Identification of a Novel Conserved Motif in the STAT Family That Is Required for Tyrosine Phosphorylation*

The rapid transcriptional activation of cellular genes by either type 1 interferons (IFNα/β) or type 2 interferon (IFNγ) is responsible for many of the pleiotropic effects of these cytokines, including their antiviral, antigrowth, and immunomodulatory activities. Interferon-stimulated gene expression is mediated by transcription factors termed Stats, which upon being tyrosine-phosphorylated, translocate to the nucleus and bind enhancers of interferon-activated genes. We have recently characterized a new Jurkat cell variant, named H123, where IFNα stimulates programmed cell death. H123 clones that are resistant to the apoptotic actions of IFNα have been selected. One of these clones (Clone 8) is defective in its responses to IFNα with regard to activation of genes that require tyrosine phosphorylation of Stat2. Stimulation of Clone 8 cells with IFNα induces normal tyrosine phosphorylation of Stat1 and Stat3. Sequencing of Stat2 RNA reveals a substitution of proline 630 located within the Src homology 2 domain of Stat2 to leucine (P630L). Pro-630 and its adjacent amino acids are conserved in all Stat family members but are absent in other proteins that contain Src homology 2 domains. Expression of Stat2 P630L in cells inhibits IFNα-stimulated gene expression. These results not only define a critical motif in Stat2 required for its transcriptional activity, but they also provide evidence that resistance to type one IFNs can be mediated by mutations in Stat2 as well as those previously described for Stat1.

Identification of components required for interferon-mediated expression of immediate early genes was greatly facilitated by selection of cells unresponsive to interferons (IFNs) 1 (1). Variants of the HT1080 fibrosarcoma cell line that do not respond to IFNs were isolated after multiple rounds of exposure to the frameshift mutagen ICR191 (2). These variants fail to express proteins encoding Jaks or Stats and have been used extensively to define important domains required for the structure and function of these proteins. Other domains known to be essential for the function of Jaks and Stats have been gleaned from the identification of mutant proteins that result in specific diseases. For instance, point mutations in Jak3 that render the kinase inactive were found in several patients with severe combined immunodeficiency (3,4). Many of these mutations occur in the N-terminal region of the protein, which aided in defining its "FERM" domain (4). Melanomas and malignant breast cancer cell lines have been reported to lack expression of Stat1, Stat2, and p48 (5). It is unclear whether there is a mutation in the gene or some other event that makes the expression of this transcription factor undetectable. A single point mutation in Stat1 (leucine 706 to serine) renders individuals susceptible to infection with mycobacteria (6). Leucine 706 has been shown to play a crucial role in Stat1 homodimerization (6). Furthermore, constitutively activated Stat3 and Stat5 have been observed in a number of hematopoietic malignancies, including T cell leukemias (7)(8)(9)(10). Mutations in Jaks or Stats associated with constitutive tyrosine phosphorylation of Stat3 and Stat5 have not yet been identified in these patients. However, it is known that in some patients there is constitutive tyrosine phosphorylation of Jak3 (10).
In cells expressing mutations in components of the Jak/Stat pathway, it remains unclear what if any selective pressure might be present to allow such mutant cells to expand. One mechanism by which mutations could arise is as a result of excessive amounts of a given cytokine or a constitutively activated cytokine receptor. This has been reported to occur in thanatophoric dysplasia type II dwarfism where Stat1 is constitutively activated due to the expression of an activated fibroblast growth factor receptor (11). Treatment of patients with type I IFNs is used therapeutically in a variety of malignancies. However, prolonged and repeated exposure to this cytokine often results in IFN resistance, suggesting that cell variants might be selected containing mutations in components of the Jak/Stat signaling pathway. To test this possibility we have taken advantage of a Jurkat cell variant (H123) obtained by chemical mutagenesis. Although parental Jurkat cells show no IFN␣-mediated apoptosis, H123 cells undergo massive death when exposed to this cytokine. 2 To determine what signaling events are required for IFN␣-stimulated apoptosis of H123 cells, we selected H123 cells that survive when exposed for extended periods to IFN␣. One IFN␣-resistant variant that we have identified from this screen (Clone 8) has a point mutation in the SH2 domain of Stat2. The characterization of this Stat2 variant not only provides us with important information concerning an uncharacterized domain in Stats, but it also demonstrates that mutations in cytokine signaling can be readily obtained by selective pressure of cytokine exposure as opposed to chemical mutagenesis. Selection of such variants can provide important information concerning domains that regulate the activity of these proteins.

