Targeted Inhibition of Interferon-γ-dependent Intercellular Adhesion Molecule-1 (ICAM-1) Expression Using Dominant-Negative Stat1*

A subset of epithelial immune-response genes (including intercellular adhesion molecule-1 (ICAM-1)) depends on an IFN-γ signal transduction pathway with the Stat1 transcription factor as a critical intermediate. Excessive local activation of this pathway may lead to airway inflammation, so we sought to selectively down-regulate the pathway using a dominant-negative strategy for inhibition of epithelial Stat1 in a primary culture airway epithelial cell model. Using a Stat1-deficient cell line, we demonstrated that transfection of wild-type Stat1 expression plasmid restored appropriate Stat1 expression and IFN-γ-dependent phosphorylation as well as consequent IFN-γ activation of cotransfected ICAM-1 promoter constructs and endogenous ICAM-1 gene expression. However, mutations of Stat1 at Tyr-701 (JAK kinase phosphorylation site), Glu-428/429 (putative DNA-binding site), His-713 (splice site resulting in Stat1β formation), or Ser-727 (MAP kinase phosphorylation site) all decreased Stat1 capacity to activate the ICAM-1 promoter. The Tyr-701 mutant (followed by the His-713 mutant) were most effective in disabling Stat1 function and in overcoming the activating effect of cotransfected wild-type Stat1 in this cell system thereby highlighting the effectiveness of blocking Stat1 homo- and hetero-dimerization. In experiments using primary culture human tracheobronchial epithelial cells (hTBECs) and each of the four Stat1 mutant plasmids, transfection with the Tyr-701 and His-713 mutants again most effectively inhibited IFN-γ activation of an ICAM-1 gene promoter construct. Then by transfecting hTBECs with wild-type or mutant Stat1 tagged with a Flag reporter sequence, we used dual immunofluorescence to show that hTBECs expressing the Tyr-701 or His-713 mutants were prevented from expressing endogenous ICAM-1 in response to IFN-γ treatment. The capacity of a specific Stat1 mutations to exert a potent dominant-negative effect on IFN-γ signal transduction provides for further definition of Stat1 structure function and a means for natural or engineered expression of mutant Stat1 to selectively down-regulate activity of this pathway in a cell type- or tissue-specific manner during immune and/or inflammatory responses.

Transcription factors generally contain at least two independent domains for DNA binding and for activation of transcription (1). Removal of the transactivation domain has been shown in many cases to result in an inactive factor that can bind and displace wild-type protein thereby creating a dominant-negative action (2). Targeting such a dominant-negative construct so that it is expressed in a specific tissue has been useful in understanding the function of specific transcription factors in different tissues. Accordingly, this strategy offers an advantage over complete deletion of the factor by homologous recombination with the endogenous gene if there is a goal of defining function in a specific tissue or cell type. In that context, we have determined that epithelial barrier tissue (and airway epithelial cells in particular) selectively activate a subset of immune response genes to mediate immunity and inflammation, and this activation is controlled by the Stat1 transcription factor (3,4). Thus, the present experiments were aimed at establishing a dominant-negative strategy for investigating the action of epithelial Stat1.
In defining this strategy, it is noteworthy that STAT 1 family proteins are somewhat more complex than other transcription factors. The STAT proteins act as critical intermediates in cytokine-dependent gene activation based on their dual capacities for signal transduction (at the cell surface) and activation of transcription (in the nucleus) (5). Signal transduction depends on programmed assembly of cytokine receptors, receptorassociated JAK kinases, and in some cases serine kinases, that recruit and activate specific STAT proteins (6 -8). Phosphorylated/activated STATs then dimerize, translocate to the nucleus, and direct transcription of specific target genes. In particular, the first member of the STAT family (designated Stat1) is critical for IFN-dependent gene activation (9). IFN-␥-dependent oligomerization of the IFN-␥ receptor and consequent crossphosphorylation of receptor-associated Jak1 and Jak2 kinases and the receptor ␣-chain leads to SH2-dependent recruitment of Stat1 (6,10). Stat1 then undergoes Tyr-701 phosphorylation and SH2-dependent release from the receptor as a homodimer (10,11) that can translocate to the nucleus and bind to a specific inverted repeat DNA element (3,12,13). Amplification steps in this pathway responsible for further specificity include MAP kinase-dependent phosphorylation of Ser-727 in the cy-toplasm (7,14) and interaction with constitutively active specificity protein 1 in the nucleus (4). Thus, the creation of a dominant-negative mutation for Stat1 is more challenging than for transcription factors with less complex function but also offers an opportunity for dissecting the relative importance of Stat1 modular function.
