Divergent Mechanisms Utilized by SOCS3 to Mediate Interleukin-10 Inhibition of Tumor Necrosis Factor α and Nitric Oxide Production by Macrophages*

The cytokine interleukin-10 (IL-10) potently inhibits macrophage function through activation of the transcription factor STAT3. The expression of SOCS3 (suppressor of cytokine signaling-3) has been shown to be induced by IL-10 in a STAT3-dependent manner. However, the relevance of SOCS3 expression to the anti-inflammatory effect of IL-10 on macrophages has been controversial. Through kinetic analysis of the requirement for SOCS3 in IL-10 inhibition of lipopolysaccharide (LPS)-stimulated tumor necrosis factor-α (TNFα) transcription and translation, SOCS3 was found to be necessary for TNFα expression during the early phase, but not the late phase of IL-10 action. SOCS3 was essential for IL-10 inhibition of LPS-stimulated production of iNOS (inducible nitric-oxide synthase) protein and nitric oxide (NO). To determine the domains of SOCS3 protein important in mediating these effects, SOCS3–/– macrophages were reconstituted with SOCS3 mutated for the SH2, KIR, SOCS box domains, and tyrosines 204 (Tyr204) and 221 (Tyr221). The SH2 domain, SOCS box, and both Tyr204 and Tyr221 were required for IL-10 inhibition of TNFα mRNA and protein expression, but interestingly the KIR domain was necessary only for IL-10 inhibition of TNFα protein expression. In contrast, Tyr204 and Tyr221 were the only structural features of SOCS3 that were necessary in mediating IL-10 inhibition of iNOS protein expression and NO production. These data define SOCS3 as an important mediator of IL-10 inhibition of macrophage activation and that SOCS3 interferes with distinct LPS-stimulated signal transduction events through differing mechanisms.

Macrophages form the first line of defense against foreign pathogens by producing pro-inflammatory mediators that either directly attack the invading organism, or recruit and activate other cells of the innate and acquired immune systems (1). Macrophage recognition of pathogens is mediated by pattern-recognition receptors, including the members of the Toll-like receptor family (TLR), 3 which recognize microbial structures forming pathogen-associated molecular patterns (2). A well characterized pathogen-associated molecular pattern is the cell wall component of Gram-negative bacteria called lipopolysaccharide (LPS). Binding of LPS to the TLR4 complex on macrophages leads to rapid macrophage activation. Whereas this inflammatory response is usually helpful and protective to the host, if left unchecked, the excessive amount of inflammatory cytokines produced leads to tissue destruction and pathologies such as septic shock (3).
Interleukin-10 (IL-10) is a key physiological negative regulator of macrophage activation (4). Activated macrophages produce various inflammatory mediators including tumor necrosis factor-␣ (TNF␣) and nitric oxide (NO), all of which are suppressed by IL-10 (5, 6). Various animal model studies have substantiated the in vivo importance of an anti-inflammatory role for IL- 10. Mice in which the IL-10 gene is deleted develop inflammatory bowel disease (7) and exaggerated immune reactions when challenged with antigen or LPS (8,9). Administration of IL-10, however, ameliorates disease in models of endotoxemia (10), transplantation (11), and autoimmunity (12). In humans, the presence of elevated endogenous IL-10 is a positive prognostic variable in autoimmune disease (13) and allogeneic transplant patients (14).
IL-10 initiates its actions on target cells by binding to the heterodimeric IL-10 receptor that belongs to the class II cytokine receptor family (4,15). Receptor binding leads to activation of Jak1 and Tyk2 kinase followed by the STAT3 transcription factor. STAT3 has been shown to be essential for the IL-10 inhibition of macrophage inflammatory mediator production (15)(16)(17) and cell cycle progression (18,19). A number of IL-10-regulated genes have been identified as being regulated by STAT3, including the NF-B family member BCL3, which is important in inhibition of TNF␣ transcription (20), and the cell cycle inhibitors p19INK 4d and p21 WIF1 , which are required for IL-10 inhibition of macrophage proliferation (19). A member of a family of negative regulators of cytokine signaling SOCS3 (suppressors of cytokine signaling-3) is also induced by IL-10 in a STAT3-dependent manner (21)(22)(23).
