Cloning and characterization of soluble and transmembrane isoforms of a novel component of the murine type I interferon receptor, IFNAR 2.

This report describes the cloning of cDNAs encoding transmembrane and soluble isoforms of a novel chain of the murine type I interferon (IFN) receptor and characterization of its capability to bind ligand and transduce signals. The transmembrane receptor (murine IFNAR 2c) has an extracellular domain of 215 amino acids and an intracellular domain of 250 amino acids, with 48% amino acid and 71% nucleotide identity with human IFNAR 2c. The cDNA for the soluble murine receptor (IFNAR 2a) encodes a 221-amino acid polypeptide identical to the first 210 amino acids of IFNAR 2c plus a novel 11 amino acids. Northern blot analyses show that murine IFNAR 2 is expressed as two transcripts of 4 kilobases encoding the transmembrane isoform and 1.5 kilobases encoding the more abundant soluble isoform. Studies using primary murine cells that lack IFNAR 1 show that IFNAR 2 is expressed, and cells bind type I IFN ligand, but do not transduce signals as detected by electrophoretic mobility shift assays of ISGF3 or GAF complexes binding to their cognate oligonucleotides. These cells show no effects on the ability of IFNgamma to activate these complexes. These studies demonstrate that the IFNAR 2 transmembrane (2c) and soluble (2a) isoforms are conserved between the human and mouse and that IFNAR 2c has intrinsic ligand binding activity, but no intrinsic signal transducing activity as measured in this study.

The type I interferons (IFNs) 1 are pleiotropic cytokines which, in all vertebrate species, can impart important signals to cells to protect against viral infection, inhibit proliferation, and activate immune effector cells (1). The human type I IFNs include multiple subtypes of IFN␣, a single IFN␤, and in some species IFN and - (2,3). The structure of type I IFNs is highly conserved, ranging from 70 -98% amino acid identity between IFN␣ subtypes to 35% identity between IFN␣ and IFN␤ (2). Not only are type I IFNs structurally and functionally related, but they also compete with each other for receptor binding and therefore share one or more common receptor component(s) (4). Although two components of the human type I IFN receptor have been cloned, their role in ligand binding and signal transduction remains unclear.
The first component of the type I IFN receptor to be cloned was human IFNAR 1 (5) which was shown to mediate response to one human IFN subtype ␣8, but not ␣2, nor ␤ when its cDNA was expressed in murine TG 9A cells (5). Subsequent studies demonstrated binding of type I IFNs to IFNAR 1 when expressed in Xenopus laevis oocytes (6) or in simian COS cells (7). By contrast, no binding of type I IFNs to IFNAR 1 was detected when IFNAR 1 was expressed in murine L929 cells (8) or hamster CHO cells (9,10). Thus, in view of these contradictory data, the role of IFNAR 1 in ligand binding remains uncertain. One possible explanation for the discrepancy between these studies may be that the human IFNAR 1 receptor component was expressed in various heterologous cell backgrounds which may either process human IFNAR 1 differently or may interact with the transfected component differently via its endogenous IFNAR chains. To define the function of receptor components clearly it would be an advantage to study the receptor function in a homologous background and without interference from other heterologous receptor components.
The cDNA encoding a second component of the human type I IFN receptor complex was cloned (11) and is now designated IFNAR 2. Subsequent studies have identified that the human IFNAR 2 gene encodes multiple mRNA transcripts which are translated into several isoforms: a soluble form designated as IFNAR 2a (11,12), a "short" transmembrane form designated as huIFNAR 2b (11,12) and a "long" transmembrane form designated as huIFNAR 2c (12,13). Although the original report proposed IFNAR 2b to be the signaling subunit (11), subsequent studies proposed that IFNAR 2c possessed signaling activity not present in IFNAR 2b (12). The reason for the existence of two transmembrane isoforms of IFNAR 2 remains unknown. Indeed the extent to which any isoforms of IFNAR 2 are expressed other than in a few cell lines derived from human tumors also remains unknown. Co-expression of human cDNAs encoding IFNAR 1 and IFNAR 2 in murine 3T3 cells gives higher binding of ligand than when either chain alone is expressed (14), but the contribution of endogenous receptor chains from the host cell as discussed above complicates interpretation of these experiments. Thus the importance of the multiple isoforms of human IFNAR 2 and their roles in ligand binding and signal transduction is still unclear.
