Lipopolysaccharide stimulates p38-dependent induction of antiviral genes in neutrophils independently of paracrine factors.

Lipopolysaccharide (LPS) induces neutrophils to synthesize and secrete pro-inflammatory cytokines and chemokines, which are regulated at both the transcriptional and translational level. We reported previously that neutrophils stimulated with LPS induce expression of genes typically expressed in response to stimulation with antiviral type I interferons (IFN), such as myxovirus resistance-1 (MX1). However, we present evidence that this response of neutrophils to lipopolysaccharide occurs in the absence of interferon-dependent signaling. Lipopolysaccharide-stimulated neutrophils do not phosphorylate the interferon-associated transcription factors signal transducer and activator of transcription-1 and -3, and medium from lipopolysaccharide-stimulated cells was unable to induce MX1 gene expression, suggesting a soluble factor is not involved. Furthermore, LPS did not alter expression of IFNA and IFNB genes. In contrast to neutrophils, LPS-stimulated human monocyte-derived macrophages induced the expression of MX1, but IFNB was induced, and medium from LPS-stimulated monocyte-derived macrophages supported MX1 induction. An inhibitor of p38 kinase blocked induction of MX1 by lipopolysaccharide, but not IFNalpha, in neutrophils, and induction of MX1 was dependent on protein synthesis. LPS, but not IFNalpha, substantially activated p38. In contrast, the induction of MX1 by LPS in monocyte-derived macrophages was insensitive to p38 inhibition, although p38 is phosphorylated in LPS-stimulated but not IFNalpha-stimulated monocyte-derived macrophages. The expression of MX1 in neutrophils and monocyte-derived macrophages is mediated by TLR4 but not TLR2. The data presented here indicate that lipopolysaccharide activates novel interferon-independent signaling pathways in neutrophils and that induction of antiviral genes is a consequence of exposure of neutrophils to bacterial products.

Neutrophils play a vital role in the innate immune response by migrating to sites of infection, ingesting microorganisms, and producing antimicrobial agents. Activation of neutrophils can occur by recognition of structural features of microbes, including the bacterial cell wall component lipopolysaccharide (LPS) 1 (1). Exposure of neutrophils and other immune cells to LPS primes the cell to respond to secondary agonists with secretion of antimicrobial peptides and formation of superoxide. In addition, it is now recognized that neutrophils are potent synthetic cells and produce cytokines and chemokines, such as TNF␣ and IL-1␤, which are coded for by the TNFA and IL-1␤ genes (2,3). Production of these immunomodulatory proteins in response to LPS is controlled at both the transcriptional and translational level.
The toll-like receptor (TLR) family is implicated in the recognition of bacteria and in the activation of cellular signaling pathways important in host defense (4,5). Genetic and biochemical data suggest that TLR4 is the major LPS receptor (4, 6 -10), although TLR2 may also play a role (11,12). LPS stimulates a pathway leading to the activation of the transcriptional regulator NFB in many cell types (4,7,8,13,14), including neutrophils (15). Whereas NFB plays a pivotal role in regulating the expression of cytokine and chemokine genes, other transcriptional regulators may also be involved in signaling from TLRs. In contrast to many other cell types, transcription factor activation is poorly understood in neutrophils.
LPS activates tyrosine and serine/threonine kinases in cells (16 -23). Among these the MAP kinase family of serine/threonine kinases (p42/p44 MAP kinase, c-Jun N-terminal kinase, and p38) plays a major role in cellular activation in a variety of cell types (24 -26). However, in neutrophils the p38 MAP kinase family is predominantly activated in response to LPS and regulates the translation of cytokines, adhesion, and migration (27)(28)(29)(30)(31). The role of p38 in regulation of gene expression is less well understood, although transcription factors, including activating transcription factor-2, SRF accessory protein-1, and CCAAT-enhancer binding protein-␤ (26), are known substrates of p38. The role of p38 in activating these transcription factors in neutrophils has not been explored. However, inhibitors of p38 have little effect on the expression of cytokine genes in neutrophils (28,32).
