Effects of Type I Interferons on the Adjuvant Properties of Plasmid Granulocyte-Macrophage Colony-Stimulating Factor In Vivo

Abstract

While administration of granulocyte-macrophage colony-stimulating factor (GM-CSF) can induce the local recruitment of activated antigen-presenting cells at the site of vaccine inoculation, this cellular recruitment is associated with a paradoxical decrease in local vaccine antigen expression and vaccine-elicited CD8⁺ T-cell responses. To clarify why this cytokine administration does not potentiate immunization, we examined the recruited cells and expressed inflammatory mediators in muscles following intramuscular administration of plasmid GM-CSF in mice. While large numbers of dendritic cells and macrophages were attracted to the site of plasmid GM-CSF inoculation, high concentrations of type I interferons were also detected in the muscles. As type I interferons have been reported to damp foreign gene expression in vivo, we examined the possibility that these local innate mediators might decrease plasmid DNA expression and therefore the immunogenicity of plasmid DNA vaccines. In fact, we found that coadministration of an anti-beta interferon monoclonal antibody with the plasmid DNA immunogen and plasmid GM-CSF restored both the local antigen expression and the CD8⁺ T-cell immunogenicity of the vaccine. These data demonstrate that local innate immune responses can change the ability of vaccines to generate robust adaptive immunity.

While recent years have witnessed enormous advances in our understanding of how immune responses are initiated and amplified, harnessing that understanding to improve the adjuvanting of vaccines has proven difficult. An adjuvant can enhance the immunogenicity of a vaccine by modulating antigen release from a depot site and by providing immunostimulatory signals. For example, when an immunogen is formulated with complete Freund's adjuvant, antigen is released from an emulsion and immunostimulation is driven by mycobacterial cell wall components. Since adjuvants such as complete Freund's adjuvant are not safe for human use, novel depot formulations and immunostimulatory cytokines are being evaluated for their ability to improve the efficacy of clinical vaccines. However, although we have a detailed understanding of the cytokine networks that expand antigen-specific lymphocyte populations, we have not yet been successful in harnessing that understanding for improving human vaccines.

Granulocyte-macrophage colony-stimulating factor (GM-CSF) has received considerable attention as a potential adjuvant (10, 13, 19). Delivered as a protein or as a plasmid, GM-CSF has been shown to recruit and activate macrophages and dendritic cells (DCs) at the site of inoculation (8, 18). Nevertheless, GM-CSF adjuvants have only modestly enhanced humoral and helper T-lymphocyte responses to plasmid DNA vaccines expressing a variety of antigens (2, 23, 31, 32). Moreover, plasmid GM-CSF has not been shown to increase major histocompatibility complex (MHC) class I-restricted cellular immune responses in animal models (2) or in human volunteers (30).

Although the administration of plasmid GM-CSF can attract antigen-presenting macrophages and DCs to the site of vaccine antigen inoculation, these additional antigen-presenting cells (APCs) are associated with little or no amplification of immune responses. In the present study, we demonstrate that the cytokines produced by these APC populations actually cause substantial damping of the vaccine-elicited immune responses. These experiments illustrate the importance of understanding the complex cellular signaling that can contribute to expanding immune responses.

MATERIALS AND METHODS

Animals and immunizations.

Six- to 8-week-old female BALB/c mice were obtained from The Jackson Laboratory (Bar Harbor, ME) and Charles River Laboratories (Wilmington, MA). Mice were maintained under pathogen-free conditions, and experimental protocols were approved by the Harvard Institutional Animal Care and Use Committee.

For plasmid DNA immunizations, 50 μg of pVRC-HIV-1 Env IIIB gp120 or the empty pVRC plasmid (NIH Vaccine Research Center, Bethesda, MD) was injected in 100 μl of sterile saline divided between the left and right quadriceps muscles (25).

Tetramer staining assay.

