Contrasting mechanisms for suppression of macrophage cytokine release by transforming growth factor-beta and interleukin-10.

Transforming growth factor (TGF)-beta and interleukin (IL)-10 inhibited lipopolysaccharide (LPS)-induced macrophage production of the inflammatory cytokines tumor necrosis factor-alpha (TNF), IL-1 alpha, and IL-1 beta by contrasting post-transcriptional mechanisms. TGF-beta acted slowly and late, as it required 12-16 h to exert a suppressive effect, and inhibited TNF production even when added 6 h after LPS. TGF-beta affected neither the level of TNF mRNA, the release of preformed TNF nor the degradation of TNF. Thus, TGF-beta appeared to inhibit translation of TNF mRNA. IL-10 not only suppressed TNF release to a 25-fold greater extent than TGF-beta, but also inhibited release of IL-1. In contrast to TGF-beta, IL-10 acted on an early step in cytokine production, its effect being maximal 3 h after addition of LPS. Unlike TGF-beta, IL-10 markedly suppressed TNF, IL-1 alpha, and IL-1 beta mRNA levels. However, this was accomplished without suppressing transcription of the corresponding genes. Moreover, cycloheximide antagonized the IL-10-dependent reduction in cytokine mRNA levels. Thus, IL-10 may induce a ribonuclease active on cytokine transcripts or may induce a protein that enhances the susceptibility of TNF, IL-1 alpha, and IL-1 beta mRNAs to ribonucleolytic action. We conclude that IL-10 and TGF-beta induce different phenotypes of macrophage deactivation, and deactivate macrophages by different mechanisms: IL-10 promotes degradation of cytokine mRNA, while TGF-beta primarily suppresses translation.

TGF-8 acted slowly and late, as it required 12-16 h to exert a suppressive effect, and inhibited TNF production even when added 6 h after LPS. TGF-8 affected neither the level of TNF mRNA, the release of preformed TNF nor the degradation of TNF. Thus, TGFfl appeared to inhibit translation of TNF mRNA. IL-10 not only suppressed TNF release to a 26-fold greater extent than TGF-8, but also inhibited release of IL-1. In contrast to TGF-8, IL-10 acted on an early step in cytokine production, its effect being maximal 3 h after addition of LPS. Unlike TGF-8, IL-10 markedly suppressed TNF, IL-la, and IL-18 mRNA levels. However, this was accomplished without suppressing transcription of the corresponding genes. Moreover, cycloheximide antagonized the IL-10-dependent reduction in cytokine mRNA levels. Thus, IL-10 may induce a ribonuclease active on cytokine transcripts or may induce a protein that enhances the susceptibility of TNF, ILla, and IL-18 mRNAs to ribonucleolytic action. We conclude that IL-10 and TGF-8 induce different phenotypes of macrophage deactivation, and deactivate macrophages by different mechanisms: IL-10 promotes degradation of cytokine mRNA, while TGF-8 primarily suppresses translation.
Activation of macrophages plays an important role in immune responses against microbes, tumors, and normal host tissues. In the last decade, many natural substances of defined composition have been identified that can increase the ability of macrophages to present antigen, secrete cytotoxic, inflammatory, or immunoregulatory molecules, and kill tumor cells or microbes. These include cytokines such as interferon (1FN)'-y and tumor necrosis factor-a (TNF), and bacterial products such as lipopolysaccharide (LPS) and staphylococcal enterotoxins. It has become equally apparent that cytokines and microbial products can also block the induction of macrophage activation or reverse a pre-existent state of activation (1,2). These processes, termed macrophage deactivation, may serve to limit immune and inflammatory responses during wound healing and tissue repair. Parasites and tumor cells appear to evade host immunity in part by producing similar mediators or by inducing their production by the host (3)(4)(5)(6)(7).
