeric peptide (white bars) are shown.

FMDV binding of sera from chimeric cyclic peptide-inoculated mice.

The lack of definitive identification of a cross-reactive epitope in the cyclic peptide containing the PRRSV xenoepitope substitution raised the question of the relevance of the broad immunity induced by this immunogen. To answer this inquiry, a virus capture (sandwich) ELISA was performed in which wild-type virus was bound to a capture antibody (coated in 96-well plates) and probed with sera from mice inoculated with the chimeric cyclic peptides (Fig. 4). Sera from mice immunized with the peptide bearing the PRRSV xenoepitope substitution bound to FMDV equally as well as those from mice inoculated with the WT cyclic peptide. Sera from mice given the cyclic peptide with the HIV xenoepitope substitution or OVA did not bind to virus, while sera from animals inoculated with the MT chimeric cyclic peptide demonstrated some capability for viral binding (although results varied greatly in this group). Hence, xenoepitope substitution appears to be a technique that can refocus immune responses to alternative epitopes found within immunogens, and, in the case of peptides based on the FMDV type O G-H loop, such alternative epitopes are found within the viral capsid.

Fig 4.

Wild-type virus binding of sera from mice inoculated with chimeric peptides. Sera collected from mice inoculated with chimeric peptides were probed against type O virus in a sandwich virus capture ELISA. Mean group responses plus 1 standard deviation are represented. Significant differences were determined by ANOVA on ranks with multiple pairwise comparisons conducted using a Student-Newman-Keuls posthoc test (α = 0.05), and groupings are indicated by A or B.

DISCUSSION

RNA viruses are inherently variable as their error-prone polymerase often makes one or more mutations per virion produced. The presence of decoy epitopes within many of these viruses is, then of no surprise. To this point, the V3 loop of HIV-1 gp120 appears to contain at least one decoy epitope as portions of it are highly variable and immunogenic, while the central region needed for coreceptor binding is fairly well conserved (18, 19, 1). The close proximity of the hypervariable region(s) to the crucial coreceptor binding sequence has made HIV vaccine development quite a challenging endeavor. Another important human pathogen, hepatitis C virus (HCV), encodes a hypervariable region in the envelope 2 protein that elicits neutralizing antibody responses. Treatment of HCV with hyperimmune serum directed against peptides derived from the hypervariable region does not protect chimpanzees from disease, which is believed to be caused by NEVs present in the initial inoculum (12). NEVs can arise after a single amino acid substitution during chronic HCV infection, and while these epitopes may still be antigenic, they are inefficient at controlling disease (11). In accordance with these other RNA/retroviruses, the highly variable nature of the G-H loop of the FMDV VP1 protein has been well described (4, 35, 7). Thus, the identification of a decoy epitope in the type O virus in this work is in keeping with previous studies. Of note, the G-H loop is considered by many investigators to be the immunodominant determinant that is primarily responsible for neutralization of virions in vivo; therefore, the identification of the decoy epitope at residues 137 to 142 is of particular concern as it may contribute to the evolution of NEVs. Indeed, NEVs have been well characterized in FMDV disease (10) and are a potential source for a vaccine breakthrough (38).

The presence of decoy epitopes in RNA viruses and their subsequent ability to escape neutralization allow these pathogens to deceptively imprint their host's immune cells with antigens that are not necessary for viral pathogenesis. The yellow fever virus (YFV) vaccine strain 17D-204 has been determined to have mutations in the envelope protein at amino acid positions 305 and 325. Reversion of these amino acids to WT sequences occurs in some NEVs found within populations of the vaccine strain. Unfortunately, strong neutralizing immunity to the 17D-204 sequence may deceptively imprint the host's memory B cells away from generating responses against revertant viruses, and this effect is compounded by increased neurovirulence in the brains of mice harboring envelop protein NEVs (34). Such mutations potentially could lead to a vaccine breakthrough although this problem has not yet been observed with the YFV vaccine. During porcine respiratory and reproductive syndrome virus (PRRSV) infection, deceptive imprinting is achieved through direction of antibody responses toward a nonneutralizing decoy epitope of gp5 for a period of at least 3 weeks before significant antibody titers are produced against the adjacent neutralizing epitope (29). This strategy likely contributes to viral transmission to new hosts by increasing viral loads and prolonging the infectious period of diseased animals. The data presented in this work indicate that FMDV potentially deceptively imprints host B cells using its G-H loop decoy epitope. This is evidenced by strong reactions in the epitope mapping experiment to the hypervariable region of residues 137 to 142 of the G-H loop by mice immunized with the WT cyclic peptide, whereas broadly cross-reactive animals inoculated with the peptide bearing the PRRSV epitope do not respond to this decoy epitope. Therefore, the avoidance of deceptive imprinting with decoy epitopes without elimination of entire immunogenic determinants appears to be achievable using xenoepitope substitution, which may also result in broader immunity to alternative regions of the virus containing more highly conserved, subdominant neutralizing epitopes.