To resolve these problems, we have undertaken to study the murine type I IFN system which is analogous to the human system in that it contains multiple IFN␣ subtypes and a single IFN␤ (2), and murine IFNAR 1 is homologous to human IFNAR 1 (15). The study of the IFN system in the mouse has several advantages. In particular the availability of transgenic and gene knock-out technology will enable functional studies on receptor function in vivo in physiological and pathological conditions. For example earlier studies by us (16) and others (17) of IFNAR 1 Ϫ/Ϫ mice demonstrated the importance of type I IFNs in antiviral defense and hemopoiesis. Furthermore the availability of cells from receptor knock-out mice will enable the study of one receptor component in cells where another component is absent. The comparison of data from the mouse and human system will also enable the identification of important isoforms or critical functional residues which are likely to be conserved across species.
In the present study we describe the cloning of two cDNAs encoding transmembrane and soluble isoforms of the second murine type I IFN receptor chain, IFNAR 2, which is highly homologous to the human IFNAR 2c and 2a isoforms, respectively. In addition, we report the conserved structural features of this receptor chain, demonstrate the expression of IFNAR 2 transcripts in normal adult mouse tissues, and demonstrate that when IFNAR 2 is expressed in primary cells in the absence of IFNAR 1 it has intrinsic binding activity, but alone does not transduce signals, as measured herein.

EXPERIMENTAL PROCEDURES
Cell Culture and Interferons-The murine interferons used in these studies were obtained from the following sources: recombinant murine IFN␤, Toray Industries, Tokyo, Japan; natural murine IFN␤ and mixed IFN␣ subtypes, Lee Biomolecular, Los Angeles, CA; recombinant murine IFN␣, Life Technologies, Inc.
Murine L cells were grown in RPMI 1640 containing 10% fetal calf serum, 100 units/ml penicillin, and 100 mg/ml streptomycin. Embryo fibroblast cell lines were grown in Dulbecco's modified Eagle's medium containing 10% fetal calf serum, penicillin, and streptomycin as above. BMM were produced by culture of bone marrow aspirates in RPMI 1640 supplemented with fetal calf serum, L-cell conditioned medium, and colony-stimulating factor 1 for 7-8 days as described previously (18).
Cloning of the MuIFNAR 2-A mouse testis cDNA library (CLON-TECH) was screened with a 757-bp huIFNAR 2 cDNA probe (nucleotides 410 to 1167 of published sequence (11). Two identical clones containing 1598 bp of insert were obtained after tertiary screens and sequenced using an automated sequencing machine (Applied Biosystems 373A DNA sequencer). The sequence of the insert corresponds to the sequence from nucleotides 622 to 2220 of human IFNAR 2a (Fig. 1). To obtain the rest of the sequence a 554-bp fragment corresponding to nucleotides 622 to 1176 was generated by PCR and used to screen a mouse lung cDNA library (Stratagene). After three rounds of screening, a clone containing an approximately 3-kb insert was obtained, and both strands of the insert were sequenced as above. The sequence (Fig. 1) contains a full open reading frame (513 amino acids) which included virtually the same sequence as the clone from the testis library.
An adult mouse liver cDNA library (Stratagene) was screened using a radiolabeled probe encoding the entire coding region of the transmembrane isoform of the muIFNAR 2. After three rounds of screening 14 positive clones were obtained, two of which encoded the transmembrane isoform of muIFNAR 2c. The remaining clones encoded an isoform of murine IFNAR 2 that is predicted to encode a soluble receptor.
Northern Blots-Poly(A) ϩ mRNA from cell lines, BMM, or organs was prepared by a modification of a published method (19). Poly(A) ϩ mRNA (3 g) was denatured with formamide, fractionated on a 1% agarose gel containing 0.67% formaldehyde, and electrophoresed in buffer containing 20 mM MOPS (pH 7.0) 5 mM sodium acetate, and 1 mM EDTA, then transferred to Hybond-C Extra (Amersham) membranes in 20 ϫ SSC according to the manufacturer's instructions. The membranes were hybridized with a 32 P-labeled murine IFNAR 2 cDNA, then stripped and rehybridized with a labeled 1.1-kb glyceraldehyde phosphate dehydrogenase cDNA probe as described previously (20).
