CpG-A-Induced Monocyte IFN-γ-Inducible Protein-10 Production Is Regulated by Plasmacytoid Dendritic Cell-Derived IFN-α¹

Innate immune activation is regulated by IFNs and chemokines. Chemokines cause changes in the shape and behavior of the specific cell subpopulations expressing their receptors, leading to chemotaxis (1) toward sites of infection (2) or injury (3, 4) and leukocyte activation (5). Together with the differential expression of chemokine receptors, chemokines may determine Th1 vs Th2 response (6). One chemokine that is important in Th1 reactions is IFN-γ-inducible protein-10 (IP-10)³ (7), a non-ELR CXC chemokine that binds to the receptor CXCR3 expressed on activated T cells and NK cells (8). IP-10, initially identified as an IFN-γ-inducible protein, is produced by a variety of cell types and is known to be inducible by both type I and type II IFNs (9, 10) and CpG DNA (11). It is expressed in response to, and has been shown to be necessary in protection against, a variety of viral and bacterial infections (12, 13, 14, 15, 16, 17). IP-10, like other non-ELR CXC chemokines, is a potent inhibitor of angiogenesis (18) and has been demonstrated to be essential in IL-12-mediated anti-tumor activity (19, 20).

Toll-like receptors (TLR) provide innate immunity with a way to detect pathogens. TLR9, a TLR involved in CpG recognition (21), is expressed in human B cells and plasmacytoid dendritic cells (pDC) (22). CpG dinucleotides are present at a lower than expected frequency and are selectively methylated in vertebrate DNA, but are present at approximately the expected random frequency and unmethylated in bacterial DNA (23). CpG motifs are recognized as danger signals and activate a wide variety of immune responses including innate, humoral, and cellular immunity (24, 25). Recently two different classes of oligodeoxynucleotides (ODN) containing CpG motifs that stimulate human cells have been described. CpG-A ODN stimulate pDC to make large amounts of IFN-α, while CpG-B ODN only weakly induce IFN-α production, despite promoting DC survival and maturation (26, 27, 28, 29, 30, 31). In contrast, CpG-B ODN strongly activate human B cells to proliferate and to secrete IL-10 but CpG-A ODN have little B cell effect (26, 27, 28, 29, 30, 31). Both CpG-A and CpG-B ODN require TLR9, because mice genetically deficient in it show no immune effects from either class (Ref. 21 and P. Payette, H. Davis, and A. M. Krieg, unpublished data). Although TLR9 thus appears to be required for the effects of both classes of CpG ODN, it is only fully sufficient for mediating the activity of the CpG-B ODN: we and others have found that human embryonic kidney cells transfected to express TLR9 with an NF-κB reporter construct showed a strong response to a CpG-B ODN (termed “K” ODN by Klinman and colleagues), but minimal response to a CpG-A ODN (termed “D” ODN by Klinman et al.) (32). Thus, the data suggest that TLR9 is required for immune stimulation in response to both ODN classes, but that TLR9 is only sufficient to reconstitute strong activation of NF-κB by CpG-B ODN, not by CpG-A ODN.

In murine macrophages, CpG DNA has been shown to induce β-chemokine protein production (33) and to up-regulate mRNAs encoding the chemokines macrophage inflammatory protein-1α and -1β, RANTES, monocyte chemoattractant protein-1, and IP-10 (34), while in pDC, CpG ODN stimulate IL-8, IL-12 (in synergy with anti-CD40), and IP-10 production (11). Thus, the overall effect of CpG ODN is to create a strong Th1-like cytokine and chemokine milieu. However, the regulation of CpG-induced chemokine expression has not been investigated previously.

In contrast to the clear responsiveness of murine monocytes and macrophages to CpG DNA, there has been some controversy over whether human monocytes are directly stimulated. Some investigators have reported that human monocytes neither express TLR9 nor respond to CpG ODN (22, 35). However, Klinman and colleagues (32, 36) have recently reported that human monocytes both express TLR9 and respond directly to CpG ODN stimulation.

