IL-1 receptor antagonist
interferon regulatory factor 3
myxovirus (influenza virus) resistance 1
- PT-real time PCR:
primary transcript real time PCR
The CXCL10 chemokine is a critical chemoattractant for the recruitment of activated Th1 and NK cells into inflammatory sites. CXCL10 is typically produced by myeloid cells in response to IFN-γ, as well as by neutrophils, though the latter require a costimulation with IFN-γ and LPS. In this study, we investigated the molecular mechanism(s) whereby IFN-γ and TLR4 ligation synergize to induce CXCL10 expression in neutrophils. By primary transcript real-time PCR analysis, we demonstrate that the CXCL10 gene is transcriptionally induced by the LPS plus IFN-γ combination in neutrophils, consistent with previous studies showing that increased CXCL10 gene expression does not reflect enhanced mRNA stability. The IFN-γ-induced STAT1 activation and the lipopolysaccharide (LPS)-induced NF-κB activation were not enhanced if neutrophils were exposed to both stimuli, whereas both transcription factors were activated by IFN-γ or LPS in monocytes. Finally, pharmacological inhibitors of NF-κB demonstrated its role in the induction of CXCL10 expression by LPS plus IFN-γ in neutrophils, and by LPS or IFN-γ in monocytes. Together, these results suggest that in neutrophils, the synergy observed between LPS and IFN-γ toward CXCL10 gene expression likely reflects the cooperative induction of the NF-κB and STAT1 transcription factors by LPS and IFN-γ, respectively.
CXCL10/IFN-γ-inducible protein 10 (IP-10) is a critical component of several immune responses, acting notably as a chemoattractant for activated NK and Th1 cells into sites of inflammation. CXCL10 mediates its various activities by binding to CXCR3 expressed on activated Th1 and NK cells 1, and influences the behavior of nonimmune cells that express a CXCR3. For instance, it inhibits endothelial cell proliferation through interfering with the function of receptors for the growth factors, basic fibroblast growth factor and vascular endothelial growth factor, resulting in the inhibition of angiogenesis 2.
Neutrophils are cells of the innate immune system that play an essential role in antimicrobial defense via the release of numerous toxic mediators 3. In addition, extensive research has now clearly established that the release of chemokines and cytokines constitutes yet another key contribution of neutrophils to innate immunity (reviewed in reference 4). Among the chemokines produced by neutrophils both in vitro and in vivo, CXCL10 stands out as being somewhat peculiar in terms of its induction characteristics, as originally shown by us 5, 6 as well as by other groups 7–9. Indeed, CXCL10 expression occurs in vitro only if neutrophils are costimulated with IFN-γ along with LPS, but not with IFN-γ or LPS alone, contrary to what happens in monocytes, macrophages, dendritic cells and other cell types 5, 6, 10. Our studies additionally revealed that neutrophils can also express and release CXCL10 in response to IFN-β in combination with LPS, or to IFN-γ together with either TNF-α or IL-1β 6. However, the molecular bases of the particular inducibility of CXCL10 in human neutrophils have only been partially elucidated thus far. In this regard, we have recently clarified that the reason explaining the inability of LPS to induce CXCL10 mRNA expression in neutrophils reflects the fact that in this cell type, LPS is unable to mobilize the intracellular MyD88-independent pathway upon TLR4 binding 11. As a result, LPS fails to induce IFN-β production and release in neutrophils, thereby preventing the autocrine/paracrine induction of STAT1 tyrosine phosphorylation that is otherwise typically triggered by the MyD88-independent pathway in other cell types, such as autologous monocytes, and which is ultimately responsible for the activation of a host of antimicrobial and antiviral genes, including CXCL10 10. Finally, we recently reported that IFN-γ (whether alone or together with LPS) is unable to drive the production of IFN-β in neutrophils, thereby establishing that the induction of CXCL10 expression and release in response to IFN-γ and LPS occurs directly, i.e. without the need for the production of endogenous IFN-β 11.
In this study, we investigated the molecular mechanism(s) by which IFN-γ and TLR4 ligation synergize to induce CXCL10 mRNA expression in neutrophils. We now report that the synergy observed between LPS and IFN-γ toward CXCL10 gene expression likely reflects the cooperative induction of the NF-κB and STAT1 transcription factors by LPS and IFN-γ, respectively.
