Insect Baculoviruses Strongly Potentiate Adaptive Immune Responses by Inducing Type I IFN¹

Baculoviruses (BVs)³ are dsDNA viruses that are pathogenic for insects. They naturally infect arthropods and have frequently been isolated from Lepidoptera, Hymenoptera, Diptera, and Crustacea. BVs have also been shown to infect mosquitoes, including adults (1, 2), and may therefore potentially be transmitted to humans. BVs cannot replicate in mammalian or other vertebrate animal cells (3). They are therefore considered safe and selective bioinsecticides and have been used around the world against many insect pests (4). However, it was recently shown that BVs can transduce mammalian cells (5).

The most studied BV is Autographa californica multicapsid nucleopolyhedrovirus (AcMNPV). The AcMNPV envelope fuses with the insect cell membrane, and this virus requires the GP64 glycoprotein for binding to and entry into insect cells. The major capsid protein VP39 and several minor proteins form the nucleocapsid (21 × 260 nm), which encloses the viral DNA and the p6.9 protein. AcMNPV has been used widely as a eukaryotic expression vector for the production of proteins requiring posttranslational modifications, such as glycosylation, proteolytic cleavage, and fatty acylation. These recombinant proteins are widely used in research and for the production of human and veterinary vaccines.

One of the major advantages of the BV expression system is that proteins produced by this expression system are not contaminated by LPS and other bacterial compounds known to activate the innate immune system through TLRs. However, the AcMNPV population is also efficiently expanded during the production of the recombinant protein. The use of purification methods that do not sharply discriminate between the recombinant products and BV particles might result in contamination of the sample with BV virions. This may be the case for crude purification methods based on size discrimination, such as gradient centrifugation, microfiltration (0.45 μm), and ultrafiltration (300 kDa)—routine protocols for the downstream processing of different BV-derived virus-like particles (VLPs) in preclinical studies (6, 7). To obtain authorisation for commercial use, a recombinant protein must fulfil a series of safety and regulatory requirements. In particular, it must be free of any genetic material, such as live BVs, to prevent the unwanted spread of genetically modified microorganisms. However, because BVs are not pathogenic in mammals, these requirements have not necessarily been strictly adhered to during the steps preceding the commercial release of BV-derived recombinant proteins. Indeed, very few preclinical studies have evaluated BV contamination of recombinant protein preparations (6, 7, 8, 9). BVs have been reported to induce cytokine production in vitro in mammalian cells and to confer protection against lethal virus infections in vivo (10, 11, 12). However, no study has yet dealt with the effects of BV contamination on the immunological properties of BV-derived recombinant proteins.

We show here that BVs have strong adjuvant properties, promoting humoral and CTL responses against coadministered Ag, dendritic cell (DC) maturation, and the production of inflammatory cytokines. BVs exert these effects by a mechanism mediated principally by IFN-αβ, although type I IFN-independent mechanisms are also involved. We also demonstrate that BVs play a major role in the strong immunogenicity of porcine parvovirus (PPV) VLPs produced in the BV expression system. This study shows that noninfectious viruses, such as insect BVs, which are widespread in the environment, may have a strong effect on the mammalian immune system.

C57BL/6 and 129 Sv mice were obtained from Charles River Laboratories. 129 mice lacking the type I IFNR (IFNARko) were purchased from Pasteur Institute. Animals were kept under specific pathogen-free conditions. Experiments involving animals were conducted according to institutional guidelines. CTLL-2 cells were obtained from the American Type Culture collection. B3Z, a CD8⁺ specific T cell hybridoma specific for the K^b-restricted OVA_257–264 epitope, was provided by N. Shastri (University of California, Berkeley, CA) and maintained in the presence of 1 mg/ml G418 and 400 μg/ml hygromycin B in complete RPMI 1640 medium (RPMI 1640 with Glutamax; Invitrogen Life Technologies) supplemented with 10% heat-inactivated FCS, 100 U/ml penicillin, 100 μg/ml streptomycin, and 5 × 10⁻⁵ M 2-ME).

