By generating a full-length infectious clone of the CVB3 genome (pBRCVB3), we had previously reported that the pBRCVB3 virus induces mainly pancreatitis, but its myocarditis-inducing ability was severely impaired11. Using this virus as the backbone, we sought to create attenuated mutant viruses that would lose their ability to induce both myocarditis and pancreatitis, thus leading us to identify one mutant virus possessing a mutation in the CAR-binding region as a vaccine candidate in the prevention of CVB3 infection.
Localization of an epitope in the VP1 region that interacts with CAR
Our initial efforts were intended to identify the immunogenic epitopes of CVB3 to track the generation of virus-specific T cell responses in infected animals12. We targeted VP1, one of the four structural proteins of CVB3 that has been identified as an immunogenic protein in other enteroviruses13,14,15. Using an overlapping peptide library, we identified three epitopes—VP1 681–700, VP1 721–740, and VP1 771–790—that induce antigen-specific, CD4 T cell responses as evaluated by MHC class II dextramers/tetramers12. Next, we took advantage of crystal structure data (PDB: 1COV)16 and, using the Discovery studio visualizer software, retrieved the footprints of the putative binding regions of CVB3 and CAR to determine the locations of viral epitopes. Using the cryo-electron microscopic model17, we mapped the VP1 to VP4 proteins as illustrated in the ribbon and surface models (Fig. 1a). These analyses led us to note that VP1 771–790 forms a part of the CAR-binding region within the canyon of CVB3 (Fig. 1b, top panel). Importantly, by analyzing the footprint of the CAR–CVB3-interacting area in the canyon, we noted that eight residues [arginine, asparagine, glycine, valine (RNGV); N, threonine (T), and NN; shown as balls] were exposed within VP1 771–790 (Fig. 1b, bottom panel with inset), whereas tyrosine, glycine, isoleucine (YGI), and leucine (L) of the N-terminal half and the C-terminal residues [leucine (L) to histidine (H)] were buried (not shown) in the peptide. Based on this information, we hypothesized that the exposed residues in the VP1 771–790 epitope are required for viral entry into the target cells through CAR interaction, whereas the buried residues are not critical for viral entry because they are not expected to interact with CAR (Fig. 1c). However, three residues toward the C-terminus, methionine (M; 783), G (784) and T (785), although buried partially, can also interact with CAR17 (Fig. 1b, bottom panel inset). Thus, we identified a region in CVB3 from which we would be able to manipulate the viral genome and derive mutant strains to test the hypothesis that the mutant viruses may retain infectivity, but lose pathogenicity and determine their phenotypes.
Creation of CVB3 mutant viruses
To create the mutant strains, we used the pBRCVB3 virus11. We sought to generate mutant viruses by targeting nine buried residues within the VP1 771–790 region. These included VP1 775, 776 and 777, and 780, 786, 787, 788, 789 and 790, corresponding respectively to two stretches of amino acids, YGI and LYARH, including VP1 780 for L at the N-terminal end of the sequence (Fig. 1c and Table S1). Specifically, via PCR-based site-directed mutagenesis using primers (Table S2), each of the above amino acids were mutated to A, except for VP1 788 residue, which was mutated to G (Table S3). In addition, two more mutants were created, each representing mutations made in stretches from positions VP1 775 to VP1 777 and from VP1 786 to VP1 790 (Table S3). After generating plasmid constructs containing the mutated CVB3 genomes, locations of mutations were confirmed by Sanger DNA sequencing prior to virus recovery (Table S4).
Recovery of infectious mutant viruses
To recover infectious viruses, we performed transfection experiments in Vero cells using the in vitro-transcribed RNA obtained from each mutant clone. After passaging twice, we ascertained expression of viral proteins by immunofluorescence, using the wt CVB3 and pBRCVB3 as positive controls, with media containing uninfected cells serving as negative controls. As shown in Fig. 2a, cells infected with wt CVB3 or pBRCVB3 exhibited viral protein expression, but as expected, no expression was detected in mock-infected cells (medium control). By comparing the intensity of immunofluorescence, we noted that the cells transfected with viral RNAs representing the mutants 2, 3, and 10 had comparable expressions of viral proteins, whereas the fluorescence intensities for mutants 4 and 8 were relatively diminished (Fig. 2a). To rule out differences in transfection efficiency, we next determined the multi- and single-step growth kinetics of the mutant viruses using multiplicity of infection (MOIs) of 0.1, and 3.0 to ascertain if the mutations introduced in VP1 affect the growth of the mutant viruses. The analyses revealed significant differences between viruses, except for mutant 3. During the initial 3 days of infection at both MOIs, viral titers for mutants 2, 4, 8, and 10 were lower than wt CVB3 or pBRCVB3 viruses (0.8 to 2.2 log) (Fig. 2b). As the days progressed (days 4 to 6), viral titers continued to be low for mutants 2 and 4 (Fig. 2b, left panel), whereas those of mutants 8 and 10 reached the levels of wt CVB3 and pBRCVB3 viruses at MOIs 0.1 and 3.0. However, at MOI 3.0, no such difference was noted for mutant 2 (Fig. 2b, right panel). Of note, viral RNA obtained from other mutants—namely, Mt 1, Mt 5, Mt 6, Mt 7, Mt 9, and Mt 11—did not yield recoverable viruses, suggesting the mutations are lethal (data not shown). Together, the data suggested that the mutations introduced in positions 776 (Mt 2), 777 (Mt 3), 788 (Mt 8), and 790 (Mt 10), with an exception of 780 (Mt 4), did not appear to significantly affect viral replication, leading us to test their infectivity in vivo.
