The type I interferon receptor: structure, function, and evolution of a family business.

Recent results indicate that coherent models of how multiple interferons (IFN) are recognized and signal selectively through a common receptor are now feasible. A proposal is made that the IFN receptor, with its subunits IFNAR-1 and IFNAR-2, presents two separate ligand binding sites, and this double structure is both necessary and sufficient to ensure that the different IFN are recognized and can act selectively. The key feature is the duplication of the extracellular domain of the IFNAR-1 subunit and the configurational geometry that this imposes on the intracellular domains of the receptor subunits and their associated tyrosine kinases.

Shared receptor components but distinct complexes for alpha and beta interferons.

The type I interferon family includes 13 alpha, one omega and one beta subtypes recognized by a complex containing the receptor subunits ifnar1 and ifnar2 and their associated Janus tyrosine kinases, Tyk2 and Jak1. To investigate the reported differences in the way that alpha and beta interferons signal through the receptor, we introduced alanine-substitutions in the ifnar2 extracellular domain, and expressed the mutants in U5A cells, lacking endogenous ifnar2. A selection, designed to recover mutants that responded preferentially to alpha or beta interferon yielded three groups: I, neutral; II, sensitive to alpha interferon, partially resistant to beta interferon; III, resistant to alpha interferon, partially sensitive to beta interferon. A mutant clone, TMK, fully resistant to alpha interferon with good sensitivity to beta interferon, was characterized in detail and compared with U5A cells complemented with wild-type ifnar2 and also with Tyk2-deficient 11.1 cells, which exhibit a similar alpha-unresponsive phenotype with a partial beta interferon response. Using anti-receptor antibodies and mutant forms of beta interferon, three distinct modes of ligand interaction could be discerned: (i) alpha interferon with ifnar1 and ifnar2; (ii) beta interferon with ifnar1 and ifnar2; (iii) beta interferon with ifnar2 alone. We conclude that alpha and beta interferons signal differently through their receptors because the two ligand subtypes interact with the receptor subunits ifnar 1 and ifnar2 in entirely different ways.

Interferon-alpha-dependent activation of Tyk2 requires phosphorylation of positive regulatory tyrosines by another kinase.

Tyk2 and JAK1, members of the Janus kinase (JAK) family of protein tyrosine kinases, are required for interferon-alpha/beta binding and signaling. Both enzymes are associated with the interferon-alpha/beta receptor, and upon ligand binding, they undergo tyrosine phosphorylation and catalytic activation in an interdependent manner. To identify residues involved in Tyk2 regulation and to understand the basis of the interdependence of Tyk2 and JAK1, six mutated versions of Tyk2 bearing single or multiple point mutations in the tyrosine kinase domain were studied in a cell line lacking endogenous Tyk2. The Y1054F/Y1055F substitutions in the putative activation loop prevented ligand-dependent activation of Tyk2, without abolishing its catalytic potential. The K930R mutation in the ATP binding site generated a kinase-negative protein, which however, still became phosphorylated upon interferon-alpha treatment. The Y1054F/Y1055F substitutions in this kinase-negative Tyk2 abolished the induced phosphorylation. These results indicate that Tyk2 is activated by phosphorylation on Tyr-1054 and/or Tyr-1055 and that this phosphorylation requires another kinase, most likely JAK1. While the Tyk2 forms mutated on Tyr-1054 and Tyr-1055 or on Lys-930 allowed some inducible gene expression, the combination of the three point mutations totally abolished signaling.

Mutant U5A cells are complemented by an interferon-alpha beta receptor subunit generated by alternative processing of a new member of a cytokine receptor gene cluster.

The cellular receptor for the alpha/beta interferons contains at least two components that interact with interferon. The ifnar1 component is well characterized and a putative ifnar2 cDNA has recently been identified. We have cloned the gene for ifnar2 and show that it produces four different transcripts encoding three different polypeptides that are generated by exon skipping, alternative splicing and differential use of polyadenylation sites. One polypeptide is likely to be secreted and two are transmembrane proteins with identical extracellular and transmembrane domains but divergent cytoplasmic tails of 67 and 251 amino acids. A mutant cell line U5A, completely defective in IFN-alpha beta binding and response, has been isolated and characterized. Expression in U5A cells of the polypeptide with the long cytoplasmic domain reconstitutes a functional receptor that restores normal interferon binding, activation of the JAK/STAT signal transduction pathway, interferon-inducible gene expression and antiviral response. The IFNAR2 gene maps at 0.5 kb from the CRFB4 gene, establishing that together IFNAR2, CRFB4, IFNAR1 and AF1 form a cluster of class II cytokine receptor genes on human chromosome 21.

