Sublocalization of the Human Interferon-y Receptor Accessory Factor Gene and Characterization of Accessory Factor Activity by Yeast Artificial Chromosomal Fragmentation*

A chromosomal fragmentation procedure was em- ployed to produce a deletion set of yeast artificial chromosomes (YACs) from a parental YAC, GART D142H8, known to map to human chromosome 21q and to encode the human interferon-y receptor (Hu-IFN-yR) accessory factor gene as well as the phosphoribosylglycinamide formyltransferase (GART) gene. When expressed in Chinese hamster ovary cells, these deleted YACs retain ac- cessory factor activity, as judged by major histocompatibility complex class I antigen inducibility, until the deletions from the acentric end exceed 390 kilobases (kb). Therefore, the accessory factor (AF-1) gene can be localized to a 150-kb region at the left (centric) end of the parental 540-kb GART YAC. Cells containing functional YACs are also able to induce the ISGF3y and y-activated factor (GAF) transcription factors, but were not pro- tected against encephalomyocarditis virus ( E M 0 upon treatment with Hu-IFN-y. Therefore, the Hu- IFN-yR and the AF-1 are sufficient for some, but not all, of the actions of Hu-IFN-y. We postulate that an additional

mosome 21 in somatic cell hamster x human hybrids does Hu-IFN-y signal transduction occur (5,6). Similarly, only when the Mu-IFN-yR encoded on mouse chromosome 10 (7) is accompanied by mouse chromosome 16 (encoding the mouse accessory factor) does the Mu-IFN-yR become biologically active in hamster x murine somatic cell hybrids (8).
The human chromosome 21 accessory factor is likely to be a membrane-associated protein as several studies have shown that the accessory factor and the extracellular domain of the IFN-yR must be from the same species for signal transduction to occur (9)(10)(11). Although interactions with the transmembrane and intracellular domains have not been ruled out, these interactions, if they exist, are not species-specific. Use of radiation-reduced somatic cell hybrids has enabled further mapping of the accessory factor gene on human chromosome 21 to a region proximal to dinucleotide repeat polymorphisms D21S58 a n d D21S223, as well as D21S224 and D21S235 (12-14). More recent data from this laboratory indicate that accessory factor activity is encoded by the yeast artificial chromosome (YAC) whose address on chromosome 21 is D142H8 (15). This YAC, designated GART D142H8 because i t also encodes the GART gene, is 540 kb in length and is located proximal to the YAC which encodes the human IFN-dp receptor (15). The present study was initiated to localize the accessory factor gene on the GART D142H8 YAC. In addition, it was anticipated that mapping the location of the accessory factor gene would facilitate isolation of the gene and coding sequence from cosmid and cDNA libraries, as experiments with cosmids and cDNA expression libraries to complement accessory factor activity have to this point been unsuccessful. In a separate report (16) we have described a procedure for fragmenting the parental GART D142H8 YAC which is, in principle, applicable to any YAC carried in a Ura-host. Application of this procedure to the GART D142H8 YAC yielded a number of fragmented YACs which have enabled us to determine the location of the accessory factor gene. Production of CHO cell lines containing a small segment of the human DNA containing the accessory factor has, in addition, enabled us to define further the properties of the accessory factor. The results of these studies are described in this communication.
EXPERIMENTAL PROCEDURES Cells-The hamsterhuman somatic cell hybrid 16-9 which contains human chromosome 6q was maintained in F-12 medium containing 10% fetal bovine serum. These cells had been transfected with a genomic clone for the class I HLA-B7 antigen (6, 15). The cell lines which were transformed by fragmented YACs were maintained in the same medium plus 450 pg/ml antibiotic G418 (Life Technologies, Inc.). The 153B7-8 cell line is a Chinese hamster ovary (CHO-K1) cell line which contains human chromosome 21q and the HLA-B7 antigen gene (6); 153B7-8LHHHI refers to the same cell line which was transfected with the Hu-IFN-yR cDNA. Both of these cell lines were maintained in 7013 F-12D containing 10% fetal bovine serum as described earlier (9,17). The 3x1s cell line is a CHO-K1 cell line which contains 1-3 Mb of human chromosome 21 (12, 18).
