Preexposure of Murine Macrophages to CpG Oligonucleotide Results in a Biphasic Tumor Necrosis Factor Alpha Response to Subsequent Lipopolysaccharide Challenge

Abstract

Bacterial DNA and synthetic oligonucleotides containing CpG sequences (CpG-DNA and CpG-ODN) provoke a proinflammatory cytokine response (tumor necrosis factor alpha [TNF-α], interleukin-12 [IL-12], and IL-6) and increased mortality in lipopolysaccharide (LPS)-challenged mice via a TNF-α-mediated mechanism. It was hypothesized that preexposure of macrophages to CpG-ODN would result in an increased TNF-α response to subsequent LPS challenge in vitro. Using the murine macrophage cell line RAW 264.7, we demonstrated both a rapid proinflammatory cytokine response (TNF-α) and a delayed inhibitory cytokine response (IL-10) with CpG-ODN. Preexposure of macrophages to CpG-ODN for brief periods (1 to 3 h) augmented TNF-α secretion and mRNA accumulation following subsequent LPS challenge (1 μg/ml). However, prolonged preexposure to CpG-ODN (6 to 9 h) resulted in suppression of the TNF-α protein and mRNA response to LPS. The addition of anti-IL-10 antibody to CpG-ODN during preexposure resulted in an increase in the LPS-induced TNF-α response over that induced by CpG-ODN preexposure alone. Thus, while brief preexposure of macrophages to CpG-ODN augments the proinflammatory cytokine response to subsequent LPS challenge, prolonged preexposure elicits IL-10 production, which inhibits the TNF-α response. Although the initial proinflammatory effects of CpG-DNA are well established, the immune response to CpG-DNA may also include autocrine or paracrine feedback mechanisms, leading to a complex interaction of proinflammatory and inhibitory cytokines.

In the past 10 years, there has been increasing recognition of the immunostimulatory properties of bacterial DNA and synthetic oligonucleotides containing an unmethylated cytosine followed by guanine (CpG-DNA and CpG-ODN). CpG-DNA was initially shown to stimulate lymphocyte proliferation, gamma interferon (IFN-γ) production, and natural killer (NK) cell tumoricidal activity (21, 29, 31–33). Subsequent studies focused on CpG-DNA stimulation of proinflammatory cytokine secretion, B-cell stimulation, and the preferential induction of a Th1-cell response. CpG-DNA and synthetic CpG-ODN stimulate the proinflammatory cytokines interleukin-6 (IL-6), IL-12, and IFN-γ in mixed splenocytes but fail to stimulate IL-2, IL-3, IL-4, IL-5, or IL-10 (15, 16, 34). In addition, prolonged incubation (12 to 24 h) with CpG-DNA or CpG-ODN stimulates tumor necrosis factor alpha TNF-α secretion in macrophage cell lines and murine peritoneal macrophages (28, 35; T. Sparwasser, T. Miethke, G. Lipford, K. Borschert, H. Hacker, K. Heeg, and H. Wagner, Letter, Nature 386:336–337, 1997). In vivo, intraperitoneal injection of CpG-ODN produces an early (1 to 2 h) increase in serum TNF-α levels while intratracheal administration of CpG-ODN results in increased TNF-α levels in lavage fluid (25, 28). Bacterial DNA and CpG-ODN cause significant mortality in d-galactosamine-sensitized mice via TNF-α-mediated liver cell apoptosis (Sparwasser et al., Letter). Additionally, in vivo preexposure with bacterial DNA followed 1 to 4 h later by lipopolysaccharide (LPS) injection results in a significant increase in serum TNF-α levels and mortality in mice with respect to LPS challenge alone (5, 28; Sparwasser et al., Letter). On the other hand, Gao et al recently demonstrated that preexposure of RAW 264.7 macrophages to CpG-ODN in vitro suppresses LPS induction of nitric oxide production with respect to that induced by LPS alone (10) and Schwartz et al demonstrated decreased pulmonary inflammation in response to LPS after systemic exposure to CpG-DNA (26). Thus, despite the potential utility of the immunostimulatory properties of the CpG-DNA, e.g., vaccine adjuvants (6, 7, 19, 27), there remains concern regarding the potentially detrimental effects of CpG-DNA-induced alterations in cytokine regulation.