Purified AcMNPV virus was obtained from Agate Bioservice SARL. Briefly, BVs were propagated in Spodoptera frugiperda (cells in HyQ-CCM3 SFM (HyClone) (72 h). The supernatant was harvested, and the cell debris was removed by centrifugation (4,000 × g, 15 min, 4°C). The viruses were collected as a pellet by centrifugation at 57,000 × g (60 min, 4°C). This pellet was then resuspended in 1 ml of PBS. The infectious titer was determined by standard plaque assay. The virus stock was free of endotoxin (<0.01 endotoxin U/ml) (Limulus amebocyte lysate test, QCL-1000) (BioWhittaker-Cambrex).

BVs (10⁹ PFU/ml) were inactivated either by exposure to UV light (2 × 10⁴ mJ/cm²) from a UV cross-linker (Amersham Biosciences) or by treatment with binary ethylenimine (BEI) or Triton X-100. BEI was freshly prepared by the cyclization of 0.2 M 2-bromoethylamine hydrobromide in 0.4 M NaOH (2 h, 37°C). BVs were incubated with 10 mM BEI (48 h, 37°C). Residual BEI was hydrolyzed with 15 mM sodium thiosulphate. For Triton X-100 treatment, BVs were incubated (30 min, 25°C) with 1% Triton X-100 and 0.3% tri-n-butyl-phosphate and dialyzed overnight against PBS. BVs (10⁹ PFU/ml) were incubated with benzonase (90 U/ml) (Novagen) and Mg₂Cl (2 mM) for 2 h at 37°C. Benzonase was inactivated with 150 mM NaCl.

DNA from BV was isolated from purified virions by treatment with proteinase K (Sigma-Aldrich) and 10% SDS (2 h, 55°C). RNA was removed by incubation with RNase A (1 h, 37°C). Viral DNA was purified by phenol-chloroform-isoamyl alcohol extraction. The precipitated DNA was collected by centrifugation at 12,000 × g and then resuspended in sterile endotoxin-free water. We detected no protein or chromosomal DNA from the insect cells.

Wild-type (WT) PPV VLPs and chimeric PPVOVA VLPs carrying the OVA_257–264 epitope were produced using a BV expression system (13). VLPs were precipitated with 20% ammonium sulfate and dialyzed. This VLP preparation is referred to as the standard preparation. The LPS endotoxin content of the standard preparation was <0.01 U/ml. In some experiments, VLPs were purified by size-exclusion chromatography on a Sephacryl S-1000 SF column (Amersham Biosciences) (9). For BV inactivation, VLPs were treated with BEI and Triton X-100, as described above and in a previous study (9).

Abs were induced in mice by a single i.v. or s.c. injection of OVA protein (10 μg) (Calbiochem), either alone or in combination with BVs or alum (1 mg). The concentration of LPS in the OVA solution was 65 EU/mg protein. For CTL priming, mice were immunized by either i.v. or s.c. injection with 10⁹ synthetic latex beads (1 μm in diameter) (Polysciences) covalently linked to the OVA_257–264 synthetic peptide (Neosystem), as previously described (14), either alone or in combination with BVs, anti-mouse CD40 Abs (100 μg) (clone 3/23; BD Biosciences), PPV VLPs (10 μg), or a mixture containing 10 μg of purified BV-DNA or CpG 2216 (Genset) and 30 μl of a cationic liposome preparation (N-[1-(2,3-dioleoyloxyl)propyl]-N,N,N-trimethylammoniummethyl sulfate (DOTAP; Roche Diagnostic Systems).

