The SOCS2 Ubiquitin Ligase Complex Regulates Growth Hormone Receptor Levels

Growth Hormone is essential for the regulation of growth and the homeostatic control of intermediary metabolism. GH actions are mediated by the Growth Hormone Receptor; a member of the cytokine receptor super family that signals chiefly through the JAK2/STAT5 pathway. Target tissue responsiveness to GH is under regulatory control to avoid excessive and off-target effects upon GHR activation. The suppressor of cytokine signalling 2 (SOCS) is a key regulator of GHR sensitivity. This is clearly shown in mice where the SOCS2 gene has been inactivated, which show 30–40% increase in body length, a phenotype that is dependent on endogenous GH secretion. SOCS2 is a GH-stimulated, STAT5b-regulated gene that acts in a negative feedback loop to downregulate GHR signalling. Since the biochemical basis for these actions is poorly understood, we studied the molecular function of SOCS2. We demonstrated that SOCS2 is part of a multimeric complex with intrinsic ubiquitin ligase activity. Mutational analysis shows that the interaction with Elongin B/C controls SOCS2 protein turnover and affects its molecular activity. Increased GHR levels were observed in livers from SOCS2−/− mice and in the absence of SOCS2 in in vitro experiments. We showed that SOCS2 regulates cellular GHR levels through direct ubiquitination and in a proteasomally dependent manner. We also confirmed the importance of the SOCS-box for the proper function of SOCS2. Finally, we identified two phosphotyrosine residues in the GHR to be responsible for the interaction with SOCS2, but only Y487 to account for the effects of SOCS2. The demonstration that SOCS2 is an ubiquitin ligase for the GHR unveils the molecular basis for its physiological actions.


Introduction
Growth Hormone (GH) is a peptide hormone of 192 amino acids secreted from the anterior pituitary [1] that plays a key role in the regulation of intermediary metabolism and longitudinal bone growth [2,3]. GH exerts its actions by binding to the Growth Hormone Receptor (GHR), which is present on cell membranes in several different organs and tissues. The GHR belongs to the cytokine receptor super family and signals chiefly through the Janus Kinase/Signal Transducer and Activator of Transcription (JAK/STAT) pathway. Upon GH binding the GHR is tyrosine phosphorylated by JAK2; this allows STAT5 to bind the receptor with its Src Homology 2 (SH2) domain. Next, STAT5 is phosphorylated by JAK2, dimerizes and migrates to the nucleus where it induces the expression of GH-responsive genes.
One of the genes induced by GH stimulation is the suppressor of cytokine signalling 2 (SOCS2). SOCS2 is an important negative regulator of GHR signalling as clearly demonstrated by studies of SOCS2 2/2 mice [4][5][6]. SOCS2 deficient mice show a 40% increase in body size due to greatly enhanced postnatal growth [6]. GH-deficient, SOCS2 2/2 mice failed to show the above-described phenotype, while treatment with exogenous GH induced excessive growth in terms of overall body weight, body and bone lengths, and the weight of internal organs and tissues. Microarray analysis on liver RNA extracts after exogenous GH administration revealed a heightened response to GH [5].