The SOCS family of proteins is transcriptionally regulated by a broad range of extracellular ligands and function in a classic feedback loop to negatively regulate signal transduction through multiple cytokine and growth factor receptors (24 -26). The family is comprised of eight proteins that possess a 70-amino acid region of homology referred to as the SOCS box, as well as a SH2 domain. SOCS1 and SOCS3 additionally possess a 12-amino acid-extended SH2 domain designated the kinase inhibitory region (KIR) (27). SOCS proteins have been described to interfere with cytokine and growth factor signal transduction through a number of mechanisms. SOCS1 and SOCS3 can bind to the catalytic site of Jak family kinases (28,29) and inhibit their activity. SOCS proteins can also interact with receptor phosphotyrosines through their SH2 domains (29 -32) and thus block subsequent binding of receptor downstream signaling molecules. SOCS proteins have also been reported to target signaling molecules for ubiquitin-mediated degradation through their SOCS box-mediated interaction with Elongins B and C (27) and this complex recruits the E2 ubiquitin-conjugating enzyme, forming the E3 ubiquitin-ligase (33). The proximity of this ubiquitin ligase to SOCS-recruited signaling molecules can result in the proteasomal degradation of the latter (34,35), although association with Elongins B/C has been shown in some systems to protect SOCS protein from degradation (33, 36 -38).
The role of SOCS3 in mediating IL-10 inhibition of macrophage function has been controversial. SOCS3 expression has been speculated to participate in inhibition of macrophage activation, not only because of its documented negative regulatory role in other systems, but also because of the observation that survival of many intracellular macrophage pathogens is associated with their ability to up-regulate SOCS3 expression (39 -41). Ectopic expression of SOCS3 in a mouse macrophage line inhibited inflammatory cytokine production and induced IL-1RA in a manner reminiscent of the effect of IL-10 on these cells (42). This study also described diminished IL-10 responsiveness in peritoneal macrophages isolated from SOCS3 Ϫ/ϩ mice. In contrast, other investigators using antisense methodology (43) or macrophage cells isolated from SOCS3 Ϫ/Ϫ animals (22,44) observe no effect of SOCS3 deletion on the ability of IL-10 to inhibit TNF␣ production.
We now show that this discrepancy may be in part because of the time point at which IL-10 function is analyzed. Our studies indicate that IL-10 switches from an early SOCS3-dependent to a later SOCS3-independent mechanism for inhibition of TNF␣. Furthermore, we use structure/function analysis of the SOCS3 protein to demonstrate that inhibition of TNF␣ transcription and protein expression, as well as iNOS and nitric oxide production, involve distinct SOCS3-dependent mechanisms.

MATERIALS AND METHODS
Reagents-All reagents were obtained from Sigma unless otherwise indicated. TNF␣ ELISA kits were obtained from BD Biosciences. Antibodies to SOCS3 protein and phospho-SOCS3 protein (phosphotyrosine 204) were kindly supplied by Dr. Nicholas Cacalano (University of California, Los Angeles). IL-10 was kindly supplied by Dr. Kevin Moore (DNAX Research Institute).
Cell Culture-The J774.1 murine macrophage cell line (American Type Tissue Culture Collection) was cultured in Dulbecco's modified Eagle's medium supplemented with 9% (v/v) fetal calf serum on tissue culture grade plates. SOCS3 ϩ/Ϫ and SOCS3 Ϫ/Ϫ macrophage cell lines were established by culturing embryonic fetal liver hematopoetic progenitor cells (45) in colony-stimulating factor-1 (CSF-1) media (Iscove's modified Dulbecco's medium supplemented with 10% (v/v) fetal calf serum and 5 ng/ml CSF-1) in the presence of retroviruses encoding the SV40 large T antigen. After 2 weeks, isolated colonies were expanded and analyzed for CD11b and CD86 expression. Clones were also tested for their ability to produce TNF␣ in response to LPS. A panel of clones was selected for both SOCS3 ϩ/Ϫ and SOCS3 Ϫ/Ϫ cells that express comparable levels of CD11b and LPS-stimulated TNF␣. All cells were grown at 37°C and 5% CO 2 in a standard tissue culture incubator.
Akihiko Yoshimura (29). Viruses were generated by transfecting these constructs individually into the Plat-E viral packaging cell line using Lipofectamine (Invitrogen)-mediated transfection as described previously (19). Viral supernatants were made 5 ng/ml in CSF-1 and 8 g/ml in protamine sulfate and layered onto a 10-cm dish containing 2 ϫ 10 6 SOCS3 Ϫ/Ϫ cells. The viral supernatant was removed after 24 h and the cells allowed to grow for several days prior to isolation of green fluorescent protein-positive cells by flow cytometry. After sorting, these green fluorescent protein-positive cells were maintained in CSF-1 media.