To further characterize the macrophage cytokine response to CpG motifs, we used a murine macrophage cell line, RAW 264.7, and elicited murine peritoneal macrophages. We hypothesized that CpG-ODN preexposure in vitro would result in a sensitization of the macrophage TNF-α response to LPS in a time-dependent manner. It was discovered, however, that although short CpG-ODN preexposure led to early sensitization of macrophages to LPS, with a resultant increase in TNF-α secretion with respect to that due to LPS alone, prolonged preexposure (6 to 9 h) resulted in desensitization of the response to LPS, with decreased levels of TNF-α mRNA and protein secretion. This desensitization was shown to be partially dependent on IL-10-mediated inhibition of TNF-α transcription, suggesting a complex system of cytokine responses to CpG-DNA that include negative-feedback mechanisms following an initial proinflammatory phase.

MATERIALS AND METHODS

Mice and peritoneal macrophages.

In vivo experiments were performed using female BALB/c mice (Hilltop Labs, Scotsdale, Pa.) weighing 20 to 25 g each. The animals were housed in a pathogen-free environment and fed laboratory chow (Purina, St. Louis, Mo.) and water ad libitum, in accordance with National Research Council Standards. All procedures were approved by the University of Virginia Animal Use Committee.

Three days prior to macrophage harvest, mice were injected intraperitoneally with 1 ml of sterilized 3% Brewer thioglycolate medium (Difco Products, Becton-Dickinson) containing 1% (each) penicillin and streptomycin. The mice were subsequently sacrificed by halothane anesthesia and cervical dislocation. Peritoneal macrophages were then harvested under sterile conditions, washed twice in phosphate-buffered saline, and resuspended in medium at the desired concentration. An α-naphthyl acetate esterase assay (Sigma, St. Louis, Mo.) was performed on a sample of the cell suspension to confirm the purity of macrophages within the cell population (they were >80% pure), and viability was determined using trypan blue exclusion.

Cell culture techniques.

For most in vitro experiments, the murine macrophage cell line RAW 264.7 (ATCC TIB 71; American Type Culture Collection, Rockville, Md.) was used. Cells were cultured in 250-ml sterile culture flasks (Corning, Inc., Corning, N.Y.) containing Dulbecco's modified Eagle's medium with 4 mM l-glutamine and 4.5 g of glucose per liter, supplemented with 1.0 mM sodium pyruvate and 10% fetal bovine serum (Gibco BRL, Life Technologies, Inc., Grand Island, N.Y.). The cells were incubated at 37°C under 5% CO₂ and, prior to each experiment, were washed twice in phosphate-buffered saline and resuspended in medium. The macrophages were placed in 96-well polystyrene plates, using 1.5 × 10⁶ macrophages/well in Dulbecco's modified Eagle's medium for in vitro assays.

CpG- and non-CpG-containing oligonucleotides and LPS.

CpG-containing oligonucleotides 5′-ATA ATC GAC GTT CAA GCA AG (CpG) and non-CpG-containing sequences 5′-ATA ATA GAG CTT CAA GCA AG (non-CpG) were synthesized on a DNase-resistant phosphorothioate backbone (Bio-Synthesis, Inc., Lewisville, Tex.) as previously described (5, 16, 25). A standard Limulus amebocyte lysate assay (Endosafe) showed that the endotoxin content of the synthesized oligonucleotides after reconstitution was less than 0.3 pg/μg of oligonucleotide (N = 3). LPS from E. coli strain O128:B12 (Sigma) was suspended in sterile 0.15 M NaCl for in vivo experiments or medium for in vitro experiments. All experiments were confirmed using a second CpG-containing oligonucleotide, 5′-TCC ATG ACG TTC CTG ATG CT. For all experiments, an ODN concentration of 1.5 μg/ml was utilized, based on a prior investigation which revealed 1.5 μg/ml to be the lowest concentration of CpG-ODN capable of eliciting consistent TNF-α secretion in RAW 264.7 cells (Fig. 1).

FIG. 1.