SOCS2 exerts its action through feedback inhibition of GH signalling. GH activated STAT5b binds the promoter of SOCS2 to promote expression, in turn SOCS2 binds at least two phosphorylated tyrosines on the GHR to negatively regulate JAK2 and STAT5b activation [5,7]. The molecular basis for these actions of SOCS2 are not entirely clear and two main mechanisms of action have been proposed, one is that SOCS2 binding to the GHR blocks the association of positive regulators [5,[8][9][10]. The other hypothesis is that SOCS2 is part of an ubiquitin ligase complex targeting interacting proteins for proteasomal degradation [8,9]. Several findings indicate that this is an important part of SOCS2 activity. SOCS2 is characterized by a SOCS-box at the Cterminal end of the protein, similarly to other ubiquitin ligases such as the Von Hippel-Lindau gene product (VHL) and SOCS1 [11][12][13]. The SOCS-box in SOCS2 interacts with Elongin B and C as revealed by the x-ray structure of the SOCS2-Elongin B/C complex [8]. In addition, another important component of SOCS ubiquitin ligases, Cullin5, has been shown to associate with SOCS2 [14]. Furthermore, deletions of the SOCS-box abrogate the inhibitory actions of SOCS2 on GH induced STAT5b activation [5]. Finally, SOCS2 was shown to be involved in the ubiquitination and proteasomal degradation of the phosphorylated proline-rich tyrosine kinase 2 (Pyk2) in Natural Killer cells in a SOCS-box dependent manner [15]. Despite all the indirect evidence, it remains undetermined whether SOCS2 actions can be explained by ubiquitination and proteasomal degradation of the key intracellular signalling components: GHR, JAK2 or STAT5b or if they involve the degradation of yet unidentified regulatory proteins. A better understanding of SOCS2 mode of action can not only shed light on the basis for its physiological functions related to GH but also to on other described activities related to inflammation, mammary gland and brain development that cannot be attributed to GH actions.
Here, we have used mutagenesis to study the influence of several domains in the SOCS2 protein on its activity. For the first time, we demonstrate a direct ubiquitin ligase activity of SOCS2 directed towards the GHR. Integrity of the SH2-domain and the SOCSbox are essential to regulate SOCS2 protein levels in mammalian cells as well as ubiquitination activity in vitro. Accordingly, point mutations in these domains are enough to impair the capacity of SOCS2 to downregulate GHR levels. Our studies also reveal the importance of tyrosine 487 on the GHR for its regulation by SOCS2 and thus provide a mechanistic explanation for the actions of SOCS2 on GH signalling.

The interaction with Elongin B and C regulates SOCS2 turnover
Several studies have failed to find SOCS2 inhibitory activity towards GH activated signalling pathways despite the clear demonstration of SOCS2 activity towards GHR signalling in vivo [16]. One possibility to explain this discrepancy is that SOCS2 associated proteins such as Elongin B and Elongin C may need to be co-expressed with SOCS2 to achieve maximal biological activity. Therefore, we first tested whether SOCS2 forms an Elongin-Cullin-SOCS (ECS) complex in cells and how different SOCS2 domains contribute to those interactions. Expression vectors with SOCS2 cDNA containing an N-terminally located FLAG-tag were used ( Figure 1A). The expression constructs were transfected into HEK293T cells and SOCS2 variants were immunoprecipitated to investigate SOCS2 interacting proteins. As shown in figure 1B, WT-SOCS2 immune precipitates were found to contain Elongin B, Elongin C, Cullin5 and Ring-box protein 2 (Rbx2) all of which are known to form part of the ECS core complex (lane 2). The interactions were confirmed by mass spectrometric analysis (data not shown). We also demonstrated that the domain responsible for the interactions is the SOCS-box. Within the SOCS-box we introduced two point mutations on wellconserved residues; L163P and C167F [8]. These residues are required for Elongin B/C binding in other SOCS-box containing proteins [17]. We did not detect Elongin B, Elongin C, Cullin5 or Rbx2 in the immunoprecipitates from the two point mutations or the SOCS-box deletion mutant (SOCSDSB) ( Figure 1B, lanes 5-7). However, the three constructs with an intact SOCS-box did coprecipitate with Elongin B, Cullin5 and Rbx2 (lanes 2-4). The SH2-domain mutant (SOCS2SH2) and the N-terminal deletion mutant (SOCS2DNT) did not interact with Elongin C. The reason why the SH2-domain mutant and the N-terminal deletion mutant fail to bind Elongin C is yet unclear. On the other hand, this is unlikely to be an artefact, since previous data from our lab demonstrate that a similar mutation in SOCS6 also abolishes Elongin C binding [18].
