The Stat3-activating Tyk2 V678F Mutant Does Not Up-regulate Signaling through the Type I Interferon Receptor but Confers Ligand Hypersensitivity to a Homodimeric Receptor*

Tyk2 is a Jak family member involved in cytokine signaling through heterodimeric-type receptors. Here, we analyzed the impact of the Val678-to-Phe substitution on Tyk2 functioning. This mutation is homologous to the Jak2 Val617-to-Phe mutation, implicated in myeloproliferative disorders. We studied ligand-independent and ligand-dependent Jak/Stat signaling in cells expressing Tyk2 V678F. Moreover, the effect of Tyk2 V678F was monitored in the context of the native heterodimeric interferon α receptor and in the context of a homodimeric receptor chimera, EpoR/R1, containing the ectodomain of the erythropoietin receptor. We show that Tyk2 V678F has increased catalytic potential in vivo and in vitro and more so when it is anchored to the homodimeric receptor. Tyk2 V678F leads to constitutive Stat3 phosphorylation but has no notable effect on the canonical interferon α-induced signaling. However, if anchored to the homodimeric EpoR/R1, the mutant confers to the cell increased sensitivity to erythropoietin. Thus, despite the catalytic gain of function of Tyk2 V678F, the effect on ligand-induced signaling is manifest only when two mutant enzymes are juxtaposed via the homodimeric receptor.

Tyk2 is a Jak family member involved in cytokine signaling through heterodimeric-type receptors. Here, we analyzed the impact of the Val 678 -to-Phe substitution on Tyk2 functioning. This mutation is homologous to the Jak2 Val 617 -to-Phe mutation, implicated in myeloproliferative disorders. We studied ligand-independent and ligand-dependent Jak/Stat signaling in cells expressing Tyk2 V678F. Moreover, the effect of Tyk2 V678F was monitored in the context of the native heterodimeric interferon ␣ receptor and in the context of a homodimeric receptor chimera, EpoR/R1, containing the ectodomain of the erythropoietin receptor. We show that Tyk2 V678F has increased catalytic potential in vivo and in vitro and more so when it is anchored to the homodimeric receptor. Tyk2 V678F leads to constitutive Stat3 phosphorylation but has no notable effect on the canonical interferon ␣-induced signaling. However, if anchored to the homodimeric EpoR/R1, the mutant confers to the cell increased sensitivity to erythropoietin. Thus, despite the catalytic gain of function of Tyk2 V678F, the effect on ligand-induced signaling is manifest only when two mutant enzymes are juxtaposed via the homodimeric receptor.
The protein tyrosine kinases of the Janus or Jak (Janus kinase) family (Tyk2, Jak1-3) initiate signaling cascades through receptors for helix-bundled cytokines. In mammals, these large enzymes are preassociated to the membrane-proximal cytosolic portion of receptors and play non-redundant functions in cytokine-regulated developmental, differentiation, and host defense processes as shown from biochemical analyses and from studies of knock-out mice (1). The present understanding of the molecular mechanism of Jak activation is still fragmen-tary: some cytokines bind in a two-step process to constitutive receptor homodimers, whereas other cytokines drive dimerization of the distinct subunits of heterodimeric receptors. Conformational and rotational changes are generated and translated across the transmembrane domains so that the juxtaposed Jaks acquire a configuration suitable for transphosphorylation and catalytic activation. This process has been best described for the erythropoietin (Epo) 5 and growth hormone homodimeric receptors, which activate two associated Jak2 molecules (2,3). Instead, in heteromeric-type receptors the mode of activation of two different Jaks and their specific role remains ill defined.
In all Jak proteins, three highly interdependent modules ensure proper regulation of the enzyme. The N-terminal region, comprising a FERM-related domain and a SH2-like domain, makes specific contacts with the cytosolic region of a cognate receptor. The N-terminal region is followed by two kinase folds, the centrally located KL (kinase-like) domain, also called pseudokinase or JH2, and the catalytic TK (tyrosine kinase) or JH1 domain. Upon ligand binding, the degree of activation of these enzymes is commonly monitored by the phosphorylation of critical tyrosines in the activation loop of the TK domain (4). Numerous studies have shown that the KL domain is central to the auto-and trans-regulation of Jak catalytic activity. On the one hand, it locks the enzyme in an unphosphorylated or low phosphorylated state, by possibly restricting the flexibility and the exposure of the activation loop. On the other hand, in response to cytokine stimulation, the KL domain plays a positive, though ill-defined, role in driving or stabilizing the protein toward an activated state (5-7).
