Re-examination of the role of suppressor of cytokine signaling 1 (SOCS1) in the regulation of toll-like receptor signaling.

Suppressor of cytokine signaling 1 (SOCS1) is an obligate negative regulator of cytokine signaling and most importantly in vivo, signaling via the interferon-gamma (IFN-gamma) receptor. SOCS1, via its Src homology 2 domain, binds to phosphotyrosine residues in its targets, reducing the amplitude of signaling from cytokine receptors. SOCS1 is also implicated in blocking Toll-like receptor (TLR) signaling in macrophages activated by TLR agonists such as lipopolysaccharide (LPS), thus regulating multiple steps in the activation of innate immune responses. To rigorously test this, we isolated macrophages from Socs1-/- mice on multiple genetic backgrounds. We found no evidence that SOCS1 blocked TLR-activated pathways, endotoxin tolerance, or nitric oxide production. However, Socs1-/-;IFN-gamma-/- mice were extremely susceptible to LPS challenge, confirming previous findings. Because LPS induces IFN-beta production from macrophages, we tested whether SOCS1 regulates IFN-alpha/beta receptor signaling. We find that SOCS1 is required to inhibit IFN-alpha/beta receptor signaling in vitro. Furthermore, the absence of a single allele encoding TYK2, a JAK (Janus kinase) family member essential IFN-alpha/beta receptor signaling, rescued Socs1-/- mice from early lethality, even in the presence of IFN-gamma. We conclude that previous reports linking SOCS1 to TLR signaling are most likely due to effects on IFN-alpha/beta receptor signaling.

SOCS1 is a member of a family of proteins that regulate cytokine signaling pathways via inhibition of key tyrosine phosphorylation events on cytokine receptors and signaling molecules such as JAK family members (1). The inhibitory effects of SOCS proteins are mediated by two domains found in all SOCS family members: an SH2 1 domain that can bind phosphorylated tyrosine residues and the SOCS box that functions as a ubiquitin E3 ligase and thus potentially directs substrate proteins to the protein degradation machinery (1). Although the biochemical events associated with SOCS function are poorly understood, overwhelming genetic evidence has demonstrated multiple members of the SOCS family are essential for the regulation of specific cytokine signal transduction events in vivo.
Mice lacking SOCS1 die 10 -20 days after birth (2)(3)(4). Dissection of the cellular and molecular mechanisms involved in the death of the Socs1Ϫ/Ϫ mice has revealed that SOCS1 is essential for the response of cells to IFN-␥ and IFN-␥ production (2,3). In the absence of SOCS1, IFN-␥ levels rise, as does the responsiveness of cells bearing IFN-␥ receptors (e.g. macrophages), causing an overwhelming inflammatory response that has been analyzed in great detail (1). The key pathway involved in this complex pathologic process is the IFN-␥-mediated phosphorylation and activation of STAT1, the central signal transduction molecule required for IFN-␥ signaling (3,5). SOCS1 itself is a target of IFN-␥-induced gene expression and thereby can function to block IFN-␥ signaling via a negative feedback loop (1). Analysis of Socs1Ϫ/Ϫ mice intercrossed with mice bearing mutations in cytokine signaling pathways has additionally revealed that SOCS1 can regulate signaling from receptors other than the IFN-␥R including those using the ␥ c chain and the IL-12 receptor (6 -10). This stands in contrast to the inhibitory effects of SOCS2 and SOCS3 that function to block a more limited number of cytokine signaling events (1).