The present study focusses on the ability of primary murine macrophages to produce TNF and IL-1, two cytokines of major importance in inflammation and immunity. We contrast TGF-@ and IL-10 as macrophage deactivators, analyze their mechanisms of action, and demonstrate that these mechanisms, while primarily post-transcriptional for both cytokines, are distinct.

MATERIALS AND METHODS
Mice-Female CD1 mice (8-12-weeks old) were from the Charles River Breeding Laboratories (Wilmington, MA).
Macrophage Cultures-Thioglycollate broth (4% Brewer's, Difco, Inc. Detroit, MI)-elicited cells were harvested from the peritoneal cavity with phosphate-buffered saline, cultured with RPMI 1640 medium (2 mM L-glutamine, 100 pg/ml gentamicin, 100 pg/ml penicillin, JRH Biosciences, Lenexa, KS) plus 10% fetal bovine serum (HyClone Laboratories, Logan, UT) in 24-well plates (106/well) (Costar, Cambridge, MA) or 78-cm2 Petri dishes (15-20 X 106/dish) (Corning Glass Works, Corning, NY) at 37 "C in 5% CO,, 95% air and enriched for macrophages by a 2-h adherence step as described (15) For estimation of the actual amount of TNF or IL-1 the data (mean OD 570 nm + S.D. or mean cpm k S.D.) from at least 10 serial 2-fold dilutions/supernatant were subjected to probit analysis using recombinant cytokine standard curves, as described (22). Finally, the fold suppression of TNF or IL-1 release in the supernatants from IL-10-or TGF-@-treated mdJ cultures was determined (nanogram/milliliter of TNF or unit/milliliter of IL-1 in control + nanogram/milliliter of TNF or unit/milliliter IL-1 in supernatant of treated culture). Preparation of Total Membranes and Cytosol-Total membranes and cytosol fractions for determination of intracellular TNF were prepared as described by Kriegler et al. (23). Briefly, macrophage monolayers from 78-cm2 tissue culture dishes were washed three times with Dulbecco's phosphate buffered saline (PBS, JRH Biosciences), scraped with a rubber policeman into PBS, and centrifuged (200 X g). The pellet was resuspended in ice-cold 10 mM K+ phosphate buffer, pH 7.5, with 1 mM phenylmethylsulfonyl fluoride and sonicated on ice for 60 s (50% duty cycle, level 4) with a Sonifier 250 cell disruptor (Branson Ultrasonics Co., Danbury, CT). The completely disrupted cells were centrifuged at 4 "C for 5 min at 500 X g, and the supernatants centrifuged for 30 min at 16,000 x g and again for 60 min at 100,000 X g. The pellet of the final centrifugation step ("total membranes") was dissolved in PBS after water-bath sonication (15 min, 4 "C) and stored together with the final supernatant ("cytosol") at -70 "C.
Preparation of Total Cell Lysates-Macrophage monolayers were washed and scraped as described above, but the cell pellet was lysed in 150 mM NaCl, 10 mM Tris, pH 8, 1 mM phenylmethylsulfonyl fluoride, 0.5 unit/ml aprotinin, and 1% Triton X-100 (60 min, rotating a t 4 T ) . The nuclei were spun down at 500 X g (10 min, 4 "C) and the supernatants finally cleared at 100,000 X g (4 "C, 60 min) (24). p5S]Cysteine Labeling, Pulse-chase Experiments, Immunoprecipitation, and SDS-Polyacrylamide Electrophoresis-The procedures folwashed three times with PBS (37 "C) and then starved for 30-60 min lowed published protocols (25). Briefly, macrophage monolayers were in cysteine-and serum-free RPMI 1640 (supplemented with 10 mM 3C. Bogdan, J. Paik, Y. Vodovotz, and C. Nathan, unpublished results. HEPES). For short-term labeling, 300 pCi of [35S]cysteine (1300 Ci/ mmol, 10 mCi/ml, Amersham Corp.) was added per 20 X lo6 macrophages. After 2-3 h, the monolayers were washed extensively with PBS, and total membranes, cytosol, or total lysates were prepared as described above. For pulse-chase experiments 10 X lo6 macrophages were labeled with 300 pCi of [35S]cysteine for 30 min, followed by addition of 2 volumes of medium containing 20 mg/ml cold L-cysteine. Total cell lysates were prepared after the time points indicated. The amount of protein-bound radioactivity was determined after trichloroacetic acid precipitation of an aliquot of the samples, Immunoprecipitation of cleared and preabsorbed lysates was performed with polyclonal rabbit anti-murine TNF serum (Genzyme, Cambridge, MA) and protein A-Sepharose CL-4B (Sigma). Immunoprecipitates were subjected to nonreducing 15% SDS-polyacrylamide gel electrophoresis, subsequent autofluorography with ENTENSIFYTM (Du Pont-New England Nuclear) and exposure to Kodak XAR-1 films (Eastman) at -70 "C.