Distinct domains of the protein tyrosine kinase tyk2 required for binding of interferon-alpha/beta and for signal transduction.

tyk2 belongs to the JAK family of nonreceptor protein tyrosine kinases recently found implicated in signaling through a large number of cytokine receptors. These proteins are characterized by a large amino-terminal region and two tandemly arranged kinase domains, a kinase-like and a tyrosine kinase domain. Genetic and biochemical evidence supports the requirement for tyk2 in interferon-alpha/beta binding and signaling. To study the role of the distinct domains of tyk2, constructs lacking one or both kinase domains were stably transfected in recipient cells lacking the endogenous protein. Removal of either or both kinase domains resulted in loss of the in vitro kinase activity. The mutant form truncated of the tyrosine kinase domain was found to reconstitute binding of interferon-alpha 8 and partial signaling. While no contribution of this protein toward interferon-beta binding was evident, increased signaling could be measured. The mutant form lacking both kinase domains did not exhibit any detectable activity. Altogether, these results show that a sequential deletion of domains engenders a sequential loss of function and that the different domains of tyk2 have distinct functions, all essential for full interferon-alpha and -beta binding and signaling.

Domains of interaction between alpha interferon and its receptor components.

We describe how constraints on the binding of human interferons (IFNs), alpha1 and alpha2 and alpha8 on mouse cells are partially relieved by the expression of the bovine (Bo) or human (Hu) IFN alpha/beta receptor (IFNAR) component in these cells. We show that, while the binding of all three is substantially increased by the transfection of Bo IFNAR, it is accompanied by an increase in activity only in the case of alpha2 and alpha8 (IFNs that otherwise have little activity on mouse cells). IFN alpha1, which shows some partial activity on mouse cells, responds to the presence of Bo IFNAR by acting, at low concentrations, as a competitive antagonist to IFNs alpha2 and alpha8. A review of published results on IFN hybrid scanning and on the effects of expressing Bo IFNAR in human cells led us to propose that an N-terminal segment of the IFN molecule interacts directly with IFNAR. Applying site-directed mutagenesis to an IFN hybrid; alpha8[60]alpha1[92]alpha8, we show that the point mutations K84 to E84 and Y90 to D90 act synergistically to cause the hybrid to behave as the parental IFN alpha8, switching the preference from Mu to Hu IFNAR in transfected mouse cells. The published structural models for IFN reveal that positions 84 and 90 span the exposed residues of the alpha-helix C of the IFN molecule. We derive a model of IFN-receptor interaction in which the A helix and the C helix of IFN interact with IFNAR and in which a binding phase can be distinguished from a binding/activity phase. We propose that the so-called "hot" domains of the IFN molecule (the AB loop and the D helix) are presented by IFNAR to interact with an additional component of the functional receptor.

Specific antiviral activities of the human alpha interferons are determined at the level of receptor (IFNAR) structure.

Differences in activity among the family of human IFNs alpha are much reduced if these ligands are assayed on bovine cells. In particular, the activity of IFN alpha D is much higher on bovine than on human cells. To examine these differences, the bovine counterpart of the human IFNAR has been cloned and expressed in a human cell line. The transfected cell line now recognizes the human IFN alpha D as a high-specific-activity IFN subtype, indicating that the differences in sensitivity between the bovine and human cells to the human IFN alpha lie in the structure of the IFNAR chain rather than in the other components of the functional receptor.

Behavior of a cloned murine interferon alpha/beta receptor expressed in homospecific or heterospecific background.

A murine interferon (IFN) alpha/beta receptor was cloned from the IFN-sensitive L1210 cell line on the basis of its homology with the human receptor. A combination of methods that includes the screening of random-primed and oligo(dT)-primed cDNA libraries and polymerase chain reactions with a single-side specificity was used. At the amino acid level, the murine IFN-alpha/beta shows 46% identity with its human counterpart. Both human WISH cells presenting a low sensitivity to mouse IFN and a murine L1210 mutant subline that does not express the receptor have been stably transfected with the murine IFN-alpha/beta receptor. Whereas transfected human cells became sensitive to a limited number of mouse IFN-alpha/beta subtypes, the transfected murine L1210 mutant was found to be fully complemented and became sensitive to all mouse IFN-alpha/beta subtypes tested, including those that were not active on transfected human cells. These results strongly suggest that the receptor described here is implicated in the mediation of the activities of all murine IFN-alpha/beta subtypes.

