Inhibition of Neutrophil Apoptosis by TLR Agonists in Whole Blood: Involvement of the Phosphoinositide 3-Kinase/Akt and NF-κB Signaling Pathways, Leading to Increased Levels of Mcl-1, A1, and Phosphorylated Bad

Polymorphonuclear neutrophils (PMN) ² are key components of the first line of defense against microbial pathogens (1). They contribute to the early innate response by rapidly migrating into inflamed tissues, where their activation triggers microbicidal mechanisms such as release of proteolytic enzymes and antimicrobial peptides, and rapid production of reactive oxygen species (ROS). PMN activation is initiated upon recognition of Ab- or complement-opsonized particles (2). PMN also directly recognize microbial products via pattern recognition receptors such as TLR (3). Ten human TLRs have so far been identified, mediating responses to pathogen-associated molecular patterns (PAMPs) shared by many microorganisms. Human PMN have been reported to express all TLRs except TLR3 (4). TLRs are members of the IL-1R superfamily and share a common activation pathway mediated by their Toll/IL-1R signaling domain, resulting in activation of NF-κB and MAPK (5). Despite these shared pathways, TLRs probably show differences in their rate, intensity, or efficiency of activation, involving unidentified mechanisms. Selective pathways are reported to be triggered by some TLRs; in particular, TLR2, TLR4, and TLR9 can activate the PI3K pathway (6, 7). Activation of cell signaling cascades triggers immune responses leading to pathogen eradication.

PMN are usually short-lived immune cells, but the prolongation of their life span is critical in their efficiency against pathogens (8). PMN activation and survival is likely to be tightly regulated, as the cytotoxic substances they release can damage adjacent healthy tissue (9). Many inflammatory mediators, including cytokines, regulate cell survival by interfering with apoptosis (10, 11, 12, 13). Regulation of PMN survival has been widely studied but remains to be fully elucidated. In particular, the impact on PMN apoptosis of PAMPs recognized by the different TLRs has rarely been investigated. LPS, a TLR4 ligand, and peptidoglycan (PGN), a TLR2 ligand, delay PMN apoptosis through poorly known mechanisms (14, 15). No data on modulation of Bcl-2 family proteins after TLR activation have been reported.

In this study we report the first analysis of the differential effects of TLR agonists on apoptosis of human PMN studied in whole blood to minimize PMN activation during their isolation and to mimic physiological conditions. We used PGN from Staphylococcus aureus (a TLR2-selective stimulus), synthetic palmitoylated mimics of bacterial lipopeptides (Pam3CSK4, a TLR1/2 heterodimer stimulus), macrophage-activating lipopeptide-2 (MALP-2, a TLR2/6 heterodimer stimulus), purified LPS (a TLR4 stimulus), bacterial flagellin (a TLR5 stimulus), loxoribine (a guanosine analog and TLR7-selective stimulus), an imidazoquinoline pharmaceutical (R848, a TLR7/8 heterodimer stimulus) and unmethylated CpG-DNA (a TLR9 stimulus). We also examined mechanisms downstream of TLR signaling pathways leading to PMN survival by studying the participation of NF-κB, PI3K/Akt, and MAPK pathways, and modulation of Bcl-2 family proteins.

The reagents used and their sources were as follows: ultrapurified LPS from Escherichia coli serotype R515 (LPS), purified flagellin from Salmonella typhimurium, and synthetic MALP-2 (Alexis); PGN from S. aureus, R-848, the guanosine analog loxoribine, and a synthetic palmitoylated mimic of bacterial lipopeptides (Pam3CSK4) (Invivogen); unmethylated CpG-DNA (HyCult Biotechnology); hydroethidine (HE; Fluka); fMLP and rottlerin (Sigma-Aldrich); cycloheximide, diphenyleneiodonium (DPI), SN50, SB203580, PD98059, genistein, wortmannin, and GF109203X (Calbiochem); 3,3′-dihexyloxacarbocyanine iodide (DiOC6) (Molecular Probes); allophycocyanin-conjugated annexin V, 7-aminoactinomycin D (7-AAD), FITC-anti-CD15, allophycocyanin-anti-CD15, anti-Akt (protein kinase B α (PKBα)), anti-phospho-Akt (S472/S473), FITC-conjugated anti-active caspase-3, FITC-conjugated anti-Bad mAbs, and FITC-conjugated goat anti-rabbit Ab (BD Pharmingen, BD Biosciences); PE-conjugated anti-CD11b mAb, PE-anti-CD45 mAb (Immunotech); FITC-conjugated goat anti-mouse Ab (Nordic Immunology); anti-phospho-IκB kinase α (IKKα) (Ser¹⁸⁰)/IKKβ (Ser¹⁸¹), anti-heat shock protein (Hsp) 27 mAbs, anti-phospho-Bad (Ser¹³⁶) and anti-phospho-Hsp27 (Ser⁸²) rabbit polyclonal Abs (Cell Signaling Technology); anti-Mcl-1 mAb and anti-A1/Bfl-1 rabbit polyclonal Ab (Santa Cruz Biotechnology); anti-TLR2 mAb, anti-TLR3 mAb, anti-TLR5, and anti-TLR7 polyclonal Abs (Imgenex).