Oligonucleotide Probes and Plasmids-30-bp single-stranded DNA oligonucleotide probes were kindly provided by Amgen (Thousand Oaks, CA) (IL-la, IL-l@, and TNF-a probes) or purchased from Clontech (Palo Alto, CA) (mouse @-actin probe). The oligoprobes were 5' end-radiolabeled with T4 polynucleotide kinase (Stratagene, La Jolla, CA) in the presence of 50 pCi of [a-"PI ATP (-6000 Ci/ mmol, 10 mCi/ml, Amersham Corp.) to a specific activity of > 2 X 10' cpm/mg. Phagemid pBluescript SK(+) containing the full-length 2-kilobase cDNA for murine IL-la was obtained from American Type Culture Collection. Plasmid pUC-9 with a 1.4-kilobase PstIIBarnHI insert of the murine TNF-a gene and plasmid pSB with a 1054-bp EcoRI fragment of mIL-l@ cDNA were kindly provided by Drs. B.
Data Presentation-Unless indicated otherwise, results are means +S.E. for the number of experiments performed.

RESULTS
IL-10 and TGF-P Suppress LPS-induced T N F Production to Different Extents and with Distinct Stimulation Requirements-Exposure of LPS-stimulated macrophages to IL-10 diminishes the accumulation of TNF in their conditioned medium (15). The first goal of the present study was to compare TGF-P to IL-10 in this regard. Throughout this study, all numerical values for TNF suppression at any concentration of TGF-P or IL-10 were determined by at least ten 2-fold dilutions of the respective supernatant ( 10) compared to a control without TGF-P or IL-10. For data involving inhibition <go%, percent suppression is the preferred presentation.
As shown in Fig. 1, when TGF-@ and LPS were added simultaneously to macrophages, TGF-(3 suppressed TNF accumulation in the supernatant 20 h later. The inhibitory effect was maximal at 10 ng/ml TGF-P, which caused 70 f 3% (4. The above comparison was based on IL-10 in conditioned media of recombinant COS cells. As the absolute amount of IL-10 in the COS cell supernatants has not been determined, we also used IL-10 purified from recombinant E. coli. At 100 ng/ml, five different preparations of this material caused >90% suppression of TNF release (Fig. 1). However, E. coliexpressed IL-10 rapidly lost activity during storage and after freezing and thawing, especially in the absence of protein carriers, so that the concentration affording 50% inhibition of TNF release varied between 0.26 ng/ml (14 PM) and 6.1 ng/ml (89 p~) . Our next goal was to investigate whether suppression of TNF release by IL-10 and TGF-@ is dependent on the stimulus used to induce TNF. SEB induces macrophage TNF production (28,29), and a combination of the phorbolester PMA and the calcium ionophore A23187 primes macrophages for tumor cell cytotoxicity, which might reflect TNF production (30). We confirmed that these agents were not contaminated with LPS, yet induced TNF release, and demonstrated that TGF-@ inhibited TNF production to a similar extent, whether TNF was induced by SEB, PMA and A23187, or LPS itself (Table  I). IL-10 strongly suppressed both LPS-and SEB-triggered TNF release but was less active when PMA/A23187 was the stimulus (Table I).