The cellular response to viral infection includes the induction of genes for the type I interferons, IFN␣ and IFN␤, which act in an autocrine or paracrine manner to stimulate the induction of interferon-stimulated genes (ISGs) (33). Induction of IFN␣/␤ protein by virus requires the activation of the IRF family of transcriptional regulators by phosphorylation, and subsequent gene expression of IFNA and IFNB (34,35), through poorly understood pathways. IFN␣/␤ binds to the cell surface IFN receptor and activates the Jak1 and Tyk2 tyrosine kinases leading to the phosphorylation of the STAT1 and STAT2 transcription factors and subsequent dimerization (33). The major regulatory factor involved in IFN␣/␤ action, ISGF3, is a complex of STAT1, STAT2, and IRF9-p48-ISGF3␥. ISGF3 binds to a conserved sequence in the 3Ј region of IFN-responsive genes known as the interferon-stimulated response element. ISGs, such as MX1, are regulated primarily by transcriptional activation of this element, and the antiviral activity of IFN␣/␤ resides in their ability to induce ISGs.
Links between anti-bacterial and antiviral activity have been suggested previously. In particular, the induction of IFN␤ by LPS has been well described (36,37). In this way, antiviral gene expression is accomplished in response to bacterial infection. We recently described (32,38) the regulation of ISGs in LPS-treated neutrophils. Here we investigated the mechanisms by which neutrophils induce ISGs in response to LPS. LPS-stimulated neutrophils do not increase IFNA or IFNB gene expression, secrete a soluble mediator to induce ISGs, or phosphorylate STAT proteins; in contrast, monocyte-derived macrophages increase IFNB expression, secrete a soluble ISGinducing factor(s), and phosphorylate STAT proteins. Furthermore, the induction of ISGs by LPS in neutrophils, but not monocyte-derived macrophages, is sensitive to inhibitors of p38, and these effects are mediated by TLR4 but not TLR2. Whereas both IFN-dependent and -independent induction of ISGs by dsRNA and virus has been described (39 -42), to our knowledge this is the first report of IFN-independent ISG induction by LPS. These data suggest a potential new role of neutrophils in innate immunity by inducing an antiviral genotype and suggest that distinct pathways are activated by LPS in neutrophils and monocyte-derived macrophages.
Cell Preparation and Stimulation-Neutrophils were isolated from citrated blood of healthy donors and purified on a Percoll gradient, as described previously (44). Contamination with monocytic cells is less than 5%. Cells (20 -25 ϫ 10 6 cells/ml) were resuspended in neutrophil medium (RPMI containing 1% heat-inactivated platelet-poor plasma and 10 mM HEPES, pH 7.6) and divided into 1.5-ml tubes. Heatinactivated platelet-poor plasma is a necessary source of LPS-binding protein (45). RT-PCR analysis utilized 20 ϫ 10 6 cells (in 1 ml). Similar RT-PCR results were obtained at lower cell concentrations. Cells were treated with the indicated concentration of LPS, IFN␣/D, and PGN and rotated continuously at 37°C for up to 4 h. In some experiments, cell supernatants were retained for conditioned media experiments or enzyme-linked immunosorbent assay (see below).
Human monocyte derived macrophages were prepared by the method of Fadok et al. (46). Briefly, peripheral monocytic cells from healthy human volunteers were washed in Hanks' balanced salt solution with calcium and magnesium, and 12 ϫ 10 6 cells in X-vivo medium (Bio-Whittaker) per well were seeded in 6-well plates. One hour after plating, non-adherent cells were removed by washing, and adherent cells were grown in X-vivo, 10% heat-inactivated human serum. The medium was changed every 3 days thereafter, and experiments were performed 7 or 8 days after plating. RAW264.7 cells, obtained from the American Type Culture Collection, were cultured as described (47).
Affymetrix Oligonucleotide Array-Preparation of total RNA (5 g) for analysis by oligonucleotide array (Affymetrix) was performed as described (32). Data were analyzed using Affymetrix GeneChip software. Time courses from three donors were analyzed.
Reverse Transcription (RT)-PCR-cDNA was prepared by reverse transcription using 2 g of total RNA, derived from 20 ϫ 10 6 neutrophils, or 0.5-2 g of RNA from monocyte-derived macrophages and treated as indicated. PCRs were performed using specific primers for MX1, PKR, TNFA, IFNA (48), IFNB, and GAPDH. Murine-specific primers were used in experiments with RAW264.7 cells.
Enzyme-linked Immunosorbent Assay-Secretion of IFN␣ and IFN␤ was quantified by enzyme-linked immunosorbent assay kits as directed by the manufacturer (R & D Systems).
