Identification of Amino Acid Residues Critical for the Src-homology 2 Domain-dependent Docking of Stat2 to the Interferon α Receptor*

The interaction between Src-homology 2 domains (SH2) domains and phosphorylated tyrosine residues serves a critical role in intracellular signaling. In addition to the phosphotyrosine, adjacent residues are critical mediators of the specificity of this interaction. Upon treatment of cells with interferon α (IFNα), the IFNaR1 subunit of the IFNα receptor becomes tyrosine phosphorylated at position 466. The region surrounding phosphorylated tyrosine 466 subsequently acts as a docking site for the SH2 domain of Stat2, facilitating phosphorylation of the latter and, thus, the transduction of the IFNα signal. In this report site-specific mutagenesis was employed to analyze the nature of the interaction between the SH2 domain of Stat2 and the region surrounding tyrosine 466 on IFNaR1. Mutation of the valine at the +1 position carboxyl-terminal to tyrosine 466 or of the serine at the +5 position inhibits the association of Stat2 with phosphorylated IFNaR1. Moreover, receptors mutated at either of these two positions act in a dominant manner to decrease IFNα signaling, as assayed by both Stat2 phosphorylation and expression of an IFNα-responsive reporter. The demonstration that these two residues are critical in mediating the interaction between Stat2 and IFNaR1 suggests that STAT proteins might utilize a structurally distinct subset of SH2 domains to mediate signal transduction from the cell surface to the nucleus.

The interaction between Src-homology 2 domains (SH2) domains and phosphorylated tyrosine residues serves a critical role in intracellular signaling. In addition to the phosphotyrosine, adjacent residues are critical mediators of the specificity of this interaction. Upon treatment of cells with interferon ␣ (IFN␣), the IFNaR1 subunit of the IFN␣ receptor becomes tyrosine phosphorylated at position 466. The region surrounding phosphorylated tyrosine 466 subsequently acts as a docking site for the SH2 domain of Stat2, facilitating phosphorylation of the latter and, thus, the transduction of the IFN␣ signal. In this report site-specific mutagenesis was employed to analyze the nature of the interaction between the SH2 domain of Stat2 and the region surrounding tyrosine 466 on IFNaR1. Mutation of the valine at the ؉1 position carboxyl-terminal to tyrosine 466 or of the serine at the ؉5 position inhibits the association of Stat2 with phosphorylated IFNaR1. Moreover, receptors mutated at either of these two positions act in a dominant manner to decrease IFN␣ signaling, as assayed by both Stat2 phosphorylation and expression of an IFN␣-responsive reporter. The demonstration that these two residues are critical in mediating the interaction between Stat2 and IFNaR1 suggests that STAT proteins might utilize a structurally distinct subset of SH2 domains to mediate signal transduction from the cell surface to the nucleus.
Interferon ␣ (IFN␣) 1 is a member of the cytokine superfamily of effector molecules. First identified as an anti-viral agent, IFN␣ has also been shown to affect cell growth by inducing the transcription of growth inhibitory genes (for a review, see Ref. 1), following interaction with widely expressed cell surface receptors. Two subunits of the IFN␣ receptor (designated IF-NaR1 and IFNaR2) have been identified (2)(3)(4)(5), both encoded by genes localized within a 400-kilobase region on chromosome 21 (6). As with other cytokine receptors, both subunits are rapidly phosphorylated on tyrosine following binding of the cognate ligand (7)(8)(9).
Ligand binding activates transcription via a signaling pathway mediated by members of the Janus kinase and signal transducers and activators of transcription (JAK and STAT) families of proteins. The JAK family of tyrosine kinases includes four mammalian members. Genetic evidence has demonstrated that two of these, Tyk2 and Jak1, are involved in IFN␣ signal transduction (10,11). Each of these JAK kinases has been shown to constitutively bind to a subunit of the IFN␣ receptor; Tyk2 associates with IFNaR1 (12)(13)(14) and Jak1 with IFNaR2 (15). In vitro, all of the JAK kinases can phosphorylate IFNaR1, predominantly on tyrosine 466 (Tyr-466) (13,16). Homodimerization of a chimeric protein composed of the CD4 extracellular domain fused to the IFNaR1 intracellular domain resulted in its tyrosine phosphorylation. This phosphorylation was, again, found to occur primarily on Tyr-466 and to be dependent on the association of IFNaR1 with Tyk2 (17). The pattern of phosphorylation of IFNaR1 under physiologic conditions, though, has not been determined.
