View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector Immunity Article

Suppression of Signaling by SOCS3: Characterization of the Mode of Inhibition and the Basis of Its Specificity

Jeffrey J. Babon,1,2,5,* Nadia J. Kershaw,1,3,5 James M. Murphy,1,2,5 Leila N. Varghese,1,2 Artem Laktyushin,1 Samuel N. Young,1 Isabelle S. Lucet,4 Raymond S. Norton,1,2,6 and Nicos A. Nicola1,2,* 1Walter and Eliza Hall Institute of Medical Research, 1G Royal Pde, Parkville, 3052, VIC, Australia 2The University of Melbourne, Royal Parade, Parkville, 3050, VIC, Australia 3Ludwig Institute for Cancer Research, Royal Pde, Parkville, 3050, VIC, Australia 4Monash University, Wellington Rd, Clayton, 3800, VIC, Australia 5These authors contributed equally to this work 6Present address: Monash Institute of Pharmaceutical Sciences, Monash University, Parkville 3052, Australia *Correspondence: [email protected] (J.J.B.), [email protected] (N.A.N.) DOI 10.1016/j.immuni.2011.12.015

SUMMARY Genetic deletion of each individual JAK leads to various immunological and hematopoietic defects; however, aberrant Janus kinases (JAKs) are key effectors in controlling activation of JAKs can be likewise pathological. Three myelopro- immune responses and maintaining hematopoiesis. liferative disorders (polycythemia vera, essential thrombocythe- SOCS3 (suppressor of cytokine signaling-3) is a mia, and primary myelofibrosis) are caused by a single point major regulator of JAK signaling and here we investi- mutation in JAK2 (JAK2V617F)(James et al., 2005; Levine et al., gate the molecular basis of its mechanism of action. 2005) that renders the kinase constitutively active and results We found that SOCS3 bound and directly inhibited in cytokine-independent activation of JAK-based signaling path- ways. An even more severe phenotype results from activation of the catalytic domains of JAK1, JAK2, and TYK2 JAK by oncogenic fusion, for example TEL-JAK2, which has but not JAK3 via an evolutionarily conserved motif been studied because of its role in childhood T and B cell acute unique to JAKs. Mutation of this motif led to the lymphoblastic leukemia (Lacronique et al., 2000). formation of an active kinase that could not be In order to prevent aberrant proliferation, JAK activity is regu- inhibited by SOCS3. Surprisingly, we found that lated in a number of ways. The primary negative regulators of the SOCS3 simultaneously bound JAK and the cytokine JAKs are a family of known as the suppressors of cyto- receptor to which it is attached, revealing how spec- kine signaling (SOCS) (Endo et al., 1997; Hilton et al., 1998; Naka ificity is generated in SOCS action and explaining et al., 1997; Starr et al., 1997) whose expression is induced by why SOCS3 inhibits only a subset of . JAK-STAT activation and they then inhibit the signaling cascade, Importantly, SOCS3 inhibited JAKs via a noncompet- creating a negative feedback loop. itive mechanism, making it a template for the devel- All eight SOCS proteins (SOCS1-7 and CIS) contain a central SH2 domain and a C-terminal SOCS box domain (Hilton et al., opment of specific and effective inhibitors to treat 1998), which interacts with elongins B and C and Cullin5 to cata- JAK-based immune and proliferative diseases. lyze the ubiquitination of bound signaling proteins (Babon et al., 2009; Kamizono et al., 2001; Zhang et al., 1999). Elegant studies INTRODUCTION performed by Yoshimura and colleagues (Sasaki et al., 1999; Yasukawa et al., 1999) showed that the two most potent Both hematopoiesis and the immune response are regulated suppressors of signaling, SOCS1 and SOCS3, contain a short by the action of cytokines through activation of the Janus motif, upstream of their SH2 domain, known as the KIR (kinase kinase-signal transducer and activator of transcription-sup- inhibitory region), which allows them to suppress signaling by pressor of cytokine signaling (JAK-STAT-SOCS) signal trans- direct inhibition of JAK catalytic activity. This is their dominant duction pathway (O’Shea and Murray, 2008). mode of action in vivo (Boyle et al., 2007; Zhang et al., 2001). There are four mammalian JAKs (JAK1-3 and TYK2), each Initial characterization of the KIR noted its amino acid sequence consisting of four domains (Figure S1 available online; Wilks similarity to the activation loop of JAKs (Sasaki et al., 1999; and Harpur, 1994). The N-terminal FERM domain binds constitu- Yasukawa et al., 1999). Like most tyrosine kinases, JAKs contain tively to the appropriate membrane-bound receptor whereas the an activation loop that blocks the catalytic cleft. Autophosphor- C-terminal kinase (catalytic) domain phosphorylates substrate ylation of this loop causes its translocation away from the cata- proteins. Between these are a noncanonical SH2 domain and lytic site and allows substrate access, thus activating the kinase. a pseudokinase domain, the most distinctive feature of the Consequently, it was proposed that the SH2 domain of SOCS1 JAK family. This domain has recently been shown to be catalyt- and SOCS3 binds the activation loop tyrosine phosphate and ically active (Ungureanu et al., 2011) and it regulates the activity the KIR acts as a pseudosubstrate to block the active site (Sasaki of the JH1 domain (Saharinen et al., 2000). et al., 1999; Yasukawa et al., 1999).

Immunity 36, 239–250, February 24, 2012 ª2012 Elsevier Inc. 239 Immunity Mechanism of Inhibition of by SOCS3

Despite the ability of SOCS proteins to bind to and inhibit to be a good substrate for JAK2JH1 phosphorylation. To verify JAKs, deletion of individual SOCS in mice has revealed that phosphorylation of SOCS3 was not by itself the cause of an exquisite specificity for particular cytokine-receptor combina- decreased gp130cyt phosphorylation, the entire reaction was tions rather than specific JAKs. For example, SOCS1 inhibits spotted onto nitrocellulose membranes, allowing total phos- interferon g signaling without affecting IL-6 signaling whereas phorylation of all components to be quantified. SOCS3 had the converse is true for SOCS3 (Alexander et al., 1999; Croker a clearly titratable inhibitory effect on JAK-catalyzed phosphor- et al., 2003), yet both systems utilize the ylation with an IC50 of 1 mM(Figure 1C). same JAKs (JAK1 and JAK2) (Murray, 2007). Moreover, the A limiting feature of these assays was that the concentration of binding affinities of the SOCS3 SH2 domain for phosphorylated SOCS3 required to inhibit JAK2JH1 was similar to the concentra- JAK peptides is several logs lower than that for certain cytokine tion of substrate. To ensure that it was not a SOCS3-substrate receptor phosphopeptides, and this binding is important in intact interaction that was responsible for inhibiting the phosphoryla- cells (Nicholson et al., 2000). tion reaction, we adopted a more robust enzyme-inhibition assay In this study we dissect both the mechanism and specificity of format where [Substrate] > > [Inhibitor] > > [Enzyme]. These JAK inhibition by SOCS3 by using biochemical, structural, and assays used high concentrations of a peptide substrate, resi- kinetic methods and resolve these apparent discrepancies. dues 693–708 of STAT5b (Saharinen et al., 2003; Zhao et al., We show that SOCS3 directly inhibits JAK1, JAK2, and TYK2 2010). SOCS3 inhibited phosphorylation of this peptide but not JAK3 because of the presence of a conserved three- substrate with the same IC50 of 1 mM(Figure 1D). These results residue motif on the former three JAK family members. By indicate that SOCS3 functions by blocking the ability of JAK2 to utilizing two distinct binding surfaces, SOCS3 is able to bind to phosphorylate substrates and is therefore a direct inhib- JAK and the cytokine receptor to which it is attached simulta- itor of its catalytic activity. neously, explaining why SOCS3 is specific for cytokines that signal through particular receptors. Intriguingly, inhibition occurs A SOCS1-SOCS3 Chimera Is a More Potent Inhibitor via a mechanism in which SOCS3 does not compete with either than SOCS3 ATP or substrate. This makes SOCS3 a noncompetitive tyrosine Replacing the KIR of SOCS3 with the corresponding region from kinase inhibitor and a new template for the future development SOCS1 resulted in a chimeric construct (termed SOCS1-3) that of a class of small-molecule JAK inhibitors with distinct advan- inhibited JAK2 kinase activity with higher affinity than did wild- tages over the current ATP analogs used to treat JAK-based type SOCS3 (IC50 values of 0.15 ± .02 and 1.2 ± 0.1 mM, respec- diseases. tively; see Figure 1D), despite there being a difference of only

