IL-10 is a potent anti-inflammatory cytokine and inhibitor of TNF-α production. The molecular pathways by which IL-10 inhibits TNF-α production are obscure, with diverse mechanisms having been published. In this study, a new approach has been taken for the study of human cells. Adenovirus was used to deliver TNF-α promoter-based luciferase reporter genes to primary human monocytic cells. The reporter genes were highly responsive to macrophage activation and appeared to mirror the behavior of the endogenous TNF-α gene. When added, either with or after the stimulus, IL-10 required the 3′ untranslated region of the TNF-α gene to inhibit luciferase mRNA and protein expression, indicating a posttranscriptional mechanism. However, if macrophages were incubated with IL-10 before activation, inhibition of gene expression was also mediated by the 5′ promoter, suggesting a transcriptional mechanism. To our knowledge, this is the first time that a dual mechanism for IL-10 function has been demonstrated. Studies to elucidate the mechanisms underlying the inhibition of TNF-α production addressed the effect of IL-10 on the activation of p38 mitogen-activated protein kinase and NF-κB. However, these studies could demonstrate no requirement for the inhibition of p38 mitogen-activated protein kinase or NF-κB activation as potential mechanisms. Overall, these results may explain the diversity previously ascribed to the complex mechanisms of IL-10 anti-inflammatory activity.
Inflammation is an essential host response to infectious challenge. However, when excessive, the inflammatory response leads to harmful, or even fatal, consequences as seen in rheumatoid arthritis, Crohn’s disease, and septic shock (1). Thus, the inflammatory response must normally be tightly regulated. Among the many factors that suppress the inflammatory response, the cytokine IL-10 is one of the most important. IL-10 is a potent inhibitor of monocyte/macrophage activation, blocking the expression of TNF-α and other proinflammatory mediators. Mice defective in IL-10 expression develop an inflammatory Crohn’s-like disease and produce enhanced amounts of TNF-α in response to LPS (2). Furthermore, in murine models, many inflammatory diseases are ameliorated by administration of exogenous IL-10 (reviewed by Donnelly et al. in Ref. 3).
Despite considerable scientific and clinical interest, the molecular pathways underlying IL-10 inhibition of TNF-α expression remain obscure. In human PBMC, IL-10 has been reported to suppress TNF-α gene transcription (4, 5), possibly by inhibiting the activation of the transcription factor NF-κB (6, 7). However, unlike the inhibition of TNF-α gene transcription (4, 5), the blockade of NF-κB activation does not appear to require IL-10-induced de novo protein synthesis (6, 7). In contrast, studies performed in murine macrophages have claimed that IL-10 acts through the posttranscriptional mechanisms by destabilizing TNF-α mRNA (8, 9), or, more recently, by inhibiting gene translation via blocking the activation of p38 mitogen-activated protein kinase (MAPK)5 (10). A role for IL-10-induced de novo protein synthesis has also been described in murine macrophages (9). Furthermore, Riley et al. (11) have demonstrated the absolute requirement of the transcription factor STAT-3 in mediating the anti-inflammatory effects of IL-10. There appears to be no simple explanation for the variety of mechanisms ascribed to IL-10-mediated inhibition of TNF-α expression except for the different systems used. Thus, it may be possible that there are major differences between cell systems in how IL-10 exerts its effect.
For understanding human physiology and disease, a most appropriate system for the study of IL-10 effects is primary human monocyte-macrophages. However, a major drawback in using these cells for signaling studies is the inability to transfect them and introduce transgenes. Recently, we have successfully overcome this problem using adenoviral vectors (12, 13). In this study, we used this approach to study IL-10 regulation of TNF-α at the gene level in primary human macrophages. In particular, adenoviral vectors were constructed, incorporating luciferase reporter genes, under the control of the TNF-α promoter, with or without the 3′ untranslated region (3′UTR). Once introduced into primary human macrophages, these reporter genes gave high levels of induction in response to LPS or zymosan. This, to our knowledge, is the first direct examination of TNF-α gene function in primary human cells. Moreover, this study showed that, depending on the length of exposure of cells to the cytokine, IL-10 could inhibit TNF-α expression by either the 5′ promoter region or the 3′UTR, suggesting that both transcriptional and posttranscriptional mechanisms could be involved in this single cell type. However, unlike previous studies in other systems, IL-10 was unable to inhibit LPS-induced activation of p38 MAPK and had only a minor effect on NF-κB-induced transcription.
