Configuration of the interferon-alpha/beta receptor complex determines the context of the biological response.

Constituents of the Type 1 interferon (IFN) receptor (IFNABR) identified to date include the α and β transmembrane subunits and the associated intracellular kinases, Jak 1 and Tyk 2. In this report, we demonstrate that a human cell type that expresses both subunits of IFNABR, together with Jak 1 and Tyk 2, exhibits a limited binding capacity for and is only partially sensitive to the effects of IFN-α/β, despite adequate levels of the cytoplasmic transcription factors Stat1, Stat2, and Stat3. Specifically, a low affinity interaction between IFN-α/β and cell surface receptors results in ISGF3 (Stat1:2) activation and an antiviral response, yet no IFN-inducible growth inhibition. Using a panel of murine cells that are variably configured with respect to the human IFNABR-α/β subunits, we provide evidence that an additional component(s) encoded on human chromosome 21 is required to confer high affinity binding and IFN-inducible growth inhibition to cells that express the α and β subunits of the IFNABR. The data indicate that transcriptional activation that leads to an antiviral response is mediated by IFN-α/β activation of IFNABR-α and IFNABR-β in the context of a low affinity interaction, yet a high affinity interaction is necessary for signal transducing events that mediate growth inhibition. We provide evidence that the extent of ISGF3 activation correlates directly with the magnitude of an antiviral but not a growth inhibitory response.

Constituents of the Type 1 interferon (IFN) receptor (IFNABR) identified to date include the ␣ and ␤ transmembrane subunits and the associated intracellular kinases, Jak 1 and Tyk 2. In this report, we demonstrate that a human cell type that expresses both subunits of IFNABR, together with Jak 1 and Tyk 2, exhibits a limited binding capacity for and is only partially sensitive to the effects of IFN-␣/␤, despite adequate levels of the cytoplasmic transcription factors Stat1, Stat2, and

Stat3. Specifically, a low affinity interaction between IFN-␣/␤ and cell surface receptors results in ISGF3 (Stat1:2) activation and an antiviral response, yet no IFN-inducible growth inhibition. Using a panel of murine cells that are variably configured with respect to the human IFNABR-␣/␤ subunits, we provide evidence that an additional component(s) encoded on human chromosome 21 is required to confer high affinity binding and IFN-inducible growth inhibition to cells that express the ␣ and ␤ subunits of the IFNABR. The data indicate that transcriptional activation that leads to an antiviral response is mediated by IFN-␣/␤ activation of IFNABR-␣ and IFNABR-␤ in the context of a low affinity
interaction, yet a high affinity interaction is necessary for signal transducing events that mediate growth inhibition. We provide evidence that the extent of ISGF3 activation correlates directly with the magnitude of an antiviral but not a growth inhibitory response.
Type 1 interferons (IFN), 1 comprised of ␣ and ␤ subtypes, are a family of biologically active related proteins that exhibit target cell specificity (1,2). This specificity is mediated by a high affinity interaction between the IFN and its cell surface receptor (3)(4)(5). Accumulating evidence indicates that the Type 1 IFN receptor is multicomponent. A cDNA coding for a human IFN-␣ receptor peptide (IFNABR-␣) was reported to code for a functional human IFN-␣ receptor when transfected into murine cells (6). However, sensitivity to the biological effects of human IFN was restricted to a specific IFN-␣ species, IFN-␣8. More recently, a distinct cell surface receptor protein has been identified (7). This receptor, IFNABR-␤, may represent the primary ligand recognition chain of the IFN receptor complex. Studies from our laboratory demonstrate that specific species of membrane glycosphingolipids containing a terminal Gal␣1-4Gal associate with IFNABR-␣ to create a functional receptor (8).