Tetrameric H-2D^d complexes folded around the human immunodeficiency virus type 1 (HIV-1) IIIB V3 loop P18 epitope peptide (P18-I10 or RGPGRAFVTI) (29) were prepared and used to stain P18-specific CD8⁺ T cells as previously described (2). Mouse blood was collected in RPMI 1640 containing 40 U/ml heparin. Following lysis of erythrocytes, 0.1 μg phycoerythrin (PE)-labeled D^d/P18 tetramer in conjunction with APC-labeled anti-mouse CD8 monoclonal antibody (MAb) (Ly-2; Caltag, San Francisco, CA) was used to stain P18-specific CD8⁺ T cells. The cells were washed in phosphate-buffered saline (PBS) containing 2% fetal bovine serum and fixed in 0.5 ml PBS containing 1.5% paraformaldehyde. Samples were analyzed by two-color flow cytometry on a FACSCalibur fluorescence-activated cell sorter (FACS [BD Biosciences, Mountain View, CA]), and gated CD8⁺ T cells were examined for staining with the D^d/P18 tetramer.

Antibody ELISAs.

Four weeks following immunization with pVRC-HIV-1 Env IIIB gp120, serum anti-gp120 antibody responses were measured in the immunized mice by enzyme-linked immunosorbent assay (ELISA) as previously described (17). Ninety-six-well plates coated overnight with 100 μl/well of 1 μg/ml recombinant IIIB gp120 (ImmunoDiagnostics, Inc., Woburn, MA) in PBS were blocked for 2 h with PBS containing 5% bovine serum albumin (BSA) and 0.05% Tween 20. Sera were then added in serial dilutions and incubated for 1 h. The plates were washed three times with PBS containing 0.05% Tween 20 and incubated for 1 h with a 1/4,000 dilution of biotinylated rat anti-mouse immunoglobulin A (IgA [clone 11-44-2]), IgM (clone 1B4B1), IgG1 (clone H143.225.8), IgG2a (clone H106.771), IgG2b (clone LO-MG2b-2), and IgG3 (clone LO-MG3-7) (SouthernBiotech Birmingham, AL), followed by a 1/2,000 dilution of streptavidin-horseradish peroxidase (Southern Biotech, Birmingham, AL). For visualization of the horseradish peroxidase conjugates, o-phenylenediamine dihydrochloride substrate (Sigma) was added. The reactions were allowed to continue for 30 min and then stopped by adding 50 μl of 2.5 M H₂SO₄ per well, and optical density (OD) values were determined at 490 nm with a Dynatech MR5000 ELISA plate reader. Samples were tested in duplicate and repeated at least twice.

Muscle cell analysis.

Muscles from plasmid GM-CSF-pretreated, immunized mice were removed and minced in RPMI 1640 using a scalpel. Tissues were then incubated by gentle shaking at 37°C for 30 min in a 5 ml solution of RPMI 1640 containing 1 mg/ml collagenase type IV (Sigma-Aldrich, St. Louis, MO). Samples were centrifuged at 1,700 rpm, washed with RPMI 1640 containing 5% fetal calf serum (R5 medium), and passed through preseparation filters (Miltenyi Biotec, Auburn, CA). Muscle cells were resuspended in 100 μl R5 medium and incubated with CD16/CD32 blocking antibody for 10 min and then with antibodies specific for F4/80, CD11c, CD3, CD4, CD8, CD19, Gr1, and B220 antigens for 15 min (eBiosciences, San Diego, CA; BD Pharmingen). Samples were analyzed using an LSR II cytometer (BD Pharmingen).

Muscle cytokine analysis.

Quadriceps muscles were harvested from inoculated mice and immediately frozen in homogenization buffer (1 ml/muscle) (40 mM K₂HPO₄, 0.8 mM EDTA, 0.8 mM dithiothreitol [DTT], 8% glycerol, and 1× Promega reporter lysis buffer [pH 7.2]). Muscles were later homogenized using a homogenizer (Fisher PowerGen 100) at high speed. Homogenates were then centrifuged (3,000 × g) for 10 min, and 50 μl of supernatant was mixed with 10 μl of mouse cytometric bead array (CBA) cytokine capture beads and 50 μl of PE detection reagent, according to the manufacturer's directions (BD Pharmingen). CBA samples were washed and analyzed by FACS Array (BD Pharmingen). Alpha interferon (IFN-α), IFN-β, and GM-CSF concentrations were assayed by ELISA. One hundred microliters (100 μl) of muscle homogenate supernatant was analyzed undiluted using mouse IFN-α and -β ELISA kits, according to the manufacturer's directions (PBL Biomedical Laboratories, Piscataway, NJ).