Materials and Methods
Human mononuclear cells were isolated from single donor plateletphoresis residues obtained from the North London Blood Transfusion Center (London, U.K.) by Ficoll-Hypaque centrifugation preceding monocyte separation in a Beckman JE6 elutriator (Beckman, High Wycombe, U.K.). Monocyte purity was routinely >85% when assessed by flow cytometry (14). The elutriated human monocytes were cultured at 1 × 106/ml in RPMI 1640 (BioWhittaker, Verviers, Belgium) with 25 mM HEPES and 2 mM l-glutamine supplemented with 10% (v/v) heat-inactivated FCS and 10 U/ml penicillin/streptavidin. To optimize infection, macrophages were derived from the monocytes by culturing the cells with M-CSF at 100 ng/ml (Genetics Institute, Boston, MA) for 48 h (13).
Human TNF-α promoter (−1173 bp) with 3′UTR of the human TNF-α gene (pGL3-TNF-α-3′UTR), or without the 3′UTR (pGL3-TNF-α) (15), were subcloned into the pAdTrack vector (16) to generate pAdTrack-p5′3′UTR and pAdTrack-p5′. KpnI/SalI fragments containing the human TNF-α promoter, the luciferase reporter gene, and the SV40 late poly(A) signal were derived from pGL3-TNF-α inserted into KpnI/SalI sites of the AdTrack vector. pAdTrack-p5′3′UTR was obtained by substituting a XbaI/BamHI fragment containing the SV40 late poly(A) signal in the pGL3-TNF-α plasmid for ∼1 kbp of 3′UTR amplified by PCR with corresponding primers: 3′UTR-F (XbaI), aattctagaGGAGGACGAACATCCAAC; and 3′UTR-R(BamHI), aatGgATcCCCAGAGTTGGAAATTC. The KpnI/SalI fragments were subsequently cloned into the pAdTrack vector.
Adenoviral vectors and their propagation
The pAdEasy-1 adenoviral plasmid was provided by Prof. B. Vogelstein (Howard Hughes Medical Institute, Baltimore, MD). Recombinant viruses were generated by homologous recombination in BJ5183 Escherichia coli transformed by heat-shock with 1 μg of each of the linearized PmeI pAdTrack constructs and 100 ng of pAdEasy-1. Kanamycin-resistant positive recombinant clones were selected and confirmed by restriction enzyme digestion. Viral DNA was transfected into HEK 293 cells. Viruses were purified by ultracentrifugation through two cesium chloride gradients, as described in He et al. (16). Plaque assays were performed by HEK 293 cells, exposing the cells to each virus for 1 h in serum-free DMEM (Life Technologies, Paisley, U.K.) and, subsequently, overlaying the cells with an agarose mixture (1.5% agarose, 2× DMEM with 4% FCS; v/v, 1/1) and incubated for 10–14 days to determine viral titer (16). The IκBα encoding virus was kindly provided by Dr. R. de Martin (University of Vienna, Vienna, Austria) and adeno-NF-κB luciferase reporter virus (AdvNF-κB-luc), previously described (17), was kindly provided by Dr. P. McCray (University of Iowa, Iowa City, IA).
Infection and IL-10 treatment
Human macrophages were plated at a density of 2 × 105 cells/well in 96-well plates and exposed to virus at the optimal multiplicity of infection (m.o.i.; 40:1 for Advp5′ and Advp5′3′UTR; 100:1 for Ad0 AdvIκBα; 200:1 for AdvNK-κB-luc) for 1 h in serum-free medium, followed by washing and reculturing in growth medium with 2% (v/v) FCS for 24 h. Infected cells were then stimulated with 10 ng/ml LPS (Salmonella typhimurium; Sigma, Poole, U.K.) or 30 μg/ml zymosan (Sigma) for 4 h, unless otherwise stated, in the presence or absence of IL-10.