The initial interaction between a Type 1 IFN and its specific cell surface receptor apparently leads to ligand-induced tyrosine phosphorylation of IFNABR-␣ (9) and a rapid phosphorylation activation, in the absence of protein synthesis, of a latent cytosolic transcription factor, ISGF3 (reviewed in Ref. 10). Complementation of IFN-resistant mutant cell lines with two members of the Janus family of nonreceptor protein tyrosine kinases (Jak), Tyk 2 (11,12) and Jak 1 (13), that are required for IFN-induced signal transduction further defines a potential role for receptor-associated phosphorylation events in the signaling cascade initiated by ligand binding. Indeed, the data imply that Jak 1 associates with IFNABR-␤ (7) and Tyk2 with IFNABR-␣ (9,11,12,14). Current models for IFN-induced signaling invoke kinase-mediated phosphorylation of the STAT proteins Stat1, Stat2, and Stat3 (15)(16)(17)(18)(19). These phosphorylated STATs dimerize via SH2-phosphotyrosyl interactions (20,21) and translocate to the nucleus where they bind to specific promoter sequences, thereby regulating gene expression. Homo-and heterodimers of Stat1 and Stat3 bind the palindromic IFN response element, pIRE, (22,23). A heterodimer of Stat1 and Stat2 associates with a DNA-binding adapter protein, p48, to form the Stat complex designated ISGF3 (24,25). ISGF3 transcriptionally activates a subset of genes that contain an IFN-stimulated response element (ISRE) (26).
The existence of multiple, distinct signaling pathways that effect different biological outcomes in response to a single IFN suggests that component configuration is critical within a multimeric transmembrane receptor complex. Members of the hematopoietic growth factor family of cytokines, which include interferon-␥, interleukins, erythropoietin, growth hormone, granulocyte colony-stimulating factor, and granulocyte-macrophage colony stimulating factor, mediate their pleiotrophic effects through interactions with multicomponent receptors (27)(28)(29)(30)(31)(32)(33)(34)(35). Generally, specific ligands exhibit low affinity binding to individual receptor components, yet high affinity binding occurs when the intact receptor complex is assembled. For certain cytokine receptors, such as the interleukin-2 receptor complex, an intermediate affinity interaction between the ligand and a single receptor component, the interleukin-2␤ chain, may lead to a limited signal transduction (36). Thus, there may be some redundancy associated with multimeric receptor complexes that allows for partial responses when specific receptor components are not expressed at the cell surface.
In agreement with affinity studies with 125 I-IFN-␣ (37-44), here we report that IFNABR-␣ by itself is not sufficient for the appropriate high affinity binding and complete biological responses effected by IFN-␣ and -␤. We provide evidence that a low affinity interaction between IFN-␣/␤ and the two receptor chains IFNABR-␣ and IFNABR-␤ is associated with ISGF3 activation and antiviral responses. Cell surface expression of human chromosome 21-encoded factors, which include IFNABR-␣ and IFNABR-␤ forms, permits a high affinity interaction between the IFN and the receptor complex, such that target cells exhibit both antiviral and growth inhibitory activities.

EXPERIMENTAL PROCEDURES
Cell Cultures and Virus-Human Daudi (lymphoblastoid derived from B cells) cells were grown in 10% RPMI 1640 medium as suspension cultures. T98G (human glial), MRC-5 (human fetal lung fibroblast), and Chinese hamster ovary cells were grown as monolayer cultures in 10% modified Eagle's medium ␣. Murine A9 and a human/mouse somatic hybrid cell type composed of murine A9 cells containing 1-5 copies of intact chromosome 21 from WI-38 human fibroblast cells were grown as monolayer cultures in 20% modified Eagle's medium. Encephalomyocarditis virus, passaged on murine L929 cells, was used for this study.
Interferons-IFN-Con 1 , as a consensus IFN-␣, was a gift from Amgen, Inc. (Thousand Oaks, CA) and had a specific activity of 3.0 ϫ 10 9 units/mg protein. IFN-Con 1 was designed to represent an average IFN-␣ with an amino acid chosen at each site that was found most frequently in the known family of IFN-␣s. IFN-␤ was provided by Biogen Inc. (Cambridge, MA) and had a specific activity of 2.0 ϫ 10 8 units/mg protein. IFN-␣2b was a gift from Schering Corp. and had a specific activity of 2.0 ϫ 10 8 units/mg protein. Murine IFN-␣ was purchased from Calbiochem Corp. and had a specific activity of 1.6 ϫ 10 6 units/mg protein.
Antiviral Assay-A description of the assay for IFN-induced anti-encephalomyocarditis virus activity on monolayer cells has been described (46).
Cell Surface Receptor Binding Assays-Iodination of IFN-␣ with 125 I and the receptor binding assays for monolayer and suspension cultures have been previously described (5,47).