Next, we contrasted the effects of IL-10 and TGF-p on TNF production under different incubation conditions. The inhibitory effect of TGF-P was only evident after 6-8 h of LPS stimulation, whereas IL-10 suppressed TNF production by 80-90% at 1 h and exerted maximal suppression by 7-10 h of LPS stimulation (Fig. 2 A ) . Exposure of macrophages to TGF-@ for 4-6 h prior to LPS strongly enhanced the capacity of TGF-8 to inhibit TNF production (Fig. 2B). For example, preincubation with 10 ng/ml TGF-@ for 8-16 h caused 92.9 f 0.9% (20 & 3-fold) suppression (24 experiments). In contrast, pretreatment of macrophages with IL-10 prior to LPS only increased suppression of TNF release if the concentration of IL-10 was otherwise suboptimal (Fig. 2B). The enhanced suppression after preincubation with TGF-@ persisted when free TGF-/3 was removed by washing the cells prior to the addition of LPS. For IL-10, in contrast, TNF suppression was strikingly reduced if the macrophages were exposed to IL-10 only prior to, but not during, LPS stimulation (Fig. 2C). Moreover, TGF-/3 caused significant suppression of TNF production even when added as late as 6 h after LPS, whereas IL-10 failed to do so if added later than 2 h (Fig. 20). Finally, IFN-7 countered the enhanced suppression of TNF release  seen after pretreatment of macrophages with TGF-@. However, IFN--y did not overcome the basal level of suppression in macrophages treated simultaneously with LPS and TGF-/3 (not shown) and was unable to reverse the suppression induced by IL-10 (Fig. 2E). These numerous differences between the effects of IL-10 and TGF-@ suggested that IL-10 and TGF-@ may suppress TNF release by different mechanisms.
TGF-@ ana' IL-10 Do Not Selectively Inhibit TNF Release, Cause Uptake of Secreted TNF, nor Induce TNF Degradation-Prior studies (15, and data not shown) ruled out that IL-10 or TGF-@ induce secretion of a TNF inhibitor or decrease the sensitivity of the target cells used to assay TNF. Therefore, we next asked if TGF-/3 and IL-10 inhibited the accumulation of TNF in the extracellular medium by decreasing its secretion, promoting its uptake, or enhancing its degradation. As shown in Table 11, both IL-10 and TGF-/3 reduced TNF activity in macrophage cytosol and membrane fractions to an extent similar to the reduction in TNF activity in the culture supernatants of the same cells. Thus, neither IL-10 nor TGF-(3 selectively inhibited the secretion of preformed TNF. Next, we added defined amounts of recombinant murine TNF to IL-10or TGF-@-treated macrophages. After 20 h, we were able to recover just as much exogenous TNF from these cytokine-treated cultures as from control cultures that had not been exposed to IL-10 or TGF-@. Thus, neither IL-10 nor TGF-@ promoted uptake or degradation of extracellular TNF. Finally, we immunoprecipitated [35S]cysteine, biosynthetically labeled TNF from lysates of macrophages stimulated with LPS in the presence or absence of TGF-/3 or IL-10. At 4 h of LPS stimulation, the amounts of 26-kDa TNF precursor Panel A , macrophages were stimulated with LPS (0.5 pg/ml) in the absence or presence of TGF-8 (10 ng/ml) or COS cell IL-10 (10 units/ ml) for the time periods indicated (n = 3). Panel B, macrophages were incubated with TGF-P (10 ng/ml), COS cell IL-10 (10 or 0.5 units/ ml), or medium alone for the times indicated, after which LPS (0.5 pg/ml) was added to the cultures for 20 h. The TNF concentration in the control supernatants for the different periods of preincubation ranged from 18 to 200 ng/ml ( n = 2). Panel C, macrophages were exposed to TGF-8 (10 ng/ml), COS cell IL-10 (10 units/ml), or medium alone. After the times indicated, the monolayers were washed extensively and then stimulated with LPS (0.5 pg/ml) for 20 h. The TNF concentration in the control supernatants for the different periods of preincubation ranged from 5 to 30 ng/ml ( n = 2). Panel D, macrophages were stimulated with LPS (0.5 pg/ml) for 20 h. TGF-8 (10 ng/ml), COS cell IL-10 (10 units/ml), or medium was added to the cultures at the indicated times after initiation of LPS stimulation (n = 3). Panel E, macrophages were stimulated with LPS (0.5 pg/ml) for 20 h in the absence of presence of TGF-P (10 ng/ml) or COS cell IL-10 (10 units/ml) and varying concentrations of IFN-7. The TNF concentration in the control supernatants (LPS plus different concentrations of IFN-7) ranged from 4 to 24 ng/ml ( n = 2). Panels A-E, n represents the number of similar experiments.

