CpG DNA Activates Survival in Murine Macrophages through TLR9 and the Phosphatidylinositol 3-Kinase-Akt Pathway¹

Cytosine guanine dinucleotide DNA potently activates mouse macrophages, dendritic cells (DC),³ and B cells (1) and promotes the development of a Th1-type acquired immune response (2). As well as triggering production of proinflammatory cytokines and chemokines and up-regulating expression of cell surface costimulatory molecules, CpG DNA also triggers survival of macrophages (3), DC (4), neutrophils (5), and B cells (6). In macrophages, activation of survival pathways by CpG DNA is likely to counteract the effects of proapoptotic mediators such as NO and TNF-α that are also produced in response to CpG DNA (7, 8), thereby allowing macrophages to coordinate an appropriate inflammatory response to pathogenic challenge. However, excessive macrophage survival during chronic inflammatory diseases such as rheumatoid arthritis is likely to contribute to disease pathology (9).

The balance between cellular apoptosis and survival is primarily controlled by members of the Bcl family of proteins that localize to the mitochondrial outer-membrane and act to control mitochondrial permeability (10, 11). When the ratio of prosurvival to proapoptotic Bcl family members is biased toward the latter, the outer mitochondrial membrane becomes permeable and allows the release of factors such as cytochrome c that initiate the apoptotic process in the cytosol (12). The PI3K-Akt and Ras-Raf-MEK-ERK pathways are important regulators of cellular survival and function by activating transcription of antiapoptotic genes including Bcl family members and by phosphorylating and inactivating proapoptotic proteins (13, 14).

Macrophage survival depends upon the growth factor, CSF-1 (CSF-1/M-CSF), which activates both the Ras-Raf-MEK-ERK (15, 16) and PI3K-Akt (17) pathways. In the absence of CSF-1, bacterial products such as LPS and CpG DNA can permit cellular survival and can activate the MEK-ERK (3) and PI3K-Akt pathways (18, 19, 20). The pathways leading to MEK-ERK activation by LPS and CSF-1 may be distinct. The involvement of the upstream Ras-Raf components in MEK-ERK activation by LPS is controversial (15, 21, 22, 23, 24), whereas the alternative MAP-3-kinase, Cot/Tpl-2 was required for LPS-induced MEK-ERK activation (25, 26). The pathways leading to PI3K-Akt activation by LPS and CpG DNA still remain poorly defined. In macrophages, both the MEK-ERK and PI3K-Akt pathways target the transcription factor Ets-2 (16, 17) that in turn induces expression of the antiapoptotic gene, Bcl-x_L (27).

Macrophage activation in response to bacterial products is primarily mediated by members of the TLR family that signal via the MyD88-IRAK-TRAF6 pathway to regulate proinflammatory responses (28). Responses to CpG DNA are dependent on TLR9 (29), an intracellular receptor that traffics to an endosomal compartment upon treatment with CpG DNA (30). In vitro, TLR9 directly interacts with single-stranded CpG DNA at a pH approximating that of the endosomal environment (31). Although inflammatory cytokine production is clearly dependent on TLR9, recent data have indicated that other CpG DNA signaling pathways could exist. Chu et al. (32) reported that responses to CpG DNA were also dependent on the serine threonine kinase, DNA-dependent protein kinase (DNA-PK) and argued that a direct interaction occurs between DNA-PK and CpG DNA. That dependence on DNA-PK for CpG DNA-induced responses was not reproduced in subsequent studies (33, 34). More recently, Chu and colleagues (20) showed that a CpG-containing phosphorothioate oligodeoxynucleotide (PS-ODN) activated Akt in macrophages in a DNA-PK-dependent and a TLR9-independent manner, and that DNA-PK directly interacted with Akt. Because of the potential involvement of TLR9, DNA-PK, MEK-ERK, and PI3K-Akt in CpG DNA-induced survival, in this study we set out to clearly define the pathways necessary for survival of macrophages in response to this stimulus.