The STAT proteins are a family of latent cytoplasmic transcription factors employed in signaling pathways activated by various cytokines and growth factors (for a review, see Ref. 18). STAT proteins share a common domain structure; each has a DNA-binding domain, a transactivation domain, a Src-homology 3 domain, a Src-homology 2 (SH2) domain for binding phosphotyrosine, and a single tyrosine phosphorylation site near the C terminus. Complementation of mutant cell lines nonresponsive to IFN␣ revealed that two STAT family members, Stat1 and Stat2, are involved in the IFN␣ signal transduction pathway (19,20). After IFN␣ stimulation, Stat2 docks to IFNaR1 via an interaction between the Stat2 SH2 domain and the phosphorylated Tyr-466 (Tyr(P)-466) of IFNaR1 (16,17). The critical role of Tyr(P)-466 and the Stat2 SH2 domain in this docking interaction has been demonstrated by in vitro binding studies (16). In addition, phosphopeptides corresponding to the region surrounding Tyr-466 and dominant-negative IFNaR1 constructs in which Tyr-466 has been mutated to phenylalanine inhibit Stat2 phosphorylation in vivo (16). Phosphotyrosine dependent recruitment by IFNaR1 positions Stat2 to become tyrosine phosphorylated, presumably by one of the associated JAK kinases. Phosphorylated Stat2, in turn, provides a docking site for the SH2 domain of Stat1, positioning it for tyrosine phosphorylation (16,19). The two STAT molecules heterodimerize via SH2-phosphotyrosine interactions, translocate to the nucleus, and effect transcription of genes under control of interferon-stimulated gene response elements (ISREs) (21)(22)(23).
SH2 domains were first identified as conserved noncatalytic domains in the Src and Fps proteins (24). These domains were subsequently found in many proteins and shown to mediate interactions between cytoplasmic proteins by binding to phosphorylated tyrosine (25). Site-specific mutations in the residues surrounding tyrosines that function as SH2 domain-docking sites (summarized in Ref. 26) and subsequent studies that identified ligands for recombinant SH2 domains from degenerate phosphopeptide libraries indicated that the residue immediately C-terminal and the residue three amino acids C-terminal to the phosphorylated tyrosine (the ϩ1 and ϩ3 positions with respect to the tyrosine), were of critical importance in binding and conference of specificity for many SH2 domains (27,28). In this report, we examine the role of the residues C-terminal to Tyr-466 of IFNaR1 in the docking of Stat2 and subsequent signaling events and demonstrate that the amino acids at positions ϩ1 and ϩ5 relative to Tyr-466 are particularly critical for these molecular events.

EXPERIMENTAL PROCEDURES
DNA Constructs-All of the constructs were expressed from the vector pMT2T (29). Plasmid constructs encoding the wild type fulllength IFNaR1, the Y466F mutant of full-length IFNaR1, and the wild type CD4-IFNaR1 chimera have been previously described (16,17). A series of five mutant constructs, which convert residues 467 through 471 to alanine were generated by a single step polymerase chain reaction approach, using a 5Ј primer containing the appropriate mutation and a wild type 3Ј primer. Mutant polymerase chain reaction products spanning positions 1462-1617 of the IFNaR1 cDNA (5) were first cloned into the pGEM-T vector (Promega) and sequenced to ensure fidelity. NsiI-MfeI restriction fragments, containing the various mutations, were transferred into versions of the receptors cloned in Bluescript (Stratagene). Subsequently, PstI or EcoRI fragments containing the entire mutant full-length or chimeric receptor, respectively, were transferred into the pMT2T expression vector. The ␤-galactosidase expression plasmid (pCH110) used to control for transfection efficiency, and the ISRE-luciferase reporter plasmid (pZ-ISRE-luc; from R. Pine, Public Health Research Institute, New York, NY), which contains four tandem copies of the ISRE, have been previously described (30,31).