Assignment of the human interferon-alpha receptor gene to chromosome 21q22.1 by in situ hybridization.

The human interferon-alpha receptor gene (IFN AR) has been assigned to the long arm of human chromosome 21 (report of the committee on the genetic constitution of chromosomes 20 and 21; Ref 1). The present report confirms the assignment and refines the mapping to the 21q22.1 band, using a cDNA probe for the human IFN AR gene and in situ hybridization to metaphase chromosomes.

The cellular receptor of the alpha-beta interferons.

This is a selective review of recent trends in research on the cellular receptor for the alpha-beta interferons. It deals mainly with work published in the last three years (1985-88), and therefore mainly with receptors for the human interferons. The binding characteristics of several human alpha interferons are examined, and the importance of in vitro experimental models for establishing the relationship between receptor binding and the cellular response is emphasized.

Estimates of normal binding of a human recombinant alpha interferon to peripheral blood mononuclear cells from a study matching healthy subjects to subjects with insulin dependent diabetes.

This study compares the binding of a human recombinant alpha-interferon to peripheral blood mononuclear cells (PBMC) from patients with insulin dependent diabetes (IDDM) mellitus and control subjects. Diurnal and longer term of variation, feeding, fasting and haemoglobin glycosylation were examinated for their influence on interferon binding to PBMC. No gross differences in binding were demonstrated, in particular no effect of glucose levels was seen on the binding of interferon alpha-2 to PBMC.

Interferon removes its own receptors as it blocks the division of Daudi cells.

The Burkitt-derived line, Daudi, whose proliferation is inhibited by human alpha-interferon (IFN-alpha), was treated with 125I-labelled recombinant human IFN-alpha A. After separation from unbound ligand, cell-bound IFN was extracted with the detergent digitonin yielding soluble and insoluble complexes of IFN and receptor, together with a certain amount of uncomplexed IFN. 1. Soluble complexes were stable enough to be separated from uncomplexed IFN by permeation chromatography. Treatment of soluble complexes with the bifunctional reagent, disuccinimidyl suberate, yielded a radioactive product separating with an Mr of 130,000 on electrophoresis in sodium dodecyl sulphate. Similar complexes could be recovered with sodium dodecyl sulphate from the digitonin-insoluble residue, treated with the bi-functional reagent. 2. The total (soluble and insoluble) of complexed IFN obtained after digitonin extraction was a constant fraction (0.62) of the total cell-bound radioactivity, being independent of the concentration of IFN added to the cells (less than pM to greater than nM), and of the time of incubation (1 min to 20 h). However, between 30 min and 3 h of incubation, the insoluble complex increased, at the expense of the soluble complex, and there appeared a cellular pool of degraded ligand. From 3 h to 20 h the distribution of ligand-derived radioactivity remained constant while the total amount decreased to less than 10% of its value at 30 min. This decrease in binding was matched by the appearance of an equivalent quantity of radiolabelled fragments in the culture medium. 3. The inhibition of cellular division due to IFN was shown to be coincident with the disappearance of cellular binding and with the cell-mediated degradation of receptor-complexed IFN. We propose that IFN removes its own receptor and, in doing so, blocks a linked function necessary for the stimulated growth of Daudi cells.

Electrostatic interactions in the cellular dynamics of the interferon-receptor complex.

