|Home | About | Journals | Submit | Contact Us | Français|
Recombinant rabies virus (RV)-based vectors have demonstrated their efficacy in generating long-term, antigen-specific immune responses in murine and monkey models. However, replication-competent viral vectors pose significant safety concerns due to vector pathogenicity. RV pathogenicity is largely attributed to its glycoprotein (RV-G), which facilitates the attachment and entry of RV into host cells. We have developed a live, single-cycle RV by deletion of the G gene from an RV vaccine vector expressing HIV-1 Gag (SPBN-ΔG-Gag). Passage of SPBN-ΔG-Gag on cells stably expressing RV-G allowed efficient propagation of the G-deleted RV. The in vivo immunogenicity data comparing single-cycle RV to a replication-competent control (BNSP-Gag) showed lower RV-specific antibodies; however, the overall isotype profiles (IgG2a/IgG1) were similar for the two vaccine vectors. Despite this difference, mice immunized with SPBN-ΔG-Gag and BNSP-Gag mounted similar levels of Gag-specific CD8+ T-cell responses as measured by major histocompatibility complex class I Gag-tetramer staining, gamma interferon-enzyme-linked immunospot assay, and cytotoxic T-cell assay. Moreover, these cellular responses were maintained equally at immunization titers as low as 103 focus-forming units for both RV vaccine vectors. CD8+ T-cell responses were significantly enhanced by a boost with a single-cycle RV complemented with a heterologous vesicular stomatitis virus glycoprotein. These findings demonstrate that single-cycle RV is an effective alternative to replication-competent RV vectors for future development of vaccines for HIV-1 and other infectious diseases.
The global spread of HIV-1 represents one of the most significant pandemics to afflict humans (22). Despite tremendous efforts to increase HIV awareness in the general population, UNAIDS reports that fewer than one in five people has access to HIV prevention strategies and many are subject to cultural stigmas thwarting such efforts (43). As such, an HIV vaccine is paramount for preventing disease transmission. It is not yet clear precisely what characteristics are critical for an effective HIV vaccine, yet evidence suggests one would need to induce both antibody and CD8+ T-cell-mediated immunity (reviewed in reference 25). Live viruses are at the forefront of HIV vaccine development (7) because they are powerful inducers of both of these arms of immunity. We previously demonstrated that replication-competent rabies virus (RV)-based vectors can induce long-lasting antigen-specific immune responses in both murine and monkey models, as well as protect rhesus macaques from an AIDS-like disease (23, 24, 26-29, 42). However, there are safety concerns with the use of any replication-competent virus for widespread immunization. To address this, we sought to develop and evaluate the immunogenicity of a safer alternative: a single-cycle RV expressing HIV-1 Gag as a model antigen.
Single-cycle viral vectors are defective in certain viral components that are required for infectious particle assembly (reviewed in reference 12). As such, the virus undergoes one complete cycle of replication in the primary infected cell and produces progeny virions that are unable to spread to a second round of cells. The progeny are noninfectious and provide inert antigen that may or may not be immunogenic (12). In contrast, so-called replication-deficient viruses do not complete that initial round of replication. These two attenuation strategies have been adopted for use with many different viruses including, but not limited to, adenovirus (Ad), vaccinia virus (VV), canarypox virus (CPV), herpes simplex virus (HSV), vesicular stomatitis virus (VSV), and, more recently, RV (4, 6, 9, 18, 21, 33, 35, 36, 38). However, the results regarding the immunogenicity of such vectors are mixed. For example, both the replication-deficient Ad5 vector and modified vaccinia Ankara (MVA) showed reduced humoral and cellular immunogenicity compared to their replication-competent counterparts, but the use of higher titers and multiple immunizations did increase such responses (18, 33, 35). In the case of CPV, the replication-deficient vector provided poor HIV-specific cellular responses, causing the termination of phase II HIV-1 vaccine trials (38). In contrast, single-cycle VSV, a rhabdovirus closely related to RV, has been shown to induce HIV-1 Env-specific CD8+ T-cell responses equivalent to full-length VSV when administered intramuscularly (36). However, protection of rhesus macaques against highly pathogenic simian immunodeficiency virus (SIV) challenge by both replication-competent and single-cycle VSV needs to be shown.
In the study described here, we generated a single-cycle RV vector expressing HIV-1 Gag (SPBN-ΔG-Gag) by deletion of the entire RV glycoprotein (RV-G) from the RV genome. RV-G was chosen due to its critical role in the attachment and entry of RV into host cells, which makes RV-G one of the most important determinants of viral pathogenicity (10, 11, 37). RV particles lacking G are unable to spread, as evidenced by intracranial infection with a G-deleted RV that remains restricted to the primary infected neurons (13, 44). It must be noted that in the absence of RV-G, virions are still capable of budding though at a 30-fold lower efficiency (32). These virions, however, are incapable of attachment and entry into a secondary host cell. Because of this, SPBN-ΔG-Gag was propagated on a trans-complementing cell line induced to express RV-G (or VSV-RV-G), effectively facilitating virus spread. To evaluate the immunogenicity of the single-cycle vector, we immunized mice and compared the humoral and cellular responses to responses generated by replication-competent RV. Our results indicate that single-cycle RV generates reduced vector-specific antibody responses but similar HIV-1 Gag-specific CD8+ T-cell responses. Moreover, these responses can be significantly enhanced by a heterologous boost with a single-cycle RV complemented with a VSV glycoprotein. Taken together, the results presented here show evidence that single-cycle RV is a promising platform for a safe, live viral vaccine for use against HIV-1 and other applications.