In addition to analysis of Stat1 structure-function ex vivo, more recent studies of Stat1 biology in the genetically Stat1deficient mouse indicate that Stat1 is required for IFN-dependent biologic responses, especially viral immunity (15,16). However, as noted above, this approach does not allow for functional assessment of tissue-specific alterations in Stat1 activity. In the case of epithelial tissue, localized assessment is essential, because it appears that selective and localized activation of IFN-␥-driven Stat1-dependent gene expression in airway epithelial cells may be a feature of respiratory viral infection as well as inflammatory diseases such as asthma (17)(18)(19). Thus, the level of Stat1-dependent gene activation may critically control immune and inflammatory responses, and regulating the level of epithelial Stat1 activity may influence the type of host response.
Accordingly, we sought a means to down-regulate the IFNdriven signal transduction pathway using a dominant-negative strategy for Stat1 in a relevant ex vivo system. We took advantage of Stat1 (and other STAT protein) structure-function relationships and a primary culture human airway epithelial cell model with a defined IFN-␥-dependent immune response gene (ICAM-1) to establish strategies for inhibition of Stat1 activity. We reasoned that mutations that may inactivate a potential Stat1 DNA-binding site at Glu-428 and Glu-429 (20), the Jak1/2 phosphorylation site at Tyr-701 (21), or the MAP kinase phosphorylation site at Ser-727 (14) or deletion of the putative Stat1 transactivation domain (the carboxyl-terminal 38 amino acids) (22) might each serve to generate a Stat1 protein that is nonfunctional (in a Stat1-deficient cell line) and may exert a dominant-negative action (in an epithelial cell model). To our knowledge, this represents the initial indication that loss of function may correlate with dominant-negative activity for Stat1 in a biologically relevant human cell model. The mutants should prove useful for regulating IFN-␥-driven gene expression in a tissue-or cell-specific manner and for further analyzing structure-function in the Stat1 pathway.
Cell Culture-Stat1-deficient U3A cells were obtained from G. Stark (Cleveland Clinic, OH) and I. Kerr (Imperiacal Cancer Research Foundation, London) were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, L-glutamine and penicillin/streptomycin as described previously (23). Human tracheobronchial epithelial cells (hTBECs) were isolated from mucosal strips by enzymatic dissociation and then cultured in LHC-8e medium on flasks coated with collagen/albumin as described previously (3,4,24 Mutant Stat1 Expression Plasmids-Selected mutations in the Stat1 coding sequence were created using the Stat1 cDNA as template in the polymerase chain reaction (PCR) with Thermus flavus DNA polymerase (Amersham Corp.) and a 55°C annealing cycle. All PCR generated fragments were sequenced using the dideoxynucleotide technique to verify sequence integrity (26).