MATERIALS AND METHODS
Cells and Cell Culture Reagents-H123-and Clone 8-derived Jurkat cells were maintained in RPMI 1640 medium supplemented with 10% fetal calf serum (Invitrogen) containing 2 mM L-glutamine, penicillin, and streptomycin (Invitrogen). The human fibrosarcoma 2fTGH mutant cell line deficient in Stat2 (U6A) (a gift from G. Stark, The Cleveland Clinic Foundation) and 293T cells were cultured in high glucose Dulbecco's modified Eagle's medium containing 10% fetal calf serum and antibiotics. Stable lines of U6A cells reconstituted with Stat2 or Stat2 P630L were generated by transfection with Superfect reagent. Cells were maintained in the presence of 450 g/ml G418. Recombinant human IFN␣-2a was a gift from Hoffmann-LaRoche.
RNase Protection Assays-RNase protection assays were performed as described previously (12,13). Briefly, total RNA was isolated with RNAzol B (Tel-Test Inc.). Antisense 32 P-labeled RNA probes were synthesized by in vitro transcription using T7 or SP6 RNA polymerase (New England Biolabs, Inc.). 10 g of RNA and 32 P-labeled probes were incubated in hybridization buffer (80% formamide, 40 mM PIPES (14), 400 mM NaCl, and 1 mM EDTA) overnight at 56°C followed by digestion with T1 RNase (Ambion) for 1 h at 37°C (15). After phenol:chloroform extraction and ethanol precipitation, protected RNA fragments were solubilized and subjected to electrophoresis on a 4.5% polyacrylamideurea gel.
RT-PCR Analysis and Generation of Stat2 P360L Construct-Total RNA from Clone 8 cells was prepared as indicated above and used to synthesize cDNA using the cDNA synthesis kit (Invitrogen). The entire Stat2 was PCR-amplified using the following primers 5Ј-ATGGCGCA-GTGGGAAATGCTGC-3Ј and 5Ј-CTAGAAGTCAGAAGGCATCAAGG-G-3Ј and cloned into PGEM-T-Easy vector (Promega) followed by DNA sequencing. The FLAG-tagged Stat2 construct in pcDNA3 (kindly provided by C. Horvath, Mount Sinai School of Medicine) was used as template DNA. Amino acid substitution of proline 630 to leucine was generated by PCR-based site-directed mutagenesis (Stratagene) following the manufacturer's instructions using the oligonucleotides 5Ј-CATCTACTCTGTGCAACTGTACACGAAGGAGGTGC-3Ј and 5Ј-GCACCTCCTTCGTGTACAGTTGCACAGAGTAGATG. Mutagenesis was confirmed by sequencing the entire Stat2 SH2 sequence.
Immunoprecipitation and Western Blot Analysis-Whole cell extracts were subjected to immunoprecipitation with antibodies against FLAG or Stat2 antibody for 2 h, followed by the addition of protein G-Sepharose beads (Amersham Biosciences) for another 2 h at 4°C. Immunoprecipitates were separated by SDS-PAGE and transferred to a polyvinylidene difluoride membrane. The membranes were probed with rabbit polyclonal antibody against STAT1, phospho-Stat1 1 (PY-701), phospho-Stat3 (Cell Signaling Technology), STAT2 (Santa Cruz Biotechnologies), phospho-STAT2 PY-690 (Upstate Biotechnology), or FLAG epitope (Sigma). Following incubation, the membranes were washed and incubated for 30 min with peroxidase-conjugated antirabbit IgG and subsequently developed by chemiluminescence using the ECL Western blotting system (Amersham Biosciences).