The plasmid encoding SPBN-Gag, described previously (27), was used to synthesize pSPBN-ΔG-Gag. pSPBN-Gag was digested with SmaI and PacI to remove the entire RV-G gene. Incubation with Klenow fragment was followed by blunt end ligation, yielding pSPBN-ΔG-Gag (Fig. (Fig.1).1). The construction of pBNSP-Gag has been described previously (26) (Fig. (Fig.11).
Recombinant RV was recovered as previously described (15) with the following modifications for single-cycle RV recovery. Briefly, BSR cells were transfected using FuGENE 6 (Roche Diagnostics) with 5 μg of pSPBN-ΔG-Gag cDNA, 1.5 μg of T7, 2.5 μg of pTIT-N, 1.25 μg of pTIT-P, 1.25 μg of pTIT-L, and 1 μg of pTIT-G (per six-well plate). Four days posttransfection the supernatant was transferred to induced BSR-RV-G cells or BSR-VSV-RV-G cells for passage to higher titers in serum-free medium. These viruses are complemented with RV-G or VSV-RV-G from the cell lines on which they are passaged and are referred to as SPBN-ΔG-GagRVG and SPBN-ΔG-GagVSV-RVG, respectively. Infectious virus titers were determined by focus-forming assay, as previously described (37). Titers of stocks of single-cycle virus were determined in parallel on noncomplementing BSR cells to verify that they were restricted to single cell infections.
We used the Tet-off inducible gene expression system (Clontech Inc.) to generate a BSR cell line (a BHK-21 cell clone) stably expressing RV-G (BSR-RVG cells). Briefly, RV-G was cloned into the pTRE2 response plasmid (Clontech Inc.) utilizing the MluI and NheI restriction enzyme sites, yielding pTRE2-RVG. BSR-RVG cells were generated by cotransfection of BSR cells using FuGENE 6 (Roche Diagnostics) with the pTet-Off transcriptional activator plasmid (Clontech Inc.) and the pTRE2-RVG plasmid. Cell clones resistant to G418 were selected and maintained in Dulbecco's modified Eagle's medium (DMEM) containing 1 μg/ml doxycycline to suppress RV-G transcription and treated with 1 mg/ml G418 at every third passage.
Similarly, VSV-RVG cells were generated by cotransfection of BSR cells with pTet-Off and pTRE2hyg (Clontech Inc.) encoding a VSV-RV-G chimera (pTRE2hyg-VSVRVG). The chimeric VSV-RV-G is composed of the ectodomain (ED) and transmembrane domain (TM) of VSV-G, Indiana strain, fused to the RV-G cytoplasmic domain (CD), as described previously (17, 29). Cell clones resistant to G418 and hygromycin were selected and maintained in DMEM containing 1 μg/ml doxycycline and treated with 1 mg/ml G418 and 250 μg/ml hygromycin selection agents at every third passage.
Anti-RV-G surface staining was used to confirm RV-G expression by the inducible BSR-RVG cell line. Cells were washed three times in PBS to remove all traces of doxycycline, cultured in doxycycline-free medium for 48 h, and fixed with 4% paraformaldehyde (Cytofix; BD Biosciences) for 30 min. Fixed cells were washed with 10 mM glycine-phosphate-buffered saline (PBS), blocked with 1% bovine serum albumin for 1 h, and stained with polyclonal rabbit anti-RV-G, followed by secondary staining with Cy-2-conjugated donkey anti-rabbit IgG (Jackson Immunoresearch). Similarly, cell surface expression of VSV-G was confirmed by staining with two monoclonal mouse anti-VSV-G antibodies (I1 and I14 ), followed by secondary staining with Cy-2-conjugated donkey anti-mouse IgG (Jackson Immunoresearch).
To quantify the relative production of p24 by each viral vector, BSR cells were infected at a multiplicity of infection (MOI) of 10. Seventy-two hours later, lysate was collected as previously described (27), and p24 was quantified using a p24 antigen enzyme-linked immunosorbent assay (ELISA) kit (ZeptoMetrix, Inc.), as described by the manufacturer.
Spread assays were conducted to analyze the capacity of the single-cycle virus to multiply on trans-complementing BSR-RVG or BSR-VSV-RVG cells, as well as on noncomplementing BSR cells. The complementing cell lines were induced by removal of doxycycline, followed by infection with the respective RV at an MOI of 0.001. After 2 h, the virus was removed, cells were washed one time in PBS, and either doxycycline-containing or doxycycline-free medium was replenished. After incubation for 72 h at 34°C, cells were fixed with 4% paraformaldehyde (Cytofix; BD Biosciences) for 30 min, and viral antigen was detected with fluorescein isothiocyanate (FITC)-conjugated anti-RV nucleoprotein (RV-N) monoclonal antibody (Centocor) using a fluorescence microscope. Expression of HIV-1 Gag was confirmed by focus-forming assay, as previously described (37). Briefly, cells were fixed with 80% acetone for 30 min and stained with monoclonal human anti-p24 antibody (71-31; NIH AIDS Research and Reference Reagent Program) at room temperature for 1 h, followed by secondary staining with Cy-2-conjugated donkey anti-human IgG (Jackson Immunoresearch).
The BSR-RVG or BSR-VSV-RVG cell line, induced, as described above, in parallel with BSR cells, was infected at an MOI of 0.01. After 2 h the virus was removed, cells were washed one time in PBS, and doxycycline-free medium was replenished. Supernatants were sampled at the time points indicated in the figures and virus titers were determined as described above.