Northern Blot Analysis-SOCS3 ϩ/Ϫ and SOCS3 Ϫ/Ϫ cells following treatment with LPS (100 ng/ml) or LPS ϩ IL-10 (10 ng/ml) were harvested in 1 ml of TRIzol (Invitrogen) and RNA was isolated according to the manufacturer's recommendations. Equivalent amounts of RNA (10 -20 g) were resolved by electrophoresis on 1% formaldehyde-agarose gels blotted onto a nylon membrane (Pall, Mississauga, Ontario) and cross-linked by exposure to UV light. The membranes were probed with [␣-32 P]dCTP random prime-labeled cDNA probes for TNF␣ and GAPDH as described previously (19).
Measurement of TNF␣ and NO Production-SOCS3 ϩ/Ϫ and SOCS3 Ϫ/Ϫ macrophages were plated at 2 ϫ 10 5 cells/well in 24-well plates in CSF-1 media and grown overnight. The media was changed to Dulbecco's modified Eagle's medium containing 9% fetal calf serum and cells were stimulated with LPS (100 ng/ml) or LPS ϩ mIL-10 (10 ng/ml) for the indicated times. Supernatants were collected and analyzed for the presence of TNF␣ protein using a mouse TNF␣ ELISA kit (BD Biosciences) according to the manufacturer's recommendation. The NO content of the supernatants were determined using the Greiss assay (48).
Statistical Analysis-Data were expressed as the mean Ϯ S.E. Statistical evaluation was performed using one-way analysis of variance and p values Յ 0.01 were considered significant.

Induction of SOCS3 Message by IL-10 in a STAT3-dependent Manner-
From a survey of SOCS family members, we found that IL-10 treatment of J774.1 macrophage cells induces SOCS3 mRNA expression. As shown in Fig. 1, SOCS3 mRNA is induced by IL-10 as early as 30 min and this expression is sustained for as long as 10 h post-stimulation (data not shown). To confirm the role of STAT3 in IL-10 regulation of SOCS3, we also examined the effect of ⌬STAT3, a dominant inhibitory STAT3 lacking the C-termi-nal transactivation domain as we described previously (18), on IL-10 induction of SOCS3. Expression of ⌬STAT3 inhibited the ability of IL-10 to induce SOCS3 mRNA. In contrast, an analogous mutant of STAT1, ⌬STAT1, did not inhibit IL-10 induction of SOCS3, indicating that in these cells SOCS3 mRNA expression is STAT3-dependent.
STAT3 activation by IL-10 requires two cytoplasmic tyrosines conserved in both the mouse and human (Tyr 427 /Tyr 477 for mouse; Tyr 446 / Tyr 496 for human) IL-10R1 subunit. Using J774.1 cells expressing the wild-type (hIL-10R:Wt) or Tyr 446 /Tyr 496 mutant hIL-10R1 (hIL-10R1: Tyr FF ), as we previously described (18), we tested the ability of the wildtype or mutant hIL-10R to induce SOCS3 mRNA. Cells were stimulated with hIL-10 in the presence of a mIL-10R1 blocking antibody to block the endogenous IL-10R1 of the mouse (18). Although hIL-10R:WT was capable of transducing a signal leading to SOCS3 mRNA expression, the hIL-10R:Tyr FF were impaired in their ability to induce SOCS3 mRNA ( Fig. 1) in response to hIL-10.
Induction of SOCS3 Protein by IL-10-We next confirmed that IL-10 induces SOCS3 protein expression. SOCS3 protein is known to be very unstable (47) and we found it necessary to pre-treat cells for 30 min with the proteasome inhibitor MG132 to prevent SOCS3 protein degradation during preparation of the cell lysate. In the presence of MG132, SOCS3 protein expression could be detected in cells treated with IL-10 alone beginning at 60 min and increasing for up to 3 h (Fig. 2). LPS treatment alone did not result in SOCS3 protein expression, but addition of LPS along with IL-10 did enhance the amount of SOCS3 present resulting in detectable SOCS3 by 30 min of stimulation.
IL-10 Induces Tyrosine Phosphorylation of SOCS3 Protein on Residue Tyr 204 -Cacalano et al. (47) were the first to report that SOCS3 protein becomes tyrosine phosphorylated in response to certain cytokines and growth factors and that this phosphorylation is important for SOCS3 function (47). Tyrosine residues 204 and 221 were shown in fibroblasts to be necessary for SOCS3 to interact with and inhibit RasGAP in response to epidermal growth factor stimulation (47). The Cacalano group has since generated specific antibodies recognizing phosphorylated Tyr 204 but not the unphosphorylated residue (Fig. 3A). We used these antibodies to examine whether IL-10 stimulation results in phosphorylation of SOCS3 on Tyr 204 . To stabilize SOCS3 tyrosine phosphorylation, J774.1 cells were pretreated with 10 M sodium orthovanadate and 5 M MG132 for 30 min prior to stimulation for 2 h with 100 ng/ml of LPS ϩ IL-10 (100 ng/ml) or IL-10 alone. As shown in Fig. 3B, IL-10 treatment resulted in phosphorylation of SOCS3 protein on Tyr 204 at 2 h.