Cell Culture and Transfection-The human embryonic kidney 293T cell line and the human osteogenic sarcoma U2OS cell line were maintained in Dulbecco's modified Eagle's medium containing 10% fetal calf serum (Atlanta Biologicals, Atlanta, GA). Transfections were performed using calcium phosphate as described previously (16).
Antibodies, Immunoprecipitation, and Immunoblotting-The monoclonal anti-CD4 antibody used for cross-linking (Leu3a) was obtained from Becton-Dickinson Immunochemistry. Rabbit polyclonal anti-CD4 antisera used for immunoblotting (32) was from R.W. Sweet (Smith Kline-Beecham Pharmaceuticals). The rabbit polyclonal anti-Stat2 antisera, used for immunoprecipitation and immunoblotting (22), was from C. Schindler (Columbia University, New York, NY). The anti-Stat2 monoclonal antibody used in the co-immunoprecipitation experiments was obtained from Transduction Laboratories (Lexington, KY). The polyclonal anti-Tyk2 antisera used for immunoprecipitation and immunoblotting has been previously described (8). The monoclonal anti-phosphotyrosine antibody, 4G10, was obtained from Upstate Biotechnology Inc. (Lake Placid, NY). Preparation of lysates, immunoprecipitation, electrophoresis, and immunoblotting were performed as described previously (16,17).
Luciferase Assays-For each individual assay, duplicate subconfluent 10-cm dishes of U2OS cells were transfected with 5 g of the ␤-galactosidase expression plasmid, 5 g of the ISRE-luciferase plasmid, and 10 g of the expression vector. 30 h after transfection, one dish was treated with 3000 units/ml of IFN␣2 (from M. Brunda; Hoffman-La Roche, Nutley, NJ) for 18 h at 37°C, and the other dish was left untreated. Cells were then washed twice with cold phosphate-buffered saline, lysed, and rapidly frozen on dry ice. Lysates were thawed, and cellular debris was pelleted by centrifugation at 10,000 ϫ g for 5 min at room temperature. Luciferase activity was measured as per the manufacturer's protocol (Promega) and ␤-galactosidase activity was measured as described (30).
Statistical Analysis-Data from the luciferase assays was analyzed using an unpaired t test to compare wild type transfectants with each of the mutants.

RESULTS
To study the contribution of various amino acid residues in the SH2-mediated docking of Stat2 to the IFNaR1 receptor, we took advantage of a chimeric receptor system that reconstitutes the initial events in IFN␣ signaling, including kinase activation, receptor phosphorylation, and STAT docking (17). More distal events, such as Stat2 tyrosine phosphorylation, are not observed in this chimeric system, presumably because additional receptor subunit(s) (33) are required. Cells transfected with chimeric molecules composed of the CD4 extracellular domain fused to the intracellular domain of IFNaR1 (CD4-IFNaR1) are tyrosine phosphorylated, predominantly at Tyr-466 of IFNaR1, following cross-linking with an anti-CD4 antibody (17). Stat2 is recruited to the phosphorylated chimera, as evidenced by the co-immunoprecipitation of the chimeric receptor by anti-Stat2 antibodies (17). Co-immunoprecipitation was demonstrated to be specific for Stat2 and dependent on the phosphorylation of Tyr-466. In agreement with the in vitro and in vivo data cited above (16), this association supports the idea that Stat2 is recruited to phosphorylated Tyr-466 on IFNaR1. Furthermore, because more distal signaling events do not occur in the chimeric system, the Stat2-IFNaR1 association is unusually stable and can be much more readily detected than the equivalent interaction occurring under physiologic conditions.