Mice were maintained at the Thomas Jefferson University Animal Facilities, and all experimental procedures were approved by the Institutional Animal Care and Use Committee. In vivo studies were conducted using 6- to 8-week-old female BALB/c mice immunized (for prime or prime-boost experiments) intramuscularly (i.m.) in the leg muscle at the time points and concentrations of virus indicated in the figures. To study the recall immune responses, mice rested at least 4 weeks postimmunization were inoculated intraperitoneal (i.p.) with 1 × 106 PFU recombinant vaccinia virus expressing HIV-1 Gag (VV-Gag) (NIH AIDS Research and Reference Reagent Program). The induced immune responses were studied by sacrificing the mice at the time points indicated in the figures postimmunization/challenge and aseptically harvesting and homogenizing the spleens into a single-cell suspension in splenocyte medium (RPMI 1640 medium supplemented with 10% fetal bovine serum [FBS], penicillin-streptomycin, 10 mM HEPES, and 50 μM β-mercaptoethanol). Red blood cells were lysed using ACK lysing buffer (Gibco), and splenocytes were counted by trypan blue exclusion. Spleen samples were kept separate for analysis. Serum was collected at the time points indicated in the figures postimmunization for antibody analysis by ELISA.
Isolated splenocytes were blocked for 1 h at 4°C with Fc Block (rat anti-mouse CD16/CD32; BD Biosciences) and unconjugated streptavidin. Cells were washed with fluorescence-activated cell sorting (FACS) buffer (2% bovine serum albumin [BSA]-PBS) and stained for surface markers for 30 min at room temperature. Cells were washed with FACS buffer and fixed with 4% paraformaldehyde (Cytofix; BD Biosciences) for 20 min at 4°C. Surface markers were detected by phycoerythrin (PE)-Gag tetramer (H-2Kd-restricted AMQMLKETI epitope; Beckman Coulter), and antibodies FITC-CD44, peridinin chlorophyll protein (PerCP)-Cy5.5-CD8α, and allophycocyanin (APC)-CD62L (BD Biosciences). Stained samples were analyzed on a BD FACSCalibur, which collected data on 200,000 events per sample. Live cells were gated by forward scatter/side scatter. Within the CD8+ population, the percentage of activated (CD62Llo) and Gag tetramer-specific (Gag-tetramer+) cells were detected.
Splenocytes were plated into round-bottom plates with or without a stimulation cocktail of 10 μg/ml AMQMLKETI Gag peptide (New England Peptide LLC), 1 μg/ml CD49d, and 1 μg/ml CD28 (both, BD Biosciences). After incubation for 2 h at 37°C in 5% CO2, all samples were treated with 1 μl/ml GolgiPlug (BD Biosciences) and allowed to incubate for four additional hours. Cells were blocked, stained for surface CD8α, and fixed as described above. Cells were washed with FACS buffer and stained for internal molecules in the presence of BD Perm/Wash Buffer (BD Biosciences) for 30 min at room temperature. Antibodies used were PE-tumor necrosis factor α (TNF-α), FITC-gamma interferon (IFN-γ), and APC-interleukin-2 (IL-2) (BD Biosciences). Cells were subsequently fixed with 4% paraformaldehyde (Cytofix; BD Biosciences) for 20 min at 4°C. Stained samples were analyzed on a BD FACSCalibur, which collected data on 400,000 events per sample. The percentage of cytokine-producing cells was calculated as the difference between Gag-stimulated and unstimulated samples [(Gag-stimulated percentage) − (unstimulated percentage)].
The 96-well enzyme-linked immunospot (ELISPOT) assay plates (MultiScreen-IP hydrophobic PVDF; Millipore) were coated with 10 μg/ml rat anti-mouse IFN-γ antibody (BD Biosciences) and incubated overnight at 4°C. Plates were washed in 0.25% Tween 20-PBS, blocked with 5% BSA-PBS for 1 h at 37°C, and incubated in splenocyte medium for 2 h. Splenocytes were plated in triplicate at three cell densities appropriate for the experiment (ranging from 104 to 106 cells/well). These cells were incubated with or without a stimulation cocktail of 10 μg/ml AMQMLKETI Gag peptide (New England Peptide LLC), 1 μg/ml CD49d, and 1 μg/ml CD28 (both, BD Biosciences) for 18 h at 37°C in 5% CO2. Plates were washed with 0.25% Tween 20-PBS followed by sterile distilled water to lyse residual cells. Wells were incubated with 5 μg/ml biotinylated rat anti-mouse IFN-γ antibody (BD Biosciences) in 1% BSA-PBS for 2 h at 37°C in 5% CO2. Following a washing step with 0.25% Tween 20-PBS, wells were incubated with 0.2 μg/ml horseradish peroxidase-conjugated streptavidin (Jackson ImmunoResearch) in 1% BSA-PBS for 2 h at 37°C in 5% CO2. IFN-γ-secreting cells were detected by developing the plates with a solution of 3,3′-diaminobenzidine and 4-chloro-1-naphthol in methanol. The plates were scanned using an ImmunoSpot reader (Cellular Technology Ltd.), and the spots were counted using ImmunoSpot software (version 4.0; Cellular Technology Ltd.).