RPMI 1640 medium (Invitrogen Life Technologies) supplemented with 10% FCS (JRH Biosciences), 20 U/ml penicillin (Invitrogen Life Technologies), 20 μg/ml streptomycin (Invitrogen Life Technologies), and 2 mM l-glutamine (Invitrogen Life Technologies) (complete medium) was used for culture of bone marrow-derived macrophages (BMM). BMM were prepared from femurs of CD1, C57BL/6, or TLR9^−/− mice in accordance with local animal ethics guidelines. Briefly, bone marrow cells were cultured for 7 days in complete medium in the presence of 10,000 U/ml CSF-1 (a gift from Chiron) on bacteriological plastic plates. Bone marrow from Prkdcscid CB-17 mice was a gift from the laboratory of Prof. M. Lavin (Queensland Institute of Medical Research, Brisbane, Australia). All primary cells were maintained in a 37°C incubator containing 5% CO₂. Re595 LPS from Salmonella minnesota (Sigma-Aldrich) was sonicated in 0.1% triethylamine and then further diluted in PBS. The synthetic tripalmitoylated lipopeptide, Pam₃CysSerLys₄ (Roche), was used at a final concentration of 10 ng/ml in cell culture. All oligonucleotides used in this study were from Geneworks. Phosphodiester oligodeoxynucleotides (PO-ODN) used were activating oligonucleotide-1 (AO-1) (5′-GCTCATGACGTTCCTGATGCTG-3′), nonactivating oligonucleotide (NAO-1) (5′-GCTCATGAGCTTCCTGATGCTG-3′), 1668 (5′-TCCATGACGTTCCTGATGCT-3′), and 1668-GC (5′-TCCATGAGCTTCCTGATGCT-3′). PS-ODN contained modifications at every linkage and are indicated in bold. PS-ODN used were 1668S (5′-TCCATGACGTTCCTGATGCT-3′), 1668S-GC (5′-TCCATGAGCTTCCTGATGCT-3′), 1018S (5′-TGACTGTGAACGTTCGAGATGA-3′), and 1018S-GC (5′-TGACTGTGAAGCTTGCAGATGA-3′). Escherichia coli genomic DNA (EC DNA) and calf thymus genomic DNA (CT DNA) (Sigma-Aldrich) was extracted with Triton X-114 followed by phenol-chloroform extraction and ethanol precipitation to remove contaminating LPS. LY294002 and PD98059 (Calbiochem; EMD Biosciences) were dissolved in DMSO and stored in aliquots at −80°C. Chloroquine (Sigma-Aldrich) was dissolved in PBS at 100 mg/ml, stored at 4°C, and used at a final concentration of 2.5 μg/ml.

The reduction of the dye MTT (Sigma-Aldrich) was used to measure relative number of viable cells. BMM (100,000 cells/well in 200 μl in a 96-well plate) were harvested and plated in CSF-1-free medium for 18 h. The desired additions were made, and the cells were incubated for an additional 48 h before assessment of MTT reduction, as described previously (3).

On day 6, BMM (2 × 10⁶) were washed twice with calcium/magnesium-free PBS and plated on 60-mm bacteriological plates. BMM were treated as indicated, harvested, and processed as outlined by Vadiveloo et al. (35). Briefly, cells were resuspended thoroughly in 50 μl of PBS, and 1 ml of ice-cold 70% ethanol was added. Cells were left overnight at 4°C and, on the following day, were pelleted, washed with PBS, and resuspended in 0.5 ml of a staining solution containing 69 μM propidium iodide and 5 μg/ml RNase A in 38 mM sodium citrate. Stained cells were analyzed using a FACSCalibur (BD Biosciences) flow cytometer. All data analysis was conducted using CellQuest (version 3.1) (BD Biosciences).

Whole-cell extracts were prepared as described previously (3) except all solutions containing 1 mM sodium vanadate were also supplemented with 1 mM sodium pyrophosphate/1 mM sodium molybdate/10 mM sodium fluoride.

BMM (2 × 10⁶) were washed with PBS and plated in 2 ml of complete medium in 60-mm bacteriological plastic dishes. Adherent and nonadherent cells were harvested and washed once with ice-cold PBS, resuspended in 20 μl of fresh ice-cold lysis buffer (50 mM PIPES/KOH (pH 6.5), 2 mM EDTA, 0.1% CHAPS, 20 mg/ml leupeptin, 10 mg/ml pepstatin, 10 mg/ml aprotinin, 1× Complete protease inhibitors (EDTA-free) (Roche), 5 mM dithioerythritol, and 1 mM PMSF). Cells were then lysed by repeated (three times) freeze-thawing on ice and a dry-ice/ethanol bath.