Oligonucleotides-A double-stranded oligonucleotide, representing nucleotides Ϫ107 to Ϫ87 of the human 2-5A synthetase gene, which contains a functional ISRE, was synthesized. The sequence is CCTTCT-GAGGAAAGGAAACCA. This oligonucleotide was synthesized with a SalI-compatible linker at the 5Ј terminus (TCGAC). Gel-purified oligonuceotide was mixed with an equimolar amount of its respective complement, heated to 65°C for 15 min, and annealed at room temperature for 18 h. This preparation was end-labeled for use in electrophoretic mobility shift assay by T4 DNA kinase in the presence of [␥-32 P]ATP. A mutant ISRE was also used, its sequence being CCTTCTGAGGCCAC-TAGAGCA.
Extract Preparation and Electrophoretic Mobility Shift Assay-Whole cell extracts were prepared from IFN-treated or untreated cells by sequential extractions in high salt buffers (8). Binding reactions for the gel retardation assay contained 5Ј-32 P-labeled oligonucleotide, nonspecific and specific competitor DNA, and whole cell extract as indicated in the figure legends. Incubations were at room temperature for 30 min. Electrophoretic mobility shift assays were performed as described previously, using native 6% polyacrylamide gels run in a Tris/ glycine/EDTA buffer, pH 8.3 (8).
RNA Purification, Gel Electrophoresis, and Northern Hybridization-Total RNA was extracted from cells by extraction with guanidinium thiocyanate and CsCl 2 ultracentrifugation (49). Northern hybridization procedures have been described elsewhere (8).
Reverse Transcription PCR-Oligonucleotide primers were designed to reverse transcribe and PCR amplify the published form of the IFNABR-␤ gene from a maximum of 2 g of total RNA isolated from cells. The sequences are as follows: 1) 5Ј-ATGCTTTTGAGCCAGAATG-3Ј, 2) 5Ј-TCACTGGGGCACAGGG-3Ј, and 3) 5Ј-TAACTGTTTGCTCT-TTATTTTCT-3Ј. Briefly, 1.5 M of primers 2 and 3 were used to reverse transcribe RNA. They were added to a solution containing 0.5 mM dNTP, 5 ϫ reaction mix (50 mM Tris-HCl, pH 8.3, 75 mM KCl, and 3 mM MgCl 2 ), and 13.3 units/l of Moloney murine leukemia virus reverse transcriptase and incubated at 42°C for 1 h. The reaction was terminated by incubation at 80°C for 10 min. Two PCR reactions of 30 cycles each were performed using Taq polymerase. The first of these (30 s at 94°C, 30 s at 49°C, 2 min at 72°C) used 2 l of reverse transcribed single strand DNA as a template and primers 1 and 3. The second (30 s at 94°C, 30 s at 54°C, 2 min at 72°C) incorporated 2 l of the first PCR reaction as template DNA. Primers 1 and 2 were used in this reaction.
MRC-5 DNA from the second PCR reaction was size-fractionated on a low melting point agarose gel, and the 992-bp band was extracted and ligated into the vector PCR II. The entire reverse transcription PCR from beginning to end was performed three individual times. Single strand sequencing was performed according to the Sequenase Version 2.0 DNA Sequencing kit. Sequences were checked against results obtained from automated DNA sequencing (Biotechnology Service Centre, University of Toronto).
Flow Cytometric Analysis of IFNABR-␣ mAbs Binding to Native IFNABR-␣ on Cells-IFNABR-␣ mAbs, AA3 and EA12, were raised against the extracellular domain of IFNABR-␣ by immunization of mice with Chinese hamster ovary cells expressing a high level of cell surface IFNABR-␣, followed by boosting with IFNABR-␣/Fc fusion protein. The antibodies were screened by enzyme-linked immunosorbent assay against IFNABR-␣/Fc fusion protein versus LFA3TIP, a fusion protein containing an identical human IgG1 Fc region, but with LFA3 domain 1 replacing the IFNABR-␣ extracellular portion (50). 0.5 ϫ 10 6 cells are incubated with fluorescence-activated cell sorter buffer (negative control) or mAb for 45 min on ice, washed, and incubated with biotin-SPconjugated F(abЈ) 2 rat anti-mouse IgG (Jackson ImmunoResearch) for a further 30 -45 min. The cells are washed and then incubated with R-phycoerythrin-conjugated streptavidin for a further 30 min. Immunofluorescence is analyzed with a Becton Dickinson FACScan (8).