Immunofluorescent Staining-Following stimulation, cells were washed twice with PBS, centrifuged onto a microscope slide for 3 min at 400 rpm, fixed in 4% paraformaldehyde for 10 min at room temperature, and washed again in PBS. Cells were then permeabilized with 0.2% Triton X-100 in PBS for 5 min followed by incubation with primary antibody diluted in blocking solution (2% goat serum, 2 mg/ml bovine serum albumin in PBS) at 4°C overnight. The slides were washed with blocking buffer and then incubated for 1 h with fluorescein isothiocyanate-labeled secondary antibody (Alexis Biochemicals). After several washes with PBS, the slides were counterstained with 4Ј,6-diamidino-2-phenylindole Vectashield mounting medium (Vector Laboratories).
Transfections and Luciferase Assay-Cells were transfected using Superfect reagent (Qiagen) with 2 g of 5ϫ ISRE luciferase reporter plasmid in combination with wild type Stat2 or Stat2 P630L plasmid DNA. To normalize for luciferase activity and control for transfection efficiency, 0.5 g of pRL-TK (Promega) was also included. After overnight incubation at 37°C, cells were left untreated or stimulated with IFN␣ for 6 h. Cell lysates were prepared and luciferase activity was measured with a Luminometer (Dynatech Laboratories) using the dualluciferase reporter system according to the manufacturer (Promega).
Measurement of Apoptosis-Cells were left untreated or stimulated with 1,000 units/ml of IFN␣ for 48 h. Cells were then harvested, resuspended in binding buffer (10 mM HEPES, pH 7.4, 140 mM NaCl, and 2.5 mM CaCl 2 ), and incubated with annexin V-fluorescein isothiocyanate (Pharmingen) and propidium iodide (Sigma) for 10 min at room temperature. Cells were collected ungated (10,000 events) and analyzed by two-color flow cytometry to discriminate between apoptotic and necrotic cells.
Flow Cytometry-Cells were incubated for 30 min on ice in staining buffer (PBS solution containing 0.2% NaN 3 and 2% fetal calf serum) containing phosphatidylethanolamine-labeled anti-major histocompatibility complex class I or isotype-matched control antibodies. Surface expression of this marker was analyzed using a FACScan TM cytometer (BD Biosciences) and further analyzed using CellQuest software (BD Biosciences).

RESULTS
To isolate H123 cells that were resistant to the apoptotic actions of type 1 interferons, cells were incubated with 1,000 units/ml of IFN␣ for 4 days. To remove dead cells, the cultures were ficolled, and surviving cells were cultured again in the presence of IFN␣ for 7 days. This selection process was repeated a total of 6 times followed by expansion of cells in the absence of IFN␣. Single colonies were further tested to ensure that exposure to IFN␣ did not induce cell death (Fig. 1A). Although incubation of H123 cells with IFN␣ for 48 h caused ϳ45% of cells to be annexin V-positive, there was no change in the number of annexin V-positive cells in Clone 8 cells exposed to IFN␣. Clone 8 cells maintained their resistance to IFN␣induced apoptosis when cultured in the absence of the cytokine for months (data not shown). To determine whether Clone 8 cells showed any responsiveness to IFN␣, we examined surface expression of major histocompatibility complex class I protein (Fig. 1B). Incubation of either H123 or Clone 8 cells with IFN␣ resulted in similar enhanced expression levels of major histocompatibility complex class I, indicating that at least some of the responses to IFN␣ are intact in Clone 8 cells.