Proteins were resolved by SDS-PAGE (8–12%), and gels were transferred to methanol-activated Immobilon-P polyvinylidene difluoride membranes (Millipore). Blots were blocked and probed with Abs recognizing the phosphorylated forms of the MAPK family members, ERK1/2 (Thr²⁰²/Tyr²⁰⁴), and Akt (Ser⁴⁷³) (Cell Signaling Technologies). Anti-mouse and rabbit IgG conjugated to HRP were purchased from Cell Signaling Technologies. The Ab against cleaved caspase-3 was purchased from Cell Signaling Technologies. Anti-mouse and rabbit IgG conjugated to HRP were purchased from either New England Biolabs or Cell Signaling Technologies. For reprobing of membranes with Abs, membranes were stripped with 63 mM Tris-Cl (pH 6.7)/2% SDS/100 mM 2-ME for 30 min at 50°C with agitation. Membranes were then washed three times with TBST for 30 min, and then blocked and probed as above except phosphatase inhibitors were omitted.

TNF-α assays were conducted in the presence of 5 μg/ml brefeldin A to block secretion of TNF-α. Adherent and nonadherent cells were harvested with 5 ml of ice-cold PBS/1 mM EDTA/0.1% NaN₃. Cells were pelleted and resuspended in 50 μl of 2.4G2 hybridoma supernatant (rat anti-FcR mAb). After 10 min on ice, 250 μl of 1% formaldehyde solution was added and incubated for a further 5 min. This was followed by addition of 1 ml of PBS/1 mM EDTA/0.1% NaN₃/0.2% heat-inactivated goat serum/0.1% BSA. Cells were then pelleted and resuspended in 50 μl of PBS/0.1% NaN₃/0.2% heat-inactivated goat serum/0.1% BSA containing a 1/250 dilution of anti-TNF-α PE-conjugated mAb (BD Pharmingen) and incubated on ice for 1 h. Cells were washed with addition of 1 ml of PBS/1 mM EDTA/0.1% NaN₃/0.2% heat-inactivated goat serum/0.1% BSA, pelleted, and resuspended in 200 μl of PBS/1 mM EDTA/0.1% NaN₃/0.2% heat-inactivated goat serum/0.1% BSA. Cells were then analyzed by flow cytometry using a FACSCalibur (BD Biosciences).

Dragoi et al. (20) reported that DNA-PK, but not TLR9, was required for Akt phosphorylation and activation in BMM in response to a CpG-containing PS-ODN. Because the PI3K-Akt pathway is an important mediator of cell survival (13), these data imply that CpG DNA induces cellular survival independently of TLR9 in BMM. We therefore assessed the involvement of TLR9 in CpG DNA-induced survival in this cell type. The CpG-containing PS-ODN used by Dragoi et al., 1018S (Fig. 1,A), as well as another commonly used CpG-containing PS-ODN, 1668S (Fig. 1 B), permitted survival of BMM as assessed by the MTT cell viability assay. The corresponding control PS-ODN with a GpC inversion (1018S-GC and 1668S-GC) did not enhance survival, thus indicating that the survival effect was mediated by the CpG-containing immunostimulatory motif. 1018S and 1668S, at doses up to the maximally stimulating dose of 0.1 μM, did not enhance survival of BMM from TLR9-deficient mice. Hence, CpG DNA-mediated survival was dependent on TLR9.