ISGF3 Activation in the Context of IFNABR-␣ and IFNABR-␤ Expression Confers an Antiviral Response-Human
MRC-5 cells, although responsive to the antiviral effects of human IFN-Con 1 and IFN-␤ (Fig. 1B), are about 10-fold less sensitive than T98G cells (Fig. 1A) or other IFN-sensitive cells, e.g. A549 and HeLa cells (data not shown). In contrast to T98G cells that are growth inhibited by IFN-Con 1 and IFN-␤, with ID 50 s of ϳ100 pg/ml (Fig. 1C), MRC-5 cells are insensitive to the antiproliferative effects of these IFN (Fig. 1D). Fig. 2 describes the steady-state receptor binding characteristics of 125 I-IFN-Con 1 on proliferating MRC-5 (A) and T98G (B) cells, at 4°C. Specific binding was resolved into the characteristic high and low affinity biphasic Scatchard plot for T98G cells, yet only the low affinity, monophasic plot was obtained for the MRC-5 cells (Fig. 2C). Indeed, the binding capacity of MRC-5 cells for IFN-␣ is significantly reduced compared with T98G cells (Fig.  2, A and B).
Total RNA from MRC-5 cells, when reverse transcribed, contained cDNA that could be amplified by PCR using IFNABR-␤-specific primers (7). Analysis of the PCR products revealed an expected 397-bp band (data not shown). To further examine IFNABR-␤ expressed in MRC-5 cells, a 992-bp fragment was subcloned into the vector PCR II and sequenced. Our results indicate that an IFNABR-␤ form expressed in MRC-5 cells is encoded by a gene whose sequence is essentially identical to the previously reported sequence, with one codon difference that would result in a conservative change to the amino acid residue at position 216: Asp to Asn. In Northern blots, however, two transcripts of 1.55 and 4.5 kilobase pairs were revealed that may represent differentially spliced products of the same gene (Fig. 2D). Whole cell extracts from MRC-5 and T98G cells exhibit comparable levels of IFNABR-␤ when Western blots are probed with a polyclonal Ab to IFNABR-␤ (Fig. 2E). Flow cytometric analysis of IFNABR-␣ mAb (Fig. 2F) binding to native IFNABR-␣ on MRC-5 and T98G cells identifies similar levels of IFNABR-␣ cell surface expression on these cell types. When viewed together, the data suggest that IFNABR-␣ and IFNABR-␤ expression are not limiting factors that affect the binding capacity of MRC-5 cells.
There is emerging evidence to suggest that the Jaks, Tyk 2 and Jak 1, constitutively associate with the intracellular regions of IFNABR-␣ and IFNABR-␤, respectively, and that this association creates a productive receptor, both in terms of ligand recognition and signal transduction. IFN-␣/␤-induced receptor-mediated ISGF3 activation requires both Tyk 2 and Jak 1. Because there is good evidence to suggest that transcriptional activation that precedes IFN-induced responses is mediated by ISGF3 activation, we examined the extent of IFNinduced ISGF3 activation in both T98G and MRC-5 cells. The results in Fig. 3A demonstrate that IFN-Con 1 /IFN-␤ treatment of both T98G and MRC-5 cells resulted in specific activation of ISGF3. Antisera to Jak 1 and Tyk 2, as well as monoclonal antibodies raised against Stat1, Stat2 and Stat3, detected these factors in extracts from unstimulated T98G and MRC-5 cells (Fig. 3B), indicating that these are not limiting in the MRC-5 cells.