IFN␣-activated expression of IFN-stimulated genes (ISGs) requires expression and activation of Jak1 and Tyk2 that results in tyrosine phosphorylation of Stat1 and Stat2. To determine whether the amount of these proteins was altered in Clone 8 cells, cell extracts were prepared from H123 and Clone 8 cells, and Western blots were probed with antisera against Stat1, Stat2, Jak1, and Tyk2 ( Fig. 2A). Although concentrations of Stat1, Jak1, and Tyk2 were the same in H123 and Clone 8 cells, the amount of Stat2 in these cells was clearly diminished.
The fact that the concentrations of Stat2 were decreased in Clone 8 cells suggested that there could be a defect in IFN␣stimulated Stat2-dependent gene expression in these cells. Examination of IFN␣-stimulated Stat1 tyrosine phosphorylation by electrophoretic mobility shift assays with an oligonucleotide that binds tyrosine-phosphorylated Stat1 (GRR) showed approximately the same amount of activated Stat1 in H123 and Clone 8 cells (Fig. 2B, upper panel). However, IFN␣-stimulated formation of the heterotrimeric transcription complex ISGF3 as assayed by EMSA using an oligonucleotide (ISRE) that binds tyrosine-phosphorylated Stat1, Stat2, and IRF9 was not detectable in nuclear extracts prepared from IFN␣-treated Clone 8 cells (Fig. 2B, lower panel). These results suggest that RNAs whose expression is dependent upon IFN␣-stimulated formation of ISGF3 binding to an ISRE are absent in Clone 8 cells. This is indeed the case; IFN␣-stimulated expression of ISG54, ISG15, and 6 -16 RNAs is not observed in Clone 8 cells after various times of incubation with IFN␣ (Fig. 2C). As seen previously, IFN␣-induced expression of these RNAs is very robust in H123 cells (lanes 1-3) as well as in wild type Jurkat cells (data not shown).
To directly examine whether there are changes in IFN␣stimulated tyrosine phosphorylation of Stats in H123 compared with Clone 8 cells, cell extracts were prepared from untreated and IFN␣-treated cells. Immunoblots were probed with specific antiserum that recognizes only the tyrosine-phosphorylated forms of Stat1, -2, or -3 (Fig. 3A). Although IFN␣ treatment of both Clone 8 and H123 cells caused equal amounts of tyrosine-phosphorylated Stat1 and Stat3 (Fig. 3A), IFN␣- Previous studies have indicated that when cells are exposed to IFN␣, an association between Stat1 and Stat2 can be detected in immunoprecipitates using either Stat1 or Stat2 antisera (16). We are unable to detect any association between Stat1 and Stat2 in IFN␣-treated Clone 8 cells (Fig. 3B, lanes 3  and 4). However, the association of these proteins is easily detectable in H123 cells incubated with IFN␣ (lanes 1 and 2).
We also examined whether nuclear translocation of Stat2 occurs in IFN␣-treated Clone 8 cells. H123 and Clone 8 cells were left untreated or incubated for 30 min with IFN␣ prior to fixation and staining cells for Stat1 and Stat2 (Fig. 3C). As expected, IFN␣-stimulated nuclear accumulation of Stat1 was readily detectable in both H123 and Clone 8 cells. In H123 cells we could also see IFN␣-dependent nuclear translocation of Stat2 using antibodies against Stat2 or tyrosine-phosphorylated Stat2. In contrast, IFN␣-stimulated nuclear translocation of Stat2 in Clone 8 cells was dramatically reduced. In some experiments there appeared to be a very modest IFN␣-induced nuclear accumulation of tyrosine-phosphorylated Stat2, whereas in other experiments the staining was not above background.
Although the level of tyrosine-phosphorylated Stat2 is diminished in Clone 8 cells compared with H123 cells, Clone 8 cells also express significantly less Stat2 protein. We have not been able to detect any IFN␣-stimulated ISG expression in Clone 8 cells. This may be because of an insufficient amount of tyrosine-phosphorylated Stat2 in Clone 8 cells or because of tyrosine-phosphorylated Stat2 that is not functional. To distinguish between these two possibilities, we incubated both H123

FIG. 2. Clone 8 cells display a defect in IFN␣-stimulated ISGF3driven responses.