Individual sera were tested for OVA protein-specific IgG by ELISA. IFN-α, IFN-β, monokine induced by IFN-γ (MIG), and IFN-γ-inducible protein-10 (IP-10) levels were detected in sera with ELISA kits (IFN-α and IFN-β: PBL Biomedical Laboratories) (MIG and IP-10: R&D Systems). Serum IL-12p40, IL-12p70, IL-6, and IFN-γ concentrations were determined by in-house ELISA (15). The frequency of OVA_257–264-specific IFN-γ- or IL-4-producing cells was determined by ELISPOT, as described previously (14). Results are expressed as the number of spot-forming cells (SFC) per million splenocytes. For each mouse, the number of OVA_257–264-specific IFN-γ SFC was determined by calculating the difference between the number of spots generated in the presence and absence of the OVA_257–264 peptide (1 μg/ml).

Naive syngenic splenocytes were pulsed with OVA_257–264 peptide (10 μg/ml) (30 min, 37°C), washed extensively and labeled with a high concentration (1.25 μM) of CFSE (Molecular Probes). The nonpulsed control population was labeled with a low concentration (0.125 μM) of CFSE. CFSE^high- and CFSE^low-labeled cells were mixed in a 1:1 ratio (5 × 10⁶ cells of each population) and injected i.v. into mice. The number of CFSE-positive cells remaining in the spleen after 20 h was determined by FACS.

Collagenase- and DNase I-treated splenocytes were incubated with anti-CD11c Ab-coated magnetic beads, either alone or together with anti-mpdca1 Ab-coated magnetic beads (Miltenyi Biotec). Cells were selected using an automated magnetic cell sorter (AutoMACS; Miltenyi Biotec).

Sorted splenic DCs from C57BL/6 mice (10⁵/well) were cocultured with B3Z (10⁵/well cells) and various concentrations of recombinant PPV VLP for 18 h. The degree of activation of B3Z T cell hybridomas was determined by measuring the amount of IL-2 in the supernatant in a standard CTLL-2 bioassay.

The frequency of OVA_257–264-specific CTL in vivo was determined by tetramer staining, using PE-conjugated H-2K^b-OVA_257–264-tetramer (Beckman Coulter). Results are expressed as the percentage of CD3⁺CD8⁺ cells that were OVATetr. Magnetically sorted DCs from stimulated mice were stained with anti-CD11c-allophycocyanin, anti-B220-PE and anti-CD40-FITC, anti-CD86-FITC, anti-K^b-FITC, anti-IA^b-FITC, or appropriate isotype control Abs. Cells were acquired in a FACSCalibur flow cytometer and analyzed using CellQuest Software (BD Biosciences).

We evaluated the possible induction of adaptive immune responses by BVs, by analyzing the Ab responses elicited by a single injection of the OVA protein, either alone or in combination with BVs. The coinjection of BVs and OVA protein resulted in the priming of a potent and long-lasting OVA-specific humoral response (Fig. 1,A). This adjuvant effect was observed with BV doses from 10³ to 10⁶ PFU (Fig. 1,B). We next investigated whether BVs also promoted CTL responses, using beads coated with the OVA_257–264 peptide (BOVAp). These beads can transfer CTL epitopes to the MHC class I (MHC-I) pathway of DCs in vivo but are devoid of stimulatory capacity and do not induce CTL responses unless an appropriate stimulatory signal is delivered simultaneously (14). Immunization with BOVAp and BVs induced a massive expansion of the OVA_257–264-specific CTL (OVATetr⁺) population, which accounted for 10% of the total number of CTLs on day 7 after injection (Fig. 2,A). CTLs generated in the presence of BVs displayed full lytic function in vivo, as shown by in vivo killing assays conducted 7 days after immunization. This lytic function remained evident 80 days after injection (Fig. 2,B). BVs had an adjuvant effect after both i.v. and s.c. immunization, although fewer OVATetr⁺ CTL were detected after s.c. immunization (Fig. 2,C). BVs elicited CTL responses in a dose-dependent manner, as shown either by in vivo killing assays (Fig. 2,D) and tetramer staining (Fig. 2,E). Despite the low frequency of OVATetr⁺ CTLs, we detected measurable lytic activity after the injection of BOVAp and 10³ PFU of BVs. The simultaneous administration of BOVAp and BVs induced a vigorous type 1 CTL response, characterized by a large number of IFN-γ-producing cells that increased with BV dose and by a small number of IL-4-producing cells (Fig. 2,F). This ability of BVs to increase the CTL response strongly was also observed with beads coated with three other peptides containing CTL epitopes (Fig. 3). Thus, BVs can provide the signals required to promote specific humoral and CTL responses against coinjected Ag.