Although the effects of PS-ODN on survival are relevant to therapeutic applications of CpG DNA, they are not relevant to responses to pathogenic challenge, where natural phosphodiester DNA will be encountered. Indeed, the requirement for different immunostimulatory sequences between human and mouse TLR9 (36) applies only to PS-ODN and not PO-ODN (37). We therefore assessed the involvement of TLR9 in survival triggered by the CpG-containing PO-ODN, 1668. As with 1668S, 1668 permitted BMM survival and this effect was not apparent in cells treated with the GpC-inverted control PO-ODN, 1668-GC (Fig. 1,C). 1668 did not trigger survival of TLR9-deficient BMM, even at doses up to 5 μM. A similar outcome was observed when the CpG-containing phosphodiester ODN AO-1 and its GpC control, NAO-1, were used (Fig. 1,D). Fig. 1,E shows that survival pathways could still be appropriately activated in CSF-1-starved TLR9-deficient BMM, because LPS, CSF-1, and bacterial lipoprotein all induced survival in both wild-type and TLR9^−/− BMM. Hence, CpG DNA-induced survival of BMM is entirely dependent on TLR9. We also confirmed that chloroquine, which inhibits responses to CpG DNA by preventing endosomal acidification, blocked survival permitted by the PO-ODN 1668, but not LPS (Fig. 2).

CpG DNA-permitted survival could be mediated by the MEK-ERK and/or PI3K-Akt pathways. We therefore investigated the requirement for TLR9 in the activation of these pathways. Phosphorylation of Ser⁴⁷³ is required for full activation of Akt (13). Fig. 3,A shows that treatment with 0.1 μM 1668S, that was sufficient to trigger maximal survival of BMM (Fig. 1,B), induced phosphorylation of Akt at Ser⁴⁷³ in wild-type, but not TLR9-deficient, BMM. In contrast, LPS triggered this event in both wild-type and TLR9-deficient BMM. As expected, ERK1/2 phosphorylation in response to a CpG-containing PS-ODN was also TLR9 dependent (Fig. 3,B). The TLR9 dependency of Akt phosphorylation was also observed with the CpG-containing phosphodiester ODNs 1668 and AO-1 (Fig. 3 C).

TLR9-dependent Akt activation (Fig. 3) contrasts with previously published data (20). In that study, TLR9-independent activation of Akt in response to a high dose (∼1.5 μM) of 1018S was reported. Because PS-ODN can have CpG-independent effects on cell function, we assessed Akt activation and survival in BMM in response to 1.5 μM 1018S and the non-CpG control, 1018S-GC. Both PS-ODN activated Akt phosphorylation, but the CpG-containing PS-ODN was the more potent stimulus (Fig. 4,A). The Akt activation observed with 1018S-GC was also apparent in TLR9^−/− BMM, but no CpG-specific response was observed in these cells. Similar results were apparent with 1668S and 1668S-GC (data not shown). Hence, the TLR9-independent activation of Akt that was previously reported (20) is likely to reflect CpG-independent effects of PS-ODN. Similarly, survival permitted by 1.5 μM 1018S was TLR9 dependent (Fig. 4 B). These doses of PS-ODN also had a modest effect on survival in TLR9^−/− BMM, but this effect was CpG independent and was variable between experiments (data not shown). Hence, PS-ODN activate Akt and permit survival of BMM in a TLR9-dependent and CpG-dependent fashion. In addition, high doses of PS-ODN have some TLR9-independent, CpG-independent effects on cell function.