In vivo bioluminescence measurement.

Animals were injected intraperitoneally with 100 μl of a 30-mg/ml solution of firefly luciferin in PBS (Xenogen, Alameda, CA), and 100 μl of a 20-mg/ml ketamine and 1.72-μg/ml xylazine mixture. After 20 min, imaging was performed using the Xenogen in vivo imaging system (IVIS) series 100 (Xenogen) with an integration time of 1 min. Luminescence measurements were made using Living Image software (version 2.50.1; Xenogen).

Statistics.

The statistical significance of differences between groups was determined as described in the figure legends using the GraphPad Prism program (version 4.03). A value of P < 0.05 was considered statistically significant. Error bars represent the standard error of the mean.

RESULTS

Adjuvanting with plasmid GM-CSF led to diminished cellular immune responses.

To examine the adjuvanting properties of plasmid GM-CSF, we first monitored protein GM-CSF expression in injected muscles following plasmid GM-CSF inoculation. We observed a peak expression on day 3 following plasmid GM-CSF inoculation (Fig. 1, A). Groups of mice were then pretreated with either PBS, a sham plasmid, or a plasmid DNA encoding GM-CSF. Three days later, the mice were immunized with a plasmid DNA vaccine encoding HIV-1 IIIB gp120. The kinetics and magnitude of the vaccine-elicited CD8⁺ T-cell responses to the dominant H-2D^d-restricted p18 epitope were monitored by tetramer staining of peripheral blood lymphocytes (2). Over the course of 4 weeks of observation, plasmid GM-CSF adjuvanting did not augment the frequency of gp120-specific CD8⁺ T cells elicited by plasmid gp120 immunization. In fact, the plasmid GM-CSF-pretreated group had a threefold-lower vaccine-elicited CD8⁺ T-cell response on day 12 following inoculation than the other two groups (P < 0.01) (Fig. 1B).

FIG. 1.

Plasmid GM-CSF intramuscular inoculation in mice. (A) GM-CSF protein expression in muscle of plasmid GM-CSF-inoculated mice. Mice (n = 4 mice; 8 muscles/group) were inoculated with 50 μg plasmid GM-CSF, sham plasmid, or PBS. At 4 h and 1, 2, 3, and 7 days postinoculation, muscles were isolated and homogenized and GM-CSF expression was measured by ELISA. A significant amount of GM-CSF protein was measured in plasmid GM-CSF-injected muscles (P < 0.01; t test) and reached a peak on day 3 following inoculation. Data are expressed as the mean ± standard error. (B) The CD8⁺ T-cell response elicited by a plasmid DNA immunogen was suppressed by plasmid GM-CSF pretreatment. Three groups of mice (n = 6 mice/group) were inoculated intramuscularly with 50 μg plasmid GM-CSF, 50 μg sham plasmid, or PBS and 3 days later with a plasmid gp120 vaccine construct. p18-specific CD8⁺ T-lymphocyte responses were then monitored by staining peripheral blood lymphocytes with a D^d/p18 tetramer and an anti-CD8 MAb. The plasmid GM-CSF-pretreated group had a threefold-lower peak vaccine-elicited CD8⁺ T-cell response on day 12 following plasmid gp120 inoculation than the other two groups (P < 0.01; t test). Data are expressed as the mean percentage of tetramer-positive CD8⁺ T cells ± standard error. The results shown are representative of three experiments performed.

Plasmid GM-CSF adjuvanting was associated with decreased plasmid DNA vaccine antigen expression.

To determine if pretreatment with plasmid GM-CSF affected the expression of plasmid DNA immunogens, we monitored plasmid DNA vaccine antigen expression by in vivo imaging (11). Animals were pretreated with plasmid GM-CSF, a sham plasmid, or PBS and then inoculated with a luciferase (Luc) plasmid. Vaccine antigen expression levels in the plasmid GM-CSF-pretreated animals were lower than those in untreated animals for at least 14 days following inoculation. This diminished level of Luc expression was seen as early as 4 h following vaccine inoculation. While expression was also decreased transiently in mice pretreated with a CpG-containing sham plasmid, this effect lasted only through day 3 following vaccination (Fig. 2).