Dose response of oligonucleotide stimulation of TNF-α secretion in RAW 264.7 cells. A total of 1.5 × 10⁶ cells were treated with different concentrations of CpG-ODN, non-CpG-ODN, or medium alone for 24 h, and the TNF-α level in supernatant was measured. Values represent the means of at least three experiments. Error bars represent the standard error of the mean. ∗, P < 0.05 versus non-CpG-ODN.

Measurement of cytokine levels.

IL-10 secretion was measured by an enzyme-linked immunosorbent assay (ELISA) using a mouse IL-10 Duoset kit (R&D/Genzyme, Cambridge, Mass.) containing primary and secondary antibodies along with horseradish peroxidase-streptavidin. TMB (3,3′,5,5′-tetramethylbenzidine) was used as a substrate (Sigma), and the plates were read at 450 nm on an ELISA plate reader. To inhibit IL-10 bioactivity in vitro, anti-mouse IL-10 monoclonal antibody (clone JES5-16E3; Pharmingen, San Diego, Calif.) was added to designated wells at a concentration of 10 μg/ml. Additional controls were performed using an isotype control antibody (rat immunoglobulin G2b [IgG2b], clone A95-1; Pharmingen). Preliminary experiments with 20 μg of anti-IL-10 antibody per ml showed no difference from those with 10 μg/ml in IL-10 neutralization for macrophages in our system. TNF-α was measured using a TNF-α ELISA Minikit (Endogen, Woburn, Mass.) containing the primary and secondary antibodies and horseradish peroxidase-streptavidin.

Measurement of cellular cytokine mRNA levels and cell surface markers.

TNF-α, IL-6, and transforming growth factor β (TGF-β) mRNA were quantified using RNase protection assays (Pharmingen). Following treatment of RAW 264.7 cells (2 × 10⁶) with CpG-ODN, non-CpG-ODN, or medium alone under various conditions, the cells were washed and total RNA was isolated and purified using an RNA purification kit (RNeasy Minikit; Qiagen, Inc., Valencia, Calif.). To generate the TNF-α mRNA probes, the MCk-3 template set (Pharmingen) (which includes a template for multiple murine cytokine mRNA probes including TNF-α; the TNF-α probe protects a 287-base sequence) was incubated with [³²P]UTP in the presence of RNasin, GACU, dithiothreitol, RNA polymerase, and transcription buffer (in vitro transcription kit, Pharmingen) and incubated for 1 h. Following treatment with DNase, we added EDTA, Tris-saturated phenol, chloroform-isoamyl alcohol (50:1), and yeast tRNA. The aqueous phase was removed, treated with 4 M ammonium acetate and ice-cold ethanol, and incubated for 30 min at −70°C; the pellet was washed with 70% ethanol, air dried, and solublized in buffer. Using a scintillation counter, representative samples were quantified (Cerenkov counts per microliter). The previously prepared RNA and an aliquot of the probe set were incubated at 56°C for 12 to 16 h in an Omnigene thermal cycler (Hybaid, Inc., Woodbridge, N.J.) and then treated with RNase. Samples were electrophoresed on an acrylamide-bisacrylamide (19:1) gel containing 40% acrylamide and 2% bisacrylamide (Bio-Rad Laboratories, Hercules, Calif.). The gel was dried and placed on film, and the film was exposed at −70°C overnight and developed. Using the undigested probes as markers, a standard curve was plotted to establish the identity of the RNase-protected bands in experimental samples. The films were developed using photodensitometry to quantify ³²P activity associated with the mRNA in each sample. Levels are reported as the ratio of TNF-α to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) to control for the amount of RNA loaded in each sample on the gel. GAPDH mRNA is constituitively expressed in these cells.

