Role of Genes That Modulate Host Immune Responses in the Immunogenicity and Pathogenicity of Vaccinia Virus

Abstract

Poxvirus vaccine vectors, although capable of eliciting potent immune responses, pose serious health risks in immunosuppressed individuals. We therefore constructed five novel recombinant vaccinia virus vectors which contained overlapping deletions of coding regions for the B5R, B8R, B12R, B13R, B14R, B16R, B18R, and B19R immunomodulatory gene products and assessed them for both immunogenicity and pathogenicity. All five of these novel vectors elicited both cellular and humoral immunity to the inserted HIV-BH10 env comparable to that induced by the parental Wyeth strain vaccinia virus. However, deletion of these immunomodulatory genes did not increase the immunogenicity of these vectors compared with the parental vaccinia virus. Furthermore, four of these vectors were slightly less virulent and one was slightly more virulent than the Wyeth strain virus in neonatal mice. Attenuated poxviruses have potential use as safer alternatives to current replication-competent vaccinia virus. Improved vaccinia virus vectors can be generated by deleting additional genes to achieve a more significant viral attenuation.

Poxviruses are attractive vaccine vectors, due in part to their ability to elicit strong and long-lasting CD4⁺ and CD8⁺ T-lymphocyte responses, as well as humoral immune responses (4, 12, 13, 17, 18, 23). However, replication-competent poxvirus vectors pose serious health risks because of their well-documented capacity to disseminate in immunosuppressed individuals (7, 9, 31, 41, 43). These risks have increased in recent years due to a rising proportion of the population living with immunosuppression resulting from human immunodeficiency virus (HIV) infection, organ transplantation, or cancer therapies. Poxviruses are being explored as vectors for vaccines against a variety of diseases, including AIDS and cancer. In addition, widespread public vaccination is being considered to counter the threat of a use of smallpox as a bioterrorist weapon. There is, therefore, a need for safer poxvirus vaccine vectors that are highly immunogenic but have minimal pathogenic potential.

Attenuated and host range-restricted poxviruses that undergo limited replication in human and other mammalian cell lines, such as modified vaccinia virus Ankara (MVA) and canarypox, have been developed as vaccine vectors for use in humans. These poxvirus vectors have been shown to be safe for use in humans (15, 21, 35, 39) and to induce protective immunity in animal models (8, 36, 38) and would therefore appear to represent the poxvirus vectors of choice for human vaccine development. However, the disappointing immunogenicity in human volunteers of the recombinant MVA-HIV vaccine constructs in the recently reported International AIDS Vaccine Initiative trials (W. Jaoko et al., AIDS Vaccines Int. Conf., abstr. 2004) as well as the high-dose recombinant canarypox-HIV administration in the HIV Vaccine Trials Network trials (P. Goepfert et al., 10th Conf. Retrovir. Opportun. Infect., abstr. 82, 2003) suggest that more potent recombinant poxvirus constructs will be needed for the production of an effective HIV vaccine. Therefore, we engineered recombinant vaccinia virus vectors by selectively deleting portions of the vaccinia virus genome in an attempt to create attenuated vectors that retain immunogenicity for use as HIV vaccines.

Recent studies have demonstrated that poxvirus genomes encode a variety of proteins that are associated with immune function. These proteins include homologs of mammalian proteins that are involved in the induction of and/or activity of various immune system components and are thus likely to contribute to the immunosuppressive and pathogenic properties of poxviruses. Moreover, it has been proposed that poxviruses may be made more immunogenic through the deletion of some or all of these immune-related genes (1, 14, 52, 53). To evaluate these hypotheses, we created a series of novel vaccinia viruses through systematic inactivation of vaccinia virus gene products that specifically interact with immune functions. Attractive targets for this strategy included the vaccinia virus genes that encode a gamma interferon (IFN-γ) receptor (2, 37), serine protease inhibitors (serpins) (10, 27, 29, 48), complement regulators (19, 24, 28), protein kinases (6), and cytokine receptors (1, 11, 47, 49-51).

In the present study, we deleted poxvirus B5R, B8R, B12R, B13R, B14R, B16R, B18R, and B19R from a Wyeth strain vaccinia virus that contained HIV-1 BH10 env (see Table 1 for a description of deleted genes and their cellular homologs). We then quantified antivector and anti-Env immune responses as well as the pathogenicity of each vector in mice with the hope of identifying new, safer poxvirus constructs that might have utility as vaccine vectors and smallpox vaccines.

TABLE 1.