Splenocyte lytic activity was measured by a nonradioactive calcein release assay, described previously (3, 34), with the following modifications. Briefly, P815 target cells (ATCC) cultured in DMEM supplemented with 5% FBS and penicillin-streptomycin were pulsed with either 10 μg/ml AMQMLKETI HIV-1 Gag peptide (H2-Kd restricted; New England Peptide LLC) or 10 μg/ml IGPGRAFYAR HIV-1 Env peptide (H2-Dd restricted; New England Peptide LLC), both immunodominant major histocompatibility complex class I (MHC-I)-restricted epitopes, for 18 h at 37°C in 5% CO2. These P815 targets were washed and resuspended in HBSSF medium (Hanks balanced salt solution without Ca2+, Mg2+, or phenol red [Gibco], supplemented with 5% FBS, 10 mM HEPES, and 4 mM l-glutamine), labeled with 10 μM calcein-AM (Molecular Probes) for 30 min in a 37°C water bath, and washed with HBSSF medium. Target cells were then plated into round-bottom plates in triplicate at 104 cells/well with various concentrations of splenocytes (effectors) to reach final effector/target (E/T) ratios of 100, 33, 11, and 3.7. Plates were incubated for 3 h at 37°C in 5% CO2. Cells were pelleted by spinning at 700 × g for 5 min, and 100 μl of the supernatant was transferred to a 96-well flat-bottom plate. Fluorescence was measured on a Victor2 1420 multilabel counter (Wallac) scanning 1.0 s/well at 485-nm/535-nm wavelengths. Spontaneous release (SR) was measured by incubation of targets with HBSSF medium alone, and maximum release (MR) was measured by incubation of targets with lysis buffer (0.1% Triton X-100 in 50 mM sodium borate, pH 9.0). Percent lysis was calculated as follows: [(sample fluorescence − SR fluorescence)/(MR fluorescence − SR fluorescence)] × 100.
All data were analyzed by Prism software (GraphPad). Unpaired two-tailed t tests were used with the addition of Welch's correction if the variances were unequal. One-way analysis of variance (ANOVA) with Bonferroni correction was used for between-group comparisons, and a two-way ANOVA was used to deduce the total effect of virus verses treatment.
The goal of this study was to establish a single-cycle RV as a vaccine vector using HIV-1 Gag as a model antigen. For this approach, we constructed a novel RV vector whereby the entire RV-G gene was deleted from pSPBN-Gag by utilizing restriction cites flanking the RV-G gene. This vector was designated SPBN-ΔG-Gag (Fig. (Fig.1B),1B), and infectious virus was recovered. We used BNSP-Gag (Fig. (Fig.1D),1D), which encodes all five of the RV genes, in addition to Gag, as a replication-competent control virus for comparative analysis. While SPBN-ΔG-Gag encodes Gag between the RV-M and RV-L genes, BNSP-Gag encodes Gag the between RV-N and RV-P genes.
Because RV-G is required for virus spread (10, 13, 37, 44), we propagated SPBN-ΔG-Gag on a BSR-RVG trans-complementing cell line. Virions collected from these harvests do not encode RV-G in their genome but are complemented with RV-G from the cell line. Virus produced in this manner is hereafter referred to as SPBN-ΔG-GagRVG. To circumvent loss of expression of RV-G (14), the complementing cell line was generated using the Tet-off inducible gene expression system whereby RV-G gene expression is off unless conditionally turned on by the removal of doxycycline (Fig. (Fig.2A2A).
Though we have previously shown that the relative position of Gag within the RV genome does not impact either virus growth kinetics or Gag-specific immunogenicity (26), we sought to quantify the relative production of HIV-1 p24 by SPBN-ΔG-Gag and BNSP-Gag viruses. Cell lysates collected from BSR cells infected at an MOI of 10 for 72 h showed that SPBN-ΔG-Gag produced slightly less p24 than BNSP-Gag (15 and 20 ng/ml, respectively).
To evaluate the replication of SPBN-ΔG-GagRVG, we analyzed virus spread on BSR and trans-complementing BSR-RVG cells. Both cell lines were inoculated at an MOI of 0.001 and 72 h later immunostained for the presence of RV-N (Fig. (Fig.2B).2B). When SPBN-ΔG-GagRVG was grown on trans-complementing BSR-RVG cells, we detected the formation of foci similar in size and morphology to those generated by replication-competent BNSP-Gag. However, when SPBN-ΔG-GagRVG was used to infect BSR cells, only single infected cells were observed. These results demonstrated that SPBN-ΔG-GagRVG can infect cells, but the infection is restricted to the initially infected cell. To further characterize the growth kinetics, we conducted multistep growth curve assays in which BSR or BSR-RVG cells were infected at an MOI of 0.01, and the presence of infectious virus was determined at various time points by focus-forming assay. The results, shown in Fig. Fig.2C,2C, demonstrate that when SPBN-ΔG-GagRVG is grown on BSR cells, no infectious virus is detected in the supernatants. However, when propagated on the trans-complementing BSR-RVG cells, SPBN-ΔG-GagRVG grew to high titers of 107 to 108 focus-forming units (FFU)/ml, similar to replication-competent BNSP-Gag.