four residues. A truncated construct of SOCS3, SOCS322-185, RESULTS that lacks the SOCS box and hence elonginB and C binding, was equally effective at inhibiting JAK2. These results show SOCS3 Inhibits Substrate Phosphorylation by the Kinase that the kinase inhibitory activity of SOCS3 relies solely upon Domain of JAK2 its KIR and extended SH2 domain and not its SOCS box domain. To examine SOCS3 inhibition of JAK kinase activity, we devel- In addition, the KIR from SOCS1, but not SOCS4 or SOCS5, can oped an in vitro kinase assay consisting of three purified, act as a functional replacement, making it likely that SOCS1 acts recombinant components: enzyme (JAK2 catalytic domain, in a similar manner to SOCS3.

JAK2JH1), substrate (gp130 cytoplasmic domain, gp130cyt), and inhibitor (SOCS3). Various chimeras of SOCS3 were also Both the KIR and SH2 Domain of SOCS3 Are Required for produced that had the key residues in the KIR, L22-S29, Kinase Inhibition but Phosphotyrosine Binding Is Not replaced by the corresponding residues of SOCS1, SOCS4, Previous work had highlighted the importance of F25A within the or SOCS5 (these are designated SOCS1-3, SOCS4-3, and KIR and R71 within the SH2 domain for SOCS3 function (Nichol- SOCS5-3; see Figure S1A). All SOCS proteins were expressed son et al., 1999; Sasaki et al., 1999). Here we confirm in vitro that and purified in complex with elongins B and C, their physiological deletion of the first eight residues in the KIR (residues 22–29) or SOCS box ligands, because they provide increased stability mutagenesis of F25 and R71 completely abrogated inhibition and solubility. JAK2JH1 was purified from insect cells. Both (Figure 1E). The KIR in isolation, as a synthetic peptide, could SOCS3 and SOCS1-3 inhibited substrate phosphorylation in not inhibit JAK2, even at concentrations 1003 the IC50 values this system (Figure 1A). Inhibition was not substrate specific as of the full-length protein (data not shown). The requirement for shown by the fact that the use of a different substrate (the acti- R71, which directly binds pTyr, implies that SOCS3 may bind vation loop of JAK2 fused to GST, termed GST-J; Figure 1B) the phosphorylated activation loop of JAK2 as part of its inhibi- led to identical results. In contrast, SOCS2, SOCS4, SOCS4-3, tory mechanism. However, the addition of a known high-affinity and SOCS5-3 had no effect on phosphorylation of any substrate (KD = 100 nM) for the SOCS3 SH2 domain, murine tested. gp130750-764 (STASTVEpYSTVVHSG) (Nicholson et al., 2000), As seen in Figure 1, all SOCS constructs were themselves at a 5-fold molar excess had no effect on JAK inhibition by phosphorylated to some extent in these assays, as was elon- SOCS3. In addition we were able to form a ternary complex of ginC. This was not unexpected because the isolated kinase JAK2JH1:SOCS3:gp130750-764 containing all three components domain of JAK2, like that of many tyrosine kinases, shows little at a stoichiometric ratio as analyzed by gel filtration and rpHPLC substrate specificity and will phosphorylate any tyrosines that (Figure S1). Therefore, although R71 may contact JAK2 when are solvent exposed and not part of ordered secondary struc- bound, these results indicate that phosphopeptide binding by ture. For example, we found a synthetic polymer, poly-Glu4Tyr, SOCS3 is undisturbed in the presence of JAK2.

240 Immunity 36, 239–250, February 24, 2012 ª2012 Elsevier Inc. Immunity Mechanism of Inhibition of Janus Kinase by SOCS3

Figure 1. SOCS3 and SOCS1-3 inhibit substrate phosphorylation by JAK2

(A) Autoradiography (top) and Coomassie-stained (bottom) SDS-PAGE analysis of JAK2JH1 catalyzed phosphorylation of the gp130 cytoplasmic domain in the presence of serial 10-fold dilutions of SOCS-elonginBC complexes. Both SOCS3 and a SOCS1-3 chimera were effective inhibitors of substrate phosphorylation. Gradient bars represent decreasing concentration of each SOCS construct (10 mM, 1 mM, 100 nM, 10 nM, left to right). (B) As in (A) except a different substrate (GST-J) was used. SOCS concentration was 5 mM, 1 mM, 200 nM, left to right in each case. (C) Quantitative kinase inhibition experiments. As in (A) except reactions were spotted onto nitrocellulose filters. The experiment was performed in quadruplicate, raw data shown above and quantitated below. The control reaction (0 SOCS) was normalized to 100%. (D) SOCS1-3 is a more effective inhibitor of JAK2 than SOCS3. As in (C) except a STAT5b peptide was used as substrate and the reaction spotted onto P81 phosphocellulose paper. Reactions contained 30, 12, 5, 2, 0.8, 0.3, 0.1, 0.04, 0.02, 0.008, 0 mM SOCS (left to right). (E) SOCS3 constructs that contained mutations in the KIR (F25A), SH2 domain (R71K), or that lacked the first eight residues of the KIR (D22-29) did not inhibit JAK2. As in (D). Reactions contained 10, 5, 2.5, 1.2, 0.6, 0.3, 0.15, 0 mM SOCS (left to right). The addition of gp130 phosphopeptide (pY) did not affect inhibition. Error bars represent ± range of two experiments. Data are normalized to no-inhibitor controls.

SOCS3 Inhibits JAK1, JAK2, and TYK2 but Not JAK3 in specific activity. The only mutations to affect SOCS3-medi- because of the Presence of a Three-Residue Motif ated inhibition were in a three-residue motif 1071-1073GQM (Fig- in the JAK Insertion Loop ure 2B). Mutating this sequence completely abolished the ability We cloned, expressed, and purified the kinase domains of all of SOCS3 to inhibit JAK2. Depending on the method of four JAKs and examined the ability of SOCS3 to inhibit them. sequence/structure alignment, these three residues correspond 1043-1045 1044-1046 SOCS3 inhibited JAK1, JAK2, and TYK2 with IC50 values of to either DVP or VPA in JAK3. Both GQM-DVP 2 mM, 1.5 mM, and 7 mM, respectively, but had no inhibitory effect and GQM-VPA mutants of JAK2 were insensitive to SOCS3. on JAK3 (Figure 2A). The GQM motif forms the final three residues of the JAK inser- A comparison of the sequence of all four kinase domains high- tion loop (residues 1052–1073; Figure 2D), an insertion that is lighted a number of residues that were conserved within JAK1, specific to JAKs (Haan et al., 2009; Lucet et al., 2006). More JAK2, and TYK2 but not JAK3. Therefore, a series of mutant detailed mutagenesis showed that the first and third of these JAK2 molecules were produced that replaced wild-type resi- residues, G1071 and M1073, were absolutely required for due(s) with the corresponding residues from JAK3. All mutant SOCS3 inhibition, whereas mutation of the central glutamine, kinases were catalytically active with no significant difference Q1072, had only a minor effect (Figure 2C). Mutation of I1074