Measurement of human TNF-α production
TNF-α levels were measured in cell supernatants by sandwich ELISA as previously reported (18).
After stimulation, cells were washed once in PBS and lysed with 100 μl of chloramphenicol acetyltransferase lysis buffer (0.65% (v/v) Nonidet P-40, 10 mM Tris-HCL (pH 8), 0.1 mM EDTA (pH 8), 150 mM NaCl). Cell lysate (50 μl) was transferred to a luminometer cuvette strip and luciferase assay buffer (220 μl) was added. Luciferase activity was measured with a luminometer (Labsystems, Chicago, IL) by dispensing 30 μl luciferin (1.5 mM; Sigma) per assay point. Cell lysates were assayed for protein concentration by Bradford assay and luciferase activity was adjusted accordingly.
RNase protection assay
After M-CSF treatment, cells were plated at 2 × 106/well in a 12-well plate and infected, as described above. In Advp5′- and Advp5′3′UTR-infected cells, luciferase and GAPDH mRNAs were detected by RNase protection assay (RPA) by using luciferase and GAPDH riboprobes, respectively. In parallel, TNF-α and GAPDH mRNAs were detected in Adv0-infected cells. Riboprobe vectors were constructed as follows. A 352-bp HincII-XbaI luciferase fragment was cloned from pGL3c (Promega, Madison, WI) into pBluescript KS-digested EcoRV and XbaI. A 268-bp TNF-α gene fragment was amplified by PCR from human genomic DNA and subcloned into the SpeI site of pBluescript KS+ (kindly provided by Dr. A. Clark, Kennedy Institute, London, U.K.). Riboprobe template constructs were linearized by appropriate restriction enzyme and purified by phenol-chloroform extraction and ethanol precipitation. Luciferase and GAPDH riboprobes were synthesized using T7 RNA polymerase and TNF-α riboprobe by using T3 RNA polymerase (Boehringer Mannheim, Indianapolis, IN) in the presence of 50 μCi of [α-32P]UTP (800 mCi/mmol; Amersham Pharmacia Biotech, Little Chalfont, U.K.). The final concentration of unlabeled UTP in the in vitro transcription reactions was 12 μM, except in the case of luciferase, where it was 2.4 μM. RPAs were conducted using the Direct Protect kit (Ambion, Austin, TX). Under the hybridization conditions DNA-RNA heteroduplexes are not detected. Protected RNA fragments were resolved by electrophoresis on denaturing 6% polyacrylamide gels, quantified by phosphor imaging (Fuji FLA-2000; Raytek Scientific, Sheffield, U.K.) and visualized by autoradiography. Each experiment was performed twice and serial dilutions of lysates were used to check that quantitations were within the linear range of the assay.
Immunoprecipitation and in vitro kinase assays
MAPKs were immunoprecipitated from cleared cell lysates, as described previously (19). In vitro kinase assays for p38 MAPK were performed using either His6-MAPKAPK-2 or GST-ATF-2 as a substrate, c-Jun N-terminal kinase (JNK) assays were performed using GST-ATF-2 as a substrate, and p42 MAPK assays were performed using myelin basic protein (Sigma) as a substrate. Immunoprecipitates were incubated with 30 μl kinase assay buffer (25 mM Tris (pH 7.5), 25 mM MgCl2, 25 mM β-glycerophosphate) containing 20 μM ATP and 0.5 μCi [γ-32P]ATP (Amersham Pharmacia Biotech) with 50 μg/ml appropriate substrate protein for 25 min at room temperature. Reactions were terminated by the addition of gel sample buffer and boiling for 5 min. All substrates were separated by SDS-PAGE. Gels were dried and phosphorylated substrates were visualized using a Fuji FLA-2000 phosphor imager and by autoradiography at −70°C.