Resistance of SOCS3 Ϫ/Ϫ Cells to IL-10 Inhibition Is Time-dependent-LPS stimulation of macrophages results in production and secretion of TNF␣ and IL-10 has been reported to inhibit this at multiple levels targeting both transcription (23,49) and post-transcriptional mechanisms (49,50). To characterize the potential role of SOCS3 protein in IL-10 inhibition of TNF␣ mRNA and/or protein expression, we compared the ability of IL-10 to inhibit TNF␣ production from LPS-stimulated SOCS3 Ϫ/Ϫ versus control SOCS3 ϩ/Ϫ macrophage cell lines. Cells were treated with LPS Ϯ IL-10 and supernatants collected after 2 or 4 h of stimulation. As shown in Fig. 4, the SOCS3 Ϫ/Ϫ and SOCS3 ϩ/Ϫ cells produce similar amounts of TNF␣. IL-10 is equally able to inhibit TNF␣ production down to 10-20% of the LPS-stimulated level in both the SOCS3 ϩ/Ϫ versus SOCS Ϫ/Ϫ macrophages at the 4-h time point (Fig. 4B). However, at the 2-h time point, IL-10 inhibits TNF␣ expression to 40% of LPS alone levels in SOCS3 ϩ/Ϫ but not in SOCS3 Ϫ/Ϫ macrophages (Fig. 4A). To confirm the delayed TNF␣ inhibition in these cells is truly because of the absence of SOCS3 protein, we reconstituted the SOCS3 Ϫ/Ϫ cells with a wildtype SOCS3 cDNA via retroviral transduction. SOCS3 Ϫ/Ϫ cells reconstituted with SOCS3 cDNA regained their responsiveness to IL-10 and were able to inhibit TNF␣ production at 2 h (Fig. 4A, SOCS3:WT). These results suggest that during the early phase of IL-10 inhibition of TNF␣ production, IL-10 requires SOCS3 to mediate its action.
Requirement of Various SOCS3 Domains in Mediating IL-10 Inhibition of TNF␣ Protein Expression-SOCS3 has three conserved regions: the SOCS box, SH2, and KIR domains, as well as two tyrosine residues shown to be important for its function in other cytokine systems. To examine the involvement of each of these domains in mediating the inhibitory effect of IL-10 on TNF␣ protein expression, we retrovirally transduced SOCS3 Ϫ/Ϫ macrophage cell clones with myc epitopetagged wild-type or mutant SOCS3 cDNAs (described in Ref. 29) using infection methods devised for high efficiency gene transfer into macro-   phage cells (18). Transduced cells were sorted based on green fluorescent protein expression and cell lysates were prepared to confirm expression of SOCS3 protein (Fig. 5, lanes 3-9). Lysates were also prepared from the parental SOCS3 ϩ/Ϫ and SOCS3 Ϫ/Ϫ cells (Fig. 5, lanes 1 and 2) treated with 100 ng/ml of IL-10 for 1 h to induce expression of endogenous SOCS3 protein. The expression of each SOCS3 protein variant was confirmed to be similar to each other and similar to endogenous levels induced by IL-10. The blots were re-probed for ERK protein to make sure there was an equal amount of protein in each sample.
We then assessed the impact of the various domain mutations on the ability of SOCS3 to mediate the inhibitory effect of IL-10 on LPS-stimulated TNF␣ protein expression. Cells were stimulated with 100 ng/ml LPS Ϯ IL-10 (10 ng/ml) for 2 h, and TNF␣ levels in the culture supernatant were determined by ELISA (Fig. 6). Only SOCS3 Ϫ/Ϫ cells reconstituted with wild-type SOCS3 regained responsiveness to IL-10, whereas those reconstituted with SH2, KIR, or the SOCS box domain, Y204F or Y221F mutants were resistant to IL-10. This suggests that for inhibition of TNF␣ protein release, IL-10 requires the KIR domain and SOCS box domain including the two tyrosines, Tyr 204 /Tyr 221 .