The negative regulatory role of the KL domain is best supported by the deregulated behavior of the Jak2 V617F mutant found in Polycythemia vera patients and implicated in a wide range of myeloproliferative disorders (8 -12). This valine residue, located in the KL domain, is conserved in Jak1 and Tyk2 but not in Jak3, where it is replaced by a methionine. All findings to date are concordant in attributing a gain of function to Jak2 V617F. Similar conclusions were drawn for the corresponding Jak1 V658F and Tyk2 V678F mutants, based on their * This work was supported by grants from the Institut Pasteur, the Association pour la Recherche sur le Cancer (ARC 3158), CNRS, and INSERM. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1  phosphorylation status and their ability to transform murine pro-B Ba/F3 cells to cytokine-independent cells (13,14). Notably, the impact of this Val-to-Phe substitution in Jakmediated signaling through a heterodimeric receptor has not been explored. Indeed, in this context, a mutant Jak would be juxtaposed to the wild-type (WT) form of another Jak family member. Here, we have addressed this issue by studying the functional consequences of Tyk2 V678F in type I IFN-induced signaling in human cells. The type I IFN receptor is a prototypic class 2 cytokine receptor consisting of two transmembrane chains, IFNAR1 and IFNAR2, which are not preassociated in the absence of ligand and which bind Tyk2 and Jak1, respectively (15). Using Tyk2-deficient cells, we show that the Tyk2 V678F mutant, expressed at nearly physiological level, is hyperphosphorylated and hyperactive with respect to the WT protein but has no effect on the profile of Jak1/Stat1/Stat2 (signal transducer and activator of transcription) activation by IFN␣. However, it does augment phosphorylation of Stat3 in a ligandindependent fashion. We have also analyzed the impact of Tyk2 V678F on signaling through an EpoR/R1 chimeric receptor forming homodimers. We demonstrate that, in this context, in addition to constitutively targeting Stat3 phosphorylation, Tyk2 V678F confers to the cell increased ligand sensitivity.
Immunoprecipitation and in Vitro Kinase Assay-Cells were lysed in 50 mM Tris, pH 6.8, 0.5% Nonidet P-40, 200 mM NaCl, 10% glycerol, 1 mM EDTA, 1 mM sodium vanadate, 1 mM sodium fluoride, 10 mM phenylmethylsulfonyl fluoride, 3 g/ml aprotinin, 3 g/ml leupeptin, and 3 g/ml pepstatin. Tyk2 was immunoprecipitated from 1 mg of lysate using R5-7 polyclonal Abs. Beads were washed three times with 50 mM Tris, pH 6.8, 400 mM NaCl, 0.5% Triton X-100, and 1 mM EDTA; once with 50 mM Tris, pH 6.8, and 200 mM NaCl; and once with kinase buffer (50 mM Hepes, pH 7.4, and 10 mM MgCl 2 ). Beads were then resuspended in 10 l of kinase buffer with or without 50 M ATP (Fermentas). The in vitro kinase reaction was performed at 30°C for 5 min in a total volume of 20 l. The reaction was stopped by boiling in Laemmli buffer, and the phosphorylated products were analyzed by Western blotting with the anti-phosphotyrosine 4G10 mAb and, after stripping, with the anti-Tyk2 mAb. Signals were acquired and quantified with a Kodak Image Station 440 CF.
Quantitative Reverse Transcription-PCR-Total RNA from control or IFN␣-treated cells was purified with RNeasy columns (Qiagen). Reverse transcriptions were primed with random primers and performed using Moloney murine leukemia virus reverse transcriptase (Invitrogen). Quantitative real-time PCR was performed using the TaqMan gene expression assays technology (Applied Biosystems) for IRF1 (Hs00233698-m1), 6-16 (Hs00242571-m1), and SOCS3 (Hs00269575-m1). Each sample was run in triplicate, normalized to 18 S RNA amplification levels in the same sample, and calculated relative to expression of the target gene in unstimulated cells.