One milliliter samples of fresh blood from healthy donors, collected onto lithium heparinate (10 U/ml) were incubated in 24-well tissue cultures plates at 37°C with 5% CO₂ for various times with PBS or the following TLR agonists: LPS (0.001–10 ng/ml), PGN (0.01–1 μg/ml), R-848 (0.01–10 μg/ml), CpG-DNA (10–100 μg/ml), Pam3CSK4 (10–500 ng/ml), MALP-2 (0.01–10 ng/ml), flagellin (0.01–100 ng/ml), or loxoribine (0.1–1000 μM). Cycloheximide (10 μg/ml) and rhGM-CSF (1000 pg/ml) were used as proapoptotic (16) and antiapoptotic (12) controls, respectively.

In some experiments, samples were pretreated for 20 min with 10 μM DPI, an inhibitor of NADPH oxidase, or for 1 h with the NF-κB inhibitor SN50 (100 μg/ml) or kinase inhibitors at optimal concentrations previously determined in whole blood (wortmannin, 2500 nM; GF109203X, 5 μM; genistein, 100 μM; PD98059, 50 μM; SB203580, 25 μM; rottlerin, 10 μM).

Human PMN were isolated in LPS-free conditions by Dextran sedimentation and Ficoll-Hypaque centrifugation of freshly drawn blood; PMN were further purified by negative selection with pan anti-human HLA class II-coated magnetic beads (Miltenyi Biotec) to deplete B lymphocytes, activated T lymphocytes, and monocytes, as previously described (17). Less than 0.5% of cells were positive by nonspecific esterase staining, and flow cytometry showed the absence of CD45⁺/CD14^high, CD45⁺/CD3⁺, or CD45⁺/CD19⁺ cells; this showed that the PMN were highly purified, without contaminating monocytes.

Apoptosis was quantified by staining with annexin V and a vital dye (7-AAD) (18). After pretreatment of whole blood with TLR agonists for 2–48 h, samples (100 μl) were washed twice in PBS, incubated on ice with FITC-anti-CD15 and PE-anti-CD45 for 15 min, and then with allophycocyanin-annexin V for 15 min. After dilution in PBS (500 μl), samples were incubated with 7-AAD at room temperature for 15 min and analyzed immediately by flow cytometry. PMN were identified on the CD15/side scatter (SSC) dot plot (see Fig. 1,B); use of the combination of allophycocyanin-annexin V and 7-AAD differentiates between early apoptotic PMN (annexin V⁺, 7-AAD⁻) and late apoptotic PMN (annexin V⁺, 7-AAD⁺) (see Fig. 1,C). Fig. 1 A (after CD45 gating) shows an accurate distribution of the different leukocyte populations on the forward scatter (FSC)/SSC dot plot after 8 h of incubation. We checked that the percentage of apoptotic PMN, determined by flow cytometry on whole blood, correlated with the percentage obtained by morphological assessment on cytospin preparation stained with May-Grünwald Giemsa (not shown).

Some experiments were performed with highly purified PMN (1 × 10⁶/ml) incubated with TLR agonists at optimal concentrations for 8 h and then with allophycocyanin-annexin V and 7-AAD before flow cytometry.

The mitochondrial apoptotic pathway was detected as a loss of the mitochondrial transmembrane potential (Δψ_m), measured as a reduction of the DiOC6 incorporation (19). After pretreatment of whole blood at 37°C for 8 h with PBS or TLR agonists, samples (100 μl) were loaded with DiOC6 (40 nM) for 15 min at 37°C and then stained with allophycocyanin-anti-CD15 for 30 min at 4°C. Samples were then suspended in PBS and analyzed immediately by flow cytometry.