Immunity 36, 239–250, February 24, 2012 ª2012 Elsevier Inc. 241 Immunity Mechanism of Inhibition of Janus Kinase by SOCS3

Figure 2. SOCS3 Inhibits JAK1, JAK2, and TYK2 but Not JAK3 because of a Three-Residue Motif in the JAK Insertion Loop

(A) Kinase inhibition assays were performed with the kinase domain of all four JAKs. Only JAK3JH1 was not inhibited by SOCS3. Error bars represent ± range from two experiments. Data were normalized to no-inhibitor controls. (B) Mutating the GQM motif makes JAK2 nonresponsive to SOCS3. All other mutations have no effect. (C) G1071 and M1073 are absolutely required for SOCS3 inhibition. As in (A) except point mutants of JAK2 were tested. (D) Sequence alignment of the GQM region of all JAKs and the KIR of SOCS3 and SOCS1. Highly conserved residues are shown boxed in gray, the GQM motif in yellow, conserved residues in the KIR in black. (E) The structure of JAK2 (PDB ID 2B7A). The GQM motif is solvent exposed and shown in yellow. The locations of mutated residues from (B) are indicated by gray arrows. *Zebrafish JAK2b is grouped with TYK2 in this figure.

also had a small effect on the IC50 of SOCS3. The GQM motif is this position. Equally, none of these organisms contained this solvent exposed (Figure 2E) as expected if it forms a direct motif in JAK3 and the corresponding sequence within this region contact with SOCS3. Although the GQM sequence is unlikely was not conserved. A sequence comparison of SOCS mirrors to represent the entire binding surface on JAK, these data indi- this phenomenon. Only vertebrates have SOCS1 and SOCS3 cate that it is an essential motif that allows JAK to be inhibited homologs, and these all have highly similar kinase inhibitory by SOCS3. regions. In contrast, insects contain only SOCS4-7 homologs. In summary, this analysis shows that all organisms that contain The GQM Motif of JAK1, JAK2, and TYK2 Is Conserved in an expanded JAK system also encode a SOCS protein with All Organisms that Contain a SOCS1 or SOCS3 Homolog a functional KIR. All organisms from insects to mammals contain at least one JAK with vertebrates typically containing four JAKs. A sequence Mutating the GQM Motif in JAK1 Leads to Prolonged IL-6 alignment of JAKs from within these organisms (Figure 2D) Signaling in Live Cells showed that the GQM motif is conserved in JAK1, JAK2, and To examine our hypotheses regarding the specificity of SOCS3 TYK2 from all vertebrate species listed with the exception of action in a physiological setting, we wished to mutate full-length zebrafish JAK2b, which contains a highly similar motif (GQT) at JAK to a form that is impervious to inhibition by SOCS3 and to

242 Immunity 36, 239–250, February 24, 2012 ª2012 Elsevier Inc. Immunity Mechanism of Inhibition of Janus Kinase by SOCS3

Figure 3. IL-6-Induced Phosphorylation of STAT3 Is Prolonged in the Presence of JAK1GQM-DVP (A) Kinase inhibition assays were performed with the kinase domain of JAK1 and JAK1GQM-DVP. JAK1GQM-DVP was not inhibited by SOCS3. Error bars represent ± range from two experiments. Data were normalized to no-inhibitor controls. (B) Wild-type and mutant (GQM-DVP) JAK1 constructs were transfected into JAK1À/À (U4A) cells. Cells were stimulated with IL-6 plus s-IL-6Ra for 15 min and then harvested at the indicated times. Lysates were analyzed by immunoblotting with antibodies specific for phosphorylated STAT3, total STAT3, SOCS3, and JAK1. examine the effect this has on IL-6 signaling. Because JAK2 is which is not susceptible to SOCS inhibition (Figure S2A). Like- dispensible for IL-6 signaling but JAK1 is not (Murray, 2007), wise, SOCS329-185, which lacks the KIR, showed no interaction we cloned and expressed JAK1GQM-DVP which we predicted, with wild-type JAK2 (Figure S2B). The SOCS3-JAK2 complex based on our JAK2 experiments, would be resistant to SOCS was in slow exchange and hence could not be assigned, but inhibition. As shown in Figure 3A, the kinase domain of JAK1 is the surface of SOCS3 that interacts with JAK2 could be mapped active in the presence of the GQM-DVP mutation but it cannot by identifying resonances from the SOCS3 spectra that shifted be inhibited by SOCS3. We then cloned full-length JAK1WT and when in complex with JAK2. In order to avoid false positives, JAK1GQM-DVP and transfected these constructs into JAK1À/À only nonoverlapped peaks that shifted by more than one peak human fibrosarcoma (U4A) cells (Guschin et al., 1995). These width (0.1 and 0.6 ppm in the 1H and 15N dimensions, respec- cells express the gp130 shared coreceptor and hence can be tively) were considered in this analysis. A number of residues, stimulated with a mixture of IL-6 and soluble IL-6Ra (Guschin including K22-S29 (the first eight residues of the KIR), had to et al., 1995). As shown in Figure 3B, JAK1GQM-DVP was able to be excluded from analysis on this basis (they are unstructured activate STAT3 after IL-6 stimulation, but this activation was pro- in the absence of JAK2 and hence their resonances reside in longed compared to wild-type JAK. pSTAT3 was still detectable a heavily overlapped region of the spectrum). Nevertheless, 21 4 hr poststimulation in the presence of the DGQM mutants backbone and 2 side chain amides were identified that shifted compared to only 2 hr in the presence of WT JAK. These results in the presence of JAK2. Several of these shifts were large, for À/À are identical to those seen in Socs3 cells, which also show example S74 had a chemical shift perturbation of >dAV = 0.67 2 2 1/2 a 2-fold increase in the persistence of pSTAT3 upon IL-6 stimu- (dAV =(Dd1H +(Dd15N/5) /2) ). Repeating this analysis on the lation (Croker et al., 2003, 2004), and indicate that SOCS3 inhibi- methyl region of 1H-13C-HMQC spectra identified a further ten tion has been completely disrupted in these cells. Collectively, residues whose methyl groups shifted in the presence of JAK2. these data demonstrate that the GQM motif is essential for The combination of these data mapped a 30 residue binding SOCS3 inhibition of JAK, both in vitro and in live cells. surface on SOCS3 (Table S2). This surface is centered on the extended SH2 subdomain (ESS) helix (Figure 4) and includes NMR Analysis Reveals that SOCS3 Can Interact its junction with the SH2 domain proper, the N-terminal portion with JAK2 and Cytokine Receptor Simultaneously, of helix aA, the BC loop, and the DE loop. via Two Adjacent Binding Surfaces Of the ten methyl-containing residues identified, six are within By using NMR, we mapped the surface of SOCS3 that binds to the ESS helix and five of these have solvent-exposed side chains JAK2 by chemical shift perturbation. Both 1H-15N-HMQC and in the unbound state, making it likely that they represent part of 1H-13C-HMQC spectra were recorded with 250 mM labeled the true binding surface. The other residue, L41, forms the junc-

SOCS322-185 ± 500 mM unlabeled JAK2JH1. The gp130 peptide tion with the SH2 domain and appears to anchor the ESS helix to was present in all experiments because it leads to increased the core of the SH2 domain by a number of hydrophobic interac- solubility of SOCS3 but does not interfere with JAK inhibition tions. This residue contains the most upfield shifted resonance in

(Figure 1). the SOCS3 spectra (d-CH3 À0.4 ppm) resulting from ring current The 15N-HMQC spectrum of SOCS3 was well dispersed with effects from Y47, F80, and F102. This shifted even further upfield narrow line-widths, but the addition of a 2-fold molar excess of in the presence of JAK2, suggesting that a subtle conformation unlabelled JAK2 led to intense line broadening and widespread change in this region moves the Leu side chain closer to one of chemical shift perturbation (Figure 4B), consistent with the these three aromatic groups. formation of a 52 kDa SOCS3-JAK2 complex. No line broad- The mapped interaction surface is adjacent to one end of the ening or peak shifts were seen upon addition of JAK2GQM-DVP, pTyr binding groove. Residues that shift upon JAK binding

Immunity 36, 239–250, February 24, 2012 ª2012 Elsevier Inc. 243 Immunity Mechanism of Inhibition of Janus Kinase by SOCS3

Figure 4. NMR Analysis Identifies the Surface of SOCS3 that Interacts with JAK2

(A) Ribbon diagram of the structure of SOCS322-185 (PDB ID 2HMH) with important secondary structural motifs indicated. The dashed line indicates the first seven residues of the KIR, which are unstructured in the absence of JAK. 1 15 1 13 (B) H- N and H- C HMQC analysis of the SOCS3-JAK2JH1 interaction are shown in upper and lower panels, respectively. The SOCS spectra are in red and the

SOCS3-JAK2JH1 spectrum in black.

