The Interferon-y Receptor Extracellular Domain - The Journal of ...

Vol. 269, No.22, Issue of June 3, pp. 15533-15539, 1994 Printed in U S A .

THEJOURNAL OF BIOLOOICAL CHEMISTRY

The Interferon-y Receptor Extracellular Domain NON- DEN TIC^ R E Q U I R E ~ E FOR ~ S LIGAND IN DING AND S I G N ~ I N G * (Received for publication, October 19, 1993,and in revised form, March 8, 1994)

Amy Axelrod$, VernaC. GibbsQ, and David V. Goeddelh From the $Department of Molecular Biology, Genentech, Inc., South Sun Francisco, California 94080 and the §Department of Surgery, University of California at Sun Franciso, Sun Francisco Veterans Administration Medical Center, Sun Francisco, California 94121

The human interferon-gamma receptor fhIFN-yR) ex- 21-encoded component does not appear toaffect the affinity of tracellular domain interacts in a species-specific man- receptor for ligand, but does interact with the EC domain of the receptor. The latter result was demonstrated by constructing ner with both its ligand and an accessory factor encoded on human chromosome 21. Mutant interferon-y recep- hybrid receptors with a human EC domain and a murine IC tors were constructed by homolog-scanning mutagen- domain. These hybrid receptors do not function in murine cell esis,replacingsegments of thehumanextracellular lines, but do function in human-murine somatic cell hybrids domain with the corresponding murine sequence. Re- containing human chromosome 21, indicating that thespeciesplacement of hIFN-yR amino acids 1-100, 100-132, 134- specific interaction between the receptor and the chromosome 183, or 183-245 abolished binding to human interferon- 2 1 component occursextracellularly (Gibbs et al., 1991; Hemmi gamma (hIFN-y). However,replacementofhIFN-yR et al., 1992). amino acids 134-209, 183-209, 134-153, 153-167, or 167The IFN-y receptor (IFN-yR) EC domain is predicted to fold 183 ordeletion of residues 156-165 affected hIFN-y bindinto a specific @stranded structure sharedby a large receptor ing only partially or not at all. Receptors that bound family (Bazan, 1990; Thoreau et al., 1991). Within this family hIFN-y were tested for their ability to signal a functional response, induction of major his~ompatibilitycom- the IFN-yR is most closely related to theIFN-a receptor, tissue factor, and CRF2-4 (Lutfalla et al., 1993). More distant relaplex classI antigen expression. Replacement of residues 134-209 greatly reduced the ability of the receptor to tives are found in the cytokine receptor I family and include receptor components for growth hormone, prolactin, erythrosignal.Thissignalingdefectcouldnotbeattributed solely to a reduction in affinity for ligand and could not poietin, granulocyte-macrophage colony-stimulating factor, granulocyte colony-stimulating factor, andthe interleukins be localized to any subregion. IL-2 (&chain), IL3, IL-4, IL-5, IL-6, and IL-7. Additional information on the structure of the IFN-yR EC domain is limited. Interferon-y (IFN-y),l a cytokine produced by activated T- Glycosylation contributes 25 kDa to the receptor’s mass, but is cells and natural killer cells, is required for an effective im- not required for binding activity (Fountoulakis et al., 19901. mune response to some parasitic and viral i n f e c t i o ~(Dalton et Disulfide bonds form between consecuti~ecysteines, and all aE., 1993; Huang et al., 1993). IFN-y modulates immune re- four disulfide bonds are required for full ligand binding activity sponses in manyways, but itspleiotropic activities appear tobe (Stuber et al., 1993). In addition, ligand binding activity is mediated through a single cell surface receptor expressed on disrupted if the EC domain is truncated by even 21 amino acids many cell types (reviewed in Farrar andSchreiber, 1993). The at the matureN terminus or 14 amino acids at themembranehuman and murine IFN-y receptors share 52% amino acid se- proximal end (Fountoulakis et al., 1991). We report here a homolog-scanning mutagenesis (Cunningquence identity andspan themembrane a single time ( b e t et al., 1988; Gray et al., 1989; Hemmi et al., 1989; Munro and ham et al., 1989) of the IFN-yR EC domain. Segments of the hIFN-yR EC domain were replaced with the corresponding Maniatis, 1989; Kumar et al., 1989). The extracellular (EC) domains share 50% amino acid sequence identity and bind sequence from the mouse IFN-yR (mIFN-yR). This homologscanning mutagenesisstratem has theadvantage of changing IFN-y with strict species-specificity. IFN-y signal transduction requiressequences within the re- multiple amino acids in a single receptor mutant while miniceptor intracellular (IC) domain (Farraret al., 1991,1992; Cook mizing the possibility of introducing a global change in the et al., 19921, a species-specific component encoded on human receptor’s conformation. We analyzed the mutantreceptors for chromosome 21 or murine chromosome 16 (Jung et a.?., 1987; both ligand binding activity and the ability to induce MHC Hibino et al., 1990), and also, most probably, ligand-induced class I antigen expression. Although there is no biochemical dimerization of two receptor molecules (Greenlund et al., 1993; assay at thistime for interaction between the receptor and the Fountoulakis et al., 1992; Dighe et al., 1993). The chromosome species-specific accessorycomponent, a defect in theinteraction between these two molecules wouldinterfere with the ability of .” * The costs of publication of this article were defrayed in part by the the receptor to mediate a functional response. Thus, a receptor be hereby marked mutant that binds ligand but does not function may indicate payment of page charges. This article must therefore “advertisement” in accordancewith 18 U.S.C. Section 1734 solelyto how the receptor interacts with the accessory component. We indicate this fact. tl To whom co~es~ndence and reprint requestsshould be addressed discuss our results with respect to the structural model proGenentech, Inc., 460 Pt. San Bruno Boulevard, S. San Francisco, CA posed (Bazan, 1990) for the IFN-yR. ”