Effect of Ultrapure LPS on CXCL10 expression in human neutrophils
Our previous studies on CXCL10 expression 5, 6 used a commercial LPS preparation that may have been contaminated with TLR2-activating agonists 12. It was therefore necessary to perform experiments with Ultrapure LPS that only activates neutrophils via TLR4. The latter experiments 11 allowed us to confirm that our earlier reports on the effects of LPS on neutrophil and monocyte mRNA expression for CXCL10, TNF α, CCL4, IL-1 receptor antagonist (IL-1ra) and IL-12p40 were unequivocally mediated via TLR4 only. Further investigation shown herein (Fig. 1), supported by additional time-course and dose-response experiments (not shown), confirmed that the effects of Ultrapure LPS (used with or without IFN-γ) on the release of CXCL10, TNF-α and IL-1ra by neutrophils were similar to those exerted not only by the original batch of LPS (from Escherichia coli, 0111:B4 strain) from which it was ultra-purified, but also by the same LPS preparations (from E. coli, 026:B6 strain) used in our previous publications 5, 6. Accordingly, real-time PCR studies confirmed that the induction of CXCL10 (Fig. 2A and B), TNF-α and IL-1ra (Fig. 2B) mRNA mediated either by the Ultrapure LPS or by the two batches of unpurified LPS, used alone or in combination with IFN-γ, were qualitatively and quantitatively comparable, both in neutrophils and autologous monocytes. Only the release of CXCL8 proved to be less potently induced by the Ultrapure LPS as compared to the two unpurified LPS preparations (Fig. 1), confirming earlier observations that the CXCL8 response by unpurified LPS-stimulated neutrophils is partially due to the contaminating TLR2-specific ligands 13, 14. Because similar results were observed with a second Ultrapure LPS (from E. coli, 0111:B4 strain), purchased from a different supplier (see Materials and methods), we used Ultrapure LPS in all subsequent experiments from this point onwards.
IFN-γ plus LPS-mediated induction of CXCL10 mRNA is dependent on NF-κB
We have previously observed that the induction of CXCL10 mRNA expression in neutrophils costimulated with IFN-γ and LPS occurs directly, i.e. without the need for the production of endogenous mediators such as TNF-α 5 or IFN-β 11. In addition, other studies had already excluded that the LPS-mediated potentiation of CXCL10 mRNA accumulation in IFN-γ-treated neutrophils might reflect changes in the stability of CXCL10 transcripts 5. Conversely, our primary transcript real-time PCR (PT-real time PCR) studies confirmed that the inducible accumulation of CXCL10 mRNA transcripts in IFN-γ- or IFN-γ plus LPS-treated neutrophils and monocytes specifically involves transcriptional events (Fig. 3). These observations prompted us to investigate the role of the transcription factors that are known to be critically involved in the regulation of CXCL10 — in particular STAT1, but also NF-κB 15, and (to a lesser extent) interferon regulatory factor 3 (IRF-3) 16 and activating transcription factor (ATF-)2/c-Jun 17. In keeping with the inability of neutrophils to produce IFN-β, we have already shown that the activation of IRF-3 is not inducible in IFN-γ plus LPS-treated neutrophils 11. We have also shown that neutrophils express little or no c-Jun, a key component of AP-1 complexes, and that consequently, AP-1 is not inducible by LPS and/or IFN-γ in these cells 11, 18. Consequently, we reasoned that a possible mechanism leading to the stronger induction of CXCL10 gene by the LPS plus IFN-γ combination in neutrophils could be a synergistic enhancement of IFN-γ-induced STAT1 tyrosine phosphorylation and, in turn, STAT1-containing DNA-binding activities exerted by LPS. This, however, proved to be wrong, since IFN-γ alone already triggers the tyrosine phosphorylation of p91 STAT1α (and to a much lesser extent, that of the modestly expressed p84 STAT1β) (Fig. 4A), and already induces STAT1-containing DNA-binding activities at optimal levels (Fig. 4B) in both neutrophils and monocytes. The combined effect of IFN-γ and LPS did not appear to involve other STAT