Optimizing HIV-1-specific CD8+ T-cell induction by recombinant BCG in prime-boost regimens with heterologous viral vectors

Authors


Abstract

The desire to induce HIV-1-specific responses soon after birth to prevent breast milk transmission of HIV-1 led us to propose a vaccine regimen which primes HIV-1-specific T cells using a recombinant Mycobacterium bovis bacillus Calmette-Guérin (rBCG) vaccine. Because attenuated live bacterial vaccines are typically not sufficiently immunogenic as stand-alone vaccines, rBCG-primed T cells will likely require boost immunization(s). Here, we compared modified Danish (AERAS-401) and Pasteur lysine auxotroph (222) strains of BCG expressing the immunogen HIVA for their potency to prime HIV-1-specific responses in adult BALB/c mice and examined four heterologous boosting HIVA vaccines for their immunogenic synergy. We found that both BCG.HIVA401 and BCG.HIVA222 primed HIV-1-specific CD8+ T-cell-mediated responses. The strongest boosts were delivered by human adenovirus-vectored HAdV5.HIVA and sheep atadenovirus-vectored OAdV7.HIVA vaccines, followed by poxvirus MVA.HIVA; the weakest was plasmid pTH.HIVA DNA. The prime-boost regimens induced T cells capable of efficient in vivo killing of sensitized target cells. We also observed that the BCG.HIVA401 and BCG.HIVA222 vaccines have broadly similar immunologic properties, but display a number of differences mainly detected through distinct profiles of soluble intercellular signaling molecules produced by immune splenocytes in response to both HIV-1- and BCG-specific stimuli. These results encourage further development of the rBCG prime-boost regimen.

Introduction

It is 30 years since the first cases of AIDS were described, yet HIV-1 infection continues to take its toll particularly in resource-poor areas such as many parts of sub-Saharan Africa 1. About half of the infected adults are women of childbearing age, who will expose their babies to HIV-1 during pregnancy, delivery and breastfeeding 2. Although antiretroviral treatment substantially decreases mother-to-child transmission (MTCT) of HIV-1 3, it is not ideal due to the cost, requirement for daily compliance, side effects and possible development of drug resistance. Because breast milk provides essential nutrients and protection against other diseases in the early days of life 4, 5, formula as an HIV-1-free alternative is recommended only if it is acceptable, feasible, affordable, sustainable and safe (AFASS) and thus it is not an option for many HIV-1-positive mothers in Africa. The best hope for protecting newborns and infants in the ‘South’ (and in the ‘North’) against acquiring HIV-1 from their infected mothers while breastfeeding remains the development of safe, effective and accessible adult and pediatric vaccines 6. For babies born to HIV-1-positive mothers, induction of anti-HIV-1 immunity as soon as possible after birth is highly desirable and may provide a basis for lifetime protection, which can be maintained by boosts later in life.

A successful HIV-1 vaccine may have to induce both T-cell and antibody responses 7, although an effective T- or B-cell vaccine alone is likely to impact the epidemic. HIVA, a first-generation immunogen, which was designed to optimize induction of T cells 8, has been used extensively in mice, rhesus macaques, and human adult 9 and neonate (manuscripts in preparation) volunteers to advance T-cell vaccine development. It is derived from African HIV-1 clade A consensus Gag p24 and p17 coupled to regions of overlapping CD8+ T-cell epitopes. To facilitate both comparison and combination of different vaccination modalities, HIVA has been expressed and presented to the immune system from plasmid DNA 8, several non-replicating mammalian viruses 8, 10 attenuated strains of mycobacteria 11–13 and as a protein on the surface of baculovirus/insect cell-expressed bluetongue virus microtubules 14, 15. In some vaccination regimens, HIVA afforded protection against model surrogate virus challenges including chimeric EcoHIV/NDK 16. In humans, broad CD8+ T-cell responses recognizing multiple epitopes in HIV-1 Gag were associated with a good control of chronic HIV-1 replication 17–19. Thus, the rationale for using a Gag-based immunogen for induction of protective T-cell immunity remains valid, although improved second-generation immunogens that better address the viral diversity and escape are now under evaluation 20, 21.