94080.

‘The abbreviations used are IFN-y,interferon-y;hIFN-y,human interferon-y; mIFN-y, murine interferon-y; hIFN-a, human interferoninteralpha; hIFN-yR, human interferon-y receptor; mIFN-yR, murine feron-y receptor; MHC, major histocompatibilitycomplex; EC, extracellular; IC, intracellular; TM, transmembrane; IL, interleukin.

MATERIALS AND METHODS

Reagents-RecombinanthIFN-y and mIFN-yweresuppliedas highly purified proteins (Genentech, Inc., manufacturing group). Recombinant hIFN-aaa was from Roche Pharmaceuticals (Nutley, NJ).

15533

15534

Interferon-? Receptor Extracellular Domain

Radioiodinated hIFN-y (specificactivity 18 pCi/pg) was providedby Avi Ashkenazi (Genentech, Inc.). Purified mAb208, a murine monoclonal antibody specific forthe hIFN-yR (Sheehan et al., 1988),and polyclonal rabbit sera generated against the purified hIFN-yR EC domain were provided by R. Schreiber (Washington University, St. Louis, MO). purified murine monoclonal antibody H-2Kk(anti-H-2Kk),specific formurine MHC class Iantigen, was purchased fromBecton-Dickinson (Mountain View, CA). Fluorescein isothiocyanate-conjugatedgoat antimouse IgG, PE-conjugated goat anti-mouse IgG, and phycoerythrinconjugated goat anti-rabbit IgG were purchased from Caltag (South San Francisco, CA). Cells and Cell Culture-The fibroblast cell lines SCC16-5 (Janssenet al., 1986)and WA17 (Raziuddinet al., 1984)are murine-human somatic cell hybrids that contain human chromosome 21 (one and three copies, respectively)as theonly human chromosome (providedby R. Schreiber, Washington University, St. Louis, MO). Cells were grown as described (Gibbs et al., 1991). Plasmid Construction-AllIFN-yR constructs were made in the mammalian cell expression vector pRK5. The IFN-yR EC domain mutants areinterspecies hybrids which werecreated by replacing regions of the hIFN-yR EC domain with the corresponding sequence from the mIFN-yR by standard methods. The nomenclature chosen indicates the amino acids of the hIFN-yR (inclusive)that have been replaced. Numbering is from the initiator methionine of the unprocessed form of the receptor (see Fig. lA). For all constructs, the sequence of DNA derived from synthetic oligonucleotides or by polymerase chain reaction was confirmed bydideoxy sequencing of supercoiled DNA (Amersham Corp.). Expression of Mutant IFNy Receptors in Mammalian Cells-For transient expression of mutant IFN-y receptors, WA17 cells (8 x lo5/ 10-cm plate) were transiently transfected by the DEAE-Dextran method as described (Ausubel et al., 1987). To analyze receptors by radioligand binding or cytofluorometric analysis, cells were harvested with EDTA 48 h after transfection. To analyze function of the receptor, cells were removed with EDTA from the 10-em plate 24 h after transfection and replated on two six-wellplates. Experiments were initiated in fresh media 24 h later. For stable expression of mutant IFN-y receptors, SCC16-5 cells(5 x 106/10-cmplate) were cotransfected by a calcium phosphate precipitation technique (Chen and Okayama, 1988) with 25 pg of expression plasmid and 0.5 pg of pRKneo DNA. After 48 h, transfectants were selected in 600pg/ml of G418 (Geneticin, LifeTechnologies,Inc.). Receptor-positive cells were isolated as described previously (Gibbs et al., 1991). The selected cells were expanded and sorted two to three additional times to obtain stable pools of receptor-positive cells, which were then maintained in 500 pg/ml G418. Radioligand Binding-Cells transiently transfected with receptor constructs were harvested and binding was performed in triplicate with 200 PM lZ5I-hIFN-yas described (Gibbset al., 1991). Nonspecificbinding was determined in triplicate by the addition of 290 nM unlabeled hIFN-y to the above reaction. For each construct, specific binding was calculated by subtracting the average nonspecific counts bound from the average total counts bound. To compare the relative binding ability of each receptor construct, the specific counts bound were dividedby the percentage of cells expressing receptor. The percent of receptor positive cells was determined by immunofluorescence flow cytometry (BectonDickinson FACScan)after stainingwith a rabbit polyclonal sera against the EC domain of the hIFN-yR (1:lOOO dilution) followedbyphycoerythrin goat anti-rabbit and varied between 10 and 27%. The relative binding ability of each receptor construct was then expressed as a percentage of the binding obtained with a wild-type EC receptor (HHM). Analysis of Receptor Affinity for hIFN-y by Competitionwith mAb208-mAb208 competes with hIFN-y for binding to the hIFN-yR (Sheehan et al., 1988). We determined the relative affinity of several receptors for hIFN-y by measuring the ability of hIFN-y to block binding of mAb208. Since this method assumes that each of the receptors has the same affinity for mAb208, we first measured the relative affinity of the receptors for mAb208. The amount of mAb208 bound to cells following incubation with various amounts of mAb208 in a lOO-pl reaction was determined by immunofluorescence flow cytometry. To obtain values for the amount of mAb208 bound, the mean fluorescence intensity of cells to which no mAb208 had been added was subtracted from each mean fluorescence intensity measured. Binding for each amount of mAb208 is expressed as a percentage of the maximal binding achieved (which was a t 2.4 pg/ml mAb208). IC,, values were determined from a four parameter curve fitof the data. To measure the ability of hIFN-y to compete for binding with mAb208, aliquots of 5 x