(C) Surface diagram of SOCS3. Residues whose resonances shift in the presence of JAK2JH1 highlighted in red. The orientation of SOCS3 on the left is identical to that in (A).

include several (T52-G54) around the pY-2 pocket as well as gp130 phosphopeptide remains bound in the presence of several in the BC loop (R71-D75), including R71, which forms JAK2 (Figure S1). a salt bridge with pTyr. However, residues that display charac- Taken together, these results indicate that the JAK2 binding teristic chemical shift perturbations when the gp130 peptide is surface on SOCS3 borders but does not overlap the phosphotyr- bound (T89, D107, R94, L178) (Babon et al., 2006) maintain these osine binding groove. This surface can be defined as consisting characteristic chemical shift positions in the presence of JAK2. of the KIR, ESS helix, and the edge of the pTyr binding groove. By In addition, it is clear from chromatographic analysis that the binding JAK and specific cytokine receptors (which are already

244 Immunity 36, 239–250, February 24, 2012 ª2012 Elsevier Inc. Immunity Mechanism of Inhibition of Janus Kinase by SOCS3

associated with JAK) simultaneously, SOCS3 becomes part of phosphate transfer. Many kinases have the ability to catalyze the a high-affinity ternary complex. A model in which this ternary transfer of a phosphate moiety to a water molecule, rather than complex underpins the specificity of SOCS3 (Figure S2) will be to tyrosine, thereby acting as an ATPase. We reasoned that if discussed. the mode of action of SOCS3 is to inhibit phosphate transfer, then it should also inhibit phosphate transfer to water and hence SOCS3 Is a Noncompetitive Inhibitor of JAK2 the ability of JAK2 to act as an ATPase. Therefore, we measured

The model for the mechanism of JAK inhibition by SOCS3 has the ATPase activity of JAK2JH1 in the presence and absence of been that the KIR acts as a pseudosubstrate and thereby blocks SOCS3. access to the active site (Yasukawa et al., 1999). Kinases have As shown in Figure 6, we unexpectedly observed a small, but two substrates: ATP (the phosphate donor) and a tyrosine-con- reproducible, activation of JAK2 ATPase activity in the presence taining substrate (the phosphate acceptor). If SOCS3 acts as of SOCS3. SOCS3 and SOCS1-3 stimulated the ATPase activity a pseudosubstrate, then this implies that it will compete with of JAK2 by nearly 2-fold. SOCS3F25A had no effect. This activity the binding of one or both of these substrates. This can be titrated with an apparent EC50 of 2 mM. These results indicate addressed by performing steady-state enzyme kinetics in the that SOCS3 specifically inhibits the ability of JAK to transfer presence of SOCS3 (Cornish-Bowden, 2004). phosphate to tyrosine but does not inhibit its ability to hydrolyze Kinetic experiments were performed at 25C, with an enzy- ATP and transfer phosphate to water. Our favored molecular me:substrate ratio > 1:1000 (10 nM enzyme:0.01–5 mM STAT model of inhibition, incorporating this information, will be peptide). Under these conditions, product formation was linear discussed. with time for >45 min, although two time points (7.5 and Because the rate-limiting step of a number of kinases is 15 min) were taken in all experiments to ensure that this was product (ADP) release (Adams, 2001), we wished to rule out the case. Results were quantified with scintillation counting the possibility that SOCS3 may act by stabilizing a JAK-ADP and phosphorimaging. When the ATP concentration was varied (Enzyme-Product) complex. Such a mechanism implies that (0–1 mM), the STAT substrate concentration was fixed at JAK would be insensitive to the presence of SOCS3 during the 1.6 mM. Conversely, when the STAT peptide concentration was first round of catalysis, when ADP is absent. However, single varied (0–2.5 mM), the ATP concentration was fixed at 2 mM. turnover experiments showed that SOCS3 was still a potent ATP peptide JAK2JH1 had KM = 140 mM and KM = 0.6 mM under inhibitor of JAK under these conditions (Figure S4A). In addition, these conditions. Initial reaction velocity was plotted against we did not observe any synergistic effect when a combination of substrate concentration at various concentrations of inhibitor SOCS3 and ADP were used in standard kinase inhibition exper- (Figure 5). iments (Figure S4B). Surprisingly, these analyses showed that SOCS3 is a Collectively, these results show that ATP is still hydrolyzed by noncompetitive inhibitor of JAK2JH1, with respect to both ATP JAK in the presence of SOCS3 and thereby confirm that SOCS3 and substrate. This was apparent by linear least-squares fitting does not compete with ATP for binding. Therefore, inhibition of of the data to a mixed inhibition model with Sigmaplot as well JAK by SOCS3 will not be affected by a high intracellular ATP as by Lineweaver-Burk reciprocal analyses (Figure 5, lower concentration. panels). Lines that intersect on the abscissa indicate noncom- petitive inhibition. Dixon plot analyses (Dixon, 1953) of these DISCUSSION data are shown in Figure S3A. These analyses were performed on three separate occasions, each time in duplicate with The prevailing model of SOCS3 action has been that it is different preparations of both enzyme and inhibitor. Fitting of recruited to specific cytokine receptors by its SH2 domain and the data yielded Ki = 1.5 ± 0.7 mM and 1.2 ± 0.3 mM versus once there would eventually engage JAK with both its SH2 substrate and ATP, respectively. These results can be con- domain and KIR. The SH2 domain would bind the phosphory- trasted both qualitatively and quantitatively to identical experi- lated activation loop of JAK and the KIR would then block ATP ments performed with ADP as inhibitor (Figure S3B), which binding (Sasaki et al., 1999). We now show that SOCS3 interacts gives rise to the expected ATP-competitive inhibition curves. with both JAK and the gp130 receptor simultaneously by utilizing

One attribute of noncompetitive inhibition is that the IC50 is not two adjacent binding surfaces and that ATP binding by JAK is affected by substrate concentration. As shown in Figure 5, unaffected. Such a mode of action explains the specificity of

SOCS3 inhibited JAK with identical IC50 values at ATP and SOCS3 and has important consequences both biologically and substrate concentrations that varied by 40-fold. In contrast, therapeutically on a number of fronts as now discussed. the IC50 of the ATP-competitive inhibitors ADP and CMP-6 First, the ability of SOCS3 to bind to JAK and simultaneously to increased in the presence of high ATP concentration, but not in the receptor to which it is attached leads to an unusual ternary the presence of high substrate concentration, as expected. complex in which each moiety is directly bound to the other Collectively, these results show that SOCS3 is a noncompetitive two. For such a ternary complex to dissociate, at least two direct inhibitor of JAK2 and therefore imply that it does not act by interactions must be broken, and consequently the overall blocking the active site of the kinase. affinity of such a complex is higher than any of the individual associations. It follows, therefore, that cytokines that utilize Mechanism of SOCS3-Mediated Suppression receptors with SOCS3 binding sites (e.g., IL-6) will be effectively of JAK/STAT Signaling inhibited by SOCS3 (Wormald et al., 2006), whereas cytokines In considering the molecular mechanism of SOCS inhibition of that signal through receptors that lack such a site (e.g., IFN-g) JAK, we thought it most likely that SOCS3 was directly inhibiting will not, even though they may signal through the same JAKs.