FIG. 2.

Pretreatment with plasmid GM-CSF suppressed expression of a plasmid DNA immunogen. Mice (n = 4 mice/group) were inoculated with 50 μg plasmid GM-CSF, sham plasmid, or PBS on day −3 and with 50 μg plasmid Luc on day 0. Luc expression was monitored by IVIS over the ensuing 21 days. The amount of Luc expression on day 7 was significantly lower in the plasmid GM-CSF-pretreated group than in the PBS group and also lower on day 14 following plasmid Luc inoculation than in both the PBS- and sham-pretreated control groups (P < 0.05; t test). Luc expression is shown as mean ± standard error. The results shown are representative of six experiments performed.

Inflammatory cells were recruited to the site of plasmid GM-CSF inoculation.

To explore the possibility that monocyte-, macrophage-, and DC-initiated innate immune responses may suppress the ability of plasmid GM-CSF to augment vaccine-elicited adaptive immune responses, we examined the cellular infiltrates that accumulated at the site of plasmid GM-CSF and vaccine inoculation. We extracted cells from collagenase-digested muscles and characterized those cells by flow cytometry. Following plasmid GM-CSF or sham plasmid inoculation, we observed significant increases in the numbers of macrophages, plasmacytoid dendritic cells (pDCs), myeloid dendritic cells (mDCs), and T cells in muscles (Fig. 3) (data not shown). No increase in the frequency of infiltrating B cells was observed in the first 7 days following plasmid DNA inoculations (data not shown). The frequencies of macrophages and pDCs were greatest from days 1 to 3 following inoculation (Fig. 3). On day 7, significantly more mDCs were observed in plasmid GM-CSF-inoculated than in sham plasmid- or PBS-inoculated muscles (P < 0.01; t test).

FIG. 3.

Plasmid DNA and plasmid GM-CSF recruited monocytes and APCs at the site of inoculation. Mice (n = 3 mice; 6 muscles/group) were inoculated intramuscularly with 50 μg sham plasmid, 50 μg plasmid GM-CSF, or PBS, and quadriceps muscles were harvested at 4 h and on days 1, 2, 3, and 7 postinoculation. Cells were isolated from muscle by collagenase digestion, stained with MAbs, and analyzed by flow cytometry. The frequencies of cells per muscle are shown for F4/80⁺ monocytes; CD3⁻, CD19⁻, CD4⁻, CD8⁻, B220⁻, Gr1⁻, and CD11c⁺ mDCs; and CD3⁻, CD19⁻, Gr1⁺, CD11c^int, and B220⁺ pDCs. mDCs were present in larger numbers in muscle of plasmid GM-CSF-inoculated mice on day 7 than in the other groups (P < 0.01; t test). The results shown are representative of two experiments performed.

Plasmid GM-CSF transiently increased the levels of inflammatory chemokines and cytokines at the site inoculation.

The large numbers of macrophages and DCs attracted to the site of plasmid GM-CSF inoculation may alter the local cytokine milieu and thus modulate the immune responses generated by plasmid DNA immunization. To determine which cytokines are present at the site of plasmid GM-CSF administration, we measured inflammatory mediators in muscle homogenates at several time points following inoculation. We used cytometric bead analysis to assay interleukin-12p70 (IL-12p70), tumor necrosis factor alpha (TNF-α), monocyte chemoattractant protein 1 (MCP-1), and IL-6 (Fig. 4). No increase of IL-12p70 concentration was observed following either sham plasmid or plasmid GM-CSF injection, while IL-6 concentrations were transiently elevated following plasmid GM-CSF administration (P < 0.01 compared to the other two groups; t test). However, concentrations of MCP-1 and TNF-α were significantly elevated for up to 3 days following injection in muscles inoculated with plasmid GM-CSF as compared with muscles inoculated with sham plasmid or PBS (P < 0.01; t test).

FIG. 4.