A surprising set of results was recently published showing SOCS1 also regulates Toll-like receptor (TLR) signaling (11,12). TLRs are a group of 10 -11 surface transmembrane proteins that contribute to the sensing of conserved microbial structures, such as lipopolysaccharide (LPS), detected by TLR4 or flagellin that signals through TLR5 (13). TLRs function in higher order signaling complexes that lead, in part, to the rapid activation of NF-B and MAP kinase signaling and the subsequent production of cytokines and chemokines by cells such as macrophages and dendritic cells (13). The activation of TLR signaling is an essential and conserved mechanism to activate the innate immune system. The authors of the aforementioned papers established that the absence of SOCS1 caused increased lethality when mice were exposed to LPS and that TLR-stimulated macrophages from Socs1Ϫ/Ϫ mice overproduced nitric oxide (NO) compared with controls (11,12). They further showed that overexpression of SOCS1 in macrophages reduced TLR4 signaling (11,12). Based on these data, SOCS1 was proposed to inhibit TLR signaling although the protein(s) in the TLR signal transduction cascade targeted by SOCS1 were not identified. However, SOCS proteins are well established to target tyrosine phosphorylation events, while the TLR signaling cascades are regulated predominantly by serine/threonine phosphorylation (13), making the SOCS1mediated inhibition of TLR signaling conceptually difficult to reconcile with existing knowledge of the regulatory events that control TLR signaling. In addition, previous studies have established that SOCS1 expression is induced by TLR4 signaling but indirectly via the autocrine/paracrine effects of IFN-␣/␤ (14).
We therefore designed experiments using Socs1Ϫ/Ϫ mice and macrophages to provide support for the notion that SOCS1 can regulate TLR signaling. The results, however, demonstrate that there is no compelling evidence for such a regulatory role for SOCS1. In contrast, we find that SOCS1 is a physiologically important regulator of IFN-␣/␤ signaling via effects of TYK2, a JAK family member required for IFN-␣/␤ signaling. Since TLR signaling induces IFN-␣/␤ from macrophages challenged with LPS (14,15), we propose that previous observations linking SOCS1 and TLR signaling are most likely mediated through the regulation of IFN-␣/␤ signaling.

MATERIALS AND METHODS
Mice-Socs1Ϫ/Ϫ mice were generated from heterozygous intercrosses and have been described in detail (2). Socs1Ϫ/Ϫ mice on IFN-␥Ϫ/Ϫ, Rag2Ϫ/Ϫ, or Stat1Ϫ/Ϫ backgrounds have been described (2,16). Tyk2Ϫ/Ϫ mice were a gift of Dr. K. Shimoda (Kyushu University Graduate School of Medical Sciences, Fukuoka, Japan). Mice lacking TYK2, SOCS1, and IFN-␥, in different combinations, were derived by intercrossing to generate Socs1Ϫ/Ϫ mice with one or two alleles of IFN-␥ and Tyk2. All mice were housed in specific pathogen free facilities and used in accordance with the guidelines of the St. Jude Institutional Animal Care and Use Committee.
TLR and Interferon Signaling Studies-Macrophages were stimulated with various TLR ligands or interferons as described (19). Immunoblotting was performed as described using antibodies described (17,20).
Endotoxin Challenges-Socs1Ϫ/Ϫ, Socs1ϩ/Ϫ, or Socs1ϩ/ϩ mice on an IFN-␥Ϫ/Ϫ background (ages 8 -13 weeks) were individually weighed and challenged with Escherichia coli LPS (made to 5 mg/ml in phosphate-buffered saline) at 40 mg/kg by the intraperitoneal route (19). Mice were monitored every 4 -6 h over a 96-h period and survival recorded at each time point. Survival data were analyzed using Kaplan-Meier statistics. Stratification based on the sex of the mice yielded equivalent results.
NO Measurements-NO levels from activated macrophages were measured using the Griess assay as described (20).
Endotoxin Tolerance-Tolerance in purified macrophages was induced according to the protocols described by Vogel and colleagues (21). Cytokine (TNF-␣, IL-6, and IL-12) levels were measured in the culture supernatants by ELISA 18 h after the LPS restimulation (22).