Recombinant poxviruses carrying deletions in the HindIIIB region

Virus	Deletion size (kbp)	Genes deleted	Immunoregulatory gene homologies (references)
vT273	No deletion	None	Parental Wyeth strain; expressing HIV-1 BH10 env
vT134	No deletion	None	Negative control containing lacZ, no HIV-1 env
vT287	2.5	ΔB5R-B8R	B5R: regulator of complement activation; membrane glycoprotein (19, 20, 24, 25)
			B8R: IFN-γ receptor (2, 37)
vT285	5.3	ΔB8R-B15R	B8R: IFN-γ receptor (2, 37)
			B12R: serine/threonine protein kinases (6)
			B13R: IL-1β convertase inhibitor, SERPIN (10, 27, 29, 48)
			B14R: IL-1β convertase inhibitor, SERPIN (10, 27, 29, 48)
			B15R: IL-1 and IL-6 receptor (1, 47, 49)
vT290	7.5	ΔB5R-B15R	Encompasses vT287 + vT285 deletions
vT281	4.0	ΔB16R-B19R	B16R: IL-1 receptor (1, 49)
			B18R: IFN-α receptor (11)
			B19R: IL-1 and IL-6 receptor, IFN inhibitor (47, 51)
vT284	8.9	ΔB8R-B19R	Encompasses vT285 + vT281 deletions

Recombinant viral constructs.

The recombinant vaccinia virus vT273 expressing HIV-1 BH10 env (gp160) was constructed by inserting the env sequence into a nonessential site in the HindIII M region (16) of the TBC-33 strain of vaccinia virus by in vivo recombination as previously described (33). In the vT273 construct, env expression is controlled by the vaccinia virus 40K (H5R) promoter (22, 42). The control recombinant vaccinia virus vT134 contains no env insertion, but does contain the lacZ gene, and thus can be used as a negative control virus when evaluating Env-associated viral effects.

The deletion mutants vT287, vT285, vT290, vT281, and vT284 were engineered from vT273 to contain a series of gene deletions in the HindIIIB region (Table 1) by replacing the native vaccinia sequences with the Escherichia coli lacZ gene under control of the C1 promoter (26). The deleted gene products included a variety of immunomodulatory proteins, some with homology to serine protease inhibitors (serpins) and the gamma interferon (IFN-γ) receptor. Deleted genes were located in the B5R-B19R region of the vaccinia virus genome (Table 1). The genomic structure of these recombinant viruses was confirmed by PCR (data not shown).

Expression of Env gp160 by each gene-deleted vaccinia construct was demonstrated by in situ immunostaining using an anti-gp41 antibody (clone Chessie 8; data not shown) and by soluble CD4 enzyme-linked immunosorbent assay. Samples were prepared by infecting 3 × 10⁶ BSC-40 cells with each recombinant viral construct or vaccinia virus vAbT-33 (negative control) at a multiplicity of infection of 10. Following a 24-hour infection, the cells and culture medium were harvested and separated by centrifugation. The culture medium (supernatant) was analyzed for expression levels of Env. The cell pellets were lysed in 0.5 ml of MPER buffer (Pierce Biotechnology, Inc., Rockford, IL), and insoluble debris were pelleted by centrifugation. The resulting supernatant (lysate) was analyzed for Env expression.

The soluble enzyme-linked immunosorbent assay was conducted by coating 96-well plates with 0.05 μg of CD4 (DuPont NEN, Boston, MA). The next day, standards (native HIV-1 IIIB gp120, ABI), and samples (supernatants and lysates) were incubated for 1 h at 37°C. Following the incubation, HIV-1 human serum (Scripps Laboratories, San Diego, CA) biotinylated in-house was added and incubated at 37°C for 1 h, plates were developed using horseradish peroxidase-streptavidin and tetramethylbenzidine microwell substrate (KPL, Inc., Gaithersburg, MD). Vaccinia virus vAbT-33 was used as the negative control in this assay. As shown in Table 2, no major differences in Env gp160 expression levels were detected among the various recombinants.

TABLE 2.

HIV-1 gp120 expression by gene-deleted vaccinia virus constructs^a

Virus	gp120 (ng)
	Lysate (ng)	Sup (ng)
vT273	901	285
vT287	526	257
vT285	895	414
vT290	356	219
vT281	1403	387
vT284	951	432
vAbT-33	—b	—

^aWe infected 3 × 10⁶ BSC-40 cells with each recombinant viral construct for 24 hours. Expression of gp120 was measured in culture medium (sup) and lysed cell pellets (lysate).

^b—, below detection.

In vivo immunogenicity.

Female BALB/c mice (8 to 10 weeks old; Charles River Laboratories, Cambridge, MA) were maintained in accordance with the guidelines of the Committee on Animals for the Harvard Medical School and the Guide for the Care and Use of Laboratory Animals. Mice were immunized with 50 × 10⁶ PFU of each recombinant vaccinia virus construct (Table 1; n = 4 per group) by the intraperitoneal route on days 0 and 56.