To mutate Stat1 at amino acids 428 and 429 (the putative DNAbinding site), an upstream primer encoding Stat1 sequence extending from the SphI site at the codon for amino acid 323 to 20 bp downstream and a downstream primer extending from the EspI site at the codon for amino acid 446 to 74 bp upstream (with the codons for amino acids 428 and 429 converted from glutamine to alanine) were used. The PCR product was inserted into SphI/EspI-digested pSK-Stat1-5UTR generating pSK-Stat1-5UTR-Glu-428/9M. To insert this mutant Stat1 sequence downstream to the CMV promoter, pSK-Stat1-5UTR-Glu428/9M was digested with PflMI/Sa-cII and the resulting Stat1 sequence encoding amino acids 1 to 472 with the mutations of amino acids 428 and 429 was inserted into PflMI/SacII-digested pcDNA3m-Stat1 generating pcDNA3 m-Stat1-Glu428/9M. To add the 8 amino acid Flag sequence downstream to this mutant Stat1 coding sequence, pcDNA3m-Stat1-Glu428/9M was digested with KpnI/EcoRI releasing the Stat1 sequence encoding amino acids 1-742 with the mutations of amino acids 428 and 429, and this sequence was inserted into KpnI/ EcoRI-digested pcDNA3m-Stat1/Flag generating pcDNA3m-Stat1-Glu428/ 9M/Flag.
To mutate Stat1 amino acid 701 (the Jak1/2 phosphorylation site), an upstream PCR primer encoding Stat1 sequence extending from the XmaI site at the codon for amino acid 606 to 20 bp downstream and a downstream primer extending from the XbaI site at the codon for amino acid 714 to 60 bp upstream (with the codon for amino acid 701 converted from tyrosine to phenylalanine) were used. The PCR product was inserted into XmaI/XbaI-digested pSK-Stat1-5UTR generating pSK-Stat1-5UTR-Tyr-701M. pSK-Stat1-5UTR-Tyr-701M was then digested with PflMI/EcoRI, and the resulting Stat1 sequence encoding amino acids 473 to 742 with mutation of Tyr701 was inserted into PflMI/EcoRI-digested pBX-Stat1 generating pBX-Stat1-Tyr-701M. To insert this mutant Stat1 sequence downstream to the CMV promoter, pBX-Stat1-Tyr-701M was digested with SacII/BamHI, and the resulting Stat1 sequence encoding amino acids 1-750 with mutation of amino acid 701 was inserted into SacII/ BamHI-digested pcDNA3m-Stat1 generating pcDNA3m-Stat1-Tyr-701 M.
The Flag sequence was inserted downstream to the mutant Stat1 sequence in pcDNA3m-Stat1-Tyr701M to generate pcDNA3m-Stat1-Tyr-701M/Flag in a manner identical to that used to generate pcDNA3m-Stat1/Flag.
To delete Stat1 sequence encoding amino acids 713 to 750 (generating Stat1␤), an upstream primer encoding Stat1 sequence extending from the HindIII site at the codon for amino acid 524 to 20 bp downstream and a downstream primer containing a BamHI and HpaI site, a stop codon, and Stat1 sequence extending from the codon at amino acid 712 to 20 bp upstream were used. The PCR product was inserted into HindIII/BamHIdigested pcDNA3m-Stat1 to generate pcDNA3m-Stat1-His-713M. To add the Flag sequence downstream to this mutant Stat1 coding sequence, the identical upstream primer and a downstream primer containing a BamHI and HpaI site, a stop codon, the Flag sequence, and Stat1 sequence extending from amino acid 712 to 20 bp upstream were used. The PCR product was inserted into HindIII/BamHI-digested pcDNA3m-Stat1 to generate pcDNA3m-Stat1-His-713M/Flag.