Conditioned Media Experiments-Supernatants were isolated from neutrophils or monocyte-derived macrophages stimulated for 4 h with 100 ng/ml LPS (LPS-CM) or non-stimulated cells (NS-CM). Naive neutrophils (20 ϫ 10 6 ) were suspended in 250 l of neutrophil medium and 750 l of NS-CM or LPS-CM supplemented with 10 g/ml polymyxin B. Monocyte-derived macrophages were stimulated for 4 h with NS-CM or LPS-CM supplemented with 10 g/ml polymyxin B, and 1.5 ml was added per well of naive monocyte-derived macrophages cultured in a 6-well plate. Cells were incubated for 4 h at 37°C; neutrophils were rotated continuously. RT-PCR was performed as described above.
Whole Cell Extraction and Immunoblotting-Neutrophils (20 ϫ 10 6 ) were stimulated for the indicated time, washed once in ice-cold phosphate-buffered saline, pH 7.4, and resuspended in 400 l of 20 mM imidazole, pH 7.4, 250 mM sucrose, 2.5 mM MgCl 2 , 5 mM EGTA, 10 g/ml leupeptin, 10 g/ml aprotinin, 1 M pepstatin, 1 mM phenylmethylsulfonyl fluoride, 2 mM p-nitrophenyl phosphate, 10 mM ␤-glycerol phosphate, and 200 M sodium orthovanadate. After incubation on ice for 5 min, cell suspensions were sonicated using a Braun-Sonic 200 sonicator twice for 10 s, extracted with 400 mM NaCl for 20 min at 4°C, and clarified by centrifugation at 18,000 ϫ g for 20 min at 4°C. Monocyte-derived macrophages were treated as indicated, washed in phosphate-buffered saline, and lysed in SDS sample buffer. Protein samples were separated by SDS-PAGE, transferred to nitrocellulose, and probed with the indicated antibody.
p38 Kinase Reactions-Whole cell extracts from stimulated cells (250 l or 400-500 g from 20 ϫ 10 6 cells) were supplemented with 750 l of 1% Triton X-100 in 50 mM Tris, pH 7.4, incubated on ice for 10 min, and centrifuged for an additional 10 min. The supernatants from this spin were incubated with anti-p38 for 1 h, and protein A-Sepharose was added for an additional 30 min. The beads were washed, and kinase activity was determined at 30°C in 20 mM Tris, pH 7.4, 10 mM MgCl 2 , 50 M [␥-32 P]ATP (4 Ci/assay), 2 g of GST-ATF-2, 1 mM dithiothreitol, and 10 mM p-nitrophenyl phosphate. Reactions were stopped after 20 min by addition of 2ϫ SDS sample buffer. Phosphorylation of ATF-2 was visualized after SDS-PAGE on a STORM 860 PhosphorImager (Amersham Biosciences), and quantification was performed using ImageQuant (Amersham Biosciences).

Expression of Interferon-stimulated Genes by LPS Is IFNindependent-
We recently demonstrated changes in gene expression in human neutrophils treated with LPS for 4 h using oligonucleotide (32) and cDNA (38) microarray analysis. Unexpectedly, increases of genes were observed that in other systems are associated with activation by the type I interferons, IFN␣ or IFN␤ (49). Further analysis of the time course of expression levels of these ISGs by oligonucleotide array indicated that LPS-induced ISG induction occurred after a prolonged lag (Fig. 1A). Little ISG induction was observed until 4 h of LPS stimulation; examples of the expression of three such genes are shown in Fig. 1A. In contrast, expression of TNFA and IL-1␤, NFB-regulated genes for proinflammatory cytokines known to be expressed in LPS-stimulated neutrophils, increased rapidly and was sustained for 4 h (Fig. 1B).
Expression levels of two ISGs, MX1 and PKR, and those of TNFA and GAPDH were confirmed by RT-PCR (Fig. 1C). Induction of MX1 and PKR paralleled the level of expression determined by oligonucleotide array, with little expression after 2 h but clearly induced 4 h after LPS. In contrast, TNFA expression was evident as early as 30 min after LPS exposure and was maintained at an elevated level throughout the 4 h of LPS stimulation (Fig. 1C). The expression of MX1 and TNFA was dose-dependent and observed at LPS concentrations as low as 1 ng/ml (Fig. 1D). Levels of the ISG MX1 induced by LPS were similar to those induced by IFN␣ (Fig. 1D), a known inducer of MX1, yet IFN␣ caused little enhancement of TNFA levels ( Fig. 1D). Furthermore, the induction of MX1 by LPS was blocked by pretreatment with polymyxin B, an inhibitor of LPS signaling (data not shown).