Reagents, Restriction Endonucleases, and Other Enzymes-Human IFN-y, with a specific activity of 1.3 x lo7 unitdmg, was isolated from Escherichia coli as described (19). Human IFN-aA/D(Bgl) was prepared as reported (20) and had a specific activity of 6.0 x lo7 unitdmg. Human IFN-P was isolated as described (21,221 and had a specific activity of 3.5 x lo7 unitdmg. Restriction endonucleases were obtained from New England Biolabs and Boehringer Mannheim. Yeast DNA and A ladder pulsed-field gel markers were from Life Technologies, Inc. and New England Biolabs, respectively. Construction of Fragmented YACs-The parental GART YAC (GART D142H8) was obtained through the "chromosome 21 joint YAC screening effort" (directed by Dr. David Patterson, Eleanor Roosevelt Institute, Denver, CO). The plasmids pSEl and pSE2 (with Alu sequences in opposite orientations) were constructed such that BamHI and EcoRI digestion releases an 8.4-kb fragment containing an Alu sequence, an SV2neo gene, a URA3 gene, and a tetrahymena rDNA telomere, in that order (16,23). The mixture of plasmids pSEl and pSE2 (12.5 pg of each) was so digested, and the 8.4-kb fragment was isolated and used to transform the yeast strain GART D142H8.ura.2 which contains the parental GART YAC and had been converted to a Ura-phenotype by 5-fluoro-orotic acid selection (16). Transformations were performed as described (23), and Ura' transformants were obtained on uracil-deficient plates.
Fusion of Fragmented YACs to Mammalian Cells-Approximately 2 x lo7 16-9 cells were fused to 10 ml of a 5% confluent yeast culture with polyethylene glycol (23). After incubation for 24-36 h in F-12 medium plus 10% fetal bovine serum and 50 pg/ml gentamicin, the cells were washed twice with phosphate-buffered saline, and the transformants were selected in the same medium containing 450 pgiml antibiotic G418.
Cytofluorography-Induction of MHC class I antigens by IFN was assayed as reported (9,241. For each sample, 10,000 cells were analyzed with a Coulter Epics Profile cytofluorograph and Cytologic software. Assay of the Danscription Factors ISGF3y and GAF-For factor ISGFBy, cells transformed by fragmented YACs were incubated with or without 100 unitdm1 IFN-y for 18 h. The cells were harvested by trypsin-EDTA treatment, washed once in F-12 plus 10% fetal bovine serum and once in PBS. Lysates were prepared from the cell pellets, and gel shift assays were conducted as described (25). GAF was assayed as described (26) after cells had been incubated with or without 100 unitdm1 IFN-y for 30 min.

RESULTS
Gene Mapping-Using hamster x human somatic cell hybrids, we determined that accessory factor activity is located on human chromosome 21q (5, 6) and on mouse chromosome 16 (8). Further cytogenetic evidence indicated that the chromosome 21 accessory factor gene resides within the 1-3 Mb of chromosome 21q which is contained in the 3x1s somatic cell hybrid (12, 16). In addition, the chromosome 21 accessory factor gene co-segregates with the IFN-odp receptor gene and is also located near the GART gene (12, 15). To map more precisely the location of the human chromosome 2 1 accessory factor gene, we obtained a series of YACs specific to chromosome 21 into which we introduced a neomycin resistance gene in order to determine their biological activity by fusion to mammalian cells (15).
These experiments indicated that the gene encoding the chromosome 2 1 accessory factor activity is located within a 540-kb YAC which also encodes the GART gene and whose address is D142H8. A second YAC, which is 160 kb in length and may partially overlap the GART YAC (Fig. l), encodes a human I F N -d p receptor gene (28) and is located at address B49F1 (15). These relationships are diagrammed in Fig. 1.