We therefore reasoned that any mutation altering a site required for SH2-receptor interaction will inhibit the co-immunoprecipitation of Stat2 and the CD4-IFNaR1 chimera. Five mutant CD4-IFNaR1 chimeric molecules were constructed, substituting alanine for each of the five amino acids immediately C-terminal to Tyr-466 (amino acids 467 through 471, corresponding to the sequence VFFPS). To ensure that the mutations do not effect induced tyrosine phosphorylation, we first performed cross-linking experiments with the mutants and measured chimeric receptor phosphorylation. As the top panel of Fig. 1 shows, anti-CD4 treatment of 293T cells transfected with either the wild type chimera or any of the five mutant constructs results in a strong induced tyrosine phosphorylation of the chimeric receptor. Stripping the anti-phosphotyrosine immunoblot and reprobing with a polyclonal anti-CD4 antibody (Fig. 1, bottom panel) showed that each of the chimeric constructs was expressed, indicating that these single amino acid substitutions do not dramatically alter protein stability. In addition, fluorescence-activated cell sorting analysis with an anti-CD4 antibody showed that the each of the various chimeric receptors was expressed on the cell surface in approximately equivalent levels (data not shown). It is important to note that, although we have observed differences in the level of expression of the various chimeric constructs, the extent of tyrosine phosphorylation is similar in all cases. IFNaR1-Stat2 docking is completely dependent on tyrosine phosphorylation of the chimeric receptor. Thus, this is the most relevant control criteria for comparing the constructs in the recruitment assay described below.
Having demonstrated that mutation of residues adjacent to Tyr-466 does not effect IFNaR1 phosphorylation, we sought to determine which, if any, of these residues influence the association of the Stat2 SH2 domain with the phosphotyrosine on IFNaR1. To test this, wild type and mutant CD4-IFNaR1 chimeras were transfected into 293T cells and cross-linked. Lysates were immunoprecipitated with an anti-Stat2 monoclonal antibody and immunoblotted with an anti-CD4 antisera. As shown in Fig. 2, co-immunoprecipitation and therefore association of CD4-IFNaR1 with Stat2 was detected at similar levels in cells transfected with the wild type construct (lane 1), as well as constructs expressing the mutants F468A (lane 5), F469A (lane 7), and P470A (lane 9). On the other hand, receptors encoding two mutant constructs, V467A (lane 3) and S471A (lane 11), were co-immunoprecipitated significantly less efficiently than the wild type. Thus, three residues (ϩ2, ϩ3, and ϩ4 with respect to Tyr-466) have little effect on the binding of the Stat2 SH2 domain to IFNaR1, whereas two others (ϩ1 and ϩ5) appear to be important for this association.
To demonstrate the importance of the ϩ1 and ϩ5 residues in a more physiologic setting, we employed an IFN␣-driven reporter gene system. As noted in the Introduction, docking of Stat2 to the phosphorylated IFNaR1 is believed to position Stat2 so that it may be tyrosine phosphorylated by one of the JAK kinases. Phosphorylated STAT molecules dimerize, translocate to the nucleus and stimulate transcription of genes located downstream of an ISRE. Overexpression of a mutant receptor proficient in IFN␣ binding but deficient in Stat2 docking, we reasoned, would have a dominant inhibitory effect on IFN␣ signaling.
Adenovirus-transformed 293T cells cannot be employed in reporter gene assays because the E1A gene product binds p300/ cAMP response element-binding protein and prevents Stat2mediated transactivation (34). Therefore, we transfected U2OS cells with a plasmid containing the firefly luciferase gene downstream of a minimal promoter; upstream of this gene are four tandem-arranged ISRE enhancer elements. Constructs overexpressing either wild type or mutant forms of the full-length IFNaR1 receptor protein, as well as a lacZ expressing construct to control for transfection efficiency, were co-transfected with this reporter gene construct. The extent of dominant negative inhibition is quantified by measuring activity of the reporter gene in the presence and absence of IFN␣. In cells transfected with either wild type IFNaR1 (Fig. 3, first bar from the left) or an empty vector (data not shown), IFN␣ treatment produces an approximately 10-fold increase in luciferase activity. Overexpression of IFNaR1-Y466F, which cannot be phosphorylated at the Stat2 docking site, inhibits IFN␣-induced luciferase activ- were transfected with either the wild type CD4-IFNaR1 chimera (WT) or one of the five constructs containing the indicated mutations. Two days later, transfectants were split and one-half was left untreated (odd numbered lanes), whereas the other half was treated with an anti-CD4 antibody (Leu3a) to cross-link the chimeric receptors (even numbered lanes). Lysates were immunoprecipitated with an anti-CD4 antibody (OKT4) and immunoblotted with an anti-phosphotyrosine antibody (upper panel). The position of phosphorylated CD4-IFNaR1 is indicated. Slower migrating bands in lanes 3 and 5 represent an artifactual cross-reactive species that has been observed previously (17). Blots were stripped and reprobed with an anti-CD4 polyclonal antibody (lower panel). The position of the receptor protein is indicated.