To quantify relative changes in cell surface CD14 and MAC-1 expression in vitro, RAW 264.7 cells were incubated for various time intervals with CpG-ODN or non-CpG-ODN. At each designated time interval, the cells were washed and labeled with fluorescein isothiocyanate-conjugated rat anti-mouse CD11b antibody (clone M1/70) or rat IgG2b isotype control antibody (clone R35-38; Pharmingen) and incubated for 15 to 30 min at 37°C. Additional cells were labeled with fluorescein isothiocyanate-conjugated rat anti-mouse CD14 antibody (clone rmC5-3) or rat IgG1 isotype control antibody (clone R3-34; Pharmingen) and incubated for 15 to 30 min at 37°C. Flow cytometric analysis was performed using a FacSTAR flow cytometer system (Becton Dickinson, Mountain View, Calif.). Unstained cells were washed and treated in a similar manner to measure the level of autofluorescence. Values are reported as mean channel fluorescence.

Statistical analysis.

Values for protein concentration and mRNA levels were compared using analysis of variance and post hoc by Tukey's honestly significant difference test to compare the means. The slopes of lines were compared using linear regression. P ≤ 0.05 was considered significant. Values are reported as the mean and standard error of the mean. All calculations were performed using statistical software (Statistica; Statsoft, Tulsa, Okla.).

RESULTS

CpG-ODN stimulates delayed secretion of IL-10 in macrophages.

Although CpG motifs augment a proinflammatory cytokine response (e.g., TNF-α, IL-12, IL-6, and IFN-γ), there are conflicting data regarding the role of CpG oligonucleotides in stimulation of inhibitory or regulatory cytokines such as IL-10. A total of 1.5 × 10⁶ RAW 264.7 cells were incubated with 1.5 μg of CpG or non-CpG-ODN per ml for various periods, and the IL-10 level in the supernatant was measured. Unlike non-CpG-ODN, CpG-ODN stimulation of RAW 264.7 cells produced an increase in IL-10 secretion after 6 and 9 h of incubation (Fig. 2). This finding prompted evaluation for a potential IL-10-mediated regulation of the proinflammatory response following the initial CpG stimulation of TNF-α. Treatment with an anti-IL-10 antibody (10 μg/ml) did not, however, alter the primary TNF-α response to CpG-ODN stimulation with respect to the response to CpG-ODN alone (Fig. 3).

FIG. 2.

CpG-ODN stimulation of IL-10 protein secretion in supernatants of RAW 264.7 cells. A total of 1.5 × 10⁶ cells were treated with CpG-ODN (1.5 μg/ml), non-CpG-ODN, or medium alone for 0.5 to 9 h, and the IL-10 level in supernatant was measured. Values represent the means of at least three experiments. Error bars represent the standard error of the mean. ∗, P < 0.05 for CpG versus other groups.

FIG. 3.

Effects of anti-IL-10 antibody on CpG-ODN stimulation of TNF-α production in RAW 264.7 cells. A total of 2.0 × 10⁶ cells were treated with CpG-ODN (1.5 μg/ml) with or without anti-IL-10 antibody (Ab) (10 μg/ml), non-CpG-ODN with or without anti-IL-10 antibody, medium, or anti-IL-10 antibody alone for 0.5 to 9 h, and the TNF-α level in supernatant was subsequently measured. Values represent the means of at least three experiments. Error bars represent the standard error of the mean. ∗, P < 0.01 for CpG or CpG plus anti-IL-10 antibody versus non-CpG-ODN, non-CpG-ODN plus anti-IL-10 antibody, anti-IL-10 antibody alone, or medium.

CpG-ODN-stimulated IL-10 inhibits late TNF-α mRNA production.

IL-10 has previously been shown to significantly regulate the LPS-induced production of TNF-α mRNA in vitro (5, 30). Although anti-IL-10 antibody had a minimal effect on CpG-induced TNF-α protein secretion during prolonged incubations, the role of IL-10 regulation of CpG-DNA-induced stimulation of TNF-α mRNA was studied. RAW 264.7 cells (2 × 10⁶) were incubated for various periods with CpG or non-CpG-ODN (1.5 μg/ml) in the presence or absence of anti-IL-10 antibody (10 μg/ml) and harvested, and the level of TNF-α mRNA was measured and quantified. During the initial CpG-ODN augmentation of TNF-α mRNA production, anti-IL-10 antibody appeared to have little effect. However, longer treatment with anti-IL-10 antibody plus CpG-ODN resulted in persistent elevation of TNF-α mRNA levels while incubation with CpG-ODN alone resulted in a decline of TNF-α mRNA levels (Fig. 4). Additionally, treatment under identical conditions with isotype control antibody revealed no significant differences from treatment with CpG-ODN alone at all time points.