proteins, insofar as these stimuli do not synergize to induce the tyrosine phosphorylation or DNA-binding activity of STAT3 and STAT5, while neither IFN-γ nor LPS induces the tyrosine phosphorylation of STAT4 or STAT6 (our unpublished data). Similarly, we did not find any increased activation of NF-κB DNA-binding activities in IFN-γ plus LPS-treated neutrophils as compared to cells stimulated with LPS only 11. Moreover, no significant NF-κB activation was observed in IFN-γ-treated neutrophils 11, 19. However, Fig. 5A shows that three NF-κB inhibitors acting by unrelated mechanisms 20–22 profoundly inhibited the induction of CXCL10 expression, thereby establishing a crucial role for this transcription factor in CXCL10 induction in neutrophils. These inhibitors were reasonably specific, as they did not affect the expression of the myxovirus (influenza virus) resistance 1 (MX1) gene under the same conditions (Fig. 5A). A similar inhibition of CXCL10 gene expression was observed when neutrophils were pretreated with other NF-κB blockers, such as SC-514, MG-262, 15-deoxy-PGJ2, or BAY 117082 (not shown). The various NF-κB blockers exerted essentially the same inhibitory effect on CXCL10 (but not on IFN-β) gene expression in LPS-stimulated monocytes (Fig. 5A), similar to previously reported observations 23, indicating that in monocytes, LPS directly induces CXCL10 via activation of NF-κB, or alternatively, that endogenous IFN-β elicits the induction of CXCL10 mRNA via activation of NF-κB, besides STAT1 10. Even though we cannot exclude the former phenomenon, the latter turned out to be true, as demonstrated by the inhibitory action of MG132 on the induction of CXCL10 mRNA mediated by IFN-β in human monocytes (Fig. 5B). In addition, worthy of note is that the up-regulatory effect of IFN-γ on CXCL10 (but not on CD64) mRNA expression was dramatically suppressed by both MG132 (Fig. 5B) and PTDC (not shown) in human monocytes. This is consistent with the capacity of both IFN-γ and IFN-β to elicit a fairly strong but transient NF-κB activation in these cells (Fig. 5C). Taken together, these data reveal that NF-κB activation is necessary for the induction of CXCL10 gene expression not only in IFN-γ plus LPS-treated neutrophils, but also in LPS-, IFN-γ- and IFN-β-stimulated monocytes.
In this study, we investigated the molecular mechanism(s) by which IFN-γ and TLR4 ligation by LPS synergize to induce CXCL10 mRNA expression in neutrophils. We had previously excluded that, under these conditions, CXCL10 induction does rely on endogenous TNF-α 5, on endogenous IFN-β 11, or on de novo protein synthesis 11. Herein, we show that CXCL10 induction is regulated at the transcriptional level. Although the CXCL10 gene represents a classical example of a gene that responds to type I/type II IFN- in a STAT1-dependent fashion 24, 25, our results show that, in neutrophils, its induction does not reflect a synergistic enhancement of STAT1 activation by IFN-γ and LPS. Furthermore, we previously showed that there is no evidence for any IRF-3 or ATF-2/c-Jun involvement regulating CXCL10 mRNA expression in IFN-γ plus LPS-treated neutrophils 11, consistent with the data excluding endogenous IFN-β as an intermediate regulating CXCL10 induction. Despite the demonstrated lack of AP-1 activation in neutrophils, we showed herein that the JNK pathway is nevertheless mobilized by LPS in these cells, with the shorter (p46) JNK isoforms being predominantly phosphorylated, whereas the longer (p54) JNK isoforms were predominantly phosphorylated in monocytes (Fig. 4A). Because our phospho-JNK (P-JNK) antibody recognizes both the long and short forms of JNK1, JNK2, and JNK3, and because it still is not clear whether neutrophils express JNK2 and JNK3, the actual identity of the p46 and p54 P-JNK bands we observed in neutrophils and monocytes remains to be ascertained. In any instance, a role for the JNK pathway in CXCL10 induction in neutrophils is unlikely, in view of our previous demonstration that JNK inhibition does not affect the onset of CXCL10 gene expression in these cells 11, 19.