Receptor Binding Assays-Murine IFN␤ (Toray Industries) was iodinated by incubation for 60 min with 125 I-labeled Bolton-Hunter reagent (ICN) (21). The antiviral activity of the labeled IFN was determined to be unchanged as tested in a standard cytopathic effectreduction bioassay using murine L cells and Semliki Forest virus essentially as described previously (22). Receptor binding assays were performed using murine L cells or adherent BMM by incubation of 125 I-IFN␤ with or without a 300-fold excess of unlabeled muIFN␤ for 2 h at 20°C essentially as described previously (20).
Electrophoretic Mobility Shift Assays (EMSA)-EMSA assays were performed on primary embryo fibroblasts obtained from IFNAR 1 homozygous normal (ϩ/ϩ), heterozygous (ϩ/Ϫ), or homozygous null mutant (Ϫ/Ϫ) mice (16). The cell lines were untreated or treated with recombinant muIFN␣ (100 IU/ml) or IFN␥ (500 IU/ml) for 1 h, nuclear lysates were prepared, and EMSA was performed using oligonucleotides from the ISG15 IFN-sensitive response element (ISRE) and/or the interferon response factor 1 ␥ activation sequence (GAS) element as described previously (23,24). To identify the components of bands on gel shifts, antibodies were incubated with the nuclear extracts at room temperature for 1 h prior to the addition of probe. The extracts contained no antibody, 3 g of normal rabbit immunoglobulin, 3 g of rabbit anti-human STAT-1 Ig (protein G-purified (25)), 1 g of rabbit anti-human STAT-2 Ig (n-17, Santa Cruz), 3 g of anti-murine STAT-3 Ig (C-20, Santa Cruz), or 3 l of rabbit anti-human serum.
Antiviral Assays-The antiviral activity of murine IFN␤ or IFN␣ subtypes (Lee Biomolecular, Los Angeles, CA) was tested on primary embryo fibroblast cell lines derived from muIFNAR 1 ϩ/ϩ, ϩ/Ϫ, or Ϫ/Ϫ mice by incubation with IFN concentrations ranging from 20 to 3,000 units/ml, overnight, then for 3 days with Semliki forest virus. Cell viability was determined by staining with the 3-(4,5-dimethylthiazol-2yl)-2,5-diphenyltetrazolium bromide dye (26). The murine IFNAR 2 shows 48% amino acid identity overall with the human IFNAR 2c and 71% nucleotide identity (Fig.  1B). A comparison of the murine and human IFNAR 2c amino acid sequences (Fig. 1B) shows that the cysteine residues are conserved between human and mouse except an extra C in muIFNAR 2 at amino acid residue 121. Intracellular tyrosine residues Tyr-268, Tyr-315, Tyr-317, Tyr-335, and Tyr-510 are conserved, but Tyr-306 and Tyr-411 are present in human not mouse, and Tyr-398 is present in mouse not human. Since functionally important residues are more likely to be conserved across species, the former tyrosines are likely to be the important ones, for example in phosphorylation associated with sig-nal transduction. The degree of amino acid sequence identity between the murine and human IFNAR 2 is similar to that between murine and human IFNAR 1, namely 48 and 46%, respectively.

Sequence of Murine
Several cDNA clones encoding a soluble isoform of IFNAR 2 were isolated from the libraries (12 of 14 clones analyzed), the sequence of a 1045-bp clone is shown in Fig. 2. This isoform is designated muIFNAR 2a because of its similarity with the soluble human isoform, huIFNAR 2a. The predicted amino acid sequence derived from the coding cDNA sequence is identical to the sequence of muIFNAR 2c for the first 237 amino acids (including the signal sequence) and contains an additional 11 amino acids which are specific to this isoform (Fig. 2).
