Receptor 2 Α the Human IFN- Ligand Binding and Signal Transmission In

Determination of Residues Involved in Ligand Binding and Signal Transmission in the Human IFN-α Receptor 2

This information is current as A. Chuntharapai, V. Gibbs, J. Lu, A. Ow, S. Marsters, A. of September 24, 2021. Ashkenazi, A. De Vos and K. Jin Kim J Immunol 1999; 163:766-773; ; http://www.jimmunol.org/content/163/2/766 Downloaded from

References This article cites 34 articles, 15 of which you can access for free at: http://www.jimmunol.org/content/163/2/766.full#ref-list-1

Why The JI? Submit online. http://www.jimmunol.org/

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

*average by guest on September 24, 2021

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 1999 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Determination of Residues Involved in Ligand Binding and Signal Transmission in the Human IFN-␣ Receptor 2

A. Chuntharapai,* V. Gibbs,† J. Lu,* A. Ow,† S. Marsters,* A. Ashkenazi,* A. De Vos,* and K. Jin Kim1*

The human IFN-␣ receptor (hIFNAR) is a complex composed of at least two chains, hIFNAR1 and hIFNAR2. We have performed a structure-function analysis of hIFNAR2 extracellular domain regions using anti-hIFNAR2 mAbs (1D3, 1F3, and 3B7) and several type I human IFNs. These mAbs block receptor activation, as determined by IFN-stimulated gene factor 3 formation, and block the antiviral cytopathic effects induced by type I IFNs. We generated alanine substitution mutants of hIFNAR2-IgG and determined that regions of hIFNAR2 are important for the binding of these blocking mAbs and hIFN-␣2/␣1. We further demonstrated that residues E78, W101, I104, and D105 are crucial for the binding of hIFN-␣2/␣1 and form a defined protrusion when these residues are mapped upon a structural model of hIFNAR2. To confirm that residues important for ligand binding are indeed Downloaded from important for IFN signal transduction, we determined the ability of mouse L929 cells expressing hIFNAR2 extracellular domain mutants to mediate hIFN signal. hIFN-␣8, previously shown to signal a response in L929 cells expressing hIFNAR1, was unable to signal in L929 cells expressing hIFNAR2. Transfected cells expressing hIFNAR2 containing mutations at residues E78, W101, I104, or D105 were unresponsive to hIFN-␣2, but remained responsive to hIFN-␤. In summary, we have identified specific residues of hIFNAR2 important for the binding to hIFN-␣2/1 and demonstrate that specific regions of the IFNAR interact with the subspecies of type I IFN in different manners. The Journal of Immunology, 1999, 163: 766–773. http://www.jimmunol.org/

ype I IFNs are a family of cytokines defined by their there may be additional species-specific components required for antiviral activities. Human IFNs include at least 14 sub- the antiproliferative response (12). Recently, Petricoin et al. (13) T species of hIFN-␣,2 one hIFN-␤, one hIFN-␻, and one reported that the antiproliferative action, but not the antiviral ac- hIFN-␶ (1, 2). These type I hIFNs share a common receptor (IF- tion or the activation of the JAK-STAT pathway, of IFN-␣ re- NAR) (3, 4), which is composed of two chains, a 135-kDa ␣ sub- quires TCR signaling components. unit (hIFNAR1) (5) and a 115-kDa ␤ subunit (hIFNAR2) (6, 7). hIFNAR2 is a 515-aa protein composed of an ECD of 217 res- Three different forms of hIFNAR2 have been reported: a 40-kDa idues. The ECD of hIFNAR2 is composed of two domains (ϳ100 soluble form designated hIFNAR2a (6), a 55-kDa short form residues/domain), domain 1 and domain 2. IFN-mediated signaling by guest on September 24, 2021 known as hIFNAR2b (6), and a 115-kDa long form known as is initiated by ligand-induced receptor dimerization via the ECD, hIFNAR2c (7). These three forms are derived by alternative splic- tyrosine phosphorylation of the Tyk2 and Jak1 tyrosine kinases, ing of the same gene. Only hIFNAR2c mediates a biological re- and subsequent phosphorylation of the Stat1 and Stat2 proteins. sponse when associated with hIFNAR1 (5) and is the form of the Activated STATs translocate to the nucleus as an IFN-stimulated receptor we have utilized in this study. response element 3 (ISGF3) complex and induce the transcription When hIFNAR1 is expressed alone in mouse cells, there is no of IFN-stimulated genes (13). There are multiple IFNs in the type significant IFN-␣ binding; however, the expression of hIFNAR2 I IFN family that initiate receptor dimerization. To understand how alone produces a low affinity ligand-binding receptor (0.5–1 nM). the same receptor interacts with these different IFN subtypes, in this The coexpression of hIFNAR1 and hIFNAR2 results in a high study we have investigated the interaction of type I IFNs with hIF- affinity receptor complex (10–100 pM) (8–10). These results dem- NAR2 using soluble hIFNAR2-IgG immunoadhesin and blocking onstrate that hIFNAR2 is the ligand-binding subunit, but hIFNAR1 mAbs. Using alanine-scanning mutagenesis, we have determined res- contributes to the formation of a high affinity receptor. It has been idues on hIFNAR2 that are important for type I IFN binding. We have shown that the coexpression of hIFNAR1 and hIFNAR2c in a extended the binding data by transfecting wild-type and mutant hIF- murine background can mediate the antiviral response of human NAR2 cDNAs into murine L929 cells and have studied the effect of IFNs, but not an antiproliferative response (11). This suggests that several type I IFNs on mediating signal transduction.

*Departments of Antibody Technology, Molecular Oncology, and Protein Engineer- Materials and Methods ing, Genentech Inc., South San Francisco, CA 94080; and †Department of Surgery, San Francisco Veterans Affairs Medical Center, San Francisco, CA 94121 Production of soluble hIFNAR1-IgG and various species of type Received for publication January 29, 1999. Accepted for publication May 4, 1999. 1 IFN The costs of publication of this article were defrayed in part by the payment of page hIFNAR2-IgG and various subspecies of hIFN-␣ were prepared as de- charges. This article must therefore be hereby marked advertisement in accordance scribed (14) with the following modification: A c-DNA encoding the hIF- with 18 U.S.C. Section 1734 solely to indicate this fact. NAR2-IgG molecules was constructed based on the ECD (residues 1–216) 1 Address correspondence and reprint requests to Dr. K. Jin Kim, Department of of hIFNAR2. hIFNAR2-IgG was expressed in 293 cells and the immuno- Antibody Technology, Genentech Inc., South San Francisco, CA 94080. E-mail ad- adhesin was purified using a protein A column. Human IFN-␣2/␣1 dress: [email protected] (IFN-␣2 residues 1–62/␣1 residues 64–166) (15, 16) were a gift from Dr. ␤ 2 Abbreviations used in this paper: hIFN, human IFN; ECD, extracellular domain; M. J. Brunda (Hoffman-LaRoche, Nutley, NJ). hIFN- was obtained from hIFNAR, hIFN-␣ receptor; ISGF3, IFN-stimulated gene factor 3; RT, room Sigma (St. Louis, MO). The specific activities of the various type I IFNs temperature. are as follows: IFN-␣2/␣1(2ϫ 107 IU/mg), IFN-␣1(3ϫ 107 IU/mg),