10' cells were incubated in 100 pl of 12%fetal bovine serdphosphatebuffered saline with a range of hIFN-y concentrations for 1h on ice,and then stained with mAb208 (2.4 pg/ml) followed by phycoerythrin-goat anti-mouse IgG. The mean fluorescent intensity measured in the absence of mAb208 was subtracted from each mean fluorescent intensity measured. The amount ofmAb208 bound a t each concentration of hIFN-y is expressed as a percentage of the maximal amount of mAb208 bound (the amount of mAb208 bound in the absence of hIFN-y). IC,, values were determined from a four parameter curve fit of the data. Functional Analysis-Expression ofMHC class I antigen following induction by hIFN-y, mIFN-y, or hIFN-cu2a was assayed by immunofluorescence flow cytometry afterstaining with anti-H-2Kk and fluorescein isothiocyanate-goat anti-mouse as described (Gibbs et al., 1991). To determine the dose response to hIFN-y, cells were treated with concentrations of hIFN-y between 0.029and 2900 m. Media alone was used to measure the basal level of MHC class I antigen expression, ) used t o measure and asaturating concentration of mIFN-y (290n ~was maximal MHC class I antigen expression. To obtain a value for the level of induction, the mean fluorescence intensity with media alone was subtracted from each of the mean fluorescence intensities after treatment with hIFN-y The induction at each hIFN-y dose is expressed as a percentage of the maximal MHC class I antigen induction achieved with mIFN-y. RESULTS

To identify region(s) of the hIFN-yR EC domain that are responsible for the receptor's species-specific interaction with ligand and with the accessory component, mutant IFN-yreceptors wereconstructed. Regions of the EC domain of the hIFN-yR were replaced with thecorresponding sequenceof the mIFN-yR. For the first setof mutant receptor constructs, five regions that together covered sequences throughout the EC domain were arbitrarily chosen for replacement. The regions replaced were amino acids 1-100 (inclusive, numbering from the initiatormethionine of the hIFN-yR) in thehybrid receptor designated HM1-100;100-132 in HM100-132; 134-209 in HM134-209; 183-209 in HM183-209; and 183-245 in HM183245 (Fig. 1).These receptors all have murine IC and TM domains, except HM134-209 and HM183-209 which have the human TM domain. We have shown previously that changing the IC or IC and TM domains of the hIFN-yR to thecorresponding murine sequence (forming the HHM and HMM receptors, respectively) does not affect affinity for hIFN-y and facilitates higher receptor expression (Gibbs et al., 1991). Binding of hIFN-y to Hybrid Human-Murine ZFN-y Receptors-The hybridreceptor constructs were transiently transfected into WA17, a mouse cell line that contains human chromosome 21, but lacks the hIFN-yR. Cell surface expression of all receptor constructs was verified by immunofluorescence flow cytometry using a polyclonal antibody against the hIFNyR. All constructs were transiently expressed on 10-27% of the cells. Binding of lZ5I-hIFN-yto the transiently transfected cells indicated that HM1-100, HM100-132, and HM183-245 did not bind hIFN-y detectably. On the other hand, HM134-209 and HM183-209 bound hIFN-y (Fig. 1). Analysis of Hybrid IFN-y Receptors That Bind hZFN-y" Since HM134-209 and HM183-209 can bind hIFN-y, we studied these receptors further to determine whether they were a functional response to hIFN-y. Pools of stable able to mediate clones that express HM134-209, HM183-209, or the native hIFN-yR were established SCC, in a murine cell line containing a single copy of human chromosome 21. The SCC cell line does notexpressthe hIFN-yR andthereforecannot respond t o hIFN-y. However, upon transfection with pRK-H-yR, this cell expression of MHC class Iantigen line (HgR) is able to increase in response to hIFN-y (Fig. 2). An increase in MHC class I antigen expression in response to hIFN-y was seenfor HM183209 but notfor HM134-209 (Fig. 2 B ) , even though both receptors were expressed at a high level (Fig. 2 A ) . Cells expressing HM134-209 were able to induce MHC class I antigen expres-