Several early reports demonstrated the production of type I interferons by LPS-stimulated cells (36,37), a response that may depend on TLR4 (50). To our knowledge LPS is known to stimulate ISGs only through the induction of type I interferons. Therefore, we explored the possibility that LPS induced IFNA or IFNB in neutrophils. In resting neutrophils, RT-PCR analysis indicated a low level of IFNA and nearly undetectable levels of IFNB ( Fig. 2A). After a 4-h exposure to LPS, neutrophils failed to change the transcript level of either cytokine ( Fig. 2A). Furthermore, several IFNA species, IFNB and IFNW, were absent from neutrophils and unchanged by LPS by gene expression analysis (data not shown). Cells responded normally with the induction of MX1 and TNFA.
Although the expression of IFN␣/␤ protein is tightly regulated by IFN gene expression, the possibility exists that neutrophils release preformed IFN protein upon stimulation by LPS. However, IFN␣ and IFN␤ were not detected in the supernatants of resting and LPS-stimulated neutrophils (data not shown). Together, these data indicate that induction of ISGs in LPS-stimulated neutrophils is independent of the transcription and production of type I interferons.

Induction of MX1 Is Independent of STAT Activation and
Release of a Soluble Factor-Although the data above indicate that type I interferons are not produced in LPS-treated neutrophils, the finding that LPS stimulation of neutrophils induces IFN-regulated genes suggested that IFN-stimulated signaling pathways are activated. The interferon-independent induction of ISG56 by dsRNA is regulated in part by STAT1␣ (51). IFN␣/␤ activate the STAT family of transcriptional regulators by Jak1/Tyk2-mediated phosphorylation of STATs on tyrosine (36,37). Whole cell extracts were probed for STAT1, STAT3, and tyrosine-phosphorylated STAT1 and STAT3. STAT1␣ (p90) and STAT3␣ (p92) were the predominant species detected in neutrophils. Exposure of neutrophils to LPS for 30 min and 4 h did not increase the level of phospho-STAT1 isoforms or phospho-STAT3 isoforms (Fig. 2B). Exposure of neutrophils to LPS for 30 min was consistently observed to decrease the basal phosphorylation level of both STAT proteins. In contrast, IFN␣ stimulation of cells led to a robust, time-dependent phosphorylation of both STAT proteins (Fig.  2B). Stimulation of neutrophils with IFN␣, but not LPS, for 4 h resulted in the loss of STAT1 but not STAT3.
Because chemokines can activate STAT proteins (52), we investigated if LPS stimulated the release of a non-interferon factor in neutrophils that mediates ISG induction. Our expression profile data and the data of others (53) (data not shown) indicated that LPS regulates expression of mRNA for oncosta- tin M. This gp130-activating ligand, associated with activation of Jak/STAT pathways, is also released from already synthesized pools (53). However, recombinant human oncostatin M failed to induce MX1 (data not shown). To investigate further the possibility that released factors were responsible for MX1 induction, cell-free supernatants from unstimulated and LPSstimulated neutrophils were added to naive neutrophils for 4 h, and expression of MX1, TNFA, and GAPDH was measured by RT-PCR. Conditioned medium from LPS-stimulated neutrophils induced TNFA message but not that of MX1 (Fig. 2C). We have not characterized further the TNFA-inducing factor released from LPS-stimulated neutrophils. Therefore, although conditioned medium from LPS-stimulated neutrophils is biologically active, it does not contain an ISG-inducing substance. These data indicate that LPS is able to induce MX1 message, but in contrast to other cells types paracrine factors are not sufficient for induction of MX1.
Different Mechanisms of ISG Induction in Neutrophils and Macrophages-In macrophages, IFN␤ production by LPS mediates ISG induction (36,37). In order to clarify the difference between neutrophils and macrophages, we produced human monocyte-derived macrophages and stimulated them with LPS. Monocyte-derived macrophages treated with LPS increased MX1 transcript levels (Fig. 3A). Like neutrophils, resting monocyte-derived macrophages expressed little IFNA, and LPS did not modulate IFNA levels (Fig. 3A), as determined by RT-PCR. However, LPS-stimulated monocyte-derived macro- phages up-regulated IFNB (Fig. 3A), consistent with previous studies (36) with murine macrophages.