Table I. General characteristics of mAbs to hIFNAR2

Afﬁnityf a b c d e Ϫ1 mAb Isotype Cytometry Epitope Western Blot IP Kd (pM) 1D3 IgG2a ϩϩ A ϩ ND 242 1F3 IgG2a ϩϩ B Ϫϩ5 3B2 IgG1 ϩϩ B ϪϩND 3B7 IgG2a ϩϩ C Ϫϩ1

a The mAb isotype was determined using an isotype-speciﬁc goat anti-mouse Ig. b All mAbs were selected for positive staining of a 9D human B cell line. c mAbs were shown to recognize different epitopes by a competitive binding ELISA. d The immunoblot was performed using hIFNAR2-IgG reduced with DTT. e U266 cells were biotinylated using NHS-LC-biotin and lysed with 1% Nonidet P-40. Biotinylated hIFNAR2 was precipitated by mAbs bound to protein-G-4B Sepharose and separated on a 7.5% SDS-PAGE gel. Biotinylated hIFNAR2 transferred onto nitrocellulose paper was detected by HRP-strepavidin. f The afﬁnity of mAbs were determined using the KinExA system.

IFN-␣2(2ϫ 107 IU/mg), IFN-␣5(8ϫ 107 IU/mg), IFN-␣8 (19 ϫ 107 the unbound mAbs. The anti-IFNAR2 mAbs bound to beads were detected IU/mg), and IFN-␤ (1.5 ϫ 105 IU/mg). by 1.5 ml of PE-labeled goat anti-mouse IgG. Unbound labeled material

was removed by drawing 4.5 ml of 0.5 M NaCl through the bead pack over Downloaded from Generation of mAbs to hIFNAR2 a 3-min period. The equilibrium constant was calculated using the software provided by the manufacturer (Sapidyne). BALB/c mice were immunized with 2.5 ␮g of hIFNAR2-IgG into each hind footpad, and mAbs were generated as described (17). Three days after the final boost, popliteal lymph nodes were fused with myeloma cell line Electrophoretic mobility shift assay (EMSA) ϫ P3 63Ag.U.1 (18). Culture supernatants were initially screened for their HeLa cells (5 ϫ 105 cells) were incubated with each hIFN-␣ (25 ng/ml) in ability to bind to hIFNAR2-IgG, but not to CD4-IgG in a capture ELISA, 200 ml of DMEM for 30 min at 37°C. In experiments using the anti- as described previously (19). The selected mAbs were further tested for hIFNAR2 mAbs, cells were incubated with 5–500 ␮g/ml for 15 min at 4°C http://www.jimmunol.org/ their ability to block ligand-receptor binding in a capture ELISA and for before addition of the hIFN-␣. Cells were lysed using 0.025% Nonidet their ability to recognize cell membrane receptors on U266 cells by flow- P-40, and gel-shift assays were performed using a 32P-labeled double- cytometric analysis, as described (19). After cloning the selected hybrid- stranded oligonucleotide containing the ISG15 (19). DNA-protein com- omas twice, their Ag specificity as well as blocking activities were con- plexes were resolved in 6% nondenaturing polyacrylamide gels (Novex, firmed in a ligand receptor-binding assay. The Western blot and isotype San Diego, CA) and analyzed by autoradiography. The specificity of the analyses were done as described (20). assay was determined by the addition of 350 ng of unlabeled ISG15 probe Capture ELISA in separate reaction mixtures. Formation of an ISGF3-specific complex was confirmed by a supershift assay using anti-murine Stat1 or Stat2 Ab (Santa Microtiter plates (Maxisorb; Nunc, Kamstrup, Denmark) were coated with Cruz Biotechnology, Santa Cruz, CA).

50 ␮l/well of 2 ␮g/ml of goat Abs specific for the Fc portion of human IgG by guest on September 24, 2021 (goat anti-hIgG Fc; Cappel, ICN, Costa Mesa, CA) in PBS overnight at 4°C Antiviral assay and blocked with 2% BSA for1hatRT.Atotal of 50 ␮l/well of 50 ng/ml of hIFNAR2-IgG (or hIFNAR2-IgG mutant) was added to each well for The assay was done as described, in duplicate 96-well microtiter plates 1 h. Plates were then incubated with 50 ␮l/well of 2 ␮g/ml of anti- using human lung carcinoma A549 cells challenged with encephalomyo- hIFNAR2 mAbs (or 50 ng/ml of hIFNAR2-IgG mutants) for 1 h. Wells carditis virus (23). Serial dilutions of mAbs were incubated with various were then incubated with 50 ␮l/well of HRP goat anti-mouse IgG. The units of type 1 IFNs for1hat37°C in a total volume of 100 ␮l. These bound enzyme was detected by the addition of TMB (3,3Ј,5,5Ј-tetramethyl- mixtures were then incubated with 5 ϫ 105 A549 cells in 100 ␮l of medium benzidin) substrate, the reaction was stopped by the addition of stop solution for 24 h. Cells were then challenged with 2 ϫ 105 PFU of encephalomyo- (Kirkegaard & Perry Laboratories, Gaithersburg, MD), and the plates were carditis virus for an additional 24 h. At the end of the incubation, cell read at 450 nm with an ELISA plate reader. Between each step, plates were viability was determined by visual microscopic examination. The neutral- washed three times in wash buffer (PBS containing 0.05% Tween-20). izing Ab titer (EC50) was defined as the concentration of Ab that blocks 50% of the antiviral cytopathic effect by 100 U/ml of type I IFNs. The units Epitope mapping using a competitive binding ELISA of type I IFNs used in this study were determined using National Institute of Health reference human rIFN-␣2 as a standard. To determine whether mAbs recognized the same or different epitopes, a competitive binding ELISA was performed using biotinylated mAbs (Bio-mAb). mAbs were biotinylated using N-hydroxyl succinimide, as described (20). Generation of various hIFNAR2-IgG mutants The cDNAs encoding 1–216 residues of the extracellular domain of type 1 Determination of the affinities of mAbs hIFNAR2 were expressed as immunoadhesins (14). Alanine substitution The equilibrium dissociation and association constant rates of anti- mutants were generated according to the method of Kunkel (24). Mutant hIFNAR2 mAbs were determined using KinExA, an automated immuno- receptor IgGs were expressed transiently in human 293 cells. Transfected assay system (Sapidyne Instruments, Boise, ID), as described, with a mod- 293 cells were grown overnight in F-12:DMEM (50:50) containing 10% ification (21, 22). Briefly, 1 ml of anti-human IgG agarose beads (56 ␮m; FCS, 2 mM L-glutamine, 100 ␮g/ml of penicillin, and 100 ␮g/ml of strep- Sigma, St. Louis, MO) was coated with 20 ␮g of hIFNAR2-IgG in PBS by tomycin, and then were placed in a serum-free media. Three days later, gentle mixing at RT for 1 h. After washing with PBS, nonspecific binding culture supernatants were collected. For the hIFNAR2-IgG mutant analy- sites were blocked by incubating with 10% human serum in PBS for1hat sis, the concentrations of immunoadhesin molecules in 293 transfected RT. The blocked beads were diluted into 30 ml of assay buffer (0.01% culture supernatants were determined by ELISA using CD4-IgG as a stan- BSA/PBS). The diluted beads (550 ␮l) were drawn through the flow cell dard, and were adjusted to be 50 ng/ml. The degree of mAb binding to with 20-␮m screen and then washed with 1 ml of running buffer (0.01% these mutants was compared with the degree of mAb binding to the wild- BSA ϩ 0.05% Tween-20 in PBS). The beads were then disrupted gently type receptor using a capture ELISA. with a brief backflush of running buffer, followed by a 20-s setting period to create a uniform and reproducible bead pack. A bead pack (ϳ4mm Plasmid construction and DNA transfection into mouse high) was created in the observation flow cell. For equilibrium measure- L929 cells ments, mAbs (5 ng/ml in 0.01% BSA/PBS) were mixed with a serial di- lution of hIFNAR2-IgG (starting from 2.5 nM to 5 pM) and were incubated A 648-bp fragment, containing the entire coding region of the hIFNAR2 at RT for 2 h. Once equilibrium was reached, 4.5 ml of this mixture was cDNA, was inserted into the mammalian cell expression vector pRSV and drawn through the beads, followed by 250 ␮l of running buffer to wash out designated pRSVHAR2. cDNAs encoding mutant receptors were produced 768 LIGAND-BINDING REGION OF hIFNAR2