Interferon- y Extracellular Receptor

15535

Domain

hIFN-y as well as HM183-209 (Fig. 1).lz5I-hIFN-y bindingto the stable cell lines expressing these receptor constructs gave similar results to those obtained with the transiently transfected cells. Scatchard analysis with theHgR cell line gave the expected affinity (Kd= 1.5 x 10-lo),and affinities 10-fold lower could be measured but the lower affinity of the HM134-209 mutant receptor could not be measured by this technique. To determine the relativeaffinities of HM134-209, HM183209, and HgR for hIFN-y, we measured theability of hIFN-y to compete with the anti-hIFN-yRmAb208 for binding to the receptor. This method of measuring the affinity for hIFN-y requires that mAb208 bind each of these receptors with equal EC affinity. mAb208 does bind these three receptors with equal affinity (EC,, = 55-90 pg/ml) (Fig. 3 A ) . The competition between hIFN-y and mAb208 is shown in Fig. 3B. The IC,, values TM for hIFN-7 were 0.80 nM for HgR, 33 nM for HM183-209, and 750 IIM for HM134-209. To determine whether HM134-209 did not function simply IC because of its reduced affinity for ligand, we measured dose responses for induction of MHC class I antigen expression by hIFN-y (Fig. 3C). For HM183-209, the reduction in function seen at lower hIFN-y concentrations (50% activity at 0.4 nM) A B C D E F can be explained by the reduced affinity of this receptor for hIFN-y. For HM134-209, only a slight response is detected and only at the highest hIFN-y concentration (3000 I“). Thus, an approximately 10,000 times higher concentration of hIFN-y is required to elicit a response from HM134-209 compared with HM183-209 even though there is only a 23-fold difference in their affinities for hIFN-y. Another way t o compare the functional response of these receptors is to measure theirability to signal at approximately equal receptor occupancy. This was performed by treating HgR with hIFN-y at 1 nM, HM183-209 with 30 nM hIFN-y, and HM134-209 with 1000 I” hIFN-y. Both HgR and HM183-209 were fully functional, whereas HM134-209 was unable toinduce MHC class I expression (Fig. 2 B ) . Thus, HM134-209 appears to be defective in another aspect of receptor function,perhaps in its ability to dimerize or to interact productively with the chromosome 21-encoded accesA B C D E F vector FIG.1. Top, schematic representation of the first set of mutant IFNy sory component. Analysis of Residues 134 to 183 of the hIFN-yR-Asecond set receptors with hybrid humadmouse EC domains. A, HMM; B , HMI100; C, HM100-132; D,HM134-209; E, HM183-209; F , HM183-245. of mutants was constructed to test whether thefunctional deBottom, binding of T-hIFN-y to cells transiently transfected with the fect in HM134-209 could be localized. The role of the C-termireceptor constructs. Means and standard deviations are shown for a representative experiment done in triplicate. Binding was compensated nal portion of the region had been examined in HM183-209, so for transfection efficiency as described under “Materials and Methods.” the additional mutants focused on the residues between 134 The transfection efficiencies were 23%(A,) 24% ( B ) , 16%(C); 23% (D); and 183. First, residues 134-183 were replaced with the corre27% (E);and 10% ( F ) . sponding sequence of the mIFN-yR to form HM134-183. The sion in response to mIFN-y indicating that the nonspeciesspecific components of the response were intact (Fig. 2 B ) . Furthermore, the inability of the HM134-209 containing cells to respond to hIFN-y was not due toloss of human chromosome 21, since another function that dependson human chromosome 21, induction of MHC class I antigen by hIFN-a2a, was intact in these cells (Fig. 2 B ) . However, HM134-209 may not bind

rniFNy FIG.2. Expressionand functional analysis of HgR (top), HM183-209 (center), and HM134-209 (bottom). Analysis is on stably transfected SCC16-5 cell populations sorted for high IFN-yR expression. Measurement of IFN-yR and MHC class I expression was by immunofluorescenceflow cytometry.A, receptor expression using mAb208. B , expression of MHC class I antigen after culture alone (dotted lines) or with IFN ligands (solid lines).hIFN-y, was added for 48 h at 1 n~ for HgR, 30 n~for HM183209, and 1000 MI for HM134-209. mIFN-y was added for 48 h a t 300 MI. hIFN-a2a was added for72 h at 1000 unitdml.