Requirement of Various SOCS3 Domains in Mediating IL-10 Inhibition of TNF␣ mRNA-To determine whether IL-10 inhibition of TNF␣ mRNA required SOCS3 protein expression, SOCS3 ϩ/Ϫ and SOCS3 Ϫ/Ϫ mutants expressing macrophages were treated with 100 ng/ml of LPS ϩ IL-10 (10 ng/ml) for 2 h and total RNA was prepared for Northern analysis. Fig. 7 shows that expression of TNF␣ mRNA is inhibited in SOCS3 ϩ/Ϫ , SOCS3 Ϫ/Ϫ reconstituted with wild-type, and KIR domain mutants, but not in parental SOCS3 Ϫ/Ϫ macrophages, SOCS3 Ϫ/Ϫ cells reconstituted with control vector, SH2 domain mutant, SOCS box domain mutant, or Y204F/Y221F. Re-probing the same Northern blot for GAPDH mRNA confirmed equal RNA loading. This indicates that IL-10 requires the SH2 domain and SOCS box domain including the two tyrosines, Tyr 204 /Tyr 221 of the SOCS3 protein to inhibit TNF␣ mRNA expression. IL-10 thus regulates TNF␣ transcription and translation through different domains of SOCS3 protein.
Tyr 204 /Tyr 221 of SOCS3 Protein Is Important for Inhibition of NO Production by IL-10-NO is another important proinflammatory mediator produced by activated macrophages, but its expression is regulated differently than that of TNF␣ (51)(52)(53). We examined whether inhibition of NO production by IL-10 is SOCS3-dependent. Parental SOCS3 ϩ/Ϫ and SOCS3 Ϫ/Ϫ macrophages were treated with 100 ng/ml of LPS ϩ IL-10 (10 ng/ml) for 24 h, and the levels of NO in the supernatants were determined. As shown in Fig. 8, IL-10 was able to inhibit NO production in SOCS3 ϩ/Ϫ cells (38% of LPS stimulated levels), but this inhibition was significantly impaired in SOCS3 Ϫ/Ϫ cells (80% of LPS stimulated levels). To ensure that this effect is due solely to the absence of SOCS3 protein, SOCS3 Ϫ/Ϫ cells reconstituted with SOCS3 cDNA were tested as well  and were shown to have a restored responsiveness to IL-10, whereas the SOCS3 Ϫ/Ϫ cells reconstituted with an empty vector behave like parental SOCS3 Ϫ/Ϫ cells. Next, SOCS3 Ϫ/Ϫ cells reconstituted with various SOCS3 domain mutants were examined to determine which of these regions are important for the ability of IL-10 to inhibit NO production. Fig. 8 shows that IL-10 is able to inhibit NO production in SOCS3 Ϫ/Ϫ macrophages reconstituted with SOCS box domain, SH2 domain, and KIR domain mutants, but not in SOCS3 Ϫ/Ϫ cells reconstituted with the Y204F and Y221F mutants.

IL-10 Requires Tyr 204 and Tyr 221 of SOCS3 for Inhibition of iNOS
Protein Expression-Because we observed NO inhibition by IL-10 in a SOCS3-dependent manner, we investigated whether the inhibition of NO by IL-10 is because of down-regulation of inducible nitric-oxide synthase (iNOS), which catalyzes the production of NO from L-arginine. Parental SOCS3 ϩ/Ϫ and SOCS3 Ϫ/Ϫ cells, as well as the SOCS3 Ϫ/Ϫ reconstituted with various SOCS3 constructs were treated with 100 ng/ml of LPS Ϯ IL-10 (10 ng/ml) for 24 h and cell lysates were prepared for determination of iNOS protein by Western analysis. As shown in Fig. 9, iNOS protein expression mirrors the NO production data with inhibition of iNOS protein by IL-10 observed in SOCS3 ϩ/Ϫ but not in SOCS3 Ϫ/Ϫ cells. IL-10 was also able to inhibit induction of iNOS expression in SOCS3 Ϫ/Ϫ macrophages reconstituted with wild-type SOCS3, SOCS box domain mutant, SH2 domain mutant, and KIR domain mutant. However, SOCS3 protein lacking either Tyr 204 or Tyr 221 was not able to support IL-10 inhibition of the iNOS protein. The blots were re-probed with antibody against ERK to confirm that equal amounts of protein were loaded for each sample. These data show that IL-10 requires Tyr 204 and Tyr 221 to inhibit iNOS expression, which are also the same residues involved in the inhibition of NO production by IL-10.