higher affinity than murine IgG2. From the results of competitive binding ELISA, we were able to group these seven mAbs into three groups depending upon regions of hIFNAR2-IgG they recognized (data not shown). Three mAbs, 1D3, 1F3, and 3B7, rep- resenting each group, were selected for our study, and the general characteristics of these three mAbs are outlined in Table I. mAb 1D3 recognizes the reduced form of hIFNAR2 in a Western blot assay (data not shown), suggesting that it may recognize a linear epitope, while mAbs 1F3 and 3B7 recognize conformational epitopes. Receptor affinities were determined using a KinExA system that allows the measurement of mAb affinities in a solution Ϫ1 phase. The affinities (Kd ) of mAbs 3B7, 1F3, and 1D3 are 1, 5, and 242 pM, respectively, demonstrating that these are relatively high affinity mAbs. The blocking ability of each mAb was determined in a competitive binding ELISA (Fig. 1). mAbs 1F3 and 3B7 at 0.6 nM (0.1 ␣ ␣ FIGURE 1. mAbs to hIFNAR2 inhibit the binding of hIFN- 2/ 1to ␮g/ml) were able to block Ͼ90% of hIFN-␣2/␣1 binding to the hIFNAR2-IgG, as determined by an ELISA. hIFNAR2-IgG was captured soluble receptor, hIFNAR2-IgG. mAb 1D3 showed no significant onto ELISA wells precoated with goat anti-human IgG Fc. Bio-hIFN- ␣2/␣1 plus various concentrations of mAbs were added into each well. The blocking activity even at a concentration 10 times higher, of 6 nM Downloaded from ␮ level of Bio-hIFN-␣2/1 bound was detected by the addition of (1 g/ml). To extend these findings in functional assays, we ana- HRP-streptavidin. lyzed the ability of mAbs 1D3, 1F3, and 3B7 to prevent receptor activation via the ISGF3 complex conformation by an EMSA and by an IFN-induced antiviral assay. in a repair deficient Escherichia coli strain using two oligonucleotide prim- In HeLa cells treated with 25 ng/ml of several type I IFNs, all ers and pRSVHAR2 as a template according to the manufacturer’s instruc- three mAbs at a concentration of 500 ␮g/ml prevent receptor ac- http://www.jimmunol.org/ tions (Clontech Laboratories, Palo Alto, CA). The accuracy of all cDNA tivation and subsequent ISGF3 complex formation (Table II). At was confirmed by supercoiled DNA sequencing with an automated DNA sequencing system. The murine fibroblast cell line L929 was cotransfected lower concentrations, only 1F3 and 3B7 are effective. These results with 1 ␮g of expression plasmid and 50 ng of pSVE neo DNA per dish by show that all three mAbs are blocking Abs, although mAb 1D3 is a liposome-mediated transfection technique (Superfect, Qiagen, Chats- relatively weak. The blocking activities of these mAbs were also worth, CA). Forty-eight hours after transfection, the cells were split and tested in an antiviral assay using A549 cells challenged with ECM transfectants were selected in G418. Twenty-four individual G418-resistant ␣ ␣ ␣ clones were analyzed for each construct. Human IFNAR2-expressing virus in the presence of 100 U/ml of hIFN- 1, hIFN- 2, hIFN- 5, clones were initially screened by RT-PCR using hIFNAR2-specific primers hIFN-␣8, hIFN-␣2/␣1, or IFN-␤ (Table III). Serial dilutions of because the mutations created in the ECD of hIFNAR2 might affect mAb mAbs in the concentration range of 0.1–30 ␮g/ml were tested in

binding to the receptor. Those clones that expressed mRNA for hIFNAR2 by guest on September 24, 2021 duplicate. The EC50 was arbitrarily determined as the mAb con- were then analyzed by flow cytometry using mAb 3B2 and PE-conjugated centration that neutralized 50% of the antiviral cytopathic effect goat anti-mouse IgG (Caltag, San Francisco, CA) to determine the level of membrane receptor expression. At least three clones of each construct were induced by 100 U/ml of type I IFN. All three mAbs blocked the tested for a functional response to several IFNs. activity of all type I IFNs tested, and the potency of these mAbs was variable depending upon the IFN subtypes tested. mAb 3B7 Յ ␮ Results demonstrated its blocking activity at an EC50 1 g/ml, while Anti-hIFNAR2 mAbs show differential blocking activities mAb 1F3 required 1–3 ␮g/ml to block activity. mAb 1D3 was able ␤ mAbs to the ECD of hIFNAR2 were generated using a soluble to block the activity of all IFNs tested, except IFN- , although a ϭ ␮ immunoadhesin as an Ag, as described in Materials and Methods. much higher Ab concentration (EC50 10–20 g/ml) was re- Initially, we selected 20 strong positive binding mAbs to hIFNAR2 quired. From these biological assays, we conclude that all three by an ELISA. After flow-cytometric analysis on 9D cells (a human mAbs are blocking Abs and mAb 3B7 is the most potent. B cell line), we selected seven mAbs that recognized membrane hIFNAR2. Six mAbs are of the IgG2 isotype and one, mAb 3B2, Determination of residues on hIFNAR2 that are important for is an IgG1. mAb 3B2 was used for routine flow cytometry to ligand binding evaluate receptor expression since our PE goat anti-mouse IgG To determine areas in the ECD of hIFNAR2 that are important for (Caltag, San Francisco, CA) appears to bind murine IgG1 with ligand binding, we generated alanine substitution hIFNAR2-IgG

Table II. Effects of anti-hIFNAR2 mAbs on ISGF3 formation in HeLa cellsa

1D3 (␮g/ml) 1F3 (␮g/ml) 3B7 (␮g/ml)