Replication-competent vectors are believed to have an immunogenic advantage over alternative vaccine strategies, including inactivated virions, due to their inherent ability to replicate and spread within the immunized host (12). It has previously been shown that replication-deficient RV generated by deletion of the RV-P retained the same balanced IgG2a/IgG1 antibody isotype profile as was observed for live replicating RV while inactivated RV induced a lower humoral response with an IgG1 bias (6). To determine if a G-deleted, single-cycle SPBN-ΔG-GagRVG virus induces antibody responses similar to replication-competent RV, we immunized groups of five BALB/c mice each with 1 × 106 FFU of SPBN-ΔG-GagRVG or BNSP-Gag. Blood was collected at 15 days postimmunization, and an ELISA for detection of total IgG, IgG1, and IgG2a antibody against RV-G was performed (Fig. 3A and B). As expected, we detected significantly less (P < 0.05) total IgG and IgG2a antibody for the SPBN-ΔG-GagRVG-immunized mice than for the BNSP-Gag-immunized mice (Fig. (Fig.3A).3A). There was no detectable difference in IgG1 levels between the immunization groups. Despite this general decrease in anti-RV-G antibodies, the ratios of IgG2a/IgG1 were not statistically different between the two groups (Fig. (Fig.3B).3B). Of note, SPBN-ΔG-GagRVG does not encode genomic RV-G. It was therefore important to measure the antibody responses against a protein encoded by both RVs. Thus, we conducted an anti-RNP ELISA for detection of antibodies against RV-N (Fig. 3C and D). Again, we detected significantly less (P < 0.001) total IgG for SPBN-ΔG-GagRVG-immunized mice than for the BNSP-Gag-immunized mice (Fig. (Fig.3C),3C), but no differences in either the IgG2a or IgG1 levels were observed. Likewise, the IgG2a/IgG1 ratios were not statistically different (Fig. (Fig.3D3D).
To determine the humoral response to different doses of these vectors, blood was collected from mice 1 month after i.m. immunization with 105, 104, 103, 102, or 101 FFU of BNSP-Gag or SPBN-ΔG-GagRVG. Serum was analyzed for anti-RV-G total IgG antibodies by ELISA (Fig. (Fig.3E).3E). Similar to what was initially observed (Fig. (Fig.3A),3A), SPBN-ΔG-GagRVG induces fewer antibodies than BNSP-Gag at titers of 105, 104, and 103 FFU (Fig. (Fig.3E).3E). Results show that the number of anti-RV-G antibodies is dose dependent as they decrease linearly with decreasing immunization titers. No antibodies were detected for immunizations with 102 and 101 FFU.
Lastly, analysis of the onset of anti-RV-G total IgG antibodies over time was conducted on serum collected from mice immunized with 1 × 105 FFU of BNSP-Gag or SPBN-ΔG-GagRVG at 7, 16, 30, and 55 days postimmunization. Though lower, the overall kinetics of antibody development by SPBN-ΔG-GagRVG is comparable to that of BNSP-Gag (Fig. (Fig.3F3F).
Together, these data demonstrate that while SPBN-ΔG-GagRVG induces lower antibody responses than BNSP-Gag, the response has the same balanced IgG isotype profile and kinetics as the replication-competent RV.
CD8+ T-cell responses are thought to be an important characteristic of a potential HIV vaccine (25). Thus, to determine such cellular responses generated by a single-cycle RV, we compared the HIV-1 Gag-specific CD8+ T-cell responses induced by priming with SPBN-ΔG-GagRVG to those induced by BNSP-Gag. For this approach, BALB/c mice immunized i.m. with 1 × 106 FFU of BNSP-Gag or SPBN-ΔG-GagRVG were sacrificed at 7, 10, 15, and 20 days postimmunization (Fig. (Fig.4A).4A). The splenocytes were then analyzed by flow cytometry (Fig. (Fig.4B)4B) and an IFN-γ-ELISPOT assay (Fig. (Fig.4C)4C) at each time point. In mice immunized with BNSP-Gag, we measured a significant increase, using one-way ANOVA with a Bonferroni correction, in both activated Gag tetramer-positive T cells and IFN-γ-producing cells at day 10 postprime relative to naïve mice. Additionally, we compared the T-cell responses of the immunized mice averaged across all time points to the responses of the naïve mice using a two-way analysis of variance comparing virus type (BNSP-Gag or SPBN-ΔG-GagRVG) to immunization (immunized or unimmunized). We found that the type of virus (replication-competent or single-cycle RV) had no effect on the response (for Gag tetramer-positive cells, P = 0.9620; for IFN-γ production, P = 0.8733), while the immunization did have a significant effect (Gag tetramer-positive cells, P = 0.0044; IFN-γ production, P = 0.0048). Together, these results indicate that priming with BNSP-Gag or SPBN-ΔG-GagRVG induces a significant number of Gag-specific CD8+ T cells and that the magnitude of this primary response is equivalent between replication-competent and single-cycle RV vectors.
To analyze whether the observed T-cell responses were dependent on the initial titers used for immunization, we performed a dose-response analysis. Groups of BALB/c mice were immunized i.m. with 106, 105, or 104 FFU of BNSP-Gag or SPBN-ΔG-GagRVG. At 10 days postprime, spleens were harvested, and splenocytes were analyzed by IFN-γ-ELISPOT assay (Fig. (Fig.4D).4D). Interestingly, for both viruses, the number of IFN-γ-producing T cells remained consistently high in mice across all three immunization titers. Though there is a decrease in IFN-γ-producing cells in mice immunized with 106 and 105 FFU of SPBN-ΔG-GagRVG, there is no significant difference either between the 106 and 104 FFU groups or the 105 and 104 FFU groups, suggesting that there is no dose response. In contrast to the antibody responses, low immunization titers (104 FFU) with either virus were sufficient to induce T-cell responses equal to titers 100-fold higher (Fig. (Fig.4D).4D). Therefore, RV-induced antibody responses are dose dependent while CD8+ T-cell responses are not.