Conversely, pharmacological inhibition of NF-κB activation by seven different NF-κB blockers markedly decreased the induction of CXCL10 mRNA in IFN-γ plus LPS-treated neutrophils, without affecting the induction of MX1 or CD64, whose induction is known to be κB independent. In view of our findings that LPS (but not IFN-γ) activates NF-κB in neutrophils 11, 19, our data indicate that the synergy between LPS and IFN-γ toward CXCL10 gene expression must reflect the cooperative induction of the NF-κB and STAT1 transcription factors by LPS and IFN-γ, respectively (Fig. 6). Such a scenario would also be consistent with the known ability of TNF-α to synergize with IFN-γ in inducing CXCL10 mRNA in neutrophils 5, 6 and other cell types 15, as well as with the known ability of TNF-α to activate NF-κB in neutrophils 19. It is also consistent with the identification of the mechanism underlying the marked synergy between LPS and IFN-γ on CXCL9 mRNA expression in mouse macrophages, which was shown to be the result of LPS-induced NF-κB and IFN-γ-induced STAT1 26. The synergy was not dependent on new protein synthesis, was independent of TNFα, and occurred at the level of gene transcription 26. On a related note, the use of NF-κB inhibitors also enabled us to demonstrate that NF-κB activation is necessary for the induction of CXCL10 and other type I/II IFN-dependent genes in monocytes (i.e. ISG15, IFIT1 and MX1, unpublished observations). Accordingly, both IFN-γ and IFN-β were shown to be able to rapidly and transiently activate NF-κB in monocytes, as demonstrated in this study and consistent with the requirement of NF-κB activity for the inducible expression of CXCL9 and CXCL10 genes by IFN-γ in human tumor cell lines 27, 28. This effect of type I and type II IFN towards NF-κB activation also helps to understand why the induction of CXCL10 mRNA/production is much weaker in human neutrophils stimulated with IFN-γ or IFN-β alone than in IFN-γ- or IFN-β-treated monocytes. Indeed, this is likely because only the STAT1-dependent pathway is mobilized in IFN--treated neutrophils, thereby leaving out the contribution of the NF-κB pathway, which is important for optimal CXCL10 induction (Fig. 6). Finally, our data do not exclude that the induction of CXCL10 mRNA expression by IFN-γ and LPS in neutrophils might also be favored because it is not counter regulated by attenuators of cytokine-induced responses, such as the SOCS proteins 29. Indeed, our previous observations have shown that in contrast to monocytes, neither SOCS-1 nor SOCS-3 protein expression is induced by LPS in neutrophils 11, 30. Conversely, it appears that monocyte responses are attenuated by LPS-activated SOCS-1/SOCS-3, since the induction of CXCL10 mRNA expression by IFN-γ plus LPS is much lower than that exerted by IFN-γ alone 6. Interestingly, the reason why SOCS-1 is not inducible by LPS in neutrophils derives from the lack of mobilization of the MyD88-independent pathway, which controls SOCS-1 gene via endogenous IFN-β 11. By comparison, SOCS-3 mRNA is induced in LPS-treated neutrophils, but not the related protein 30. It is also interesting to mention that, in preliminary experiments, neutrophils preincubated with LPS for up to 6 h prior to the addition of IFN-γ for further 18 h, or vice versa, produced less CXCL10 than neutrophils costimulated with IFN-γ plus LPS for the same overall time. Taken together, these data suggest that SOCS proteins might not be the only regulators directly or indirectly controlling CXCL10 mRNA expression, stressing the necessity of further studies to clarify this issue. Whatever the case may be, our current findings clearly indicate that the mechanisms governing the expression of CXCL10 in neutrophils and monocytes are cell type-specific and under the control of distinct regulatory pathways. These differences between primary monocytes and neutrophils also illustrate the relevance of studying the expression of physiologically relevant genes in primary cell types.
Materials and methods
Cell purification and culture
Highly purified granulocytes (neutrophils >96.5%, eosinophils <3%) and Percoll-purified monocytes were isolated and prepared under endotoxin-free conditions from buffy coats of healthy donors as previously described 11, 31. In some experiments, monocytes were also isolated by negative selection (to >95% purity) using MACS® separation (Monocyte Isolation kit II, from Miltenyi Biotec, Bergisch Gladbach, Germany). Immediately after purification, neutrophils and monocytes were suspended in RPMI-1640 medium supplemented with 10% low endotoxin FBS (<0.5 EU/mL, BioWhittaker, Verviers, Belgium), treated with or without 100 ng/mL Ultrapure E. coli LPS (0111:B4 strain, from Invivogen, San Diego, CA or from Alexis, Lausen, Switzerland) and/or 100 U/mL IFN-γ (R&D Systems, Minneapolis, MN) or 1000 U/mL IFN-β [Betaferon(R), Schering, Berlin, Germany], and then plated either in 6/24-well tissue culture plates (Nunc, Roskilde, Denmark), or in polystyrene flasks (Orange, Trasadingen, Switzerland) before culture at 37°C, 5% CO2 atmosphere. In selected experiments, neutrophils and monocytes were also incubated with standard unpurified E. coli LPS strains (026:B6 from Sigma, or 0111:B4 from Invivogen). In other experiments, neutrophils and monocytes were preincubated for 30 min with different NF-κB inhibitors, including 10 µM MG-132, 3 µM MG-262, 30 µM 15-deoxy-PGJ2, 300 μM pyrrolidine dithiocarbamate (PDTC) (all from Sigma), 5 μM BAY 117082, 100 μM SC-514 (Calbiochem, San Diego, CA), 100 μg/mL SN50 peptide and its related control (BIOMOL Plymouth Meeting, PA). After the desired incubation period, cells were collected and spun at 300 × g for 5 min. The resulting supernatants were immediately frozen in liquid nitrogen and stored at –80°C, while the corresponding pellets were either extracted for total RNA or lysed for protein analysis as described below. All reagents used were of the highest available grade and were dissolved in pyrogen-free water for clinical use 31.