The Tyr 204 and Tyr 221 Mutants Are More Stable Than Wild-type SOCS3-Unexpectedly, although both Tyr 204 and Tyr 221 located in the SOCS box domain are required for IL-10 inhibition of NO and iNOS protein, a SOCS3 protein entirely lacking the SOCS box was able to mediate IL-10 inhibition (Figs. 8 and 9). One explanation might lie in the involvement of Tyr 204 and Tyr 221 phosphorylation in regulating SOCS3 interaction with Elongin C in mouse macrophage and other cell lines (38). Tyrosine phosphorylation of Tyr 204 and Tyr 221 disrupts the SOCS box-mediated interaction with Elongin C, resulting in the degradation of SOCS3 (38). Thus the Y204F or Y221F mutants can still associate with Elongin C complex through their SOCS box domain, but because neither Y204F nor Y221F can be phosphorylated, SOCS3 and associated signaling proteins it may have targeted are not released for proteasomemediated degradation. The sustained interaction with Elongin C may also sequester SOCS3 and thus make it unavailable for inhibitory mech-   anisms mediated through other domains (i.e. KIR, SH2). The SOCS box domain-deleted SOCS3, on the other hand, is not able to interact with Elongin B/C complex and target proteins for ubiquitination, but the other domains of SOCS3 (i.e. KIR, SH2) remain available for mediating interference with LPS-activated signaling proteins through other inhibitory mechanisms (28 -32). In support of this hypothesis we found that the Tyr 204 and Tyr 221 SOCS3 mutant proteins are indeed more stable than the wild-type SOCS3 (Fig. 10). SOCS3 Ϫ/Ϫ cells reconstituted with wild-type, Tyr 204 , or Tyr 221 mutants were treated for 30 min with 5 M MG132 to inhibit proteasome-mediated proteolysis, or left untreated with MG312. All cells were then stimulated 100 ng/ml of IL-10 for 2 h at 37°C prior to preparation of cell lysates. The wild-type SOCS3 was clearly degraded in the absence of MG132, but the Tyr 204 and Tyr 221 proteins remain intact.

DISCUSSION
LPS potently activates macrophages by binding to the TLR4 receptor complex and initiating a series of intracellular signaling (reviewed in Refs. 54 and 55). NF-B is activated and translocates into the nucleus where it regulates the expression of TNF␣ and numerous other inflammatory mediators. The ERK and p38 MAPK Ser/Thr kinases are also acti-vated by LPS and are important for TNF␣ expression by regulating cytoplasmic transport of TNF␣ mRNA (56), mRNA stability (57,58), and removal of translational silencers (50). IL-10 has been shown by some investigators to inhibit TNF␣ expression by interfering with NF-B (59 -61) or p38 MAPK kinase activity (50), whereas other investigators have reported no effect by IL-10 on either (49,62), perhaps reflecting differences in the macrophage cell types studied (reviewed in Ref. 15). Nevertheless, one consistent consensus is that IL-10 requires activation of STAT3 to exert its anti-inflammatory action (15,23,63,64). Research has thus focused on identifying genes induced by IL-10 in a STAT3-dependent manner to define the molecular pathway by which IL-10 inhibits macrophage activation.
Whereas SOCS proteins were first described as classic feedback inhibitors of cytokine receptor signaling, it was appreciated that their induction and activation by one receptor system could serve to crossregulate a heterologous system. We and others soon discovered that IL-10 rapidly induced SOCS3 (21,42,43,65,66) in a STAT3-dependent manner (21)(22)(23). Because STAT3 activation is absolutely required for the anti-inflammatory action of IL-10 (15,23,63,64), the possibility that SOCS3 was also critical for IL-10 inhibition of macrophage function was studied by a number of investigators. Berlato et al. (42) found that con-  Fig. 5) were treated with 100 ng/ml of LPS and 10 ng/ml of IL-10 for 24 h. Lysates were prepared and subjected to immunoblot analysis with ␣-iNOS protein antibody. The blot was re-probed with ␣-ERK protein antibody to confirm equal protein loading in each sample. B, the bands were quantitated by densitometry and the ratios for iNOS protein/ERK protein were plotted. n ϭ 3. FIGURE 10. The Tyr 204 and Tyr 221 mutants are more stable than wild-type SOCS3. SOCS3 Ϫ/Ϫ cells reconstituted with wild-type, Tyr 204 , or Tyr 221 mutants were treated for 30 min with 5 M MG132 or not and then all cells were stimulated with 100 ng/ml of IL-10 for 2 h at 37°C prior to preparation of cell lysates. A, lysates were resolved by SDS-PAGE and subjected to immunoblot analysis with anti-myc to detect the SOCS3 protein or STAT3 protein as a loading control. B, the amount of SOCS3 protein remaining in the MG132untreated samples was quantitated with the Licor Odyssey imager and expressed as a percentage of the amount of protein in the MG132-treated sample. n ϭ 3.
stitutive expression of SOCS3 diminished the amount of TNF␣ protein and NO produced by the transduced cells in response to LPS stimulation. Crespo et al. (67) and Baetz et al. (68) similarly reported that constitutive expression of SOCS3 inhibited transcription of the iNOS gene. The negative role for SOCS3 in macrophage function is also suggested by the frequently observed induction of this gene by intracellular macrophage pathogens. Induced expression of SOCS3 is postulated to contribute to survival of mycobacteria (39), leishmania (40), listeria (41), and RNA viruses (69) in the macrophage host.