Murine IFNAR 2 Expression-Northern blot analysis of mRNA from normal mouse liver and placenta using a fulllength murine IFNAR 2 cDNA probe indicates two major transcripts of approximately 4.0 and 1.5 kb in size (Fig. 3A), similar to the sizes reported for human IFNAR 2 transcripts. An identical filter was simultaneously hybridized with a probe encompassing nucleotides 917-1579 of the cDNA which encodes the cytoplasmic domain of murine IFNAR 2c. This cytoplasmic domain probe hybridized with only the larger 4.0-kb transcript. Hybridization with a probe that contained nucleotides 913 to 1012 of the IFNAR 2a cDNA detected only the 1.5-kb band. This probe contained nucleotide sequences that are absent in the IFNAR 2c cDNA, namely the 3Ј-untranslated region and the nucleotides encoding the soluble-specific amino acids.
These data indicate that the 1.5-kb transcript contains mRNA encoding the soluble muIFNAR 2a, whereas the 4.0-kb transcript encodes the transmembrane IFNAR 2c. This situation appears to be different from the human in which the soluble receptor isoform is encoded by a 4-kb transcript, whereas the shorter transcript encodes the short C-terminally truncated transmembrane isoform, IFNAR 2b (12). Interestingly, we are yet to isolate any murine cDNA clones homologous to the short C-terminally truncated transmembrane human IFNAR 2b isoform (from more than 20 clones from three different cDNA libraries). Because the human IFNAR 2b is proposed to have arisen because of an Alu sequence in the gene (27), the existence of this isoform may be specific to human species.
Northern blot analyses of poly(A) ϩ mRNA obtained from several organs of adult mice show that both the 4.0-and 1.5-kb transcripts are present in all tissues (Fig. 3B), as is also the case in the murine cell lines examined (data not shown). Occasionally a weak, larger transcript was observed at about 9 kb in some organs (e.g. liver, thymus, lung, Fig. 3B); this may represent unprocessed mRNA or another, minor transcript. It is notable from the Northern blots that the 1.5-kb transcript was more intense relative to the 4-kb transcript in most tissues analyzed, implying that the soluble receptor encoded by this transcript may be functionally important.
Function of Murine IFNAR 2 in Ligand Binding and Signal Transduction-To perform functional studies on murine IF-NAR 2, we have utilized the murine IFNAR 1 Ϫ/Ϫ mice previ- ously generated in our laboratory (16). As shown in Fig. 4A, BMM derived from IFNAR 1 ϩ/ϩ, ϩ/Ϫ, and Ϫ/Ϫ mice express both the 1.5-and 4.0-kb transcripts of murine IFNAR 2 at similar proportions to each other and relative to glyceraldehyde-3-phosphate dehydrogenase. Thus these cells provide an ideal opportunity to examine the function of IFNAR 2 in a primary cell line in the presence or absence of IFNAR 1.
While murine IFNs have proven difficult to radiolabel with 125 I, we have succeeded in labeling murine IFN␤ which maintained biological activity for a sufficient time to perform binding assays. This type I IFN bound specifically to BMM in a dose-dependent manner. However, the ability of BMM to bind 125 I-labeled murine IFN␤ is reduced in IFNAR 1 Ϫ/Ϫ cells to approximately one-half that in IFNAR 1 ϩ/ϩ cells (Fig. 4B). These data imply that murine IFNAR 1 contributes to ligand binding, but in its absence, IFNAR 2 is expressed and significant binding of ligand occurs.