Following stimulation, cells were scraped into ice-cold PBS and lysed in hypotonic lysis buffer (0.125% Nonidet P-40, 5 mM HEPES (pH 7.9), 10 mM KCl, 1.5 mM MgCl2), and nuclei were harvested by centrifugation (13,000 × g for 30 s). Nuclear protein extracts were prepared by incubating the nuclei in hypertonic extraction buffer (5 mM HEPES (pH 7.9), 25% glycerol, 500 mM NaCl, 1.5 mM MgCl2, 0.2 mM EDTA) for 2 h with constant agitation. NF-κB DNA binding activities were determined by incubating 1–3 μg of each extract with [γ-32P]ATP-labeled double-stranded NF-κB consensus oligonucleotide (Promega), followed by resolution on a 5% (w/v) nondenaturing polyacrylamide gel. Gels were dried on to filter paper and retarded DNA; protein complexes were visualized using Hyperfilm MP (Amersham Pharmacia Biotech).
LPS and zymosan potently activate TNF-α reporter genes delivered by adenoviral infection into primary human macrophages and demonstrate the involvement of the 3′UTR in mRNA stability
The 5′ promoter and 5′ promoter-3′UTR constructs, described previously (15), were incorporated into recombinant adenoviruses (reporter viruses, Advp5′ and Advp5′-3′UTR, respectively), as previously described by He et al. (16). Primary human macrophages were infected with these adenoviral constructs at a m.o.i. of 40:1, as previous studies had indicated that this concentration resulted in the successful infection of >90% of cells (Ref. 13 and data not shown). LPS activation of the reporter gene resulted in a potent stimulation of both the 5′ and 5′3′UTR constructs, respectively (Fig. 1). The reporter gene also responded equally well to the yeast product, zymosan (Fig. 1), an alternative stimulus of TNF-α production with a similar potency to LPS (20). A consistent finding in all experiments was the lower absolute response of constructs containing the 3′UTR. These data support the view, obtained from previous studies in murine macrophage cell lines, that the 3′UTR is generally suppressive to TNF-α expression (21).
Kinetic studies were also performed to compare the behavior of the reporter genes with the endogenous gene. In response to LPS, TNF-α production reached a maximum 4 h after stimulation, and thereafter decreased slowly with a t1/2 of ∼18.5 h (Fig. 2,A). The kinetics of luciferase activity from Advp5′- or Advp5′3′UTR-infected LPS-stimulated human macrophages followed a similar profile to TNF-α, with maximum expression at 4 h and apparent t1/2 estimated to be 23.5 and 12 h, respectively (Fig. 2, B and C). The shorter half-life of luciferase, when under the additional control of the 3′UTR, might be expected from the overall destabilizing effect of this element on TNF-α mRNA (21, 22) and explain why the absolute signals produced in the presence of the 3′UTR are consistently lower (Fig. 1). To confirm this, studies on luciferase mRNA stability were performed and showed that the presence of the 3′UTR does indeed increase the rate of decay of luciferase mRNA (Fig. 2,D). In the presence of the 3′UTR the mRNA had a t1/2 of 56 min; however, in the absence of this element there was little decay (<15%) within the same timeframe. Similar studies on TNF-α mRNA indicate a t1/2 of ∼30 min, similar to luciferase mRNA from the 3′UTR constructs (Fig. 2 D). One might have expected that the half-life of the endogenous TNF-α protein would have been comparable to the 5′3′UTR rather than the 5′ construct. However, these data do not take into account potential differences in the biological half-lives of the TNF-α and luciferase proteins, and one can only assume that the half-life of the endogenous TNF-α would be longer in the absence of the 3′UTR in humans, as shown previously in mouse cells with deletions of the 3′UTR on the AU-rich region of the 3′UTR (23).