Tyk2 V678F Has a High Basal Autophosphorylation Activity-
To study the effect of the V678F substitution, Tyk2WT and Tyk2 V678F were transiently expressed in 293T cells, and their level of phosphorylation on tyrosine was compared. Cell lysates were analyzed by Western blotting using two phospho-specific Abs, one directed to the activation loop tyrosines (Tyr 1054 and Tyr 1055 ) and the other directed to Tyr 292 . This latter residue is located in a stretch of 37 residues unique to the Tyk2 FERM domain and is a phosphorylation target identified in phosphoproteomic analyses (18,19). Using either Ab, Tyk2 V678F was found to be considerably more phosphorylated than the WT protein, despite comparable expression level (Fig. 1A). Similar data were obtained when the two proteins were transiently Tyk2 Val 678 in Heterodimeric and Homodimeric Receptors JULY 4, 2008 • VOLUME 283 • NUMBER 27 expressed in Tyk2-deficient 11,1 cells (Fig. 1B). To exclude a possible role for endogenous Jak1 in the basal phosphorylation of Tyk2 V678F, Jak1-deficient U4C cells were also transfected with the two constructs. Similarly, Tyk2 V678F was found to be hyperphosphorylated (Fig. 1B). These results show that the V678F mutation in the KL domain alters the phosphorylation state of the enzyme.
In protein kinases, the phosphorylation state of the activation loop is a critical, though not absolute, determinant of catalytic potential. Therefore, we asked whether the hyperphosphorylation of Tyk2 V678F on the loop correlated with an augmented intrinsic autophosphorylation activity. For this and further studies, we turned to stable neo R transfectants derived from 11,1 cells. We chose clones expressing the WT and V678F at comparable levels approaching the physiological level of Tyk2 in parental 2fTGH cells. To measure the intrinsic autophosphorylation ability of WT and mutant proteins in vitro, Tyk2 was immunoprecipitated from two pairs of clones and incubated in the presence or absence of cold ATP. The reaction products were analyzed by blotting with the 4G10 mAb, and bands were quantified. As shown in Fig. 1C, the autophosphorylation activity of mutant V678F was consistently higher than that of the WT protein. Together, these results show that the V678F mutation augments Tyk2 phosphorylation on the activation loop as well as the autocatalytic potential of the enzyme.
Tyk2 V678F Augments Stat3 Phosphorylation but Not the Canonical IFN␣-induced Signaling-We previously reported that Tyk2 sustains the level of IFNAR1, one of the two subunits of the type I IFN receptor (20). Based on this, we analyzed the content of IFNAR1 in cells stably expressing Tyk2WT (WT5 cells) or Tyk2 V678F (VF5 cells). As can be seen in Fig. 2 (bottom panels), IFNAR1 levels were similarly reconstituted, indicating that the V678F mutation in the KL domain does not impair the ability of Tyk2 to interact with the receptor. As anticipated The next day, cell lysates were analyzed in parallel by Western blotting with an Ab specific to phospho-Tyr 1054 /Tyr 1055 in the activation loop, an Ab specific to phospho-Tyr 292 , and an anti-Tyk2 mAb. Note that the endogenous Tyk2 band in vector-transfected cells is barely detected. B, Tyk2-deficient 11,1 cells and Jak1-deficient U4C cells were transiently transfected with either Tyk2WT or Tyk2 V678F and analyzed as described in A. One membrane was analyzed with Abs specific to phospho-Tyr 1054 /Tyr 1055 , stripped, and reprobed with anti-Tyk2 mAb. Another membrane was revealed with Abs specific to phospho-Tyr 292 . C, Tyk2 V678F phosphorylation in vitro is shown. WT5 and WT9 or VF5 and VF6 neo R clones were derived from 11,1 cells and express Tyk2WT or Tyk2 V678F, respectively. Tyk2 was immunoprecipitated (IP) from the indicated clone and subjected to an in vitro kinase reaction in the presence or absence of 50 M ATP for 5 min at 30°C. The phosphorylation level was analyzed by blotting the reaction products with anti-phosphotyrosine mAb (p-Tyr). The membrane was stripped and reprobed for total Tyk2. Results from five independent assays were quantified as follows. The amount of phosphorylated Tyk2 was normalized for total Tyk2 protein. The value of the reaction without ATP was subtracted from the value of the reaction with ATP. For each experiment, the values for the V678F were normalized by the WT (-fold increase), and the mean of the fold increases obtained from five experiments is represented with S.D. The phosphorylation level of the indicated proteins was analyzed by Western blotting with phosphotyrosine-specific Abs. Note that the anti-phospho-Jak1 Ab used detects phosphorylated Jak1 (indicated by the arrow) and also phosphorylated Tyk2. The membranes were reprobed for total protein level. Total IFNAR1 levels are shown in the bottom panels.