The up-regulation of IFNB message in response to LPS suggested that STAT signaling would be activated. To confirm the activation of STAT proteins in LPS-stimulated monocyte-derived macrophages, we measured the levels of phospho-STAT1 and -STAT3. LPS failed to induce phosphorylation of either STAT1 or STAT3 after 30 min of exposure; however, prolonged exposure (4 h) of monocyte-derived macrophages to LPS stimulated the phosphorylation of STAT1 and STAT3 (Fig. 3B). Furthermore, a reproducible decrease in STAT proteins was observed in monocyte-derived macrophages stimulated with LPS for 4 h.
To determine whether IFNB induction and delayed STAT activation are associated with ISG induction in monocyte-derived macrophages, we performed conditioned medium experiments. Media harvested from resting and LPS-stimulated monocyte-derived macrophages were treated with polymyxin B to inactivate LPS, added to naive monocyte-derived macrophages for 4 h, and RT-PCR was used to determine the activity of the conditioned medium. Consistent with the transcriptional increase of IFNB levels and the activation of STAT proteins in LPS-stimulated monocyte-derived macrophages, conditioned medium from LPS-treated monocyte-derived macrophages upregulated MX1 mRNA to an equivalent level as control, LPS-stimulated monocyte-derived macrophages (Fig. 3C). Conditioned medium from LPS-treated monocyte-derived macrophages also demonstrated a weak TNFA-inducing activity; the inactivation of residual LPS by polymyxin B is also demonstrated by the weaker induction of TNFA in conditioned medium than in control, LPS-stimulated monocyte-derived macrophages. Therefore, whereas monocyte-derived macrophages respond to LPS by induction of IFNB, STAT phosphorylation, and transfer of MX1-inducing activity into the supernatant, neutrophils fail to display these traits.

MX1 Induction in Neutrophils Is Sensitive to p38
Inhibition-Our recent report (32) on gene expression in LPS-stimulated neutrophils suggests that the major transcriptional target of the p38 inhibitor, SB203580, is the ISGs. Induction of ISGs was primarily reduced in the presence of the inhibitor, whereas most genes were left unchanged. To explore the signal transduction pathways utilized by LPS for induction of MX1, cells were treated with kinase inhibitors. Pretreatment with the p38 MAP kinase inhibitor SB203580 blocked induction of MX1, as shown by RT-PCR analysis (Fig. 4A), in agreement with oligonucleotide array data (32). The specificity of SB203580 for p38 was supported by the findings that SB202474, an inactive analog of SB203580, failed to inhibit MX1 induction, and another more specific p38 inhibitor, M39 (43), was also active in inhibiting MX1 induction (Fig. 4B). Furthermore, the dose-dependent inhibition of MX1 induction by SB203580 was consistent with inhibition of p38 (data not shown). In contrast to LPS, the expression of MX1 by IFN␣ was not affected by SB203580 (Fig. 4A). Consistent with the lack of effect of SB203580, IFN␣ stimulated little p38 phosphorylation and did so with delayed kinetics (Fig. 4C). However, LPS leads to the rapid and sustained phosphorylation of p38 in neutrophils (Fig. 4B) (28). We have shown previously (27) that phosphorylation of p38 correlates with the activation of p38 kinase in LPS-stimulated neutrophils (data not shown). Levels of p38 protein are not changed by the treatments. An inhibitor of the mitogen-activated protein kinase/extracellular signal-regulated kinase kinase/MAP kinase pathway, PD98059, failed to alter levels of MX1 in LPS-stimulated cells (data not shown). Therefore, the activation of p38 is an important signal for ISG induction by LPS but is dispensable for signaling by IFN␣. Differences in ISG induction described between neutrophils and monocyte-derived macrophages prompted us to explore the role of p38 in LPS-stimulated ISG induction in monocyte-derived macrophages. Expression of MX1 in response to LPS was not altered by preincubation of monocyte-derived macrophages with SB203580 (Fig. 5A). In addition, transcription of TNFA was not inhibited by SB203580. Macrophage-like RAW 264 cells also induced MX1 in response to LPS and were similarly refractory to SB203580 (data not shown). The lack of inhibition by SB203580 of MX1 induction in monocyte-derived macrophages is not due to an inability to activate p38, as LPS treatment leads to the time-dependent phosphorylation of p38 (Fig.  5B), as seen in RAW 264 cells (data not shown) (47). Therefore, these data support alternative pathways for ISG induction by LPS in neutrophils and macrophages.
Protein Synthesis Is Required for MX1 Induction by LPS-The necessity for protein synthesis in the induction of MX1 by LPS was explored using cycloheximide, a general protein synthesis inhibitor. Neutrophils pretreated with cycloheximide and stimulated with LPS did not induce MX1 message, whereas TNFA levels were partially inhibited (data not shown). Thus, whereas LPS does not induce the synthesis of a secreted ISG-inducing factor, synthesis of an intracellular protein may be necessary for this response.