TLR Signaling to Activate ERK and IB␣ Phosphorylation Is
Unaffected in the Absence of SOCS1-A series of unpublished preliminary studies performed in our laboratory using macrophages from Socs1Ϫ/Ϫ mice suggested SOCS1 was unlikely to regulate TLR signaling. In response to recent suggestions that the absence of SOCS1 exacerbated TLR signaling, we accordingly established assay systems to systematically test this concept. We measured TLR signaling by first testing the activation, in Socs1Ϫ/Ϫ macrophages isolated from the bone marrow of 7-9-day-old mice, of two important pathways of TLR activity: the activation of NF-B through the phosphorylation and degradation of IB␣ and the activation of p42/44 ERK phosphorylation. BMDMs from Socs1Ϫ/Ϫ or Socs1ϩ/ϩ macrophages were stimulated with LPS, CpG DNA, or dsRNA, agonists of TLR4, TLR9, and TLR3, respectively. When measured by immunoblotting using antibodies specific for the activated, phosphorylated forms of IB␣ or ERK, the absence of SOCS1 had no significant effect on the normal kinetics or qualitative amounts on the phoshorylated forms of either proteins (Fig. 1). We next isolated BMDMs from adult Socs1Ϫ/Ϫ mice on three different backgrounds to further attempt to establish if there were any differences in TLR signaling in the absence of SOCS1. BMDMs were isolated from Socs1Ϫ/Ϫ mice on Rag2Ϫ/Ϫ, IFN-␥Ϫ/Ϫ, or Stat1Ϫ/Ϫ backgrounds and stimulated with LPS or CpG. IB␣ and ERK phosphorylation was measured as described above (Fig. 2). As we found for Socs1Ϫ/Ϫ macrophages, there appeared to be insufficient evidence to support the notion that the absence of SOCS1 affected TLR signaling in terms of IB␣ or ERK phosphorylation. SOCS1 Does Not Control the Ability of Macrophages to Produce NO-Previous studies had claimed the SOCS1 was an inhibitor of NO production (11,12): a pathway dependent on the integration between TLR and IFN-␥ signaling. To test this, we isolated BMDMs from Socs1Ϫ/Ϫ mice or Socs1ϩ/ϩ mice (Fig. 3a), peritoneal inflammatory macrophages (Fig. 3b) from Socs1Ϫ/Ϫ;IFN-␥Ϫ/Ϫ mice or BMDMs from Socs1ϩ/ϩ or Socs1Ϫ/Ϫ on Rag2Ϫ/Ϫ, IFN-␥Ϫ/Ϫ, or Stat1Ϫ/Ϫ backgrounds (Fig. 3c) and stimulated them with different TLR agonists in the presence or absence of IFN-␥. NO levels were measured indirectly using the Griess method (Fig. 3). As predicted, NO output was highest in macrophages co-stimulated with TLR agonists and IFN-␥ in each macrophage population. However, the absence of SOCS1 did not appear to significantly enhance NO production in any circumstance.
SOCS1 Does Not Control the Ability of Macrophages to Become Endotoxin Tolerant-Endotoxin tolerance occurs when macrophages are exposed to LPS for hours to days and then washed and restimulated with LPS. In normal macrophages, the second stimulation is associated with a depressed ability to produce cytokines. The refractory nature of macrophages after the primary stimulation has been considered to be a significant component of host defense against systemic bacterial challenge, such as in septic shock. Previous research noted that SOCS1-deficient macrophages failed to become endotoxin tolerant (11). To test this, we performed tolerance assays using well established assay systems (21). BMDMs were stimulated with LPS for 18 h or left unstimulated, then washed and restimulated for an additional 18 h. Levels of inflammatory cytokines were measured in the supernatants by ELISA. In contrast to previous findings, we could find no evidence that the absence of SOCS1 affected the ability of macrophages to acquire tolerance to endotoxin as shown by the inhibition of IL-12 ( Fig. 4) or IL-6 and TNF-␣ (data not shown) by pretreatment with LPS.