Leukocytes (obtained after red cell lysis with FACS lysing solution) were fixed with 2% paraformaldehyde-PBS for 10 min at 37°C. After one wash with PBS, leukocytes were incubated in ice-cold PBS-90% methanol in the dark for 30 min at 4°C to permeabilize the membranes as previously described (20). After one wash with PBS-human serum albumin (HSA) (2%), cells were stained with FITC-conjugated anti-active caspase-3 for 1 h at room temperature. After one wash (400 × g for 5 min), cells were resuspended in 1% paraformaldehyde-PBS and analyzed by flow cytometry.

Superoxide anion () production was measured with a flow cytometric assay derived from the HE oxidation technique (21): HE diffuses into cells and, during the PMN oxidative burst, nonfluorescent intracellular HE is oxidized by to highly fluorescent ethidium (E⁺), that is trapped in the nucleus by intercalation into DNA. After the different treatments, whole blood (1 ml) was loaded for 15 min with HE (1500 ng/ml) at 37°C, followed by PBS or 10⁻⁶ M fMLP for 5 min. The reaction was stopped and RBC were lysed as described above. After one wash, leukocytes were resuspended in 1% paraformaldehyde-PBS.

After the different treatments, whole blood was incubated at 37°C with PBS or 10⁻⁶ M fMLP for 5 min. Samples (100 μl) were stained at 4°C for 30 min with FITC-anti-human CD11b.

After incubation of whole blood with TLR agonists or PBS for various times at 37°C, leukocytes were permeabilized in 90% methanol as previously described (20). Cells were then stained with anti-Akt, anti-Akt phosphospecific, anti-Hsp27, anti-Hsp27 phosphospecific, anti-Mcl-1, anti-A1/Bfl-1, or anti-Bad phosphospecific Abs for 1 h at room temperature and washed once in PBS-2% HSA. Samples were then incubated for 30 min with FITC-goat anti-mouse or anti-rabbit Ab. Bad content was studied by staining with FITC-conjugated anti-Bad. After one wash, leukocytes were resuspended in 1% paraformaldehyde-PBS and analyzed by flow cytometry.

We used a BD Immunocytometry Systems FACSCalibur. To measure apoptosis in whole blood, PMN were identified on the CD15/SSC dot plot and 2 × 10⁵ events were counted per sample. In other experiments, FSC and SSC were used to identify the PMN population and to gate out other cells and debris; 10,000 events were counted per sample. All the results were obtained with a constant photomultiplier gain. The data were analyzed using CellQuest software (BD Biosciences), and mean fluorescence intensity (MFI) was used to quantify the responses. Caspase-3 activity was expressed in relative units (percentage of caspase-3^high cells × MFI). Nonspecific Ab binding was determined on cells incubated with the same concentration of the corresponding isotype control or with nonimmune serum.

Suspensions of 40 × 10⁶ PMN/ml in PBS buffer were treated with 2.7 mM diisopropylfluorophosphate for 20 min at 4°C and pelleted at 400 × g for 8 min at 4°C. The pellet was resuspended in CHAPS solubilization buffer containing 50 mM Tris, pH 7.5, 15 mM CHAPS, 1 mM EDTA and antiproteases. The cells were incubated on ice and the suspension was then centrifuged at 1500 × g for 5 min. Following SDS-PAGE on 10% acrylamide gels, proteins were transferred to nitrocellulose filters. The filters were incubated for 1 h at room temperature in 50 mM Tris, 150 mM NaCl, 0.1% Tween 20 (TBST) containing 5% (w/v) fat-free dried milk. Nitrocellulose membranes were incubated overnight with specific Abs against TLR2, TLR3, TLR5, and TLR7 at 1/500 dilution. Following five washes with TBST, the membranes were incubated with goat anti-mouse or goat anti-rabbit Abs conjugated to HRP. After five washes with TBST, revelation was performed with a chemiluminescence method (ECL; Amersham Life Sciences) following the manufacturer’s instructions.

Data are reported as means ± SEM. Comparisons were based on ANOVA and Tukey’s posthoc test, using Prism 3.0 software (Graph Pad Software).