Immunity 36, 239–250, February 24, 2012 ª2012 Elsevier Inc. 245 Immunity Mechanism of Inhibition of Janus Kinase by SOCS3

Figure 5. SOCS3 Is a Noncompetitive Inhibitor of JAK2

(A) Michaelis-Menten (top) and Lineweaver-Burk (bottom) analysis of SOCS3 inhibition of JAK2JH1. ATP (left) and peptide substrate (right) titrations show that SOCS3 alters Vmax but not Km, indicating noncompetitive inhibition. Inhibition experiments were performed with 10 nM JAK and 0, 0.5, 1, 2, 4, 8 mM SOCS3

246 Immunity 36, 239–250, February 24, 2012 ª2012 Elsevier Inc. Immunity Mechanism of Inhibition of Janus Kinase by SOCS3

gp130Y757F compared to wild-type gp130 (Jenkins et al., 2005; Nicholson et al., 2000). Likewise, it has been shown that IFN-g signaling is only poorly suppressed by SOCS1 when its binding site on the IFN-g receptor (pTyr441) has been mutated (Starr et al., 2009), and we suggest that SOCS1 and SOCS3 both share this mechanism of receptor dependence to gain high specificity and efficacy toward particular cytokines. Even in the absence of receptor, SOCS3 is highly specific toward JAKs, rather than other tyrosine kinases. This is high- lighted by the fact that it displays selectivity even within the JAK family. Selectivity arises from the fact that SOCS3 interacts with a motif within the JAK insertion loop. By interacting with this region, it is able to specifically target JAKs (over other kinases) and, by targeting the GQM motif, it is able to distinguish JAK1, JAK2, and TYK2 from JAK3. An evolutionary comparison of JAK and SOCS sequences is telling in this regard. Only verte- brates have evolved an expanded JAK system (four JAKs) and it seems they have also evolved the ability to directly and specif- ically inhibit three of them. Although there may be another protein that functions to inhibit JAK3, it is unlikely to do so via the same mechanism because JAK3 shows no evolutionary conservation in the GQM-equivalent region. No other human kinases contain GQM at this position and therefore SOCS3 would not be ex- pected to directly inhibit any other kinases in the genome. This is exemplified by the fact that, in intact cells, JAK1 becomes unresponsive to SOCS3 if the GQM motif is mutated, even though it remains tethered next to the kinase on the gp130 receptor. This indicates that JAK3, which lacks GQM, will not be inhibited by SOCS3 even if they were to be bound to the same receptor complex. The SOCS3 binding site on the gp130 receptor, pY757, is also the binding site of the phosphatase SHP-2, which is involved in stimulating the Ras/ERK and PI-3K/Akt signaling pathways (Neel et al., 2003). Mutating this site in mice leads to a Th1 Figure 6. SOCS3 Has an Activating Effect on JAK2JH1 ATPase Activity cell-biased immune response, autoimmune arthritis, and the (A) JAK-catalysed ATPase assays were performed in the presence of various development of gastric adenoma because of the combined concentrations of SOCS3/elonginBC constructs, analyzed by thin-layer effect of dysregulating both the JAK-STAT and SHP2-ERK path- chromotagraphy, and quantified. SOCS3 and SOCS1-3 enhanced the ATPase ways (Ernst and Jenkins, 2004). In this context, our JAKGQM-DVP activity of JAK2 but the KIR mutant (F25A) did not. mutant provides a useful resource because, unlike the (B) Raw data used to generate the EC50 curves shown in (A). JAK2JH1 Y757F hydrolyzed ATP at 0.05 sÀ1 under these conditions compared to 0.09 sÀ1 in the gp130 system, only JAK/STAT signaling (and not SHP2- presence of 15 mM SOCS3. Error bars represent ± range, two experiments. ERK signaling) is dysregulated. Data are normalized to no-inhibitor controls. Therapeutically, our data have important consequences because, to our knowledge, SOCS3 is the only biological kinase inhibitor that acts noncompetitively as regards both ATP and Importantly, we show that although SOCS3 can inhibit JAK1, substrate. Other inhibitors act by competitive mechanisms, JAK2, and TYK2 in the absence of receptor, it does so with rela- either by blocking the active site directly, for example p27KIP1 tively poor (micromolar) affinity. Therefore, when artificially over- (Russo et al., 1996) and RKIP (Yeung et al., 1999), or by disrupt- expressed, SOCS3 can be predicted to inhibit signaling via any ing it allosterically, for example JIP (Barr et al., 2004; Heo et al., cytokine that utilizes JAK1, JAK2, or TYK2, but when present 2004). By virtue of its noncompetitive mechanism, the inhibitory at physiological levels, SOCS3 will inhibit only cytokines that function of SOCS3 is unaffected by high intracellular ATP signal through particular receptors. This has been seen in concentration. Structure-guided drug design has historically numerous studies and explains why SOCS3 shows detectable, targeted the ATP binding site as the most amenable for inhibitor but greatly impaired, inhibition of JAK-STAT signaling through development. All current JAK inhibitors (Ghoreschi et al., 2009)

(top to bottom curves, respectively, in upper panels) with varying concentrations of ATP and substrate. Lineweaver-Burk curves that intersect on the abscissa indicate noncompetitive inhibition. Error bars represent ± range from two experiments. (B) JAK2 inhibition experiments were performed in the presence of SOCS3 (top, two independent experiments) or in the presence of ATP competitive inhibitors

ADP and CMP-6 (bottom). The IC50 values of SOCS3 were unaffected by ATP concentration unlike the IC50 values of ADP and CMP-6. Error bars represent ± range from two experiments. Data are normalized to no-inhibitor controls.

Immunity 36, 239–250, February 24, 2012 ª2012 Elsevier Inc. 247 Immunity Mechanism of Inhibition of Janus Kinase by SOCS3