To further examine the specific interaction between Type 1 IFN and IFNABR-␣, we investigated the biologic effects of a number of different human Type 1 IFN in murine A9 cells that were stably transfected with the human IFNABR-␣. Northern blots confirmed that human IFNABR-␣ mRNA is expressed in the transfectants (data not shown), and flow cytometric analysis of IFNABR-␣ mAb binding to human IFNABR-␣ on A9Ϫ transfectants containing plasmid only and the A9ϩ transfectants containing IFNABR-␣ identified IFNABR-␣ cell surface expression on the A9ϩ transfectants alone (Fig. 4). The results in Fig. 5 demonstrate that murine A9ϩ cells exhibiting cell surface expression of the human IFNABR-␣ are restricted in their antiviral responsiveness to selected human Type 1 IFN. Whereas the parental A9 cells are relatively insensitive to the antiviral effects of human IFN-␣2b (ID 50 ϭ ϳ2.5 g/ml/5 ϫ 10 5 units/ml), IFN-Con 1 (ID 50 ϭ ϳ50 ng/ml/1.5 ϫ 10 5 units/ml), and IFN-␤ (ID 50 ϭ ϳ100 ng/ml/2 ϫ 10 4 units/ml) compared with murine IFN-␣ (60 ng/ml/90U/ml), when challenged with the same infecting dose of encephalomyocarditis virus (Fig.  5A), the A9ϩ cells exhibit a 200-fold increase in sensitivity to the antiviral effects of IFN-Con 1 (ID 50 ϭ ϳ0.5 ng/ml/1.5 ϫ 10 3 units/ml) (Fig. 5B). The A9ϩ transfectants remain relatively unresponsive to the antiviral effects of human IFN-␣2b and IFN-␤, with dose-response curves similar to those depicted in Fig. 5A. Examination of the sensitivity of the transfectant A9ϩ cells to the growth inhibitory effects of IFN-␣2b, IFN-Con 1 , and IFN-␤ revealed that in contrast to the partial antiviral responsiveness, the A9ϩ cells remained refractory to the antiproliferative effects of each of these IFN (Fig. 6). Gel retardation assays to determine IFN-induced ISGF3 activation in the transfectant versus the A9Ϫ cells showed that there was a direct correlation between IFN-induced ISGF3 activation in the transfectants and IFN-induced antiviral activity; whereas both IFN-␣2b and IFN-␤ failed to induce ISGF3 activation in the A9ϩ cells and failed to elicit an antiviral response in these cells, IFN-Con 1 induced ISGF3 activation and an antiviral response in the A9ϩ transfectants (Fig. 7).
In an attempt to restore complete sensitivity to the full range of biologic responses inducible by Type 1 IFN, we examined the influence of chromosome 21-encoded factors, in addition to IFNABR-␣ and IFNABR-␤ expression, on biological outcome. We examined the biological effects of different human Type 1 IFN, both IFN-␣s and IFN-␤, in murine A9 cells that contain 1-5 copies of intact human chromosome 21, designated A9ϩ21. At the outset we determined that the levels of cell surface expression for IFNABR-␣ were comparable in A9ϩ and A9ϩ21 cells (Fig. 4). These data confirmed that the extent of cell surface expression for IFNABR-␣ was not a limiting factor in the A9ϩ transfectants. Next, we examined cells for IFNABR-␤ expression, and our results demonstrate that in contrast to the IFNABR-␣-containing A9ϩ cells, A9ϩ21 cells express IFNABR-␤ RNA (Fig. 8).
In subsequent studies we demonstrated that those additional factors encoded on chromosome 21 that are expressed in the A9ϩ21 cells are sufficient to enhance the antiviral effectiveness of IFN-Con 1 by 150-fold over that detected in the A9ϩ cells and to restore sensitivity to the antiviral effects of IFN-␣2b and IFN-␤ (Fig. 5). Moreover, the A9ϩ21 cells were fully responsive to the growth inhibitory effects of the different human IFN-␣s and IFN-␤ (Fig. 6). Whereas the growth inhibitory effects of murine IFN-␣ remained unchanged in the parental A9 and A9ϩ cells (ID 50 ϭ ϳ10 3 units/ml) due to the rapid doubling time of the A9ϩ21 cells (ϳ12 h), ϳ10 4 units/ml murine IFN-␣ are required to achieve an ID 50 in these cells (data not shown). Similarly, the effectiveness of the human IFN in the A9ϩ21 cells is apparently reduced by approximately 1 order of magnitude compared with T98G cells. Of note, our data indicate a correlation again between antiviral activity and ISGF3 activation, because we were able to restore ISGF3 activation in response to IFN-␣2b and IFN-␤ in the A9ϩ21 cells (Fig. 7).
The Interaction between IFN-Con 1 and IFNABR-␣ Is a Low Affinity Event-Because the level of cell surface expression of IFNABR-␣ on the murine A9ϩ cells does not appear to be a limiting factor, we examined the binding characteristics of 125 I-IFN-Con 1 on A9ϩ cells. Whether binding capacity was assessed at 4 or 37°C, we were unable to detect any specific high affinity binding, as determined by competition with 100fold excess of unlabeled ligand. Apparently, any interaction between IFN and the IFNABR-␣ on the surface of the A9ϩ cells must be a low affinity event that is undetectable in our standard binding reaction. We modified the experimental conditions of our binding studies by altering the incubation time with 125 I-IFN-␣ and increasing the ligand:cell ratio, but with identical outcome: no significant specific high affinity binding was

FIG. 2. Analysis of IFNABR-␣ and IFNABR-␤ expression on MRC-5 cells.