5

I 1

11

E

HM 3 183-209 =

[30nM]

hlFNa2a

A i :

I

3

I I

HM 134-209

Receptor Expression

i

:

.

[lo00 nM:

MHC Class I induction

Interferon-y Receptor Extracellular Domain

15536

A

loo

and 167-183 were replaced with the corresponding sequence 1 from the mIFN-yR to form HM134-153,HM153-167, and

mAb208 (ng/ml) n

I

I

B Y "

80 60 J

40

-

J"

20 -

0-

HM167-183, respectively. In addition a deletion that removed residues 156 through165was constructed and designated HA156-165. Residues 156-165 are not conserved between the human and murinesequences and map to an extended loop on the structural model proposed by Bazan (Bazan, 1990). This loop contains residues that are critical for ligand binding in three otherreceptors (IL-3-binding proteinAIC2A (Wang et al., 19921, IL2 receptor P-chain (Imler et al., 1992), and IL-6 receptor (Yawata et al., 1993)) that are distant relatives in the cytokine receptor family. The IFN-y receptors with changes in the 134-183 region were analyzedafter the constructs were transiently transfected into the WA17 mouse cell line containing humanchromosome 21. Allthe receptors with changes in this region bound mAb208 with similar affinity (Fig. 4B). HM134-153 and HM153-167 bound hIFN-ywith wild-type (HHM) affinity, HM167-183 bound hIFN-y with lower than wild-type affinity, and HA156165 bound hIFN-y with an affinity that was only slightly reduced relative to wild-type (Fig.4C).Since these receptors were able tobind hIFN-y,we tested whether theywere capable of transmitting a functional response to hIFN-y. HM134-153, HM153-167, HM167-183, and HA156-165 all induced expression of MHC class I antigen inresponse to hIFN-y. HM167-183 did not respond as well as the other receptor constructs, consistent with HM167-183's reduced ability to bind hIFN-y. Control HHM (wild-type) respondedto hIFN-y, whereas vector and HM134-183, which was unable to bind hIFN-y,did not respond to hIFN-y (Fig. 40).Thus, none of the residues between 134 and 183 is species-specific for functional signaling by the hIFNyR. The functional defect observed for HM134-209, which was not duesolely to loss of affinity forhIFN-y could not be mapped t o any particularresidues. This defect could be due to either an overall conformational change in receptor structure or to the loss of interactions inmore than one of the subregions. DISCUSSION

FIG.3. Dose responses for HgR (squares),HM163-209 (circbs), and HM134-209 (triangles). Analysis is on stably transfected SCC16-5 cell populations sorted for high IFN-y receptor expression with polyclonal anti-hIFN-yR antibody.All measurements were by flow cytometry, and values were calculated as described under "Materials and Methods." A, binding of mAb208 at a range of concentrations, shown as a percent of maximal mAb208 binding. B, blocking of mAb208 binding sites by hIFN-y. The amount of mAb208 bound at concentrations of hIFN-y between 0.3 and 1000 rm is shown as a percent of the amount of mAb208 bound in the absence of hIFN-y. C , induction of MHC class I antigen by hIFN-y concentrations between 0.3 and 1000 rm is shown as a percent of the maximum MHC class I antigen induction that was measured after induction with mIFN-y.

affinity of this receptor construct for hIFN-y was tested in transiently transfectedWA17 cells by assaying displacement of mAb208 by hIFN-y as described above. Results for HM134-209 and HM183-209 were indistinguishable from the results described above withstablytransfected SCC cells (datanot shown). HM134-183 bound mAb208 with the same affinity as a wild-type receptor construct (HHM) (Fig.4B),but its binding t o mAb208 wasnot inhibited by hIFN-y(Fig.4C). Thus, HM134-183 does not bind hIFN-7 detectably,even though HM134-209, which replaces a n additional 26 residues, does. This suggests that the smaller replacement in HM134-183 does not simply change residues that make important contacts with hIFN-y, but changes the conformation of the receptor so that it no longer binds hIFN-y. An alternative approach taken t o analyze amino acid residues 134-183 was t o replace smaller groups of amino acids within this region (Fig. 4A).Three regions 134-153, 153-167,