In this study, we investigated CpG-dependent IP-10 production. CpG-A ODN were found to be potent inducers of IP-10 secretion, effecting pDC directly, while in monocytes the effect was indirect and mediated by pDC-derived IFN-α. We found no evidence for direct stimulation of highly purified human monocytes by any CpG ODN. Neither T cells nor B cells produced significant amounts of IP-10 in response to CpG-A ODN. IFN-α alone was able to induce the secretion of small amounts of IP-10 in PBMCs and was found to synergize with low concentrations of IFN-γ.

ODN were provided by Coley Pharmaceutical Group (Wellesley, MA). Sequences are shown 5′-3′; lower case letters, phosphorothioate linkage; upper case letters, phosphodiester linkage 3′ of base: ODN 2006: tcgtcgttttgtcgttttgtcgtT; ODN 2137: tgctgcttttgtgcttttgtgctT; ODN 2216: ggGGGACGATCGTCgggggG; ODN 2243: ggGGGAGCATGCTGgggggG. ODN were diluted in TE buffer (10 mM Tris, pH 8.0, 1 mM EDTA) and stored at 4°C. Dilutions of original stocks were made in PBS (Life Technologies, Grand Island, NY) and stored at 4°C until use.

PBMCs were isolated from healthy donors from either whole blood or buffy coats (Elmer L. DeGowin Blood Center, University of Iowa, Iowa City, IA) by Ficoll-Hypaque density gradient (Lymphoprep; Invitrogen, Carlsbad, CA) as previously described (11). Cells were cultured in RPMI 1640 supplemented with 10% heat-inactivated (56°C, 1 h) FCS, 1.5 mM l-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, 0.1 mM nonessential amino acids, and 1 mM sodium pyruvate (all from Life Technologies, Grand Island, NY) (complete medium) in a 5% CO₂ humidified incubator at 37°C in a 96-well plate (200 μl/well) unless specified otherwise.

pDC (CD123⁺⁺/HLA-DR⁺/lin⁻ cells) were isolated from PBMCs with a VarioMACS (Miltenyi Biotec, Auburn, CA) according to the manufacturer’s protocol. Briefly, PBMCs were isolated from a buffy coat as described above and washed one time with MACS buffer (PBS supplemented with 0.5% BSA and 2 mM EDTA). PBMCs were then resuspended in MACS buffer (300 μl/10⁸ PBMCs) and incubated with FcR blocking reagent (100 μl/10⁸ cells) and anti-BDCA-4 Ab conjugated to magnetic microbeads (100 μl/10⁸ cells) for 15 min at 4°C, washed, and passed over a positive selection column in a magnetic field. The column was washed three times with MACS buffer and the pDC were collected by plunging the MACS buffer through the column. The pDC were then passed over a second positive selection column using the same method to increase the purity of the resulting pDC. The resulting purified pDC were cultured at 4.4 × 10⁵/ml in complete medium supplemented with 10 ng/ml IL-3 (R&D Systems, Minneapolis, MN) in a 96-well plate. pDC purity was >90%. The results shown in this manuscript are representative of >50 experiments using purified primary pDCs, in which various CpG ODN were found to act directly on the pDC using a range of different assays.

Monocytes were also isolated from PBMCs with a VarioMACS according to the manufacturer’s protocol. Briefly, PBMCs were incubated with a mixture of hapten-conjugated anti-CD3, -CD7 -CD19, -CD45RA, -CD56, and -IgE Abs and an FcR blocking reagent. The cells were then washed, incubated with anti-hapten microbeads, and passed over a depletion column to deplete T cells, NK cells, B cells, some DCs, and basophils. Cells in the flow-through, mainly monocytes, were then incubated with microbead-conjugated anti-bromodichloroacetate-4 Ab and passed over a depletion column to remove any remaining pDC. Cells were cultured at 5 × 10⁵/ml in complete medium in a 96-well plate. The purity of monocytes after depletion of pDC was over 97.5%. The results shown in this manuscript are representative of >20 experiments using purified primary monocytes, showing a failure of CpG ODN to activate directly.

