The E3L gene of vaccinia virus contributes to its pathogenesis [27
]. Both the N and C terminal domains of the E3L protein have been found to be necessary for full virulence in mouse models [27
]. In this report we demonstrate that when administered by scarification, viruses containing mutations in E3L can induce potent Th1 dominated immune responses that are protective against intra-nasal challenge with wtVACV. These viruses are highly attenuated when administered by scarification even in SCID mice, with VACVΔE3L and VACV-vIF2αH, causing little detectable pathology or disease even after 6 months of infection. Thus, these viruses have the potential for use as replication-competent, attenuated vaccines for protection against smallpox and as replication-competent, attenuated vaccine vectors for protection against heterologous diseases.
We have evaluated four viruses for pathogenicity and efficacy: VACVE3LΔ7C, VACVE3LΔ54N, VACVΔE3L and VACV-vIF2αH. VACVE3LΔ7C, VACVE3LΔ54N are replication competent, partially attenuated viruses. Both replicate to near wt titers in the nasal mucosa [22
] and in the dermis, but are attenuated by 4–5 logs after IN infection of wt mice [22
]. Both have reduced virulence in SCID mice, when infected either IN [22
]or by scarification, and both replicate to intermediate titers in ovaries after IP infection. For VACVE3LΔ7C this is likely due to the decrease in affinity that E3LΔ7C protein has for dsRNA [17
]. The properties of VACVE3LΔ54N are likely due to a combination of properties of E3LΔ54N: lack of a functional Z-DNA-binding domain [27
], and rapid turn-over of E3LΔ54N protein (). Induction of pathogenesis after a long latent period by VACVE3LΔ54N suggests that this virus is replicating at low level for a long period of time in the SCID mice. Such prolonged replication could lead to selection of a more pathogenic variant. Since E3L proteins with larger deletions in the N-terminus than Δ54N are stable, a mutation that yields a stable protein could lead to a more pathogenic virus that could have caused disease after a long latent period. Both VACVΔE3L and VACV-vIF2αH, are much more highly attenuated than VACVE3LΔ7C and VACVE3LΔ54N. Neither virus caused detectable weight loss in SCID mice after IN [22
] or scarification infection, even up to six months after infection. VACVΔE3L was not detectable in ovaries after IP infection. For VACV-vIF2αH, virus was detected in the ovaries of only one of three mice, although at 6 logs lower titer than wtVACV. Both viruses did replicate either after IN infection [22
] or scarification infection, although titers were 3–4 logs lower than in wtVACV infected mice. The detection of VACVΔE3L replication in the nasal mucosa and skin but not in ovaries is surprising, since ovaries are thought to be the most sensitive organ to detect VACV replication. It is possible that the unique sensitivity of VACVΔE3L and VACV-vIF2αH to IFN is compromising replication in the ovaries to a greater extent than in the nasal mucosa and in the skin. While the ATV eIF2α homologue partially complements deletion of E3L in cells in culture (Talasela et al., manuscript in preparation), there was very little decrease in attenuation detected in virus containing the eIF2α homologue in place of E3L. Again, it is possible that the IFN-sensitivity of VACV-vIF2αH (Talasela et al., manuscript in preparation), may be limiting virus replication and pathogenicity in the whole animal.
Somewhat surprisingly, all of the attenuated viruses tested in this study produced detectable skin reactions (takes) that no matter how mild, were protective against challenge with wtVACV. All of the viruses produced takes in all of the animals tested, when administered by scarification at 106 pfu. At 105 pfu all of the viruses except VACVΔE3L produced takes in all of the animals tested: VACVΔE3L produced takes in 7 of 10 animals vaccinated at this dose. The severity of the take directly correlated with pathogenicity of the virus tested and the extent of virus replication in the dermis, with the order being: wtVACV>VACVE3LΔ7C>VACVE3LΔ54N>VACV-vIF2αH=VACVΔE3L. For VACVΔE3L, which gave takes on about 70% of the animals at 105 pfu, those animals that had a take were fully protected, while the animals that didn’t have a take were largely unprotected. Thus, the very mild skin lesions produced by the highly attenuated viruses, VACVΔE3L and VACV- vIF2αH, were fully protective against a challenge with wtVACV, even when the animals were challenged three months after vaccination (data not shown).