Whole blood PMN cultured at 37°C died by apoptosis: ∼26 and 70% of cells were annexin V⁺ after 8 and 24 h, respectively (Fig. 1,D). As previously reported with isolated PMN (12, 16), apoptosis was accelerated by cycloheximide (65% of annexin V⁺ cells after 8 h), and delayed by GM-CSF (4% of annexin V⁺ cells after 8 h). As shown in Fig. 2, the percentage of total annexin V⁺ cells, and the percentage of annexin V⁺, 7-AAD⁻ cells, fell significantly, in a concentration-dependent manner after 8 h of treatment with all TLR agonists except flagellin (TLR5) and loxoribine (TLR7). We checked that environmental LPS did not contribute to the effect of the TLR agonists (except for the TLR4 agonist, which is purified LPS). In fact, the prolongation of PMN survival by PGN, R-848, CpG-DNA, Pam3CSK4, and MALP-2 was not modified by preincubation with a TLR4 neutralizing Ab, while the effect of LPS on PMN survival was completely abolished (not shown). PMN apoptosis, measured in terms of total annexin V⁺ PMN, fell to a similar extent after incubation for 8 h with LPS, PGN, R-848, and CpG-DNA at optimal concentrations. This inhibition was significantly stronger than that induced by Pam3CSK4 and MALP-2 (Table I). Similar levels of inhibition were found in the early stage of PMN apoptosis (annexin V⁺/7AAD⁻ cells; not shown). A kinetic study showed that LPS, PGN, R-848, and CpG-DNA induced ∼50% inhibition of PMN apoptosis after 2 h, rising to a maximum of 80% after 8 h. In samples treated with all the TLR agonists except flagellin and loxoribine, 80% of PMN were annexin V⁺ after 48 h incubation, compared with 24 h in PBS-treated samples, further showing that TLR agonists delay PMN apoptosis (Fig. 3).

As cytokine production by contaminating monocytes has been implicated in TLR agonist-induced inhibition of PMN apoptosis at later times (22 h) but not at earlier times (4 h) (22), we examined the effect of 8 h of incubation with TLR agonists on highly purified PMN depleted of all monocytes as previously described (17). TLR agonist-induced inhibition of PMN apoptosis was similar whether apoptosis was studied in whole blood or in highly purified preparations, suggesting a direct effect of TLR agonists on PMN apoptosis. However, apoptosis inhibition was slightly less effective with purified PMN than with whole blood: the protective agonists induced 38–69% inhibition of PMN apoptosis in purified preparations and 46–87% in whole blood (Table I). To minimize cell activation during isolation and to better mimic physiological conditions, the following experiments were performed with PMN in their whole blood environment.

After 8 h of incubation, PBS-treated PMN exhibited a reduced capacity to produce superoxide anion () and to increase surface CD11b expression in response to fMLP relative to baseline (Fig. 4). This impairment was found with both annexin V⁻ and annexin V⁺ PMN, although annexin V⁺ cells were less potent than annexin V⁻ cells. In contrast, incubation for 8 h with the TLR agonists found to delay apoptosis increased production in the basal state and also in response to fMLP (LPS was the most potent agonist in terms of increasing production) (Fig. 4,A). This impact of the TLR agonists on ROS production did not differ significantly according to annexin V staining, although the stimulation index (MFI of fMLP-stimulated sample/MFI of unstimulated sample) was slightly lower in annexin V⁺ than annexin V⁻ populations. Similar results were observed concerning PMN surface CD11b expression (Fig. 4 B). Flagellin and loxoribine did not significantly modify PMN responses. This increased ROS production was not involved in the effects of TLR agonists on PMN apoptosis, as incubation with DPI, a NADPH oxidase inhibitor, altered neither the intensity nor the kinetics of apoptosis inhibition (not shown).

Thus, TLR agonists found to delay apoptosis also extended the functional life span of PMN.

As flagellin (a TLR5 agonist) and loxoribine (a TLR7 agonist) did not delay PMN apoptosis, in contrast to the other TLR agonists, we checked that TLR5 and TLR7 were expressed in PMN. TLR2 and TLR3 were used as positive and negative controls, respectively (4). Fig. 5 shows that TLR2, TLR5, and TLR7 were expressed in PMN, contrary to TLR3.