Socs1Ϫ/Ϫ;IFN-␥Ϫ/Ϫ Mice Are Extremely Susceptible to Endotoxin Challenge-Published studies have shown that Socs1Ϫ/Ϫ mice are highly susceptible to systemic endotoxin challenge (11,12). In addition, Socs1ϩ/Ϫ;IFN-␥Ϫ/Ϫ mice were also more susceptible than Socs1ϩ/ϩ;IFN-␥Ϫ/Ϫ mice, suggesting that gene dosage of Socs1 leads to a significant effect in the response of the host to septic challenge. We performed further experiments to test these concepts. In these experiments, we used adult Socs1Ϫ/Ϫ mice on an IFN-␥Ϫ/Ϫ background to obviate the need to use mice less than 10 -15 days of age, since newborn mice are more sensitive to LPS challenge than adults (19). Groups of male and female mice were accurately weighed and injected with 40 mg/kg LPS and survival monitored over 4 days. A relatively high dose was necessary, since the absence of IFN-␥ signaling enhances endotoxin resistance (23). In agreement with previous studies, we found that Socs1Ϫ/Ϫ; IFN-␥Ϫ/Ϫ mice rapidly died compared with controls (Fig. 5). However, there was no effect of Socs1 heterozygosity in contrast to claims that loss of a single allele of Socs1 on an IFN-␥Ϫ/Ϫ background increased LPS-induced lethality (11). In contrast to previous experiments that used limited numbers of mice (n ϭ 4), our experiments used 40 Socs1ϩ/Ϫ;IFN-␥Ϫ/Ϫ mice compared with 23 Socs1ϩ/ϩ;IFN-␥Ϫ/Ϫ mice to establish that there was no evidence for any effects of Socs1 heterozygosity on the ability to withstand systemic endotoxin challenge.
SOCS1 Regulates IFN-␣/␤ Signaling-The extreme susceptibility of Socs1Ϫ/Ϫ;IFN-␥Ϫ/Ϫ mice to endotoxin challenge suggested that SOCS1 was essential to negatively regulate a pathway(s) independent of IFN-␥ that was crucial in controlling the systemic response to endotoxin. Since compelling evidence for a role of SOCS1 in directly regulating TLR signaling was lacking, we sought to substantiate a role for SOCS1 in alternative pathways relevant to innate immune responses that could contribute to endotoxin sensitivity. Upon exposure to LPS, macrophages produce high levels of IFN-␣/␤ that act as an autocrine/paracine regulators of macrophages, NK cells, and dendritic cells, all of which contribute to the innate response to LPS (14,15). IFN-␣/␤ production plays a crucial role in the regulation of the systemic inflammatory response to LPS because mice lacking mice lacking IFN-␤, which signals through the IFN-␣/␤R, are significantly resistant to LPS challenge suggesting that IFN-␤ signaling plays a pathogenic role in sepsis (24). Furthermore, TYK2, a JAK family member that regulates cytokine signaling from the IFN-␣/␤R is also involved in endotoxin susceptibility, since Tyk2Ϫ/Ϫ mice are highly resistant to LPS challenge (24,25). Therefore, we hypothesized that SOCS1 may regulate IFN-␣/␤ signaling in activated macrophages through TYK2, and this may account for the lethality observed in LPS-challenged Socs1Ϫ/Ϫ;IFN-␥Ϫ/Ϫ mice. To test this, we isolated BMDMs from Socs1Ϫ/Ϫ mice and stimulated them with IFN-␤ and measured STAT1 phosphorylation by immunoblotting. We found that the absence of SOCS1 causes a slight increase in STAT1 phosphorylation when macrophages are pulsed with IFN-␤ over time (data not shown). To confirm this finding, we next stimulated Socs1Ϫ/Ϫ macrophages with IFN-␥ or IFN-␤ for 10 min and then washed away the cytokines. The decay in total STAT1 phosphorylation was followed over time. We found that STAT1 phosphorylation induced by IFN-␤ was increased in Socs1Ϫ/Ϫ cells compared with wild-type cells, confirming the notion that SOCS1 is an inhibitor of IFN-␣/␤ signaling (Fig. 6a) in addition to its essential role in blocking IFN-␥ signaling (Fig. 6b).