We checked that the adjuvant effect of the BV preparations was not due to contaminants from the insect cells by immunizing mice with BOVAp and supernatant from BV-infected or uninfected Sf9 cells. OVA_257–264-specific CTL were primed only in mice immunized with BOVAp and supernatant from BV-infected cells (Fig. 4 A). The BV purification method was mimicked by collecting the pellet after the centrifugation of supernatants from noninfected Sf9 cells (120 ml) and resuspending it in PBS (1 ml). We detected no CTL response after immunization with BOVAp and 200 μl of this “purified” suspension (data not shown).

For further confirmation that BVs were responsible for this strong stimulation of adaptive immune responses, BV preparations (10⁹ PFU/ml) were subjected to three different virus inactivation protocols: UV irradiation, treatment with Triton X-100, and alkylation with BEI (9). The inactivation of BVs by UV irradiation has been shown to damage viral DNA and also to induce denaturation of the GP64 envelope glycoprotein (10). The nonionic detergent Triton X-100, together with the solvent tri-n-butyl-phosphate, disturbs the lipid envelope of BVs and other enveloped viruses without affecting their protein or nucleic acid components. BEI is a potent alkylating agent that selectively reacts with nucleic acids but not with proteins (16). We detected no infectious BVs after UV light or BEI treatment and only a residual titer of 7 × 10³ PFU/ml after Triton X-100 treatment. These treatments strongly impaired the capacity of BVs to promote CTL responses (Fig. 4,B). We investigated whether the adjuvant property of BVs was due to soluble contaminant oligonucleotides (ODNs), by treating purified BVs with benzonase, which digests both DNA and RNA ODNs. Due to the size of benzonase (60 kDa), we expected it to digest only soluble ODNs and not those inside the virion. Consistent with this hypothesis, the viral titer of the BV suspension was not strongly affected by benzonase treatment (1 × 10⁹ and 1.6 × 10⁸ PFU/ml before and after treatment, respectively). BVs digested with benzonase had similar adjuvant properties to untreated viruses (Fig. 4 B). Thus, a compound intrinsically associated with BVs, rather than a soluble contaminant from the Sf9 cell culture, was responsible for the adjuvant properties of the BV preparations. We also found that the strong adjuvant properties of BVs depended on the ability of these viruses to infect insect cells because disruption of the viral envelope and/or damage to the viral DNA abolished these properties.

It has been reported that baculoviral DNA (BV-DNA) contains CpG motifs (10). Thus, we examined the effect of purified BV-DNA on the induction of CTL responses by immunizing mice with BOVAp together with BV-DNA. No CTL response was detected after priming mice with BOVAp, with either CpG-ODN or BV-DNA (Fig. 4,C). We suspected that inadequate uptake and/or a rapid degradation of the naked DNA was preventing the induction of an effective CTL response. Therefore, we immunized mice with BOVAp together with either BV-DNA or CpG-ODN mixed with the cationic liposomal preparation DOTAP, which facilitates the uptake of DNA and prolongs its life in vivo (17). A strong specific CTL response was detected in both cases (Fig. 4 C), whereas no CTL response was detected after immunization with BOVAp and DOTAP alone. We detected no CTL response after immunization with BOVAp- and benzonase-treated BV-DNA mixed with DOTAP (data not shown). Thus, BV-DNA delivered by an appropriate vehicle can promote CTL responses, suggesting that the viral DNA may be responsible for the adjuvant properties of BVs.