Because CpG DNA activates the prosurvival PI3K-Akt and MEK-ERK pathways in several cell types including BMM, we determined whether these pathways were required for CpG DNA-induced survival in BMM. We focused on the effects of natural phosphodiester DNA (in the form of either EC DNA or the CpG-containing PO-ODN, AO-1), given its relevance to natural infection and host-pathogen interactions. Using the appearance of sub-G₀/G₁ DNA as an indicator of apoptosis in CSF-1-deprived BMM (Fig. 5, A and B), we observed a dose-dependent capacity of EC DNA and AO-1 to prevent apoptosis (Fig. 5,C). For comparison, we also monitored the antiapoptotic effect of CSF-1 and LPS (Fig. 5,C). CSF-1-deprived BMM were then pretreated with the PI3K inhibitor, LY294002, or the MEK1/2 inhibitor, PD98059, and the ability of CpG DNA to reduce the degree of apoptosis was monitored. LY294002, in a dose-dependent manner, reversed the ability of EC DNA, a CpG-containing PO-ODN (AO-1), LPS, and CSF-1 to block DNA fragmentation in BMM (Fig. 6,A). LY294002 alone enhanced the apoptosis of CSF-1-starved BMM somewhat, probably via reversal of PI3K-dependent survival promoted by factors within serum as previously observed (38). In contrast to the findings with LY294002, PD98059 did not reverse the ability of CpG DNA, LPS, and CSF-1 to block DNA fragmentation in BMM (Fig. 6,B). Consistent with these data, cleaved caspase-3, a key mediator of apoptosis, was not detected in CSF-1-starved BMM after treatment with a CpG-containing PO-ODN, LPS, or CSF-1 for 4 h (Fig. 6,C). However, cleaved caspase-3 was detectable if BMM were pretreated with LY294002, but not PD98059, before the addition of prosurvival stimuli (Fig. 6,C). We confirmed independently that treatment of BMM with LY294002 blocked CpG DNA-induced phosphorylation of Akt at Ser⁴⁷³ and that PD98059 blocked CpG DNA-induced phosphorylation of ERK1/2 (data not shown). Hence, PI3K-Akt, but not MEK-ERK, appeared to be required for LPS, CSF-1, and CpG DNA-mediated survival of BMM. The effect of LY294002, particularly at lower doses (10 and 30 μM), was more striking on the CpG DNA response than the LPS or CSF-1 response (Fig. 6 A). We hypothesized that this may be due to the ability of PI3K inhibitors such as wortmannin and LY294002 to block the uptake of DNA (33) (D. P. Sester, unpublished data). Hence, we determined whether LY294002 affected other CpG DNA responses (proinflammatory cytokine production) or specifically targeted survival pathways in BMM.

CSF-1-starved BMM were pretreated with LY294002, and then stimulated with either LPS, EC DNA, or the CpG-containing PO-ODN, AO-1, for 4 h in the presence of brefeldin A, followed by detection of intracellular TNF-α via flow cytometry. Pretreatment with 100 μM LY294002 had no significant effect on LPS-induced TNF-α production in BMM (either in terms of number of cells responding or amount of TNF-α produced per cell) (Fig. 7, A and B). Although LY294002 did reduce the amount of TNF-α produced per cell in response to CpG DNA (∼50%) (Fig. 7,B), it did not have a significant effect on the number of responding cells producing TNF-α (Fig. 7 A). Thus, although PI3K inhibitors do inhibit CpG DNA uptake (33), they do not abolish all responses to CpG DNA. We conclude that the PI3K-Akt pathway is required selectively for CpG DNA-mediated survival in BMM.

The above data show that natural CpG-containing phosphodiester DNA signals via TLR9 and PI3K-Akt to induce survival in BMM. To determine whether the catalytic activity of DNA-PK was also necessary, we assessed CpG DNA-induced survival in BMM from SCID mice that have a natural mutation in the gene encoding the catalytic subunit of DNA-PK that ablates function (39). Fig. 8,A shows that there was no obvious defect in the ability of EC DNA to promote the survival of BMM from SCID mice. Moreover, no difference in the dose-response for survival mediated by the CpG-containing PO-ODN, AO-1, was noted between BMM from wild-type and SCID mice (Fig. 8 B). Thus, bacterial CpG DNA-induced survival was not compromised in BMM from SCID mice.

In this study, we assessed the contribution of TLR9, DNA-PK, and the PI3K-Akt and MEK-ERK pathways to macrophage survival in response to CpG DNA. We found that survival of BMM in response to CpG-containing PO-ODN was dependent on TLR9 (Fig. 1, C and D) and PI3K/Akt (Fig. 6, A and C) and was independent of DNA-PK catalytic activity (Fig. 8) and MEK-ERK (Fig. 6,B). Similarly, CpG-containing PS-ODN activated Akt (Fig. 3,A) and survival (Fig. 1, A and B) via TLR9. High concentrations of CpG-containing PS-ODN did trigger low-level Akt activation and some survival independently of TLR9, but these effects were CpG independent and were minor compared with TLR9-dependent responses (Fig. 4, A and B). TLR9-independent effects of natural phosphodiester DNA on immune cells have also been reported, but these effects were also independent of the CpG motif (40, 41). Thus, to our knowledge, TLR9 is required for all CpG-dependent effects of DNA on immune cells. A very recent study has suggested that activation of Src family kinases occurs in response to detection of DNA at the cell surface, and that this response is CpG dependent and occurs upstream of TLR9 (42). We are currently investigating these findings.