To identify the stage of the apoptotic program at which TLR agonists act to delay apoptosis, we measured the effects of TLR agonists on caspase-3 activity and Δψ_m. As shown in Table II caspase-3 activity was detected during spontaneous PMN apoptosis after 8 h incubation with PBS at 37°C, and was significantly reduced by the TLR agonists that delayed PMN apoptosis (LPS, PGN, R-848, CpG-DNA, Pam3CSK4, MALP-2). In contrast, flagellin and loxoribine did not significantly modify caspase-3 activity relative to PBS. The proportion of DiOC6^low (loss of Δψ_m) cells was significantly reduced after treatment with the active TLR agonists, relative to both PBS, flagellin and loxoribine. Furthermore, both caspase-3 activity and the percentage of DiOC6^low cells correlated with the percentage of annexin V⁺ cells.

These results suggest that apoptosis was delayed, at least in part, through an effect at or upstream of the mitochondrial membrane permeabilization.

NF-κB activation has been implicated in PMN survival (23), and particularly in LPS-induced PMN survival. However, the involvement of the NF-κB pathway in the effects of TLR agonists that delay PMN apoptosis has not been comparatively investigated.

To determine the involvement of the NF-κB pathway in whole blood, we first used an inhibitor-based approach. The NF-κB inhibitor (SN50) had no significant effect on spontaneous apoptosis: the percentage of annexin V⁺ cells was 24.2 ± 4.5 and 25.1 ± 3.8 in samples incubated with PBS and SN50, respectively. As shown in Fig. 6,A, SN50 (100 μg/ml), which inhibits nuclear translocation of NF-κB, significantly reduced TLR-mediated inhibition of PMN apoptosis at 8 h, suggesting a role of NF-κB. The control peptide SN50M did not significantly alter PMN apoptosis, whatever the TLR agonist used. Another NF-κB inhibitor, the sesquiterpene lactone parthenolide (10 μM) also inhibited TLR-induced inhibition of PMN apoptosis (not shown). To confirm these findings, we used flow cytometry and a mouse anti-phospho-IKK mAb, and found that the protective agonists significantly induced IKK phosphorylation relative to PBS-treated samples (Fig. 6 B).

However, the NF-κB inhibitors only partially reversed TLR agonist-induced inhibition of PMN apoptosis, suggesting the involvement of other signaling pathways.

TLRs may activate many other signaling pathways, a number of which have already been implicated in the regulation of PMN life span, including MAPKs (24, 25), PI3K (26, 27), and tyrosine kinases (28).

To investigate the signaling pathways involved in TLR agonist-induced inhibition of apoptosis, we first examined the effects of various kinase inhibitors on PMN in whole blood incubated with TLR agonists for 8 h. We checked that kinase inhibitors alone did not alter PMN apoptosis; in particular, the percentage of annexin V⁺ cells was 25.3 ± 5.2 and 23.2 ± 4.7 in samples incubated with PBS and wortmannin, respectively. As shown in Fig. 7,A, only wortmannin (a PI3K inhibitor) attenuated LPS-dependent inhibition of apoptosis. Similar effect was observed with a second PI3K inhibitor, LY2940002 (25 μM) (not shown). Blockade of both the NF-κB and PI3K pathways potentiated this effect (18% apoptosis inhibition with LPS + wortmannin + SN50). GF109203X (a protein kinase C (PKC) inhibitor), genistein (a broadly specific tyrosine kinase inhibitor), PD98059 (a MEK1/2 kinase inhibitor), SB203580 (a p38MAPK inhibitor), and rottlerin (a PKCδ inhibitor) had no effects on LPS-induced PMN survival whatever the concentrations used. Similar results were observed with the other TLR agonists that delayed neutrophil apoptosis (not shown). As shown in Fig. 7 B, wortmannin has a similar inhibitory effect on LPS, PGN, R-848, CpG-DNA, Pam3CSK4, and MALP-2-induced PMN survival. Wortmannin also inhibited the effect of TLR agonists on caspase-3 activity, and reduced the percentage of DiOC6^low cells (not shown).

One mechanism involved in PI3K prevention of apoptosis is activation of the PKB/Akt pathway (29). We therefore studied the phospho-Akt content of intact PMN treated in whole blood, by means of flow cytometry with a mouse anti-human-phospho-Akt mAb. As shown in Fig. 8,A, incubation of whole blood with LPS significantly increased Akt phosphorylation after as little as 2 min, as compared with PBS. Similar results were observed with PGN, R-848, CpG-DNA, Pam3CSK4, and MALP-2 (Fig. 8 B). In contrast, pretreatment with flagellin and loxoribine did not modify Akt phosphorylation. Total Akt content, measured with a mouse anti-human Akt mAb in the same conditions, was not modified by treatment with TLR agonists (not shown).