Administration of plasmid GM-CSF transiently increased levels of inflammatory chemokines and cytokines in the muscle at the site of inoculation. Mice (n = 4 mice; 8 muscles/group) were inoculated with 50 μg plasmid GM-CSF, sham plasmid, or PBS. At 4 h and on days 1, 2, 3, and 7 following inoculations, muscles were isolated, homogenized, and assessed for IL-6, MCP-1, TNF-α, and IL-12p70 using a CBA assay. Significant amounts of cytokines were measured for IL-6, MCP-1, and TNF-α but not IL-12p70 at 4 h following the inoculation in the GM-CSF group (P < 0.01; t test). Increased amounts of MCP-1 and TNF-α were measured for 3 days following inoculation. Data are expressed as the mean ± standard error. The results shown are representative of four experiments performed.

Plasmid GM-CSF increased the levels of IFNs at the site of inoculation.

Type I IFNs have been shown to inhibit gene expression from a plasmid DNA (26). Accordingly, we examined type I IFN and IFN-γ expression levels following the inoculation of plasmid GM-CSF or sham plasmid (Fig. 5). We observed greater elevations in the local concentrations of both IFN-α and IFN-β cytokines, but not IFN-γ, following plasmid GM-CSF inoculation compared to the sham-inoculated group (P < 0.01; t test). Only transient expression of IFN-β was observed in the sham-inoculated mice. However, both IFN-α and IFN-β expression reached high levels on day 3 following inoculation of plasmid GM-CSF and decreased thereafter (Fig. 5).

FIG. 5.

Administration of plasmid GM-CSF increased levels of IFN-β in the muscle at the site of inoculation. Mice (n = 4 mice; 8 muscles/group) were inoculated with 50 μg plasmid GM-CSF, sham plasmid, or PBS. At 4 h and 1, 2, 3, and 7 days postinoculation, muscles were isolated and homogenized and IFN-α, IFN-β, and IFN-γ were measured by ELISA or CBA assay. Significant amounts of IFN-α and IFN-β secretion were measured in plasmid GM-CSF-injected muscles (P < 0.01; t test). Data are expressed as the mean ± standard error. The results shown are representative of four experiments performed.

Administration of anti-type I IFN-β MAbs potentiated antigen expression.

The studies of the kinetics of cytokine production in plasmid GM-CSF-adjuvanted plasmid DNA immunization suggested that IFN-β might be mediating the suppression of plasmid DNA vaccine antigen expression. To determine the role of IFN-α and IFN-β in this suppression, we codelivered an anti-IFN-α (clone Rmma-1; PBL Biomedical Laboratories, Piscataway, NJ) or anti-IFN-β (clone RMMB-1; PBL Biomedical Laboratories, Piscataway, NJ) MAb with a plasmid Luc DNA construct in plasmid GM-CSF-pretreated mice. Gene expression was monitored by in vivo imaging for 28 days (Fig. 6B). We observed higher levels of Luc expression in the anti-IFN-β-treated mice than in the control antibody-treated mice (clone 11711; R&D Systems, MN) for 14 days following immunization (P < 0.05; t test). Luc expression was not elevated in mice not pretreated with plasmid GM-CSF before injection of plasmid Luc, even when anti-IFN-β was inoculated coincidently with the plasmid Luc (Fig. 6A). Importantly, at 7 days following inoculation, in vivo Luc expression levels in the anti-IFN-β-treated animals were comparable to those in mice receiving plasmid Luc without plasmid GM-CSF pretreatment.

FIG. 6.

Administration of anti-IFN-α and anti-IFN-β MAbs enhanced expression of plasmid Luc in mice pretreated with plasmid GM-CSF. (A) Three groups of mice (n = 4 mice/group) were administered 50 μg plasmid Luc by intramuscular inoculation and were also given 250 μg anti-IFN-β or isotype control (ISO) MAb intraperitoneally at the same time. Luc expression was monitored by IVIS over the ensuing 21 days, and expression is expressed as the mean ± standard error. No significant differences in Luc expression were observed between the three groups of mice (P = 0.49; t test). (B) Mice (n = 4 mice/group) were inoculated intramuscularly with 50 μg plasmid GM-CSF on day −3 and 50 μg plasmid Luc on day 0. Coincident with the plasmid Luc inoculation, 250 μg anti-mouse IFN-α, anti-mouse IFN-β, or isotype control MAb was injected intraperitoneally, Luc was monitored by IVIS over the ensuing 28 days and is expressed as the mean ± standard error. Compared to the isotype control group, Luc expression was significantly increased for 14 days in the anti-IFN-β- but not anti-IFN-α-administer group (P < 0.05; t test). The results shown are representative of four experiments performed.