As priming by both single-cycle and replication-competent RVs induce the formation of HIV-1 Gag-specific CD8+ T cells, we next sought to determine the capacity of these T cells to expand and assume a cytolytic phenotype during a recall response to the HIV-1 Gag antigen. Groups of five BALB/c mice were immunized i.m. with 2 × 106 FFU of either virus, rested 10 weeks, and then challenged i.p. with 1 × 106 PFU of vaccinia virus expressing HIV-1 Gag (VV-Gag) (Fig. (Fig.5A).5A). Spleens were harvested 4.5 days postchallenge, and splenocytes were analyzed by flow cytometry and IFN-γ-ELISPOT assay. Immunization by BNSP-Gag and SPBN-ΔG-GagRVG induced similar numbers of activated Gag tetramer-positive T cells, as depicted in Fig. Fig.5B5B for single representative mice and in Fig. Fig.5C5C for all mice analyzed (P = 0.8353). As with the tetramer staining, we obtained similar numbers of IFN-γ-producing cells (Fig. (Fig.5D)5D) (P = 0.9433). Unstimulated splenocytes had no positive responders, demonstrating the specificity of the assay (data not shown). This indicates that single-cycle RV induces antigen-specific T cells that are functional in cytokine secretion.
However, CD8+ T-cell antiviral activity is mediated by both cytokine secretion and lytic activity. These two effector functions can be differentially regulated (1, 5, 19) yielding T cells that are antigen specific yet not functionally cytolytic. Thus, to determine the cytolytic capacity of the T cells generated by the single-cycle RV, BALB/c mice were immunized i.m. with 1 × 106 FFU of SPBN-ΔG-GagRVG or BNSP-Gag, rested 4 weeks, and challenged i.p. with 1 × 106 PFU of VV-Gag. Spleens were harvested 4.5 days postchallenge, and splenocytes were assayed for cytolytic activity by a nonradioactive CTL assay (calcein release assay). Splenocytes from both immunization groups exhibited lytic activity when cultured with Gag-stimulated target cells (Fig. (Fig.5E)5E) and remained unresponsive when cultured with irrelevant Env-stimulated target cells (Fig. (Fig.5E).5E). Naïve mice challenged with VV-Gag exhibited no detectable Gag-specific cytolytic activity. The percent specific lysis detected from BNSP-Gag immunization was equal to that of SPBN-ΔG-GagRVG (P = 0.8121) at an E/T ratio of 100. Data are representative of three independent experiments. Of note, we found that the percent specific lysis positively correlated to the percent activated Gag tetramer-positive T cells (r2 = 0.9298) (Fig. (Fig.5F5F).
To evaluate the proliferative capacity of T cells generated from low-dose RV immunizations, we immunized groups of BALB/c mice i.m. with 105, 104, 103, 102, or 101 FFU of BNSP-Gag or SPBN-ΔG-GagRVG. Five weeks postimmunization, mice were challenged i.p. with 1 × 106 PFU of VV-Gag, and spleens were harvested 4.5 days postchallenge. Splenocytes were analyzed by flow cytometry (Fig. (Fig.6A)6A) and IFN-γ ELISPOT assay (Fig. (Fig.6B).6B). Interestingly, for both viruses, the number of activated Gag tetramer-positive T cells remained consistently high in mice immunized with 105, 104, and 103 FFU and then dropped to background levels, as seen in naïve mice, with the titers 102 and 101 (Fig. (Fig.6A).6A). Similar results were found for the number of IFN-γ-producing cells (Fig. (Fig.6B).6B). Statistical analysis showed a significant difference in IFN-γ-producing cells in mice immunized with 105 verses 103 FFU of SPBN-ΔG-GagRVG (Fig. (Fig.6B).6B). However, direct comparison of BNSP-Gag-immunized mice with SPBN-ΔG-GagRVG-immunized mice at each specific titer (i.e., 105 versus 105) showed no significant differences, indicating that the single-cycle vector is just as potent as the replication-competent vector, even at very low titers. The complete lack of a cellular response in animals immunized with 102 or 101 FFU may be due to the fact that titers determined in vitro do not translate into infectivity in vivo. This correlates to the absence of detectable antibodies at these low titers (Fig. (Fig.3E).3E). This dose response analysis demonstrates that, upon VV-Gag challenge, low-titer immunizations with single-cycle RV induce CD8+ T-cell responses capable of expanding to levels as robust as high-dose replication-competent immunizations.
In summary, our results demonstrate that single-cycle RVs induce CD8+ T cells with effector functions, cytokine secretion and lytic activity, equally as potent and proliferative as those observed for the replication-competent vector, BNSP-Gag. This further illustrates the potential of single-cycle RV as a vaccine vector.