To mutate Stat1 at amino acid 727 (the MAPK phosphorylation site), an upstream primer encoding Stat1 sequence extending from the XbaI site at the codon for amino acid 714 to 55 bp downstream (with the codon for amino acid 727 converted from serine to alanine) and a downstream primer extending from the XbaI site in the multiple cloning site to 20 bp upstream were used. The PCR product was inserted into XbaI-digested pBX-Stat1 generating pBX-Stat1-Ser-727M. pBX-Stat1-Ser-727M was then digested with PflMI/Bsu36I, and the resulting Stat1 sequence encoding amino acids 473 to 728 with mutation of Ser-727 was inserted into PflMI/Bsu36I-digested pcDNA3m-Stat1 generating pcDNA3m-Stat1-Ser727M. The Flag sequence was inserted downstream to the mutant Stat1 sequence in pcDNA3m-Stat1-Ser-727M to generate pcDNA3m-Stat1-Ser727M/Flag as described above for pcDNA3m-Stat1-Glu-428/9M/Flag. DNA Transfection-Plasmid DNA for wild-type and each mutant Stat1 was purified by two successive centrifugations through cesium chloride and then used to transfect duplicate U3A or hTBEC monolayers at 60 -80% confluency. Monolayers (in a 35-mm tissue culture plate) were treated with 2.0 ml of serum-free media containing 3 g of DNA and 12 g of Lipofectin (Life Technologies, Inc.) for 16 h at 37°C and then with 2 ml of complete media containing IFN-␥ (10 or 100 units/ml) for 1 h (to assess phosphorylation or nuclear translocation of Stat1) or 18 -48 h (to assess ICAM-1 level or luciferase reporter gene activity) at 37°C. For cotransfection experiments, the levels of reporter and expression plasmids were titered to provide similar levels of basal promoter activity among experiments (1.5-2.5 g of reporter plasmid and 0.05-1.5 g of expression plasmid per 30-mm culture plate). The construction of the luciferase reporter plasmid with ICAM-1 gene sequence (Ϫ130 to Ϫ35) driving the Photinus pyralis luciferase gene (pBH-130 -35ICAM-1-luc) was described previously (3). Controls for transfection efficiency were performed using cotransfection with plasmids containing RSV-CAT as described previously (3,4,27). In addition, multiple DNA transfections (at least three experiments done in duplicate) were performed for each experimental condition to define consistent results.
Immunoprecipitation and Immunoblotting-To monitor expression of wild-type and mutant Stat1 constructs in transfected cells, the cellular proteins were subjected to immunoblotting using anti-Stat1 and anti-phosphotyrosine mAbs. Cell protein extracts were prepared by lysing cells in 50 mM Tris, pH 8, 150 mM NaCl, 0.5% Nonidet P-40, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 0.75 mM dithiothreitol, 1 mM sodium orthovanadate, 10 mM sodium fluoride, 2 mM sodium pyrophosphate, 10 mg/ml leupeptin, and 10 g/ml aprotinin (28). For Stat1 immunoblotting, aliquots of cell protein were subjected to SDS-polyacrylamide gel electrophoresis in 7% polyacrylamide, electrophoretically transferred to polyvinylidene difluoride membrane (Immobilon-P, Millipore, Bedford, MA), which were then washed, blocked with 5% dry milk, and incubated with anti-Stat1 or anti-Stat3 mAb (0.5 g/ml). Primary antibody binding was detected using anti-mouse IgG 1 horseradish peroxidase conjugate and an enhanced chemiluminescence detection system (Amersham). Specificity for Stat1 immunoblotting and equality of protein loading was demonstrated by re-blotting against anti-Stat3 mAb. To detect phosphorylated Stat1, 250 g of cell protein was incubated for 1 h at 4°C with 10 g of anti-Stat1 mAb conjugated to agarose beads. The resulting immune complex was subjected to SDS-polyacrylamide gel electrophoresis in 7% polyacrylamide and then electrophoretically transferred to polyvinylidene difluoride membrane for immunoblotting against anti-phosphotyrosine mAb conjugated to horseradish peroxidase (0.2 g/ml) and detection by enhanced chemiluminescence.
Reporter Gene Assay-Reporter gene assays (for luciferase and CAT activities) were performed as described previously (3,4,27). For the luciferase assay, cell monolayers were washed and then lysed by addition of 50 mM Tris-MES, pH 7.8, containing 1% Triton X-100 and 1 mM dithiothreitol. The lysate was cleared by centrifugation, and a lysate volume corresponding to 75 g of total protein was then added to 200 l of reaction buffer (50 mM Tris-MES, pH 7.8, 10 mM magnesium acetate, 1 mM ATP) in a luminometer cuvette. Luciferase activity was determined after addition of 100 l of 1 mM luciferin using a Monolight 2010 luminometer (Analytical Luminescence Laboratory, San Diego, CA). All results are the average of duplicate samples and are representative of three experiments.