Induction of MX1 Is Specific for LPS and IFN and Is TLR4specific-As shown above, both LPS and IFN␣ stimulate the expression of MX1 message. The ability of other relevant ligands to induce MX1 in neutrophils was determined. The proinflammatory cytokines TNF␣ and IL-1␤, fMet-Leu-Phe, and OSM failed to induce MX1 expression (data not shown).
TLR4 is recognized as the LPS receptor. To determine whether other TLRs can mediate the induction of ISGs, we stimulated neutrophils and monocyte-derived macrophages with peptidoglycan (PGN), a TLR2-specific ligand, and determined gene expression using RT-PCR. In neutrophils, LPS and PGN stimulated the expression of TNFA, demonstrating the biological activity of PGN on neutrophils (Fig. 6A). However, SB203580 had no effect on the expression of TNFA, indicating the inhibition of MX1 expression in LPS-stimulated neutrophils was specific. In contrast to the effect of LPS on neutrophils, PGN was unable to induce expression of MX1 (Fig. 6A). As shown previously, SB203580 inhibited MX1 induction by LPS, but SB203580 did not modify the expression profile induced by PGN. Similarly, lipoteichoic acid, another TLR2-activating ligand, failed to induce MX1 (data not shown). Likewise, PGN-stimulated monocyte-derived macrophages did not express MX1 (Fig. 6B), although TNFA was up-regulated by both LPS and PGN. These data indicate that ISG induction is dependent on TLR4-specific signaling events.
The ability of SB203580 to block ISG induction in LPSstimulated neutrophils indicates that p38 plays a role and suggests that the inability of PGN to induce ISG expression might be explained, at least in part, by inadequate p38 activation. Therefore, we determined p38 activity in response to PGN. PGN enhanced p38 activity in a time-dependent manner, with p38 activity observed after 10 min and maximal activity at 40 min; p38 activity declined by 60 min post-stimulation (Fig.  6C). However, the extent of PGN-stimulated p38 activity was less than that observed with LPS (Fig. 6C), and the time course of PGN-stimulated p38 activation was delayed compared with LPS. DISCUSSION We previously identified ISGs as p38-dependent gene targets in LPS-stimulated neutrophils (32). The present study was undertaken to clarify the role of p38 in LPS-stimulated ISG induction and the mechanism of this response. The delayed expression of ISGs in LPS-stimulated neutrophils suggested that LPS was activating a secondary, indirect response, as had been observed in macrophage(-like) cells (36,37). However, several lines of evidence point toward a mechanism in neutrophils distinct from those known to be involved in the expression of ISGs in macrophages. First, LPS did not induce expression of IFNA or IFNB mRNA, as determined by RT-PCR and gene expression analysis. Furthermore, secretion of IFN␣ or IFN␤ was not observed in the supernatants of LPS-treated neutrophils. Second, phosphorylation of STAT proteins, a well characterized marker of STAT activation, was not observed in neutrophils stimulated with LPS. Finally, the supernatant from LPS-stimulated neutrophils was unable to support induction of MX1, indicating that a paracrine mechanism was not responsible. In contrast, LPS stimulation in other cell types is associated with interferon-dependent induction of ISGs and signaling events (36,37).
LPS is well known to enhance the expression of type I IFNs in macrophages, particularly of IFN␤ (36,37). Data presented here support the enhanced expression of the IFNB gene by LPS, enhanced STAT phosphorylation that is delayed with respect to STAT phosphorylation in IFN␣-treated monocytederived macrophages, and the ability of the supernatant from LPS-stimulated monocyte-derived macrophages to support MX1 induction in naive monocyte-derived macrophages. That LPS stimulates induction of ISGs in neutrophils in the absence of type I interferons or interferon-dependent signaling has not, to our knowledge, been observed previously. Therefore, a clear distinction exists between induction of ISGs in neutrophils and monocyte-derived macrophages after exposure to LPS. We conclude that in neutrophils LPS activates a novel and intrinsic ISG-inducing signaling pathway. Interferon-independent induction of ISGs can occur in response to dsRNA, although the dependence on IFN␣/␤ is cell type-dependent (40,51,54).