The human and murine IFN-yR EC domains share 50% amino acid sequence identity, but are species-specific for interaction with both IFN-y and theaccessory component (encoded on human chromosome 21 or murine chromosome 16). We constructed hybrid murinehuman IFN-yR EC domains to determine regions important for these species-specific interactions. Of the 10 mutant IFN-y receptors constructed, sixretained the ability to bind hIFN-y at least partially, whereas four were unable t o bind hIFN-y. The six receptors able to bind hIFN-y were tested further t o determine whether they were able to mediate a functional response, induction of MHC class I antigen. An inability to mediatea functional response mayindicate an inability to interact with the accessory component, for which there is no biochemical assay at this time. The six IFN-yR mutants that were able tobind hIFN-y all differ from the wild-type hIFN-yR between residues 134 and 209. None of the mutants we constructed within this region affected the ability of the receptor to mediate a functional response, but, interestingly, changing the entire region 134-209 to the corresponding murine sequence interfered with the receptor's ability to signal. The inability to signaldoes not appear to becaused by a defect in theability of the receptor to dimerize,since qualitatively both HM183-209 and HM134-209 formed dimers which could be detected by cross-linking analysis (performed as in Greenlund et al., 1993; data not shown) One scenario consistent with these results is that thespeciesspecific accessory component makes multiple contacts in the 134-209 region. Thus, theaccessory componentheceptor interaction would be largely unaffectedby small changes within the

Interferon- y Receptor Extracellular Domain

C

15537

D

10-2

10-1

100

101

102 104

103

0 Media IhlFNr

a

b

c

d

e

f

vector

hlFNy (nM) FIG.4. Characterization of mutant IF’N-y receptors in transiently transfectedWA17 cells. All measurements were by flow cytometry, and values were calculated as described under ”Materials and Methods.” In A and D , receptors are labeled as a-f, and in B and C, these receptors are assigned symbols. HHM ( a , open squares) HM134-183 ( b , solid squares) HM134-153 (c,open circles) HM153-167 ( d ,closed circles) HM167183 (e, open triangles) HA156-165 ( f , closed triangles). A, schematic representation of the second set of mutant IFN-y receptors. B , binding of mAb208 as in Fig. 3 A . C, blocking of mAb208-binding sites by hIFN-7 as in Fig. 3B D,induction of MHC class I antigen by media, hIFN-y, or mIFN-y. Mean channel shifts are shown.

134-209 region, but inhibited by changing the entire region. Replacement of 134-183 resulted in a receptor with no detectable ligand binding activity, a puzzling result given that replacement of the larger region 134-209 only impaired binding partially. These two results are understandable if region 134-183 must interact correctly with region 183-209 in order to form a receptor capable of binding ligand. Specifically,if murine region 134-183 recognizesspecies-specific determinants in region 183-209, then replacing just 134-183 would cripple the receptor, whereas replacing the entire 134-209 region would be tolerable. Furthermore, human region 134-183 may interact with conserved determinants within region 183209, thus permitting functional replacement of region 183-209 with murine sequence. Ligand binding was disrupted by several different replacements, suggesting that the species-specific determinants for ligand binding may map to multiple regions. However,it isalso possible that the replacements (particularly the larger ones) disrupt the overall natural conformation of the receptor. Although two mutants unable to bind hIFN-y, HMlx100, and HM183-245, contain large replacements (100 and 62 amino acids, respectively), a third mutant,HM100-132, that was unable t o bind hIFN-y, had only a few residues replaced. The changes in HM100-132 consist of six conservative replacements and 5 amino acid replacements that change electrical charge (A112D, E117K, R123L, D124K, and K131G), suggesting that some of these charged residues may form speciesspecific salt bridges with ligand. The location of these critical charged residues in a structural model for the hIFN-yR de-

scribed below is consistent with the ability of these residues to interact with ligand. The structural model proposedby Bazan (1990)predicts that the IFN-yR foldsinto two tandem barrelsthat arepacked with their axis at anangle to form a V-shaped crevice.Each barrel is comprised of seven P-strands and the loops connecting them. This structural model applies to a large family of receptors and is supported by a crystallographic study of one family member, the human growth hormone receptor (De Vos et al., 1992). Structure-fimction studies of four family members (the IL-2 receptor (p-chain) (Imler et al., 19921, the IL-3 receptor (Wang et al., 1992), the IL-6 receptor (Yawata et al., 19931, and the growth hormone receptor (Bass et al., 1991) and co-crystallization of growth hormone with its receptor (De Vos et al., 1992) support Bazan’s prediction that ligand binds in the V-shaped crevice, although apparently the location of critical binding residues varies between these receptors. The IFN-yR is a distant cousin of these other receptors even though it belongs to the same broad structural family. Thus, it is of interest to determine whether ligand binding determinants are similarly located in the IFN-yR. The location of amino acids for the hIFN-yR are shown upon the predicted structure in Fig. 5. The species-specific charged residues noted above (between 100 and 132) are located in positions spanning the lastp-sheet of the first barrel(G in Fig. 5). The growth hormone receptor has two charged residues important for ligand binding in this vicinity (Bass et al., 1991). The IL-2, IL-3,and IL-6 receptors have important binding residues in theloop between By and Cy. This By-Cy loop(residues