We further examined the interaction between SOCS2 and Elongin B/C. Experiments shown in Figure 1C demonstrate that SOCS2 protein content in transfected cells becomes significantly higher if Elongin C and B levels are also increased by cotransfection. The same is true for the two mutated versions of SOCS2 with an intact SOCS-box. As expected the protein levels of the mutants that lack or have a mutated SOCS-box and are not able to bind Elongin B/C are not affected by the presence of Elongin B/C ( Figure 1D).
To evaluate whether or not the increased SOCS2 level in the presence of Elongins is a due to a folding or a degradation event we treated transfected cells with cycloheximide to block protein synthesis ( Figure 1E). SOCS2 levels decrease faster when Elongins B/C are not co-transfected and remain stable for a longer time in the presence of Elongin B/C. From these experiments we concluded that the SOCS-box is essential for the formation of the ECS-complex and that formation stabilizes SOCS2 and reduces its degradation.

SOCS2 displays E3 ubiquitin ligase activity in vitro
SOCS2s ability to bind to Elongin B and C, Cullin5 and Rbx2 led us to examine the functional relevance of this interaction. Since these four proteins are known to interact with other SOCS-box containing ubiquitin ligases [19][20][21], we wanted to investigate if the SOCS2 complex has a capacity as an ubiquitin ligase. HEK293T cells were transiently transfected with either FLAG-SOCS2 or an empty vector in concert with Elongin B and C. Cells were lysed and FLAG-tagged proteins were immunoprecipitated, and the immunoprecipitates were incubated at 30uC with human recombinant E1, UbcH5b (E2), HA-ubiquitin (Ub) and an ATP regenerating system for 30 minutes. When analyzed with a Western blot, high molecular weight signals corresponding to HA-Ub covalently attached to high molecular weight protein complexes are detected when all components of the system are present (Figure 2A, lane 2), an indication of ubiquitin ligase activity in the precipitates. We also assayed the activity of the various mutated SOCS2 constructs using the above mentioned system ( Figure 2B). Our results show that the SOCS2DSB mutant ( Figure 2B, lane 4) exhibits clearly decreased ubiquitin ligase activity as compared to the wild-type ( Figure 2B, lane 2). In line with its capacity to bind Cullin5/Rbx2, we found the SOCS2-SH2 mutant also displayed ubiquitin ligase activity, although at a lower level when compared to the wild type variant. This is in line with recently published data on SOCS2 where WT-SOCS2 was shown to precipitate with ubiquitinated Pyk2 whilst SOCS2DSB only precipitated with the un-ubiqutinated protein [15]. We thus concluded that SOCS2 is an ubiqutin ligase and that the SOCSbox is essential for this function.

SOCS2 regulates Growth Hormone Receptor levels
Ubiquitination is now recognized as a regulatory posttranslational modification controlling protein-protein interactions and thereby influencing a variety of cellular processes. Given the inhibitory role of SOCS2 in GHR signalling and since SOCS2 exhibits ubiquitin ligase activity and can bind to phosphorylated tyrosines on the GHR [5,10], we hypothesized that SOCS2 specifically regulates GHR turnover through proteasomal degradation. To test this hypothesis, we co-transfected SOCS2, Elongin C and Elongin B expression vectors with a GHR expression vector in HEK293T cells and analysed GHR levels by Western Blot. Prior to lysis cells were cultivated for 4 hours in DMEM without FBS and stimulated with GH for the indicated times. As shown in Figure 3A, we detected a major band and a weaker one corresponding to the immature GHR (GHR i; top left panel) and after a longer exposure a band of approximately 130 kDa corresponding to the mature, glycosylated form of the receptor (GHR m) [22], which is present in much lower concentrations. The expression of SOCS2 resulted in a significant reduction of both GHR forms, even in the absence of exogenously added GH. In contrast, cells transfected with SOCS2DSB did not differ from control cells regarding GHR levels. The SOCS2-SH2 mutant variant does have an effect on GHR levels but not as pronounced as the one of WT-SOCS2.