Because primary BMM still bind type I IFN in the absence of IFNAR 1, and because the IFNAR 2 chain was still expressed, the question arises whether a signal can be transduced in response to the bound type I IFN in IFNAR 1 Ϫ/Ϫ cells? We have previously shown that primary embryo fibroblast (PEF) cell lines were sensitive to the antiviral responses of type I IFNs, so we have used these cell lines to examine signal transduction by testing for the formation of the transcriptional activation complexes, ISGF3 (IFN-stimulated gene factor) and GAF (␥ activated factor) following treatment with murine IFN␣ and as a control, IFN␥. EMSA studies show that IFN␣ induces the formation of an ISGF 3 complex in IFNAR 1 ϩ/ϩ PEF (Fig.  5A), but in IFNAR 1 Ϫ/Ϫ PEF there is no formation of the ISGF3 complex after induction by IFN␣ (Fig. 5B). It was surprising to find evidence of a low level of ISGF3 formation and binding to its cognate element after IFN␥ treatment of both IFNAR 1 ϩ/ϩ and Ϫ/Ϫ cells (Fig. 5, A and B). Nevertheless, this IFN␥-induced ISGF3 formation was reproducible, and its formation in IFNAR 1 Ϫ/Ϫ cells in response to IFN␥ discounts any indirect induction of ISGF3 via IFN␣ or -␤ production. The identity of the "shifted" bands in Fig. 5, A and B, were confirmed by preincubation with antibodies to STAT-1 and STAT-2 and subsequent reduction in the intensity of the ISGF 3 band (Fig. 5, C and D). Antibodies to human p48 also produced a minor reduction in intensity of the ISGF3 band formed in response to IFN␣ or -␥ (data not shown).
As expected, activation of GAF binding to the GAS element after induction by IFN␥ was similar in IFNAR 1 ϩ/ϩ and IFNAR 1 Ϫ/Ϫ cells (Fig. 6, A and B). By contrast, the induction of GAF in response to IFN␣ was observed in IFNAR ϩ/ϩ PEF (Fig. 6A), but not in IFNAR 1 Ϫ/Ϫ cells (Fig. 6B). The multiple bands of GAF represent STAT-1 and STAT-3 homo-and heterodimers as shown by preincubation of extracts with antibodies to these factors (Fig. 6, C and D). The specificity of the binding of the transcription factor complexes is demonstrated by the competitive inhibition by an excess of unlabeled oligonucleotide. These results demonstrate that IFNAR 2 alone (in the absence of IFNAR 1) is unable to transduce the signal(s) required to activate the formation of ISGF3 or GAF, even though it is sufficient to mediate significant binding of the type I IFN ligand. DISCUSSION The results provided herein detail the cloning and characterization of the transmembrane and soluble forms of the murine type I IFN receptor which shows significant homology with the human receptor, IFNAR 2. The conservation of important structural motifs in the extracellular domain such as the num- ber and location of cysteine residues probably indicates their importance in forming intramolecular disulfide bonds to maintain the appropriate folding required for ligand interaction. This similarity in architecture of murine IFNAR 2 places it in the class 2 family of cytokines, along with all other IFN receptor chains. In the intracellular domain of IFNAR 2 there are several putative sites for interaction with candidate signaling molecules, but the definition of these will require extensive studies using mutation of potential active sites. Among these are 5 tyrosine residues which are conserved with the human IFNAR 2 isoform, since these are potential targets for phosphorylation by tyrosine kinases and docking of signaling molecules such as JAK kinases and STAT proteins (28). The species conservation highlights these residues as the most likely to be functionally important. Furthermore the conservation of a soluble form of this receptor suggests that it may also have an important biological function.
For many cytokine receptors it has been demonstrated that two or three chains constitute the functional receptor. These chains have been defined as primary ligand binding (or ␣) chains, accessory (non-binding or ␤), or signal transducing or co-binding chains (29). This definition has been an important step in understanding how the receptor works in ligand interaction and signal transduction and in defining the action of soluble receptors (e.g. whether they inhibit interaction with a functional receptor or facilitate interaction by presenting the ligand to an accessory or transducing chain). However, as yet there has been no convincing definition of the role of the receptor chains in the type I IFN system. The availability of IFNAR 1 Ϫ/Ϫ mice from our previous work provides an ideal opportunity to study the function of IFNAR 2 in primary cells lacking the other component. In this study we have shown that in primary bone marrow-derived macrophages which lack IFNAR 1, but express IFNAR 2, there is significant binding of ligand. These data indicate for the first time, unequivocally, the intrinsic ligand binding capability of IFNAR 2. The observation that binding is reduced in IFNAR 1 Ϫ/Ϫ cells relative to IFNAR 1 ϩ/ϩ cells indicates clearly that IFNAR 1 also contributes to binding. This result also clarifies the contradictory data from human studies of the role of IFNAR 1 in ligand binding (see introduction).