15538

Interferon-y Receptor Extracellular Domain

C domain

N.domain 153

100

245

A B E D C F G FIG.5. Structural model for the IFN-yR EC domain. Zbp, location of amino acid residues changed in the hybrid receptor constructs are shown on the p-stranded structure proposed by Bazan. The model shows twodomains, each comprised of seven /3-strands and the loops connecting them. Bottom, each set of seven p-strands ispredicted to forma barrel, and the barrels are predicted to pack with axis at an angle, thereby creating a V-shapedcrevicewhichmayform the binding site for IFN-y (Bazan, 1990). IFN-y is a dimer and is predicted to bind receptor through its conservedpolycationic C terminus and a Species-specific determinant in theN terminus (Griggs et al., 1992). Consistent with these models for IFN-y and its receptor, appropriate sequences are present in the receptor crevice (p-sheet G and loop Dy-Ey)to interact with ligand as modeled.

A' 6' E' D' C' F' G'

""'3

+ +++

Cytoplasm COOH

153-167) is extended in the IFN-yR, but appears to be unimportant for either ligand binding or signaling by the receptor, since deletion of residues 156-165 affected neither activity. HM134-209 was still able to bind ligand, indicating that none of the residues from Ay through Ey participates inan essential species-specificinteraction with ligand. However,weak speciesspecific interactions with ligand may occur between 167 and 183 (Cy) and between 183 and 209 (Dy through Ey). The IFN-y ligand exists as a dimer with the N terminus of one molecule located adjacent to the C terminus of the other molecule (Ealick et al., 19911, so that the potential N- and C-terminal binding regions of the ligand are located adjacent to each other. A structure-function study of hIFN-y has suggested a model in which the N terminus of the ligand binds speciesspecific determinants of the receptor, and the polycationic C

terminus of the ligand binds a separateconserved epitope of the receptor (Griggs et al., 1992). Since we have shown that replacingthe species-specific charged residue(s) located around p-sheet G of the receptor (amino acids 112, 117, 123, 124, or 131) interferes with ligand binding, these residues may interact with species-specific determinants in the ligand, perhaps with charged residue(s) within the IFN-y ligand N terminus. Interestingly, a conserved polyanionicloop in the receptor, residues 193-201 (loop Dy-Ey),is located nearby in the structural model for the receptor and could provide a nonspecies-specific binding site for the cationic tail of the ligand (Fig. 5). This study has identified regions of the hIFN-yR EC domain that are important for function, suggesting how this receptor may interact with its ligand. Furthermore, alarge region of the EC domain was found to participate in some aspect of IFN-y

Interferon- y Receptor Extracellular Domain signal transduction in addition to ligand binding, possibly interaction with the accessory (chromosome 21 encoded) component. The framework developed in this study will aid future structure-function studies which can test smaller regions and specific residues tofurther our understanding of how the IFN-yR interacts with its ligand and accessory component to initiate intracellular signals. Acknowledgments-We are grateful to Robert Schreiber for helpful discussions and advice. We thank Jeff Hooley and Jim Chin for help with the flow cytometry experiments. We also thank Kerrie Andow for computer graphics workandTracey Rivas for help in preparing the manuscript.

REFERENCES Aguet, M., Dembic, Z., and Merlin, G. (1988) Cell 56, 273-280 Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seidman, J. G., Smith, J. A,. and Struhl, K. (1988) Current Protocols in Molecular Biology, Greene Publishing, New York Bass, S. H., Mulkemn, M. G., and Wells, J. A. (1991) Proc. Natl. Acad. Sei. U. S. A. 88,44984502 Bazan, J. F. (1990) Proc. Natl. Acad. Sci. U. S. A. 87, 6934-6938 Cao, Y. (1990) Technique 2, 109-111 Chen, C. A,, and Okayama, H. (1988)BioTechniques 6,632-638 Cook, J. R., Jung, V., Schwartz, B., Wang. P.,and Pestka, S. (1992)Proc.Natl. Acad. Sci. U. S. A. 8 9 , 11317-11321 Cunningham, B. C., Jhurani, P., Ng, P., and Wells, J. A. (1989) Science 243,13301536 ~. "