Immunofluorescent Cytochemistry-hTBECs and U3A cells were grown to 80% confluency on 8-chamber slides (Nunc, Naperville, IL) and then were transfected with 0.3 g of wild-type or mutant Stat1 expression plasmids as described above. After treatment with or without IFN-␥ for 18 -48 h, cells were fixed in methanol, pretreated with 2% gelatin in phosphate-buffered saline for 1 h at 25°C to block nonspecific Ig binding, and then treated with anti-Stat1 Ab 91T (1:900 v/v), anti-Flag mAb (1 g/ml), biotin-conjugated anti-Flag mAb (1 g/ml), or anti-ICAM-1 mAb 84H10 (0.2 g/ml) for 1 h at 25°C. Cells were rinsed and then were treated with FITC-or CY3-conjugated secondary Ab for 1 h at 25°C. Slides were mounted and then photographed using an epifluorescence photomicrography system (model D-7082, Carl Zeiss, Thornwood, NY).

Wild-type and Mutant Stat1
Behavior in Stat1-complemented Cells-To first determine the characteristics of wildtype and mutant Stat1 expression plasmids, we examined the behavior of transiently expressed Stat1 proteins in a Stat1deficient cell line derived by somatic mutation from HT1080 fibrosarcoma cells (23). Immunoblot analysis of newly expressed wild-type and mutant Stat1 proteins in this system using anti-Stat1 and anti-phosphotyrosine mAbs indicated high level expression of wild-type Stat1 that was appropriately phosphorylated in response to IFN-␥ treatment (Fig. 1). Mutations at Glu-428/429 (a putative DNA-binding site) or Ser-727 (a MAP kinase phosphorylation site) gave a similar level of expression and phosphorylation as wild-type, whereas mutation at Tyr-701 correctly prevented phosphorylation. Consistent with the absence of tyrosine phosphorylation, we also demonstrated that wild-type but not Tyr-701 mutant Stat1 translocated to the nucleus in response to IFN-␥ in this cell system (data not shown). Mutation at His-713 (causing truncation at Val-712) resulted in an appropriately smaller (84-kDa) Stat1␤ species, which was unseparated from the phosphorylated species by this electrophoresis technique (as is also true of endogenous Stat1␤) (22). Each mutant Stat1 species appeared to be expressed at a similar level to the wild-type Stat1 in this expression system (consistent with similar levels of transfection efficiency for each plasmid construct).
Modification of ICAM-1 Promoter Activity in Stat1-complemented Cells-To next determine the capacity of wild-type and mutant Stat1 to regulate IFN-␥-dependent gene activation in this same cell system, we cotransfected each of the Stat1 expression constructs with a reporter plasmid that contained the ICAM-1 gene interferon-␥-response element (IRE) driving a luciferase reporter gene (3,4). The ICAM-1 promoter was chosen in preparation for similar experiments with hTBEC cells (as noted below).
Expression plasmids encoding wild-type Stat1 conferred IFN-␥ responsiveness in this system, whereas each of the four mutations disabled Stat1 function to varying degrees ( Fig. 2A). The Tyr-701 mutation was the most effective and the Ser-727 mutation was the least effective in inactivating Stat1 function ( Fig. 2A). In their possible role as dominant-negative mutations, each of the four Stat1 mutants exhibited the same rank order of potency for antagonizing wild-type Stat1 action as assessed by inhibition of IRE-driven gene activation in Stat1complemented cells with Tyr-701 and His-713 mutants as the most effective (Fig. 2B). This system was designed with a ratio of mutant to wild-type Stat1 expression plasmid of 10:1 after titration of this ratio indicated a maximal effect for the Tyr-701 and His-713 mutants (Fig. 3).