FIG. 2. Mutations in amino acids near the major site of tyrosine phosphorylation of IFNaR1 affect co-immunoprecipitation with Stat2. Cells (293T) were transfected with either the wild type CD4-IFNaR1 chimera (WT) or one of the five constructs containing the indicated mutations. Two days later, transfectants were split and one-half was treated with an anti-CD4 antibody (Leu3a) to cross-link the chimeric receptors (odd numbered lanes), whereas the other half was left untreated (even numbered lanes). Lysates were immunoprecipitated with an anti-Stat2 monoclonal antibody, electrophoresed on a nonreducing gel, and immunoblotted with an anti-CD4 antibody to assay co-immunoprecipitation of the chimeric molecules with Stat2 (upper panel). The position of the chimera is indicated. The filter was stripped and reprobed with a polyclonal anti-Stat2 antibody to control for the efficiency of immunoprecipitation (lower panel). The position of the Stat2 protein is indicated. The intense band seen in some lanes of the lower panel, which migrates more slowly than Stat2, is nonreduced immunoglobulin. ity 70% (Fig. 3, seventh bar). In cells transfected with the construct containing IFNaR1 mutated at the ϩ3 site with respect to the tyrosine (F469A), an increase in luciferase activity similar to that seen in cells transfected with the wild type is observed following treatment (Fig. 3, fourth bar). In contrast, constructs mutated at either the ϩ1 (V467A, second bar) or ϩ5 (S471A, sixth bar) positions produced an unequivocal dominant negative effect on induced luciferase activity, though less pronounced than the Y466F mutant construct. Finally, mutations at the ϩ2 and ϩ4 positions (F468A and P470A, respectively), which did not appreciably effect the ability of Stat2 to coimmunoprecipitate the receptor (Fig. 2), produced intermediate effects on reporter gene activity when overexpressed in U2OS cells.
To determine if the dominant inhibitory effect of either the ϩ2 or ϩ4 mutant is significant, an unpaired t test was used to compare the induction of reporter gene activity observed in cells transfected with wild type IFNaR1 and the induction observed in cells transfected with each of the mutants. As expected from inspection of Fig. 3, a significant difference in induced luciferase activity was observed between the wild type and cells overexpressing either the ϩ1 or ϩ5 mutants (p Ͻ 0.0005), whereas the induction of reporter gene activity in cells overexpressing the ϩ3 mutant was not different from the wild type (p Ͼ 0.05). Interestingly, the induction of luciferase activity in cells overexpressing the ϩ4 mutant was found to be significantly different from that in wild type IFNaR1 transfected cells (p Ͻ 0.005), indicating some inhibitory effect of this mutation. Statistical analysis of the data for the ϩ2 mutant indicates no difference from wild type (p Ͼ 0.05). Thus, the reporter gene assay and the co-immunoprecipitation data both suggest that residues at the ϩ1 and ϩ5 positions are critically involved in the interaction between the Tyr-466 docking site and the Stat2 SH2 domain. Mutation at the ϩ4 position may have a smaller effect on signaling that is not revealed in the co-immunoprecipitation experiment.