Although the GART YAC is known to be oriented as shown in  receptor YAC a n d the GART YAC (dashed line, Fig. 1) has not been defined. To date, no IFN-a receptor probe has hybridized to the GART YAC.' This may indicate that the IFN-a receptor YAC is oriented as shown in Fig. 1 or that the IFN-a receptor gene is proximal to the GART YAC, but there is a gap between these two YACs. If there is such a gap, it is likely to be less than 50 kb in size. Based on digests of human DNA, Cheng et al. (29) have shown that the IFN-a receptor gene, the CRF2-4 gene (a gene encoding a receptor-like protein whose function is unknown; Ref. 301, and the GART gene are all contained within a 400-kb MluI restriction endonuclease fragment. These data are consistent with our map shown in Fig.   1. The 3' end of the GART gene has been found to be close to 150 kb from the left (centric) end of the GART YAC. The GART gene itself is 40 kb in length (29); and the IFN-a receptor YAC is 160 kb. These numbers account for 350 kb of the 400-kb MluI fragment.

MHC Class ZAntigen Induction in Cell Lines 'I).ansformed by
Fragmented YACs-To map more closely the location of the human chromosome 21 accessory factor gene on the GART YAC and to facilitate its identification, a method was devised to fragment the GARTYAC from its acentric end (16,311. Briefly, a fragmentation vector was constructed which contains a BamHI-EcoRI fragment, including an Alu sequence, a neophosphotransferase (neo) gene, a URA3 gene, and a telomere. Yeast cells containing the parental GART YAC were converted to a Uraphenotype by 5-fluoro-orotic acid selection. Recombination of the Alu sequence with homologous sequences in the parental GART YAC produced truncated YACs which were selected on uracil-deficient plates. DNA was isolated and was characterized by pulsed-field gel electrophoresis. YAC size was generally found to range from 80 to 500 kb (Fig. 2). Several YACs were shown to contain the GART gene based upon hy-  Flc. 2. Schematic illustration of YACs used to transform 1 6 9 cells. The phenotype of all the yeast strains containing the YACs shown is Ura', T r p * , neo'. Hybridization to a probe (:'"P-labeled by random priming) specific to the left (centric) end ofthe GART YAC (pGC8.10E6) indicated that all YACs shown contained an intact centric end. Fusion ofyeast containing YACs to 16-9 cells was performed as described under "Experimental Procedures." A "+" indicates that MHC class I antigen induction with Hu-IFN-y was observed; "-" indicates no observed induction a t 51,000 unitshl IFN-y. Sizes of YACs were determined by pulsed-field gel electrophoresis with yeast DNA and/or A ladder pulsedfield gel markers. Smnll circles represent telomeres; oun/s indicate location of centromeres. determined to be UraS', neo', Trp'. The transformants were next fused to recipient 16-9 somatic cell hybrids, which contained the human IFN-y receptor as well as the HLA-B7 gene. These fusions produced variable numbers of antibiotic G418resistant colonies. Certain YACs did not produce any viable cells. However, for the cell lines reported here, all fusions resulted in at least 100 antibiotic G418-resistant colonies. In one instance, a limiting dilution procedure was used to produce a clonal J29/16-9 cell line. Antibiotic G418-resistant cells were assayed for chromosome 21 accessory factor activity as described previously (24). As is shown in Figs. 2 and 3, cell lines J29/16-9, J28/16-9, J16/16-9, and 518116-9 were responsive to Hu-IFN-y as seen by induction of MHC class I antigen gene expression. However, cell lines J20/16-9 and J6/16-9 did not respond to Hu-IFN-y. These results indicate that the chromosome 21 accessory factor gene is located within 150 kb from the centric end of the GART YAC. Restriction mapping of the YACs 520 and 518, both approximately 150 kb in size, indicated that these YACs are virtually identical. However, two additional fusions of yeast cells containing YAC 520 have not produced IFN-y-responsive cells. We therefore conclude that YAC 520 contains a small deletion relative to YAC 518 which inactivates the accessory factor gene. Yeast cells containing YAC J 6 h a v e been fused twice to 16-9 cells without isolation of any Hu-IFNy-responsive cells. In addition, another 80-kb YAC was found to be inactive (data not shown). YACs less than 150 kb in size (Le. with more than 390 kb of the acentric end deleted) were inactive. YAC J18 is the smallest YAC that retained accessory factor activity.