Although non-interferon ligands can tyrosine-phosphorylate and activate STAT proteins (52), we have found no evidence for this in LPS-stimulated neutrophils. Another post-translational modification of STAT1, arginine methylation, was reported to be important for STAT1 transactivation (55). However, the observation that tyrosine phosphorylation of STATs is still required for transactivation and translocation to the nucleus suggests that STAT activation by these additional mechanisms is not responsible for induction of ISGs. Together with the lack of detectable type I interferon release or the secretion of a paracrine factor, the failure of LPS to enhance STAT phosphorylation in neutrophils points toward a unique mechanism of ISG induction. This finding is reminiscent of the observation that certain viruses and dsRNA induce ISGs in the absence of IFN secretion and independently of STAT activation (40,41). Induction of ISGs by virus is dependent on members of the IRF family of transcriptional regulators including IRF3, IRF5, and IRF7 (34,35,56). IRF family members are phosphorylated in response to virus on C-terminal Ser residues (48, 56 -59), a modification necessary for transactivation (60), and identified by a retardation of electrophoretic mobility of IRF3 (60). These investigators were unable to observe activation of IRF3 by LPS in numerous cell types. However, LPS does alter the electrophoretic mobility of IRF3 in neutrophils, 2 suggesting the involvement of IRF3 in LPS signal transduction in neutrophils, but this gel shift is not altered by SB203580. 2 A recent publication has indicated the importance of IRF3 and NFB in the induction of ISGs by LPS (61). However, our studies differ in several respects. Doyle et al. (61) describe a set of early "interferon-stimulated genes" whose induction is dependent on NFB, including IFNB. The early response is additionally modulated by IRF3; however, subsequent late ISG induction (including that of MX1) is dependent on IFN␤ (61) as seen in virus-infected cells (58). Due to complications in genetically manipulating neutrophils, we are unable at this time to modulate this pathway. Further experiments are necessary to resolve the role of IRF3, and other IRF family members, in the transcriptional response of neutrophils to LPS. Although we have not determined the importance of NFB on the response in human neutrophils, IFN␤ production does not appear to be involved. However, it is possible that the protein synthesissensitive factor is a newly synthesized NFB-dependent gene (see below).
LPS Induction of ISGs Is p38-dependent-The p38 pathway has a role in cytokine production, adhesion, and migration of LPS-stimulated neutrophils (27,28,30,31,62). The finding that transcription factors are substrates for p38 (or p38-dependent kinases) implicates p38 in the regulation of gene expression (26). Furthermore, STAT1 phosphorylation on Ser-727, a modification necessary for the transactivation potential of STAT1, is dependent on p38 in some cell types (63). However, p38 inhibition with SB203580 had no effect on IFN␣-stimulated MX1 induction in neutrophils, further indicating that p38 activation and STAT regulation are independent signaling events in neutrophils. IFN␣ is a weak inducer of p38 phospho-rylation in neutrophils, which again indicates that the LPS effect on ISG induction differs from that stimulated by IFN␣. p38 has been implicated in the induction of ISG54 in response to LPS in astrocytes, and IRF3 was shown to translocate to the nucleus after several hours of LPS exposure (64); however, these investigators did not link IRF3 activation to p38 activity or comment on IFN␣/␤ production.
A definite role of p38 is obscured by the possibility of other targets for SB203580, including Raf, JNK2␣, TGFBR-I and -II, and Lck (43,65). However, the IC 50 values of SB203580 for Lck (20 M) and TGFBR-I (40 M) are greater than that used in these studies (10 M), and a JNK inhibitor, SP600125, did not alter MX1 gene expression (data not shown). Whereas Raf is inhibited in vitro at a lower concentration of SB203580 (IC 50 2 M), no inhibition is evident in vivo (66), and an inhibitor of the Raf substrate mitogen-activated protein kinase/extracellular signal-regulated kinase kinase, PD98059, has no effect on MX1 induction by LPS. Finally, SB202474, an inactive analog of SB203580, was ineffective at blocking MX1 induction, whereas a more specific inhibitor of p38, M39 (43), also was effective at inhibiting MX1 induction. Although it is necessary at present to use inhibitors to modulate kinases in neutrophils, the data indicate that SB203580 is acting by the inhibition of p38.