Blocking IFN-β skewed the humoral immune responses to a Th1 phenotype.

To evaluate the effects of IFN-β on the humoral immune responses generated in plasmid GM-CSF-pretreated, vaccinated mice, anti-gp120 Ig subtypes were measured in mice that were pretreated with or without GM-CSF and then vaccinated with a plasmid gp120 immunogen coadministered with either anti-IFN-β or an isotype control MAb. Antibodies were administered 4 h before plasmid DNA vaccination. Sera were sampled 4 weeks following immunization. Adjuvanting with plasmid GM-CSF increased anti-gp120 antibody titers. Plasmid DNA gp120 immunization induced IgA, IgM, and IgG1 as well as IgG2a and IgG2b anti-gp120 antibodies. Humoral immune responses following coadministration of a control MAb were not different from those of mice receiving the GM-CSF plasmid alone. However, coadministration of anti-IFN-β antibody led to decreased anti-gp120 IgA, IgM, and IgG1 antibody titers and increased IgG2a and IgG2b titers (P < 0.01; t test) (Fig. 7). Weak gp120-specific IgG3 titers were detected in all groups (Fig. 7). This increase in IgG2/IgG1 ratio is consistent with a skewing towards a Th1 T-cell response (12, 14).

FIG. 7.

Anti-gp120 antibody responses in plasmid GM-CSF-pretreated, plasmid gp120 DNA-vaccinated mice. Four groups of mice (n = 6 mice/group) were inoculated on day −3 with either 50 μg plasmid GM-CSF or sham plasmid alone. On day 0, all mice were inoculated with 50 μg plasmid gp120 DNA. Groups of plasmid GM-CSF-pretreated mice also received 250 μg anti-IFN-β MAb or the same quantity of isotype control (Cont) antibody intraperitoneally. On day 28, sera were obtained from the mice and anti-gp120 antibody titers were evaluated by ELISA. The geometric mean titer (GMT) ± standard error is shown for each group. The results shown are representative of three experiments performed.

Coadministration of anti-IFN-β MAb increased plasmid GM-CSF-adjuvanted plasmid DNA vaccine-elicited cellular immune responses.

Finally, we examined whether anti-IFN-β MAb treatment would augment cellular immune responses in GM-CSF-adjuvanted plasmid DNA-vaccinated mice. First, plasmid gp120 was injected with either PBS, isotype antibody control, or anti-IFN-β. H-2D^d/p18 tetramer staining was monitored, and we observed no difference in levels of tetramer-binding CD8⁺ T cells among the three groups (Fig. 8A). Then mice were divided into three groups and were treated with plasmid GM-CSF 3 days prior to plasmid DNA gp120 immunization. One group received anti-IFN-β MAb, another group received an isotype control antibody, and the third group received no antibody. Four hours later, all mice were immunized with plasmid DNA gp120. H-2D^d/p18 tetramer staining was performed on peripheral blood lymphocytes of these mice over the ensuing 4 weeks. Peak immune responses were significantly higher in the anti-IFN-β antibody-treated group than in the two groups of control mice (P < 0.01; t test). These elevated cellular immune responses persisted throughout the duration of the experiment (Fig. 8B).

FIG. 8.

Anti-IFN-β antibody treatment augmented CD8⁺ T-cell responses. Three groups of mice (n = 6 mice/group) received either no pretreatment (A) or pretreatment with 50 μg plasmid GM-CSF on day −3 (B). All mice were then immunized with 50 μg plasmid gp120 on day 0. Mice then either were untreated or were inoculated with 250 μg anti-mouse IFN-β antibody or an isotype control (Iso Cont) antibody. gp120-specific cellular immune responses were monitored by D^d/p18 tetramer staining of peripheral blood CD8⁺ T cells. Peak immune responses were significantly higher in the anti-IFN-β antibody-treated group (P < 0.01; t test). Data are expressed as the mean ± standard error. The results shown are representative of three experiments performed.