A-C, 125 I-IFN-Con 1 binding to receptors on MRC-5 and T98G cells. IFN-Con 1 , as a consensus IFN-␣, with a specific activity of 3 ϫ 10 9 units/mg protein, was used. The steady-state receptor binding characteristics were determined. 2. detected. In ligand-receptor cross-linking studies with 125 I-IFN-Con 1 we were able to detect trace amounts of IFN receptor complexes (data not shown). Comparison of the cross-linked complexes from A9ϩ and Daudi cells that were separated by polyacrylamide gel electrophoresis revealed identical bands. Similar results have been reported with other species of murine cells transfected with human IFNABR-␣ (44). In the final series of cell surface receptor binding experiments, we examined the relative binding characteristics of IFN-Con 1 , IFN-␣2b, and IFN-␤ for cell surface receptors on A9ϩ21 cells (Fig. 9). The results indicate that each of the Type 1 IFN species is able to recognize the receptors on these cells.
Our data would suggest that different Type 1 IFN, both ␣ and ␤, exhibit poor affinity for IFNABR-␣ that is not associated with an intact, functionally competent receptor complex; A9ϩ murine cells that lack human IFNABR-␤ do not bind these IFN with high affinity. Accordingly, we examined the ability of different IFN to bind to a fusion protein that comprised the extracellular portion of IFNABR-␣ linked to the Fc region of IgG1. The conformational integrity of the IFNABR-␣/Fc fusion protein is suggested by the ability of 2 distinct mAbs, which recognize native cell surface IFNABR-␣ in flow cytometry (51), to immunoprecipitate the fusion protein (data not shown). Differential recognition of the IFNABR-␣ fusion protein by these antibodies, together with the ability of the fusion protein to associate with a specific glycosphingolipid species normally associated with cell surface IFNABR-␣ (8), suggests that this fusion protein is conformationally similar to native cell surface IFNABR-␣. Using 125 I-IFN-Con 1 we were unable to identify specific, competable binding to the fusion protein to any significant extent, nor were we able to demonstrate that IFN-␣2b or IFN-␤ exhibited any appreciable affinity for this fusion protein (data not shown). Our approach was to incubate the fusion protein for 2 h with varying doses of 125 I-IFN-Con 1 at 4°C, as per our standard binding reaction, in the presence or the absence of a 100-fold excess of unlabeled IFN-Con 1 or varying doses of competitor IFN, IFN-␣2b, or IFN-␤. The amount of 125 I-IFN-Con 1 specifically bound to the fusion protein was then determined by either (i) conjugating the fusion protein to protein A-Sepharose and measuring bound cpm in the pellet fraction separated by centrifugation or by (ii) covalently crosslinking bound 125 I-IFN to the fusion protein with disuccinimidyl suberimidate, affinity purifying the fusion protein on protein A-Sepharose, and then visualizing the salteluted fraction by SDS-polyacrylamide gel electrophoresis. Failure to detect 125 I-IFN-Con 1 bound to the fusion protein by either method supports our findings that IFNABR-␣ alone does FIG. 3. STAT and Jak levels in MRC-5 cells. MRC-5 cells were either left untreated or treated with 10 ng/ml of IFN-Con 1 (con 1 ) or IFN-␤ (␤). T98G cells were similarly either left untreated or treated with 1 ng/ml of IFN-Con 1 or IFN-␤. A, 5 g of whole cell extracts prepared as described under "Experimental Procedures" were reacted with 30,000 cpm of a 32 P-end-labeled ISRE, representing nucleotides Ϫ107 to Ϫ87 of the human 2-5A synthetase gene, which contains a functional ISRE. Complexes were resolved by using native gel electrophoresis and visualized by autoradiography. Mobility of ISGF3 is indicated. Specific complexes were identified by the addition of 100-fold excess of unlabeled ISRE (ISRE) or mutant ISRE (mut ISRE) to the reaction. FP, free probe. B, 4 g of whole cell extracts from untreated MRC-5 and T98G cells were separated by gel electrophoresis, transferred to filters, and immunoblotted with monoclonal Abs to Stat1, Stat2, Stat3, Jak 1, and Tyk 2. Immunoreactive bands were visualized by the ECL Western blotting system. Stat1 exists as two isoforms, Stat2 exists as a single isoform, and Stat3 exists as three isoforms. MRC-5 cells constitutively express one isoform of Stat3, whereas T98G cells express two isoforms of Stat3. not constitute a binding receptor. Apparently contradictory reports that suggest that IFNABR-␣ is a binding receptor (52, 53) did not evaluate the binding capacity of the receptor chain alone.