We also examined whether the mutant versions of SOCS2 were able to bind the receptor. As expected WT-SOCS2 and SOCS2DSB both interact with the GHR, but as figure 3B shows so does the SOCS2-SH2 construct although to a significantly lesser extent. The weak interaction is still detected which suggests that SOCS2 might be able to interact with the GHR using an extended interface, not entirely dependent on phosphotyrosine binding by the SH2-domain or alternatively involving a third partner.
The interaction between WT-SOCS2 and the GHR generally seems to be facilitated by stimulation with GH, most likely since this leads to increased tyrosine-phosphorylation of the receptor as indicated by the fact that the SOCS2-SH2 mutant binds significantly less GHR after GH stimulation. We also notice that the SOCS2-SH2 residual GHR binding activity is also GH stimulated. As for the reasons for this observation, we can only speculate but its relevance is arguably minor since the amount of GHR bound by the SOCS SH2-domain is just a minor fraction of that bound by the WT-SOCS2. Although we cannot entirely explain the unexpected behaviour of the SOCS2-SH2 mutant, we can certainly conclude from these experiments that the SH2domain is a significant contributor to interaction of SOCS2 with the GHR in cells.
We did not observe a truly ligand dependent decrease in GHR levels in SOCS2 overexpressing cells ( Figure 3A). Non stimulated cells also exhibit significantly decreased levels of the receptor. This could be related to the fact that significant amounts of GHR tyrosine phosphorylation are observed in GH untreated cells likely due to spontaneous dimerization and activation of GHR in the transfected cells.
To ascertain if the decreased GHR levels observed in SOCS2 expressing cells are due to an increase in degradation, we treated the cells with cycloheximide and examined the GHR levels in the presence and absence of SOCS2 ( Figure 3C). In the absence of SOCS2 the levels of the mature GHR start to diminish after approximately 2 hours whilst mature GHR levels are undetectable after 1 hour in the presence of SOCS2 demonstrating that SOCS2 actively promotes the degradation of the mature form of the GHR.
In order to further substantiate the role of SOCS2 in the regulation of GHR protein levels we used a previously validated siRNA [23] to knockdown SOCS2 and examined the effect on GHR levels. Knockdown of SOCS2 increases the levels of GHR present in the cell, an effect that is more pronounced in the mature GHR form. Consequently, the phosphorylation of STAT5 upon GH treatment is increased in SOCS2 siRNA treated cells ( Figure 4A). Similar results were observed when liver from SOCS2 2/2 mice were analyzed. The levels of GHR were also enhanced in SOCS2 2/2 mice livers as compared to WT controls ( Figure 4B). In conclusion, in the absence of SOCS2 the mature GHR levels and downstream GHR signalling are increased ( Figure 4A).
From these results we ascertained that SOCS2 indeed regulates the level of GHR present in the cell and that this is an effect dependent on a functional SOCS-box ( Figure 3A). The primary effect of SOCS2 seems to be increased degradation of the mature signalling form of GHR ( Figure 3C).

SOCS2 targets the GHR for proteasomal degradation
It has been reported earlier that the proteasome and the ubiquitin system is involved in the endocytosis of GHR [24]. Furthermore SOCS2 has been shown to regulate the levels of Pyk2 and to increase its ubiquitination [15]. To elucidate if SOCS2 regulates GHR availability by a proteasomal pathway, we treated cells transfected with SOCS2, Elongins and GHR with proteasome inhibitors MG132 and Bortezomib ( Figure 5A). Both proteasome inhibitors counteracted the reducing effect of SOCS2 on receptor levels. Again, the effect is mainly noticeable for the mature receptor. This suggests that SOCS2s regulation of GHR levels is dependent on a functional proteasome.