The safety of HIV-1 vaccines prioritizes the use of non-replicating vaccine vectors 14, 22. Because such modalities are less immunogenic as stand-alone vaccines, they are combined into heterologous prime-boost regimens delivering a common, pathogen-derived immunogen such as HIVA to maximize the induction of transgene-specific responses while avoiding build-up of anti-vector immunity, which dampens vaccine uptake 10, 23, 24. Each vector and vector–immunogen combination is likely to have unique biological and immunological properties. While some vectors are better for priming, others excel in boosting 23–25 and although some general rules are emerging, combining vaccines into more complex heterologous regimens remains largely empirical.

Attenuated Mycobacterium bovis bacillus Calmette-Guérin (BCG) is administered to a majority of African infants at birth as a vaccine against tuberculosis. Because of its many attractive features, BCG is explored as a vector for subunit vaccines against other illnesses particularly for developing countries 26. In experimental models, recombinant BCG (rBCG) elicited protective immunity against a number of important human pathogens 27–33. However, in the only efficacy trial in human volunteers to date, rBCG failed to demonstrate consistent protection against Lyme disease 34. Nevertheless, effort continues to be invested into development of BCG-vectored vaccines 26.

Originally, virulent M. bovis was attenuated by growth on potato slices imbibed with glycerol, which after 198 passages resulted in non-virulent bacilli 35. Further propagation of BCG in various laboratories around the world prior to the introduction of archival seed lots in 1960s 36 led to diversification of BCG into genotypically distinct daughter strains. It has been postulated that the resultant phenotypic differences contribute to variable anti-tuberculosis vaccine efficacy 37, although in humans real evidence for this is lacking 38, 39. Previously, we reported the construction of two candidate HIV-1 vaccines BCG.HIVA40113 and BCG.HIVA22211, which employed two distinct BCG strains that were rationally engineered to increase safety and T-cell immunogenicity. In vaccine BCG.HIVA401, the HIVA gene is vectored by AERAS-401, Danish strain BCG1331ΔureC::ΩpfoA(G137Q)40. AERAS-401 expresses mutated detoxified perfringolysin O, which facilitates BCG escape from the endosome and thus increases availability of the transgene product for antigen processing and presentation. The HIVA gene is integrated into the mycobacterial chromosome and its transcription is driven by the hsp60 promoter. The second vaccine, BCG.HIVA222 is derived from a lysine auxotroph strain of BCG Pasteur mc21604ΔlysA5::res, which lacks an essential enzyme catalyzing the last step in lysine biosynthesis 41. This enzyme is supplied from transformed episomal plasmid pJH222.HIVA, which also carries the HIVA transgene under control of the α-antigen promoter. In both vaccines, the HIVA open-reading frame contains GC-rich codons similar to those used by mycobacterium 42 and is coupled at its 5′-end to a gene coding for a 19-kDa lipoprotein or Ag85B leader sequence, which facilitates transport of the HIVA protein across the mycobacterial membrane. Both BCG.HIVA vaccines elicited HIV-1-specific CD8+ T-cell responses in BALB/c mice or rhesus macaques 11–13, 43, 44. Here, we compare BCG.HIVA401 and BCG.HIVA222 head-to-head for induction of HIV-1-specific CD8+ T-cell responses when combined into heterologous prime-boost regimens with vaccines vectored by human adenovirus serotype 5 (HAdV-5), ovine atadenovirus serotype 7 (OAdV-7), modified vaccinia virus Ankara (MVA) and plasmid DNA.

