A gene on human chromosome 21 located in the region 21q22.2 to 21q22.3 encodes a factor necessary for signal transduction and antiviral response to type I interferons.

The type I interferons (IFNs) are a family of multifunctional cytokines which includes the 15 IFN alpha subtypes and IFN beta. These IFNs compete for binding to cell surface receptors. However, murine cells transfected with a cDNA for a human IFN alpha receptor (IFNAR) developed an antiviral response only to human IFN alpha B, but not to human IFN alpha 2 nor -beta(1). In this study we show, using a panel of CHO-human chromosome 21 hybrid cell lines which all express IFNAR, that only those containing the region 21q22.2 to 21q22.3 transduce signals for IFN responses. Two such hybrid cell lines responded to IFNs alpha 2, -alpha B and -beta by induction of 2'-5' oligoadenylate synthetase and resistance to viral infection. Other hybrid cell lines, that lacked the region 21q22.2-3, failed to transduce signals as above; even though they expressed IFNAR and bound human IFN alpha 2, -alpha B, and -beta. These data demonstrate that a gene(s) located in the region 21q22.2-3 encodes a factor(s) which is necessary for signaling but does not influence ligand binding. This factor is not the cofactor required for IFN gamma signaling which is located in the region 21p to 21q22.1(2).

The type I interferons (IFNs) are a family of multifunctional cytokines which includes the 15 IFNa subtypes and IFNp. These IFNs compete for binding to cell surface receptors. However, murine cells transfected with a cDNA for a human IFNa receptor (IFNAR) developed an antiviral response only to human IFNcUB, but not to human IFNa2 nor -/3 (1). In this study we show, using a panel of CHO-human chromosome 21 hybrid cell lines which all express IFNAR, that only those containing the region 21q22.2 to 21q22.3 transduce signals for IFN responses. Two such hybrid cell lines responded to IFNsd, -aB and -p by induction of 2'4' oligoadenylate synthetase and resistance to viral infection. Other hybrid cell lines, that lacked the region 21q22.2-3, failed to transduce signals as above; even though they expressed IFNAR and bound human IFNa2, -aB, and -6. These data demonstrate that a gene(s) located in the region 21q22.2-3 encodes a factods) which is necessary for signaling but does not influence ligand binding. This factor is not the cofactor required for IFNy signaling which is located in the region 21p to 21q22.1 (2).
The interferons (IFNs)' are a large family of cytokines whose functions include the protection of cells against viral infection, the regulation of cell proliferation and differentiation, and activation of cells (such as NK cells and macrophages) (3). The type I IFNs include the 15 subtypes of IFNa, one IFNP, two IFNo (and several trophoblast proteins in some species) (4, 5). This classification is based on similarities in structure (4) and biological activities (31, a shared receptor or receptor component (6), and induction in response to virus and location of the genes on chromosome 9, (3). Type I1 IFN, or IFNy, differs from Type I IFNs in all of the above respects, except that some overlap in biological function occurs. The structures of the type I IFNs are conserved with 75-95% amino acid identity between IFNa subtypes, 35% identity between IFNa and -P, and 50-65% identity between IFNa and the remaining type I IFNs (4). The structural differences among the type I IFNs are reflected in functional differences, for example in antigenicity (7) and in biological specific activities which can vary by up to 1000-fold * This study was funded by the National Health and Medical Research Council ofAustralia. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ To whom correspondence should be addressed. between subtypes (8). Receptor binding has been shown to be necessary, but not suficient, for the biological activities of type I IFNs (3,6). Post-receptor binding events such as signal transduction and induction of IFN response genes are also necessary for biological responses. Many type I IFN-induced genes have been identified, and some, in particular those encoding the double-stranded RNAdependent 2'-5' oligoadenylate synthetase, p68 kinase, and Mx proteins have been assigned a role in antiviral andor antiproliferative effects of IFNs (9). Recently, early events in the induction of certain IFN-sensitive genes have been characterized. The protein kinases tyk-2 (10) and  are involved in signal transduction by type I IFNs and a transcription factor complex, termed ISGF3a, is activated by phosphorylation as early as 2 min after receptor binding of IFNa (12,13). However the precise mechanisms by which IFN ligand-receptor interaction transduces signals to activate the intracellular pathways for antiviral or antiproliferative effects remain to be elucidated.
Most of the type I IFNs have been presumed to bind to a common cell surface receptor component based on their ability to compete with each other for receptor binding (6). Furthermore the gene encoding the receptor (or a component of the receptor) for the type I IFNs has been shown to be localized on human chromosome 21 in the region 21q22.1 (1,6,14). The cDNA encoding IFNAR has been isolated (1). Unexpectedly, murine cells transfected with IFNAR developed an antiviral response only to human IFNaB, but not to IFNa2 nor IFNP (1).