are ATP analogs or competitors and bind to the active site of the yield. Because we were unable to completely displace this inhibitor after puri- enzyme, which has two major drawbacks: (1) ATP in the cell, fication, all other assays used JAK expressed in the absence of this inhibitor. which can be as high as 10 mM, competes with inhibitor binding Cloning and Overexpression of JAK1 in JAK1–/– Cell Lines and leads to reduced efficacy in vivo and (2) the ATP binding site The JAK1-deficient human fibrosarcoma cell line U4A has been described of tyrosine kinases are all structurally similar and thus specificity previously (McKendry et al., 1991; Guschin et al., 1995) and was maintained is difficult to achieve. In contrast, a noncompetitive JAK inhibitor in Dulbecco’s modified Eagle’s medium, supplemented with 10% (v/v) heat- would retain its potency in vivo and may take advantage of the inactivated fetal calf serum and hygromycin (250 mg/ml). The mutation of greater structural variation in regions outside the ATP binding murine JAK1 resulting in the amino acid mutations GQM > DVP for expression site to gain greater specificity for JAK over the rest of the kinome in U4A cells was produced via oligonucleotide-directed PCR mutagenesis. (Cozza et al., 2009). As a specific, noncompetitive JAK inhibitor Wild-type and mutant JAK1 cDNAs were cloned into the puromycin-resistant plasmid pEF-IRES-P (Hobbs et al., 1998) and transfected into the U4A cell line that does not bind to the active site of the enzyme, SOCS3 is with FuGENE HD Transfection Reagent (Roche), according to the manufac- the ideal template for the development of a new class of thera- turer’s instructions. Stable cell lines overexpressing either wild-type JAK1 peutic JAK inhibitors. or the JAK1GQM-DVP mutant were selected with puromycin (2 mg/ml) and Having targeted JAK via both the receptor to which it is tested for JAK1 expression by protein immunoblot with an antibody to JAK1 attached and its GQM motif, what then is the molecular mecha- (Santa Cruz). nism of SOCS3? The noncompetitive nature of inhibition by Cytokine Stimulation and Protein Immunoblotting SOCS3, and the fact that it does not block phosphate transfer U4A cells and their derivatives (2 3 106) were plated overnight in 6-well plates to water, implies that it does not block or destroy the structure and pulsed with 400 ng/ml human recombinant IL-6 and 500 ng/ml sIL-6Ra of the kinase active site. We propose a model in which SOCS3 (R&D Systems, MN) for 15 min. Cells were washed in PBS and lysed for binding alters the conformation of JAK in such a way that the 30 min in 50 ml ice-cold KALB lysis buffer (150 mM NaCl, 50 mM Tris [pH 7.5], distance between the ATP terminal phosphate and the acceptor 1% [v/v] Triton X-100, 1 mM EDTA, 1 mM Na3VO4, 1 mM NaF, 1 mM PMSF) con- tyrosine hydroxyl group, or their relative geometry, is affected. taining protease inhibitors (Complete cocktail tablets; Roche). Lysates were cleared by centrifugation (13,000 3 g) for 10 min at 4C and supernatants The GQM motif is located within 8 A˚ of the substrate binding boiled in 43 reducing sample buffer. A 15 ml sample was separated by SDS- site of JAK2 and therefore SOCS3 binding may distort its loca- PAGE, transferred onto polyvinylidene difluoride membranes (Millipore), and tion. Even a small shift in their relative positions could dramati- examined for phosphorylated STAT3 (Cell Signaling), total STAT3 (Santa cally affect phosphate transfer from ATP to the tyrosine hydroxyl, Cruz), and JAK1 expression by protein immunoblot. because these moieties need to be positioned within 3 A˚ to allow nucleophilic attack within the developing transition state JAK2 Kinase Inhibition Assays by Protein Substrates  (Madhusudan et al., 1994). Only the structure of a SOCS-JAK 1 mg/ml protein substrate was incubated with 50 nM JAK2JH1 at 25 C for 30 min in 20 mM Tris (pH 8.0), 100 mM NaCl, 1 mM DTT, 2 mM ATP, and complex will enable this hypothesis to be examined. 4 mM MgCl2 (kinase buffer) and various concentrations of SOCS-elonginBC By regulating cytokine signaling, SOCS3 plays a key role in complexes. 1 mCi g-[32P]ATP was included to allow visualization of phosphor- maintaining the hematopoietic system and controlling the ylation via autoradiography and phosphorimaging. After incubation, the reac- immune response. Our results reveal the basis for both the effi- tions were either boiled and subjected to analysis by SDS-PAGE or terminated cacy and specificity of SOCS3 action and explain how it is able with 50 mM EDTA and spotted onto a nitrocellulose membrane. Membranes 3 to regulate signaling via a particular subset of cytokines. Finally, were washed extensively (4 200 ml, 15 min) with PBS and subsequently exposed to a phosphorimager plate (Fuji). our experiments show that unlike all currently available JAK inhibitors, SOCS3 inhibits JAK via a mechanism in which it is JAK2 Kinase Inhibition Assays via Peptide Substrates not affected by high intracellular ATP levels, thereby suggesting  0–2 mM substrate peptide was incubated with 10 nM JAK2JH1 at 25 C for that it is the ideal template upon which to base the development 10–20 min in kinase buffer and 1 mCi g-[32P]ATP. After incubation, the reactions of a new class of therapeutic JAK inhibitors. were spotted onto P81 phosphocellulose paper and quenched in 5% H3PO4. The paper was washed extensively (4 3 200 ml, 15 min) with 5% H3PO4 and exposed to a phosphorimager plate (Fuji). EXPERIMENTAL PROCEDURES Steady-State Kinetics Cloning and Expression Michaelis/Menten analysis requires the use of a high enzyme-to-substrate All SOCS3 constructs lack the first 21 amino acids and have the PEST motif ratio so that product formation is linearly proportional to time and product inhi- R (residues 129–163) replaced by a Gly-Serx4 linker, and these modifications bition is negligible. Substrate concentration must be KM to approach satura- enhance its stability and solubility (Babon et al., 2005, 2006). This parent tion and allow accurate determination of Vmax. Therefore, 2 nM JAK2JH1 was construct, SOCS322-225DPEST, was used as a template for all further muta- used to phosphorylate 0–5 mM STAT5b peptide in these assays. Inhibitor, genesis and is henceforth referred to as SOCS3. Coexpression and purifica- SOCS3-elonginBC, was included at 0–10 mM final concentration. Reactions tion of SOCS3 with elongins B and C was as previously described (Babon were performed in kinase buffer except that both ATP and STAT5b peptide et al., 2008) Plasmids encoding SOCS2 and SOCS4 were kind gifts of A. were titrated independently. 0.1 mg/ml BSA and 1 mCi g-[32P]ATP were added Bullock (SGC, Oxford, UK). The sequence of all constructs is given in Supple- at 25C. 7.5 and 15 min time points were used to ensure that product formation mental Information. was linear with time. When ATP was titrated, STAT5b was at 1.6 mM (KM); JAK1, JAK2, JAK3, and TYK2 (kinase domains) were cloned into pFASTBAC when STAT5b was titrated, ATP was at 2 mM (15 3 Km). After incubation, (Invitrogen) and expressed as 6xHIS-tagged proteins. JAK2 mutants were the reactions were spotted onto P81 phosphocellulose paper (Millipore) and produced with oligonucleotide-directed PCR mutagenesis. JAK2 was also treated as described in the previous paragraph. Control experiments showed expressed as a GST fusion by cloning into pDEST-20 (Invitrogen). All JAK that retention onto P81 paper was linear to >5 mM peptide. constructs were expressed and purified as previously described (Lucet et al.,

2006) except that JAK2JH1 used for NMR analysis was expressed in the pres- ATPase Assays ence of 0.4 mM of the JAK inhibitor 2-(1,1-Dimethylethyl)-9-fluoro-3,6-dihy- ATPase assays were performed by incubating 0.25 mM JAK2JH1 in kinase dro-7H-benz[h]-imidaz[4,5-f]isoquinolin-7-one (EMD biosciences) to increase buffer (except that ATP concentration was 0.1 mM) and 1 mCi g-[32P]ATP for

248 Immunity 36, 239–250, February 24, 2012 ª2012 Elsevier Inc. Immunity Mechanism of Inhibition of Janus Kinase by SOCS3

30 min at 25C. SOCS3-elonginBC was present in the reaction at 0–10 mM SOCS box of suppressor of cytokine signaling-3 contributes to the control of concentration. Reactions were stopped with 25 mM EDTA, spotted onto G-CSF responsiveness in vivo. Blood 110, 1466–1474. PEI-cellulose TLC plates, and developed in 1 M LiCl, 1 M formic acid for Cornish-Bowden, A. (2004). Fundamentals of Enzyme Kinetics (London: 45 min, air-dried, and exposed to a phosphorimager plate. Portland Press). Cozza, G., Bortolato, A., Menta, E., Cavalletti, E., Spinelli, S., and Moro, S. NMR  (2009). ATP non-competitive Ser/Thr kinase inhibitors as potential anticancer SOFAST-HMQC experiments were recorded at 37 C in 20 mM MES, 1 mM agents. Anticancer. Agents Med. Chem. 9, 778–786. DTT on a Bruker Avance 600 MHz spectrometer equipped with cryoprobe. Croker, B.A., Krebs, D.L., Zhang, J.G., Wormald, S., Willson, T.A., Stanley, E.G., Robb, L., Greenhalgh, C.J., Fo¨ rster, I., Clausen, B.E., et al. (2003). SUPPLEMENTAL INFORMATION SOCS3 negatively regulates IL-6 signaling in vivo. Nat. Immunol. 4, 540–545. Croker, B.A., Metcalf, D., Robb, L., Wei, W., Mifsud, S., DiRago, L., Cluse, L.A., Supplemental Information includes Supplemental Experimental Procedures, Sutherland, K.D., Hartley, L., Williams, E., et al. (2004). SOCS3 is a critical four figures, and one table and can be found with this article online at physiological negative regulator of G-CSF signaling and emergency granulo- doi:10.1016/j.immuni.2011.12.015. poiesis. Immunity 20, 153–165.