HeLa Cells 5 50 500 5 50 500 5 50 500

hIFN-␣2/␣1 ϪϪϩϩϩϩϩϩϩ hIFN-␣1 ϪϪ ϩϩ/Ϫϩ ϩϩϩ ϩ hIFN-␣2 Ϫϩ/Ϫϩ ϩ ϩϩϩϩϩ hIFN-␣5 ϪϪϩϩϩϩϩϩϩ hIFN-␣8 ϪϩϩϪϩϩϩϩϩ hIFN-␤ ϪϪϩϩϩϩϪϩϩ

a EMSA detecting ISGF3 complex formation during IFN activation was carried out using HeLa cells treated with 25 ng/ml of human type 1 IFNs plus 5–500 mg/ml of mAbs as described in Materials and Methods. Results were expressed as complete blocking (ϩ), partial blocking (ϩ/Ϫ), or no blocking (Ϫ). The Journal of Immunology 769

Table III. Effects of anti-hIFNAR2 mAbs on the antiviral effects of type I IFNsa

␮ EC50 of mAb ( g/ml)

mAb hIFN-␣2/␣1 hIFN-␣1 hIFN-␣2 hIFN-␣5 hIFN-␣8 hIFN-␤

1D3 20 10 20 10 20 NB 1F3 3 2 3 1 3 2 3B7 0.6 0.1 1 0.1 0.3 0.3

a The neutralizing Ab titer (EC50) was deﬁned as the concentration of Ab that neutralizes 50% of the antiviral cytopathic effects induced by 100 U/ml of type I IFNs on A549 cells. The experiment was done using serial dilutions of mAbs in the range of 0.1–30 ␮g/ml in duplicate. When mAbs at 30 ␮g/ml showed no blocking effect in this assay, we arbitrarily designated these as a nonblocking mAb (NB).

mutants. We initially selected 12 charged areas for analysis and binding area of mAb 1D3 is in a small area, including residues substituted clusters of two to seven residues with alanines (Table 133–139 and 153–157, in domain 2 of hIFNAR2. IV). Eleven of the twelve mutants could be expressed as immu- To deﬁne speciﬁc residues of hIFNAR2 that bind hIFN-␣2/␣1 noadhesins, as detected in an anti-human IgG-Fc ELISA. Mutant 8 and interact with the blocking mAbs, 26 single alanine mutants (residues 145–149: EEQSE/AAQSA) did not express at all, sug- were generated in the region encompassing residues 49–156 (Ta-

gesting that this region is important for maintenance of the struc- ble V). Compared with the wild-type binding, hIFN-␣2/␣1 lost Downloaded from tural integrity of hIFNAR2. The binding of hIFN-␣2/␣1 (50 nM) Ͼ75% binding to the hIFNAR2 mutants, D68A, E78A, W101A, and mAbs (5 nM) to hIFNAR2-IgG mutants (0.5 nM) was deter- I104A, and D105A. All of these mutants except D68A retained the mined in a capture ELISA. The degree of binding to the receptor high binding to the mAbs, demonstrating their structural integrity. mutants was compared with the degree of binding obtained with In contrast, the binding of all three mAbs to mutant D68A was the wild-type receptor. Alanine substitution mutants, 1 (7–11), 2 signiﬁcantly diminished. This suggests that D68 may be structur-

(29–33), 10 (159–163), 11 (172–173), and 12 (187–192), showed ally important, as mutations of it affect the binding of ligand and http://www.jimmunol.org/ significant binding (Ͼ70% of the wild-type control) to Bio-hIFN- all three mAbs. The binding ELISA was done by capturing 0.5 nM ␣2/␣1, while 3 (49–55), 4 (68–72), 5 (74–78), 6 (105–109), 7 hIFNAR2-IgG mutants onto an anti-human IgG-Fc-coated well, (133–139), and 9 (153–157) showed an 80% reduction in their followed by the incubation with 10-fold molar excess (5 nM) of ability to bind Bio-hIFN-␣2/␣1. Residues 49–157 are in the mid- Bio-hIFN-␣2/␣1. Under these conditions, we should be able to dle portion of hIFNAR2 occupying portions of domains 1 and 2, detect the majority of the receptor-ligand interactions if there is a suggesting that this area interacts with ligand. We also determined significant affinity. However, to confirm the result shown in Table the binding of the three anti-hIFNAR2 mAbs, 1D3, 1F3, and 3B7, III, we compared the binding of these IFNAR2-IgG mutants/wild to these 11 hIFNAR2 mutants. In comparison with the wild-type type to various concentrations (0.5–50 nM) of Bio-hIFN-␣2/␣1

hIFNAR2-IgG, all of these mutants show 35–100% binding to at (Fig. 2). Compared with the binding of wild hIFNAR2-IgG to by guest on September 24, 2021 least one of these mAbs, suggesting that these multiple mutants various doses of Bio-hIFN-␣2/␣1, mutants W101A, I104A, and retain the general structural integrity of hIFNAR2. When the ala- D105A, and E78A demonstrate Ͻ50% binding of the wild type at nine substitution, in particular residues of hIFNAR2-IgG, resulted various concentrations of IFN, while K49A and K54A show in Ͻ25% of the wild-type binding, we arbitrarily considered these Ͻ75% binding. From these results, we concluded that the most residues as being important for the mAb binding. The most potent crucial residues in IFNAR2 for the interaction with hIFN-␣2/␣1 blocking mAb, 3B7, recognizes an epitope of the receptor in a are E78, W101, I104, and D105, while K49 and K54 have some small area containing residues 49–55 and 68–72. The areas rec- inﬂuence on this interaction. Residues E78, W101, I104, and ognized by mAb 1F3 span a larger region in residues 49–55, 68– D105, which are important for binding to the ligand, comprise a 72, 74–78, 105–109, 133–139, and 153–156, which closely over- small region of the receptor according to our proposed computer laps with the ligand-binding region of hIFN-␣2/␣1. The main model (Fig. 5A).

Table IV. Binding of hIFN-␣2/␣1 and mAbs to multiple alanine hIFNAR2 mutantsa

% Wild-Type hIFNAR2 Binding

No. Residues Alanine Substitution IFN-2/1 1D3 1F3 3B7

1 7–11 DYTDE/AYTAA 73 Ϯ 10 105 Ϯ 13 96 Ϯ 1 101 Ϯ 13 2 29–33 ELKNH/ALANA 87 Ϯ 22 80 Ϯ 187Ϯ 682Ϯ 7 3 49–55 KPEDLK/APAALA 18 Ϯ 239Ϯ 16Ϯ 04Ϯ 0 4 68–72 DLTDE/ALTAA 16 Ϯ 138Ϯ 25Ϯ 03Ϯ 0 5 74–78 RSTHE/ASTAA 16 Ϯ 195Ϯ 116Ϯ 189Ϯ 2 6 105–109 DMSFE/AMSFA 19 Ϯ 265Ϯ 18Ϯ 140Ϯ 1 7 133–139 EEELQFD/AAALQFA 16 Ϯ 15Ϯ 19Ϯ 035Ϯ 1 8 145–149 EEQSE/AAQSA —b ——— 9 153 Ϯ 157 KKHKP/AAHAP 19 Ϯ 114Ϯ 114Ϯ 035Ϯ 1 10 159–163 EIKGN/AIAGN 92 Ϯ 589Ϯ 796Ϯ 593Ϯ 3 11 172–173 DK/AA 73 Ϯ 773Ϯ 591Ϯ 780Ϯ 3 12 187–192 HEWED/AASAAQ 82 Ϯ 382Ϯ 10 83 Ϯ 966Ϯ 10

a Wild-type and mutant hIFNAR2-IgGs (0.5 nM) were captured with goat anti-human IgG Fc reagent precoated onto microtiter wells. Biotinylated hIFN-␣2/␣1 (50 nM) or mAbs (5 nM) were allowed to interact for 1 h. After washing, the amounts of the ligand or mAbs bound were determined by the addition of HRP-streptavidin or HRP-goat anti-mouse IgG, respectively. b No expression. 770 LIGAND-BINDING REGION OF hIFNAR2