In order to correlate protection with immune response, we characterized induction of neutralizing antibodies, comet formation inhibiting antibodies and induction of cell mediated immunity. At 2 weeks post vaccination, high titers of neutralizing antibodies were detected in serum from animals immunized with wtVACV and VACVE3LΔ7C. The less pathogenic viruses, VACVΔE3L, VACV-vIF2αH and VACVE3LΔ54N, as well as wtVACV (NYCBH), produced lower levels of neutralizing antibodies than VACVE3LΔ7C(WR) and wtVACV(WR). The ability to induce neutralizing antibodies correlated best with the replication capacity of these viruses, as measured either on the skin after scarification, or in ovaries after IP infection. Thus, it may be that the dose of viral antigen is the key determinant of induction of neutralizing antibodies. However, infection with the more highly attenuated viruses apparently primed for induction of both neutralizing and comet formation inhibiting antibodies, because high titers of these antibodies were detected after a single boost with VACVΔE3L. Despite inducing a poor primary antibody response these viruses all protected against challenge with wtVACV. Thus, either the low levels of neutralizing antibodies were sufficient for protection, or the small number of antibody producing cells generated during primary vaccination were sufficient to expand and protect the animals upon challenge, or other components of the immune system, likely cell mediated immunity, were compensating for this low antibody response. Interestingly, B cell deficient mice vaccinated with E3L deletion mutant viruses are also protected against challenge with pathogenic vaccinia virus despite an inability to produce antibodies. These results are in contrast to other previously reported results [34
], demonstrating a critical role for antibodies in protection against challenge of mice with vaccinia virus.
To begin to evaluate induction of cell mediated immunity, we measured cytokines released from ex vivo restimulated splenocytes from vaccinated animals. These assays provided the first information that would begin to explain why VACVΔE3L and VACV-vIF2αH are protective in vaccination experiments despite the absence of substantial levels of neutralizing antibodies. Restimulated splenocytes from mice vaccinated with viruses deleted of all or part of the E3L gene produced roughly twice as much IFN-γ and TNF-α as splenocytes from wtVACV vaccinated mice. Splenocytes from VACVΔE3L vaccinated mice also secreted more IL-2. Splenocytes from all of the vaccinated animals secreted very low levels of IL-4 and IL-5. Thus, these highly attenuated viruses induce potent Th1 dominated immune responses.
Intracellular cytokine staining was also used to identify which cell types were activated during vaccination. Animals vaccinated with either wtVACV or any of the attenuated viruses described in this manuscript had similar percentages of CD4+/IFNγ+ T cells (about 5% of CD4+ T cells), and similar percentages of CD8+/IFNγ+ T cells (about 10% of the CD8+ T cells). The fact that these experiments showed similar percentages of IFN-γ positive CD4+ and CD8+ cells from the spleens of both VACVΔE3L vaccinated mice and wtVACV vaccinated mice indicates that it is possible that T cells are not the only cells from the spleens of vaccinated mice that are producing IFN-γ, or that T cells from VACVΔE3L vaccinated mice are secreting more IFNγ on a per cell basis than splenocytes from wtVACV vaccinated animals.
VACVΔE3L induces a potent Th1 dominated cell mediated immune response despite replicating poorly on the skin of mice. We believe this is due to induction of pro-inflammatory signal transduction and pro-inflammatory gene expression in cells infected with VACVΔE3L [36
]. The E3L gene products can bind to double helical nucleic acids and sequester viral induced PAMPs. Thus, infection with wtVACV does not lead to activation of the p38 MAP kinase pathway, and does not lead to activation of the IKK complex [36
]. Consistent with this, we detected no activation of the pro-inflammatory transcription factors ATF-2, NF-κB or IRF-3 and very little increased pro-inflammatory gene expression in cells infected with wtVACV [36
]. Virus deleted for E3L leads to signaling through the p38 MAP kinase pathway leading to phosphorylation of ATF-2, leads to phosphorylation of IKKα/βand IKKε, and leads to nuclear accumulation and activation of both NF-κB and IRF-3 [36
]. Microarray experiments have shown that infection with VACVΔE3L induces upregulation of a large number of proinflammatory gene products [36
]. We believe that this massive increase in virus-induced signaling leading to activation of proinflammatory transcription factors and proinflammatory gene expression is responsible for the efficacy of VACVΔE3L as a vaccine despite its attenuation and limited replication capacity in infected mice.
Most of the work described in this manuscript was done with the WR strain of VACV. VACV(WR) is a highly neurovirulent, mouse adapted strain of VACV, and while it is relevant for studies in the mouse model, it is not appropriate for use in humans [23
]. To test attenuation and efficacy of attenuating mutations in E3L in strains of vaccinia virus more appropriate for use in humans, experiments were performed to test vaccination efficacy and immunological characteristics in two additional strains. The strains chosen were the Copenhagen and New York City Board of Health strains, both of which have been previously used in human vaccination campaigns [1
]. These strains were found to more attenuated than the VACV(WR) in the mouse models tested [22
]. Introducing the E3L deletion mutation into these viruses led to further attenuation in mice. VACVΔE3L in both a Copenhagen and a NYCBH background were fully able to induce very mild protective pocks in vaccinated mice. As was seen with VACVΔE3L(WR), VACVΔE3L(NYCBH) did not induce the synthesis of neutralizing antibodies, but did prime for synthesis of neutralizing antibodies. Splenocytes from VACVΔE3L(NYCBH) vaccinated mice also secreted somewhat more IFNγ, similar to splenocytes from VACVΔE3L(WR) vaccinated mice.