ACKNOWLEDGMENTS Dixon, M. (1953). The determination of enzyme inhibitor constants. Biochem. J. 55, 170–171. We thank S. Yao for assistance with NMR, A. Bullock for plasmids, J.-G. Zhang Endo, T.A., Masuhara, M., Yokouchi, M., Suzuki, R., Sakamoto, H., Mitsui, K., for recombinant IL-6, and the ACRF Biomolecular Resource Facility for Matsumoto, A., Tanimura, S., Ohtsubo, M., Misawa, H., et al. (1997). A new sequencing. We are grateful to P. Hertzog for the gift of U4A cells. This work protein containing an SH2 domain that inhibits JAK kinases. Nature 387, was supported in part by the National Health and Medical Research Council 921–924. of Australia (program grant nos. 461219 and 487922 and project grant no. Ernst, M., and Jenkins, B.J. (2004). Acquiring signalling specificity from the 1011804), the U.S. National Institutes of Health (grant no. CA22556), the cytokine receptor gp130. Trends Genet. 20, 23–32. Victorian State Government Operational Infrastructure Support Grant, and Ghoreschi, K., Laurence, A., and O’Shea, J.J. (2009). Selectivity and thera- the NHMRC Independent Research Institutes Infrastructure Support Scheme peutic inhibition of kinases: to be or not to be? Nat. Immunol. 10, 356–360. (361646). J.J.B., R.S.N., and N.A.N. acknowledge fellowship support from the Guschin, D., Rogers, N., Briscoe, J., Witthuhn, B., Watling, D., Horn, F., National Health and Medical Research Council, and J.M.M. from the Australian Pellegrini, S., Yasukawa, K., Heinrich, P., Stark, G.R., et al. (1995). A major Research Council. L.N.V. acknowledges scholarship support from the role for the protein tyrosine kinase JAK1 in the JAK/STAT signal transduction Leukaemia Foundation of Australia and the Australian Stem Cell Centre. pathway in response to interleukin-6. EMBO J. 14, 1421–1429. N.A.N. is a founder and member of the scientific advisory board of MuriGen Pty Ltd. Haan, C., Kroy, D.C., Wu¨ ller, S., Sommer, U., No¨ cker, T., Rolvering, C., Behrmann, I., Heinrich, P.C., and Haan, S. (2009). An unusual insertion in Received: March 22, 2011 Jak2 is crucial for kinase activity and differentially affects cytokine responses. 182 Revised: November 24, 2011 J. Immunol. , 2969–2977. Accepted: December 21, 2011 Heo, Y.S., Kim, S.K., Seo, C.I., Kim, Y.K., Sung, B.J., Lee, H.S., Lee, J.I., Park, Published online: February 16, 2012 S.Y., Kim, J.H., Hwang, K.Y., et al. (2004). Structural basis for the selective inhi- bition of JNK1 by the scaffolding protein JIP1 and SP600125. EMBO J. 23, REFERENCES 2185–2195. Hilton, D.J., Richardson, R.T., Alexander, W.S., Viney, E.M., Willson, T.A., Adams, J.A. (2001). Kinetic and catalytic mechanisms of protein kinases. Sprigg, N.S., Starr, R., Nicholson, S.E., Metcalf, D., and Nicola, N.A. (1998). 101 Chem. Rev. , 2271–2290. Twenty proteins containing a C-terminal SOCS box form five structural Alexander, W.S., Starr, R., Fenner, J.E., Scott, C.L., Handman, E., Sprigg, classes. Proc. Natl. Acad. Sci. USA 95, 114–119. N.S., Corbin, J.E., Cornish, A.L., Darwiche, R., Owczarek, C.M., et al. (1999). Hobbs, S., Jitrapakdee, S., and Wallace, J.C. (1998). Development of a bicis- SOCS1 is a critical inhibitor of interferon gamma signaling and prevents the tronic vector driven by the human polypeptide chain elongation factor potentially fatal neonatal actions of this cytokine. Cell 98, 597–608. 1alpha promoter for creation of stable mammalian cell lines that express Babon, J.J., Yao, S., DeSouza, D.P., Harrison, C.F., Fabri, L.J., Liepinsh, E., very high levels of recombinant proteins. Biochem. Biophys. Res. Commun. Scrofani, S.D., Baca, M., and Norton, R.S. (2005). Secondary structure assign- 252, 368–372. ment of mouse SOCS3 by NMR defines the domain boundaries and identifies James, C., Ugo, V., Le Coue´ dic, J.P., Staerk, J., Delhommeau, F., Lacout, C., an unstructured insertion in the SH2 domain. FEBS J. 272, 6120–6130. Garc¸on, L., Raslova, H., Berger, R., Bennaceur-Griscelli, A., et al. (2005). A Babon, J.J., McManus, E.J., Yao, S., DeSouza, D.P., Mielke, L.A., Sprigg, N.S., unique clonal JAK2 mutation leading to constitutive signalling causes polycy- Willson, T.A., Hilton, D.J., Nicola, N.A., Baca, M., et al. (2006). The structure of thaemia vera. Nature 434, 1144–1148. SOCS3 reveals the basis of the extended SH2 domain function and identifies Jenkins, B.J., Roberts, A.W., Najdovska, M., Grail, D., and Ernst, M. (2005). 22 an unstructured insertion that regulates stability. Mol. Cell , 205–216. The threshold of gp130-dependent STAT3 signaling is critical for normal regu- Babon, J.J., Sabo, J.K., Soetopo, A., Yao, S., Bailey, M.F., Zhang, J.G., Nicola, lation of hematopoiesis. Blood 105, 3512–3520. N.A., and Norton, R.S. (2008). The SOCS box domain of SOCS3: structure and Kamizono, S., Hanada, T., Yasukawa, H., Minoguchi, S., Kato, R., Minoguchi, 381 interaction with the elonginBC-cullin5 ubiquitin ligase. J. Mol. Biol. , M., Hattori, K., Hatakeyama, S., Yada, M., Morita, S., et al. (2001). The SOCS 928–940. box of SOCS-1 accelerates ubiquitin-dependent proteolysis of TEL-JAK2. Babon, J.J., Sabo, J.K., Zhang, J.G., Nicola, N.A., and Norton, R.S. (2009). The J. Biol. Chem. 276, 12530–12538. SOCS box encodes a hierarchy of affinities for Cullin5: implications for ubiqui- Lacronique, V., Boureux, A., Monni, R., Dumon, S., Mauchauffe´ , M., Mayeux, 387 tin ligase formation and cytokine signalling suppression. J. Mol. Biol. , P., Gouilleux, F., Berger, R., Gisselbrecht, S., Ghysdael, J., and Bernard, O.A. 162–174. (2000). Transforming properties of chimeric TEL-JAK proteins in Ba/F3 cells. Barr, R.K., Boehm, I., Attwood, P.V., Watt, P.M., and Bogoyevitch, M.A. (2004). Blood 95, 2076–2083. The critical features and the mechanism of inhibition of a kinase interaction Levine, R.L., Wadleigh, M., Cools, J., Ebert, B.L., Wernig, G., Huntly, B.J., 279 motif-based peptide inhibitor of JNK. J. Biol. Chem. , 36327–36338. Boggon, T.J., Wlodarska, I., Clark, J.J., Moore, S., et al. (2005). Activating Boyle, K., Egan, P., Rakar, S., Willson, T.A., Wicks, I.P., Metcalf, D., Hilton, mutation in the tyrosine kinase JAK2 in polycythemia vera, essential thrombo- D.J., Nicola, N.A., Alexander, W.S., Roberts, A.W., and Robb, L. (2007). The cythemia, and myeloid metaplasia with myelofibrosis. Cancer Cell 7, 387–397.