Although the magnitude of MHC class I antigen induction in cell lines J29/16-9 and J28/16-9 does appear somewhat greater than that of J18/16-9 in Fig. 3, subsequent assays showed that all cell lines responding to IFN-y seemed to do so to approximately the same extent (Fig. 4). Therefore, it is unlikely that fragmentation has deleted a relevant gene at the acentric end of the GART YAC. In addition, it is significant to note that the MHC class I antigen response of the J16/16-9, J18/16-9, J28/ 16-9, and J29/16-9 cell lines as a function of Hu-IFN-y concentration are comparable with that of certain human cell lines and hamster x human somatic cell hybrids containing human lJ16'16-9 I chromosomes 21 (or 21q) and 6 (or 6q or the cloned Hu-IFN-yR, Refs. 5,6, 17, and 18). Typically, a human cell line treated with 10-1.000 units/ml of IFN-y for 2-3 days will exhibit 1.5-3-fold higher fluorescence as a population when analyzed by cytofluorography. The quantification of fluorescence for the individual cell lines produced by YAC fusion are all within this range (Fig.  4).
MHC class I antigen induction as a function of IFN-y concentration with the cell lines produced by the fusions was examined. As shown in Fig. 4 hamster x human somatic cell hybrids (5,6,17,18).
Danscriptional Activation-We examined the induction of two transcription factors in these hamster cells. The transcriptional activation of interferon-inducible genes has been shown to be stimulated by transcription factors which accumulate in the cytoplasm and migrate to the nucleus (25,32,33). Levels of transcription factor ISGF3y have been shown to increase in the cytoplasm after IFN-y treatment (34). To determine how much of the IFN-y pathway is functional in the YAC cell lines, we measured ISGF3y in IFN-y-treated and untreated cells by gel retardation assays with labeled interferon stimulated response element (ISRE; Refs. 25, 32, and 33). As is shown in Fig. 5A, ISGF3y levels are not induced in 56116-9 and 16-9 cells treated with Hu-IFN-y. However, Hu-IFN-y induced ISGF3y in 518/ 16-9 cells. The reason for the high background in 56/16-9 cells is not clear. The unlabeled ISRE competitor oligonucleotide effectively competed out the ISGF3y signal. We also examined the induction of GAF, which is induced much more rapidly than ISGF3y (26). Incubation of 16-9, GART/16-9, 56/16-9, and 518/ 16-9 cells with IFN-y for 30 min induced GAF in the GART/16-9 and 518116-9 cells, but not in 16-9 and 56/16-9 cells (Fig. 5B). This is consistent with the ISGF3y data. Antiviral Activity-The Hu-IFN-y receptor and chromosome 21q have been shown to be sufficient for an antiviral (EMCV) response in CHO cells (24). In addition, the 3x1s cell line, containing only 1-3 Mb of chromosome 21q, when transfected with the Hu-IFN-yR, is also protected by Hu-IFN-y when challenged by EMCV (15). To determine whether the receptor and the accessory factor encoded on the YACs are sufficient to produce antiviral effects in CHO 16-9, 518/16-9, and 529/16-9 cells, the cells were challenged with EMCV after treatment with various interferons. Human IFN-/3 and IFN-aA/D(Bgl) were active in protecting the 16-9, J18/16-9, and J29/16-9 cells against EMCV (Table I). However, even at >3250 units/ml, Hu-IFN-y was unable to induce protection against EMCV on 16-9 and 518/16-9 lines (Table I); at >13,000 units/ml, Hu-IFN-y was unable to induce any protection against EMCV on 529/16-9 cells. The 153B7-8 cell line (containing human chro-mosome 21) transfected with the Hu-IFN-yR was found to have high antiviral activity against EMCV with half-maximal protection at 3 unitdm1 of IFN-y. This value closely agrees with the dose-response data for MHC class I antigen induction in the 16-9 cell lines containing the YAC clones (Fig. 4). These data indicate that anti-EMCV responsiveness requires another human gene which is not present on either the 518 or 529 YAC. This gene is, however, located within the 1-3 Mb of chromosome 21, since the 3x1s and 153B7-8 cells transfected with the human IFN-y receptor cDNA are protected from EMCV infection a t similar levels of Hu-IFN-y (15). The 16-9 cells containing YACs 518 and 529 exhibited a slight increased sensitivity compared with parental 16-9 cells to both Hu-IFN-& and Hu-IFN-/3 of about 2-fold and 4-6-fold, respectively. The significance of this increased sensitivity remains to be determined. DISCUSSION The results obtained in this study indicate that it is feasible to map eukaryotic genes by fragmentation of YACs followed by the expression of fragmented YACs in a host cell which is capable of providing a specific assay for the gene product. Although a number of fragmentation vectors have been developed (35)(36)(37), we are not aware of any studies in which yeast artificial chromosomal fragmentation has been used to map genes based on complementation of gene function. In principle, the same conclusions could be obtained by transfection of cosmid clones. However, given the size constraints of cosmid vectors, it may be difficult to obtain a single cosmid which contains a complete and functional gene. None of the cosmids which we have mapped to the region of chromosome 21 under consideration produced accessory factor activity (18). The localization of accessory factor activity on the GART YAC will enable us to identify the AF-1 gene.