More recently, a number of studies cast doubt on the involvement of SOCS3 in mediating IL-10 inhibition of macrophage activation. Neither Kuwata et al. (20) nor Baetz et al. (68) reported significant effects of ectopic SOCS3 expression on LPS-stimulated TNF␣ levels. A inhibitory role for SOCS3 was also discounted by studies in which SOCS3 expression was suppressed using antisense oligonucleotides (43) or gene disruption (22,44).
Our finding that IL-10 inhibition of TNF␣ requires SOCS3 only during the early phase (Ͻ2 h) of TNF␣ inhibition provides a possible explanation for these discrepant reports. Lang (22), Kuwata (20), Yasukawa (44), and Baetz et al. (68) measured TNF␣ after 5, 6, 9, and 24 h after LPS stimulation, respectively. Lee and Chau (43), who study a late phase IL-10-induced gene called heme oxygenase-1 (HO-1), extensively pretreated their cells with IL-10 for 2-4 h prior to addition of LPS for an additional 2 h. We hypothesize that during the early phase of IL-10 signaling, IL-10 utilizes SOCS3-dependent processes to limit TNF␣ mRNA and protein production. However, at later time points (Ͼ2 h) other mechanisms are recruited. One such alternate pathway may involve HO-1, an IL-10 late response gene (43). It is not clear how mechanistically HO-1 exerts this inhibition but pre-treatment of cells and mice with IL-10 for 2-4 h prior to LPS challenge is required to best observe the contribution of HO-1 to the anti-inflammatory action of IL-10 (43). Another alternate pathway for mediating the sustained inhibitory effect of IL-10 on TNF␣ expression involves the IB family member BCL-3. Like SOCS3, BCL-3 is induced by IL-10 in a STAT3dependent manner (20) and it has been shown to be a negative regulator of TNF␣ transcription (70). Macrophages isolated from BCL-3 Ϫ/Ϫ mice are impaired in their ability to be inhibited by IL-10, but similar to HO-1, the greatest difference between wild-type and BCL-3 Ϫ/Ϫ cells is observed if cells are pretreated with IL-10 for 18 h prior to stimulation with LPS (20). One might question the physiological relevance of the contribution of SOCS3 to inhibition of TNF␣ because it appears that IL-10 can effectively, with time, restrain TNF␣ production in SOCS3 Ϫ/Ϫ mice. This could, however, simply represent the multiple levels of redundancy involved in regulating a critical physiological response.
The study of signal transduction in macrophages is further complicated by their rapid elaboration of autocrine/paracrine cytokines. For example, LPS stimulation leads to production of IFN␤, which acts in an autocrine/paracrine manner to amplify macrophage responses (68,71). Because of this effect, LPS-induced SOCS1 expression has been speculated to be involved in either directly inhibiting TLR4 signaling (72) and/or inhibiting IFN␤-stimulated pathways (73). By analogy, IL-10induced proteins may either act directly on TLR4 pathways (SOCS3, the early response) or on pathways stimulated by autocrine inflammatory cytokines (HO-1 and BCL-3, the late response). Autocrine IFN␤ is especially important for macrophage production of NO (51-53) and we found that unlike TNF␣, the IL-10 inhibition of NO is dependent on SOCS3 even though we measure NO at 24 h after stimulation, the earliest time point at which NO is detectable. The dependence on SOCS3 for IL-10 inhibition of NO is consistent with data from the only two reports that assessed macrophage NO production (42,68). Because LPS-stimulated NO production in macrophages is highly dependent on autocrine production of IFN␤ (51-53), IL-10 inhibition of NO may differ fundamentally from its regulation of TNF␣ in that it may do so via SOCS3 interference with IFN␤ signaling rather than with TLR4 signaling pathways. Our structure/function analyses indicating different domains of SOCS3 are important for IL-10 inhibition of TNF␣ mRNA, protein, or NO production lends support for the existence of divergent mechanisms.