Vaccination strategies commonly include multiple inoculations, or boosters, to increase the antigen-specific immune responses and better protect the immunized host. In this next step, we sought to examine the capacity of our single-cycle RV as a booster vaccine. As RV-G is highly immunogenic, mice previously immunized with an RV-G expressing RV are likely to avoid reinfection with the same vector, as previously shown for replication-competent RV in mice and monkeys (29, 42). Thus, we sought to generate single-cycle RV for boosting that is complemented with a heterologous glycoprotein. For this we used a VSV-RVG chimera in which the ectodomain and transmembrane domains of the VSV-G (Indiana strain) are fused to the RV-G cytoplasmic domain (CD) (17). Previously, it was shown that replacing the CD of a foreign glycoprotein with that of RV-G promotes its incorporation into the RV envelope (30). To produce a vector containing this heterologous glycoprotein, we propagated SPBN-ΔG-Gag on a BSR-VSV-RVG trans-complementing cell line. Like the BSR-RVG cells (Fig. (Fig.2),2), this cell line was generated using the Tet-off inducible gene expression system whereby VSV-RVG gene expression is off unless conditionally turned on by the removal of doxycycline (Fig. (Fig.7A).7A). Virus produced in this manner was designated SPBN-ΔG-GagVSV-RVG. To evaluate replication of SPBN-ΔG-GagVSV-RVG on BSR-VSV-RVG cells, we analyzed virus spread (Fig. (Fig.7B)7B) on induced or uninduced cells by inoculating them at an MOI of 0.001 and 72 h later immunostaining the cells for the presence of RV-N. Foci were detected only when SPBN-ΔG-Gag was grown on induced BSR-VSV-RVG cells (Fig. (Fig.7B,7B, top). This demonstrates that SPBN-ΔG-GagVSV-RVG is a single-cycle virus like SPBN-ΔG-GagRVG. These findings were confirmed by multistep growth curve assays (Fig. (Fig.7C),7C), which demonstrate that when SPBN-ΔG-Gag is grown on BSR cells, no infectious virus is produced (Fig. (Fig.7C);7C); however, when propagated on trans-complementing BSR-VSV-RVG cells, SPBN-ΔG-Gag grew to titers of 106 to 107 FFU /ml, approximately 10-fold lower than growth of BNSP-Gag (Fig. (Fig.7C).7C). This 10-fold decrease in viral titer may be due to less efficient budding due to the chimeric VSV-RV-G, as was previously observed (17).
To evaluate the capacity of heterologous single-cycle RV to increase HIV-1 Gag-specific immune responses, groups of five BALB/c mice were immunized i.m. with 1 × 106 FFU of SPBN-ΔG-GagRVG and boosted i.m. at 6 weeks postimmunization with 1 × 106 FFU of SPBN-ΔG-GagVSV-RVG (heterologous boost), 1 × 106 FFU of SPBN-ΔG-GagRVG (homologous boost), or PBS (mock boost) (Fig. (Fig.8A).8A). At 0, 5, 10, and 15 days postboost, mice were sacrificed, and splenocytes were assayed for IFN-γ production by ELISPOT assay (Fig. (Fig.8B),8B), for activated Gag tetramer-positive T cells by flow cytometry (Fig. (Fig.8C),8C), and for generally activated CD8+ T cells (Fig. (Fig.8D).8D). Additionally, cells were assayed for cytokine production by ICS (Fig. 8E to H). In all assays, we found that both the homologous boost and mock boost did not increase the numbers of Gag-specific CD8+ T cells. This suggests that primary immunization by the single-cycle virus induces neutralizing antibodies that prevent a successful homologous boost, as shown previously (29, 42). However, the heterologous boost induced a significantly higher number of IFN-γ-producing cells (Fig. (Fig.8B),8B), activated Gag tetramer-positive T cells (Fig. (Fig.8C),8C), and generally activated CD8+ T cells (Fig. (Fig.8D)8D) than the homologously boosted mice. Importantly, the Gag-specific cells are also functional, as evidenced by their capacity to produce IFN-γ, IL-2, TNF-α, and the dual expression of both TNF-α and IFN-γ (Fig. 8B and E to H). Together, this demonstrates that the immune response generated by single-cycle RV can be significantly enhanced by boosting with a single-cycle RV complemented with a heterologous glycoprotein.
The need for an effective HIV-1 vaccine is greater than ever. Live viral vectors are at the forefront of vaccine development due to their capacity to generate strong humoral and cellular immune responses. However, the safety concerns associated with live, replicating vectors have proven to be a major obstacle in their approval for use in humans. As such, recent vaccine development has focused on making substantially attenuated vectors by truncation or deletion of viral genes directly involved with virus replication and spread (8, 18, 26, 33, 35, 36, 38). Here, we describe the development of a single-cycle RV, attenuated by deletion of RV-G and expressing HIV-1 Gag as a model antigen. Our immunogenicity studies demonstrate that single-cycle RV induced equally potent CD8+ T-cell responses but lower vector-specific antibody responses than replication-competent RV following i.m. injection. Even though the induced humoral responses were lower, they were still potent enough to block a second infection with the same vector, indicating the potential of such vectors to induce neutralizing responses against other pathogens.
The decision to delete the RV-G gene was made based on what is known about its critical role in the attachment and entry of RV into host cells. RV-G is one of the most important determinants of viral pathogenicity (10, 11, 37), and without it, RV particles are unable to spread in vitro or in vivo (13, 44). However, because RV-G has little known involvement in genome transcription and replication, a G-deleted vector would theoretically infect a cell and express its gene products just as efficiently as a fully replication-competent RV. This intracellular gene expression would provide ample antigen for presentation to both MHC-I and MHC-II pathways thought to contribute to live viral vector immunogenicity. Therefore, we hypothesized that a G-deleted, single-cycle RV would provide a safe, yet potent, alternative to our prototypic RV vector. The safety benefits are enhanced by the fact that, unlike viruses that are attenuated by point mutations, a single-cycle virus missing a large section of the genome is unlikely to revert back to its virulent form. This is especially true for negative-stranded viruses, including RV, where recombination is extremely rare and has been described for only respiratory syncytial virus (RSV) (41).