Effect of TGF-0 and IL-IO on TNF bioactivity in macrophage cytosol, membranes, and culture supernatant
Macrophages (20 X lo6) were stimulated with LPS (0.5 pg/ml) in the absence or presence of TGF-@ (10 ng/ml) or in COS cell-expressed    (23) and 17-kDa mature TNF were markedly reduced by IL-10 treatment and slightly reduced by TGF-/3. Neither cytokine caused the appearance of immunoprecipitable degradation products of TNF (Fig. 3). In pulse-chase experiments, neither IL-10 nor TGF-/3 altered the stability of TNF protein (data not shown).

TGF-/3 and IL-10 Differentially Affect Steady-state TNF
mRNA Levels-The foregoing experiments established that TGF-/3 and IL-10 inhibit accumulation of TNF in media conditioned by LPS-stimulated macrophages by decreasing their synthesis of TNF. Thus, we next analyzed the effects of TGF-/3 and IL-10 on the expression of TNF mRNA. In accordance with previous data (31,32), stimulation of macrophages with LPS alone led to a strong expression of TNF mRNA, the levels peaking at 2-3 h. IL-10 dose dependently inhibited this increase (Fig. 4, A and B). The TNF mRNA levels were strikingly reduced even during the onset of TNF mRNA expression (Fig. 4C). Cycloheximide, which blocks protein synthesis and superinduces TNF mRNA (33), abrogated the inhibitory effect of IL-10 on TNF mRNA expression (Fig. 40). When added at the same time as LPS, TGF-/3 suppressed TNF release without decreasing TNF mRNA (Fig.  4B). When added prior to LPS, TGF-8 did suppress TNF mRNA expression, but only at later time points (210-12 h) of stimulation (Fig. 4E).
The Rapid Down-regulation of TNF mRNA Expression by IL-10 Is Not Due to Transcriptional Inhibition-LPS was reported to enhance macrophage TNF gene transcription 3fold, which contributes to the 100-fold elevated TNF mRNA levels observed after stimulation with LPS (32,34). Therefore, we performed nuclear run-on assays to determine whether IL-10 inhibits LPS-induced TNF gene transcription. In five experiments, LPS either did not alter TNF gene transcription or increased it by a factor of 52. Nuclei from macrophages stimulated with LPS in the presence of IL-10 synthesized equal or only slightly less [~~-~*P]UTP-labeled TNF mRNA than nuclei from macrophages stimulated with LPS alone (Fig. 5, A and B). These results demonstrate that the rapid and marked reduction of TNF mRNA expression after IL-10 treatment (Fig. 4, A-C) is unlikely to result from modulation of TNF gene transcription.