Hsp27 has been reported to associate with Akt (30). Furthermore, phosphorylation of Hsp27 by Akt results in its dissociation from Akt and may participate to the Hsp27-induced delay in PMN apoptosis (31). We therefore studied the effects of TLR agonists on Hsp27 phosphorylation by means of flow cytometry. Phospho-Hsp27 content was significantly increased after 5 min incubation with LPS (Fig. 8,C) and with the other TLR agonists which increased Akt phosphorylation (Fig. 8,D). No modification of total Hsp27 content was observed. The TLR agonist-induced increase in phospho-Hsp27 and phospho-Akt content was completely reversed by pretreatment of whole blood with wortmannin (Fig. 8, B and D).

It is now generally accepted that blood PMN do not express the classical antiapoptotic protein Bcl-2 (32). In contrast, there is a very close correlation between levels of the antiapoptotic protein Mcl-1 and PMN survival (32). As previously reported with isolated PMN (33), we found that Mcl-1 levels fell during constitutive PMN apoptosis in whole blood (Fig. 9,A). Treatment of whole blood with LPS significantly increased Mcl-1 levels after 1 h as compared with PBS controls; values peaked at 2 h (Fig. 9,A). Similar results were observed with PGN, R-848, CpG-DNA, Pam3CSK4, and MALP-2 (Fig. 9 B). The TLR-induced increase in Mcl-1 was significantly reversed by 30 min preincubation with wortmannin (not shown).

The only other antiapoptotic factor so far implicated in PMN survival is A1/Bfl-1. In contrast to Mcl-1, A1 levels did not fall during spontaneous PMN apoptosis. However, the A1 level increased significantly after incubation with TLR agonists that delayed apoptosis. This effect appeared as early as 1 h and was maximal after 2 h; flagellin and loxoribine had no effect (Fig. 9, C and D). As synthesis of the antiapoptotic protein A1 depends on NF-κB activation (34), we analyzed the effect of SN50 on TLR-induced A1 levels. We found that the TLR agonist-induced increase in A1 was reversed by sample preincubation with the NF-κB inhibitor SN50. In fact, MFI of the sample pretreated with PBS and then incubated with LPS was 209.0 ± 12.7, while MFI of the sample pretreated with SN50 and then incubated with LPS was 149.2 ± 5 and did not differ from that of the sample incubated with PBS alone. Similar results were observed with the other TLR agonists (not shown). Preincubation with wortmannin only partially reversed the effect of TLR agonists on A1 levels (not shown).

Most reports show that levels of proapoptotic proteins in PMN remain fairly constant when the cells age naturally and when their apoptosis is experimentally delayed or accelerated (32). Bad phosphorylation by Akt/PKB can induce an antiapoptotic effect by Bad dissociation from antiapoptotic proteins of the Bcl-2 family, also resulting in delayed apoptosis (35). We therefore examined whether TLR agonist-induced phosphorylation of Akt was associated with Bad phosphorylation. As shown in Fig. 9,E, levels of phospho-Bad increased significantly during incubation (30 and 60 min) with LPS, while total Bad content was unaffected (not shown). Similar results were observed with other TLR agonists that delayed apoptosis (Fig. 9 F). This increase was completely reversed by preincubation with wortmannin (not shown).

Our results show that all TLR agonists except for flagellin (TLR5) and loxoribine (TLR7) delay spontaneous apoptosis of human PMN and extend their functional life span in whole blood. The antiapoptotic action of TLR agonists required activation of NF-κB and PI3K. Furthermore, flow cytometry of intact cells showed that the TLR agonists which delayed PMN apoptosis induced phosphorylation of Akt, a major target of PI3K, as well as Hsp27 phosphorylation. Finally, the TLR-induced delay in PMN apoptosis was associated with increased levels of Mcl-1 and A1, both of which are antiapoptotic members of the Bcl-2 family. These effects were reversed by PI3K and NF-κB inhibitors, respectively. In addition, TLR activation led to PI3K-dependent phosphorylation of the proapoptotic protein Bad, which is known to result in delayed PMN apoptosis.