We and others (18, 19, 20, 21, 22, 23, 24) have shown that VLPs produced by the BV expression system are highly efficient vehicles for delivering heterologous CTL epitopes to the MHC-I pathway because they can induce potent immune responses after a single injection in the absence of exogenous adjuvant. We have also recently shown that PPV VLPs are strong adjuvants when independently provided with Ag (14). Recombinant BVs have been detected in standard PPV VLP preparations, reaching titers greater than 10⁷ PFU/ml (9). The data presented here may therefore suggest that the strong immunogenicity and adjuvant properties of PPV VLPs produced in the BV expression system are due to the adjuvant properties of BVs. Therefore, we used various procedures, such as size exclusion chromatography (SEC) to remove BVs from VLP preparations, or chemical treatment with BEI or Triton X-100 to inactivate them. SEC can easily remove BVs because BVs and PPV VLPs differ considerably in size (9). The BV titer in preparations of PPV VLPs carrying the OVA_257–264 epitope (PPVOVA) was 5.2 × 10⁷ PFU/ml and that in standard control PPV VLPs preparations was 9.3 × 10⁷ PFU/ml. We detected no BVs by plaque assay in preparations treated with BEI or Triton X-100 or after purification by SEC, confirming previous results (9). These treatments were previously shown to have no effect on the physical integrity of PPV VLPs (9). Purification by SEC or treatment with BEI or Triton X-100 did not affect the capacity of PPVOVA VLPs to deliver the OVA epitope to the MHC-I pathway, as shown by a classical in vitro presentation assay (Fig. 5,A). In contrast, these treatments strongly reduced the ability of PPVOVA VLPs to induce CTL responses following in vivo immunization (Fig. 5,B). However, when mice were immunized with BEI-treated PPVOVA together with BVs, the CTL response was restored in a BV dose-dependent manner, confirming that BV removal had affected the adjuvant properties but not the delivery capacity of PPVOVA. The VLP adjuvant effect was also greatly reduced in mice immunized with BOVAp in combination with PPV VLPs purified by SEC or treated with BEI or Triton X-100 (Fig. 5 C). These data clearly indicate that the adjuvant properties of standard PPV VLP preparations were due to the presence of BVs.

Type I IFNs are produced by most, if not all cells, in response to virus infection (25) and are an essential link between innate and adaptive immunity (11, 26). Therefore, we thought that these molecules might be involved in the adjuvant effect of BVs. The injection of BVs into mice led to a dose-dependent increase in IFN-β and IFN-α levels in serum (Fig. 6,A) that was fully abolished by BV inactivation (Fig. 6,B). We were unable to detect IFN-α after the injection of 200 μl of supernatant from uninfected Sf9 cells, whereas we detected high levels of IFN-α (17.2 ± 4.3 ng/ml) after injection of the same volume of a supernatant from BV-infected Sf9 cells (BV dose: 10⁶ PFU) (data not shown), showing that the in vivo production of IFN-αβ was due to BVs and not to Sf9 cell contaminants. We investigated the way in which the type I IFNAR regulated the induction by BVs of IFN-αβ production in vivo in WT (129Sv) and IFNAR-deficient (IFNARko) mice. Sustained levels of IFN-β were detected in the sera of WT mice between 2 and 8 h after injection, whereas a rapid, massive increase in IFN-β levels followed by a marked decrease was observed in IFNARko mice, with IFN-β being almost undetectable 6 h after injection (Fig. 6,C). The absence of the type I IFNR, resulting in a lack of secreted IFN-β sequestration, may account for the high level of IFN-β observed in IFNARko mice. By contrast, levels of IFN-α production were markedly lower in IFNARko mice than in WT mice (Fig. 6 C). Thus, the in vivo production of IFN-αβ in response to BVs is strongly regulated by IFNAR. The kinetics of production were consistent with the positive-feedback model proposed by Taniguchi and Takaoka (27).