To obtain fragmented derivatives of the parental GARTYAC, we developed vectors (pSE1 and pSE2) which fragment from the acentric end of the YAC. Homologous recombination of vector DNA with the YAC results in the addition of a URA3 and a neo' gene to the acentric end. Yeast clones containing fragmented YACs were then fused to 16-9 cells and transformants were selected with antibiotic G418. Although this procedure requires selection for Ura-mutants, the frequency of fragmentation is high, with 78% of the yeast transformants containing a YAC smaller than the parental GART YAC. The Ura3' selection procedure has the further advantage that it can be used with YACs in both YF'H252 and AB1380 yeast strains. The fragmented YACs obtained with this procedure were shown to be truncated from the acentric end only based upon hybridization to a probe (pGC8.10E6) which is specific for the left end of the parental YAC. Restriction mapping of the fragmented YACs indicated that some fragments are shared by all of the YACs, due presumably to their shared centric ends.
The present study localizes the human chromosome 21 accessory factor to a position which is within 310 kb of the human IFN-aIP receptor gene. This estimate is derived from the fact that the YAC containing the IFN-dP receptor gene is 160 kb in size and the 518 fragmented YAC, which encodes accessory factor activity, is 150 kb in size. The distance between the accessory factor gene and the GART gene has not been established with precision. The GART gene is not encoded by 518 (data not shown), but mapping of cosmids derived from the GART and 518 YACs indicates that the acentric end of 518 and the GART gene are present in a single cosmid clone.3 Precise localization of the accessory factor gene is not yet possible, since we do not have any YACs which are larger than 56 (80 kb) J. Soh and S. Pestka, unpublished data. and smaller than 518 (150 kb). 520, which is also 150 kb in size, was negative for MHC class I antigen induction in several assays. If these data result from the fact that YAC 520 is actually slightly more truncated than YAC 518, the accessory factor gene would have to be bounded by the right end of YAC 518. On the other hand, we cannot exclude that the inactivity of YAC 520 is due to a small internal deletion, undetectable by our gross restriction mapping analyses. We have assayed an additional small YAC which is the same size as 56 (59, 80 kb). This YAC was also negative for MHC induction.

Antiviral (EMCV) activitv of interferons in cell lines stably fused with YACs containing the human chromosome 21 accessory factor gene
Since hamster interferons and hamster interferon-induced genes are not available, we chose to examine whether two transcription factors which have been shown to be induced by IFN-y in human cells can be induced in a hamster cell background. ISGF3y is a 48-kb protein which is the DNA binding component of the ISGF3 complex (32). Three other proteins associate with ISGF3y to form ISGF3. The other transcription factor, y-activated factor (or GAF), is rapidly induced by IFN-y and binds to a DNA sequence different from that to which ISGF3y binds (26). We found that there is a correlation between MHC class I antigen inducibility and induction of hamster ISGF3y and GAF in the cell lines transformed by the fragmented YACs; the cell lines in which Hu-IFN-y can induce MHC class I surface antigens are also inducible for ISGF3y and GAF in response to Hu-IFN-y (Fig. 5 ) . This indicates that the chromosome 21 accessory factor is required and sufficient for ISGF3y and GAF induction.