Table V. Binding of hIFN-␣2/␣1 and anti-hIFNAR2 mAbs to hIFNAR2-IgG mutantsa

% Wild hIFNAR2 Binding

Mutants hIFN-␣ 2/1 1D3 1F3 3B7 Polyclonal

K49A 68 Ϯ 10 101 Ϯ 1 104 Ϯ 699Ϯ 198Ϯ 13 E51A 95 Ϯ 9 103 Ϯ 4 101 Ϯ 467Ϯ 092Ϯ 3 D52A 97 Ϯ 11 101 Ϯ 1 107 Ϯ 683Ϯ 294Ϯ 15 K54A 53 Ϯ 187Ϯ 183Ϯ 287Ϯ 12 91 Ϯ 13 K57A 111 Ϯ 2 143 Ϯ 2 260 Ϯ 15 144 Ϯ 1 107 Ϯ 5 D68A 0 41 Ϯ 18Ϯ 18Ϯ 038Ϯ 1 D71A 113 Ϯ 26 100 Ϯ 291Ϯ 3 110 Ϯ 13 103 Ϯ 12 E72A 129 Ϯ 29 91 Ϯ 7 106 Ϯ 19 90 Ϯ 13 94 Ϯ 2 R74A 122 Ϯ 28 92 Ϯ 2 103 Ϯ 12 93 Ϯ 16 93 Ϯ 13 H77A 83 Ϯ 19 150 Ϯ 4 121 Ϯ 137 132 Ϯ 16 150 Ϯ 12 E78A 1 Ϯ 192Ϯ 354Ϯ 494Ϯ 289Ϯ 14 W101A 0 99 Ϯ 490Ϯ 298Ϯ 2 100 Ϯ 13 I104A 23 Ϯ 2 104 Ϯ 194Ϯ 6 104 Ϯ 199Ϯ 13 D105A 8 Ϯ 289Ϯ 362Ϯ 887Ϯ 297Ϯ 14 E109A 99 Ϯ 10 99 Ϯ 4 107 Ϯ 19 99 Ϯ 4 100 Ϯ 17 E133A 99 Ϯ 992Ϯ 396Ϯ 12 147 Ϯ 580Ϯ 12 E134A 86 Ϯ 387Ϯ 182Ϯ 388Ϯ 12 92 Ϯ 3 Downloaded from E135A 72 Ϯ 288Ϯ 174Ϯ 180Ϯ 12 90 Ϯ 1 Q137A 85 Ϯ 12 63 Ϯ 155Ϯ 461Ϯ 14 82 Ϯ 1 D139A 109 Ϯ 187Ϯ 100 Ϯ 4 139 Ϯ 589Ϯ 14 E145A 78 Ϯ 984Ϯ 282Ϯ 12 93 Ϯ 12 91 Ϯ 11 E146A 87 Ϯ 493ϮϮ 92 Ϯ 398Ϯ 12 97 Ϯ 5 E149A 97 Ϯ 895Ϯ 192Ϯ 3 104 Ϯ 11 99 Ϯ 12 K153A 71 Ϯ 191Ϯ 186Ϯ 596Ϯ 11 93 Ϯ 13 http://www.jimmunol.org/ K154A 82 Ϯ 18 90 Ϯ 178Ϯ 689Ϯ 19 91 Ϯ 4 K156A 106 Ϯ 283Ϯ 11 97 Ϯ 10 87 Ϯ 796Ϯ 17

a Experiments were carried out as described in Table IV.

In our initial analysis of the multiple alanine substitution hIF- Residues on hIFNAR2 that are important for ligand binding are NAR2-IgG mutants, we demonstrated a significant reduction in the functionally important binding of the anti-hIFNAR2 mAbs to some of these mutants. Soluble hIFNAR2 immunoadhesin molecules may assume a tertiary However, in the more detailed analysis of single alanine substitu- by guest on September 24, 2021 structure in solution that is different from the receptor cellular mem- tion mutants, we were not able to detect any single residue that had brane structure. To extend the findings obtained with the soluble hIF- a significant effect on the binding of these mAbs. This suggests NAR2-IgG, we generated murine L929 cells that express wild-type that multiple regions of hIFNAR2-IgG are involved in mAb bind- and mutant human IFN receptors. The species specificity of the IFN ing and that the regions on hIFNAR2 interacting with mAbs are system has been widely used as a means to understand the receptor much larger than the region interacting with hIFN-␣2/␣1 as shown function. Murine cells, such as L929 cells, that express mIFNAR1 and in other systems (25). mIFNAR2 will respond to all murine IFNs and some, but not all, human IFNs. For example, hIFN-␣8, hIFN-␣2, and hIFN-␤ signal with human IFNAR, while hIFN-␣1, hIFN-␣5, and hIFN-␣10 can signal with human or murine IFNARs. Mouse L929 cells transfected with hIFNAR1 respond to hIFN-␣8, but do not respond to hIFN-␣2 or hIFN-␤, demonstrating that the mIFNAR2 can interact with hIF- NAR1 to effect a signaling complex (5). We utilized this system to test the signaling ability of the hIFNAR2 mutants. We transfected L929 cells with a vector that encodes the cDNA for hIFNAR2 and established a stable cell line that expresses the full-length hIFNAR2 chain. The level of hIFNAR2 expression from a representative clone (L929.R2.19) was determined by flow cytometry using mAb 3B2 (Fig. 3A). To demonstrate that the expressed human receptor is able to bind hIFN-␣2, we determined the binding of biotinylated hIFN-␣2 (Bio-hIFN-␣2) to the L929.R2.19 cells by flow cytometry (Fig. 3B). The L929.R2.19 cell line was then tested for its ability to respond to type I IFNs in an EMSA (Fig. 3C). Mouse fibroblast L929 cells constitutively express murine IFN receptors. Thus, L929.R2.19 cells respond to murine IFN-␣11 (mIFN-␣11) with formation of an activated ISGF3 complex (Fig. 3C). Human IFN-␣1 is recognized by the FIGURE 2. The binding of hIFN-␣2/␣1 to hIFNAR2-IgG alanine mu- murine receptor, so mouse L929 cells will respond to the hybrid tants, as determined by an ELISA. Experiments were performed as de- human type I IFN (hIFN-␣2/␣1) we used in the binding studies scribed in Fig. 1 using various concentrations of Bio-hIFN-␣2/␣1. cited above. Therefore, this hybrid human IFN could not be used The Journal of Immunology 771 Downloaded from