Finally, to elucidate the molecular nature of the inhibition observed in cells overexpressing IFNaR1 mutants, we examined the phosphorylation state of signaling cascade components following IFN␣ treatment. Wild type and mutant IFNaR1 constructs were overexpressed in 293T cells. These cells were used because they are IFN␣ responsive and can be transfected at the very high levels required to permit the proteins produced by the transfected genes to displace endogenous molecules involved in signaling interactions (17). Tyrosine phosphorylation of signaling components was assayed by immunoblotting anti-Tyk2 and anti-Stat2 immunoprecipitates with an anti-phosphotyrosine monoclonal antibody (4G10). In cells transfected with either wild type IFNaR1 (Fig. 4A, lane 1) or any of the five IFNaR1 constructs mutated near Tyr-466 (Fig. 4A, lanes 3, 5, 7 , 9, and 11), IFN␣ treatment induces the tyrosine phosphorylation of Tyk2. On the other hand, when Stat2 phosphorylation was assayed, differences were observed between the various mutants. Stat2 phosphorylation in cells transfected with the mutants F468A, F469A, and P470A (Fig. 4B, lanes 5, 7, and 9, respectively) occurs at levels similar to that of wild type IF-NaR1 (Fig. 4B, lane 1). On the other hand, cells transfected with mutants V467A and S471A (Fig. 4B, lanes 3 and 11, respectively) showed a decreased induction of Stat2 tyrosine phosphorylation, relative to the wild type. Tyrosine phosphorylation of the Stat1 protein, which co-immunoprecipitates with Stat2, is also diminished.
As shown previously (16), overexpression of the Y466F mutant of IFNaR1 has a stronger effect on IFN␣ induced Stat2 and Stat1 tyrosine phosphorylation (Fig. 4C). Thus, consistent with the idea that SH2-dependent docking on IFNaR1 positions Stat2 for tyrosine phosphorylation, mutations at the ϩ1 and ϩ5 positions show a dominant inhibitory effect on Stat2 activation, independent of Tyk2 activation. Furthermore, as in Fig. 3, IFNaR1 constructs containing mutations in the carboxyl-terminal flanking amino acids produce a more modest dominant negative effect compared with that observed with a construct in which the critical tyrosine has been mutated to phenylalanine. DISCUSSION We have previously demonstrated that following IFN␣ treatment, Stat2 is specifically recruited to Tyr(P)-466 on the IF-NaR1 subunit of the receptor (16,17). SH2-mediated recruitment of STATs to phosphorylated cytokine receptors is believed to be the mechanism by which STAT specificity is determined (35). To confirm that the region surrounding Tyr-466 is required for Stat2 recruitment, and to begin to characterize the residues that confer specificity for Stat2 binding, we systematically mutated the five amino acids C-terminal to Tyr-466 on the IFNaR1 receptor subunit. We employed two experimental approaches previously used to characterize the IFN␣-signaling Lysates were prepared and analyzed for luciferase and ␤-galactosidase activities. Plots represent fold increase in IFN␣-dependent luciferase activity relative to untreated cells, normalized for ␤-galactosidase activity. Error bars represent one standard error. Numbers above the bars represent the number of replicate experiments for each particular construct. Asterisks over bars indicate those transfectants displaying a statistically significant inhibition in induced luciferase activity relative to the wild type construct (see text for details). pathway: (i) inducible activation of a chimeric receptor by anti-CD4 antibodies (17), and (ii) overexpression of dominant inhibitory versions of IFNaR1 (16). Our results indicate that residues at the ϩ1 and ϩ5 positions C-terminal to Tyr-466 are critically important in mediating the binding of Stat2 to Tyr(P)-466. Specifically, mutation of either of these residues to alanine did not effect the anti-CD4 induced tyrosine phosphorylation of the chimeric receptor ( Fig. 1) but did drastically disrupt the ability of the receptor to co-immunoprecipitate Stat2 (Fig. 2). Second, full-length receptors with alanine at either position function in a dominant inhibitory fashion to suppress both ISRE-dependent reporter gene activity (Fig. 3) and Stat2 tyrosine phosphorylation (Fig. 4B) in response to IFN␣. Significantly, overexpression of constructs with these same mutations did not suppress Tyk2 tyrosine phosphorylation (Fig. 4A), consistent with the idea that the effect we have observed is occurring at the level of Stat2 docking to the IFNaR1 subunit and not in earlier steps in the signaling cascade, such as kinase activation.