Taken together, these findings provided relatively good correlation between the failure to mediate Stat1-dependent gene activation and the capacity to antagonize wild-type Stat1-dependent gene activation. The findings also offered initial evidence that Glu-428/429 was a critical site for Stat1 action (analogous to Glu-434/435 for Stat3) (20) and that Ser-727 phosphorylation was required for maximal but not partial Stat1 action. The degree of decrement in Stat1 function with Ser-727 mutation was less than observed for other gene promoter constructs (14), suggesting some gene-specific differences in the importance of this site for Stat1 function.
To further determine whether Stat1 expression constructs modify endogenous ICAM-1 promoter activity in U3A cells, we also monitored ICAM-1 expression that is transcriptionally induced by IFN-␥ treatment (3). Using dual immunofluorescence labeling for Stat1 and ICAM-1, we found that cells containing wild-type Stat1 (but not Stat1-Tyr-701M) expressed ICAM-1 after IFN-␥ stimulation (Fig. 4). The lower level of ICAM-1 expression relative to ICAM-1 promoter activity in U3A cells (compared with hTBECs) may reflect an additional post-transcriptional influence on ICAM-1 levels in hTBECs. 2 Each of these approaches (monitoring ICAM-1 promoter activ-ity and tracking ICAM-1 expression) was then extended to studies of primary culture hTBECs to determine whether Stat1 mutants exerted a dominant-negative action on endogenous Stat1 in a biologically relevant cell system.
Capacity of Stat1 Mutants to Inhibit ICAM-1 Gene Promoter Activity in hTBECs-In contrast to experiments using complemented U3A cells, we now aimed to compete native Stat1 activity by overexpressing mutant Stat1 proteins using an airway epithelial cell model system (hTBECs) with well defined IFN-␥-dependent transcriptional control for the ICAM-1 gene promoter (3,4,24). The initial approach was similar to that used for U3A cells, in which ICAM-1 promoter activity was monitored using a cotransfected ICAM-1 luciferase reporter plasmid. As was the case for Stat1-complemented U3A cells, the Tyr-701M construct most effectively inhibited IFN-␥-dependent promoter activity (Fig. 5). The His-713M construct was nearly as effective in blocking endogenous Stat1 action, whereas the Glu-428/9M and Ser-727M constructs had no significant effect in antagonizing Stat1 action on ICAM-1 promoter activity. This system was designed with a ratio of expression to reporter plasmid of 1:1, after experiments with higher ratios of expression to reporter plasmid offered no ad- ditional inhibitory effect for any of the four mutant Stat1 plasmids (because higher levels of expression plasmid resulted in promoter competition that extinguished reporter signal). Each expression plasmid construct had an equivalent effect on basal promoter activity of the reporter plasmid suggesting that each construct was transcribed at a similar level in this cell system (consistent with equivalent levels of transfection efficiency and protein expression for each construct as noted above).
Capacity of Stat1 Mutants to Inhibit Endogenous ICAM-1 Gene Expression in hTBECs-We next aimed to determine whether inhibition of Stat1 activity by mutant Stat1 constructs were able to also block expression of the endogenous ICAM-1 gene. We again used a strategy similar to the one used for U3A cells, but in this case, it was necessary to separate transiently expressed from native Stat1 protein. Accordingly, we generated constructs encoding for wild-type and Tyr-701M Stat1 fused to an 8-amino acid epitope (Flag) that could be used to track expression. Validation experiments in U3A cells indicated that the wild-type Stat1/Flag fusion protein was appropriately expressed, phosphorylated, translocated, and was capable of mediating activation of the ICAM-1 promoter, whereas the Stat1-Tyr-701M/Flag fusion protein was expressed at the same level but failed to undergo phosphorylation or translocation or to activate the ICAM-1 promoter and was effective in blocking wild-type Stat1 activity when transfected into cells at a ratio of mutant to wild-type Stat1 of 10:1 ( Fig. 1 and data not shown).