To further examine the mechanism by which SOCS2 affects the GHR we performed an in vitro ubiquitination assay with SOCS2 that had been co-precipitated with GHR. After the completion of the ubiquitination assay, SOCS2 was eluted from the beads and the growth hormone receptor was immunoprecipitated with a myc-antibody and its ubiquitination level was examined by probing with an anti HA-tag antibody. As shown in Figure 5B, high molecular weight Myc-GHR/HA-Ubiquitin conjugates are more abundant when SOCS2 and GHR are co-precipitated and SOCS2 effects are mediated through Tyr487 on the GHR Since SOCS2 contains an SH2-domain capable of binding phosphorylated tyrosines we used mutational analysis to determine which tyrosines on the GHR are important for SOCS2s regulation of the receptor. It has previously been reported that the SH2domain of SOCS2 binds to phosphopeptides corresponding to the residues Y487 and Y595 on the GHR [5,10]. We employed four different GHR constructs (Y487F, Y534F, Y595F and Y487/ 595F) and assayed the effect of SOCS2 on the levels of these mutant receptors ( Figure 6A). The levels of two of the mutants are clearly less affected by the presence of SOCS2; namely Y487F and the double mutant Y487/595F. This demonstrates that tyrosine  (lanes 3,4). Cells were starved and treated with GH for 0 (2) or 10 (+) minutes prior to lysis. Lysates were immunoprecipitated as described in the materials and methods and results visualized with western blot. GHR was detected with an anti-Myc antibody. (C) HEK293T cells transfected with either Myc-GHR, FLAG-SOCS2 and Elongin B/C or GHR and Elongin B/C alone were treated with cycloheximide (100 mg/ml) and GH (2 mg/ml) for the times indicated in the figure before lysis. Cell lysates were visualized by western blot. C denotes the control which was transfected with Empty vector instead of GHR. GHR was detected with an anti-GHR antibody. doi:10.1371/journal.pone.0025358.g003 487 is of great functional importance for SOCS2 mediated regulation of GHR. This is in line with previous work from our lab that has demonstrated that STAT5b signalling for the Y487F and Y487/595F GHR mutants are unaffected by increasing concentrations of SOCS2 [5]. Surprisingly, the SOCS2 SH2-domain has a higher affinity for peptides corresponding to phosphorylated Tyr595 than for those containing phosphorylated Tyr487 [5] but these differences in affinity doesn't seem relevant for the function of SOCS2 in the control of GHR levels.

Discussion
In this work we provide a mechanistic explanation for how SOCS2 acts as a negative regulator of GHR signalling [5,10,25]. SOCS2 assembles a canonical ECS complex with intrinsic ubiquitin ligase activity through interactions with Cullin5, Rbx2, and Elongins C and B. The SOCS2 ubiquitin ligase complex has the ability to ubiquitinate the GHR in vitro and downregulate GHR in a proteasomally dependent manner. The downregulatory activity of SOCS2 on the GHR is mediated by both a functional SOCS-box and by its interaction with Tyr487 in the intracellular domain of the GHR.
The structure of Elongin B/C in complex with the SOCS-box domain of SOCS2, SOCS3 and VHL has been deciphered and predict a very strong affinity between Elongin C and the SOCSbox which has lead to the proposal that Elongin B/C interaction is important for the correct folding and expression of the SOCS proteins [8,[26][27][28]. Our data showing the reduced expression of the SOCS2 L163P and C167F SOCS-box and the R73E SH2domain mutants, which are defective in Elongin C interaction lend support to these hypotheses and is in line with previous data from studies with VHL tumour suppressor SOCS-box point mutations [29]. However, the entire deletion of the C-terminal end of SOCS2 (SOCS2DSB) resulted in dramatically increased levels of SOCS2 as compared to the WT. For the SOCS-box deficient SOCS2DSB, the stability is not influenced in any way by the presence of the two Elongins. We can also note that its protein level remains constantly high for all experiments. The simplest explanation for this observation is that the SOCS2 C-terminus also contains signals for degradation which reside outside of the SOCS-box. A possibility is that Cullin5, whose binding motif in SOCS2 lays outside the Elongin B/C binding domain [30] mediates the autoubiquitination and degradation of SOCS2. Indeed, autoubiquitination of adaptor subunits has been described for the Vif-Elongin B/C-Cullin5 complex [31] and the structurally related Skp-Cullin-F-box family of E3 ligases [32]. However, we have not detected the appearance of SOCS2-ubiquitin complexes during in vitro ubiquitination assays and further studies are needed before a definite conclusion can be drawn.