Stark, Kerr, and colleagues (36) have produced mutant cell lines that are deficient in various components of the IFN␣ signaling pathway. Complementation of these mutant lines proved invaluable in establishing the role of these components under physiologic conditions. Unfortunately, no IFNaR1-deficient mutants have been identified, preventing us from performing similar experiments. In addition, the murine IFNaR1 receptor does not contain an obvious region of homology to the Tyr-466 docking site on the human version, nor has a functionally equivalent site been identified (37) (it is unclear how Stat2 is recruited to the murine receptor). Thus, complementation experiments, which might confirm our results, await the development of human cell lines that are null at the IFNaR1 locus.
Although residues carboxyl-terminal to the tyrosine contribute to the specificity of an SH2-phosphotyrosine interaction, the strongest binding occurs between the SH2 domain and the phosphotyrosine itself, which fits into a deep pocket containing a positively charged arginine residue (38). Thus, Tyr(P)-466 is the main determinant of affinity for Stat2 when it binds the docking site on IFNaR1. As such, mutation of residues carboxyl-terminal to the phosphorylated tyrosine will likely weaken, but not entirely abrogate Stat2 binding. This is consistent with our observation of some residual co-immunoprecipitation for the ϩ1 and ϩ5 mutant constructs in Fig. 2. Other explanations, such as low level binding of Stat2 to the region surrounding Tyr-481 (16) or an incomplete dominant negative effect because of technical limitations might also account, at least in Stat2 is a nonspecific cross-reactive artifact that has been noted previously (16). The filter was stripped and reprobed with an anti-Stat2 antibody (lower panel). Panel C, a separate experiment was performed in the same manner as panel B, except that the wild type and Y466F mutant IFNaR1 constructs were employed. part, for the residual co-immunoprecipitation. The secondary role of the carboxyl-terminal residues is also well demonstrated in the reporter gene assay, where the Y466F mutant is clearly a stronger inhibitor than either the ϩ1 or ϩ5 mutants (Fig. 3), as well as the dominant inhibition of Stat2 tyrosine phosphorylation (Fig. 4).
The proline residue at the ϩ4 position may also contribute to the Stat2 SH2 domain binding site on IFNaR1. This is evidenced by the modest, but statistically significant dominant inhibitory effect of an alanine mutation at this position on the ISRE-driven reporter gene assay (Fig. 3). In contrast, two biochemical measures of the interaction, Stat2-IFNaR1 co-immunoprecipitation (Fig. 2) and Stat2 tyrosine phosphorylation (Fig. 4), showed little or no difference for the ϩ4 alanine mutation when compared with the wild type controls. One possible explanation for the difference between the reporter assay and the biochemical assays may lie in the increased sensitivity of reporter gene activation to small alterations in phenotype. Another possibility is that the biochemical measurements are made at a single point in time and do not necessarily reflect the kinetics of signaling, whereas reporter gene activation measures the accumulated activation of the signaling pathway.
It should be noted that our results are based entirely on the use of alanine substitution to identify residues that are critical for the Stat2-IFNaR1 interaction. Although alanine is widely employed as a substitute amino acid because its methyl group side chain is unlikely to participate in most hydrophobic or hydrophilic interactions, we cannot rule out the possibility that mutation of the residues in question to amino acids other than alanine might affect the interaction. It is of interest to note that experiments aimed at identifying a consensus sequence for Stat2 SH2 domain binding partners, using a partially degenerate library of phosphopeptides, were not successful because of technical limitations (28). Thus, additional data that might guide the choice of other substitutions is lacking at present.