We then investigated whether the strong adjuvant properties of BVs depended on IFN-αβ. WT and IFNARko mice were immunized with the OVA protein, alone or in combination with BVs. As in C57BL/6 mice, BVs markedly increased the OVA-specific IgG response in 129Sv mice but not in IFNARko mice (Fig. 6,D). However, we still observed a small increase in Ab titers in IFNARko mice, indicating that BVs can promote the induction of Abs independently of IFN-αβ. We assessed the effect of IFNAR signaling on the adjuvant effect of BVs on the CTL response by immunizing 129Sv and IFNARko mice with BOVAp in combination with BVs. A much weaker, but nonetheless detectable, adjuvant effect of BVs on CTL responses was observed in IFNARko mice (Fig. 6,E). We investigated whether the impaired immune response observed in IFNARko mice was due to the lack of IFNAR signaling rather than a general defect in this strain by immunizing WT and IFNARko mice with the OVA protein mixed with alum or with BOVAp, together with anti-CD40 mAbs. IFNARko mice mounted humoral (Fig. 6,F) and CTL (Fig. 6 G) responses similar to those of control mice.

Several recombinant proteins produced in the BV expression system have been shown to activate DCs and to produce large amounts of inflammatory cytokines in vivo (28, 29, 30, 31, 32). We investigated whether BVs could induce DC maturation by injecting C57BL/6 mice with PBS or BVs and analyzing phenotypic activation markers on splenic plasmacytoid DC (pDC) and conventional DC (cDC) subsets. BVs strongly activated pDCs and cDCs in vivo (Fig. 7,A). This activation was totally abolished by BV inactivation (Fig. 7,B) and was strongly dependent upon IFNAR signaling (Fig. 7,C). Finally, we analyzed the ability of BVs to induce the in vivo secretion of inflammatory mediators (Fig. 8). High levels of IL-12p40, IL-12p70, IL-6, IFN-γ, MIG, and IP-10 were detected in the sera of mice injected with BVs. The secretion of these inflammatory mediators was totally abolished if BVs were inactivated by UV light. In the absence of IFN-αβ signaling, the BV-induced production of IL-6, IFN-γ, MIG, and IP-10 was strongly impaired, whereas the secretion of IL-12p40 and p70 was enhanced.

This study demonstrates that nonpathogenic insect BVs may have a strong effect on mammalian immune responses. We show that BVs have strong adjuvant properties, promoting humoral and CTL responses against coadministered Ag, DC maturation, and the production of inflammatory mediators through mechanisms primarily mediated by IFN-αβ.

It has been shown that lymphocytic choriomeningitis virus (LCMV) promotes CD8⁺ T cell priming against an independently coadministered exogenous Ag through mechanisms highly dependent on type I IFN (33). IFN-α injection has also been shown to enhance humoral and CTL responses against soluble protein (33, 34). The immunostimulatory activity of IFN-α results, at least partly, from its ability to induce DC maturation directly (35). Several laboratories have recently demonstrated that type I IFNs also act directly on naive B cells and CD4⁺ and CD8⁺ T cells, leading to clonal expansion and differentiation into effector (36, 37, 38, 39) and memory cells (37). However, IFNAR-signaling independent mechanisms are also involved in the adaptive immune responses promoted by BVs. DC licensing through CD40 ligation and/or signaling through other inflammatory cytokines, such as IL-12, may contribute to the greater stimulatory capacity of BVs.