Effects on TNF mRNA Stability-At the earliest time points after LPS stimulation at which we detected weak expression of TNF mRNA (45 min), i.e. well before the maximal expression of TNF mRNA, IL-10 had already exerted its strong inhibitory effect (Fig. 4C). There was no time point where macrophages were cultured in medium or stimulated with LPS (0.5 pg/ml) for 1.5 h in the absence or presence of different concentrations of COS cell-IL-10.10 pg of total RNA were electrophoresed on a 0.8% agarose, formaldehyde gel and analyzed by Northern blot hybridization with the indicated radiolabeled probe. Blots were exposed to xray film for 20 h (TNF) or 40 h (@-actin). The amount of TNF in the culture supernatants was determined as detailed in the legend to Fig.   1. Panel B, macrophages were stimulated with LPS (0.5 pg/ml) in the absence or presence of COS cell IL-IO (10 units/ml) or TGF-0 (10 ng/ml) for 3, 10, or 20 h. 10 pg of total RNA were processed as described above. The same blot was hybridized sequentially with TNF-, @-actin-, IL-l@-and IL-la-specific probes (exposure time 21, 24, 16, and 72 h, respectively). Panel C, macrophages were cultured in medium or stimulated with LPS (0.5 pg/ml) in the absence or presence of COS cell IL-10 (10 units/ml) for 45, 90, or 125 min. 10 pg of total RNA were processed as described above and exposed to xray film for 15 (TNF), 34 (&la), 5 (&I@), and 36 h @-actin). Panel D, macrophages were stimulated with LPS (0.5 pg/ml) for 3 h in the absence or presence of IL-10 and cycloheximide (CHX) (10 pglml). 12 pg of total RNA were processed as described above and exposed to x-ray film for 20 h. The ethidiumbromide (EtBr) stain of the agarose gel demonstrates uniformity of RNA loading. Panel E, Macrophages were cultured in medium alone or with TGF-0 (10 ng/ ml) for 12 h prior to stimulation with LPS (0.5 pg/ml) for 20 h. 15 pg of total RNA were analyzed as described above and exposed to xray film for 48 h (TNFprobe) or 5 d (@-actin probe). TNF concentration in the culture supernatants was determined as in the legend to Fig. 1. TNF mRNA was equally expressed in control and IL-10 treated m4, and therefore it was not possible to determine reliably the half-life of TNF transcripts during the onset of IL-10-dependent suppression. However, such experiments were feasible with TGF-j3, which only diminished the level of TNF mRNA late after the onset of stimulation of macrophages with LPS, provided the macrophages were pretreated with TGF-j3. Accordingly, we preincubated macrophages in medium or TGF-j3 (10 ng/ml) for 2-13 h, stimulated them with LPS for 70-120 min, and then inhibited RNA synthesis with actinomycin D for different periods of time. The amount of TNF mRNA was then estimated by Northern analysis. Fig.  6 illustrates one of four such experiments in which preincubation with TGF-j3 decreased the half-life of LPS-induced TNF mRNA about 2-fold.

IL-10, but Not TGF-j3, Is a Potent Inhibitor of Macrophage
IL-1 Production-Supernatants from macrophages stimulated with LPS in the presence of IL-10 or TGF-8 were tested for IL-1 bioactivity in the D10G4.1 assay. Control experiments demonstrated that the concanavalin A/IL-1-triggered proliferation of the D10G4.1 T-cell clone was unaffected by either COS cell-expressed IL-10 or TGF-j3.3 IL-10, but not supernatant from mock-transfected COS cells, inhibited macrophage IL-1 release in a concentration-dependent manner (Fig. 7). The extent of suppression was the same at 3,9, and 20 h of LPS stimulation (Fig. 4B). The concentration of IL-10 required for 50% suppression averaged 0.35 +: 0.12 unit/ ml (three experiments) and was therefore approximately 10fold higher than for suppression of TNF release (0.04 unit/ m l k 0.01 unit/ml, four experiments). TNF has been reported to induce IL-1 in murine macrophages (35,36). Therefore, we tested the possibility that the suppression of IL-1 release by IL-10 might be mediated through its suppression of TNF production. In our hands, recombinant TNF (1.6-160 ng/ml) did not induce IL-1 release, nor did the addition of anti-TNF antibody to the culture of LPS-stimulated macrophages diminish the production of IL-1, although the antibody neutralized the TNF in the same supernatants (not shown). Recombinant TNF (1.6-160 ng/ ml) failed to reverse the suppression of IL-1 release by IL-10 (not shown). Thus, the suppression of IL-1 production by IL-10 was independent of its suppression of TNF release.