DISCUSSION
Human IFNABR-␣ stably expressed in murine cells apparently confers sensitivity only to select human IFN-␣s, IFN-␣8 (6,44,54,55), and IFN-Con 1 (our data, 9,55). In contrast, a monoclonal antibody to human IFNABR-␣ inhibits the biologic activity of several species of human IFN-␣, IFN-␤, and IFN-in human cells (56), suggesting a structural heterogeneity of the IFNABR-␣ that would allow for differential affinities amongst the different Type 1 IFN, dependent on the presentation/accessibility of IFNABR-␣ on the cell type in question. The implications are that additional components influence the in- teraction between different Type 1 IFN and IFNABR-␣. Studies from this laboratory have identified a requirement for glycosphingolipid modification of IFNABR-␣ (8). The relation of IFNABR-␣ to IFNABR-␤ is unclear, yet accumulating evidence indicates that co-expression is essential for IFN-induced activation of Tyk 2 and Jak 1.
There is considerable sequence identity between the murine and human IFNABR-␣ peptides (and bovine IFNABR-␣) and between the murine and human Type 1 IFN, yet species specificity is determined at the level of receptor recognition. An earlier report from this laboratory suggested that within the Type 1 IFN molecule are two epitopes associated with receptor recognition: one that is conserved among the different species of IFN-␣s and IFN-␤s and the other that may be variable and associated with species specificity and separation of IFN-␣/␤ properties (57). It is likely that the different Type 1 IFN share a similar structural conformation and that the IFNABR-␣ peptide, when optimally configured at the cell surface in associa-tion with IFNABR-␤, is able to accommodate for the minimal structural variations. The preceding data would suggest that the origin of the cell type dictates the accessibility of IFNABR-␣ to the different Type 1 IFN. Interestingly, the IFNABR-␣ mAb that we employed for flow cytometry is able to distinguish human IFNABR-␣, regardless of the host cell, implying that the epitope(s) to which it is directed on the extracellular portion of IFNABR-␣ is unaffected by cell type yet is specific for human IFNABR-␣ alone. We infer that the IFN recognition epitope(s) on IFNABR-␣ are influenced by cell factors.
IFNABR-␣ expressed on human MRC-5 cells that also express IFNABR-␤ is able to interact with IFN-␣/␤ to invoke ISGF3 activation and an antiviral response, yet the limited binding capacity of MRC-5 cells and the low affinity ligandreceptor interaction apparently precludes their sensitivity to the growth inhibitory effects of the same IFN. The implications are that MRC-5 cells express functionally deficient Type 1 IFN receptors. We observed that a 10-fold increase in IFN-␣/␤ dose is required in MRC-5 cells to achieve a similar level of ISGF3 activation to that obtained in T98G cells and that there is a direct correlation between the extent of IFN-induced ISGF3 activation and the magnitude of an antiviral response. Using monoclonal antibodies directed against individual Jaks and STATs, we have shown that cytoplasmic extracts from MRC-5 cells express Jak 1, Tyk 2, Stat1, Stat2, and Stat3, thus we infer that none of these factors are limiting in the MRC-5 cells. Apparently, the receptor configuration in the MRC-5 cells precludes the appropriate IFN-induced activation of at least Stat1␣ and Stat2. Furthermore, our data indicate that ISGF3 activation mediated by IFN-␣/␤-induced activation of IFNABR-␣ and IFNABR-␤ does not necessarily invoke an antiproliferative response. When viewed solely in the context of the affinity characteristics of the ligand-receptor interaction, the implications are that the threshold of sensitivity to the antiviral effects of IFN is lower than that required to induce a growth inhibitory response. We have demonstrated that MRC-5 cells express comparable levels of cell surface IFNABR-␣ and IFNABR-␤ with Daudi (8) and T98G cells that are responsive to the growth inhibitory effects of Type 1 IFN. These observations suggest that accessory factors directly influence the binding capacity of cells for IFN ␣ and ␤, perhaps through an interaction with IFNABR-␣ and/or IFNABR-␤, thereby affecting signal transduction events that mediate an antiproliferative response. In a similar manner to the Jaks, the accessory component may contribute to the overall affinity of the IFN receptor complex through an association with an intracellular domain in either IFNABR-␣ and/or IFNABR-␤. The implications are that this accessory component mediates signal transduction that leads to a growth inhibitory response.