FIG. 4.

Effects of anti-IL-10 antibody on CpG-ODN stimulation of TNF-α mRNA production in RAW 264.7 cells. A total of 2.0 × 10⁶ cells were treated with CpG-ODN (1.5 μg/ml) with or without anti-IL-10 antibody (αIL-10) or isotype control antibody (10 μg/ml), non-CpG-ODN with or without anti-IL-10 antibody, medium (M) with or without anti-IL-10 antibody, or isotype control antibody for 0.5 to 9 h. The cells were harvested, and the TNF-α mRNA level was measured using the RNase protection assay. (a) Representative gel from at least three experiments. Each experiment was performed under identical conditions with similar results. Lanes: M, medium; ∗, medium plus anti-IL-10 antibody; †, medium plus isotype control. (b) Graphical depiction of TNF-α mRNA level as a function of time normalized to GAPDH [³²P]mRNA expression. Error bars represent the standard error of the mean. ∗, P < 0.05 for CpG or CpG plus anti IL-10 antibody (Ab) versus non-CpG-ODN and non-CpG-ODN plus anti-IL-10 antibody; †, P < 0.05 for CpG plus anti IL-10 antibody versus non-CpG-ODN and non-CpG-ODN plus anti-IL-10 antibody.

In vitro preexposure of macrophages to CpG-ODN results in a biphasic TNF-α response to subsequent LPS challenge.

Following the demonstration of the direct immunomodulatory activity of CpG-ODN, subsequent analyses were performed to characterize the effects of CpG-ODN preexposure of macrophages on the response to a second proinflammatory stimulus, LPS. Based on previous in vivo data (5, 28; Sparwasser et al., Letter), it was hypothesized that CpG-ODN preexposure of macrophages in vitro would result in a time-dependent sensitization of the TNF-α response to LPS challenge. RAW 264.7 cells (1.5 × 10⁶) were incubated with CpG-ODN or non-CpG-ODN (1.5 μg/ml) for various periods in medium, washed, and incubated with 1 μg of LPS per ml for 1.5 h, and the TNF-α level in supernatant was immediately measured. In the early periods (1 to 3 h), preexposure to CpG-ODN did result in a significant increase in LPS-induced TNF-α levels in supernatant with respect to exposure to non-CpG-ODN and LPS alone; however, preexposure to CpG-ODN for longer periods (6 to 9 h) resulted in a decreased LPS-induced TNF-α response with respect to controls (Fig. 5a). To examine the potential regulatory role of IL-10 in this process, RAW 264.7 cells were treated with CpG-ODN plus 10 μg of anti-IL-10 antibody per ml for various periods and subjected to LPS stimulation. The addition of anti-IL-10 antibody during CpG-ODN preexposure for 6 and 9 h augmented LPS-induced TNF-α secretion with respect to CpG-ODN preexposure alone, suggesting that in this system local IL-10 at least partially mediates macrophage insensitivity to LPS as measured by TNF-α secretion (Fig. 5b). Additional experiments revealed no difference in the LPS-induced TNF-α response following preexposure to CpG-ODN alone and CpG-ODN plus isotype control antibody over all time points (data not shown). Furthermore, cell viability remained greater than 85% in the 6- and 9-h groups, thus excluding cell death as the cause of decreased TNF-α production at these time points.

FIG. 5.