Since receptor binding data were not presented for IFNa2 or -P it, remains unclear whether the failure to elicit an antiviral response in these transfected cells by these two ligands was due to an inability t o bind the cloned receptor component or whether another factor(s) is required to transduce signals.
In the present study we have examined the receptor binding and biological responses of several human type I IFNs. We have used a panel of CHO-human chromosome 21 hybrid cell lines which contain different portions of human chromosome 21 to examine (i) the binding of IFNs and (ii) the biological consequence of such binding as manifest by the induction of 2 ' 4 ' oligoadenylate synthetase activity and the development of antiviral responses. Our studies show that a gene located on the distal portion of human chromosome 21 in the region q22.2 to q22.3 encodes a factor which does not affect ligand-receptor binding, but is necessary for signal transduction by type I IFNs. Furthermore this factor is distinct from a signaling factor required for responses to type I1 IFNy, the gene for which maps to a different part of chromosome 21.
Northern Blots-Cells were grown to mid-log phase, harvested, and RNA extracted using guanidinium isothiocyanate (16). Approximately 30 pg of RNA in 50% formamide were run on 1% agarose gels, transferred to Hybond-N, hybridized with a 32P-labeled 1.2-kilobase pair NcoI fragment of the human IFNAR cDNA a t 55 "C overnight and washed at 65 "C, 0.1 x SSPE, 0.1% SDS. As a control for RNA loading, filters were stripped and rehybridized with a "P-labeled 1.1-kilobase pair glyceraldehyde phosphate dehydrogenase cDNA probe (17).
Receptor Binding Stuclies"IFNa2a and IFNaB were iodinated using chloramine T as described previously (18) to a radioactive specific activity of approximately 50 pCi/pg of protein. IFNP,, was iodinated also using chloramine T (19) to a radioactive specific activity of approximately 100 pCi/pg of protein. Integrity of the iodinated IFNs was monitored by comparative antiviral and receptor binding activities relative to unlabeled IFNs and was shown to be unaffected for the period of use. Binding assays were performed using monolayers of CHO or hybrid cell lines in six-well dishes containing 1-3 x lo6 cells/well. Duplicate wells were incubated with each of a range of concentrations of 1251-labeled IFN in 0.5 ml of binding medium (Dulbecco's modified Eagle's medium containing 10% fetal calf serum, 50 mM Hepes, 1.0 mM CaCl,, pH 7.4) with and without excess unlabeled IFN (100 x for IFNa2a and -aB or 1000 x for IFN p, , as appropriate) to determine nonspecific binding.
The plates were incubated a t 20 "C for 2 h with moderate shaking. The incubation mixture was then removed and the monolayers washed four times with cold binding medium. The cells were harvested and counted in a y-counter. Scatchard analyses of the binding curves were performed using the LIGAND program. In all cases where significant binding was detected, a "one-site" fit was statistically significant ( p < 0.05) and a "two-site" tit was not (20). The binding of 12sI-labeled IFNa2a and IFNaB to the CHO cell line was so low that it could not be resolved by a Scatchard plot. Binding curves and Scatchard plots were produced using the INPLOT program. Antiviral Responses-The hybrid cell lines and CHO-K1 were each plated into wells of a 96-well microtiter plate so as to form confluent monolayers (approximately 1 x lo5 cells/ml). IFNs were added to four wells and serially diluted in Dulbecco's modified Eagle's medium, 3% fetal calf serum, penicillin, and streptomycin down the plate in 2-fold steps, to give a range of IFN activity from approximately 20 to 3000 IU/ml. Cell monolayers were incubated with IFN overnight, supernatants removed, and 200 pVwell of Semliki forest virus (SFV) added at a dilution of 100 x TCID-50. The virus titers were predetermined and did not differ significantly between cell lines. Virus and cells were incubated a t 37 "C in 5% CO, in air for three days. The supernatants were then removed and cell monolayers stained with 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide as described previously (21). IFN titers were determined as the concentration of IFN to provide protection of 50% of the cells. The ratio of IFN titers needed for antiviral activity for the various cell lines was determined within each experiment, and these ratios from three to five experiments performed on different days are expressed as mean S.E.