A, whole cell extracts from unstimulated cells were resolved by SDS-PAGE, and immunoblot analysis was performed with antibodies against Stat1, Stat2, Jak1, and Tyk2 proteins. Immunoblots were reprobed with actin antibody to control for equal loading of protein (lower panel). B, nuclear extracts from cells incubated with or without IFN␣ for 30 min were assayed for Stat1 homodimer or ISGF3 formation by EMSA with a GRR (top panel) or ISRE probe (bottom panel), respectively. C, H123 and Clone 8 Jurkat cells were left untreated or stimulated with IFN␣ for the indicated times. Total RNA was isolated and expression of ISG54, ISG56, and 6 -16 RNAs was determined by RNase protection assay. Actin RNA was used as an internal control to normalize for the equal amounts of RNA used in each hybridization.

FIG. 3. Clone 8 cells have impaired tyrosine phosphorylation and nuclear translocation of Stat2. H123 or Clone 8 Jurkat cells
were left untreated or stimulated with IFN␣ for 30 min. A, whole cell extracts were prepared, resolved by SDS-PAGE, and immunoblot analysis was performed with phospho-specific antibodies against Stat1, -2, and -3 antibodies. Membranes were reprobed with antibodies against Stat1, -2, and -3 to verify equal loading of protein. B, whole cell extracts were immunoprecipitated with anti-Stat1 antisera. The immunoprecipitates were separated by SDS-PAGE, and immunoblots were probed with anti-Stat2 (upper panel) and anti-phosphotyrosine Stat1 antibodies (middle panel). The membrane was reprobed with anti-Stat1 antibody to ensure equal levels of immunoprecipitated Stat1 (lower panel). C, subcellular distribution of Stat1 and Stat2 was analyzed by immunofluorescence microscopy. Cells were permeabilized and stained with anti-Stat1, anti-Stat2, or anti-phosphotyrosine Stat2 antibodies and counterstained with 4Ј,6-diamidino-2-phenylindole. and Clone 8 cells with different concentrations of IFN␣ and examined tyrosine phosphorylation of Stat2 after 30 min and ISG RNA expression after 6 h. These incubation times have been shown to be optimal for analysis of these two events (17). Incubation of cells with 10 units/ml of IFN␣ caused tyrosine phosphorylation of Stat2 in H123 cells, which was about the same amount as that seen in Clone 8 cells incubated with 100 units/ml (Fig. 4A). At 10 units/ml, IFN␣-stimulated expression of the 6 -16 RNA was readily detectable, whereas induction of ISG54 and ISG56 was weak (Fig. 4B). The differences in the expression of these RNAs may be due to a variety of reasons, including the fact that the half-lives of ISG54 and ISG56 are very short compared with 6 -16, which is very stable (data not shown). Incubation of Clone 8 cells with 1000 units/ml of IFN␣ causes a slightly larger amount of tyrosine-phosphorylated Stat2 than 100 units/ml and is between that observed with 10 and 100 units/ml of IFN␣ in H123 cells. Incubation of H123 cells with 100 units/ml of IFN␣ induced maximal accumulation of ISG RNAs, whereas neither 100 nor 1000 units/ml of IFN␣ stimulated ISG RNA expression in Clone 8 cells. IFN␣-stimulated tyrosine phosphorylation of Stat1 was approximately the same in both cell lines. These results strongly suggest that not only is there diminished tyrosine phosphorylation of Stat2 in Clone 8 cells incubated with IFN␣, but there is likely a defect in the protein because the amount of tyrosine-phosphorylated Stat2 detected in Clone 8 cells clearly induces ISG expression in H123 cells but not in Clone 8 cells. The fact that we cannot detect ISGF3 formation or an association of Stat1 and Stat2 in Clone 8 cells treated with IFN␣ supports this hypothesis.