Signaling through the IFN-␣/␤ receptor requires that activ- ity of both JAK1 and TYK2. Since we had obtained biochemical evidence for a role of SOCS1 in regulating IFN-␣/␤ signaling in macrophages, we next sought genetic evidence for a role of SOCS1 in regulating IFN-␣/␤ signaling. We crossed Socs1ϩ/Ϫ;IFN-␥ϩ/Ϫ;Tyk2ϩ/Ϫ mice to generate mice lacking SOCS1, TYK2, and IFN-␥ in various combinations. The breeding strategy incorporated the IFN-␥ null allele to facilitate breeding for the circumstance where TYK2 loss could not rescue the lethal phenotype resulting from the SOCS1 deficiency (3,5).
Two effects of loss of TYK2 on the lethality caused by SOCS1 deficiency were observed in the progeny of these mice (Table I). First, loss of a single allele of Tyk2 rescued Socs1Ϫ/Ϫ mice that were also IFN-␥ϩ/ϩ. These mice were weaned and still alive 40 days after birth. By comparison, Socs1Ϫ/Ϫ mice are all dead at the time of weaning, and most die by 10 days after birth. Therefore, Tyk2 heterozygosity is sufficient to overcome the early lethal effects of SOCS1 deficiency. A second effect was observed on mice that lacked SOCS1 but were IFN-␥ϩ/Ϫ. These animals die between 30 -70 days but from a pathological syndrome distinct from the Socs1Ϫ/Ϫ mice (26). We could generate Socs1Ϫ/Ϫ;IFN-␥ϩ/Ϫ;Tyk2ϩ/Ϫ or Socs1Ϫ/Ϫ;IFN-␥ϩ/Ϫ;Tyk2Ϫ/Ϫ mice that were healthy and fertile. All mice born with either of these genotypes has survived, some to well over 100 days (Table I). This suggests that loss of one or two alleles of Tyk2 can also inhibit the lethal effects of IFN-␥ in the Socs1Ϫ/Ϫ;IFN-␥ϩ/Ϫ background. Taken together, these data suggest that elimination of TYK2 rescues lethality caused by the absence of SOCS1. This suggests that SOCS1 negatively regulates TYK2 signaling and therefore IFN-␣/␤ signaling. DISCUSSION Our results suggest that SOCS1 plays no direct role in TLR signaling. We found no evidence for any effects of the absence of SOCS1 on TLR-induced signaling pathways critical for the establishment of innate immune responses. In performing these studies, we found an unanticipated role for SOCS1 in regulating IFN-␣/␤ signaling. This pathway most likely accounts for previously observed effects of the absence of SOCS1 in systemic LPS challenges.
Previous published studies (11,12) linking SOCS1 to the inhibition of TLR signaling could not be reproduced in our hands. We have applied genetic and biochemical approaches to show that BMDMs isolated from Socs1Ϫ/Ϫ mice or Socs1Ϫ/Ϫ mice on three different genetic backgrounds had normal TLR signaling to different TLR agonists. This result was expected, since there was no prior evidence to suggest that TLR signal-ing, a pathway primarily regulated by serine/threonine phosphorylation, could be inhibited by any SOCS protein, all of which bind tyrosine-phosphorylated substrates via their SH2 domains. Published data purporting a role for SOCS1 in regulating TLR signaling were in part derived from SOCS1 overexpression in macrophages (11,12). Numerous genetic studies have established that enforced expression of SOCS family members often leads to artifactual effects on many cytokine and other signaling pathways (1). Thus, the effects of SOCS1 overexpression on TLR signaling appear to fulfill a familiar pattern.

SOCS1 Regulates IFN-␣/␤ Signaling in Innate Immunity
SOCS1 was also claimed to regulate systemic endotoxin sensitivity (11,12). We found similar results using a rigorously defined endotoxin challenge protocol. We showed that Socs1Ϫ/Ϫ;IFN-␥Ϫ/Ϫ mice were highly sensitive to challenge. In contrast to previous data, however, we could find no evidence for any effects of Socs1 gene dosage on endotoxin sensitivity (11). Most likely, this stems from the low numbers of mice used by other investigators and the arbitrary administration of LPS dosage based on sex (12). In contrast, we used large numbers of mice using exact dosages based on animal weight.