Surface Ag staining was performed as previously described (26) and live cells were gated on based on morphology (side light scatter vs forward light scatter). mAbs against lineage markers (FITC-conjugated lineage mixture consisting of anti-CD3, -CD14, -CD16, -CD19, -CD20, -CD56) were purchased from BD Biosciences (San Diego, CA). Anti-CD3-FITC, anti-CD14-FITC, and anti-CD19-FITC were purchased from BD PharMingen (San Diego, CA). Flow cytometric data were acquired on a FACScan (BD Immunocytometry Systems, San Jose, CA). Data were analyzed using the computer program FlowJo (version 2.7.4; Tree Star, Stanford, CA).

For IL-6 staining, cells were treated for 12 h, after which PMA (50 ng/ml) and calcium ionophore (500 ng/ml) (both from Sigma-Aldrich, St. Louis, MO) were added. After 3 h further incubation, brefeldin A (1 μg/ml) (Sigma-Aldrich, St. Louis, MO) was added to block cytokine secretion. This was followed by another 3 h of incubation. For IP-10 staining, no PMA or calcium ionophore was used. Brefeldin A was added after 12 h of treatment, followed by another 6-h incubation. For both IL-6 and IP-10, the cells were then harvested and incubated with 5 μl/well of either anti-CD19-FITC or anti-CD14-FITC Ab on ice in the dark for 15 min. Cells were washed with PBS and resuspended in 100 μl of medium A (Fix and Perm kit; Caltag Laboratories, Burlingame, CA), incubated at room temperature in the dark for 15 min and washed with PBS. Cells were then resuspended in 100 μl of medium B (Fix and Perm kit) and 8 μl of either anti-IP-10-PE or anti-IL-6-PE (BD PharMingen) were added, followed by a 15-min room temperature incubation in the dark. Cells were washed and examined by flow cytometry.

Cells were treated with ODN for 44 h, and supernatants were collected and stored at −20°C until use. Capture and detection Abs and IP-10 standard were purchased from BD PharMingen. Capture Ab (catalog no. 23161D) was diluted to 0.5 μg/ml in PBS, added to 96-well Nunc-Immunoplates at 100 μl/well, and the sealed plates were incubated overnight at 4°C. The following day capture Ab was flicked out of plates and wells were blocked with 5% nonfat milk in PBS at 37°C for 1 h followed by four washes with PBS supplemented with 0.05% Tween 20 (TPBS; Fisher Biotec, Fair Lawn, NJ). Dried plates were sealed and stored at −20°C until use. IP-10 standard (catalog no. 24541V) was diluted to 5 μg/ml in 1% BSA in PBS and stored in aliquots at −80°C until use. When performing ELISAs, plates were thawed and standards were diluted in 10% FCS in PBS “buffer.” One hundred microliters of standards or samples (cell supernatants) were added to wells and incubated overnight. The next day, plates were washed five times with TPBS, after which 100 μl of biotinylated detection Ab (catalog no. 555048) (diluted to 0.5 μg/ml in buffer) was added, followed by a 45-min room temperature incubation and five washes with TPBS. Avidin-peroxidase (Sigma-Aldrich) (previously diluted to 1 mg/ml in PBS and stored in aliquots at −20°C) was diluted in buffer to 2.5 μg/ml and 100 μl/well was added to plates, followed by a 30-min room temperature incubation. Plates were then washed as before and 100 μl/well room temperature tetramethylbenzidine substrate (catalog no. T8665; Sigma-Aldrich) was added and incubated until color developed (∼15–20 min), after which, 100 μl/well 0.67N H₂SO₄ were added to stop the reaction. Plates were read at 450 nm with a Molecular Devices Kinetic microplate reader and analyzed with SOFTmax computer software (Molecular Devices, Sunnyvale, CA).

IFN-α ELISAs were performed using a kit from Pestka Biomedical Laboratories (New Brunswick, NJ) as previously described (26). Briefly, 100-μl standards were added to wells and incubated for 1 h (all incubations were at room temperature) after which plates were washed two times with wash buffer and 100 μl of Ab solution was added. Following another 1-h incubation, plates were washed three times and 100 μl of HRP conjugate was added and plates were incubated for 1 more hour followed by four washes and the addition of 100-μl room temperature tetramethylbenzidine substrate. After another 1-h incubation, 100 μl of stop solution was added and plates were read at 450 nm in the same plate reader as above.