Identification of Mt 10 as a vaccine candidate
We infected groups of mice with five mutant viruses (Mt 2, Mt 3, Mt 4, Mt 8, and Mt 10) individually; at termination on day 21 post-infection, hearts and pancreata were collected for histology (Fig. 3a). In these experiments, wt CVB3 and pBRCVB3, and saline recipients were used as positive and negative controls, respectively. Figure 3b, left panel, shows that wt CVB3-infected mice, but not saline-recipients, had lost body weight by day 9 post-infection [p.i] (p ≤ 0.0001), whereas those infected with pBRCVB3 had a similar decline in relation to saline recipients (p ≤ 0.0001), but it occurred progressively as expected11. In contrast, animals infected with the mutant viruses remained clinically healthy and did not lose body weight compared to wt CVB3 or pBRCVB3 groups (p ≤ 0.0001), except for mice infected with Mt 3. We then compared mortalities between groups, and, expectedly, all animals infected with the wt CVB3 died by day 9 p.i (100%), as opposed to 20% (2/10) of those infected with pBRCVB3 virus (Fig. 3b, right panel, and Table 1, top panel). However, none of the animals infected with the mutant viruses died, except for one animal in the Mt 4 group (~ 11%) that had succumbed to the disease (Fig. 3b, right panel, and Table 1, top panel), suggesting that the degree of pathogenicity of mutant viruses may vary.
We then examined hearts and pancreata from the above groups for microscopic inflammatory changes. Expectedly, histological analyses revealed that 100% (10/10) of animals infected with the wt CVB3, and 80% (8/10) of animals infected with pBRCVB3 had myocardial lesions as indicated by inflammatory foci of 34.2 ± 13.4 and 11.8 ± 7.3, respectively (Table 1, top panel), with infiltrates primarily consisting of mononuclear cells (MNCs) (Fig. 3c, top panel). By comparison, isolated lesions were noted in heart sections only in groups that received Mt 3, Mt 4, and Mt 8, whereas sections from other groups (Mt 2 and Mt 10), including the saline group, were free of myocarditic lesions (Table 1, top panel). By analyzing pancreatic sections, we noted that mice infected with wt CVB3 or pBRCVB3 virus had comparable pancreatitis as revealed by atrophy, inflammation, necrosis, and mineralization (Fig. 3c, and Table 1, bottom panels); as expected, pancreatic sections from the saline group were devoid of such lesions. However, sections from groups infected with mutant viruses revealed incidence of pancreatitis to be similar for Mt 2 (~ 89%) or Mt 3 (100%) viruses, whereas incidence was relatively low in the Mt 4 virus group (~ 56%), followed by the Mt 8 virus group (~ 31%), but necrosis in the latter was also lacking (Table 1, bottom panel). Strikingly, however, sections from recipients of Mt 10 virus did not reveal pancreatitis, which is similar to sections from their healthy counterparts (Fig. 3c, and Table 1, bottom panels). Since animals infected with Mt 10 virus were free of both myocarditis and pancreatitis, we decided to evaluate whether the Mt 10 virus can be used as a vaccine candidate.
Mt 10 virus offers protection against wt CVB3 virus in challenge studies
We performed challenge studies with an expectation that animals primed with Mt 10 virus would be protected from wt CVB3 challenge. We tested this hypothesis by immunizing animals with a single dose of Mt 10 virus and challenging them 14 days later with wt CVB3. After 21 days post-challenge, animals were euthanized and tissues were collected for histopathology. Saline groups were used as controls (Fig. 4a). Clinically, non-vaccinated animals infected with wt CVB3 virus lost body weight significantly starting ~ day 3 post-infection. Similar to the saline control, animals in the Mt 10 vaccine group, and more importantly, animals in the Mt 10 + wt CVB3 group did not lose body weight and remained clinically healthy during the length of the experiment (p ≤ 0.0001) (Fig. 4b, left panel). These observations were further captured by analyzing mortality rates; 50% (9/18) of animals in the wt CVB3 alone group died, but no mortalities were noted in saline, Mt 10-alone, and Mt 10/wt CVB3-challenged groups (Fig. 4b, right panel, and Table 2, top panel). Additionally, we did not observe sex-based differences with vaccine protection, but it was critical to evaluate tissues for inflammatory changes, if any, in the challenged animals.