Co-crystallization of the Src SH2 domain with a phosphotyrosine containing peptide ligand corresponding to a known high affinity binding site revealed not only a conserved binding pocket for phosphotyrosine but also a set of interactions between the SH2 domain and residues of the peptide carboxylterminal to the tyrosine (39). The amino acid immediately carboxyl-terminal to the tyrosine (the ϩ1 residue) was found to make direct contact with the side chains of two amino acids (␤D3 and ␤D5) within the ␤D-strand of the central ␤-sheet conserved in SH2 domains. SH2 domains with glutamine at the ␤D3 position preferentially bind peptides with valine, leucine, or isoleucine at the ϩ1 position (27). Stat2 contains a glutamine at the position corresponding to ␤D3 (40) and binds to a phosphotyrosine motif containing a valine at the ϩ1 position. As shown in the Fig. 2, mutation of this valine to alanine substantially reduces the association between Stat2 and phosphorylated IFNaR1. The Stat2 SH2 domain also interacts with a phosphorylated tyrosine on the Stat1 protein (Tyr-701) when the Stat1-Stat2 heterodimer is formed (21,41). Interestingly, the ϩ1 position with respect to Tyr-701 of Stat1 is an isoleucine, one of the amino acids predicted to interact with a ␤D3 glutamine (27). Thus, our finding that Val-467 on IFNaR1 is important for Stat2 SH2 binding fits with one of the predicted structural features of the Stat2 SH2 domain.
X-ray crystallographic data also revealed that the SH2 domains of both Src (38,39) and Lck (42) fold to create a hydrophobic pocket into which the side chain of the ϩ3 amino acid fits. Other SH2 domains were subsequently found to make direct contacts with the ϩ3 position. However, mutation of the ϩ3 site in IFNaR1 (Phe-469) to alanine does not effect association of the receptor with Stat2 (Fig. 2), nor does it effect downstream signaling (Figs. 3 and 4). In contrast, our data suggest that the serine in the ϩ5 position is a likely site of contact between the Stat2 SH2 domain and IFNaR1. The involvement of a residue at the ϩ5 position in SH2-phosphotyrosine interaction is not without precedent. For example, x-ray crystallographic analysis of the N-terminal SH2 domain of Syp complexed with high affinity peptides revealed strong interaction between the SH2 domain and the ϩ5 residue of the phosphopeptide (43). This SH2 domain, however, also forms a hydrophobic channel that contacts the ϩ1 and ϩ3 residues, in a manner similar to Src. In fact, the side chain of the ϩ3 residue projects deep within the SH2 domain of Syp, providing extensive contacts and stability. Thus, the structure of the SH2 domain of Stat2 is unlikely to be analogous to that of the N-terminal SH2 domain of Syp.
The SH2 domain of Stat1 was also recently shown to display a distinctive binding specificity. In a manner similar to IFN␣, IFN␥ induces tyrosine phosphorylation of its receptor at a specific site, tyrosine 440, which is subsequently used as a docking site for the SH2 domain of Stat1 (44). Using changes in surface plasmon resonance as a measure of binding affinities, Greenlund et al. (45) showed that mutation of the ϩ1 and ϩ4 positions to alanine reduced the affinity of the Stat1 SH2 domain for the IFN␥ receptor. The effect of mutation of the ϩ5 position, however, was not tested in this system, and thus at this time it is premature to draw conclusions about the similarity of the binding specificities of these SH2 domains. Marengere et al. (46) classify SH2 domains into two groups -those that select hydrophobic residues at the ϩ1, ϩ2, and ϩ3 positions; and those that select hydrophilic residues at the ϩ1 and ϩ2 positions, and a hydrophobic amino acid at the ϩ3 position. The data for selectivity of both Stat1 and Stat2, though, indicate that they may be structurally distinct and define a new class or classes of SH2 domains.
In most cases, the role of the SH2 domain is to stably localize a protein to the inner surface of the cell membrane via an interaction with phosphotyrosine on either catalytic receptors or adaptor molecules. The Stat2 SH2 domain, on the other hand, must interact sequentially with at least two separate phosphotyrosines. First, it transiently binds Tyr(P)-466 on IF-NaR1. Subsequently, it uncouples from the receptor and dimerizes with Stat1, prior to translocating to the nucleus. Thus, functionally, the STAT SH2 domains are distinct from previously characterized SH2 domains. Although additional studies, including crystallographic analyses, will be required to confirm our hypothesis, the data presented in this report suggest that the SH2 domain of Stat2 may be structurally distinct from previously characterized SH2 domains.