Dalton, D. K., Pitts-Meek, S., Keshav, S., Figari, I. S., Bradley, A., and Stewart, T. A. (1993) Science 269, 1739-1742 De VOS,A. M., Ultsch, M., and Kossiakoff, A. A. (1992) Science 248, 739-742 Dighe, A. S., Farrar, M. A., and Schreiber, R. D. (1993) J . Biol. Chem. 268,1064510653 Ealick, S. E., Cook, W. J., Vijay-Kumar, S., Carson, M., Nagabhushan, T. L., Trotta, P. P., and Bugg, C. E. (1991) Science 252,698-702 Farrar, M. A., and Schreiber, R. D. (1993)Annu. Reu. Immunol. 11,571411 Farrar, M. A,, Femadez-Luna,J., and Schreiber, R. D. (1991) J . Biol. Chem. 266, 19626-19635 Farrar, M. A,, Campbell, J. D., and Schreiber, R. D. (1992) Proc. Natl. Acad. Sci. U. S. A. 8 8 , 11706-11710

15539

Fountodakis, M., Juranville, J., Stuher, D., Weibel, E. K., and Garotta,G. 11990) J. Biol. Chem. 266,13268-13275 Fountoulakis, M., Lahm, H., Maris, A., Friedlein, A., Manneberg, M., Stueber, D., and Garotta, G. (1991) J. Biol. Chem. 266,14970-14977 Fountoulakis, M., Zulauf, M., Lustig, A,, and Garotta, G. (1992) Eur. J . Biochem. 208,7811-787 Gibbs, V. C., Williams, S. R., Gray, P.W., Schreiber, R. D., Rice, G., and Goeddel, D. V. (1991) Mol. Cell. Biol. 11, 5860-5866 Gray, P-W., b o n g , S., Fennie, E. H., Farrar, M. A,, Pingel, J. T., Fernandez-Luna, J., and Schreiber, R. D. (1989) Proc. Natl. Acad. Sci. U. S. A. 86,8497-8501 Greenlund, A. C., Schreiber, R. D., Goeddel, D. V., and Pennica, D. (1993) J . B i d . Chem. 268,18103-18110 Griggs, N. D., Jarpe, M. A,, Pace, J. L., Russell, S. W., and Johnson, H. M. (1992) J. Immunol. 149,517-520 Hemmi, S., Peghini, P., Metzler, M., Merlin, G., Dembic, Z., and Aguet, M. (1989) Proc. Natl. Acad. Sci. U. S. A. 86, 9901-9905 Hemmi, S., Merlin, G., and Aguet, M. (1992) Proc. Natl. Acad. Sci. I l . S. A. 89, 2737-2741 Hibino,Y., Mariano, T. M., Kumar, C. S., Kosak, C. A.,and Pestka, S. (1991)J. Biol. Chem. 266,6948-6951 Huang, S., Hendriks, W., Althage, A., Hemmi, S., Bluethmann, H., Kamijo, R., Vilcek, J., Zinkernagel, R. M., and Aguet, M. (1993) Science 269, 1742-1745 Imler, J., Miyajima, A,, and Zurawski, G. (1992) EMBO J . 11,2047-2053 Janssen, J.w.G., Collard, J. G., Tulp,A., Cox, D.,Millington-Ward,A., and Pearson, P. (1986) Cytometry 7 , 4 1 1 4 1 7 Jung, V., Rashidbaigi, A,, Jones, C., Tischfield, J. A., Shows, T. B., and Pestka, S. (1987) Proc. Natl. Acad. Sci. U.S. A. 84,41514155 Kumar, C. S., Muthukumaran, G., Frost, L. J.,Noe, M.,Ahn,Y. H., Mariano, T. M., and Pestka, S. (1989) J. Bid. Chem. 264, 17939-17946 Lutfalla, G., Gardiner, K , and Uze, G . (1993) Genomics 16,366-373 Munro, S., and Maniatis, T. (1989) Proc. Natl. Acad. Sci. U.S. A. 66, 8248-9252 Raziuddin, A,, Sarkar, F. H.,Dutkowski, R., Shulman, L., Ruddle, F. H., and Gupta, S. L. (1984) Proc. Natl. Acad. Sci. U. S. A. 81, 5504-5508 Sheehan, K. C.F., Calderon, J., and Schreiber, R.D. (1988) J. Immunol. 140, 42314237 Stuber, D., Friedlein, A., Fountoulakis, M., Lahm, H., and Garotta, G. (1993) Biochemistry 32,2423-2430 Thoreau, E., Petridou, B., Kelly, P. A,, Djiane, J., and Mornon, J. P. (1991) FEBS Lett. 282, 26-31 Wang,H.,Ogorochi,T., Arai, K., and Miyajima, A. (1992) J . Biol. Chem. 267, 979-983 Yawata, H., Yasukawa, K , Natsuka, S., Murakami, M., Yamasaki, K., Hibi, M., Taga, T., and Kishimoto, T.(1993) EMBO J . 12, 1705-1712 Yon, J., and Fried, M. (1989) Nucleic Acids Res. 17, 4895