(a) Effect of CpG-ODN preexposure of RAW 264.7 cells on TNF-α secretion following subsequent LPS challenge. A total of 1.5 × 10⁶ cells were treated with 1.5 μg of CpG-ODN or non-CpG-ODN per ml or medium alone for 0.5 to 9 h. After each period, the cells were washed and treated with LPS (1 μg/ml) for an additional 1.5 h (excluding the medium-no LPS series in which cells were treated similarly except for LPS exposure), and the TNF-α level in supernatant was measured. Values represent the means of at least three experiments. Error bars represent the standard error of the mean. ∗, P < 0.05 for CpG versus other groups; †, P < 0.05 for CpG or medium-no LPS versus all other groups. (b) LPS-induced TNF-α secretion following preexposure of RAW 264.7 cells to CpG-ODN in the presence or absence of anti-IL-10 antibody (Ab). A total of 2.0 × 10⁶ cells were treated with CpG-ODN (1.5 μg/ml) with or without anti-IL-10 antibody (10 μg/ml), non-CpG-ODN with or without anti-IL-10 antibody, medium, or anti-IL-10 alone for 0.5 to 9 h. After each period, the cells were washed and treated with LPS (1 μg/ml) for an additional 1.5 h. The TNF-α level in supernatant was measured. Values represent the means of at least three experiments. Error bars represent the standard error of the mean. ∗, P < 0.05 for CpG versus other groups; †, P < 0.05 CpG plus anti-IL-10 antibody versus other groups.

Figure 6 demonstrates a similar biphasic response in murine peritoneal macrophages, although the time course of subsequent insensitivity to LPS measured by TNF-α release was slightly delayed. After 12 h of preexposure to CpG-ODN, there was significant suppression of LPS-induced TNF-α secretion compared to the secretion after preexposure to medium alone. In addition, preexposure to CpG-ODN plus anti-IL10 antibody for 12 h resulted in augmentation of the LPS-induced TNF-α response, presumably due to the elimination of the IL-10-mediated suppression seen in the CpG-ODN-alone and CpG-ODN–plus–isotype control antibody groups.

FIG. 6.

LPS-induced TNF-α secretion following preexposure of murine peritoneal macrophages to CpG-ODN in the presence or absence of anti-IL-10 antibody. A total of 1.5 × 10⁶ cells were treated with CpG-ODN (1.5 μg/ml) with or without anti-IL-10 antibody (Ab) (10 μg/ml) or isotype control antibody, non-CpG-ODN (1.5 μg/ml) with or without anti-IL-10 antibody (10 μg/ml), or medium alone for 1 to 12 h. After each period, the cells were washed and treated with LPS (1 μg/ml) for an additional 1.5 h. The TNF-α level in supernatant was measured. Values represent the means of at least three experiments. Error bars represent the standard error of the mean. ∗, P < 0.05 for CpG and (CpG plus isotype antibody) versus other groups.

Further investigation was undertaken to investigate the relationship between CpG-ODN preexposure dose and modulation of the macrophage response to subsequent LPS challenge. RAW 264.7 cells (1.5 × 10⁶) were incubated with CpG-ODN at different concentrations for 3 and 9 h in medium washed, and incubated with 1 μg of LPS per ml for 1.5 h. The TNF-α level in supernatant was then immediately measured. The CpG-ODN dose required for both augmentation of LPS-induced TNF-α production following 3 h of CpG-ODN preexposure (Fig. 7a) and suppression of LPS-induced TNF-α production following 9 h of CpG-ODN preexposure (Fig. 7b) was similar to that required for direct CpG-ODN stimulation of TNF-α from RAW 264.7 cells (Fig. 1).

FIG. 7.

Impact of the CpG-ODN dose on modification of LPS-induced TNF-α secretion following CpG-ODN preexposure in RAW 264.7 cells. A total of 1.5 × 10⁶ cells were treated with 0 to 6 μg of CpG-ODN per ml for 3 and 9 h, washed, and then treated with LPS (1 μg/ml) for an additional 1.5 h. The TNF-α level in supernatant was measured. Values represent the means of at least three experiments. Error bars represent the standard error of the mean. (a) Dose-response curve for stimulation of LPS-induced TNF-α secretion in RAW 264.7 cells following a 3-h CpG-ODN preexposure. (b) Dose-response curve for suppression of LPS-induced TNF-α secretion in RAW 264.7 cells following a 9-h CpG-ODN preexposure.

IL-10 inhibits TNF-α mRNA transcription following prolonged CpG-ODN preexposure with subsequent LPS challenge.