One particularity of this study is that we analyzed PMN apoptosis by flow cytometry in whole blood. This avoided PMN isolation procedures, which have been shown to induce surface expression of molecules that are not detected in whole blood and may thereby alter PMN responses (36). Such artifacts might contribute to the acceleration of spontaneous apoptosis that we observed after PMN purification. In addition, analysis of TLR agonist-induced modulation of PMN survival in whole blood mimics physiological conditions more closely than the use of isolated cells. In particular, interactions between cellular elements have been reported to be important in maintaining PMN viability (37). However, our data do not rule out a contribution of PBMC to the kinetics of spontaneous PMN apoptosis and to the TLR agonist-induced delay in PMN apoptosis. Release of survival factors such as cytokines by activated monocytes (22, 38) could contribute to the stronger agonist-induced inhibition of PMN apoptosis that we observed in whole blood as compared with highly purified PMN.

In keeping with previous data (12, 14, 15, 39), we found that LPS delayed PMN apoptosis in whole blood. We also extended this observation to PGN, R-848, CpG-DNA, Pam3CSK, and MALP-2. MALP-2 (TLR2/6) and Pam3CSK4 (TLR1/2) had milder effects on PMN survival. TLR1 and TLR6 act as coreceptors for TLR2 and have been reported to inhibit cellular responsiveness to activating ligands (40, 41). The lack of effect of flagellin and loxoribine could be related to a lack of triggering of their signaling pathways, itself possibly due to weak expression of TLR5 and TLR7, in keeping with previous reports that the corresponding mRNAs are weakly expressed in PMN (41, 42).

We observed, for the first time, that treatment of PMN in whole blood with TLR agonists which strongly delayed PMN apoptosis at 8 h maintained and even enhanced PMN functions, both in the basal state and in response to a bacterial product (fMLP), as compared with values obtained at baseline (T0h) and with untreated samples at 8 h. TLRs play a central role in innate immunity by mediating PAMP recognition, as reflected by the increased susceptibility to infections of children with deficient TLR transduction (43). The prolonged functional life span induced by LPS, PGN, R-848, CpG-DNA and, to a lesser degree, Pam3CSK and MALP-2 may represent a crucial enhancement of PMN defenses against microbial pathogens. Furthermore, our data, obtained with whole blood, suggest that in pathological situations such as sepsis the delayed PMN apoptosis induced by TLR agonists at the systemic level could potentiate inflammatory reactions and lead to vessel damage.

Many groups have investigated the signaling pathways involved in PMN survival, but the specific roles of individual pathways involved in the response to individual TLR ligands remain to be clarified. We first used NF-κB inhibitors commonly used to inhibit NF-κB nuclear translocation (44, 45). The NF-κB control peptide SN50M had only minor effects on TLR-induced PMN survival, whereas the active SN50 NF-κB inhibitor significantly prevented TLR agonist-induced survival. Similar effects were observed with a second NF-κB inhibitor, the sesquiterpene lactone parthenolide. In addition, using an anti-phosphorylated-IKK Ab, we found that TLR-agonists protecting against apoptosis induced IKK phosphorylation, which permits IκB phosphorylation, leading to its degradation and NF-κB nuclear translocation (46). These data are in keeping with those previously reported on LPS-treated isolated PMN (15, 23). In this study, we extend this observation to all TLR agonists capable of delaying PMN apoptosis, pointing to a major role of NF-κB in TLR-mediated PMN survival.

Our results also demonstrate that TLR agonists delay PMN apoptosis via a PI3K-dependent pathway, although our data do not rule out the possibility that PI3K inhibitors could also act by preventing, at least in part, the generation of survival signals by other cells in whole blood. In contrast, tyrosine kinases, MAPK and PKC do not appear to be involved. The lipid products of PI3K–predominantly phosphatidylinositol 3,4,5-triphosphate–induce translocation of Akt/PKB to the plasma membrane, where it is phosphorylated and activated by phosphatidyl-inositol 3,4,5-phosphate-dependent protein kinase (PDK1), and this pathway has been forwarded as a major mediator downstream of PI3K (29, 47). Akt/PKB activity has been shown to prevent apoptosis induced by cytokines and growth factors, cellular stress, chemotherapeutic agents, and irradiation (48). Our data showing that TLR agonists capable of delaying PMN apoptosis increase PI3K-dependent Akt phosphorylation strongly suggest a central role of Akt in this effect. Furthermore, we show for the first time that the same TLR agonists induce a PI3K-dependent increase in Hsp27 phosphorylation. It has recently been shown that Hsp27 phosphorylation by Akt leads to disruption of the Akt-Hsp27 interaction, and it has been suggested that released Hsp27 may promote independent survival signals (31). It has also been postulated that Hsp27 inhibits apoptosis through inactivation of caspase-3, caspase-9, and inhibition of cytochrome c release (49, 50).