Histologically, heart sections from ~ 67% (12/18) of mice infected with wt CVB3 had myocardial lesions (34.5 ± 13.3), whereas inflammatory changes were lacking in both vaccine-alone and vaccine/challenged groups (Fig. 4c and Table 2, top panel). Similarly, 89% (16/18) of animals infected with wt CVB3 had pancreatitis as indicated by atrophy (89%) and inflammation (89%), whereas sections from the vaccine or vaccine/challenged groups did not reveal such changes (Fig. 4c and Table 2, bottom panel), except that a small area of atrophy was noted in only one animal in the vaccine group (Table 2, bottom panel). Overall, the finding that the vaccine recipients challenged with wt CVB3 were completely protected from both myocarditis and pancreatitis implies that disease protection might have been mediated by the Mt 10 virus. Thus, we identified Mt 10 as an attenuated, avirulent CVB3 vaccine virus, but the data raised questions as to the underlying mechanisms of disease protection.
Mt 10 virus induces virus nABs predominantly of IgG isotypes
An important component of the adaptive immune response is the production of antibodies that are critical to preventing colonization and spread of infections. First, we measured nABs based on cytopathic effect (CPE), which involved assessment of percent neutralization of wt CVB3 in Vero cells exposed to serum collected from saline or Mt 10 vaccinated groups. Figure 5a shows that the virus was not neutralized when cells were exposed to serum from saline recipient animals, as expected. In contrast, sera harvested from vaccine recipients on day 14 showed 100% virus neutralization at 1:160 dilution, and virus neutralization of ~ 50% or more was still evident at dilutions up to 1:2560. Similar trends were observed with sera collected at day 35 post-vaccination, but no striking differences were noted between days 14 and 35 post-vaccination (Fig. 5a). While these data indicated that Mt 10 virus has the ability to induce nABs, the possibility remained that qualitative variations may exist between different antibody isotypes specific to the virus.
To that end, we measured total Ig, IgG1, IgG2a, IgG2b, IgG3, IgM, IgA, and IgE by enzyme-linked immunosorbent assay (ELISA) using CVB3 VP1 as a specific viral antigen, with Keyhole limpet hemocyanin (KLH) serving as a negative control, and we made a few observations (Fig. 5b). (i) Serum from both saline and Mt 10 groups did not react to KLH for any of the isotypes indicated above, ruling out the possibility of non-specific reactivity. (ii) Reactivity of sera for IgE and IgA, and to a lesser extent for IgM except at day 35, was lacking for VP1 in the vaccine group, suggesting that these isotypes are not the major components of the antibody response against the virus. (iii) Vaccine recipients had elevated levels of VP1-reactive total Igs by day 14 post-vaccination, and their levels remained elevated up to day 35. (iv) As to the various IgG isotypes, they were elevated only in the vaccinated groups, but not in controls, and occurred in the order of IgG2a > > IgG2b > IgG3 > IgG1 (Fig. 5b). Since vaccine-induced antibody responses were dominated by IgG isotypes, the data pointed to a role for simultaneous induction of virus-specific T cell responses as T cell-derived cytokines are critical for isotype switching18.
T cell responses induced by Mt 10 virus are antigen-specific and involve mainly Th1 cytokines
We recently reported identification of two major CD4 T cell epitopes within CVB3 VP1, VP1 681–700, and VP1 721–740, and construction of MHC class II tetramers and dextramers to evaluate their antigen-specificity12. Using these tools, we analyzed virus-specific T cells in animals injected with saline or vaccinated with Mt 10 virus. In brief, a pool of spleens and lymph nodes, hereafter termed lymphocytes, harvested from these groups were stimulated with VP1 681–700 or VP1 721–740 peptides, and staining was performed by flow cytometry using IAk/VP1 681–700 dextramers and VP1 721–740 tetramers and their corresponding controls (Bovine ribonuclease [RNase] 43–56)12. As expected, CD4 T cells from saline-recipient animals did not bind VP1 681–700 or VP1 721–740 dextramers or tetramers, and their staining intensities were comparable to those of RNase 43–56 (control) (Fig. 6a, left panel). Similar analysis in vaccine recipients revealed staining of CD4 T cells for both VP1 681–700 dextramers (0.7 ± 0.2; p ≤ 0.05) and VP1 721–740 tetramers (1.9 ± 0.8; p ≤ 0.05), whereas staining for RNase 43–56 was negligible (Fig. 6a, right panel). While these data lend support for the occurrence of antigen-specific T cell expansion in response to vaccine virus, additional studies were needed to investigate the nature of cytokine responses.