IFN-␥RϪ/Ϫ mice are very resistant to endotoxin challenge suggesting an important role for IFN-␥ signaling in driving the lethal inflammation in sepsis (23). We were thus surprised that Socs1Ϫ/Ϫ;IFN-␥Ϫ/Ϫ mice were highly susceptible to LPS challenge. This suggested that SOCS1 regulated IFN-␥-independent pathway(s) in the systemic inflammatory response. Mutations other than the loss of IFN-␥ or its signaling components can rescue the SOCS1 deficiency, including loss of STAT6 or STAT4 (10,16). The absence of STAT4 on a Socs1Ϫ/Ϫ background showed that SOCS1 is a significant physiological regulator of IL-12 signaling, since this STAT member is uniquely utilized by the IL-12R (10). However, the IL-12R also uses TYK2 to activate the receptor and promote STAT4 phosphorylation (27,28). TYK2 is also used by the IFN-␣/␤R and, like IFN-␤, plays an important role in the systemic pathogenic response to endotoxin, since Tyk2Ϫ/Ϫ or IFN-␤Ϫ/Ϫ mice are highly resistant to LPS challenge (24,25). Based on these data, we concluded that TYK2 could be a target of SOCS1, accounting the sensitivity of Socs1Ϫ/Ϫ;IFN-␥Ϫ/Ϫ mice to endotoxin. To test this, we first isolated macrophages from Socs1Ϫ/Ϫ mice and challenged them with IFN-␤ and measured STAT1 phosphorylation. STAT1 is activated following TYK2-mediated phosphorylation of the ligand-bound IFN-␣/␤R. We found that STAT1 phosphorylation is extended in this assay, providing biochemical evidence that SOCS1 plays an obligate role in IFN-␣/␤ signaling in addition to its established role in inhibiting IFN-␥ signaling. Previous overexpression studies in HeLa cells had suggested that SOCS1 was capable of inhibiting STAT1 phosphorylation from the IFN-␣/␤R (29). However, using the same assay system, these authors also showed SOCS3 could inhibit both IFN-␣/␤R and IFN-␥R signaling (29), a phenotype of SOCS overexpression not supported by subsequent studies in Socs3Ϫ/Ϫ cells and mice (1).
Finally, we obtained genetic evidence that TYK2 is involved in SOCS1 function by showing that Socs1Ϫ/Ϫ mice can be rescued by the absence of a single allele of Tyk2 and that loss of TYK2 can rescue the lethal phenotype of Socs1Ϫ/Ϫ;IFN-␥ϩ/Ϫ mice. At present it is not possible to exclude contributions from both IL-12 and IFN-␣/␤ in this system, since loss of STAT4 can also partially rescue Socs1Ϫ/Ϫ mice from early lethality (10). This study also does not address the possible long term effects of the absence of TYK2 on the ability of mice to make and respond to IFN-␥. It is possible that the extended life span of Socs1Ϫ/Ϫ;IFN-␥ϩ/Ϫ;Tyk2ϩ/Ϫ and Socs1Ϫ/Ϫ;IFN-␥ϩ/Ϫ; Tyk2Ϫ/Ϫ mice reflects effects on IFN-␥ homeostasis in the whole animal that enable the mice to escape the excessive inflammatory response that occurs in the absence of SOCS1, in addition to specific effects on IL-12 and IFN-␣/␤ signaling as well as yet undiscovered cytokine receptors that could utilize TYK2.
In conclusion, we find that SOCS1, an SH2-containing protein that binds phosphotryosines in its targets, does not regulate TLR signaling, a set of pathways regulated largely by serine/threonine phosphorylation. Our data suggest that SOCS1 regulates IFN-␣/␤ signaling through effects on TYK2 and that previous studies attributing effects of SOCS1 on TLR signaling are more likely the result of priming or concurrent stimulation with IFN-␣/␤ induced by TLR signaling. This pathway could be a significant target in the search for novel modifiers of the systemic inflammatory response and thereby be a useful target for sepsis therapeutics.