Whole PBMCs at 1 × 10⁶/ml were incubated with ODN 2216, ODN 2243 (at 1 μg/ml), or no ODN in complete medium for 44 h. Supernatants were carefully removed, avoiding cells, and spun in a Labnet Spectrafuge in 1.5-ml tubes at 5000 rpm for 30 s to spin down any remaining cells. The top portion of the supernatants was transferred to fresh Eppendorf tubes and either used right away or stored at −20°C before use. When ready to use, cell samples were resuspended in 50 μl of complete medium and 150 μl of conditioned medium was added to this for incubation.

Two different classes of stimulatory CpG ODN and their GpC controls were tested for IP-10 induction in whole PBMCs. ODN 2006, a phosphorothioate-modified ODN with three CpG motifs, is representative of class B. ODN 2216, which has a chimeric backbone (diester in the middle and thioate on the ends) and a palindromic sequence containing three CpG motifs, flanked by poly(G) sequences at the 5′ and 3′ ends, is representative of class A. PBMCs were incubated for 48 h with ODNs in varying concentrations and their supernatants were assayed by ELISA for IP-10 (Fig. 1). ODN 2216 caused IP-10 induction at all concentrations used, the highest being 3450 pg/ml IP-10 with 2 μg/ml ODN added, while its GpC control 2243 caused no detectable levels at any of these concentrations. ODN 2006 induced no detectable IP-10 until the ODN concentration reached 0.25 μg/ml and then stayed about the same for the three highest ODN concentrations at ∼800 pg/ml IP-10, while its GpC control 2137 induced no significant production of IP-10.

To ascertain what cell type(s) in peripheral blood were producing IP-10 in response to CpG, whole PBMCs were incubated for 18 h with these same ODN. Four different cell types were identified using the surface markers lin 1 (lineage-negative cells, Fig. 2,A), CD3 (T cells, Fig. 2,B), CD14 (monocytes, Fig. 2, C and E), or CD19 (B cells, Fig. 2 D). The cells were then stained intracellularly for IP-10 and analyzed by flow cytometry. In response to ODN 2216, IP-10 levels were increased by ∼8-fold in monocytes and 4-fold in the lineage-negative population but did not change in B cells or T cells.

It has been demonstrated that TLR9 is responsible for the recognition of CpG DNA by the vertebrate immune system (21). Human B cells and pDC express high levels of TLR9 while monocytes do not (22), leading us to hypothesize that the effect on monocytes may be indirect. Therefore, we tested whether isolated monocytes would produce IP-10 in response to CpG. Furthermore, because the preceding experiment indicated that some lineage-negative cell type(s) were also making IP-10 and because it is known that pDC can respond directly to CpG, we hypothesized that CpG-stimulated isolated pDC may produce IP-10. To test these theories, monocytes and pDC were isolated from PBMCs by magnetic cell sorting and incubated for 18 h with ODN and then stained intracellularly for IP-10. As expected, isolated monocytes did not produce IP-10 in response to CpG (Fig. 3,A) while both ODNs 2006 and 2216 directly stimulated pDC to produce IP-10 (Fig. 3 B).

The preceding experiment demonstrated that CpG-induced IP-10 production in monocytes is an indirect effect. To determine whether this requires cell-to-cell contact with some other cell type within whole PBMCs or if conditioned medium alone would be sufficient, both whole PBMCs and isolated monocytes were incubated with either conditioned medium or ODN for 18 h. Whole PBMCs were surface stained for CD14, after which these and the isolated monocytes were stained intracellularly for IP-10 and analyzed by flow cytometry. Results are reported both as IP-10 mean fluorescence intensity (MFI) and the percent of cells staining positive for IP-10. The CD14⁺ cells within whole PBMCs responded to both conditioned medium and CpG DNA while isolated monocytes produced IP-10 only in response to conditioned medium, indicating that cell-to-cell contact is not required for CpG-induced IP-10 in monocytes (Fig. 4).