We adopted multiplex bead array analysis to analyze cytokine responses in culture supernatants harvested from antigenic stimulations described above for a panel of T helper (Th) 1, Th2, and Th17, including interleukin (IL)-6, tumor necrosis factor (TNF)-α, and anti-inflammatory (IL-10) cytokines. The analyses revealed detection of mainly interferon (IFN)-γ and IL-2 [Th1], followed by IL-22, representative of Th17 subset, and IL-5, IL-4, and IL-13 (Th2) in cultures derived from vaccine-recipients as compared to saline-recipient animals without antigenic stimulations, suggesting that the virus-primed lymphocytes can spontaneously produce these cytokines (Fig. 6b). While IL-9, IL-17A, and IL-17F were consistently absent, other inflammatory cytokines (IL-6 and TNF-α) were also detected in low amounts in vaccine-recipients with a tendency for IL-10 to be high in the vaccine groups (Fig. 6b). However, upon stimulation with viral peptides (VP1 681–700 and VP1 721–740), we noted upregulation of only IFN-γ, indicating that the VP1-specific T cells capable of producing IFN-γ can form one component of virus-reactive cells. It is possible that the virus-sensitized cells may contain T cell specificities for other viral antigens that we have not examined in this study. Furthermore, we also examined a panel of anti-viral cytokines in serum samples collected at various time points (from day 0 to day 35) post-vaccination. These analyses revealed significant elevation of chemokines such as IFN-γ -induced protein (IP)-10 (p ≤ 0.0001), monocyte chemoattractant protein (MCP)-1 (p ≤ 0.001), and keratinocyte-derived chemokine (KC) (p ≤ 0.05) on day 4 after vaccination, whereas granulocyte–macrophage colony-stimulating factor (GM-CSF) (p ≤ 0.001) was detected at day 21 when compared with saline recipients (Fig. 7). In contrast, other cytokines (IFN-γ, TNF-α, IL-12p70, Regulated upon Activation, Normal T Cell Expressed and Presumably Secreted [RANTES], IL-1β, IL-10, IFN-α, IFN-β, and IL-6), although detected in varied amounts, were not significant between groups (Fig. 7). Taken together, our data suggest that the Mt 10 virus can induce antigen-specific T cell responses capable of producing mainly IFN-γ, a critical cytokine in fighting intracellular pathogens like viruses, in addition to promoting other anti-viral cytokines described above. While these components together with virus nABs might have contributed to disease protection mediated by the Mt 10 virus, safety concerns, if any, needed to be evaluated.
Animals vaccinated with Mt 10 virus did not reveal autoimmune response and cardiac injury
We had previously reported that CVB3 infection can lead to the induction of pathogenic T cells reacting to cardiac myosin as a secondary event10,19. To ensure that the Mt 10 virus is safe and does not induce side effects, we considered two readouts: assessment of cardiac myosin-reactive T cells to determine autoimmunity20, and measurement of cardiac troponin I (cTnI) as a cardiac injury marker21. Using T-cell proliferation assay, we noted myosin reactivity in animals infected with the wt CVB3, as indicated by the appearance of T cells responding to cardiac myosin heavy chain (Myhc)-α 334–352 (p ≤ 0.001) (Fig. 8a, left panel), an immunodominant epitope that induces autoimmune myocarditis in A/J mice10,20. Such reactivity was lacking in both vaccine-recipients (Fig. 8a, middle panel) and those challenged with wt CVB3 (Fig. 8a, right panel). Under similar conditions, we analyzed sera samples by ELISA for cTnI. As shown in Fig. 8b, cTnI was detected only in the wt CVB3-infected group (p ≤ 0.05), but not in the vaccine or vaccine/challenge groups. Taken together, the data suggest that the Mt 10 virus, despite being live attenuated, is unlikely to induce side effects described above, since cardiac injury is expected to occur as a consequence of direct virus-induced damage to the cardiac tissue22,23,24.
Vaccine-recipients are free of wt CVB3 in challenge studies
The disease course induced by CVB3 in infected mice assumes viral and non-viral phases that occur in continuum, with the possibility that viral nucleic acids can be present in chronically infected animals25. We sought to evaluate this phenomenon by looking for viral RNA in hearts and pancreata harvested from animals inoculated with wt CVB3, vaccine alone, and vaccine-challenged groups. This was achieved by quantitative/real-time RT-PCR using the probes and primers specific to CVB3-VP1. Expectedly, hearts and pancreata obtained from the wt CVB3 alone group contained CVB3 viral nucleic acids (p ≤ 0.01) (Fig. 9). Importantly, viral RNA was not detected in the vaccine group, including animals challenged with the wt CVB3 (Fig. 9), suggesting that immune responses generated by the Mt 10 virus may be sufficient to prevent infection by the wt CVB3.