Our data also show that an intact KIR domain is required for IL-10 to inhibit TNF␣ protein, although not mRNA expression. TNF␣ expression in macrophages is regulated at multiple levels post-transcriptionally including: mRNA transport into the cytoplasm through a Tpl2/ERKdependent pathway (56), protein translation through p38 MAPK (50), and TNF␣ converting enzyme-mediated cleavage and release of active cytokine through as yet uncharacterized kinase cascades (74). However, as has been observed in other macrophage cell types (15), neither ERK nor p38 MAPK appear to be inhibited by IL-10 in these macrophages in our hands (data not shown). The KIR domain is the region of SOCS3 that interacts with and inhibits the kinase domain of Jak2 (28). Jak2 was recently reported to be activated by within 1 min of LPS addition to RAW264.7 macrophages and shown to be necessary for activation of phosphatidylinositol 3-kinase, c-Jun N-terminal kinase (75), and autocrine production of IFN␤. Thus, SOCS3 might interfere with LPS signaling by inhibiting Jak2 activation. However, we have not been able to detect Jak2 activation in our cells. Furthermore, IFN␤ is essential for NO production and we did not observe any effect of KIR domain mutation on the ability of IL-10 to inhibit NO. Work is continuing to identify the target of the KIR domain in regulating TNF␣ expression.
Mere constitutive expression of SOCS3 protein in macrophages is not sufficient to inhibit TNF␣ protein production to the same extent as in IL-10-treated cells (Berlato et al. (42) and Fig. 2). This suggests that signaling pathways in addition to SOCS3, or an IL-10-dependent posttranslational modification of SOCS3 might be involved. In this study we have shown for the first time that IL-10, but not LPS (data not shown), induces tyrosine phosphorylation of SOCS3 and that tyrosine residues 204 and 221 are essential for inhibition of TNF␣ mRNA and protein release, as well as inhibition of NO production and iNOS expression by IL-10. Tyrosine phosphorylation of SOCS3 has been shown to be important for interaction with certain signaling molecules in other cytokine and growth factor systems (47,76). Tyrosine phosphorylation has also been shown to decrease the half-life of SOCS3 protein by disrupting the interaction between SOCS3 and Elongin C (38) releasing it for proteasome-mediated degradation.
Unexpectedly, although both Tyr 204 and Tyr 221 located in the SOCS box domain are required for IL-10 inhibition of NO and iNOS protein, a SOCS3 protein entirely lacking the SOCS box was able to mediate IL-10 inhibition (Figs. 8 and 9). As discussed under "Results, " this interesting observation may be because of the requirement of Tyr 204 and Tyr 221 phosphorylation for SOCS3 to dissociate from the Elongin C complex and subsequent degradation (38). The sustained association of SOCS3 with Elongin C may interfere with SOCS3 inhibition of LPS signaling in two ways: first, blocking SOCS3-targeted degradation of signaling molecules and second, interference with the ability of other SOCS3 domains (i.e. SH2 and KIR domains) to mediate their inhibitory actions (i.e. blocking phosphotyrosyl interations and kinase enzymatic activity, respectively) (29 -32, 77). The SOCS box deleted mutant is also not able to target proteins for degradation, because of this lack of the Elongins B/C interaction domain, however, the other domains (i.e. SH2 and KIR) of the SOCS3 protein remain available for other inhibitory mechanisms (28 -32). In support of this hypothesis we confirm that the Tyr 204 and Tyr 221 mutants are more stable than the wild-type protein.
Studies are now underway to identify LPS-activated signaling molecules that SOCS3 may be targeting.
Regardless of the molecular basis for the apparently contradictory requirement for Tyr 204 /Tyr 241 , but not the SOCS box in IL-10 inhibition of NO, it is of note that both Tyr 204 /Tyr 221 and the SOCS box are required for IL-10 inhibition of TNF␣ production. This suggests that substantially different mechanisms are involved in the SOCS3-dependent inhibition of TNF␣ and NO. This is consistent with our previous studies that suggest that the pathway by which IL-10 inhibits NO production and iNOS expression is distinct from the one that mediates inhibition of TNF␣ (18).
Our studies indicate the importance of SOCS3 protein in the antiinflammatory action of IL-10 and also have shown that inhibition of NO and TNF␣ by IL-10 is dependent on different domains of SOCS3 protein. This is the first demonstration of alternate mechanisms of action of SOCS3 protein on divergent pathways activated by the same stimulus. Work is underway to identify proteins targeted by SOCS3. A better understanding of how IL-10 inhibits macrophage activation will provide insight into development of novel therapeutic strategies for treatment of inflammatory diseases.