Our results with murine cells transfected with the human IFNABR-␣ indicate that the subtle structural variations among the different human IFN-␣s and IFN-␤ cannot be accommodated readily when human IFNABR-␣ is expressed on murine cells and, indeed, receptor recognition is dramatically restricted. Not sursprisingly, only the low affinity signaling pathway associated with an antiviral response is effected in the murine transfectants. Human IFNABR-␣, perhaps associated with murine specific components, will accommodate IFN-Con 1 . This may be a consequence of the inherent higher specific affinity of IFN-Con 1 for the Type 1 IFN receptor, per se, that allows recognition of IFNABR-␣ on the murine background. By contrast, IFN-␣8, which has been shown to be active on murine cells transfected with human IFNABR-␣ (6,52,55), is distinct from the other human IFN-␣ species in that region associated with species-specific receptor recognition (5, 57) and may indeed resemble a murine IFN in this region of the molecule. Full   FIG. 8. IFNABR-␤ expression in A9؉21 cells. A, using a PCRbased protocol, 2 ng of poly(A)ϩ RNA extracted from A9ϩ and A9ϩ21 cells were reverse transcribed, and the resultant cDNA probed for IFNABR-␤ using specific primers (7). Analysis of the PCR products revealed an expected 397-bp band in the A9ϩ21 and not the A9ϩ lane. B, Northern analysis of IFNABR expression in A9Ϫ, A9ϩ, and A9ϩ21 cells. 10 g of total RNA were separated by gel electrophoresis and transferred to filters that were probed for IFNABR-␤ by Northern hybridization. IFNABR-␤ cDNA was a 992-bp insert in plasmid PCR II. The different lanes were probed for ␤-actin (2-kilobase pair insert in pUC18) to assess loading of RNA.
FIG. 9. Competitive displacement of 125 I-IFN-Con 1 from A9؉21 cell surface receptors. 1.2 ϫ 10 5 cells were incubated at 4°C for 2 h with 10 ng/ml of 125 I-IFN-Con 1 containing no unlabeled competitor (100% bound) or the indicated concentrations of different IFN. The values shown were obtained by subtracting nonspecific cpm from total cpm bound. Nonspecific binding was determined in the presence of a 100-fold excess of unlabeled IFN. The points represent the means of triplicate cultures, and error bars denote the S.D. from the means. Ⅺ, IFN-Con 1 ; f, IFN-␤; E, IFN-␣2b. biological sensitivity was restored to IFNABR-␣ by complementation with chromosome 21-encoded factors that include IFNABR-␤. From our experimental results we infer that distinct signaling pathways invoke antiviral and growth inhibitory responses.
The challenge is to elaborate those regions on IFNABR-␣ and IFNABR-␤ that interact with the ligand IFN, that are possibly involved in receptor-receptor interactions, and that associate with accessory factors (e.g. Jaks) that are necessary for full ligand-receptor activation. Our data indicate that co-expression of IFNABR-␣ and IFNABR-␤ is not sufficient to invoke a complete biological response to IFN-␣/␤. Apparently, ISGF3 activation that leads to an antiviral response is mediated by IFN-␣/␤ activation of IFNABR-␣ and IFNABR-␤ in the context of a low affinity interaction, yet a high affinity interaction is necessary for signal transducing events that mediate a growth inhibitory response. Moreover, we confirm that IFNABR-␣ alone is not a binding receptor chain. The present study provides evidence that a fully productive Type 1 IFN transmembrane receptor complex is comprised of at least one other accessory factor encoded on chromosome 21 in addition to IFNABR-␣, IFNABR-␤, Tyk 2, and Jak 1.