Preventing deceptive imprinting by immunodominant decoy epitopes is an essential component of the rational design of next-generation vaccines directed against RNA viruses; however, immune responses then need to be redirected to subdominant epitopes in order to induce protective immunity. One approach to immune refocusing is to reduce the availability of a decoy epitope through immune dampening. Masking of HIV capsid epitopes through altered glycosylation patterns has been used as a way to dampen the immune response directed toward the immunodominant V3 loop decoy epitope. Refocused animals are still capable of neutralizing virus, but epitope mapping experiments indicate that a significant response can be redirected toward the subdominant V1 loop in these animals (17). Unfortunately, the V1 loop is also variable, and so further refocusing attempts are being conducted using altered glycosylation patterns for other regions of the HIV capsid as well (39).

Immune refocusing has also been used to direct humoral responses away from the G-H loop of FMDV. Capsid-based DNA vaccines bearing full or partial replacement of the G-H loop (including the decoy epitope) with glycine residues have been found to greatly enhance cross-reactivity of sera from inoculated mice to different serotypes of FMDV. Complete replacement of the G-H loop was inefficient at protecting mice from homologous challenge, but partial G-H loop replacement appears to have prevented viremia and death in vaccinated animals (16). In a related experiment, Fowler et al. (14) determined that partial loss of the FMDV serotype A G-H loop, immediately upstream of RGD and through the downstream alpha-helix (the structure of which was predicted to be lost by computational analysis), in inactivated vaccines did not generate broadened immunity or immune refocusing in cattle. The differences in results between these two studies were attributed to either different vaccination approaches (DNA vaccine versus inactivated virus) or the animal species tested (mice versus cattle). However, the use of xenoepitope substitution of the decoy epitope in this work does appear to have broadened immunity to different epitopes found in the native virus (in the peptide group containing the PRRSV substitution), which is in keeping with the results of Frimann et al. (16).Two similarities of our study with Frimann et al., which differ from those of Fowler et al., were the animal species used (mice versus cattle) and serotype of FMDV used (type O versus type A). Given the immense variation in G-H loop amino acid sequence between serotypes (with the exception of the RGD integrin-binding motif), it is possible that the region upstream of RGD is not as important for immunity against type A viruses (and may not even contain a decoy epitope). Thus, selection of xenoepitopes to be used for substitution and careful identification of the target strains(s) are both important factors when xenoepitope substitution is used in vaccine design and should be given careful consideration before its implementation.

In this work we identified a decoy epitope in the G-H loop of type O of FMDV and replaced it with immunogenic epitopes from other sources in peptide immunogens using xenoepitope substitution. Swapping the decoy epitope out with one from the PRRSV gp5 glycoprotein resulted in immune refocusing to a presumed conformational epitope, thereby broadening immunity in mice to all peptides tested. Sera from these animals were capable of binding to wild-type virus, indicating that the refocused epitope is present in the virus. Unfortunately, it is widely accepted that protection from FMDV challenge is dependent on the presence of multiple neutralizing epitopes in vaccines, indicating that the peptides used in this study will not be sufficient for vaccination in the field. Indeed, sera collected from mice inoculated with similar cyclic peptides in a pilot study were not capable of neutralizing virus in vitro (data not shown). The most likely contribution xenoepitope substitution has to offer the field of FMDV vaccine design will be in the genetic replacement of the decoy epitope with the PRRSV gp5 epitope in the context of current capsid-based vaccines delivered via the replication-defective adenovirus (26, 32). However, the peptide approach employed here may still be useful for other RNA viruses, particularly when the utilization of whole inactivated virus is not feasible. Also of note in this study, xenoepitope substitution with the HIV V3 loop epitope resulted in focusing of immunity specifically to this sequence, and substitution with a presumptive tolerogen from murine albumin appears to have broken tolerance. It can thus be concluded that the cyclic peptides utilized in this study are capable of not only broadening immunity but also focusing humoral responses or breaking tolerance to self-antigens. This may have implications for other applications for these peptides beyond immune refocusing.