Tyk2 Val 678 in Heterodimeric and Homodimeric Receptors
from the transient analyses described in Fig. 1A, stably expressed Tyk2 V678F, but not Tyk2WT, was basally phosphorylated on the activation loop (Fig. 2, top panel; compare lanes 1  and 5 or lanes 9 and 14). Moreover, a notable difference between WT5 and VF5 cells concerned Stat3. Indeed, in the absence of ligand, Stat3 was weakly but detectably phosphorylated only in VF5 (Fig. 2, Stat3-P; compare lanes 1 and 5 or lanes  9 and 14). Having shown that Tyk2 V678F is catalytically more active than the WT protein (Fig. 1C), we compared the kinetic and dose-response phosphorylation profiles in WT5 and VF5 cells stimulated with IFN␣. Jak1, Stat1, and Stat2 were similarly phosphorylated in the two clones, whereas Stat3 and Stat5 were somewhat more phosphorylated in VF5 cells (Fig. 2). In conclusion, these data demonstrate that Tyk2 V678F has gained the capacity to phosphorylate endogenous Stat3 in the absence of added cytokine. However, Tyk2 V678F rescues signaling through the heterodimeric IFN receptor to the same extent as the WT protein.
To investigate further the impact of Tyk2 V678F, we monitored the extent of induction of IFN-stimulated genes in WT5 and VF5 cells treated for different times with 500 pM IFN␣. Transcripts of SOCS3, a Stat3-dependent gene, IRF1, a Stat1dependent gene, and 6-16, an ISGF3-dependent gene, were quantified by real-time PCR (Fig. 3). The only consistent difference between the two clones was a 2-fold higher level of SOCS3 transcripts in VF5 cells treated with IFN␣ for 1 h. This result further supports the finding that Tyk2 V678F targets specifically Stat3.
Cells were also subject to a sensitive assay (survival in hypoxanthine/aminopterin/thymidine-or 6-thioguanine-containing medium), which monitors canonical (ISGF3-mediated) IFN␣ response in 2fTGH-derived mutants (21). Concordant with the above data, WT-and V678F-expressing cells showed comparable IFN sensitivity phenotypes in these media (data not shown).
Tyk2 V678F Has Increased Basal Activity when Associated to a Homodimeric Receptor-It has been proposed that, when Jak2 V617F is expressed at nearly physiological levels, transformation of Ba/F3 cells to cytokine-independent proliferation requires coexpression of a homodimeric receptor (22). Based on these results, we investigated whether the Tyk2 V678F mutant would have a stronger signaling output than the WT protein when it is anchored to a homodimeric-type receptor. To this end, we made use of 11,1 cells that express a chimeric receptor (EpoR/R1) containing the ectodomain of EpoR fused to the transmembrane and the intracellular regions of IFNAR1 (16). These cells (denoted EpoR/R1 cl.19) were stably reconstituted with either Tyk2WT or Tyk2 V678F, and several clones were chosen based on comparable level of Tyk2 (see scheme in Fig. 4A). Moreover, we confirmed that the chosen clones expressed the same level of EpoR/R1 receptor as the parental EpoR/R1 cl.19 cells (16). Fig. 4B shows a direct comparison of the level of phosphorylated Tyk2 in a panel of WT-or V678F-expressing clones that either lacked or expressed the EpoR/R1 chimera. In all of the clones analyzed, Tyk2WT was not phosphorylated, whereas the V678F mutant was hyperphosphorylated. Importantly, the extent of phosphorylation of the V678F mutant was consistently higher in cells expressing EpoR/R1. This result suggests that the mere presence of the homodimeric EpoR/R1 receptor boosts the already deregulated state of the mutant protein, whereas it does not affect the WT protein. The basal level of Stat3 phosphorylation in these clones was also analyzed (Fig. 4B). Irrespective of the presence of EpoR/R1, Stat3 was not phosphorylated in WTexpressing clones. On the other hand, Stat3 was phosphorylated in all V678F-expressing clones, and the presence of EpoR/R1 did not consistently enhance it.