Northern blot analysis showed that IL-10 potently inhibited the LPS-induced accumulation of IL-la and IL-lj3 mRNA before and at the time of their maximal expression, as well as indicated, total RNA was isolated, equal amounts were slot-blotted to nitrocellulose, hybridized to the TNF probe, and exposed to x-ray film for 12 h. The autoradiographic signal was quantified by videodensitometry. The plotted values represent % of the initial hybridization signal after 120 min of LPS stimulation (before addition of actinomycin D). later during the course of stimulation (Fig. 4, B and C). The marked increase in IL-la and IL-1P mRNA after LPS stimulation is at least partially due to post-transcriptional mechanisms, as LPS reportedly elicits only a slight increase of ILl a and IL-lP gene transcription (37). Our transcriptional assays confirmed these findings for IL-la, and we observed only a modest decrease in IL-la gene transcription after IL-10 treatment (Fig. 5, A and B). In the case of IL-1B, there was a marked increase of transcriptional activity after LPS stimulation, but this was also not significantly altered in the presence of IL-10 (Fig. 5). Thus, it appears that IL-10 regulates LPS-induced IL-la and IL-10 mRNA expression predominantly at a post-transcriptional level.
In contrast to IL-10, TGF-/? (0.001-100 ng/ml) added to the macrophages simultaneously with LPS did not affect macrophage IL-1 production ( Fig. 4B and Fig. 7) and IL-la and IL-lP mRNA expression (Fig. 4B). IL-la and IL-1P gene transcriptions were also not altered (not shown). If the macrophages were preincubated with 10 ng/ml TGF-P for 15 h prior to stimulation with 0.25 or 1.0 pg/ml LPS, IL-1 release was suppressed slightly (53.2 f 5.4% in three experiments).

DISCUSSION
The suppressive effects of TGF-/3 and IL-10 on the capacity of primary macrophages to produce TNF and IL-1 differed by at least seven functional criteria. 1) At their most effective concentrations, IL-10 caused an approximately 25-fold greater suppression of TNF production than TGF-P. 2) For TGF-P, suppression of TNF release was substantial only after 12-16 h of stimulation, whereas for IL-10, TNF release was suppressed 90% or more as soon as 3 h after stimulation with LPS. 3) Maximal suppression of TNF production by TGF-P, but not by IL-10, required pretreatment with the cytokine prior to LPS stimulation, 4) TGF-P caused the same degree of TNF suppression when macrophages were exposed to TGF-0 only prior to LPS stimulation or throughout the period of exposure to LPS. TNF suppression by IL-10, in contrast, was largely dependent on the simultaneous presence of LPS. 5) TGF-P suppressed TNF production even if added up to 6 h after LPS, whereas macrophages stimulated by LPS for more than 2 h were completely resistant to the suppressive effect of IL-10. 6) IFN-? partially reversed TGF-P-but not IL-10mediated suppression of TNF production. 7) IL-10, but not TGF-8, potently suppressed the production of IL-1.