Levels of circulating anti-gp120 and antivector antibodies elicited by each gene-deleted vaccinia virus in the peripheral blood of mice 0 and 4 weeks after the prime inoculation as well as 2 and 15 weeks after the boost inoculation were assessed by enzyme-linked immunosorbent assay (Fig. 1). Anti-gp120 and antivector antibody responses in naïve animals are not shown, as all absorbances from these serum samples fell below the cutoff used for the determination of antibody titer values in immunized animals. Despite significant modification of the viral genome by deletion of genes previously reported to affect host immune responses, the anti-gp120 and antivector antibody responses detected in each of the experimentally immunized groups were similar to those detected in the mice immunized with the parental vT273 (Fig. 1; exact Wilcoxon rank sum test; all P > 0.1 except as discussed below).

FIG. 1.

Serum anti-vaccinia virus (A) and anti-gp120 (B) antibody titers in mice following immunization with gene-deleted recombinant vaccinia virus constructs. Serum samples were obtained at 0 weeks, 4 weeks after prime (pi), and at 2 and 15 weeks postboost (pb) immunization with 50 × 10⁶ PFU vaccinia by the intraperitoneal route. All anti-gp120 and antivector antibody responses in naïve (week 0) animals are not shown, as all absorbances from these serum samples fell below the cutoff used for the determination of antibody titer values in immunized animals. Antibody titers are expressed as the geometric mean (± standard error) of 4 mice per group. Symbols are as defined in the figure key.

However, the antivector antibody responses detected in the groups immunized with vT281 (ΔB16R-B19R; P = 0.03), vT284 (ΔB8R-B19R; P = 0.03), and vT290 (ΔB5R-B15R; P = 0.02) were lower than the responses detected in mice immunized with vT273 at week 4 postinoculation. This may indicate slower kinetics of the humoral immune response to the vaccinia virus in mice immunized with these vectors compared with the parental vector. Further, the antivector antibody responses detected in the groups immunized with vT281 (ΔB16R-B19R; P = 0.08 at week 2 postboost) and vT284 (ΔB8R-B19R; P = 0.05 at week 2 postboost and P = 0.03 at week 15 postboost) continued to be lower than those detected in mice immunized with the parental vT273 following boost immunization (Fig. 1). Meanwhile, with the exception of mice immunized with vT290 at week 4 postinoculation (P = 0.03), the anti-gp120 antibody responses elicited in all the groups of mice experimentally immunized were comparable to responses elicited in mice immunized with the parental vT273 at all time points assayed.

CD8⁺ T lymphocytes elicited by each recombinant vaccinia virus construct were assessed (Fig. 2) by tetramer staining followed by flow cytometric analysis of peripheral blood samples. Briefly, red blood cells from approximately 100 μl of whole blood were lysed in a solution of NH₄Cl-Tris. Peripheral blood mononuclear cells were then washed and stained with 0.1 to 0.2 μg phycoerythrin (PE)-labeled H-2D^d/p18 tetramer complexes in conjunction with anti-CD8α-PE-Cy5 monoclonal antibodies (CT-CD8a; Caltag, Burlingame, CA) as previously described (44). This allowed the detection of CD8⁺ T lymphocytes that recognize the H-2D^d-restricted dominant p18 epitope (RGPGRAFVTI), derived from the V3 loop of HIV-1 BH10 Env (40, 46). The PE-labeled H-2D^d/p18 tetramer complexes were constructed as described (3, 30). Washed peripheral blood mononuclear cells were fixed in 0.5 ml of phosphate-buffered saline containing 1.5% formaldehyde and analyzed on a Coulter EPICS XL (Beckman Coulter, Fullerton, CA).

FIG. 2.

Immunogenicity of novel recombinant vaccinia virus constructs after prime (day 0) and boost (day 56) vaccinations with 50 × 10⁶ PFU by the intraperitoneal route. Peripheral blood samples were stained with H-2D^d/p18 tetramer and anti-CD8 antibody and analyzed by flow cytometry. Levels of p18/Env-specific CD8⁺ cells are expressed as the median (± standard error) of 4 mice per group. Symbols: Δ, vT134; *, vT273; ▴, vT287; •, vT285; ▪, vT290; ♦, vT281; —, vT284.