The autophosphorylation activity of WT and mutant Tyk2 anchored to EpoR/R1 was tested in vitro using two EpoR/R1 clones (WT9 and VF8). As shown in Fig. 4C, the V678F mutant in VF8 cells exhibited Ͼ14-fold increased basal activity with respect to the WT protein in WT9 cells. Thus, the impact of the V678F point mutation on the basal autophosphorylation activity of the enzyme is higher when the mutant protein is placed in the context of a cognate homodimeric receptor (compare Figs. 1C and 4C).
Tyk2 V678F Confers Ligand Hypersensitivity When Associated to a Homodimeric Receptor-Having shown that the basal gain of function of Tyk2 V678F is greater when the mutated enzyme coexists with the homodimeric EpoR/R1 receptor, we investigated the consequences on Epo signaling. Basal and Epoinduced Jak/Stat phosphorylation profiles were monitored in several clones. As shown in Fig. 5, in the absence of stimulation Tyk2 and Stat3 were phosphorylated in V678F-expressing clones, but not in control WT clones. In response to Epo, neither Jak1 nor Stat2 was activated (data not shown). The absence of Jak1 activation was expected because this enzyme cannot associate to the cytosolic portion of the EpoR/R1 chimera. The absence of Stat2 activation is most likely due to the lack of suitable docking sites on the chimeric receptor (16,23). On the other hand, Tyk2 and, to a lesser extent, Stat1, Stat3, and Stat5 were phosphorylated in response to Epo. Importantly, the induced phosphorylation was more robust in clones expressing Tyk2 V678F than in clones expressing Tyk2WT, and this was particularly evident in the dose-response profile (Fig. 5B). Together, these results indicate that the presence of the homodimeric receptor does indeed amplify the signaling potential of the Tyk2 V678F mutant.

DISCUSSION
We report here on the functional characterization of the Tyk2 V678F mutant protein expressed at near-physiological levels in the most appropriate context of Tyk2-deficient cells. This point mutation in the Tyk2 KL domain has a clear effect on the functioning of the enzyme. On the one hand, the mutation does not affect overall protein stability nor perturbs the ability of Tyk2 to sustain its cognate receptor, IFNAR1. On the other hand, the mutation has a clear (auto)activating effect, operationally defined as an increased ligand-independent (basal) phosphorylation of critical tyrosines in the activation loop both in vivo and in vitro. A direct consequence of this gain of function is the increased basal phosphorylation of Tyr 292 , located in the FERM domain and conserved in Tyk2 orthologues but not in other Jak family members. Although the role of Tyr 292 remains elusive, we have confirmed previous evidence (18,19) of its phosphorylation in cells stimulated with IFN␣. Moreover, phospho-Tyr 292 results, at least in part, from autophosphorylation because it is undetected in the kinase-dead K930R mutant. 6 We found that, when placed in the context of a cognate homodimeric-type receptor, the V678F mutant has enhanced capacity to autophosphorylate basally, whereas wild-type Tyk2 remains unaffected. We believe that this is due to the propensity of EpoR/R1 and associated kinase to form dimers even in the absence of ligand, as was described for EpoR (3,24,25). We infer from these findings that the V678F mutation somehow enables ligand-independent phosphorylation of the activation loop and that this effect is magnified when two mutant enzymes are juxtaposed via homodimerizing receptors. However, 6 J. Ragimbeau and S. Pellegrini, unpublished observations. In bold are clones that express the EpoR/R1 chimera. Phosphorylation was analyzed by Western blotting with phosphotyrosinespecific Abs. The membrane was reprobed for total Tyk2 content. C, basal in vitro autophosphorylation activity of Tyk2WT and Tyk2 V678F placed in the context of the EpoR/R1 chimeric receptor. WT9 and VF8 clones were derived from EpoR/R1 cl.19 cells and express Tyk2WT and Tyk2 V678F, respectively. The autophosphorylation activity of immunoprecipitated (IP) Tyk2 was assayed as described in the legend to Fig. 1C.