Immunity 36, 239–250, February 24, 2012 ª2012 Elsevier Inc. 249 Immunity Mechanism of Inhibition of Janus Kinase by SOCS3

Lucet, I.S., Fantino, E., Styles, M., Bamert, R., Patel, O., Broughton, S.E., SH2 protein-3 (CIS3/SOCS3) inhibits Janus tyrosine kinase by binding through Walter, M., Burns, C.J., Treutlein, H., Wilks, A.F., and Rossjohn, J. (2006). the N-terminal kinase inhibitory region as well as SH2 domain. Genes Cells 4, The structural basis of inhibition by a potent and specific 339–351. pan-Janus kinase inhibitor. Blood 107, 176–183. Starr, R., Willson, T.A., Viney, E.M., Murray, L.J., Rayner, J.R., Jenkins, B.J., Madhusudan, J., Trafny, E.A., Xuong, N.H., Adams, J.A., Ten Eyck, L.F., Gonda, T.J., Alexander, W.S., Metcalf, D., Nicola, N.A., and Hilton, D.J. Taylor, S.S., and Sowadski, J.M. (1994). cAMP-dependent protein kinase: (1997). A family of cytokine-inducible inhibitors of signalling. Nature 387, crystallographic insights into substrate recognition and phosphotransfer. 917–921. Protein Sci. 3, 176–187. Starr, R., Fuchsberger, M., Lau, L.S., Uldrich, A.P., Goradia, A., Willson, T.A., McKendry, R., John, J., Flavell, D., Mu¨ ller, M., Kerr, I.M., and Stark, G.R. Verhagen, A.M., Alexander, W.S., and Smyth, M.J. (2009). SOCS-1 binding to (1991). High-frequency mutagenesis of human cells and characterization of tyrosine 441 of IFN-gamma receptor subunit 1 contributes to the attenuation of a mutant unresponsive to both alpha and gamma interferons. Proc. Natl. IFN-gamma signaling in vivo. J. Immunol. 183, 4537–4544. Acad. Sci. USA 88, 11455–11459. Murray, P.J. (2007). The JAK-STAT signaling pathway: input and output inte- Ungureanu, D., Wu, J., Pekkala, T., Niranjan, Y., Young, C., Jensen, O.N., Xu, gration. J. Immunol. 178, 2623–2629. C.F., Neubert, T.A., Skoda, R.C., Hubbard, S.R., and Silvennoinen, O. (2011). The pseudokinase domain of JAK2 is a dual-specificity protein kinase that Naka, T., Narazaki, M., Hirata, M., Matsumoto, T., Minamoto, S., Aono, A., negatively regulates cytokine signaling. Nat. Struct. Mol. Biol. 18, 971–976. Nishimoto, N., Kajita, T., Taga, T., Yoshizaki, K., et al. (1997). Structure and function of a new STAT-induced STAT inhibitor. Nature 387, 924–929. Wilks, A.F., and Harpur, A.G. (1994). Cytokine signal transduction and the JAK 16 Neel, B.G., Gu, H., and Pao, L. (2003). The ‘Shp’ing news: SH2 domain-con- family of protein tyrosine kinases. Bioessays , 313–320. taining tyrosine phosphatases in cell signaling. Trends Biochem. Sci. 28, Wormald, S., Zhang, J.G., Krebs, D.L., Mielke, L.A., Silver, J., Alexander, W.S., 284–293. Speed, T.P., Nicola, N.A., and Hilton, D.J. (2006). The comparative roles of Nicholson, S.E., Willson, T.A., Farley, A., Starr, R., Zhang, J.G., Baca, M., suppressor of cytokine signaling-1 and -3 in the inhibition and desensitization Alexander, W.S., Metcalf, D., Hilton, D.J., and Nicola, N.A. (1999). Mutational of cytokine signaling. J. Biol. Chem. 281, 11135–11143. analyses of the SOCS proteins suggest a dual domain requirement but distinct Yasukawa, H., Misawa, H., Sakamoto, H., Masuhara, M., Sasaki, A., Wakioka, mechanisms for inhibition of LIF and IL-6 signal transduction. EMBO J. 18, T., Ohtsuka, S., Imaizumi, T., Matsuda, T., Ihle, J.N., and Yoshimura, A. (1999). 375–385. The JAK-binding protein JAB inhibits Janus tyrosine kinase activity through Nicholson, S.E., De Souza, D., Fabri, L.J., Corbin, J., Willson, T.A., Zhang, J.G., binding in the activation loop. EMBO J. 18, 1309–1320. Silva, A., Asimakis, M., Farley, A., Nash, A.D., et al. (2000). Suppressor of cyto- kine signaling-3 preferentially binds to the SHP-2-binding site on the shared Yeung, K., Seitz, T., Li, S., Janosch, P., McFerran, B., Kaiser, C., Fee, F., cytokine receptor subunit gp130. Proc. Natl. Acad. Sci. USA 97, 6493–6498. Katsanakis, K.D., Rose, D.W., Mischak, H., et al. (1999). Suppression of Raf-1 kinase activity and MAP kinase signalling by RKIP. Nature 401, 173–177. O’Shea, J.J., and Murray, P.J. (2008). Cytokine signaling modules in inflamma- tory responses. Immunity 28, 477–487. Zhang, J.G., Farley, A., Nicholson, S.E., Willson, T.A., Zugaro, L.M., Simpson, Russo, A.A., Jeffrey, P.D., Patten, A.K., Massague´ , J., and Pavletich, N.P. R.J., Moritz, R.L., Cary, D., Richardson, R., Hausmann, G., et al. (1999). The (1996). Crystal structure of the p27Kip1 cyclin-dependent-kinase inhibitor conserved SOCS box motif in suppressors of cytokine signaling binds to bound to the cyclin A-Cdk2 complex. Nature 382, 325–331. elongins B and C and may couple bound proteins to proteasomal degradation. Proc. Natl. Acad. Sci. USA 96, 2071–2076. Saharinen, P., Takaluoma, K., and Silvennoinen, O. (2000). Regulation of the Jak2 tyrosine kinase by its pseudokinase domain. Mol. Cell. Biol. 20, 3387– Zhang, J.G., Metcalf, D., Rakar, S., Asimakis, M., Greenhalgh, C.J., Willson, 3395. T.A., Starr, R., Nicholson, S.E., Carter, W., Alexander, W.S., et al. (2001). Saharinen, P., Vihinen, M., and Silvennoinen, O. (2003). Autoinhibition of Jak2 The SOCS box of suppressor of cytokine signaling-1 is important for inhibition 98 tyrosine kinase is dependent on specific regions in its pseudokinase domain. of cytokine action in vivo. Proc. Natl. Acad. Sci. USA , 13261–13265. 14 Mol. Biol. Cell , 1448–1459. Zhao, L., Ma, Y., Seemann, J., and Huang, L.J. (2010). A regulating role of the Sasaki, A., Yasukawa, H., Suzuki, A., Kamizono, S., Syoda, T., Kinjyo, I., JAK2 FERM domain in hyperactivation of JAK2(V617F). Biochem. J. 426, Sasaki, M., Johnston, J.A., and Yoshimura, A. (1999). Cytokine-inducible 91–98.

250 Immunity 36, 239–250, February 24, 2012 ª2012 Elsevier Inc.