These functional differences suggested that IL-10 and TGF-B affect macrophage cytokine production by distinct mechanisms. The huge increase of TNF production after LPS stimulation of macrophages has been ascribed to three independent processes: acceleration of TNF gene transcription (factor of 3), increase of TNF mRNA levels (factor of loo), and rise in TNF protein secretion (factor of 10,000) (31,32,34). Consequently, we studied the effect of IL-10 and TGF-@ on TNF at the transcriptional, post-transcriptional, and translational/post-translational levels. Suppression of TNF protein in the extracellular and intracellular compartments after IL-10 treatment could be attributed largely to the marked (15-40-fold) reduction of TNF mRNA levels from early time points (45 min) of stimulation onward. Likewise, the suppression of IL-1 release by IL-10 was associated with strong reduction in IL-la and IL-16 mRNAs. The exact mechanism of suppression of TNF and IL-1 mRNA expression by IL-10 is not yet clear. However, the process depends on de mu0 protein synthesis and is not attributable to inhibition of TNF or IL-1 gene transcription. It seems highly likely, therefore, that IL-10 promotes the rapid degradation of TNF and IL-1 transcripts. This notion is supported by short time kinetic experiments, which demonstrate that IL-10 inhibits the LPS-induced cytokine mRNA up-regulation at its onset rather than just causing degradation of already expressed mRNA. Enhanced TNF mRNA expression after LPS stimulation is largely caused by post-transcriptional events (32,34). The 3"untranslated regions of TNF, IL-la, and IL-lP mRNA contain several UA-rich octameric units (UUAUUUAU), which confer instability of the messages (38-41). LPS-stimulated macrophages express a ribonucleolytic activity which selectively degrades mRNAs with the conserved UA-rich sequence element in their 3"untranslated regions (41). It is possible that IL-10 increases the expression of this nuclease or the expression of a protein that enhances the susceptibility of TNF, IL-la, and IL-16 mRNAs to its action.
For TGF-P, the mechanism of suppression of TNF produc-tion differed depending on the stimulation conditions. Without preincubation, TGF-@-mediated suppression of TNF release was not accompanied by changes in TNF mRNA. Under those conditions, TGF-@ most likely reduced TNF mRNA translation. This notion is supported by the fact that TGF-(3mediated suppression of TNF production increased during the period of stimulation with LPS and was also observed if TGF-(3 was added to the macrophages considerably later than LPS. Preincubation with TGF-@, in contrast, led to a reduction in TNF mRNA levels at later time points of stimulation, in association with decreased stability of TNF mRNA. It is possible that pretreatment with TGF-P induces the same nucleolytic activity as postulated above for IL-10, but to a much lesser extent. TGF-(3 has been reported to stimulate human mononuclear leukocytes to accumulate mRNA for TNF and IL-1 (42-45) as well as to secrete TNF and IL-1 protein (42). In inflammatory peritoneal murine macrophages, TGF-(3 (10 ng/ml) neither elicited IL-1 nor TNF protein production nor induced the respective cytokine mRNAs at 1, 3, 8 or 20 h.3 When human peripheral blood mononuclear cells were stimulated with LPS, the release of TNF and IL-1 was inhibited by TGF-/3 (45), as with murine macrophages in this and another study (10). Thus, the action of TGF-@ appears to vary with the activation state and species of the mononuclear cells. A similar situation may pertain for IL-10. Although IL-10 appears to suppress cytokine production by LPS-stimulated human monocytes (14) and resting (13) and inflammatory murine macrophages (15, and this paper) equally well, IL-10 downregulates class I1 major histocompatibility complex antigens on human monocytes (46) but not on mouse macrophages (47). TGF-(3 and IL-10 have a broad spectrum of functions. Both display inhibitory as well as stimulatory effects on diverse cells (11,48) and cannot be categorized simply as "suppressors'' or "activators." However, on inflammatory murine macrophages, TGF-(3 and IL-10 appear to have a predominantly deactivating effect. Nonetheless, TGF-@ and IL-10 suppress macrophage cytokine production under different conditions and by different mechanisms. Thus, IL-10 and TGF-@ might complement each other during the regulation of macrophagedependent responses in vivo. de Chastonay (BACHEM Bioscience, Philadelphia), T. Hamilton