To examine the autocrine or paracrine effects of IL-10 on TNF-α mRNA formation following CpG-ODN preexposure and LPS challenge, 2 × 10⁶ RAW 264.7 cells were incubated with CpG or non-CpG-ODN (1.5 μg/ml) in the presence or absence of anti-IL-10 antibody (10 μg/ml) for various periods, washed, and incubated with LPS (1 μg/ml) for an additional 1.5 h. Long-term treatment (6 to 9 h) with CpG-ODN plus anti-IL-10 antibody resulted in a return of TNF-α mRNA levels to the same level as in controls (medium or non-CpG-ODN preexposure followed by LPS exposure), in contrast to suppression of TNF-α mRNA by CpG-ODN preexposure (Fig. 8). Further study revealed no significant difference in TNF-α mRNA levels when CpG-ODN preexposure was compared to CpG-ODN plus isotype control antibody (10 μg/ml) preexposure. In reference to other cytokines, additional experiments demonstrated no difference in LPS-stimulated IL-6 and TGF-β mRNA levels following a 6-h preexposure to either CpG-ODN or non-CpG-ODN (n = 3) (data not shown).

FIG. 8.

TNF-α mRNA expression after preexposure of RAW 264.7 to CpG-ODN in the presence or absence of anti-IL-10 antibody (αIL-10) or isotype control antibody, followed by LPS stimulation. A total of 2.0 × 10⁶ cells were treated with CpG-ODN (1.5 μg/ml), non-CpG-ODN (1.5 μg/ml), or medium (M) with or without anti-IL-10 antibody or isotype control antibody (10 μg/ml) for 0.5 to 9 h. After each period, the cells were washed and treated with LPS (1 μg/ml) for an additional 1.5 h, and then the TNF-α mRNA level was measured using the RNase protection assay. (a) Representative gel from at least three experiments. Each experiment was performed under identical conditions with similar results. Lanes: M, medium; ∗, medium plus anti-IL-10 antibody; †, medium plus isotype control antibody; ‡, medium alone, no LPS stimulation. (B) Graphical depiction of TNF-α mRNA as a function of time, normalized to GAPDH [³²P]mRNA expression. Error bars represent the standard error of the mean. ∗, P < 0.05 for CpG versus other groups.

To rule out the possibility that the diminished LPS response in CpG-ODN-preexposed cells was due to a decreased number of receptors, both major (CD14) and minor (CD11b/CD18) LPS coreceptor expression was assessed after 2-, 3-, and 8-h incubations. There was no significant difference in CD14 expression between medium-, CpG-ODN-, and non-CpG-ODN-exposed cells at 2 h (CD14 mean channel number, 12.0 ± 1.9, 13.7 ± 2.3, and 12.3 ± 2.3 respectively [P > 0.05 comparing all groups]) or 3 h (11.9 ± 2.0, 11.3 ± 1.9, and 11.3 ± 2.0 [P > 0.05]). Interestingly, CD14 expression was increased after 8 h of incubation with 1.5 μg of CpG-ODN per ml compared to medium or non-CpG-ODN incubation (15.3 ± 1.8 versus 10.0 ± 1.2 or 9.4 ± 0.9, respectively [P < 0.05 for CpG-ODN versus either group]). There was no significant difference in CD11b/CD18 expression between groups at 2, 3, and 8 h (data not shown).

DISCUSSION

Although CpG-DNA is known to have a stimulatory effect on the macrophage/monocyte system, there are few in vitro data regarding the alteration of the macrophage response to LPS following CpG-DNA preexposure. This paper demonstrates that in addition to stimulation of TNF-α, CpG-ODN stimulates significant IL-10 production in vitro, leading to a late (6 to 9 h) suppression of LPS-inducible TNF-α secretion via regulation of TNF-α mRNA transcription. Thus, CpG-ODN stimulation of IL-10 may serve a counterregulatory function that results in suppression of macrophage sensitivity to LPS.