Results

BCG.HIVA priming augments CD8+ T cells induced by heterologous vaccines

As part of our long-term aim to prevent MTCT of HIV-1 through breast milk by early pediatric vaccination, we set out to compare directly two BCG.HIVAs each engineered in its own way to increase vaccine safety and induction of T cells. In addition, four different, previously described vaccines, human adenovirus HAdV5.HIVA, ovine atadenovirus OAdV7.HIVA, poxvirus MVA.HIVA, and plasmid pTH.HIVA DNA 8, 10, were compared for their ability to boost rBCG-primed CD8+ T-cell responses. Thus, BALB/c mice were primed with BCG.HIVA401, BCG.HIVA222 or left unimmunized on week 0 and boosted with one of four heterologous vaccines or PBS on wk 12. Animals were sacrificed 4 wks later and analyzed individually for HIV-1-specific CD8+T-cell responses (Fig. 1A). The HIVA protein contains two immunodominant epitopes H (RGPGRAFVTI from Env restricted by H-2D/Ld45) and P (IFQSSMTKI from viral polymerase restricted by H-2Kd16), and the corresponding peptides were employed in all T-cell assays. First, an ELISPOT assay was utilized to estimated the frequencies of T cells producing IFN-γ upon HIV-1-specific restimulation (Fig. 1B). This showed that BCG.HIVA222 primed for higher frequencies of H-specific T cells than BCG.HIVA401 when HAdV5.HIVA (p=0.02) and MVA.HIVA (p=0.07) vaccines were used as boosts and, consistently with our previous publication 11, BCG.HIVA222 enhanced significantly MVA.HIVA alone vaccination (p=0.04). Mean H-specific frequencies ranged from approximately 100–350 spot-forming units (SFU) per 106 cells, while responses to peptide P were much lower as P epitope is sub-dominant to H 16. Vaccination with either rBCG alone failed to induce detectable H-specific responses in this assay (Fig. 1). Accumulation of IFN-γ in the tissue culture supernatant upon H peptide restimulation was also examined in a Bio-Plex assay, which confirmed the priming capacity of both BCG.HIVA vaccines, i.e. the IFN-γ concentrations in the culture supernatant were increased by prior rBCG administration relative to all four OAdV7.HIVA, HAdV5.HIVA, MVA.HIVA and pTH.HIVA DNA vaccines alone with the highest mean reaching over 4200 pg/mL (Fig. 1C). Finally, splenocytes from the same experiment were subjected to a polychromatic flow cytometry assay following peptide H restimulation for parallel detection of IFN-γ, TNF-α, CD107a (degranulation) and IL-2 (Fig. 1D). While boosting with plasmid pTH.HIVA DNA elicited very low frequencies of plurifunctional H-specific T cells, recombinant viral vaccines induced robust responses with frequencies of bifunctional IFN-γ+CD107a+ cells reaching 4%, 3% and 2% of total CD3+CD8+ cells for the OAdV7.HIVA, HAdV5.HIVA and MVA.HIVA vaccines, respectively (Fig. 1D light gray); proportions of cells displaying all four functions were approximately half of the bifunctional frequencies (Fig. 1D black). Overall, boosting capacities of the attenuated virus-based vaccines tested were largely comparable and superior than that of the plasmid DNA.

Figure 1.

Augmentation of HIV-1-specific CD8+ T-cell responses by BCG.HIVA401 or BCG.HIVA222 priming. (A) Groups of 5 BALB/c mice were left unimmunized or inoculated i.p. with 106 CFU of BCG.HIVA401 or BCG.HIVA222, or PBS, and boosted 12 wk later with one of the following vaccines delivered i.m.: 106 IU of HAdV5.HIVA, 107 IU of OAdV7.HIVA, 108 PFU of MVA.HIVA, 100 μg of plasmid pTH.HIVA DNA or PBS as a negative control; mice were sacrificed at wk 16. (B) Frequencies of IFN-γ-producing splenocytes upon H (black) and P (grey) peptide restimulation were determined in an ELISPOT assay. B401 – BCG.HIVA401; B222 – BCG.HIVA222; A – HAdV5.HIVA; O – OAdV7.HIVA; M – MVA.HIVA; and D – pTH.HIVA DNA. (C) Using H peptide, IFN-γ secretion into the culture supernatant over 24 h was determined in a Bioplex assay. (D) Multicolour flow cytometry was used to estimate frequencies of oligofunctional H-specific T cells following BCG.HIVA401 or BCG.HIVA222 priming. The gating strategy is shown in Supporting Information Fig. 1. Cells positive for IFN-γ, IL-2, TNF-α and CD107a (black), IFN-γ, IL-2, and CD107a (dark grey), and IFN-γ and CD107a (light grey) are shown. (B–D) All assays were performed for individual animals, unstimulated release was subtracted and the results are shown as mean+SD (n=5). (D) Statistical significance was determined by Student's t-test. These results are representative of at least two independent experiments.