In the last experiment it was shown that some substance(s) in the conditioned medium caused IP-10 secretion in isolated monocytes. IP-10 has previously been demonstrated to be inducible by IFN-α (11) and class A ODN have been shown to induce IFN-α in PBMCs. Therefore, we examined the supernatants that contained the highest amounts of IP-10 (from Fig. 1) for IFN-α by ELISA. As expected, PBMCs treated with ODN 2216 secreted high levels of IFN-α (∼7500 pg/ml), while those treated with ODN 2006 secreted very little (∼50 pg/ml) and PBMCs treated with control ODN secreted no detectable IFN-α (Fig. 5).

Based on the results of the last two experiments we hypothesized that 2216 was causing the pDC to make IFN-α and that this IFN-α was then up-regulating IP-10 expression in monocytes. To address this, whole PBMCs were treated with ODN 2216 and varying amounts of the IFN-α neutralizing Abs MMHA-2 (anti-IFN-α) or anti-CD118 (anti-IFN-αβ receptor). CpG-induced IP-10 expression in the CD14⁺ population was abrogated with both Abs in a dose-dependent manner (Fig. 6, A and B). The Ab treatment specifically lowered IP-10 production, as CpG-induced IL-6 production in the CD19⁺ population was not effected (Fig. 6 C).

The preceding experiment demonstrated that IFN-α is essential for CpG-induced IP-10 secretion in monocytes. To determine whether IFN-α alone is sufficient to induce high levels of IP-10, whole PBMCs were plated with varying amounts of IFN-α−2b. After 2 days the level of IP-10 in the supernatants was determined by ELISA. IFN-α−2b alone did induce some IP-10 production, but not at the high levels seen following CpG treatment (Fig. 7).

IP-10 is highly inducible by IFN-γ. Whole PBMCs were treated with a low concentration of IFN-γ and varying amounts of IFN-α-2b for 40 h, followed by an IP-10 ELISA on the supernatants. The IFN-α-2b did show synergy with IFN-γ for IP-10 production under these conditions (Fig. 8), however, when using higher amounts of IFN-γ we did not see this same effect (data not shown).

CpG-DNA motifs are recognized by the innate immune system through TLR9 as nonself and can activate a wide variety of immune responses (for review, see Refs. 25 , 37 , and38). Chemokines play an important role in the direction of immune activation, with Th1 reactions being dependent on the CXC chemokines IP-10, T cell-α chemoattractant, and monokine induced by IFN-γ (6). In this study, we define an immunoregulatory circuit in PBMCs leading from CpG-A ODN stimulation to IFN-α secretion by pDCs to IP-10 secretion by monocytes. Unexpectedly, IFN-γ seems to play a minimal role in CpG-A-induced IP-10 while IFN-α is essential.

The CpG-A ODN 2216 induced the production of large amounts of IP-10 in PBMCs (at 2 × 10⁶/ml) in a dose-dependent manner with IP-10 increasing steadily from an ODN concentration of 0.05–2 μg/ml. Compared with CpG-A 2216, the CpG-B ODN 2006 induced <25% as much IP-10, with detectable amounts not being reached until an ODN concentration of 0.25 μg/ml, which then stayed level from ODN concentrations of 0.5–2 μg/ml. When this same assay was performed with PBMCs at 1 × 10⁶/ml, ODN 2006 induced little to no IP-10, while ODN 2216 still did induce IP-10 (data not shown) suggesting that the two different classes of ODN exhibit differential dependence on cell concentration. After looking at different cell populations within whole PBMCs, it was determined that monocytes produce the major portion of IP-10 in response to CpG stimulation, that the lineage negative cell population produces ∼one-third as much, and that neither T cells nor B cells produce significant amounts of IP-10 under these experimental conditions.

Some investigators have reported that human monocytes express very low levels of TLR9 and are insensitive to CpG-mediated up-regulation of CD80, CD86, and HLA-DR in the absence of pDC (22, 35), but another group has reported the opposite finding (32, 36). Although we did not examine monocyte expression of TLR9, our data from >10 different normal donors establish that highly purified human monocytes do not produce IP-10 in response to CpG ODN, nor show any other response (Fig. 3 and data not shown). In the course of our studies we found that extreme measures are required to avoid pDC contamination of monocyte preps, because failure to perform complete pDC depletions from PBMC often led to apparent direct CpG responses from a “monocyte” population which appeared to be 99% pure. The reports of direct monocyte activation by CpG ODN used elutriated monocytes, which were only 99% pure (32, 36). Because the normal pDC level in human PBMC is only ∼0.1%, great care appears to be required for such studies of intercellular interactions.