To determine whether there is a mutation in the Stat2 expressed in Clone 8 cells, RNA was isolated and RT-PCR was used to obtain Stat2 cDNAs. Four individual clones were sequenced, and a single point mutation (Cys to Thr) was detected in the coding region (Fig. 5A). This mutation causes a substitution of proline 630 to leucine (P630L). Sequence homology alignment shows that Pro-630 is present in all mammalian Stat proteins as well as in Drosophila Stat (see Fig. 5B). Several of the amino acids adjacent to Pro-630 are also conserved in other Stat family members, suggesting its importance in Stat structure and/or activity. It is notable that the crystal structure of Stat1 indicates that the hydrophobic side chains of Val-641 and Ala-642 are located in the dimer interface of Stat1 (18).
To verify that the defect in IFN␣-stimulated ISRE-dependent gene expression could be restored in Clone 8 cells by expression of Stat2, ISRE reporter assays were performed. Clone 8 cells were transfected with an ISRE-luciferase reporter construct and increasing amounts of wild type Stat2. A cytomegalovirus Renilla luciferase was also included to normalize for transfection efficiency (Fig. 5D). Transfection of increasing amounts of wild type Stat2 into Clone 8 cells induced a dosedependent increase in IFN␣-stimulated ISRE-luciferase activity that attained nearly the same level that was observed in H123 cells (compare lanes 3 and 4 with 5). As expected, transfection of a vector that expresses green fluorescent protein did not restore IFN␣-stimulated ISRE-dependent luciferase activ- To examine the function of Stat2 P630L in the absence of endogenous wild type Stat2, we expressed either FLAG-tagged wild type or Stat2 P630L in Stat2-null 2fTGH human fibrosarcoma cells (U6A). Pools of cells that stably express Stat2 protein were selected and incubated with or without IFN␣ for 30 min. Whole cell extracts were immunoblotted with phosphospecific Stat1 and Stat2 antibodies (Fig. 6A). Similar to Clone 8 cells, IFN␣-stimulated tyrosine phosphorylation of Stat2 in U6A cells that express Stat2 P630L was greatly diminished compared with cells expressing wild type protein (Fig. 6A). IFN␣-stimulated tyrosine phosphorylation of Stat1 has also been shown to be diminished in U6A cells, probably because Stat2 provides a docking site for Stat1 to effectively engage Jak1 and Tyk2 when they are bound to the receptor (19). Interestingly, whereas IFN␣-stimulated tyrosine phosphorylation of Stat1 is enhanced in U6A cells expressing wild type Stat2, in cells expressing Stat2 P630L the level of tyrosinephosphorylated Stat1 remains about the same as cells that do not express Stat2 (Fig. 6, compare lanes 2, 4, and 6). Consistent with Clone 8 cells treated with IFN␣ (see Fig. 3B), we are also unable to detect any association between Stat1 and Stat2 in IFN␣-treated U6A cells reconstituted with Stat2 P630L (Fig.  6B, lane 2 compared with lane 4, lower panel).