Efficient induction of in vivo killing by BCG.HIVA401-prime regimens

Killing of HIV-1 infected cells is likely to be one of the critical properties of protective T-cell responses. Here, we determined the efficacy of vaccine-elicited T cells to kill sensitized target cells in vivo. Thus, BALB/c mice were primed with BCG.HIVA401 at wk 0 and boosted with one of the four heterologous vaccines 14 wk later (Fig. 2A). First at wk 16, mice were bled and for each group, their isolated PBMCs were combined and assayed in multicolour flow cytometry for parallel production of IFN-γ and surface expression of CD107a. This test confirmed the vaccine take prior to the in vivo killing assay (Fig. 2B). At wk 18, naïve splenocytes were pulsed with peptides H and P or left unpulsed, each cell population was labeled differentially with CFSE, combined and transferred back into vaccinated or naive mice, and the cell survival was evaluated 5 h later. The boosting vaccines of BCG.HIVA401-primed responses ranked from the most to the least efficient killing of the H- and P-pulsed targets as HAdV5.HIVA, OAdV7.HIVA, MVA.HIVA and pTH.HIVA DNA (Fig. 2C). Thus, the ability of vaccine-induced cells to lyse in vivo target cells was demonstrated and shown to correlate well with the IFN-γ/CD107a expression.

Figure 2.

Efficient in vivo killing induced by heterologous regimens. (A) Groups of 5 BALB/c mice were administered i.p. with 106 CFU of BCG.HIVA401 and boosted i.m. with 107 IU of HAdV5.HIVA, 107 IU of OAdV7.HIVA (10-fold higher than in Fig. 1), 108 PFU of MVA.HIVA or 100 μg of plasmid pTH.HIVA DNA using the indicated schedule. (B) At wk 16, mice were bled and their PBMCs were combined for an intracellular cytokine staining analysis using H peptide restimulation. Gate was first applied on live lymphocytes and then on CD3+CD8+ cells before enumeration of the IFN-γ+CD107a+ cells. (C) At wk 18, vaccinated animals were injected i.v. with differentially labeled H (black) or P (grey) peptide-pulsed on unpulsed syngeneic splenocytes, which were re-isolated after 5 h and enumerated. Gate was first applied on live CMTMR-labeled lymphocytes followed by gating directly on differentially labeled populations. The in vivo killing experiment was performed twice and representative data are shown as mean+SD of 5 independently evaluated mice in each group.

BCG.HIVA401 and BCG.HIVA222 imprint differently production of intercellular signaling molecules

Next, we decided to compile more complete functional profiles of the T cells induced by each of the tested vaccines alone and compare them to those induced by heterologous regimens. The levels of various signaling molecules released into the supernatants were quantified using a Bio-Plex assay following a 24-h in vitro restimulation with peptide H. First, the data for single vaccine immunizations indicated that each vaccine modality induced T cells producing a unique pattern of soluble factors (Table 1). While all HIVA vaccines programmed production of some pro-inflammatory cytokine(s), their levels and types differed. In addition to pro-inflammatory cytokines, HAdV5.HIVA uniquely primed the induction of HIV-1 entry-blocking β-chemokines CCL3 (MIP-1α), CCL4 (MIP-1β) and CCL5 (RANTES). MVA.HIVA potentiated secretion of IL-1β, and both BCG strains and interestingly OAdV7.HIVA generated memory cells producing IL-17. Notably, BCG.HIVA222 vaccination led to memory cells with much broader and stronger H peptide-stimulated cytokine release compared with BCG.HIVA401.

Table 1. Profiles of soluble signaling molecules produced by vaccine-elicited, HIV-1-specific CD8+ T cellsa)
VaccineBCG.HIVA401BCG.HIVA222HAdV5.HIVAOAdV7.HIVAMVA.HIVApTH.HIVA
  1. a

    a) Groups of 5 BALB/c mice were immunized with individual vaccines, their splenocytes were isolated either 16 wk later for BCG.HIVAs or 4 wk later for the other modalities and stimulated with peptide H for 24 h, and the production of soluble factors was determined in the culture supernatants using a Bioplex assay. Values are shown as median concentrations (pg/mL).