In primary human macrophages, IL-10 requires the 3′UTR to suppress TNF-α production but has no effect on p38 MAPK activation
Adenovirus delivery of TNF-α-based reporter gene constructs to primary human macrophages was then used to investigate what role the 5′ and 3′ regions may play in the IL-10 inhibition of TNF-α expression. Adeno-reporter virus-infected cells were simultaneously treated with LPS and various concentrations of IL-10 for 4 h, after which time TNF-α production and luciferase activities were assayed (Fig. 3). IL-10 inhibited TNF-α expression to a maximum of 80% at 10 ng/ml (Fig. 3). However, the responses of the two reporter constructs were quite distinct. The 5′ construct was only weakly inhibited by IL-10 (∼10%), whereas the 5′3′UTR construct showed a dose response profile similar to the endogenous TNF-α, although the maximum inhibition attained was less (60% at 10 ng/ml). The IC50 for IL-10 on TNF-α protein expression was 0.2–0.3 ng/ml, compared with 2–3 ng/ml for the reporter gene. However, if the concentration of half-maximal inhibition is calculated, then the activity of IL-10 is similar for the endogenous gene (∼0.1 ng/ml) and the 5′3′UTR construct (0.2–0.3 ng/ml). This suggests that aspects of the inhibitory activity of IL-10 on the reporter gene and the endogenous gene are similar. The kinetics of IL-10 inhibition of TNF-α expression and luciferase expression were also compared using the optimal concentration of 10 ng/ml (Fig. 4). Over the 2- to 24-h period poststimulation, the expression of TNF-α and the 5′3′UTR construct gave very similar profiles with very little activation detected in the presence of IL-10 (Fig. 4, A and B). The effect of IL-10 on the 5′ construct was again much weaker, with no significant inhibition of luciferase activity over the time course (Fig. 4,C). These data suggest that the major inhibitory effect of IL-10 is mediated via the 3′UTR and that there appears to be little effect on the transcription of the gene when IL-10 is administered at the same time as LPS. To gain a further insight into the mechanism of IL-10 activity, luciferase mRNA levels were analyzed by RPA. As shown in Fig. 5, simultaneous addition of IL-10 caused a marked reduction of luciferase mRNA from the 5′3′UTR construct, whereas there was only a marginal effect on mRNA from the 5′ construct. These data indicate that IL-10 causes a reduction in mRNA levels via the 3′UTR. Data obtained with TNF-α mRNA showed similar results to the 5′3′UTR construct (Fig. 5).
As the 3′UTR is associated with posttranscriptional control of TNF-α expression, the data so far suggest that IL-10 mediates its activity at this level. If this is so, IL-10 should still be able to inhibit TNF-α production, at least for a period, if added after LPS. Reporter virus-infected macrophages were stimulated with LPS, IL-10 was added for periods of up to 2 h postactivation, and the cells were harvested at 4 h for assay. As expected, IL-10 had little effect on the activity of the 5′ construct, regardless of when it was added (Fig. 6). The inhibitory effect of IL-10 on TNF-α and the expression of the 5′3′UTR construct was maintained, even if IL-10 was added 1 h after LPS activation, but was greatly reduced if the cytokine was added 2 h postactivation (Fig. 6). These data suggest that IL-10 does not inhibit the early events that are involved in TNF-α production (e.g., transcription) and instead targets later events (e.g., posttranscriptional). These data support the hypothesis that posttranscriptional control mediated via the 3′UTR is the target of IL-10.
As p38 MAPK has been implicated in the posttranscriptional control of TNF-α expression (24, 25) and has very recently been shown to be inhibited by IL-10 in murine macrophages (10), we investigated whether this kinase was a target for IL-10 in human macrophages. As shown in Fig. 7, IL-10 was unable to inhibit LPS-induced p38 MAPK activation. As the other related p42/44 extracellular signal-regulated kinase, MAPKs, and p54 JNK are also involved in regulating TNF-α expression, the effect of IL-10 on these kinases was also studied. However, like p38 MAPK, there was no inhibitory effect on the activation of these kinases (Fig. 7).