The p38 kinase pathway may also regulate the activity of other components necessary for the transcriptional complex, such as a non-IRF transcription factor and/or co-activators. Kinase pathways are also known to modify chromatin and thus are implicated in chromatin remodeling. p38 phosphorylates histone H3 on Ser-28 (67), and the p38-activated kinase MSK-1 phosphorylates H3 on Ser-10 and HMG-14 (68). Recently, nucleosome remodeling was shown to occur as a consequence of TLR2 activation (69). A role for p38 in histone acetylation, another chromatin modification, has not been described, although ATF-2, a p38 substrate, displays histone acetyltransferase activity (70). At present, we cannot distinguish the mechanism(s) by which SB203580 inhibits ISG expression.
ISG Induction by LPS Requires Protein Synthesis-The lack of a secreted ISG-inducing factor from LPS-stimulated neutrophils and the sensitivity of this response to cycloheximide suggest that the synthesis of an intracellular protein is required. One candidate is IRF7, a protein induced by IFN␣/␤ and virus (71,72) and necessary for a full genetic response to infection (71)(72)(73). However, we were unable to detect IRF7 under any conditions in neutrophils. Alternatively, the maintenance of a constitutively expressed, but labile, protein may be the target of cycloheximide. The interferon-independent induction of ISGs by dsRNA is generally independent of de novo protein synthesis (54,74). The known inhibitory effect of SB203580 on cytokine synthesis supports the possibility that SB203580 and cycloheximide act to inhibit production of the same factor.
The Induction of ISG by LPS in Neutrophils Occurs through the Activation of p38 and TLR4 but Not TLR2-Stimulation of neutrophils with a TLR2-activating ligand, PGN, had no effect on transcript levels of MX1; however, PGN was active, as demonstrated by the induction of TNFA. A similar observation was made recently in murine macrophages (50). In addition, the induction of MX1 in human monocyte-derived macrophages is also mediated by TLR4 but not TLR2. Therefore, induction of ISGs is confined to activation of a subset of TLR family members. However, the mechanisms by which MX1 induction occurs is different between the two cell types. Induction of MX1 is sensitive to SB203580 only in neutrophils. Although PGN activates p38 in neutrophils, it is to a lesser degree than does LPS. This can be interpreted to suggest that the level or duration of p38 activity by PGN and TLR2 is insufficient to support ISG induction; however, whereas low LPS concentrations in-duced MX1 expression (Fig. 1C), this dose of LPS activated p38 poorly (28). Thus, our findings do not exclude the possibility that a non-p38, TLR4-specific pathway is required and that TLR2, while activating p38 sufficiently, does not activate this putative second signal. Although p38 plays a necessary role in LPS-induced ISG expression in neutrophils, p38 activation may not be sufficient for ISG expression. As described above and suggested previously (61) (data not shown), IRF3 may fulfill this role.
Indirect evidence supports a role for neutrophils in antiviral defense. Viral infection promotes neutrophilic inflammation (75,76), and neutrophils express anti-viral defensins (77). Neutrophils exposed to LPS also synthesize and secrete large quantities of the CC-chemokines MIP-1␣ and MIP-1␤ that may compete with viral co-receptors. Macrophages exposed to LPS display antiviral activity to human immunodeficiency virus by secretion of MIP-1␣ and MIP-1␤ (78,79) and down-regulation of CCR5 (80). The induction of antiviral ISGs by LPS described here suggests that exposure of neutrophils to bacterial infection primes neutrophils for an antiviral role. However, this study shows that macrophages and neutrophils have different mechanisms of ISG induction by LPS, suggesting that antiviral mechanisms may also differ. The finding that LPS induces antiviral ISGs suggests that anti-bacterial pathways and antiviral pathways are linked in neutrophils. Alternatively, ISG products could function in the more established antibacterial activity of neutrophils. Recently, the F protein of respiratory syncytial virus was shown to recognize TLR4 (81); whether this affected induction of ISGs was not determined. This may not represent a universal mechanism of antiviral response because many viruses require internalization and replication before ISGs are induced (41,82).
In summary, gene expression analysis of LPS-stimulated neutrophils has led to the detection of an unexpected set of antiviral genes (32,38). In contrast to other cell types (36,37,83), such as macrophages, the LPS induction of ISGs in neutrophils does not involve secretion of mediators, including IFN␣/␤, or STAT activation. Furthermore, the signal transduction pathways necessary for induction of ISGs in neutrophils involves p38, is activated by a TLR4 ligand, and is not activated by a TLR2 ligand. Together, our results suggest a new mechanism for neutrophils in antiviral defense and that novel signaling pathways are activated in response to LPS, which differ substantially from pathways activated by LPS in macrophages. These studies may aid our understanding of the mechanisms of TLR activation and of innate immunity.