Median (pg/mL)
IL-1α42144805094270
IL-1β00003760
IL-20217105614835460
IL-30165001780
IL-40840000
IL-50205048100
IL-6001360250
IL-9000000
IL-101101010141095
IL-12p40000000
IL-12p70119151198000
IL-13000000
IL-1784146036500
Eotaxin000000
G-CSF000000
GM-CSF000000
IFN-γ0156109124705160
CXCL1 (KC)000000
CCL2 (MCP1)21614511759001542
CCL3 (MIP1α)006620000
CCL4 (MIP1β)010182804800
CCL5 (RANTES)0031449400
TNF-α78712930012160

Second, we examined the impact of BCG.HIVA401 or BCG.HIVA222 priming on the cytokine signatures produced by the heterologous vaccine boosts. Thus, BALB/c mice were vaccinated as in Fig. 1A. Examination of the HAdV5.HIVA regimen responses showed increases in IL-12p70, IFN-γ and CCL5 production when primed with either of the rBCG vaccines. BCG.HIVA222-primed mice showed an increase in CCL3 and CCL4, while priming with BCG.HIVA401 led to an increase in IL-6 and a significant decrease was seen for CCL2 (MCP1) (Fig. 3A). For OAdV7.HIVA regimens, a reduction was seen in IL-17 levels for both rBCG-primed groups. BCG.HIVA401 priming resulted in a reduction for IL-1α, IL-2 and IL-5, while BCG.HIVA222 prime significantly increased levels of IL-10, IFN-γ, and CCL5 (Fig. 3B). Boosting with MVA.HIVA led to significant differences with both rBCG primes for the cytokines IL-6, CCL4 and CCL5, but only BCG.HIVA222 priming led to broad responses with increased levels of IL-1β, IL-2, IL-3, IL-4, IFN-γ and CCL3 secretion (Fig. 3C). Minimal levels of cytokines were measured for animals immunized with pTH.HIVA DNA with significant increases in IL-4, IL-10 and CCL4, and a decrease in CCL2 levels (Fig. 3D), perhaps, because of the generally low frequencies of induced HIV-1-specific T cells.

Figure 3.

Impact of BCG.HIVA401 or BCG.HIVA222 priming on production of soluble factors following heterologous boost. Groups of 5 BALB/c mice were immunized as in Fig. 1A, their splenocytes were isolated and stimulated with peptide H for 24 h and the intercellular signaling molecule concentrations in the supernatants were determined using a Bioplex assay. Only factors affected by rBCG priming are shown grouped by the use of (A) HAdV5.HIVA, (B) OAdV7.HIVA, (C) MVA.HIVA and (D) pTH.HIVA DNA. All assays were carried out separately for individual animals, unstimulated release was subtracted and the results are shown as mean+SD (n=5) of one experiment. Statistical significance is indicated as determined by Student's t-test. B401 – BCG.HIVA401; B222 – BCG.HIVA222; A – HAdV5.HIVA; O – OAdV.HIVA; M – MVA.HIVA; and D – pTH.HIVA DNA.

Finally, being intrigued by the differences in the molecular signaling patterns observed between the two rBCG vaccines, we examined these profiles following restimulation of BCG.HIVA401- or BCG.HIVA222-immune splenocytes by the homologous rBCG vaccine at wks 0 (naïve), 4 and 12 (Table 2). Once again, significant differences were observed. Thus, overall, stimulation of naïve splenocytes with BCG.HIVA222 induced the highest cytokine concentrations with the exception of IL-12p40, the latter responses being the strongest for BCG.HIVA401. While the soluble factor productions after rBCG vaccination were generally higher at wk 4 compared with wk 12, for BCG.HIVA401 the levels of IL-4 and IL-5 rose. Also of note was the initially five-fold higher level of IL-10 induced by BCG.HIVA222 compared with BCG.HIVA401, while at wks 4 and 12 the opposite was true, with BCG.HIVA401 inducing approximately 35-fold higher levels of this inhibitory cytokine. Thus, multiple differences between the rBCG vaccines were detected as shown by the amounts, type and kinetics of soluble factor production.

Table 2. Soluble signaling molecule profiles produced by homologous rBCG-stimulated naïve and BCG.HIVA401/222-immune mouse splenocytesa)
Vaccination/restimulationBCG.HIVA401BCG.HIVA222
Wk04120412
  1. a

    a) Splenocytes were isolated from groups of 3 naïve or 5 BCG.HIVA401/222-immunized BALB/c mice 4 or 12 wk after mock or vaccine administration, restimulated with the same rBCG in vitro for 24 h and the production of soluble factors was determined in the supernatants using a Bioplex assay. Values are shown as median concentrations (pg/mL).