FIGURE 3. Receptor expression, binding of hIFN-␣2, and signaling ability of L929 cells expressing hIFNAR2. A, The level of expression of a representative clone, L929.R2.19, was determined using FITC-mAb 3B3 (solid line, FITC-mAb 3B3; dotted line, unstained control).B, The ligand- binding ability was determined using Bio-hIFN-␣2, followed by PE- streptavidin (solid line, Bio-hIFN-␣2 plus PE-streptavidin; dotted line, PE- http://www.jimmunol.org/ streptavidin alone). C, The ability of L929.R2.19 cells to mediate the IFN signal was determined by EMSA, which detects ISGF3 complex formation. As a control, L929. R1 cell line expressing hIFNAR1 is included. in the studies utilizing the transfected L929 cells. To examine the response to other human IFNs, hIFN-␣2, hIFN-␣8, and hIFN-␤ were tested. After incubation of L929.R2.19 cells with 25 ng of hIFN-␣2/ml or 100 U of hIFN-␤/ml, an ISGF3 complex is ob- by guest on September 24, 2021 served. This complex is the same as that formed in response to mIFN-␣11 and can be shifted by Abs to murine Stat1 or murine Stat2 (data not shown). In contrast, there is no response of L929.R2.19 cells to hIFN-␣8 (Fig. 3C). The hIFN-␣8 is biologi- FIGURE 4. Receptor expression, binding of hIFN-␣2, and signaling cally active, because in a L929.R1 cell line expressing the hIF- ability of L929-transfected cells expressing hIFNAR2 mutants. All of the NAR1 chain, an ISGF3 complex is observed after hIFN-␣8 treat- experiments were conducted as described in Fig. 3. L929.R2.74–78, L929 ment (Fig. 3C). Thus, from this experiment, we conclude that in cells expressing hIFNAR2 with multiple mutation in residues 74–78; the presence of the hIFNAR2 chain, hIFN-␣2 and hIFN-␤, but not L929.R2.78, L929 cells expressing hIFNAR2 with a single alanine mutation in residue E78; L929.R2.74, L929 cells expressing hIFNAR2 with a hIFN-␣8, will signal a biological response in a mouse cell line. single alanine mutant in residue R74. Two cDNAs were constructed that encode receptor mutants with alanine substitutions for multiple hydrophobic residues (2 mutant, residues 29–33, and 5 mutant, residues 74–78) in the with single alanine substitutions (R74A, E78A, W101A, I104, ECD. Six cDNAs were constructed that encode receptor mutants D105A, and E109A) in the ECD. Each construct was transfected into the mouse L929 cell line, and stable cell lines were established (Table VI). Fig. 4 illustrates the level of hIFNAR2 expression using mAb Table VI. Summary of signal transduction of L929 hIFNAR2 ECD 3B2 and the binding of biotinylated hIFN-␣2 (Bio-hIFN-␣2) of rep- mutantsa resentative clones (L929.R2.74–78, L929.R2.78, and L929.R2.74) from three mutant cell lines, as determined by ﬂow cytometry. IGSF Complex Formation hIFNAR2 Receptor All of these clones expressed hIFNAR2, although the level of Mutants Expression mIFN-␣11 hIFN-␣2 hIFN-␤ receptor expression was variable among all of the mutant clones examined due to the clonal variation (Fig. 4A). The L929.R2.74 WT ϩϩϩ ϩ ϩ ϩ 29–33 ϩϩ ϩ ϩ ϩ mutant receptor cell line expressed a low level of Bio- 74–78 ϩϩϪϩ/Ϫ hIFN-␣2 binding, while this binding was undetectable on the R74 ϩϩ ϩ ϩ ϩ L929.R2.74–78 and the L929.R2.78 cell line (Fig. 4B). All clones, E78 ϩϩ ϩ Ϫ ϩ ϩϩϪϩincluding L929.R2.19 (Fig. 3B), tested demonstrated a low level of W101 ␣ I104 ϩϩϪϩBio-hIFN- 2 binding that did not correlate well with the level of D105 ϩϩϪϩreceptor expression. This difference may be a result of the bioti- E109 ϩϩϩϩnylation of the IFN. aThe degrees of response (Ϫ, ϩ/Ϫ, ϩ, ϩϩ, and ϩϩϩ) were arbitrarily deter- The cell lines that express a mutant hIFNAR2 were tested for mined by comparing the results shown in Fig. 4. their ability to respond to mIFN-␣11, hIFN-␣2, and hIFN-␤ in an 772 LIGAND-BINDING REGION OF hIFNAR2 Downloaded from

FIGURE 5. A, Location of residues on hIFNAR2 that are important for binding of hIFN-␣2/␣1. The model of hIFNAR2 was made by displaying its sequence on the structure of tissue factor (28). The crucial binding residues for hIFN-␣2/␣1 are shown in orange. B, Hypothetical model for the hIFNAR1 and hIFNAR2 interaction. C, Regions on hIFNAR2 important for binding of mAbs 1D3, 1F3, and 3B7. The binding regions of 1D3 and 3B7 are shown in dark blue and royal blue, respectively. The binding region of 1F3 includes all blue colors. http://www.jimmunol.org/

EMSA. The L929.R2.74–78 cell line that contains the receptor Discussion with alanine substitutions in residues 74–78 does not respond to On the basis of their structural homology, the cytokine receptor hIFN-␣2 and weakly responds to hIFN-␤, yet remains responsive superfamily can be divided into two classes (26, 27). The ﬁrst class to mIFN-␣11, indicating that there is no defect in the general path- of the cytokine superfamily includes the receptors for human way of ISGF3 formation. The single alanine substitution mutants of