FIGURE 5. Epo-induced Jak/Stat activation in clones expressing
Tyk2WT or Tyk2 V678F. A, two sets of EpoR/R1 clones expressing either Tyk2WT (WT11 and WT4) or V678F (VF23 and VF18) were stimulated with a single dose (2 units/ml) of Epo for the indicated time. Phosphorylation of Tyk2, Stat1, and Stat3 was analyzed by Western blotting with phospho-specific Abs. The membrane was stripped and analyzed for total Tyk2 content. Note that the phospho-Stat1-specific Abs used in these blots weakly detect unphosphorylated Stat1, which is the bottom band pointed to by the arrow. B, the EpoR/R1 clones shown in A were stimulated with increasing doses of Epo (units/ml) for 15 min and analyzed for phosphorylation of Tyk2, Stat1, Stat3, and Stat5 by Western blotting with phospho-specific Abs. The membrane was stripped and analyzed for total Stat3 content.

Tyk2 Val 678 in Heterodimeric and Homodimeric Receptors
despite a basal catalytic gain, the V678F mutant is still "activable" upon ligand binding in either receptor context. Thus, the mutant may exist in a partially activated conformation, which is intermediate between that of a resting and a fully activated enzyme.
What are the consequences of the Tyk2 V678F deregulation in terms of ligand-independent and ligand-dependent Jak/Stat signaling? An important finding is that the V678F mutant augments specifically the level of tyrosine phosphorylated Stat3 in a ligand-independent fashion and in cells that do not require cytokines for proliferation, thus ruling out the requirement of a selection step for activation of signaling. These results are concordant with previous findings obtained in 11,1 and Ba/F3 cells expressing Tyk2 V678F, TpoR, and Stat3 (14). In the cytokinedependent Ba/F3 cells, Tyk2 V678F also activated Stat5, whereas in the cells used here a modest increment of Stat5 phosphorylation was observed only in response to IFN. We suggest that in a cell context where Stat5 can drive proliferation, further selection may occur, which leads to constitutive Stat5 activation. Interestingly, we found that Tyk2 V678F leads to an increased Stat3 phosphorylation in IFN-stimulated cells, and this correlates with enhanced early induction of the SOCS3 gene. Conversely, Tyk2 V678F has no remarkable effect on the IFN-induced phosphorylation of Jak1, Stat1, and Stat2. Indeed, the mutant rescued IFN␣-induced signaling and Stat1/1-or Stat1/2-mediated gene induction to virtually the same extent as the WT protein. This result suggests that Jak1, which is juxtaposed to Tyk2 in the heterodimeric type I IFN receptor, prevails over or controls the deregulated Tyk2 mutant. This is in accordance with previous work that proposed a hierarchy of Jak function in the IFN receptor (17,26).
We do not know how the privileged interplay of Tyk2 V678F and Stat3 is achieved. One possibility is the direct docking of Stat3 onto hyperphosphorylated Tyk2 V678F, for instance via the phospho-Tyr 292 in the FERM domain or the C-terminal Tyr 1176 located in a Stat3 consensus binding site. Alternatively, Tyk2 V678F could basally phosphorylate a Stat3 docking site accessible on IFNAR1 and EpoR/R1.
The analyses of the EpoR/R1 clones showed that Epo-induced signaling was robustly enhanced in the V678F context. Indeed, Tyk2WT-expressing cells never achieved the level of Epo-induced Stat phosphorylation detected in V678F-expressing cells (Fig. 5). Such a remarkable effect may be consequent to juxtaposition of two partially activated V678F mutants anchored to the homodimeric receptor. Despite Epo-induced Stat1/3 activation, we could not measure substantial induction of SOCS3, IRF1, or CIS transcripts in these V678F clones (data not shown), and we believe that this is due to the rather weak level of phosphorylated Stats formed. Indeed, when compared in the same blot, this level is considerably lower than that observed in control cells treated with IFN (data not shown). Thus, as previously suggested in Ref. 27, it is likely that the overall architecture of the chimeric receptor complexed with Tyk2 may not allow efficient recruitment, phosphorylation, and/or release of the Stats.