DISCUSSION

Previous studies have shown that plasmid GM-CSF administered as an adjuvant increases the numbers of APCs at the site of inoculation but does not enhance MHC class I-restricted cellular immune responses to plasmid DNA immunogens (2, 23, 30-32). Here we show that the failure of plasmid GM-CSF to augment plasmid DNA vaccine-elicited immune responses may be attributable to local increases in the production of certain cytokines. We demonstrate that the GM-CSF-mediated suppression of cellular immune responses is reversed by the administration of a monoclonal anti-IFN-β. Thus increased type I IFN activity suppresses the immunogenicity of plasmid DNA immunogens in this setting.

A number of mechanisms may contribute to plasmid GM-CSF induction of type I IFNs. The production of type I IFNs by activated macrophages, DCs, and pDCs has been extensively studied (5, 27). GM-CSF may induce the expression of type I IFNs by recruiting and activating these cells (8, 18, 21). At the same time, most plasmid DNA constructs contain unmethylated CpG motifs that can stimulate pDCs and induce the production of type I IFNs (3, 20). Indeed, this effect mediated by CpG motifs may explain the modest inhibition of vaccine immunnogenicity seen in sham plasmid-inoculated animals in some of the present studies.

Other mechanisms may also contribute to plasmid GM-CSF-mediated suppression of cellular immune responses. Since IFN-α has been shown to induce regulatory T-cell differentiation, type I IFNs may suppress cellular immune responses by activating regulatory T cells (15). Alternatively, DCs have also been shown to suppress cellular immune responses in some contexts (1, 28).

IFNs may decrease vaccine immunogenicity by suppressing antigen expression. In fact, other studies have shown type I IFNs can inhibit gene expression from liposome-plasmid DNA complexes (26). Consistent with this mechanism, we observed decreased levels of antigen expression when employing a plasmid GM-CSF adjuvant in the present study. IFNs can also activate cellular ribonucleases, inhibit protein translation (26), and directly suppress expression from viral promoters, including the cytomegalovirus promoters used in this study (6, 24). Interestingly, recently we have shown that pretreatment of mice with plasmid GM-CSF was not associated with a damping of Luc expression from a plasmid construct with a Rous sarcoma virus promoter (data not shown). Since type I IFNs can inhibit cytomegalovirus promoter activity through suppression of immediate-early gene expression by down regulating NF-κB activity (7), this finding suggests that the Rous sarcoma virus promoter activity may be NF-κB independent.

Type I IFNs can also diminish cellular immune responses by skewing helper T cells towards a Th2 phenotype. Indeed, we observed in the present study that blocking IFN-β with a MAb led to an increased IgG2/IgG1 ratio in the anti-gp120 antibody response, consistent with a bias toward a Th1 phenotype. Thus, the present results are in agreement with previous reports suggesting a role for type I IFNs in Th2 skewing (4, 12, 14).

Although we focused our attention on the suppressive effects of IFN-β in this study, a number of other soluble factors are upregulated by plasmid GM-CSF that may mediate a similar effect. TNF-α can induce the suppression of gene expression in the setting of lipid-complexed plasmid DNA delivery to the lung (16). Although we observed only a transient increase in local TNF-α concentrations in the present study, it is possible that this cytokine acts like IFN-α/β in suppressing plasmid DNA expression. Type I IFNs and IFN-γ may also have similar activities (9, 22, 24). It is therefore possible that multiple innate effector molecules can contribute to the suppression of gene expression. Elimination of these mediators may lead to greater increases in immunogenicity than were seen in the present study following administration of anti-IFN-β MAb.

Plasmid GM-CSF adjuvants for plasmid DNA vaccines have been advanced into human clinical trials despite their unimpressive activity in the augmentation of MHC class I-restricted cellular immune responses in preclinical studies (30). The findings in the present experiments suggest that the failure of plasmid GM-CSF to adjuvant CD8⁺ T-cell responses may be a consequence of enhanced innate immune responses mediated by GM-CSF. Transient blocking of the activity of selected mediators of this innate immune response, exemplified here by the use of an anti-IFN-β MAb, may improve the immunogenicity of plasmid GM-CSF-adjuvanted DNA vaccines.

These results highlight the importance of understanding the complexities of cytokine biology in developing cytokine-based vaccine adjuvants and suggest potential strategies for enhancing plasmid DNA vaccine immunogenicity.