In contrast with the lack of CpG response from isolated monocytes, isolated pDC are able to respond directly to CpG ODN by producing IP-10. Within whole PBMCs, the myeloid DCs (lineage-negative, HLA-DR⁺, CD11c⁺ cells) produce more IP-10 in response to class A CpG ODN than the pDC (data not shown), but their lack of TLR9 expression and similarity to monocytes suggests that this effect may also be indirect.

To determine the mechanism of CpG-induced IP-10 production by monocytes, we tested whether cell-to-cell contact with some other cell type was required. When whole PBMCs were incubated with either CpG ODN or supernatants from PBMCs that were treated with CpG ODN, they responded to both equally. Although it appears as though conditioned medium is a little less effective, this is misleading because only three-fourths of the medium used was conditioned, as one-fourth was the regular medium in which the cells were originally resuspended. However, when isolated monocytes were treated in like manner they only responded to the conditioned medium, not the ODN, suggesting that: 1) no contaminating cells that are involved in CpG-induced IP-10 were present, 2) something secreted into the medium by another cell type is responsible for the effect, and 3) no cell-to-cell contact with any other cell type is required for monocytes to produce IP-10.

Previous studies indicate that both IFN-α and IFN-γ can induce IP-10 production, but we have been able to detect only very low levels of IFN-γ in CpG-treated PBMCs under the experimental conditions used in this study (data not shown). Because class A CpG ODN are known to induce IFN-α in pDC (26), we tested the supernatants from the PBMCs treated with the highest amounts of 2216 and 2006 for IFN-α. At these ODN concentrations, PBMCs stimulated with 2216 secreted ∼4-fold more IP-10 than those stimulated with 2006, but 150-fold more IFN-α. To ascertain whether the high levels of IFN-α induced by ODN 2216 are required for IP-10 production in monocytes, whole PBMCs were treated with ODN 2216 and varying amounts of neutralizing Abs to IFN-α. Abs to IFN-α or to IFN-αβ receptor both abrogate the ability of CpG ODN to induce IP-10 in the CD14⁺ population of whole PBMCs but have no significant effect on CpG-induced IL-6 production in the CD19⁺ population. The ability of CpG A to induce IP-10 in monocytes is clearly mediated by IFN-α.

The regulation of CpG-induced IP-10 production appears to be complex, with multiple sources including pDCs, myeloid DCs, monocytes, and possibly other cell types. CpG-A ODN induce the production of high levels of IP-10 in a dose-dependent manner, along with high levels of IFN-α. CpG-B ODN induce less IP-10 production, with levels reaching a plateau very quickly, along with very low levels of IFN-α. Although CpG-B ODN can induce IP-10 directly in pDC, they do not seem to have the ability to induce the more potent secondary response in monocytes. Taken together, these data suggest that there may be a different mechanism of action between CpG-A and CpG-B-induced IP-10 production. Although IFN-α is clearly necessary for CpG-A-induced IP-10 production on monocytes, it does not appear to be completely sufficient. PBMCs stimulated with up to 1000 U/ml IFN-α-2b secreted only approximately one-tenth as much IP-10 as seen with activation by 2 μg/ml 2216. This ODN has been shown to induce a small amount of IFN-γ production in PBMCs (26), and we demonstrate in this study that IFN-α will synergize with a small amount of IFN-γ for inducing IP-10 secretion. However, it remains possible that other factors, as yet unknown, are also involved in mediating the IP-10 expression in CpG-stimulated monocytes.

Because of the role of IP-10 in coordinating anti-infective responses and mediating anti-tumor activity, these findings could provide insight into the therapeutic applications for CpG DNA. CpG DNA is presently in human clinical trials and has been shown to be useful as an adjuvant in vaccines (38). It has been shown to be effective in a variety of mouse tumor models, as well as for immunotherapy of allergy and infectious disease (25, 39). It seems likely that CpG-induced IP-10 production contributes to its therapeutic activities.

We thank Sharon Setterquist for technical assistance and Jörg Vollmer and Gunther Hartmann for helpful discussions.