Preincubation of human macrophages with IL-10 before LPS stimulation reveals a second mechanism of inhibiting TNF-α production through the 5′ promoter
As time of exposure to IL-10 could obviously have a bearing on its function, this study was extended to investigate the effect of adding IL-10 before LPS for periods of up to 24 h (Fig. 8). Preincubating macrophages with IL-10 for 12 h produced a modest increase in the inhibition of endogenous TNF-α production or luciferase activity from the 5′3′UTR reporter gene, when compared with the effect of simultaneous addition of LPS and IL-10 (Fig. 8). Further periods of preincubation, up to 24 h, did not elicit any major additional effect. However, pre-exposure of cells to IL-10 had a profound effect on the expression of the 5′ reporter. Preincubation of macrophages with IL-10 for 12 h before LPS stimulus resulted in 50% inhibition of the 5′ construct that increased to nearly 70% when the preincubation period was extended to 24 h (Fig. 8). This was compared with the 80% inhibition of endogenous TNF-α production. The effect of preincubating macrophages for 24 h with different concentrations of IL-10 was also examined. As shown in Fig. 9, LPS-induced TNF-α protein expression was inhibited to a maximum of 90% (10 ng/ml IL-10) with an IC50 of ∼0.1 ng/ml, regardless of which reporter construct had been infected into the macrophages. IL-10 also inhibited the expression of the 5′3′UTR reporter to an identical degree to the endogenous gene (Fig. 9,B). However, in contrast to data in Fig. 3,A, preincubation for 24 h with IL-10 now produced a dose-dependent inhibition of the 5′ construct that showed a maximum inhibition of 60% at 10 ng/ml (Fig. 9,A). The IC50 for IL-10 on the 5′ construct was 5 ng/ml, but this reduced to 0.5 ng/ml if the half-maximal inhibition was again calculated. These data suggest that, in addition to posttranscriptional regulation of the TNF-α gene, IL-10 can also inhibit transcription of the TNF-α gene if cells are exposed to this inhibitory factor for a sufficient period. Indeed, studies on mRNA levels showed that preincubation for 24 h with IL-10 resulted in a decrease in luciferase mRNA, regardless of the presence of the 3′UTR (Fig. 5). However, we were unable to perform nuclear run-on experiments to confirm an effect on transcription, as we cannot obtain sufficient cells from a single donor to perform fully controlled experiments.
IL-10 inhibition of LPS-induced TNF-α production is independent of NF-κB
An inhibitory mechanism that involves the 5′ promoter region of the TNF-α gene suggest that IL-10 may be interfering with the function of a transcription factor. As NF-κB has been implicated previously as a target for IL-10, the effect of the cytokine on the activation of this transcription factor was investigated. As shown in Fig. 10,A, IL-10 had no effect on the activation of NF-κB by LPS, as measured by EMSA, regardless of the length of the exposure to the cytokine. As would be expected from this result, we also observed no effect of IL-10 on IκBα degradation (data not shown). Next, using an adenovirus encoding an NF-κB-driven luciferase reporter gene previously described by Sanlioglu et al. (17), the effect of IL-10 on the transactivating activity of the transcription factor was also examined. As shown in Fig. 10,B, preincubation with IL-10 for 24 h had no effect on the NF-κB transcriptional activity, although there was a slight inhibition of the NF-κB reporter gene (∼20%) when IL-10 was added simultaneously with LPS. As expected, the coinfection of macrophages with the IκBα-encoding adenovirus (AdvIκBα) inhibited gene expression by >90%, whereas a control virus had no effect. We extended this study further by addressing whether a role for NF-κB in TNF-α expression was essential for IL-10 inhibition. We have previously shown using AdvIκBα that NF-κB is not a requirement for zymosan-induced TNF-α production (20). In this study, similar data were obtained with the reporter genes that showed that coinfection of the cells with the reporter viruses and AdvIκBα resulted in an 80–90% inhibition of the response of both constructs to LPS. In contrast, we observed no significant inhibition in response to zymosan (Fig. 10,C). A control virus Ad0 had no effect on responses to either stimulus. However, IL-10 showed the same inhibitory profile to zymosan-induced reporter gene activity as seen above with LPS, namely inhibition via the 3′UTR, when cytokine and stimulus were added simultaneously, and an effect on the 5′ region only when cells were preincubated with IL-10 (Fig. 10,D). IL-10 inhibited zymosan-induced production of TNF-α protein to levels similar to those obtained with the reporter genes, except, of course, in the case of the t0 time point and the 5′ construct (Fig. 10 D). These data support the conclusion that IL-10-induced inhibition of TNF-α expression is NF-κB independent.