growth hormone, erythropoietin, GM-CSF, IL-3, IL-4, IL-6, and by guest on September 24, 2021 residue R74 and E78 were examined to determine which of these two IL-7, while the class 2 cytokine receptor family includes the re- residues are important for signal transmission. The L929.R2.74 cell ceptors for IFN-␥, IFN-␣, and IL-10. The structures of the human line, which contains a hIFNAR2 with a single alanine substitution at growth hormone receptor (28) and tissue factor, which belong to residue R74, is responsive to hIFN-␣2 and hIFN-␤, even though the class 1 and class 2 cytokine receptor superfamily, respectively, ␣ binding of the Bio-hIFN- 2 is low. The L929.R2.78 cell line does not have been well characterized. The main difference between these ␣ ␤ respond to hIFN- 2, yet remains responsive to hIFN- . These data two receptors is the angle between ϳ100-aa domains. The angle indicate that residue E78 in this small mutated region is involved in between domains in the class 1 receptor family is ϳ85°, while that ␣ mediating a functional response to hIFN- 2. of class 2 receptor family is ϳ120°. Using the structure of tissue The results obtained in all L929 transfectants, including multiple factor as a backbone (29), we have constructed a computer model mutants (residues 29–33 and residues 74–78) and single mutants of the ECD of hIFNAR2 (Fig. 5A) to have a better understanding (R74, E78, W101, I104, D105, and E109), are summarized in Ta- of the possible structure of hIFNAR2. The multiple alanine sub- ble VI. All of these mutants expressed the ECD-hIFNAR2 on the stitutions in hIFNAR2 residues 49–54, 68–72, 74–78, 105–109, cell surface membrane, as determined by flow cytometry, and re- 133–139, and 153–156 completely abolished the binding of hIFN- ␣ ␤ sponded to mIFN- 11 and hIFN- . While all single mutants re- ␣2/1. Although it appears to be puzzling that so many residues ␤ sponded to hIFN- , only two single mutants, R74A and E109A, from various portions of the hIFNAR2 sequence contribute to the responded to hIFN-␣2. These results further support the finding binding of hIFN-␣2/1, our computer model of hIFNAR2 reveals that residues E78, W101, I104, and D105 are crucial for mediating that all of these residues come together to form a protrusion. Fur- the hIFN-␣2 response. These findings in conjunction with the re- thermore, results obtained with single mutant analysis (Table V sults from the ligand-binding studies performed with the soluble and Fig. 2) demonstrate that residues E78, W101, I104, and D105 immunoadhesin suggest that the conformation of the soluble hIF- are crucial for the binding of hIFN-␣ 2/1, and these residues form NAR2-IgG closely mimics the structure of the membrane-bound a small protrusion in our model (Fig. 5A). receptor. We demonstrate that the regions on the soluble receptor In our previous work describing the neutralizing epitope on hIF- that are recognized by the hybrid hIFN-␣2/␣1 are the same as NAR1 recognized by blocking mAbs, we proposed a structural those recognized by the hIFN-␣2 on the hIFNAR2-expressing cell model of hIFNAR1 (19). The 409-residue ECD of hIFNAR1 is lines. While all type I IFNs utilize IFNAR2 to signal, we have almost twice as big as the 217-residue ECD of hIFNAR2. It would demonstrated that different receptor regions are utilized by differ- be interesting to know how these hIFNAR1/hIFNAR2 come to- ent IFNs in generating a biological response. Thus, in this system, gether in the presence of IFN. To visualize the spatial interaction hIFN-␣8 is unable to signal without the hIFNAR1 receptor, while of these two receptor chains with an IFN, we considered the fol- hIFN-␤ most likely uses residues on hIFNAR2 that are distinct lowing information. First, our present study demonstrates that both from those utilized by hIFN-␣2. domains of IFNAR2 interact with the ligand, as proposed by Seto The Journal of Immunology 773 et al. (30). Second, the second and third domains of hIFNAR1 may 5. Uze, G., G. Lutfalla, and I. Gresser. 1990. Genetic transfer of a functional human inter- ␣ be involved in mediating an IFN signal, and the region around the feron receptor into mouse cells: cloning and expression of its cDNA. Cell 60:225. 6. Novick, D., B. Cohen, and M. Rubinstein. 1994. The human interferon ␣/␤ re- residue K249 is crucial (19). Finally, hIFNAR2 has ϳ15 residues ceptor: characterization and molecular cloning. Cell 77:391. at the C-terminal portion of the ECD that may allow the hIFNAR2 7. Domanski, P., M. Witte, M. Kellum, M. Rubinstein, R. Hackett, P. Pitha, and O. R. Colamonici. 1995. Cloning and expression of a long form of the ␤ subunit of to come into close proximity to the second and third domains of the interferon ␣␤ receptor that is required for signaling. J. Biol. Chem. 270:21606. hIFNAR1. Based on this information, we designed a model that 8. Cohen, B., D. Novick, S. Barak, and M. Rubinstein. 1995. Ligand-induced as- shows how hIFNAR1 and hIFNAR2 may come together in the sociation of the type I interferon receptor components. Mol. Cell. Biol. 15:4208. 9. Russell-Harde, D., H. Pu, M. Betts, R. N. Harkins, H. D. Perez, and E. Croze. presence of an IFN (Fig. 5B). This needs to be confirmed using 1995. Reconstitution of a high affinity binding site for type I interferons. J. Biol. crystal structure analysis. Chem. 270:26033. The results shown in Table IV and Fig. 5C demonstrate that the 10. Cutrone, E. C., and J. A. Langer. 1997. Contributions of cloned type I interferon receptor subunits to differential ligand binding. FEBS Lett. 404:197. binding of mAb 1F3 overlaps most closely with that of IFN-␣ 2/1. 11. Platanias, L. C., P. Domanski, O. W. Nadeau, T. Yi, S. Uddin, E. Fish, These binding areas are located both on domain 1 and 2 of IF- B. G. Neel, and O. R. Colamonici. 1998. Identification of a domain in the ␤ subunit of the type I interferon (IFN) receptor that exhibits a negative regulatory NAR2. The main area recognized by mAb 3B7 includes residues effect in the growth inhibitory action of type I IFNs. J. Biol. Chem. 273:557. 49–55 and 68–72 in domain 1 of hIFNAR2, while mAb 1D3 rec- 12. Petricoin, E. F., S. Ito, B. L. Williams, S. Audet, L. F. Stancato, A. Gamero, ognizes mainly residues 133–139 and 153–157 in domain 2. Res- K. Clouse, P. Grimley, A. Weiss, J. Beeler, et al. 1997. Antiproliferative action of interferon-␣ requires components of T-cell-receptor signalling. Nature 390:629. idues important for the most potent blocking mAb 3B7 overlap 13. Darnell, J. E. 1994. STATs and gene regulation. Science 277:1630. only partially, but are in close proximity to the crucial binding 14. Haak-Frendscho, M., S. A. Marsters, J. Mordenti, S. Brady, N. A. Gillett, S. A. Chen, ␣ region of IFN-␣ 2/1 (Fig. 5C). These results suggest that mAb 3B7 and A. Ashkenazi. 1994. Inhibition of TNF by TNF receptor immunoadhesin: Ϫ1 comparison to an anti-TNF monoclonal antibody. J. Immunol. 152:1347. ϭ Downloaded from (Kd 1 pM) exerts its potent blocking activity due to its high 15. Rehberg, E., B. Kelder, E. G. Hoal, and S. Pestka. 1982. Specific molecular activities affinity to hIFNAR2 and its close proximity to the crucial ligand- of recombinant and hybrid leukocyte interferons. J. Biol. Chem. 257:11497. 16. Horisberger, M. A., and S. Di Marco. 1995. Interferon-␣ hybrids. Pharmacol. binding region on hIFNAR2. In contrast, mAb 1D3, which binds Ther. 66:507. to a region in domain 2 of hIFNAR2, showed no blocking activity in 17. Chuntharapai, A., and K. J. Kim. 1997. Generation of monoclonal antibodies to the ligand receptor-binding ELISA, but demonstrated weak blocking chemokine receptors. Methods Enzymol. 288:15. 18. Yelton, D. E., B. A. Diamond, S. P. Kwan, and M. D. Scharff. 1978. Fusion of activities in the antiviral assay and in EMSA. This suggests the pos- mouse myeloma with spleen cells. Curr. Top. Microbiol. Immunol. 81:1.