We have demonstrated for the first time that the SOCS2 complex exhibits ubiquitin ligase activity in vitro, specifically towards the GHR. The analysis of mutant variants of SOCS2, confirm that SOCS2 activity is dependent on the interaction with Elongin B/C and Cullin5/Rbx2 through its C-terminus. The SH2-mutant showed ubiqutin ligase activity, which is in line with the finding that this mutant retains the ability to interact with Cullin5 and Rbx2 although it shows reduced binding to Elongin C. This may explain the residual ubiqutin ligase activity observed in the in vitro ubiquitination assays for this mutant since the ring finger protein Rbx2 is considered an essential component of the enzymatic reaction leading to the transfer of ubiquitin moieties to the target proteins. We have also shown that overexpression of SOCS2 downregulates GHR levels, especially the mature form of the receptor. This downregulation is not evident when the SOCS2DSB mutant is overexpressed, even though this mutant retains the ability to bind to the receptor. This demonstrates that the ability to form a complete E3 ligase complex is essential for SOCS2 ability to regulate GHR levels. However, the SH2-domain mutant retains both a slight ability to bind to the receptor and has an effect on GHR levels; this suggests that the SH2-domain might not be the only domain involved in GHR binding. The existence of target binding motifs outside the SH2-domain has also been demonstrated for SOCS6, where an additional interacting domain found in the N-terminus is involved in the recognition and ubiquitination of the p56 lck tyrosine kinase [33].
Previous publications have shown that the proteasome plays an important role in GHR endocytosis and degradation [34]. We also investigated the effect of proteasome inhibitors on GHR levels and found that these counteract SOCS2 and lead to increased GHR levels. Together with the evidence that SOCS2 is able to ubiquitinate the GHR in vitro, it would suggest that SOCS2 mediated ubiquitination leads to endocytosis and proteasomal degradation of the receptor. On the other hand SOCS2 is not the only ubiquitin ligase acting on the GHR. Van Kerkhof et al. have showed how the ubiquitin ligase ß-TrCP is involved in the regulation of GHR endocytosis and degradation in a ligand independent manner [35]. We have also noted in our experiments that siRNA based ß-TrCP downregulation increase GHR levels. In this background, SOCS2 retains its ability to downregulate GHR levels, arguing for an independent effect of SOCS2 on GHR turnover (data not shown). The dependency of SOCS2 mediated downregulation on GHR Tyr487 phosphorylation, add support for the existence of parallel pathways of GHR ubiquitination driven by ß-TrCP and SOCS2. The fact that SOCS2 seems to preferentially regulate mature, activated receptors seem to be a crucial aspect of its physiological actions as manifested in the phenotypic gigantism observed in SOCS2 2/2 mice in comparison with the normal growth rate reported for ß-TrCP1 2/2 mice [4,36].