The Env-specific CD8⁺ cellular responses elicited by these viruses were comparable after priming immunization (Fig. 2; exact Wilcoxon rank sum test; all P > 0.1 at days 7 and 49 postinoculation). Following boosting, vT290 (ΔB5R-B15R) only elicited lower levels of CD8⁺ T lymphocytes (P = 0.03 at days 7 and 53 postboost) compared with mice immunized with the parental vT273 (Fig. 2). Levels of CD8⁺ T lymphocytes elicited after boost in mice immunized with vT287 (ΔB5R-B8R) were not statistically different (P > 0.1 at days 7 and 53 postboost) than those elicited in mice immunized with the parental vT273 (Fig. 2).

In vivo pathogenicity.

Having shown that these gene-deleted viral vectors exhibited little to no loss of immunogenicity (Fig. 1 and 2), we evaluated whether these gene deletions resulted in the attenuation of their pathogenicity. For these studies, weanling BALB/c AnNTac mice (n = 10 per group; Taconic, Germantown, N.Y.) were inoculated intracranially with each recombinant virus (Table 1) in serial dilutions to determine the 50% lethal dose (LD₅₀) over a period of 13 days. All the recombinant viruses were able to kill the mice at the highest dose tested between 3 and 5 days after inoculation. The number of live and dead mice was recorded and the LD₅₀ was determined. The LD₅₀s calculated from one experiment are shown in Table 3. All gene-deleted vectors but vT281 exhibited modestly (<1 log) reduced pathogenicity compared to the parental vT273. Similar results were obtained in a second experiment (data not shown).

TABLE 3.

LD₅₀ of gene-deleted vaccinia virus constructs^a

Virus	LD50 (PFU)
vT273	108.29
vT287	109.00
vT285	108.71
vT290	109.15
vT281	108.06
vT284	108.87

^aWeanling BALB/c AnNTac mice (n = 10 per group) were inoculated intracranially with recombinant virus in serial dilutions and observed daily for death until day 13.

A subtle association was seen in this study between the pathogenicity and immunogenicity of these gene-deleted constructs. The lowest LD₅₀ was seen for vT290 (ΔB5R-B15R). This virus elicited cellular anti-Env responses comparable to the wild-type vT273 after a priming immunization. However, after boosting, p18-binding CD8⁺ cell levels in mice immunized with vT290 were significantly lower than in mice immunized with vT273. Conversely, although lower after priming, humoral responses in mice immunized with vT290 were comparable to those in mice immunized with vT273 after boost. These results, when contrasted with the LD₅₀ and magnitude of immune responses seen in mice inoculated with vT285 (ΔB8R-B15R) that were comparable to those seen following vT273 inoculation, may implicate B5R, a complement activation regulatory protein homolog (19, 20, 24), as a stronger influence on host immune function and vector pathogenicity than are B8R, B12R, B13R, B14R, and B15R. The deletion of immunomodulatory genes in vT290 in addition to B5R may have increased the attenuation of this vector.

The present findings do not support the contention that the immune-related genes carried by vaccinia virus contribute to decreasing the immunogenic properties of the virus. Thus, deletion of B5R, B8R, B12R, B13R, B14R, B16R, B18R, and B19R gene products does not improve the capacity of vaccinia virus to augment immune responses to a heterologous HIV-1 Env protein carried by the virus. This is perhaps not surprising, since these specific gene deletions do not likely change the kinetics of vaccinia virus replication in vivo or the expression of the heterologous gene. In fact, we have recently shown that vaccinia virus is comparable in its immunogenicity in rhesus macaques to MVA (45), an attenuated vaccinia virus that was derived by over 500 serial passages of the Ankara strain on primary chicken embryo fibroblasts, resulting in multiple genomic deletions totaling approximately 31 kb (5, 34).

The observations in this study also do not argue for a role of these immune-related genes in the pathogenic potential of vaccinia virus. While a reduction in the pathogenicity of selected gene-deleted vaccinia virus constructs was seen in the present experiments, this reduction was modest in magnitude. It is possible that other in vivo assays for pathogenicity might uncover differences between these new vaccinia virus constructs and the wild-type virus. In fact, Legrand et al. (32) have recently reported that nude mice infected with vaccinia viruses with focused serpin-related gene B13R and B22R deletions have a longer survival than nude mice infected with wild-type virus. These same mutants manifested lower virus titers and weight loss in vivo in immunocompetent CB6F₁ mice as well (32). In that work, vaccinia virus strain WR was used, which is more pathogenic than the Wyeth strain which has been used in the present study. This may also account for the higher pathogenicity differences observed by Legrand et al. (32). Nevertheless, the present study suggests that any such decrease in pathogenicity may be modest in other vaccinia strains and/or model-dependent.

Why poxviruses have picked up a variety of host genes and continued to carry these genes is not readily apparent. These genes may confer subtle selective advantages for the viruses, but the nature of those advantages is not clear. Continued study of this important vaccine and immunizing vector will be important to improve its immunogenicity and pathogenic profile.