Median (pg/mL)
IL-1α1710433512011
IL-1β43182933780031214
IL-20010002
IL-3000210
IL-404301300
IL-532141110
IL-6137634123930223642
IL-9511010156420
IL-1056869176280255
IL-12p401784300120
IL-12p7010292512102
IL-13381171863500
IL-17443743 089014 43625390
Eotaxin967226725023384530
G-CSF382510 034401845881072259
GM-CSF30793771168
IFN-γ62405312184
CXCL1 (KC)15 30716 53323133 897275494
CCL2 (MCP1)2373559087194108
CCL3 (MIP1α)526859 993263512 3853851659
CCL4 (MIP1β)27778248380693303929822
CCL5 (RANTES)235122632644520745
TNF-α124876579633348256181

In summary, we demonstrated in the BALB/c mice that (i) each subunit HIV-1 vaccine stimulated a different quality HIV-1-specific CD8+ T cells, (ii) rBCG priming induced production of a number of soluble intercellular signaling molecules that was often stimulated by a subsequent heterologous vaccine boost, and (iii) the vaccination with BCG.HIVA401 and BCG.HIVA222 recombinants clearly imparted different imprints on the H peptide- and BCG-stimulated profiles of soluble factors.

Discussion

In the course of this work, two candidate HIV-1 vaccines BCG.HIVA222 and BCG.HIVA401 were compared in adult BALB/c mice and found to have overall largely similar immunological properties with several notable differences in vaccine-induced T-cell functionality. Both rBCG vaccines primed HIV-1-specific CD8+ T cells, which could be potently expanded by a heterologous boost delivered by human adenovirus-vectored HAdV5.HIVA, sheep atadenovirus-vectored OAdV7.HIVA and poxvirus MVA.HIVA vaccines.

IFN-γ ELISPOT assay is a well-standardized readout for T-cell vaccine immunogenicity and remains the first-line test used to compare vaccine regimens across many clinical trials 46. However, measurements of IFN-γ production alone failed to capture the complex vaccine-induced patterns of soluble factors and may underestimate the total frequencies of vaccine-elicited T cells 47, 48. Moreover, different viruses induce T cells with different qualities 49, 50 and, by analogy, vaccines vectored by different attenuated viruses induce phenotypically and functionally distinct insert-specific memory CD8+ T cells. Here, following administration of five vaccine modalities, we measured the capacities of peptide H-specific memory T cells to produce 23 soluble factors and found that indeed each vector type had imprinted a distinct cytokine signature. Differences were also found between the two rBCG vaccines. For example, when naïve or rBCG-immune splenocytes were stimulated with homologous rBCG, in the long-term BCG.HIVA401 induced higher levels of IL-4 and IL-10 than BCG.HIVA222. In the past, several BCG strains were compared as vaccine vectors for expression of foreign antigens, induction of cytokines and in vivo replication and persistence with some 51–53, but not all 54 studies reporting differences in individual parameters in various systems. However, it is not clear how these inter-BCG strain dissimilarities detected in adult mice would transfer to human infants, in whom differences in anti-tuberculosis efficacy of various BCG strains have not been confirmed 38, 39.

The BCG.HIVA401/222 vaccines used in priming also impacted in different ways on the boost vector-induced cytokine production. Thus, both rBCG vaccines were able to effectively prime HIV-1-specific CD8+ cells, which were boosted by heterologous HAdV5.HIVA, OAdV7.HIVA and MVA.HIVA vaccines. Notable was the potentiation by BCG.HIVA222 of β-chemokine release stimulated by MVA.HIVA; MIP-1α, MIP-1β and RANTES were all shown to compete with HIV-1 for CCR5 binding, and their binding led to down-regulation of the CCR5 receptor and further inhibition of HIV-1 infection 55, 56. The BCG.HIVA401–HAdV5.HIVA and BCG.HIVA401/222–MVA.HIVA regimens also induced significant increases in IL-6 production relative to boosting vaccines alone. IL-6 is a pro-inammatory cytokine involved in the transition from the innate into adaptive response and has an important role in T-cell growth and differentiation. However, IL-6 production also augmented HIV-1 expression in monocytic cells 57. Overall, we demonstrated that at least in the BALB/c mice, administration of lysine auxotroph of Pasteur strain and Danish strain-derived AERAS-401 led to induction of functionally distinct HIV-1-specific memory T cells. It remains a possibility that the above described differences between the two rBCG vaccines might have been contributed by the episomal (about five copies) and chromosome-integrated HIVA transgene in BCG.HIVA222 and BCG.HIVA401, respectively. Clearly, modulation of cytokines either by the choice of a vaccine regimen, or by cytokine delivery or antibody blocking is feasible; however, more complete knowledge of the balance between immune protection/pathogenesis and inhibition/enhancement of primary and chronic HIV-1 replication is required 58 before any firm recommendations can be drawn from this or any other preclinical studies.