In order to establish the relative immunogenicity of a single-cycle RV vector, we had to choose an appropriate replication-competent control virus for comparative analysis. VSV, a closely related rhabdovirus, has documented transcriptional attenuation of gene products in the 3′ to 5′ direction (20), and a similar mechanism of transcriptional attenuation has been suggested for RV (16). In order to utilize the most stringent control with equal or greater expression of the model antigen, HIV-1 Gag, we chose BNSP-Gag as our replication control virus. Indeed, when infected cell lysate was assayed for p24, we confirmed that SPBN-ΔG-Gag produces just slightly less p24 than BNSP-Gag (15 and 20 ng/ml, respectively). This decrease may be attributed to changes in virus transcription that occur from removal of an entire gene segment. However, we have previously shown that the relative position of Gag within the RV genome does not impact virus growth kinetics (26), and this was confirmed in growth curves for SPBN-ΔG-Gag and BNSP-Gag (Fig. (Fig.2C2C).
Vector-specific antibody responses were lower for the single-cycle vector than for the replication-competent vector. However, unlike inactivated RV (20), the isotype profiles generated by single-cycle vectors were identical to those of replication-competent vectors, indicating the same balanced Th1/Th2 immune responses. Th1 responses are particularly important for the efficacy of a vaccine targeted for intracellular pathogens due to their known cytokine-mediated antiviral activities and macrophage activation. We showed that anti-RV-G antibodies are dose dependent (Fig. (Fig.3E),3E), suggesting that higher immunization titers could be used if high antibody levels are sought, as is done for other replication-deficient vectors (18, 33, 35). It is still interesting that a G-deleted RV induces such substantial anti-RV-G antibodies, and based on the ineffective homologous boost, these antibodies appear to be neutralizing. Considering that the same was observed for anti-RNP antibodies, this would suggest that comparable humoral responses may be generated against other vector-encoded proteins, including foreign antigens.
Despite the lower humoral response to single-cycle RV, the CD8+ T-cell responses were equal in quality and quantity to those of replication-competent RV. While the primary T-cell response is relatively low, it is clear that these Gag-specific T-cell precursors effectively expand in a robust secondary recall response in the event of a VV-Gag challenge, yielding highly functional, cytokine-secreting CTLs. Of note, these recall responses were reached at titers as low as 103 FFU of single-cycle or replication-competent RV. This is in contrast to immunization titers in the order of 109 PFU used for recombinant VV (rVV) and its replication-deficient counterpart, rMVA, as well as the replication-competent and -deficient Ad5 vectors (18, 33, 35). The CD8+ T cells induced by single-cycle RV exhibited potent CTL function as well as the capacity to expand to heterologous boost or challenge. This is significant because other RNA viruses, including lymphocytic choriomeningitis virus (LCMV) and hepatitis C virus (HCV), are known to induce functionally unresponsive T cells, usually a consequence of low-level or persistent infections (1, 5, 19). Furthermore, because rhabdoviruses can incorporate foreign glycoproteins into viral particles (17, 30, 31), we were able to develop a single-cycle booster virus complemented with VSV-RVG that significantly enhances the cellular immune response. We showed that boosting with a single-cycle RV enhances frequency and functionality of antigen-specific T cells. Enhanced functionality was observed by an increase in polyfunctional IL-2-, IFN-γ-, and TNF-α-expressing T cells, which has been loosely correlated to lower viral loads in human HIV nonprogressors (2). A VSV-based single-cycle vector encoding HIV-1 Env has also been successful in a heterologous prime/boost schedule; however, if ever used in primates, the Env protein could permit virus spread, resulting in a replication-competent instead of a single-cycle vector (36).
It is generally believed that the strength of the immune response positively correlates with a vector's ability to replicate and spread (40). However, when injected i.m., replication-deficient/single-cycle RV- and VSV-based vectors (8, 21, 36) fare just as well as their replication-competent counterparts. We hypothesize that replication-competent RV does not extensively spread in the muscle, rendering it phenotypically similar to the single-cycle RV in vivo. In fact, quantification of viral mRNA from infected muscle samples by real-time PCR showed no significant difference in message between BNSP-Gag or SPBN-ΔG-GagRVG at either 48 or 120 h postinfection (data not shown). These data support our hypothesis that the RV vaccine strain does not spread extensively in vivo. Therefore, it would appear that the local intramuscular replication of both replication-competent and single-cycle RV provides the “danger signal” sufficient for effective recruitment of immune cells (e.g., dendritic cells and macrophages) to induce potent immune responses. The failure of single-cycle RV-based vectors to induce a potent immune response via mucosal routes (data not shown) might be explained by the inability of the vector to elicit such danger signals from the initially infected target cell on a mucosal surface. In contrast, a replication-competent virus could spread beyond the initially infected target cell to reach an immune-inductive site where a potent immune response is elicited.
In conclusion, we have generated a single-cycle RV for prime/boost vaccine regimens that elicits HIV-1 Gag-specific CD8+ T-cell responses equally robust as those of replication-competent RV. This vector is a safe, effective alternative to replication-competent RV vectors for HIV-1 and other infectious diseases and warrants further development.
The anti-p24 antibody (71-31) and recombinant vaccinia virus expressing HIV-1 Gag (vP1287) were obtained through the NIH AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH.
This study was supported by NIH grants R01AI049153 and P01AI082325 to M.J.S.
Published ahead of print on 6 January 2010.