Using membrane preparations of the interferon receptor, prepared from cells of the Burkitt line, Daudi, we have examined the binding of three human recombinant alpha-interferons. 1. We discovered a binding titration of the interferons IFN-alpha A and IFN-alpha D in the pH range 6-9. Receptor binding, negligible at pH 6, rises to a maximum close to pH 9. We have shown that binding of IFN-alpha A at basic pH is to the same receptors as at neutrality and that IFN-receptor complexes extracted with digitonin are more stable at basic pH than they are at neutrality. 2. The recombinant interferon, IFN-alpha B, shows little change of binding in the pH range 6-9. At its basic optimum the binding of IFN-alpha A approaches that of IFN-alpha B, while at neutral pH the binding of IFN-alpha A is 3-4 times less. This difference at neutral pH is seen on intact cells as well as on membrane preparations. The specific activity of IFN-alpha B is close to that of IFN-alpha A, both of which are 10-20 times more active than IFN-alpha D; and the binding titration is, therefore, independent of the initial binding affinities. 3. Using hybrid IFNs constructed from the DNA sequences of alpha D and alpha B, we have isolated the sequence responsible for the binding titration to the segment comprising amino acids 61-92. Examination of these sequences reveals that Lys-84 is present in all the IFN-alpha except IFN-alpha B where it is replaced by Glu; and Tyr-90, present in most of the common IFN-alpha including alpha A and alpha D, is replaced by Asp in IFN-alpha B. Lys and Tyr would normally titrate in the pH range 6-9. We conclude that the binding titration is due to an electrostatic interaction and we propose that the interaction is between IFN-receptor complexes. The role of the interaction in the binding losses that accompany the antiproliferative effects of IFN is discussed.

[Immunization against interferon in man]. / Immunisation anti-interféron chez l'homme.

We have summarized 14 human cases with antibodies to interferon (anti-IFN) reported in the literature. In 4 cases, the patients developed auto-antibodies without any injection of exogenous interferon (IFN). In most cases, antibodies were directed against alpha IFN (13/14) and especially against recombinant alpha 2 IFN (8/14). When tested, antibodies to IFN were always IgG. In a few subjects, the IFN system has been studied more completely with the following results: the competent cells are able, after induction, to produce IFN; the endogenous IFN is neutralized by the antibodies to IFN; the antibodies to IFN are able to inhibit the interaction of IFN with its specific cellular receptors. Clinically, the raising of an anti-IFN immunization does not seem to impair the antiviral or antitumoral effects of IFN. These data can be related to the relative low neutralizing titres of antibodies to IFN with regard to levels of IFN produced at the site of the viral infection. We discuss the putative benefit of antibodies on the availability of circulating IFN and the possible role of antibodies to IFN in autoimmune diseases.

Biological activities and receptor binding of two human recombinant interferons and their hybrids.

Two human recombinant lymphoblastoid interferon-alpha subtypes, LyIFN-B (alpha 8) and LyIFN-D (alpha 1), and 10 hybrids generated therefrom were produced in Escherichia coli and purified. The antiviral and antiproliferative activities and the induction of (2'-5')oligoadenylate synthetase were compared to their receptor binding affinities. The IFN subtypes and their hybrids had similar specific antiviral activities on bovine cells. On human cells both the specific antiviral and antiproliferative activities of LyIFN-B were about 30-fold higher than those of LyIFN-D. This difference in activity could be attributed partly to the N-terminal amino acids 1 to 60 and partly to amino acids 61 to 92. A third domain affecting the biological activities was found within the carboxy-proximal segment from amino acids 93 to 150. The differences in these activities were found to correlate with their ability to bind the receptor, suggesting that the differences in activity might be due to altered binding of the IFNs to the cellular receptors. In contrast, the induction of (2'-5')oligoadenylate synthetase did not follow the same activity profile. On mouse cells, the efficiency of the hybrids was affected by at least four sites on the IFN protein. A hybrid with the N-terminal segment 1 to 60 from IFN-B and amino acids 61 to 166 from IFN-D had a specific antiviral activity on mouse cells as high as on human cells corresponding to a 500- and 5000-fold increase in specific activity compared to IFN-D and IFN-B, respectively. We suggest that on mouse cells the IFN activity may be more dependent on conformational differences than on human cells, which in turn might reflect a less precise fit to the mouse receptor than to the human receptor.

Effects of long-term treatment of human carcinoma cells with interferon alpha.

Three tumor cell lines derived from human colon, urinary bladder and ovarian carcinoma were serially passaged in the continuous presence of human interferon alpha for extended periods of time. Phenotypic changes induced by interferon differed among these three cell lines. Thus interferon enhanced colon tumor cell aggregation but inhibited the aggregation of bladder tumor cells. The antiproliferative activity of interferon was more pronounced in bladder and ovarian cells than in colonic cells. However, the tumorigenicity of parental and cloned colon tumor cells injected i.p. or s.c. was markedly reduced by passage of the cells with interferon. Interferon treatment reduced the tumorigenicity of ovarian tumor cells when these cells were injected i.p. but not when injected s.c. The tumorigenicity of bladder tumor cells was not affected by interferon.