Our studies show that despite this intrinsic ligand binding activity, IFNAR 2 cannot transduce signals in the absence of IFNAR 1. Co-expression of both receptor chains is therefore necessary for these signals to be transduced. We demonstrated that IFN␣ activated the formation of the classical ISGF3 complex as well as GAS-binding factors that contained STAT-1 and STAT-3. However none of these activated factors were observed in IFNAR 1 Ϫ/Ϫ cells despite IFN binding to these cells. IFN␥ signaling was intact in PEF cells independent of the presence or absence of IFNAR 1. It was particularly noteworthy that IFN␥ signaling was demonstrated not only by the conventional activation of GAS-binding factors but also by activation of ISGF3, the identity of which was demonstrated and confirmed with appropriate antibodies. This is the first demonstration that ISGF3 can be activated in response to IFN␥, and its presence in IFNAR 1 Ϫ/Ϫ cells rules out the possibility that it could be due to activation of IFN␣ signaling.
The existence of two isoforms of murine IFNAR 2 is different FIG. 4. A, Northern blot of poly(A) ϩ mRNA from BMM of mice which were either normal (ϩ/ϩ), heterozygous (ϩ/Ϫ), or homozygous (Ϫ/Ϫ) for a null mutation in the IFNAR 1 gene. The blots were hybridized with a cDNA probe for IFNAR 2 then glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as a control for loading. B, receptor binding curve for 125 I-labeled IFN␤ specifically bound to BMM isolated from IFNAR 1 ϩ/ϩ (ࡗ) or Ϫ/Ϫ (Ⅺ) mice. The data are presented as the mean (Ϯ SD, n ϭ 3) moles of IFN bound per cell versus moles of IFN added.
FIG. 5. Electrophoretic mobility shift assays of cell lysates prepared from PEF cells, before (؊) and after (؉) treatment with 1000 IU/ml of recombinant IFN␣ or IFN␥. Incubations of lysate and labeled ISRE oligonucleotide were performed in the absence (Ϫ) or presence (ϩ) of unlabeled oligonucleotide (ISRE oligo). The band representing the ISGF3 complex is indicated with an arrow. A, cells from IFNAR 1 ϩ/ϩ mice. B, cells from IFNAR 1 Ϫ/Ϫ mice. C, lysates from cells from IFNAR 1 ϩ/ϩ mice treated with antibodies to STAT-1 or STAT-2 or control antiserum (n) prior to gel shift. D, lysates from cells from IFNAR 1 Ϫ/Ϫ mice preincubated with antibodies as above.
from the situation in humans where there are three isoforms including abundant transcripts of a short transmembrane isoform, IFNAR 2b. The screening of several murine cDNA libraries has so far failed to detect a murine IFNAR 2b, 2 so the existence of this isoform may be restricted to the human where this transcript arises because of an Alu sequence in the human gene (27). Thus the murine mRNA transcripts of IFNAR 2 appear to be different from those of human, in which case the IFNAR 2c is encoded by a large (approximately 4.0 kb) and a small (approximately 1.7 kb) transcript, IFNAR 2a is encoded by a 4.0-kb transcript, and IFNAR 2b is encoded by the smaller transcript (12). However it should also be noted that the results for human IFNAR 2 expression were obtained using a tumorderived cell line, whereas our current results represent the first studies of IFNAR 2 expression in normal adult organs. Our data also indicate that the shorter transcript is the more abundant of the two in most tissues analyzed. Confirmation of similar differences in expression for the different IFNAR 2 isoforms at the protein level would suggest an important role for the soluble receptor in regulating the IFN response. The likelihood that a soluble isoform has biological activity is strengthened by our finding that the transmembrane isoform of IFNAR 2 has intrinsic ligand binding activity, thereby enhancing the probability that the soluble isoform will also bind ligand.
Thus this study demonstrates the sequence and important characteristics of murine IFNAR 2 as having a major role in binding type I IFN ligand but requiring the cooperation of IFNAR 1 for complete binding and signal transduction. These results will form the basis for significant future studies on understanding the regulation of the type I IFN system in vivo.