In the case of hTBECs, expression of wild-type Stat1/Flag was again similar to wild-type Stat1 as assessed by the level of protein expression and pattern of cellular immunofluorescence. In addition, hTBECs transfected with wild-type Stat1/Flag responded to IFN-␥ with a loss of the nuclear halo (indicating nuclear translocation of Stat1/Flag) and expression of ICAM-1 (indicating ICAM-1 gene activation by Stat1/Flag) (Fig. 6). By contrast, hTBECs expressing Stat1-Tyr-701M/Flag showed no change in Stat1 immunofluorescence pattern and failed to express ICAM-1 in response to IFN-␥ treatment. Additional experiments with each of the other three Stat1 mutants indicated that their effects on ICAM-1 promoter activity also correlated well with effects on endogenous ICAM-1 expression (Fig. 7). Thus, the Tyr-701 and His-713 mutants but not Glu-428/9 and Ser-727 mutants were effective in blocking both ICAM-1 pro-moter/reporter activity and ICAM-1 expression induced by IFN-␥. DISCUSSION This report represents the initial examination of dominantnegative action for mutant Stat1 on endogenous gene expression in a primary culture cell model. In that context, however, several STAT family members have been examined for dominant-negative activities on reporter gene and in some cases endogenous gene expression in transformed cell models. Stat3 bears the closest homology to Stat1 and mutations at its DNAbinding site (Glu-434/435) or JAK kinase phosphorylation site (Tyr-705) as well as deletion of the carboxyl-terminal portion that includes the transactivation domain (yielding Stat3␤) appear to suppress the activity of Stat3-responsive reporter constructs transfected into transformed cell lines (29 -31). This approach required cotransfection of wild-type Stat3 to reconstitute the complete activation pathway (in COS cells) or the presence of multiple copies of the Stat3 DNA response element joined to a thymidine kinase or junB minimal promoter in the 5Ј flanking region of the reporter (in HepG2 or M1 cells). Other recent reports indicate that deletion of the carboxyl-terminal transactivation domain of Stat5 also results in sustained DNA binding and a dominant-negative effect on reporter gene activity of Stat5-responsive constructs (in COS and Ba/F3 cell lines) and expression level of interleukin-3 early response genes (in interleukin-3-dependent Ba/F3 or 32Dc1-Epo1 wild-type cells) (32)(33)(34). In addition, an analysis of Stat2 indicates that mutations at the JAK kinase phosphorylation site or an internal deletion involving this site can at least partially inhibit reporter gene activity of IFN-␣-responsive constructs in WI-38 cells (35). We also found that loss of function correlated closely with dominant-negative action for Stat1 in a Stat1-complemented U3A cell line. However, these correlations were less apparent in studies of endogenous gene activation and expression in primary culture human airway epithelial cells. Thus, each of these studies highlights the need for investigating STAT function in primary culture human cell models with defined pathways for STAT-dependent regulation of endogenous cytokine-dependent genes.
Our data indicate that four distinct mutations of Stat1 cause loss of function, and at least two of these mutations (Tyr-701  the activated IFN-␥ receptor. Thus, Stat1 and Stat1-Tyr-701M exhibit equivalent capacities for recruitment to the phosphorylated receptor (via native SH2 domains). However, only wildtype Stat1 is then likely to be capable of dissociating from the receptor, because this dissociation step depends on Tyr-701 phosphorylation of Stat1 and consequent affinity-driven association with a companion Stat1 in preference to the IFN-␥ receptor ␣-chain (10). The marked effectiveness of Tyr-701 mutation in down-regulating Stat1-dependent gene activation indicates that interfering with this relatively early step in IFN-␥ signal transduction (i.e. Stat1 homodimerization) is critical for efficiently blocking this pathway.