In conclusion, our results add to the cumulative knowledge on combining multiple vaccine components into a single vaccination regimen, highlighted differences between two similar, yet distinct candidate rBCG vaccines and encourages further development of the rBCG prime-recombinant virus boost strategy as an anti-HIV-1 vaccine platform for human neonates. While, as the first step towards translation, MVA.HIVA has been safely administered to healthy African infants born to both HIV-1-negative and positive mothers (manuscripts in preparation), more immunogenic candidate vaccines vectored by adenoviruses of non-human origin have been in or entered over 10 clinical trials in the United Kingdom and Africa and may in future either substitute or further enhance the rMVA boost. rBCG remains for now the earliest vaccine prime of anti-HIV-1 responses.

Materials and methods

Preparation of the BCG.HIVAequation image vaccine

Construction of BCG.HIVA401 was described previously 13. BCG.HIVA401 was grown in protein-free 7H9 medium or in Middlebrook 7H9 medium supplemented with 10% OADC enrichment (BD Biosciences) plus 0.05% (vol/vol) tyloxapol (Sigma) at 37°C, and the titers and viabilities of the frozen stocks were determined on 7H10 agar (BD Biosciences) plates by serial dilution.

Preparation of the BCG.HIVAequation image vaccine

Construction and preparation of the BCG.HIVA222 stock was described previously 11. Briefly, mycobacterial cultures were grown in Middlebrook 7H9 broth medium or on Middlebrook agar 7H10 medium supplemented with albumin–dextrose complex (ADC, Difco) and containing 0.05% Tween 80 and 25 μg/mL kanamycin. The L-lysine monohydrochloride (Sigma) was dissolved in distilled water and used at a concentration of 40 μg/mL.

Preparation of the HAdV5.HIVA vaccine

Construction and preparation of E1-deleted, GFP-expressing HAdV5.HIVA was describe previously 10. Briefly, working virus socks were grown on HEK 293 T cells, purified using column chromatography, titred by determining the number of infectious units (IU) determined as GFP-expressing cells and stored at −80°C until use.

Preparation of the OAdV7.HIVA vaccine

Construction of OAdV7.HIVA was described previously 10. OAdV7.HIVA was grown on permissive CSL503 ovine fetal lung cells, titred using published procedures 59 and stored at −80°C until use.

Preparation of the MVA.HIVA vaccine

Construction of MVA.HIVA was described previously 8. Working vaccine stock was grown in chicken embryo broblast cells using DMEM supplemented with 10% FCS, penicillin/streptomycin and glutamine, purified on a 36% sucrose cushion, titred and stored at −80°C until use.

Preparation of the plasmid pTH.HIVA DNA vaccine

Plasmid pTH.HIVA DNA 8 was prepared using the Endo-Free Gigaprep (Qiagen) and stored at −80°C until use.

Mouse immunizations and isolation of splenocytes

Six- to 8-wk-old female BALB/c mice were immunized either without anaesthesia i.p. with BCG.HIVA401/222 or under general anaesthesia i.m. with HAdV5.HIVA, OAdV7.HIVA, MVA.HIVA or pTH.HIVA at doses and schedules outlined in the Figure legends. On the day of sacrifice, individual spleens were collected and splenocytes were isolated by pressing spleens through a cell strainer (Falcon) using a 5-mL syringe rubber plunger. Following the removal of red blood cells with RBC Lysis Buffer (Sigma), splenocytes were washed and resuspended in RPMI 1640 supplemented with 10% FCS, penicillin/streptomycin (R-10). All animal procedures and care conformed strictly to the United Kingdom Home Office Guidelines.