sibility that mAb 1D3 binding may interfere with the interaction of 19. Lu, J., A. Chuntharapai, J. Beck, S. Bass, A. Ow, A. M. De Vos, V. Gibbs, and http://www.jimmunol.org/ hIFNAR1 and hIFNAR2 since the residues that are important for K. J. Kim. 1998. Structure-function study of the extracellular domain of the human IFN-␣ receptor (hIFNAR1) using blocking monoclonal antibodies: the mAb 1D3 binding are in domain 2, a lower portion of hIFNAR2. role of domains 1 and 2. J. Immunol. 160:1782. It is well known that all of the type 1 IFNs, hIFN-␣, hIFN-␤, and 20. Kim, K. J., M. Alphonso, C. H. Schmelzer, and D. Lowe. 1992. Detection of ␻ human leukemia inhibitory factor by monoclonal antibody based ELISA. J. Im- hIFN- , bind to the same cell surface receptor, but there is some munol. Methods 156:9. disparity in the biological effects demonstrated by different mem- 21. Blake, D. A., P. Chakrabarti, M. Khosraviani, F. M. Hatcher, C. M. Westhoff, bers of the type 1 IFN family. For example, IFN-␤ is more effec- P. Goebel, D. E. Wylie, and R. C. Blake II. 1996. Metal binding properties of a monoclonal antibody directed toward metal-chelate complexes. J. Biol. Chem. tive than other type 1 IFNs in the treatment of multiple sclerosis 271:27677. (31, 32) and induces a very strong cytoplasmic association be- 22. Craig, L., P. C. Sanschagrin, A. Rozek, S. Lackie, L. A. Kuhn, and J. K. Scott. tween hIFNAR1 and hIFNAR2 of the type 1 IFNAR (33, 34). It 1998. The role of structure in antibody cross-reactivity between the peptides and by guest on September 24, 2021 folded proteins. J. Mol. Biol. 281:183. has also been shown that hIFN-␣2 and hIFN-␤ require distinct 23. Measurement of antiviral activity induced by interferons, ␣, ␤, and ␥. 1992. In intracytoplasmic regions of the hIFNAR2 for signaling (35, 36). Current Protocols in Immunology, Vol. 1. J. E. Coligan, A. M. Kruisbeek, D. H. Margulies, E. M. Shevach, and W. Strober, eds. Wiley-Interscience, New The results shown in Fig. 5 demonstrate that there are differences York. Unit 6.9.1. in the hIFNAR2 residues interacting with hIFN-␣2 and those with 24. Kunkel, T. A., J. D. Roberts, and R. A. Zakour. 1987. Rapid and efficient site- hIFN-␤. Our findings further support the notion that there are sub- specific mutagenesis without phenotypic selection. Methods Enzymol. 154:367. ␣ ␤ 25. Cunningham, B. C., and J. A. Wells. 1991. Rational design of receptor-specific tle but important differences between IFN- and IFN- in the way variants of human growth hormone. Proc. Natl. Acad. Sci. USA 88:3407. they interact with IFNAR. 26. Bazan, J. F. 1990. Structural design and molecular evolution of a cytokine re- Results shown in Fig. 3 demonstrate that murine L929-trans- ceptor superfamily. Proc. Natl. Acad. Sci. USA 87:6934. 27. Bazan, J. F. 1990. Haemopoietic receptors and helical cytokines. Immunol. Today fected cells with hIFNAR2 respond to hIFN-␣2 and hIFN-␤, but 11:350. not hIFN-␣8. The fact that hIFN-␣8 mediates a signal through 28. De Vos, A. M., M. Ultsch, and A. A. Kossiakoff. 1992. Human growth hormone and extracellular domain of its receptor: crystal structure of the complex. Science hIFNAR1 expressed on murine L929 cells, but not through hIF- 255:306. NAR2 expressed on L919 cells, indicates the important role of 29. Muller, Y. A., M. H. Ultsch, and A. M. de Vos. 1996. The crystal structure of the hIFNAR1 in mediating the hIFN-␣8 signal on the murine cell extracellular domain of human tissue factor refined to 1.7 Å resolution. J. Mol. Biol. 256:144. background. By understanding the subtle differences among dif- 30. Seto, M. H., R. N. Harkins, M. Adler, M. Whitlow, W. B. Church, and E. Croze. ferent subspecies of the type 1 IFNs, we may have a better under- 1995. Homology model of human interferon-␣8 and its receptor complex. Protein standing of how a single receptor sorts the various type 1 IFN Sci. 4:655. 31. Anonymous. 1989. Interferon-␣ and transfer factor in the treatment of multiple signals. This is an area for future investigation. We believe that the sclerosis: a double-blind, placebo-controlled trial. J. Neurol. Neurosurg. Psychi- information provided in this study will facilitate the understanding atry 52:566. 32. Bever, C. T. J., H. S. Panitch, H. B. Levy, D. E. McFarlin, and K. P. Johnson. of type 1 IFN receptor structure. 1990. Induction of ␥ interferon in patients with chronic progressive multiple sclerosis did not cause clinical deterioration. Neurology 40:259. Acknowledgments 33. Platanias, L. C., S. Uddin, P. Domanski, and O. R. Colamonici. 1996. Differences We thank Brian Fendly for careful reading of the manuscript and A. Mi- in interferon ␣ and ␤ signaling: interferon ␤ selectively induces the interaction of ␣ ␤ ronov for secretarial assistance. the and L subunits of the type I interferon receptor. J. Biol. Chem. 271:23630. 34. Rani, M. R. S., G. R. Foster, S. Leung, D. Leaman, G. R. Stark, and R. M. Ransohoff. 1996. Characterization of ␤-R1, a gene that is selectively in- References duced by interferon ␤ (IFN-␤) compared with IFN-␣. J. Biol. Chem. 271:22878. 1. Weissmann, C. 1986. The interferon genes. Prog. Nucleic Acid Res. Mol. Biol. 35. Croze, E., D. Russell-Harde, T. C. Wagner, H. Pu, L. M. Pfeffer, and H. D. Perez. 33:251. 1996. The human type I interferon receptor: identification of the interferon ␤-spe- 2. Pestka, S., J. A. Langer, K. C. Zoon, and C. E. Samuel. 1987. Interferons and their cific receptor-associated phosphoprotein. J. Biol. Chem. 271:33165. actions. Annu. Rev. Biochem. 56:727. 36. Abramovich, D., L. M. Shulman, E. Ratovitski, S. Harroch, M. Tovey, P. Eid, and 3. Pestka, S. 1997. The interferon receptors. Semin. Oncol. 24, no. 3 Suppl 9:s9–18. M. Revel. 1994. Differential tyrosine phosphorylation of the IFNAR chain of the 4. Domanski, P., and O. R. Colamonici. 1996. The type-I interferon receptor: the type I interferon receptor and of an associated surface protein in response to long and the short of it. Cytokine Growth Factor Rev. 7:143. IFN-␣ and IFN-␤. EMBO J. 13:5871.