The precise mechanism through which BVs exert their adjuvant effects remains unclear. It has been shown that BVs efficiently transduce a wide range of mammalian cells (5) but do not replicate in these cells (3). BVs are thus now recognized as useful viral vectors for gene delivery to mammalian cells. The nature of the molecules involved in BV uptake remains unclear. Several studies have shown that the viral envelope glycoprotein, GP64, plays a key role in the interaction of BVs with mammalian cells (10, 12, 40). The GP64 glycoprotein contains mannose, fucose, and N-acetyl-glucosamine residues (41). It has been suggested that GP64 interacts with the mannose receptor, which is expressed principally on macrophages and DCs. Mannose receptor is involved in endocytosis/phagocytosis and plays an important role in host defenses and the induction of innate immunity (42). However, there are several lines of evidence to suggest that GP64 is not the principal actor in the stimulatory properties of BVs. Abe et al. (10) reported that BVs activated the production of inflammatory cytokines in peritoneal exudate cells (macrophages in their majority) and splenic cDCs via the TLR9/MyD88-dependent pathway and that RAW264.7 cells “infected” with BVs or transfected with BV-DNA secreted inflammatory cytokines, whereas the removal of GP64 from the BV virion and pretreatment with purified GP64 glycoprotein or lysosomotropic agents, such as chloroquine, decreased cytokine production. These data strongly suggest that the internalisation of viral DNA mediated by GP64 glycoprotein and endosomal maturation, which may release the viral DNA into TLR9-expressing cellular compartments, make important contributions to the stimulatory properties of BVs in vitro. Our data support this hypothesis, as the treatment of BVs with BEI, a potent alkylating agent that reacts selectively with nucleic acids, but not with proteins (16), strongly decreased the ability of BVs to induce the cross-priming of CD8⁺ T cells. Moreover, immunization with BOVAp plus BV-DNA in cationic liposomes was sufficient to induce a CTL response. However, significant levels of IFN-α production are detectable in TLR9- and MyD88-deficient peritoneal exudate cells and DCs stimulated in vitro with BVs (10), suggesting that viral components other than BV-DNA or TLR9/MyD88-independent BV-DNA recognition pathways are also involved in the immunostimulatory properties of BVs. Studies are currently underway in our laboratory to decipher the mechanisms through which BVs exert their adjuvant properties.

The adjuvant effect of BVs was observed with numbers of virions (10⁴-10³ PFU) within the range potentially contaminating recombinant protein preparations generated in BV expression systems, such as VLPs (8, 9). Surprisingly, the potential effect of contaminating insect viruses on the immune responses induced by recombinant proteins produced in BV expression systems has never been studied. Recombinant VLPs are most commonly purified by size discrimination methods, such as gradient centrifugation, microfiltration, and ultrafiltration, which do not readily discriminate between VLPs and BVs (6, 7). Therefore, BVs are likely to be present in many of these preparations. The lack of BV pathogenicity, at least in mammals, may have led to a lack of concern about the possible presence of BVs in BV-derived recombinant protein preparations. Indeed, very few studies have provided accurate information about the level of BV contamination in their protein preparation (8, 9).

Therefore, the data presented here strongly suggest that some of the immunological properties attributed to proteins produced in the BV expression system may be due to contaminant BVs. Indeed, several BV-derived VLPs have been shown to induce the production of inflammatory cytokines in vivo and to activate macrophages and DCs through MyD88 pathway-dependent mechanisms (28, 29, 30, 31, 32). These stimulatory properties have been attributed to the particulate structure of the VLPs, certain structural components, or to internal contaminants that may be recognized by mammalian cells as pathogen-associated molecules. However, we clearly demonstrate here that, at least for PPV VLPs, the contaminating BVs present in the preparations are strictly required for the induction of CTL responses by such recombinant VLPs. The adjuvant effect of these particles was also strongly affected by the removal or inactivation of BVs.

Finally, many of the studies of innate immune response activation by microbial products to date have focused on molecules derived from pathogens. This study demonstrates that, although insect viruses are not harmful to mammals, they can have a strong effect on mammalian innate and adaptive immune responses. It remains to be determined whether humans are exposed to such viruses through inoculation by insect vectors or through the release of these viruses into the environment and whether BVs affect the human immune system.

We thank Nicolas Escriou for assistance with UV virus inactivation and Richard Lo-Man for critical discussions.

The authors have no financial conflict of interest.