RESULTS
Expression of ZFNAR and Ligand Binding-In initial experiments we examined the four different CHO-human chromo- Northern blot analysis of these four cell lines as well as the CHO cell line demonstrated that all the hybrid cell lines expressed a transcript of approximately 2.7 kilobase pairs for human IFNAR, whereas the CHO cell line did not express the human transcript ( Fig. l b 1. The transcript size was as expected on the basis of sequence and published data, including Northern blots of RNA from Daudi cells (from which the IFNAR cDNA was cloned) (1). The apparent difference in mobility of the transcript from the 21q' cell line is an artifact of the running conditions used for this gel, thus the GAPDH transcript also has a similarly affected mobility (Fig. lb). Furthermore in other Northern blot experiments where RNA from 21q+ and 72532x6 only were tested, the IFNAR transcript had the same mobility (data not shown).
Analyses of the binding of human type I IFNs to receptors on the hybrid cell lines are presented as binding curves and Scatchard plots (insets) (Fig. 2). Scatchard analysis of the binding of IFNs a2, aB, and p to the hybrid cell lines was resolved into a one-site fit in all cases (p < 0.05) (Fig. 2). The very low level of binding of lZ5I-IFNs a 2 or aB to the CHO cell line could not be resolved by Scatchard analysis (Fig. 2a). The binding of lZ5I-IFNp to the CHO cell line was also lower than to the hybrid cell lines. Although there may appear to be some high affinity binding component evident from the Scatchard plots, this could not be significantly resolved at these very low levels of binding (and even if it was, the numbers would have been very small and similar for the four hybrid cell lines). Such a low level of high affinity binding may be a consequence of some low level inter-

CONCENTRATION (pM)
action of the human IFN ligand-human IFNAR complex with various hybrid cell lines. For instance, the binding of IFNaB to other components of the endogenous hamster type I IFN recep-72532x6 and 21q+ had dissociation constants of 170 and 150 PM, tor(s). respectively, and the number of receptors per cell were 300 and All four hybrid lines bound '251-IFNa2 with similar dissocia-400, respectively (Fig. 2b). For IFNP, binding to 72532x6 and tion constants of about 200 PM and had similar number of 21q+ dissociation constants were 54 and 68 PM, respectively, receptors per cell, about 800 (Table I and Fig. 2a). Similarly, the and the number of receptors per cell were 1300 and 720, rebinding of '251-IFNaB and '251-IFNP was comparable in the spectively.

Type I Interferon
Induction of 2'-5' Oligoadenylate Synthetase-The ability of the receptor-bound IFNs 012, a8, and p to transduce signals in the hybrid cell lines was measured first by the induction of 2'-5' oligoadenylate synthetase, a double-stranded RNA-dependent IFN-induced enzyme which is induced early and plays an important role in the antiviral response (24). IFNa2 significantly induced 2 ' 4 ' oligoadenylate synthetase activity in 72532x6 and 643C-13, but surprisingly not in 21q+ nor MRC 2G nor the CHO cell lines (Fig. 3a). The enzyme activity of the untreated cells in nanomoles of Pi incorporated per pg of protein was: 108 f 49 (mean f S.E., n = 5 ) for CHO, 56 +. 37 (n = 4) for 21q+, 62 f 46 (n = 3) for 72532x6, 73 2 35 (n = 4) for MRC 2G and 62 +. 27 (n = 4) for 643C-13. The induction of 2'-5' oligoadenylate synthetase occurred in an apparent dose-related fashion and was already significantly increased in 72532x6 by 100 IU/ml of IFNa2 (Fig. 3a). On the contrary, the treatment of 21q+, MRC 2G, and CHO cell lines, which all lack the region of human chromosome 21q22.2-3, showed no (or very little) induction of 2'-5' oligoadenylate synthetase activity, even at concentrations of IFNa2 up to 1000 IU/ml (Fig. 3a). Induction of 2 ' 6 ' oligoadenylate synthetase by IFNaB and IFNp was also only observed in 72532x6, which contains the entire human chromosome 21, but not in the 21q+ cell lines (Fig. 3, b and c ) . Furthermore, we also found in one experiment that IFNp (1000 IU/ml) increased the activity of 2'45' oligoadenylate synthetase in 643C-13 cells (increase of 360 nmol of Pjpg of protein), but not in MRC 2G cells. These data demonstrate that a factor(s) encoded by a gene(s) located at 21q22.2-3 is necessary for eliciting a response resulting from IFN ligandreceptor interaction and show that the factor is required prior to induction of IFN-induced genes such as 2'-5' oligoadenylate synthetase which occurs early in the response.

Antiviral Responses-Even though all the hybrid cell lines
Receptor binding of IFNa2a to hybrid cell lines    (Fig. 4). Interestingly, the 21q+ cell line, which lacks the human chromosome 21 region 21q22. 2-q22.3, was as insensitive to IFNa2 as the CHO cell line (Fig. 4).