Peptides

Peptides RGPGRAFVTI and IFQSSMTKI designated H and P, respectively, were synthesized in an in-house facility (Weatherall Institute of Molecular Medicine, Oxford, UK), dissolved in DMSO (Sigma) at a concentration of 10 mg/mL, stored at −80°C and used at final assay concentration of 2 μg/mL unless otherwise stated.

Polychromatic flow cytometry assay

Two million cells per well of a 96-well round-bottomed plate (Falcon) were pulsed with peptide together with 50 ng anti-CD107a-FITC antibody and incubated at 37°C, 5% CO2 for 90 min before addition of GolgiStop™. After further 5 h, the reaction was terminated, the cells were washed with FACS wash buffer (PBS, 2% FCS, 0.01% azide) and a live/dead stain was applied (LIVE/DEAD Fixable Violet Dead Cell Stain Kit for 405 nm excitation, Invitrogen), at 4°C for 30 min, followed by two washes and blocked with 100 ng anti-CD16/32 at 4°C for 30 min. All subsequent antibody incubations were performed using the same conditions. The cells were washed two times and stained with 25 ng of anti-mouse CD19-Pacic Blue™, 100 ng anti mouse-CD3 PerCP-Cy5.5 (eBioscience) and 200 ng anti-mouse CD8α-PE-Texas Red (Abcam). The cells were washed two times with FACS buffer and permeabilized with BD Cytox/Cytoperm (BD Biosciences), washed two times with BD Perm/Wash buffer, before staining with 25 ng of anti-mouse IFN-γ-PE-Cy-7, 25 ng of anti-mouse TNF-α-PE, 25 ng of anti-mouse IL-2-Alexa Fluor 647. Cells were washed with BD Perm/Wash buffer and fixed with CellFIX™ and stored at 4°C until analysis. Samples were acquired on Cyan FACS machine (Dako) and the results were analyzed using Flowjo software (Tree Star).

Measurement of supernatant cytokines by Bioplex

To measure supernatant cytokine concentrations, 2×106 of mouse splenocytes in 100 μL of R-10 were added to each well of a 96-well round-bottomed plate (Costar), pulsed with the H-peptide at 2 μg/mL, rBCG at 0.015 MOI, or medium alone, and incubated at 37°C in 5% CO2 for 24 h. Plates were then spun at 500 g for 5 min, and the supernatant was removed and stored at −80°C until use. Upon thawing, the supernatant samples were spun at 17×000×g for 10 min and the cytokine assay was performed using a Bio-Plex mouse cytokine 23-Plex panel (1×96-well) kit (Bio-Rad) according to manufacturer's protocol.

In vivo killing assay

Naïve syngeneic mice were sacrificed, the splenocytes prepared as above and incubated with or without peptides in R-10 at 37°C, 5% CO2 for 90 min and washed three times. Peptide-unpulsed target cells were labeled with CMTMR (Cell Tracker Orange, Molecular Probes) only, while peptide-pulsed target cells were labeled with CFSE (Molecular Probes) at 160 nM (H) and 80 nM (P). Differentially labeled cell cultures were washed, resuspended in PBS and combined for intravenous adoptive transfer, with each animal receiving approximately 2×106 cells of each population. After 5 h, animals were sacrificed, and their splenocytes were isolated and analyzed using flow cytometry. Cytotoxicity was calculated using the following formula: Adjusted % survival=100×(% survival of peptide-pulsed cells/mean % survival of peptide unpulsed cells), followed by the calculation of % specific lysis=100−adjusted % survival 16.

Statistical analysis

Statistical significance was determined using a paired Student's t-test with a two-tailed distribution, and two way ANOVAs on group immunization data using Prism software and Excel software. Differences were considered as significant at p≤0.05.

Acknowledgements

The work was supported by MRC UK

Conflict of interest: GWB, who is an inventor of the ovine atadenovirus vector system and a former Chief Scientific Officer of Biotech Equity Partners Pty Ltd, which has licensed the vector from the CSIRO.

Ancillary