The published literature on the mechanisms of the anti-inflammatory action of IL-10 is somewhat contradictory. The proposed mechanisms, although incompletely characterized, appear to be diverse and depend on the nature of the stimulus, the gene of interest, and the cell system investigated. In the present study, a genetic approach was taken by investigating which regions of the TNF-α gene mediate the suppressive effect of IL-10. A particular aspect to this study was the harnessing of our previously successful experience of using adenovirus to deliver transgenes into primary human monocytic cells (12, 13, 20, 26). Using this approach, we were able to deliver, for the first time to our knowledge, TNF-α gene-based reporter constructs into primary human macrophages and investigate TNF-α gene regulation in a system highly relevant to human pathology. Unlike previous studies, the data showed that IL-10 apparently uses two independent mechanisms for inhibiting TNF-α expression, by targeting either the 5′ promoter or the 3′UTR.
The description of a potential posttranscriptional mechanism via the 3′UTR for the IL-10-mediated inhibition of TNF-α production in human cells is novel, as previous studies in human cells have implicated a transcriptional target (4, 5, 7, 27). However, none of these studies used the approach of analyzing gene function. These data are in agreement with studies in murine macrophages, where a posttranscriptional mechanism has been proposed (9, 28). Previous studies on the 3′UTR have shown that this region has an overall suppressive effect (21) on TNF-α expression (which this study has now confirmed in human macrophages). However, although we were able to show that IL-10 was able to decrease mRNA levels via the 3′UTR, it was not possible to confirm previous findings in murine macrophages (9) that this effect was mediated by destabilizing TNF-α mRNA (data not shown). A potential target for IL-10 in this context would be p38 MAPK, as this kinase is involved in the posttranscriptional control of TNF-α expression (24). However, we were unable to show any effect of IL-10 on the activation of p38 MAPK, the related p54/JNK, or p42/44 MAPK. These results agree with previous findings that IL-10 did not inhibit LPS-induced phosphorylation of p38 MAPK in human monocytes (3). A recent study has suggested that cytokines and growth factors could affect mRNA splicing (29). The luciferase reporter gene used in this work is, of course, devoid of introns; however, the apparent similarity of the responses of the endogenous TNF-α and reporter genes to IL-10 indicate that introns and mRNA splicing are not targets for IL-10 activity. Therefore, the mechanism by which IL-10 may inhibit TNF-α expression posttranscriptionally is still unclear. A potential target is the AU-rich ARE regions found in the 3′UTRs of TNF-α and many other cytokine genes. These regions have been implicated in the regulation of mRNA stability and turnover (30), as well as in mRNA translation (31, 32, 33, 34). The recent study in murine macrophages by Kontoyiannis et al. (10) has identified this region as a target for IL-10-mediated suppression of TNF-α gene translation by a mechanism involving inhibition of p38 MAPK. However, IL-10 inhibition of TNF-α in human macrophages appears to be grossly different, as there is no inhibition of p38 MAPK. Also, as TNF-α mRNA levels are inhibited, this precludes any major role for an effect on TNF-α gene translation. The full understanding of the mechanism for the posttranscriptional regulation of TNF-α by IL-10 in human cells will obviously require a more extensive study and may require the discovery of as yet unknown pathways involved in the general posttranscriptional control of this gene.