Analysis of mediator concentration
Cytokine concentrations in cell-free supernatants were measured by specific ELISA kits for CXCL10, IL-1ra and TNFα (R&D Systems) and CXCL8 (Euroclone, Wetherby West Yorkshire, UK). Detection limits were 15 pg/mL for CXCL8, CXCL10 and TNF-α and 50 pg/mL for IL-1ra.
Real-time PCR and PT real time PCR studies
Real-time PCR was performed and analyzed exactly as previously described 32 using 1 μg total RNA (usually extracted from 107 neutrophils or 2–106 monocytes), with gene-specific primers (purchased from Invitrogen, Carlsbad, CA) available in the public database RTPrimerDB (http://medgen.UGent.be/rtprimerdb/) under the following entry codes: β2m (3534), CD64 (3536), CXCL10/IP-10 (3537), CXCL8/IL-8 (3553), GAPDH (3539), IFIT1 (3540), IFN-β (3542), IL-12p40 (3543), IL-1ra (3544), IL-6 (3545), MX1 (3549), TNF-α (3551). The reaction conditions were identical for all primer sets, as it follows: 50°C for 2 min, 95°C for 2 min, and then 40 cycles of 95°C for 15 s and 60°C for 1 min. β2m was selected as a normalizing gene, according to its stable expression levels in leukocytes 14. Data were calculated with Q-Gene software (www.BioTechniques.com) and are expressed as mean normalized expression (MNE) units after β2m normalization. PT-real time PCR was conducted essentially like real time PCR, according to the protocol previously described 33. The only methodological modifications were: (i) a second DNase I treatment of the RNA samples, before reverse transcription; (ii) the use of primers designed from an intron region of the selected genes (see public database RTPrimerDB (http://medgen.UGent.be/rtprimerdb/) under the following entry codes: PT-CXCL10 (3538) and β2m (3534). The same RNA samples were processed in the absence of reverse transcriptase and served as controls for genomic DNA contamination.
Cellular and nuclear extracts
After stimulation for the indicated times, neutrophils (0.8–1 × 108/condition) were diluted in ice-cold PBS and centrifuged twice at 300 × g for 5 min at 4°C. The cells were then suspended in relaxation buffer containing anti-phosphatase and anti-protease cocktails and disrupted in a nitrogen bomb (Parr Instruments, Mobile, IL) to prepare nuclear and cytoplasmic extracts, exactly as described 19. Nuclear extracts from neutrophils and monocytes were also prepared using a modification of Dignam et al. method 34, as previously described 19. Alternatively, cellular extracts were prepared by lysis with RIPA buffer [(25 mM Tris pH 7.5, 150 mM NaCl, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS), containing 1 mM DTT and anti-protease and anti-phosphatase cocktails]. Nuclear, cytoplasmic and whole cell extracts were frozen and stored at –80°C. Small aliquots of the various samples were routinely processed for protein content determination, by using a protein assay kit (Bio-Rad, Hercules, CA.
For Western blot analysis, nuclear, cytoplasmic or whole cell extracts were subjected to immunoblots by standard procedures. Nitrocellulose membranes were firstly blocked for 1 h at room temperature in TBS/T (20 mM Tris-HCl pH 7.6, 150 mM NaCl, 0.1% Tween 20) containing 5% BSA, and then incubated overnight at 4°C, in the presence of specific primary antibodies in the same buffer. Antibody binding was detected by using horseradish peroxidase-conjugated anti-mouse or anti-rabbit IgG (Amersham), and revealed using the enhanced chemiluminescence system (ECL, Amersham), according to the manufacturer's instructions. Antibodies against phospho-tyrosine STAT1 (#9171) and phospho-tyrosine JNK (#4671) were purchased from Cell Signaling Technologies (Beverly, MA, USA), whereas mAb for native JNK1 were purchased from Santa Cruz.
Transcription factor binding analyses were performed by incubating nuclear extracts in binding buffer in the presence of labeled oligonucleotide probes, exactly as described 35.
Data are expressed as means ± SE. Statistical evaluation was performed by the Student's t-test for paired data and considered to be significant if p <0.05.
This work was supported by grants from Ministero dell'Istruzione, dell'Università e della Ricerca (PRIN 2005, FIRB and 60%), Fondazione Cassa di Risparmio di Verona-Vicenza-Ancona e Belluno, Associazione Italiana per la Ricerca sul Cancro (AIRC) and the Canadian Arthritis Society. P. P. McDonald is a Scholar of the Fonds de la recherche en santé du Québec.