We have examined the ability of Stat2 P630L to restore IFN␣-stimulated transcription in U6A cells (Fig. 7A). U6A cells stably expressing wild type or Stat2 P630L were transfected with an ISRE luciferase reporter and a cytomegalovirus-driven Renilla luciferase plasmid to normalize for transfection effi-ciency. Cells were incubated with or without IFN␣ for 6 h, and extracts were assayed for luciferase activity. Transfection of cells with wild type Stat2 permitted IFN␣ to induce luciferase activity about 10-fold compared with untreated cells. This degree of IFN␣-stimulated ISRE reporter activity has been previously observed (20,21). Transfection of Stat2 P630L is unable to restore an IFN␣-stimulated ISRE-dependent gene expression. To determine whether Stat2 P630L functions to block the actions of wild type Stat2, we have expressed Stat2 P630L with either wild type Stat2 or by itself in H123 cells or in 293T cells. In all cellular contexts, transfection of increasing amounts of Stat2 P630L plasmid blocked IFN␣-stimulated ISRE luciferase reporter activity. These results suggest that the Stat2 P630L not only is ineffective as a transcriptional activator, but its expression can inhibit the actions of wild type Stat2. DISCUSSION Chemical mutagenesis and selection of cells lines that do not express Jaks or Stats have contributed greatly to our knowledge of the components of the Jak/Stat pathway required for activation of ISGs by interferons. Using these cell lines, it has been possible to identify domains within Jaks and Stats required for their function (1). The determination of individual residues that are important for the function of these proteins has been obtained through examination of sequences that are conserved between members of the family as well as between species. In addition, the crystal structure of the Stats has provided useful information concerning important residues for Stat function (18). However, the number of conserved regions within the Jak and Stat family of proteins are numerous, and analysis of each conserved residue by point mutations and expression in null cells is a daunting task. The other approach to determine important amino acids and/or domains required for the function of proteins is the examination of diseases that result from mutations in known signaling molecules. Point mutations in Jak3 in humans cause severe combined immunodeficiency, and analysis of a variety of these naturally occurring mutations has provided important information concerning the structure/function of Jak3 as well as other members of the family (3,4). So far, three mutations in Stat1 have been described in humans that are associated with impairment of immunity to mycobacterial and viral infection (6). One mutation (L706S) is an amino acid that is required for Stat1 homodimerization, whereas the other mutations (L600P and a two-nucleotide deletion in exon 20 that generates a premature stop codon at position 603) both reside within the SH2 domain (6).
Although identification of naturally occurring mutants of Jaks and Stats that are associated with diseases has been very informative in terms of understanding the functions of these proteins, these mutations are very rare, limiting the information that can be obtained from this approach. To define both physiologically and structurally important mutants in the Jaks and Stats, we have used a selection strategy to isolate IFN␣resistant cells, which normally do not survive in the presence of this cytokine. Using this approach we hoped to isolate IFN␣resistant variants that will provide us with information concerning critical previously unrecognized residues that are important for Jak/or Stat function. One IFN␣-resistant clone obtained from this screen expresses a Stat2 protein with an amino acid substitution in proline 630 to leucine. It is notable that Pro-630 is conserved in all mammalian Stats as well as in Drosophila Stat. One amino acid N-terminal and three amino acids C-terminal to the Pro-630 are also highly conserved (Fig.  5). Furthermore, the hydrophobic side chains of two of these residues, alanine 641 and valine 642, pack at the interface of Stat1 homodimers. The fact that we are unable to detect Stat1/ Stat2 dimers by either EMSA or coimmunoprecipitation suggests that Pro-630 in Stat2 is also critical for heterodimer formation. We are presently examining whether Stat1 homodimer formation requires Pro-633. It is reasonable to speculate that the folding of the protein might also be disrupted in Stat2 P630L because it is poorly tyrosine-phosphorylated by treatment of cells with IFN␣ and is not significantly translocated to the nucleus. The lack of tyrosine phosphorylation and nuclear translocation of Stat1 is also observed with a mutation in the conserved arginine 602 in the SH2 domain of Stat1 (18,22). Interestingly, examination of the crystal structure of Stat1 also shows that the conserved proline 630 in the SH2 domain is in close proximity to arginine 602. It is possible that both amino acid residues might be required to form a high affinity binding site for the phosphorylated tyrosyl residue in Stats for dimerization. Although Pro-630 and the surrounding conserved amino acids are located in the SH2 domains of Stat proteins, we have not identified this sequence in the domains of any other SH2 domain-containing proteins.
There are presently a number of malignancies that are being treated with type I IFNs, including chronic myelogenous leukemia, non-Hodgkin's lymphomas, and melanoma. Clinical resistance to IFN␣ occurs in all of these malignancies. Now that we have identified a mutation in Stat2 that was selected by chronic exposure of the human leukemic Jurkat variant line to IFN␣, we are in a position to determine whether this mutation might be also seen in patients who develop resistance to IFN␣ as a result of extended therapy with this cytokine.