Akt activation could also explain, at least in part, the role of NF-κB in TLR-induced PMN survival. Indeed, it was recently reported that Akt modulates NF-κB-dependent transcription in TLR2-stimulated PMN by modifying phosphorylation of the p65 subunit (7). Nevertheless, Akt is necessary but not sufficient for NF-κB activation by TLR (51).

PI3K/Akt activation could be an alternative pathway to the IL-1R-associated kinase/IKK/NF-κB pathway, leading to independent modulation of Bcl-2 family proteins (52). Increased levels of Mcl-1 have been implicated in PMN survival induced by proinflammatory cytokines such as GM-CSF, IL-1, TNF-α, and IL-15 (12, 33, 53, 54). We found that PMN treatment for 1 h with TLR agonists capable of delaying neutrophil apoptosis prevented the loss of Mcl-1. In keeping with previous data demonstrating that increased Mcl-1 translation depends on the PI3K/Akt pathway (55), we found that the TLR-induced Mcl-1 elevation was reversed by pretreatment with wortmannin. The only other antiapoptotic gene product so far implicated in PMN survival is A1. Previous studies have been restricted to mRNA, and show that A1 transcripts are cytokine-regulated in human PMN (34). Using a novel method–flow cytometry–to investigate A1 protein expression by intact permeabilized PMN, we showed for the first time that PMN activation by TLR agonists is associated with an increased intracellular content of A1 protein. In keeping with previous data (56), this effect was regulated by NF-κB and partially inhibited by wortmannin. Finally, kinetic analysis showed an early increase in Mcl-1 and A1 levels and a return to control value by 240 min. These results strongly suggest that this early increase in Mcl-1 and A1 levels inhibits the cellular apoptosis machinery and gives an advantage to TLR agonist-stimulated cells.

Rapid regulation of PMN survival could be achieved by posttranslational protein modifications. In particular, increased Bad (Bcl-x_L/Bcl-2-associated death promoter homologue) phosphorylation has been implicated in PMN survival induced by GM-CSF (13, 24). Phosphorylated Bad interacts with 14-3-3 protein, and the resulting Bad sequestration diminishes Bad binding to diverse antiapoptotic Bcl-2 proteins anchored to the mitochondrial membrane (57). Increased amounts of antiapoptotic proteins are then free to bind to Bax and to prevent its proapoptotic activity, leading to cell survival. Recently, it was reported that underphosphorylated Bad interacts with all antiapoptotic Bcl-2 family members, and particularly A1 and Mcl-1 (58). Our results demonstrating that TLR agonists which inhibit PMN apoptosis increase Bad phosphorylation from 30 to 60 min strongly suggest that this phenomenon is involved in PMN survival induced by TLR activation. In keeping with previous data showing that Bad phosphorylation on serine 136 is induced by phospho-Akt (59), we observed that TLR agonist-induced Bad phosphorylation was PI3K-dependent.

Taken together, our findings demonstrate that relatively specific agonists of TLR, namely LPS (TLR4), PGN (TLR2), R-848 (TLR7/8), CpG-DNA (TLR9), Pam3CSK4 (TLR1/2), MALP-2 (TLR2/6), with the exception of flagellin (TLR5) and loxoribine (TLR7), are able to delay PMN apoptosis and extend the PMN functional life span in whole blood. Our results also point to the involvement of the PI3K/Akt and NF-κB pathways in PMN survival induced by TLR activation. PI3K-dependent phosphorylation of Akt may be strongly involved in the increased levels of the antiapoptotic protein Mcl-1 and the increased phosphorylation of the proapoptotic protein Bad. The antiapoptotic action of TLR agonists may be facilitated by Akt-dependent phosphorylation of Hsp27. In addition, NF-κB activation may lead to increased levels of the antiapoptotic protein A1. To our knowledge, this is the first report that Bcl-2 family proteins are modulated upon TLR activation. Ongoing studies may identify new therapeutic targets for regulating TLR-induced PMN survival and functions in inflammatory disorders in which these cells contribute to bystander tissue damage.

The authors have no financial conflict of interest.