The observation that cells 518116-9 and 529116-9 are not protected from EMCV infection even in the presence of very high (>3,000 units/ml) Hu-IFN-y concentrations indicates that, although the accessory factor and the Hu-IFN-yR are adequate to induce MHC class I antigens and to activate the ISGF3y and GAF pathways, an additional factor is required to generate resistance to EMCV. This factor is clearly not located within the 518 or 529 YACs. Examination of the parental GART YAC has also shown that this YAC is not sufficient to produce anti-EMCV activity when expressed in CHO cells (15). We have designated this chromosome 21 accessory factor that induces MHC class I antigens and transcription factors ISGF3y and GAF as AF-1 (Fig. 6). The second factor, which is also encoded on chromosome 21q and is required for EMCV resistance, has been designated AF-2. We cannot exclude the possibility that the activity we ascribe to AF-2 is due to a dosage effect with respect to AF-1: that is, it is conceivable that what we have designated AF-2 may reflect quantitative differences in the expression of AF-1. We will examine this possibility with a cDNA encoding AF-1 and/or with antibodies to AF-1. However, the amount of AF-1 produced is clearly sufficient for ISGF3y, GAF, and MHC class I antigen induction in 518/16-9 cells (Figs. 2-5). Furthermore, the magnitude of the inductions observed in 518116-9 cells was similar to those in human cell lines; doseresponse characteristics were comparable as well. Although AF-2 is required for the antiviral EMCV activity in response to Hu-IFN-y, it may function alone or together with AF-1. The FIG. 6. Diagrammatic representation of the location and interaction of the Hu-IFN-y receptor accessory factors located on chromosome 21q and elsewhere. A F -1 refers to the chromosome 21 accessory factor whose activity and location have been described previously (5, 6,9,17,18). AF-2 is an accessory factor which is also located on chromosome 21q and which confers EMCV resistance to cells in response to Hu-IFN-y. AF-3 is an accessory factor which is required for VSV resistance (24) in response to Hu-IFN-y, but not located on chromosome 21q. Dashed arrows indicate potential cooperative interactions between accessory factors. accessory factor designated AF-3 was defined in a previous study (24) in which it was shown that CHO cells containing chromosome 21q (and expressing both AF-1 and AF-2) are incapable of generating full VSV resistance in response to Hu-IFN-y. AF-3 is therefore located on a chromosome other than 21q. As with AF-2, although AF-3 is required for the antiviral VSV activity in response to Hu-IFN-y, AF-3 may function alone or together with AF-1 and/or AF-2. AF-1, AF-2, and AF-3 are species-specific in their activity.
The conclusions obtained in this study with fragmented YACs which map to human chromosome 21 should be compared with those recently reported by Kalina et al. (38) using human x mouse somatic cell hybrid fibroblasts. These chromosome 81-containing mouse fibroblasts transfected with the Hu-IFN-yR cDNA were found to be resistant to EMCV upon IFN-y treatment; 2',5'-oligoadenylate-~ynthetase and MHC class I antigens were also induced by Hu-IFN-y. However, when individual clones were examined, it was found that EMCV protection was absent in 67% of the clonal cell lines. Therefore, the genes present on human chromosome 21 may be insufficient to produce EMCV protection in all cases, parts of chromosome 21 may be missing in some clones, or there may be variable expression of the factor encoded on chromosome 21 responsible for EMCV protection. Our experiments with WA17 cells (a mouse x human somatic hybrid cell line which is trisomic for human chromosome 21; Refs. 39  assay systems employed in the present study in mouse cells. In any case, although the data of Kalina et al. (38) support the hypothesis that other species-specific proteins besides AF-1 are required for anti-EMCV activity, the present study indicates that a factor (AF-2) encoded by chromosome 21 complements AF-1 to produce EMCV resistance. In summary, the fact that a specific chromosomal fragment containing a gene for AF-1, which is functional in CHO cells (as judged by MHC class I antigen, ISGFSy, and GAF induction), is not sufficient for either anti-EMCV or anti-VSV activities indicates that other factors are required in order to produce the full spectrum of IFN-y activities. Taken as a whole, our results support the hypothesis that multiple accessory factors ("1, AF-2, and AF-3) are required for the full functional activity (including anti-EMCV and anti-VSV activity) of Hu-IFN-y and its receptor. Further identification and characterization of these accessory factors will be necessary to understand the IFN-yR signal transduction mechanism.