In contrast to the Tyr-701 mutant, Stat1␤ effectively undergoes tyrosine phosphorylation with consequent homodimerization (with itself) or heterodimerization (with full-length Stat1) and translocation to the nucleus with binding to the IRE. However, once bound, the truncated version of Stat1 is incapable of activating transcription. The dominant-negative action of Stat1␤ appears to be based on its failure to activate transcription as a homodimer and its capacity to "decoy" full-length Stat1 (Stat1␣) to form a less effective Stat1/Stat1␤ heterodimer. Accordingly, the degree of blockade may depend on the relative affinities of Stat1␤ for Stat1 in comparison to Stat1 for itself. In fact, the slight decrease in dominant-negative activity for Stat1␤ (compared with the Tyr-701 mutant; Fig. 3) may reflect the possibility that phosphorylated Stat1 has a higher affinity for activated Stat1 than for activated Stat1␤. It is also possible that Stat1␤ bound to the IRE may recruit additional Stat1 through its amino-terminal end as described for promoters containing multiple IRE sites (36). In either case, Stat1␤ exerts significant dominant-negative activity only when its expression exceeds that for native Stat1, and maximal inhibition of Stat1 action was not observed until the ratio of Stat1␤ to Stat1 was 10:1. This finding argues against the possibility that Stat1␤ might function as a naturally occurring negative regulator of Stat1 activity until its expression levels exceed that for Stat1. To date, cellular levels of Stat1␤ are twoto three-fold lower than for Stat1 in airway epithelial cells 3 and other cell types (22). However, the ratio of full-length Stat1 to Stat1␤ and possible dysregulation of this ratio has not yet been examined during in vivo conditions with altered activity of IFN-␥.
Mutations of the putative DNA-binding site (Glu-428 and -429) or the MAP kinase phosphorylation site (Ser-727) were significantly less effective than Tyr-701 and His-713 in antagonizing transiently-expressed or endogenous Stat1 action. In the case of Ser-727M, it appears that the mutation is relatively ineffective in causing loss of Stat1 function, so it is not surprising that the mutation also exerts a relatively weak dominantnegative action. However, in the case of Glu-428/9M, it appears that the mutation causes loss of Stat1 function (due to decreased DNA-binding activity) but is relatively inactive in antagonizing the activity of transiently expressed or endogenous Stat1. We found no difference in the level of expression for this construct (or other constructs) to account for decreased action, but it is still possible that subtle decreases in expression (that are difficult to define during transient expression) might account for diminished dominant-negative activity. Nonetheless, it seems more likely that the Glu-428/9 mutant may form a heterodimer with Stat1 that is still capable of causing gene activation. This possibility again highlights the basis for the effectiveness of the Tyr-701 mutation in inhibiting native Stat1 activity because this mutation (unlike the others) completely prevents the formation of partially active Stat1 heterodimers.
In summary, our findings indicate that overexpression of mutant Stat1 (especially Tyr-701 or His-713 mutation) interrupts IFN-␥-driven signal transduction in an airway epithelial cell model that has proved highly characteristic of epithelial behavior in vivo (37). The biologic context for the present work includes the finding that IFN-␥-driven expression of airway epithelial ICAM-1 (as well as interferon regulatory factor-1 and RANTES) is a major determinant of epithelial immune cell interaction (24, 38 -41). In addition, it appears that the activity of this IFN-␥-driven pathway (monitored by Stat1 activation and target gene expression) is increased during the immune response to respiratory viruses and the inflammatory response that is characteristic of asthma (17)(18)(19). It is likely therefore that an IFN-␥-driven, Stat1-dependent subset of immune response genes (including ICAM-1, interferon regulatory factor-1, RANTES, and Stat1 itself) provides a molecular link between respiratory viral infection and the development of asthma. Thus, the present attempt to down-regulate Stat1-dependent gene activation provides another step toward better defining the role of this activation pathway in models of airway immunity and inflammation. Studies of Stat1-deficient mice indicate that Stat1 is required for immunity to nonrespiratory viruses (15,16), but the role of Stat1-dependent gene activation in response to inhaled agents (infectious or allergic) or in the immune function of the airway epithelium is still uncertain. The present results imply that naturally occurring or genetically engineered defects in phosphorylation-dependent activation of Stat1 will provide the most potent insight into the role of Stat1 in mediating airway epithelial immunity and inflammation.