Several studies indicate that in addition to its actions on GHR signalling, SOCS2 also has important actions in neuron development [37], the regulation of metabolism [38], mammary gland development and cancer [39] as well as the immune response [39][40][41]. Although it seems unlikely that the various functions of SOCS2 are all GHR mediated, the wide tissue distribution of SOCS2 [42] together with our demonstration that SOCS2 controls GHR levels invite a further exploration of these phenotypes to assess the possible role of GHR signalling. Likewise, future studies of SOCS2 mechanisms of action should include an assessment of its ubiquitin ligase activity as an end-point measurement. Already, several polymorphisms of the SOCS2 gene has been identified in humans, one linked to height [43] while the other one is linked to type 2 diabetes [44]. Our results suggest that disorders associated to SOCS2 lack of function may possibly be treated by GHR antagonists, one of which is already in clinical use. This is an exiting opportunity that awaits further validation of the role of SOCS2 in human disease. Animal experiments SOCS2 2/2 mice (C57BL/6J) have been described [6]. Mice were housed under controlled temperature (23uC) and lighting (12 hr-light and 12 hr-dark cycles) with free access to water and food. Male SOCS2 2/2 and wild-type (WT) mice between 27-28 weeks old were allocated into groups (3-7 mice per cage) and were fed a standard control diet. Mice were sacrificed by CO 2 gas followed by cervical dislocation. Livers were excised and snapfrozen in liquid nitrogen before being stored at 280uC. 20-50 mg of liver tissue was homogenized in 1mL ice-cold buffer containing: 50 mM Tris HCl pH 7.5/150 mM NaCl/5 mM EDTA/0.5% Igepal-40/1 mM Na 3 VO 4 /20 mM NaF/1 mM dithiothretiol (DTT)/1 mM phenylmethanesulphonylfluoride (PMSF)/16 Cocktail inhibitor (Complete mini, Roche)/16 phosphatase inhibitor cocktail 2 (Sigma). Homogenates were cleared by centrifugation at 14,0006 g for 15 minutes at 4uC.

Transfection
Amounts of between 3-10 mg of plasmid DNA was used for transfections. Transfections were carried out by employing SuperFect (Qiagen) according to the manufacturer's directions. SOCS2, Elongin C and Elongin B were transfected at a 2:1:1 ratio. Samples lacking any of the components were brought to the same amount of DNA by addition of an empty vector.

Immunoprecipitation
HEK 293T cells were cultivated in Ø100 mm plates and transfected with 8-10 mg of DNA as described above. Cells were lysed in 50 mM Tris HCl, pH 7.5/150 mM NaCl/5 mM EDTA/ 0.5% Igepal-40/1 mM Na 3 VO 4 /20 mM NaF/1 mM PMSF/16 Cocktail inhibitor (Roche). Lysates were centrifuged at 14,0006 g for 15 minutes at 4uC to remove cell debris. Supernatants from identical samples were pooled and rotated with 10 ml sepharose beads (Amersham) per sample for 1 h at 4uC to remove proteins binding to sepharose. Samples were spun down for 2 minutes at 1000 rpm at 4uC. For immunoprecipitation with anti FLAG- antibody supernatants were rotated with 10 ml Flag-conjugated agarose beads (Sigma) per sample for 1 h at 4uC. For immunoprecipitation with anti-Myc antibody, supernatants were rotated with 2 mg of antibody over night at 4uC. After supernatants had been removed the beads were washed in Equalizing buffer; 50 mM Tris HCl, pH 7.5/150 mM NaCl/ 5 mM EDTA.

In vitro ubiquitination assay
Following immunoprecipitation beads were washed three times in HEB buffer; 20 mM Tris HCl, pH 7.5/5 mM KCl/1.5 mM MgCl 2 /1 mM DTT/10 mM MG132. Beads were resuspended in 20 ml HEB buffer per sample. The packed and washed Flagconjugated beads containing FLAG-WT or FLAG-mutant SOCS2 complexes were incubated with 1 ml E1 (0.5 mg/ml His 6tagged, human recombinant, Biomol)/1.5 ml E2 (1 mg/ml His 6tagged, human recombinant, Biomol)/8 ml 200 mM Ub/5 ml 100 mM HA-Ub/8 ml ATP regenerating system (10 mM ATP/ 20 mM Tris HCl, pH 7.5/10 mM MgCl 2 /300 mM creatinine phosphate (Roche)/0.5 ng/ml rabbit creatinine kinase (Calbiochem)) at 30uC for 30 minutes. The reaction was carried out in HEB buffer and the total reaction volume was 100 ml. Reactions were either stopped by the addition of 46SDS sample buffer (Invitrogen) followed by heating to 70uC or by the addition of 20 mM EDTA. For samples stopped with EDTA, proteins bound to the beads were eluted by a 2 hour incubation with FLAGpeptide (100 mg/ml) followed by immunoprecipitation with anti-Myc antibody as described above.