Additional experiments demonstrated that only 72532x6, which contains the entire human chromosome 21, demonstrated efficient antiviral activity irregardless of type I IFN ligand used or the cytopathic virus used. Thus, about 50-fold less IFNp (47 2 27, n = 3) was required to protect 72532x6 (relative to 21s' and CHO) from the cytopathic effect of SFV. Also, when encephalomyocarditis virus was used as the challenge virus, 21q+ was as insensitive as CHO to the antiviral effects of IFNa2 (10-fold more IFN required than 72532x6). Together, these data demonstrate that a gene(s) encoded in the region 21q22.2-3 is necessary for antiviral responses triggered by the binding of these type I IFNs to the receptor. These results confirm those on the induction of 2'-5' oligoadenylate synthetase. DISCUSSION These studies demonstrate that a factor encoded on human chromosome 21 q22.2 to q22.3 is required for transducing signals necessary for antiviral response induced by several type I IFNs, namely a2, aB, and 0, The hybrid cell lines that contained the entire (or virtually entire) human chromosome 21 showed an up-regulation in 2'4' oligoadenylate synthetase activity and an antiviral response. Hybrid cell lines that contained human chromosome 21 but lacked the region q22.2 to q22.3 failed to show evidence of a biological response, even though they bound the three type I IFN ligands. The localization of the gene encoding this factor to human chromosome 21q22. 2-22.3 and not to other random human genomic fragments that may be present in such interspecies hybrids (25) is confirmed by its presence in two hybrid cell lines containing then SFV for 3 days. Thereafter, viable cells were stained (red) using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide as described 72532x6, and 21q'. Quadruplicate wells of each cell line were exposed to serial dilutions of IFNol2a ranging from approximately 20 to 3000 IU/ml, in the text.
this region as well its absence from two hybrid cell lines which lack this region. These findings are also significant because they show that the factor(s) necessary for signal transduction of an antiviral response is apparently not involved in receptor binding. The cloning of this molecule is currently in progress and will be aided by the recent mapping of the entire human chromosome 21 and the availability of yeast artificial chromosomes spanning this chromosome (26). The data in this study also give insight into the failure of murine cells transfected with IFNAR to respond to human IFNs a2 and /3 as reported by UzB et al. (1). The absence of the human chromosome 21-encoded factor identified herein may, in part at least, explain why the murine IFNAR transfectants did not develop an antiviral response to IFNs a2 and /3 (1). The development of a response only to IFNaB reported in those studies could be explained if IFNaB, but not IFNa2 and -P, interacted with the murine homologue(s) of the "signaling" factor($, located a t human chromosome 21q22.2 to 22.3. For this to occur, IFNaB should have structural differences from both IFNa2 and -P and a similarity with muIFNa. Indeed IFNaB has 26 amino acid residues different from IFNa2 and many more differences from IFNP; 6 of these residues are conserved between human IFNaB and the majority of murine IFNas, namely Asn", Ile14, Proz6, Gln5', Ala53, and Metloo (4). Perhaps one or more of these residues is important for interactions with the murine signaling factor(s).
In the present study the hamster homologue of the human chromosome 21-encoded signaling factor did not reconstitute signal transduction by the human IFN ligand (012, aB, nor pl-hurnan receptor complex, indicating the species specificity of this interaction. It is unlikely that this factor is one of the JAK family of protein tyrosine kinases known to be involved in type I IFN signaling (10, 11). The activity of JAK-1 is not speciesspecific, since human cells defective in IFN signal transduction exhibit reconstituted signaling when transfected with murine JAK-1 cDNA(11). Another JAK family member, tyk-2, is known to be encoded by a gene located on human chromosome 19 (27). Furthermore, none of the known JAK family members have been localized to human chromosome 21 (37).
The finding that a type I IFN signaling molecule in addition IFNyR interacts with a transmembrane component which does not affect ligand binding but is essential for signal transduction (35,36). This component (GAC) is encoded by a gene also on the proximal portion of chromosome 21 (28).
to IFNAR (and an IFNy signaling molecule (28) and IFN-inducible Mx genes (15)) are located on chromosome 21 may also have relevance to Down's syndrome (trisomy 21). This may explain why cells from individuals with Down's syndrome (trisomy 21) have shown approximately %fold greater antiviral response to IFN (29), which is greater than a 1.5-fold gene dosage effect, which would be expected from the overexpression of IFNAR by itself. Some of the features of Down's syndrome, such as inhibited growth and immunological abnormalities, may be due (in part) to dysregulation of the antiproliferative and immunoregulatory actions of IFNs. tain all components required to give the described receptor