Our findings contrast with the conclusions of Dragoi et al. (20), who argued that CpG DNA-induced activation of Akt was TLR9 independent and DNA-PK dependent in BMM. It should be noted that, in that study, high concentrations (∼1.5 μM) of the CpG-containing PS-ODN 1018S were used in cellular activation assays and no GpC control PS-ODN was included. Hence, it is likely that, as we describe here, phosphorothioate-modified DNA rather than the CpG motif triggered TLR9-independent Akt activation. Others have also reported that PS-ODN triggered Akt phosphorylation in mouse resident peritoneal macrophages independently of the CpG motif (43). Although not pertinent to the cellular mechanisms of activation by CpG motifs, DNA-PK-mediated activation of Akt and survival in response to PS-ODN is still likely to be relevant to the therapeutic applications of phosphorothioate-modified CpG DNA. It is possible that other CpG-independent effects of PS-ODN such as chemotaxis (43) and proliferation (44) may be mediated through activation of DNA-PK. Such properties may contribute to the immunomodulatory effects of CpG-containing PS-ODN in therapeutic settings.

Apart from the role of TLR9 in CpG DNA-mediated survival, we also focused on the downstream prosurvival pathways, MEK-ERK and PI3K-Akt. We previously observed that, whereas CpG DNA, LPS, and CSF-1 all efficiently activated ERK1/2, CpG DNA and LPS were relatively poor inducers of Akt phosphorylation compared with CSF-1 in BMM (45). Consistent with these findings in mouse macrophages, CSF-1 was also a more potent activator of Akt kinase activity than other survival factors, including LPS, in human monocytes (18). Despite these comparatively weak effects of bacterial products on Akt activation, we have shown that the PI3K inhibitor LY294002, which prevented Akt phosphorylation in BMM (data not shown), inhibited survival in response to LPS, CpG DNA, and CSF-1. This suggests that the PI3K-Akt pathway is required for survival in response to all three stimuli in BMM (Fig. 6, A and C). The PI3K-Akt pathway was previously implicated in CSF-1-mediated survival of human monocytes (46) and the murine macrophage cell line BAC1.2F5 (38). In addition, Park et al. (4) showed that CpG DNA-induced survival of DC was PI3K dependent, whereas others have demonstrated that Akt activation occurred in response to LPS in macrophages and DC (19, 47, 48).

In this study, we have shown that, in BMM, Akt activation in response to CpG DNA occurs downstream of TLR9 (Fig. 3, A and C) and is required for survival (Fig. 6, A and C). Despite its requirement for survival, PI3K was dispensable for the production of TNF-α in response to LPS and played only a partial role in regulating CpG DNA-mediated TNF-α production (Fig. 7, A and B). This was probably due to a partial blockade of all responses to bacterial/CpG DNA by PI3K inhibitors that is associated with reduced uptake of DNA (33) (D. P. Sester, unpublished data). This additional inhibitory effect may also underlie the increased potency of LY294002 in reversing survival mediated by EC DNA and AO-1, when compared with LPS, in BMM (Fig. 6 A).

Numerous reports show a role for the MEK1/2-ERK1/2 module for survival in a number of cell systems (49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63). In contrast to the PI3K-Akt pathway, MEK1/2 activity was dispensable for the antiapoptotic action of CpG DNA, LPS, or CSF-1 in BMM (Fig. 6, B and C). This is in agreement with previous studies showing no role for MEK-dependent ERK1/2 activation in BMM survival promoted by CSF-1 (64, 65), nor in the antiapoptotic action of CpG ODN and LPS on murine splenic DC (4). In summary, we have clearly delineated the TLR9-PI3K-Akt pathway as the mediator of CpG DNA-induced survival in mouse macrophages and, in contrast to previous workers (20), find no evidence for TLR9-independent responses to CpG DNA in Akt activation.

We thank Prof. Shizuo Akira for providing TLR9^−/− mice for these studies. We also thank Prof. Martin Lavin for providing bone marrow from Prkdcscid CB-17 mice and Prof. Alan Aderem for the provision of the 2.4G2 hybridoma.

The authors have no financial conflict of interest.