The second inhibitory mechanism of IL-10 in primary human macrophages required the 5′ promoter region, suggesting an inhibition of gene transcription. However, the amount of material available from individual donors prevented any meaningful run-on experiments being performed to confirm this. In contrast to the posttranscriptional mechanisms, this inhibitory mechanism required a prolonged pre-exposure to IL-10. The inference from these data would be that there is a requirement for gradual changes in cell physiology, e.g., protein synthesis, to mediate this effect of IL-10. A requirement for IL-10-directed protein synthesis for the inhibition of LPS-induced TNF-α gene transcription is supported by previous studies in human cells, showing that the protein synthesis inhibitor cycloheximide inhibits IL-10 function (4, 5). However, so far our own studies with cycloheximide have proved inconclusive, possibly because the general inhibition of protein synthesis leads to toxicity. It has been reported in human PBMC that IL-10 can suppress the activation of the key proinflammatory transcription factor, NF-κB (7), possibly by inhibiting IκBα kinase activity (6) or NF-κB DNA binding (6). Previously, we have also observed IL-10-mediated inhibition of LPS-induced NF-κB activity in the murine cell lines RAW 264.7 and the pre-B cell line 70Z/3. However, the concentrations required to achieve this effect were high (an IC50 of ∼100 ng/ml), far in excess of that required to inhibit TNF-α expression (an IC50 of ∼0.3 ng/ml) (28). However, our studies in human macrophages have failed to show any effect of IL-10 (10 ng/ml) on NF-κB activation, as judged by EMSA. Additional studies using an adeno-NF-κB reporter virus showed that preincubation with IL-10 had no effect on transcriptional function. The only effect of IL-10 on NF-κB function we observed was a 20% decrease in LPS-induced NF-κB reporter gene activity when IL-10 was added simultaneously with LPS. This does not appear to account for a mechanism that operates via the 3′UTR or the potent suppression of IL-10 on TNF-α expression. Moreover, our studies with zymosan show that a NF-κB-dependent mechanism of gene induction is not an essential requirement for the IL-10 inhibition of TNF-α expression. The studies showing inhibition of NF-κB reported that IL-10 inhibited activation of the factor with either minimal (5 min) or no pre-exposure of the cells to the cytokine before stimulus. This would not fit with the observations in this study and the studies of others (4, 5) showing the requirement for prolonged exposure to IL-10 (4, 5), which indicated de novo protein synthesis was required for IL-10 inhibition of TNF-α gene transcription. In summary, our data using primary human macrophages do not agree with previous studies that IL-10 has a profound effect on NF-κB activation and that this is a mechanism by which IL-10 inhibits TNF-α expression. We can only conclude that, to some extent, the differences in these results may be due to our use of primary cells rather than cell lines and that changes in signaling mechanisms can occur between such systems. Such a conclusion is supported by recent studies on NF-κB-inducing kinase. This kinase was proposed to play an essential role in NF-κB activation by many stimuli, including LPS, TNF-α, and IL-1, in cell lines (35). However, this kinase has subsequently been shown to play no such role in response to these stimuli in primary human or murine cells (36, 37, 38). Our data indicate that another transcriptional mechanism is the target of IL-10 and that the mechanism involved is more likely to be indirect, probably requiring the expression of IL-10-induced proteins. The identification of the transcriptional target for IL-10 may be dependent on a greater understanding of the general mechanisms controlling TNF-α gene transcription.
The previous studies in human cells focusing on transcriptional regulation have overlooked posttranscriptional mechanisms operating via the 3′UTR. The advantage of this study is that, by correlating the effect of IL-10 with different regions of the TNF-α promoter, it was possible to identify and delineate both mechanisms and show, for the first time, that both mechanisms could coexist. Also, by using the adenoviral system to undertake this study in primary cells, any potential problems with performing studies in transformed cell lines could be avoided. This report provides some key insights as to why there has been much disagreement in the field of IL-10 function, by showing that multiple mechanisms can coexist and that the nature of the system studied has a major impact on the result.
We thank Drs. A. Clark, J. Dean, and U. Sarma, L. Bradley, and Profs. M. Feldmann, J. Saklatvala, R. D. Schreiber, and H. Nagase for their helpful reading of the manuscript, S. Evans for typing the manuscript, Dr. R. de Martin for the AdvIκBα virus, and Dr. Paul McCray for the AdvNF-κB-luc.
This work was funded by the Medical Research Council, the Arthritis Research Campaign, and the European Union. A.D. was supported by a European Union Training, Mobility, and Research award.
Abbreviations used in this paper: MAPK, mitogen-activated protein kinase; 3′UTR, 3′ untranslated region; m.o.i., multiplicity of infection; AdvNF-κB-luc, adeno-NF-κB luciferase reporter virus; JNK, c-Jun N-terminal kinase; RPA, RNase protection assay.