Although research has focused primarily on CpG-DNA stimulation of proinflammatory cytokines such as TNF-α, IL-6, and IL-12, a small number of conflicting reports have been published regarding induction of the counterregulatory cytokines such as IL-10 and IL-4 by CpG-DNA. Early studies reported a failure of CpG-ODN to stimulate IL-10 secretion in various cell populations (16, 20), while Anitescu et al have recently demonstrated stimulation of IL-10 secretion in vitro and in vivo by CpG-ODN and have shown that this late secretion of IL-10 is an essential component of a feedback mechanism resulting in the inhibition of IL-12 activity (2). In addition, Huang et al. demonstrated that CpG-DNA stimulation of IL-10 in vitro downregulated the Th1-like cytokine response to heat-killed bacteria or LPS (14). These findings are consistent with our demonstration of a delayed CpG-ODN stimulation of IL-10 acting as an inhibitor of TNF-α secretion following subsequent LPS challenge.

The inhibitory role of IL-10 in other systems is well established. IL-10 is known to work in an autocrine fashion to inhibit cytokines associated with a Th1-like response (IL-12 and IL-2) in favor of a Th2-like response to antigen challenge (9). In addition, IL-10 inhibits monocyte/macrophage TNF-α secretion in vitro (8, 24, 30) and antagonizes LPS-induced TNF-α release in vivo (11). These findings are consistent with our demonstration of inhibition of LPS-induced TNF-α by CpG-ODN-stimulated IL-10. There are, however, several unanswered questions regarding the suppression of TNF-α in response to LPS following prolonged preexposure with CpG-ODN. While the addition of anti-IL-10 antibody during preexposure of cells to CpG-ODN was able to augment TNF-α mRNA back to control levels compared to preexposure to CpG-ODN alone, TNF-α protein secretion was only partially restored at the later times. This suggests that there are other factors activated by CpG-ODN that also play a role in the desensitization of the macrophage to LPS. Such factors could potentially include other anti-inflammatory mediators such as TGF-β or TNF-α-inhibiting factor, both of which are important in other models of monocyte hyporesponsiveness to LPS challenge (4, 24), and unrelated changes in later steps of TNF-α cleavage and secretion. Further studies are also necessary to identify the role of delayed CpG-DNA stimulation of IL-10 in vivo.

There are many similarities between our in vitro results and classical endotoxin tolerance. It is well established that under the proper conditions, preexposure of macrophages in vitro to LPS can induce suppression of the proinflammatory cytokine response to subsequent LPS challenge. Potential cellular mechanisms of endotoxin tolerance include alterations in nuclear factor κB (NF-κB) activity, alterations in mitogen-activated protein kinases and p38 kinase, and alteration in Toll-like receptor 4 (Tlr4) expression (1, 3, 17, 18, 22, 23). We are currently examining these mechanisms to evaluate their importance to the LPS insensitivity induced by CpG-ODN in our system.

Interestingly, the in vivo correlate of endotoxin tolerance is characterized by decreased LPS-induced mortality following in vivo preexposure to LPS; this is in contrast to the increase in LPS-induced mortality seen following in vivo preexposure to CpG-DNA (28). Thus, while there is much overlap between endotoxin tolerance and the effects of CpG-DNA preexposure, there are still significant differences in these systems that preclude firm conclusions regarding their exact relationship.

We demonstrated the biphasic LPS-induced TNF-α response to CpG preexposure in both RAW 264.7 cells and murine peritoneal macrophages, with similar late augmentation of TNF-α on addition of an anti-IL-10 antibody. Despite the similar responses seen between murine cell lines and primary murine macrophages, there remains the potential for important discrepancies between murine and human models of endotoxin responsiveness, which are beyond the scope of this paper. It is now clear, however, that human monocytes respond to CpG-DNA in vitro by secreting TNF-α via a mechanism that is at least temporally distinct from that of LPS-induced TNF-α secretion (12, 13).

CpG-DNA produces a rapid stimulatory effect on macrophages and monocytes, resulting in the release of several proinflammatory cytokines. The relationship between CpG-DNA, bacterial pathogenesis, and the innate immune response, however, remains unclear. Although there is room for optimism regarding the potential clinical benefit of using these oligonucleotides as immunomodulators, particularly for vaccines, caution is necessary to avoid perhaps unforseen, delayed effects on the immune response to subsequent infections. On the other hand, the well-timed addition or deletion of CpG-DNA activity in the actively infected patient might have a broad enough effect to be useful as a new therapeutic modality in the treatment of human sepsis and septic shock.