Recently, several authors have debated on whether the transforming function of the Jak2 V617F mutant, monitored in the Ba/F3 cell system, requires anchoring to a homodimeric recep-tor and whether overexpression can bypass the need for a receptor scaffold (14,22,28). Although Tyk2 and Jak2 serve distinct functions by interacting with different cytokine receptors, they appear to share a high degree of similarity in their regulatory mode, as suggested from the present analyses of the Tyk2 V678F mutant. What can we learn from these Tyk2 studies? We have repeatedly described the tight physical and functional interdependence existing between Tyk2 and the shortlived IFNAR1 receptor chain. On the one hand, Tyk2 sustains IFNAR1 at the cell surface, preventing its endocytosis and degradation, and contributes to formation of high affinity binding sites for the low affinity IFN␣ ligand (20,29). Reciprocally, IFNAR1 anchors Tyk2 at the membrane, and in so doing it may influence its overall spatial configuration and "shape" its catalytic behavior. On this basis, one would conclude that the gainof-function phenotype of Tyk2 V678F is necessarily dependent on, and undissociable from, the non-dimerizing IFNAR1 scaffold. However, when a Jak is overexpressed, endogenous cognate receptors may become limiting, and alternative modes of "scaffolding" may bypass the receptor need. Upon overexpression, the receptor binding behavior of the Jak may be affected, low affinity scaffolding proteins other than receptors could intervene, or a high local concentration of Jaks may facilitate their dynamic contact. Recently, Pradhan et al. (30) drew attention to the possibility that high expression of a single chain of a heterodimeric cytokine receptor (notably IL-27 receptor-␣) could be involved in transformation of myeloid progenitor cells via ligand-independent activation of Jak2. If this is the case, a receptor chain is not inert but can "prime" the associated WT kinase and even more the Val-to-Phe mutant. This possibility leads to the question of what distinguishes IL-27 receptor-␣ from other single chain receptors. It is conceivable that the specific interaction between a receptor scaffold and the FERM domain of a cognate Jak ultimately determines the degree of flexibility of the kinase domains and the degree of substrate accessibility (31,32).
We have previously reported on a functional screen of Tyk2 randomly mutated in the KL domain (6). The screen was intended to identify substitutions in KL that positively or negatively disturbed Tyk2 function and the cellular response to IFN. Our inability to select gain-of-function Tyk2 mutants may now be interpreted knowing the properties of Tyk2 V678F, which, despite deregulated catalytic activity, did not alter IFNinduced signaling. The detailed analysis of loss-of-function KL mutants obtained from the screen led us to attribute to the KL domain two distinct functions: it keeps the enzyme in a low phosphorylated resting configuration, and it contributes to the shift to, and/or the maintenance of, the activated configuration (see the model in Fig. 5 of Ref. 6). Interestingly, two KL mutants (H669P and R856G) were basally phosphorylated on the activation loop, like the V678F is. However, unlike V678F, they were catalytically inactive. This difference strongly suggests that the Val 678 -to-Phe substitution loosens the negative control exerted intramolecularly by the KL on the TK domain without perturbing the ill-defined positive role exerted by the KL in the activated enzyme.
Crystallographic analyses are required to understand the structural basis of the interaction between the two kinase

Tyk2 Val 678 in Heterodimeric and Homodimeric Receptors
domains and the effect of the Val-to-Phe mutation. However, some attractive predictions emerged from the model by Lindauer et al. (33) of the relative (intramolecular) orientation of the JH2 and JH1 domains in Jak2. The model was inspired by the (intermolecular) dimerization possibilities of the kinase domain of the fibroblast growth factor receptor (34). The authors located Val 617 at the basis of a loop connecting two ␤ strands of the small lobe of the JH2 domain. In the proposed JH2-JH1 arrangement, Val 617 is at one of two interfaces between the two kinase domains, and by making contact with the JH1 domain it may prevent motion of the activation loop. The replacement of Val 617 by Phe may displace the KL domain, rendering the activation loop more accessible to phosphorylation.
Two nucleotide changes are required to generate the V678F substitution in Tyk2, making this mutant unlikely to be found in human pathologies (13). However, V678F represents the first activating mutation described in Tyk2 and serves as a study model to analyze signaling and potential pathogenic consequences of a natural hyperactive Tyk2 mutant. Having shown that Tyk2 V678F targets specifically Stat3, and given the well described implication of Stat3 in tumor formation and progression (35), our finding suggests that an activating mutation in Tyk2 could represent a susceptibility factor in tumor development (36 -38). Tyk2 has been implicated in physiological responses to a large subset of helical cytokines, through its ability to interact with the IL-10 receptor-2 and IL-12 receptor-␤1 chains. Thus, Tyk2 is a critical mediator of the IL-10 (IL-10, IL-22, IL-26), the IL-12 (IL-12, IL-23), and the IFN families of cytokines, which regulate multiple cell types of the innate and adaptive immune response (39 -41). How the deregulated Tyk2 V678F would behave in these heteromeric receptors and what would be the functional consequences in the development of immune responses remain interesting and open questions.